2. DPPIV Inhibition: Promising Therapy
`for the Treatment of Type 2 Diabetes
`
`PAUL E. WIEDEMAN
`
`Abbott Laboratories, Department R4CP, Building AP9B, 100 Abbott Park
`Road, Abbott Park, IL 60064-6113, USA
`
`INTRODUCTION
`Type 2 Diabetes Pandemic
`Inhibition of DPPIV as a Strategy to Enhance the Incretin Effect
`Dipeptidyl Peptidase IV
`Potential for Disease Modification with DPPIV Inhibitors
`Importance of DPPIV Selectivity
`Structural Biology
`
`MEDICINAL CHEMISTRY AND PRECLINICAL STUDIES
`Early DPPIV Inhibitors
`Covalent DPPIV Inhibitors: P2 Secondary Amine Analogues
`Covalent DPPIV Inhibitors: P2 Primary Amine Analogues
`Non-Covalent DPPIV Inhibitors: P2 Primary Amine Analogues
`Peptidic DPPIV Inhibitors that Extend into P10
`Non-Peptidic/Non-Covalent DPPIV Inhibitors
`
`CLINICAL DATA FOR ORAL DPPIV INHIBITORS
`
`CONCLUSIONS
`
`AUTHORS NOTE
`
`REFERENCES
`
`64
`64
`64
`66
`68
`68
`69
`
`71
`71
`74
`80
`84
`89
`90
`
`96
`
`101
`
`102
`
`102
`
`Progress in Medicinal Chemistry – Vol. 45
`Edited by F.D. King and G. Lawton
`DOI: 10.1016/S0079-6468(06)45502-8
`
`63
`
`r 2007 Elsevier B.V.
`All rights reserved.
`
`AstraZeneca Exhibit 2162
`Mylan v. AstraZeneca
`IPR2015-01340
`
`Page 1 of 47
`
`

`
`64
`
`DPPIV INHIBITION
`
`INTRODUCTION
`
`TYPE 2 DIABETES PANDEMIC
`
`Diabetes is a major health problem encountered across the globe. Nearly
`200 million individuals worldwide suffer from diabetes of which 90–95% are
`type 2 diabetics [1]. Some staggering figures include India, home to 35.5
`million diabetics, China with 23.8 million, the United States with 16 million,
`Russia with 9.7 million and Japan with 6.7 million. The incidence of di-
`abetes has not peaked. By 2025, the diabetic population of Africa, the
`Eastern Mediterranean, the Middle East and Southeast Asia is expected to
`increase by 100%. Rises are anticipated in other regions as well, including
`Central and South America (85%), the Western Pacific (75%), North
`America (50%) and Europe (20%). Already diabetes is the fourth main
`cause of mortality in the majority of developed countries. Healthcare
`systems will certainly be strained to meet the growing demand of this pan-
`demic, as even now 50% of patients are undiagnosed. The growing diabetic
`population requires additional therapies with alternative mechanisms of
`action and improved tolerability. There are currently five main classes of
`oral antidiabetics, each limited in one way or other by the degree of efficacy
`and side effects [2]. Concerns with sulfonylureas centre on hypoglycemia
`and weight gain. Non-sulfonylurea secretagogues have the same issues along
`with a more complex dosing schedule. Patients on thiazolidinediones are
`prone to weight gain and oedema. Biguanides and a-glucosidase inhibitors
`often produce significant gastrointestinal distress. Inhibition of dipeptidyl
`peptidase IV (DPPIV) cleavage of glucagon-like peptide-1 (GLP-1) is a
`highly validated target for the treatment of type 2 diabetes. Several DPPIV
`inhibitors are in clinical development, and the first requests for regulatory
`review have been filed. The reported clinical data have established proof of
`concept in man, confirming the possibility that DPPIV inhibitors will be the
`next major new class of oral antidiabetic drug.
`
`INHIBITION OF DPPIV AS A STRATEGY TO ENHANCE THE INCRETIN EFFECT
`
`The phenomenon of increased insulin secretion following oral administra-
`tion of glucose compared to intravenous administration is known as the
`incretin effect. An agent responsible for this effect is the incretin hormone,
`GLP-1. In response to the oral ingestion of nutrients, proglucagon is proc-
`essed and GLP-1 is released from enteroendocrine L-cells in the distal small
`intestine and colon. Binding of GLP-1 to its G-protein-coupled receptor on
`pancreatic b-cells increases glucose-stimulated insulin secretion [3, 4].
