Suppression of Factor-Dependent Transcription Termination by Antiterminator RNA

Nascent transcripts of the phage HK022 put sites modify the transcriptionelongation complex so that it terminates less efficiently atintrinsic transcription terminators and accelerates throughpause sites. We show here that the modification also suppressestermination in vivo at two factor-dependent terminators, onethat depends on the bacterial Rho protein and a second thatdepends on the HK022-encoded Nun protein. Suppression was efficientwhen the termination factors were present at physiological levels,but an increase in the intracellular concentration of Nun increasedtermination both in the presence and absence of put. put-mediated antiterminationthus shows no apparent terminator specificity, suggesting thatput inhibits a step that is common to termination at the differenttypes of terminator.

INTRODUCTION

Top

Introduction

After initiating RNA synthesis, RNA polymerase (RNAP) continuesto elongate the transcript until it reaches a termination site.At such sites, the enzyme has a high probability of dissociatingfrom the transcript and the template (reviewed in references 26and 32). Bacteria have two basic types of transcription terminationsignals, which differ in their requirements for halting elongation.Intrinsic terminators can stop transcription through the actionof the nascent transcript. Formation of an RNA stem-loop immediatelyupstream of a U-rich stretch in nascent RNA disrupts RNA-DNAbase pairs within the transcription elongation complex, andthis destabilizes the complex (15, 20, 47). By contrast, factor-dependentterminators recruit a termination factor to the nascent transcript.Two termination factors have been well characterized: the bacterialRho protein and the bacteriophage-encoded Nun protein. Afterbinding to nascent transcripts, they both act on the nearbyelongation complex. Rho has an ATP-driven RNA-DNA helicase activity,which is thought to destabilize the elongation complex (7, 30).Nun is transferred from its RNA binding site to the elongationcomplex, where it is thought to anchor RNAP to the DNA templatewithin a few hundred nucleotides downstream of the binding site (16, 39, 43).Dissociation of Nun-arrested polymerase from the template andthe transcript has not been observed in vitro and appears torequire an additional factor or factors. Recent evidence suggeststhat the Escherichia coli Mfd protein can stimulate the dissociationof Nun-arrested complexes (42).

E. coli and its bacteriophages alter the efficiency of transcription terminationin order to control the expression of genes located downstreamof terminators (reviewed in reference 44). For example, the phage ${lambda}$ antitermination proteins N and Q modify RNAP so that it readsthrough intrinsic and rho-dependent terminators. Both N andQ recognize specific phage sequences (nut and qut, respectively)before they modify polymerase, and this limits antiterminationto RNAP molecules that are transcribing phage DNA. ElongatingRNAP that has been modified by interaction with either proteinretains the modification as it translocates, as shown by its abilityto read through multiple, sequential terminators. The actionof both proteins is enhanced by host-encoded factors. Transcriptionof the rRNA operons in E. coli is also subject to antitermination controland, as in the case of ${lambda}$ N and Q, cis-acting sequences (boxAsites) located near rRNA promoters limit antitermination topolymerase molecules that are transcribing rRNA operons (6, 11).Ribosomal antitermination also requires trans-acting factors (41, 48).

Bacteriophage HK022 is a relative of ${lambda}$ that also antiterminatestranscription in order to increase the expression of genes locateddownstream of termination sites (reviewed in reference 45).However, in contrast to the factor-dependent antiterminationmechanisms outlined above, transcription of cis-acting, promoter-proximalphage sequences (put sites) is sufficient to convert RNAP intoa termination-resistant form; no dedicated factors are absolutely required.We refer to this as intrinsic antitermination. put differs insequence from the nut, qut, and ribosomal operon boxA sites.Computer modeling and enzymatic probing of RNAs synthesizedin vitro suggest that the put transcripts fold into two stem-loopsseparated by an unpaired base (4). The stems are required forput function, since mutations that prevent base pairing reduceantitermination, and additional mutations that reestablish basepairing but not the original sequence restore antitermination(18). Nascent put RNA binds to the transcription elongationcomplex and remains associated with it through subsequent translocation.Stable binding is required for antitermination (35).

