OUP user menu

Generation of multiple cell types in Bacillus subtilis

Daniel Lopez, Hera Vlamakis, Roberto Kolter
DOI: http://dx.doi.org/10.1111/j.1574-6976.2008.00148.x 152-163 First published online: 1 January 2009


Bacillus subtilis is a Gram-positive bacterium that is well known for its ability to differentiate into metabolically inactive spores that are highly resistant to environmental stresses. In fact, populations of genetically identical B. subtilis comprise numerous distinct cell types. In addition to spores, cells can become genetically competent, motile, produce extracellular matrix or degradative enzymes, or secrete toxins that allow them to cannibalize their neighbors. Many of the cell fates listed above appear to be mutually exclusive. In this review, we discuss how individual cells within a population control their gene expression to ensure that proper regulation of differentiation occurs. These different cell fates are regulated by an intricate network that relies primarily on the activity of three major transcriptional regulators: Spo0A, DegU, and ComK. While individual cells must choose distinct cell fates, the population as a whole exhibits a spectrum of phenotypes whose diversity may increase fitness.

  • Bacillus subtilis
  • differentiation
  • cell fate
  • development


One of the tenets of biology is that when a medium is inoculated with bacteria, nutrients allow the inoculum to grow by multiplying its members. The population of cells reaches a certain density, exhausts the nutrients, and stops growing. All of these cells are presumed to behave identically because they come from a clonal ancestor and have identical genomes. This represents the best scenario, where bacterial cultures are viewed as a suspension of billions of individual microorganisms, identical but independent of the rest. In the worst-case scenario, bacterial cells are treated as small bags, where interesting chemical reactions occur or simply, as tools to produce and purify plasmids or enzymes.

Microbiologists, however, have long appreciated the exceptions to this view of bacteria. One of the best-studied soil microorganisms, Bacillus subtilis, is well known for its ability to differentiate from vegetatively growing cells into metabolically inactive spores that allow it to survive under extreme environmental conditions (Piggot & Hilbert, 2004). Bacillus is not alone in its ability to differentiate. For example, the aquatic organism Caulobacter cresentus has innate heterogeneity built into its population. In order to replicate, Caulobacter divides asymmetrically to yield two genetically identical yet morphologically distinct cell types: the swarmer cells, which are able to swim using a flagellum, and the stalked cells, which attach to surfaces (Holtzendorff, 2004). Only the stalked cells can replicate their DNA and therefore the swarmer cells must first differentiate into stalked cells before replicating (Hottes, 2005). Another example is the photosynthetic Cyanobacteria such as Anabeana sp. that grow in filaments and, under nitrogen-limiting conditions, about 5–10% of the cells per filament differentiate into heterocysts that are specialized nitrogen-fixing cells (Adams, 2000). This differentiation is essential because vegetative cells produce oxygen through photosynthesis and the heterocyst protects the oxygen-sensitive nitrogen-fixing enzymes from the O2 produced by its neighbors (Zhang, 2006). There is interdependence between these two cell types as the heterocysts supply vegetative cells with fixed nitrogen and yet they rely on the photosynthetic population to provide them with carbon and reductants (Golden & Yoon, 2003). Even more complex are the Myxobacteria, which are social microorganisms that, upon starvation, begin an intricate dance in which thousands of cells coordinate their behavior to form mounds termed fruiting bodies (Curtis, 2007; Kroos, 2007). Only cells within mounds are capable of differentiating into metabolically inactive spores. However, even when the fruiting bodies have formed, a subset of cells termed peripheral rods maintain vegetative growth outside of the fruiting bodies, presumably ready to take advantage of any sudden changes in nutrient availability (O'Connor & Zusman, 1991).

The examples discussed above demonstrate how the ability to differentiate benefits the microorganism. Essentially, more efficient proliferation or survival of the community results from a division of labor of its members. In this way, the community can optimize its resources by differentiating into distinct cell types that have numerous metabolic processes activated at the same time, but not in all the same cells.

Bacillus subtilis is a master of differentiation as it has the ability to display a multitude of distinct cell types. Extensive research has resulted in a detailed molecular characterization of the physiology of several distinct cell types in dispersed cultures. For example, at the onset of the stationary phase, B. subtilis can differentiate into competent cells capable of taking up DNA from the environment (Dubnau, 1991; Dubnau & Provvedi, 2000), or as described above, it can differentiate into dormant spores that are highly resistant to external stresses (Rudner & Losick, 2001; Piggot & Hilbert, 2004). Additionally, a subset of cells can produce an extracellular ‘killing factor’ and toxin that functions to kill (or cannibalize) cells that have not yet begun sporulation, thereby allowing the cannibalistic cells to delay their own commitment to enter into the sporulation pathway (Gonzalez-Pastor, 2003). A fourth cell type is observed during biofilm formation when a portion of the population produces extracellular matrix material that holds cells together (Vlamakis, 2008). Differentiation has also been observed in cells growing exponentially. Only a fraction of cells express sigD, the sigma factor necessary for flagellar production, resulting in heterogeneity in motility (Kearns, 2005). Some of these cell types can be distinguished from their sister cells because their altered gene expression results in morphological changes that are visible under the microscope. Others can only be distinguished by measuring cell-type-specific gene expression using reporter molecules (Fig. 1).

