CZ20002339A3 - Insoluble compositions for affecting blood glucose - Google Patents

Abstract

The invention relates to insoluble compositions comprising a protein selected from the group consisting of insulin, insulin analogues and proinsulins; a derivatised protein selected from the group consisting of derivatised insulin, derivatised insulin analogue and derivatised proinsulin; a complexing compound; a hexamer-stabilising compound; an adjunct metal cation. The formulations comprising the insoluble composition are suitable for parenteral and non-parenteral administration for the treatment of hyperglycaemia and diabetes. The microcrystalline form of the insoluble precipitate is pharmaceutically analogous to the neutral protamine Hagedorn (NPH) form of insulin crystals. It has surprisingly been found that suspension formulations comprising such insoluble compositions have unique and controllable dissolution properties which enable therapeutically advantageous glucodynamics compared to NPH insulin formulations.

Description

Insoluble compositions for influencing blood glucose

Technical field

The invention relates to insoluble compositions for influencing blood glucose. More generally, the invention relates to the field of pharmaceutical treatment of diabetes and hyperglycemia.

State of the art

For a long time, the goal of insulin therapy has been to mimic the pattern of endogenous insulin secretion in normal individuals. The daily physiological need for insulin varies and can be divided into two phases:

1) the absorption phase, requiring an influx of insulin to process the blood glucose dose related to food intake, and

2) the post-absorptive phase, requiring sustained insulin delivery to regulate hepatic glucose output to maintain optimal blood glucose levels during the non-meal period. Accordingly, effective therapy for people with diabetes usually involves the combined use of two types of exogenous insulin preparations: a fast-acting mealtime insulin given by a single injection, and a long-acting, so-called basal insulin given by injection once or twice a day to control blood glucose levels between meals. The ideal basal insulin should provide a prolonged and flat duration of action, that is, it will control blood glucose levels for at least 12 hours and preferably for 24 hours or more, without the risk of hypoglycemia. In addition, the ideal basal insulin should be miscible with soluble mealtime insulin and should not cause irritation or reaction at the injection site. Ultimately, it should be possible for the patient to resuspend basal insulin easily and uniformly before administration.

It is understood by those skilled in the art that long-acting insulin preparations are obtained by formulating normal insulin into microcrystalline suspensions for subcutaneous injection. Examples of commercial basal insulin preparations include NPH (Neutral Protamine Hagedorn) insulin, protamine zinc insulin (PZI) and ultralente (UL).

Early versions of the current commercial NPH insulins, which contained large amounts of protamine, were developed in the 1930s by Scott et al. [J. Pharmacol. Exp. Ther. 58:78 ff. (1936)] and Hagedorn et al. [J. Am. Med. Assoc. 106:177-180 (1936)]. In 1946, NPH insulin was developed that had an isophane ratio of insulin to protamine, along with zinc, by Krayenbuhl et al. [Rep. Sten Mam. Hosp. Nord. Insulinlab. 1:60 ff. (1946)]. These researchers found that when insulin and protamine were combined in the so-called isophane ratio at neutral pH, in the presence of zinc and a phenolic compound, an amorphous precipitate formed and that upon standing the amorphous precipitate transformed into elongated tetragonal crystals that had pyramidal-shaped ends. These crystals have been described as rodlike. The isophane ratio of insulin to protamine sulfate has been observed to be approximately 0.09 mg protamine sulfate per mg insulin. Zinc is required in an amount of at least approximately 3.5 μς per mg insulin and the phenolic compound in a concentration of • 9

I *

9 <

» 9 9 <

99 higher than about 0.1%.

NPH insulin is the most commonly used insulin preparation, accounting for 50 to 70 percent of the insulin used worldwide. It is a suspension of a microcrystalline complex of insulin, zinc, protamine, and one or more phenolic preservatives. NPH insulin preparations that are commercially available contain human insulin, porcine insulin, bovine insulin, or mixtures thereof. NPH-like preparations containing a monomeric insulin analog, LysB298,ProB29-analog of human insulin, are also known in the art [and are hereinafter referred to as NPL; De Felippis, M. R., U.S. Patent No. 5,461,031, issued October 24, 1995; De Felippis, M. R., U.S. Patent No. 5,650,486, issued July 22, 1997; and De Felippis, M. R., U.S. Patent No. 5,650,486, issued July 22, 1997; Patent No. 5,747,642, issued May 5, 1998]. It is generally accepted that NPH insulin provides prolonged blood glucose control compared to regular insulin because the insulin must first dissolve from the NPH insulin microcrystals before it is absorbed. With regular insulin, no dissolution is required prior to absorption. With NPH insulin, dissolution is the rate-controlling factor in determining pharmacodynamics and pharmacokinetics.

NPH insulin microcrystals have a distinct rod-like morphology with typical dimensions of approximately 5 microns in length, 1 micron in thickness, and 1 micron in width. The prolonged duration of action of NPH insulin microcrystals is due to their slow absorption from the subcutaneous injection site.

However, current NPH insulin regimens do not provide the ideal flat pharmacokinetics required to maintain optimal blood glucose levels during fasting for extended periods between meals. As a result, NPH insulin therapy may lead to undesirably high blood insulin levels, which may cause life-threatening hypoglycemia.

In addition to not achieving an ideal flat pharmacokinetic profile, the duration of action of NPH insulin is also not ideal. Specifically, the biggest problem associated with NPH therapy is the so-called dawn phenomenon, which is hyperglycemia that leads to a loss of effective glucose control overnight while the patient is sleeping. These control deficiencies contribute to serious long-term complications of diabetes and bring significant disadvantages to the quality of life of the patient.

glycemic medical and reduction

Protamine zinc insulin (PZI) has a composition similar to NPH but contains higher amounts of protamine and zinc than NPH. PZI preparations can be formulated as amorphous precipitates with intermediate duration of action or as a crystalline material with long duration of action. However, PZI is not an ideal basal insulin because it is not miscible with soluble mealtime insulin and the high zinc and protamine contents can cause irritation or injection site reactions.

does not contain built-in

Human insulin ultralente is a microcrystalline insulin preparation that contains a higher level of zinc than NPH and neither protamine nor phenolic preservatives in the microcrystal. Human ultralente preparations provide an intermediate duration of action that is not suitably flat and • » 99

I « · « » · 99 • · · · · · • 9 do not form stable mixtures with insulin. In addition, they are difficult to resuspend.

Attempts have been made to overcome the persistent shortcomings of known insulin suspensions. Fatty acid acylated insulins have been investigated for basal blood glucose control [Havelund, S. et al., WIPO published application WO95/07931, March 23, 1995]. Their prolonged duration of action is due to the binding of the fatty acid portion of these molecules to serum albumin. The fatty acyl chain length of these molecules is such as to take advantage of the fatty acid binding to serum albumin. The fatty acid chains used in fatty acid acylated insulins are typically longer than about ten carbon atoms, and a chain length of fourteen to sixteen carbon atoms is optimal for fatty acid binding to serum albumin and for prolonged duration of action.

Unlike NPH insulin, which is insoluble, the aforementioned fatty acid acylated insulins are soluble at usual therapeutic insulin concentrations. However, the duration of action of these preparations may not be sufficiently short or flat to provide ideal basal control, and they are weaker than insulin, requiring the administration of larger amounts of drug [Radziuk, J. et al., Diabetologia 41:116-120, 489-490 (1998)].

Whittingham, J. L. et al. [Biochemistry 36:2826-2831 (1997)] crystallized the B29-Ns-tetradecanoyl-des(B30)-analog of human insulin as a hexameric complex with zinc and phenol for structural studies by X-ray crystallography. The hexamer was found to have an R6 conformation and to have certain properties that distinguish it from human insulin hexamers. Whittingham et al. do not describe any pharmaceutical or pharmacological properties of the crystals that were formed, nor do they suggest that the crystals would have beneficial properties for the treatment of diabetes or hyperglycemia. It is not possible to predict from the work of Whittingham et al. that protamine-containing NPH-type crystals could be formed from derivatized insulins and insulin analogs or what the pharmacokinetic or pharmacodynamic response of such crystals might be.

There remains a need to identify insulin preparations that have a flatter and longer duration of action than NPH insulin, that are miscible with water-soluble mealtime insulins, that are easily resuspended, and that do not present a risk of irritation or reaction at the site of administration. The applicant has unexpectedly discovered that these properties can be provided by the use of insoluble compositions that contain a derivatized protein, a non-derivatized protein, zinc, protamine, and a phenolic preservative. In addition to the above properties, the insoluble compositions of the invention have the flexibility of duration of control and shape of the glucodynamic response profile. They are considered to be compositions that operate with controlled release and in which the rate of release is controlled by the ratio and nature of the derivatized protein. One object of the present invention is therefore an insoluble composition containing a non-derivatized protein, a derivatized protein, a complexing compound, a hexamer-stabilizing compound, and a divalent metal cation. Other objects of the present invention are ...

- 7 • · · · of the invention, which will be discussed further, are the preparation, formulation of pharmaceutical compositions and the use of such compositions.

The applicant is not aware of any examples of mixtures of derivatized and non-derivatized insulins as those terms are understood herein. Crystals consisting of proinsulin and insulin are known [Steiner, D. F., Nature 243:528530 (1973); Low, B. W. et al., Nature 248:339-340 (1974)] and crystals consisting of insulin or an insulin analog having approximately the same isoelectric point as insulin and an insulin analog containing additional amino acids [Dörschug, M. et al., U.S.

5,028,587, issued July 2, 1991].

basic patent no.

preparations attributed

Steiner grew crystals consisting of proinsulin and insulin in molar ratios of approximately 1:11, 1:5, 1:2, and 1:1 (i.e., 0.5, 1, 2, and 3 moles of proinsulin to 6 moles of total insulin and proinsulin) in 0.095 M sodium citrate, pH 6.0, 0.03 M NaCl, 0.012 M ZnCl ₂ , and 16% acetone. The amount of proinsulin strongly influenced the rate of crystallization. The crystals were very different from crystals of pure insulin under the same conditions and were characterized as rhombohedral crystals with rounded edges. There was great variation within and between individuals. The usefulness of crystallization of proinsulin and insulin was that it facilitated the isolation of small amounts of proinsulin and related structures from pancreatic extracts. The author considered whether crystallizations could occur between precursors and the resulting peptides and between other closely related proteins.

···· Φ· φ · φ • · · φ φ φ · φφ φφ

Low, B. W. et al. formed very large crystals consisting of equimolar proportions of bovine or porcine insulin and their corresponding proinsulins, with the proinsulin and insulin being formed into homogeneous hexamers prior to crystallization. X-ray crystallographic analysis and quantitative electrophoresis supported the conclusion that the unit cells in the crystals were formed by twelve insulin hexamers and twelve proinsulin hexamers. Specifically, it was stated that no studies were known to indicate that insulin and proinsulin form mixed dimers and hexamers in solution.

at least one at the C-terminal

Dórschug, M. et al. have described crystals consisting of insulin, des(PheB1) insulin, des(ThrB30) human insulin or des(AlaB30) bovine insulin and insulin having a basic modification at the B-terminal end (modified insulin). Such modified insulins are described, for example, in European Patent Application No. 132,765. Globin or protamine sulfate were mentioned as auxiliary compounds that can be used in the crystalline preparations. No examples of the use of protamine are given nor is there any indication that the inventors considered the effect of adding such compounds. Furthermore, the modified insulins used in Dórschug et al. are different from the derivatives used in the present invention.

As mentioned above, the applicant unexpectedly observed that if a protein selected from the group consisting of insulin, insulin analog and proinsulin is made less soluble in f 00

0 0 0 • · 00 • 0 0 0 ·

0· 0

0 0 0

0000 ♦ ·

aqueous solvent or more lipophilic derivatization of a group at one or more of its reactive sites, the derivatized protein and the non-derivatized protein selected from the group consisting of insulin, insulin analog and proinsulin can be incorporated into insoluble precipitates and into NPH-like crystals containing protamine. When such proteins are co-precipitated or crystallized to form insoluble compositions, the dissolution rate at which the proteins dissolve from the insoluble compositions is greatly reduced compared to the dissolution rate at which physically similar insoluble compositions containing the non-derivatized protein dissolve.

The applicant has also discovered that both amorphous precipitates and microcrystals containing derivatized protein, protein, complexing compound, divalent metal cation and hexamer-stabilizing compound provide flatter and longer duration of action than physically similar microcrystals consisting solely of non-derivatized protein. Furthermore, the applicant has unexpectedly discovered that the advantage of flatter and longer duration of action can be achieved even with amorphous precipitates containing one of the proteins and the derivatized protein.

The essence of the invention

The present invention relates in its broadest scope to insoluble compositions comprising a derivatized protein selected from the group consisting of an insulin derivative, an insulin analog derivative, and a proinsulin derivative, a protein selected from the group consisting of insulin, insulin analogs ···· ·· • · · · · · · · • ♦ · These insoluble compositions are useful for the treatment of diabetes and hyperglycemia and offer the advantage of a flatter and longer duration of action than NPH insulin. The insoluble compositions are admixable into a formulation with a soluble protein or a soluble derivatized protein or both. By varying the ratio of protein to derivatized protein, the amount of prolongation of the duration of action can be finely controlled over a very wide range of durations of action from about the same as NPH insulin to far greater than NPH insulin.

acylated acid including fatty

More specifically, the present invention provides insoluble protein and fatty acid acylated protein compositions useful for the treatment of diabetes and hyperglycemia. These compositions comprise a fatty protein selected from the group consisting of an acid acylated insulin, a fatty acid acylated insulin analog, and a fatty acid acylated proinsulin, a protein selected from the group consisting of insulin, insulin analogs, and proinsulin, protamine, a phenolic preservative, and zinc. The present invention differs from prior art fatty acid acylated insulin compositions in that the extended duration of action of the present invention is not necessarily related to albumin binding, although albumin binding may further extend the duration of action of certain compositions of the present invention.

The present invention relates to microcrystals comprising a protein selected from the group consisting of insulin, an insulin analog and proinsulin, a derivatized protein selected from the group consisting of derivatized insulin, a derivatized insulin analog and a derivatized proinsulin, a complexing compound, a divalent metal cation and a hexamer-stabilizing compound. The microcrystals of the present invention are useful for treating diabetes and for controlling blood glucose levels in a patient in need thereof.

precipitate comprising derivatized proinsulin; comprising

The present invention relates to an amorphous comprising a protein selected from the group consisting of insulin, an insulin analog, and a protein selected from the group consisting of insulin, a derivatized insulin analog, and a derivatized proinsulin, a complexing compound, a divalent metal cation, and a hexamer-stabilizing compound. The amorphous of the present invention are useful in diabetes and for controlling blood glucose levels in a patient in need thereof. They are also useful as intermediates for forming the microcrystals of the present invention.

precipitates for treatment

The present invention relates to aqueous suspension preparations comprising an insoluble composition and an aqueous solvent.

···· 4· ·· ♦· ·* ' • φ · · 4 * · 4 · ·

4 4 · · ·· · · · ·· 4 4 4 4 4 4 · · ··

One such aqueous suspension preparation consists of a microcrystalline composition according to the present invention and an aqueous solvent. Another such aqueous suspension preparation consists of an amorphous precipitate according to the present invention and an aqueous solvent. The soluble aqueous phase of the disclosed suspension preparations may optionally consist of a protein such as human insulin or a soluble human insulin analog, such as a monomeric insulin analog, which controls blood glucose levels immediately after a meal and may additionally or alternatively contain a derivatized protein. The preparations according to the present invention have significantly improved pharmacodynamics compared to human NPH insulin and their duration of action can be deliberately chosen to be wide ranging, from a small prolongation compared to human NPH insulin to a very prolonged one compared to human NPH insulin.

The present invention also relates to methods for preparing hybrid hexamers, mixed hexamers, amorphous precipitates and cocrystals of the present invention,

The present invention relates to a method of treating diabetes or hyperglycemia which comprises administering to a patient in need thereof a sufficient amount of an insoluble composition of the present invention to regulate the patient's blood glucose level.

The present invention includes hybrid hexamer compositions consisting of a protein selected from the group consisting of insulin, an insulin analog, and proinsulin; a derivatized protein selected from the group consisting of derivatized insulin, derivatized insulin analogs, and derivatized

4444 ·· • · • · • 4 • · 4 • 4 44 • 4

· · 4 • 4 44 • · 4 4

4 4 4

44 proinsulins and zinc. The hybrid hexamers of the present invention are useful for treating diabetes and for controlling blood glucose levels in a patient in need thereof. They are also useful as intermediates in the formation of insoluble compositions of the present invention, which themselves are useful for treating diabetes and for controlling blood glucose levels in a patient in need thereof. It is believed that the hybrid hexamers are formed when the protein and the derivatized protein are first mixed together under conditions that strongly promote dissolution in lower aggregation states than the hexameric state and then the conditions are changed to promote the hexameric aggregation state.

The present invention includes mixed hexamer compositions comprising zinc hexamers of a protein selected from the group consisting of insulin, an insulin analog, or proinsulin and zinc hexamers of a derivatized protein selected from the group consisting of derivatized insulin, a derivatized insulin analog, or a derivatized proinsulin. The mixed hexamers of the present invention are useful for treating diabetes and for controlling blood glucose levels in a patient in need thereof. They are also useful as intermediates in the formation of insoluble compositions of the present invention, which themselves are useful for treating diabetes and for controlling blood glucose levels in a patient in need thereof. It is believed that the mixed hexamers are formed when the protein and derivatized protein are first separately dissolved under conditions that promote the hexameric aggregation state and then mixed together under conditions that further promote the hexameric aggregation state.

·· <· • ·

The present invention includes the use of insoluble compositions of the present invention for the preparation of a medicament for the treatment of diabetes or hyperglycemia.

Overview of images in drawings

The invention will be explained in more detail by means of specific examples of embodiments shown in the drawings, in which

Fig. 1 shows the dissolution over a period of five hours for co-crystals of the present invention having a ratio of human insulin to B29-NE-octanoyl-human insulin of 3:1 (dashed line), 1:1 (solid line), and 1:3 (thick solid line), compared to the dissolution of a preparation of human insulin crystals and protamine (dotted line).

Fig. 2 shows the dissolution data for the same experiments as Fig. 1, but with the Timeline extended to show the data for the 1:3 co-crystals up to about 13.5 hours. The dissolution of co-crystals of the present invention having a ratio of human insulin to B29-NE-octanoyl-human insulin of 3:1 (dashed line), 1:1 (solid line), and 1:3 (thick solid line) is compared to the dissolution of a preparation of human insulin crystals and protamine (dotted line).

Fig. 3 shows the dissolution of co-crystals according to the present invention, which have a ratio of human insulin to β29-Νεoctanoyl-human insulin of 1:1 (thin solid line) and 1:3 (thicker solid line), compared to the dissolution of microcrystals consisting only of β29-Νε-octanoyl-human insulin.

99 • 9 · · · ···· 99

9 9

9 9 • 9 9

9 9 9

99 insulin (thick solid line) or a preparation of human insulin crystals and protamine (dotted line).

Fig. 4 shows the dissolution of co-crystals of the present invention having a ratio of 1:3 human insulin to β29-Νεoctanoyl-human insulin (thick solid line), compared to crystal preparations of human insulin and protamine (dotted line), human insulin ultralente (thin solid line), and bovine ultralente (solid line).

Detailed description of the present invention

The term mixed hexamers refers to a mixture of protein hexamers and derivatized protein hexamers, in which the protein hexamers consist of zinc and a protein selected from the group consisting of insulin, insulin analogs, and proinsulins, and in which the derivatized protein hexamers consist of zinc and a derivatized protein selected from the group consisting of derivatized insulin, derivatized insulin analogs, and derivatized proinsulins. The level of zinc actually incorporated into the hexamers is in the range of between about 2 and about 4 zinc atoms per hexamer.

The term hybrid hexamer refers to a hexamer consisting of six monomers and zinc, wherein at least one monomer is selected from the group consisting of insulin, insulin analogs, and proinsulins, and at least one monomer is selected from the group consisting of derivatized insulin, derivatized insulin analogs, and derivatized proinsulins. The level of zinc actually incorporated into the hexamers is typically in the range of between about 2 and about 4 zinc atoms per

44

4 4 4

4 4 4

4 4 4 4

4 4 4 <4 44

4« 44

4 4 4

4 44

4 4 4

4 4 4

44

4*4* 44

4 4 • · · • 4 4 • 4 4 · • 4 44 one hexamer.

As used herein, the term cocrystal refers to a microcrystal according to the present invention.

The term insoluble composition refers to a substance in either a microcrystalline state or an amorphous precipitate state. The presence of microcrystals or amorphous precipitate can be detected visually by examination under a microscope. Solubility depends on the solvent and a given composition may be insoluble in one solvent but soluble in another.

The term microcrystal refers to a solid that consists primarily of matter in a crystalline state, with individual crystals having a predominantly single crystallographic composition and microscopic size, typically with a largest dimension ranging from 1 micron to 100 microns. The term microcrystalline refers to the state of being a microcrystal.

The term rod-like refers to a particular microcrystalline morphology, which can also be described as pyramidal-terminated tetragonal rods. The morphology of the microcrystals of the present invention can be readily determined by examination under a microscope.

The term irregular morphology is a characterization of microcrystals whose morphology, as determined by microscopic examination, is not readily classifiable by any well-known crystal type, does not represent a single type of crystal morphology, or is not readily identifiable due to the fact that

4» 44

4 4 4

4 4 4 « 4 4

4 4 4 • 4 44 ·««· »9

4 ·

4 4

4 4 4 » 4 ·

44

4# 94 • 4 4 4 • 4 44 « 4 4 4 • 4 4 »

4* 44 crystals are too small for a definite classification.

The term amorphous precipitate refers to an insoluble material that is not in crystalline form. A person of ordinary skill in the art can distinguish crystals from amorphous precipitates. The amorphous precipitates of the present invention have advantageous pharmacological properties in themselves and are also intermediates in the formation of the microcrystals of the present invention.

The term protein may have its usual meaning, i.e., a polymer of amino acids. The term protein as used herein also has a narrower meaning, i.e., a protein selected from the group consisting of insulin, insulin analogs, and proinsulins. The term non-derivatized protein also refers to a protein selected from the group consisting of insulin, insulin analogs, and proinsulins.

As used in the claims and elsewhere where the context so requires, the term total protein refers to the aggregate amount of protein (insulin, insulin analog, or proinsulin) and derivatized protein (derivatized insulin, derivatized insulin analog, or derivatized proinsulin). Although protamine and other known complexing compounds are also proteins in the broadest sense, the term total protein does not include them.

The term derivatized protein refers to a protein selected from the group consisting of derivatized insulin, derivatized insulin analogs, and derivatized proinsulin, which are derivatized with a functional group such that: • *

that the derivatized protein is either less soluble in an aqueous solvent than the non-derivatized protein or is more lipophilic than non-derivatized insulin or forms a complex with zinc and protamine that is less soluble than the corresponding complex with the non-derivatized protein. Determining the solubility or lipophilicity of proteins and derivatized proteins is well known to those skilled in the art. The solubility of derivatized proteins and proteins in zinc and protamine complexes can be readily determined by well-known methods [Graham and Pomeroy, in Pharm. Pharmacol. 36:427-430 (1983), as modified in DeFelippis, M. R. and Frank, B., EP 735,048] or the procedures described herein.

Many examples of such derivatized proteins are known in the art, including benzoyl, p-tolylsulfonamide, carbonyl and indolyl insulins and insulin analogs [Havelund, S.

WO95/07931, published by Derivatives et al.,

March 23, 1995];

alkyloxycarbonyl derivatives of insulin [Geiger, R. et al., 1972; September issued August 15 3,907,763, issued 23

U.S. Patent No. 3,684,791,

Brandenberg, D. et al., U.S.

1975]; aryloxycarbonyl derivatives of insulin [Brandenberg, D. et al., U.S. 3,907,763, issued September 23, 1975]; alkylcarbamyl derivatives [Smyth, D. G., U.S. Patent No. 3,864,325, issued February 4, 1975; Lindsay, D. G. et al., U.S. Patent No. 3,950,517, issued April 13, 1976]; carbamyl, O-acetyl derivatives of insulin [Smyth, D. G., U.S. Patent No. 3,864,325 issued February 4, 1975]; crosslinked alkyldicarboxyl derivatives [Brandenberg, D. et al., U.S. Patent No. 3,907,763, issued September 23, 1975]; N-carbamyl, O-acetylated insulin derivatives [Smyth, D. G., U.S.

• · »··· ·» • · Patent No. 3,868,356, issued February 25, 1975]; various 0-alkyl esters [Markussen, J. , U.S. Patent No. 4,343,898, issued August 10, 1982; Morihara, K. et al., U.S. Patent No. 4,400,465, issued August 23, 1983; Morihara, K. et al., U.S. Patent No. 4,401,757, issued August 30, 1983; Markussen, J.,

U.S. Patent No. 4,489,159, issued to Obermeier, R. et al., U.S. Patent No. July 1986; and Andresen, F. H. and 4,601,979, issued July 22, insulin derivatives [Balschmidt, P 5,430,016, issued July 4, insulin derivatives [Lindsay, D. G., U.S. Patent No. 3,869,437, issued March 4, 1975]; and fatty acid acylated proteins which are described in this application.

December 18, 1984; 4,601,852, issued 22nd ed., U.S. Patent No. 1986]; alkylamide et al., U.S. Patent No. 1995]; various others

The term acylated protein, as used herein, refers to a derivatized protein selected from the group consisting of insulin, insulin analogs, and proinsulin that is acylated with an organic acid that is linked to the protein via an amide bond formed between an acidic group of the organic acid and an amino group of the protein. Generally, the amino group can be the α-amino group of an N-terminal amino acid of the protein or can be the ε-amino group of a Lys residue of the protein. The acylated protein can be acylated at one or more of the three amino groups that are present in insulin and most insulin analogs. Monoacylated proteins are acylated at a single amino group. Diacylated proteins are acylated at two amino groups. Triacylated proteins are acylated at three amino groups. The organic acidic compound can be, for example, a fatty acid, « · • ·

φ · · • ♦ φ · · <, · · · aromatic acid or any other organic compound that bears a carboxylic acid functional group that can form an amide bond with an amino group of a protein and that will reduce the aqueous solubility, increase the lipophilicity, or reduce the solubility of zinc-protamine complexes of the derivatized protein compared to the non-derivatized protein.

The term fatty acid acylated protein refers to an acylated protein selected from the group consisting of insulin, insulin analogs and proinsulins, which is acylated by a fatty acid that is linked to the protein via an amide bond formed between an acidic group of the fatty acid and an amino group of the protein. Generally, the amino group may be the α-amino group of the N-terminal amino acid of the protein or it may be the ε-amino group of the Lys residue of the protein. A fatty acid acylated protein may be acylated at one or more of the three amino groups present in insulin and most insulin analogs. Monoacylated proteins are acylated at a single amino group. Diacylated proteins are acylated at two amino groups. Triacylated proteins are acylated at three amino groups. Fatty acid acylated insulin is described in Japanese Patent Application 1-254,699. See also Hashimoto, M. et al., Pharmaceutical Research, 6:171-176 (1989) and Lindsay, D. G. et al., Biochemical J. 121:737-745 (1971). Further disclosures of fatty acid acylated insulins and fatty acid acylated insulin analogs and methods for their synthesis can be found in Baker, J. C. et al., U.S. 08/342,931, filed November 17, 1994 and issued as U.S.

• 0

4 Patent No. 5,693,609, December 2, 1997; Havelund, S. et al., WO95/07931, published March 23, 1995 and corresponding U.S. Patent No. 5,750,497, May 12, 1998; and Jonassen, I. et al., WO96/29342, published September 26, 1996. These disclosures are expressly incorporated herein by reference as describing fatty acid acylated insulins and fatty acid acylated insulin analogs and enabling the synthesis of these substances.

The term fatty acid acylated protein includes pharmaceutically acceptable salts and complexes of fatty acid acylated proteins. The term fatty acid acylated protein also includes preparations of acylated proteins in which the population of acylated protein molecules is homogeneous with respect to the site or sites of acylation. For example, Νεmonoacylated protein, Bl-Na-monoacylated protein, Al-Namonoacylated protein, Al, Bl-Nct-diacylated protein, Νε,ΑΙΝα,diacylated protein, Νε, Bl-Nct,diacylated protein and Νε,Al,Bl-Να,triacylated protein are all included under the term fatty acid acylated protein for the purposes of describing the present invention. The term also refers to preparations of acylated protein molecules having

In the latter case, the term acylated protein includes mixtures of diacylated proteins, mixtures of triacylated proteins, mixtures of triacylated proteins, and mixtures of acylation populations.

in which heterogeneous fatty acid monoacylated and monoacylated and diacylated and monoacylated, diacylated and triacylated proteins.

