Role of σ^D in Regulating Genes and Signals during Myxococcus xanthus Development

Abstract

Starvation-induced development of Myxococcus xanthus is an excellent model for biofilm formation because it involves cell-cell signaling to coordinate formation of multicellular mounds, gene expression, and cellular differentiation into spores. The role of σ^D, an alternative σ factor important for viability in stationary phase and for stress responses, was investigated during development by measuring signal production, gene expression, and sporulation of a sigD null mutant alone and upon codevelopment with wild-type cells or signaling mutants. The sigD mutant responded to starvation by inducing (p)ppGpp synthesis normally but was impaired for production of A-signal, an early cell density signal, and for production of the morphogenetic C-signal. Induction of early developmental genes was greatly reduced, and expression of those that depend on A-signal was not restored by codevelopment with wild-type cells, indicating that σ^D is needed for cellular responses to A-signal. Despite these early developmental defects, the sigD mutant responded to C-signal supplied by codeveloping wild-type cells by inducing a subset of late developmental genes. σ^D RNA polymerase is dispensable for transcription of this subset, but a distinct regulatory class, which includes genes essential for sporulation, requires σ^D RNA polymerase or a gene under its control, cell autonomously. The level of sigD transcript in a relA mutant during growth is much lower than in wild-type cells, suggesting that (p)ppGpp positively regulates sigD transcription in growing cells. The sigD transcript level drops in wild-type cells after 20 min of starvation and remains low after 40 min but rises in a relA mutant after 40 min, suggesting that (p)ppGpp negatively regulates sigD transcription early in development. We conclude that σ^D synthesized during growth occupies a position near the top of a regulatory hierarchy governing M. xanthus development, analogous to σ factors that control biofilm formation of other bacteria.

In nature, many bacteria exist in biofilms, within which cells send signals to each other and respond by changing expression of certain genes (67, 68). This not only alters the physiology of individual cells but also often causes macroscopic change in the morphology of the biofilm. Understanding how cells interact to bring about such change is a fundamental challenge of broad practical significance (48).

Myxococcus xanthus is a gram-negative soil bacterium that provides an excellent experimental model for biofilm research (53). Upon sensing nutrient limitation, certain genes are induced and signals are produced (29, 31, 64). Cells respond by altering their pattern of gliding movements and piling on top of one another to form mounds. Within the mounds further signaling and gene expression cause cells to differentiate from metabolically active rods into dormant, spherical spores.

Decades of research on M. xanthus have elucidated numerous molecular events that occur during the developmental process (13). The events tracked in this study are depicted in Fig. 1. Starvation induces a stringent response that involves RelA-dependent synthesis of (p)ppGpp (23, 49, 50, 63). This intracellular signal induces early developmental genes such as csgA (10) and sdeK (23, 63). CsgA is a 25-kDa protein that helps maintain the stringent response within cells (11) and appears to serve as an extracellular signal (C-signal) after being cleaved to a 17-kDa form at the cell surface (34, 47, 62). SdeK is a histidine kinase essential for expression of many developmental genes (18, 56), but neither its input nor its substrate is known. (p)ppGpp also initiates production of A-signal (23), a mixture of peptides and amino acids generated by extracellular protease activity (41, 55). If cells are at a high enough density (42), sufficient A-signal is made to trigger expression of genes such as Ω4521 (also known as spi), Ω4514 (40), and fruA (52). The genes Ω4521 and Ω4514 were identified by insertion of Tn5 lac (38) (as is the case for other numbered genes and operons described below). Ω4521 expression has been used to develop an assay for A-signal (39, 40) and to facilitate genetic screens that have elucidated the A-signal transduction pathway (21, 76, 77). The Ω4514 operon is expressed slightly later during development than Ω4521 due to negative autoregulation by the product of the first gene in the operon (22). FruA is a response regulator necessary for responses to C-signal (14, 52), but its putative kinase has not been found. C-signaling is needed for progression beyond the morphological stage of forming broad, low mounds and for full expression of most genes that begin to be transcribed after about 6 h into development (36). Some genes, such as Ω4400 and the Ω4499 and dev operons, are expressed weakly in the absence of C-signaling but require C-signaling for full expression (4, 36). Of these, only dev is important for sporulation (37, 38, 71). Genes expressed later, such as Ω4403 and the Ω7536 operon, fail to be expressed in the absence of C-signaling, and Ω7536 is required for sporulation (36, 46).

FIG. 1.

Timeline of events during M. xanthus development. Signal inputs are below the line and genes induced are above.

