Dissociation of the Pf1 nucleoprotein assembly complex and characterisation of the DNA binding protein

739 (1983) 216-224 Elsevier BiomedicalPress

216

Biochimica et Biophysica A cta,

BBA 91184

DISSOCIATION OF THE Pfl NUCLEOPROTEIN ASSEMBLY COMPLEX AND CHARACTERISATION OF THE DNA BINDING PROTEIN G. GEOFFREY KNEALE European Molecular Biology Laboratory, Postfach 10. 2209, D - 6900 Heidelberg (F. R. G.)

(Received September27th, 1982)

Key words: DATA-protein interaction," Nucleoprotein; Virus replication," Bacteriophage Pfl

During replication of bacteriophage Pfl, progeny viral strands are complexed with a single-stranded DNA binding protein, analogous to the gene 5 protein of bacteriophage fd. Using fluorescence spectroscopy, ultracentrifugation and DNA-cellulose chromatography, conditions for dissociation of the nucleoprotein have been investigated. The Pfl protein is unusual in that it is not released from the DNA by 2 M NaCI. Complete separation occurs in 0.6-1.0 M MgCI2, leading to a procedure for the purification of the protein. Two subfractions of the protein can be isolated of isoelectric points 5.9 and 6.4. The molecular weight of the native DNA binding protein has been studied by gel filtration and sedimentation. The major species in solution has a sedimentation coefficient of 2.3 S and a diffusion coefficient of 7.8-10 -7 cm2"s -1, corresponding to a protein dimer ( M r -- 30800). Protein tetramers are induced in the presence of octanucleotides, but not tetranucleotides. Analysis of the ultraviolet spectra of the DNA binding protein and the native nucleoprotein complex indicates a stoichiometry of 3.9 + 0.4 nucleotides per protein subunit. The molar extinction coefficient of the DNA when bound to the protein (c26o = 8100) suggests that the binding protein maintains the DNA in an extended (unstacked) conformation similar to that found in the mature Pfl virion.

Introduction One of the major proteins synthesised after infection of P s e u d o m o n a s a e r u g i n o s a with the filamentous bacteriophage Pfl is a single-stranded DNA binding protein [1] analogous to the gene 5 protein of filamentous bacteriophage fd [2,3]. The protein can be extracted from infected cells as a helical complex with single-stranded viral DNA, having dimensions approx. 1 #m × 100 A, although the detailed structures of the fd and Pfl complex are different [4]. Unlike other singlestranded binding proteins, both the Pfl and fd proteins bring together two single DNA strands, regardless of sequence, into a single morphological assembly [4]. As the viral DNA is a topological closed circle, the two DNA strands must run the length of the complex in an antiparallel manner.

Genetic and biochemical experiments show that the fd gene 5 protein is required for the synthesis of viral DNA strands [5] apparently by binding stoichiometrically to single-strands, thereby inhibiting synthesis of the complementary strand [6]. The gene 5 protein seems to play a further role in packaging the viral DNA in a form suitable for assembly of the viron at the cell membrane [7] during which the gene 5 protein is released from the complex [8]. The Pfl DNA binding protein presumably plays a similar role in viral replication and assembly. X-ray diffraction studies of fibres of the native Pfl nucleoprotein complex [9] suggest that the protein subunits are arranged as dimers following a helical path of pitch 45 .A, with 6.0 dimers per turn. It was proposed that the subunits are oriented in a direction approximately perpendicular

