Structure/Function Relationships in the Inhibition of Thimet Oligopcpti,Tase by Carboxyphenyl Propyl-Peptides

Home , Thermolysin, Thimet oligopeptidase

Volume 294. number 3. 183-186 FEBS 10458 Dc~:cmlx-r IO91 ~ 1991 Federation of European Biochemical Societies 00145793/91/$3._~t)

Structure/function relationships in the inhibition of thimet oligopcpti,tase by carboxyphenyl propyl-peptides

C. Graham Knight and Alan J. Barrett

Der~rcs~ncnl of Bi¢~chemistrt'. Strangeway$ Research Laboratory. EVorts Causeway. Cambridge CBI 4RN. UN

Received 26 September I(~I; revised version received 17 October 1991

Some novel N-[l(RS)-carboxy-3-phenylpropyl]trilaeptide p-aminobcnzoates hawe bexm synthesised as inhibitors of lhimet oligopeptidas¢ (EC 3.4.24.151. The~e compounds are considered to bind as substrate analogues with the Cpp group in SI and the l~ptide portion in the S" ~ite~. The most Dotc,t inhibitor is Cpp-Ala-Pro-Phe-pAb. which has a K. - 7 nM. Substitution of (~ly for Ala at Pl" leads to weaker binding which can be ascribed to increa.~d rotational tr~x'dom. Good substrates often have Pro at P2" and Pro is favoured over Ala at this position in the inhihitors, too. When P2" is Pro. Phe is preferr~ over Tyr and Tip in P3 +. The p-amlnobenToate group makes an ;ml~,rtant contribution to the binding, probably by forming a .salt bridge, and removal of the ('-terminal negative 4zhargc results in much less potcnl inhibitors.

EC 3.4.24. I S; Fm, ym© inhibition; Subsite specificity

i. INTRODUCTION tide-pAb are competitive inhibitors, and since the most potent have peptide structures analogous to those of Thimet oligopeptidase (EC 3.4.24.15) is ~" thiol-de- good substrates [4], the amino acid sidechains pre- pendent metallo-endopeptidase widely distributed in sumably occupy the same specificity subsites on the dells and tissues, and previously known by the names enzyme. By convention [7], these subsites are labelled S Pz-peptidase, endo-oligopeptidase A and soluble endo- to the N-terminal side of the scissile bond and S" to the peptidase 24.15 [I]. Characteristic features of the en- C-terminal side, numbering being away from the bond zyme are the inhibition by both chelators and thiol- cleaved, the scissile bond. The amino acid residues reactive reagents and activation by low concentrations which occupy these subsites are labelled P and P', re- of thiols. The biological function of thimct oligo- spectively. Although it is not yet possible to define the peptidase is unknown, but is capable in vitro of generat- characteristics of the subsites precisely [I], good sub- ing or destroying a number of pharmacologically active strates and inhibitors of thimet oligopeptidase com- peptides and is found in the highest concentrations in monly have hydrophobic residues at Pi and PY [2,4]. the soluble protein fraction of testis, brain and pituitary In addition to their use in biological experiments, [2]. lnhibitors may provide the best approach to inves- inhibitors more potent than those currently available tigating the physiological role of the enzyme [3]. are needed for active site titration of the enzyme, so that Thimet oligopeptidase is selectively and reversibly in- kinetic studies can be placed on a quantitative basis. hibited by N-[l(RS)-carboxy-3-phenylpropyi]tripeptide Also needed are inhibitors more resistant to degrada- p-aminobenzoates (Cpp-peptide-pAb) [4,5]. The mecha- tion, which currently complicates their use in studies of nism of inhibition may be similar to that proposed for the role of thimet oligopeptidase in vivo [3]. We now thermolysin and Cpp-Leu-Trp [6], where X-ray crystal- report our initial studies to investigate the relationships lography has revealed the coordination of the carboxyl between inhibitor structure and function. ol the Cpp group to the Zn atom in the active site of the enzyme. In thimet oligopeptidase, interaction of the 3- 2. EXPERIMENTAL phenyl ring with the hydrophobic S I subsite also makes 2.]. Maaet4ais an important contribution to the binding [4] Cpp-pep- Boc-amino acid N-hydroxysuc~nimide esters, p-aminobenzoic acid and L-phenylalanine derivatives ~ from Sigma. Uoc-amino acid Abbrevlatlons: Ahx, L-2-aminohexanoic acid; Boc, t-butyloxycarbo- p.aminobenzoates were made as de~ribed [8]. Ethyl; 2-keto-4-phenyl- ttyl; CI~, N-[It(RSk-¢arboxy-3-phenylpropyli; Drip, dinitrophenyl; butanoat© from Scbeizerhall (South Plainfield, NJ 07080) was saponi- Moc. 7-melhoxyooumarin-3