In this study, we used the molecular markers described above to investigate the role of σ^D in regulating genes and signals during M. xanthus development. σ^D is somewhat similar to M. xanthus σ^A, the major vegetative σ factor in the σ⁷⁰ family, and to three alternative M. xanthus σ factors (σ^B, σ^C, and σ^E), although σ^D and the other alternative σ factors are much smaller than σ^A (73). σ^D is essential for viability in stationary phase, helps cells resist oxidative and osmotic stress, and is required for normal development (73). A sigD null mutant exhibits altered patterns of protein synthesis compared to wild-type cells during the late-log phase of growth and early in development (73); however, none of the differentially expressed proteins was identified. Here, we employ well-characterized markers to examine the role of σ^D in development. Under the conditions we used, the sigD null mutant failed to form mounds containing spores (fruiting bodies) within 72 h, whereas wild-type cells formed mounds by 12 h and some spores by 24 h, with the spore number increasing over the next 2 days. We found that the sigD mutant undergoes a stringent response to starvation but fails to induce developmental genes and signals. Surprisingly, the sigD mutant was able to respond to codeveloping wild-type cells by inducing a subset of C-signal-dependent genes; however, the results suggest that σ^D RNA polymerase or a gene under its control is required cell autonomously to increase transcription of the dev operon in response to C-signal, an event that is important for sporulation. We also discovered that the level of sigD transcripts in cells decreases more than 100-fold after 20 min of starvation, so it appears that σ^D synthesized during growth fulfills its early role in development.

MATERIALS AND METHODS

Bacterial strains and plasmids.

M. xanthus strains that were used in this work are listed in Table 1. Plasmids were introduced into M. xanthus by electroporation (32). The plasmid pPVsigD was used as a control for quantitative PCR (QPCR; see below). To construct pPVsigD, a 951-bp DNA fragment including the sigD gene and 200 bp of upstream DNA was amplified from M. xanthus DK1622 chromosomal DNA using the primer pair 5′-CCGAAGCTTTCGTGGCCGACGCGGCTC-3′ and 5′-GCCGGATCCTCACGCCGCGTCGGCGG-3′. The DNA fragment was purified by agarose gel extraction and cloned into pCR2.1-TOPO according to the manufacturer's instructions. The DNA insert was sequenced at the Michigan State University Genomics Technology Support Facility to ensure that the correct sequence was obtained. Tn5 lac fusions were introduced into M. xanthus by generalized transduction; Mx4 phage stocks (6, 19, 28) were prepared on donor strains, and phages were mixed with recipients (10⁸ cells) at multiplicities of infection of 0.1, 0.5, 1.0, and 2.0.

TABLE 1.

M. xanthus strains used in this study

Strain	Relevant characteristic(s)a	Source or reference
DK101	pilQ1	25
DK1622	Wild type	30
DK4322	Tn5 lac (Kmr) Ω4521	40
DK4324	Tn5 lac (Kmr) Ω4521 asgB480	40
DK4368	Tn5 lac (Kmr) Ω4403	38
DK4408	Tn5 lac (Kmr) Ω4408 (insertion in sdeK)	40
DK5057	Tn5 (Kmr) Ω4560 asgA476	39
DK5208	csgA::Tn5-132 (Tcr) Ω205	61
DK5270	Tn5 lac (Kmr) Ω4403 csgA::Tn5-132 (Tcr) Ω205	36
DK5279	Tn5 lac (Kmr) Ω4414 (insertion in devR)	36
DK5287	Tn5 lac (Kmr) Ω4414 csgA::Tn5-132 (Tcr) Ω205	36
DK10524	Tn5 lac (Kmr) Ω7536	46
JPB40030	attB::pJB40030 (Kmr) (Ω4400-lacZ fusion)	78
MDS12	attB::pDS12.P (Kmr) (fruA-lacZ fusion)	65
MDS200	csgA::pJM200 (Kmr) (csgA-lacZ fusion)	59
MDY101	attB::pDY101 (Kmr) (Ω4499-lacZ fusion)	79
MDY4400.SD	ΔsigD Tn5 lac (Kmr) Ω4400	78
MDY4499.SD	ΔsigD attB::pDY101 (Kmr)	79
MS1000	DK101 ΔrelA1	12
MTH1-3	attB::pTH1-6 (Kmr) (Ω4514-lacZ fusion)	22
ΔsigD	sigD deletion	73
sigDcsgA	ΔsigD csgA::pJM200 (Kmr)	This study
sigD4408	ΔsigD Tn5 lac (Kmr) Ω4408	This study
sigD4521	ΔsigD Tn5 lac (Kmr) Ω4521	This study
sigDfruA	ΔsigD attB::pDS12.P (Kmr)	This study
sigD4514	ΔsigD attB::pTH1-6 (Kmr)	This study
sigDdev	ΔsigD Tn5 lac (Kmr) Ω4414	This study
sigD7536	ΔsigD Tn5 lac (Kmr) Ω7536	This study
sigD4403	ΔsigD Tn5 lac (Kmr) Ω4403	This study
ΔfruA	fruA deletion/replacement (Smr)	T. Ueki and S. Inouye
sigD4400fruA	ΔsigD Tn5 lac (Kmr) Ω4400 ΔfruA (Smr)	This study
sigD4403fruA	ΔsigD Tn5 lac (Kmr) Ω4403 ΔfruA (Smr)	This study
sigD4499fruA	ΔsigD attB::pDY101 (Kmr) ΔfruA (Smr)	This study

^aKm^r, kanamycin resistant; Tc^r, oxytetracycline resistant; Sm^r, streptomycin resistant.

