HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Supplementary data Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Steil, L. Articles by Völker, U. Search for Related Content PubMed PubMed Citation Articles by Steil, L. Articles by Völker, U. Agricola Articles by Steil, L. Articles by Völker, U.

Genome-wide analysis of temporally regulated and compartment-specific gene expression in sporulating cells of Bacillus subtilis

¹ Philipps-University Marburg, Department of Biology, Laboratory for Microbiology, D-35032 Marburg, Germany
² Max-Planck-Institute for Terrestrial Microbiology, D-35043 Marburg, Germany
³ Ernst-Moritz-Arndt-University, Medical School, Laboratory for Functional Genomics, Walther-Rathenau-Str. 49A, D-17487 Greifswald, Germany
⁴ Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Av. da República, Apartado 127, 2781-901 Oeiras Codex, Portugal

Correspondence
Uwe Völker
voelker{at}uni-greifswald.de

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
Temporal and compartment-specific control of gene expression during sporulation in Bacillus subtilis is governed by a cascade of four RNA polymerase subunits. ^F in the prespore and ^E in the mother cell control early stages of development, and are replaced at later stages by ^G and ^K, respectively. Ultimately, a comprehensive description of the molecular mechanisms underlying spore morphogenesis requires the knowledge of all the intervening genes and their assignment to specific regulons. Here, in an extension of earlier work, DNA macroarrays have been used, and members of the four compartment-specific sporulation regulons have been identified. Genes were identified and grouped based on: i) their temporal expression profile and ii) the use of mutants for each of the four sigma factors and a bofA allele, which allows ^K activation in the absence of ^G. As a further test, artificial production of active alleles of the sigma factors in non-sporulating cells was employed. A total of 439 genes were found, including previously characterized genes whose transcription is induced during sporulation: 55 in the ^F regulon, 154 ^E-governed genes, 113 ^G-dependent genes, and 132 genes under ^K control. The results strengthen the view that the activities of ^F, ^E, ^G and ^K are largely compartmentalized, both temporally as well as spatially, and that the major vegetative sigma factor (^A) is active throughout sporulation. The results provide a dynamic picture of the changes in the overall pattern of gene expression in the two compartments of the sporulating cell, and offer insight into the roles of the prespore and the mother cell at different times of spore morphogenesis.

The descriptions of all regulon members as well as complete sets of raw and normalized data are available as supplementary data with the online version of this paper at http://mic.sgmjournals.org.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
During the early stages of endospore development in the bacterium Bacillus subtilis, the rod-shaped cell is partitioned into a small prespore and a much larger mother cell. Each cell type receives a copy of the bacterial chromosome, and deploys specific but interdependent genetic programmes controlled by the successive appearance of the ^F, ^E, ^G and ^K subunits of RNA polymerase (reviewed by Errington, 2003; Hilbert & Piggot, 2004; Stragier & Losick, 1996). Entry into sporulation is induced by nutrient starvation, and is mainly controlled through phosphorylation of the Spo0A response regulator (Burbulys et al., 1991; Phillips & Strauch, 2002; Sonenshein, 2000). Spo0AP controls the expression of a large regulon which includes the genes encoding the first compartment-specific regulators ^F and ^E, as well as genes required for the asymmetric partitioning of the cell (Errington, 2003; Hilbert & Piggot, 2004; Stragier & Losick, 1996). Following asymmetric division, the prespore is first engulfed by the mother cell and then encased in several protective structures, including a primordial germ cell wall, which will become the cell wall of the germinating cell, a cortex layer essential for heat resistance, consisting of a modified form of peptidoglycan, and a double-layered protein coat which confers protection against harsh chemicals and lysozyme, and which influences germination (Driks, 1999; Henriques & Moran, 2000). Lysis of the mother cell at the end of the sporulation process releases the mature spore into the environment.

Activation of ^F occurs in the prespore immediately after the polar division of the sporulating cell. ^F controls gene expression during the early stages of prespore development, and directs transcription of the gene encoding ^G, which replaces it during later post-engulfment stages of prespore development. Conversely, ^E drives transcription of early mother-cell genes, including the structural gene for ^K, which replaces ^E following engulfment of the prespore by the mother cell. Activation of ^F is coupled to the formation of the polar septum, and ^F activity is required for the mother-cell-specific activation of ^E (Errington, 2003; Hilbert & Piggot, 2004; Stragier & Losick, 1996). The transcriptional activity of ^E is then required for the activation of ^G in the prespore, which is somehow coupled to the completion of the engulfment process (Partridge & Errington, 1993; Sun et al., 2000). Lastly, ^G triggers a signalling pathway that activates ^K in the mother cell (Losick & Pero, 1981; Piggot & Losick, 2002). These cell–cell signalling mechanisms ensure that the forespore and mother-cell-specific programmes of gene expression are kept in pace and in register with the course of morphogenesis and are essential to ensure that the differentiation process takes place with high fidelity (Errington, 2003; Hilbert & Piggot, 2004; Stragier & Losick, 1996).

