Proc. Natl. Acad. Sci. USAVol. 82, pp. 3078-3081, May 1985Biochemistry

Rapid "footprinting" on supercoiled DNA(supercoiling/lac promoter)

JAY D. GRALLADepartment of Chemistry and Biochemistry and the Molecular Biology Institute, University of California at Los Angeles, Los Angeles, CA 90024

Communicated by Paul D. Boyer, December 26, 1984

ABSTRACT A DNase protection technique is describedand applied to the interaction of three lac control proteins withsupercoiled lac DNA. The technique uses end-labeled oligonu-cleotide primers to probe specific DNA regions as an alterna-tive to protocols requiring restriction endonuclease cleavage orblotting. Thus DNA may be probed with high resolution in itsnative state. It is demonstrated that the introduction of super-coiling into DNA accelerates the rate of lac ps promoter bind-ing by RNA polymerase but does not alter the positions atwhich polymerase, cAMP-binding protein, or lac repressorbind to lac DNA.

There is accumulating evidence that higher-order DNAstructure is important for the regulation of expression ofmany genes. DNA supercoiling has been implicated as beingimportant in both prokaryotes and eukaryotes (1, 2). Alter-ations in chromatin structure also appear to be an integralpart of gene activation in eukaryotes (3). However, detailedknowledge of the manner of interaction of regulatory pro-teins with supercoiled or chromatin templates has been diffi-cult to obtain. This is because most high-resolution probeexperiments require cleavage with restriction enzymes, aprocedure that destroys the native structure. Techniquesthat do not involve restriction cleavage generally yield dataof lower resolution.

Various reagents have been used to probe DNA bound byregulatory proteins. These include chemicals and drugs (4,5), light (6), and nucleases (7). The protocols have technicallimitations that have hindered high-resolution characteriza-tion of DNA-protein complexes under conditions approxi-mating closely those inside cells. They require the use ofDNA cleaved with restriction enzymes to allow introductionof a specific end-label and also require proteins to be of suffi-cient purity such that the end-label is not removed by con-taminating enzyme activities. Alternatively, one can performthe DNA cleavage on intact DNA and then attempt to purifyand end-label a small nicked restriction fragment containingthe region of interest (6).Below, a technique is described that does not require the

use of any restriction enzymes or the purification of DNAfragments. It yields information that is of quality comparableto that of the best current procedures, and it can be appliedto natural, supercoiled, DNA templates. Since the techniquerequires neither digesting and labeling the original templatenor blotting with probes, it is extremely rapid and gives datawith high resolution.DNase protection patterns ("footprints") of the three lac

operon regulatory proteins [RNA polymerase, lac repressor,and cAMP-binding protein CAP)] interacting with super-coiled DNA are presented. The template chosen was the lacpS promoter, in which a point mutation in the -10 regionallows efficient transcription in the absence of the CAP-cAMP system; CAP, however, can further stimulate lac pS

transcription. DNA supercoiling is shown to decrease thetime required for promoter binding by RNA polymerase butnot to affect the ultimate positioning of the three regulatoryproteins on lac ps DNA.

MATERIALS AND METHODSMaterials. Escherichia coli RNA polymerase, CAP, lac re-

pressor, and lac plasmids were all purified in this laboratoryand were kind gifts of James Borowiec and Anita Meikle-john. Commercial sources were used for pBR322 sequencingprimers (Pharmacia P-L Biochemicals), DNase I (Worthing-ton), and the Klenow fragment ofDNA polymerase (Bethes-da Research Laboratories).

