JOURNAL OF BACTERIOLOGY, May 1995, p. 2663?C2672
0021-9193/95/$04.00 0 Copyright 1995, American Society for Microbiology
Vol. 177, No. 10
The Product of the Pleiotropic Escherichia coli Gene csrA Modulates Glycogen Biosynthesis via Effects on mRNA Stability
MU YA LIU,1 HONGHUI YANG,2
TONY ROMEO1, 2*
Department of Microbiology and Immunology1 and Department of Biochemistry and Molecular Biology,2 University of North Texas Health Science Center at Fort Worth, Fort Worth, Texas 76107-2699
Received 8 September 1994/Accepted 13 March 1995
The carbon storage regulator gene, csrA, modulates the expression of genes in the glycogen biosynthesis and gluconeogenesis pathways in Escherichia coli and has been cloned, mapped and sequenced (T. Romeo, M. Gong, M. Y. Liu, and A. M. Brun-Zinkernagel, J. Bacteriol. 175:4744?C4755, 1993; T. Romeo and M. Gong, J. Bacteriol. 175:5740?C5741, 1993). We have now conducted experiments that begin to elucidate a unique mechanism for csrA-mediated regulation. Steady-state levels of glgC transcripts, encoding ADP-glucose pyrophosphorylase, were elevated by up to sixfold in a csrA::kanR mutant and were less than 6.5% of wild-type levels in a strain containing pCSR10 (csrA ), as shown by S1 nuclease protection analysis. The rate of chemical decay of these transcripts after adding rifampin to cultures was dramatically reduced by the csrA::kanR mutation. Deletion studies of a glgC - lacZ translational fusion demonstrated that the region surrounding the initiation codon was important for csrA-mediated regulation and indicated that neither csrA-mediated regulation nor stationary phase induction of glgC expression originates at the level of transcript initiation. Cell-free (S-200) extracts containing the CsrA gene product potently and speci?cally inhibited the in vitro transcription-translation of glg genes. The deduced amino acid sequence of CsrA was found to contain the KH motif, which characterizes a subset of diverse RNA-binding proteins. The results indicate that CsrA accelerates net 5 -to-3 degradation of glg transcripts, potentially through selective RNA binding. During the transition from exponential growth into the stationary phase, Escherichia coli and many other bacteria convert available carbon into -1,6-branched -1,4-D-glucan or glycogen, which subsequently is degraded as an endogenous source of
carbon and energy. Glycogen biosynthesis and utilization depend on several structural and regulatory genes (for reviews, see references 27 to 29). In E. coli, two adjacent operons, glgBX and glgCAY contain genes that are essential for glycogen synthesis. The glgB gene encodes glycogen branching enzyme (EC 18.104.22.168) (5), glgC encodes ADP-glucose pyrophosphorylase (EC 22.214.171.124) (4), and glgA encodes glycogen synthase (126.96.36.199) (19). The coding regions of glgC and glgA overlap by 1 bp, and each gene is preceded by a Shine-Dalgarno sequence indicative of a ribosome-binding site (40). Interestingly, two genes which apparently encode enzymes involved in glycogen degradation are also encoded in this gene cluster, glycogen phosphorylase (EC 2.4.21) is encoded by glgY or glgP (33, 46), and glgX encodes a putative glucanotransferase or hydrolase (33). A third unlinked monocistronic operon consists of the gene glgS, which stimulates glycogen synthesis by an unknown mechanism (15). The expression of the glg structural genes in part determines the amount of glycogen that is accumulated by cultures. The expression of the glgCAY operon is induced in stationary phase and is positively regulated by cyclic AMP (cAMP)-cAMP receptor protein (CRP) and by ppGpp, which mediate