Direct interaction between two Escherichia coli transcription antitermination factors, NusB and ribosomal protein S10

J. Mot. Biol. (1992) 223, 55-66

Direct Interaction Between Two Escherichia coli Transcription Antitermination Factors, NusB and Ribosomal Protein SlO Stephen W. Mason, Joyce Li and Jack GreenblattjDepartment of Molecular and Medical Genetics and Banting and Best Department of Medical Research [Jniversity of Toronto, Toronto, Canada (Received 19 April

1991; accepted 9 September 1991)

The Escheriehia coli proteins NusB and ribosomal protein SlO are important for transcription antitermination by the bacteriophage lambda N protein. We have used sucrose gradient co-sedimentation and affinity chromatography with immobilized ribosomal protein SlO, a glutathione S-transferase-SlO fusion protein, and NusB to show that NusB binds directly and very selectively to SlO. The interaction is non-ionic and has an estimated Kd value of 10d7 M. We hypothesize that NusB binds to N-modified transcription complexes primarily by interacting with SlO. Keywords:

antitermination;

N protein; NusB; SlO; protein-protein

interactions

temperatures (Keppel et al., 1974; lower Georgopoulos et al., 1980). The nusE71 mutation, in the gene that codes for SlO, interferes with the function of N at high temperature (Friedman et al., 1981). There are mutations in nusB (Ward et al., 1983) that suppress the antitermination defects caused by the nusA1 (Friedman & Baron, 1974) and nusE?‘l (Friedman et al., 1981) mutations. Though these particular suppressor mutations are not allelespecific, such mutations could be indicative of direct or indirect functional interactions among proteins in a multi-protein complex (Jarvick & Botstein, 1975), including those involved in transcriptional antitermination (Greenblatt & Li, 1981a). Transcription complexes assembled in vitro in the presence of either E. eoli extracts or purified transcription factors contain N, NusA, NusB. SlO and NusG and are stabilized by multiple protein-protein interactions (Greenblatt & Li, 1981a,b; Horwitz et al., 1987; Mason & Greenblatt, 1991). We show here that NusB binds directly and selectively to ribosomal protein SlO. We hypothesize that NusB binds to N-modified transcription complexes primarily by interacting with SlO. at

1. Introduction Antitermination by the bacteriophage lambda N protein is an important mechanism of transcriptional regulation in the lytic development of the phage. The N protein modifies RNA polymerase transcribing the phage early operons so that transcriptional termination signals are not recognized (for a review, see Friedman, 1988). This modification of RNA polymerase by N requires the phageencoded nut (N-utilization) site (Friedman et al., 1973; Adhya et aZ., 1974; Franklin, 1974; Salstrom & Szybalski, 1978; Olson et al., 1982). Efficient antitermination also involves the host Escherichia coli proteins NusA (Friedman & Baron, 1974), NusB (Keppel et al., 1974; Friedman et al., 1976), NusG (Downing et al., 1990; M. Gottesman, personal communication; J. Li, R. Horwitz, S. McCracken & J. Greenblatt, unpublished results), and ribosomal protein SlO (Friedman et al., 1981; Das et al., 1985; Horwitz et al., 1987). NusB and SlO were originally implicated in antitermination by N on the basis of E. coli mutations that prevent N-dependent lambda growth on nusB and nusE mutant strains (Keppel et al., 1974; Friedman et al., 1976, 1981). The nusB gene product regulates the synthesis of rRNA (Sharrock et al., 1985) and is essential for the growth of E. coli

2. Materials

and Methods

(a) Materials KusB was purified as described (Swindle et al., 1988). Purified SlO was a generous gift from Dr V. Nowotny. Anti-NusB was purified as described by Horwitz et al. (1987). E. coli SIOO extracts were prepared from the wild-

t Author to whom all correspondence should be addressed at: 112 College St, Toronto, Ontario, Canada, M5G lL6. 55 0022-2836/92/010055-12

$o3.oojo

0

1992 Academic

Press Limited

t,i.fw strain ,I(: 11X ((:rrrnl,latt. I!GL). anti thr 2 nrutant strains .J(:-CciG (uusE?/ drrivnt,ivr of .J(: 14X c~onstrrtc~trcl 1)~ I)r I). 1. Friedman). and .J(:U!) (~/u&j tlerivativv of .1(:11X) as described (Coda & Grernblatt. I!IXf,). .4n E. co/i S%OO extract of strain .JGl4X was prepared by the same met hod as the SIN) extract. except that it was cyntrifug4 at 200.000 g for 2 h in a Beckman type 65 rotor. The c~oncetttration of protein itt the cxtra(.t,s iist~l in ttresrb t~xl)erirnents was drtrrminrd by using f
(b)

