J. Mol. Riol. (l&O) 211, 211-220

Neutron

and Light-scattering Studies of DNA Gyrase and its Complex with DNA

Susan Krueger’, Giuseppe Zaccai’, Alexander Wlodawer3, Joerg Langowski4 Mary O’Dea5, Anthony Maxwell6 and Martin GellerP ‘National

Institute of Standards and Technology Gaithersburg, MD 20899, U.S.A. Laue-Langevin 38042 Grenoble Cedex, France

21nstitut

156X,

‘NCI-Frederick BRI - Basic Research Program,

Cancer Research Facility P.O. Box B, Frederick, MD

21701, U.S.A.

4European Molecular Biology Laboratory 156X, 38042 Grenoble Cedex, France National

5Laboratory of Molecular Biology, NIDDK Institutes of Health, Bethesda, MD 20892, U.S.A.

University (Received 26 April

‘Department of Biochemistry of Leicester, Leicester LEl 1989, and in revised form

7 RH, U.K. 24 August

1989)

The solution structure of Escherichia coli DNA gyrase, an enzyme that catalyzes the ATP-dependent supercoiling of DNA, has been characterized by small-angle neutron scattering (SANS) and dynamic light-scattering (DLS). The enzyme and its complex with a 172 base-pair fragment of duplex DNA, in H,O or ‘HZ0 solvent, were studied by contrast variation and the measurement of hydrodynamic parameters as a function of scattering angle. The complex was also measured in the presence of 5’.adenylyl-fi,y-imidodiphosphate (ADPNP), a non-hydrolyzable ATP analog that is known to support limited supercoiling. R, = 67 A, from SANS and the hydrodynamic The values of the radius of gyration, radius, R,, = 64 A, from DLS predict a larger than expected volume for the enzyme. supporting the notion of channels or cavities within the molecule. In addition, several classes of models were rejected based on SANS data obtained in 2H20 at larger scattering angles. The best fit to both the SANS and DLS data is obtained for oblate, inhomogeneous particles approximately 175 A wide and 52 A thick. Such particles provide a large surface area for DNA interaction. Both R, and R,, values change very little upon addition of DNA, suggesting that’ DNA binds in a manner that does not significantly change the shape of the protein. No appreciable change in structure is found with the addition of ADPNP. However: the higherangle SANS data indicate a slight rearrangement of the enzyme in the presence of nucleotide.

1. Introduction

consists of two copies of each subunit (A,B,). The mechanism of DNA supercoiling by gyrase appears to involve the wrapping of a segment> of DNA (approx. 120 bpt) around the enzyme followed by

Bacterial DNA gyrase catalyzes the ATP-dependent supercoiling of DNA (Gellert, 1981; Cozzarelli, 1980). The enzyme consists of two subunits, A and B, of molecular weight 97,000 and 90,000, respectively (Swanberg & Wang, 1987; Yamagishi et al., 1986; Adachi, et al., 1987). Both are essential for the DNA supercoiling reaction and the active enzyme 011 0022%2836/90/01021

I-10

$03.00/O

t Abbreviations used: bp, base-pair(s); ADPNP. 5’.adenylyl-p,y-imidodiphosphate; SAM, small-angle neutron scattering; DLS, dynamic light-scatt,ering; ACF, autocorrelation function. 0

1990 Acadrmk

Press Limited

21%

s. tiruegrr et al.

