INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Upon temperature downshift, Escherichia coli cells rapidly but transiently produce a selective set of proteins called cold shock proteins (CSPs), which are considered to be essential for cellular adaptation to low temperature (33, 60). At 15°C the synthesis of CSPs dramatically increases during the first hour (induction stage) and then is reduced to new basal levels (repression stage). During this period (acclimation phase), including both the induction and repression stages, cell growth is arrested, indicating that cellular events occurring during the acclimation phase are essential for cellular adaptation to low temperature (55, 60, 61). Unlike the heat shock response, in which a heat shock sigma factor (³²) plays a major role in the induction of heat shock proteins (66), a specific sigma factor for cold shock response has not been identified. It has been shown that no de novo protein synthesis is required for the induction of CSPs (17).

CspA is considered to be a major cold shock protein (22, 63), and its expression is regulated in a complex manner at the levels of transcription, mRNA stability, and translation efficiency (60). Among these, mRNA stability is considered to play a major role in cold shock induction (6, 19, 20). Importantly, the cspA mRNA possesses an unusually long 5' untranslated region (5'-UTR) consisting of 159 bases (54), which is crucial for the mRNA stability (6, 19, 20, 64). Although cspA is well transcribed at 37°C (21, 44), CspA is almost undetectable at this temperature due to extreme instability of the cspA mRNA except for immediately after dilution of a stationary-phase culture into a fresh rich medium and upon nutritional upshift (7, 63). The cspA mRNA is dramatically stabilized upon cold shock, allowing a high CspA production (6, 19, 20). RNase E was shown to be involved in the extreme instability of cspA mRNA at high temperatures (19).

The 5'-UTR of the cspA mRNA contains a unique sequence called the cold box (29). When fragments containing the cold-box sequence were overproduced at 15°C, CspA production became constitutive. However, when CspA was simultaneously overproduced, the normal transient expression was resumed (29). Moreover, when a cspA gene without the cold-box sequence was reintroduced into a cspA deletion mutant, CspA production became constitutive due to poor repression at the end of the acclimation phase (2, 18). These results indicated that the cold-box sequence plays an important role in autoregulation of the cspA expression. In addition, mRNA stability was also suggested to play a role in the repression of cspA expression in the repression stage of the acclimation phase on the basis that the half-life of the cspA mRNA became shorter (20).

The facts that mRNA stability is the major factor for regulation of CSP expression (6, 19, 20) and that polynucleotide phosphorylase (PNPase) is a cold shock-inducible exoribonuclease and a component of RNA degradosomes involved in mRNA degradation (1, 3, 4, 16, 33, 43, 49, 55, 60) led us to examine whether PNPase is involved in the regulation of CSP expression upon cold shock. Here, we report that PNPase is essential for the adaptive growth resumption of cold shock-treated cells by selectively degrading mRNAs for CSPs at the end of the acclimation phase. In a pnp mutant, the induction of CSPs upon cold shock was normally observed as in the wild-type strain. However, their production no longer autoregulated and was maintained at high levels throughout cold shock treatment. This resulted in cell growth arrest and a dramatic reduction of colony-forming ability below 25°C. Importantly, during preparation of this paper it was reported that a PNPase-deficient mutant of Yersinia enterocolitica was unable to degrade cspA1/A2 mRNA properly after cold shock (46). Taken together, these results indicate that the PNPase function is required for the critical transition from the acclimation phase to cell growth resumption after cold shock.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. E. coli strains MG1693 (thyA715) (1), SK5691 (thyA715 pnp-7) (1), SH3208 (his trpE5 ) (35), BZ452 (his trpE5 smbB105 zcb::Tn10 ) (35), MC4100 [F araD139 (argF-lac)U169 rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsR] and AR134 (MC4100 but pcnB80) (25) were used in this study. pnp-7 appeared to be a nonsense mutation (1). smbB is an allele of rne, and smbB105 also possesses a nonsense mutation to produce C-terminal half-truncated RNase E (35). MG1693 and SK5691 were provided by S. R. Kushner (University of Georgia), and SH3208 and BZ452 were provided by S. Hiraga (Kumamoto University, Japan).