Pub/is&ed by E./sev~r Sck,nce Pub/t~'rs B. g,'. 183 Votume .~4. number 3 UEBS LETTERS December 1991

2.2. 5;Ynth,'s:.~ ¢,I Cpp-.4la-Pro-Ph*'-p.4~ was controlled, and data colh_.cted and analys~d, hy use of the This was made step-wire by standard ~ltstion-pha~ mctht~cls [4. I0]. VI.USYS software Dackag~ [I I ]. Whe en,wm¢ was pro-activated with Reactions ~crt" ft~llowt~J by HP[.C on a C1~4 L'olumn, clulcd with tt S mM dithiothreitol in assay buffer (50 mM Tris-HCI, DH 7.g, contain- gradient of 5 100~4;, aL~tomtrilc containing 0 I¢IF trIfluorowuctlc acid Ing tlOSqF Itr U iS) for 5 rain at 21"( and then diluted IO*fold and kept (Iml,min) over 25 rain. Tht, colulnn ¢Itiant ~Jas monitored at ".10 rim. at 0"('. A~ay~, were made with OR nM em, yme. based on protein I~--Phe-p,Ab 1384 rag. I retool) gas treatL,~J with trifluolat~llccYic ;lt:iti/ concentration. For the determination of K,. the rate of substrat¢ hy- water I 19 I. ~'~'v. :q) ml) for 20 n~it~ at 21 ~'C to remove the B¢~.- group droly~i~ in the absem.'¢ of inhibitor was first recorded, and then the The acid was removed at 44)°(" in vacuo and the t~sidue was dissolved inhibitor was add~ in a negligible volume and the new steady stale in dimcthylfomlamidc (Sail, To this ,~tuiion was addc~l B~'-Pro- was monitored Rapid equilibration was observed in all cases. Ap- ONSu (330 rag, I I mmolL l-hydroxyl~n~otria~c~le (I 35 g, I ntm~i) parent inhibition constants A'.=r~., were calculated by fitting Ihe frac- and di-isopropylethylamine (O.7 ml. 4 mmolL After stirring overnight, tional activity darn to II1¢ Mlorrisort L~quation [I 2| by nonlin~r relgl~o analysis by HPLC sho~k~t."d the reaction to the complete. The dimethyl- sion, using the p¢ogt am En~litter ! I-Ascvier-Biosofl, Cambridge, UK). formam/d~ was evaporated at 40°(" in vacuo and chloroform (~) ml) Values wct'¢ corrected lor the effoet ofsubstrate by using equation (t). was adld~. After eglra~liom with ~ tw/v} citric acid (2 ~< ~ mlL water 120 ml) and saturaled Na('l (20 roll the chloroform solution was K, = K.~.~/(I ÷ [SFK.) (l) dried (M8SO4) and evaporated to leave crude Boc-Pro-Phe-pAh_ This was Ptx-rystalli~L'd from ethi*nol/water (yielding 0.46 g. 95~[-I and A value of 16 .u M was dctcrmin*~! for Km. The different:c~ in K, va|tt~ treated with tri~luoroa,cctic acid/water as described above, The residue I'~twL~n hthibitors were converted into differences in the fr~ energy after evaporation was dissolved in dimethylformamkle and reacted of binding t~(~. d hy equation (2), with Eloc.Ala.~)NS¢ (2~ mr. I mmoh. I-hydroxybenzotriazote and di-isop~xspylethylamine. As before, the solvent was evaporated after --. -2.303RT- Iogl reac*ion overnight and the residue dissolved in chloroform, extracted 66G~,.a K,,tK',:) (2) and dried. The crude Ek~-Ala-Pro-Phe-pAb was 97q~, pure by HPL(" analysis and was treated with trifluoroa~i~tic acid without further purification. The residue after evaporation was dissolved in methanol/ 3. RESULTS AND DISCUSSION water ( 1:1 v/v, ~ ml) and brought to pH 7 with 5 M NaOH. Sodium 3. I. A lanine is fitvoured in Pl" 3-phcnyl-2-ketobutyrate ( I g, 5 retool) and sodium cyanoborohydride (0.32 g. 5 retool) were added and She mixture stirr~ at 2 I~'C. HPLC Good peptide substrates of thimct oligopeptidase analysis showed that reaction was ~.~.