Inhibitors of Dipeptidyl Peptidase 4
`
`Stephen L. Gwaltney, II and Jeffrey A. Stafford
`
`Takeda San Diego, Inc., 10410 Science Center Drive, San Diego, CA 92121
`
`Contents
`1. Introduction
`1.1. Function of DPP4
`1.2. Structure of DPP4
`1.3. Therapeutic significance
`2. Preclinical DPP4 inhibitors
`3. DPP4 inhibitors in clinical development
`4. Alternative indications for DPP4 inhibitors
`References
`
`1. INTRODUCTION
`
`149
`149
`150
`150
`151
`156
`160
`161
`
`Dipeptidyl peptidase 4 (EC 3.4.14.5, DPP-IV, DPP4, CD26) is a ubiquitous serine
`protease that modulates the biological activities of numerous peptides, including
`glucagon-like peptide-1 (GLP-1). GLP-1 plays an important role in the control of
`post-prandial glucose levels by potentiating glucose-stimulated insulin release and
`inhibiting the release of glucagon. Other actions of GLP-1 include delaying gastric
`emptying, inducing satiety and increasing beta cell mass. GLP-1 has shown efficacy
`in diabetics, but suffers from a very short physiological half-life (t1/2 2 min) due to
`DPP4-mediated cleavage of the active peptide (7-36 amide or 7-37) to an inactive
`form (9-36 amide or 9-37). Intense research in the pharmaceutical industry aims to
`discover and develop stable GLP-1 analogs, exogenous agonists of the GLP-1 re-
`ceptor or small-molecule inhibitors of DPP4. This research has been buoyed re-
`cently by positive clinical trial data on GLP-1 analogs and DPP4 inhibitors. The
`field of DPP4 inhibition has been reviewed extensively [1–12]. This review attempts
`to provide an update to the previous ARMC article on DPP4 inhibitors [13] cov-
`ering the primary literature from 2001 through the end of March 2005. It is not the
`intent of the authors to provide another review of the pharmacology of DPP4, but
`to concentrate on the medicinal chemistry in the field.
`
`1.1. Function of DPP4
`
`DPP4 functions as a serine protease and cleaves the amino-terminal dipeptide from
`oligopeptides with a proline or alanine at the penultimate position. Peptides with
`residues other than Pro or Ala at the penultimate position may also be low-affinity
`substrates for DPP4. In contrast, DPP4 is not selective with respect to the N-terminal
`
`ANNUAL REPORTS IN MEDICINAL CHEMISTRY, VOLUME 40
`ISSN: 0065-7743 DOI 10.1016/S0065-7743(05)40010-X
`
`r 2005 Elsevier Inc.
`All rights reserved
`
`MYLAN Ex. 1011, Page 1
`
`

`
`150
`
`S.L. Gwaltney, II and J.A. Stafford
`
`residue [14] and shows little discrimination of various prime-side residues [15,16]. A
`number of biologically important peptides are substrates for DPP4 in vitro [17,18].
`
`1.2. Structure of DPP4
`
`DPP4 is a 110-kDa glycoprotein expressed on the cell surface and widely distributed
`throughout the body. Cleavage of the extracellular portion of DPP4 from the
`22-residue transmembrane section results in a soluble, circulating form of approxi-
`mately 100 kDa. Functional DPP4 is a homodimer, although an active heterodimer
`with fibroblast activation protein has been observed [19]. The consensus sequence for
`DPP4 is G-W-S-Y-G and the catalytic triad is made up of Ser630, Asp708 and
`His740. It has been shown that the glycosylation state of the enzyme is not important
`for enzyme activity, dimerization, and adenosine deaminase binding [20].
`Several groups have reported crystal structures of human DPP4 [15,21–24], and
`one group has reported the structure of porcine DPP4 [25]. These structures show
`the dimeric nature of the enzyme and reveal that the catalytic site is located in a
`cavity between the a/b hydrolase domain and an eight-bladed propeller domain.
`Also revealed is the oxyanion hole, which is composed of the backbone NH of
`Tyr631 and the OH of Tyr47. A co-complex of DPP4 and the inhibitor Val-
`pyrrolidide demonstrates that two glutamates in the active site play an important
`role in substrate binding by forming a salt bridge with the N-terminus of a peptide
`substrate. The pyrrolidine of the inhibitor effectively fills a hydrophobic pocket that
`will only accommodate small residues. This pocket engenders DPP4’s selectivity for
`proline at P1. This work also revealed that two openings in the enzyme may provide
`access to and egress from the catalytic site for some substrates and products [21].
`The importance of Tyr547 in the stabilization of the intermediate oxyanion was
`confirmed through site-directed mutagenesis [26]. Most authors agree that peptides
`enter the larger side opening to access the active site [15]. It has been postulated that
`the dipeptide product is expelled through the narrow b-propeller opening [21,24].
