J. Mol. Bid.

(1991) 220, 401-424

Lysozyme Revisited: Crystallographic Evidence for Distortion of an IV-Acetylmuramic Acid Residue Bound in Site D Natalie C. J. Strynadka and Michael N. G. James Medical Research Council of Canada Group in Protein Structure and Function Department of Biochemistry, University of Alberta Edmonton, Alberta, Canada T6G 2H7 (Received

14 September 1990; accepted 28 February

1991)

A structure of the trisaccharide 2-acetamido-2-deoxy-n-muramic acid-& 1 + 4)-2-acetamidoacid (NAM-NAG-NAM), 2-deoxy-n-glucose-/I( 1 -+ 4)2-acetamido-2-deoxy-n-muramic bound to subsites B, C and D in the active-site cleft of hen egg-white lysozyme has been determined and refined at 1.5 A resolution. The resulting atomic co-ordinates indicate that the NAM residue in site D is distorted from the full 4C, chair conformation to one in which the ring atoms C-l, C-2, O-5 and C-5 are approximately coplanar, and the hydroxymethyl group is positioned axially (a conformation best described as a sofa). This finding supports the original proposals that suggested the ground-state conformation of the sugar bound in site D is strained to one that more closely resembles the geometry required for the oxocarbonium-ion transition state, the next step along the reaction pathway. Additionally, detailed analysis at 1.5 A resolution of the environments of the catalytic residues Glu35 and Asp52 provides new information on the properties that may allow lysozyme to promote the stabilization of an unusually long-lived oxocarbonium-ion transition state. Intermolecular interactions between the N-acetylmuramic acid residue in site D and the lysozyme molecule that contribute to the saccharide ring distortion include: close packing of the O-3’ la&y1 group with a hydrogen-bonded “platform” of enzyme residues (Asp52, Asn46, Asn59, Ser50 and Asp48), a close contact between the hydroxymethyl group of ring D and the 2’-acetamido group of ring C and a strong hydrogen-bonded interaction between the NH group of Val109 and O-6 of ring D that stabilizes the observed quasi-axial orientation of the -CH,OH group. Additionally, the structure of this complex shows a strong hydrogen bond between the carboxyl group of Glu35 and the /I-anomeric hydroxyl group of the NAM residue in site D. The hydrogen-bonded environment of Asp52 in the native enzyme and in the complex coupled with the very unfavorable direction of approach of the potential carboxylate nucleophile makes it most unlikely that there is a covalent glycosylenzyme intermediate on the hydrolysis pathway of hen egg-white lysozyme. Keywords: chicken egg-white lysozyme; carbohydrate/protein interactions; glycosidic hydrolysis; oxocarbonium ion intermediate

iV-acetylglucosamine (NAG) in the bacterial cell wall peptidoglycan (Salton & Ghuysen, 1959, 1960; Jeanloz et al., 1963, see Fig. 1). The best characterized member of the C-type lysozyme family is that isolated from hen egg-white. The complete primary sequence of this molecule has been determined both by protein and by DNA sequencing techniques (Jolles & ,Jolles, 1961; Jolles et al., 1963; Canfield, 1963; Nguyen-Huu et al., 1979). In common with all type C lysozymes, hen egg-white lysozyme (HEWL) is comprised of 129 amino acid residues (Jolles &

OH

Enrymat~c cleavage I I

$ ,

NAM

E

F OH

&AM

NAG

Figure 1. The hexasaccharide composed of alternating 2-acetamido-Sdeoxy-u-glucose (NAG) and 5acrtamido-2deoxy-n-muramic acid (NAM) residues as found in the natural cell wall peptidoglycan of gram negative bacteria. NHAv refers to the N-acetyl or acetamido moiety found in both NAG and NAM sugars. R refers to the lactyl or OCH(CH,)COOH moiety found only in NAM sugars. The 6 saccharide units are thought to bind to successivr sites on lysozyme-labeled A thrdugh F. 1967a,b; Phillips, 1966, 1967). Subsequently, the structure of HEWL has been determined from tetragonal crystals under a variety of conditions of & Richards, temperature and pressure (Kundrot 1987) and from triclinic, monoclinic and orthorhombic crystal forms (Joynson et al., 1970; Moult et al., 1976; Hogle et aE., 1981; Artymiuk et al., 1982; Hodsdon et al., 1990). All of these studies indicate that the conformation of lysozyme is essentially the same under the different conditions and crystalline environments. The overall fold of lysozyme is shown in Figure 2. A deep crevice or cleft containing the active site divides the molecule into two domains; one of them is almost entirely p-sheet structure (encompassing residues 40 to 85), whereas the other is comprised of the N and C-terminal segments (residues 1 to 39 and 101 to 129) and is more helical in nature. The two domains are linked by an a-helix (residues 89 to 99).

The currently accepted mechanism of lysozyme often termed general acid catalysis, was initially proposed from observations gleaned through the modeling of polysaccharide substrates into the active-site cleft of the enzyme (Phillips, 1966, 1967; Blake et aE., 1967a,b). These authors envisaged that six sugar units (termed A through F) of an oligosaccharide substrate could be accommodated within the HEWL active-site cleft (see Fig. 1). This proposal was supported by subsequent action,

chemical

data,

which

showed

that

the

natural

bat*-

terial cell wall substrate is indeed a hexasaccharide residues. alternating NAG and NAM of Additionally, the maximal rates of lysozyme action occur when either (NAG-NAM), or the related chitin hexasaccharide (NAG), are the substrates. The modeled position of the hexasaccharide within the catalytic cleft led to the development of the lysozyme hydrolytic pathway as summarized in

Figure 2. A stereographic representation highlighting the overall fold of hen egg-white lysozyme. For clarity, only the atoms of main-chain and selected side-chains within the active site region are shown.

Crystallographic

Evidence for Distortion

of Bound N-Acetylmuramic

ProposedMechanism

Supporting Evidence

‘lb two acidic r&&es, Glu35 and Asp52. flank the glycosidic oxygen atom that links the NAM residue at site D and the NAG residue at site E The specikity should be for tbe cleavage between the D and E sugars.

- All rqumccs of type< iycozyma bwe a Glu md Asp in analogcur positions to Glu35 and Asp52

Due to the hydrophobic nature of its envkmment in HEWL. Glu35 would be pmtonated at the pH of maximal enzymatic activity @H5). Asp52 would be ionized as it is in a hydqzhilic mvimnmmt

- Qanid modificatim inactivtion.

403

Acid Reference

d Asp52 leads to

PmmLl& Rqwy. 1*I*,d..1%9,1970

1969

- Site specific mutagenesis of GM5 and Asp52 tcmlly inactivates lysozymes ability to cleave tbe natural haasaccbaridc sub&ate, ‘ll~e qecificity of mzymic hydrolysis is for tbe bmd bctwvn the fcwtb and fifth sug.r tits fran the non-reducing md of thenatuml rubstme. pH diffemtce speara of native HEWL and duivativea with citber GM35 or Asp52 ester&d indicate pffi values of 6.1-6.5 for GM5 and 4.7-5.1 for Asp52. - Neutron diffraction to 1.4 A resolution shows a pzak of negative density (ie. a hydrogen) off of the urboxylate oxygen of G135.

An qnimal fit of the hexasaccharide ahmate into tbe active site cleft nquirrs a stetic distortim d the sugar ring at site D into the bigher energy half &air cmfom~ation thereby weakming the bmd beavem atoms C-1 and O-4 of residues D and E (i.e. creating P higher energy ground state).

TheclyatallogmQhic de.tcmlinatioo d I lactone

A positive fne encqy term was asscciated biiding of 8 NAM aaccharidc at site D.

with the

Glu35 acting as a genenl acid catalyst would dmate a pmon to the weakened glycosidic bond between sugar residues D Ed E. Bond d~vpge

is cp~

lesding

10 the

fomwion of a transitian state in which C-l and O-5 OTCpositively cbwged (oxocarba~ium ion). The formation of the wtanium ion is favowed by tbe lowered activatim energy arising fran the bigha ground state energy induced in step (b).

Pord~td..

1974

daivuiveoftcvr~AG)bamdtoritaAu,Dd lyrozyme indicates the most p&able ccafonnation of site D was a sofa or boat Not possible to diitinguish between the two due to low rcsolutim and lad; of refinement.

u-hydrogen isctq effect musumnents on tti NAG, and GlcNAc@(l-4)Glc@OPh showed. change towds ap’ hybridizaticm in the gmmeny of the ructim cnltre between free substrate and the trmsitim state.

ChipIlM & Sharon. 1969; C&6naa> I97I; Rqday# 1967; lmto dI d., 1972

Dahfqtiad.,l%96 sIni* l r al., 1973 Rambebrrg & Kimh. 198

The positive citarge on the oxourrbmium ionis stabilizedbytkiadzedoegatively charged carboxylate group d Asp52.

-

‘he liktimc of tbis o,.ocarb.,,,i,,,i, _ Asp52 ion pair shwld be long mough for tk leaving gmup to diffuse away and the glycosyl aczqot (water) u) diffuse in and mdt the carbmium ion at C-l before the ion-pair Couapses to a covplmt acylal intem~ediate.. The ~@tan cm Gld5 would also be replmished by this water.

