`
`Current Opinion in Drug Discovery & Development 2008 11(4):512-532
`© The Thomson Corporation ISSN 1367-6733
`
`From the bench to the bedside: Dipeptidyl peptidase IV inhibitors, a
`new class of oral antihyperglycemic agents
`Zhonghua Pei
`Address
`Department of Discovery Chemistry, Small Molecular Drug Discovery, Genentech Inc,
`1 DNA Way, South San Francisco, CA 94080, USA
`Email: pei.zhonghua@gene.com
`
`New therapeutic agents are needed to combat the ever-increasing prevalence of diabetes. The two incretins glucagon-like peptide-1 (7-36)
`(GLP-1(7-36)) amide and glucose-dependent insulinotropic peptide (GIP) are released from the small intestine in response to the ingestion of
`nutrients and regulate glucose homeostasis in a glucose-dependent fashion; however, the action of both incretins is terminated by the rapid
`N-terminal cleavage of two amino acid residues of GLP-1 and GIP by dipeptidyl peptidase-IV (DPP-IV). The preservation of active GLP-1 and
`GIP by inhibiting DPP-IV activity is an attractive strategy for the treatment of diabetes in patients who exhibit a reduced incretin response.
`This strategy has resulted in the launch of two DPP-IV inhibitor drugs; sitagliptin in North America, several European territories, and various
`other countries, and vildagliptin in the EU as well as various countries. This article provides an overview of the recent advances in and the lessons
`learned from the design of potent and selective small-molecule inhibitors of DPP-IV for the treatment of type 2 diabetes.
`
`Keywords Dipeptidyl peptidase IV, DPP-IV, incretin effect, glucagon-like peptide-1, GLP-1, glucose-dependent insulinotropic peptide,
`GIP, diabetes, vildagliptin, sitagliptin
`
`Abbreviations
`CYP cytochrome P450, DPP-IV dipeptidyl peptidase-
`IV, FPG fasting plasma glucose, GIP glucose-dependent
`insulinotropic peptide, GLP-1 glucagon-like peptide-1,
`HbA1c glycosylated hemoglobin A1c, HTS high-throughput
`screening, PK pharmacokinetic, PPCE post-proline cleaving
`enzyme
`
`Introduction
`Diabetes
`Diabetes is a progressive disease characterized by a lack of
`production or improper use of insulin by the body, resulting
`in increased blood glucose levels. According to the American
`Diabetes Association, a person is diagnosed diabetic if any
`of the following criteria apply: (i) fasting plasma glucose
`(FPG) is ≥ 126 mg/dl; (ii) diabetes symptoms exist and casual
`plasma glucose is ≥ 200 mg/dl; or (iii) plasma glucose is
`≥ 200 mg/dl in an oral glucose tolerance test [1].
`
`The prevalence of diabetes, particularly type 2 diabetes, is
`rapidly increasing and affects hundreds of millions of people
`worldwide. For example, 20.8 million children and adults in
`the US (7% of the population) currently have diabetes. This
`increase in prevalence of diabetes has spread to countries
`such as China and India – two countries with large populations
`that used to boast low incidence rates of the disease [2]. The
`social and economic burden associated with diabetes and its
`complications are very significant [1].
`
`The primary goal for diabetes intervention is to control blood
`glucose levels in order to minimize diabetic complications
`such as coronary disease, heart attack, stroke, heart
`
`failure, kidney failure, blindness, erectile dysfunction,
`neuropathy, gangrene and gastroparesis. Glycemic control
`is monitored by either blood glucose levels, fructosamine
`levels or glycosylated hemoglobin A1c (HbA1c) levels. While
`blood glucose levels change constantly (within a certain
`range) to meet the needs of the body (eg, blood glucose
`levels surge after each meal), the HbA1c level reflects the
`average blood glucose level during the 120-day life span of
`red blood cells: the higher the blood glucose levels over this
`period, the higher the HbA1c levels. The American Diabetes
`Association recommends that the HbA1c levels of diabetics
`are maintained below 7%, and it has been demonstrated
`that even a small reduction in HbA1c level translates into
`a significantly better health outcome [3]. Fructosamine
`(glycated albumin) levels reflect the average glucose levels
`over the past 1 to 3 weeks and are usually measured to
`determine whether a particular therapy is effective. The
`relationship between the three parameters of glycemic
`control allows inter-conversion between them to be made
`[4]. Postprandial blood glucose (PPG) level is another
`useful parameter for the diagnosis and management of
`diabetes, because elevated PPG levels often develop early in
`the disease even before elevated FPG is noted.
`
`Several possible medical interventions for type 2 diabetes
`exist beside change
`in
`lifestyle, which
`is the only
`intervention with no side effects, but has very low patient
`compliance. While insulin, repaglinide and exenatide all
`share the limitation of administration by injection, oral
`antihyperglycemic agents are limited in terms of efficacy
`and/or side effects (Table 1) [5,6]. For example, episodes
`of hypoglycemia that are associated with sulfonylurea
`treatment are very dangerous and can be life threatening.
