|Home | About | Journals | Submit | Contact Us | Français|
A strain of Bacillus subtilis lacking two 3′-to-5′ exoribonucleases, polynucleotide phosphorylase (PNPase) and RNase R, was used to purify another 3′-to-5′ exoribonuclease, which is encoded by the yhaM gene. YhaM was active in the presence of Mn2+ (or Co2+), was inactive in the presence of Mg2+, and could also degrade single-stranded DNA. The half-life of bulk mRNA in a mutant lacking PNPase, RNase R, and YhaM was not significantly different from that of the wild type, suggesting the existence of additional activities that can participate in mRNA turnover. Sequence homologues of YhaM were found only in gram-positive organisms. The Staphylococcus aureus homologue, CBF1, which had been characterized as a double-stranded DNA binding protein involved in plasmid replication, was also shown to be an Mn2+-dependent exoribonuclease. YhaM protein has a C-terminal “HD domain,” found in metal-dependent phosphohydrolases. By structure modeling, it was shown that YhaM also contains an N-terminal “OB-fold,” present in many oligosaccharide- and oligonucleotide-binding proteins. The combination of these two domains is unique. Thus, YhaM and 10 related proteins from gram-positive organisms constitute a new exonuclease family.
Degradation of mRNA in bacteria is an important element in the control of gene expression and is an essential function. According to current models of the process of mRNA decay in Escherichia coli, decay can initiate with cleavage by the 5′-end-dependent endoribonuclease RNase E (17, 41). The upstream fragment of such cleavage is subject to 3′-to-5′ exonucleolytic degradation by the two major 3′-to-5′ exoribonucleases, RNase II and polynucleotide phosphorylase (PNPase). Subsequent RNase E cleavages yield more RNA fragments that can be degraded in the 3′-to-5′ direction. Alternatively, an mRNA can be degraded from the otherwise resistant 3′ end by the action of 3′-to-5′ exonucleases working in combination with poly(A) polymerase and RNA helicase (18, 41). The final turnover of oligonucleotide decay products to mononucleotides is achieved by oligoribonuclease (27). Inactivation of both RNase II and PNPase results in cell death (23). Similarly, the orn gene, encoding oligoribonuclease, is essential (27).
We have been studying mRNA decay in Bacillus subtilis. In contrast to E. coli, where RNase II is the major 3′-to-5′ exoribonuclease activity in cell extracts, in B. subtilis cell extracts PNPase is the major 3′-to-5′ exoribonuclease (22). Other differences between RNA processing enzymes in B. subtilis and E. coli are known from the genome sequences: there are no sequence homologues in B. subtilis for genes encoding oligoribonuclease, RNase II, or RNase E (although there is indirect evidence for an RNase E-like activity in B. subtilis ). Thus, the RNases that are involved in regulating mRNA half-life may be very different in E. coli and B. subtilis.
Despite the major role that PNPase appears to have in B. subtilis mRNA decay, we were able to construct a strain with a disruption of pnpA, the gene encoding PNPase (48). The pnpA strain grows more slowly than the wild type and has several other interesting phenotypes: competence deficiency, temperature sensitivity (35), altered morphology, tetracycline sensitivity (48), and accumulation of 5′-proximal mRNA decay fragments (9). Extracts from the pnpA strain were used to identify a second 3′-to-5′ exoribonuclease in B. subtilis (40), which we called RNase R, based on its similarity to RNase R of E. coli (16). B. subtilis RNase R is encoded by the rnr gene (previously designated yvaJ). Although no phenotype is detectable with an rnr disruption strain, a pnpA rnr double mutant grows even more slowly than the pnpA single mutant. Preliminary studies following decay of a specific mRNA revealed no differences in mRNA processing between the pnpA and pnpA rnr strains. Presumably, one or more as yet unidentified exoribonuclease activities are present in B. subtilis that can compensate for the lack of PNPase and RNase R. In this study, the pnpA rnr strain was used as the starting material for purification of yet another 3′-to-5′ exoribonuclease activity.
The wild-type B. subtilis host was BG1, which is trpC2 thr-5. RNase mutant strains were constructed by transformation with chromosomal DNA isolated from the chloramphenicol-resistant pnpA mutant strain BG116 (48) and the spectinomycin-resistant rnr mutant strain BG295 (40). To disrupt the yhaM gene, an internal BsaBI-ClaI fragment (nucleotides [nt] 530 to 704 of the yhaM coding sequence) was replaced with a PCR-generated PvuII-BstBI fragment containing the pUB110 phleomycin resistance gene (44). This DNA was used to transform BG1 to phleomycin resistance (1 μg/ml). B. subtilis growth media and competent B. subtilis cultures were prepared as described previously (24). E. coli strain DH5α (29) was the host for plasmid constructions.
