DNA Binding Regions of Q Proteins of Phages λ and φ80

Abstract

Bacteriophage λ gene Q protein and the related proteins of other lambdoid phages are transcription antiterminators that interact both with DNA in the late gene promoter segment and with RNA polymerase subunits. Using hybrids between Q of λ and the related Q of phage 80, we characterized elements of both Q and DNA that contribute to the DNA binding function. In particular, we found a C-terminal segment of the protein that is responsible for binding specificity and an ∼15 residue segment on a predicted alpha helix within this segment at which alanine substitutions decrease DNA binding. We identified a six-nucleotide segment located between the −35 and −10 promoter elements that confers binding specificity and is the site of point mutants that impair binding, and we isolated suppressors in λ Q that restore binding function by increasing the overall binding affinity. We also identified putative zinc finger structures in both proteins.

The Q proteins of phage λ and its relatives are transcription antiterminators that interact with DNA, with the RNA polymerase (RNAP) σ70 initiation subunit, and with RNAP core subunits (11, 13, 14). Their role in phage growth is to modify transcription initiated at the promoter of a single long transcription unit for phage late genes (5) and, particularly, to extend transcription beyond a terminator immediately downstream of this promoter. Each Q protein acts at a promoter-associated engagement site (named qut), with a species specificity based at least in part on specific DNA binding, and becomes a subunit of the elongating complex (20).

The qut site consists of a DNA binding element, located between the −35 and −10 promoter elements (Fig. 1B), and a segment downstream of +1 containing a promoter −10 element-like sequence that serves to capture elongating RNAP 16 to 25 nucleotides from the RNA start site in a σ70-induced paused state that is primed for Q recognition (12, 21). During this pause, Q binds DNA just upstream of the elongating enzyme (Fig. 1A); binding is sequence specific for various members of the Q family (19, 21). Phage λQ (at least) simultaneously contacts region 4 of σ70, displacing it forward to a promoter −35 element-like sequence located between the −2 and −7 positions of the promoter (11) (Fig. 1). Exchange of segments between qut sites of λQ and phage 82Q shows that either pause-inducing sequence is functional in combination with either Q-binding element, which in turn confers Q specificity on the qut (3); we confirm this location of the specificity element for 80Q below. Although direct evidence is missing, essential contacts with the core enzyme likely also are made as Q binds, or at least during the early stages of RNAP emergence from the pause. Sometime during subsequent elongation σ70 is likely released, leaving Q as a subunit of the elongating complex; in vitro analysis shows that Q of phage 82 becomes a firmly affixed subunit of the elongating core downstream of the promoter and that σ70 is missing at this stage (20).

FIG. 1.

Elements of phage λ and phage 80 Q function. (A) Diagram of the paused transcription complex before and after binding Q, showing the disposition of σ70 subunits, and the displacement forward of σ70 region 4 by Q (11, 12). (B) Sequence alignment of phage λ and phage 80 Q proteins (CLUSTAL, MegAlign, and DNASTAR). Predicted alpha-helical regions (PredictProtein; http://www.embl-heidelberg.de/predictprotein/predictprotein.html) are indicated by black underlines or overlines, and the determined region of specificity is shown as a red line between the sequences. Sites of cysteines of the putative zinc finger (black dots) and of strong (red triangles) and weak (open triangles) suppressors of the qut binding region mutations are shown. (C) The promoter and initial transcribed regions of qutλ and qut80, including the DNA binding element. The −35 and −10 promoter elements are shown as lines between the sequences. Overlines designate binding sites assigned for the phage 82 Q protein (9), which plausibly coincide with those of λQ and 80Q (21). Dots are sites of ethylnitrosourea interference of λQ binding (1), red bases are sites of a double-base-pair substitution that impairs Q function and DNA binding (21), and blue bases mark the main region of sequence difference between qutλ and qut80. The pause-inducing −10 promoter-like sequence is underlined (12). The initial transcribed base (designated +1) is known only for λ.

