Vol. 171, No. 9

Induction of Anaerobic Gene Expression in Rhodobacter capsulatusIs Not Accompanied by a Local Change in Chromosomal

Supercoiling as Measured by a Novel AssayDAVID N. COOK, GREGORY A. ARMSTRONG, AND JOHN E. HEARST*

Department of Chemistry, University of California, Berkeley, and Laboratory of Chemical Biodynamics, LawrenceBerkeley Laboratory, Berkeley, California 94720

Received 21 March 1989/Accepted 10 June 1989

In the photosynthetic bacterium Rhodobacter capsulatus, the enzyme DNA gyrase has been implicated in theexpression of genes for anaerobic metabolic processes such as nitrogen fixation and photosynthesis. To assessthe involvement of supercoiling in anaerobic gene expression, we have developed an assay to detect in vivochanges in superhelicity of small regions of the bacterial chromosome. Our method is based on the preferentialintercalaction of psoralen into supercoiled versus relaxed DNA, and we have demonstrated the sensitivity of theassay in vivo on chromosomal regions from 2 to 10 kiobases in size. In experiments with inhibitors of gyrase,the reactivity of individual chromosomal fragments to psoralen decreases by a factor of 1.8 compared withDNA from control cultures. We used our assay to determine whether there is a change in superhelicity near thegenes coding for essential proteins for photosynthesis upon a shift from respiratory to anaerobic photosyntheticgrowth. For comparison, we also examined a restriction fragment containing theJbc operon, which codes forthe subunits of cytochrome bcl, a membrane-bound electron transport complex utilized during both aerobicand anaerobic photosynthetic growth. During this shift in growth conditions, the pufand puh mRNAs, codingfor structural polypeptides of the photosynthetic apparatus, underwent a six- to eightfold induction, while theamount of mRNA from thejbc locus remained constant. However, we detected no change in the superhelicityof either the genes for photosynthesis or those for the bc, complex during this metabolic transition. Our datathus do not support a model in which stable changes in chromosomal superhelicity regulate anaerobic geneexpression. We suggest instead that the requirement for DNA gyrase in the transcription of photosynthesisgenes results from the requirement for a swivel near heavily transcribed regions of the chromosome.

A recent model for the general control of gene expressionfor aerobic and anaerobic metabolism has postulated a rolefor DNA topology in regulating the transition betweenmetabolic modes (40). The model proposes that DNA gy-rase, which introduces negative supercoils into the chromo-some, is essential for anaerobic gene expression, whiletopoisomerase I, which relaxes the chromosome, is requiredfor expression of genes for aerobic metabolism. This hypoth-esis is based on the isolation of obligate aerobic strains ofSalmonella typhimurium which were shown to have muta-tions in one of the genes for DNA gyrase (gyrA or gyrB) andof obligate anaerobes with mutations mapping to the gene fortopoisomerase I (topA) (40). This model of a gyrase-inducedswitch in metabolic modes has been extended to otherfacultative anaerobes as well through the study of the effectsof gyrase inhibitors on expression of essential genes foranaerobic metabolism. In Klebsiella pneumoniae, expres-sion of the nifHDK genes, which code for nitrogenase andnitrogenase reductase, is blocked by drugs which inhibitgyrase (17). In Bradyrhizobium japonicum, expression ofenzymes for hydrogen metabolism, an anaerobic process, isalso repressed by drugs which target gyrase (22).For the purple photosynthetic bacterium Rhodobacter

capsulatus, recent papers have documented the inhibition ofanaerobic gene expression by drugs which inhibit gyrase.Under anaerobic conditions in the light, R. capsulatusdevelops an extensive photosynthetic membrane system inwhich pigment-protein complexes carry out light-drivenelectron transport to generate metabolic energy. R. capsu-

* Corresponding author.

latus is also capable of nitrogen fixation under anaerobic,nitrogen-limiting conditions. Kranz and Haselkorn (17) haveshown that synthesis of the R. capsulatus nifHDK geneproducts is inhibited by a 5-h treatment with the gyraseinhibitor novobiocin, whereas synthesis of most major solu-ble proteins appears to be unaffected asjudged by 3H-labeledamino acid incorporation and sodium dodecyl sulfate-poly-acrylamide gel electrophoresis (SDS-PAGE). In a study ofmRNA accumulation from essential genes for photosynthe-sis, Zhu and Hearst (43) showed that inhibition of gyraseresults in the loss of detectable mRNA from the puhA locus,which codes for the H subunit of the photosynthetic reactioncenter, and from the puf operon, which codes for thepigment-binding reaction center subunits L and M and forthe light-harvesting I antenna polypeptides. Loss of detect-able puhA mRNA occurred within 15 min after treatmentwith gyrase inhibitors, a time comparable with the half-life ofthe message. In contrast, mRNA levels for the fbc operoncoding for the cytochrome bc, complex, which is utilized forboth respiration and photosynthesis, are unaffected by gy-rase inhibitor treatment. The rapidity of the decrease inpuhA and pufmRNA has been interpreted to imply that theobserved effects result directly from gyrase inhibition andare not a secondary response to drug treatment (43).An attractive hypothesis is that DNA superhelicity in the

region of the genes for photosynthesis is altered by DNAgyrase, leading to repression or derepression of transcription(43). This model might explain why most of the knownessential genes for photosynthesis are clustered on a 46-kilobase (kb) section of the chromosome (20), as shown inFig. 1. This section of the chromosome codes for at least

4836

JOURNAL OF BACTERIOLOGY, Sept. 1989, p. 4836-48430021-9193/89/094836-08$02.00/0Copyright © 1989, American Society for Microbiology

CHROMOSOMAL SUPERCOILING AND ANAEROBIC GENE EXPRESSION 4837

puhA

4-

bch crt bch

BamHI

crt bch

N K Fr r 'i: .y :i}:: i:: i:: i:s -- i-ER--E :iSE::iED:yiS: :: 7:::: T T sw s

:::::::::: :: :::::: :::::: ::: o / ::::::::: :: ::::::: :i:: iL:7:X: :i: v # ! v / /::: ::::::: ::::::::::::::::::::: :::::::::::::::: ::::: :::::::: ::: :::: :::::: ::::::::::::::::::::::::: ::::::: :: > >:::E +E iEE iL.: 7:E :0: :X: :i:::i:|:E: ::: S S

. . . .