Construction of a csdA null mutant. The kanamycin-resistant gene (1.3-kb HincII fragment) from pUC7Km(Pst) was inserted into the middle of the coding region of the csdA gene (alternatively, deaD and mssB) at the EcoRV site in pKX164 (65), yielding pKNJ9026. The linearized pKNJ9026 DNA fragment containing the disrupted csdA gene (csdA::kan) was then introduced into the chromosome of a recD mutant FS1576 (recD1009 thi-1 thr-1 leuB6 lacY1 tonA21 supE44). Kanamycin-resistant transformants were isolated, and the disruption of csdA on the chromosome was confirmed by Southern hybridization (data not shown). The csdA::kan mutation was transduced into the wild-type strain MC4100, yielding KNJ130 (MC4100 but csdA::kan).

Protein labeling experiments. Cells were grown in M9 medium supplemented with glucose, 19 amino acids (no methionine), and thymine at 37°C and then were transferred to 15°C. Cells were labeled with [³⁵S]methionine (1,092 Ci/mmol; Amersham) for 10 min. For a functional stability assay of mRNAs, rifampin was added to a final concentration of 200 µg/ml at 1 h after transfer to 15°C and then cells were labeled with [³⁵S]methionine for 5 min at 0, 20, 40, and 80 min after the addition of rifampin. Cell lysates were prepared and processed by two-dimensional gel electrophoresis as described previously (31). The first dimension was carried out within the pH range of 3.5 (right) and 10 (left).

Western blot analysis. Cells were grown in M9 medium supplemented with glucose, Casamino Acids, L-tryptophan, and thymine at 37°C and then were transferred to 15°C. Cells were collected at 0, 4, and 20 h after the temperature downshift, and colony-forming abilities were examined by using Luria-Bertani (LB) plates containing thymine at 37°C. Equal numbers of cells (10⁷ cells) were analyzed by a tricine-sodium dodecyl sulfate (SDS)-16.5% polyacrylamide gel (52) for CspA or by an SDS-12.5% polyacrylamide gel for Era and enolase. After transfer to a nitrocellulose membrane (BA85), CspA, Era, and enolase were detected by using anti-CspA antiserum (54), anti-Era antiserum (42), and anti-enolase antiserum (gift from A. Carpousis), respectively, as described previously (63). Alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin G (IgG; Sigma) was used as a second antibody. Detection was carried out by using the chromogenic substrates 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) and 4-nitroblue tetrazolium chloride (NBT) (Boehringer Mannheim).

Isolation of RNA and primer extension analysis. Cells were grown in M9 medium supplemented with glucose, Casamino Acids, L-tryptophan, and thymine at 37°C and then were transferred to 15°C. Cells were collected at 0, 0.5, and 4 h after the temperature downshift, and RNA was extracted by the hot-phenol method as described previously (62). Primer no. 8285, 5'-ACATAGTGTATTACCTTTAA-3', which corresponds to the complementary strand of +144 to +163 of cspA, where the transcription start site was assigned as +1 (54), was labeled with [-³²P]ATP (>5,000 Ci/mmol; DuPont-New England Nuclear) by T4 polynucleotide kinase (Gibco BRL). Primer extension was carried out using avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim) with 5 µg of RNA as described previously (62). Primer extension products were analyzed on a 6% polyacrylamide gel under denaturing conditions in 1× Tris-borate-EDTA (TBE) and 6 M urea.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of a pnp mutation on protein expression upon cold shock. In the wild-type strain, CSPs, such as CspA, RbfA, and CsdA, were almost undetectable at 37°C (Fig. 1a) but were dramatically induced upon temperature downshift to 15°C (Fig. 1b). At 4 h (Fig. 1c) and 20 h (Fig. 1d) after the downshift, the expression of individual CSPs in the wild-type cells was reduced to a new basal level. In contrast to CSP expression, the synthesis of most of the non-cold shock proteins (non-CSPs) was transiently inhibited during the acclimation phase. However, after the acclimation phase their synthesis was resumed with concomitant repression of the CSP expression (see Fig. 1c). This is consistent with the notion that cell growth is resumed approximately 2 h after cold shock to 15°C (22).