The term insulin as used herein refers to human insulin, the amino acid sequence and specific structure of which are well known. Human insulin consists of twenty-one « 9 <

·

9 ♦ · · · ··

99 »99 *

V 9 9 4 ► « · ’ » · · *

9.

amino acids of the A-chain and thirty amino acids of the B-chain, which are cross-linked by disulfide bonds. Properly cross-linked insulin contains three disulfide bridges: one between position 7 of the A-chain and position 7 of the B-chain, a second between position 20 of the A-chain and position 19 of the B-chain, and a third between positions 6 and 11 of the A-chain.

The term insulin analog refers to proteins that have an A-chain and a B-chain that have essentially the same amino acid sequences as the A-chain and B-chain of human insulin, but differ from the A-chain and B-chain of human insulin by including one or more amino acid deletions, one or more amino acid substitutions, and/or one or more amino acid additions that do not destroy the insulin activity of the insulin analog.

Animal insulins are analogs of human insulin and are therefore insulin analogs as defined above. Four such animal insulins are rabbit, porcine, bovine and ovine insulin. The amino acid substitutions that distinguish these animal insulins from human insulin are listed below for the convenience of the reader.

• tititi * * • ti • · > « ti ti • ti t · ti ti· ”

Location amino acids A8 A9 A10 B30 human insulin Thr Shit Ile Thr . rabbit insulin Thr Shit Ile Shit porcine insulin Thr Shit Ile Ala bovine insulin Ala Shit Val Ala sheep insulin Ala Gly Val Ala

Another type of insulin analog, a monomeric insulin analog, is well known in the art. Monomeric insulin analogs are structurally similar to human insulin and have activity that is similar or the same as human insulin, but contain one or more amino acid deletions, substitutions, or additions that cause a relaxation of the contact involved in dimerization and hexamerization, resulting in a reduced tendency to form higher aggregated states. Monomeric insulin analogs are fast-acting analogs of human insulin and are described, for example, in Chance, R. E. et al., U.S. Patent No. 5,514,646, May 7, 1996; Brems, D. N. et al. Protein Engineering, 5:527-533 (1992); Brange, J. J. V. et al., EPO Publication No. 214,826, published March 18, 1987; Brange, J. J. V. et al., U.S. Patent No. 5,514,646, May 7, 1996; Patent No. 5,618,913, Apr. 8, 1997; and Brange, J. et al., Current Opinion in Structural Biology 1:934-940 (1991). An example of a monomeric insulin analog is described as human insulin in which the Pro at position B28 is replaced by an Asp, Lys, Leu, Val or Ala residue and where the Lys at position B29 is Lys or is replaced by a Pro residue, and further AlaB26-human insulin, des(B28-B30)-human insulin and des(B27)-human insulin. Monomeric insulin analogs used as derivatives in crystals or used

·· ·· • · · · • · · · · • · * . ·· ·« non-derivatized in the solution phase of suspension formulations are correctly cross-linked at the same positions as human insulin.

Another group of insulin analogs for use in the present invention are analogs in which the isoelectric point of the insulin analog is in the range of about 7.0 to about 8.0. These analogs are referred to as p1-shifted insulin analogs. Examples of such insulin analogs include ArgB31,ArgB32-human insulin, GlyA21,ArgB31,ArgB32-human insulin, ArgAO,ArgB31,ArgB32-human insulin, and ArgAO,GlyA21,ArgB31,ArgB32-human insulin.

Another group of insulin analogs are insulin analogs that have one or more amino acid deletions that do not significantly impair the activity of the molecule. This group of insulin analogs is hereinafter referred to as deletion analogs. For example, insulin analogs with a deletion of one or more amino acids at positions B1-B3 are active. Similarly, insulin analogs with a deletion of one or more amino acids at positions B28-B30 are active. Examples of deletion analogs include des(B30)-human insulin, desPhe(B1)-human insulin, des(B27)-human insulin, des(B28B30)-human insulin, and des(B1-B3)-human insulin. Deletion analogs used as a derivative in the present crystals or used undiveritasified in the solution phase of suspension preparations are properly cross-linked at the same positions as human insulin.

Amidated amino acids, and particularly asparagine residues in insulin, are known to be chemically unstable [Jorgensen, K. H. et al. U.S. Patent No. 5,008,241, issued April 16, 1991;

···· *♦ • * « • 9 <

· • »9 ¹ • 9 99 • · · • 9 · • 9 · · • 9 ·♦ e ' · ♦ • 999 · · · » · · 9 «9 · f

Dorschug, M., U.S. Patent No. 5,656,722, issued August 12, 1997] . They are particularly susceptible to deamidation and various rearrangement reactions under certain conditions which are well known. For this reason, the insulin analog may optionally be insulin or an insulin analog which has one or more amidated residues replaced by other amino acids in order to achieve chemical stability. For example, Asn or Gln may be replaced by non-amidated amino acids. Preferred amino acid replacements for Asn or Gln are Gly, Ser, Thr, Asp or Glu. It is preferred to replace one or more Asn residues.

In particular, AsnA18, AsnA21 or AsnB3 or any combination of these residues may be replaced, for example, by a Gly, Asp, or Glu residue. Also, GlnA15 or GlnB4 or both may be replaced by either Asp or Glu. Preferred substitutions are Asp at B21 and Asp at B3. Also preferred are substitutions that do not change the charge of the protein molecule, so that replacement of Asn or Gln by neutral amino acids is also preferred.

The term proinsulin refers to a single-chain peptide molecule that is a precursor of insulin. Proinsulin can be converted to insulin or an insulin analog by chemical or preferably enzyme-catalyzed reactions. The proper disulfide bonds are formed in proinsulin as described above. Proinsulin comprises insulin or an insulin analog and a linker or linker peptide. The linker peptide has between 1 and about 35 amino acids. The linker or linker peptide is connected to the terminal amino acid of the A-chain and to the terminal amino acid of the B-chain by an α-amide bond or two α-amide bonds. Preferably, there is no

4 • 4 ··

• 4 4 4 4 • t • · • ♦ • · of the amino acids in the linker peptide is cysteine. Preferably, the C-terminal amino acid of the linker peptide is Lys or Arg. The proinsulin may have the general formula X-B-C-A-Y or may have the general formula X-A-C-B-Y, wherein X is hydrogen or is a peptide of from 1 to about 100 amino acids having Lys or Arg as the C-terminal amino acid, Y is hydroxy or is a peptide of from 1 to about 100 amino acids having either Lys or Arg as the N-terminal amino acid, A is the A-chain of insulin or the A-chain of an insulin analog, C is a peptide of from 1 to about 35 amino acids, none of which are cysteine, wherein the C-terminal amino acid is Lys or Arg, and B is the B-chain of insulin or the B-chain of an insulin analog.

Pharmaceutically acceptable salt means a salt formed between any one or more charged groups in a protein and any one or more pharmaceutically acceptable non-toxic cations or anions. Organic and inorganic salts include, for example, salts prepared from acids such as hydrochloric, sulfuric, sulfonic, tartaric, fumaric, hydrobromic, glycolic, citric, maleic, phosphoric, succinic, acetic, nitric, benzoic, ascorbic, p-toluenesulfonic, benzenesulfonic, naphthalenesulfonic, propionic, carbonic and the like, or, for example, ammonium, sodium, potassium, calcium or magnesium salts.

The verb acylate means to form an amide bond between a fatty acid and the amino group of a protein. A protein is acylated when one or more of its amino groups are combined by an amide bond with the acidic group of a fatty acid.

• 9 • 9

The term fatty acid means a saturated or unsaturated straight-chain or branched-chain fatty acid containing from one to eighteen carbon atoms.

The term C1 to C18 fatty acid refers to a saturated or unsaturated straight chain or branched chain fatty acid containing from one to eighteen carbon atoms.

The term divalent metal cation refers to an ion or ions that participate in the formation of a complex with a plurality of protein molecules. Transition metals, alkali metals, and alkaline earth metals are examples of metals that are known to form complexes with insulin. Transition metals are preferred. Zinc is particularly preferred. Other transition metals that may be pharmaceutically acceptable for forming complexes with insulin proteins include copper, cobalt, and iron.

The term complex has two meanings in the present invention. In the first, the term refers to a complex formed between one or more atoms in the proteins that form the complex and one or more divalent metal cations. The atoms in the proteins serve as electron-donating ligands. Proteins typically form hexameric complexes with divalent transition metal cations. The second meaning of the term complex in the present invention is the association between a complexing compound and the hexamers. A complexing compound is an organic molecule that typically has a higher number of positive charges that binds or complexes with

• · ·· ♦ * ♦ · • · ·· • · * 1 • · ♦ · ·* «♦ hexamers in an insoluble composition, thereby stabilizing them against dissolution. Examples of complexing compounds suitable according to the present invention include protamine, surfen, various globin proteins [Brange, J., Galenics of Insulin, Springer-Verlag, Berlin Heidelberg (1987)] and various polycationic polymeric compounds known to complex with insulin.

The term protamine refers to a mixture of strongly basic proteins obtained from fish sperm. The average molecular weight of the proteins in protamine is approximately 4,200 [Hoffmann, J. A. et al., Protein Expression and Purification, 1:127-133 (1990)]. Protamine may refer to relatively salt-free preparations of proteins, often called protamine bases. Protamine also refers to preparations containing protein salts. Commercial preparations vary greatly in their salt content.

Protamines are well known to those skilled in the art of insulin and are commonly incorporated into NPH insulin products. Pure protamine fractions are useful in the present invention, as are mixtures of protamines. Commercial protamine preparations, however, are not typically homogeneous with respect to the proteins administered. They are nevertheless useful in the present invention. Protamine consisting of protamine base is useful in the present invention, as are protamine preparations containing protamine salts and preparations which are mixtures of protamine base and protamine salts. Protamine sulfate is a commonly used protamine salt. All weight ratios relating to protamine are determined relative to protamine free

9··· ·» ·· ·» ·« ·· « ·«·» • · · · · · · »»·« • « · « · » · * * * · · ·««« »·»« · · · * 99 9 9 9 9 9 9 9 9· 99 base. One skilled in the art can determine the amount of other protamine preparations that have a specific weight ratio relative to protamine.

The term suspension refers to a mixture of a liquid phase and a solid phase that contains insoluble or sparingly soluble particles that are larger than colloidal size. Mixtures of NPH microcrystals and an aqueous solvent form suspensions. Mixtures of an amorphous precipitate and an aqueous solvent also form suspensions. The term suspension preparation means a pharmaceutical composition in which the active agent is present in a solid phase, for example a microcrystalline solid, an amorphous precipitate, or both, which is finely dispersed in an aqueous solvent. Finely dispersed is such that the highly dispersed solid can be suspended in a uniform manner in the mixture, a uniform suspension, from the aqueous solvent by the action of which results in the formation of a reasonable volume from which a therapeutic dose can be withdrawn. Examples of commercially available insulin suspension preparations include, for example, NPH, PZI, and ultralente. A small proportion of the solid in a microcrystalline suspension preparation may be amorphous. Preferably, the proportion of amorphous material is less than 10% and most preferably less than 1% of the solids in the microcrystalline suspension. Similarly, a small amount of the material in the amorphous precipitate suspension may be microcrystalline.

NPH insulin refers to the Neutral Protamine Hagedorn insulin preparation. Synonyms include human NPH insulin and NPH insulin, among many others. Humulin N is a commercial preparation of NPH insulin. A related term is NPL,

4· · 4 9 9 4 4 4 · ·· ·· » · · 9 9 9 * · ·« ·

Λ 9 9 9 9 99 9 9 9 9

9 9 9 9 9 9 9 9 9 9 · ·· 99 9· 9 9 99 4 4 which refers to an NPH-like preparation of LysB28,ProB29analog of human insulin. The meaning of these terms and the methods of said insulin preparations are well known to those skilled in the art of insulin preparations.

The term aqueous solvent refers to a liquid solvent that contains water. An aqueous solvent system may be comprised entirely of water, may contain water and one or more water-miscible solvents, and may contain solutes. The most commonly used miscible solvents are short-chain organic alcohols such as methanol, ethanol, propanol, short-chain ketones such as acetone, and polyols such as glycerol.

An isotonic agent is a compound that is physiologically tolerated and provides a suitable tonicity to a preparation to prevent the flow of water across cell membranes that are in contact with the administered preparation. Glycerol, also known as glycerin, is commonly used as an isotonic agent. Other isotonic agents include salts, such as sodium chloride, and monosaccharides, such as dextrose and lactose.

The insoluble compositions of the present invention comprise a hexamer-stabilizing compound. The term hexamer-stabilizing compound refers to low molecular weight non-proteinaceous compounds that stabilize a protein or derivatized protein in a hexameric aggregate state. Phenolic compounds, particularly phenolic preservatives, are the best known stabilizing compounds for insulin and insulin derivatives. Hexamer···· ·· • 9

99 • 9 9

99

99

9 9 · • · 99

9 » ·

9 9 9

9 ·· stabilizing compounds stabilize the hexamer by binding to it through specific intermolecular contacts. Examples of hexamer-stabilizing agents include: various phenolic compounds, phenolic preservatives, resorcinol, 4'-hydroxyacetanilide, 4-hydroxybenzamide and 2,7-dihydroxynaphthalene. The multi-purpose insoluble compositions of the present invention contain a preservative in addition to the hexamer-stabilizing compound.

The phenolic preservative used in the preparations of the present invention may be the same or different from the hexamer preservative and may be a stabilizing agent.

The term preservative refers to a compound added to a pharmaceutical preparation to act as an antimicrobial agent. A parenteral preparation must meet the requirements for preservative efficacy in order to be useful as a commercially viable product for various uses. Among the preservatives known in the art to be effective and acceptable for parenteral preparations are benzalkonium chloride, benzethonium, chlorhexidine, phenol, m-cresol, benzyl alcohol, methylparaben, chlorobutanol, o-cresol, p-cresol, chlorocresol, phenylmercuric nitrate, thimerosal, benzoic acid, and various mixtures thereof. See, for example, Wallhäusser, K.-H., Develop. Biol. Standard, 24:9-28 (1974) (S. Krager, Basel).

The term phenolic preservative includes the compounds phenol, m-cresol, o-cresol, p-cresol, chlorocresol, methylparaben and mixtures thereof. Certain phenolic preservatives such as phenol and m-cresol are known to be

• ♦ • 4 • 4 4« • «4 4 • 4 ·4 • 4 4 ·

4 4 4

4· bind to insulin-like molecules and thereby induce conformational changes that increase either physical or chemical stability or both [Birnbaum, D. T. et al., Pharmaceutical Res. 14:25-36 (1997); Rahuel-Clermont, S. et al., Biochemistry 36:5837-5845 (1997)].

The term buffer or pharmaceutically acceptable buffer refers to compounds that are known to be safe for use in insulin formulations and that act to control the pH of the formulation and maintain it at the pH desired for the formulation. The pH of the formulation of the present invention is from about 6.0 to about 8.0. Preferably, the formulations of the present invention have a pH in the range of about 6.8 to about 7.8. Pharmaceutically acceptable buffers for maintaining the pH at values from slightly acidic pH to slightly basic pH include compounds such as phosphates, acetates, citrates, arginine, TRIS and histidine. TRIS refers to 2-amino-2-hydroxymethyl-1,3-propanediol and any pharmacologically acceptable salt thereof. The free base and hydrochloride forms are two common forms of TRIS. TRIS is also known in the art as trimethylolaminomethane, tromethamine and tris(hydroxymethyl)aminomethane. Other buffers which are pharmaceutically acceptable and which are suitable for controlling pH at a desired level are known to the ordinary chemist.

The term "administer" means to introduce a composition of the present invention into the body of a patient in need thereof for the treatment of a disease or condition.

The term treatment refers to the care of a patient who has diabetes or hyperglycemia or another condition in which

AND

9 9 9 • 9 9 «

9 99

9 9 ·

9 9 9

9· 9·

99

9 9 9

9 9 9

9 9 9

9 9 9

99 indicated the administration of insulin for the purpose of managing or alleviating symptoms and complications associated with the condition. Treatment includes the administration of a composition of the present invention to prevent the onset of symptoms or complications, alleviate symptoms or complications, or eliminate the disease, condition, or disorder.

As noted above, the present invention relates to insoluble compositions that have properties similar in some respects to NPH insulin and superior in other respects to NPH insulin. They are similar to NPH insulin in that they have similar physical properties as set forth below. A light microscope equipped with an oil immersion objective and a cross polarizer was used to examine the microcrystals comprising B29-Ns-octanoyl-human insulin, insulin, zinc, protamine and phenol prepared according to the present invention. Examination at 1000x magnification showed that these microcrystals were single and rod-shaped and exhibited a uniform crystal morphology. The sizes of these microcrystals generally ranged from about 2 microns to 8 microns in length. Direct comparison using this microscope showed that the morphology of these microcrystals appeared to be similar to that of commercially produced porcine NPH microcrystals, which have been described elsewhere as rod-shaped. The size range of these microcrystals was also similar to that of commercially produced NPH microcrystals, which generally have a mean length of approximately 5 microns. The commercial manufacturing specification for the mean length of NPH microcrystals is from 1 micron to 40 microns.

However, the microcrystals of the present invention are ···· ·· • φ ♦ · · φ · · · • · φ φ· ·· • Φ φφ ♦ · · · • · φ • · · · • φ φ φφ φφ φφ φφ • φ · · φ φ φ * • φ φ φ · φ φ φ φ φ φφ φφ unexpectedly and unpredictably different from NPH insulin crystals in their dissolution properties and in their duration of action. In particular, the microcrystals of the present invention dissolve much more slowly under conditions simulating physiological conditions than do NPH insulin crystals and provide a longer and flatter blood glucose control profile than does NPH insulin. This was demonstrated by the following experiments.

It has been found that certain derivatized proteins, in soluble form, have durations of action that are not significantly different from those of regular human insulin. Three groups of animals were used. Each animal in the first group received a dose (0.75 nmol/kg) of Humulin (soluble human insulin), each animal in the second group received a dose (0.75 nmol/kg) of soluble B29Νε-octanoyl-human insulin (C8-hI), and each animal in the third group received a dose (0.75 nmol/kg) of soluble B29Νε-decanoyl-human insulin (C10-hI). The experiments were carried out essentially as described in Example 5, with five dogs per group. The proteins were administered subcutaneously. Blood glucose concentrations were determined and are shown in the table below.

Table 1. Blood glucose concentrations before and after administration of Humulin, soluble B29-Ns-octanoyl-human insulin (C8-hI), or soluble B29-Ns-decanoyl-human insulin (C10-hI) in normal dogs simultaneously with administration of somatostatin to create a transient diabetic state. Mean values ± standard error are given.

···· ** • ·

Blood Glucose Concentration (mg/dl) Time (h) Humulin® Soluble C8-hI Soluble C10- hl -0.5 110 ± 2 115 ± 4 108 ± 2 0 101 ± 2 101 ± 7 96 ± 4 0.5 83 ± 5 80 ± 5 85 ± 6 1 54 ± 6 52 ± 4 70 ± 5 1.5 4 9 + 4 51 ± 2 57 ± 4 2 48 ± 4 51 ± 2 52 ± 3 2.5 55 ± 4 60 ± 3 56 ± 4 3 59 ± 2 65 ± 4 58 ± 4 3, 5 65 ± 2 73 ± 5 63 ± 4 4 71 ± 2 85 ± 6 68 ± 4 5 87 ± 2 110 ± 8 79 ± 3 6 104 ± 3 124 ± 4 91 ± 7 7 119 ± 8 145 ± 14 106 ± 8 8 144 ±+ 5 153 ± 16 119 ± 11

These data clearly demonstrate that soluble B29-Ns-octanoyl-human insulin and B29-N8-decanoyl-human insulin, administered subcutaneously to normal dogs in a transient diabetic state, produce glucose reductions roughly comparable to those achieved with soluble human insulin. In particular, soluble B29-N8-octanoyl-human insulin exhibits a more rapid onset and shorter duration of action than human insulin.

In the second experiment, it was found that the dissolution rate of cocrystals of insulin and B29-N8-octanoyl-human insulin,

fcfcfcfc ·· • · 4 • · <

• · <

• ·· ¹ • fc ·· prepared according to the present invention is noticeably longer than that of commercially produced NPH-porcine insulin. This was most surprising in light of the above data. The dissolution rate can be measured using well-known procedures [Graham and Pomeroy, in Pharm. Pharmacol. 36:427-430 (1983), as modified in De Felippis, MR and Frank, Β., EP 735,048] or using the procedures described herein.

The dissolution rate of porcine NPH insulin microcrystals was determined by placing 5 microliters of U100 porcine NPH insulin in 3 mL of Dulbecco's phosphate-buffered saline (without calcium or magnesium) in a 1 cm square silicon cuvette at 22°C and stirring at a constant speed using a stirrer. Absorbance measurements at 320 nm were made at one-minute intervals. Absorbance at 320 nm corresponds to light scattered by insoluble particles present in the aqueous suspension. Consequently, as the microcrystals dissolve, the absorbance approaches zero. The porcine NPH insulin microcrystals were completely dissolved after approximately 20 minutes.

This magnetic solution

It was found that the dissolution rate of protamine-zinc human insulin crystals that did not contain co-crystallized acylated human insulin was also approximately 20 minutes, as determined by the above procedure. These NPH human insulin crystals were prepared essentially in the manner of Preparation No. 1 below, except that no acylated protein was used and 7 parts of human insulin were used. The data obtained from this experiment are shown in Fig. 1 as a dashed line.

·♦ ·· • ·

• · · · · • · · · · · ·

Cocrystals of the present invention, consisting of human insulin and B29-Ns-octanoyl-LysB29 human insulin, were prepared as described in Preparations 2 _of 4 and 5 below. In these preparations, human insulin and acylated insulin were used with precrystallization weight ratios of 3:1, 1:1 and 1:3.

The method described above was used to measure the dissolution rate of these cocrystals. Briefly, a volume of 12 microliters of each cocrystal suspension containing protamine, zinc, B29-NE-octanoyl-LysB29 human insulin, and human insulin (containing not more than 50 U/ml) was placed in 3 ml of Dulbecco's phosphate-buffered saline (free of calcium or magnesium) in a 1 cm square silicon cuvette. This solution was stirred at the same constant speed at the same temperature of 22°C. The data obtained from this experiment are shown in Fig. 1 and show that the 3:1 cocrystals required more than 100 minutes to dissolve, that the 1:1 cocrystals required more than 150 minutes to dissolve, and that the 1:3 cocrystals did not completely dissolve even after 400 minutes.

The time required for the absorbance during dissolution to reach half the value between the initial absorbance and the final absorbance is defined as the ti/ ₂ value. The ti/ ₂ values for these preparations are given below in Table 2.

Table 2. Dissolution ti/ ₂ values for Iletin NPH, human insulin NPH, and cocrystals of B29-octanoyl-human insulin and human insulin ♦ <

• · • »

9999

9 ·· *

• · · • · · · ·

Substance %B29- octanoyl- human insulin % human insulin % pork insulin values tl/2 (minutes) Iletin NPH 0 0 100 6 Human NPH 0 100 0 7 3:1 cocrystal 25 75 0 38 1:1 cocrystal 50 50 0 117 1:3 cocrystal 75 25 0 >400

These experiments show that in Dulbecco's phosphate-buffered saline (free of calcium and magnesium), a solution that in some respects mimics interstitial fluid, the dissolution rates of the 3:1, 1:1, and 1:3 co-crystals are significantly slower than those of porcine NPH microcrystals. These experiments also demonstrate that the dissolution rates of microcrystals containing protamine, zinc, and human insulin are very similar to those of porcine NPH insulin (Iletin NPH). These results further indicate that the dissolution rate of the protamine, zinc, B29-Ne-octanoyl-LysB29-human insulin and human insulin cocrystals depends on the ratio of human insulin to B29-Ne-octanoyl-LysB29-human insulin present in the cocrystals and in particular that the dissolution rate of the protamine, zinc, B29-Ne-octanoylLysB29 human insulin and human insulin cocrystals decreases as the proportion of human insulin decreases.

Another dissolution method, based on HPLC determination of dissolved protein and derivatized protein, was used to compare the dissolution rate of the 1:3 microcrystals «4 *· rt · « « <· described above with commercial preparations of human NPH insulin, human insulin ultralente and bovine ultralente and also to study the effects on the dissolution rate of various changes in the ratio of protein to derivatized protein.

expressed as a percentage of time, with 100% being unfiltered samples, unfiltered were linearly

A volume (0.5 ml) of U100 preparation was suspended in 200 ml of Dulbecco's phosphate-buffered saline adjusted to pH 7.4 in a water-jacketed dissolution apparatus maintained at 25 °C. The dissolution buffer also contained 1 mg/ml human serum albumin to prevent adsorptive losses of the dissolved insulins. The dissolution medium was stirred at a constant speed of 180 rpm. At regular intervals, 3.5 ml of this solution was withdrawn and filtered through a 0.22 μm filter that bound low molecular weight proteins. The first 0.5 ml of the filtrate was discarded and the next 1.5 ml of the filtrate was acidified with 4 ml of 5 N HCl and subjected to HPLC analysis. The concentrations of dissolved insulins were determined based on HPLC peak areas and the results were solutes as a function of the total area of insulins in If the total area of the unfiltered samples decreased slightly with time, the corrected value was used as 100% to calculate the percentage dissolved at each time point. The results of these studies are shown in Fig. 3 and 4.

The data in Fig. 3 show that the greater the fraction of derivatized protein incorporated into the microcrystals, the slower the dissolution rate. The data in Fig. 4 show that the 1:3 microcrystal of human insulin and B29-octanoyl-human insulin dissolve significantly more slowly than both human NPH insulin and human ultralente insulin. Most significantly, the 1:3 microcrystal has a dissolution rate very similar to that of bovine ultralente. Bovine ultralente has long been considered a near-ideal long-acting insulin preparation, both because of its very prolonged biological activity and also because of the flatness of the pharmacodynamic response after administration. It is therefore predicted that these microcrystals may approach or surpass bovine ultralente in their ability to provide basal glucose control over very long time intervals.

Since the duration of action profile of NPH preparations is very strongly related to the rate of dissolution of the microcrystals in the subcutaneous interstitial fluid, it is inferred from the above experiments that the microcrystalline compositions of the present invention exhibit a more prolonged duration of action when administered subcutaneously to diabetic patients than do existing commercial NHP insulin preparations. Importantly, these results also demonstrate that the present invention allows for the control and even optimization of the duration of action in patients by manipulating the ratio of protein to derivatized protein.

This ability to manipulate the duration of action by simple changes in the ratios of protein and derivatized protein is most significant when viewed in light of the history of problems with insulin preparations. Historically, a major obstacle to the development of insulin preparations for basal glucose control has been that their time of action has been invariantly tied to their own

0000 0*

0' · · · · 0 0 «0 molecular properties of the protein, such as albumin binding affinity, isoelectric point or solubility. As a result, only a single duration of action was possible for each molecule or preparation, and the only possible approach to improving pharmacokinetics was to further modify the molecule.

A long-sought goal has been the development of controlled-release systems whose pharmacokinetics can be precisely and easily adjusted by manipulation of the formulation matrix. Much academic and industrial research in this area has focused on insulin over the past 15 years, but the goal of controlled release has proven particularly difficult to achieve due to the extremely narrow therapeutic index of insulin, the requirement for its chronic use, as well as economic considerations that work against sophisticated and expensive manufacturing processes and formulations.

the pharmacological efficacy of the present invention is

The main advantage of the present invention is that it provides a controlled release system in which the pharmacokinetics of insulin release can be more easily controlled than in other proposed controlled release technology systems. Insoluble compositions composed of insulin protein together with derivatized protein provide an advantageous means of adjusting the dissolution rate and thus allowing for controlled release. While not intended to be simplistic or limiting, it is contemplated that the insoluble compositions of the present invention are based on the slow release of a consistent ratio of protein to derivatized protein from the composition. It is further contemplated, without being limited thereby, that • ·

0 0 · · 0 0 0 0

00 the invention is not limited in any way, that a significant essential feature of the present invention is the complete or nearly complete homogeneity of the insoluble compositions.

