Proc. Natl. Acad. Sci. USAVol. 92, pp. 3444-3448, April 1995Biochemistry

Heterogeneity in molecular recognition by restrictionendonucleases: Osmotic and hydrostatic pressureeffects on BamHI, Pvu II, and EcoRV specificity

(protein-DNA recognition/star activity/bound water)

CLIFFORD R. ROBINSONt AND STEPHEN G. SLIGARtSchool of Chemical Sciences and Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign, Urbana, IL 61801

Communicated by Jiri Jonas, University of Illinois, Urbana, IL, December 5, 1994 (received for review August 18, 1994)

ABSTRACT The cleavage specificity of the Pvu II andBamHI restriction endonucleases is found to be dramaticallyreduced at elevated osmotic pressure. Relaxation in specificityof these otherwise highly accurate and specific enzymes,previously termed "star activity," is uniquely correlated withosmotic pressure between 0 and 100 atmospheres. No othercolligative solvent property exhibits a uniform correlationwith star activity for all of the compounds tested. Applicationof hydrostatic pressure counteracts the effects of osmoticpressure and restores the natural selectivity of the enzymes fortheir canonical recognition sequences. These results indicatethat water solvation plays an important role in the site-specific recognition of DNA by many restriction enzymes.Osmotic pressure did not induce an analogous effect on thespecificity of the EcoRV endonuclease, implying that selectivehydration effects do not participate in DNA recognition in thissystem. Hydrostatic pressure was found to have little effect onthe star activity induced by changes in ionic strength, pH, ordivalent cation, suggesting that distinct mechanisms may existfor these observed alterations in specificity. Recent evidencehas indicated that BamHI and EcoRI share similar structuralmotifs, while Pvu II and EcoRV belong to a different structuralfamily. Evidently, the use of hydration water to assist insite-specific recognition is a motif neither limited to nordefined by structural families.

Type II restriction endonucleases, along with their companionmethyltransferases, make up bacterial restriction/modificationsystems, which protect bacteria from foreign DNA. Theytypically consist of symmetric homodimers, which recognizesymmetric DNA sequences of four to eight bases, and requireonly Mg2+ as a cofactor to cleave duplex DNA (1). Despite thevast number of type II enzymes identified to date (nearly 2400)no homology and little sequence similarity exist (2). Despitethis lack of homology, recent reports have identified significantstructural similarities among these enzymes (3), and a newclassification of restriction enzymes has been proposed on thebasis of the location of the scissile bond (4). Recent structuralevidence has led to appreciation of structural and functionalroles for water in site-specific interactions in protein-DNAcomplexes (5-8).The use of osmotic pressure represents a powerful method

for identifying and studying the participation of hydrationwater in biomolecular systems (9). By examining the correla-tion between functional or structural properties and osmoticpressure, solvent water has been shown to play key roles inbiochemical processes, such as substrate binding (10), protein-protein interactions (11), allosteric effects and conformationalchanges (12), catalysis (13), protein stability (14), and ionchannel formation (15). We have previously utilized osmotic-

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

and hydrostatic-pressure techniques to study the role of waterin mediating site-specific interactions of the EcoRI endonu-clease with canonical and alternate DNA sequences (16, 18).To investigate whether this nontraditional role for water that

was discovered in EcoRI represents a common feature ofrestriction enzyme systems, we have examined the Pvu II,BamHI, and EcoRV endonucleases. Along with EcoRI, theyare the only members of the type II restriction endonucleasefamily to be characterized by high-resolution structural anal-ysis (4, 19-23). For these three enzymes, cleavage at alternatesites has been observed previously, either by addition oforganic solvents to the reaction buffer or by changes in ionicstrength, pH, or divalent cation concentration or identity(24-28). Cleavage at alternate sites by these and other restric-tion enzymes has been termed "star activity" by analogy withEcoRI star (EcoRI*) activity (29).Each subunit of the Pvu II endonuclease from Proteus

