J. Mol. Hid. (1992) 227. 917-933

Structure of a Hinge-bending Bacteriophage Mutant, Ile3 -+Pro

T4 Lysozyme

M. M. Dixon, H. Nicholsont, L. Shewchuk W. A. Baase and B. W. Matthews1 Howard

Institute of Molecular Biology Hughes Medical Institute and Department of Physics University of Oregon, Eugene, OR 97403, U.S.A.

(Received 23 January

1992; accepted 29 May

1992)

The mutant T4 phage lysozyme in which isoleucine 3 is replaced by proline (13P) crystallizes in an orthorhombic form with two independent molecules in the asymmetric unit. Relative to wild-type lysozyme, which crystallizes in a trigonal form, the two 13P molecules undergo large hinge-bending displacements with the alignments of the amino-terminal and carboxyterminal domains changed by 28.9” and 32.9”, respectively. The introduction of the mutation, together with the hinge-bending displacement, is associated with repacking of the side-chains of Phe4, Phe67 and Phe104. These aromatic residues are clustered close to the site of the mutation and are at the junction between the amino and carboxyl-terminal domains. As a result of this structural rearrangement the side-chain of Phe4 moves from a relatively solvent-exposed conformation to one that is largely buried. Mutant 13P also crystallizes in the same trigonal form as wild-type and, in this case, the observed structural changes are restricted to the immediate vicinity of the replacement. The main change is a shift of 63 to 95 A in the backbone of residues 1 to 5. The ability to crystallize 13P under similar conditions but in substantially different conformations suggests that the molecule undergoes large-scale hinge-bending displacements in solution. It is also likely that these conformational excursions are associated with repacking at the junction of the N-terminal and C-terminal domains. On the other hand, the analysis is complicated by possible effects of crystal packing. The different 13P crystal structures show substantial differences in the binding of solvent, both at the site of the Tle3+Pro replacement and at other internal sites. Keywords: lysozyme; dynamics; domain motion; bound water; cavities

1. Introduction

addition,

The lysozyme from bacteriophage T4 is a globular protein of 164 residues consisting of a carboxylterminal domain, made up mainly of a-helices, and an amino-terminal domain, made up of both b-sheet and a-helices (Fig. 1). The wild-type protein and a number of mutants have been well characterized by a variety of techniques (Hawkes et al., 1984; Weaver & Matthews, 1987; Matthews, 1987; Chen & Schellman, 1989; Chen et al., 1989; McIntosh et al., 1990; Bell et al., 1990). Over 80 o/o of the T4 lysozyme mutants purified to date have been crystallized isomorphously with the wild-type lysozyme under a narrow range of conditions (Wozniak et al., 1990; Brennan et al., 1988). In t Present address: Department of Chemistry and Biochemistry, Massey University, Palmer&on North, New Zealand. $ Author to whom all correspondence should be addressed. 0022-2836/92/190917-17

$08.cto/0

917

the wild-type

protein

has been crystallized

isomorphously under very different conditions, e.g. low salt as opposed to N 2 M-phosphate (Bell et al., 1991). In contrast, several mutants have been found that crystallize both isomorphously and nonisomorphously with wild-type under conditions virtually identical to wild-type. Recently, the crystal structure of one of these mutants, Met6 + He (M61); has been reported (Faber & Matthews, 1990). While the isomorphous structure of M61 is very similar to that of wild-type lysozyme, the nonisomorphous structure displays four alternate molecular conformations, with different “hingebending” displacements that leave the N and C-terminal domains essentially unchanged but effectively enlarge the active-site cleft. Here, the structure of a second T4 lysozyme mutant, Ile3+Pro (13P), which crystallizes both isomorphously with wild-type and non-isomorphously in yet another crystal form, is also found to display a large hinge-bending displacement. The 0 1992 Academic Press Limited

918

M. M. Dixon et al.

Figure 1. Backbone tracing of’ wild-type T4 Iysozyrne showing the positions of thr nlutation~ Iltb:r + 1’t.c) a11t1 Met,6 + Ile. The filled circles show the locations of the 4 essentialIS-itlt,rrrl~l solvent tnolrc~ulrs I~our~tl to wilti-type fysozyme (Sol 1 to Sol 4) and t,hr open (aircle shows the solvent molecult~ srrn at thtb rt$ac~rmrnt site in thca isomorphous crystals of 13P.

structure of the Ile3 -+Pro mutant, that is isomorphous with wild-type we will designate I3P,. The crystal form that is non-isomorphous with wild-type contains two independent molecules and we denotr these 13P, and 13P,. I3P is one of a series of directed mutations of T4 lysozyme originally created in an effort to stabilize the enzyme by introducing proline residues at predetermined sites (Matthews et al., 1987). The backbone conformation at residue 3 suggested that a. proline residue could be substituted at this site without significantly altering the protein structure and therefore contribute to enhanced stability. At, the same time it was recognized that the removal of the isoleucine sidechain would result in the loss of stabilizing hydrophobic and van der Waals’ interactions (Matsumura et al.. 1988, 1989). The observation that the net stability of the mutant is less than the wild-type shows that the loss of st’ability due t,o

removal of the isoleucine is greater by replacement! with proline.

