In Vitro and In Vivo Analysis of the Role of PrrA in Rhodobacter sphaeroides 2.4.1 hemA Gene Expression

Abstract

The hemA gene codes for one of two synthases in Rhodobacter sphaeroides 2.4.1 which catalyze the formation of 5-aminolevulinic acid. We have examined the role of PrrA, a DNA binding protein that is associated with the metabolic switch between aerobic growth and anoxygenic photosynthetic growth, in hemA expression and found that hemA transcription is directly activated by PrrA. Using electrophoretic mobility shift assays and DNase I protection assays, we have mapped two binding sites for PrrA within the hemA upstream sequences, each of which contains an identical 9-bp motif. Using lacZ transcription reporter plasmids in wild-type strain 2.4.1 and PrrA⁻ mutant strain PRRA2, we showed that PrrA was required for maximal expression. We also found that the relative impacts of altering DNA sequences within the two binding sites are different depending on whether cells are growing aerobically or anaerobically. This reveals a greater level of complexity associated with PrrA-mediated regulation of transcription than has been heretofore described. Our findings are of particular importance with respect to those genes regulated by PrrA having more than one upstream binding site. In the case of the hemA gene, we discuss possibilities as to how these new insights can be accommodated within the context of what has already been established for hemA transcription regulation in R. sphaeroides.

The α-proteobacterium Rhodobacter sphaeroides 2.4.1 can obtain energy by aerobic and anaerobic respiration and by anoxygenic photosynthesis. Such catabolic flexibility is possible because this organism is equipped with the ability to synthesize different tetrapyrroles and other molecular components necessary to carry out these processes. Hemes and bacteriochlorophyll are representative tetrapyrroles of R. sphaeroides that are indispensable to energy metabolism, and their absolute and relative levels vary according to environmental conditions. Thus, under aerobic conditions, hemes, required as part of cytochromes for electron transfer, are present but the cell does not produce bacteriochlorophyll (reviewed in references 21 and 24). As oxygen tensions fall, tetrapyrrole levels increase by more than 100-fold, predominantly in the form of bacteriochlorophyll, which serves in pigment-protein complexes that capture light energy during photosynthesis (24). In the absence of light or oxygen, hemes support growth by anaerobic respiration using alternate electron acceptors. We are interested in the genetic circuits that allow this facultative bacterium to adjust tetrapyrrole levels to suit its metabolic needs. Tetrapyrrole biosynthesis begins with the formation of 5-aminolevulinic acid (ALA), and ALA production parallels the cell's requirements for various amounts of tetrapyrroles (24). Therefore, investigations directed towards understanding how the cell produces the appropriate species and amounts of these critical molecules should benefit by defining the mechanisms by which ALA formation is regulated.

In R. sphaeroides 2.4.1, ALA is formed by the Shemin pathway, i.e., condensation of glycine with succinyl-coenzyme A, a reaction that is catalyzed by ALA synthase activity, assisted by the cofactor pyridoxal phosphate (21, 24). R. sphaeroides 2.4.1 has two ALA synthases, coded for by the hemA and hemT genes, which are differentially expressed (29, 39). The hemA gene is the primary target for regulation in response to changes in several environmental parameters, including changes in oxygen tension, and increased expression of hemA when oxygen tensions are reduced, in preparation for the metabolic switch to photosynthesis, is both dramatic and complex (13, 29).

