Use of a Riboswitch-controlled Conditional HypomorphicMutation to Uncover a Role for the Essential csrA Gene inBacterial Autoaggregation*

Received for publication, June 1, 2009, and in revised form, August 4, 2009 Published, JBC Papers in Press, August 25, 2009, DOI 10.1074/jbc.M109.028076

Ye Jin‡, Rory M. Watt§, Antoine Danchin¶1, and Jian-dong Huang‡2

From the ‡Department of Biochemistry, Li Ka Shing Faculty of Medicine, and §Oral Biosciences, Faculty of Dentistry, University ofHong Kong, Pok Fu Lam, Hong Kong, China and ¶Genetics of Bacterial Genomes, CNRS URA2171, Institut Pasteur, 28 Rue duDocteur Roux, 75724 Cedex 15 Paris, France

Essential genes encode biological functions critical for cellsurvival. Correspondingly, their null mutants are often difficultto obtain, which impedes subsequent genetic and functionalanalysis. Here, we describe the development and utility of atheophylline-responsive riboswitch that enables target geneexpression to be specifically “tuned” from low to high levels,which may be used to generate conditional hypomorphicmutants. Low levels of gene activity in the absence of the ligand(theophylline) permit cell survival, enabling gene activities to beinvestigated. Normal gene expression levels and wild-type phe-notypes can be restored by the addition of the ligand. We dem-onstrate the utility of this approach with csrA, an essential genein Escherichia coli that encodes the global regulatory proteinCsrA.We placed the theophylline-responsive riboswitch imme-diately upstream of the csrA ribosome binding site, with theresultingmutant named switch-csrA.Hypomorphismof switch-csrA and its specific responsiveness to theophylline were veri-fied by phenotypic examination and translation analysis. Theutility of switch-csrA revealed a previously unidentified func-tion for CsrA, namely its role as a repressor of cellular autoag-gregation. Specifically, switch-csrA in the non-ligand-boundform produced low levels of CsrA, and its cells autoaggregated.Theophylline binding induced conformational changes in theriboswitch and permitted efficient csrA translation; conse-quently, autoaggregation did not occur. Our results indicatethat CsrA modulates autoaggregation via the polysaccharideadhesin poly-�-1,6-N-acetyl-D-glucosamine. In summary, theuse of ligand-responsive riboswitches to construct conditionalhypomorphic mutants represents a novel approach for investi-gating the activities of essential genes, which effectively comple-ments traditional genetic approaches.

In Escherichia coli, at least 620 open reading frames arerequired for viability in rich growth media (1). These essentialgenes are of particular interest because they play critical roles inbasic cellular functions; they are potential components for cre-

ating a minimal genome (2), and their mutation provides apromising approach to drug target validation in antibacterialdrug discovery.However, genetic studies of essential genes have been

impeded by mutation-caused lethality. The effects of overex-pressing essential genesmay be easily examined, but investigat-ing the effects of lowering their expression levels may be tech-nically difficult to achieve. This problem may be solved byconstructing insertional mutants or by deleting a section of theessential genes so that the resulting mutants are partially activein gene functions. However, the construction of such a hypo-morphicmutant is not feasible when its integrity is essential forfull gene functionality. Alternatively, one can construct condi-tional lethal amber mutations in essential genes (3). With thismethod, an amber stop codon is introduced into a target gene toconstruct a nonsensemutation, which results in the productionof a truncated protein. A plasmid is used to produce an ambersuppressor tRNA, which recognizes the amber stop codon andallows translation to “read through” the codon and produce afull-length protein. However, like the insertion mutation andpartial gene deletion discussed above, the amber mutation pro-duces truncated proteins, which are usually non-functional andtherefore not ideal for constructing hypomorphic mutations.Another approach is to place the target gene under the controlof an inducible promoter (4–6) to turn gene expression on andoff in response to the presence or absence of promoter induc-ers, respectively. However, with this method, the wild-typeexpression levels and the natural transcriptional regulation ofthe gene can be overridden by the inducible promoter.Here, we have developed a riboswitch-based approach for

constructing hypomorphic mutations in essential genes. Ourapproach has several advantages over previously reportedmethods, in that the riboswitch-controlled hypomorphicmutants produce full-length functional proteins, the productlevels are controlled within a “natural” expression range, andriboswitches enable the specific and reversible control of geneexpression in a ligand-dependentmanner. To test the feasibilityof our methodology, we selected the csrA gene, which is knownto be an essential gene in E. coli (6). We rationally designed atheophylline-responsive riboswitch with in vivo activity andused this to construct a hypomorphic mutation of csrA (here-after referred to as switch-csrA). Results from �-galactosidaseassays involving csrA-lacZ translational fusions constructed onthe chromosome of the switch-csrA strain confirmed that only

* This work was supported by the University of Hong Kong Committee onResearch and Conference Grants Seed Funding Program for Basic Researchand by Research Grants Council of Hong Kong Grant HKU 7485/06M (toJ. D. H.).

1 Supported by PROBACTYS Program Grant CT-2006-029104 in an effort todefine genes essential for the construction of a synthetic cell.

