BORONIC ACI D IN HI BITORS OF* D IPEPTIDYL PEPTIDASE iV: A NEW CLASS OF IMMUNOSUPPRESSIVE AGENTS Roger J. Snow and William W. Bachovchin Io II. HI. IV. V. VI. VII. VIII. IX. X. XI. XlI. Introduction ............................ 150 Dipeptidyl Peptidase IV as a Target for Immunosuppression... 150 Substrate Specificity of DPP lV .................. 152 Inhibitors of DPP IV ........................ 153 Boronic Acids as Serine Protease Inhibitors ........... 155 Chemistry of the Proline Boronic Acid Dipeptides ........ 156 Synthesis of L-Proline Boronic Acid ............... 156 Structural Studies on Proline Boronic Acid Dipeptides ...... 159 Structural Studies of Boronic Acids ................ 161 Studies on Pinanediol Hydrolysis ................. 163 NMR Studies of Cyclization Rates ................ 164 Enzyme Inhibition Studies .................... 167 t This chapter is dedicated to the menmry of Dr. Simon Courts. Advances in Medicinal Chemistry Volume 3, pages 149-177 Copyright © 1995 by JAI Press Inc. All fights of reproduction in any form reserved. ISBN: 1-$$938-798-X 149
`
`AstraZeneca Exhibit 2150
`Mylan v. AstraZeneca
`IPR2015-01340
`
`Page 1 of 29
`
`

`
`1 50 ROGER I. SNOW and WILLIAM W. BACHOVCHIN XHI. XIV. XV. Enzyme Kinetics .......................... 170 Biological Activity of boroPro Dipeptides ............. 170 Issues Concerning the Development of boroPro Dipeptides .... 172 Acknowledgments ......................... 172 References .............................. 173 I. INTRODUCTION The aim of this review is to discuss the design and development of a new class of immunosuppresive agents, which are inhibitors of the serine protease dipeptidyl peptidase IV (DPP IV). In particular, it will focus on the most potent class of inhibitors of this enzyme, dipeptides of proline boronic acid, which have been studied extensively in a collaborative effort between groups at Tufts University and at Boehringer Ingelheim. ii. DIPEPTIDYL PEPTIDASE IV AS A TARGET FOR IMM U NOSUPPRESSION Dipeptidyl peptidase IV ~ is a membrane-bound enzyme found on the surface of a wide vari.'.'ety of cell types, including liver, kidney, and intestine. ~ It acts by cleaving a dipeptide from the free N-terminus of a polypeptide where the second residue is proline. Its function in these tissues is generally thought to be involved in the catabolism of proline- containing peptides. 3 DPP IV is also found on the surface of T-lympho- cytes; its expression correlates with a memory phenotype, and it is highly expressed on double-negative thymocytes and natural killer cells: In the mid-1980s several groups showed that the amount of DPP IV on lym- phocytes increased upon activation. 5 Sch6n et al. subsequently showed that inhibitors of the enzyme block the proliferative response of T-cells to antigenic stimulation 6 and suppress IL-2 production. 7 DPP IV was later shown to be identical to the T-cell marker antigen known as CD26, s'9 which had been identified as a costimulatory molecule in T-cell activa- tion. ~° Cross-linking of CD26 with antibodies can potentiate the reponse to suboptimal stimulation of the T-cell receptor by anti-CD3. Recent reviews have appeared coveting biochemical it and immunological ~2 aspects of DPP IV. This review will focus on inhibitor design. Details of the involvement of DPP IV in T-cell activation have not yet been elucidated. Early on it was noted that several cytokines, including IL-I~, IL-2, TNF-~, and GM-CSF, have N-terminal sequences that should be substmtes for DPP IV. Hoffmann ~a has shown that short
`
`Page 2 of 29
`
`

