INTRODUCTION

Transcription termination factor Rho is a strongly conserved component in bacteria (1). To cause termination, it first binds to the nascent RNA and then uses the energy derived from ATP hydrolysis to mediate the dissociation of the RNA (2). The Rho polypeptide has two distinct domains, one for binding RNA, the other for binding ATP (3, 4). A comparison of the rho genes isolated from several different phylogenetic groups has revealed that the ATP binding domain is more highly conserved than the RNA binding domain (1).

The most radical divergence that has been discovered so far is the presence of a 150-250-residue extra sequence in the RNA binding domain of the Rhos in organisms from the high G + C Gram-positive group (5, 6). This group includes Micrococcus luteus, the Streptomycetes, and the Mycobacteria. Four Rho sequences from this group are known and they all have similar extra sequences. In M. luteus, the main extra sequence is between the phylogenetically conserved residues Ile⁴⁸ and Gly³¹², and it is unusual in its high proportion of Arg, Glu, and Asp residues and the very low abundance of hydrophobic residues. We refer to it as the arginine-rich extra sequence. Although the sequences of these extra residues in the four known examples are not highly conserved, their compositions and properties are very similar (6), suggesting that they have a similar function and that the extra sequences arose through adaptive evolution. Since the RNA transcripts in these organisms would have a high proportion of G residues, they are likely to have high levels of base-paired secondary structures. One possible function for the insert is to give Rho the ability to bind to and cause termination of transcripts that would have a high degree of secondary structure (5).

Rho factor from M. luteus was found to cause termination of transcription in a heterologous system with Escherichia coli RNA polymerase and a  DNA template (5). The M. luteus Rho differed from the E. coli Rho in being able to terminate transcription at far upstream sites on that template, at sites that were not normally used by E. coli Rho. Since E. coli Rho binds preferentially to RNA that is largely single-stranded and rich in unpaired cytosine residues (2) and since the  cro transcript that was available for interaction with Rho when RNA polymerase was at the upstream sites of  cro is predicted to have a substantial degree of base-paired secondary structure (7), these results suggest that M. luteus Rho was using these sites because it was capable of binding to highly structured transcripts.

In this report we test the role of the arginine-rich extra sequence in the RNA binding domain of M. luteus Rho on the specificity of the termination by comparing the properties of wild-type M. luteus Rho with those of a variant, called des(60-300) Rho, that lacks most of the extra sequence. We also test the role of the RNA secondary structure on the process by using ITP to replace GTP for the synthesis of RNA, which would have less stable base pairing. The results indicate that the extra sequence does have a major role in allowing the M. luteus Rho to bind to RNA with a relatively high degree of base-paired secondary structures.

EXPERIMENTAL PROCEDURES

E. coli Rho protein was provided by Lislott Richardson (Indiana University). M. luteus Rho was purified as described elsewhere (8). NusG was a gift from Barbara Stitt (Temple University). Bicyclomycin was obtained form Fujisawa Pharmaceutical Co. Ltd. (Osaka, Japan). E. coli RNA polymerase was purchased from Epicentre Technologies. Enzymes used in DNA manipulations were purchased from New England Biolabs. Radioactive nucleotides were from ICN Radiochemicals. Ribonucleotides, deoxynucleotides, and dideoxynucleotides were purchased from Pharmacia Biotech Inc. Polynucleotides were from Miles Laboratories.

Ligation of the 2021-bp¹ BamHI/SphI fragment (+113 to +2076) from pMLRHOSK+ (1, 9) into the approximately 4-kilobase BamHI/SphI of pBN10 (5) provided the entire M. luteus rho gene on plasmid pTB1 (1200 to +2076). pBN15, a plasmid containing a deletion of bp +1150 to +2076 of the 5 rho sequence in pMLRHOSK+ was prepared by ligation of the large EcoRI/BglII fragment of pMLRHOSK+ that had been blunt ended by treatment with Klenow DNA polymerase. pBN15 was then digested with BamHI/AluI releasing a 268-bp fragment. This fragment was cloned into the 2994-bp BamHI/RsrII fragment of pMLRHOSK+ creating the in-frame deletion of Arg⁶⁰ to Glu³⁰⁰ (rho bp +180 to +900) in plasmid pBN16.

