Biochemistry 1999, 38, 11597-11603
`
`11597
`
`Articles
`
`NVP-DPP728
`(1-[[[2-[(5-Cyanopyridin-2-yl)amino]ethyl]amino]acetyl]-2-cyano-(S)-pyrrolidine), a
`Slow-Binding Inhibitor of Dipeptidyl Peptidase IV
`
`Thomas E. Hughes,* Manisha D. Mone, Mary E. Russell, Stephen C. Weldon, and Edwin B. Villhauer
`Metabolic and CardioVascular Diseases Research, NoVartis Institute for Biomedical Research, 556 Morris AVenue,
`Summit, New Jersey 07901-1398
`ReceiVed April 12, 1999; ReVised Manuscript ReceiVed July 2, 1999
`
`ABSTRACT: Inhibition of dipeptidyl peptidase IV (DPP-IV) has been proposed recently as a therapeutic
`approach to the treatment of type 2 diabetes. N-Substituted-glycyl-2-cyanopyrrolidide compounds, typified
`by NVP-DPP728 (1-[[[2-[(5-cyanopyridin-2-yl)amino]ethyl]amino]acetyl]-2-cyano-(S)-pyrrolidine), inhibit
`degradation of glucagon-like peptide-1 (GLP-1) and thereby potentiate insulin release in response to glucose-
`containing meals. In the present study NVP-DPP728 was found to inhibit human DPP-IV amidolytic
`activity with a Ki of 11 nM, a kon value of 1.3 (cid:2) 105 M-1 s-1, and a koff of 1.3 (cid:2) 10-3 s-1. Purified bovine
`kidney DPP-IV bound 1 mol/mol [14C]-NVP-DPP728 with high affinity (12 nM Kd). The dissociation
`constant, koff, was 1.0 (cid:2) 10-3 and 1.6 (cid:2) 10-3 s-1 in the presence of 0 and 200 (cid:237)M H-Gly-Pro-AMC,
`respectively (dissociation t1/2 (cid:24)10 min). Through kinetic evaluation of DPP-IV inhibition by the D-antipode,
`des-cyano, and amide analogues of NVP-DPP728, it was determined that the nitrile functionality at the
`2-pyrrolidine position is required, in the L-configuration, for maximal activity (Ki of 11 nM vs Ki values
`of 5.6 to >300 (cid:237)M for the other analogues tested). Surprisingly, it was found that the D-antipode, despite
`being (cid:24)500-fold less potent than NVP-DPP728, displayed identical dissociation kinetics (koff of 1.5 (cid:2)
`10-3 s-1). NVP-DPP728 inhibited DPP-IV in a manner consistent with a two-step inhibition mechanism.
`Taken together, these data suggest that NVP-DPP728 inhibits DPP-IV through formation of a novel,
`reversible, nitrile-dependent complex with transition state characteristics.
`
`Dipeptidyl peptidase IV (DPP-IV, EC 3.4.14.5)1 is a post-
`proline cleaving serine protease with significant sequence
`and structural similarity to other R-(cid:226)-hydrolases (e.g., prolyl
`oligopeptidase, acetylcholinesterase). DPP-IV is found through-
`out the body, both circulating in plasma and as a type II
`membrane protein produced by a variety of tissues, including
`kidney, liver, and intestine. DPP-IV may play a role in
`cleavage and inactivation of biologically active peptides with
`accessible amino-terminal Xaa-Pro- or Xaa-Ala- sequences
`(1, 2). Indeed, DPP-IV degrades and regulates the activity
`of several regulatory peptides in man (including the gut
`peptide “incretin” hormone glucagon-like peptide-1 (GLP-
`1), growth hormone-releasing hormone, and gastric inhibitory
`polypeptide). Due to the impressive antidiabetic actions of
`GLP-1, DPP-IV inhibition has been proposed as an intriguing
`new approach to the therapy of type 2 diabetes mellitus (3).
`Several classes of DPP-IV inhibitors bearing transition
`state mimics have been identified, and their kinetic properties
`have been extensively investigated. Peptidyl (R-aminoalkyl)-
`
`* Author for correspondence. Tel: 908-277-7336. Fax: 908-277-
`7728. E-mail:
`thomase.hughes@pharma.novartis.com.
`1 Abbreviations: NVP-DPP728, 1-[[[2-[(5-cyanopyridin-2-yl)amino]-
`ethyl]amino]acetyl]-2-cyano-(S)-pyrrolidine; DPP-IV, dipeptidyl pep-
`tidase IV; pNA, p-nitroaniline.