Sporulation involves the expression of a large number of genes. Many loci have been identified following chemical mutagenesis of a sporulation-proficient strain, based on the property that on sporulation plates colonies of a wild-type (Spo⁺) but not those of many asporogenous (Spo^–) or oligosporogenous mutants produce a dark-brown pigment (Piggot & Coote, 1976). Sporulation loci have also been identified by transposon mutagenesis (Sandman et al., 1987) or by the use of integrational plasmids for gene disruption, by reverse genetics, as encoding components of the spore or some of its structures (e.g. Donovan et al., 1987; Kuwana et al., 2002; Lai et al., 2003), or by expression-based screens designed to find members of specific sporulation regulons (e.g. Beall et al., 1993). Prior to sequencing of the B. subtilis genome, about 100 genes were listed as being involved in sporulation, and about 60 of those were known to be dependent on the activity of one of the four compartment-specific sigma factors (Stragier & Losick, 1996). The availability of the B. subtilis genome sequence (Kunst et al., 1997) has made possible the use of DNA arrays to study the profile of gene expression during sporulation at a genome-wide level. Three recent studies have employed DNA arrays and expression profiling to study sporulation. Fawcett et al. (2000) have characterized the transcription profile of the early to middle stages of sporulation induced by nutrient exhaustion in Difco sporulation medium (DSM). These authors have compared transcripts present during growth, at the onset of sporulation, and 2 h after the initiation of sporulation of a wild-type strain, with transcripts present in mutants for spo0A and sigF. The transcription of 66 genes was found to be dependent on both Spo0A and ^F, including several genes known to be under the control of ^F or ^E. The use of hidden Markov models trained to find known promoter elements allowed the assignment of 11 new genes to the ^F regulon, and the assignment of 22 to control by ^E (Fawcett et al., 2000). Two studies have provided information on the composition of the ^E regulon when sporulation is induced by growth and resuspension in a poorer synthetic medium. Eichenberger et al. (2003) have reported transcriptional profiling and bioinformatics data to support their assignment of 253 genes to the ^E regulon, including 181 new genes. Disruption of 12 of the newly identified genes produced a sporulation phenotype (Eichenberger et al., 2003). Feucht et al. (2003) have found a total of 171 ^E-dependent transcripts, 101 of which were previously unknown, and of these, mutations in about 10 diminished the efficiency of sporulation.

In the present study, we wanted to extend expression-profiling studies to the late-prespore and mother-cell-specific regulons (under ^G and ^K control, respectively), and simultaneously provide an overall picture of the changes in the pattern of gene expression, over time, when sporulation is induced by growth followed by resuspension in a defined medium. We made use of DNA macroarrays to identify genes governed by ^F, ^E, ^G and ^K. Most of the previously characterized sporulation genes were found and assigned to the correct regulon. We report on the identification of a total of 439 sporulation genes, including 185 new genes. Our results strongly support the view that the different sporulation regulons are largely differentiated both temporally and spatially, with little overlap between consecutive prespore- or mother-cell-specific regulons (Li & Piggot, 2001). The results also provide insight into the specific contributions of the prespore and mother-cell types to spore development and spore properties.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
Bacterial strains, media and growth conditions.
Escherichia coli DH5 was used for routine cloning experiments. The B. subtilis strains used in this work are listed in Table 1. The transcriptional profiling experiments were performed with strain JH642 (BGSC1A96; a kind gift from J. Hoch, La Jolla, USA) and isogenic derivatives with defects in particular sporulation-specific sigma factors. For these experiments, bacteria were routinely grown with vigorous agitation in growth medium until an OD at 540 nm of 0·75, at which time they were collected by centrifugation at room temperature for 15 min at 6000 g. Immediately afterwards, cell pellets were resuspended in twice the original volume of resuspension medium to induce sporulation (Boschwitz & Yudkin, 1983; Nicholson & Setlow, 1990). For RNA preparation, samples were collected at 30, 90, 150, 210, 270 and 330 min after resuspension. Cells were harvested by mixing 20 ml culture with an equal volume of frozen killing buffer (20 mM NaN₃, 20 mM Tris/HCl, pH 7·5, 5 mM MgCl₂) and subsequent centrifugation for 10 min at 10 000 g at 4 °C. After washing with 1 ml killing buffer, the cell pellets were stored at –80 °C until further use for total RNA preparation.

Cell lysis, RNA isolation and dot-blot/Northern analysis.
RNA was isolated according to the acid phenol method described by Völker et al. (1994). Aliquots of the total RNA prepared for the DNA macroarray experiments were used for the dot-blot and Northern analysis of the expression profiles of the spoIIR, spoIID, spoIIID, sspE and gerE genes. Digoxigenin- (DIG) labelled anti-sense RNA probes were generated by in vitro transcription using a StripEZ-kit (Ambion) and gene-specific PCR products as templates. The following PCR products were generated for the production of antisense RNA probes: a 544 bp spoIIR fragment using primers spoIIR-for (5'-CTGGCAAACAGCGATAGTG-3') and spoIIR-rev (5'-TAATACGACTCACTATAGGGAGGTCGGAAATCCATTCG-3'), an 891 bp spoIID fragment using primers spoIID-for (5'-CACTATCCGTACTATGTGC-3') and spoIID-rev (5'-TAATACGACTCACTATAGGGAGGCCAAATCCTCTCGTC-3'), a 307 bp spoIIID fragment using primers spoIIID-for (5'-GTGGTGTGCACGATTACATC-3') and spoIIID-rev (5'-TAATACGACTCACTATAGGGAGGCGATTGCTGAACAGGCTC-3'), a 335 bp sspE fragment using primers sspE-for (5'-GAGAAAGCTTTACGATCACCTGCACATTC-3') and sspE-rev (5'-TAATACGACTCACTATAGGGAGGAGTGATTAGCTGTTTTGTTG-3') and a 186 bp gerE fragment using primers gerE-for (5'-TCGAAGCCGTCGCTAACG-3') and gerE-rev (5'-TAATACGACTCACTATAGGGAGGCTCTAGCTCACCCATTC-3'). Because, in each of the PCR reactions with chromosomal DNA from strain JH642, the reverse (rev) primers carried the sequence of the T7 promoter, the PCR fragments could be used for in vitro RNA synthesis with T7 RNA polymerase (Ambion). This yielded hybridization probes internal to the genes. Denaturing RNA electrophoresis on agarose gels, RNA transfer by diffusion onto a nylon membrane (NY13N; Schleicher & Schuell), hybridization to gene-specific probes and signal detection were performed as described by Scharf et al. (1998).

