EP0537185A1 - An enzymatic process for the preparation of derivatives of growth hormone releasing factor and peptides useful as intermediates in the process - Google Patents

Abstract

Derivatives of the growth hormone releasing factor GRF(1-29)NH2 and its analogs are prepared by reacting a substrate component of the formula GRF'-Met-Ser-X, wherein GRF' represents the initial sequence GRF(1-26) or its analogues, comprising fragments of GRF(n-26), in which n is between 1 and 8 and X represents an acyclic alpha-amino carboxylic acid residue having a side chain uncharged hydrophilic which exhibits at least the size of a methyl group, preferably Ala, Thr, Ser, Asn or Gln, with H-Arg-NH2 as the nucleophilic component, in the presence of an L-specific serine or thiolcarboxypeptidase enzyme obtained from yeast or of animal, plant or other microbial origin, preferably CPD-Y from yeast, in an aqueous solution or dispersion having a pH between 6 and 9 and, if necessary, coupling the desired fragment termination N (1-(n-1)) by chemical or enzymatic process. Also described are GRF-related peptides of the formula GRF"-Met-Ser-X, where GRF" represents the initial sequence GRF(1-26), in which 1 to 4 of the amino acid residues may be replaced or deleted, or their des-alpha-NH2 derivatives, X corresponding to the description above, and which constitute effective intermediate products. The method allows for selective transpeptidation, yielding satisfactory amounts of biologically effective GRF-related amides.

Description

AN ENZYMATIC PROCESS FOR THE PREPARATION OF DERIVATIVES OF GROWTH HORMONE RELEASING FACTOR AND PEPTIDES USEFUL AS INTERMEDIATES IN THE PROCESS.

The present invention relates to.a process for the preparation of GRF(1-29)NH₂, an interesting biologically active derivative of growth hormone releasing factor, and analogs thereof as well as new peptides which are useful as intermediates in this process.

The use of biologically active peptides for pharmaceutical purposes and in agriculture has increased the importance of being able to synthesize such compounds in bulk-scale. Three methods are available: (a) chemical synthesis, (b) enzymatic synthesis, and (c) fermentation with genetically manipulated microorganisms. While the methods (a) and (b), or a combination of them, are the preferred for short peptides it becomes more and more apparent that long peptides in the future will be produced through exploitation of the advances in the methods dealing with recombi- nant DNA. However, these methods do not permit a number of modications such as incorporation of D-amino acids and C- terminal amide groups which may be of importance for the biological activity. A subsequent enzymatic modification is therefore highly desirable and such reactions have only been studied to a limited extent.

An enzyme has been described which catalyses the hydroxyl- ation of C-terminal glycyl residues which subsequently decomposes leaving the penultimate residue amidated (US

Patent No. 4.708.934, EP 308067A, DK Application No.

4489/88). This glycine oxidase enzyme which is dependent on Cu²⁺, O₂ and ascorbate as cofactors is considered to be the enzyme responsible for in vivo formation of peptide amides. It has been utilised for amidation of peptides in small scale but as it exhibits low activity its applicability for large scale work is still questionable. The enzyme, as isolated from natural sources like rat medullary thyroid carcinoma is very costly.

Amidation may also by achieved by protease-catalysed condensation reactions using an amino acid amide or peptide amide as nucleophile. The yields of condensation reactions are generally low even in the presence of organic solvents unless the product precipitates in the reaction mixture and this is often not the case with long peptides. In addition, the precursor peptide may exhibit poor solubility in such media. However, serine or thiol- protease catalysed transpeptidation reactions may be carried out in high yield but it is a prerequisite that the enzyme exhibits specificity for a peptide bond close to the C-terminus. Endopeptidases are not generally suitable since they usually will cleave at other positions in the peptide chain as well. Serine carboxypeptidases, on the other hand, exhibit strict specificity for the C- terminal peptide bond and are able to catalyse the exchange of the C-terminal amino acid with an amino acid amide, added to the reaction medium to compete as nucleophile with water.

This property of serine carboxypeptidases was realized by a group of researchers at Carlsberg Research Center and lead to a family of patents assigned to the present assignee based on DK application No. 1443/79 represented by EP Patent No. 17485, US Patent No. 4.806.534 and its parent US patent No. 4.339.534 and WO 80/02151, which lead to i.a. DK Patent No. 155613. These patents were based on the at that time surprising finding that exopeptidases were suitable as catalysts for enzymatic peptide synthesis, while the prior art dealt exclusively with endopeptidases. Dependent on the nature of the reactants (substrate and nucleophile components), and the reaction conditions, particularly the pH, serine and thiol carboxypeptidases may catalyze peptide synthesis by chain elongation or by transpeptidation. The preferred enzyme is carboxypeptidase Y (CPD-Y) from yeast.

The underlying and subsequent research has been further described in a number of articles (Ref 14-18), which together with the above-mentioned patents are all incorporated by reference.

The general principle of enzymatic peptide synthesis by transpeptidation in the presence of serine or thiol carboxypeptidases is disclosed in US Patent No. 4.806.473 and its parallel Danish Patent No. 155613. With particular reference to the production of peptide amides these patents generally disclose and claim the production of peptide amides A-B-NH₂ where A represent an N-terminal protected amino acid residue or an optionally N-terminal protected peptide residue and B-represents an L-amino acid residue, by reacting as substrate component an optionally N-terminal protected peptide A-X-OH where A is as defined above and X represents an amino acid, with a nucleophile (amine) component H-B-NH₂ in the presence of an L-specific serine or thiol carboxypeptidase enzyme from yeast, or of animal, vegetable or microbial origin in an aqueous solution of dispersion being a pH from 5 to 10.5. As further explained in Ref. 14 the preferred pH is about neutral if the formation of a peptide amide is desired.

As a representative example from US Patent No. 4,806,473 may be mentioned the reaction of Bz-Phe-Gly with Leu-NH₂ in the presence of CPD-Y at pH 7.6 and 25°C leading to the formation of Bz-Phe-Leu-NH₂ in a 90% yield. Further examples are incorporated in DK Patent No. 155 613 i.a. the reaction of Ac(Ala)₄ with Leu-NH₂ in the presence of CPD-Y leading to Ac(Ala)₃-Leu-NH₂ in a yield of 70%.

Further experiments are disclosed in Refs. 14-18 which support the pioneer character of these early patents and the general applicability of serine carboxypetidases as catalysts for C-terminal modification of peptides.

In order to provide a better understanding of the present invention which is described in more detail below a general discussion of the transpeptidation principle and the competing reactions is deemed proper.

C-terminal amidation of a peptide by means of a serine carboxypeptidase catalysed transpeptidation reaction is dependent on: a) solubility of the peptide in an aqueous medium in which the enzyme is active and relatively stable, b) accessibility of the C-terminus to enzymatic cleavage and c) availability of a serine carboxypeptidase of suitable substrate preference.