`Additional desirable effects of GLP-1 include increased insulin gene
`
`Page 2 of 47
`
`

`
`P. E. WIEDEMAN
`
`65
`
`expression [5], and increased pancreatic b-cell proliferation and islet neo-
`genesis [6]. By contrast and also of benefit are the inhibition of glucagon
`secretion [7, 8] and decreased gastric emptying that results in the slowed rate
`of nutrient absorption [9,10]. GLP-1 receptors are also expressed in hypo-
`thalamic nuclei responsible for modulating feeding behaviour, and periph-
`eral administration of GLP-1 promotes satiety and inhibits food intake in
`man [11, 12]. In total these effects demonstrate that GLP-1 (7–36) amide has
`multiple biological effects that contribute to glucose homeostasis and pro-
`motes normalization of post-meal glucose levels. It is important to recognize
`that GLP-1 augments insulin secretion in a glucose-dependent manner.
`Unlike sulfonylurea drugs or insulin, enhancing endogenous GLP-1 levels
`does not increase the risk of hypoglycemia.
`Infusion of active GLP-1 and GLP-1 (7–36) amide reduces post-meal and
`fasting glycemia in patients with non-insulin-dependent diabetes mellitus;
`thus establishing the potential of GLP-1-based therapy for the treatment of
`type 2 diabetes [13–15]. This effect occurs despite a blunting of the incretin
`effect in type 2 diabetics [16]. The key problem, however, is that active GLP-
`1 (7–36) amide is rapidly converted to inactive GLP-1 (9–36) amide by the
`action of DPPIV via the cleavage of the N-terminal dipeptide (His-Ala) of
`GLP-1 (7–36) amide [17, 18]. The short half-life of GLP-1 (7–36) amide in
`the circulation (o2 min) makes it impractical as a therapeutic agent and has
`led to the development of alternative strategies to enhance the anti-
`diabetogenic activity of GLP-1. One successful approach that will not be
`covered in this chapter is the development of GLP-1 receptor agonists that
`are resistant to DPPIV cleavage [19]. This approach remains an active area
`of research and development [20]. Exenatide (Byetta) is the first marketed
`GLP-1 receptor agonist. It is effective in reducing glycosylated haemo-
`globin, HbA1c (biomarker for glycemic control) levels in type 2 diabetics but
`is a twice-a-day injectable peptide with nausea as a prominent side effect
`[21]. Another strategy that is the focus of this chapter is to increase the
`circulating half-life of endogenous GLP-1 by inhibiting its enzymatic de-
`gradation by DPPIV [17] (Fig. 2.1).
`Another incretin hormone, glucose-dependent insulinotropic polypeptide
`(GIP), is also degraded by DPPIV [18]. Similar to GLP-1, GIP is a 42-amino
`acid peptide secreted by endocrine K cells of the duodenum in response to
`ingestion of nutrients [22]. The physiological actions of GIP include glucose-
`dependent potentiation of insulin secretion and regulation of insulin gene
`transcription. In contrast to GLP-1, glucose tolerance is not improved in
`type 2 diabetics treated with exogenous GIP [23]. However, it has been
`reported that the response to GIP improves in diabetic patients treated with
`glyburide to reduce fasting glucose levels [22]. Studies with hyperglycemic
`diabetic rats or normal rats made hyperglycemic by glucose clamp have
`
`Page 3 of 47
`
`

`
`66
`
`DPPIV INHIBITION
`
`Fig. 2.1 Inhibition of DPPIV sustains endogenous GLP-1 and modulates the incretin effect.
`
`decreased expression of pancreatic GIP receptor, which in turn causes a loss
`of insulinotropic response to GIP [24]. Taken together, these data support
`the notion that the initial effect of DPPIV inhibition in human diabetes is
`primarily by GLP-1, but raises the possibility that in the long run, when
`glucose levels fall, treatment with DPPIV inhibitors could also improve the
`insulinotropic action of GIP.
`
`DIPEPTIDYL PEPTIDASE IV
`
`DPPIV (EC 3.4.14.5; also known as lymphocyte cell surface protein CD26)
`was first described in 1967 [25]. DPPIV is a post-proline cleaving serine
`protease with a catalytic triad of Ser-Asp-His oriented inversely to classical
`serine proteases and with significant homology to other a,b-hydroxylases.
