Transcription Activation at Class II CAP-Dependent Promoters: Two Interactions between CAP and RNA Polymerase

At Class II catabolite activator protein (CAP)-dependent promoters, CAP activates transcription from a DNA site overlapping the DNA site for RNA polymerase. We show that transcription activation at Class II CAP-dependent promoters requires not only the previously characterized interaction between an activating region of CAP and the RNA polymerase α subunit C-terminal domain, but also an interaction between a second, promoter-class-specific activating region of CAP and the RNA polymerase α subunit N-terminal domain. We further show that the two interactions affect different steps in transcription initiation. Transcription activation at Class II CAP-dependent promoters provides a paradigm for understanding how an activator can make multiple interactions with the transcription machinery, each interaction being responsible for a specific mechanistic consequence.

In this report, we show that transcription activation at Class II CAP-dependent promoters requires a second, Class II–specific activating region, define the critical residues of the second activating region, define the subunit orientation of the second activating region, present evidence that the second activating region functions through interaction with RNAP, and define the target in RNAP. In addition, we show that the previously characterized and second activating regions affect different steps in transcription.

At Class I CAP-dependent promoters, interaction between the activating region and αCTD (and concomitant recruitment of RNAP to promoter DNA) appears to be the entire basis of transcription activation. Thus, at Class I CAP-dependent promoters, removal of αCTD eliminates transcription activation ( Igarashi and Ishihama, 1991 ).

Simple CAP-dependent promoters (those that require only CAP for transcription activation) can be grouped into two classes ( Ebright, 1993 ). In Class I CAP-dependent promoters, the DNA site for CAP is located upstream of the DNA site for RNA polymerase (RNAP). The best characterized Class I CAP-dependent promoters are the lac promoter and the artificial promoter CC(−61.5), each of which has a DNA site for CAP centered at position −61.5. In contrast, in Class II CAP-dependent promoters, the DNA site for CAP overlaps the DNA site for RNAP. The best characterized Class II CAP-dependent promoters are the galP1 promoter and the artificial promoter CC(−41.5), each of which has a DNA site for CAP centered at position −41.5.

Escherichia coli catabolite activator protein (CAP; also referred to as the cAMP receptor protein, CRP) is a structurally characterized transcription activator protein ( Kolb et al., 1993a ). CAP functions by binding, in the presence of the allosteric effector cAMP, to specific DNA sites located near or in CAP-dependent promoters.

Results

CAP Has a Class II–Specific Activating Region (“AR2”)

The objective of the first set of experiments was to determine whether CAP contains a second activating region specifically required to activate transcription at Class II CAP-dependent promoters, and, if so, to define its location within CAP. Our approach was to perform random mutagenesis of the gene encoding CAP, followed by a screen, to isolate mutants specifically defective in transcription activation at Class II CAP-dependent promoters, i.e., defective in transcription activation at Class II CAP-dependent promoters, but not defective in transcription activation at Class I CAP-dependent promoters, and not defective in DNA binding. We designate such mutants “crppc,CAP,II,” where “crp” denotes the gene encoding CAP, “pc” denotes positive-control-defective, and “CAP, II” denotes Class II CAP-dependent promoters.

For our screen, we used strain XE82/CC(−41.5). This strain contained a deletion of crp and contained two reporter constructs. The first reporter construct had lacZ fused to the Class II CAP-dependent promoter CC(−41.5) and served as an indicator of Class II CAP- dependent transcription; the second reporter fusion had lacZ fused to the artificial CAP-repressed promoter lac-PUV5-OCAP (Irwin and Ptashne, 1987; Zhou et al., 1993a) and served as an indicator of repression, and thus of DNA binding.

We performed random mutagenesis of the crp gene of plasmid pYZCRP using error-prone PCR, introduced the mutagenized plasmid DNA into strain XE82/pRW 2CC(−41.5), and identified transformants defective in Class II CAP-dependent transcription but not defective in DNA binding as red colonies on lactose/tetrazolium agar.

