J. Med. Chem. 2004, 47, 2587-2598
`
`2587
`
`Synthesis of Novel Potent Dipeptidyl Peptidase IV Inhibitors with Enhanced
`Chemical Stability: Interplay between the N-Terminal Amino Acid Alkyl Side
`Chain and the Cyclopropyl Group of
`r-Aminoacyl-L-cis-4,5-methanoprolinenitrile-Based Inhibitors
`
`David R. Magnin,† Jeffrey A. Robl,*,† Richard B. Sulsky,† David J. Augeri,†,^ Yanting Huang,†
`Ligaya M. Simpkins,† Prakash C. Taunk,† David A. Betebenner,†,# James G. Robertson,‡ Benoni E. Abboa-Offei,‡
`Aiying Wang,‡ Michael Cap,‡ Li Xin,‡ Li Tao,[ Doree F. Sitkoff,¥ Mary F. Malley,O Jack Z. Gougoutas,O
`Ashish Khanna,§ Qi Huang,‡ Song-Ping Han,‡ Rex A. Parker,‡ and Lawrence G. Hamann*,†
`Departments of Discovery Chemistry, Metabolic Research, Exploratory Pharmaceutics, Computer-Assisted Drug Design,
`Solid State Chemistry, and Pharmaceutical Candidate Optimization, Bristol-Myers Squibb, Pharmaceutical Research Institute,
`P.O. Box 5400, Princeton, New Jersey 08543-5400
`
`Received January 27, 2004
`
`A series of methanoprolinenitrile-containing dipeptide mimetics were synthesized and assayed
`as inhibitors of the N-terminal sequence-specific serine protease dipeptidyl peptidase IV (DPP-
`IV). The catalytic action of DPP-IV is the principle means of degradation of glucagon-like
`peptide-1, a key mediator of glucose-stimulated insulin secretion, and DPP-IV inhibition shows
`clinical benefit as a novel mechanism for treatment of type 2 diabetes. However, many of the
`reversible inhibitors to date suffer from chemical instability stemming from an amine to nitrile
`intramolecular cyclization. Installation of a cyclopropyl moiety at either the 3,4- or 4,5-position
`of traditional 2-cyanopyrrolidide proline mimetics led to compounds with potent inhibitory
`activity against the enzyme. Additionally, cis-4,5-methanoprolinenitriles with (cid:226)-branching in
`the N-terminal amino acid provided enhanced chemical stability and high inhibitory potency.
`This class of inhibitors also exhibited the ability to suppress prandial glucose elevations after
`an oral glucose challenge in male Zucker rats.
`
`Introduction
`With the spread of Western lifestyles, the prevalence
`of type 2 diabetes in the world’s population is rising.1
`Current treatment strategies include reducing insulin
`resistance using glitazones,2 supplementing insulin
`supplies with exogenous insulin,3 increasing insulin
`secretion with sulfonylureas,4 reducing hepatic glucose
`output with biguanides,5 and limiting glucose absorption
`with glucosidase inhibitors.6 Promising new targets for
`drug development are also emerging. Of particular
`interest is the pharmacology surrounding the incretin
`hormone glucagon-like peptide 1 (GLP-1).7 GLP-1 is
`known to function as a mediator of glucose-stimulated
`insulin secretion, and several clinical studies have
`shown that administration of the peptide or its ana-
`logues results in antidiabetic action in subjects with type
`2 diabetes.8 Although GLP-1 is secreted as GLP-1 (7-
`36) amide from the small and large intestines in
`response to dietary signals, it is rapidly truncated to
`
`* To whom correspondence should be addressed. For J.A.R.: tele-
`phone, 609-818-5048; fax, 609-818-3550; e-mail, jeffrey.robl@bms.com.
`ForL.G.H.: telephone,609-818-5526;fax,609-818-3550;e-mail: lawrence.
`hamann@bms.com.
`† Department of Discovery Chemistry.
`^ Present address: Lexicon Pharmaceuticals, 350 Carter Road,
`Princeton, NJ 08540.
`# Present address: Pharmaceutical Discovery Division, Abbott
`Laboratories, Abbott Park, IL 60064.
`[ Department of Exploratory Pharmaceutics.
`¥ Department of Computer-Assisted Drug Design.
`O Department of Solid State Chemistry.
`§ Department of Pharmaceutical Candidate Optimization.
`‡ Department of Metabolic Research.