`
`phosphate diphenyl ester inhibitors of DPP-IV bind with low
`affinity (10-4 M IC50 values) but rapidly form highly stable
`covalent complexes with the active site serine residue of
`DPP-IV (4). Pro-boroPro and related analogues bind with
`high affinity (Ki (cid:24)10-11 M) in a reversible manner (half-
`life of enzyme-inhibitor complex (cid:24) 150 min; 5). The
`boronic acid analogues, however, are unstable in solution
`due to reversible intramolecular cyclization, and also inhibit
`dipeptidyl peptidase II, a related serine protease. Irreversible
`“suicide substrate” methylsulfonio cyclopeptide inhibitors
`have been described (6) as mechanistic tools but may not
`be suitable for therapeutic use. Although a series of (N-
`hydroxyacyl amide) aminodicarboxylic acid pyrrolidides
`have been described, only relatively unselective inhibitors
`with micromolar potency have been prepared (7). Confor-
`mationally constrained fluoroolefin-containing peptidyl-hy-
`droxylamine inhibitors also have been described (8), but
`isolation of enantiomerically pure compounds requires
`tedious separation of diastereomers. For these reasons, new
`chemical classes of selective and potent DPP-IV inhibitors
`are of interest and needed in order to evaluate the feasibility
`and efficacy of DPP-IV inhibition as a therapeutic approach.
`Recently, DPP-IV inhibitors with 2-cyanopyrrolidide P1
`substituents have been reported (9, 10). These compounds
`
`10.1021/bi990852f CCC: $18.00 © 1999 American Chemical Society
`Published on Web 08/18/1999
`
`SAXA-DEF-00399
`
`Page 1 of 7
`
`AstraZeneca Exhibit 2016
` Mylan v. AstraZeneca
` IPR2015-01340
`
`

`
`Hughes et al.
`
`11598 Biochemistry, Vol. 38, No. 36, 1999
`
`Chart 1. Structures of NVP-DPP728 and Analogs
`
`bind to DPP-IV several orders of magnitude more tightly
`than the corresponding pyrrolidide analogues (e.g., Ki values
`for isoleucine-2-cyanopyrrolidide and isoleucine pyrrolidide
`are 2 and 400 nM, respectively; 10, 11). Similar potency is
`observed with a new class of cyanopyrrolidide inhibitors,
`termed N-substituted-glycyl-2-cyanopyrrolidide compounds
`(12). Recently, NVP-DPP728 (see Chart 1), a novel deriva-
`tive of this class, has been identified as a potent and selective
`DPP-IV inhibitor for use in the treatment of diabetes mellitus
`(example no. 5 in ref 12, described also in ref 13). NVP-
`DPP728 inhibits human and rat plasma DPP-IV with IC50
`values in the range of 5-10 nM with >15 000-fold selectiv-
`ity relative to DPP-II and a range of proline-cleaving
`proteases (13). This compound shows promise as an antidia-
`betic agent due to its ability to preserve the integrity of
`GLP-1 (13) and improve glucose tolerance (14).
`While dipeptide-like pyrrolidide compounds (e.g., valine
`pyrrolidide) inhibit DPP-IV through simple reversible com-
`petitive binding (11),
`the kinetic properties of nitrile-
`containing inhibitors have not been rigorously evaluated. In
`a brief communication, Li and colleagues (9) reported that
`dipeptide cyanopyrrolidide compounds generate competitive
`or mixed inhibition profiles and postulated that the mecha-
`nism of inhibition involves formation of an imidate inter-
`mediate, (comparable to the thioimidate intermediate state
`known for nitrile-cysteine protease complexes; 15, 16).
`We hypothesized that imidate formation for nitrile DPP-
`IV inhibitors, if analogous to cysteine protease inhibition,
`should display slow-binding inhibition kinetics. A series of
`studies, focused on NVP-DPP728, have been undertaken to
`define the kinetic and molecular mechanisms for responsible
`for the potent inhibition of DPP-IV by cyanopyrrolidine
`compounds. Here we report that NVP-DPP728 inhibits DPP-
`IV by a slow-binding mechanism and that
`the rate of
`inactivation is dependent upon L-chirality of the pyrrolidine
`nitrile functionality. The binding affinity and rate of dis-
`sociation of bound inhibitor determined by kinetic experi-
`ments were further confirmed by direct binding measure-
`ments using radiolabeled NVP-DPP728 and DPP-IV highly
`purified from bovine kidney cortex.
`
`EXPERIMENTAL PROCEDURES
`
`A ) V
`
`and p-nitroaniline were from Sigma (St. Louis, MO). H-Ala-
`Pro-pNA was from Bachem (King of Prussia, PA). Bovine
`kidney cortices were obtained from Pell Freeze Biological
`(Rogers, AR). The human colonic carcinoma cell line Caco-2
`was obtained from the American Type Culture Collection
`(ATCC HTB 37).
