Vol. 175, No. 7JOURNAL OF BACTERIOLOGY, Apr. 1993, p. 2037-20450021-9193/93/072037-09$02.00/0Copyright X 1993, American Society for Microbiology

Analysis of the Promoter and Regulatory Sequences of an

Oxygen-Regulated bch Operon in Rhodobacter capsulatus bySite-Directed Mutagenesis

DZWOKAI MA, DAVID N. COOK, DAVID A. O'BRIEN, AND JOHN E. HEARST*

Department of Chemistry, University of California, Berkeley, and Division of Structural Biology,Lawrence Berkeley Laboratory, Berkeley, California 94720

Received 28 July 1992/Accepted 12 January 1993

The biosynthesis of pigments (carotenoids and bacteriochlorophylls) in the photosynthetic bacteriumRhodobacter capsulatus is regulated by the oxygen concentration in the environment. However, the mechanismof this regulation has remained obscure. In this study, transcriptional fusions of the bchCXYZ promoter regionto lacZ were used to identify the promoter and regulatory sequences governing transcription of thesebacteriochlorophyll biosynthesis genes. The promoter region was identified in vivo by making deletions andsite-directed mutations. The 50 bp upstream of the promoter region was shown to be required for theoxygen-dependent transcriptional regulation of bchCXYZ. A previously described palindrome sequence is alsolikely involved in the regulation. A gel mobility shift assay further defined the interaction of transcriptionregulators with these DNA sequence elements in vitro and demonstrated that a DNA-protein complex is formedat this promoter region. Since the suggested promoter sequence and the palindrome sequence are foundupstream of several other bch and crt operons, these sequences may be responsible for regulating oxygen-

dependent pigment biosynthesis at the level of transcription in R. capsulatus. In addition, these cis-acting DNAelements are not found upstream of puh and puf operons, which encode the structural polypeptides of thereaction center and light-harvesting I complexes. This observation supports the model of different regulatorymechanisms for the pigment biosynthesis enzymes and structural polypeptides required for the production ofthe photosynthetic apparatus.

Rhodobacter capsulatus is a purple nonsulfur bacteriumwhich can grow by multiple metabolic modes (12, 19). Whenthe concentration of dissolved oxygen in the environment islow, this bacterium forms an intracytoplasmic photosyn-thetic membrane that can transduce light energy into chem-ical energy. The photosynthetic apparatus consists of threemajor pigment-protein complexes, including a reaction cen-ter and two types of light-harvesting complexes, I and II.Each of these complexes is composed of several structuralpolypeptides that bind bacteriochlorophyll and carotenoidpigments. The structural polypeptides are encoded by thepuf, puh, and puc operons, and the pigment biosynthesisenzymes are encoded by multiple bch and crt operons.Oxygen tension is a major factor controlling the coordi-

nate expression of these genes. At least part of the oxygen-dependent regulation of pigment biosynthesis occurs at thetranscriptional level (6, 20, 27). Although the genes encodingthe structural polypeptides and pigment biosynthesis en-zymes are both triggered by reduced oxygen tension, twolines of evidence suggest that the regulatory mechanisms forthese two classes of operons may be different. First, Arm-strong et al. (1) have noted that several DNA sequencemotifs, a &70-like sequence and a palindrome sequence, areconserved in the regions upstream from many pigmentbiosynthesis genes but not upstream from the puf and puhoperons. Second, mutations in a newly discovered trans-regulatory factor, regA, reduce anaerobic induction of struc-tural polypeptide biosynthesis significantly but do not haveany significant deleterious effect on pigment gene expressionat the level of transcription (20). This is the most direct

* Corresponding author.

evidence for the differential regulation of photosynthesisgenes in R. capsulatus.Although a great deal has been learned recently about the

regulatory mechanism of structural polypeptide biosynthe-sis, much less is known about regulation of pigment biosyn-thesis. At least part of the reason for this discrepancy is thatthe pigment biosynthesis genes are induced to a smallerextent (two- to fivefold) at the transcriptional level thanthose encoding structural polypeptides (10- to 30-fold) (20)and are therefore more difficult to study. Recent studies havefocused on the superoperonal organization of pigment bio-synthesis and structural genes and the functional importanceof this organization for the adaptation from aerobic toanaerobic environmental conditions (reviewed in reference22). Despite the apparent importance of transcription initia-tion of pigment biosynthesis genes, both for their ownexpression and for the expression of downstream structuralgene products, the DNA regulatory sequences controllingtranscription have not yet been identified in detail for anypigment biosynthesis operon.The most intensively studied region from the regulatory

point of view is the bchCXYZ operon (formerly bchCA [3a];EMBL Data Library accession number Z11165), whichcodes for several enzymes necessary for the synthesis ofbacteriochlorophyll and is part of a superoperon that in-cludes crtEF andpufQBALMX (24, 25). Transcription initi-ation at the bchCXYZ promoter is augmented three- tofourfold by reduced oxygen tension (2, 23, 25). Young et al.(25) mapped the promoter and regulatory signals for thisoperon to a 134-bp fragment by deletion analysis. Wellingtonand Beatty (23) identified the 5' end of an RNA species fromthis region. This mRNA species was proposed to result fromtranscription initiation because its 5' site maps to a site with

2037

2038 MA ET AL.

sequence similarity to puf operon promoter sequences (23).However, Armstrong et al. (1) proposed that different se-quences in this region with similarity to an Escherichiacoli-type cr70 promoter and to a variety of prokaryotictranscription factor binding sites might function in initiatingand regulating bchCXYZ expression.

This study identifies the regulatory sequences for tran-scription initiation at the bchCXYZ promoter. lacZ tran-scriptional fusions define a minimal 85-bp DNA fragmentthat confers wild-type regulation of bchCXYZ transcriptionand demonstrate that a promoter region which has sequencesimilarity with the E. coli u70 promoter is utilized in tran-scription initiation. Deletions and site-specific mutationssuggest that a palindrome sequence and an AT-rich regionare required for proper expression. A gel mobility shift assaydemonstrates that a DNA-protein complex forms throughspecific binding to the above regions.

