Previous Article | Next Article ![]()
Eukaryotic Cell, August 2006, p. 1266-1275, Vol. 5, No. 8
1535-9778/06/$08.00+0 doi:10.1128/EC.00022-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Department of Microbiology, University of Tennessee, Knoxville, Tennessee
Received 25 January 2006/ Accepted 12 June 2006
|
|
|---|
mutant was found to exhibit reduced FLO11 levels. Consequently, the opi1
mutant perturbs the FLO11-dependent phenotype of invasive growth. The opi1
mutant's defects in mat formation and invasive growth are dependent on the transcriptional activator Ino2p. These results indicate that Opi1p affects mat formation and invasive growth by participating in the regulation of FLO11. |
|
|---|
The bakers' yeast Saccharomyces cerevisiae has been used as a powerful genetic model system for studying the yeasts. S. cerevisiae is able to generate a complex multicellular structure called a mat on the surface of low-density agar (0.3%) petri plates made with rich (yeast extract-peptone-dextrose [YEPD]) medium (called low-agar plates) (54). A mat is formed on low-agar plates through a process called sliding motility, where the cells spread as they grow over the surface of the medium (42, 53). As the mat grows, it spreads and generates patterns in the middle (called the hub), while the leading edge (called the rim) remains smooth in appearance (Fig. 1).
![]() View larger version (131K): [in a new window] |
FIG. 1. S. cerevisiae forms a mat on low-agar plates. Wild-type yeast cells were inoculated on the center of YEPD plates containing 0.3% agar with a toothpick and grown at 23°C for 5 days.
|
Multicellular behaviors of microorganisms appear to be dependent on both the nutrient environment and the surface properties of the substrate on which they are growing (36, 46). An important nutrient cue that drives mat formation in S. cerevisiae is glucose deprivation (T. B. Reynolds, A. Jansen, X. Peng, and G. R. Fink, unpublished results). In haploid bakers' yeast, glucose deprivation also drives invasive growth on 2% agar YEPD plates (11) as well as biofilm formation on polystyrene (54).
The role that the surface plays in regulating multicellular behavior in S. cerevisiae remains to be explored. In this report, transcriptional profiling was used to compare cells grown in a mat to cells grown in liquid medium in suspension cultures (planktonic cells). This analysis revealed several genes that appear to be expressed in response to surface association. One of these genes is INO1, which is regulated by the inositol regulon. The inositol regulon consists of three transcription factors, Ino2p, Ino4p, and Opi1p, that control expression of INO1 and other phospholipid biosynthetic genes (3, 24, 31). This communication demonstrates that Opi1p is required for the full expression of FLO11. The effect of Opi1p on FLO11 expression is dependent on the transcriptional activator Ino2p.
|
|
|---|
1278b. They are all isogenic to the strain TBR1 (54). The strains and their genotypes are shown in Table 1. With the exception of the ino1
mutant, the mutants in this study were made by PCR-based gene disruptions using specific deletion mutants from the Yeast Knock Out Collection (Open Biosystems/Research Genetics) as templates (78). Each strain in the Yeast Knock Out Collection carries a gene disruption in which a particular open reading frame (ORF) is replaced by the G418 resistance cassette kanMX4 (76). Primers were used to amplify DNA encompassing the kanMX4-disrupted ORF of interest plus
300 base pairs flanking either side of the ORF (Table 2). The resulting PCR product was then transformed into TBR1 by the lithium acetate transformation method (20), and transformants were selected on YEPD plates (71) containing 200 µg/ml G418 (54). PCR with primers that annealed outside of the disruption construct (Table 2) and inside the TEF promoter of the kanMX4 cassette (TRO369) (Table 2) (76) were used to confirm each construct. The ino1
mutant was made as described previously (22) using the primers listed in Table 2. The ino1
opi1
mutant was made by disrupting OPI1 in the ino1
mutant as described above. The ino2
opi1
double mutant was created by standard tetrad dissection techniques (63). Strains were maintained on YEPD plates (71), and experiments were performed on YEPD plates or in liquid YEPD medium (71) as indicated. Low-agar plates (54) (YEPD plates with only 0.3% agar) were used for mat assays, Northern blots, and microarray analysis as indicated. |
View this table: [in a new window] |
TABLE 1. Yeast strains used in this study
|
|
View this table: [in a new window] |
TABLE 2. Primers used in this study
|
Microarray analysis comparing mat and planktonic cells.