`
`Mylan EX 1015, Page 1
`
`
`
`Dipeptidyl peptidase IV inhibitors Pei 513513
`
`The body weight gain that is associated with chronic treatment
`with sulfonylurea and thiazolidinedione agents can also result
`in further metabolic stress on patients in the long term.
`Therefore, new therapeutic agents with improved efficacy
`and/or an improved tolerability/safety profile are needed to
`combat the ever-increasing prevalence of diabetes.
`
`GLP-1 and GIP actions
`A greater stimulation of insulin is provoked following an oral
`administration of glucose compared with the same amount
`of glucose infused intravenously, a phenomenon known as
`the incretin effect [7•,8,9]. The contribution of incretins
`to glucose-stimulated insulin increase is reduced to 8% in
`patients with type 2 diabetes, compared with 58% in healthy
`patients [10].
`
`The two major incretins are glucagon-like peptide-1 (GLP-1)
`and glucose-dependent insulinotropic peptide (GIP).
`GLP-1(7-36) amide (the active form) is processed from
`proglucagon and released from enteroendocrine L-cells
`in the distal small intestine and colon in response to the
`oral ingestion of nutrients. GLP-1(7-36) amide has multiple
`biological effects that contribute to glucose homeostasis
`and promote the normalization of blood glucose levels, as
`demonstrated by the following observations: the binding
`of GLP-1(7-36) amide to its G-protein-coupled receptor
`on pancreatic β-cells increased glucose-stimulated insulin
`secretion [11•]; GLP-1(7-36) amide stimulated insulin gene
`expression [12] and inhibited glucagon secretion from islet
`cells [13]; GLP-1 slowed gastric emptying, thereby reducing
`the rate that nutrients were absorbed into the circulation [14];
`GLP-1 receptor mRNA is widely expressed in hypothalamic
`nuclei that are responsible for modulating feeding behavior
`[15,16]; and the peripheral administration of GLP-1 promotes
`satiety and inhibits food intake in man [17,18].
`
`GIP is a 42-amino acid peptide that is secreted by endocrine
`K-cells of the duodenum in response to the ingestion of
`nutrients [19]. The physiological actions of GIP include
`glucose-dependent potentiation of insulin secretion and
`regulation of insulin gene transcription. In contrast to GLP-1,
`GIP does not inhibit glucagon secretion or influence gastric
`emptying in humans [20•,21].
`
`DPP-IV activity
`Both GLP-1(7-36) amide and GIP(1-42) are rapidly inactivated
`by dipeptidyl peptidase-IV (DPP-IV, EC 3.4.14.5; also known
`as lymphocyte cell surface protein CD26 or adenosine
`deaminase [ADA]-binding protein) via the cleavage of
`two amino acid residues from the N-terminus [22,23,24•].
`DPP-IV was first
`identified by Hopsu-Havu and
`Glenner in 1966 as an enzyme that possesses glycyl-
`proline-β-naphthylamidase activity [25•], and it was later
`demonstrated that DPP-IV cleaves prolyl and alanyl peptide
`bonds at the penultimate position from the N-terminus
`[26-28]. DPP-IV is a serine protease with the catalytic
`triad Ser630-Asp708-His740 (all residue numberings are from
`human DPP-IV) oriented in a non-classical amino acid
`sequence order, and with significant sequence similarity to
`other α,β-hydroxylases (eg, prolyl oligopeptidase [POP]).
`DPP-IV is expressed as a glycoprotein on the surface of cells
`of most tissues, including kidney, liver, intestine, placenta,
`prostate, skin, lymphocytes and endothelial cells. The
`proteolytic cleavage of DPP-IV from cell surfaces results in
`a soluble circulating form of DPP-IV with a monomeric mass
`of approximately 100 kDa. DPP-IV is catalytically active
`as a dimer. The X-ray crystal structures of human and rat
`DPP-IV have been solved [29,30]. The human DPP-IV
`monomer consists of two domains comprised of an N-terminal
`eight-blade β-propeller and a C-terminal catalytic domain
`that adopts an α,β-hydrolase fold. Similar to POP, the active
`site is positioned inside a large solvent-filled cavity that
`
`Table 1. Currently available antihyperglycemic interventions and their limitations.