the ATP-driven passage of anot,her segment, of DNA through a double-stranded break in the wrapped segment (Maxwell $ Gellert. 1986). ATP hydrolysis by the B subunit is required for catalytic supercoiling but the non-hydrolyzable XTP analog ADPSP (5’.adenylyl-b,y-imidodiphosphate) will support stoichiometrit supercoiling by gyrnse (Sugino rt al. 1 1978). This has been interpret,ed as suggesting that nucleotide binding is sufficient, to support a single cycle of supercoiling (a reduction of 2 in linking number) but that hydrolysis is required for enzyme turnover. KThilc it is clearly important to gain a,n understanding of t)he supercoiling mechanism. relativeI) little structural informat,ion is caurrently available for DNA gyrase and its complex with DXA. Several types of biochemica,l experiments including nuclease prot&ion studies have suggested that a segment ot I)?L’A is wrapped around t,he outside of a protein core with a positive superhelical sense (Liu K! \F’ang. I9i8a,h: Fisher pf nl.. 1981: tiirkegaard 8r Wang. 1981: Morrison & (‘ozzarelli. 1981: Ran et nl.. 19Si). Measurement of the translational fric%onal coefficient of gyrase c~omplesed with a DNA molec*ule of about 140 bp suggested that t’he pa,rticle was globular and the ratio of t)he frictional coefficient to that of the unhydrated equivalent) sphere gave a value of 1.9 (Liu & LVang, 19780). which indicated extensive bound and entrained solvent,. Elect’ron microscopic st,udies of gyraseel>NA complexes (Rau rt ~1.. 1987; I&her d nl., 19X4: Moore rf nl.. IUX3: Kirchhausen rt (11.. 1985: Maxwell et nl.. 1989) have generally shown roughly spherical particles of diameter 150 to 210 x (I .%= @I nm). At high resolut ion (Kirchhausen it 01.. 1985) these part,icl& have been interpreted as having a heart-shaped structure with t.he A subunits forming the upper lobes of the heart. ,\nalysis of gyrasr-l)N&4 complexes by transient elec+ricsdichroism support*ed the notion that a single turn of 1)NA of about 120 bp was wrapped around a roughly globular protein part,ic:le (Rau et crl.. 1987). Relaxation time measurements from t’he dichroism decaay c’urres of R complex between gyrase and 127 bp 1)NA fragments gave an equivalent sphere radius of X4 3 compared wit,h the calculated value of 66 4 for a fully hydrated protein and I)NA. It is apparent from the anomalously high frictiona~l coefficient, of the gyrase-DNA complex. its large sizr from elect’ron microscopy and it)s large rot.ational diffusion coefficient. that, a considerable discrepancy exists between the observed and predict’ed size of the complex. One suggestion t’o account for this discrepanq- is t)he existence of hydrated cavities in the enzyme structure (Rau rt (~1.. 1987). Indeed some mechanistic models of gyrase action invoke channels or cavities in t,hr enzyme (Wang et al., 19X1: Morrison et cd., 1981). Small-angle neutron scatt)ering (SAKS) is a tech niclue t,hat can be applied to the structural analysis of macromolecules (Jacrot. 1976) and is particularlp appropriat,e for analysis of t’wo-component systems where the individual components

exhibit differing rleutl,ori-s~att~~riJig lengths. Because hydrogen a,nd deut,erium (2H) have ver> different scat,tering lengths ( -0.3742 x 10 ’ * ml for H and 0.6671 x 10- l2 cam for “H), the scattc>ring length of the solvent casn be varied over H wide range by simply adjusting t’he ‘H,O/H,O rat.io. This method of c*ontrast’ variation has been successfully applied to the nu(*leosome (sore particle (Pardon rf nl.. 1975: Suau vt al.. 1977). among other syst,ems. 11 was shown that’ the radius of gyration of the IjS.4 was great,cr than that of the protein. supporting t-he tlotion t.hat t.he rtucleosorna,l I)Sr\ \vits wrapp~~d around the outside of the histone oc~tamer. The obvious similarities between t,he nu(~leosorn(~alrtl the gyrasepl)XA-4 (~omplt~x \\rould ~l~gg~cast that SASS could equally

well IF apptietl

to thr Itltt(xr systct~~.

Informa,tion c~omplernrntar~ to S.\NS (‘ill1 1w obtained by dynamics light -scaatlterinp ( l>l,S). This method determines the translational diffusion c~oeflicirnt of particles in solution by ntriasnrinp t tit, autocborrelat,ion func+tion i AC’F) (2 the sc*wtierrd light intensity. From the tliffGon c,ol+icietit t trts hydrodynamic radius. i?,,. is oht~ained. L!‘hile f tke radius of gyration is det~erminrcl by the mass tlist t+ K, nlolritors itI tirst bution of the whole partic+. al)I)roximatiori its overall (bxfensiori. In ;ttltlition. it is very sensitive 10 polydisprrsit,v and j)rovicicv it cahrc*kthat the solut,iorr does not, c,ont;cin aggregutecl rnakrial.

\yrry

(Aen.

usetill

stiap

information

vit11

be deducaed from simultaneous Ill~ilSlll’~I~lt’IIt~S of’ Rh und IZ,. Therefore. the c*ombined rcisults from I )l,S itnd

SAXkS \vere c~trlplovetl in t.tiis stutiy

t’o c~otlst ruc*t

a model for the strurtuml organization gyrasc and its c~omplex wit,11 DNA.