Three DNA binding proteins that are associated with the changes in gene expression that accompany alterations in oxygen tension have been described for R. sphaeroides 2.4.1: PpsR, FnrL, and PrrA (reviewed in reference 40). Transcriptome profiling studies predict that hemA is not a member of the PpsR regulon (17, 28). With respect to FnrL, in vivo measurements have shown that an intact fnrL gene is required for anaerobic induction of transcription from both the P1 and the P2 promoter of hemA (13, 38). The role of PrrA in hemA regulation is less certain, since there are different conclusions on the relative role of this transcription factor in controlling expression of this gene (11, 30, 39). Here, we undertook an investigation of the role of PrrA in hemA expression for the purpose of resolving this issue.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1, and a description of relevant hemA sequences is shown in Fig. 1. Escherichia coli was grown in Luria-Bertani media (31), and R. sphaeroides was cultured in Sistrom's succinate minimal medium A (5, 7, 32). Reagent-grade antibiotics purchased from Sigma Chemical Co. (St. Louis, MO) were used for strain selection or plasmid maintenance. For E. coli, final concentrations were 100 μg/ml of ampicillin (Ap), 15 μg/ml of tetracycline (Tc), and 50 μg/ml of kanamycin or spectinomycin (Sp) and streptomycin (St). For R. sphaeroides, final concentrations were 50 μg/ml of Sp and St and 0.75 μg/ml of Tc.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant characteristic(s)a	Reference(s) or source
Strains
E. coli
DH5α	F− (φ80dlacZΔM15) recA1 endA1 hsdR17 supE44 thi-1 gyrA96 relA1 deoR Δ(lacZYA-argF)U169	20
DH5αphe	DH5α with phe::Tn10d; Cmr	11
ER2566	F−fhuA2 lon ompT lacZ::T7 gene1 gal sulA11 Δ(mcrC-mrr)114::IS10R[mcr-73::mini-Tn10(Tcs)]2 R[zgb-210::Tn10(Tcs)] endA1 dcm	New England Biolabs
HB101	F− Δ(gpt-proA)62 leuB6 supE44 ara-14 galK2 lacY1; used in triparental matings	6
XL1-Blue	[F′::Tn10(Tcr) proAB lacIq Δ(lacZ)M15] recA1 endA1 gyrA96 thi-1 supE44 relA1 lac	Stratagene
R. sphaeroides
2.4.1	Wild type	W. Sistrom
PRRA2	ΔprrA(BstBI-PstI)::ΩSprStr; cannot grow photosynthetically	10
Plasmids
pBRO9	pUI1925 with MUT-D mutation (Fig. 1); Apr	This study
pBRO10	pUI1925 with MUT-C mutation (Fig. 1); Apr	This study
pBRO11	pUI1925 with MUT-B mutation (Fig. 1); Apr	This study
pBRO12	pUI1925 with MUT-A mutation (Fig. 1); Apr	This study
pBRO29	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO12 carrying MUT-A mutation; Tcr Spr/Str	This study
pBRO34	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO11 carrying MUT-B mutation; Tcr Spr/Str	This study
pBRO36	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO9 carrying MUT-D mutation; Tcr Spr/Str	This study
pBRO40	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO10 carrying MUT-C mutation; Tcr Spr/Str	This study
pBRO42	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pUI1925 carrying wild-type sequence; Tcr Spr/Str	This study
pBRO44	pSB3 with MUT-E mutation (Fig. 1); Apr	This study
pBRO49	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO44 carrying MUT-E mutation; Tcr Spr/Str	This study
pBRO53	pBRO44 with MUT-B mutation (Fig. 1); Apr	This study
pBRO54	pSB3 with MUT-B mutation (Fig. 1); Apr	This study
pBRO55	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO54 carrying MUT-B mutation; Tcr Spr/Str	This study
pBRO56	hemA::lacZ fusion in pCF1010 with PstI-XbaI fragment from pBRO53 carrying MUT-B and MUT-E mutations; Tcr Spr/Str	This study
pBSIISK+	ColE1; Apr	Stratagene
pCF1010	RSF1010 derivative; used for creating lacZ transcriptional fusions; Tcr Spr/Str	25
pJC407	PrrA intein/chitin-binding domain expression plasmid; Apr	4
pRK2013	IncP1 ColE1 Tra+ of RK2; Knr; used as a helper plasmid in triparental matings	6, 15
pSB3	316 bp of hemA::lacZ upstream sequences (Fig. 1) in pUI1087; Apr	This study
pSB4	pCF1010 with PstI-XbaI fragment from pSB3; Tcr Spr/Str	This study
pUI1015	2-kb NaeI-BamHI fragment with hemA ligated to pUC19(HincII-BamHI); includes 246 bp of upstream hemA sequences (Fig. 1); Apr	29
pUI1087	pBSIISK+ with modified polylinker; Apr	41
pUI1925	246 bp of hemA upstream sequences (Fig. 1) in pUI1087; Apr	13
pRKK146	In vitro transcription template containing cycA P2 fused to the spf′ terminator and carrying a control promoter (RNA-1); Apr	22
pJZ52	In vitro transcription template containing hemA::lacZ promoter sequences (Fig. 1) fused to the spf′ terminator and carrying a control promoter (RNA-1); Apr	This study
pUC19spf′	pUC19 derivative with ColE1 (RNA-1 transcript) and spf′ sequences that include the Spot42 transcription terminator; Apr	1, 12

^aCm, chloramphenicol; Kn, kanamycin.

FIG. 1.

Description of the relevant DNA sequences and oligonucleotides used. The hemA sequences carried on plasmid pSB3 extend 316 bp upstream of the translation initiation codon (the first residue of the codon is numbered +1 and labeled MET-HemA), and those hemA sequences carried on plasmids pJZ52 (used as a template in the transcription assays), pUI1015 (used as a template to generate labeled DNA for EMSAs), and pUI1925 (used as a template in oligonucleotide-directed mutagenesis) extend 246 bp upstream of the translation initiation codon; the vector DNA sequences for pSB3 and pUI1015 are displayed offset relative to the hemA DNA sequences. The +1 sites of transcription from hemA promoters P1 and P2 (13, 29) are labeled (P1) and (P2). The FNR consensus sequence is also labeled. Sequences of one partner of each pair of mutagenic oligonucleotides are as shown (the partner has the complementary sequence), and the mutations created are shown in bold and underlined. Primers used to synthesize biotin-labeled DNA that was examined by EMSAs are indicated by arrows above the DNA sequences. Fluorescently labeled (FAM) DNA used in the DNase I protection assays was generated using primers whose sequences are indicated by bold and italics. Note that the DNA was labeled on only one strand, and so a FAM-labeled primer was used in combination with an unlabeled primer in each labeling reaction. The sequences protected from DNase I cleavage by AP-PrrA are highlighted in gray boxes, and a 9-bp motif that is present within each of these regions is shown with white letters. For further details regarding the oligonucleotides and their use, see Materials and Methods.

Conjugations and transformations.

Mobilization of plasmids into R. sphaeroides was performed by triparental matings with HB101(pRK2013), as described previously by Davis et al. (5). E. coli cells were prepared for transformation by CaCl₂ treatment (31), with the exception of XL1-Blue competent cells included in a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA).

