Stephanie H. Shanblatt and Arnold Revzin

Transcription

1 THE JOURNAL OF BOLOGCAL CHEMSTRY by The American Society for Biochemistry and Molecular Biology, nc. Vol. 262, No. 24, ssue of August 25, pp ,1987 Printed in W.S.A. nteractions of the Catabolite Activator Protein (CAP) at the Galactose and Lactose Promoters of Escherichia coli Probed by Hydroxyl Radical Footprinting THE SECOND CAP MOLECULE WHCH BNDS AT gal AND THE ONE CAP AT lac MAY ACT TO STMULATE TRANSCRPTON N THE SAME WAY* Stephanie H. Shanblatt and Arnold Revzin From the Department of Biochemistry, Michigan State University, East Lansing, Michigan (Received for publication, March 2, 1987) The catabolite activator protein (CAP) binding sites of the Escherichia coli galactose and lactose operons were probedby hydroxyl radical footprinting. This method reveals each base that is protected by the bound protein. The patterns of protection seen for the primary CAP sites at gal and lac were virtually identical. n the presence of RNA polymerase the footprint of the second CAP molecule at gal was found to be very similar to those at the other two sites. This upstream site in gal aligns perfectly with the lac CAP site with respect to the start of P1 transcription. Replacing most of the gal second CAP site DNA with heterologous sequences did not abolish binding although it became noticeably weaker. n vitro transcription studies of this hybrid gal promoter DNA further demonstrated the reduced affinity of the second CAP. These results are consistent with molecular models proposed for specific CAP binding and suggest that the second CAP at gal may be responsible for overall stimulation of transcription at this operon. Thus, in spite of differences in stoichiometry, the mechanisms of activation by CAP at gal and lac may be quite similar. The catabolite activator protein (CAP) of Escherichia coli stimulates transcription from many promoters, notably those involved in sugar metabolism. Much evidence suggests that it does so by interacting with RNA polymerase after CAP binds at a specific site (1). The effect of CAP on transcription from the gal promoter is seen both on formation of a closed complex between polymerase and the double helical promoter region, and on the rate of isomerization to the open, transcriptionally competent form of the enzyme-dna complex (2). At the lac promoter it has been reported that the main effect of CAP is to enhance the binding of RNA polymerase to form a closed complex (3), although there are suggestions that CAP may alter the isomerization rate as well (4). However, because CAP binds at various locations relative to the polymerase-promoter site, it is not known whether a common mechanism of activation functions at different operons. The protein binds between -30 and -50 at gal, where +1 is the start of transcription (5, 6). The CAP-binding site in *This research was supported by Grant GM from the National nstitutes 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 advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: CAP, catabolite activator protein; bp, base pair. the lac operon control region spans the -50 to -70 region (7). nterestingly, at gal a second CAP molecule binds, in the presence of RNA polymerase, at the -50 to -70 location, exactly where it does so at lac (2, 8). Both the gal and lac operons have overlapping RNA polymerase-binding sites (9-12). n each case the P1 promoter is stimulated by CAP while the upstream P2 site is inhibited by it. The lac P2 promoter initiates in vitro 22 bp upstream of P, thus placing its -35 region in the middle of the CAP-binding domain. The gal P2-35 sequence and primary CAP site also overlap since gal P2 begins transcription only 5 bp upstream of P. The binding of the two CAP molecules at gal is ordered CAP binds first at -30 to -50, followed by polymerase, then the second activator protein binds cooperatively (2). When DNA upstream of -60 is removed so that the second CAP cannot bind, a CAP molecule bound at the primary site will still induce RNA polymerase to bind at P1 and initiate transcription. However, the resulting level of RNA synthesis is considerably lower than that obtained when both CAP molecules are able to bind (2, 13). That is, while the first CAP is able to inhibit P2, presumably by occluding polymerase from that