J. Med. Chem. 2005, 48, 5025-5037
`
`5025
`
`Discovery and Preclinical Profile of Saxagliptin (BMS-477118): A Highly Potent,
`Long-Acting, Orally Active Dipeptidyl Peptidase IV Inhibitor for the Treatment
`of Type 2 Diabetes
`
`David J. Augeri,*tt Jeffrey A. Robl,t David A. Betebenner,t David R. Magnint Ashish Khanna,
`James G. Robertson, Aiying Wang,' Ligaya M. Simpkins,t Prakash Taunk,t Qi Huang,- Song-Ping Han,-
`Benoni Abboa-Offei,'I Michael Cap, Li Xin,- Li Tao,# Efflie Tozzo,- Gustav E. Welzel, Donald M. Egan,'
`Jovita Marcinkeviciene," Shu Y. Chang," Scott A. Biller,t Mark S. Kirby,' Rex A. Parker,' and
`Lawrence G. Hamaim*
`
`Departments of Discovery Chemistry, Metabolic Diseases, Pharmaceutical Candidate Optimization,
`Exploratory Pharmiaceutics, Chemical Enzymologv, Bristol-Myers Squibb, Phacrmaceutical Research Institute,
`P.O. Box 5400, Princeton, New Jersey 08543-5400
`
`Received March 22, 2005
`
`Efforts to further elucidate structure -activity relationships (SAR) within our previously
`disclosed series of 3-quaternary amino acid linked L-cis-4,5-methanoprolinenitrile dipeptidyl
`peptidase IV (DPP-IV) inhibitors led to the investigation of vinyl substitution at the /-position
`of t.-cycloalkyl-substituted glycines. Despite poor systemic exposure, vinyl-substituted com-
`pounds showed extended duration of action in acute rat ex vivo plasma DPP-IV inhibition
`models. Oxygenated putative metabolites were prepared and were shown to exhibit the potency
`and extended duration of action of their precursors in efficacy models measuring glucose
`clearance in Zuckerf"'f rats. Extension of this approach to adamantylglycine-derived inhibitors
`led to the discovery of highly potent inhibitors, including hydroxyadamantyl compound BMS-
`477118 (saxagliptin), a highly efficacious, stable, and long-acting DPP-IV inhibitor, which is
`currently undergoing clinical trials for treatment of type 2 diabetes.
`
`Introduction
`Primary defects in insulin secretion, along with
`development of insulin resistance, contribute to the
`etiology of type 2 diabetes mellitus. Diminished post-
`prandial insulin secretion resulting from both functional
`defects and
`loss of survival of pancreatic
`/3-cells
`progresses into hyperglycemia and declining insulin
`sensitivity. As lifestyle trends and dietary factors have
`contributed to an alarming rise in the incidence of type
`2 diabetes,' the search for novel mechanistic approaches
`to control this chronic metabolic disease has intensified
`in parallel. To complement the currently available
`diabetes treatments,2 approaches operating within the
`enteroinsular axis through the incretin hormone glu-
`cagon-like peptide 1 (GLP-1), alone or in combination
`with other agents, are beginning to show promise in the
`treatment of diabetes.3 GLP-1 is a major component of
`the prandial nutrient-sensing mechanism regulating
`insulin secretion
`following meals.4 Intact, active
`GLP-1(7-36) amide is secreted into the circulation from
`intestinal L-cells in response to dietary signals. Con-
`centrations of GLP-1(7-36) amide sufficient to activate
`
`fax:
`
`609-818-5526
`* Corresponding
`Telephone:
`author.
`609-818-3550. E-mail: lawrenice.hamannbm s-com.
`I Discovery Chemistry.
`Present address: Lexicon Pharmaceuticals, 350 Carter Road,
`Princeton, NJ 08540.
`Present address: Pharmaceutical Discovery Division, Abbott
`LaboraLories, Abbott Park, IL 60064
`1 Pharmaceutical Candidate Optimization.
`IMetabolic Diseases,
`4 Exploratory Pharmaceutics.
`Chemical Enzymology.
`Present address: Novartis tnstitute for BioMedical Research,
`250 Massachusetts Avenue, Cambridge, MA 02139
`
`the GLP-1 receptor expressed on pancreatic /3-cells
`result in increased insulin secretion, delayed glucose
`absorption, and reduced hepatic glucose production. All
`of these components work in concert to modulate blood
`glucose levels. Because GLP-1 release is nutrient stimu-
`lated, this mechanism promotes insulin secretion under
`prandial glycemia conditions, minimizing the potential
`for hypoglycemia. Recent reports have further demon-
`strated a beneficial effect of agents acting through the
`GLP-1 axis on the preservation and/or restoration of
`/3-cell function in animals,' suggesting the exciting
`possibility that emerging drugs acting in this pathway
`may lead to improvement of the diabetic condition.