For microcrystals, it is assumed that each individual microcrystal in suspension consists of approximately the same ratio of protein to derivatized protein. This ratio closely mirrors the ratio of combined proteins in solution to derivatized protein prior to crystallization. Also, as the microcrystals dissolve, a determined portion of the protein and derivatized protein are released throughout the dissolution period. The significance of this type of behavior is significant, as it results in a constant but reduced rate of release of the two active molecules from the injection site into the bloodstream. To achieve this goal, special attention must be paid to the method of preparation of the compositions.

The present discovery that it is possible to crystallize a protein with a derivatized protein and thereby obtain homogeneous cocrystals was a surprising finding, considering the complexity of the parameters affecting protein crystallization: specific and nonspecific intermolecular interactions such as hydrogen bonding, electrostatic interactions, hydrophobic interactions, van der Waals forces, the excluded volume effect, solubility, and steric bulk. Although the basic understanding of the crystallization process of small molecules is relatively advanced, for complex macromolecules it is still a qualitative science, and current practice consists of empirically applying a large set of crystallization recipes.

• · · ) titi · • ti *· • ti ) « » « • ti • · ti · • · ti · • ti ti · • · titi · • titi · • · titi

The hydrophobic effect has been proposed as a major mechanism for protein association [Chothia, C. et al., Nature 256:705-708 (1975)]. In addition, a strong correlation between surface hydrophobicity and protein-protein contact has been observed [Young, L. et al., Protein Science 3:717-729 (1994)]. Furthermore, given the complexity of protein crystallization, the discovery that derivatized insulins can cocrystallize with an underivatized protein in a manner isomorphic or nearly isomorphic to normal insulin is surprising.

Co-crystallization of two related therapeutic proteins with the goal of modifying therapeutic behavior is a novel concept. This is largely because the vast majority of work in the field of protein crystallization is motivated by the goal of achieving single, large, and homogeneous crystals suitable for X-ray diffraction structural analysis. Although protein co-crystals are known, such co-crystal systems are characterized as host-substrate complexes, such as protein and its receptor, protein-DNA complexes, and protein-receptor complexes. These host-substrate co-crystal complexes are fundamentally different from the co-crystals of the present invention, as no complex is formed between the protein and the derivatized protein.

The predictability of protein properties and homogeneity of the present invention were demonstrated by measuring (using HPLC) protein concentration and dissolution studies using protein and insoluble derivatized compositions according to two microcrystal preparations,

• 4

44 •··· 94

4 4 which were prepared from known amounts of human insulin (protein) and B29-octanoyl-human insulin (derivatized protein). In the first preparation, the weight ratio of protein added to derivatized protein was 1:3. In the second preparation, this ratio was 55:45. For the 1:3 microcrystals, B29-octanoyl-human insulin represented between about 73% and 76% of the total protein in 14 measurements over a period of about 10 hours during dissolution under the conditions described for the dissolution test above. No trend was observed in the data. For the 55:45 microcrystals, B29-octanoyl-human insulin represented between about 43% and 47% of the total protein for 9 measurements made over a period of about 3.75 hours during dissolution under the conditions described for the dissolution test above. Again, no trend was observed in the data.

The insoluble compositions of the present invention may form crystals with rod-shaped or irregular morphology or may be represented by amorphous precipitates. Preferred insoluble compositions are comprised of acylated insulin or acylated insulin analog, zinc ions present in an amount of about 0.3 to about 0.7 moles per mole of total protein, and a phenolic preservative selected from the group consisting of phenol, m-cresol, okresol, p-cresol, chlorocresol, methylparaben, and mixtures thereof, and present in an amount sufficient relative to the total protein to stabilize the T3R3 or R6 hexameric conformation, and protamine present in an amount of about 0.15 to about 0.7 moles per mole of total protein.

Preferred group of insulin analogues for preparation • · • · • · ♦ ♦ ·· * · · 4 • · · * • · · · · • · ·

9· «►· derivatized insulin analogs used to form insoluble compositions consist of animal insulins, deletion analogs, and p1-shifted analogs. A more preferred group consists of animal insulins and deletion analogs. Deletion analogs are even more preferred.

Another preferred group of insulin analogs for use in the microcrystals of the present invention comprises a monomeric monomer residue at amino acid position B28.

Particularly preferred are those in which the amino acid is Val or Ala, is Lys or is His or

For,

Asp, is Asp, Lys, Leu, the residue at position B29, the residue at position BIO, the amino acid residue at position B1 is Phe, Asp or is deleted alone or in combination with deletion of the residue at position B2, the amino acid residue at position B30 is Thr, Ala, Ser or is deleted, and the amino acid residue at position B9 is Ser or Asp; provided that either position B28 or B29 is Lys.

Another preferred group of insulin analogs for use in the present invention are those in which the isoelectric point of the insulin analog is between about 7.0 and about 8.0. These analogs are referred to as p1-shifted insulin analogs. Examples of p1-shifted insulin analogs include, for example, ArgB31,ArgB32-human insulin, GlyA21,ArgB31,ArgB32-human insulin, ArgAO,ArgB31,ArgB32-human insulin, and ArgAO,GlyA21,ArgB31,ArgB32-human insulin.

Another preferred group of insulin analogs includes LysB28,ProB29-human insulin (B28 is Lys; B29 is Pro); AspB28-human insulin (B28 is Asp), AspB1-human insulin, ·· ·· • «

ArgB31,ArgB32-human insulin, ArgAO-human insulin, AspB1,GluB13-human insulin, AlaB26-human insulin, GlyA21-human insulin, des(ThrB3O)-human insulin and GlyA21,ArgB31,ArgB32-human insulin.

Particularly preferred insulin analogs include LysB28,ProB29-human insulin, des(ThrB30)-human insulin, AspB28-human insulin, and AlaB26-human insulin. Another particularly preferred insulin analog is GlyA21,ArgB31,ArgB32-human insulin [Dorschug, M., U. S. Patent No. 5,656,722, August 12, 1997]. The most preferred insulin analog is LysB28,ProB29-human insulin.

Preferred derivatized proteins are acylated proteins and preferred acylated proteins for the microcrystals and formulations of the present invention are fatty acid acylated insulins and fatty acid acylated insulin analogs. Fatty acid acylated human insulin is highly preferred. Fatty acid acylated insulin analogs are equally highly preferred.

The particular group used to derivatize insulin, insulin analog, or proinsulin (collectively referred to as a protein) can be any chemical group that does not significantly reduce the biological activity of the protein, is non-toxic unless bound to the protein, and most significantly reduces the aqueous solubility, increases lipophilicity, or reduces the solubility of zinc-protamine complexes of the derivatized protein.

One preferred group of acylation groups are fatty

4· 44

4 4 • 44

4 · • 4 ·

44 · 4 ♦

4 4 · • ·· ♦

4 4 <

4· ·· acids that have a straight chain and are saturated. This

group include methane acid (Cl) , ethane acid (C2) , propane acid (C3) , n-butane acid (C4) , n-pentane acid (C5), n-hexane acid (C6) , n-heptane acid (C7) , n-octane acid (C8) , n-nonane acid (C9) , n-decane acid (CIO), n -undecane acid (Cil), n- -dodecanoic acid (C12), n-tridecanoic acid (C13), n-

tetradecanoic acid (C14), n-pentadecanoic acid (C15), n-hexadecanoic acid (C16), n-heptadecanoic acid (C17) and n-octadecanoic acid (C18). Corresponding to this are formyl (Cl), acetyl (C2), propionyl (C3), butyryl (C4), pentanoyl (C5), hexanoyl (C6), heptanoyl (C7), octanoyl (C8), nonanoyl (C9), decanoyl (CIO), undecanoyl (Cil), dodecanoyl (C12), tridecanoyl (C13), tetradecanoyl (C14) or myristoyl, pentadecanoyl (C15), hexadecanoyl (C16) or palmitoyl, heptadecanoyl (C17) and octadecanoyl (C18) or stearoyl.

The preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having an even number of carbon atoms - that is, C2, C4, C6, C8, C10, C12, C14, C16 and C18 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having an odd number of carbon atoms - that is, C1, C3, C5, C7, C9, C11, C13, C15 and C17 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having more than 5 carbon atoms - that is, C6, C7, C8, C9, C10, C11, C12, C13, C14, C15, C16, C17 and C18 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having less than 9 carbon atoms - that is, C1, C2, C3, C4, C5, C6, C7 and C8 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having between 6 and 8 carbon atoms - that is, C6, C7 and C8 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having more than between 4 and 6 carbon atoms - that is, C4, C5 and C6 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having more than between 2 and 4 carbon atoms - that is, C2,

C3 and C4 saturated fatty acids.

·«·· ··

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having less than 6 carbon atoms - that is, C1, C2, C3, C4 and C5 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having less than 4 carbon atoms - that is, C1, C2 and C3 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having more than 9 carbon atoms - that is, C10, C11, C12, C13, C14, C15, C16, C17 and C18 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having an even number of carbon atoms and more than 9 carbon atoms - that is, C10, C12, C14, C16 and C18 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having 12, 14 or 16 carbon atoms, i.e. C12, C14 and C16 saturated fatty acids.

···· ·» • · φ φ φ φ • φ φ φ φ · ·· φ· »· Φ· • φ φ · φ φ ·· • φ φ · φ φ φ · φφ ·· φφ ·· • φ φ · φ φ · ♦ φ φ · φ φ · · • Φ φφ

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having 14 or 16 carbon atoms, i.e. C14 and C16 saturated fatty acids. A fatty acid with 14 carbon atoms is particularly preferred. A fatty acid with 16 carbon atoms is also particularly preferred.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of saturated fatty acids having between 4 and 10 carbon atoms, i.e. C4, C5, C6, C7, C8, C9 and C10 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of saturated fatty acids having an even number of carbon atoms between 4 and 10 carbon atoms, i.e. C4, C6, C8 and C10 saturated fatty acids.

Another preferred group of fatty acids for forming fatty acid acylated proteins used in the microcrystals of the present invention consists of fatty acids having between 6, 8 or 10 carbon atoms. A fatty acid with 6 carbon atoms is particularly preferred. A fatty acid with 8 carbon atoms is also particularly preferred. A fatty acid with 10 carbon atoms is particularly preferred.

It is clear to the skilled person that narrower preferred groups are formed by combining preferred fatty acid groups • · • 44 • •44 44 • 4 ·

4 4

4 4 ·4

4 4 described above.

Another preferred group of acylation groups consists of saturated fatty acids that are branched. A branched fatty acid has at least two branches. The branch length of a branched fatty acid can be described by the number of carbon atoms in the branch, starting from the carboxyl carbon. For example, the branched fatty acid 3-ethyl-5-methylhexanoic acid has three branches whose lengths are five, six, and six carbon atoms. In this case, the longest branch contains six carbon atoms. As another example, 2,3,4,5-tetraethyloctanoic acid has five branches whose lengths are 4, 5, 6, 7, and 8 carbon atoms. The longest branch contains eight carbon atoms. A preferred group of branched fatty acids are acids that have from three to ten carbon atoms in the longest branch.

A representative list of such saturated fatty acids will be given to avoid any misunderstanding of the range of fatty acids that can be used as acylating groups for proteins according to the present invention: 2-methyl-propionic acid, 2-methyl-butyric acid, 3-methyl-butyric acid, 2,2-dimethyl-propionic acid, 2-methyl-pentanoic acid, 3-methyl-pentanoic acid, 4-methyl-pentanoic acid, 2,2-dimethyl-butyric acid, 2,3-dimethyl-butyric acid, 3,3-dimethyl-butyric acid, 2-ethyl-butyric acid, 2-methyl-hexanoic acid, 5-methyl-hexanoic acid, 2,2-dimethyl-pentanoic acid, 2,4-dimethyl-pentanoic acid, 2-ethyl-3-methyl-butyric acid, 2-ethyl-pentanoic acid, 3-ethylpentanoic acid, 2,2-dimethyl-3-methyl-butyric φφ • * · φ φ · φ · · φ · · • φφφ « • φ

acid,2-methylheptanoic acid, 3-methylheptanoic acid, 4-methylheptanoic acid, 5-methylheptanoic acid, 6-methylheptanoic acid, 2,2-dimethylhexanoic acid, 2,3-dimethylhexanoic acid, 2,4-dimethylhexanoic acid, 2,5-dimethylhexanoic acid, 3,3dimethylhexanoic acid, 3,4-dimethylhexanoic acid, 3,5-dimethylhexanoic acid, 4,4-dimethylhexanoic acid, 2-ethylhexanoic acid, 3-ethylhexanoic acid, acid, acid, butyric acid, acid,

4-ethylhexanoic acid, 2-(1-propyl)pentanoic acid 2-(2-propyl)pentanoic acid, 2,2-diethyl acid, 2,3,4-trimethylpentanoic acid, 24-ethylhexanoic acid 2-ethylhexanoic acid

2-Propyl-pentanoic 3-ethyl-hexane methyl-octanoic acid, 4-methyl-octanoic acid, 7-methyl-octanoic acid, 2,2-dimethyl-heptanoic acid, 2,β-dimethyl-heptanoic acid, 2-ethyl-2-methyl-hexanoic acid, 3-ethyl-5-methyl-hexanoic acid, 3-(1-propyl)hexanoic acid, 2-(2-butyl)-pentanoic acid, 2-(2-(2methylpropyl))pentanoic acid, acid, 8-methyl-nonanoic acid, acid, 4-(1-propyl)-heptanoic acid, 5-(2-propyl)heptanoic acid, 3-methyl-undecanoic acid, 2-pentylheptanoic acid, 2,3,4,5,6-pentamethyl-heptanoic acid, 2,6-diethyl-octanoic acid, 2-hexyl-octanoic acid, 2,3,4,5,6,7-hexamethyl-octanoic acid diethyl-4,4-diethyl-hexanoic acid, acid, 2,3,4,5-tetraethyl-octanoic acid, 2-octyldecanoic acid and 2-(1-propyl)-3-(1-propyl)-4,5-diethyl-6methyl-heptanoic acid.

2-methyl-nonanoic acid 6-ethyl-octanoic acid, 3,32-heptyl-nonanoic acid

Yet another preferred group of acylating groups includes

·· ·* ► ♦ · · ř · ·· » 9 · « » · · « ·· ·· • 9 • · * • · · • · · • 9 9 • 9 « cyclic alkyl acids containing from 5 to 24 carbon atoms, in which the cyclic alkyl moiety or moieties contain 5 to 7 carbon atoms. A representative list of such cyclic alkyl acids will be given to avoid any misunderstanding of the range of fatty acids that can be used as acylating groups for proteins according to the present invention: cyclopentyl formic acid, cyclohexyl formic acid, 1-cyclopentyl acetic acid, 2-cyclohexyl acetic acid, 1,2-dicyclopentyl acetic acid and the like.

A preferred group of derivatized proteins for use in the insoluble compositions of the present invention include monoacylated proteins. Monoacylation at the ε-amine group is most preferred. In the case of insulin, monoacylation at LysB29 is preferred. Similarly, for certain insulin analogs, such as LysB28,ProB29-analog of human insulin, monoacylation at the ε-amine group of LysB28 is most preferred. Monoacylation at the amine group of the B-chain (B1) is also preferred. Monoacylation at the α-amine group of the A-chain (A1) is also preferred.

Another preferred group of acylated proteins for use in the insoluble compositions of the present invention includes diacylated proteins. The diacylation may be, for example, at the ε-amine group of Lys and the α-amine group of the B-chain, or may be at the ε-amine group of Lys and the α-amine group of the A-chain, or may be at the α-amine group of the A-chain and the amine group of the B-chain.

Another preferred group of acylated proteins for use in

0

0

0 0 ···· 00 • 0 • 0

0 0

0

00

0« 0

0 ·

0 · insoluble compositions of the present invention include triacylated proteins. Triacylated proteins are proteins that are acylated at the ε-amine group of Lys, at the amine group of the B-chain, and at the α-amine group of the A-chain.

It is also advantageous to use acylated proteins that are a mixture of monoacylated and diacylated proteins.

It is also advantageous to use acylated proteins that are a mixture of monoacylated and triacylated proteins.

Another preferred group of acylated proteins consists of a mixture of diacylated and triacylated proteins.

It is also advantageous to use acylated proteins that are a mixture of monoacylated, diacylated and triacylated proteins.

Certain fatty acid acylated proteins used in the microcrystals of the present invention will be listed to ensure an understanding of the scope of the present invention. This list is illustrative and the fact that certain fatty acid acylated proteins are not listed does not mean that a microcrystal containing such a protein is not within the scope of the present invention.

B29-Ns-formyl-human insulin.

BΙ-Να-formyl-human insulin.

Al-Na-formyl-human insulin.

B29-Ns-formyl-, B1-Na-formyl-human insulin.

• ·

9 ·9· • 999 *9 • 9

B29-Ne-formyl-,Al-Nα-formyl-human insulin.

ΑΙ-Να-formyl-,ΒΙ-Να-formyl-human insulin.

B29-Ns-formyl-,Al-Nα-formyl-,B1-Na-formyl-human insulin. B29-Ne-acetyl-human insulin.

B1-Na-acetyl-human insulin.

Al-Na-acetyl-human insulin.

B29-Ns-acetyl-, Bl-Na-acetyl-human insulin.

B29-NE-acetyl~,Al-Na-acetyl-human insulin.

Al-Na-acetyl-, Bl-Na-acetyl-human insulin.

B29-Ns-acetyl-,A1-Na-acetyl-,B1-Na-acetyl-human insulin. B29-Ns-propionyl-human insulin.

B1-Na-propionyl-human insulin.

Al-Na-propionyl-human insulin.

B29-Ns-propionyl-,B1-Na-propionyl-human insulin.

B29-Ns-propionyl-,Al-Na-propionyl-human insulin.

Al-Na-propionyl-, Bl-Na-propionyl-human insulin.

B29-Ns-propionyl-,Al-Na-propionyl-,Bl-Na-propionyl-human insulin.

B29-Non-butyryl-human insulin.

Bl-Na-butyryl-human insulin.

Al-Na-butyryl-human insulin.

B29-Ne-butyryl-, Bl-Na-butyryl-human insulin.

B29-Ne-butyryl-,Al-Na-butyryl-human insulin.

ΑΙ-Na-butyryl-,Bl-Na-butyryl-human insulin.

B29-N-butyryl-,A1-N-butyryl-,B1-N-butyryl-human insulin. B29-N8-pentanoyl-human insulin.

B1-Na-pentanoyl-human insulin.

In «

0 * • · 0

0 *

Al-Na-pentanoyl-human insulin.

B29-N8-pentanoyl-,ΒΙ-Να-pentanoyl-human insulin. B29-NE-pentanoyl-,ΑΙ-Να-pentanoyl-human insulin. ΑΙ-Να-pentanoyl-,ΒΙ-Να-pentanoyl-human insulin. B29-Ne-pentanoyl-,ΑΙ-Να-pentanoyl-human insulin.

B29-NE-hexanoyl-human insulin.

BΙ-Να-hexanoyl-human insulin.

Al-N-hexanoyl-human insulin.

B29-N-hexanoyl-,ΒΙ-Να-hexanoyl-human insulin.

B29-N-E-hexanoyl-,A1-Nα-hexanoyl-human insulin.

ΑΙ-Να-hexanoyl-,ΒΙ-Να-hexanoyl-human insulin.

B29-Νε-hexanoyl-,ΑΙ-Να-hexanoyl-,ΒΙ-Να-hexanoyl-human insulin.

B29-NE-heptanoyl-human insulin.

BΙ-Να-heptanoyl-human insulin.

Al-Na-heptanoyl-human insulin.

B29-Nε-heptanoyl-,ΒΙ-Να-heptanoyl-human insulin. B29-Ns-heptanoyl-,ΑΙ-Να-heptanoyl-human insulin. ΑΙ-Να-heptanoyl-,ΒΙ-Να-heptanoyl-human insulin. B29-Ns-heptanoyl-,ΑΙ-Να-heptanoyl-human insulin.

B29-Non-octanoyl-human insulin.

ΒΙ-Να-octanoyl-human insulin.

ΑΙ-Να-octanoyl-human insulin.

B29-N8-octanoyl-, B1-Na-octanoyl-human insulin.

B29-Νε-οktanoy1-, Al-Na-octanoyl-human insulin.

ΑΙ-Να-octanoyl-,ΒΙ-Να-octanoyl-human insulin. B29-Ns-octanoyl-,ΑΙ-Να-octanoyl-, Βΐ-Να-octanoyl-human insulin.

B29-Non-nonanoyl-human insulin.

BΙ-Να-nonanoyl-human insulin.

ΑΙ-Να-nonanoyl-human insulin.

B29-Ns-nonanoyl-,ΒΙ-Να-nonanoyl-human insulin. B29-Ns-nonanoyl-,AI-Να-nonanoyl-human insulin. ΑΙ-Να-nonanoyl-,ΒΙ-Να-nonanoyl-human insulin. B29-Ne-nonanoyl-,ΑΙ-Να-nonanoyl-,Bl-Na-nonanoyl-human insulin.

B29-Non-decanoyl-human insulin.

BΙ-Να-decanoyl-human insulin.

Al-Na-decanoyl-human insulin.

B29 -Νε-decanoyl-,Bl,Να-decanoyl-human insulin. B29-Ne-decanoyl-,ΑΙ-Να-decanoyl-human insulin. ΑΙ-Να-decanoyl-,ΒΙ-Να-decanoyl-human insulin. B29-Ne-decanoyl~,ΑΙ-Να-decanoyl-,Bl-Να-decanoyl-human insulin.

B29-Ne-undecanoyl-human insulin.

BΙ-Να-undecanoyl-human insulin.

Al-Na-undecanoyl-human insulin.

B29-Ns-dodecanoyl-human insulin.

BΙ-Να-dodecanoyl-human insulin.

Al-Na-dodecanoyl-human insulin.

B29-Nε-tridecanoyl-human insulin.

β1-Na-tridecanoyl-human insulin.

• · • · ♦ * » · · • · · • · · · • · · 9 • · · · · · · · · · • · · · » · · • 4 · · · • · · · · ·

Al-Να-tridecanoyl-human insulin.

B29-Ns-tetradecanoyl-human insulin.

BΙ-Να-tetradecanoyl-human insulin.

Al-Na-tetradecanoyl-human insulin.

B29-Non-pentadecanoyl-human insulin.

ΒΙ-Να-pentadecanoyl-human insulin.

ΑΙ-Να-pentadecanoyl-human insulin.

B29-Νε-hexadecanoyl-human insulin.

BΙ-Να-hexadecanoyl-human insulin.

Al-Na-hexadecanoyl-human insulin.

B29-Ns-heptadecanoyl-human insulin.

Bΐ-Να-heptadecanoyl-human insulin.

Al-Na-heptadecanoyl-human insulin.

B29-Ns-octadecanoyl-human insulin.

BΙ-Να-octadecanoyl-human insulin.

Al-Na-octadecanoyl-human insulin.

B28-Ns-formyl-LysB28,ProB29-human insulin analog. Bl-Na-formyl-LysB28,ProB29-human insulin analog. Al-Na-formyl-LysB28,ProB29-human insulin analog. B28-Ns-formyl-, Bl-Na-formyl-LysB28, ProB29-human insulin analog.

B28-Ns-formyl-,Al-Nα-formyl-LysB28,ProB29-analog of human insulin.

ΑΙ-Na-formyl-,Bl-Na-formyl-LysB28,ProB29-analog of human insulin.

B28-N8-formyl~, Al-Na-formyl-,B1-Na-formyl-LysB28,ProB29analog of human insulin.

• you • you « you · you

Even you · « » ·· * • « » *

B28-NE-acetyl-LysB28, ProB29-human insulin analog. Bl-Na-acetyl-LysB28,ProB29-human insulin analog. Al-Na-acetyl-LysB28,ProB29-human insulin analog.

B28-Ns-acetyl-, Bl-Na-acetyl-LysB28,ProB29-analog of human insulin.

B28-Ns-acetyl-,Al-Na-acetyl-LysB28,ProB29-analog of human insulin.

Al-Na-acetyl-, Bl-Na-acetyl-LysB28, ProB29-analog of human insulin.

B28-Νε-acetyl-, ΑΙ-Να-acetyl-,ΒΙ-Να-acetyl-LysB28,ProB29 analog of human insulin.

B28-Ne-propionyl-LysB28,ProB29-human insulin analog. Bl-Na-propionyl-LysB28,ProB29-human insulin analog. Al-Na-propionyl-LysB28,ProB29-human insulin analog. B28-Ns-propionyl-,Bl-Na-propionyl-LysB28,ProB29-human insulin analog.

B28-Ne-propionyl-,Al-Na-propionyl-LysB28,ProB29-analog of human insulin.

Al-Na-propionyl-,Bl-Na-propionyl-LysB28,ProB29-analog of human insulin.

B28-NE-propionyl-,ΑΙ-Να-propionyl-,Bl-Na-propionylLysB28,ProB29-human insulin analog. B28-Ns-butyryl-LysB28,ProB29-human insulin analog. Bl-Na-butyryl-LysB28,ProB29-human insulin analog. Al-Na-butyryl-LysB28,ProB29-human insulin analog.

B28-Ne-butyryl-, B1-Na-butyryl-LysB28,ProB29-analog of human insulin.

B28-Ns-butyryl-,Al-Na-butyryl-LysB28,ProB29-analog of human insulin.

ΑΙ-Να-butyryl-,Bl-Na-butyryl-LysB28,ProB29-analogue of human insulin.

B28-Ne-butyryl-,ΑΙ-Να-butyryl-,Bl-Na-butyryl-LysB28,ProB29 analog of human insulin.

B28-NE-pentanoyl-LysB28,ProB29-human insulin analog. Bl-Na-pentanoyl-LysB28,ProB29-human insulin analog. Al-Na~pentanoyl-LysB28,ProB29-human insulin analog. B28-Ns-pentanoyl-,Bl-Na-pentanoyl-LysB28,ProB29-human insulin analog.

B28-Νε-pentanoyl-, Al-Na-pentanoyl-LysB28, ProB29-analog of human insulin.

Al-Na-pentanoyl-,Bl-Na-pentanoyl-LysB28,ProB29-analog of human insulin.

B28-Νε-pentanoyl-,Αΐ-Να-pentanoyl-,Bl-Να-pentanoylLysB28,ProB29-analog of human insulin.

B28-Ns-hexanoyl-LysB28,ProB29-human insulin analog. Bl-Na-hexanoyl-LysB28,ProB29-human insulin analog. Al-Na-hexanoyl-LysB28,ProB29-human insulin analog. B28-Ne-hexanoyl-,Bl-Na-hexanoyl-LysB28,ProB29-human insulin analog.

B28-N8-hexanoyl-,Al-Na-hexanoyl-LysB28,ProB29-analog of human insulin.

A1-Nα-hexanoyl-,B1-Nα-hexanoyl-LysB28,ProB29-analog of human insulin.

B28-Νε-hexanoyl-,Αΐ-Να-hexanoyl-,Bl-Να-hexanoylLysB28,ProB29-analog of human insulin.

B28-N8-heptanoyl-LysB28,ProB29-human insulin analog.

999« ««

9

9 9 9

9 9 9 • 9 9 9

99 · 9 · • 9 9 «

9 9 ·

9 9 « «9 «9

Bl-Na-heptanoyl-LysB28, ProB29-human insulin analog. Al-Na-heptanoyl-LysB28,ProB29-human insulin analog. B28-N8-heptanoyl-,Bl-Na-heptanoyl-LysB28,ProB29-human insulin analog.

B28-Ns-heptanoyl-,Al-Na-heptanoyl-LysB28,ProB29-analog of human insulin.

Al-Na-heptanoyl-,Bl-Na-heptanoyl-LysB28,ProB29-analog of human insulin.

B28-Ne-heptanoyl-,Αΐ-Να-heptanoyl-,Bl-Na-heptanoylLysB28,ProB29-analog of human insulin.

B28-Ne-octanoyl-LysB28,ProB29-analogue of human insulin. B1-Na-octanoyl-LysB28,ProB29-human insulin analog. Al-Na-octanoyl-LysB28,ProB29-analogue of human insulin. B28-Ns-octanoyl-, B1-Na-octanoyl-LysB28,ProB29-analogue of human insulin.