0167-4781/83/0000-0000/$03.00 © 1983 ElsevierSciencePublishers

217

to the helix so that each subunit of the dimer binds to four nucleotides of opposing strands; the two DNA strands would be held together by proteinprotein interactions between the two subunits. A similar scheme was proposed for the fd nucleoprotein complex on the basis of hydrodynamic studies of the fd gene 5 protein [10] and the crystalline packing of subunits in the presence of oligonucleotides [11]. However, the Pfl DNA binding protein (M r = 15400) shows no obvious sequence homology with the fd gene 5 protein (M r = 9690) [12,13]. The fd gene 5 protein can be isolated in large quantities by DNA-cellulose chromatography [2] and has been studied by a variety of physical and chemical techniques (reviewed in Ref. 14). However, attempts in a number of laboratories to isolate the Pfl DNA binding protein by similar techniques have been unsuccessful (Anderson, E.A. and Marvin, D.A., personal communication; Day, L.A., personal communication). Although the protein can be obtained from the nucleoprotein complex by isopycnic centrifugation on CsC1 gradients [1], the amount of dissociation is variable. Undissociated material will band at a density close to that of the free protein, and thus the method is not suitable for preparation of pure protein for physicochemical studies. For an understanding of the structure and assembly of the Pfl nucleoprotein complex, it is important to establish the number of subunits in the free protein and its complex with oligonucleotides, the number of nucleotides bound per protein subunit, and the nature of the protein-nucleic acid interaction. In this paper, conditions for quantitative release of the protein are investigated and a large scale purification procedure is given. Characterisation of the binding protein and its interaction with oligonucleotides and viral DNA, using hydrodynamic and spectroscopic methods, is reported. Materials and Methods

Preparation of the Pfl nucleoprotein complex. Pfl infected cell extracts of Pseudomonas aeruginosa were prepared and depleted of free DNA as described earlier [1]. The complex was pelleted at 40 000 rev./min for 4 h in a Beckman 75 Ti rotor and resuspended overnight in 10 mM Tris-HC1/1

mM EDTA, pH 7.5 (Buffer A). The suspension was clarified by centrifugation at 20000 rev./min for 1 h, layered on 5-25% sucrose gradients and centrifuged at 27 000 rev./min for 4 h in a Beckman SW27 rotor. Fractions were taken and monitored by SDS-polyacrylamide gel electrophoresis. Fractions containing the complex were pooled and dialysed against Buffer A containing 0.2 M NaCI. The complex was precipitated with 10% poly(ethylene glycol) 6000, resuspended in Buffer A, and centrifuged once more on sucrose gradients. Fractions pure in the nucleoprotein complex, according to agarose [1] and SDS-polyacrylamide gel electrophoresis, were dialysed against Buffer A and concentrated by high speed centrifugation. Sedimentation analysis. Sedimentation experiments were performed at 4°C on 5-25% linear sucrose gradients at 40 000 rev./min in a Beckman SW 40 rotor. The pure DNA binding protein (100 /~g) alone, or in the presence of excess oligonucleotides (30 ttg), was layered onto each gradient and fractions (0.7 ml) were monitored for protein concentration using the Biorad protein assay. Fractions from parallel gradients containing marker proteins were assayed by gel electrophoresis as described in the legend to Fig. 8. The linearity of the sucrose gradients was checked by refractometry. Ultraviolet spectroscopy. Ultraviolet absorption spectra were recorded on a Cary 118 scanning spectrophotometer. The base line was adjusted to zero absorbance for wavelengths between 230 and 350 nm. Extinction coefficients were determined by amino acid analysis of the Pfl DNA binding protein and the Pfl nucleoprotein complex, both of known absorbance, assuming 18 mol alanine per mol protein [12]. Difference spectra were obtained by subtracting the protein spectrum from the nucleoprotein spectrum, at equimolar (subunit) concentrations in the same buffer. This was done under native conditions (10 mM Tris-HC1, pH 7.5) and in dissociation buffer (10 mM Tris-HCl/1 M MgC1z, pH 7.5). The molar extinction coefficient for Pfl viral DNA in dissociation buffer was estimated as c260 = 7440 (per nucleotide) by comparison with the previously published value (c260 -- 7700) in 0.15 M NaC1/0.015 M sodium citrate [15]. Fluorescence spectroscopy. Fluorescence emis-

218

Prof. J.H. van Boom, as described [17]. The oligonucleotides lack both 3'- and 5'-terminal phosphates. Extinction coefficients of ~260 = 7730 and c260 = 7860 were used for tetra- and octanucleotides, respectively.