the catabolite repression and stringent response global regulatory systems, respectively (30, 35). The transcription of glgCAY depends on 70 RNA polymerase (29) and is not regulated by the alternative sigma factors s (which is the gene product of rpoS or katF ), 54 (29, 35), or 32 (29). The 5 termini of four stationary-phase-induced transcripts have been mapped within a 0.5-kb noncoding region upstream from glgC, further suggesting complex transcriptional control (35). The expression of the glgBX operon is also induced in stationary phase but is not in?uenced by cAMP or ppGpp. Transcription of the glgS gene has been shown to involve both cAMP-CRP and s (15). We recently described the molecular cloning, mapping, and characterization of a pleiotropic gene, csrA, which dramatically alters the level of glycogen that is accumulated under a variety of growth conditions and which also affects gluconeogenesis and cell surface properties (31, 32). The csrA gene was shown to encode a 61-amino-acid polypeptide which somehow negatively regulates the expression of glgB, glgC, and pckA (encoding the gluconeogenic enzyme phosphoenolpyruvate carboxy kinase [EC 188.8.131.52]). Each of these genes was still induced in the stationary phase in a csrA::kanR insertion mutant, indicating that csrA-mediated regulation is superimposed on the growth-phase regulation. The expression of glgC was strongly regulated via csrA; 10-fold higher levels of ADP-glucose pyrophosphorylase were present in a csrA::kanR insertion mutant. The effects of csrA on glycogen synthesis were mediated independently of the catabolite repression and stringent response systems, and it was suggested that csrA may encode a component of a novel global regulatory system. The present study
explores the possible mechanism of csrA-mediated regulation and suggests that the CsrA gene product is a factor which controls messenger RNA stability. (Some of the experiments described herein were conducted in partial ful?llment of the Master of Science Degree by H. Yang at the University of North Texas Health Science Center at Fort Worth.) MATERIALS AND METHODS
Chemicals and reagents. Radiolabeled [ -32P]ATP, -35S-dATP, and translation grade [35S]methionine were purchased from Dupont NEN (Wilmington, Del.). Rifampin was purchased from Sigma Chemical Co. (St. Louis, Mo.). The CRP, ppGpp, cAMP, S1 nuclease, and enzymes for DNA manipulation were
* Corresponding author. Phone: (817) 735-2121. Fax: (817) 7352118. 2663
LIU ET AL. TABLE 1. Bacterial strains and plasmids used in this study
Strain or plasmid
Source and/or reference
E. coli K-12 BW3414 TR1-5BW3414 G6MD3 HB101 DH5 GM161 MBM7060 Plasmids pUC19 pOP12 pPR1 pPR2 pPR2b pMLB1034 pCSR10 pCZ3-3 pCV1 p CZ pT CZ pC Z p C Z509 pMLC1
lacU169 BW3414 csrA::kanR Hfr his thi Strs (malA-asd) supE44 hsdS20(rB mB ) recA13 ara-14 proA2 lacY1 galK2 rpsL20 xyl-5 mtl-1 supE44 lacU169 ( 80 lacZ M15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1 F thr-1 leuB6 dam-4 thi-1 hsdS1 lacY1 tonA21 supE44 F araC Am araD lacU169 trp Am malB Am rpsL relA thi supF ( p 1048) Cloning vector, high copy number, Ampr Contains asd and glgBXCA in pBR322, Tetr 0.5 kb of glgC ?anking DNA in pUC19, Ampr glgC and glgA in pUC19, Ampr Same as pPR2 except for insert orientation, Ampr Vector for making lacZ translational fusions, Ampr csrA subcloned into pUC19, Ampr glgC - lacZ in pMLB1034, Ampr Vector for making glgC - lacZ deletions, Ampr Pre?x designation for clones with nested upstream deletions from glgC - lacZ, Ampr Pre?x designation for ?ve glgC - lacZ deletion clones with an upstream trpA terminator, Ampr Pre?x designation for clones with glgC coding region deletions from glgC - lacZ, Ampr glgC - lacZ containing 50 bp of glgC upstream ?anking DNA and 9 bp of coding DNA, Ampr Contains the lacZ promoter and part of coding region from pUC19 cloned into pMLB1034, Ampr