Preparation

of

GST and

G1\‘T-S10

protrim

A @rpsJ fusion plasrnid was construc%ed by cloning the rpd gene. encoding ribosomal protein SlO. into t.hr vrc%or p(:EX-3X (Smith Bi ,Johnson. 1988: Pharmacia). M hich contains the caoding sequence for the 26 kI)a glutathione S-transferasr (GSTt) protein under the control of a tcxc promot,er. This GST-SIO expression plasmid. p-1 1,:3X. was made in 2 steps: first an Au1 fragment (520 basepairs), containing the rps./ gene from pLL36 (Freedman rt al.. 19%) was ligat.4 int)o a Klrnowtreatrd RnntHT sitts of the pATHI I multiple cloning site (Dirckman 6 Tzagoloff. 1985: Sadowski rt (~1.. 1986). Plasrnids ~3ith both orientations of inserts were isolated. .A construc*t with the rpsJ reading frame in the reverse orirnt,ation relative to the reading frame of the trpE gene of pATH I I NW cut at its single HindIII sitr. treated with Kleno~ enzyme to produce blunt ends. and ligated to Ir:coKl (IO-mer) linkers. This was then cut with EcoRI to give a ,%O base-pair fragment that was inserted into the EcoRI sit,e in the polylinker of p(:ICX-3X. This produced the plasmid pJL38 with the reading frame of rp.J fused irifrarnr at the (‘-terminal coding end of t,hr yst gene. GST and QST23IO were expressed in JMlOl by induction of log-phase cells with addition of IPTG to 0.1 mht (Srnit,h HL .Johnson. 1988). The cells (2% g front a I I caulturr) were resuspended in 40 ml of PBS (phosphate saline: 1.50 mM-Ba(‘t. butTered 16 miv-?ia,HPO,. 4 mM-xaH,I’O,, pH 73), l.OO,C, (v/v) Triton X-100. lysed by sonication. and centrifuged at 10.000 g in a Sorval SS31 rotor. The crude supernatant’ (40 ml) was absorbed in a hatch. with gentle agitation. to 1 ml of glutathionrSepharose beads at 4°C’ for 30 min. The beads were gently pelleted by centrifugation at 1000 revs/min in Sorval RT6000 centrifuge and washed twice with 2.5 tnl and onct‘ with 10 ml of PKS (IO min each at 4’(‘. with gentle agitation). The specifically bound (:ST or GST--SlO was rluted from the glutathionr-Sepharose beads twice with 1 ml and once with 2 ml of buffer containing 10 mM-Hepes (pH 7.9). 20 mn-glutathione. IO mM-dithiothreitol. The purified GST and (:ST-SlO were then dialyzed against ,\(‘I3 (affinity column huff’er: IO mM-Hepes (pH 5.0). 10°,, (M./V) glycerol. 01 mM-EDTA. 1.0 mm-dithiothreitol) c.ontaining 500 mM-5aC‘t.

t Abbreviations used: GST, glutathione S-transferase: SDS/PAGE, SDS/pol~acrylamide gel electrophoresis: IPTG. isopropyl-1-thlo-j-D-galactoside.

\‘ariorts atttorttlts of SIO. (:$I SIO 01’ 2ti,t< \((‘ri~ c~~uplrtl to L’OOtng of .Afli-(:(iI IO ill ii total \-1111irt1~~01 0% ml of A(‘I3. (b.5 M-xa(‘l. .\fter c~oupling ovtlrtlight at 4 ‘C’. any remaining ester groups on the gel MPW I~loc~kt~~l by the addition of 20 ,uI of I ~I-t,thanolarninr HC’I (pH 8.01 and inc*ubatiott \!\a~ c~ontittur~cl for at least I It. Thch (‘ont’t’t1~ tratiotr of inrmohilizt~tl l)rotvin evils clf~trrnittlc~fl 13~ gc.1 t~lt:c~trophort~xis folloNcrtl I)>, stirinitrg mith (‘oottli~ic~ Irltit. to coniJ)are the starting tttatt~riitl with the, itttlolltlt 111 uttbountl pr‘olrirr Iefi at thv tint1 of tht, c~ortl)lirlg rcbac.tiorr. The 20 ~1 tnicro-alEnit)c~olurtttls \+tltv J)t,t~parc(I it1 IO0 01 200 PI capillary tubes ((~rr~trblatt AY I,I. l!Nl~/). ‘I’hcs columns ~rre \vashr(l with .\(‘I< c*otttainity 1.0 wS:L(‘l and then rquilibratrd wit It :\(‘I% c.ontaining IO0 tttlr Na(‘l (loading t)ttfGr). Ext,ract (40. 80 or IO0 htl) that hilcl I)1lelt dialvssrd against A(‘K c~ont,ainttrg IO0 tir+Sit( ‘I_ or I pg of’ l~ur~fietl Susl% tlilutrtl into 40 ~1 of’ .\(‘K. IO0 tttlr Na(‘I c.cmtaining 200 pg/mI of Ifoth insulitt anal l~~\-itlr. sertlnt albumin. ~a6 Ioatlrtl onto tht, colutttns antI tlrtsj \\vr’f’ Mashed \vith 10 colurntt volumes of ;\(‘I<. I(H) ttl\l-N;t(‘l. The c*oltttrltls \\ert t.htLn c~lutt~d srquerttially with .I( ‘I<. I4 nt-Sa(‘I mtl AC’R. I+“,, SI)S (100 /A eavh) Al)ottt half of the elution volume was thett subjrc.tetl to ILIW+ ro llhoresis ilt the l,rcsrncae of SI)S (I~at~rtimli. l!UO). f’r~ott~irts were visualized by stainirtp Lvith silver (Jlorrissc~> I!)XI ) or they \vert‘ transftirreti to nitrocellulose and l)roi)tvl \vit h 0.5 pg affinity-J)urifird anti-Nusl3 or atlti-SlO,nll ‘I’hc, bound J)rimary antihodv was tletvc+~vl I),! ~1111111,110~ staining \vitt1 goat ;Itlti-ral)l,it-I#(: alkalint~ l)hosphatast. (*ottjugatt. (l%io~ Ratl) cl> tlvsl~ribc~tl I)\ tht, n1atinfac4 t~rvr ((I)

t’rrp~~ralion

f!f’

ujiir~itg

purijiud

unli-~4’11)

crt//i/mrl!/

Rabbit ant,isrrum raised against puritietl ribosotnal protein SJO was a generous gift from L)r I,. Kahan, .\rtti-SIO was purified from crude serum by affinity cfIrotriatography at room t,rmprrat,urt*. Serurtt (1 ml) was l)assed over a 0.5 ml column cont,aining 2 mg of C:ST~SIO irttrnobilized on Affi-C:el IO. Tflr column was thoroughly Lvashrd and the antibodies were t,lutt:d with 100 tn>T-glyc.irrv HC’I (pH 2.5). immediateI?neutralized wit,h I M-Tris. HC’I (pH 8.0). and dialyzed against phosphate tmfl’rretl saline.

