Proc. Nati. Acad. Sci. USA
Vol. 73, No. 6, pp. 1917-1920, June 1976
Biochemistry
Cyanobacterial DNA-binding protein related to Escherichia coli HU (low-molecular-weight protein/histone-like protein/lysine-rich protein/prokaryotic DNA-binding protein)
ROBERT HASELKORN*t AND JOSETTE ROUVIERE-YANIVt * Service de Physiologie
Microbienne and t D~partement de Biologie Moleculaire, Institut Pasteur, Paris, 75015
Communicated by Donald F. Steiner, March 25, 1976 ABSTRACT
A DNA-binding protein has been isolated from
(blue-green algae): Anabaena sp and Aphanocapsa sp. It has a molecular weight as determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis of about 10,000, is rich in lysine and arginine, and lacks tyrosine, tryptophan, and cysteine. The proteins from both strains show immunological identity with a similar DNA-binding protein from Escherichia coli, when tested by immunodiffusion with an antiserum prepared against the E. coli protein. Both the Aphanocapsa and E. coli proteins form compact, rapidly sedimenting complexes with cyanophage or bacteriophage DNA. The similarities between the proteins from cyanobacteria and E. coli suggest a degree of evolutionary conservation comparable to that of the histones of eukaryotes. two cyanobacteria
Models for the regulation of transcription in eukaryotes have to take into account the structure of DNA in chromosomes. That structure is determined to a considerable degree by histones and
nonhistone proteins in chromatin. Prokaryotic transcription and replication have only recently been studied from the point of view of the structural considerations imposed by the arrangement of the DNA in the bacterial nucleoid (1-3). The histones have been strongly conserved during the evolution of eukaryotes. For example, the amino acid sequences of histones H3 and H4 from a wide variety of sources are virtually identical (4). Histones H2a and H2b from man, pig, chicken, and lobster are sufficiently related antigenically to be indistinguishable by the microcomplement fixation test (5). With the recent discovery in Escherichta coli of a low-molecular-weight DNA-binding protein having a histone-like amino acid composition (6), it became of interest to seek a similar protein in other prokaryotes. We report here the characterization of such a protein, which we cQll HU, from two representative cyanobacteria (blue-green algae). MATERIALS AND METHODS Biological Material. Protein HU was prepared from two strains from the collection of the Service de Physiologie Microbienne, Institut Pasteur. Strain 7120 is an Anabaena species which is filamentous, forms heterocysts, and fixes nitrogen. Strain 6701, from which the protein most extensively characterized was obtained, is an unicellular organism, Aphanocapsa. Both strains were grown under axenic conditions in 10 liter carboys of medium GN (7) modified by a 10-fold increase in the concentration of Na2CO3 and gassed with 1% CO2 in air. Carboys were illuminated by banks of Osram-L-interna fluorescent lamps which provided an intensity of 5000 lux (Im/m2) at the surface of the carboy. Purification of DNA-Binding Proteins. The procedure was essentially the one described for E. coli HU protein (6). The frozen paste of strain 6701 cells (20 g) was diluted with 60 ml of the following buffer: 20 mM Tris-HCI at pH 8.1, 2 mM Permanent address: Department of Biophysics and Theoretical Biology, University of Chicago, Chicago, Ill. 60637.
1917
CaCI2, 10 mM MgCI2, 1 mM EDTA, 1 mM mercaptoethanol containing 20 Atg/ml of DNase. The diluted paste was passed through a French pressure cell once at 20,000 pounds/inch2 (140 MPa), incubated for 15 min at room temperature, and then centrifuged successively for 10 min at 8000 rpm and 180 min at 40,000 rpm. The final supernatant was dialyzed against 20 mM Tris-HCI at pH 8.1,5 mM EDTA, 50 mM NaCl, 0.5 mM mercaptoethanol, made 10% (wt/vol) in glycerol, and centrifuged again for 10 min at 8000 rpm. The supernatant, about 70 ml, was applied to a DNA-cellulose column prepared as described previously (6), washed extensively with 20 mM Tris-HCI at pH 8.1,5 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol, 50 mM NaCl, and then developed with successive steps of this buffer containing 0.15 M, 0.4 M, 0.6 M, and 2.0 M NaCl. Material absorbing at 280 nm was obtained in the first fractions following each change of salt concentration. These fractions were analyzed by sodium dodecyl sulfate gel electrophoresis on a 15% polyacrylamide gel as described in ref. 6. The material eluting at 0.4 M salt displayed at least eight bands on the gels, but over 80% of the Coomassie-blue-staining material was contained in one band corresponding to a molecular weight of about 10,000. Fractions containing this band were pooled, concentrated, and dialyzed against 20 mM TrisHCI at pH 8.1, 1 mM EDTA, 0.3 M NaCl, 10% glycerol. The pooled fractions were then passed over a column of Sephadex G-100 equilibrated and developed with the same buffer. The column (1.5 X 15 cm) was calibrated with chymotrypsinogen A, myoglobin, and cytochrome c; for each run Blue Dextran 2000 and tryptophan were added to indicate the void and included volume, respectively. The pooled fractions from the DNA-cellulose column fractionated on the G-100 column as follows. The chlorophyll a, together with essentially all other material absorbing at 280 nm, emerged with the void volume. A second small peak of material, detected only by absorption at 240 nm, emerged at a position corresponding to that of chymotrypsinogen A. The voided fractions contained no protein detectable by polyacrylamide gel electrophoresis. The fractions absorbing at 240 nm contained material which yielded a single band on sodium dodecyl sulfate polyacrylamide gel electrophoresis corresponding to a molecular weight of about 10,000. These fractions were pooled and comprised the material used for determination of amino acid composition and for DNA binding experiments. Immunodiffusion. Rabbits were immunized with pure E. coli HU protein (6), as described by Guiso and Truffa-Bacchi (8). The y-globulins were concentrated 3-fold by ammonium sulfate precipitation. Immunodiffusion (Fig. 1C) and immunoelectrophoresis (J. Rouviere-Yaniv, unpublished work) revealed a single precipitin band between the antiserum and E. coli HU protein. Amino Acid Composition. Protein samples were Iyophilized, dissolved in 6 M HCI, sealed under vacuum, and hydrolyzed at 1100 for 24 and 48 hr. Samples containing 400 nmol of total
1918
Biochemistry: Haselkorn and Rouviere-Yaniv
Proc. Nati. Acad. Sci. USA 73 (1976) Table 1. Amino acid composition of HU proteins
B
Mole %
c
al.