I’;;

.I

‘:~~;:~;..~~.‘~;.~~;.~~;.~~~.~~~.~~~.~~~.~~;~ Thecatalytic: e”& is -pk&d k*

., . .I.,.. -;. ::.:.: .,.... -:. .: .: ... ...... .,.. . -_...-;,....,..... ..(,.,. ..:y.,: _.. _.._.. . . -. .: . . . -..: ..,. .... . :... ..\.....:. au 35 D A 0 0

Figure

retention

of cmfiguntion

1 the C-l cwbott.

3. The general acid catalytic

pathway

Figure 3. Although several points concluded from these elegant analyses were later verified or supported by other experimental data (see Fig. 3), some aspects remained largely speculative. Most controversial of these are the proposed steps involving the ground-state distortion of the site D

Quantib of the cleavage of T-Melled disaxhatidcr by HEWL showed the mxticm 99.7% with retention d ccafiguntion.

of HEWL,

Dahiqtit

ad..

19690

ocilltx

as proposed by Phillips (1966).

NAM residue in order to avoid a steric clash with the side-chain of Trpl08, and the electrostatic stabilization of an unusually long-lived oxocarbonium ion intermediate by Asp52. An alternative mechanism for lysozyme action arises from the more general (e-e) glycosidase mech-

anism first proposed by Koshland (1953) and elaborated upon by Sinnott (1987). This mechanistic st,ream has been validated at all steps for a number Escheriehia coli lac% of glycosidases including P-galactosidase and the /?-glucosidase from Aspergillus wentii (Sinnott, 1987). Comparison of the Koshland (1953) and Phillips (1966, 1967) proposals indicates that the fundamental difference between the two mechanisms lies in whether there is stabilization of a long-lived carboxylate-oxocarbonium ion pair between Asp52 and the anomeric carbon atom of the NAM residue in site D (Phillips, 1966, 1967), or format.ion of a short-lived oxocarbonium t,ransition state that, could collapse into a tet,rahedral covalent intermediate (Koshland. 1953). Early on. the modeling studies reported by Phillips and co-workers suggested that the formation of a covalent bond would be prohibited simply by t,hr long distance from Asp52 to the (‘-1 carbon atom of XAM (-3 8). However, there has never been a high-resolution refined structural determination of a saccharide bound at site I> to verify this cblaitn. Furthermore. the existing experimental data describing the lysozyme hydrolytic, pathway. is equally supportive of either mechanistic proposal (Sinnott, 1987). Unfortunately, the extremely difficult, and time-consuming synthesis of lysonyme substrat’es and possible analogues in addition to the complicat,ions arising from transglycosylation effects (Kravachenko, 1967) and the format,ion of non-productive complexes (Sharon & Seifter, 1964: Holler et al.; 19756) have severely limited the kinetic: analysis required to definit,ively choose one mccharristic scheme over another. In order to probe these controversial issues more fully, a trisaccharide (2.acetamido-2-deoxgI)-muramic acid)-/I(1 + 4)-(2-acetamido-2-deoxgu-glucosyl)-B( 1 + 4).(2.acetamido-2-deoxy-d-muramic acid), MGM. was co-crystallized with HEWL and t’he structure determined at 2.5 13 resolut,ion (Kelly et al.. 1979). Due to the steric constraints imposed by the la&y1 groups on O-3 of the first, and third sugar rings, MGM has only one strongly favored binding mode to HEWL in sites B, (1 and 1) (Patt et al., 1978). Although the above crystallographic work provided an exciting first glimpse of a NAM resitlur bound at the catalytic subsite 1). t,he relatively low

Table 1 Summary

of observed

Eysozyme-MGM

intensities complex

No. of

No. 0f

possible reflections

measured reflections

for

the

No. of ltesolution range (A) m-25 2~FrZ~O 2c1.77 1.77 -1.60 1Mb I .49

4285 4077 4049 3995 3963

4280 4069 4023 3960 2874

reflections (I r 20,) P/;,) 97 96 92 87 57

resolut.ion and lack ot’ rt+inrtnent. ltrc~~lntlt~~l :III objective and ac:curat,r intrrpretatfiorl 01’ 1I)(* st.ruc, ture. Figure 3 shows that subt.lr> tfifferc~nc~c~s ill atomic positions can eit,hrr support or IqS;tt tl was similar t.o that describckd previously (Suor~g rl (11.. 1985). Data processing, including local scaling, Lorfwtz and polarization corrections. and merging to a uniyucl data set werta carried out with t,he program pa(bkagfa of Howard rl crl. (1985). For nativr Iysozymr. a. total of 60.233 ohs~~r’vations were measured caorresponding to a unique st,t, of I I ,23F, reflections to I.75 A rrsollltion. Thta overall merging

c~omputrd for all equivalent measurements was W44 and the estimated decay olrsrrvcd during the data collrc%ion IS caomplrx. a total of I19.782 observations were mcasurtad corresponding to a unique set, of 18.923 reflections to 15 A resolution. Table 1 gives the breakdown of observed data with resolution. The overall merging K c*omputrd over all measurements was oG4 and the estimated dera> observed during the data c~ollection was approximately 18%. The structure factor amplitudes for both the native and t.he enzyme+trisaccharide complex were brought to absolute scale using the scale factors determined from Wilson-type plots of the c.orrectcd intensity data (Wilson. 1942). The appropriately scaled data sets had an agreement factor of 0.12 (C(IFMGMI - IF,(I/CIF,(. where IFMoM and lFNl are the structure factor amplit,udes measurrd

Crystallographic from the MGM-lysozyme crystals, respectively). (b) Refinement

complex

of native

Evidence for Distorticm of Round N-Acetylmuramic and native lysozyme

180

lysozymr

120

refinement least-squares Restrained-parameter (Hendrickson & Konnert, 1980) of the native HEWL structure was done using the data collected to 175 A resolution (Table 2). The initial model used was set 2LYM from the Brookhaven Protein Data Bank (Bernstein et al., 1977) corresponding to the 1 atmosphere pressure refined structure of native lysozyme (Kundrot & Richards, 1987). The model consisted of 1000 protein atoms and 19 of the water molecules having B-factors less than 25 8’. Minor adjustments to the initial model and the selection of additional solvent molecules were made from examining density maps IF,1 - IFA and ‘LIFO1- IF,1 electron computed at various stages during the refinement. The manual adjustments were made using FRODO (Jones, 1985) adapted to run on a Silicon Graphics Iris 3030 workstation (Cambillau, 1987). During the least-squares refinement. restraints on bond lengths, bond angles, planar groups and chiral centers for the protein were applied. Restraints on non-bonded contacts were weak and the weights on thermal parameter restraints were relaxed towards the final stages of refinement. ?u’orestraints on conformational angles or possible hydrogen bonds were applied at any stage.

405

Acid

60

-60

+ , 0

-120

I

I

I

!

-60

0

60

120

1 D

$J (deg.1

Figure 4. A plot of C$ ver8u.s tj (Ramakrishnan & Ramachandran, 1965) for the refined trisaccharideHEWL complex structure. Gly residues are denoted by + : Pro b,y 0; b-branched residues by V and all others by 0.

solution and rejinement’of the HE WGMGM complex

(c) Structure

The lysozyme-MGM complex was solved by difference Fourier methods using the resulting fully refined mode1 of native lysozyme. Every precaution was taken to remove all solvent molecules from the active-site region of the native lysozyme model before proceeding. From the remaining set, only water molecules with temperature factors, B, less than 250 A2 were included initially. The refinement strategy for the sugar/enzyme data was similar t.o that described for the native lysozyme. A trisaccharide template was amalgamated into the PROLSQ dictionary using a modified version of COKEXK (Pahler & Hendrickson, 1990). The molecular template for the NAM and NAG residues in the trisaccharide was constructed from data of the crystallo-

Table 2 Rejkement statistics _ ~. .--

__

Resolution range (A) R= ZlFoll~,lPl~,l) No. of reflections (IFI 2 40,) fl (all data) No. of protein atoms No. of sugar atoms

No. of solvent atoms r.m.s. deviations from ideal values root-mean-square error Bond distances (A) Angle distances (A) Planar 14 distances (A) Planar groups (A) (:hiral centers (A3) Non-bonded contacts

(a) single torsions (A) (h) multiple torsions (A) (c) hydrogen bonds (A)

Native 8x&1.75 @I54 10,315 I%9 loo0

Complex

127

80-1.5 0.169 17,386 191 1000 53 139

0017 O-037 @040 0010 o-162

0023 0.043 0.046 0.017 0.179

o-235 O-190 o-236

0241 @232 0238

graphic determinations of the isolated sugar residues (Knox & Murthy, 1974; MO & Jensen, 1978). In these structures, the glucopyranose rings are in the full 4C, chair conformation. In order to ensure that the final refined conformation observed for the D ring was not biased, the following additional tests were done at the conclusion of the refinement. Firstly, the mode1 of the iii’-acetylmuramic acid in site D was changed back to the full 4C, chair conformation with the hydroxymethyl C,,,-O,,, group in an equatorial position and 6 further cvcles of refinement were run. Secondly, with the site D GAM residue in the initial 4C, chair conformation, a water molecule was placed in the density that corresponded to the refined O-6 position when it was quasiaxial. Six further refinement cycles with this as the “initial” model were also carried out. Finally, the site D SAM residue was completely removed from the input co-ordinate set and 5 cycles of restrained least-squares refinement were run. IF,1 - IF,1 and 2)F,,) - IF,] maps were calculated with the resulting phases from all 3 test cases. The co-ordinates for both the native and complex lysozyme structures have been deposited in the Brookhaven Prot.ein Data Bank (Bernstein et al.. 1977).