Preparation of labelled cDNA, array hybridization and DNA macroarray regeneration.
Prior to the cDNA labelling, the overall integrity of the total RNA preparation was verified by Northern-blot analysis with digoxigenin-labelled probes directed against known members of the four compartment-specific sporulation regulons. The DNA macroarray analyses employed commercially available Panorama B. subtilis DNA macroarrays from Sigma Genosys which carry duplicate spots of PCR products representing 4107 B. subtilis genes, as well as the corresponding commercial primer mix (Sigma Genosys), which consists of 4107 specific oligonucleotide primers complementary to the 3' ends of all mRNA-encoding B. subtilis genes. cDNA synthesis, probe hybridization and washing of the filters were performed as described by Steil et al. (2003). Arrays were exposed to storage phosphor screens (Molecular Dynamics) for two to four days, and subsequently scanned with a Storm 840/860 phosphorimager (Molecular Dynamics) at a resolution of 50 µm and a colour depth of 16 bit. Bound cDNA was stripped off the DNA-macroarray membranes by three washing cycles involving a short (1 min) washing step with 250 ml boiling buffer (5 mM sodium phosphate, pH 7·5, 0·1 % SDS) and an incubation in 250 ml fresh buffer at 95 °C for 20 min.

Data analysis.
Data analysis followed a three-step procedure. First, the ArrayVision software Version 6.1 (Imaging Research) was used for the quantification of the hybridization signals after direct import of the phosphorimager files. The analysis yielded the artifact-removed volumes (ARVol) and background values, calculated from the median of a line surrounding each group of eight spots on the array. These data were then used in a second step in Microsoft Excel to calculate, for every spot on the array, a quality score that reflected the ratio between the signal intensity and the background intensity (further details available at http://www.medizin.uni-greifswald.de/funkgenom/supplemental_material). This quality score was utilized to identify hybridization signals close to the detection limit, thereby avoiding artificially high induction ratios for those genes. Data normalization and data analysis were done in a third step with GeneSpring (Version 5.02) (Silicon Genetics). Gene expression for a particular comparison of conditions was considered to be changed when three criteria were fulfilled: i) expression of the gene had to exceed the background signal level by a threshold determined as described (further details available at http://www.medizin.uni-greifswald.de/funkgenom/supplemental_material); ii) changes in expression of the gene had to be statistically significant, as defined in a statistical group comparison of the values of the selected conditions with a parametric test (ANOVA) and a Benjamini and Hochberg False Discovery Rate correction with a P value cut-off of 0·05, as defined in the GeneSpring software package; iii) the change in expression had to exceed a factor of three. Calculations of ratios were done with means of the parallel spots on the filters.

Web access.
The complete dataset for all growth conditions investigated is available online (http://www.medizin.uni-greifswald.de/funkgenom/supplemental_material).

Construction of gfp transcriptional fusions inserted into the amyE locus.
The gfp gene was amplified using primers gfpD (5'-CCCAAGCTTGGGGGATCCGGGAAAAGGTGGTGA-3') and gfpR (5'-GGCGAATTCTTATTTGTATAGTTCATCCATGC-3'), and plasmid pEA18 (a gift from Alan Grossman) as the template. The 744 bp PCR fragment was digested with HindIII and EcoRI and ligated to pMLK83 (Karow & Piggot, 1995) which had been digested with the same enzymes, yielding pMS157. To create transcriptional fusions of the yuiC, yhaX, yhcV and yxeE promoter regions to gfp, the following PCR products were first generated: a 468 bp fragment encompassing the yuiC promoter using primers yuiC-53D (5'-CATGCTGCTCGAGAATGTCTTGGATTATGGC-3') and yuiC-521R (5'-GTTCCTGGATCCCATTTTGACAAGTCCTTCGC-3'); a 516 bp fragment containing the yhaX promoter using primers yhaX-67D (5'-GGAAAACTCGAGATAATAACATTGAAAGCGCC-3') and yhaX-583R (5'-GGCATCAAGCTTTAGCGATTTCGC-3'); a 485 bp yhcV fragment with primers yhcV-30D (5'-AAATAACTCGAGTTATTACCAAGGAAC-3') and yhcV-515R (5'-CAACGGGGATCCGCCCCGACGTTATGC-3'); and a 409 bp fragment carrying the yxeE promoter using primers yxeE-42D (5'-GACCCTCGAGTGCTTTGGGAAATCACC-3') and yxeE-451R (5'-GTAAGGATCCTGCTGAGGCAGCTGAGGGC-3'). The PCR fragments carrying the yuiC, yhcV and yxeE promoter regions were digested with XhoI and BamHI and ligated to SalI- and BamHI-digested pMS157, to produce pMS174, pMS173 and pMS172, respectively (Table 1). The fragment encompassing the yhaX promoter region was digested with XhoI and HindIII and ligated to pMS157 which had been digested with SalI and HindIII, yielding pMS175 (Table 1). Samples of ScaI-digested pM172, pM173, pM174 and pM175 were used to transform the parental strain MB24, as well as a panel of congenic strains mutant for ^F, ^E, ^G and ^K, selecting for kanamycin resistance. AmyE^– transformants, the result of a double cross-over at the amyE locus, were kept for further analysis (see Table 1).