This complex pathway of Q engagement with RNAP implies a variety of sequential interactions among nucleic acids and proteins. There is little information about the structural basis of these interactions or about the structure of Q at all. The family of Q proteins, defined through their activity and positions in the genome, contains both related members and those apparently unrelated in primary sequence (16). A genomic analysis suggests an underlying dimeric structure of the λQ protomer (BLAST analysis of λQ; National Center for Biotechnology Information [NCBI], http://www.ncbi.nlm.nih.gov/), and the DNA-bound form of λQ is believed to be a dimer of the promoter (1). There is apparent homology of λQ and the zinc finger region of the chaperone DnaJ (http://www.ncbi.nlm.nih.gov/).

We have made mutants of both Q protein and its site of action in the DNA in order to understand the interacting elements of each. To guide the analysis, we used two Q proteins that are moderately but distinctly related, those of phages λ (22.5 kDa) and 80 (29.5 kDa); they are 47% identical over the span of λQ, although 80Q has a nonhomologous insertion of ca. 80 residues in the middle (Fig. 1B). Each acts specifically on its own qut (Q utilization) site according to activity in a reporter system (see below). We identify specificity elements of the DNA binding element for each Q. We also show that the C-terminal region of each protein and, in particular, the segment underlined in red (Fig. 1B) is responsible for specificity. Structure prediction (PredictProtein; http://www.embl-heidelberg.de/predictprotein/predictprotein.html; Fig. 1B,black overline) of this region proposes nearly identical alpha-helical and loop regions, despite some substantial differences in amino acid sequence. We identify in the λQ and 80Q proteins essential cysteines that may contribute to a zinc finger-like element.

MATERIALS AND METHODS

Strains, plasmids, and mutagenesis. The sequence of phage 80 gene Q and surrounding genes was obtained in this laboratory by L. Matthews and D. Kadosh (unpublished data). All plasmids were made with standard molecular biology techniques. For reporter construction, BamHI/SmaI fragments containing either qutλ or qut80 combined with the tandem terminators t₈₂t_o (6) were cloned into pRS528 (15). The qutλ point mutants and λ/80 qut fusions were made by PCR and cloned into p′RS528 so as to replace the promoter segment; p′RS528 is pRS528 [λ+49 (wt)t₈₂t_o] with the EcoRI site destroyed. The λ+49 qut fragment contains λ sequence from the HindIII site upstream of the late gene promoter pR′ to nucleotide 49 of the late transcript, followed by an introduced EcoRI site (17). The qut-lacZ fusions in pRS528 were transferred to λRS45 (15) and then in single copy to the chromosome of E. coli SG20250 as described previously (15); E. coli strain SG20250 [F⁻ araΔ139 Δ(lacIPOZY)U169 strA thi] was from S. Gottesman (National Cancer Institute, National Institutes of Health, Bethesda, Md.). The parental reporter strains are designated JWR1002 (qutλ) and JWR1023 (qut80). In the absence of a Q source, these reporters gave background levels of ca. 20 to 30 β-galactosidase units, and white color on MacConkey lactose plates after 16 h of incubation at 37°C. To generate templates for in vitro Q activity assay, qut fragments were subcloned into p′XY306, which carries λ+49 (wt)t₈₂t_o in the pXY306 backbone (17).

λQ expression.

pJG100 was constructed to express λQ upon IPTG (isopropyl-β-d-thiogalactopyranoside) induction for in vivo activity measurements. In brief, pJG100 is a low-copy plasmid conferring spectinomycin resistance and containing the phage T7 promoter A1 with two lac operators between −35 and +20 (from pUHE21-2; D. H. Bujard, unpublished data) and the lacI^q gene from pGEX-2T (Pharmacia Biotech) (4); a 760-bp fragment containing λQ was cloned under control of the T7 A1 promoter. pJG102 is an equivalent construct expressing 80Q. Induction by IPTG of pJG100 transformed into JWR1002 gave 2,000 β-galactosidase units and a deep red color on MacConkey lactose plates; pJG102 transformed into JWR1023 gave 500 β-galactosidase units and a pink color.

To overexpress λQ for purification, the gene was subcloned into p′QE-30, which was constructed from pQE-30 (Qiagen) to contain the lacI^q gene from pJG100 and to express native λQ (lacking the histidine tag of the original pQE-30 vector).

β-Galactosidase assay of Q function in vivo.

