This Article Abstract Full Text (PDF) Supplemental material Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Bergman, N. H. Articles by Hanna, P. C. Search for Related Content PubMed PubMed Citation Articles by Bergman, N. H. Articles by Hanna, P. C.

Transcriptional Profiling of the Bacillus anthracis Life Cycle In Vitro and an Implied Model for Regulation of Spore Formation

Department of Microbiology and Immunology,¹ Bioinformatics Program, University of Michigan Medical School, Ann Arbor, Michigan 48109²

Received 19 May 2006/ Accepted 26 June 2006

	ABSTRACT

Top Abstract Introduction Materials and Methods Results and Discussion References

 
The life cycle of Bacillus anthracis includes both vegetative and endospore morphologies which alternate based on nutrient availability, and there is considerable evidence indicating that the ability of this organism to cause anthrax depends on its ability to progress through this life cycle in a regulated manner. Here we report the use of a custom B. anthracis GeneChip in defining the gene expression patterns that occur throughout the entire life cycle in vitro. Nearly 5,000 genes were expressed in five distinct waves of transcription as the bacteria progressed from germination through sporulation, and we identified a specific set of functions represented within each wave. We also used these data to define the temporal expression of the spore proteome, and in doing so we have demonstrated that much of the spore's protein content is not synthesized de novo during sporulation but rather is packaged from preexisting stocks. We explored several potential mechanisms by which the cell could control which proteins are packaged into the developing spore, and our analyses were most consistent with a model in which B. anthracis regulates the composition of the spore proteome based on protein stability. This study is by far the most comprehensive survey yet of the B. anthracis life cycle and serves as a useful resource in defining the growth-phase-dependent expression patterns of each gene. Additionally, the data and accompanying bioinformatics analyses suggest a model for sporulation that has broad implications for B. anthracis biology and offer new possibilities for microbial forensics and detection.

	INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

 
The members of the Bacillus genus are spore-forming bacteria that switch between two morphologies—the dormant endospore and the vegetative cell—depending on whether the local environment will facilitate growth. In spore form, the bacterium is resistant to a variety of physical extremes, including heat, desiccation, UV and -irradiation, and oxidation, and the ability to switch between this cell type and the rapidly dividing vegetative form provides the bacilli with a highly effective strategy for persistence in the environment (28). Given the different functions of each, it is unsurprising that the two morphologies are microscopically and biochemically distinct, and the processes by which one form converts to the other, termed germination and sporulation, are known to be highly complex (10, 30, 35).

In several Bacillus species, the two forms of the bacterium play very different roles in the establishment and progression of disease. The most prominent example of this is Bacillus anthracis, the causative agent of anthrax; in this case, the spore morphology allows the bacterium to evade host defenses and establish infection, and the disease state is then perpetuated by growth of (and toxin/capsule production by) vegetative cells within the body (9). Because the two morphologies each play a distinct role, the transitions between germination and sporulation represent key phases in the pathogenesis of this organism, and it is clear that the regulated progression of this cycle is critical for successful establishment of disease and subsequent survival of the organism after death of the host (11, 22). Given this, a complete understanding of B. anthracis virulence hinges on a detailed definition of the complete Bacillus life cycle.

For a number of reasons, many studies have examined the Bacillus life cycle using the common model organism Bacillus subtilis (37). However, the sequencing of the B. subtilis and B. anthracis genomes has revealed that the two species are genetically very different, with more than 1,000 genes in the B. anthracis genome with no clear B. subtilis homolog (23, 32). Thus, despite the gross morphological similarities between the two, it seems likely that there are subtle differences that could help explain the unique ability of B. anthracis to cause disease. Because of this, the life cycle of B. anthracis itself is receiving increased interest, and the spore and vegetative morphologies as well as the processes by which they interconvert have been examined in recent studies (3, 6, 11-13, 26, 41). These reports have provided detailed proteomic characterizations of both cell types and identified a number of key loci involved in both germination and sporulation, but significant gaps remain in our understanding of the changes in gene expression that occur throughout the cycle.

Given the clear importance of the timing and progression of the B. anthracis life cycle in the pathogenesis of this organism, we sought to define the patterns of gene expression that occur throughout the complete cycle. DNA microarrays provide an efficient means of gaining insights into changes occurring in gene expression over time, but previous studies using microarrays to study B. anthracis expression generally focused on only a limited portion of the life cycle or were hindered by the fact that the arrays were based on incompletely annotated genome sequence (4, 26). Here we report the design and construction of an improved B. anthracis microarray and its use in examining gene expression trends through the entire life cycle in vitro, something that has not been done previously for any Bacillus species. We identified and defined five different waves of gene expression that occur during this cycle, each with a distinct functional profile. Additionally, we performed statistical analyses comparing these waves to recently generated proteomic data, and we have gained broad insights into the process of sporulation that have allowed us to formulate a new model for how this process occurs.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

 
Bacterial cultures. All work described in this report was done using the Sterne 34F2 strain (pXO1⁺ pXO2^–) of Bacillus anthracis. Initial spore stocks were prepared as follows: a single B. anthracis colony was used to inoculate brain heart infusion medium containing 5% glycerol. This culture was grown overnight at 37°C, and 1 ml was used to inoculate 500 ml of modified G medium as described in reference 8. Growth was measured by spectrophotometry at 600 nm. Progress through sporulation was monitored microscopically by scoring for the presence of phase-bright spores and also by measuring the percentage of cells in a culture that were capable of surviving an extended heat treatment (65°C for 30 min). Spores were purified as described previously (26), and the resulting stock was >99% pure when assayed by heat sensitivity assays and contained no appreciable vegetative debris.