2. Experimental l’roredures

for

than that pained

Procedures

c*loning. squc~ncing antI f,uritic,atiori k Smith. 19X4: Alatsumura rt nl.. 1988; 1989: Alber h Matthews. f 987: Potrettb c,t f&cl.. 1991). Two different, crystal forn~s were rrbta.incJd using star1 dard wild-typr qvstallization caonditions (\Vtaavtbr & Matthews. 1987; Brennan 01 (11.. 1988). C’rystals wt’rt’ grown by microbatch of vapor difl’usion mrthods with a protein conc*entration of I5 m&ml in .. 2 xl-phosphattb. 5 IO rn.M-/~-mc~rc~a~)tortha,lc,l. WI M-Ka(‘l. md to Wild-t.ype-like trigonal c.ryst.als in sparf’ grotrp I’Y2:! I (Table I ) grew over 2 to 3 weeks at 4°C’. a typical growth t,imr for mutant Iysozyrnr (7.ystaIs. Vnder thr samt- contlitions. orthorhombic cr>-stals originally took al)out 1 ?‘rar to grow. although subsryurnt bat,chrs grciv\ much mc)rt’

were as described (Zoller

A Hinge-bending

Mutant of T4 Lysozyme

919

Table 1 Data collection and rejinement statistics Form I (‘rystal

form

\Vild-typet

1’3,21 spaw group (‘ell dimensions 0 (.Q 61.2 b (A) ti1.P c (-9) 96.8 M0lrr11lrs:a.u. 1 lVM(AX/Da) %.X Resolution (-%) 60-1.7 Knerge PC,) 91 FLIS0(“0) Refinement statistics R-factor (O) 16.7 r.m.s. deviations from ideal values: Bonds (A) 0018 Angles (0) 2%

Form III I3P

I3P

13P(HgCl,)

Form II MGIf

P3,2 I

P%,2,2

P2,2,2

1’2,2,2,

60.9 60.9 97.3 1 28 6G2.0 7.0 27 1

86.5 966 392 2 2.2 20-20 6.7

86.5 96.6 39.2 ” 2.2 20.-4.0 5.9 20.0

72.2 73.8 150.5 4 2.7 6G2.1 12.5

16.7

19.0

23.6

0.019 2.6

0.020 3.0

(PO13 2.7

t From \vearer Br Matthe\vs (1987). $ From Faber & Matthews (1990). 1, is the crystal packing parameter (Matthews, 1968). Emergeis the agreement between repeated intensity measurements, Rise is the average difference between structure amplitudes measured for wild-type and an isomorphous mutant or heavy-atom derivative. IZ-fkrtor is the csrystallographir residual for the refined structure. a.u.. asymmetric unit.

rapidly. over a period of 1 to 2 months. These crystals were in space group P2,2,2 with apparently 2 molecules/ asymmetric unit (Table 1). X-ray data for both crystal forms were collected on a San Diego Multiwire Mark II area detector (Hamlin. 1985) after crystals were stabilized in a mother liquor of 2.7 M-phosphate (pH 66), (b23 M-lr’a(‘l and I.4 rnM#-mercaptoethanol. (b) Isomorphous

structure

The isomorphous structure (13P,) was determined direct,ly. The co-ordinates of wild-type T4 lysozyme (Weaver & Matthews, 1987) were used as the starting point with the mutant side-chain initially placed (Jones, 1982) from inspection of the (2F,,,,,,,,Fwi,d~,ype) electron density map. In the early stage of refinement (Tronrud et a/.. 1987) the starting model was allowed to relax to conform to the observed mutant crystallographic data. Stereochemical restraints were then applied to achieve the desired agreement with ideal bond lengths and angles (see Dao-pin et nl., 1991 for a typical refinement protocol). (c) Son-isomorphous

structure

Starting models for the 2 independent molecules in the orthorhombic crystal form (13P, and 13PB) were determined by a combination of single isomorphous replacement (SIR?) and molecular replacement techniques. Conventional rotation and translation-function approaches located 1 but not the 2nd of the 2 lysozyme molecules in the asymmetric unit. A single heavy-atom derivative was therefore obtained by soaking the crystals in 1.0 mM-HgCl, for 2 days (Table 2). A difference Patterson map was readily interpreted in terms of 2 sites of substitution (1 and 2), which were refined using the HEAVY (Terwilliger & Eisenberg, 1983). program A native electron density map calculated at 50 A resolut Abbreviations replacement; WT,

used: SIR, single isomorphous wild-type; r.m.s., root-mean-square.