`GLP-1 is rapidly truncated during its secretion in the
`ileum by the dipeptidyl peptidase IV (DPP-IV, EC
`3.4.14.5) located on the capillary endothelium proximal
`to the L-cells where GLP-1 is secreted The efficient
`cleavage by DPP-IV of the N-terminal dipeptide His-
`Ala from GLP-1(7-36) amide yields GLP-1(9-36) amide,
`a weak antagonist of the receptor, and this cleavage
`has been demonstrated to be the primary physiological
`route of degradation of GLP-1(7-36) amide in both
`humans and animals.7 The rapid cleavage by DPP-IV
`results in an apparent elimination half-life of only 60-
`90 s for GLP-1(7-36) amide, and peak circulating levels
`of intact GLP-1(7-36) amide typically do not exceed
`5-10 pM, a range bracketing its K. as GLP-1 receptor
`agonist. Inhibition of DPP-IV prevents the degradation
`of the incretin hormones GLP- 1 and glucose-dependent
`insulinotropic peptide (GIP) and has been demonstrated
`to potentiate the levels of these peptides in multiple
`species."
`
`10.1021/jm050261p CCC: $30.25 @ 2005 Ameican Chemical Society
`Published on Web 06/24/2005
`
`AstraZeneca Exhibit 2171
`Mylan v. AstraZeneca
`IPR2015-01340
`
`Page 1 of 13
`
`

`
`5026 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15
`
`Augeri et al.
`
`Scheme 1
`
`R2
`
`Bo N
`
`b, c
`
`H2
`
`0
`7a-g
`
`CONH2
`
`2
`R
`
`a, b^
`
`OH + 5
`
`CN
`
`0
`Sa-g
`
`R2
`2 ]
`
`H2N
`
`N
`
`CN
`10a, b, d, g
`
`Ri
`
`BocHN
`
`o
`9a, b, d, g
`
`BocHN mrOH
`
`6a-g
`
`TFAa
`CONH 2
`
`d
`
`a. R, = R 2 = Me
`b. R, = R2 - Et
`c. R 1,R 2 = -9tH 2)3-
`d. R,R 2 = -(CH24-
`e. R1,R2 =-H2)5-
`f. RiR 2 -
`-(CH) 6 -
`g. R1 ,R2 = -(CH 2CH 2OCH 2CH 2)-
`DM'; 'b) POC13 , pyridine, imidazole, -20
`
`I (a) EDAC, HOBT,
`
`DPP-IV is a 240 kDa, 766 residue N-terminal dipep-
`tidyl exopeptidase that is composed of two 110 kDa
`subunits' and exists as both a membrane-bound protein
`and as a soluble protein in plasma. It is a nonclassical
`serine protease that exhibits high specificity for peptides
`with proline or alanine in the P1 position. Any amino
`acid can occupy the P2 position so long as the P2 -P1
`peptide bond can adopt a trans configuration."0 The
`membrane-bound form of DPP-TV is expressed in several
`tissues, including kidney, liver, the brush border mem-
`branes of intestinal enterocytes, on the pancreatic duct
`epithelia, and in vascular endothelial cells. In these
`tissues DPP-IV is N-terminally bound to the membrane
`with its catalytic activity located in the extracellular
`domain. The soluble, circulating form of DPP-IV is shed
`from cell surfaces by proteolytic cleavage releasing a
`fully active soluble form minus the 29 amino acids of
`the N-terminus.
`Clinical evidence has shown that small molecule
`inhibitors of DPP-IV lower blood glucose levels, increase
`glucose tolerance, and improve insulin response to oral
`glucose in patients with type 2 diabetes-" Reversible
`small-molecule inhibitors of DPP-IV have been studied
`for the past several years, and a large body of structure-
`activity relationship (SAR) data has been generated. 3b,12
`Until the very recent disclosures of several nonpeptidic
`chemotypes," 14 the known inhibitors had all been
`dipeptidomimetic in nature, bearing structural resem-
`blance to the N-terminal dipeptide of the enzyme
`substrates. For this class of inhibitors, the penultimate
`N-terminal proline or proline mimetic, generally a
`thiazolidine (1),ls a C-substituted or N-substituted
`
`R
`
`SR
`
`H2N
`
`'
`0
`1
`
`H2N
`
`'
`0
`2
`
`qN
`
`RN
`
`CN
`
`CN
`
`H
`3
`
`R,
`
`R2 R
`
`H2N
`
`N
`0 CN
`4
`
`cyanopyrrolidine (2, 3),16,17 or a cyclopropanated cyan-
`opyrrolidine (4)," is appended to an amino acid or an
`amino acid surrogate. Many inhibitors in this cyano-
`pyrrolidine structural class have suffered from varying
`degrees of chemical instability which have hampered
`formulation efforts. In addition, many examples of this
`class exhibit limited pharmacodynamic duration of
`action. We report herein the discovery of highly effica-
`cious long-acting inhibitors of DPP-IV that have led to
`the identification of compound 26 (BMS-477 118, saxa-
`
`'C; (c) TFA, CH2C1 2, rt; (d) 5% Pd/C, 112 1 atm, MeOHl.