MATERIALS AND METHODS

Bacterial strains and conijugations. The R. capsulatusstrain used in these experiments was B100 (15). E. coli DH5a(18) was used for cloning and lacZ assays. E. coli strainsHB101 and NECO200 (18) were used in conjugation as thedonor and the helper, respectively. Plasmids were mobilizedfrom E. coli to R. capsulatus by triparental matings withpRK2013 carried in NECO200 as a helper plasmid, asdescribed previously (28). Transconjugants were repurifiedthree times on RCV plates (21) and two times on PYE plates(15) with appropriate concentrations of antibiotics (see be-low).Growth conditions. All E. coli strains were grown in LC

medium (or plates) at 370C, and all R. capsulatus strainswere grown in RCV medium (or plates) supplemented with0.1% thiamine at 320C. Antibiotics were used at the follow-ing concentrations: for E. coli, kanamycin, 50 ,ug/ml, andspectinomycin, 100 ,ug/ml; forR capsulatus, kanamycin, 10pug/ml, and spectinomycin, 10 ,ug/ml. To measure inductionof lacZ transcription from the bchCXYZ promoter, R. cap-sulatus cultures (50 ml total) harboring various plasmidswere grown aerobically in tubes (2.3 cm inner diameter by 19cm) in the dark until the optical density at 680 nm (OD680)reached 0.05. The cultures were then shifted to anaerobiccondition in the presence of light (500 W/m2) provided by abank of Lumiline lamps. Aerobic cultures were sparged witha mixture of N2-02-CO2 (80:20:2), and anaerobic cultureswere sparged with a mixture of N2-CO2 (80:2). Gas flow andcomposition were controlled with a Matheson Gas ProductsMultiple Dyna-blender, model 8219. Growth rate was mon-itored on a Bausch & Lomb Spectronic 21 spectrophotom-eter.

Plasmid construction. We constructed promoter expres-sion vectors, designed specifically for organisms with DNAof high GC content, in which a multiple cloning site (MCS)was flanked upstream by thepuc* termination hairpin (5) anddownstream by the puhA ribosome-binding sequence (26)fused to a promoterless lacZ gene derived from pMC1871(Fig. 1B) (4). A pBR322 derivative of this configuration,designated pDC400, was used for all primary cloning, includ-ing polymerase chain reaction (PCR) modification of thebchCXYZ promoter region. This MCS and promoterlesslacZ were inserted into the BglII site of plasmid pRK290 (8),which had been modified by the addition of a kanamycinresistance cartridge at the Sall site, to generate pZM400(Fig. 1A). Cloned DNA segments were moved from pDC400

puc* transcription terminator BstEII SnaBi

GCCCA CCGGC ACCCG TCGGT GGGCG CTTTT GGTAA CCTAC GTAGC

Sphl Sstil Apal puhAShine Dalgarno met gly

ATGCC CGCGG GGGCC CTAAG CTAAA GGAGG ACTAA CATGG GC

stop codons IcZ

FIG. 1. Promoter expression vector used to study the regulationof transcriptional initiation of bchCXYZ in R.* capsulatus. (A)pZM400, a promoter expression vector suitable for organisms witha high GC base content, was derived from the broad-host-rangeplasmid pRK290 (8) by insertion of a kanamycin resistance gene anda promoterless lacZ gene flanked by a GC-rich MCS. DNA se-quences upstream of bchCXYZ were fused to lacZ at the MCS sothat LacZ protein (P-galactosidase) could be monitored in a directassay for transcription initiated from the inserted DNA sequences.The construction of pZM400 and the related plasmid pZM500(spectinomycin resistance) are described in Materials and Methods.1T,puc* transcription terminator (5); SC, translational stop codonsin three reading frames; RBS, puhA ribosome-binding site. (B)Sequence of the MCS, including puc* transcription terminator,restriction enzyme sequences, translational stop codons in threereading frames, and the puhA ribosome-binding site (Shine Dal-garno).

into pZM400 by cutting both plasmids with BstEII and SstI,which cuts uniquely in the lacZ coding sequence.

Other pZM plasmids were constructed in an analogousmanner. pZM500 and its derivatives differ from pZM400 inthat they contain a spectinomycin resistance cartridge in-serted into the kanamycin resistance gene of pZM400. Wefound that spectinomycin resistance was more easily se-lected in liquid RCV cultures. For deletion analysis experi-ments, DNA upstream from bchCXYZ was cloned from theApaI site at position +79 relative to the 5' mRNA endmapped by Wellington and Beatty (23) to various other sitesin the MCS. Briefly, pZM410 contains the fragment fromApaI to SmaI at position -85; pZM420 contains the frag-ment from ApaI to SphI at position -399; and pZM430contains the fragment fromApaI to BstEII at position - 1989(Fig. 2) (1). All site-specific mutations in plasmids pZM511 topZM519 were constructed in pDC410, a pBR322 analog ofpZM410, by PCR mutagenesis (10). The nucleotide sequence

A

B

J. BACTERIOL.

R. CAPSULATUS OXYGEN-REGULATED bch PROMOTER 2039

A

-1989

-i3

-399

E 6

-85 +1 +79S ~~~~~~00 0 0

1YtE crtF

1

pZM400

pZM410

pZM420

pZM430

B

Q

0

bo

Cd

pZM400 pZM410 pZM420 pZM430FIG. 2. Deletion analysis of the bchCXYZ promoter. (A)

pZM400, pZM410, pZM420, and pZM430 contain a lacZ gene

preceded by 0, 133, 399, and 1,989 bp of DNA sequence upstreamfrom the transcription start site of bchCXYZ, respectively. (B)Relative P-galactosidase activities of the above plasmids underaerobic (+) and anaerobic (-) conditions. The activity is the mean

of three independent assays and is normalized to the value forpZM410 under aerobic conditions. Error bars represent standarddeviations.

of each mutation was verified by double-stranded sequenc-

ing.13-Galactosidase assays. P-Galactosidase (LacZ) activity

assays were performed essentially as described by Miller(17) with the following modifications. Aerobically grown

samples (1 ml) were removed from the bacterial culture 60,30, and 2 min before the shift to anaerobic conditions.Anaerobically grown samples (1 ml) were removed 4, 4.5,and 5 h after the shift. In a separate experiment, it was

determined that P-galactosidase activity reached a steadystate 4 h after the shift to anaerobic conditions (unpublisheddata). Aliquots were frozen in liquid N2 immediately aftersampling. LacZ assays were performed in duplicate for eachsample, and a mean value for the three time points foraerobic and anaerobic cultures was calculated.