tion using phases based on the 2 mercury sites, had features indicating likely cc-helices and solvent regions but could not be immediately interpreted in terms of 2 lysozyme molecules. It was expected (Matthews & Remington, 1974; Faber & Matthews, 1990) that the mercury would bind to either or both of the 2 cysteine residues in the molecule. Therefore. a series of restricted 6-dimensional searches of the SIR map were performed by rotating and translating, in turn, models of the backbone atoms of the wild-type and each of the 4 mutant Met6 + Ile molecules (Faber & Matthews, 1990) through the map, keeping either Cys54 or Cys97 in the vicinity of the 2 mercury sites. Agreement was measured by

Table 2 Rejined heavy-atom parametera Fractional Site 1 2

Relative occupancy 2.00 208

5

co-ordinates ^

Y

0.0470 -0.0190 01689 0.0499 0.0897 0.6239 Centric data Acentric data

Resolution (A) Number of reflections Average phasing power Centric R-value Mean figure of merit (all data)

40 440 1.27 055

40 1915 1.60 0.35

The phasing power is defined as the r.m.s. of the heavy-atom scattering divided by the r.m.s. of the lack of closure error, i.e. .fr.m.../4.m... where:

fH is the heavy-atom scattering factor calculated from the heavyatom positions, lFpl and IFpHI are the observed amplitudes from the protein and derivative crystals, respectively, and ?z is the number of reflections.

920

M. M. Dixon

et al.

Table 3 Determination Search model A Asite I WT MHI, M61, M61,

M61,

of molecular

orientation, I’rali height

Y (‘9 - 1% - I.8 - I.8 -1.8 - 1.8

4.2 4.2 4.2 4.2 4.2

H. Final orientation N-term 40 (!-term 50 t:. Site 2 WT M61, M61, M61, M61,

45 4.5 4-5 45 4-5

2.54 25.0 250 250 250

D. Ir‘innl orientatim N-term 43 C-term 34

24.5 21.5

The Table summarizes the search through the SIR SO A resolution electron density map using different search models as tlesc~rib~cl in the text. In the first part of the Table, site 1 of the heavy-atom derivative (co-ordinates T = 42 11. y = -- I.8 .A. 2 = 6.8 A) is assumt~~ lo correspond to a mercury atom bound to Cys,54, as described by Matthews & Remington (1974). Th c search model was then rotated in steps of 10” with this site as the center of rotation. $ and 4 define t,he orientation of thr rotation axis. with $ being the inclination to thr y-axis and 4 the azimut,hal angle between the %:-axis and the projection of the rotation axis on the .r-: plane. K is the anglr of rotation about the ($,+) axis. The highest peak is given as the number of standard deviations above the r.m.s value. This gives t,he approximate orientation of 1 of the 13P molecules in the asymmetric unit. For comparison. the final orientation parameters for the amino-t.erminnl domain and the carboxyl-terminal domain are also given. The second part of the Table summarizes the determination of the alignment of the 2nd UP mole~uir. again obtain& by assuming that site 2 of the heavy-atom derivative corresponds to a mercury atom bound to Cys54. Parallel searches were also performed assuming that both sites 1 and 2 corresponded to mercury bound to Cys97. In this case the ma,ximum peak heights. for the Ram? motirls show-n in the Table, were between 3.7~~and 44~.

the SIR map density at the positions of the backbone and p-carbon atoms of the search model as it was moved through the map, using the PROTEIN system of programs (Steigemann. 1974). The location of the 2 molecules was clearlv indicated hy peaks 5 to 70 ahovr background (Table 3). The 4 M6T search models seemed t,o he slight’ly superior to wild-type, hut the approximate position and orientation of the 2 ISI’ molecules would have been clear no matter what model was used (Table 3). The N and V-terminal domains wew then oriented separately using separate local searches in the regions near the orientations found in the searches using the entire molecule. Tn t,he starting model for refinement, the side-chains were taken to he in the conformations of the &I61 molecule that gave the highest peak-to-noise ratio in the SIR map summing

search. except for residues arily modeled as alanine.

3 and 6. which

were tempor-

The start.ing K-factor for all data between do3 and 2.0 A resolution was 45.0”/,, and dropped t,o 4@00/,, after 9 cagrles of rigid-body refinement. treating the four domains as separate units. The stereochemistry was then relaxed. and SO cycles of refinement with fixed B-values, alternated with rebuilding hy inspeetion of electron density and difference maps, reduced the

K-fac%or to BX)(~/,. Over the R-values

were refined,

and

water

next

50 cycles individual

molecules

were

added

to

the model if they appeared in the difference maps as peaks of at least 50, and had at least 1 potential hydrogenbonding partner. The final R-factor was 19.0%. Crystallographic and refinement, statistics are summarized in Table I. Co-ordinates of the T3Po. T3P, and I3P,

structures

have

been deposit,rd

in the Brookhaven

IJat)a

Hank (Atx~ession num hers 1TS6. 1LBS).