binding site, and to increase polymerase binding at P, the second CAP appears to be required for stimulation of transcription from P1. One common feature of these operons is the sequence of the CAP-binding site, which is remarkably similar at all sites so far examined, the consensus being 5 -AA-TGTGA--T--- TCA-ATW (W is either A an or T) (14). The protein, a dimer of identical subunits, has been crystallized (15), and models describing how it binds to its specific site have been proposed (14, 16). These models are consistent with DNase footprint and methylation/ethylation interference data of the lac and gal CAP sites (5, 8, 17, 18). One portion, helix F, of each monomer is thought to bind to one-half of the sequence across the major groove of right-handed B-DNA. The dimeric protein covers about 20 base pairs of DNA, two turns of the helix. t is of interest to examine in detail the nature of the second CAP binding at the galactose operon. While there are sequences in this region which are homologous to the CAP consensus, it has also been reported that gal DNA upstream of -60 can be replaced by heterologous sequences without adversely affecting CAP stimulation of P1 transcription (5, 13). DNase footprinting (19), while defining a region of binding, does not provide detailed information at the base pair level. Methylation and ethylation interference experiments were, by nature, unable to yield results on the binding of the second CAP (see Ref. 18 for further discussion). We therefore turned to hydroxyl radical footprinting; in this approach, the reactions are done preformed on complexes and

2 CAP nteractions at the galand lac Promoters the radical can attack every base withlittle sequence specificity (20). Footprints were performed on the complete gal regulatory region as well as on a fragment in which the DNA upstream of -60 was replaced by heterologous sequences. The lac CAP site was also studied by this method and compared with the two gal CAP-binding sites. All three sites examined showed marked similarities inthe patternsof protection. The data suggest that CAP bound at itsprimary site a t gal serves mainly to inhibit RNA polymerase binding at P2, while the CAP bound upstream may function stimulate to transcription in a manneranalogous to thatof the lone CAP at lac. 1 FG. 2. Densitometer scans of gal lower and upper strand footprints. Fig. 1 were Lanes 2-4 and 6-8 of scanned. Panels A-C are of the lower strand and D-F are the upper strand. Panels A and D aredna only reactions, pawls B and E show CAP-DNA complexes, and panels C and F are CAPRNA polymerase-dna reactions. Regions a-e and a + show bases protected by CAPfrom attack by the hydroxyl radical. 3 4 e d C MATERALS ANDMETHODS DNA Fragments-Fragments containing the galactose promoter region were subcloned into the EcoR and Hind sites of pbr322. The -90 fragment extended from the Cfo site a t -90 (replaced by an EcoR linker) to the BstE site a t +38 (Hind linker). The -60 fragment was generated by deleting gal sequences from -90 to the Hinf site at -60; an EcoR linker was then placed a t -60. The fragments were excised from the recombinant plasmids with Hind and Dde (the site at base 4290 of pbr3221, yielding fragments with 72 bp of pbr322 DNA upstream of the gal sequences. Fragments were 5 labeled a t one or the other of these ends (-166, -136, or +38) using [y3*p]atp andpolynucleotide kinase. Lactose promoter DNA was labeled at either -123 or +67 as described elsewhere (18). Proteins-E. coli RNA polymerase and CAP protein were purified by the methods of Burgess and Jendrisak (21) and Eilen et al. (22), respectively, as described (18). Hydroxyl Radical Footprinting-Binding reactions were performed as described (18)in a total volume of 34 pl; no nonspecific DNA was added. Binding buffer contained 20 mm Tris-C1, ph 8.0, 10 mm MgC12, 100 mm KCl, 0.1 mm dithiothreitol, 0.1 mm EDTA, 200 p~ CAMP.CAP and RNA polymerase solutions were dialyzed into binding buffer in order to remove all glycerol from them prior to use. Footprinting reactions were done as described by Tullius and Dombroski (20); 2 pl of each reagent were added to 34pl of binding reactions. Reactions containing RNA polymerase were challenged with 100 pg/ml heparin immediately before footprinting. Samples d b +r uu -70. a FG. 1. Hydroxyl radical footprintof the gal CAP sites. The authentic gal fragment was 5 -end-labeled at either +38 or -166 to visualize the lower (lunes 1-4) or upper (lunes 5-8) strand. Reactions containing DNA only (lanes 2 and 6 ),DNA and CAP (lanes 3 and 3, or DNA with CAP and RNA polymerase (lanes 4 and 8) were treated to generate the hydroxyl radical. Products were separated on a sequencing gel and visualized by autoradiography. Lanes 1 and 5 are Gsequencing reactions. Regions marked a-c and a -c were protected by CAP only. n the presence of polymerase, regions d, e, d, and e were also protected. 1 b E a e C c b ( -30 a