`Route of
`Intervention
`Hypoglycemia
`Edema
`Expected HbA1c
`administration
`reduction (%)
`1.0 to 2.0
`
`−
`
`−
`
`BW
`gain
`−
`
`GI side
`effects
`−
`
`Lactic
`acidosis
`−
`
`Liver
`monitoring
`−
`
`Injection
`
`1.5 to 2.5
`
`(diet
`Lifestyle
`and exercise)
`Insulina
`
`Repaglinide
`
`1.0 to 1.5
`
`Exenatide
`
`Oral
`
`AGIs
`
`Biguanide
`(metformin)
`Sulfonylureas
`
`0.5 to 1.0
`
`0.5 to 0.8
`
`1.0 to 1.5
`
`1.5
`
`TZDs
`
`0.5 to 1.4
`
`√
`
`√
`
`−
`
`−
`
`−
`
`√
`
`−
`
`√
`
`−
`
`−
`
`−
`
`−
`
`—
`
`√
`
`√
`
`√
`
`Loss
`
`−
`
`−
`
`√
`
`√
`
`−
`
`−
`
`√
`
`√
`
`√
`
`−
`
`−
`
`−
`
`−
`
`−
`
`−
`
`√
`
`−
`
`−
`
`−
`
`−
`
`−
`
`−
`
`−
`
`−
`
`√
`
`aInhalable insulin was briefly introduced to the market, but was removed shortly afterwards.
`AGI α-glucosidase inhibitor, BW body weight, GI gastrointestinal, TZD thiazolidinedione.
`
`Mylan EX 1015, Page 2
`
`
`
`514 Current Opinion in Drug Discovery & Development 2008 Vol 11 No 4
`
`is surrounded by the β-propeller domain. The β-propeller
`domain forms two channels for substrates to access the
`active site: a central β-propeller pore with a diameter of
`approximately 13 Å and a larger pore with a diameter of
`approximately 20 Å located on the side. Peptides that are
`larger than these pores cannot enter the catalytic site of
`DPP-IV.
`
`In addition to cleaving GLP-1 and GIP, DPP-IV may play a
`role in the cleavage of substrates with the preferred amino-
`terminal Xaa-Pro- or Xaa-Ala- dipeptide sequences (where
`Xaa is any amino acid residue), resulting in the inactivation
`or alteration of the biological activities of such substrates.
`One such substrate is GLP-2, which is a 33-amino acid
`peptide that is co-secreted with GLP-1 from the intestines.
`The administration of GLP-2 in humans increases glucagon
`secretion, which may counteract the glucagonostatic effect
`of GLP-1 [31]. Other potential DPP-IV substrates include
`growth hormone-releasing hormone (GHRH), substance
`P, bradykinin, gastrin-releasing peptide, neuropeptide Y
`(NPY), peptide YY (PYY) and certain chemokines such as
`RANTES (Regulated on Activation, Normal T-cell Expressed
`and Secreted), stromal cell-derived factor-1, eotaxin and
`macrophage-derived chemokine, all of which may modulate
`immune function [32]. Unlike in the case of GLP-1 and
`GIP, little is known about whether the inhibition of DPP-IV
`activity will increase the endogenous circulating levels of
`the above mentioned intact peptides, and whether increased
`levels of the intact peptides will alter the various associated
`downstream pathways [33].
`
`Two different approaches to augmenting the effects of
`GLP-1 and GIP have emerged: the administration of GLP-1
`and GIP analogs that are resistant to DPP-IV degradation,
`and the inhibition of DPP-IV activities. Each approach has
`advantages and limitations [34,35]. DPP-IV inhibition has
`been demonstrated to be a viable approach to the treatment
`of diabetes through years of intensive research, including
`GLP-1 infusion studies [36], DPP-IV knockout studies [37•],
`small-molecule inhibitor studies (vide infra) and, most
`recently, the human clinical trials (vide infra) that validate
`DPP-IV as a novel target in the management of diabetes.
`
`The buzz: DPP-IV inhibitors potentially modify
`disease
`Studies in Zucker diabetic rats suggested that chronic
`exposure to GLP-1 may increase β-cell mass by promoting
`growth and differentiation and by inhibiting apoptosis
`(decreased functional β-cell mass is linked with type 1 and
`2 diabetes) [38]. Chronic treatment with DPP-IV inhibitors
`preserved islet function in diabetic mice [39] and improved
`β-cell survival and islet cell neogenesis in streptozotocin-
`induced diabetic rats [40]. In a mouse model with impaired
`insulin sensitivity and secretion arising from the combination
`of a high-fat diet and a streptozotocin-induced β-cell mass
`reduction, treatment with a des-fluoro sitagliptin analog for
`3 months increased the number of insulin-positive β-cells.
`This resulted in the normalization of both β-cell mass and
`the ratio of β-cells to α-cells, and the restoration of normal
`islet architecture [41]. Treatment of neonatal rats with the
`
`DPP-IV inhibitor vildagliptin for 3 weeks increased β-cell
`replication and inhibited apoptosis, leading to increased
`β-cell mass [42]. Furthermore, the GLP-1 mimetic exendin-4
`induced pancreatic β-cell proliferation and islet neogenesis
`[43,44]. All these data suggest that DPP-IV inhibitors could
`potentially delay or prevent the loss of functional β-cell
`mass and therefore slow the progression of the disease.