3. Results (a) Quality of the rejlned structures In all, 41 cycles of refinement were carried out on native lysozyme and an additional 32 cycles on the 1ysozymeMGM complex. Table 2 gives a summary of the refinement statistics for both the native and complexed structures. Figure 4 depicts the mainchain conformational angles of the lysozyme component of the complex in the form of a Ramachandran plot (Ramakrishnan & Ramachandran, 1965). There are six non-glycine

O~~..~l...~.~‘~“‘l’ Kl R 21

Q 41

R '61

DlOl

s 61

” a121-I

b 2 3

-

0 -10 -20

1,

1 .301,,,,,,...,,,,,,, R 21 Kl

I

,

i

Q 41

R 61

s 81

.,,.,,I DlOl

(b) -1 0121

Protein sequence Figure 5. (a) A plot of the mean temperature factors fbr the 1000 protein atoms of the lysozyme molecule in the trisaccharide complex form. Main-chain atoms are indicated by thick lines and side-chain atoms by thin lines. (b) A plot of the differences in mean temperature factors for the saccharide complexed and native lysozyme forms. The AB value plotted at each residue position corresponds to the R value of the complex minus the ZZ value of the native.

residues with $J,+ conformational angles close to the left-handed a-helical region, aL. All the other amino acid residues in the region with C$ from +50” t’o + 120” correspond to glycine residues. There is no other residue in the refined HEWL with conformation corresponding to unallowed regions of $,$ space. Three of the residues with 4,$ values close to the aL region are amino acid types seldom found wit.h positive 4 values, Arg21, Phe38 and Glu57. Arg21 occurs in two successive p-turns, Asnl9 + Gly22 and Tyr20 + Tyr23. The first is a type IIt turn and t,he second is a type I’. Phe38 is the third residue in a type III’ turn that runs from Ser36 to Asn39. Gln57 is the last residue in a t,ype I turn; this t,urn is buried behind the catalytic residue Glu35. Gln57 is probably involved in substrate binding interact’ions in site E. The average temperature factor for the at.oms of’ native lysozyme is 18.3 A’, and for thr at,oms of lysozyme in the sugar cotnplex it is 154 -4’. Figure 5 compares the main-chain and side-chain H-factors between lysozyme and the lysozymet~risaccharide complex. As seen in Figure 5 and from t)he lower average temperature factor of the complex, the whole lysozyme molecule becomes slightly more ordered upon sugar binding. The atoms exhibiting the greatest decreases in thermal t Turns have been classified following Crawford st al. (1973).

mot.ion art’ lhose from rclsidurs ArgtiH fo Ar.g73 and generally those lining the acktivr-iitch calrti l%rlitlr molecular dynamics trajrtcatory simulatiori~ of’ f’r~ix and substratcx-bound Iysozyme had mr~dutic~cl that ;i region of t,hca ac%ivc site (Asp101 to l\la107) had significantly reduced motion. presumably due to interac+ions wit’h the substra1.e (T’osi r/ rrl 19X6. 1989). In general, the most poorly ordered rcJsi(1uc.s 01’ HEWL in the native and complexed slruc+urchs involve the s&-chains of lysinr. argininta or gluta mine residues that, lie on the molecular surf&~. In the nativr structure. t,hcb sidechains c,f’ I,ysl:l. Argl4, Arg21. Arg61. Arg73, I,ys97. ,\splOl_ Argl 12, Gln121. Arg125, Arg128 and t,hcl (‘-t,rrmin;LI residue I,ru 129 were not well defined (tern ptbrat.urfb factors of over 35 A’; weak electSron density for t hr atoms at the end of t.hc side-chains). Tht. sidrchains of ArgBl. Arg73, (ilnl21. Argl25. Argl2X and Leu129 arc similarly affrc*trd in t hi t ri saccharide bound struct~ur~~. Alternative conformations of th(h sidr-chain position werp observed in t.he final electron densit,y for, Val109 in both the native and the complex protein structures. These alterna.tivc conformations (*orr(‘spond to side-chain torsion angles of’ A.‘,“. -- 55’ and 178”. In the nativfl struc*turt>, yallo9 is thsposetl to solvent: noma of the thrrltb possible modtas appears to have an r:nergrtic* advant,agr over t hr oth(Tr. Thus. the equivalencr of thtl t,hrrcJ lobes of rlf~c~trotr density, suggest,ing a hindered rotor-likck mot ion of’ the side-czhsin. may not br surprising. In thus (‘as? of’ the trisacc:hnritle-Iysoz~Irl~~ complex. \‘a1 109 (‘an form hvdrophobic cont.a,c:ts with thfh I) ring only ilr t,he 6.5” or - 55” c*onfortnat ions. In I)ot,h of’ thrxst> cases. onfb of the methyl groups makes contac.1 with the site 1) sugar and the other is solvent -cxI)os~~d. There is no significant st.at.ist ical difl’rrcnc:c~ or hcAt,tt:r tit of’ t+catro,l clensit)y brtwt~c~n thr t ~‘0 possit)t(x conformations u~m1 refinrlmrtit , so we ibssunlf~ f~f1llal orcupall~!~’ for twh. lo subst~ant~iat~~~further discaussion on In ortlc~~ enzymatic. pathways and tictails of i hf, intc3r’ac.t ions betwtlen HEM’I, arld M(2M:I. it is im1)ort;rrll io vcarif\ the c1ualitJ, of the struci urtls ott whicah t hcastsdiscussions art’ I)asetl. ‘l’hrl obsf,rved elt~(*lrorl drnsit ips fi)r Il~r sitl~~-c*hai~~s of rrsitiucts in 1I)(’ ac.1i\:tt-sitch tqiorl of lysoz\.mcs ill the nativca itrId c~orriplt~xt~d fOrnl ill’t shown 'it) li’igur(t A(a) ;trd (I)). wspfv*ti\-ely. Ah surmised f'rom these Figures. all at.orns of thr proposed c*atalytic residues, (:I~35 and AspSP. arc’ clearly tlrfimd in both structures. Furthrrmorfa. Asp&i, .AspM. Gln57. Asn59. TrpBZ, Trp63. Asp101 Asn103 and AlalOi. all residues t,hat direct1.v contact t.hr trisaccharidr XIGM, are also clearly observed in the elect.ron density maps. Thtb low(‘r R-factors (Fig, 5) and rntiancetl elrc.tron densities for XsplOl and TrpBZ (Fig. 6) indicatr> increased order for #lest% rf?sidws it\ th liya.ndrd form of‘ Iysozym~~. Thrrv is no satisfactory w>ty of’ detc~rmining t htt accurscy of the at.ornic (,o-ortlinat.f~s resulting from rrstraincd-l)aranit?trr Itlast -squarrs rf+inemrnt Cj’cx

Crystallographic

Evidence for Distortion

of Bound N-Acetylmuramic

Acid

407

(b)

Figure 6. The active-site region of HEWL in the (a) native and (b) MGM complexed forms. For clarity. only the electron density of selected side-chains is shown. The electron density has been calculated with coefficient 21F,I - IF,1 and t,he phases of the final least-squares cycle. The map is contoured at +@40 eA3.

felt that the approach of (Iruickshank (1949, 1954. 1967). which estimates the co-0rdinat.e error of each atom t.ypr a.s a function of its temperature factor, would be particularly appropriate in t.he present work as it emphasizes the fact that co-ordinat’e errors in regions of low B-factors are small, whereas t,he co-ordinates of those portions of the molecules or solvent with high R-factors are less reliable (see also Read et al.. 1983: Fujinaga et al., 1985). The calculated root-mean-square error for the protein atoms and associated solvent in the native lysozyme structure is O.lti a. and in t’he lysozyme-MGM complex is 0.14 4. In a more general sense. though, the co-ordinate accuracy of a particular atom can be estimated from its observed electron density and refined individual temperature factors. The observed electron density for each of the atoms of MGM and thfk corresponding group average

temperature factors are given in Figure 7 and Table 3. respectively. The relatively low temperature factors and clearly defined elect.ron density associated with the atoms of ring C show t,hat this is the most highly ordered of the three rings, with the buried %aceta,mido group exhibiting H-factors that are as low as some of the most well-defined groups in the protein. The atomic positions of the site CJNAG are thus determined with a high accuracv. On the other hand, t,he solvent-exposed d-acet’ammo group of ring D and the lactyl group of ring B are poorly defined in the electron density map and have the largest B-factors of all the atoms in the MGM trisaccharide (Table 3). The atomic positions of those groups are thus determined wit.h lower accuracy. The acetamido group of ring B and the lactyl group of ring D, although bet,ter defined t,han the site B lactyl and site 1) N-acetamido groups (Fig. 7(b)),