`
`Boronic Acid Inhibitors of Dipeptidyl Peptidase IV 151 peptides corresponding to the N-terminal sequence of these molecules are cleaved by the enzyme, but the full-length cytokines are not. Further- more, short N-terminal deletions of either IL-I~ ~4 or IL-2 ~5 have little effect on their biological activity, arguing against this as a possible mechanism. To date there are no convincing candidates, either for the endogenous substrate, or for a possible ligand, such as another cell surface molecule. DPP IV has been reported to be associated both with CD45, ~6 a tyrosine phosphatase known to be essential for T-cell activa- tion, and with adenosine deaminase. ~7 Mutations in adenosine deaminase result in a form of severe combined immune deficiency, but the majority of the enzyme is normally found in the cytosol, and the function of the membrane-associated form is not yet clear. Inhibitors of DPP IV have no effect on the association of ADA. Antibodies against CD26 induce tyrosine phosphorylation, ts'~9 The human immunodeficiency virus (HIV) tat protein has been shown to bind to DPP IV and to inhibit its catalytic activity2°; this may be involved in the pathogenesis of the virus. DPP IV was also reported to function as a receptor for infection of cells by HIV, 21 although other groups have presented data that cast doubt on this observation. 22 More work is required to elucidate the mechanism of action of DPP IV, but the effectiveness of DPP IV inhibitors in blocking proliferation, which is discussed in more detail below, indicates that it plays an important role in T-cell activation. DPP IV is a 110-kDa homodimeric cell surface glycoprotein, with an intracellular N-terminus and a cytoplasmic domain of only six residues. Highly homologous cDNAs have been cloned from rat, 23 mouse, 24 and human cells, 25 which code for proteins of 767, 760, and 766 amino acids, respectively. The active site serine has been identified as Ser631 in the rat sequence 26 and Ser624 in mouse. In mouse, Asp702 and His734 have been shown to be the other members of the catalytic triad. 27 Based on this, Ser630, Asp702, and His740 comprise the catalytic triad of the human enzyme. DPP IV is not related to the classical serine proteases such as chymotrypsin, in either sequence or arrangement of catalytic residues, but is more similar to a family of peptidases and lipasesY and in particular to prolyl endopeptidase (PEP). 29 There has been some debate over whether the catalytic activity of DPP IV is essential for its function; antibodies against DPP IV such as IF7, which block T-cell activation, 1°~ have no effect on its enzymatic activity. Morimoto 3° has reported mutagenesis studies in which the active-site serine was replaced by alanine, which reduced the costimulatory activity of the molecule. Hegen, 31 studying CD26-transfected Jurkat cells, saw little effect on IL-2
`
`Page 3 of 29
`
`

`
`152 ROGER j. SNOW and WILLIAM W. BACHOVCHIN production or proliferation with reversible inhibitors and concluded that enzyme inhibition has no effect. It is possible that more complete, or irreversible, inhibition is required to block the response. Recently it was observed that addition of a soluble form of DPP IV to cell cultures could potentiate the response to antigen, 32 but no effect was seen when the DPP IV was f'urst inactivated with the irreversible inhibitor DFP. Although details of the mechanism of action of DPP IV in T-cell activation have still to be elucidated, the ability ofinhibitors to block this response makes it an attractive target for therapeutic intervention. I!1. SUBSTRATE SPECIFICITY OF DPP IV Most of the information about the steric requirements of the active site has come from studies on artificial substrates, in most cases dipeptide p-nitroanilides. 11'33-35 These are reactive amides, and may not reflect the substrate preference of peptides. These studies showed that there is an absolute requirement for a free N-terminus (P2 site, Fig. 1). There is a strong preference for proline at the second position (P] site), but artificial substrates with hydroxyproline or alanine are also cleaved, between 100- and 1000-ft~ld less efficiently, as judged by comparing k~=lK m. The five-membered ring of proline can be unsaturated, or contain an O or S atom without losing activity. 34 The azetidine analog is a slightly better substrate, and piperidine is about fivefold poorer than proline. ~ Almost any amino acid can occupy the terminal position (P2). 33c'35 There is a slight preference for hydrophobic side chains, but among natural amino acids there is only about a 10-fold X = O,S,CHOH n - 0,1.2 accepted Hydrophobic Ala 10OOx weaker preferred ~, '~k X Any except R ~ \ Pro ! I !cN: j N-terminuS required ~ H=N N O P2 P1 Pl" Figure 1. Substrate specificity of DPP IV.
`
`Page 4 of 29
`
`