`If this effect is confirmed in humans, DPP-IV inhibitors will
`represent a revolutionary approach to combating diabetes.
`
`DPP-IV inhibitors
`The medicinal chemistry of DPP-IV inhibitors has been
`[45-47,48•,49•],
`reviewed extensively
`therefore, an
`exhaustive review is not intended here. This section provides
`a brief overview of the most important small-molecule
`DPP-IV inhibitors, with emphasis on the lessons that can be
`learned from this intensive endeavor by both academia and
`industry. Selected recent developments are also included.
`
`Profiles of 'optimal' inhibitors
`Selectivity
`The side effects observed after the administration of the
`reversible DPP-IV inhibitor P32/98 to various species led
`scientists at Merck & Co Inc to suspect that the inhibition
`of other dipeptidyl peptidases might be occurring. When
`a selective DPP-VIII/IX dual inhibitor (IC50 for DPP-IV,
`DPP-VIII and DPP-IX = 30,000, 63 and 68 nM, respectively)
`was identified and administered to rats for 2 weeks,
`significant toxic effects such as alopecia, thrombocytopenia,
`anemia, enlarged spleen, multiple histological pathologies
`and death were observed [50••]. In an acute study in dogs,
`the DPP-VIII/IX inhibitor caused bloody diarrhea, emesis
`and tenesmus. The same toxicity was observed in both
`animal models when a different nonselective inhibitor was
`administered. In contrast, no adverse effects were observed
`with a selective DPP-IV inhibitor (IC50 for DPP-IV, DPP-VIII
`and DPP-IX = 27, 69,000 and > 100,000 nM, respectively).
`Therefore, it appears that the inhibition of DPP-VIII/IX at
`pharmacologically relevant drug levels should be avoided in
`order to minimize off-target adverse effects.
`
`Duration of inhibition
`Because DPP-IV
`is distributed ubiquitously and may
`also cleave a number of substrates that are important in
`regulating other body functions (vide supra), an early
`hypothesis by researchers was that inhibitors with short
`half-lives may be preferred in order to minimize potential
`side effects. Such duration of DPP-IV inhibition should be
`long enough to reduce postprandial glucose excursion,
`yet short enough to allow enough time for the activities
`of other endogenous substrates of DPP-IV to restore to
`normal levels. However, the results from one clinical trial
`showed the opposite trend. In this trial diabetic patients
`(HbA1c = 8.8 to 10.1%) who were poorly controlled
`on sulfonylurea treatment were
`infused with active
`GLP-1 either for 16 or 24 h per day for 7 days [51•]. The
`outcome for patients undergoing 24-h infusion (eg, as
`measured by FPG) was better than that for the patients
`undergoing 16-h infusion, indicating the preference for
`long-acting DPP-IV inhibition in order to achieve maximal
`efficacy.
`
`Mylan EX 1015, Page 3
`
`
`
`Dipeptidyl peptidase IV inhibitors Pei 515515
`
`developed into a marketed drug for diabetes because of
`its side effects, it was the first molecule to demonstrate
`efficacy in both animal and human studies, and the need
`for selectivity over DPP-VIII/IX (vide supra). The treatment
`of Wistar rats with P32/98 resulted in a dramatic increase
`of active GLP-1 (from 13.4 to 90%), as determined by using
`[125I]GLP-1(7-36) and the observation of increased glucose
`tolerance [53]. In a later study, orally administered P32/98
`in both lean and obese Zucker rats blocked the degradation
`of both GIP and GLP-1, inhibited DPP-IV activity (65%),
`increased insulin secretion (150% increase in obese animals
`and 27% increase in lean animals) and decreased plasma
`glucose in a glucose tolerance test [54]. In a clinical trial
`involving 24 diabetic patients previously on a controlled diet,
`or regular acarbose or glibenclamide, treatment with P32/98
`increased active GLP-1 by 5.8-fold and improved glucose
`tolerance by 22 to 33% [55].
`
`DPP-728
`The next most intensively studied DPP-IV inhibitor is
`DPP-728 (2; Figure 2), which was discovered by researchers
`at Novartis AG. DPP-728 incorporates 2-cyanopyrrolidine as
`the P1 surrogate, a functional group that was first used in
`prolyl endopeptidase inhibitors [201] and then used by other
`research groups in the design of DPP-IV inhibitors [56•].
`
`Substrate-like inhibitors
`inhibitors, either covalent
`The substrate-like DPP-IV
`or non-covalent, are dipeptidic in nature and have the
`generic structures A or B shown in Figure 1, where the
`P1
`substituents occupy
`the S1 pocket and
`the
`P2 substituents occupy the S2 pocket of DDP-IV.
