Interaction of CbbR and RegA* transcription regulators with the Rhodobacter sphaeroides cbbIPromoter-operator region.

The form I (cbb(I)) Calvin-Benson-Bassham (CBB) reductive pentose phosphate cycle operon of Rhodobacter sphaeroides is regulated by both the transcriptional activator CbbR and the RegA/PrrA (RegB/PrrB) two-component signal transduction system. DNase I footprint analyses indicated that R. sphaeroides CbbR binds to the cbb(I) promoter between -10 and -70 base pairs (bp) relative to the cbb(I) transcription start. A cosmid carrying the R. capsulatus reg locus was capable of complementing an R. sphaeroides regA-deficient mutant to phototrophic growth with restored regulated synthesis of both photopigments and ribulose-bisphosphate carboxylase/oxygenase (Rubisco). DNase I footprint analyses, using R. capsulatus RegA*, a constitutively active mutant version of RegA, detected four RegA* binding sites within the cbb(I) promoter. Two sites were found within a previously identified cbb(I) promoter proximal regulatory region from -61 to -110 bp. One of these proximal RegA* binding sites overlapped that of CbbR. Two sites were within a previously identified promoter distal positive regulatory region between -301 and -415 bp. Expression from promoter insertion mutants showed that the function of the promoter distal regulatory region was helical phase-dependent. These results indicated that RegA exerts its regulatory affect on cbb(I) expression through direct interaction with the cbb(I) promoter.

The nonsulfur purple bacterium Rhodobacter sphaeroides is capable of both dark aerobic chemoheterotrophic growth and anoxic photosynthetic growth. Under photoautotrophic growth conditions, where CO 2 functions as the sole carbon source, the Calvin-Benson-Bassham (CBB) 1 reductive pentose phosphate cycle provides nearly all cellular carbon. Photosynthetic growth in the presence of fixed carbon sources (i.e. photoheterotrophic growth) changes the primary role of the CBB cycle. Under these growth conditions, the CBB cycle facilitates the use of CO 2 as an electron sink and terminal electron acceptor for reducing equivalents generated by carbon oxidation and photosynthesis (1). In R. sphaeroides, control of CBB cycle gene (cbb) expression is achieved through the regulated expression of two major cbb operons, denoted form I (cbb I ) and form II (cbb II ) (2,3). These operons are located on separate genetic elements in this organism (4,5). The cbb I operon comprises structural genes that encode CBB cycle enzymes, including fructose 1,6-sedoheptulose 1,7-bisphosphatase (cbbF I ), phosphoribulokinase (cbbP I ), fructose 1,6-sedoheptulose 1,7bisphosphate aldolase (cbbA I ), as well as the large and small subunit genes of form I (L 8 S 8 ) ribulose-bisphosphate carboxylase/oxygenase (Rubisco) (cbbL I cbbS I ) (6). The cbb II operon encodes homologs of cbbF I , P I , and A I as well as the genes for transketolase (cbbT II ), glyceraldehyde-3-phosphate dehydrogenase (cbbG II ) and the large subunit of the form II-type Rubisco (cbbM II ) (7). Studies over the years have shown that the regulation of cbb gene expression in R. sphaeroides is complex (2). Expression of both the cbb I and cbb II operons is quite low under dark aerobic chemoheterotrophic conditions. However, under photosynthetic growth conditions, expression of the genes from both operons is derepressed, with each operon responding independently to a number of environmental parameters such as the level of CO 2 and the reduction state of organic carbon compounds supplied for growth (8 -12). In general, growth under photoheterotrophic conditions, with a fixed (organic) carbon source, results in an excess of cbb II expression over cbb I . Maximal expression from both operons is observed under photoautotrophic conditions; i.e. when CO 2 is used as the sole source of carbon, with cbb I operon expression exceeding that for the cbb II operon (12). In addition to this apparent independent control of cbb I and cbb II gene expression, some form of interdependent control also exists, manifest by a compensatory increase in the expression of one operon upon the inactivation of the other (6,8,10,11). The cbbR gene, which is located immediately upstream and in the opposite orientation to cbbF I (13), mediates this compensatory effect, with the product of the cbbR gene shown to positively regulate the expression of both the cbb I and cbb II operons (13,14). Superimposed on the requirement for cbbR is the regA-regB (prrA-prrB) twocomponent regulatory system, encoding sensor kinase RegB (PrrB) and response regulator RegA (PrrA). This system has also been shown to play a role in cbb regulation, based primarily on genetic studies in which a R. sphaeroides regB insertion mutant was found to exhibit reduced cbb I and cbb II expression during photoautotrophic growth in a 1.5% CO 2 /98.5% H 2 atmosphere (15). Because the reg (prr) genes had originally been shown to regulate the anaerobic activation of operons encoding structural genes of the photosynthetic reaction center (puh) and light-harvesting complexes (puf and puc) of both R. capsulatus (16,17) and R. sphaeroides (18), it was most surprising to find that this same two-component system controls cbb expression. In addition, the reg genes, and close homologs, have also been recently shown to be involved in the regulation of nitrogen fixation in R. sphaeroides (19,20) and Bradyrhizobium japonicum (21) as well as in controlling an operon involved in the oxidation of formaldehyde (22). Current investigations are directed at elucidating mechanisms of gene regulation and cbb gene expression, with the R. sphaeroides cbb I operon serving as the primary model system for our studies on CO 2 fixation. Previous work, using cbb I ::lacZ promoter fusions, indicated that the cbb I promoter is comprised of a promoter proximal region (Ϫ100 to ϩ1 base pairs (bp)) that is sufficient to confer low level and CbbR-dependent regulated expression of cbb I (14). CbbR was shown to bind to this region of the promoter in gel mobility shift assays. An additional promoter distal upstream activating region was identified, between Ϫ280 and Ϫ636 bp that significantly enhanced cbb I expression under all growth conditions tested. To further characterize the regulatory mechanism and to identify CbbR binding site(s) within the cbb I promoter region, in vitro studies with purified CbbR are required. Moreover, germane to our studies on cbb control was the isolation of an R. capsulatus regA mutant that showed increased photosynthesis gene expression under both aerobic and anaerobic growth conditions (23). This mutant encodes a RegA protein (RegA*) that exerts its effect independent of the cognate sensor kinase, RegB. Moreover, RegA* was shown to possess enhanced DNA binding activity relative to wild-type RegA. The increased DNA binding activity of RegA* made it possible to demonstrate the direct interaction of RegA* with discreet sites within the puc and puf promoters (23). Because the interaction of RegA with the cbb system has not been investigated beyond the physiological and genetic studies previously described (15), the potential to use purified RegA* for detailed in vitro studies was attractive. Thus, in this communication, we first identified binding sites for CbbR within the cbb I promoter proximal regulatory domain using DNase I footprinting experiments. Furthermore, after demonstrating the functional complementation of an R. sphaeroides regA (prrA) insertion mutant with R. capsulatus regA, in vitro studies with purified RegA* firmly established the presence of RegA* binding site(s) within the cbb I promoter-operator region. RegA* binding sites were found within both the cbb I promoter proximal regulatory region and the promoter distal upstream activating region. Moreover, the function of the promoter distal upstream activating region was found to be dependent on proper helical phasing with respect to downstream promoter elements. These results suggest that RegA interacts with the cbb I promoter at both the promoter proximal region (overlapping the CbbR binding site) and the previously described upstream activating region (14). A DNA-looping mechanism for cbb I activation is discussed.
Spectral Analysis of Photopigments and ␤-Galactosidase Assays-Cultures were first grown to late exponential phase under chemoheterotrophic conditions. The cells were harvested, washed in minimal media, and transferred to either photoautotrophic or carbon starvation growth conditions for 72 h. Sonicated extracts of R. sphaeroides CAC strains were generated in a buffer containing 10 mM Tris, pH 8.0, 1 mM EDTA, and 5 mM ␤-mercaptoethanol. Samples were diluted to a protein concentration of 0.1 mg/ml, and the absorbance was scanned from 400 to 900 nm using a Beckman DU-70 spectrophotometer. Total protein was determined with a protein assay dye-binding reagent (Bio-Rad, Hercules, CA). ␤-Galactosidase assays were performed as described previously (14).
