Previous Article | Next Article ![]()
Infection and Immunity, July 2003, p. 4151-4158, Vol. 71, No. 7
0019-9567/03/$08.00+0 DOI: 10.1128/IAI.71.7.4151-4158.2003
Copyright © 2003, American Society for Microbiology. All Rights Reserved.
Microbiology and Tumorbiology Center, Karolinska Institutet, SE-171 77 Stockholm, Sweden,1 Divisions of Cell Biology and Immunology,2 Structural Biology, Gesellschaft für Biotechnologische Forschung, D-38124 Braunschweig, Germany3
Received 25 November 2002/ Returned for modification 29 January 2003/ Accepted 7 April 2003
|
|
|---|
|
|
|---|
The rdar morphotype is a multicellular behavior commonly expressed by Salmonella enterica serotype Typhimurium isolates (26) and certain isolates of Escherichia coli (2, 24). The rdar morphotype mediates different types of multicellular behavior, for example, cell aggregation in liquid culture, pellicle formation at the air-liquid interface, and biofilm formation at liquid-solid interfaces (17). The rdar morphotype produces an extracellular matrix consisting of cellulose and curli fimbriae, which is the major determinant of cell-cell interactions and cell adherence to hydrophilic and hydrophobic abiotic surfaces. Besides their distinct roles in bacterial self-organization, features related to virulence and transmission such as adherence and invasion of epithelial cells and chlorine resistance have been assigned to curli fimbriae and cellulose, respectively (3, 8, 13, 21, 22).
Gene required for cellulose biosynthesis are encoded by the bcsABZC operon. While bcsA encodes the cellulose synthase, the exact functions of the other genes remain to be determined. Structural genes for curli fimbriae are encoded by the csgBA(C) operon. Cellulose and curli fimbria biosynthesis is commonly regulated, either through adrA or directly by csgD, a transcriptional activator, divergently transcribed to csgBA(C).
Performing this study, we wanted to gain insights into the ability of clinically important Enterobacteriaceae besides E. coli and serotype Typhimurium to express cellulose and curli fimbriae. The isolates were collected from the gastrointestinal tract as the major human reservoir for dissemination of isolates.
Isolation and phenotypic characterization of bacterial strains. Since we suspected that cellulose and curli fimbria expression are unstable phenotypes upon subculture, we collected primary isolates of bacterial strains from 21 volunteers who provided fecal samples. Fecal swabs were immediately inoculated on MacConkey agar plates (Oxoid) and incubated overnight at 37°C. The confluent microbial outgrowth was collected and resuspended, and appropriate dilutions were plated on MacConkey agar plates for single colonies. A total of 25 non-E. coli isolates which belonged to 13 species were identified with the API 20E kit (BioMerieux, Nürtingen, Germany): Citrobacter sp. (isolated independently two times), Citrobacter freundii (n = 3), Citrobacter koseri/farmeri (n = 1), Enterobacter sp. (n = 1), Enterobacter aerogenes (n = 2), Enterobacter cloacae (n = 3), Enterobacter sakazakii (n = 2), Klebsiella sp. (n = 2), K. oxytoca (n = 2), K. pneumoniae (n = 2), Klyvera sp. (n = 2), Proteus mirabilis (n = 1), and Raoultella ornithinolytica (n = 2). When the fecal strains were grown on Congo Red (CR) plates (17), we observed that several of the isolates selectively bound CR, thereby resembling the rdar, bdar, or pdar morphotype of serotype Typhimurium reference strains (Fig. 1). Serotype Typhimurium colonies which show the rdar morphotype express cellulose and curli fimbriae, while colonies which show the pdar morphotype express cellulose and colonies which show the bdar morphotype express curli fimbriae (17, 26).
