Proteolytic Adaptor for Transfer-Messenger RNA-Tagged Proteins from α-Proteobacteria

Abstract

We have identified an analog of SspB, the proteolytic adaptor for transfer-messenger RNA (tmRNA)-tagged proteins, in Caulobacter crescentus. C. crescentus SspB shares limited sequence similarity with Escherichia coli SspB but binds the tmRNA tag in vitro and is required for optimal proteolysis of tagged proteins in vivo.

AAA+ proteases such as ClpXP degrade a wide variety of substrates and are important for many physiological events in bacteria (9, 11, 13, 14, 27, 31). To enhance the rate of proteolysis and to increase the specificity of substrate selection, these proteases can employ adaptor proteins that promote or inhibit degradation of certain classes of substrates (5, 26, 32). The best-characterized of these proteolytic adaptor proteins is SspB from Escherichia coli, which promotes the degradation of several ClpXP substrates, including proteins tagged by transfer-messenger RNA (tmRNA) during translation (10, 26). tmRNA is a specialized RNA that can enter the ribosome during translation and mediate the addition of a short peptide, encoded within tmRNA itself, to the C terminus of the nascent protein (22). Many proteins tagged by tmRNA are made from mRNAs that are damaged or do not have an in-frame stop codon, and the tagged proteins are rapidly degraded in vivo (18, 22). Genes encoding tmRNA and SmpB, a protein that binds to tmRNA and is required for its activity, have been identified in all bacteria with sequenced genomes, as well as some chloroplasts and mitochondria (15, 21). ClpXP is also widely conserved, but SspB has been identified only in γ- and β-proteobacteria. Here we report the identification of a functional analog of SspB in Caulobacter crescentus and all α-proteobacteria.

Deletion of the gene encoding tmRNA in C. crescentus results in a cell cycle defect, and this defect is not complemented by tmRNA-DD, a variant of tmRNA that tags proteins but does not target them for degradation (20). These results suggested that proteolysis of tmRNA-tagged proteins is required for proper timing of the cell cycle in C. crescentus (20) and instigated a search for genes that might be involved in this proteolysis. The conserved hypothetical protein CC2102 was investigated because it is immediately 3′ of ssrA, the gene encoding tmRNA. Sequence similarity searches revealed that CC2102 is conserved throughout the α-proteobacteria and contains a domain of unknown function, DUF1321, that is unique to these proteins (28). In fact, CC2102 was proposed as a gene that is diagnostic for α-proteobacteria (17). Three iterations of PSI-BLAST (1) identified some sequence similarity between CC2102 and SspB proteins from γ- and β-proteobacteria (Fig. 1A). Based on this sequence similarity and the characterization below, we refer to CC2102 as SspB. The identification of SspB homologs here extends the phylogenetic distribution of SspB through the α-, β-, and γ-proteobacteria and raises the possibility that other bacteria contain analogs of SspB that are sufficiently divergent to make them difficult to identify by similarity searches alone. The PSI-BLAST search that linked CC2102 to SspB also identified genes from δ-proteobacteria and Bacteroidetes species that are related to SspB, including Mxan_2058, which is adjacent to smpB in the Myxococcus xanthus genome.

FIG. 1.

Alignment of SspB and tmRNA tag sequences. SspB sequences (A) and tmRNA tag peptide sequences (B) from bacterial species were aligned using CLUSTALW (4). Residues conserved in all three species are shaded, and those that are identical in all sequences are in boldface. Dots indicate SspB residues that make side chain contacts to the tmRNA tag sequence in published crystal structures (24, 25, 29). Triangles indicate tmRNA tag residues that are required for SspB binding (10).

Mutagenesis studies of the E. coli tmRNA tag peptide revealed that residues 1 to 4 and 7 are important for SspB binding (10) and that these residues are conserved in the C. crescentus tag (Fig. 1B). Cocrystal structures of SspB from Haemophilus influenzae and E. coli with the cognate tmRNA tag peptides showed that many of the SspB residues that directly contact the tmRNA tag peptide are in three loops: L1, L2, and L3 (25, 29). The sequence KFGGQP in C. crescentus SspB is similar to those in L3 of E. coli and H. influenzae SspB (Fig. 1A), but the residues in L1 and L2 are not conserved, nor are other residues that lead to hydrogen bonding or hydrophobic contacts with the tag peptide. Therefore, despite the similarities in the tmRNA tag peptides between C. crescentus, E. coli, and H. influenzae, it appears that many of the SspB residues that are in contact with the tag peptide are not conserved.

One possible explanation for the lack of sequence conservation in the substrate binding domain of SspB is that this domain is also used to recognize diverse peptides in substrates that are not tagged by tmRNA. In E. coli, SspB also acts as a proteolytic adaptor for proteins with little resemblance to the tmRNA tag peptide, including a fragment of RseA (9). Remarkably, SspB binds RseA in an orientation opposite to that of the tmRNA tag peptide, indicating that overlapping substrate specificities are encoded in the same binding pocket (24). It is not yet known whether C. crescentus SspB is an adaptor for substrates in addition to tmRNA-tagged proteins, but the divergence in SspB sequences may be due in part to coevolution with substrates that do not use the tmRNA tag.