Cultures for RNA extraction were begun by adding 10⁸ spores to 500 ml of modified G medium (8) in a 2-liter Erlenmeyer flask. The culture was grown at 37°C with shaking at 300 rpm, and 20-ml samples were removed for RNA extraction at the appropriate time point. Heat sensitivity assays were done by dividing a culture into two equal parts, incubating them at either 22°C or 65°C for 30 min, and then growing the surviving cells on brain heart infusion agar plate and comparing the CFU counts from heated and unheated samples.

RNA extraction. Bacteria were harvested by vacuum filtration at the appropriate time point and incubated in boiling lysis buffer (2% sodium dodecyl sulfate, 16 mM EDTA [pH 8.0], 20 mM NaCl) for 3 min. Following lysis, the mixture was extracted successively in phenol (65°C, twice), phenol (22°C), a 25:24:1 mix of phenol:CHCl₃:isoamyl alcohol, and finally 24:1 CHCl₃:isoamyl alcohol. RNA was then precipitated by addition of 2.5 volumes of 100% ethanol and incubation at –20°C. Pellets were washed with 70% ethanol and resuspended in 200 µl H₂O. The resulting RNA was further purified using an RNeasy kit (QIAGEN), and concentrations were measured by UV spectrophotometry. RNA quality was assessed by measuring the ratio of absorbance at 260 and 280 nm, as well as by visualization on an Agilent 2100 bioanalyzer or denaturing agarose gels.

Microarray design. The B. anthracis microarray described in this study was developed based on the Affymetrix GeneChip platform. These arrays contain 25-mer probes corresponding to each of the 5,815 open reading frames in the B. anthracis Ames Ancestor sequence (GB accession numbers NC_007530, NC_007322, and NC_007322, corresponding to the chromosome, pXO1, and pXO2, respectively), at a density of 16 probes per gene. Further details, as well as probe sequences, are available upon request.

Microarray sample processing and data collection. RNA samples were reverse transcribed, and the corresponding cDNA samples were purified, fragmented, and labeled according to Affymetrix recommended protocols (available at http://www.affymetrix.com/support/downloads/manuals/expression_s3_manual.pdf) at the UM Comprehensive Cancer Center Microarray Core Facility. Hybridization to the custom B. anthracis GeneChips, as well as scanning of the arrays, was also done according to standard Affymetrix protocols. At this point, several quality control steps were done in order to ensure that the raw data were of sufficient quality to proceed. First, the distributions of perfect match probe intensities for each chip were compared, since the robust multichip average procedure used for normalization and background correction is based on the assumption that these distributions are very similar. Once this assumption was verified, a plot of average probe intensity versus position within a gene was generated for each sample. This plot shows whether there is a systematic skew within a given data set toward probes that lie near the end of each gene, which would imply an abnormal amount of RNA degradation or a problem with the reverse transcription step. Once we had verified that all the samples showed a similar 5'-3' profile, we used the robust multichip average method to subtract background, normalize the data, and compute a single probe set summary for each gene (2, 20, 21). Principal-components analysis verified that biological replicates were very similar to each other and formed relatively tight clusters (data not shown).

Finally, in order to verify the accuracy of the data obtained from the newly developed GeneChip, we compared the data presented in this study to data recently generated in our laboratory in experiments examining the growth phase-dependent regulation of seven genes scattered throughout the B. anthracis genome. In every case, the microarray data were consistent with the trends observed by either reporter gene expression or quantitative reverse transcription-PCR (data not shown).

Data analysis. Statistical analysis and clustering of microarray data was done using the TM4 suite of programs (http://www.tm4.org/) (34). Several different methods were used for clustering, and we found empirically that the five waves of expression were visually apparent regardless of the method (hierarchical, K-median, self-organizing map [SOM]) used. Prior to using divisive methods like K-median and SOM in which the user supplies the overall number of clusters/nodes at the outset, we performed a figure-of-merit calculation to gauge the fit of the data to various numbers of clusters (40). This calculation confirmed that further divisions beyond five yielded only marginal improvements to how the data fit into the cluster framework and suggested that our visual observation that there were five major clusters was correct (data not shown). The SOM algorithm was used to generate figures (e.g., see Fig. 2) because it allowed us to order genes within a given group based on how stringently they were regulated.

View larger version (50K): [in this window] [in a new window]   FIG. 2. Cluster analysis of gene expression patterns occurring during the Bacillus anthracis life cycle in vitro. The expression levels for a given gene are shown relative to that gene's median across the entire time course. Green indicates up-regulation, red indicates down-regulation, and color intensity is proportional to the level of regulation, with the most intense shades of green and red corresponding to a difference in expression level of at least eightfold. Tick marks above the cluster diagram indicate the first replicate corresponding to times of 1, 2, 3 h, etc., up to h after inoculation.

The custom B. anthracis microarrays are available from Affymetrix with permission from the developers (for research purposes); further information can be obtained by contacting the authors.