![]() View larger version (103K): [in a new window] |
FIG. 1. Congo red and Calcofluor staining of morphotypes of representatives of enterobacterial species isolated from feces. The strains were grown for 48 h at 28°C and for 24 h at 37°C. Morphotypes were compared with the phenotype of serotype Typhimurium UMR1 (ATCC 14028-1s, Nal r) and its mutants, which have been streaked at the lower part of the panel. ROWS: 1, Citrobacter sp. strain Fec2; 2, C. freundii Fec4; 3, C. koseri/farmeri Fec157; 4, Enterobacter sp. strain Fec125; 5, E. aerogenes Fec135; 6, E. cloacae Fec36; 7, E. sakazakii Fec39; 8, Klebsiella sp. strain Fec164; 9, K. oxytoca Fec139; 10, K. pneumoniae Fec141; 11, R. ornithinolytica Fec153; 12, P. mirabilis Fec162; 13, serotype Typhimurium UMR1 (cellulose 28+, curli 28+); 14, serotype Typhimurium MAE1 (cellulose 28+, curli -); 15, serotype Typhimurium MAE222 (cellulose -, curli 28+); 16, serotype Typhimurium MAE51 (cellulose -, curli -).
|
Expression of curli fimbriae and cellulose. Three strains with characteristic morphotypes, Citrobacter sp. strain Fec2, C. freundii Fec4, and E. sakazakii Fec39, were chosen for more detailed analyses. Citrobacter sp. strain Fec2 expressed the bdar morphotype at 28°C, C. freundii Fec4 expressed the rdar morphotype at 28°C, and E. sakazakii Fec39 expressed the pdar morphotype at 28°C and 37°C (Fig. 1). However, as displayed in Fig. 1, during storage for several months at -70°C, Citrobacter sp. strain Fec2 changed morphotype to rdar at 28°C and pdar at 37°C, indicating enhanced cellulose biosynthesis. However, since those changes occurred after the molecular study of the isolates, they will not be considered here.
After an enrichment procedure employed for plate-grown cultures of Citrobacter sp. strain Fec2 (18a), the putative curli fimbria subunit was detected as a prominent band at
16 kDa that was cut out, trypsin digested, and subjected to sequence analysis by quadruple time-of-flight mass spectrometry (20). Sequence comparison of three peptides with the database indicated that the protein was homologous to CsgA, the structural subunit of curli fimbriae expressed by E. coli and serotype Typhimurium. Western blot analysis detected curli fimbriae after formic acid treatment (5) produced by Citrobacter sp. strain Fec2 and C. freundii Fec4 at 28°C, while E. sakazakii Fec39 did not produce any curli fimbriae (data not shown).
Calcofluor binding indicated variable production of cellulose, a 1,4-ß-glucan, in the three strains (Fig. 1). Since Calcofluor binding is not absolutely specific for cellulose, we confirmed cellulose production by isolating crystalline cellulose by the Updegraff method with a mixture of hot 58% acetic acid and 19% nitric acid (25, 26). After hydrolysis of the isolated polymer, the glucose monomers detected by high-pressure liquid chromatography (HPLC) (Fig. 2) were indicative of cellulose production. While glucose was detected when samples from E. sakazakii Fec39 grown at 28°C and 37°C and C. freundii Fec4 grown at 28°C were analyzed, Citrobacter sp. strain Fec2 samples did not show any significant glucose peak.
![]() View larger version (16K): [in a new window] |
FIG. 2. Detection of glucose monomers by HPLC after isolation of crystalline cellulose. Results for (A) E. sakazakii Fec39 and (B) serotype Typhimurium MAE51 are shown as examples of positive and negative outcomes, respectively. Crystalline cellulose was hydrolyzed and sugar monomers were detected by HPLC. Glucose shows two peaks representing the - and ß-anomers (retention time, 13.77 and 13.95 min, respectively); 1 µg of myoinositol was used as an internal standard (retention time, 15.75 min).
