Compromised Factor-Dependent Transcription Termination in a nusA Mutant of Escherichia coli: Spectrum of Termination Efficiencies Generated by Perturbations of Rho, NusG, NusA, and H-NS Family Proteins ^▿

Bacterial strains, plasmids, and growth media.

The strains of E. coli used in the present study are listed in Table 1. Phage λNK1324, used for Tn10dCm transposition, has been described earlier (37), and phage λcI857 was from our lab stock. Media and growth conditions were largely as described in the accompanying study (47). Where necessary, the medium was supplemented with ampicillin (Amp), tetracycline (Tet), chloramphenicol (Cm), kanamycin (Kan), and IPTG (isopropyl-β-d-thiogalactopyranoside) at the described concentrations (47). Spectinomycin (Sp) supplementation was at 50 μg/ml.

Several plasmids were included (with salient features in parentheses): pACYC184 (p15A vector, Cm^r Tet^r [12]), pLG339 and pLG-H-NSΔ64 (pSC101 replicon vector and its derivative encoding H-NSΔ64, respectively, Kan^r [56]), pCP20 (pSC101-based Ts replicon encoding Flp recombinase, Amp^r [18]), pHYD2509 (pSC101 replicon encoding H-NSΔ64, Amp^r [47]), pHYD2546 (p15A replicon carrying ydgT, Cm^r [47]), and two plasmids that are IPTG dependent for replication (29), pHYD751 (nusG⁺, Amp^r) and pHYD1201 (rho⁺, Amp^r). The following three plasmids with nusA⁺ were used: pHYD2554 (Amp^r, carrying the 10-kb EcoRI-HindIII fragment between kilobase coordinates 3310.06 and 3320.08 of the E. coli genome [45] cloned in a ColE1-based replicon, and obtained from Manjula Reddy); pHYD2557 (Cm^r, carrying a 2.3-kb PCR-amplified region between genomic nucleotide coordinates 3314061 and 3316393 [45] cloned in a pSC101-based Ts replicon and obtained from Ranjan Sen); and pHYD2556 (Sp^r, carrying the minimal nusA⁺ open reading frame with its native ribosome-binding site between genomic nucleotide coordinates 3314061and 3315548 [45] cloned downstream of the ara regulatory region in a pSC101-based replicon and obtained from Ranjan Sen).

Isolation and genetic mapping of suppressors of rho and nusG derivatives expressing H-NSΔ64.

For the selection of suppressors in which transcriptional polarity relief was restored, the parental strains were pLG-H-NSΔ64 derivatives of GJ6520 and GJ6524 (rho and nusG, respectively, both carrying the lacZ_U118 [Am] and trpE9777 [Fr] mutations). (Although a procedure for Tn10dTet transposon mutagenesis [37] was attempted in obtaining the suppressors, subsequent experiments demonstrated that suppression was not caused by the transposon insertion [data not shown].) Selections were for Anth⁺ (utilization of anthranilate [Anth] at 20 μg/ml to satisfy the auxotrophic requirement for tryptophan [Trp]), Mel⁺ (utilization of melibiose as sole C source at 39°C), or both (47), that is, for restoration of transcriptional polarity relief of the rho and nusG derivatives even in the presence of H-NSΔ64.

The suppressors obtained were then analyzed as follows. Plasmid preparations from them were transformed into GJ6520 or GJ6524, as appropriate, to determine whether any of the mutations was on the pLG-H-NSΔ64 plasmid itself. For the non-plasmid-borne suppressors, phage P1 lysates prepared on the mutants were used to transduce pLG-H-NSΔ64-bearing derivatives of GJ7361 (rho) and GJ7362 (nusG) to Ilv⁺, and GJ7360 (rho) and GJ7363 (nusG) to Arg⁺ followed by scoring for the polarity-relief phenotypes. The first pair of strains carries an ilv::Tn10 mutation and the second pair an argE::Tn10 mutation that are linked to the rho and nusG loci, respectively; in this manner, suppressors mapping to either rho or nusG were sought to be identified.