`
`Figure 1. Generic structures of substrate-like DPP-IV inhibitors.
`
`P1
`
`O
`
`NH
`
`P2
`
`A
`
`NH2
`
`P1
`
`B
`
`P2
`
`O
`
`P32/98
`One of the earliest reported DPP-IV inhibitors, P32/98
`(1; Figure 2), which was licensed from Probiodrug AG to
`Merck, used thiazolidide as the P1 surrogate. P32/98 had
`modest inhibitory potency (Ki = 126 nM for DPP-IV isolated
`from pig kidney [52], IC50 = 420 nM against human DPP-IV
`[50••]) and modest selectivity over other peptidases such
`as amyloid precursor protein, fibroblast activation protein-α
`and prolidase. Although P32/98 was not successfully
`
`Figure 2. Structures of substrate-like DPP-IV inhibitors.
`
`CH3
`
`NH2
`
`CH3
`H
`
`N
`
`O
`
`R
`
`3 R = H, Ki = 410 nM
`4 R = CN, Ki = 2.2 nM
`
`H
`
`OH
`
`H
`
`H
`
`H
`
`C
`N
`
`N
`
`O
`
`NH
`
`NH
`
`N
`
`C
`
`N
`
`2 DPP-728
`
`IC50 = 22 nM
`
`H
`
`R
`
`CH3
`
`NH2
`
`CH3
`H
`
`N
`
`O
`
`S
`
`1 P32/98
`
`IC50 = 420 nM
`Ki = 126 nM
`
`H
`
`H
`
`C N
`
`N
`
`O
`
`NH2
`
`C N
`
`N
`
`O
`
`NH
`
`H
`
`C N
`
`N
`
`O
`
`NH
`
`H
`
`H
`
`5
`
`IC50 = 13 nM
`
`6 R = H, IC50 = 3 nM
`7 vildagliptin R = OH, IC50 = 3.5 nM
`
`8 saxagliptin
`(Bristol-Myers Squibb/AstraZeneca/
`Otsuka Pharmaceutical)
`
`Ki = 0.6 nM
`
`N
`
`C
`
`N
`
`C
`N
`
`O
`
`11
`
`NH
`
`OH
`
`IC50 = 104 nM
`
`CH
`
`C
`
`N
`
`O
`
`C
`N
`
`CH3
`
`NH
`
`F
`
`F
`
`N
`
`N
`
`OH
`
`O
`
`10 ABT-279
`(Abbott Laboratories)
`
`Ki = 1 nM
`
`C N
`
`N
`
`O
`
`NH2
`
`9 denagliptin
`
`Ki = 22 nM
`
`F
`
`Mylan EX 1015, Page 4
`
`
`
`516 Current Opinion in Drug Discovery & Development 2008 Vol 11 No 4
`
`Scheme 1. Intramolecular cyclization of DPP-IV inhibitors.
`
`H
`
`O
`
`R
`
`N N
`
`O
`
`H2O
`
`H
`
`NH
`
`R
`
`N N
`
`O
`
`Neutral or
`basic pH
`
`C N
`
`N
`
`O
`
`NH
`
`R
`
`The cyano group forms a covalent bond (albeit reversibly)
`with the hydroxyl group of the serine residue in the catalytic
`site of DPP-IV. The introduction of a 2-cyano group
`usually boosts the potency of the inhibitors significantly,
`as evidenced by the direct comparison of the des-
`cyano analog 3 (IC50 = 410 nM) and the cyano analog 4
`(IC50 = 2.2 nM) [56•]. This early observation led to the
`widespread use of 2-cyanopyrrolidine as the P1 'warhead'.
`Other less popular P1 warheads include 2-cyanoazetidine
`[57] and 2-boronopyrrolidine [58]. Inhibitors derived
`from 2-cyanoazetidine are less potent than those derived
`from 2-cyanopyrrolidine, and
`inhibitors derived
`from
`2-boronopyrrolidine are generally less selective against
`other proteases and not chemically stable, as a result of
`intramolecular cyclization between the boron and the free
`amine (vide infra).
`
`DPP-728 exhibited an IC50 value of 22 and 7 nM against
`DPP-IV extracted from Caco2 cells and human plasma,
`respectively, and displayed excellent selectivity over
`DPP-II and post-proline cleaving enzyme (PPCE) [59].
`DPP-728 is a slow-binding inhibitor of DPP-IV with a kon of
`1.3 x 105 M-1s-1 and a koff of 1.3 x 10-3 s-1 [60]. The corresponding
`des-cyano analog of DPP-728 had a much weaker potency
`(Ki = 15,600 nM) than DPP-728, again revealing the
`significance of the 2-cyano group on the inhibitory potency
`of this type of inhibitor.
