The Role of RbfA in 16 S rRNA Processing and CellGrowth at Low Temperature in Escherichia coli

Bing Xia†, Haiping Ke†, Ujwal Shinde and Masayori Inouye*

Department of BiochemistryUMDNJ-Robert Wood JohnsonMedical School, 675 Hoes LanePiscataway, NJ 08854-5635USA

RbfA, a 30 S ribosome-binding factor, is a multicopy suppressor of a cold-sensitive C23U mutation of the 16 S rRNA and is required for efficientprocessing of the 16 S rRNA. At 37 8C, DrbfA cells show accumulation ofribosomal subunits and 16 S rRNA precursor with a significantly reducedpolysome profile in comparison with wild-type cells. RbfA is also a cold-shock protein essential for Escherichia coli cells to adapt to low tempera-ture. In this study, we examined its association with the ribosome and itsrole in 16 S rRNA processing and ribosome profiles at low temperature.In wild-type cells, following cold shock at 15 8C, the amount of free RbfAremained largely stable, while that of its 30 S subunit-associated formbecame several times greater than that at 37 8C and a larger fraction oftotal 30 S subunits was detected to be RbfA-containing. In DrbfA cells, thepre-16 S rRNA amount increased after cold shock with a concomitantreduction of the mature 16 S rRNA amount and the formation of poly-somes was further reduced. A closer examination revealed that 30 S ribo-somal subunits of DrbfA cells at low temperature contained primarilypre-16 S rRNA and little mature 16 S rRNA. Our results indicate that thecold sensitivity of DrbfA cells is directly related to their lack of translationinitiation-capable 30 S subunits containing mature 16 S rRNA at low tem-perature. Importantly, when the C-terminal 25 residue sequence wasdeleted, the resulting RbfAD25 lost the abilities to stably associate withthe 30 S subunit and to suppress the dominant-negative, cold-sensitivephenotype of the C23U mutation in 16 S rRNA but was able to suppressthe 16 S rRNA processing defect and the cold-sensitive phenotype of theDrbfA cells, suggesting that RbfA may interact with the 30 S ribosome atmore than one site or function in more than one fashion in assisting the16 S rRNA maturation at low temperature.

q 2003 Elsevier Ltd. All rights reserved.

Keywords: RbfA; 16 S rRNA processing; ribosome; cold shock*Corresponding author

Introduction

Cold shock presumably imposes two major chal-lenges on bacterial cells at the level of translation: astabilization of secondary structures in mRNAsand a transient inactivation of at least a large por-

tion of cellular ribosomes. The translational blockis manifested by a transient diminishment of poly-somes, an increase of 70 S monosomes and a sharpreduction of the synthesis of most cellular proteinsimmediately after a steep temperature downshift.1,2

In Escherichia coli, a family of CspA-like RNAchaperones are induced upon temperature down-shift and may prevent secondary structure for-mation in mRNAs.3 It has been demonstrated thatthese RNA chaperones are essential for cells toadapt to low temperature.4 CspA, the major cold-shock proteins of E. coli, binds single-strandedRNA without any sequence specificity and is ableto render the RNA substrate more susceptible toribonuclease digestion.5 It is conceivable that theseRNA chaperones, dramatically induced upon coldshock to a level of approximately 2 £ 106 molecules

0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.

† B.X. and H.K. contributed equally to this work.Present addresses: B. Xia, Dana-Farber Cancer Institute

and Harvard Medical School, 44 Binney Street, BostonMA 02115, USA; U. Shinde, MRB 631, Department ofBiochemistry and Molecular Biology, Oregon HealthSciences University, 3181 SW Sam Jackson Park Road,Portland, OR 97201, USA.

E-mail address of the corresponding author:inouye@umdnj.edu

Abbreviation used: RbfA, ribosome-binding factor A.

doi:10.1016/S0022-2836(03)00953-7 J. Mol. Biol. (2003) 332, 575–584

per cell,4 would extensively interact with a myriadof different RNAs and prevent or modulate the for-mation of potential secondary structures, thusinfluencing multiple RNA-related cellular pro-cesses, such as translation, transcription and RNAdecay.

Two ribosome-associated factors, CsdA andRbfA, are induced after cold shock and have beenproposed to assist ribosomal functions at lowtemperature.2,6 CsdA was initially identified as asuppressor of a mutation in rpsB encoding thesmall ribosomal subunit protein S2.7 Later, it wasshown to be a cold shock-inducible RNA helicasethat unwinds double-stranded RNA.6 Such anactivity is likely to facilitate either translationinitiation/elongation or ribosomal maturation/assembly, both of which may be impeded by thestabilization of RNA secondary structures at lowtemperature.