H H H H

FIG. 1. Physical-genetic map of the R. capsulatus photosynthe-sis gene cluster. BamHl sites are denoted by solid circles, andfragment designations are shown in capital letters. Bacteriochloro-phyll biosynthesis genes (bch) are indicated by stippled regions, andcarotenoid biosynthesis genes (crt) are indicated by crosshatchedregions. Directions of transcription for the puf and puh genes are

denoted by the arrows. Note that 35 kb of DNA separates the pufand puh promoter regions even though both operons code forstructural proteins of the photosynthetic reaction center. Locationsof M13 probes used in both Southern and RNA dot blotting are

indicated below by bars. For a more detailed map, see reference 41.

five structural proteins of the photosynthetic apparatus(pufBALM and puhA) as well as all known enzymes forcarotenoid and bacteriochlorophyll biosynthesis (32, 45).Nonphotosynthetic functions have not been mapped to thissection of the genome (45). Since the bacterial chromosomehas been shown to be composed of topologically indepen-dent domains of torsionally strained DNA (30, 38), thephotosynthesis cluster might conceivably be contained on a

single domain and thus be subject to facile regulation bygyrase. Topological isolation of the photosynthesis gene

cluster might allow it to be regulated by gyrase indepen-dently of other regions, such as the Jbc locus.

Heretofore, only an overall change in superhelicity ofchromosomes has been amenable to analysis either byethidium titration (7) or by isotopically labeled psoralenbinding (29). These methods allow one to detect changes inbulk chromosomal superhelicity but cannot test whether thetorsional state of several domains changes independentlywithout altering the average superhelical density of thechromosome. We have therefore developed an assay thatuses psoralen cross-linking to detect changes in the super-

helicity of small segments of the chromosome. Our resultsindicate that there is no stable change in the unrestrainedsuperhelicity of DNA in the photosynthesis gene clusterduring a shift of growth conditions in which genes forphotosynthesis are induced six- to eightfold. Similarly, we

observed no change in the superhelicity of DNA coding forthe constitutively expressed cytochrome bc1 complex. Ourdata thus do not support a model in which gyrase activatesgene expression by adjusting local superhelix density underanaerobic conditions. We speculate that the requirement forDNA gyrase in the transcription of anaerobically inducedgenes may result from the topological requirement for a

DNA swivel near heavily transcribed regions of the chromo-some.

MATERIALS AND METHODS

Bacterial strains and growth conditions. R. capsulatusSB1003 was grown in 150 ml of RCV medium (37) in 250-mlside-arm flasks at 32°C. Aerobically grown cultures were

sparged in the dark with a mixture of N2, 02, and CO2 in a

ratio of 80:20:2. Photosynthetic cultures were illuminated at15 W m-2 by a bank of General Electric Lumiline lamps andsparged with a mixture of N2 and CO2 in a ratio of 80:2. Gasflow rates were approximately 250 cm3 min-1 for eachculture. Gas flow rate and composition were controlled witha Matheson Gas Products Multiple Dyna-blender, model8219. Growth rates were monitored at 680 nm on a Bausch &Lomb Spectronic 21.

Irradiation of cells. Logarithmically growing cultures werequickly cooled in an ethanol-dry ice bath and pelleted for 6min at 12,000 x g in an SS34 rotor. All subsequent stepswere performed in dim light to eliminate undesired trimeth-ylpsoralen (TMP) cross-linking. Cells were suspended inice-cold TES buffer (50 mM Tris hydrochloride [pH 8], 10mM EDTA, and 50 mM NaCl) and 0.25 .g ofTMP per ml for10 min on ice. Irradiations were performed on 6-ml samplesin a six-well tissue culture plate on ice. Long-wavelength UVlight at a power density of 2 mW cm-2 was provided by abank of blacklight bulbs (Southern New England UltravioletCo.; Xmax = 350 nm). Irradiated samples were pelleted,suspended in a lysis solution consisting of 0.5% SDS, 10 mMEDTA, and 23 pLg of proteinase K per ml, and stored at-70°C for later use.DNA isolation. Cells were lysed by three freeze-thaw

cycles between a 37°C bath and dry ice. After a 20-minincubation at 37°C, DNA was extracted once with phenol-chloroform-isoamyl alcohol (25:24:1) and twice with chloro-form-isoamyl alcohol (24:1). All steps though the first chlo-roform extraction were performed in dim light. Nucleic acidswere precipitated with ethanol and suspended in 0.5 ml ofTES. RNase A digestion (30 p.g ml-1) for 15 min at 37°C wasfollowed by proteinase K digestion (50 pLg ml-') for 30 min at55°C. Samples were again extracted with phenol-chloroform-isoamyl alcohol and chloroform-isoamyl alcohol, followedby ethanol precipitation. Before restriction, samples wereirradiated with 36 kJ of UV-A per m2to cross-link any TMPmonoadducts in potentially cross-linkable sites.