`
`After an oral dose of 10 µmol/kg in rats, DPP-728 displayed
`a Cmax value of 3.65 µM, a short half-life (0.85 h), moderate
`clearance (28 ml/min/kg) and an oral bioavailability of
`74% [59]. In a 4-week study involving 93 patients with
`early-stage type 2 diabetes, HbA1c was reduced from
`7.4 to 6.9% after DPP-728 treatment (100 mg three times
`daily) [61]. Interestingly, the mean 24-h insulin was reduced
`by 26 pmol/l.
`
`As a free amino group is required to interact with the Glu205
`and Glu206 residues of DPP-IV, one potential drawback of
`2-cyanopyrrolidines is that the cyano group undergoes
`intramolecular cyclization with the free amino group to
`form the inactive amidine, and further reaction to form
`the diketopiperazine under neutral or basic pH conditions
`(Scheme 1). The rate of cyclization can be modulated by
`the size of the groups on or next to the nitrogen atom of
`the free amino group: sterically bulky groups slow down the
`intramolecular cyclization and thus increase the chemical
`stability of the inhibitors [59]. This cyclization problem is
`worse when the 2-cyano group is replaced by a boronic acid,
`which was used in the design of some of the early DPP-IV
`inhibitors.
`
`Bi- and tricyclic DPP-IV inhibitors and other
`structurally similar compounds
`After the discovery of DPP-728, efforts continued at
`Novartis to identify DPP-IV inhibitors with improved chemical
`stability and a longer half-life. The observation that steric
`bulk adjacent to the P2 amine is beneficial to both chemical
`stability and potency led to the incorporation of bi- and
`tricyclic
`rings
`into
`the design. For example,
`the
`2-adamantylamine analog 5 (IC50 = 13 nM; Figure 2) is less
`potent than the corresponding 1-adamantylamine analog
`6 (IC50 = 3 nM). Examination of the primary metabolites
`of adamantylamine analogs 5 and 6 suggested that
`monohydroxylation on the adamantyl ring might be well
`tolerated, hence
`the 3-hydroxyl-1-adamantyl analog
`vildagliptin (7, LAF-237; Figure 2) was synthesized. Vildagliptin
`had slightly better potency and, more importantly, much
`improved pharmacokinetic (PK) properties compared with
`DPP-728. Vildagliptin had an IC50 value of 3.5 and 2.7 nM
`against DPP-IV extracted from Caco2 cells and from human
`plasma, respectively, and displayed excellent selectivity over
`DPP-II (IC50 = 210 µM) and PPCE (IC50 > 500 µM) [62•]. As can
`be seen from the data in Table 2, vildagliptin had improved
`PK properties compared to DPP-728, displaying a lower
`Cmax and a longer half-life. This difference in PK properties
`translated well into the observed DPP-IV inhibition in
`cynomolgus monkey studies. While DPP-728 provided
`≥ 50% inhibition of DPP-IV activity for only 4 to 5 h post-dose
`(1 µmol/kg po), vildagliptin provided ≥ 50% inhibition for
`≥ 10 h at the same dose level [62•].
`
`Other structurally similar DPP-IV inhibitors have subsequently
`been discovered and entered into human clinical trials.
`These include saxagliptin (8, Bristol-Myers Squibb Co/
`AstraZeneca plc/Otsuka Pharmaceutical Co Ltd; Figure 2)
`[63], denagliptin (9; Figure 2) [202] and ABT-279 (10, Abbott
`Laboratories; Figure 2) [64]. A common structural theme of
`these inhibitors is the introduction of small substituents on
`
`Table 2. Comparison of the pharmacokinetic properties between
`DPP-278 and vildagliptin in cynomolgus monkeys.
`DPP-728
`Vildagliptin
`805
`293
`1.3
`1.5
`0.8
`0.7
`35
`90
`> 90
`> 90
`
`Cmax (nM)
`CL (l/kg/h)
`Vss (l/kg)
`t1/2 (min)
`F (%)
`
`CL clearance, Cmax maximum plasma concentration, F oral
`bioavailability, t1/2 half-life, Vss volume of distribution at steady state.
`
`Mylan EX 1015, Page 5
`
`
`
`Dipeptidyl peptidase IV inhibitors Pei 517517
`
`the 2-cyanopyrrolidine ring to improve/maintain potency, as
`well as to establish IP position. Large substituents on the
`2-cyanopyrrolidine ring are generally not tolerated because
`the S1 pocket is relatively small. Since vildagliptin, denagliptin
`and saxagliptin were discovered before the potential liability
`of DPP-VIII/IX inhibition was established, selectivity over
`DPP-VIII/IX was not optimized for these three compounds.
`ABT-279 (Ki = 1 nM), which has a proprietary acetylenyl group
`that confers excellent potency and selectivity, has been
`demonstrated to be efficacious and safe in animal studies.
