0 1986 by The American Society of Biological Chemists, Inc. THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol . 261, No. 23, Issue of August 15, pp . 10885-10890,1986

Printed in U.S.A.

The Binding of Catabolite Activator Protein and RNA Polymerase to the Escherichia coli Galactose and Lactose Promoters Probed by Alkylation Interference Studies*

(Received for publication, February 4, 1986)

Stephanie H. Shanblatt and Arnold Revzin From the Department of Biochemistry, Michigan State University, East Lansing, Michigan 48824

The Escherichia coli galactose and lactose promoter regions have been studied by alkylation interference experiments. The data reveal those bases or phosphate groups which, when modified, prevent the binding of the catabolite activator protein (CAP) or RNA polym- erase and hence are presumably in contact with the proteins. Interference contacts made by CAP at its primary binding sites at g d and lac are quite similar, indicating that CAP-CAMP uses the same mode of bind- ing at these two operons. RNA polymerase, when bound in the presence of CAP-CAMP, exhibits contacts at the gal and lac P1-10 regions very much like those of the lac UV5 and T7 A3 promoters (Siebenlist, U., Simpson, R. B., and Gilbert, W. (1980) Cell 20, 269- 281). CAP, therefore, does not detectably alter the structure of the open complex. The binding sites for CAP and RNA polymerase at lac, as deduced from interference experiments, do not overlap. However, at gal a CAP molecule is found much closer to the enzyme, and there is competition for a set of mutual contacts. These experiments thus reveal both similarities and differences in the mechanisms whereby CAP activates transcription at catabolite-sensitive operons.

DNA-binding proteins such as the Escherichia coli catabo- lite activator protein (CAP’) and RNA polymerase interact with unique target DNA sequences and form stable complexes in the performance of their primary functions. A description of specific contacts made by such proteins with their sites of action on DNA is an integral step toward understanding how the proteins act. Protection experiments, such as DNase I footprinting, can provide information as to where on the DNA a protein is bound but often do not reveal much detail about the nucleic acid-protein interactions involved (Schmitz and Galas, 1979). Alkylation interference studies permit analysis of those bases and phosphates which are important for the formation of a specific DNA-protein complex; the experi- ments involve modifying each site and then testing whether a particular modification affects binding (Siebenlist and Gil- bert, 1980). Such interference sites define the “contacts,” points at which the DNA and protein are presumably in close proximity; taken together a set of contacts provides an imprint of the protein.

The CAP protein selectively binds upstream of many pro- moters and activates transcription of these catabolite-sensi-

*This research was supported by Grant GM 25498 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviation used is: CAP, catabolite activator protein.

tive operons when the intracellular CAMP level rises (de- Crombrugghe et al., 1984). At both lac and gal, CAP-CAMP prevents binding of RNA polymerase to alternative positions (called P2 sites) and enhances initiation from the P1 promoter at +1 (Musso et al., 1977; Malan and McClure, 1984; Spassky et al., 1984; Peterson and Reznikoff, 1985). A 21-base pair palindromic consensus sequence has been proposed as the CAP-binding site at several operons (Ebright et al., 1984). The CAP site at the lactose promoter has been well defined both genetically and biochemically; its location is between base pairs -50 and -70 relative to the transcriptional start site at +l. Based on the x-ray crystallographic structure of CAP (McKay et al., 1982) and mutational analysis, models have been proposed describing the precise way in which CAP binds to lac DNA (Ebright et al., 1984; Weber and Steitz, 1984). The salient features of these models predict that each CAP monomer of the dimeric protein interacts with successive major grooves in right-handed B-DNA in a nearly symmetric way.

Interactions of CAP with sites other than lac have as yet not been studied in as much detail. At the galactose promoter, CAP binds between -30 and -50 (Taniguchi et al., 1979; Busby et al., 1982), much closer to RNA polymerase than in the lac case. Given this difference in spacing, a precise descrip- tion of CAP binding to the gal site may shed light on the mechanism of activation at this operon as well as on the universality of proposed models for CAP action. We report here interference studies of the binding of CAP to its primary site at the gal promoter.

