Promoter Selectivity of the Bradyrhizobium japonicum RpoH Transcription Factors In Vivo and In Vitro

Abstract

Expression of the dnaKJ and groESL₁ heat shock operons of Bradyrhizobium japonicum depends on a ς³²-like transcription factor. Three such factors (RpoH₁, RpoH₂, and RpoH₃) have previously been identified in this organism. We report here that they direct transcription from some but not all ς³²-type promoters when the respective rpoH genes are expressed in Escherichia coli. All three RpoH factors were purified as soluble C-terminally histidine-tagged proteins, although the bulk of overproduced RpoH₃ was insoluble. The purified proteins were recognized by an anti-E. coli ς³² serum. While RpoH₁ and RpoH₂ productively interacted with E. coli core RNA polymerase and produced E. coli groE transcript in vitro, RpoH₃ was unable to do so. B. japonicum core RNA polymerase was prepared and reconstituted with the RpoH proteins. Again, RpoH₁ and RpoH₂ were active, and they initiated transcription at the B. japonicum groESL₁ and dnaKJ promoters. In all cases, the in vitro start site was shown to be identical to the start site determined in vivo. Promoter competition experiments revealed that the B. japonicum dnaKJ and groESL₁ promoters were suboptimal for transcription by RpoH₁- or RpoH₂-containing RNA polymerase from B. japonicum. In a mixture of different templates, the E. coli groESL promoter was preferred over any other promoter. Differences were observed in the specificities of both sigma factors toward B. japonicum rpoH-dependent promoters. We conclude that the primary function of RpoH₂ is to supply the cell with DnaKJ under normal growth conditions whereas RpoH₁ is responsible mainly for increasing the level of GroESL₁ after a heat shock.

The ς subunit confers promoter recognition ability upon bacterial RNA polymerases. Sigma factors specifically interact with the catalytic, four-subunit core RNA polymerase (α₂ββ′). Only after assembly with ς to form the so-called holo-RNA polymerase can the enzyme recognize promoter sequences and initiate transcription accurately. Upon initiation, the ς factor is released and the core enzyme elongates the RNA chain (reviewed in reference 9).

Bacteria often contain multiple sigma factors (for reviews, see references 8, 10, and 11). Under normal growth conditions, transcription of most genes in Escherichia coli is initiated by RNA polymerase containing the major sigma factor ς⁷⁰ (also called ς^D). Alternative sigma factors coordinate transcription of distinct subsets of coregulated genes which, for example, are required for nitrogen assimilation (ς⁵⁴ [ς^N]), for flagellum gene expression (ς²⁸ [ς^F]), during stationary phase (ς³⁸ [ς^S]), or under conditions of heat stress (either ς³² [ς^H] or ς²⁴ [ς^E], depending on whether the stress signal is elicited in the cytoplasm or the periplasm) (17).

We are studying the regulation of heat shock genes in Bradyrhizobium japonicum, the nitrogen-fixing root nodule symbiont of the soybean plant. This organism uses three different strategies to control the expression of heat shock proteins (Hsps) at elevated temperatures. Two mechanisms rely on conserved DNA elements (CIRCE and ROSE) which are located immediately downstream of the transcription start sites of some heat shock operons (2, 18, 20). In both cases, it is likely that a putative repressor protein binds to these sites and prevents transcription under normal growth conditions. Regulation of a third class of heat shock genes is mediated by ς³²-like factors (2, 16). B. japonicum is the first known organism which contains an rpoH gene family coding for three different ς³²-like factors (RpoH₁, RpoH₂, and RpoH₃) (18, 19). Notably, these proteins differ not only in their primary sequence (the number of identical amino acids varies between 49 and 66%) but also in their significance to survival and in the way they are regulated. RpoH₂ appeared to be crucial for survival under all conditions because several attempts to disrupt the corresponding gene failed. By contrast, RpoH₁ and RpoH₃ could be deleted individually. Even the rpoH₁-rpoH₃ double mutant had no obvious growth defect under various conditions.

All three RpoH proteins were capable of producing high levels of β-galactosidase from a groE-lacZ fusion in a ς³²-deficient E. coli strain (19). In fact, rpoH₁ had originally been discovered by searching for this phenotype in a complementation approach (18). A puzzling observation was made when the abilities of the RpoH factors to complement the temperature-sensitive phenotype of the rpoH mutant were compared. Only RpoH₂ reached a similar complementation capacity to E. coli ς³² in allowing the strain to grow at otherwise restrictive temperatures (19). While RpoH₁ still permitted growth at intermediate temperatures, RpoH₃ almost completely failed. Taken together, these results could be taken as a hint of a somewhat overlapping but also divergent promoter specificity of the three RpoH factors of B. japonicum.