I
Al
FIG. 1. Immunodiffusion of HU proteins. The central well contained anti-E. coli HU 'y-globulin in each case. (A) Peak fractions of Anabaena protein eluted from DNA-cellulose with 0.4 M NaCl (wells a, b, and c). (B) Peak fractions of Aphanocapsa protein eluted similarly (a-f). (C) ec: purified E. coli HU, 20 ug/ml; al: Aphanocapsa HU purified through Sephadex G-100 (see Fig. 2, sample C), approximately 150 ,g/ml. The last two unlabeled wells contained coliphage T4 gene 32 protein at 1 mg/ml or 20 tg/ml.
amino acids were analyzed by G. LeBras and M. Gouyvenaux Technicon Analyzer. Tryptophan was shown to be absent by the lack of absorption in the ultraviolet (see above). Cysteine was determined as cysteic acid on a sample oxidized with performic acid prior to hydrolysis; none was found. Ultracentrifugation. The sedimentation coefficients of DNA from coliphage X and cyanophage N-1 (9), as well as those of complexes of the DNA with protein, were determined on an MSE ultrascan analytical ultracentrifuge using cells with an optical path of 1 cm and solutions having an absorbance at 260 nm of 0.4. Complexes were prepared by mixing DNA and protein in 0.01 M sodium phosphate at pH 7.1, 0.1 M KC1. Lambda DNA was prepared as described previously (6). N-1 was grown on strain 7120 and purified, and the DNA was prepared by phenol extraction, as described previously (9).
on a
RESULTS
Cyanobacterial proteins serologically related to E. coli HU
The fractions isolated from Anabaena (7120) extracts by DNA-cellulose chromatography were tested for cross-reaction with an antiserum prepared against E. coli HU using Ouchterlony immunodiffusion tests in agar. The column fractions eluted at 0.4 M NaCl contained material yielding precipitin bands (Fig. 1A), whereas those eluted at 0.15 M, 0.6 M, and 2 M NaCl did not. The fractions containing cross-reacting antigen do not absorb ultraviolet light at 280 nm, but do show a very weak phenylalanine-like absorption around 260 nm. They also contain a single major protein revealed by polyacrylamide gel electrophoresis, having a mobility greater than cytochrome c.
Amino acid
Aphanocapsa
E. coli (6)
Lysine Histidine Arginine Aspartate + asparagine Threonine Serine Glutamate + glutamine Proline Glycine Alanine Valine Isoleucine Leucine Tyrosine Phenylalanine Methionine Cysteine Tryptophan
14.0 1.5 5.4
14.0 1.5 5.1
8.0 6.0 8.1
8.1 6.0 4.4
8.1 4.5 7.5 11.5 10.0 3.3 3.4 0 4.0 2.5 0 0
9.6 3.0 7.4 16.3 6.0 6.0 6.6 0 3.0 1.5 0 0
These determinations exhausted the supply of protein from Anabaena.