`
`Boronic Acid inhibitors of Dipeptidyl Peptidase IV 153 difference between the best and the poorest substrate. Monomethylation of the amine is tolerated in Sar-Pro, but further methylation abolishes activity. The enzyme has been shown to recognize the trans form of the H-Xaa-Pro peptide bond. 36 The P] residue can be anything except a second proline. As might be expected, peptides containing D-amino acids are not cleaved. IV. INHIBITORS OF DPP IV The first inhibitors of DPP IV to be described were the LysPro derivative I and the acyl hydroxamic acid 2. sb These were not particularly potent, but 2, which is an irreversible inhibitor, with K i of 30 gM and kin ~ of 6.2 x 10-4s -~, was the first compound used to demonstrate blockade of T-cell function by DPP IV inhibition. The mechanism of inhibition of DPP IV by 2 and its analogs has been studied in some detail. 37"38 The acyl hydroxamic acids are suicide substrates, which are converted to a reac- tive intermediate, probably an isocyanate, in the active site. However, they are cleaved by the enzyme more efficiently than they are activated to generate the reactive species, which is the reason for their modest potency. Compound 1 reflects the inherently limited afrmity of a prod- uct-based inhibitor. The tripeptides diprotin A and B were reported to be inhibitors 39 but in fact are poor substrates. 4° Azapeptide esters such as 3 have been shown to inactivate DPP IV at 20-100 IJM, 4~ presumably by forming a stable acyl enzyme intermediate. Reactivation is quite rapid, however, and the compounds are chemically unstable in solution at pH O HN o No: L~ v - i NO2 H2N~ N~ ' ~O~ 20 O CO2H 1 H2N OPh H2 N~,~/ O O 3 4 X = CH 2, S
`
`Page 5 of 29
`
`

`
`154 ROGER J. SNOW and WILLIAM W. BACHOVCHIN H2N 0 -- r-(CH~° =~c, H2N % N~pp 1OAr Ar 0 0" ~'OAr 5 6a n=1 6b n=2 7. Although X-Pro dipeptides are weak inhibitors, simply removing the carboxylic acid gave a series of pyrrolidides and thiazolidides (4) 42 that are competitive, reversible inhibitors with ICs0s of 2-6 IxM. These have been shown to inhibit mitogen-induced lymphocyte proliferation 42 and suppress IL-2 and IL-6 production. 43 Analogous pyrrolidides have been reported as inhibitors of the related enzyme PEP. 44 Many inhibitors of serine proteases employ an electrophilic group such as an aldehyde or trifluoromethyl ketone to interact with the active site serine, but since inhibitors of DPP IV are necessarily dipeptides with a free N-terminus, use of these groups is limited because they will react with the terminal amine to give a cyclic product. The only report of this type of compound is the trimethylammonium methyl ketone 5, 45 which is a time-dependent inhibitor with an apparent K i of 9 ixM and kJK i of 2700 M-{-s -~ but has a half-life of only 15 minutes at pH 7.6. A series of inhibitors with greater stability are the phosphonate dipeptides 6. 46 These have moderate potency, the best compound (6b) having a reported K i of 236 ~tM, and ki~ t of 0.353 s -~, but they produce long-lasting inhibition, presumably by forming a phosphonate ester with the serine. Interestingly, the piperidine analogs (e.g., 6b) are better inhibitors than pyrrolidines (6a) in this case. The active-site directed group that turns out to be particularly useful for inhibition of DPP IV is the boronic acid. Proline boronic acid (boroPro) was first used by Bachovchin 47 in inhibitors of IgA protease and later incorporated into the DPP IV inhibitors Ala-boro- Pro 7 and Pro-boroPro 8. 4s These were shown to be extremely potent, time-dependent inhibitors of DPP IV and had immunosuppressive activ- 0 B -.. 0 0 , B "" 0 7 | g
`
`Page 6 of 29
`
`