In the present study, we set out to systematically test the promoter specificities of the RpoH factors. To complete our previous studies, we checked which Hsps were produced in each complemented E. coli strain. We were particularly interested in establishing an in vitro transcription system to approach this question. By the successful use of purified components, we show that RpoH₁ and RpoH₂ indeed favor different promoters.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

E. coli cells were grown in Luria-Bertani (LB) medium (15) supplemented with ampicillin (200 μg/ml) or kanamycin (30 μg/ml) if required. E. coli A7448 (4, 30) was grown at 28°C; the routine growth temperature for other E. coli strains was 37°C. B. japonicum was grown aerobically at 28°C in PSY medium (22) supplemented with 0.1% (wt/vol) arabinose and 100 μg of spectinomycin per ml.

Plasmid constructions.

Recombinant DNA techniques were performed by standard methods (23). The individual rpoH genes were amplified from plasmids containing these genes (18, 19) by PCR with Vent DNA polymerase as specified by the manufacturer (New England Biolabs, Beverly, Mass.). The oligonucleotides used as primers were designed so that they introduced a NdeI recognition site overlapping the start codon of each gene and either a XhoI site immediately downstream of the 3′ ends of rpoH₁ and rpoH₂ or a NotI site at the equivalent position of rpoH₃. The PCR-generated fragments were cut with NdeI-XhoI or NdeI-NotI and ligated into pET24b(+) (Novagen, ams Biotechnology, Lugano, Switzerland) digested with the same enzymes. The resulting plasmids, pRJ5086, pRJ5101, and pRJ5102, encoded C-terminally histidine-tagged RpoH₁, RpoH₂, and RpoH₃ proteins, respectively. The vector-encoded amino acids leucine and glutamic acid were introduced between the carboxy-terminal amino acid of RpoH₁ or RpoH₂ and the histidine tag. A stretch of five amino acids (AAALE) was introduced at the corresponding position of RpoH₃-His₆.

Plasmids used as transcription templates were based on pRJ9519, which contains the B. japonicum rrn terminator. It derives from pRJ9601 (3) from which the rrn promoter and an internal EcoRI restriction site upstream of the terminator have been deleted. Plasmid pRJ5099 carries a 0.9-kb EcoRI-BglII fragment containing the B. japonicum dnaKJ promoter region (16). The B. japonicum groESL₁ template pRJ5134 contains a BglII-EcoRV-digested 0.4-kb PCR fragment originating from pRJ8067 (2). While the BglII site was present in the sequence upstream of groESL₁, the EcoRV site was introduced into the groES₁ gene by the PCR primer (Sig106, 5′-GGCTTGATATCGATCGGGATC-3′ [recognition site underlined]). Plasmid pEC5162 bears a 0.4-kb E. coli groESL promoter fragment that was PCR amplified from pEC8182 (kindly provided by H.-M. Fischer, Zürich, Switzerland), a subclone of pND5 (12). EcoRI restriction sites at either end of the fragment were created by the PCR primers (Sig122, 5′-GCGAATTCCCTGGGCCAGCCC-3′; Sig123, TCGAATTCTTTACGCTTGACG-3′ [EcoRI sites underlined]).

The correct nucleotide sequences of all PCR-amplified fragments introduced into pET24b(+) or the transcription vector pRJ9519 were confirmed by sequencing. It revealed for pEC5162 that in this particular case the EcoRI site in pRJ9519, which was assumed to be destroyed (see above), was present and had been cut by EcoRI, resulting in a product that was approximately 100 bp shorter than expected.

Overproduction and purification of RpoH-His₆ proteins.

Freshly transformed E. coli BL21/pLys cells (26) carrying pRJ5086, pRJ5101, or pRJ5102 were used for overproduction of RpoH proteins. When the cells had reached an optical density at 600 nm of 0.7 at 28°C, production of the recombinant proteins was induced by the addition of 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG). After 2 h, the cells were harvested and disrupted in a French pressure cell, and proteins were purified from the soluble fraction of a 500-ml cell culture (RpoH₁ and RpoH₂) or 2,000-ml culture (RpoH₃). The extracts were loaded onto a 1.5-ml Ni-nitrilotriacetic acid-agarose column (Qiagen, Hilden, Germany), and chromatography was performed as specified in the pET System Manual (Novagen). Protein fractions were dialyzed against storage buffer (20 mM Tris-HCl [pH 8.0], 200 mM NaCl, 50% glycerol) and then stored at −20°C.