3 11424 CAP nteractions the at were electrophoresed on 10% polyacrylamide gels containing 7 M urea in Tris-Borate-EDTA buffer. Maxam-Gilbert sequencing reactions (23) of each fragment were run in adjacent lanes. Dried gels were autoradiographed using Kodak XAR-5 film and the films scanned with a Hoefer GS300 scanning densitometer. n Vitro Transcription-Reactions were performed as previously described (2) in binding buffer and the indicated concentrations of CAMP. The 3ZP-labeled P1 and P2 transcripts were separated on a sequencing gel and quantified by liquid scintillation counting. RESULTS Hydroxyl Radical Footprinting-This method, similar in principle to DNase footprinting, employs hydroxyl radicals generated in the test tube by reaction of hydrogen peroxide with an iron(1)-edta complex. The radical is thought to react with the sugar moieties of DNA, leading to strand scission at the site of attack. Reaction conditions are such that each DNA molecule is broken only once, resulting in a ladder of products differing by only one base. Because of the small size of the reagent it can reach regions of the DNA in very close proximity to the bound protein, e.g. CAP, much better than can the larger DNase reagent. The net result, however, is qualitatively the same. Bases that interact with the bound protein protect the DNA from reaction with the hydroxyl radical, thus leaving bands of reduced intensity when A C. 1 l e dl C b a l FG. 3. Hydroxyl radical footprint of the gal hybrid fragment. This fragment has gal sequences upstream of -60 replaced with pbr322 DNA. t was labeled at -136 on the upper strand, mixed with CAP or CAP plus RNA polymerase, reacted with the footprinting reagents, and the products resolved on a sequencing gel. Densitometer scans of the autoradiomam. shown in this figure, are of DNA only (panel A), CAP-DNA-(pa&l E), and CAP-RNA polymerase-dna (panel C). Regions protected by CAP are labeled a -e, as in Fig. 2. gal and lac Promoters the products are separated on a DNA-sequencing gel. The gal CAP Sites-The primary CAP binding site at the gal operon, denoted gal-1, was examined first. A DNA restriction fragment containing gal sequences from -90 to +38 was 32P-labeled at one or the other end. This -90 fragment was mixed with CAP-CAMP to form complexes and then reacted with the hydroxyl radical. The products were separated on a sequencing gel and visualized by autoradiography. Data for the lower (lanes 1-4) and upper (lunes 5-8) strands are shown in Fig. 1. Lanes 2 and 6 show reactions which contained labeled DNA, but no added protein; bands of approximately equal intensity are seen at each base. When CAP is added (lunes 3 and 7) three regions on each strand, labeled a-c and a -c, are protected from attack. The addition of RNA polymerase to CAP-gal DNAcomplexesallows the second CAP molecule to bind to the upstream gal-2 site. Two additional regions on each strand, d, e, d and e, are protected by the binding of the second CAP. These protected regions are visualized clearly on densitometer scans of the footprint gels. Fig. 2 shows the lower (panels A-C) and upper (panels D-F) strands of the gal region. Each base appears as a peak, while regions protected by CAP from attack by the hydroxyl radical show significantly reduced peak intensity. Although not every base in the DNA only tracings (panels A and D) has quite the same amplitude, comparison of these panels with either the CAP-DNA scans (panels B and E) or those from the CAP-polymerase-DNA reactions (panels C and F) shows precisely which bases are protected by the bound protein. Thus, on each strand the three regions protected by CAP alone are centered 10 base pairs apart; the protected areas