β-Galactosidase was assayed in permeabilized cells as described previously (10), after induction of Q by 1 mM IPTG for 1 h during exponential growth. Background activity without Q induction was negligible (∼2% of induced levels). Activity determinations generally represent an average of three to four independent experiments.

Construction of λ/80 fusion Q proteins and λQ alanine scanning mutants.

All λ and 80 fusion Q proteins were produced by a multistep PCR protocol. Briefly, a Q N-terminal or C-terminal PCR primer was used, along with an internal PCR primer, to make two overlapping Q fragments, which were then purified, combined in equimolar amounts, and used as a template in a further PCR in which appropriate C-terminal and N-terminal primers amplified the desired product; this was cloned into the pJG100 vector backbone, and its structure verified by sequencing.

Most of the 33 λQ alanine scan mutants were made by the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, Calif.) in the p′QE-30 vector; some were produced by the multistep PCR method described above. Point mutants were made by either procedure.

PCR mutagenesis of λQ protein and screening.

PCR mutagenesis was done separately on the N-terminal (bp 1 to 256) and C-terminal (bp 256 to 621) fragments by standard PCR. Twelve independent PCRs were pooled, ligated into the pJG100 backbone, and transformed into ElectroMAX DH10B cells (10¹⁰ transformants/μg of pUC19 DNA) to yield >10⁵ individual colonies. Plasmid DNA representing the λQ mutant library was extracted from the pooled colonies. Then, 10 ng library DNA was electroporated into appropriate reporter cells to produce 10⁵ transformants for screening λQ mutants. Defective λQ mutants were identified as white or pink colonies with a large white halo on MacConkey Lac plates compared to the red-color phenotype of wild-type λQ (expressed by pJG100) in wild-type qutλ reporter cells (JWR1002). Candidate clones were selected by colony color on the MacConkey lactose indicator plates and confirmed by streaking-out and reincubation. Q mutants were further characterized by β-galactosidase assay on relevant reporter cells and by Western blot analysis to verify expression of the Q polypeptide. In the screen for defectives, 27 candidates from the 50,000 λQ N-terminal mutagenized pool and 61 clones from the 50,000 λQ C-terminal mutagenized pool were identified and sequenced. λQ suppressors of qutλ mutants were identified to restore red color to the two qutλ mutant (−25G&ρɛγ;C and −22T&ρɛγ;G) reporter cells; seven potential λQ suppressors were isolated from 500,000 λQ C-terminal mutant colonies, and DNA sequencing identified five different suppressor mutations.

Q overexpression and purification.

Q proteins were overexpressed to 10 to 20% of total cell protein upon IPTG induction of the p′QE-30 vector in E. coli BL21 cells and purified as described previously (18). Peak fractions from phosphocellulose chromatography were identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, pooled, dialyzed into λQ storage buffer (10 mM K phosphate [pH 6.5], 1 mM dithiothreitol [DTT], 1 mM EDTA, 1 M KCl, 50% glycerol), and stored at −20°C.

Electrophoretic mobility shift assay of λQ-DNA binding.

A 115-bp 5′-end-labeled double-stranded DNA probe containing λ pR′ −35 to +35 was made by PCR with a γ³²-P-labeled primer and then gel purified. λQ protein in 1 μl of λQ buffer (10 mM Tris-HCl [pH 7.5], 1 mM DTT, 0.1 mM EDTA, 200 mM KCl, 50% glycerol, and 1 mg of bovine serum albumin/ml) was added to 0.01 nM purified probe (5,000 cpm) and a 100-fold excess of sonicated calf thymus DNA (competitor DNA) in a 25-μl binding reaction; the final buffer was 20 mM Tris-HCl (pH 8.0), 25 mM KCl, 12% glycerol, 1 mM DTT, 0.1 mM EDTA, and 40 μg of BSA/ml. Binding reactions were incubated 10 min at room temperature and stored on ice before loading onto a prechilled and prerun 5% polyacrylamide gel in 0.5× TAE (20 mM Tris-acetate [pH 8.5], 1 mM sodium EDTA) running buffer. Samples were electrophoresed at 4 to 8°C in a Bio-Rad Protean II xi apparatus (Bio-Rad, Hercules, Calif.) with a circulating ice-H₂O bath. Gels were dried and scanned with a PhosphorImager, and radioactive bands were quantified with ImageQuant software. The K_d is reported as the concentration of λQ protein giving 50% shift.