Purified SusB, purified SIO. or both. at various (‘or1ve’1itrations. plus 0.5 pg of myoplobin and I00 pg of bovint~ serum alburnirt were incaubatrd for I5 min at 37 (’ in it total volume of 200 PI AC’R. and then layred onto a 4.0 ml 100,, to 300+, (w/v) linear su(arost’ gradient in :\(‘H. The gradients were centrifuged at 55.000 revs;mitt for 14 h at 1°C‘ in a Beckman S\VBOTi rotor. I’our-drop fractions ( - 2’45 PI) were c~ollrct8rd from the bottom of the centrifuge tube and the whole fraction was IyoJ)hilizetl. resuspended in 100 ~1 of Laernrnli (IWO) sample buffer. subjected to SI)S/polvacr\;lamitie gel t~lrc,troph”rt,sis. ant] analyzed by staining with silver (Morrissry. l9Rt ).

3. Results (a) Afinity

chromatography f&on protein

using a UN--,Yl(I

\+‘e have used affinity chromatography protein-protein interactions involved

to itientif’v in antitertn-

nation by the lambda N protein ((:reenhlatt 1981a.h). Tn order t,o define a role for SlO

& Li, in this

,!usB-SlO

system. we used affinity chromatography to determine whether any antitermination factor in a crude E’. coli extract could bind to SlO. Analysis by SDS/PAGE followed by staining with Coomassie brilliant blue R of th e purified proteins used as chromatography in these experiligands for affinity ments is shown in Figure 1. Since SlO is relatively difficult to purify on a large scale, we initially used a GST-S10 fusion protein as the ligand. A gst-rpsJ fusion gene in the plasmid pJL38 was created by subcloning thr complete SlO-encoding rpsd gene into the polylinker of pGEX-3X (Smith & Johnson, 1988), which contains the sequence encoding glutathione S-transferase from Schistosoma japonicum under the control of the E. coli tat promoter. After induction with TPTG of E. coli strains containing pGEX-3X or pJL38, GST and GST-SlO were produced as approximately 25% of the total cellular protein. The GST was totally soluble whereas the GST-SlO fusion protein was approximately 50 y0 soluble (data not shown). GST and GST-SlO were purified by chromatography on glutathioneSepharose beads followed by elution of bound proteins with buffer containing free glutathione (Smith & Johnson, 1988), and the resulting preparations were nearly homogeneous (Fig. l(b), lanes 2 and 3). Affinity columns were made by covalently coupling purified GST or various concentrations of GST-S10 to Affi-Gel 10. These columns were loaded with an

97.4 68 -

Interaction

37

E. coli 8100 post-ribosomal supernatant fraction in order to avoid detecting ribosomal proteins that bind to SlO. The columns were washed with loading buffer, and eluted with buffer containing 1.0 M-?ll’aCl followed by buffer containing 1.0% SDS. The proteins were analyzed by SDS/PAGE followed by staining with silver (Fig. 2(a)). There were no E. co& proteins that bound to GST-SlO columns and not to GST columns that could be eluted wit,h 1 M-KaCl (data not shown). Several proteins were eluted with l.Oo/o SDS from the GST-SlO columns (Fig. 2(a)> lanes 2 to 8). However, all but one of these were also eluted from a GST-SlO column that had not been loaded with extract, (lane lo), and were, therefore, small amounts of protein ligand that were not covalently coupled to the matrix and could be released with SDS. Of particular interest in the GST-SlO column eluates was a protein with an apparent’ M, of 16,000. This protein, which was not derived from the MT-S10 ligand preparation (compare lanes 4 to 8 with lane lo), bound in large amounts only to the GST-SlO columns with the highest ligand concentrations (lanes 5 to 8). The 16,000 M, protein had the same electrophoretic mobility as purified NusB that was run in parallel on the same gel (lane B). We have prepared affinity-purified anti-NusB antibody (Horwitz et al., 1987) that recognizes only NusB and a 40.000 M, cross-reacting E. coli prot,ein in Western blots of crude extracts (see Fig. 6(b), below). West-

97.4 68

31

-

21-5

-

14.4

-

Figure 1. Purified proteins used in these experiments. (a) The purified proteins NusB (2 pg in lane 2 and 3 pg in lane 4) and SlO (2 pg in lane 3 and 3 pg in lane 5) were electrophoresed on a 135o/o (w/v) polyacrylamide/SDS gel and the proteins visualized by staining with Coomassie brilliant blue. (b) The purified proteins (2 pg of each) GST-SlO and GST were analyzed as in (a). M in both (a) and (b): 2pg each of phosphorylase B (97,400&f,), bovine serum albumin (68,000 A&), ovalbumin (43,000 M,), carbonic anhydrase (31,000 M,). soybean trypsin inhibitor (21.500 Afr), and lysozyme (14,400 M,),

G-SIO-,

r I

G

G-SIO

+-,+ 0

0.8

I -2

2.0

-

2.8

4.2

5.6

7.0

5.8

G

: L iqand

-

: E.xtroct lmmobilil llqond

5.8

7.0

31-

GST-SIO

-

GST

I(

10-J M )

-

-

14.4

-

!ed

2

I

3

4

5

2.0

2.8

6

9

6

'

Ii

IO

t-3

(b)

46-

,-G-S10 0 Cc)

-, 0.0

I.2

4.2

G 5.6

7.0

5.8

: Ligand :[‘m;;;;;zed]

(~o-“M)

9072 46-

2&5-

10.5-

NusB

15.5-

Y

I

23456789E

IO

20

I

40

60

00 I

I Purified

NusB

(ng)