Gel electrophoresis and Southern blotting. Gels and DNAsamples were handled essentially by the method of Vos andHannawalt (35). From 4 to 5 pLg of BamHI-restricted ge-nomic DNA was denatured with 0.1 volume of 1 M NaOH at55°C for 2 min, rapidly neutralized with 0.12 volume of 1 MTris hydrochloride (pH 4.2), and placed on ice. Sampleswere run on Tris-phosphate-agarose gels and blotted tonitrocellulose overnight. Southern blots were probed withprimer-extended DNA made from the single-stranded M13subclones shown in Fig. 1. The BamHI D fragment of thephotosynthesis gene cluster (20, 32) and a 9-kb BamHIfragment hybridizing tojbc were nick translated on plasmidscontaining these inserts. The probe forfbc was constructedfrom pRSF1 (8) by removing the BamHI fragments which donot contain the fbc genes. The resulting plasmid was namedpDC100. After hybridization and autoradiography, radioac-tive bands were cut from the filter, dissolved in 1 ml of ethylacetate, and counted in 5 ml of Opti-fluor (Packard Co.) in aPackard model 3385 scintillation counter.RNA isolation and dot blotting. RNA was extracted by a

scaled-down version of the procedure of Zhu and Kaplan(44). Equal amounts of purified RNA, about 5 ,ug per timepoint, were dot blotted onto a Gene Screen membrane (NewEngland Nuclear Corp.) with a Minifold dot blot apparatus(Schleicher & Schuell) by the method of Schloss et al. (28).Probes were made as stated above. In order to account forfluctuations in the amount of RNA dotted, we also prepareddots from serial dilutions with about 3 ng of RNA per timepoint. These diluted samples were probed with either labeled

VOL. 171, 1989

D L 01 E H G MJ-f

4838 COOK ET AL.

pRC1 (42), which encodes one set of R. capsulatims rRNAgenes, or with nick-translated R. capsulatus chromosomalDNA. After hybridization and autoradiography, dots werequantitated by scintillation counting as described above,except that the nylon membranes were not dissolved in ethylacetate. Data for the puf, puh, and fbc probes were normal-ized for both rDNA (see Fig. 4B and C) and chromosomalDNA hybridization (data not shown), yielding similar re-sults.

Chemicals. TMP and novobiocin were purchased fromSigma Chemical Co., St. Louis, Mo.

RESULTS

Assay for local changes in chromosomal superhelicity. DNAintercalators unwind the double helix and therefore bindmore readily to negatively supercoiled than to relaxed DNA.The psoralens are a class of intercalators which are capableof reacting photochemically with DNA (for a review, seereference 3). Upon irradiation with long-wavelength UVlight (UV-A, 320 to 380 nm), psoralens form cyclobutanemonoadducts and interstrand cross-links with pyrimidinebases of the DNA. The rate of formation of these adductshas been shown to be a function of DNA superhelicity bothin vitro and in vivo (14, 29). In experiments with Escherichiacoli, the rate of addition to chromosomal DNA in controlcells is nearly twice that of cells either treated with thegyrase inhibitor novobiocin or irradiated by a 'Co y-emit-ting source (29). These experiments utilized a radiolabeledderivative, [4,5',8-3H]TMP, to determine incorporation intototal DNA and therefore could not determine levels ofphotoaddition to individual sequences.We have extended the usefulness of psoralens for detect-

ing supercoiling in vivo by exploiting the ability of psoralen-cross-linked DNA to reanneal rapidly after denaturation(15). Our method allows us to determine the rate of TMPcross-linking to any specific DNA restriction fragment in thegenome (Fig. 2). After in vivo cross-linking at low levels(less than 1 TMP cross-link per kb), genomic DNA is purifiedand denatured. Under these alkaline conditions, cross-linkedstrands melt but are held in register by TMP so that theyreanneal rapidly upon a return to neutral pH. The cross-linkprovides a nucleation site for helix formation, whereasunmodified or monoadducted DNA is irreversibly denaturedin our procedure. Electrophoresis under native conditionsseparates cross-linked from un-cross-linked DNA, and theamount of cross-linking of any particular restriction frag-ment in a genomic digest can be quantified by Southernblotting (35).

Figure 3 shows the results of novobiocin-induced relax-ation of the chromosome as measured in the TMP assay. R.capsulatus SB1003 was grown aerobically in the dark asdescribed in Materials and Methods. A portion of the culturewas quickly cooled, pelleted, and suspended in an ice-coldbuffer containing TMP. Cells were incubated in the dark onice to allow the TMP to equilibrate and subsequently irradi-ated while on ice. The remainder of the culture was treatedwith novobiocin, harvested, and irradiated as describedabove. After isolation of genomic DNA and removal of freeTMP by organic extraction, samples were reirradiated withUV-A to cross-link any psoralen monoadducts in potentiallycross-linkable sites. Since our assay is specific for cross-links, this in vitro irradiation step substantially increases thesensitivity and improves the reproducibility of the assay.The DNA samples were then restricted and denatured im-mediately prior to gel electrophoresis. Neutralized samples