2 To whom correspondence should be addressed. Tel.: 852-2819-2810; Fax:852-2855-1254; E-mail: jdhuang@hkucc.hku.hk.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 284, NO. 42, pp. 28738 –28745, October 16, 2009© 2009 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

28738 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 • NUMBER 42 • OCTOBER 16, 2009

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

small amounts of CsrA were expressed in the absence of theo-phylline, but there were nearly wild type gene expression levelsin the presence of the ligand. Examination of cell motility onsemisolid agar further validated the riboswitch-controlled hy-pomorphy of csrA.In addition to controlling the csrA hypomorphic mutation,

the use of the artificial riboswitch enabled us to identify a newfunction for the CsrA protein, namely that it acts as a repressorof cellular autoaggregation (AG).3 Analogously, the overpro-duction of the CsrB small non-coding RNA (sncRNA), which isan antagonist of CsrA, induced AG similar to that observed forthe riboswitch-controlled hypomorphic mutant of csrA. Ourresults further indicate that CsrA mediates its effects upon AGby regulating the biosynthesis of the biofilm adhesin poly-�-1,6-N-acetyl-D-glucosamine (PGA). Thereby, we demonstratethat ligand-responsive riboswitches can be effectively used tocomplement traditional approaches for investigating genefunctions when null mutations are lethal.

EXPERIMENTAL PROCEDURES

Bacterial Strains, Plasmids, and Growth Conditions—Thebacterial strains and plasmids used in this study are listed in

Table 1. All bacterial strains were grown at 37 °C, with shak-ing at 220 rpm, in Luria-Bertani (LB) medium or M63 mini-mal medium supplemented with 0.2% glucose. The antibiot-ics ampicillin (50 �g/ml), kanamycin (50 �g/ml), andchloramphenicol (12.5 �g/ml) were used for selection whenappropriate.Mutagenesis—Gene deletion mutants were constructed

using the �red recombination system (7). E. coli-K12 MG1655was transformed with plasmid pSim6 (a gift from Dr. DonaldCourt) (8), from which the expression of the � recombinationproteins is induced at 42 °C. PCR fragments encompassing aloxP-cm-loxP or FRT-kan-FRT cassette with homology (45 nt)to the regions immediately flanking each deletion locus weretransformed via electroporation into MG1655 cells harboringpSim6. After induction of �red functions, recombinants wereselected for chloramphenicol (cm) or kanamycin (kan) resist-ance and were further verified by colony PCR. Primers used toconstruct the deletion mutants are listed in Table 2.csrB Cloning—The selectable loxP-cm-loxP cassette was

chromosomally inserted immediately downstream of the stopcodon of csrB (as described above). The gene (with its nativepromoter) and adjacent loxP-cm-loxP cassette were then PCR-amplified using primers that contained homology (45 nucleo-tides) to the plasmid insertion site. The PCR product andpET32a(�) expression vector (Novagen) were co-transformed

3 The abbreviations used are: AG, autoaggregation; PGA, poly-�-1,6-N-acetyl-D-glucosamine; RBS, ribosome binding site; sncRNA, small noncoding RNA;UTR, untranslated region; cm, chloramphenicol; kan, kanamycin.

TABLE 1Strains and plasmids used in this study

Strain/plasmid Relevant characteristicsa Source

StrainsMG1655 E. coli K-12 reference strain Laboratory collectionDY330 W3110 lacU169 ��cI857ts � (cro-bioA)� Laboratory collectionMG1655 �flu �flu::cat, Cmr This studyMG1655 switch-csrA cat-riboswitch-csrA, Cmr This studyMG1655 switch-csrA-lacZ cat-riboswitch-csrA-lacZ translational fusion in chromosome This studyMG1655 �pgaB �pgaB::cat, Cmr This studyMG1655 �pgaB switch-csrA �pgaB cat-riboswitch-csrA, Cmr This studyMG1655 cat-pga cat-pgaABCD, Cmr This studyMG1655 cat-pga-lacZ cat-pgaAB-lacZ translational fusion in chromosome, Cmr This studyMG1655 pga-lacZ pgaAB-lacZ translational fusion in chromosome This study

PlasmidspSim6 pSC101, Ampr Ref. 8pCm Empty multicopy (pET) vector, with the fragment between the T7-lac

promoter and terminator replaced with a Cmr geneRef. 9

pCsrB csrB gene fromMG1655 in a multicopy pET vector under the control ofthe native promoter of csrB

This study

a Cmr, chloramphenical resistance; Ampr, ampicillin resistance.

TABLE 2Oligonucleotide primers used to construct the E. coli MG1655 mutants used in this study

Name Sequence (5�–3�)