In vitro transcription.

RNAP and NusA protein were purified as described previously (21). Transcripts were made by preincubating 1 nM DNA template (280 bp, yielding a 113-nucleotide terminated transcript and a 204-nucleotide runoff transcript) and 20 nM RNAP (and 150 nM NusA for experiments with λQ protein) for 5 min at 37°C in 20 mM Tris-HCl (pH 8.0); 0.1 mM EDTA; 1.0 mM DTT; 50 mM KCl; 10% glycerol; 200 μM ATP, CTP, and GTP; 50 μM UTP; and 0.5 μCi of [α³²-P]UTP/μl in a 25-μl reaction volume. RNA synthesis was initiated by the addition of 2.5 μl of 40 mM MgCl₂ and 100 μg of rifampin/ml. In reactions where it was present, λQ protein in 2.5 μl of Q buffer was added 30 s before the addition of MgCl₂ and rifampin. Reactions were stopped with 125 μl of stop buffer (12 mM EDTA, 600 mM Tris-HCl [pH 8.0], 120 μg of tRNA/ml) and extracted with 150 μl of phenol-chloroform-isoamyl alcohol (25:24:1). RNA was precipitated with 2.5 volumes of 100% ethanol and resuspended in 4 μl of 80% formamide, 1× Tris-buffered saline, and 0.05% bromophenol blue. RNA products were separated on a 12%, 8 M urea-polyacrylamide gel and scanned with a PhosphorImager. Percent terminator readthrough was calculated after quantification of the terminated and readthrough transcripts with ImageQuant software, considering the U content of the transcripts. The K_m is reported as the concentration of λQ protein giving 50% of the maximum readthrough activity.

RESULTS

Mutational analysis of the Q binding regions. Previous work showed that λQ function and DNA binding are impaired by the double point mutant −13A −15A of qutλ (1, 21) and by deletions of DNA farther upstream, although not by a substitution upstream of −26 that (necessarily) retains a functional −35 element (see Fig. 1C). The DNA binding function of qutλ in vitro is not impaired by substitution of sequences located between positions −12 and −1 (1). These results suggest that the essential λQ DNA binding sequences are located approximately between positions −13 and −26, although sequences farther downstream (in addition to the pause-inducing sequences in the early transcribed region) still could be required for Q function. DNA footprinting experiments are consistent with this assignment; thus, Q proteins of phages λ, 82, and 21 protect the cognate qut DNA approximately between positions −30 and −10 against cleavage by MPE [methidium propyl EDTA · Fe(III)] (21), and the binding of λQ is inhibited by the ethylnitrosourea reaction of phosphates in the vicinity of −12 (template), −15 (nontemplate), −22 (template), and −25 (nontemplate) (1).

The nontranscribed regions of qut80 and qutλ show the most extensive difference in the segment −22 to −28 (Fig. 1C), suggesting that this span may encompass contacts that are essential for binding specificity and that provide important binding energy. To investigate its function, we changed each base in the qutλ segment from −22 to −26 and assayed for Q function in vivo by using a reporter assay (7). Only three changes reduced Q-dependent reporter activity as much as twofold: −25C (25%), −22A (52%), and −22G (36%) (Table 1). As expected, if these mutations affect antitermination but not promoter function, background activity without induction of Q was not changed from that of the wild type (data not shown). We assayed mutant function by in vitro transcription and measured mutant effects on DNA binding by electrophoretic mobility shift assay, including also the previously characterized qutλ −15A −13A double mutant as a control. Representative experiments are illustrated in Fig. 2, and the results are summarized in Table 1. All of the mutants are about twofold defective in antitermination in vitro relative to the wild type and require an ∼10-fold-higher concentration of Q for maximum antitermination activity. Furthermore, all are defective in DNA binding; for the −22 mutants, essentially no binding occurred at 50 times the concentration required for maximal shift of wild-type DNA.

TABLE 1.