`DPPIV is expressed as a 110 kDa glycoprotein on the surface of cells of most
`tissues including kidney, liver, intestine, placenta, prostate, skin, lymph-
`ocytes and endothelial cells. DPPIV is catalytically active as a dimer. Pro-
`teolytic cleavage of DPPIV from cell surfaces results in a soluble circulating
`form with a monomeric mass of approximately 100 kDa. In addition to
`cleaving GLP-1, DPPIV may play a role in the cleavage of other substrates
`with accessible amino-terminal Xaa-Pro- or Xaa-Ala-dipeptide sequences,
`resulting in their inactivation or alteration in their biological activities. Po-
`tential DPPIV substrates include growth hormone releasing hormone, GIP,
`pituitary adenylate cyclase-activating polypeptide 38 (PACAP38), substance
`P, bradykinin, gastrin releasing peptide, neuropeptide Y, peptide YY,
`
`Page 4 of 47
`
`

`
`P. E. WIEDEMAN
`
`67
`
`certain chemokines such as RANTES (regulated on activation normal T cell
`expressed and secreted), stromal cell-derived factor, eotaxin and macroph-
`age-derived chemokines [26]. Long-term safety concerns have arisen because
`these possible DPPIV substrates include chemokines, vasoactive peptides,
`neuropeptides and gastrointestinal peptides. Despite in vitro cleavage of
`these peptides by DPPIV, many of the activities associated with these pep-
`tides appear not to be physiologically regulated in vivo by DPPIV action.
`Those that appear to be regulated by DPPIV peptidase activity are described
`below.
`GLP-1 and GIP have been validated as in vivo substrates of DPPIV in
`DPPIV knockout (DPPIV KO) mice and DPPIV-deficient Fisher rats [27,
`28]. The importance of GLP-1 and GIP in the gluco-regulatory action of
`DPPIV inhibitors has been shown in double incretin receptor (i.e., GLP-1R
`and GIPR) knockout mice where the glucose-lowering effect of DPPIV
`inhibitors is abolished [29]. DPPIV-deficient mice are healthy, have normal
`blood glucose levels in the fasted state but reduced glucose excursion after a
`glucose challenge [27]. The active, insulinotropic form of GLP-1 and glu-
`cose-dependent insulin levels are both increased in DPPIV KO mice com-
`pared to wild-type littermates. Similarly, DPPIV-deficient Fisher rats have a
`phenotype of improved glucose tolerance and enhanced glucose-dependent
`insulin secretion [28].
`Similarly, PACAP38 has been validated as an additional target of in vivo
`DPPIV peptidase activity. PACAP38 is a neuropeptide that is involved with
`signalling to pancreatic nerves and is therefore associated with neural reg-
`ulation of islet function. Preservation of endogenous levels of PACAP38
`with a DPPIV inhibitor may be an additional way that the inhibitors en-
`hance antidiabetic effects. PACAP38 was administered exogenously to both
`wild-type and DPPIV-deficient mice [30]. In the DPPIV-deficient mice, the
`rate of PACAP38 clearance was reduced and little of the DPPIV metabolite,
`PACAP(3–38) was observed. In another study in mice, PACAP38 was ad-
`ministered intravenously with glucose following previous administration of
`a DPPIV inhibitor [31]. As was observed in the same study with GLP-1, the
`PACAP38-treated animals showed increased insulin levels and a greater rate
`of glucose elimination.
`Although DPPIV clearly regulates GLP-1 (and probably GIP and
`PACAP38) action in vivo, the DPPIV enzyme may have a broader role in
`metabolic control. In this regard, DPPIV KO mice show resistance to diet-
`induced obesity with a concomitant reduction in adiposity compared to
`wild-type mice [32]. Consequently, long-term inhibition of DPPIV may have
`the potential to decrease the body weight. However, DPPIV inhibitors cur-
`rently in clinical development have not demonstrated weight reduction, but
`several trials have shown weight neutrality.
`
`Page 5 of 47
`
`

`
`68
`
`DPPIV INHIBITION
`
`POTENTIAL FOR DISEASE MODIFICATION WITH DPPIV INHIBITORS
`
`GLP-1 has trophic effects on b-cells [33]. Not only does it stimulate b-cell
`proliferation but it also inhibits apoptosis of b-cells [34]. GLP-1 is capable of
`enhancing the differentiation of new b-cells from pancreatic ductal epithelium
`[6, 35]. DPPIV-mediated preservation of endogenous GLP-1 may also
`enhance the effects of GLP-1 on b-cell rescue/prevention of apoptosis.
`Indeed, chronic treatment with DPPIV inhibitors has been shown to preserve
`islet function in diabetic mice and to improve b-cell survival and islet cell
`neogenesis in streptozotocin-induced diabetic rats [36, 37]. It has been reported
`that humans with type 2 diabetes have a b-cell deficit most likely due to
`increased b-cell apoptosis [38]. Consequently, by increasing GLP-1 levels,
`DPPIV inhibitors may be capable of providing new b-cells in patients with
`type 2 diabetes and thus, prevent the worsening of the disease. Furthermore,
`the treatment of
`insulin-resistant
`individuals (impaired glucose-tolerant
`pre-diabetics) with DPPIV inhibitors may delay or prevent the onset of
`diabetes. This would be a major medical breakthrough in the treatment of type
`2 diabetes and impaired glucose tolerance. However, clinical studies to date
`have not provided conclusive evidence that the b-cell sparing activity observed
`with DPPIV inhibitors or GLP-1 analogues in diabetic animal models will also
`occur in man. The diagnostics to directly measure b-cell mass in the clinical
`setting simply do not exist yet. However, improvements in b-cell function can
`and have been assessed. On an optimistic note, an apparent decrease in the
`rate of progression of type 2 diabetes has been observed with a DPPIV in-
`hibitor in combination with metformin in a long-term clinical study [39].
