2100
`
`J. Med. Chem. 1998, 41, 2100-2110
`
`Specific and Irreversible Cyclopeptide Inhibitors of Dipeptidyl Peptidase IV
`Activity of the T-Cell Activation Antigen CD26
`
`Coralie Nguyen,† Julia` Blanco,‡ Jean-Paul Mazaleyrat,† Bernard Krust,‡ Christian Callebaut,‡ Etienne Jacotot,‡
`Ara G. Hovanessian,‡ and Michel Wakselman*,†
`SIRCOB, Universite´ de Versailles, Baˆtiment Lavoisier, 45 avenue des Etats Unis, F-78000 Versailles, France, and Unite´ de
`Virologie et Immunologie Cellulaire, ERS 572 CNRS, Institut Pasteur, 28 rue du Dr. Roux, F-75724 Paris Cedex 15, France
`
`Received September 26, 1997
`
`The dipeptidyl peptidase IV (DPP IV) activity of CD26 is characterized by its post-proline-
`cleaving capacity that plays an important but not yet understood role in biological processes.
`Here we describe a new family of specific and irreversible inhibitors of this enzyme. Taking
`into account the substrate specificity of DPP IV for P2-P1><-P1¢ cleavage, we have designed
`and synthesized cyclopeptides c[(RH2N+)-Lys-Pro-Aba-(6-CH2-S+R2)-Glyn] 2TFA- (Aba ) 3-ami-
`nobenzoic acid, R ) alkyl) possessing a proline at the P1 position and a lysine in the P2 position,
`which allows the closing of the cycle on its side chain. These molecules show a free N-terminus,
`necessary for binding to the CD26 catalytic site, and a latent quinoniminium methide
`electrophile, responsible for inactivation. Treatment of c[RZ-Lys-Pro-Aba-(6-CH2-OC6H5)-Glyn],
`obtained by peptide synthesis in solution, with R2S/TFA simutaneously cleaved the Z protecting
`group and the phenyl ether function and led to a series of cyclopeptide sulfonium salts. These
`cyclopeptides inhibited rapidly and irreversibly the DPP IV activity of CD26, with IC50 values
`in the nanomolar range. Further studies were carried out to investigate the effect of the
`modification of the ring size (n ) 2 or 4) and the nature of the sulfur substituents (R ) Me,
`Bu, Oct). Cycle enlargement improved the inhibitory activity of the methylsulfonio cyclopeptide,
`whereas the increase of the alkyl chain length on the sulfur atom had no apparent effect. Other
`aminopeptidases were not inhibited, and a much weaker activity was observed on a novel
`isoform of DPP IV referred to as DPP IV-(cid:226). Thus, this new family of irreversible inhibitors of
`DPP IV is highly specific to the peptidase activity of CD26.
`
`Introduction
`Dipeptidyl peptidase IV (DPP IV, EC 3.4.14.5), a
`membrane-bound exopeptidase, has been classically
`associated to the T-cell activation antigen CD26, a
`multifunctional sialoglycoprotein expressed on a variety
`of different epithelia and also by different hematopoietic
`cell types (for some recent reviews, see Yaron and
`Naider.1 and Fleischer et al.2). Dipeptidyl peptidase IV
`(CD26) is a serine protease which has the unique
`specificity to cleave dipeptides from the N-terminus of
`polypeptides provided that proline is the penultimate
`residue.1 In HIV infection, CD26 has been implicated
`in the viral entry process and its cytopathic effect.3
`Furthermore, DPP IV activity inhibition by HIV-1 Tat
`protein has been proposed as the mechanism of the lack
`of response to recall antigens observed in early stages
`of HIV infection;4 however, the relevance of this inhibi-
`tion in physiological conditions is unclear.5 Whatever
`is the case, the addition of soluble CD26 can restore
`response to recall antigens of HIV-infected individuals
`in vitro.6
`Irrespective of its peptidase activity, CD26 is associ-
`ated with other molecules on the cell surface. It has
`been shown to be the main receptor of adenosine
`deaminase,7 and on T-lymphocytes, CD26 is associated
`with CD45,8 a cell-surface-expressed phosphotyrosine
`
`phosphatase involved in signal transduction. Both
`features of CD26, its signaling capacity and its pepti-
`dase activity, contribute to the costimulatory function
`of CD26 in T-cell activation events. However, the role
`of DPP IV activity of CD26 in these events is unclear.