vulgaris has 157 amino acids, and the homodimer recognizessymmetric CAGCTG duplex DNA sequences, cleaving bothstrands in the middle of the site to leave blunt ends (30). PvuII* sites include all single base substitutions at the four internalpositions (CNGCTG, CANCTG, CAGNTG, and CAGCNG),as well as A or G at the first position and T or C at the sixthposition (24). The dimeric BamHI endonuclease from Bacillusamyloliquefaciens H consists of identical 213-amino acid resi-due subunits, and cleaves GGATCC duplex DNA sequencesbetween the two guanine bases on both strands, leavingfour-base 5' overhangs (31). Reported BamHI* sites includeGGAACC, GGCTCC, GAATCC, GGATCN, and comple-mentary sequences. The EcoRV endonuclease derives fromthe restriction/modification system in Escherichia coli. EcoRVis a dimer of identical subunits of 244 amino acids. It recog-nizes the symmetric GATATC duplex DNA sequences andcleaves both strands in the middle of the site, leaving blunt ends(32). EcoRV has been reported to cut at alternate sites withsingle base changes from the canonical site, including GT-TATC, AATATC, GANATC, and complementary sequences(33, 34).

In this study we test whether the participation of water is ageneral phenomenon in restriction enzyme-DNA interactions,thereby seeking to advance a general understanding of themesin protein-DNA recognition and to provide molecular expla-nations for the flawed molecular recognition events that aremanifested as restriction enzyme star activity.

MATERIALS AND METHODSFor each enzyme, a standard buffer was defined in whichcleavage was observed only at the canonical site. Sets of

Abbreviations: DMSO, dimethyl sulfoxide; atm, atmosphere.tPresent address: Department of Biology, Massachusetts Institute ofTechnology, Cambridge, MA 02139ITo whom reprint requests should be addressed.

3444

Proc. Nati Acad Sci. USA 92 (1995) 3445

reactions were performed in the presence of increasing con-centrations of the various compounds (osmolytes) added to thestandard buffer. Enzyme reactions were initiated by additionof enzyme to the solution. Order of addition had no observableeffect. Enzymes were incubated with DNA containing canon-ical and alternate ("star") sites in total volumes of 20 p.l at 37°Cfor 4 h and then quenched by addition of EDTA to a finalconcentration of 20 mM.

Restriction enzymes were purchased from GIBCO/BRL.Purified pUC18 DNA was prepared by using Promega MagicMega-Preps, from 1 liter cultures of E. coli grown in mediumcontaining 200 mg of ampicillin. Reaction mixtures contained1 unit/,ul of enzyme and 0.5 ,g/,ul of pUC18 DNA, unlessotherwise noted. For EcoRV, enzymatic activity at the canon-ical GATATC site was investigated by using purified A DNAobtained from GIBCO/BRL. EcoRV* activity was tested byusing linear or circular pUC18 or A DNA. All three forms ofDNA gave equivalent results.The standard buffers which allowed canonical site cleavages

were: Pvu II, 50 mM Tris HCl, pH 7.4/6 mM MgCl2/50 mMKCl/50 mM NaCl; BamHI, 50 mM Tris HCl, pH 8.0/2 mM2-mercaptoethanol/10 mM MgCl2/100 mM NaCl; EcoRV, 50mM Tris HCl, pH 7.5/100 mM NaCl/10 mM MgCl2/10 mM2-mercaptoethanol/100 ,tg of bovine serum albumin per ml.Glycerol, ethylene glycol, dimethyl sulfoxide (DMSO), etha-nol, 2-propanol, sucrose, and dextrose were added to thestandard buffers to final concentrations of 0-6 M to induceosmotic pressures from 0 to 150 atmospheres (atm; 1 atm =101.3 kPa).

Reactions at elevated hydrostatic pressures ranging from 1to 1600 atm were performed as described (16) in 400-,lIpolypropylene tubes (Cole-Palmer) filled with 0.2 g of zirco-nium oxide beads (Biospec Products, Bartlesville, OK). Solu-tions of enzyme and DNA were separated by a thin layer ofmineral oil (Sigma) until the samples reached the desiredpressure and temperature, at which time the apparatus wasinverted to mix the solutions. Following a 4-h incubation at37°C, the enzymes were inactivated by heating to 65°C andEDTA was added to 20 mM.The determinations of canonical site activity as a function of

hydrostatic pressure were performed by using the standardbuffer for each enzyme. For the analysis of hydrostatic pres-sure effects upon star activity induced by osmotic pressure inPvu II and BamHI, glycerol was added to the standard bufferin concentrations from 0 to 4.8 M to induce osmotic pressuresfrom 0 to 120 atm. The analysis of hydrostatic pressure effectsupon star activity induced by changes in ionic strength, pH, ordivalent cation concentration or identity was performed in thefollowing buffers, with closed, circular pUC18 DNA as thesubstrate: Pvu II, 10 mM Tris-HCl, pH 8.5/15 mM MgCl2;BamHI, 20 mM Tris HCl, pH 8.8/2 mM MnCl2; EcoRV, 50mM Tris-HCl, pH 8.5/10 mM MnCl2/10 mM 2-mercaptoetha-nol/100 ,tg of bovine serum albumin per ml.