View larger version (84K): [in this window] [in a new window]   FIG. 1.   Two-dimensional analysis of total cellular proteins upon cold shock in pnp+ and pnp-7 cells. Cells were labeled with [35S]methionine at 37°C (a and e) and then at 0.5 h (b and f), 4 h (c and g), and 20 h (d and h) after temperature downshift from 37 to 15°C. Cell lysates were analyzed by two-dimensional gel electrophoresis as described previously (31). The first dimension was carried out within the range of 3.5 (right) and 10 (left). (a to d) The wild-type strain MG1693. (e to h) The pnp-7 mutant SK5691. Typical CSPs are circled. Spot no. 1, PNPase (33); no. 2, CsdA (32); no. 3, RbfA (31); no. 4, CspA and CspG (53); no. 5, CspI (53); and no. 6, CspB (53).

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Primer extension analysis of the cspA mRNA. RNA extraction and primer extension analysis were carried out as described in Materials and Methods. The primer extension products were analyzed on a 6% denatured polyacrylamide gel. Lanes 1, 4, 7, and 10, RNA extracted from exponentially growing cells at 37°C; lanes 2, 5, 8, and 11, RNA extracted from cells at 0.5 h after cold shock; lanes 3, 6, 9, and 12, RNA extracted from cells at 4 h after cold shock. Lanes 1 to 3, MG1693 (pnp+); lanes 4 to 6, SK5691 (pnp-7); lanes 7 to 9, SH3208 (rne+); lanes 10 to 12, BZ452 (smbB105). The relative cspA mRNA amounts were calculated using the transcript at 0.5 h of each strain as 100%.

View larger version (46K): [in this window] [in a new window]   FIG. 3.   Immunological detection of CspA, Era, and enolase in pnp+ and pnp-7 cells after temperature downshift. Cells were grown in M9 medium supplemented with glucose, Casamino Acids, L-tryptophan, and thymine at 37°C to a mid-exponential phase and then were transferred to 15°C. Cells were collected at 0 (lanes 1 and 4), 4 (lanes 2 and 5), and 20 (lanes 3 and 6) h after temperature downshift, and equal numbers of cells (107 cells) per lane were analyzed by a tricine-SDS-16.5% polyacrylamide gel (52) for CspA and by an SDS-12.5% polyacrylamide gel for Era and enolase. After transfer to nitrocellulose membrane (BA85), proteins were detected by using anti-CspA antiserum, anti-Era antiserum, or anti-enolase antiserum. Alkaline phosphatase-conjugated goat anti-rabbit IgG (Sigma) was used as a second antibody. Detection was carried out by using the chromogenic substrates BCIP and NBT. Lanes 1 to 3, MG1693 (pnp+); lanes 4 to 6, SK5691 (pnp-7).

Selective degradation of mRNAs by PNPase. The results described above indicate that the cspA expression in the pnp-7 mutant is normally induced upon cold shock but is not autoregulated even in the presence of a very high concentration of CspA. In the wild-type strain, CspA production upon cold shock is transient and the rate of CspA synthesis is rapidly reduced in the later half of the acclimation phase (22). Since the change in the rate of CspA synthesis has been shown to be parallel to the change in the amount of cspA mRNA, the cspA autoregulation during the acclimation phase is considered to be at the level of mRNA.