on the lower strand are spaced in between those on the upper strand. The second CAP molecule yields only two additional binding regions on each strand, although the 10-base pair spacing continues. The first CAP does not shift position in response to the binding of either polymerase or the second CAP. The c and c regions appear to be further reduced in the CAP plus polymerase scans, probably because these regions are shared by the two bound CAP d mers. Binding of the Second CAP to Heterologous Sequences-To examine the role of specific sequences on the binding of the second CAP molecule, gal DNA downstream of -60 was joined by an EcoR linker to the 70-bp Dde-EcoR fragment of pbr322 (bases ). This gal hybrid -60 fragment has three of the four regions normally protected by the second CAP replaced with pbr322 DNA. Hydroxyl radical footprinting of this fragment in the presence of CAP alone showed the same pattern of protection as seen with the authentic gal DNA, as illustrated in panel B of Fig. 3. n the presence of polymerase, the second CAP does bind to the hybrid fragment, but the binding is noticeably weaker. n both the d and e regions, only two bases are protected, and the e peaks are not as diminished as on the normal gal fragment (compare Fig. 2, panel F and Fig. 3, panel C). Region d on the lower strand (data not shown) was protected to about the same extent as on the -90 fragment. However, only moderate protection of a single base was observed in the e region. Transcription: Authentic Versus Hybrid gal Frugments- Possible effects of the weaker binding of the second CAP to the hybrid gal DNA fragment were examined by in vitro transcription. The second CAP stabilizes the CAP-polymerase-gal DNA complex, particularly at low camp concentrations (2). This effect is manifested in the amount of camp required to switch transcription from P2 to P1. The hybrid -60 fragment described above and the authentic gal DNA to which the pbr322 Dde-EcoR fragment was joined upstream of -90, were used as templates in runoff transcription exper-

4 nteractions CAP the at gal and lac Promoters camp. JJM FG. 4. camp titrations of authentic and hybrid gal fragments. The authentic (panel A) and hybrid (panel B) gal promoter DNA fragments were used as templates for in vitro runoff transcription. 32P-Labeled P1 and P2 transcripts were synthesized in the presence of the indicated camp concentrations. Data are presented as the percentage of P2 (open circles) or P1 (closedcircles) RNA synthesized. Results are the average of three experiments. iments. Reactions were done at varying camp concentrations while maintaining constant CAP, polymerase, and DNA concentrations. 32P-Labeled P1 and P2 transcripts were separated on a sequencing gel and each band quantified. The results of this camp titration using fragments containing either the authentic or the hybrid gal promoter are shown in Fig. 4. There was about a 1.5-fold difference in the camp concentration required to cross-over from P2- to P-initiated transcripts between the two templates. This would appear to reflect the weaker binding of the second CAP to the hybrid fragment, as also evidenced by the diminished footprint. The lac CAP-binding Site-A comparison of the hydroxyl radical footprint of the lac site with that of gal revealed many similarities. Fig. 5 shows the densitometer scans of the lac upper strand DNA footprints with CAP (panel B) and CAP plus polymerase (panel C). As with the primary gal CAP site three regions, spaced 10 bp apart, are protected from attack. n the presence of polymerase, however, no additional protected bases were observed. The valley at -40 is most prominent in the DNA only trace (panel A); this is a reproducible anomaly in the electrophoretic behavior of these products. The scans in panels B and C can be superimposed on panel A and are not indicative of protection at -40 by added protein. The lac lower strand also revealed three protected regions spaced in between those of the upper strand (data not shown; see Fig. 6). DSCUSSON Specific CAP Binding gal at and &-The four CAP-binding sites probed by hydroxyl radical footprinting are presented in Fig. 6, each aligned with respect to the CAP consensus sequence. The primary gal site, labeled gal-1, shows essentially the same pattern of protection as the lac