Cummings Life S&race Cetbter ...... 7',Ms~M; (II), slow exchange with TIM zrM: and (III). a region where no ...... Mtice relaxatiott times must Iw c*onsidtwd.
,J. Mol.
Riol.
(1983) 163.
Catalytic
G-105
Conformation
of Carboxypeptidase
A
The Structure of a True Reaction Intermediate Stabilized at Subzero Temperatures
Ljepartment
of Biophysics a rzd Theoretical Cummings Life S&race Cetbter The University of Chicago Chicago, Ill. 60637. C.S.A. (Krceived
IO May
Biology
1982)
The structure of the mixed anhydride; acyl-enzyme intermediate of the rsterolptic reaction of carboxypeptidase A-is characterized by applicat,ion of cryoenzymologic. magnetic resonance. and molecular graphics methods with use of the Co2’substituted enzyme and the specific spin-label ester substrate O-3-(2,2.;i.h-tet,ramcthylpyrrolinyl-1 -oxyl)-propen-2-oyl-I,-fl-phrnyllactate. A radial separation of 7.7 pi between the active site Co’+ and the nitroxide group in the low trmperaturestabilized acyl-enzyme intermediate is determined on the basis of their spin-spin (dipole ~dipole) interactions. Application of molecular graphics techniques shows that the only configuration of the substrate t’hat is strrically accommodated bv the active site vields a calculated metal ion-to-nit’roxide distance of 7.8 !I. iteric accommodation of the spin-label in the activr site requires severe torsional distortion around the aliphatic double bond of t#he propenoyl side-chain. Examirlation of the structure of the enzyme: spin-label intermediate reveals that thr distort’ion arises from steric interactions of t’he pyrrolinyl group with the protein at a position that corresponds to the site occupied by the penultimate amide residue of at1 oligopeptide substrate from the site of cleavage. Together wit’h kinetic data showing that, hydrolysis of the spin-label is governed by rat,e-limiting deacylation, the results indicate that geometric distortion of substratts by secondary interactions with thr t~rlzynlt’. in general. is an obligat~ory part of the caat,alytir action of c.arl)ox?-l)er~tidase A. LVhen viowetl with resf)ect to requirements for stereoelertronie control of bond cleavage in tetrahedral adducts of esters and amides (Deslongchamps. 197,5) the results suggest that torsional distortion during c*at)al>-sisresults in rotation around t,he sc>issilebond of t,hc,sul)stratt>. and t,hat this rotation is required to form the mixed anhydride reaction intermediate. Thtw findings further support t)he interpretation that thushydrolysis of esters and amides (aatalyzrd by ~arl)oxvf)el)t’idas~, A proceeds ac*c.ording to similar mechanisms except that formation of the rnixrd anhydride is r;rte-dt,t,c,l,rrti,ritlg in I)el)tide h,vdrolysis whik drac*ylat,ion of the mixed anh,vdridr is r,atc%-limiting in ester, hydrolysis. Additionally. in this study application of thr extension of the theory of the Solomon-Bloembergen-Morgan rquat,ions derived by Lindner ( 1965) for paramagnetic metal ions with 8 2 1 demonstrates that the zero-field splitting of th(x high-spin Co2’ m the Ineta]-sui)stitutc,d enzytne has no significant influence in determination of the relaxation enhancement of solvent prot,ons by the active sitch metal ion.
1. Introduction There are three dist,inc+ regions of’erlz~rlle---sllbstr,ate int,erac+ions ill the ac$i\,tl site of carboxypeptidase A (peptidyl-r,-amino acid hydrolase: EC 3.4 17 2): t,he hydrophobic pocket with Arg145 for binding the COOH-terminal residue of thta substrat,c: t,he hond cleavage site. consisting of the zinc, ion and t,h(l y-c*wrhox~l;ttt~ group of Glu270, at which splitting of the scissile amide or ester bond O(YUI’S: ant1 four subsites near the entrance into the active sitta t,ermed thr sites of secorrclarq substrate recognition These regions are whematic~ally illustratcad in Figuw I Chemical (Waldschmidt-L&z. 1931 : Stahmann c~fa/ 1946) and S-ray diffractiorl (Lipscomb et al., 1968; Quiocho et al., 197 1) studies have sho\z n that t,he hydrophobic. pocket and Argl45 bind the COOH-terminal group of the substrate. st,abilizing its configuration in the active site for bond cileavage, while t,he hydrolytic st’ep itself requires co-ordination of the carbonyl oxygen of t)hc scissiltl bond to the m&al iolr (Felber et al . 1962: Lipwomb et al , 1968). In bond (aleavagSc>.($lu270 a,c:ts as a nucleophile to form a mixed-anhydride, acyl-enzyme reaction int.ermediatrl, alrti deacylation is subsequently catalyzed by a. I’enta-c.o~ordinat,~, mrt’al-hydrositl~~ species to whic*h the wrbonyl oxygen of the scissilr bond is ligated (Ku0 k Makinen, 1982; Makinen et al.. 1982) The catalyt,ic role of the sites of scv*ondaq. substrate recognition is less clear. (:hcmiaal studies \+,ith synth&c oligopt~ptidt~ substrates show that catalytic reactirit)y is wrrelat,ed \vibh t,he degrw of stereospecificity of the penultimate ( P2) amide residut> towards thr fwzyxrtch (Abramouitz et al., 1967; Schechtrr. 1970) The rtwlt,s of model building studio indicate that the side-chains of residues Arg7 I. Argl27, Tyr 19X, and l’he279 constituting the secondary recognition sit)es may bind and assist in thth anchoring ot’ oligopeptide substrates (Quiocho & Lipscomb. 197 I) To sptlc*ifh mow prwist~ly t’h(, catalytic role of these interactions requires detailed st,ru&ural c.harac:tcriza,t.ioII ot’a kinetically competent intermediat’e of the enzyme-c:at,alyzetl reaction Hitherto. a comparable objective has not been achieved for any rnzymcb
Flc: 1 Schematic3 illustration ofthe binding int,erartions of’an oligopeptide substjrat.c ill the actiw sitr of carboxypeptidase A The active site of the enzyme is ~~m~~~~svdof 5 nubsitw (S, to S, and S,‘) located on both sides of t,he c&alytic sit,? (Abramowitz et a/ 1967. Sehwht~er. 1970) The cat,alytic sit.e is I amino ;tc,id wsiclri~~. alrd lbtl with bot.h (‘K’l’l, and TEPOPL by get filtration chromatography at) subzero temperature:: (Makinen et al . 1982)