B28-Νε-octanoyl-, Al-Na-octanoyl-LysB28,ProB29-analogue of human insulin.

ΑΙ-Να-octanoyl-,Bl-Na-octanoyl-LysB28,ProB29-analogue of human insulin.

B28-Νε-octanoyl-, ΑΙ-Να-octanoyl-,Bl-Na-octanoylLysB28,ProB29-analogue of human insulin.

B28-Ne-nonanoyl-LysB28,ProB29-human insulin analog. Bl-Na-nonanoyl-LysB28,ProB29-human insulin analog. Al-Na-nonanoyl-LysB28,ProB29-human insulin analog. B28-N8-nonanoyl~,Bl-Na-nonanoyl-LysB28,ProB29-human insulin analog.

B28-Ne-nonanoyl-,Al-Na-nonanoyl-LysB28,ProB29-analog of human insulin.

• · · · · * • · · · • · · ·

9 99

9 9 9

9 9 9 ·9 * · · « • · · · • 9 9 9

9 9 9

99

ΑΙ-Να-nonanoyl-,Bl-Na-nonanoyl-LysB28,ProB29-analogue of human insulin.

B28-Ne-nonanoyl-,ΑΙ-Να-nonanoyl-,Bl-Na-nonanoylLysB28,ProB29-analogue of human insulin.

B28-Ne-decanoyl-LysB28,ProB29-human insulin analog. Bl-Na-decanoyl-LysB28,ProB29-human insulin analog. Al-Na-decanoyl-LysB28,ProB29-human insulin analog. B28-Ns-decanoyl-, Bl-Na-decanoyl-LysB28,ProB29-human insulin analog.

B28-Ne-decanoyl-, Al-Na-decanoyl-LysB28,ProB29-analog of human insulin.

Al-Na-decanoyl-,Bl-Na-decanoyl-LysB28,ProB29-analog of human insulin.

B28-N8-decanoyl-, A1-Nα-decanoyl-, B1-Na-decanoylLysB28, ProB29-analog of human insulin.

B28-Ns-undecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-undecanoyl-LysB28,ProB29-human insulin analog. Al-Na-undecanoyl-LysB28,ProB29-human insulin analog. B28-N8-dodecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-dodecanoyl-LysB28,ProB29-human insulin analog. Al-Na-dodecanoyl-LysB28,ProB29-human insulin analog. B28-N8-tridecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-tridecanoyl-LysB28,ProB29-human insulin analog. Al-Na-tridecanoyl-LysB28,ProB29-human insulin analog. B28-N8-tetradecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-tetradecanoyl-LysB28,ProB29-human insulin analog. Al-Na-tetradecanoyl-LysB28,ProB29-human insulin analog.

• φφφ · « φ φ * • · · φ φ φ φ · · φ φ φ φ φ φ < · φ • Φ φφ

B28-Not-pentadecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-pentadecanoyl-LysB28,ProB29-human insulin analog. Al-Na-pentadecanoyl-LysB28,ProB29-human insulin analog. B28-Ns-hexadecanoyl-LysB28,ProB29-human insulin analog. Bl-Not-hexadecanoyl-LysB28,ProB29-human insulin analog. Al-Na-hexadecanoyl-LysB28,ProB29-human insulin analog. B28-Ns-heptadecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-heptadecanoyl-LysB28,ProB29-human insulin analog. Al-Na-heptadecanoyl-LysB28,ProB29-human insulin analog. B28-Ns-octadecanoyl-LysB28,ProB29-human insulin analog. Bl-Na-octadecanoyl-LysB28,ProB29-human insulin analog. Al-Na-octadecanoyl-LysB28,ProB29-human insulin analog. B29-Ns-pentanoyl-GlyA21,ArgB31,ArgB32-human insulin. Bl-Na-hexanoyl-GlyA21,ArgB31,ArgB32-human insulin. Al-Na-heptanoyl-GlyA21,ArgB31,ArgB32-human insulin. B29-Nc-octanoyl-,Bl-Na-octanoyl-GlyA21,ArgB31,ArgB32-human insulin.

B29-Ne-propionyl- _from Al-Na-propionyl-GlyA21,ArgB31,ArgB32human insulin.

Al-Na-acetyl,Bl-Na-acetyl-GlyA21,ArgB31,ArgB32-human insulin.

B29-NE-formyl~,Al-Na-formyl1-,B1-Na-formylGlyA21,ArgB31,ArgB32-human insulin.

B29-Ne-formyl-des(TyrB26)-human insulin.

B1-Na-acetyl-AspB28-human insulin.

B29-N-propionyl-,A1-N-propionyl-,B1-N-propionylAspB1,AspB3,AspB21-human insulin.

• 9 9 · Φ Φ

9 φ· ΦΦ ΦΦ

ΦΦΦΦ *ΦΦ· • Φ ·· Φ Φ ® ·

Φ ΦΦ ΦΦ · * ·'

ΦΦΦΦ

Al-Na-butyryl-AspBlΟ-human insulin.

B29-N-pentanoyl-GlyA21-human insulin.

B1-Na-hexanoyl-GlyA21-human insulin.

Al-Na-heptanoyl-GlyA21-human insulin.

B29-Ne-octanoyl-,B1-Na-octanoyl-GlyA21-human insulin.

B29-N8-propionyl-,Al-Na-propionyl-GlyA21-human insulin. Al-Na-acetyl,B1-Na-acetyl-GlyA21-human insulin.

B2 9-Ns-formyl-, Al-Na-formyl-, B1-Na-formyl-GlyA21-human insulin.

B29-N-butyryl-des(ThrB30)-human insulin.

B1-Na-butyryl-des(ThrB30)-human insulin.

Al-Na-butyryl-des(ThrB30)-human insulin.

B29-Ne-butyryl~, Bl-Na-butyryl-des(ThrB30)-human insulin. B29-Ns-butyryl~, Al-Na-butyryl-des(ThrB30)-human insulin. ΑΙ-Na-butyryl-,Bl-Na-butyryl-des(ThrB30)-human insulin. B29-Ne-butyryl-,Al-Na-butyryl-,Bl-Na-butyryl-des(ThrB30)human insulin.

Aqueous compositions containing water such as are preferred. Aqueous suspensions, in a solvent are highly preferred.

the main solvent of which is water

The compositions of the present invention are used for the treatment of patients who have diabetes or hyperglycemia. The compositions of the present invention typically provide derivatized protein in concentrations from about 1 mg/ml to about 10 mg/ml. Current insulin product compositions are typically characterized using a concentration of insulin activity units (units/ml), such as o ·· • * · · • 9 · · • · · ♦ • 9 9 9

99

U40, U50, U100 and the like, which roughly correspond to approximately 1.4, 1.75 and 3.5 mg/ml preparations. The dose, route of administration and number of administrations per day will be determined by the physician, taking into account such factors as the therapeutic objective, the nature and cause of the patient's disease, the patient's sex and weight, level of training, eating habits, administration procedure and other factors known to the trained physician. Roughly speaking, a daily dose in the range of about 1 nmol/kg body weight to about 6 nmol/kg body weight (6 nmol is considered equivalent to about 1 unit of insulin activity) can be given. A dose in the range of about 2 to about 3 nmol/kg is typical for current insulin therapy.

A physician of ordinary skill in the treatment of diabetes is able to select the most therapeutically advantageous means for administering the compositions of the present invention.

Parenteral administration is preferred. Typical routes for parenteral administration of insulin suspension preparations are subcutaneous and intramuscular. The compositions and preparations of the present invention may also be administered by the nasal, buccal, pulmonary or ocular routes.

Glycerol at a concentration of from about 12 mg/ml to about 25 mg/ml is preferred as an isotonic agent. Even more preferred as an isotonic agent is the use of glycerol at a concentration of from about 15 mg/ml to about 17 mg/ml.

M-cresol and phenol, or mixtures thereof, are preferred preservatives in the compositions of the present invention.

0

0* 0·

0 0 ·

0 0 0

0· ·

0 0 · • 0 00

Insulin, insulin analogs, or proinsulins used for the preparation of derivatized proteins can be prepared by any of a number of recognized peptide synthesis techniques including classical (solution) methods, solid phase synthesis, semisynthetic methods, and more recently recombinant DNA methods. See, for example, Chance, R. B. et al., U.S. Patent No. 5,514,646, May 7, 1996; EPO Publication No. 383,472, February 7, 1996; Brange, J. J. V. et al., EPO Publication No. 214,826, March 18, 1987; and Belagaje, R. M. et al., U.S. Patent No. 5,304,473, April 19, 1994, which describe the preparation of various proinsulins and insulin analogs. These references are expressly incorporated herein by reference.

In general, derivatized proteins are prepared using methods known in the art. The above-referenced list of publications referred to above describes derivatized proteins and contains suitable methods for preparing derivatized proteins. These references are expressly incorporated herein by reference for methods for preparing derivatized proteins. To prepare acylated proteins, the protein is reacted with an activated organic acid, such as an activated fatty acid. Activated fatty acids are derivatives of commonly used acylating agents and include activated fatty acid esters, fatty acid halides, activated fatty acid amides, such as an activated azolide derivative [Hansen, L. B., WIPO Publication No. 98/02460, 22 January 1998] and fatty acid anhydrides. The use of activated esters, especially N-hydroxysuccinimide esters of fatty acids is a particularly preferred means for acylating a free amino acid with a fatty acid. Lapidot et al. describe the preparation of N-hydroxysuccinimide esters and • · 0· • · * · · • · · ·· φ · · 9 9 9

9 9 9 9

999

6Ί their use in the preparation of N-lauroyl-glycine, N-lauroyl-L-serine and N-lauroyl-L-glutamic acid. The term activated fatty acid ester means a fatty acid that has been activated using general techniques known in the art [Riordan, J. F. and Vallee, B. L., Methods in Enzymology, XXV:494-499 (1972); Lapidot, Y. et al., J. Lipid Res. 8:142-145 (1967)]. Hydroxybenzotriazide (HOBT), N-hydroxysuccinimide and their derivatives are particularly well known for generating activated acids for peptide synthesis.

For selective acylation of the ε-amine group, various protecting groups can be used to block the α-amine groups during coupling. The choice of a suitable protecting group is known to those skilled in the art and includes p-methoxybenzoxycarbonyl (pmZ). Preferably, the ε-amine group is acylated in a one-step synthesis without the use of amine protecting groups. A method for selective acylation of the Νε-amine group of a Lys residue is described and claimed in Baker, J. C. et al., U.S. Patent No. 5,646,242, July 8, 1997, which is expressly incorporated herein by reference in its entirety. A method for preparing a dry powder of acylated protein is described and claimed in Baker, J. C. et al., U.S. Patent No. 5,700,904, December 23, 1997, which is expressly incorporated herein by reference in its entirety.

The primary role of zinc in the present invention is to facilitate the formation of zinc hexamers of protein and derivatized protein, either separately as mixed hexamers or together as hybrid hexamers. Zinc facilitates the formation of hexamers of insulin and insulin analogs. Zinc also aids in the formation of hexamers of derivatized insulins and insulin analogs.

4*44 ·» • * 4 • 4 4

4 4

4 4 4

44

4* 44 * * 4

4 44 • 4 4 4 4 • 4 4 4

44 *4

4 4 4

4 4 4

4 4 4

4 4 4

44 analogs. Hexamer formation is conveniently achieved by adjusting the pH of a solution containing the protein or derivatized protein.

or both to the neutral region of divalent zinc or by adding pH adjustment to the neutral region.

in the presence of divalent zinc ions after further extension of time

For efficient microcrystal or amorphous precipitate yield, the molar ratio of zinc to total protein in the microcrystal and amorphous precipitate of the present invention is limited to a lower limit of about 0.33, i.e., about two zinc atoms per hexamer, which are required for efficient hexamerization. The composition of the microcrystal and amorphous precipitate is suitably formed with about 2 to about 4-6 zinc atoms in the absence of any other compound that competes with insulin for zinc. Even more zinc may be used in the process if a compound that competes with the protein for zinc binding, such as a citrate or phosphate-containing compound, is present. An excess of zinc above the minimum amount required for efficient hexamerization may be desirable to facilitate faster hexamerization. An excess of zinc above the minimum amount may also be present in the composition of the present invention and may be desirable to improve chemical and physical properties such as stability, to improve suspendability and possibly also efficacy. As a result

3θ for j are allowable zinc:protein ratios over very wide ranges for use in the insoluble compositions, methods and preparations of the present invention.

In accordance with the present invention, zinc is present

·

4 · · ·

• 4 ·

in the preparation in an amount of from about 0.3 moles to about 7 moles per mole of total protein, and more preferably from about 0.3 moles to about 1.0 moles of total protein. Even more preferably, the ratio of zinc to derivatized protein is in the range of from about 0.3 to about 0.7 moles of zinc atoms per mole of total protein. Most preferably, the ratio of zinc to total protein is from about 0.30 to about 0.55 moles of zinc atoms per mole of total protein.

In preparations with higher zinc content, which are similar to PZI preparations, the zinc ratio is from about 5 to about 7 moles of zinc per mole of total protein.

The zinc compounds that provide zinc according to the present invention can be any pharmaceutically acceptable zinc compound. The addition of zinc to insulin preparations is known in the art, as are pharmaceutically acceptable sources of zinc. Preferred zinc compounds for providing zinc according to the present invention include zinc chloride, zinc acetate, zinc citrate, zinc oxide and zinc nitrate.

The complexing compound is required for the formation of microcrystals and precipitates according to the present invention. The complexing compound must be present in sufficient amounts to cause substantial precipitation and crystallization of the hexamers. Such amounts can be readily determined for each particular preparation of a particular complexing compound by simple titration experiments. Ideally, the concentration of the complexing compound is adjusted so that only a negligible amount of the complexing compound remains in the

0000 ·· • 9

0 0 0

0 0 0

0·

0· 0 0 0

0 0 soluble phase after the precipitation and crystallization processes are complete. This requires the use of a complexing compound, in an amount based on an experimentally determined isophane ratio. This ratio is expected to be very similar to that for NPH and NPL. However, it may be slightly different, as derivatization can affect the nature of the protein-protamine interaction.

When protamine is a complexing compound, it is present in the microcrystal in an amount of from about 0.15 mg to about 0.5 mg per 3.5 mg of total protein. The ratio of protamine to total protein is preferably from about 0.25 to about 0.40 (mg/mg). More preferably, this ratio is from about 0.25 to about 0.38 (mg/mg). Preferably, protamine is used in an amount of from 0.05 mg to about 0.2 mg per mg of total protein, and more preferably from about 0.05 to about 0.15 milligrams of protamine per milligram of total protein. Protamine sulfate is a preferred salt form of protamine for use in the present invention. If protamine sulfate or another salt of protamine is used, the weight of the salt used must be adjusted relative to the weight of protamine free base that would be used in the same preparation depending on the ratio of the molecular weights of the salt and protamine.

To further extend the duration of action of the compositions of the present invention or to improve their suspendability, additional protamine and zinc may be used after crystallization. Therefore, formulations containing a higher amount of protamine than the isophane ratio are also within the scope of the present invention. In these formulations, the protamine ratio is from 0.25 mg to about 0.5 mg of protamine per • 4 44

4 4 4

4 44

4 4 4

4 4 4

44 ···· ·· • 4

4 4

4 4 · 4

4 4 one mg of total protein.

A desired component of the microcrystals and precipitates of the present invention is a hexamer stabilizing compound.

The structures of three hexamer conformations, designated T6, T3R3 and R6, have been characterized in the literature. In the presence of a hexamer stabilizing compound, such as various phenolic compounds, the R6 conformation is stabilized. It is therefore highly likely that the hexamers are in the R6 conformation or in the T3R3 conformation in crystals and precipitates formed in the presence of a hexamer stabilizing compound, such as phenol or mcresol, among others. A wide range of hexamer stabilizing compounds is suitable. These compounds must be present in sufficient proportions relative to the total protein to stabilize the R6 hexamer conformation. For this, at least 2 or at least 3 moles of hexamer stabilizing compound per hexamer are required for effective stabilization of the hexamers. It is preferred that at least 3 moles of hexamer stabilizing compound per hexamer are present in the microcrystals and precipitates of the present invention. The presence of higher ratios of hexamer stabilizing compound, at least up to 25-50-fold, in the solution from which microcrystals and precipitates are formed does not adversely affect the stabilization of the hexamers.

A preservative may be present in the compositions of the present invention, particularly if the composition is intended for multiple use. As noted above, a wide variety of preservatives are known. Preferably, the preservative is present in the solution in an amount suitable to provide an antimicrobial effect sufficient to meet the pharmacological requirements.

Preferred preservatives are the phenolic preservatives listed above. Preferred phenolic preservatives are from about 2 mg to about 5 mg per milliliter of aqueous suspension preparation. These concentrations are based on the total weight of phenolic preservatives, as mixtures of individual phenolic preservatives are also contemplated. Suitable phenolic preservatives include, for example, phenol, m-cresol, and methylparaben. Preferred phenolic compounds are phenol and m-cresol. Mixtures of phenolic compounds, such as phenol and m-cresol, are also contemplated and are highly preferred. Examples of mixtures of phenolic compounds are 0.6 mg/ml phenol and 1.6 mg/ml m-cresol or 0.7 mg/ml phenol and 1.8 mg/ml m-cresol.

The microcrystals of the present invention are preferably elongated, also called rod-shaped, single crystals that are composed of a protein, a derivatized protein, a divalent cation, and include a complexing compound and a hexamer-stabilizing compound. The mean length of the microcrystals of the present invention is preferably in the size range of 1 micron to 40 microns, and more preferably in the size range of 3 microns to 15 microns.

Preferred compositions contain from about 3 mg to about 6 mg of protamine sulfate per 35 mg of total protein and from about 0.1 to about 0.4 mg of zinc per 35 mg of total protein. Other preferred compositions contain from about 10 mg to about 17 mg of protamine sulfate per 35 mg of total protein and from about 2.0 to ·· ·· • 4 · ·

44·· 44 • 4 · • · · · • · · · · ·

4»

4 4

4 4

4 4

4 4 approximately 2.5 mg zinc per 35 mg total protein. Other preferred compositions contain per ml 0.34-0.38 mg. protamine sulfate; 0.01-0.04 mg zinc; and 3.2-3.8 mg total protein.

Both non-derivatized protein and derivatized protein are required for the co-crystals and amorphous precipitates of the invention. The weight ratio of the two proteins determines the amount of time extension of the preparations. A preferred ratio of moles of protein to moles of derivatized protein is between about 1:100 and about 100:1. Another preferred ratio of moles of protein to moles of derivatized protein is between about 1:1 and about 100:1. Another preferred ratio of moles of protein to moles of derivatized protein is between about 1:1 and about 20:1. Still other preferred ratios of moles of protein to moles of derivatized protein: between about 2:1 and about 20:1; between about 2:1 and 10:1; between about 2:1 and 5:1; between about 3:1 and 5:1; between 1:1 and 1:20; between 1:1 and 1:10; between about 1:2 and about 1:20; between about 1:2 and 1:10; between about 1:2 and 1:5; between about 1:3 and 1:5; between about 10:1 and about 1:10; between about 9:1 and about 1:9; between about 5:1 and about 1:5; and between about 3:1 and about 1:3.

The present invention relates to methods of preparing the compositions. It is also contemplated that the insoluble compositions of the invention may be used to prepare medicaments for controlling blood glucose levels and for treating diabetes or hyperglycemia. The amorphous precipitates and microcrystals of the present invention «« *· • · · · • » ·· « · · · ·*«· ·* ·· • · · · • · ♦ • · · · · · · · »· may be prepared for use in medicaments or for other uses in a variety of ways.

In summary, suitable methods generally consist of steps in one of the following sequences: solubilization (if starting from dry material), hexamerization, homogenization, complexation, precipitation, crystallization and optionally preparation; or solubilization (if starting from dry material), homogenization, hexamerization, complexation, precipitation, crystallization and optionally preparation.

Solubilization refers to the dissolution of the derivatized protein and protein sufficiently to form hexamers. Hexamerization refers to the process in which protein and derivatized protein molecules bind with divalent zinc atoms to form hexamers. Complexation refers to the formation of insoluble complexes between the hexamers and protamine. Precipitation typically occurs upon formation of insoluble complexes. Crystallization involves the conversion of the precipitated hexamer-protamine complexes into crystals, which are typically rod-shaped crystals.

Solubilization is carried out by dissolving the derivatized protein and the protein in an aqueous solvent. The aqueous solvent may be, for example, an acidic solution, a neutral solution, or a basic solution. The aqueous solvent may partially contain a water-miscible organic solvent, such as ethanol, acetonitrile, dimethyl sulfoxide, and the like. Acidic solutions may be, for example, HCl solutions, preferably from about 0.01 N HCl to about 1.0 N HCl. Other acids that are pharmaceutically acceptable may also be used. Basic •9 · * ·

The solutions may be, for example, NaOH solutions, preferably from about 0.01 N NaOH to about 1.0 N NaOH or higher. Other bases which are pharmaceutically acceptable may also be used. For reasons of protein stability, the concentration of the acid or base is preferably as low as possible while still being effective to sufficiently solubilize the protein and derivatized protein.

Most proteins (insulin, insulin analogs, and proinsulins) and many derivatized proteins can be dissolved to a suitable concentration at neutral pH. Solutions for derivatized proteins at neutral pH may contain a buffer and optionally one or more additional agents such as phenolic compounds, zinc, and isotonic agents.

If hexamerization occurs before homogenization, two populations of homogeneous hexamers are formed first and then the populations are mixed, forming mixed hexamers. If homogenization occurs first, hexamerization yields hybrid hexamers. As noted above, to prepare insoluble compositions consisting of hybrid hexamers, the protein and derivatized protein are homogenized under conditions that promote dissociation into monomeric or dimeric aggregation states prior to hexamerization with a divalent metal cation. To achieve the necessary dissociation, the protein and derivatized protein may be mixed under strongly acidic or strongly basic conditions. The degree of dissociation, and thus the degree of homogenization, is influenced by the solution conditions chosen for this step. Insulin and related proteins readily self-associate in a series of reactions producing dimers, hexamers, and other associated forms. The equilibrium distribution of these association forms is fc · • · • fc « » · · « ·· »· • fc «fc • · · · • fcfc ♦ • · fcfc fc fc· dependent on many parameters including pH. It is generally assumed that these association reactions involve a primary, monomer-dimer-hexamer system. Consequently, depending on the chosen solution conditions, homogenization should achieve mixing of monomers, dimers, or mixtures thereof. Homogenization in, for example, 1 N HCl, may involve higher fractions of monomer mixtures than in 0.1 N HCl, which would likely involve more dimer mixtures. For the preparation of compositions consisting of hybrid hexamers, the homogenization process will be effective provided that only a very small or negligible fraction of homogeneous hexamers of the protein or derivatized protein exist under the homogenization conditions used.

Compositions consisting of mixed hexamers comprise predominantly two types of hexamers, namely protein hexamers and derivatized protein hexamers. In this case, the homogenization step occurs after the hexamerization step and achieves homogenization of the hexamers before complexation with a complexing compound is performed. Consequently, the homogenization step is performed under solution conditions that stabilize the zinc divalent insulin hexamer. Solution conditions that stabilize insulin hexamers are well known in the literature.

The solution conditions required for hexamerization are those that allow the formation of hybrid hexamers or mixed hexamers in solution. These conditions will be identical to or very similar to the conditions under which insulin or insulin analogs are hexamerizable. Typically, hexamerization requires zinc and a neutral to slightly basic pH, meaning from about pH 6.8 to about pH 8.4. Hexamer-stabilizing compounds favor hexamerization by promoting the formation of the R6 conformation.

The presence of a T3R3 derivatized protein and a protein. In certain monomeric hexamer-stabilizing compounds, hexamers.

under certain circumstances also insulin analogues are necessary for the creation

For compositions consisting of hybrid hexamers, seven hexameric species are expected: P ₆ , P5D1, P4D2, P3D3, P ₂ D ₄ , P1D5 and D6 , where P represents the protein monomer and D represents the derivatized protein monomer. The statistical distribution of hexamers is expected to follow a Poisson distribution and is influenced by the relative ratio of protein to derivatized protein and the degree of dissociation prior to hexamerization. For example, from a homogenized solution consisting predominantly of dimers, four major species of hybrid hexamers are expected to form: P ₆ , P ₄ D ₂ , P2D ₄ and D ₆ . For compositions consisting of mixed hexamers, only two species of hexamers are expected to be predominant: P ₆ and D ₆ .

The complexation step must involve combining the complexing compound with the hexamer under solution conditions under which they are soluble. This can be achieved by combining separate solutions of the hexamers and protamine or by first forming a solution of the protein, derivatized protein, and protamine at an acidic or basic pH and then shifting the pH to the neutral region.

During crystallization, the solution conditions must stabilize the crystallizing species and facilitate the transformation of precipitates into • ti · ti · · • ti > · ti ti » · * » ti ·· ti * · ' » titi ti ti ti · ti . ti · ti ti · · ¹ •4 ti* ·· · * crystals. The solution conditions therefore determine the rate and final product of crystallization. Crystallization usually involves a complex equilibrium consisting of precipitate, dissolved complexes of non-crystalline hexamer-protamine, and crystals, the conditions of the crystals. In addition to the assumed equilibrium, the derivatized protein

To obtain microcrystals, the crystallization conditions must be chosen to lead to equilibrium formation. From a perspective based on the assumption that solubility profoundly affects the rate of crystallization and crystal size, since lower solubility is likely to reduce the net conversion of precipitate from solution to crystal. Furthermore, it is well known that slowing the rate of crystallization often leads to the formation of larger crystals. Therefore, it is assumed that the rate of crystallization and crystal size depend on the size and nature of the derivatizing moiety of the derivatized protein.

Crystallization parameters that affect the rate of crystallization and crystal size of the present invention are: the size and nature of the acyl group; temperature; the presence and concentration of compounds that compete with the protein and derivatized protein for zinc, such as citrate, phosphate, and the like; the nature and concentration of the phenolic compound or compounds; the concentration of zinc; the presence and concentration of water-miscible organic solvents; the time allowed for crystallization; pH and ionic strength; the nature and concentration of the buffer; the concentration of precipitants; the presence of materials; the shape and material of the vessel; the speed of agitation; and the total protein concentration. The temperature and concentration of compounds competing for zinc are considered to be particularly important.

concentration of vaccinators • •44 4» • 4 • «4 · • 4 4 · • · 4 4 4 • 4 4 · · 4 4

Compounds in competition for zinc, such as citrate, can affect the rate of crystal formation and indirectly the size and quality of the crystals.

These compounds may exert their effect by forming coordination complexes with zinc in solution, thereby competing with the relatively weak zinc binding sites on the surface of zinc hexamers. Occupancy of these weak surface binding sites is likely to prevent crystallization. In addition, many derivatized proteins are partially insoluble in the presence of little more than 0.333 zinc per mole of derivatized protein, and the presence of the competitor compounds restores solubility and permits crystallization. The optimal concentration of the competitor compounds can be determined using conventional techniques for any combination of protein and derivatized protein. The upper limit is, of course, the concentration at which zinc is precipitated by the competitor compound or the concentration at which residual competitor compound would be pharmaceutically unacceptable, such as if it caused pain or irritation at the site of administration.

The following is an example of a method for preparing precipitates and crystals according to the present invention. A measured amount of derivatized protein and a measured amount of protein are dissolved or otherwise combined to form a solution in an aqueous solvent containing a hexamer-stabilizing compound, such as a phenolic compound. To this solution is added a solution of zinc in the form of its soluble salts, for example ZnCl ₂ , to provide about 0.3 moles of zinc per mole of derivatized insulin or up to about 0.7 moles or up to 1.0 moles of zinc per mole of total protein, (protein + derivatized protein). Absolute ethanol or other miscible organic solvent may optionally be added to this solution in an amount necessary to provide the solution with about 5% to about 10% by volume of organic solvent. This solution may then be filtered through a 0.22 micron filter, which binds low protein components. The protamine solution is prepared by dissolving a measured amount of protamine in an aqueous solvent. This solution may be filtered through a 0.22 micron filter, which binds low protein components. The protein and derivatized protein solution and the protamine solution are combined, after which a precipitate is first formed. The resulting suspension is slowly stirred at ambient temperature (typically about 20-25°C) and then microcrystals are formed over a period of time from about 4 hours to about 10 days.