c)f’ I)r\‘A

2. Materials and Methods DNA\ gyrasr was prepared as drscritwd (Mizuuc.hi rt t/i.. 1984). The A and K suhunitjs we’re extensivrl>. dialyzed

against H-buffet (50 rnM-Tris. HC’I (PH 7.5). 55 rnll-Ii(‘l. 5 rn_M-tlithiothreitoI, 5”,, giycwol) or I)-buffer (like H-buffer but in ‘H,(J). Tht, A and I3 subunits wert’ in(sw bated together at ?f,‘(’ for 30 min to revonstitut~r the A,B, r)x’i\ gyrastl trtramer. :\ 172 bp DNA fragnit~tit derived from thr sea urchin 5 S rihosomal rR;VA gene ~.as as drwrihrd (>litXWt’ll prvparv(I f?orIl plasmid p.jS172-2.5 & Grllert. 1984). Gyrasr--DNA cwmplrxrs WPW assemblrd by incubation of the 172 hp fragment with 11X.4 gyrasr ill H-bulk or D-bufkr in t,he prese~~w of 2 lnwNlg(‘1, fog 1 h at 2:iC(‘. [‘ndrr these wndibions. the enzyme is acativr.

Wherr indicated. ;\l>PiYI’ U’BYitl~luded >tt a t*onc,entr;rtion of 0.35 111~.The proportions of the i\ suhu~rit. IS subunit and l).X:l, ust:d in thrscs t~~pwimwt:, ww &tw mined empirically by t,itrating increasing amounts 1)f’ I cwmponcntj against fixed amount,s of t.hc otjhrr 2 wd detrrmining. I)y I”)l?-~tc~1,?-laniiclr ql cslrac.1 vc~f)hot.(3is (Maxarll b (kllrrt. 19X4). the clxtrnt of c~~nrplrx ii)rniw tion. ThtL final c.otlc,rntratiolts used for both Stlpili: MKI DLS experimPnt,s were 0.78 mp A subunit,/ml (804 pi) Q90 mg R subunit/nil (10~0 pM). and 0-43 mg l)Nri/rttl (3.74 PM). (That, thr subunit concerlt,rat,ions rxcwd the DXA concentrations in molar terms probabl>, rrtircts t.he presmve of a small amount of inactivr protrin in the gyrase preparatjions.) The &ability of t,hesr cwmplexes was rvidenwd by thr cvnsistmcy of the scattjering data with

Scattering

Studies of DNA

time and was verified by electrophoresis following SANS experiments. Neutron-scattering measurements were made on the Dll instrument at the Institut Laue-Langevin. Grenoble. France (Ibel, 1976). The wavelength used was 1 = 103 A with a spread of An/n = soi,. Samples were measured at 6°C in quartz spectrophotometer cuvettes of Cl00 cm or 0.200 cm path-lengths. Sample-to-detector distances of I.5 m. 2.5 m or 535 m were used in order to cover thr of Bragg wave number, Q = appropriat’e range 4ni-’ sin 0, where 20 is the scattering angle. lntensity curves were corrected for uniformity of detector response and were placed on an absolute scale by dividing by the incoherent scattering from 1 mm of water measured under the same conditions ( i.e. the molecule has an average uniform scattering density, pv. The contrast between the molecule and the solvent is then defined as: (3)

Is = Pv-P,

and can be varied by changing the solvent. For example, ps = 1000/b 2H20 solvents and ps = 100% H,O solvents. Equation (2) can be written. (Guinier & Fournet, 1955):

the 2H20 content of 6.41 x 10” cme2 for - 5.62 x 1O9 cm 2 for for small Q values, as

Z(Q) = Z(0) exp (-iQ”Z?,“).

(1)

(4)

where Z(0) = (FL’)2 and the radius of gyration. of a particle of uniform density is defined as: 1 R2g = -1,’

R,,

r2d3r.

(5)

sV

Equation (4) is known as the Guinirr approximation. The radius of gyration and the scattering intensity at zero angle, Z(O), can be found directly from a plot of In (Z(Q)) versus Q2. Typically, the angular range of validity (Jacrot & Zaccai, 1981) for this approximation is for QR, < 1. No simple model such as a sphere, ellipsoid, etc. will be an exact’ model of the particle, as actual molecules are not solid objects with well-defined boundaries. However, in most casts, simple geometric shapes can be used as *a good first) approximation (Kratky, 1982). The entire measured scattering curve is compared to model scattering curves for various shapes in order to determine the best fit. (h) Dynamic light-scattering The int#ensity autocorrelation function, U’2’(~), for the light scattered from a dilute solution of small R xprc*trti X,K, str&ture. The ratio of molecular weights, M, (c~omplex)/M, (gyrase) is very calosr to 1.3. indcaat,ing that the gyrnseeJ>NA complex of’ M, z 489.000 is bring measured. However. its radius of gyration is essentially the same as t,hat, of the gyrast’ protein alone (see Table 1). A plot, of log (I(Q)/l(O)) ZWSU~S Q& for both gyrase and the gyrascb-~J)NA complex (Fig. 2) shows that t,he t,wo curves are not significantly different> in the Guinier region. This confirms that the overall shape of t’he particle does not change when the complex forms. The same conclusion is reached for the complex in thr present~e of ADPNP as it’s log (I(&)//(O)) !*~~‘.susQR, plot is identical with that of the complrs.