We performed 60 independent mutagenesis reactions, screened 15,000 transformants, and isolated 21 independent mutants defective in Class II CAP-dependent transcription but not defective in DNA binding. For each mutant, we prepared plasmid DNA, introduced the plasmid DNA into strains XE65.2/pRWCC(−41.5), XE65.2/pRWCC(−61.5), and XE82, and performed in vivo assays of Class II CAP-dependent transcription, Class I CAP-dependent transcription, and DNA binding. Based on the results, the mutants could be divided into two groups. The first group, consisting of eight mutants, was defective in both Class II and Class I CAP-dependent transcription. The second group, consisting of 11 mutants, was defective solely in Class II CAP-dependent transcription.

summarizes the sequences and phenotypes of the first group of mutants, i.e., those defective in both Class II and Class I CAP-dependent transcription (crppc,CAP,I). These mutants map to the previously defined activating region (Zhou et al., 1993a, 1994a; Niu et al., 1994). Thus, substitutions were obtained at amino acids 158, 159, 160, 162, and 164. Each of these mutants is defective in Class II CAP-dependent transcription, defective in Class I CAP-dependent transcription, but not defective in DNA binding.

Table 1

Amino Acid SubstitutionCodon SubstitutionNumber of IsolatesActivation, II (%)aActivation, I (%)aRepression (%)a158 Thr→AlaACT→GCT2b152.0120158 Thr→lleACT→ATT1135.7120159 His→ArgCAC→CGC18.84.6120160 Pro→ThrCCG→ACG1c2320140162 Gly→AspGGT→GAT15.31.9110162 Gly→ValGGT→GTT15.23.4160164 Gln→ArgCAA→CGA16.710110Open in a separate window

summarizes the sequences and phenotypes of the second, new group of mutants, i.e., those defective solely in Class II CAP-dependent transcription (crppc,CAP,II). These mutants map to the N-terminal cAMP binding domain of CAP, far from the previously defined activating region. Thus, substitutions were obtained at amino acids 19, 21, and 101. Each of these mutants is defective in Class II CAP-dependent transcription, but not defective in Class I CAP-dependent transcription and DNA binding. Each of these mutants is defective in Class II CAP-dependent transcription at all Class II promoters tested (CC[−41.5], melR, and galP1 [ and data not shown; V. Rhodius and S. Busby, personal communication]).

Table 2

Amino Acid SubstitutionCodon SubstitutionNumber of IsolatesActivation, II (%)aActivation, I (%)aRepression (%)a19 His→LeuCAC→CTC26.617012019 His→TyrCAC→TAC310.317012021 His→LeuCAT→CTT17.2150180101 Lys→GluAAA→GAA5b4.9120120Open in a separate window

To confirm the results of the in vivo assays, we purified three of the mutant CAP derivatives and assessed Class II CAP-dependent transcription, Class I CAP-dependent transcription, DNA binding, and DNA bending in vitro.

To assess Class II and Class I CAP-dependent transcription, we performed abortive initiation in vitro transcription experiments with the CC(−41.5) promoter and the CC(−61.5) promoter, respectively. The results in show that the mutant CAP derivatives are defective in Class II CAP-dependent transcription, with defects from 2- to 7-fold, but are not defective in Class I CAP-dependent transcription.

An external file that holds a picture, illustration, etc.
Object name is nihms686864f1.jpgOpen in a separate window

To assess DNA binding, we performed equilibrium DNA binding experiments with a DNA fragment containing the consensus DNA site for CAP. The results in show that wild-type CAP and the mutant CAP derivatives exhibit indistinguishable DNA binding affinities.

To assess DNA bending, we performed electrophoretic mobility shift experiments with four circularly permuted DNA fragments containing the consensus DNA site for CAP. The results in show that wild-type CAP and the mutant CAP derivatives result in indistinguishable DNA bend angles.

Taken together, our results establish that amino acids 19, 21, and 101 of CAP are critical for Class II CAP-dependent transcription, but are not critical for Class I CAP-dependent transcription, DNA binding, and DNA bending. The simplest interpretation of the results is that Class II CAP-dependent transcription requires a second interaction between CAP and RNAP, in addition to the interaction by the previously defined activating region, and that amino acids 19, 21, and 101 constitute the determinant of CAP for this second interaction.

In the primary structure of CAP, amino acids 19 and 21 are distant from amino acid 101. However, in the three-dimensional structure of CAP, amino acids 19, 21, and 101 are adjacent to each other ( ). Amino acids 19, 21, and 101 are distant from the amino acids of CAP involved in DNA binding and DNA bending and are located in a prominently accessible and protruding portion of CAP ( ).

