Gene Expression Regulation by the Curli Activator CsgD Protein: Modulation of Cellulose Biosynthesis and Control of Negative Determinants fo
http://www.100md.com
《细菌学杂志》
Swiss Federal Institute of Environmental Technology (EAWAG), uberlandstrasse 133, CH-8600 Dubendorf, Switzerland,University of Milan, Via Celoria 26, 20100 Milan, Italy,CNRS-INSA-UCBL, 10, rue Dubois, 69622 Villeurbanne Cedex, France
ABSTRACT
Curli fibers, encoded by the csgBAC genes, promote biofilm formation in Escherichia coli and other enterobacteria. Curli production is dependent on the CsgD transcription activator, which also promotes cellulose biosynthesis. In this study, we investigated the effects of CsgD expression from a weak constitutive promoter in the biofilm formation-deficient PHL565 strain of E. coli. We found that despite its function as a transcription activator, the CsgD protein is localized in the cytoplasmic membrane. Constitutive CsgD expression promotes biofilm formation by PHL565 and activates transcription from the csgBAC promoter; however, csgBAC expression remains dependent on temperature and the growth medium. Constitutive expression of the CsgD protein results in altered transcription patterns for at least 24 novel genes, in addition to the previously identified CsgD-dependent genes. The cspA and fecR genes, encoding regulatory proteins responding to cold shock and to iron, respectively, and yoaD, encoding a putative negative regulator of cellulose biosynthesis, were found to be some of the novel CsgD-regulated genes. Consistent with the predicted functional role, increased expression of the yoaD gene negatively affects cell aggregation, while yoaD inactivation results in stimulation of cell aggregation and leads to increased cellulose production. Inactivation of fecR results in significant increases in both cell aggregation and biofilm formation, while the effects of cspA are not as strong in the conditions tested. Our results indicate that CsgD can modulate cellulose biosynthesis through activation of the yoaD gene. In addition, the positive effect of CsgD on biofilm formation might be enhanced by repression of the fecR gene.
INTRODUCTION
Most bacteria are capable of surface colonization and biofilm formation through the production of specific adhesins and extracellular structures. Curli fibers (also known as thin aggregative fimbriae) are a major factor in adhesion to surfaces, cell aggregation, and biofilm formation in many enterobacteria (11, 36, 42, 43, 53). Expression of curli is linked to cellulose biosynthesis, which leads to the production of an extracellular matrix and results in tight cell-cell and cell-surface interactions and in the so-called rdar morphotype in Salmonella (45, 57, 58). Expression of both curli and cellulose depends on the CsgD protein, a putative transcription regulator of the LuxR family, which activates transcription of the csgBAC operon (2), which encodes curli structural subunits, and transcription of the adrA gene, a positive effector of cellulose biosynthesis (45). In addition to csgBAC activation by CsgD, production of curli is subject to complex regulation, which affects both the csgDEFG operon (encoding the CsgD transcription regulator and the CsgEFG curli-specific transport system) and the csgBAC operon (encoding curli structural subunits) (8, 20, 22, 53). Curli expression is dependent on different environmental and physiological cues, such as a low growth temperature (<32°C), low osmolarity, and slow growth or starvation (i.e., conditions usually encountered by the bacteria outside the mammalian host) (19, 22, 31, 36, 44). However, curli are an important virulence factor in some Salmonella and pathogenic Escherichia coli strains, in which temperature-dependent regulation can be bypassed and curli expression can also take place at 37°C (3, 4, 37, 38). In contrast, curli operons are cryptic in a large number of both clinical and environmental E. coli isolates, as well as in laboratory strains, despite the presence of functional csg genes. However, mutations either in the specific promoters (44, 52) or in global regulatory genes, such as hns (2) or ompR (53), can restore the expression of curli-encoding genes.
The product of the CsgD-dependent adrA gene is a member of the GGDEF protein family (16, 50). The AdrA protein can catalyze the synthesis of bis-(3',5')-cyclic diguanylic acid (cyclic di-GMP), which in turn stimulates the enzymes responsible for cellulose production (48). In addition to the genes coding for factors directly involved in the curli-cellulose extracellular matrix, the CsgD protein positively regulates glyA, which encodes the glycine biosynthethic enzyme serine hydroxymethyltransferase (10), and represses the dipeptidase-encoding pepD gene (7). The promoters of the csgBAC operon and of the adrA and pepD genes share a conserved 11-bp sequence (CGGGKGAKNKA), which is necessary for CsgD-dependent regulation (7).
The pepD gene was identified as a CsgD-dependent gene using a whole-genome expression approach in which the following two laboratory strains of E. coli were compared: PHL565, which is unable to produce curli, and a spontaneous curli-producing mutant, PHL628 (7). The PHL628 strain has a mutation in the ompR gene that results in a single leucine-to-arginine substitution (ompR234 allele); the ompR234 mutation increases transcription of ompR-dependent genes, including csgD (53), and stimulates specific DNA binding by the OmpR234 protein at the csgD promoter (25, 42). However, comparisons of laboratory strains unable to express curli with curli-producing mutants with mutations in global regulatory genes, such as ompR, do not allow precise evaluation of the direct contribution of the CsgD protein to gene regulation. To circumvent this problem, in this work we transformed the curli-deficient PHL565 strain either with a plasmid that allowed constitutive, low-level expression of the csgD gene or with a control vector, and we compared whole-genome expression in the two strains. We found that the CsgD protein controls the modulation of cellulose production through activation of yoaD, a putative cyclic di-GMP esterase-encoding gene. Constitutive CsgD expression affects the transcription of a number of genes and represses the fecR and cspA regulatory genes, which play a negative role in biofilm formation and cell-cell aggregation.
MATERIALS AND METHODS
Bacterial strains and plasmids. For this study, we used E. coli K-12 laboratory strain PHL565 (53) and derivatives of this strain (Table 1). Bacterial cells were grown either in Luria-Bertani broth (LB) or in M9Glu/sup (M9 minimal medium supplemented with 0.4% glucose and 2.5% LB). When necessary, ampicillin (100 μg/ml), kanamycin (50 μg/ml), or chloramphenicol (25 μg/ml) was added. For csgD expression, the csgD gene was cloned from the pCP900 plasmid (42) into the pT7-7 plasmid using the NdeI and PstI sites to obtain the pT7-CsgD plasmid, in which csgD was under the control of a phage T7 RNA polymerase-dependent promoter. For yoaD overexpression studies, the yoaD gene was amplified by PCR using PHL565 genomic DNA as the template and the yoaDFw and yoaDRev primers (Table 2). The PCR product was cloned into pGEM-T Easy, in which the yoaD gene was placed under the control of the Plac promoter, using the following primers: yoaDfwr (5'-ATGCAAAAAGCACAACGG-3') and yoaDrev (5'-GTTAACGTAACGGCATAATG-3'). MG1655 mutant strains carrying either yoaD, fecR, or cspA mutant alleles were obtained from the laboratory of F. Blattner, University of Wisconsin (http://www.genome.wisc.edu/functional/tnmutagenesis.htm). The mutant alleles were transferred into PHL565 by P1 transduction (33).
CsgD localization experiments. Cell fractionation was performed as described previously (12). Five hundred-milliliter cultures of PHL565/pT7-7 and PHL565/pT7-CsgD were grown in M9Glu/sup at 30°C for 15 h. The cells were harvested by centrifugation at 7,000 rpm for 10 min at 4°C, washed, and resuspended in 20 ml phosphate-buffered saline (PBS). Cells were disintegrated by sonication and centrifuged as described above to remove unbroken cells. The supernatant from the low-speed centrifugation was centrifuged at 100,000 x g for 1 h at 4°C to separate the cytoplasm (supernatant) and the membrane fraction (pellet). The pellet was washed with 2 ml of 2% Sarkosyl in PBS, left for 20 min at room temperature, and centrifuged at 40,000 x g at 10°C for 10 min to remove ribosomes and cytoplasmic proteins that were still associated with the membrane fraction. The pellet was resuspended in 1 ml of 1% Sarkosyl, incubated, and centrifuged as described above. The supernatant, corresponding to inner membrane proteins, was collected, and the pellet, corresponding to outer membrane proteins, was resuspended in 0.5 ml H2O. Protein concentrations were determined, and either 40 μg (for cytoplasmic fractions) or 20 μg (for membrane fractions) of total proteins was loaded onto a 12.5% sodium dodecyl sulfate (SDS)-polyacrylamide gel. Specific bands were identified by mass spectrometry of the peptide products after in-gel trypsin digestion (9).
Biofilm formation and cell aggregation assays. Biofilm formation in microtiter plates was determined essentially as described previously (15). Cells were grown in liquid cultures in microtiter plates (0.2 ml) for 18 to 20 h either in M9Glu/sup or in LB at 30°C; the liquid medium was removed, and the cell density was determined spectrophotometrically by determining the optical density at 600 nm (OD600). Cells attached to the microtiter plates were washed twice gently with PBS and stained for 20 min with 1% crystal violet (CV) in ethanol. The stained biofilms were washed with tap water and dried. For semiquantitative determination of biofilms, CV-stained cells were resuspended in 0.2 ml of 70% ethanol by vigorous pipetting. The A600 of each sample was determined and normalized to the OD600 of the corresponding liquid cultures. For cell aggregation assays, cultures were grown overnight in 3 ml of M9Glu/sup at 30°C in 15-ml Falcon tubes. Overnight cultures were left to stand for 24 h at room temperature to allow sedimentation of cell aggregates. In order to obtain better visualization of sedimented cells, the supernatant was removed by gentle pipetting, and the cell aggregates were fixed and stained with 1% crystal violet for 15 min. After destaining with H2O, the CV-fixed and -stained cell aggregates were dissolved in 70% ethanol for semiquantitative analysis by using a procedure similar to the procedure used for adhesion assays.
RNA isolation, cDNA labeling, and microarray data analysis. Total RNA from E. coli cells that were grown for 15 h in M9Glu/sup at 30°C to the stationary phase was isolated as described by Sambrook et al. (46). RNA samples were quantified spectrophotometrically, and the quality of the samples was checked by electrophoresis on agarose gels. For gene array experiments, fluorescently labeled cDNA from 50 μg of total RNA was produced using a CyScribe first-strand cDNA labeling kit (Amersham Biosciences) with either Cy3- or Cy5-dCTP. Labeled cDNA was pooled, purified with a Minielute PCR purification kit (QIAGEN), and then concentrated with a Microcon-30 (Millipore) prior to addition of the hybridization buffer. The resulting cDNA was hybridized to an E. coli K-12 V2 array (MWG) by following the manufacturer's instructions. Microarray slides were scanned using an Affimetrix 428 array scanner. Spots and corresponding background signals of the resulting 16-bit TIFF images were quantified using the Affimetrix Jaguar software, version 2. Subsequent data analysis was performed using the program GeneSpring 4.1 from Silicon Genetics. Induction factors (PHL565/pT7CsgD compared to PHL565/pT7-7) were calculated using the Cy3 and Cy5 signal intensities of each spot. Spots with signals whose values were less than 50 were excluded from the analysis. Normalization was performed using the 50th percentile distribution of the spots remaining after background correction, and genes with expression levels similar to the background level were excluded prior to normalization. Data from three independent gene array experiments were averaged. Only spots whose final average ratios were higher than or equal to 4 and that had a signal ratio greater than 2.5 in all three experiments were considered significant. Finally, single genes that showed significant induction but belonged to an operon in which no other gene was affected by CsgD expression were also discarded. The function and sequence homology of genes of interest were determined using the following databases: Colibri (http://genolist.pasteur.fr/Colibri/genome.cgi) (32), Swiss-Prot (http://www.expasy.org/) (18), and EcoCyc (http://www.ecocyc.org/) (26).
Real-time PCR analysis. Real-time PCR was performed with the same RNA that was used for the gene array experiments. Reverse transcription was carried out according to the manufacturer's instructions (Applied Biosystems) by using MultiScribe reverse transcriptase (62.5 U/μg) of total RNA in the presence of 1.25 μM random hexamers. Real-time PCRs were performed using the SYBR Green PCR master mixture, and the results were determined with an ABI Prism 7000 sequence detection system. In a 25-μl reaction mixture, the cDNA produced from 20 ng of total RNA was used. The primers used are listed in Table 2. All reactions were performed in triplicate, along with reactions with negative control samples using DNase I-digested RNA as templates in order to verify the lack of residual DNA. The relative amounts of the transcripts were determined by normalization to 16S rRNA. For comparison of yoaD expression in PHL628 and yoaD expression in PHL565, RNA was isolated from cells harvested in the exponential phase (OD600, 0.25), at the onset of the stationary phase (OD600, 0.8), and in the stationary phase (OD600, 2.5) in M9sup.