The type of reactions which may take place are outlined in Scheme 1 with the substrate R-NH-A-B-C-OH and the nucleophile H-D-NH₂. The enzyme attacks the C-terminal peptide bond producing an acyl-enzyme intermediate which sub- sequently may be deacylated by the nucleophile, producing an amidated transpeptidation product (T1), in competition with water, producing a hydrolysis product (H1). The ratio of T1 to H1 may be increased by increasing the concentration of the reactive, deprotonated form of the nucleo- phile, i.e. by increasing either pH or the added amount of nucleophile. The nature of the leaving group, H-C-OH, has, in the case of CPD-Y, been shown to significantly influence this ratio as well (14, 15) albeit with no obvious trend. Since this amino acid residue does not constitute part of the desired amidated product (Tl) it may in principle be chosen freely. A serine carboxypeptidase may also catalyse the formation of an elongated condensation product (C1) in competition with H1 and T1. However, at the optimal pH-values for transpeptidation such reactions are in aqueous solution energically unfavourable and consequently, at equilibrium, the amount of C1 which may be formed is limited to a few per cent, even at nucleophile concentrations exceeding 1 M. Furthermore, the rates of such reactions are normally very low compared to the competing attack on the C- terminal peptide bond, and thus, condensation products are rarely a problem. However, this relation may be reversed in cases where the C-terminal sequence of the peptide does not match the substrate preference of the enzyme with the consequence that small amounts of condensation products are formed in the initial phases of the reaction and then, with time, disappears as the substrate is consumed (16).

The hydrolysis product (H1) may be transformed via the same type of reactions into the products T2, H2 and T1, followed by similar reactions with H2, etc. However, new products may also arise when the enzyme acts on the C- terminal amide bond of T1, i.e. ammonia functions as leaving group, producing another transpeptidation product (TT1) or hydrolysis product (HT1). For these products to appear in significant amounts it is required that the amidase activity towards Tl is significant compared to the peptidase activity towards the substrate and this is normally only the case at pH>9 where the peptidase activity is low and the amidase activity is high or in cases where the peptidase activity is low due to lack of preference for the C-terminal peptide bond.

It is apparent that the action of a serine carboxy- peptidase on the C-terminus of a peptide in the presence of an amino acid amide may lead to numerous products of which only T1 is the desired.

While reactions with N-blocked dipeptides of a suitable composition relative to the substrate preference of the enzyme in some case lead to yields of T1 approaching 100% (14, 15, 17, 18) the situation is different with larger molecules.

Thus in the patent family represented by US patent no. 4,645,740 and Danish patent no. 148714 assigned to the present assignee and incorporated herein by reference a process is described for enzymatic replacement of the B-30 amino acid in insulins, in particular conversion of porcine insulin into human insulin by replacement of the B-30 amino acid alanine with threonine using a serine carboxypeptidase, preferably carboxypeptidase Y from yeast (CPD-Y) or carboxypeptidase P from Penicillium janthinellum (CPD-P) as catalyst.

When porcine insulin was reacted with threonine amide in the presence of CPD-Y at pH 7.5 a complicated mixture of reaction products was obtained ( Example 4 of the patents). The mixture was analysed by ion exchange chromatography and three peaks were detected. The peaks were further analyzed by enzymatic digestion and amino acid analysis leading to the following composition (Ins-Pro-Lys-Ala-OH being used to denominate the porcine insulin starting material), and the symbols from Scheme I are used as illustration.

Peak 1: 21% of which

65% Ins-Pro-Lys-Thr-Thr-NH₂ (TT1)

35% Ins-Pro-Lys-Thr-NH₂ (T1) Peak 2: 61% of which

52% Ins-Pro-Lys-Ala-OH

(unreacted porcine insulin) (S) 26% Ins-Pro-Lys-Thr-OH

(human insulin) (HT1)

22% Ins-Pro-Thr-NH₂ (T2)

Peak 3: 18%

Breakdown products

Thus in the known process only 21% of a mixture of T1 and TT1 was obtained. The predominant fraction consisted of a mixture of unreacted S, HT1 and T2.

Also a significant amount of breakdown products was obtained.

In another example carried out at pH 9.5 80% of the porcine insulin was reacted under formation of what appeared to be human insulin amide (T1) which, however, was further hydrolyzed in situ to human insulin (HT1). A significant amount (20%) had reacted to form H2.

The above-mentioned insulin modification is further analyzed in Refs. 23 and 16. While therefore the formation of human insulin amide by means of the general principle of transpeptidation described above (S -> T1) is certainly possible, it is not easily monitored. Evidently reaction conditions may be selected which on the face of it seem favourable to the formation of T1, but such attempts have not been reported. Also the separation of the various reactants may be more or less difficult. As T1 lacks the negative charge at the C-terminus it may easily be separated from H1 in preparative scale by ion exchange chromatography. However, provided that the side-chain of the C-terminal amino acid residue is uncharged, T2, T3 etc. as well as C1 and TT1 exhibit the same charge and can therefore not be separated from T1 by ion exchange chromatography. In such cases preparative HPLC with reverse-phase columns appears to be the only, and a much less attractive, alternative. It is thus more desirable to attempt to suppress the formation of such products by selecting the optimal pH, concentration of nucleophile and most important, the serine carboxypeptidase of the most suitable substrate preference (see Table I) which rapidly and in the highest possible yield converts the substrate in T1 and produces the lowest possible yield, if any, of the undesirable side-products.

To recapitulate the essence of the above observations, incorporation of C-terminal amide groups in peptides by transpeptidation in the presence of a serine carboxypeptidase using the proper amino acid amide as the nucleophile as broadly described and claimed in US patent no. 4,806,473 and the other family members is a very appropriate method virtually applicable for any peptide.

However, the process is not always sufficiently selective and necessitates purification procedures in order to remove products of various side reactions in particular when longer peptides are used, in which case the optimal reaction conditions for suppressing the side reactions are difficult to establish.

In the early articles by the original inventors published shortly after filing of the above patent applications, attempts were made to analyze the influence of the C- terminal (leaving group) amino acid and the penultimate amino acid. Thus in Ref. 14 Breddam et al using Leu-NH₂ as the nucleophile and Gly, Ala, Ser, Val, Leu and Phe as the leaving groups suggested on the basis of the obtained yields that only in cases where the leaving group is one of the smallest amino acids, i.e. Gly, Ala or Ser is the reaction successful, and at least for the simple substrates tested there was no dependence on the penultimate residue (being Ala, Phe and Gly).

In Ref. 15 Breddam et al using Gly-NH₂ as the nucleophile and Z-Ala-X as the substrate, where X was Gly, Ala, Ser, Arg, Pro, Lys, Asn, His, Val, Met, Phe and Asp modified the earlier statement to the effect that when using Gly- NH₂ as the nucleophile, the yield is strongly dependent on the nature of the C-terminal (leaving group) amino acid. The yields varied from 10 to 100% with the lowest yields obtained with substrates where a hydrophobic acid (Val, Met, Phe) serves as leaving group. It should be noted that the yields with basic acids (Arg, Lys) are comparable to the yields with the hydrophilic acids (Ala, Ser), Lys being even better than Ser.

As for the penultimate amino acid residue of the peptide substrates the influence was investigated using a series of N-blocked dipeptides with different penultimate amino acids (Ala, Val, Leu, lie, Phe and Val) as the leaving group. Using Gly-NH₂ as the nucleophile, it was apparent that the coupling yield which varied from 45% for lie to 5% for Phe is dependent on the penultimate amino acid residue, but no obvious trend could be found.