West et al. (1993) have shown that substitution of amino acid 96 of CAP improves transcription activation at Class II CAP-dependent promoters. Amino acids 19, 21, and 101 are adjacent to amino 96, and with this amino acid, form a surface with dimensions of 20 × 8 Å ( ). We designate these four amino acids “activating region 2” (AR2), and we redesignate the previously defined activating region “activating region 1” (AR1). We propose that both AR1 and AR2 interact with RNAP in transcription activation at Class II CAP-dependent promoters.

AR2 Requires Positive Charge

To identify critical side-chain determinants within AR2, we performed alanine scanning (Cunningham and Wells, 1989). We substituted each surface amino acid from 16–23 and 96–108 of CAP, one-by-one, by alanine. We then assessed Class II CAP-dependent transcription, Class I CAP-dependent transcription, and DNA binding in vivo.

The results for Class II CAP-dependent transcription are presented in . Alanine substitution of amino acids 19, 21, and 101 resulted in ≥5-fold defects in Class II CAP-dependent transcription. We conclude that, for each of these amino acids, side-chain atoms beyond Cβ make favorable interactions in Class II CAP-dependent transcription. Alanine substitution of amino acid 96, an amino acid at which non-alanine substitutions improve Class II CAP-dependent transcription (West et al., 1993), resulted in a 2- to 3-fold increase in Class II CAP-dependent transcription. We conclude that, for this amino acid, side-chain atoms beyond Cβ make unfavorable interactions in Class II CAP-dependent transcription. Alanine substitution of no other amino acid in the region tested had a significant, reproducible effect. We conclude that for no amino acid other than 19, 21, 101, and 96 do side-chain atoms beyond Cβ make significant interactions in Class II CAP-dependent transcription.

An external file that holds a picture, illustration, etc.
Object name is nihms686864f2.jpgOpen in a separate window

The results for Class I CAP-dependent transcription and DNA binding are presented in . Alanine substitution of no amino acid in the region tested had a significant, reproducible effect on Class I CAP-dependent transcription or DNA binding, confirming that this region is not involved in these processes.

It is striking that all three amino acids that make favorable interactions are positively charged (His-19, His-21, and Lys-101), and that the amino acid that makes an unfavorable interaction is negatively charged (Glu-96). Based on the correlation of positive charge with function, we suggest that the most important structural feature of AR2 is positive charge and that AR2 is likely to interact with a target having a complementary charge, i.e., a negative charge.

AR2 Functions in the Downstream Subunit of CAP

To determine which subunit of the CAP dimer functionally presents AR2, we performed “oriented-heterodimer” analysis ( ; Zhou et al., 1993b, 1994b). We constructed CAP heterodimers having one subunit with a functional AR2 and one subunit with a nonfunctional AR2, we oriented the heterodimers on Class II promoter DNA using appropriate DNA-binding specificity mutants in protein and DNA, and we assessed the abilities of the oriented heterodimers to activate transcription.

An external file that holds a picture, illustration, etc.
Object name is nihms686864f3.jpgOpen in a separate window

We performed oriented-heterodimer analysis for each of three AR2− mutants: [Leu-19]CAP, [Tyr-19]CAP, and [Glu-101]CAP ( ). In each case, the result was the same. In the orientation in which the functional AR2 was in the downstream subunit, the heterodimer was functional in transcription activation (CX in ); in contrast, in the orientation in which the nonfunctional AR2 was in the downstream subunit, the heterodimer was nonfunctional in transcription activation (XC in ). We conclude that Class II CAP-dependent transcription requires a functional AR2 in only one subunit, i.e., the downstream subunit. Consistent with this conclusion, the super-activating effect of substitution of amino acid 96 occurs when the substitution is present in only one subunit, i.e., the downstream subunit (Williams et al., 1996).

AR2 Functions through Protein–Protein Interaction with RNAP

To test directly the hypothesis that AR2 interacts with RNAP, we measured CAP–RNAP interaction in solution.