`
`Boronic Acid Inhibitors of Dipeptidyl Peptidase IV 155 ity both in cell culture 4a and in vivo. 49 However, these compounds were also reported to lose their inhibitory activity in aqueous solution at neutral pH. 48 This has created the impression that the compounds decom- pose. As will be seen, this is not the case; the change in activity is now known to be due to reversible interconversion with a cyclic species 9 s°'5~ analogous to a diketopiperazine. This turns out to be a major advantage over compounds containing a carbonyl at the active site, which undergo irreversible cyclization to pyrazinones. V. BORONIC ACIDS AS SERINE PROTEASE INHIBITORS The idea that boronic acids, which readily form tetrahedral borates, might mimic the tetrahedral transition state in serine proteases was first put forward in the early 1970s. Simple aryl or alkylboronic acids were shown to be weak inhibitors of chymotrypsin. 52 Greater affinity was achieved by incorporating the boronic acid into an amino acid analog. 53 This has subsequently been extended to other proteases of greater them- peutic interest, such as human neutrophil elastase (HNE) 53b and thrombin. Kettner 54 introduced extremely potent tetrapeptide boronic acid inhibitors of HNE, and he 55 and others 56 have reported potent thrombin inhibitors. Boronic acid inhibitors of elastase have been reviewed recently. 57 The high affinity observed with many peptide boronic acids is due to the formation of a tetmhedral borate, covalently bonded to the active site serine. There is good evidence for the formation of this kind of adduct from X-ray crystallography of chymotrypsin, 5a ct-lytic protease, 59 and 6~-64 elastase. 6° NMR observations support this structure for certain in- hibitors of these enzymes, as well as for trypsin. In some cases, however, the boron is coordinated to the active site histidine. 6.-66 It has been proposed 61'63'64'66 that boronic acids that are able to occupy the specificity subsites of the enzyme bind to the serine, whereas non-peptide boronic acids, or peptides that are unable to occupy these sites bind to histidine. In general the highest affinity inhibitors form serine adducts. In several cases boronic acids show slow-binding kinetics. This is thought to be due to two-step binding with formation of a pre-complex, followed by slow conversion to the tetrahedral serine adduct. Tight binding of the tetrahedral borate can lead to extremely slow off-rates for some inhibi- tors, as is the case for the DPP IV inhibitors, which in turn are responsible for the high affinity.
`
`Page 7 of 29
`
`

`
`156 ROGER j. SNOW and WILLIAM W. BACHOVCHIN VI. CHEMISTRY OF THE PROLINE BORONIC ACID DIPEPTIDES At the outset, using Ala-boroPro and Pro-boroPro as leads, the chemistry group had three objectives: (1) to develop a practical synthesis of the active enantiomer of praline boronic acid (boroPro); (2) to prepare analogs to improve the activity and develop structure-activity relation- ships (SAR); (3) to understand the reason for the decrease in inhibitory activity and, if possible, find ways to minimize it. VII. SYNTHESIS OF t-PROLINE BORONIC ACID The first goal was to devise a synthesis ofproline boronic acid that could provide the active enantiomer on a large scale. The original synthesis of 7 and 8 used mcemic boroPro 47 obtained by the Matheson dichlo- romethyllithium insertion approach, 67 which has been used for almost all examples of aminomethyl boronic acids. The diasteroisomers of Pro- boroPro have since been separated by HPLC. 68 An enantioselective version of this synthesis is possible, with pinanediol as the chiral auxil- Iary, and has been used to make L-boroArg ~5'67b via an intermediate that could also be used for boroPro. A limitation of this chemistry is the need to carry out the reaction at very low temperature, at least if the dichlo- romethyllithium is generated by using butyllithium, which is difficult to maintain on a large scale. We decided it would be more convenient to start from a precursor already containing the five-membered ring. The f'urst approach started from pyrrole (Scheme 1), 69 which was protected with a Boc group. This serves two functions: first it activates the ring for lithiation at the 2-position, which was then reacted with trimethyl borate, forming 10. Second, it activates the ring for catalytic hydrogenation, which proceeded smoothly to yield the Boc-protected racemic boroPro U. The free boronic acid was protected with (+)-pinanediol, which not only serves as a chiral auxiliary for separating the enantiomers, but is a particularly stable protecting group. In this system we found that other protecting groups for the boronate, such as pinacol, were lost to some extent during chromatography on silica gel. This method could be used to prepare boroPro on a 100-g scale. An even shorter synthesis was achieved with Boc pyrrolidine, which was lithiated with sec-butyl lith- ium as described by Beak 7° and quenched with trimethyl borate to give 11 directly. This reaction can be carded out at temperatures as high as -40°C and has been used successfully on a 400-g scale. Removal of the
`
`Page 8 of 29
`
`