Protocol. One microliter of 0.5 mg/ml solution of purifiedpBR322-lac DNA was diluted to 14 1.l in transcription buffer[buffer I: 30 mM Tris HCl, pH 8.0/100 mM KCl/3 mMMgCl2/0.2 mM dithiothreitol/0.1 mM EDTA/3% (vol/vol)glycerol/100 pkg of bovine serum albumin per ml]. The solu-tion was warmed to 370C, control proteins were added asdescribed below, and 1 pi of freshly prepared DNase I at0.03 mg/ml was added. (The nuclease concentration shouldbe determined by titration.) After 1 min, 15 gl of cold redis-tilled neutralized phenol and 2 ,l of 0.2 M EDTA were mixedwith the sample and incubated on ice for at least 2 min. Thesample was heated to approximately 85-90'C for 2 min andcentrifuged, and the phenol layer was removed. The bufferwas exchanged by loading directly onto a prespun SephadexG-50 column in a 1-ml syringe, followed by centrifugation toallow recovery of the DNA in the original volume (ref. 8, pp.464-466). The column had been equilibrated with 50 mMTris'HCl, pH 7.2/10 mM MgCl2. Approximately 1 ng of theappropriate 32P-end-labeled oligonucleotide primer (16-mer)was added, and the sample was heated for 4 min at 90'C todenature the template DNA and then for 15 min at 450C toallow primer hybridization. Dithiothreitol (1/20 vol of 4 mMstock) and all four deoxynucleoside triphosphates ('/1o vol ofa combined stock, 5 mM each) were added and the sampleswere equilibrated briefly at 370C. One microliter of freshKlenow DNA polymerase at 0.4 unit/,lp was added, and in-cubation was continued for 10 min at 37°C. Ten microlitersof cold stop mix (4 M ammonium acetate/20 mM EDTA)was added, followed by 3 vol of cold ethanol. Samples werethen precipitated and electrophoresed on (6% polyacrylam-ide) DNA sequencing denaturing gels according to standardprocedures (ref. 8, p. 478).

Stocks of regulatory proteins were prepared by slow dilu-tion into buffer I. One microliter of freshly prepared stockswas mixed slowly with 14-,ul DNA samples (above) to givefinal concentrations of 80 nM for RNA polymerase (activeconcentration), 1 AM nominal concentration for repressor(estimated 25% active), and 40 nM protein plus 200 AMcAMP for CAP. Incubation of CAP or repressor was for ap-proximately 5 min at 37°C. Standard incubations with poly-merase were for 12-14 min at 370C. In the case of polymer-

Abbreviations: CAP, cAMP-binding protein.

3078

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. NatL Acad Sci. USA 82 (1985) 3079

ase, heparin was then added to 20 /ug/ml and the incubationwas continued for 2-3 min prior to addition of DNase andprocessing as described above.

RESULTS

The rationale of the technique is diagrammed in Fig. 1. Su-percoiled plasmid DNA containing the region to be charac-terized is lightly digested with DNase I. This creates a col-lection of fragments with relatively random 5' and 3' termini.The DNA is deproteinized, denatured, and incubated brieflywith a synthetic 32P-end-labeled DNA primer oligonucleo-tide that hybridizes to a unique location near a small targetsite. In the example shown in Fig. 1, two commercially avail-able pBR322 primers can be used to probe the two strands ofthe inserted cloned lac operon DNA. The primer is extendedwith Klenow DNA polymerase to copy every hybridizedstrand to the 5' terminus created by nuclease digestion.Since the only source of radioactivity is the label at the endof the primer, the only radioactive strands are those with the5' end of the primer and a 3' end corresponding to a cleavagesite in the original supercoiled template. If protein coveredany of these cleavage sites on the original supercoiled DNA,then cleavage at this position may be altered and show up as

a band of altered intensity on a denaturing gel as in a conven-tional DNase partial protection experiment (7).