Figure 1

Differentiation of cell types in Bacillus subtilis can be monitored in cells harboring transcriptional reporter fusions for each cell type. Images are overlays of transmitted light (grey) with fluorescent reporters. Blue indicates motile cells (Phag-cfp), red indicates matrix-producing cells (PyqxM-yfp), green indicates sporulating cells (PsspB-yfp) and yellow indicates competent cells (PcomG-yfp). Each cell type presents a characteristic phenotype that differentiates it from the rest of the population.

In this review, we have compiled the new advances describing the generation of distinct cell types of B. subtilis. We have highlighted the most important points in the differentiation process of each cell type, describing and presenting a personal interpretation of the challenges presented in the current investigations as well as the questions remaining for future studies.

Heterogeneity in gene expression

The classic approach to monitor variations in gene expression within a bacterial population is to measure average values in response to the presence of a specific signal. The traditional transcriptional fusion of gene promoters to the β-galactosidase gene (lacZ) is measured by a quantitative in vitro enzymatic assay of expression in a bulk population of cells. This technique is ideal if the pattern of gene expression is identical throughout the population, assuming that all cells respond similarly to the presence of the stimulus to express a certain set of genes. However, if the population comprises a heterogenous group of cells, averaging expression values will result in a gross miscalculation of what is occurring at the single-cell level.

In the past decade, many tools have become available to facilitate gene-expression studies at the single-cell level and the result has been a burst of research analyzing differentiation in heterogenous populations of bacteria (Brehm-Stecher & Johnson, 2004; Veening, 2004). Improvements in fluorescent protein technology have resulted in numerous reporters with differential excitation and emission spectra that allow monitoring of multiple fluorophores in a population of cells. Fluorescence can be monitored using microscopy to visualize individual cells directly or using flow cytometry to assay the distribution of fluorescence throughout a population. In this manner, a Gaussian distribution at the level of gene expression can be observed in a population of cells. This distribution is considered a unimodal differential expression of the gene, given that the cells in one side of the curve are not expressing at the same intensity as the cells distributed at the other side. The variation in expression is due to noise, or random fluctuations in the regulation of the expression of the gene.

In other cases, the distribution of gene expression clearly bifurcates into two distinct subpopulations, in which a subpopulation of activated cells expresses the reporter fusion and a subpopulation of nonactivated cells does not express the reporter (Fig. 2). Both subpopulations can be easily differentiated by fluorescence microscopy or flow cytometry (see Figs 1 and 2c). This phenomenon has been referred to as ‘bistability’, due to the dual stable pattern of gene expression in a population of genetically identical cells. The basis of the bistable response stems from a nonlinear induction of gene expression. Specifically, induction of the gene leads to a stronger induction of the gene and thus, once cells reach a certain threshold of gene expression, a hypersensitive response in the induction of the gene occurs (Dubnau & Losick, 2006).

Figure 2

Mechanism of bistable switch. (a) Positive autoregulation. The regulator induces its own expression. (b) Double repression. The regulator inhibits a repressor of the regulator. (c) Flow cytometry of a cell harboring the reporter fusion PyqxM-yfp to monitor the subpopulation of matrix-producing cells.

Two mechanisms have been described to control this hypersensitive-inducible gene expression (Ferrell, 2002). The first is a positive autoregulation (Fig. 2a), in which the product of the gene induces its own expression. In this situation, the level of gene expression is enhanced after a certain threshold of expression is reached, resulting in a positive feedback loop. In this manner, if a regulatory protein is slightly activated in a cell due to random fluctuations in gene expression, the feedback loop will result in high levels of expression of that regulator, and genes controlled by the regulator will thus be activated in that cell. However, cells that did not reach the threshold of expression to induce the positive feedback mechanism will remain inactive for the response.

This positive autoregulation is the case for competence activation in B. subtilis. The cascade of genes required for competence is induced by the regulator ComK; thus, an induction of comK expression will trigger activation of the competence pathway (see Fig. 3). Similar to the mechanism described above, the expression of comK is self-induced by the ComK regulator, and so when levels of ComK reach a certain threshold, the expression of its own gene comK increases nonlinearly (Maamar & Dubnau, 2005; Smits, 2005). Interestingly, the induction of comK can be achieved either by inducing expression of the gene comK or by promoting the stabilization of the ComK protein (Avery, 2005). Both cases result in a nonlinear induction of comK and an activation of competence.

Figure 3

Network of the different genetics pathways related to cell differentiation in Bacillus subtilis. Genes related specifically for each differentiation process are located within the specific frame.