`Some authors have described that small synthetic
`inhibitors of DPP IV impair mitogen and antigen
`stimulation of PBMC and other lymphocytic cell types.9
`In contrast, others have found no effect of DPP IV
`inhibitors on stimulated T-cells.10 Similarly, contradic-
`tory results on the function of the DPP IV activity of
`CD26 have been reported by using cell lines expressing
`a mutated, catalytically inactive form of CD26.11,12 The
`failure to understand the role of DPP IV activity is, in
`part, due to the fact that no physiological substrates
`have been identified. However, a broad spectrum of
`bioactive peptides, including some interleukines, chemok-
`ines, neuropeptides, and growth factors, can be poten-
`tially cleaved by DPP IV.
`The availability of stable specific irreversible inhibi-
`tors or highly potent reversible inhibitors of DPP IV
`should be useful in studies for the determination of the
`physiological and pathological role(s) of this enzyme.
`Several competitive, tight-binding, or irreversible in-
`hibitors of the enzyme are already known: oligopeptides
`with the N-terminal X-Pro sequence (X ) various amino
`acid residues) such as the diprotins A and B (Ile-Pro-
`Ile or Val-Pro-Leu),13,14 X-pyrrolidides and X-thiazo-
`lidides,14-16 X-cyanoPro and X-cyanoThia,17-19 X-phospho-
`noPro or Pip aryl esters,20,21 X-boroPro,22-25 X-ProCH2N+-
`S0022-2623(97)00640-7 CCC: $15.00 © 1998 American Chemical Society
`Published on Web 05/13/1998
`
`* Corresponding author. Tel: 33 1 39254365. Fax: 33 1 39254452.
`† Universite´ de Versailles.
`‡ Institut Pasteur.
`
`Page 1 of 11
`
`AstraZeneca Exhibit 2017
` Mylan v. AstraZeneca
` IPR2015-01340
`
`

`
`Specific Irreversible Inhibitors of DPP IV
`
`Journal of Medicinal Chemistry, 1998, Vol. 41, No. 12 2101
`
`Chart 1. General Structure of the Cyclopeptides A
`Designed for Irreversible Inhibition of the DPP IV
`Activity of CD26
`
`Me3,26 X-azaPro derivatives,27 X-Pro-N-(arylcarbonyloxy)-
`fluoro olefin
`amides,28,29 and one of
`its ª(CFdC)
`isosteres.30,31 Owing to the presence of a free amino
`N-terminus and the flexibility of the imino peptide bond,
`several of these inhibitors easily cyclize and are not very
`stable in solution.32 The diacylated hydroxylamines are
`mechanism-based irreversible inhibitors of the pro-
`tease: a demasked acylnitrene can react directly with
`an active site nucleophile or can lead, through a Lossen
`rearrangement, to an electrophilic isocyanate.33
`We have previously designed and studied function-
`alized cyclopeptides containing a latent quinoniminium
`methide electrophile as suicide substrates for serine
`proteases.34 Taking into account the substrate specific-
`ity of the DPP IV enzyme for P2-Pro><-P¢ 1 cleavages
`(P1 ) Pro; Schechter and Berger notation35), we have
`now designed, synthesized, and studied cyclopeptides
`A (Chart 1): c[(RH2N+)-Lys-Pro-Aba-(6-CH2-S+R2)-Glyn]
`2TFA- (Aba ) 3-aminobenzoic acid, R ) alkyl), pos-
`sessing the same latent electrophile, as selective suicide
`substrates for this exoprotease. These cyclopeptides
`were able to induce the complete, rapid, and irreversible
`inhibition of the DPP IV activity of CD26 with IC50 in
`the nanomolar range. Their specificity was demon-
`strated by the lack of its effect on the activity of other
`peptidases, including the cell-surface-expressed DPP IV-
`(cid:226), the recently described protein with typical DPP IV
`activity.36
`
`Results
`From the relative weak importance of the nature of
`the P2 side chain on the rate of hydrolysis of the enzyme
`substrates (vide infra), we hypothesized that the cy-
`clization to the (cid:15)-amino function of a P2 lysine residue
`would result in new molecules able to bind the catalytic
`site of the DPP IV activity in CD26. Moreover, this
`cyclization would leave the N-terminal R-amino group
`free and protonated, a necessary condition for the
`recognition by the enzyme (Chart 1). In macrocycles
`A, the substituted P¢ 1 aminobenzoic acid residue Aba-
`(6-CH2-S+R2) is a precursor of the latent electrophilic
`quinoniminium methide cation. The presence of differ-
`ent numbers of glycine residues (n) in the cyclopeptides
`A allows the variation of the ring size, whereas the
`nature of the sulfur substituents (R) will modify the bulk
`and the lipophilicity in this part of the molecule. The
`influence of the leaving group ability on the inhibition
`efficiency and selectivity has been also examined by
`replacing the benzylic sulfonium substituent by an
`acetate group.