Cells were isolated from mat cultures as follows. Mats were grown at
25°C on low-agar (0.3%) YEPD plates (54) for 7 days. Mats were collected from the plate by scooping up two whole mats per sample with a spoon and placing the cells in a 50-ml conical centrifuge tube (Fisher) containing 5 ml of ice-cold water and vortexing to homogenize the cells and agar. The cells were then pelleted by centrifugation for 5 min at 4°C. The supernatant and most of the agar were decanted, and the cell pellet and residual agar were flash-frozen in a bath of dry ice and ethanol. Planktonic cells were grown to logarithmic (log) (optical density at 600 nm [OD600],
0.8) or post-logarithmic (post-log) (OD600,
8.0) phase in 100 ml of liquid YEPD medium in a shaker at 25°C. The cells were collected by centrifugation, and the cell pellets were flash-frozen. Two biological replicates were performed for each condition described above. Total RNA and mRNA were isolated and labeled as described previously (39). mRNA was prepared and probed on Affymetrix microarrays as described previously (58). Microarray data were normalized and floored to 20 as described previously (58), replicate data sets were averaged, and ratios of gene expression were generated by comparing the biofilm cultures and post-log-phase cultures to the log-phase cultures. These data were then clustered using the software Cluster and TreeView (15). During clustering, the data were filtered such that genes were defined as upregulated or downregulated if there were reliable data for at least 80% of the conditions and the genes had expression ratios greater than a minimum variance of 2.0 log2 units (4.0-fold) between lowest and highest expression values (38). Probes on the yeast genome S98 chips that did not correspond to specific ORFs or known genomic elements were disregarded in this analysis. The whole set of normalized data is available in the supplemental material.
Northern blotting.
Northern blotting was performed as described in the work of Sambrook and Russell (59) by use of Church buffer for prehybridization and hybridization steps (8). A Techne Hybrigene oven set to 60 or 65°C was used for all incubation and wash steps. For Northern blots of planktonically grown cells, the cells were grown in liquid YEPD medium (2% glucose) to logarithmic phase (OD600 of between 0.7 and 1.0). The cells were collected by centrifugation, aspirated, flash-frozen in a bath of dry ice and methanol, and stored at 80°C. For Northern blots of mats, cells were collected with a spatula from mats grown for 5 days at 23°C and pelleted in a microfuge, and the agar was decanted. Total RNA was collected by acid-phenol extraction (32), and 10 µg of total RNA was subjected to Northern blotting. A PCR product (primers were TRO367, ATGCAAAGACCATTTCTACTCG, and TRO368, TGCCAGGAGCTTGCATATTGAG) corresponding to the first 484 bp of the FLO11 ORF was used to probe the total RNA. The probe was labeled with [
-32P]ATP using a Prime-It II random primer kit (Stratagene), and following probing, the blots were visualized on a Storm phosphorimager. The data were quantified using ImageQuant software. FLO11 expression was normalized to the expression level of ACT1, which was probed on the same membrane. The ACT1 probe was generated with the primers TRO480, GCCGGTTTTGCCGGTGACGAC, and TRO 481, GCCGGTTTTGCCGGTGACGAC. The primer TRO481 was later discovered to contain two mismatches with the ACT1 sequence; however, the resulting probe was sequenced and revealed to be specific for the yeast ACT1 gene.
|
|
|---|
One large cluster consisted of 99 genes that were upregulated in the mat and post-log-phase cultures compared to log-phase cultures (Fig. 2A). These genes largely overlapped (85%) with genes that have been shown to be upregulated during diauxic shift and stationary phase (12, 19). This overlap suggested that the gene expression profile of cells in a mat is similar to that of cells that are experiencing diauxic shift or entering stationary phase. Despite this similarity, the mat continued to expand, indicating that the cells in the mat, or at least those in the rim, continued to grow. The post-log-phase cells, conversely, experienced less than one doubling before arresting growth.