`The co-complex of DPP4 and a compound related to NVP-DPP728 [23] confirms
`that cyanopyrrolidine inhibitors form an imidate with the active site serine, con-
`sistent with a model proposed earlier [27]. Two groups have observed the trapping
`of tetrahedral intermediates in co-complexes of peptides with DPP4 [15,24].
`
`1.3. Therapeutic significance
`
`Relative to wild-type controls, DPP4-deficient mice are resistant to the development
`of obesity and hyperinsulinemia when fed a high-fat diet [28]. DPP4 knockout mice
`also show elevated GLP-1 levels and improved metabolic control. Relative to DPP4
`positive controls, DPP4-deficient Fischer rats show improved glucose tolerance
`following an oral glucose challenge due to enhanced insulin release mediated by
`high levels of active GLP-1 [29,30]. In these studies, the authors note that fasting
`and post-challenge glucose levels in both strains are similar, supporting previous
`assertions that hypoglycemia is unlikely during treatment with DPP4 inhibitors.
`
`MYLAN Ex. 1011, Page 2
`
`

`
`Inhibitors of Dipeptidyl Peptidase 4
`
`151
`
`The use of GLP-1 and its analogs in the treatment of diabetes has been reviewed
`recently [31,32]. It has been shown that DPP4 inhibition prevents the degradation of
`endogenous GLP-1 and glucose-dependent insulinotropic polypeptide (GIP) in dogs,
`thereby preserving the insulinotropic effects of these peptides [33]. In the same study, it
`was noted that total incretin secretion was reduced, suggesting that feedback mech-
`anisms restrict the secretion of incretins when levels of active peptide are elevated. It
`has been demonstrated that agonism of the GLP-1 receptor results in growth and
`differentiation of pancreatic islet beta cells [34–36]. If realized in humans, such an
`effect may result in preservation or restoration of b-cell function in diabetics. In
`human clinical trials, infusion of GLP-1 led to such beneficial effects as decreases in
`post-prandial glucose excursions, increases in post-prandial insulin, reductions in
`HbA1c, weight loss, enhanced insulin sensitivity and improved b-cell function [37,38].
`Administration of the GLP-1 analogs exendin-4, CJC-1131 and NN2211 resulted in
`similar beneficial effects [31,32]. Notably, DPP4 inhibition has been shown to augment
`the insulin secretion effects of not only GLP-1 and GIP, but also pituitary adenylate
`cyclase-activating polypeptide (PACAP) and gastrin-releasing peptide (GRP) [39].
`
`2. PRECLINICAL DPP4 INHIBITORS
`
`Early DPP4 inhibitors closely mimicked DPP4 substrates, as exemplified by valine-
`pyrrolidide (Val-Pyr, 1), P32/98 (2) and FE 999011 (3). A large body of data has
`been reported for these compounds and provided early biological validation for the
`use of DPP4 inhibitors as an approach to the treatment of diabetes.
`
`N
`
`CN
`
`O 3
`
`H2N
`
`S
`
`N
`
`H2N
`
`O
`
`2
`
`N
`
`O 1
`
`H2N
`
`Treatment of six-week-old db/db mice with Val-Pyr resulted in increased endo-
`genous GLP-1 levels, potentiated insulin secretion and improved glucose tolerance;
`however, while the effects on GLP-1 and insulin were maintained in mice at 23
`weeks of age, the improved glucose control was lost [40]. Studies in rats demon-
`strated that combining Val-Pyr with metformin leads to reduced food intake and
`body weight gain, improved glucose tolerance and increases in active plasma GLP-1
`and that these effects are absent or less significant when using either drug as
`monotherapy [41,42]. In related work, treatment of rats with metformin or piog-
`litazone resulted in reduced serum DPP4 activity. Since the authors found that these
`agents are not inhibitors of DPP4 in vitro, they suggested that the effect resulted
`from reduced DPP4 secretion [43].
`Double incretin receptor knockout (DIRKO) mice are genetically altered to
`lack both the GLP-1 receptor and the GIP receptor. A study in these animals with
`
`MYLAN Ex. 1011, Page 3
`
`

`
`152
`
`S.L. Gwaltney, II and J.A. Stafford
`
`Val-Pyr and a structurally unrelated inhibitor, SYR106124, showed that while these
`inhibitors provide improved glucose tolerance and increased insulin levels in wild-
`type and single incretin receptor knock out mice, these effects were lost in the
`DIRKO mice. This result points to the essential nature of the incretin receptors in
`the actions of DPP4 inhibitors [44].
`While inhibitors such as 4 (Ki ¼ 6.03 mM) and 5 (IC50 ¼ 12 mM) are related to
`the cyanopyrrolidine DPP4 inhibitors through the use of the fluoroolefin amide
`isostere, these compounds are only weak inhibitors of the enzyme [45–47].