The central component in the activation of transcription by CAP is, of course, its effect on RNA polymerase. The formation of a complex between RNA polymerase and pro- moter DNA is a multistep process. Kinetic studies have been interpreted as indicating that CAP enhances the rate at which RNA polymerase binds to form “closed” complexes at the lac promoter but that the isomerization to a transcriptionally competent “open” complex is unaffected (Malan et al., 1984). It is not known whether an open complex formed at a pro- moter is different when CAP participates in the process. Determination of the contacts made by RNA polymerase at the wild type lac and gal promoters in the presence of CAP enables us to address this issue, as well as to elucidate possible regions of overlap as these proteins bind simultaneously to DNA.

EXPERIMENTAL PROCEDURES

DNA Restriction Fragments-The galactose promoter region be- tween the C/o1 sites at -90 and +145 was subcloned from pBdCl (DiLauro et aL, 1979) into the EcoRI site of pBR322 using EcoRI linkers. The fragment used from this subclone extended from the artificial EcoRI site at -96 to a natural BstEII site at +38 and was

10885

10886 CAP and RNA Polymerase Contacts at gal and lac DNA

uniquely labeled using [Y-~'P]ATP and polynucleotide kinase at one or the other 5' end.

The lactose promoter DNA used was the 203-base pair piece from the POP-203 plasmid; the fragment extends from -140 to +63 with synthetic EcoRI sites at each end. To produce DNA uniquely labeled at one 5' end either the EcoRI site at +67 or the PuuII site at -123 was labeled.

Proteins-E. coli RNA polymerase was purified from strain PR7 by the method of Burgess and Jendrisak (1975) as modified by Lowe et al. (1979). Purified holoenzyme was saturated with excess purified sigma factor. The protein was judged to be 45% active as determined by titration of lac L8-UV5 promoter DNA using a gel electrophoresis assay (Garner and Revzin, 1981).

E. coli CAP protein was purified from cells containing the cloned crp gene carried on pHA7 by the method of Eilen et al. (1978). The purified protein was at least 95% pure as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and was about 25% active in binding to wild type lac DNA.

DNA Modifications-End-labeled DNA was modified either with dimethyl sulfate to methylate guanine N-7 and adenine N-3 or with ethylnitrosourea to ethylate the backbone phosphates (Siebenlist and Gilbert, 1980; Hendrickson and Schleif, 1985). DNA, typically 0.1- 0.5 pg, was reacted with 1 pl of dimethyl sulfate in 200 p1 of 50 mM sodium cacodylate, 1 mM EDTA at 25 "C for 1 min. The reaction was quenched with 50 p1 of dimethyl sulfate stop buffer (Maxam and Gilbert, 1980) and ethanol precipitated twice. For the phosphate modifications 100 pl of ethanol saturated with ethylnitrosourea was added to 100 p1 of DNA in 50 mM sodium cacodylate, 1 mM EDTA, heated at 50 "C for 45 min, and subsequently ethanol precipitated twice. In both cases the dried pellets from the second precipitation were suspended in a small volume of 20 mM Tris-C1 (pH 8.0), 0.1 mM EDTA and were used directly in the binding reactions. Dimethyl sulfate protection experiments were performed as described for the interferences except that proteins were bound to DNA prior to the modification.