Fig. 2 shows the application of this protocol to cloned lacDNA. Lane 2 shows a series of bands resulting from labeledprimer 1 (see Fig. 1) extension through the lac region of plas-mid DNA lightly digested with DNase. The pattern of lightand dark bands is similar to results obtained with end-labeledrestriction fragments; these intensity differences occur dueto sequence-dependent variations in the rate of DNA cleav-age. Note that this pattern itself is not influenced by thepresence of DNA supercoiling (compare lanes 2 and 5). Theuse of primer 2 (see Fig. 1) yields a different but specificpattern of digestion products (lane 8 of Fig. 2). This repre-sents the pattern of cleavages on the opposite strand. It isimportant to note that the same physical DNA sample is be-ing probed in both cases; the sample was simply divided af-ter DNase digestion and incubated separately with eachprimer. This means that any strand-specific differences ob-served during this procedure cannot be influenced by the useof different end-labeled fragments for the two strands.The amounts of material used in this experiment are quite

small; 0.5 ,g of whole plasmid DNA and approximately 1 ngof 16-mer primer per lane. Increasing the amount of labeledprimer leads to an increase in signal intensity (not shown).The small amounts of material used are adequate to generatean excellent signal with an overnight exposure of the autora-

A2 3

B45 6

C78 9

A

IDNase I,denaturation,hybridization

32p .-

B32 p -

32 p -

jKlenow DNApolymerose

C ---P32P

deraturing gelI electrophoresis

DNase cleavage pattern

FIG. 1. Scheme for measuring protection from DNase. (A) Plas-mid pAS-21 (ref. 9), containing a 200-base-pair lac insert, is digestedlightly with DNase I. The two arrows refer to the positions ofpBR322 sequencing primers P1 and P2, which hybridize uniquely topBR322 sequences upstream from the EcoRI and HindIII sites, re-spectively. (B) One 32P-end-labeled primer is shown hybridized tothe denatured, digested DNA; fragments that do not contain a prim-er binding site are not shown. (C) The primers are extended withDNA polymerase, creating strands with lengths representinguniquely the distance between the 5' end of the primer and the pre-cise position on the plasmid DNA that was digested. Denaturing gelanalysis then reveals whether the cleavage frequency at this positionwas altered in parallel experiments in which control protein wasbound.

go z <Lscr- z~~~~~~~~~~~-

"eC~~~~~~~Li_-

N

a-

'eelbam.

FIG. 2. Protection patterns on linear and supercoiled DNA. Pat-terns were obtained with DNA linearized by BamHI cleavage (lanes1-3 and 7-9) or supercoiled DNA (lanes 4-6). Protein additions werelac repressor (lanes 1, 4, and 7), RNA polymerase (lanes 3, 6, and 9),or none (lanes 2, 5, and 8). Lanes 1-6 were probed with primer 1 andlanes 7-9 were probed with primer 2. REP 1 refers to the lac opera-tor protection and REP 2 corresponds to protection of a secondaryrepressor binding site. RNAP represents the lac promoter pattern,with an arrowhead pointing in the direction of transcription. Theautoradiograph is of 6% acrylamide/50% urea sequencing gels,which were calibrated by running the products of known sequencereactions in adjacent lanes.

Biochemistry: Gralla

ii..

Proc. Natl. Acad. Sci USA 82 (1985)

diogram. The signal is very strong because the end-labelingof short single-stranded oligonucleotide primers is very effi-cient.

Fig. 2 also shows the alterations in the digestion patternthat occur when lac repressor is bound to linear plasmidDNA prior to digestion with DNase. In the case of repressorthere is a region of strong contiguous protection, which canbe seen on both strands (Fig. 2, lanes 1 and 7) and covers thesame region identified previously (7) using end-labeled DNA(REP 1). The use of probe P1 reveals another strongly pro-tected region in the upstream flanking sequences (REP 2).This protection is very strong when the probe P1 is used(compare lanes 1 and 2) and likely corresponds to an opera-tor-like sequence detected in other experiments (10, 11). Useof the probe P2 on the very same DNA sample, however,reveals little protection of residues on the opposite strand(compare lanes 7 and 8). Gels which were run longer to ex-pand this region (not shown) consistently show weak protec-tion, and a clear strand preference seems to exist.DNase protection patterns of RNA polymerase on both

strands of lac pS DNA were also obtained (compare lanes 2with 3 and 7 with 8). As demonstrated previously (7), RNApolymerase interacts with at least 70 base pairs of promoterDNA. On both strands there is very strong protection of theapproximately 40-base-pair stretch surrounding the start-point of transcription. This corresponds to the "protected"fragments obtained in early experiments that used vast ex-cesses of DNase I to digest RNA polymerase-promotercomplexes. The upstream 30 base pairs covered by the poly-merase include DNase-hypersensitive sites interspersedwith regions of weak and strong protection. These results aresimilar to those obtained on lac UV5 DNA with end-labeledrestriction fragments (7)-.