`Inhibitors. NVP-DPP728 was prepared as described (com-
`pound no. 5 in ref 12). The D-antipode, des-cyano, and amide
`analogues of NVP-DPP728 were prepared as described (13).
`Dr. Tapan Ray (Novartis Radiosynthesis Laboratory), in-
`corporating the label at the carbonyl carbon, kindly provided
`[14C]-NVP-DPP728 (specific activity 49 mCi/mmol).
`Preparation of Human and BoVine DPP-IV. Where
`indicated, human DPP-IV preparations consisted of extracts
`of Caco-2 cells (17), cultured as previously described to
`induce differentiation (18). Cell extract containing human
`DPP-IV was prepared from cells solubilized in 10 mM Tris-
`HCl, 0.15 M NaCl, 0.04 tiu aprotinin, 0.5% nonidet-P40,
`pH 8.0, by centrifugation at 35 000g for 30 min at 4 (cid:176)C to
`remove cell debris. The preparations contained approximately
`30 mU DPP-IV/mg ((cid:24)0.6 (cid:237)g/mg of protein; 1 unit cleaves
`1 (cid:237)mol of H-Ala-Pro-pNA/min; enzyme content derived from
`Vmax determined using Gly-Pro-4-nitroaniline, using a theo-
`retical maximal activity of 55 U/mg as described (19)).
`Bovine DPP-IV was purified from kidney cortex using
`adenosine deaminase (ADA) affinity chromatography as
`previously described (20). Following digestion of a micro-
`somal membrane fraction with bromelain,
`the resulting
`soluble protein was resolved by sequential Q-Sepharose,
`ADA-Sepharose 4B, and Mono-Q chromatography to yield
`a >90% pure DPP-IV enzyme preparation with a molecular
`weight by SDS-PAGE of 105 kDa (specific activity was
`20 units/mg of protein).
`Kinetics of Inhibition of DPP-IV. The progress of DPP-
`IV inhibition by the indicated compounds was measured
`under pseudo-first-order inhibition conditions, i.e., [I0] g 10-
`[E0], by reacting DPP-IV with a mixture of inhibitor and
`substrate and recording the liberation of free pNA at 405
`nm. Unless otherwise indicated, all reactions were conducted
`using 20 (cid:237)g of extract protein in 25 mM Tris-HCl, 140 mM
`NaCl, 10 mM KCl, 1% bovine serum albumin, pH 7.4, at
`25 (cid:176) C (referred to as “assay buffer”). Under these conditions,
`Km for H-Ala-Pro-pNA was 73 (cid:237)M. Reaction progress was
`monitored using a Molecular Devices SpectraMax Plus
`microplate spectrophotometer (Sunnyvale CA). Reactions
`were 0.15 mL of final volume, initiated by the addition of a
`5 (cid:237)L aliquot of enzyme stock and mixed using the automated
`mixing feature of the SpectraMax reader. Total elapsed time
`between enzyme addition and the initiation of data collection
`was less than 30 s. Readings were taken every 10 s for a
`total of 1000 s, and initial (blank) absorbance values were
`subtracted from the data prior to subsequent calculations.
`Data were exported to Microsoft Excel and subsequently into
`the data analysis package Origin (Microcal Software Inc.,
`Northampton, MA) where curve fitting was performed. Data
`were fitted to the integrated rate equation for slow binding
`inhibition (eq 1) according to the method described by
`s)(1 - e-k’t)/k¢ + A0
`st + (V
`
`- V
`
`0
`
`Materials. Bovine serum albumin, bromelain, calf intes-
`tinal adenosine deaminase, CNBr-activated Sepharose 4B,
`
`Williams and Morrison (1979), by nonlinear regression
`analysis. Values for V0 (initial rate), Vs (final steady-state rate),
`
`SAXA-DEF-00400
`
`Page 2 of 7
`
`

`
`DPP-IV Inhibition by NVP-DPP728
`
`Biochemistry, Vol. 38, No. 36, 1999 11599
`
`FIGURE 1: Dose-response curve of DPP-IV inhibition by NVP-
`DPP728. Following a 10-min preincubation of human DPP-IV with
`the indicated concentrations of inhibitor, the reaction was initiated
`by the addition of H-Ala-Pro-pNA (166 (cid:237)M final concentration).
`Values are means ( SEM of three independent experiments. The
`line represents the logistic function with upper and lower asymptotes
`and slope fixed at 100, 0, and 1, respectively. The IC50 value derived
`from these data is 14 nM.
`
`k¢ (apparent rate constant for the transition from V0 to Vs),
`and A0 (the initial absorbance at 405 nm) were obtained for
`each progress curve. These values were subsequently used
`to generate kon (association rate constant), koff (dissociation
`rate constant), and Ki values as described in the Results.