Mobility shift assays. R. capsulatus cells were grown as

described above to anOD680 of 0.25 before being harvestedor shifted to photosynthetic growth conditions. Shifted cellswere then grown photosynthetically for 60 min before har-vesting. Cells were first cooled in an ice-water bath for 15min with continuous sparging with the same gas mixtureused for growth and then centrifuged for 15 min at 6,000x g

at 4'C. Each culture was resuspended in 1% of its originalvolume in 50 mM Tris-HCl (pH 7.5)-25 mM KCl-5 mMEDTA-20% (vol/vol) glycerol-0.1% Surfact-amp nonionicsurfactant (Pierce)-1 mM dithiothreitol-0.5 mM phenylme-thylsulfonyl fluoride (Sigma). Cell lysates were obtained by

one passage through a French pressure cell at 13,000 lb/in2.Large cell debris and unbroken cells were removed bycentrifugation at 4,500 x g for 10 min. Lysates were thendivided into aliquots, frozen in liquid N2, and stored at-70'C. Protein concentrations were determined by theBradford assay (Bio-Rad).Probes for mobility shift assays were prepared by cutting

the appropriate pZM plasmids with BstEII and ApaI, whichremoved the promoter insert, and purifying the desired DNAfragment in an agarose gel with a Mermaid Kit (Bio 101). Theprobe containing only the AT-rich region was synthesized byPCR and purified in the same way. These DNA fragmentswere 32p labeled with T4 DNA kinase and [_y-32P]ATP andpurified through a Sephadex G-50 spin column; 2 Rl of32P-labeled DNA probe (approximately 20 fmol) was addedto a mixture of 8 AI of nonspecific competitor DNA and 10 RIof cell lysate containing various amounts of protein. In allcases, the competitor DNA sample contained a 500-foldexcess (by weight) of poly(dI- dC) and a 200-fold excess (byweight) of sonicated pZM500. For the binding competitionexperiment, a 50-fold molar excess of specific unlabeledcompetitor DNA was also included. After incubation for 20min at room temperature, samples were loaded on a native4.5% polyacrylamide gel, run for 4 h at 8 V/cm, and thenvisualized by exposure on a Molecular Dynamics PhosphorImager.

RESULTS

Promoter expression vector for analyzing bchCXYZ tran-scriptional regulation. To study quantitatively the regulationof transcription initiation at the bchCXYZ promoter in R.capsulatus, two lacZ transcriptional fusion vectors, pZM400and pZM500, were constructed (Fig. 1A and B and Materialsand Methods). These vectors are derived from pRK290,whose broad-host-range, low-copy-number (three to fivecopies per cell [11]) RK2 replicon is stably maintained inboth E. coli and R. capsulatus (8; unpublished data). SincelacZ fusions utilize the puhA translation initiation site in allconstructs, a direct comparison of promoter strength can bemade between different promoters with these vectors (2). Byinserting various R. capsulatus DNA fragments into theMCS and by making deletions and site-specific mutations,the cis-acting regulatory elements which control expressionof the bchCXYZ operon were identified as described below.The synthesis of bacteriochlorophyll is also regulated by

light under anaerobic conditions. Under low-light condi-tions, or when growing bacterial cultures reached a suffi-ciently high density to interfere with light transmittance,transcription of the bchCXYZ operon was higher than thefourfold induction level observed in this study (data notshown). In order to separate the effects of oxygen and lighton gene regulation, the experiments reported below wereperformed at a low cell density (OD680 of 0.05) and constanthigh light intensity (500 W/m2). Under these conditions, lightwas not a growth-limiting factor, and the induction ofbchCXYZ expression was due only to a change in oxygenavailability (data not shown).

Deletions define the minimal regulatory region. To identifythe promoter and regulatory sites for transcription ofbchCXYZ, a series of plasmids were made which containeddifferent lengths of DNA upstream from the transcriptionalstart site. Each of these constructs had a 3' terminus at theApaI site at nucleotide +79 relative to the transcription startsite, as identified in experiments by Wellington and Beatty(23). Bacterial cells harboring these plasmids were sparged

VOL. 175, 1993

2040 MA ET AL.

crtE crtF bchC bchX bchY bchZ

AT-rich region

TCAGAAAGTCGCACATCCGTCTGTCGCAAAAGTGTCTAATCAAA

-35 -10 +1

TTGACA GTCGGGCG GTTCAATG CACA+++++ .... ++++

FIG. 3. Sequence of 85 bp of DNA upstream from the bchCXYZoperon. Arrows show the direction of transcription; + 1 indicates thetranscription initiation site mapped by Wellington and Beatty (23). Acr70-like sequence (including -35 and -10 regions) is marked withasterisks, and the palindrome sequence is boxed. Upstream of the-35 sequence is an AT-rich region which spans about 50 bp (1).Possible promoter or regulatory sequences identified by Wellingtonand Beatty (23) are centered at about -12 and -26 bp and aremarked by plus signs.