E’ir.st-tl~‘l.i\.;tti\.t’ k‘.s.r. ;tt)sorl)tiorr slw’tl,:l w’r, wc~ordt~tl \t ith ;I Hrllk(‘l, E:lt%~~f) s1lt’c.t ro mode at 9.4 GHz and 100 kHz frequrnc,v modulation and meter operated in the TE,,, equipped \vith a Varian E-257 variable temperature accessory The sample temperature in thr e.s r. cavity was determined directly with a copper-constantan thermocouple probe attached to a Fl\tkt, X600.\ digital multimeter. ‘l’h(l sample tc~t11l)er;~trlre \vas maintained to within &-I deg.(‘ fof, the saml)t(J votumt~ of O%K~mt in t,hfi cluartx ticwar insert of the viiri;tt+ tempwat~uw acwssot~~. (iii) Xuclear
magnetic
rdsonar~ce
skit&c8
Measurement,s of the longitudinal relaxation rate of’ wat,er protons at 24.3 and 100 JlHz were carried out in the laboratory of Dr il S. Mildran at the Institute for Cancer Research in Philadelphia, Pennsylvania. The pulsed method of Carr & Purcell (1954) was employed \vit,h use of an ShlK Slwciatties PStiOM’ spectrometer equipped with an n.m.r’. I’K-04 c~roswti-c,oil probe. This technique has been discussed by Mildvan 8r Gupt,a (1978) Water proton relaxation rates were also measured by the pulsed Fourier t,ransform n m r method (Pung et al . 1974) with a Ovarian XL-100.FT spectrometer at 100 MHz and wit,h a Brukrr HS-270 pulsed Fourier transform spectrometer at 270 MHz For all measurements the samplr tc~mlwraturr was maintained to within + 1 drg.(‘ lvith a g;rwc,us nitrogtln flo\v systrllr. (iv) Model
builditq
artd computer
yraphics
studirs
Initial assessment of t)he structural relationships of the spill-label substrate in t hr a&\r site of &CPA was made uit,h a LABQUIP (Reading. England) atomic model of the enzyme constructed in this laboratory. Inspection of the covalent reacstion intermediat,e was carrietl out by replacing t,he ester bond of the substrate molecule with a mixed anhydride linkage, between the y-carboxylate group of Glu270 and t,hr carbonyl group of S~(2.2.5.5 tetramethylpyrrolinyt-1 -oxyl)propen-2-oic acid Quantitative evaluation of the binding relationships of (‘l(‘f’l, and TEPOPL in the acativcs site of ZnCPA was subsequently made with the use of the computer graphiw facilities of’ Professor (1 David Barry of the Computer Research Laboratory at Washington University in St. Louis, Missouri The atomic co-ordinates of the enzyme at 2.0 A resolut,ion (Quioc,ho & Lipscomb, 1971) were obtained from the Brookhaven Protein Data Bank This dat.a set also includes the atomic co-ordinates of the inhibitor Glp-L-Tyr in the act,ivr site The atomic numbering scheme and a comparison of the chemical bonding structures of TEPOPL and (‘t(‘Pf, are provided in Fig. 2. The atomic co-ordinates of the substratw \c~Lw generated from molecular fragments and are listed in Appendix A For t,he substrat,e ( ‘f(‘P1,. the (*o-ordinates were calculated for the p-chforocinnamoyl moiety on the basis of standard bond lengths and angles. This fragment was joined to the O,,, atom of #phenyllactic* avid (Cesario & Guilhem, 1975) in the standard configuration of an est,er bond defined by X-ra? diffraction studies (Makinen & Isaacs, 1978) For t,he spin-label ester substrat,e. the VOordinates were similarly constructed from the X-ray determined atomic co-ordinatrs of ~.d.T,.;i.-tetrarnrt~~~~t-~~-carl~ami~lo-l~~rrolin~~-f -0xyl (Turfcy Hc Hoer. 1972) and t,~/Sphrw~l lactic acid, the propenoyl moiety being added according to standard bond lengths and angles The co-ordinates of hydrogen atoms were calculated on the basis of a C-H bond length of 1.00 A and idealized valence angles. For TEPOPL the set of co-ordinates in .\!)f) distinguishing characteristic of TEPOPL is that, the conjugated system of double bonds results in moderately strong absorption wit,h a maximum change in absorptivity at 274 nm upon hydrolysis. The change in absorptivity equal t)o 5190 M-l cm-’ provides a convenient means to determine kinetic. paramrt,ers spectrophotometrically (Koch it aE . 1979). In addition. an important, structural feature of TEPOPL is the present of the unsaturated bond in t,he pyrrolinyl ring This eliminates the possibility of enantiomers that may exhilbit differential cbatalytic. react,ivit>..
as in the
I-vhyrnot~ry~psin
(Fiohr
case of’ saturat,etl d (11.. 1975:
pyrrolidine Ka,uer
si)in-lahrl
& Kedine~~.
srllwtratr
andogs
of’
IW!)).
Kinetic parameters for the hydrolysis of TEPOPL cat)alyzrd by %nVf’A and CoCPA determined on the basis of initial velocity data are compared in Table 2 The values of k,,, sho\v that the spin-label substrate is considerably less reactive to\\ ar& hydrolysis than is ClCPL while the values of K, are essentially identical Nonetheless, the value of k,,JK, of - IO4 M-is’ for TEPOPL is comparable t)o that of other specific peptide and ester substrates of &CPA (Makinen d al , 1970 : Suh & Kaiser, 1976; Bunting ct al., 197-C: Auld & Vallee. 1970a,h. 1971) Figurct 4 illustrates the temperature dependence of k,,, for the hydrolysis of TEPOPL in both aqueous medium and in cryosolvent mixtures The activation energy of 17.3 f ().!I
C’om,parison
of kinetic
parameters
the hydrolysis of TEPOPL by carboxypeptidase A
governing
catalyzed
Substrate
Enzyme
k,,,(s- ‘)
TEPOPL
ZnCPA (‘o(‘PA
3.3*0.1 14+0.1
(‘1cPLt
Z&P-4 (‘o(‘PA
73.0 * 5 11f?o*4
t From lllakinen
an,d CK’PL
et al (1979)
19.8 kcol /mol
I/TX IO3 (Pi FIG: 3 Temperature dependence of the rate ofhydrolysis of TEPOPL catalyzed by ZnW4 The initial substrate concentrat,ion was 1.8 x 10m4 M, and the enzyme concentration was 3.3 x IO-’ M In separate experiments these conditions were established to correspond to the steady-state approximation Thv reaction mixtures contain 0.5 M-sodium chloride buffered to pH 7.5 with 0.01 M-sodium oacodylate The activation energies estimated from the least-squares slope are indicated for each reaction mixture (0) .\quw~s: (0) rthx+nr glycwl/H20 (1 : I ).
kcal/mol for the hydrolysis of TEPOPL is equivalent to that of 16.5 + 0% kcal/mol reported earlier for the hydrolysis of ClCPL in aqueous medium (Makinen et al . 1979). The difference in the activation energy of 19.X k 3.7 kcal/mol observed for hydrolysis of TEPOPL in cryosolvent mixtures compared to that in aqueous medium is not statistically significant for the number of data presented.