`Radiolabeled Inhibitor Binding. Binding and dissociation
`of [14C]-NVP-DPP728 were studied by incubating 2.5 (cid:237)g
`(23 pmol) of purified bovine kidney DPP-IV with inhibitor
`in a volume of 4 mL of 50 mM Tris-HCl, pH 8.0, for 5 min
`at 25 (cid:176) C, followed by capture on DEAE cellulose membrane
`disks (25 mm diameter, Schleicher & Schuell). Bound
`enzyme was rapidly washed with 1 mL of the same buffer
`at 4 (cid:176)C, and both bound and eluted 14C were quantified by
`liquid scintillation counting in a Beckman (Columbia, MD)
`LS6000IC scintillation counter with quench correction
`(counting time was 20 min or 2% of (cid:243)). Nonspecific binding,
`less than 10% of the total bound activity, was determined in
`the presence of a 1000-fold excess of nonradioactive NVP-
`DPP728. For determination of dissociation rates of the
`enzyme-inhibitor complex, bovine kidney DPP-IV was
`incubated (2.5 (cid:237)g/time point) as above with 1000 nM [14C]-
`NVP-DPP728 for 10 min, followed by capture with 100 (cid:237)L
`of a 5:1 (gel:buffer) slurry of ADA-Sepharose 4B. The
`samples were then incubated with mixing for 20 min, and
`enzyme-bound inhibitor was collected on a 0.45 (cid:237)m nylon-
`66 membrane (Rainin, Woburn MA). The resin (with
`immobilized labeled inhibitor) was resuspended in 10 mL
`of buffer ([EI] after dilution was 2.3 nM). At the indicated
`time points, samples were removed and quickly filtered
`through Whatman type 1 filter paper disks (2.5 cm). The
`trapped resin was rapidly washed with 1 mL of ice-cold assay
`buffer, and both trapped (enzyme bound) and eluted (free)
`inhibitor were quantified by scintillation counting. Blanks
`containing radiolabeled inhibitor, but no enzyme, were
`subtracted from both the bound and free counts and were
`less than 10% of the total radioactivity. Dissociation curves
`were plotted as the log of the fraction of initial bound enzyme
`versus time following dilution. Off-rates were calculated as
`the slope of these plots.
`Inhibitor Stability. Under the conditions employed, NVP-
`DPP728 undergoes intramolecular cyclization, yielding a
`
`FIGURE 2: Lineweaver-Burk plot of DPP-IV activity measured
`in the presence of varied concentrations of NVP-DPP728 and
`substrate. Inhibitor effects were assessed as described in the legend
`to Figure 1, except both inhibitor and substrate concentrations were
`varied. Symbols correspond to different inhibitor concentrations
`as indicated in the legend. Values are means of triplicate determina-
`tions in which the standard deviations were less than 5% of the
`mean values. Lines shown are the least-squares linear regression
`lines.
`
`FIGURE 3: Slow-binding kinetics for the inhibition of DPP-IV by
`NVP-DPP728. Progress curves for pNA generation were recorded
`over 1000 s (16.7 min) at 405 nm. Measurement was done in 25
`mM Tris-HCl pH 7.4, 140 mM NaCl, 10 mM KCl, and 1 wt %/vol
`bovine serum albumin in the presence of 166 (cid:237)M H-Ala-Pro-pNA.
`Values are shown corrected for background absorbance (ap-
`proximately 0.03 AU). Symbols correspond to different inhibitor
`concentrations as indicated in the legend. Values are from one of
`three replicate studies.
`
`cyclic imidate product, with a half-life of approximately 72
`h. Accordingly, less than 1% of the compound is expected
`to cyclize during the time frame of the current investigations.
`
`RESULTS
`
`NVP-DPP728 fully inhibited H-Ala-Pro-pNA cleavage by
`DPP-IV derived from human colonic adenocarcinoma cells
`with an IC50 value of 14 nM (Figure 1). NVP-DPP728
`displayed complex inhibition kinetics when assessed graphi-
`cally by Lineweaver-Burk analysis (shown for illustrative
`purposes in Figure 2), consistent with results reported (9)
`for Xaa-cyanopyrrolidide compounds. Assessment of reaction
`progress curves in the presence of varied inhibitor concentra-
`tions revealed a clear time-dependent approach to steady
`state, characteristic of slow binding inhibition kinetics (Figure
`3). These progress curves were fitted to eq 1 to determine
`values for kon¢, the association rate constant for inhibitor
`binding. Values for k¢ were plotted against the inhibitor
`concentration, [I0] (Figure 4). A linear dependency between
`[I0] and k¢ was observed and fitted (eq 2) to obtain estimates
`
`SAXA-DEF-00401
`
`Page 3 of 7
`
`

`
`11600 Biochemistry, Vol. 38, No. 36, 1999
`
`Hughes et al.