Microarray data accession number. All microarray data described in this study are freely available from the NIAID Administrative Resource for Biodefense Proteomics Research Programs (http://www.proteomicsresource.org) or within the ArrayExpress database (http://www.ebi.ac.uk/arrayexpress), accession number E-MEXP-788.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

 
Timing of the B. anthracis life cycle in vitro. On a microscopic level, the B. anthracis life cycle is relatively well defined. Upon exposure to a sufficient concentration of germinant(s), the dormant endospore begins the process of germination and transforms into a rapidly growing vegetative cell. The bacterium then continues to grow quickly until nutrient depletion occurs, at which point the cell begins the process of sporulation. The cycle concludes with the release of a new endospore, which persists in the environment until nutrient concentrations rise again to levels that will support further growth. Our initial experiments showed that in a common laboratory medium, the bacteria progressed in a highly synchronous manner through an entire life cycle in a period of roughly 8 h (Fig. 1). In order to verify that the cycle was complete and that the cells had finished sporulation and formed functional endospores, we tested samples taken at each time point for heat sensitivity, since it is well known that Bacillus spores are able to easily survive heat treatments in which vegetative cells are efficiently killed (27). Our initial inoculum was microscopically verified to be >98% phase-bright spores (data not shown), and we observed that these cells were virtually all heat resistant (>95% survival after incubation at 65°C for 30 min). Fifteen minutes after inoculation, we observed that the cells had lost essentially all heat resistance (<5% survival), and the survival rates remained below 5% until 450 min postinoculation, at which point the culture had again reached >95% survival (phase-bright spores within mother cells were first visible roughly 30 min before this point). This finding showed that the culture was, under laboratory conditions, progressing through an entire life cycle in the 8-h time frame and also that the cells within the culture were highly synchronized and amenable to analysis by microarray.

View larger version (9K): [in this window] [in a new window]   FIG. 1. Growth of Bacillus anthracis in vitro. Bacterial growth is shown measured by optical density at 600 nm (OD600). Filled squares indicate those time points at which less than 5% of the culture was able to survive a 30-min incubation at 65°C. Open squares indicate a survival rate under the same conditions of >95%. Growth curves were done in triplicate; a representative sample is shown.

Interestingly, only 37 of the 4,956 genes were encoded on pXO1; 175 would be expected by chance if the 4,956 genes were uniformly distributed across pXO1 and the chromosome, so the observed number represents a highly significant underrepresentation (Fisher's exact test, P < 1e–122). This may be explained by the fact that within this plasmid is a disproportionately large number of genes associated with pathogenesis, and these genes may be regulated in response to environmental cues that are never present in vitro. Consistent with this idea, we observed that many of the major virulence factors (e.g., lef and pagA) were identified by the ANOVA filter as unchanging throughout the life cycle, though the baseline expression levels of all of these genes were well above background. It should also be noted that inaccurate annotation of the pXO1 sequence may contribute to the observed underrepresentation in pXO1 genes as well; a large fraction of the genes on this plasmid are hypothetical, and if the total gene count is artificially high because a number of these genes are in fact unexpressed pseudogenes, it could partially explain the observed underrepresentation.

Five waves of gene expression during the B. anthracis life cycle. In order to visualize the data and identify large-scale patterns occurring during the B. anthracis life cycle, we converted the data to relative expression values, such that the expression of each gene is measured relative to its median expression level throughout the entire life cycle. The data were then clustered using a self-organizing map algorithm (see Materials and Methods for details). It was visually apparent that there were five major clusters (Fig. 2), and the results obtained from the clustering procedure were consistent with this; the SOM algorithm requires that an overall cluster number be specified a priori, and increasing this number simply results in subclusters that are highly similar (data not shown). We had previously observed large waves of gene expression during the B. anthracis life cycle (26), but in that case the groups of genes were much smaller because we were only examining 2,090 genes (as opposed to 4,956 in the current study). On a gene-by-gene basis, the data were extremely consistent between the two studies, though the waves in the current work were much larger because of the huge increase in the number of genes being considered, and they were also better defined because the extended time frame of the current study (8 h, covering a complete life cycle, versus 5 h, covering late log phase and sporulation) allowed us to more precisely differentiate between similar expression profiles, especially in considering genes that were up-regulated in the early stages of outgrowth.

The five clusters varied in size, timing, and the length of time for which the genes within each was up-regulated, and complete lists of the genes within each can be found in Tables S1 to S5 in the supplemental material. In order to identify the functional families that were overrepresented in each wave and thus the pathways and functions up-regulated and presumably important to the cell during each phase, we performed a pathway analysis using the EASE algorithm (19). Briefly, this algorithm identifies Gene Ontology (GO) (http://www.geneontology.org/) and/or TIGRFAM (http://www.tigr.org/TIGRFAMs/) terms that are overrepresented at or beyond a specified statistical threshold within a particular subset of genes. The results of this analysis are shown in Table 1.

The genes in the third wave were up-regulated only after the cells had been growing for roughly 2 h, and these include loci associated with both rapid growth (e.g., energy metabolism, electron transport, and glycogen biosynthesis) and the various pathways involved in responding to an increasingly toxic environment (nitrate assimilation and phosphate starvation). Consistent with this, we observed that while these genes were up-regulated the cellular growth rate slowed gradually until approximately 2 h later, when a plateau was reached and the fourth wave of expression began. Wave 4 remained up-regulated through the end of the experimental time course, and it contains a number of genes whose homologs in other species have been previously associated with sporulation, as well as loci involved in responding to oxidative stress (e.g., catalase and superoxide dismutase). Finally, roughly 30 min after the beginning of wave 4, the fifth group of genes was up-regulated. This group is relatively similar to wave 4 in that in contains a large number of sporulation-associated loci, but it is interesting to note that the genes in wave 5 show an expression profile that suggests that their mRNAs are also up-regulated in the cell during the earliest stages of germination and outgrowth. It appears either that these mRNAs are expressed during germination (perhaps because of residual transcriptional activators that have survived within the spore) or that the mRNAs themselves are left over from sporulation and were perhaps protected from degradation by ribosomes or other RNA-binding proteins.