The suppressor mutation in nusA was mapped with the aid of a transposon-tagging approach, as follows. Random transpositions of Tn10dCm were generated in GJ7395 following infection with λNK1324, as described previously (37). (GJ7395 is the derivative of the original nusA suppressor GJ7387 in which the FRT-flanked Cm^r insertion in Δ(yefM-yoeB) [13] was excised by site-specific recombination with the aid of plasmid pCP20, as described previously [18].) A P1 lysate prepared on a population of Cm^r cells was used for transduction into GJ6511/pLG-H-NSΔ64, with double selection for Cm^r Anth⁺ colonies. One such clone was shown in subsequent transduction experiments to have the Tn10dCm insertion ca. 86% linked to the suppressor mutation. Molecular mapping of the transposon insertion was undertaken with the aid of an inverse-PCR approach as described previously (30), and the insertion was shown to be located after bp coordinate 3310756 on the E. coli genome (45), downstream of codon 14 in the truB open reading frame at the 71-min region of the chromosome. In additional crosses, the suppressor mutation was also shown to be 92% cotransducible with an auxotrophic ΔargG::Kan mutation (2) in this region (data not shown).

Scoring for synthetic lethal phenotypes in rho and nusG derivatives.

Tests for synthetic lethality conferred by additional mutations in rho and nusG strains were performed with derivatives carrying the cognate rho⁺ or nusG⁺ genes on the IPTG-dependent replicons pHYD1201 or pHYD751, respectively, and then testing for growth on medium not supplemented with IPTG.

Other techniques.

Procedures for P1 transduction (37), in vitro DNA manipulations and transformation (46), and β-galactosidase assays (37) were as described. The scoring of phenotypes associated with transcriptional polarity and its relief (including quantitation of the extent of polarity relief with the aid of P_lac-lacZ and P_lac-t_R1-lacZ fusion construct pairs wherein t_R1 is a Rho-dependent terminator from lambdoid phage H19B), as well as determination of in vivo transcription elongation rates were undertaken as described in the accompanying study (47). The chromosomal rho and nusA genes were PCR amplified with the primer pairs 5′-TCCTCGACGCTAACCTGGC-3′/5′-ACATCGCCAGCGCGGCAT-3′ and 5′-TTACGCTTCGTCACC-3′/5′-GAAGGCGAAGCTTGTTCCCCACTT-3′, respectively, and sequenced on both strands with these primers, along with additional primers internal to the genes. The plasmid-borne hns gene was PCR amplified with a pair of flanking vector-based primers 5′-AAGTGCGGCGACGATAGT-3′ and 5′-CCGTCTTTCATTGCCATAC-3′ and sequenced on both strands with these primers.

Selection for suppressors restoring polarity relief in rho and nusG derivatives with H-NSΔ64.

As described earlier (29, 47), the dominant-negative H-NSΔ64 variant (that comprises the N-terminal segment of 63 amino acids, from the 137-amino-acid native protein) restores transcriptional polarity in the rho(A243E) or nusG(G146D) mutants. In the present study, we obtained suppressors of the hns effect on rho and nusG, by selection for mutants in these backgrounds in which polarity relief had been reestablished at trp and lac loci bearing polar mutations in the first genes of each of the two operons (trpE9777 [frameshift] and lacZ_U118 [amber], respectively). Derivatives exhibiting relief of transcriptional polarity at these loci were selected as Anth⁺ and Mel⁺ (at 39°C), respectively, as described above. A phenotype correlated with transcriptional polarity relief in the rho and nusG mutants is that of R-loop-dependent lethality with ColE1 family plasmids such as pACYC184, which is also reversed by H-NSΔ64 (26, 29, 47), and all of the Mel⁺ Anth⁺ suppressors obtained in the present study had also reverted to the pACYC184 lethality phenotype.

The rho and nusG strains used in the Anth⁺ selections above were hns⁺ on the chromosome and carried the gene encoding H-NSΔ64 on a pSC101-based plasmid (56). In several of the spontaneous Mel⁺ Anth⁺ derivatives that were obtained, the suppressor mutations were shown to be plasmid-borne, and sequencing of the hns gene on the plasmid revealed that the suppressor mutation in each of them had either (i) reverted the frameshift allele encoding H-NSΔ64 to hns⁺ or (ii) led to a total loss of hns function, including the dominant-negative effect of H-NSΔ64 (data not shown). That the two sets of mutant plasmids had now become wild type and null, respectively, for H-NS function and had lost their dominant-negative character was confirmed by measurements of expression from a proU-lac transcriptional fusion (which is known to be H-NS regulated [56]), in derivatives that were chromosomally hns⁺ or null hns and carrying the mutant plasmids (data not shown). These results served to reaffirm the strong effect of H-NSΔ64 in restoration of transcriptional polarity in the rho and nusG mutants.