pCsrB-F CGAGATCGATCTCGATCCCGCGAAATTAATACGACTCACTATAGGCTCAAATCTTGCGGAATTCCpCsrB-R AAAAACCCCTCAAGACCCGTTTAGAGGCCCCAAGGGGTTATGCTAGGATAGCAGGAATAAAAAAAGGGpCm-F CGAGATCGATCTCGATCCCGCGAAATTAATACGACTCACTATAGGAATAGGCGTATCACGAGGCpCm-R AAAAACCCCTCAAGACCCGTTTAGAGGCCCCAAGGGGTTATGCTATAGTGAACCTCTTCGAGGGACflu-F GTGGTGGATGACCTCAGCCATGCCGGACAGATTCATTTCACCTCCAATAGGCGTATCACGAGGCflu-R GCATAACCGGCCTGAACGCCCAGGGTGATATTTTCCCGGACACGGTAGTGAACCTCTTCGAGGGACSwitch-csrA-F TTATATGATGGATAATGCCGGGATACAGAGAGACCCGACTCTTTTAATAGGCGTATCACGAGGCSwitch-csrA-R GAACTGCCAAGGGCATCAAGACGATGCTGGTATTTCAGTTTGTTTTAGTGAACCTCTTCGAGGGACpgaB-F ATGTTACGTAATGGAAATAAATATCTCCTGATGCTGGTGAGTATAAATAGGCGTATCACGAGGCpgaB-R TTTCGGATACCAGGCTGTTGAAAACTCAGGACGAATAAGGTCTATTAGTGAACCTCTTCGAGGGACcat-pga-F GGAATTTACTGATTTAATTATTTTAATCCTAATTTATTTTGAAAAAATAGGCGTATCACGAGGCcat-pga-R AGCCCATTTGGTTTTCGGGCACCTTTTTCTGCTACTTGAATACATAAATTACGCCCCGCCCTGCcsrA-lacZ-F ATCTACCAGCGTATCCAGGCTGAAAAATCCCAGCAGTCCAGTTACCTGGCCGTCGTTTTACAACcsrA-lacZ-R AGAAATTTTGAGGGTGCGTCTCACCGATAAAGATGAGACGCGGAATAGTGAACCTCTTCGAGGGACpga-lacZ-F CTTATTCGTCCTGAGTTTTCAACAGCCTGGTATCCGAAAAATGATCTGGCCGTCGTTTTACAACpga-lacZ-R GGTATGCATAACACCAGACATAATATAAAAAACGATACGATGCGATAGTGAACCTCTTCGAGGGAC

Role for csrA in Bacterial Autoaggregation

OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28739

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

into E. coli DY330 that had previously been induced for �redrecombination functions (7). Recombinants were selected forchloramphenicol resistance and verified by PCR and sequenc-ing. The native promoter drove CsrB overproduction inMG1655 (without requiring induction). Primers used for csrBcloning are listed in Table 2.Semiquantitative Reverse Transcription-PCR—Total RNA

was isolated from overnight cultures in LB medium. Subse-quently, 2�g of RNAwas reverse transcribed in a total reactionvolume of 20 �l. Each reaction was incubated at 55 °C for 50min, followed by 15 min at 70 °C. 2 �l of the resulting reversetranscript products (cDNA) were then used for 22, 24, 26, 28,and 30 rounds of PCR (30 s each at 94, 55, and 72 °C) with ExTaqDNA polymerase (Takara Bio Inc.) and primers comple-mentary to csrB (RT-CsrB-F, 5�-GTCAGACAACGAAGTGA-ACATCAGG-3�; RT-CsrB-R, 5�-GGAGCACTGTATTCACA-GCGCT-3�) and the 16 S rRNA gene (RT-16S-F, 5�-CTCCTA-CGGGAGGCAGCAG-3�; RT-16S-R, 5�-CTCCGTATTACC-GCGGCTG-3�) to co-amplify the gene of interest and theinternal control. PCR products were separated in 1.5% agarosegel.Construction of Chromosomal lacZ Translational Fusions

and �-Galactosidase Assays—The loxP-cm-loxP-selectablecassette was inserted immediately after the stop codon of thelacZ gene on theMG1655 chromosome using �red recombina-tion (as described above). Next, the lacZ-loxP-cm-loxP cassettewas PCR-amplified and inserted (in frame) immediately priorto the stop codon of the target gene on the chromosome. Theinserted lacZ fragment started from the eighth codon of thelacZ gene and was co-transcribed and translated with the fusedgenes. The expression of gene-lacZ fusions was quantifiedusing a �-galactosidase assay, as described previously (9). In allof the lacZ fusion constructs used in this study, the endoge-neous lacZ gene was not deleted. Our pilot studies showed thatthe endogeneous lacZ levels were extremely low, and thereforeits presence did not affect our lacZ fusion results. Primers usedto construct the lacZ fusions are listed in Table 2.Autoaggregation Assay—Overnight cultures were diluted

1:200 and incubated at 37 °C, with shaking at 220 rpm, in liquidLBmedium.When cells aggregated (visualizedmacroscopicallyby the clumping or “fluffing” of cells in liquid cultures), 100�l ofeach cell suspension was transferred to flat-bottom 96-wellplates (Iwaki, Tokyo, Japan), and the images of the cell aggre-gates were captured by scanning. To better visualize the cellaggregates, 1 �l of crystal violet (0.1%) was added to each cellsuspension (100 �l) immediately prior to image capture.Motility Assay—Overnight cultures in LBmediumwere used

to assay cell motility. 2 �l of the bacterial suspensions wereplaced onto the centers of the motility plates (LB medium con-taining 0.3% agar). Plates were then incubated at 37 °C for16–20 h. Motility was assessed by examining the colony sizeson the semisolid agar medium.Statistical Analysis—Paired t tests were used to compare the

two means obtained from �-galactosidase assays; one-wayANOVA was used for the comparison of multiple means. pvalues of �0.05 were considered statistically significant.