06

06

C8/

(b)

Figure 7. (a) The final, refined enzyme-bound conformation of the trisaccharide NAM-KAG-NAM. All atoms within sugar units B, C and D are labeled appropriately for reference in the text. (b) The electron density map computed in the cycle. The map is region of the trisaccharide using coefficients 21F,I - IF,1 and phases CL, from the final least-squares caontoured at +030 e.k3.

also have relatively large B-factors (Table 3). The atomic positions for these groups are thus less accu-

rate than those of the corresponding groups on ring C. Calculation of the molar ratios of MGM to lysozyme MGM

indicate that in solution there to complex with approximately

dissolved enzyme. It is likely, however, that the crystallization process has selected out only the complexed molecules. This assumption is supported by the very different habit that crystals of the enzyme-sugar complex adopt and by the significant

is sufficient

differences

87%

native

of the

in

unit

lysozyme

cell

parameters

crystals.

relative

The

to

the

trisaccharide

Crystallographic

Evidence for Distortion

of Bound N-Acetylmuramic

409

Acid

Figure 8. An overlap of the refined native lysozyme (thin lines) and sugar-bound lysozyme (thick lines), emphasizing the verv subtle conformational changes that the enzvme molecule undergoes to accommodate the trisaccharide. The result ii al narrowing of the cleft by lpprox. 1.0 A overall

displaces seven ordered water molecules from the active-site cleft of lysozyme (as seen in the refined nat,ive structure). If the occupancy of the sugar were much less than 100o/o, one might expect a contribution from these water molecules to the observed electron density for the MGM molecule in the lysozyme complex. For instance, the NAG sugar bound at site C displaces three water molecules, one positioned 074 w from C-8 (occ. = 1, B = 33), one positioned @39 A from O-7 (occ. = 1, B = 31) and one positioned approximately 2.2 A from C-2 and C-4 of the sugar ring (occ. = 1, B = 41). Examination of the electron density maps of the lysozyme-trisaccharide complex at a contour level close to the noise level of the map (0.15 eA3) indicates there is no observable density corresponding to the positions of the native water molecules that it displaces. Furthermore, the conformation of the site

C sugar is, within experimental error, identical to that observed in previous studies involving NAG and (NAG), binding to HEWL (Perkins et al., 1978; J. C. Cheetham, P. J. Artymiuk, D. C. Phillips, unpublished results). These points, coupled with the well-defined electron density and low B-factors (Table 3) for NAG in site C (see Fig. 7(b)), support the argument that the occupancy of MGM in the crystals is near 1.0. Given the above, it is also unlikely that contributions from the bound water molecules in the native structure influence the electron density of the site D sugar. The structure of the complex has been refined on this basis, since only the B-factors of the MGM molecule were varied; the occupancy factors for all atoms of MGM were kept at 1.0.

Table 3

(i) Protein atoms The native lysozyme atomic co-ordinates were compared with those of the lysozyme molecule in the structure of the complex using a least-squares fitting program (W. Bennett, unpublished results). The r.m.s. deviation for the 1000 protein atoms is 079 A; for the 517 main-chain atoms the r.m.s. difference is 0.28 A. The overlap of the two lysozyme structures is shown in Figure 8. In general, the changes in the molecular conformation are subtle and involve a narrowing of the cleft as the residues and the secondary structural units that line the sugar binding site move in towards the bound trisaccharide. This result correlates well to an earlier theoretical prediction employing normal mode analysis, which concluded that the motions of lysozyme associated with the low-frequency modes (the dominant motions) involved a slight opening and closing of the active-site cleft via concerted move-

Average temperature factors for saccharides bound to sites B, C and D of HEWL Sugar rn0iet.y Ring at,oms ((‘-I, C-2. (I-3, C-4, c-5, O-5) Exocyvlic oxygen atoms 0-I O-3 o-4 Hydroxymethyl atoms (‘-6 O-6 Awtamido atoms (N-2. c-7, O-7, (I-8) Lwtyl atoms (C-9, (i-10. (‘-Il. O-10, O-12)

&m

n to planar the csarboxylate group into significant density ((‘(2)-(‘(3,-0,3)-C~9,~ 89”: (1(3,-0,3)-(‘,9,-(l,10). - 3L’ ‘). In this conformation. the lactyl carboxylate group makes il hydrogen bond to Asn 103 Na2 (Fig. 10(a)). Other hydrogen-bondt>d interactions that involve the SAM residue in site 13are from the AsplOl carboxylate group to O-6 anal from O-6 alId Table 4 (ieometry of glyeosidic linkngest Sugar

4 (“1

M(:bl II (’ (‘- I)

- 108

(:(x:$ A Ii I approximately 3.8 Lk on average. The stacking of a,romatic side-chains on to the sugar rings of pro-

t,ein-bound rnonosacchari(l~,s has betan obsc~rv~tl ill both the high-resolution structural tit~tc,~tttirtatioIrs of r,-arabinosr binding prot~ein and t)-galwc+)se binding protein (Quiocho, 1988. 19X9). It has heen proposed that this type of interaction may br it general

feat,ure

of

protein

savc~haritlc~

hiriclittg

(Quiocho. 1986). (ii) Site Site

C’ ?r’-ncetylglucosaminr (’

plays

cleavage specificity

Fig. 10.

a key role of HEWL

in determining the (Fig. 10(b)). There is

Crystallographic

Evidence for Distortion

of Bound

N- Acetylmuramic

413

Acid

Figure 10. Hydrogen-bonding interactions between protein atoms and sugar residues within (a) sites B. (b) C and (c) D. Lysozyme side-chains are shown as thin lines, main-chain and sugar as thicker lines. Hydrogen bonds are shown as broken l&es. Water molecules are depicted as filled circles. The hydrogen bond lengths are given in Table 5.

a deep pocket on the surface of HEWL into which the 2-acetamido group fits (Fig. 9). This determines which side of the sugar ring faces the enzyme, since any substituent on O-3 is precluded on steric grounds. Thus, only N-acetylglucosamine residues will fit into site C (Phillips, 1966; Blake et aZ., 19676; Johnson et al., 1988). The ring of the N-acetylglucosamine adopts the full 4C:, chair conformation, as was observed for the SAM residue in site B. Similar conformations for both of t)hese rings have been observed in the highresolution, refined structure of a (NAG), complex with HEWL (Cheetham, Artymiuk & Phillips, unpublished results). The plane of the N-acetyl group is approximately perpendicular to the mean plane of the pyranose ring as was observed for NAM

in site B. The conformational angle C(,,-Cc,,-No,-C,,, is 115,“. The 2-acetamido moiety of the NAG ring in site C forms two hydrogen bonds with main-chain atoms of HEWL, from N-2 to the carbonyl oxygen atom of Ala107 and from the main-chain NH group of Asn59 to the carbonyl oxygen atom O-7 (Table 5). In addition, the N” atom of Trp63 donates a hydrogen bond to O-3 of the ring and the N” at,om of Trp62 donates a hydrogen bond to O-6. Hydrophobic contacts are made from the acetamido group on C-2 to the indole ring of Trp108 and the side-chains of Ile58 and Ile98 (Fig. IO(b)). The carbohydrate-prot,ein interactions we observe for the NAG residue bound to subsite C of HEWL are similar to those found in previously

Table 5 Summary

of hydrogen-bonded

contacts in the HE WLMGM

Direct hydrogen bond to protein Atom

Atom

Dist. (A)

complex

Water-mediated hydrogen bond to protein ~~~ Dist. Atom Atom (A)