Bacterial growth and development.

Escherichia coli was grown at 37°C in Luria-Bertani medium (60) containing 50 μg/ml of ampicillin. M. xanthus was grown at 32°C in CTT medium (24) (1% casitone, 10 mM Tris-HCl [pH 8.0], 1 mM KH₂PO₄ · K₂HPO₄, 8 mM MgSO₄ [final pH 7.6]) in liquid cultures or on agar (1.5%) plates. Forty micrograms of kanamycin per milliliter, 12.5 μg/ml of oxytetracycline, or 200 μg/ml of streptomycin was used when required for selective growth. For starvation induction of M. xanthus in liquid cultures, cells were grown to a density of approximately 5 × 10⁸ cells/ml, the culture was centrifuged, and the cell pellet was resuspended in prewarmed TPM (10 mM Tris-HCl [pH 8.0], 1 mM KH₂PO₄ · K₂HPO₄, 8 mM MgSO₄ [final pH 7.6]) liquid medium to a density of approximately 5 × 10⁸ cells/ml, as described previously (63). Fruiting body development was performed on TPM agar (1.5%) plates as described previously (38). Developmental β-galactosidase activity was measured as described previously (38). Codevelopment (also called extracellular complementation) was performed by mixing equal optical density units of cells as described previously (36). For mutants alone or upon codevelopment with wild-type cells, developmental β-galactosidase activity was measured for three isolates, and points on the graphs show the average while error bars show one standard deviation of the data (in many cases the error bar is so small that it is hidden by the symbol representing the point on the graph). In each experiment, developmental β-galactosidase activity from the corresponding lacZ fusion in wild type (one isolate) was measured as a positive control (which agreed with previously published results) and is plotted on the graph. Extracellular complementation of sporulation was performed as described previously (36), except the production of heat- and sonication-resistant spores was assayed as follows. Cells from spots of five 20-μl aliquots were scraped from TPM agar and suspended in 500 μl of sterile distilled water. The suspension was sonicated four times for 10 s with intermittent cooling and then incubated in a water bath at 50°C for 2 h. Tenfold dilutions of the resulting suspension in distilled water were plated in CTT soft agar and incubated at 32°C for 4 days before colonies were counted.

Quantification of guanosine nucleotides.

In vivo analysis of guanosine nucleotides was performed as described previously (49) with the following modifications. Cells were grown in CTT medium without the addition of 1 mM KH₂PO₄ · K₂HPO₄. The concentration of inorganic phosphate in this medium was determined to be 0.75 mM using a method described previously (2). After 6 h of growth in the presence of 100 μCi/ml [³²P]orthophosphate (NEX053H carrier free, 285.6 Ci/mg; Perkin Elmer) to equilibrate the phosphate pools at a cell density of 100 Klett units, a 1-ml aliquot was removed as the time zero reference sample, and nucleotides were extracted as described previously (49). The remaining culture was centrifuged, and the cell pellet was resuspended in TPM medium modified to contain 0.75 mM KH₂PO₄ · K₂HPO₄ and 100 μCi/ml [³²P]orthophosphate to maintain the same specific activity. One-milliliter aliquots were removed at 15, 30, 45, and 60 min, and nucleotides were extracted and analyzed by thin-layer chromatography as described previously (49). The amount of radioactivity in each spot was then quantified on a PhosphorImager 400S (Molecular Dynamics, Sunnyvale, CA) equipped with the ImageQuant, version 5, program (Molecular Dynamics, Sunnyvale, CA).

A-signal production and assay.

A-factor was harvested from developing cells as described previously (39, 40). A-factor was assayed by measuring the restored expression of Tn5 lac Ω4521 in the asgB480 mutant background (M. xanthus DK4324) as described previously (39). One unit of A-factor produced 1 nmol of o-nitrophenyl-β-d-galactopyranoside per min, for an entire test well containing 1.25 × 10⁸ cells, above the background. To compare the activities of A-factor harvested from different bacterial strains, dilutions of each cell-free supernatant were assayed, and the linear portion of an activity-versus-volume curve was used to estimate A-factor activity (39).

Measurement of sigD transcript levels by QPCR.

Total RNA samples were extracted by the hot phenol method (60) from approximately 10¹⁰ quick-frozen M. xanthus cells. These cells were harvested either during vegetative growth at a cell density of approximately 5 × 10⁸ cells/ml or after 20 or 40 min poststarvation as described above. Analysis was performed as described by Diodati et al. (12). Expression of the sigD gene was calculated from DNA quantities obtained for concurrent standard curve reactions using pPVsigD. Typically, a standard curve was generated by reacting a 10-fold dilution series of plasmid DNA, ranging from 10¹⁰ to 10¹ copies/μl, with the following sigD forward and reverse primer pair: 5′-CATGGCCAATTCGACGAAGT-3′ and 5′-GCCTTCATGAGACCGACGTT-3′, respectively. The DyNAmo HS SYBR Green QPCR Master Mix (Finnzymes, Espoo, Finland) was used as described by the manufacturer. Each sample was assayed in triplicate, and at least two biological samples were assayed.