DNA manipulations and DNA sequence analysis.

DNA isolation, restriction endonuclease treatment, and other enzymatic treatment of DNA fragments and plasmids were done according to standard protocols (31) or manufacturers' instructions. Enzymes were purchased from New England Biolabs, Inc. (Beverly, MA), Gibco-BRL/Life Technologies, Inc. (Gaithersburg, MD), and Promega (Madison, WI). DNA was analyzed by standard electrophoretic techniques (31), and isolation of DNA from agarose was performed using a Zymoclean purification kit (Zymo Research Co., Orange, CA). DNA sequencing was performed using an ABI Prism 310 genetic analyzer with an ABI Prism BigDye Terminator cycle sequencing ready reaction kit (Applied Biosystems, Inc., Foster City, CA) and primers purchased from Integrated DNA Technologies (Coralville, IA). Sequencing reaction mixtures were prepared according to manufacturers' instructions. To improve primer extensions due to the high G+C content of the R. sphaeroides genome, dimethyl sulfoxide (DMSO) was added to the sequencing reaction mixtures at a final concentration of 5% prior to thermal cycling.

Mutagenesis.

Oligonucleotide-directed mutagenesis was carried out using a QuikChange site-directed mutagenesis kit (Stratagene) and primers purchased from Integrated DNA Technologies. By incorporating restriction endonuclease recognition sites within the mutagenized sequences, we could prescreen plasmid candidates. Positive candidates were then confirmed by DNA sequence analysis of all of the hemA sequences on both strands. The sequences of the mutagenic primers used and the mutations introduced into the hemA sequences are shown in Fig. 1.

PrrA purification and phosphorylation.

The PrrA protein was purified from E. coli strain ER2566 with plasmid pJC407 (4), which expresses a PrrA-intein/chitin binding domain fusion protein, by use of an IMPACT T7 one-step protein purification system (New England Biolabs, Waltham, MA). The fusion protein was bound to chitin beads under conditions described previously (4), followed by overnight incubation with dithiothreitol, during which the PrrA protein was released by the intein cleavage reaction. The purified protein consists of the complete PrrA amino acid sequences extended by C-terminal Pro-Gly, which were added to optimize stability and cleavage (4). After reaching a concentration of approximately 1.5 to 1.8 mg/ml by use of Amicon Ultra-15 centrifugal filter units (Millipore Corp., Billerica, MA), the purified PrrA protein was dialyzed against storage buffer (40 mM Tris-HCl, pH 7.9, 50 mM KCl, 5 mM MgCl₂, and 1 mM dithiothreitol) and stored in 10-μl aliquots at −80°C. Acetyl phosphate treatment of PrrA was performed as described by Comolli et al. (4); the reaction mixtures contained 30 μM PrrA, 25 mM acetyl phosphate, and 20 mM MgCl₂ in a total reaction volume of 20 μl and were incubated for 1 hour at 30°C.

In vitro transcription assays.

In vitro transcription assays were performed using crude R. sphaeroides 2.4.1 RNA polymerase holoenzyme (4). Template DNA included 246 bp of hemA sequences upstream of the translation initiation site (Fig. 1) contained on a PstI-XbaI fragment isolated from plasmid pUI1925 (Table 1) and inserted upstream of transcription terminators in plasmid pUC19spf′ (1), creating plasmid pJZ52. Assays were also performed using as a template the plasmid pRKK146 (22), having the same vector backbone but with the R. sphaeroides 2.4.1 cytochrome c₂ gene cycA P2 sequences, as a positive control for PrrA and acetyl phosphate-treated PrrA (AP-PrrA) activation of transcription. As detailed in the assay protocol (4), the RNA-1 transcript that is also transcribed from the plasmids was used to evaluate relative abundance of hemA or cycA transcripts generated in the presence and absence of PrrA or AP-PrrA.

EMSAs.

Biotin-labeled PCR products containing various amounts of hemA upstream sequences were generated with primer pairs (purchased from Integrated DNA Technologies and with the sequences shown in Fig. 1) in which one of the primers is 5′ end labeled with biotin. Suitable concentrations of biotin end-labeled DNA were determined by performing dot blot analysis of 1:10, 1:100, 1:1,000, and 1:10,000 dilutions of DNA and using components of a Pierce chemiluminescent electrophoretic mobility shift assay (EMSA) kit (Pierce Chemical Company, Rockford, IL). Mobility shift assays were then carried out according to the manufacturer's instructions, with the substitution of 50 ng/μl poly(dA · dT) purchased from Sigma Chemical Co., as recommended for high-G+C genomes to reduce nonspecific binding. The binding reaction mixtures contained approximately 150 nM biotin-labeled DNA, 8.6 μM phosphorylated PrrA, 2.5% glycerol, 50 ng/μl poly(dA · dT), 40 mM Tris (pH 7.9), 50 mM MgCl₂, and 2 mM EDTA in a total volume of 20 μl and were incubated for 20 min at room temperature, followed by the addition of 5 μl loading buffer (Pierce chemiluminescent EMSA kit). Manufacturer's instructions were also followed for sample electrophoresis, transfer to nylon membranes (Pierce Chemical Company), and detection of the biotin-labeled DNA with Kodak BioMax XAR film (Rochester, NY). To calibrate relative mobilities, prebiotinylated size standards (New England Biolabs) were electrophoresed with all of the samples.