O-6 O-IO o-12

Asp101 ODl Asn103 ND2 AsnlOY ND2

2.4 2.8

O-6 O-236 O-163 O-163 o-7 O-270

O-236 o- 163 Glyl04 0 Ile98 0 O-270 O-236

2% 2% 2.9 2.9 32 2.8

O-3 O-6 N-2 o-7

Trp63 Trp62 Ala107 Asn59

NE1 NE1 0 N

30 2.9 2.9 2.9

O-6 O-205

O-205 Asp48 OD2

31 29

O-l O-6 N-2 N-2

Glu35 OEl Va1109 N Asp52 OD2 Asn46 ODI

29 2.8 3.2 3.4

o-1 O-181 O-10 O-200 O-200 O-10 O-205

O-181 Asn57 O-200 Asp48 Thr47 O-205 Asp48

2.7 2+ 2.3 3.1 32 24 2.9

‘3.2

OEl ODl OGl OD2

/

;-::g$g

Cti;OH

OH

H3 (a)
srnt refir~tl st rucstjurrh shows t,hat) these residues are from the ?\‘A!If rcGciue in site I) and do not, contribute to the distortion. As shown in Table 3, the at)oms of the 2-ac~rt.amitlo group on NAG have very low H-facators. and ar(’ as well ordered and tightly bound as the surrounding protein atoms. An c~quatorial orirntat.iorr f’ot C(6j-O(6j on ring I) would makcb serious t.oo-c*losc~ contacts with the highly ordered acetamido S-2 ot ring C and so the hydroxymet,hyl group is forctkd into the observed quasi-axial position. The first refinement t,est (Materials and bfethods, section (c)) ~otifirmed that st,artitrg from an initial six c*yc*fes of Irastfull ‘Y1 chair conformation squares refinement resulted once again in it distorted sofa conformation for the NAM ring. The C-6 and O-6 atoms in the cxquat.orial starting positions were 1.0 and I.4 A. respectively, from thr final refined quasi-axial positions. In the srcaontl refinrment test,. a water moft~cuft~ was placed in the density that corresponded to thr axial O-6 atom in addition to assuming t,hat the full 4(‘1 (*hair c.otrformation with (’,(6j-O(6j equatorial was once again t hc initial conformation of thta NAM residue. Thr rlec*tron density in the region of the sit~tk I) NA;L1 saccharide resulting from the addit’ionaf six c*yc~lesot least-squares refinement was virtually indistinguishable from that after the first. t,est and from t ho final 2lF0l - I.r(‘,/ electron drnsitv distribution resulting from the main body of’ refinetnt~nt III t.his s~orlcl test, however, t)hc clquatjorial ( ‘(6)-0(6, hytlrox!~methyl group did not mo\-(1 into its form~~r dc>nsit.y; the extra wat*er molecule remained in its SI artiny position and no new density was obsrrvrd for t htx Co,-Oc,, atoms in an c,quatorial positioli. Finall>., Figure 12 shows the iI+‘“(:I- IIil,( rlec+ron dcansit.>, resulting from the t,hird tc>st. i.e. removal 01’ the, entire I) sugar from the input mod~~l followtld t)y tivc~ cycles of least-squares refinetnent. The tiif%~rc~nc~e density observed at site I) c~lcarfy supports t h(b distorted caonformation obtained from our init,ial refinement and indicates t,hat this result is Ilot a manifestation of model bias. (loffrc*tively, t tie results of these three refinement tests st rongl\- support thra interpretation that t,he sugar ring tn sit’ca 1) is distorted in the ground stat.e towards tht, sof’a or half-chair caonformation. There are several intermofecbufar caontacts between lysozyme and the NA;M residue in site 1) t,hat, would stabilize this distort,ed conformation. One of these “stabilizing” int,eract.ions is the hydrogen bond from the main-chain NH of Vaf109 to the oxygen at’orn. O-6. of t,hr quasi-axial tlydroxymethyl group (Table 5, Fig. 13). This hydrogen

Crystallographic

Evidence for Distortion

of Hound N-Acetylmuramic

Acid

415

Figure 12. The 21F,I - IF,], phases a,. diflerence electron density resulting after 5 cycles of restrained least-squares refinement on an initial model with the sugar in subsite D removed. The final refined conformation of the subsite I) SAM residue we report in this paper is shown in dark lines. The map is contoured at f&15 eA3.

bond would not be so favorably oriented if the site I) NAM residue had the full 4C, chair conformation

(Fig. 13). The conformation of the exocyclic (‘(5)-C1c6jbond of the site 1) NAM residue is gauchegauche. one that is commonly observed in 4C, glucopyranose rings (Jeffrey, 1990). This is also the conformation adopted by the C(,,-Cc,, bond in both of the glucopyranose rings bound in sites R and ( (Fig. 7). The conformational angle C(,,-C(,,-Cc,,-O(,, for each of the three rings is - 63”, - 59” and - 63” (K, (1 and D. respectively). This gauche-gauchr conformation for O-6 in the site D NAM residue probably also stabilizes an intramolecular hydrogen bond O-6-H O-5. The O-6 . O-5 contact distance of 25 .A is short Sor a non-bonded contact but the too-close rontact would be partially offset by the favorable hydrogen-bonding interaction. The gnu& + conformation is the most likely position is for the H-6 proton on O-6 as a tram position

precluded by thr strong hydrogen bond from Va1109

NH t>oO-6 (Figs 10(c) and 13). An additional favorable electrostatic interaction from OE:! of the Glu35 carboxyl group supports this proposal for the hydrogen-bonded position of the H-6 proton. The importance of this hydrogen bond is underscored by the results of two independent studies. Ballardie rt al. (1977) tested a series of aryl fi-1.4-linked oligosaccharides of (NAG),,, with n = 2. 3 and 4 as substrates of HEWL. Kinetic parameters and the pH dependence of these substrate analogs paralleled the kinet.ic behavior of (NAG),. However, compounds of the type (NAG),_, (SAX)-DNP (DNP is dinitrophenol) in which the 2-acetamido 2deoxy-b-n-xylose lacks the hydroxymet.hyl group on (21 were not hydrolyzed by Iysozvme. Even though the equivalent oligosaccharide ( comes to within 2.3 A of C-l of the ?AM residue in site I) wit,hout rnoving 06’ from its hydrogen-bonded position (xl = -78”, x2 = - 171”: see Fig. 16). Thus, it, would seem to be possible to form a covalent bond between a nucleophilic Asp52 and the anomeric carbon atom of the NAM residue. Thr

Crystallographic

Evidence for Distortion

of Round N- Acety1muram.k

Acid

419

Figure 16. The environment of the catalytic residue Asp52. Main-chain and sugar groups are shown as thick lines, side-chains as thin lines and hydrogen bonds as thin, broken lines. Only those lysozyme side-chains that are in the region of Asp52 have been inrluded to ensure clarity. (a) The closest approach of the Asp52 carboxylate group to the (‘-1 anomeric carbon atom of the site D sugar that still retains the surrounding, conserved hydrogen-bonding networks. (b) Shown in thicak broken lines, the position Asp52 would have to occupy in order to provide the most favorahle stereoelectronic configuration for the promotion of nucleophilic attack on the C-l carbon atom of the site D sugar. It is ohvious from this Figure that such a position for Asp52 would require disruption of the hydrogen-bonded network to Asn46 and Asn59 as well as disrupting the surrounding S-stranded &sheet within which Asp52 is anchwd.

implications of this are discussed below. The residues that comprise the hydrogen-bonded network of the platform are all highly conserved in the sequences of C-type lysozymes (Jolles & Jolles, 1984). Thus, this network probably has a common function in the hydrolytic mechanism of these enzymes. (f) 1mplicntion.s of the HEW GMGM complex for th,r catalytic mechanism of lysozym~c (i)

Distortion,

in xitv D

An essential component of the mechanism wherehy HEWI, catalyzes the hydrolysis of oligosaccharides as put forward by Phillips and co-

workers (Phillips, 1966, 1967; Blake et al.. 1967a,b: Imoto et al., 1972) is the distortion of the NAM (or NAG) residue bound in site 1) (Fig. 3). Thus, the formation of the oxocarbonium transition state which, by definition requires the atoms (1-5, O-5, C-l and C-2 to be coplanar, is favored by the pretransition ground-state distortion of the sugar ring in site 11 into a twist-boat, B,,,, or half-chain conformation. A variety of methods have been used to procure evidenee for substrate distortion in the lysozymr mechanism. Early on, free energies of a,ssociation of a variety of sugars binding to the several subsites of HEWI, were determined by fluorescence-titration experiment,s (Chipman et al.. 1967. 1968; (‘hipman. 1971).