Analysis of the reaction products was as described (16, 17).The fraction of star activity was defined for each reaction as

0~

._

1.0

Osmotic pressure, atm

the intensity of the bands produced by cleavage at star sitesdivided by the intensity of all bands (canonical site fragmentsplus star site fragments). A fraction star activity of 0.5 meansthat half of the plasmids cut at the canonical sites were also cutat a star site. In all cases in which fraction star activities weredetermined, restriction mapping confirmed that cleavage re-actions proceeded to completion at the canonical sites beforeany star sites were cut. Osmotic pressures were determined asdescribed previously (16) by using a UIC 70 vapor-phaseosmometer or by calculation from tabulated data (35). Vis-cosities and dielectric constants were taken from tabulateddata (35, 36).

RESULTSThe pUC18 plasmid contains two canonical Pvu II sites(CAGCTG), at positions 306 and 628. Cleavage at these sitesyields fragments of 2368 and 318 bp. The most readily cleavedPvu II* site in pUC18 was CAGCGG at position 112, gener-ating fragments of 2174, 318, and 194 bp. The presence ofosmolytes in the reaction buffer promoted cleavage at alter-nate sites in pUC18 yielding additional fragments. DMSO,ethylene glycol, glycerol, ethanol, and 2-propanol promotedPvu II* activity. Star activity is tightly correlated with osmoticpressure (Fig. 1A) but correlated less well with dielectricconstant (Fig. 2A) or relative viscosity (Fig. 2B).

Cleavage at the unique canonical BamHI site in pUC18(position 429) linearizes the plasmid. There are 11 potentialBamHI* sites in pUC18. Addition ofDMSO, glycerol, ethyleneglycol, sucrose, or dextrose promoted BamHI* site cleavage,yielding a distinctive band pattern consisting of eight newDNAfragments. The fraction of star activity was quantitated fromthese bands and was tightly correlated with osmotic pressure(Fig. 1B) but not well correlated with dielectric constant (Fig.2C) or relative viscosity (Fig. 2D).No canonical EcoRV sites exist within pUC18, but the

plasmid contains seven EcoRV* sites. Cleavage at these siteswas not promoted by any of the compounds tested exceptDMSO, regardless of whether circular or linear pUC18 wasused as the substrate. None of the compounds tested appearedto inhibit EcoRV canonical site activity, as judged by analysisof EcoRV cleavage of A DNA (data not shown).We sought to apply hydrostatic pressure as a probe of

restriction enzyme specificity. First, the effect of elevatedhydrostatic pressure upon the activity of the three enzymes attheir canonical sites was tested under standard buffer condi-tions (Table 1). Pvu II and BamHI were unaffected by increas-ing the pressure from 0 to 750 atm, showed only small effectsat 1000 atm, and still retained most of their activity above 1500atm. EcoRV was more sensitive to hydrostatic pressure, butretained most of its activity up to 1000 atm.When hydrostatic pressure is applied to reactions with

elevated osmotic pressure, a dramatic reversal of the osmoticpressure effects upon specificity is observed for Pvu II andBamHI (Fig. 3). Hydrostatic pressure applied from 1 to 500

FIG. 1. Effects of osmotic pres-sure generated by various sub-stances on the enzyme specificityof Pvu II (A) or BamHI (B). 0,DMSO; *, ethylene glycol; O, glyc-erol; *, ethanol; X, 2-propanol; +,sucrose; and a, dextrose. Lines rep-resent the best fit by least squaresanalysis, with the correlation coef-ficients indicated.