RESULTS

Construction of a Riboswitch-controlled Hypomorphic Mu-tant of csrA—It has previously been reported that csrA isrequired for the growth of the E. coli DJ624 strain (a MG1655derivative, lacX74malP::lacIq) in LBmedium (6). To determinewhether the csrA gene was also essential in E. coli K-12MG1655, extensive attempts were made to delete this gene inthis strain. No csrA null mutants could be successfully con-structed in LB, although a large number of recombineeringexperiments were performed. This indicated that the csrA genewas essential for the viability of E. coliMG1655 under the con-ditions used.We therefore selected the csrA gene for the devel-opment of a riboswitch-controlled conditional hypomorphicmutation, to investigatewhether this approach could be used tofacilitate the genetic study of essential genes.To construct such a hypomorphic mutation, we sought to

rationally design a ligand-responsive riboswitch that wouldhave the ability to modulate expression levels when placedimmediately upstreamof the ribosome binding site (RBS) of thetarget gene. This riboswitch should meet the following fourcriteria: (a) the riboswitch should adopt multiple conforma-tions, and those with lower free energy would not permit geneexpression, so that the target gene expression would berepressed in the absence of a specific ligand; (b) at least one ofthese riboswitch conformations would permit limited geneexpression, so that the target gene expression would not beentirely “shut off” in the absence of the ligand; (c) the addition ofthe ligand would favor the formation of the conformation thatpermits gene expression, so that gene expression could be spe-cifically derepressed by adding the ligand; and (d) in the ligand-bound form, the riboswitch could increase the target geneexpression to nearly wild-type levels.Based on these criteria, we designed a theophylline-respon-

sive riboswitch and integrated this directly upstreamof the csrARBS (Fig. 1). The resultingmutant was named switch-csrA. Thesecondary structures formed by the construct and their freeenergy were predicted using RNAstructure version 4.6 (10, 11).This cis regulatory element is composed of a theophyllineaptamer (12, 13) and a short nucleotide sequence, such thatfour conformations (A, B, C, and D) coexist at equilibrium. Inconformations A and B, the RBS is base-paired, and thereforethe gene translation is blocked (14, 15). In conformation C, theRBS is immediately adjacent to a stem loop. Consequently, geneexpression levels are likely to be low, since it has previouslybeen demonstrated that a stem-loop structure directly besidethe RBS interferes with ribosome accessibility (16). In contrast,conformationD, inwhich the RBS is single-stranded, putativelyallows for efficient ribosome binding and gene translation.According to the prediction of free energy using RNAstructure4.6, conformations A (�21.9 kcal/mol) and B (�21.4 kcal/mol)have relatively lower free energy than conformations C (�21.0kcal/mol) and D (�20.9 kcal/mol) and therefore are proposedto predominate in the absence of theophylline. Consequently,the overall csrA translation levels would be expected to be low.However, when theophylline binds to the riboswitch, the equi-librium distribution would be shifted to favor the formation of

Role for csrA in Bacterial Autoaggregation

28740 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 • NUMBER 42 • OCTOBER 16, 2009

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

conformations C andD. Increased levels of the conformationDshould significantly enhance the expression of csrA.CsrA Expression Is Tightly Controlled by the Artificial

Riboswitch—To determine how effectively the riboswitch reg-ulated the expression of CsrA, we constructed a chromosomalcsrA-lacZ translational fusion and then integrated the ribos-witch device into this construct, forming switch-csrA-lacZ. Inthe absence of theophylline but in the presence of 5mM caffeine(serving as a control), switch-csrA-lacZ had significantly lowerlevels of �-galactosidase activity than csrA-lacZ.When caffeinewas replaced with 5 mM theophylline, the levels of �-galacto-sidase activity in switch-csrA-lacZ increased significantly andwere only marginally lower (p � 0.05) than those of the csrA-lacZ translational fusion (Fig. 2). This indicated that the bind-ing of theophylline induced a conformational change in theriboswitch region, “freeing up” the ribosomal binding site forefficient CsrA expression. There were extremely low levels of�-galactosidase activity in the negative control, switch-csrA(which lacks the lacZ fusion), in the presence of either caffeine

or theophylline (Fig. 2). This excluded the possibility that theaddition of theophylline or caffeine influences (endogenous)lacZ expression levels either in the presence or absence of theriboswitch. Taken together, these results demonstrated that theriboswitch was specifically responsive toward theophylline andthat in the absence of this ligand, the riboswitch significantlyreduced csrA expression levels.To further validate the ability of the riboswitch to control

csrA expression, we examined the motility phenotype of theswitch-csrA strain, taking advantage of the fact that CsrAmod-ulates the motility of E. coli cells (17). We found that switch-csrA cells were immotile in 0.3% semisolid LB agar in theabsence of theophylline (Fig. 1).When theophylline was added,the cells regained their motility (Fig. 1), providing additionalconfirmation that the riboswitch could effectively regulate csrAexpression.Riboswitch-controlledHypomorphicMutation of csrA Results

in Cell Autoaggregation—Interestingly, the switch-csrA cellsthat contained low levels of CsrA (i.e. grown in the presence of

FIGURE 1. Structures of the four predicted conformations (A–D) adopted by the riboswitch in the switch-csrA strain and the motility phenotypes ofswitch-csrA cells in the presence or absence of theophylline (5 mM). When theophylline was not added, the closely related xanthine compound caffeine (5mM) was used as a negative control (whose chemical structure differs from theophylline only by the additional presence of one methyl group). In conformationsA and B, the RBS is base-paired, and therefore gene translation is likely to be effectively blocked. In conformation C, the RBS is immediately adjacent to a stemloop, and the gene expression levels are also likely to be low. In conformation D, the RBS is single-stranded which putatively allows for efficient ribosomebinding and gene translation. We speculate that when theophylline binds to the riboswitch, the equilibrium distribution shifts to favor the formation ofconformations C and D. Increased levels of conformation D should activate csrA expression. The motility of E. coli cells was examined on semisolid (0.3%)Luria-Bertani agar after 16 h of culture at 37 °C. The color scheme used was as follows. Red, ribosome binding sites; green, AUG start codon.