The distinction between intrinsic and factor-dependent antiterminationis highlighted by the following observations. First, E. coli mutantsthat are defective in put-mediated antitermination supportedN-mediated, Q-mediated, and ribosomal operon antitermination (10).Conversely, host mutants that are defective in ${lambda}$ N-mediated antitermination supportedthe growth of HK022 (3). The two types of mutations changeddifferent host proteins: those defective in HK022 antiterminationaltered the ß'-subunit of RNAP, and those defectivein ${lambda}$ antitermination altered the E. coli Nus proteins or theß-subunit of RNAP (13, 24). Finally, purified wild-typepolymerase efficiently read through multiple sequential intrinsicterminators that were fused to a wild-type put site. Efficientreadthrough did not require additional protein factors but wasprevented by a ß' mutation that is defective for antiterminationin vivo (10, 18).

The different factor and site requirements of the antiterminationsystems cited above could influence the spectrum of terminatorsthat each is capable of suppressing. The N/nut and Q/qut pathwaysprevent termination at both intrinsic and Rho-dependent terminators,suggesting that they interfere with a step that is common toboth types. The ribosomal boxA pathway promotes efficient readthroughof Rho-dependent terminators but is ineffective or less effectiveagainst intrinsic terminators (1). The put pathway is knownto suppress several intrinsic terminators. Here we show thatput also promotes readthrough of three factor-dependent terminators,one that requires Rho ( ${lambda}$ T_R1) and two that require Nun ( ${lambda}$ nutLand ${lambda}$ nutR).

MATERIALS AND METHODS

Top

Materials and Methods

Bacteria, phages, and plasmids. The strains, phages, and plasmids used in this study are listedin Table 1.

View this table:
[in this window]
[in a new window]
TABLE 1. Bacteria, plasmids, and phage strains^a

Bacterial growth, media, biochemicals, and antibiotics. Cell cultures were grown in Luria-Bertani (LB) or tryptone broth(TB) (25). Antibiotics were added (when required) at the followingconcentrations: ampicillin, 100 µg/ml; kanamycin, 50 µg/ml;chloramphenicol, 30 µg/ml; spectinomycin, 25 µg/ml.Fusions that contained the P_TAC or P_Lac promoters were inducedwith isopropyl-ß-D-thiogalactopyranoside (IPTG; 1mM final concentration; purchased from Gold Biotechnologies). o-Nitrophenyl-ß-D-galactopyranosidewas purchased from Sigma. Restriction enzymes, Klenow, and ligasewere purchased from New England Biolabs. Bacteriophage weregrown and assayed as described previously (2). Oligonucleotideswere purchased from BioServe (Laurel, Md.).

Cloning of the ${lambda}$ T_R1 terminator. A fragment containing the ${lambda}$ T_R1 terminator was amplified by PCRfrom ${lambda}$ DNA using oligos RK76 (5'-CATCGGATCCTGGAACAACGCATAACCC-3') andRK78 (5'-TGCAGGATCCCTATGTAAGTATTTCC-3'). RK78 primes downstreamof the ${lambda}$ T_R1 near the cII translation start site. RK76 primesat the beginning of the cro coding sequence and changes the initiatingcodon to prevent cro translation. The incorporated BamHI restrictionsites used for cloning are underlined. The amplified fragmentwas digested with BamHI and cloned into the reporter constructsshown in Table 1. All fusions made in this study were sequencedat the University of Maryland Biopolymer Laboratory. All fusionsthat contain the cro-TR1 region of lambda are signified by aT_R1 notation.

Cloning the ${lambda}$ nutL and nutR sites. The ${lambda}$ nutL site was amplified from pSB513 with primers RK88 (5'-CAGCGAATTCTGAAGGTGACGCTCTTAAAAATT-3') andSBS59 (5'-CGCCGGAGATCTCTGCAGTGGAGCGGGCAGCGGG-3'). The incorporatedEcoRI site in RK88 is underlined. The purified PCR productswere digested with EcoRI, and the ends of the resulting 64-bpnutL-containing fragment were filled with Klenow and clonedinto the SmaI site of pRAK31 (18).