`Chemistry. Cyclopeptides A were prepared accord-
`ing to the reaction sequence shown in Scheme 1.
`
`Peptide synthesis in solution, using DCC/HOBt for the
`formation of the amide bonds, was applied for the
`synthesis of the linear precursors of these macrocycles.
`The strategy involved the use of a nitro substituent as
`a latent amino group.37 Coupling of the 2-(phenoxym-
`ethyl)-5-nitrobenzoic acid (Nba-[CH2OPh], 538) with
`either ethyl diglycinate or ethyl tetraglycinate gave the
`substituted ethyl nitrobenzoyl polyglycinates 6a,b. Se-
`lective catalytic hydrogenation of the nitro substituent,
`without hydrogenolysis of the benzyl ether function,
`occurred in the presence of platinium oxide in MeOH/
`DMF. The unstable aminobenzoyl polyglycine deriva-
`tives were rapidly acylated with N-Boc-L-proline to
`generate compounds Boc-Pro-Aba-[CH2OPh]-Glyn-OEt
`7a,b. Cleavage of the N-protecting group occurred in
`trifluoroacetic acid. In the following coupling step, the
`orthogonally diprotected Z-L-Lys(Boc)-OH derivative
`was used. The linear peptides 8a,b were obtained in
`good yields. Cyclization of these precursors was achieved
`by the azide method: hydrazinolysis of the ester func-
`tion leading to protected hydrazides 9a,b, selective
`cleavage of the N(cid:15)-Boc group of the lysine residue in the
`presence of the NR-Z protecting group, treatment of the
`resulting hydrazides with an alkyl nitrite, and dilution
`in DMF in the presence of a tertiary amine. The
`cyclization yields were 45% and 44% for 10a,b, respec-
`tively.
`Organic sulfides, such as thioanisole, in trifluoroacetic
`acid are known to deprotect O-benzyltyrosine without
`the formation of O-to-C rearrangement products39 and
`also to cleave the N-benzyloxycarbonyl protecting group.40
`The reactions occur by a “push-pull mechanism”:
`nucleophilic attack of the sulfide lone pair on the
`benzylic carbon of a protonated ether or carbamate
`function. The byproducts in these deprotection reac-
`tions are sulfonium salts. We reasoned that treatment
`of the c[RZ-Lys-Pro-Aba-(6-CH2-OC6H5)-Glyn] cyclopep-
`tides with various dialkyl sulfides in trifluoroacetic acid
`should cleave both the Z protecting group and the
`phenyl ether function (Scheme 2), thus leading to
`sulfonium salts having a protonated N-terminus. Ef-
`fectively, such a treatment gave the bis trifluoroacetate
`salts A: 11a (n ) 2, R ) Me, 62% yield), 11b (n ) 4, R
`) Me, 60% yield), 12a (n ) 2, R ) Bu, 74.5% yield),
`13a (n ) 2, R ) Oct, 71% yield).
`For the preparation of the cyclopeptide having a
`benzylic acetoxy substituent, the corresponding dim-
`ethylsulfonium salt 11a was treated with potassium
`acetate in dry DMF. The resulting acetate was not very
`stable and decomposed during purification. The crude
`product was therefore reacted with di-tert-butyl dicar-
`bonate to give the stable N-protected derivative which
`was easily purified by chromatography. Treatment with
`trifluoroacetic acid cleaved the Boc protecting group and
`led to the expected acetate salt 15a in 76% yield.
`Finally, selective hydrogenolysis of the Z protecting
`group of the cyclopeptide in the presence of the benzylic
`phenyl ether function was achieved by using a pal-
`ladium on carbon catalyst in aqueous methanol and
`gave the cyclopeptide 16a possessing a phenoxy sub-
`stituent (Scheme 2).
`For comparison, simplified linear analogues of the
`cyclopeptides A were studied. The trifluoroacetate salts
`+-Lys(H+)-H2+-Ala-Pro-Aba(6-CH2-S+Me2)-OMe and H2
`
`
`
`Page 2 of 11
`
`

`
`2102 Journal of Medicinal Chemistry, 1998, Vol. 41, No. 12
`
`Nguyen et al.