The Arg⁶⁰ to Glu³⁰⁰ region (rho bp +180 to +900) in plasmid pBN16 was flanked by two unique restriction sites, SphI and NcoI. Digestion of pBN16 with these restriction endonucleases released a 183-bp fragment, which was swapped for the 906-bp SphI/NcoI fragment in pTB1, introducing the desired deletion into the M. luteus rho gene (pTB3). The approximately 2550-bp BamHI/EcoRI fragment from pTB3 was ligated in the overexpression vector pRSETa (Invitrogen) to create pBN18.

Primers WN7 (5-CATATGGCTAGCATGACCGAATCCACGGAACAG-3) and WN3 (5-GGGCGGTCTCGGCGGCGGGG-3) were used to amplify with a polymerase chain reaction a 247-bp DNA fragment in M. luteus rho (12 to +235). Primer WN7 includes the addition of NdeI and NheI sites at the 5 end of the M. luteus rho gene. Additionally, the M. luteus translation initiation codon at +1, GTG, was changed to ATG. Digestion of the polymerase chain reaction product with NheI and SphI yielded a 118-bp fragment, which was ligated into the 4130-bp NheI/SphI fragment of plasmid pBN18 to create pBN19. This resulted in the in-frame placement of M. luteus des(60-300) rho into plasmid pRSETa and added the primary amino acid sequence MRGSHHHHHHGMASM to the N terminus. Plasmid pBN21 was constructed by ligation of the 1398-bp NheI/EcoRI fragment from pBN19 into the 5330-bp NheI/EcoRI fragment of vector pET28a (Novagen) which added the primary amino acid sequence MGSSHHHHHHSSGLVPRGSHMASM to the N terminus. Verification of the sequence changes were confirmed by DNA sequencing (10).

Plasmid pBN21 containing the M. luteus rho gene with amino acid positions Arg⁶⁰-Glu³⁰⁰ deleted (numbering includes Met¹)² was transformed into the protease-deficient (ompT) E. coli strain BL21DE3[pLysS]. A colony of BL21DE3[pBN21; pLysS] was used to inoculate 150 ml of 2 × YT containing chloramphenicol (25 µg/ml) and kanamycin (50 µg/ml). The culture was grown to A₆₀₀ = 0.4. Overexpression was induced by the addition of IPTG to 1 mM. The culture was aerated an additional 3 h at 37 °C with moderate shaking. Cells were pelleted by centrifugation (10 min, 3000 × g), washed once in STE (100 mM NaCl, 10 mM Tris-HCl, pH 7.8, 1 mM EDTA), pelleted as above and stored at 20 °C for up to 2 months.

The cell pellet (1 g) was resuspended in 5 ml of grinding buffer (50 mM Tris-HCl, pH 7.8, 0.1 M NaCl, 5% glycerol, 2 mM EDTA, and 0.1 mM dithiothreitol) and blended at low speed with a Virtis homogenizer for 30 s. Phenylmethylsulfonyl fluoride (1 mM) and lysozyme (2 mg) were added, the cell mixture was again blended, and cell lysis was allowed to continue at room temperature for 15 min. Sodium deoxycholate was added to a final concentration of 0.05%, and the mixture was kept at room temperature an additional 10 min. After chilling the lysate to 4 °C, MgCl₂ and DNase were added to final concentrations of 40 mM and 10 µg/ml, respectively. The mixture was kept at 4 °C, for 30 min. The cell extract was diluted with 5 ml of grinding buffer, and the cell debris was pelleted by centrifugation (15 min, 12000 × g). The supernatant was centrifuged (30 min, 100,000 × g) in a Beckman 50Ti rotor.