Figure 2. Affinity chromatography with immobilized GST-SlO. Affinity columns containing immobilized GST-WI0 at ligand concentrations indicated at the top of the Fig., and control columns containing either no immobilized protein (lane 1) or immobilized GST (58 x lo-’ M; lanes 9 and 11) were loaded each with 80 ~1 (@8 mg protein) SlOO extract (lanes 1 to 9) or ACB, 100 mM-NaCl (lanes 10 and ll), washed with ACB, 100 mM-NaCl, and eluted sequentially with ACB, 1.0 M-N&cl (not shown) and ACB, 1.0% SDS. The SDS eluates were divided into 2 equal portions, subjected to electrophoresis on 13.5% (w/v) polyacrylamide/SDS gels and the proteins were visualized by (a) staining with silver, or (b) following electrophoresis the proteins were transferred to nitrocellulose and probed with affinity-purified anti-NusB. (c) One half of the flow-throughs from each column were electrophoresed, transferred, and probed as in (b). Lane B in (a) and (b): 40 ng of purified NusB. Lane E in (c): 40 ~1 of extract which is one half of the amount loaded onto the column. The positions of the marker proteins, as for Fig. 1, are indicated on t,he left of (a). The positions of BRL pre-stained marker proteins are indicated on the left of (b) and (c): ovalbumin (46,000 Mf), carbonic anhydrase (28,500 M.), /?-lactoglobin (18.400 M,), lysozyme (15,400 M,).

:

Elution

I

0

I M-NaCI

I

Extract

O-25

2

I

0.5

(a)

I I

Purified NusB

1 %SDS

Extract I

0 0.25 0.5 0 0.25 0.5

3456789

SlO

NusB

18.4

46=- 28.5 -

I

I

Extract

(b)

I

0.25 0.5

1 % SDS

0

:

:

:

-

Elution

load

Immobilized SlO (mg/m I )

NusB

I

Figure 3. AiffinitT- c.hrotnatopraphy with ima~obilizrtl SIO. Attrinity cwlumns c~orttaininp immobilizrd SIO. at the indicated ligand concentrations. MPW loatlrtf with 40 pl (0.34 mp protein) 0; S200 cbxtrwt (lanes I to 6) 01’ I pp of purified Susl3 (lawn 7 to !3) washed with :\(‘K. 100 mwSa(‘l. and rlutrtl srquenti;illv with ;\(‘I3 c,ontaininp txithrr as for Fig. L)(a). ‘I’)w l wsitions of SusR and SlO arv indicated on the right. I.0 M-XaCI (lanes I to 3) or l~O”, SDS (lanrs 4 to !)). (a) Eluates uerv (,l(,(.trophorrsetl and visualiwl The positions of protein markers. as in Fig. 1. are indicated on the left. (b) Thr SIX3 rluates shown in (a). lanes 4 to ti. wert’ electrophoresed as in (a). transferred to nitrocellulose. and probed with aSnit?-purified anti-KusH. NusH standards: 10 and 40 ng of purified SusK. as indicbated. The position of SusB is indicated on the right. Thr positions of y-e-stained marker protems. as for Fig. I(b). with the addition of bovinr swum alhurnin (i2.000 MC). are intlic*ated on thr left. Immobilized S10 c:orlc~erltra,tiorIs: 026 mg/ml = 2.3 x 10m5 M and 0.5 mg/ml = 4.6 x IO-’ M.

:

1:

load

Immobilized SlO (mq/ml)

-

-

1

cbrn blotting with the aflinity-purified ant~i-NnsH antibody identified the 16,000 M, (:ST -SlO binding protein as NusR (Fig. 2(b)). The amount of NnsH i hat bound to ckac*h GST-SlO column was dett>rmined by measuring t,hr density of cLac*h band in a digitized comput,er irnage of the Western blot) shown in Figure B(b) and comparing thern to the densit,ies of other NusK st.andards that were present on thr CVestern blot,, but not shown in Figure 2(b). The amount of NusH in each lane in Figure P(b) was: 1 iig in lane 1, I1 ng in lane 5. 31 ng in lane 6. 41 ng in lane 7. and 4-t ng in lane 8. X small amount of a protein with the same electrophorrtic mobili)? as NusH bound to caolumns containing no immobrllzrd (:S’l ligand (Fig. 2(a). Iam 1) or Imrnobillzed (Fig. 2(a), lane 9). but Western blotting revealed t,hat t,his background E. coli protein binding t)o the column matrix was not NusR (Fig. 2(b). lanes I to 9). To further demonstrate that NusIZ binds quant,itativcly to SIO. the flowthroughs from th( GST-SlO columns were Western-blotted and probcld wit,h anti-NusK. Tt) was found that essentially all of t,hr KusB in the extract bound to the> (:ST--SIO columns with t,he highest> immobilized ligand COW c~rntrat,ians (lanes 7 and 8. Fig. 2(c)). while the Howt>hrough from the two control columns (lanes 1 and 9. Fig. 2(o)) were not appreciably depleted. Therrhfore, t,he experiment shown in Figure 2 implied t,hat t,hr only non-ribosomal Ii:. cd prot,ein t.hat binds selectively and quant,it,atively t,o SlO is NusH.

(b) A,finity

chromatography

u,sing

immohilizd

SlO

To establish t,hat the binding of NusK to (XT-S10 columns was not, influenced by the GST rnoiety of the fusion protein. affinity chromatographv was also performed using immobilized SlO. The SlO used in these experiments was nearly homogeneous (Fig. l(a), lanes 3 and 5). Although we have not detected ribosomal proteins that bind to SlO in the SlOO extracts that were used in other experiments, the E. coli S200 extract was used in this experiment to ensure that all ribosomes were removed from the extract. SlO columns and a) control column containing no immobilized protein were loaded with E. coli S200 extract. After the columns were washed with t’en column volumes of loading buffer they were eluted successively with buffers containing 1.0 M-NaCl and l.O”i, SDS and the eluted proteins were analyzed l)g SDS/PAGE followed by staining with silver (Fig. 3(a)). There was no protein in t,he crude extract, that bound t)o the SlO columns, and not the control column. that could be eluted with 1 M-NaCl (lanes 1 to 3). The only proteins eluted with 1.0% SDS from an SlO column (lanes 5 and 6) and not from a control column (lane 4) corresponded in elect’rophoretic mobility to SlO (12,000 M,, lanes 5 and 6) and NusH (16,000 H,, lane 6). The 12,000 M, protein released with SDS specifically from the SlO columns represent,ed a small percentage of the SlO protein ligand that’ was not covalently coupled t,o t)he column