![]() View larger version (69K): [in a new window] |
FIG. 2. Hierarchical clustering analysis revealed a unique transcriptional profile associated with mat growth on a low-agar plate. mRNA was collected from growing mats on low-agar plates or from liquid batch cultures (planktonic) at log or post-log phase. The mRNA was probed on Affymetrix yeast genome S98 chips, and ratios were generated by comparing data from mat and post-log-phase cultures to those from log-phase cultures. The ratios were subjected to hierarchical clustering via the programs Cluster and TreeView (http://rana.lbl.gov/EisenSoftware.htm) (15). See Materials and Methods for more details. Yeast systematic ORF names are shown to the right of clusters, and commonly accepted gene names are shown to the right of the systematic names where applicable. (A) A cluster of genes that was upregulated in both post-log-phase and mat cultures. (B) A cluster of genes that exhibited a large downregulation in post-log-phase cultures but not in mat cultures. (C) A cluster of genes that showed a large upregulation in mat cultures but not in post-log-phase cultures. The gene expression conditions shown are post-log phase (lanes 1), hub (lanes 2), rim (lanes 3), and whole mats (lanes 4). An asterisk marks genes that appear twice in a cluster due to more than one set of probes on the Affymetrix array for those select genes.
|
There was a cluster of eight genes that were highly upregulated in the mat cultures but not in the post-log-phase cultures (Fig. 2C). These eight genes exhibited an average 38-fold increase in gene expression in the mat but an average 1.9-fold increase in the post-log-phase cultures. All eight genes except STR3 were disrupted, and the mutants were examined to determine if these mutations affected mat formation. None of these mutants exhibited a defect in mat formation compared to the wild-type strain.
The opi1
mutant exhibits a defect in mat formation and invasive growth.
Three of the genes (INO1, PAU2, and DAN1) upregulated in the mats but not in the post-log-phase cultures (Fig. 2C) have been shown to be regulated by members of what this report refers to as the inositol regulon (3, 24, 31). This prompted a further investigation into the role of this regulon in mat formation. The inositol regulon consists of three transcription factors, Ino2p, Ino4p, and Opi1p, that control expression of several phospholipid biosynthetic genes (3, 24, 31). The most heavily regulated of these genes is INO1, which encodes the myo-inositol-1-phosphate synthase. Ino1p converts glucose-6-phophate into inositol-1-phosphate (13, 41). Ino2p and Ino4p form a heterodimeric transcriptional activator of INO1 expression (3), while Opi1p acts antagonistically as a transcriptional repressor of INO1 (24).
Gene disruptions of INO2, INO4, and OPI1 revealed that OPI1 has a significant effect on mat formation, and INO2 and INO4 have minor effects. During mat formation, the wild-type strain exhibits development of a distinct hub and rim, forms spokes, and expands over the surface to a width of 41 ± 2.4 mm by day 5 (Fig. 1 and 3). The opi1
mutant was retarded in morphological development and spreading compared to the wild-type strain (Fig. 3). This mutant did not develop spokes, had a poorly developed hub, and did not spread well, expanding to only 26 ± 2 mm by day 5.
![]() View larger version (68K): [in a new window] |
FIG. 3. Disruption of OPI1 perturbs mat formation. Wild-type (WT) and mutant strains were inoculated onto low-agar plates with toothpicks and grown for 5 days at 23°C. Photographs were taken at 3 days (top panels) and 5 days (bottom panels) of growth. Bars, 1 cm.
|
and ino4
mutants had only a slight defect in mat formation, producing mats that did not spread as well as the wild-type strain but had a similar morphology (Fig. 3). The ino2
and ino4
mutant mats spread to only 35 ± 1.2 mm and 35 ± 1.7 mm, respectively, by day 5. Development of the spokes in the ino2
and ino4
mutants appeared to precede that in the wild-type strain slightly but reproducibly (Fig. 3, 3 days).
The effects of the opi1
mutant on mat formation seemed likely to be a consequence of a novel effect on FLO11 regulation. Northern blotting revealed that in liquid medium the opi1
mutant had a significant reduction in FLO11 expression (Fig. 4). The defect in FLO11 expression shown by the opi1
mutant in planktonic culture could also be seen when FLO11 expression was assessed in the mat (Fig. 5). Conversely, examination of the ino2
and ino4
mutants revealed that they exhibited modest increases in FLO11 expression compared to that seen for the wild-type strain when grown planktonically (Fig. 4) or as mats (Fig. 5).