To measure growth rates of B. subtilis strains, overnight cultures that were grown in Luria-Bertani (LB) broth without selection were spun down, washed, and diluted 1:100 in fresh LB broth. Growth at 37°C was monitored by measuring absorbance at 600 nm. Suspensions of overnight cultures were appropriately diluted in LB broth for plating on LB agar in order to assay colony formation at room temperature.
Uniformly labeled RNA substrates for the RNase assay were synthesized by T7 RNA polymerase transcription (MAXIscript T7 kit; Ambion) in the presence of [α-32P]UTP. The template was a gel-purified DNA fragment that was generated by PCR amplification of the 187-nt sequence that starts at the Bs-RNase III cleavage site in EG242 and extends to the HindIII site (40). The upstream primer in the PCRs contained the 17-nt T7 RNA polymerase promoter sequence at its 5′ end. The products were gel purified as described previously (1). In some experiments, RNA was 5′-end labeled with T4 polynucleotide kinase and [γ-32P]ATP. To assay hydrolysis of single-stranded DNA, oligonucleotides were 5′-end labeled. To assay hydrolysis of double-stranded DNA, 5′-end-labeled TaqI DNA fragments from plasmid pSE420 (13) were used.
The standard RNase reaction mixture contained 50 mM Na-Tricine (pH 8.0), 100 mM KCl, 1 mM MnCl2, and 0.2 to 0.5 pmol (2 × 105 to 4 × 105 cpm/pmol) of labeled substrate. (Tricine was found to be the most effective buffering agent for YhaM activity; YhaM was less active in Tris and morpholinepropanesulfonic acid [MOPS] buffers and was completely inactive in HEPES buffer.) The reaction mixture (100 μl) was incubated for 20 min at 37°C and then extracted with an equal volume of phenol-chloroform. A 10-μl volume of the aqueous phase was mixed with gel-loading buffer (Ambion) and electrophoresed on a 20% denaturing polyacrylamide gel to resolve mononucleotides. The amount of mononucleotide product was quantitated on a PhosphorImager instrument (Molecular Dynamics) or, for more precise quantitation, by excising the radioactive spot and determining radioactive counts. To observe the degradation of larger RNA fragments, 50 μl of the phenol-chloroform-extracted aqueous phase was ethanol precipitated, washed with 75% ethanol, dried, dissolved in gel-loading buffer, and separated by electrophoresis on a 6% denaturing polyacrylamide gel.
The protocol for assaying bulk mRNA decay in vivo was essentially as described previously (48). Decay of 3H-pulse-labeled RNA was followed by measuring trichloroacetic acid (TCA)-precipitable counts remaining at times after the stopping of transcription by addition of actinomycin D to exponential-growth-phase cultures. Incorporation into stable RNA was determined by measuring TCA-precipitable counts remaining 60 min after inhibition of transcription, and these counts were subtracted from each time point. In addition, counts from nonspecific filter binding were subtracted from each time point. Bulk mRNA half-lives were determined by linear regression analysis on semilogarithmic plots of percent RNA remaining versus time. The values in the text are times for 50% RNA remaining.