Binding Reactions-Reactions were performed in 40 mM Tris-C1 (pH 8.0), 10 mM MgClz, 100 mM KCl, 0.1 mM EDTA, 0.1 mM dithiothreitol, 1% glycerol, and 20 p M CAMP. In the dimethyl sulfate protection experiments the Tris buffer in the binding reactions was replaced by 50 mM sodium cacodylate (pH 8.0). Modified DNA was present at 5 nM, CAP and RNA polymerase at 50 nM. DNA was incubated with CAP for 10 min at 37 "C and, where indicated, subsequently with RNA polymerase for an additional 30 min. Heparin was added at 80 pg/ml to reactions containing RNA polymerase to remove nonspecifically bound enzyme. The binding reactions were then loaded onto a 5% polyacrylamide gel in 1 X TBE buffer (90 mM Tris, 90 mM &Boa, 2.5 mM Na2EDTA) containing 5 p M CAMP, with 0.1 volume of 25% Ficoll as the loading reagent. The gels were electrophoresed until the free DNA was well separated from the DNA-protein complexes. Following autoradiography the bands were excised and the DNA eluted overnight at 37 'C by the crush-and- soak method (Maxam and Gilbert, 1980). The recovered DNA was extracted with pheno1:chloroform (l:l), ethanol precipitated, then suspended in 43 pl of 10 mM NaPO, (pH 7) plus 7.5 p1 of 1 N NaOH and heated at 90 "C for 30 min to achieve strand scission at the site of the modification. Following neutralization with HCl and ethanol precipitation, the products were suspended in a small volume of 90% formamide and electrophoresed on 8% (201) polyacrylamide, 7 M urea sequencing gels. Maxam-Gilbert sequencing reactions of each fragment were electrophoresed in parallel (Maxam and Gilbert, 1980); the A reaction is A + G and the T reaction is C + T. Sequencing gels were dried and autoradiographed with Kodak XAR-5 film.

RESULTS

Interference experiments involve partially modifying the DNA such that each molecule in the population has no more than one modification, as in Maxam-Gilbert chemical se- quencing of DNA. The protein under study is then allowed to interact with the modified DNA. If the modification has occurred at a site that makes an essential contact with the protein, then the protein is unable to bind. Conversely, the protein is able to form stable complexes with any DNA molecule that is not alkylated at a site required for the DNA- protein interaction. The unbound DNA is then separated electrophoretically from the DNA-protein complexes. The

experiment is done under conditions of protein excess so that any free DNA is free only because it had been modified at a site essential for protein binding. Following strand scission a t the site of the modification the free and bound DNAs are compared on a sequencing gel; bands which appear in the free DNA fraction and are concomitantly absent in the bound represent points of essential contact between the protein and its target DNA-binding site (Siebenlist et al., 1980).

Contacts a t the gal Operon-Interference experiments were performed with galactose promoter DNA, 5'-end labeled at either -96 or +38; this allows visualization of the upper and lower strands, respectively. The DNA was modified either with dimethyl sulfate, which methylates the N-7 of guanines in the major groove and the N-3 of adenines in the minor groove, or with ethylnitrosourea to ethylate the backbone phosphates. The modified DNA was then mixed with either CAP-CAMP alone or CAP-CAMP followed by RNA polymer- ase. As shown in Fig. lA, lanes 2 and 3, the G at -39 is the only essential purine on the gal upper strand that interacts with CAP. There are three regions of phosphate contact with CAP (Fig. lA, lanes 12 and 131, one centered at -30, one at -39, and another at -49. Each set of these phosphate contacts is about one turn of the helix apart from the others. The lower strand shows guanine contacts at -35 and -37 (Fig. 1 B, lanes 1 and 2) as well as regions of phosphate interaction around -35 and -45 (Fig. lB, lanes 11 and 12).

When RNA polymerase is added to preformed CAMP-CAP- DNA complexes, open complexes are formed only at the P1 position, as deduced from transcription (Musso et al., 1977)' and exonuclease I11 protection (Shanblatt and Revzin, 1983) experiments. RNA polymerase at P1 makes essential contacts on the upper strand with the G a t -14 and the A at -11 (Fig. lA, lanes 4 and 5) and with phosphates at positions -9, -11, -12, -13, -14, -17, and -18 (Fig. lA, lanes 10 and 11). On the lower strand (Fig. 1B) the guanines a t -13 and -33 (lanes 3 and 4) and the phosphates at -32 and-33 (lanes 9 and 10) show evidence of strong interaction with RNA polymerase.