The microcrystals can then be separated from the mother liquor and placed in a different solvent for storage and administration to the patient. Examples of suitable aqueous solvents are as follows: water for injection containing 25 mM TRIS, 5 mg/ml phenol and 16 mg/ml glycerol; water for injection containing 2 mg/ml dibasic sodium phosphate, 1.6 mg/ml m-cresol, 0.65 mg/ml phenol and 16 mg/ml glycerol; and water for injection containing 25 mM TRIS, 5 mg/ml phenol, 0.1 M sodium citrate and 16 mg/ml glycerol.

In another method of preparing insoluble compositions of the present invention, for example, a measured amount of dry derivatized protein and a measured amount of dry protein are dissolved together in an acidic aqueous solvent such as 0.1 N - 1.0 N HCl. This solution is stirred to ensure thorough mixing of the derivatized protein and protein. The ratio of derivatized protein powder to protein powder in this mixture is predetermined to achieve a similar ratio of derivatized protein to protein in the insoluble composition to be formed. A separately prepared aqueous solution consisting of a phenolic preservative and optionally a pharmaceutically acceptable buffer is combined with the acidic protein solution. The pH of the resulting solution is then adjusted to about 6.8 to about 8.4, preferably from about 6.8 to about 8.0, or preferably to a pH of from about 7.2 to about 7.8, and most preferably from about 7.4 to about 7.8. To this solution is added a solution of zinc in the form of its soluble salt, for example ZnCl ₂ , to provide from about 0.3 moles of zinc per mole of total insulin up to about 4 moles of zinc per mole of total insulin. The pH of this solution is adjusted to the value indicated above and preferably to about 7.4-7.6 and may then be filtered through a 0.22 micron filter that binds low protein components. A protamine solution is prepared by dissolving a measured amount of protamine in an aqueous solvent. The protamine solution may be filtered through a 0.22 micron filter that binds low protein components. The protein and derivatized protein solution and the protamine solution are combined and a precipitate then begins to form. The resulting suspension is slowly stirred at ambient temperature (typically about 20-25 °C) , after which microcrystals are formed over a period of about 4 hours to about 10 days.

In another method of preparing insoluble compositions according to • · 9 · · · « *

9

♦ * ♦ * • · * * of the present invention, a measured amount of derivatized protein is first dissolved in an aqueous solvent containing a phenolic preservative. To this solution is added a solution of zinc in the form of its soluble salt, for example ZnCl ₂ , to provide about 0.3 moles of zinc per mole of derivatized protein up to about 4 moles of zinc per mole of derivatized protein. The pH of the resulting solution is then adjusted to a value of about 6.8 to about 8.4, preferably from about 6.8 to about 8.0, or preferably to a pH of about 7.2 to about 7.8, and most preferably from about 7.4 to about 7.8. A second solution is prepared separately, wherein a measured amount of a protein selected from the group consisting of insulin, insulin analogs, and proinsulin is dissolved in an aqueous solvent containing a phenolic preservative. To this solution is added a solution of zinc in the form of its soluble salt, for example ZnCl ₂ , to provide from about 0.3 moles of zinc per mole of protein to about 4 moles of zinc per mole of protein. The pH of the resulting solution is then adjusted to a value of about 6.8 to about 8.4, preferably from about 6.8 to about 8.0, or preferably to a pH of about 7.2 to about 7.8, and most preferably from about 7.4 to about 7.8 or 7.4-7.6. Portions of the derivatized protein solution and the protein solution are then combined in a ratio that is predefined to achieve a similar ratio of derivatized protein to protein in the insoluble composition. This solution is mixed to ensure thorough mixing of the derivatized protein and the protein. This solution is then adjusted to a pH of about 7.6 and may then be filtered through a 0.22 micron filter that binds low protein components. Protamine solution is prepared separately by dissolving a measured amount of protamine in an aqueous solvent. This protamine solution can be filtered

Φ··· φ* • φ φ φ * ·

::

·♦ ♦· through a 0.22 micron filter that binds low protein components. The protein and derivatized protein solution and the protamine solution are combined and then an initial precipitate is formed. The resulting suspension is stirred slowly at ambient temperature (typically about 20-25 °C) and then microcrystals are formed over a period of about 4 hours to about 10 days.

While not all of the numerous methods that lead to the preparation of insoluble compositions of the present invention will be described, the following procedures provide additional methods of the present invention:

dissolving the protein, derivatized protein, hexamer stabilizing compound, and divalent metal cation in an aqueous solvent having a pH that allows for hexamer formation and adding a complexing compound;

dissolving the protein, derivatized protein, hexamer stabilizing compound, and divalent metal cation in an aqueous solvent having a pH that does not permit hexamer formation, adjusting the pH to a value in the range of between about 6.8 and about 7.8, and adding the complexing compound;

dissolving the protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent having a pH allowing the formation of hexamers, separately dissolving the derivatized protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent having a pH allowing the formation of hexamers, thoroughly mixing the two solutions with each other and then adding the complexing compound;

dissolution of protein, hexamer-stabilizing compound, divalent metal cation and complexing compound in ♦ · · · · ♦ • ·

- 84 in an aqueous solvent, where the resulting solution has a pH at which precipitation does not occur, separately dissolving the derivatized protein, the hexamer-stabilizing compound, the divalent metal cation and the complexing compound in an aqueous solvent, where the resulting solution has a pH at which precipitation does not occur, thoroughly mixing these two solutions with each other and adjusting the pH of the solution to a value at which precipitation occurs;

dissolving the protein, derivatized protein, hexamer stabilizing compound, divalent metal cation, and complexing compound in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur, and adjusting the pH of the solution to a value at which precipitation occurs;

dissolving the protein, derivatized protein, hexamer stabilizing compound and divalent metal cation in an aqueous solvent, the resulting solution having a pH at which precipitation does not occur when a complexing agent is added, adding the complexing compound and adjusting the pH of the solution from step b) to a value at which precipitation occurs;

dissolving the protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, the resulting solution having a pH at which precipitation does not occur when a complexing compound is added, separately dissolving the derivatized protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, the resulting solution having a pH at which precipitation does not occur when a complexing compound is added, thoroughly mixing the two solutions with each other, adding a complexing compound to the solution and adjusting the pH to

I • 999 · · «9 «

9 9 • «

9· • 9 9 9 ♦ · «

9 9 9

9 9 9

9 9 ·

9 ·· the value at which precipitation occurs;

dissolving the protein, the protein derivative, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, the resulting solution having a pH at which precipitation does not occur when a complexing compound is added, adjusting the pH of the solution to a value at which precipitation occurs when a complexing compound is added and adding the complexing compound to the solution; dissolving the protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, the resulting solution having a pH at which precipitation does not occur when a complexing compound is added, separately dissolving the derivatized protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, the resulting solution having a pH at which precipitation does not occur when a complexing compound is added; thoroughly mixing the two solutions with each other, adjusting the pH of the solution from step c) to a value at which precipitation occurs when a complexing compound is added and adding the complexing compound to the solution;

In a preferred embodiment, the microcrystals are prepared in a manner that eliminates the need to separate the microcrystals from the mother fluid. It is preferred that the mother fluid itself be suitable for administration to a patient or that the mother fluid can be made into a fluid suitable for administration by dilution with a suitable diluent. The term diluent herein refers to a solution containing an aqueous solvent in which pharmaceutically acceptable excipients are present, including but not limited to a buffer, an isotonic preservative, protamine, and the like.

dissolved various in it including reagent, zinc, ♦··· ·· * * • · · · ·

9 · ·· <» · 9 9 9

9 9 9 9

9 9 9

99

9 9 · • · · ·

9 9 · • · · ·

9 9 9

In addition to the protein, derivatized protein, divalent cation, complexing compound, and hexamer-stabilizing compound, the pharmaceutical compositions adapted for parenteral administration of the present invention may include additional excipients and carriers such as water-miscible organic solvents such as glycerol, sesame oil, aqueous propylene glycol, and the like. When present, such agents are typically used in amounts of less than about 2.0% by weight, based on the final formulation. For further information on a number of techniques using conventional excipients or carriers for parenteral products, see Remington's Pharmaceutical Sciences, 17th Edition, Mack Publishing Company, Easton, PA, USA (1985), which is incorporated herein by reference.

In the general application of the present invention, it is also contemplated that the formulation may comprise a mixture of microcrystals and a soluble protein fraction selected from the group consisting of insulin, derivatized insulin, insulin analogs and derivatized insulin analogs. Examples of such pharmaceutical compositions include sterile, isotonic, aqueous saline solutions of insulin, insulin analog, derivatized insulin or derivatized insulin analog, buffered with a pharmaceutically acceptable buffer and non-pyrogenic. Preferred soluble phases are insulin or a rapid-acting insulin analog such as LysB28,ProB29-human insulin or AspB28-human insulin. Such mixtures are intended to provide a combination of mealtime glucose control provided by soluble insulin and basal glucose control provided by insoluble insulin. The ratio

9999 99 • · ·

9 9 • · <

• · 9 • 9 • · 9 9 • · 99

9 « 9 9 • · 9 9

9 99

9 9

9 9 ·9 ·· ·

The ratio of total protein (protein plus derivatized protein) in the insoluble phase to total protein in the soluble phase is in the range of about 9:1 to about 1:9. A preferred range for this ratio is from about 9:1 to about 1:1, and more preferably about 7:3. Other possible ratios are 1:1 and 3:7.

Examples of embodiments of the invention

The following preparations and examples illustrate and explain the present invention. The scope of the present invention is not limited by these preparations and examples. References to parts for solids mean parts by weight. References to parts for liquids mean parts by volume. Percentages, when used to express concentrations, mean weight per unit volume (x 100). All temperatures are in degrees Celsius (°C). TRIS refers to 2-amino-2-hydroxymethyl-1,3-propanediol. A 1000 parts per million (ppm) zinc solution was prepared by diluting 1.00 ml of a 10,000 ppm zinc atomic absorption standard solution [Ricca Chemical Company, zinc in dilute nitric acid] with water to a final volume of 10.00 ml.

In many of the preparations listed below, the yield of precipitate and crystals was estimated. The yield estimate refers to the determination of the amount of total protein in the precipitate or crystals and an estimate of the amount of protein that was initially in solution. To determine the amount of total protein, samples of the redissolved precipitate or crystals and the supernatant above the precipitate or crystals were analyzed by reverse-phase gradient HPLC as described below.

···· ·* » ♦ · « · ·

9 9

9

Briefly, the analytical system consisted of a CS reversed phase column at 23°C. The flow rate was 1.0 mL/min and UV detection was performed at 214 nm. Solvent A was 0.1% (v/v) trifluoroacetic acid in a 10:90 (v/v) acetonitrile:water mixture. Solvent B was 0.1% (v/v) trifluoroacetic acid in a 90:10 (v/v) acetonitrile:water mixture. The development program was (minutes, %B): (0.1; 0); (45.1; 75); (50.1; 100); (55; 100); (57; 0); (72; 0). All changes were linear. The skilled artisan can use other analytical systems to achieve the same goal.

To prepare for HPLC analysis, aliquots of the well-mixed suspension were dissolved by dilution with either 0.01 N HCl or 0.03 N HCl. The HPLC analysis results of these solutions allowed for the calculation of total protein. Aliquots of the suspensions were centrifuged for approximately 5 minutes in an Eppendorf 5415C microcentrifuge at 14,000 rpm. The settled supernatant was diluted with either 0.01 N or 0.1 N HCl and analyzed by HPLC. The precipitate was washed by resuspension in Dulbecco's phosphate-buffered saline (free of calcium or magnesium) and pelleted by centrifugation. The buffer was decanted and the solid was redissolved in 0.01 N HCl. The redissolved precipitate was analyzed by HPLC.

HPLC was used to confirm the presence of the expected proteins in the acidified suspension, the redissolved precipitate and the supernatant and also to determine the protein concentrations. The retention times of the peaks in the chromatograms of the redissolved precipitates were compared with the retention times observed for protamine and the active compounds used for the preparation. The agreement of the retention times was always good, ·· ·· • · ·· * · 9 9 • · ·· • 9 « · · • ♦ ♦ · ·· ·<

· 9 • · · · · 9 • · · · · · · which demonstrated that protamine, protein, and derivatized proteins were indeed incorporated into the microcrystals. The concentrations of protein and derivatized protein were determined by comparing the corresponding peak areas with standards. A 0.22 mg/ml solution of derivatized insulin was used as a standard. A standard containing protamine was also used, but only to determine the retention time. Protamine concentrations were not determined.

In many of the preparations described below, a standard spectrophotometric assay was used to determine how quickly the crystals dissolved in Dulbecco's phosphate-buffered saline (pH 7.4) at ambient temperature. Significant deviations from the procedure described immediately below are noted, where necessary, in the description of the individual preparations. A spectrophotometer suitable for measurement in the ultraviolet region and equipped with a 1 cm cuvette and magnetic stirrer was used in all dissolution assays. A cuvette, containing a small stir bar and 3.00 ml of phosphate-buffered saline (PBS), was placed in the compartment of the spectrophotometer. The instrument was set at 320 nm and zeroed against the buffer alone. Then, 4.0 microliters of a well-suspended preparation, usually containing a concentration approximately equivalent to the U50 of the preparation or approximately 1.6 to 1.8 mg/ml, was added to the cuvette. After waiting 1.0 min for mixing, the optical density at 320 nm was recorded. Since the proteins reported in this work do not absorb light at 320 nm, the decrease in optical density was due to a decrease in light scattering as the crystals dissolved. The time taken for the optical density to fall to half its initial value (t1/ ₂ ) is usually reported. As a control, 2.0 microliters of U100 Humulin N (t1/2) were added.

4444 **

4 · • 4 4

4 4 • 4 4 · • 4 44

44

4 · 4

4 «4

4 4 4

4 4 4

44

4

4 4 ·4 ··

4 4 • 4 44 •

is human NPH insulin, which is also known as human NPH insulin) into 3.00 ml of PBS buffer and the optical density at 320 nm was monitored as above. The dissolution half-life (t1/2) for Humulin® was approximately 6 minutes.

Preparation 1

9:1 co-crystals of human insulin and B29-Ns-octanoyl-human insulin

human insulin (0.7 human insulin (6.3 in 1000 parts volume from 50 mM TRIS, 0 , 1 M at pH 7 ,6. To this

parts by weight) were dissolved in an aqueous solvent consisting of sodium citrate and 10 mg/ml phenol solution, 75 parts of a 15.3 mM zinc chloride solution were added. The pH was adjusted to 7.6 with 1 N HCl and/or 1 N NaOH. This solution was filtered through a 0.22 micron filter, which binds low protein components. A second solution was prepared by dissolving 7 parts by weight of protamine sulfate in 10,000 parts by volume of water and filtering through a 0.22 micron filter, which then binds low protein components. Equal volumes of the insulin-containing solution and the acylated insulin and protamine sulfate solution were combined. An amorphous precipitate initially formed. This suspension was allowed to stand for about 24 hours at ambient temperature (typically about 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 2 • · • · · • · · ·

• fc ·· fcfc fc fcfc · • · ·

3:1 co-crystals of human insulin and B29-Ns-octanoyl-human insulin

The method of Preparation 1 was repeated, but using

1.75 parts by weight of dry powder of B29-Ns-octanoyl-LysB29 human insulin and 5.25 parts by weight of dry powder of human insulin. After combining equal volumes of the solution containing insulin and acylated insulin and the protamine sulfate solution, an amorphous precipitate formed. This suspension was left to stand for about 24 hours at ambient temperature (typically about 22 °C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 3

3:1 preparation of cocrystals of human insulin and β29-Νεoctanoyl-human insulin

The co-crystalline microcrystals prepared in Preparation 1 were separated from the supernatant and recovered by conventional solid-liquid phase separation methods such as filtration, centrifugation or sedimentation. The co-crystalline microcrystals obtained were then suspended in a solution containing 25 mM TRIS, 5 mg/ml phenol and 16 mg/ml glycerol, pH 7.8, so that the resulting insulin activity concentration was approximately 100 U/ml.

Preparation 4

1:1 co-crystals of human insulin and B29-Ne-octanoyl-human insulin

The method of Preparation 1 was repeated, but 3.5 • 0 ·· • · 0 · • · ·· • 0 0 · « 0· ·

00 » 0

0 0

0 0

0 0

0 parts by weight of dry powder of B29-N8-octanoyl-LysB29 human insulin and 3.5 parts by weight of dry powder of human insulin. After mixing equal parts of the solution containing insulin and acylated insulin and the protamine sulfate solution, an amorphous precipitate formed. This suspension was left to stand for about 24 hours at ambient temperature (typically about 22 °C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 5

1:3 co-crystals of human insulin and B29-Ns-octanoyl-human insulin

The procedure of Preparation 1 was repeated, but 5.25 parts by weight of dry powder of B29-Ns-octanoyl-LysB29 human insulin and 1.75 parts by weight of dry powder of human insulin were used. After mixing equal parts of the solution containing insulin and acylated insulin and the protamine sulfate solution, an amorphous precipitate was formed. This suspension was allowed to stand for about 24 hours at ambient temperature (typically about 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 6

3:1 co-crystals of human insulin and B29-Ne-hexanoyl-human insulin

The method of Preparation 1 was repeated, but using

1.75 parts by weight of dry powder of B29-N8-hexanoyl-LysB29 human insulin and 5.25 parts by weight of dry powder of human insulin. After mixing equal parts of the solution containing insulin,

An amorphous precipitate formed in the acylated insulin protamine sulfate solution. This suspension was allowed to stand for approximately 24 hours at ambient temperature (typically approximately 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 7

3:1 co-crystals of human insulin and B29-Ns-butyryl-human insulin

The method of Preparation 1 was repeated, but using

1.75 parts by weight of dry powder of B29-Ns-butyryl-LysB29 human insulin and 5.25 parts by weight of dry powder of human insulin. After mixing equal parts of the solution containing insulin and acylated insulin and the protamine sulfate solution, an amorphous precipitate formed. This suspension was left to stand for about 24 hours at ambient temperature (typically about 22 °C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 8

Co-crystalline microcrystals of protamine, zinc, β29-Νεoctanoyl-human insulin and human insulin

B29-Ns-octanoyl-LysB29 human insulin (20.1 mg) was dissolved in 1 ml of a solvent consisting of 0.1 N HCl. Human insulin (19.3 mg) was dissolved in 1 ml of a solvent consisting of

073

0.1 N HCl. Five solutions containing different ratios of B29-Nε- 94 ··

Φ · φφ • φ φ φ ♦ · • φ φ φφ ♦ φ « φ φ ♦ φ φ · φ φ φ octanoyl-LysB29 human insulin to human insulin were prepared by combining the volumes of each solution in the ratios given below.

Table 3. Volumes of human insulin and β29-Νεoctanoyl-human insulin solutions used for the preparation of precipitates and microcrystals.

Volume (μΐ) Ratio of human insulin to acylated human insulin 1:0 3:1 1:1 1:3 0:1 human insulin solution 400 300 200 100 0 solution B29-Ne-octanoyl- human insulin 0 100 200 300 400

To each of these five solutions was added 1.6 ml of a solvent consisting of 50 mM TRIS buffer, 0.1 M sodium citrate, and 10 mg/ml phenol at 7.6. To each of these five solutions was added 0.15 ml of a 15.3 mM zinc chloride solution. The pH of each of the resulting five solutions was adjusted to 7.6 with 1 N NaOH. Each of the resulting five solutions was filtered through a 0.22 micron filter, which binds low protein components. An additional solution was prepared by dissolving 3.50 mg of protamine sulfate in 10 ml of water and then filtered through a 0.22 micron filter, which binds low protein components. A volume of 1.3 ml of each of the five solutions and 1.9 ml of the protamine sulfate solution were combined, and an amorphous precipitate immediately appeared in each of the five solutions. These five solutions were left undisturbed for 24 hours at ambient temperature (approximately 22°C). This procedure ···· **

9 *

9 9 *9 9 • 9 9 · ·· «9 99

I* 9 9 9

9 99

9 9 9 9 • 9 · · • I 99

9 9 9

9 ·· resulted in the formation of a white to off-white microcrystalline solid in each of the five solutions.

Preparation 9

9:1 co-crystals of human insulin and B29-NE-octanoyl-human insulin

Dry powder of B29-NE-octanoyl-LysB29 human insulin (0.7 parts by weight) was dissolved in 100 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 7.5 parts of a 15.3 mM zinc chloride solution. A second solution was prepared by dissolving dry powder of human insulin (6.3 parts by weight) in 900 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 67.5 parts of a 15.3 mM zinc chloride solution. The acylated insulin solution and the insulin solution were combined together and stirred to ensure mixing of the two solutions. This solution was filtered through a 0.22 micron filter, which binds low protein components. The protamine solution was prepared by dissolving 7 parts by weight of protamine sulfate in 10,000 parts by volume of water and then filtering through a 0.22 micron filter, which binds low protein components. Equal volumes of the acylated insulin solution and the protamine sulfate solution were combined. An amorphous precipitate formed. This suspension was allowed to stand for approximately 24 hours at ambient temperature (typically approximately 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

φφ «· φ · » · • r φφ φ · « φ « » ·« «*

Preparation 10

3:1 co-crystals of human insulin and B29-NE-octanoyl-human insulin

Dry powder of B29-Ne-octanoyl-LysB29 human insulin (1.75 parts by weight) was dissolved in 250 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 18.75 parts of a 15.3 mM zinc chloride solution. A second solution was prepared from dry powder of human insulin (5.25 parts by weight) which was dissolved in 750 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 56.25 parts of a 15.3 mM zinc chloride solution. The acylated insulin solution and the insulin solution were combined with each other and stirred to ensure mixing of both solutions. This solution was filtered through a 0.22 micron filter, which binds low protein components. The protamine solution was prepared by dissolving 7 parts by weight of protamine sulfate in 10,000 parts by volume of water and then filtering through a 0.22 micron filter, which binds low protein components. Equal volumes of the acylated insulin solution and the protamine sulfate solution were combined. An amorphous precipitate formed. This suspension was allowed to stand for approximately 24 hours at ambient temperature (typically approximately 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 11

1:1 co-crystals of human insulin and B29-N8-octanoyl-human insulin • · • · · · · • · · · · · · · · powder of human dissolved in 500 filtered through 0.22 low protein component.

Dry powder of B29-Ne-octanoyl-LysB29 human insulin (3.5 parts by weight) was dissolved in 500 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 1.75 parts of a 15.3 mM zinc chloride solution. A second solution was prepared in which dry insulin (3.5 parts by weight) was dissolved in 500 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 37.5 parts of a 15.3 mM zinc chloride solution. The acylated insulin solution and the insulin solution were combined together and stirred to ensure mixing of the two solutions. This solution was a micron filter that binds Protamine solution was prepared by dissolving 7 parts by weight of protamine sulfate in 10,000 parts by volume of water and then filtering through a 0.22 micron filter that binds low protein components. Equal volumes of the acylated insulin solution and the protamine sulfate solution were combined. An amorphous precipitate formed. This suspension was allowed to stand for about 24 hours at ambient temperature (typically about 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 12

1:3 co-crystals of human insulin and B29-Ns-octanoyl-human insulin

Dry powder of B29-Ne-octanoyl-LysB29 human insulin (5.25 parts by weight) was dissolved in 750 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 56.25 parts of a 15.3 mM zinc chloride solution. A second solution was prepared in which dry powder of human insulin (1.75 parts by weight) was dissolved in 250 parts by volume of an aqueous solvent consisting of 50 mM TRIS, 0.1 M sodium citrate and 10 mg/ml phenol at pH 7.6. To this solution was added 18.75 parts of a 15.3 mM zinc chloride solution. The acylated insulin solution and the insulin solution were combined and stirred to ensure that the two solutions were mixed. This solution was filtered through a 0.22 micron filter, which binds low protein components. The protamine solution was prepared by dissolving 7 parts by weight of protamine sulfate in 10,000 parts by volume of water and then filtering through a 0.22 micron filter, which binds low protein components. Equal volumes of the acylated insulin solution and the protamine sulfate solution were combined. An amorphous precipitate formed. This suspension was allowed to stand for about 24 hours at ambient temperature (typically about 22°C). The amorphous precipitate transformed into a co-crystalline microcrystalline solid.

Preparation 13

Cocrystals of human insulin and B29-N8-hexanoyl-human insulin

An acidic solution of B29-Ne-hexanoyl-human insulin was prepared by dissolving 12.3 mg of B29-NE-hexanoyl-human insulin in 0.3 ml of 0.1 N HCl. An acidic solution of human insulin was prepared by dissolving 4.6 mg of human insulin (zinc crystals) in 0.1 ml of 0.1 N HCl. The two solutions were combined to a total volume of 0.4 ml. This resulting solution was stirred for approximately 5 minutes. To this resulting solution was added 0.150 ml of a 1000 ppm zinc divalent solution with stirring. A crystallization diluent was prepared containing 32 mg/ml glycerol, 50 mM tris buffer, 10 mg/ml phenol and 100 mM sodium citrate at pH 7.6. 1.6 ml of crystallization diluent was added to the insulin solution. The pH of the solution was adjusted to 7.59 using 1 N NaOH and 1 N HCl. The solution was filtered through a 0.22 micron filter, which binds low protein components. The protamine solution was prepared by dissolving 7.47 mg of protamine sulfate in 10 mL of water. Two milliliters (2 mL) of the protamine solution was added to 2 mL of the insulin solution. The resulting solution was allowed to stand for 18 hours at a controlled temperature of 25°C.

Microscopic inspection (after 18 hours) revealed that crystallization had occurred and the preparation yielded uniformly formed individual rod-shaped crystals with a mean length of approximately 3 microns.

Four milliliters (4 ml) of the crystalline preparation formed above after 18 hours were allowed to stand overnight and the crystals were allowed to completely settle. The supernatant was then removed and replaced with 4 ml of diluent consisting of 16 mg/ml glycerol, 20 mM Tris buffer, 1.6 mg/ml m-cresol, 0.65 mg/ml phenol and 40 mM sodium citrate, pH 7.6. The crystals were then resuspended and allowed to settle again. This procedure was repeated three times except that the third time the supernatant was replaced with only 3 ml of diluent.

The rate of crystal dissolution was measured by placing 0.005 ml ti··· ♦ · titi ti • ti

- 100 uniformly suspended preparation in 3 ml of Dulbecco's phosphate-buffered saline (free of calcium or magnesium) in a 1 cm square silicon cuvette at 22°C. This solution was stirred at a constant speed using a magnetic stirrer. Absorbance measurements at 320 nm were made at one minute intervals. Absorbance at 320 nm corresponds to light scattered by insoluble particles present in the aqueous suspension. Consequently, as the microcrystals dissolve, the absorbance decreases to zero. The time required to dissolve 0.005 ml of this preparation was greater than 150 minutes. The time required to dissolve 0.005 ml of a U100 sample of commercial Humulin N, exposed to the same conditions, was approximately 10 minutes.

The amount of total protein in the preparation was analyzed by HPLC to determine the total potency. Total potency refers to the total concentration of human insulin and β29-Νεhexanoyl-human insulin. An aliquot (0.050 ml) of the fully resuspended preparation was dissolved in 0.950 ml of 0.01 M HCl and subjected to HPLC analysis as described below. The total potency determined by this analysis was 4.54 mg/ml.

The following conditions were used for HPLC analysis: C8 reversed-phase column; constant temperature 23 °C; 1.0 ml/min, detection at 214 nm; solvent A=10% acetonitrile (v/v) in 0.1% aqueous trifluoroacetic acid; solvent B=90% acetonitrile (v/v) in 0.1% aqueous trifluoroacetic acid; linear gradients (0.1 min, 0%B; 45.1 min, 75%B; 50.1 min, 100%B; 55 min 100%B; 57 min, 0%B; 72 min, 0%B). Standards were prepared by dissolving insulin and acylated insulin in 0.01 N HCl. The concentration of each

4 < ·

- 101 • •44 • 4

The standard was determined by UV spectroscopy. A solution of 1.0 mg/mL human insulin in a 1 cm cuvette was assumed to have an absorbance of 1.05 optical density units at the maximum wavelength (approximately 276 nm). This corresponds to a molar extinction coefficient of 6098. Acylated insulins were assumed to have the same molar extinction coefficient as human insulin. The UV-calibrated solutions were then diluted to obtain standards at 0.220, 0.147, 0.073, and 0.022 mg/mL. The standards were analyzed by HPLC and a standard curve of area versus concentration was obtained.