L J D |'DC

FIG. 2. Schematic illustration of the TMP assay. (A) DNA isirradiated in vivo to a level of less than 1 cross-linkable TMP adductper kb. After isolation, samples are reirradiated in vitro to ensurethe maximum yield of psoralen cross-links. (B) Following endonu-clease restriction, DNA is denatured briefly with alkali. (C) Uponneutralization, TMP-cross-linked DNA reanneals rapidly, whereasun-cross-linked DNA is irreversibly denatured. (D) Samples areelectrophoresed in an agarose gel, blotted to nitrocellulose, andprobed for the specific sequence of interest. The level of cross-linking for any fragment in the genome can be determined in thismanner. XL, Cross-linked DNA; SS, single-stranded DNA.

were run on native agarose gels and blotted, and the frag-ments of interest were visualized by hybridization.These blots characteristically contained two bands in each

lane: a lower band which ran at the position of the single-stranded DNA, and an upper cross-linked band which mi-grated at the position of double-stranded DNA. Samplesincubated with TMP but not irradiated showed no detectablecross-linked band (Fig. 3A). Longer exposure to light re-sulted in a higher percentage of cross-linked DNA. Mostimportantly, the rate of appearance of cross-linking wasnotably slower when cells were treated with novobiocin,indicating that less TMP was intercalated in the relaxedchromosome.

Blots can be quantitated by determining the fraction ofcounts at the single-stranded position for each time point.When plotted on a semilog scale, these data yielded a linearrelationship (Fig. 3B). This behavior would be expected fora Poisson-type process, in which a single psoralen cross-linkin a DNA fragment is sufficient to cause the DNA to run asa double strand (35). The ratio of cross-linking rates betweencontrol and novobiocin-relaxed DNA for restriction frag-ments in this and other experiments was 1.8 (Fig. 3B anddata not shown). We examined a number of BamHI restric-tion fragments from the photosynthesis gene cluster, includ-ing BamHI-C, -D, -F, -G, -H, -J, and -K (Fig. 1). Theserestriction fragments contain genes coding for carotenoidand bacteriochlorophyll biosynthetic enzymes and for thestructural polypeptides which bind these pigments and carryout light harvesting and primary photochemistry in thephotosynthetic membrane (32, 45). These fragments rangedin size between 2 and 10 kb, and each was sensitive to thecross-linking assay in the presence of novobiocin. Larger

J. BACTERIOL.

15:-).3

-04*-

B

CHROMOSOMAL SUPERCOILING AND ANAEROBIC GENE EXPRESSION

(+) NOVO0 .5 1 2 4 8

A

1.0

me~_

.8

0

04

Ss -t e

B

z

(I)

U')

0

C4)

(L3

wi

.2

0 2 4 6 a

Time of irradiation (min)FIG. 3. (A) Southern blot of DNA irradiated in vivo for various

amounts of time, digested with BamHI, and probed for the BamHI-K fragment of the photosynthesis gene cluster (32, 45) without (-)

and with (+) novobiocin (NOVO) treatment. Cells were grownaerobically to early log phase, and half the culture was harvested foran irradiation time course at 2 mW of UV-A per cm2 for the numberof minutes shown above each lane. The remaining cells were treatedwith 100 mg of novobiocin per ml for 15 min and then irradiated. (B)Quantitation of results from the blot in panel A and a like blot for theBamHI F fragment. Symbols: *, control culture; +, novobiocin-treated culture. SS, Single stranded; XL, cross-linked.

fragments cross-linked at a faster rate, since they repre-sented a larger target for cross-linking (Fig. 3B).mRNA accumulation during the shift from aerobic to an-

aerobic photosynthetic growth. TMP is a hydrophobic com-pound (aqueous solubility of 0.6 ,ug ml-') and, when addedto an anaerobic photosynthetic cell culture, would preferen-tially reside in the extensive photosynthetic membrane sys-tem of R. capsulatus. This fact makes a direct comparison ofchromosomal superhelicity between aerobic and photosyn-thetic cultures a difficult task. We chose instead to performthe assay during a shift from aerobic to anaerobic conditionsto alleviate any potential problems caused by sequestering ofTMP in the mnembrane of steady-state photosynthetic cells.Cultures shifted from 20 to 0% oxygen exhibited little or nogrowth for the first hour and thereafter resumed growth at aslightly faster rate than under aerobic conditions (Fig. 4A).This result is in agreement with the work of Gray (12), whoshowed that, in an identical shift from aerobic-dark toanaerobic-light conditions, Rhodobacter sphaeroides under-

-2 0 2 4 6

Time after shift (hours)

B

ZsE

056-2,

0 4

'a)

4~ 2

cr-

Shift *

*~~~~0.i*;h i f t 0

-J I I~~~~~~1

-40 0 40 80 l 20

Time after shift (min)

C

-20 -1 0 0 20 30 45 60 90 1 20

puhA

pufLN * *

fbc

FIG. 4. (A) Growth curves for two different cultures upon a shiftfrom aerobic to anaerobic photosynthetic growth conditions. (B)mRNA accumulation time course for a shifted cell culture for pufLand pufM (*), puhA (El), andfbcF, fbcB, and ]bcC (K). Data shownwere normalized to rRNA, but similar results were obtained whennormalized to total RNA (see Materials and Methods). (C) Autora-diograms of the samples from panel B.

A (-) NOVO0 .5 1 2 4 8

XL me4.iWU4

VOL. 171, 1989 4839

4840 COOK ET AL.

goes a 90-min lag phase. During this period there are no

significant increases in cell number or cell mass or in totalcellular RNA, DNA, or protein content (12).