|
Multicellular behavior of isolates. The expression of the extracellular matrix components cellulose and curli fimbriae is associated with biofilm formation and other modes of multicellular behavior (17). We used a steady-state model, incubation of the strains in glass tubes (17), to test biofilm formation by Citrobacter sp. strain Fec2, C. freundii Fec4, and E. sakazakii Fec39. Optimal biofilm formation was achieved with different incubation conditions for each strain (Fig. 3A and data not shown), which indicated that the regulatory patterns of biofilm formation with respect to oxygen tensions were different between the strains. Pellicle formation was observed for all three strains (data not shown). While E. sakazakii Fec39 formed a pellicle after just 24 h of incubation at 28°C in standing cultures, pellicle formation in cultures of Citrobacter sp. strain Fec2 and C. freundii Fec4 required 48 h to develop. In liquid culture at 28°C after incubation for 24 h, all three strains formed clumps, although the extent of clumping and the consistency of the clumps varied between the strains (Fig. 3B and data not shown). When observed by fluorescence microscopy, Calcofluor-stained cellulose fibrils could be observed for E. sakazakii Fec39 and C. freundii Fec4 (Fig. 3B). Free-floating cellulose fibrils associated with cell clumps were observed for E. sakazakii Fec39, as in serotype Typhimurium when cellulose is the sole extracellular matrix component (26). On the other hand, C. freundii Fec4 produced cellulose fibrils which were tightly wrapped around the cells, as in serotype Typhimurium when cellulose is coexpressed with curli fimbriae (17). Electron microscopy studies of the extracellular matrix of plate-grown colonies of E. sakazakii Fec39 and C. freundii Fec4 further supported the view that the extracellular matrix components cellulose and curli formed structures similar to those in serotype Typhimurium (Fig. 3C).
![]() ![]() ![]() View larger version (437K): [in a new window] |
FIG. 3. Phenotypes of Citrobacter sp. strain Fec2, C. freundii Fec4, and E. sakazakii Fec39. (A) Biofilm formation. Cultures were grown at 28°C for 48 h in Luria broth without salt and with shaking. Optimal biofilm formation was achieved at 180 and 210 rpm for Citrobacter sp. strain Fec2, C. freundii Fec4, and E. sakazakii Fec39, respectively. (B) Cell clumping and arrangement of Calcofluor-stained cellulose fibrils in C. freundii Fec4 (top) and E. sakazakii Fec39 (bottom). Note that C. freundii Fec4, due to the expression of cellulose and curli fimbriae, formed tighter clumps than E. sakazakii Fec39, which only expressed cellulose. Left, phase contrast; right, fluorescence microscopy. Magnification, x600. (C) Scanning electron microscopy of plate-grown cells. Upper panel: C. freundii Fec4 (left) in comparison with serotype Typhimurium MAE52 (right; cellulose +, curli +). Lower panel: E. sakazakii Fec39 (left) in comparison with serotype Typhimurium MAE97 (right; cellulose +, curli -). Bars, 1 µm.
|
For bcsA, the following consensus primers were used: BCSA74A (CTTCCGTATTGGCAGTCAGGTTCAGGACG) and BCSA70 (GCGCCAGCGGGTTAAACGGCTG) for the N terminus, BCSA74 (GCAACAGATTCAATTTCTGCCCTTC) and BCSA86 (GCACCCGC-GCTGGCAGCGTATTCG) for the middle part of the gene, and BCSA62 (TGGGTCTTCTACAACCTGATTA) and BCSA62A (GCGGCGGTGCAATTTGCGCAAAGGT) for the C terminus.
Sequence data were submitted to the EMBL data library under the following accession numbers: Citrobacter sp. strain Fec2, csgD-csgBA, AJ515700; bcsA, AJ515698; C. freundii Fec4, csgD-csgBA, AJ515701; E. sakazakii Fec39, csgD-csgBA, AJ515702; bcsA, AJ515699. Although E. sakazakii Fec39 did not express curli fimbriae, it contained the csgD-csgBA region with intact genes.
CsgD, the central regulator for multicellular behavior, was highly conserved among the species (data not shown). Most diverse were the CsgD proteins from E. sakazakii Fec39 and serotype Typhimurium ATCC 14028, with a homology of 94%. While several amino acid substitutions were found in the N-terminal domain, a single amino acid substitution was found in the DNA binding domain.
csgD expression is highly regulated by environmental conditions mediated by response regulators and DNA architectural proteins (9, 12), which bind in the intergenic region. Surprisingly, binding sites for OmpR and IHF are not located in the most conserved regions (data not shown). This finding suggested that OmpR and IHF binding might differ among the species and consequently the response to environmental conditions. The 521-bp csgD-csgBA intergenic region can be divided into four regions, IR1 to IR4 (17). The sequence of the IR3 region is highly diverse among the species, but its length is conserved in all strains, suggesting structural importance.