YhaM was purified as follows. The pnpA rnr strain was grown in YT medium (8.0 g of tryptone/liter, 8.0 g of yeast extract/liter, 2.5 g of NaCl/liter) until the end of the exponential-growth phase. Twenty-five grams of cells (wet weight) was washed in buffer A (20 mM Tris-HCl [pH 7.75], 100 mM KCl, 0.2 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) and resuspended in buffer A at 0.1 g of cells per ml. Cells were lysed by incubation with 0.2 mg of lysozyme/ml for 25 min at 37°C, followed by disruption in a French pressure cell. All subsequent procedures were performed at 4°C. Cell debris was cleared by centrifugation at 10,000 × g for 15 min. The supernatant was centrifuged at 200,000 × g for 2 h. Nucleic acids were precipitated from the resulting supernatant (containing ~600 mg of protein) by slow addition of 15 ml of a 0.2-g/ml solution of streptomycin sulfate and were removed by centrifugation at 27,000 × g for 15 min. The Mn2+-dependent ribonucleolytic activity was precipitated with 60% ammonium sulfate, collected by centrifugation at 27,000 × g for 15 min, and resuspended in 50 ml of buffer A containing 10% glycerol. After overnight dialysis against “starting buffer” (buffer A containing 150 mM KCl and 10% glycerol), proteins (~190 mg) were passed through DEAE-Sepharose CL-6B (Sigma). Unbound protein (~43 mg) was collected and fractionated on Affi-Gel Blue (Bio-Rad). The matrix was washed extensively with starting buffer and then with starting buffer containing 400 mM KCl. The Mn2+-dependent activity was eluted with 1 M KCl. The eluate was dialyzed against 400 mM KCl buffer, concentrated to approximately 55 ml by using a Centriprep YM-30 filter (Millipore), and dialyzed overnight against 50 mM KCl buffer. Part of the recovered sample (3.7 of 4.4 mg) was fractionated by fast protein liquid chromatography (FPLC) (LKB Pharmacia LCC-500 Plus) on a Mono Q column that was equilibrated to 20 mM Tris-HCl (pH 7.75)-50 mM KCl. The column was developed with a two-step linear gradient (0 to 10% in 5 ml and 10 to 30% in 20 ml of 20 mM Tris-HCl [pH 7.75]-1 M KCl). Fractions (1 ml) were collected and assayed for RNase activity. The three most active fractions were eluted at approximately 200 mM KCl and together contained about 400 μg of protein.
After concentration by precipitation with acetone, half of the sample was resolved by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis (SDS-10% PAGE). Proteins were transferred to a polyvinylidene fluoride (PVDF) membrane (Bio-Rad) in 50 mM Tris-50 mM borate buffer for 18 h at 12 V. The membrane was stained with 0.2% amido black and destained in water. Individual protein bands were cut out of the membrane and eluted in 50 mM Tris-HCl (pH 8.7)-1% Triton X-100, as described previously (21), and assayed for RNase activity. The activity of the purified YhaM protein was stable when it was stored frozen at −80°C in a buffer containing 25% glycerol or at −20°C in a buffer containing 50% glycerol. For protein sequence analysis, 15 μg of protein from the most active Mono Q fraction was resolved on an SDS-10% PAGE gel and stained with Coomassie blue. The 39-kDa protein band, containing approximately 1 μg of protein, was excised and submitted to the Columbia University Protein Chemistry Core Facility. The protein was reduced, alkylated with iodoacetamide, and digested with trypsin, and the masses of the resulting peptide fragments were analyzed by matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS).
His-tagged versions of B. subtilis YhaM and Staphylococcus aureus CBF1 were purified on a Ni2+-nitrilotriacetic acid (NTA) column under denaturing conditions, as described in the Qiagen overexpression protocol. The yhaM and cbf1 coding sequences were amplified by PCR, by using as a template genomic DNA from B. subtilis strain BG1 and S. aureus strain RN4220 (34), respectively, and cloned between the NcoI and BglII sites of plasmid pQE60 (Qiagen). In both cases, the cloning resulted in two additional residues at the C terminus, followed by six histidine residues. In addition, the second residue in CBF1 was changed from Arg to Gly. These constructs were verified by sequence analysis, performed by the Mount Sinai Department of Human Genetics DNA sequencing facility. For further purification of the His-tagged proteins (as shown in Fig. Fig.5A),5A), the proteins were resolved on an SDS-10% PAGE gel and blotted onto a PVDF membrane. Elution from the membrane was done either under nonreducing conditions, as described above, or under reducing conditions created by addition of β-mercaptoethanol.
In previous work, a pnpA rnr double mutant was constructed (40). By use of a uniformly labeled, 187-nt RNA substrate (see Materials and Methods), an Mn2+-dependent exoribonuclease activity was detectable in extracts prepared from the pnpA rnr strain. Protein extracts of this strain were fractionated by ammonium sulfate precipitation and DEAE-Sepharose and Affi-Gel Blue column chromatography, as described in Materials and Methods. The most active fraction was further purified on a Mono Q column, resulting in a preparation with several prominent proteins (Fig. (Fig.1A,1A, lane 1). A gel in which this preparation was resolved was blotted onto a PVDF membrane, and prominent protein bands were eluted and tested for exoribonuclease activity. The band that migrated at about 39 kDa (Fig. (Fig.1A,1A, lane 2) was the only one that contained RNase activity. More of this protein was obtained by preparative gel electrophoresis, and the amino acid sequence was determined by mass spectrometry analysis (see Materials and Methods). This analysis indicated that a single species was the major protein, with only slight traces of other proteins. The sequence of the major protein matched that encoded by the B. subtilis yhaM gene, the function of whose gene product was previously unknown. The predicted molecular size of the 314-amino-acid YhaM protein is 35.5 kDa, but the protein displays a somewhat higher molecular size of about 39 kDa on SDS-PAGE gels.