In Ref. 23 some of the experiments underlying US patent no. 4,645,740 and its family members were discussed. Here porcine insulin Ins-Pro-Lys-Ala was reacted with i.a. Thr-

NH₂ and it was concluded that Ins-Pro-Lys-OH was a better substrate than Ins-Pro-Lys-Ala-OH, since Ins-Pro-Thr-NH₂ was formed in greater yields than Ins-Pro-Lys-Thr-NH₂. By inference Lys in this reaction was a better leaving group than Ala.

Also a significant oligomerization under formation of Ins-

Pro-Lys-Thr-Thr-NH₂ occurred.

These results were further confirmed in Ref. 16 using Bz- Lys-Ala-OH as a model peptide alongside with porcine insulin. The conclusive message was that for the future use of CPD-Y (the serine carboxypeptidase used in the experiments) in transpeptidation reactions it is important to be aware of the possibility that side products may be formed.

Besides the above investigations of the applicability of serine carboxypeptidases in C-terminal modifications of insulin, a further experiment with amidation of longer peptides using CPD-Y as a catalyst has been reported.

Thus in EP-B2-197794 and the parallel US patent no. 4,709,014 (Tamaoki) human calcitonin-Leu peptide was reacted with ammonia as the nucleophile using CPD-Y as the catalyst under conditions otherwise similar to those used by Breddam et al in Ref. 15. (The discussion of Ref. 15 above has been limited to the use of amino acid amides as nucleophiles, while in fact the main purpose of the article was to compare various nucleophiles also including ammonia).

Tamaoki obtained human calcitonin amide in a yield of 24.7%, leaving 57% unreacted substrate and 17.2% non- amidated side products (including human calcitonin).

In its more general aspects the Tamaoki patents disclose a process for the preparation of a peptide having a C- terminal proline amide, which comprises reacting in aqueous solution a peptide substrate having C-terminal Pro-Leu, Pro-lle, Pro-Val or Pro-Phe with carboxypeptidase Y in the presence of ammonia.

Without in any way wanting to endorse the statements made by Tamaoki, it should be mentioned that he claims that contrary to the findings of Breddam et al in Ref. 15, where a preference for hydrophilic C-terminal amino acids as leaving groups is expressed, the use of hydrophobic amino acids (Leu, lle, Val and Phe) gives better yields than Gly, when Pro is the penultimate amino acid. Nevertheless, the yields of the amidation products in Tamaoki's examples using Cbz-Ala-Pro-X-OH as the substrate, where X is Leu, Leu, Val, Phe and lle, were only 35,1%, 43%, 15,4%, 13,4%, 22,6%, respectively. The remainder was - to the extent reported - unreacted starting material and non-amidated side-products Cbz-Ala- Pro-OH.

As mentioned earlier, the present invention relates to a process for the preparation of derivatives of growth hormone releasing factor and analogs thereof.

Growth hormone releasing factor (GRF) or somatocrinin, stimulating the release of growth hormone in the pituitary, is a 44-residue amidated peptide 5 10

H₂N-Tyr-Ala-Asp-Ala-Ile-Phe-Thr-Asn-Ser-Tyr-Arg-Lys-

15 20 25

Val-Leu-Gly-Gln-Leu-Ser-Ala-Arg-Lys-Leu-Leu-Gln-Asp- 30 35

Ile-Met-Ser-Arg-Gln-Gln-Gly-Glu-Ser-Asn-Gln-Glu-Arg-

40

Gly-Ala-Arg-Ala-Arg-Leu-NH₂ which may be used to stimulate growth of vertebrates (31). Both the amide group as well as the 15 amino acid residues closest to the C-terminus can be removed without abolishing the activity, but each results in a gradual decrease in the biological activity. However, a large part of the activity may be regained by amidation of these truncated peptides. One such derivative of particular interest is GRF (1-29)NH₂, which can be regarded as the bioactive core of GRF. Since the determination of the activity of GRF(1-29)NH₂ numerous structural linear and cyclic analogs have been made by exchanging, deleting or modifying one or more of the 29 L-amino acids in the native GRF-chain. Particularly interesting analogs have been made by Heimer et al (32), Roche Research Center, New Jersey, USA, viz. [desNH₂Tyr¹,D-Ala²,Ala¹⁵] -GRF(1-29)NH₂ which is 10 times more potent than the parent compound in vivo and Cyclo- (Asp⁸-Lys¹²)-[Ala¹⁵]GRF(1-29)NH₂, which is less potent than the linear compound in vitro, but retained significant bioactivity and induced marked growth hormone in pigs.

The Heimer et al article is incorporated by reference.

Heimer et al also focused on a number of other modifica tions. Replacement of Gly¹⁵ with helix forming amino acids

Val, Leu, Aib and Ala resulted in analogs with increasing biological activity, while replacement by Sar, a helix- breaking residue resulted in a profound loss of biological activity.

Analogs of GRF(1-29)NH₂ with replacement of Tyr by desNH₂Tyr¹ or incorporation of D-Ala² also resulted in increased biological activity. Other analogs are described in the patents and patent applications listed below, all incorporated herein by reference.

EP-A-352014 (Salk Inst. for Biol. Studies) showing a number of possible modifications of GRF and GRF(1-29)NH₂ in the 1, 2, 3, 8, 12, 13, 15, 18, 21, 22, 24, 25, 27 and

28 positions and where of interest in the present context Arg²⁹ may be replaced by Cys, Abu, Asp, Glu, Orn, Lys, Dab or Dap (all L or D). Ostensibly these peptides are more active with better resistance to enzymatic degradation than native GRF.

EP-A-307860 (Roche) describes a number of linear and cyclic GRF analogs related to the ones earlier discussed. EP-A-292334 (Salk) describes a number of possible modifications of GRF and GRF(1-29)NH₂ in the 1, 2, 3, 8, 10, 12, 13, 15, 18, 21, 22, 24, 25, 27 and 28 positions.

FR-A-2594832 (Sanofi) describes modified GRF(1-29)NH₂, where Val 19 may be exchanged with lle, Met²⁷ may be exchanged with Nle and/or Ser²⁸ may be exchanged with Asn and in which at least 1 and up to 6 amino acids, preferably in positions 2, 5, 12, 17, 21 or 27, are replaced by an α-amino alkynoic acid, preferably α-amino- 4-pentynoic acid, α-amino-4-hexynoic acid and 2,6-diamino-

4-hexynoic acid, which may be substituted. EP-A-216517 (Salk) corresponding to US-A-4689318 describes a number of GRF(1-29)NH₂ analogs where at least 4 of amino acids 5-13, 15, 18-21 and 26 are different from the native sequence and preferably containing Ala²⁸.

US-A-4628043 (Salk) describes another group of analogs, where i.a. Tyr¹ may be replaced by Met, Leu, D-Tyr, D-His or His and Gly¹⁵ may be replaced by D-Ala.

Other analogs are described in EP-A-188214 (Tulane Educational Fund) [D-Ala²,Arg^12,21]GRF(1-29)NH₂, EP-A-

177819 (Roche), US-A-4528190 (Salk), US-A-4518586 (Salk), EP-A-121764 (Roche), with Met²⁷ replaced by Leu or Nle, JP-A-63/287799 (Tulane Educational Fund) disclosing i.a.

[iPr-Tyr¹, iPr-Lys^12,21]GRF(1-29)NH₂ as 106 times more potent than GRF( 1-29 )NH₂.