Note that some CAP contacts are not seen in the CAP plus RNA polymerase lanes; this arises because of the presence of the overlapping CAP-independent gal P2 promoter (Musso et al., 1977). (The analogous situation is also encountered at the lac operon due to the overlap between the CAP and P2 polymerase binding sites there (Malan and McClure, 1984; Peterson and Reznikoff, 1985).) However, any binding at P2 does not cause erroneous identification of P1 polymerase contacts despite the fact that DNA-RNA polymerase and DNA-RNA polymerase-CAP complexes migrate to the same position on a binding gel. Recall that all contacts are visual- ized by the appearance of a band in the free DNA fraction. If CAP is unable to bind because of a modification then polym- erase will not bind a t P1; the enzyme may or may not be able to interact at P2, depending upon whether that modification also interferes with P2. If polymerase does bind at P2 the "CAP contact" is lost from the free DNA lane in the CAP plus RNA polymerase reaction. However, the CAP contacts are separately determined in reactions containing only CAP so that those lost contacts are already known (e.g. the -39G at gal). A modification which prevents CAP binding and, therefore, P1 may also prevent P2 binding. This DNA band will appear in the free DNA fraction because CAP does not bind; the fact that polymerase also does not interact at P2 is moot. Finally, if CAP binds but polymerase cannot bind a t P1 due to modification of a site essential for P1 polymerase binding, then the addition of heparin to this binding reaction will remove the CAP and the DNA will appear in the free

S. H. Shanblatt and A. Revzin (1986) Biochemistry, in press.

CAP a n d RNA Polymerase Contacts at gal a n d lac DNA 10887

1 2 3 4 5 6 7 8 9 1 0 1 1 12

F B F B G A T C B F B F

.+ I

. -10

9-20

*-30

0-60

- 70

8-60

-50

.- 40

.-20

.-lo

.+l

FIG. 1. CAP and RNA polymerase interference contacts at the gal operon. A , the upper strand of gal DNA, 5'-end labeled at -96, is visualized. Numbers at right are relative to the start of P1 transcription at +l. Lunes 1-5 show purine contacts; the DNA was pretreated with dimethyl sulfate. Lunes 10-14 show phosphate contacts determined after pretreatment with ethylnitrosourea. Lunes 6-9 are Maxam-Gilbert sequencing reactions. Lanes 2 and 3, and 13 and 12, are the free ( F ) and bound ( B ) fractions of CAP-only reactions; lanes 4 and 5, and I 1 and 10, show the free and bound DNA from CAP plus RNA polymerase reactions. Control lanes I and 14 (Co) show modified DNA which was used in the binding experiments. Solid circles between each set of free and bound DNA bands indicate sites of interference. B, the lower strand, 5'-end labeled at +38, where +1 is the gal P1 transcriptional start, is shown. Lanes 1-4 reveal purine contacts, 5-8 are Maxam-Gilbert sequencing reactions, and 9-12 yield phosphate contacts. Lunes I and 2 and 12 and I 1 are the free ( F ) and bound ( B ) fractions of CAP- only reactions; lanes 3 and 4, and 10 and 9 contain the free and bound DNA from the CAP plus RNA polymerase experiments.

fraction because of the P1 interference contact, which is the desired result. In summary then, because the binding of CAP leads exclusively to polymerase interacting at P1 (i.e. no P2 complexes formed) and the CAP contacts are independently determined in the absence of polymerase, no P1 contacts are either obscured or falsely attributed to P1 as a consequence of the presence of the overlapping P2 site.

Contacts at the lac Operon-The analogous experiment to that depicted in Fig. 1 was also performed on the lactose promoter region. Fig. 2 shows a gel illustrating the contacts between CAP and RNA polymerase and the upper strand of lac DNA. The G at -68, when methylated, interferes with CAP binding as do three phosphates in its vicinity and three

more located 10 base pairs downstream. These data agree well with those previously reported by Majors (1975a). RNA po- lymerase, when bound at the lac P1 promoter in the presence of CAP, makes two guanine and several phosphate contacts in the -10 region of the promoter. One additional guanine and three phosphate contacts were also observed in the -35 region on the upper strand of lac DNA. No phosphate or purine contacts attributable to RNA polymerase were ob- served on the lac lower strand.