`
`CN
`
`F
`
`5
`
`NH
`
`H2N
`
`CN
`
`F
`
`4
`
`Several recent papers have examined the effects of long-term treatment with P32/
`98 (2) in rodent models of diabetes. A three-month treatment regimen provided
`sustained improvements in glucose tolerance, increased b-cell responsiveness and
`improved peripheral insulin sensitivity in Zucker fa/fa rats [48,49]. The same in-
`vestigators have shown that 7 weeks of treatment with 2 enhances b-cell survival
`and islet neogenesis in a streptozotocin-induced diabetes model [50]. A study de-
`signed to compare the effects of 2 with those of rosiglitazone and to the effects of
`the combination of the two agents found that the DPP4 inhibitor provided im-
`proved glucose tolerance in both prediabetic and diabetic animals. While rosiglita-
`zone resulted in increased body weight, 2 was body-weight neutral. However,
`neither agent was very effective at improving the diabetic condition of older ZDF
`rats [51]. Studies have shown that the metabolism of 2 is dominated by oxidation of
`the sulfur atom and glucuronidation of the primary amine [52].
`In rodent models of diabetes, chronic treatment with FE 999011 (3) provided
`improved glucose tolerance, postponed the progression to hyperglycemia by 21
`days, reduced hypertrigylyceridemia and prevented a rise in circulating free fatty
`acids [53].
`Rodent studies using NVP-DPP728 (6, IC50 ¼ 7 nM) [54] and the structurally
`related K579 (7, IC50 ¼ 5 nM) have demonstrated similar pharmacological effects
`as those seen with the inhibitors discussed above. In a comparative study, 7 ap-
`peared to provide better control of DPP4 activity and glucose excursions than did 6
`[55]. Combination of 7 with glibenclamide further enhanced the glucose control
`without significant hypoglycemia [56].
`
`N
`
`O
`
`CN
`
`NH
`
`7
`
`N
`
`N
`
`N
`
`N
`
`O
`
`CN
`
`NH
`
`6
`
`HN
`
`N
`
`NC
`
`MYLAN Ex. 1011, Page 4
`
`

`
`Inhibitors of Dipeptidyl Peptidase 4
`
`153
`
`The 2-CN pyrrolidine present in 6 can be substituted by a cyanopyrazoline, but
`this results in a less potent compound (8, IC50 ¼ 360 nM) [57]. A pyrazolidine
`heterocycle has also been examined (9, IC50 ¼ 1.56 mM) [58].
`
`NO2
`
`HN
`
`N
`
`N
`
`N
`
`N
`
`O
`
`CN
`
`H2N
`
`O
`
`O
`
`9
`
`NH
`
`8
`
`HN
`
`N
`
`NC
`
`Several groups have examined substituted pyrrolidines in an effort to improve
`potency or stability of the inhibitors. Attempted incorporation of hydroxy or met-
`hoxy substituents at various positions on the ring led to reduced potency, but
`fluorination at
`the 4-position gave increased potency as in compound 10
`(IC50 ¼ 0.6 nM). This compound also displayed increased plasma drug concentra-
`tions relative to the unsubstituted inhibitor [59]. In an examination of pyrrolidines
`cyclopropanated at either the 3,4 or 4,5 positions, it was found that while intro-
`duction of the cyclopropane on the face of the pyrrolidine trans to the cyano group
`’ s, the cis-3,4-methano and cis-4,5-methano
`led to compounds with micromolar IC50
`moieties were well tolerated. One goal of this work was to reduce the intramolecular
`amine-nitrile cyclization that plagues many cyanopyrrolidine DPP4 inhibitors.
`Bulky substituents on the amino acid and the cyclopropane moiety provided im-
`pressive improvements in solution stability. Compound 11 (IC50 ¼ 1.5 nM) has a
`half-life of 5 hours, while compound 12 (Ki ¼ 8 nM) has one of 27 hours and com-
`pound 13 (Ki ¼ 7 nM), 42 hours. Compound 13 reduced glucose excursions fol-
`lowing an oral glucose tolerance test (OGTT) in Zucker fa/fa rats [60].
`
`F
`
`N
`
`O
`
`10
`
`CN
`
`H2N
`
`CN
`
`N
`
`O
`
`11
`
`CN
`
`N
`
`O
`
`12
`
`H2N
`
`CN
`
`N
`
`O
`
`13
`
`H2N
`
`H2N
`
`Ketopyrrolidines and ketoazetidines, which replace the cyano group with a
`heteroaryl ketone, have also been examined as DPP4 inhibitors. Heteroaryl ketones
`have been used extensively as reversible serine protease inhibitors and act by pro-
`viding an electrophilic carbonyl that can form a tetrahedral species with the active
`site serine. An examination of rings from four to six atoms revealed that only the
`piperidine derivatives were not inhibitors of the enzyme. 2-Thiazolyl and 2-ben-
`zothiazolyl substituents provided sufficient activation of the carbonyl to give low
`nanomolar inhibitors such as 14 (IC50 ¼ 30–42 nM). These compounds suffer from
`an internal cyclization followed by oxidation to give dihydroketopyrazines such as
`15 [61].