The supernatant was analyzed to determine the total concentration of soluble human insulin and B29-Ns-hexanoyl-human insulin present in the preparation. To 0.040 ml of the supernatant was added 0.160 ml of 0.01 N HCl. The acidified supernatant was analyzed by HPLC as described above. The concentration of soluble human insulin and B29-Ns-hexanoyl-human insulin in the supernatant was determined to be 0.07 mg/ml.

The ratios of the amounts of B29-Nε-hexanoyl-human insulin and human insulin in the crystals were determined by sedimenting an aliquot (0.100 ml) of the preparation using a benchtop centrifuge, settling the supernatant, resuspending the crystals in 0.400 ml of Dulbecco's phosphate-buffered saline, centrifuging again, removing the supernatant, and finally dissolving the crystals in 1.50 ml of 0.01 N HCl. HPLC analysis was performed as described above. The result of this analysis was 84.2% B29-Nεhexanoyl-human insulin and 15.8% human insulin.

• ·

- 102

9 9 * « · · 9 • 9 »99 • 9 · · • 9 »

9 <

«9 · • 9 9 »

Preparation 14

Co-crystalline suspension preparation containing human insulin and B29-Ns-decanoyl-human insulin

An acidic solution of B29-Ne-decanoyl-human insulin was prepared by dissolving 10.4 mg of B29-Ns-decanoyl-human insulin in 0.25 ml of 0.1 N HCl. An acidic solution of human insulin was prepared by dissolving 30.3 mg of human insulin (zinc crystals) in 0.75 ml of 0.1 N HCl. The two solutions were combined to a total volume of 1 ml. This resulting solution was stirred for approximately 5 minutes. To this solution was added 0.305 ml of a 1000 ppm zinc divalent solution with stirring. To the resulting solution was added 4 ml of crystallization diluent (40 mg/ml glycerol, 50 mM tris buffer, 4 mg/ml m-cresol, 1.625 mg/ml phenol, 100 mM sodium citrate, pH 7.4). The pH of the resulting solution was adjusted to 7.58. This solution was filtered through a 0.22 micron filter, which binds low protein components. Five milliliters (5 mL) of protamine solution (37.6 mg protamine sulfate in 50 mL water) was added to 5 mL of the filtered solution. The resulting solution was allowed to stand for 63 hours at a controlled temperature of 25°C.

Microscopic inspection (after 63 hours) revealed that crystallization had occurred and that the preparation yielded uniform single rod-shaped crystals with an approximate mean length of 8 microns.

The rate of crystal dissolution was measured by placing 0.006 ml of uniformly suspended crystal preparation into 3 ml of Dulbecco's phosphate-buffered saline (without

calcium or magnesium content) in a 1 cm square silicon cuvette at 22°C. The time required to dissolve 0.006 ml of this crystalline preparation was greater than 300 minutes. The time required to dissolve 0.005 ml of a U100 sample of commercial Humulin N for dissolution under the same conditions was approximately 10 minutes.

For HPLC analysis, the crystals were sedimented by allowing the preparation to stand still. Eight milliliters (8 ml) of the supernatant was then removed and replaced with 8 ml of diluent [16 mg/ml glycerol, 20 mM Tris buffer, 1.6 mg/ml mcresol, 0.65 mg/ml phenol, and 40 mM sodium citrate, pH 7.6]. The co-crystals were then resuspended. This procedure was repeated three times, except that the third time, 8 ml of the supernatant was replaced with 7 ml of diluent.

The potency of the crystalline preparation and supernatant was analyzed by HPLC, essentially as described in Preparation

13. The total potency determined by this analysis was 3.87 mg/ml. The concentration of soluble human insulin and B29-Nε-decanoyl-human insulin in the supernatant was determined to be 0.06 mg/ml. The ratios of human insulin and B29-Nε-decanoyl-human insulin in the crystalline phase were determined by the method of Preparation 13 to be 74.3% human insulin and 25.7% B29-Nε-decanoyl-human insulin.

Particle size determination was performed on a sample of the formulation using a particle sizer (Multisizer Model IIE, Coulter Corp., Miami, FL 33116-9015). To perform this measurement, 0.25 ml of the crystalline formulation was added to 100 ml of diluent consisting of 14 mM dibasic ·· ► 4 4 * • 4 ·· ·

- 104 • · * ♦ · ’ • 4 · · * ·♦ ·· * 4 4 · · • 4«

4 4

9 sodium phosphate, 16 mM glycerol, 1.6 mg/ml m-cresol and 0.65 mg/ml phenol, pH 7.4. The nozzle size of the apparatus was 50 microns. Particle size data were collected for 50 seconds. This measurement showed that the mean particle diameter of the crystals was approximately 6 microns with an approximately normal distribution, in which particles of size from approximately 2 microns to approximately 9 microns fell. This result is similar to the particle size distribution of commercial NPH determined using an analogous method [DeFelippis, M. R. et al. J. Pharmaceut. Sci. 87:170-176 (1998)].

Preparation 15

Co-crystalline suspension preparation containing human insulin and B29-NE-octanoyl-human insulin.

An acidic solution of B29-Ns-octanoyl-human insulin was prepared by dissolving 30.3 mg of B29-NE-octanoyl-human insulin in 0.75 mL of 0.1 N HCl. An acidic solution of human insulin was prepared by dissolving 59.7 mg of human insulin (zinc crystals) in 1.5 mL of 0.1 N HCl. An aliquot (0.25 mL) of the human insulin solution was combined with a 0.75 mL solution of B29-NE-octanoyl-human insulin to give a volume of 1 mL, which was stirred for approximately 5 minutes. To this solution was added 0.365 mL of a 1000 ppm zinc divalent solution while stirring. 4 ml of crystallization diluent (40 mg/ml glycerol, 35 mM dibasic sodium phosphate buffer, 4 mg/ml mcresol, 1.625 mg/ml phenol, 15 mM sodium citrate, pH 7.4) was added to the insulin-zinc solution.

filtered through a 0.22 micron filter that binds low protein components. Five milliliters (5 mL) of protamine ♦ · ·· ♦ ·

105

9 · • · · ♦ ♦ · · ·

4· solution (37.9 mg protamine sulfate in 50 ml water) was added to 5 ml of filtered insulin-zinc solution. The resulting solution was allowed to stand for 48 hours at a controlled temperature of 25°C.

Microscopic inspection (after 48 hours) revealed that crystallization had occurred and that the preparation yielded uniform single rod-shaped crystals with an approximate mean length of 5 microns.

For HPLC analysis and dissolution testing, the crystals were sedimented by allowing the preparation to stand still. Eight milliliters (8 ml) of the supernatant was then removed and replaced with 8 ml of diluent [16 mg/ml glycerol, 14 mM dibasic sodium phosphate buffer, 1.6 mg/ml m-cresol, 0.65 mg/ml phenol, 6 mM sodium citrate, pH 7.6]. The crystals were then resuspended. This procedure was performed in the same manner three times, except that the third time, 8 ml of the supernatant was replaced with 7 ml of diluent.

The dissolution rate was determined essentially as described in Preparation 13 above. The mean time for dissolution of 0.005 ml of the preparation was over 300 minutes. The time for dissolution of 0.005 ml of a U100 sample of commercial Humulin N under the same conditions was approximately 10 minutes.

The total potency and the potency in the supernatant were determined by HPLC, essentially as described in Preparation 13. The total potency was 3.44 mg/ml. The concentration of soluble human insulin and B29-Ne-octanoyl-human insulin in the crystalline preparation was determined to be 0.01 mg/ml. The ratio of the amount of human insulin to B29-Ne-octanoyl-human insulin in the crystalline preparation was determined to be 0.01 mg/ml.

0 0

0«

0 0 • 0 0

0 ·

- 106 ···· ·0 0 ·

00 0 0 · 0 0 0 0 0 ♦·

0 0 * · · • 0 0 0 0

00 ·· phase were determined essentially in the manner of Preparation 13 as 25.5% human insulin and 74.5% B29-Ns-octanoyl-human insulin.

The mean particle diameter of the crystals, determined as described in Preparation 14, was approximately 6 microns, with an approximately normal distribution ranging from approximately 2 microns to approximately 12 microns. This result is similar to the particle size distribution of commercial NPH as described in DeFelippis, M. R. et al. supra.

Preparation 16

Three co-crystalline preparations compared to an insulin preparation

An acidic solution of B29-Ns-octanoyl-human insulin was prepared by dissolving 24.18 mg of B29-N8-octanoyl-human insulin in 0.6 mL of 0.1 N HCl. An acidic solution of human insulin was prepared by dissolving 41.1 mg of human insulin (as zinc crystals) in 1 mL of 0.1 N HCl. Four 0.4 mL solutions were prepared by combining different volumes of B29-Ns-octanoyl-human insulin and human insulin solution as shown in Table 4 below.

Table 4. Preparation of microcrystal preparations

Preparation D C B A Nominal weight percentage B29-Ne-octanoyl-human insulin 75 50 25 0

4 ·*»· ·· • · · « · · • · ·

9 9 4

4 9 4

4 4 4 ·· ·

44

107

Volume of B29-NE-octanoyl- human insulin solution added (μΐ) 300 200 100 0 Volume of human insulin solution added (μΐ) 100 200 300 400

To each of the four 0.4 mL solutions was added 0.15 mL of 1000 ppm zinc divalent solution. To each of the four 0.55 mL solutions was added 1.6 mL of crystallization diluent (50 mM Tris buffer, 10 mg/mL phenol, 100 mM sodium citrate, pH 7.6). Each of the four solutions was adjusted to pH 7.6 with a small amount of 1 N NaOH and 0.1 N HCl. Each solution was filtered through a 0.22 micron filter, which binds low protein components. Two milliliters (2 mL) of each of the four protein solutions were combined with 2 mL of protamine solution (7.34 mg protamine sulfate in 10 mL water). In each case, a precipitate formed immediately. The four 4 mL suspensions were allowed to stand at ambient temperature (approximately 22°C) for 16 hours.

Microscopic inspection (after 16 hours) revealed that each of the four preparations formed uniform, single, rod-shaped crystals with an approximate mean length of about 10 microns.

Each 4 ml preparation was transferred to a test tube and centrifuged in a benchtop centrifuge at 3000 rpm for 20 minutes to completely sediment the crystals. 3 ml of supernatant was removed from each preparation and replaced with 3 ml of diluent (25 mM Tris buffer, 5 mg/ml phenol, 16 mg/ml glycerol, pH 7.4). The crystals were then resuspended. This procedure was performed three times with the same

except that the third time, 3 ml of supernatant was replaced with 2.5 ml of diluent for each preparation.

Each of the four preparations was analyzed by HPLC to quantitatively determine the total potencies of the preparations and the composition of the individual crystals, essentially as described above. Total potency refers to the total concentration of human insulin and B29-N8-octanoyl-human insulin. Total potency and percentage of B29-N8-octanoyl-human insulin were determined by analyzing an aliquot of the uniformly suspended preparation. The supernatant was analyzed to determine the total concentration of soluble human insulin and soluble B29-N8-octanoyl-human insulin present in each preparation. The results of these analyses are given below. Dissolution times were determined as described above in Preparation 13.

109

Table 5. Properties of microcrystal preparations.

Preparation D C B A NPH B29-Ns-octanoyl-human insulin in crystals (%) 77.7 51.4 23, 7 0 Human insulin crystals (%) 22.3 48.6 76.3 100 Total Potency (mg/ml) 3.21 3.48 3.38 3.43 — Supernatant Potency (mg/ml) <0.01 <0.01 <0.01 <0.01 Dissolution time (min) 300 120 50 20 10

Preparation 17

Preparation of insoluble compositions

The following preparation is an outline of another method that was used to prepare precipitates and microcrystals according to the present invention. It should be read in conjunction with the data in Table 6 below.

A measured amount of derivatized protein, prepared as described above, was dissolved in 0.6 ml of 0.1 N HCl. A measured amount of protein was dissolved in 0.2 ml of 0.1 N HCl (zinc crystals of human insulin or LysB28,Pro29analog of human insulin). Both solutions were thoroughly mixed with each other by stirring for five to ten minutes. A volume (0.32 ml) of an aqueous solution containing 1000 ppm divalent zinc and a volume (3.2 ml) of a diluent solution (approximately 50 mM Tris reagent, approximately 10 mg/ml phenol, approximately 16 mg/ml

9999 ·» • 9

110 •

• 9 · • 9 9 · ·♦

99

9 9 · ··

9 9

9 9

99

9 9

9 9

9 ·

9 « glycerol and approximately 29.5 mg/ml sodium citrate) were added to the mixture of both proteins. The pH of the resulting solution was adjusted to approximately 7.6 (7.55-7.64) using 1 N HCl or 1 N NaOH.

The pH-adjusted solution was filtered through a 0.22 micron filter, which binds low protein components. Four milliliters of a solution of protamine in water (approximately 37.3 mg protamine sulfate per 100 ml, range 37.18-37.48) was added to four milliliters of the filtrate. A precipitate formed immediately after the addition of the protamine solution. The preparation was allowed to stand at 25°C. Dissolution tests were performed as described above.

Under the same conditions, NPH insulin dissolved in about 6 minutes.

Table 6. Preparation of insoluble compositions.

Protein Lys (B28),Pro(B29)-human insulin analog Protein weight (mg) 4.28 4.02 3.84 3, 96 4.15 Derivatized protein B29acylated-human insulin Derivative group butanoyl pentanoyl hexanoyl nonanoyl decanoyl Weight of derivatized protein (mg) 11, 90 12.1 12.08 12.13 12.20 Crystal shape rod-shaped rod-shaped rod-shaped rod-shaped rod-shaped Yield (%) >80 >90 >90 >90 >90 Derivatized protein in insoluble phase (%) 77.5 75.1 77 76.5 76, 6 Dissolution time (min) 23-24 31 54 67 37-38

44

4 4 ·

4 4 *

4 4 4

4 4 4

44 ···· 44

- 111 • 4 44 ·· • * · * • · · · • 4 4 *

4 4 4 «4 44

Protein human insulin Protein weight (mg) 12.09 12.18 12.12 12.21 12.27 Derivatized protein B28-acylated-Lys(B28),Pro(B29)- human insulin analog AI,B2 8-diacylated- Lys(B28),Pro(B29)- human insulin Derivative group butanoyl hexanoyl octanoyl dibutanoyl dihexanoyl Weight of derivatized protein (mg) 4.39 4.21 4.28 4.11 4.23 Crystal shape rod-shaped rod-shaped rod-shaped rod-shaped rod-shaped Yield (%) >80 >90 >90 >90 >90 Derivatized protein in the insoluble phase (%) 27, 6 25 27 26.8 24.1 Dissolution time (min) 16-17 10-11 27-28 10-11 20-21

-087-

Protein human insulin Protein weight (mg) 4.26 4.26 4.13 4.23 12.09 Derivatized protein B29acylated-human insulin Derivative group butanoyl pentanoyl hexanoyl nonanoyl tetradecanoyl Weight of derivatized protein (mg) 12.39 12, 39 12.03 12.06 4.16 Crystal shape rod-shaped rod-shaped rod th rod th rod-shaped Yield (%) >90 72 >90 >90 >90 Derivatized protein in the insoluble phase (%) 76.3 76.5 75 72.5 25.07 Dissolution time (min) 45-46 62-63 77-78 77-78 61

φφ ♦· β φ · φ » · • φ · • ·« φ · *

- 112 ··«· · ·«

Protein human insulin Protein weight (mg) 12.19 Derivatized protein B29acylated-human insulin Derivative group hexadecanoyl Weight of derivatized protein (mg) 3, 99 Crystal shape rod-shaped Yield (%) >90 Derivatized protein in insoluble phase (%) 24.7 Dissolution time (min) 71-72

The following preparation is an outline of another method that was used to prepare precipitates and microcrystals according to the present invention. It should be read in conjunction with the data in Table 7 below.

A measured amount of derivatized protein, prepared as a 10 mg/ml solution of 29.5 mg/ml as described above, was dissolved in 3.2 ml of diluent (approximately 50 mM Tris reagent, approximately phenol, approximately 16 mg/ml glycerol, and approximately sodium citrate). A measured amount of protein was dissolved in 0.6 ml of 0.1 N HCl (zinc crystals of human insulin or LysB28, Pro29-analog of human insulin). The two solutions were thoroughly mixed with each other by stirring for five to ten minutes. The pH of the resulting solution was adjusted to approximately 7.6 (7.55-7.64) using 1 N HCl or 1 N NaOH. The pH-adjusted solution was filtered through a 0.22 micron filter, which binds low-protein components. An equal volume of protamine solution in water (approximately 37.3 mg protamine sulfate per 100 ml, range 37.18-37.48) was added to the filtrate volume.

• · • · ♦ · ··

113

A precipitate formed immediately after the addition of the protamine solution. The preparation was allowed to stand at 25°C. Dissolution tests were performed as described above. Under the same conditions, NPH insulin dissolved after approximately 6 minutes.

Table 7. Preparation of insoluble compositions.

Protein human insulin Weight of protein 12.13 11, 87 12.15 12.28 12.22 Derivatized protein AI,B29-diacyl-human insulin A1,B28- diacyl- LysB28, ProB29 human insulin B29acyl- AspB28 analogue of human insulin Derivative i ci group dioctanoyl dinonanoyl didecanoyl dioctanoyl octanoyl Weight of derivatized protein 4.29 4.07 4.06 4.29 3, 98 Crystal shape rod-shaped rod-shaped rod-shaped rod-shaped rod-shaped Yield (%) >90 >90 >90 >90 >90 Derivatized protein in the insoluble phase (%) 25.2 24.6 26, 6 27.4 24.4 Dissolution time (min) 73-74 25 31-32 47-48 43-44

The following preparation is an outline of another method that was used to prepare precipitates and microcrystals according to

of the present invention. It should be read in conjunction with the data in Table 8 below.

A measured amount of derivatized protein, prepared as described above, was dissolved in a measured volume of 0.1 N HCl. A measured amount of protein was dissolved in a measured volume of 0.1 N HCl (zinc crystals of human insulin or LysB28,Pro29-analog of human insulin). The measured volumes of both solutions were thoroughly mixed with each other by stirring for five to ten minutes. Measured volumes of an aqueous solution containing 1000 ppm divalent zinc and a diluent solution (approximately 50 mM Tris reagent, approximately 10 mg/ml phenol, approximately 32 mg/ml glycerol and approximately 30 mg/ml sodium citrate dihydrate, pH 8.47) were added to the mixture of both proteins. The pH of the resulting solution was adjusted to approximately 7.6 (7.58-7.63) using 1 N HCl or 1 N NaOH. The pH-adjusted solution was filtered through a 0.22 micron filter, which binds low protein components. Two milliliters of a solution of protamine in water (approximately 37.5 mg protamine sulfate per 100 ml) were added to two milliliters of the filtrate. A precipitate formed immediately after the addition of the protamine solution. The preparation was allowed to stand at 25°C. Dissolution tests were performed as described above. Under the same conditions, NPH insulin dissolved in approximately 6 minutes.

- 115 • ·

• · · · • · β ·

Table 8. Preparation of insoluble compositions.

Protein human insulin Protein weight (mg) 11.3 11.3 33, 6 33.6 16.5 Volume of 0.1 N HC1 0.57 0.57 1.68 1.68 0.83 Derivatized protein B29acylated-human insulin Derivative group 2-methyl- hexanoyl 2-ethyl- hexanoyl 4-methyl- octanoyl 3-methyl- decanoyl dodecanoyl Weight of derivatized protein (mg) 6, 07 6.3 6.12 2.12 6.3 Volume 0.1 N HCl (ml) 0.3 0.3 0.3 0.1 0.3 ml of protein solution combined with ml of derivatized protein solution 0.10 + 0.30 0.10 + 0.30 0.10 + 0.30 0.30 + 0.10 0.10 + 0.30 ml 1000 ppm zinc 0.152 0.152 0.112 0.096 0.152 ml of diluent 1.6 1.6 1.6 1.6 1.6 Crystal shape rod-shaped rod-shaped small irregular rod-shaped rod-shaped Yield (%) >90 >90 >80 Derivatized protein in the insoluble phase (%) 74.1 75.5 81, 9 Dissolution time (min) 116 236 40

• 9 ·

- 116

9 9 9 ·· ·· · · • · 9 9 · 9 ♦ 9 · · · ·· * 9 9 9 9 9 >1 9 9 9 9 9 9

Protein human insulin Protein weight (mg) 22.3 22.3 22.3 22, 3 Volume of 0.1 N HC1 1.12 1, 12 1.12 1.12 Derivatized protein B29acylated- rabbit insulin B29acylated- porcine insulin B2 9acylated- sheep insulin B2 9acylated- bovine insulin Derivatizing group octanoyl octanoyl hexanoyl hexanoyl Weight of derivatized protein (mg) 6, 21 6.07 6.07 6.21 ml protein solution combined with ml derivatized protein solution 0.1 + 0.3 0.1 + 0.3 0.1 + 0.3 0.1 + 0.3 ml 1000 ppm zinc 0.152 0.152 0.152 0.152 ml of added diluent 1.6 1.6 1.6 1.6 Crystal shape rod-shaped rod-shaped small irregular rod-shaped Yield (%) >90 >90 >90 >90 Derivatized protein in the insoluble phase (%) 74.6 74.7 75.4 75, 9 Dissolution Time (min) 111 >300 227 >300

• those

- 117 * · · • ti ti· * · · · <· ·· • · · ti ti ·· ti tititi· · ti ti · • titi tititi titi · ti ti· · · ·· · • ti »ti ·· titi

Protein human insulin Protein Weight (mg) 33, 6 2.4 22, 6 Volume of 0.1 N HC1 1, 68 0.1 1, 12 Derivatized protein B29acylated- Gly(A21),Arg(B31),Arg(B32)-analog of human insulin B29acylated- human insulin B29acylated- desThr(B30)- human insulin analog Derivatizing group decanoyl 1,4-dichloro- phenyl-thioacetyl octanoyl Weight of derivatized protein (mg) 2.23 5.1 6, 06 Volume 0.1 N HC1 0.1 0.3 0.3 ml of protein solution combined with ml of derivatized protein solution 0.3 + 0.1 0.1 + 0.3 0.1 + 0.3 ml 1000 ppm zinc 0.096 0.152 0.152 ml of added diluent 1.6* 1.6 1.6 Crystal shape rod-shaped small irregular crystals rod-shaped Yield (%) >90 >90 >90 Derivatized protein in the insoluble phase (%) 20, 4 71.8 74.2 Dissolution time (min) 63 66 >300

• ·

- 118 • · · 9 9 » • · • 9 • 9 * • 9 · · * 2.096 ml of 37.5 mg/100 ml protamine sulfate solution was added before the diluent was added. The pH was adjusted after the diluent was added.

Preparation 18

Preparation of amorphous suspension

A measured amount (13.84 mg protein) of solid B28-tetradecanoyl-Lys(B28),Pro(B29) human insulin analog was dissolved in 0.375 mL of 0.1 N HCl. A measured amount of zinc human insulin (7.40 mg protein) was dissolved in 207 microliters of 0.1 N HCl. An aliquot (125 μΐ) of insulin solution (containing 4.47 mg human insulin) was added to the B28-tetradecanoyl-Lys(B28),Pro(B29) human insulin analog solution. A volume (180 μΐ) of 1000 ppm zinc and 2.0 ml of diluent (1.6 mg/ml phenol, 4 mg/ml mcresol, 40 mg/ml glycerol, 5 mg/ml anhydrous dibasic phosphate, 7.5 mg/ml disodium phosphate dihydrate, pH 7.6) were added. The pH was raised from 5.6 to 8.0 with 100 microliters of 1N NaOH and lowered back to 7.59 with 20 microliters of 1N HCl and 1N NaOH. The concentration of B28-tetradecanoyl-Lys(B28),Pro(B29) human insulin analog was 4.94 mg/ml and the concentration of human insulin was 1.60 mg/ml. The solution was passed through a 0.22 micron filter that binds low protein components and refrigerated overnight. The following morning, no precipitate was present in the solution. To 2.50 mL of the solution was added 2.88 mL of protamine solution (0.75 mg/mL solid protamine sulfate dissolved in water). An amorphous precipitate formed upon addition of protamine.

After the addition of protamine, the concentration of B28 was determined again.

- 119 • Φ · ♦ φ φ φφ β · · · φ φφφφ • φφφφ φφφφ φφφ «φφ φφ φ φφφφ φφφφ φφ φφ φφ φφ φφ φφ φφ φφ φφ tetradecanoyl-Lys(Β28),Pro(Β29) human insulin analog and human insulin in the soluble phase. Samples for HPLC analysis were prepared immediately after the addition of protamine. From the retention times of the peaks, HPLC analysis showed that the insoluble material in the suspension contained protamine, B28-tetradecanoylLys(B28) , Pro(B29) human insulin analog and human insulin. The concentration of B28-tetradecanoyl-Lys(B28),Pro(B29) human insulin analog in the soluble phase was 2.30 mg/ml and the concentration of human insulin was 0.74 mg/ml.

The concentrations of B28-tetradecanoyl-Lys(B28),Pro(B29) human insulin analogue and human insulin in acidified suspension, supernatant and precipitate samples were determined and are shown in the following table. These values are in reasonable agreement with the expected values. Protamine concentrations were not determined.

Table 9. Preparation of insoluble compositions.

Concentration (mg/ml) Weight (%) Sample B28-tetradecanoyl- Lys (B28),Pro(B29) human insulin analog human insulin Suspension 2.52 0.79 76, 2 Supernatant 0.006 0.057 8.9 Precipitate 2.29 0.60 79.2

• β 4 · «4 ·· 4 4 · · ·· · « 4 * » · ♦ · · 4

4 4 « · · · 4 » · *

4 ··· 4 4 444 44 4

44 4 4 44 4 44 4

- 120 - ............

Preparation 19

Preparation of amorphous suspensions

The following preparation is an outline of another method that was used to prepare precipitates according to the present invention. The method was used to prepare amorphous insulin precipitate preparations from each of the following three derivatized proteins: B29-N8-octanoyl-human insulin; B29-NE-nonanoyl-human insulin; and B28-Ns-octanoylLysB28,ProB29-human insulin analog.

A measured amount of solid derivatized protein was dissolved in 3 mL of 0.1 N HCl to form a solution containing approximately 16 mg/mL of derivatized protein. A measured amount of zinc crystals of human insulin (73 mg, of which 67.17 mg was protein) was dissolved in 4.198 mL of 0.1 N HCl to form a solution containing approximately 16 mg/mL of insulin. Three mL of derivatized protein solution and one mL of insulin solution were combined and mixed thoroughly. Measured volumes of 1000 ppm zinc solution (1.137 mL) and diluent (16 mL containing per mL: 1.625 mg phenol, 4 mg m-cresol, 40 mg glycerol, 5 mg anhydrous dibasic sodium phosphate, 7.5 mg sodium citrate dihydrate, pH 7.6) were added. The pH was adjusted to approximately 7.6 (7.58-7.61) using 5 N NaOH and 5 N HCl solutions. The volume added during pH adjustment was 0.11 to 0.12 ml. The solution was filtered through a 0.22 micron filter, which binds low protein components, and refrigerated overnight. The following morning, no precipitate was present in the solution. The solution consisted of protein and derivatized protein (approximately 1:3 weight ratio) and the total protein concentration was

121 ···· ·· »· · · • · * · ♦ · • · * · · · · ♦ · · · ♦ · · «· · · · · · * · · · · · · · · « · · · • · · · · equivalent to approximately 85 units per milliliter. Just before testing in rats, equal volumes of the solution and protamine sulfate solution (0.352 mg/mL) were combined and mixed thoroughly. An amorphous precipitate formed immediately. A sample of the suspension preparation containing the amorphous precipitate was immediately injected into the test animals. After mixing with protamine, the total protein concentration was approximately 42.4 units/mL.