I

Biochemistry: Robinson and Sligar

3446 Biochemistry: Robinson and Sligar

74 76 78Dielectric constant, E,

c)

0

.!2cn

V.,

1.2

._

,, 0.8CuXu 0.60

0 0.4cu

v 0.2

00 1 2 3 4 5 6

Relative viscosity, 71/710

- FIG. 2. Effects of viscosity (Band D) and dielectric constant (Aand C) on enzyme specificity. (Aand B) Pvu II. (C and D) BamHI.0, DMSO; *, ethylene glycol; O,glycerol; 0, ethanol; X, 2-propanol;+, sucrose; and A\, dextrose. Solidlines represent best fits for eachcompound. Dashed lines representthe best fit considering data fromall compounds as a single data set,

7 8 with the correlation coefficients in-dicated.

atm inhibits and ultimately eliminates the Pvu II* and BamHI*activities induced by osmotic pressure from 0 to 100 atm.

Intriguingly, alterations in enzyme specificity induced byother changes in buffer conditions are not sensitive to hydro-static pressure. When star activity is induced by decreasedionic strength, increased pH, or changes in divalent cationconcentration or identity, hydrostatic pressure up to 500 atminhibits only a small amount (<15%) of the Pvu II* andBamHI* activities. Above 500 atm no further inhibition isobserved up to 1400 atm. ForEcoRV, hydrostatic pressures upto 1200 atm did not measurably inhibit the star activity inducedby increased pH (from 7.5 to 8.5), decreased ionic strength(eliminating 100mM NaCl), or substitution of Mn2+ for Mg2+.

DISCUSSIONIn the cases of BamHI and Pvu II we observe a change inspecificity upon the addition of osmolyte, resulting in cleavageat alternate, noncanonical sites. The extent of this cleavage istightly correlated with osmotic pressure generated by theaddition of neutral solutes or cosolvents. All compounds thatwere effective at promoting cleavage at star sites induced afraction star activity uniformly proportional to the osmoticpressure (Fig. 1). The addition of these compounds to thereaction buffer changes other colligative solution properties,such as viscosity and dielectric constant, which have beenshown to play a role in biomolecular interactions (37, 38). Aninterpretation of the results as being related to osmotic effectsrequires an analysis of these other properties. Viscosity anddielectric constant are poorly correlated with the observedfraction star activity over the concentration range tested (Fig.2). The variety of neutral solutes and cosolvents tested effec-tively rules out a specific interaction between the compounds

Table 1. Inactivation of restriction enzymes byhydrostatic pressure

Percent of ambient pressure activity atvarious pressures*

Enzyme 750 atm 1000 atm 1600 atm

Pvu II 100 98 90BamHI 100 96 79EcoRV 94 82 41

*Values have errors of approximately 7 percentage points.

and the enzymes or DNA. The simplest interpretation of theseresults is that a population of bound waters, sequestered frombulk solvent, is present in the complexes between the restric-tion enzyme and canonical site DNA but is not present whenthe enzyme is associated with the star sites. Release of thesewaters to bulk solvent is promoted by elevated osmotic pres-sure in the solution, and enhances the affinity or activity for theenzyme at alternate sites. These data are consistent with thestudies of Spolar and Record, who suggested that a couplingexists between local folding, water release, and site-specificbinding ofDNA in the EcoRI endonuclease and other protein-DNA complexes (39). As reported previously, an analogouseffect occurs with EcoRI (16, 17). No such effects wereobserved with the EcoRV endonuclease, suggesting that hy-dration waters are not utilized in the same way in the EcoRV-DNA complex.At elevated hydrostatic pressures, the influence of osmotic

agents on the specificity of Pvu II and BamHI is greatlydiminished. The natural selectivity of the enzymes observedunder standard conditions is restored, even in the presence ofan osmotic pressure of 100 atm, by application of hydrostaticpressure (Fig. 3). One possibility is that the effects of hydro-static pressures to 500 atm are due to a general inhibition ofthe enzyme; however, overall activity of Pvu II and BamHI atthe canonical site was unchanged between 0 and 750 atm(Table 1). Our interpretation is that the return of selectivitypromoted by hydrostatic pressure is due to specific solvationeffects which alter the total volume of the system. Hydrostaticpressure has been shown in many cases to preferentiallyhydrate biomacromolecules and complexes (40, 41). A simpleexplanation for our results is that waters of solvation, releasedby elevated osmotic pressure, are restored by hydrostaticpressure, perhaps due to the participation of discrete waters insite discrimination by Pvu II and BamHI. The inhibition of theoverall enzymatic activity at the canonical sites observed athydrostatic pressures in excess of 1500 atm is more likely torepresent dissociation of the protein-DNA or dimeric protein-protein complexes (or both), as has been observed previously(40, 42, 43). Strikingly, hydrostatic pressure has very differenteffects upon star activity that is induced by changes in ionicstrength, pH, or divalent cations. This contrast between theresponses of the two types of star activity could indicatedistinct origins for the changes in specificity promoted by thetwo methods.

