The protein ORF80 from the acidophilic and thermophilic archaeon Sulfolobus islandicus binds highly site-specifically to double-stranded DNA and represents a novel type of basic leucine zipper protein

Abstract

The cryptic high copy number plasmid pRN1 from the thermophilic and acidophilic crenarchaeote Sulfolobus islandicus shares three conserved open reading frames with other S.islandicus plasmids. One of the open reading frames, namely orf80, encodes a 9.5 kDa protein that has no homology to any characterised protein. Recombinant ORF80 purified from Escherichia coli binds to double-stranded DNA in a sequence-specific manner as suggested by EMSA experiments and DNase I footprints. Two highly symmetrical binding sites separated by ∼60 bp were found upstream of the orf80 gene. Both binding sites contain two TTAA motifs as well as other conserved bases. Fluorescence measurements show that short duplex DNAs derived from a single binding site sequence are bound with submicromolar affinity and moderate cooperativity by ORF80. On DNA fragments carrying both binding sites, a rather large protein–DNA complex is formed in a highly cooperative manner. ORF80 contains an N-terminal leucine zipper motif and a highly basic domain at its C-terminus. Compared to all known basic leucine zipper proteins the order of the domains is reversed in ORF80. ORF80 may therefore constitute a new subclass of basic leucine zipper DNA-binding proteins.

INTRODUCTION

Plasmid pRN1 from the acidophilic and thermophilic archaeon Sulfolobus islandicus was first described in 1994 by Zillig et al. (1). pRN1 belongs to the archaeal plasmid family pRN (2), which now includes five members: plasmids pRN2 (3), pSSVx (4) and pHEN7 (2) have been isolated from S.islandicus strains; another member of this family, pDL10, is found in the crenarchaeote Acidianus ambivalens.

The replication mode of the pRN plasmid family has not yet been investigated in detail. Indirect evidence exists that pDL10 replicates via a rolling circle mechanism (5). The pRN plasmids share a core of three highly conserved open reading frames. The shortest gene, orf56 in pRN1, has low homology to plasmid copy control proteins of several eubacterial plasmids (3). We could show that purified recombinant ORF56 binds with high affinity to an imperfect inverted repeat within its promoter (6), suggesting that ORF56 may inhibit its own expression and expression of the overlapping gene orf904, which does not seem to have a promoter of its own. Another short open reading frame, orf80 in pRN1, displays no homology to any characterised protein. It was tentatively named plasmid regulatory protein (plr) (5). The third conserved open reading frame codes for a large protein of about 900 amino acids named ORF904. One domain of ORF904 shows homology to the helicase domain of the bacteriophage P4 primase and ORF904 is possibly directly involved in plasmid replication. Although this plasmid family has been very well analysed at the sequence level, information on the biochemical properties and functions of the gene products is still lacking. The plasmid family and especially the relatively small plasmids pRN1 and pRN2 are an attractive backbone for the construction of a high copy number Sulfolobus–Escherichia coli shuttle vector.

In this report we focus on the product of the orf80 gene from pRN1. Homologues of the gene orf80 are also found in the conjugative plasmids pNOB8 (7) and pING1 (8) and within the genome of Sulfolobus solfataricus. We expressed orf80 in E.coli and characterised the purified recombinant protein. ORF80 binds specifically to two sites located upstream of its own gene. The two binding sites have distinct DNA elements in common and are of palindromic structure. Interestingly, ORF80 could constitute a new subclass of basic leucine zipper DNA-binding proteins, where the N-terminal leucine zipper is followed by a C-terminal basic domain.

MATERIALS AND METHODS

Cloning and expression of orf80

All cloning steps were performed in E.coli XL1-Blue. The orf80 gene was amplified from plasmid pUC18-pRN1 (a gift from David Faguy, Dalhousie University, Halifax, Canada) by PCR with forward primer 5′-AAGAATTCCATGGGATCCCATATGAGTGA and reverse primer 5′-GCCGTGACGAACTCCTGTGACTCGAGCTCAACC. The PCR product was cut with NcoI and XhoI (New England Biolabs) and ligated into the expression vector pET28c (Novagen) cut with the same enzymes. The identity of the insert of the resulting plasmid (pET28-orf80) was confirmed by DNA sequencing. The final construct was transformed into E.coli BL21(DE3) cells, which were used for overexpression. Aliquots of 4 ml of an overnight culture were inoculated into 1 l of LB medium supplemented with 50 µg/ml kanamycin and grown at 37°C. After 3 h, 1 mM IPTG was added and the culture was fermented for a further 3 h. The cells were collected by centrifugation and kept frozen at –70°C until use.