![]() View larger version (53K): [in a new window] |
FIG. 4. The inositol regulon mutants misregulate FLO11 expression during planktonic growth. (A) Strains were grown in YEPD liquid medium to log phase, collected, and subjected to Northern blotting against FLO11. ACT1 was reprobed on the same membrane as a loading control. (B) ImageQuant software was used to quantify the expression of FLO11 by the strains, and ACT1 was used to normalize for loading differences. Three replicate experiments are represented in the graph, and the results are presented as percentages of the level of FLO11 expression by the wild-type (WT) strain. An asterisk indicates that the mutant shows a statistically significant difference from the wild-type strain.
|
![]() View larger version (51K): [in a new window] |
FIG. 5. The inositol regulon mutants misregulate FLO11 expression during mat formation. (A) Strains were inoculated on low-agar plates with toothpicks and grown for 5 days at 23°C. Cells were collected and subjected to Northern blotting against FLO11. ACT1 was reprobed on the same membrane as a loading control. (B) ImageQuant software was used to quantify the expression of FLO11 by the strains, and ACT1 was used to normalize for loading differences. Three replicate experiments are represented in the graph, and the results are presented as percentages of the level of FLO11 expression by the wild-type (WT) strain. An asterisk indicates that the mutant shows a statistically significant difference from the wild-type strain.
|
mutant cells were removed (Fig. 6B). Conversely, the ino2
and ino4
mutants exhibited a slight increase in invasive growth compared to the that of the wild type. This could not be seen when the cells are washed from the plate with water alone (Fig. 6B). However, when the plate was rubbed with a gloved finger to more thoroughly remove noninvasive cells, it was revealed that the ino2
and ino4
mutants exhibited a level of invasive growth slightly greater than that of the wild type (Fig. 6C).
![]() View larger version (128K): [in a new window] |
FIG. 6. Mutations in the inositol regulon affect invasive growth. (A) The strains were streaked onto YEPD plates and grown at 30°C for 5 days. (B) The plates were then washed with water to remove noninvasive cells. (C) The plates were washed with water and rubbed with a gloved finger to more thoroughly remove noninvasive cells. The diagram at lower right indicates the placement of the different strains on the plates shown in B. WT, wild type.
|
mutant's defect in mat formation and invasive growth is dependent on INO2.
Overexpression of INO1 by the opi1
mutant is dependent on the Ino2p transcriptional activator. An ino2
mutant fails to overexpress INO1 in an opi1
background (23). The opi1
mutant's failure to undergo invasive growth (Fig. 6) and mat formation (Fig. 3) was likewise dependent on INO2. An ino2
opi1
double mutant behaved like an ino2
mutant rather than an opi1
mutant in both mat formation (Fig. 7A) and invasive growth (Fig. 7B).
![]() View larger version (75K): [in a new window] |
FIG. 7. The ino2 mutation is epistatic to the opi1 mutation for controlling mat formation and invasive growth (A) Mat formation. The wild type, the opi1 and ino2 mutants, and the opi1 ino2 double mutant were inoculated on the center of low-agar plates with toothpicks and grown at 23°C for 5 days. (B) Invasive growth. The strains were streaked onto YEPD plates and grown at 30°C for 5 days (left panel). The plates were then washed with water to remove noninvasive cells (middle panel). The plates were washed with water and rubbed with a gloved finger to more thoroughly remove noninvasive cells (right panel). The diagram to the far right indicates the placement of the different strains on the plates shown in B. WT, wild type.
|
mutant on mat formation and invasive growth was that increased production of INO1 and inositol in the opi1
mutant might regulate these phenotypes (45). If this were the case, then an ino1
mutation should suppress defects in mat formation and invasive growth exhibited by an opi1
mutant. However, analysis of an opi1
ino1
double mutant revealed that it did not suppress the mat formation or invasive growth defects exhibited by an opi1
mutant (Fig. 8).