Fig. 2 also shows the results obtained by using supercoiledDNA as a template for binding of lac repressor and RNApolymerase (lanes 4-6). Careful inspection of the results forboth strands (only one is shown) reveal no differences in in-teractions between linear and supercoiled DNA. Supercoil-ing does not "cure" areas of weak protection nor does iteliminate DNase-hypersensitive sites. The pattern due toCAP binding to supercoiled DNA is shown in Fig. 3. Again,the protection pattern is similar to that obtained previouslywith end-labeled restriction fragments (12).

Since DNA supercoiling can enhance the rate of lac tran-scription (13, 14), an experiment was performed to seewhether supercoiling enhances the rate at which the lac pspromoter can be bound by RNA polymerases SupercoiledDNA was treated with BamHI restriction enzyme, which lin-earizes the template by cleavage far from the lac promoterregion. Linear and supercoiled DNA were incubated sepa-rately with RNA polymerase for a very short time, 30 sec,much shorter than the 12 min used to generate the protectionpatterns shown in Fig. 2. Fig. 3 shows that after this shorttime, there is little or no protection when the linear DNA isused (compare lanes 1 and 3), as expected from previoustranscription studies (15). This indicates that RNA polymer-ase has had insufficient time to bind tightly to the promoter.By contrast, an identical short incubation with supercoiledDNA leads to significant protection from DNase (comparelanes 2 and 3), indicating that supercoiling has allowed poly-merase to become promoter bound very rapidly.The patterns shown in Figs. 2 and 3 also show a strong

region of protection by RNA polymerase downstream fromthe lac promoter. RNA polymerase protection is clearly seenin the region across the junction point of lac and pBR322DNA(Fig. 3, broken line). It is known that certain inserts inthis position of pBR322 partially restore promoter activity,presumably due to a partial reconstruction of the plasmid tetpromoter (16). Apparently, the reconstructed promoter inthis case binds RNA polymerase rapidly, since a strong pro-

2 3 4

"I

Or U 0_CD

FIG. 3. Effect of supercoiling on the rate of promoter binding.Protection patterns were obtained after allowing only 30 sec for thebinding of RNA polymerase to linear DNA (lane 1) or supercoiledDNA (lane 2). Lane 3 has no protein added and lane 4 has CAP asthe only protein addition. RNAP refers to protection of the lac pro-moter by RNA polymerase and CAP refers to protection of the lacCAP-binding site. The broken line refers to the RNA polymerasecovering the lac-pBR322 junction.

tection pattern is obtained even after only a 30-sec incuba-tion.

DISCUSSIONThese experiments demonstrate a method of probing DNAbinding sites that can be applied to native supercoiled DNA.The technique is much faster than previous ones and suffersfrom no loss of resolution because there is no requirementfor blotting. Using this technique, one can begin with smallamounts of whole unlabeled plasmid DNA and obtain high-quality protection patterns in a single day. Although the ap-plication in these experiments was the use of DNase on su-percoiled DNA, the general applicability should be muchwider. Since the target DNA need not be restricted or la-beled, DNA from any source should be amenable to probing.There are only two obvious requirements. First, a primermust be available that hybridizes uniquely near the targetsequence. With the current advances in oligonucleotide syn-thesis, this requirement should be met easily for most or allgenes whose sequences are known or that are cloned in avector of known sequence. Second, a method must be avail-able for lightly digesting the DNA. The use of nucleases iscommon both in vitro and in isolated nuclei, but nucleases donot penetrate living cells. However, there are various re-agents available that probably can cleave DNA inside livingcells. Therefore, it should be possible to do ultrarapid foot-printing by breaking DNA in vivo, deproteinizing, and prob-ing in vitro with labeled primer. The ability to perform thelabeling step after the nuclease digestion step also allows ex-