RbfA (ribosome-binding factor A) was identifiedas a multicopy suppressor of a C23U mutation inthe 50-end helix of the 16 S ribosomal RNA.8 TheC23U mutation is thought to weaken the stabilityof the 50-end helix in the mature form and lead tomore efficient formation of an alternative helix inthe precursor.9 Interestingly, DrbfA mutant cellsand C23U mutant cells showed identical pheno-types, as both of them are cold sensitive andaccumulate a large amount of ribosomal sub-units.8,9 The same authors also discovered thatRbfA was associated with the 30 S ribosomalsubunit.8 These results strongly indicate that RbfAmight be involved in the maturation of the 16 Sribosomal RNA. Indeed, a more recent studyshows that RbfA is required for efficient 16 SrRNA processing.10

We have previously shown that RbfA is a cold-shock protein and the DrbfA mutant cells display aconstitutive cold-shock response and are unable toadapt to low temperature.2 Recently we havedemonstrated the NMR solution structure of anRbfA to be a typical KH domain known to interactwith RNA.11,12 In this study, to better understandthe biological function of RbfA in cold-shock adap-tation, we carried out comparative studies onseveral relevant cellular functions using bothwild-type and DrbfA mutant strains at both optimaland low temperatures. We found that in DrbfAcells, the amount of mature 16 S rRNA and the for-mation of polysomes were further reduced follow-ing cold shock from their already lower levelscompared with those in wild-type cells. The 30 Sribosomal subunits of DrbfA cells at low tempera-ture contained primarily the 16 S rRNA precursor,indicating that the cold sensitivity of the DrbfAstrain is due to its lack of functional 30 S subunitsat low temperature. We also found that theC-terminal region of RbfA was susceptible to pro-tease digestion and was required for its stableinteraction with the ribosome. Truncated RbfAwith its C-terminal 25 residues deleted (RbfAD25)completely suppressed the accumulation of the16 S rRNA precursor in the DrbfA cells at 37 8C as

well as at 15 8C and also largely rescued the cold-sensitive phenotype of the DrbfA cells. Interest-ingly, RbfAD25 was unable to suppress thedominant-negative, cold-sensitive phenotype ofthe cells expressing the 16 S rRNA C23U mutant,suggesting that RbfA may be multifunctional in16 S rRNA maturation, interacting with 16 S rRNAor some other component of the 30 S subunit atmore than one site.

Results

Association of RbfA with 30 S ribosomes at37 8C and after cold shock

Using an HA-tagged RbfA construct, Dammeland Noller have shown that RbfA specificallyassociates with the 30 S ribosomal subunit in thecell.8 In this work, we first carried out Westernblotting to detect the endogenous RbfA at 37 8Cand 15 8C, using a rabbit polyclonal antibodyagainst full-length RbfA. Cellular componentswere separated by sucrose density-gradient cen-trifugation and then fractionated. Subsequently,the fractions were subjected to Western blottinganalysis to detect RbfA. As shown in Figure 1A,approximately one-third of the total cellular RbfAwas detected in fractions corresponding to 30 Sribosomal subunits, while the remaining two-thirds of the protein was exclusively localized incytosolic fractions at 37 8C.

Shortly after cold shock at 15 8C, cells entered agrowth lag period (acclimation phase) when thecellular content of polysomes drastically decreasedbut 70 S monosomes as well as the 50 S and 30 Ssubunits accumulated (not shown). Six hours aftercold shock, cells became adapted to the low tem-perature and a normal level of polysomes wasreformed (Figure 1B), reflecting the active trans-lation in the cell. At this point, the RbfAs per cellamount had been induced approximately twofold(not shown) and around two-thirds, comparedwith roughly one-third at 37 8C, of its total amountwas associated with the 30 S subunits. In otherwords, the amount of 30 S ribosome-bound RbfAincreased approximately fourfold, six hours aftercold shock and the remaining RbfA pool, whichwas exclusively detected in the cytosolic fractions,was maintained at the same level. At both 37 8Cand 15 8C, no RbfA was detected in any other ribo-somal fractions.

Stoichiometry of RbfA–30 S ribosomalsubunit association

To gain further insight into the fashion in whichRbfA associates with the 30 S ribosomal subunit atboth 37 8C and low temperature, we determinedthe stoichiometry of the association. The 30 S sub-units-containing fractions of sucrose gradients,similar to those shown in Figure 1A and B, wereloaded along with purified RbfA and subjected toWestern blotting analysis. We estimated that lanes

576 RbfA in Cold Shock Adaptation

11 and 15 of Figure 1C, each equivalent to 8 ml ofthe 30 S subunit peak of the 37 8C or cold-shockedcells, contained 4 ng and 12 ng of RbfA, respec-tively. The A260 values of these two fractions were2.06 and 3.17, respectively. Assuming that eachA260 unit corresponds to a ribosome concen-tration of 60 mg/ml and the molecular mass ofthe 30 S subunit is 800 kDa, we were able to calcu-late the molar ratios of RbfA and the 30 S ribosome

subunit in the peak fractions. Under the conditionused, approximately 20% of free 30 S subunits con-tained RbfA at 37 8C and this ratio increased toapproximately 40% at six hours after cold shock.Although the exact ratios may vary under differentconditions, it seems plausible that only a subpopu-lation of free 30 S subunits contain RbfA and that agreater fraction of free 30 S subunits contain RbfAafter cold shock than at 37 8C.