To verify that the yhaM gene was specifying the observed RNase activity, the yhaM coding sequence was cloned on a His tag expression vector into E. coli (see Materials and Methods). Expression of His-tagged YhaM was induced, and the protein was purified and assayed (Fig. (Fig.1B).1B). The results show an Mn2+-dependent exonuclease activity that is specifically induced in the strain expressing the His-tagged YhaM protein.
From many experiments similar to the one for which results are shown in Fig. Fig.1B,1B, where mononucleotides are the only small products detected after treatment with YhaM, we infer that the activity is exoribonucleolytic. To differentiate between 5′-to-3′ versus 3′-to-5′ exonucleolytic activity, the sensitivities of RNA substrates with different secondary structures were determined. Figure Figure2A2A is a diagram of the 187-nt substrate that was used to assay for exonuclease activity. While this substrate contains an 80-nt 5′-terminal sequence and a 50-nt 3′-terminal sequence that are not predicted to form any secondary structure that would significantly hinder exonuclease digestion, there is an internal sequence that is predicted to form a strong secondary structure (predicted free energy, −20.2 kcal/mol), as shown. When the products of YhaM degradation are separated on a 20% denaturing polyacrylamide gel (e.g., Fig. Fig.1B),1B), only the mononucleotide product is seen clearly, and the degraded large products migrate as a dark, broad band near the top of the gel. However, when the 187-nt RNA was treated with YhaM and the products were separated on a 6% polyacrylamide gel, a degradation intermediate was seen (Fig. (Fig.2B,2B, arrow). Importantly, when a 5′-end-labeled 187-nt substrate was used, the identical degradation intermediate was seen (data not shown). The size of this intermediate is consistent with a block to 3′-to-5′ exoribonuclease activity at the base of the strong internal secondary structure. When a shorter substrate, which terminated at the top of the loop structure shown in Fig. Fig.2A,2A, was used in the RNase assay, no decay intermediates were detected (Fig. (Fig.2B,2B, 110-nt substrate). Thus, we conclude that YhaM specifies a 3′-to-5′ exoribonuclease activity.
Several biochemical parameters of YhaM activity were determined in vitro by using the YhaM protein purified from B. subtilis. YhaM was active in the presence of Mn2+ or Co2+ but was inactive in the presence of Mg2+ (Fig. (Fig.3A).3A). No activity was observed in the presence of Mn2+ at a concentration of <10 μM. YhaM did not require monovalent cations for activity, but its activity was stimulated in the presence of monovalent cations (Na+ or K+) at 50 to 100 mM. Activity was inhibited at monovalent cation concentrations of >200 mM (<10% activity relative to that with 50 mM). YhaM was active in a broad pH range; we observed comparable activity from pH 6 to pH 9.
The substrate specificity of YhaM was tested by using 5′-end-labeled DNA. One example of such an experiment is shown in Fig. Fig.3B,3B, in which a 38-nt DNA oligonucleotide was the substrate. The results showed that YhaM was capable of degrading single-stranded DNA to mononucleotides in the presence of Mn2+. Interestingly, in the presence of Co2+, the final degradation product was larger than a mononucleotide, which was not the case with RNA substrates (Fig. (Fig.1B1B and and3A).3A). Since the DNA substrates were 5′-end labeled, this result also showed that the single-stranded DNA exonuclease activity is 3′-to-5′. To test for activity on double-stranded DNA, DNA fragments generated by TaqI digestion of plasmid pSE420 (13) were 5′-end labeled. These DNA fragments were not degraded by YhaM in the presence of either Mn2+, Co2+, or Mg2+ (data not shown).