Barry Wanner 32 39 3 3 Arnold Revzin (1) 41 44 23 35 35 This 41 32 30 This This This This This This
study study study study study study
from the sources previously indicated (35). Polynucleotide kinase,
Sequenase 2.0, DNA sequencing reagents, and Bal 31 exonuclease were from U.S. Biochemical Corp. (Cleveland, Ohio). Moloney murine leukemia virus reverse transcriptase was from Promega (Madison, Wis.). Protein and DNA molecular weight standards were purchased from Bethesda Research Laboratories (Gaithersburg, Md.). Bacterial strains and plasmids. Table 1 lists the strains and plasmids that were used in this study, their sources, and relevant genotypes. Growth conditions. Kornberg medium (1.1% K2HPO4, 0.85% KH2PO4, 0.6% yeast extract, 0.5% glucose ) was used to grow strains for all glg gene expression studies, and LB medium (1% tryptone, 1% NaCl, 0.5% yeast extract, 0.2% glucose; pH 7.4 ) was used for routine laboratory cultures. Liquid cultures were grown at 37 C with gyratory shaking (250 rpm). For growth curve experiments, cultures were inoculated with 1 volume of an overnight culture per 400 volumes of fresh medium. Solid Kornberg medium containing 1% glucose was routinely used to grow colonies for semiquantitative staining of glycogen with iodine vapor (33). Antibiotics were used at the following concentrations (in micrograms per milliliter): ampicillin, 100; tetracycline, 10; kanamycin, 100; and rifampin, 200. Molecular biology and nucleotide sequencing. Standard procedures were used for isolation of plasmid DNA and restriction fragments, restriction mapping, transformation, and molecular cloning, as previously described (33, 35). Dideoxynucleotide sequencing (37) was performed using the Sequenase version 2.0 kit under the conditions described by the manufacturer (U.S. Biochemical Corp.). For sequencing plasmid DNA containing the upstream glgC - lacZ deletions, the pBR322 EcoRI Clockwise Primer (Bethesda Research Laboratories) was used; for sequencing the glgC - lacZ fusions that had deletions in the glgC coding region, a primer that anneals within lacZ (GATGTGCTGCAAGGC GATTAAGTTGGGTAACG) was used. Transcript mapping and stability studies. The appropriate conditions for quantitative S1 nuclease protection analysis of chromosomally encoded glgC transcripts were previously determined and described in detail, including RNA isolation, hybridization and S1 nuclease reactions, resolution of protected fragments on 4% polyacrylamide gels containing 6 M urea, and autoradiography (35). In experiments measuring the effect of csrA on steady-state levels of glgC transcripts, the RNA was puri?ed beyond the standard procedure by an additional extraction with phenol and was dissolved in GT solution (4.0 M guanidinium isothiocyanate, 0.1 M Tris [pH 7.5], 1% -mercaptoethanol) and centrifuged through a 5.7 M CsCl cushion (3). For the RNA stability studies, exponentially growing cultures were treated with rifampin to inhibit the initiation of transcription (38) and were sampled at 2-min intervals. The cells were harvested at 14,000 rpm in a microcentrifuge and frozen in solid CO2-ethanol, with no more than 2 min allowed to elapse between sampling
and freezing. The rRNA species present in each RNA preparation were examined by formaldehyde agarose gel electrophoresis (20) to assess the general quality of the RNA. In order to ensure that probe DNA was free from nicks, the probes were examined by denaturing polyacrylamide gel electrophoresis. To ensure that the protected (S1