sion spectra were recorded using a Perkin-Elmer MPF-44A spectrometer in the ratio mode. The excitation wavelength was 285 nm with a 5 nm bandpass for excitation and emission. Gel electrophoresis. Electrophoresis was performed on 15% polyacrylamide slab gels with a 5% stacking gel, in the presence of 0.2% SDS, as previously described [1 ]. Samples were diluted with an equal volume of electrophoresis buffer and heated to 95°C for 5 rain prior to loading. Gels were run at 60 mA for 3 h, and stained with Coomassie brilliant blue. Biorad low molecular weight protein standards were run in a parallel track, giving rise to six bands of 98, 66, 45, 30, 20 and 14 kDa. DNA and oligonucleotides. Single-stranded Pfl D N A was prepared from Pfl bacteriophage by phenol extraction as described earlier [16]. Poly(dT) was purchased from Miles Laboratories. The oligodeoxynucleotides d(C-G-C-A) and d(G-C-GT-T-G-C-G) were synthesised in the laboratory of

I

I

I

I

Results Dissociation of the nucleoprotein complex Fluorescence spectroscopy was used to monitor the dissociation. The emission spectrum of the Pfl nucleoprotein complex has a maximum at 330 nm arising from the single tryptophan in the protein (Trp 14) and is characteristic of a buried tryptophan. In high concentrations of MgC12 the spectrum is shifted towards higher wavelength (kmax = 340 nm), identical to that of the free protein (Fig. 1). The Rayleigh scattering peak at 285 nm is greatly reduced in 1 M MgC12, corresponding to the reduced molecular weight of the aggregate. A convenient measure of the spectral shift is the ratio of

I

0.~

4

bo.e

ii[ /

g •~ 0.4 E 0.9

o.ff/

o-f

-

-

02~

,/

U_ ,

290

I

I

310

330

Aem(nm)

I

I

I

3.50 370 390

0.2

0.t,

06

Concentration (N)

08

• 1 ~ *

0.2

I

04

•

I

0.6

•

I

0.8

f

Concentration (M)

Fig. 1. Fluorescence emission spectra of the Pfl nucleoprotein complex. The complex was dissolved to a final concentration of 0.2 m g / m l in the following solutions: (a) 10 mM Tris-HCl, pH 7.5; (b) 1 M MgCI2/10 mM Tris-HCl, pH 7.5; (c) 4 M urea. Spectra were recorded at 22°C with an excitation wavelength of 285 nm. Fig. 2. Effect of Mg 2÷ and Na ÷ on the fluorescence spectrum. The ratio of fluorescence intensity at 350 nm to that at 330 nm was measured at various concentrations of MgCI 2 (e) and NaCI (O). Pfl nucleoprotein complex at 0.2 m g / m l in 10 mM Tris-HCI, pH 7.5, was diluted with 3 M MgCI: or 4 M NaCI to the appropriate salt concentration. Fig. 3. The effect of Mg 2+ and Na + on the sedimentation of Pfl complex. The complex (1 m g / m l in 10 mM Tris-HCl, pH 7.5) was diluted to the appropriate concentration of MgCI 2 ( O ) or NaCI (e), then centrifuged for 2 h in the Beckman Airfuge rotor AI00 at 160000×g. Under these conditions both the complex and free DNA are pelleted. The fraction of protein remaining in the supernatant was determined by measurement of -4595, before and after centrifugation, using the Biorad protein assay.

219

the fluorescence signal at 350 nm to that at 330 nm. Fig. 2 shows that the apparent dissociation of the complex by MgC12 is almost complete at a concentration of 0.6 M. Much smaller fluorescence changes were observed with NaC1, suggesting little or no dissociation. Similar measurements as a function of urea concentration showed that the fluorescence is highly quenched and has a peak at approx. 350 nm in concentrations higher than 3 M, indicating complete exposure of the tryptophan, typical of a denatured protein (Fig. 1). No stable intermediate states were observed and it was concluded that in urea, unfolding of the protein occurs at the same time as dissociation from the DNA. In order to see if MgCI 2 could be used to quantitatively separate the viral DNA from the Pfl DNA binding protein, high speed centrifugation studies were performed. Fig. 3 shows that at concentrations approaching 1 M MgC12, quantitative separation is achieved. No appreciable release of the protein from the DNA was found in concentrations of up to 2 M NaC1. It was also found that 2 M LiC1 or 2 M NaSCN were effective in separating the protein from the DNA by centrifugation (results not shown).