Purification of B. japonicum RNA polymerase core enzyme.

RNA polymerase was prepared as described previously (3). Core RNA polymerase was separated from holoenzyme on a Resource Q column (6 ml; Pharmacia, Uppsala, Sweden). Up to 1.4 mg of RNA polymerase was loaded onto the column equilibrated with T₅₀GED (50 mM Tris-HCl [pH 8.0], 10% glycerol, 1 mM EDTA, 0.1 mM dithiothreitol) containing 0.1 M NaCl. After the column was washed with the same buffer, bound protein was eluted with 170 ml of a linear 0.1 to 0.325 M NaCl gradient. Peak fractions which were devoid of sigma factor (judged by sodium dodecyl sulfate [SDS]-polyacrylamide gel electrophoresis [12% polyacrylamide]) were pooled, dialyzed against T₁₀GED buffer containing 0.02 M NaCl and 50% (vol/vol) glycerol, and concentrated by ultrafiltration.

Western blot (immunoblot) analysis.

Crude extracts of E. coli cells were prepared, separated on SDS–12% polyacrylamide gels, and transferred to nitrocellulose membranes as described previously (18). The following antisera were used: anti-E. coli ς³², DnaK, and GrpE sera (B. Bukau, Freiburg, Germany; 3,000-fold dilutions [7]); anti-E. coli GroEL serum (Epicentre Technologies, Madison, Wis.; 20,000-fold dilution); and anti-E. coli IbpA serum, which recognizes the small Hsps IbpA and IbpB (A. Easton, St. Louis, Mo.; 1,500-fold dilution [1]). The primary rabbit antibodies were detected with a chemiluminescence Western blotting kit (Boehringer GmbH, Mannheim, Germany).

In vitro transcription.

Multiple-round transcription assays with RNA polymerase holoenzyme were carried out in a volume of 20 μl under standard conditions as described previously (3). RNA polymerase core enzyme (1.15 μg of B. japonicum enzyme or 1 U of E. coli enzyme [Boehringer Mannheim] per assay) was reconstituted with RpoH factors (0.1 μg per assay; molar ratio of core RNA polymerase and RpoH protein, 1:1) for 20 min at 4°C in a volume of 10 μl containing 40 mM Tris-HCl (pH 8.0), 10 mM MgCl₂, 0.1 mM EDTA, 0.1 mM dithiothreitol, 150 mM KCl, 0.4 mM K₃PO₄, and 0.2 mg of bovine serum albumin per ml. The reaction was started by adding the reconstituted enzyme to a 10-μl sample containing the same components as the reconstitution buffer plus 1 mM of each nucleoside triphosphate, 1 μCi of [α-³²P]UTP (800 Ci/mmol; DuPont, Bad Homburg, Germany), 1 U of RNase inhibitor per μl, and 20 nM DNA template (final concentrations).

For single-round transcription experiments, the DNA template was added to the reconstitution sample before the remaining components together with 100 ng of heparin per ml (final concentration) were added. The template concentration in promoter competition experiments was 10 nM for each plasmid.

Suitable RNA size markers were synthesized in vitro with T7 or T3 RNA polymerase with linearized pBluescript-based plasmids as templates.

Transcript mapping.

RNA isolation and primer extension analysis was performed as described elsewhere (2). The start site of transcripts synthesized in vitro was determined as described previously (3). The oligonucleotides 702 (2) and DnaK12 (16) were used to determine the start sites of in vitro-synthesized B. japonicum groESL₁ or dnaK transcripts. Oligonucleotide Sig123, which was used to amplify the promoter region, was also used to determine the in vivo and in vitro start site of the E. coli groESL operon.

RESULTS

Plasmid-encoded RpoH proteins of B. japonicum selectively transcribe heat shock promoters in E. coli.