The supernatant was collected (11 ml), and the conductivity was determined. This fraction was applied to a 10-ml (1 cm × 12.5 cm) column of Biorex-70 resin (Bio-Rad), which had been equilibrated in buffer TGED (20 mM Tris-HCl, pH 7.8, 10% glycerol, 1 mM EDTA, and 0.1 mM dithiothreitol) containing 0.1 M NaCl. Unbound protein was washed through the column with buffer TGED containing 0.1 M NaCl. M. luteus des(60-300) Rho protein was eluted with a 150-ml NaCl gradient (0.1 M to 1.5 M) in buffer TGED. Fractions containing poly(C)-dependent ATPase activity (5) were pooled and dialyzed overnight against buffer D (50 mM Tris-HCl, pH 7.8, 10% glycerol, 0.3 M NaCl).

The dialyzed pool was applied to a 1-ml (0.5 cm × 5 cm) Ni-nitrilotriacetic acid resin (Qiagen) column and washed with 10 column volumes of buffer D. The His-tagged protein was eluted from the resin in a 20-ml imidazole gradient (in buffer D) from 5 mM to 500 mM imidazole. Fractions containing poly(C) dependent ATPase activity were pooled, concentrated with a Centriprep-100 (Amicon), and dialyzed overnight against 1 liter of buffer F (50 mM Tris-HCl, pH 7.8, 0.15 M NaCl, 0.1 mM EDTA, 0.1 mM dithiothreitol, and 50% glycerol). The protein was stored at 20 °C.

cro RNA transcripts were synthesized in vitro with T7 RNA polymerase (11). Plasmid pIF2, which contains the cro gene with a modified promoter recognized by T7 RNA polymerase, was digested with either TaqI or HaeIII restriction endonuclease to use as a template to synthesize runoff transcripts of 378 or 115 nucleotides, respectively. 3 pmol of template were added to a 100-µl reaction containing reaction buffer (80 mM Tris acetate, pH 7.8, 20 mM Mg(OAc)₂, 1 mM spermidine, 10 mM dithiothreitol, 100 mg/ml acetylated bovine serum albumin, 2 mM each of UTP, CTP, GTP, and 0.125 mM [-³²P]ATP (350 nCi/pmol), and 40 units of RNasin (Promega)). To synthesize RNA molecules in which the guanosine residues were replaced with inosine residues, 2 mM ITP was substituted for 2 mM GTP in the polymerization reaction. The addition of 5 mM GMP was found to enhance the product yield. Synthesis was initiated by the addition of 200 units of T7 RNA polymerase, and the reaction was incubated for 2 h at 37 °C. RNA of the desired length was gel-purified on a 6% polyacrylamide gel containing 7 M urea after electrophoresis at 30 watts for 2 h. The RNA was electroeluted from the gel slice, precipitated with 0.1 volume of 5 M ammonium acetate and 3 volumes of EtOH, collected by centrifugation, resuspended in double-distilled H₂O, and stored at 80 °C. Typical yields were 20 pmol for guanosine-incorporated RNA or 100 fmol for inosine-substituted RNA.

³²P-Labeled RNA obtained by in vitro synthesis was diluted into 50 µl of transcription reaction mix (150 mM potassium glutamate, pH 7.8, 40 mM Tris acetate, pH 7.8, 4 mM Mg(OAc)₂, 1 mM dithiothreitol, 0.02% Nonidet P-40, 0.002% acetylated bovine serum albumin, 1% glycerol) to a final concentration of 0.2 nM. This mixture was added to various concentrations of Rho protein (final concentration from 0.05 to 100 nM hexamer) and incubated at 25 °C for 10 min. Reactions were filtered under vacuum on Biotrace NT filters (Gelman) which had been prewetted in binding buffer. The filters were washed with 6 volumes of wash buffer (binding buffer without the acetylated bovine serum albumin), dried, and assayed for radioactivity by scintillation counting. Sample values were adjusted for the fraction of actively binding protein which was determined by the method of Witherell and Uhlenbeck (12). Data were analyzed with the GraFit program v. 3.0 (Erithacus Software Ltd).