Q Activity of mutant qutλ regions^a

qutλ DNA	Q activity
	In vivo (% WT)	In vitro RT		DNA binding (Kd [nM])
		Maximum RT (% WT)	Km (nM)
Wild type	100	100	4	1
−13/−15A	15	41	30	10
−25C	25	48	35	15
−22G	36	45	40	>100
−22A	52	48	30	>100
−22A/−25C (λ/80/λ−3)	2	4		5
−22T/−25G in qut80 (80/λ/80−3)	2	3		1.5

^aRT, terminator readthrough; WT, wild type.

FIG. 2.

Defects of qut site mutations in λQ function and DNA binding. (A) Gel analysis of λQ antitermination activity on wild-type and two mutant qut DNAs: −13A/−15A (21) and −25C (the present study). λQ was present at 2, 30, 50, 100, and 200 nM, and RNA synthesis was carried out for 2.5 min. The last lane shows the pattern of synthesis in the absence of λQ, which is identical for all three DNAs. RT, readthrough RNA; t82, RNA stopped at the t82 intrinsic terminator in the template DNA. The pause, Q-dependent pause, and abortive products were as described previously (2). (B) Graph of the data in panel A. (C) Bandshift analysis of λQ binding to wild-type and mutant DNAs. λQ was present at 0, 1, 2, 5, 10, and 20 nM, except for the −13 −15 DNA, for which λQ was present at 1, 2, 5, 10, 20, and 50 nM. (D) Graph of some of the data in panel C.

We conclude that the segment from positions −22 to −26, as well as the bases at positions −13 and −15, are part of the DNA binding site and that defects in binding impair Q function. However, the relationship is not simple; saturating concentrations of Q do not overcome the binding deficiency, suggesting that the wild-type binding condition per se—e.g., the residence time of Q on DNA or the configuration of bound Q—is required for optimal function. The effect of a tighter binding mutant of Q (see below) is consistent with this view.

To determine whether the segment of greatest difference between qutλ and qut80 (positions −22 to −28) confers specificity for each Q, we made a set of exchanges in this segment between the two sites and assayed the hybrid sites for Q function in vivo and for DNA binding activity in vitro. Since we are unable to detect activity of purified 80Q in vitro, either in antitermination or DNA binding, we assayed only DNA binding by λQ. Figure 3 summarizes the results of these exchanges. First note that each qut site shows the correct specificity for each Q (Fig. 3, lines 1 and 5). Second, replacement of the λ segment from −22 to −28 by that from 80 switches the specificity to 80 (line 2); thus, this segment confers specificity. However, the presence of this segment is not sufficient for activity because the converse replacement (line 6) shows no activity with λQ, even though it binds λQ approximately as well as do single point mutants of qutλ that impair Q function only two- to threefold. Note the lack of any simple relation of antitermination activity to Q binding activity in vitro; thus, qut80 is inactive with λQ in vivo, but it binds λQ as well as do point mutants of qutλ that still have substantial activity. Furthermore, replacement of qut80 bases with the critical −22T and −25G of qutλ increases the binding of λQ to nearly equal that of qutλ, although this substitution has no activity with λQ in vivo. This result confirms the role of these nucleotides, and this segment in λQ binding and also the insufficiency of tight binding for Q activity.

FIG. 3.

Function of qutλ and qut80 hybrids recombinant in the segment from −28 to −22. λ sequences are shown in black or gray, and 80 sequences are indicated in red or pink. The DNAs were analyzed for activity in vivo as described in the text and for DNA binding in vitro by bandshift analysis.

Hybrids of λQ and 80Q proteins identify a specificity region.

The two Q proteins are sufficiently similar in primary sequence that reciprocal exchanges along the linear sequence should provide structurally congruent hybrid proteins that can be assayed meaningfully. We made such exchanges to investigate the portion of each protein required for specificity. The central region, where 80Q has an insertion of ca. 80 amino acids relative to λQ, required consideration. Because both reciprocal exchanges of outer segments were active in conjunction with the λ central sequence (λ amino acids 113 to 120), whereas the 80 central sequence (80 amino acids from positions 111 to 183) yielded active hybrid Q protein only in conjunction with the 80 rightward segment, we used the λ central sequence with all hybrids.