We have previously described that all three RpoH proteins of B. japonicum were functionally active in a ς³²-negative E. coli reporter strain which contains a chromosomally integrated E. coli groE-lacZ fusion (19). This strain produced high β-galactosidase activity when it was complemented with any of the three plasmids harboring an rpoH gene under the control of the lac promoter. We now monitored the ability of these complemented strains to produce heat shock proteins in E. coli. To this end, we performed immunoblot analyses with antibodies raised against different Hsps. Extracts from normally grown and heat-shocked cells of E. coli MC4100 and the ς³²-negative strain containing pUC18 without insert served as controls. The typical heat-inducible synthesis of the Hsps GroEL, DnaK, GrpE, and IbpA plus IbpB was observed in the wild type but not in the mutant (Fig. 1). As expected from the activation of the groE-lacZ fusion, all strains complemented by rpoH produced high levels of GroEL. However, the amounts of other Hsps in these strains were quite different. While E. coli ς³² and B. japonicum RpoH₂ produced comparable, high levels of DnaK, GrpE, and small Hsps, RpoH₁ and in particular RpoH₃ produced only small or undetectable amounts of DnaK and GrpE. This result indicated that the sigma factors conferred different promoter specificities upon E. coli core RNA polymerase.

FIG. 1.

Detection of heat shock proteins produced by E. coli MC4100 (WT) and E. coli A7448 (rpoH⁻). E. coli A7448 transformed with pUC18 (−), pRJ5000 (rpoH₁, indicated as 1), pRJ5002 (rpoH₂, indicated as 2), pRJ5040 (rpoH₃, indicated as 3), or pEC5007 (E. coli rpoH, indicated as E) was grown in the presence of 0.5 mM IPTG. Crude cell extracts were prepared from cells grown at 28°C (0) or heat-shocked cells (heat shock [HS] from 28 to 43°C for 5 or 10 min as indicated) and separated on a SDS–12% polyacrylamide gel. A Coomassie blue-stained gel is shown at the top. The apparent molecular masses (in kilodaltons) of relevant marker proteins are indicated to the left of the gel. Similar gels with suitably diluted samples (1/5, 1/10, 1/3, and 1/15 of the amounts loaded onto the stained gel) were used for immunodetection of the Hsps DnaK, GroEL, GrpE, and IbpA plus IbpB, respectively.

The B. japonicum RpoH factors use the same in vivo start site as E. coli ς³².

Since all four RpoH factors synthesized GroEL protein efficiently, we analyzed whether the heterologous proteins initiated transcription at the same position as E. coli ς³². Total RNA was isolated from each complemented strain, and the 5′ end of the groESL transcript was determined by primer extension. This strain contains two groESL promoter regions: the original operon and a chromosomally integrated groE-lacZ fusion. Transcripts of both operons were detected by the primer used in these experiments. The transcripts formed by all four RpoH proteins start at the identical position, which corresponds to the previously described ς³²-dependent site (Fig. 2) (5). Small amounts of a shorter product probably represent prematurely terminated reverse transcripts. RNA from the mutant containing only the vector resulted in a faint signal that represents transcripts originating from the ς⁷⁰-dependent promoter (Fig. 2) (30). This transcript was hardly observed with RNA from the rpoH⁺ strains, presumably because the abundant RpoH proteins prevented transcription from the ς⁷⁰ promoter by promoter occlusion. It has been shown that this promoter is very weak and virtually inactive in rpoH⁺ strains (14, 30).

FIG. 2.

Primer extension analysis of groESL transcripts synthesized in E. coli A7448. Reverse transcripts were obtained with 15 μg of total RNA isolated from the strain complemented with B. japonicum rpoH₁ (lane 1), rpoH₂ (lane 2), or rpoH₃ (lane 3) or E. coli rpoH (lane E) and 30 μg of RNA from the same strain transformed with pUC18 (lane −). Oligonucleotide Sig123 was used as the primer for the primer extension and corresponding sequencing reactions (lanes T, C, G, and A). The transcription start site used by the ς³²-like proteins is indicated by a solid arrow; the open arrow marks the transcript initiated at the ς⁷⁰-dependent promoter.

Overexpression and purification of histidine-tagged RpoH₁, RpoH₂, and RpoH₃.

Knowing that amino-terminally or carboxy-terminally histidine-tagged ς³² of E. coli is active in vivo and in vitro (7), we constructed plasmids to produce the B. japonicum RpoH proteins with a histidine tag attached to their C termini. Crude cell extracts containing these proteins were prepared and separated into pellet and supernatant fractions (Fig. 3). Surprisingly, the overproduction efficiency and the solubility of the RpoH proteins varied over a wide range. RpoH_2His and RpoH_3His were overproduced to the highest levels. However, more than 95% of RpoH_3His was found in the pellet fraction, indicating that the majority of this protein was insoluble. By contrast, approximately 50% of overproduced RpoH_1His and RpoH_2His remained in the supernatant. All three proteins were purified from the soluble supernatant fraction by nickel-nitrilotriacetic acid chromatography. The proteins exhibited different binding affinities to the column resin. While some RpoH_2His had already eluted during the wash step with an imidazole concentration of 50 mM, RpoH_1His and RpoH_3His were still retained at this concentration (data not shown). The purification profile for RpoH_2His (Fig. 4A) serves as an example to show that the nickel affinity chromatography technique allowed efficient purification of histidine-tagged B. japonicum RpoH proteins.