Three chromosomal mutants obtained as Anth⁺ Mel⁺ in the selections above (Fig. 1) were further characterized. One of these, obtained in the nusG derivative with H-NSΔ64, was mapped to the rho locus and then identified as a mutation leading to a Q32R substitution in Rho; when the single mutant rho(Q32R) was constructed and tested, it exhibited moderate polarity relief (Anth⁺ with trpE9777) and was killed with pACYC184, with both phenotypes being suppressed by H-NSΔ64 (Fig. 2). Another suppressor mutation, obtained in the parental rho-4, that is, rho(A243E), strain with H-NSΔ64, was also in rho, resulting in an additional R102S substitution in the protein (Fig. 1). These two results suggested that one method of achieving polarity relief in H-NSΔ64-bearing derivatives is to increase the severity of deficiency in transcription termination beyond that conferred by the rho-4 or nusG(G146D) mutations themselves. The rho(R102S, A243E) double mutant exhibited pronounced rich medium (LB) sensitivity that was not affected by the presence or absence of H-NSΔ64 (data not shown), and all experiments with this mutant were performed in minimal medium.

Identification of a transcription termination-defective nusA mutation.

The third chromosomal suppressor mutation (that was obtained in the nusG parent with H-NSΔ64, Fig. 1) was shown by phage P1 crosses to be in neither rho nor nusG (data not shown). The mutation was then mapped, with the aid of a transposon-tagging approach as described in Materials and Methods, to the 71-min region of the E. coli chromosome, ca. 90% linked to argG. By transduction experiments, it was shown that the tagged mutation could also suppress the effects of H-NSΔ64 in a rho-4 background (GJ7417) for both polarity relief and plasmid pACYC184 lethality (data not shown). Since nusA is in this location, we tested whether the suppressor mutation was in nusA. Introduction of any of three plasmids carrying nusA⁺ (pHYD2554, -2557, or -2556, of which the last is a minimal nusA⁺ construct) reversed the phenotypes of polarity relief and lethality with plasmids of the ColE1 family in the suppressor-bearing strain GJ7395 (data not shown). The nusA locus of the suppressor was then PCR amplified and sequenced, which showed that the gene had a mutation resulting in an Arg-to-Cys substitution at amino acid residue 258 in the KH1 domain of the protein (R258C).

Single mutant derivatives of nusA (without concomitant expression of H-NSΔ64) were then constructed and they exhibited several phenotypes. At the t_R1 terminator (of lambdoid phage H19B) positioned upstream of a lacZ reporter gene on the chromosome, the nusA(R258C) mutation conferred moderate transcriptional polarity relief (Table 2) compared to that by the rho(A243E) or nusG(G146D) mutations (47). The nusA strain was also polarity relieved at trpE9777 (Anth⁺) and killed following pACYC184 transformation (Fig. 2). All of these phenotypes were suppressed by expression of H-NSΔ64 (see Table 2 and Fig. 2), as had been shown earlier also for the rho and nusG mutants, and the strain was also complemented by plasmid pHYD2556 (carrying minimal nusA⁺ under P_ara) in an l-arabinose-dependent manner (Fig. 3).

As described in the accompanying study, overexpression of another protein of the H-NS family YdgT is associated with suppression of polarity relief and pACYC184 lethality phenotypes in rho and nusG strains (47) and, likewise, the nusA mutant phenotypes were also suppressed upon YdgT overexpression (Table 2, Fig. 2). These results established that the nusA(R258C) mutant is defective for Rho-dependent transcription termination.

We also tested the effect of the rpoB8 mutation in the nusA strain. RNA polymerase bearing the RpoB8 subunit exhibits a reduced rate of transcription elongation which is associated with suppression of transcription termination deficiency of rho mutants (31, 51). We found that, even in the nusA(R258C) strain, rpoB8 suppressed both the Anth⁺ phenotype associated with trpE9777 and plasmid pACYC184 lethality (Fig. 2).

The nusA(R258C) mutant exhibited a normal growth phenotype and also was unaffected in its ability to support the growth of phage λ at all temperatures tested (see Fig. 4 for 42°C, data not shown for 30 and 37°C), whereas the nusA1 mutant was λ resistant as previously reported (51).