Although our own work (both this study and reference 27) has shown that mRNA can be isolated from newly formed spores, the possibility that mRNA might persist for long periods of time within the dormant spore has not been addressed. In order to explore this issue, we performed an RNA isolation procedure on a large number of spores (10¹¹) that had been isolated and purified as described previously (27) and then stored at room temperature for 1 month. The total RNA yield from this procedure was more than 1,000-fold lower than the yields obtained from a similar number of vegetative cells (data not shown), and this finding was reproducible across three independent spore stocks. As noted above, several studies have indicated that the RNA isolation procedure used here is fully capable of purifying mRNA from within newly formed (but nonetheless heat resistant and thus mature) spores (27, 29), and there were no indications of incomplete lysis/solubilization of the spores during the procedure, so we attribute the extremely low yield to a relative lack of mRNA in the dormant spore. It follows, then, that the expression of the genes in wave 5 during the early stages of outgrowth is more likely due to residual transcription factors within the spore that direct de novo mRNA synthesis, and this is consistent with our previous finding that there are at least 20 known or putative transcriptional regulators present within the spore proteome (27).

Although the timing of expression differs between the fourth and fifth waves, the two groups are quite similar in terms of gene content, especially compared to the first three waves of expression. These two final waves show a much higher proportion of hypothetical genes (47.9% and 46.2%, respectively, compared to a range of 26.8% to 36.3% in the first three waves; Table 1), and the genes within them are generally more stringently regulated (there are >100 genes in each of waves four and five for which maximum and minimum expression levels differ by >150-fold, compared to 8 such genes in waves 1 to 3 combined). Presumably, both of these findings reflect the fact that sporulation is a highly regulated process and requires a large number of genes without characterized homologs in other model organisms, and they highlight the utility of the microarray data in linking more than 2,000 hypothetical genes in the B. anthracis genome to a particular growth phase. In general, very little is known about these genes, and their association with a phase of the life cycle allows them to be connected to a set of likely biological functions for the first time.

Although the precise regulatory mechanisms directing the timing and duration of the different waves of expression remain unclear, previous studies of Bacillus subtilis and other gram-positive species have shown that this type of high-level control of gene expression is often achieved by the use of different RNA polymerase sigma factors, each of which is specifically responsible for a particular program of expression. There are 25 putative sigma factors that have been identified within the B. anthracis genome, and 19 showed growth-phase-dependent expression patterns in the current study, with at least 2 of these genes expressed within each of the five waves (see Table S6 in the supplemental material). Some of these proteins (8) have clear homologs in Bacillus subtilis, and these typically showed expression profiles consistent with those reported previously for their better-characterized homologs (e.g., the sporulation-specific sigma factors ^E, ^F, ^G, and ^K are expressed with other sporulation-associated genes in the later waves). Interestingly, the B. anthracis ^A homolog also followed this trend and was expressed in wave 2 during rapid growth. Previous work using cDNA arrays had suggested that this locus might have been incorrectly annotated, since its expression was found to peak during late sporulation. We believe this result was due to an error in the construction of those microarrays, and our current data match what was expected from a number of studies with B. subtilis (18).

Apart from these eight, the majority of the putative sigma factors in B. anthracis are completely uncharacterized and have no clear counterpart in B. subtilis. The microarray-based identification of the temporal expression pattern and associated functional trends thus provides a useful starting point in beginning to link these genes to putative homologs in other species. Perhaps most significantly, we note that the regulatory sequences recognized by each of the 25 sigma factors remain undefined, and since the array data show which genes are closely coexpressed with each sigma factor, they also serve as a valuable basis for predicting each sigma factor's regulon.

Expression of the spore proteome. Recently, a proteomic survey identified more than 750 protein components of the B. anthracis endospore, and accompanying experiments examining mRNA expression during sporulation suggested that some of these proteins were actually synthesized during early steps of the life cycle rather than during sporulation (26). Since the array data described in the present study comprised an entire life cycle, they allowed us to examine the temporal expression of the spore proteome in a much more comprehensive way than had been previously possible. Of the 765 proteins identified in the spore proteome, 690 corresponded to genes that were identified in the microarray analysis as showing some growth-phase-dependent regulation (i.e., they passed the ANOVA filter described above), and these loci were found in all five waves of expression. Interestingly, we found that the genes whose products make up the spore proteome were overrepresented in waves 1 and 2 to a high statistical significance (calculated by Fisher's exact test) (Table 2) and underrepresented in waves 4 and 5. This profile was somewhat surprising, since it indicates that more than half of the spore proteome genes are actually up-regulated during the early phases of the B. anthracis life cycle, and it suggests that during sporulation the majority of the spore proteome is packaged from preexisting stocks rather than synthesized de novo.

Potential mechanisms for regulating the composition of the spore proteome. A mechanism for the type of "directed packaging" that is implied by the microarray data could take several different forms. Perhaps the most obvious possibility is that the cell could use an active transport mechanism to specifically target certain proteins to the forespore during sporulation. However, there does not appear to be any recognizable signal or "tag" in the spore proteome protein sequences that could mediate this transport, nor is there any evidence for the existence of the transport machinery itself. Consequently, although this scenario is difficult to rule out completely, it seems somewhat unlikely.