Purification of ORF80

The frozen cells were resuspended in 20 ml of lysis buffer (50 mM sodium phosphate, 1 mM EDTA, pH 8.0). Lysozyme (100 µg/ml) and Triton X-100 (0.1%) were added and the cells were incubated for 15 min at 4°C. After sonication the lysate was clarified by centrifugation for 30 min at 4°C and 40 000 g. The pH of the supernatant was adjusted to pH 6.0 and then loaded onto a fractogel EMD-DEAE column (16 ml). The column was developed with a linear salt gradient from 0.2 to 2 M Na₂SO₄ in 25 mM citrate, pH 6.0. ORF80 eluted at ∼1 M Na₂SO₄. Protein-containing fractions were pooled and diluted to 0.3 M Na₂SO₄ with double distilled water. The pool was then loaded onto a 1 ml hydroxyapatite column (Macroprep; Bio-Rad), which was developed with a linear salt gradient from 0.2 to 2 M NaCl in 50 mM urea, 5 mM potassium phosphate, pH 7.0. ORF80 eluted as a sharp peak at 1 M NaCl. ORF80-containing fractions were pooled and dialysed against 30% glycerol in 25 mM sodium phosphate, pH 8.0. The yield from this purification was highly variable and rather low, with ∼0.3 mg ORF80 per l expression culture, as the majority of ORF80 aggregated during purification and had to be removed by centrifugation.

During the course of the study we learned that ORF80 could also be purified using an alternative approach. Cell extracts were precipitated with 20% saturated ammonium sulphate. The precipitated protein was washed with water and the water-insoluble ORF80 was solubilised in 6 M urea, 300 mM NaCl, 50 mM sodium phosphate, pH 7.0, and purified on an EMD-SO^3– column in the presence of 6 M urea. ORF80 purified by this protocol was indistinguishable in electrophoretic mobility shift assays (EMSAs) from ORF80 purified by the former protocol. The alternative purification scheme allowed us to obtain and store ORF80 at concentrations >50 µM, which we needed to perform the CD and fluorescence measurements.

The concentration of the purified protein was calculated using the theoretical extinction coefficient ɛ = 11 744 M^–1 cm^–1.

EMSAs

First, the DNA-binding activity of ORF80 was assayed with restriction fragments obtained from pUC18-pRN1, which contains the complete sequence of pRN1. ORF80 was incubated for 2 min with the restriction digests in 20 mM Tris–acetate, 50 mM potassium acetate, 10 mM magnesium acetate, 1 mM DTT, pH 7.9. DNA fragments were then separated on 1% agarose gels in 0.5× TBE (45 mM Tris–borate, 0.5 mM EDTA, pH 8.3) and stained with ethidium bromide.

After identification of the binding site by footprinting experiments, ³²P-labelled oligodeoxynucleotides encompassing the ORF80 binding site were hybridised with a complementary oligodeoxynucleotide and used as probes in EMSA (Table 1). These reactions were carried out in 50 mM Tris–HCl, pH 8.0, in the presence of varying amounts of ORF80 and competitor DNA. The binding reactions were analysed on 6% polyacrylamide gels run in 0.5× TBE.

Table 1. Short DNA substrates used in this study.

[CRC Table-insert tagged image=gke651t01]

The sequences of the respective forward oligonucleotides are shown. These oligonucleotides were hybridised with their complementary oligonucleotides and used as double-stranded substrates in EMSA experiments. BS-31bp contains one complete ORF80 binding site and was used for most experiments. The fluorescent dye Texas Red is attached at the 3′-end of BS-36bpTR. The asterisks above the DNA substrates indicate the region protected by ORF80 as revealed by DNase I footprinting. The bases in bold are conserved in both binding sites and are palindromic (see also Fig. 3C). HS-20bpTR contains only one half-site of the ORF80-binding site.

A 149 bp DNA fragment carrying both binding sites (Fig. 1B) was prepared by PCR with a ³²P-labelled forward primer (5′-GCTAACACGATAAGGCAAAC) and an unlabelled reverse primer (5′-AGCTTTTCCTTCAGATCAC). These binding experiments were analysed in 10% polyacrylamide gels.

Figure 1.

Genomic map of the cryptic high copy number plasmid pRN1. (A) The plasmid pRN1 is a circular plasmid of 5350 bp. The three open reading frames orf56, orf80 and orf904 have homologues in other archaeal plasmids and are indicated by thick dark grey lines. An archaeal promoter is found upstream of orf56 and orf80. The short open reading frames orf90a, orf72 and orf90b (light grey) have no homology to any known protein and may not code for proteins. (B) Details of the region flanking orf80. Vertical arrows indicate restriction sites. The thick horizontal arrow shows the localisation of the orf80 coding region. The numbers refer to the base positions of GenBank database entry U36383.

Gels were analysed by electronic autoradiography with an Instant Imager (Canberra Packard). The image intensity is linear with respect to the signal. Quantification of binding was performed with the instrument’s software.