temperatures that reaction of both C’lCPL and TEl’OPL \vit,h %nCPX at -A()“(‘ leads to the accumulation of a covalent reaction int,ermediate formed by acylat,ion of the y-carboxylat,e group of ($11~270 (Makinen of nl . 1982) I’ndrr c~ornparahlt~ cqosolvent conditions. addition of TEPOPL to a solution of %n(‘I’A at --BW‘~ ’ results in a change in its ultraviolet spectrum, as illustrated in Figure 4 Ko spe(+raI chhange is induced further if the tightly bound inhibitor I,-~,rIizvlsucclillat,e (Kyers & Il:olfenden. 1973) is added to the enzyme solution af%er the formation of the tit’\\ of I,-~,erlz?-lsuc,c,itlat,c, prior to substratt~ spectrally defined species. hut, addition prcvmts the appearanw of t,he spertral vhangc Affer formation of the spt+4 ri-111~~ identified species at -60°C’. a decwasr in ahsorhanw at, 271 nni attrihtablt~ t,o the hydrolysis of TEPOPL is ol~servrd if the temperature of’ the solution is r(uic.kl>, raised to -WY Comparable observations have Iwett not’tvl \\ ith t’wpc~~t t,o thr IOU tempwature stabilized rea,vtion int,ermediat)e formed \\+t.h C‘N‘l’l, (Makinrtl /I (11 1976.1979) of’ t hts spit) .lalwl sukmtratt~ The absolute absorption spectrum of t’he acyl rnoiriy bound to the enzyme is illustrated in Figure 4(b) The spwtrurn is wnstruc+t4 on t ht, 0.2
2 a
(0)
Products 0.1 --\ Intermedlote ‘\, 0 !~ Substrate ‘1, \
-0.1
\
/’ \
I I’
x-----
I’ ‘\._I’
-0.2
~
0.5 /
fbi
0-.1II-.L ---I 220
240
260
280
300
320
340
C’ATALYTI(’
A(“‘TIOX
BY (IAKBO)iPPF,P’l‘ln.;\SE
A
71
basis of the difference spectrum in Figure 4(a) and the spectrum of the free subst,rat,r (Koch et al., 1979). This spectrum reveals a shift in the absorption maximum from 258 nm to 253 nm. Since the absorptivity of TEPOPL in this spectral region arises from the cross-conjugated system of olefinic bonds with the carboxyl group, t-he shift in the absorption maximum signifies a structural distortion of the substrate in tht> active site upon formation of the covalent reaction intermediate (b) Magnetic
resonance studies of POPPA
(i) Electron spin resonance of the enzyme: spin-label
reaction
in,termedia,te
The (first-derivative) e.s.r absorption spectrum of TEPOPL react,ed in 1 : 1 stoichiometry with ZnCPA at -60°C is compared in Figure 6 to the spectrum of TEPOPL reacted with CoCPA There is no alteration in linewidth when the sample temperature is lowered to - 90°C: the freezing point of the cryosolvent, and a careful inspection of the two spectra shows no detectable difference in lineshape. There is a noticeable decrease in the peak-to-peak amplitude of the nitroxide radical for TEPOPL reacted with the Co2+ -enzyme. On the other hand, the peak-to-peak amplitude of free TEPOPL in the presence of an equimolar concentration of the fret (*(tbiilt ion is unchanged from that, in the absence of ( ‘o’+. Also. when I, brnzylsuwina~~e is added to (‘o(‘P.4 before the addition of the spin-label substrate. the si)ec’trum is identical in amplit.ude to that of an equivalent concentration of free stjin-label. These observations indicate that the change in amplitude in Figure 5(a) is caused by a specific interact.ion between t.he cobalt ion in t,he active sit,e and the
(b)
FIG 5 First-derivative e.s.r absorption spectra of the acyl-enzyme reaction intermediate of carboxypeptidase A formed with TEPOPL at - 60°C The reaction intermediate was prepared directly in q~~arta sample tubes under the conditions described for Fig. 1. Thr tinal enzyme and srtbstratc, concentrations were each 4.5 x 10m4 M, and the total sample volume was maintained at 0200 ml (a) Continuous line, spectrum of the acyl-enzyme formed with Z&PA: broken line, spectrum of the acylenzyme formed with (‘o(‘P.~. In (b) both spectra were rwordrd after, addition of 2.8 x IO-” WI.benzylsuccinate to the acyl-enzyme reaction mixture For (b) the continuous line represents the spectrum of the acyl-enzyme formed with Z&PA: the broken line is the spectrum of the acyl-enzyme formed with CCCP.~
i2
11
C’owtprisort
fi1ro
ET
A I2
of’ thr peak-to-peak signal amp&& of’ ‘I’EPOPL carboxyprptidanP A at - h’O”C’
t Reaction and sryosolvrrrt
An
(‘.
ititportarrt
conditions
ohcar\-ation
rractd
with
are outlined in the legends to Figs 1 anti 5
in
otit’
cxpet’itnettts
is that
:uhlitiott
of’ 1.-hw2yl
after reaction of the spin-label substrate Mith M’PA leads tjo a furthct decrease in the peak-to-peak amplitude (cf Table 3) No changtb is observed undw identical conditions \f,hen TEPOPL is reacttad \vit,h %nC”PA ‘l’hta spectra arc’ compared in Figure 5(b). The lack of change in signal amplitude \\ith ZnCPA shot\ s that the unpaired elect’ron of the nitroxidr group has not been delocalized onto protein residuw nor has the free radical been chemically altered Displacrmmt oft hr spin-label substratr from t,he enzyme by t,-benzylsu~cinat,e or an increase in thv succinate
distance induwd
ht\vwn
arrtplit~udr~. tvac*tion
Thwfow. to of tlir
gror~l)
binding
n-ould
tjlw decease
ititrt~tncdiatr
t.llck inhibitor pcrtuthx~ion
the ttitroxide
1)~ inltil)itor.
rtfmi the
irt the
acTi\-c
in an incwase
in the signal
additjion
~tlz~‘tnr
I~ourid
rcwlt
atnl)litudc~
of I,-hetiz!-lsric,c.iti~t~,
wwtion
tmagttct~ic~ intetwtions
sitr
in tlw of’tlw
th
cwl)alt
iott signal
(‘o( ‘P-l\ : sl)itt-IalA
is inducwl
intet~ttltYlii~tc~ arltl hetn-em
;ttrtl
lwak-to-iwak
rllllst
the nitrosidc~
by t)itttlitig resirrlt
gtwuf)
atitl
of
f’tutt1
it
metal
ion. These changes demonstrate that the inhibitor t,-hettz~lsttc,c~irt~~t,t,does not displaw the spin-label substrate from t,hr actiw site upon formation of the low Since ii non-vovalently I~oiintl tetn~)eratitt~e~st~a~)ilizetl reaction intermediate. substrate would be eventually displaced by a tightly boutlti inhibitor srtcah as benzylsuwinate (Kt - IO-’ nt). t,he results affirm that the spectra in li‘iguw 5 arise only front the acyl-enzyme reaction int,ertnediate ac~vumrrlated at subzero t~etnperatures. As will be demonstrated later, the change in the peak-to-peak amplitude of t,he
(‘.4T;1LYTT(’
A(‘TIOS
BY CAKBOSYI’EP’I’I1>.~sE
;\
711
e.s.r spectrum of TEPOPL reacted with CoCPA can be accounted for by the dipolar interaction between the metal ion and the nitroxide group, modulated by the spinlattice relaxation of the cobalt) ion. The theory of this interaction (Leigh. 1970) lrads t’o t’he relat’ionships : 6H g(H) =
(H-H,)*+8H2
6H=C(l-3cos*
’
(1)
8;)*+SHo.
