J. Mol. Biol. (1991) 219, 747-755

Spatial Arrangement of a-Factor and Core Enzyme of Escherichia coli RNA Polymerase A Neutron

Solution

Scattering Study

Hermann Lederer’, Kell Mortensen’, Roland P. May3, Gisela Baer’ Henry L. Crespi4 Dominik Dersch’ and Hermann Heumann’j‘Max-Planck-Institut fiir Biochemie, D-8033 Martinsried, Germany 2Ris~ National institute, DK-4000 Roskilde, Denmark 31nstitut Laue- Lange&, F-38042 Grenoble Cedex, France 4Argonne National Laboratory, IL 60439, U.S.A. (Received 1 December 1990; accepted 25 February

1991)

By means of neutron solution scattering we determined the position and orientation of core enzyme and a-factor within the Escherichia coli RNA polymerase holoenzyme with the aim of improving existing models. The individual components, core enzyme (E) and a-factor (G), were highlighted by deuterium labeling and their center-to-center distances determined in the monomeric and the dimeric holoenzyme. The following distance parameters were obtained: dEIPa, = 8.6( f 1) nm, &m-E2 = 11.5(f 1) nm, d,,_,, = 12.0(+@7) nm, dE,-s 1 = 9( f 3) nm. Using a triangulation procedure the position of the o-factors, o1 and 02, were determined with respect to the mass center of the core enzyme molecules, E, and E,, assuming a symmetrical arrangement of the holoenzyme molecules in the dimer (C2 symmetry). In addition, the orientation of the a-factor with respect to core enzyme was estimated by means of model calculations. The obtained model of holoenzyme depicts the Q-factor as buried in a groove of core enzyme, probably between the large subunits /?’ and fl. Keywords: RNA polymerase; neutron scattering; sigma factor; deuteration

1. Introduction Initiation of transcription in Eubacteria requires a specificity factor, the a-factor (for a review, see Helmann & Chamberlin, 1988; von Hippel et al., 1982). This a-factor binds reversibly to RNA polymerase core enzyme (subunit composition: pbcr,) and forms the holoenzyme. Only holoenzyme is capable of a specific initiation at the promotor. The a-factor dissociates from the transcribing complex after a short RNA chain consisting of about nine to ten baseshas been synthesized (Hansen & McClure, 1980). The free a-factor can rebind to core enzyme and reform a transcription-competent holoenzyme. This series of reactions during initiation has been termed the a-cycle (Burgess & Travers, 1971). There are several different types of o-factors enabling RNA polymerase to recognize specific promoters (Helmann & Chamberlin, 1988). The majority of promoters in Escherichia coli are recognized by the holoenzyme (molecular weight: 449,000) containing t Author to whom all correspondenceshould be addressed 747 0022%2836/91/12074iJJ9

$03.00/O

the o”, which has a molecular weight of 70,000 (Burton et al., 1981). To better understand the function of the o-factor, information about the structural relationship of core enzyme and the o-factor is required. Neutron small-angle scattering was used to study the internal structure of the RNA polymerase holoenzyme (Stijckel et al., 1979, 1980a,b). X-ray small-angle scattering was used to determine the overall shape of the isolated o-factor, the core enzyme and the holoenzyme (Meisenberger et al., 1980a,b, 1981). From these studies a model of the spatial arrangement of the o-factor and core enzyme was proposed. The X-ray approach is unsatisfactory due to its inability to determine the structure of the components o-factor and core enzyme in situ as part’ of the complex. Neutron solution scattering can overcome this problem. Single components of the protein complex can be visualized by deuteration, whereby the scattering contribution of the rest of the protein is suppressed by contrast variation, using a mixture of H,O and ‘H,O as solution buffer (Engelman & Moore, 1972: Koch & Stuhrmann. 1979; for a review. see Timmins & Zaccai, 1988). In 41 y. 2H20 the contrast of proto0

1991

AAoadnmic

Press

Limited

748

H. Lederer

nated protein is matched (Lederer et al., 1986). This technique was used to determine the center-tocenter distances between subunits of RNA polymerase (Stiickel et al., 1979, 1980aJ). We have continued this approach with the aim of determining the spatial arrangement of the a-factor and core enzyme. We show that in a two-component system, such as core enzyme and o-factor, a low resolution model can be derived when the molecule under study dimerizes as a symmetric structure.