DNA-cellulose chromatography From the above results it is clear why the standard conditions of DNA-cellulose chromatography have not been successful in the isolation of the Pfl DNA binding protein. As the protein is not released from viral DNA by the standard extraction buffer (containing 1 M NaC1) then there will be little free protein to bind to the column. Furthermore any free protein that does bind will not be released using the standard elution buffer [18], assuming the interaction with DNA-cellulose is the same as that in the in vivo complex; this was subsequently checked by running the pure protein down a DNA-cellulose column. Although some protein came through unretarded, almost none was eluted with 2 M NaC1; the majority was eluted by a 0-1 M MgCI 2 gradient (Fig. 4). Attempts to purify the DNA binding protein from cell lysates by DNA cellulose chromatography followed the method of Alberts and Herrick [18] except that 1 M MgC12 was used in place of 1 M NaC1, for both dissociation of the complex in

0.t~ 03

60

.~ 0.2

~o

20 ~

o.1 o.o

- o-x-x,-~-~ 5

. . . .

10 15 20 Froction number

25

Fig. 4. DNA-cellulose chromatography of the purified Pfl D N A binding protein. 0.5 g of DNA-cellulose was packed into a 4 × 0 . 7 cm column and washed with 50 ml of elution buffer (50 m M NaC1/20 m M T r i s - H C l / l m M E D T A / 1 m M 2mercaptoethanol, pH 8.0). 0.75 mg of pure Pfl D N A binding protein (1 ml) was applied to the column, followed by washes with (i) 3 ml of elution buffer, (ii) 3 ml of elution buffer containing 2 M NaCI, (iii) a further 2 ml of elution buffer, and finally (iv) 10 ml of a 0-1 M MgCI 2 gradient in the elution buffer. The flow rate was 2.5 m l / h . Protein concentrations (e) of each fraction (0.6 ml) were determined using the Biorad protein assay (.4595). The ionic strength (X) of the fractions was monitored with a conductivity meter.

cell lysates and for elution of the protein from the column (results not shown). Pure Pfl DNA binding protein could be obtained by this method but in relatively low yield. Because of this and the possibility of contamination of the protein by DNA in the presence of Mg 2÷ , an improved purification procedure was sought.

Purification of the DNA binding protein For routine purification an alternative procedure was devised which exploits the strong in vivo association between the protein and viral DNA. Partially pure nucleoprotein complex was prepared by high speed centrifugation from cell free extracts of infected Ps. aeruginosa, as described in Materials and Methods. Pellets from 50 g cells were resuspended in 150 ml 10 mM Tris-HC1/1 M MgC12, pH 7.5, to dissociate the complex, then centrifuged at 40 000 rev./min for 4 h in a Beckman 45 Ti rotor to remove the released viral DNA. The supernatant was dialysed against two changes of 10 mM Tris-HCl, pH 8.0, and the protein precipitated by the addition of an equal volume of saturated ammonium sulphate. The sample at this stage was approx. 50% pure in the DNA binding

220

]:

3

4

: ~

B

N

80 1.5

7.0 1.o

6.0 0.5

7r-t

D

i~~O..o.o..o.o..o 5.0 10

20

30

t,0

50

Fraction number Fig. 5. Purification of the Pfl DNA binding protein on a chromatofocusing column. The protein extract was dialysed against 25 mM imidazole-HCl buffer, pH 7.5, and applied to a column of Pharmacia PBE 74 (30 × 1 cm). The pH gradient was developed by elution with 300 ml of Polybuffer 74 (pH 4.0) at a flow rate of 36 m l / h . Fractions (6 ml) were monitored by absorbance at 280 nm (e) and by SDS-polyacrylamide gel electrophoresis. Peaks l and 2 both contained a single protein band of M r = 15 400. The pH of each fraction was also measured (O). Fig. 6. SDS-polyacrylarnide gel electrophoresis at various stages during the purification procedure. Samples were taken at the following stages: (1) cell-free extract; (2) pellet after high speed centrifugation; (3) ammonium sulphate precipitate of the supernatant after centrifugation in 1 M MgC12; (4) pooled fractions 22-24 and (5) fractions 27-30 after chromatofocusing (see Fig. 5). Electrophoresis was performed on 15% polyacrylamide slab gels containing 0.2% SDS. The positions of the molecular weight marker bands are also shown (6).