`
`GLP-1 (9-36) by cleavage of the N-terminal dipeptide
`residues. The truncated metabolite has antagonist
`activity against the GLP-1 receptor both in vitro and
`in vivo.9 The principle enzyme responsible for the
`cleavage of GLP-1 (7-36) amide to GLP-1 (9-36) amide
`is dipeptidyl peptidase IV (DPP-IV, EC 3.4.14.5), a
`nonclassical, sequence-specific serine protease that
`catalyzes the cleavage of dipeptides from the N-termi-
`nus of proteins with the sequence H-X-Pro-Y or H-X-
`Ala-Y (where X, Y ) any amino acid; Y * Pro).10
`Inhibition of DPP-IV has been shown to be effective at
`sustaining circulating levels of GLP-1 (7-36) and
`therefore offers a new therapeutic approach for the
`treatment of type 2 diabetes.11
`Early reports of DPP-IV inhibitors included proline-
`based dipeptide mimics bearing boronic acid12 (1) or
`diphenyl phosphonate substituents (2).13 These com-
`pounds were irreversible inhibitors of DPP-IV or were
`slow to dissociate from the enzyme. Several first-
`generation dipeptide surrogates have been disclosed as
`reversible inhibitors of DPP-IV,
`including both C-
`substituted (3) and N-substituted (4) glycinylprolineni-
`trile dipeptide analogues.14,15 These compound classes
`include many potent inhibitors of the enzyme, but all
`suffer from chemical instability whereby the N-terminal
`amine intramolecularly cyclizes onto the nitrile, forming
`inactive cyclic imidates and/or their diketopiperazine
`hydrolysis products. However, more recent publications
`have disclosed a series of more hindered N-alkylamines
`(5) that have much greater chemical stability.16 Thia-
`
`10.1021/jm049924d CCC: $27.50 © 2004 American Chemical Society
`Published on Web 04/13/2004
`
`AstraZeneca Exhibit 2002
`Mylan v. AstraZeneca
`IPR2015-01340
`
`Page 1 of 12
`
`

`
`2588 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 10
`
`Magnin et al.
`
`Scheme 1a
`
`Figure 1. Known proline-derived dipeptidyl peptidase IV
`inhibitors.
`
`zolidide (6) and pyrrolidide (7) based inhibitors have also
`been disclosed that lack the nitrile moiety and thus
`possess greater stability, though at the cost of reduced
`potency17a-c with a few very recent notable excep-
`tions.17d-f These compounds are shown in Figure 1.
`We endeavored to synthesize novel dipeptide sur-
`rogates containing a proline mimic linked to an N-
`terminal amino acid that would act as reversible
`inhibitors of DPP-IV with maximum potency and en-
`hanced solution stability. Hanessian has reported that
`installation of a 4,5-methano moiety into a proline
`residue has the effect of flattening the five-membered
`ring,18 and one might extend a similar conformational
`argument to other cyclopropanated prolines as well. By
`inference, the cyclopropane bridge may occupy the
`space-filling region that would normally be given to the
`methylene group in the puckered or “envelope” confor-
`mation of a typical five-member ring. With this prece-
`dent we sought to establish whether a methanoproline
`derivative could serve as a viable proline surrogate in
`a DPP-IV inhibitor. Our strategy was to prepare dipep-
`tide surrogates containing a cyclopropanated prolineni-
`trile derived from L-proline. The regio- and stereochem-
`ical disposition of the cyclopropane bridge was varied
`in order to identify a compound with maximal stability
`and potency. Herein, we report the discovery of L-cis-
`4,5-methanoprolinenitrile dipeptides as potent inhibi-
`tors of DPP-IV with increased chemical stability and
`high potency. In addition, we also present data demon-
`strating that these novel DPP-IV inhibitors effectively
`lower plasma glucose after a glucose challenge in rodent
`models.
`
`Chemistry
`Dipeptides (12-15) composed of N-terminal isoleucine
`appended to cyclopropylprolineamides or nitriles derived
`from L-proline (8-11) were targeted to probe the
`potential for this type of inhibitor scaffold. Isoleucine
`was selected as the N-terminal residue because it was
`the most potent natural amino acid reported in the
`2-cyanopyrrolidide series.14a These inhibitors were ex-
`pected to provide a dependable inhibitory benchmark
`for the differing methanoproline structures. The syn-
`thetic routes used to prepare the dipeptides derived from
`L-cis- and L-trans-4,5-methanoproline,18b L-cis-3,4-meth-
`
`a (a) N-Boc-amino acid, EDAC, DMAP or PyBop, 50-90%; (b)
`(for intermediates 8-10) POCl3, pyridine, imidazole, 70-90%; (c)
`TFA, CH2Cl2, TFA or HCl in Et2O or EtOAc, 70-90%.
`
`anoproline,19 and L-2,3-methanoproline20 are shown in
`Scheme 1.