and
In these equations, g(H) describes the lineshape of the nitroxide radical, SH, is th Enhancement of the relaxation rate of water protons by (‘o(!PA was observed at 20°C in both aqueous buffers and in cryosolvent, mixtures containing SO: 30 (v;\-) perdeuterated ethylene glycol. At lower temperatures, however. the paramagnt+cb contribution of CoCPA in pure aqueous solutions was too small to be quant,itat~rtl accurately. Therefore, the temperature and frequency dependence of’ l/7’,, of \\‘at,cr protons was determined in 50: 50 (v/v) aqueous solutions of perdeuterated cthylcnc~ glycolt. These conditions correspond more c~losely to t’he cryosolwnt mixtures 1 II thv 20’ (I to - 20 ( required for stabilization of the tXWc+iOrl irltrt,lnetliatcl. range, the paramagnetic contribution of CoCPA. i.e. l/f, p. in the cosolvent mixt,rrw increases with decreasing temperature Belo\\ - 25°C. the paramagnetic cwtltr~ bution decreases with decreasing temperature t In stjeady-state kinetic studies, we have demonstrated that ethylrrw glycol oh a norl~cwnpetli IL t. inhibitor of both ZnWA and CoCPL4 while methanol becomes a competitive Inhibitor upon metal io11 substitution (Makinen it a2 , 1979) W e c*onclude that ethylene glyvol is unlikely to vo-ordinat,r to ttw active site metal ion. since its inhibitory behavior is compatible with binding to the enzyme together sith substrate On this basis the observed nuclear relaxation enhancement is due only to water protons of’thv exchangeable solvent molecules in the inner c*o-ordination sphere of the active site metal ion
~‘ATALYTI(’
A(‘TIOS
BY ~‘Al~BOS1’PEP’I’ID.~SF:
A
7.-I
Figure 6(a) illustrates the general theoretical temperature dependence of l/TIP \t,hile Figure 6(b) illustrates the temperature and frequency dependence of l/T,, of \vater protons as defined specifically by equation (6) In Figure 6(a) there are three tlistinc:t regions in tile plot. each cwrresponding to R limiting form of equation (6) (vf’. Dnek. 1972). Region I corresponds to the region of fast chemical exchange ; region I I corresponds to the region of slow exchange; and region III corresponds to a region of no exchange of metal-bound water molecules with the bulk solvent. Comparison of L’igure 6(a) and (b) shows that, the metal-bound water in (‘o(“P=\ is in t,htb fast rsc*hangv rrgion in the dOV(’ t)o - 20°C’ range. ‘l’his wnc4usion is supl~wt.etl I)>. ttw olwrvation that l/T,, is frequency-dependent. Furthermore, in the 20’(’ to --PO (’ On the tjasis of range. d( l/T,,)/d( l/T) or equivalently d(l/T,,/d( 1,/T) 14 ‘. 1wsitive. the analysis of the temperature dependence of l/T,. outlined by 1)wek (1972.19731. tlris c*ontlition (San arise only if 7,e is frec~uenc,?--indel)endcnt.
I
I 3.4
I
/ 3.8
I
J
4.2
l/TxlO”(K-‘I WC; 6 The temperature dependence of l/7’,, of solvent protons in the presence ofcarboxypept,idase A (a) S(,hvmatic rrpreaentation of the theoretical trmprratuw drp~~ndenc~~of thv longitudinal lwototl relaxation rate for a paramagnetic ion system undergoing chemical exchange. as governed by equat)ion (5) (of Dwek, 1972) The Roman numerals designate regions corresponding to (I), fast exchange with 7’,Ms~M; (II), slow exchange with TIM zrM: and (III). a region where no chemical exchange betlreen nwtal~bound and bulk solvent molrcrilr~s O-latcl in e.s.r. vxperirncwts to eliminate the prominent wsonanw cw~t~rihutior~ of the methyl grorlp~+ in cwwlylic acid. The llatir wfi~t~ to measuwmmts made at 11.3 hlHz (0): 100 MHz (0): and %iO \lHz (A).
The
observation
exc*hange
t)hat’ the metal-hound
itt the 20 (’ to -20
(’ twIgr
water
and that
in (‘o(‘l’A
is it1 fast
cherni(.ai
T,~ is tl~c~cIti~~tt(~~-iitdrl)~il(l~ilt
itt t ttv
23.4 to 270 MHz region provides a basis to estimate T-,~of the (‘o* + in Co(‘PA LZ’tb employ the parameters q = It and r = L*‘.7 I A, calculat~rd for the distance bet\+ een t,hc, cdxtlt
ion
arid
txv-definetl
it tJrotoit
of’ it tltt~tal-c.o-ot~tiitl~tt(~(l
ptt~arltetrtx
dfivtive
ill;tpiwtic~
of’ ttloittt~tlt
sttsc*el)til)ilitJ~
stlltlirs
(.T) we
I)>. sttbstittttioil
obtain
tttrtal p
itttlitlo wl)ot~ttd
\vitttar wicl for,
ttiolt~(~tttt~
c~Otll[~l~sc~s (‘d’l’.\
ott
is 4.71 B.M. (Rosentwt~g ut t/l.. l!E;i). :
011 tlttl
imsis
t 11th
I)asis
of’ S
l!l(ii).
(1~l~1Y~llliltl. 01’
12rottr cvluatiotls
‘l’tI,s
t~li~~llt’t
it.