:~Iso. kitkit. ;LI~~I~M~s of’ t.hts t*ltA;L\riLpth t’at.tLa tbt’ ii \-arit~t.~ of oligosnt~t:h;tritlrs were ust~l to J)robe hind irig iL#itlitit~s of the subsites (Kupley ct (II.. 1967: lmoto ~1 ~1.. 1!)7%). These two studies t:onc~Jutletl that’ the binding of sugar residues to sit)ta I) \\‘;I$ t~tic~rgrtically unfavorable by approximat,ely 3 t,o 6 kc#mol (1 (*al = 5.185 .I). Tn both studies it was c~ont~ludrd that grountl-stat,cb distt&on of thcb sac’charide in sit,{1 1) was the t*aus~. HowcJvckr. more recent studies att,ribut,e the maj0rit.y of the unfavorable binding free energy t,o the presence of a lactyl group on t.he sugar bound in site D (Schindler. 1977: SfY below). Yore direct st.ructural data for the distortion at sit,e 1) has relied on the crystal structurr of a cmplrx between (NAG),-NAL and HEWL (Ford rt 01.. 1979). The &la&one of NAG was bound to site I> and this transition-state mimic provided strong rvidrnc+r that site D favored sugar rings with sofa or K t*onformations. Such conformations bring the C’i:$)(6) hydroxymet,hyl group into a quasi-axial t~onformatittn (Fig. I I) and t.his was t.he observed t.onformation of the &lactone. linfortunat.ely. this st,ructure determination was at a relatively low rest)Iution (2.5 A) and it has not been refined (see abovr). A number of other studies do not support the idea of ground-state distortion of the saccharide bound in sit,e D. These studies are often cited as evidence t,hat there is no distortion of the D ring (e.g. Post & Karplus, 1986). The initial report of the structure of the complex of HEWGMGM (Kelly et al., 1979) t*laimed that there was no distortion in the D ring. I’nfortunately, that< analysis was also based on an unrefined. relatively low-resolution (2.5 A) tarystal structure. We have shown with the present study that t.his initial interpretation was incorrect. The set~ontl structural study, often quoted as not supporting distortion of ring is an n.rn.r. analysis of thr binding of MGM to HEWL in solution. Patt it nl. (1978) did not observe any change in the tboupling constant between CC,,-H and C&,-H of t’he reducing sugar NAM when it was bound to lysozyme. These authors measured the coupling constant between H”; H, and found that, its value (2% Hz) t~orresponded to the a-anomer (axial-OH) of NAM. Nonetheless, as discussed above. sec%ion (ta)(iii), and depicted in Figure 13, thtl cr-anomer of t,he NAM residue in site D cannot be accommodated in the Jjresent crystallographically determined NA&l-binding mode due to the potentially extreme st,tbrit: conflicts with Asp%. Theoretical chemical approaches have also attempted to evaluate the importance of strain in the catalytic hydrolysis of oligosaccharides by H EW I,. i:onformational energy calculations performed on oligosaccharide substrate models “binding” in subsites R, C and I> suggested that they could do SO without distortion of the ring in site D (Pincus & Sheraga, 1979, 1981). A similar conclusion was also reached by Levitt from energy minimizat ion calculations done on a model of a hexasaccharidr “binding” in all six subsites (A through F) of

r),

HEL\Tl, (l,evitt, 197-C: \Varshel CV I,t,v~tt.. 1!)70). Although t,hrse IWO studies stressed t.hrs poss~h~llt~ of the saccharide hindinp in subsittb I) in a11 titltiistorted conformat,ion. t hr c*alt~ulatiotis JJrotillt.c*d other t~t~t~ft~rmatitms of this ring that wt’rt’ (list ortchtl from thtb full t.hair and let1 to ht,Lter. intt’rat~lionh with the t’nz?me and had t~otnparablt~ cotriJJutt~t1 hinding rnergles. It is rxtremel,v dif%c~ult to (IitY~rt~rrt,iat,r accuratel~~ among the many Itm~t~ntq,v strut’t.ures J)roducetl ilt thest, t~alculations. t~spc~c~iall~ in light of the approximations rcquirrd in ortkr I hat present t*omput,ationat engines can perform t httrn in a rcasonablr t.imr. The lack of solvent. in t,htb IY,IV~IIS tions. the tlchat&r nature of’ t ht. pott~tltial f’~~nc tions and t hc lack of atntnuratr rrtint~l [)lX~~)OSiLlof an altrrnativc pathwax in which t hc t~ntlot~yc~lit~ OC5,-C,,, bond of ring I) is broken initially. All.hough this is an interesting possibility. t,hrrtl ;Lrt’ marry uncert,ainties in such a pathway. It is not at iill (*Ieat how l:hr st~cond subst,rat,e of the reathl ion. H ,O. t:ould approateh thr anomerit~ t~arhon at ant ( ‘-I of ring I) Jjrior to the departure of’ t hr leaving groul) sa.c:charide in site F: it) order to t~~mJ)l~~ir~ t trtl hydrolysis. The original modeling of a hexasacoharitlt~ in t.hth binding cleft of HEW], suggested that thcx hytiroxymethyl group on C-5 of the residue in site 1) was the likely source of ground-state distortion in thth suhst’rate (Phillips, 1966. 1967; Blake d crl., 19670.6). a detailed comparison of unitary Subsequently. binding free-energies derived from measured bintling const*ants for a series of oligosat:t+aridt~s i tnplicat,& t hr unfavorable binding free-cnerg,v ot + 2.9 kt*al/mol to t.he O+lactyl group of’ muramit. acid whrn t’his residue was in site D (Schindter it nl.. 1977). K8elat,ive to the trisaccharide. NA(; N\‘AhINAti. that was assumed to bind in subsitrs A. IS and (‘, an addit,ional ?jAC residue in site 1) enhan& thr binding hy a factor of 2.9 and an additional x.4X residue in site I) enhanced binding by 1.5 timrts. On the other hand. the additional NAM residut, in sittl I) destabilized t’he binding hy a factor of 7.X x IO ‘. The unexpet%ed finding that ,V-at:etyIglut~os;rrninr and Il;-a~etylxylosamine both have favorahlf~ hintling frrr-energies of sitnilar magnitude for sit,ta I) t-an be rat,iunalizrd itr terms of the NAM-NAt i--NAM binding we have observed in this study. Favtlrablt~ interactions are possible for the XAX residue as it Jacks bot.h the O-3 lact’yl group and t,ho (1-5

Crystallographic

Evidence .for Distortion

hydroxymethyl group of a muramic acid residue. Thus, in this case, intimate contacts t’o residues on the surface of site D in lysozyme are possible without invoking distortion of the saccharide ring. It should be kept in mind, however, that saccharides with N,4X residues in site D are not substrates (Ballardie et al.. 1977). For a NAG residue in site U. it is possible to position an undistorted N-acetylglucosamine so that there are no unfavorable contact’s between C-5 hydroxymethyl group and the Sacetamido group of the NAG residue in site C and no unfavorable contacts between atoms of the sugar ring and the residues forming the platform (Asn59. Asp46, etc.). This is because the atoms that comprise t’he O-S-la&y1 group are not present’ in a NAG residue. This orientation can be achieved by rotations about the glycosidic bond joining the residues in site C and site D, and results in a fa,vorable hydrogen-bonded interaction between the hydroxyl group a,t the O-6 atom of NAG in site T> and the amide nitrogen atom of Va1109 while maint,aining the undist’orted NAG conformation. Such an orientation is not possible by simple rotations in the case of NAM-NAG-NAM because of the presence of the bulky 0-3-lactyl group in the NAM residues. Even though a NAG ring can be modeled in an undistorted fashion in site D: there is strong evidence t,hat it is accommodated better in a conformation that more closely mimics the transition state. The tightest binding tetrasaccharide known has the Sacaetamidoglucosamine la&one (NAI,) in site I> (Secemski & Lienhard, 1971: Secemski ef al.. 1978). With the sofa-like conformation of the ring (C-2. (‘-1, O-5, (‘-5 caoplanar) and C,,,- 0 C6)in an axial orientation. thf> tactone is a good mimic of the proposed transition st from the 4(:, chair towards a sofa-like conformation include those with the hydroxymethyl group on C-.5 and those with the lactyl group on O-3. We envisage that, as a NAM residue approaches site 11, two close contacts wit.h residues t,hat form the platform (Asrr46, .4sp48, Ser50, Asp.52 and AsnFi9) would form. Thus, as the favorable binding interactions with Iysozyme and the other carbohydrate residues form. t.he lactyt group moves int)o the observed (*onformation between ring (’ and ring 1) by rot>ations about Short contacts between (1-l 1 (‘(3,-o(3) and OC3J-(~(9). of t.he lactyl group and O-5 of NAG in site (’ rnake the observed conformation relatively unfavorable but presumably still of lower energy t.han t.hr one that would occur if the lactyl group remained in a conformation like that, in the isolated a-anomer described by Knox & Murthy (1974, see Fig. 13). Thus. the present structure confirms that the O-3 lactyl group on t,he NAM residue in site D probably contributes to the unfavorable binding energy of NAM in this site (Schindler et aE., 1977). The well-defined electron density (Fig. 7(b)) and the relatively low B-factors associated with the atoms of the NAG residue in site C (Table 3) are consistent with this site atone providing approximately one-half of the binding energy of a tetrasaccharide (Schindler et (II., 1977). The Sacetamido group is rigidly held by hydrogen bonding and hydrophobic int*eractions (Fig. 10(b)). If the NAM residue in site 11 were to bind with the full 4C, conformation and the CH,OH group on (‘-5 in an equatorial position, there would be too-close contacts made with the relatively rigid Zacetamido group on ring (‘. Thus, the hydroxymethyl group adopt,s t’hr quasi-axial orientation of the distorted sofalike conformation that we observe for the NAM residue in site 1). Additional favorable st,abilizing interactions for this distorted conformat,ion are realized by the hydrogen bond from t,he amide NH group of Vat109 to the hydroxymethyl oxygen atom O-6 (Fig. 10(c)). The binding mode for saccharide rings in site 1) that we have described could explain the results of several lines of research. The very tight binding of (NAG),NAI, (Recemski & l,ienhard, 1971: Secemski et al.. 1972; Schindler d ~1.. 197i) compared with t,hat of (NAG), is due to the existing sofa-like conformation of the la&one and the resut,ant quasi-axial conformation of the CH,OH group tha,t allows for its formation of a favorable hydrogen bond to the NH group of Vatl09. The importambe of the site I) CC,,-OC6)interactions to l.ysozyme in the t,ransition state of the reaction are also emphasized by the results of Ballardie et a,l. (1977). Their analysis of the kinetic parameters that resulted from the hydrolysis of p-nitrophenyl groups from a series of NAG and NAX containing oligosaccaharides showed that there must, be an energetically favorable interaction between site 1) of HEWL and C(6)-O(6, in the transition state to account’ for the observed trend in k,,JK,,, values. Our results suggest that lack of hydrolytic activity when a substrat,r with a NAX residue is bound in sit,e 11 is