Other methods. -Glucuronidase assays were performed as described previously (33), except that hydrolysis of para-nitrophenyl--glucoronide was determined spectrophotometrically at A405. Congo red binding was performed on agar media plates as previously described (45). For determination of cellulose production cells were grown on calcofluor agar medium (45), gently scraped off the plates, and resuspended to an OD600 of 0.2, and the OD366 of the suspension was determined.
RESULTS
CsgD constitutive expression and cell localization. In the pT7-CsgD plasmid, the csgD gene is under the control of a T7 RNA polymerase-dependent promoter. However, in E. coli strains such as PHL565, which do not carry the T7 RNA polymerase-encoding gene, detectable csgD transcription can still take place and most likely depends on recognition by bacterial RNA polymerase of promoter-like sequences upstream of the csgD gene. According to real-time PCR experiments, in PHL565/pT7-CsgD the amount of csgD transcript was roughly 100-fold greater than the amount in the PHL565 strain carrying the control vector pT7-7 (Table 3), in which csgD expression was negligible. The level of the csgD transcript in PHL565/pT7-CsgD was 1.5- to 2-fold higher than the levels in other curli-proficient E. coli strains, such as the ompR234 mutant PHL628 (53) or WK2, a curli-producing environmental isolate, as determined by real-time PCR (Landini, unpublished data). However, unlike curli-producing strains in which csgD expression is driven from its own promoter, the levels of csgD transcription in PHL565/pT7-CsgD do not vary significantly in different growth conditions, again as determined by real-time PCR experiments (data not shown). Thus, we concluded that pT7-CsgD allows constitutive csgD expression totally uncoupled from physiological and environmental signals, such as growth phase and osmolarity, which, in contrast, control the expression of the csgDEFG promoter (2, 8, 42). No band corresponding to the CsgD protein was detectable in crude extracts of PHL565/pT7-CsgD as determined by SDS-polyacrylamide gel electrophoresis (PAGE) (data not shown). However, analysis of the different cell compartments (cytoplasm, inner membrane, and outer membrane) after fractionation of cells grown overnight at 30°C in M9Glu/sup led to identification of a 25-kDa band present only in the cytoplasmic membrane fraction of PHL565/pT7-CsgD, where it accounted for only a small percentage of the total proteins (Fig. 1). In-gel trypsin digestion of the protein followed by mass spectrometry analysis confirmed that this band indeed corresponded to CsgD. Unlike CsgD, proteins that form inclusion bodies, such as green fluorescent protein, are not readily solubilized by Sarkosyl treatment and are not found in the inner membrane after cell fractionation (data not shown). Thus, CsgD localization in the cytoplasmic membrane did not appear to be due to the formation of inclusion bodies or to other artifacts that depended on nonphysiological CsgD expression. Expression of CsgD from pT7-CsgD did not result in any major changes in protein expression at a level detectable by one-dimensional SDS-PAGE analysis. Interestingly, however, the Dps protein, found in the outer membrane protein compartment in PHL565 transformed with the pT7-7 vector, was present at a much lower level in PHL565/pT7-CsgD.
Effects of constitutive csgD expression on adhesion and curli expression. We tested how production of CsgD from the pT7-CsgD plasmid affects surface attachment in the PHL565 strain. As shown in Fig. 2A, CsgD expression from the pT7-CsgD plasmid resulted in a fourfold increase in surface attachment by the PHL565 strain in M9Glu/sup and in a roughly twofold increase in LB. A similar degree of surface attachment stimulation was observed previously for PHL628, an ompR234 mutant derivative of PHL565 (42). The ompR234 mutation results in increased transcription from the csgD promoter, thus allowing CsgD-directed curli and cellulose biosynthesis and biofilm formation (53). The PHL565/pT7-CsgD strain formed red colonies when it was plated on growth medium supplemented with the amyloid protein-binding dye Congo red (Fig. 2B), suggesting that curli production was induced in this strain. Increased Congo red binding and surface colonization induced by pT7-CsgD were indeed dependent on curli, since transformation with pT7-CsgD of the PHL856 strain, a PHL565 derivative in which the csgA gene encoding the main curli subunit has been inactivated, did not result in surface attachment (Fig. 2A) or in Congo red binding (Fig. 2B) by this strain. Thus, csgD expression from the pT7-CsgD plasmid led to the production of a functional CsgD protein and conferred an adherent, curli-expressing phenotype to the PHL565 strain.
Since CsgD expression from the pT7-CsgD plasmid was independent of the csgDEFG promoter, the observation that pT7-CsgD induced biofilm formation by PHL565 more efficiently in M9Glu/sup than in LB strongly suggested that medium-dependent curli regulation takes place at a step after csgD transcription. Thus, we tested the effects of different media, as well as temperature and osmolarity (the main environmental signals regulating curli production), on transcription from the csgBAC promoter in the presence of constitutively expressed CsgD protein. Transcription from the csgBAC promoter was determined in the PHL856 strain transformed with either pT7-7 or pT7-CsgD; in PHL856 the csgA gene is interrupted by a uidA gene encoding -glucuronidase. csgBAC transcription was significantly lower (up to 25-fold) in LB (Fig. 3B) than in M9Glu/sup (Fig. 3A), particularly during the exponential phase of growth. Only in the late stationary phase did the level of csgBAC transcription in LB increase to almost one-half the level of transcription in M9Glu/sup.
Growth at 37°C resulted in a reduction in csgBAC transcription in both LB and M9Glu/sup (Fig. 3), and this effect again appeared to be more significant in the exponential phase (up to 10-fold reduction observed) than in the late stationary phase (2.5- to 3-fold reduction). Thus, temperature-dependent regulation did not appear to take place at the level of csgD transcription. Indeed, the levels of transcription from the csgDEFG promoter appeared to be similar at 30°C and 37°C in PHL857, a CsgD-expressing mutant derivative of PHL565 (data not shown). Recently published observations suggest that in E. coli temperature-dependent regulation of curli is mediated by the product of the crl gene, which acts as the temperature sensor at the csgBAC promoter (6). To further investigate this possibility, we transformed the EB9 strain, a crl920::cam derivative (40) of PHL856, with either pT7-7 or pT7-CsgD, and we measured csgBAC transcription. The crl mutation resulted in a clear reduction in csgBAC transcription at 30°C, while it did not have any effect at 37°C, in agreement with the proposed role of the crl gene (Fig. 3).
In contrast to the substantial changes induced by media and temperature, the effects of growth medium osmolarity on csgBAC transcription in PHL565/pT7-CsgD were modest. Addition of up to 0.25 M NaCl to M9Glu/sup resulted in a less-than-twofold reduction in csgBAC transcription, which was, in contrast, totally abolished in the PHL857 strain, in which CsgD was expressed from its own promoter (Fig. 4).
Effects of constitutive CsgD expression on whole-genome transcription. The results of both adhesion and csgBAC transcription experiments showed that constitutive expression from pT7-CsgD led to production of an active CsgD protein. Thus, we performed a whole-genome transcription assay in which we compared PHL565 strains that were transformed with either pT7-7 or pT7-CsgD and were grown in M9Glu/sup at 30°C (i.e., the optimal conditions for curli expression). We considered an average difference in gene expression that was equal to or greater than fourfold significant (see Materials and Methods). Ten genes were found to be up-regulated and 14 genes were found to be down-regulated in the PHL565 strain transformed with pT7-CsgD (Table 3). Among the up-regulated genes we found, as expected, csgBAC and adrA, which are known to be CsgD regulated, and csgD itself. Increased csgD transcription was exclusively due to the presence of the pT7-CsgD expression vector and was independent of the csgDEFG promoter, as indicated by a lack of any increase in csgEFG transcripts (see the results for real-time PCR experiments). In addition to known CsgD-dependent genes, we found three genes with as-yet-unknown functions (yaiB, yjgW, and ytfI) and two genes (ymdA and yoaD) whose putative functions can be predicted based on the amino acid sequences of their products. The ymdA gene encodes a hypothetical protein similar to proteins in the fimA/papA fimbrial protein family and is located 120 bp downstream of the csgBAC operon. The yoaD gene encodes a member of the EAL protein family, which is thought to be responsible for degradation of cyclic di-GMP, a signal molecule able to trigger cellulose biosynthesis, biofilm formation, and different cellular processes in several gram-negative bacteria (23). Interestingly, the CsgD-activated adrA gene encodes a diguanylate cyclase, the biosynthetic enzyme for cyclic di-GMP (45, 48, 57), suggesting that both intracellular accumulation and degradation of cyclic di-GMP are mediated by CsgD-regulated genes. In addition to adrA and yoaD, a third GMP-related gene, gsk, was found to be more highly expressed in PHL565/pT7-CsgD. The Gsk protein is a GMP synthase belonging to the nucleoside salvage pathway (24, 27) and might be involved in either repletion or maintenance of the GMP cellular pool.
All 14 genes that were down-regulated in PHL565/pT7-CsgD have known functions, and none of them has yet been shown to be regulated by CsgD. The pyrBI operon encodes the two subunits of aspartate carbamoyl transferase, an enzyme that is part of the pyrimidine biosynthetic pathway (28, 55). The gatA, gatC, and gatZ genes listed in Table 3 belong to the gatYZABCDR operon, encoding a phosphoenolpyruvate-dependent phosphotransferase system transporter specific for the sugar galactitol (35, 39). Although the gatY, gatB, gatD, and gatR genes are not listed in Table 3, they were all down-regulated in PHL565/pT7-CsgD by factors ranging from 2.7- to 3.3-fold, suggesting that the whole gat transcription unit is indeed repressed in a CsgD-dependent fashion (data not shown). Two genes encoding outer membrane proteins, the main OmpF porin and the OmpT protease, as well as the methionine biosynthesis metA gene, were also down-regulated in PHL565/pT7-CsgD (Table 3).
The five remaining genes which are down-regulated by constitutively expressed CsgD belong to the following two functional groups: iron-sensing genes and cold shock-responding genes. Our assays showed that there was 4.5-fold repression of transcription of both the fecR and fhuE genes in PHL565/pT7-CsgD (Table 3). The outer membrane FhuE protein serves as a receptor for ferric coprogen and ferric rhodotorulic acid, which upon binding by FhuE can be taken up via the TonB system (47). The periplasmic FecR protein plays a role in iron sensing and in regulation of the alternative sigma factor FecI. The two genes are cotranscribed in the fecIR operon (41, 49), and the level of the fecI transcript is 2.5-fold lower in PHL565/pT7-CsgD (data not shown), which is consistent with the possibility that there is either direct or indirect transcriptional repression of the whole fecIR operon by CsgD. Finally, three of the main cold shock-induced genes in E. coli (cspA, cspB, and cspG) (29, 54) appeared to be down-regulated by factors of four- to sevenfold in PHL565/pT7-CsgD. In addition, the level of the transcript of the infA gene, which is known to respond to cold shock (21), was also fourfold lower in the CsgD-expressing strain.
Real-time PCR analysis of genes differentially expressed in PHL565/pT7-CsgD. A selection of genes that were found to be differentially expressed in the whole-genome transcription assay were tested in real-time PCR experiments. First, we tested genes belonging to the csgBAC and csgDEFG operons and genes proposed to be CsgD dependent (adrA, glyA, and pepD). Real-time PCR experiments showed that the levels of expression of the csg genes were 10- to 100-fold-higher than the levels in the whole-genome transcription experiment (Table 3). The large differences were probably due to the difficulty of evaluating precisely the very low levels of expression of csg genes in the PHL565 strain. In contrast, real-time PCR and whole-genome transcription assays yielded very similar results for all other genes tested (Table 3).
Unlike the levels of transcription of the csgBAC operon, the levels of transcription of the csgE, csgF and csgG genes (i.e., the csgDEFG operon) were not altered by the presence of the pT7-CsgD plasmid, strongly suggesting that the CsgD protein does not regulate its own gene and indicating that, despite the presence of a functional chromosomal copy of csgD in the PHL565/pT7-CsgD strain, csgD expression is solely dependent on the plasmid copy of the gene.
Transcription of the glyA gene was only weakly stimulated (1.3-fold) by the CsgD protein according to the results of our real-time experiments (Table 3). Although CsgD has been shown to suppress glycine autotrophy and to stimulate serine hydroxymethyltransferase activity via the glyA gene (10), little CsgD-dependent stimulation (1.2- to 2-fold) of glyA transcription was detectable when cells were grown in minimal medium supplemented with amino acids (10), which is consistent with our results. In contrast, the pepD gene appeared to be 4-fold negatively regulated by CsgD in real-time PCR experiments, similar to the 4-fold repression observed for the curli-expressing PHL628 derivative of PHL565 (7) and to the 2.5-fold repression observed for the PHL565/pT7-CsgD gene array compared with the PHL565/pT7-7 gene array (Table 3).