Role for csrA in Bacterial Autoaggregation

OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28741

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

caffeine but without theophylline) flocculated when grown inLB medium (Fig. 3), which is a typical AG phenotype. Theswitch-csrA cells also autoaggregated when they were incu-bated in M63 medium containing 0.2% glucose in the absenceof theophylline (data not shown). Conversely, the theophylline-treated cells, whose CsrA expression levels were comparablewith those of the wild type, did not aggregate (Fig. 3). Theseresults indicate that CsrA acts as a negative regulator of AG.CsrB Overproduction Induces Cell Autoaggregation in E. coli—

The CsrB sncRNA has previously been demonstrated toantagonize the activities of the CsrA protein by sequestering it

(18). Thus, CsrB-overproducing E. coli cells have lower levels ofCsrA activity.We hypothesized that if AGwas repressed by theCsrA protein, then CsrB overproduction would result in AGlevels analogous to those observed for the switch-csrA cells (inthe absence of theophylline). We cloned CsrB together with itsnative promoter into a high copy pET vector, forming plasmidpCsrB that constitutively transcribed CsrB inMG1655. reversetranscription-PCR verified that CsrB was abundantly producedin the cells carrying pCsrB, whereas this sncRNA was barelydetectable in cells containing an empty vector control (Fig. 4).Weobserved that the plasmid-encodedCsrB inducedAG in thewild-type MG1655 strain in LB (Fig. 3) as well as in M63medium (data not shown). Thus, reducing intracellular CsrAactivities via the overproduction of CsrB results in cellular AG.This supplies additional evidence that the AG phenotypeobserved for the switch-csrA strain is due to reduced CsrA lev-els and that CsrA negatively modulates AG.PGA Is an Autoaggregation Factor Responsible for CsrA-me-

diated Autoaggregation—Ag43 (encoded by flu) (19) has previ-ously been established as an AG-inducing factor in E. coli K-12strains. However, we observed that CsrB overexpression drivenby its native promoter conferred AG to a flu mutant (data notshown). Thus, Ag43 is not required for the AG phenotype dis-cussed in this study.CsrA has been shown to regulate the expression of the glg-

CAP operon (20), cstA (21), the pgaABCD operon (22), hfq (20),flhDC (17), and ycdT and ydeH (23) at the post-transcriptionallevel. Of these targets, the pgaABCD operon encodes for pro-teins involved in the biosynthesis and translocation of the cell-bound polysaccharide PGA, which is a biofilm adhesin. Wespeculated that this adhesin might be a potential AG-inducingfactor that was part of the CsrA-mediated AG pathway. To testthis possibility, we first examined the effects that differing levelsof CsrAhad on the expression of a pga-lacZ translational fusionon the MG1655 chromosome. Overproducing CsrB, whichsequesters theCsrAprotein, significantly increased the levels ofthe pga-lacZ fusion (Fig. 5). This is consistent with the previ-ously reported finding that CsrA negatively regulates the bio-

FIGURE 2. Regulation of intracellular CsrA expression levels by the theo-phylline-responsive riboswitch. LacZ expression levels in the switch-csrA-lacZ, csrA-lacZ, and switch-csrA strains were determined using �-galactosid-ase assays in the presence of 5 mM caffeine (ca) or 5 mM theophylline (th).Switch-csrA, which does not contain the csrA-lacZ fusion, served as a negativecontrol, indicating endogenous lacZ expression levels. �-Galactosidase activ-ities (relative units) were normalized to those of switch-csrA-lacZ in the pres-ence of 5 mM caffeine. Error bars, S.D. (*, p � 0.05).

FIGURE 3. Cell flocculation of various mutant strains of E. coli MG1655.Overnight cultures were diluted 1:200 and incubated at 37 °C, with shaking at220 rpm. The switch-csrA strain was incubated in LB medium supplementedwith 5 mM theophylline (switch-csrA theophylline) or 5 mM caffeine as a con-trol (switch-csrA caffeine), whereas other strains were grown in (unsupple-mented) LB containing antibiotics where applicable. When cells aggregated(evaluated by visual inspection), 100 �l of each cell suspension was trans-ferred to a 96-well plate. To make the cell aggregates more visible, 1 �l ofcrystal violet (0.1%) was added to each cell suspension immediately prior toimage capture. wt, wild type MG1655; wt pCsrB, MG1655 strain carrying theCsrB expression plasmid pCsrB; wt pCm, MG1655 carrying control plasmidpCm; cat-pga, a MG1655 strain in which the 5�-UTR region of the pga operonis replaced by a chloramphenicol resistance cassette; pgaB pCsrB, a MG1655strain containing a pgaB mutation harboring the pCsrB plasmid; pgaB switch-csrA, a MG1655 strain containing both the pgaB mutation and the switch-csrAconstruction. wt, wild type.