DISCUSSION

Comparison of the gal and lac CAP-binding Sites-The contacts observed on both DNA strands of the two promoters

10888 CAP and RNA Polymerase Contacts at gal and lac DNA

1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3

F B F B G A T C B F B F C o -

,+ 1

+lo

”0

1-30

B- 40

450

b-60

B-70

p-80

FIG. 2. CAP and RNA polymerase interference contacts at the lac operon. The lac DNA was 5’-end labeled at -123; thus data for the upper strand are seen here. Numbers at right are relative to the P1 transcriptional start site a t +l. In lanes 1-4 the DNA was pretreated with dimethyl sulfate and in lanes 9-13 with ethylnitro- sourea; lanes 5-8 are sequencing reactions. Lanes 1 and 2, and 12 and 11 are the free ( F ) and bound ( B ) fractions of CAP-only binding reactions; lanes 3 and 4, and 10 and 9, the free and bound DNA from CAP plus RNA polymerase experiments. The control (Co) , lane 13, is the ethylated DNA used in the binding reactions of lanes 9-12. Interference sites are indicated by solid circles between each set of free and bound DNA lanes.

are depicted in Fig. 3. The sequences have been aligned according to the 21-base pair CAP-binding site consensus of Ebright et al. (1984), that sequence being A-A-N-T-G-T-G- A-N-N-T-N-N-N-T-C-A-N-A-T-W where N is unspecified and W is either an adenine or thymine. This sequence is partially symmetric, with its center between bases 11 and 12. The gal sequence is oriented opposite to lac; the essential

-75 & .1*1 O-86 &JJ -OS

lac G T T A A T T A C A C T C A A T C G A ~ T @ A G T C A A T T A A T O T G A G T T A G C T C A C T C A

TT 1 ’ fT 0 O t t FIG. 3. Comparison of gal and lac CAP-binding sites. The

sequences have been aligned according to the CAP consensus of Ebright et al. (1984), denoted by the numbering in the center. In each case a 5‘-end is at the upper left of the sequence. They are in reverse orientation relative to the start of P1 transcription at +l. Arrows are sites of phosphate interference, circled bases indicate strong purine interference, and dotted circles weaker interference. Closed circles above or below a base denote protection of purines from attack by dimethyl sulfate (these protection experiments are described under “Experimental Procedures”; data not shown).

TGTGA regions (bases 4-8 of the consensus) observed at all CAP-binding sites are aligned. The pattern and spacing of both the purine and phosphate contacts are strikingly similar for both gal and lac, providing strong evidence that CAP binds the same way a t these two sites. This result supports the notion that the models for CAP binding proposed by Weber and Steitz (1984) and Ebright et al. (1984) are correct not just for lac, but for other CAP sites as well.

It has been observed that the binding of CAP to its primary gal site is much weaker than to the lac CAP site (Kolb et al., 1983). Despite the apparent overall similarity of the contacts at these two sites, two important contacts are missing in gal. The two guanines at consensus base pairs 16 and 18 of lac are absent from the gal sequence, and CAP cannot interact in the same way with the adenine and thymine which replace them. Note in Fig. 3 that these two guanine residues are at symmet- rically opposite positions in the palindrome to the guanines at positions 5 and 7, which are part of the highly conserved TGTGA sequence. CAP interacts in the major groove at both sets of guanines in lac, but, because the sequences differ, can make but one set of contacts at gal. The inability of CAP to make these important contacts with gal DNA may account for its 10-fold reduced affinity for this region relative to the lac site.

Binding of RNA Polymerase to the gal and lac PI Pro- moters-The interactions between RNA polymerase and pro- moter sites have been studied in detail. Siebenlist et al. (1980) have reported interference data for the T7 A3 and lac UV5 promoters; strong contacts were observed at the -10 and -35 regions with good homology between the two promoters. We were interested in whether CAP has any effect on the essential polymerase contacts. That is, does polymerase bind to the promoter site in the same manner irrespective of whether CAP is prebound upstream of the enzyme-binding site. The purine and phosphate contacts made by polymerase at the gal and lac promoters in the presence of CAP are illustrated in Fig. 4. The lac UV5 contacts, determined in the absence of CAP by Siebenlist et al. (1980), are included for comparison. The wild type lac promoter exhibited only those contacts also observed on lac UV5. No new sites were detected, though a few of the UV5 contacts were not seen on lac+, e.g. phosphates a t -18, -31, and -39 and purines at -11 and -32. No