SspB binds to the C. crescentus tmRNA tag peptide.

The E. coli SspB protein is a dimer in solution, and this dimer binds to two molecules of tmRNA-tagged protein in vitro (30). To determine whether the C. crescentus SspB homolog could bind the C. crescentus tmRNA tag, a variant of SspB with six histidine residues at its N terminus was produced in E. coli and purified. Gel filtration of the purified protein on a Superose 6 column resulted in a single peak corresponding to a molecular mass of approximately 40 kDa, consistent with a dimer of the 18-kDa protein SspB (Fig. 2). When C. crescentus SspB was incubated with equimolar concentrations of purified green fluorescent protein (GFP) containing the tmRNA-encoded peptide AANDNFAEEFAVAA (GFP-tag), a single peak eluted from the gel filtration column at approximately 100 kDa. The protein from this peak was separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and stained with silver, revealing that both SspB and the GFP tag were present (Fig. 2). Quantification of the silver-stained bands using purified GFP tag standards showed that the ratio of GFP tag to SspB was 1.0 to 1.1. Therefore, it is likely that this peak corresponds to a mixed tetramer containing two molecules of SspB and two molecules of GFP tag. Gel filtration of SspB incubated with GFP lacking the tmRNA tag resulted in two peaks, corresponding to a dimer of SspB and a monomer of GFP, indicating that SspB requires the tmRNA tag peptide for binding to GFP (not shown). These data demonstrate that C. crescentus SspB binds to the cognate tmRNA tag peptide and are consistent with a protease adaptor mechanism similar to that of E. coli SspB.

FIG. 2.

SspB binds to the tmRNA tag peptide. SspB from C. crescentus (6 μM), GFP tag (6 μM), and a mixture containing a 6 μM concentration of each protein were analyzed by gel filtration. Proteins in the peak fractions were separated on SDS-polyacrylamide gels and stained with silver.

SspB is required for optimal degradation of tmRNA-tagged proteins in vivo.

To determine whether C. crescentus SspB affects the proteolysis of tmRNA-tagged proteins in vivo, the degradation of a λ repressor reporter protein (λ-cI-N-trpAt) (22) was assayed in the wild type and strains with sspB deleted. Induction of this reporter results in an mRNA that lacks a stop codon, and protein made from this mRNA is targeted to the tmRNA pathway. Pulse-chase assays wild-type C. crescentus strains showed that most of the λ-cI-N-trpAt was degraded in the dead time of the experiment (Fig. 3). Based on the complete elimination of labeled protein by the 2-min point, the half-life of λ-cI-N-trpAt in the wild type is estimated to be less than 0.5 min. To verify that degradation was caused by tmRNA tagging and not intrinsic instability of λ-cI-N-trpAt, the experiment was repeated in cells containing tmRNA-DD. tmRNA-DD-tagged λ-cI-N-trpAt had a half-life of 60 ± 4 min, confirming that addition of the wild-type tmRNA tag targets the protein for degradation. In cells lacking SspB, the half-life of λ-cI-N-trpAt was 4 ± 0.2 min, an increase of more than eightfold compared to cells containing SspB (Fig. 3). This increase in stability is similar to the increase observed for an identical reporter in the E. coli sspB strain (0.5 min in the wild type versus 5 min in the ΔsspB strain) (10, 26), suggesting that C. crescentus SspB is a proteolytic adaptor for tmRNA-tagged proteins.

FIG. 3.

Deletion of sspB increases the half-life (t_1/2) of tmRNA-tagged protein in vivo. Results of pulse-chase assays of λ-cI-N-trpAt tagged by tmRNA in a wild-type strain, a ΔsspB strain, and a wild-type strain producing tmRNA-DD are shown. The C. crescentus sspB gene was deleted using sacB counterselection (12). Production of λ-cI-N-trpAt was induced by addition of IPTG (isopropyl-β-d-thiogalactopyranoside) for 30 min, and cultures were pulsed for 2 min with 150 μCi [³⁵S]methionine before the addition of excess unlabeled methionine. Cells were harvested and lysed at the indicated times, and λ-cI-N-trpAt was partially purified by affinity chromatography over Ni-nitrilotriacetic acid agarose (QIAGEN). Degradation of the labeled protein was monitored by separating the proteins using SDS-polyacrylamide gel electrophoresis and monitoring the loss of the band corresponding to the λ-cI-N-trpAt (arrows). The average half-life and standard deviation from at least three independent experiments are indicated. For the wild type the half-life was estimated to be <30 s based on the complete absence of labeled protein by 2 min.

SspB is not coregulated with tmRNA and SmpB.