(3) i~tttl
C’A’I’ALYTI(’
AP’I’IOS
BY (‘ARBOXYPEPTIDASE
A
77
systems. Also, in no previous study with macromolecules has the validity of the application of the Solomon-Bloembergen relationships been discussed in the case of the high-spin Co’+ . However, we have demonstrated through application of steadgstate and cryoenzymologic methods that the Co2+ -substituted enzyme is kinet)ically and mechanistically identical to native ZnCPA and that its catalytic reaction is governed by the same rate-limiting step (Makinen et aZ., 1979). The Co2+substituted enzyme is, therefore, the paramagnetic derivative of choice to caharacterize the structure of the acyf-enzyme intermediate. for we cannot be certain that other paramagnetic, metal-substituted derivatives of carboxypeptidase A are mechanistically similar to the native enzyme. Therrfore. in r\ppendix II we have outlined those factor’s that justify the valid&>of this approach for the Co2+ in CoCPA. The most important consideration is thr magnitude of the zero-field splitting (214), f or in the derivation of equation (3) the zero-field splitting is ignored and application of the Solomon-Bloembergen relationships may no longer be valid under the conditions %ID(>T~’ >ws. Tlo evaluate t’he influence of the zero field splitting, we have applied the theory of the enhancement of solvent proton relaxation derived by Lindner (1965) in which the zero-field splitting of paramagnetic ions with ,S 2 1 is explicitly taken into (WIsideration. r\s documented in Appendix TT. the value of 7c estimated on the basis of the theory of Lindner (1965) is equivalent within experimental error to that obtained directly through the use of equation (3). On the basis of our estimate of thr magnitudeofBlD1 (~8 cm-‘)fortheCo 2+ in CoCPA (Kuo, 1981; L. C. Kuo & M. M’. Makinen, unpublished results), the result of the analysis is that the zero-field splitting of the high-spin Co 2f in CoCPA has little influence on f/‘Y,,. Therefore. UY c+onc*ludr that the results of our proton relaxation enhancement studies are talidf~al)l)lied to estimate t,hr spin-lattice r’elaxat,ion time of the (‘0” in ( ‘o(‘P~~. (iii) Estimation of the distancr between thr C’02+ and the n&oxide group in the acylpnzym,e reaction intermediate There are se\-era1 assumptions implicit in application of the relationships derived by Leigh (1970) to calculate the distance between the nitroxide group and a paramagnetic ion In the specific case of the CoCPA : spin-label reaction intermediate. we address t)hem as follows. The 1 : 1 stoichiomrtry of t’he cobalt ion to nitroxide is ensured since only equimolar concrnt’rat’ions of (‘o(‘P.4 and spin-label pster were employed. The absence of a change in lineshape in the spectrum of thta enzyme : spin-label int,ermediate upon a decarease in t,emperature to -9O’(’ shows that motional freedom of t,fle covafently bound nitroxide is not significant at -60 (’ and that the assumption of t’he rigid-lattice approximation has been appropriately made. =Zn additional imljortant considerat,ion is that the flucstuating magnetic tield at the nitroxide-free radical center produced by the rapidly relaxing metal ion may provide a mechanism of spin-fatt,icr relaxation for the spin-labs-t (Hyde (4 ~1.. 1979). These conditions may obtain with use of the high-spin cobalt ion with a short, spin-lattice relaxation time under the conditions t)hat 7,? al)proximates the reciprocal of the electron Larmor precession frequenc,v of thr sf)in-label. (‘omparison of the e.s.r. spectra of the spin-label in the acyl-enzyme formed with the Zn2+- and (‘02+-enzymes shows no measnrabfe change in
I,. (‘. KlJO
7x
ET .-I /.
lineshape. On t,hrt basis of wlationshil)s for the dipolar Hamiltonian t)hat desc.rilw the relaxation behavior of two interacting sf)ins (cf. Hytlr (4 r/l.. 19X). ~II relax:rt,iotl approximate lower limit of’ 8 X lo-’ s is estimat,ed for the spin-lattice time of the nitroxide-free radical with a value for 71e of 1OF I1 s fat, thrl c.ob:rlt ion (Kuo. 1981 ). This \alue is not unreasonable for a slo\v-tumbling. ititt.oxitlt~-f’l,~,~, radicaal in a viscous. tluid solut,ion at -60 (’ ( \Yt~rdun Pt II/.. I972 : IGwd. I!)i(i : Kusumi et t/l.. 19X2). Thew two wnsiderations argw strongly against :I fwr(aeptiblr, cahanga in the relaxatjion mwhanism of the spin-label substrata. On this basis. \\x* cwnclude that t,he dwrrase in signal antl)litudr of t)w spin-labrl in t)tle ac~yl-m~~~n~~~ foimed with (‘o(‘I’;\. a.s illustrated in I%gure 5. reflwts only the. dipolar itiflumc+t~ ot t)he nearby paramagnetic* wbalt ion. Substit’ution of the experimentally determined \.alur of (8.2 & 1.3) x IO lL s fi)r 7,c in tllcl in equation (2). together. with t’he value for (’ of I.376 C: for a S”,, tlec~~~ase nitroxide signal amplitude (Dwek. 1973). Itlads to a c~al(~ulatr4 distjanw of i-7 .A +03 p, b&veen the wbalt ion and the fret radicsal writer of the nitroside grotrf) in with t,lle resrrlts of S l'i1.L' the acyl-enzyrnc> reaction int)eimediate. Together crystallographic studies that define thr general binding relationships of inhibitors and sluggishly wactivc substrates in t)lw activcx site of .Zn(‘P;\ (Quiocho 4 r/l.. I!)71 Quiocho cY- I,ipscomb. 1971 : Rrrs Sr LipswrnIJ. 19X1 ). tlris result iwo\.iti(ss 8 wnfiyut~a~tioti of the slJiti-la1~~~l quantitative parattietei~ to assign ttic binding substrat,e by tnodel building and computer gtaphichs tncthotls. (c) Detwmination
of th.r con$yurat,ion using
oj’ TEPOPL
computrr
ire the acticv sik
c~f %n(‘t’.-l
graphics
For an initial assessment ofthe binding configuration of TEl’Ot’L, the spill-label substrate in its extended trans configuration \vas constructed into thr active site of' Zn(‘P.4 with t’hr IIS~ of mole~rllat~ models. Tile t,tbl.tnin;t] I~-8-I,herl?.lla(.tatf, I,(lsi(]ll(l was superpositioned onto the tryosine residue of the (ily-I,-Tyr inhihit,or, as defined by the set of Brookhaven co-ordinates, \vith the carbonyl oxygen of the scissile bond co-ordinated to the metal ion. With this configuration of the spin-label. strong steric. interactions between the pyrrolinyl moiety and enzyme residues near Tyrl98 and Phe279 prevent accommodation of the substrate molecule. Relief of the st~eric~ hindrance requires rotation around the C=(’ bond of thr propettoyl sidr-chaitt Assessment ofthe accommodation of the acyl moiety of TEPOPL in the active sitr of the covalent reaction intermediate was made by connecting t,he caarbonyl group of the propenoyl side-chain to the y-carboxylate group of Glu270 in a mixed anhydride linkage. For the mixed anhydride reaction intermediate. t,here is similarly st.rorlg steric hindrance with amino acid side-chain residues near Tyr198 and Phe279 t,hat prevent accommodation of the acyl group in its trawls configuration This steric. hindrance is relieved only by rotation around the (‘=(’ bond of t,htB propenoyl sidtachain. If the bulky tetramethyl-pyrrolinyl moiety is constructed to protrude out’ oi’ the active site cleft into bulk solvent, the carbonyl oxygen of the sub&ate no longw remains co-ordinated to the metal ion. In addition, the metal iota-t,o-nitroxittr distance is not less than I2 A Thus, the only configuration of the acyl moiety of’Gw spin-label substrate in the mixed-anhydride intermediat>e that meets the requiw ment of co-ordination of the carbonyl oxygen to t,hr metal ion (Kuo &Z Makinen.