CAP and RNA Polymerase Contacts at gal and lac DNA 10889

-m 0

v

FIG. 4. Comparison of gal and lac P1 promoter interactions with RNA polymerase. The sequences are aligned relative to the start site of P1 transcription at +l. The lac UV5 contacts (Siebenlist et al., 1980) are included for comparison. Only the upper strand of lac+ is shown, as no contacts were observed on the lower strand. Arrows indicate sites of strong phosphate interference, dotted arrows weaker contacts; circled bases show purine interference. Closed circles indicate protection of purines from dimethyl sulfate modification, and carats denote sites of enhanced purine modification by dimethyl sulfate (the protection experiments are described under “Experimental Procedures”; data not shown).

interference sites were detected between the CAP site and -38, implying that polymerase does not first bind at some upstream site, such as P2, and then slide into the P1 position (Siebenlist et al., 1980; Spassky et al., 1984). The overall similarity of the interference patterns for the lac wild type and UV5 promoters suggests that CAP does not alter the contacts essential for lac P1 open complex formation.

The galactose P1 promoter contacts also show strong ho- mology with those of lac UV5 in the -10 region (Fig. 4). No contacts attributable to polymerase were detected in the -35 region of gal P1 (see below). This is not surprising given the placement of CAP at gal. Comparison with Fig. 3 reveals that CAP contacts around gal -35 are quite similar to those made by RNA polymerase near lac -35. (Note that the upper and lower strands as well as the numbering orientation of the gal DNA are reversed in Fig. 3 compared to Fig. 4.) Despite the fact that CAP may occlude the polymerase -35 binding site a t gal, the pattern of contacts observed in the -10 region strongly suggests that CAP does not change the overall mode of polymerase binding at P1 or the interactions required for forming an open complex.

Implications for the Mechanism of CAP Action at the lac and gal Promoters: CAP-RNA Polymerase Interactions-The precise way in which CAP stimulates transcription at catab- olite-sensitive operons is not well understood. As indicated above, both the lac and gal control regions have overlapping P1 and P2 RNA polymerase binding sites. In each case, prebinding of CAP prevents polymerase from forming open complexes at P2. The DNA-protein contacts reported here shed light on a number of interesting features of CAP-RNA polymerase interactions at these prototype catabolite-sensi- tive operons.

We find that RNA polymerase makes mainly the same contacts a t lac P1 as at the mutant CAP-independent L8- UV5 promoter. Therefore, the conclusions drawn by Sieben- list et al. (1980) remain valid for the wild type site. The enzyme and activator protein do not have overlapping con- tacts when they bind simultaneously to the DNA. Their linear positioning is such that protein-protein interactions are fea- sible.

At gal the situation is somewhat different. Here the primary CAP-binding site is between base pairs -30 and -50, but a second CAP molecule is bound at the -50 to -68 region when transcriptionally competent complexes form at P1 (Shanblatt

and Revzin, 1983). Polymerase contacts at the gal P1 -10 region are very similar to those for lac P1, lac UV5, and T7 A3. By analogy with those promoters one would expect phos- phate contacts a t -37 to -40 of gal PI. However, these are also positions which, when modified, exclude CAP binding at gal.