with a mixture of N2-02-CO2 (80:20:2) and sampled for,B-galactosidase activity. Cells were then shifted to rigor-ously anaerobic conditions (N2-CO2, 80:2) and sampled overthe next few hours until the IP-galactosidase activity reacheda new steady state. The results of this deletion experimentare shown in Fig. 2.The basal level of P-galactosidase activity from pZM400

was quite low, confirming that thepuc* transcription termi-nator was preventing significant readthrough transcriptionfrom the vector. The 85 bp of DNA upstream of thetranscription start site (Fig. 3) contained in pZM410 wassufficient to confer the same regulation of 3-galactosidaseactivity as observed for the 2,000-bp DNA fragment con-tained in pZM430 or the 399-bp DNA fragment in pZM420(Fig. 2). These data define the minimal regulatory region andare consistent with an earlier report (25). Further deletionsinto this 85-bp region resulted in a loss of inducible geneexpression (see below). Therefore, the promoter and essen-tial regulatory elements for the bchCXYZ operon are con-tained within this 85-bp DNA region.To investigate the specific promoter and regulatory se-

quence elements in detail, site-specific mutations and anadditional deletion were constructed as described below.The identity of each of the site-specific mutations wasverified by DNA sequencing, and the R. capsulatus strainsthat harbored these mutated plasmids exhibited identicalphenotypes and growth rates (data not shown).bchCXYZ transcription utilizes a promoter which has se-

quence similarity with E. coli 70 promoter. Two promotersequences have been postulated for the bchCXYZ operon(Fig. 3). Armstrong et al. (1) proposed that the sequenceTTGACA(N)16AATGAT might constitute the -35 and -10elements of an E. coli-type a7O promoter, while Wellingtonand Beatty (23) pointed out that sequences upstream of aputative transcription initiation site were similar to se-quences near the puf operon promoter. These proposedpromoter sequences overlap extensively. In order to distin-guish between these models, site-specific changes wereintroduced into each region (Fig. 4A). Certain mutations

AC (pZM514) T (pZM511)

-35 -10 +1

(AT-rich) TTGACA GT CGGGC GTGTAAG TTCAATGATACA CACA

C (pZM512) C (pZM513)

B

*.0(U(U'ZI

*0

u)cm

Cd

a).w

10-

8-

6-

4-l+- +- +- +- +-

pZM510 pZM511 pZMS12 pZM513 pZM514

FIG. 4. Effects of mutations in the putative promoter ofbchCXYZ. (A) Arrows show the locations of specific mutationsmade within the `70-like sequence (pZM511, pZM512, and pZM513)and the puf-like sequence (pZM511 and pZM514). For example, inpZM511, an A to T substitution was introduced at position -12. (B)Relative P-galactosidase activities of the putative promoter muta-tions under aerobic (+) and anaerobic (-) conditions.

increased the similarity with the consensus sequences, whileothers decreased the similarity.An A--T mutation at position -12 in pZM511 brings the

-10 element into closer agreement with the E. coli cT70consensus sequence but decreases the similarity with thepuf-like sequence. Thus, one would predict an enhancementof expression if a cr0-like promoter is utilized at bchCXYZ ora diminution of expression if the puf-like sequences areimportant for transcription initiation. The mutation inpZM511 significantly enhanced the strength of constitutivelacZ expression (Fig. 4B). T at this position is one of themost highly conserved nucleotides in the cr70 consensussequence (16), consistent with a role for the u0-like se-quence in transcription at bchCXYZ. Conversely, a T-*Cchange at position -35 in pZM512 and an A-+C mutation atposition -11 in pZM513 alter highly conserved nucleotideswithin the postulated -35 and -10 elements and decreasethe similarity of the &70-like sequence with the consensuspromoter. Both changes reduced transcription to almost anull level (Fig. 4B). In addition, constructs bearing the70-like sequence were efficiently expressed in E. coli,

although at a lower level than in R. capsulatus. Mutations inthe -10 and -35 elements had the same qualitative effectson 0-galactosidase expression in E. coli and R. capsulatus(data not shown). On the other hand, a G--C mutation atposition -24 in pZM514 decreased the similarity to thepuf-like sequence, yet it did not significantly alter the level oftranscription compared with the wild type. Taken together,these results suggest that the Cr70-like sequence, and not thepuf-like sequence, is utilized for the initiation of transcrip-tion at bchCXYZ.

Palindrome sequence required for normal regulation. Arm-strong et al. (1) also noted the existence of a nearly palin-dromic sequence, TGTAA(N)8ATACA, that has sequenceidentity with the binding sites of many prokaryotic transcrip-

J. BACTERIOL.

R. CAPSULATUS OXYGEN-REGULATED bch PROMOTER 2041

ApZM59) CC (pZM517) - -TC

l( :-35 -10 +1

(AT-rich) TAATCAAA ?TGACA GTCGGGCG [ GTTCAATG CACA

(pZM518) CC(pZM)515) I:(pZM516)

B

6

4-

2-

I+.. ..

pZMS10 pZMS15 pZM516 pZM517 pZMS18 pZM519

FIG. 5. Effects of mutations on the regulation of transcriptioninitiation. Site-specific changes were introduced into the bchCXYZpromoter region by PCR (10) and assayed for their effects on

3-galactosidase expression under aerobic and anaerobic conditions.(A) Locations of various site-specific mutations and their corre-

sponding sequences. pZM515 contains a dinucleotide substitution inthe left box of the palindrome sequence; pZM516 contains thissubstitution in the right box. pZM517 carries both of these muta-tions. pZM518 contains a deletion of the AT-rich sequences up-

stream from the -35 region of the promoter. The starting point ofthe deletion is at position -41. In pZM519, a G residue is inserted infront of the A at position -41. (B) Relative j-galactosidase activitiesof the various mutants under aerobic (+) and anaerobic (-) condi-tions. Impaired inductions of twofold or less, compared with four-fold in the wild-type pZM510, were observed for each of thesemutants.