CsgA and CsgB, the structural proteins of curli fimbriae, were also highly conserved (Fig. 4A and data not shown). Most diverse were the CsgA and CsgB proteins from C. freundii Fec4 and E. sakazakii Fec39 with a homology of 80% and 78%, respectively. Polymerized CsgA of Salmonella enteritidis possesses two domains, an N-terminal domain of 22 residues, which is proteinase K susceptible, and a C-terminal core domain, which is proteinase K resistant (6). The highest sequence diversity among the CsgA proteins was found in the glycine-rich N-terminal domain, which is proteinase K susceptible. The C-terminal domain of each protein consisted of five tandemly arranged sequences each containing the SX5QXGX2NXAX3Q consensus sequence (6).
![]() ![]() ![]() View larger version (147K): [in a new window] |
FIG. 4. Comparative analysis of structural genes required for curli and cellulose biosynthesis. (A) Multiple alignment of CsgA sequences from different Enterobacteriaceae. The alignment was carried out with ClustalX in default mode (23). Amino acids identical in all the proteins are displayed in bold. An arrow indicates the cleavage site for the signaling peptide. The five internal repeats (C5a to C5e) are boxed. Proteinase-sensitive (light bar) and proteinase-resistant (dark bar) regions are indicated above the alignment. N- and C-terminal amino acid stretches identified in E. coli to mediate protein binding (15) are indicated by bars under the alignment. Three tryptic peptides of CsgA from Citrobacter sp. strain Fec2 were subjected to mass spectrometric sequence analysis, and the partial sequences (underlined), identical within the limits of technical accuracy to the translated csgA gene product, were detected. From the molecular mass of tryptic peptide 1, M+H: 3,990.7 (theoretic mass: 3,990.9), the amino terminus of the protein could be deduced as GSVP. CsgA sequences of E. coli K-12 (P28307) and serotype Typhimurium ATCC 14028 (P55225) were taken from the Swiss Prot database. Abbrevations: Sathy, serotype Typhimurium UMR1; Ecoli, E. coli K-12; Citsp, Citrobacter sp. strain Fec2; Citfr, C. freundii Fec4; Entsa, E. sakazakii. (B) Multiple alignment of the core region of enterobacterial BcsA, the catalytic subunit of the cellulose synthase. Conserved amino acids are displayed as described for panel A. Conserved motifs are indicated by boxes above and below the sequence. Abbrevations: Kpneu, K. pneumoniae; Psepu, Pseudomonas putida KT2440. BcsA sequences of E. coli K-12 (P37653) and serotype Typhimurium ATCC14028 (Q93IN2) were taken from the Swiss Prot database. BcsA from P. putida (http://www.tigr.org) and K. pneumoniae (http://genome.wustl.edu/projects/bacterial/?kpneumoniae=1) were from unfinished genome projects. (C) Phylogeny for BcsA sequences with BcsA from P. putida as an outgroup. After alignment of sequence, the tree was constructed by the neighbor-joining method subjected to 1,000 bootstrap trials. The tree was drawn with TreeView.
|
Sequence alignment of BcsA, the cellulose synthase, is shown in Fig. 4B. Previously, sequence comparison of all available bacterial cellulose synthases detected a highly conserved core region of 350 residues which contains the D3D2D35QRXRWA motif, common to processive ß-glycosyltransferases, and five additional motifs (16). Although the overall sequence was highly conserved among the species over the whole length of the protein, again BcsA of E. sakazakii Fec39 showed the highest sequence diversity (Fig. 4C). Most remarkable, the PVDPYE and the FFCGS motif were replaced by a different sequence (Fig. 4B).
Detection of csgA, csgD, and bcsA in other enterobacterial species isolated from feces. Conserved primers were used to detect the presence of csgA, csgD, and bcsA genes in other species isolated from feces (Fig. 5 and data not shown). bcsA was present in all species. In E. aerogenes Fec135 and R. ornithinolytica Fec153, csgD could not be detected, while csgA was present. We expect csgD to be present but not detectable by the primer pair used, because csgA is present. In summary, the genes required for the biosynthesis of curli fimbriae were present in all species investigated besides K. pneumoniae and K. oxytoca. Klebsiella ornithinolytica, which has recently been shown to belong to a phylogenetic line distinct from K. pneumoniae and K. oxytoca and therefore has been renamed R. ornithinolytica (10), harbors the gene cluster for curli biosynthesis.