Light microscopy and image processing.
Samples (0·5 ml) of cells in resuspension medium were collected throughout sporulation, and resuspended in the same volume of PBS supplemented with 10 µg ml^–1 4',6'-diamidino-2-phenylindole (DAPI). Microscope slides were prepared as described previously (Serrano et al., 2004). Images were acquired using a cooled charge couple device (Cooke) on a multi-wavelength wide-field three-dimensional microscopy system (63x/1·4 OIL Plan Apochromat objective, Zeiss 100M, Intelligent Imaging Innovation). Standard filters for fluorescein isothiocyanate (for green fluorescence protein, GFP) and DAPI were used.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
Experimental strategy
Our transcriptional profiling approach employed two different strategies to assign genes to the control of ^F, ^E, ^G or ^K. First, the temporal expression pattern after initiation of sporulation by resuspension in a defined medium (Boschwitz & Yudkin, 1983) was recorded for well-characterized members of all four regulons in the Bacillus subtilis wild-type strain JH642. Second, congenic mutants of JH642 defective in the production of each of the four sigma factors or the BofA protein were utilized to assign genes to a particular regulon, even if the temporal expression patterns showed partial overlap. Membership of one of the four sporulation regulons was also substantiated by locating the genes to operons and searching for conserved sequences recognized by the respective sigma factor in front of the potential transcription units. As a further test, we also made use of fusions of the genes encoding active forms of the four compartment-specific sigma factors to the IPTG-inducible P_spac promoter to force their synthesis in vegetatively growing cells, in the absence of all other sporulation proteins.

Prior to the transcriptional profiling experiments, bona fide members of each regulon were selected, and their temporal expression was analysed by RNA dot-blot experiments in the B. subtilis wild-type strain JH642, in mutants lacking one of the four sporulation-specific sigma factors or the regulatory protein BofA, and in strains allowing artificial expression of active forms of the sigma factors in vegetative cells. This allowed us to determine the time points for the best discrimination among the four regulons (Fig. 1).

View larger version (35K): [in this window] [in a new window]   Fig. 1. Expression pattern of canonical members of the four sporulation-specific sigma regulons in wild-type cells, various sporulation mutants, and upon IPTG induction of sigma factor production in vegetatively growing cells. The B. subtilis wild-type strain JH642 and its isogenic mutant strains MO1073 (sigF), MO512 (sigE), MO1074 (sigG), MO1027 (sigK) and Marb30 (sigG bofA) were grown in growth medium and sporulation was induced by collection of the cells by centrifugation and resuspension in twice the volume of resuspension medium after they had reached an OD540 of 0·75. For the experiments involving production of active alleles of the sporulation sigma factors in vegetatively growing cells, strains were grown in LB. During exponential growth, these cultures were divided, one half was induced with 1 mM IPTG and after 30 min cultivation with or without IPTG samples were collected for preparation of RNA. Total RNA was prepared after IPTG induction or at the time points indicated after resuspension, as described in Methods. The RNA samples were transferred to nylon membranes by dot-blotting and the RNA blots were then hybridized with specific, digoxigenin-labelled probes internal to the structural genes of spoIIR, spoIID, spoIIID, sspE and gerE. After fluorescence detection with a STORM860 fluorimager and quantification with the Imagequant program (Amersham Biosciences), induction ratios were calculated by dividing the signal intensities of each particular time point by the intensity measured with RNA from samples harvested immediately after resuspension (wild-type and sporulation mutants) or by the value of the culture incubated without IPTG (IPTG induction of active sigma factor alleles). (a) Early prespore and mother-cell-specific gene-expression patterns (SigF and SigE regulon); (b) late prespore and mother-cell-specific gene-expression patterns (SigG and SigK regulon). The dots above each bar graph indicate the samples which were chosen for the discrimination of the different regulons in the analysis of the DNA-array data. After normalization of the DNA arrays, data ratios were determined by dividing the signal intensities of samples marked with filled dots by the intensities of samples from the same section (wild-type temporal expression, mutant analysis and IPTG induction) marked with open dots.

Differentiation of the compartment-specific sporulation regulons
For their inclusion in the early prespore- or mother-cell-specific ^F or ^E regulons, genes had to fulfil the threefold induction criterion at both 90 and 150 min, compared to 30 min, after initiation of sporulation in the wild-type (Figs 1a and 2a), as well as for the comparison of the expression of the wild-type and the sigF mutant 90 min after initiation of sporulation (Figs 1a and 2b). We were able to discriminate between the ^F and ^E regulons on the basis of their expression pattern in the sigE mutant. While genes assigned to the ^E regulon displayed at least threefold higher expression at 90 min after initiation of sporulation in the wild-type compared to the sigE mutant, members of the ^F regulon did not show this induction (Figs 1a and 2b) but were instead required to display at least threefold higher expression at 90 min in the sigE mutant compared to the sigF mutant (Fig. 1a).