3. Results

The change in therm&ability of 131’ relat.ivr to wild-type (WT) was drt.ermined 1)~ using Grcu1a.r dichroism to monitor the unfolding as a function ot temperature (Dao-pin rt ~1.. 1990). At pH 3.01 itI 25 mM-K(“l. 3 m>~-H3t’04. IS m~~-KH2t’04 thy melting temperature is W2( +O~-t)“(’ which is 7.3( fO%)deg.(’ lower than the WT undo the sam(J conditions. The change in ent.halpy of unfolding. AIlI, was 102 kcal mol 1 (1 Cal = 1.lM,J) anti the entropy of unfolding. A#. was 319 ~1 drg~ ’ tnot ’ The corresponding values for the G’T Iysozytrrt~ under the same caonditions are AII = I30 kval mol ’ and A8 = 398 cd (kg 1 mol 1 (Kriksson it 4.. 1992). From Heckt’el & Schellman’s (I 987) relation this corresponds t,o a reduction in st,ahilit>y of’ t htb mutant protein relative to t,he M:T of’P%( iO.4) kfxl mol ‘. The unfolding of the protein c~orresponds to a two-state transit,ion with no suggestion of stablo intermediates. The activit.y of the mutant protrin. drtx+rtninetl hy the method described h-y Tsugita et nl. (19(S)

A Hinge-bending

Mutant of T4 Lysozyme

921

(a)

Y

CI

GLU 5

GLU S

(b)

Figure 2. (a) Electron density map showing the difference in density between isomorphous 13P, and WT lysozyme. Amplitudes are P’,,,,,,.,, -Fwild.typs) and ph ases are calculated from the refined WT structure (Weaver & Matthews, 1987). Resolution is 2.0 8. Positive (continuous) and negative (broken) contours are drawn at levels of *4a, where u is the r.m.s. density throughout the unit cell, (b) Superposition of the 13P, crystal structure (open bonds) on that of WT (filled bonds). The solvent molecule SOL OH1 is present only in the mutant structure. The other solvent (not labeled) is present

in WT but not in the mutant. appeared to be indistinguishable native enzyme.

from that of the

(b) Isomorphous I3P structure The map showing the difference in electron density between the mutant and the WT structure is illustrated in Figure 2(a). Negative density near the CY and Cd’ atoms of the side-chain of the Ile3 in the WT lysozyme and the presence of a small positive feature adjacent to the main-chain N, C” and CB atoms confirms the replacement of Ile with Pro. A negative density feature indicates the displacement of a solvent molecule hydrogen-bonded to the backbone amide group of Ile3.

Figure 2(b) shows the superposition of the refined 13Pc structure on the WT lysozyme. To eliminate effects due to the changes in the unit cell dimensions of the mutant crystals, the refined co-ordinates of the 13P mutant were rotated and translated so as to minimize the discrepancy (@18 A for the backbone atoms). The main structural adjustment is an essentially rigid-body shift ( -0.3 to 0.5 A) in the backbone of residues 1 to 5 such that the pyrrolidine ring of Pro3 moves toward the space vacated by the Ile3 side-chain. The crystallographic thermal factors of the atoms in the vicinity of Pro3 are very similar to those in the WT structure. The refinement suggests that a buried water molecule occupies the cavity created

922

M. M. Dixon et al.

(a)

(b) Figure 3. (a) Backbone of the 2 independent molecules 13P, and 13P, in the non-isomorphous crystal structure (thick bonds) superimposed on the backbone of WT lysozyme (thin bonds) according to the N-terminal domains (residues 15 to 60). (b) Backbone of the 2 independent molecules in the non-isomorphous structure (thick bonds) superimposed on the backbone of WT lysozyme (thin bonds) according to the C-terminal domains (residues 80 to 160). by the Ile3 -+ Pro replacement (Fig. Z(b)). The solvent molecule participates in several hydrogen bonds (see Discussion, below). It has a crystallographic thermal factor of 15 A2, well below the average for solvent in other T4 lysozyme structures (-48 A’), indicating that it is well ordered. The electron density difference map is essentially featureless farther away from the site of the substitution, indicating that the 13P mutant is very similar to WT. (c) Non-isomorphous

I3P structures

(i) Overall conformation The most striking aspect of the two lysozyme molecules in the non-isomorphous I3P crystal form (13P, and 13P,) is the relative movement of the N-terminal and C-terminal domains. While the con-

formation within each domain is very similar to the WT structure (root-mean-square (r.m.s.) in backbone atoms