Refined structure of Cro repressor protein from bacteriophage λ suggests both flexibility and...

Article No. mb981849 J. Mol. Biol. (1998) 280, 129±136

Refined Structure of Cro Repressor Protein fromBacteriophage lll Suggests both Flexibilityand Plasticity

Douglas H. Ohlendorf, Dale E. Tronrud and Brian W. Matthews*

Institute of Molecular BiologyHoward Hughes MedicalInstitute and Department ofPhysics, University of OregonEugene, OR 97403-1229, USA

Present address: D. H. OhlendorBiochemistry, University of MinnesMedicine, 4-225 Millard Hall, Minn0347, USA.

0022±2836/98/260129±08 $30.00/0

The structure of the Cro repressor protein from phage l has been re®nedto a crystallographic R-value of 19.3% at 2.3 AÊ resolution. The re-®ned model supports the structure as originally described in 1981 andprovides a basis for comparison with the Cro-operator complex describedin the accompanying paper. Changes in structure seen in different crystalforms and modi®cations of Cro suggest that the individual subunits aresomewhat plastic in nature. In addition, the dimer of Cro suggests a highdegree of ¯exibility, which may be important in forming the Cro-DNAcomplex.

The structure of the Cro subunit as determined by NMR agrees reason-ably well with that in the crystals (root-mean-square discrepancy ofabout 2 AÊ for all atoms). There are, however, only a limited number ofintersubunit distance constraints and, presumably for this reason, thedifferent NMR models for the dimer vary substantially among them-selves (discrepancies of 1.3 to 5.5 AÊ ). Because of this variation it is notpossible to say whether the range of discrepancies between the X-ray andNMR Cro dimers (2.9 to 7.5 AÊ ) represent a signi®cant difference betweenthe X-ray and solution structures.

It has previously been proposed that substitutions of Tyr26 in Croincrease thermal stability by the ``reverse hydrophobic effect'', i.e. byexposing 40% more hydrophobic surface to solvent in the folded formthan in the unfolded state. The re®ned structure, however, suggests thatTyr26 is equally solvent exposed in the folded and unfolded states. Themost stabilizing substitution is Tyr26! Asp and in this case it appearsthat interaction with an a-helix dipole is at least partly responsible forthe enhanced stability.

# 1998 Academic Press

Keywords: Cro repressor protein; bacteriophage l; crystallographicstructure; ¯exibility; plasticity
*Corresponding author
Introduction

The genome of bacteriophage l codes for tworepressor proteins, the Cro repressor (Cro) and thel or cI repressor. The function of Cro is to turn offearly gene transcription during lytic growth, whilel repressor is required to maintain lysogenicgrowth (for reviews, see Ptashne, 1986; Pabo &Sauer, 1984).

The structure of Cro was determined byAnderson et al. (1981) and a detailed model was

f, Department ofota School ofeapolis, MN 55455-

proposed for its interaction with DNA byOhlendorf et al. (1982). Cro is one of the prototypi-cal representatives of the ``helix-turn-helix'' class ofDNA-binding proteins (Anderson et al., 1982;Ohlendorf et al., 1983; Brennan & Matthews, 1989).

The crystal structure of a Cro-operator complex(Brennan et al., 1990) supported the generalfeatures of the Cro-DNA model (Ohlendorf et al.,1982). This complex was, however, at low resol-ution (nominally 3.9 AÊ ) and not re®ned. Recently,it has been possible to determine and re®ne thestructure of a different Cro-DNA complex fromcrystals that diffract to 3.0 AÊ resolution. This hasmade it possible, for the ®rst time, to visualize thespeci®c interactions between the protein and theDNA (Albright & Matthews, 1998). In order to

# 1998 Academic Press

130 Cro Repressor Protein from Bacteriophage �

describe, in detail, the structural changes thataccompany DNA binding, we here describe there®ned structure of the native protein.

Results and Discussion

The crystal structure of Cro contains fourmonomers (O, A, B and C) arranged around fourmutually intersecting non-crystallographic 2-foldaxes. The protein is dimeric in solution (Takedaet al., 1986), the physiological dimer correspond-ing to the O-B (or A-C) dimer (Figure 1(a)),which excludes 680 AÊ 2 of surface area per mono-mer.