`Standard conditions were employed to couple the
`enantiomerically pure L-cis-4,5-cyclopropylprolineamide
`nucleus (8) to give the corresponding Boc-protected
`dipeptides in good to excellent yield.21 Dehydration14b,22
`(POCl3, pyridine, imidazole) of the amide and removal23
`of the N-terminal Boc protecting group (TFA or HCl)
`gave the proline dipeptide (12) in high yield. Similar
`protocols were used for the L-cis-3,4-methanoproline
`fragment 10 and the L-4,5-trans-methanoproline com-
`pound 9. The known (-)-(2S,3R)-methanoprolinenitrile
`(11) was prepared according to the method of Hercout.20a
`In this instance, the dehydration reaction was per-
`formed prior to introduction of the isoleucine fragment.
`Preparation of a library was undertaken to probe
`structure-activity relationships between the N-termi-
`nal amino acid residue and the cis-4,5-methanopro-
`linenitriles. These inhibitors were generated in a three-
`step sequence in parallel array format in a manner
`similar to that described in Scheme 1. Initial reaction
`of the Boc-protected amino acid with methanoprolinea-
`mide, EDAC, and DMAP in dichloromethane, and
`subsequent purification through an SCX ion exchange
`cartridge, gave good to excellent yields of coupled
`dipeptides. Dehydration and acid-promoted deprotection
`(TFA in dichloromethane) yielded the inhibitors as TFA
`salts. Further purification of the final products was
`easily accomplished by preparative reverse-phase HPLC
`or by trituration with Et2O.
`Holmberg has prepared tert-alkylmalonic acid deriva-
`tives through a TiCl4-mediated Knoevenagel process
`and subsequent copper-assisted conjugate addition of a
`Grignard reagent or conjugate reduction. Following this
`method (Scheme 2), diesters can be converted to the
`protected amino acids in three steps.24 This methodology
`provided an approach to the cycloalkylglycine amino
`acid derivatives 20. First, the malonic acid diesters 17
`were subjected to conjugate addition of a methyl group
`or hydride and then hydrolyzed to the corresponding
`monoacids 18. Subsequent Curtius rearrangement of 18
`by treatment with diphenylphosphoryl azide, followed
`by trapping of the intermediate isocyanate with benzyl
`alcohol, provided the Cbz adducts 19. Finally, the esters
`were hydrolyzed to give the amino acids 20 in racemic
`form. The racemic Boc-protected acids 20 were then
`coupled to the L-cis-4,5-methanoprolineamide group.
`
`Page 2 of 12
`
`

`
`Dipeptidyl Peptidase IV Inhibitors
`
`Journal of Medicinal Chemistry, 2004, Vol. 47, No. 10 2589
`
`Scheme 2a
`
`Table 1. In Vitro Inhibition Constants for Porcine DPP-IV and
`Solution Stability Half-Lives for Prolinenitrile DPP-4 Inhibitors
`
`a (a) TiCl4, THF, CCl4, diethyl malonate, 0 °C, then pyridine, 0
`°C to room temp, 68%; (b) MeMgI, CuCl, Et2O, 0 °C, 69%; (c)
`NaOH, THF, EtOH, 78%; (d) (i) diphenylphosphorylazide, NEt3,
`PhH, reflux; (ii) BnOH, reflux, 18 h, 100%; (e) H2, Pd(OH)2, EtOAc,
`100%; (f) (Boc)2O, K2CO3, THF, H2O, 92%, two steps; (g) NaOH,
`MeOH, THF, room temp, then aqueous HCl, 95%.
`
`Processing as previously described (Scheme 1) gave a
`mixture of diastereomers that were separated at the
`nitrile stage by silica gel flash chromatography. The
`later-eluting isomer in each case was consistently
`identified as the desired L,L-dipeptide.25
`
`Results and Discussion
`
`All compounds were tested in vitro against purified
`porcine DPP-IV. Inhibition was determined against the
`substrate H-Ala-Pro-pNA. Production of p-nitroaniline
`(pNA) was measured at 405 nm over 15 min. A com-
`parison of methanoproline-derived isoleucine N-termi-
`nal dipeptides shows that the enzyme inhibitory activity
`is critically dependent on both the position and the
`orientation of the methano bridge (Table 1). The pres-
`ence of the methano bridge on the trans side of the
`prolinenitrile (e.g., 22, 23) resulted in a significant loss
`of activity relative to the unsubstituted prolinenitrile
`21. This result was consistent with early reports where
`insertion of a methyl group at the 2-postion of 2-cyan-
`opyrrolidides resulted in a 2000-fold decrease in activ-
`ity.26 In contrast, when the methano bridge was oriented
`cis to the nitrile at either the 4,5- or 3,4-position,
`inhibitory activity was only slightly diminished. This
`is clearly illustrated with cis-4,5-methano inhibitors 24
`and 26 and cis-3,4-methano inhibitors 25 and 27, where
`inhibitory potency resides in the 20 nM range. A more
`complete exploration of N-terminal amino acids reveals
`that increasing the degree of (cid:226)-branching to that of tert-
`Leu (e.g., 29) further increases potency in the 4,5-
`methano series to that of the prolinenitrile version (28).