Western Immunoblot Analysis-Proteins were resolved by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore, Bedford, MA) using a Bio-Rad Transblot semidry cell, as directed by the manufacturer. The blots were probed with antibodies specific for either R. sphaeroides form I or form II Rubisco (8), developed using the Vistra ECF fluorescent detec- , and chemoheterotrophic (C) conditions. In A and B, each strain was grown chemoheterotrophically to mid-exponential phase and then transferred to minimal media as described under "Experimental Procedures" and incubated for 72 h in the light. In C, each strain was grown under chemoheterotrophic conditions. tion system (Amersham Pharmacia Biotech) and visualized using a Storm 840 phosphorimaging system (Molecular Dynamics, Sunnyvale, CA). (27) carrying the CbbR expression plasmid, pET11R-11, were prepared as described previously (14). Extracts prepared in this way were unsuitable for use in DNase I footprinting experiments due to the presence of significant levels of phosphatase. To remove phosphatase and other potential interfering substances, CbbR was purified by ion exchange chromatography using high performance Q-Sepharose (Amersham Pharmacia Biotech). CbbR was eluted with a 0.1-1.5 M KCl gradient in buffer B (10 mM Tris-HCl, pH 8.5, 300 mM potassium glutamate, 1 mM dithiothreitol, 30% glycerol) (14). Fractions containing CbbR were pooled and subjected to gel filtration chromatography with a Superose 6 (Amersham Pharmacia Biotech) column (1 x 30 cm) using buffer B. The fractions containing CbbR were pooled and then dialyzed against buffer B and stored at Ϫ20°C.

Synthesis of Recombinant R. sphaeroides CbbR in E. coli and Preparation of Extracts-Extracts
Purification of RegA* and RegBЉ-RegA* was purified from E. coli strain BL21(DE3) carrying the plasmid pET29CBD::regA* using a method previously described (23).
DNaseI Footprint Analysis-Probes for DNase I footprint analyses were prepared by polymerase chain reaction (PCR) amplification of selected regions of the cbb I operon promoter of R. sphaeroides, using p12EH (13) as a template. Selective labeling of DNA strands was performed by 5Ј-end labeling of one of the oligonucleotide primers with T4 polynucleotide kinase (New England BioLabs, Beverly, MA) and [␥-32 P]ATP (7000 Ci/nmol, ICN Biochemicals, Costa Mesa, CA) prior to amplification. The PCR reactions consisted of 1 mol of each labeled and unlabeled oligonucleotide primer, 28 ng of template DNA and 2.5 units of Taq polymerase (Life Technologies, Inc., Gaithersburg, MD). Amplification was performed as follows: denaturation at 95°C for 5 min, then 30 cycles of 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C. A final elongation step was performed for 7 min at 72°C. PCR fragments were purified on nondenaturing polyacrylamide gels followed by electroelution. The isolated probe DNA was then ethanol-precipitated and resuspended in 100 l of buffer containing 50 mM HEPES (pH 8.0) and 100 mM sodium acetate.
Construction of cbb I ::lacZ Promoter Fusion Plasmids Containing Small Insertions-In general, insertions in the cbb I promoter were introduced into the XhoI site of pUCC1, which carries a 720-bp EcoRI/ BamHI fragment from the cbb I ::lacZ translational promoter fusion pVKC1 (14). This fragment spans 664 bp 5Ј to the cbbF I translation start, as well as 56 bp of the cbbF I coding sequence and contains the complete cbb I promoter. EcoRI/BamHI fragments containing insertions were then ligated into EcoRI/BamHI-digested pMC1403. The pMC1403 derivatives were digested with EcoRI and ligated into the EcoRI site of the conjugative plasmid pVK101 (30). The plasmid pVKXFI was constructed by digestion of pUCC1 with XhoI followed by filling in the overhanging ends using Klenow polymerase and deoxynucleoside triphosphates to generate a 4-bp insertion. In the construction of 10-bp (pVKH10), 15-bp (pVKH15), and 21-bp (pVKH21) insertions, oligo pairs were synthesized such that, when they were annealed, each contained a central BglII site and single stranded, XhoI complementary, overhanging ends. The annealed pairs were ligated into XhoI-digested pUCC1. The resulting insertion derivatives were then digested with BglII, and the overhanging ends were blunted using mung bean nuclease (Life Technologies, Inc.). The sequences of the oligonucleotide pairs used in the construction of the cbb I promoter insertions are as follows: pVKH10, 5Ј-TCGACCAGATCTAG-3Ј and 5Ј-TCGACTAGATCTGG-3Ј; pVKH15, 5Ј-TCGACCTGAGATCTGTCAG-3Ј and 5Ј-TCGACTGACAG-ATCTCAGG-3Ј; pVKH21, 5Ј-TCGACCTGTCGAGATCTGCAGTCAG-3Ј and 5Ј-TCGATCGACTGCAGATCTCGACAGG-3Ј.