FIGURE 4. Detection of intracellular CsrB small noncoding RNA levels bysemiquantitative reverse transcription-PCR. Total RNA was isolated fromovernight cultures of (wild-type) MG1655 carrying the CsrB expression plas-mid pCsrB and MG1655 containing pCm (empty vector control). 2 �g of RNAwas reverse transcribed, and the resulting cDNA was amplified for 22, 24, 26,28, and 30 cycles (n number of cycles used) using gene-specific primers. 16S rRNA was used as an internal control, with its reverse transcript co-amplifiedwith that of CsrB. PCR products formed after 22, 24, 26, and 28 cycles areincluded in the agarose gel image shown in the figure, along with a DNAladder (100, 200, 300, and 400 bp bands indicated). Densitometry indicatedthat for both the CsrB and 16 S rRNA products, cycles 22, 24, and 26 werewithin the linear range for the PCR amplification (data not shown). Expectedproduct sizes for CsrB and 16 S rRNA are 347 and 200 bp, respectively.

Role for csrA in Bacterial Autoaggregation

28742 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 • NUMBER 42 • OCTOBER 16, 2009

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

synthesis of PGA (22).We also used recombineering (which is ahomologous recombination-based technique) to accuratelyintegrate a “cm-switch” cassette (comprising a chlorampheni-col resistance gene adjacent to the riboswitch) immediatelyupstream of the RBS of the csrA gene of themutant carrying thepga-lacZ fusion. The resulting mutant strain contained boththe switch-csrA and pga-lacZ fusions, enabling us to examinehow the riboswitch regulated the production of PgaAB. As pre-dicted, the switch significantly up-regulated the levels of thepga-lacZ fusion in the absence of theophylline (i.e. in the pres-ence of caffeine), putatively due to repressed csrA expression(Fig. 6). When theophylline was present, the pgaAB expressionlevels were comparable with those of cells that contained thepga-lacZ fusion but lacked the riboswitch (Fig. 6). Thus, bothCsrB-overproducing cells and non-ligand-activated switch-csrA cells, which are both AG-positive, have increased levels ofPGA production.We then deleted pgaB to disrupt PGA expression in a switch-

csrA background. No AG was evident in the pgaB switch-csrAmutant (Fig. 3). The disruption of PGA synthesis also blockedAG in the cells carrying pCsrB (i.e. overexpressing CsrB; Fig. 3).These results suggest that PGA is required for CsrA-mediatedAG. Next, we replaced the pgaA 5�-untranslated region (UTR)with a chloramphenicol resistance gene (cat) fragment, in orderto constitutively express the genes encoding the PGA biosyn-

thesis and translocation machinery located downstream. Thenative promoter of the pga operon was left intact. The 5�-UTRof the pgaAmRNA harbors six CsrA-binding sites. By bindingto these sites, CsrA represses the translation of the pgaABCDtranscript (22). Thus, replacing the pgaA leader with the catgene cassette should “protect” the pga operon against CsrAinhibition, thereby increasing the amounts of PGA biosynthe-sized. To test whether the hybrid cat-pga strain had increasedlevels of pgaABCD expression, we inserted the lacZ codingsequence immediately prior to the stop codon of pgaB (inframe) to construct a chromosomally located translational pga-lacZ fusion. �-galactosidase assays verified that replacing thepgaA leader with the cat fragment increased the levels of thepga-lacZ fusions 4-fold (Fig. 7). Furthermore, we observedthat the hybrid cat-pga strain autoaggregated in liquid culture(Fig. 3), supplying additional evidence supporting the operationof PGA as an AG-promoting factor. Taken together, theseresults indicate that PGA plays a pivotal role in AG observedwith switch-csrA.

DISCUSSION

Here, we report the development of a new approach formak-ing hypomorphic mutations in essential genes and successfullyapply it to study the activities of the csrA gene in E. coli. Toachieve this, we created a theophylline-responsive riboswitchthat could be used to express CsrA at reduced levels within thecell, in a highly reproducible and controllable manner. Usingthis riboswitch-controlled hypomorphic mutation of csrA, wedemonstrate that the CsrA protein negatively mediates AG of

FIGURE 5. Regulatory effects of the sncRNA CsrB on the expression of anE. coli MG1655 chromosomal pga-lacZ translational fusion. CsrB produc-tion was driven by its native promoter on a high copy plasmid pCsrB. �-Ga-lactosidase activities due to the translation of the chromosomal pga-lacZfusions were normalized to those of an MG1655 strain containing a pga-lacZfusion and a control vector (pCm). Error bars, S.D. (*, p � 0.05).

FIGURE 6. The theophylline-responsive switch in switch-csrA cells con-trols pga expression. �-Galactosidase levels (relative units) resulting fromexpression of a chromosomally located pga-lacZ translational fusion weremeasured for strains where CsrA levels are regulated by a theophylline-re-sponsive riboswitch (pga-lacZ switch-csrA) or are under native control (pga-lacZ). �-Galactosidase activities were normalized to those of the (pga-lacZ)control in the presence of 5 mM caffeine (ca). PGA-LacZ levels were signifi-cantly increased in the pga-lacZ switch-csrA strain in the presence of 5 mM

caffeine (CsrA repressed) and were relatively unaffected in the presence of 5mM theophylline (th; CsrA de-repressed). Error bars, S.D. (*, p � 0.05).

FIGURE 7. Replacing the pgaA 5�-UTR with a chloramphenicol resistancecassette significantly enhances PGA production. A, �-galactosidase activ-ities (relative units) were determined for two MG1655 strains containing chro-mosomal pga-lacZ fusions: one with the native pgaA 5�-UTR (pga-lacZ) andanother where the 5�-UTR had been replaced with a chloramphenicol-resis-tance cassette fragment (cat) with the native promoter of the pga operon(pgaAp) being left intact (cat-pga-lacZ). B, the pgaAB-lacZ expression levels inthe pga-lacZ strain (normalized to 1), which contains the native arrangementof CsrA binding motifs within the 5�-UTR, are significantly lower than those ofthe cat-pga-lacZ strain, which lacks these CsrA repressor binding sites. Errorbars, S.D. (*, p � 0.05).