`Replacement of these alkyl substituents with aromatic
`residues such as Phe (31) or Trp (32) in the N-terminal
`position significantly eroded potency. Alkyl substitution
`on the terminal amine was generally ill-tolerated in this
`series (e.g., 33-35), though it would appear that the
`relatively more accessible N-terminus found in proline
`derivative 34 is capable of restoring a modest degree of
`inhibitory activity. Replacement of Leu and tert-Leu side
`chains with medium-ring cycloalkyl and methylcy-
`cloalkyl (37-42) generally led to inhibitors with com-
`parable or slightly improved activity, though potency
`dropped steadily with increasing larger ring size (data
`not shown). The Cp-Gly and Me-Cp-Gly derived ana-
`logues 38 and 41 were among the most potent compound
`prepared in this series, exhibiting Ki values of 4 and 7
`nM, respectively.
`
`N-terminal
`amino acida
`
`stability,
`t1/2 (h)c
`5
`
`22
`4
`28
`2
`27
`42
`4
`
`substituted
`prolinenitrile Ki (nM)b
`compd
`prolinenitrile 2 ( 0.5
`21
`Ile
`1620 ( 80
`22
`trans-4,5-
`Ile
`7500 ( 200
`23
`trans-2,3-
`Ile
`25 ( 1
`24
`cis-4,5-
`Ile
`15 ( 1
`25
`cis-3,4-
`Ile
`29 ( 1
`26
`cis-4,5-
`Val
`12 ( 1
`27
`cis-3,4-
`Val
`prolinenitrile 8 ( 0.5
`28
`tert-Leu
`7 ( 0.5
`29
`cis-4,5-
`tert-Leu
`14 ( 1
`30
`cis-3,4-
`tert-Leu
`65 ( 3
`31
`cis-4,5-
`Phe
`230 ( 10
`32
`cis-4,5-
`Trp
`1940 ( 80
`33 N-Me-Val
`cis-4,5-
`107 ( 5
`34
`cis-4,5-
`Pro
`35
`cis-4,5-
`10000
`Pip
`135 ( 10
`36 Met
`cis-4,5-
`12 ( 0.5
`37
`cis-4,5-
`Cb-Gly
`4 ( 0.5
`38
`cis-4,5-
`Cp-Gly
`15 ( 1
`39
`cis-4,5-
`Ch-Gly
`11 ( 0.5
`40
`(1-Me-Cb-1-yl)-Gly cis-4,5-
`7 ( 0.5
`41
`(1-Me-Cp-1-yl)-Gly cis-4,5-
`8 ( 0.5
`42
`(1-Me-Ch-1-yl)-Gly cis-4,5-
`a Cb ) cyclobutyl; Cp ) cyclopentyl; Ch ) cyclohexyl. b Values
`represent the mean ( SEM and are at least triplicate determina-
`tions. c Solution stability data are measured at 39.5 °C and pH
`7.2 in phosphate buffer.
`
`19
`
`24
`
`Scheme 3
`
`Solution Stability. After the discovery that the L-cis-
`4,5- and L-cis-3,4-methanoprolinenitriles were potent
`inhibitors of DPP-IV in vitro, a comparative study was
`initiated to investigate aqueous solution stability of
`these analogues. The N-terminal valine, isoleucine, and
`tert-leucine dipeptide nitriles were selected in order to
`make direct comparisons between methanoprolinenitrile
`and unsubstituted prolinenitrile compounds with re-
`spect to solution stability. Reaction rates were moni-
`tored by following the disappearance of starting mate-
`rial on reverse-phase HPLC at pH 7.2 and 39.5 °C in
`phosphate buffer. HPLC mass spectral analysis revealed
`that the two major products that formed during the
`stability experiments had either an identical mass or
`an M + 1 mass to the parent starting material. These
`data are consistent with the formation of intramolecular
`cyclization products, with the initial cyclic imidate (X
`) NH) surrendering to the diketopiperazine (X ) O)
`upon hydrolysis (Scheme 3).
`Two interesting aspects of the solution half-life data
`should be noted. The first is that there is increased
`solution stability associated with the cis-4,5-methano-
`prolinenitrile derivatives when compared to either the
`
`Page 3 of 12
`
`

`
`2590 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 10
`
`Magnin et al.