`Scheme 2"
`R1
`
`a
`
`0
`
`R2
`
`11a-g
`
`R1
`R2
`O 2Et
`12a-g
`
`b
`
`b 2 R
`13a-g
`
`OH
`
`Rj
`R2
`
`0
`
`NHBoc
`
`R2
`
`OH
`
`d
`
`BocHN
`
`0
`14a-g
`6a-g
`'C
`triethylphosphcnoacetate, NaH, THF 0
`I (a)
`to rt;
`(b) DIBAL-H, toluene, -78
`'C to rt; (c) N-Boc glycine, DCC,
`DMAP, CH 2C12, Irt; (d) ZnC12, THY, LDA, -78
`-C to rt.
`
`gliptin), which is currently undergoing clinical evalua-
`tion for the treatment of type 2 diabetes.
`
`Chemistry
`To further our understanding of the SAR surrounding
`/3-quaternary N-terminal amino acid-containing inhibi-
`tors, we focused on elaboration of our previously dis-
`closed cyanomethanopyrrolidine-based scaffold 3 to pro-
`duce long-acting inhibitors structurally related to the
`prototype scaffold 4. A general synthesis route was
`chosen that incorporated at the
`-position a vinyl
`substituent amenable to functionalization for further
`elucidation of SAR. Standard peptide coupling condi-
`tions were employed to link enantiomerically pure
`L,-methanoprolinamide core fragment 5 with various
`racemic vinyl-substituted amino acids Ca-g to give
`dipeptides 7a-g in yields of 85-95% (Scheme 1).
`Dehydration of the resultant amides using TFAA or
`POCI3 gave the corresponding nitriles.20 Chromato-
`graphic isolation of the bioactive L-isomer was generally
`carried out at the stage of the Boc-protected nitrile. 21
`Finally, removal of the N-terminal Boc using TFA gave
`inhibitors 8a-g in high yield. The vinyl groups of Ca-f
`could be reduced (Pd/C, H2 ) to afford the corresponding
`ethyl compounds, which were similarly elaborated to
`dipeptides lOa,b,d,g.
`Amino acids possessing a J-quaternary vinyl group
`were prepared in a manner complementary to the
`malonate Knoevenagel/Michael addition sequence used
`previously." Lewis acid-mediated ester enolate Claisen
`rearrangement of substituted glycinyl allylic esters2 led
`directly to /-vinyl amino acids 6a-g in 58-85% overall
`yields (Scheme 2). The requisite Claisen precursors were
`readily prepared in three steps from the appropriate
`ketones 11a-g. Horner-Enmons olefination ofketones
`
`Page 2 of 13
`
`

`
`Discovery and Preclin ical Profile of Saxagliptin
`
`Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 5027
`
`Scheme 30
`
`BocHN
`
`)n
`
`N
`
`O
`15c-e
`
`CN
`
`1. aorbc-
`
`2. d
`
`Hoc 0
`H N
`Bo/
`
`N
`CN
`
`If
`O
`
`17d
`18d
`to 0 'C, 60-79%; (b) Os04, NMNO, THF/H20 1:1, rt, 47-63%; (c) NaIO 4;
`'C; then NaBH 4, -78
`(a) 03, MeOH/CH 2C2 10:4, -78
`rt, 56%; (d) TFAICH2Cla1:2, O 'C to rt-
`workup, then NaBH 4, MeO
`
`d.n=1
`d. n = 1
`e. ni = 2
`
`H2N
`
`I/
`
`N.
`
`O
`16c-e
`
`CN
`
`d
`
`CN
`
`N
`
`r 0
`
`HO
`HO
`H 2N
`
`Scheme 40
`
`H3 CO 2 C
`19
`
`a, bc
`
`HO
`
`g, h, i
`
`20
`
`21
`
`H2N
`
`N
`H
`
`C0 2H
`
`-tN_
`0
`CN
`23
`22
`a (a) LAH, THF, 0 'C to rt, 96%; (b) (CICO)2 , DMSO, CH2 C12,
`'C, 98%; c (-(-)-2-phenylgycinol, NaHSO 3, KCN, 65%;
`-78
`(d) 12 M HCI, HOAc, 80 'C, 16 h, 78%; (c) 20% Pd(OH)2, 50 psi
`H2 , MeOH/HOAc 5:1; (f) (Boc20, KXCO, DMF, 92%, two steps;
`(g) 5, EDAC, HOBT, DMF, 92%; (h) POC13, pyridine, imidazole,
`'C;
`i) TFA, CH9 C12, r, quanL.
`-20
`
`11a-g with the ylide generated from triethylphospho-
`noacetate gave the a,3-unsaturated esters 12a-g in
`92-98% yield. Esters 12a-g were then reduced with
`DIBAL to the corresponding allylic alcohols 13a-g and
`condensed with N-Boc glycine using DCC/DMAP to give
`esters 14a-g in 79-87% yield over two steps. ZnClz-
`mediated Claisen rearrangement of the LIDA-generated
`enolate of glycine esters 14a-g proceeded at low tem-
`perature to give the desired /3-vinyl amino acids 6a-g
`in 65-90% yield.