Ribosomal profiles of wild-type and DrbfAmutant cells at 37 8C and after cold shock

It has been shown that DrbfA cells accumulate alarge amount of ribosomal subunits and display16 S rRNA processing defect.8,10 As mutant cellswere viable at 37 8C but sensitive to low tempera-ture, we performed comparative studies on theribosomal profiles of wild-type (MC4100) andDrbfA mutant (BX41) cells at both 37 8C and aftercold shock. When the ribosomal profiles of theabove cells growing at 37 8C were compared itwas found that BX41 cells contained a much largeramount of free subunits and a significantly smalleramount of polysomes than did MC4100 cells grow-ing under the same condition (compare Figure 2Cwith A). The level of the 70 S ribosomes was alsohigher in BX41 cells, albeit to a lesser extent. Fourhours after cold shock, while cold-adapted wild-type cells displayed a normal polysome contentpractically identical with that at 37 8C (Figure 2Band A, respectively), the polysome content ofmutant cells further decreased at 15 8C (compareFigure 2D with C).

16 S Ribosomal RNA processing defects ofDrbfA cells

Next we checked the states of the 16 S rRNA pro-cessing of mutant as well as wild-type cells at both37 8C and low temperature. Total RNAs wereextracted and subjected to primer extension analy-sis, using a primer that anneals to a region approxi-mately 50 bases downstream of the 50-end ofthe mature 16 S rRNA. The results demonstratedthat BX41 cells contained a substantially greateramount of 16 S rRNA precursor likely to be thepre-16 S rRNA processed at the RNase III site115 nt upstream of the 50 end of its mature form10

and concomitantly a much reduced amount ofmature 16 S rRNA (bands a and b in Figure 3A,respectively) than did MC4100 cells. At 37 8C, theamount of the pre-16 S rRNA was approximatelythree times greater in BX41 cells than in MC4100cells and the amount of the mature 16 S rRNA inBX41 cells was approximately half of the wild-typelevel (Figure 3A). Upon cold shock, the level ofmature 16 S rRNA was maintained in MC4100 cellsalthough the pre-16 S rRNA amount increased.However, in BX41 cells the mature 16 S rRNA levelgradually decreased further (Figure 3A).

In order to examine the status of 16 S rRNA pro-cessing in different ribosomal components, we

Figure 1. Localization of RbfA in the cell. Two MC4100(wild-type) cell cultures were grown in 100 ml of LBmedium at 37 8C to an exponential phase. Then, cells inone culture were immediately collected and cells in theother were collected after a cold-shock treatment at15 8C for six hours. Ribosomal profiles of the cells wereanalyzed by sucrose gradient centrifugation asdescribed.21 The gradients were fractionated into 24 frac-tions whose RbfA contents were analyzed by Westernblotting. A and B, Ribosomal profiles and cellular localiz-ations of RbfA at 37 8C and six hours after cold shock,respectively. Different ribosome particles are markedabove corresponding peaks. RbfA Western blots areshown below the ribosomal profiles. C, Stoichiometry ofRbfA association to 30 S ribosomal subunits. Sucrose gra-dient fractions corresponding to 30 S ribosomal subunitsfrom 37 8C cells and cold-shocked cells were loadedalong with indicated amounts of purified RbfA andthen subjected to Western blotting analysis.

RbfA in Cold Shock Adaptation 577

Figure 2. Ribosomal profiles ofwild-type (MC4100) and DrbfA(BX41) cells at 37 8C and after coldshock. A and B, Ribosomal profilesof MC4100 cells at 37 8C and fourhours after cold shock at 15 8C,respectively. C and D, Ribosomalprofiles of BX41 cells at 37 8C andfour hours after temperature down-shift to 15 8C, respectively. Differentribosome particles are markedabove corresponding peaks in A.

Figure 3. 16 S ribosomal RNA processing in MC4100 and BX41 cells. A, Status of the 16 S rRNA processing, as ana-lyzed by primer extension, at 37 8C and 0.5, 1, 3, 6, and 18 hours after cold shock in MC4100 and BX41 cells. The pre-16 S rRNA bands are indicated by a, and mature 16 S rRNA bands by b. B and C, The maturation status of the 16 SrRNA in different ribosomal particles of MC4100 and BX41 cells, four hours after temperature downshift to 15 8C.RNAs were extracted from fractions of the sucrose gradients shown in Figure 2B and D.

578 RbfA in Cold Shock Adaptation

extracted RNAs from individual fractions of thesucrose gradients shown in Figure 2 and subjectedthem to primer extension analysis using the sameprimer. Four hours after cold shock, 30 S ribosomalsubunits of MC4100 cells contained a hetero-geneous population of 16 S rRNAs, with approxi-mately 50% of the population being the maturespecies (the lane marked by a triangle inFigure 3B). At 37 8C, the whole pattern was practi-cally identical (not shown). On the other hand,BX41 cells contained much more 16 S rRNA pre-cursor than did MC4100 cells and the majority ofthe 16 S rRNA in the 30 S subunits in these cellswas found to be in the precursor form with an esti-mated less than 20% of the total population beingmature, at four hours after cold shock (the twolanes marked by triangles in Figure 3C). The 37 8Cpattern was quite similar except that slightly moremature 16 S rRNA species were observed at theregion corresponding to higher-order polysomeslocated toward the bottom of the gradient (notshown).