2. Materials and Methods (a) Sample

preparation

Growth of E’. coli B and preparation of DPu’A-dependent RXA polymerase (EC 2.7.7.6) core enzyme was performed according to Zillig et al. (1970) with slight modifications. The RNA polymerase subunit CJwas prepared from the M5219/pMRG8 according t,o overproducing strain Gribskov & Burgess (1983). Fully deuterated subunit c was obtained by growing the cells in *H,O medium using deuterated algal hydrolysate as carbon source (Crespi rt al.. 1960). RNA polymerase holoenzyme was reconstituted from stoichiometric amounts of core enzyme and a-factor as described by Heumann et al. (1986). The final purification step of holoenzyme was chromatography on a mono S-column (Pharmacia) in 50% (v/v) glycerol. The protein concentrations were determined according to Heil & Zillig (1970). The activity of the enzyme was determined as described (Heumann et al., 1986). The dialysis buffer that matches the scattering length density of protonated RPU’A polymerase core enzyme contained 590/b (v/v) H,O, 41%) (v/v) 2H,0. 10 mM-Tris.HCl (pH 7.9), 0.05 M-XaCl (low salt) 01 W5 M-NaCl (high salt) and 10 mM-P-mercaptoethanol. (b) Data

collection

and

4ll

& = -- sin($/2). A The distance distribution functions p(r) were calculated with a program from Glatter (1977). Radii of gyration were determined from area and second moment of the p(r) funct,ions> as well as from Guinier plots (ln(1(0)) zlersus Q2). Cross-sectional radii of gyration were derived from cross-sectional Guinier plots (ln(&*I(Q)) oersus &‘). Interdistance distribution functions were obtained b) subtracting from the pair distance distribution function of 2 components the distance distribution functions of the individual components. Model calculations were performed with the Fortran program MULTIBODY of Glatter (1980). determination labeled

between

radius of gyration K of the complex parallel axes theorem of mechanics:

according

to thr

R2 = z,Rf+~~R;+q~~d:,~.

where zi denotes the weighting with the molar mass Mi: 2, =

Mi/(iWl

(1)

factors of the subunit + MII,).

(d) The distance neighboring

between core enzyme (E,) a-factor (03 in the holoenzywhe

i (2)

and thr dimpr

The determination of the distance between E, and (r2 requires knowledge of the corresponding scattering funtion, which was obtained as a linear combination of the scattering functions of the following structures. the dimrr (E,a, E,o,), the core-core substructure (E, EC,), the ~-0 substructure (a, a,). the core--B substructure (E, n,) and the subskuctures of (‘ore enzyme (E) and the a-factor (cr) according to:

(E,u2)2 = ; X~(~,a,li:2u2)2-2(E,oI)2 -(li:,l(:2)2-(u.1u2)2+4(~:1)Lf4(u,)2~.

(3)

The distance (d,, i) between the mass cherlter (S,) of’ thr dimer and the csomponmt i was determined hy the general parallel axis theorem of mec.hanics: t-11

with IZ,. the radius of gyration of the dimrr. of gyration of the component i. and:

/ti the racliuh

processing

The small-angle neutron scattering measurements were done at the Dll small-angle spectrometer at the Institut Laue-Langevin (Grenoble, France) and the Riscl SARIS facility (Roskilde, Denmark). The samples were equilibrated to 37°C during beam exposition. Data were processed and corrected as described by Heumann et al. (1988). From the combined scattering rurves Z(Q) (with the momentum transfer):