DNase I footprinting

Aliquots of 30 pmol footprint forward primer (5′-CGCCACTTGGCGAGAAATTTGCTCAAAG) were labelled with 6 pmol [γ-³²P]ATP (1.2 MBq; Hartmann, Germany) by incubation for 15 min with T4 polynucleotide kinase (Fermentas) at 37°C. The labelled primer was used directly in PCR with unlabelled reverse primer (5′-TCTAACTCTTCCAGGGTTTGAC). The annealing temperature was 60°C and the template for amplification was pUC-pRN1. The resulting PCR fragment of 470 bp was pure as judged by denaturing gel electrophoresis and was purified by adsorption to silica (QiaQuick; Qiagen).

Aliquots of 1–5 pmol ORF80, 100 ng poly[d(I-C)·d(I-C)] and the labelled PCR fragment (∼5000 Bq, 125 fmol) were combined in a 4 µl binding reaction in 50 mM Tris–HCl, pH 7.0, and incubated for 5 min at 37°C. Then 3 µl of 0.05 or 0.1 mU/µl DNase I diluted in 20 mM Tris–HCl, 5 mM CaCl₂, 10 mM MgCl₂, pH 7.5, were added to the binding reaction. After 2 min at room temperature, the reaction was stopped by adding 12 µl of stop solution (80% formamide, 0.1% bromphenol blue, 0.1% xylene cyanol, 10 mM EDTA). A sample of 5 µl of the solution was incubated for 5 min at 80°C and then loaded on an 8% polyacrylamide sequencing gel. The dried gel was exposed to storage screens and analysed on a PhosphorImager (Molecular Dynamics).

Fluorescence titrations

All fluorescence titrations were carried out on a LS-50B fluorospectrometer from Perkin Elmer equipped with a thermostat and a stirred jacket. The analytical titrations were done in 1 ml cuvettes with 5 nM BS-36bpTR (Table 1) in standard buffer (25 mM sodium phosphate, 100 mM NaCl, 0.01% Tween-20, pH 7.0). The fluorescence of free DNA was recorded at 613 nm with a slit width of 20 nm (excitation at 585 nm with a slit width of 15 nm). ORF80 was added to a final concentration of 1 µM. Then, 300 µl of the solution in the cuvette was serially replaced (typically 12 times) by 300 µl of a 5 nM DNA solution in standard buffer.

For the competition experiments 5 nM BS-36bp was allowed to bind to 250 nM ORF80. Then increasing amounts of unlabelled BS-36bp and of calf thymus DNA were added. For the non-specific competitor DNA we assumed that the number of binding sites equals the number of base pairs, implying overlapping binding sites, which is allowed as the number of binding sites exceeds by far the protein concentration.

For the stoichiometric titrations we used 0.5 µM BS-36bpTR and 0.5 µM HS-20bpTR (Table 1). The titrations were carried out in standard buffer and a 100 µl cuvette. For these measurements the same excitation and emission wavelengths were used, but the slit widths were narrowed to 5 and 6.5 nm, respectively.

Dimethylsuberimidate (DMS) crosslinks

Aliquots of 300 pmol ORF80 were incubated with and without 300 pmol double-stranded binding site DNA (BS-31bp) in the presence of 10 mM DMS in 0.2 M triethanolamine, pH 8.0, for 2 h at room temperature in a reaction volume of 30 µl. Aliquots of the reactions were separated on a 13% polyacrylamide peptide gel according to Schagger and von Jagow (9).

RESULTS

Expression and purification

ORF80 was overexpressed in E.coli using the T7 expression system. Purification of ORF80 proved to be difficult due to the low expression of soluble ORF80 and its high tendency to aggregate. We finally worked out a two column strategy, which yielded essentially pure ORF80 (see Fig. 7, lane 1). From the retention of purified ORF80 in SDS–PAGE gels an apparent molecular mass of 11 kDa was determined, which agrees with the theoretical molecular mass of the full-length protein of 9.6 kDa. The N-terminus of the purified protein was determined to be GSHMSD, indicating that the initiator methionine is removed by the E.coli host cell during expression.

Figure 7.

Crosslinking of ORF80 with DMS. ORF80 was incubated with and without DMS and analysed on a denaturing peptide gel. Lane 1, no DMS; lane 2, 10 mM DMS added. Chemical crosslinking was repeated in the presence of binding site DNA (BS-31bp). Lane 3, no DMS; lane 4, 10 mM DMS.

When purified ORF80 is heated for only 1 min at 95°C in the presence of 0.5% SDS and 0.25% β-mercaptoethanol prior to loading onto a denaturing peptide gel, a second protein band with an apparent molecular mass of 23 kDa is observed (data not shown). We ascribe this protein band to incompletely denatured dimers of ORF80. Unfortunately we could not analyse the quaternary structure of ORF80 by gel filtration as the protein aggregated on the column, even under slightly denaturing conditions (1 M urea). Additional evidence that ORF80 is dimeric in solution comes from our DMS and glutaraldehyde crossslinks (see below).