Scattering

Studies of DNA

Gyrase

215

02 j-

0.c zc 0 9 0 -0.1 < %F _J - I.(

i+y

90

I

I 0.05

I

I I 0.10 u* (lo-2

I

iI I

0.15

- / .e

I I.1

I

0

P,

I

I I.7

I

I 2.3

I 9

QRg

(a)

Figure 2. Log (I(Q)/Z(O)) uer~~~ QRgplot of SAKS data taken for gyrase alone (0) and the gyrase-DNA complex (0). The similarity of the 2 curves suggests that no major change in shape occurs when DNA is bound to the enzyme.

0

-Ie zc 0 4 5

;;__

Dynamic light-scattering measurements were made on the same three samples in H,O. Since the

0

-2

samples consisted of homogeneous solutions of gyrase or gyrase-DNA complexes, with a possible contamination by larger aggregates or dust, the data were analyzed using a two-component model (see eqn (7)). This model was sufficient for all samples measured and no third component had to be assumed.

0 0 0

-3

99

I

-4 0.00

I 0.05

I

I I 0.10 02 (10-Q A-2, (b)

(ai+

0 (,

I 0, I 5

I

Table 1

0.20

R, and 1(0)/c

values from Guinier H,O data

1(0)/r (cd/g)

C

-I

zz 0 2 -2 5

jits to SANS

Gyrase alone Oyrase-DNA complex Qyrase-DNA complex + ADPNP

R, (A)

Measured

67k2 69k2 69+2

0.19 * 0.02 0.32 & 042 0.32 & 092

(:alculated 0.20 O-30 0.30

Measured R, and 1(0)/c values are averages based on several different experiments. I(0) is normalized to the incoherent scattering from water and c represents the total wncentration of both protein and DNA.

-2

Figure 1. Guinier fit to the SAKS data taken in H,O for -4

(a) gyrase alone: (b) the gyrase-DKA complex: and (c) the complex in the presence of ADPNP. The corresponding radii of gyration, calculated from the slopes of the fitted (continuous) lines, are (a) R, = 67( + 1) 8; (b) R, = 69(fl) A; and (c) R, = 71(*1) 8.

216

S. Krueyer

et al

- 0.E

- 2.4

451 25

-3.2

55

85 28

145

115

Figure 3. Plot of the hydrodynamic radius, R,, LWSUS the scattering angle, 20, obtained from L)I,S data for DrJA gyrase in H,O (0). The broken line represents the average (R,, = 64( +3) 8) of all of the measured values.

Two-component exponent’ial fits to the DLS data showed. in all cases. that, the contribution from aggregated particles or dust was small compared t’o the sample signal. Since no significant variation of the measured diffusion coefficients with scattering angle could be detected (as shown in Fig. 3) the diffusion coefficients were averaged over all angles measured and over all equivalent samples. Table 2 shows the calculated hydrodynamic radii, Rh, for the three samples. No change in Rh is found, within experimental error, v.pon DNA binding or in the presence of ADPNP. This is consistent with the similar R, values observed with SANS for the same samples. However, evidence for increased aggregation was observed for the complex with ADPKP.

(b) SANS datain 2H,o The data in H,O do not extend beyond QRg z 3, because of the high background due to the incoherent scattering of H, and the low concentration of protein. For a protein in 2H20 solvents, however, the combination of higher contrast, lower background, larger sample volume and higher transmission yields a signal-to-noise ratio that is about 20-fold more favorable than for a solution of the same concentra-

Table 2 Hydrodynamic

radii from exponential to DLS data

Qyrase alone Qyrase-DNA complex Gyrase-DNA complex + ADPNP

fits

I

I 5.4

I

I

I

I

I

6

Figure 4. Log (I (C&/l(O)) versus QR, plot,s of SAIL’S data in ‘H,O for ggrase alone (!J), the gyrase+DPu’A complex (0) and the complex in the presence of AUPNP (A). The small difference in the latter scat,tering curve. which czannot be attributed to systematic error. represent’s a small conformational change of the protein moiety in the presence of nucleotide.