nuclease-resistant) products were dependent on glgC expression, RNA from the glg deletion strain G6MD3 was hybridized to probes. The probe
that was used for S1 mapping of glgC transcripts in steady-state RNA analyses was a previously described uniquely labeled 5 -32P-labeled BamHI-BglI restriction fragment of pPR2 (35). The RNA stability studies used a uniquely labeled 5 -32P-labeled BamHI-HincII fragment of pPR2b, which is 70 bp longer than the BamHI-BglI fragment. For all S1 mapping experiments, labeled probe was hybridized to 50 g of total RNA at a ratio of greater than 100-fold excess relative to the glgC mRNA. The protected fragments which were generated within each reaction were applied to a single well for electrophoretic analysis. Steady-state analysis and mRNA stability experiments were conducted twice. The labeled fragments that were protected from S1 nuclease digestion were quanti?ed by densitometric analysis of the autoradiograms on a Discovery Series scanning densitometer utilizing RFLPrint version 2.0 software (PDI, Inc., New York). Several exposures of each gel were prepared and scanned to ensure that the data were collected within the linear ranges of the ?lm and the densitometer. Transcripts encoding the glgC - lacZ fusion of pT CZ40 were mapped by primer extension analysis (3) with the oligodeoxynucleotide primer CCCAGT CACGACGTTGTAAAACG. Enzyme and protein assays. Total cell protein and -galactosidase activity were quanti?ed as previously described (32). S-30 coupled transcription-translation. Experiments to measure the effects of CsrA-containing extracts on the in vitro transcription-translation of plasmidencoded genes were conducted with S-30 extracts, as previously described (35). S-30 extracts were centrifuged at 4 C for 1 h at 200,000 g, and the supernatant solutions were stored at 80 C to provide S-200 extracts. Proteins were labeled during in vitro synthesis by incorporation of [35S]methionine and denatured, and equal volumes of each reaction were subjected to electrophoresis on 9.5% sodium dodecyl sulfate-polyacrylamide slab gels. Radiolabeled proteins were detected by ?uorography using sodium salicylate (8). Preparation of deletion derivatives of the glgC - lacZ translational fusion. Plasmid clones containing nested 5 deletions from an in-frame glgC - lacZ translational fusion were prepared by ?rstconstructing a cloning vector, pCV1. This was accomplished by ligating three restriction fragments together in a single reaction: the 0.6-kb EcoRI-KpnI fragment of pCZ3-3 (30), the 35-bp EcoRIHincII polylinker fragment of pUC19, and the 140-bp KpnI-BglI fragment from pPR1. The last fragment was made blunt at the
BglI end by using the Klenow fragment prior to ligation (35). The DNA to be inserted into pCV1 was prepared by linearizing pCZ3-3 (grown in GM161) with EcoRI and by digestion for various times with Bal 31 exonuclease to generate nested deletions. The DNA was then treated with Klenow fragment to generate blunt ends and was digested with BclI. Fragments of less than 2 kb in length were isolated by electrophoresis through low-melting-point agarose and were ligated into the 5-kb BclI-SmaI fragment of pCV1 (grown in GM161). Approximately 200 of the resulting clones were analyzed by restriction analysis with HincII, and selected clones were sequenced to determine the precise endpoint of each deletion. Each of these clones was given the pre?x designation p CZ, which was followed by numbers indicating the extents of glgC upstream noncoding DNA (in base pairs) that were present.