Generally speaking a person skilled in the art looking for a process for amidation of GRF(1-29)OH and its analogs would certainly expect the general process according to US patent no. 4,806,473 using GRF(1-28)-X, where X is an L- amino acid, as a substrate and Arg-NH₂ as a nucleophile in the presence of a serine carboxypeptidase to work, but he would have no obvious choice of the applicable amino acids X, and he would expect a number of competing reactions to occur leaving him uncertain as to the yields and the necessary purification procedures in order to arrive at the desired amidated end-product.

The invention is based on the surprising finding that by selecting as the leaving group X in the starting material GRF(1-28)X a number of amino acids which does not logically fit into any of the preference patterns suggested in the earlier works by Breddam et al and Tamaoki discussed above, viz. the acyclic α-aminocarboxylic acids having an uncharged hydrophilic side chain of at least the size of a methyl group, it is possible to obtain GRF(1-29)NH₂ in a surprisingly high yield. Preferred amino acids are Thr, Ala, Ser, Gln and Asn, in particular Thr and Ala. Very small amounts of unreacted substrate and byproducts resulting from the competing reactions are observed, in particular when the experiences with human insulin amide described above and Tamaoki's results with calcitonin are taken into account.

To a certain extent, these observations are based on experiments with model tripeptides containing the native GRF(27-28) sequence and various leaving groups, as well as a synthetic GRF(1-28)Ala-peptide, as further described below.

Based on these experiments and the general correlation between model peptides and longer peptides shown in the above articles, it is therefore reasonable to conclude that the process is not limited to production of peptides containing the native GRF(1-29) sequence but also at least to analogs where one or more of the amino acids in positions 1 to 26 are replaced by other amino acids both in D- and L-form and cyclic amino acids, which may be protected or derivatized, e.g. analogs of the types disclosed in the above-mentioned patents incorporated by reference.

Accordingly, the present invention relates to a process for the preparation of derivatives of Growth Hormone Releasing Factor, viz. GRF(1-29)NH₂ and analogs thereof, characterized by reacting a substrate component of the formula GRF'-Met-Ser-X wherein GRF' denominates the native GRF(1-26) sequence or analogs thereof including GRF(n-26) fragments, where n is from 1 to 8, and X is an acyclic α-amino carboxylic acid residue having an uncharged hydrophilic side chain of at least the size of a methyl group, with H-Arg-NH₂ as nucleophile component in the presence of an L-specific serine or thiolcarboxypeptidase enzyme from yeast or of animal, vegetable or other microbial origin in an aqueous solution or dispersion having a pH of from 6 to 9, and if necessary coupling the desired N-terminal (1-(n-1)) fragment chemically or enzymatically.

For the enzymatic fragment coupling an endopeptidase enzyme may be used having the proper specificity with regard to the C-terminal of the (1-(n-1)) fragment, e.g. papain or chymotrypsin.

A chemical fragment coupling may be carried out following introduction of suitable protective groups in solution phase in the presence of suitable catalysts, e.g. DCC together with additives, e.g. HOBt.

A preferred group of substrate components are those of the formula GRF"-Met-Ser-X, wherein GRF" denominates the native GRF(1-26) sequence, wherein from 1 to 4 of the amino acid residues may be replaced or deleted or the des- α-NH₂ derivatives thereof and X has the above meaning.

The applicable carboxypeptidases in the process of the invention are L-specific serine or thiol carboxypeptidases. Such enzymes can be produced by yeast fungi, or they may be of animal, vegetable or other microbial origin. A particularly expedient enzyme is carboxypeptidase Y from yeast fungi (CPD-Y). This enzyme is described in the earlier patents i.a. with reference to Johansen et al (Ref. 28) who developed a particularly expedient purification method by affinity chromatography on an affinity resin comprising a polymeric resin matrix with coupled benzylsuccinyl grups. CPD-Y, which is a serine enzyme is available in large amounts and displays relatively great stability. Further details are given in Ref. 14.

In addition to CPD-Y, which is the preferred enzyme at present, the process of the invention is feasible with other carboxypeptidases, such as those listed in the following survey:

Enzyme Origin

Fungi

Carboxypeptidase S-1 Penicillium janthinellum Carboxypeptidase S-2 Penicillium janthinellum Carboxypeptidase(s) from Aspergillus saitoi

Carboxypeptidase(s) from Aspergillus oryzae

Plants

Carboxypeptidase(s) C Orange leaves

Orange peels

Carboxypeptidase C_N Citrus natsudaidai Hayata Phaseolain French bean leaves

Carboxypeptidases M-I and M-II Germinating berlay

Carboxypeptidase W-II Wheat bran

Carboxypeptidases from Germinating cotton plants

Tomatoes

Watermelons

Bromelain(pineapple)powder

The close relationship between a number of the above carboxypeptidases is discussed by Kubota et al (Ref. 33). As stated above the process of the invention is carried out at pH 6.0 to 9.0, preferably at pH 6.5 to 8.0, most preferably from 7.5 to 8.0. The preferred pH-value, which is often within a very narrow range, depends upon the enzyme used. The selected pH-value should preferably be maintained throughout the reaction.

The pH-control may be provided for by incorporating a suitable buffer for the selected pH-range in the reaction medium.

The pH-value may also be maintained by adding an acid, such as HCl, or a base, such as NaOH, during the reaction. This may conveniently be done by using a pH-stat.

However, the conditions may also be influenced upon by varying the enzyme concentration, reaction time, etc.

The reaction is carried out in an aqueous reaction medium which, if desired, may contain up to 80% by volume of an organic solvent. Preferred organic solvents are alkanols, e.g. methanol and ethanol, glycols, e.g. ethylene glycol or polyethylene glycols, glycerol, alkanoic acids, e.g. acetic acid, dimethyl formamide, dimethyl sulfoxide, tetrahydrofurane, dioxane and dimethoxyethane.

The selection of the composition of the reaction medium depends particularly upon the solubility, temperature and pH of the reaction components and the reaction products involved and upon the stability of the enzyme.

The reaction medium may also comprise a component that renders the enzyme insoluble, but retains a considerable part of the enzyme activity, such as an ion exchanger resin. Alternatively, the enzyme may be immobilized in known manner, e.g. by bonding to a matrix, such as a cross-linked dextran or agarose, or to a silica, polyamide or cellulose, or by encapsulating in polyacrylamide, alginates or fibres. Besides, the enzyme may be modified by chemical means to improve its stability or enzymatic properties.

Preferably a chelating agent e.g. EDTA is included in the reaction medium which may also comprise salts. The concentration of the two participants in the reaction may vary within wide limits, as explained below. A preferred starting concentration for the substrate component is 0,1-20 mM, preferably 1-10 rnM, and for the nucleophile component 0.02 to 2 M, preferably 0.2 - 1.5 M.

The enzyme activity may vary as well, but the concentration is preferably 10^-8 to 10^-4 M. The most advantageous activity depends i.a. on the substrate chain and concentration, the nucleophile concentration, the reaction time, the reaction temperature, the pH, and the presence of organic solvents and/or salts. The amount of enzyme should in each case be adjusted to give the necessary degree of conversion for formation of an optimal absolute amount of product.