---. tlucb to the lack of the hydrogen-bonded interaction between (:C6J-OC6I on the substrat,e and the amide nit,rogen atom of Val109 with concomitant loss of those interactions that help to stabilize the sofa-like conformation of the transition state. Finally. t’hc previously mentioned study in which a lysozymr mutant that had a Val109 Pro substitution (i.e. a prolyl residue in position 109) also resulted in a cbatalytically incompetent enzyme (Tnaka rt al., 1990). The cyclic imino group of a prolyl residue precludes a hydrogen bond from the enzyme to the quasi-axially positioned hydroxymethgl group. Thus, the sofa-like conformation of the transition state of a good substrate is not stabilized by the Val109 Pro mutant. With the distorted, transition-state-like geometry of the NAM residue in site D stabilized by the many interactions we have described for the ground-state structure and the carboxyl group of Glu35 forming a hydrogen bond with the glycosidic oxygen atom of the substrate (Fig. 10(c)), the groups are almostj perfect,ly oriented for the transfer of a proton from Cl1135 to the glycosidic oxygen atom. This proton transfer is assisted by the large elect’rostatic* field across the active-site cleft (Dao-Pin rt al.. 1989). This field results from the charge distribut’ion in the enzyme as a whole and could contribute as much as 9 kcal/mol towards the catalytic rate enhancement. This electrostatic component would assist. in thr charge separation during the catalytic3 cleavage of of the C(,,p OC4,) bond that results in the formation the oxocarbonium ion intermediate (Fig. 3(c)). Small shifts of the atoms of the NAM ring place (J-5, O-5, C-l and C-d in a plane with t,he (1C6j-OC6,group quasi-axial, a conformation that was imposed by the binding t’o site 11. Since the potential barrier for such a ring distortion is approximately 6 to 10 kcal/mol (Amt. B Brown, 1967), the enzyme has a profound role in catalysis by inducing and stabilizing the conformational change in the substrate. Th~h oxocarbonium ion intermediate is stabilized bv t’hr proximity of the formal negat’ive cbharges on i:lu35 and Asp%. Tn turn, the negative charge OII the Clu35 carboxylate group is stabilized by the hydrogen bond from the NH group of Ala1 IO and by t,he fact that it. is positioned at the positive pole of a helix dipole (Val109-Arg114). The negative charge on Asp52 is stabilized by hydrogen-bonding interactions from the amide side-chains of Asn46 and Asn59. Following the cleavage of the glycosidic bond, t,he residues in sites E and F are free to depart and the glycosidic oxygen at’om would be replaced h\- a water molecule. The collapse of the oxocarbonium ion with the water molecule could t’ake place in a concerted fashion with the departure of the lC,F dimer product. The water molecule should be positioned appropriately for nucleophilic attack of the OH ion on the carbonium ion C-l of the sugar residue in site 1). The negat)ively charged carboxy late group of Glu35 remains appropriately positioned to abstract the proton from the water. The electrostatic field that assisted in the charge separa-

lion during the general ~Lc,iti-c~at,al~z(,(~ part trt’ t h(b reaction \voultl most likely bc revc~rsc~tl in I trc, prta sence of the oxocarboniurn ion on the irltl,t’trlc,tli;ti(~ and the negativrl~; charged carbosylat,r grotrp (#I’ Glu35. The proton IS abstrac+ed frorn t.tre rntrrc’ basic, syn dire&on on t,he carl)oxylate group of (~lu3.5 (Gandour, 1981). This direction of’attac*k by t hci 011 ion results in retention of configuratic,n of t htb c*qu;~t#orial /I-anomer of the resulting /J-r+rriuraniic~ ;icGl in site I) (Dahlquist rt /I/.. 1969c(). (ii) Ix thwe II c:or&nt

gl~ycos~y1+nz~yrrrr intrrrt~rdiatu!

There has been muclh wr‘it’ten on the mrc~hanisnl of HEWL and whether or not an oxocarboniurn ion intermedia& st’abilized by the proximity of t hr negatively charged carboxylate group of LAsp5% is feasible (Sinnot, 198’7). The mechanism involving covalent, caatalysis through a glycosylLenzyme inter.mediate was first. elaborated by Koshland (1953). All of the necessary rlement,s for hydrolysis cGc/this pathway seem to br in place in the lysozymc syytern. Tn addit,ion, there are several (e -+ e) /I-glycostdases that almost c*ert)ainly catalyze t.he hydrolysis of substrates through iI c~oralentl,v at t.ac~hrcl infermediat,tA (Sinnot. 1987). \z’t> tlil,\r c~xplorc~tl thch possibility of forttiing a covalrntly attac*hetl glycosyl enzyme through tht* formation of a bond from Asp (!OO62 (the nucleophile) to thtx (‘-1 atom of’ the oxocaarborriurn ion intermediate. As we have pointed out in scxct.ion (c)(ii). it is possible to adjust. x1 and x2 of’ i\sp52 to allow a (+)stb approach of 2.3 LA between t hr 0”’ atom of Asp52 and t,hr anorneric~ caarbon atom of the SAM r&duck in site I) (Fig. 16(a)). Howtsvtbr, an 0.8 .4 rnovoment is additional approximateI> required befor? a covalent, bond can be f’ormetl. This movement is analogous t,o that) required in the, formation of thr covalent tet)rahedral intertnrcliate on t,he cat~alytic pat,hway of thr serinc protrinases (James Pt al.. 1980). In that. pathway. the rruc+leophilic alkoxidc ion, (‘HZ-O 1 of Serl95 moves int’o position as the carbonyl caarbon at’orn of t.he hcissilr bond approaches and takes on tetrahedral c*harac*ter. The resultant covalent (1%) bond is of the ordrr of 1.5 ‘4 in Iengtjti. Figure 16(a) shows the possible direcation of approach of the 0”’ atom of Asp.52 to the anomeric* carbon atom (‘-1, =\sph% 0 ” is relatively tixt~tl in its hydrogen-t)onded position to t,hcb platfortn. Thus, unless the ti~tlrogeri-t)ontleti interactions to Asn4B and *&rr59 t)hat maintain this position itt’t’ disrupted. the carboxy group cannot approach mart’ chlosely than 2.3 x to the (‘-I atom of NAZI. It is not likely that the NAM residue will rnovo rnucah mart’ towards Asp52 eit’her. as too-short non-bonded (‘otlt,acts would then occur between the hydrophobic. side of the rtng ((‘C3j- H atld C:(z)-H) and the rtbsidues of the platform (Fig. If)(c)). Thus. thtx distance of closest approach appears t,o be approximately 2.0 il. A src~~rrtl argument iLpillst forming a cY~v;tlentt~ attached gly~)syl intermediate corntbs f’rorn t hc, direction of approach of the 06* atom of’ Asp.52 in forming the hypothetical acylal intermediatt~. The

Crystalloyraphic

Evidence

for

Distortion

formation of esters of carboxylic acids takes place almost exclusively from the syn direction rather than the anti as a result of the stronger basicity of the carboxylate nucleophile in this direction (Candour, 1981). It can he surmised from an examination of Figure 16(a) that the syn lone-pair orbitals on the Asp52 carboxylate group are not favorably oriented for forming a covalent acylal bond to t,he (‘-1 atom of the n sugar. The direction of approach of thta carboxylate oxygen atom 06’ (i.e. ( ~y-062 (‘-1) is 170” rather than the most favorable angle of 120”. Tn addition, the most favorable (*onformation for t’he acylal adduct is that shown in Figure 16(b). It would not be possible to achieve such a conformation wit,hout completely disrupting the observed inteiactions formed by Asp52 and dramatically changing t,hr conformation of the supporting st’rands in f.he B-sheet. Although present techniques in X-ra!. crystallography cannot) give information about lifetimes of intermediates or about how rapidly enzyme-catalyzed

reactions

take place,

the results

of this most

powerful technique provide details of the relative disposition of atoms in enzyme-substrate analogue complexes that must be taken into account when caontemplating various reaction mechanisms. The newly emerging Laue diffraction techniques give promise of determining the structures of active enzyme---substrate complexes in the crystalline state (Hajdu $ Johnson, 1990). The results of the present crystallographic analysis of the complex between HEWI, and the trisaccharide MGM are most consistent with the general acid hydrolysis mechanism originally suggested by Phillips and co-workers (Phillips, 1966, 1967; Blake et al.. 1967n,6). Tt also shows

how

important,

it

is to determine

crystal

st.ru(+turrs of enzymes and their complexes at as high a resolution as possible and to refine these strucl,urrs to the limit, of the available diffraction data. Dr Brian Sykes kindly provided the sample of NAM-XAG-YiAM. We thank Stanley Moore for growing the crystals used in this study and for assistance in data caollection. Dr Anita Sielecki provided advice during refinement stages. We thank Dr J. Cheetham, Oxford, for sending us a copy of her paper prior to its publication. Mae Wylie expertly and patiently dealt with many revisions of the text. This work was supported in part by the Medical Research Council of Canada. E.C.J.S. is a reoipient of an Alberta Heritage Foundation for Medical Research Studentship.

oj Bou.nd

N-Acetylmuramic

423

Acid

Bernstein. F. C., Koetzle, T. F.. Williams, G. J. B., Meyer, E. F., ,Jr, Brice, M. D., Rodgers, ,J. R., Kennard, 0.. Shimanouchi. T. & Tasumi. M. (1977). .I. Mol. Hiol. 112, m5542. 1Uake. C. C. F., Koenig, D. F.. Mair. G. .4.. Sorth, A. C. ‘I’.. Phillips, D. C. & Sarma. V. R. (1965). ,VaturP (London),

206, 757.-761.