`
`IMPORTANCE OF DPPIV SELECTIVITY
`
`DPPIV is a member of a family of closely related enzymes that share DPP-
`IV-like catalytic activity [40]. Consequently, there is potential for adverse
`events or toxicity associated with non-selective DPPIV inhibitors. Two
`members of the DPPIV family, DPP8 [41] and DPP9 [42, 43], were recently
`discovered and characterized. Their exact physiologic role remains under
`investigation. However, profound toxicities in rodent and adverse side
`effects in dog have been reported with a DPP8/9 inhibitor [44]. Merck sci-
`entists reported in 2-week rat toxicology studies that a selective dual DPP8/
`DPP9 inhibitor (1), IC50 ¼ 38 and 55 nM, respectively, caused alopecia,
`thrombocytopenia, anaemia, enlarged spleen, multiple histological path-
`ologies and death. In an acute dog study, the DPP8/9 inhibitor produced
`bloody diarrhea, emesis and tenesmus. The same toxicity was observed in
`both models with a DPPIV inhibitor (2) non-selective against DPP8/9, but
`was not observed with a selective DPPIV inhibitor MK-0431 (3) (DPPIV
`
`Page 6 of 47
`
`

`
`P. E. WIEDEMAN
`
`69
`
`F
`
`F
`
`NH2
`
`O
`
`N
`
`N
`
`N
`
`F
`(3)
`DPPIV IC50 = 27 nM
`DPP8 IC50 > 69000 nM
`DPP9 IC50 > 100000 nM
`
`N
`
`CF3
`
`O
`
`N
`
`NH2
`
`HN
`
`O
`
`O
`
`(5)
`DPPIV IC50 = 1300 nM
`DPP8 IC50 = 154 nM
`DPP9 IC50 = 165 nM
`
`Me
`
`Me
`
`O
`
`N
`
`NH2
`
`(1)
`DPPIV IC50 = 30000 nM
`DPP8 IC50 > 38 nM
`DPP9 IC50 > 55 nM
`Me
`
`O
`
`N
`
`S
`
`Me
`
`N
`
`NH2
`
`Me
`
`Me
`
`O
`
`NH2
`
`B(OH)2
`
`O2N
`
`(2) (allo-isomer)
`DPPIV IC50 = 460 nM
`DPP8 IC50 = 220 nM
`DPP9 IC50 = 320 nM
`
`(4)
`DPPIV IC50 < 4 nM
`DPP8 IC50 =4 nM
`DPP9 IC50 = 11 nM
`
`Fig. 2.2 Selectivity for DPPIV over related peptidases.
`
`IC50 ¼ 27 nM, DPP8 ¼ 69mM, DPP9>100 mM). Importantly, several DPP-
`IV inhibitors reported in earlier studies were found to lack selectivity for
`DPPIV over DPP8/9. Some of the compounds such as (4) and (5) with poor
`selectivity were active in in vitro models of T-cell activation while DPPIV
`selective compounds exhibited no response. This data supports the premise
`that non-selective DPPIV inhibitors likely exhibit off-target
`toxicities.
`Additionally, skin lesions have been observed in monkeys treated with DPP-
`IV inhibitors. Although the mechanism of action has yet to be determined, the
`U.S. Food and Drug Administration has insisted that all DPPIV inhibitors be
`tested in monkeys [45]. This data is relatively recent and a significant amount
`of the medicinal chemistry efforts described herein occurred prior to this
`understanding. It has been placed here to put the subsequent research in
`context and because the selectivity between prolylpeptidases may be a differ-
`entiating factor for drug candidates. Regardless of the outcome of all these
`evaluations,
`it would be prudent to remain cautious as development of
`DPPIV inhibitors for the treatment of type 2 diabetes proceeds (Fig. 2.2).