![]() View larger version (55K): [in a new window] |
FIG. 5. Amplification of csgA (A) and bcsA (B) genes by PCR in enterobacterial species isolated from feces. PCR products of 1,301 bp and 860 bp were detected for csgA and bcsA, respectively. Lanes: M, size markers (SmartLadder), 200 to 10,000 kbp (Eurogentec); C, control serotype Typhimurium MAE52; 1, E. cloacae Fec36; 2, E. aerogenes Fec135; 3, C. koseri/farmeri Fec157; 4, K. oxytoca Fec139; 5, K. pneumoniae Fec141; 6, R. ornithinolytica Fec153. Primers for csgA were AGFA66 and AGFA60. bcsA primers were ES-BCSA162 (GACGATCTCTACCAGGTCTGG) and ES-BCSA681 (GTAACGCCACCAGATATAGCGG).
|
In the family Enterobacteriaceae, cellulose biosynthesis has been shown to occur in Salmonella spp., E. coli, and K. pneumoniae (21, 26). In this study, we found that most of the Enterobacter isolates produced cellulose. In particular, two unrelated isolates of E. sakazakii showed pronounced cellulose production at 28°C and 37°C (Fig. 1 and data not shown). Clinical isolates of E. sakazakii have been reported to display a peculiar wrinkled, leathery colony morphology, which, upon storage as agar stocks, readily dissociated into the standard smooth colony morphology (11). Based on the results from this study, this colony morphology is due to cellulose production, which is retained by E. sakazakii upon infection. Species from the genus Klebsiella and R. ornithinolytica isolated from feces did not show elevated expression of cellulose (Fig. 1), although the genetic information for the biosynthesis of cellulose is present (Fig. 5 and data not shown). However, as a possibility, cellulose biosynthesis might be induced in response to specific environmental signals, as shown for the plant symbiont Rhizobium leguminosarum bv. trifolii by contact with roots (1, 7). The observed species-specific expression pattern of curli fimbriae and cellulose in fecal strains raises the question of whether species-specific and consequently morphotype-specific niches exist in the gastrointestinal tract. Alternatively, a habitat-specific morphotype for each species is already determined outside the human body and retained when a strain passes through the gastrointestinal tract.
Multicellular behavior, including biofilm formation, is a fundamental life style of bacteria. It is reasonable to assume that such ancient behavior has common principles among bacteria. The rdar morphotype, defined as the expression of cellulose and curli fimbriae, represents a basic multicellular behavior of Enterobacteriaceae closely related to serotype Typhimurium and E. coli, but individual components of the rdar morphotype and its expression pattern are subject to species-specific adaptation for which no simple pathogen-commensal soil organism classification scheme can be proposed. For example, although semiconstitutive expression of the rdar morphotype is observed (17), the vast majority of pathogenic serovar Typhimurium and serotype Enteritidis strains express the rdar morphotype highly regulated by environmental conditions and only at ambient temperature (18, 18a). Otherwise, Shigella spp. and enteroinvasive E. coli, which cause invasive gastrointestinal disease (19), have lost the rdar morphotype and multicellular behavior. The rdar morphotype was also not found in the Klebsiella species analyzed in this study. However, constitutive expression of cellulose was observed in E. sakazakii, which can cause newborn meningitis, and the majority of E. coli strains from sepsis produce curli fimbriae or the rdar morphotype constitutively (2). Therefore, the expression of cellulose and curli fimbriae is complex and might be highly determined by the environmental microniche of the organism under study. In any case, the capacity to produce cellulose and curli fimbriae in Citrobacter, Enterobacter, and potentially Klebsiella species might contribute to the significant role those species play in biofilm-related infections.
The excellent technical assistance of J. Majewski and A. Tiepold in mass spectrometric analysis is gratefully acknowledged. We thank Anne von Euler-Matell for electron microscopy of strains.
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2010 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»