The larger preparation of DNA-binding protein from Aphanocapsa (6701) was characterized more extensively. Fig. 1 shows the serological cross-reaction with the fractions eluted by 0.4 M NaCl from the DNA-cellulose column (B) and with the fractions corresponding to the peak of absorbance at 240 nm from the G-100 column (C). No cross-reacting material was found in the other fractions from the DNA-cellulose column and none was in the void volume from the G-100 column, which contains all of the material absorbing at 280 nm. The absence of spurs between the precipitin bands formed with the E. coli and cyanobacterial proteins (Fig. IC) demonstrates that the two proteins are closely related. Amino acid composition The ultraviolet absorption spectrum of the material eluted from the G-100 column is absolutely flat above 270 nm, indicating the absence of tryptophan and tyrosine. No cysteic acid was found in a performic acid oxidized sample after hydrolysis and ion-exchange chromatography, indicating the absence of cysteine. The relative molar concentrations of the remaining amino acids are listed in Table 1, together with the corresponding figures for the E. coli HU. Both proteins are strikingly similar with respect to lysine, arginine, and histidine content, and neither contains cysteine, tryptophan, or tyrosine. The two proteins have similar ratios of basic to acidic amino acids. The decrease in alanine content of the cyanobacterial protein is balanced by an increase in serine content. The percentages of hydrophobic amino acids are also similar, although the cyanobacterial protein contains more valine than isoleucine and leucine. Molecular weight The cyanobacterial protein elutes from the G-100 column at a position identical to that of chymotrypsinogen A. Thus its maximum molecular weight is 25,000, since all the calibrating proteins are effectively spherical. On polyacrylamide gels containing sodium dodecyl sulfate the protein moves faster than
Proc. Nati. Acad. Sci. USA 73 (1976)
Biochemistry: Haselkorn and Rouvie're-Yaniv
1919
Table 2. Sedimentation coefficients of viral DNA-HU complexes Sedimentation coefficient, 200 DNA source
N-1 X
DNA
DNA + cyanobacterial HU
DNA + E. coli HU
32 28
44 48
44 73
The solvent in all cases was 0.1 M KCl, 0.01 M NaPO4, pH 7.1. DNA and protein concentrations were each 20 1Ag/ml.
algal
___
_
HU
HU~ A
~ B
C
i
_ ~
E.
coli
B
FIG. 2. Polyacrylamide gel electrophoresis of Aphanocapsa DNA-binding protein. (A) Shows the peak fraction eluted from the DNA-cellulose column with 0.4 M salt; (B) is E. coli HU purified through the phosphocellulose step (6); (C) contains material from three pooled fractions from the G-100 column corresponding to the peak of absorbancy at 240 nm (see Materials and Methods).
cytochrome c but slightly slower than E. coli HU (Fig. 2), and thus has a molecular weight as determined by this method of 10,000. A small amount of the protein was reacted with 2 mg/nl of dimethyl suberimidate in 0.2 M triethanolamine, pH 8.5, for 2 hr at room temperature (10). More than half of the product migrated on a sodium dodecyl sulfate gel slightly more slowly than unreacted material, while the remainder, perhaps 2('0%, had a molecular weight, by this method, of 25,000. We interpret these results to mean that the cyanobacterial protein exists, in dilute solution, as a dimer. When reacted under the same conditions, E. coli HU gave the dimer as a major product and a trace of tetramer J. Rouviere-Yaniv, unpublished results). Sedimentation of DNA-protein complexes The protein was originally isolated by binding and elution from a column containing heterologous (calf thymus) DNA (6). We have studied the interaction of the protein in solution with DNA preparations from a bacteriophage (coliphage X) and a cyanophage (N-1). These were mixed with roughly equal weights of protein and analyzed by boundary centrifugation. The results, summarized in Table 2, are indicative of substantial interaction. The sedimentation profiles for A DNA with the two proteins are shown in Fig. 3. All the DNA molecules, under these conditions which correspond to protein excess, bind roughly comparable amounts of protein.
has a similar amino acid composition, and likewise forms rapidly sedimenting complexes with DNA from bacterial viruses. Approximately 800 ,ug of protein HU were isolated after 0-100 chromatography from 20 g (wet weight) of strain 6701, corresponding to 7.5 X 1011 cells. Therefore the minimum number of HU monomers per cell is 60,000. In view of likely losses during preparation of the crude extract the real number could be substantially higher. Cells of 6701 contain an average of two chromosomes, each with kinetic complexity corresponding to a molecular weight of 2.7 X IOP (M. Herdman, personal communication) or approximately 107 base pairs per cell. The minimum ratio of HU to DNA is thus one dimer of HU for every 300 base pairs. The studies of HU binding by electron microscopy (6) reveal that HU binds to all parts of the DNA molecules, but not uniformly. The images observed suggest a compaction of the DNA structure. These observations are confirmed by ultracentrifugation studies: the sedimentation boundaries formed by HUDNA complexes are not much broader than those of free DNA, and the DNA-HU complexes are much more compact than free DNA. The addition of protein to DNA results in a complex of increased partial specific volume, the increase being proportional to the amount of protein. For a complex with equal weight of protein and DNA, the partial specific volume is increased about 25% compared to DNA alone. For homologous DNA molecules the sedimentation coefficient, s, is proportional to Mro 35 where Mr is molecular weight; doubling the molecular weight increases s by about 30%. Since this increase is nearly compensated by the change in partial specific volume, one expects the sedimentation coefficient of HU-DNA complexes 0.2
A
0.1
E C
~.0.21 LO
(N4 11
0.1
.0 .0
0.2' 0.1
.0 0.1
-W
DISCUSSION of DNA-binding proteins from the isolation We have described two different cyanobacteria, each representative of a major sub-class of these organisns. In spite of the evolutionary distance between E. colf and cyanobacteria, the cyanobacterial DNAbiding
protein is serologically related to its E. col counterpart,
Oj
VV
FIG. 3. Ultracentrifugation of X DNA and its complexes with HU. Each panel shows successive scans at 4 min intervals after the rotor reached a speed of 35,000 rpm. (A) X DNA, 20 gg/ml in 0.1 M KCI, 0.01 M NaPO4; (B) X DNA, 20 jg/ml plus Aphanocapsa HU, 20 sg/mL (C) X DNA, 20 jg/ml plus E. coli HU, 20 sg/ml. The direction of sedimentation is from left to right.