`
`cis-3,4-methanoprolinenitrile or the unadorned pro-
`linenitrile compounds. For instance, in the case where
`the N-terminal amino acid is isoleucine, the isomeric
`4,5-methano-substituted compound 24 has a solution
`half-life of 22 h, which is 5.5-fold longer than the
`analogous 3,4-methano-substituted compound 25 and
`approximately 4.5-fold longer than the unadorned pro-
`linenitrile analogue 21. The second feature is that there
`is a strong correlation between steric size at the (cid:226)-posi-
`tion of the N-terminal amino acid and relative solution
`stability. Increasing the degree of branching at the
`(cid:226)-position of the alkylglycine substituent increases
`solution stability. The most stable compounds within
`their respective prolinenitrile series have the tert-Leu
`N-terminal amino acid fragment in common. For ex-
`ample, unadorned prolinenitrile 28 (t1/2 ) 27 h) is more
`stable than its less branched isomer 21 (t1/2 ) 5 h).
`Similarly, cis-4,5-methanoprolinenitrile 29 (t1/2 ) 42 h)
`is more stable than the less branched dipeptidenitriles
`24 and 39 (t1/2 ) 22 and 19 h, respectively), though the
`magnitude of the effect of increased (cid:226)-branching in this
`series appears to be blunted by the inherent baseline
`stability imparted by the methano bridge. This observa-
`tion is in agreement with data reported by Coutes12a
`and Snow27 where the rates of cyclization for Xaa-
`boroproline dipeptides were shown to be Gly-BoroPro
`> Ala-BoroPro > Val-BoroPro.
`Computational Analysis. Computational analysis
`was undertaken to more fully understand the relative
`stabilities of the methanoprolinenitriles. Ground-state
`conformations were generated for methanoprolinenitrile
`and prolinenitrile forms of the N-terminal tert-leucine
`dipeptide compounds. The calculated ground-state struc-
`ture for the tert-leucine dipeptidenitrile is identical to
`the conformation observed through single-crystal X-ray
`structural analysis28 of the TFA salt of 29 (rms ) 0.1 Å
`for calculated and observed heavy atoms; see Figure 2,
`lower structure). Both have the same syn conformation
`around the amide bond, characterized by a small C(2)-
`N-C(8)dO torsional angle (-5° and +2° for the TFA
`structure). It is of additional interest that a similar
`conformation has been observed in several recently
`disclosed cocrystal structures of DPP-IV/inhibitor com-
`plexes.29
`In addition to the syn conformation, there is a
`calculated local low-energy minimum where the reactive
`amine and nitrile are close to each other; in this anti
`conformation (Figure 2, upper structure), the C(2)-N-
`C(8)dO torsional angle is 180°. Moreover, the angle
`between the amine N and the C(cid:17)N group is 109° ( 1°
`and the distance between the these reactive partners
`is 2.95 Å. It is therefore reasonable to assume that the
`observed intramolecular cyclization is initiated from this
`conformation. The value of 109° between the amine
`group and the nitrile is in close agreement with the
`hypothetical angle of attack of at least 108° reported
`by Baxter and Connor.30 It was envisioned that the
`relative energetic differences between the global mini-
`mum and the reactive local minimum would represent
`a means to evaluate the relative stabilities of compounds
`in solution.
`The calculated conformations and their relative ener-
`gies can be used to examine the basis for two aspects of
`the experimental compound stability: the increase in
`
`Figure 2. The upper structure depicts the local low-energy
`minimum for 29 where the reactive amine is close to the nitrile
`(anti conformation). The lower structure is the solid-state
`conformation observed in the X-ray crystallographic structure
`of the TFA salt of compound 29 (syn conformation). Two
`independent cations were observed in the crystal structure,
`though because these have identical conformations, only one
`is shown for purposes of clarity. Carbon atoms C(2) and C(8)
`are labeled.
`
`stability due to side chain bulk (specifically (cid:226)-branching)
`and the increase in stability upon conversion from
`prolinenitrile to cis-4,5-methanoprolinenitrile. Calcu-
`lated relative energies between the ground-state con-
`formation and the local low-energy minimum confor-
`mation that brings the reactive amine and nitrile in
`proximity are presented in Table 2 for the methanopro-
`linenitrile and prolinenitrile forms of N-terminal tert-
`leucine and alanine dipeptidenitriles, as well as for the
`same prolinenitriles with a simple acetamide cap.