protein (Fig. 6, track 3). The precipitate was redissolved in 5 ml of 25 mM imidazole-HCl buffer, p H 7.5, and after dialysis against this buffer applied to a Pharmacia chromatofocusing column, as described in the legend to Fig. 5. This resulted in two pure protein peaks, one eluting at pH 5.9 and a minor peak at pH 6.4 (Fig. 5). Both peaks were identical electrophoretically to the Pfl DNA binding protein (Fig. 6, tracks 4 and 5). It is possible that the major peak arises from deamidation of Gln 34, as suggested by a comparison of the amino acid sequence and the corresponding nucleotide sequence [12]. The two peaks containing pure protein were separately pooled. Both peaks showed the same absorption and fluorescence spectra; only the major peak ( p I = 5.9) was used for further experiments. Polybuffer was removed by ammonium sulphate precipitation of the protein, which was then dialysed against the required buffer

and finally centrifuged at 10000 rev./min for 30 min to remove any insoluble material. Typical yields of 15-25 mg of pure protein per 50 g of bacteria were obtained following this procedure.

Physicochemical characterisation From the amino acid sequence the subunit molecular weight of the DNA binding protein is 15400 [12]. To obtain information on the shape and size of the native protein, gel filtration was carried out on Ultragel ACA 54 in 0.1 M Tris-HCl, pH 8.0. The majority of the protein is eluted at ~ / ~ = 0.48, with a much smaller peak at ~ / I / t = 0.37 (Fig. 7). For proteins of unknown shape the elution volume is not related simply to M r, but can be correlated with diffusion coefficient D [19,20]. For the major peak one obtains D = 7.8 • 10 - 7 c m 2 . s - 1 and for the minor peak D = 6 . 0 10 - 7 c m 2" s - 1 . From the diffusion coefficients the

221

1.02.2s

!

0.8t.d

r-r~ t..

0.2

32s/~

06-

o t/I

0t~-

,{¢ 0.1

0.2-

.......

0.2

I\

0t~ 0.6 Me/Mr

,

0.8

,

!i;i/x_

0.0 5 10 Frocfion number

15

Fig. 7. Gel filtration of the Pfl D N A binding protein on Ultragel A C A 54. 8 mg of the pure protein was applied to a 100x2.5 cm column of A C A 54 and eluted with 0.1 M Tris-HCl, pH 7.5, at 50 m l / h . Fractions (5 ml) were collected and their absorbance at 280 n m was measured. The void volume, ~ (160 ml) and total volume, I/t (530 ml) were measured by calibration with Dextran blue and dinitrophenyl alanine. The marker proteins used were ribonuclease A (D = 11.9.10-7 cm 2. s - I ) , chymotrypsinogen (D = 9.5.10-7 cm 2- S-l), ovalbumin (D = 7.8-10-7 cm 2. S l) and bovine serum albumin (D = 5.9.10-7 cm 2. s - i ) . Fig. 8. Sedimentation of the Pfl D N A binding protein on sucrose gradients. Sedimentation was performed at 4°C on 5-25% sucrose gradients in 0.1 M Tris-HC1, p H 7.5. The gradients were centrifuged for 40 h at 40000 r e v . / m i n in an SW40 rotor. Fractions (0.7 ml) were taken from the bottom and assayed for protein by the Biorad protein assay (A595). Three separate gradients were run containing 200/~g of the D N A binding protein alone (O), and in the presence of 30/Lg of the tetranucleotide C-G-C-A ( O ) or 3 0 / t g of the octanucleotide G-C-G-T-T-G-C-G (A). In a parallel experiment the gradients were calibrated by mixing the D N A binding protein with the following marker proteins: ribonuclease A (s = 1.85), cytochrome c (s = 1.9), myoglobin (s = 2.04), chymotrypsinogen (s = 2.54) and bovine serum albumin (s = 4.31). The positions of marker proteins were determined by SDS-polyacrylamide gel electrophoresis of the fractions. The presence of these proteins had no effect on the sedimentation of the D N A binding protein.