3080 Biochemistry: Gralla

Proc. Natl. Acad ScL USA 82 (1985) 3081

periments to be conducted with crude extracts, which nor-mally contain activities that remove the end-label from re-striction fragments.The principal limitation on the sensitivity of the technique

is likely to be obtaining sufficient amounts of template DNAfor genes that are not cloned or carried by viruses. The signalcan be increased by increasing the amount of labeled primersince the hybridization efficiency is quite low. Because thesignal is directly proportional to the specific activity of theprimer, alternative methods of creating labeled primershould also allow an increase in sensitivity. It is not yetknown whether such modifications will ultimately allowprobing of binding sites in their native chromosomal loca-tions.The initial application described here provides DNase pro-

tection patterns of regulatory proteins interacting with su-percoiled DNA. Supercoiling was shown to shorten the timerequired for lac pS promoter binding, in agreement with thereported stimulation of lac pS transcription in vitro (14). Thelac pS promoter is extremely sensitive to supercoiling, withthe transcription rate on relaxed DNA being only 2-3% ofthat attainable on fully supercoiled templates. Other promot-ers probably differ in their sensitivity to supercoiling, and alikely example of this is revealed by the experiment shown inFig. 3. Note the clear indication of polymerase bound to thepromoter fortuitously reconstructed across the lac-pBR322junction (broken line, lane 1). Protection of this sequence isvery strong after only 30 sec on linear DNA. In the same laneone can see that the lac pS promoter is not significantlybound on the very same DNA. This reconstructed sequencethus does not require supercoiling for rapid polymerase bind-ing. It is curious, however, that this reconstructed promoteris rather weak in vivo (9). It might, in fact, represent a pro-moter that is inhibited by the supercoiling present in vivo,since the protection does seem slightly weaker on super-coiled DNA after a 30-sec binding time (Fig. 3, lane 2).The patterns of lac repressor-DNA complexes reveal pro-

tection outside the lac operator of an approximately 28-base-pair segment centered near -78. The protection is quite spe-cific but asymmetric on the two strands (Fig. 2, REP 2). Thisresult is consistent with previous reports of an operator-likeelement in this general region (10, 11). These sequences havevery significant homology to the central, essential, region ofthe lac operator located 90 base pairs downstream. The ho-mology consists of two perfect matches of 7-base-pair and 5-base-pair segments separated by the same number of basepairs in the authentic and pseudo-operators. This secondarysite was not detected in the original operator-protected-frag-ment experiments (17). An explanation for this apparent dis-crepancy may be that one of the two DNA strands is onlyweakly protected, even though the site is fully occupied atmoderate repressor concentrations (Fig. 2).

It is interesting to note that this minor operator (02) over-laps somewhat with a minor promoter (P2) also present inthe lac region that initiates transcription in vitro approxi-mately 22 base pairs upstream from the normal start site (13,18, 19). Since operator (20) and promoter (4) mutations liewithin the P1/01 operator/promoter and demonstrably in-terfere with repressor (21) or RNA polymerase (15) functionsat these sites, a physiological role of the alternate P2/02 sys-tem seems unlikely. However, since a repressor molecule at02 could in principle act to repress transcription from P2,these alternative elements might represent an ancestral regu-lated lac control system. This may have been replaced even-tually by the current P1/01 apparatus, which has the advan-

tage of being subject to further fine tuning by the catabolite-control apparatus.These results, taken together with previous experiments,

