Advertisement

CspA, the Major Cold-shock Protein of Escherichia coli, Is an RNA Chaperone*

  • Weining Jiang
    Affiliations
    Department of Biochemistry, Robert Wood Johnson Medical School, Piscataway, New Jersey 08854
    Search for articles by this author
  • Yan Hou
    Affiliations
    Department of Biochemistry, Robert Wood Johnson Medical School, Piscataway, New Jersey 08854
    Search for articles by this author
  • Masayori Inouye
    Correspondence
    To whom correspondence should be addressed
    Affiliations
    Department of Biochemistry, Robert Wood Johnson Medical School, Piscataway, New Jersey 08854
    Search for articles by this author
  • Author Footnotes
    * This work was supported by National Institutes of Health Grant GM19043. The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
      CspA, the major cold-shock protein of Escherichia coli, is dramatically induced during the cold-shock response. The amino acid sequence of CspA shows 43% identity to the “cold-shock domain” of the eukaryotic Y-box protein family, which interacts with RNA and DNA to regulate their functions. Here, we demonstrate that CspA binds to RNA as a chaperone. First, CspA cooperatively binds to heat-denatured single-stranded RNA if it is larger than 74 bases, causing a supershift in gel electrophoresis. A minimal concentration of CspA at 2.7 × 10−5M is absolutely required for this cooperative binding, which is sufficiently lower than the estimated cellular concentration of CspA (10−4M) in cold-shocked cells. No specific RNA sequences for CspA binding were identified, indicating that it has a broad sequence specificity for its binding. When the 142-base 5′-untranslated region of the cspA mRNA was used as a substrate for ribonucleases A and T1, the addition of CspA significantly stimulated RNA hydrolysis by preventing the formation of RNase-resistant bands due to stable secondary structures in the 5′-untranslated region. These results indicate that binding of CspA to RNA destabilizes RNA secondary structures to make them susceptible to ribonucleases. We propose that CspA functions as an RNA chaperone to prevent the formation of secondary structures in RNA molecules at low temperature. Such a function may be crucial for efficient translation of mRNAs at low temperatures and may also have an effect on transcription.

      INTRODUCTION

      The cold-shock response is a physiological response of living cells to temperature downshift, which is considered to be essential for cells to survive at low temperatures (for review, see
      • Jones P.G.
      • Inouye M.
      ). When an exponentially growing culture of Escherichia coli is shifted from 37 to 10°C, there is a lag period of cell growth for approximately 4 h before cell growth is resumed (
      • Jones P.G.
      • VanBogelen R.A.
      • Neidhardt F.C.
      ). Cold-shock proteins are induced during this lag period, and a total of 17 cold-shock proteins have been identified by two-dimensional gel electrophoresis (
      • Jones P.G.
      • VanBogelen R.A.
      • Neidhardt F.C.
      ,
      • Jones P.G.
      • Krah R.
      • Tafuri S.R.
      • Wolffe A.P.
      ,
      • Lee S.J.
      • Xie A.
      • Jiang W.
      • Etchegaray J.-P.
      • Jones P.G.
      • Inouye M.
      ,
      • Jones P.G.
      • Mitta M.
      • Kim Y.
      • Jiang W.
      • Inouye M.
      ).
      CspA is the major cold-shock protein in Escherichia coli, and its production reaches more than 10% of total cellular protein synthesized during the cold shock response (
      • Goldstein J.
      • Pollitt N.S.
      • Inouye M.
      ). It has been suggested that the expression of cspA is regulated at the level of transcription (
      • Goldstein J.
      • Pollitt N.S.
      • Inouye M.
      ,
      • Tanabe H.
      • Goldstein J.
      • Yang M.
      • Inouye M.
      ,
      • Jiang W.
      • Jones P.G.
      • Inouye M.
      ,
      • Etchegaray J.P.
      • Jones P.G.
      • Inouye M.
      ) as well as at the level of post-transcription (
      • Brandi A.
      • Pietroni P.
      • Gualerzi C.O.
      • Pon C.L.
      ,
      • Goldenberg D.
      • Azar I.
      • Oppenheim A.B.
      ). CspA consists of 70 amino acid residues with a molecular mass of 7.4 kDa (
      • Goldstein J.
      • Pollitt N.S.
      • Inouye M.
      ). The three-dimensional structure of CspA has been determined both by NMR (
      • Newkirk K.
      • Feng W.
      • Jiang W.
      • Tejero R.
      • Emerson S.D.
      • Inouye M.
      • Montelione G.
      ) and x-ray crystallography (
      • Schindelin H.
      • Jiang W.
      • Inouye M.
      • Heinemann U.
      ). CspA consists of five anti-parallel β sheets, which form a β-barrel structure. RNA binding motifs RNP1 (Phe18-Gly19-Phe20) and RNP2 (Phe31-Val32-His33) are located on β-2- and β-3-sheet, respectively (
      • Newkirk K.
      • Feng W.
      • Jiang W.
      • Tejero R.
      • Emerson S.D.
      • Inouye M.
      • Montelione G.
      ). Seven of eight aromatic residues are distributed on the same surface of the CspA molecule, suggesting potential interactions between CspA and nucleic acids through hydrophobic interactions. Indeed, it has been shown that CspA interacts with a 24-base ssDNA
      The abbreviations used are: ssDNA
      single-stranded DNA
      ssRNA
      single-stranded RNA
      dsDNA
      double-stranded DNA
      CSD
      cold-shock domain
      UTR
      untranslated region.
      (
      • Newkirk K.
      • Feng W.
      • Jiang W.
      • Tejero R.
      • Emerson S.D.
      • Inouye M.
      • Montelione G.
      ).
      Interestingly, the amino acid sequence of CspA shows 43% identity to the “cold-shock domain” (CSD) of the eukaryotic Y-box protein family. All Y-box proteins characterized to date show varied nucleic acid binding activities (for review, see
      • Wolffe A.P.
      • Tafuri S.
      • Ranjan M.
      • Familari M.
      ,
      • Wolffe A.P.
      ,
      • Sommerville J.
      • Ladonery M.
      ). Some Y-box proteins bind to DNA to affect gene transcription, DNA replication, and DNA repair, while other Y-box proteins interact with mRNA to affect translational efficiency. Furthermore, it has been demonstrated that the CSD is directly responsible for nucleic acid binding activities (
      • Tafuri S.R.
      • Wolffe A.P.
      ,
      • Bouvet P.
      • Matsumoto K.
      • Wolffe A.P.
      ).
      In this paper, we demonstrate that CspA cooperatively binds to ssRNA. A certain length of ssRNA (more than 74 bases) is required for the cooperative binding. CspA binding to ssRNA displays a low sequence specificity. A minimal concentration of CspA required for RNA binding is 2.7 × 10−5M, which is sufficiently lower than the cellular concentration of CspA in cold-shocked cells (10−4M). In the presence of CspA, ssRNA became much more sensitive to ribonucleases, and furthermore, secondary structures formed in the RNA molecule that are resistant to ribonucleases became highly susceptible to ribonuclease digestion. These results indicate that CspA binding to RNA results in unfolding RNA secondary structures. We propose that CspA functions at low temperatures as an RNA chaperone, which binds to mRNAs with a low sequence specificity to prevent the formation of secondary structures in mRNAs. Such a function is considered to be important for efficient translation of mRNAs at low temperatures.