`
`Scheme 1. Synthesis of the Cyclopeptides 10a,b Having a Benzylic Phenoxy Substituent and a Terminal N-Z
`Protecting Groupa
`
`a (i) GlynOEt, DCC/HOBt; (ii) 1. H2/PtO2, MeOH-DMF, 2. Boc-Pro-OH, DCC/HOBt; (iii) 1. CF3CO2H or HCl/CH2Cl2, 2. Z-Lys(Boc)-OH,
`DCC/HOBt; (iv) NH2NH2/MeOH; (v) 1. CF3CO2H/CH2Cl2, 2. HCl/DMF-THF, 3. i-PrNEt2/DMF.
`Scheme 2. Last Steps of the Synthesis of Cyclopeptides 11a-16aa
`
`a (i) R2S/CF3CO2H; (ii) 1. KOAc/DMF, 2. Boc2O; (iii) CF3CO2H; (iv) H2/Pd/C, MeOH.
`Pro-Aba(6-CH2-S+Me2)-OMe (18a,b) were prepared as
`+-X-Pro-
`above from the corresponding benzylic ethers H2
`Aba(6-CH2-OC6H5)-OMe (17a,b).
`Biological Studies.
`In all the studies described
`here, the dipeptidyl peptidase activity of CD26 and DPP
`IV-(cid:226) was investigated using the enzymes in their
`natural habitat, i.e., by using either crude cell extracts
`or intact cells36 as described in the Experimental
`Section. Indeed, CD26 and DPP IV-(cid:226) are expressed on
`
`the cell surface, and thus their enzymatic activity could
`be assayed by using intact cells.36 As a source of CD26
`and DPP IV-(cid:226), we used either human CEM cells
`overexpressing CD26 by transfection49 or human C8166
`cells which express only DPP IV-(cid:226).36 In some experi-
`ments partially purified enzyme preparations were also
`used.51 In the case of CD26, we also used immunoaf-
`finity-purified preparations of CD26 as we have de-
`scribed previously.36 For the measurement of the
`
`Page 3 of 11
`
`

`
`Specific Irreversible Inhibitors of DPP IV
`
`Journal of Medicinal Chemistry, 1998, Vol. 41, No. 12 2103
`
`Figure 1. Specificity of molecule 11a. Effect of inhibitor 11a
`on different types of peptidase activities. Crude MOLT4 cell
`extracts36 were assayed for the effect of 10 (cid:237)M compound 11a
`on different aminopeptidase activities. The peptidases tested
`were as follows: DPP IV by the cleavage of GP-pNA, RP-pNA,
`and AP-pNA; Arg-peptidase by the cleavage of R-pNA; Ala-
`peptidase by the cleavage of A-pNA; and Pro-peptidase by the
`cleavage of P-pNA. At 10 (cid:237)M 11a, the DPP IV activity against
`different substrates was inhibited by more than 75%, while
`no apparent effect on the other peptidases was observed.
`Abbreviations of the amino acids: G ) glycine, P ) proline, A
`) alanine, and R ) arginine.
`
`activity of different peptidases, extracts from human
`MOLT4 cells were used.36
`Preliminary results pointed out that molecule 11a
`was a potent inhibitor of DPP IV activity of CD26.
`Indeed, the IC50 value for the inhibition of the DPP IV
`activity of a purified preparation of CD2636 was found
`to be 0.01 (cid:237)M. For this reason, we first studied the
`specificity of this inhibitor by determining its effect on
`different aminopeptidases found in crude cell extracts.
`The peptidases tested were as follows: arginine-pepti-
`dase (EC 3.4.11.6) by the cleavage of Arg-pNA, alanine-
`peptidase (EC 3.4.11.2) by the cleavage of Ala-pNA,
`proline-peptidase (EC 3.4.11.5) by the cleavage of Pro-
`pNA, and DPP IV by the cleavage of different substrates
`(Gly-Pro-, Arg-Pro-, and Ala-Pro-pNA).36 At 10 (cid:237)M
`molecule 11a, which is 3 orders of magnitude higher
`than its IC50, the DPP IV activity monitored with the
`different substrates was inhibited by more than 80%
`(Figure 1). The remaining residual 20% activity was
`probably due to the background (i.e., a nonspecific and
`CD26-independent cleavage of the substrates) since it
`was observed even at 100 (cid:237)M molecule 11a (not shown).
`In contrast, molecule 11a exerted no inhibitory effect
`on the activity of arginine-, alanine-, and proline-
`aminopeptidase (Figure 1). These results therefore
`demonstrated the specific nature of the DPP IV inhibi-
`tion by molecule 11a.