tion in H,O. The scattering curves in ‘H,O for the prot,ein and the complexes are shown in Figure 4. Due to systematic errors arising from small differences between thr ‘H,O content of the sample and buffer, it) was judged preferable to adjust, the measured solvent background by applying Pored’s law to I(Q) at, t,he highest Q values (Luzzat’i & 1980). The only assumption in this Tardieu, approach is that Z(Q) has been measured sufficientI\, far in Q that the law applies. The ‘H,O scattering curves can be divided into three sections. In the region where 0 I QR, < 3. the scattering curves of gyrase. the complex and the complex wit,h ADPNP follow t’he Cuinier equation (see eqn (4)) quite well. just as in the case of the H,O data. In the next, region, where 3 < QK, < 5.5. a flat part of each curve exists beyond the first maximum. While the curves for gyrase and its complex with DNA are identical, there is a small difference in that for the complex in the presence of ADPNP. This change. although small, cannot, be simulated by a systematic error in the background subtraction or in R since all three curves come back together again !h the third region where 5.5 < QR, d 9. In this final region, a peak corresponding to a second subsidiary maximum. related to the domain structure inside the shape, is clearly seen.

The experimentally determined values of Kg and R,, were used to obtam initial models of the gyrase enzyme based on simple geometric shapes. Some models were then eliminated using the higher-angle

Scattering Studies qf DNA Gyrase

217

4-

.0 -

4-

II. , , , 0.00 0.75

I I QR, I.50

I 2.25

, 3

Figure 5. Log (I(Q)) versus QR, plots of SANS data from gyrase in H,O (0) along with that calculated from (.....) a prolate model with semi-axes 6 : 2 : 1, as well as those calculated from ( - - -) a flat toroidal model. and t,oroidal models in which one-half of the torus is rotated .45” and (-.-) 90” out of the toroidal plane. (-)

SANS data. The final models are those that best fit all available SANS and DLS data. The Rg value obtained for gyrase immediately suggests a deviation from a globular conformation. The radius of gyration of a compact globular protein of molecular weight 400,000 and partial specific volume, V = 6735 cm3/g, is calculated to be 43 A. For example, the Fl moiety of bacterial ATPase (an a3p3 structure of 380,000 molecular weight) has R, = 46 A (Satre & Zaccai, 1979) and is therefore nearly globular. The R, value for gyrase is quite large in comparison, so gyrase must deviate considerably from this type of structure. A similar conclusion can be drawn from the DLS data. A solid sphere with R,, = 64 A and the partial specific volume mentioned above would have a molecular weight of 883,000. Since the molecular weight of gyrase is less than half of this value, it must be either non-spherical or non-homogeneous. Both prolate and oblate models consistent with the experimental R, and R,, values as well as the known gyrase volume were constructed using a finite number of spherical elements (Glatter, 1980; modification of 0. Glatter’s original program, MULTIBODY, available from R. May at the Instit,ute Laue-Langevin). As illustrated in Figure 5, a doughnut or toroidal shape, similar to that of the Rho protein (Oda & Takanami, 1972; Yager, T., Tardieu, A & Calmettes, P., personal communication), with an inner radius of 46 A and an outer radius of 84 A, fits the SANS data in the Guinier region

quite

well

out, to

QR, = 2.5. On the other

hand, a prolate model, with semi-axes a/b/c: equal to 6 : 2 : 1, deviates from the data for QR, > 1. Hydrodynamic radii of the same model structures were

Figure 6. Log (I(Q)) versus QR, plot8 of SAN’S data taken from gyrase in ‘H,O ([7), along with the same curve calculated from a flat toroidal model (broken line). Note that QR, extends to 10.0 for the ‘H,O data rather than to only 30 as for the H,O data (Fig. 5). The data fall below the curve for the toroidal model. indicating that the gyrase enzyme is not as “hollow” as the torus.

computed using a program based on a modification of the Kirkwood algorithm (de Haen et al., 1983). Those obtained for the oblate models were in the best agreement

with

the experimentally

measured

values. Furthermore, neither the SANS nor the I)LS data restrict the central axis of the torus to a single plane. For example, if the torus was sliced in half by passing a plane through its diameter, one of the two sections could be rotated out of the toroidal plane. Even if the angle of rotation was 90”, R,, would only change by approximately 1 A, which is within experimental error (see Table 2). While R, would decrease by approximately 100/b for a 90” rotation angle, the inner and outer radii of the model torus could be increased slightly in order to fit the measured value of Rg. Such calculated scattering curves are also shown along with the data in Figure 5. They do not differ appreciably, in the same & range in which data were obtained, for angles of rotation up to 90”. While the toroidal models fit t)he SANS data out) to QR, = 2.5, they do not adequately describe the data taken in 2H,0 beyond QR, = 3. as is clearly illustrated in Figure 6, where log (Z(Q)/](O)) ‘UerRUS QR, for both the data and the model are plotted. The data points fall below the model curve, indicating that’ the gyrase enzyme is not as “hollow” as the toroidal model scatterers. Tt, is more likely that gyrase contains several smaller cavities throughout its interior, rather than one large cavity. The family of models that best fit all available SAN’S and DLS data consists of four subunits arranged in an oblate structure approximately 175 A