      EXPERIMENTAL PROCEDURES

       Reagents and Enzymes

      All NTPs were purchased from Boehringer Mannheim. [α-32P]CTP (400 Ci/mmol) was purchased from Amersham Corp. SP6 RNA polymerase was from New England Biolabs. T7 RNA polymerase was from Promega. RNase inhibitor was purchased from Boehringer Mannheim. All restriction enzymes were from New England Biolabs.

       Purification of CspA

      CspA was purified as described previously (
      • Chatterjee S.
      • Jiang W.
      • Emerson S.D.
      • Inouye M.
      ).

       Preparation of RNA Fragments

      RNA fragments (fragments A-F; see Fig. 1, Fig. 2) were obtained as described previously (
      • Lee C.
      • Hurwitz J.
      ). Basically, RNA fragment A was prepared by transcribing the pGEMI (Promega) plasmid cut with PvuII using SP6 RNA polymerase. Fragment B was prepared by transcribing the plasmid pSP65 cut with XbaI using SP6 RNA polymerase. Fragment C was made by transcription with SP6 RNA polymerase of pGEM3cs− plasmid cut with HaeII (
      • Scheffner M.
      • Knippers R.
      • Stahl H.
      ). Fragment D was made by transcription with SP6 RNA polymerase of pGEM3 plasmid cut with SacI. Fragment E was prepared by transcribing pGEM3cs− cut with RsaI using T7 RNA polymerase. Fragment F was synthesized by transcribing pGEM3 cut with AccI using T7 RNA polymerase.
      Figure thumbnail gr1
      Fig. 1CspA binding to a partial double-stranded RNA. RNA was prepared by annealing nonradioactive fragment A and 32P-labeled fragment B as shown at the top. The RNA binding assay was carried out as described in “Experimental Procedures.” Lane 1, RNA without heat denaturing; lane 2, RNA with heat denaturing; lanes 3-6, increasing amounts of CspA (0.5, 1, 2, and 3 μg, respectively) was added to RNA; lanes 7 and 8, BSA (1 and 3 μg, respectively).
      Figure thumbnail gr2
      Fig. 2CspA binding to different heat-denatured single-stranded RNA templates. All ssRNA fragments were synthesized by in vitro transcription as described under “Experimental Procedures.” RNA sequences of fragments C, D, E, and F are shown at the top. The RNA binding assay was carried out with 3 μg of CspA as described in . RNA fragments were heat-denatured. Lanes 1 and 2, fragment A; lanes 3 and 4, fragment B; lanes 5 and 6, fragment C; lanes 7 and 8, fragment D; lanes 9 and 10, fragment E; lanes 11 and 12, fragment F.
      The RNA fragments, from A1 to A6, were synthesized in the way shown in Fig. 3A. Briefly, fragment A1 was made by transcribing pGEMI cut with HindIII by SP6 RNA polymerase. Fragment A2 was made by transcribing pGEMI cut with EcoRI by SP6 RNA polymerase. Fragment A3 was made as follows. pGEMI was digested by HindIII and EcoRI, followed by Klenow fill-in and self-ligation. The resulting plasmid was cut with PvuII and transcribed by SP6 RNA polymerase. Fragment A4 was prepared by cutting pGEMI with XbaI and EcoRI, followed by self-ligation. The resulting plasmid was cut with PvuII and transcribed by SP6 RNA polymerase. Fragment A5 was made as follows. The 60-base pair fragment from pTH1 digested with SacI (
      • Musso R.E.
      • Hodam T.
      ) was inserted into the SacI site of pGEMI. The resulting plasmid was digested by HindIII and transcribed by SP6 RNA polymerase. Fragment A6 was made by transcribing the same plasmid used for A5 cut with PvuII using SP6 RNA polymerase. All labeled RNA fragments were labeled with [α-32P]CTP. In vitro transcription was carried out using the protocols provided by the manufacturer.
      Figure thumbnail gr3
      Fig. 3Length and sequence requirements for CspA binding to RNA. A, RNA fragments A1-A6 were prepared as described. The construction of each DNA fragment is described under “Experimental Procedures.” The length of each fragment is indicated. B, binding assay of all RNA fragments described in A. RNA fragments used in the binding assay and the addition of CspA are indicated above each lane. 1 μg of CspA was added in lane 8, and 3 μg of CspA was added in all other reactions.