(c) Distance

et al.

thr mass

centers

of two

components

The center-t,o-center distance (d,,2) of 2 subunits with radii of gyration I?, and R, was determined from the

3. Results

The analysis of the spatial arrangement of WIY~ enzyme and a-factor first required knowledge of the c*onformational differences of core enzyme and c-factor. in t,he free and the hound state. To attail this information. t’he scattering curves of the cornponents, the a-fact’or and the core enzyme. isolated and in complex. were caomparetl. Holoenzyme was reconstituted from core en+-mc and o-factor with one of the components deuterated (see Materials and Methods). The scattering contribution of the protonatrd part was suppressed, using a solution buffer containing 41 o. *H ,O (Lederer pf al., 1986). Figure l(a) shows the scattering curvy’s ot’ the o-fac%or isolat,ed and in the complex with ww enzyme, Figure 1 (b) shows thr corresponding Fourier transform, the dist,ance distribution fumtion, p(r). This function represents the frequency of distances between any two volume elements of the scat,tering molecule. It is often pasier for the judgcament of the cluality of a model to cLompare the distance distribution funckion rather Ihan the scatt,ering curve (Glatter. 1980). Figure 1 (h) includes the

Quaternury

Structure

of E. coli RNA

Polymerase

749

0 I.0 o-8

-I $j

-

-IE

0.6

-2

-3

I IO

-0.2i

2

I 20

0

r (nm)

0 (nm?

Figure 2. The distance distribution function of the core enzyme, free (broken line), and in situ (continuous line). The inset shows the top view of the model for core enzyme proposed by Meisenberger et al. (1980b).

(a) I.2 I.0 0.0 L; 5

0.6 04

0.2 C r (nm) (b)

Figure 1. (a) The scattering curve of the o-factor, isolated (n), and in situ (I) (bound to core enzyme). (b) The distance distribution function of the free a-factor (broken line) and the model function (continuous line). The inset shows the top view of the corresponding model proposed by Meisenberger et al. (198Oa).

nated o-factor. The measurements were performed again under contrast matching conditions, i.e. in 41 y. 2H20. Figure 2 shows the distance distribution function of core enzyme, with and without a-factor. The maximum of p(r), of the core enzyme bound to the a-factor, is slightly shifted towards shorter distances, indicating that binding of the a-factor leads to a more compact’ molecule. LA model of the overall shape of core enzyme has been proposed on the basis of X-ray and neutron small-angle scattering experiments showing that the core enzyme is a flat triangular-shaped molecule (inset of Fig. 2) (Stiickel et al., 1980a; Meisenberger et a.l., 1980b).

(c) The spatial p(r) function of the model of isolated a-factor as well as the proposed model (shown as inset, according to Meisenberger et al., 1980a). The scattering curves of the a-factor, free and in complex with core enzyme, agree within the error margin, indicating no gross conformational change during complex formation. (b) Thr

structural parameters of core enzyme frrr and bound to the a-factor

To visualize stituted from

core enzyme, holoenzyme deuterated core enz.vme

was reconand proto-

arrangement

of core enzyme and

o-factor in the monomeric holoenzyme Figure 3(a) shows the distance distribution function, p(r), of holoenzyme. A comparison with the p(r) of core enzyme (Fig. 2) shows that holoenzyme and core enzyme have the same maximum dimenmeans that the a-factor does not sion, which increase the overall size of the holoenzyme. Neither the radii of gyration (see Table 1) nor the crosssectional radii of gyration (R, = 3.5( +@3) nm; see Fig. 3(b)) of core enzyme and holoenzyme (Ea) differ significantly, implying that the a-factor is at least partly buried in the core enzyme. To render this qualitative statement into a more

Table 1 Molecular

weights

and radii components

of gyration of the monomeric of E. coli RNA polymerase

o-Factor Molecular

weight

R o. munomer, free (nm) R G. monomer, in situ (nm) RG. dimcr (nm) The ((iribskov gyration

70,wJ 4.2( f 02) 40(+@2) 7.3(f02)

molecular weights obtained by biochemical 8: Burgess, 1983; Ovchinnikov etal., were obtained as described in Materials

and the dimeric

Holoenzyme

C0re

379,000 5%( * 02) 6.2(+@2) 8.2(*0.2) methods and neutron 1977, 1981, 1982; Lederer and Methods.