It should be noted also that a total of four“mature” 16 S ribosomal RNA species, differingfrom each other by only one base in length, wereobserved in both strains and at both temperatures.All these species were present in actively translat-ing polysomes (Figure 3B and C), indicating thatall of them are functional.

Limited trypsin proteolysis of RbfA

Since RbfA overexpressed in E. coli wasunstable, generating a smaller fragment duringpurification (not shown), we speculated thatRbfA may have a considerably flexible region ora tail that would be unstructured when the pro-teins exists in solution free from ribosomes, ashas been found for many ribosomal proteins.13

A limited trypsin proteolysis revealed that RbfAindeed contained an easily degradable portionand a relatively compact region resistant tomoderate proteolysis (Figure 4A). The molecularmass of the trypsin-resistant fragment was deter-mined to be 12,143 Da by mass spectroscopy,which corresponds to an RbfA fragment with itsC-terminal 25 residues truncated. This core RbfAfragment, designated RbfAD25, showed a typicalCD spectrum for a/b proteins (not shown) andwas highly heat-stable. As shown in Figure 4B,the melting temperature of the full-length RbfAwas approximately 78 8C, while the truncatedprotein was even more stable with a melting tem-perature of around 82 8C. Recently, we havedetermined the NMR structure of this RbfA corefragment to be a KH domain.12

Inability of RbfAD25 to stably bindto ribosomes

As mentioned before, many ribosomal proteinspossess long N-terminal or C-terminal extensionswhich frequently serve as major ribosomal-bindingmotifs by reaching deep into the core of a sub-unit.13 In order to test whether the C-terminalregions of RbfA share a similar function, westudied the cellular localization of RbfAD25 bysucrose density-gradient centrifugation followedby Western blotting. In cells expressing exogenousfull-length RbfA, although the majority of theprotein existed in soluble fractions, a significantamount co-fractionated with 30 S ribosomal sub-units (Figure 5A). In contrast, the truncated pro-tein existed exclusively in the soluble fractions(Figure 5B), indicating that the C-terminal regionof RbfA is required for its stable binding to 30 Ssubunits. However, a possible weak or transientinteraction between RbfAD25 and the 30 S subunitwhich was competed by endogenous RbfA cannotbe ruled out. A weak band (marked X, Figure 5A)migrating faster than RbfA at a position similar tothat of RbfAD25 was detected, which was presentonly under over-expression conditions and con-sidered to be an RbfA degradation product.

Figure 4. Biochemical analysis of the RbfA protein.A, Limited trypsin proteolysis of the RbfA protein.RbfA was digested with trypsin for indicated timeperiods and then the digestion reactions were loadedonto a 16.5% tricine gel to resolve the products: 10 mg ofRbfA and 50 ng of trypsin were used in each reaction.The reactions were carried out in a 10 ml volume in a buf-fer containing 20 mM Tris–HCl (pH 8.0) and 1 mMEDTA. B, Melting curve of the full-length RbfA and theRbfAD25. The circular dichroism (CD) measurementwas performed as described.23

RbfA in Cold Shock Adaptation 579

Complementation of the DrbfA mutantphenotypes by RbfAD25

Consistent with the previous report,8 the rbfAgene is not essential at 37 8C although BX41 cellsformed smaller colonies compared to MC4100cells at this temperature (Figure 6A) and wereunable to form colonies on LB-agar plates at 15 8C.In liquid LB medium, BX41 cells grew slower thandid MC4100 cells at 37 8C and were unable togrow after cold shock at 15 8C (Figure 6B). To testwhether RbfAD25 is biologically functional, apINIII plasmid expression system was constructedand tested for its ability to rescue above mutantphenotypes. It was found that the small colony for-mation and slow growth at 37 8C were completelysuppressed (Figure 6A and B). BX41 cells harbor-ing pINIIIrbfAD25 were also able to form coloniesat 15 8C, but they formed later and were smallerthan cells with pINIIIrbfA (Figure 6A). At 15 8C,the growth of BX41 cells expressing RbfAD25 at15 8C was also recovered but they grew slower thancells expressing the full-length RbfA (Figure 6B).Expression of the truncated RbfA did not showany discernible effect on the growth of wild-typeMC4100 cells at both temperatures (Figure 6B).Interestingly, overexpression of the intact RbfA at37 8C delayed the entry of MC4100 cells into expo-nential phase after dilution of an overnight cultureand also inhibited that of BX41 cells to a lesserextent (Figure 6B). However, both cells displayeda normal growth rate after cold shock (Figure 6B).