1920
Proc. Nad. Acad. Sci. USA 73 (1976)
Biochemistry: Haselkorn and Rouvibre-Yaniv
to be nearly the same as that of the DNA alone. In fact, s increases greatly (Table 2), so the complexes must be relatively compact. In addition to HU, several other DNA-binding proteins of low molecular weight have been described from E. coli (11-14). Like HU, they stimulate transcription; however, this stimulation appears to be specific, whereas the stimulation by HU is not (6). Therefore their function in the cell could be different from that of HU. By contrast, a few DNA-binding proteins, possibly related to HU, have been isolated from other bacteria and characterized. Geiduschek and coworkers (15) described a protein, TFI, which is synthesized during phage SP01 infection of Ba-
cillus subtilis. TFI is a dimer with monomer molecular weight about 11,500. Its amino acid composition resembles that of HU, but it has a lower ratio of basic to acidic amino acids and a different lysine:arginine ratio. There are at least 105 molecules per infected cell (16). Its function in SPOI infection is unknown, but it has been shown to inhibit the transcription of SPO1 DNA in vitro. On the other hand, E. coli HU stimulates the transcription of X DNA in vitro (6). Another histone-like protein was recently isolated from Thermoplasma acidophilus (17), a member of the mycoplasma group that grows at a very low pH and at high temperature, and contains DNA of unusually high AT content. This protein is strongly associated with DNA and is soluble in acid. Its amino acid composition is similar to that of HU, but its molecular weight is higher. The comparison of HU protein and primitive eukaryote chromosomal proteins may throw more light on the evolutionary relation between prokaryotes and eukaryotes. For example, it will be interesting to know the extent of relatedness between HU and the acid-soluble low molecular weight protein isolated from dinoflagellate chromatin by Rizzo and Nooden (18). The cellular function of HU still remains conjectural, although it seems probable that it plays a role in determining the structure of the prokaryotic chromosome. In such a structure, HU protein could prevent the RNA polymerase from binding to nonproductive sites on DNA, as is suggested by the stimulatory effect of HU on X DNA transcription that is observed only when DNA is in excess (6). Finally, it will be of interest to determine whether HU is responsible for maintaining the chro-
matin-like appearance of the E. coil chromosome observed by Griffith (19). We thank G. Cohen-Bazire and F. Schaeffer for the gift of cyanobacterial cells. We are indebted to M. Goldberg, N. Leveque, 0. Croissant, P. Truffa-Bacchi, and M. Yaniv for advice and help during the course of this work. R.H. is grateful to Roger Stanier and the members of the Service de Physiologie Microbienne for their hospitality and to the John Simon Guggenheim Foundation for a fellowship enabling him to work in Paris. J.R:-Y. is grateful to F. Gros for his interest in this work and for providing laboratory facilities. The work was supported by grants from the Centre National de la Recherche Scientifique. 1. Rouviere, J., Lederberg, S., Granboulan, P. & Gros, F. (1969) J.
Mol. Biol. 46,413-430.
Pettijohn, D. W. (1971) Proc. Nati. Acad. Sci. USA 68,609. Worcel, A. & Burgi, E. (1972) J. Mol. Biol. 72, 127-147. Elgin, S. C. R. & Weintraub, H. (1975) Annu. Rev. Biochem. 44, 725-774. Stollar, B. D. & Ward, M. (1970) J. Biol. Chem. 245,1261-1266. Rouviere-Yaniv, J. & Gros, F. (1975) Proc. Natl. Acad. Sci. USA 72,3428-3432. Stanier, R. Y., Kunisawa, R., Mandel, M. & Cohen-Bazire, G. (1971) Bacteriol. Rev. 35, 171-205. Guiso, N. & Truffa-Bacchi, P. (1974) Eur. J. Biochem. 42, 401-4.04. Adolph, K. W. & Haselkorn, R. (1971) Virology 46,200-208. Davies, G. E. & Stark, G. R. (1970) Proc. Natl. Acad. Sci. USA 66,651-656. Cukier-Kahn, R., Jacquet, M. & Gros. F. (1972) Proc. Nati. Acad. Sci. USA 69,3643-3647. Ghosh, S. & Echols, H. (1972) Proc. Natl. Acad. Sci. USA 69, 3660-3664. Franze de Fernandez, M. T., Eoyang, L. & August, J. T. (1968) Nature 219,558-590. Ramakrishnan, T. & Echols, H. (1973) J. Mol. Biol. 78,675-686. Wilson, D. L. & Geiduschek, E. P. (1969) Proc. Nati. Acad. Sci. USA 62,514-520. Johnson, G. G. & Geiduschek, E. P. (1972) J. Biol. Chem. 247, 3571-578. Searcy, D. G. (1975) Biochim. Biophys. Acta 395,535-547. Rizzo, P. J. & Nooden, L. D. (1974) Biochim. Biophys. Acta 349, 415-427. Griffith, J. D. (1976) Proc. Natt. Acad. Sci. USA 73, 563-57.
2. Stonington, 0. G. &
3. 4. 5. 6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16. 17. 18.