DNase I protection assays.

DNase I protection assays were performed based on methods described by Yindeeyoungyeon and Schell (36), using an ABI Prism 310 genetic analyzer to detect fluorescently labeled cleavage products. The sequences to be examined were end labeled on one strand by thermal cycling a reaction mixture containing 30 μM of pSB3 template DNA and 20 pmol of each primer (sequences are shown in Fig. 1). Fluorescently 5′-end 6-carboxyfluorscein (FAM)-labeled primers were purchased from Applied Biosystems. The labeled DNA products were then gel purified and drop dialyzed against water by using 0.025-μm filter disks (Millipore Corp.) for 1 h at room temperature.

In order to generate an optimal fragmentation pattern, the labeled DNA was treated with various concentrations of freshly diluted RNase-free DNase I (Roche Molecular Biochemicals, Indianapolis, IN). The 10-U/μl stock of enzyme was serially diluted in D buffer (10 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 5 mM CaCl₂, and 0.1 mg/ml bovine serum albumin), as described by Yindeeyoungyeon and Schell (36). Each 20-μl reaction mixture contained approximately 18 μM of FAM-labeled DNA, to which 5 μl of diluted DNase I was added. Final enzyme concentrations ranged from 1 × 10⁻⁵ to 3 × 10⁻⁸ U/μl. The reaction mixtures were incubated at 20°C for 2 min and then shifted to 98°C for 10 min in order to heat inactivate the enzyme. Following the recommendations of Yindeeyoungyeon and Schell (36), optimal DNase I concentrations were those that left approximately 40% of full-length labeled PCR product. We also found that this treatment generated the most consistent fragmentation pattern and clearly resolved product peaks.

The protection assays were then performed by adding 5 μl of the optimal concentration of DNase I to 20-μl samples that contained 18 μM FAM-labeled DNA and 0, 100, 250, or 500 nM of PrrA or AP-PrrA and that had been preincubated at room temperature for 20 min to allow the protein to bind to the DNA (these are the same conditions as those used for the EMSAs). Following a 2-min incubation at 20°C, the DNase I was heat inactivated as described previously. In order to correct for sample loss during subsequent processing, approximately 0.7 μM of intact, FAM-labeled but untreated DNA of a length that is dissimilar to that of the test DNA (generated with unlabeled “3a” and FAM-labeled “DOWN” primers, shown in Fig. 1) was added to each assay mixture. Enzyme was removed from the samples by using phenol:chloroform (3:1) extraction, and the aqueous layer was further treated according to the Applied Biosystems protocols for preparing sequencing reaction mixtures, except that 1.6 fmol of 500-TAMRA (6-carboxytetramethylrhodamine) size standards (purchased from Applied Biosystems) was added to each sample. The samples were loaded onto the ABI Prism 310 genetic analyzer and then run and analyzed using the same parameters as previously described (36), with the exception of substituting 500-TAMRA size standards.

The untreated labeled DNA that had been added before sample processing was used to standardize the samples, allowing us to directly compare the results in terms of relative peak fluorescence intensities. Phosphodiester bonds that are protected from or hypersensitive to DNase I cleavage in the presence of increasing amounts of acetyl phosphate-treated PrrA protein were identified on the basis of a decrease or an increase in relative fluorescence intensity measured for each cleavage product, relative to results for DNA treated in the absence of PrrA.

Construction of the β-galactosidase reporter plasmids.

Transcription fusion plasmids were constructed by moving hemA sequences contained on PstI-XbaI DNA fragments isolated from plasmid pUI1925 or pSB3 or their mutagenized derivatives into plasmid pCF1010 (Table 1). In this way, the wild-type or mutant hemA upstream sequences are positioned in front of a promoterless lacZ gene.

β-Galactosidase activity assays.

Assays of β-galactosidase activity were performed using extracts of cells that had been grown aerobically by sparging liquid cultures with a mixture of 30% oxygen, 2% carbon dioxide, and 68% nitrogen or under anaerobic-dark conditions in screw-cap tubes completely filled with Sistrom's succinate medium supplemented with yeast extract (final concentration of 1%, wt/vol) and with DMSO added as an alternate electron acceptor to a final concentration of 0.06 M. Preparation of cleared cell lysates and enzyme assays were performed as previously described (34). All spectrophotometric measurements were made using a U-2010 UV/Vis spectrophotometer (Hitachi High Technologies America, Inc., Schaumburg, IL).

Protein concentrations.

A Pierce bicinchoninic acid protein assay reagent or a Bio-Rad protein assay dye reagent concentrate (Hercules, CA) was used to determine protein concentrations. Bovine serum albumin was used as a standard in all cases.

RESULTS

Transcription activation by PrrA and AP-PrrA in vitro.