Among the novel identified genes whose expression was affected by CsgD, we tested the EAL protein-encoding yoaD gene, as well as the fecR and cspA genes, in real-time PCR experiments (Table 3). For these genes, the real-time PCR results closely reflected the results of the whole-genome transcription assays, showing that there was an eightfold increase for the yoaD transcript and a roughly fourfold decrease for both the cspA and fecR transcripts in PHL565/pT7-CsgD.
Effects of novel identified CsgD-regulated genes on surface attachment and cell aggregation. Since the CsgD protein activates curli and cellulose production, which are factors that are involved in biofilm formation and in cell-cell interaction, we tested the possibility that the yoaD, cspA, and fecR genes could also play a role in these processes. Biofilm formation was measured by determining the ability to attach to a solid surface, while cell-cell interaction was determined by a cell aggregate sedimentation test, as described in Materials and Methods. In aggregation tests, sedimented pellets were fixed and stained with 1% crystal violet, which allowed semiquantitative measurement of cell aggregation. As shown in Fig. 5A, inactivation of either yoaD, cspA, or fecR led to a significant increase in cell aggregation, and the results for dissolution of crystal violet-stained pellets in ethanol ranged from a 5-fold increase for the cspA mutant strain to an almost 12-fold increase for the yoaD derivative of PHL565. Neither the growth rates nor the viabilities of the mutant strains differed significantly from the value for PHL565, suggesting that increased cell aggregation does not depend on cellular stress. Strong stimulation of cell aggregation following inactivation of yoaD would be consistent with the putative role of the YoaD protein as a negative regulator of cellulose biosynthesis (48, 57). Indeed, PHL565yoaD cells grown on calcofluor agar plates exhibited a 3.5-fold increase in absorbance at 366 nm, which indicated that there was increased cellulose production, while neither the cspA nor fecR mutations had similar effects (Fig. 5B). The positive effect of yoaD inactivation on cell aggregation does indeed depend on the YoaD protein, since introduction of a functional yoaD gene on the pGEMTyoaD plasmid in the PHL565yoaD strain totally abolished both cell sedimentation (Fig. 5A) and calcofluor binding (Fig. 5B).
Despite the stimulatory effect of yoaD or cspA on cell aggregation, inactivation of either yoaD or cspA did not result in a significant increase in surface attachment (Fig. 5C). In contrast, inactivation of the fecR gene positively affected cell adhesion; the effects of the fecR mutation were more pronounced when cells were grown in LB, in which this mutation led to a six- to eightfold increase (Fig. 5C).
The effect of the yoaD gene on cell aggregation was also tested in the PHL628 strain, an ompR234 mutant derivative of PHL565 able to express the csgDEFG operon (42, 53). The PHL628 strain formed cell aggregates that were clearly detectable in our sedimentation assays (Fig. 6). Transformation of PHL628 either with a control vector or with pGEMTyoaD, in the absence of isopropyl--D-thiogalactopyranoside (IPTG) induction, did not affect cell aggregation. However, upon full Plac induction by addition of 0.5 mM IPTG, which maximized yoaD expression from pGEMTyoaD, PHL628 cell aggregation was completely inhibited (Fig. 6). In contrast, surface adhesion to microtiter plates by PHL628 was not affected by the presence of the pGEMTyoaD plasmid, even in the presence of IPTG (data not shown).
Expression of the yoaD gene in PHL628. Whole-genome transcription analysis experiments suggested that the yoaD gene is activated in a CsgD-dependent fashion. To confirm this, we measured yoaD expression, using real-time PCR, in the CsgD-expressing PHL628 strain and compared it to the expression in PHL565. Samples were taken in different growth phases, and, as a control, we determined the expression of the CsgD-dependent adrA (yaiC) gene in the same conditions. The levels of yoaD transcripts were 10- to 12-fold higher in PHL628 than in PHL565 in stationary-phase cells, while the differences in expression levels were less than 2-fold in exponentially growing cells (Fig. 7). In contrast, growth phase-dependent expression could not be detected for adrA, whose PHL628/PHL565 expression ratio only increased from 15 to 22.
DISCUSSION
In this work, we tested the effects of low-level, constitutive expression of the CsgD protein, a positive regulator of curli and cellulose, on cell adhesion, transcription of curli genes, and global gene expression in the nonadherent PHL565 strain of E. coli. The CsgD protein was found to be associated with the cytoplasmic membrane after cell fractionation (Fig. 1). Although localization of CsgD in the cytoplasmic membrane may be unusual for a transcription factor, other membrane-associated proteins, such as the ToxR protein of Vibrio cholerae (14) and CadC in E. coli (13), can act as transcription regulators. Alternatively, some regulatory proteins, such as Mlc (30, 51), can be found in an active conformation in the cytoplasm and can be temporarily inactivated by sequestration to the inner membrane. In future experiments we will define the nature of the interaction between the cytoplasmic membrane and CsgD.
Constitutive expression of the csgD gene from the pT7-CsgD plasmid results in a fully active CsgD protein that is able to promote surface colonization by PHL565 in a curli-dependent fashion (Fig. 2) and to activate transcription at the csgBAC promoter (Fig. 3). By expressing the CsgD protein independent of its own promoter we could determine if important environmental signals in curli regulation, such as osmolarity, temperature, and growth medium, target either the csgDEFG promoter or subsequent steps in the curli regulation cascade. Constitutively expressed CsgD could still efficiently activate csgBAC transcription at high osmolarity (Fig. 4), in agreement with previous data showing that osmolarity control of curli production occurs mainly at the csgDEFG promoter via OmpR-dependent regulation (19, 42, 53). Unlike osmolarity-dependent regulation, temperature-dependent regulation of curli seems to take place at a step later than csgD transcription, as indicated by the observation that activation of csgBAC transcription by constitutively expressed CsgD is strongly inhibited at 37°C. This suggests that curli temperature regulation involves different mechanisms in E. coli and in Salmonella. In Salmonella, no csgD transcription takes place at 37°C; however, mutations in the csgDEFG promoter region can restore both csgD transcription and curli production at nonpermissive temperatures (44). In contrast, our observations suggest that temperature regulation comes into play at the csgBAC promoter (Fig. 3). This result is consistent with the hypothesis that the product of the crl gene (40) is the main temperature sensor for curli expression in E. coli (6). Indeed, crl mutations strongly impaired CsgD-dependent csgBAC transcription at 30°C, although our results suggest that additional temperature-dependent regulatory mechanisms could also affect csgBAC transcription (Fig. 3). Finally, growth medium-dependent curli regulation also takes place at a step later than csgD transcription (Fig. 3), suggesting that reduced csgBAC transcription in LB does not depend on the higher osmolarity of the LB but is due to yet another mechanism for environmental control of csgBAC transcription.
From the results of the whole-genome transcription assays we concluded that constitutive CsgD expression might affect the expression of about 30 genes in the conditions that we tested. The results of our experiments did not allow us to conclude that all genes that showed differential expression in PHL565/pT7-CsgD are indeed directly regulated by the CsgD protein. However, the altered expression in PHL565/pT7-CsgD of several genes involved in the modulation of intracellular nucleotide and nucleoside pools (gsk and pyrBI genes) and in membrane transport (gatYZABCDR operon) does support the notion that, in addition to the regulation of biofilm genes, CsgD might play a role in the regulation of metabolic processes. Indeed, mutations that inactivate the csgD gene have been shown to affect the nutritional requirements of environmental isolates of E. coli (52), and CsgD overexpression can overcome glycine auxotrophy of a folA mutant of E. coli MG1655 (10). The CsgD regulatory network is summarized in Fig. 8.
The main role of the CsgD protein, however, appears to be related to biofilm formation and cell-cell interaction and to adaptation to the biofilm way of life (Fig. 8). This notion is supported by the observation that the newly identified csgD-regulated genes yoaD (up-regulated by CsgD), fecR, and cspA (down-regulated) negatively affect cell aggregation and/or surface attachment (Fig. 5). Repression of fecR and cspA by CsgD would be consistent with the role of the CsgD protein as a positive determinant for biofilm formation and cell-cell interaction. The fecR gene encodes a periplasmic protein involved in iron sensing, in regulation of the FecI sigma factor, and in iron uptake. Iron availability is a major signal for biofilm formation in several gram-negative bacteria, such as Pseudomonas aeruginosa (1, 5). The presence of iron-sensing genes, such as fecR and fhuE, in the CsgD regulon might allow modulation of the intracellular iron concentration during the transition from planktonic cells to attached cells. Interestingly, we also observed a reduced amount of the Dps protein in the outer membrane protein fraction of PHL565/pT7-CsgD (Fig. 1) and a twofold reduction in dps transcription in the PHL565/pT7-CsgD strain (data not shown), suggesting that there is possible negative control of the dps gene by CsgD. Dps is a bacterial ferritin that is important for iron storage, particularly in slowly growing cells (34, 56). The main location of Dps is in the cytoplasm, but the Dps protein can also be found in the outer membrane fraction in several E. coli strains (Landini, unpublished data). We are currently investigating the mechanisms of control of surface attachment and Dps regulation by fecR.
Our data suggest that constitutive CsgD expression has a strong effect on the cold shock regulon (Table 3) and that cspA, which encodes the major cold adaptation protein in E. coli, may be involved in the negative regulation of cell aggregation (Fig. 5A). The role of cspA in cell aggregation and adhesion will be evaluated after a temperature downshift (i.e., in conditions in which the cold shock response protein is fully activated).
In contrast to fecR and cspA, the yoaD gene is activated by the CsgD protein (Table 3), although it also negatively affects cell aggregation (Fig. 5A). The yoaD gene belongs to a single-gene transcription unit and encodes a putative 59.5-kDa protein carrying the cyclic di-GMP phosphodiesterase EAL domain. Proteins belonging to the EAL family are involved in the degradation of cyclic di-GMP, a signal molecule that triggers several cellular processes, including cellulose biosynthesis (17, 48). Consistent with the putative role of the yoaD gene, inactivation of this gene stimulates cell aggregation (Fig. 5A) and results in increased cellulose biosynthesis (Fig. 5B), while overexpression of the gene negatively affects cell aggregation in a curli-producing strain of E. coli (Fig. 6). The YoaD protein might be expressed in a CsgD-dependent fashion in order to modulate cellulose biosynthesis by counteracting the positive effect of the adrA gene, which is also controlled by CsgD and encodes a GGDEF protein responsible for cyclic di-GMP biosynthesis and activation of cellulose production (48). YoaD-mediated modulation of cellulose biosynthesis may depend on the cell's need to prevent excessive consumption of glucose and/or GMP. Consistent with this hypothesis, the timing of yoaD expression is delayed with respect to adrA expression and is limited to the stationary phase of growth in the CsgD-expressing PHL628 strain (Fig. 7). Since yoaD, like adrA (45), plays a role in the regulation of cellulose production and is regulated in a CsgD-dependent fashion, we propose that yoaD should be reannotated as adrB.
ACKNOWLEDGMENTS
Financial support for this work was provided by the Swiss National Science Foundation (SNF grant 3100-058871).
We thank Valerie James and Jerome Carson for correcting the manuscript.
REFERENCES
Arevalo-Ferro, C., M. Hentzer, G. Reil, A. Gorg, S. Kjellenberg, M. Givskov, K. Riedel, and L. Eberl. 2003. Identification of quorum-sensing regulated proteins in the opportunistic pathogen Pseudomonas aeruginosa by proteomics. Environ. Microbiol. 5:1350-1369.
Arnqvist, A., A. Olsen, and S. Normark. 1994. s-Dependent growth-phase induction of the csgBA promoter in Escherichia coli can be achieved in vivo by 70 in the absence of the nucleoid-associated protein H-NS. Mol. Microbiol. 13:1021-1032.
Ben Nasr, A., A. Olsen, U. Sjobring, W. Muller-Esterl, and L. Bjorck. 1996. Assembly of human contact phase proteins and release of bradykinin at the surface of curli-expressing Escherichia coli. Mol. Microbiol. 20:927-935.
Bian, Z., A. Brauner, Y. Li, and S. Normark. 2000. Expression of and cytokine activation by Escherichia coli curli fibers in human sepsis. J. Infect. Dis. 181:602-612.