In the first 2 h after the shift, transcription of the structuralgenes for photosynthetic reaction centers (coded for bypufLM and puhA) and the light-harvesting I antennae (codedfor by pufAB) was induced approximately six- to eightfold(Fig. 4B and data not shown). A time course of mRNAaccumulation demonstrated that these genes were inducedshortly after a rigorous aerobic to anaerobic shift and thatmRNA accumulated to a maximum approximately 90 minafter the shift, about the same time at which cell growthresumed (Fig. 4A). In comparison, mRNA for the cy-

tochrome bc, complex (coded for by the 11c operon) did not

change significantly from preshift levels (Fig. 4B). Thiscomplex is utilized for respiratory as well as photosyntheticgrowth (4) and is therefore already present in the cellularmembrane.The amount of total RNA used in dot blots for the

environmental shift time course experiment described abovewas normalized to give equal loadings by hybridization withan R. capsulatus rRNA probe, pRC1 (42). Normalization ofour data by hybridization with nick-translated genomic DNAalso gave essentially the same results (unpublished data).This was expected, since rRNA accounts for the vast

majority of total cellular RNA. Chemoheterotrophic R.capsulatus cultures shifted to low oxygen tension have beenclaimed to increase their rRNA content sevenfold during thefirst 140 min after the shift (16). We have not observed a

measurable increase in either total RNA or rRNA content inR. capsulatus cells during the time course of our environ-mental shift experiments (unpublished data). In fact, if suchan increase in rRNA were actually occurring, one wouldhave to argue that thefbc mRNA must also be coordinatelyinduced with the rRNA genes in order to account for theapparently constitutive expression pattern observed (Fig. 4Band C).Chromosomal supercoiling during the induction of genes for

photosynthetic metabolism. We used our assay to testwhether stable changes in superhelicity accompany tran-

scription of genes for photosynthesis in the switch fromaerobic to anaerobic metabolism. R. capsulatus was grown

aerobically in the dark to mid-log phase. A portion of theculture was harvested and irradiated with UV-A and TMP,while the remainder was shifted to anaerobic conditions inthe light. Forty-five minutes into the shift, when the rate ofmRNA accumulation for photosynthesis gene was maximal(Fig. 4B), the procedure was repeated on the induced cells.There was no detectable change in superhelicity for theDNA of the photosynthesis gene cluster (Fig. 5). The BamHIF fragment shown in Fig. 5 contained puhA, which codes forthe H subunit of the photochemical reaction center, whileBamHI-K was immediately downstream of puhA (Fig. 1).We also probed BamHI-J, -G, and -D to confirm this resultfor other regions of the photosynthesis gene cluster. Simi-larly, there was no detectable change in superhelicity of a

9-kb restriction fragment containing the fbc operon encodingthe bc1 complex. We repeated this experiment three timesand examined six different BamHI restriction fragmentsfrom the R. capsulatus chromosome. The deviation in cross-

linking rates between aerobic and shifted anaerobic culturesaveraged 5 3%. These results demonstrate that neitherhighly regulated genes such as pufLM and puhA nor consti-tutive genes such as the fbc operon undergo a change insuperhelicity after a shift from aerobic to anaerobic photo-synthetic growth.

-00L)

ci

ci)

W

0

C.)

U-

0 4 8

Time of Irradiation (min)FIG. 5. Comparison of cross-linking rates of aerobic (*) and

shifted anaerobic (+) cells. The irradiation of shifted cells was

performed 45 min into the adaptation period to photosyntheticgrowth. The BamHI K and F fragments are part of the photosyn-thetic gene cluster and are 2 and 4 kb in size, respectively. The bc,probe is a 9-kb restriction fragment containing all three genes fromthe Jbc operon.

DISCUSSION

We have described a novel method for detecting changesin the superhelicity of specific sequences in the bacterialchromosome. The ability to detect changes in superhelicityin local regions of the chromosome allows direct appraisal ofsome long-standing questions about the relationship betweenDNA supercoiling, chromosome structure, and transcrip-tional activation of genes. The bacterial chromosome hasbeen shown, both in vivo and in vitro, to be composed oftopologically independent domains of torsionally strainedDNA (30, 38). Complete relaxation of E. coli chromosomalsupercoils by DNase I in vitro (38) or by 'Cofy-irradiation invivo (30) indicates that there are approximately 40 of thesedomains. The number of domains correlates well with thenumber of chromosomal cleavages induced by treatment ofE. coli with the gyrase inhibitor oxolinic acid (31). It isintriguing to postulate that the superhelicity of some do-mains might be maintained at a different level from that ofother domains (13, 30). Differential superhelicity among

domains might be important in regulating transcription, sincethe utilization of a variety of promoters by RNA polymeraseboth in vivo and in vitro has been shown to be a function ofthe superhelical density of the DNA template (reviewed inreference 6).

Previously, studies involving superhelicity have been per-

formed either on total chromosomal DNA (7, 29) or on

plasmids (for example, see reference 19). The former workprovided important information about whole-chromosomestructure but did not address questions concerning specificregions of the chromosome. The latter studies used theexquisite sensitivity of gel systems which can resolve indi-vidual topoisomers and provide exact information on thelinking number of plasmids. However, studies of bacterialstrains with mutations in one of the topoisomerase genesdemonstrate that the same mutation can have differenteffects on gene expression depending on whether the gene ofinterest is carried on a plasmid or on the chromosome. Forinstance, mutations in the topA gene of S. typhimurium can

BamHI-K

\BamHI-F

bcl

I 2~~%

J. BACTERIOL.

2 6

CHROMOSOMAL SUPERCOILING AND ANAEROBIC GENE EXPRESSION 4841

suppress the leu-500 promoter mutation when the leu-SOOpromoter is found on the chromosome but not when it iscloned into a plasmid (27). Similarly, topA missense muta-tions in S. typhimurium increase both plasmid superhelicityand expression from the proU locus, which is responsible fortransport of glycine betaine in response to osmotic stress(13). In contrast, topA deletions, which also increase plas-mid superhelicity, fail to induce expression of chromoso-mally encoded proU (13).