The ${lambda}$ nutR sequence was amplified from plasmid pJT6 with RK95 (5-TACGGATATCAATAACCCCGCTCTTAC-3') andRK96 (5'-GCTGGATCCGTTTAATTTGATGCCCTTTTTC-3'). The EcoRV andBamHI sites used for the cloning are underlined. The purifiedproducts were digested with EcoRV and BamHI and cloned into SmaI-BamHI-digestedpRAK31. All fusions that contain only the nutR-TR1 region aresignified by a nutR-T_R1 notation.

Crossing lacZ fusions onto ${lambda}$ . The lacZ fusions in pRS415 were crossed from the plasmids inwhich they were constructed onto ${lambda}$ RS88 as described previously (18, 38).The copy number of ${lambda}$ prophage was determined as described elsewhere (31).

ß-Galactosidase assays. ß-Galactosidase activities of cells carrying P_TAC-lacZfusions were assayed in microtiter plates as described previously (18).Overnight TB broth cultures were diluted into fresh TB containing antibioticsand incubated for approximately 2 h. The exponentially growingcultures were then diluted into TB alone or TB supplementedwith 1 mM IPTG. The cultures were incubated for 1 h more andthen assayed for ß-galactosidase activity. ß-Galactosidaseactivities of cells carrying P_L(HK022)-lacZ fusions were determinedas described elsewhere (25). Overnight cultures were dilutedinto LB broth with or without 1 mM IPTG and grown for 3 to 4h to cell densities of 2 x 10⁸ to 5 x 10⁸/ml before assay.

Plasmid curing. To measure steady-state levels of ß-galactosidasein strains that carried single-copy HKP_L-putL-nutL-lacZ (or nutR-lacZ)fusions, the plasmid that expresses HK022 repressor (pRAK262)was removed by growing cultures at 42°C in LB for severalgenerations. The loss of pRAK262 was confirmed by screeningfor chloramphenicol sensitivity and for immunity to HK022.

RESULTS

Top

Results

put-mediated suppression of a Rho-dependent transcription terminator. Lambda T_R1 is a well-characterized Rho-dependent terminator (14, 21, 34).We measured the efficiency of termination at T_R1 by comparingthe activity of ß-galactosidase produced after inductionby cells containing a single copy of a P_TAC-T_R1-lacZ transcriptionfusion to that produced by cells containing a P_TAC-lacZ transcriptionfusion. We measured put-mediated antitermination of T_R1 in twoways. First, we compared ß-galactosidase activityproduced by a P_TAC-putL-T_R1-lacZ fusion to that produced bycomparable fusions that lacked a functional putL site. Second,we compared ß-galactosidase activity produced by a P_TAC-putL-T_R1-lacZ fusionin rpoC⁺ cells to that produced by the same fusion in rpoC-Y75Ncells. This mutation, which alters the ß'-subunitof RNAP, prevents put-mediated antitermination at intrinsicterminators (10, 18). The two methods gave similar results.

T_R1 terminated transcription with an efficiency of 94 to 95%in the absence of put or when the cells contained the rpoC-Y75Nmutation (Fig. 1 and Table 2). The presence of a functionalputL site between the promoter and T_R1 in rpoC⁺ cells reducedtermination to 39%. We conclude that put partially suppressestermination at T_R1 and that suppression is prevented by rpoC-Y75N.These conclusions are supported by measurements made with anotherfusion (see below).

View larger version (34K):
[in this window]
[in a new window]
FIG. 1. put-mediated antitermination of the Rho-dependent terminator, ${lambda}$ T_R1. The strains were rpoC⁺ or rpoC-Y75N derivatives of TAP114 and contained single copies of a P_TAC-putL-T_R1-lacZ or a P_TAC-putL-lacZ fusion, as indicated. The mutant putL site has a multiple base substitution that prevents antitermination of Rho-independent terminators (linker scanning mutation G [18]). ß-Galactosidase activities (in arbitrary units) were measured 1 h after addition of IPTG to growing cultures and are the means of at least four independent assays. The open bars ( ${Delta}$ T_R1) report the activities from fusions lacking T_R1, the shaded bars (T_R1 plus Psu) report the activities from those carrying T_R1 in cells that contained a plasmid with a P_TAC-psu fusion (pNL150), and the filled bars (T_R1) report the activities from fusions containing T_R1 in cells that contained a P_TAC- ${Delta}$ psu fusion (pNL151). The standard error of the mean was less than 20%.