rrrat,ris. The SIW rluat)es showli ii) Figrirc~ 3(a). lit~ios 1 to 6. we’re \I’t~stt,i,rl-l)lottt,d with affinit), puritic,d an&SusH. whicah r<‘vtbaled that the 16.000 M, SlO binding prot,rin was indrrcl SusH (Fig. 3(l)i) ‘I’hr~rc~ fore, SIO ~l and independently of ilny otti(br R. CO/~ molt~(.ul~~. purified Sust% was also c,hro,n;ttc,pray)ti(~cl on (7)~~trol and SIO (.oIumns. The Nusl3 usc~i in t h(ah(sca.xptsri ments had been purified to homogenr>ity (S\vindtr et 01.. 19X8: see Fig. 1(a). lanes 2 and 4). So prlrific~tl XusB was eluted from SlO columns 1)~ I \I-Nii(‘l (data not sho\vn). but NusK was rlutrcl n.ii tr I+“,, SDS from the SlO columns (Fig, 3(a). lam’s ?( arrcl 9) and not from a control c*olumn (la,no 7). log(~t hc~r with some of the SlO ligand. These rtasults t hc~rt~forc~ indicated that) NusR binds directly alld s&c~t iv(hl~ to SlO.

To confirm the srlrc%ivit>- of the NusM SIO inter action. afliniQ- c~tirornatoglaptl~ experimerlt s \vtarfa also performed using immobilized XusH. ( ‘olurnn5 containing inimobilizrd Nusl3 and c~ontrol columns containing no immobilizc~d protc4tl wf’r(’ Ioatl~~l wit tr E. co/i SlOO extract. washt4 Lvith ttill (.otnmn volumes of loading buffer. atIt vlutrtl s~~clut’t~t~iall> with bufft~rs caontaining t a~-Na(‘l at~ti I4”,, SI)S. The eluatrs were analyztld t)y SI)S:T’A(: E followt,d by silver staining (Fig, 4(a)) and l\‘rJstclrn l)lot tillp u:ith affinity-purified anti-S10 (Fig. 1(b)). Thr~rc~ \VHS no protein that bound specifically to thp SusH column (Fig. 4(a). Iam L’) and not to thr, control column (lane 1) and was eluted wit,h I MM-Na(‘l. but il single 15’.coli protein t.hat c*o-migrated Fvit h purifircl SlO (SlO lanes in Fig. l(a)) was elut,ed wit8h I+“,, SDS from the NusH csolumn (lane 1). atltl not from the cont,rol column (lane 3). \Vhen thth sam(x (4nat(~h were electrophoresed on an SI)S-polyac~t.vlarrlitit, gt.1. followed by transfer to nitrocellulosc> and probing with an affinity-purified anti-S10 antibody that recognizes only SlO and a c-ross-react ive 66.000 !)I, E. coli protein in a Wrst,ern blot (see Fig. 6(a). below). it) was found that the 12.000 M, I:‘. co/i KusB-binding protein was SlO (Fig. -C(h). lancb 4). Therefore, just as SlO columns select.ivc~ly binci KusB. XusB affinity columns selec~tivrty I)illtl SIO.

In order to detrrmint~ the strength 01’ t,hcl SlO-NusB interaction and to detect it with a method that is independent of affinity ctrtomat,ography we used sucrostb gradient sedimentation. Purified SusK and/or SlO. at, various cone-thntrations. were incubated and loaded onto lOl;, to 30”,,, sucrose gradients. A large amount of bovine serum albumin was included in all the mixtures to serve as a carrier protein, and myoglobin was included to

97.7 66 43

21.5 21.5

CBCB II C

B

C

B

NaCl

INaCl

SDS (bl

SDS (a)

Figure 4. Afinity chromatography with immobilized RiusB. Afinity columns containing either ((J) no immobilized protein or (13) 05 mg (3.2 x 10m5 M) immobilized h’usB/ml were loaded with 100 ~1 (1.0 mg protein) of 17. coli 8100 extract. scashrd with ACB. 100 mM-RjaCI, and successively eluted with ACR containing 1 M-NaCl (lanes 1 and 2 of both as for Fig. 2 and panels) and A(‘B containing 1.0 7; SDS (lanes 3 and 4 of both panels). Eluates were electrophoresed proteins were visualized by (a) silver staining or by (b) Western blot probed with affinity-purified anti-SlO. S10 standard (both panels): 1 I, 22 and 88 ng purified SlO, as indicated. Extract in (b): 2 or 10 ~1 of E. coli SlOO extract, (%C)or 100 pg of protein) was loaded directly on the gel. Because such a large amount of extract was loaded onto the gel, a protein with an :Vr of 65.000 that cross-reacted with the anti-S10 (b) spread during electrophoresis so that it appeared to be present in the SIO standard lane. The 65,000 M, protein is not present, in purified SlO (see (a)). M in (a), marker protr+ns as for Fig. 1. M in (b). pr+sta.ined marker proteins as for Fig. B(b) and 3(b).