`Conformational Stability Due to Side Chain
`Bulk. The values in the first column of Table 2 compare
`the conformational energy differences between the
`ground state and the geometry where internal cycliza-
`tion could occur for the unsubstituted prolinenitrile
`series. The ab initio (G98) results are expected to be
`considerably more accurate than the force field values
`and indicate that the energy required to assume the anti
`conformation increases with side chain bulk (e.g., 0.3,
`1.9, 2.8 kcal/mol for no side chain, the alanine side
`chain, and the tert-leucine side chain, respectively). The
`force field energy results agree qualitatively with the
`ab initio values and suggest that the primary contribu-
`tion is due to van der Waals interactions. Examination
`of the structures reveals extremely close contacts be-
`tween two of the side chain methyl groups and the
`carbonyl oxygen in the anti conformation (approxi-
`mately 3 Å each) that would increase the energy barrier
`for internal cyclization and thus lead to greater stability.
`A similar trend is observed for increasing side chain
`bulk in the methano-substituted series (column 2), and
`these values are in good agreement with the solution
`data where increased (cid:226)-branching enhances stability.
`Conformational Stability Due to Methano Sub-
`stitution. As shown in the third column of Table 2, the
`ab initio (G98) energy calculations indicate that the
`
`Page 4 of 12
`
`

`
`Dipeptidyl Peptidase IV Inhibitors
`
`Journal of Medicinal Chemistry, 2004, Vol. 47, No. 10 2591
`
`Table 2. Calculated Energya Differences between the
`Calculated Ground State and the Local Low-Energy Minimum
`Required for Cyclization
`
`28 (Prolinenitrile) and 29 (Methanoprolinenitrile)
`(A ) tert-Leu)
`¢H(28) b ¢H(29) b ¢¢H(29 - 28)
`2.8
`3.4
`0.6
`
`3.5
`3.3
`0.2
`
`4.2
`3.9
`0.3
`
`0.7
`0.6
`0.1
`
`method interaction
`G98 ab initio
`Insight CFF force field
`total nonbond
`van der Waals
`electrostatic
`
`43 (Prolinenitrile) and 44 (Methanoprolinenitrile)
`(A ) Ala)
`¢H(43) b ¢H(44) b ¢¢H (44 - 43)
`1.9
`2.6
`0.6
`
`3.3
`2.9
`0.4
`
`3.9
`3.4
`0.4
`
`0.6
`0.6
`0.0
`
`method interaction
`G98 ab initio
`Insight CFF force field
`total nonbond
`van der Waals
`electrostatic
`
`45 (Prolinenitrile) and 46 (Methanoprolinenitrile)
`(A ) Ac)
`¢H(45) b ¢H(46) b ¢¢H (46 - 45)
`0.3
`0.9
`0.6
`
`method interaction
`G98 ab initio
`Insight CFF force field
`-0.2
`0.2
`0.0
`total nonbond
`-0.2
`-0.4
`0.2
`van der Waals
`0.1
`0.3
`0.2
`electrostatic
`a Energies are for the cis-4,5-methanoprolinenitrile and pro-
`linenitrile versions of the compounds and are given in kcal/mol.
`b Energies are for the anti conformation (Scheme 3 and Figure 2
`upper structure) relative to the global minimum conformation,
`which corresponds to the syn geometry in Scheme 3 and Figure 2
`lower structure.
`
`addition of the methano bridge increases the energy
`barrier toward adopting the conformation required for
`cyclization by approximately 0.6 kcal/mol for the N-
`terminal tert-leucine dipeptide compound as well as for
`the alanine dipeptide and acetamide compounds. The
`lack of dependence on the nature of the side chain
`suggests that the observed increase in stability of the
`cis-4,5-methano compounds is due to a local interaction
`between the methano bridge and the N-terminal amino
`acid that favors the ground state relative to the anti
`conformation where cyclization is initiated. The force
`field energy analysis (Insight CFF) results also agree
`qualitatively, implying that van der Waals interactions
`are primarily responsible for the increase in energy.
`These conclusions are supported by the cyclization rate
`differences between the isomeric isoleucine derivatives
`24 (t1/2 ) 22 h) and 21 (t1/2 ) 5 h), where a 4.5-fold
`increase in stability is observed for the methanopro-
`linenitrile. This increase in stability is in fair agreement
`with the calculated value of 0.6 kcal/mol. The calculated
`¢¢H and observed solution half-life results support
`contributions to stability from both (cid:226)-branching and the
`presence of the 4,5-methano bridge on the pyrrolidine
`ring.
`In Vivo Activity. In vivo evaluation of DPP-IV
`inhibitors has supported the connection between DPP-
`
`Figure 3. Effects of inhibitors 29 (b) and 41 (9) dosed at 3
`(cid:237)mol/kg po versus vehicle control (O) on plasma DPP-IV
`activity in Zuckerfa/fa rats.