218

S. Krueger et’ al.

w

+-+

13.58

(a)

Figure 8. Log (Z(Q)) versus QrZ, plot of SASS dat,a in ‘H,O from the gyrase--DNA complex in the presence of ADPNP (A). along with the same curve (broken line) calculated for the 2 possible oblate models (Fig. 7). which give almost identical scattering curves.

13.5 a

(b)

Figure 7. Top-view drawings of 2 simple models for DNA gyrase that feature 4 subunits arranged in an ablate structure approximately 175 a wide and 52 a thick, with cavities. grooves or indentations of the order of 15 8. Possible cavities or indentations are illustrated in (a), while fewer. but deeper, grooves are shown in (b). Both models agree equally well with all available SAN8 and DLS data. The broken circles represent the radius of gvration (R, E 67 A) of the models. Subunit volumes are slightly larger than those calculated for fully compact particles.

wide and 52 J! t,hick with cavities; grooves or indentations of the order of 15 L!. Two simple models of this type, which give almost identical scattering curves, are illustrated in Figure 7. Possible cavities of indentat’ions are shown in Figure 7(a). while fewer. but deeper. grooves are featured in

Figure 7(b). The broken circles correspond to the radius of gyration of the model (Rg 2 67 .A). Sinw the A and B subunits

t,his resolution.

cannot

four identical

be distinguished

subunits

at

have been

drawn for illustrativtl purposes. The calculated volume of each subunit in the model is slightly larger than calculated for a fully compact particle of the same molecular weight. indicating domain strucBture within the subunits. Indeed, the existence of discrete domains within the A subunit is supported by recent experimental evidence (Reece & Maxwell. 1989). A disk of diameter 185 L%and height 52 L! can be circumscribed around each model. Just as in the case of the torus. one half of the disk can be rotated out of the plane slightly without affecting t,hr scat.tering curve appreciably. The log (I(Q)/Z(O)) VW-susOh!* KANS curve for thca oblate models is plotted in Figure 8. along with t,he data obtained in 2H,0 from the DNA gyrase complex in the presence of ADPNP. The data follow the model curves much more closely than in t’hfb toroidal case shown in Figure 6. Thus, the family of oblate. inhomogeneous models illust,rated in Figure 7 more closely resemble the gyrase enzyme and its complexes. The computed hvdrodgnamics radius for these models is 69 B, implyrng that the? are also consistent’ with the DLS data. It can be concluded from Figures 4, 6 and 7 that the small change in the 2H20 scattering curve seen between 3 < QR, < 5.5 for this sample is due to a small conformatlonal change in the protein. l)?;A or bot’h, which increases the inhomogeneity of the particle. Since the contrast of DNA is much smaller than that of protein in 2H20 solvent, the contribution of the scattering from DR;A to t,hr t,ot’al scattering intensity is small. Therefore, it is more likeI> that the small change in the scattering curve upon addition of ADPKP is due to a rearra,ngement of t,hc protein moiety rather than DNA.