`
`I BOC a.b ~ OH ( OH BOC 10 ~ OH I OH BOC d,b I BOC o - ~ 8'0 B O = ' .e HCI HCI 13a 13 R;,,N R e , °'-C.---/'_.., I 14 I IS n k w o--- 12 i J 2 R I 1| 1 OH R ~ ~,. _.._.~ R ~ ~ B=.-OH " "" m / .1 2 BOC O / B-~OH O / B'-OH O~ NHR HO HO R 1 19 17 18 • Ala R 1 =Me, R2=H b Pro R 1.R 2='(CH2) 3 c Val R l=iPr,R 2=H Scheme 1. Reagents: (a) LiTMP, THF, -78°C; (b) (MeO)3B,-78°C to rt; (c) H 2, Pt/C, EtOAc, 50 psi; (d) sBuLi, TMEDA, THF,-70°C; (e) (+)-pinanediol, CH2Cl 2, rt; (f) HCl, Et20, 0°C-rt; (g) fractional crys- tallization; (h) BOC-amino acid, EDC, HOBT, N-methyl mor- pholine, CH2Cl2, 0°C-rt; b); (i) Na2CO 3 (aq), CH2Cl2; (j) maleic acid or MeSO3H , CH2ClJMeOH; (k) PhB(OH) 2, H20, hexane, rt; (I) Dowex 50X2-200, elute with aq. NH3; (m) MeSO3H, MeCN; (n) NalO 4, NH4OAc, H20, acetone, ft. 157
`
`Page 9 of 29
`
`

`
`158 ROGER I. SNOW and WILLIAM W. BACHOVCHIN Boc group with HCI gave the salt 13. Separation of the diastereoisomers of the fully protected boroPro 12 could be achieved, but only by HPLC. In a few instances (e.g., Val-boroPro), the diastereoisomers of the dipep- tide after coupling 13 to a Boo-amino acid were separable by column chromatography. A general solution was found when it was noted that the diastereoisomers of the hydrochloride salt 13 were readily separable by fractional crystallization. 69 The required L-isomer 12a is much less soluble and was obtained in >98% de after two crystallizations. The absolute configuration of 13a 69 and 15e 5~ was confirmed by X-ray crystallography. Several ways to obtain asymmetric induction were considered. Beak has reported an enantioselective version of the Boc-pyrrolidine lithia- tion, using sparteine in place of TMEDA71; unfortunately, the available enantiomer of sparteine gives the o configuration. Hydrogenation of the pinanediol ester of 10 gave only very slight selectivity, 69 although a C2-symmetric diol might be better. Since the racemic synthesis and isomer separation work well, these proved to be the most practical methods for obtaining L-boroPro in quantity. BoroPro pinanediol ester hydrochloride 13a was coupled with a variety of Boc-protected amino acids. The best results were obtained using the water-soluble carbodiimide EDC and HOBT, which typically gave 1,4 in >95% yield. Deprotection with HC1 gave the dipeptide 15 with the boronic acid still protected, which in some instances were used directly in biological assays. We also wanted to prepare the free boronic acids. Pinanediol bomnates are reported 72 to be very stable to hydrolysis, and the conditions described for their deprotection are incompatible with the other functional groups. Little formation of 19 was seen on treating 13 under a variety of hydrolytic conditions. To overcome this problem, two methods for removing pinanediol were developed by Coutts. 73a In the case of the fully protected dipeptide 13, it was noted that the problem is not that the hydrolysis is slow, but that the equilibrium strongly favors the boronate. The equilibrium could be driven to the boronic acid by removing the pinanediol by cleavage with periodate, providing 19 in good yield. It is noteworthy that loss of the boronic acid, which is susceptible to nucleophilic oxidants, is not observed with periodate, an electrophilic oxidant. In the case of the free amine dipeptides (lb'), efficient deprotection was achieved by transesterification with phenyl- boronic acid in a two-phase mixture of water at low pH and hexane. Both starting material and product are soluble only in the aqueous layer, whereas pinanediol phenyl boronate is soluble in hexane and was recovered
`
`Page 10 of 29
`
`