Synthetic lethal phenotypes between rho, nusG, and nusA mutations.

Our findings that the suppressor selection for polarity relief in the parental rho-4 or nusG derivatives with H-NSΔ64 yielded mutations that were themselves associated with defective transcription termination led us to examine the synthetic phenotypes, if any, of the various mutant combinations in absence or presence of H-NSΔ64. The mutant combinations were assembled by P1 transductions in strains carrying the cognate rho⁺ or nusG⁺ gene on an IPTG-dependent replicon, so that the synthetic phenotypes (of both viability and transcriptional polarity relief) could then be determined on growth media not supplemented with IPTG.

By this assay, the mutant combinations rho(A243E) nusA(R258C) and nusG(G146D) nusA(R258C) exhibited synthetic lethality, and furthermore, transformation of these strains with a plasmid encoding H-NSΔ64 restored viability, but the resultant derivatives exhibited relief of transcriptional polarity at trpE and lethality with plasmid pACYC184 (Fig. 5 A). Likewise, the rho(A243E) nusG(G146D) combination has previously been reported to be inviable (29), which was reproduced in the present study, and its H-NSΔ64 derivative was viable but polarity relieved (Fig. 5A). The combinations rho(R102S, A243E) nusG(G146D) and rho(R102S, A243E) nusA(R258C) were also inviable, and the latter was rescued by expression of H-NSΔ64 (data not shown). The rho(Q32R) mutation conferred a slow-growth phenotype with nusA(R258C) but not nusG(G146D), and the former was suppressed by H-NSΔ64 (Fig. 5A).

Two triple mutant combinations were tested (Fig. 5B): rho(Q32R) nusG(G146D) nusA(R258C) was viable only in the presence of H-NSΔ64, whereas rho(A243E) nusG(G146D) nusA(R258C) was inviable even with H-NSΔ64, that is, it could be maintained only in the simultaneous presence of both the IPTG-dependent replicon with rho⁺ or nusG⁺ and the plasmid expressing H-NSΔ64.

To quantitate the extent of polarity relief in many of the viable double and triple mutant combinations (with or without expression of H-NSΔ64), β-galactosidase activities were determined from pairs of P_lac-lacZ fusion derivatives with or without the intervening Rho-dependent t_R1 terminator of lambdoid phage H19B, as described previously (47). The data from these experiments, presented in Table 3, were consistent with the notions that combining the individual mutations leads to additive effects on polarity relief and that H-NSΔ64 expression is ameliorative in this regard. Thus, for example, the rho(Q32R) mutation conferred polarity relief to the extents of, respectively, 30, 34, and 53% by itself and in combination with the nusG and nusA mutations in the absence of H-NSΔ64, which were then reduced to 15, 18, and 32%, respectively, in the presence of H-NSΔ64. The rho(Q32R) nusA nusG triple mutant (which is lethal without H-NSΔ64) exhibited 34% polarity relief with H-NSΔ64. Likewise, the nusA rho(A243E) and nusA nusG mutants with H-NSΔ64 were more polarity relieved (40 and 30%, respectively) than any of the corresponding single mutants with H-NSΔ64 [nusA, 5%; rho(A243E), 18%; and nusG, 17%] (see Table 2 and reference 47). In strains bearing the rho(R102S, A243E) allele alone or in combination with nusA, the same trend was observed. However, since the corresponding derivatives without the t_R1 terminator exhibited 2- to 3-fold lower levels of lac expression than those not carrying this rho allele, the calculated extent of transcriptional readthrough was apparently elevated in these strains. [As mentioned above, strains with rho(R102S, A243E) are LB sensitive, and hence the derivatives had been cultured in a different minimal A medium for the enzyme assays.]

H-NSΔ64 was not able to suppress either the sickness of ΔnusG or ΔnusA mutants in the reduced-genome E. coli strain MDS42 (9, 42) or the inviability associated with these mutations or with Δrho in wild-type E. coli (data not shown).

Transcription elongation rate is unaffected in nusA mutant.

The NusA protein has been implicated in modulating the rate of transcription elongation (40, 44, 48) and thus in ensuring the coupling of transcription and translation (58); furthermore, the kinetic coupling model (31) posits that the efficiency of Rho-dependent termination is inversely related to the elongation rate of transcription. Hence, we considered the possibility that the nusA(R258C) mutant is termination-defective because of an inappropriately increased transcription elongation rate in the mutant leading to a mismatch in coupling of RNA polymerase with the pioneer ribosome and with Rho, which would then also explain its suppression by the rpoB8 mutation (Fig. 2).