~mplete after 24 h. The solveats commonly have hydrophobic residuesin PI and P3". and were evaporated, water (50 ml) was added and the ,solution acidified arginine is often found in Pl" [2], suggesting that the SI" with 6 M HCI ( 1.5 ml) to precipitate Cpp-Ala-Pro-Phe-pAb.HCI. The site of the enzyme is accessible to large side chains. This crude prodtu:t wa,~ washed with ethyl acetate, and reerystallised from ethanol and water, yielding 0.33g ¢49~-), m,p. IS? 188°C. Analysis: conclusion is supported by studies with inhibitors, since calculated for C,,H~N~O~.H(?I (M, 679.2) C. 62.72; H, 6.O4; N. 8.60; the K, of Cpp-Ph¢-Ala-Phe-pAb is only 3-fold that of t~)und (', 62.98: H, 6.1 I; N. 8.60. Cpp-Ala-Phe-pAb [4] (Table 1). Leucine. however, is The other inhibitors w~re mad~ by similar methods. The identities found to be an unfavourable residue in Pi'. of the final l,~ptide products were confirmed by elemental anatysis, It has been observed that replacement of Ala by Gly NMR or fast atom bombardment mass spectrometry. All the phenyla- lanin~-¢ontaining inhibitors displayed characteristic double peaks on in Pl" to give Cpp-Giy-Ala-Phe-pAb results in a far less HPLC. attributabh: to the Daetially resolvt'd R and S diastereomers potential inhibitor [4.5], and we find similarly weak 141. binding when P2" is Pro (Table !). in principle, the weaker binding of the Cpp-Gly- inhibitors could be due 2.3, D¢ls,rttaOl¢it~on ol K, vcllut',s Thimet oligoraepttdase activity was assayed at 300( `` with the to the lack of an interaction of the alanine side chain in quenched fluorc,s~cat frcptidc Mc~.'-Pro-L~u ~(~ly-Pro-D-Lys(Onp1191, the SI" pocket of the enzyme, or to greater rotational in ~,hich the symbol + represents the s¢issile bond. The fluorlmeter freedom allowing the compounds to adopt unfavourable conformations. However. the K, differenoes Table ! between Cpp-Ala-Ala-Phe-pAb and Cpp-(31y-Aia-Phe- Effi.~ct of inhibitor structurt: at PI' pAb. and Cpp-Al,~-Pro-Phe-pAb and Cpp-Gly-l~o- Phe-pAb correspond to differences in binding free en- i + l" 2' 3' 4" K, (nM) ergy AAGbind of !.9 and 2.2 kcal/M, respectively. By contrast, the free energy change on transfer of a methyl Cpp Phc Ala Phe pAb 81 -~ 10" group from octan-l-ol to water is only 0.7 kcal/M [13], Cpp - Leu Pro Phe pAb 279 _+ 10 CPlP Ala Ala PIR - pAb 30.6 _+ 1.0 which sug~oests that the hydrophobi¢ pocket model is CI~ -GIy- AI2 -- ~- pAb 774 _+ 16 inadequate to explain the differences in gi and that the Cpp - Ala ~ Pro - Phe - pAb 7.3 + 0.3 increased flexibility of the gly¢ine derivatives is re- Cpp - Gly -- Pro Phe -- pAb 289 ± g sponsible for their poorer binding. Such flexibility has Cpp Sar • Ala . PI~ pAb 15700 ± 200 CI~ Sat Pro Phe pAb 3410 ± 130 been shown to be responsible for the weaker binding of analogues of Cpp-Ala-Pro (enalaprilat) to peptidyl-di- The Cpl~peptide-pAb inhibitors nre substrate nnalogues and are peptidase A (angiotensin converting enzyme) [14]. With thought to bind with the Cl~ group in the Sl subtile of the enzyme. these compounds, binding could be enhanced as much The residues have been numtmred accordingly, and the symbol *+' in- as 2500-fold by conformational restriction, so it seems dicates the bond corresponding to the scissile bond in a substrate. The error va|uea are standard errors es:imated by the E~fitter pro- reasonable to suggest that restricted rotation around the gram. alanine N-Ca bond plays a role in orienting the Cpp *Value from 13l carboxyl group towards the Zn atom in the active site