`
`STRUCTURAL BIOLOGY
`
`DPPIV has proven to be a drug target that has benefited from structural
`biology guidance, in particular crystallography. Initially, several homology
`models were constructed that provided preliminary insight into possible
`structural features [46–48]. Following was an extensive effort that provided
`
`Page 7 of 47
`
`

`
`70
`
`DPPIV INHIBITION
`
`X-ray crystal structures of recombinant human DPPIV [49–53], natural
`source porcine DPPIV [54], natural source rat DPPIV [55] and bacterial
`DPPIV [56]. Both the apo structure and those with bound inhibitors were
`examined. Many features were found as previously predicted by comparison
`with related peptidases. However, there were certainly some surprises. To
`summarize, DPPIV exists as a dimer with two domains, an a/b hydrolase
`domain and a rather rare eight-bladed propeller domain [57] with a 30–45 A˚
`cavity between them. Inhibitors bind within the cavity next to the catalytic
`triad. Two openings access the cavity, a funnel-shaped opening through the
`b-propeller and a larger opening between the hydrolase and propeller do-
`mains that is believed to be the entry route of substrates. Both are negatively
`charged, which attract the positively charged amine moiety found in all
`inhibitors. Glutamic acid residues 205 and 206 form a salt bridge to the
`amine group in the P2 portion of inhibitors [58] and this interaction is
`responsible for orienting the N-terminal portion of peptide substrates for
`cleavage. Steric constraints limit the enzyme to dipeptide cleavage. Aspara-
`gine 710 and arginine 125 hydrogen bond to the carbonyl of a P2 amino
`acid. A small hydrophobic pocket is optimally filled with small flat rings,
`such as pyrrolidine or phenyl of the inhibitor, and the oxyanion hole is
`formed by two tyrosines. Tyr547 was found to be as critical as any of the
`catalytic triad for full enzymatic activity for its ability to stablilize the oxy-
`anion intermediate [59]. Interactions for specific inhibitors will be elaborated
`in the medicinal chemistry discussions of those compounds (Fig. 2.3).
`As previously mentioned, the active form of the enzyme is a dimer and once
`formed, there is no equilibration back to the monomeric species [60]. Hydro-
`phobic forces were determined to be responsible for dimer formation with
`residues Phe713, Trp734 and Tyr735, all found to be critical. The extent of
`glycosylation was determined not to be a factor affecting dimer formation [61].
`DPPIV is an amazingly dynamic enzyme. As mentioned above, there are
`two channels which access the catalytic region, and analogy to the similar
`enzymes tricorn protease and prolyl oligopeptidase suggested that substrate
`entrance and cleavage products’ exit could be through the b-propeller tunnel
`
`HN
`
`O
`
`P1'
`
`R3
`
`NH
`
`O
`
`P1
`
`R2
`
`X
`
`N
`
`O
`
`R1
`
`N
`H
`
`P2
`
`Fig. 2.3 Schematic of DPPIV substrate.
`
`Page 8 of 47
`
`

`
`P. E. WIEDEMAN
`
`71
`
`[54]. More recent studies have determined that the side channel more likely
`acts as both entrance and exit, with an expansion of the channel to allow
`both ready access and egress [62]. Conformational change was also noted
`for Tyr547 (Tyr548 rat) in the active site that accommodated various in-
`hibitor types [55, 62]. The flexibility of this tyrosine is specific to DPPIV and
`is likely a selectivity determinant over the similar enzymes DPP8 and
`DPP9. Further selectivity may be achieved with polar residues in the P10
`position [63].
`The firm footing on structural understanding of DPPIV has allowed vir-
`tual docking exercises to suggest novel inhibitors [64, 65]. Subsequent eval-
`uation in in vitro assays established which chemotypes were suitable starting
`points for future structure-based design efforts.
`
`MEDICINAL CHEMISTRY AND PRECLINICAL STUDIES
`
`Inhibition DPPIV has proven to be a drug target that has successfully
`brought together all the drug discovery disciplines. Many institutions and
`companies have contributed to the wealth of information both in the pub-
`lished and patent literature. A quick search of the patent literature will find
`in excess of 300 patents filed by more than 60 companies and institutions.
`That body of information is too extensive to be included in this chapter. The
`study of DPPIV inhibition has been quite active over the past decade and
`the medicinal chemistry effort has been reviewed at time points throughout
`[66–71]. Here we will concentrate on journal publications, some meeting
`abstracts and the occasional patent application to illustrate the drug dis-
`covery story of DPPIV inhibition. A brief section will describe some of the
`early ground breaking studies that set the stage for the explosive research
`development in this field. Structural class will loosely collect the more con-
`temporary medicinal chemistry and preclinical information.
`
`EARLY DPPIV INHIBITORS
`
`Inhibitors of DPPIV have long been sought as tools to elucidate the func-
`tional significance of the enzyme. The first inhibitors were characterized in
`the late 1980s and 1990s. Each was important in establishing an early
`structure–activity relationship (SAR) for subsequent investigation. It should
`be noted that the inhibitors fall into two main classes, those that interact
`covalently with DPPIV and those that do not.