Scattering Studies of DNA Gyrase

5. Conclusions Measurements of Escherichia coli DNA gyrase in H,O solution by SANS and DLS yielded values for R, and R,, of 67 A and 64 8, respectively. Furtherof molecular more, an absolute determination weight by SANS confirmed the A$, tetrameric structure of the molecule. These results strongly support the existence of cavities or grooves in the gyrase enzyme, which would provide a large surface area for interaction with DNA. Models that best fit all available SANS and DLS data consist of four subunits arranged in an oblate structure approximately 175 A wide and 52 A thick (see Fig. 7). High-resolution electron micrographs of DNA gyrase (Kirchhausen et al., 1985) have been interpreted as suggesting a “heart-shaped” structure for the A,EI, tetramer. The measured diameter of the particle was about 210 A, a value consistent with other electron microscopy studies (Rau et al., 1987; Lother et al., 1984; Moore et al., 1983; Maxwell et al., 1989). Given the potential for disruption of samples during preparation for electron microscopy, this value is in reasonable agreement with the 175 A value derived from the SANS and DLS data. In addition, if a portion of the oblate particle is rotated out of the plane, the image can be shown to be similar to the “heart-shaped” model described above when viewed from certain angles. Transient electric dichroism measurements of complexes between DNA gyrase and short linear DNA fragments (Rau et al., 1987) suggested that the protein complex is roughly globular and that the DNA is wrapped around the protein core in a single turn. The equivalent sphere radius of a complex between gyrase and a 127 bp DNA fragment, as determined by relaxation time measurements, was found to be 84 8, in good agreement with the radius calculated for the oblate model in this paper. The measured particle dimensions also exceeded the expected dimensions in the electric dichroism experiments, suggesting the existence of cavities or grooves in the complex. In addition, the measurements indicated a significant conformational change in the structure of the complex in the presence of ADPNP, involving the binding of the free DNA tails to the protein core (Rau et al., 1987). Such a change results in the reduction of the hydrodynamic radius of the complex with a 172 bp DNA fragment, seemingly inconsistent with the SANS and DLS results (Tables 1 and 2). Since no appreciable change occurs in either R, or R,, in the presence of DNA, both the protein and DNA must have approximately the same radius of gyration. Additionally, the shape of the SANS scattering curve in H,O does not change upon DNA binding, indicating that the centers of mass of both the enzyme and the DNA must be located close to each other. Therefore, a ring of DNA must bind in the approximate location of the broken circle of fB E 67 A shown in Figure 7. The DNA configuration suggested by the electric dichroism experiments in the absence of ADPNP would be consistent with these results.

219

The small change observed in the large-angle SANS scattering curve upon addition of ADPNP to the DNA-gyrase complex in ‘Hz0 solvent is indicative of a conformational rearrangement of the protein moiety which increases the inhomogeneity of the particle, with the DNA possibly settling into, and thus widening, a groove in-between subunits. However, no change in the overall particle shape is detected by either SANS or DLS. The rotational diffusion coefficient, measured in the electric dichroism experiments, is extremely sensitive to the long dimension of the particle, 1, as D,,, CC13, whereas the translational diffusion coefficient, measured in the DLS experiments is related to the long dimension as Dt,,, oc 1. Therefore, a small change in the long dimension of the particle, while easily detected by electric dichroism measurements, possibly would not be detected with SANS or DLS. Current models of gyrase action involve the passage of a segment of DNA through a doublestranded DNA break, held open by interaction with the protein (Maxwell & Gellert, 1986). Implicit in these models is the requirement for the DNA to be passed through at least part of the protein. The SANS and DLS data strongly support the existence of cavities or grooves in the gyrase tetramer that would facilitate this process. Collaborative aspects of this project were supported in part by grant no. 86/0479 from the ?u’orth Atlantic Treaty Organization. Partial sponsorship was also provided by the National Cancer Institute under contract no. NOl-CO74101 with Bionetics Research Inc. Work in the laboratory of A.M. was supported by the SERC (U.K.). The views expressed in this article do not necessarily represent the position of the National Institute of Standards and Technology or the National Institutes of Health. Certain commercial equipment, instruments or materials are identified in order to adequately specify the experimental procedure. Such identification does not imply recommendation or endorsement by the above organizations, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.

References Adachi, T., Mizuuchi, M., Robinson. E. A., Appella, E., O’Dea, M. H., Gellert, M. & Mizuuchi. K. (1987). Nucl. Acids Res. 15, 771-784. Berne, B. J. & Pecora, R. (1976). Dynamic Light Scattering, J. Wiley and Sons, New York. Cozzarelli, N. R. (1980). Science, 207, 953-960.

de Haen, C., Easterly, R. A. & Teller. D. C. (1983). Biopolymers,

22, 1133-1143.

Fisher, L. M., Mizuuchi, K., O’Dea, M. H.. Ohmori, H. & Gellert, M. (1981). Proc. Nut. Acad. 8ci., U.S.A. 78, 4165-4169. Gellert, M. (1981). Annu. Rev. Biochem. 50, 879-910. Glatter, 0. (1980). Acta Phys. Austriaca, 52, 243-256. Guinier, A. & Fournet, G. (1955). Small Angle Scattering of X-Rays, J. Wiley and Sons, New York. Ibel, K. (1976). J. Appl. Crystallogr. 9, 630-643. Jacrot, B. (1976). Rep. Prog. Phys. 39, 911-953. Jacrot, B. & Zaccai, G. (1981). Riopolymers, 20, 2413-2426.