184 Volume 294, number 3 FEBS LETTERS December 1991 of thimet oligopeptidase, although other interactions Table !11 probably contribute to efficient binding. For example, I'~I1/~X-I aft inhibitor ~lruclu'~ :It P4" in the complex hetwcen thermolysin and Cpp-Leu-Trp (K, = 50 nM [15]). both oxygens of the Cpp carboxyl I 4 I' 2 ~ 3" 4" K, (nM) group arc coordinated to the zinc, but in addition e~ch Cpp Ala Ala Phe pAb 30.6 ~ 1.0 forms two hydrogen bonds to other residues lining the Cpp Ala Ala Phe OH 155 ± 5 active site [6]. Similarly. the protonated secondary Cpp Ala Ala Phe NHPh t860 + 50 amino group forms three hydrogen bonds to surround- Cpp Ala Ala Phe NHPhNO~ 10~ ± 40 ing residues [6]. The importance of this nitrogen in the Cpp Ala Ala Phe - NH: 5790 ± 220 interaction of Cpp-inhibitors with thimet oligopeptidase is shown clearly by the substitution of sar- cosine (Sar) for Gly (Table I) which results in very interactions at the C-terminus with a series of reagents marked losses of activity. in which the Cpp-Ala-Ala-Phe peptide part was retained (Table !II). The most potent inhibitor remains the p- 3,2. Binding is favoured by proline in P2" aminobenzoate, but surprisingly, removal of" the pAb Several good substrates by thimet oligopeptidase group to yield the Free acid results only in a 5-fold have Pro in P2". including bradykinin and neurotensin increase in K,, corresponding to the loss of 1.0 kcal/M [2]. Bz-Gly-Phe+Aia-Pro-Phe-pAb [16]. Pz-Pro-Leu+ of binding free energy. By contrast, replacement of the Giy-pro-D-Arg [! 7] and the quenched fluorescent pep- carboxyl group by a hydrogen atom to give the un- tide, Dnp-Pro-Leu+Gly-Pro-Trp-D-Lys [18]. These charged anilide results in binding weaker by 2.5 kcai/M. data suggest that the binding of peptides to thimet oli- The p-nitroanilide inhibitor, with the dipolar nitro gopeptidas¢ may be favoured not only by hydrophobic group in place of the carboxylate, is only marginally residues in Pl and P3", but also by Pro in P2". To inves- more effective than the anilide. The important contribu- tigate the effect of pro in this position, we have com- lion of the C-terminal negative charge to the binding is pared some members of the series Cpp-Ala-Aia-X-pAb confirmed with the amide Cpp-Ala-Ala-Phe-NH2, and Cpp-Ala-pro-X-pAb, where X is an aromatic or which binds more weakly than its parent acid by 2.2 large hydrophobic amino acid (Table II). Substitution kcal/M. These free energies are in the range of those of Pro for Ala leads to slightly tighter binding and Cpp observed for