Preparation 20

Gly(A21),Arg(B31),Arg(B32)-human insulin analogue

Gly(A21)Arg(B31)Arg(B32)-human insulin was obtained by fermentation of E. coli in which the Gly(A21)-human proinsulin precursor molecule was overexpressed into inclusion bodies. A portion (94.7 g) of the inclusion bodies was solubilized in 500 ml of 6 M guanidine hydrochloride containing 0.1 M TRIS, 0.27 M sodium sulfite and 0.1 M sodium tetrathionate, pH 10.5 at ambient temperature. The pH was rapidly lowered to 8.8 with 12 N HCl. After vigorous stirring in an open vessel for 45 min, the pH was lowered to 2.1 with phosphoric acid and the sample was centrifuged overnight at 4 °C. The supernatant was decanted and stored at 4 °C for further processing. The pellet was re-extracted with 200 ml of additional pH 10.5 solution described above (see above) and then centrifuged for 3 hours at 4°C. This and the previously obtained supernatant were each diluted 4X with 100 mM sodium phosphate, pH 4, thereby precipitating the product and other acidic components. The precipitate was then allowed to settle, most of the supernatant was decanted and discarded. The resulting suspension was centrifuged, followed by decantation and discarding.

122

00*0 ** *0 00 ·0 ·0 • < « · «» · · 00 «

000 00 00 000· • · · · 0 00 000 00 0 * 00 0 0 00 0 ·00 0

00 00 00 0· ·0 of the additional supernatant was removed, leaving a wet pellet of the crude Gly(A21)-human proinsulin S-sulfonate precursor. The pellets were solubilized in 1.5 liters of 7 M deionized urea, the pH was adjusted to 8 with 5 N NaOH, and the mixture was stirred for several hours at 4 °C. Salt (NaCl) was then added to achieve a 1 M concentration and the sample was loaded onto an XAD-7 column (14 cm X 20 cm, Toso-Haas,

Montgomeryville, PA) which had previously been rinsed with mixtures of 50% acetonitrile/50% 50 mM ammonium bicarbonate, 10% acetonitrile/90% 50 mM ammonium bicarbonate, and finally 7 M deionized urea/1M NaCl/20 mM TRIS, pH 8. Once the column was loaded, 4.5 liters of 7 M deionized urea/1 M NaCl/20 mM TRIS, pH 8 solution was pumped into it, followed by 2.8 liters of 50 mM ammonium bicarbonate/1 M NaCl, and 6.5 liters of 50 mM ammonium bicarbonate. The column was eluted with a linear gradient of acetonitrile in 50 mM ammonium bicarbonate while the eluate was monitored by UV at 280 nm. The desired peak, namely the partially purified Gly(A21) human proinsulin S-sulfonate precursor, was collected, lyophilized and subjected to the following shaping and disulfide bond formation procedure. A portion (5.4 g) of the precursor was dissolved in 3 liters of 20 mM glycine, pH 10.5, 4°C. Then, 15 ml of 240 mM cysteine HCl was added with stirring, maintaining the pH at 10.5 and the temperature at 4°C. The reaction solution was gently stirred at 4°C for 27 hours and then the reaction was stopped by lowering the pH to 3.1 by adding phosphoric acid. Acetonitrile (155 mL) was added and the solution was then loaded onto a 5 x 25 cm C4 reversed phase column which was flushed with 60% acetonitrile/40% water/0.1 L TFA and equilibrated in 10% acetonitrile/90% water/0.1% TFA. Once

- 123 ···· »* • « · • · * ·· ·· » « · 9 » · 9 4

I · 9 4» » · · · · · ·· the column was filled, 1 liter of 17.5% acetonitrile/82.5% water/0.1% TFA was pumped into it, then it was eluted with a linear gradient of acetonitrile in 0.1% TFA while monitoring at 280 nm. Selected fractions were collected and lyophilized, yielding 714 mg. To convert the proinsulin precursor to the desired insulin analog, 697 mg of human proinsulin Gly(A21) precursor was dissolved in 70 ml of 50 mM ammonium bicarbonate, then cooled to 4 °C, pH 8.3. A portion (0.14 mL) of a 1 mg/mL solution of porcine trypsin (Sigma Chemical Company, St. Louis, MO) in 0.01 N HCl was added to the sample solution, which was gently stirred at 4°C for approximately 24 hours. An additional 0.14 mL of the trypsin solution was then added to the reaction solution, which was then stirred for an additional 21 hours and 45 minutes. The reaction was stopped by lowering the pH to 3.2 by adding 0.7 mL of glacial acetic acid and 0.3 mL of phosphoric acid. The resulting Gly(A21)Arg(B31)Arg(B32)-human insulin solution sample from the trypsin digestion reaction was diluted 4X with 30% acetonitrile/70% 50 mM acetic acid, pH 3.1 and loaded onto a 1 x 30 cm S HyperD F cation exchange column (Biosepra, Marlborough, MA) pre-pumped with 30% acetonitrile/70% 50 mM acetic acid/500 mM NaCl, pH 3.3 and equilibrated in 30% acetonitrile/70% 50 mM acetic acid. After loading, approximately 50 mL of 30% acetonitrile/70% 50 mM acetic acid was pumped into the column, then eluted with a linear NaCl gradient in 30% acetonitrile/50 mM acetic acid while monitoring the eluate at 276 nm. Selected fractions containing Gly(A21)Arg(B31)Arg(B32)-human insulin were collected, diluted 3X with purified water, and loaded onto a 2.2 cm C4 reversed-phase column (Vydac, Hesperia, CA)

- 124 φ · · · φφ φ φ

ΦΦ ► « φ · ·♦ ·· φ φ φ φ φ φ φ φ • · φ φ φ φ φ φ φ pre-filled with a mixture of 60% acetonitrile/40% water/0.1% TFA and then 10% acetonitrile/90% water/0.1% TFA. After filling, approximately 200 ml of a mixture of 10% acetonitrile/90% water/0.1% TFA was pumped into the column, then it was eluted with a linear gradient of acetonitrile in 0.1% TFA. Selected fractions were collected and lyophilized to yield 101 mg. Analytical HPLC demonstrated a purity of greater than 95% for the main peak. Analysis of the purified protein by electrospray mass spectroscopy (ESMS) determined a molecular weight of 6062.9 (theory 6063.0).

Preparation 21

Des(B30)-human insulin

Des(B30)-human insulin was prepared from human proinsulin by severe trypsin hydrolysis. A portion (2 g) of human proinsulin, which had been produced by biological synthesis in recombinant E. coli and purified by conventional methods [Frank, BH et al., in PEPTIDES: Synthesis-StructureFunction. Proceedings of the Seventh American Peptide Symposium, Rich, DH and Gross, E. (eds.), Pierce Chemical Company, Rockford, pp. 729-738, 1981; see also Frank, BH, U.S. Patent No. 4,430,266, issued February 7, 1984; both citations are incorporated by reference] was dissolved in 400 ml of 0.1 M, pH 7.5 HEPES buffer. After adding 8 ml of 1 M CaCl ₂ (in water) and adjusting the pH to 7.5 with 5 N NaOH, 2 ml of a 10 mg/ml solution of porcine trypsin (Sigma) in 0.01 N HCl was transferred to the sample solution with gentle stirring. The reaction solution was left stirring at ambient temperature for 2 hours and 42 minutes and after this time it was transferred to a 37 °C environment with occasional stirring. After 1

- 125 99 99 > * « 1 » 9 9 <

The product was monitored containing purified for 45 hours at 37°C. The enzymatic reaction was stopped by lowering the pH to 3.0 with phosphoric acid and lowering the temperature to 4°C for storage. The solution was then brought to ambient temperature and diluted with 50 ml of acetonitrile and then to a final volume of 500 ml with purified water, then loaded onto a 2.5 x 58 cm CG-161 (Toso-Haas) column that had been previously flushed with 1 column volume of 40% acetonitrile/60% 0.1 M ammonium sulfate, pH 2.5 and 2 column volumes of 10% acetonitrile/90% 0.1 M ammonium sulfate, pH 2.5. Once the column was filled, 1 column volume of 10% acetonitrile/90% 0.1 M ammonium sulfate, pH 2.5 was pumped into the column. The column was eluted with a linear gradient of acetonitrile in 0.1 M ammonium sulfate, pH 2.5, while monitoring the eluate at 276 nm. The desired peak, partially purified des(B30)-human insulin, was obtained by collecting a selected fraction. This partially purified des(B30)-human insulin sample was diluted to 1.28 liters with purified water, pH 3.5, and loaded onto a 1 x 29 cm S HyperD F cation exchange column (Biosepra) prepacked with 1 column volume of 30% acetonitrile/70% 0.1% TFA/0.5 M NaCl, pH 1.9 and 2 column volumes of 30% acetonitrile/70% 0.1% TFA, pH 2.3. After loading, the column was filled with 1 column volume of 30% acetonitrile/70% 0.1% TFA, pH 2.3, then eluted with a linear gradient of NaCl in 30% acetonitrile/70% 0.1% TFA, pH 1.9 to 2.3, while eluting at 276 nm. Selected des(B30)-human insulin fractions were collected, diluted 2.5X with purified water, and loaded onto a 35-c.c. CS SepPak column (Waters, Milford, MA), pre-cleaned and conditioned with 2 column volumes of acetonitrile, 2 column volumes of 60% acetonitrile/40% 0.1% TFA, and 2 column volumes of • 9

9 • * ♦ 9

9 99

- 126

9 9 9 9 9 9 9 9 9 9 9

9 9 99 99 99 ·· 10% acetonitrile/90% 0.1% TFA. After loading the SepPak column, it was rinsed with 3 column volumes of 10% acetonitrile/90% 0.1% TFA and then eluted with 2 column volumes of 60% acetonitrile/40% 0.1% TFA. The lyophilized eluted product yielded 500 mg. Analytical HPLC showed a major peak with >95%. Electrospray mass spectroscopy (ESMS) of the purified protein showed a molecular weight of 5706.5 (theory 5707).

Preparation 22 Rabbit insulin

Rabbit insulin was prepared as described in Chance, R. E. et al. [Proinsulin, Insulin, C-Peptide, Baba, S. et al. (Editors), Excerpta Medica, Amsterdam-Oxford, pp. 99-105 (1979)].

Preparation 23

Asp(B28)-human insulin analogue

Asp(B28)-human insulin was prepared and purified essentially according to Examples 31 and 32 of Chance, R. E. et al. (U. S. Patent No. 5,700,662, issued December 23, 1997); the reference is expressly made to insulin [Bromer, W. W.

as a reference, and Chance, R. E.

Acta, 133:219-223 (1967), which is

Des(B23-30)-human , Biochim. Biophys. incorporated herein by reference] and the synthetic octapeptide Gly-Phe-Phe-Tyr-Thr-AspLys(Tfa)-Thr were condensed using supported semisynthesis, purified by reverse-phase gel HPLC, treated with 15% trypsin filtration and ammonium hydroxide (v/v) for four hours at ambient temperature for • ··

- 127 ·♦·· ·· • · · • · · ·· · 9 9 9 9

9 99 99 99 removal of the trifluoroacetate (Tfa) blocking group of Lys(B29), purified by reversed-phase HPLC and lyophilized.

Preparation 24

Syntheses of derivatized proteins

The following is an outline of the synthesis of certain derivatized proteins used to prepare precipitates and microcrystals according to the present invention. The text should be read in conjunction with the data in Table 10 below.

A measured amount of purified insulin or insulin analog was dissolved in a measured volume of dimethyl sulfoxide (DMSO) with stirring. A measured volume of tetramethylguanidine hydrochloride (TMG) was then added and the solution was mixed thoroughly. In a separate vessel, a measured amount of N-acylsuccinimide (NAS) was dissolved in a measured volume of DMSO. A measured volume of the second solution was added to the first solution. The reaction was carried out at ambient temperature and the progress of the reaction was monitored by analyzing samples of the reaction mixture using HPLC. The reaction was stopped with a measured volume of ethanolamine and then acidified to pH 2-3.

The reaction mixture was then purified using reversed phase chromatography alone or using a combination of cation exchange chromatography followed by reversed phase chromatography. Purification by reversed phase chromatography was performed using an FPLC system (Pharmacia) with UV detection at 214 nm or at 280 nm, fraction collector, 2.2 x 25 cm or 5 x 30 cm C18 • ·

- 128

9 9 9 99 9 9

9 9 9 · 9

9 9 9 9 • 99 9 9 99 9 9 9 99 9

99 99 99 99 99 column, flow rate 2.5 or 5 ml/min at ambient temperature. The liquid phases were mixtures of Solution A [10.1% trifluoroacetic acid (TFA) in a 10:50 acetonitrile:water (v:v)] and Solution B [0.1% trifluoroacetic acid (TFA) in a 70:30 acetonitrile:water (v:v)], which were suitable for elution and separation of the target compounds. Typically, the column was equilibrated and filled with 100% Solution A. Then, a linear gradient into a portion of Solution B was used for the corresponding separation of the reaction products. Fractions containing the product were collected. Variations in the purification methods are within the ordinary skill in the art.

Table 10 below provides experimental data, obtained according to the method outlined above, for the synthesis of derivatized proteins that were used to prepare various embodiments of the present invention. The starting proteins were prepared as described above or by conventional methods. Conventional purification methods were used to obtain highly pure starting proteins for the syntheses listed below. The synthesis of insulin, insulin analogs and proinsulin is within the scope of conventional procedures in the art and can be achieved using recombinant expression, semisynthesis or solid phase synthesis and subsequent chain combination. Purification of the synthesized proteins to the purity required for the preparation of derivatives used in the present invention is accomplished by conventional purification techniques.

The molecular weight of the purified derivative was confirmed by mass spectrometry, specifically electrospray mass analysis (ESMS). The determination of the acylation site was based either on chromatographic analysis (HPLC) or on the N-terminal

- 129

99 • 9 9 9

9 9 9 analysis (N-terminal) or both.

Table 10. Summary of the synthesis of various derivatized proteins.

Starting protein human insulin human insulin human insulin Protein Weight (mg) 141.3 1,080 120 DMSO (ml) 42 30 36 TMG (μί) 30.5 233 25, 9 NAS acylation chain n-hexanoyl n-octanoyl n-dodecanoyl NAS Weight (mg) 7.76 85.7 9.22 DMSO Volume (ml) 1.0 1.0 0.701 Volume of added NAS solution (ml) 0.494 0.785 0.701 Response time (min) 40 105 40 Volume of ethanolamine (μί) 20 100 120 Total yield (%) 40 33 36 Molecular mass (theory) 5906.0 5933.3 5990.0 Molecular mass (ESMS) 5906.8 5933.9 5990.0 HPLC purity (%) 96 94 98 Acylation site (HPLC) No No No Acylation site (N- terminal) No No No

«· ·· • * • ·« • « • *

- 130

99 99 ·· ··

99

Starting protein human insulin human insulin human insulin Protein Weight (mg) 194 2040 2050 DMSO (ml) 60 62 58 TMG (μΐ) 41, 9 441 443 NAS acylation chain n-tetra- decanoyl n-butyryl n-hexanoyl NAS Weight (mg) 23, 4 269.3 209 DMSO Volume (ml) 1.0 1.0 2.0 Volume of added NAS solution (ml) 0.756 0.29 1.44 Response time (min) 20 30 30 Volume of ethanolamine (μΐ) 5 100 100 Total yield (%) 45 27* 22 Molecular mass (theory) 6018.1 5877.8 5905.9 Molecular mass (ESMS) 6018.2 5877.8 5906.0 HPLC purity (%) 98 94 93 Acylation site (HPLC) New New New Acylation site (N- terminal) New New New

★ purification involved first reversed-phase HPLC, then cation-exchange HPLC, then reversed-phase HPLC

131 • fc«« ·* ·« fc· fcfc fcfc fcfcfc ···· fcfcfcfc fcfcfc · · fcfc · fc fc · • fc fcfcfc fcfc fc*»· » · fc fcfc·· fcfcfcfc fcfc fcfc fcfc fcfc fcfc fcfc

The following is an outline of the synthesis of additional derivatized proteins. The outline should be read in conjunction with the data in Table 11 below, which provides a complete description of the synthetic purified insulin or its dissolution by addition to the schemes. A measured dose of insulin analog was added to a measured volume of 50 mM boric acid, pH 2.57. A measured volume of acetonitrile, equal to the volume of boric acid solution, was then slowly added with stirring. The volume of solvent is the sum of the volumes of boric acid and acetonitrile. The pH of the solution was adjusted to between 10.2 and 10.5 using NaOH. In a separate vessel, a measured amount of N-acylsuccinimide (NAS) was dissolved in a measured volume of DMSO. A measured volume of the second solution was added to the first solution. The reaction was carried out at ambient temperature, the pH was maintained above 10.2 as necessary, and the progress of the reaction was monitored by analyzing samples of the reaction mixture using HPLC. The reaction was stopped by acidification to pH 2-3. The reaction mixture was then subjected to purification using reversed phase chromatography as described above.

Table 11 presents the experimental data obtained in the above manner for the synthesis of derivatized proteins which were used to prepare various embodiments of the present invention. The molecular weight of the purified derivative was confirmed by mass spectrometry, specifically electrospray mass analysis (ESMS). The determination of the acylation site was based on either chromatographic analysis (HPLC) or N-terminal analysis (N-terminal) or both.

Table 11. Overview of the synthesis of various derivatized • 4

4 44 ·· 4 4

4 4 4 4 4 4

4 4 4 4 44

- 132 • 4 4 4 44 4 4 44 ·

44 44 44 · · 44 proteins.

Starting protein human insulin human insulin human insulin Protein Weight (mg) 2170 2420 2250 Solvent (ml) 200 240 200 NAS acylation chain n-butyryl n-pentanoyl n-octanoyl Weight of N-acyl- succinimide (mg) 108.7 1155 173 DMSO Volume (ml) l<0 5 1.0 Volume of added NAS solution (ml) 0.955 0.719 0.81 Response time (min) 40 40 40 Total yield (%) 25 12 37 Molecular mass (theory) 5877.8 5891.8 5933.9 Molecular mass (ESMS) 5877.7 5891.9 5933.8 HPLC purity (%) 96 95 96 Acylation site (HPLC) New New New

- 133

ΦΦΦΦ 9 9 • 9 · · * * ·

Φ · · Φ Φ Φ · • Φ «· 99 Φ· • φ φφ • · · φ φ 9 Φ ·

Φ Φ Φ Φ

Φ Φ Φ Φ ·· ·0

Starting protein human insulin human insulin human insulin Protein Weight (mg) 1960 2750 1040 Solvent (ml) 200 200 200 NAS acylation chain n-nonanoyl n-dodecanoyl n-tetra- decanoyl Weight of N-acyl- succinimide (mg) 145.8 19, 9 10, 3 DMSO Volume (ml) 1.0 1.0 1.0 Volume of added NAS solution (ml) 0.887 0.771 0.885 Response time (min) 40 30 35 Total yield (%) 35 14 39 Molecular mass (theory) 5947.9 5990.0 6018.1 Molecular mass (ESMS) 5948.1 5989.9 6018.1 HPLC purity (%) 94 93 94 Acylation site (HPLC) New New New

- 134

44 · · ·· ·· • · · · • * · · · · « »

Starting protein sheep insulin bovine insulin porcine insulin Protein Weight (mg) 312 275 200 Solvent (ml) 100 100 100 NAS acylation chain n-hexanoyl n-hexanoyl n-octanoyl Weight of N-acyl- succinimide (mg) 27.2 19.9 16, 4 DMSO Volume (ml) 1.0 1.0 1.0 Volume of added NAS solution (ml) 0.644 0.771 0.764 Response time (min) 45 30 82 Total yield (%) 31 50 41 Molecular mass (theory) 5801.7 5831.8 5903.9 Molecular mass (ESMS) 5801.8 5831.7 5903.9 HPLC purity (%) 96 96 96 Acylation site (HPLC) No No No

• * «

- 135

Starting protein rabbit insulin des(B30)- human insulin AspB28- human insulin Protein Weight (mg) 211.4 205, 3 132.3 Solvent (ml) 100 20 20 NAS acylation chain n-octanoyl n-octanoyl n-octanoyl Weight of N-acyl- succinimide (mg) 16.8 21, 5 11, 5 DMSO Volume (ml) 1.0 0.5 1.0 Volume of added NAS solution (ml) 0.786 0.303 0.715 Response time (min) 57 40 85 Total yield (%) 39 47 32 Molecular mass (theory) 5919.9 5833.6 5951.9 Molecular mass (ESMS) 5920.0 5832.7 5952.2 HPLC purity (%) 95 96 94 Acylation site (HPLC) New New New

- 136 «· «· · * · · φ « · φ » φ · φ · • · · φ · · · φ · ·«· · » ♦ · · Φ· · • · · φ ·· · ' ·· · • Α ♦ » 4» Φ· ·» ·*

Starting protein GlyA21,ArgB3 1,ArgB32- human insulin analog human insulin des(B27)- human insulin analog Protein Weight (mg) 86.2 134.8 44.8 Solvent (ml) 10 20 7 NAS acylation chain n-octanoyl 2-methnyl- hexanoyl n-octanoyl Weight of N-acyl- succinimide (mg) 22.4 749 3, 6 DMSO Volume (ml) 0.5 4.93* 1.0 Volume of added NAS solution (ml) 0.115 0.052 0.993 Response time (min) 40 45 40 Total yield (%) 45 45 53 Molecular mass (theory) 6189.2 5919.9 5832.8 Molecular mass (ESMS) 6189.2 5919.9 5832.9 HPLC purity (%) 97 96 93 Acylation site (HPLC) New New New

* Dissolved in acetonitrile instead of DMSO.

137

Starting protein human insulin human insulin human insulin Protein Weight (mg) 160 147.7 2080 Solvent (ml) 20 20 200 NAS acylation chain 4-methyl- octanoyl 3-methnyl- decanoyl n-octanoyl Weight of N-acyl- succinimide (mg) 715 22.5 146, 7 DMSO Volume (ml) 4.97* 1.0* 1.0* Volume of added NAS solution (ml) 0.0734 0.609 0.884 Response time (min) 60 45 40 Total yield (%) 54 38 5, 3** Molecular mass (theory) 5947.9 5976.0 6060.1 Molecular mass (MS) *** 5947.8 5976.2 6060.5 HPLC purity (%) 96 96 92 Acylation site (HPLC) New New Al-Nα,Nε

* Dissolved in acetonitrile instead of DMSO.

** Yield of Al-Να,B29-Ns-diacyl-human insulin derivative *** Determined by Matrix-Assisted Laser Desorption Ionization (MALDI) mass spectroscopy instead of electrospray mass spectroscopy

- 138 «··· 4 9 9 4 9 4

4 9 4 * · • « W · · 44

4 4

Starting protein des(B30)- human insulin analog human insulin human insulin Protein Weight (mg) 205, 3 1960 2110 Solvent (ml) 20 200 200 NAS acylation chain n-octanoyl n-nonanoyl n-decanoyl Weight of N-acyl- succinimide (mg) 21.5 145.8 150, 5 DMSO Volume (ml) 0.5 1.0 1.0 Volume of added NAS solution (ml) 0.089 0.0887 0975 Response time (min) 40 40 60 Total yield (%) 11.0 11.5 11.1 Molecular mass (theory) 5959.5 6088.2 6116.2 Molecular mass (MS) 5959.3 6088.3 6116.4 HPLC purity (%) 96 92 92 Acylation site (HPLC) Α1-Να,Νε Α1-Να,Νε Α1-Να,Νε

The following is a general outline of the synthetic scheme for additional derivatized proteins. In specific instances, the outline should be read in conjunction with the data in Table 12 below, which provides a complete description of the synthetic schemes for specific derivatized proteins. A measured amount of purified insulin or insulin analog was dissolved by adding it to a measured volume of DMSO. The pH of the solution was adjusted with 10 equivalents of tetramethylguanidine.

• ·

- 139 • ti ti · titi • · « · » · ti • · titi ti ti titi • ti «· ti ti ti ti • ti ti ti • titi ti ti ti * ti • ti titi

In a separate vessel, a measured amount of N-acyl succinimide (NAS) was dissolved in a measured volume of DMSO. A measured volume of the second solution was added to the first solution to give a 1.9-fold molar excess of N-acyl succinimide. The reaction was carried out at ambient temperature and the progress of the reaction was monitored by analyzing aliquots of the reaction mixture using HPLC. The reaction was quenched with 20 microliters of ethanolamine, cooled in an ice/water bath, and diluted 2.1-fold with 0.1N HCl. The reaction mixture was then subjected to desalination on a reverse phase chromatography column using the following protocol:

1) the column was wetted with 100% acetonitrile, then washed using three or four column volumes of 0.1% TFA/70% acetonitrile (Buffer B); and finally washed using four to five column volumes of 0.1% TFA/10% acetonitrile (Buffer A);

2) diluted reaction mixtures were loaded and the column was washed again with five to six column volumes of Buffer A; and

3) The derivatized protein was eluted by passing five to six column volumes of Buffer B through the column. The liquid collected during the elution was frozen and then lyophilized. The lyophilized crude product (86.1 mg) was then subjected to repurification using reversed phase chromatography as described above.

Table 12 presents the experimental data obtained by the above method for the synthesis of derivatized proteins that were used to prepare various embodiments of the present invention. The molecular weight of the purified derivative was confirmed by mass spectrometry, namely

140 by electrospray mass spectrometry (ESMS). The determination of the acylation site was based either on chromatographic analysis (HPLC).

Table 12 proteins.

Overview of the synthesis of various derivatized

Starting protein human insulin Protein weight (mg) 179 DMSO for dissolving human insulin (ml) 20 NAS acylation chain 1,4-dichlorophenylthio- acetyl Weight of N-acyl succinimide (mg) 30 DMSO Volume (ml) 1.0 Volume of added NAS solution (ml) 0.412 Response time (min) 50 Total yield (%) 32.2* Molecular mass (theory) 6026.78 Molecular mass (ESMS) 6026, 9 HPLC purity (%) 96 Acylation site (HPLC) New

* Yield calculated based on the weight of the desalinated and lyophilized crude product.

• ·

141 «··· 99

9 «

Attempt 1

Duration of action of cocrystals in dogs

The duration of action of the three co-crystalline compositions of the present invention was determined in normal dogs receiving a constant infusion of somatostatin to produce a transient diabetic state. The first co-crystalline preparation, containing human insulin and B29-Ns-octanoyl-insulin, was prepared essentially as described for Preparation D in Preparation 16 above and was administered subcutaneously at a dose of 3 nmol/kg (8753). The second co-crystalline preparation, containing human insulin and B29-Ns-octanoyl-human insulin, was prepared as described for Preparation D in Preparation 16 above and was administered subcutaneously at a dose of 2.5 nmol/kg (8152.5). Finally, a third co-crystalline preparation, containing human insulin and B29-Ne-decanoyl-human insulin, was prepared as described in Preparation 14 above and administered subcutaneously at a dose of 2.5 nmol/kg (10252.5). The data were compared with data obtained with the same route of administration of Humulin N (2.0 nmol/kg NPH), bovine/porcine Ultralente insulin (3 nmol/kg, BP-UL), and saline.

Experiments were performed in chronically cannulated male and female beagle dogs (Marshall Farms, North Rose, NY) weighing 1017 kg without anesthesia and after an overnight fast. At least ten days before the study, the animals were anesthetized with isoflurane (Anaquest, Madison, WI), and silicone catheters connected to vascular access ports (V-A-P™, Access Technologies, Norfolk Medical, Skokie, IL) were inserted into the femoral artery and femoral vein. The catheters were filled with a glycerol/heparin solution (3:1, by volume; final

142 • fcffc fcffc

fc · · · fcfcfc · · · • ffcfcfcc heparin concentration of 250 KlU/ml; glycerol from Sigma Chemical Co., St. Louis, MO and heparin from Elkins-Sinn, lne., Cherry Hill, NJ) to prevent catheter occlusion and injury was closed. Kefzol (Eli Lilly & Co., Indianapolis, IN) was administered preoperatively (20 mg/kg, IV and 20 mg/kg, I.M.) and Keflex was administered postoperatively (250 mg, p.o. once daily for seven days) to prevent infection. Torbugesic (1.5 mg/kg, I.M.) was administered postoperatively to relieve pain.

Blood was collected just prior to the study to determine the animal's health status. Only animals with hematocrits above 38% and leukocyte counts below 16,000/mm ³ were used in the experiment (hematology analyzer: Cell-Dyn 900, Sequoia-Turner, Mountain View, CA).