0.8:E< 0.6

- 0.4

v 0.2

0

.2

Q 0.8

Cu

- 0.6

0

5:

0 0.4

Cu

0.2 :

0772

Proc. NatL AcatL Sci. USA 92 (1995)

r

Proc. NatL Acad Sci. USA 92 (1995) 3447

it,

C13

._Cee

0.8

0.6

0.4

0.2

Osmotic pressure, atm

We can speculate that for Pvu II and BamHI, elevatedosmotic pressures result in conformational changes accompa-nied by release of bound waters, which are reversed byhydrostatic pressure. For the other class of changes whichinduced star activity, the small change in Pvu II and BamHIstar activity observed with pressure could indicate that alimited conformation change is induced by the ionic/pH/cation changes (and reversible by hydrostatic pressure) but thatmost of the effects are electrostatic.

In addition to identifying a role for bound water in recog-nition by Pvu II and BamHI, the results presented here arguethat fundamental differences exist in the mechanisms of sitediscrimination among type II restriction enzymes. EcoRI haspreviously been shown to display the same behavior as Pvu IIand BamHI (16, 17). For all three enzymes, increases in staractivity accompany increases in osmotic pressure, while hy-drostatic pressure restores specificity. In contrast, EcoRV*activity was not induced by a simple application of osmoticpressure. We conclude that a differential solvation is linked tothe recognition ofDNA by EcoRI, BamHI, and Pvu II, but notEcoRV. Furthermore, the response of EcoRV to hydrostaticpressures is also different from the other three restrictionenzymes. For Pvu II and BamHI, hydrostatic pressure not onlyreverses the effects of osmotic pressure on specificity, it alsohas a small but measurable impact on star activity induced bychanges in ionic strength, pH, or divalent cation. This obser-vation is also true for EcoRI (16, 18). By contrast, in theEcoRV system, hydrostatic pressure has no observable effectupon ionic strength-, pH-, or cation-induced star activity,supporting the hypothesis that fundamental differences existbetween the mechanisms of molecular recognition in theserestriction enzyme-DNA complexes.EcoRV differs substantially from other type II restriction

enzymes in that in the absence of the "catalytic cofactor"Mg2+, it binds with equal affinity to cognate and noncognateDNA. Binding specificity (and cleavage) is only achieved in thepresence of Mg2+ (or Mn2+ or other divalent cations; refs. 34and 44). Other type II enzymes can discriminate cognate fromnoncognate sites with high selectivity (103-105) in the absenceof magnesium (2). ThusEcoRV may rely upon interaction withthe divalent cation to help direct its specificity, while the otherenzymes utilize bound waters for a similar function.

Despite the lack of sequence homology among restrictionenzymes, recent x-ray crystallographic studies of the EcoRI,Pvu II, BamHI, and EcoRV endonucleases have indicated thatEcoRI and BamHI belong to the same structural class, whilePvu II and EcoRV appear to be members of a distinctstructural family (3, 4, 19-23). Another feature which maydefine these classes of enzymes is the cleavage position withintheir recognition sequences. EcoRI and BamHI cut after thefirst base in a six-base site, leaving four-base 5' overhangs,while Pvu II and EcoRV cut both stands in the middle of asix-base site, leaving blunt ends. Given the apparent correla-tion between structure and functional properties, it might havebeen expected that DNA-recognition motifs would be alsopreserved within these putative enzyme families. However, the

0 20 40 60 80 100 120 140

FIG. 3. Effect of hydrostaticpressure on altered specificity in-duced by osmotic pressure for PvuII (A) or BamHI (B). 0, Ambientpressure; 0, 100 atm; *, 200 atm;K, 300 atm; X, 400 atm; and A, 500atm. Lines represent the best fit byleast squares analysis.