ORF80 is a DNA-binding protein

First, the DNA-binding activity of recombinant ORF80 was assayed with various restriction fragments of pUC18-pRN1. Figure 2 shows a restriction with ClaI/SphI yielding a 472 bp fragment whose mobility decreases in the presence of ORF80. More importantly, when ORF80 was probed with a TaqI restriction digest that yields numerous restriction fragments only one fragment of ∼610 bp was specifically shifted. This fragment and the 472 bp ClaI/SphI fragment originate from the same region of pRN1 around the gene orf80 (Fig. 1B). The decrease in mobility after complexation with ORF80 is very large, suggesting that a large protein–DNA complex is formed. Indeed, binding of a single molecule of a 9.6 kDa protein to a DNA fragment of several hundred base pairs length is unlikely to result in a significant decrease in mobility of the complexed DNA compared with the unbound DNA.

Figure 2.

ORF80 binds specifically double-stranded DNA. Plasmid pUC-pRN1 was digested with various restriction endonucleases. The digests were then probed with purified ORF80. The samples were separated on a 1% agarose gel and the DNA visualised with ethidium bromide. Lane M, size marker. Double restriction with ClaI and SphI (lanes 1 and 2) yields a 472 bp restriction fragment whose mobility decreases upon incubation with 8 µM purified ORF80 (lane 2). In lanes 3 and 4 digestion with TaqI is shown, which yields 14 fragments. A fragment of 613 bp is specifically shifted upon addition of 8 µM ORF80 (lane 4, see arrows on the right). The 472 and 613 bp fragments correspond to a region encompassing the orf80 gene (Fig. 1B).

To further narrow down the exact binding site of ORF80 we performed a DNase I footprint using a probe encompassing this region. The footprint shown in Figure 3A clearly demonstrates that ORF80 binds to two sites within this probe. The two binding sites are separated by about 60 bases. At each binding site about 27 bases are protected against DNase I digestion. In addition, several hypersensitive DNase I sites are generated within each binding site. Figure 3B shows the precise location of the two binding sites relative to the initiator methionine of orf80. The two binding sites have a very similar structure. Each contains two TTAA motifs whose centres are separated by 11 bp, i.e. about one helix turn. Both binding sites have a pronounced palindromic structure. The TTAA motifs are themselves palindromic and two of the TTAA boxes together with four other nucleotides are arranged in a highly symmetrical manner (see boxed bases in Fig. 3C).

Figure 3.

Footprint analysis of ORF80. (A) DNase I footprint of ORF80 bound to a 380 bp DNA fragment covering most of the orf80 gene (see Fig. 1B). Lanes 1 and 2, sequencing ladder termination reaction ddG and ddT, respectively; lanes 3 and 8, control reaction, no ORF80 added; lanes 4 and 7, footprint reaction with 0.4 µM ORF80; lanes 5 and 6, footprint with 1.2 µM ORF80. Lanes 3–5, 0.05 mU/µl DNase I; lanes 6–8, 0.1 mU/µl DNase I. No DNase I protection was seen outside the region shown. (B) Nucleotide sequence of pRN1 upstream of the orf80 coding region. The thick black lines illustrate the region that is protected against DNase I hydrolysis. Vertical arrows show the positions of the DNase I hypersensitive sites generated through ORF80 binding. BRE, TATA box and box B are the key elements of a typical archaeal promoter (17,18). (C) The two binding sites observed in the footprint are aligned. Vertical bars indicate identical nucleotides. The boxed nucleotides are conserved in both binding sites and are palindromic, with the centre of symmetry indicated by a black oval. The thick black horizontal line illustrates the region of both sites protected against DNase I digestion.

To study the DNA-binding properties of ORF80 in more detail we investigated the binding of ORF80 to short double- and single-stranded oligonucleotides whose sequences were derived from the binding site that is more distal to the presumed transcription start site. In binding reactions containing 500 nM ORF80 and a 31 bp double-stranded oligonucleotide that encompasses the complete binding site we observed a large protein–DNA aggregate that was not able to enter the gel (Fig. 4A, lane 2). When large amounts of competitor DNA, either calf thymus DNA or poly[d(I-C)·d(I-C)], were added to the binding reaction, a defined protein–DNA complex was formed that was capable of entering the gel (lanes 3 and 7). In the presence of unlabelled binding site DNA this protein–DNA complex was no longer detectable (lane 4). These EMSAs clearly show that ORF80 binds specifically to short double-stranded DNA probes derived from the region protected against DNase I digestion.

Figure 4.