`
`Boronic Acid Inhibitors of Dipeptidyl Peptidase IV 159 quantitatively. This method has also been descd~ by Kettner. 73b The free boronic acids were isolated by passing the aqueous layer through an ion exchange column and eluting the product with aqueous ammonia. This yielded the amine free bases, which exist as the cyclic form 18 (see below). The open compound 17 was prepared either by adding dilute HCI and lyophilizing, or in certain cases by adding methanesulfonic acid to an acetonitrile solution of 18, which precipitated the salt, 17. VIII. STRUCTURAL STUDIES ON PROLINE BORONIC ACID DIPEPTIDES When the decrease in activity of the boronic acids was first observed, a plausible explanation seemed to be the formation of a cyclic species such as 9. Initial studies to identify the proposed cyclic structure were carried out on the pinanediol boronates. If the cyclization were occurring, the change from trigonal to tetrahedral boron should be readily discernible by ~B NMR. The dipeptide salt 15c showed the expected XtB signal for a trigonal boronate as a broad peak centered at 30 ppm downfield from BF 3 etherate. On converting 15c to the free base, the ~B NMR spectrum consisted of a single, much sharper peak at 7.8 ~i, consistent with a tetrahedral boron, and provided strong evidence that this material has the cyclic structure 1re. ~H NMR of 16c showed it to be predominantly one diastereoisomer (90%) at the newly formed chiral center at boron. Further evidence that 16¢ was indeed cyclic came from the 2D NOESY spectrum, which revealed NOEs between an NH and a methyl group of pinanediol (see Fig. 2), as well as between the boroPro tx proton and one H " H Figure 2. Selected NOEs observed in the 2 D NOESY spectra of 16a (R 1 = Me), dashed arrow, and 16c (R ~ = iPr), solid arrows. From Ref. 51.
`
`Page 11 of 29
`
`

`
`160 ROGER I. SNOW and WILLIAM W. BACHOVCHIN of the pinanediol methylene bridge protons. This also revealed the configuration at boron to be as shown in Figure 2, resulting from attack by the nitrogen on the face of the boronate ring that bears the pinanediol methyl group. This is the expected diastereoisomer, since the methylene bridge of the pinanediol shields the face of the boronate ring adjacent to it and prevents the approach of the nitrogen from this side. The cyclic pinanediol esters of Ala-boroPro 16a and Pro-boroPro 16b showed liB NMR signals at 9.2 and 8.9 5, respectively. In the case of 16a, a NOE was observed between the two 0~ protons (Fig. 2), again consistent with the cyclic structure. Proof of the cyclic structure and stereochemistry was obtained by X-my crystallography. The cyclic form of the proline analog 16b crystallized as a chloroform solvate, shown in Figure 3. s~ This clearly shows a nitrogen-boron bond of 1.74 ,/k and the tetrahedral geometry of both boron and the proline nitrogen, with the S configu- ration in each case. The B-N bond length is consistent with the closest C(7) C(6) 0(3) N(2) C(8) C(9) B(1) C(14) C(1 0(1) C(13) C(15) N(1) C(1) C(4) Cc3) C(19) C(16) C(18) C(2) Figure 3. ORTEP drawing of the X-ray structure of 16b. The unit cell contains two independent molecules of chloroform, which were disordered, and are omitted for clarity. From Ref. 51.
`
`Page 12 of 29
`
`