`To investigate the irreversible nature of the inhibitory
`molecule 11a, its effect on the DPP IV activity of CD26
`expressed on the cell surface was studied. For this
`purpose, we established by transfection of CD26 cDNA,
`human CEM cells which express high levels of CD26,
`i.e., cells which express high levels of DPP IV activity
`on the cell surface. Consequently, intact cells (clone
`H01) could be assayed for DPP IV activity by incubation
`with an appropriate substrate, such as Gly-Pro-pNA.36
`Under these experimental conditions, molecule 11a was
`found to be a potent inhibitor of cell-surface DPP IV
`activity, with maximum inhibition occurring at 2 (cid:237)M.
`First, we investigated the kinetics of inhibition of cell-
`surface DPP IV activity in these CEM cells preincubated
`
`Figure 2. Kinetics of molecule 11a-mediated inhibition of the
`cell-surface-expressed DPP IV activity of CD26. Intact CEM
`cells expressing high levels of CD26 (clone H01)49 were
`incubated for 5, 10, 15, and 30 min in PBS with or without 5
`(cid:237)M compound 11a at 37 °C and then washed twice with PBS.
`DPP IV activity of CD26 was then monitored by the cleavage
`of GP-pNA as described in the Experimental Section. The
`binding of the inhibitor 11a to DPP IV active site is very rapid
`since 5 min is sufficient to obtain more than 75% inhibition.
`
`Figure 3. Irreversibility of DPP IV inhibition by molecule
`11a demonstrated by the cell-surface-expressed CD26. Intact
`CEM cells overexpressing CD26 (clone H01) were preincubated
`for 15 min in PBS in the absence or presence of 5 (cid:237)M 11a at
`37 °C, then washed twice in PBS, and cultured in RPMI
`supplemented medium. Aliquots of cells were taken at the
`indicated times, washed twice with PBS, and then assayed for
`DPP IV activity by incubating cells with the substrate GP-
`pNA for 1 h at 37 °C. The samples at time 0 represent the
`DPP IV activity just after the 15 min of preincubation.
`with 5 (cid:237)M molecule 11a. At times of 5, 10, 15, and 30
`min, cells were washed extensively to remove unbound
`molecule 11a before assay of the DPP IV activity. As
`shown in Figure 2, inhibition was almost maximal after
`5 min of incubation, since the degree of inhibition was
`only slightly increased at 30 min. These results dem-
`onstrated that molecule 11a has the capacity to bind
`CD26 rapidly and thus inhibit irreversibly its DPP IV
`activity. Second, to confirm the irreversible nature of
`molecule 11a, cells were incubated with 5 (cid:237)M inhibitor
`for 15 min before extensive washing and further incuba-
`tion in the culture medium for up to 3 days. At different
`times during this period, cells were monitored for cell-
`surface-expressed DPP IV activity (Figure 3). The
`maximum inhibition observed following the 15-min
`incubation of cells with molecule 11a (time 0 h) was
`found to last for several hours. At 6 h, there was a very
`slight difference on the maximum inhibition, whereas
`at 24 h, there was about 50% inhibition. Interestingly,
`this 50% reduction of the inhibition at 24 h coincided
`with the doubling time of cells, which results in the
`
`Page 4 of 11
`
`

`
`2104 Journal of Medicinal Chemistry, 1998, Vol. 41, No. 12
`
`Nguyen et al.
`
`Table 1.
`Inhibition of DPP IV Activity by the Cyclopeptide
`Compoundsa
`
`inhibition of
`Gly-Pro-pNa
`cyclopeptides A
`hydrolysis (IC50, (cid:237)M)
`CD26
`DPP IV-(cid:226)
`glycine (n)
`leaving group
`ref
`S+Me2
`11a
`0.012
`1
`2
`S+Me2
`11b
`4
`0.003
`0.61
`S+Bu2
`12a
`2
`0.02
`0.57
`S+Oct2
`13a
`2
`0.03
`0.63
`15a
`1.5
`3.5
`2
`OAc
`16a
`10
`25
`2
`OPh
`a The DPP IV activity was monitored by the cleavage of Gly-
`Pro-pNA in purified preparations of CD26 and DPP IV-(cid:226) (Experi-
`mental Section). CD26 was purified using extracts of CEM H01
`cells expressing very high levels of CD26.49 DPP IV-(cid:226) was purified
`using extracts of CD26-negative C8166 cells.36 Both purification
`procedures were as described elsewhere.51 Such purified prepara-
`tions of CD26 and DPP IV-(cid:226) were free of any significant contami-
`nation by other peptidases. CD26 and DPP IV-(cid:226) preparations were
`preincubated for 15 min with different concentrations of each
`inhibitor ranging from 1 nM to 10 (cid:237)M before adding the substrate
`Gly-Pro-pNA as indicated.