RESULTS
Complementation Studies-Purified RegA, more specifically RegA*, is only available from R. capsulatus. To use this protein for in vitro DNA binding studies with the R. sphaeroides cbb system, it was necessary to establish that the R. capsulatus regA gene could indeed complement a R. sphaeroides regA mutant. Cosmid pCSM9d, carrying the R. capsulatus reg locus (31), was introduced into the R. sphaeroides regA insertionmutant CAC::regA⍀ (32). Although R. sphaeroides CAC::regA⍀ was unable to grow under phototrophic conditions, the strain complemented with the R. capsulatus regA gene, R. sphaeroides CAC⍀-C, regained the ability to grow under both photoheterotrophic and photoautotrophic conditions (data not shown). These results suggested that complementation with the R. capsulatus gene enabled R. sphaeroides CAC::regA⍀ to use this source of RegA to support photosynthetic growth.
Because RegA (PrrA) clearly regulates photopigment expression in both R. sphaeroides (18) and R. capsulatus (16,17), comparisons of the relative levels of photosynthetic pigment formation in R. sphaeroides CAC⍀-C, R. sphaeroides CAC::regA⍀, and R. sphaeroides CAC were made. Spectral analyses were performed on cell extracts of the three strains incubated under chemoheterotrophic, photoautotrophic, and anaerobic "carbon starvation" growth conditions (Fig. 1). Although R. sphaeroides CAC::regA⍀ is unable to grow photoautotrophically, sufficient cell mass was added in the initial inoculum to allow for sampling of the culture and extract preparation. Incubation under aerobic chemoheterotrophic conditions resulted in low levels of photopigment synthesis in R. sphaeroides CAC, with slightly lower levels in R. sphaeroides CAC⍀-C (Fig. 1C). R. sphaeroides CAC::regA⍀ possessed virtually no detectable photopigments (Fig. 1C). R. sphaeroides CAC::regA⍀ also did not induce photopigment synthesis after incubating dark chemoheterotrophically grown cells under photoautotrophic growth conditions (Fig. 1A); however, both strains CAC and CAC::⍀-C synthesized high levels of photopigments under photoautotrophic conditions. The level of photopigment synthesis for the complemented strain, R. sphaeroides CAC⍀-C, was Ͼ50% of that observed for R. sphaeroides CAC. Given that R. sphaeroides CAC::regA⍀ is unable to grow photoautotrophically, photopigment induction was examined under conditions where all strains could be compared relative to their ability to support photopigment synthesis. This was accomplished by first growing all strains (R. sphaeroides CAC⍀-C, R. sphaeroides CAC::regA⍀, and wild-type strain CAC) to mid-exponential phase under chemoheterotrophic conditions in the dark as before. These cells were then transferred to minimal media and incubated 72 h in the light while bubbling with 100% argon. Under these anaerobic, carbon-starved conditions, the pattern of photopigment induction in the three strains was identical to that found under photoautotrophic conditions, with wild-type strain CAC showing the highest level of photopigment induction, strain CAC⍀-C showing somewhat lower induction, and strain CAC::regA⍀ exhibiting no detectable induction (Fig. 1B).