Figure 1. (a) Ribbon diagram showing the structure ofthe O-B (or A-C) Cro dimer as seen in the crystal struc-ture. Phe58 of each subunit (shown in yellow) contrib-utes to the core of the partner subunit. Residues 61 to66 at the carboxy terminus are disordered and not seenin the crystal structure. (Figure courtesy of RonAlbright.) (b) Comparison of the backbone structures ofa subunit of wild-type Cro (blue), Cro in complex withoperator DNA (yellow) and engineered monomeric Cro(red). (Figure courtesy of Ron Albright.)

Secondary structure

Cro has a rather simple secondary structure con-sisting of three a-helices, a1 (7 to 14), a2 (16 to 23)and a3 (27 to 36) and b-sheet strands, b1 (2 to 6),b2 (39 to 45) and b3 (49 to 56). Gly15 connectsdirectly from helix a1 to helix a2. In order to do soit has the unusual dihedral angles (f � 85�,c � 150�; see Figure 2). Helices a2 and a3 comprisethe well-known helix-turn-helix motif. Helix a3ends with the fairly common secondary structureof hydrogen bonds from residue n � 1 to residuen ÿ 4 and from residue n to residue n ÿ 3. As isoften the case, residue n is glycine (Gly37) in theleft-handed a conformation (Schellman, 1980).Helix a1 also ends with glycine (Gly15) in the left-handed a conformation.

Strands b1, b2 and b3 together form an antipar-allel b-pleated sheet with the conventional left-handed twist. The hairpin bend between strandsb2 and b3 is a type I reverse turn (Venkatachalam,1968) with Gly48 in this turn in the left-handed aconformation. The interaction of strand b3 withstrand b2 ends at Glu53 and the continuation ofthe strand (residues 54 to 58) forms an antiparallelinteraction with strand b30 from the 2-fold-relatedmonomer (Figure 1(a)). At the end of strand b3there is a proline residue at position 57 followedby a bend that is formed by a cis peptide link toPro59 (Figure 3(a)). The conformation exhibited inthis turn has been called type VIb by Richardson(1981). At the apex of the turn, Phe58 from onemonomer interdigitates into the core of the 2-fold-related monomer (Figures 1(a) and 3(b)).

Figure 2. Ramachandran diagram, calculated using theprogram of Laskowski et al. (1993), showing the back-bone conformational angles for the four Cro monomers.Glycine residues are shown as triangles, non-glycineresidues as squares. Ser60 is toward the carboxy termi-nus of the molecule at the point where the electron den-sity becomes weak, indicative of disorder.

Figure 3. Representative sections of the electron density map following re®nement. Coef®cients are 2Fo ÿ Fc, whereFo are the observed amplitudes and Fc are those calculated from the re®ned model. Phases also are from the re®nedmodel. (a) The region near cisPro59. The Figure includes the conserved water molecule, labeled Sol. (b) The regionwhere Phe58 of monomer A (labeled A58) penetrates into the hydrophobic core of monomer C and is surrounded byresidues, including Leu7, Leu23, Val25, Ile30, Arg38 and Ile40.

Cro Repressor Protein from Bacteriophage � 131

Different Cro structures

The structure of Cro has been determined in ®vedistinct modi®cations. Here, we brie¯y comparethese different variants. First there is the crystalform of the DNA-free protein, as described herein,including four crystallographically independentsubunits, O, A, B and C. The discrepanciesbetween the coordinates of representative mono-mers are included in Table 1. The discrepancybetween backbone atoms (not shown) is 0.33 AÊ to0.46 AÊ , which is comparable with a number ofother cases in which the structure of the same pro-tein has been determined in different crystal forms(e.g. see Chothia & Lesk, 1986; Wagner et al., 1992;Zhang et al., 1995). The second and third structuresof Cro are complexes with operator DNA (Brennanet al., 1990; Albright & Matthews, 1998). We hereuse the higher-resolution coordinates of the latter.Fourth there is an engineered monomeric form of

Cro, the structure of which has recently been deter-mined at high resolution (Albright et al., 1996).Finally, the solution structure of wild-type Cro hasbeen determined by NMR (Matsuo et al., 1995).