A guide to the hemA upstream sequences, its relevant features, and additional information is provided in Fig. 1. Using as template DNA either plasmid pJZ52, which includes 246 bp of hemA sequences upstream of the translation initiation site or 168 bp upstream of the P1 transcription start site (Fig. 1 and Table 1), or plasmid pRKK146, having the P2 promoter sequences of the cycA gene of R. sphaeroides, coding for cytochrome c₂, and serving as a positive control for PrrA-activated transcription (22), we performed transcription assays in the presence of various concentrations of PrrA and AP-PrrA. As is true for other response regulators (14, 19, 27), acetyl phosphate treatment has been shown to generate phosphorylated PrrA (4), making it possible to modify the protein purified from an E. coli expression system in the absence of its cognate kinase protein, PrrB. The transcription results shown in Fig. 2 indicate that, while the RNA-1 transcript generated from either plasmid could be detected in all of the lanes regardless of the presence or absence of untreated PrrA or AP-PrrA, hemA transcripts were produced at detectable levels only in the presence of PrrA, and AP-PrrA further stimulated transcription. Since no hemA transcripts could be detected in the absence of added PrrA, the degree to which PrrA activates transcription could not be determined. However, we could compare the relative levels of transcripts generated in reaction mixtures that included PrrA to those generated in mixtures that contained AP-PrrA. At the highest concentrations of activator used, we calculate that hemA transcript levels are approximately 17- to 20-fold higher in reaction mixtures that included AP-PrrA than in those that included PrrA.

FIG. 2.

Image of autoradiograph of transcripts generated in vitro. The left four lanes are transcripts generated using plasmid pRKK146 as a template, and the right five lanes are transcripts generated using plasmid pJZ52 as a template (Table 1). The RNA products are as indicated, and the amounts of PrrA or AP-PrrA used are as shown. For further details regarding the reaction conditions, see Materials and Methods.

The hemA upstream DNA present in plasmid pJZ52 contains two promoters, P1 and P2 (Fig. 1), which have been examined previously both in vitro (29) and in vivo (13). The relative sizes of the cycA P2 and hemA transcripts indicate that PrrA activation is targeting the P1 promoter of hemA, since the hemA transcripts generated from the P1 promoter of hemA are calculated to be 38 nucleotides (nt) longer than cycA P2 transcripts, whereas transcripts from the P2 promoter of hemA would be 4 nt shorter than the cycA P2 transcripts (13, 22, 29). The absence of detectable transcripts from the hemA P2 promoter is consistent with previous in vivo results that showed that transcription from the P2 promoter requires an intact fnrL gene (13). With respect to the P1 promoter, although this promoter also requires fnrL for induced expression in response to lowering oxygen tensions, it was proposed that this induction probably occurs via an indirect mechanism (13). Such a mechanism is compatible with the data presented here, which indicate that transcription from P1 requires PrrA.

Localizing the PrrA binding sites within the hemA upstream sequences by use of EMSAs.

Having demonstrated that PrrA directly activates hemA transcription, we undertook to determine where PrrA binds within the hemA upstream sequences. We examined whether or not EMSAs could be used to identify hemA upstream sequences required for PrrA binding by performing a competition assay using the same sequences that had been examined in the transcription assays, i.e., 246 bp upstream of the translation initiation site (Fig. 1). As shown in Fig. 3, the presence of AP-PrrA generated a slow-migrating species, and formation of this species could be abolished by the addition of unlabeled specific competitor DNA.

FIG. 3.

EMSA results for DNA corresponding to hemA residues −246 to +1, relative to the translation initiation codon. DNA was generated using “UP” and biotin-labeled “DOWN” primers (Fig. 1). The presence or absence of AP-PrrA is as indicated, as is the amount of unlabeled competitor DNA added to the assay.

By performing EMSAs using end-labeled products corresponding to consecutively overlapping portions of the hemA upstream sequences, we then scanned across the hemA upstream sequences to further delimit the sequences required for PrrA binding. As shown in Fig. 4, two possible binding sites, corresponding to hemA sequences spanning residues −246 to −187 and −161 to −87 relative to the translation initiation codon, were detected, since DNA containing those sequences is shifted in the presence of AP-PrrA. This positions the two binding sites −168 to −131 bp and −83 to −50 bp relative to the P1 transcriptional start site.

FIG. 4.

EMSA results for DNA containing various hemA upstream sequences. (A) Positions of key coordinates within the hemA upstream sequences and corresponding diagram of the labeled DNA samples (PCR products generated using the primers indicated; see Fig. 1 for primer sequences) analyzed in the assays. Shaded regions indicate sequences that are shifted in the presence of AP-PrrA. (B) Images of the mobility shifts detected using the PCR products described in the legend for panel A. The DNA is identified by the primer pairs used to generate the labeled DNA, where “U” denotes the “UP” primer and “D” indicates the “DOWN” primer. A shift in mobility indicates DNA binding by AP-PrrA protein and is marked with a “+” below the images, “+/−” indicates that only a portion of the labeled DNA is shifted, and “−” indicates that none of the DNA is shifted.

DNase I protection assays.

Initially, PrrA-mediated protection from DNase I cleavage was examined using amplified DNA having the same hemA sequences that had been evaluated by the transcription assays and the EMSAs, containing 246 bp upstream of the translation initiation site of hemA. Those protection results (not shown) indicated that the upstream PrrA binding site might continue beyond the sequences contained on that DNA template. Therefore, we repeated the assays using DNA containing 316 bp of hemA sequences upstream of the translation initiation site (Fig. 1). For those sequences, the DNase I cleavage patterns generated in the presence and absence of AP-PrrA are shown in Fig. 5, together with a graphic comparison of the relative peak intensities indicating the amounts of cleavage products. The results indicate that, as was also observed with the EMSAs, there are two regions that are protected: an upstream binding site (binding site I) that spans residues −272 to −209 relative to the hemA translation initiation codon and a downstream binding site (binding site II) that spans residues −161 to −128 relative to the translation initiation codon. Both of the binding sites are located upstream of the hemA P1 promoter, which is consistent with ascribing a transcription activator function to PrrA.