Bollinger, N., D. J. Hasset, B. H. Iglewski, J. W. Costerton, and T. R. McDermott. 2001. Gene expression in Pseudomonas aeruginosa: evidence of iron override effects on quorum sensing and biofilm-specific gene regulation. J. Bacteriol. 183:1990-1996.
Bougdour, A., C. Lelong, and J. Geiselmann. 2004. Crl, a low temperature induced protein in Escherichia coli that binds directly to the stationary phase sigma subunit of RNA polymerase. J. Biol. Chem. 279:19540-19550.
Brombacher, E., C. Dorel, A. J. Zehnder, and P. Landini. 2003. The curli biosynthesis regulator CsgD co-ordinates the expression of both positive and negative determinants for biofilm formation in Escherichia coli. Microbiology 149:2847-2857.
Brown, P. K., C. M. Dozois, C. A. Nickerson, A. Zuppardo, J. Terlonge, and R. Curtiss 3rd. 2001. MlrA, a novel regulator of curli (AgF) and extracellular matrix synthesis by Escherichia coli and Salmonella enterica serovar Typhimurium. Mol. Microbiol. 41:349-363.
Chen, X., L. M. Smith, and E. M. Bradbury. 2000. Site-specific mass tagging with stable isotopes in proteins for accurate and efficient protein identification Anal. Chem. 72:1134-1143.
Chirwa, N. T., and M. B. Herrington. 2003. CsgD, a regulator of curli and cellulose synthesis, also regulates serine hydroxymethyltransferase synthesis in Escherichia coli K-12. Microbiology 149:525-535.
Cookson, A. L., W. A. Cooley, and M. J. Woodward. 2002. The role of type 1 and curli fimbriae of Shiga toxin-producing Escherichia coli in adherence to abiotic surfaces. Int. J. Med. Microbiol. 292:195-205.
DeFlaun, M. F., B. M. Marshall, E.-P. Kulle, and S. B. Levy. 1994. Tn5 insertion mutants of Pseudomonas fluorescens defective in adhesion to soil and seeds. Appl. Environ. Microbiol. 60:2637-2642.
Dell, C. L., M. N. Neely, and E. R. Olson. 1994. Altered pH and lysine signalling mutants of cadC, a gene encoding a membrane-bound transcriptional activator of the Escherichia coli cadBA operon. Mol. Microbiol. 14:7-16.
DiRita, V. J., and J. J. Mekalanos. 1991. Periplasmic interaction between two membrane regulatory proteins, ToxR and ToxS, results in signal transduction and transcriptional activation. Cell 64:29-37.
Dorel, C., O. Vidal, C. Prigent-Combaret, I. Vallet, and P. Lejeune. 1999. Involvement of the Cpx signal transduction pathway of E. coli in biofilm formation. FEMS Microbiol. Lett. 178:169-175.
Galperin, M. Y., A. N. Nikolskaya, and E. V. Koonin. 2001. Novel domains of the prokaryotic two-component signal transduction systems. FEMS Microbiol. Lett. 203:11-21.
Garcia, B., C. Latasa, C. Solano, F. Garcia del Portillo, C. Gamazo, and I. Lasa. 2004. Role of the GGDEF protein family in Salmonella cellulose biosynthesis and biofilm formation. Mol. Microbiol. 54:264-277.
Gasteiger, E., A. Gattiker, C. Hoogland, I. Ivanyi, R. D. Appel, and A. Bairoch. 2003. ExPASy: the proteomics server for in-depth protein knowledge and analysis. Nucleic Acids Res. 31:3784-3788.
Gerstel, U., and U. Romling. 2001. Oxygen tension and nutrient starvation are major signals that regulate agfD promoter activity and expression of the multicellular morphotype in Salmonella typhimurium. Environ. Microbiol. 3:638-648.
Gerstel, U., C. Park, and U. Romling. 2003. Complex regulation of csgD promoter activity by global regulatory proteins. Mol. Microbiol. 49:639-654.
Gualerzi, C. O., A. M. Giuliodori, and C. L. Pon. 2003. Transcriptional and post-transcriptional control of cold-shock genes. J. Mol. Biol. 331:527-539.
Hammar, M., A. Arnqvist, Z. Bian, A. Olsen, and S. Normark. 1995. Expression of two csg operons is required for production of fibronectin- and Congo red-binding curli polymers in Escherichia coli K-12. Mol. Microbiol. 18:661-670.
Jenal, U. 2004. Cyclic di-guanosine-monophosphate comes of age: a novel secondary messenger involved in modulating cell surface structures in bacteria Curr. Opin. Microbiol. 7:185-191.
Jochimsen, B., P. Nygaard, and T. Vestergaard. 1975. Location on the chromosome of Escherichia coli of genes governing purine metabolism. Adenosine deaminase (add), guanosine kinase (gsk) and hypoxanthine phosphoribosyltransferase (hpt). Mol. Gen. Genet. 143:85-91.
Jubelin, G., A. Vianney, C. Beloin, J. M. Ghigo, J. C. Lazzaroni, P. Lejeune, and C. Dorel. 2005. CpxR/OmpR interplay regulates curli gene expression in response to osmolarity in Escherichia coli. J. Bacteriol. 187:2038-2049.
Karp, P. D., M. Riley, S. M. Paley, A. Pellegrini-Toole, and M. Krummenaker. 1999. Eco Cyc: encyclopedia of Escherichia coli genes and metabolism. Nucleic Acids Res. 27:55-58.
Kawasaki, H., M. Shiamaoka, Y. Usuda, and T. Utagawa. 2000. End-product regulation and kinetic mechanism of guanosine-inosine kinase from Escherichia coli. Biosci. Biotechnol. Biochem. 64:972-979.
Ke, H. M., R. B. Honzatko, and W. N. Lipscomb. 1984. Structure of unligated aspartate carbamoyltransferase of Escherichia coli at 2.6-A resolution. Proc. Natl. Acad. Sci. USA 81:4037-4040.
Lee, S. J., A. Xie, W. Jiang, J. P. Etchegaray, P. G. Jones, and M. Inouye. 1994. Family of the major cold-shock protein, CspA (CS7.4), of Escherichia coli, whose members show a high sequence similarity with the eukaryotic Y-box binding proteins. Mol. Microbiol. 11:833-839.
Lee, S. J., W. Boos, J. P. Bouche, and J. Plumbridge. 2000. Signal transduction between a membrane-bound transporter, PtsG, and a soluble transcription factor, Mlc, of Escherichia coli. EMBO J. 19:5353-5361.
Maurer, J. J., T. P. Brown, W. L. Steffens, and S. G. Thayer. 1998. The occurrence of ambient temperature-regulated adhesins, curli, and the temperature-sensitive hemagglutinin tsh among avian Escherichia coli. Avian Dis. 42:106-118.
Medigue, C., A. Viari, A. Henaut, and A. Danchin. 1993. Colibri: a functional database for the Escherichia coli genome. Microbiol. Rev. 57:623-654.
Miller, J. H. (ed.). 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
Nair, S., and S. E. Finkel. 2004. Dps protects cells against multiple stresses during stationary phase. J. Bacteriol. 186:4192-4198.
Nobelmann, B., and J. W. Lengeler. 1996. Molecular analysis of the gat genes from Escherichia coli and of their roles in galactitol transport and metabolism. J. Bacteriol. 178:6790-6795.
Olsen, A., A. Arnqvist, M. Hammar, and S. Normark. 1993. Environmental regulation of curli production in Escherichia coli. Infect. Agents Dis. 2:272-274.
Olsen, A., M. J. Wick, M. Morgelin, and L. Bjorck. 1998. Curli, fibrous surface proteins of Escherichia coli, interact with major histocompatibility complex class I molecules. Infect. Immun. 66:944-949.
Persson, K., W. Russell, M. Morgelin, and H. Herwald. 2003. The conversion of fibrinogen to fibrin at the surface of curliated Escherichia coli bacteria leads to the generation of proinflammatory fibrinopeptides. J. Biol. Chem. 278:31884-31890.
Postma, P. W., J. W. Lengeler, and G. R. Jacobson. 1993. Phosphoenolpyruvate:carbohydrate phosphotransferase systems of bacteria. Microbiol. Rev. 57:543-594.
Pratt, L. A., and T. J. Silhavy. 1998. Crl stimulates RpoS activity during stationary phase. Mol. Microbiol. 29:1225-1236.
Pressler, U., H. Staudenmeier, L. Zimmerman, and V. Braun. 1988. Genetics of the iron dicitrate transport system of Escherichia coli. J. Bacteriol. 170:2716-2724.
Prigent-Combaret, C., E. Brombacher, O. Vidal, A. Ambert, P. Lejeune, P. Landini, and C. Dorel. 2001. Complex regulatory network controls initial adhesion and biofilm formation in Escherichia coli via regulation of the csgD gene. J. Bacteriol. 183:7213-7223.
Romling, U., Z. Bian, M. Hammar, W. D. Sierralta, and S. Normark. 1998. Curli fibers are highly conserved between Salmonella typhimurium and Escherichia coli with respect to operon structure and regulation. J. Bacteriol. 180:722-731.
Romling, U., W. D. Sierralta, K. Eriksson, and S. Normark. 1998. Multicellular and aggregative behaviour of Salmonella typhimurium strains is controlled by mutations in the agfD promoter. Mol. Microbiol. 28:249-264.
Romling, U., M. Rohde, A. Olsen, S. Normark, and J. Reinkoster. 2000. AgfD, the checkpoint of multicellular and aggregative behaviour in Salmonella typhimurium, regulates at least two independent pathways. Mol. Microbiol. 36:10-23.
Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Sauer, M., K. Hantke, and V. Braun. 1987. Ferric-coprogen receptor FhuE of Escherichia coli: processing and sequence common to all TonB-dependent outer membrane receptor proteins. J. Bacteriol. 169:2044-2049.
Simm, R., M. Morr, A. Kader, M. Nimtz, and U. Romling. 2003. GGDEF and EAL domains inversely regulate cyclic di-GMP levels and transition from sessility to motility. Mol. Microbiol. 53:1123-1134.
Stiefel, A., S. Mahren, M. Ochs, P. T. Schindler, S. Enz, and V. Braun. 2001. Control of the ferric citrate transport system of Escherichia coli: mutations in region 2.1 of the FecI extracytoplasmic-function sigma factor suppress mutations in the FecR transmembrane regulatory protein. J. Bacteriol. 183:162-170.
Tal, R., H. C. Wong, R. Calhoon, D. Gelfand, A. L. Fear, G. Volman, R. Mayer, P. Ross, D. Amikam, H. Weinhouse, A. Cohen, S. Sapir, P. Ohana, and M. Benziman. 1998. Three cdg operons control cellular turnover of cyclic di-GMP in Acetobacter xylinum: genetic organization and occurrence of conserved domains in isoenzymes. J. Bacteriol. 180:4416-4425.
Tanaka, Y., K. Kimata, and H. Aiba. 2000. A novel regulatory role of glucose transporter of Escherichia coli: membrane sequestration of a global repressor Mlc. EMBO J. 19:5344-5352.
Uhlich, G. A., J. E. Keen, and R. O. Elder. 2001. Mutations in the csgD promoter associated with variations in curli expression in certain strains of Escherichia coli O157:H7. Appl. Environ. Microbiol. 67:2367-2370.
Vidal, O., R. Longin, C. Prigent-Combaret, C. Dorel, M. Hooreman, and P. Lejeune. 1998. Isolation of an Escherichia coli K-12 mutant strain able to form biofilms on inert surfaces: involvement of a new ompR allele that increases curli expression. J. Bacteriol. 180:2442-2449.
Yamanaka, K., and M. Inouye. 1998. Identification and characterization of five cspA homologous genes from Myxococcus xanthus. Biochim. Biophys. Acta 1447:347-365.
Zhang, Y., and E. R. Kantrowitz. 1991. The synergistic inhibition of Escherichia coli aspartate carbamoyltransferase by UTP in the presence of CTP is due to the binding of UTP to the low affinity CTP sites. J. Biol. Chem. 266:22154-22158.
Zhao, G., P. Ceci, A. Ilari, L. Giangiacomo, T. M. Laue, E. Chiancone, and N. D. Chasteen. 2002. Iron and hydrogen peroxide detoxification properties of DNA-binding protein from starved cells. A ferritin-like DNA-binding protein of Escherichia coli. J. Biol. Chem. 277:27689-27696.
Zogaj, X., M. Nimtz, M. Rohde, W. Bokranz, and U. Romling. 2001. The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as the second component of the extracellular matrix. Mol. Microbiol. 39:1452-1463.