VOL. 177, 1995
In experiments that were designed to block potential read-through transcription from the plasmid vector into the glgC - lacZ translational fusion, the trpA Rho-independent terminator (9) was cloned upstream from the glgC - lacZ fusion in ?ve of the clones to generate the clones which were designated by the pre?x pT CZ. For these experiments, a synthetic oligodeoxynucleotide (AAT
TCAGCCCGCCTAATGAGCGGGCTTTTTTTTGGATCCG) was annealed with a complementary oligonucleotide to generate a double-stranded fragment containing 5 overhangs (prepared by Biosynthesis, Inc., Lewisville, Tex.) and was phosphorylated and cloned into the dephosphorylated EcoRI sites of the plasmids. Nucleotide sequencing was used to identify clones that contained only a single terminator inserted in the desired orientation. Plasmid clones containing in-frame deletions within the coding region of the glgC-lacZ fusion were obtained by transforming strain MBM7060 with a plasmid containing an out-of-frame glgC - lacZ translational fusion and selecting for spontaneous in vivo deletions that confer the Lac phenotype, as described for pCZ3-3 isolation (30). The resulting 120 clones were restriction mapped with HincII and BamHI, and 12 of them were sequenced to determine the extent of the deletions. Each of these clones was given the pre?x designation pC Z which was followed by the amount of glgC coding DNA (in base pairs) that each contained. A clone that contained only 50 bp of glgC upstream ?anking DNA and 8 bp of coding DNA, p C Z508, was constructed by subcloning the BamHIAvaII fragment, which was made blunt at the AvaII end with the Klenow fragment, from pC Z8 into pMLB1034. The control plasmid pMLC1 was constructed by subcloning the 0.2-kb BamHI-PvuII fragment of pUC19 into pMBL1034. pMLC1 contains the lacZ promoter region upstream from an inframe lacZ coding region. Computer-assisted secondary-structure analysis. The secondary structures of KH proteins were analyzed by the method described by Garnier et al. (12) on a Macintosh computer using
GeneWorks (Intelligenetics, Inc.). Sliding windows of 17 residues were chosen for the predictions.
EFFECTS OF csrA ON glg TRANSCRIPT STABILITY
RESULTS Effects of the carbon storage regulator gene of E. coli, csrA, on steady-state levels of glgC transcripts. Previous measurements of the levels of ADP-glucose pyrophosphorylase and glgC - lacZ-encoded -galactosidase in csrA and csrA::kanR strains indicated that the gene product of csrA (CsrA) strongly regulates glgC expression. Therefore, glgC was chosen as the model gene for studying the regulatory mechanism of CsrA. The 5 ends of chromosomally encoded glgC transcripts were mapped by S1 nuclease protection analysis, and their relative levels of abundance were determined for four isogenic strains that differed in their csrA genotypes (Fig. 1 and 2) (Table 2). The overall pattern of transcription was observed to be identical to that of E. coli B and E. coli K-12 3000 (35); four transcripts occur within the upstream ?anking region of glgC, each of which is present in higher levels in early-stationaryphase versus exponential-phase growth. The levels of these transcripts were found to be negatively affected by csrA. Strain TR1-5BW3414 (csrA::kanR) accumulated approximately 4- to 6-fold higher levels of transcripts A, B, and C than did BW3414 (csrA ). The probe that was used to map transcripts in Fig. 1 was not resolved from the fragment protected by transcript D. However, densitometric analysis of data from the experiment shown in Fig. 2 indicated that transcript D was elevated 1.7fold by the csrA::kanR mutation. Transformation of E. coli with the csrA-encoding plasmid pCSR10 was previously shown to strongly inhibit accumulation of glycogen (32). This plasmid caused a severe decrease in the steady-state levels of the glgC transcripts. The relative concentration of the major glgC transcript (transcript B) was less than 2% in a pCSR10-containing strain compared with that of an isogenic csrA::kanR strain and was less than 6.5% of that of a csrA strain. Consistently with the previous observation that csrA affects the expression of glgB, glgC, and pckA in both the exponential and the stationary phases of growth (32), the csrA::kanR mutation affected glgC transcript levels in both growth phases. Effects of csrA on glgC mRNA stability. The increase in levels of glgC transcripts in the csrA::kanR strain could be explained by an increase in the rate of synthesis of these transcripts or by a decrease in the rates of their degradation, i.e., by increased
FIG. 1. S1 nuclease protection analysis of glgC transcripts from isogenic strains BW3414 (csrA ), TR1-5BW3414 (csrA::kanR), TR1-5BW3414(pUC19), and TR1-5BW3414(pCSR10) (csrA overexpressing). Total RNA was extracted from BW3414 (lanes 1 and 6), TR1-5BW3414 (lanes 2 and 7), TR1-5BW 3414(pUC19) (lanes 3 and 8), TR1-5BW3414(pCSR10)
(lanes 4 and 9), and G6MD3 ( glg) (lane 5), and transcripts were analyzed by S1 nuclease protection as described in Materials and Methods. Samples in lanes 1 through 5 show results from cells harvested in log phase; lanes 6 through 9 were from early-stationaryphase cells. The letters A through D represent protected fragments. In this analysis, the fragment protected by transcript D was not resolved from the full-length BamHI-BglI probe. The indicated size markers consisted of the intact probe (618 bp) and fragments prepared from the probe by digestion with HinfI (488 bp) or AvaII (245 bp).