FIG. 3.

Solubility of overproduced B. japonicum RpoH proteins. The supernatant (Sup) and pellet (Pellet) fraction of crude cell extracts prepared before (−) or after (+) IPTG induction were separated by centrifugation at 15,000 rpm (SS-34 rotor) at 4°C for 30 min. Control cells contain the vector pET24b(+) without insert. RpoH_1His, RpoH_2His, and RpoH_3His were produced from plasmids pRJ5086, pRJ5101, and pRJ5102, respectively. After electrophoresis in an SDS–12% polyacrylamide gel, the proteins were stained with Coomassie blue. The apparent molecular masses (in kilodaltons) of marker proteins are indicated to the left of the gel. Arrows point to the overproduced RpoH_His protein.

FIG. 4.

Purification of RpoH_2His and immunodetection of B. japonicum RpoH_His proteins. (A) Aliquots of the RpoH_2His supernatant fraction which was loaded onto the column (load), the flowthrough (flow), the different wash (wash), and the elution (eluate) fractions were separated on an SDS–12% polyacrylamide gel, and the proteins were stained with Coomassie blue. The imidazole concentration in the wash and elution buffers is indicated. The apparent molecular masses (in kilodaltons) of the marker proteins are indicated to the left of the gel. The arrow points to the protein fraction which was subsequently used for in vitro experiments. (B) Coomassie blue-stained SDS–12% polyacrylamide gel of 100 ng of each purified B. japonicum RpoH_His protein. The apparent molecular masses of two marker proteins are indicated to the right of the gel. (C) Immunodetection of purified B. japonicum RpoH proteins with anti-E. coli ς³² serum. For each protein, a 5-ng sample was subjected to Western blot analysis.

In previous work, we showed that RpoH₁ or RpoH₂ produced in E. coli A7448 was detectable by an anti-E. coli ς³² serum in an immunoblot (18, 19) but that RpoH₃ did not consistently react with the antiserum. By using the purified His-tagged proteins (Fig. 4B), we could now show that this serum also recognizes RpoH_3His but clearly does so less well than it recognizes the other proteins (Fig. 4C).

Multiple-round in vitro transcription assays.

In a first attempt to test whether the purified RpoH proteins were transcriptionally active in vitro, we simulated the situation in the E. coli rpoH mutant in which the proteins had been shown to function in vivo. E. coli core RNA polymerase was reconstituted with the individual RpoH_His proteins, and the ability to transcribe the E. coli groESL promoter was monitored in a multiple-round transcription assay. The core enzyme alone produced small amounts of transcript (Fig. 5A), most probably as a result of some residual ς³² protein in the preparation that was detectable by Western blot analysis (data not shown). Transcription was stimulated by the addition of RpoH_1His and in particular by RpoH_2His. RpoH_3His appeared to be inactive because the reconstituted enzyme did not produce more transcript than core RNA polymerase alone. Since RpoH₃ was active at the E. coli groESL promoter in vivo (compare Fig. 1) we conclude that the purified protein was in an inactive state, probably because the small portion of soluble protein was in an inactive conformation. Transcriptional activity of RpoH₁ and RpoH₂ required the presence of RNA polymerase because the sigma factors alone were unable to synthesize transcript. As positive controls, we used E. coli and B. japonicum holo-RNA polymerases, both of which contain significant amounts of RpoH protein as revealed by Western blot analysis (data not shown). The transcripts produced by these enzymes were indistinguishable from the ones formed by the reconstituted enzymes, indicating that transcription had been initiated at the same position (Fig. 5A).

FIG. 5.

Transcription of the E. coli groESL promoter by E. coli RNA polymerase core enzyme and B. japonicum RpoH factors. (A) The enzyme combinations tested are indicated. Core, E. coli RNA polymerase core enzyme; RpoH, purified RpoH_His protein; Ec holo, E. coli RNA polymerase holoenzyme; Bj holo, B. japonicum RNA polymerase holoenzyme. (B) Primer extension analysis of in vitro-synthesized RNA. The enzyme with which the transcript was generated is indicated. The primer extension and sequencing reactions (TCGA) were performed with oligonucleotide Sig123. The transcription start site is depicted by an arrow.