220

8. Krueger

Kern, D., Zaccai. G. & Giege, R. (1980). Biochemistry, 19. 3158-3164. Kirchhausen, T., Wang, J. C. bi Harrison. S. (1. (1985). Cell, 41, 933-943. Kirkegaard. K. & Wang, J. (1. (1981). Cell, 23. 721-729. Kratky. 0. (1982). In Small Angb S-ray &‘cat&ring (Glatter, 0. & Kratky. O., eds), pp. 3-15. Academic Press, Eew York. Liu. L. F. & Wang, J. C. (1978a). I’roc. ~V’at. ilcarl. Sci.. 1’S.A. 75, 2098-2102. Liu. L. F. & Wang. J. C. (19783). C’ell. 15. 979-984. Lother. H., Lurz, R. & Orr. E. (1984). M~cl. ilcids RP.s. 12. 904-914. Luzzati. V. & Tardieu. A. (1980). Annu. Rer. Riophys. Bioeng. 9. l-29. Maxwell. A. & Gellert. $1. (1984). ./. Hiol. C’h~m. 259. 14472-14480. Maxwell. A. 8: Gellert. 11. (1986). Advan. Prokin Chrm. 38, 69-107. Maxwell. A.. Gellert? $1. & >lcTurk. I’. (19S9). In Hiyhlights of Modern Bioch,wni.rtry, (Kot)ky, X.. Skoda. .J.. Paces. V. & Kostka. 17.. eds). pp. 97-114. VSI’ International Science Publishers. Zeist). Eizuuchi, K.. Mizuuchi. M., O’Dea. M. H. &. t&Ilert.. M. (1984). J. Biol. Chem. 259. 919999201. Moore. (1. I,.. Klevan. L.. Wang. ,I. (‘. 8 (;rilIith. J. I). (1983). J. Biol. Clhem. 258, 4612-4617. Morrison, A. Cy:Cozzarelli. h-. R. (1981). I’roc. iVat. ilcnd. Ah’.. 1’S.A. 78, 1416-1460. Morrison. A.. Brown, I’. 0.. Kreuzer. I(. ?;.. Otter. IZ.. Cerrard. 8. P. & (lozzarelli. X. R’. (1981). In Mechnnistic Stu,dies of DSA Replicution and (&n&c Reeonr-

et al. (Alberts. H.M. & Fox. (‘. IT.. feds). pp. 785-806, Academic Press. Kew York. Oda, T. &. Takanami, M. (197”). ./. Mol. Rio/. 71. 799-802. Pardon, ,J. F.. Worcester. I). I,.. Wooley. ,J. C‘.. ‘l’atvheli. K.. Van Holde. K. & Richards, l. %I. (1975). .2’ur/. Acids R/w. 11. 2 163 -2 176. Reece. R. *J. & Maxwell, A. (1989). ,1. Hiol. (‘hum. In the press. Rau, I). C’., Gellert. M.. Thoma. F. $ ~laxwc~ll. A. (19X7). J. Mol. l?iol. 193, 555-569. .? 102. Sat,re, M. & Zacoai. (*. ( 1979). FE&S IXm, 244-248. Stuhrmann, H. Il. (1974). ,/. --Ipp/. ~‘r//stu/lr,(/t~. 7. 173-178. Stuhrmann. H. 1~. Hr Millrr. ii. (1978). .J. .1//p/. ~‘r,q.~/rr/logr. 11. 365-345. Suau. I’.. Kneale. (:. (i.. Rraddock. G. W.. RaIdwin .I. I’ & Bradbury. E. M. (1977). S’clcl. .4cirl,s Kes. 4. 3769-3786. Sugino. A.. Higgins. X. I’.. Brown. P. 0.. I’rrbles, (‘. I,. & Cozzarelli. N. R. (1978). Pror. Nat. Acad. Sri.. f -.S.A, 75. 4838-4842. Swanberg. 8. I,. Ji Wang. *J. (‘. (1987). .1. .Ilol. /Co!. 197. 72S736. Wang, *J. C’., Gumport. R. I.. .Javaherian. K.. Kirkegaard. K., Klevan, L.. Kotewitz. M. I,. d Tse. T.-C. (1981). In Mechanistic &udies of TINA Rcylicat’ion and Genetic Recombination (Alberts, 1%.M. 8r Fox. (3. F.. eds), pp. 769-784, Academic Press. New York. Yamagishi, J-I.. Yoshida. H.. Yama,yashi. .\I. & Sakamura. S. (1986). Mol. Gpn. (knot. 204, 367 353. bination