The second mechanism to generate a bimodal population results from a pair of mutually repressing repressors (Fig. 2b). In this case, the expression of a repressor will repress the expression of another repressor. The effect may be similar to a positive autoregulation if the second repressor acts over the first one. Then, the repression of a repressor would mean activation. In this scenario, the system of mutually repressing repressors would be translated into a high expression of the first repressor when the system is repressed and a high expression of the second repressor when the system is activated.

In B. subtilis, the master transcriptional regulator, Spo0A, is actually regulated using both of the mechanisms described above. It is under the control of a positive feedback loop through its activation of itself as well as by a modified double-repression system. In this case, the repressor (Spo0A) represses abrB and AbrB represses sigH, an activator of spo0A. Therefore, until a threshold of Spo0A is reached, AbrB will repress sigH, the activator of spo0A, keeping Spo0A levels low (Fig. 3). A slight induction of Spo0A will alleviate the AbrB-mediated repression of SigH and the balance will be shifted to induce further spo0A expression (Strauch, 1990; Predich, 1992).

Cell fate determination in B. subtilis

A common mechanism used by Bacillus to ensure proper cell fate is to utilize the same regulator that activates one cell fate to inhibit a mutually exclusive cell fate. This is exemplified in the case of Spo0A and AbrB. As described in the previous section, Spo0A inhibits the expression of AbrB and AbrB inhibits the expression of Spo0A (through SigH). This genetic model predicts an inverse regulation in which cells expressing spo0A would not express abrB and vice versa. This was, in fact, the case as these two cell types were visualized at the single-cell level in an isogenic population of B. subtilis harboring yfp under the control of the abrB promoter and cfp under the control of the promoter of an Spo0A-P-regulated gene, spoIIA (Veening, 2004).

Differential gene expression has been observed previously in B. subtilis in studies of the sporulation pathway. As sporulation initiates, cells divide asymmetrically to produce two compartments: a mother cell and a forespore. Once this asymmetric division occurs, a cascade of sigma factors is triggered. Different sets of genes are activated by the sigma factors, resulting in a discrete expression profile for each compartment. For example, after the asymmetric division, SigF is activated in the forespore and SigF-regulated genes trigger SigE activation in the mother cell. The products of genes under the control of SigE and SigF are necessary for engulfment of the forespore. Once the forespore is engulfed, SigG becomes active in the forespore, leading to the activation of the SigK in the mother cell, which results in spore maturation, producing the cortex and the spore coat (Piggot & Hilbert, 2004). Because there are specific morphological changes that occur during sporulation it was possible to monitor the stage of sporulation using fluorescent membrane dyes. These dyes, coupled with fluorescence microscopy and gfp fusions to sigma factor-specific promoters, has allowed for beautiful visualization of gene expression in each compartment (Rudner & Losick, 2001).

Most of the situations we discuss in the following sections of this review arise from the activity of three major transcriptional regulators: DegU, Spo0A, and ComK. DegU and Spo0A are response regulators whose activities depend on their phosphorylation state. DegU becomes phosphorylated directly by its cognate histidine kinase, DegS (Kunst, 1994). As mentioned in the previous section, Spo0A is controlled at the transcriptional level by multiple pathways. In addition, Spo0A is regulated at the post-translational level by five histidine kinases (KinA, KinB, KinC, KinD, and KinE) that influence its phosphorylation state either directly or indirectly via the phosphorelay proteins Spo0F and Spo0B (LeDeaux, 1995; Jiang, 2000). Furthermore, several phosphatases interact with the phosphorelay proteins Spo0F and Spo0B to stimulate autodephosphorylation and thus decrease the available phosphate for Spo0A. These phosphatases are termed Rap proteins and they in turn are regulated by Phr peptides (Perego & Brannigan, 2001; Pottathil & Lazazzera, 2003) (Fig. 3).

One of these Rap proteins, RapA, has been shown to be essential for maintaining the bistable expression of Spo0A (Veening, 2005). A direct target of Spo0A-P, PspoIIA-gfp, was monitored for a bistable response in a strain where spo0A was expressed from an IPTG-inducible promoter. Despite an increase in Spo0A protein levels, there was no increase in PspoIIA-gfp expression when cells were harvested during vegetative growth, a time when the phosphorelay is inactive. An Spo0A-sad67 mutant that is in a constitutively active state expressed from the same IPTG-inducible promoter stimulates a bistable response from PspoIIA-gfp. Thus Spo0A phosphorylation is essential for the bistable response. Mutants in either RapA, which dephosphorylates the phosphorelay protein Spo0F, or Spo0E, a phosphatase that directly dephosphorylates Spo0A-P, are unable to elicit a bistable response in PspoIIA-gfp expression. Instead, both of these mutants express very high levels of the reporter, presumably because most cells have high levels of Spo0A-P. To our knowledge, mutants in other Rap proteins that affect the Spo0A phosphorelay have not been assayed for a bistable expression pattern.