`
`MYLAN Ex. 1011, Page 5
`
`

`
`154
`
`S.L. Gwaltney, II and J.A. Stafford
`
`S
`
`N
`
`N
`
`O
`
`O
`
`N
`
`N
`
`S
`
`N
`
`H2N
`
`O
`15
`14
`Substituted cycloalkylglycine thiazolidides and pyrrolidides are potent DPP4 in-
`hibitors. Compound 16 (IC50 ¼ 88 nM) demonstrated good PK in both the rat and
`dog with bioavailabilities of 36% and 100%, respectively [62]. An extensive exam-
`ination of the SAR surrounding cyclopentyl and cyclohexylglycine derived pyrro-
`lidides and thiazolidides has been reported. The cyclopentylglycine derivatives were
`found to be more potent than their cyclohexyl counterparts. While the thiazolidides
`provided greater potency, these compounds suffered from reduced metabolic sta-
`bility. Compound 17 (IC50 ¼ 13 nM) was found to be a potent inhibitor selective for
`DPP4 over QPP and PEP [63]. In a series of mono or disubstituted pyrrolidides with
`fluorine at the 3 and 4 positions, the monofluorinated compounds were more potent
`than the difluoro analogs. Compound 18 (IC50 ¼ 48 nM) was bioavailable in rat
`and dog and gave a 42% reduction in glucose excursion following an OGTT in lean
`mice [64]. This compound undergoes metabolic activation and subsequent conju-
`gation with biological nucleophiles. This is believed to occur through oxidation and
`defluorination events, which produce an enal that acts as a Michael acceptor [65].
`Compounds 19 (IC50 ¼ 6 nM) and 20 (IC50 ¼ 6 nM) are potent inhibitors of DPP4
`that also incorporate the monofluorinated pyrrolidine [66].
`
`F
`
`N
`
`O
`
`HN
`
`S
`
`F
`
`O O
`
`F
`
`HN
`
`S
`
`O O
`
`O
`
`O
`
`S
`
`HN
`
`S
`
`F
`
`O O
`
`F
`
`H
`H2N
`
`18
`
`F
`
`N
`
`O
`
`H2N
`
`N
`
`HN
`
`O
`
`S
`
`N
`
`O
`
`H
`H2N
`
`17
`
`F
`
`F3C
`
`N
`
`O
`
`H
`H2N
`
`16
`
`F3C
`
`N
`
`S
`
`F
`
`N
`
`O
`
`H2N
`
`19
`20
`Starting from high-throughput screening (HTS) hit 21 (IC50 ¼ 1.9 mM), a series of
`b-homophenylalanine thiazolidides was developed [67]. Substitution of fluorine at the
`
`MYLAN Ex. 1011, Page 6
`
`

`
`Inhibitors of Dipeptidyl Peptidase 4
`
`155
`
`2-position of the phenyl ring was found to provide an approximately 3-fold improvement
`in potency. The most potent compound reported in this series was 22 (IC50 ¼ 119 nM).
`This work was extended to a series of proline and thiazolidine amides such as 23
`(IC50 ¼ 0.48 nM). While very potent, these analogs demonstrate poor PK properties
`[68]. Investigation of the SAR in a series of related piperazines represented by 24
`(IC50 ¼ 19 nM) revealed that the R-benzyl group was important for potency [69]. These
`analogs also suffer from short metabolic half-lives due to oxidation of the piperazine ring
`and poor pharmacokinetics. These liabilities were addressed through the discovery of
`MK-0431, which will be discussed in the section on clinical DPP4 inhibitors.
`
`F
`
`CO2H
`
`F
`
`F
`
`F
`
`F
`
`F
`
`NH2
`
`O
`
`22
`
`N
`
`S
`
`NH2
`
`O
`
`N
`
`24
`
`Ph
`
`NH
`
`Cl
`
`O
`
`NH
`
`HN
`
`O
`
`NH2
`
`O
`
`N
`
`21
`
`O
`
`HN
`
`O
`
`NH2
`
`O
`
`N
`
`23
`
`Ph
`
`F
`
`F
`
`Sulphostin (25) is a natural product with an IC50 of 6.0 ng/mL, which corres-
`ponds to approximately 20 nM [70,71]. In an examination of the structure-activity
`relationships for analogs of sulphostin, it was found that the carbonyl and C-3
`amino group of the parent structure were important for maintaining potency, as
`was the absolute configuration at phosphorus. Heterocycle ring sizes of 5–7 atoms
`were well tolerated. The sulfonic acid moiety could be removed while maintaining
`potency, but deletion of this group negatively impacts the stability of these analogs.
`Compound 26 is an 11 nM inhibitor of DPP4 [72].