`Further elaboration of vinyl-containing dipeptides
`7c-e was accomplished at the stage of the dehydrated
`cyano-containing compounds 15c-e (Scheme 3). Oidda-
`tive cleavage of the vinyl substituent to prepare hy-
`droxymethyl compounds 16c-e was achieved either by
`ozonolysis/NaBH 4 reduction or by catalytic 0s04-
`NMNOINaIO4/NaBH 4 conditions, followed by acidic
`deprotection of the Boc group. Additionally, 15d was
`converted to the corresponding diol and deprotected to
`give 18d.
`A logical extension of our previously observed SAR
`trends favoring f-branched P2 units led us to explore
`rigidly bridged polycyclic systems such as adamantyl.
`Analogues bearing an adamantyl ring at the N-terminal
`a-carbon were synthetically derived from a common
`homochiral adamantylglycine intermediate prepared
`using asymmetric Strecker chemistry (Scheme 4).3
`Reduction of commercially available adamantane car-
`boxylic acid methyl ester 19 by LAH, followed by Sworn
`
`oxidation, afforded the requisite aldehyde. which was
`then subjected to asymmetric Strecker conditions (con-
`densation with (R)-(-)-2-phenylglycinol with addition
`of KCN) to give the desired homochiral R,S diastere-
`omer 20 in 65% yield. Hydrolysis of the nitrile group to
`give acid 21, followed by hydrogenolysis of the chiral
`auxiliary, afforded the enantiomerically pure amino acid
`22. Boc protection of the resulting primary amine,
`followed by coupling to methanoprolinamide core 5,
`dehydration of the amide to nitrile, and deprotection,
`afforded the adamantylglycine containing inhibitor 23
`in good overall yield-
`Hydroxylation of N-Boc-adamantylglycine 22 at the
`bridgehead was accomplished using KMnO 4 in 2%
`aqueous KOH at elevated temperature to give N-Boc
`hydroxyadamantyl glycinc 24 in 51% yield (Scheme 5).24
`Standard acylation conditions were used to couple 24
`to methanoprolinamide core 5, furnishing anide 25 in
`high yield. Amide 25 was subsequently elaborated to
`provide two additional analogues. Dehydration of amide
`25 with TFAA, followed by in situ basic hydrolysis of
`the resulting trifluoroacetate and deprotection of the
`N-terminus, gave hydroxy derivative 26 in 87q yield
`over three steps. The hydroxy group of 25 was subjected
`to fluoride substitution using DAST, 24 and subsequent
`dehydration using POC 3 in pyridine, followed by depro-
`tection of the terminal nitrogen, provided fluoroada-
`mantylglycine analogue 30 in 73% overall yield for three
`steps. Prolonged exposure of protected adamantylglycine
`22 to KMnO4 in 2% aqueous KOH provided the dihy-
`droxyadainantylglycine derivative 27. Coupling of 27 to
`5, followed by dehydration of the resultant prolineamide
`with TFAA, in situ basic hydrolysis of the bis-trifluoro-
`acetate, and removal of the terminal Boc group using
`TFA, afforded dihydroxyadamantyl analogue 28 in 74%
`overall yield.
`In Vitro and in Vivo Biological Activity. DPP-
`IV Inhibitory Activity in Vitro and ex Vivo. The
`DPP-IV inhibitory activity of analogues in the present
`series was measured against human DPP-IV using
`standard assays as described in the Experimental
`Section (Table 1). Many of the compounds in this series
`were potent inhibitors of DPP-IV in vitro, several with
`Ki's in the sub-nanomolar range. Additionally, several
`inhibitors in this series exhibited significant slow, tight-
`binding kinetics. 5
`A finer discrimination between the most potent
`compounds within this structurally related series of
`inhibitors with respect to pharmacodynamic effects and
`
`Page 3 of 13
`
`

`
`5028 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15
`
`Augeri et al.
`
`Scheme 50
`
`/a
`
`HO
`
`b
`
`c, d
`
`BOC-HN
`
`COH
`
`BOC-HN
`
`CO 2H
`
`BOC
`
`22
`
`24
`
`a, 90 oin
`
`HO
`
`JOH
`
`HO
`
`,OH
`
`BOC-HN CO2H
`
`H2 N
`
`Ie
`
`f,d
`
`F
`
`H
`2
`
`CN
`
`O0
`27
`30
`(-f>NH2
`28
`29
`(a) KMnO 4, 2% aq KOH, 60 to 90 0C, 60 min, 51%, (b) 5, EDAC, HOBT, DMF, 77-85%, (c) (CF 3COhO, pyridine, THF 0 'C to rt, then
`10% aq K 2C0 3 in MeOH, 89-92%; (d) TFA, CH2C12 , rt, 89-95/; (e) DAST, CH 2C12, _78 C, 94%; f) POC1:, pyridine, imidazole, 82%.