Bacillus subtilis encodes 8 Phr peptides and 11 Rap proteins. Each Phr peptide is encoded in an operon with its cognate Rap protein (Lazazzera, 2001; McQuade, 2001). Phr peptides are initially translated as preproteins with an N-terminal secretion signal. After secretion and processing to a pentapeptide, they are imported via the Opp oligopeptide permease. Inside the cell, the Phr peptides specifically inhibit one or more cognate Rap regulatory proteins, thereby altering gene expression. Notably, not all Rap homologs function as phosphatases; some inhibit their target's ability to bind DNA by interacting directly with the proteins. Some of the processes regulated by the Rap proteins include sporulation, competence, transposition, and antibiotic and exoenzyme production (Perego & Brannigan, 2001; Auchtung, 2005, 2006).

The molecular mechanisms of B. subtilis regulation have been teased apart over the years and as a result of numerous excellent researchers, we have a detailed understanding of the pathways of genes that are involved in cellular differentiation. In the sections below we have focused on describing pairs of cell types whose fates are determined by a common regulator. We provide a number of examples in which this inverse regulation occurs.

Exoprotease production and motility

A remarkable feature of B. subtilis is its ability to produce large quantities of extracellular degradative enzymes, such as exoproteases (for instance, the exoprotease subtilisin encoded by the gene aprE and bacillopeptidase encoded by the gene bpr) or exopolysaccharases (such as the levansucrase encoded by the gene sacB). Aside from acting on products secreted by the cells, these enzymes can also degrade products released by dead cells within the stationary-phase culture. The putative role assigned to this armament of exogenous degradative enzymes is to provide extra nutrients to the community by degrading exogenous proteins into small peptides and polysaccharides into mono- or disaccharides that can more readily be assimilated by the cells (Msadek, 1999).

Expression of the degradative enzymes is under the control of the two-component system DegS–DegU (Mader, 2002; Ogura, 2003). DegS is a cytoplasmic kinase that phosphorylates the master regulator DegU. Once the regulator is phosphorylated, it initiates the expression of the set of genes involved in the production and secretion of the exoenzymes. The exact biochemical nature of the signal to activate DegS to phosphorylate DegU is unknown, but it has been proposed that DegS is activated by starvation or salt stress (Kunst & Rapoport, 1995).

The induction of exoenzyme expression requires a high level of DegU phosphorylation. In contrast, high levels of DegU-P repress motility (Ogura, 2003; Amati, 2004; Kobayashi, 2007; Verhamme, 2007). It has also been proposed that DegU gene regulation varies depending on the amount of phosphorylated protein in the cell and that low levels of DegU-P activate motility (Kobayashi, 2007; Verhamme, 2007). Ultimately, conditions that generate a high proportion of DegU-P correspondingly inhibit motility and induce exoprotease production. One can imagine the benefit of repressing motility when a cell produces an extracellular resource such as these exoenzymes. If the cell is to benefit from the degradation products, it must remain in the local environment where the enzymes are functioning.

Exoprotease production in cultures of B. subtilis has been monitored at the single-cell level with transcriptional fusions of fluorescent proteins to the promoter of the gene aprE (subtilisin). Using fluorescence microscopy, it was demonstrated that aprE is only expressed in a fraction of the cells. The heterogeneous expression of aprE has been attributed to heterogeneity in expression of the response regulator degU. Indeed, degU expression is also specifically activated during the stationary growth phase, and quantification of aprE-gfp and degSU-gfp reporters showed a coordinate increment with increasing time (Veening, 2008) (Fig. 3).

Contrary to the production of exoproteases, motility in B. subtilis occurs during the exponential phase and decreases during the stationary phase (Kearns & Losick, 2005). However, single-cell experiments showing that motility and exoprotease production are mutually exclusive in the population have yet to be performed. Because both processes are triggered by different phosphorylation states of the regulator DegU, it is reasonable to propose that the onset of one subpopulation defines the elimination of the other. An alternate hypothesis could be that both cell types are maintained in the community due to variations in the phosphorylation levels of the master regulator DegU for each type of cell.

Motility and matrix production

As will be discussed in detail later, multicellular communities of B. subtilis, termed biofilms, comprise numerous cell types, including motile cells and matrix-producing cells. These two processes are highly regulated at both the transcriptional and the post-translational levels. Transcription of genes important for motility and matrix production is regulated by Spo0A. High levels of Spo0A-P repress the fla/che motility operon, whereas Spo0A-P is required for extracellular matrix gene expression via the activation of the regulatory protein SinI (Fujita, 2005). As would be predicted by the inverse regulation, the subpopulation of cells differentiated to produce extracellular matrix does not exhibit induction of the genes related to motility. This was shown using time-lapse microscopy to monitor the expression of motility (Phag-cfp) and matrix-specific (PyqxM-yfp) promoters, where two discrete subpopulations of cells were observed. Under the conditions analyzed, cells either expressed the motility reporter, the matrix reporter, or no reporter; very few cells showed sustained coexpression, although at any given point some cells were in the process of transitioning from one state to the other (Vlamakis, 2008).