To confirm that the reconstituted enzymes use the same start site as in vivo, we determined the 5′ end of in vitro-synthesized RNA by primer extension analysis (Fig. 5B). It was found to be identical to the start site of the ς³²-dependent groESL (5) (compare Fig. 2). The presence of comparable faint signals produced by RNA polymerase core enzyme and RpoH₃-containing enzyme again reflected the notion that the latter was not active.

Next we tested a physiologically more relevant combination by reconstituting B. japonicum core RNA polymerase with the purified RpoH factors and transcribing the B. japonicum groESL₁ and dnaKJ promoters, which are dependent on a ς³²-type factor in vivo (2, 16). Again, RpoH_3His did not stimulate transcription (data not shown), but transcription from both promoters was initiated when RNA polymerase had been reconstituted with RpoH_1His or RpoH_2His (Fig. 6A and B). The groESL₁ and dnaKJ transcripts are of the expected lengths (401 and 358 nucleotides, respectively), as judged from the RNA size marker. To demonstrate that these transcripts had been initiated at the same site as in vivo, in vitro-synthesized RNA was subjected to primer extension analysis. The determined 5′ ends of the groESL₁ and dnaKJ transcripts (Fig. 6C and D, respectively) were mapped to the same positions as described previously (2, 16).

FIG. 6.

Transcription of the B. japonicum groESL₁ promoter and the dnaKJ promoter by B. japonicum RNA polymerase reconstituted with purified RpoH factors. (A) Transcription of the B. japonicum groESL₁ promoter. The enzyme combinations tested are indicated. Core, B. japonicum RNA polymerase core enzyme; RpoH, purified RpoH_His protein. The arrow points to an RNA size marker of 406 nucleotides. (B) Transcription of the B. japonicum dnaKJ promoter. The enzyme combinations are as indicated in panel A. (C) Primer extension analysis of in vitro-synthesized groESL₁ transcript. Oligonucleotide 702 was used for the sequencing (TCGA) and primer extension reaction. The triangle points to the start site, which has been determined in vivo (2). The origin of small amounts of slower-migrating bands is not known because corresponding products were not observed after in vitro transcription. (D) Primer extension analysis of in vitro-synthesized B. japonicum dnaKJ transcript. Oligonucleotide DnaK12 was used for the sequencing (TCGA) and primer extension reaction. The 5′ end of the reverse transcript (indicated by a triangle) corresponds to the major start site found in vivo (16).

Single-round in vitro transcription assays.

To gain more insight into possible promoter specificities of RpoH₁ and RpoH₂, it was important to establish a single-round transcription system. B. japonicum core RNA polymerase was reconstituted with RpoH_2His, and transcription at five different promoters was assayed under conditions which allowed for the production of only one transcript by each preformed RNA polymerase-template complex. First we tested the E. coli groESL promoter, which was transcribed very well by this enzyme combination (Fig. 7A). The B. japonicum dnaKJ and groESL₁ promoters were also transcribed specifically (Fig. 7B). As a control, we examined the hspA rpoH₁ and rrn promoters, which are expected to be transcribed by B. japonicum ς⁸⁰, the ς⁷⁰ homolog (3, 18). The corresponding products would be 625 and 240 nucleotides long, respectively. Neither promoter was transcribed by the RNA polymerase containing RpoH_2His, indicating that this sigma factor specifically recognizes ς³²-like promoters.

FIG. 7.

Single-round transcription of different promoters by B. japonicum core RNA polymerase reconstituted with RpoH_2His. The E. coli (A) or B. japonicum (B) promoters which were transcribed are indicated. Experiments were performed with core RNA polymerase alone (−) and the same enzyme reconstituted with RpoH_2His (+). Arrows point to the transcripts; the position and length (in nucleotides) of RNA size markers are indicated to the right.

Promoter competition experiments.