Stokes radii of the major and minor peaks are calculated as 2.8 and 3.6 nm, respectively. Sedimentation of the native protein on sucrose gradients shows a h o m o g e n o u s peak at 2.3 S (Fig. 8). If one assumes that this b a n d corresponds to the major peak on gel filtration ( D = 7.8 • 10 - 7 c m 2. s - 1), substitution into the Svedberg equation M r = R T S / D ( 1 - ~ p ) allows the native molecular weight to be determined. Taking T = 293 K and p = 1.00 g. c m - 3 , and assuming a partial specific volume ~ = 0.76 [9] one obtains M r = 29 900, close to the value expected for a protein dimer. C o m b i n ing the Stokes radius, a, with the molecular weight allows the frictional ratio f / f o to be determined, from the equation f i f o = a / ( 3 ~ M r / 4 q r N ) t/3. F o r

a protein dimer of M r = 30 800 and a = 2.8 n m a frictional ratio of 1.31 is obtained, which is a fairly typical value for globular proteins. If the minor c o m p o n e n t with a = 3.6 n m is a protein tetramer ( M r = 61 600), this would have a frictional ratio of 1.34 and a sedimentation coefficient s = 3.6. Addition of tetranucleotides has no effect on the sedimentation of the D N A binding protein, but in the presence of octanucleotides the peak broadens and shifts to s = 3.2 (Fig. 8). The increased breadth of the peak and the slightly slower sedimentation than expected for a tetramer suggest that some of the protein remains as dimers, which are not resolved on the gradient.

222

Ultraviolet spectroscopy The ultraviolet absorption spectrum of the Pfl D N A binding protein has a maximum at 278 nm, a shoulder a 283 nm and a minimum at 251 nm (Fig. 9). The absorption spectrum of the Pfl nucleoprotein complex has a broad maximum at 260-270 nm with a minimum at 246 nm (Fig. 10). The concentrations of the protein and nucleoprotein complex were determined by amino acid analysis, assuming 18 alanines per protein subunit, which allows their molar extinction coefficients to be determined. Thus, for the D N A binding protein ~278 = 11 800 and for the nucleoprotein complex cz60 = 38200 (per mol of protein subunits). The absorbance .4278 ~ of a 1% solution of the Pfl D N A binding protein is therefore 7.7. If there are four nucleotides per protein subunit then the molecular weight of the nucleoprotein unit is 16 600 [9]. Thus for the nucleoprotein complex, ,4260 = 23. The ultraviolet spectrum of a nucleoprotein complex can be used to determine the stoichiometry of its components [21]. For this it is necessary

06t

'

,

06I

Discussion

041

~ o4

o2

oo

3+0 3 0"

WaveIength(nm)

to dissociate the complex to remove any hyperchromic effects so that the resulting spectrum can be analysed as the sum of the spectra of the isolated components. For the Pfl nucleoprotein complex it is convenient to use 1 M MgC12 for dissociation. When the protein spectrum is subtracted from the spectrum of the nucleoprotein (both in 1 M MgCl 2, with an equal concentration of protein subunits) the resulting difference spectrum has a maximum at 260 nm (Fig. 10) as for DNA. Knowing the molar concentration of subunits in the sample and the extinction coefficient of Pfl D N A in 1 M MgC12 (c260 = 7440) a nucleotide/protein ratio of 3.9 _+ 0.4 is obtained. The uncertainty in this parameter is primarily due to an estimated uncertainty of approx. 5% in the extinction coefficients of protein and complex. The nucleotide/protein ratio indicates that there is 8% ( w / w ) D N A in the nucleoprotein complex. When the experiment is repeated under non-dissociating conditions (10 m M Tris-HC1, p H 7.5) the difference spectrum shows a 9% increase in A260 (Fig. 10). Thus for the D N A in the nucleoprotein complex ~260 = 8100.

260 280 300 320

Wavelength(nrn)

Fig. 9. Ultraviolet absorption spectrum of the Pfl D N A binding protein. The spectrum of the protein at 0.6 m g / m l in 10 m M Tris-HCl, pH 7.5, was measured in a 1 cm path length cuvette. Fig. 10. Ultraviolet absorption spectrum of the Pfl nucleoprotein complex. Upper curve: spectrum of the native complex in 10 m M Tris-HCl, pH 7.5. Middle curve: difference spectrum after subtracting the protein spectrum from the native complex spectrum, both in 10 m M Tris-HC1, pH 7.5. Lower curve: difference spectrum after subtracting the protein spectrum from the dissociated complex spectrum, both in 1 M M g C I 2 / 1 0 m M Tris-HC1, pH 7.5. All samples were at a protein concentration of 0.23 m g / m l in a 1 cm path length cuvette at 22°C.