View larger version (67K): [in this window] [in a new window]   FIG. 4.   Analysis of the functional stability of mRNA. Cells were grown at 37°C to mid-exponential phase, transferred to 15°C, and then further incubated for 1 h. Rifampin was added to a final concentration of 200 µg/ml. Cells were labeled with [35S]methionine for 5 min at 0 min (a and e), 20 min (b and f), 40 min (c and g) and 80 min (d and h) after the addition of rifampin. Cell lysates were analyzed by two-dimensional gel electrophoresis under the same conditions as in Fig. 1. (a to d) The wild-type strain MG1693. (e to h) The pnp-7 mutant SK5691. Typical CSPs are circled. Spot no. 1, PNPase; no. 2, CsdA; no. 3, RbfA; no. 4, CspA and CspG; no. 5, CspI; no. 6, CspB. Forty spots were densitometrically measured, and functional half-lives were calculated from two experiments.

Requirement of PNPase for cell growth at low temperature. Since the synthesis of CSPs was not repressed in the pnp mutant (Fig. 1 and 2), it is speculated that the pnp mutant may not be able to adapt to low temperature, resulting in a cell growth defect. As shown in Fig. 5, the pnp mutant was extremely sensitive to low temperature and was unable to form colonies below 25°C, which agrees well with the results reported by Luttinger et al. (41); the colony-forming ability dropped to 10⁷. The cold-sensitive growth phenotype of the pnp mutant was fully complemented by transforming cells with plasmid pKX150 carrying the pnp⁺ gene (65), indicating that PNPase function is essential for growth at low temperature but not at high temperature.

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Colony-forming abilities of the pnp mutant at various temperatures. Cultures of strains were grown in LB medium containing thymine (50 µg/ml) and chloramphenicol (30 µg/ml) at 37°C overnight. After appropriate dilutions, cells were plated on LB agar plates containing thymine (50 µg/ml) and chloramphenicol (30 µg/ml) and were incubated at 42, 37, 30, 25, 20, and 15°C. , the wild-type strain MG1693 carrying the vector pHSG575; , MG1693 carrying the pnp+ plasmid pKX150 (65); , the pnp-7 mutant SK5691 carrying pHSG575; , SK5691 carrying pKX150.

Effect of C-terminal half-truncated RNase E on CSP expression. PNPase is known to interact with the C-terminal region of RNase E and to form the RNA degradosome in association with RhlB (an RNA helicase), DnaK, and enolase (43, 49, 57). Recently it was reported that only RNase E, PNPase, and RhlB are enough to reconstitute a functional RNA degradosome (13). At low temperature, more stable stem-loop structures are formed in mRNAs, which makes them more resistant to degradation by exoribonucleases, such as PNPase. Among the RNA degradosome components, only PNPase is known to be cold shock inducible. Therefore, it is interesting to determine whether cold shock-induced PNPase constitutes a functional component of the RNA degradosome. To test this question, we used the smbB105 mutant, in which the C-terminal half of RNase E is truncated (35), resulting in no formation of the RNA degradosome (35, 52). As shown in Fig. 6, CSPs were not detected at 37°C in either wild-type or smbB105 strains (Fig. 6a and 6d, respectively), while they were dramatically induced immediately upon cold shock in both strains (Fig. 6b and 6e, respectively). At 4 h after temperature downshift, their expression in the mutant was repressed as in the wild-type strain (Fig. 6f and 6c, respectively). The changes in the amounts of cspA mRNA in the smbB105 mutant upon cold shock (Fig. 2, lanes 10 to 12) were also very similar to those observed in the wild-type strain (Fig. 2, lanes 7 to 9). Furthermore, cell growth was not affected by the smbB105 mutation at either high or low temperatures (data not shown). These results indicate that the association of PNPase with RNase E or the degradosome formation is not required for the adaptive degradation of CSP mRNAs by PNPase.

View larger version (98K): [in this window] [in a new window]   FIG. 6.   Two-dimensional analysis of total cellular proteins upon cold shock in rne+ and smbB105 cells. Cells were labeled with [35S]methionine at 37°C (a and d) and then at 0.5 h (b and e) and 4 h (c and f) after temperature downshift from 37 to 15°C. Cell lysates were analyzed by two-dimensional gel electrophoresis under the same conditions as in Fig. 1. (a to c) The wild-type strain SH3208. (d to f) The smbB105 mutant BZ452. Typical CSPs are circled in b and e. Spot no. 1, PNPase; no. 2, CsdA; no. 3, RbfA; no. 4, CspA and CspG; no. 5, CspI; no. 6, CspB.