The same extracts used in the photopigment expression studies were also used for Western immunoblots to determine the level of synthesis of both form I and form II Rubisco (Fig. 2,  A and B). R sphaeroides strain CAC::regA⍀ did not synthesize significant levels of form I Rubisco under any of the conditions tested ( Fig. 2A), whereas R. sphaeroides CAC and R. sphaeroides CAC⍀-C produced form I protein under both photoautotrophic and carbon starvation conditions. Western immunoblotting experiments using antibodies specific for R. sphaeroides form II Rubisco gave similar results (Fig. 2B).
Binding of R. sphaeroides CbbR to the cbb I Promoter-Previous gel mobility shift studies had indicated that CbbR binds to the cbb I promoter within 100 bp of the cbb I transcription start (14). To more clearly define the site of CbbR binding, DNase I protection assays, using 32 P-labeled probes spanning the region from ϩ40 to Ϫ133 bp relative to the cbb I transcription start were performed. CbbR protected two closely spaced regions, as seen by DNaseI digestion (Fig. 3). The first site (site A) is located from Ϫ14 to Ϫ35 bp, and the second site (site B) spanned from Ϫ41 to Ϫ65 bp. Similar regions of protection were detected on the opposite strand (data not shown).
Binding of R. capsulatus RegA* to the R. sphaeroides cbb I Promoter-Previous genetic evidence indicates that the RegA-RegB (PrrA-PrrB) system is involved in cbb regulation in R. sphaeroides (15). Because a constitutively active RegA protein (RegA*) had previously been isolated from the related nonsulfur purple bacterium R. capsulatus and shown to bind specific regions of the puf and puc operon promoters (23), we employed R. capsulatus RegA* in DNase I footprinting experiments of the cbb I promoter-operator. These studies were buttressed by the ability of the R. capsulatus regA gene to restore the ability of a R. sphaeroides CAC::regA⍀ to synthesize both form I and form II Rubisco (Fig. 2). Using 32 P-labeled probes covering a region of the R. sphaeroides cbb I promoter from ϩ61 to Ϫ600 bp, it was shown that RegA* did indeed bind to the R. sphaeroides cbb I promoter at four distinct sites (Figs. 4, A-C). Protection at the first site (site 1) was found from Ϫ67 to Ϫ80 bp with the additional protection of two bands at Ϫ83 and Ϫ84 bp (Fig. 4A). DNase I-hypersensitive sites were found within site 1 at Ϫ66 and Ϫ67 bp. The second site (site 2) was found in close proximity to site 1 and consisted of protection from Ϫ92 to Ϫ109 bp with a hypersensitive site located at Ϫ95 bp (Fig. 4A). The third RegA* binding site (site 3) protected the region from Ϫ302 to Ϫ327 bp with hypersensitive sites at Ϫ302 and Ϫ320 bp (Fig. 4B). Site 3 always exhibited protection at much lower protein concentrations relative to the other sites, suggesting that this site may have a higher binding affinity for RegA*. The fourth site (site 4) is comprised of an area of protection from Ϫ400 to Ϫ414 bp with DNase I-hypersensitive sites at Ϫ408 and Ϫ411 bp (Fig. 4C). We also detected a potential RegA* binding site overlapping the cbbF I transcription start, however, we were unable to clearly map this site in DNase I protection experiments (data not shown).