`Since DPPIV is a protease, it is not unexpected that inhibitors would
`likely have a peptidic nature and this theme has carried through to
`
`Page 9 of 47
`
`

`
`72
`
`DPPIV INHIBITION
`
`contemporary research. A number of non-covalent inhibitors were charac-
`terized in the same paper [72]. The tripeptides diprotin A (Ile-Pro-Ile (6))
`and diprotin B (Val-Pro-Leu (7)) were identified as modest inhibitors with
`IC50s of 8 and 920 mM, respectively. Although these simple peptides were
`not anticipated to be metabolically stable, the recognition element of proline
`in the penultimate position was established. Lys[Z-4-NO2]-pyrrolidide (5)
`indicated that a simple, rather flat heterocycle was sufficient to occupy the
`P1 site where DPPIV favours proline and alanine. The IC50 was again
`modest at 2 mM, and the compound showed significant toxicity. Two other
`compounds with five-membered heterocycles mimicking proline in the P1
`pocket were Val-pyrrolidide (9) and Ile-thiazolidide ((10), P32/98) with
`IC50s of 6.0 and 2.8 mM, respectively. These compounds would become two
`of the more extensively studied inhibitors.
`A natural product isolate, TMC-2 (11), from Aspergillus orzae A374 was
`identified as a specific inhibitor (7.7 mM) of DPPIV [73]. A solid-phase
`combinatorial chemistry effort reduced the size and determined the critical
`core structure, TSL-225 (12), that maintained equivalent potency (5.7 mM)
`(Fig. 2.4).
`
`O
`
`OH
`
`iPr
`
`HN
`
`O
`
`O
`
`N
`
`H2N
`
`iPr
`
`O
`
`H2N
`
`N
`
`S
`
`Et
`
`Me
`
`P32/98 (10)
`
`diprotin B (7)
`
`O
`
`iPr
`
`N
`
`(9)
`
`H2N
`
`H
`N
`
`H2N
`
`N
`
`O
`
`O
`
`OH
`
`OH
`
`O
`
`HN
`
`Et
`
`Me
`
`O
`
`N
`
`H2N
`
`Et
`
`Me
`
`O
`
`diprotin A (6)
`
`O
`
`N
`
`NH2
`
`HN
`
`O
`
`O
`
`(5)
`
`OMe
`
`HO
`
`OH
`
`H
`N
`
`OH
`
`OH
`
`OH
`
`O
`
`NH
`
`H2N
`
`N
`
`O
`
`O
`
`O2N
`
`TMC-2 (11)
`
`TSL-225 (12)
`
`Fig. 2.4 Peptidic and non-covalent inhibitors.
`
`Page 10 of 47
`
`

`
`P. E. WIEDEMAN
`
`73
`
`Inhibitors that interacted covalently with DPPIV were also investigated at
`an early stage. N-Peptidyl-O-aroyl hydroxylamines exemplified by (13) were
`weak, irreversible inhibitors found to suffer from hydrolytic instability [74].
`A series of dipeptide phosphonates were likewise irreversible inhibitors [75].
`Although, the potency was weak, Ki ¼ 236 mM for (14), the compounds
`addressed the issue faced by many transition-state inhibitors, the intramo-
`lecular cyclization of the N-terminus to the electrophilic moiety requisite for
`the active site serine. Phosphonate esters react poorly with nitrogen
`nucleophiles and the members of this series were stable in pH 7.8 buffer
`for hours to days. Selectivity was observed for DPPIV over the proteases
`and esterases chymotrypsin, trypsin, human leukocyte elastase, porcine
`pancreatic elastase, acetylcholinesterase, papain and cathepsin B.
`Applying work established previously with inhibitors of another class of
`post-proline-cleaving serine proteases, the IgA proteases, dipeptides incor-
`porating the a-amino boronic acid analogue boroPro were the first truly
`potent inhibitors with Ki values in the low nanomolar range [76]. The SAR
`established in this work confirmed the requirement of a basic primary or
`secondary amine at the N-terminus. The boronic acid electrophile exhibited
`slow binding kinetics characteristic of many transition state DPPIV inhib-
`itors. Unfortunately, the compounds were extremely unstable to intramo-
`lecular cyclization in a neutral aqueous environment with Pro-boroPro (15)
`having a t1/2 of 1.5 h. There has been some recent work extending the
`boroPro motif into a P2 N-alkyl-glycine that becomes a common group to
`be described below [77].
`A real breakthrough occurred with the discovery of 2-cyanopyrrolidines
`as the first potent, selective and chemically stable inhibitors [78–80]. These
`competitive inhibitors were presumed to react with the active site serine,
`thus
`forming an imidate adduct. 2-Cyanopyrrolidine
`(16) had a
`Ki ¼ 200 nM while showing weak inhibition against the related enzymes
`DPPII (110 mM) and prolyl oligopeptidase (22% inhibition at 400 mM).