449,000 6.4(*0.2) 8.0( + ocq scattering etal..

agree within 5?,, 1986). The radii of

750

H. Lederer et al I.2 I-0 0.8 2

0.6 0.4 0.2 20

30

rinmj

(a 1

0

-0.6

*0 ‘;:

-0.8

0” J

- I.0

I -1.2

(nml

(b)

Figure 4. (a) The distance distribution t’unctiou of the dimeric holoenzymr. (b) The interdistancr distribution function of the 2 holoenzyme molecules in thr dirnrr.

cl *(“m-2) (b)

=

rE,a,Em

between 0.06,

the

represents holoenzyme

the venter-to-center molecules.

disbancr

I

0.05 0.04 r 19

0.03 0.02 o-0

I

o-00 -0.01

0

IO

20

30

.r inm) (cl

Figure 3. (a) The distance distribution function of the holoenzyme. (b) The cross-sectional Guinier plot of the core enzyme (0) and the holoenzyme (m). (c) The interdistance distribution function of the core enzyme and the g-factor (in the monomeric holoenzyme). The maximum of the function at T = vElal represents the center-t,ocenter distance between the core enzyme and the a-factor in the holoenzyme (see Table 3).

yuantitative one, information on the position and orientation of the o-factor with respect to core enzyme is required. The most frequent distance, d Ill, E-U> between core enzyme and a-factor can be obtained directly from the corresponding interdistance distribution function (see Fig. 3(c)). This fumtion represents all distances connecting volume elements in core enzyme with volume elements in the a-factor and should closely resemble the centerto-center distance between the components. The distance, d,, E d = 8.0( + 2) nm. obtained from the

interdistance distribution function. is ~losc~ to the value; d,-, = 8%( + 1) nm, obtained by the parallel axis theorem of mechanics (see Materials and Methods, and Tables 2 and 3). The determination of the position of the a-factor by model fitting was unsatisfactory. even if the information about the center-to-center distancr, between the components was taken into account. The sensitivity of the procedure is not sufficient to ascertain both the position and the orientation of the t’wo components. Addit,ional information was required to determine the position of the a-factor.

(d) 7’he spatial

arrangement th,e dimeric

of thr holovnzyme

cwmpunent~

is

As shown for the dimeric tetracycline repressor. distance parameters in symmetrically arranged molecules have proven to be very useful in confining the spatial arrangement of the components (Lederer rt al., 1989). Therefore. we determined the center,-tocenter distances of the components in the dimeric. holoenzyme with the aim of ascert.aininp the position of the a-factors by triangulation. RNA polymerase holoenzyme dimerizrs at low ionic strength (Berg Br Chamberlin, 1970: Heumann et al.. 1982a; Shaner et al., 1982). X-ra; small-angle scattering studies have shown that the overall structture of the dimeric holoenzyme tan be approximated by a symmetrical side-by-side arrangement

Quaternary

of E. coli RNA

Structure

751

Polymerase

Table 2 Center-to-center holoenzyme

u-Factor u-Factor u-Factor Core( E2) Holo(E,u,)

in the monomeric and dimeric axes theorem of mechanics

distances of the components determined by the parallel (a,)

Core enzyme(E,)

Holoenzyme(E,a,)