Motility requires the induction of a large fla-che operon, which contains 31 genes encoding for proteins that make up the basal body of the flagella, the chemotaxis system, and the sigma factor SigD. SigD is encoded at the end of the fla-che operon and is required for the expression of the hag locus, which encodes flagellin, the protein comprising the actual flagellar filament, as well as for the motA and motB genes, which encode for the motor proteins necessary for flagellar rotation (Marquez-Magana & Chamberlin, 1994). SigD also controls the expression of autolysins (lytA, lytD, and lytF) that function in cell separation, thus insuring that motile cells are nonfilamentous.

The transition from a motile to a sessile state is mediated by a biased bistable switch from the SigD sigma factor. The activation of SigD is driven by the master regulator DegU, which activates the expression of the operon in its unphosphorylated form. The unphosphorylated form of DegU directly binds the regulatory region of the fla/che operon promoter (Tsukahara & Ogura, 2008). The bias in SigD activity relies on the SwrA protein, which enhances expression of the fla-che operon and thus sigD (Kearns & Losick, 2005). It has recently been proposed that SwrA is required for DegU binding and thus for the expression of sigD. This finding reveals that swrA expression is controlled by a positive feedback loop, because the expression of swrAA is induced by SigD itself (Calvio, 2008).

Flagellar motility requires the expression of numerous genes and the activation of many proteins. This process requires energy and may be costly for the cell. However, it has been observed that some cells inhibit motility on behalf of other processes, such as matrix formation. Extracellular matrix production requires the activation of two operons: the 15-gene epsA-O operon, responsible for the production of the exopolysaccharide component, and the yqxM-sipW-tasA operon, responsible for the production and secretion of the major protein component of the matrix, TasA (Branda, 2004; Kearns, 2005). A recent report from Blair (2008) exquisitely explains a mechanism that inhibits motility post-translationally once matrix genes are expressed.

During exponential growth, the eps and yqxM operons are repressed in all cells by an inhibitor, SinR. SinI, an antagonist of SinR, is expressed only in a subpopulation of cells in which Spo0A has been activated (Chai, 2008). SinI antagonizes SinR by binding directly to SinR in a 1: 1 stoichiometry. Once SinI has been bound to SinR, the repressor cannot bind DNA and it is inactivated (Lewis, 1996, 1998). In this way, as levels of Spo0A-P increase in the cell, SinR-mediated repression is relieved, allowing the expression of the eps operon, and proteins needed for exopolysaccharide are produced. One of these proteins, EpsE, has a putative function as a family II glycosyltransferase, but also acts as an inhibitor of motility. When EpsE interacts with a specific flagellum protein, FliG, it inhibits flagellar rotation, similar to a clutch (Blair, 2008).

The clutch provided by EpsE might act as a fail-safe mechanism to prevent any wasted energy; it guarantees the stop of flagella rotation when cells are producing extracellular matrix, which would presumably interfere with rotation. Moreover, if conditions switch to favor motility over matrix production, flagella could be reactivated, allowing cells to reuse the investment of energy in flagella synthesis.

Matrix production and sporulation

Upon starvation or under other harsh conditions, cells differentiate to produce an endospore that functions to preserve DNA and essential proteins ultimately needed for germination when conditions become favorable (Piggot & Hilbert, 2004). Under some conditions, before sporulation, cells form biofilms, where they become encased in a self-produced extracellular matrix (Branda, 2001). Cells within biofilms are more resistant to biocides and antibiotics and part of this resistance is attributed to protection provided by the extracellular matrix. Both sporulation and matrix production are genetically triggered by phosphorylation of the master regulator, Spo0A. Additionally, not all of the cells in the population produce a matrix or sporulate (Dubnau & Losick, 2006; Chai, 2008).

Heterogeneity in the expression of the genes required for sporulation or matrix production is due to a bistable switch in the expression of the master regulator Spo0A. As discussed previously, Spo0A regulation is highly complex and involves many feedback loops. Once Spo0A is phosphorylated, the expression of the sigma factor SigH is stimulated through relief of AbrB-mediated repression (Hahn, 1995). SigH then stimulates transcription of other genes required for Spo0A phosphorylation (such as the kinase, kinA) and required for spo0A expression (such as spo0A itself). In this way, the phosphorylation of Spo0A (Spo0A-P) turns on a cycle that stimulates both its synthesis and its phosphorylation (Chung, 1994).

While spores and matrix producers are regulated by Spo0A, they are quite distinct cell types. What dictates the cell fate of a matrix producer or a spore? Spo0A regulates different genes when it is present at different levels. A high threshold level of Spo0A is needed to trigger sporulation and a lower threshold of the regulator is needed to induce sinI expression and derepress matrix genes (Fujita, 2005). Phosphorylation of the regulator Spo0A is driven by five different histidine kinases (A, B, C, D, and E), activated by a myriad of stimuli, most of them still unknown (Jiang, 2000). Of these five kinases that can ultimately phosphorylate Spo0A, KinA is essential for spore formation, while KinC and KinD are required for biofilm formation (Hamon & Lazazzera, 2001; Kobayashi, 2008). The difference between these kinases lies in their kinase activity. In vitro analysis of purified kinases showed that KinA is a much more efficient kinase than KinC or D (LeDeaux, 1995; Jiang, 2000). Thus, it is possible that depending on which kinase is stimulated, different levels of Spo0A activation will occur, resulting in a different cell fate.