`
`production of cells that express newly synthesized CD26
`molecules. At 3 days, no inhibition of cell-surface DPP
`IV activity was observed. Taken together, these data
`indicate that the inhibitory effect of molecule 11a is
`irreversible, since the DPP IV activity could be resumed
`only by the newly synthesized CD26 (Figure 3). CD26
`being a cell-surface-expressed protein, its expression on
`the cell surface could be modified with respect to the
`different phases of the cell cycle. Consequently at 24
`and 72 h, the DPP IV activity was variable in control
`cells (without preincubation with the inhibitor) (Figure
`3). It should also be noted that the cell-surface expres-
`sion of CD26 was not affected by coupling with molecule
`11a. Indeed, cells in the absence or presence of prein-
`cubation with molecule 11a manifested similar levels
`of cell-surface-expressed CD26, as monitored by FACS
`analysis using anti-CD26-specific monoclonal antibodies
`(data not shown).
`The other cyclopeptides, possessing different leaving
`groups (12a, 13a, 15a, 16a) or a larger ring (11b) were
`also assayed for their capacity to inhibit the DPP IV
`activity of CD26 using crude extracts from CEM cells
`(clone H01). The IC50 values for the inhibition of Gly-
`Pro-pNA hydrolysis are given in Table 1. All of these
`cyclopeptides showed different IC50 values in the nano-
`molar or micromolar range. However, no correlation
`was observed between these IC50 values and the length
`or lipophilicity of the leaving group. Molecules 15a and
`16a with leaving groups as acetate and phenoxy,
`respectively, manifested significantly reduced IC50 val-
`ues. In contrast, increasing the ring size in a given
`molecule generated a compound with an increased
`inhibitory activity (Table 1, compare molecules 11a,b).
`By use of the CD26-negative T-lymphoblastoid cell
`line C8166, we have recently described a CD26-like cell-
`surface protein with typical DPP IV activity.36 This
`novel form of DPP IV, referred to as DPP IV-(cid:226), was
`found to be distinct from CD26. However the pH
`optimum and the profiles for substrate molecules were
`found to be indistinguishable for both CD26 and DPP
`IV-(cid:226). Similarly, several previously described inhibitors
`of DPP IV exerted a very similar mode of action on both
`CD26 and DPP IV-(cid:226).36 Consequently, it remained
`
`Table 2. Characterization of the Inhibition Kinetics of CD26
`and DPP IV-(cid:226) by the Cyclopeptide 11a
`inhibition constantsb
`kinact (s-1)
`KI ((cid:237)M)
`enzymea
`11. 10-4
`0.085
`CD26
`22. 10-5
`0.470
`DPP IV-(cid:226)
`a The DPP IV activities of CD26 and DPP IV-(cid:226) were assayed
`by the use of extracts from CEM H01 (cells expressing very high
`levels of recombinant CD26) and C8166 cells (cells expressing only
`DPP IV-(cid:226)), respectively. The cleavage of the substrate Gly-Pro-
`pNA (0.5 mM) was monitored in the presence of increasing
`concentrations of inhibitor 11a by measuring absorbance at 405
`nm. Preparation of extracts and assay conditions were as described
`before36 and as in the Experimental Section. b Calculations were
`done as indicated in the Experimental Section. KI is the equilib-
`rium constant of the inhibitor binding to the enzyme, whereas kinact
`is the constant of the irreversible reaction that leads to the
`inactivation of the enzyme.
`
`essential to assess the action of the cyclopeptide inhibi-
`tors on DPP IV-(cid:226). The results given in Table 1 show
`that the inhibitory effect of the different irreversible
`cyclopeptide inhibitors is much more pronounced for
`it is
`CD26 compared to DPP IV-(cid:226). For example,
`interesting to note that the inhibitory effect of molecules
`11a,b is 83- and 203-fold higher on the DPP IV activity
`of CD26 compared to that of DPP IV-(cid:226), respectively
`(Table 1). In contrast to such a significant selectivity,
`molecules 15a and 16a exerted only about 2-fold dif-
`ference between the two enzymes. These data might
`suggest that the higher specificity of the inhibitors on
`the DPP IV activity of CD26 could be related to the
`reactivity of the cyclopeptide sulfonium salts.