Exchange of flanking arms (Fig. 4, lines 3 and 4) showed clearly that the specificity resides in the C-terminal region. Fine structure mapping of this region by further reciprocal exchanges (examples shown in Fig. 4, lines 9 to 16) and then by exchange of a small segment defined by the activities of reciprocally exchanged hybrids (Fig. 4, lines 5 to 8) showed that the specificity resides in the segments from positions 155 to 181 of λQ and positions 218 to 243 of 80Q; this segment is indicated by a red line in Fig. 1B. Structure prediction suggests that this region consists of two alpha helices separated by a loop of about five amino acids. Note that this is the segment of greatest sequence difference between the two proteins in the right half, a finding consistent with a role in binding specificity.

FIG. 4.

Function of hybrids between λQ and 80Q. For each hybrid protein, the number above indicates the boundary residue from 80Q, and the number below indicates the boundary residue from λQ. In vitro activity was determined at a saturating concentration of Q (500 nM).

Finally, the N-terminal region is also required to maintain full λQ activity (Fig. 4, lines 5 and 6); it may be required to maintain the structural integrity of λQ protein, since hybrid proteins with the 80Q N-terminal regions did not refold efficiently (data not shown).

Mutational analysis of the λQ polypeptide.

The region of the putative specificity element appeared prominently in a screen for defective Q function in a library of randomly mutagenized λQ genes, which identified mutable sites throughout the gene (data not shown). Thus, we obtained defective mutants in 28 codons in the segment containing amino acids 87 to 206 and, of these, 14 codons were in the interval from positions 155 to 190 that covers the predicted helix-loop-helix segment. This mutagenesis was not saturated but was sufficiently dense so the distribution is likely to be meaningful: there were two occurrences in seven of the 28 codons, three occurrences in two of the codons, and five occurrences in one codon.

To determine in more detail regions active in DNA binding, we constructed point mutants by alanine scanning mutagenesis throughout the region of specificity. Figure 5 shows the result of in vivo and in vitro activity assays and DNA binding assays of purified proteins modified by alanine substitution. It is clear that the region of the second predicted helix (residues 173 to 190) and probably the adjacent upstream loop are required for both activity and DNA binding. Some of these substitutions probably disrupt protein structure, but others (e.g., S177, R178, and K181) are likely to remove solvent-exposed functional groups.

FIG. 5.

Activity of alanine substitutions in the region of specificity of λQ. The activity in vivo was determined as described. Activity in vitro was determined as the percent readthrough of the terminator in transcription by 500 nM purified Q, normalized to 100 for wild-type λQ. DNA binding was measured by bandshift assay of purified protein and is given as 100 × (K_d with wild-type protein)/(K_d with mutant protein), except that the lowest values are plotted arbitrarily as 3 in order to indicate where the assays were done.

Mutations that increase λQ activity and DNA binding.

DNA binding or other functional elements may also be defined by mutations that increase activity. To seek such mutants, we screened randomly mutagenized Q for greater activity on the point mutants −25C and −22G of the Q binding element. We identified three strong suppressors, E134K, V189E, and H192Y, defined as giving 90 to 100% of wild-type activity on the −22A mutant, and three weaker suppressors, W136R, V189A, and N200D, defined as giving 60 to 70% wild-type activity on −22A (Fig. 1C). Although suppressors specific for one mutant might have identified base-specific contacts, we found that the three mutants −22A, −22G, and −25C were suppressed similarly by all six suppressors (Table 2 and data not shown). Thus, the suppressors likely have general effects on DNA binding. None are in the specificity element, but all surround it on the C-terminal half of λQ, suggesting that an extended region of the C-terminal half contributes to DNA binding activity.

TABLE 2.

Activity of suppressor λQ proteins on wild type and mutant qutλ^a

Activity type	qutλ DNA	λQ
		WT	E134K	H192Y
In vivo activity (% WTλQ)	WT	100	125	112
	−22A	52	100	92
	−22G	36	92	72
	−25C	25	76	62
	−13/−15A	15	13	35
In vitro activity (maximum % RT, Km [nM])	WT	44, 3.8	50, 0.6	49, 1.6
	−22A	21, 28	57, 2.7	49, 10
	−22G	20, 39	55, 2.5	44, 21
	−25C	21, 35	54, 9	40, 27
	−13/−15A	18, 29	30, ND	42, ND
DNA-binding activity (Kd [nM])	WT	1	0.03	0.4
	−25C	15	0.9	3.8
	−13/−15A	10	1.6	3.1
	−22A or −22G	ND	ND	ND

^aWT, wild type; ND, not determined; RT, terminator readthrough.