19.
Vol. 73, No. 6, pp. 1917-1920, June 1976
Biochemistry
Cyanobacterial DNA-binding protein related to Escherichia coli HU (low-molecular-weight protein/histone-like protein/lysine-rich protein/prokaryotic DNA-binding protein)
ROBERT HASELKORN*t AND JOSETTE ROUVIERE-YANIVt * Service de Physiologie
Microbienne and t D~partement de Biologie Moleculaire, Institut Pasteur, Paris, 75015
Communicated by Donald F. Steiner, March 25, 1976 ABSTRACT
A DNA-binding protein has been isolated from
(blue-green algae): Anabaena sp and Aphanocapsa sp. It has a molecular weight as determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis of about 10,000, is rich in lysine and arginine, and lacks tyrosine, tryptophan, and cysteine. The proteins from both strains show immunological identity with a similar DNA-binding protein from Escherichia coli, when tested by immunodiffusion with an antiserum prepared against the E. coli protein. Both the Aphanocapsa and E. coli proteins form compact, rapidly sedimenting complexes with cyanophage or bacteriophage DNA. The similarities between the proteins from cyanobacteria and E. coli suggest a degree of evolutionary conservation comparable to that of the histones of eukaryotes. two cyanobacteria
Models for the regulation of transcription in eukaryotes have to take into account the structure of DNA in chromosomes. That structure is determined to a considerable degree by histones and
nonhistone proteins in chromatin. Prokaryotic transcription and replication have only recently been studied from the point of view of the structural considerations imposed by the arrangement of the DNA in the bacterial nucleoid (1-3). The histones have been strongly conserved during the evolution of eukaryotes. For example, the amino acid sequences of histones H3 and H4 from a wide variety of sources are virtually identical (4). Histones H2a and H2b from man, pig, chicken, and lobster are sufficiently related antigenically to be indistinguishable by the microcomplement fixation test (5). With the recent discovery in Escherichta coli of a low-molecular-weight DNA-binding protein having a histone-like amino acid composition (6), it became of interest to seek a similar protein in other prokaryotes. We report here the characterization of such a protein, which we cQll HU, from two representative cyanobacteria (blue-green algae). MATERIALS AND METHODS Biological Material. Protein HU was prepared from two strains from the collection of the Service de Physiologie Microbienne, Institut Pasteur. Strain 7120 is an Anabaena species which is filamentous, forms heterocysts, and fixes nitrogen. Strain 6701, from which the protein most extensively characterized was obtained, is an unicellular organism, Aphanocapsa. Both strains were grown under axenic conditions in 10 liter carboys of medium GN (7) modified by a 10-fold increase in the concentration of Na2CO3 and gassed with 1% CO2 in air. Carboys were illuminated by banks of Osram-L-interna fluorescent lamps which provided an intensity of 5000 lux (Im/m2) at the surface of the carboy. Purification of DNA-Binding Proteins. The procedure was essentially the one described for E. coli HU protein (6). The frozen paste of strain 6701 cells (20 g) was diluted with 60 ml of the following buffer: 20 mM Tris-HCI at pH 8.1, 2 mM Permanent address: Department of Biophysics and Theoretical Biology, University of Chicago, Chicago, Ill. 60637.
1917
CaCI2, 10 mM MgCI2, 1 mM EDTA, 1 mM mercaptoethanol containing 20 Atg/ml of DNase. The diluted paste was passed through a French pressure cell once at 20,000 pounds/inch2 (140 MPa), incubated for 15 min at room temperature, and then centrifuged successively for 10 min at 8000 rpm and 180 min at 40,000 rpm. The final supernatant was dialyzed against 20 mM Tris-HCI at pH 8.1,5 mM EDTA, 50 mM NaCl, 0.5 mM mercaptoethanol, made 10% (wt/vol) in glycerol, and centrifuged again for 10 min at 8000 rpm. The supernatant, about 70 ml, was applied to a DNA-cellulose column prepared as described previously (6), washed extensively with 20 mM Tris-HCI at pH 8.1,5 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol, 50 mM NaCl, and then developed with successive steps of this buffer containing 0.15 M, 0.4 M, 0.6 M, and 2.0 M NaCl. Material absorbing at 280 nm was obtained in the first fractions following each change of salt concentration. These fractions were analyzed by sodium dodecyl sulfate gel electrophoresis on a 15% polyacrylamide gel as described in ref. 6. The material eluting at 0.4 M salt displayed at least eight bands on the gels, but over 80% of the Coomassie-blue-staining material was contained in one band corresponding to a molecular weight of about 10,000. Fractions containing this band were pooled, concentrated, and dialyzed against 20 mM TrisHCI at pH 8.1, 1 mM EDTA, 0.3 M NaCl, 10% glycerol. The pooled fractions were then passed over a column of Sephadex G-100 equilibrated and developed with the same buffer. The column (1.5 X 15 cm) was calibrated with chymotrypsinogen A, myoglobin, and cytochrome c; for each run Blue Dextran 2000 and tryptophan were added to indicate the void and included volume, respectively. The pooled fractions from the DNA-cellulose column fractionated on the G-100 column as follows. The chlorophyll a, together with essentially all other material absorbing at 280 nm, emerged with the void volume. A second small peak of material, detected only by absorption at 240 nm, emerged at a position corresponding to that of chymotrypsinogen A. The voided fractions contained no protein detectable by polyacrylamide gel electrophoresis. The fractions absorbing at 240 nm contained material which yielded a single band on sodium dodecyl sulfate polyacrylamide gel electrophoresis corresponding to a molecular weight of about 10,000. These fractions were pooled and comprised the material used for determination of amino acid composition and for DNA binding experiments. Immunodiffusion. Rabbits were immunized with pure E. coli HU protein (6), as described by Guiso and Truffa-Bacchi (8). The y-globulins were concentrated 3-fold by ammonium sulfate precipitation. Immunodiffusion (Fig. 1C) and immunoelectrophoresis (J. Rouviere-Yaniv, unpublished work) revealed a single precipitin band between the antiserum and E. coli HU protein. Amino Acid Composition. Protein samples were Iyophilized, dissolved in 6 M HCI, sealed under vacuum, and hydrolyzed at 1100 for 24 and 48 hr. Samples containing 400 nmol of total
1918
Biochemistry: Haselkorn and Rouviere-Yaniv
Proc. Nati. Acad. Sci. USA 73 (1976) Table 1. Amino acid composition of HU proteins
B
Mole %
c
al.