(‘,\TALYTI(’
;\(“I’105
BY (‘AKBC)SYl’EI’TIDASE
Ai
79
1982) also requires rotation around the olefinic bond of the propenoyl side-chain for steric compatibility with the active site. These preliminary model building studies show that the intact spin-label and the covalently attached acyl moiety of the substrate exhibit equivalent steric relationships with active site residues. A stereoview of the extended van der Waals’ surface of the active site of ZnCPS is illustrated in Figure 7. To facilitate the computational requirements of the algorithms employed, only the residues that define the active site surface were employed. These residues were selected by direct examination of the molecular model of Zn(‘P&A%and the listing of the amino acid residues employed for calculating the surface is provided in Table 5. With the active site surface represented by the contours, the substrate is viewed as a wire skeletal figure. In fitting the substrate into the active site, constrained only by allowed, non-bonded interactions with protein residues, the requirement is that the non-hydrogen atoms of the substrate must lie within the contoured surface. Forbidden contacts are recognized by the penetration of a non-hydrogen atom through the contours of the extended van der Waals’ surface. The structure of Gly-r-Tyr in the active site of ZnCPA is illustrated in Figure 8. The wire image representation of the inhibitor: shown according to the X-raydefined co-ordinates, is observed to fit within the contours of the active site surface. With the carbonyl group of the inhibitor co-ordinated to the metal ion, only the oxygen atom of the tyrosine hydroxyl group penetrates slightly through the surface contours. This slight violation of the allowed contacts is probably the result of fitting the dipeptide inhibitor int,o the contoured surface calculated for the free enzyme. Small shifts of the side-chains of Arg145, Glu270 and Ile247 in addition to the prominent movement of Tyr248 from its “up” position to its “down” position upon binding of the inhibitor have been noted in X-ray crystallographic studie:G (Lipscomb et al., 1968: Quiocho & Lipscomb, 1971; Bees & Lipscomb, 1981). Shown in Figure 9 is the spin-label substrate with the t’erminal L-P-phenyllactate residue superpositioned onto the terminal tyrosine residue of the Gly-L-Tyr in the active: site. This diagram illustrates that the terminal L-/-phenyllactate residue of the substrate is stereochemically similar to the COOH-terminal residue of the Gly-L-'l'yr inhibitor molecule. Since the structural relationships for accommodation of the terminal L-/-phenyllactate residue are nearly identical to those of L-tyrosine in tht: Gly-r-Tyr inhibitor complex, as shown in Figure 9, the non-bonded interaction:; Amino
a,cid residues employed in calculation of the extended van der Waals’ surfalc~~ qf the active site of ZnCPA IleO His89 kg7 1 Glu72 Asp142 Am144 kg145 Glu163 Thrl64
His196 Serl97 Tyr198 Ser199 GlIl200 Lru201 LW202 Leu203 Tyr204
Ile243 Thr246 Ile247 l’yr248 Ala250 SerPB 1 Qly252 Gly253 Ser254
lle255 Asp256 Ser266 Thr268 c:1u270 Let127 1 Thr274 PheP79
FIG; 8 Stereo virw ofthe binding configuration of’(:ly-I-‘l’yr N itbin thr van dw LVaals‘ surf’a(,r cwntourof’the actjive site of’%nCPA The inhibitor is represented as a \\ire skeletal figure and is defined hy the IY)ordinates obtained through the Brookhaven Protein Data Bank. The carbonyl oxygen of’thc wissilr bowl is vo-ordinated to the metal ion For purposes of’ clarity. thr wir? skeletal image of’ the protein residaw surrounding the surface of’ the active site arc omitted
(‘ATALYTI(’
A(‘TIOX
BY (‘ARBOXYPEPTTDASE
A
81
FIN: 9 Stereo view of TEPOPL in its extended lrans configuration The spin-label substrate wa>. visually positioned into the active site of Z&PA so that the L-j%phenyllaetate residue was superimposed onto the COOH-terminal residue of the Gly-L-Tyr molecule, as illustrated in Fig 8 The spin-label is represented by continuous lines, while the positions ofthe corresponding non-hydrogen atoms of the GlyL-Tyr inhibitor are connected by broken lines The van der Waals’ surface contours, except for those of the spherical Zn’+. are omitted for purposes of clarity. Difference Fourier studies have shown that I,& phenyllactate binds in the active site in a similar configuration to that of L-amino acid residues with aromatic side-chains (Quiocho etal , 1971; Rees & Lipscomb, 1981)
with this portion of the molecule will not be discussed. The general features of the stereochemical interactions of the L-tyrosine moiety have been described in numerous publications by Lipscomb and co-workers. Tllustrated in Figure 10 is a comparison of the extended trans configuration of’ TEPOPL in the active site of ZnCPA and the configuration obtained by rotation around bonds in the propenoyl side-chain to achieve steric accommodation of the spin-label by active site residues. Penetration of the surface contours by the spinlabel molecule in its tram configuration shows that this configuration cannot be accommodated by the active site. The torsionally distorted spin-label, on the other hand, exhibits no forbidden contacts with residues of the active site. Steric accommodation of the spin-label substrate required a rotation of 89” around the C,,,,--Cc,S, double bond as well as rotations of 115” around the Cc3zJ4,4S) bond and 90” around the C(1SJ---C~2,) bond immediately adjacent to the olefinic bond of the propenoyl side-chain (cf. Fig. 2). Table 6 provides a listing of the non-bonded contacts of atoms of the acyl moiety with protein residues for the spin-label in this distorted configuration. The calculated contact distances in Table 6 show that steric accommodation of the distorted configuration of TEPOPL is narrowly restricted by? residues near the entrance into the active site of ZnCPAt. The contacts of the tram t More recent X-ray studies of ZnCPA with phase refinement at 1.7.5 A resolution have shown that the amino acid residues Ser197, Tyr198, Pro205, Tyr206, Arg272 and Asp273 near the active site have peptide bonds in a cis configuration that were not apparent in the data that define the earlier Brookhaven co-ordinates (Rees etal , 1981) The first two residues are included in Table 5 in defining the active site surface Subsequent re-examination of the possible influence of the remaining four shows that only Asp273 exhibits contacts by calculation with TEPOPL, the shortest occurring with the (:(a.) atom (~4.7 A) of the pyrrolinyl ring and only for the extended Iran.r configuration We conclude that the change in the configuration of the amide group of these protein residues from that defined by the Brookhaven co-ordinates does not alter the conclusions derived here for the assessment of stern, interactions of TEPOPL with active site residues
(b)
spn-label substrate wthm the extended van der M’aals‘ surfaw of thrs actirt~ sitra ‘I’hr spirl~labvl substrate is shown in its extended trarts caonfiguration (broken lirrw) and in a iorslonally tiistort.wl c,onfiguration (continuous lines) achieved by rotation around the (‘,z.,-(Y13 , tlouhlt~ bond ofthe propwo~l side-chain and by rotation around thr adjawnt (*,, ,-(*,2 ) and (‘,3 )-CT,+,, aliphatica Iw~~tls as (Irswitwtl 111 the text (h) f’omparison of the extended fraws and torsionally distorted configurations of 7’EPOl’l. bound in the active site of %nCPA and shown in a similar projevtlon to t,hat in (a) ‘I’hr ac*tiw site surfa(~~ contours aw omitted to illustrate thr structural constraints of 1blac71rg the slain-lwhcsl t),v srll~~~t.f~~)sitlollitl~ of the trrrninal (‘OOH rwidur auto 1,.t\,rosinr as in Fig. $1alltl r~l,-l,~tlitlcti~l~~ of tlw ca~~I~on\-l OK~~VII 111 thra
scissilv
horrtl
to
tlx,
zinc
iota
configuration of the spin-label are too numerous to be listed meaningfully. \\-C. therefore, provide in Table 7 only a listing of the amino acids that, exhibit strong steric hindrance preventing accommodation of the spin-label. Of particular importance is that numerous, forbidden contact,s with portions of t,he polypeptidtl backbone are observed. The atomic co-ordinates of the aepl moiety of t)he t’orsionally dist,orted spirl-label are providrd in Appendix TTT. For this contigrwation. the cAcrllatrd metal iowto-
(‘.ATALYTIC
ACTION
BY CARBOXYPEPTIDASE
A
X3
PIG: 11 Stereo view ofthe extended van der Waals surface of the torsionally distorted configuration ot TEPOPL in the active site of ZnCPA The surface contours were generated for the spin-label substrate in the distorted configuration shown in Fig. 10. The amino acid residues constituting the active site and the immediately neighboring regions are shown as wire skeletal figures The view is from the region correspondmg to bulk solvent looking into the active site cleft of the enzyme By carefully tracing the succession of non-bonded interactions on all parts of the substrate, amino acid residues in van der Waals‘ contact with the terminal COOH residue can be also seen These occur primarily with amino acid residues Asn144, Arg145, Ile243, Ala250, Gly253, Ser254, Ile25.5, Asp256 and Thr26R and are comparable to those observed \Lith the COOH-terminal residue of Gly-1..Tyr The metal ion has been omitted for purposes of clarit?