Although the expression system was IPTG-inducible, we observed that the leaky expressionof the truncated proteins in the LB medium with-out IPTG was sufficient to complement the variousgrowth defects. Under this condition, we measuredthe expression level of RbfAD25 and RbfA byWestern blotting. As shown in Figure 6C, the levelsof both proteins were constant in both strains andat both 37 8C and 15 8C. The level of the full-length

RbfA expressed from the plasmid appeared tobe five to ten times higher than that of theendogenous protein (compare lanes 1 and 2, or 7and 8), while the amount of RbfAD25 seemed tobe between one-third and one-half of that of theoverexpressed full-length RbfA or approximatelytwo to three times the endogenous RbfA level inMC4100 cells. It should be noted that truncation ofthe C-terminal 25 residues may cause a loss ofa potential epitope(s) for antibody recognition,leading to weaker Western blotting signals. Asexpected, RbfA was undetectable in the mutantstrain (lanes 4 and 10).

Suppression of the 16 S rRNA processingdefect of DrbfA cells by RbfAD25

Next we examined whether RbfAD25 could sup-press the 16 S rRNA processing defect caused byRbfA deletion. The status of 16 S rRNA processingof BX41 cells expressing RbfAD25 was comparedwith those expressing the full-length RbfA as wellas those carrying the vector alone by primer exten-sion as described earlier. As shown in Figure 7A, at37 8C, BX41 cells contained approximately threetimes more 16 S rRNA precursors and less maturespecies than did MC4100 cells (compare lanes 1and 4). When RbfAD25 was expressed in BX41cells (lane 6), the amount of the precursors waseffectively reduced to a level nearly identical withthose in either MC4100 or BX41 cells expressingthe full-length RbfA (lanes 1 and 5, respectively).It should be noted that overexpression of neitherRbfA nor RbfAD25 in wild-type cells (lanes 2 and3) further promoted the processing of the 16 SrRNA to completion, suggesting that RbfA is notthe rate-limiting factor for the 16 S rRNA process-ing. At 15 8C, expression of RbfAD25 also restoredthe efficient processing of 16 S rRNA in BX41 cells,as evidenced by a significant increase of theamount of mature species and a reduction of thatof pre-16 S rRNA (Figure 7B).

Effect of RbfAD25 expression on the (C23U)mutation in 16 S rRNA

RbfA was originally found as a suppressor for adominant-negative cold-sensitive phenotype ofcells carrying a C to U mutation at position 23(C23U) of 16 S rRNA.8,9 We, therefore, examined ifRbfAD25 was able to suppress the cold-sensitivephenotype of BW25141 cells harboring the p23Uplasmid, which expresses a 16 S rRNA with theC23U mutation. As shown in Figure 8, BW25141cells co-transformed with p23U and pACYC184(vector) formed smaller colonies at 42 8C andno colonies at 30 8C. When co-transformed withp23U and pACYC184rbfA, BW25141 cells formedcolonies not only at 42 8C but also at 30 8C.However, cells co-transformed with p23U andpACYC184rbfAD25 did not form colonies at 30 8C.

Figure 5. Inability of RbfAD25 to stably bind the 30 Sribosomal subunit. A and B, Localization of over-expressed RbfA and RbfAD25, respectively, in the cell.X indicates a possible RbfA degradation productobserved under overexpression condition.

580 RbfA in Cold Shock Adaptation

Discussion

When E. coli cells exponentially growing at 37 8Care shifted to a low temperature (i.e. 15 8C), cellsenter a growth lag period called the acclimationphase, during which a specific set of cold-shockproteins are induced while the production of mostcellular proteins is temporarily blocked.14 RbfA, a30 S ribosome-associated factor, is one such cold-shock protein.2,8 In this study, we showed at37 8C, approximately one-third of total cellularRbfA was associated with 30 S ribosomal subunits(Figure 1A), while about two-thirds became associ-ated with 30 S subunits, six hours after cold shock(Figure 1B). Only a subpopulation of free 30 Ssubunits contained RbfA and this population, as apercentage of total free 30 S subunits, increasedafter cold shock (Figure 1C). The induction of

RbfA and the shift of its distribution toward the30 S associated form after cold shock indicate thata more extensive RbfA–30 S subunit interaction isnecessary for efficient protein production at lowtemperature.

RbfA was initially proposed to be a factorinvolved in either the late maturation of the 30 Ssubunit or initiation of translation.8 Later, its rolein the 16 S rRNA processing was reported.10 In thepresent study, we demonstrated that BX41 cells(DrbfA) contained approximately three times more16 S rRNA precursors and significantly less maturespecies than did wild-type cells, especially at lowtemperature (Figure 3A). Consistent with theearly findings that the final stage of 16 S rRNAprocessing occurs after the assembly of the 30 Ssubunit but before the subunit can be further incor-porated into 70 S ribosomes and polysomes,15 – 17

Figure 6. Suppression of theDrbfA mutant phenotypes by RbfAand RbfAD25. A, Plates showingthe suppression of the small colonyformation phenotype of BX41 cellsby RbfAD25 at 37 8C. B, Growthcurve showing the suppression ofthe cold-sensitive growth of BX41cells by RbfA and RbfAD25.C, Western blot showing the levelof endogenous RbfA as well as theoverexpressed RbfA and RbfAD25in both MC4100 and BX41 cells at37 8C and after cold shock. Lanes 1,4, 7 and 10 are cells with the controlpINIII vector; lanes 2, 5, 8 and 11are cells with pINrbfA; lanes 3, 6, 9and 12 are cells with pINrbfAD25.Cell strains and temperatures areindicated above the gel.