site. Previous methylation and ethylation interference studies of these two sites also revealed strong similarities (18). n addition, the results obtained by hydroxyl radical footprinting and ethylation interference are virtually identical. That is, protection was observed at the same bases whose phosphate modification C FG. 5. Hydroxyl radical footprint of the lac CAP site. A lac DNA restriction fragment, 5'-end-labeled at -123 (upper strand), was subjected to footprint analysis in the presence of CAP with or without RNA polymerase. Densitometer scans of the sequencing gel are presented. Panel A is the reaction with DNA only, panel E is that of CAP-DNA complexes, and panel C that of CAP-RNA polymerase- DNA complexes. Regions protected by CAP from attack are indicated as 8 C'. Consensus 5 ' - A A - T E T G A - - T T b - A T W T T - A C B C T - - A b E T ~ T A Z **** *** *** A A G T P T G A C A T G G A A T b A A T T gal T T C A C A G T G T A C C T T A T T T A A *** **** **** *** *** ++++ T C T T E T E T A A A C G A T T C C A C T -55 A G A A C A C A T T T G C T A A G E T G A gal **** *** gal-2 (hybrid) ** ** C T T C A A E A A T T C C A T T C C A C T G A A G T T- C T T A A G G T A A G G T G A " +++ * *** *** *** *** T A A T E T E A G T T A G C T C A C T C A lac A T T A C A C T C A A T C G A E T Q A G T *** *** *** FG. 6. gal and lac CAP sites defined by hydroxyl radical footprinting. Each CAP-binding site is aligned with the consensus sequence shown at the top. gal-1 is the primary CAP site and gal-2 is the upstream binding region; gal-2 (hybrid) is the second site with heterologous DNA replacing gal sequences upstream of -60. The lac site shown is the primary CAP-binding site at that operon. Numbering of the four promoter sequences is relative to the start of P1 transcription at +l. Note that gal-1 is in reverse order compared with the others; this orientation provides the best comparison with the consensus. Bases protected by CAP from hydroxyl radical attack are marked with an *. The +s at the gal-2 sites from -54 to -48 are bases protected by the binding of the first CAP to gal-1, but are presumed to be shared when the second CAP molecule binds. Underlined bases are those proposed to hydrogen bond with CAP (see text).

5 11426 CAP nteractions at the gal and lac Promoters prevented binding of the protein. As described in detail by Weber and Steitz (16), each monomer of the CAP dimer interacts with one-half of the - 30 DNA binding site. While there are several potential phosphate-cap bonds, these authors propose that only 4 base pairs in each half-site are involved in hydrogen bonding to the protein. They are 5'-GTGA, and are underlined in Fig. 6. The specified interactions are with the G of bp 5 and 18 in the consensus numbering (corresponding to -68 and -55 of lac), the A of bp 6 and 17, the G and C of bp 7 and 16, and the A of bp 8 and 15. Thus, at lac there are 10 hydrogen bonds between CAP and DNA bases. At gal, only 6 of the 10 bases proposed to hydrogen bond to CAP are present, those at bp 5,6, 7, 8, and 17. The substitutions at bp 15, 16, and 18 of the gal-1 site may not permit additional hydrogen bonding to occur. This correlates well with the reduced affinity of CAP for the primary gal site - 50 compared with lac (24) and with the report that mutations which abolish CAP binding at gal map to bp 5 (-35) and 7 (-37) (6). The alteration at bp 7 is analogous to the lac L8 mutation (25). The pattern of protection from hydroxyl radical attack we observed is consistent with this model and a similar one proposed by Ebright et al. (14). All of the bases protected by CAP at gal-1 (between -30 and -50) lie on one face of the helix as illustrated in Fig. 7. Those bases listed above that hydrogen bond to CAP lie in between the protected regions; they are marked with Xs in Fig. 7. As CAP spans the major groove between protected areas, cy-helix F reaches in perpendicular to the plane and makes its hydrogen bonds with the - 70 exposed bases. Nature of the gal Second CAP Site-The protection pattern observed for the upstream gal CAP site, gal-2, looks very similar to that seen at the primary gal and lac sites, as displayed in Fig. 6. t thus appears from these data that the second CAP binds in the proposed fashion at this site also. Only four additional regions are