nitroxide distance is 7.8 8. The calculated radial separation of the spin centers. estimated on the basis of their dipolar coupling interactions, was 7.7 A &- 0.2 8. WC emphasize that the calculated metal ion-to-nitroxide separation obtained by computer graphics results purely from a search for the configuration of the spin-label substrate that is sterically accommodated by the active site. Shown in Figure 11 is a stereodiagram of the molecular shape of the torsionally distorted configuration of TEPOPL in the active site of ZnCPA. In this diagram, the extended van der Waals’ surface of the substrate has been calculated to define the volume occupied by the spin-label in the active site. The diagram illustrates the close steric interactions that restrict the accommodation of the spin-label with respect to both the pyrrolinyl ring as well as the COOH-terminal residue. No protein residues are found within the van der Waals’ volume calculated for the torsionally distorted molecule. as anticipated by the listing of the non-bonded contacts in Table 6. \Ve have similarly calculated the extended van der Waals’ surface of TEPOPL in its extended tram configuration in the active site of ZnCPA to illustrate the strong steric hindrance with residues listed in Table 7. The contents of the cavity of the tram configuration of the molecule can be seen in Figure 12. Of the residues listed in Table 7, those seen within the cavity in this projection are Glu72, Ser197, Tyr198. Ser 199 and Phe279. The numerous forbidden contacts that each residue makes Q ith the atoms of the spin-label moiety of the substrate in its tram configuration are readily seen. Except for Glu72, which serves as a donor ligand to the Zn2+, the other amino acid residues are located near the entrance to the active site cleft of the enzyme and have been designated as sites of secondary substrate recognition
H is69
S,!
(:lu72
’ ‘, 1
A L’g127
NH,
(’
s, 0
s 0
(‘,, 0
Amino acid residue in &CPA
Atom of amino acid residue
Atom of substratet
a (4 4.x 3% 3.4 42
(42
4.9 4.2 4.5 3.2 3.9 4.8
4.9
Phed79
Allowed. non-bonded interactions of the terminal COOH residue of TEPOPL are discussed in the legend t,o Fig 11 These are comparable to those observed with +vvsiw in the inhibitor complex of &CPA formed with Gly-I,-Tyr, that have been described in X-ray studies (Lipscomb et al 196X: Rees 8: Lipscomb, 1981) t Atom designations are according to the numbering scheme in Pig 2 $ .\I1 distances 1.53 1\ are listed
=1r:tive site amin,o
acid residues of Z&PA preventing accommodation con$guration of the acyl moiety of TEPOPL
of
the trans
His69 Arg7 1 Glu72? His196 Ser197t Tyr1981 ser199t Glu270 Phed79 For this listing only amino acid residues with atoms calculated to lie at contact distance ~4.0 a with atoms of the acyl moiety are considered Invariably each residue listed exhibits numerous forbidden 1, RuhinstcGn cut(xl (197 I) have dernonst,rated t’ha,t tjhwcx is ottly otw spitr-latticcb relaxatiort time to cwttsidcr under t)hr c:ondit,iott C&T, < I. hut if w~,,Z 3 I I u o qtitr Mtice relaxatiott times must Iw c*onsidtwd 111 a mac~rornolecwle, the metal iott is genrrally not acwssiblt~ tIo hulk sol\,tlttt , Rtt(l by the slov tumhiing motion oi. 1II(~ spitt-orbit iriterac+oti is modulated rnacromolewk~ The parameter T! no lottgrr has a \\cll dcfitwd physic*al mtaattittg as for aqua-liganded complexes in solut,ion For a nrtmhrr of Mu2 f-cwzyrnt~s, hou e\‘vt’ the values of T,~arc similar and ttot) different f’rom that fhund f’or 1’rw 9111’~ itt alt acjttvotts solution (lory has I~(Y*II dwiwd to vxplaitt the physicsal mc~atring of’ tht) c~otwlatiotl I imt, 7,. for it. t.t~l;ttc~ct analyses iii t~~u~~romolr~c~ular systems tllt’till>
(‘;\‘L’;\LY’I’I(’
r\(‘l’lOS
13). ~‘.~ICLloS~~I’EI”I’II~~~SE
A
O!)