I
Al
FIG. 1. Immunodiffusion of HU proteins. The central well contained anti-E. coli HU 'y-globulin in each case. (A) Peak fractions of Anabaena protein eluted from DNA-cellulose with 0.4 M NaCl (wells a, b, and c). (B) Peak fractions of Aphanocapsa protein eluted similarly (a-f). (C) ec: purified E. coli HU, 20 ug/ml; al: Aphanocapsa HU purified through Sephadex G-100 (see Fig. 2, sample C), approximately 150 ,g/ml. The last two unlabeled wells contained coliphage T4 gene 32 protein at 1 mg/ml or 20 tg/ml.
amino acids were analyzed by G. LeBras and M. Gouyvenaux Technicon Analyzer. Tryptophan was shown to be absent by the lack of absorption in the ultraviolet (see above). Cysteine was determined as cysteic acid on a sample oxidized with performic acid prior to hydrolysis; none was found. Ultracentrifugation. The sedimentation coefficients of DNA from coliphage X and cyanophage N-1 (9), as well as those of complexes of the DNA with protein, were determined on an MSE ultrascan analytical ultracentrifuge using cells with an optical path of 1 cm and solutions having an absorbance at 260 nm of 0.4. Complexes were prepared by mixing DNA and protein in 0.01 M sodium phosphate at pH 7.1, 0.1 M KC1. Lambda DNA was prepared as described previously (6). N-1 was grown on strain 7120 and purified, and the DNA was prepared by phenol extraction, as described previously (9).
on a
RESULTS
Cyanobacterial proteins serologically related to E. coli HU
The fractions isolated from Anabaena (7120) extracts by DNA-cellulose chromatography were tested for cross-reaction with an antiserum prepared against E. coli HU using Ouchterlony immunodiffusion tests in agar. The column fractions eluted at 0.4 M NaCl contained material yielding precipitin bands (Fig. 1A), whereas those eluted at 0.15 M, 0.6 M, and 2 M NaCl did not. The fractions containing cross-reacting antigen do not absorb ultraviolet light at 280 nm, but do show a very weak phenylalanine-like absorption around 260 nm. They also contain a single major protein revealed by polyacrylamide gel electrophoresis, having a mobility greater than cytochrome c.
Amino acid
Aphanocapsa
E. coli (6)
Lysine Histidine Arginine Aspartate + asparagine Threonine Serine Glutamate + glutamine Proline Glycine Alanine Valine Isoleucine Leucine Tyrosine Phenylalanine Methionine Cysteine Tryptophan
14.0 1.5 5.4
14.0 1.5 5.1
8.0 6.0 8.1
8.1 6.0 4.4
8.1 4.5 7.5 11.5 10.0 3.3 3.4 0 4.0 2.5 0 0
9.6 3.0 7.4 16.3 6.0 6.0 6.6 0 3.0 1.5 0 0
These determinations exhausted the supply of protein from Anabaena.