Furthermore, observed changes in plasmid superhelicityare not always equivalent to simultaneous alterations inchromosomal supercoiling. Gyrase inhibition in E. coli re-laxes the chromosome within 30 min (7), but the sametreatment causes pBR322 to become positively supercoiled(19, 39). In a AtopA gyrB double mutant of E. coli, thechromosome is less negatively supercoiled than in the wildtype, but pBR322 from the same strain is significantly moreunderwound than plasmids isolated from the wild type, withnegative superhelicity exceeding that which can be producedin vitro by purified gyrase (24, 26). These plasmid anomalieshave recently been explained as transcriptionally inducedeffects caused by the inactivation of topoisomerase proteins(18, 33, 39). For some purposes, such as the study oftopological questions involving DNA replication or chromo-somal domain structure, the TMP assay described hereprovides a useful alternative to studies on plasmids.We can estimate the maximum possible change in chro-

mosomal supercoiling which might escape detection by ourassay. Relaxation of the chromosome by novobiocin in R.capsulatus is complete in about 20 min, as measured by theTMP assay, and changes in superhelicity can be consistentlydetected at much shorter times when the change in TMPreactivity is small (data not shown). We can reliably detect achange in the rate of cross-linking which is 15% of themaximal change caused by novobiocin (Fig. 3). This limit isgreater than the 5% random variation observed betweenaerobic and anaerobic cultures in Fig. 5. Sinden et al. (29)have shown that the rate of TMP photoreaction is linearlyproportional to the unrestrained superhelical density over arange comparable to that found in vivo. Thus, a 15%threshhold in TMP cross-linking rates would represent amaximal undetectable change of unrestrained superhelicityin vivo of 15% during a shift from aerobic to anaerobicgrowth.

Bliska and Cozzarelli (1) have recently demonstrated,using a recombinational assay, that 60% of the supercoils invivo are probably restrained by interactions with proteinsand other cellular components. They estimate that the unre-strained superhelical density in vivo is about -0.02. Thisnumber agrees with earlier estimates that about half of the invivo supercoils are restrained (23). Since intercalators aresensitive to unrestrained supercoils but not to DNA wrappedby protein, a 15% threshold in detection of chromosomalsupercoiling translates to a maximum undetectable differ-ence in unrestrained superhelical density of 0.003.Yamomoto and Droffner (40) suggest that the transition

from an aerobic to an anaerobic environment results in grosschanges in chromosomal superhelicity, which then activateexpression of a large number of genes. This model hasreceived wide attention in the scientific literature, since atrue physiological role for changes in chromosomal super-coiling as a mechanism to regulate gene expression has notyet been conclusively demonstrated (for a discussion ofsupercoiling as a response to environmental stress, seereference 13). In a recent paper, Dorman et al. (5) presentedgenetic evidence against a global role for DNA supercoiling

in the regulation of anaerobic gene expression. They mea-sured the novobiocin sensitivity of several transcriptionallacZ fusions to oxygen-regulated genes in the chromosomeof S. typhimurium. While one aerobically expressed genefusion (tonB-lacZ) is stimulated by gyrase inhibition, most ofthese oxygen-regulated fusions are unaffected by inhibitionof DNA gyrase. Furthermore, at least one anaerobicallyexpressed gene fusion is stimulated by gyrase inhibition (5).One can argue instead, based on evidence in the literature,

that the bacterial cell has evolved a homeostatic mechanismfor maintaining relatively constant chromosomal superhelic-ity (25). Expression of topoisomerase genes can respond tochanges in DNA superhelicity and thereby increase ordecrease the level of topoisomerase proteins. Transcriptionof gyrA and gyrB is stimulated by relaxed DNA (21),whereas transcription of topA is activated by increasednegative superhelicity of the template (34). This regulation isexactly that which would be required in the homeostasismodel (21). In addition, the DNA-binding affinities of the twotopoisomerases are dependent on the superhelicity of theDNA substrate, with gyrase binding best to relaxed DNAmolecules (10) while topoisomerase I prefers negativelysupercoiled DNA (36). This biochemical evidence indicatesthat the topoisomerases might be titrated over the chromo-some so as to minimize local differences in DNA superhe-licity.Our data for R. capsulatus argue that the requirement for

DNA gyrase in the transcription of photosynthesis genes (43)is not related to a gyrase-induced change in chromosomalDNA superhelicity which activates transcription. The loss ofmRNAs for photosynthesis after treatment with gyraseinhibitors could be an indirect result of inhibitor treatment.Gyrase is clearly an essential enzyme for DNA metabolism,and loss of gyrase activity might have numerous and com-plex effects on the cell which do not directly reflect itsimmediate role in bacterial physiology. On the other hand,data are accumulating that gyrase may play a direct role inthe transcription of strongly expressed genes and so may beessential to the expression of the puf and puh operons (seebelow).We emphasize that our experiments are designed to detect

stable changes in chromosomal superhelicity due to changesin gyrase activity. In control experiments, we can easilydetect the relaxation of the chromosome after gyrase inhibi-tion (Fig. 2 and data not shown). These experiments, inwhich irradiations are carried out on ice, might not besensitive to transient changes in superhelicity induced bytranscription since RNA polymerase is itself inactive underthese conditions.We favor an alternative hypothesis to that advanced by