Although the fd gene 5 protein is released from D N A in 0.4 M NaCI [8], there is little or no release of the Pfl D N A binding protein even in 2 M NaC1. From the results of fluorescence spectroscopy and ultracentrifugation it is clear that high concentrations (0.6-1.0 M) of Mg 2+ can be used to dissociate the protein and D N A components of the Pfl complex. However divalent cations are not an absolute requirement for dissociation, as both LiC1 and N a S C N disrupt the complex - - although higher concentrations are required. These salts are chaotropic, which suggests that there are substantial hydrophobic, as well as electrostatic, interactions involved. The greater efficacy of Mg z+ compared to N a + in dissociation is not unusual; MgC12 is 5-10-times more effective than NaC1 in dissociating complexes of D N A with the single-stranded binding proteins of bacteriophage T4 (the gene 32 protein) and of E. coli [22]. However, for these two proteins dissociation is complete at 0.2 M MgC12. This effect of l~lg 2÷ on dissociation applies not

223 only to the native complex, as high concentrations of Mg 2+ are also required to elute the Pfl DNA binding protein from DNA-cellulose columns. The changes observed in the fluorescence emission spectrum of the Pfl nucleoprotein complex as a function of Mg 2+ concentration show that fluorescence may be used to monitor dissociation. The tryptophan environment becomes more polar on dissociation, as judged by the shift of the emission peak to 340 nm, although the total intensity remains fairly constant. As the tryptophan fluorescence is highly quenched on exposure to 4 M urea, the quantum yield is clearly high in both the nucleoprotein complex and the free protein. A more detailed study using time resolved fluorescence techniques will be reported elsewhere. From measurements of the total mass of the Pfl nucleoprotein complex a stoichiometry of 4.2 + 0.5 nucleotides per protein subunit has been reported [9] on the assumption that the complex contains a single molecule of viral DNA. That assumption is confirmed by the nucleotide/protein ratio of 3.9 _ 0.4 determined by a direct spectroscopic method, assuming only the individual extinction coefficients of the protein and DNA. It has been estimated that the number of nucleotides per subunit of fd gene 5 protein is approximately five in the native fd complex [8,23], although for in vitro reconstitutes this number is closer to four [2,24]. The p r o t e i n / D N A ratio (w/w) is therefore about twice as high for Pfl complex (13:1) as for fd complex (6.5: 1). Although it may be fortuitous, these numbers are very close to those one calculates for the Pfl virion (15 : l) and the fd virion (7.5: 1) respectively. The hyperchromism of the DNA in the Pfl nucleoprotein complex is attributed to unstacking of the bases of the single-stranded Pfl DNA. The extinction coefficient per nucleotide, ~260= 8100, is typical of unstacked DNA and is close to that of the DNA in Pfl virions (~260 = 8000). For the latter there is evidence from fluorescence that the D N A bases are stacked with tyrosine residues of the coat protein [25]. Thus in the transition from the Pfl nucleoprotein complex to the assembled Pfl virion, it seems that the D N A is maintained in an extended form by intercalation of aromatic amino acid side chains. This is in marked contrast

to the assembly of fd virions. The DNA in the fd nucleoprotein complex is also hyperchromic (~260 = 8200) and its bases are stacked with tyrosine and phenylalanine residues [25-27]; however, the D N A in the fd virion is hypochromic (E260 = 6700) and evidence from X-ray fibre diffraction suggests that there are no specific interactions between viral DNA and the fd coat protein subunits [25,28]. Assembly of the virion is coupled with dissociation of the DNA binding protein from the viral DNA. The high ionic strength required for in vitro dissociation is not encountered in vivo, where more specific interactions, involving displacement by the major coat protein, will be involved. The increased stability of the Pfl nucleoprotein complex in high salt, compared to the fd complex, is thus consistent with a stronger specific interaction between the coat protein and DNA in the Pfl virion. The hydrodynamic properties of the Pfl DNA binding protein support the arrangement of subunits proposed on the basis of X-ray fibre diffraction of the nucleoprotein complex [9]. In free solution the Pfl protein exists as a dimer consisting of two symmetry related subunits, each of which can bind a tetranucleotide. Binding of the octanucleotide bridges adjacent binding sites of two protein dimers to form a tetramer; binding of DNA induces linear aggregates of dimers which, in the native complex, are arranged as a helix with six dimers per turn. Similar results have also been found for the fd gene 5 protein [10,11], providing further evidence that the two proteins perform the same role in replication and packaging of the viral DNA.