Blake. (1. (‘. F.. Mair, G. A., Xorth. A. (‘. T., Phillips. I). (‘. & Sarma. V. R. (1967a). Proc. Roy sYoc. ser. B, 167. 365-377. Blake, (‘. C. F.. Johnson, 1,. pu’.. Mair. G. A., Xorth. A. (1. T., Phillips, D. C. & Sarma, V. R. (t967h). Proc. Roy. Sot. ser. H, 167. 378-388. C. M. (1987). Documentation .for TOM, (:arnbillau. Mnlrcular (Graphics Prograwl for the IRIS. CRMC2, (:ampus Luminy. Case 913, Marseille, France. (‘anfield, R. E. (1963). ,I. Biol. Chem. 238. 2698-2707. C’hipman. 1). M. (1971). Biochemistry, 10, 171.5. 1722. (‘hipman. 1). M. & Sharon, N. (1969). Sciun,cP, 167, 454-~469. (‘hipman, 11. M.. (irisaro, V. 8: Sharon. S. (1967). .J. Niol. Chum. 242. 4388-4393.

(‘hipman.

1). >I.. Pollock,

.J. Hioi.

(‘rawford.

J .I. & Sharon. N. (1968).

(Ihem. 243, 487-491,

.I. I,.. Lipscomb,

W. M. & Schellman.

(1. (:.

( 1973). f’roc. fiat. Acad. Sci., I’.N.A 70. 538-592. (‘ruic.kshank, I). \Y. J. (1949). Acta ~‘rystallogr. 2. 65-W. (‘rllic,kshank, I). \V. ,J. (1954). Actn (‘r?ystallogr. 7.

5 19%523. (‘ruicakshank, I). b’. .J. (1967). In International Tables for X-my (‘rystallography (Kasper. ,J. S. bi J,onsdale, K., eds). vol 2. pp. 318-340, KynocBh Press. Rirmingham. England. Dahlquist, F. W.. Borders, (‘. L.. Jacobson. (:. & Raftery, lf. il. (1969a). Riochemistry, 8. 694 699. Dahlquist. F. IV.. Rand-Muir. T. &r Raftrry. M. A. (19696).

f%och,emistry,

8, 4214-4221.

Dao-Pin. S.. Liao. I>. & Remington. S. (1989). Proc. :Vnt. .3cnd. Sri.. t3.,4. 86. 5361-5365. Ford. I,. 0.. Johnson. 1~.K., Machin, I’. A.. Phillips, I). (‘. & Tjian. R. (1974). J. 1Mol. Biol. 88. 349 $71. Fujinaga. M.. ljelbaere. L. T. ,J., Brayer. (:. 1). & James. 11. IT. (:. (1985). J. Mol. Hiol. 183. -1-79501. Gandour. It. D. (1981). Hioorg. Chem. 10. 169. 176. 29, Hajdu, J. & Johnson. L. 9. (1990). Rioc&mistry, 1669-1678. Hamlin, R. (198.5). Methods Enzyw/,ol. 114. .I l&452. Hendrickson, $V. A. & Konnert. ,J. H, (1980). In Crystallography (JXamond. R., ( ‘ompu,ting in Rarnaseshan. S. & Venkatesan. K.. eds), pp. 13.0113.23, Indian Academy of Scienc~rs. Jntrrnational I’nion of (‘rystallography, fsangalorr. India. Hodsdon, rJ. :M., Brown, G. M., Sieker. 1,. (‘. & ,Jensen, I,. H. (1990). ilcta Crystallogr. sect. H. 46. 54-62. HopIe, ,J., Rao. S. T.. Mallikarjunan. 31.. ISeddell, (‘.. McMullan, It. K. & Sundralingam. M. (1981). .4rtn C‘rystallogr.

Holler.

sect. B, 37, 591 597.

E.. Rupley. J. A. Bi HOW. (:. 1’ (IF)‘i.%). 14. 1088-lO!N. Rupley. J. A. & Hess. (:.

Riochemistry,

References Anet. F. 14. 1~.bi Brown, J. R. (1967). J. Awaer. (‘hem. Sot. 89, 76&765. Artymiuk. 1’. ,J., Blake, C. c’. F., Rice, 1). W. & Wilson, K. S. (1982). Acta Crystallogr. sect. B, 38, 778-783. Ballardie. F. W., Capon, B.. Cuthbert, M. W. & Dearie, W. M. (1977). Rioorg. Ghem. 6, 483609. Banjeree, S. K., Holler, E., Hess, G. P. & Rupley, J. A. (1975). J. RioI. Chem. 250, 4355-4367.

Holler.

E.,

Biochemistry.

I’.

(19i5h).

14. 2377--2385.

Howard,

A. .J.. Nelson, C. 8: Xuong. S. 11. (19%). 114, 452-472. Tmot’o, I.. Johnson. 1,. N., Sorth. A. (‘. T.. Phillips. L). (1. Kr Rupley, J. i\. (1972). In The Enzymes (fio,ver. I’., tad.). 3rd edit. vol. 7. pp. 665--86X. Academic3 Press. IXew York. Inaka. K.. Matsushima, M. Br Morikana. li. ( 1990). AMethods Enzymol.

(‘ollrcf~d

;I t/.&acts

of the Xc’

(‘ongress

and (&era1

.Jarws. M. X. C.. Sielecki. A. It. $Z Moutt. .I. (198:1). ItI f’eptidrs. i\‘truc’turv and sanction (Hrubv. Y. & Ric*h. I).. rtls). pp. 5~l~Fi30, Piers (‘hrmic~al (‘o.. Roc,kforcl. TT,. .Jamw M. N. G. Kr Sielwki. A. R. (1987). In /lif~loyj~~~/ ~~l~~~rfJmoluc~lrs nnd A ssemblirs, A ctiw S&r : of A.. tds). vol. Enzymes (,Jurnak, F. A. h JkPhrrson. 3, pp. 413-486. .John Wiley & Sons. Xew York. ,leantoz, R. W.. Sharon. IV. & Flow-ers. H. 11. (1968). Niochem. Riophys. Hes. (‘omnaun. 13. ZO- 25. *Jeffrey. (:. A. C Taylor. K. (1980). J. C’oncput. C’hrm. 1. 99--t 09. Jt+Trey. (i. A-\. (1990). Acta. C‘rystalloyr. set? H, 46, 89~~105. Johnston. 1,. iY.. (Iheetham, ,J.. McLaughlin. I’. ,J.. :\caharva. K. I’., Hatfold. I). & Phillips, I). C’. (19X8). (‘urr. 1’opic.s Nicrobiol. Imtnunol. 139, Xl 119. .Joltts. *J. Kr ~JollPs. I’. (1961). (‘A’ Acad. A'ci.. Paris. 253. 2773~-2775. I. K: ,Joltts. I’. .Jotl$s. .I.. ,Jauregu-Adetl, J.. Kernier, (1963). BiochinL. Biophys. rlcta. 78, 668-689. .Joltts. T’. & ,Jolt&i, .I. (1984). Mol. (‘~11. Hiw/wrr/. 63. I W-l 80. . . ,lonrs. T. A. (1985). Methods /Cnzyntol. 115. 157-17 I. .lovnson. M. A., Xorth. A. (‘. T.. Sarma. V.. I)ickrrson. RI. E. & Strinrouf. 1,. K. (1970). -1. Mol. Riol. 50.

I’att,.

S. I,.. I~al~lo. .J. H.. t~ot~krlhritlr. K.. \VrisL. (;. & 56. (i”,i (i”!). Sykes. Il. I). (197x). (‘c/r/~/. 91. HifJchrn/. t’txkins. S. .I.. .Johnson. I,. S.. RIa(*hill. I’. .j. cU;I’hillips. I). (‘. (l!GX). /lifJchrn. .I. 173. tiO7 +I& Phillips. I). (‘. (196(i). Sri. ,-