`
`Table 1: Kinetic Constants for DPP-IV Inhibition by Pyrrolidine Compoundsa
`P1¢ substituentb
`S
`
`koff (s-1)
`kon (103 M-1 s-1)
`Ki ((cid:237)M)
`EI half-life (h)
`R
`compd
`(1.3 ( 0.2) (cid:2) 10-3
`127 ( 27
`0.011 ( 0.004
`0.14
`H
`CN
`NVP-DPP728
`(1.5 ( 0.2) (cid:2) 10-3
`0.27 ( 0.03
`5.6 ( 1.4
`0.13
`CN
`H
`D-antipode
`<0.01
`15.6 ( 3.6
`rapid
`rapid
`H
`H
`des-cyano
`320 ( 118
`CONH2
`amide
`ND
`ND
`ND
`H
`0.078 (cid:2) 10-3
`2.5
`5000
`0.000016
`H
`B(OH)2
`Pro-boroProc
`>672
`Pro-Pro(OPh)2
`irreversible
`0.02
`70
`H
`Pro(OPh)2
`a Reactions were performed at 25 (cid:176)C in 25 mM Tris-HCl, pH 7.4, containing 140 mM NaCl, 10 mM KCl, 1% bovine serum albumin, and 166
`(cid:237)M H-Ala-Pro-pNA.Values are means ( standard deviations for three experiments. b Functional group present at the pyrrolidine-2-position. c,d Values
`are from Gutheil and Bachovchin (5) and from Lambeir et al. (23), respectively. ND: not determined. EI half-life values were calculated as the
`ratio of 0.693/koff.
`
`d
`
`FIGURE 4: Determination of the association rate constant kon¢ from
`a plot of k¢ vs [Io]. The line represents a least squares linear fit of
`the indicated k¢ and I (NVP-DPP728 concentration) values. k¢ values
`were calculated according to eq 2. The line predicts a slope (kon¢)
`of 0.47 (cid:237)M-1 s-1. Values are means ( SEM of three separate
`experiments.
`
`kon
`
`(3)
`
`k¢ ) koff
`
`+ kon¢[I 0]
`(2)
`of kon¢ and koff. The rate constant kon¢ was subsequently
`corrected for the competition of the substrate using eq 3,
`
`) kon¢(1 + [S0]/Km)
`where [S0] is the concentration of the chromogenic substrate
`and Km is the separately determined Michaelis-Menton
`constant. Ki¢ values were determined using a direct, non-
`linearizing plot of Vs vs I, fitted to eq 4.
`¢) + 1)
`) v0/(([I0]/Ki
`V
`(4)
`s
`Ki was subsequently calculated from Ki¢ according to eq
`
`5.
`
`Ki
`
`(5)
`
`) Ki¢/(1 + [S0]/Km)
`
`The inhibition constants for NVP-DPP728, its D-antipode,
`and its des-cyano and amide analogues, determined in three
`separate experiments, are shown in Table 1. Potency of NVP-
`DPP728 was strongly dependent upon the presence and
`chirality of the P1 nitrile functionality. By alteration of the
`orientation (L- to D-) of the nitrile-pyrrolidine bond,
`approximately 500-fold loss of potency was observed. By
`removal of the nitrile substituent altogether (hydrogen
`replacement), a 1000-fold loss of potency resulted.
`Similarly, placement of a more bulky amide substituent
`with substantially less dipole character in place of the nitrile
`resulted in a 30 000-fold loss of potency. These results
`
`FIGURE 5: Dissociation of the NVP-DPP728-DPP-IV complex
`following dilution into substrate. An aliquot of DPP-IV enzyme
`previously incubated in the presence of 0 (squares) or 300 (circles)
`nM L-NVP-DPP728 was diluted 100-fold into 1 mM H-Ala-Pro-
`pNA, in assay buffer. Dissociation was monitored by substrate
`hydrolysis (absorbance at 405 nm). Absorbance readings were taken
`every 15 s for 30 min.
`
`indicated that the nitrile functionality, in the L- (or S-)
`configuration, imparts approximately 3.9 kcal/mol of binding
`energy compared to the des-cyano (-H) analogue.
`Inhibitor dissociation was studied by diluting the pre-
`formed NVP-DPP728/DPP-IV complex into a concentrated
`substrate solution such that the complex concentration was
`approximately 150-fold less than Ki¢ and the S/Km ratio was
`>10. Figure 5 shows that the DPP-IV enzymatic activity
`was slowly recovered from the inhibitory complex, indicated
`by the nonlinear increase in rate relative to the control curve.