We assayed purified mutant E134K and H192Y proteins for antitermination activity and for DNA binding (Table 2 and Fig. 6). In general, there is good correspondence between the increase in function in vivo and in vitro; furthermore, the mutants bind both mutant and wild-type DNA more tightly than does wild-type λQ. It is reasonable to conclude that the primary defect of qut mutants at positions −22 and −25 is λQ binding and that the suppressors act by strengthening λQ binding. It is noteworthy that the activity of the −13A −15A mutant is poorly suppressed, particularly by E134K, which has the tightest binding constant and which restores the band shift with the −13A −15A mutant to nearly wild-type strength. This result indicates that increased binding does not suppress the −13A −15A defect, suggesting a conformational disruption in the complex caused by these mutations, in addition to their overall effect on the binding constant. It is also noteworthy that both suppressors slightly but reproducibly increase antitermination on wild-type DNA at saturating Q concentration, a finding consistent with the conjecture above that strength of binding is important independent of DNA occupancy by Q.

FIG. 6.

Antitermination and DNA binding activity of suppressor λQ. (A) Antitermination activity in vitro of wild-type λQ on wild-type and mutant qutλ; (B) antitermination activity in vitro of E134K λQ on wild-type and mutant qutλ; (C) DNA binding activity of E134K λQ and H192Y λQ on wild-type qutλ.

A putative zinc binding module in Q.

A set of four cysteines in λQ at 118, 121, 144, and 147 are arrayed similarly to those in characterized zinc fingers (8), here with the configuration CX₂CX₂₂CX₂C; this segment has a distinct similarity to the zinc finger of DnaJ, which is identified in an NCBI BLAST search of λQ (although not with high overall significance). The cysteine motif is especially similar in the occurrence of two glycine residues downstream from each cysteine pair, giving for each pair the configuration CX₂CXGXG. Like the specificity region, the putative zinc finger was prominent in the randomly mutagenized library: of the 28 mutated codons found in the 124-amino-acid C-terminal segment, 5 were among the four cysteines and four glycines of these motifs. To further characterize these regions, we made conservative serine substitutions of each cysteine, as well as histidine substitutions, which are functional in some zinc finger structures. All of these substitutions are inactive (Table 3). In contrast, substitutions at three other cysteines in λQ have little or modest effect on activity (Table 3). In 80Q, the sequence homologous to the upstream λQ (CX₂C) motif at residues 118 to 125 has only one residue between the first two cysteines, but there are two more potential elements of configuration CX₂CXGXG within the 80Q “insertion,” marked by parentheses in Fig. 1C. Thus, there are several possible arrangements of zinc finger structures in 80Q.

TABLE 3.

Effect of changes in potential zinc finger residues on Q activity

Amino acidsubstitutiona	In vivo activity (% WT λQ)b
C118S	9.2
C121S	8.9
C144S	9.1
C147S	7.6
C118H	8.0
C121H	7.1
C144H	8.0
C147H	8.4
C121A	6.5
C53S	50.0
C108S	85.0
C191A	86.4

^aCodon changes are TGC(Cys)→AGC(Ser), TGC(Cys)→CAC(His), and TGC(Cys)→ GCC(Ala).

^bWT, wild type.

Unlike the specificity mutants, the putative zinc finger mutants of λQ are not preferentially defective in binding qutλ site mutants, on either side of the Q binding region (Table 4). Thus, the putative zinc finger mutants that are about 10- to 20-fold defective relative to wild-type Q in binding wild-type qut generally show only 2- to 10-fold further defect in binding either the −13 −15A mutant or the −25C mutant. In contrast, the specificity mutants that are comparably defective for wild-type DNA (Y161A, T175A, and V180A) are essentially unable to bind either mutant qutλ. One consistent interpretation is that the zinc finger motif provides a relatively small fraction of the binding energy, and the specificity region interacting with bases in the interval from −13 to −27 provides most of the binding energy.

TABLE 4.