`The X-ray crystal structure of ABT-279 bound to human
`DPP-IV confirmed that, similar to other 2-cyanopyrrolidine-
`based DPP-IV inhibitors, the 2-cyano group forms a covalent
`bond with the hydroxyl group of the Ser630 residue, and the
`amino group forms salt bridges with the Glu205 and Glu206
`residues (Figure 3A). The 5-acetylenyl group fits tightly in
`a narrow tunnel formed by the Phe357 and Tyr547 residues
`(Figure 3B). The pyridyl group stacks with the Arg125 residue
`and the carboxylate moiety forms a salt bridge with the
`His126 residue [65].
`
`The 2,5-dicyanopyrrolidine-based inhibitor 11 (IC50 = 104 nM)
`was reported by Pfizer Inc to increase active GLP-1 levels
`by approximately 2-fold in beagles [66], and des-cyano
`analogs of saxagliptin were recently reported to be potent
`DPP-IV inhibitors [67].
`
`DPP-IV inhibitors with conformationally restricted
`P2 groups
`Conformational restriction of the side chain of DPP-728
`resulted in the investigation of C(4)- and C(5)-substituted
`prolinyl-2-cyanopyrrolidines (pro-pro). Either an amino
`group (12; Figure 4) [68] or a carbonyl group (13; Figure
`4) [69] at the C(4) position of the P2 proline moiety resulted
`in highly potent inhibitors, with compound 12 exhibiting an
`IC50 value of 0.13 nM. Extension of the linker by one
`methylene unit led to analogs such as compounds 14
`(Ki = 5 nM; Figure 4) [70] and 15 (Ki = 1.7 nM; Figure 4) [71].
`Interestingly, the 2-cyanopyrrolidine moiety in compounds 14
`and 15 was replaced by a thiazolidine or a difluoropyrrolidine,
`
`form covalent
`respectively, both of which do not
`bonds with the serine residue of DPP-IV. Compounds 16
`(IC50 = 1.5 nM; Figure 4) [72] and 17 (IC50 = 14 nM;
`Figure 4) [73] are two other similar C(4)-substituted
`analogs that lack the electrophilic nitrile group. A slightly
`different conformational
`restriction strategy
`led
`to
`C(5)-substituted P2 pyrrolidines, such as compound 18
`(Ki = 3.1 nM; Figure 4) [74]. In all these pro-pro systems,
`the relative stereochemistry between C(4)/C(5) and C(2)
`is cis. All pro-pro inhibitors with a 2-cyano group have
`the potential issue of chemical instability, as discussed
`earlier. The C(5) pro-pro compounds have an advantage
`over
`the C(4) pro-pro
`compounds
`in
`that
`the
`substituents on C(5) are closer to the P2 proline nitrogen
`atom than the substituents on C(4), making the nitrogen
`atom more sterically hindered and therefore slowing the
`rate of intramolecular cyclization. Another way in which
`the side chain has been made more rigid is through
`the introduction of a cis-3-aminopyrrolidine ring as the
`P2 group, resulting in analogs such as compound 19
`(IC50 = 1.3 nM; Figure 4) [75].
`
`Substrate-like, non-covalent DPP-IV inhibitors
`After the discovery of sitagliptin (vide infra), research
`the α-amino amide series continued at Merck.
`on
`Replacement of the ethyl group in P32/98 (1; Figure 2) with an
`aromatic group and
`switching
`the oxidation-prone
`thiazolidide ring for a more metabolically stable fluorinated
`pyrrolidine ring led to the identification of amide 20
`(IC50 = 64 nM; Figure 5) [76]. Extension of the β-methyl
`group on compound 20 to a dimethylamido group improved
`the potency and decreased hERG channel binding.
`Optimization of the PK properties of the dimethylamido
`analog of compound 20 by introducing heterocycles led
`to compound 21 (Figure 5), which exhibited an IC50 value
`of 8.8 nM against DPP-IV, > 10,000-fold selectivity over
`other dipeptidyl peptidases and a good PK profile [77]. The
`X-ray crystal structure of compound 21 co-complexed with
`DPP-IV indicates that the dimethylamide carbonyl group of
`the inhibitor forms a hydrogen bond with the hydroxyl group
`
`Figure 3. X-ray crystal structure of the 5-alkynyl cyanopyrrolidine ABT-279 bound to human DPP-IV viewed from two different angles.
`A
`B
`
`Phe357
`
`Tyr547
`
`Glu206
`
`Glu205
`
`His126
`
`Arg125
`
`Residues Glu205 and Glu206 in panel A, Arg125 and His126 in panel B are highlighted as sticks.
`(Protein Data Bank DOI: 10.2210.pdb2i30/pdb).