To test the sensitivity of 3′ DNA overhangs and the resistance of 5′ DNA overhangs to attack by YhaM, plasmid pGEM-3Zf(+) (Promega) was digested with restriction endonuclease EcoRI (which leaves a 5′ overhang) or SacI (which leaves a 3′ overhang). The cleaved DNA was treated with YhaM for 1 h at 37°C in the presence of Mn2+ and was then ligated and used to transform E. coli DH5α, with plating for blue/white screening. For the EcoRI-treated DNA (5′ overhang), 0.2% of the colonies recovered were white, whereas for the SacI-treated DNA (3′ overhang), 32.4% of the colonies recovered were white. Four of the white colonies derived from the SacI-treated DNA were sequenced. In each case the 4-nt overhang was precisely deleted due to the treatment with YhaM. These results indicate that YhaM activity is specific for single-stranded DNA in the 3′-to-5′ direction.
Initial attempts were made to understand the function of YhaM in vivo. An internal portion of the yhaM coding sequence was replaced with a phleomycin resistance gene from plasmid pUB110 (see Materials and Methods), resulting in the loss of 58 out of 314 codons. The YhaM knockout, which was easily obtained, was confirmed by PCR. No obvious mutant phenotype was observed; the growth rate at 30 or 37°C in a rich or defined medium was the same as that of the wild type. Double and triple RNase mutants were then made, using the previously constructed pnpA and rnr knockout strains, and the doubling times were measured in LB medium at 37°C (Table (Table1).1). Among the single RNase knockout strains, only the pnpA strain showed a significantly longer doubling time (31.5 min) than the wild type (24.5 min). The absence of yhaM did not significantly affect the doubling times when combined either with the pnpA mutation alone (doubling times, 31.5 min for the pnpA strain and 31.0 min for the pnpA yhaM strain) or with both the pnpA and rnr mutations (doubling times, 41.0 min for the pnpA rnr strain and 42.0 min for the pnpA rnr yhaM strain). These results suggest that yhaM is not required for a wild-type growth rate during the exponential-growth phase. However, it was observed reproducibly that the lag phase for the pnpA rnr yhaM triple mutant was extended by about 40 min relative to the lag phase for the pnpA rnr double mutant (Fig. (Fig.44).
Since a cold-sensitive phenotype has been observed with the pnpA deletion mutant (9), the growth of yhaM mutant strains at a lower temperature was tested by plating them on LB agar and growing them at room temperature. The results demonstrated that the absence of yhaM in the pnpA rnr background had a profound effect on growth at a low temperature: After 72 h at room temperature, colonies from the pnpA rnr strain were visible, but no colonies were seen for the pnpA rnr yhaM strain. After 96 h (Table (Table1),1), colonies were barely visible on the plates containing the pnpA rnr yhaM triple mutant, whereas the pnpA rnr double mutant gave 0.5-mm-diameter colonies at this time.
To determine whether the degradation of mRNA was affected in RNase mutant strains, bulk mRNA decay was measured (see Materials and Methods). Average bulk mRNA half-lives (from four experiments with duplicate samples in each) were 3.52 ± 0.18 min for the wild type, 3.87 ± 0.26 min for the yhaM mutant, 3.30 ± 0.49 min for the pnpA rnr mutant, and 3.64 ± 0.31 min for the pnpA rnr yhaM mutant. These data show that no significant differences in bulk mRNA half-life were observed between the wild type and these three mutant strains. Thus, PNPase, RNase R, and YhaM are not required to maintain a wild-type rate of general mRNA decay, suggesting the existence of other activities that either are primary or can compensate for the absence of these three activities.
A close homologue of yhaM was previously identified in S. aureus and was named cbf1 because it encodes cmp-binding factor 1 (52). CBF1 protein (313 amino acids) was shown to bind specifically to cmp, the replication enhancer of plasmid pT181 (26). A BLAST search (2) of the protein databases, using the entire YhaM amino acid sequence, revealed a number of gram-positive organisms containing proteins identified as “similar to S. aureus CMP-binding protein” which were at least 40% identical to YhaM. To test whether the CMP-binding proteins had RNase activity, the S. aureus cbf1 coding sequence was expressed from a His tag vector in E. coli, and CBF1 protein with a C-terminal His tag was purified. It was observed previously that overexpression of cbf1 and purification by DNA affinity chromatography led to the isolation of full-length CBF1 protein as well as a lower-molecular-weight form that apparently was the result of translation from an internal start codon (52). Similarly, we found that the His tag protein purification protocol resulted in one major protein for YhaM but two proteins for CBF1 (Fig. (Fig.5A,5A, lanes 1 and 4). The band migrating between the two marked CBF1 proteins probably represents an unrelated protein that copurifies with CBF1. Although the predicted molecular size of the His-tagged CBF1 protein is 36.7 kDa, it displays a mobility of around 41 kDa on an SDS-PAGE gel.