`
`NH2
`
`NH2
`
`NH2
`
`NH2
`
`N
`
`N
`
`Cl
`
`Ar
`
`N
`
`Cl
`
`28: Ar = phenyl
`29: Ar = 3,5-dimethoxyphenyl
`
`OO
`
`N
`
`27
`
`NH2
`
`N
`
`O
`
`PO
`
`NHSO3H
`
`NH2
`
`25
`
`N
`PO
`
`H2N
`
`NH2
`
`O
`
`NH2
`
`26
`
`Starting from HTS hit 27 (IC50 ¼ 10 mM), the potency in a series of amino-
`methylpyrimidines was improved 100,000 fold through modification of the two aryl
`
`MYLAN Ex. 1011, Page 7
`
`

`
`156
`
`S.L. Gwaltney, II and J.A. Stafford
`
`substituents [73]. Ortho and para substituents on the C-6 phenyl ring were tolerated,
`whereas meta substitution generally led to loss of potency. A breakthrough was
`realized when the 2,4-dichlorophenyl derivative 28 (IC50 ¼ 10 nM) was prepared.
`Optimization of 28 through modification of the C-2 phenyl group led to compound
`29, reported to be a 100-picomolar inhibitor. An X-ray crystal structure of 29 in
`DPP4 reveals that the dichlorophenyl group effectively fills S1, the pyrimidine ring
`forms a cation-p interaction with Arg125, the aminomethyl group interacts with
`Tyr662 and the two active-site glutamates, and the anilino nitrogen forms an ad-
`ditional H-bond with the backbone carbonyl of Glu205.
`Analogs of these compounds where the pyrimidine is replaced with a pyridine
`(e.g. 30) were explored. It was found that potency is improved by reducing the
`torsion angle between the phenyl ring and the pyridine core. Optimization in this
`series led to compound 34 [74].
`
`Cmpd
`
`30
`31
`32
`33
`34
`
`n
`
`0
`3
`2
`1
`
`NH2 NH2
`
`N
`
`Cl
`
`(CH2)n
`
`30-33
`
`MeO
`
`Cl
`
`MeO
`
`Torsion (calc.)
`
`IC50 (mM)
`
`321
`221
`01
`
`NH2 NH2
`
`N
`
`Cl
`
`0.92
`0.24
`0.045
`0.039
`0.007
`
`NH2 NH2
`
`N
`
`Cl
`
`O
`
`Cl
`
`N
`
`N
`
`35
`
`34
`
`Cl
`
`Compound 28 is an inhibitor of CYP450 3A4 with an IC50 of 5.4 mM. This
`compound also caused phospholipidosis in cultured fibroblasts. It was anticipated
`that by reducing the lipophilicity of these compounds, improved properties would
`be realized. While replacing the 2-phenyl group with small groups such as Me, OMe
`and NH2 led to significant decreases in potency, groups such as 4-thiomorpholinyl
`and N-hydroxyethyl, N-methylamino were well tolerated. Compound 35 has an
`IC50 of 9 nM versus DPP4, does not induce phospholipidosis and has an IC50 of 30
`mM versus CYP450 3A4 [75].
`
`3. DPP4 INHIBITORS IN CLINICAL DEVELOPMENT
`
`Small-molecule DPP4 inhibitors have advanced into clinical trials. First-generation
`inhibitors P32/98 (2) [76] and NVP-DPP728 (6) [77] were important tools to vali-
`date the concept that DPP4 inhibition is an effective method to improve glucose
`control in diabetic patients through an increase in active GLP-1 levels. In a 4-week
`study evaluating ninety-three type 2 diabetic patients, NVP-DPP728 (6), at doses of
`100 mg three times daily and 150 mg twice daily, demonstrated meaningful reduc-
`tions of plasma glucose, insulin, and the glycohemoglobin, HbA1c [78].
`
`MYLAN Ex. 1011, Page 8
`
`

`
`Inhibitors of Dipeptidyl Peptidase 4
`
`157
`
`The early agents, however, have been replaced by a second generation of DPP4
`inhibitors, which have improved potency, selectivity, and pharmacokinetics over
`their pioneering predecessors. These include LAF237 (36, vildagliptin), MK-0431
`(37, Sitagliptin), BMS-477118 (38, saxagliptin), and GSK23A (39). Other DPP4
`inhibitors that are known to have entered clinical trials are P93/01 and SYR322,
`though the chemical structures of these compounds have not been disclosed [79,80].