Reactions were performed with E. coli RNA polymerase-initiated cro DNA complexes as described previously (5). When inosine-substituted RNA was characterized, ITP was substituted for GTP in the reaction mixtures with the following modifications. ITP was used in the formation of the A²⁴ complex at 40 µM rather than 4 µM. Additionally, during elongation NTP concentrations were increased to 2.7 mM (ATP), 1.4 mM (UTP), 0.7 mM (CTP), and 1.1 mM (ITP). Rho concentrations were saturating (28 nM) under these conditions.

RESULTS

To address the potential role of the arginine-rich amino acid extra sequence in the M. luteus Rho protein, a mutant form of the factor, M. luteus des(60-300) Rho, which lacks most of that extra sequence was prepared and used for functional assays. To allow rapid purification and separation from endogenous E. coli Rho, the gene for expressing M. luteus des(60-300) Rho was engineered to encode an N-terminal hexahistidine sequence. After expression from a T7 RNA polymerase promoter by induction with IPTG, this variant of M. luteus Rho comprised greater than 40% of the total cellular protein (data not shown). Purification to homogeneity was achieved by cation exchange chromatography over Biorex-70 and metal ion affinity chromatography on a nickel-nitrilotriacetic acid-agarose. Table I shows the summary of the purification of 0.77 mg of M. luteus des(60-300) Rho protein from 1 g of induced cells.

Table I. Summary of purification of M. luteus des(60-300) Rho Step Protein Activity* Specific activity Yield mg units units/µg % Crude extract 23.0 198,000 8.5 100 Biorex-70 pool 3.8 36,800 9.7 19 Ni-NTA pool 1.5 30,000 20.2 15 Dialyzed protein 0.77 11,700 23.0 6 * One unit corresponds to the release of 1 nmol of Pi per min at 37 °C.

The purified des(60-300) Rho protein consisted of a single Coomassie Blue R250-stainable component which migrated at the position expected for a 51-kDa polypeptide (Fig. 1, lane 3), or slightly slower than the 47-kDa E. coli Rho polypeptide (Fig. 1, lane 2). As was found previously (5), the 76-kDa wild-type M. luteus Rho protein migrated anomalously at the position for a 95-kDa protein (Fig. 1, lane 4), presumably due to the arginine-rich composition of its extra sequence (13).

lane Ml

To determine if the 241-amino acid deletion from within the RNA binding domain of the M. luteus protein significantly altered the ATPase activity of M. luteus des(60-300) Rho, ATPase assays were performed using homopolymer RNA cofactors which had been used previously to characterize E. coli (14) and M. luteus Rho (5). The results (Table II) indicated that removal of the insert had no effect of the specific activity of M. luteus des(60-300) Rho compared to the wild-type protein when poly(C) was the cofactor. Furthermore, M. luteus des(60-300) Rho was activated to nearly the same extents as the wild-type protein by the other homopolymers tested. Finally, removal of the insert did not abolish the requirement for an RNA cofactor as no ATPase activity was detected in the absence of RNA.

Table II. Activation of Rho ATPase with different RNA homopolymers Rates were measured in standard assays using 25 ng of Rho protein, 20 µg/ml of the indicated polynucleotide, and 1 mM ATP and are presented per nmol of hexameric protein. Polynucleotide ATPase rate E. coli Rho M. luteus des (60-300) Rho M. luteus Rho µmol ATP hydrolyzed min1 nmol1 None <0.1 <0.1 <0.1 Poly(C) 11.2  ± 0.3 14.2  ± 0.3 14.5  ± 0.1 Poly(U) 2.2  ± 0.1 7.7  ± 0.4 6.2  ± 0.3 Poly(A) <0.1 2.4  ± 0.1 5.0  ± 0.1 Poly(I) 0.2  ± 0.1 1.1  ± 0.1 1.1  ± 0.2

M. luteus des(60-300) Rho was indistinguishable from the wild-type M. luteus Rho with respect to its ability to hydrolyze the three other NTPs and dATP and its sensitivity to the Rho-specific inhibitor, bicyclomycin (data not shown).