`
`Boronic Acid Inhibitors of Dipeptidyl Peptidase IV 161 examples in the Cambridge Crystallographic Database. The configura- tion at boron is in agreement with that determined for 15c by NMR. To determine whether the cyclization is reversible, 16e was treated with acid in an organic solvent, whereupon it was reconverted to the salt of 15e. IX. STRUCTURAL STUDIES OF BORONIC ACIDS Similar experiments were performed on the free boronic acids. Although these could be isolated as solids either as 17 or 18, it was not possible to obtain suitable crystals for X-ray analysis. The lIB NMR spectrum of the cyclic form of the flee boronic acid of Val-boroPro (1Be) in D20 showed a single peak at 2.7 ~5, and ~H NMR showed the valine tx proton at 3.55 ~i. On acidification of the solution, the liB signal moved to 28 8, and the valine ct proton moved downfield to 4.14 ~i. Both these observations closely parallel the changes seen in the conversion of 16e to 15e. The IR spectra were also consistent with the structures shown; 17e displayed a strong band at 1350 cm -~, typical of a trigonal boronic acid, which was absent in 1Be. The solution structures and conformational behavior of Pro-boroPro 74 and Val-boroPro 75 have been rigorously defined by ~H NMR in H20. The preferred conformations and populations of each conformer were determined for the open and cyclic forms, using both the L and o diastereoisomers of boroPro, which also led to an independent determination of absolute configuration. The open form of Pro-boroPro 17b exists as a mixture of two conformers (Fig. 4A), which differ only in the flip of the Pro ring (referred to as North and South), both having the boroPro ring in a conformation where the boronic acid is pseudo- equatorial. Val-boroPro 17e (Fig. 4C) has a very similar conformation of the boroPro ring, and only one conformer of the Val residue is observed. The observed solution conformations are quite similar both to the solid-state conformations of lSb and 15e, s~ as well as to the calculated minimum energy conformation of several related dipeptides. 76 NOE assignments on lgb and lge provided conclusive evidence for the cyclic structures (Fig. 4, B and D). Both compounds exist as predomi- nantly one conformer in solution. The conformations observed for the boroPro ring and six-membered ring in lgb and lge are almost identical and are similar to the conformation of these rings in 16e, as determined by X-ray.
`
`Page 13 of 29
`
`

`
`B 50% Figure 4. Stereodiagrams of solution structures determined by NMR for (A) open Pro-boroPro 17b; (B) cyclic Pro-boroPro 18b, from Ref. 74; (C) open Val-boroPro 17c; (D) cyclic Val-boroPro 18c, from Ref. 75. 162
`
`Page 14 of 29
`
`

`
`Boronic Acid lnhibitors of Dipeptidyl Peptidase iV 163 X. STUDIES ON PINANEDIOL HYDROLYSIS Peptide boronic acid enzyme inhibitors have generally been prepared as boronate esters, with pinacol or pinanediol, which are assumed to hydro- lyze to the boronic acid. This has been demonstrated for the pinacol ester of a chymotrypsin inhibitor, 54 which gains activity on standing in aque- ous solution. Several thrombin inhibitors show potency as the pinanediol ester comparable to the boronic acid. 55,56 Kettner 55 proposed that this is due to hydrolysis but also noted the reported stability of pinanediol boronates. The DPP IV inhibitors also showed almost the same potency as either the pinanediol ester 15 or the boronic acid 17 (Table 3). We wished to determine whether the activity is due to hydrolysis to the boronic acid, or whether the esters themselves are active. 5~ This was studied initially by IH NMR using a solution of 15e in phosphate buffer in H20 at pH 7.8. Even at early time points, the formation of free pinanediol was apparent, along with peaks corresponding to the free boronic acid 17e. The amount ofpinanediol increased over several hours, and two new sets of peaks appeared, one of which was 16e, and the other was the cyclic boronic acid 1Be. After 24 hours, the only species present by NMR were 16e, 1Be, and pinanediol, and no further change was observed after this time. The extent of pinanediol formation was esti- mated to be 40-50%. Qualitatively similar results were obtained by using a gas chromatography assay to measure pinanediol; approximately 20% hydrolysis was observed after 5 minutes, which leveled off at 50--60% after 5 hours. From these observations we concluded that the process illustrated in Scheme 2 was occurring. Hydrolysis of pinanediol does occur but is incomplete. Both the boronate ester and the free boronic acid undergo cyclization at neutral pH, with the equilibrium strongly favoring the cyclic form. The cyclic pinanediol ester 16e does not interconvert directly with lge, since the mechanism for hydrolysis of a boronate presumably involves attack by water on the vacant orbital of boron as the first step. In 16e this coordination site is occupied by the amine. At lower pH (-3) cyclization does not occur, but the rate of hydrolysis as determined by ~H NMR is much slower than at pH 7.8. The extent of hydrolysis at pH 7.8 was surprising, in view of the reported stability of pinanediol boronates, n Since hydrolysis to the free boronic acid clearly occurs in water, it seemed likely that this could account for the inhibition seen with the pinanediol ester 15e. To confirm this the rate of inhibition of DPP IV by lSe and 17e was measured by using a continuous enzyme assay, in the presence of a large excess ofpinanediol, which should drive
`
`Page 15 of 29
`
`