FIG. 5.

Schematic diagram of DNase I protection assay data. (A and D) Bar graph displays of the fractional peak fluorescence intensities corresponding to cleavage between consecutive residues of hemA generated in the presence of AP-PrrA in the concentrations indicated, relative to peak intensities in the absence of AP-PrrA (assigned the value of 1 on the y axes). The DNA residues are indicated on the x axes of the graphs, and the sequences that are protected from cleavage are underlined and in bold. (B and C) Relevant portions of the electropherograms of the samples. For further details regarding measurements, see Materials and Methods.

In vivo analysis of PrrA regulation of hemA under anaerobic (inducing) conditions.

In order to examine the role of PrrA with respect to hemA transcription in vivo and evaluate the relative levels of importance of the two PrrA binding sites, we constructed lacZ transcriptional reporter plasmids carrying intact or mutated hemA upstream sequences that are altered within the two PrrA binding sites defined by the in vitro analyses (Fig. 1). We evaluated binding site II alone first, using reporter plasmids having either the intact or altered sequences corresponding to plasmid pUI1925, which contains 246 bp of hemA sequences upstream of the translation initiation site. We mutated the hemA sequences by using mutagenic oligonucleotides that introduced changes encompassing 6 residues at a time, scanning across the DNase I-protected region. While studies have demonstrated that the sequences associated with PrrA binding are highly variable (8, 9, 22, 23, 26), the one consistent feature is a preponderance of G/C nucleotides. Therefore, we deliberately chose alterations that are AT rich. To accommodate these changes and to incorporate a restriction endonuclease recognition sequence so as to facilitate screening of mutant plasmid candidates, each mutated sequence contains an EcoRI site (mutations are designated MUT-A to MUT-D [Fig. 1]). The intact or altered hemA sequences were isolated on DNA fragments from plasmid pUI1925 and the mutagenized derivatives, which were generated using flanking PstI and XbaI restriction sites within the vector sequences. These DNA fragments were then positioned upstream of a promoterless lacZ gene to create the corresponding reporter plasmids, which were subsequently moved into wild-type strain 2.4.1 and PrrA⁻ mutant strain PRRA2. We cultured the resulting exconjugants under anaerobic conditions, which are inducing for hemA expression (29, 38), and then assayed cell extracts for β-galactosidase activity. The hemA sequences examined and the corresponding measurements of β-galactosidase activity are reported in Fig. 6. The data indicate that, while β-galactosidase activity in extracts of wild-type strain 2.4.1 with plasmid pBRO42 having intact hemA sequences is approximately 6.9-fold higher than that in extracts of mutant strain PRRA2 with the same reporter plasmid, the activities vary no more than 1.4-fold in extracts of the two strains having any of the reporter plasmids with altered hemA sequences, i.e., plasmids pBRO29, -34, -36, and -40. Thus, apparently all of the hemA sequence changes introduced by mutagenesis negatively impact PrrA binding at binding site II.

FIG. 6.

β-Galactosidase activities in extracts of wild-type strain 2.4.1 (black bars) and mutant strain PRRA2 (gray bars) with lacZ transcription reporter plasmids having intact or altered hemA upstream sequences. For further details regarding the hemA sequences present on the plasmids, see Table 1. Cells were cultured under anaerobic-dark/DMSO conditions, and values represent duplicate assays of a minimum of three independent growth experiments. Vertical bars represent the standard deviations from the means, and the numerical values are provided in parentheses. One unit of enzyme activity is defined as 1 μmol of o-nitrophenyl-β-d-galactopyranoside hydrolyzed per minute.