The question of whether CAP or RNA polymerase actually makes DNA contacts around -38 cannot be rigorously re- solved by these data. We note that DNase I protection studies at the gal regulatory region show marked enhancement of the -49A band both when CAP is bound by itself and when RNA polymerase is also present at PI (Shanblatt and Revzin, 1983). Thus, CAP remains at its primary site when polymerase binds. In addition, the presence of the enzyme probably does not grossly distort the CAP molecule; if it did the CAP configu- ration which favors DNase I digestion at -49 would be altered and the enhancement would disappear. Of course one could make a similar argument with respect to RNA polymerase, which makes the correct contacts at the gal P1 -10 region when CAP is present, a situation which might not prevail if the proper polymerase contacts around -38 are prohibited by CAP binding. At this time we speculate that CAP, not RNA polymerase, makes the contacts around -38. This conclusion derives from the idea that the much larger RNA polymerase molecule may have more flexibility, so that distortions around -38 might not be transmitted to that part of the protein which is in proximity to the -10 region. Also, CAP is known to be excluded from binding if the -37 to -40 region is modified we are precluded from obtaining direct evidence that this is so for RNA polymerase at P1 by the fact that the enzyme binds stably at P2 in the absence of CAP.

Regardless of whether CAP or RNA polymerase adjoins the DNA around -38, competition for contacts there will reduce the binding energy of the polymerase-P1 promoter interac- tion. This is more than overcome by favorable CAP-polym- erase interactions and by CAP-CAP interactions which occur as the second CAP molecule binds and stabilizes open com- plexes at gal P1.

Fig. 5 is a schematic representation of CAP and RNA polymerase bound at the lac and gal P1 promoters. The end- on diagram indicates that CAP has roughly the same orien- tation about the helix axis in both cases, but the side view shows the different locations of CAP along the DNA with respect to polymerase a t P1. It is of interest that CAP has a

10890 CAP and RNA Polymerase Contacts at gal and lac DNA I

-50 +I I I

n FIG. 5. Schematic diagram of CAP and RNA polymerase

positions at the lactose and galactose promoters. Upper, view down helix axis. Horizontal hatching represents CAP at its primary site at gal; uertical hatching symbolizes CAP at lac. Polymerase binding is indicated by two open areus denoting contacts in the -10 and -35 regions of the P1 promoter. Lower, side view showing RNA polymerase (oual) plus two CAP molecules (hexagons) at gal, one CAP at lac. No polymerase contacts are detected upstream of -40 (or downstream of +l); hence extension of the enzyme to the -50 to -55 region (and to +20) shown here is based on DNase I protection data (Schmitz and Galas, 1979; Shanblatt and Revzin, 1983).

rather larger stimulatory effect on transcription at the lac P1 promoter (-20-fold) (Majors, 1975b) than at gal P1 (-6-fold) (Musso et al., 1977). Malan and McClure (1984) concluded that CAP affects only the binding constant (Kg) for RNA polymerase closed complex formation at the lac P1 position, not the isomerization rate ( k t ) of closed to open complexes. This result seems consistent with the modest amount of protein-protein contact likely at lac. However, Lorimer and Revzin (1986) interpret their data on CAP-polymerase bind- ing in terms of an effect on k2 as well. Shanblatt and Revzin' resolved the limited CAP stimulation at gal P1 into contri- butions from both K g and k2.

Finally, the interference data reveal contacts which may be essential for the exclusion of P2 by a bound CAP molecule. We find that CAP binding is prohibited by phosphate eth-

ylation at bases -36 of gal and -54 of lac and by modification of the -37G of gal and the -55G of lac. RNA polymerase contacts at gal P2 have been determined (data not shown) and as expected are consistent with those at the other pro- moters described above. Since gal P2 initiates at nucleotide -5 and lac P2 at -22, there are crucial contacts which prevent CAP binding located in each case about 32 base pairs from the start point of transcription. We propose that these are essential contacts for polymerase binding at P2 (measured for gal P2, inferred for lac P2 assuming that contacts there resemble those at the other promoters). Thus, if CAP is bound a t either lac or gal, RNA polymerase cannot make the critical contacts and does not bind to the P2 site. Meiklejohn and Gralla (1985) showed that high concentrations of CAP can increase the rate at which polymerase bound at P2 is re- 1eased.This may involve fluctuations in enzyme conformation which allow CAP to usurp the -32, -33 contacts, thus en- hancing RNA polymerase dissociation from the P2 position.