According to the invention the reaction temperature is 10° to 50°C, preferably 20° to 40°C. The most appropriate reaction temperature for a given synthesis can be determined by experiments, but depends particularly upon the concentration of the nucleophile component and the enzyme concentration. An appropriate temperature will usually be about 20° to 30°C, preferably about 25°C, taking into account due consideration for enzyme activity and stability.

Similar variations occur for the reaction time which depends very much upon the other reaction parameters, especially the enzyme concentration. The standard reaction time in the process of the invention is about 1 - 3 hours. The abbreviations of- amino acids, amino acid derivatives and peptides are according to Guidelines of the IUPAC-IUB Commission on Biochemical Nomenclature and the amino acids are on L-form unless otherwise stipulated. The binding site for the C-terminal amino acid residue of the sub- strate is denoted S₁ ' and those for the amino acid residues in the amino-terminal direction away from the scissile bond are denoted S₁, S₂, ...... S_i. The substrate positions are all denoted P₁', P₁, P₂ ..... P_i in correspondence with the binding sites (Ref. 1).

The following additional abbreviations are used: AcOH, acetic acid; Bz, N-benzoyl; DCCI, N,N'-dicyclohexyl- carbodiimide; Dhbt, 3,4-dihydro-4-oxo-1,2,3-benzotriazin- 3-yl; DICI, diisopropylcarbodiimide; DMAP, 4-N,N-dimethyl- aminopyridin, DMF, N,N-dimethylformamide; EDTA, ethylene diamine tetraacetic acid; Fmoc, fluorenyl-9-ylmethyl- oxycarbonyl; Ft, fraction of transpeptidation; GRF, growth hormone releasing factor; HPLC, high performance liquid chromatography; Mtr, 2,3,6-trimethyl-4-methoxyphenyl- sulfonyl; Pfp, pentafluorophenyl; TFA, trifluoro acetic acid; THF, tetrahydrofuran.

Before the process of the invention will be illustrated by examples, starting materials, methods of measurement, etc. will be explained in general terms.

Starting materials

DMF was purified by fractional distillation in vacuo. THF was passed through active Alumina prior to use. Fmoc-amino acids were purchased from Milligen and were converted into Dhbt esters by reaction with carbodiimides and Dhbt-OH (Fluka) in THF (20). The peptide synthesis resin (Macrosorb SPR) was purchased from Sterling Organics. Sequence analysis was carried out on an ABI gas phase sequencer and d Durrum D-500 instrument was used for amino acid analysis. The HPLC equipment was from Waters Associates. Samples were hydrolyzed for 24 h in 6 N hydrochloric acid and evaporated, prior to amino acid analysis.

H-Arg-NH₂-2HCl, Bz-Met-OH, other amino acids and amino acid derivatives and GRF (1-29)-NH₂ were obtained from Bachem, Switzerland. CPD-Y is commercially available from the applicants. The substrates Bz-Met-Ser-Ala-OH, Bz-Met- Ser-Leu-OH, Bz-Met-Ser-Thr-OH, Bz-Met-Ser-Arg-OH, Bz-Met- Ser-Gly-OH and Bz-Met-Ser-Ser-OH were synthesized as described below.. All other reagents and solvents were from Merck, W. Germany. CPD-W-II and CPD-S-1 were prepared as previously described (18, 21).

General procedure for the Preparation of Tripeptide Substrates of the Formula BzMetSerX, X = Ala, Gly, Ser, Thr, Leu, Arg, Asn, Gln, Met 10.7 g benzoylmethionine methylester was dissolved in 240 ml DMF and 36.6 g serineisopropylester was added, followed by 400 ml H₂O. The pH was then adjusted to 8.5 using aqueous sodium hydroxide solution and 0.8 ml of 0.1 M EDTA was added. The reaction was then initiated by addition of 10 ml of a 0.2% (w/w) solution of crude papain preactivated by incubation with 0.1 M betamercaptoethanol at 35°C. The reaction was stirred for 3 hours and after 30 minutes 25 ml of enzyme solution was added and after one hour further 25 ml of enzyme solution was added. pH was then lowered to 3 by addition of aqueous HCl solution and the mixture was filtered and purified by reverse phase HPLC on Waters Preppak 500 60 μm C₁₈ columns using alcohol gradients at pH 3 in 50 mM acetic acid. Combined fractions containing the pure product benzoylmethionylserine isopropylester were combined and taken to dryness under reduced pressure to yield: 9.0 g BzMetSerOiPr (54%).

Example of Synthesis of BzMetSerThr:

2.0 g BzMetSerOiPr was dissolved in 30 ml of DMF and added to a solution of 24.0 g Threonine in 34 ml of H₂O and 2 ml of 0.1 N EDTA in which pH had been adjusted to 9.0 using sodium hydroxide solution. The reaction was then initiated by addition of 2.8 ml of a 0.35 mM solution of CPD-Y and was stirred at constant pH for 4.5 hours at room temperature. The reaction was then stopped by addition of 10 M HCl solution to pH 3 and following filtration, the mixture was purified by reverse phase HPLC on Waters Deltapak 300 A 15 um C₁₈ columns using an alcohol gradient in 50 mM acetic acid pH 3.

Combined fractions containing pure products were taken to dryness under reduced pressure to yield 0.63 g of BzMetSerThrOH (34%) with an analytical purity of more than 95% by HPLC at 220 nm.

BzMetSerAla, BzMetSerSer, BzMetSerGly, BzMetSerLeu, BzMetSerArg, BzMetSerAsn, BzMetSerGln and BzMetSerMet were prepared in an analogous manner, using the free amino acids Ala, Ser, Gly, Leu, Arg, Asn, Gin and Met as nucleophiles instead of Thr, respectively, and similarly pure products were obtained.

Assembly of human GRF(1-28)-Ala-OH A column was packed with kiselguhr supported poly dimethyl acryl amide methyl ester resin, Macrosorb SPR-100 (500 mg, 0.05 mmole functional groups) derivatized with ethylene diamine and acylated with 4-hydroxylmethyl phenoxyacetic acid Dhbt ester. Fmoc-Ala-O-Pfp (137 mg, 275 αmole) and DMAP (6 mg, 5 αmole) was dissolved in DMF and recirculated through the resin (500 mg) overnight. The reaction mixture was displaced with DMF and acetic anhydride (50 μl) and DMAP (3 mg) was added. After 10 min of recirculation the reaction mixture was again displaced and the circuit was washed thoroughly with DMF.

A standard synthesis cycle consisted of a sequence of 10 min deprotection with piperidine in DMF (20%), a wash with

DMF, the introduction and recirculation of acylating agent and a wash was entered at the deprotection point. During the washing periods all valves and loops were exercised. All couplings were performed with a single addition of Fmoc amino acid Dhbt esters (3 eqv.) as described by Atherton et al. (20) with t-butyl based side chain protection and Mtr protection of arginine. Coupling times were determined by the fading of the intense yellow colour developed on the resin by formation of an ion pair between amino groups and Dhbt-OH (22). Two hours coupling time was sufficient for most of the reactions. Slow couplings were observed for Arg(20) (6 h) and for the sequence from Ala(4) to Leu(14) 5-10 h).