Effects of substitutions in putative zinc finger residues and specificity element residues on DNA binding by λQ^a

λQ protein	Kd (nM)
	qutλWT	qutλ−13/−15A	qutλ−25C
WT	1	10	15
C118G	15	31	39
G123S	9	29	45
G125D	9	24	28
C144G	24	117	159
C144R	10	18	33
C147Y	22	57	81
Y161A	20	≥1,000	≥1,000
T175A	6	≥1,000	≥1,000
V180A	8	≥500	≥500

^aWT, wild type.

DISCUSSION

We have identified regions of both λ and 80 Q proteins and their DNA binding sites that confer specificity for Q function. These protein regions also are required for DNA binding; thus, replacement of the specificity region of qutλ with that of 80 both changes qut specificity and destroys binding activity for λQ. Furthermore, point mutations of qutλ impair both Q function and Q binding in vitro. However, Q function depends on DNA binding in a complex way. Thus, the defect of certain qut mutants that bind Q weakly is not overcome by higher concentrations of Q, and some combinations of mutant Q and qut show nearly wild-type binding activity but no function. We conclude that binding is necessary but not sufficient for activity. What else, then, is required? We suggest two possibilities that are not exclusive. First, a precise alignment of the Q polypeptide may be required for engagement with the σ70-containing paused RNAP elongation complex. The steps of engagement are not known, except that λQ displaces σ70 region 4 from a binding site in the Q binding element to a −35-like element located at positions −7 to −2; a specific protein-protein interaction between Q and region 4 of σ70 facilitates this displacement (11). Second, the dissociation rate of Q might need to be sufficiently slow to allow a sequence of subsequent steps to occur; a dissociation rate increased by mutation might not be overcome by a higher Q concentration.

We can discern distinct DNA binding elements of the Q polypeptide and, possibly, of the qut site. The protein specificity regions, segments from positions 155 to 181 of λ Q and positions 218 to 243 of 80Q, likely contribute to the binding in the region of qut from positions −22 to −28: some mutations that change these segments of either Q or qut have no detectable DNA binding activity, and some combinations of a partially defective protein with mutant DNA give a greatly enhanced defect. The putative zinc finger mutants behave differently; they all have a modest effect on DNA binding, and they do not act synergistically with mutations in the Q binding element. We suggest that this element either affects overall structure of the DNA binding domain of Q or it contacts DNA at an unidentified locus to provide a small amount of binding energy. The location of two mutations increasing DNA binding (E134K and W136R) within the potential zinc binding motif, as well as their positive charge, is consistent with a distinct role in DNA binding.

The qut mutations at positions −13 and −15 may represent a binding site with a function distinct from that in the specificity segment from positions −22 to −28. First, the mutations at positions −13 and −15 change the footprint of λQ on DNA in a qualitative manner: whereas λQ protects wild-type DNA from cleavage by MPE approximately uniformly from positions −10 to −30, the −13/−15 double mutant DNA is protected only from positions −18 to −30 (21). Second, the −13 and −15 double mutation is suppressed poorly by substitutions screened to suppress mutations at −22 and −25, even though in vitro binding of λQ is restored nearly to the wild-type binding constant. Thus, the conformation or persistence of the structure bound in the vicinity of positions −13 and −15 may be important independently of the residence of the whole Q polypeptide on DNA.

It remains to be shown whether a metal is in fact bound to the putative zinc finger domain, although treatment of Q with a chelating agent destroys DNA binding activity, and this can be restored by addition of zinc ions (data not shown); we could not, however, restore antitermination activity. Since the Q protein is prepared through steps of denaturation in guanidine and refolding in buffers without explicit addition of zinc, an essential metal would have to be a trace component of the buffers used.

The random library yielded some more information about the distribution of mutable sites in the λQ polypeptide. In addition to the 28 mutated codons in the portion from residues 87 to 207, we obtained 12 mutated codons in the portion from positions 1 to 86; there were six single codon occurrences, four double occurrences, one triple occurrence, and one quadruple occurrence. Seven mutated codons were in the region from positions 48 to 58, a segment mostly of predicted alpha helix. Another notable feature of the mutant collection is the presence of lethal substitutions at either extreme of the polypeptide: both V6E and T206P gave ca. 5% wild-type activity. The functions of these terminal regions remain to be determined.