`
`Table 1. In Vitro Inhibition Constants for Human DPP-IV and
`cx Vive Plasma DPP-IV Inhibition in Normal Rats
`
`Table 2. Potency and Duration of Effect of Compounds 16d
`and 26 in the ex Vive Rat Plasma DPP-IV Inhibition Model
`
`ED, emol/kg at time postdosel
`
`compd
`
`0.5 h
`
`2 h
`
`l6d
`26
`
`0.4 + 0.15
`0.12 + 0.04
`
`3.2 z 1.2
`0.2 + 0.07
`
`4 h
`5.0 & 1.9
`0.3 + 0.10
`
`6 h
`
`11 + 4.2
`0.5 + 0.15
`
`' Compounds dosed po to fasted normal SD rats at the indicated
`times postdose, plasma aliquots were isolated, and DPP-IV inhibi-
`tion was assayed using the fluorogenic peptide assay. EDso is the
`50%. inhibitory dose calculated from the plot s of percent inhibition
`vs dose aL each time point.
`
`postdose to assay plasma (prepared with EDTA) DPP-
`IV activity in vitro using the fluorogenic DPP-IV-specific
`substrate Ala-Pro-AFC. Plasma DPP-IV activity deter-
`minations were calculated by linear regression from
`plots of product vs time (initial 20 min). Data were
`calculated as mean percent inhibition vs controls receiv-
`ing water vehicle. Maximal inhibition of plasma DPP-
`IV under the conditions of this assay reached 85 -90%
`(Table 1). In a dose- relationship mode, ED 5o's were
`determined for select compounds at multiple time points
`of 0.5, 2, 4, and 6 h postdose (Table 2).
`
`Results and Discussion
`The SAR described in our previous account culminat-
`ing in 4,5-methanoprolinenitrile analogues 4 revealed
`a strong preference for compounds with lipophilic
`N-terminal fl-quaternary amino acids." In the course
`of further studies exploring SAR around fl-quaternary
`cycloalkylglycine-based inhibitors, we encountered un-
`expectedly potent activity and extended duration of
`action in ex vivo DPP-IV inhibition studies with com-
`pound 8d, which contains a (vinylcyclopentyl)glycine
`amino acid fragment. However, metabolism and phar-
`macokinetic studies with 8d revealed uncharacteristi-
`cally poor oral bioavailability (F = 5.3%) and high rat
`liver microsomal turnover rate [0.55 nmol/inin/mg pro-
`tein for 8d vs 0.32 for compound 4 (where R1 and R 2
`taken together - cyclopentyl, and R3 - Me)]. Similar
`observations were made for other vinyl-containing
`analogues 8c,c,g, and these results suggested conver-
`sion to an active metabolite in vivo. As the vinyl
`substituent seemed a likely site of metabolism, synthe-
`sis of oxygenated analogues (16d and 18d) derived from
`chemical modification of the olEfin moiety was under-
`
`% plasma DPP-IV inhibn at
`4 jimol/kg po, normal rats
`
`4 h
`10
`20
`32
`64
`60
`66
`nd
`
`0n
`
`d
`44
`nd
`17
`56
`
`8n
`
`d
`83
`87
`57
`61
`
`30 min
`13
`39
`42
`71
`76
`77
`nd5
`
`0n
`
`d
`40
`nd
`36
`69
`17
`nd
`84
`87
`62
`80
`
`human
`DPP-IV
`K, (nM
`
`57+8
`25 + 4
`12 + 0.9
`3.9
`0.6
`1.4 + 0.06
`10 + 3
`10 ± 2
`7.1
`0.7
`31 ± 2
`5.5 - 0.7
`21 ± 0.6
`42 ± 4
`7.4 ± 1.1
`8.0 + 0.4
`143 ± 15
`0.9 + 0.32
`0.6 + 0.06,
`2.1 ± 0.3
`1.8 + 0.5
`
`compd
`
`8a
`8b
`Sc
`8d
`8e
`Sf
`8g
`10a
`10b
`10d
`log
`16c
`16d
`16e,
`l8d
`23
`26
`28
`30
`
`' Values represent the mean - SEM and are at least triplicate
`determinations. nd = not determined. ' Compound 26 did not
`show any significant inhibition of dipeptidyl peptidase II
`(DPP-I) at concentrations up to 30 pM.