RESULTS

σ^D is required for normal expression of early, A-signal-independent genes.

The csgA and sdeK genes are induced at the onset of development, and their expression does not depend on A-signaling (10, 40, 45). Developmental expression of both genes was severely impaired in a sigD null mutant (Fig. 2). The defect in expression was cell autonomous; it could not be rescued extracellularly by codevelopment of the sigD mutant with wild-type cells (Fig. 2).

FIG. 2.

Expression of A-signal-independent genes in a sigD mutant. β-Galactosidase specific activity during development was measured for lacZ fusions to csgA (A) and sdeK (B) in wild type (▪) or in a sigD mutant alone (•) or upon codevelopment of the fusion-containing sigD mutant with an equal number of fusionless wild-type cells (○).

A sigD mutant responds normally to starvation by synthesizing ppGpp.

Developmental induction of csgA and sdeK depends on RelA (10, 23), which mediates the stringent response to amino acid limitation by synthesizing (p)ppGpp (23). To determine whether σ^D is required for the stringent response, ppGpp and GTP (from which ppGpp is synthesized) levels were measured during nutritional downshift. The sigD null mutant was indistinguishable from wild type (Fig. 3). We conclude that σ^D RNA polymerase is not required for the stringent response to starvation but is directly or indirectly needed for increased transcription of csgA and sdeK early in development (Fig. 2).

FIG. 3.

ppGpp and GTP levels during nutritional downshift of wild type and a sigD mutant. Guanosine nucleotides were measured in extracts of wild-type cells (squares) or a sigD mutant (circles) as described in Materials and Methods during growth (time zero) or after resuspension in starvation buffer. ppGpp (closed symbols) and GTP (open symbols) levels are expressed relative to the time zero level in a representative experiment.

σ^D is important for expression of A-signal-dependent genes.

Three reporter fusions, Ω4521, fruA, and Ω4514, that are normally induced early in development and whose expression depends on A-signaling (40, 52), also failed to be expressed normally in the sigD null mutant (Fig. 4). Supplying A-signal by mixing wild-type cells with the reporter fusion-bearing sigD mutants and allowing them to codevelop did not restore developmental expression (Fig. 4). These results suggest that σ^D RNA polymerase is required in a cell autonomous fashion for normal expression of A-signal-dependent genes.

FIG. 4.

Expression of A-signal-dependent genes in a sigD mutant. β-Galactosidase specific activity during development was measured for lacZ fusions to Ω4521 (A), fruA (B), and Ω4514 (C) in wild type (▪) or in a sigD mutant alone (•) or upon codevelopment of the fusion-containing sigD mutant with an equal number of fusionless wild-type cells (○).

A sigD mutant is defective in producing A-signal.

Although the sigD mutant did not respond properly to A-signal (Fig. 4), it was possible that the sigD mutant could produce this signal. We tested this possibility in two ways. First, the sigD mutant was codeveloped with an asgA mutant that is unable to make A-signal. Alone, the asgA mutant is unable to sporulate (Table 2). Upon mixing with the sigD mutant and codevelopment, spores formed, and all the spores were derived from the asgA mutant, but the number was about 10⁴-fold less than the number of asgA-derived spores produced upon codevelopment with wild-type cells (Table 2). This suggests that the sigD mutant makes A-signal at a reduced level. In agreement, the sigD mutant released about fourfold less A-signal than wild-type cells (after the background value observed for the asgA mutant was subtracted), as measured by the ability of medium from starved cells to stimulate Ω4521 expression in asgA mutant cells (Table 3). We conclude that σ^D is required both to produce the normal level of A-signal and to respond to it by directly or indirectly increasing transcription of A-signal-dependent genes.

TABLE 2.

Extracellular complementation of sporulation

Strain or mixture (description)a	Sporulation (% wild type)b	Rescue (%)c
DK1622 (wild type)	100
ΔsigD (sigD mutant)	<0.001
DK5057 (asgA mutant)	<0.001
DK1622+DK5057	140 ± 38	18
ΔsigD+DK5057	0.0023 ± 0.0007	100
DK5208 (csgA mutant)	<0.001
DK1622+DK5208	130 ± 18	100
ΔsigD+DK5208	15 ± 11	100

^aFor strain mixtures (+), the second listed strain was unable to sporulate alone and contained a marker gene that confers resistance to kanamycin (DK5057) or oxytetracycline (DK5208).

^bProduction of heat- and sonication-resistant spores was assayed as described in Materials and Methods after 4 days of development at 32°C. Wild type produced 5 × 10⁶ spores. Numbers indicate averages and standard deviations of three experiments.

^cThe ability of the first strain to rescue sporulation of the second was assessed by transferring cells from spore-derived colonies onto CTT agar containing the appropriate antibiotic. The number that grew after 3 days at 32°C is expressed as a percentage of the number tested, which was 200.

TABLE 3.