The data suggest that there are distinct structuralchanges within the core of the Cro subunit, bothon binding DNA, and on engineering to promotethe formation of the monomer. Figure 1(b) com-pares the backbone structures of the engineeredmonomer and subunits of Cro in the free andDNA-bound forms. As can be seen, the wild-type(free) structure is distinct from the other two. Thus,it appears that the structure of Cro, even within asingle monomer, is relatively plastic.

The determination of the solution structure ofCro by NMR is summarized by an ensemble of 20different sets of coordinates (Matsuo et al., 1995).Table 1 gives the range of values obtained whenthese coordinate sets are compared with the X-ray

Table 1. Comparisons of the crystal and NMR structures ofdifferent modi®cations of Cro

Discrepancy(AÊ )

A. Comparisons of monomersFree Cro, subunit O versus subunit A 1.2Free Cro, subunit O versus subunit B 1.3Free Cro, subunit O versus subunit C 1.1Free Cro, subunit O, versus bound Cro 1.7Free Cro, subunit O, versus engineered Cro monomer 1.8Bound Cro versus engineered Cro monomer 1.7Free Cro, subunit O versus free Cro (NMR) 1.9-2.4Bound Cro versus free Cro (NMR) 1.9-2.4Engineered Cro monomer versus Cro (NMR) 2.1-2.6Cro (NMR) versus Cro (NMR) 1.0-1.8

B. Comparisons of dimersFree Cro, OB versus AC 1.2Free Cro, OB versus free Cro (NMR) 2.9-7.5Free Cro (NMR) versus free Cro (NMR) 1.3-5.5Bound Cro versus free Cro, OB 4.5Bound Cro versus free Cro (NMR) 3.4-5.4

The Table summarizes the root-mean-square discrepancies between thestructures indicated. The comparisons include both main-chain andside-chain atoms but are restricted to residues 3 to 56, since these arethe ones that are reasonably well de®ned (i.e. non-disordered) in all ofthe structures being compared. (Residues 57 to 59 are present in theNMR structures but have widely divergent conformations (Matsuo et al.,1995).) Free Cro, subunits O, A, B and C correspond to the four subu-nits in the crystals of wild-type Cro (this work). Bound Cro is Cro incomplex with operator DNA (Albright & Matthews, 1998). EngineeredCro monomer is an engineered monomeric form of Cro (Albright et al.,1996). Free Cro (NMR) corresponds to an ensemble of 20 coordinatesets for wild-type Cro determined by NMR (Matsuo et al., 1995).


structures, and shows the internal consistencybetween the NMR structures. Within the accuracyof the NMR coordinates it is not possible to deter-mine whether the structure within a subunit of Croin solution corresponds most closely to the crystalstructure of free Cro, of DNA-bound Cro or ofengineered monomeric Cro.

There are, however, substantial differencesbetween the coordinates for the various dimericstructures of Cro. The two dimers of Cro present inthe crystal structure of the wild-type protein (i.e.OA versus AC) agree moderately well (1.2 AÊ dis-crepancy for all atoms). There is a substantialdifference (4.5 AÊ ) between the crystal structures ofthe wild-type OB dimer and that in complex withDNA (Albright & Matthews, 1998). This is due tothe large conformational change of the dimer onbinding (Brennan et al., 1990; Albright &Matthews, 1998). The ensemble of NMR-deter-mined structures of the Cro dimer differ substan-tially among themselves (from 1.3 AÊ to 5.5 AÊ ).They also differ from the crystallographicallydetermined dimer (2.9 AÊ to 7.5 AÊ ). As noted byMatsuo et al. (1995), the determination of the con-formation of the dimer by NMR is dif®cult becauseof the small number of inter-subunit distance con-straints. In addition, the only inter-subunit NOEsthat could be found in the NMR spectra of Crowere predicted by the X-ray structure, and aretherefore consistent with it. Thus, it is not clearwhether the apparent discrepancy of 2.9 to 7.5 AÊ

between the Cro dimer in the crystals, and thatdetermined by NMR, represents a real difference,or simply re¯ects the indeterminancy in the NMRstructures for the dimer.