Yamamoto and Droffner (40) for the observed dependenceon gyrase in the expression of genes for anaerobic metabo-lism in R. capsulatus (17, 43). Redirection of the cellularmetabolism to photosynthesis or nitrogen fixation involves asubstantial commitment by cells to the production of newenzyme systems. This induction, in turn, requires a highlevel of transcription of the genes for the structural proteinsinvolved. Recent evidence indicates that on highly ex-pressed regions of topologically anchored DNA, RNA poly-merase divides the template into domains and introducespositive and negative superturns by rotating the DNAthrough the transcription complex (2, 9, 11, 18, 33, 39). Therequirement for gyrase under these conditions could simplyreflect the need to maintain the chromosome at its steady-state superhelicity by restoring negative supercoils removedby the process of transcription. Thus, instead of activating

VOL. 171, 1989

4842 COOK ET AL.

transcription by altering DNA topology, gyrase could berequired, in certain circumstances, as a consequence oftranscription. We are currently performing experiments totest this model.

ACKNOWLEDGMENTS

We thank G. Drews for providing pRC1 and N. Gabellini forproviding pRSF1. We also thank D. Burke-Aguero, D. O'Brien, J.Gingrich, M. Alberti, J. Kahn, D. Falvey, and P. Spielmann forcritically reading the manuscript and Melanie Beikman for valuabletechnical assistance.

This material is based upon work supported under a NationalInstitutes of Health training grant to D.N.C. and a National ScienceFoundation Graduate Fellowship to G.A.A. This work was alsosupported in part by Public Health Service grant GM 30786 from theNational Institutes of Health and by the Office of Basic EnergySciences, Biological Energy Division, Department of Energy, undercontract DE-ACO30-76F00098.

LITERATURE CITED

1. Bliska, J. B., and N. R. Cozarelli. 1987. Use of site-specificrecombination as a probe of DNA structure and metabolism invivo. J. Mol. Biol. 194:205-218.

2. Brill, S. J., and R. Sternglanz. 1988. Transcription-dependentDNA supercoiling in yeast DNA topoisomerase mutants. Cell54:403-411.

3. Cimino, G. D., H. B. Gamper, S. T. Isaacs, and J. E. Hearst.1985. Psoralens as photoactive probes of nucleic acid structureand function: organic chemistry, photochemistry, and biochem-istry. Annu. Rev. Biochem. 54:1151-1193.

4. Daldal, F., E. Davidson, and S. Cheng. 1987. Isolation of thestructural genes for the Rieske Fe-S protein, cytochrome b andcytochrome c1, all components of the ubiquinol:cytochrome c2oxidoreductase complex of Rhodopseudomonas capsulata. J.Mol. Biol. 195:1-12.

5. Dorman, C. J., G. C. Barr, N. N. Bhriain, and C. F. Higgins.1988. DNA supercoiling and the anaerobic and growth phaseregulation of tonB gene expression. J. Bacteriol. 170:2816-2826.

6. Drlica, K. 1984. Biology of bacterial deoxyribonucleic acidtopoisomerases. Microbiol. Rev. 48:273-289.

7. Drlica, K., and M. Snyder. 1978. Superhelical Escherichia coliDNA: relaxation by coumermycin. J. Mol. Biol. 120:145-154.

8. Gabellini, N., and W. Sebald. 1986. Nucleotide sequence andtranscription of the fbc operon from Rhodopseudomonassphaeroides. Eur. J. Biochem. 154:569-579.

9. Gamper, H. B., and J. E. Hearst. 1982. A topological model fortranscription based on unwinding angle analysis of E. coli RNApolymerase binary, initiation, and ternary complexes. Cell 29:81-90.

10. Gellert, M., K. Mizuuchi, M. H. O'Dea, and H. A. Nash. 1976.An enzyme that introduces superhelical turns into DNA. Proc.Natl. Acad. Sci. USA 73:3872-3876.

11. Giaever, G. N., and J. C. Wang. 1988. Supercoiling of intracel-lular DNA can occur in eukaryotic cells. Cell 55:849-856.

12. Gray, E. D. 1967. Studies on the adaptive formation of photsyn-thetic structures in Rhodopseudomonas sphaeroides. I. Synthe-sis of macromolecules. Biochim. Biophys. Acta 138:550-563.

13. Higgins, C. F., C. J. Dorman, D. A. Stirling, L. Waddell, I. R.Booth, G. May, and E. Bremer. 1988. A physiological role forDNA supercoiling in the osmotic regulation of gene expressionin S. typhimurium and E. coli. Cell 52:569-584.

14. Hyde, J. E., and J. E. Hearst. 1978. Binding of psoralenderivatives to DNA and chromatin: influence of the ionicenvironment on dark binding and photoreactivity. Biochemistry17:1251-1257.

15. Johnston, B. H., M. A. Johnson, C. B. Moore, and J. E. Hearst.1977. Psoralen-DNA photoreaction: controlled production ofmono- and diadducts with nanosecond ultraviolet laser pulses.Science 197:906-908.

16. Klug, G., N. Kaufmann, and G. Drews. 1984. The expression ofgenes encoding proteins of B800-850 antenna pigment complex

and ribosomal RNA of Rhodopseudomonas capsulata. FEBSLett. 177:61-65.

17. Kranz, R. G., and R. Haselkorn. 1986. Anaerobic regulation ofnitrogen-fixation genes in Rhodopseudomonas capsulata. Proc.Natl. Acad. Sci. USA 83:6805-6809.

18. Liu, L. F., and J. C. Wang. 1987. Supercoiling of the DNAtemplate during transcription. Proc. Natl. Acad. Sci. USA84:7024-7027.