results presented in this communication indicate that the roleof differential hydration in recognition is a widely used motif,neither limited to nor defined by the emerging subclasses oftype-II restriction endonucleases, and one which is not readilydetected by structural analysis.While the existence of structurally or functionally significant

waters in these systems is clear, we can as yet only speculate ontheir exact locations and roles. One possibility is that boundwaters are directly involved in contacts between the proteinand DNA, as observed in the tip repressor complex withcognate DNA (6, 7). Although information on the presence ofwaters in the protein-DNA interface of the Pvu II and EcoRVstructures was not available at the time of this writing, twowaters do mediate base-specific contacts in the EcoRI-DNAcomplex (45). EcoRV binds to noncognate DNA with modestaffinity in a relatively loose conformation believed to benecessary for facilitated diffusion of the enzyme along theDNA helix (32). When shifting from noncognate to cognateDNA, the protein tightens its grip, completely surrounding theDNA helix and making specific contacts in the major andminor grooves. This transition results in an increase in theburied surface area of >1800 A2, including 230 A2 from newprotein-protein contacts (21). Since an increase in buriedsurface area is often accompanied by release of bound water(10-12, 39, 46), elevated osmotic pressures could serve toenhance the affinity of this complex. Pvu II, BamHI, andEcoRI bind with low affinity to noncognate DNA (2) andmake specific contacts primarily in the major groove. For theseenzymes, recognition might be facilitated by formation of arelatively "loose" complex (47, 48) or one in which confor-mational heterogeneity is required for optimal binding (49).Release of water from the complex could result in an altered,tighter, configuration conferring a different specificity. Theobservation of discrete waters in the EcoRI-DNA interface(45) is consistent with this model.Another possibility is that hydrostatic and osmotic pressures

are influencing the conformational equilibrium of the twomonomer subunits of these dimeric enzymes and their com-plexes with cognate and noncognate sequences. For example,in many oligomeric proteins, pressure-induced dissociationleads to conformational drifting of the monomers (50, 51).Upon reassociation, these proteins or enzymes can displaysubtly different characteristics and activities (52).

Solvent molecules might serve to orient the two enzymemonomers with respect to each other or direct the conforma-tion of individual enzyme subunits. This idea is supported bythe structural evidence. The Pvu II structure is similar toEcoRV only in its DNA-binding domain; the dimerizationdomains of the two proteins are very different (20). Thestructure of BamHI is most similar to EcoRI in the dimeriza-tion domain (22).Water can also play a role in controlling the conformation

of DNA by affecting the susceptibility of a particular sequenceto deformation, binding, or cleavage. DNA conformation isextremely sensitive to solution conditions, and changes insolvation can have dramatic effects upon its deformability (53,

B

0 W.

Biochemistry: Robinson and Sligar

3448 Biochemistry: Robinson and Sligar

54). In the EcoRV-cognate DNA complex, unstacking of thecentral 2 bp results in a sharp bend of the helix and a narrowingof the major groove, while the minor groove is widened. Thisdistortion of the B-form helix axis, coupled with binding byEcoRV in the minor groove, may displace the minor groovehydration waters (55). Osmotic pressure would then serve tofacilitate site-specific binding. In contrast, Pvu II and EcoRI donot distort the DNA helix axis and may therefore preservehydration of the minor groove. Release of these waters couldalter the preferred conformation of cognate and noncognateDNA, altering enzyme specificity.

It has been proposed that the effect of cosolvents to enhancecleavage at noncanonical sites simply results from increasedaffinity of the enzymes for DNA at all positions due to globaldehydration or excluded volume effects (56). Three lines ofevidence argue against this explanation. (i) In both the Pvu IIand BamHI systems reported here and in the EcoRI systemreported previously (16), the rate of cleavage at the canonicalsite decreases as osmotic pressure increases, while cleavage atthe star site increases. This shift in specificity would not beexpected if the enzymes were simply displaying enhancedbinding or cleavage at all sites. (ii) No decrease in activity isobserved at the canonical site under standard buffer condi-tions when hydrostatic pressure is applied. This implies that themechanism by which hydrostatic pressure reverses the effectsof cosolvents does not involve dissociation of the complex. (iii)Although under standard conditions EcoRV discriminatesbetween canonical and alternate sites to a similar degree as theother three enzymes, its specificity is unaffected by the addi-tion of neutral solutes or cosolvents. If these compounds weresimply lowering the dissociation constant at all sites by a feworders of. magnitude, EcoRV would be expected to show anenhanced level of star activity as well.