`A value for k¢ was determined from the upwardly concave
`curve by fitting the data to eq 1, in which k¢ represents the
`rate for reestablishment of the steady-state equilibrium
`between DPP-IV and NVP-DPP728/DPP-IV complexes
`following dilution. A value for k-2 (Table 2, 1.4 ( 0.5 (cid:2)
`10-3 s-1) was then derived by linear regression from a plot
`of k¢ against I (not shown), where the y-intercept is taken as
`the rate constant for decay of the NVP-DPP728/DPP-IV
`complex, as described (22). Comparison of disassociation
`rates calculated for the des-cyano analogue of NVP-DPP728
`and the amide analogue (Table 1) indicate that the presence
`of the nitrile functionality of NVP-DPP728 imparts potency
`by promoting formation of a relatively long-lived complex.
`Equilibrium binding experiments were carried out using
`[14C]-labeled NVP-DPP728 in order to confirm the results
`obtained by kinetic methods and to assess the potential for
`effects of substrate on enzyme-inhibitor dissociation. The
`compound was bound to bovine kidney DPP-IV, and EI
`
`SAXA-DEF-00402
`
`Page 4 of 7
`
`

`
`DPP-IV Inhibition by NVP-DPP728
`
`Biochemistry, Vol. 38, No. 36, 1999 11601
`
`method
`
`Table 2: NVP-DPP728 Affinity and Dissociation Constants
`Determined by Kinetic and Binding Methods
`[S0]
`koff
`Ki
`(s-1)
`((cid:237)M)
`(nM)
`Kinetic Methods
`(1.3 ( 0.2) (cid:2) 10-3
`11 ( 4
`166
`initiated by enzyme additiona
`(1.5 ( 0.3) (cid:2) 10-3
`15 ( 1
`1000
`(1.5 ( 0.5) (cid:2) 10-3
`16 ( 4
`166
`initiated by dilution of EIb
`Bindingc Methods
`(1.0 ( 0.1) (cid:2) 10-3
`12 ( 2
`0
`without substrate
`(1.6 ( 0.1) (cid:2) 10-3 c
`200 ND
`with substrate
`a Reactions were initiated by the addition of enzyme and were
`performed at 25 (cid:176)C in 25 mM Tris-HCl, pH 7.4, containing 140 mM
`NaCl, 10 mM KCl, 1% bovine serum albumin, and H-Ala-Pro-pNA
`as indicated. b Reactions were initiated by 100-fold dilution of enzyme
`preincubated for 60 min with NVP-DPP728 into buffer containing
`substrate. Inhibition constants were determined as described in the
`Results. c Binding and dissociation of [14C]-NVP-DPP728 to purified
`bovine kidney DPP-IV was measured at 25 (cid:176)C in 50 mM Tris-HCl,
`pH 8.0, and 150 mM NaCl. Following washing of DPP-IV saturated
`with [14C]-NVP-DPP728 and dilution into buffer containing 0 or 200
`(cid:237)M Gly-Pro-AMC, bound and free inhibitor were collected by filtration
`and quantitated by scintillation counting. c Significantly different from
`0 substrate value (p < 0.0001 by Student’s t test). ND: not determined.
`
`FIGURE 6: Equilibrium binding of [14C]-NVP-DPP728 to bovine
`kidney DPP-IV. Aliquots (2.5 (cid:237)g) of bovine DPP-IV were diluted
`into 50 mM Tris-HCl, pH 8.0, containing 0-300 nM [14C]-NVP-
`DPP728. Following a 10-min incubation, enzyme-bound inhibitor
`was separated from free inhibitor and quantified by scintillation
`counting. The data represent the mean (SEM) of three independent
`experiments.
`
`complexes were adsorbed onto DEAE cellulose disks. After
`subtraction of nonspecific binding, the data were fit (Figure
`6) according to eq 6, where [EI] is the concentration of
`+ [I])
`[EI] ) [Etotal][I]/(Kd
`(6)
`enzyme-inhibitor complex, [I] is the free inhibitor concen-
`tration, [Etotal] is the enzyme concentration, and Kd is the
`dissociation constant (equivalent to Ki).
`The calculated Kd and [Etotal] derived from these data were
`11.5 ( 1.8 nM and 10.0 ( 0.4 nmol/mg of protein,
`respectively. On the basis of a unit molecular weight of
`110 000 Da, approximately 1 mol of binding was observed
`per mol of enzyme. The data also were plotted according to
`the method of Scatchard (24) (see eq 7, inset to Figure 6),
`[EI]/[I] ) ([EI]max
`- [EI])/Kd
`where the slope of the fit line is equal to 1/Kd and the
`x-intercept is equal to [EI]max. The Kd value obtained by this
`method was 8.6 nM, in agreement with the value obtained
`by both the saturation binding and kinetic methods. The
`
`(7)
`
`Scheme 1
`
`x-intercept (6.3 nM), equal to the concentration of binding
`sites, agreed well with the enzyme concentration of 5.7 nM.