19. Lockshon, D., and D. R. Morris. 1983. Positively supercoiledplasmid DNA is produced by treatment of Escherichia coli withDNA gyrase inhibitors. Nucleic Acids Res. 11:2999-3017.

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

21. Menzel, R., and M. Gellert. 1983. Regulation of the genes for E.coli DNA gyrase: homeostatic control of DNA supercoiling.Cell 34:105-113.

22. Novak, P. D., and R. J. Maier. 1987. Inhibition of hydrogenasesynthesis by DNA gyrase inhibitors in Bradyrhizobium japoni-cum. J. Bacteriol. 169:2708-2712.

23. Pettijohn, D. E., and 0. Pfenninger. 1980. Supercoils in prokary-otic DNA restrained in vivo. Proc. Natl. Acad. Sci. USA77:1331-1335.

24. Pruss, G. J. 1985. DNA topoisomerase I mutants: increasedheterogeneity in linking number and other replicon-dependentchanges in DNA supercoiling. J. Mol. Biol. 185:51-63.

25. Pruss, G. J., and K. Drlica. 1989. DNA supercoiling andprokaryotic transcription. Cell 56:521-523.

26. Pruss, G. J., S. H. Manes, and K. Drlica. 1982. Escherichia coliDNA topoisomerase I mutants: increased supercoiling is cor-rected by mutations near gyrase genes. Cell 31:35-42.

27. Richardson, S. M. H., C. F. Higgins, and D. M. J. Lilley. 1988.DNA supercoiling and the leu-500 promoter mutation of Salmo-nella typhimurium. EMBO J. 7:1863-1869.

28. Schloss, J. A., C. D. Silflow, and J. L. Rosenbaum. 1984. mRNAabundance changes during flagellar regeneration in Chlamydom-onas reinhardtii. Mol. Cell. Biol. 4:424-434.

29. Sinden, R. R., J. 0. Carlson, and D. E. PettiJohn. 1980.Torsional tension in the DNA double helix measured withtrimethylpsoralen in living E. coli cells: analogous measure-ments in insect and human cells. Cell 21:773-783.

30. Sinden, R. R., and D. E. Pettijohn. 1981. Chromosomes in livingEscherichia coli cells are segregated into domains of supercoil-ing. Proc. Natl. Acad. Sci. USA 78:224-228.

31. Snyder, M., and K. Drlica. 1979. DNA gyrase on the bacterialchromosome: DNA cleavage induced by oxolinic acid. J. Mol.Biol. 131:287-302.

32. Taylor, D. P., S. N. Cohen, W. G. Clark, and B. L. Marrs. 1983.Alignment of genetic and restriction maps of the photosynthesisregion of the Rhodopseudomonas capsulata chromosome by a

conjugation-mediated marker rescue technique. J. Bacteriol.154:580-590.

33. Tsao, Y.-P., H.-C. Wu, and L. F. Liu. 1989. Transcription-driven supercoiling of DNA: direct biochemical evidence fromin vitro studies. Cell 56:111-118.

34. Tse-Dinh, Y.-C. 1985. Regulation of the Escherichia coli DNAtopoisomerase I gene by DNA supercoiling. Nucleic Acids Res.13:4751-4763.

35. Vos, J. H., and P. C. Hannawalt. 1987. Processing of psoralenadducts in an active human gene: repair and replication of DNAcontaining monoadducts and interstrand cross-links. Cell 50:789-799.

36. Wang, J. C. 1971. Interaction between DNA and an Escherichiacoli protein omega. J. Mol. Biol. 55:523-533.

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

38. Worcel, A., and E. Burgi. 1972. On the structure of the foldedchromosome of Escherichia coli. J. Mol. Biol. 71:127-147.

39. Wu, H.-Y., S. Shyy, and J. C. Wang, and L. F. Liu. 1988.Transcription generates positively and negatively supercoileddomains in the template. Cell 53:433-440.

40. Yamamoto, N., and M. L. Droffner. 1985. Mechanisms deter-

J. BACTERIOL.

CHROMOSOMAL SUPERCOILING AND ANAEROBIC GENE EXPRESSION 4843

mining aerobic or anaerobic growth in the facultative anaerobeSalmonella typhimurium. Proc. Natl. Acad. Sci. USA 82:2077-2081.

41. Youvan, D. C., and B. L. Marrs. 1985. Photosynthetic apparatusgenes from Rhodopseudomonas capsulata, p. 173-181. In K. E.Steinback, S. Bonitz, C. J. Arntzen, and L. Bogorad (ed.),Molecular biology of the photosynthetic apparatus. Cold SpringHarbor Laboratory, Cold Spring Harbor, N.Y.

42. Yu, P. L., B. Hohn, H. Falk, and G. Drews. 1982. Molecularcloning of the ribosomal RNA genes of the photosyntheticbacterium Rhodopseudomonas capsulata. Mol. Gen. Genet.188:392-398.

43. Zhu, Y. S., and J. E. Hearst. 1988. Transcription of oxygen-regulated photosynthetic genes requires DNA gyrase inRhodobacter capsulatus. Proc. Natl. Acad. Sci. USA 85:4209-4213.

44. Zhu, Y. S., and S. Kaplan. 1985. Effects of light, oxygen, andsubstrates on steady-state levels of mRNA coding for ribulose-1,5-bisphosphate carboxylase and light-harvesting and reactioncenter polypeptides in Rhodopseudomonas sphaeroides. J. Bac-teriol. 162:925-932.

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

VOL. 171, 1989