tion factors and which is repeated at several other intergenicpositions in the photosynthesis gene cluster, overlapping theputative -10 site of the promoter region. Three mutationswhich reduce the symmetry of this palindrome were madeand analyzed for their'effects on lacZ expression (Fig. 5).Two of the changes were dinucleotide substitutions in eitherthe left half (GT-vCC in pZM515) or the right half (CA- TCin pZM516) of the palindrome. The mutation in the left half(pZMS15) reduced the anaerobic induction of the bchCXYZpromoter by twofold and elevated the overall level of lacZexpression (Fig. SB). The mutation in the right half(pZM516) also reduced anaerobic induction by twofold butdecreased the overall level of lacZ expression. The same

dinucleotide mutations were introduced simultaneously intothe palindrome region in construct pZMS17 (Fig. SA). Thischange drastically altered the sequence of the palindrome,presumably making it unrecognizable as a binding site for a

sequence-specific transcription regulator. The aerobic levelof lacZ activity in this mutant was midway between those ofthe half-site mutants (Fig. SB), and the twofold inductionobserved with either independent half-site mutant was re-

tained in the double mutant.The lacZ data presented thus far do not delineate a clear

role for the palindrome sequence in transcription of

bchCXYZ. The increase in aerobic expression in pZM515 isconsistent with a role for binding a repressor under aerobicconditions. On the other hand, the decrease in anaerobicexpression in pZM516 argues for a role in binding anactivator under anaerobic conditions. Furthermore, it isdifficult to explain why all of the palindrome mutants,including the pZM517 double mutant retain some level oftranscription stimulation if the palindrome alone is central toregulation of bchCXYZ. However, the gel mobility shiftassays described below and the impaired lacZ expressionobserved for all three mutations make us believe that thepalindrome sequence is part of the regulation apparatus.

AT-rich region plays an essential role in regulation. Up-stream of the cr70-like sequence is a 50-bp region that has anAT content of 55%, compared with a genome average of only33% AT in R. capsulatus (Fig. 3). Similar AT-rich regionshave been identified in other intergenic regions of the pho-tosynthesis cluster (1). To determine whether this sequenceplays a role in the regulation of transcription, the AT-richregion was deleted in construct pZM518, but the a70-likesequence and the palindrome sequence were left intact (Fig.5A). This deletion produced noninducible P-galactosidaseactivity (Fig. SB). Thus, the AT-rich region was required toconfer regulated expression on the bchCXYZ promoter.Moreover, a G residue was inserted within the AT-richregion between positions -41 and -42 to produce pZM519(Fig. 5A). This mutation resulted in increased levels of bothaerobic and anaerobic gene expression and reduced theinduction of the bchCXYZ promoter to twofold (Fig. SB).

In vitro analysis of the bchCXYZ promoter region by gelmobility shift assays. From the in vivo analysis, several DNAsequence motifs were assumed to be important for theregulation of bchCXYZ transcription. In order to furtherelucidate the role(s) of these sequences, gel mobility shiftassays were performed with extracts of either aerobically orshifted anaerobically grown cells. To directly compare thebinding efficiency of aerobic and anaerobic extracts, R.capsulatus was grown under vigorous aeration as describedin Materials and Methods. After a portion of the culture washarvested for preparation of an aerobic extract, the remain-ing cells were shifted to anaerobic conditions in the light andgrown for an additional 60 min. During this interval, cellgrowth is significantly slowed while the culture adapts to thenewly required metabolic conditions, and genes for photo-synthesis are activated (7). Sampling the anaerobic extractduring this adaptation period facilitates a direct comparisonbetween the aerobic and anaerobic states, since the sameculture is used over a short time interval to make bothextracts.A stable DNA-protein complex was observed when a

radioactive DNA probe containing the 85-bp wild-type pro-moter region was incubated with the crude cell lysates (Fig.6). This shifted complex was resistant to competition by alarge excess (700-fold by weight) of nonspecific DNA, dem-onstrating that the complex was formed by the binding of asequence-specific protein(s) to the probe. Under identicalassay conditions, a higher yield of complex seemed to beobtained with the aerobic cell lysate than with the shiftedanaerobic lysate (Fig. 6).To identify which sequences were involved in complex

formation, the binding experiments were also performedwith DNA probes with a variety of mutations in the pro-moter region (Fig. 7). The gel mobility shift assay with thewild-type sequence was repeated as a control (Fig. 7,pZM510, lanes 10 to 12). A DNA-protein complex withmobility identical to that of the one above was also observed

VOL. 175, 1993

2042 MA ET AL.

aerobic cell lysate anaerobic cell lysate

1/16 1/8 1/4 1/2 ix 1/16 1/6 1/4 1/2 ixlane 1 2 3 4 5 6 7 8 9 10 11 12

well _b- ' n o A

iLALboundprobe-_pu

free probe -- R

FIG. 6. In vitro mobility shift assay by cell lysates from aerobicand shifted anaerobic cultures. Equal molar amounts of 32P-labeledwild-type DNA probe from pZM510 were titrated with increasingconcentrations of cell lysate from a bacterial culture grown underaerobic conditions and then shifted to anaerobic conditions for 1 h.A value of 1x represents a total protein concentration of 5.6 mg/ml,as measured in a Bradford assay. At the same total protein content,the aerobic cell lysate exhibited an approximately twofold-higherbinding affinity for the probe DNA. Lanes 1 to 5, aerobic cell lysate;lanes 7 to 11, anaerobic cell lysate; lanes 6 and 12, probe DNA withno cell lysate added. At the highest protein concentrations, theprobe was retained in the wells (e.g., lanes 4, 5, and 11).