A-factor assay

Strain	A-factor (U)a
DK1622 (wild type)	11.5 ± 0.1
DK5057 (asgA mutant)	3.3 ± 0.1
sigD4403 (sigD mutant)	5.2 ± 0.2

^aA-factor released from starved cells was measured as described in Materials and Methods. Numbers are the average and standard deviation of three experiments.

σ^D is required for expression of C-signal-dependent genes, but for some genes the requirement is not cell autonomous.

Five genes that depend not only on A-signaling (40) but also on C-signaling (36, 46) failed to be expressed normally during development in the sigD null mutant (Fig. 5). Expression of two, dev (37, 38, 71) and Ω7536 (46), is important for development and was not rescued by codeveloping the sigD mutant with wild-type cells (Fig. 5). Surprisingly, expression of the other three, Ω4400, Ω4403, and Ω4499, increased significantly in the sigD mutant upon codevelopment with wild-type cells (Fig. 5). These results indicate that σ^D is needed in a cell autonomous fashion for expression of some C-signal-dependent genes (dev and Ω7536), but others (Ω4400, Ω4403, and Ω4499) do not require σ^D within the cell for their expression if extracellular signals are supplied by codeveloping wild-type cells.

FIG. 5.

Expression of C-signal-dependent genes in a sigD mutant. β-Galactosidase specific activity during development was measured for lacZ fusions to dev (A), Ω7536 (B), Ω4400 (C), Ω4403 (D), and Ω4499 (E) in wild type (▪) or in a sigD mutant alone (•) or upon codevelopment of the fusion-containing sigD mutant with an equal number of fusionless wild-type (○) or csgA mutant (▵) cells.

Expression of at least five essential developmental genes (csgA, sdeK, fruA, dev, and Ω7536) depends on σ^D cell autonomously (Fig. 2, 4, and 5), so it is not surprising that wild-type cells failed to rescue sporulation of the sigD null mutant (data not shown). On the other hand, the sigD mutant was capable of rescuing a csgA mutant that does not make C-signal, although the number of csgA-derived spores was about eightfold less than the number in mixtures with wild-type cells (Table 2). A reduced level of C-signal from the sigD mutant is expected because csgA expression is reduced (Fig. 2). It is worth noting that even in the mixture with wild-type cells, all the spores were derived from the csgA mutant (Table 2). This behavior, called developmental cheating, in which the mutant sporulates disproportionately, has been observed for the csgA mutant previously (75).

Rescue of gene expression in the sigD mutant depends on csgA and fruA.

To determine whether the rescue of expression of certain C-signal-dependent genes in the sigD mutant upon codevelopment with wild-type cells (Fig. 5) depends on the normal C-signal transduction pathway, we performed two experiments. First, we substituted a csgA mutant for wild type in the codevelopment experiments with the sigD mutants containing the Ω4400, Ω4403, or Ω4499 reporter. There was no rescue of gene expression (Fig. 5, triangles), demonstrating that rescue depends on csgA, which is needed to produce C-signal (34, 62). In the second experiment, we constructed sigD fruA double mutants bearing the Ω4400, Ω4403, or Ω4499 reporter. FruA mediates responses to C-signal during development (14, 52). The fruA mutation impaired rescue of Ω4400 expression and prevented rescue of Ω4403 and Ω4499 expression when the double mutants were codeveloped with wild-type cells (Fig. 6; note the changes in the values along the y axes compared to Fig. 5). Taken together, the results suggest that the sigD mutant responds to C-signaling by wild-type cells (Fig. 5) via the normal pathway involving FruA. Expression of Ω4403 and Ω4499 relies completely on this pathway, and so does expression of Ω4400 for the most part, but the latter appears to be induced very weakly by another mechanism in the sigD fruA double mutant upon codevelopment with wild-type cells (Fig. 6).

FIG. 6.

Rescue of Ω4400, Ω4403, and Ω4499 expression in the sigD mutant depends on fruA. β-Galactosidase specific activity during development was measured for lacZ fusions to Ω4400 (A), Ω4403 (B), and Ω4499 (C) in the sigD fruA double mutant alone (•) or upon codevelopment of the fusion-containing sigD fruA mutant with an equal number of fusionless wild-type cells (○).

Inability to rescue dev expression in the sigD mutant is not due to a requirement for a high level of C-signaling.

Why was expression of only certain C-signal-dependent genes rescued in the sigD null mutant upon codevelopment with wild-type cells (Fig. 5)? Perhaps C-signaling in the mixture is suboptimal, and certain genes require a higher level of C-signaling than others in order to be induced. To test this hypothesis, we made use of our observations that csgA expression (Fig. 2) and C-signaling (Table 2) are reduced in the sigD mutant. The sigD mutant was used as a suboptimal donor of C-signal in mixing experiments with csgA mutants bearing reporters. Contrary to the hypothesis, we found that the sigD mutant partially rescued expression of dev, but not Ω4403, in the csgA mutant background (Fig. 7). Partial rescue of dev expression, which is important for sporulation, is consistent with the finding that the sigD mutant partially restored sporulation of the csgA mutant upon codevelopment (Table 2). Expression of Ω4403 is not required for sporulation (38). The failure of the sigD mutant to rescue expression of Ω4403 in the csgA mutant background suggests that Ω4403 expression requires a higher level of C-signaling than dev expression. Wild-type cells can supply sufficient C-signal to rescue expression of Ω4403 in the sigD mutant background, but even the high level of C-signal supplied by wild type is insufficient to rescue dev expression in sigD mutant cells (Fig. 5). Taken together, our results suggest that developing cells require σ^D RNA polymerase or a gene under its control to increase transcription of dev in response to C-signal.