       Preparation of DNA Fragments

      dsDNA fragment A′ was made as follows. pGEMI was cut with HindIII and end-labeled by Klenow filled-in with [α-32P]dATP. The linear DNA fragment was then cut with PvuII followed by gel purification of the resulting 88-base pair fragment A′. As a result, only the minus strand DNA was labeled. The labeled minus strand DNA was then obtained by heat-denaturing the labeled dsDNA fragment A′. The plus strand DNA was labeled with [γ-32P]ATP by T4 kinase after heat-denaturing dsDNA fragment A′. This procedure also labeled the 3′-end-labeled minus strand DNA at its 5′-end as well.

       RNA and DNA Binding Assay

      RNA binding assay was carried out in a 15-μl reaction containing 10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 50 mM KCl, 7.4% glycerol, 1 × 106 cpm RNA template, and purified protein. Mixtures were incubated on ice for 20 min and then loaded on a 8% polyacrylamide gel. The running buffer was 1 × TBE (50 mM Tris-Cl, 1 mM EDTA, 50 mM boric acid). Electrophoresis was carried out at 130 V for 3.5 h at 4°C. The gel was vacuum-dried and visualized by autoradiography. The DNA binding assay was carried out in exactly the same manner.

       Ribonuclease Assay

      32P-Labeled RNA substrates were incubated with 3 μg of purified CspA in the binding buffer for 15 min on ice. The volume of the final reaction mixture was 15 μl. RNase A or RNase T1 was added to the RNA-substrate solution, and the mixture was kept on ice for 15 min in the case of fragment B or at 15°C for 5 min in the case of the 5′-UTR of the cspA mRNA. Half of the mixture was loaded onto a 8% acrylamide gel. In the case of fragment B, the remaining half was heat-denatured for 5 min and subjected to 7 M urea, 10% acrylamide gel electrophoresis. The region corresponding to the 5′-UTR from +1 to +142 of the cspA mRNA was placed under T7 promoter using polymerase chain reaction, and in vitro transcription of the polymerase chain reaction-amplified template was carried out in the presence of [α-32P]CTP according to the protocol provided by the manufacturer.

      RESULTS

       CspA Binds to RNA

      To examine whether CspA can bind to RNA, we first utilized a partially double-stranded RNA as shown in Fig. 1. RNA fragments A and B were synthesized by in vitro transcription, and fragment B was labeled with [α-32P]CTP. When fragments A and B were annealed, the product had a 29-base pair double-stranded region with single-stranded regions at the 5′-end (28 bases) and the 3′-end (41 bases) on fragment A and at the 5′-end of fragment B (9 bases) (Fig. 1). Using the annealed complex, gel retardation experiments were carried out with different amounts of the purified CspA as shown in Fig. 1. When the annealed complex that migrated at position b (lane 1) was heat-denatured and quickly cooled on ice before loading on the gel, 32P-labeled fragment B was separated from nonradioactive fragment A and migrated at position a (lane 2). When increasing amounts of CspA were added to the annealed complex, its migration became retarded shifting from position b to c (lanes 3-6). As a control, the same amounts of bovine serum albumin (BSA) were added, where no shifted band was observed (lanes 7 and 8). It is interesting to note that the amounts of CspA used in this assay were rather high; 0.5, 1, 2, and 3 μg of CspA was added in a final 15-μl reaction mixture in lanes 3, 4, 5, and 6 in Fig. 1, respectively. In addition, the shift caused by CspA was discontinuous as the band was suddenly shifted from position b to c at 2 μg of CspA (lane 5) and more notably at 3 μg (lane 6). The CspA concentration in the reaction mixture at 3 μg was calculated to be 2.7 × 10−5M. At 1 μg (lane 4), no shift was observed. This result suggests a cooperative binding of CspA to RNA.