In published work, we have shown using fluorescence-anisotropy measurements that a binary complex of CAP and a fluorochrome-labeled short DNA fragment containing the DNA site for CAP and containing no specific determinants for binding of RNAP is able to interact with RNAP to form a ternary complex (Heyduk et al., 1993). The interaction exhibits an equilibrium binding constant of ≈5 × 107 M−1 and requires AR1 (Heyduk et al., 1993; ). In subsequent work, we have shown that the interaction is strengthened 6-fold upon substitution of Lys-52 by asparagine (a substitution that results in improved transcription activation at Class II, but not Class I, CAP-dependent promoters [Bell et al., 1990; Williams et al., 1991; see Discussion]), indicating that the ternary complex analyzed in fluorescence-anisotropy measurements is a Class II ternary complex ( ).

An external file that holds a picture, illustration, etc.
Object name is nihms686864f4.jpgOpen in a separate window

In the current work, we performed analogous experiments with each of three AR2− mutants: [Leu-19]CAP, [Tyr-19]CAP, and [Glu-101]CAP ( ). The AR2− mutants exhibited 8- to 20-fold defects in interaction with RNAP. We conclude that AR2 makes direct, favorable protein–protein interaction with RNAP and that amino acids 19 and 101 contribute ≈1–2 kcal/mol each toward CAP–RNAP interaction.

AR2 Interacts with the RNAP α Subunit N-Terminal Domain

As a first step to identify the target of AR2 within RNAP, we performed site-specific protein–protein photocross-linking (Chen et al., 1994). We constructed a CAP derivative having a photoactivatible cross-linking agent site-specifically incorporated at a single, defined amino acid adjacent to AR2, i.e., amino acid 17. We then formed the ternary complex of CAP derivative, RNAP, and Class II CAP-dependent promoter CC(−41.5), UV-irradiated the ternary complex to initiate cross-linking, and determined the site at which cross-linking occurred. To facilitate identification of the site at which cross-linking occurred, we used a photoactivatible cross-linking agent that contained a radiolabel and that was attached to CAP through a disulfide linkage. This permitted, after UV-irradiation, cleavage of the cross-link and transfer of radiolabel to the site at which cross-linking occurred.

The results are presented in . CAP→RNAP cross-linking occurred primarily in the RNAP α subunit (efficiency ≈ 20%). (Smaller amounts of cross-linking occurred in the RNAP β or β′ subunit [or in both] [efficiency ≈ 4–5%], and much smaller amounts of cross-linking occurred in the RNAP σ subunit [efficiency ≈ 1%].) Control experiments established that CAP→RNAP cross-linking required UV-irradiation, promoter DNA, RNAP, and cAMP (the allosteric effector required for specific DNA binding by CAP) ( and data not shown).

An external file that holds a picture, illustration, etc.
Object name is nihms686864f5.jpgOpen in a separate window

To define the site in α at which cross-linking occurs, we performed proteolytic mapping with hydroxylamine, which cleaves α into fragments consisting of amino acids 1–208 and 209–329 ( ). The results establish that cross-linking occurs within amino acids 1–208. This region of α corresponds to the α N-terminal domain (αNTD; amino acids 8–235; Blatter et al., 1994; Busby and Ebright, 1994).

We conclude that, in the ternary complex of CAP, RNAP, and Class II CAP-dependent promoter, AR2 of CAP is in direct physical proximity to αNTD, and we propose that AR2 interacts with αNTD.

AR2 Interacts with Amino Acids 162–165 within the RNAP α Subunit N-Terminal Domain

To define further the target of AR2, we performed random mutagenesis of the gene encoding α and screened for mutants specifically defective in Class II CAP-dependent transcription, i.e., defective in Class II CAP-dependent transcription but not defective in Class I CAP-dependent transcription and CAP-independent transcription. We designate such mutants “rpoApct,CAP,II,” where “rpoA” denotes the gene encoding α, “pct” denotes positive-control-target-defective, and “CAP,II” denotes Class II CAP-dependent transcription.

Our screen tested two phenotypes: first, defect in Class II CAP-dependent transcription, and second, absence of a defect in CAP-independent transcription. To test the first phenotype, the screen scored lacZ expression from a PCC(−41.5)–lacZ fusion. To test the second phenotype, the screen scored viability. We reasoned that mutants of α specifically defective in Class II CAP-dependent transcription, like mutants lacking CAP (Sabourin and Beckwith, 1975), would be viable on rich media, but that mutants of α defective in both Class II CAP-dependent and CAP-independent transcription would be inviable (cf. Tang et al., 1994).