salt bridges in proteins [I 3] and suggest Ala-Pro-Phe-pAb is the most potent inhibitor of thimet that, in addition to binding to the active site Zn, the oligopeptidase so far described with Ki = 7 nM. p-aminobenzoate inhibitors have an important elec- Orlowski et al. [4] have observed that the binding of trostatic interaction with |ysine or arginine side chains Cpp-Ala-Ala-Tyr-pAb to thimet oligopeptidase is twice in $4 ° or $5" subsites. as tight as that of the Phe analogue, but we do not see this effect when the P2" residue is Pro and both Cpp- 4. CONCLUSION Ala-Pro-Tyr-pAb and Cpp-Ala-Ala-Tyr-pAb are com- parable in inhibitory activity. By contrast, the inhibitor Although the studies described here have resulted m with Leu in PY shows a marked decrease in inhibitory an inhibitor of thimet oligopeptidase that binds more potency (T~ble !1). tightly than those described before. Cpp-Ala-Pro-Phe- pAb binds too weakly to be used as an active site titrant, 3.3. Potent inhibition requires a negalively charged C- and a further decrease in Ki of about 20-fold will be terminus necessary to achieve this aim. One approach to tighter All the inhibitors discussed so far have been p-amino- inhibitors may be to explore alternatives to the Cpp benzoates. Stewart [19] has pointed out that the pAb group, perhaps by introducing substituents into the group is structurally related to a dipeptide and this phenyl ring, as the 70-fold difference in inhibitory group seems likely to make a major contribution to the potency between carboxyphenyipropyl and carbox- binding of the inhibitors. ~'e explored the influence of yphenylethyl derivatives of Ala-Ala-Phc-pAb {4] il- lustrates the dramatic effects of small changes in struc- Table !! ture in this region of the molecule. Effect of inhibitor structure at P2' and PY Cpp-Ala-Pro-Phe-pAb may prove to be a good inhibitor for in vivo studies of thimet oligopeptidase. Cpp- I ÷ l" - 2" 3" 4" K, (riM) Ala-Ala-PheopAb is cleaved by neutral endopeptidase (EC 3.4.24.1 I) at the Ala-Phe bond [3], whereas the Cpp- Ala -- Ala - Ph¢ - pAb 30.6 _+ 1.0 Ct~ - Ala -- Ala - Tyr - pAb 14.4 + 0.5 Pro-Phe bond may be more resistant to hydrolysis. This Cl~ - Ala - Pro - Ph¢ - I~Ab 7.3 4- 0.3 bond may be cleaved, however, by prolyl oligopeptidase Cpp - Ala - Pro - Tyr - pAb I 1.8 + 0.4 (EC 3.4.21.26) and control experiments with specific Cpp - Aim. - Pro - Trip - pAb 17.6 _-4 1.0 inhibitors [20] of this enzyme will be necessary, Cpp - Ala - Pro - Lee - pAb 76.8 + 3.0 An advantage of the Cpp-peptide-pAb inhibitors is