protected, but they provide two more major grooves with which CAP can interact. This FG. 7. Helix representation of gal CAP-binding sites. The is depicted in Fig. 7, where all the protected bases at gal are two CAP-binding sites at the galactose operon are schematically represented on a helix of right-handed B-DNA. Each of the bases represented. Thirty-four bases involved in CAP binding were protected from hydroxyl radical attack by CAP in the presence of resistant to attack by hydroxyl radical; none are hidden from RNA polymerase at gal is indicated by a closed circle on the helix view in Fig. 7. This clearly shows that CAP binds t only one backbone. The five bases at each half-site thought to hydrogen bond side of the helix and that at gal both CAP molecules are on with CAP are indicated by Xs. The first CAP binds in the -30 to the same face. -50 region and the second between -50 and -70. When CAP binds to the gal-1 site it appears affect to about 26 base pairs. When both CAP molecules are bound, the critical, and it is most likely in the same location at the gal-2 region from -48 to -54 is presumed to be shared by the ends site. of the two dimers. This seems reasonable since it is the very Effects of Altering the Second CAP-binding Site-When gal end of each protein molecule that stretches to these limits. DNA upstream of -60 is replaced by heterologous sequences (Note that CAP and RNA polymerase appear to share con- the number of protected bases is decreased (gal-2 (hybrid) in tacts around -39 (18).) This alignment of CAP at gal-2 allows Fig. 6). Only three hydrogen bonds can presumably form in bases -68 to -65 (consensus bp 5-8) and -58 to -55 (bp 15- the -68 to -65 region. Notably, the G at -68 (bp 5 of the 18) to hydrogen bond to the protein in the proposed manner. consensus) has been replaced by an A. The importance of this Sequence constraints limit the probable number of hydrogen G in specific binding has been strongly implicated by ethylabonds at gal-2 to six (bp 5, 6, 7, 15, and 18). tion interference experiments on the primary lac and gal CAP- As is evident in Fig. 6, not only are the protection patterns binding sites (7, 18). Without specific hydrogen bonding to of the gal second site and the lac CAP site very similar, but this G, the end of the protein may not be as firmly anchored the patterns also are precisely aligned with respect to the upstream of -70. The DNA may even be less bent without start of P1 transcription. This means that at both operons a this interaction so that the protein is unable to reach the CAP molecule is located at the same linear and rotational bases beyond -70. Replacement of the A at -67 may be less position with respect to RNA polymerase. n order for CAP crucial since the amino acid proposed to hydrogen bond to it to protect across more than 20 base pairs, the DNA must (Glu-181) also interacts with the G at -66, which is not assume a bent or kinked conformation (16). The CAP-induced altered in the hybrid construct. bend in the DNA is centered at -60 in lac (26); the energy The weaker binding seen with the hybrid -60 fragment was stored in this conformation could be important in CAP acti- also manifested in the in vitro CAMP titration transcription vation of RNA polymerase. The position of this bend may be experiment (Fig. 4). At saturating CAP and polymerase con-

6 centrations, the amount of camp becomes critical in determining whether P2 or P1 transcripts are synthesized (2). When the camp concentration is too low, CAP will not remain bound at its primary site long enough for RNA polymerase to find P1 before CAP dissociates and the enzyme binds to the P2 promoter instead. At slightly higher camp concentrations, this event as well as dissociation of the CAP-polymerase-DNA closed complex are less likely to occur, in part because binding of the second CAP stabilizes the complex. When the upstream CAP site is removed (and not replaced with heterologous DNA), high camp concentrations, e.g. greater than 10 KM, are needed to effect the P2 to P1 switch (2, 13). The gal hybrid -60 fragment described in this paper exhibited intermediate behavior. t required about 1.5 times more camp to switch from P2 to P1 transcription than did the authentic gal DNA, but much less than that required by a template without a second CAP-binding