View this table:
[in this window]
[in a new window]
TABLE 2. Rho-mediated termination and put-mediated antitermination at ${lambda}$ T_R1^a

The apparent termination efficiency of T_R1 was severalfold higherin our fusions than in phage ${lambda}$ (12). It has previously been observedthat the activity of T_R1 depends on its context (14, 29; D.Court, personal communication), and this might explain the difference.Nevertheless, to confirm that our fusions indicate Rho-dependenttermination rather than termination at an uncharacterized intrinsicterminator, we measured ß-galactosidase activity inthe presence of phage P4 Psu, a protein that antagonizes Rhoactivity (22, 23). We found that Psu significantly increasedß-galactosidase activity if the reporter fusion carriedan inactive put site or if the strain carried the rpoC-Y75Nmutation (Fig. 1). Therefore, the T_R1-containing reporter fusionsdo, indeed, report Rho-dependent termination. Psu had no sucheffect on a put⁺ fusion in rpoC⁺ cells, as expected if put suppressesRho-dependent termination (Fig. 1). Psu did not increase lacZexpression as much as did deletion of T_R1, perhaps because Psudid not completely prevent Rho termination, or not enough timewas allowed to reach the steady-state level of ß-galactosidaseafter Rho action was blocked. The psu gene was derepressed foronly 1 h before ß-galactosidase was measured, becausecontinuous high-level expression is lethal (23). Another possibility,discussed more fully below, is that deletion of T_R1 increasesthe stability or translation efficiency of the lacZ message.

We note that a deletion and a base substitution mutation ofput increased the accumulation of ß-galactosidase1.5- to 3.3-fold in fusions that lack T_R1 (Fig. 1). (We considerpossible explanations below.) The increased activity did notsignificantly change our estimate of T_R1 termination, whichwas 95 to 96% regardless of whether antitermination was preventedby a put mutation or by rpoC-Y75N (Table 2).

put suppresses Nun-dependent termination. The phage HK022-encoded Nun protein terminates transcriptionafter binding to a nascent transcript of the ${lambda}$ nutL or nutR sites(see introduction). To measure Nun-dependent termination, wecompared the steady-state levels of ß-galactosidaseproduced by cells containing a single copy of a P_L(HK022)-putL-nutL-lacZ transcriptionfusion in the presence and absence of Nun. In this fusion, theputL site is in its natural location, immediately downstreamof the HK022 P_L promoter. To see if put suppresses Nun-dependenttermination, we inactivated putL by mutation or prevented putaction with the rpoC-Y75N mutation. A plasmid that containsa P_Lac-nun fusion (pNUN) provided Nun at either low or highconcentration according to whether the culture was grown inthe absence or presence, respectively, of the lac operon inducerIPTG. The low concentration was similar to that found in a single-copyHK022 lysogen, and the high concentration was about 100 timesgreater (reference 19 and other data not shown).

The efficiency of termination at a low Nun concentration was33%, and mutation of rpoC or putL increased this efficiencyto 94 or 83%, respectively (Fig. 2 and Table 3). The efficiencyof termination at a high Nun concentration was 83%, and mutation of rpoC or putL increased this efficiency to 99.7 or 99.5%,respectively. We conclude that put antiterminates Nun-dependenttermination at nutL, and that increasing the Nun concentrationincreases termination in both the presence and absence of put-mediated antitermination.

View larger version (46K):
[in this window]
[in a new window]
FIG. 2. Antitermination of Nun-dependent terminators. The strains were derivatives of TAP114 carrying a single copy of a P_L(HK022)-putL-lacZ fusion. Different fusions contained nutL, nutR, or neither nut site between putL and lacZ, and Nun was provided by a plasmid containing a P_Lac-nun fusion (pNUN). The shaded and filled bars show enzyme activity from mid-log cultures carrying pNUN grown in the absence and presence, respectively, of 1 mM IPTG. The open bars show enzyme activity from mid-log cultures carrying the vector plasmid (pNUN ${Delta}$ ) grown for 3 to 4 h in the presence of 1 mM IPTG. The mutant putL site (linker scanning mutation G [18]) has a multiple base substitution that prevents antitermination of Rho-independent terminators. Activities are averages of assays of at least two independent cultures at two different times during exponential growth. The standard errors of the mean ranged from 3 to 33%.