with a DNA probe containing a point mutation in the -35region of the cr0-like sequence (Fig. 7, pZM512, lanes 13 to15) even though this point mutation abolished transcriptionalinitiation completely. Therefore, the shifted band did notarise from the binding of RNA polymerase to its cognatepromoter. On the other hand, mutations which were clearlynot in the o&0-like sequence affected binding. The complexdid not assemble when probes that lacked the 50-bp AT-richregion (Fig. 7, pZM518, lanes 4 to 6) or contained substitu-tions within the palindrome sequence (Fig. 7, pZM515,pZM516, and pZM517, lanes 16 to 24) were used. Insertionof a single G at position -41 within the AT-rich regionessentially abolished complex formation (Fig. 7, pZM419,lanes 7 to 9). The results from these experiments areconsistent with the previous genetic analysis, since both thepalindrome sequence and the AT-rich region were requiredfor properly regulated expression of lacZ fusions. However,a probe containing the AT-rich region alone was also notsufficient to bind protein under our experimental conditions(Fig. 7, lanes 1 to 3).To study further the DNA-protein complex, a binding

competition experiment was performed (Fig. 8). Unlabeledwild-type DNA competed efficiently with wild-type labeledDNA so that, at a 50-fold molar excess, only very littlecomplex was detected in a mobility shift experiment (com-pare lanes 2 to 4, Fig. 8). This control demonstrated that thespecific DNA-binding proteins in the cell extract were lim-iting in this experiment. An unlabeled DNA with a mutationin the -35 region also effectively competed with the labeledprobe (Fig. 8, lane 5). However, DNA fragments that eitherlacked the AT-rich region (Fig. 8, lane 7) or containedmutations in the palindrome sequence (Fig. 8, lanes 9 to 11)did not effectively compete for binding, even at a 50-foldmolar excess over labeled probe. The AT-rich region alone(Fig. 8, lane 8) and a DNA fragment which contained a G

inserted into the AT-rich region (Fig. 8, lane 6) also failed toinhibit specific complex formation. Finally, even when boththe AT-rich region and the palindrome sequence werepresent on separate DNA fragments at a 50-fold excess, theshifted wild-type band still formed (Fig. 8, lane 12).

DISCUSSION

This article presents the results of experiments designed toinvestigate the DNA sequences that govern oxygen-depen-dent regulation of pigment gene expression in R. capsulatus.Site-specific mutations show that bchCXYZ utilizes a pro-moter which has sequence similarity with the E. coli u70promoter. Two distinct promoters (one constitutive and oneinducible) have been proposed for both thepuf (3) and crtEF(25) operons. The evidence presented here, however, im-plies that only a single promoter drives bchCXYZ expres-sion. Two different point mutations (pZM512 and pZM513)were made within the -35 and -10 regions of the suggestedpromoter element, and each eliminated both aerobic andanaerobic transcription. This result excludes the presence ofa second promoter during either aerobic or anaerobicgrowth. A third promoter mutation (pZM511) increasedpromoter strength but resulted in a smaller ratio of inducedto uninduced P-galactosidase activity. It has been shownpreviously that changing the kinetic parameters of the RNApolymerase-promoter interaction can simultaneously alterthe effectiveness of a regulator (13). Therefore, one possibleexplanation for the decreased induction in pZM511 is thatthis mutation makes the promoter so potent that its ability toinitiate transcription becomes less dependent on the actionof other regulators.

It is noteworthy that many of the promoter mutations inthis study were chosen from a sequence comparison to theE. coli &70 consensus sequence (Fig. 9). Even though theeffects of these mutations are consistent with the existenceof a o70-like promoter in R. capsulatus, other possibilitiescannot be ruled out. For instance, the actual RNA poly-merase recognition element may encompass additional se-quences in R. capsulatus which are not found in the E. coliOJ0 consensus. Another possibility is that the optimal R.capsulatus promoter differs from the a-70 sequence at resi-dues not examined in this study. More extensive methods,such as random mutagenesis or the isolation of spontaneouspromoter mutants, are necessary to determine whethertranscription at bchCXYZ utilizes a true a70-like promoter.With respect to the role of the palindrome sequence, our in

vivo lacZ data are ambiguous as to whether it functions asthe binding site for either a repressor or an activator protein.The fact that all three palindrome mutants still retain partialinduction argues that this palindrome cannot be the solecis-acting sequence in regulation, if it has any such function.Meanwhile, interpretation of the in vivo lacZ data is com-plicated by the possible secondary effects of these palin-drome mutations on the nearby promoter, especially for asystem like bchCXYZ that shows only a total fourfoldinduction. Actually, many of the mutations reported herehave minor or moderate effects on the level of bchCXYZtranscription. Even so, the results from the gel mobility shiftassays do suggest that the palindrome sequence is involvedin the formation of DNA-protein complex (Fig. 7 and 8).Formation of a DNA-protein complex in our gel shift exper-iment occurs only with an intact palindrome sequence. Theexact relationship between this in vitro complex and tran-scription of bchCXYZ has not been solved yet (see below).Interestingly, the same palindrome sequence was also iden-

J. BACTERIOL.

VOL. 175, 1993 R. CAPSULATUS OXYGEN-REGULATED bch PROMOTER 2043

lane

bound probe -

free probe I

AT-rich

- + *1 2 3

pZM518

- + *4 5 6

pZM519

+ *7 8 9

pZM510

+ *10 11 12

B probe

lane

pZM512 pZM515 pZM516

13 14 15 16 17 18 19 20 21

pZM517

22 23 24

bound probe -o-

free probe -O

FIG. 7. In vitro mobility shift assay with DNA probes derived from the various mutated promoter fragments. Equal amounts of wild-typeor mutant 32P-labeled DNA probes were incubated with the same amount of cell lysate (1.4 mg of total protein per ml of binding mixture).Each probe was tested against an aerobic (+) or shifted anaerobic (-) lysate for complex formation; lanes marked * contain only the probeDNA with no lysate or other protein source. Binding assays, including incubations, were done simultaneously but loaded on separate gels.The migration of free probe varied because of the length of the DNA.

A

A probe

I

t

II

--BP-

--ON-

AL L

0,40,;, ;a

A.4

A

T77..