reveal an extraordinary packing of protein binding sites with-in the lac operon control region. Within a 100-base-pair re-gion there are six sites, two each for RNA polymerase andlac repressor as discussed above (7, 12, 18, this work) andtwo for CAP, including a very weak site within the lac opera-tor (12). Although this may represent solely current and an-cestral regulatory systems, it may also reflect the possibilitythat regulatory proteins in general recognize common coreDNA sequences and thus have some affinity for heterolo-gous elements. The trinucleotide T-G-T appears nine timesin the 100-base-pair region and could be such a common ele-ment. In fact, the consensus recognition sequences for CAP[5'-A-A-N-T-G-T-G-A-3', in which N represents a nonspeci-fied nucleotide (22)] and the RNA polymerase -35 region(5'-T-G-T-C-A-A-3', bottom strand) have T-G-T-N-A incommon. It would be interesting to learn whether these twoproteins and certain repressors have evolved to recognizemodified forms of the T-G-T sequence motif.

I am very grateful to Leonora Poljak for advice, Selina Dwight forassistance, and James Borowiec and Anita Meiklejohn for gifts ofmaterial. This research was supported by grants from the U.S. Pub-lic Health Service (CA19941), the National Science Foundation(PCM8118712), and the Cancer Research Coordinating Committee.

1. Gellert, M. (1981) Annu. Rev. Biochem. 50, 879-910.2. Harland, R., Weintraub, H. & McKnight, S. (1982) Cold

Spring Harbor Symp. Quant. Biol. 47, 958-963.3. Mathis, D., Oudet, P. & Chambon, P. (1981) Prog. Nucleic

Acid Res. Mol. Biol. 24, 1-55.4. Sibenlist, U., Simpson, R. & Gilbert, W. (1980) Cell 20, 269-

281.5. Ross, W., Landy, A., Kikuchi, Y. & Nash, H. (1978) Cell 18,

297-307.6. Becker, M. & Wang, J. C. (1984) Nature (London) 309, 682-

687.7. Galas, D. & Schmitz, A. (1978) Nucleic Acids Res. 5, 3157-

3170.8. Maniatis, T., Fritsch, E. & Sambrook, J. (1982) Molecular

Cloning: A Laboratory Manual (Cold Spring Harbor Labora-tory, Cold Spring Harbor, NY).

9. Ackerson, J. & Gralla, J. (1983) Cold Spring Harbor Symp.Quant. Biol. 47, 473-476.

10. Gilbert, W., Majors, J. & Maxam, A. (1976) in Dahlem Work-shop on Chromosomes (Abakon, Berlin), p. 176.

11. Fried, M. & Crothers, D. M. (1981) Nucleic Acids Res. 9,6505-6526.

12. Schmitz, A. (1981) Nucleic Acids Res. 9, 277-291.13. McClure, W., Hawley, D. & Malan, T. (1982) in Promoters,

ed. Rodriguez, R. (Praeger, New York), pp. 111-120.14. Borowiec, J. & Gralla, J. (1985) J. Mol. Biol., in press.15. Stefano, J. & Gralla, J. (1982) J. Biol. Chem. 257, 13924-13929.16. Rodriguez, R., West, R., Heyneker, H., Bolivar, F. & Boyer,

H. (1979) Nucleic Acids Res. 6, 3267-3287.17. Gilbert, W. & Maxam, A. (1973) Proc. Natl. Acad. Sci. USA

70, 3581-3584.18. Spassky, A., Busby, S. & Buc, H. (1984) EMBO J. 3, 43-50.19. Reznikoff, W., Maquat, L., Munson, L., Johnson, R. & Man-

decki, W. (1982) in Promoters, ed. Rodriguez, R. (Praeger,New York), pp. 80-93.

20. Gilbert, W., Gralla, J., Majors, J. & Maxam, A. (1975) in Pro-tein-Ligand Interactions, ed. Sund, H. (de Gruyter, Berlin),pp. 193-206.

21. Jobe, A., Sadler, J. & Bourgeois, S. (1974) J. Mol. Biol. 85,231-248.

22. de Crombrugghe, B., Busby, S. & Buc, H. (1984) Science 224,831-838.

Biochemistry: Gralla