`
`NH2
`
`O
`
`N
`
`37
`
`N
`
`N
`
`N
`
`CF3
`
`O
`
`CN
`
`N
`
`NH2
`
`S
`
`O O
`
`39
`
`F
`
`F
`
`F
`
`F
`
`NC
`
`O
`
`N
`
`NH
`
`36
`
`O
`
`CN
`
`HO
`
`HO
`
`N
`
`NH2
`
`38
`
`CH3O
`
`The chemical architecture of the DPP4 inhibitors that have been advanced into
`clinical trials is interesting both for their common structural features and those
`features that make them unique. A cyanopyrrolidine amide is a frequently repeating
`motif in small-molecule DPP4 inhibitors. In each of the compounds possessing this
`functionality, the cyano group undergoes nucleophilic addition by the catalytic se-
`rine, resulting in covalent modification of the DPP4 enzyme. Introductions to the
`pyrrolidine moiety of fluorine substitution (e.g., 39) or a fused ring (e.g., 38) are
`reported to provide improved in vivo and/or stability properties. An a-amino acid
`fragment is also common among DPP4 inhibitors, and it has been revealed from
`crystallography studies that the basic, protonated amino group interacts with a pair
`of Glu residues in the DPP4 active site. Like the modifications on the pyrrolidine
`ring, the installation of a quaternary center adjacent to this essential amino group
`serves the purpose of inhibiting the internal cyclization reaction, which gives rise to a
`bicyclic amidine that undergoes further hydrolysis to a diketopiperazine. In the case
`of NVP-DPP728, the observed by-product of this decomposition pathway is 40 [81].
`
`O
`
`H
`
`N
`
`N
`
`O
`
`40
`
`NH
`
`N
`
`NC
`
`MYLAN Ex. 1011, Page 9
`
`

`
`158
`
`S.L. Gwaltney, II and J.A. Stafford
`
`LAF237 (36) is a 4 nM inhibitor of DPP4 with 410,000-fold selectivity over post-
`proline converting enzyme (PPCE) and DPP-II [81]. The identification of the
`1-amino-3-hydroxyadamantane ring system was the result of careful and systematic
`SAR studies on the P-2 site, wherein a loss of enzymatic potency was observed with
`both carbamate and ester derivatives of the 3-hydroxy group. As expected, the steric
`encumbrance of the adamantane retards the rate of intramolecular cyclization by
`about 30-fold. Compound 36 demonstrated oral efficacy in a standard OGTT
`model using obese Zucker fa/fa rats.
`In a 12-week, placebo-controlled phase II trial, 36 was effective at improving
`glycemic control when given as monotherapy to drug-naı¨ ve patients [82]. The mean
`change in HbA1c was 0.6% from a baseline of 8.0. The most common adverse event
`was hypoglycaemia, which was observed in 7 patients (10%) and was considered
`mild.
`In a double-blinded trial, 107 patients with type 2 diabetes being treated with
`metformin, patients were randomized to receive 36 or placebo [83]. After 12 weeks
`of combination treatment, patients completing the 12-week therapy study were
`eligible to extend treatment to 52 weeks. In the 56 patients randomized to 36 (50 mg
`once daily), HbA1c levels after 12 weeks were reduced by 0.6% from an average
`baseline of 7.7%. In contrast, HbA1c levels were unchanged in patients (n ¼ 107)
`receiving placebo. In the 42 patients that progressed to the extended study, HbA1c
`levels in the 36-treated groups were unchanged from 12 weeks to 52 weeks. Patients
`in the metformin-plus-placebo group (n ¼ 29) showed a gradual increase in HbA1c
`over the 40-week extension, resulting in a between-group average difference of
`–1.1% HbA1c levels. These 52-week data on 36 in combination with metformin
`provide compelling evidence that DPP4 inhibition represents a robust method for
`longer-term glycemic control.
`The discovery of MK-0431 (37, Sitagliptin), and the incorporation of a b-amino
`acid moiety, represented a notable departure from the characteristic a-amino acid
`fragment featured in most reported DPP4 inhibitors [84]. Indeed, X-ray evidence
`suggests a binding orientation of the amide carbonyl that is opposite to its a-amino
`acid progenitors, with the b-amino group retaining the predicted interactions to the
`Glu 205/206 pair. Compound 37 is a non-covalent, 18 nM inhibitor of DPP4 that
`possesses excellent selectivity (2000 to 5000-fold) over related peptidases DPP2,
`DPP8, and DPP9. The desired animal pharmacokinetics of 37 came from modi-
`fications of piperazine amides such as 24, which undergo extensive metabolism yet
`are also potent DPP4 inhibitors [69]. Interestingly, and perhaps generally pertinent
`to the development of safe and effective DPP4 inhibitors for chronic administra-
`tion, DPP8 and DPP9 were specifically highlighted for cross-reactivity as it has
`been reported that the DPP8/9 selective inhibitor 41 is associated with ‘‘multi-organ
`pathology and mortality’’ when administered to rats for 2 weeks at a dose of
`100 mg/kg/day. Moreover, these effects were observed in both wild-type and DPP4-
`deficient mice, suggesting that the toxicities were independent of DPP4 inhibition
`[85]. The preclinical oral efficacy of 37 was demonstrated in an OGTT model using
`C57BL/6N male mice.