frequency used in the proton relaxation experiments was 270 MHz, the electron Larmor frequency is smaller than the zero-field splitting of CoCPA. Correspondingly, the validity of the application of these relat’ionships to CoCPA becomes questionable Lindner (1965) has extended the theory of the Solomon-Bloembergen relationships to consider the relaxation behavior of nuclear spins under the influence of a paramagnetic relaxant with 82 1. TXVO situations were considered depending on \\.het,her the zero-field splitting is great,er or smaller than the electronic Zeeman energy As the zero-field splitting of the Co2+ in CoCPA is much larger than the Zeeman energy. only the former case will be discussed here For the spin Hamilt80nian #,=/3~H~g~S+S~fi~S with the Zeeman energy term small relative to the S.fi *S term, Lindner (1965) has derived the general relationship given 1)~ equation (A2) t,o describe the longitudinal relaxation time of a prot’on in the vieinit\ of a high-spin paramagnetic ion : 1 T
IM
--
2p2y:
10
1
1,‘,r6 c y+“F(D,u,Jc)
(A2) 1
\\ here F(D,
us, 7,) = l/4
(A3)
and v = ws/lDI and is I02,~ = ws7,, and d T = 1 + XV. To explicitly examine the effect of the zero-field splitting on TIM we have plot’tedt equation (A2) in Figure 18(a) in terms of (TIM~c)-’ versus log ws for values of the zero-field splitting (2101) constant between 0.025 and 25 cm-‘. As IDI increases, l/TIM decreases at) values of WIT, < 1 lvithout changing significantly at ~~7, $1. Therefore, the value of l/TIM calculated on the basis of equation (3) \vith equation (A2) as compared to that calculated depends on the relative magnitude of IDI and ws when the zero-field splitting is larger t,han 7, ’ In Figure 18(b), a value of 8 cm- ’ 1sused for the zero-field splitting energy. andTCisrariedfrom 10-10t010~‘2 s. The curve calculated for D = 0 at 7C= lo- I2 s is also illustrated. The hump in the curves in Figure 18(b) appears at TV> -6 X lo-l2 s. It is, thus, apparent that for values of’~~ -mes. ISac+ case must I)e individually analyzed as the inflllenw of t Ire zero-field spliUing on I /7',. delwmls 011 both the magnitude of T,~ and the niagnitutlr of :‘I). In Table 10 we have compared the values of 7C for CoCI’A calculated directly on the basis of equation (A2) t,o those obtained by equation (3). For the fast exchange region, t)he difference in the values of 7, obtained by the two different equations is not significant. Extrapolation of’these new, calculated values of TCto -60°C yields a value of ( 10.1 + 1.6) x 1O- l2 s for the spin-lattice relaxation time. The corresponding calculated met’al ion-to-nitroxide distance in the acyl-enzyme reaction intermediatt
L. (‘. KC:0
E7‘ .-I !.
FIG: 18 Graphical plots ofthe value 0f’(7’~~~~)~~’ as a fimction of t,he Lartnor frequrn~~y of thr rlwtrou ws The plot,s are generat,ed according to equation (AZ) (a) Plots are illustrated for values of 21111of 2.5. 8.0 and 25.0 cm-’ with TV.= 1Om’2 s The arro\, indicates the Larmor frequency at, which UI,“T~= 1 (b) Plots are generated according t,o equation (Al) with 2101=8.0 cm-’ at different values of TV. ah indicated The broken curve denotes the value 0f(7’~,~,)- ’ assfunction ofw,under the condition 2lDj = 11
TAIH,E 10 Comparison
of values of 7C calculated
(T,)24-3 (Tc)lOO (Tc)270
according
3.35 (335) 2%5 (2.53) 1,:iI (I ,54)
to rpatl;orr
4.01 (3.51) 2% (WI)
(d%)
and rquatiort
(J)
4.25 (349) 3.3-l (3.111
is then 7.9 il f 1.3 A, still in excellent agreement with that. obtained by computjer graphics. Koenig (197X) has also derived relationships to account for the zero-fi& splitting ofparamagnetic cations in solvent relaxation enhancement studies and has applied this treatment to Nln2+-proteins. For the manganese ion. the zero-field splitting is significanbly smaller in magnitude than for t’he high-slain (‘02+ ion.
(‘ATALYTl(’
;\(‘TIOS
BY
(‘AIZBOX~F’F,l’TID;\SF,
.A
(c) The magnitude of the hyperfine interaction Equation (3) may be invalid if the hyperfine interaction between the electronic* and the nuclear spins of the paramagnetic ion is of the same order of magnitude as ws. This is because the nuclear hyperfine term is neglected in the derivation of the lSolomon-Bloembergen relationships. We have determined the average hyperfine splitting of the (I = 7/2) “Co nucleus for a variety of polycrystalline complexes of the high-spin Co2+ of four-, five-, anti sixfold co-ordination. The values range from 12 G (100 MHz) in Co(%picoline-Soxide),(ClO,), to 58 G (545 MHz) in Co(L-histidine), at,pH 12 (Kuo, 1981; L. C. Kuo 8 M. W. Makinen, unpublished results). The hyperfine splitting of the Co2+ in (YoCPA, as estimated on the basis of the bandwidths of spectra of the acyl-enzyme reaction intermediate and of free CoCPA (Kuo & Makinen, 1982), is within this range. Since ws=660 w,, the hyperfine interaction between the electronic and nuclear spins of high spin Co ’ + is insignificant relative to ws for proton frequencies >3 MHz. On this basis, the application of equation (3) remains valid for CoCPA
(d) The electronic g-factor The electronic g-factor enters into the Solomon-Bloembergen substitution into equation (A4) : (p)
= gfqs(As+
I)]+.
equation
only upon
(A4)
where (p) is the expectation value of the magnetic moment of the paramagnetic metal ion complex. This equation is often applied since it is more convenient to obtain the g value from an e.s.r. spectrum than to measure directly the effective magnetic moment of a system. This approach is valid as long as there is little anisotropy in the magnetic moment of the system. Equation (=\4) is not applicable for paramagnetic systems with considerable gvalue anisotropy such as in high-spin Co2+ complexes, under which conditions the geometrical factors related to the electronic spin-proton nuclear spin interaction art‘ complicated and difficult to apply (Sternlicht, 1965). In the case of CoCPA, we have avoided this problem by employing the value of p (4.77 B. M.) determined as the effec+ive rnagnet’ic moment through magnetic susceptibility studies (Rosenberg et (Il.. 1975).
(e) Anisotropic
rotation
Use of equation (3) assumes that the paramagnetic ion undergoes isotropic rotation characterized by a single rotational correlation time. When the para~ magnetic ion is bound to a macromolecule, the overall tumbling time is taken to be the rotational correlation time of the macromolecule. However, internal rotational motion such as that of a segment of the macromolecule, or of the metal ion, if it is bound by only one residue of the macromolecule, may render the rotational motion anisotropic and the rotational correlation time a function of the re-orientation times about different axes. Since &CPA is a single-chained, roughly spherical globular molecule in which the metal ion is co-ordinated t)o three protein ligands (Lipscomb 4
lo:!
L. (‘. KU0
87’ .-l I,
et al., 1968) anisotropic tumbling and internal rotations are not likely to contribute significantly to t,he overall rotational correlation time. Moreover. in order t.cr influence the dipolar interaction between the electronic and nuclear spins. an internal rotation (or the overall anisotropic tumbling) must be sufficiently rapid relative to the other relaxation processes. This is extremely unlikely for macro and spin-spin relaxation molecules containing high-spin Co2+ as the spin-lattice times of the paramagnetic electrons are the dominating contributions t’o th(l correlation times.
(f) Exponential
correlation
tirrws
Use of equation (3) assumes that the correlation functions for the different types of motions, characterized by TV,TV, and 7,e; are exponential. Since the electronic spin lattice relaxation time is the dominating correlation time for Co2+ systems. this assumption only concerns T,~, which has been discussed above.
(g) Point
dipok
approximation
A detailed discussion has been given by Burt’on of al. (1979) concerning the of the point dipole approximation in the application of the validity Solomon-Bloembergen equations. Waysbort & Navon (1975) have shown that the assumption of the point-dipole approximation for Mn2+-proteins results in an underestimate of the metal-proton distance r by