View this table:
[in this window]
[in a new window]
TABLE 3. Nun-mediated termination and put-mediated antitermination at the ${lambda}$ nutL site^a

The effect of Nun on lacZ expression was, as expected, sitespecific: Nun, even when present at a high concentration, reducedthe activity of a reporter fusion that lacked a nut site byonly a small amount, and the rpoC-Y75N mutation did not furtherreduce ß-galactosidase production (Fig. 2 and Table 3).Unexpectedly, ß-galactosidase activity in the fusionlacking nutL was 0.3 to 0.4 that of the nutL⁺ fusion in theabsence of Nun (Fig. 2). Therefore, we did not attempt to estimatethe efficiency of Nun-dependent termination by comparing theactivities of these two fusions in the presence of Nun (seebelow).

We used a P_L(HK022)-putL-nutR-T_R1-lacZ fusion to measure putsuppression of Nun-dependent termination at the ${lambda}$ nutR site.The nutR region also contains the Rho-dependent ${lambda}$ T_R1 terminator; therutA and rutB sequences, which are required for Rho action,lie immediately upstream and downstream, respectively, of theboxB element of nutR (14). nutL lacks rut sequences but is otherwisevery similar to nutR. We hoped that comparison of the nutL-and nutR-containing fusions would tell us if Rho affects terminationor antitermination at the nutR site. We note that the fusionsused for these experiments have a different promoter, a differenttranscription start, and less transcribed DNA between putL andnutR than the P_TAC-putL-T_R1-lacZ fusion used to measure Rho-dependenttermination in the experiments of Fig. 1 (see Materials and Methods).These differences could, in principle, alter termination and antiterminationefficiencies.

We estimated Nun-dependent termination and put-mediated antiterminationby measuring ß-galactosidase activities in the presenceor absence of Nun in rpoC⁺ or rpoC-Y75N cells, as describedearlier for nutL. The efficiencies of Nun-dependent terminationat nutR at low and high Nun concentrations were similar to thoseobserved at nutL, and put suppressed Nun action at both sitesto approximately the same extent (Fig. 2 and Table 4). Therefore,the presence of rut sites and their interaction with Rho donot appreciably alter Nun termination or put antiterminationat nut (see also reference 33).

View this table:
[in this window]
[in a new window]
TABLE 4. put-mediated antitermination of interdigitated Nun- and Rho-dependent terminators^a

We estimated the efficiency of Rho-dependent termination andput-mediated antitermination at T_R1 by measuring ß-galactosidaseproduced by fusions with or without the nutR-T_R1 region in rpoC⁺or rpoC-Y75N cells. Since these cells did not contain Nun, therewas no Nun-dependent termination at nutR. The estimated efficiencyof termination in the rpoC mutant host was 80%, and terminationwas completely suppressed in the wild-type host (Table 4). Theseestimates are somewhat lower than those presented in Table 2,in which different fusions were used. The differences mightbe the result of differences in the fusions, but there are alsouncertainties in our estimates of termination efficiency, asdiscussed below. Nun caused an additional reduction in ß-galactosidaseactivity produced by fusions containing the nutR-T_R1 region(Fig. 2), but the experimental uncertainties preclude quantitativeestimates of any effect of Nun on Rho-dependent termination.

We wish to call attention to several unexpected observations.Mutation or deletion of putL increased the specific activityof ß-galactosidase 1.5- to 3-fold in certain fusions(Fig. 1 and Table 2 footnote [ ${Delta}$ T_R1 fusions]). In other fusions,deletion of nutL or nutR-T_R1 decreased the specific activity ofß-galactosidase to 0.3 to 0.4 (Fig. 2, No Nun). Wespeculate that these changes increased the stability and/orthe translation efficiency of the lacZ message, but we haveno independent evidence for this hypothesis. Because of this,estimates of termination and antitermination that are basedon comparison of fusions that either contain or lack the terminatoror antiterminator sites, respectively, should be consideredas approximations. In many cases we estimated efficiencies oftermination and antitermination by comparing identical fusionsin the presence and absence of a trans-acting protein, suchas Nun, Psu, or wild-type RNAP, and these estimates are probably morereliable. For example, our estimate that unsuppressed Rho-mediated terminationefficiency at T_R1 is 80% was based on comparison of differentfusions (Fig. 2 and Table 4). Comparison of the amount of ß-galactosidaseproduced by the fusion containing T_R1 in rpoC⁺ and rpoC-Y75Ncells (Fig. 2) suggested that the true termination efficiencyin this fusion could be 90% or more.