View larger version (68K): [in this window] [in a new window]   Fig. 2. Temporal and spatial compartmentalization of the four sporulation-specific sigma regulons. The log2 values of the ratios between the normalized signal strength of the individual conditions depicted in the axis descriptions are plotted. The intensity ratios of all 4107 genes represented on the Panorama B. subtilis DNA macroarrays from Sigma Genosys are indicated as small, light-grey shaded diamonds in the background. These data are not filtered to remove spurious induction ratios, and thus light-grey background symbols displaying seemingly strong regulation are statistically not significant. The following groups of genes are emphasized with specific symbols: (a) differentiation of prespore-specific sporulation regulons (expression ratio 90 to 30 min, y axis; expression ratio 210 to 30 min, x axis); black crosses, F-regulon members, previously characterized by non-chip approaches; white crosses, G-regulon members, previously characterized by non-chip approaches; large open circles, F-regulon members identified in this study; large filled circles, G-regulon members identified in this study; (b) differentiation of early prespore and mother-cell-specific gene expression (expression ratio of wild-type versus sigF mutant at 90 min, y axis; expression ratio of wild-type versus sigE mutant at 90 min, x axis); black crosses, F-regulon members, previously characterized by non-chip approaches; white crosses, E-regulon members, previously characterized by non-chip approaches; large open circles, F-regulon members identified in this study; large filled diamonds, E-regulon members identified in this study; (c) analysis of mother-cell-specific gene expression (expression ratio 90 to 30 min, y axis; expression ratio 270 to 150 min, x axis); white crosses, E-regulon members, previously characterized by non-chip approaches; black crosses, K-regulon members, previously characterized by non-chip approaches; large black diamonds, E-regulon members identified in this study; large open diamonds, K-regulon members identified in this study; (d) differentiation of late prespore and mother-cell-specific gene expression (expression ratio of wild-type 270 min to 30 min, y axis; expression ratio of sigG bofA mutant versus sigK mutant at 270 min, x axis); white crosses, G-regulon members, previously characterized by non-chip approaches; black crosses, K-regulon members, previously characterized by non-chip approaches; large open diamonds, K-regulon members identified in this study; large filled circles, G-regulon members identified in this study. Genes belonging to a particular sporulation regulon and displaying significant induction in one of the conditions displayed are represented by open or filled large symbols with superimposed crosses. WT, Wild-type.

In addition to recording the temporal expression in the wild-type strain and the effect of individual sigma-factor knock-outs on the expression profile, we also tested induction of genes following expression of the four sporulation-specific sigma factors during exponential growth. This series of experiments employed strains in which expression of the sigma-factor-encoding genes was governed by the IPTG-inducible promoter P_SPAC. Activation of the mother-cell-specific sigma factors ^E and ^K during sporulation requires their proteolytic processing from inactive pre-proteins, and thus this study made use of specific sigE and sigK alleles that are active without processing (Oke & Losick, 1993; Stragier et al., 1988). Before our stringent selection criteria were applied to the data derived from the artificial induction of the sigma factors, we utilized the group of sporulation genes previously characterized by biochemical and genetic approaches as test set. This analysis yielded quite different reassignment rates for the four different sporulation regulons. Whereas 76 % and 75 % of the 21 ^F- and 57 ^K-dependent genes discovered by non-chip approaches displayed at least threefold higher expression upon induction of the respective sigma factor allele during growth, only 20 % and 43 % of the 59 ^E-dependent and 46 ^G-dependent genes were reassigned using this procedure. Furthermore, the induction ratios observed after artificial induction during growth were in general much lower than those observed in sporulating cells. These lower induction ratios might be a reflection either of the leakiness of the P_SPAC promoter or of the only partial activity of the sigma factor alleles used. In the case of sigE, the limited activity of a sigE copy from which pro-amino acid sequences have been removed has been observed before (Eichenberger et al., 2003). Failure to induce the whole set of sporulation genes during growth is not unexpected because many sporulation genes might require other regulatory inputs that are not provided during growth. As a consequence of this analysis of a well-defined screening set of known sporulation genes, we decided to utilize the data of the artificial induction experiments merely as supportive information (column P_SPACIPTG/co in Supplementary Tables S1–S4, available online as supplementary data with the online version of this paper at http://mic.sgmjournals.org), but not as discriminative information, in order to avoid large numbers of false-negative candidates.

The fact that we were able to find conditions that permitted the separation of most of the genes in consecutive regulons expressed in different compartments of the sporulating cell, as for ^F and ^E (Fig. 2b), ^E and ^G (Figs 1a, b and 2) or ^G and ^K (Fig. 2d), or in the same compartment, as for ^F and ^G in the prespore (Fig. 2a), or ^E and ^K in the mother cell (Fig. 2c), supports the conclusion of an earlier study which indicated that sporulation-specific gene expression is largely compartmentalized, both temporally and spatially (Li & Piggot, 2001). Nevertheless, as illustrated by the group of genes lying almost at the diagonals of the comparisons depicted in Fig. 2a, c, some genes displayed an ambiguous behaviour. This group of genes could result from limitations of our experimental analysis, or could reflect a biological property of the system, for example, that the expression of some genes is governed by more than one factor (see also below).