Mobility

The crystallographic thermal factors for the back-bone atoms within the four monomers are com-pared in Figure 4(a). The side-chain thermal factorsare compared in Figure 4(b). For the side-chains, inparticular, the consistency is good (cf. Zhang et al.,1995). The correlations between the backbone ther-mal factors range from 0.76 to 0.53, while thosebetween the side-chain atoms are between 0.76 and0.63. This suggests that the mobilities seen crystal-lographically are indicative of those in solution. Itshould, however, be remembered that the structurein the crystal is tetrameric, whereas that in solutionis dimeric. Thus, additional motion may well occurin solution that is not anticipated in the crystals. Inparticular, the connection between the subunits inthe physiological dimer (Figure 1) seems, by itsnature, to be especially ¯exible, and to allow bothbending and twisting. Indeed, a large confor-mational change is seen when Cro binds to DNA(Brennan et al., 1990; Albright & Matthews, 1998).

Figure 4(a) gives an overall impression of thedynamic behavior of the monomer within the crys-tal. The most rigid region (or perhaps one shouldsay the least mobile) corresponds to the b2 and b3

Figure 4. Crystallographic thermal factors of the fourCro monomers, O, A, B and C. The locations of thesecondary structure elements are indicated in the bot-tom panel. (a) Main-chain thermal factors. The valueplotted for each residue is the average of the values forthe backbone atoms. (b) Side-chain thermal factors. Thevalue plotted for each residue is the average of theB-factor of the side-chain atoms.


strands, which can be considered as the ``core'' ofthe physiological dimer. In general, the connectionsbetween the elements of secondary structure tendto be more mobile than the helices and sheetsthemselves. The solvent-exposed hairpin bend inthe vicinity of Ala46, Asp47 and Gly48 is especiallymobile. The carboxy-terminal part of the protein(residues 60 to 66) is very mobile and distinct elec-tron density cannot be seen for the residuesbeyond Ser60.

Bound solvent

There are only two places where a solvent mol-ecule is bound at an equivalent location (within1 AÊ ) in all four monomers. In the ®rst case the sol-vent is hydrogen bonded to the carbonyl oxygenatom of Phe58, which is within the only cis peptidebond in the structure (Phe58-cisPro59; Figure 3(a)).The second, conserved, solvent molecule is hydro-gen bonded by the hydroxyl group of Thr43 andthe carbonyl oxygen atom of Phe41. This solvent islocated at an interface between two monomerssuch that, in the crystal structure, it is essentiallyinaccessible to bulk solvent.

Thermostable Cro mutants

Pakula & Sauer (1989, 1990) identi®ed two Cromutants, Tyr26! Cys and Gln16! Leu, thatincrease the melting temperature of the protein by11 to 14 deg. C. Such variants, especially at sol-vent-exposed sites, are rare and led Pakula Sauer(1990) to construct other variants at position 26,several of which also increased thermostability(Val, Leu, Gln, His and Asp). Replacements withPhe and Trp, however, had stabilities close towild-type. Pakula & Sauer (1990) attributed theenhanced stabilization at site 26 to the ``reversehydrophobic effect'', i.e. Tyr26 in the wild-typestructure of Cro was assumed to adopt a confor-mation such that it exposed more hydrophobic sur-face to solvent than it does in the unfolded form.The present structure determination does not sup-port this suggestion. Although the side-chain ofTyr26 is exposed to solvent in the folded protein,the degree of exposure is not unusual. The fourindependent copies of Tyr26 in the crystal structureof the wild-type proteins have backbone confor-mational angles of f � ÿ95�, c � 140�. This corre-sponds to a commonly observed extendedconformation (Figure 1). The solvent accessibilitiesof the four tyrosine side-chains, calculated usingEDPDB (Zhang & Matthews, 1995) followingthe method of Lee & Richards (1971) are 165 AÊ 2 to191 AÊ 2 and that for a model extended peptide ofthe form Ala-Tyr-Ala is 190 AÊ 2. Taking the latter asa model for the unfolded state (Lee & Richards,1971) the fractional solvent accessibility of the fourTyr26 side-chains are, respectively, 0.87, 1.00, 0.90and 1.01. In other words, the calculated solventaccessibility of Tyr26 is essentially the same in thefolded and extended states. In contrast, Pakula &