We then evaluated the upstream binding site (binding site I), using reporter plasmids having either intact or altered sequences corresponding to plasmid pSB3 (Table 1), which contains 316 bp of hemA sequences upstream of the translation initiation site (Fig. 1). An inspection of the DNA sequences spanning binding site II revealed that they contain a 9-bp motif, GCTGGCGGT, which is exactly repeated within the sequences of binding site I (Fig. 1). Since the β-galactosidase activity assays indicate that altering those sequences within binding site II abolishes PrrA binding, as reported using plasmid pBRO34 (Fig. 6), the upstream repeat of the motif within binding site I was also targeted for mutagenesis. Again, we introduced changes that replace GC-rich sequences with AT-rich sequences (designated MUT-E [Fig. 1]); this time we created a new AseI restriction site, which was exploited to screen the plasmid candidates for the presence of the altered sequences. The changes to the upstream 9-bp motif were made alone and also in combination with changes to the downstream motif (MUT-B [Fig. 1]). Then, exploiting the flanking PstI and XbaI restriction sites within the vector DNA, the wild-type or mutant hemA upstream sequences were positioned upstream of a promoterless lacZ gene. Using these lacZ reporter plasmids having the intact and altered sequences, we evaluated the level of transcription in wild-type strain 2.4.1 versus that in mutant strain PRRA2 in extracts of cells grown under inducing, anaerobic conditions for hemA expression. Those β-galactosidase activity measurements are shown in Fig. 7. The data indicate that the level of transcription from wild-type hemA upstream sequences containing both PrrA binding sites, monitored using plasmid pSB4, is approximately 1.8-fold higher than that from sequences that contain only the intact binding site II, monitored using reporter plasmid pBRO42. Furthermore, alterations to the 9-bp motif within binding site I represented on plasmid pBRO49 result in a reduction of β-galactosidase activity to nearly the same levels as those reported using plasmid pBRO42 having binding site II alone; the relative differences in extracts of wild-type versus PrrA⁻ cells for pBRO49 and pBRO42 are approximately 7.7- and 6.9-fold, respectively. By contrast, the assay results for plasmid pBRO55 indicate that altering the downstream motif within binding site II alone apparently abolishes all PrrA regulation, since the β-galactosidase activity measurements differ by approximately 1.1-fold in extracts of wild-type 2.4.1 cells versus extracts of PrrA⁻ mutant cells. Similarly, the activities measured in extracts of wild-type and PrrA⁻ mutant cells with plasmid pBRO56, in which both motifs are altered, differ by approximately 1.3-fold.

FIG. 7.

β-Galactosidase activities in extracts of wild-type strain 2.4.1 (black bars) and mutant strain PRRA2 (gray bars) with lacZ transcription reporter plasmids having intact or altered hemA upstream sequences. Alterations to the hemA sequences are within the two PrrA binding sites identified using EMSAs and DNase I protection assays and are indicated on the x axis. The hemA sequences present on the plasmids are detailed in Fig. 1, and additional information about the plasmids is included in Table 1. (A) Activities measured in extracts of cells with the plasmids indicated that had been cultured under anaerobic-dark/DMSO conditions. (B) Activities measured in extracts of cells with the plasmids indicated that had been grown aerobically by sparging liquid cultures with a mixture of 30% oxygen, 68% nitrogen, and 2% carbon dioxide. Vertical bars represent the standard deviations from the means, and the numerical values are provided in parentheses. Values represent duplicate assays of a minimum of three independent growth experiments. One unit of enzyme activity is defined as 1 μmol of o-nitrophenyl-β-d-galactopyranoside hydrolyzed per minute.

In vivo analysis of PrrA regulation of hemA under aerobic conditions.

We also assayed extracts of aerobically grown (sparging with 30% oxygen) wild-type and PrrA⁻ mutant cells with reporter plasmids either having the intact hemA upstream sequences containing both PrrA binding sites or having changes in either binding site alone or in both binding sites together. The data, shown in Fig. 7, indicate that transcription from the hemA upstream sequences is reduced by approximately 5.4-fold in the absence of PrrA. Further, in contrast to the situation with anaerobically grown cells, in which binding site II plays a greater role, under aerobic conditions both binding sites apparently need to be intact in order to achieve full expression of hemA, since alterations in either binding site I or binding site II reduce the level of activity to that measured for the reporter plasmid having intact hemA sequences in the PrrA⁻ mutant cells.

In combination with the results obtained for anaerobically grown cells, this analysis also made it possible to assess the contribution of fnrL to hemA expression by comparing the β-galactosidase activities in extracts of PrrA⁻ cells having the reporter plasmid pSB4 with intact hemA upstream sequences grown aerobically versus anaerobically. Assuming that FnrL is the only other transcription factor influencing transcription from these sequences in response to lowering oxygen tensions, it is capable of inducing hemA transcription by approximately 11.3-fold.

DISCUSSION

On the basis of our in vitro and in vivo results, we conclude that PrrA directly activates hemA transcription, that this protein binds to two sites within the hemA upstream sequences, and that this binding is required for transcription from the hemA P1 promoter under both aerobic and anaerobic conditions. Our findings further indicate that the relative contributions of the two PrrA binding sites change according to the growth conditions; whereas binding sites I and II are equally important in PrrA-dependent hemA transcription under aerobic conditions, binding site II is more important than binding site I under anaerobic conditions. Similar to our findings for the PrrA requirement for transcription from the P1 promoter of hemA, several workers have previously reported that PrrA or its Rhodobacter capsulatus homologue, RegA, is required for both aerobic basal transcription and induction of transcription in response to changes in oxygen tensions for other genes (2, 4, 18, 19, 22). We believe this is the first reported observation of a differential effect for two PrrA binding sites associated with the same gene. Among many potential mechanisms that could account for an apparent change in relative importance of these binding sites, we favor the possibility that the change reflects the state of phosphorylation of PrrA. We propose that under aerobic conditions PrrA is predominantly unphosphorylated and that in this form it binds cooperatively at each of the two sites. Under anaerobic conditions, PrrA is predominantly phosphorylated, and this form of PrrA has a higher affinity for binding site II.