Temporal differentiation within the ^F and ^G regulons
In addition to the key role of the four compartment-specific sigma factors in establishing the overall pattern of temporal and compartment-specific gene expression during sporulation, gene expression within each regulon can also be classified in several epistatic classes, in part because of the influence of ancillary transcription factors that may function as repressors or activators (Errington, 2003; Hilbert & Piggot, 2004). For example, some ^F-dependent genes are expressed soon after asymmetric division of the sporangial cell and the concomitant prespore-specific activation of ^F (Karow et al., 1995; Londono-Vallejo & Stragier, 1995; Londono-Vallejo et al., 1997; Wu & Errington, 2003), while expression of the dacF or spoIIIG genes, for example, appears delayed relative to the first wave of ^F-directed genes (Karow et al., 1995; Partridge & Errington, 1993; Schuch & Piggot, 1994; see below). Therefore, we wanted to test whether or not, based on our data, we could discriminate between groups of genes with a common expression profile within each of the main regulons. Based upon their temporal pattern of expression, we were able to differentiate the 55 genes of the ^F regulon into two classes (Fig. 3a and Supplementary Table S1). Class 1 comprises 36 genes whose expression peaked at around 90 or 150 min and decreased thereafter, suggesting that their transcription is switched off. Class 2 included 19 genes whose main period of expression was centred 270 min after the onset of sporulation. The spoIIR, spoIIQ, lonB and rsfA genes were assigned to class 1, in agreement with experimental data (Karow et al., 1995; Londono-Vallejo & Stragier, 1995; Londono-Vallejo et al., 1997; Serrano et al., 2001; Wu & Errington, 2000), whereas dacF, gpr, spIVB, sspN and tlp (Cabrera-Hernandez et al., 1999; Schuch & Piggot, 1994; Sussman & Setlow, 1991) were assigned to class 2 (Supplementary Table S1). ^F and ^G have overlapping promoter specificities, and therefore some ^F-dependent genes are only recognized by ^F, whereas other genes are recognized by both ^F and ^G (Amaya et al., 2001; Haldenwang, 1995; Helmann & Moran, 2002). The class of late ^F-dependent genes, which presumably coincides with class 2 as defined here (Fig. 3 and Supplementary Table S1), appears to include genes under the dual control of ^F and ^G, such as gpr and dacF (Schuch & Piggot, 1994; Sussman & Setlow, 1991). Most of the ^F-dependent genes that deviate significantly from the y axis towards the diagonal in Fig. 2a, which depicts the separation of the ^F and ^G regulons, also cluster in class 2, which has a temporal definition, and in that sense defines a partial overlap between the ^F and ^G regulons.

View larger version (35K): [in this window] [in a new window]   Fig. 3. Characterization of subclasses of the F (a), E (b), G (c) and K (d) regulons. K-means clustering of the expression patterns of genes assigned to the four sporulation regulons was utilized to detect subclasses in each regulon displaying particular expression patterns. This clustering allowed between two to four classes, employed standard correlation, and a maximum of 100 iterations. Two distinct subclasses were recognized for the prespore-specific regulons (F and G), and mother-cell-specific regulons (E and K) could be divided into three subclasses. At least two different reasons seemed to contribute to the extremely high induction ratios determined for some of the K-regulon members. Firstly, expression of some K-dependent genes is extremely low in the early phase of sporulation. Secondly, at the latest time point analysed (330 min), gene expression of the cells is mainly confined to a rather small group of genes, thus yielding extremely high levels for those individual genes.

It should be noted that the mechanism by which expression of certain ^F genes is delayed relative to the first class of ^F-dependent genes is not clear (Errington, 2003; Hilbert & Piggot, 2004). Expression of the late ^F-governed gene spoIIIG appears to require the activity of ^E in the mother cell (Partridge & Errington, 1993). Presumably, ^E activity could lead to the activation of a prespore-specific factor required for spoIIIG transcription, or otherwise cause the inactivation or removal of a repressor. However, this putative signalling pathway has not been examined in detail, and it is not known whether other class 2 genes are also ^E-dependent. Recently, the ^F-dependent gene rsfA has been shown to be involved in the control of prespore-specific ^F-dependent gene expression (Wu & Errington, 2000). Disruption of rsfA had different effects on class 1 genes: it caused increased expression of spoIIR, but had no effect on the expression of spoIIQ or of rsfA itself (Wu & Errington, 2000). Evidently, rsfA does not appear to be the main regulatory factor in the temporal differentiation of the ^F regulon, or in specifying ^F only or dual ^F/^G control.