`
`Mylan EX 1015, Page 6
`
`
`
`518 Current Opinion in Drug Discovery & Development 2008 Vol 11 No 4
`
`Figure 4. Structures of substrate-like DPP-IV inhibitors with conformationally restricted P2 groups.
`
`N
`
`S
`
`O
`
`NH
`
`14
`
`Ki = 5 nM
`
`NH
`
`O
`
`N
`
`S
`
`O
`
`N
`
`C N
`
`N
`
`O
`
`NH
`
`13
`
`IC50 = 1.7 nM
`
`C N
`
`4
`
`NH
`
`2
`
`N
`
`O
`
`NH
`
`C
`
`N
`
`Cl
`
`12
`
`IC50 = 0.13 nM
`
`F
`
`F
`
`F
`
`N
`
`N
`
`N
`
`F F
`
`N
`
`O
`
`NH
`
`S
`
`O
`
`O
`
`O
`
`NH
`
`S
`
`NH
`
`CH3
`
`N
`
`N
`
`S
`
`O
`
`NH
`
`16
`
`IC50 = 1.5 nM
`
`NH2
`
`O
`
`N
`
`CN
`
`N
`
`19
`
`IC50 = 1.3 nM
`
`CH3
`
`C
`
`N
`
`15
`
`Ki = 1.7 nM
`
`C N
`
`2
`
`N
`
`O
`
`NH
`
`5
`
`O
`
`CH3
`
`CH3
`
`18
`
`Ki = 3.1 nM
`
`OH
`
`O
`
`N
`
`S
`
`O
`
`NH
`
`O
`
`N
`
`CH3
`
`S
`
`N
`
`N
`
`N
`
`CH3
`
`17
`
`IC50 = 14 nM
`
`of the Tyr547 side chain. Compound 21 is an 'insurance back-
`up' to sitagliptin in case the latter falters in development.
`Saturation of the middle phenyl ring of compound 21 led to
`the potent and selective inhibitor 22 (IC50 = 4.8 nM), which
`demonstrated excellent PK properties in multiple species
`[78]. In these series exemplified by compounds 21 and
`22, the difluoropyrrolidine analogs generally display better
`PK profiles than the corresponding monofluoropyrrolidine
`analogs.
`
`Starting from the high-throughput screening (HTS) hit
`23 (IC50 = 84 nM; Figure 5), researchers at Eli Lilly & Co
`conducted an optimization program that resulted in the
`identification of the more potent compound 24 (IC50 = 5 nM;
`Figure 5). The thiophene group of compound 24 occupies
`the S1 pocket of DPP-IV. The molecules from this series
`adopt a U-shaped binding conformation and the residue
`Tyr547 undergoes a 70° side chain rotation to accommodate
`the inhibitor, thus allowing access to a previously unexposed
`area of the protein backbone for hydrogen bonding [79].
`
`Non-substrate-like inhibitors
`Non-substrate-like inhibitors are non-covalent and do not
`resemble the dipeptidic nature of DPP-IV substrates; instead,
`non-substrate-like inhibitors typically incorporate an aromatic
`
`ring (instead of proline mimetics, such as pyrrolidine,
`fluoropyrrolidine, 2-cyanopyrrolidine or thiazolidine) that
`occupies the S1 pocket of DPP-IV.
`
`DPP-IV inhibitors from Merck
`Two of the few hits identified by HTS at Merck were
`β-amidoacyl amide 25 and piperazine 26, which showed
`modest DPP-IV inhibitory potency (IC50 = 1.9 and 11 µM,
`respectively; both Figure 6). Amide 25 was originally
`synthesized for a thrombin inhibitor program and has an
`IC50 value of 52 nM against thrombin. Extensive optimization
`of the β-amidoacyl amide series led to the identification of
`extremely potent and selective compounds such as acid 27
`(IC50 = 0.48 nM; Figure 6); however, acid 27 demonstrated
`poor absorption and high clearance. A new lead was
`identified through the combination of the two HTS series,
`leading to inhibitor 28 (IC50 = 19 nM; Figure 6) [80]. Efforts
`to overcome the extensive metabolism of the piperazine
`ring led to the discovery of sitagliptin (29, MK-431;
`Figure 6) through the introduction of the key trifluoromethyl-
`substituted triazolopiperazine group [81,82•]. Sitagliptin
`had an IC50 value of 18 nM against DPP-IV and
`demonstrated remarkable selectivity over DPP-VIII and
`DPP-IX (IC50 > 45 µM for both). The X-ray crystal structure of
`sitagliptin bound to DPP-IV revealed that the trifluorophenyl
`
`Mylan EX 1015, Page 7
`
`
`
`Dipeptidyl peptidase IV inhibitors Pei 519519
`
`F
`
`F
`
`O
`
`O
`
`NC
`
`H3
`
`CH3
`
`H
`
`N
`
`H
`
`NH2
`
`22
`
`IC50 = 4.