Effect of a pcnB mutation on CSP expression. It is reasonable to assume that the key for the selective degradation of a group of mRNAs exists in the 3' end of mRNAs, since PNPase is a 3' to 5' exoribonuclease. Not only in eukaryotes but also in prokaryotes, polyadenylation is a well-known, posttranscriptional modification at the 3' end of mRNAs (51). E. coli contains poly(A) polymerase encoded by pcnB (9), and poly(A) tail has been suggested to regulate mRNA degradation (5, 11, 12, 24, 26, 28, 47). Moreover, purified PNPase preferentially degrades polyadenylated RNA but not nonpolyadenylated RNA when these two species exist simultaneously (39). If CSP mRNAs are selectively polyadenylated, one can expect that CSP synthesis should be prolonged in a pcnB mutant due to reduced degradation. To examine if poly(A) tail is involved in CSP expression, we carried out the pulse-labeling experiment using the pcnB80 mutant and subsequently Western blotting analysis with the anti-CspA antibody (25, 40). Interestingly, CspA expression was clearly detected in the pcnB80 mutant at 37°C (Fig. 7e), while that in the wild-type strain was hardly detected (Fig. 7a). Indeed, the amount of CspA in the pcnB80 mutant was a little higher than that in the wild-type strain (data not shown). These results suggest that polyadenylation is involved in CspA expression to some extent; that is, polyadenylation might enhance the degradation of CspA mRNA. Upon temperature downshift, CSPs were dramatically induced in the wild-type cells (Fig. 7b) as well as in the pcnB80 mutant cells (Fig. 7f). However, importantly the synthesis of CSPs in the pcnB80 mutant was prolonged to 4 h after temperature downshift (compare Fig. 7g to 7c), and at 20 h it dropped to a new basal level, which is a level similar to that in the wild-type strain (Fig. 7d and 7h). Consistently, it was found that the amount of CspA in the pcnB80 mutant was significantly higher at 20 h after temperature downshift than that in the wild-type strain (data not shown). In contrast, the amount of enolase, a non-CSP, was unchanged upon temperature downshift (data not shown). These results indicate that polyadenylation plays an important role in the degradation of CSP mRNAs. It should be mentioned, however, that in contrast to PNPase, the pcnB80 mutant did not show the cold-sensitive growth phenotype (data not shown) and poly(A) polymerase was not a CSP from the analysis of pcnB::lacZ fusion constructs (data not shown).

View larger version (51K): [in this window] [in a new window]   FIG. 7.   Two-dimensional analysis of total cellular proteins upon cold shock in pcnB80 and csdA mutant cells. Cells were labeled with [35S]methionine at 37°C (a, e, and i) and then at 0.5 h (b, f, and j), 4 h (c, g, and k), and 20 h (d, h, and l) after temperature downshift from 37 to 15°C. Cell lysates were analyzed by two-dimensional gel electrophoresis under the same conditions as in Fig. 1. Only a portion corresponding to low-molecular-weight proteins is shown. (a to d) The wild-type strain MC4100. (e to h) The pcnB80 mutant AR134. (i to l) The csdA::kan mutant KNJ130. Typical CSPs are indicated by an arrow. Spot no. 1, RbfA; no. 2, CspA and CspG; no. 3, CspI; no. 4, CspB.