Time-lapse microscopy was used to follow differentiation of cells harboring both a matrix-specific promoter (PyqxM) fused to CFP and a sporulation-specific (PsspB) promoter fused to YFP. By analyzing cells as they began to express these cell-type-specific promoters, sporulating cells were observed to arise mostly from matrix-producing mother cells. This seems reasonable, assuming that in order to reach high threshold levels of Spo0A-P, a cell must first go through a low threshold state, resulting in sequential activation of matrix and sporulation genes in the same cells. However, it has been observed that some sporulating cells arise suddenly without first expressing matrix genes (Vlamakis, 2008). We might speculate that conditions exist where high threshold levels of Spo0A-P are reached directly, or in a rapid response, and in these situations there is not enough Spo0A-P at a low threshold to activate matrix gene expression first.

Sporulation and cannibalism

As we have explained before, nutrient limitation or cell stresses can trigger sporulation in a subpopulation of B. subtilis cells via the phosphorylation of Spo0A. We also discussed how gradual increases in the levels of Spo0A-P activate or repress different sets of genes. Similar to what was observed with extracellular matrix production, the activation of lower levels of Spo0A additionally triggers the expression of two operons: skf (sporulation-killing factor) and sdp (sporulation-delaying protein). The proteins produced from these operons are bacteriocins that cause a large reduction in the number of viable B. subtilis cells before cells are irreversibly committed to sporulation. This enables the population of cells to literally cannibalize their siblings in order to provide food to the cells resistant to the toxins and to delay the one-way process to the formation of the spores (Gonzalez-Pastor, 2003).

Cells susceptible to the action of the Skf and Sdp peptidic toxins (Spo0A OFF cells) are killed and through their lysis they release nutrients into the extracellular milieu. The Spo0A ON cells are immune to the action of the peptides due to Spo0A-P-induced expression of the machinery to detoxify cells from the peptides (Ellermeier, 2006). Consequently, B. subtilis cells that do not produce Spo0A will be sensitive to the action of the killing factors, while cells producing Spo0A are resistant.

Cells committed to sporulation are also cannibalistic and they obtain their food by killing their siblings that are not committed to the formation of spores. Because of the irreversibility of the sporulation pathway, delaying sporulation before commitment might be beneficial in some situations.

Sporulation and competence

The ability of B. subtilis to develop natural competence is very well known and it has been the subject of study for many years. Competence can be defined as a physiological state in which cells are able to uptake and assimilate exogenous DNA in a process controlled by the cell itself. Only a small fraction of cells shows natural competence (around 10% of the total population). Remarkably, the competent cells can be separated from the rest using a sucrose gradient, supporting the idea that these cells are, actually, morphologically different from the other members of the community (Singh, 1967) (Fig. 1).

The subpopulation of competent cells within the community arises due to the bistable regulation of the master regulator ComK, which controls the initiation of the cascade of the genes involved in competence (Dubnau, 1991; van Sinderen, 1995). The comK gene can be stimulated by its own product ComK (Maamar & Dubnau, 2005); thus, comK expression increases nonlinearly due to a positive feedback loop. This results in a heterogeneous population of cells in which some cells express ComK and others do not (Suel, 2006). Importantly, promoting the stabilization of ComK protein can also induce the bistable response, due to the induction of the comK expression that ComK protein exerts (Avery, 2005).

The activation of the expression of comK is driven by the induction of the quorum-sensing pathway ComX-ComP-ComA (Magnuson, 1994). Briefly, the pheromone ComX activates the histidine kinase ComP to phosphorylate the master regulator ComA. Once ComA is in its active form, ComA-P, it triggers the expression of the regulator ComK to initiate the pathway to competence.

Remarkably, ComA-P can also regulate the expression of other features that are associated with sporulation. For instance, the phosphorylated form of ComA can induce the expression of the proteins RapA, RapB, and RapE (Auchtung, 2006). These proteins belong to the peptidic system of quorum sensing of B. subtilis and they specifically block the phosphorelay route to phosphorylate the master regulator Spo0A. Thus, cells differentiated to be competent have inhibited the process of Spo0A phosphorylation and they will not sporulate.

Moreover, activation of the expression of the competence regulator ComK also acts in a second role of inhibiting the expression of the gene spo0A gene (Berka, 2002; Claverys & Havarstein, 2007). In this case, the activation of the competence leads to the inhibition of Spo0A expression. Because of this regulation, cells committed to be competent are maintained as a distinct population and can only initiate sporulation once the levels of ComK have decreased within the cell. Flow cytometry analysis of cells harboring gfp fusions to either a sporulation-specific promoter (PspoIIA-gfp) or a competence promoter (PcomG-gfp) showed that competence and sporulation occur in only a subpopulation of cells and are sequential processes (Veening, 2006b). In the same study, it was shown that competent cells, capable of taking up an antibiotic resistance marker, were able to develop into mature spores.