ORF80 binds specifically to its binding site. (A) ORF80 binding analysed with a short duplex DNA probe derived from the DNase I footprint (lanes 1–7) and to single-stranded oligodeoxynucleotides of the same sequence (lanes 8–10). The single-stranded (BS-31bp, see Table 1) and double-stranded probes were labelled with ³²P and used at a concentration of 5 nM in the gel shift binding reaction. Lane 1, no protein; lane 2, 500 nM ORF80; lane 3, 500 nM ORF80 and 1.5 µM bp calf-thymus DNA; lane 4, 500 nM ORF80 and 500 nM unlabelled duplex probe; lanes 5–7, 500 nM ORF80 and 50, 500 and 5000 nM bp poly[d(I-C)·d(I-C)], respectively; lane 8, no protein; lane 9, 500 nM ORF80; lane 10, 500 nM ORF80 and 1.5 µM bp ssM13 DNA. (B) Binding to short double-stranded DNAs of varying length. Lanes 1, 3, 5 and 7, no protein added; lanes 2, 4, 6 and 8, binding reaction in the presence of 500 nM ORF80. All binding reactions contained 5 nM double-stranded ³²P-labelled DNA and 5 µM poly[d(I-C)·d(I-C)]. The probes were BS-36bp (lanes 1 and 2), BS-31bp (lanes 3 and 4), BS-27bp (lanes 5 and 6) and BS-21bp (lanes 7 and 8) (see Table 1 for detailed sequence information). ds and ss represent unbound double- and single-stranded DNA, B is bound DNA and W stands for bound and aggregated DNA that remained in the wells of the gels. The amount of aggregation was somewhat variable [compare lane 7 of (A) with lane 4 of (B)].

In binding experiments with a 31 nt single-stranded probe derived from the binding site we observed the formation of a large complex that could not enter the gel when 500 nM ORF80 was present (Fig. 4A, lane 9). The addition of single-stranded competitor DNA (1.5 µM bases of M13 DNA) to this binding reaction completely abolished complex formation with the probe (lane 10), indicating that ORF80 does not bind to this single-stranded form of the binding site in a specific way.

In the absence of competitor DNA we always observed formation of a large protein–DNA complex, which was unable to enter the gel. This observation is certainly related to the high tendency of ORF80 to form aggregates, a property that was a severe obstacle in the purification scheme. Apparently, the presence of competitor DNA reduces the concentration of free ORF80 due to non-specific affinity of ORF80 for the competitor. In order to observe defined protein–DNA complexes in the gel shift assays we therefore routinely included 5 µM base pairs of poly[d(I-C)·d(I-C)] as competitor. In the presence of this amount of competitor DNA half-maximal binding to the double-stranded binding site was achieved at ∼700 nM ORF80 (Fig. 5A). Quantitative analysis of the binding isotherm suggests that the binding is cooperative, with a Hill constant of ∼2 (Fig. 5C).

Figure 5.

Quantitative EMSA. (A) Binding to the short duplex binding site DNA. The radioactive probe (5 nM BS-31bp) was incubated with 0, 0.08, 0.12, 0.18, 0.26, 0.40, 0.59, 0.89, 1.33 and 2 µM ORF80 in the presence of 5 µM bp poly[d(I-C)·d(I-C)]. (B) Binding to a 149 bp DNA fragment containing both binding sites. The radioactive PCR fragment (∼5 nM) was probed with 0, 0.18, 0.26, 0.40, 0.59, 0.89 and 1.33 µM ORF80 in the presence of 5 µM bp poly[d(I-C)·d(I-C)]. ds represents unbound double-stranded DNA, B is bound DNA and W stands for bound and aggregated DNA that remained in the wells of the gels. (C) Quantitative analysis of the EMSA data obtained from (A) and (B). Open circle, single binding site DNA (31 bp); filled circle, 149 bp DNA fragment with two binding sites. Both titrations show positive cooperativity. The fits are based on the Hill model of binding. The Hill constant n_Hill is 2.1 for the single site substrate and 4.2 for the long substrate. Half-maximal binding is ∼700 nM for both substrates.

In further gel shift experiments we investigated the minimal size of the binding site that allows specific binding in the presence of competitor DNA. In the DNA probes of 36, 31, 27 and 21 bp length the binding site is stepwise shortened (Table 1). A distinct protein–DNA complex was observed for the 36 and 31 bp probes, whereas with the smaller probes only a small amount of aggregate was formed (Fig. 4B). This observation is in line with the footprinting data, as the 31 bp substrate just covers the protected region (Table 1), indicating that the whole footprinting area is required for specific binding. Deletion of the border regions of the footprinting area apparently leads to a loss of contacts that are necessary for formation of a stable specific complex.

The footprinting data have revealed the existence of two ORF80 binding sites that are separated by ∼60 bp. In order to address the question of whether there is cooperativity between the adjacent binding sites, we prepared a 149 bp PCR fragment carrying both sites and analysed ORF80 binding to this probe by gel shift titration experiments. As shown in Figure 5B, several complexes were formed in a cooperative manner. At high concentrations of ORF80 the probe was saturated and a distinct protein–DNA complex of rather low mobility was formed. Intermediary complexes were also visible; however, these complexes were only weakly populated. Analysis of these titrations suggests that the cooperativity of this binding reaction has a Hill coefficient of ∼4 (Fig. 5C), which gives the lower limit of the number of ORF80 molecules that bind cooperatively to the two sites. This result is in line with the large decrease in mobility of the restriction fragments upon ORF80 binding (see Fig. 2).