As for ^F, the 113 genes assigned to the ^G regulon could be divided into two classes. Class 1 groups 73 genes whose expression is increased at 150 min and tends to decrease at later time points, whereas class 2 includes 40 genes whose expression is induced by 210 min (Fig. 3c). Class 1 includes genes such as gerAA, gerAB and spoVT, previously characterized as ^G dependent (Bagyan et al., 1996; Feavers et al., 1990), as well as the spoIIIG gene (Sun et al., 1991). That spoIIIG was found in the ^G and not in the ^F regulon (Supplementary Tables S1 and S3) is consistent with the report that most spoIIIG transcription stems from an auto-catalytic loop in which, following its activation, ^G recognizes the promoter for its own gene (Sun et al., 1991). This positive feedback regulatory scheme is thought to result in a rapid increase in the cellular level of ^G triggered by its activation following completion of engulfment. Note however that some ^F-dependent transcription of spoIIIG does occur (Sun et al., 1991), albeit not to sufficient levels to pass our stringent criteria for its inclusion in the ^F regulon (see above). Class 2 includes, for example, sspA (Mason et al., 1988), most of the genes of the spoVA operon (Mouldover et al., 1994), and sspF, previously shown to be transcribed about one hour later than other genes in the ^G regulon (Panzer et al., 1989). There are two known transcriptional regulators within the ^G regulon, SpoVT and SplA (Bagyan et al., 1996; Fajardo-Cavazos & Nicholson, 2000). The role of the TRAP-like SplA protein may be limited to the modulation of the level of expression of the gene encoding the SplB spore photoproduct lyase (Fajardo-Cavazos & Nicholson, 2000), whereas the spoVT gene has been shown to encode an AbrB-like transcription factor with a more global role in the control of ^G-controlled gene expression (Bagyan et al., 1996). Expression of the spoVT gene itself, as well as that of spoIIIG and the gerA operon, is normally repressed in a spoVT-dependent manner, whereas expression of the spoVA operon and of sspA requires spoVT (Bagyan et al., 1996). Prolonged expression of several genes, including spoIIIG and gerA, was observed in cells of a spoVT mutant, suggesting that SpoVT may normally shut off transcription of these genes. These observations are in agreement with the overall expression profile of the genes grouped in class 1. Presumably, class 1 mostly includes genes whose expression is not critically dependent on SpoVT or that are repressed by SpoVT, whereas class 2 includes genes whose expression depends to various extents on SpoVT.

Temporal differentiation within the ^E and ^K regulons
A similar analysis of the 154 ^E-dependent genes allowed the differentiation of three distinct temporal classes (Fig. 3b and Supplementary Table S2). Class 1 includes 41 genes whose expression is rapidly induced and peaks at 90 min, to decrease thereafter. As for the first class of ^F-dependent genes, this suggests that transcription of this group of genes is switched off (see above). In contrast, the 93 genes in class 2 show a slower rate of induction, and prolonged expression, which peaks between 150 and 210 min. A third class includes 20 genes which are maximally induced around 210 min, and whose expression persists at later times in development. spoIID (Rong et al., 1986) and the SpoIIID-repressed spoIIIA operon (Illing & Errington, 1991) are both found in class 1. Expression of the spoIIIA operon occurs transiently, prior to the accumulation of SpoIIID to high cellular levels (Illing & Errington, 1991), which fits well with the overall pattern of class 1 genes, and suggests that other genes in this class may be subjected to repression. In contrast, the spoIIID gene itself and spoIVCB (encoding the N-terminal half of ^K), both of which are known to be SpoIIID dependent (Kunkel et al., 1989; Sato et al., 1994; Stevens & Errington, 1990), are found in class 2. Presumably, efficient expression of the class 2 genes requires the accumulation of SpoIIID above a certain threshold level, as was suggested for spoIIID and spoIVCB (Kunkel et al., 1989; Sato et al., 1994; Stevens & Errington, 1990). Class 3 includes the coat morphogenetic gene cotE, which is expressed from two tandem ^E-dependent promoters, one of which (P2) is additionally dependent on SpoIIID (Zheng & Losick, 1990), spoVJ, known to be expressed from tandem ^E- and ^K-dependent promoters (Foulger & Errington, 1991), and at least one gene, csk22, previously reported to be under ^K control (Henriques et al., 1997). The distinction between classes 1 and 2 may be attributable mostly to the effects of the regulatory protein SpoIIID upon ^E-dependent gene expression (Zheng & Losick, 1990). SpoIIID-independent genes are expressed early, and SpoIIID-dependent genes are expressed later, while the expression of some of the early genes is repressed or switched off (Halberg & Kroos, 1994; Illing & Errington, 1991; Kroos et al., 1989; Kunkel et al., 1989). However, class 3 may represent an overlap between the ^E and ^K regulons, either because genes in this class have multiple promoters utilized by ^E or ^K, or because atypical ^E-type promoters can also be recognized by ^K (Helmann & Moran, 2002). In any case, class 3 represents a partial overlap between the mother-cell-specific ^E and ^K regulons.

Additional regulators may contribute to the fine-tuning of gene expression within the ^E regulon. For example, Wu & Errington (2000) reported that the ^E-dependent gene ylbO, which other studies also placed in the ^E regulon (Eichenberger et al., 2003; Feucht et al., 2003; Supplementary Table S2), encodes a putative transcriptional regulator, highly similar to RsfA. YlbO may work together with ^E in the mother cell, in much the same way that RsfA regulates ^F-dependent gene expression in the prespore (Wu & Errington, 2000). Unfortunately, the effects of a ylbO mutation on mother-cell-specific gene expression have not yet been examined. Moreover, in addition to SpoIIID and, hypothetically, to YlbO, the expression of some ^E-dependent genes may be influenced by other factors. Several known or putative transcriptional regulators have been assigned to the ^E regulon: purR, encoding the repressor of the purine operons; birA, a biotin acetyl-CoA-carboxylase synthetase and transcriptional regulator (Eichenberger et al., 2003); yhgD (TetR/AcrR family) and ytzE (DeoR family) (Feucht et al., 2003). Note, however, that none of these four genes could be assigned to the ^E regulon under our experimental conditions (Fig. 4b and Supplementary Table S2). In yet another example, the expression of the mmg operon, which encodes proteins with similarity to fatty-acid-metabolizing enzymes, is under ^E control, but is subjected to catabolite repression in a ccpA-dependent manner (Bryan et al., 1996). It is not known whether CcpA regulates other mother-cell-specific genes.