Another possibility is that the proteins that are synthesized early in the life cycle but also become part of the spore may be simply expressed at higher levels; later, a simple random packaging of all available proteins would lead to the spore contents being biased toward the more highly expressed subset. Because it is difficult to compare the absolute expression values of different genes via microarray analysis (due to differences in probe hybridization efficiencies), we examined this possibility by taking advantage of the fact that relative expression levels can be estimated from the codon usage within each gene. Briefly, although the genetic code is redundant, with most amino acids encoded by two or more codons, a given organism typically prefers one synonym over the other(s). The degree to which a gene adheres to these genome-wide preferences can be measured as a CAI, and several groups have shown that the CAI is highly correlated with the overall expression level, with the more highly expressed genes in a given bacterial genome tending to use the preferred codons (7, 15, 36). With this in mind, we compared the CAI values from the genes that produce the spore proteome to the values from the rest of the genome to see if there was a significant difference. We found that although there was a slight difference, it was not statistically significant (Mann-Whitney test, P = 0.347), and when we considered the genes within each wave separately we found similar results (for instance, comparing the subset of wave 2 genes that produce spore proteome proteins to the rest of the wave 2 genes yielded a statistically insignificant difference; Mann-Whitney test, P = 0.249). Together, these results suggest that the spore proteome's corresponding mRNAs are not expressed at a higher average level than the rest of the genes in the genome and that this probably is not the primary method employed by the cell in regulating the composition of the spore proteome.

If mRNA expression levels are generally similar, we reasoned that the cell could have allowed individual genes within the genome to evolve such that the spore proteome sequences are more stable than the average gene; this would allow the cell to synthesize a variety of proteins at similar levels early in its life cycle but also to have a significantly biased pool from which to package the spore proteome when sporulation begins several hours later. Experimental measurement of protein stability on a genomewide scale is not currently feasible, but several studies have shown that both the physical stability (i.e., susceptibility to physical or chemical denaturation) and biochemical stability (i.e., susceptibility to degradation by cellular enzymes) of a given protein can be accurately predicted (1, 16, 17, 38). With this in mind, we sought to test whether the proteins in the spore proteome would be expected to be significantly more stable (either physically or biochemically) than the average protein encoded within the B. anthracis genome.

Although the precise biophysical reasons for this finding are still being elucidated, it is known from previous studies that the dipeptide distributions in physically stable proteins differ significantly from those occurring in physically unstable proteins (17, 33). The dipeptide frequencies found in these two populations have been defined extensively (33), and an algorithm for predicting the stability of a given protein based on these frequencies is in wide use (http://www.expasy.org/tools/protparam.html) (14). We used this algorithm to generate predicted "instability index" values for every protein-coding gene in the B. anthracis genome (see Table S7 in the supplemental material), and we found that on average the spore proteome sequences were predicted to be more stable than the rest of the genome (Mann-Whitney test, P = 0.0042). None of the other genomic subgroups tested (e.g., the five expression waves) showed a significant deviation from the rest of the genome. We noted, however, that when the spore proteome was broken into five subgroups based on the waves of expression, the most significant differences between the spore proteome subset and the rest of the genes occurred in waves 1 and 2 (P = 0.0006). In short, the spore proteome components are generally predicted to be more physically stable than the average B. anthracis protein, and those expressed earliest in the life cycle were most likely to have a higher predicted physical stability.

Given the apparent differences in physical stability between the spore proteome and the rest of the B. anthracis genome, we adopted a parallel approach to determine whether there might be a similar difference in predicted biochemical stability. The majority of intracellular protein degradation is known to depend on N-terminal sequences, and the sequence elements that determine a protein's half-life within a bacterial cell have been precisely defined in previous studies (1, 38). These sequence rules (collectively termed the "N-end rule") have been built into an algorithm that predicts the half-life of a given protein in vivo (http://www.expasy.org/tools/protparam.html [14]), and we used this tool to calculate the predicted in vivo half-life for each protein encoded in the B. anthracis genome. We then compared the predicted values for the spore proteome proteins to those from the rest of the genome (note that prior to calculating the in vivo half-life, the N-terminal methionine was removed from all protein sequences; studies have shown that the majority of bacterial proteins are processed in this way [5, 25], and since the precise substrate specificities of the eight putative aminopeptidases in the B. anthracis genome are not known, we took the unbiased approach of treating all proteins as potential substrates). We found that the proteins in the spore proteome had an average predicted half-life that was significantly (nearly an hour) longer than the average predicted half-life for the rest of the B. anthracis proteins (Mann-Whitney, P = 7.9e^–7). As in the case of physical stability, none of the five waves of expression showed a statistically significant deviation from the rest of the genome.

Together, these bioinformatics analyses suggested that the proteins that make up the spore are both physically and biochemically more stable than the average protein encoded by the B. anthracis genome. We note that although it remains conceivable that these differences reflect a bias in the proteomic analysis that prevented the identification of less-stable proteins, we believe this possibility is unlikely for several reasons. First, the multidimensional protein identification technique for proteomic analysis of the spore does not depend on proteins surviving intact through an extensive purification protocol; early in the protocol, the proteins are intentionally degraded for an efficient chromatographic separation (24, 26). Second, several of the proteins positively identified as part of the spore proteome are generally thought to be present at a level of less than 10 copies per cell (12, 26, 39), and at least one of these (GerY_A) is predicted to have an extremely short (2- to 3-min) half-life in the cell, so the method is clearly able to identify scarce proteins. Based on these facts, it seems that an easily degraded protein, even if present only at low levels, would have an excellent chance of being detected by the multidimensional protein identification technique protocol. Thus, we conclude that the bias toward physical and biochemical stability predicted here is not an experimental artifact but rather is a real property of the B. anthracis spore proteome.