Sauer (1990) suggested that the side-chain of Tyr26in the folded structure was 1.4 times more solvent-exposed than in the extended form. We do notknow the origin of this discrepancy, although wenote that the solvent accessibility quoted by Lee &Richards (1971) for a tyrosine side-chain in anextended Ala-Tyr-Ala peptide is 75% less than thatdetermined here, presumably due to differences inatomic radii that were used in the earlier calcu-lations. In any event, the present calculations donot show any evidence for a ``reverse hydrophobiceffect'' at site 26.

What, then, is the cause of the enhanced stabiliz-ation at site 26? In the case of Asp26, which is themost up-stabilizing variant, energy simulationanalysis (Tidor, 1994) suggests that the helix dipoleeffect could be important. Residue 26 is at theamino terminus of an a-helix and Nicholson et al.(1988, 1991) have also shown experimentally thatthe introduction of aspartic acid at such sites inphage T4 lysozyme consistently increases the stab-ility of the protein.

For the other stabilizing mutation in Cro,Gln16! Leu (Pakula & Sauer, 1989), it wouldseem that the effect could be due simply to theincreased burial of hydrophobic surface in themutant structure, in particular at the interfacewhere the side-chain of residue 16 is in contactwith the side-chain of Ile30.

Methods

Data collection

Cro protein was puri®ed and crystallized as described(Takeda et al., 1977; Anderson et al., 1979). The spacegroup is R32 with cell dimensions a � b � 91.6 AÊ ,c � 268.5 AÊ , and the solvent content parameter isVM � 3.7 AÊ 3/dalton.

The principal means of data collection was oscillationphotography (Rossmann, 1979; Schmid et al., 1981) usingmonochromatized X-rays from an Elliott GX-21 rotatinganode generator. Because the crystals grow as long tri-angular prisms with the c-axis along the prism axis, crys-tals were always mounted such that the c-axis wascoincident with the axis of oscillation.

Most of the high-resolution data were collected from asingle, unusually large, crystal measuring about0.6 mm � 0.6 mm � 1.2 mm. These data were sup-plemented by some precession ®lm data plus lower-res-olution oscillation data (Anderson et al., 1981). Overall,80% of the data were measured to 2.2 AÊ resolution, witha merging R-factor on intensities of 9.6%. For the ®nalre®nement, however (see below), the data were restrictedto 2.3 AÊ resolution (84% complete).

Refinement

The crystal structure has four Cro monomers perasymmetric unit related to each other by non-crystallo-graphic 222 symmetry (Anderson et al., 1981). A multipleisomorphous replacement map at 3.5 AÊ resolution wasimproved by averaging over these four independentregions and used for the initial structure determination(Anderson et al., 1981; Ohlendorf et al., 1982).

The four Cro monomers, each of 66 amino acid resi-dues, are labeled O, A, B and C (O-B and A-C are thephysiological dimers). A model for residues 1 to 59 ofmonomer O was built into the electron density map andthen expanded, using the non-crystallographic sym-metry, to obtain starting models for the correspondingparts of monomers A, B and C. From this point on, there®nement of all four monomers was independent.

The starting R-factor, at 3.5 AÊ resolution, was 52.4%.Initial re®nement was with the program EREF (Jack &Levitt, 1978). A series of ten or more cycles was run at3.5 AÊ and then at 2.8 AÊ resolution, subsequently extend-ing to 2.5 AÊ and ultimately 2.2 AÊ resolution. Checks onthe molecular stereochemistry were made and question-able conformations were checked against appropriatedifference density maps as well as the multiple isomor-phous replacement map. Re®nement of thermal factors,and placement of bound solvent, was not attempteduntil after 75 cycles of re®nement.