as a sedimentation standard. When SlO was sediment’ed alone. its peak was in the ninth fraction (Fig. 5(a), lane 0). one and one-half fractions behind the peak of myoglobin (lanes 7 and 8). Centrifugation of XusH produced a peak in the eighth fraction (Fig. 5(b). lane 8). one-half fraction behind myoglobin (lanes 7 and 8). When the two proteins at the same concentrations (6 x lo-’ M, 1 : 1 molar ratio) were mixed prior to centrifugation, the peak of SlO co-sedimented wit.h t,he peak of NusB in fraction seven (Fig. 5(c). lane 7), one-half fraction in front of the peak of myoglobin (lane 7 and 8). LJsing both bovine serum albumin (4.5 S) and myoglobin (2 S) as sedimentation standards, the approximate sedimentation constant’s were 1.5 for SlO (apparent MT = 11.000). I.9 for NusB (apparent M, = lS,OOO), and 2.1 for the NusB-SlO complex (apparent M, = 18,000). Thus. our data are most consistent with the formation of a highly asymmetric complex containing one molecule each of SlO and NusR. The c*onc*entrations of SusB and SlO in the serve

were systemat~ically mixt.ure wit,h decreased [NusB] = [SlO] in order to determine t,he approximat,e concentration at which the complex dissociat.es (Fig. 5, (c) to (e)). Though the two proteins were present in equimolar amounts in (c) to (e), the amount of SlO relative to NusB appeared t’o be less because SlO stains less well with silver (see equal amounts of NusB and SlO run as standards in all panels). As mentioned above, all of the SlO still c’osedimented with KusB when the concentration of both proteins was 6 x 10V7 M (Fig. 5(c)). but the SIO moved somewhat more slowly at 3 x 10. 7 M and much more slowly (half way (Fig. 5(d)), between free SlO and the SlO-XusB complex) at 1.5 x lO-’ M (Fig. 5(e)) due to dissociation of the complex. Thus. approximately one-half of the was complex dissociated at approximately 2 x 10Y7 M. If the two- to threefold dilution of the samples during centrifugation is taken into account, the XusB-SlO interaction has a Kd value of about

lo- ’ M.

Sedimentation

4 1

2

3

5

4

6

7

8

9 10 11 12

13

14

S BM

(a) SIO

6xlo-7 M -

Myoglobin NusB SlO

1

2

3

4

5

6

7

8

9 10 11 12

13

14

S BM

(b) NusB 6x10-7 M

1

2

3

4

5

6

7

8

9 10 11 12

13

14

-

M yoglobin NusB

-

SlO

S BM

w S 10 & NusB 6xlo'7M Myoglobin NusB SlO Fig. 5.

NusB-810

1 2

3

4

5

6

7

8

Interacticm

9 10 11 12

13

14

63

S BM

id) SIO & NusB 3x10” M M yoglobin NusB SlO

1

2

3

4

5

6

7

8

9 10 11 12

13 14

S BM

SlO & NusB 1*5x10-‘M M yoglobin NusB -

SlO

Figure 5. Sucrose gradient, co-sedimentation of NusK and SlO. Either (a) SIO, (b) NusK. or both SIO and SusH together at (CL)6 x 10m7 M; (d) 3 x 10m7 M; (e) 1.5 x lo-’ Y. were mixed with 100 /~g of bovine serum albumin. and 0.5 ~(a of Fractions myoglobin in 200 ~1 of buffer, incubated, loaded onto lO(& to 30’>C, linear sucrose gradients and centrifuged. were collected from the bottoms of the tubes and analyzed by sIW/PAOE followed by staining with silver. The direction of sedimentation from top to bottom is indicated by the arrow. Breakdown products of the bovine serum albumin added tjo each mixture as a carrier protein are visible in some fractions. especially lanes I to 4. These peptides are also present in untreated bovine serum albumin (data not shown). S, 55 ng of purified SIO; IS. 35 np of purified NusB: 11. 50 ng of myoglobin.

(e) 7’ho effect

of the

nusIC71 and nusB.5 mutations

The rtusE71 mutation in rpsJ (Friedman et al.. 19X1) and nusK.5 mutation (Friedman et al.? 1976) impair antitermination by N in viva (Friedman et al.. 1976. 1981) and in vitro (Das & Wolska, 1984; (ioda 8: Greenblatt, 1985: Horwitz et al.. 1987). The nusE77 mutation has been shown to alter the of SlO (Friedman et al., electrophoretic* mobility 1981). To t,nst whether the nusE71 defect, (Friedman et ccl.. 1981) affects the NusB-SlO interaction we used immobilized NusB for affinity chromatography wit,h crude E. coli extracts from wild-type and nusE7l mutant strains. The extracts were loaded onto control columns containing no immobilized protein and columns containing various concentrat,ions of immobilized NusB. The columns were run as hefmr. Armlysis of the SDS eluates from these

columns by SDS/PAGE followed by transfer to nitrocellulose and probing with affinity-purified anti-S10 antibody (Fig. 6(a)) revealed that SlO from the wild-type ext’ract bound to all three NusB columns (Fig. 6(a), lanes 2 to 4) and not to a control column (lane 1). Interestingly, a protein with slightly lower electrophoretic mobility than SlO, present only in the nusE77 extract, also bound specifically to the NusB columns and reacted with the affinity-purified anti-S10 (lanes 5 to 8). Though the mobilities of the two NusB-binding proteins det,ected in Figure 6(a) were different, we are confident that the NusB-binding protein in t,he mutant extract is SlO because of the specificity of the afinity-purified antibody used in t,his experiment. It detects only SlO in wild-type extract, the protein with altered mobilitv in mutant extract. and a 65,000 M, cross-reacting protein present in both

M

Inlmobiliz ed NusB [

Extract

12

1. .

:

0

0.25

3

4

0.5

1.0 I

I

5

6

0

0.25

78

o-5 1.0

Wild-type

(b)