On the morning of the experiment, ports (Access Technologies, Norfolk Medical, Skokie, IL) were used; the catheter contents were aspirated; the catheters were flushed with saline (Baxter Healthcare Corp., Deerfield, IL); the dog was placed in a cage; and extension lines (protected by stainless steel connectors and connected to a swivel system [Instech Laboratories, Plymouth Meeting, PA]) were connected to the port lines.

Dogs were allowed at least 10 min to acclimate to the cage environment before arterial blood samples were collected for fasting insulin, glucose, and glucagon concentrations (time = -30 min). At this time, a continuous IV infusion of cyclic somatostatin (0.65 μg/kg/min; BACHEM California, Torrance, CA) was initiated and continued for the next 30.5 h. Thirty min after

143

9*9 • 9 9 9 * ♦ ·

9 9

9 « «9 *· initiation of infusion (time = 0 minutes), an arterial blood sample was collected and a subcutaneous single injection of test substance or vehicle was injected into the dorsal neck. Arterial blood samples were collected every 3 hours thereafter for determination of plasma glucose and insulin concentrations and every 6 hours for determination of plasma glucagon concentrations. The entire study lasted 30 hours.

Arterial blood samples were stored in vacuum blood collection tubes containing disodium EDTA (Terumo Medical Corp., Elkton, ND) and immediately placed on ice. An aliquot (1.5 ml) of the blood sample was transferred to a polypropylene tube containing 40 μΐ aprotinin (10,000 KlU/ml; Trasylol, Miles, Inc., Diagnostics Division, Kankakee, IL) in preparation for determination of plasma glucagon concentration. The samples were centrifuged, and the resulting plasma was transferred to polypropylene test tubes and stored on ice for the duration of the study.

Plasma glucose concentrations were determined on the day of the study using glucose oxidase with a commercial glucose analyzer. Samples for further testing were stored at -80°C until analysis. Insulin concentrations were determined using a double antibody radioimmunoassay. Glucagon concentrations were determined using a radioimmunoassay kit (LINCO Research, Inc., St. Charles, MO).

At the end of the experiment, the catheters were flushed with fresh saline, treated with Kefzol (20 mg/kg) and filled with a glycerol/heparin mixture; the antibiotic (Keflex; 250 mg) was *

·

9

- 144

9999 9*

9 · • 9 9 • 9 9 9 * 9 9 · • · 9 * • 9 9 9

9 9 9

9 9 9

9 9 9

9· administered p.o. To minimize the number of animals and to allow for matching in the database, animals were used multiple times when possible. Repeated experiments with animals were performed at least weekly intervals.

The 1:3 cocrystal of human insulin and B29-Ne-octanoyl-human insulin had a duration of action that was 9 hours longer than NPH human insulin (24 hours versus 15 hours) at the higher dose (8753) and a duration of action that was 6 hours longer than NPH human insulin (21 hours versus 15 hours) at the lower dose (8752.5). The duration of action was determined by statistically comparing the mean glucose level with the mean level of the control group (saline). The glucose profiles for the 1:3 cocrystal of human insulin and B29-Ne-octanoyl-human insulin preparations were closer to the expected basal insulin profile than was the profile for NPH-human insulin. The 1:3 cocrystals also had a higher activity and a more desirable glucose profile than did bovine/porcine Ultralente insulin. The reduction in blood glucose compared to the control group (saline) caused by the co-crystal preparations persisted longer than the reduction caused by this bovine/porcine Ultralente insulin.

A 3:1 co-crystal preparation of human insulin and β29-Νεdecanoyl-human insulin (10252.5) had a duration of action that was 9 hours longer than NPH human insulin (24 hours versus 15 hours). The difference was statistically significant (p<0.05). In the same animals, a 3:1 co-crystal preparation of human insulin and β29-Νεdecanoyl-human insulin (2.5 nmol/kg, SC) had a duration of action that was 6 hours longer than Humulin U or bovine U preparations (24 hours versus 18 hours). The differences were also

9

9 9 »9

9 9 9

145

9999 99

9 9

9 9

9 9

9 9 ·

9« 99

· 9 9 9 9 9 9 statistically significant (p<0.05). The glucose profile for the co-crystal formulations was closer to the expected basal insulin profile than was the profile for NPH-human insulin. In addition, the variability in duration of action in dogs was smallest when the 3:1 co-crystals were administered.

In conclusion, the experiment demonstrated that the cocrystals of the present invention are effective for controlling glucose levels over a prolonged period of time in dogs. The experiments also support the conclusion that the cocrystals of the present invention will be effective for controlling glucose levels overnight in patients with type 2 diabetes or as a basal component of basal/single-dose insulin therapy in patients with type 1 or type 2 diabetes. The experiments also suggest that these formulations may produce less variable responses than commercially available insulin formulations.

Attempt 2

Duration of cocrystals in pigs

The studies were performed in normal female pigs weighing 17-25 kg and without anesthesia. An arterial (carotid or femoral artery) catheter was surgically preimplanted for sampling along with a jugular vein line for somatostatin administration. The cannulated pigs were fasted for 22-24 hours before the experiment. Subcutaneous insulin injections were given in the soft skin area behind the ear at a dose of 3.0 nmol/kg (0.5 units/kg). Somatostatin was administered simultaneously at a dose of 0.3 μg/kg/min (dissolved in 0.9% NaCl containing 1% human serum albumin, Miles Canada, Etobicoke, ON) to suppress

• 4

4

4

9 • 44

4

146 • 4 4 ·« · 4 endogenous insulin secretion. Nearly complete normoglycemia was maintained by infusion of 20% dextrose at various rates, with frequent monitoring of glucose concentrations. Plasma glucose levels were determined in fresh plasma samples on the day of the study using the glucose oxidase method using a commercial glucose analyzer.

In a euglycemic clamp study, a microcrystal formulation of the invention, containing 1:3 insulin:Β29-Νεoctanoyl-human insulin, was administered subcutaneously at a dose of 0.5 U/kg (equivalent to approximately 3 nmol/kg) at the start of the study (time 0) to five pigs. The glucose infusion rate required to maintain euglycemia (set point = approximately 90 mg/dl) was determined continuously. The control group received Humulin N (U100) subcutaneously and at the same dose (n=6). A concurrent infusion of somatostatin (0.3 μg/kg/min) was maintained throughout the duration of the experiment. With the microcrystals of the present invention, the glucose infusion rate increased steadily for the first two hours, reaching a maximum of approximately 7 mg/kg/min. From that point on, for a further approximately 17.5 hours, the glucose infusion rate decreased very regularly to approximately 0.5 mg/kg/min. For most of the time between 17.5 hours and the end of the study at 24 hours, the glucose infusion rate remained between approximately 0.5 and approximately 2 mg/kg/min. In comparison, the mean glucose infusion rate in the control group (Humulin NPH) continued to increase, reaching a maximum of approximately 14 mg/kg/min approximately 3 hours after administration. From then on, the glucose infusion rate decreased to approximately 7 mg/kg/min at approximately 4.5 hours and to approximately 5 mg/kg/min at 13 hours after administration. No further data were determined for the control group. These results are consistent with the conclusion that a microcrystalline preparation containing insulin and • 4

147 ···«

4 • 4

4

4 4 • 4 ♦ « ♦ 4 4

4·

4 4

4 4 *4

44

4 4 4 4 4 4 4 4

B29-Ns-octanoyl-human insulin in a 1:3 molar ratio had a flatter glucodynamic profile than NPH insulin.

Attempt 3

Duration of exposure to cocrystals in rats

The microcrystal formulation of the invention consisted of 1:3 insulin:B29-Ne-octanoyl-human insulin and was tested in BBDP/Wor rats, a genetically characterized animal model that is bred and available from the University of Massachusetts Medical Center (Worchester, MA) in association with Biomedical Research Models, Inc. (Rutland, MA). The DPBB/Wor rat line is prone to diabetes and exhibits insulin-dependent (autoimmune) diabetes mellitus. All formulations were administered subcutaneously at a dose of 0.9 U per 100 g body weight.

Male BBDP/Wor rats, 4-5 months old, maintained on long-acting insulin (LTI), were randomly divided into five experimental groups A, B, C, D and E. Group A (n=22) was treated for three days with U40 human insulin ultralente (Humulin UL); group B (n=18) was treated for three days with U40 Iletin Ultralente (65% bovine insulin, 35% porcine insulin); group C (n=10) was treated for three days with a microcrystal preparation containing 1:3 insulin:Β29-Νεoctanoyl-human insulin, prepared as described above; group D (n=21) was treated with a microcrystal preparation containing 100% B29-Nt-octanoyl-human insulin; and group E /

received U40 bovine-porcine PZI insulin (PZI). Each rat was given daily injections of its group's preparation for two days before blood glucose was determined and on the day it was

148 • 944

9 • « • 9 • 4 9 • 4 • 4 Φ «· • « Φ

Φ Φ Φ « Φ

ΦΦΦ φ Φ Φ ·

Φ · Φ · φ Φ Φ Φ

Φ Φ Φ ·

ΦΦ ·Φ blood glucose determined.

Blood was drawn half an hour before administration of the test preparations. Samples of the preparations were injected at 11:30. Blood was obtained by tail prick (without anesthesia). Samples were kept briefly on ice, then centrifuged and glucose was determined using a Beckman II glucose analyzer. Blood samples were drawn immediately before administration of the test preparations and 2, 4, 6, 8, 12, 16, 20 and 24 hours after their administration. Assuming that adequate action is indicated by a blood glucose level of less than 200 mg/dl, the preparations were effective for about 9.5 hours (Humulin UL), about 12 hours (Iletin U), about 15.5 hours (the present invention), about 20.5 hours (100% B29-Ns-octanoyl-human insulin) and about 21.5 hours control (PZI). It follows that the microcrystalline preparation of the present invention controlled blood glucose for a longer period of time than Humulin UL and Iletin U and for a shorter period of time than the microcrystalline preparation of 100% B29-Ns-octanoyl-human insulin or the PZI preparation.

Attempt 4

Exposure times of amorphous precipitates in rats

Amorphous precipitate preparations containing 1:3 insulin:B29-Ns-octanoyl-human insulin, 1:3 insulin:B28-Nεoctanoyl-Lys(B28),Pro(B29)-human insulin, and 1:3 insulin:B29Nε-nonanoyl-human insulin, prepared as described in Preparation 19, were tested in BBDP/Wor rats. All preparations were administered subcutaneously at a dose of 0.9 U per 100 g body weight.

00 • 0

»00

0 φ ♦ t ·

0 0

149

Male BBDP/Wor rats, 4-5 months old and maintained on long-acting insulin (PZI), were randomly divided into five experimental groups A, B, C, D and E. Group A (n=7) was treated with U40 NPH (Humulin N). Group B (n=8) was treated with U42.4 1:3 insulin:B29-Ns-octanoyl-human insulin. Group C (n=8) was treated with U42.6 1:3 insulin:B28-Ns-octanoylLys(B28),Pro(B29)-human insulin. Group D (n=8) was treated with U42.7 1:3 insulin:B29-Ns-nonanoyl-insulin. Group E was treated with U40 bovine-porcine PZI insulin (PZI).

Blood was collected half an hour before the administration of the test preparations. Samples of the preparations were injected subcutaneously (0.9 U/100 g body weight) at 11:30. Blood was obtained by tail prick (without anesthesia). Samples were briefly kept on ice, then centrifuged and glucose was determined using a commercial glucose analyzer. Blood samples were collected immediately before the administration of the test preparations and 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22 and 24 hours after their administration. Assuming that adequate action is indicated by a blood glucose level of less than 200 mg/dl, the preparations were effective for approximately 7.5 hours (1:3 insulin:B28Νε-octanoyl-Lys(B28),Pro(B29)-human insulin), approximately 9 hours (NPH), approximately 15.5 hours (1:3 insulin:B29-Ns-octanoyl-Lys(B28),Pro(B29)-human insulin), approximately 16 hours (1:3 insulin:Β29-Νεnonanoyl-human insulin), and approximately 22.5 hours control (PZI).

Principles, advantageous designs and working methods according to • ·

- 150 ··«· • · • · · • · ·♦ ·· ·· • · · · · • · ·*

9 9 9 9

9 9 ·

99 • 9 9

9 9

9 9 9

9 9 of the present invention have been described in the foregoing description of the invention. The present invention, the protection of which is claimed by the present application, is not limited in any way to any particular form described, since the described preferred embodiments and examples are considered illustrative and not restrictive. Those skilled in the art can make variations and changes without departing from the scope of the present invention.

Represents:

Dr. Otakar Svorcik

00

0 0 0

0 00

0 0 ·

0 0 ·

00

JUDr. Otakar Švorčík, attorney at law

120 00 Prague 2, Halkova 2

Claims (7)

PATENT CLAIMS 1. An insoluble composition characterized in that it contains: a) a protein selected from the group consisting of insulin, insulin analog and proinsulin; b) a derivatized protein selected from the group consisting of derivatized insulin, derivatized insulin analog, and derivatized proinsulin; c) a complexing compound; d) a hexamer-stabilizing compound; and e) divalent metal cation. 2. The composition according to claim 1, characterized in that it is an amorphous precipitate. 3. The composition according to claim 1, characterized in that it is a microcrystal. • 4 • * · · 4 4 - 162 49 44 44 a) dissolving the protein, derivatized protein, hexamer stabilizing compound and divalent metal cation in an aqueous solvent having a pH that allows for hexamer formation, and b) adding a complexing compound. 74. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, derivatized protein, hexamer stabilizing compound and divalent metal cation in an aqueous solvent having a pH that does not allow the formation of hexamers, b) adjusting the pH to a value in the range of between about 6.8 and about 7.8; and c) adding a complexing compound. 75. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent having a pH that allows for the formation of hexamers; b) separately dissolving the derivatized protein, the hexamer stabilizing compound and the divalent metal cation in an aqueous solvent having a pH allowing the formation of • · • · · · • » · · • · · 9 - 163 • 9 99 9 · ♦ · 9 9 9 9 9 9 9 9 9 9 9 9 99 99 99 99 9 9 9 9 hexamers; c) thoroughly mixing the solutions from steps a) and b); and d) adding a complexing compound to the solution formed in step c). 76. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, the hexamer-stabilizing compound, the divalent metal cation and the complexing compound in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur; b) separately dissolving the derivatized protein, the hexamer stabilizing compound, the divalent metal cation, and the complexing compound in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur; c) thoroughly mixing the solutions from steps a) and b); and d) adjusting the pH of the solution from step c) to a value at which precipitation occurs. 77. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, derivatized protein, hexamer stabilizing compound, divalent metal cation and complexing compound in an aqueous solvent, where the resulting • 9 ♦· ·· ·9 • 9 9 9 9 999« 9 99 99 9999 - 164 *··· 99 • · 99 9 « 99 9 9 99 9 9 9 99 99 9* 99 »9 the solution has a pH at which precipitation does not occur; and b) adjusting the pH of the solution from step a) to a value at which precipitation occurs. 78. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, derivatized protein, hexamer stabilizing compound and divalent metal cation in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur when a complexing agent is added; b) adding a complexing compound; and c) adjusting the pH of the solution from step b) to a value at which precipitation occurs. 75. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur when the complexing compound is added; b) separately dissolving the derivatized protein, the hexamer stabilizing compound and the divalent metal cation in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur when the complexing compound is added; • · - 165 c) thoroughly mixing the solutions from steps a) and b); d) adding a complexing compound to the solution from step c); and e) adjusting the pH of the solution from step d) to a value at which precipitation occurs. 80. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, protein derivative, hexamer stabilizing compound and divalent metal cation in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur when the complexing compound is added; b) adjusting the pH of the solution from step a) to a value at which precipitation occurs when a complexing compound is added; and c) adding a complexing compound to the solution from step b). 81. A method for preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: a) dissolving the protein, the hexamer-stabilizing compound and the divalent metal cation in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur when the complexing compound is added; • 0 0« 00 00 0 000 0 000· - 166 0·0· «0 0 0 b) separately dissolving the derivatized protein, the hexamer stabilizing compound and the divalent metal cation in an aqueous solvent, wherein the resulting solution has a pH at which precipitation does not occur when the complexing compound is added; c) thoroughly mixing the solutions from steps a) and b); d) adjusting the pH of the solution from step c) to a value at which precipitation occurs when a complexing compound is added; and e) adding a complexing compound to the solution from step d). 82. A method of preparing hybrid hexamers, comprising dissolving a protein, a derivatized protein, a hexamer-stabilizing compound, and a divalent metal cation in an aqueous solvent having a pH that allows for the formation of hexamers. 83. A method for preparing hybrid hexamers, comprising the following steps: • you· • · • you - 167 ♦ ti ti • ti ti • · · • ti ti • ti titi • 9·· tititi· • ··· ti ti · · • titi ··· ti· ti • titi ti ti titi · «· titi ti· ·« a) dissolving the protein, derivatized protein, hexamer stabilizing compound and divalent metal cation in an aqueous solvent having a pH that does not allow hexamers to form, and b) adjusting the pH to a value between approximately 6.8 and approximately 4· ·* 44 44 • 4 · 4 4 »44« • · 4 4 · ♦ 4 4 4 4 4 • · 4 · 4 · 4 4 · 4 44 4 4444 4444 4444 •4 44 44 44 44 69. A method of treating diabetes, comprising administering the composition of claim 1 to a patient in need thereof in an amount sufficient to control the patient's blood glucose level. 70. A method of treating hyperglycemia comprising administering to a patient in need thereof a composition according to claim 54 in an amount sufficient to control the patient's blood glucose level. 71. A hybrid hexamer composition comprising six monomers and zinc, wherein at least one monomer is selected from the group consisting of insulin, insulin analogs, and proinsulins, and at least one monomer is selected from the group consisting of derivatized insulin, derivatized insulin analogs, and derivatized proinsulins. 72. A mixed hexamer composition comprising zinc-protein hexamers and zinc-derivatized protein hexamers, wherein the zinc-protein hexamers comprise zinc and a protein selected from the group consisting of insulin, insulin analogs, and proinsulins, and wherein the zinc-derivatized protein hexamers comprise zinc and a derivatized protein selected from the group consisting of derivatized insulin, derivatized insulin analogs, and derivatized proinsulins. 73. A method of preparing an insoluble composition according to claim 1, characterized in that it comprises the following steps: 4 · ··· 4 4 4 · · 4 · · 9 9 9 9 · β* · · ·· · «4 * 9 9 9 9 9 9 9 99 63. The suspension formulation of claim 54, wherein the solution phase further comprises a protein selected from the group consisting of insulin and insulin analogs and a derivatized protein selected from the group consisting of derivatized insulin and derivatized insulin analogs. 64. The suspension preparation of claim 63, wherein the protein in the solution phase is the same as the protein in the insoluble phase and the derivatized protein in the solution phase is the same as the derivatized protein in the insoluble phase. 65. A suspension preparation according to claim 54, characterized in that the insoluble phase consists essentially of an amorphous precipitate. 66. A suspension preparation according to claim 54, characterized in that the insoluble phase consists essentially of microcrystals. 67. A suspension preparation according to claim 54, characterized in that the insoluble phase consists of a mixture of amorphous precipitate and microcrystals. 68. The suspension preparation of claim 54, wherein the solution phase further comprises zinc and protamine, wherein the ratio of zinc to total protein in the suspension preparation is from about 5 to about 7 moles of zinc atoms per mole of total protein and the ratio of protamine to total protein in the suspension preparation is from about 0.25 mg to about 0.5 mg per mg of total protein. • ·44 ·· • 4 - 161 4. The composition according to claim 3, characterized in that the microcrystal has a rod-shaped morphology. 5. The composition of claim 3, wherein the microcrystal has an irregular morphology. 6. The composition of claim 1, wherein the derivatized protein is selected from the group consisting of fatty acid acylated insulin, fatty acid acylated insulin analogs, and fatty acid acylated proinsulins. 7. The composition according to claim 6, characterized in that 152 φφ φφ « φφφφ φ φ φφ φφφ ♦ · · φ · φ φ φ φφ φφ φφ φφ φφ the complexing compound is protamine, the hexamer-stabilizing compound is a phenolic preservative, and the divalent . metal cation is zinc. 8. The composition of claim 1, wherein the derivatized protein is selected from the group consisting of fatty acid acylated insulin and fatty acid acylated insulin analogs. 9. The composition of claim 8, wherein the complexing compound is protamine, the hexamer-stabilizing compound is a phenolic preservative, and the divalent metal cation is zinc. 10. The composition of claim 1, wherein the derivatized protein is selected from the group consisting of acylated insulin, acylated insulin analogs, and acylated proinsulins. 11. The composition of claim 1, wherein the derivatized protein is selected from the group consisting of acylated insulin and acylated insulin analogs. 12. Composition according to claim at, characteristic with by that, that derivatized protein is monoacylated in their Lys -New- amine group • 13. Composition according to claim 12, determining with by that, that The complexing compound is protamine, which is present in an amount of about 0.15 mg to about 0.5 mg per 3.5 mg of total protein. 153 • fcfcfc ·· • · · • fc fc · • · • fc ·· • fcfc · • fcfcfc fc fcfc · • fcfc · fcfc fcfc • fc « · · • · · • · fc • fc fc 14. The composition of claim 13, wherein the divalent metal cation is zinc, which is present in an amount of about 0.3 moles to about 0.7 moles per mole of total protein. 15. The composition of claim 14, wherein the hexamer-stabilizing compound is a phenolic preservative selected from the group consisting of phenol, m-cresol, okresol, p-cresol, chlorocresol, methylparaben, and mixtures thereof and is present in a ratio of at least 3 moles of phenolic preservative per 6 moles of total protein. 16. Composition according to claim 15, determining with that, that the protein is selected insulin analogues. from the file, including insulin and 17. The composition according to protein is insulin. claim 16, determining with that, that 18. Composition according to claim 17, characteristic with that, that The derivatized protein is insulin that is monoacylated at the LysB29-NE-amine group. 19. Composition according to claim 18, characterized by by that, that derivatized protein is insulin, which is acylated saturated fatty acid with direct chain. 20. Composition according to claim 19, characterized by by that, that saturated fat acid with direct chain is chosen from of a set including n-hexanoic acid, n-heptanoic « · • φ 154 • · φ · φφ ·· acid, n-octanoic acid, n-nonanoic acid and ndecanoic acid. 21. The composition of claim 20, wherein the molar ratio between protein and derivatized protein is from about 1:9 to about 9:1. 22. The composition of claim 21, wherein the straight-chain saturated fatty acid is selected from the group consisting of n-hexanoic acid, n-octanoic acid, and n-decanoic acid. 23. The composition of claim 22, wherein the straight-chain saturated fatty acid is selected from the group consisting of n-octanoic acid and n-decanoic acid. 24. The composition of claim 23, wherein the straight-chain saturated fatty acid is n-octanoic acid. 25. The composition of claim 24, wherein the molar ratio between protein and derivatized protein is from about 1:3 to about 3:1. 26. The composition of claim 22, wherein the molar ratio between protein and derivatized protein is from about 1:9 to about 1:1. 27. The composition of claim 16, wherein the protein is an insulin analog. 9 9 155 9·« 4 «9 • 9 V 9 9 9 9 9 · 9 > 9 · 9» *· 9 » · 9' 99 • 9 » • 9 · 9« 99 99 • 9 9 9 9 9 9 9 *9 9 9 « • 28. The composition of claim 27, wherein the protein is a LysB28,ProB29-analog of human insulin. 29. The composition of claim 27, wherein the protein is an AspB28-analog of human insulin. 30. The composition of claim 1, wherein the protein is insulin or an insulin analog and the derivatized protein is an acylated protein selected from the group consisting of acylated insulin and acylated insulin analogs. 31. The composition of claim 1, wherein the derivatized protein is insulin that is monoacylated at its LysB29-NE-amino group with a straight-chain saturated fatty acid. 32. The composition of claim 31, wherein the saturated straight-chain fatty acid is selected from the group consisting of n-hexanoic acid, n-heptanoic acid, n-octanoic acid, n-nonanoic acid, and ndecanoic acid. 33. The composition of claim 32, wherein the derivatized protein is selected from the group consisting of B29-NE-hexanoyl-human insulin, B29-NE-octanoyl-human insulin, and B29-NE-decanoyl-human insulin. 34. The composition of claim 31, wherein the straight-chain saturated fatty acid is selected from - 156 of the set, comprising n-dodecanoic acid, tetradecanoic acid and n-hexadecanoic acid. n35. The composition of claim 1, wherein the derivatized protein is diacylated at the Lys-Ne-amine group and is also acylated at one N-terminal Na-amine group and wherein the fatty acid is selected from the group consisting of n-hexanoic acid, n-heptanoic acid, noctanoic acid, n-nonanoic acid, and n-decanoic acid. 36. The composition of claim 1, wherein the derivatized protein is acylated with a saturated branched chain acid. characterized by being oily 37. The composition of claim 36, wherein the derivatized protein is acylated with a branched chain saturated fatty acid having from three to ten carbon atoms in its longest branch. 38. The composition of claim 1, wherein the derivatized protein is an insulin analog that is monoacylated at its Lys-N-amino group with a straight-chain saturated fatty acid. 39. The composition of claim 38, wherein the fatty acid is selected from the group consisting of n-hexanoic acid, n-heptanoic acid, n-octanoic acid, n-nonanoic acid, and n-decanoic acid. - 157 <> · ι: * · • · 40. The derivatized monoacylated composition of claim 38, wherein the protein is an insulin analog that has a fatty acid in the Νε-amino group selected from the group consisting of n-dodecanoic acid, n-tradecanoic acid, and n-hexadecanoic acid. 41. The composition of claim 38, wherein the derivatized protein is selected from the group consisting of fatty acid acylated animal insulins, fatty acid acylated monomeric insulin analogs, fatty acid acylated deletion analogs, and fatty acid acylated β1-shifted insulin analogs. 42. The composition of claim 41, wherein the derivatized protein is a fatty acid acylated des(B30)-analog of human insulin, a fatty acid acylated LysB28,ProB29-analog of human insulin, or a fatty acid acylated AspB28-analog of human insulin. 43. The composition of claim 42, wherein the derivatized protein is a fatty acid-acylated des(B30)-analog of human insulin. 44. The composition of claim 43, wherein the derivatized protein is a B29-Ns-myristoyl-des(B30)-analog of human insulin. 45. The composition of claim 42, wherein the derivatized protein is a fatty acid-acylated LysB28,ProB29-analog of human insulin. • · · · · · - 158 46. The composition of claim 45, wherein the derivatized protein is a B28-NE-myristoyl-LysB28,ProB29 human insulin analog. 47. The composition of claim 42, wherein the derivatized protein is a fatty acid-acylated AspB28-analog of human insulin. 48. The composition of claim 1, wherein the molar ratio between protein and derivatized protein is from about 1:9 to about 9:1. 49. The composition of claim 48, wherein the ratio is from about 1:3 to about 3:1. 50. The composition of claim 48, wherein the ratio is from about 1:9 to about 1:1. 51. The composition of claim 1, wherein the protein is insulin. 52. The composition of claim 1, wherein the protein is an insulin analog. 53. The composition of claim 52, wherein the protein is a monomeric insulin analog. 54. A suspension preparation, characterized in that it comprises an insoluble phase and a solution phase, wherein the insoluble phase comprises the insoluble composition according to claim 1 and the solution phase comprises the insoluble composition according to claim 1. - 159 phase includes an aqueous solvent. 55. The suspension formulation of claim 54, wherein the solution phase further comprises a phenolic preservative at a concentration of about 0.5 mg per mL to about 6 mg per mL of solution, a pharmaceutically acceptable buffer, and an isotonic agent. 56. The suspension formulation of claim 54, wherein the solution phase further comprises insulin, an insulin analog, a derivatized insulin, or a derivatized insulin analog. 57. The suspension preparation of claim 56, wherein the solution phase comprises insulin. 58. The suspension preparation of claim 56, wherein the solution phase comprises derivatized insulin. 59. The suspension preparation of claim 56, wherein the solution phase comprises an insulin analog. 60. The suspension preparation of claim 56, wherein the insulin analog is a monomeric insulin analog. 61. The suspension preparation according to claim 56, wherein the insulin analog is a LysB28,ProB29-analog of human insulin. 62. The suspension preparation according to claim 56, wherein the insulin analog is an AspB28-analog of human insulin. - 160 • · · · ·· ·· ·· ·· ·· • · · · · · · · · 4 · « · and · · · and ···· 7.8.