To determine what effect the alteration of the RNA binding domain had on the ability of the protein to terminate transcription, we tested its function in vitro with E. coli RNA polymerase transcribing a  cro DNA fragment, which has the Rho-dependent terminator, tR1 (15). Strong E. coli Rho termination stop points occur at sites I (296 nucleotides), II (318 nucleotides), and III (351 nucleotides) within the tR1 terminator. When a saturating amount of M. luteus des(60-300) Rho was added to the transcription mixture, RNA transcripts were synthesized that ended at several positions, but primarily at nt 240 and nt 260 (Fig. 2a, lane 4). These preferred end point positions were very different from those for the transcripts made when wild-type M. luteus Rho was present (lane 6). This result shows that the des(60-300) Rho has termination activity but with a different specificity than that of the full-length M. luteus Rho. Its specificity also differed from that of E. coli Rho (Fig. 2a, lane 2) but was very similar to that of E. coli Rho with NusG present (lane 3). Although M. luteus des(60-300) Rho caused termination primarily at more upstream sites, it was able to recognize signals within tR1, causing RNA polymerase molecules that had passed through those upstream sites to terminate transcription at sites I, II, and III (Fig. 2, lane 4). We do not know whether the observation of the close similarity of the termination specificity of M. luteus des(60-300) Rho and E. coli Rho with NusG present is merely a coincidence or is an indication that M. luteus Rho normally circumvents the requirement for NusG. E. coli NusG had very little effect on the termination specificity with M. luteus des(60-300) Rho or with M. luteus Rho (Fig. 2a, lanes 5 and 7). We did not have M. luteus NusG available to test its function. However, in further experiments (16) termination by M. luteus Rhos could be enhanced by E. coli NusG, but not nearly to the same extent as E. coli Rho.

lanes Ml

We verified that the shorter transcripts synthesized in the presence of the Rho factors were not produced by RNase degradation by doing control experiments in which the factors were added to reaction mixtures which had synthesized full-length runoff transcripts prior to adding the factors (data not shown). Titrations of the individual Rho factors revealed that the amount of Rho that was half-maximal for termination activity was approximately 2 nM for each of the proteins (data not shown).

The ability of these three Rho factors to terminate transcription was also assessed for tiZ1, the first intragenic terminator in lacZ (17) and the results closely paralleled those for the  tR1 terminator; M. luteus Rho caused transcripts to terminate at points well upstream from the first stop points used by E. coli Rho, and the M. luteus des(60-300) Rho terminated transcription at points just slightly upstream from the E. coli stop points (data not shown). This indicates that the extra sequence in M. luteus Rho has a general effect on the specificity of termination.

To determine whether the differences in the specificities of the three Rho factors is dependent on the degree of stability of the base-paired secondary structures in the nascent RNAs, we tested their function on the transcription of  cro DNA in a reaction mixture containing ITP in place of GTP. ITP can replace GTP as a substrate for E. coli RNA polymerase (18) and because inosine forms only two hydrogen bonds with cytosine, the base-paired secondary structures in inosine-substituted RNA will be less stable than the corresponding structures in RNA with guanosine residues (18, 19). The effect of the substitution of GTP with ITP on Rho action was first recognized by Adhya et al. (19), and Morgan et al. (20, 21) showed that the substitution caused E. coli Rho to terminate transcription at new sites well upstream from the normal sites on  cro DNA.