A model for sporulation. In total, the microarray and subsequent bioinformatics analyses presented in this study suggest a model for sporulation in which the spore structure is synthesized during sporulation, while the remainder of the spore proteome is made up largely of preexisting components that are in many cases synthesized much earlier in the B. anthracis life cycle. Additionally, it appears that the components of the spore proteome are on average more stable than the other proteins encoded by the genome, and this may allow the cell to synthesize a large set of proteins early in its life cycle and yet consistently package only a specific subset of those proteins into the developing spore several hours later. A stability-based selection for the spore proteome seems particularly intuitive given the spore's well-documented ability to survive for long periods of time in harsh environments and also because it provides a regulatory mechanism that is relatively passive at a time when the cell is highly stressed; during sporulation, packaging a random selection of available proteins will result in the desired biased pool simply because less-stable proteins will have been degraded.

Perhaps the most interesting implication of this model is that it suggests that the spore proteome is likely to be at least partially reflective of the growth environment encountered by the bacteria during earlier phases of growth. Although many of the same genes are induced during rapid growth on a variety of media, the amount to which individual genes are up- or down-regulated changes significantly depending on the specific biochemical environment (N. H. Bergman and P. C. Hanna, unpublished data), and this presumably would result in a corresponding change in the levels of at least some of the proteins that become part of the spore proteome. Further experiments will be necessary to further explore this possibility and to test whether these changes in proteomic composition might be useful from a forensic standpoint or whether they might result in changes in pathogenic potential (e.g., a difference in virulence between spores produced from cells grown in vitro and spores produced from cells grown in a mammalian host).

Conclusions. The microarray data presented in this study allowed us to construct what is by far the most comprehensive description of the transcriptional patterns throughout the entire B. anthracis life cycle in vitro and to identify the functions and pathways utilized by the cell during different phases of this cycle. Our findings identified potential roles for a large number (>2,000) of hypothetical genes and highlighted a number of genes that were up-regulated during phases that are relevant to anthrax, raising the possibility that they may be useful targets for therapeutics. Collectively, the data serve as a useful comparative resource for further studies of B. anthracis gene expression under a variety of other conditions (including growth inside the mammalian host), and a number of such experiments are currently ongoing in our laboratory. Finally, we also note that the custom B. anthracis microarrays developed for this study will no doubt be useful to others in the scientific community, and these arrays are now available from Affymetrix.

	ACKNOWLEDGMENTS

 
We thank Jeremy Bergman, Helena Prieto, Emily Chen, Daniel Cociorva, John Yates III, and members of the Hanna lab for valuable discussions. We also thank James MacDonald of the UM Comprehensive Cancer Center Microarray Core Facility for advice regarding normalization procedures for microarray data.

This work was supported by HHS contract N266200400059C-N01-AI-40059.

	FOOTNOTES

 
* Corresponding author. Mailing address: University of Michigan Medical School, Bioinformatics Program and Department of Microbiology and Immunology, 6605H Medical Sciences Bldg II, 1150 W. Medical Center Dr., Ann Arbor, MI 48109-0620. Phone: (734) 615-2154. Fax: (734) 764-3562. E-mail: niber{at}umich.edu.

Supplemental material for this article may be found at http://jb.asm.org/.

	REFERENCES

Top Abstract Introduction Materials and Methods Results and Discussion References

 