RbfA in Cold Shock Adaptation 581

the 16 S rRNA precursor was detected almostexclusively in the 30 S subunits (Figure 3B and C).Evidently, DrbfA mutant cells accumulated a largeamount of not only the 30 S but also the 50 S sub-units and contained a much reduced amount ofpolysomes (Figure 2).8 The accumulation of 30 Ssubunits may be caused by the inability of theimmature particles to participate in translationinitiation, resulting in concomitant accumulationof 50 S subunits.

The limited proteolysis data (Figure 4A) suggestthat RbfA may have a C-terminal extension that ispoorly structured at least when the protein existsfree from ribosomes. We demonstrated that theC-terminal extension of 25 residues is not essentialfor basic RbfA function, since RbfAD25 was ableto suppress the cold-sensitive growth of DrbfA

cells (Figure 6) as well as the accumulation of thepre-16 S rRNA in these cells (Figure 7). The factthat deletion of this C-terminal extension abolishedthe ability of RbfA to stably associate with the 30 Ssubunit (Figure 5) but not the protein’s activity tosuppress the 16 S rRNA processing defect of theDrbfA cells (Figure 7) indicates that the role ofRbfA in the processing is not absolutely dependenton its stable ribosomal association in the case ofwild-type 16 S rRNA. However, in contrast tointact RbfA, RbfAD25 was unable to suppress thedominant-negative cold-sensitive phenotype ofcells expressing the C23U mutant 16 S rRNA(Figure 8), indicating that a stronger association ofRbfA with ribosome may be required for the pro-cessing of pre-16 S rRNAs containing the C23Umutation in helix I. Cells always maintain a stablepool of free RbfA (Figure 1A) regardless of thegrowing temperatures and this soluble populationmay contribute to the facilitation of 16 S rRNAprocessing.

Recently we have determined the NMR-solutionstructure of RbfAD25 consisting of two a-helicesand three b-strands forming a typical KH domain,which is known to interact with RNA.11,12 It istempting to speculate that RbfA may interact notonly with helix I of 16 S rRNA but also with othersites, which may be important for 16 S rRNAmaturation. The interaction of RbfA with 30 S sub-units may also involve protein–protein interactionrather than sole protein–RNA interaction. AsRbfA is not associated with 70 S ribosomes, RbfAmolecules interacting with 30 S subunits appear tobe removed upon 70 S ribosome formation.

The fact that RbfAD25 is able to effectivelyrestore 16 S rRNA processing and to suppress thecold-sensitivity of BX41 cells indicates that thegrowth defect caused by loss of RbfA is directlyrelated to the deficiency of 16 S rRNA processingin DrbfA cells. It seems that the cold sensitivitycaused by the deletion of rbfA is due to a limitedamount of mature 30 S subunits formed in mutantcells, as 30 S ribosomes can participate in trans-lation only after the processing of pre-16 S rRNA.It has been reported that RimM, a 21 kDa proteinspecifically bound to free 30 S ribosomes is essen-tial for efficient processing of 16 S rRNA, and that

Figure 7. Suppression of the 16 S rRNA processingdefect of BX41 cells by RbfA and RbfAD25 at (A) 37 8Cand (B) 15 8C. The maturation status of the 16 S ribo-somal RNA was examined by primer extension analysis.Lanes 1–6 correspond to MC4100/pINIII, MC4100/pINIIIrbfA, MC4100/pINrbfAD25, BX41/pINIII, BX41/pINIIIrbfA and BX41/pINIIIrbfAD25, respectively.

Figure 8. Suppression of thedominant-negative phenotype ofBW25141 cells harboring p23Uwhich were transformed withpACYC184, pACYC184rbfA, andpACYC184 rbfAD25, and colonyformation was examined at 42 8Cand 30 8C.

582 RbfA in Cold Shock Adaptation

rbfA functions as a suppressor for a DrimM strain.10

Recently we found that the cold-sensitive growthphenotype of DrbfA cells can be suppressed byoverproduction of Era, an essential GTPase con-taining a KH domain.18 Further characterization ofthe roles of these 30 S ribosome-associating pro-teins in 16 S rRNA processing is expected to shedlight into the precise molecular mechanism of 16 SrRNA processing leading to the formation of thefully functional 30 S ribosomes.

Materials and Methods

Bacterial strains and culture medium

E. coli strain MC4100 [F2 araD139D(argF-lac)U169rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsR ] was usedas a wild-type strain. The DrbfA mutant strain,named BX41, was constructed by transducing adisrupted rbfA allele from strain CD288 to MC4100using P1 phage. The transductants were selectedby 50 mg/ml of kanamycin and candidates werefurther confirmed to be cold sensitive and lackingRbfA protein in the cell as examined by Westernblotting. Strain BW25141 (laclq rrnBT14 DlacZWJ16

DphoBR580 hsdR514 DaraBADAH33 DrhaBADLD78

galU95 endABT333 uidA (DMluI < pirþ recA1)19 wasused to examine the C23U dominant-negativegrowth.