Benefits of population heterogeneity: altruism or cheating?

The advantages of a heterogeneous population in the bacterial community can be easily postulated for a better adaptation to unexpected environmental fluctuations. By expressing a diverse set of genes in different cell types, the division of labor also permits the high optimization of resources. The maintenance of different cell types allows for simultaneous expression of different metabolic pathways with a minimal cost of energy, compared with the hypothetical cost needed if the expression would occur in all of the cells.

A clear example of this is the ability of B. subtilis to produce and secrete large quantities of extracellular proteases when cultures reach the stationary phase. These secreted enzymes are scavenging proteins that catabolize the degradation of other proteins into smaller peptides. The small peptides produced provide to the community of cells an extra resource of nutrients under starvation conditions, as long as they can be taken up and used as an alternative nutrient source by the bacteria (Msadek, 1999).

It was recently reported that these extracellular proteases follow a bistable expression pattern and only a fraction of the cells highly expresses the genes encoding the exoproteases (Veening, 2008). Interestingly, although only a subpopulation of cells is responsible for the production and secretion of these exoproteases, the whole community benefits from this feature, in terms of the release of the small peptides derived from the degradation of proteins, and can be used as nutrients and metabolized by all the cells, regardless of whether they belong to the subpopulation that actually produced the exoproteases.

Another example of the beneficial effect of division of labor occurs during the formation of the multicellular communities of cells known as biofilms. In the process of biofilm formation, cells form long chains that are held together in bundles. The ‘glue’ that holds the cells together is a self-produced extracellular matrix, which is composed of exopolysaccharides and a major protein component (TasA) (Kearns, 2005; Chu, 2006).

There are two operons responsible for matrix production (eps and yqxM operons). These operons are coordinately expressed in cells to produce exopolysaccharides and TasA, which are both essential for the integrity of the extracellular matrix. The expression of both operons follows a bistable pattern and thus heterogeneity in the expression of these genes has been reported (Chai, 2008; Vlamakis, 2008). Although the production of the extracellular matrix is observed in only a subpopulation of cells, all the cells that form the biofilm are encased in this extracellular matrix, regardless of whether they produced the components to assemble the matrix.

The examples highlighted above clearly demonstrate the potential benefits of having a heterogeneous population. The high cost in energy to produce exoproteases and extracellular matrix is paid only by a subpopulation of cells. Because these products are extracellular, their benefit is equally shared by the subpopulation of nonproducing cells.

Some interesting questions might arise from this concept. For instance, it can be argued whether this behavior might constitute a form of altruism in bacteria. The fact that the subpopulation of cells responsible for producing the common benefit needs to spend a large amount of energy to metabolically produce a common benefit might support this idea. However, more experiments are needed to determine whether the producing subpopulation really suffers from this behavior and concomitantly, if the rest of the population benefits by taking advantage of the resources. If that is correct, we might speculate that the subpopulation of cells receiving the benefit might constitute a subpopulation of ‘cheaters’ or, in contrast, if they do not take advantage of the situation, they might show a cooperative behavior (Fiegna & Velicer, 2005; Fiegna, 2006).

We are inclined to think that the division of labor in B. subtilis is more likely to be a co-operative behavior instead of cheating. If cooperation is occurring in these communities, all of the subpopulations that constitute the community must produce some beneficial features for the population, and not many cells would be free of this duty. Supporting this idea, we can examine the production of extracellular matrix to observe that the subpopulation of cells that save energy in this process does not overcome in the population as we would expect from a typical cheater behavior. It is more likely that this energy is spent in other processes that might benefit the community in other different ways.

Spatiotemporal determination of cell types within the biofilm

We have described the occurrence of different cell types within the B. subtilis cultures, leading to a division of labor among members of the population as well as how this division of labor can benefit cells within the population. How do these populations manifest in structured communities such as biofilms?

Less than 10 years ago, B. subtilis was first described to form multicellular communities known as biofilms (Miller & Diaz-Torres, 1999; Branda, 2001; Hamon & Lazazzera, 2001). Within the biofilm, cells differentiate from a predominantly unicellular motile state to a mixture of cell types, including chains of cells that are held together by a self-produced extracellular matrix. These biofilms are architecturally complex structures and eventually, sporulation occurs preferentially in macroscopic aerial structures termed fruiting bodies (Branda, 2001). The localization of sporulating cells was initially visualized by Branda and colleagues, who used a transcriptional fusion with lacZ to demonstrate the preferential expression of the sporulation-specific gene (sspE) in the tips of these aerial structures. Veening (2006a) took this a step further and used fluorescence microscopy to visualize the localization of cells expressing gfp under the control of different promoters. In this study, they showed that abrB is expressed in the base of the biofilm and sporulation-specific promoters were expressed in the aerial structures. They further analyzed mutants lacking either the phosphatase, RapA, or the transcriptional regulator SinR, and showed that the timing and spatial organization of PspoIIA-gfp expression were dramatically altered in these mutants. Although these approaches defined the specific areas where a certain pattern of gene expression was overwhelming, it did not permit the visualization of these gene patterns at the single-cell level (Veening, 2006a).