FIG. 7.

Rescue of dev but not Ω4403 expression in a csgA mutant upon codevelopment with the sigD mutant. β-Galactosidase specific activity during development was measured for lacZ fusions to dev (A) and Ω4403 (B) in a csgA mutant upon codevelopment with an equal number of fusionless wild-type (○), sigD mutant (□), or csgA mutant (▵) cells.

The level of sigD transcript drops rapidly during starvation, and sigD transcript levels are altered in a relA mutant during growth and starvation.

The foregoing results demonstrate that a sigD mutant responds to starvation by synthesizing ppGpp normally but is severely defective for developmental gene expression, including expression of early, A-signal-independent genes such as csgA and sdeK. Expression of these genes also depends on the relA gene, whose product synthesizes (p)ppGpp in response to starvation (10, 23). Does expression of sigD depend on relA? To investigate this question, we compared sigD transcript levels in wild-type and relA mutant cells during growth and shortly after starvation. The level of sigD transcript in growing cells was greatly decreased in the relA mutant compared to wild type (Table 4, time zero). This suggests that the low level of (p)ppGpp synthesized by RelA in growing cells (Fig. 3) positively regulates sigD transcription. Upon starvation, the level of sigD transcript decreased more than 100-fold in wild type by 20 min and remained low after 40 min (Table 4). The sigD transcript level decreased to a similar low level in the relA mutant after 20 min of starvation but then rose to an intermediate level by 40 min. The intermediate level was about 10-fold lower than in growing wild-type cells but about 10-fold higher than in wild-type cells that had been starving for 40 min. The latter observation suggests that RelA-dependent synthesis of (p)ppGpp in starving wild-type cells (Fig. 3) negatively regulates transcription of sigD. Taken together, these results suggest that σ^D synthesized during growth plays a critical role early in development.

TABLE 4.

sigD transcript levels

Time (min)a	sigD transcript level (%)b
	DK101 (wild type)	MS1000 (ΔrelA1)
0	100 ± 28	4 ± 4
20	0.3 ± 0.2	0.3 ± 0.1
40	0.7 ± 0.2	12 ± 3

^aAmount of time cells were in starvation buffer, as described in Materials and Methods.

^bNumbers indicate the average and standard deviation of three biological replicates, expressed as a percentage of the DK101 vegetative (time 0) value, for which six biological replicates were performed.

DISCUSSION

Our results provide several new insights into the role of σ^D in regulating genes and signals during M. xanthus development. σ^D is required very early during development for full expression of key genes such as csgA and sdeK that depend on the intracellular signal (p)ppGpp but not the extracellular A-signal. This requirement for σ^D does not appear to reflect a role in the stringent response to starvation, because a sigD null mutant accumulates (p)ppGpp normally when starved. On the other hand, the low level of (p)ppGpp synthesized in growing cells appears to be necessary for sigD transcription, and σ^D made during growth appears to fulfill its early role in development. σ^D is required both for normal production of A-signal and for cellular responses to A-signal. Supplying A-signal by codeveloping the sigD mutant with wild-type cells did not restore normal expression of A-signal-dependent genes in the sigD mutant. Another insight was quite surprising in light of the requirement of σ^D for so many early events in development; we discovered that the sigD mutant can respond to C-signal from codeveloping wild-type cells by inducing a subset of late developmental genes. This subset of genes apparently does not require σ^D RNA polymerase for expression. On the other hand, a second class of C-signal-dependent genes, including the dev operon that is important for sporulation, was not induced. In the case of the dev operon, this is not due to a requirement for a higher level of C-signaling, but rather a cell autonomous requirement for σ^D. Below, we discuss the implications of these new insights and present a model that depicts the role of σ^D in regulating genes and signals during M. xanthus development.

Role of σ^D early in development.

Fig. 8 shows the position of σ^D near the top of a regulatory hierarchy built from the results of this study and others cited below. Accumulation of sigD transcripts in growing cells depends strongly on relA (Table 4). In E. coli, RelA is responsible for ribosome-dependent synthesis of (p)ppGpp (reviewed in reference 7). The M. xanthus relA gene complements an E. coli relA mutant for production of ppGpp, suggesting that it has a similar function (23). Only a low level of ppGpp is detected in growing M. xanthus (49). Presumably, this depends on RelA activity, and the resulting low concentration of (p)ppGpp positively regulates sigD transcript accumulation. We suggest that (p)ppGpp positively regulates sigD transcription, by analogy with the effect of (p)ppGpp on rpoS transcription in E. coli, which appears to be at the level of elongation rather than initiation (44). Also, (p)ppGpp affects transcription of many genes in E. coli (reviewed in references 51 and 54). Alternatively, (p)ppGpp might influence sigD mRNA stability.