mRNA stability. Therefore, we examined the effect of csrA on the chemical decay of the glgC message after the addition of rifampin to exponentially growing cultures of BW3414 (csrA ) or TR1-5 (csrA::kanR). Steady-state levels of glgC transcripts in the pCSR10-containing strain were not suf?cient for half-life analysis. RNA isolated from the cultures was subjected to S1 nuclease protection mapping and was quanti?ed by densitometry. As shown in Fig. 2 and 3, the csrA::kanR mutation had a striking effect of on the stability of glgC transcripts in these strains. The decay curves of the transcripts exhibited biphasic decay, with a lag period or period of slow decay that was followed by exponential decay. The major transcripts A and B did not exhibit exponential decay within 12 min after the addition of rifampin (approximately one-half of a generation). In two independent experiments, it was observed that 90% of transcripts A and B were degraded in the csrA strain before any signi?cant changes were seen in the csrA::kanR mutant. Figure 2 also shows that some of the full-length probe was protected against S1 digestion. This may have been due to read-through transcription from the upstream glgBX operon, since it was not observed when RNA was prepared from a strain from which the glg genes had been deleted (Fig. 1). The stability of the full-length probe was less dramatically affected than the glgC proximal transcripts by the csrA::kanR mutation (Fig. 2; densitometry data not shown), which is consistent with the observation that glgB expression is less strongly regulated than that of glgC (32). A series of transcripts which appear to be degradation products resulting from endonucleolytic cleavage of the primary transcripts were observed on extended exposure of the autoradiogram (Fig. 2B). The overall patterns of these products were identical in the two strains, although their levels were higher in the csrA::kanR strain and their rate of decay was greater in the csrA strain.
LIU ET AL.
FIG. 2. Stability of glgC transcripts in csrA and csrA::kanR strains. Total RNA was isolated and analyzed by S1 nuclease protection from exponential-phase cells (A600 1.0) of BW3414 (csrA ) (lanes 1 through
6) and TR1-5BW3414 (csrA::kanR) (lanes 7 through 12). The times elapsed (in minutes) after the addition of rifampin are indicated above each lane. Shown are 6 (A) and 24 (B) h of exposure of the same polyacrylamide gel, such that apparent degradation products are allowed to be visualized. Note that in this analysis the fragment protected by transcript D was resolved from the full-length BamHI-HincII probe. The indicated size markers consisted of the intact probe (688 bp) and fragments prepared from the probe by digestion with HinfI (488 bp) or AvaII (245 bp).