185 Volume 294. number 3 F|~BS L i'.-FFFRS December 1991

their rapid binding to thimct oligor~ptida~e, so that 15] llafretl. A.I and Br,~uwn. MA. (Iqg0) Biochem. J. 271, 701 706. equilibration is inslantaneous on the 10 rain time scale |61 Mon;'ingo. A |:, ~nd Matthew,,. B.W. (1984) Biochemistry 27, of our as.~ays. By contrast. N-{ phenylethyiphosphonyl)- ~,724 572,~ 17l ,%¢h¢'~.'t¢r. I and BerBer. A. (1967} Biz~hem. Biophyg. Res;. Coitt- (31y-Pro-Ahx, designed as an inhibitor of CIo.vtridium nt~m. 27. 157 162. hLgtolyth'u~ coIlagena~ [21], requires several minutes |8] Oiil~wski. M, M~chaud. (" and Chu, T_CL (1983) l~ur. J, to reach equilibrium with thimet o|igopeptida.,~e, al- Bttx:hem. 13~. 8l 88. though the final K, (7 nML is similar {A.J. Barrett and [0l l'i~ljar, ILl., Knight, C.G. and Barrett, A.J, t 1990) Anal. Biocltcm. M.A. Brown. unpublished observations). Likewise, 18¢,, 112 ;15. [ ! 0 ] Bodans~'k y, M. and l~,¢lans~ky, A. (1984) The Practice of Pcplicl¢ phosphoramidon ( N-[[(6-deoxy-~- L-ma nnopyranosyl)- Synthesis. Sprinl~c,r-Verlag. Berlin oxy]hydroxyphosphinyl]-Leu-Trp) is a slow binding in- II I1 Bawlings. N,D. and Barlrctt. A J. (I~JN))Computer Appl. Bio~ci, hibitor of thermolysin [221. whereas Cpp-Leu-Trp binds 6, lib 119. normally { 15]. 112] Morrison, J,F. (1%9) Biochim. Biophys. Acta 185. 209=286. | 13] Fcrshl. A. (1985) Enzyme Structure and Mechanism, 2nd ¢d.. pp. 29~ 310. ,4¢~nowlcd, g,¢m~,nts: We ate fratefu] to Dr. M. Ortowski l~r providing [14] Thorsett. E.D. t1086) Actual. Chim. |her. 13, 257 268_ ('pp-A|a-Ala-Tyr-pAb, to l...sut.~e Handlord and Hotly Spcatmg for i15] Nlaycock. A.L.. [kSou,~a. D,M.. Payne. L.G.. ten Broekc. J.. expert technical assistance ant: to l)r, tLCC. ticywo¢~l for obtaining Wu. M.~. and Patch¢tt, A.A. (1981) Biochfm. Biophy~. R¢~, i+nd tnterpt¢tl~g the NMR aild nlass sp~.x:tra. Commun. 102, 963 969. [10] Chu. T.(~. and Orlowski, N~. (!~85} [.~ndocrino|ogy 116, 141g 1425. REFERENCES [17] Barrett. A,J, ~nd Ti~ljar, U. (19B9) BicMchem, J 261, 1047 1050. [I] Barrett, A,J. (11)90) Biol. Chem Hoppe-Seylcr 371 (Suppl.), 31 ! [lgl Knight. C',(L (l~ll Btochem J 274. 45 48. 320. [191 Stewart. F.H.('. (19K3} Aa~tr. J, ('hum. 36, 1629 If~3X 12] Orlowski. M., Reznik, S.. Ayala, J. and Picrottl. A.R (1989) [201 T~,ttru, I~., Voshimoto, T., Koriyama, N. and Furukawa, S. Biochem. J 261. "~51 95R. (19RSl J. Biochcm. 104. 580 586. 13] Lasdun. A., Res.nik, S.. Moliaeau~. ('.J. and Orlow.~k i, M. (19891 I211 Dive, V,. Yiotakis, A.. ]~lic'olaou. A. and Toma, I:. (1~)0) Eur J. Phannacol. [~ap. Ther, 251, 439 447. J. Bier:hem. 191,685 693. 14| Orlowskl. M., Michaud. C. and Molineaux. (',J. (1988) E~i~hcni- [22] Kam, C_-M.. Nigh|no, N, and Powers, J.C. t 19"/9) Biochemistry istr~' 27. 597 trY)2. 18, 3033 3038.

186