Subsequently, the re®nement was continued with theTNT re®nement package (Tronrud et al., 1987; Tronrud,1992). Because of the long duration of the re®nement ofthis model, several versions of TNT were used with the®nal cycles performed by ``release 5D''. Due to the poorquality of the high-resolution diffraction data, the resol-ution of the data were restricted to 2.3 AÊ . In view of thislimit, a number of restraints was applied to the model,including restraints on the individual thermal valuesbased on stereochemistry (Tronrud, 1996) and restraintson the pseudorotation angles of the eight proline resi-dues.

A very conservative approach to modeling solventwas adopted. Water molecules were included only if thefollowing criteria were met: (a) density was observed inthe Fo ÿ Fc map greater than 3s, (b) density wasobserved in the 2Fo ÿ Fc map greater than 1s, (c) thedensity was essentially spherical in both maps, and (d)at least one hydrogen bond with good stereochemistrywas possible. During subsequent re®nement, water mol-ecules were deleted if their 2Fo ÿ Fc density fell below1s, or if the solvent moved out of electron density or outof hydrogen bonding distance. A water molecule wasnever rejected solely on the basis of its thermal factor,although it was found that atoms with very large ther-mal factors usually failed one of the other tests. The ®nalmodel contained only 33 water molecules, i.e. abouteight per Cro monomer.

In the initial isomorphous replacement electron den-sity map, the electron density for the seven carboxy-terminal residues (60 to 66) was very weak, suggestingthat this part of the molecule was disordered, or partiallydisordered (Anderson et al., 1981; Ohlendorf et al., 1982).During re®nement, each monomer was truncated to resi-due 59 and considerable effort was spent inspectingFo ÿ Fc and 2Fo ÿ Fc maps for evidence for the preferredconformation of the remaining seven residues. In each ofthe four monomers, however, the electron densityremained very weak beyond residue 60 or 61, suggestingthat the remaining carboxy-terminal residues are, in fact,disordered, both in the crystals and in solution.

Considerable dif®culty was encountered ininterpreting the electron density centered on the crystal-lographic 3-fold axis. The ®nal model contains a phos-phate ion (labeled 1PO4) located on the 3-fold axis andanother phosphate ion (labeled 2PO4) located nearby butoffset from the axis of symmetry. The molecular 3-foldaxis of the ion 1PO4 is not coincident with the crystallo-graphic 3-fold axis. Each of the phosphates ions was

Table 2. Re®nement statistics

Resolution (AÊ ) 20-2.3Reflections 17,141Completeness of data (%) 84Number of atoms 1940Crystallographic R-value (%) 19.3Average bond length discrepancy (AÊ ) 0.019Average bond angle discrepancy (deg.) 3.3Average torsion angle discrepancya (deg.) 19.6Average deviation from planarity (trigonal groups) (AÊ ) 0.008Average deviation from planarity (general) (AÊ ) 0.014Close van der Waals contacts 35B-factor discrepancyb (AÊ 2) 7.3

a Torsion angles were not restrained during re®nement.b The B-factor discrepancy gives the average deviation of the

thermal factors from expected values based on well-determined,high-resolution crystal structures (Tronrud, 1996).


modeled with its occupancy set to 1/6. 1PO4 makes ahydrogen bond to the hydroxyl group of Ser28 of mono-mer O, and another hydrogen bond to a water molecule.In addition, it makes a salt-bridge to the side-chain ofLys32 of monomer O. Since the ion is located on thecrystallographic 3-fold axis, these interactions can, inprinciple, be repeated three times, although a singlephosphate ion cannot engage in all nine interactions sim-ultaneously. Phosphate ion 2PO4 makes a salt link withLys32 from monomer O, and possibly with Lys32 in asymmetry-related monomer.