CsrA-containing S-200 extracts inhibit the in vitro
transcription-translation of glg genes. We previously observed that the expression of glgC and glgA genes in S-30 extracts was activated by the trans-acting factors cAMP, CRP, and ppGpp (35). In order to reconstruct csrA-mediated regulation in vitro, S-30 extracts were prepared from the csrA::kanR strain TR15BW3414, and high-speed (S-200) extracts were prepared from this strain and from a strain that overexpressed the csrA gene, TR1-5BW3414(pCSR10). Extracts from the latter strain contained elevated levels of CsrA protein, as determined by Western blot (immunoblot) analysis using polyclonal antiserum prepared against a synthetic peptide composed of residues 38 to 48 of the deduced amino acid sequence of CsrA, KEVSVHREEIY (data not shown). The S-30 extract contained the nondialyzable cellular factors needed for transcription-translation of plasmid-encoded glg genes; S-200 extracts were further centrifuged to remove ribosomes and other large macromolec-
TABLE 2. S1 nuclease protection analysis of the effects of csrA on glgC transcriptsa
Strain Growth phase Relative levels of transcripts ( range)b A B C
BW3414 TR1-5BW3414 TR1-5BW3414 (pUC19) TR1-5BW3414 (pCSR10)
Exponential Stationary Exponential Stationary Exponential Stationary Exponential Stationary
6.1 (0.7) 17 (4) 38 (6) 49 (1) 24 (4) 56 (7) 1.5 1.5
12 (1) 27 (4) 69 (6) 93 (7) 57 (1) 93 (8) 1.5 1.5
2.0 (0.4) 7.0 (1.7) 6.5 (1.9) 20 (3) 4.2 (0.4) 19 (1) 1.5 1.5
a BW3414 and TR1-5BW3414 are csrA and csrA::kanR strains, respectively; pCSR10 contains the wild-type csrA gene cloned into pUC19 and results in overexpression of the CsrA protein. b Densitometry data were collected from the experiment shown in Fig. 1 and from an identical, independently conducted experiment. Arbitrary integration units were normalized for each datum set with respect to the highest value of that experiment (designated as 100), and the mean values and ranges were determined.
ular complexes. Both kinds of extracts were treated with micrococcal
nuclease to degrade endogenous nucleic acids (35), and expression in the S-30 extracts was observed to be completely plasmid dependent (data not shown). As shown in Fig. 4, when the S-30 extracts were programmed with pOP12 DNA, expression of glgB and glgC was clearly observed, as was that of the asd gene, which encodes aspartate semialdehyde dehydrogenase, an enzyme that is not involved in glycogen synthesis (35). As previously observed, the expression of glgC under these conditions was stimulated upon the addition of the activators cAMP-CRP, and ppGpp, while the expression of glgB was not affected. However, the expression of glgA was weaker than that previously observed by using S-30 extracts prepared from csrA strains (35). Addition of the CsrA-containing S-200 extract severely inhibited glgC expression in both the basal reactions and in reactions which were activated via cAMP, CRP, and ppGpp. The CsrA-containing extract also inhibited glgB expression but appeared to cause little or no inhibition of glgA expression, indicating that relative expression of glgA versus glgC was greater in the presence of CsrA. In contrast to these results, the in vivo expression of a chromosomally encoded glgA - lacZ fusion exhibited strong negative regulation via csrA (43). The expression of asd showed little or no effect of the CsrAcontaining extract in the absence of the glgCA activators cAMP-CRP and ppGpp. The expression of asd was enhanced by the addition of the CsrA-containing extract to reaction mixtures which contained these activators. We suspect that in the latter case, inhibition of glg gene expression by CsrA simply relieves competition between asd and the glg genes for one or more components of the transcription-translation reaction, since no stimulatory effect on asd occurred in the absence of the activators, in which case glg expression represented a smallerfraction of the total expression. The speci?c inhibitory effects of CsrA-containing extracts on glg gene expression have been reproducibly observed in several experiments and in the two different S-30 extracts which have been tested. The deduced amino acid sequence of CsrA contains the KH motif, a putative RNA-binding domain. Previously, we were unable to identify genes or proteins that are homologous to
VOL. 177, 1995
EFFECTS OF csrA ON glg TRANSCRIPT STABILITY
FIG. 3. Densitometric analysis of glgC transcript stability. Autoradiograms from the experiment shown in Fig. 2 were analyzed by densitometric scanning, as described in Materials and Methods. The values for a given transcript in each strain were normalized relative to the amount of that transcript at the time of addition of rifampin ( T 0 min). (A through D) Results for transcripts A through D, respectively, from strains BW3414 (csrA ) (open squares) and