Plasmid construction

The full-length coding sequence (from ATG toTAA) of the rbfA gene was amplified by PCRand cloned between the NdeI and BamHI sitesof pET11a and an NdeI-containing derivativeof a pINIII vector,20 generating pETrbfA andpINIIIrbfA. The expression vectors used to pro-duce the C-terminally truncated RbfA protein,pETrbfAD25 and pINrbfAD25, were then obtainedby changing the 109th codon (ATG) of the wild-type rbfA gene to a TAA stop codon by site-directed mutagenesis. The mutagenesis was per-formed following the protocol described in theQuickChange Mutagenesis Kit (Stratagene). Inorder to transform BW25141 cells carrying p23U(Ampr)8,9 with a vector expressing RbfA andRbfAD25, the rbfA gene and the rbfAD25 genewere transferred from pETrbfA and pETrbfAD25to pACYC184 (Cmr) (New England Biolabs) to con-struct pACYC184rbfA and pACYC184rbfAD25,respectively. The p23U plasmid is for theexpression of 16 S rRNA having a C to U mutationat position 23 (C23U), which causes a dominantcold-sensitive phenotype.8,9

Protein expression and purification

To overexpress the RbfA proteins, E. coli BL21cells transformed with pETrbfA and pETrbfA.25were grown in 1.5 l of M9 medium containing0.4% (w/v) glucose and 0.2% (w/v) casein

amino acids at 37 8C with vigorous shaking.When the A600 of the cultures reached 0.5 to 0.6,IPTG was added to a 1.5 mM final concentrationto induce the expression of target proteins. Thecultures were continued for another three hoursand then cells were collected by centrifugation at4 8C. The full-length and truncated RbfA proteinswere purified following the same protocol asbelow.

Cells from each 1.5 l culture were resuspendedin 30 ml of buffer I (20 mM phosphate (pH 5.8),50 mM KCl, 5 mM b-mercaptoethanol, 1 mM pro-tease inhibitor PMSF), washed once and then dis-rupted by two passages through a French press inthe same buffer. The resulting cell lysate was clari-fied by centrifugation at 12,000 rpm (17, 211 g) forten minutes at 4 8C. Proteins in the supernatantwas precipitated by adding (NH4)2SO4 to 60%saturation. Protein precipitates were collected bycentrifugation and dissolved in 5 ml of bufferI. The resulted protein solution was centrifuged at12,000 rpm for ten minutes and then loaded onto agel-filtration column (Sephacryl S-100, AmershamPharmacia Biotech) and eluted using buffer I.Fractions containing partially purified RbfA werethen pooled and loaded onto an SP-Sepharosecolumn (Amersham Pharmacia Biotech) and elutedby a linear gradient of 50 mM–500 mM KCl inbuffer I. Fractions containing sufficiently pureRbfA proteins (.98%) were pooled and dialyzedagainst a 20 mM sodium acetate buffer (pH 4.0).Then, the protein was concentrated and stored inthe same low pH buffer.

Primer extension analysis of the 16 Sribosomal RNA processing

To examine the maturation status of the 16 Sribosomal RNA in different ribosomal particles,cellular components were first separated bysucrose gradient ultra-centrifugation as describedbefore.21 The gradients were then fractionated into24 fractions from which RNAs were individuallyobtained by means of extraction with phenol/chloroform and precipitation with ethanol. TheRNAs were dissolved in 20 ml of water and 1 ml ofeach RNA was used in each primer extensionexperiment.

To analyze the maturation status of the unfrac-tionated 16 S ribosomal RNA, 2 mg of total RNAextracted by a hot phenol method22 was used ineach primer extension reaction. OligonucleotidePE16 S (50-CGACTTGCATGTGTTAGG-30), whichanneals to a region approximately 50 basesfrom the 50-end of the mature 16 S ribosomalRNA, was labeled at the 50-end with [g-32P]ATPusing phage T4 polynucleotide kinase and used asa primer. Two picomoles of the labeled primerwas used in each reaction, which was carriedout at 42 8C for one hour using AMV-RT in thepresence of RNase inhibitor (both from RocheMolecular Biochemicals).

RbfA in Cold Shock Adaptation 583

Acknowledgements

We thank Dr H. Noller for p23U. This work issupported by a grant (GM19043) from the NationalInstitute of Health.

References

1. Jones, P. G., VanBogelen, R. A. & Neidhardt, F. C.(1987). Induction of proteins in response to low tem-perature in Escherichia coli. J. Bacteriol. 169,2092–2095.

2. Jones, P. G. & Inouye, M. (1996). RbfA, a 30 S ribo-somal binding factor, is a cold-shock protein whoseabsence triggers the cold-shock response. Mol. Micro-biol. 21, 1207–1218.