The larger preparation of DNA-binding protein from Aphanocapsa (6701) was characterized more extensively. Fig. 1 shows the serological cross-reaction with the fractions eluted by 0.4 M NaCl from the DNA-cellulose column (B) and with the fractions corresponding to the peak of absorbance at 240 nm from the G-100 column (C). No cross-reacting material was found in the other fractions from the DNA-cellulose column and none was in the void volume from the G-100 column, which contains all of the material absorbing at 280 nm. The absence of spurs between the precipitin bands formed with the E. coli and cyanobacterial proteins (Fig. IC) demonstrates that the two proteins are closely related. Amino acid composition The ultraviolet absorption spectrum of the material eluted from the G-100 column is absolutely flat above 270 nm, indicating the absence of tryptophan and tyrosine. No cysteic acid was found in a performic acid oxidized sample after hydrolysis and ion-exchange chromatography, indicating the absence of cysteine. The relative molar concentrations of the remaining amino acids are listed in Table 1, together with the corresponding figures for the E. coli HU. Both proteins are strikingly similar with respect to lysine, arginine, and histidine content, and neither contains cysteine, tryptophan, or tyrosine. The two proteins have similar ratios of basic to acidic amino acids. The decrease in alanine content of the cyanobacterial protein is balanced by an increase in serine content. The percentages of hydrophobic amino acids are also similar, although the cyanobacterial protein contains more valine than isoleucine and leucine. Molecular weight The cyanobacterial protein elutes from the G-100 column at a position identical to that of chymotrypsinogen A. Thus its maximum molecular weight is 25,000, since all the calibrating proteins are effectively spherical. On polyacrylamide gels containing sodium dodecyl sulfate the protein moves faster than
Proc. Nati. Acad. Sci. USA 73 (1976)
Biochemistry: Haselkorn and Rouvie're-Yaniv
1919
Table 2. Sedimentation coefficients of viral DNA-HU complexes Sedimentation coefficient, 200 DNA source
N-1 X
DNA
DNA + cyanobacterial HU
DNA + E. coli HU
32 28
44 48
44 73
The solvent in all cases was 0.1 M KCl, 0.01 M NaPO4, pH 7.1. DNA and protein concentrations were each 20 1Ag/ml.
algal
___
_
HU
HU~ A
~ B
C
i
_ ~
E.
coli
B
FIG. 2. Polyacrylamide gel electrophoresis of Aphanocapsa DNA-binding protein. (A) Shows the peak fraction eluted from the DNA-cellulose column with 0.4 M salt; (B) is E. coli HU purified through the phosphocellulose step (6); (C) contains material from three pooled fractions from the G-100 column corresponding to the peak of absorbancy at 240 nm (see Materials and Methods).
cytochrome c but slightly slower than E. coli HU (Fig. 2), and thus has a molecular weight as determined by this method of 10,000. A small amount of the protein was reacted with 2 mg/nl of dimethyl suberimidate in 0.2 M triethanolamine, pH 8.5, for 2 hr at room temperature (10). More than half of the product migrated on a sodium dodecyl sulfate gel slightly more slowly than unreacted material, while the remainder, perhaps 2('0%, had a molecular weight, by this method, of 25,000. We interpret these results to mean that the cyanobacterial protein exists, in dilute solution, as a dimer. When reacted under the same conditions, E. coli HU gave the dimer as a major product and a trace of tetramer J. Rouviere-Yaniv, unpublished results). Sedimentation of DNA-protein complexes The protein was originally isolated by binding and elution from a column containing heterologous (calf thymus) DNA (6). We have studied the interaction of the protein in solution with DNA preparations from a bacteriophage (coliphage X) and a cyanophage (N-1). These were mixed with roughly equal weights of protein and analyzed by boundary centrifugation. The results, summarized in Table 2, are indicative of substantial interaction. The sedimentation profiles for A DNA with the two proteins are shown in Fig. 3. All the DNA molecules, under these conditions which correspond to protein excess, bind roughly comparable amounts of protein.
has a similar amino acid composition, and likewise forms rapidly sedimenting complexes with DNA from bacterial viruses. Approximately 800 ,ug of protein HU were isolated after 0-100 chromatography from 20 g (wet weight) of strain 6701, corresponding to 7.5 X 1011 cells. Therefore the minimum number of HU monomers per cell is 60,000. In view of likely losses during preparation of the crude extract the real number could be substantially higher. Cells of 6701 contain an average of two chromosomes, each with kinetic complexity corresponding to a molecular weight of 2.7 X IOP (M. Herdman, personal communication) or approximately 107 base pairs per cell. The minimum ratio of HU to DNA is thus one dimer of HU for every 300 base pairs. The studies of HU binding by electron microscopy (6) reveal that HU binds to all parts of the DNA molecules, but not uniformly. The images observed suggest a compaction of the DNA structure. These observations are confirmed by ultracentrifugation studies: the sedimentation boundaries formed by HUDNA complexes are not much broader than those of free DNA, and the DNA-HU complexes are much more compact than free DNA. The addition of protein to DNA results in a complex of increased partial specific volume, the increase being proportional to the amount of protein. For a complex with equal weight of protein and DNA, the partial specific volume is increased about 25% compared to DNA alone. For homologous DNA molecules the sedimentation coefficient, s, is proportional to Mro 35 where Mr is molecular weight; doubling the molecular weight increases s by about 30%. Since this increase is nearly compensated by the change in partial specific volume, one expects the sedimentation coefficient of HU-DNA complexes 0.2
A
0.1
E C
~.0.21 LO
(N4 11
0.1
.0 .0
0.2' 0.1
.0 0.1
-W
DISCUSSION of DNA-binding proteins from the isolation We have described two different cyanobacteria, each representative of a major sub-class of these organisns. In spite of the evolutionary distance between E. colf and cyanobacteria, the cyanobacterial DNAbiding
protein is serologically related to its E. col counterpart,
Oj
VV
FIG. 3. Ultracentrifugation of X DNA and its complexes with HU. Each panel shows successive scans at 4 min intervals after the rotor reached a speed of 35,000 rpm. (A) X DNA, 20 gg/ml in 0.1 M KCI, 0.01 M NaPO4; (B) X DNA, 20 jg/ml plus Aphanocapsa HU, 20 sg/mL (C) X DNA, 20 jg/ml plus E. coli HU, 20 sg/ml. The direction of sedimentation is from left to right.