`
`IV inhibition, increases in plasma insulin levels, and
`an improvement in glucose tolerance.31 Compounds 29
`and 41 were potent inhibitors of DPP-IV in vitro and
`demonstrated excellent solution stability. As such, these
`inhibitors were selected to determine the effects of DPP-
`IV inhibition in vivo on glucose tolerance and insulin
`secretion in Zuckerfa/fa rats. An oral glucose tolerance
`test (oGTT) in the Zuckerfa/fa rat is a frequently used
`model of hyperglycemia in type 2 diabetes and obesity
`research. Zuckerfa/fa rats are severely hyperphagic,
`extremely obese, markedly insulin-resistant, and mildly
`hyperglycemic because of a mutation and loss of function
`of the leptin receptor gene.32,33 Fasted male Zuckerfa/fa
`rats were dosed orally with water or with inhibitors 29
`and 41 (3 (cid:237)mol/kg), and an oGTT was conducted 0.5 h
`after the dosing. Plasma DPP-IV activity, glucose, and
`insulin levels were then monitored over a 2 h period.
`Figures 3-5 show the ex vivo plasma DPP-IV activity,
`insulin response, and glucose excursion curves in re-
`sponse to an oral glucose challenge (2 g/kg). Animals in
`the control group reached peak plasma glucose levels
`60 min after glucose administration, at which point the
`drug-treated animals exhibited a 30-35% decrease in
`glucose levels compared to controls (control animals, 356
`mg/dL; compound 29 treated animals, 226 mg/dL;
`compound 41 treated animals, 245 mg/dL). Glucose
`levels were significantly reduced in the drug-treated
`animals from 30 min onward, with maximal reductions
`in glucose observed at 90 min (-34% to -38%). Plasma
`DPP-IV activity was maximally suppressed (60%) 30
`min after dosing (Figure 3), and the effects of these
`inhibitors were sustained throughout the course of the
`experiment (60-35%). The insulin response to oral
`glucose was also enhanced by treatment with DPP-IV
`inhibitors (Figure 4), demonstrating the link between
`the glucose-lowering effects and DPP-IV inhibition of
`these compounds.
`Interestingly, a significant decrease in plasma glucose
`levels occurred when DPP-IV activity in plasma was
`inhibited only by 35-60%. This finding suggests that
`it may not be necessary to completely suppress plasma
`DPP-IV activity in order to achieve antihyperglycemic
`efficacy in type 2 diabetics. However, differences may
`exist in the inhibition of the turnover of native sub-
`strates such as GLP-1 compared with that of the
`pseudosubstrate used in this assay, and potential assay
`
`Page 5 of 12
`
`

`
`2592 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 10
`
`Magnin et al.
`
`Experimental Section
`General Chemical Procedures. All reactions were carried
`out using oven-dried or flame-dried round-bottomed (rb) flasks
`and glassware under a static atmosphere of argon or nitrogen,
`and the mixtures were stirred magnetically unless otherwise
`noted. All reagents used were of commercial quality and were
`obtained from Aldrich Chemical Co., Sigma Chemical Co.,
`Lancaster Chemical Co., or Acros Chemical Co. All reactions
`were carried out using commercially available anhydrous
`solvents from Aldrich Chemical Co. or EM Science Chemical
`Co. unless otherwise noted. “Brine” refers to a saturated
`aqueous solution of NaCl. Unless otherwise specified, solutions
`of common inorganic salts used in workups are aqueous
`solutions. 1H (500 MHz) and 13C (125 MHz) NMR spectra were
`recorded on a JEOL JNM-ECP500 spectrometer, and 1H (400
`MHz) and 13C (100 MHz) spectra were recorded on a JEOL
`GSX400 spectrometer. Chemical shifts are given in parts per
`million (ppm) downfield from internal reference tetramethyl-
`silane in (cid:228) units, and coupling constants (J values) are given
`in hertz (Hz). Selected data are reported in the following
`manner:
`chemical shift; multiplicity; coupling constants;
`assignment. Elemental analyses were performed by the Ana-
`lytical Chemistry department at Bristol-Myers Squibb. Melting
`points were taken on a Hoover Uni-melt melting point ap-
`paratus and are uncorrected. Boiling points are reported
`uncorrected. Ku¨ gelrohr distillations were performed using a