Zogaj, X., W. Bokranz, M. Nimtz, and U. Romling. 2003. Production of cellulose and curli fimbriae by members of the family Enterobacteriaceae isolated from the human gastrointestinal tract. Infect. Immun. 71:4151-4158.(Eva Brombacher, Andrea Ba)
ABSTRACT
Curli fibers, encoded by the csgBAC genes, promote biofilm formation in Escherichia coli and other enterobacteria. Curli production is dependent on the CsgD transcription activator, which also promotes cellulose biosynthesis. In this study, we investigated the effects of CsgD expression from a weak constitutive promoter in the biofilm formation-deficient PHL565 strain of E. coli. We found that despite its function as a transcription activator, the CsgD protein is localized in the cytoplasmic membrane. Constitutive CsgD expression promotes biofilm formation by PHL565 and activates transcription from the csgBAC promoter; however, csgBAC expression remains dependent on temperature and the growth medium. Constitutive expression of the CsgD protein results in altered transcription patterns for at least 24 novel genes, in addition to the previously identified CsgD-dependent genes. The cspA and fecR genes, encoding regulatory proteins responding to cold shock and to iron, respectively, and yoaD, encoding a putative negative regulator of cellulose biosynthesis, were found to be some of the novel CsgD-regulated genes. Consistent with the predicted functional role, increased expression of the yoaD gene negatively affects cell aggregation, while yoaD inactivation results in stimulation of cell aggregation and leads to increased cellulose production. Inactivation of fecR results in significant increases in both cell aggregation and biofilm formation, while the effects of cspA are not as strong in the conditions tested. Our results indicate that CsgD can modulate cellulose biosynthesis through activation of the yoaD gene. In addition, the positive effect of CsgD on biofilm formation might be enhanced by repression of the fecR gene.
INTRODUCTION
Most bacteria are capable of surface colonization and biofilm formation through the production of specific adhesins and extracellular structures. Curli fibers (also known as thin aggregative fimbriae) are a major factor in adhesion to surfaces, cell aggregation, and biofilm formation in many enterobacteria (11, 36, 42, 43, 53). Expression of curli is linked to cellulose biosynthesis, which leads to the production of an extracellular matrix and results in tight cell-cell and cell-surface interactions and in the so-called rdar morphotype in Salmonella (45, 57, 58). Expression of both curli and cellulose depends on the CsgD protein, a putative transcription regulator of the LuxR family, which activates transcription of the csgBAC operon (2), which encodes curli structural subunits, and transcription of the adrA gene, a positive effector of cellulose biosynthesis (45). In addition to csgBAC activation by CsgD, production of curli is subject to complex regulation, which affects both the csgDEFG operon (encoding the CsgD transcription regulator and the CsgEFG curli-specific transport system) and the csgBAC operon (encoding curli structural subunits) (8, 20, 22, 53). Curli expression is dependent on different environmental and physiological cues, such as a low growth temperature (<32°C), low osmolarity, and slow growth or starvation (i.e., conditions usually encountered by the bacteria outside the mammalian host) (19, 22, 31, 36, 44). However, curli are an important virulence factor in some Salmonella and pathogenic Escherichia coli strains, in which temperature-dependent regulation can be bypassed and curli expression can also take place at 37°C (3, 4, 37, 38). In contrast, curli operons are cryptic in a large number of both clinical and environmental E. coli isolates, as well as in laboratory strains, despite the presence of functional csg genes. However, mutations either in the specific promoters (44, 52) or in global regulatory genes, such as hns (2) or ompR (53), can restore the expression of curli-encoding genes.
The product of the CsgD-dependent adrA gene is a member of the GGDEF protein family (16, 50). The AdrA protein can catalyze the synthesis of bis-(3',5')-cyclic diguanylic acid (cyclic di-GMP), which in turn stimulates the enzymes responsible for cellulose production (48). In addition to the genes coding for factors directly involved in the curli-cellulose extracellular matrix, the CsgD protein positively regulates glyA, which encodes the glycine biosynthethic enzyme serine hydroxymethyltransferase (10), and represses the dipeptidase-encoding pepD gene (7). The promoters of the csgBAC operon and of the adrA and pepD genes share a conserved 11-bp sequence (CGGGKGAKNKA), which is necessary for CsgD-dependent regulation (7).
The pepD gene was identified as a CsgD-dependent gene using a whole-genome expression approach in which the following two laboratory strains of E. coli were compared: PHL565, which is unable to produce curli, and a spontaneous curli-producing mutant, PHL628 (7). The PHL628 strain has a mutation in the ompR gene that results in a single leucine-to-arginine substitution (ompR234 allele); the ompR234 mutation increases transcription of ompR-dependent genes, including csgD (53), and stimulates specific DNA binding by the OmpR234 protein at the csgD promoter (25, 42). However, comparisons of laboratory strains unable to express curli with curli-producing mutants with mutations in global regulatory genes, such as ompR, do not allow precise evaluation of the direct contribution of the CsgD protein to gene regulation. To circumvent this problem, in this work we transformed the curli-deficient PHL565 strain either with a plasmid that allowed constitutive, low-level expression of the csgD gene or with a control vector, and we compared whole-genome expression in the two strains. We found that the CsgD protein controls the modulation of cellulose production through activation of yoaD, a putative cyclic di-GMP esterase-encoding gene. Constitutive CsgD expression affects the transcription of a number of genes and represses the fecR and cspA regulatory genes, which play a negative role in biofilm formation and cell-cell aggregation.
MATERIALS AND METHODS
Bacterial strains and plasmids. For this study, we used E. coli K-12 laboratory strain PHL565 (53) and derivatives of this strain (Table 1). Bacterial cells were grown either in Luria-Bertani broth (LB) or in M9Glu/sup (M9 minimal medium supplemented with 0.4% glucose and 2.5% LB). When necessary, ampicillin (100 μg/ml), kanamycin (50 μg/ml), or chloramphenicol (25 μg/ml) was added. For csgD expression, the csgD gene was cloned from the pCP900 plasmid (42) into the pT7-7 plasmid using the NdeI and PstI sites to obtain the pT7-CsgD plasmid, in which csgD was under the control of a phage T7 RNA polymerase-dependent promoter. For yoaD overexpression studies, the yoaD gene was amplified by PCR using PHL565 genomic DNA as the template and the yoaDFw and yoaDRev primers (Table 2). The PCR product was cloned into pGEM-T Easy, in which the yoaD gene was placed under the control of the Plac promoter, using the following primers: yoaDfwr (5'-ATGCAAAAAGCACAACGG-3') and yoaDrev (5'-GTTAACGTAACGGCATAATG-3'). MG1655 mutant strains carrying either yoaD, fecR, or cspA mutant alleles were obtained from the laboratory of F. Blattner, University of Wisconsin (http://www.genome.wisc.edu/functional/tnmutagenesis.htm). The mutant alleles were transferred into PHL565 by P1 transduction (33).
CsgD localization experiments. Cell fractionation was performed as described previously (12). Five hundred-milliliter cultures of PHL565/pT7-7 and PHL565/pT7-CsgD were grown in M9Glu/sup at 30°C for 15 h. The cells were harvested by centrifugation at 7,000 rpm for 10 min at 4°C, washed, and resuspended in 20 ml phosphate-buffered saline (PBS). Cells were disintegrated by sonication and centrifuged as described above to remove unbroken cells. The supernatant from the low-speed centrifugation was centrifuged at 100,000 x g for 1 h at 4°C to separate the cytoplasm (supernatant) and the membrane fraction (pellet). The pellet was washed with 2 ml of 2% Sarkosyl in PBS, left for 20 min at room temperature, and centrifuged at 40,000 x g at 10°C for 10 min to remove ribosomes and cytoplasmic proteins that were still associated with the membrane fraction. The pellet was resuspended in 1 ml of 1% Sarkosyl, incubated, and centrifuged as described above. The supernatant, corresponding to inner membrane proteins, was collected, and the pellet, corresponding to outer membrane proteins, was resuspended in 0.5 ml H2O. Protein concentrations were determined, and either 40 μg (for cytoplasmic fractions) or 20 μg (for membrane fractions) of total proteins was loaded onto a 12.5% sodium dodecyl sulfate (SDS)-polyacrylamide gel. Specific bands were identified by mass spectrometry of the peptide products after in-gel trypsin digestion (9).
Biofilm formation and cell aggregation assays. Biofilm formation in microtiter plates was determined essentially as described previously (15). Cells were grown in liquid cultures in microtiter plates (0.2 ml) for 18 to 20 h either in M9Glu/sup or in LB at 30°C; the liquid medium was removed, and the cell density was determined spectrophotometrically by determining the optical density at 600 nm (OD600). Cells attached to the microtiter plates were washed twice gently with PBS and stained for 20 min with 1% crystal violet (CV) in ethanol. The stained biofilms were washed with tap water and dried. For semiquantitative determination of biofilms, CV-stained cells were resuspended in 0.2 ml of 70% ethanol by vigorous pipetting. The A600 of each sample was determined and normalized to the OD600 of the corresponding liquid cultures. For cell aggregation assays, cultures were grown overnight in 3 ml of M9Glu/sup at 30°C in 15-ml Falcon tubes. Overnight cultures were left to stand for 24 h at room temperature to allow sedimentation of cell aggregates. In order to obtain better visualization of sedimented cells, the supernatant was removed by gentle pipetting, and the cell aggregates were fixed and stained with 1% crystal violet for 15 min. After destaining with H2O, the CV-fixed and -stained cell aggregates were dissolved in 70% ethanol for semiquantitative analysis by using a procedure similar to the procedure used for adhesion assays.
RNA isolation, cDNA labeling, and microarray data analysis. Total RNA from E. coli cells that were grown for 15 h in M9Glu/sup at 30°C to the stationary phase was isolated as described by Sambrook et al. (46). RNA samples were quantified spectrophotometrically, and the quality of the samples was checked by electrophoresis on agarose gels. For gene array experiments, fluorescently labeled cDNA from 50 μg of total RNA was produced using a CyScribe first-strand cDNA labeling kit (Amersham Biosciences) with either Cy3- or Cy5-dCTP. Labeled cDNA was pooled, purified with a Minielute PCR purification kit (QIAGEN), and then concentrated with a Microcon-30 (Millipore) prior to addition of the hybridization buffer. The resulting cDNA was hybridized to an E. coli K-12 V2 array (MWG) by following the manufacturer's instructions. Microarray slides were scanned using an Affimetrix 428 array scanner. Spots and corresponding background signals of the resulting 16-bit TIFF images were quantified using the Affimetrix Jaguar software, version 2. Subsequent data analysis was performed using the program GeneSpring 4.1 from Silicon Genetics. Induction factors (PHL565/pT7CsgD compared to PHL565/pT7-7) were calculated using the Cy3 and Cy5 signal intensities of each spot. Spots with signals whose values were less than 50 were excluded from the analysis. Normalization was performed using the 50th percentile distribution of the spots remaining after background correction, and genes with expression levels similar to the background level were excluded prior to normalization. Data from three independent gene array experiments were averaged. Only spots whose final average ratios were higher than or equal to 4 and that had a signal ratio greater than 2.5 in all three experiments were considered significant. Finally, single genes that showed significant induction but belonged to an operon in which no other gene was affected by CsgD expression were also discarded. The function and sequence homology of genes of interest were determined using the following databases: Colibri (http://genolist.pasteur.fr/Colibri/genome.cgi) (32), Swiss-Prot (http://www.expasy.org/) (18), and EcoCyc (http://www.ecocyc.org/) (26).
Real-time PCR analysis. Real-time PCR was performed with the same RNA that was used for the gene array experiments. Reverse transcription was carried out according to the manufacturer's instructions (Applied Biosystems) by using MultiScribe reverse transcriptase (62.5 U/μg) of total RNA in the presence of 1.25 μM random hexamers. Real-time PCRs were performed using the SYBR Green PCR master mixture, and the results were determined with an ABI Prism 7000 sequence detection system. In a 25-μl reaction mixture, the cDNA produced from 20 ng of total RNA was used. The primers used are listed in Table 2. All reactions were performed in triplicate, along with reactions with negative control samples using DNase I-digested RNA as templates in order to verify the lack of residual DNA. The relative amounts of the transcripts were determined by normalization to 16S rRNA. For comparison of yoaD expression in PHL628 and yoaD expression in PHL565, RNA was isolated from cells harvested in the exponential phase (OD600, 0.25), at the onset of the stationary phase (OD600, 0.8), and in the stationary phase (OD600, 2.5) in M9sup.