As shown in Figure 2, all of the non-glycine residuesincluded in the re®ned model except Ser60 are withinthe allowed regions of the Ramachandran diagram, 94%being in the most-favored regions as de®ned byLaskowski et al. (1993).

Representative portions of the ®nal 2Fo ÿ Fc electrondensity map are shown in Figure 3. These include Pro59,which has a cis conformation (Figure 3(a)), and the side-chain of Phe58 of one monomer, which is buried withinthe core of a 2-fold-related monomer (Figure 3(b)). There®nement statistics are presented in Table 2.

The re®ned coordinates have been deposited in theBrookhaven Protein Data Bank for immediate release(access code 5CRO).

Acknowledgements

We thank Drs Larry Weaver, Jim Remington andTerry Gray for discussion and suggestions through thecourse of this project, and Ron Albright for the prep-aration of Figure 1. This work was supported in part bya grant from the National Institutes of Health (GM20066)to B.W.M. and a postdoctoral fellowship from theNational Institutes of Health (GM08403) to D.H.O.

References

Albright, R. A. & Matthews, B. W. (1998). Crystal struc-ture of l-Cro bound to a consensus operator at3.0 AÊ resolution. J. Mol. Biol. 280, 137±151.

Albright, R. A., Mossing, M. C. & Matthews, B. W.(1996). High-resolution structure of an engineeredCro monomer shows changes in conformation rela-tive to the native dimer. Biochemistry, 35, 735±742.

Anderson, W. F., Takeda, Y., Echols, H. & Matthews,B. W. (1979). The structure of a repressor: Crystallo-

graphic data for the cro regulatory protein of bac-teriophage l. J. Mol. Biol. 130, 507±510.

Anderson, W. F., Ohlendorf, D. H., Takeda, Y. &Matthews, B. W. (1981). Structure of the cro repres-sor from bacteriophage l and its interaction withDNA. Nature, 290, 754±758.

Anderson, W. F., Takeda, Y., Ohlendorf, D. H. &Matthews, B. W. (1982). Proposed a-helical super-secondary structure associated with protein-DNArecognition. J. Mol. Biol. 159, 745±751.

Brennan, R. G. & Matthews, B. W. (1989). The helix-turn-helix DNA-binding motif. J. Biol. Chem. 264,1903±1906.

Brennan, R. G., Roderick, S. L., Takeda, Y. & Matthews,B. W. (1990). Protein-DNA conformational changesin the crystal structure of a lambda Cro-operatorcomplex. Proc. Natl Acad. Sci. USA, 87, 8165±8169.

Chothia, C. & Lesk, A. M. (1986). The relation betweenthe divergence of sequence and structure in pro-teins. EMBO J. 5, 823±826.

Jack, A. & Levitt, M. (1978). Re®nement of large struc-tures by simultaneous minimization of energy andR factor. Acta Crystallog. sect. A, 34, 931±935.

Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein struc-tures. J. Appl. Crystallog. 26, 283±291.

Lee, B. & Richards, F. M. (1971). The interpretation ofprotein structures: estimation of static accessibility.J. Mol. Biol. 55, 379±400.

Matsuo, H., Shirakawa, M. & Kyogoku, Y. (1995).Three-dimensional dimer structure of the l-Crorepressor in solution as determined by heteronuc-lear multidimensional NMR. J. Mol. Biol. 254, 668±680.

Nicholson, H., Becktel, W. J. & Matthews, B. W. (1988).Enhanced protein thermostability from designedmutations that interact with alpha-helix dipoles.Nature, 336, 651±656.

Nicholson, H., Anderson, D. E., Dao-pin, S. &Matthews, B. W. (1991). Analysis of the interactionbetween charged side-chains and the a-helix dipoleusing designed thermostable mutants of phage T4lysozyme. Biochemistry, 30, 9816±9828.

Ohlendorf, D. H., Anderson, W. F., Fisher, R. G.,Takeda, Y. & Matthews, B. W. (1982). The molecu-lar basis of DNA-protein recognition inferred fromthe structure of cro repressor. Nature, 298, 718±723.