`
`duration of action required utilization of a medium-
`throughput acute efficacy model measuring a surrogate
`biomarker expected to be predictive of downstream
`antihyperglycemic effects. As DPP-IV is found in plasma
`and on the surfaces of blood and tissue cells, it was
`reasoned that measurement of inhibition of the circulat-
`ing enzyme in plasma might provide a convenient
`biomarker for the degree of preservation of plasma
`incretin hormone levels. Though the relative contribu-
`tion of these enzyme loci to the physiological degrada-
`tion of GLP-1(7 -36) amide important for antihyper-
`glycemic effects is not fully understood, it was further
`envisioned that plasma enzyme inhibition measured cx
`vivo after an oral dose of test compound might be used
`to develop pharrmacokinctic- pharmacodynamic rela-
`tionships and provide information regarding duration
`of action. Compounds were administered orally in water
`vehicle at 4 jmol/kg to normal Sprague-Dawley rats,
`and blood samples were taken at 30 min and 4 h
`
`Page 4 of 13
`
`

`
`Discov~ery and Preclin ical Profile of Saxagliptin
`
`Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 5029
`
`taken. Dio1 1Sd showed only weak inhibitory activity;
`however, hydroxymethyl analogue 16d exhibited po-
`tency similar to that of the vinyl analogue in both in
`vitro and ex vivo assays, restored rat liver micros omal
`turnover rate to a more moderate level (0.16 nmollmin/
`mg protein), and restored oral bioavailability to within
`the range characteristic for other structurally related
`analogues in the series (F = 59%). Unequivocal char-
`acterization of 16d as the active metabolite of Sd was
`never established, though the behavior of 16d mirrored
`that achieved upon administration of Sd. Accordingly,
`DPP-IV inhibitors 16c and 16e were prepared. A similar
`trend toward reconnection of pharmacokinetic proper-
`ties with pharmacodynamic measurements was ob-
`served for the homologous pairs of inhibitors Sc16c and
`8e/16e. Despite this latter observation, the five-mem-
`bered ring compound 16d stood out as significantly more
`effective in the rat ex vivo plasma DPP-IV inhibition
`assay.
`A more striking observation of metabolic conversion
`was seen with the highly potent adamantylglycine-
`containing analogue 23 (Ki = 0.9 nM). Although this
`compound afforded potent plasma DPP-IV inhibition
`after oral administration to rats (84% at 0 5 h, 83% at
`4 h), it exhibited poor absolute bioavailability (F = 2%)
`after oral dosing and rapid turnover in rat liver micro-
`somes. Interestingly, compound 23 also weakly inhibited
`CYP3A4 with an IC 5 0 of 20 pM, where previous closely
`related analogues were devoid of any CYP inhibitory
`activity. Preparation of the bridgehead-hydroxylated
`analogue 26 gave a compound with a virtually identical
`in vitro (Ki = 0.6 nM) and ex vivo (87% inhibition of
`plasma DPP-IV at 0.5 and 4 h) profile, a slow rat liver
`microsomal turnover rate, no CYP3A4 inhibition up to
`100 pM, and good oral exposure (F = 75%, tl/2 = 2.1 h).
`Two other substituted adamantyl-derived compounds
`were also synthesized and investigated. Dihydroxyada-
`mantyl compound 28, while still reasonably active in
`the ex vivo assay, exhibited extremely high aqueous
`solubility but exhibited low oral exposure in rats,
`presumably resulting from very poor absorption. Though
`fluoroadamantyl compound 30 was also effective ex vivo,
`it exhibited very low oral exposure and had a rat liver
`microsomal turnover rate indicative of extensive me-
`tabolism, similar to that of compound 23. Due to its
`exceptional plasma inhibitory potency and pharmaco-
`dynamic duration of action in this preliminary ex vivo
`assay (ED5 0 for 26 at 6 h = 0.5 ,imol/kg vs ED5 0 for 16d
`at 6 h = 11 pmol/kg, Table 2, compound 26 was chosen
`for further study in acute efficacy models.
`Oral Glucose Tolerance in Zuckerfafa Rats.
`Zuckert f" rats are a well-established genetically modi-
`fled rodent model of obesity-induced insulin resistance 26
`and provide a background to measure the effects of
`DPP-IV inhibitors in a prediabetic animal.2 7 The
`nutrient-induced incretin secretion component of the
`GLP-1-dependent mechanism makes this a suitable
`model with which to study postprandial glucose excur-
`sions after administration of an oral glucose tolerance
`test (oGTT). DPP-IV inhibitor 26 was chosen for futher
`study in this animal model by virtue of its highly potent
`effects in vitro and ex vivo. Compound 26 was admin-
`istered orally to Zuckerfait
`rats at 0.5 h pnor to oGTT,
`consisting of a glucose challenge (2.0 g/kg), followed by
`
`300
`
`-3,
`
`qehic. 0.3
`
`1
`
`3.
`
`Figure 1. Effects of inhibitor 26 dosed at 0.3, 1, and 3 umoli
`kg po versus vehicle control on plasma glucose clearance after
`an oGTT given 4h postdose in 5ucker'
`[a rats.
`
`blood sampling at intervals over the next 2 h for plasma
`glucose measurenents (data not shown). Maximal re-
`sponses in glucose excursion in this model were associ-
`ated with plasma DPP-IV inhibition of approximately
`60% vs control, and no additional antihvperglycemic
`effects were seen at higher percent inhibition On the
`basis of these preliminary findings at a single dose,
`compound 26 was further evaluated in the Zuckerftuit
`rat model with an oGTT performed 4 h after oral
`administration of test compound in a dose-response
`format. Postprandial plasma glucose and insulin levels
`were again measured at intervals over 2 h following the
`glucose challenge. Compound 26 was highly effective at
`eliciting marked dose-dependent enhancements in glu-
`the dose range 0.3-3 ymol/kg
`cose clearance
`in
`(0.13-1.3 mg/kg) in this model relative to controls
`(Figure 1).