       Requirements of CspA Binding to Heat-denatured ssRNA

      In order to find out which strand in the annealed complex (A or B) is responsible for CspA binding, fragments A and B were labeled with [α-32P]CTP separately. Both fragments were heat-denatured before the addition of CspA. As shown in Fig. 2, the same amount of CspA used in Fig. 1 was able to cause a mobility shift of fragment A (lanes 1 and 2) but not fragment B (lanes 3 and 4). To investigate whether CspA binding to heat-denatured ssRNA has any sequence specificity, four other RNA fragments (fragments C, D, E, and F) shown in Fig. 2 were synthesized by in vitro transcription. They were also labeled with [α-32P]CTP, and a gel retardation assay was carried out as described above. Among the new RNA fragments, fragments C (lanes 5 and 6) and E (lanes 9 and 10) were found to form complexes with CspA, as judged from their retarded migration in the gel. In contrast, CspA did not bind to fragments D (lanes 7 and 8) and F (lanes 11 and 12).
      Those fragments incapable of CspA binding (fragments B, D, and F) were shorter in length (39, 52, and 46 bases, respectively) than those fragments to which CspA was able to bind (fragments A, C, and F are 98, 107, and 103 bases, respectively). Such a size dependence of RNA binding has also been observed with FRGY2, a Xenopus Y-box protein (
      • Bouvet P.
      • Matsumoto K.
      • Wolffe A.P.
      ). In order to examine the effect of the length of ssRNA on CspA binding, a number of deletions or insertions were created based on fragment A as shown in Fig. 3A. The results of a gel retardation assay performed with these fragments are shown in Fig. 3B. CspA could not bind to fragment A1 (32 nucleotides) and A2 (56 nucleotides), which resulted from deleting 66 (fragment A1) and 42 bases (fragment A2) from the 3′-end of fragment A (lanes 1 and 2 and lanes 3 and 4, respectively). When an internal 46-base region was removed, leaving the 42-base 3′-end sequence that was deleted in fragment A2, a minor retardation was observed with the resulting fragment A3 (52 nucleotides; lanes 5 and 6). When the internal deletion became shorter by 22 bases, CspA was able to bind well to the resulting fragment A4 (74 nucleotides), causing a supershift (lanes 12 and 13) as observed with fragment A (lanes 7-9). These results demonstrate that CspA binding to ssRNA requires a certain length of the target RNA. Fragment A4 is the shortest template that is capable of CspA binding.
      Based on the fact that CspA caused a minor retardation of fragment A3, it was postulated that CspA binding to heat-denatured ssRNA may also have some sequence specificity that exists in the 3′-end region of fragment A. To test this possibility, a 60-base sequence from pTH1 (
      • Musso R.E.
      • Hodam T.
      ) digested with SacI was added to the 3′-end of fragment A2, creating fragment A5 of 116 bases in length, which is longer than fragment A (see Fig. 3A). CspA was found to hardly bind to fragment A5 (Fig. 3B, lanes 14 and 15). This result indicates that there is a preferable CspA binding sequence located at the 3′-end region of fragment A in addition to the length required for CspA binding. The 56-base 5′-end sequence of fragment A (fragment A2) does not contain such a sequence as required for the CspA binding. Note that the 60-base sequence added to fragment A2 did not have any negative effect on the CspA binding, since the insertion of the sequence at the middle of fragment A resulting in fragment A6 did not inhibit the supershift of fragment A6 (lanes 16 and 17). When the nucleotide sequences of fragments A, C, and E, all of which showed specific interaction with CspA, were compared with each other, a 15-base sequence common to all three fragments was found: CUAUAGUGAGUCGUA, from base 66 to 80 in fragment A and from base 25 to 39 in fragment C and CUAUAGUGucUCcUA in fragment E from base 25 to 39 (Fig. 2; mismatched bases in lowercase type). This region in fragment A (8 nucleotides at a time) was randomized, and the SELEX method (
      • Tuerk C.
      • Gold L.
      ) was applied to screen the most preferable sequences for CspA binding. However, this approach was unsuccessful, since CspA was still able to bind to the fragments with the randomized sequences.

       Cooperative Binding of CspA to Heat-denatured ssRNA

      We further examined the cooperation of CspA binding to ssRNA using fragment A. As shown in Fig. 4, when 1 μg of CspA was added to 32P-labeled fragment A in a 15-μl reaction mixture, a minor retardation was observed (compare lane 2 with lane 1). However, when 3 μg of CspA were added in the same final volume (15 μl) as in the case of lane 2, the band was supershifted (lane 3). This supershifted band did not change its migration even when a 24-fold excess of nonradioactive fragment A was added to the reaction mixture (lane 4), suggesting that all RNA molecules added were shifted. A similar result was obtained when a 5-fold (lane 6) or 24-fold (not shown) excess of 32P-labeled fragment A was added, indicating that CspA binding to ssRNA is cooperative. This cooperative binding is only dependent on the CspA concentration (2.7 × 10−5M) and not dependent on the molar ratio of CspA to RNA. This was demonstrated by the following experiment; when 1 μg of CspA, which could not cause the supershift in lane 2, was used in a 5-μl reaction mixture (one-third of the reaction volume used in the experiment in lane 2), the supershift was observed as shown in lane 5. In this reaction, the CspA concentration became identical to that with 3 μg of CspA in the 15-μl reaction mixture. This critical concentration can be calculated to be approximately 2.7 × 10−5M. Note that the concentration of CspA inside the cold-shocked cell is estimated at 10−4M (
      • Goldstein J.
      • Pollitt N.S.
      • Inouye M.
      ), which is higher than the critical concentration obtained above.
      Figure thumbnail gr4
      Fig. 4Cooperative RNA binding of CspA. RNA binding assay was carried out as described under “Experimental Procedures.” The relative amounts of CspA (1 relative unit (U) = 1 μg), 32P-labeled ssRNA fragment A (1 relative unit = 1 × 106 cpm and 0.045 nmol), and nonradioactive fragment A (a total of 1.08 mmol) used are shown above the gel. The reaction volumes are also listed there. The labeled RNA was first added to the reaction buffer, followed by the addition of CspA. The cold fragment A was first mixed by the labeled fragment followed by the addition of CspA. Lane 1, labeled fragment A alone; lanes 2 and 3, 1 and 3 μg of CspA were added, respectively; lane 4, a 24-fold excess of cold fragment A was added to the labeled fragment A, followed by the addition of 3 μg of CspA; lane 5, 1 μg of CspA was added, but the reaction volume was reduced from 15 to 5 μl; lane 6, a 5-fold excess of labeled fragment A was used in the binding assay with 3 μg of CspA.