1. Bachmair, A., D. Finley, and A. Varshavsky. 1986. In vivo half-life of a protein is a function of its amino-terminal residue. Science 234:179-186.[Abstract/Free Full Text]
2. Bolstad, B. M., R. A. Irizarry, M. Astrand, and T. P. Speed. 2003. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19:185-193.[Abstract/Free Full Text]
3. Bongiorni, C., R. Stoessel, D. Shoemaker, and M. Perego. 2006. Rap phosphatase of virulence plasmid pXO1 inhibits Bacillus anthracis sporulation. J. Bacteriol. 188:487-498.[Abstract/Free Full Text]
4. Bourgogne, A., M. Drysdale, S. G. Hilsenbeck, S. N. Peterson, and T. M. Koehler. 2003. Global effects of virulence gene regulators in a Bacillus anthracis strain with both virulence plasmids. Infect. Immun. 71:2736-2743.[Abstract/Free Full Text]
5. Bradshaw, R. A., W. W. Brickey, and K. W. Walker. 1998. N-Terminal processing: the methionine aminopeptidases and N()-acetyl transferase families. Trends Biochem. Sci. 23:263-267.[CrossRef][Medline]
6. Brunsing, R. L., C. La Clair, S. Tang, C. Chiang, L. E. Hancock, M. Perego, and J. A. Hoch. 2005. Characterization of sporulation histidine kinases of Bacillus anthracis. J. Bacteriol. 187:6972-6981.[Abstract/Free Full Text]
7. Carbone, A., A. Zinovyev, and F. Kepes. 2003. Codon adaptation index as a measure of dominating codon bias. Bioinformatics 19:2005-2015.[Abstract/Free Full Text]
8. Dixon, T., A. Fahd, T. Koehler, J. Swanson, and P. Hanna. 2000. Early events in anthrax pathogenesis: intracellular survival of B. anthracis and its escape from RAW264.7 macrophages. Cell. Microb. 2:453-463.
9. Dixon, T., M. Meselson, J. Guillemin, and P. Hanna. 1999. Medical progress: anthrax. N. Engl. J. Med. 341:815-826.[Free Full Text]
10. Dworkin, J., and R. Losick. 2005. Developmental commitment in a bacterium. Cell 121:401-409.[CrossRef][Medline]
11. Fisher, N. 2006. On the role of cellular differentiation in the establishment of anthrax. University of Michigan, Ann. Arbor, Mich.
12. Fisher, N., and P. Hanna. 2005. Characterization of Bacillus anthracis germinant receptors in vitro. J. Bacteriol. 187:8055-8062.[Abstract/Free Full Text]
13. Francis, A. W., C. E. Ruggiero, A. T. Koppisch, J. Dong, J. Song, T. Brettin, and S. Iyer. 2005. Proteomic analysis of Bacillus anthracis Sterne vegetative cells. Biochim. Biophys. Acta 1748:191-200.
14. Gasteiger, E., C. Hoogland, A. Gattiker, S. Duvaud, M. R. Wilkins, R. D. Appel, and A. Bairoch. 2005. Protein identification and analysis tools on the ExPASy server, p. 571-607. In J. M. Walker (ed.), The proteomics protocols handbook. Humana Press, Totowa, N.J.
15. Goetz, R. M., and A. Fuglsang. 2005. Correlation of codon bias measures with mRNA levels: analysis of transcriptome data from Escherichia coli. Biochem. Biophys. Res. Commun. 327:4-7.[CrossRef][Medline]
16. Gonda, D. K., A. Bachmair, I. Wunning, J. W. Tobias, W. S. Lane, and A. Varshavsky. 1989. Universality and structure of the N-end rule. J. Biol. Chem. 264:16700-16712.[Abstract/Free Full Text]
17. Guruprasad, K., B. V. Reddy, and M. W. Pandit. 1990. Correlation between stability of a protein and its dipeptide composition: a novel approach for predicting in vivo stability of a protein from its primary sequence. Protein Eng. 4:155-161.[Abstract/Free Full Text]
18. Horsburgh, M. J., P. D. Thackray, and A. Moir. 2001. Transcriptional responses during outgrowth of Bacillus subtilis endospores. Microbiology 147:2933-2941.[Abstract/Free Full Text]
19. Hosack, D. A., G. Dennis, Jr., B. T. Sherman, H. C. Lane, and R. A. Lempicki. 2003. Identifying biological themes within lists of genes with EASE. Genome Biol. 4:R70.[CrossRef][Medline]
20. Irizarry, R. A., B. M. Bolstad, F. Collin, L. M. Cope, B. Hobbs, and T. P. Speed. 2003. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res. 31:e15.[Abstract/Free Full Text]
21. Irizarry, R. A., B. Hobbs, F. Collin, Y. D. Beazer-Barclay, K. J. Antonellis, U. Scherf, and T. P. Speed. 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4:249-264.[Abstract]
22. Kang, T. J., M. J. Fenton, M. A. Weiner, S. Hibbs, S. Basu, L. Baillie, and A. S. Cross. 2005. Murine macrophages kill the vegetative form of Bacillus anthracis. Infect. Immun. 73:7495-7501.[Abstract/Free Full Text]
23. Kunst, F., N. Ogasawara, I. Moszer, A. M. Albertini, G. Alloni, V. Azevedo, M. G. Bertero, P. Bessieres, A. Bolotin, S. Borchert, R. Borriss, L. Boursier, A. Brans, M. Braun, S. C. Brignell, S. Bron, S. Brouillet, C. V. Bruschi, B. Caldwell, V. Capuano, N. M. Carter, S. K. Choi, J. J. Codani, I. F. Connerton, A. Danchin, et al. 1997. The complete genome sequence of the gram-positive bacterium Bacillus subtilis. Nature 390:249-256.[CrossRef][Medline]
24. Link, A. J., J. Eng, D. M. Schieltz, E. Carmack, G. J. Mize, D. R. Morris, B. M. Garvik, and J. R. Yates III. 1999. Direct analysis of protein complexes using mass spectrometry. Nat. Biotechnol. 17:676-682.[CrossRef][Medline]
25. Link, A. J., K. Robison, and G. M. Church. 1997. Comparing the predicted and observed properties of proteins encoded in the genome of Escherichia coli K-12. Electrophoresis 18:1259-1313.[CrossRef][Medline]
26. Liu, H., N. H. Bergman, B. Thomason, S. Shallom, A. Hazen, J. Crossno, D. A. Rasko, J. Ravel, T. D. Read, S. N. Peterson, J. Yates III, and P. C. Hanna. 2004. Formation and composition of the Bacillus anthracis endospore. J. Bacteriol. 186:164-178.[Abstract/Free Full Text]
27. Montville, T. J., R. Dengrove, T. De Siano, M. Bonnet, and D. W. Schaffner. 2005. Thermal resistance of spores from virulent strains of Bacillus anthracis and potential surrogates. J. Food Prot. 68:2362-2366.[Medline]
28. Nicholson, W. L., N. Munakata, G. Horneck, H. J. Melosh, and P. Setlow. 2000. Resistance of Bacillus endospores to extreme terrestrial and extraterrestrial environments. Microbiol. Mol. Biol. Rev. 64:548-572.[Abstract/Free Full Text]
29. Passalacqua, K. D., N. H. Bergman, A. Herring-Palmer, and P. Hanna. 2006. The superoxide dismutases of Bacillus anthracis do not cooperatively protect against endogenous superoxide stress. J. Bacteriol. 188:3837-3848.[Abstract/Free Full Text]
30. Piggot, P. J., and D. W. Hilbert. 2004. Sporulation of Bacillus subtilis. Curr. Opin. Microbiol. 7:579-586.[CrossRef][Medline]
31. Price, M., K. Huang, E. Alm, and A. Arkin. 2005. A novel method for accurate operon predictions in all sequenced prokaryotes. Nucleic Acids Res. 33:880-892.[Abstract/Free Full Text]
32. Read, T. D., S. N. Peterson, N. Tourasse, L. W. Baillie, I. T. Paulsen, K. E. Nelson, H. Tettelin, D. E. Fouts, J. A. Eisen, S. R. Gill, E. Holtzapple, O. A. Okstad, E. Helgason, J. Rilstone, M. Wu, J. F. Kolonay, M. J. Beanan, R. J. Dodson, L. M. Brinkac, M. Gwinn, R. T. DeBoy, R. Madpu, S. C. Daugherty, A. S. Durkin, D. H. Haft, W. C. Nelson, J. D. Peterson, M. Pop, H. M. Khouri, D. Radune, J. L. Benton, Y. Mahamoud, L. Jiang, I. R. Hance, J. F. Weidman, K. J. Berry, R. D. Plaut, A. M. Wolf, K. L. Watkins, W. C. Nierman, A. Hazen, R. Cline, C. Redmond, J. E. Thwaite, O. White, S. L. Salzberg, B. Thomason, A. M. Friedlander, T. M. Koehler, P. C. Hanna, A. B. Kolsto, and C. M. Fraser. 2003. The genome sequence of Bacillus anthracis Ames and comparison to closely-related bacteria. Nature 423: 81-86.[CrossRef][Medline]
33. Reddy, B. V. 1996. Structural distribution of dipeptides that are identified to be determinants of intracellular protein stability. J. Biomol. Struct. Dyn. 14:201-210.[Medline]
34. Saeed, A. I., V. Sharov, J. White, J. Li, W. Liang, N. Bhagabati, J. Braisted, M. Klapa, T. Currier, M. Thiagarajan, A. Sturn, M. Snuffin, A. Rezantsev, D. Popov, A. Ryltsov, E. Kostukovich, I. Borisovsky, Z. Liu, A. Vinsavich, V. Trush, and J. Quackenbush. 2003. TM4: a free, open-source system for microarray data management and analysis. BioTechniques 34:374-378.[Medline]
35. Setlow, P. 2003. Spore germination. Curr. Opin. Microbiol. 6:550-556.[CrossRef][Medline]
36. Sharp, P. M., and W. H. Li. 1987. The codon adaptation index—a measure of directional synonymous codon usage bias, and its potential applications. Nucleic Acids Res. 15:1281-1295.[Abstract/Free Full Text]
37. Sonenshein, A. L., J. A. Hoch, and R. Losick. 2001. Bacillus subtilis and its closest relative: from genes to cells. American Society for Microbiology, Washington, D.C.
38. Tobias, J. W., T. E. Shrader, G. Rocap, and A. Varshavsky. 1991. The N-end rule in bacteria. Science 254:1374-1377.[Abstract/Free Full Text]
39. Woese, C. R., J. C. Vary, and H. O. Halvorson. 1968. A kinetic model for bacterial spore germination. Proc. Natl. Acad. Sci. USA 59:869-875.[Free Full Text]
40. Yeung, K. Y., D. R. Haynor, and W. L. Ruzzo. 2001. Validating clustering for gene expression data. Bioinformatics 17:309-318.[Abstract/Free Full Text]
41. Zaman, M. S., A. Goyal, G. P. Dubey, P. K. Gupta, H. Chandra, T. K. Das, M. Ganguli, and Y. Singh. 2005. Imaging and analysis of Bacillus anthracis spore germination. Microsc. Res. Tech. 66:307-311.[CrossRef][Medline]