Fluorescence titrations

To study more precisely the binding affinity and stoichiometry of the DNA–ORF80 protein complex, we developed a fluorescence-based DNA-binding assay for ORF80. We discovered that BS-36bp labelled with the fluorescent dye Texas Red at the 3′-end of the forward strand is a suitable probe for fluorescence titration. As can be seen in Figure 6A, the fluorescence of Texas Red is quenched upon ORF80 binding to about one-third of the initial value. At concentrations >500 nM a slight increase in fluorescence intensity was observed, which is ascribed to the formation of aggregates and/or complexes of higher molecular order. The lower part of the titration, however, could be very well fitted to the Hill equation and yielded a Hill coefficient of n_Hill = 1.9 with half-maximal binding at 136 nM. The observed fluorescence intensity change is not caused by simple binding of ORF80 to the dye, as revealed by competition experiments with BS-36bp (specific DNA) and with calf thymus DNA (non-specific DNA). As expected, the specific DNA was a good competitor, whereas high amounts of the non-specific DNA are needed to displace ORF80 from the fluorescently labelled specific DNA probe (Fig. 6B). The competition experiments also clearly demonstrate that the type of inhibition is different between the specific DNA and the non-specific competitor. The slope of the curve is much steeper for the specific DNA, indicating that the specific competitor binds ORF80 in a cooperative manner. The rapid increase in fluorescence intensity from 20 to 40 nM competitor indicates a rapid decrease in free ORF80 in this concentration range caused by cooperative binding of ORF80 to the specific competitor.

Figure 6.

Fluorescence titrations with Texas Red-labelled binding site DNA. (A) Aliquots of 5 nM Texas Red-labelled BS-36bp were titrated with increasing amounts of ORF80. The solid line is a fit according to the Hill equation (n_Hill = 1.93, k_Hill = 138 nM). (B) Competition experiments with 250 nM ORF80 bound to 5 nM DNA. Increasing amounts of competitor DNA were added. Filled circles, unlabelled specific DNA (BS-36bp); open circles, calf thymus DNA. (C) Stoichiometric titration with 0.5 µM Texas Red-labelled BS-36bp (filled circles) and with HS-20bp (open circles). Stoichiometric equivalence was reached at protein/DNA ratios of about 12 and 6, respectively.

In order to determine the stoichiometry of the protein–DNA complexes we performed fluorescence titrations with 0.5 µM DNA. Under these conditions we expected a nearly linear decrease in fluorescence intensity until the DNA probe was completely saturated with protein. With the half-site binding DNA (HS-20bpTR) stoichiometric equivalence was reached at a protein/DNA ratio of about six. Twice the amount, i.e. about 12 monomers of ORF80, are required to saturate the DNA with one complete binding site (BS-31bpTR). We have no indication whether the protein is fully active. However, we obtained similar titration curves with different protein preparations. The experiment therefore further supports the hypothesis that a rather large complex consisting of several ORF80 monomers is formed on the binding site DNA (see also Discussion).

Oligomer structure of ORF80 and ORF80–DNA complexes

To study the oligomer structure of free and bound ORF80 we performed crosslinking experiments with the bifunctional amine-reactive reagent DMS. When ORF80 was allowed to react with DMS in the absence of binding site DNA, we observed an additional protein band with an apparent molecular mass of 24 kDa in denaturing peptide gels (Fig. 7, lane 2). This protein band is ascribed to a crosslink between two ORF80 molecules. In crosslinking reactions in the presence of DNA the dimeric protein band was stronger and in addition protein bands corresponding to crosslinking of three and four ORF80 polypeptide chains are visible (lane 4). The same pattern is seen when crosslinking was done with glutaraldehyde (data not shown). These experiments indicate that ORF80 is a dimer in solution and assembles into a larger complex when bound to DNA.

Is ORF80 a basic leucine zipper?

Analysis of the primary structure of ORF80 revealed several important properties (Fig. 8). The 30 N-terminal amino acids of ORF80 have a high coiled coil score (10) and are predicted to be helical. In addition, there are five leucines completely conserved among the ORF80 homologues from pRN1, pRN2, pDL10, pSSVx, pING1 and SSO8948. Four of them are in the correct register to form a leucine zipper (positions a and d in a helical wheel presentation). The ORF80 homologue from pNOB8 has an isoleucine in one of the five conserved leucine positions and the homologue from pHEN7 has a methionine in one position. In contrast, two of the genomic ORF80 homologues (SSO8758 and SSO5209) have none or two of these leucines conserved and might therefore be unable to form a leucine zipper. These data strongly suggest that the plasmid subsets of ORF80 homologues as well as SSO8949 can multimerise via the N-terminal domains to form a leucine zipper. For ORF80 the existence of dimers is demonstrated by SDS–PAGE experiments and chemical crosslinking. CD measurements allow an estimation of the secondary structure composition of a protein. The negative ellipticity in the wavelength region between 210 and 250 nm reveals that ORF80 has a significant fraction of helical regions (Fig. 9). Deconvolution of the CD spectrum with a neural network (11) yields an α-helical content of 36% and a β-strand content of 17.1%. These values agree quite well with the GOR IV predictions (43 and 13%, respectively) and thus further support the hypothesis that the N-terminus could be helical, which is a requirement for formation of a leucine zipper.