Other methods. -Glucuronidase assays were performed as described previously (33), except that hydrolysis of para-nitrophenyl--glucoronide was determined spectrophotometrically at A405. Congo red binding was performed on agar media plates as previously described (45). For determination of cellulose production cells were grown on calcofluor agar medium (45), gently scraped off the plates, and resuspended to an OD600 of 0.2, and the OD366 of the suspension was determined.
RESULTS
CsgD constitutive expression and cell localization. In the pT7-CsgD plasmid, the csgD gene is under the control of a T7 RNA polymerase-dependent promoter. However, in E. coli strains such as PHL565, which do not carry the T7 RNA polymerase-encoding gene, detectable csgD transcription can still take place and most likely depends on recognition by bacterial RNA polymerase of promoter-like sequences upstream of the csgD gene. According to real-time PCR experiments, in PHL565/pT7-CsgD the amount of csgD transcript was roughly 100-fold greater than the amount in the PHL565 strain carrying the control vector pT7-7 (Table 3), in which csgD expression was negligible. The level of the csgD transcript in PHL565/pT7-CsgD was 1.5- to 2-fold higher than the levels in other curli-proficient E. coli strains, such as the ompR234 mutant PHL628 (53) or WK2, a curli-producing environmental isolate, as determined by real-time PCR (Landini, unpublished data). However, unlike curli-producing strains in which csgD expression is driven from its own promoter, the levels of csgD transcription in PHL565/pT7-CsgD do not vary significantly in different growth conditions, again as determined by real-time PCR experiments (data not shown). Thus, we concluded that pT7-CsgD allows constitutive csgD expression totally uncoupled from physiological and environmental signals, such as growth phase and osmolarity, which, in contrast, control the expression of the csgDEFG promoter (2, 8, 42). No band corresponding to the CsgD protein was detectable in crude extracts of PHL565/pT7-CsgD as determined by SDS-polyacrylamide gel electrophoresis (PAGE) (data not shown). However, analysis of the different cell compartments (cytoplasm, inner membrane, and outer membrane) after fractionation of cells grown overnight at 30°C in M9Glu/sup led to identification of a 25-kDa band present only in the cytoplasmic membrane fraction of PHL565/pT7-CsgD, where it accounted for only a small percentage of the total proteins (Fig. 1). In-gel trypsin digestion of the protein followed by mass spectrometry analysis confirmed that this band indeed corresponded to CsgD. Unlike CsgD, proteins that form inclusion bodies, such as green fluorescent protein, are not readily solubilized by Sarkosyl treatment and are not found in the inner membrane after cell fractionation (data not shown). Thus, CsgD localization in the cytoplasmic membrane did not appear to be due to the formation of inclusion bodies or to other artifacts that depended on nonphysiological CsgD expression. Expression of CsgD from pT7-CsgD did not result in any major changes in protein expression at a level detectable by one-dimensional SDS-PAGE analysis. Interestingly, however, the Dps protein, found in the outer membrane protein compartment in PHL565 transformed with the pT7-7 vector, was present at a much lower level in PHL565/pT7-CsgD.
Effects of constitutive csgD expression on adhesion and curli expression. We tested how production of CsgD from the pT7-CsgD plasmid affects surface attachment in the PHL565 strain. As shown in Fig. 2A, CsgD expression from the pT7-CsgD plasmid resulted in a fourfold increase in surface attachment by the PHL565 strain in M9Glu/sup and in a roughly twofold increase in LB. A similar degree of surface attachment stimulation was observed previously for PHL628, an ompR234 mutant derivative of PHL565 (42). The ompR234 mutation results in increased transcription from the csgD promoter, thus allowing CsgD-directed curli and cellulose biosynthesis and biofilm formation (53). The PHL565/pT7-CsgD strain formed red colonies when it was plated on growth medium supplemented with the amyloid protein-binding dye Congo red (Fig. 2B), suggesting that curli production was induced in this strain. Increased Congo red binding and surface colonization induced by pT7-CsgD were indeed dependent on curli, since transformation with pT7-CsgD of the PHL856 strain, a PHL565 derivative in which the csgA gene encoding the main curli subunit has been inactivated, did not result in surface attachment (Fig. 2A) or in Congo red binding (Fig. 2B) by this strain. Thus, csgD expression from the pT7-CsgD plasmid led to the production of a functional CsgD protein and conferred an adherent, curli-expressing phenotype to the PHL565 strain.
Since CsgD expression from the pT7-CsgD plasmid was independent of the csgDEFG promoter, the observation that pT7-CsgD induced biofilm formation by PHL565 more efficiently in M9Glu/sup than in LB strongly suggested that medium-dependent curli regulation takes place at a step after csgD transcription. Thus, we tested the effects of different media, as well as temperature and osmolarity (the main environmental signals regulating curli production), on transcription from the csgBAC promoter in the presence of constitutively expressed CsgD protein. Transcription from the csgBAC promoter was determined in the PHL856 strain transformed with either pT7-7 or pT7-CsgD; in PHL856 the csgA gene is interrupted by a uidA gene encoding -glucuronidase. csgBAC transcription was significantly lower (up to 25-fold) in LB (Fig. 3B) than in M9Glu/sup (Fig. 3A), particularly during the exponential phase of growth. Only in the late stationary phase did the level of csgBAC transcription in LB increase to almost one-half the level of transcription in M9Glu/sup.
Growth at 37°C resulted in a reduction in csgBAC transcription in both LB and M9Glu/sup (Fig. 3), and this effect again appeared to be more significant in the exponential phase (up to 10-fold reduction observed) than in the late stationary phase (2.5- to 3-fold reduction). Thus, temperature-dependent regulation did not appear to take place at the level of csgD transcription. Indeed, the levels of transcription from the csgDEFG promoter appeared to be similar at 30°C and 37°C in PHL857, a CsgD-expressing mutant derivative of PHL565 (data not shown). Recently published observations suggest that in E. coli temperature-dependent regulation of curli is mediated by the product of the crl gene, which acts as the temperature sensor at the csgBAC promoter (6). To further investigate this possibility, we transformed the EB9 strain, a crl920::cam derivative (40) of PHL856, with either pT7-7 or pT7-CsgD, and we measured csgBAC transcription. The crl mutation resulted in a clear reduction in csgBAC transcription at 30°C, while it did not have any effect at 37°C, in agreement with the proposed role of the crl gene (Fig. 3).
In contrast to the substantial changes induced by media and temperature, the effects of growth medium osmolarity on csgBAC transcription in PHL565/pT7-CsgD were modest. Addition of up to 0.25 M NaCl to M9Glu/sup resulted in a less-than-twofold reduction in csgBAC transcription, which was, in contrast, totally abolished in the PHL857 strain, in which CsgD was expressed from its own promoter (Fig. 4).
Effects of constitutive CsgD expression on whole-genome transcription. The results of both adhesion and csgBAC transcription experiments showed that constitutive expression from pT7-CsgD led to production of an active CsgD protein. Thus, we performed a whole-genome transcription assay in which we compared PHL565 strains that were transformed with either pT7-7 or pT7-CsgD and were grown in M9Glu/sup at 30°C (i.e., the optimal conditions for curli expression). We considered an average difference in gene expression that was equal to or greater than fourfold significant (see Materials and Methods). Ten genes were found to be up-regulated and 14 genes were found to be down-regulated in the PHL565 strain transformed with pT7-CsgD (Table 3). Among the up-regulated genes we found, as expected, csgBAC and adrA, which are known to be CsgD regulated, and csgD itself. Increased csgD transcription was exclusively due to the presence of the pT7-CsgD expression vector and was independent of the csgDEFG promoter, as indicated by a lack of any increase in csgEFG transcripts (see the results for real-time PCR experiments). In addition to known CsgD-dependent genes, we found three genes with as-yet-unknown functions (yaiB, yjgW, and ytfI) and two genes (ymdA and yoaD) whose putative functions can be predicted based on the amino acid sequences of their products. The ymdA gene encodes a hypothetical protein similar to proteins in the fimA/papA fimbrial protein family and is located 120 bp downstream of the csgBAC operon. The yoaD gene encodes a member of the EAL protein family, which is thought to be responsible for degradation of cyclic di-GMP, a signal molecule able to trigger cellulose biosynthesis, biofilm formation, and different cellular processes in several gram-negative bacteria (23). Interestingly, the CsgD-activated adrA gene encodes a diguanylate cyclase, the biosynthetic enzyme for cyclic di-GMP (45, 48, 57), suggesting that both intracellular accumulation and degradation of cyclic di-GMP are mediated by CsgD-regulated genes. In addition to adrA and yoaD, a third GMP-related gene, gsk, was found to be more highly expressed in PHL565/pT7-CsgD. The Gsk protein is a GMP synthase belonging to the nucleoside salvage pathway (24, 27) and might be involved in either repletion or maintenance of the GMP cellular pool.
All 14 genes that were down-regulated in PHL565/pT7-CsgD have known functions, and none of them has yet been shown to be regulated by CsgD. The pyrBI operon encodes the two subunits of aspartate carbamoyl transferase, an enzyme that is part of the pyrimidine biosynthetic pathway (28, 55). The gatA, gatC, and gatZ genes listed in Table 3 belong to the gatYZABCDR operon, encoding a phosphoenolpyruvate-dependent phosphotransferase system transporter specific for the sugar galactitol (35, 39). Although the gatY, gatB, gatD, and gatR genes are not listed in Table 3, they were all down-regulated in PHL565/pT7-CsgD by factors ranging from 2.7- to 3.3-fold, suggesting that the whole gat transcription unit is indeed repressed in a CsgD-dependent fashion (data not shown). Two genes encoding outer membrane proteins, the main OmpF porin and the OmpT protease, as well as the methionine biosynthesis metA gene, were also down-regulated in PHL565/pT7-CsgD (Table 3).
The five remaining genes which are down-regulated by constitutively expressed CsgD belong to the following two functional groups: iron-sensing genes and cold shock-responding genes. Our assays showed that there was 4.5-fold repression of transcription of both the fecR and fhuE genes in PHL565/pT7-CsgD (Table 3). The outer membrane FhuE protein serves as a receptor for ferric coprogen and ferric rhodotorulic acid, which upon binding by FhuE can be taken up via the TonB system (47). The periplasmic FecR protein plays a role in iron sensing and in regulation of the alternative sigma factor FecI. The two genes are cotranscribed in the fecIR operon (41, 49), and the level of the fecI transcript is 2.5-fold lower in PHL565/pT7-CsgD (data not shown), which is consistent with the possibility that there is either direct or indirect transcriptional repression of the whole fecIR operon by CsgD. Finally, three of the main cold shock-induced genes in E. coli (cspA, cspB, and cspG) (29, 54) appeared to be down-regulated by factors of four- to sevenfold in PHL565/pT7-CsgD. In addition, the level of the transcript of the infA gene, which is known to respond to cold shock (21), was also fourfold lower in the CsgD-expressing strain.
Real-time PCR analysis of genes differentially expressed in PHL565/pT7-CsgD. A selection of genes that were found to be differentially expressed in the whole-genome transcription assay were tested in real-time PCR experiments. First, we tested genes belonging to the csgBAC and csgDEFG operons and genes proposed to be CsgD dependent (adrA, glyA, and pepD). Real-time PCR experiments showed that the levels of expression of the csg genes were 10- to 100-fold-higher than the levels in the whole-genome transcription experiment (Table 3). The large differences were probably due to the difficulty of evaluating precisely the very low levels of expression of csg genes in the PHL565 strain. In contrast, real-time PCR and whole-genome transcription assays yielded very similar results for all other genes tested (Table 3).
Unlike the levels of transcription of the csgBAC operon, the levels of transcription of the csgE, csgF and csgG genes (i.e., the csgDEFG operon) were not altered by the presence of the pT7-CsgD plasmid, strongly suggesting that the CsgD protein does not regulate its own gene and indicating that, despite the presence of a functional chromosomal copy of csgD in the PHL565/pT7-CsgD strain, csgD expression is solely dependent on the plasmid copy of the gene.
Transcription of the glyA gene was only weakly stimulated (1.3-fold) by the CsgD protein according to the results of our real-time experiments (Table 3). Although CsgD has been shown to suppress glycine autotrophy and to stimulate serine hydroxymethyltransferase activity via the glyA gene (10), little CsgD-dependent stimulation (1.2- to 2-fold) of glyA transcription was detectable when cells were grown in minimal medium supplemented with amino acids (10), which is consistent with our results. In contrast, the pepD gene appeared to be 4-fold negatively regulated by CsgD in real-time PCR experiments, similar to the 4-fold repression observed for the curli-expressing PHL628 derivative of PHL565 (7) and to the 2.5-fold repression observed for the PHL565/pT7-CsgD gene array compared with the PHL565/pT7-7 gene array (Table 3).