Ohlendorf, D. H., Anderson, W. F. & Matthews, B. W.(1983). Many gene-regulatory proteins appear tohave a similar a-helical fold that binds DNA andevolved from a common precursor. J. Mol. Evol. 19,109±114.

Pabo, C. O. & Sauer, R. T. (1984). Protein-DNA recog-nition. Annu. Rev. Biochem. 53, 293±321.

Pakula, A. A. & Sauer, R. T. (1989). Amino acid substi-tutions that increase the thermal stability of the lCro protein. Proteins: Struct. Funct. Genet. 5, 202±210.

Pakula, A. A. & Sauer, R. T. (1990). Reverse hydro-phobic effects relieved by amino-acid substitutionsat a protein surface. Nature, 344, 363±364.

Ptashne, M. (1986). A Genetic Switch. Gene Control andPhage l, Blackwell, Palo Alto, CA.

Richardson, J. S. (1981). The anatomy and taxonomy ofprotein structure. Advan. Protein Chem. 34, 167±339.

Rossmann, M. G. (1979). Processing oscillation diffrac-tion data for very large unit cells with an automatic


convolution technique and pro®le ®tting. J. Appl.Crystallog. 12, 225±239.

Schellman, C. (1980). The aL conformation at the ends ofhelices. In Protein Folding (Jaenicke, R., ed.),pp. 53±61, Elsevier/North-Holland BiomedicalPress, New York.

Schmid, M. F., Weaver, L. H., Holmes, M. A., GruÈ tter,M. G., Ohlendorf, D. H., Reynolds, R. A.,Remington, S. J. & Matthews, B. W. (1981). Anoscillation data collection system for high-resolutionprotein crystallography. Acta Crystallog. sect. A, 37,701±710.

Takeda, Y., Folkmanis, A. & Echols, H. (1977). Cro regu-latory protein speci®ed by bacteriophage lambda.Structure, DNA-binding and repression of RNAsynthesis. J. Biol. Chem. 252, 6177±6183.

Takeda, Y., Kim, J., Caday, C. G., Steers, E., Jr,Ohlendorf, D. H., Anderson, W. F. & Matthews,B. W. (1986). Different interactions used by Crorepressor in speci®c and nonspeci®c DNA binding.J. Biol. Chem. 261, 8608±8616.

Tidor, B. (1994). Helix-capping interaction in (Cro pro-tein: a free energy simulation analysis. Proteins:Struct. Funct. Genet. 19, 310±323.

Tronrud, D. E. (1992). Conjugate-direction minimization:an improved method for the re®nement of macro-molecules. Acta Crystallog. sect. A, 48, 912±916.

Tronrud, D. E. (1996). Knowledge-based B-factorrestraints for the re®nement of proteins. J. Appl.Crystallog. 29, 100±104.

Tronrud, D. E., Ten, Eyck L. F. & Matthews, B. W.(1987). An ef®cient general-purpose least-squaresre®nement program for macromolecular structures.Acta Crystallog. sect. A, 43, 489±503.

Venkatachalam, C. M. (1968). Stereochemical criteria forpolypeptides and proteins. V. Conformation of asystem of three linked peptide units. Biopolymers, 6,1425±1436.

Wagner, G., Hyberts, S. G. & Havel, T. F. (1992). NMRstructure determination in solution: a critique andcomparison with X-ray crystallography. Annu. Rev.Biophys. Biomol. Struct. 21, 167±198.

Zhang, X.-J. & Matthews, B. W. (1995). EDPDB: a multi-functional tool for protein structure analysis. J. Appl.Crystallog. 28, 624±630.

Zhang, X.-J., Wozniak, J. A. & Matthews, B. W. (1995).Protein ¯exibility and adaptability seen in 25 crystalforms of T4 lysozyme. J. Mol. Biol. 250, 527±552.

Edited by P. E. Wright

(Received 24 December 1997; received in revised form 3 April 1998; accepted 8 April 1998)

Refined structure of Cro repressor protein from bacteriophage λ suggests both flexibility and...

Documents

Transcript of Refined structure of Cro repressor protein from bacteriophage λ suggests both flexibility and...