       CspA Binding to ssDNA

      It is known that most Y-box proteins that contain a cold-shock domain are capable of nucleic acid binding (for review, see
      • Wolffe A.P.
      • Tafuri S.
      • Ranjan M.
      • Familari M.
      ,
      • Wolffe A.P.
      ,
      • Sommerville J.
      • Ladonery M.
      ). Some of them bind to RNA, while others bind to DNA. Therefore, it was interesting to investigate whether CspA is also able to bind to ssDNA. To test this, a dsDNA (fragment A′) was generated from the template DNA used to generate fragment A RNA (see “Experimental Procedures”). Fragment A′ was labeled by [α-32P]dATP at the 3′-end of the minus strand by Klenow fill-in. As shown in Fig. 5, CspA failed to bind the dsDNA fragment A′ (lanes 1 and 2). When this dsDNA was heat-denatured, the labeled minus strand was separated from the plus strand and migrated at position b (lane 3). The addition of CspA caused a minor shift of the minus strand (compare lane 4 with lane 3). As a result of a minor shift of the minus strand in the presence of CspA, it migrated at the same position as double-stranded fragment A′ at position c. Therefore, we subsequently performed the following experiment. In addition to the 3′-end labeling, the 5′-ends of both strands were also labeled with [γ-32P]ATP by T4 kinase. When this dsDNA was heat-denatured, it yielded two distinct bands at positions a and b (lane 5). The band at position b corresponded to the minus strand according to lane 3, where only the minus strand was labeled. Thus, the band at position a corresponded to the plus strand. When CspA was added to the heat-denatured fragment A′, two bands at positions c and d appeared (lane 6). The band at position c appears to correspond to the bands observed in lane 4, resulted from a minor shift of the minus band caused by CspA. Therefore, the band at position d was caused by a supershift, which resulted from the CspA binding to the plus strand, the band at position a in lane 5. These results indicate that CspA is able to bind to both ssRNA and ssDNA with the same sequence specificity. However, CspA cannot bind to dsDNA.
      Figure thumbnail gr5
      Fig. 5CspA binding to DNA. Preparation of the dsDNA fragment A′, which served as the template for ssRNA fragment A, was carried out as described under “Experimental Procedures.” First, only the minus strand was labeled at the 3′-end by Klenow filled-in with [α-32P]dATP (lanes 1-4). For lanes 5 and 6, both minus and plus strands were labeled at their 5′-ends with T4 kinase with [γ-32P]ATP after heat-denaturing the Klenow-labeled fragment A′. The DNA binding assay was carried out as described under “Experimental Procedures.” 3 μg of CspA was used. Lane 1, dsDNA fragment A′ alone; lane 2, fragment A′ with 3 μg of CspA; lane 3, heat-denatured fragment A′ containing the 32P-labeled minus strand; lane 4, heat-denatured fragment A′ as in lane 3 was incubated with 3 μg of CspA; lane 5, both plus and minus strand DNA were labeled by T4 kinase with [γ-32P]ATP after heat denaturing; lane 6, plus and minus strand DNA as in lane 5 were incubated with 3 μg of CspA.