This article has been cited by other articles:

• (2008). Bacillus anthracis-derived nitric oxide is essential for pathogen virulence and survival in macrophages. Proc. Natl. Acad. Sci. USA 105: 1009-1013  
• Zhang, C., Crasta, O., Cammer, S., Will, R., Kenyon, R., Sullivan, D., Yu, Q., Sun, W., Jha, R., Liu, D., Xue, T., Zhang, Y., Moore, M., McGarvey, P., Huang, H., Chen, Y., Zhang, J., Mazumder, R., Wu, C., Sobral, B. (2008). An emerging cyberinfrastructure for biodefense pathogen and pathogen host data. Nucleic Acids Res 36: D884-D891 [Abstract] [Full Text]  
• Thompson, B. M., Waller, L. N., Fox, K. F., Fox, A., Stewart, G. C. (2007). The BclB Glycoprotein of Bacillus anthracis Is Involved in Exosporium Integrity. J. Bacteriol. 189: 6704-6713 [Abstract] [Full Text]  
• Bergman, N. H., Anderson, E. C., Swenson, E. E., Janes, B. K., Fisher, N., Niemeyer, M. M., Miyoshi, A. D., Hanna, P. C. (2007). Transcriptional Profiling of Bacillus anthracis during Infection of Host Macrophages. Infect. Immun. 75: 3434-3444 [Abstract] [Full Text]  
• Passalacqua, K. D., Bergman, N. H., Lee, J. Y., Sherman, D. H., Hanna, P. C. (2007). The Global Transcriptional Responses of Bacillus anthracis Sterne (34F2) and a {Delta}sodA1 Mutant to Paraquat Reveal Metal Ion Homeostasis Imbalances during Endogenous Superoxide Stress. J. Bacteriol. 189: 3996-4013 [Abstract] [Full Text]  

This Article Abstract Full Text (PDF) Supplemental material Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Bergman, N. H. Articles by Hanna, P. C. Search for Related Content PubMed PubMed Citation Articles by Bergman, N. H. Articles by Hanna, P. C.