`Thus, using equilibrium binding measurements with radio-
`labeled compound, it was possible to confirm the affinity
`measurements obtained using kinetic methods and to sub-
`stantiate a model for single-site, competitive binding.
`Since the both the D-antipode and the des-cyano analogue
`of NVP-DPP728 were found to have similar, low potency
`(Ki values of 5.6 and 15.6 (cid:237)M, respectively), it appeared
`that L-chirality was required for high-affinity binding of the
`nitrile functionality. Surprisingly, evaluation of the inhibition
`kinetics for the D-antipode revealed essentially identical
`dissociation rates (koff values of 1.5 and 1.3 (cid:2) 10-3 s-1 for
`the D- and L-isomers, respectively). Because the pair of
`inhibitors bind with markedly different association rates (127
`(cid:2) 103 vs 0.27 (cid:2) 103 M-1 s-1), but dissociate with identical
`kinetics, a series of experiments were performed to dissect
`the mechanism of slow binding.
`As described above, reaction progress curves obtained in
`the presence of a range of inhibitor concentrations indicated
`that NVP-DPP728 obeyed slow-binding inhibition kinetics.
`This behavior was indicated by the observation that NVP-
`DPP728-mediated inhibition of H-Ala-Pro-pNA cleavage
`approached steady-state equilibrium on a time scale of
`minutes under the conditions employed and the data could
`be fitted robustly to the slow-binding equation (22). Three
`mechanisms have been proposed that describe slow-binding
`behavior (Scheme 1, after Cha (25)).
`In mechanism A, enzyme (E) binds to the inhibitor (I) in
`a slow step to form a tight EI complex. In mechanism B, a
`loose EI complex forms rapidly and is followed by a
`(relatively) slow isomerization to a tight EI* complex.
`Mechanism C describes a slow isomerization of free enzyme
`(E) to form E* which can rapidly and tightly bind I, forming
`a tight EI* complex. To discriminate between the binding
`mechanisms, the relationships observed between I and k¢,
`and between I and V0, were assessed. The initial velocity
`(V0) was found to be significantly inhibited in proportion to
`the inhibitor concentration (Figure 7) for NVP-DPP728, a
`finding inconsistent with mechanism A in which V0 is
`predicted to be unaffected by the concentration of inhibitor
`(22).
`the first-order rate constant k¢
`The observation that
`increased with increasing inhibitor concentration (illustrated
`in Figure 4) was consistent with mechanisms A and B but
`not mechanism C, in which k¢ should decline with increasing
`inhibitor concentration. For this reason, mechanism B appears
`to best explain inhibition of DPP-IV by NVP-DPP728.
`Dissociation studies employing radiolabeled NVP-DPP728
`were conducted with purified bovine DPP-IV to confirm the
`binding constants determined by kinetic means using DPP-
`IV contained in cell extracts. For these studies, purified
`
`SAXA-DEF-00403
`
`Page 5 of 7
`
`

`
`11602 Biochemistry, Vol. 38, No. 36, 1999
`
`Hughes et al.
`
`Scheme 2. Proposed Model for Inhibition of DPP-IV by
`NVP-DPP728
`
`FIGURE 7: Dependence of initial velocities (V0) on NVP-DPP728
`concentration. Data represent initial velocity (Vi) values derived
`using eq 1 from progress curves measured as described in Figure
`3. Measurement was done in 25 mM Tris-HCl pH 7.4, 140 mM
`NaCl, 10 mM KCl, and 1 wt %/vol bovine serum albumin in the
`presence of 1 mM H-Ala-Pro-pNA. Values shown are means (SEM)
`of three experiments.
`
`-kt
`
`bovine kidney DPP-IV was saturated with [14C]-labeled
`NVP-DPP728, trapped with ADA-Sepharose, washed, and
`resuspended in buffer (with or without 0.2 mM H-Gly-Pro-
`AMC) such that the concentrations of enzyme and inhibitor
`were e 0.2 Ki. The concentrations of free and bound inhibitor
`were determined at 0, 1, 2, 3, 4, 6, 8, and 10 min after
`resuspension in buffer. The data (percent bound vs time) were
`fitted to a single-exponential decay curve (eq 8), where [EI],
`[EI] ) [EI]0e
`(8)
`[EI]0, and k are the concentration of enzyme-inhibitor
`complex at time t, the concentration of complex at time 0,
`and the rate constant, respectively. The half-life for inhibitor
`dissociation was taken as the natural log of 2 (0.693) divided
`by the rate constant and was determined to be 11.9 and 7.2
`min in the absence and presence of substrate, respectively
`(p < 0.0001 by Student’s t test, Table 2). These results
`indicated that, although the effects were relatively minor and
`although the dissociation of the inhibitor from the enzyme-
`inhibitor complex occurred more rapidly in the presence of
`the dipeptide substrate than in its absence, the enzymatically
`determined kinetic results represented a reasonable prediction
`of dissociation kinetics.