FIG. 8.

Model of regulatory events governing M. xanthus development. Arrows indicate positive regulation, which may be direct or indirect. The dashed arrows indicate regulatory inputs that this study suggests might be different for dev than for Ω4400, Ω4403, and Ω4499.

The level of sigD transcripts in wild-type cells decreased more than 100-fold after 20 min of starvation (Table 4). Starvation induces (p)ppGpp synthesis in M. xanthus (49, 50), and this depends on RelA (23) but not on σ^D (Fig. 3 and 8). The resulting high concentration of (p)ppGpp appears to be at least partly responsible for negatively regulating sigD transcript accumulation after starvation, as depicted in Fig. 8, because the sigD transcript level was higher in relA mutant cells than in wild-type cells after 40 min of starvation (Table 4). Whether (p)ppGpp alone negatively regulates sigD transcript accumulation or works in combination with other starvation-induced signals is an open question, as is the mechanism(s) of regulation. Previously, sigD expression was investigated with transcriptional and translational fusions to lacZ (73). β-Galactosidase activity was observed during growth and increased about twofold as cells entered stationary phase or were placed under conditions that induce development. The stability of β-galactosidase, the longer times prior to sampling, and/or a difference in the starvation conditions might explain why a decrease in sigD expression was not detected previously. It will be interesting to measure sigD transcript and σ^D protein levels under different starvation conditions and after longer periods of starvation. The rapid drop in sigD transcript levels after starvation suggests that σ^D made during growth fulfills its early role in development.

σ^D is needed for full expression of csgA and sdeK (Fig. 2) and to produce the wild-type level of A-signal (Tables 2 and 3). As depicted in Fig. 8, induction of (p)ppGpp is also required for these events (10, 23). This could mean that (p)ppGpp and σ^D RNA polymerase affect the same step(s) very early in development. A simple model would be that (p)ppGpp stimulates σ^D RNA polymerase to transcribe one or more genes, analogous to stimulation of σ^S RNA polymerase activity by (p)ppGpp in stationary phase E. coli (43). However, based on studies in E. coli, the effects of (p)ppGpp on RNA polymerase are likely to be complex (reviewed in references 51 and 54). We think it is unlikely that σ^D RNA polymerase transcribes sdeK, because the promoter of this gene has sequences in the −24 and −12 regions similar to the consensus for σ⁵⁴ (18). Since σ⁵⁴ RNA polymerase requires an activator protein to initiate transcription, perhaps σ^D RNA polymerase transcribes the gene encoding the activator. Likewise, it seems likely that σ^D RNA polymerase transcribes an activator of csgA, rather than directly transcribing csgA, for several reasons. First, the csgA promoter has sequences centered at −35 and −10 similar to the consensus for σ^A (3, 45). Second, deletion analysis of the csgA promoter region suggests the involvement of one or more activator proteins (45). Third, csgA is expressed, albeit poorly, in the sigD null mutant (Fig. 2). With respect to the role of σ^D in producing A-signal, at least five genes are known to be involved in A-signal production (8, 17, 39), but transcription of only one, asgE, has been studied in detail (16). asgE appears to be transcribed from a promoter similar to the σ⁵⁴ consensus during growth and from two promoters during development. One of the putative developmental promoters is similar to the σ⁵⁴ consensus, but the other is not similar to the σ⁵⁴ or σ^A consensus; so it is possible that σ^D RNA polymerase transcribes asgE from this promoter.

Our investigation also revealed that σ^D is important for cellular responses to A-signal (Fig. 4), which include induction of Ω4521, fruA, and Ω4514 early in development. Ω4521 is very likely to be transcribed by σ⁵⁴ RNA polymerase, based on mutational analysis of its promoter (33). Also, an activator protein appears to bind somewhere between −146 and −90 relative to the transcriptional start site (20). A candidate for the activator protein is SasR, an NtrC-like response regulator identified in a genetic screen as a positive regulator of Ω4521 (21). Transcription of sasR and/or several other genes involved in Ω4521 regulation (21) might be the direct target(s) of σ^D RNA polymerase activity that explain the failure of the sigD mutant to induce Ω4521 expression upon codevelopment with wild-type cells that supply A-signal (Fig. 4). Transcription of fruA appears to be regulated differently than that of Ω4521. It involves MrpC, a cyclic AMP receptor protein-like activator (70), which binds upstream of the fruA promoter (72). Expression of mrpC is reduced in the absence of A-signaling (69), which might explain the dependence of fruA expression on A-signaling (14, 52). Since expression of mrpC depends on MrpB, an NtrC-like response regulator (70), transcription of the mrpAB operon might be the direct target of σ^D RNA polymerase that explains the cell autonomous requirement of σ^D for fruA induction in response to A-signal. Alternatively or in addition, the fruA promoter might be a target of σ^D RNA polymerase activity, since mutational analysis indicates that the −35 and −10 regions are important (65), but the sequences do not match the σ^A consensus very well. In contrast, the promoter of the Ω4514 operon is similar to the σ^A consensus and can be transcribed in vitro by σ^A RNA polymerase (22). Multiple upstream DNA elements are necessary for full induction of the Ω4514 operon, suggesting the involvement or one or more transcriptional activator proteins, whose transcription might be the direct target(s) of σ^D RNA polymerase. Identification of the direct targets of σ^D RNA polymerase activity is an important future goal in order to better understand the role of σ^D early in M. xanthus development.