Role for csrA in Bacterial Autoaggregation

OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28743

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

E. coli cells. Our results further indicate that PGA plays a piv-otal role in the CsrA-mediated AG process.Essential genes encode critical cellular functions and thus

tend to be of high scientific interest. However, they are oftenhard to study genetically, since their deletion is lethal and thusnull mutations cannot be constructed. The creation of hypo-morphicmutations is an alternative strategy by which the func-tions of essential genes may be investigated in vivo. Severalapproaches have been developed for this purpose, but each hasits own disadvantages. Insertionmutagenesis, partial gene dele-tion, and ambermutations can result in the production of trun-cated proteins. However, a truncated protein may be com-pletely inactive and thus not hypomorphic. There is theadditional possibility that the functions of a truncated proteinmay differ significantly from those of its full-length counter-part. Another approach is to place the essential gene under thecontrol of an inducible promoter (4–6), but then the nativecopy number and/or regulation of the gene may be disturbed.Given these considerations, we investigated the feasibility ofusing a cis-acting inhibitory RNA device to control the expres-sion of essential genes at the translational level, so that the full-length proteins would be produced at reduced levels. The use ofa riboswitch-based hypomorphic mutation has the additionalbenefit of leaving the native promoter intact, thus preservingthe natural transcriptional levels and regulation of the targetgene.Riboswitches are structured RNAs typically located in the

5�-UTR of bacterial mRNAs, which adopt altered conforma-tions in response to an environmental signal in the form of ametabolite. The conformational changes result in the inhibitionor activation of (downstream) target gene expression. Here, weengineered a riboswitch that adopts four putative conforma-tions when inserted immediately upstream of the RBS of thecsrA gene. csrA was chosen because it is an essential gene inE. coliK12 (6) and is a global regulator controlling awide varietyof biological processes, thus meriting further detailed investi-gation. In the absence of the theophylline ligand, conforma-tions A and B predominate and interfere with ribosome bind-ing. Only one of the two predicted theophylline-boundconformations (conformation D in Fig. 1) allows for relativelyunimpeded csrA expression. Therefore, in the non-ligand-bound form, switch-csrA has low but not zero CsrA levels,whereas theophylline binding increases the csrA expression tonearly wild-type levels. The proposed control of the riboswitchover the downstream csrA expression was verified by csrA-lacZtranslational fusions and by examining the cell motility pheno-type. Thus, riboswitches, in addition to their more commonapplications in modular and universal gene regulation, can beused to construct more “subtle” hypomorphic gene mutants.Using our approach, analogous riboswitch-controlled condi-tional hypomorphic mutations may be constructed for otheressential genes, enabling their functions to be dissected. Suchriboswitches may be composed of a theophylline aptamerand a short nucleotide sequence, which, together with theRBS region of the essential gene of interest, would formmul-tiple conformations at equilibrium (after transcription). Thesequence of the theopylline aptamer would be conserved (12,13), whereas the composition of short nucleotide sequence

would be gene-specific. This short sequence should bedesigned such that the RBS of the target gene is blocked inthe RNA conformations adopted in the absence of theophyl-line, thus repressing gene translation, but is “freed up” aftertheophylline binding, facilitating gene expression. It is note-worthy that the riboswitches used for constructing hypo-morphic mutations should not completely “shut off” geneexpression in the absence of ligand (i.e.when “switched off”),and gene expression levels should be restored to those of thewild-type cells in the presence of ligand (switched on). Theswitched on cells serve as a perfect control for their hypo-morphic switched off counterparts.Notably, our use of the riboswitch-controlled hypomorphic

mutation enabled us to identify an important new function ofthe CsrA protein; it represses AG in E. coli. AG is visualizedmacroscopically by the clumping or “fluffing” of cells in liquidcultures or by the cells settling (or flocculating) upon standingwithout agitation. AG is an important virulence factor, since ithas been associated with bacterial adhesion and biofilm forma-tion (24–26) as well as enhanced protection against hostdefenses (27). Only a few AG factors have thus far been identi-fied in E. coli. These include the autotransporter protein family(19, 24, 25) and some cell surface structures, such as type IV pili(28). As described earlier, the csrA deletionmutant is not viable,and therefore the regulatory role of CsrA in AGmay have beeneasily overlooked by relying solely on traditional genetic tech-niques. In contrast, the switch-csrA strain, whose csrA expres-sion was specifically repressed by the theophylline-responsiveriboswitch, evidently displayed AG. The addition of theophyl-line, which increased the CsrA levels in the switch-csrA strain,abolished the AG phenotype. Thus, we demonstrate that ribos-witch-controlled hypomorphs can be used to identify genefunctions that traditional mutagenesis approaches may beunable to reveal.

Acknowledgment—We thank D. L. Court for providing plasmidpSim6 for the recombineering experiments.