`
`164 ROGER J. SNOW and WILLIAM W. BACHOVCHIN .o N -_- ~i~ H2 • " ~ H 2 + HO---J/ 0 0 /B-~oH /\ HO 11;c 17(; IT IT O~'~NH O~'~ NH= lllc 111c Scheme 2. Equilibrium established on dissolving Val-boroPro pi- nanediol ester 15¢ in water at pH 7.8. the equilibrium in favor of the boronate. ~j The results indicated very similar levels of inhibition by either 15(: or 17¢. Addition of 3 mM (+)pinanediol reduced the level of inhibition significantly, and to the same e~tent in each case, suggesting that 17c is considerably more inhibitory than the pinanediol ester 15c. Xl. NMR STUDIES OF CYCLIZATION RATES In order to understand the factors governing the rate of cyclization, the interconversion of the boronic acids between the open and cyclic forms was studied by ]H NMR, using solutions of 17a-c in phosphate buffer in D20, pD 7.8. 5~ In each case a smooth conversion from the protonated open form to the cyclic form was observed (Fig. 5). After 24 hours, equilibrium had been reached, with none of the open form detectable by NMR. Cyclization of Ala-boroPro 17a was rapid, (fi/2 = 44 minutes), whereas cyclization of Pro-boroPro (17b) and Val-boroPro (17e) was significantly slower (fin = 160 and 190 minutes, respectively). These rates, measured in D20, are slower than those measured in H20, either by NMR (fir2 = 30 minutes for Val-boroPro) 5° or by HPLC (t~r 2 = 35-55 6s minutes for Pro-boroPro). This is due to a solvent deuterium isotope effect, which leads to differences in pK~ irtD20 compared to H20. At pD 4 or below no change was seen, with either Ala-boroPro or Val-boroPro,
`
`Page 16 of 29
`
`

`
`E k 0 r- iB Q. 0 Boronic Acid Inhibitors of Dipeptidyl Peptidase IV 165 100' 80 60 40 2O I ~.N~ Vatee "-'~ ProBP .--e--- AlaBP - v- Dm A F ,% "11 ~ "-. \V "'",,.. T v ....... I ......................... ~ • " - .~-~. ............. ."" ..... 0 t 00 200 .300 400 500 600 700 800 Time (mill) Figure 5. ~H NMR measurement of the cyclization of 17a, 17b, and 17c, in 0.5 M phosphate in D20 at pD 7.8, 25°C. Curves indicate fit of data to Eq. 1 by nonlinear regression. From Ref. 51. but above pD 4 the cyclic form 18 started to appear. Pro-boroPro cyclizes to the extent of 15% even at pD 3, despite the higher pK~ of proline. By allowing the cyclization to reach equilibrium at a range of different pHs, 77 it was shown that the tendency to cyclize was greatest for Pro-boroPro and least for Val-boroPro. Analogous NMR experiments were carried out storting from the cyclic compounds. 5~ As expected, no change was seen at pD 7.8. On dissolving 18a or lge in D20 at pD 3.0, disappearance of the cyclic form was almost immediate. In addition to the open trans form, a new species was observed, which is thought to be the protonated, open cis form (cAH in Scheme 3). This is assumed to be formed rapidly as the B-N bond is broken by protonation, and then isomerizes to trans over time. In the case of 18b, ring opening was markedly slower, with the cyclic form still present after 2 hours. From these observations the mechanism in Scheme 3 was proposed. The protonated dipeptides exist predominantly