`Oral Glucose Tolerance in ob/ob Mice. Evidence
`from both inhibitor studies and knock-out animals
`support that the mouse is also a suitable species in
`which to study the effects of DPP-IV inhibition on
`glucose clearance and insulin potentiation. 2 s To this end,
`the effects of compound 26 on glucose clearance and
`enhancement of insulin secretion was studied in the
`ob/ob mouse. In this model the oGTT was performed at
`1 h after oral administration of 26 at 1, 3, or 10 ymoll
`kg (Figure 2). The data show that compound 26 dose-
`dependently elevated plasma insulin significantly at
`15 min post-oGTT, with concomitant improvement in
`the glucose clearance curves at 60 min post-oGTT.
`Compound 26 exhibited robust glucose-lowering ef-
`fects in a dose-relational manner in the Zuckert Ir rat
`oGTT model, even when the glucose challenge was
`administered 4 h postdose of compound. Similarly
`outstanding efficacy was observed in reducing postpran-
`dial glucose AUC in ob lob mice. This compound also
`proved quite effective in elevating insulin levels after
`an oGTT
`in ob/ob mice, further demonstrating the
`effectiveness of potentiating GLP-1-induced insulin
`secretion as a key component mediating the antihyper-
`glycemic actions of this potent DPP-IV inhibitor. It is
`anticipated that compound 26, given its extended phar-
`macodynamic response, will be amenable to once daily
`dosing in humans.
`
`Conclusion
`A series of f-quaternary cycloalkylglycine amino acid
`residues were incorporated into our previously disclosed
`4,5-mothanoprolinenitrile scaffold, and many of these
`compounds showed potent DPP-IV inhibitory activity.
`Several compounds containing a vinyl functionality also
`
`Page 5 of 13
`
`

`
`5030 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15
`
`Augeri et al.
`
`~60C
`-.-
`--.-v.--. 261,06510
`---. 3.---..
`0 0,ooASo
`26 0~m51~
`
`1506
`
`1400i),0
`
`-.50
`
`II
`
`Figure 2. Effects of inhibitor 26 dosed at 1, 3, and 10 pmol/kg po versus vehicle control on plasma insulin (left panel) and
`plasma glucose (right panel) after an oGTT in oblob mice. Compound 26 significantly lowered plasma glucose levels
`(vs vehicle) at the 60 min time point at 3 and 10 jmol/kg (p < 0.05) and significantly increased plasma insulin levels (vs vehicle)
`at the 15 min time point at 10 umol/kg (p < 0.05).
`
`exhibited extended duration of action in an ex vivo
`plasma DPP-IV inhibition model in normal rats relative
`to closely related analogues lacking this moiety. These
`analogues, however, also showed markedly reduced
`systemic exposure after oral dosing and rapid rat liver
`microsomal turnover rates where related small alkyl-
`substituted analogues did not. Efforts to define the role
`of suspected metabolites resulted in the synthesis of
`several hydroxymethyleycloalkyl-based analogues that
`maintained in vitro and, for some, in vivo activity. These
`hydroxymethyl analogues also displayed favorable phar-
`macokinetic properties with a tighter correlation of
`pharmacokinctics to pharmacodynamics. Analogously,
`hydroxylation of a similarly disposed adamantylglycme-
`based inhibitor yielded a compound (26) with in vivo
`potency and duration of action superior to that of any
`compound from this series. Consequently, this com-
`pound was chosen for development and is currently
`under clinical investigation for the treatment of type 2
`diabetes. The basis for the enhanced efficacy observed
`for the present compounds in animal models relative
`to other agents may be due to contributions from
`multiple factors, including exquisite enzyme inhibitory
`potency and compound distribution to the tissue com-
`partment potentially critical for maximal antihyper-
`glycemic effects (V_ = 5.2 L/kg for compound 26 in the
`rat). Further studies are underway in these laboratores
`to more fully understand and quantitatively character-
`ize the physicochemical basis for the observed findings.
`
`Experimental Section
`All reactions were carried out under a static atmosphere of
`argon or nitrogen and 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- 'H (400 MHz)
`and 53C (100 MHz) NMR spectra were recorded on a JEOL
`GSX400 spectrometer using Me 4Si as an internal standard
`unless otherwise noted. 'H (500 MHz) and "C (125 MHz) NMR
`spectra were recorded on a JEOL JNM-ECPoO spectrometer.