Since the efficiency of transcription termination at a particular site is very sensitive to the RNA chain elongation rate (22), and since that rate is about 5-fold lower in reaction mixtures containing ITP than in reaction mixtures containing GTP at the same concentration (16), we increased the levels of NTP concentrations in the reaction mixtures for our termination studies to a level that allowed the RNA chains to grow at the same average rate as in the standard reaction mixtures with GTP. With these conditions E. coli Rho terminated transcription at more upstream sites than were used during synthesis of guanosine-containing RNAs (compare Fig. 2, panel b, lane 1, with panel a, lane 2). The sizes of these transcripts, as estimated by gel migration, were similar to those reported by Morgan et al. (20). Likewise, smaller, novel transcripts were produced when M. luteus des(60-300) Rho was present for the synthesis of inosine-containing RNA than for the synthesis of guanosine-containing RNA (Fig. 2, panel b, lane 2). The termination pattern was in fact very similar to that produced by E. coli Rho (compare Fig. 2, panel b, lanes 1 and 2), although there was a slightly greater preference for terminating at earlier points. The wild-type M. luteus protein was also functional on inosine-substituted RNA and the overall preference of termination stop points with it was similar to that produced with the guanosine-containing RNA at nucleotide positions 96, 120, 145, 170, and 240 (compare Fig. 2, panel a, lane 6, with panel b, lane 3, which has a darker exposure). A control reaction in which no Rho factor was added indicated that most of the polymerase molecules were able to read through the template to produce the run-off cro transcript (Fig. 2, panel b, lane 4).

The effect of ITP substitution was also assessed with the tiZ1 terminator in lacZ. The results (data not shown) closely paralleled those for the cro template. When transcription was carried out with ITP in place of GTP, termination was shifted to more upstream positions with des(60-300) Rho and E. coli Rho but not with M. luteus Rho.

To determine if the inability of M. luteus des(60-300) Rho to terminate at the upstream sites used by the wild-type M. luteus protein during transcription with GTP was due to a lack of protein-RNA interaction in the highly structured 5 end of the RNA molecule, RNA binding studies with  cro RNAs of various lengths and compositions were performed. Using the same procedures as described by Faus and Richardson (11), we confirmed that the equilibrium dissociation constant for the complexes of E. coli Rho with the full-length, 378-nucleotide  cro RNA was approximately 1 nM (Table III). With M. luteus Rho and M. luteus des(60-300) Rhos, the K_d values for their complexes with the same RNA were both about 10-fold lower (Table III). When the RNA was a partial transcript of the cro gene of only 115 nucleotides, which is about the size of the shorter transcripts produced by action of M. luteus Rho but shorter than the transcript terminated by the other Rhos, M. luteus Rho bound to it with the same affinity as to the full-length cro transcript, while M. luteus des(60-300) Rho bound to it with nearly 20-fold lower affinity, and E. coli Rho gave barely detectable binding (Table III).

Table III. Kd values for Rho-cro RNA binary complexes The RNAs are indicated by their length in nucleotides (nt) and whether they contain GMP or IMP residues. The method of Witherell and Uhlenbeck (12) was used to evaluate the fraction of protein that was active in RNA binding. The values determined were: E. coli Rho, 100%; M. luteus des(60-300) Rho, 20%; and wild-type M. luteus Rho, 70%. Rho Kd of cro RNA 378 nt (GMP) 115 nt (GMP) 115 nt (IMP) nM E. coli 1.3  ± 0.3 >50 0.3  ± 0.05 M. luteus des(60-300) 0.2  ± 0.1 4.4  ± 0.5 0.2  ± 0.05 M. luteus 0.1  ± 0.05 0.1  ± 0.05 0.1  ± 0.05

However, when a 115-nt partial cro transcript which has been synthesized using ITP instead of GTP was used, M. luteus Rho again bound to it with the same affinity as to the guanosine-containing transcript, while M. luteus des(60-300) Rho and E. coli Rho were able to bind to it almost as well as the M. luteus Rho (Table III). These results strongly suggest that higher ordered nucleic acid structure near the 5 end of the cro RNA is inhibitory to binding of Rho factors lacking the arginine-rich insertion within the RNA-binding domain.

DISCUSSION

We have shown that the arginine-rich subdomain of M. luteus Rho confers on it the ability to bind RNA molecules that do not interact well with two Rho proteins which lack that subdomain, E. coli and M. luteus des(60-300) Rhos, and that this difference in RNA binding properties accounts for the altered transcription termination specificity of M. luteus Rho. We then showed that the differences in RNA binding and termination selectivity are largely eliminated when the RNA transcripts contain inosine residues instead of guanosine residues. Together, these results support the contention that the subdomain in M. luteus Rho gives it the ability to terminate transcription of its G-rich transcripts.