3. Yamanaka, K., Fang, L. & Inouye, M. (1998). TheCspA family in Escherichia coli: multiple gene dupli-cation for stress adaptation. Mol. Microbiol. 27,247–255.

4. Xia, B., Ke, H. & Inouye, M. (2001). Acquirement ofcold sensitivity by quadruple deletion of the cspAfamily and its suppression by PNPase S1 domain inEscherichia coli. Mol. Microbiol. 40, 179–188.

5. Jiang, W., Hou, Y. & Inouye, M. (1997). CspA, themajor cold-shock protein of Escherichia coli, is anRNA chaperone. J. Biol. Chem. 272, 196–202.

6. Jones, P. G., Mitta, M., Kim, Y., Jiang, W. & Inouye,M. (1996). Cold shock induces a major ribosomal-associated protein that unwinds double-strandedRNA in Escherichia coli. Proc. Natl Acad. Sci. USA, 93,76–80.

7. Toone, W. M., Rudd, K. E. & Friesen, J. D. (1991).deaD, a new Escherichia coli gene encoding a pre-sumed ATP-dependent RNA helicase, can suppressa mutation in rpsB, the gene encoding ribosomal pro-tein S2. J. Bacteriol. 173, 3291–3302.

8. Dammel, C. S. & Noller, H. F. (1995). Suppression ofa cold-sensitive mutation in 16 S rRNA by over-expression of a novel ribosome-binding factor,RbfA. Genes Dev. 9, 626–637.

9. Dammel, C. S. & Noller, H. F. (1993). A cold-sensitivemutation in 16 S rRNA provides evidence for helicalswitching in ribosome assembly. Genes Dev., 7,660–670.

10. Bylund, G. O., Wipemo, L. C., Lundberg, L. A. &Wikstrom, P. M. (1998). RimM and RbfA are essentialfor efficient processing of 16 S rRNA in Escherichiacoli. J. Bacteriol. 180, 73–82.

11. Swapna, G. T., Shukla, K., Huang, Y. J, Ke, H., Xia, B.,Inouye, M. & Montelione, G. (2001). Resonanceaasignments for cold shock protein ribosome-bindingfactor A(RbfA) from Escherichia coli. J. Biomol. NMR,21, 389–390.

12. Huang, Y. J., Swapna, G. V., Rajan, P. K., Ke, H., Xia,B., Shukla, K. et al. (2003). Solution NMR structureof ribosome-binding factor A (RbfA), a cold-shockadaptation protein from Escherichia coli. J. Mol. Biol.327, 521–536.

13. Wimberly, B. T., Brodersen, D. E., Clemons, W. M., Jr,Morgan-Warren, R. J., Carter, A. P., Vonrhein, C. et al.(2000). Structure of the 30 S ribosomal subunit.Nature, 407, 327–339.

14. Yamanaka, K. (1999). Cold shock response in Escheri-chia coli. J. Mol. Microbiol. Biotechnol. 1, 193–202.

15. Nierhaus, K. H. (1991). The assembly of prokaryoticribosomes. Biochimie, 73, 739–755.

16. Lindahl, L. (1975). Intermediates and time kinetics ofthe in vivo assembly of Escherichia coli ribosomes.J. Mol. Biol. 92, 15–37.

17. Lindahl, L. (1973). Two new ribosomal precursorparticles in E. coli. Nature New Biol. 243, 170–172.

18. Inoue, K., Alsina, J., Chen, J. & Inouye, M. (2003).Suppression of defective ribosome assembly in arbfA deletion mutant by overexpression of Era, anessential GTPase in Escherichia coli. Mol. Microbiol.48, 1005–1016.

19. Datsenko, K. A. & Wanner, B. L. (2000). One-stepinactivation of chromosomal genes in Escherichia coliK-12 using PCR products. Proc. Natl Acad. Sci. USA,97, 6640–6645.

20. Masui, Y., Coleman, J. & Inouye, M. (1983). Multi-purpose expression cloning vehicles in Escherichiacoli. In Experimental Manipulation of Gene Expression(Inouye, M., ed.), pp. 15–32, Academic Press, NewYork.

21. Xia, B., Etchegaray, J. P. & Inouye, M. (2001). Non-sense mutations in cspA cause ribosome trappingleading to complete growth inhibition and celldeath at low temperature in Escherichia coli. J. Biol.Chem. 276, 35581–35588.

22. Sarmientos, P., Sylvester, J. E., Contente, S. & Cashel,M. (1983). Differential stringent control of the tan-dem E. coli ribosomal RNA promoters from therrnA operon expressed in vivo in multicopy plas-mids. Cell, 32, 1337–1346.

23. Shinde, U. P., Liu, J. J. & Inouye, M. (1997). Proteinmemory through altered folding mediated by intra-molecular chaperones. Nature, 389, 520–522.

Edited by D. E. Draper

(Received 16 June 2003; received in revised form 17 July 2003; accepted 22 July 2003)

584 RbfA in Cold Shock Adaptation