       CspA Functions as an RNA Chaperone

      “RNA chaperone” refers to proteins that prevent RNA misfolding and resolve misfolded RNAs, thereby ensuring their biological functions (
      • Herschlag D.
      ). Based on RNA binding characteristics of CspA presented above, which include low sequence specificity, cooperative binding, and low binding affinity to RNA (2.7 × 10−5M), it is speculated that CspA may function as an RNA chaperone.
      First, we attempted to examine whether CspA interacts with secondary structures in RNA. For this purpose, we tested the effect of CspA on non-heat-denatured fragment B. As shown in Fig. 2, CspA does not interact with heat-denatured fragment B. When fragment B was analyzed by acrylamide gel electrophoresis without heat denaturation, two additional bands appeared at positions b and d (Fig. 6A, lane 1), in addition to the band at position a, which is considered to be the same as heat-denatured fragment B (lane 2). RNAs at positions b and d are likely to be dimers and trimers of fragment B, respectively, which resulted from intermolecular hybrid formation using a number of short inverted repeats in fragment B. Interestingly, when increased amounts of CspA were added to non-heat-denatured fragment B (0.2, 0.5, 1, 2, and 3 μg for lanes 3, 4, 5, 6, and 7, respectively), band b shifted to c and band d disappeared with concomitant appearance of band e. Band a hardly moved as previously shown in Fig. 2, and the addition of the same amounts of BSA (1, 2 and 3 μg for lanes 8, 9, and 10, respectively) did not change the pattern. In order to confirm that CspA did not modify the RNA molecule, the same reaction mixtures used above were loaded on a 7 M urea-10% acrylamide gel, and all gave a single band at the same position as shown from lane 1 to 10 in Fig. 6B. These results indicate that CspA can also interact with certain structures formed by intermolecular interaction between RNA molecules.
      Figure thumbnail gr6
      Fig. 6Interaction of CspA with secondary structures formed by fragment B. A, interaction of CspA with fragment B. The binding assay is described under “Experimental Procedures.” Lane 1, fragment B alone without heat-denaturation; lane 2, heat-denatured fragment B; lanes 3-7, increasing amounts of CspA (0.2, 0.5, 1, 2, and 3 μg of CspA, respectively) were added to undenatured fragment B; lanes 8-10, increasing amounts of BSA (1, 2, and 3 μg of BSA, respectively) were added to undenatured fragment B; lane 11, digestion of the undenatured fragment B by RNase A (1 μg); Lane 12, digestion of undenatured fragment B by RNase A in the presence of CspA; lane 13, digestion of undenatured fragment B by RNase A in the presence of BSA. B, the same reaction mixtures used in A were subjected to 7 M urea, 10% acrylamide gel electrophoresis.
      Next, the ribonuclease assay was carried out to examine whether CspA binding to RNA secondary structure facilitates the RNA susceptibility to ribonuclease A. In the absence of CspA, fragment B was degraded by RNase A into two major products, indicated as X and Y (Fig. 6B, lane 11). However, when 3 μg of CspA was added to fragment B followed by the addition of RNase A, product Y was further degraded and migrated to the bottom of the gel (not shown in Fig. 6B), while product X was unchanged (lane 12). Note that CspA itself did not contain any ribonuclease activity (Fig. 6B, lanes 3-10), this result indicates that fragment B becomes more sensitive to the RNase A in the presence of CspA. When the equal amount of BSA was added with RNase A, no enhancement of RNase A digestion was observed (lane 13).
      On the basis of the results obtained above, we next used a longer RNA from the cspA mRNA, which has a long 5′-UTR consisting of 159 bases (
      • Goldstein J.
      • Pollitt N.S.
      • Inouye M.
      ). Out of the 159-base 5′-UTR, the RNA from +1 to +142 was produced under a T7 promoter in the presence of [α-32P]CTP. As shown in Fig. 7, this RNA migrated at position h in 8% acrylamide gel electrophoresis (lane 1). When CspA (3 μg in the final volume of 15 μl) was added, it was supershifted to position j (lane 2). When 0.5 μg of RNase A was added to the RNA and the mixture was incubated for 5 min at 15°C, five RNase A-resistant bands appeared at positions a, b, c, d, and e (lane 3). As the concentration of RNase was reduced to 0.25 and 0.125 μg, the density of band e increased, while the density of band d decreased (lanes 4 and 5, respectively). Further reduction of RNase A to 0.063 μg resulted in substantially long RNA fragments at positions f and g without producing shorter fragments at positions c, d, and e (lane 6). However, when 3 μg of CspA was added before the addition of RNase A, RNase-resistant bands at positions c, d, and e disappeared as shown in lanes 7, 8, and 9, respectively, leaving mainly bands a and b. At the lowest RNase A concentration in the presence of CspA (lane 10), a broad, supershifted band at position i as well as very broad bands around position f were observed.
      Figure thumbnail gr7
      Fig. 7Enhancement of RNase activities by CspA. Ribonuclease assays were carried out as described under “Experimental Procedures.” The RNA substrate used was derived from the 5′-UTR of the cspA mRNA, and the region from +1 to +142 was placed under a T7 promoter. The RNA was labeled with [α-32P]CTP. The RNase assays were carried out for 5 min at 15°C. The amounts of RNases used are shown at the top in micrograms. When CspA was used (indicated by +), 3 μg was added. The final volume was 15 μl. Lane 1, RNA substrate alone; lane 2, RNA plus CspA (3 μg); lanes 3-6, 0.5, 0.25, 0.125, and 0.063 μg of RNase A, respectively, were added; lanes 7-10, the same as lanes 3-6, respectively, except that CspA (3 μg) was added before adding RNase A; lanes 11-13, 0.4, 0.2, and 0.1 μg of RNase T1, respectively, were added; lanes 14-16, the same as lanes 11-13, respectively, except that 3 μg of CspA was added to each reaction. Immediately after the reaction, the reaction mixture was applied to an 8% acrylamide gel, and electrophoresis was carried out at 150 V at 4°C. The MspI digest of pBR322 DNA was used as size markers. The numbers with a bar indicate lengths of double-stranded DNA in base numbers.
      Similar results were obtained with RNase T1; in the absence of CspA, digestion of the RNA with 0.4, 0.2, and 0.1 μg of RNase T1 (lanes 11, 12, and 13, respectively) resulted in the formation of a stable RNA fragment at positions o and l. In the presence of 0.4 μg of RNase T1, part of the band at position o was further digested to produce a band at position m. When CspA was added into the reaction mixture, the band at position o completely disappeared, leaving a very stable band at position l in all concentrations of RNase T1 (lanes 14-16). It should be noted that in the presence of 0.4 μg of RNase T1 (lane 14), the background in the region higher than band l was very low, and a faint band at position n observed in the absence of CspA (lanes 11-13) was not detected. Instead, another stable band at position k appeared. It should also be noted that CspA itself did not have RNase activity as shown in lane 2 of Fig. 7, where RNA was quantitatively supershifted.
      The results described above demonstrate that CspA indeed enhances susceptibility of RNA to RNases; in the case of RNase A, it was estimated that CspA stimulates the RNase activity approximately 10-fold (compare lane 9 with lane 3 in Fig. 7). RNase-resistant bands observed in Fig. 7 are considered to be due to secondary structures formed in the RNA molecule. Fig. 8 shows a predicted secondary structure of the 5′-UTR of the cspA mRNA from +1 to +142. In such a structure, loop regions are sensitive to RNase cleavage, which generates a number of fragments of different sizes like bands a, b, c, d, and e in RNase A digestion and k, l, m, and n in RNase T1 digestion (cleavage at G residues circled in Fig. 8). The fact that these bands disappeared in the presence of CspA indicates that CspA destabilizes the secondary structures existing in RNA molecules to further digest them into smaller fragments. We, therefore, propose that CspA functions as an RNA chaperone.
      Figure thumbnail gr8
      Fig. 8Proposed secondary structure of the cspA 5′-UTR from +1 to +142. The secondary structure is from previous studies (
      • Jiang W.
      • Jones P.G.
      • Inouye M.
      ). Those G residues susceptible to RNase T1 are circled.