Acknowledgements I thank Dr. D.A. Marvin for his support in this work and Dr. K.O. Greulich and Prof. A. Tsugita for discussion. Skilled technical assistance was provided by K. Nilsson. Oligonucleotides were kindly provided by Prof. J.H. van Boom.

References 1 Kneale,G.G. and Marvin, D.A. (1982) Virology 116, 53-60 2 Alberts, B.M., Frey, L. and Delius, H. (1972) J. Mol. Biol. 68, 139-152 30ey, J.L. and Knippers, R. (1972)J. Mol. Biol.68, 125-138

224 4 Gray, C.W., Kneale, G.G., Leonard, K.R., Siegrist, H. and Marvin, D.A. (1982) Virology 116, 40-52 5 Henry, T.J. and Pratt, D. (1969) Proc. Natl. Acad. Sci. U.S.A. 62, 800-807 6 Salstrom, J.S. and Pratt, D. (1971) J. Mol. Biol. 61,489-501 7 Marvin, D.A. and Wachtel, E.J. (1976) Philos. Trans. R. Soc. London, Ser. B276, 81-98 8 Pratt, D., Laws, P. and Griffiths, J. (1974) J. Mol. Biol. 82, 425-539 9 Kneale, G.G., Freeman, R and Marvin, D.A. (1982) J. Mol. Biol. 156, 279-292 10 Cavalieri, S.J., Neet, K.E. and Goldthwait, D.A. (1976) J. Mol. Biol. 102, 697-711 11 McPherson, A., Jurnak, F., Wang, A., Kolpak, F., Rich, A., Molineux, I. and Fitzgerald, P. (1980)Biophys. J. 32, 155-173 12 Maeda, K., Kneale, G.G., Tsugita, A., Short, N.J., Perham, R.N., Hill, D.F. and Petersen, G.B. (1982) Eur. Mol. Biol. Org. J. 1,255-261 13 Nakashima, Y., Dunker, A.K., Marvin, D.A. and Konigsberg, W. (1974) FEBS Lett. 40, 290-292 and errata 43, 125 14 Coleman, J.E. and Oakley, J.L. (1980) Crit. Rev. Biochem. 7, 247-289 15 Wiseman, R.L and Day, L.A. (1977) J. Mol. Biol. 116, 607-611 16 Wiseman, R.L., Berkowitz, S.A. and Day, L.A. (1976) J. Mol. Biol. 102, 549-561

17 Van der Marel, G.A., Van Boeckel, C.A.A., Wilie, G. and Van Boom, J.H. (1982) Nucleic Acids Res. 10, 2337-2342 18 Alberts, B.M. and Herrick, G. (1970) Methods Enzymol. 21, 198-217 19 Andrews, P. (1965) Biochem. J. 96, 595-605 20 Van der Vliet, P.C., Keegstra, W. and Jansz, H.S. (1978) Eur. J. Biochem. 86, 389-398 21 Berkowitz, S.A. and Day, L.A. (1976) J. Mol. Biol. 102, 531-547 22 Anderson, R.A. and Coleman, J.E. (1975) Biochemistry 14, 5485-5491 23 Pretorius, H.T., Klein, M. and Day, L.A. (1975) J. Biol. Chem. 250, 9262-9269 24 Day, L.A. (1973) Biochemistry 12, 5329-5339 25 Day, L.A., Wiseman, R.L. and Marzec, C.J. (1979) Nucleic, Acids Res. 6, 1393-1403 26 Coleman, J.E. and Armitage, I.M. (1978) Biochemistry 17, 5038-5045 27 Hilbers, C.W., Garssen, G.J., Kaptein, R., Schoenmakers, J.G.G. and Van Boom, J.H. (1978) in Nuclear Magnetic Resonance Spectroscopy in Molecular Biology (Pullman, B., ed.), pp. 351-364, Reidel, Dordrecht, The Netherlands 28 Banner, D.W., Nave, C. and Marvin, D.A. (1981) Nature 289, 814-816