I

I nu.sE

71

(a) Figure 6. Hffects of the nush7’1 and nusH3 mutations. (a) (lolumns c.ontaining immobilized XuslS at the indicaatrtl concentrations (mg/ml) were loaded with 100 ~1 (I.0 mg protein) of either wilti-t,ype SIOO extract, (lanes I to 4) or n~&T7 mut,ant extract (lanes 5 to 8), washed with ACH, 100 mM-PjaCII. and eluted successively with AC’R, I.0 WE&Y (not shown) and ACR, l.Ooi, SDS. The SDS eluates were subjecbed to SDS/PAGE. transferred to nitrocellulosr. and probed with affinity-purified anti-810. SlO standards: 12 and 88 ng of purified SIO. as indicakd: 11. prr-stained markrr protrins as for Fig. 2(c). Extract standards: 10 ~1 of’SlO extract from wild-type and n~sE’i1 strains. as irrdic.atrtl. Both extracts (Gontained a 85,000 M, protein that cross-reacted with anti-SIO. The apparrnt presen~r of this same bard ill thcl XX ng of purified S10 standard lane is an artifact due to widening of the extract lane next to the SlO st,andartl lane. (b) Bxtrac+s (10 ~1) from wild-type and nusB5 mutant cells were subjected t,o SDS/I’A(:E: followed by Western blot, analysis with affinity-purified anti-NusR. NusR standard: 80 ng of purified SusB; M, prr-stained molecula,r weight st~andartls as fol, Fig. d(h). Immobilized Ku& concentrations: 025 mg/ml = I+5 x IO.-’ w. 0.5 mg/ml = 3.2 x IO-’ YIP and I.0 mg/ml = 6.4 x 1W5 M.

extracts (see Fig. (i(a)). The change in the SIN polyacrylamide gel mobility of the mutant 810 in the nusE71 strain may be caused by a change in it,s isoelectric point (Friedman et al., 1981). In any case. the nusE71 mutation does not directly affect the interaction between NusB and SlO. We also wanted t,o assess the effect of the nusR5 mutation on the NusB-Sl0 interaction. However. when crude extracts from nusB5 and wild-type strains were compared in Western blots probed with affinity-purified anti-NusB antibody. we found that

no XusH could he tlet,ec+rtl nusHS skain (Fig. B( 1))).

in an extract

tiorri

;I

4. Discussion The assembly of elongation complex protein interac:t,ions. have been detected graphy. We used this to NusA (Creenblatt

an N-modified transcription involves multipie proLein-Several of these interactions by protein aflinity chromatomethod to show that Iv binds & Li. 1981a) and that, NusA

~YutsB-SIO Interaction binds to RNA polymerase (Greenblatt $ Li, 19816). More recently. we found that immobilized NusG binds directly and selectively to the t,ermination factor Rho (J. Greenblatt & J. Li, unpublished results). Tn eacsh case, the immobilized protein ligand selectively bound to only one or two proteins that were present’ in an unpurified E. coli extract. Therefore, we are confident that the protein-protein interactions that are identified in this way are very likely to be indicative of functional interact)ions. W’e have shown here, that NusB binds directly and with very high srleot’ivity to SIO. Indeed, when crude K. coli extra& were used for affinity chromatography. SlO columns bound only NusH. and NusK caolumns bound only 810. (:enetic: st.udicLs have shown that mutations in the VULSR gene (Wa,rd et al.. 1983) can suppress the nusE71 mut)ation in 610 (Friedman et al.. 1981). which prevents antiterminat’ion by N. These types of suppressor mutations are often evidence of a functional interaction between the proteins encoded by the mutated genes (Jarvick & Hotst,ein, 1975; (:rernblatt dt I,i, 19Xla). The experiments described here suggest that NusR binds directly to SlO in N-modified transcription complexes. However, the nusE71 mutation in SIO does not’ affect t)he binding of NusB to SIO in affinit)v chromatography experiments. That is, perhaps, not surprising. since the same mut,ations in nusH that’ suppress n~sE71 also suppress the n~ssA 1 mutation (W’ard et al.. 1983). Therefore, suppressing mutations in the nusI? gene may compensate for alterations in NusA or SlO by providing additional contacts between NusB and another rnolrcnlt~ in thr N-modified elongation c~omplt~x. (Friedman ef nl.. 1976) also The nus/l,5 mutation impairs antitermination by N. We have found that rxtracts from a nuaR5 strain contain an undetect,able amount of NusB. We think that the nusRS mut,ation makes the Xusl3 protein unstable. The antitcrrnination defect in t,he nusRS mutant’ strain is almost certainly caused by bhe decreased amount of NusH available for antitermination. The dissociat,ion constant for the NusBPS1O interaction can be estimated from the concentration of immobilized GSTPSIO that was reyuirrd for the efficient binding of NusB to the column (see Fig. 2). The minimum cboncrntration of immobilized ligand that was required to efficiently bind NusB was approximately 3 x 10 ’ M. Taking into account that our columns were washed with ten column volumes of bufGr prior to t#heir elution with l.OO,, SDS, and assuming that all of the SlO l&and was capable of for the dissociation binding Nusl%. an estimate csonstant is about 3 x IO 6 M, However. the K,, value rlst,itnated from sucrose gradient co-sedimentation (Fig, 5) was approximately 1OV’ M. This value was derived from the concentration at which approximately half of t.hr Nus&SlO complexes were dissoc*iat>etl. This Kd value is similar to the dissociation constants of the NusA-RNA polymerasr (- 10- ’ 31: (:reenblatt & I,i, l981b: Schmidt bi (‘hamberlin. 1984) and tht> N-SusA (- 10m6 YVI;Creellblatt & Li.

65

1981n) interactions

of this system. The difference in by sucrose gradient co-sedimentat,ion and protein affinity chromatography can be reconciled by assuming that not all of the (:ST-SIO ligand used for affinity chromatography was accessible for binding NusB. Therefore, the true dissociation constant for the NusB-SlO interaction is probably that estimated from the sucrose gradient experiments (Fig. 5). The interaction between IVusB and SIO is prohably entirely non-ionic. Although thr init,ial binding of SusH to SlO is slightly affected by salt (datea not shown), the final complex is probably held toget)her only by hydrophobic interact’ions since it is not sensitive to 1 M-NaCl (Figs Z and 3). elsewhere, (hlason & 1 11 work described Crrenblatt. 199l), we found that’ SlO also binds weakly. but specifically, t,o RNA polymerase. Therefore, the stable association of SlO with the N-modified transcription complex (Mason & Grrenblatt, 1991) is probably produced 1)s t,he sum of two weak interactions, one with RSA polymerase and one with XusB. Similarly, the stable assoviat,ion of SusK with N-modified elongation complexes (Horwitz rt al.. 1987; Mason Rr Greenblatt. 1991) must) involve SlO, but could not be produced solely 1)~ the weak binding of NusB to SIC). The presence of Nusll in the isolated elongation complex depends not only on SlO, but also on the nut site and ot’her proteins (Horwitz Pt al.. 1987; Mason & (irernblatt. 1991). One or more of these fact’ors, or t’hr vrut site RNA, and in particular hoxrl (Friedman Pt crl.. 1990), may also contact’ NusR wit bin the N-modified elongation complex. Therefore. we envision that a,n S-modified elongat~ion caomplex is forrned in a highly co-operat,ive manner and is stabilized by many weak l)rott,in~~l’rolein and prott~inPK.SA int,eractions.