The upper and lower CBF1 bands were blotted onto a PVDF membrane, isolated separately (shown in Fig. Fig.5A,5A, lanes 5 and 7), and then tested for RNase activity by using the 110-nt RNA substrate described above. The full-length CBF1 protein gave weak exoribonuclease activity in the presence of Mn2+ (Fig. (Fig.5B,5B, lane 6), but the lower-molecular-weight protein had no detectable RNase activity (data not shown).
There are no cysteine residues in YhaM, but there are two cysteine residues in CBF1, which are located in the C-terminal half of the protein. It was hypothesized that the conditions for CBF1 isolation were such that oxidation of the cysteine residues could occur, resulting in poor RNase activity. Therefore, the YhaM and CBF1 proteins were isolated again from PVDF membranes, but under reducing conditions (see Materials and Methods). The isolated proteins are shown in Fig. Fig.5A,5A, lanes 6 and 8. Interestingly, a single major band was obtained when the higher- and lower-molecular-weight forms of CBF1 were isolated under reducing conditions, whereas a smaller protein band was present when the forms of CBF1 were isolated under nonreducing conditions (Fig. (Fig.5A,5A, lanes 5 and 7). We suppose that these additional proteins are oxidized forms of CBF1.
CBF1 protein that was isolated under reducing conditions gave a more robust RNase activity (Fig. (Fig.5B;5B; compare lanes 6 and 8). There was no difference between the activities of YhaM isolated under nonreducing versus reducing conditions (Fig. (Fig.5B,5B, lanes 2 and 4). We conclude that CBF1 protein, which was formerly characterized as a DNA-binding protein, is similar to YhaM in that it is a 3′-to-5′ exoribonuclease which is Mn2+ dependent in vitro.
Simple sequence similarity searches using BLAST (2) against protein sequence databases identified only one region of the YhaM sequence as having similarity to other proteins of known function, namely, the HD domain (4) extending from residues 160 to 279. The HD domain is the defining feature of a family of metal-dependent phosphohydrolases, so named because of the conserved His and Asp residues that are predicted to be involved in catalysis (4). In order to further characterize the YhaM sequence, profile searches were attempted using PSI-BLAST (3) and HMMER (25). PSI-BLAST searches against the SWISS-PROT and TrEMBL databases (7) produced no significant hits for segments outside the HD domain. A HMMER search against the PFAM collection (8) of Hidden Markov Models (HMM) (25) for protein families resulted in a borderline significant hit with the “tRNA_anti” family of PFAM, which corresponds to OB-fold nucleic acid binding domains (38). The OB-fold hit covered the region between residues 17 and 90 of YhaM, with an E-value of 0.037. Since this E-value was not statistically significant, the fold assignment was confirmed by structural modeling and model evaluation.
Even in the absence of clear sequence similarity, it is sometimes possible to confirm structural similarity between a protein of known structure and a sequence by constructing a three-dimensional model of the sequence based on the known structure and evaluating the quality of the model with energy functions (43). To confirm the OB-fold assignment of the 17-90 segment of YhaM, two models were built using two different nucleic acid binding OB-fold structures: the anticodon-binding domain of lysyl-tRNA synthase of E. coli (19) (Protein Data Bank code 1KRS) and the RPA32 subunit of replication protein A (11) (Protein Data Bank code 1QUQ, chain C). Models were built using MODELLER (42) and were evaluated by a procedure based on statistical potentials of mean force (43, 45). The evaluation procedure calculates the probability that the modeled sequence adopts the same structural fold as the template used to model it. This probability is called pG, and a pG above 0.7 is considered significant (43). The two best models for the 17-90 segment of YhaM, based on the anticodon-binding domain of lysyl-tRNA synthase and the RPA32 subunit of replication protein A, had pG values of 0.94 and 0.99, respectively. These results indicated that the 17-90 segment of YhaM is very likely to adopt the OB-fold conformation.