      DISCUSSION

      CspA shows 43% identity with the CSD of eukaryotic Y-box proteins that have RNA or DNA binding activities (for review, see
      • Wolffe A.P.
      • Tafuri S.
      • Ranjan M.
      • Familari M.
      ,
      • Wolffe A.P.
      ,
      • Sommerville J.
      • Ladonery M.
      ). In previous studies, it has been suggested that CspA binds to DNA and is involved in gene regulation (
      • Jones P.G.
      • Krah R.
      • Tafuri S.R.
      • Wolffe A.P.
      ,
      • La Teana A.
      • Brandi A.
      • Falconi M.
      • Spurio R.
      • Pon C.C.
      • Gualerzi C.O.
      ). In addition, the three-dimensional structure of CspA strongly indicates that CspA is a nucleic acid-binding protein (
      • Newkirk K.
      • Feng W.
      • Jiang W.
      • Tejero R.
      • Emerson S.D.
      • Inouye M.
      • Montelione G.
      ). In this paper, we have demonstrated that CspA binds cooperatively to heat-denatured ssRNA as well as ssDNA with a low sequence specificity, while CspA cannot bind to dsDNA. A certain length of ssRNA (74 nucleotides) and the minimal concentration of CspA at 2.7 × 10−5M are required for the supershift of the RNA caused by CspA. Since the CspA concentration in cold-shocked cells was estimated to be 10−4M (
      • Goldstein J.
      • Pollitt N.S.
      • Inouye M.
      ), CspA in cold-shocked cells is able to cooperatively bind ssRNA. In addition to CspA binding to heat-denatured ssRNA, we demonstrated that CspA is able to interact with RNA secondary structures formed either intermolecularly or intramolecularly. As a result, RNA becomes more susceptible to RNase digestion, and RNase-resistant double-stranded RNA fragments disappear. All of these properties of CspA support a notion that CspA functions as an RNA chaperone, which weakly interacts with RNA to affect RNA folding. Such a property is considered to be essential for efficient translation of mRNA at low temperatures (
      • Jones P.G.
      • Inouye M.
      ). Recently, a new cold-shock protein, CsdA, was found, which is exclusively associated with ribosomes and has double-stranded RNA unwinding activity (
      • Jones P.G.
      • Mitta M.
      • Kim Y.
      • Jiang W.
      • Inouye M.
      ). It is possible that CsdA unwinds stable secondary structures formed in mRNAs at low temperatures and that CspA, as an RNA chaperone, binds to unwound mRNAs. Alternatively, CspA first binds to RNA to assist the CsdA activity to unwind folded RNA. The weak binding of CspA to mRNAs, however, probably does not interfere with the translational activity of ribosomes, but rather may facilitate mRNA translation.
      It is possible that CspA may share some functional similarities with a Xenopus Y-box protein, FRGY2, which is known to mask mRNA in oocytes (
      • Richter J.D.
      • Smith L.D.
      ,
      • Ranjan M.
      • Tafuri S.R.
      • Wolffe A.P.
      ,
      • Bouvet P.
      • Wolffe A.P.
      ). FRGY2, however, binds to a 95-base RNA having a specific 6-base sequence at a Kd value of 2 × 10−8M, 1000-fold lower than the concentration of CspA required for its binding to RNA (
      • Bouvet P.
      • Matsumoto K.
      • Wolffe A.P.
      ). Its CSD of 77 residues by itself was shown to still retain a reasonably low Kd value (about 10−7M) with the sequence-specific RNA binding ability. In contrast, our attempts to determine the specific CspA-binding sequences by SELEX (
      • Tuerk C.
      • Gold L.
      ) were unsuccessful. At present, it is not known what accounts for the differences in RNA binding and sequence specificity between CspA and the FRGY2 CSD. One can speculate that FRGY2 acquired the sequence-specific binding ability as well as the ability to bind to RNA more tightly during evolution. Also, other domains of FRGY2 may contribute to the higher affinity to RNA as well as the sequence specificity for its binding. It is important to note that FRGY2 is a major component of masked maternal mRNA in Xenopus oocytes, while CspA is speculated in the present paper to be an RNA chaperone to facilitate mRNA translation at low temperatures. Nevertheless, these two proteins also share some similar properties such as the size requirement of RNA and the ability to bind nonspecific RNA (
      • Bouvet P.
      • Matsumoto K.
      • Wolffe A.P.
      and this study). It has been speculated that nonspecific binding of CspA to RNA may occur through the hydrophobic interactions between nucleotide bases and the seven aromatic side chains regularly aligned on the surface of the CspA molecule (
      • Newkirk K.
      • Feng W.
      • Jiang W.
      • Tejero R.
      • Emerson S.D.
      • Inouye M.
      • Montelione G.
      ). Mutations of these aromatic residues were indeed shown to affect the interaction between RNA and proteins for CspB from Bacillus subtilis (
      • Schroder K.
      • Graumann P.
      • Schnuchel A.
      • Holak T.A.
      • Marahiel M.A.
      ) and for FRGY2 (
      • Bouvet P.
      • Matsumoto K.
      • Wolffe A.P.
      ).