`
`MYLAN Ex. 1011, Page 10
`
`

`
`Inhibitors of Dipeptidyl Peptidase 4
`
`159
`
`H3C
`
`CH3
`
`O
`
`N
`
`NH2
`
`41
`
`A randomized, placebo-controlled OGTT in fifty-six type 2 diabetics was con-
`ducted to assess the glucose-lowering activity and safety/tolerability of 37 following
`a single oral dose of 25- or 200-mg. As compared to placebo, incremental glucose
`AUC was reduced by approximately 22% and 26%, for the 25- and 200-mg
`doses respectively [86]. Dose-responsive plasma increases in insulin, c-peptide, and
`GLP-1, and reductions in glucagon were also noted, leading to the conclusion that
`pharmacologic proof-of-concept had been achieved.
`GSK23A (39) is a penicillinamine-based inhibitor of DPP4 with a Ki value of
`53 nM [87]. The compound contains a 4-fluoro substituent on the cyanopyrrolidine
`ring, which confers unique biochemical and physical properties versus its des-fluoro
`analog 42. For example, 39 has a half-time to onset of DPP4 inhibition of 120 min,
`as measured in human plasma, compared to o20 min for compound 42. In ad-
`dition, 39 has a half-time for internal cyclization (37 1C, pH 7.2) of 1733 hr versus
`360 hr for 42. In a standard OGTT in ob/ob mice 39 showed an expected lowering
`of plasma glucose with an increase in both GLP-1 and insulin. By contrast, in the
`db/db mouse model, serum levels of glucose were unchanged following 8 weeks of
`treatment with 39, presumably due to the severe insulin resistance of these animals.
`
`O
`
`CN
`
`S
`
`O O
`
`N
`
`NH2
`
`42
`
`CH3O
`
`BMS-477118 (38) is a methanoproline-based DPP4 inhibitor [88,89]. The cis-
`fused, 4,5-methano bridge on the pyrrolidine ring appears to have been borrowed
`from the existing ACE inhibitor literature, wherein it has been shown that captopril
`analogs such as 43 having a fused cyclopropane ring retain the full ACE inhibitory
`potency of captopril (44) itself [90]. This modification is highlighted as a structural
`tool to enhance the chemical stability of the compound by retarding the rate of the
`internal cyclization reaction. BMS-477118 is a potent inhibitor of DPP4 with a
`reported IC50 value of 0.45 nM. It is reported to be selective over the related pep-
`tidases DPP2, DPP8, DPP9, and fibroblast-activating protein (FAP). In preclinical
`OGTT studies using both Zucker fa/fa rats and ob/ob mice, BMS-477118 displays
`the phenotypic profiles characteristics of DPP4 inhibition including transient lowe-
`ring of glucose with concomitant increases in plasma insulin.
`
`MYLAN Ex. 1011, Page 11
`
`

`
`160
`
`S.L. Gwaltney, II and J.A. Stafford
`
`O
`
`CO2H
`
`O
`
`CO2H
`
`HS
`
`N
`
`HS
`
`N
`
`CH3
`
`43
`
`CH3
`
`44
`
`It is widely accepted that the level of glycohemoglobin, HbA1c, is the best mea-
`sure of long-term glycemic control and should be a primary endpoint for assessing
`the effectiveness of diabetes therapy [91]. A key topic that remains unanswered is
`the relationship between in vivo DPP4 inhibition over time and its effects on both
`HbA1c and safety, though the early reports are highly encouraging that sustained
`DPP4 inhibition is both efficacious and well tolerated. Based on the large volume of
`primary journal literature, patent literature, and reports from scientific meetings, it
`is fair to speculate that there are a number of other small-molecule DPP4 inhibitors
`that are undergoing preclinical and clinical evaluation.
`
`4. ALTERNATIVE INDICATIONS FOR DPP4 INHIBITORS
`
`Numerous studies suggest DPP4 inhibition for pharmacological uses other than the
`restoration of glycemic control and type 2 diabetes. The common underlying hy-
`potheses of these studies are divided into two categories that describe fundamental
`properties of DPP4. CD26, a membrane-associated peptidase that has DPP4 ac-
`tivity has been extensively studied in relation to its role in regulating T-cell phys-
`iology. As such, DPP4 activity has been studied in the context of T-cell activation
`and immune function [92]. Activated T-cells are known to have increased cell-
`surface expression of DPP4 [93]. Furthermore, cytokines such as RANTES,
`SDF-1a, MCP-2, and TNF-a have all been characterized as substrates for DPP4, so
`it is a reasonable hypothesis that DPP4 plays an immunomodulatory role [94].