View larger version (108K): [in this window] [in a new window]   Fig. 4. Genomic organization of the F (a), E (b), G (c) and K (d) regulons. Genes and operons known to be controlled by the four sporulation-specific sigma factors F, E, G and K from traditional genetic and biochemical approaches are displayed inside the circles. Those regulon members that could be reassigned to the corresponding regulon based on the expression pattern recorded in this study are indicated by black diamonds, and those that did not pass the selection criteria adopted in our study are indicated in grey. Newly identified members of the four sporulation regulons are displayed outside the circles. For the E regulon, the display also contains genes previously reported by Eichenberger et al. (2003): genes displayed outside without any additional label were discovered in this study only, genes that are underlined were reported in the studies of Eichenberger et al. (2003) only, and those marked with an additional asterisk were discovered in this study and one of the two previous publications (Eichenberger et al., 2003; Feucht et al., 2003). For the display of the E, G and K regulons, sections of the chromosome displaying a particularly high density of sporulation genes are enlarged to increase visibility. In the genomic display of the F regulon, the portion of the chromosome located in the prespore at the time of F activation is filled with grey colour.

As noted above for the ^E regulon, the expression of the ^K regulon may be subject to additional levels of control. Recently, the yjcC gene (renamed spoVIF; Supplementary Table S4) was found to be required for the formation of heat- and lysozyme-resistant spores, and it has been suggested that it could play a role in modulating the expression of other ^K-governed genes (Kuwana et al., 2003). Moreover, our analysis identifies a transcriptional regulator of the MarR family (ysmB), upstream of the gene (racE) encoding a glutamate racemase, both in class 2 (Fig. 4d and Supplementary Table S4). It is not known whether expression of the racE gene can be modulated in response to the availability of glutamate or some other factor via YsmB. However, these observations suggest that, even at a late stage in spore morphogenesis, the mother-cell-specific line of gene expression is responsive to environmental stimuli.

An overview of the four compartment-specific regulons
The ^F regulon.
Several regulatory functions can be attributed to ^F-directed gene expression. First, transcription of the spoIIR gene is required to signal ^E activation in the mother cell (Karow et al., 1995; Londono-Vallejo & Stragier, 1995), and transcription of bofC and spoIVB is required for proper signalling of ^K activation (Cutting et al., 1991b; Gomez & Cutting, 1996). ^F also drives expression of at least two genes involved in the control of its own activity, rsfA and lonB (Serrano et al., 2001; Wu & Errington, 2000), of the spoIIIG gene (Sun et al., 1989, 1991) and of the spoIIQ gene, which is required for efficient expression of spoIIIG (Londono-Vallejo et al., 1997; Sun et al., 2000). In addition to its regulatory role, spoIIQ is also involved in the engulfment process, although only under certain nutritional conditions (Sun et al., 2000). With the exception of the spoIIIG gene, which did not pass our selection criteria (see above), all these regulatory genes were found in the present study (Supplementary Table S1 and Fig. 4a). Several genes in the ^F regulon have functions in spore protection or spore germination: the katX-encoded catalase, for example, is implicated in spore protection against hydrogen peroxide (Bagyan et al., 1998); the mutTA gene encodes an antimutator 8-oxo-dGTPase (Ramirez et al., 2004); the sspN and tlp genes, which code for small acid-soluble spore proteins (SASP), act by shielding the prespore chromosome (Cabrera-Hernandez et al., 1999); gerD is required for efficient germination in response to L-alanine and to a mixture of glucose, fructose, L-asparagine and KCl (Kemp et al., 1991); and gpr encodes a protease involved in SASP protein degradation during spore germination (Sussman & Setlow, 1991). Interestingly, our analysis identified the gene (yyaC) for a second possible GPR-like protease in the ^F regulon (Fig. 4a and Supplementary Table S1), suggesting a scenario of partial redundancy. Another function of the ^F regulon may be to contribute to the morphogenesis of the spore protective layers. For example, the dacF gene encodes a penicillin-binding protein (PBP), with D-alanyl-D-alanine carboxypeptidase activity, which is involved in regulating the degree of cross-linking of the spore peptidoglycan (Popham et al., 1999). No obvious alteration in spore peptidoglycan structure was found for a dacF single insertional mutant (Wu et al., 1992), but our study identifies a second PBP-encoding gene (yrrR, in an operon with a gene of unknown function, yrrS) in the ^F regulon (Fig. 4a and Supplementary Table S1). It will be interesting to analyse a yrrR mutant, as well as a strain doubly mutant for dacF and yrrR. The finding of ripX, which encodes a site-specific integrase/recombinase involved in proper chromosome partitioning (Sciochetti et al., 1999), and yqhH, predicted to code for an SNF2-type helicase (Supplementary Table S1), suggests that their products may prepare the spore for the resumption of growth following germination and outgrowth. Seve