Effect of a csdA mutation on CSP expression. It is interesting to note that poly(A) polymerase of E. coli was reported to interact with RNA, RNase E, and DEAD-box RNA helicases of RhlE, SrmB, and CsdA by means of far Western analysis (50). Among them, CsdA is a cold shock-inducible protein and has an activity to unwind double-stranded RNA in the absence of ATP (32). CsdA function was proposed to be essential for ribosome function to increase translational efficiencies of mRNAs by unwinding stable secondary structures formed at low temperature (32). To examine whether CsdA is involved in the selective degradation of CSP mRNAs after the acclimation phase, we prepared a csdA null mutant as described in Materials and Methods. CSPs were hardly detected in the csdA mutant at 37°C (Fig. 7i), as in the wild-type strain (Fig. 7a). Consistently, the amount of CspA in the csdA mutant at 37°C was almost undetectable (data not shown). Upon temperature downshift to 15°C, CSPs were dramatically induced in the csdA mutant (Fig. 7j). Similar to the pcnB mutant described above, CSP expression in the csdA mutant was prolonged to 4 h after temperature downshift (Fig. 7k) and it was dropped to the new basal level at 20 h (Fig. 7l). The amount of CspA in the csdA mutant was consistent with these results (data not shown). The amount of enolase was again not affected by the csdA mutation (data not shown). It should also be mentioned that the csdA mutant did grow at low temperature with a reduced growth rate (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The present work demonstrates that PNPase is involved in selective degradation of CSP mRNAs at the stage of repression during the acclimation phase. Very recently it was reported that a PNPase-deficient mutant of Y. enterocolitica was unable to degrade cspA1/A2 mRNA properly after cold shock (46). The authors suggested that after synthesis of CSPs and cold adaptation of the cells, CSP mRNAs must be degraded; otherwise, they trap ribosomes, prevent translation of bulk mRNAs, and thus inhibit growth of the bacterium at low temperatures (46). Our results demonstrated here are fully consistent with their observations, and these results together suggest that PNPase regulates gene expression by selectively degrading specific mRNAs. Such a loss of autoregulation of CSP genes during the process of cold shock adaptation has also been observed in another CSP mutant, an rbfA mutant. rbfA encodes a cold shock protein required for ribosomal maturation and/or translation initiation. A constitutive induction of the cold shock response occurs in this mutant in a manner similar to that found in the present study, resulting in slower cell growth (31). It is thus apparent that continuous high expression of CSPs is deleterious to cells and that for cell growth resumption, their expression has to be repressed once the cellular concentrations of individual CSPs reach optimal levels.

PNPase was reported to be cold shock inducible in E. coli (33), Y. enterocolitica, a psychrotrophic bacterium (23), and Photorhabdus sp. (family Enterobacteriaceae) (10). In E. coli grown at 37°C, PNPase was shown to be responsible for only 10% of the total processive 3' to 5' mRNA degradation (15). It is important to note that the phosphorolytic degradation of mRNA by PNPase is less energetically expensive than the hydrolytic degradation by RNase II, another exoribonuclease in E. coli, as the high-energy phosphate bonds of the ribonucleotide are retained in the PNPase reaction (14, 15). Thus, PNPase may be more favorable than RNase II for mRNA degradation in E. coli at low temperature. In a pnp mutant of Y. enterocolitica, cell growth was severely restricted at 5°C but not at 30°C (23), and in the case of Bacillus subtilis, a pnp mutation caused cold-sensitive cell growth (41, 59). Luttinger et al. (41) and we also (in Fig. 5) demonstrated that the E. coli pnp mutant showed a cold-sensitive growth phenotype. All these results strongly indicate the importance of PNPase under the low-temperature growth condition. It should be noted that in B. subtilis, PNPase has been suggested to be essential for the expression of specific genes at the posttranscriptional level, mRNA stability, or translation (41).