Helical Phase-dependent Function of cbb I Promoter Upstream Activating Sequences-Earlier studies using cbb I ::lacZ promoter fusions showed that a 4-bp deletion from Ϫ282 to Ϫ285 bp, approximately half a helical turn, resulted in a significant reduction in cbb I expression under all growth conditions tested (14). We wished to determine if the effect of the 4-bp deletion was due to the rotation of upstream regulatory sites to the opposite face of the DNA relative to downstream promoter elements. Small insertions of 4, 10, 15, and 21 bp were introduced between Ϫ281 and Ϫ282 bp in the promoter fragment of the cbb I ::lacZ fusion pVKC1 generating fusions pVKXFI, pVKH10, pVKH15, and pVKH21, respectively (Fig.  5A). The fusions were transferred into R. sphaeroides CAC and assayed for ␤-galactosidase activity under photoautotrophic growth conditions (Fig. 5B). Interestingly, insertions of 1 (pVKH10) and 2 (pVKH21) helical turns resulted in levels of cbb I expression that were equivalent to the wild-type promoter (pVKC1). By contrast, strains containing fusion plasmids car-rying insertions equivalent to approximately 0.5 (pVKXFI) and 1.5 (pVKH15) helical turns showed a 5-to 6-fold reduction in cbb I expression. These results indicated that optimal function of regulatory elements upstream of Ϫ282 bp in the cbb I promoter is helical phase-dependent. DISCUSSION A number of in vivo studies have shown that the product of the cbbR gene is involved in positive regulation of CBB cycle gene expression in both phototrophic (13,24,33) and chemoautotrophic bacteria (34,35). In the case of the chemoautotrophic bacteria Xanthobacter flavus (36) and Ralstonia eutropha (Alcaligenes eutrophus) (37), in vitro DNase I footprinting experiments have shown that CbbR protected two closely spaced regions within 100 bp of the cbb promoter. Each of these sites contains one or more copies of the LysR-type consensus DNA binding motif T-N 11 -A (38). Our footprinting experiments show that the binding of CbbR to the R. sphaeroides cbb I promoter followed a similar pattern, with dual regions of protection, each containing one or more T-N 11 -A motifs, separated by a short spacer sequence.
Beyond the in vitro binding of CbbR to the promoter operator region of the cbb I operon of R. sphaeroides, we were also inter- ested in providing an in vitro framework for previous studies, which indicated that the global response regulator RegA (PrrA) also positively affected cbb gene transcription in this organism. It was first shown that R. capsulatus regA complemented an R. sphaeroides regA-insertion mutant to phototrophic growth and restored regulated expression of both photopigment production and Rubisco synthesis. It was important to establish complementation because regA-deficient strains of these two organisms have different phenotypes despite the high degree of deduced amino acid sequence identity (85%) between the RegA/ PrrA proteins of these two organisms (17,18). Thus a clear rationale existed for using the constitutively active R. capsulatus RegA* to identify RegA binding sites within the R. sphaeroides cbb I promoter-operator region. The results of DNase I footprinting experiments reported here clearly demonstrated that R. capsulatus RegA* binds to multiple sites within the R. sphaeroides cbb I promoter region. The logical question to ask is whether RegA binding sites detected by the DNase I footprinting experiments represent functional regulatory sites. In this context, previous in vivo studies suggest that several of the newly identified RegA* binding sites within the cbb I promoter region do in fact serve a regulatory function. In particular, cbb I ::lacZ promoter fusion studies (14) indicated a positive regulatory role for binding sites 2, 3, and 4 (see Fig. 5). A cbb I ::lacZ fusion terminating at position Ϫ103 bp (BssHII site), containing RegA site 1, the CbbR binding sites, and only a portion of RegA site 2, showed 1.7-fold lower cbb I expression under photoautotrophic growth conditions than that with a longer fusion containing all of RegA site 2 (up to Ϫ281 of Fig. 5). Moreover, the fact that a fusion ending at the BssHII site still supported regulated cbb I expression is indicative that RegA site 1 is sufficient to confer RegA-mediated regulation of cbb I expression. The 50-bp region between the XhoI and PflM1 sites, that spans the "high affinity" RegA site 3, enhanced photoautotrophic cbb I expression an additional 13.8-fold (14). Furthermore, addition of RegA site 4, in a fusion containing a sequence of up to Ϫ637 bp, yielded a further 3-fold enhancement of photoautotrophic cbb I expression (14).