We therefore determined the rate of transcription elongation in vivo at the native lacZ locus following induction of the operon with IPTG, as previously described (31, 47), but no difference was observed between the pair of isogenic nusA⁺ and nusA(R258C) strains (Fig. 6). Transcription elongation rates in the rpoB8 nusA⁺ and rpoB8 nusA(R258C) pair of strains were also determined; although rpoB8 itself was associated with a decreased elongation rate as has been reported earlier (31), once again the nusA mutation had no additional effect (Fig. 6). We conclude that NusA(R258C) does not grossly alter the transcription elongation properties of RNA polymerase in vivo.

A nusA missense mutant defective in Rho-dependent transcription termination.

By using a suppressor selection approach in rho or nusG strains expressing H-NSΔ64, we have identified new mutations that affect the efficiency of Rho-dependent transcription termination, including a recessive missense mutation (R258C) in the KH1 domain-encoding region of the nusA gene. The nusA mutant is specifically affected for Rho-dependent termination, without any other apparent phenotype, such as growth sensitivity at low or high temperatures, λ phage resistance, or alteration of transcription elongation rates in vivo.

An earlier study had shown that a G253D substitution in the same KH1 RNA-binding domain is associated with phenotypes of cold sensitivity and λ resistance (59). The nusA1 and nusA11 substitution mutations, both of which affect the S1 domain of the protein, confer transcriptional polarity relief, but they are associated with additional phenotypes such as λ resistance (nusA1) and temperature sensitivity (nusA11) (23, 39, 53). Our findings also support those of Cardinale et al. (9), who showed that a nusA insertion-deletion mutation, in the reduced-genome E. coli strain MDS42, renders transcription termination defective in the strain.

NusA in E. coli has been best studied as a transcription elongation factor and as a protein required for transcription antitermination in rrn operons and lambdoid phages (6, 17, 40, 44, 48). Previous reports have also suggested that NusA can indirectly affect Rho-dependent termination by its influence on the rate of transcription elongation and thereby on transcription-translation coupling (31, 58). However, we found that the nusA(R258C) mutant is unaltered for the rate of transcription elongation at the lacZ locus in vivo, which is consistent with recent findings that NusA's roles in transcriptional pausing and intrinsic termination are mediated solely by its N-terminal domain and are unaffected by substitutions in the KH1 or KH2 domains (28). Therefore, the simplest model to explain our results and those of Cardinale et al. (9) is that NusA participates directly in the process of Rho-dependent transcription termination, rather than indirectly through its effects on transcription elongation.

Spectrum of efficiencies of Rho-dependent termination and their associated phenotypes.

Our studies demonstrate that, by combining different mutations in rho, nusG, and nusA, along with expression of YdgT or of variants of H-NS such as H-NSΔ64, one can generate a wide spectrum of efficiencies of Rho-dependent termination in vivo. As depicted schematically in Fig. 7, the corresponding phenotypes would extend from viable cells exhibiting full transcriptional polarity (highest efficiency) through viable but polarity-relieved cells (intermediate efficiency) to inviable cells (lowest efficiency). Perturbations involving the H-NS family of proteins, as with expression of H-NSΔ64 or of YdgT, are associated with a shift in the spectrum from lower to higher efficiency, such as to convert synthetic lethal combinations to viable but polarity-relieved states, and polarity-relieved derivatives to those exhibiting full polarity. In the accompanying study (47), we suggested a model in which the polymeric H-NS scaffold is able to modulate Rho-dependent termination.

Some workers have speculated that Rho's essential function in E. coli may be different from its role in transcription termination (10, 54), and the fact that Cardinale et al. (9) could obtain viable insertion-deletion mutants of nusA or nusG, but not of rho, in the background of strain MDS42 may be seen as lending support to this notion. On the other hand, the existence of a spectral continuum as depicted in Fig. 7 suggests that inviability is merely an extreme consequence of decreasing efficiency of Rho-dependent termination. This is supported also by the observation that ΔnusA and ΔnusG derivatives of MDS42 exhibit hypersensitivity to the Rho inhibitor bicyclomycin (9, 54).