Kd values derived

The authors thank all the members of the laboratory for helpful discussion and advice. In particular we thank Drs S. ,McC!racken and L. Desbarats for t,echnical and theoretical advicae and discussion. We also thank Dr V. Irjowotny for the gift of purified SlO, Dr L. Kahan for the gift of anti-S10 antiserum, and Dr ,J. Segall and Dr H. Krause for a critical review of the manuscript. This work was supported by the Medical Research (‘ouncil of (‘anada.

References Adhya. S.. Gottesman. M. & de Cromhrugghe, K. (1974). Release of polarity in Escherichan. co/i b? gene :V of phage 1: termination and antitermination of’ transcription. I’roc. Xat. .4cad. Sci.. C...\‘..3. 71. “534-2.538.

Dab. A. & Wolska. K. (1984). Transcription antitermination in, Ctro by lambda S gene product: requirement for a phagr nut site and the products of host nusA, nusB. and nusE genes. 0~11, 38. 165-173. Das. A.. Ghosh. B., Barik. S. & Wolska. K. (1985). Evidence that, ribosomal prot,ein SlO itself is a cellular component necessary for transcription antitermirration by phage 1 h’ protein. Proc. ,Vrrt. .Icnd. Sci., I:.S.d. 82. 4070~~4074. 1)iecakma.n. C’. 1,. & Tzagoloff. A. (198.5). Assembly of the

nlitoc.hondriaI mc~tnbranr~ systvtn. (‘1~1%. a > wst nrcc~f~~ar gytltl ttwwstwy f’or synthesis of’ cytochromts h ./. Hid (‘Hun,. 260, 151:~ 1.W). I)owning. iV. I,.. Sullivan. S. I,.. Gottestnatr. M. E. c‘+ f)entiis. I’. I’. (1990). Sryuenw and transcript~iotit~l ICschPrichin m/i .srrKnerd: ftattvrn of thr rssential ~~peron, .J. Nnctrriol. 172. Ili%l 1627. Franklin. ;2. (‘. (1971). Altrrrtf reading of genetic8 signals f’usd to thtl .I’ ofwron of Itactt~riophagr i,: getwtic rvitlrnc,r for tnotlific~ation of ftolymrrasr by the ftrcjtvitr ftrodu~~t of the .V grnr. ./. Mol. /{id. 89. :1:3 4X. .I. At. K: 1,ititfahI. I,. (f!M). Frrrdtnan. I,. I’.. Zrtrgal. (:t~nc~tic* disstv*tiott of strittgtwt cwntrol atttf nritrof thr h!.schvrichirx co/i St0 tional shift-ttf) rrsfx~rtst~ rittosottial protein ofwroti. -1. Mol. Rid. 185. 70 I i 11. Frircfman. I). I. (1988). R~epulation of f’hage gene c.xftression by t,ermination and antitermination of tranwrifttion. Tn 7’hr Hacf~rioph.rryrs ((‘alrndar I< tsd.). vol. 2, f’lenutn. XPW York. f~‘rirtlmatt. I). I. & Rarotl. I,. S. (1971). (:rnrtic, (.hwactw izat.ion of’s hac~tc~ria.1 loc.us ittvolvtd in thtx ac,ti\-ity of thcb AV frnction of f’hagr E.. C’iroloyy, 58, I-Cl 11X. f~‘rirtltttan, I). I.. 12’ilgrts. (:. S. & 1fural. I<. .J. (l!IT:l). (1eitt~ S wgulator f’ttnc+ion of’ f)hagt‘ Iatnhtla i,r/rf/YI: vvitlt~nc~e that a sitr of’ .I’ action tfiffrrs f’rom a sitr ot .V rvcwgnitiotl. -1. MO!. Niol. 81. 505 516. f~‘riecfman. I). I.. fLumann. 31. F. & 12arott. I,. S. (I!G(i). (‘oo~wrativ~ t+frc+s of hac~trrial ttrutatiotis afbxtitig Iatnf~cia S grnr rxf)rrssiotr. I. Isolation and c,har;tc,trriz;rtic,rr of’ a rcrcsH ttlutant. I’irology. 73. 11’1 f”7. f~rirtlttiati, I). I.. Svhaurr. X. T.. fbumann. 1Lf. I<., f%aron. b:vidtwc~c~ that rihosomal I,. s. K: Adhya. s. I,. (1981). f,r.ottAin St0 ftartic~ipatrs in c.ontrol of transc,rifttioti Sd. .J,md. ,Vci.. 1 ..S..-l 78. trrmination. I’mr. 1115 111x. Frirdtt~an. I). I.. Olson. E. R.. .Johnsott. 1,. I,.. Alrssi. I). Kr Yl. (:. (1990). Trans~rifttiotr-deprrltfrttt (~I~11\‘PII. c~c)mJwtitiotr f’or a host ftwtor’: thr fuwtiorr and ol)titnal styuonvv of the phape ,&Lc=~ transcrif)tiotr ant,itc~t,triinat,ic,n signal. Oenm I)pl~(i/oj~. 4. 22 fO~-L~d~. (&wrgof)oulos. (‘. I’.. Swintflr. .I.. Kepprl. F.. 13allivtbt. Xl.. f