DISCUSSION

Top

Discussion

We have shown that put suppresses Nun- and Rho-dependent transcription terminatorsin vivo. Suppression was efficient when the termination factorswere present at physiological levels, but increasing the concentrationof Nun increased termination (on templates containing a nutsite) both in the presence and absence of a functional put site.We previously demonstrated efficient suppression of severalstrong intrinsic terminators in vivo and in vitro. We also showedthat the HK022 put sites could replace the function of the ${lambda}$ N gene and nut sites in ${lambda}$ -HK022 hybrid phages (10, 18, 28). Thehybrid phages showed no obvious defect in lytic growth, lysogenization,or lysogenic induction. It therefore appears that the put sitessuppress the numerous ${lambda}$ terminators in the early operons to theextent required for normal phage growth. put-mediated antitermination thusshows no demonstrable terminator specificity. The ${lambda}$ N/nut andQ/qut antitermination pathways also suppress intrinsic and Rho-dependentterminators, and N suppresses Nun-dependent termination by competingfor binding to nut RNA (5, 8, 16, 33, 46). Q has not yet been testedon a Nun-dependent terminator. This apparent lack of terminator specificitysuggests that put inhibits a step that is common to terminationat the different types of terminator. What might this step be?

A short region of RNA-DNA hybrid at the 3' end of the nascentRNA chain is believed to play a critical role in stabilizingthe transcript elongation complex (27, 36). Current models proposethat intrinsic and Rho-dependent termination are a consequence ofhybrid disruption (15, 20, 40, 47). It is, therefore, temptingto suggest that put RNA acts by increasing the stability ofthe hybrid or by preventing a step that follows hybrid disruption.However, this suggestion fails to account for put suppressionof Nun-dependent terminators. Nun binds to nascent nut RNA andis then delivered to the nearby elongation complex, where itarrests translocation. Arrest is believed to be the result ofa Nun-DNA interaction that anchors the elongation complex to thetemplate so that it can no longer translocate (17, 43). Sincethe arrested complex is stable and catalytically active in vitro,it is not obvious how further stabilization by put RNA couldprevent arrest or restart translocation once arrest had occurredor, if it did, how increasing the concentration of Nun wouldovercome this effect.

We offer two models to explain suppression of Nun-dependenttermination. First, the efficiency of Nun arrest in vitro isdecreased by conditions that increase the rate of translocation (16).It is possible that rapid translocation of RNAP away from theNun binding site decreases the efficiency of transfer of Nunto the elongation complex. If so, put-mediated accelerationof the elongation complex through pause sites (18) could havethe same effect on Nun transfer. Elevating the Nun concentration shouldquicken its binding to nut RNA and thus increase the probabilityof transfer. In the second model, put RNA impedes or delaysbinding of Nun to nut RNA. We have shown that nascent put RNAbinds to elongating polymerase and that this complex persistsas polymerase translocates (35). If the nascent put transcriptbinds close to the product RNA exit channel in RNAP, it mightdelay the binding of Nun to the nascent nut transcript untilthe elongation complex is too distant for efficient transferof Nun to RNAP. This effect would be mitigated by increasing theNun concentration. Either of these models can be adapted toexplain the suppression of other classes of terminators by putRNA. However, it is possible that put suppresses different typesof terminators in different ways and that no single mechanismsuffices to account for its action.

ACKNOWLEDGMENTS

We are grateful to Max Gottesman and Don Court for their commentson the manuscript. We also acknowledge Jennie Tang for constructingplasmid pJT6.

FOOTNOTES

* Corresponding author. Mailing address: National Institute of Child Health and Human Development, NIH, Bldg. 6B, Room 3B308, Bethesda, MD 20892-2785. Phone: (301) 496-3555. Fax: (301) 496-0243. E-mail: rweisberg{at}nih.gov.

Present address: Biology Department, Western Kentucky University,Bowling Green, KY 42101.

REFERENCES

Top