Figure 8.

Alignment of the ORF80 homologues. The plasmid counterparts of ORF80 and three ORF80 homologues from the S.solfataricus genome are aligned. Five conserved leucine residues are boxed, four of which might form a leucine zipper (positions a and d). The coiled coil score was calculated according to Lupas (10). The characters a–g refer to a helical wheel representation; amino acids at positions a and d are occupied by leucine residues in a typical leucine zipper protein. The secondary structure of ORF80 was evaluated by GOR IV (19). Charge + denotes an arginine or lysine in the consensus sequence whereas – represents a conserved aspartic or glutamic acid.

Figure 9.

CD spectroscopy of ORF80. CD spectra of 12 µM ORF80 in 50 mM sodium phosphate, pH 7.0, and of 15 µM ORF80 in 9.7 M urea, 50 mM sodium phosphate, pH 7.0, are shown. Remarkably, ORF80 is not completely unfolded at 9.7 M urea.

C-terminal to the putative leucine zipper domain there is a less conserved region of about 10 amino acids, which is followed by a highly conserved and positively charged region extending up to the C-terminus of the protein (Fig. 8). The charge and high conservation of this domain suggest that it functions as a DNA-binding region. As compared to the classic basic leucine zipper proteins, such as the transcription factor GCN4, where the basic region is N-terminal to the leucine zipper, these domains are arranged in reverse order in ORF80. Therefore, ORF80 could constitute a new subclass of leucine zipper DNA-binding proteins.

Structure of the ORF80-binding site

All ORF80 homologues have a highly conserved and highly basic C-terminal domain, which is presumed to be involved in DNA binding. Given the high conservation of this putative DNA-binding domain we searched the nucleotide sequence of the respective plasmids for ORF80 binding sites. Upstream of all orf80 genes we find two potential ORF80 binding sites that are all separated by the same distance (Fig. 10). However, the distances between the binding sites and the start codons differ. On plasmids pRN1, pRN2, pHEN7 and pSSVx and on the genomic copy SSO5209 binding of ORF80 to the less upstream binding site could occlude the BRE and TATA box (see also Fig. 3). Both elements are required for transcription (12,13). It therefore seems plausible that ORF80 down-regulates its own synthesis on these plasmids. For plasmids pDL10 and pING1, as well as for the genomic copies SSO8948 and SSO8758, the distance is larger and it is less likely that binding of ORF80 to these sites directly influences transcription of orf80. The most remarkable feature of the ORF80 binding sites is the strong conservation of the spacing of the two binding elements, indicating a close functional and/or structural relationship between the two sites.

Figure 10.

Alignment of the ORF80 putative binding sites. The promoter sequences were aligned with PILEUP (GCG program package) and the consensus calculated with PRETTY. The highly conserved TTAA motifs are boxed. The centres of the two TTAA motifs within one binding site are separated by 11 bp in pRN1, pRN2, pSSVx and pHEN7. The number in parentheses at the beginning of each line refers to the base count of the database entry. The sequences between the two binding sites are poorly conserved, indicating that this region may not contribute to the binding of ORF80. In contrast, the region upstream of the first TTAA motif is highly conserved and purine rich, whereas the sequence downstream of the fourth TTAA motif is pyrimidine rich and also highly conserved. These additional conserved regions are palindromic, with the centre of symmetry between the two binding sites (black oval).

DISCUSSION

The cryptic high copy number plasmid pRN1 from S.islandicus harbours three genes that are conserved among the plasmid family pRN. One of these genes, orf80 from pRN1, has no homology to other known proteins. It is, however, related to genes of the pRN family as well as to other Sulfolobus spp. plasmids. In fact, the orf80 gene is found in all Sulfolobus plasmids characterised so far, suggesting that orf80 codes for an essential plasmid function required for replication and maintenance of the plasmids in the host cell. It is noteworthy that ORF80 homologues were found neither outside the genus Sulfolobus (except pDL10) nor in Sulfolobus virus sequences. orf80 genes are found on the conjugative plasmids pING and pNOB8, suggesting that orf80 is not only restricted to the pRN plasmid family but is also required by conjugative plasmids from Sulfolobus strains. The orf80 gene has not been annotated in the virus–plasmid hybrid pSSVx, but the high similarity of the gene product and the conserved upstream region with the putative ORF80 binding sites strongly suggest that the orf80 homologue is also functional in this plasmid. Searching the complete genome sequence of S.solfataricus P2 we discovered three further homologues of orf80. The gene SSO8948 has high similarity to the plasmid orf80 genes, whereas the remaining two genes, SSO8758 and SSO5209, are more distantly related. The latter two genes code for proteins that do not possess the highly conserved leucine residues and the upstream region of SSO5209 does not contain the typical ORF80 binding site motif TTAA(N)₇TTAA. It is unknown whether the three genes are functional and what function they may carry out. Given the high plasticity of the Sulfolobus genome (14), it is possible that gene integration events are responsible for the occurrence of orf80-like genes within the genome. SSO8758 and SSO5209 possibly result from an early integration event, whereas SSO8948 integrated more recently. It is a matter of speculation how the host orf80 gene (and a possible heterodimer) could interfere with the plasmid borne orf80 gene.