We thank Professor Gregorio Weber, Professor Richard Gumport,Dr. Eric Fisher, Jason Rockhill, and Dr. Anne Skaja Robinson forhelpful discussions and advice; Professor Steven Zimmerman for theuse of his vapor pressure osmometer; and Ms. Aretta Weber forassistance in manuscript preparation. This work was supported byNational Institutes of Health Grants GM31756 and GM33775 (S.G.S.)and by National Institutes of Health National Research Service Awardin Molecular Biophysics Grant GM08276 (C.R.R.).

1. Fisher, E. W. & Gumport, R. I. (1995) in The Encyclopedia ofMolecular Biology and Biotechnology: Fundamentals and Appli-cations, ed. Meyers, R. A. (VCH, Weinhein, Germany), in press.

2. Bennet, S. P. & Halford, S. E. (1989) Curr. Top. Cell. Regul. 30,57-104.

3. Riddihough, G. (1994) Nature (London) 370, 78.4. Newman, M., Strzelecka, T., Dorner, L. F., Schildkraut, I. &

Aggarwal, A. K. (1994) Structure 2, 439-452.5. Ha, J.-H., Capp, M. W., Hehenwalter, M. D., Baskerville, M. &

Record, M. T., Jr. (1992) J. Mol. Biol. 228, 252-264.6. Lawson, C. L. & Carey, J. (1993) Nature (London) 366, 178-182.7. Otwinowski, Z., Schevitz, R. W., Zhang, R.-G., Lawson, C. L.,

Joachimiak, A., Marmorstein, R. Q., Luisi, B. F. & Sigler, P. B.(1988) Nature (London) 335, 321-329.

8. Feng, J. A., Johnson, R. C. & Dickerson, R. E. (1994) Science263, 348-355.

9. Parsegian, V. A., Rand, R. P., Fuller, N. L. & Rau, D. C. (1986)Methods Enzymol. 127, 400-437.

10. Rand, R. P., Fuller, N. L., Butko, P., Francis, G. & Nicholls, P.(1993) Biochemistry 32, 5925-5929.

11. Kornblatt, J. A., Kornblatt, M. J., Hui Bon Hoa, G. & Mauk,A. G. (1993) Biophys. J. 65, 1059-1065.

12. Colombo, M. F., Rau, D. C. & Parsegian, V. A. (1992) Science256, 655-659.

13. Di Primo, C., Sligar, S. G., Hui Bon Hoa, G. & Douzou, P. (1992)FEBS Lett. 312, 252-254.

14. Santoro, M. M., Liu, Y., Khan, S. M. A., Hou, L.-X. & Bolen,D. W. (1992) Biochemistry 31, 5278-5283.

15. Zimmerberg, J. & Parsegian, V. A. (1986) Nature (London) 323,36-38.

16. Robinson, C. R. & Sligar, S. G. (1994) Biochemistry 33, 3787-3793.

17. Robinson, C. R. & Sligar, S. G. (1993) J. Mol. Biol. 234, 302-306.18. Robinson, C. R. & Sligar, S. G. (1995) Methods Enzymol., in

press.19. Kim, Y., Grable, J. C., Love, R., Greene, P. J. & Rosenberg, J. M.

(1990) Science 249, 1307-1309.20. Athanasiadis, A., Vlassi, M., Kotsifaki, D., Tucker, P. A., Wilson,

K. S. & Kokkinidis, M. (1994) Nat. Struct. Biol. 1, 469-475.21. Winkler, F. K., Banner, D. W., Oefner, C., Tsernoglou, D.,

Brown, R. S., Heathman, S. P., Bryan, R. K., Martin, P. D.,Petratos, K. & Wilson, K. S. (1993) EMBO J. 12, 1781-1795.

22. Newman, M., Strzelecka, T., Dorner, L. F., Schildkraut, I. &Aggarwal, A. K. (1994) Nature (London) 368, 660-664.

23. Cheng, X., Balendirian, K., Schildkraut, I. & Anderson, J. E.(1994) EMBO J. 13, 3927-3935.

24. Nasri, M. & Thomas, D. (1987) NucleicAcids Res. 15, 7677-7687.25. George, J. & Chirikjian, J. G. (1982) Proc. Natl. Acad. Sci. USA

79, 2432-2436.26. George, J., Blakesley, R. W. & Chirikjian, J. G. (1981) J. Biol.