Of all the known OB-fold structures, RPA32 is the most similar to YhaM (17-90) at the sequence level. RPA32 binds single-stranded DNA, and residues W107 and F135 have been predicted to be important for interaction with single-stranded DNA by similarity to the structure of the RPA70 subunit of replication protein A (10). The equivalent residues in YhaM are W51 and Y76 (see Fig. Fig.6A6A and B), suggesting that these residues may play a role in the interaction of YhaM with nucleic acids. To further investigate the relevance of these two residues, their conservation among homologues of YhaM was explored. Homologues of YhaM were searched by identifying protein sequences that contain an OB-fold domain and an HD domain. Searches were performed with HMMER (25), using the HMMs defined for the nucleotide binding OB-fold and the HD domains in PFAM (8). A permissive E-value cutoff was used for the OB-fold search because of the borderline E-value of the YhaM OB-fold domain (see above). The search identified 11 sequences in 10 gram-positive organisms (Fig. (Fig.6C)6C) which contain an OB-fold domain and an HD domain. Sequences annotated as “similar to S. aureus CMP-binding protein” were included among those identified. All sequences have lengths and OB-fold-HD domain spacing that are very similar to those of YhaM. The alignment of the OB-fold regions of all 11 sequences shows that W51 of YhaM is completely conserved and Y76 is conserved except in the Clostridium acetobutylicum and Fervidobacterium islandicum proteins, where conservative mutations to W and F are observed. This conservation lends strong support to the OB-fold assignment and indicates that W51 and Y76 may be important in the interaction of YhaM with single-stranded nucleic acids. It has been pointed out that stacking of aromatic residues and bases is a common theme in RNA-binding domains that are formed by β-sheets (37).
Another OB-fold domain known to interact with RNA is the S1 RNA-binding domain (37). The residues predicted to be important for RNA binding in S1 (14) are not conserved in YhaM (data not shown), and the S1 HMM of PFAM (8) does not identify the 17-90 region of YhaM. These results indicate that the OB-fold domain in YhaM is probably more closely related to the RPA32 subunit of replication protein A than to S1, in spite of the fact that it is an RNA binding domain like S1.
Our initial goal in purifying additional RNases from B. subtilis was to identify those activities responsible for initiation and completion of bulk mRNA decay. Earlier studies with B. subtilis strains with deletions of the genes encoding PNPase and RNase R, the two 3′-to-5′ exoribonucleases identified previously, indicated that these RNases are dispensable with regard to bulk mRNA decay (48, 40). YhaM, the RNase isolated in the present study, turned out to be an interesting exoribonuclease activity that is unique to gram-positive organisms. Our findings confirm the recent speculation of Aravind and Koonin (5), based on predictions of protein domain structure, that YhaM is an RNase. However, because the in vivo function of YhaM remains to be determined, we prefer to defer renaming the yhaM gene and the protein it encodes.
We observed no growth phenotype or significant change in bulk mRNA half-life associated with the absence of YhaM. The absence of YhaM in a strain that also lacks PNPase and RNase R did result in a failure to grow at room temperature and an extended lag phase at 37°C. The absence of individual RNases also affected growth at a low temperature (Table (Table1),1), but not as severely as the triple RNase mutant. It has been reported that the cold shock protein cspA of E. coli, which was named because of its massive induction of expression upon cold shock (28), is also induced when a stationary-phase culture is diluted with fresh medium (12, 49). Thus, the phenotypes of the triple mutant that we observed (cold sensitivity and extended lag phase) may be related.
In E. coli, expression of PNPase is induced by cold shock (31), and a pnp strain of E. coli is severely deficient in colony formation at temperatures below ambient temperature (35, 50). It has been shown recently that the function of PNPase in growth at cold temperatures is to participate in turnover of mRNAs that are involved in the cold adaptation program (50). Since B. subtilis homologues of the major E. coli cold shock proteins are also regulated posttranscriptionally (32), the increased cold sensitivity of the B. subtilis triple RNase mutant may suggest that YhaM is indeed involved in at least some aspects of mRNA decay.
After preparation of this article, a report by Noirot-Gros et al. on protein interactions of the B. subtilis DNA replication apparatus was published (39). Using yeast two-hybrid screening, these authors found a specific interaction of YhaM with DnaC, a DNA helicase that is a component of the replisome. It is possible that the single-stranded DNase activity of YhaM, which we have shown here, could be involved in DNA replication. On the other hand, based on the similarities of YhaM and the S. aureus CBF1 protein, we can speculate that the relevance of YhaM to DNA replication may be a double-stranded DNA binding function. Although we showed that CBF1 protein has exoribonuclease activity that is similar to that of YhaM in its divalent cation dependence, CBF1 was first described as a DNA-binding protein specific for the cmp region of plasmid pT181 (52). While it might seem surprising that an exoribonuclease would have double-stranded DNA-binding capacity, there is a precedent for this observation. It has been shown that E. coli PNPase also can bind specific double-stranded DNA sequences (51).