Among the novel identified genes whose expression was affected by CsgD, we tested the EAL protein-encoding yoaD gene, as well as the fecR and cspA genes, in real-time PCR experiments (Table 3). For these genes, the real-time PCR results closely reflected the results of the whole-genome transcription assays, showing that there was an eightfold increase for the yoaD transcript and a roughly fourfold decrease for both the cspA and fecR transcripts in PHL565/pT7-CsgD.
Effects of novel identified CsgD-regulated genes on surface attachment and cell aggregation. Since the CsgD protein activates curli and cellulose production, which are factors that are involved in biofilm formation and in cell-cell interaction, we tested the possibility that the yoaD, cspA, and fecR genes could also play a role in these processes. Biofilm formation was measured by determining the ability to attach to a solid surface, while cell-cell interaction was determined by a cell aggregate sedimentation test, as described in Materials and Methods. In aggregation tests, sedimented pellets were fixed and stained with 1% crystal violet, which allowed semiquantitative measurement of cell aggregation. As shown in Fig. 5A, inactivation of either yoaD, cspA, or fecR led to a significant increase in cell aggregation, and the results for dissolution of crystal violet-stained pellets in ethanol ranged from a 5-fold increase for the cspA mutant strain to an almost 12-fold increase for the yoaD derivative of PHL565. Neither the growth rates nor the viabilities of the mutant strains differed significantly from the value for PHL565, suggesting that increased cell aggregation does not depend on cellular stress. Strong stimulation of cell aggregation following inactivation of yoaD would be consistent with the putative role of the YoaD protein as a negative regulator of cellulose biosynthesis (48, 57). Indeed, PHL565yoaD cells grown on calcofluor agar plates exhibited a 3.5-fold increase in absorbance at 366 nm, which indicated that there was increased cellulose production, while neither the cspA nor fecR mutations had similar effects (Fig. 5B). The positive effect of yoaD inactivation on cell aggregation does indeed depend on the YoaD protein, since introduction of a functional yoaD gene on the pGEMTyoaD plasmid in the PHL565yoaD strain totally abolished both cell sedimentation (Fig. 5A) and calcofluor binding (Fig. 5B).
Despite the stimulatory effect of yoaD or cspA on cell aggregation, inactivation of either yoaD or cspA did not result in a significant increase in surface attachment (Fig. 5C). In contrast, inactivation of the fecR gene positively affected cell adhesion; the effects of the fecR mutation were more pronounced when cells were grown in LB, in which this mutation led to a six- to eightfold increase (Fig. 5C).
The effect of the yoaD gene on cell aggregation was also tested in the PHL628 strain, an ompR234 mutant derivative of PHL565 able to express the csgDEFG operon (42, 53). The PHL628 strain formed cell aggregates that were clearly detectable in our sedimentation assays (Fig. 6). Transformation of PHL628 either with a control vector or with pGEMTyoaD, in the absence of isopropyl--D-thiogalactopyranoside (IPTG) induction, did not affect cell aggregation. However, upon full Plac induction by addition of 0.5 mM IPTG, which maximized yoaD expression from pGEMTyoaD, PHL628 cell aggregation was completely inhibited (Fig. 6). In contrast, surface adhesion to microtiter plates by PHL628 was not affected by the presence of the pGEMTyoaD plasmid, even in the presence of IPTG (data not shown).
Expression of the yoaD gene in PHL628. Whole-genome transcription analysis experiments suggested that the yoaD gene is activated in a CsgD-dependent fashion. To confirm this, we measured yoaD expression, using real-time PCR, in the CsgD-expressing PHL628 strain and compared it to the expression in PHL565. Samples were taken in different growth phases, and, as a control, we determined the expression of the CsgD-dependent adrA (yaiC) gene in the same conditions. The levels of yoaD transcripts were 10- to 12-fold higher in PHL628 than in PHL565 in stationary-phase cells, while the differences in expression levels were less than 2-fold in exponentially growing cells (Fig. 7). In contrast, growth phase-dependent expression could not be detected for adrA, whose PHL628/PHL565 expression ratio only increased from 15 to 22.
DISCUSSION
In this work, we tested the effects of low-level, constitutive expression of the CsgD protein, a positive regulator of curli and cellulose, on cell adhesion, transcription of curli genes, and global gene expression in the nonadherent PHL565 strain of E. coli. The CsgD protein was found to be associated with the cytoplasmic membrane after cell fractionation (Fig. 1). Although localization of CsgD in the cytoplasmic membrane may be unusual for a transcription factor, other membrane-associated proteins, such as the ToxR protein of Vibrio cholerae (14) and CadC in E. coli (13), can act as transcription regulators. Alternatively, some regulatory proteins, such as Mlc (30, 51), can be found in an active conformation in the cytoplasm and can be temporarily inactivated by sequestration to the inner membrane. In future experiments we will define the nature of the interaction between the cytoplasmic membrane and CsgD.
Constitutive expression of the csgD gene from the pT7-CsgD plasmid results in a fully active CsgD protein that is able to promote surface colonization by PHL565 in a curli-dependent fashion (Fig. 2) and to activate transcription at the csgBAC promoter (Fig. 3). By expressing the CsgD protein independent of its own promoter we could determine if important environmental signals in curli regulation, such as osmolarity, temperature, and growth medium, target either the csgDEFG promoter or subsequent steps in the curli regulation cascade. Constitutively expressed CsgD could still efficiently activate csgBAC transcription at high osmolarity (Fig. 4), in agreement with previous data showing that osmolarity control of curli production occurs mainly at the csgDEFG promoter via OmpR-dependent regulation (19, 42, 53). Unlike osmolarity-dependent regulation, temperature-dependent regulation of curli seems to take place at a step later than csgD transcription, as indicated by the observation that activation of csgBAC transcription by constitutively expressed CsgD is strongly inhibited at 37°C. This suggests that curli temperature regulation involves different mechanisms in E. coli and in Salmonella. In Salmonella, no csgD transcription takes place at 37°C; however, mutations in the csgDEFG promoter region can restore both csgD transcription and curli production at nonpermissive temperatures (44). In contrast, our observations suggest that temperature regulation comes into play at the csgBAC promoter (Fig. 3). This result is consistent with the hypothesis that the product of the crl gene (40) is the main temperature sensor for curli expression in E. coli (6). Indeed, crl mutations strongly impaired CsgD-dependent csgBAC transcription at 30°C, although our results suggest that additional temperature-dependent regulatory mechanisms could also affect csgBAC transcription (Fig. 3). Finally, growth medium-dependent curli regulation also takes place at a step later than csgD transcription (Fig. 3), suggesting that reduced csgBAC transcription in LB does not depend on the higher osmolarity of the LB but is due to yet another mechanism for environmental control of csgBAC transcription.
From the results of the whole-genome transcription assays we concluded that constitutive CsgD expression might affect the expression of about 30 genes in the conditions that we tested. The results of our experiments did not allow us to conclude that all genes that showed differential expression in PHL565/pT7-CsgD are indeed directly regulated by the CsgD protein. However, the altered expression in PHL565/pT7-CsgD of several genes involved in the modulation of intracellular nucleotide and nucleoside pools (gsk and pyrBI genes) and in membrane transport (gatYZABCDR operon) does support the notion that, in addition to the regulation of biofilm genes, CsgD might play a role in the regulation of metabolic processes. Indeed, mutations that inactivate the csgD gene have been shown to affect the nutritional requirements of environmental isolates of E. coli (52), and CsgD overexpression can overcome glycine auxotrophy of a folA mutant of E. coli MG1655 (10). The CsgD regulatory network is summarized in Fig. 8.
The main role of the CsgD protein, however, appears to be related to biofilm formation and cell-cell interaction and to adaptation to the biofilm way of life (Fig. 8). This notion is supported by the observation that the newly identified csgD-regulated genes yoaD (up-regulated by CsgD), fecR, and cspA (down-regulated) negatively affect cell aggregation and/or surface attachment (Fig. 5). Repression of fecR and cspA by CsgD would be consistent with the role of the CsgD protein as a positive determinant for biofilm formation and cell-cell interaction. The fecR gene encodes a periplasmic protein involved in iron sensing, in regulation of the FecI sigma factor, and in iron uptake. Iron availability is a major signal for biofilm formation in several gram-negative bacteria, such as Pseudomonas aeruginosa (1, 5). The presence of iron-sensing genes, such as fecR and fhuE, in the CsgD regulon might allow modulation of the intracellular iron concentration during the transition from planktonic cells to attached cells. Interestingly, we also observed a reduced amount of the Dps protein in the outer membrane protein fraction of PHL565/pT7-CsgD (Fig. 1) and a twofold reduction in dps transcription in the PHL565/pT7-CsgD strain (data not shown), suggesting that there is possible negative control of the dps gene by CsgD. Dps is a bacterial ferritin that is important for iron storage, particularly in slowly growing cells (34, 56). The main location of Dps is in the cytoplasm, but the Dps protein can also be found in the outer membrane fraction in several E. coli strains (Landini, unpublished data). We are currently investigating the mechanisms of control of surface attachment and Dps regulation by fecR.
Our data suggest that constitutive CsgD expression has a strong effect on the cold shock regulon (Table 3) and that cspA, which encodes the major cold adaptation protein in E. coli, may be involved in the negative regulation of cell aggregation (Fig. 5A). The role of cspA in cell aggregation and adhesion will be evaluated after a temperature downshift (i.e., in conditions in which the cold shock response protein is fully activated).
In contrast to fecR and cspA, the yoaD gene is activated by the CsgD protein (Table 3), although it also negatively affects cell aggregation (Fig. 5A). The yoaD gene belongs to a single-gene transcription unit and encodes a putative 59.5-kDa protein carrying the cyclic di-GMP phosphodiesterase EAL domain. Proteins belonging to the EAL family are involved in the degradation of cyclic di-GMP, a signal molecule that triggers several cellular processes, including cellulose biosynthesis (17, 48). Consistent with the putative role of the yoaD gene, inactivation of this gene stimulates cell aggregation (Fig. 5A) and results in increased cellulose biosynthesis (Fig. 5B), while overexpression of the gene negatively affects cell aggregation in a curli-producing strain of E. coli (Fig. 6). The YoaD protein might be expressed in a CsgD-dependent fashion in order to modulate cellulose biosynthesis by counteracting the positive effect of the adrA gene, which is also controlled by CsgD and encodes a GGDEF protein responsible for cyclic di-GMP biosynthesis and activation of cellulose production (48). YoaD-mediated modulation of cellulose biosynthesis may depend on the cell's need to prevent excessive consumption of glucose and/or GMP. Consistent with this hypothesis, the timing of yoaD expression is delayed with respect to adrA expression and is limited to the stationary phase of growth in the CsgD-expressing PHL628 strain (Fig. 7). Since yoaD, like adrA (45), plays a role in the regulation of cellulose production and is regulated in a CsgD-dependent fashion, we propose that yoaD should be reannotated as adrB.
ACKNOWLEDGMENTS
Financial support for this work was provided by the Swiss National Science Foundation (SNF grant 3100-058871).
We thank Valerie James and Jerome Carson for correcting the manuscript.
REFERENCES
Arevalo-Ferro, C., M. Hentzer, G. Reil, A. Gorg, S. Kjellenberg, M. Givskov, K. Riedel, and L. Eberl. 2003. Identification of quorum-sensing regulated proteins in the opportunistic pathogen Pseudomonas aeruginosa by proteomics. Environ. Microbiol. 5:1350-1369.
Arnqvist, A., A. Olsen, and S. Normark. 1994. s-Dependent growth-phase induction of the csgBA promoter in Escherichia coli can be achieved in vivo by 70 in the absence of the nucleoid-associated protein H-NS. Mol. Microbiol. 13:1021-1032.
Ben Nasr, A., A. Olsen, U. Sjobring, W. Muller-Esterl, and L. Bjorck. 1996. Assembly of human contact phase proteins and release of bradykinin at the surface of curli-expressing Escherichia coli. Mol. Microbiol. 20:927-935.
Bian, Z., A. Brauner, Y. Li, and S. Normark. 2000. Expression of and cytokine activation by Escherichia coli curli fibers in human sepsis. J. Infect. Dis. 181:602-612.
Bollinger, N., D. J. Hasset, B. H. Iglewski, J. W. Costerton, and T. R. McDermott. 2001. Gene expression in Pseudomonas aeruginosa: evidence of iron override effects on quorum sensing and biofilm-specific gene regulation. J. Bacteriol. 183:1990-1996.