`
`DISCUSSION
`
`We have identified a new class of potent cyanopyrrolidine
`inhibitors in which a glycyl Xaa amine moiety is substituted
`with aliphatic and aromatic substituents (12, 13). These
`inhibitors are remarkably specific for inhibition of DPP-IV
`relative to other post-proline and -alanine cleaving enzymes
`(e.g., prolyl oligopeptidase, aminopeptidase P, and DPP-II).
`We have assessed kinetic behavior of this series in detail,
`focusing in this report on NVP-DPP728.
`Through kinetic evaluation of DPP-IV inhibition by NVP-
`DPP728, as well as by direct measurement of radiolabeled
`inhibitor binding in the presence and absence of substrate,
`we have established that NVP-DPP728 derives its potency
`through a slow-binding inhibition mechanism. Formation of
`the high-affinity complex is dependent upon the nitrile
`functionality within this series. Substitution with a variety
`of other substituents (e.g., amide, hydrogen) is associated
`with a significant loss of inhibitory potency as well as a clear
`
`loss of time-dependent function. Interestingly, while moving
`the nitrile from the L- to the D-configuration substantially
`reduces the overall potency of the compound, this loss of
`potency is due to a (cid:24)500-fold slower binding rate (kon, Table
`the dissociation kinetics for the L- and D-
`1). Indeed,
`enantiomers are identical,
`indicating that once formed,
`reversal of the high-affinity complex is independent of the
`nitrile orientation.
`While it is not presently possible to precisely determine
`the mechanism of binding of NVP-DPP728 to DPP-IV,
`structure-activity relationships support the involvement of
`several key interactions. First, the pyrrolidide ring interacts
`with the S1 pocket, through van der Waals or hydrophobic
`interactions. Second, hydrogen-bonding and ionic interactions
`stabilize the peptide bond carbonyl and the P2 site basic
`nitrogen functionality,
`respectively. Third, hydrophobic
`interactions stabilize P2 site side-chain binding in the S2
`pocket. These interactions may occur equally with nitrile and
`non-nitrile inhibitors. The negative charge derived from the
`acid-base-nucleophile (Asp-His-Ser) charge relay in the
`vicinity of the nitrile carbon drives a dipole-hydrogen bond
`interaction (interactions with both a hydrogen bond donor
`and the negatively charged active site serine) or transient
`imidate intermediate. The free energy change associated with
`the nitrile functionality, approximately 3.9 kcal/mol, may be
`adequately explained by either approach. These alternative
`high affinity state models are depicted in Scheme 2.
`Although additional and novel inhibition mechanisms can
`potentially be forwarded, several consequences of the model
`shown in Scheme 2 can be feasibly approached and will be
`
`SAXA-DEF-00404
`
`Page 6 of 7
`
`

`
`DPP-IV Inhibition by NVP-DPP728
`
`Biochemistry, Vol. 38, No. 36, 1999 11603
`
`addressed in subsequent communications. First, formation
`of an imidate intermediate should frequently proceed via
`hydration to yield a transformed amide byproduct (as
`observed for nitrile cysteine protease inhibitors; 26). A
`hydrogen-bond-stabilized dipole interaction would, in con-
`trast, be readily reversible, and inhibitor dissociation should
`generate unchanged parent compound only. Second, it should
`be possible to identify, through site-directed mutagenesis or
`through X-ray crystallography, the involvement of residues
`acting as hydrogen bond donors, capable of stabilizing nitrile
`inhibitor interactions. Recently, a high-resolution X-ray
`crystallographic structure of prolyl oligopeptidase has been
`reported (27) in which a tyrosine hydroxyl residue has been
`demonstrated to participate in stabilization of the oxyanion
`intermediate formed during binding of Z-pro-prolinaldehyde,
`a highly potent slow-binding inhibitor. Prolyl oligopeptidase
`is a member of the R,(cid:226)-hydrolase family, closely related to
`DPP-IV. The observation that hydrogen-bonding interactions
`may contribute to the stabilization of catalytic intermediates
`could potentially extrapolate to DPP-IV. Indeed, the finding
`that DPP-IV and prolyl oligopeptidase (28), both serine
`proteases, are strongly inhibited by nitrile-based inhibitors
`indicates that significant mechanistic differences may emerge
`which distinguish DPP-IV and other R