Chem. 255, 6521-6524.27. Kolesnikov, V. A., Zinoviev, V. V., Yashina, L. N., Karginov,

V. A., Baclanov, M. M. & Malygin, E. G. (1981) FEBS Lett. 132,101-104.

28. Malyguine, E., Vannier, P. & Yot, P. (1980) Gene 8, 163-177.29. Polisky, B., Greene, P., Garfin, D. E., McCarthy, B. J., Goodman,

H. M. & Boyer, H. W. (1975) Proc. Natl. Acad. Sci. USA 72,3310-3314.

30. Gingeras, T. R., Greenough, L., Schildkraut, L. & Roberts, R. I.(1981) Nucleic Acids Res. 9, 4525-4536.

31. Roberts, R. J., Wilson, G.A. & Young, F. E. (1977) Nature(London) 265, 82-84.

32. Vipond, I. B. & Halford, S. E. (1993) Mol. Microbiol. 9, 225-231.33. Luke, P. A., McCallum, S. A. & Halford, S. E. (1987) in Gene

Amplification and Analysis, ed. Chirikjian, J. G. (Elsevier, NewYork), Vol. 5, pp. 183-205.

34. Taylor, J. D. & Halford, S. E. (1989) Biochemistry 28, 6198-6207.35. Wolf, A. V., Brown, M. G. & Prentiss, P. C. (1987) in CRC

Handbook of Chemistry and Physics: 1986-1987, ed. Weast, R. T.(CRC, Boca Raton, FL), pp. D227-D271.

36. Dean, J. A. (1985) in Lange's Handbook of Chemistry, ed. Lange,N. A. (McGraw-Hill, St. Louis), Vol. 13, pp. 10-97-10-115.

37. Dzingeleski, G. D. & Wolfenden, R. (1993) Biochemistry 32,9143-9147.

38. Beece, D., Eisenstein, L., Frauenfelder, H., Good, D., Marden,M. C., Reinisch, L., Reynolds, A. H., Sorensen, L. B. & Yue,K. T. (1980) Biochemistry 19, 5147-5157.

39. Spolar, R. S. & Record, M. T., Jr. (1994) Science 263, 777-784.40. Weber, G. (1993) J. Phys. Chem. 97, 7108-7115.41. Weber, G. & Drickamer, H. G. (1983) Quart. Rev. Biophys. 16,

89-112.42. Silva, J. L. & Siliveira, C. F. (1993) Protein Sci. 2, 945-950.43. Peng, X., Jonas, J. & Silva, J. L. (1993) Proc. Natl. Acad. Sci. USA

90, 1776-1780.44. Vermote, C. L. M. & Halford, S. E. (1992) Biochemistry 31,

6082-6089.45. Rosenberg, J. (1991) Curr. Opin. Struct. Biol. 1, 104-113.46. Rand, R. P. (1992) Science 256, 618.47. Douzou, P. (1994) Proc. Natl. Acad. Sci. USA 91, 1657-1661.48. Hoffman, B. M. & Ratner, M. A. (1987) J. Am. Chem. Soc. 109,

6237-6243.49. Weber, G. (1992) Protein Interactions (Chapman & Hall, New

York), p. 293.50. King, L. & Weber, G. (1986) Biochemistry 25, 3632-3637.51. Ruan, K. & Weber, G. (1993) Biochemistry 32, 6295-6301.52. Silva, J. L. & Weber, G. (1993) Annu. Rev. Phys. Chem. 44,

89-113.53. Kimura, T., Asai, T. & Imai, M. (1989) Mol. Gen. Genet. 219,

69-78.54. Terry, B. J. & Modrich, P. (1983) J. Biol. Chem. 253, 9820-9825.55. Liepinsh, E., Otting, G. & Wuthrich, K. (1992) Nucleic Acids Res.

20, 6549-6557.56. Nasri, M. & Thomas, D. (1986) Biomed. Biochim. Acta 45,

997-1005.

Proc. Natl. Acad Sci. USA 92 (1995)