Bougdour, A., C. Lelong, and J. Geiselmann. 2004. Crl, a low temperature induced protein in Escherichia coli that binds directly to the stationary phase sigma subunit of RNA polymerase. J. Biol. Chem. 279:19540-19550.
Brombacher, E., C. Dorel, A. J. Zehnder, and P. Landini. 2003. The curli biosynthesis regulator CsgD co-ordinates the expression of both positive and negative determinants for biofilm formation in Escherichia coli. Microbiology 149:2847-2857.
Brown, P. K., C. M. Dozois, C. A. Nickerson, A. Zuppardo, J. Terlonge, and R. Curtiss 3rd. 2001. MlrA, a novel regulator of curli (AgF) and extracellular matrix synthesis by Escherichia coli and Salmonella enterica serovar Typhimurium. Mol. Microbiol. 41:349-363.
Chen, X., L. M. Smith, and E. M. Bradbury. 2000. Site-specific mass tagging with stable isotopes in proteins for accurate and efficient protein identification Anal. Chem. 72:1134-1143.
Chirwa, N. T., and M. B. Herrington. 2003. CsgD, a regulator of curli and cellulose synthesis, also regulates serine hydroxymethyltransferase synthesis in Escherichia coli K-12. Microbiology 149:525-535.
Cookson, A. L., W. A. Cooley, and M. J. Woodward. 2002. The role of type 1 and curli fimbriae of Shiga toxin-producing Escherichia coli in adherence to abiotic surfaces. Int. J. Med. Microbiol. 292:195-205.
DeFlaun, M. F., B. M. Marshall, E.-P. Kulle, and S. B. Levy. 1994. Tn5 insertion mutants of Pseudomonas fluorescens defective in adhesion to soil and seeds. Appl. Environ. Microbiol. 60:2637-2642.
Dell, C. L., M. N. Neely, and E. R. Olson. 1994. Altered pH and lysine signalling mutants of cadC, a gene encoding a membrane-bound transcriptional activator of the Escherichia coli cadBA operon. Mol. Microbiol. 14:7-16.
DiRita, V. J., and J. J. Mekalanos. 1991. Periplasmic interaction between two membrane regulatory proteins, ToxR and ToxS, results in signal transduction and transcriptional activation. Cell 64:29-37.
Dorel, C., O. Vidal, C. Prigent-Combaret, I. Vallet, and P. Lejeune. 1999. Involvement of the Cpx signal transduction pathway of E. coli in biofilm formation. FEMS Microbiol. Lett. 178:169-175.
Galperin, M. Y., A. N. Nikolskaya, and E. V. Koonin. 2001. Novel domains of the prokaryotic two-component signal transduction systems. FEMS Microbiol. Lett. 203:11-21.
Garcia, B., C. Latasa, C. Solano, F. Garcia del Portillo, C. Gamazo, and I. Lasa. 2004. Role of the GGDEF protein family in Salmonella cellulose biosynthesis and biofilm formation. Mol. Microbiol. 54:264-277.
Gasteiger, E., A. Gattiker, C. Hoogland, I. Ivanyi, R. D. Appel, and A. Bairoch. 2003. ExPASy: the proteomics server for in-depth protein knowledge and analysis. Nucleic Acids Res. 31:3784-3788.
Gerstel, U., and U. Romling. 2001. Oxygen tension and nutrient starvation are major signals that regulate agfD promoter activity and expression of the multicellular morphotype in Salmonella typhimurium. Environ. Microbiol. 3:638-648.
Gerstel, U., C. Park, and U. Romling. 2003. Complex regulation of csgD promoter activity by global regulatory proteins. Mol. Microbiol. 49:639-654.
Gualerzi, C. O., A. M. Giuliodori, and C. L. Pon. 2003. Transcriptional and post-transcriptional control of cold-shock genes. J. Mol. Biol. 331:527-539.
Hammar, M., A. Arnqvist, Z. Bian, A. Olsen, and S. Normark. 1995. Expression of two csg operons is required for production of fibronectin- and Congo red-binding curli polymers in Escherichia coli K-12. Mol. Microbiol. 18:661-670.
Jenal, U. 2004. Cyclic di-guanosine-monophosphate comes of age: a novel secondary messenger involved in modulating cell surface structures in bacteria Curr. Opin. Microbiol. 7:185-191.
Jochimsen, B., P. Nygaard, and T. Vestergaard. 1975. Location on the chromosome of Escherichia coli of genes governing purine metabolism. Adenosine deaminase (add), guanosine kinase (gsk) and hypoxanthine phosphoribosyltransferase (hpt). Mol. Gen. Genet. 143:85-91.
Jubelin, G., A. Vianney, C. Beloin, J. M. Ghigo, J. C. Lazzaroni, P. Lejeune, and C. Dorel. 2005. CpxR/OmpR interplay regulates curli gene expression in response to osmolarity in Escherichia coli. J. Bacteriol. 187:2038-2049.
Karp, P. D., M. Riley, S. M. Paley, A. Pellegrini-Toole, and M. Krummenaker. 1999. Eco Cyc: encyclopedia of Escherichia coli genes and metabolism. Nucleic Acids Res. 27:55-58.
Kawasaki, H., M. Shiamaoka, Y. Usuda, and T. Utagawa. 2000. End-product regulation and kinetic mechanism of guanosine-inosine kinase from Escherichia coli. Biosci. Biotechnol. Biochem. 64:972-979.
Ke, H. M., R. B. Honzatko, and W. N. Lipscomb. 1984. Structure of unligated aspartate carbamoyltransferase of Escherichia coli at 2.6-A resolution. Proc. Natl. Acad. Sci. USA 81:4037-4040.
Lee, S. J., A. Xie, W. Jiang, J. P. Etchegaray, P. G. Jones, and M. Inouye. 1994. Family of the major cold-shock protein, CspA (CS7.4), of Escherichia coli, whose members show a high sequence similarity with the eukaryotic Y-box binding proteins. Mol. Microbiol. 11:833-839.
Lee, S. J., W. Boos, J. P. Bouche, and J. Plumbridge. 2000. Signal transduction between a membrane-bound transporter, PtsG, and a soluble transcription factor, Mlc, of Escherichia coli. EMBO J. 19:5353-5361.
Maurer, J. J., T. P. Brown, W. L. Steffens, and S. G. Thayer. 1998. The occurrence of ambient temperature-regulated adhesins, curli, and the temperature-sensitive hemagglutinin tsh among avian Escherichia coli. Avian Dis. 42:106-118.
Medigue, C., A. Viari, A. Henaut, and A. Danchin. 1993. Colibri: a functional database for the Escherichia coli genome. Microbiol. Rev. 57:623-654.
Miller, J. H. (ed.). 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
Nair, S., and S. E. Finkel. 2004. Dps protects cells against multiple stresses during stationary phase. J. Bacteriol. 186:4192-4198.
Nobelmann, B., and J. W. Lengeler. 1996. Molecular analysis of the gat genes from Escherichia coli and of their roles in galactitol transport and metabolism. J. Bacteriol. 178:6790-6795.
Olsen, A., A. Arnqvist, M. Hammar, and S. Normark. 1993. Environmental regulation of curli production in Escherichia coli. Infect. Agents Dis. 2:272-274.
Olsen, A., M. J. Wick, M. Morgelin, and L. Bjorck. 1998. Curli, fibrous surface proteins of Escherichia coli, interact with major histocompatibility complex class I molecules. Infect. Immun. 66:944-949.
Persson, K., W. Russell, M. Morgelin, and H. Herwald. 2003. The conversion of fibrinogen to fibrin at the surface of curliated Escherichia coli bacteria leads to the generation of proinflammatory fibrinopeptides. J. Biol. Chem. 278:31884-31890.
Postma, P. W., J. W. Lengeler, and G. R. Jacobson. 1993. Phosphoenolpyruvate:carbohydrate phosphotransferase systems of bacteria. Microbiol. Rev. 57:543-594.
Pratt, L. A., and T. J. Silhavy. 1998. Crl stimulates RpoS activity during stationary phase. Mol. Microbiol. 29:1225-1236.
Pressler, U., H. Staudenmeier, L. Zimmerman, and V. Braun. 1988. Genetics of the iron dicitrate transport system of Escherichia coli. J. Bacteriol. 170:2716-2724.
Prigent-Combaret, C., E. Brombacher, O. Vidal, A. Ambert, P. Lejeune, P. Landini, and C. Dorel. 2001. Complex regulatory network controls initial adhesion and biofilm formation in Escherichia coli via regulation of the csgD gene. J. Bacteriol. 183:7213-7223.
Romling, U., Z. Bian, M. Hammar, W. D. Sierralta, and S. Normark. 1998. Curli fibers are highly conserved between Salmonella typhimurium and Escherichia coli with respect to operon structure and regulation. J. Bacteriol. 180:722-731.
Romling, U., W. D. Sierralta, K. Eriksson, and S. Normark. 1998. Multicellular and aggregative behaviour of Salmonella typhimurium strains is controlled by mutations in the agfD promoter. Mol. Microbiol. 28:249-264.
Romling, U., M. Rohde, A. Olsen, S. Normark, and J. Reinkoster. 2000. AgfD, the checkpoint of multicellular and aggregative behaviour in Salmonella typhimurium, regulates at least two independent pathways. Mol. Microbiol. 36:10-23.
Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Sauer, M., K. Hantke, and V. Braun. 1987. Ferric-coprogen receptor FhuE of Escherichia coli: processing and sequence common to all TonB-dependent outer membrane receptor proteins. J. Bacteriol. 169:2044-2049.
Simm, R., M. Morr, A. Kader, M. Nimtz, and U. Romling. 2003. GGDEF and EAL domains inversely regulate cyclic di-GMP levels and transition from sessility to motility. Mol. Microbiol. 53:1123-1134.
Stiefel, A., S. Mahren, M. Ochs, P. T. Schindler, S. Enz, and V. Braun. 2001. Control of the ferric citrate transport system of Escherichia coli: mutations in region 2.1 of the FecI extracytoplasmic-function sigma factor suppress mutations in the FecR transmembrane regulatory protein. J. Bacteriol. 183:162-170.
Tal, R., H. C. Wong, R. Calhoon, D. Gelfand, A. L. Fear, G. Volman, R. Mayer, P. Ross, D. Amikam, H. Weinhouse, A. Cohen, S. Sapir, P. Ohana, and M. Benziman. 1998. Three cdg operons control cellular turnover of cyclic di-GMP in Acetobacter xylinum: genetic organization and occurrence of conserved domains in isoenzymes. J. Bacteriol. 180:4416-4425.
Tanaka, Y., K. Kimata, and H. Aiba. 2000. A novel regulatory role of glucose transporter of Escherichia coli: membrane sequestration of a global repressor Mlc. EMBO J. 19:5344-5352.
Uhlich, G. A., J. E. Keen, and R. O. Elder. 2001. Mutations in the csgD promoter associated with variations in curli expression in certain strains of Escherichia coli O157:H7. Appl. Environ. Microbiol. 67:2367-2370.
Vidal, O., R. Longin, C. Prigent-Combaret, C. Dorel, M. Hooreman, and P. Lejeune. 1998. Isolation of an Escherichia coli K-12 mutant strain able to form biofilms on inert surfaces: involvement of a new ompR allele that increases curli expression. J. Bacteriol. 180:2442-2449.
Yamanaka, K., and M. Inouye. 1998. Identification and characterization of five cspA homologous genes from Myxococcus xanthus. Biochim. Biophys. Acta 1447:347-365.
Zhang, Y., and E. R. Kantrowitz. 1991. The synergistic inhibition of Escherichia coli aspartate carbamoyltransferase by UTP in the presence of CTP is due to the binding of UTP to the low affinity CTP sites. J. Biol. Chem. 266:22154-22158.
Zhao, G., P. Ceci, A. Ilari, L. Giangiacomo, T. M. Laue, E. Chiancone, and N. D. Chasteen. 2002. Iron and hydrogen peroxide detoxification properties of DNA-binding protein from starved cells. A ferritin-like DNA-binding protein of Escherichia coli. J. Biol. Chem. 277:27689-27696.
Zogaj, X., M. Nimtz, M. Rohde, W. Bokranz, and U. Romling. 2001. The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as the second component of the extracellular matrix. Mol. Microbiol. 39:1452-1463.
Zogaj, X., W. Bokranz, M. Nimtz, and U. Romling. 2003. Production of cellulose and curli fimbriae by members of the family Enterobacteriaceae isolated from the human gastrointestinal tract. Infect. Immun. 71:4151-4158.(Eva Brombacher, Andrea Ba)