00

2044 MA ET AL.

type

specific coldDNA competitor

ratio

o o N an U n D r- Cu.4 .4 .-4H -HH_-4 -4 <-H

z z z Z ZN N N N N E-1 N N N N E-4

04 04A A 04 04 0404A

0 lOx 50x 50x 50x 50x 50x 50x 50x 50x 50x

lane 1 2 3 4 5 6 7

well -_

bound probe -_N- __

8 9 10 11 12

-lw

_w1--m--,"~IRWMwTMPPw7-m-loppmr-411w

_ _ ___11to

free probe -_ i hI * 9 & ._'FIG. 8. Competition binding assay between labeled wild-type probe and unlabeled competitor DNA from the various mutant plasmids. A

50-fold molar excess of unlabeled DNA was added to an aerobic cell lysate to compete with labeled wild-type DNA probe in the binding ofprotein. Lanes: 1, free probe; 2, complex formed in the absence of specific competitor DNA; 3 and 4, wild-type sequence from pZM410; 5,-35 mutation from pZM412; 6, G residue inserted into the AT-rich region from pZM419; 7, the promoter and palindrome sequences frompZM418; 8, the AT-rich region alone, produced by PCR amplification; 9 and 10, left and right half mutations in the palindrome, from pZM415and pZM416, respectively; 11, simultaneous mutation of both halves of the palindrome from pZM417; 12, the AT-rich region and thepromoter-palindrome sequences on separate DNA fragments. In all lanes, nonspecific competitor DNA was included as outlined in Materialsand Methods.

ATTGTAA N16 TATCAT ... 35 bp ... crtlTTGGCA N16 TAAACT ... 77 bp...crtDTTGACA N17 AATGAT ... 54 bp ... bchCXYZ

TTGACA N15-19 TATAAT E. coli consensus

BN8 TGACAN8 TTACAN8 TTACAN8 TTACAN8 ATACAN8 TTACA

... 63 bp... bchB

... 22 bp ... bchE

... 20 bp... crtA

.51(69)bp... crtE(crtD)

... 51 bp... bchCXYZ

...143 bp...puc

N8 TTACA R. capsulatus consensus

N6-10 ACACA prokaryotic consensus

FIG. 9. Locations of r70-like sequences and palindromes in thephotosynthesis gene cluster of R capsulatus (1) (EMBL DataLibrary accession number Z11165). (A) Comparison of the U70-likesequences found 5' to the bch and crt operons in R. capsulatus withthe consensus Cr70 promoter of E. coli (16). N represents any

nucleotide. The right half of the figure indicates the number of basepairs between these or70-like sequences and the start codon of thenearest downstream gene. Note that both the -35 and -10 se-

quences and also the spacing between them are conserved in allthree cases. (B) Palindrome sequences found 5' to bch, crt, andpucoperons in R. capsulatus and the consensus derived from them.These sequences are compared with the consensus sequence ofmany prokaryotic transcription regulators, including CAP, LacI,AraC, GalR, LexA, and NifA (1, 9).

tified upstream of the R. capsulatuspuc operon (1). A recentstudy of the puc operon from a closely related organism, R.sphaeroides, indicates a possible negative regulatory role fora similar palindrome sequence (14). This raises the possibil-ity that these palindrome sequences are involved in thecoordinate regulation of pigment biosynthetic enzymes andthe structural polypeptides of the light-harvesting II com-plex, which are the major binding site for photosyntheticpigments.

Unlike the promoter mutations in pZM511 throughpZM513, which drastically increase or decrease the level ofboth aerobic and anaerobic transcription, deletion of theAT-rich region leads to constitutive gene expression. Thereare several ways by which the AT-rich region might affecttranscription. However, the most straight-forward interpre-tation, based on our data, is that the AT-rich region isnecessary for the binding of a regulatory protein(s). Thishypothesis is consistent with our gel mobility shift assays,which show that the AT-rich region is required for theformation of a DNA-protein complex. Interestingly, a sec-ond nearly palindromic sequence, TGTCT(N)8TGACA,which is similar to the one identified near the -10 element ofthe promoter, overlaps the -35 region and the promoter-proximal portion of the AT-rich region. It remains to betested whether this second palindrome sequence is essentialto the function of the AT-rich region. If this proves to be thecase, then the regulatory defects observed in pZM518 andpZM519 may be attributable to the disruption of this site.

Important questions remain to be answered about themechanism of transcriptional regulation of pigment biosyn-thesis in R. capsulatus. A key question is whether theprotein-DNA complex observed in our experiments plays a

TGTAATGTCATGTAATGTAATGTAATGTAA

TGTAA

TGTGT

F

J. BACTERIOL.

- - Impopp- I- --w- --Mmw 7-ww

R. CAPSULATUS OXYGEN-REGULATED bch PROMOTER 2045

role in repression or activation of transcription. Since thebinding experiments were performed in crude extracts thatwere isolated in the presence of oxygen, some crucial factorsmay have been lost or inactivated. The fact that mutations inboth the downstream palindrome and the upstream AT-richregion diminish the yield of complex suggests that a compli-cated nucleoprotein structure is formed at this promoter.Additional methods, such as UV cross-linking of proteins toDNA and footprinting of the protein-DNA complex in vitroand in vivo, will be required to understand more fully theregulation of transcription.

ACKNOWLEDGMENTS

We are grateful to Marie Alberti for sequencing all the constructsand to both Marie Alberti and Donald H. Burke for critically readingthe manuscript. We also thank David Koh for synthesizing oligonu-cleotides used in sequencing and in PCR mutagenesis.

This work was supported in part by National Institutes of Healthgrant GM 30786 and by the Office of Basic Energy Sciences,Biological Energy Division, Department of Energy, under contractDE-ACO30-76F00098.

REFERENCES1. Armstrong, G. A., M. Alberti, F. Leach, and J. E. Hearst. 1989.

Nucleotide sequence, organization, and nature of protein prod-ucts of the carotenoid biosynthesis gene cluster of Rhodobactercapsulatus. Mol. Gen. Genet. 216:254-268.