To date the rho genes from 15 organisms from several subgroups of the bacteria have been sequenced. These genes contain a number of highly conserved sequence motifs. The arginine-rich subdomain of M. luteus Rho is an unusual, extended sequence in an unconserved segment between two well conserved sequences in the RNA binding domain. So far, similar extended insert-like sequences have been found only in the proteins encoded by genes from organisms in the same phylogenetic group that includes M. luteus, the high G + C Gram-positive group. Sequences for four members of that group have been reported and all have similar extra sequences. They are similar in having greater than 150 residues, in being rich in Arg residues with balancing Glu and Asp residues, in being poor in hydrophobic residues, and in being in the same position, just preceding the highly conserved RNP1-like sequence. However, the four sequences show very little conservation among themselves with the greatest similarity in the Arg- and Gly-rich C-terminal parts (6). In spite of the lack of sequence conservation, the presence of this similar extra subdomain in all of the known putative Rho factors in members of this phylogenetic group suggests that it arose as an evolutionary adaptation.

This extra sequence is apparently not a characteristic of all the Gram-positive bacteria nor of all organisms with a high G + C content, because it is not present in the putative Rho homologs from the low G + C Gram-positive bacterium, Bacillus subtilis (23), or from Rhodobacter sphaeroides (24), a Gram-negative organism with a G + C content of 69%. Rather, the extra sequence appears to have evolved along the lineage of the high G + C Gram-positive bacteria.

The abundance of arginine residues in the extra sequence of some Rho factors make those factors members of the arginine-rich motif family of RNA-binding protein. Even though these extra sequences are also rich in Gly residues, they do not have many multiples of the RGG motif, thus they are not members of the family with that feature. Since the Rho factors with the extra sequence also contain all of the highly conserved RNP1-like sequence motifs found in all Rho factors, they are also members of the RNP motif family (25). The finding that both forms of M. luteus Rho could bind to the full-length  cro RNA with a higher affinity than E. coli Rho could indicate that an important part of the difference in termination selectivity between M. luteus and E. coli Rho resides in the sequences that were still present in the des(60-300) derivative, that is within the RNP1-like sequences. The residual difference in binding affinity is probably the reason for the ability of M. luteus des(60-300) Rho to terminate transcription at sites upstream from those used by E. coli Rho in the absence of NusG.

Arginine residues in RNA-binding proteins are known to interact with specific RNA sequence elements (26) and nonspecifically with the RNA phosphate backbone (27). Although the arginines within the subdomain could be providing either or both of these interactions to the RNA binding properties of M. luteus Rho, we suggest they could have another possible function. The lack of hydrophobic residues, which are important organizing elements for -helical and -strand secondary structures in globular proteins, indicates that the subdomain does not have an ordinary globular structure but instead has a loosely ordered structure of largely solvent accessible residues stabilized by salt bridges. Thus, this subdomain could be providing a local solvent environment that could affect the properties of the RNA as the protein is making initial contacts with it through the conserved elements of the RNA binding domain. Taking into consideration the partial specific volumes of the residues included in the extra sequence, we estimate that the concentration of guanido groups from the arginine residues in this 3.4 × 10²⁰ cm³ subdomain to be approximately 2.8 M. At this concentration these guanido groups could be acting as a localized general denaturant causing the disruption of base-paired RNA secondary structure. The disruption of base-paired structures could then allow M. luteus Rho to make strong functional contacts with single-stranded segments of the RNA using the conserved elements of the RNA binding domain. Studies on how the subdomain affects the structure of the bound RNA and on how specific Arg residues affect the binding specificity should be able to distinguish whether the subdomain is merely serving to alter the local solvent properties or is contributing more directly to the binding process through side chain-nucleotide interactions.