site. This intermediate behavior suggests that CAP binding to the hybrid site is somewhat weaker than it is to the -90 fragment, consistent with the reduced intensities in the hybrid fragment hydroxyl radical footprint. These subtle differences between the wild type and hybrid gal promoter sequences were not detected in previous experiments (13). Any reasonable mechanism for transcriptional regulation at the gal promoter must involve both CAP-CAP and CAP- RNA polymerase interactions (2, 8, 18). Since the second CAP molecule binds in a cooperative manner, the exact DNA sequence in the -50 to -70 region may play a smaller role than does the corresponding -30 to -50 sequence to which the first CAP binds. Nevertheless, deleting gal DNA to -60 weakens binding of the second CAP; we are now replacing additional gal-2 bases implicated in hydrogen bonding to clarify the importance of sequence in the binding of the upstream CAP. Comparing transcripts from gal DNA fragments which end at -90 and at -60 indicates that binding of the first CAP molecule is sufficient to inhibit the P2 promoter but not to stimulate transcription from P1 (2). At lac, both exclusion from P2 and enhancement at P1 are achieved by a single CAP dimer bound between -50 and -70. These observations, along with our finding of precise alignment of the upstream gal CAP site with that of lac, lead to the hypothesis that at both operons CAP must occupy the -50 to -70 region in order to stimulate P1 initiation. This presumably involves similar contacts between RNA polymerase and the CAP molecule bound there. The primary CAP at gal would thus serve mainly to position polymerase at P1. No CAP molecule is needed at the lac -30 to -50 region to prevent polymerase binding at lac P2 since this site is further upstream than is gal P2. Thus, despite different CAP stoichiometries at lac and gal there may be important similarities in the mechanism whereby the activator molecule functions. Various mutant lac CAP nteractions at the gal and lac Promoters and gal promoters are now being prepared by site-directed mutagenesis; these will allow accurate measurement of transcription levels, with one or two CAP molecules present and without interference from P2. The constructs will also permit us to assess whether protein-induced DNA bending may play a role. The results of such studies should provide a rigorous test of the model proposed here. Acknowledgments-We thank Diane Cryderman for expert technical assistance and Jon Kaguni for use of the densitometer. REFERENCES 1. decrombrugghe, B., Busby, S., and Buc, H. (1984) Science 224, Shanblatt, S. H., and Revzin, A. (1986) Biochemistry 25, Malan, T. P., Kolb, A., Buc, H., and McClure, W. R. (1984) J. Mot. Biol. 180, Lorimer, D. D., and Revzin, A. (1986) Nucleic Acids Res. 14, Taniguchi, T., O'Neill, M., and decrombrugghe, B. (1979) Proc. Nutl. Acud. Sci. U. S. A. 76, Busby, S., Aiba, H., and decrombrugghe, B. (1982) J. Mol. Biol. 164, Majors, J. (1975) Nature 256, Shanblatt, S. H., and Revzin, A. (1983) Proc. Natl. Acud. Sci. U. S. A. 80, Musso, R. E., DiLauro, R., Adhya, S., and decrombrugghe, B. (1977) Cell 12, Malan, T. P., and McClure, W. R. (1984) Cell 39, Spassky, A., Busby, S., and Buc, H. (1984) EMBO J. 3, Peterson, M. L., and Reznikoff, W. S. (1985) J. Mol. Biol. 185, "- "- 13. Taniguchi, T., and decrombrugghe, B. (1983) Nucleic Acids Res. 11, Ebright, R. H., Cossart, P., Gicquel-Sanzey, B., and Beckwith, J. (1984) Proc. Natl. Acud. Sci. U. S. A. 81, McKay, D. B., Weber,. T., and Steitz, T. A. (1982) J. Biol. Chem. 257, Weber,. T., and Steitz, T. A. (1984) Proc. Natl. Acud. Sci. U. S. A. 81, Majors, J. (1975) Ph.D. dissertation, Harvard University 18. Shanblatt, S. H., and Revzin, A. (1986) J. Biol. Chem. 261, Schmitz, A., and Galas, D. J. (1979) Nucleic Acids Res. 6, Tullius, T. D., and Dombroski, B. A. (1986) Proc. Natl. Acud. Sci. U. S. A. 83, Burgess, R. R., and Jendrisak, J. J. (1975) Biochemistry 14, Eilen, E., Pampeno, C., and Krakow, J. S. (1978) Biochemistry 17, Maxam, A. M., and Gilbert, W. (1980) Methods Enzymol. 65, Kolb, A., Spassky, A., Chapon, C., Blazy, B., and Buc, H. (1983) Nucleic Acids Res. 11, ppen, K., Miller, J. H., Scaife, J., and Beckwith, J. (1968) Nature 217, Liu-Johnson, H.-N., Gartenberg, M. R., and Crothers, D. M. (1986) Cell 47,