1920
Proc. Nad. Acad. Sci. USA 73 (1976)
Biochemistry: Haselkorn and Rouvibre-Yaniv
to be nearly the same as that of the DNA alone. In fact, s increases greatly (Table 2), so the complexes must be relatively compact. In addition to HU, several other DNA-binding proteins of low molecular weight have been described from E. coli (11-14). Like HU, they stimulate transcription; however, this stimulation appears to be specific, whereas the stimulation by HU is not (6). Therefore their function in the cell could be different from that of HU. By contrast, a few DNA-binding proteins, possibly related to HU, have been isolated from other bacteria and characterized. Geiduschek and coworkers (15) described a protein, TFI, which is synthesized during phage SP01 infection of Ba-
cillus subtilis. TFI is a dimer with monomer molecular weight about 11,500. Its amino acid composition resembles that of HU, but it has a lower ratio of basic to acidic amino acids and a different lysine:arginine ratio. There are at least 105 molecules per infected cell (16). Its function in SPOI infection is unknown, but it has been shown to inhibit the transcription of SPO1 DNA in vitro. On the other hand, E. coli HU stimulates the transcription of X DNA in vitro (6). Another histone-like protein was recently isolated from Thermoplasma acidophilus (17), a member of the mycoplasma group that grows at a very low pH and at high temperature, and contains DNA of unusually high AT content. This protein is strongly associated with DNA and is soluble in acid. Its amino acid composition is similar to that of HU, but its molecular weight is higher. The comparison of HU protein and primitive eukaryote chromosomal proteins may throw more light on the evolutionary relation between prokaryotes and eukaryotes. For example, it will be interesting to know the extent of relatedness between HU and the acid-soluble low molecular weight protein isolated from dinoflagellate chromatin by Rizzo and Nooden (18). The cellular function of HU still remains conjectural, although it seems probable that it plays a role in determining the structure of the prokaryotic chromosome. In such a structure, HU protein could prevent the RNA polymerase from binding to nonproductive sites on DNA, as is suggested by the stimulatory effect of HU on X DNA transcription that is observed only when DNA is in excess (6). Finally, it will be of interest to determine whether HU is responsible for maintaining the chro-
matin-like appearance of the E. coil chromosome observed by Griffith (19). We thank G. Cohen-Bazire and F. Schaeffer for the gift of cyanobacterial cells. We are indebted to M. Goldberg, N. Leveque, 0. Croissant, P. Truffa-Bacchi, and M. Yaniv for advice and help during the course of this work. R.H. is grateful to Roger Stanier and the members of the Service de Physiologie Microbienne for their hospitality and to the John Simon Guggenheim Foundation for a fellowship enabling him to work in Paris. J.R:-Y. is grateful to F. Gros for his interest in this work and for providing laboratory facilities. The work was supported by grants from the Centre National de la Recherche Scientifique. 1. Rouviere, J., Lederberg, S., Granboulan, P. & Gros, F. (1969) J.
Mol. Biol. 46,413-430.
Pettijohn, D. W. (1971) Proc. Nati. Acad. Sci. USA 68,609. Worcel, A. & Burgi, E. (1972) J. Mol. Biol. 72, 127-147. Elgin, S. C. R. & Weintraub, H. (1975) Annu. Rev. Biochem. 44, 725-774. Stollar, B. D. & Ward, M. (1970) J. Biol. Chem. 245,1261-1266. Rouviere-Yaniv, J. & Gros, F. (1975) Proc. Natl. Acad. Sci. USA 72,3428-3432. Stanier, R. Y., Kunisawa, R., Mandel, M. & Cohen-Bazire, G. (1971) Bacteriol. Rev. 35, 171-205. Guiso, N. & Truffa-Bacchi, P. (1974) Eur. J. Biochem. 42, 401-4.04. Adolph, K. W. & Haselkorn, R. (1971) Virology 46,200-208. Davies, G. E. & Stark, G. R. (1970) Proc. Natl. Acad. Sci. USA 66,651-656. Cukier-Kahn, R., Jacquet, M. & Gros. F. (1972) Proc. Nati. Acad. Sci. USA 69,3643-3647. Ghosh, S. & Echols, H. (1972) Proc. Natl. Acad. Sci. USA 69, 3660-3664. Franze de Fernandez, M. T., Eoyang, L. & August, J. T. (1968) Nature 219,558-590. Ramakrishnan, T. & Echols, H. (1973) J. Mol. Biol. 78,675-686. Wilson, D. L. & Geiduschek, E. P. (1969) Proc. Nati. Acad. Sci. USA 62,514-520. Johnson, G. G. & Geiduschek, E. P. (1972) J. Biol. Chem. 247, 3571-578. Searcy, D. G. (1975) Biochim. Biophys. Acta 395,535-547. Rizzo, P. J. & Nooden, L. D. (1974) Biochim. Biophys. Acta 349, 415-427. Griffith, J. D. (1976) Proc. Natt. Acad. Sci. USA 73, 563-57.
2. Stonington, 0. G. &
3. 4. 5. 6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16. 17. 18.
19.