We performed random mutagenesis of the entire rpoA gene of plasmid pREIIα using error-prone polymerase chain reaction (PCR), introduced the mutagenized plasmid DNA into strain XE3000, and identified transformants defective in Class II CAP-dependent transcription but not defective in CAP-independent transcription as red colonies on lactose/tetrazolium agar. We performed 60 independent mutagenesis reactions, screened 100,000 transformants, and isolated three independent mutants.

All three mutants mapped to a single amino acid within αNTD and resulted in the same substitution: Glu-165→Lys ( ). Based on in vivo assays, the mutants were defective in Class II CAP-dependent transcription (assayed at CC[−41.5]), but not defective in Class I CAP-dependent transcription (assayed at CC[−61.5], lac, and rbs), and not defective in CAP-independent transcription (assayed at lacPL8-UV5) ( and data not shown).

An external file that holds a picture, illustration, etc.
Object name is nihms686864f6.jpgOpen in a separate window

To confirm the results of the in vivo assays, we reconstituted [Lys-165]α-RNAP from purified [Lys-165]α and wild-type β, β′, and σ70, and we analyzed transcription in vitro. The results are presented in . [Lys-165]α-RNAP was 5-fold defective in Class II CAP-dependent transcription, but not defective in Class I CAP-dependent transcription and CAP-independent transcription. We conclude that amino acid 165 within αNTD is essential for Class II CAP-dependent transcription, but not for Class I CAP-dependent transcription and CAP-independent transcription.

Glu-165 is a negatively charged amino acid, and the Glu-165→Lys substitution results in charge reversal, replacing this negatively charged amino acid by a positively charged amino acid. The negative charge of Glu-165 is consistent with the hypothesis that this amino acid interacts with AR2, which carries, and requires, positive charge.

Glu-165 is located in a string of four consecutive negatively charged amino acids, i.e., Glu-162, Glu-163, Asp-164, and Glu-165 ( ). For each of these amino acids, we constructed a single alanine substitution and analyzed the effect on Class II CAP-dependent transcription at CC(−41.5) ( ). In each case, alanine substitution resulted in a small, but reproducible, defect in Class II CAP-dependent transcription. Simultaneous alanine substitution of all four amino acids resulted in a large defect in Class II CAP-dependent transcription, a defect comparable with that upon charge reversal at amino acid 165. We conclude that, for each of these four consecutive negatively charged amino acids, side-chain atoms beyond Cβ are important for Class II CAP-dependent transcription, and we propose that it is the negative charge of the side-chain atoms beyond Cβ that is the critical functional determinant.

Based on the correspondence between the photocross-linking results demonstrating that AR2 of CAP is in direct physical proximity to αNTD in the ternary complex of CAP, RNAP, and Class II CAP-dependent promoter ( ), the genetic results demonstrating that amino acids 162–165 within αNTD are important for Class II CAP-dependent transcription ( ), and the charge complementarity between AR2 and amino acids 162–165 within αNTD, we conclude that transcription activation at Class II CAP-dependent promoters involves interaction between AR2 and amino acids 162–165 within αNTD.

AR1 and AR2 Affect Different Steps in Transcription Initiation

We have analyzed the effects of substitution of AR1 and AR2 on the kinetics of Class II CAP-dependent transcription (methods as in McClure, 1980; ). Substitution of AR1 resulted in an 8-fold decrease in the binding constant, KB, for interaction of RNAP with promoter DNA to form the closed complex. Substitution of AR2 resulted in a 10-fold decrease in the rate constant, kf, for isomerization of the closed complex to the open complex with single-stranded DNA in the RNAP active site. Similar results have been obtained with at least one other Class II CAP-dependent promoter (melRcon; V. Rhodius and S. Busby, personal communication). We conclude that AR1 facilitates formation of closed complex and that AR2 facilitates isomerization of closed complex to open complex.

An external file that holds a picture, illustration, etc.
Object name is nihms686864f7.jpgOpen in a separate window