The amounts of tmRNA and SmpB are regulated through the C. crescentus cell cycle, increasing during the transition from G₁ to S phase and decreasing rapidly in early S phase (16, 19). To assess the cell cycle regulation of SspB, a polyclonal antibody was raised against purified SspB and used to probe Western blots of cell lysates from synchronized cultures (Fig. 4). Unlike the amount of tmRNA and SmpB, the amount of SspB protein was nearly constant throughout the cell cycle, indicating that the amount of SspB in the cell is proportional to the cell volume and independent of the cell cycle.

FIG. 4.

Cell cycle regulation of sspB expression. C. crescentus swarmer cells (G₁ phase) were isolated from Ludox density gradients and allowed to pass synchronously through the cell cycle (7). The schematic diagram indicates the timing of differentiation and DNA replication as monitored according to light microscopy and flow cytometry (20). Levels of SspB protein were analyzed by Western blotting, quantified, and normalized to the amount at time zero. A representative Western blot with an arrow indicating the SspB band (top) and quantification (bottom, triangles) are shown. sspB mRNA levels were measured by Q-PCR using Taqman probes (Applied Biosystems) and normalized to 16S rRNA to control for small differences in the efficiency of RNA preparation and reverse transcription. The average of three independent Q-PCRs performed on RNA extracted from a representative synchronized culture is shown (bottom, circles) with error bars indicating the standard deviation. Other synchronized cultures had mRNA levels with very similar trends and magnitudes.

Quantitative PCR (Q-PCR) measurements showed that sspB mRNA levels fluctuated less than 2.5-fold during the cell cycle, with peak expression at 45 to 60 min (Fig. 4). These data are consistent with results obtained from DNA microarray studies (23). The sspB mRNA accumulated later in the cell cycle than tmRNA, which peaks at 30 min and is at its minimum level at 45 min (16, 19). Why does the sspB mRNA level fluctuate if the protein level is constant? One possibility is that SspB protein is degraded during the cell cycle, and transcription increases at the corresponding time to maintain constant protein levels. Alternatively, the relatively small fluctuation in sspB mRNA may not result in increased protein production.

SspB levels were not affected by the growth phase of the culture or by the activity of tmRNA. Western blots of cell lysates taken from early exponential phase to stationary phase showed that SspB is present at approximately 43,000 copies per cell and varies by less than 5% from early exponential phase to stationary phase (not shown). The steady-state level of SspB protein was 40,000 to 45,000 molecules per cell in ΔssrA and ΔsmpB strains (not shown), indicating that the SspB protein level does not depend on the tmRNA pathway. Conversely, the steady-state amount of tmRNA and the cell cycle regulation of tmRNA levels were not altered in ΔsspB cells (not shown), indicating that tmRNA expression does not depend on SspB. Therefore, no regulatory interactions between expression of sspB and tmRNA or SmpB could be detected.

SspB is not required for the tmRNA phenotype.

C. crescentus strains lacking tmRNA or SmpB grow slowly due to a delay in the cell cycle and do not maintain some broad-host-range plasmids (20). Strains in which sspB was either replaced by a spectinomycin resistance cassette or removed, leaving the two 5′ codons and two 3′ codons in frame, were characterized to determine whether they had the same phenotypes as the ΔssrA and ΔsmpB strains. Cells lacking SspB grew at the same rate as the wild type and much faster than the ΔssrA and ΔsmpB strains, both in complex medium and in defined medium. Synchronized populations of the sspB cells initiated DNA replication at the same time as the wild type and showed no defects in the timing of other cell cycle-regulated events characteristic of the ΔssrA and ΔsmpB strains. The sspB strain also maintained plasmids as efficiently as the wild type did. Therefore, cells lacking SspB activity do not have the phenotypes associated with defects in tmRNA tagging.

If degradation of tmRNA-tagged proteins is required for optimal growth and coordination of the cell cycle in C. crescentus, why is there no phenotype in cells lacking SspB? Presumably, SspB is not required because degradation of tmRNA-tagged proteins with a 4-min half-life is sufficiently rapid to prevent cell cycle disruption. Similarly, no growth phenotype has been reported in E. coli strains with sspB deleted. However, C. crescentus invests in over 40,000 copies of SspB per cell, compared to ∼300 copies of SspB per E. coli cell (8). It is possible that SspB is important for proteolysis of tmRNA-tagged proteins or other ClpXP substrates in C. crescentus under growth conditions not examined in this study or that C. crescentus strains with SspB have a competitive advantage that is not detected in clonal cultures.

The only reported phenotype associated with SspB is from Francisella novicida, in which the SspB homolog, MglB, is required for intramacrophage growth (2). Interestingly, Salmonella enterica serovar Typhimurium strains lacking tmRNA have a similar defect in intramacrophage growth (3), and Bradyrhizobium japonicum strains lacking tmRNA cannot grow in root nodules (6), raising the possibility that SspB-mediated proteolysis of tmRNA-tagged proteins is required for pathogenesis and symbiosis or under some types of severe stress conditions.