`However, recent studies have raised questions on the dependence of DPP4’s pro-
`teolytic activity to T-cell activation and other functions such as proliferation and
`cytokine release. In a study using compounds with varying selectivity profiles for
`DPP4 and related peptidases, such as QPP, DPP8, and DPP9, it was found that
`both a DPP4 and a QPP inhibitor had no effects in in vitro assays measuring the
`immune responses of T-cell proliferation and IL-2 release [95]. By contrast, less
`selective compounds, such as Val-boro-Pro (45) and Lys[Z(NO2)]-pyrrolidide (46),
`which also show inhibitory activity against DPP8 and DPP9, were effective in these
`assays. The authors conclude that the T-cell-mediated effects previously assigned to
`the inhibition of DPP4 might actually be a consequence of DPP8 and/or DPP9
`activity. In addition, the uncompetitive DPP4 inhibitor, TMC-2A (47), [96] has
`been shown to suppress paw swelling in a rat adjuvant model for arthritis [97]. The
`authors of this study also note that DPP4 inhibition, per se, does not affect T-cell
`function, and that mice with mutated DPP4 lacking enzymatic activity still show a
`normal immune response. It is suggested that the binding of TMC-2A (47) may
`
`MYLAN Ex. 1011, Page 12
`
`

`
`Inhibitors of Dipeptidyl Peptidase 4
`
`161
`
`affect the function of other proteins that associate with CD26, specifically the PTPase
`activity of CD45 [98]. Compound 45 is also known as talabostat or PT-100 and is in
`clinical trials for hematological malignancies and hematopoiesis [99]. The efficacy of
`45 in these therapeutic indications may be derived from the compound’s inhibition of
`FAP [100]. Numerous other reports have focused on the involvement of DPP4 and/
`or use of DPP4 inhibitors in various inflammatory and autoimmune diseases and
`pathologies, such as Crohn’s disease, [92] and organ transplantation [101].
`
`OMe
`
`HO
`
`OH
`
`OH
`
`OH
`
`OH
`
`O
`
`NH
`
`47
`
`N
`
`O
`
`O
`
`HN
`
`H2N
`
`O
`
`O
`
`NH
`
`H2N
`
`46
`
`N
`
`O
`
`–O
`
`N+
`O
`
`H2N
`
`N
`
`O
`
`45
`
`B(OH)2
`
`In addition to its identity with the membrane-associated protein CD26, DPP4 is
`ubiquitously distributed, and many important biomolecules other than GLP-1, in-
`cluding hormones, neuropeptides, chemokines, and cytokines, have been charac-
`terized as DPP4 substrates [3]. Among these substrates for DPP4, the peptide
`hormone GLP-2 has received considerable attention recently for its activity as an
`intestinal growth factor [102]. This activity has led to the hypothesis that a DPP4
`inhibitor could show intestinotrophic effects that may be useful in the treatment of
`inflammatory bowel disease (IBD). Indeed GLP-2 itself has shown efficacy in a
`rodent model of IBD [103].
`
`REFERENCES
`
`[1] D. J. Drucker, Exp. Opin. Investig. Drugs, 2003, 12, 87.
`[2] P. E. Wiedeman and J. M. Trevillyan, Curr. Opin. Investig. Drugs, 2003, 4, 412.
`[3] A.-M. Lambeir, C. Durinx, S. Scharpe´ and I. De Meester, Crit. Rev. Clin. Lab. Sci.,
`2003, 40, 209.
`[4] J. J. Holst, Exp. Opin. Emerg. Drugs, 2004, 9, 155.
`[5] C. F. Deacon, B. Ahre´ n and J. J. Holst, Exp. Opin. Investig. Drugs, 2004, 13, 1091.
`[6] J. J. Holst and C. F. Deacon, Curr. Opin. Pharmacol., 2004, 4, 589.
`[7] A. E. Weber, J. Med. Chem., 2004, 47, 4135.
`[8] B. Ahre´ n, Current Enzyme Inhibition, 2005, 1, 65.
`[9] C.H.S. McIntosh, H.-U. Demuth, J.A. Pospisilik and R. Pederson, Regul. Pept., 2005,
`159.
`[10] E. M. Sinclair and D. J. Drucker, Curr. Opin. Endocrinol. Diabet., 2005, 12, 146.
`[11] R. Mentlein, Exp. Opin. Investig. Drugs, 2005, 14, 57.
`[12] T. Hoffman and H.-U. Demuth, in Ectopeptidases, (eds Langner and Ansorge), Kluwer
`Academic/Plenum Publishers, New York, 2002, p. 259.
`[13] E. B. Villhauer, G. M. Coppola and T. E. Hughes, Annu. Rep. Med. Chem., 2001, 36,
`191.
`[14] B. Leiting, K. D. Pryor, J. K. Wu, F. Marsilio, R. A. Patel, C. S. Craik, J. A. Ellman,
`R. T. Cummings and N. A. Thornberry, Biochem. J., 2003, 371, 525.
`
`MYLAN Ex. 1011, Page 13
`
`

`
`162
`
`S.L. Gwaltne