`Bu¨ chi GKR-51 apparatus, and reported boiling points cor-
`respond to uncorrected oven air bath temperatures. Optical
`rotations were obtained on a Perkin-Elmer 241 polarimeter
`using a cell 1 dm in length and are reported as follows:
`[R]temp
`wavelength (concentration in g/100 mL, solvent). All flash
`chromatographic separations were performed using E. Merck
`silica gel (particle size, 0.040-0.063 mm). Reactions were
`monitored by TLC using 0.25 mm E. Merck silica gel plates
`(60 F254) and were visualized using UV light and 5% phospho-
`molybdic acid in 95% EtOH, or by sequential treatment with
`1 N HCl in CH3OH followed by ninhydrin staining. LC/MS
`data were recorded on a Shimadzu LC-10AT equipped with
`an SIL-10A injector, an SPD-10AV detector normally operating
`at 220 nm and interfaced to a Micromass ZMD mass spec-
`trometer. LC/MS or HPLC retention times are reported using
`a Phenomenex Luna C-18 4.6 mm (cid:2) 50 mm column, with
`elution with a 4 min gradient of 0-100% solvent B, where
`solvent A is 10:90:0.1 CH3OH-H2O-TFA and solvent B is 90:
`10:0.1 CH3OH-H2O-TFA (HPLC, method 1). Other HPLC
`methods include the following: method 2, YMC S-5 C-18 4.6
`mm (cid:2) 50 mm, 0-100% B, elution with a 4 min gradient of
`0-100% solvent B, where solvent A is 10:90:0.1 CH3CN-H2O-
`TFA and solvent B is 90:10:0.1 CH3CN-H2O-TFA; method
`3, Zorbax SB C-18 4.6 mm (cid:2) 75 mm column, elution with an
`8 min gradient of 0-100% solvent B, where solvent A is 10:
`90:0.1 CH3OH-H2O-H3PO4 and solvent B is 90:10:0.1 CH3-
`OH-H2O-H3PO4. All solvents were removed by rotary evapo-
`ration under vacuum using a standard rotary evaporator
`equipped with a dry ice condenser. All
`filtrations were
`performed using vacuum unless otherwise noted.
`Representative Example of Preparation: General
`Method A. N-Boc-L-cis-4,5-methanoprolineamide. To a
`solution of N-Boc-4,5-methanoproline18b (1.20 g, 5.28 mmol)
`in THF (20 mL) at -15 °C was added 4-methylmorpholine
`(0.71 mL, 6.50 mmol) and then isobutyl chloroformate (0.78
`mL, 6.00 mmol) over 5 min. The reaction mixture was stirred
`at -15 °C for 30 min, cooled to -30 °C, and then treated with
`a solution of NH3 in dioxane (50 mL, 25 mmol). The reaction
`mixture was stirred at -30 °C for 0.5 h, warmed to room
`temperature, and stirred overnight. The reaction mixture was
`quenched with citric acid solution (pH 4) and extracted with
`Et2O (3(cid:2) 50 mL). The combined organic fractions were washed
`with brine, dried (Na2SO4), and concentrated. The residue was
`purified by flash column chromatography on silica gel with
`EtOAc to give 1.00 g (84%) of the title compound. 1H NMR
`and 13C NMR signals were very broad and poorly defined as a
`result of carbamate rotamers. Anal. (C11H18N2O(cid:226)0.4H2O) C, H,
`N.
`
`Figure 4. Effects of inhibitors 29 (b) and 41 (9) dosed at 3
`(cid:237)mol/kg po versus vehicle control (O) on plasma glucose after
`an oGTT in Zuckerfa/fa rats.
`
`Figure 5. Effects of inhibitors 29 (b) and 41 (9) dosed at 3
`(cid:237)mol/kg po versus vehicle control (O) on plasma insulin after
`an oGTT in Zuckerfa/fa rats.
`
`artifacts relating to the kinetic aspects of certain inhibi-
`tors cannot be ruled out at present.
`
`Conclusion
`We have demonstrated that the prolinenitrile frag-
`ment of previously reported DPP-IV inhibitors can be
`replaced with either a cis-3,4-methano- or a cis-4,5-
`methanoprolinenitrile ring system to provide novel and
`highly potent DPP-IV inhibitors. Solution stability
`studies demonstrate that introduction of either a steri-
`cally bulky amino acid side chain on the N-terminal
`amino acid or a cis-4,5-methano bridge to the prolineni-
`trile moiety significantly enhances solution stability,
`minimizing a known intramolecular cyclization path-
`way. The greatest improvement in stability is observed
`when both of these structural features are present and
`at work in concert. This added stability has the potential
`to beneficially impact the chemical and formulation
`stability of cyanopyrrolidide-based pharmaceuticals. In
`many cases, the presence of a (cid:226)-branched amino acid
`also provides increased inhibitory potency as well as
`solution sta