Deprotection and cleavage from the resin was carried out by a 16 h treatment with TFA containing phenol (5%). Evaporation and trituration with diethyl ether (3 times 25 ml) resulted in isolation of the crude peptide (210 mg). This was analyzed by reverse phase HPLC (5 α Vydac) using a gradient of 30% to 70% of B (A = 0.1% TFA, B = 10% water and 0.1% TFA in acetonitrile). Two components in equal amounts were found, probably due to the presence of methionine sulfoxide in one of them. The peptide (200 mg) was treated with TFA containing thioanisole (10%) for 2 hours. After this, only one major component was found by analytical HPLC as described above. After evaporation and trituration with diethyl ether the peptide was dissolved in 10 ml DMF/1% AcOH (2/3, 10 ml) and chromatographed on a G15 column with 1% AcOH. The fractions containing the major component was collected and lyophilized. This was separated by preparative HPLC on a 25 mm x 300 mm 15 μ reversed phase column (Waters deltaprep.). The sample was divided into two portions each of which were dissolved in 20 ml 15% DMF in water. The sample was applied to the column and eluted at 10 ml/min first for 10 minutes with 30% B and then with a gradient (20% -70% B) over 30 min. The main peak eluted after 32 min and was collected and lyophilized to yield a total of 44 mg of peptide. The amino acid analysis showed: Asp 2.93; Thr 0.94; Ser 2.82; Glu 2.23; Gly 1.08; Ala 3.91; Val 1.05; Met 1.00; lle 2.05; Leu 4.07; Tyr 1.67; Phe 0.90; Lys 2.16; Arg 2.31. The structure of the peptide was confirmed by gas phase sequence analysis.

Transpeptidation reactions

Transpeptidation reactions reported in table II below were carried out in the following way: the nucleophile (H-Arg- NH₂,2HCl) was dissolved in 5 mM EDTA and pH was adjusted with NaOH solution to the selected reaction pH. The substrate was then added (25 μl per 1 ml solution of nucleophile from a 40-80 mM solution in either DMF (the Bz-peptides) or 5% acetic acid (GRF(1-28)-Ala-OH), followed by enzyme from an aqueous solution. The reactions with the benzoylated tripeptides were carried out in a pH stat (1 ml) whereas the reactions with GRF(1-28)-Ala-OH were carried out in small scale (0.1 ml) with the nucleophile acting as buffer. When the substrate was added from 5% acetic acid a drop in pH was observed. To compensate, pH of the nucleophile solution was elevated prior to addition of the substrate. These reaction conditions were slightly modified in the experiments reported in tables III and IV as explained below. HPLC-analysis

During the reaction 10μl aliquots were removed from the reaction mixture and the reactant composition was determined by HPLC. The following eluant system was used: 50 mM triethylammonium phosphate, pH 3.0 (A-buffer) and 50 mM triethylammonium phosphate, pH 3.0 in 80% CH₃CN (B- buffer) employing various linear and concave gradients. For the N-blocked tripeptides a Nova-Pak 5 μ C-18 reverse phase column from Waters was used whereas a Vydac C-18 reverse phase column was used for the GRF(1-28)-Ala-OH. A gradient was found which could separate the N-benzoylated tripeptide substrates as well as the possible products of the reaction including the by-products reported. This permitted a detailed analysis of the reaction mixture. A similar system was not developed for the longer peptides. However, maximal resolution was attempted and each product was collected and subjected to acid hydrolysis and amino acid analysis. The separations were carried out at room temperature and monitored by UV-absorbance at 254 nm (Bz- peptides) or 280 nm (GRF).

The per cent composition of the reaction mixture was determined directly from the integrated peak areas since all components had the Bz- or, in the case of GRF, tyrosine as dominant chromophore at the respective wavelengths. The fraction of aminolysis (Ft) was expressed as the ratio between Tl and all other products formed, i.e. uncomsumed substrate was disregarded in the calculation. The invention is further illustrated in the accompanying drawings in which Fig. 1: Time course of the CPD-Y catalysed transpeptidation on GRF(1-28)-Ala-OH using H-Arg-NH₂ as nucleophile. The reaction conditions are listed in Table II. GRF(l-28)-Ala-0H (S); Δ

GRF(1-29)-NH₂ (T1); , GRF(l-28)-0H (H1);

GRF(1-28)-Arg-Arg-NH₂ (TT1).

Fig. 2: Separation of the products from GRF(1-28)-Ala-OH after treatment with CPD-Y for 147 minutes in the presence of H-Arg-NH₂. The separation was carried out on a Vydac C-18,5μ column using the TEAP buffer system described above and a 30 to 45% B gradient over 30 minutes. 1: GRF(1-28)-Arg-Arg- NH₂, 2: GRF(1-29)-NH₂, 3: GRF(1-28)-OH, 4: GRF(1-

28)-Ala-OH.

RESULTS AND DISCUSSION In the initial screening experiments the serine carboxypeptidase catalysed transpeptidation reaction leading to GRF (1-29)-NH₂ was studied in model reactions with the following N-benzoylated tripeptides: Bz-Met-Ser-Arg-OH, Bz-Met-Ser-Leu-OH and Bz-Met-Ser-Ala-OH using H-Arg-NH₂ as nucleophile. The transpeptidation product is in all three cases Bz-Met-Ser-Arg-NH₂. corresponding to Bz-GRF (27-29)- NH₂. The initial choice of leaving groups was motivated by the high catalytic activity found from the expected hydrolytic preference of available serine carboxypeptidases as listed in Table I combined with their expected suitability in transpeptidation reactions. Thus - Arg-OH was considered appropriate for CPD-W-II (17,21,26) and possibly CPD-S-1, for which good results had previously been obtained using also LeuOH as a leaving group with ArgNH₂ as a nucleophile (18), while SerOH had performed poorly. Ala-OH and possibly Leu-OH were deemed suitable for CPD-Y (14,15), from an activity standpoint (14,15). Regarding transpeptidation suitability of these groups, the prior art left no unambiguous or unequivocal guidelines for CPD-Y, as previously described. Thus, while in some references (14,15) the preferred leaving groups were small hydrophilic amino acid residues, other references (U.S. Patent 4,709,014) claimed a particular group of large hydrophobic amino acid residues, and Ala, which with respect to size and hydrophilicity pertains an intermediate position among the natural amino acids, had performed fairly poorly in the only report for a long peptide, i.e. Porcine insulin (U.S. Patent 4,645,740), where the overall yield was less than 10% of the correct product.

It should be noted in this respect, that hydrophilic amino acids are defined as amino acids having side chains more hydrophilic than cysteine, indicated by the hydrophilicity values assigned by Hopp and Woods (Proc. Nat. Acad. Sci. USA, 78 (6), pp. 3824-3828 (1981) (Ref. 34). The good initial results with Ala versus Leu in this model warranted further studies using different sizes of hydrophilic amino acid residues (i.e. Gly, Ser, Thr, Asn and Gin) to establish the size requirements for possible successful product formation, and also an establishment of the degree of hydrophilicity needed by testing Met as leaving group, which has a hydrophilicity value between Leu and Ala.

Table I. Substrate preference of serine carboxypeptidases for hydrolysis of P-X-Y-OH peptide substrates.

The preferences are listed accordingly to k_cat/K_m values in the following order: A > B > C > D > E, i.e. A corresponds to very fast hydrolysis and E to extremely slow hydrolysis.

The data are from Refs. 17, 18, 21, 26, 30.