`To further investigate the significant differences in
`the effect of the mostly studied inhibitorory molecule11a
`on DPP IV activity of CD26 and DPP IV-(cid:226) (Table 1), we
`studied the kinetics of inhibition of both enzymes, by
`using the approach previously described for irreversible
`inhibitors of trypsin-like proteases.34 This approach
`considers two steps in the inhibition process: first, the
`reversible binding of the inhibitor to the enzyme and,
`second, the cleavage and subsequent irreversible cova-
`lent modification of the enzyme leading to the loss of
`catalytic activity. The inhibition kinetics of the DPP
`IV activities of CD26 and DPP IV-(cid:226) fit well to this
`model, and the equilibrium constant of the first step (KI)
`as well as the kinetic constant of the second step, was
`calculated. The results are summarized in Table 2 and
`show that the higher potency of inhibitor 11a on the
`DPP IV activity of CD26 is the consequence of a higher
`affinity for this enzyme as demonstrated by the KI
`values, along with a faster inactivation rate as pointed
`out by the kinact values found for CD26.
`The difference in the inhibitory effect of irreversible
`inhibitors was further emphasized by testing the cell-
`surface-expressed enzymes. For this purpose, we used
`CEM cells expressing high levels of CD26 (clone H01)
`as a source of CD26 and C8166 cells as a source of DPP
`IV-(cid:226). Figure 4 shows the effect of different concentra-
`tions of molecule 11a and a previously described revers-
`ible inhibitor, Lys-[Z(NO2)]-pyrrolidide. As we had
`reported previously,36 Lys-[Z(NO2)]-pyrrolidide inhibited
`to a similar extent the DPP IV activity of both CD26
`and DPP IV-(cid:226), with comparable IC50 values. On the
`other hand, the molecule 11a completely inhibited the
`DPP IV activity of cell-surface-expressed CD26 without
`
`Page 5 of 11
`
`

`
`Specific Irreversible Inhibitors of DPP IV
`
`Journal of Medicinal Chemistry, 1998, Vol. 41, No. 12 2105
`
`Figure 5. Postulated mechanism of enzyme inactivation. Two
`covalent bonds are formed between the inhibitor and the
`enzyme (double-hit mechanism). First, an acyl-enzyme is
`formed which should involve the catalytic serine 630 of CD26
`and proline residue P1 of the inhibitory molecule (step 1).
`Second, by means of the unmasked quinoniminium methide
`(step 2), a second covalent bond is formed implicating a
`nucleophile residue Nu in the vicinity of the catalytic serine
`(step 3). Consequently, this second bond is responsible for the
`irreversible blockade of the catalytic site of the enzyme.
`
`IT1 by formation of an oxazolidine ring (IT2). Proton
`transfer from the terminal ammonium function leads
`to a third tetrahedral intermediate (IT3) and then
`preferentially to a cis acyl-enzyme which isomerizes to
`the trans isomer.
`For the herein described cyclopeptides A, the postu-
`lated mechanism of the observed enzyme inactivation
`is presented in Figure 5. Provided that the protease
`can accommodate the large molecule A, formation of an
`acyl-enzyme, by selective nucleophilic attack of the
`hydroxyl function of serine 630 on the P1 proline
`carbonyl of the cyclopeptide (step 1), would unmask a
`P¢ 1-substituted aniline having a good benzylic leaving
`group. Owing to the strong electron-releasing property
`) -1.31, compared
`of the amino substituent ((cid:243)+
`p-NH2
`p-NHCOR ) -0.6943 in the starting cyclopeptide),
`to (cid:243) +
`a fast elimination of a neutral dialkyl sulfide (or an
`anionic acetate) should give a substitued quinoniminium
`methide ion34 (step 2) tethered in the active site during
`the lifetime of the acyl-enzyme by means of the peptide
`chain. A Michael-type addition of a second nucleophile
`Nu, in the vicinity of the active serine, on this demasked
`electrophile (step 3) would establish a second covalent
`enzyme-inhibitor bond (elimination-addition pro-
`cess).44 Such a bridge between the active site serine and
`another nucleophilic residue active site leads to an
`irreversible loss of the enzymatic activity. Moreover,
`the inactivation is expected to be very selective as the
`inhibitory activity of suicide substrates (mechanism-
`based inhibitors) requires discrimination in the binding
`steps, catalytic activation by the target enzyme, and
`irreversible modification of the active center.45
`It should be emphasized that the macrocyclic sulfo-
`nium salts A are obtained in just one step by reacting
`an alkyl sulfide in trifluoroacetic acid with the cyclo-
`peptides having a benzylic phenoxy substituent and an
`
`Figure 4. Molecule 11a inhibits the DPP IV activity of the
`cell-surface-expressed CD26 but not that of DPP IV-(cid:226). Intact
`CEM cells overexpressing CD26 (clone H01) and C8166 cells
`expressing only DPP IV-(cid:226) were preincubated for 15 min in PBS
`containing different concentrations of molecule 11a (upper
`panel) or Lys-[Z(NO2)]-pyrrolidide (lower panel), before adding
`GP-pNA to monitor residual DPP IV activity. Control activity
`(100%) was determined in the absence of inhibitor.