Role of σ^D late in development.

Given the impaired expression of csgA (Fig. 2), whose product appears to be the C-signal (34, 47, 62), and fruA (Fig. 4), whose product mediates responses to C-signal (14, 52), it is not surprising that the sigD mutant failed to express all five C-signal-dependent genes tested (Fig. 5). Moreover, expression of three of these, Ω4400, Ω4403, and dev, has been shown to be affected by a mutation in sdeK (56) (Ω4499 and Ω7536 were not tested), and sdeK expression is impaired in the sigD mutant (Fig. 2).

Despite the impaired early developmental gene expression, the sigD mutant expressed Ω4403 normally, Ω4499 in a delayed fashion, and Ω4400 partially, upon codevelopment with wild-type cells (Fig. 5). This rescue depends on the normal C-signal transduction pathway, since it required the donor cells to be csgA⁺ (Fig. 5) and the recipient sigD mutant cells to be fruA⁺ (Fig. 6). We infer that sigD mutant cells synthesize enough FruA to mediate some responses to C-signal supplied by wild-type cells during codevelopment. The amount of FruA in the sigD mutant is likely to be low, since very little expression of a fruA-lacZ fusion was detected (Fig. 4). We also infer that a small amount of SdeK is made in the sigD mutant, since sdeK-lacZ expression is low (Fig. 2). Apparently, this amount is sufficient for expression of Ω4400, Ω4403, and Ω4499, and σ^D RNA polymerase is dispensable for transcription of these genes (Fig. 5).

No rescue of dev expression in the sigD mutant was observed upon codevelopment with wild-type cells (Fig. 5). This is not due to a requirement for a higher level of C-signaling, since dev, but not W4403, was expressed in csgA mutant cells codeveloped with sigD mutant cells (serving as donor of a low level of C-signal) (Fig. 7). Rather, the developmental increase in dev transcription appears to require σ^D RNA polymerase or a gene under its control for a regulatory input distinct from that required by Ω4400, Ω4403, and Ω4499. Several possibilities are depicted by dashed lines in Fig. 8. One possibility is that dev transcription requires a higher level of the SdeK histidine kinase and/or a downstream component in the SdeK signal transduction pathway. Another possibility is that dev transcription requires a higher level of phosphorylated FruA (FruA-P). The idea that a lower level of FruA-P might suffice for expression of C-signal-independent genes and a higher level might be needed for expression of C-signal-dependent genes has been proposed previously (26, 72). Likewise, different levels of FruA-P might be necessary to activate different C-signal-dependent genes. A third intriguing possibility is that σ^D RNA polymerase transcribes dev, or that an unknown activator protein under σ^D control (Fig. 8, question mark) is needed for dev transcription. In any case, our analysis of late developmental gene expression in the sigD mutant separated C-signal-dependent genes into two distinct regulatory classes. The Ω7536 operon is in the class with the dev operon, as expected, because expression of Ω7536 depends on dev (46).

Role of σ^D in M. xanthus development compared with that of σ factors in other bacteria that form biofilms.

Like σ^S in E. coli, M. xanthus σ^D is important for cell viability in stationary phase (73). Also, the amino acid sequence of σ^D region 2 is similar to region 2 of stationary-phase σ factors (73). However, whereas σ^D clearly plays an overall positive role in M. xanthus development, the overall role of σ^S in E. coli biofilm formation can be positive (1) or negative (9), depending on the experimental conditions. This probably reflects the involvement of σ^S in a complex regulatory network governing E. coli biofilm formation (57).

σ^D also mediates the responses of M. xanthus to osmotic and oxidative stress (73). Osmotic stress induced biofilm formation by a Staphylococcus aureus mucosal isolate, and σ^B was essential for the response (58); however, in another study, σ^B was not essential for S. aureus biofilm formation (74). Here again, a complex regulatory network is involved (27). In Staphylococcus epidermidis, σ^B plays an overall positive role in biofilm formation, but this appears to involve production of a repressor of icaR, which encodes a negative regulator of genes required for biofilm formation (35). Likewise, in Bacillus subtilis, σ^H regulates many genes during the transition from stationary phase to sporulation or biofilm formation via Spo0A-mediated removal of AbrB repression (5, 15, 66, 67). It would not be surprising if the overall positive role of σ^D in M. xanthus development also involves removal of repression, as this appears to be a common theme in the regulation of genes required for biofilm formation. In any case, our results demonstrate that σ^D plays a key role in the complex regulatory network governing M. xanthus development.