REFERENCES1. Gerdes, S. Y., Scholle, M. D., Campbell, J. W., Balazsi, G., Ravasz, E.,

Daugherty, M. D., Somera, A. L., Kyrpides, N. C., Anderson, I., Gelfand,M. S., Bhattacharya, A., Kapatral, V., D’Souza, M., Baev, M. V., Grechkin,Y., Mseeh, F., Fonstein, M. Y., Overbeek, R., Barabasi, A. L., Oltvai, Z. N.,and Osterman, A. L. (2003) J. Bacteriol. 185, 5673–5684

2. Mushegian, A. (1999) Curr. Opin. Genet. Dev. 9, 709–7143. Herring, C. D., and Blattner, F. R. (2004) J. Bacteriol. 186, 2673–26814. Guzman, L.M., Belin, D., Carson,M. J., and Beckwith, J. (1995) J. Bacteriol.

177, 4121–41305. Judson, N., and Mekalanos, J. J. (2000) Nat. Biotechnol. 18, 740–7456. Timmermans, J., andVanMelderen, L. (2009) J. Bacteriol.191, 1722–17247. Yu, D., Ellis, H. M., Lee, E. C., Jenkins, N. A., Copeland, N. G., and Court,

D. L. (2000) Proc. Natl. Acad. Sci. U.S.A. 97, 5978–59838. Datta, S., Costantino, N., and Court, D. L. (2006) Gene 379, 109–1159. Jin, Y., Watt, R. M., Danchin, A., and Huang, J. D. (2009) BMC Genomics

10, 16510. Mathews, D.H., Sabina, J., Zuker,M., andTurner, D.H. (1999) J.Mol. Biol.

288, 911–94011. Mathews, D.H., Disney,M.D., Childs, J. L., Schroeder, S. J., Zuker,M., and

Turner, D. H. (2004) Proc. Natl. Acad. Sci. U.S.A. 101, 7287–729212. Zimmermann,G. R.,Wick, C. L., Shields, T. P., Jenison, R. D., and Pardi, A.

Role for csrA in Bacterial Autoaggregation

28744 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 • NUMBER 42 • OCTOBER 16, 2009

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

(2000) RNA 6, 659–66713. Win, M. N., and Smolke, C. D. (2007) Proc. Natl. Acad. Sci. U.S.A. 104,

14283–1428814. Brunel, C., Romby, P., Sacerdot, C., de Smit,M.,Graffe,M., Dondon, J., van

Duin, J., Ehresmann, B., Ehresmann, C., and Springer, M. (1995) J. Mol.Biol. 253, 277–290

15. Isaacs, F. J., Dwyer, D. J., Ding, C., Pervouchine, D. D., Cantor, C. R., andCollins, J. J. (2004) Nat. Biotechnol. 22, 841–847

16. Suess, B., Fink, B., Berens, C., Stentz, R., and Hillen, W. (2004) NucleicAcids Res. 32, 1610–1614

17. Wei, B. L., Brun-Zinkernagel, A.M., Simecka, J.W., Pruss, B.M., Babitzke,P., and Romeo, T. (2001)Mol. Microbiol. 40, 245–256

18. Liu, M. Y., Gui, G., Wei, B., Preston, J. F., 3rd, Oakford, L., Yuksel, U.,Giedroc, D. P., and Romeo, T. (1997) J. Biol. Chem. 272, 17502–17510

19. Henderson, I. R., Meehan, M., and Owen, P. (1997) FEMSMicrobiol. Lett.149, 115–120

20. Baker, C. S., Eory, L. A., Yakhnin, H., Mercante, J., Romeo, T., and

Babitzke, P. (2007) J. Bacteriol. 189, 5472–548121. Dubey, A. K., Baker, C. S., Suzuki, K., Jones, A. D., Pandit, P., Romeo, T.,

and Babitzke, P. (2003) J. Bacteriol. 185, 4450–446022. Wang, X., Dubey, A. K., Suzuki, K., Baker, C. S., Babitzke, P., and Romeo,

T. (2005)Mol. Microbiol. 56, 1648–166323. Jonas, K., Edwards, A. N., Simm, R., Romeo, T., Romling, U., andMelefors,

O. (2008)Mol. Microbiol. 70, 236–25724. Sherlock, O., Vejborg, R. M., and Klemm, P. (2005) Infect. Immun. 73,

1954–196325. Sherlock, O., Schembri, M. A., Reisner, A., and Klemm, P. (2004) J. Bacte-

riol. 186, 8058–806526. Danese, P. N., Pratt, L. A., Dove, S. L., and Kolter, R. (2000)Mol.Microbiol.

37, 424–43227. Ochiai, K., Kurita-Ochiai, T., Kamino, Y., and Ikeda, T. (1993) J. Med.

Microbiol. 39, 183–19028. Bieber, D., Ramer, S.W.,Wu, C. Y.,Murray,W. J., Tobe, T., Fernandez, R.,

and Schoolnik, G. K. (1998) Science 280, 2114–2118

Role for csrA in Bacterial Autoaggregation

OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28745

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from

Ye Jin, Rory M. Watt, Antoine Danchin and Jian-dong Huang Gene in Bacterial AutoaggregationcsrARole for the Essential

Use of a Riboswitch-controlled Conditional Hypomorphic Mutation to Uncover a

doi: 10.1074/jbc.M109.028076 originally published online August 25, 20092009, 284:28738-28745.J. Biol. Chem. 

  10.1074/jbc.M109.028076Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/284/42/28738.full.html#ref-list-1

This article cites 28 references, 14 of which can be accessed free at

by guest on Decem

ber 2, 2020http://w

ww

.jbc.org/D

ownloaded from