To further investigate the promoter specificities of RpoH₁ and RpoH₂, we performed experiments in which an equimolar mix of different templates was provided for transcription. The first mix contained a composite of all five promoters which had been tested individually in the last experiment. B. japonicum core RNA polymerase was reconstituted with RpoH_1His and RpoH_2His, and transcription was assayed under single-round conditions (Fig. 8). As suggested by the experiment in Fig. 7, RNA polymerase containing RpoH_2His preferentially transcribed from the E. coli groESL promoter. (The preference for the E. coli promoter may even be underestimated, because the E. coli groESL transcript contains 50 uracils that can be radiolabeled whereas the B. japonicum dnaKJ and groESL₁ transcripts contain 81 and 80 uracils, respectively.) Transcription from the B. japonicum ς³²-type promoters was much lower with the dnaKJ promoter, yielding slightly more transcript than the groESL₁ promoter. RNA polymerase containing RpoH_1His also transcribed the E. coli groESL promoter to the highest level. Small amounts of the groESL₁ transcript and very little dnaKJ transcript were obtained, which implies that RpoH₁ and RpoH₂ have different promoter specificities. The hspA rpoH₁ and rrn promoters that were also present in the mixture were not transcribed by the reconstituted enzymes.

FIG. 8.

Single-round promoter competition experiment with B. japonicum core RNA polymerase reconstituted with RpoH_1His or RpoH_2His. The E. coli groESL promoter and the B. japonicum dnaKJ, groESL₁, hspA rpoH₁, and rrn promoters were provided at equimolar concentrations (10 nM each; the template concentration is not limiting in this assay). The transcripts obtained are indicated.

After this initial comparative assessment of the promoter specificities conferred by the different RpoH factors, we performed another template competition experiment in which only B. japonicum promoters were provided. Not only B. japonicum core but also holo-RNA polymerase was mixed with the individual purified RpoH factors. The experiment with holoenzyme showed that the added sigma factors can associate with core enzyme present in this preparation (Fig. 9). As expected, the ς⁸⁰ holoenzyme produced some hspA rpoH₁ and rrn transcript and also an 80-nucleotide reference transcript that is encoded by each vector carrying one of the promoter fragments (3). These three transcripts were not obtained when core polymerase was reconstituted with RpoH proteins. The experiments with core enzyme and with holoenzyme confirmed the different promoter specificities of RpoH₁ and RpoH₂. While the latter enzyme transcribed the groESL₁ and dnaKJ promoter well and the dnaKJ promoter slightly better than groESL₁, the RpoH₁-containing RNA polymerase produced less transcript and favored the groESL₁ promoter (Fig. 9; see also Fig. 8).

FIG. 9.

Single-round promoter competition experiment with B. japonicum RNA polymerase core enzyme and holoenzyme reconstituted with purified RpoH proteins. The B. japonicum dnaKJ, groESL₁, hspA rpoH₁, and rrn promoters were provided at equimolar concentrations (10 nM each). The corresponding transcripts are indicated. The vector-encoded reference transcript (ref) was produced only in the presence of the holoenzyme.

DISCUSSION

Different promoter specificities of B. japonicum RpoH factors.

After the previous identification of three disparately regulated rpoH genes in B. japonicum (18, 19), we were left with the question of why this organism uses a gene family for functions that in other bacteria (e.g., E. coli) are accomplished by a single rpoH gene. In principle, an organism can use two different strategies to achieve coordinate regulation of a gene under different environmental conditions: (i) The gene can be put under the control of different promoters, or (ii) several copies of this gene may be present, each having a distinctly regulated promoter. A paradigm for the differential regulation of reiterated genes is the groESL multigene family of B. japonicum. The expression of five groESL operons appears to be controlled by four mechanisms, two of which do not contribute to chaperonin production under heat shock conditions (low constitutive expression of groESL₂ and NifA-dependent expression of groESL₃) (6). Heat-induced expression of groESL genes depends on two separate mechanisms, namely, RpoH factor(s) for groESL₁ and CIRCE (controlling inverted repeat of chaperone expression) (31) for groESL₄ and groESL₅ (2). The rpoH genes of B. japonicum were also defined as a disparately regulated gene family. Transcription of rpoH₁, for instance, is heat inducible by a novel mechanism (18, 20), whereas rpoH₂ is transcribed constitutively from a ς⁷⁰-type promoter under normal growth conditions and induced from a ς^E-type promoter under severe heat shock conditions (19).

However, the option to differentially regulate each member does not seem to be the only reason for the presence of an rpoH multigene family in B. japonicum. The situation becomes more complex because the individual rpoH genes are not functionally equivalent and have (at least to some extent) different promoter specificities in vivo and in vitro. The finding that all RpoH proteins were capable of synthesizing GroEL in an E. coli reporter strain agrees well with their previously shown activation of a groE-lacZ fusion. The limited capacity of RpoH₁ and RpoH₃ to produce other Hsps, especially GrpE, provides a possible answer to why these proteins were insufficient in complementing the temperature-sensitive phenotype of an E. coli ς³² mutant (19).