To conclude then, these data indicate that interactions of CAP and RNA polymerase at the gal and lac operons have many similarities in spite of obvious differences in the posi- tions of the primary CAP-binding sites and in the stoichi- ometry of the systems (Garner and Revzin, 1982; Fried and Crothers, 1983; Shanblatt and Revzin, 1983). Essential polym- erase contacts in open complexes appear to be the same for all promoters tested, and the CAP contacts are homologous a t lac andgal (though the linear DNA sequences are reversed). A major difference is the set of mutual contact points for CAP and RNA polymerase at gal P1, which presumably relate to places on the polypeptides where protein-protein interactions are occurring.

Acknowledgments-We thank Sankar Adhya and Forrest Fuller for plasmids, and Lloyd LeCureux for expert technical assistance.

REFERENCES Bur ess R R , and Jendrisak, J. J. (1975) Biochemistry 14,4634-4638

decrombrugghe, B., Busby, S., and BUC, H. (19&1) Science 224,831-838 Bus%y, &, Aiba, H., and decrombrugghe, B. (1982) J. Mol. Biol. 154, 211-227

DiLauro, R., Taniguchi, T., Musso, R., and decrombrugghe, B. (1979) Nature 279.494-500

Ebright, R. H., Cossart, P., Gicquel-Sanzey, B., and Beckwith, J. (1984) Proc.

Eilen, E., Pampeno, C., and Krakow, J. S. (1978) Biochemistry 17,2469-2473 Fried, M. G., and Crothers, D. M. (1983) N+ic Acids Res. 1 1 , 141-158 Garner, M. M., and Revzin, A. (1981) N + x c Acids Res. 9,3047-3060

Hendrickson, W., and Schleif, R. (1985) Proc. Natl. Acad. Sci. U. S. A. 8 2 , Garner, M. M., and Revzin, A. (1982) Bcochernwtry 21,6032-6036

Natl. Acad. Sa. U. S. A. 81, 7274-7278

3129-3133 Kolb, A., Spassky, A., Chapon, C., Blazy, B., and Buc, H. (1983) Nucleic Acids

Lorimer, D. D., and Revzin, A. (1986) Nucleic Acids Res. 14,2921-2938 Lowe, P. A,, Hager, D. A, and Burgess, R. R. (1979) Biochemistry 18, 1344-

. - -. . -. .

Res. 11, 7833-7852

."" Majors, J. (1975a) Ph.D. dissertation, Harvard University Majors, J. (1975b) Proc. Natl. Acad. Sci. U. S. A. 72,4394-4398 Malan, T. P., and McClure, W. R. (1984) Cell 39,173-180 Malan, T. P., Kolh, A., Buc, H., and McClure, W. R. (1984) J . Mol. Biol. 180,

Maxam, A. M., and Gilbert, W. (1980) Methods Enzyrnol. 65,499-560 McKay, D. B., Weber, I. T., and Steitz, T. A. (1982) J . Biol. Chern. 257,9518-

Meiklejohn, A. L., and Gralla, J. D. (1985) Cell 4 3 , 769-776 Musso, R. E., DiLauro, R., Adhya, S., and decrombrugghe, B. (1977) Cell 12,

IJSZ

881-909

9524

QA7-XKA Peterson, M. L., and Reznikoff W. S. (1985) J . Mol. Biol. 185,535-543 Schmitz, A,, and Galas D. J. (i979) Nucleic Acids Res. 6 111-137 Shanblatt, S. H., and Revzin, A. (1983) Proc. Natl. A&. Sci. U. S. A . 80,

"(. ""

1 R4A-1 !iQQ

Siebenlist, U., and Gilbert, W. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, 122-

Siebenlist, U., Simpson, R. B., and Gilbert, W. (1980) Cell 20,269-281 Spassky, A., Busby, S., and Buc, H. (1984) EMBO J . 3,43-50 Tanipchi, T., O'Neill, M., and decromhrugghe, B. (1979) Proc. Natl. Acad.

Weber, I. T., and Steitz, T. A. (1984) Proc. Natl. Acad. Sci. U. S. A . 81,3973-

""_ ""- 126

Scr. U. S. A. 76,5090-5094

3977