While it may be attractive to conclude that transcription from the P1 promoter relies on PrrA alone and that FnrL mediates transcription exclusively from P2, we have previously shown that anaerobic induction from both P1 and P2 requires an intact fnrL gene (13). We also noted that the position of the presumed FnrL binding site, an FNR consensus-like sequence that actually spans the +1 site of transcription from the P1 promoter, makes it doubtful that bound FnrL is activating P1 transcription; rather, we proposed that FnrL-mediated anaerobic induction from P1 might occur by an indirect mechanism (13). Defining PrrA as a direct activator of hemA transcription raises the possibility that an indirect effect of FnrL on expression of this gene could operate through PrrA. Since the prrA upstream sequences lack an FNR consensus-like sequence, it is unlikely that FnrL regulates prrA transcription. More likely is the possibility that the presence of an active FnrL protein somehow increases PrrA activity. Thus, collectively, our results to date lead us to propose that (i) PrrA is required to activate hemA P1 transcription, (ii) FnrL acts as an anaerobic inducer of hemA transcription, and (iii) FnrL-mediated induction involves a direct mechanism with respect to P2 and an indirect mechanism for P1 that is likely centered on the effects of FnrL on PrrA activity. Finally, the work of Oh et al. (30) tells us that this indirect mechanism apparently occurs by some means other than FnrL modulation of the inhibitory signal emanating from the cbb₃ cytochrome oxidase complex that maintains PrrA in an unphosphorylated state under aerobic conditions.

Attempts to describe a consensus DNA binding sequence for PrrA have been hampered in part by the apparent variable lengths of the sequences. However, reports by several investigators agree that the most conserved DNA sequence motif is a GCG inverted repeat with variable spacing that is between 0 and 12 nt (23, 26). This same sequence motif had been described previously as the recognition sequence for RegA (33), the R. capsulatus PrrA homologue that is 100% identical at the amino acid sequence level in the DNA binding domain. While several GCG inverted repeats can be identified in the hemA upstream sequences, one example is present outside the DNase I-protected sequences, indicating that DNA sequence information alone may not be sufficient to predict PrrA binding sites. It may be that DNA structure is an important parameter in PrrA binding, as has been suggested by Laguri et al. (23). Using the program bend.it (35), we compared the predicted bending and curvature of the two PrrA binding sites of hemA P1 identified here to those of cycA P2 and found that hemA P1 sequences comprising PrrA binding sites I and II are characterized by a high degree of curvature at each of the half sites and that centered within each of the binding sites is a peak in predicted curvature. These characteristics are in good agreement with the description by Laguri et al. (23) of the PrrA binding site of cycA (22). We then analyzed cycA sequences having the 4-bp mutations that Karls et al. (22) demonstrated abolish PrrA-mediated activation of cycA P2 transcription and found that the altered sequences are predicted to be dramatically changed in curvature and, to a lesser extent, reduced in bendability within the PrrA binding site. Similar changes in structure (altered curvature and reduced bendability) are predicted from the analysis of the hemA sequences having mutations within the two PrrA binding sites.

The previously unknown level of complexity regarding the activities of FnrL and PrrA towards hemA transcription revealed by these studies, as well as other new findings (28, 37), has added new layers of control to the overall pattern of regulation of the tetrapyrrole biosynthesis pathway (30). However, it remains true that regulation of hemA transcription by both FnrL and PrrA provides considerable and necessary flexibility in controlling the production of all tetrapyrroles. Indeed, we estimate from our in vivo assays that the total range in the level of hemA transcription that could be achieved through the combined activities of these DNA binding proteins is more than 59-fold. For enzymes such as HemA, which have a very low rate of turnover (3), this responsiveness could be especially important, since enzyme availability may be a very significant means to increase the amount of tetrapyrrole formed. In the case of the HemA product, ALA, such an increase is indispensable for the production of sufficient amounts of bacteriochlorophyll for photosynthesis. Furthermore, with organisms such as R. sphaeroides, which use the Shemin pathway for ALA formation, as has already been elegantly described by Lascelles (24), we should keep in mind the need for the cell to appropriately partition succinyl coenzyme A between the tricarboxylic acid cycle and ALA production in order to maintain a balance between carbon metabolism and the synthesis of molecules essential for energy metabolism, i.e., tetrapyrroles. Depending on growth conditions, the cell must have the ability to redirect flow at this branch point, and in keeping with its known involvement in regulating carbon flow (16), PrrA may function in that capacity.

Separate from hemA regulation by PrrA and FnrL, the differences among the levels of β-galactosidase activities measured here for the hemA mutant sequences point to the possibility that transcription factors other than PrrA may interact at those sequences. FnrL is not a suitable candidate, based on the fact that the FNR consensus-like sequence within the hemA upstream sequences is 42 bp downstream from any of the residues that were altered in this study. Certainly, these differences, as well as the relationship between FnrL- and PrrA-mediated regulation, need to be addressed to fully understand the regulation of hemA expression.

In summary, we believe this study resolves the question as to whether or not PrrA directly regulates hemA expression, as we have found that PrrA is required for hemA transcription from the P1 promoter in vitro and that this involves binding of PrrA at two sites centered −163 bp and −67 bp upstream of the P1 transcription initiation site. Our data lead us to propose that the relative importance of PrrA binding at the two sites depends on the phosphorylation state of PrrA, which could provide a finer degree of responsiveness in the level of transcription than can be achieved by a single binding site or by a single mode of PrrA binding. We believe the latter finding may contribute to our understanding as to why there is variation with respect to the number of PrrA binding sites associated with different genes in R. sphaeroides.