![]() View larger version (73K): [in a new window] |
FIG. 8. The opi1 mutation does not affect mat formation or invasive growth by overexpressing the INO1 gene. (A) Mat formation. The wild type, the opi1 and ino1 mutants, and the opi1 ino1 double mutant were inoculated on the center of low-agar plates with toothpicks and grown at 23°C for 5 days. (B) Invasive growth. The strains were streaked onto YEPD plates and grown at 30°C for 5 days (left panel). The plates were then washed with water to remove noninvasive cells (middle panel). The plates were washed with water and rubbed with a gloved finger to more thoroughly remove noninvasive cells (right panel). The diagram to the far right indicates the placement of the different strains on the plates shown in B. WT, wild type.
|
|
|
|---|
Mats exhibit a unique transcriptional profile. There was a large cluster of genes upregulated in the mat that were also upregulated in post-log-phase cells (Fig. 2A), and most of these genes are known to be upregulated in diauxic shift and stationary phase (12, 19). This suggests that compared to log-phase cells, cells throughout the mat are experiencing nutrient deprivation. However, in contrast to cells entering diauxic shift/stationary phase, mat cells did not experience a large downregulation of protein synthesis genes (Fig. 2B). One explanation for this discrepancy is that cells in the mat are continuing to actively grow or that at least a subpopulation of cells in the mat are clearly dividing as the mat continues to spread on the plate over the course of several days.
These results share some similarity with observations reported for the surface-associated phenomenon of biofilm formation in the fungus Candida albicans (18). A comparison of C. albicans biofilm cells to planktonic cells by transcriptional profiling revealed that the biofilm cells expressed higher levels of protein synthesis genes. The reasons for the upregulation of protein synthesis genes in C. albicans biofilms are not known. Upregulation of protein synthesis genes may be common to many biofilms, because it has been observed for Escherichia coli biofilms as well (18, 60).
One gene that was unexpectedly absent from the upregulated genes in the cluster analysis is FLO11. Examination of FLO11 expression in the microarray data revealed that it showed only a modest increase in post-log-phase cells (1.8-fold) or in mat cells (average of 2.0-fold) compared to log-phase cells, which explains why it was not included in the clusters. Robust FLO11 expression in log-phase cells grown in YEPD (Fig. 4) is most likely the reason for these small fold increases.
Surface-specific gene expression suggests an anaerobic environment in the mat. Eight genes were upregulated in a surface-specific manner (Fig. 2C), and among these were three genes (TIR4, DAN1, and PAU2) that encode secreted/cell wall proteins that are upregulated in response to anaerobic growth (1, 52). The upregulation of TIR4, DAN1, and PAU2 suggests that the cells in the mat are experiencing conditions that are more anaerobic than those experienced by planktonic cells. This is consistent with what has been observed for E. coli biofilms (51). The role that these anaerobic response genes have in mat formation is unclear, as disruption of PAU2, TIR4, or DAN1, or TIR4 and DAN1 simultaneously, did not have an effect on mat formation. There are many homologs for all three of these genes in S. cerevisiae that may be redundant in function (1, 52). However, disruption of the transcription factor UPC2, which controls expression of DAN1, TIR4, and their family members (2), had no effect on mat formation (T. B. Reynolds, unpublished data).
Opi1p interacts with several signal transduction pathways that regulate FLO11. A disruption of INO1 revealed no obvious role for this gene in mat formation (Fig. 8). Despite the lack of a defined function for INO1 or inositol in this phenotype, the OPI1 gene affects mat formation by activating FLO11 expression (Fig. 4 and 5). The role that Opi1p plays in activating FLO11 expression may be multifaceted, as Opi1p has been found to interact with components of several pathways that control invasive or pseudohyphal growth. These pathways include the protein kinase A (PKA) pathway (44, 48, 57), the Snf1 kinase pathway (11, 33, 34), and the unfolded protein response (UPR) pathway (5, 61, 67, 68).
The PKA pathway is required to activate invasive growth (44, 48, 57). Disruption of TPK2, a catalytic subunit of the PKA pathway, prevents cells from undergoing invasive growth because Tpk2p phosphorylates the transcription factors Flo8p and Sfl1p, which regulate FLO11 (49). Tpk2p phosphorylation of Flo8p, a transcriptional activator required for upregulation of FLO11, promotes Flo8p binding of the FLO11 promoter (49). Tpk2p phosphorylation of Sfl1p, a repressor of FLO11 transcription, inhibits Sfl1p binding of the FLO11 promoter (49). Opi1p has been shown to be a substrate for PKA phosphorylation, which enhances its activities as a repressor of INO1 (69). Thus, Tpk2p could act to enhance Opi1p activity, which may contribute to FLO11 upregulation. However, it is not clear which PKA subunit actually phosphorylates Opi1p in vivo.