In other organisms, confocal microscopy has yielded three-dimensional biofilm images with a single-cell resolution (Lawrence & Neu, 1999). These images have been used to analyze spatial gradients of oxygen, pH, and the localization of live and dead cells within the biofilm (Webb, 2003; Werner, 2004). Such images have for the most part been achieved by analyzing cells in flow chambers where biofilms are attached to glass or plastic surfaces and fresh medium is continually circulated through the chamber (Busscher & van der Mei, 1995). Flow-cell analysis is not possible for B. subtilis because biofilms are very hydrophobic and do not adhere to glass or plastic surfaces well, but rather form on an agar plate or on the surface of a liquid. In addition, most studies using flow cells and confocal microscopy have been performed with organisms that produce relatively thin biofilms, about 50 μm in depth. In contrast, mature B. subtilis biofilm can be up to 1 mm in depth, presenting major difficulties for microscopy because the working distance of most high-magnification objectives is <200 μm (Lawrence & Neu, 1999).

Imaging of B. subtilis biofilms at a single-cell resolution was achieved using a thin-sectioning technique and high-magnification fluorescence microscopy (Vlamakis, 2008). Transcriptional fusions to fluorescent proteins were used to monitor the spatial pattern of cells expressing representative genes within the population. An extraordinary spatiotemporal organization was observed within the B. subtilis biofilm. Motile, matrix-producing, and sporulating cells were localized individually and in pairs relative to each other. As seen in the previous studies, sporulating cells localized towards the top and the center of the biofilm. In addition, motile cells localize to the base and the edge of the biofilm, while matrix-producing cells localize in patches throughout the biofilm. This patchy localization is consistent with the extracellular matrix being a shared resource that must function to hold together all of the cells in the biofilm, regardless of whether or not they produce matrix. By complementing the thin-sectioning technique with flow cytometry and analyzing biofilms over time as they developed, a quantitative study of the temporal expression pattern of each cell type was achieved. The major finding of this study was that mature biofilms comprise at least three different cell types, that these cell types are localized to discrete regions of the community, and that the spatiotemporal expression patterns of different genes are dynamic throughout biofilm development. In addition, the architectural complexity of the biofilms was essential for this differentiation. Mutants that did not produce the extracellular matrix formed featureless biofilms that lacked a sporulating subpopulation.

From these studies, we postulate a novel level of complexity in the spatiotemporal distribution of different cell types within spatially constrained multicellular communities of B. subtilis. Contrary to our prior thinking that in order to form a biofilm cells must differentiate from motile to matrix-producing, we now understand that the biofilm does require matrix-producing cells, but that even in mature biofilms a motile population persists. It is likely that as more cell types are analyzed, we will further appreciate the diversity of a biofilm's constituent cells.

Concluding remarks

The bacterium B. subtilis has a remarkably complex network to regulate its gene expression and to ensure that the genetically identical population can display a multitude of behaviors. Many of these behaviors are mutually exclusive and the regulatory system must take into account the alternative paths a cell might take and devise mechanisms to prevent coexpression of the wrong genes. Ultimately, this finely choreographed population is composed of different cells, each with specific purposes. We have observed several of these cell types coexisting within the structured biofilm community. It is highly likely that further research will reveal an even more intricate process than what we have come to appreciate. This organization of different cell types in the biofilm closely resembles the cellular differentiation and spatial organization observed in structures formed by multicellular eukaryotes. For example, in certain fungi, the specialization of cells to form an aerial mycelium might be comparable to the formation of the fruiting bodies by sporulating cells in the biofilm. In addition, the core of the biofilm is formed by specialized cells, able to produce the extracellular matrix, similar to the development of certain fungi, where mycelium is formed by a dense mass of filaments called hyphae, which are composed by many cells joined together in chains and separated by septa.

Similarities in development processes between multicellular eukaryotes and bacteria are strongly encouraging to advance investigation in this field. There are numerous questions that remain. Is cell–cell communication involved in the developmental process to coordinate the growth and the spatial distribution of the different cell types within the biofilm? How do other different microorganisms present in the same ecological niche, the soil, affect cellular differentiation? Does population heterogeneity provide an advantage in natural settings? We anticipate that the answers to these questions will provide an insight into how an organism controls its differentiation processes and furthermore on what evolutionary advantage is provided by maintaining a heterogeneous population. Bacillus subtilis is an ideal system in which to ask these types of questions because there is a plethora of information regarding the regulation of differentiation, and the ease of working with a microorganism with a very short doubling time that is amenable to genetic modifications should facilitate future studies.


  • Editor: Victor de Lorenzo


View Abstract