Neither the distribution of the gene nor homology to known genes enabled us to forecast the function of orf80. In order to understand the plasmid pRN1 at the molecular level we therefore sought to express and purify ORF80. We could show that purified recombinant ORF80 is a sequence-specific double-stranded DNA-binding protein. Of a large number of TaqI restriction fragments derived from pRN1 only one was bound by ORF80 in EMSA. This fragment encompasses the complete gene orf80. Our footprinting analysis narrowed the binding site down to two sites of similar structure upstream of the transcription start site. These sites are highly conserved upstream of all plasmid orf80 genes. Fluorescently labelled short DNA comprising the complete protected region of the binding site were half-maximal bound at 140 nM ORF80. The DNA-binding site is of an unusual structure and symmetry. In addition to two TTAA motifs, two other conserved nucleotides are arranged in a palindromic way. The palindromic structure suggests that an ORF80 oligomer of 2-fold symmetry is the active DNA-binding species. A search of the TFSITE database did not reveal a transcription factor that binds to a bipartite TTAA motif.

The EMSA experiments and the site titrations suggest that a large but specific protein–DNA complex is formed when ORF80 binds to its target DNA. It is possible that two ORF80 dimers could bind to short DNA sequences containing a TTAA motif and assemble into a protein–DNA complex with eight ORF80 subunits and a TTAA(N)₇TTAA motif. Consequently, 16 or more subunits would then bind to a complete binding site with two TTAA(N)₇TTAA motifs. In these protein–DNA complexes probably not all ORF80 molecules can make direct DNA contacts to the rather short DNAs. Given the high aggregation tendency of ORF80 it is not unlikely that some ORF80 molecules are only bound to the complex via protein–protein interaction. It should be stressed, however, that the protein–DNA complex is nonetheless specific and seems to have a defined stoichiometry. Otherwise we would have observed smeared shifted bands in the EMSA experiments. At this point we cannot provide firm experimental data for this binding model but our site titrations suggest that about six subunits are required to saturate a single TTAA motif and that about 12 subunits can bind to a TTAA(N)₇TTAA motif. These values are based on the assumption that the protein is fully active. The protein for these experiments was prepared according to the alternative purification scheme during which the protein was purified in the presence of 6 M urea. As this amount of urea does not unfold the protein substantially (9.7 M urea was needed to reduce the CD signal to 50%; Fig. 9) and as the CD spectrum of ORF80 was reasonable it can be assumed that most of the protein was folded and active in these assays.

What might be the physiological role of ORF80? ORF80 may have a structural and/or regulatory function. One of the ORF80-binding sites is close to the initiator methionine and indeed encompasses the archaeal TATA box. It can therefore be speculated that ORF80 represses its own transcription in an autoregulatory manner. This bacterial-like transcriptional repression has recently been reported for the Lrs14 protein from S.solfataricus (15). However, on plasmids pDL10 and pING the binding sites are more distal to the initiator methionine, and the putative archaeal TATA box of these two genes might not be blocked by binding of ORF80 to its binding site.

The function of the conserved second binding site more upstream of orf80 is even more unclear. In view of the strict conservation of the distance between the two sites and in view of the observed cooperativity in protein binding that leads to formation of a large protein–DNA complex, we assume a specific structural role for these complexes. Possibly the ORF80 complexes impart a distinct structure that is necessary for recruitment of, for example, the replication initiation machinery. Several authors have suggested that the region upstream of orf80 contains the double-stranded origin of replication (2,5). Experimental evidence supporting this hypothesis is, however, still lacking.

In preliminary experiments we asked whether partially purified recombinant ORF904, the presumed replication initiator protein, is able to interact with the ORF80–DNA complex. In EMSA, however, we were unable to demonstrate a supershift of the ORF80–DNA complex in the presence of ORF904 (G.Lipps, unpublished results).

Sequence analysis of the orf80 coding region suggests that dimerization is mediated by a leucine zipper motif. Remarkably, as compared to all other known basic leucine zipper proteins, the basic domain and the leucine zipper are swapped in ORF80. Our present work suggests that both orientations allow specific recognition of DNA. Basic leucine zippers have mostly been found in eukaryal transcription factors. To our knowledge the transcriptional activator GvpE of Halobacterium salinarum has up to now been the only archaeal protein with a basic leucine zipper motif (16). In contrast to GvpE, which is a 21 kDa protein, ORF80 is of smaller size and seems to have no additional binding site for ligands that could regulate its function. ORF80 may therefore represent a minimal model protein for this new subclass of basic leucine zipper proteins.