The optimal conditions for serine carboxypeptidase catalysed transpeptidation reactions can be established by experiments in each case. The yield of transpeptidation is generally beneficially influenced by an increase in pH due to deprotonation of the amino group of the amino acid amide nucleophile but simultaneously the peptidase activity decreases, leading to lower rates of conversion, and the amidase activity increases, leading to degradation of the amidated product. An increase in pH can be com- pensated by an increase in the concentration of nucleophile but this also reduces the rate of conversion since the nucleophile acts as an inhibitor (25). With respect to pH the compromise usually is between 7.5 and 8.0 with the exception of CPD-S-1 which is unstable above pH 7.0. The transpeptidation reaction with CPD-W-II and Bz-Met- Ser-Arg-OH was initially carried out at pH 7.6 and 10 mM H-Arg-NH₂, the low concentration of nucleophile being motivated by the efficient binding of nucleophiles with positively charged side chains to this enzyme. After 47 minutes, 91% of the substrate had been converted into the following products (see Table II): 64% Bz-Met-Ser-Arg-NH₂ (Tl), 20% Bz-Met-Ser-OH- (H1) and 6% Bz-Met-OH (H2). Bz- Met-Arg-NH₂ (T2) as well as other potential products (see Scheme 1) were absent. The fraction of transpeptidation was 0.71 and this value was also obtained at pH 8.0. With 2, 5 and 25 mM H-Arg-NH₂ the fractions of transpeptidation were 0.55, 0.64 and 0.72, respectively, suggesting that a value around 0.7 cannot be exceeded. The transpeptidation reactions with CPD-Y on Bz-Met-Ser- Ala-OH and Bz-Met-Ser-Leu-OH were initially carried out at pH 7.5 and 0.5 M H-Arg-NH₂, a rather high concentration of nucleophile due to an expected poor binding to CPD-Y which has a preference for nucleophiles with hydrophobic side chains. The fraction of transpeptidation was 0.91 with Bz- Met-Ser-Ala-OH and only 0.41 with Bz-Met-Ser-Leu-OH. The general trend of dependence on the nature of the amino acid leaving group is consistent with those of the previous observations where the fraction of transpeptidation with CPD-Y in most cases has been significantly and sometimes drastically higher when the leaving group is small and/or hydrophilic, e.g. -Ala-OH, as compared with hydrophobic, e.g. -Leu-OH (14, 15). However, the yields for Leu is in contrast to the results reported in Ref. 14, p. 243, where Leu was used as a leaving group leading to a 0% yield of transpeptidation. Thus, initially only the reaction with Bz-Met-Ser-Ala-OH was pursued at other reaction conditions and as basis for the further investigations of transamidation of longer peptides. At 0.025 M and 0.1 M H-Arg-NH₂ the fractions of transpeptidation were lower, i.e. 0.77 and 0.87, respectively, and a concentration of 0.5 M was therefore maintained. However, in spite of the high fraction of transpeptidation at this concentration a small amount (2%) of the undesirable Bz-Met-Arg-NH₂ (T2) was produced. An increase in pH to 7.8 prevented the formation of this product and simultaneously the fraction of transpeptidation was increased to 0.98.

CPD-S-1 was also tested in transpeptidation reactions with Bz-Met-Ser-Arg-OH and Bz-Met-Ser-Leu-OH. With 0.2 M H-Arg- NH₂ and pH 6.5 the fractions of transpeptidation were 0.38 and 0.48, respectively. Higher concentrations of nucleophile had no beneficial effect. The initial results of the transpeptidation reaction in the screening tests with the N-benzoylated tripeptides seemed to indicate that the most efficient amidation leading to a good yield of T1 and suppression of the formation of undesirable peptides such as T2, could be achieved with CPD-Y and with -Ala-OH as leaving group.

Consequently, GRF( 1-28)-Ala-OH was selected as test intermediate and was prepared by continuous flow peptide synthesis. The assembly of GRF(1-28)-Ala-OH (2) was carried out by the solid phase methodology described above and cleaved from the resin by a 20 h treatment with TFA and phenol.

The reaction of GRF(1-28)-Ala-OH with CPD-Y in the presence of H-Arg-NH₂ was carried out under the conditions stated in Table II. The reaction was followed by HPLC and the time course is shown in Figure 1. After approximately 2 1/2 hours essentially all substrate had been converted into the products GRF(1-28)-Arg-NH₂ (T1), GRF(1-28)-OH (H1) and GRF(1-28)-Arg-Arg-NH₂ (TT1). The HPLC chromatogram of the final reaction mixture is shown in Figure 2. The identity of GRF(1-29)-NH₂ as the dominant product was further investigated by co-chromatography with authentic GRF(l-29)-NH₂: only a single peak was observed. Separate experiments with no H-Arg-NH- added confirmed the retention time of H1. Minor components eluting around the position of GRF( 1-28)-Ala-OH, each constituting less than 2% of the reaction mixture, were collected but the amino acid analysis did not suggest any of them to be C- terminally modified derivatives of GRF(1-29).

At 147 minutes the reaction mixture contained 84% GRF(1- 28)-Arg-NH₂, i.e. GRF(1-29)-NH₂. Thus, with 1 g of CPD-Y approximately 2.7 kg of GRF(1-29)-NH₂ can be produced using the reaction times and conditions listed in Table II- However, CPD-Y is quite stable at pH 8.0 and 22°C and it is therefore possible to employ a much longer reaction time and correspondingly reduced enzyme concentration. Since CPD-Y is easily isolated from baker's yeast after autolysis (Ref. 28) or from the medium of genetically manipulated yeast cells (Ref. 29) the cost of the enzyme is rather low and the procedure described here therefore seems to be a valuable alternative to the use of the much more rare glycine oxidase.

As mentioned earlier the initial results with the N- benzoylated tripeptides Bz-Met-Ser-Ala-OH, Bz-Met-Ser-Arg- OH and Bz-Met-Ser-Leu-OH seemed to indicate that among these three model peptides the most efficient amidation would be achieved with Ala-OH as the leaving group, and this approach was tested further with the complete peptide sequence. In order to get a more thorough analysis of the appropriate leaving groups for the amidation a further series of model peptides containing hydrophilic amino acids of different size as leaving group were synthesized. Thus the small Gly, the intermediate Ser and the larger Thr were chosen, and transpeptidation tests were carried out using Bz-Met-Ser-Ala-OH as comparison. The results appear from Table III below.

Concentration of substrate: 2mM

Concentration of ArgNH₂: 1000mM

pH: 7.90

It is seen that as far as product yield and fractions of transpeptidation are concerned, surprisingly Thr which has not earlier been reported as leaving group is comparable to Ala despite its larger size. Despite a higher enzyme concentration Ser leads to somewhat less selective reac tion while Gly leaves a greater amount of unreacted substrate and a significant amount of byproducts.