8.6( k 1 .O) nm 9(+3) nm 11.5(+@9) nm

120( * 0.7) nm 9(*3) nm

9O(_fl.I)

of the triangular-shaped core enzyme with C2 symmetry (Heumann et al., 1982b; see inset of Fig. 4(a)). (i) The center-to-center distances of the components in the dimer The distances between the holoenzyme molecules, the core enzyme molecules and the a-factors in the dimeric holoenzyme were determined by means of the parallel axes theorem of mechanics, as described in Materials and Methods. This approach requires knowledge of the radii of gyration of the components in situ, which were derived from the corresponding distance distribution functions. Figure 4(a) shows the distance distribution function of the dimeric holoenzyme, Figure 5(a) the distance distribution function of the core(E,)-core(E,) substructure and Figure 6(a) the distance distribution functions of the c~-(T* substructure. The latter two distance distribution functions were obtained from the scattering curves of reconstituted holoenzyme, in which either the core enzyme molecules or the a-factors were deuterated. Alternatively, the distances were obtained directly from the interdistance distribution functions, which are shown in Figures4(b), 5(b), and 6(b). The values obtained with both methods agree within the error margin (compare Tables 2 and 3). In addition we have calculated the distance d E,-(I2, the center-to-center distance between core enzyme (E,) and the neighboring o-factor (cJ*)in the dimeric holoenzyme. Assuming a symmetrical arrangement of the holoenzyme in the dimer (see Discussion) the distances dElpa2 and dE1-a, are equal. These distances can be determined if the interdistance distribution function or the radius of gyration of the substructure (E,o,) (or (E,o,)) is known. The scattering curve of this substructure is, however, not accessible experimentally, because specific deuteration of E, and g2 alone is impossible.

nm

We can obtain the scattering curve and the corresponding distance distribution function of the substructure (E,a,) or (E,a,) by a linear combination of the known scattering functions of the holoenzyme, core enzyme and a-factor in the dimeric and monomeric form, as described in Materials and Methods (eqn (3)). Figure 6(b) shows the obtained interdistance distribution function. There is only one prominent peak, in agreement with our assumption that dE,-g2 and dE2-,,, are equal, due to a symmetrical arrangement of the dimer. From the position of the peak a distance of d,, --n2 = 8( f 4) nm was estimated. The radius of gyration (see Materials and Methods), RE,O1 = 6.2 nm, was determined from the corresponding distance distribution function (Fig. 6(a)). This value was used to calculate the distance (dEteal = 9( 53) nm) by means of the parallel axis theorem of mechanics. The values for d E,-(r2 obtained by the two methods agree within experimental error. However, the large error margins are unsatisfactory. Therefore, a less errorprone approach was used as an alternative. which is described in section (c), below. (ii) The a-factor center of mass The position of the o-factors in the dimeric holoenzyme can be determined by triangulation using the following six distances: the cent,er-to-center distance between (1) the core molecules E, and E,, d E1--E2;(2) the core enzyme E, and t,he a-factor gl: dEleal; (3) the core enzyme E, and t,he o-factor cr2, d Elma2;(4) the core molecule E, and the a-factor cr2 of the neighboring core enzyme (E2), dElmol; (5) the core molecule E, and the o-factor O, of the neighboring core enzyme (E,), dEzea,; and (6) the r~factors o1 and cZ, d,,_,,. In this context the assumption was made that the distance between core enzyme and a-factor d,_, in the monomeric holoenzyme is the same as in the dimeric holoenzyme: d,-, = dEleaI = dE2-g2 (Fig. 7).

Table 3 The distances of the components in the monomeric and dimeric determined by the interdistance distribution function a-Factor u-Factor u-Factor Core( E,) Holo(E,u,)

.~. 12(*3)nm 8(+4) nm

Core enzyme(E,)

holoenzyme

Holoenzyme(E,

8(&2)nm 8(*4) nm 12( f 2) nm lO( + 3) nm

ul)

752

H. Lederer

et al.

0.8 ,

r (nml

r(nm)

(a)

(0 I

04

r