`Little loss of activity was observed after incubation of solutions at 37 1C for
`20 h. The cyanopyrrolidine was to become the sought serine hook that pro-
`vided sufficient selectivity and solution stability for drug development.
`
`Ferring
`
`There were two papers by this Research Institute in 1996 that firmly es-
`tablished the 2S-cyanopyrrolidide as a potent pharmacophore for DPPIV
`inhibition. These studies were directed at finding modulators of T-cell
`activity, a reflection of most DPPIV research at that time. The first paper
`established the superiority of the cycloalkyl side chain in the L-configuration
`in the P2 position delivering both potency and stability, as exemplified by
`
`Page 11 of 47
`
`

`
`74
`
`DPPIV INHIBITION
`
`O
`
`H
`N
`
`O
`
`O
`
`O
`
`N
`
`NH2
`
`Cl
`
`O
`P
`
`O
`
`H2N
`
`O
`
`O
`
`N
`
`Me
`
`NO2
`
`Cl
`
`(14)
`
`(13)
`
`HN
`
`O
`
`Ph
`
`O
`
`O
`
`CN
`
`O
`
`CN
`
`O
`
`CN
`
`H2N
`
`N
`
`(17)
`
`H2N
`
`N
`
`Et
`
`Me
`
`S
`
`(18)
`
`N
`
`Me
`
`(16)
`
`N
`
`H2N
`
`O
`
`HO
`
`B OH
`
`NH
`
`Pro-boroPro (15)
`
`Fig. 2.5 Covalent inhibitors.
`
`(17) [79]. The second provided an exploration of the SAR associated with
`the P1 position. Various five- and six-membered saturated heterocycles
`containing a combination of nitrogen, sulfur and oxygen were prepared,
`each having a nitrile at the 2-position [80]. The most potent compound was
`the thiazolidide (18).
`To reinforce a key difference, the DPPIV inhibitors discussed herein are
`divisible into two classes, those that form a covalent interaction with the
`enzyme and those that do not. This distinction will persist throughout the
`remaining part of this chapter (Fig. 2.5).
`
`COVALENT DPPIV INHIBITORS: P2 SECONDARY AMINE ANALOGUES
`
`Novartis
`
`These investigators were aware of the potency and selectivity achieved with
`the dipeptide mimetic cyanopyrrolidides having a primary a-amino acid in
`the P2 binding site described above. They also knew that N-methylglycine
`was accepted by DPPIV in the P2 site [81]. Combining these two features
`established the N-alkylated-glycylcyanopyrrolidides as a novel DPPIV in-
`hibitor chemotype and established DPPIV as a drug target for large pharma
`[82]. Libraries of 2-(S)-pyrrolidinecarbonitriles were produced efficiently
`with either five-step solid-phase or three-step solution-phase sequences.
`When the P2 glycine was N-alkylated with 2-aminoethyl-5-nitropyridine, a
`potent (IC50 ¼ 8 nM) inhibitor (19) in this new chemotype was achieved.
`SAR exploration led to NVP-DPP728 (20) (IC50 ¼ 22 nM). It proved to
`
`Page 12 of 47
`
`

`
`P. E. WIEDEMAN
`
`75
`
`be chemically stable against intramolecular cyclization and selective against
`other closely related peptidases known at the time. Oral administration of
`(20) (1 mmol/kg) produced a successful result in an oral glucose-tolerance
`test (OGTT) in cynomolgus monkey. DPPIV activity was inhibited by
`89% while both peak plasma glucose levels and glucose area under the curve
`(AUC) were reduced. Oral bioavailability was >70% in both rat and mon-
`key. Steady-state volumes of distribution suggested that the drug was
`primarily distributed to body fluids. Clearance was moderate, and the half-
`life of 0.85 h in both monkey and man suggested that (20) might be best
`dosed in conjunction with a meal.
`Early published results had reported several key SAR and biochemical
`features of NVP-DPP728 (20) [83]. The D-antipode (Ki ¼ 5.6 mM), des-
`cyano (Ki ¼ 15.6 mM) and amide (Ki ¼ 320 mM) analogues of NVP-DPP728
`(Ki ¼ 0.11 mM) were compared. There was 3.9 kcal/mol of binding energy
`associated with the cyano group in the L-configuration. The inhibition
`kinetics was consistent with a two-step mechanism where an initial loose
`inhibitor–enzyme complex slowly isomerizes to a tight complex. SAR
`features included an understanding of hydrophobic or van der Waals in-
`teractions of the pyrrolidine in the P1 pocket. H-bonding and ionic inter-
`actions stabilized the P2 carbonyl and amine groups, respectively. P2 site
`side-chain interactions were hydrophobic in the S2 pocket. The nitrile was
`either considered to form a transient imidate with catalytic site serine or
`form some type of dipole–hydrogen bond interaction, either explained the
`binding energy.