In some of the reactions of the method according to the present invention, the absolute amount of product formed depends on the degree of conversion and passes through an optimum, which would justify stopping the reaction and recycling the remaining substrate, while in others, running the reaction to virtual completion is fully justified. In the Bz-Met-Ser-X tripeptide model experiments performed, Ala and Thr as the leaving groups X appear to be in the latter category, while Asn and Gin as leaving groups would appear to be in the former. In Table IV below some results at incomplete conversion are listed using the large, but hydrophobic Met as a reference example:

Concentration of substrate: 2 mM, concentration of ArgNH₂: 1000 mM, pH: 7.9

It is seen that, at between 50% and 70% conversion, good amounts of product are formed relative to the amount of substrate consumed using the large, hydrophilic amides Asn and Gln as leaving groups, while Met performs as poorly as Leu and Gly in Tables II and III and gives rise to a lot of hydrolysis byproduct. It is to be noted that this is in spite of the fact that, as stated in Table I, where Met is grouped together with Leu and Ala, Met is an active substrate as a leaving group, reducing the enzyme requirement as apparent from Table IV. It can be noted that Table I also groups Ser, Thr, Gin and Asn together with Gly as leaving groups, but in a completely different category than Ala, Leu, Met. Thus, the results in Tables II, III and IV illustrate the lack of possibility to predict the synthetic yields solely based on the hydrolytic preferences listed in Table I.

A logical standard for evaluation of the applicability of the various possible leaving groups would be the native GRF sequence, in which the C-terminal sequence is -Met- Ser-Arg-OH.

Based on the above results it can be concluded that a transamidation can be carried out with a sufficient selectivity using Ala, Thr, Ser, Asn and Gin as the leaving groups, while Leu, Gly and Met give rise to a significant amount of byproducts. It thus appears that a hydrophilic side-chain of at least the size of a methyl group is necessary for the desired selectivity and improved yield over the native GRF sequence. This result could not be foreseen from the articles discussed above and indicate that it is indeed possible to select substrates having greater feasibility as intermediates for transamidation than the natural GRF (1-29).

REFERENCES:

1. Schechter, I. & Berger, B. (1967) Biochem. Bhiphys, Res. Commun. 27, 157-162

14. Breddam, K., Widmer, F. & Johansen, J.T. (1980) Carlsberg Res. Commun. 45, 237-247

15. Breddam, K., Widmer, F. & Johansen, J.T. (1981) Carlsberg Res. Commun. 46, 121-128

16. Breddam, K., Johansen, J.T. & Ottesen, M. (1984) Carlsberg Res. Commun. 49, 457-462

17. Breddam, K. (1985) Carlsberg Res. Commun. 50, 309-323 18. Breddam, K. (1988) Carlsberg Res. Commun. 53, 309-320 20. Atherton, E., Holder, J.L., Meldal, M., Sheppard, R.C.

& Valerio, R.M. (1988) J. Chem. Soc. Perkin Trans. 1, 2887-2894

21. Breddam, K. Sørensen, S.B. & Svendsen, I. (1987) Carlsberg Res. Commun. 52, 297-311

23. Breddam, K., Widmer, F. & Johansen, J.T. (1981) Carlsberg Res. Commun. 46, 361-372

25. Breddam, K. & Ottesen, M. (1984) Carlsberg Res.

Commun. 49, 473-481

26. Breddam, K., Sørensen, S.B. & Ottesen, M. (1985) Carlsberg Res. Commun. 50, 199-209

28. Johansen, J.T., Breddam, K. & Ottesen, M. (1976) Carlsberg Res. Commun. 41, 1-14

29, Nielsen, T.L., Holmberg, S. & Petersen, J.G.L. Appl.

Microbiol. Biotechnol. (in print)

30. Breddam, K., Sørensen, S.B. & Ottesen, M. (1983) Carlsberg Res. Commun. 48, 217-230

31. Brain Peptides, Ed. Krieger, Brownstein, Martin, John Wiley & Sons (1983), 976-980

32. Synthetic peptides: Approaches to Biological Problems, Heimer et al, alan R. Liss, Inc. (1989) 309-319

33. Kubota et al. Carboxypeptidase C (1973), J. Biochem.

74, no. 4, 757-770

34. Hopp and Woods, Proc.Nat.Acad.Sci. USA, 78 (6),

pp. 3824-3828 (1981)

Claims

P A T E N T C L A I M S

1. Process for the preparation of derivatives of Growth

Hormone Releasing Factor, viz. GRF(1-29)NH₂ and analogs thereof,

c h a r a c t e r i z e d by reacting a substrate component of the formula

GRF'-Met-Ser-X wherein GRF' denominates the native GRF(1-26) sequence or analogs thereof including GRF(n-26) fragments , where n is from 1 to 8, and X is an acyclic α-amino carboxylic acid residue having an uncharged hydrophilic side chain of at least the size of a methyl group, with H-Arg-NH₂ as nucleophile component in the presence of an L-specific serine or thiolcarboxypeptidase enzyme from yeast or of animal, vegetable or other microbial origin in an aqueous solution or dispersion having a pH of from 6 to 9, and if necessary coupling the desired N-terminal (1-(n-1)) fragment chemically or enzymatically.

2. The process according to claim 1, wherein the substrate component used is GRF"-Met-Ser-X, wherein GRF" denominates the native GRF(1-26) sequence, wherein from 1 to 6 of the amino acid residues may be replaced or deleted or the des- α-NH₂ derivatives thereof and X has the above meaning.

3. The process according to claims 1 or 2, wherein X is selected from Ala, Ser, Thr, Asn and Gln.

4. The process according to claims 1 to 3, wherein the carboxypeptidase enzyme used is carboxypeptidase Y from yeast.

5. The process according to claim 4, wherein a carboxy peptidase Y is used which has been purified by affinity chromatography on an affinity resin comprising a polymeric resin matrix with a plurality of coupled benzylsuccinyl groups.

6. The process according to any of claim 1 to 3, wherein the carboxypeptidase enzyme used is selected from the group consisting of Carboxypeptidase S-l and S-2 from Penicillium janthinellum, carboxypeptidases from Aspergillus saitoi or Aspergillus oryzae, carboxypeptidases C from orange leaves or orange peels, carboxypeptidase C_N from Citrus natsudaidai Hayata, phaseolain from bean leaves, carboxypeptidases M-I and M-II from germinating barley, carboxypeptidase W-II from wheat bran, carboxy- peptidases from germinating cotton plants, tomatoes, watermelons and Bromelein (pineapple) powder.

7. The process according to any of the preceding claims, wherein an immobilized carboxypeptidase enzyme is used.

8. The process according to any of the preceding claims, wherein an aqueous reaction solution containing from 0 to 80% of organic solvent is used.

9. The process according to claim 8, wherein the organic solvent used is selected from the group consisting of alkanols, alkanoic acids, dimethyl sulfoxide, dimethyl formamide, dioxane, tetrahydrofurane, dimethoxy ethane, glycerol, ethylene glycol and polyethylene glycols.

10. The process according to claim 1, wherein a GRF'-Met- Ser-X is used, which has been produced enzymatically, by recombinant DNA-methods, by chemical synthesis or a combination of these.

11. A GRF-related peptide of the formula

GRF"-Met-Ser-X wherein GRF" denominates the native GRF(1-26) sequence, wherein from 1 to 4 amino acid residues may be replaced or deleted or the des-α-NH₂ derivatives thereof, and X is an acyclic α-aminocarboxylic acid residue having an uncharged hydrophilic side chain of at least the size of a methyl group.

12. A GRF related peptide according to claim 11, wherein GRF" denominates the native GRF(1-26) sequence and X has the above meaning.

13. A GRF related peptide according to claim 11 or 12, wherein X is Ala, Thr, Ser, Asn or Gin.

14. A GRF related peptide according to claim 13, wherein X is Ala or Thr.