No clear sequence similarity was identified for the RegA binding sites within the R. sphaeroides cbb I promoter, when compared with one another or to RegA binding sites previously identified within the R. capsulatus puf and puc promoters (23). This lack of binding site similarity reinforces a previous suggestion that RegA may recognize secondary structural motifs of FIG. 6. Summary of DNase I footprinting results of CbbR and RegA* binding to the R. sphaeroides cbb I promoter-operator region. The sequence is numbered relative to the major cbb I transcription start at ϩ1. Brackets indicate regions of protection on the top (above) and bottom (below) strands with DNase I-hypersensitive sites indicated by asterisks. Translation start sites for cbbF I and cbbR are indicated along with the cbb I transcription start sites (Ͼ). Bold numbers associated with the vertical bars and arrows indicate the fold induction of cbb I expression contributed by that particular section of DNA between it and the previous vertical bar and arrow, as determined using cbb I ::lacZ fusions (14). the DNA rather than specific sequences (23). The placement of RegA binding sites within the promoter-proximal regulatory region of the R. sphaeroides cbb I promoter is very interesting in that RegA site 1 overlaps a portion of the distal CbbR binding site (see Fig. 6). This overlap of binding sites has also been observed in the puc promoter of R. capsulatus (23), where a RegA binding site overlaps the binding site of the aerobic repressor CrtJ (39). The overlap of RegA site 1 with the CbbR binding site may suggest a direct physical interaction between these two positive regulators during cbb I activation. Alternatively, CbbR and RegA could simply bind to opposite faces of the DNA. It is also conceivable that coinducer binding to CbbR could cause conformational changes that result in shortening of the CbbR binding site thus exposing the flanking RegA binding site. Shortening of a DNase I-protected region in response to the presence of a coinducer has been observed for the LysR-type regulators OccR (40) and OxyR (41), both of which are related to CbbR. RegA binding sites 4 and 5, in the promoter distal upstream activating region, together account for a combined 41-fold induction of cbb I expression under photoautotrophic growth conditions (14). It is also clear that the ability of binding sites 3 and 4 to function was helical phase-sensitive (see Fig. 5). Insertions of 4 bp (ϳ0.5 helical turn) and 15 bp (ϳ1.5 helical turns) between the promoter distal and promoter proximal regulatory regions at Ϫ282 bp would rotate any bound regulatory proteins to the opposite face of the DNA relative to downstream regulatory sites. These insertions reduced photoautotrophic induction of cbb I expression by more than 5-fold. In support of the helical phase interpretation, insertion of 10 bp (ϳ1 helical turn) and 21 bp (2 helical turns) at the same position restored full photoautotrophic cbb I expression. This helical phase-dependent function of the promoter distal regulatory region suggests that face of the helix contact(s) between proteins bound to the promoter proximal and promoter distal regulatory regions is important for cbb activation. Also, the large distance between the two regulatory regions (190 bp) indicated that contact between proteins bound at these widely separated sites may occur through DNA loop formation (Fig. 7). There are many examples where DNA loop formation is involved in prokaryotic gene regulation (42). DNA loop formation in the cbb I promoter may only require RegA to bind and hold the ends of the loop in a manner similar to the DNA loop formed during AraC-mediated repression of the E. coli araBAD operon (43). Alternatively, additional factors may play a role in DNA loop formation or stabilization. In Azotobacter vinelandii the NifA homolog VnfA binds to two direct repeat sequences Ϫ140 and Ϫ170 bp upstream of the consensus 54 -promoter of vnfH, where it activates vnfH transcription through interaction with the 54 -holoenzyme at the downstream promoter by means of DNA loop formation. This process is facilitated by an intrinsic DNA bend in the intervening region between the promoter and the VnfA binding site(s) (44). In other NifA-regulated promoters, the DNA bending protein IHF plays a role in loop formation (45). With regard to the cbb I system, the evidence for DNA looping has encouraged us to currently probe factors that play a role in DNA loop formation. Although another study (36) indicates that CbbR can bend DNA, it is not known if RegA shares this ability.
An important first step in elucidating the precise mechanism by which CbbR and RegA regulate cbb expression is to define the RegA and CbbR binding sites in vitro. The results of the current investigation clearly show that regA exerts its positive regulatory affect on cbb I expression directly, by binding at multiple sites within both the promoter proximal and promoter distal regulatory regions. Moreover, the helical phasing between the promoter proximal and promoter distal regulatory regions is necessary for normal promoter function. Studies are now underway to measure the affinities of the various binding sites as well as to determine if there are any synergistic interactions between RegA and CbbR at the cbb I promoter. Ultimately, these studies will provide answers as to how photosynthetic bacteria employ both global (RegA) and specific (CbbR) transcriptional activators to control CO 2 assimilation.