`
`any apparent effect on cell-surface-expressed DPP
`IV-(cid:226), even when used at high concentrations such as
`10-20 (cid:237)M (Figure 4). It should be noted that molecule
`11a was active against the soluble form of DPP IV-(cid:226)
`with an IC50 value of 1 (cid:237)M. Thus, the inability of
`molecule 11a to affect cell-surface-expressed DPP IV-(cid:226)
`appears to be correlated with the structure of this
`protein when expressed on the cell surface. Whatever
`is the case, these results emphasize specific differences
`in the overall structure of DPP IV-(cid:226) compared to that
`of CD26.
`
`Discussion
`The DPP IV enzyme has an absolute requirement for
`the L-configuration of the amino acid residues, both in
`the penultimate and in the N-terminal positions. The
`amide bond between the N-terminal P2 residue and the
`P1 proline residue must be in a trans conformation. A
`protonated amino end is necessary for enzymatic hy-
`drolysis. Amino acids with aliphatic side chains at P2
`are slightly favored over aromatic ones, but the effect
`is not very large. On the basis of these literature data
`and the results of directed mutagenesis studies, Brandt
`et al. have recently proposed a model of the enzyme
`active site41 and a new catalytic mechanism for this
`glycoprotein,42 which has not yet been crystallized.
`Compared to usual serine proteases, the main difference
`is the stabilization of the first tetrahedral intermediate
`
`Page 6 of 11
`
`

`
`2106 Journal of Medicinal Chemistry, 1998, Vol. 41, No. 12
`
`Nguyen et al.
`
`NR-Z protecting group, themselves readily obtained by
`peptide synthesis in solution and cyclized by the azide
`method. These cyclopeptide sulfonium salts, as all
`compounds 11-16, belonging to the A series of mol-
`ecules, are stable and water-soluble, and their activity
`does not decrease even after several months in solution.
`Their cyclic structure prevents their decomposition by
`cyclization due to the attack of the free amine on the
`carbonyl of the proline. Moreover, the toxicity of these
`compounds on CEM cells was observed only at concen-
`trations higher than 250 (cid:237)M.
`Similar inhibitory activities were observed among the
`compounds 11a, 12a, and 13a, suggesting that the
`length of the alkyl chains on the sulfonium group has
`little or no influence on the inhibition of DPP IV activity
`of CD26.
`Interestingly, the size of the macrocycle
`appeared to play an important role, since a strong
`inhibition was observed with compound 11b which
`contains four glycine residues in its cycle. The inhibi-
`tory activity of this compound is 4 times more effective
`than that of its homologue molecule 11a, which pos-
`sesses only two glycine residues. The weak inhibition
`observed with compounds 15a and 16a is not surprising,
`since the acetate and the phenoxy groups are not good
`leaving groups compared to the alkyl sulfide ones. For
`the linear analogues of the cyclopeptides, no inhibition
`was observed at 100 (cid:237)M. In this case, a fast diffusion
`of the quinoniminium methide cation out of the active
`center may occur due to the lack of the peptide chain
`which tethers the electrophile in the active site. Taken
`together, our results show that large substituents on
`the sulfur atom did not significantly modify the inhibi-
`tion reaction. In contrast the ring size enlargement
`improved it, probably by allowing more favorable in-
`teractions with the enzyme active center and/or facili-
`tating the approach to the nucleophilic group Nu in the
`active site.
`Comparative studies on the inhibition of the DPP IV
`activity of CD26 and DPP IV-(cid:226) revealed that the two
`enzymes should be distinct from each other. In soluble
`enzyme preparations, the irreversible cyclopeptide in-
`hibitors were less active on DPP IV-(cid:226) compared to
`CD26. This latter is most probably due to both a
`different affinity and a different inactivation rate, as
`we have demonstrated to be the case with the cyclo-
`peptide 11a (Table 2). Consequently