An important feature of PNPase is that it contains two RNA-binding domains, a KH domain and an S1 domain (8). The S1 domain was originally identified in ribosomal protein S1 (53). The three-dimensional structure of the S1 domain of PNPase has been determined and found to contain a -barrel structure similar to that of CspA (8). CspA is an RNA-binding protein and has been proposed to function as an RNA chaperone (30). Furthermore, PNPase (39) and ribosome protein S1 (34) were reported to preferentially bind to the poly(A) sequence, and PNPase preferably degrades polyadenylated RNA (39). In plant chloroplasts, it has also been shown that PNPase preferentially degrades polyadenylated RNA (38) and that PNPase is a component of the poly(A) polymerase complex (36). Very recently, Mohanty and Kushner reported that PNPase functions as both an exonuclease and a poly(A) polymerase in E. coli (45). While there are few sequence homologies at the 3'-UTRs among cspA, cspB, cspG, and cspI, the addition of poly(A) tail to CSP mRNAs may result in selective degradation of these mRNAs by PNPase at low temperature (see Fig. 7). Therefore, at least at low temperature, PNPase in concert with poly(A) polymerase is reasonably considered to play an important role in the selective degradation of CSP mRNAs. However, it is also possible that an additional unknown factor(s) may selectively recognize mRNAs for CSPs to enhance their degradation by PNPase or that PNPase may selectively recognize a common sequence such as the cold box in the 5'-UTR of CSP mRNAs to enhance their degradation. In any case, the precise mechanism of selective degradation of CSP mRNAs by PNPase at low temperature awaits further characterization.

PNPase is known to form the RNA degradosome together with at least RNase E and RhlB (13, 57). The endoribonuclease RNase E and the RNA helicase RhlB are speculated to cooperatively act with PNPase for mRNA degradation. In the present study, however, we demonstrated that the interaction of PNPase with RNase E and RhlB is not required for the function of PNPase and cell growth at low temperature (see Fig. 2 and 6). Therefore, the RNA degradosomes are unlikely to play major roles in cells growing at low temperature. Very recently it was reported that although RNA degradosomes indeed exist in vivo in E. coli as multicomponent complexes, PNPase and enolase are present in E. coli in large excess relative to RNase E (37). Therefore, PNPase is detected in cells largely as molecules unlinked to the RNase E (37). Furthermore, the assembly of the RNase E-based degradosome of E. coli was reported not to be required for normal mRNA decay in vivo (48). Our results presented here are fully consistent with these results. It should be mentioned that poly(A) polymerase was also shown to interact with RNA, RNase E, and DEAD-box RNA helicases (RhlE, SrmB, and CsdA) (50), among which CsdA is cold shock inducible (32). Indeed, in the csdA mutant used in this study, CSP expression was significantly prolonged (see Fig. 7), suggesting that CsdA may play a role in unwinding RNA structures that impede the processive activity of PNPase. Note that the CsdA unwinding activity is independent of ATP (32), which might be an advantage in terms of energy consumption at low temperature. Taken together, PNPase, poly(A) polymerase, and CsdA are involved in the efficient and selective degradation of CSP mRNAs, which is essential for the cold shock adaptation of E. coli.

In the pnp mutant used in the present work, CSP mRNAs were specifically stabilized approximately twofold (Fig. 4), suggesting that PNPase is associated with the degradation of CSP mRNAs. Interestingly, in contrast to CSP mRNAs, non-CSP mRNAs became more unstable in the pnp mutant than in the wild-type strain. It is tempting to speculate that increased concentrations of CspA and other CspA homologues in the mutant cells destabilize non-CSP mRNAs as CspA and its homologues function as RNA chaperones (30).

In conclusion, the present results together with a recent report (46) represent that a specific ribonuclease plays an essential role in stress adaptation by selectively degrading mRNAs for stress-response proteins. As the overproduction of stress-response proteins appears to be deleterious to cells, the overproduction is prevented by a ribonuclease, which is induced by the same stress. In the case of the heat shock response, ³² plays the major role (66) and it has been shown that ³² is degraded by a specific protease, FtsH, which is also induced by heat shock, to regulate the heat shock response (27, 56). Thus, it is important to mention that a specific ribonuclease plays a critical role in cold shock response and adaptation in contrast to heat shock response and adaptation, where a specific protease plays a major role. This is consistent with the fact that the fate of individual mRNAs for each CSP plays a central role in cold shock. Cold shock stress is likely to be a major stress for most of the prokaryotes in nature, and PNPase may be essential for low-temperature survival and proliferation of not only E. coli and B. subtilis but also many other prokaryotes.