The properties of purified YhaM are of interest. First, we can rule out the possibility that YhaM is one of the previously isolated hydrolytic B. subtilis RNases. The descriptions of two intracellular RNases that have been reported clearly differentiate those activities from YhaM: Kennell and colleagues reported a pyrimidine-specific endoribonuclease that had a molecular size of about 15 kDa, was active in the presence of Mg2+, and was unable to degrade single-stranded DNA (36). An earlier report described an intracellular exoribonuclease that was present at low levels in vegetative cells and was induced in sporulating cells (33). The size of this protein was estimated to be 72 kDa.
Second, while an absolute requirement for divalent cations (usually Mg2+) is common among the RNases (53), the dependence on Mn2+ (or Co2+) is quite unusual. Deutscher and colleagues have reported that E. coli RNase BN, a 3′ exoribonuclease specific for tRNA, is most active in the presence of Co2+ (15). However, both Mg2+ and Mn2+ could also serve as cofactors for RNase BN. We could not detect any YhaM RNase activity in the presence of a wide range of Mg2+ concentrations. YhaM was active in vitro at low levels of Mn2+ (10 μM), which may be within the range of intracellular Mn2+ concentrations (the actual free Mn2+ concentration is not known).
Third, this is the first report of an RNase that combines two domains—the OB-fold and the HD domain—that are otherwise independently associated with proteins engaged in ribonucleotide biochemistry. None of the proteins annotated as “RNase” in the SWISS-PROT and TrEMBL databases contains an OB-fold that is similar to the one in YhaM. While the S1 domain, which is present in several RNases, is a member of the OB-fold superfamily (14), the YhaM 17-90 region does not resemble an S1 domain; furthermore, the OB-fold in the S1 domain does not resemble the one in YhaM. In addition, none of the S1-containing proteins has an HD domain. Although RNase T of E. coli is similar to YhaM in that it also is able to degrade single-stranded DNA exonucleolytically (53), we have searched for but have not found evidence that YhaM is related to RNase T at the structural or sequence level. Thus, the predicted domain structure of YhaM and the other CBF1 proteins is unique.
The predicted structure of the YhaM OB-fold very clearly suggests this as a single-stranded nucleic acid binding protein, with the exposed aromatic side chains likely to be critical for binding (37). It would be interesting to test whether the W51 and Y76 residues are important for the ability of YhaM to recognize both RNA and single-stranded DNA as substrates. Based on the sequence similarity and modeling calculations, YhaM and its homologues appear to have one OB-fold domain in the 17-90 region and one HD domain in the 160-279 region (YhaM numbering), as shown in Fig. Fig.6C.6C. The region between the OB-fold and HD domains does not match any region of known structure or function in proteins outside this group of OB-plus-HD proteins. The region from position 147 to position 160 contains several highly conserved positions. This may be an extension of the HD domain (light green block in Fig. Fig.6C),6C), which would make the interdomain region even shorter. In addition, the OB-fold domain may cover more than the 17-90 region, because the remote sequence similarity match used to define it is likely to correspond only to the most conserved core of the domain, leaving more-variable regions undetected. With the present data we cannot distinguish absolutely whether the 91-146 interdomain region is merely a linker connecting the OB-fold and HD domains or is a third domain in the OB-plus-HD family, but the small size of this region and the lack of a match to any domains of known structure or function make the latter possibility less likely.
From the bacterial genomes already sequenced, we found that YhaM/CBF1 binding protein sequences are present only in gram-positive organisms. As mentioned above, the B. subtilis genome lacks clear sequence homologues with several important E. coli RNases—RNase E, RNase II, and oligoribonuclease. In addition, it has been shown that RNase III is essential in B. subtilis (30), although it is dispensable in E. coli (6, 46). Thus, the discovery of YhaM further highlights potential differences in the nature of nuclease activities found in gram-negative and gram-positive organisms.
We thank D. Hicks for valuable advice on protein purification and S. Tamasdan for instruction on the use of FPLC.
This work was supported by United States Public Health Service grant GM-48804 from the National Institutes of Health.