      Acknowledgments

      We thank Li Fang for helpful assistance and discussions throughout this work. We also thank Drs. Guy Montelione and Beate Schwer for advice.

      REFERENCES

        • Jones P.G.
        • Inouye M.
        Mol. Microbiol. 1994; 11: 811-818
        • Jones P.G.
        • VanBogelen R.A.
        • Neidhardt F.C.
        J. Bacteriol. 1987; 169: 2092-2095
        • Jones P.G.
        • Krah R.
        • Tafuri S.R.
        • Wolffe A.P.
        J. Bacteriol. 1992; 174: 5798-5802
        • Lee S.J.
        • Xie A.
        • Jiang W.
        • Etchegaray J.-P.
        • Jones P.G.
        • Inouye M.
        Mol. Microbiol. 1994; 11: 833-839
        • Jones P.G.
        • Mitta M.
        • Kim Y.
        • Jiang W.
        • Inouye M.
        Proc. Natl. Acad. Sci. 1995; 93: 76-80
        • Goldstein J.
        • Pollitt N.S.
        • Inouye M.
        Proc. Natl. Acad. Sci. 1990; 87: 283-287
        • Tanabe H.
        • Goldstein J.
        • Yang M.
        • Inouye M.
        J. Bacteriol. 1992; 174: 3867-3873
        • Jiang W.
        • Jones P.G.
        • Inouye M.
        J. Bacteriol. 1993; 175: 5824-5828
        • Etchegaray J.P.
        • Jones P.G.
        • Inouye M.
        Genes Cells. 1996; 1: 171-178
        • Brandi A.
        • Pietroni P.
        • Gualerzi C.O.
        • Pon C.L.
        Mol. Microbiol. 1996; 19: 231-240
        • Goldenberg D.
        • Azar I.
        • Oppenheim A.B.
        Mol. Microbiol. 1996; 19: 241-248
        • Newkirk K.
        • Feng W.
        • Jiang W.
        • Tejero R.
        • Emerson S.D.
        • Inouye M.
        • Montelione G.
        Proc. Natl. Acad. Sci. 1994; 91: 5114-5118
        • Schindelin H.
        • Jiang W.
        • Inouye M.
        • Heinemann U.
        Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 5119-5123
        • Wolffe A.P.
        • Tafuri S.
        • Ranjan M.
        • Familari M.
        New Biol. 1992; 4: 290-298
        • Wolffe A.P.
        BioEssays. 1994; 16: 245-251
        • Sommerville J.
        • Ladonery M.
        FASEB J. 1996; 10: 435-443
        • Tafuri S.R.
        • Wolffe A.P.
        New Biol. 1992; 4: 349-359
        • Bouvet P.
        • Matsumoto K.
        • Wolffe A.P.
        J. Biol. Chem. 1995; 270: 28297-28303
        • Chatterjee S.
        • Jiang W.
        • Emerson S.D.
        • Inouye M.
        J. Biochem. (Tokyo). 1993; 114: 663-669
        • Lee C.
        • Hurwitz J.
        J. Biol. Chem. 1992; 267: 4398-4407
        • Scheffner M.
        • Knippers R.
        • Stahl H.
        Cell. 1989; 57: 955-963
        • Musso R.E.
        • Hodam T.
        Gene (Amst.). 1989; 85: 205-207
        • Tuerk C.
        • Gold L.
        Science. 1990; 249: 505-510
        • Herschlag D.
        J. Biol. Chem. 1995; 270: 20871-20874
        • La Teana A.
        • Brandi A.
        • Falconi M.
        • Spurio R.
        • Pon C.C.
        • Gualerzi C.O.
        Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 10907-10911
        • Richter J.D.
        • Smith L.D.
        Nature. 1984; 309: 378-380
        • Ranjan M.
        • Tafuri S.R.
        • Wolffe A.P.
        Genes & Dev. 1993; 7: 1725-1736
        • Bouvet P.
        • Wolffe A.P.
        Cell. 1994; 77: 931-941
        • Schroder K.
        • Graumann P.
        • Schnuchel A.
        • Holak T.A.
        • Marahiel M.A.
        Mol. Microbiol. 1995; 16: 699-708