`Additional SAR of NVP-DPP728 and the path to LAF237 (21) was
`combined in one paper [84]. Within NVP-DPP728, the P2 glycine was
`greatly preferred over the b-alanine analogue. The diamine chain length of
`two carbons was optimal. In addition to the nitro group, chlorine and nitrile
`were accepted on the terminal pyridine. Replacement of the 2-aminopyri-
`dine with pyrimidine, aniline, or phenyl was deleterious to binding potency.
`Based on selectivity considerations, NVP-DPP728 became the first devel-
`opment candidate.
`SAR continued with larger aliphatic side chains and rings in the P2 site,
`with good potency maintained with a fully substituted carbon adjacent to
`the P2 amine. This progressed to the investigation of multiple cyclic systems
`and culminated with the discovery that adamantyl groups furnished
`low nanomolar inhibitors. Studies on metabolites indicated that hydro-
`xylation of the adamantyl groups was well tolerated. This led to the syn-
`thesis of the 3-hydroxylated-1-aminoadamantane analogue (21) (LAF237,
`IC50 ¼ 3.5 nM), which avoided the introduction of a chiral centre. LAF237
`had the greatest solution stability to intramolecular cyclization of any tran-
`sition-state mimetic synthesized at that time. Further functionalization of
`
`Page 13 of 47
`
`

`
`76
`
`DPPIV INHIBITION
`
`Table 2.1 PEPTIDASE SELECTIVITY OF P2 PRIMARY AND SECONDARY AMINE
`INHIBITORS
`
`Compound
`
`DPPIV IC50 (nM)
`
`DPP8 IC50 (nM)
`
`DPPII IC50 (nM)
`
`68,985
`39,873
`26,520
`>100,000
`
`O
`
`CN
`
`N
`
`HN
`
`NH
`
`N
`
`NVP-DPP728 (20)
`DPPIV IC50 = 22 nM
`
`MeO
`
`Me
`
`N
`
`O
`
`CN
`
`NH
`
`N
`
`(22)
`DPPIV IC50 = 49 nM
`DPP8 IC50 >100000 nM
`
`2,283
`27
`4,573
`14,219
`
`NC
`
`P32/98 (10)
`(17)
`NVP-DPP728 (20)
`LAF237 (21)
`
`1,660
`12
`53
`51
`
`O
`
`CN
`
`N
`
`HN
`
`NH
`
`N
`
`O2N
`
`(19)
`DPPIV IC50 = 8 nM
`
`O
`
`CN
`
`N
`
`HN
`
`LAF237, vildagliptin (21)
`DPPIV IC50 = 3.5 nM
`
`HO
`
`Fig. 2.6 N-alkylated glycyl-2-cyanopyrrolidides.
`
`the adamantyl hydroxyl group decreased potency. LAF237 was selective
`over similar peptidases screened at the time and a recent report stated
`LAF237 to be a 9 mM (IC50) inhibitor of DPP8 [85]. [Note: The IC50 value
`reported here for DPPIV and DPP8 inhibition differs from that reported in
`Table 2.1. as a reflection of variance in procedures between laboratories.
`The constant, Ki, is preferred for comparison of data when available.] The
`pharmacokinetics of LAF237 was similar to NVP-DPP728, except the half-
`life was 2.6-fold longer. Using obese Zuckerfa/fa rats as a model of type 2
`diabetes in OGTT, and administration of LAF237 (10 mmol/kg) resulted in
`90% DPPIV inhibition over the 90-min study, a 60% higher level of GLP-1,
`a decrease in glucose excursion and a doubling of peak insulin levels. Com-
`parable studies in cynomologus monkey dosed with LAF237 (1 mmol/kg)
`showed >50% DPPIV inhibition for at least 10 h compared to 4–5 h for
`NVP-DPP728. LAF237 was brought forward as a development candidate
`suitable for once daily dosing (Fig. 2.6).
`
`Page 14 of 47
`
`

`
`P. E. WIEDEMAN
`
`77
`
`National Health Research Institutes, Taiwan
`
`These researchers used a selectivity comparison of known DPPIV inhibitors
`as their starting premise. The key comparison was between DPPIV and
`DPP8, as other proteases were not inhibited to a significant degree. It was
`found that P2 primary amino analogues (P32/98 (10) and (17)) were not
`selective over DPP8 while two P2 N-substituted glycine analogues were
`(Table 2.1) [86]. Based on these results, several