2. Armstrong, G. A., D. N. Cook, D. Ma, M. Alberti, D. H. Burke,and J. E. Hearst. J. Gen. Microbiol., in press.

3. Bauer, C. E., D. A. Young, and B. L. Marrs. 1988. Analysis ofthe Rhodobacter capsulatus puf operon. J. Biol. Chem. 263:4820-4827.

3a.Burke, D. H., M. Alberti, and J. E. Hearst. J. Bacteriol., inpress.

4. Casadaban, M. 1983. P-Galactosidase gene fusions for analyzinggene expression in Escherichia coli and yeast. Methods Enzy-mol. 100:293-308.

5. Chen, C.-Y., J. T. Beatty, S. N. Cohen, and J. G. Belasco. 1988.An intercistronic stem-loop structure functions as an mRNAdecay terminator necessary but insufficient for puf mRNAstability. Cell 52:609-619.

6. Clark, W. G., E. Davidson, and B. L. Marrs. 1984. Variation oflevels of mRNA coding for antenna and reaction center poly-peptides in Rhodopseudomonas capsulata in response tochanges in oxygen concentration. J. Bacteriol. 157:945-948.

7. Cook, D. N., G. A. Armstrong, and J. E. Hearst. 1989. Inductionof anaerobic gene expression in Rhodobacter capsulatus is notaccompanied by a local change in chromosomal supercoiling asmeasured by a novel assay. J. Bacteriol. 171:4836-4843.

8. Ditta, G., T. Schmidhauser, E. Yakobsen, P. Lu, X.-W. Liang,D. R. Finlay, D. Guiney, and D. R. Helinski. 1985. Plasmidsrelated to the broad host range vector, pRK290, useful for genecloning and for monitoring gene expression. Plasmid 13:149-153.

9. Gicquel-Sanzey, B., and P. Cossart. 1982. Homologies betweendifferent procaryotic DNA-binding regulatory proteins and be-tween their sites of action. EMBO J. 1:591-595.

10. Ho, S. N., H. D. Hunt, R. M. Horton, J. K. Pullen, and L. R.

Pease. 1989. Site-directed mutagenesis by overlap extensionusing the polymerase chain reaction. Gene 77:51-59.

11. Kahn, M., R. Kolter, C. Thomas, D. Figurski, R. Meyer, E.Remaut, and D. R. Helinski. 1979. Plasmid cloning vehiclesderived from plasmnids ColEl, F, R6K, and RK2. MethodsEnzymol. 68:268-280.

12. Kiley, P. J., and S. Kaplan. 1988. Molecular genetics of photo-synthetic membrane biosynthesis in Rhodobacter sphaeroides.Microbiol. Rev. 52:59-69.

13. Lanzer, M., and H. Bujard. 1988. Promoters largely determinethe efficiency of repressor action. Proc. Natl. Acad. Sci. USA85:8973-8977.

14. Lee, J. K., and S. Kaplan. 1992. cis-Acting regulatory elementsinvolved in oxygen and light control of puc. J. Bacteriol.174:1146-1157.

15. Marrs, B. 1981. Mobilization of the genes for photosynthesisfrom Rhodopseudomonas capsulata by a promiscuous plasmid.J. Bacteriol. 146:1003-1012.

16. McClure, W. R. 1985. Mechanism and control of transcriptioninitiation in prokaryotes. Annu. Rev. Biochem. 54:171-204.

17. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

18. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

19. Scolnik, P. A., and B. L. Marrs. 1987. Genetic research withphotosynthetic bacteria. Annu. Rev. Microbiol. 41:703-726.

20. Sganga, M. W., and C. E. Bauer. 1992. Regulatory factorscontrolling photosynthetic reaction center and light harvestinggene expression in Rhodobacter capsulatus. Cell 68:945-954.

21. Weaver, P. F., J. D. Wall, and H. Gest. 1975. Characterizationof Rhodopseudomonas capsulata. Arch. Microbiol. 105:207-216.

22. Wellington, C. L., C. E. Bauer, and J. T. Beatty. 1992. Photo-synthesis gene superoperons in purple nonsulfur bacteria-thetip of the iceberg. Can. J. Microbiol. 38:20-27.

23. Wellington, C. L., and J. T. Beatty. 1989. Promoter mapping andnucleotide sequence of the bchC bacteriochlorophyll biosynthe-sis gene from Rhodobacter capsulatus. Gene 83:251-261.

24. Wellington, C. L., A. K. P. Taggart, and J. T. Beatty. 1991.Functional significance of overlapping transcripts of crtEF,bchCA, and puf photosynthesis gene operons in Rhodobactercapsulatus. J. Bacteriol. 173:2954-2961.

25. Young, D. A., C. E. Bauer, J. C. Williams, and B. L. Marrs.1989. Genetic evidence for superoperonal organization of genesfor photosynthetic pigments and pigment-binding proteins inRhodobacter capsulatus. Mol. Gen. Genet. 218:1-12.

26. Youvan, D. C., E. J. Bylina, M. Alberti, H. Begush, and J. E.Hearst. 1984. Nucleotide and deduced polypeptide sequences ofthe photosynthetic reaction-center, B870 antenna, and flankingpolypeptides from R. capsulata. Cell 37:949-957.

27. Zhu, Y. S., D. N. Cook, F. Leach, G. A. Armstrong, M. Alberti,and J. E. Hearst. 1986. Oxygen-regulated mRNAs for light-harvesting and reaction center complexes and for bacteriochlo-rophyll and carotenoid biosynthesis in Rhodobacter capsulatusduring shift from anaerobic to aerobic growth. J. Bacteriol.168:1180-1188.

28. Zsebo, K. M., and J. E. Hearst. 1984. Genetic-physical mappingof a photosynthetic gene cluster from R. capsulata. Cell 37:937-947.

VOL. 175, 1993