`Chemical shifts are given in parts per million (ppm) downfield
`from internal reference tetramethylsilane
`in 5-units, and
`coupling constants (J-values) are given in hertz (Hz). Selected
`data are reported in the following manner: chemical shift,
`multiplicity, coupling constants, and assignment. All reactions
`were carried out using commercially available anhydrous
`solvents from Aldrich Chemical Co. or EM Science Chemical
`Co. unless otherwise noted. All flash chromatographic separa-
`tions 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 gel plates (60 F2 51) and were
`visualized with UV light, with 5% phosphomolybdic acid in
`
`95% EtOH, or by a sequential treatment with I N HCI/MeOH
`followed by ninhydrin staining. LCIMS data were recorded on
`a Shimadzu LC-1OAT equipped with a SIL-10A injector, a
`SPD-IOAV detector, normally operating at 220 am, and
`interfaced with a Micromass ZMD mass spectrometer. LCIMS
`or HPLC retention times, unless otherwise noted, are reported
`using a Phenomenex Luna C-18 4.6 mm x 50 mm column
`eluted with a 4 min gradient from 0 to 100% B, where A =
`10% MeOH-90% H2 0-0.1% TFA and B = 90% MeOH-10%
`H 2 0-0.1% TFA. All solvents were removed by rotary evapora-
`tion under vacuum using a standard rotovap equipped with a
`dry ice condenser- All filtrations were performed with a
`vacuum unless otherwise noted.
`General Method A. Peptide Coupling to Enantiomeri-
`eally Pure L-cis-4,5-Methanoprolinamide 5, Amide De-
`hydration, and Deprotection. Methanoprolinamide 5 was
`coupled to a variety of racemic quaternary protected amino
`acids using HOBT/EDC in DMF at room temperature to give
`a D/L mixture of diastereomers at the N-terminal amino acid.
`The desired L-diastercomer was most often chromatographi-
`cally isolated as the N-Boc-protected nitrile, obtained by
`treatment of amide 7 with POCI/imidazole in pyridine at
`'C. The final target compounds Sa-g were obtained by
`-20
`deproection using TFA in CH 2 C12.
`(S)-2-(l-Ethenylcyclopent-1-yl)glycine-L-cis-4,5-metha-
`noprolinamide (7d). 4,5-Methanoprolinamide 5 (877 mg,
`3.65 mmol ) and N-Boc cyclopentylvinylamino acid Gd (1.13 g,
`4.20 mmol), described in general method B, were dissolved in
`20 mL of DMF and cooled to 0 'C and to this mixture were
`added EDAC (1.62 g, 8.4 mmol), HOET hydrate (2.54 g,
`12.6 mmol), and TEA (1.27 g, 12.6 mmol). The reaction was
`allowed
`to warm to room temperature and stirred for 24 h.
`The reaction mixture was taken up in 100 mL of EtOAc,
`washed with H 20 (3 x 20 mL), dried (NaSO4], and purified
`by flash chromatography (100% EtOAc) to give 1.38 g (86%)
`1H NMR (500 MHz,
`of 7d as a mixture of diastereomers
`CDC13) 1.02-108 (m, 2H), 143 (s, 9H), 1.54-1.80 (in, 7H),
`184-L94 (m, 2H), 236 (dd, J - 13.6, 2.6, 1H), 258 (m, 1H),
`3.90
`(in,
`(d, J = 9.7,
`1H), 4.63
`1H), 5.04
`(dd,
`J= 10.5, 2.2, 1H), 514 (d, J= 17.6, 1H), 5.23 (d, J= 11, 1H),
`5.97 (dd, J = 17.6, 11, 1H,, MS niz 378 [M + H]
`.
`(S)-2-(1-Ethenylcyclopent-1-yl)glycine-L-cis-4,5-metha-
`noprolinenitrile (8d). Diastereomeric amide 7d (68 mg,
`0.18 mmol, 1 equiv) and imidazole (26 mg, 0.38 mmol, 2.1
`equiv) were dissolved in 2 mL of pyridine at -30
`'C, to which
`POC 3 (0.070 mL, 0.739 mmol, 4.10 cquiv) was added. After
`stirring at -30
`'C for 40 min, the solvent was removed and
`the residue thoroughly dried under vacuum. Purification by
`flash chromatography (10% EtOAclCH2C 2) afforded the de-
`sired slower eluting diastereomerically pure N-Buc-protected
`mtrile as a white solid (34 mg, 0.0946 mmol, 53%): 1H NMR
`(500 MHz, CDC 3 ) L02-1.07 (m, 2H), 142 (s, 9H), 1.55-1.75
`(m, 7H), 1.83-1.93 (in, 2H), 2.36 (dd, J = 13.6, 2.6, IH), 2.58
`(m, iH), 3.90 (in, 1H), 4.62 (d, J= 9.2, 1H), 5.03 (dd, J= 10.5,
`2.2, 1H), 513 (d, J= 17,6, 1H), 5,23 (d, J= 11, 1H1), 5 96 (dd,
`J= 17.6, 11, 1H); MS rm/z 360 [M + H]. The N-Boc protected
`
`Page 6 of 13
`
`

`
`Discovery and Preclinical Profil