The promoter selectivity of RpoH₁ and RpoH₂ was further corroborated by in vitro studies. Purified RpoH_3His protein appeared to be inactive because it showed very little (if any) activity at all promoters tested. RpoH_3His also was much less soluble than RpoH_1His and RpoH_2His. Although the small fraction of soluble protein was purified from the crude supernatant, it may still be in an aggregated and transcription-incompetent form. By contrast, RpoH_1His and, in particular, RpoH_2His were highly active in transcribing ς³²-dependent promoters. Much to our surprise, the endogenous B. japonicum groESL₁ and dnaKJ promoters were transcribed less well than the heterologous E. coli groESL promoter. This raises the question why the known ς³²-type promoters of B. japonicum contain a suboptimal promoter sequence. It is particularly puzzling because a heat shock promoter specific for the alpha subdivision of the proteobacteria (to which B. japonicum belongs) was proposed on the basis of a comparison of groESL and dnaKJ promoters from different members of this subdivision (16, 24). The consensus sequence differs from the typical ς³²-dependent heat shock promoter of E. coli at several positions. However, a comparative functional analysis of heat shock promoters from the alpha and gamma proteobacteria has not been performed yet. Our first attempt in this direction implies that the heat shock promoters of the alpha proteobacteria are not necessarily optimized for recognition by the endogenous sigma factors. In keeping with this finding are the results of a mutational analysis of the ς³²-dependent P1 promoter of the Caulobacter crescentus rpoH gene, which is autoregulated (29). This promoter also is not optimal, and its activity can be increased substantially by a particular point mutation in the −35 region. Generally, an efficient expression of heat shock genes under normal growth conditions is not required and may even be deleterious. It has been reported for E. coli that cells overproducing Hsps are sick (27). Thus, rather weak promoters may have evolved to reduce expression. After a heat shock, the abundant RpoH protein (in B. japonicum mainly RpoH₁) may override the weak promoter activities.

The purified RpoH factors directed transcription exclusively from ς³²-type promoters. A similar strict promoter dependence was also described for the corresponding enzymes of E. coli and C. crescentus (5, 28) with one notable exception: the P1 promoter of rrnB, one of seven rrn operons in E. coli, can be transcribed from RNA polymerase containing either ς⁷⁰ or ς³² (21). This is thought to be important for ribosome synthesis at high temperatures. Neither the in vitro transcription performed in this work nor the earlier characterization of the rrn operon of B. japonicum (3, 13) gave any indication that a similar situation applies here. This is particularly surprising because B. japonicum contains only a single rRNA operon.

Function of RpoH₁ and RpoH₂ in B. japonicum.

On the basis of results obtained previously it was proposed that RpoH₂ is essential for the synthesis of ς³²-dependent proteins under normal growth conditions whereas RpoH₁ provides their synthesis under stress conditions (19). The function of RpoH₃ remains obscure. Combining all data known at present, we arrive at the following, refined model.

(i) Normal growth conditions.

Although the rpoH₂ gene is transcribed to a considerable extent under these conditions, very little RpoH protein is detectable in cell extracts (18, 19). The corresponding protein is probably unstable during normal growth, as is E. coli ς³² (25). The observation that neither the rpoH₂ nor the dnaK gene could be deleted had already suggested a direct dependence of the DnaK chaperone, which is required under all conditions, on this sigma factor (16, 19). The in vitro transcription data confirm that RpoH₂ directs the transcription of the dnaKJ promoter efficiently. The groESL₁ promoter could also be transcribed in vitro, but this does not seem to play a role in vivo because the groESL₁ transcript is almost undetectable under normal growth conditions (2).

(ii) Heat shock conditions.

A temperature upshift greatly stimulates the production of RpoH₁ in B. japonicum (18, 19). The parallel induction of the groESL₁ transcript (2) indicated that it occurred as a result of an increase in RpoH₁ production. The in vitro activity of RpoH₁ supports this notion. By comparison, the dnaKJ promoter was weakly transcribed, which raises the question how heat induction of DnaK is accomplished. Since maximal induction of this protein is already reached a few minutes after a temperature upshift (16), it is highly possible that RpoH₂ is responsible for this increase. The RpoH₂ level in the cell increases in the initial phase after a heat shock before it rapidly decreases (19). Under continuous stress conditions, the weak activity of RpoH₁ may then suffice to keep the DnaK concentration at an elevated level.