The Snf1 kinase, like Opi1p, has been shown to regulate both INO1 and FLO11 transcription. Snf1p is required for upregulation of INO1, and it does so by regulating the binding of the TATA binding protein to the promoter of INO1 (64-66). Snf1p upregulates FLO11 by inactivating the transcriptional repressors Nrg1p and Nrg2p, thus derepressing transcription of FLO11 (33, 34). Microarray experiments have indicated that Opi1p can regulate the expression of the NRG2 repressor gene (30). It is possible that Opi1p contributes to the regulation of FLO11 by affecting NRG2 expression.
The UPR pathway regulates both INO1 and pseudohyphal growth (7, 9, 61). In the UPR pathway, the endoplasmic reticulum resident protein Ire1p senses accumulated unfolded proteins in the endoplasmic reticulum. Ire1p then catalyzes the splicing of an intron out of the HAC1 mRNA (HAC1u) to generate the spliced form of HAC1 (HAC1i). HAC1i is translated into the transcription factor Hac1i, which upregulates UPR target genes (40). It has been reported that in hac1
/hac1
and ire1
/ire1
mutants, pseudohyphal growth is not properly repressed (61). In addition, hac1
and ire1
mutants do not express INO1 (7, 9). Hac1i has been shown to antagonize the function of the Opi1p repressor with regards to the repression of INO1 (5). It may have a similar relationship to Opi1p for regulating FLO11 and mat formation.
The position that Opi1p and the inositol regulon take in influencing FLO11 expression, invasive growth, and mat formation in relationship to these pathways is currently under investigation. In addition, it is possible that there are as-yet-undiscovered connections to other pathways known to regulate filamentous growth (16, 17, 47, 74), such as the filamentous growth mitogen-activated protein kinase cascade (10, 55, 56, 74).
Opi1p may affect FLO11 expression by an indirect mechanism. Opi1p has been defined as a transcriptional repressor, so it was unexpected to find that it is required to activate FLO11 expression. This suggests a model in which Opi1p acts to repress another downstream repressor gene that acts directly on FLO11. Several known repressors of FLO11, including Nrg1p, Nrg2p, Sf1lp, and Sok2p (33, 34, 49, 50, 57), are attractive candidates. This, however, does not rule out the existence of a novel repressor. These possibilities are currently under investigation.
An alternative explanation is that Opi1p has a previously undefined role as a transcriptional activator. However, this seems unlikely, because the opi1
mutant requires a functional Ino2p transcriptional activator to perturb mat formation and invasive growth (Fig. 7). Genetic analysis of the ino2
and opi1
mutations revealed that the ino2
mutation is epistatic to the opi1
mutation for controlling mat formation and invasive growth (Fig. 7). The ino2
opi1
double mutant behaves like an ino2
mutant. This is similar to what has been observed for the regulation of INO1 by Ino2p and Opi1p. Previous studies have shown that an ino2
opi1
double mutant behaves like an ino2
single mutant and is unable to grow in the absence of inositol (23). Thus, Opi1p is mostly likely acting as a repressor of a gene upregulated by Ino2p. This target gene may encode a repressor of FLO11 or regulate a repressor of FLO11.
One possible candidate for this downstream gene is INO1. Overexpression of INO1 might influence intracellular signaling and affect FLO11 expression due to the production of large concentrations of inositol. Pseudohyphal growth in Candida tropicalis can be repressed by the addition of extracellular inositol (45). If this were the case, then a disruption of INO1 in the opi1
mutant should restore the wild-type phenotype. However, this was not the case, as the ino1
opi1
double mutant continued to behave like the opi1
mutant (Fig. 8).
The role of inositol regulon orthologs in other fungi. Homologs of Opi1p and other inositol regulon components are found in other fungi, including pathogens such as Candida albicans (27). This regulon may play an important role in controlling the expression of glycosylphosphatidylinositol-anchored adhesins in these fungi as well. Some glycosylphosphatidylinositol-anchored adhesins in the Candida species control biofilm formation and serve as virulence factors (6, 72, 73). Therefore, orthologs of inositol regulon components may play a role in virulence in these fungal pathogens.
This work was supported by NIH grants GM35010, GM40266, and F32GM020565.
Supplemental material for this article may be found at http://ec.asm.org/. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»