|Home | About | Journals | Submit | Contact Us | Français|
In a developmental strategy designed to efficiently exploit and colonize sparse oligotrophic environments, Caulobacter crescentus cells divide asymmetrically, yielding a motile swarmer cell and a sessile stalked cell. After a relatively fixed time period under typical culture conditions, the swarmer cell differentiates into a replicative stalked cell. Since differentiation into the stalked cell type is irreversible, it is likely that environmental factors such as the availability of essential nutrients would influence the timing of the decision to abandon motility and adopt a sessile lifestyle. We measured two different parameters in nutrient-limited chemostat cultures, biomass concentration and the ratio of nonstalked to stalked cells, over a range of flow rates and found that nitrogen limitation significantly extended the swarmer cell life span. The transcriptional profiling experiments described here generate the first comprehensive picture of the global regulatory strategies used by an oligotroph when confronted with an environment where key macronutrients are sparse. The pattern of regulated gene expression in nitrogen- and carbon-limited cells shares some features in common with most copiotrophic organisms, but critical differences suggest that Caulobacter, and perhaps other oligotrophs, have evolved regulatory strategies to deal distinctly with their natural environments. We hypothesize that nitrogen limitation extends the swarmer cell lifetime by delaying the onset of a sequence of differentiation events, which when initiated by the correct combination of external environmental cues, sets the swarmer cell on a path to differentiate into a stalked cell within a fixed time period.
Caulobacter crescentus, an aquatic oligotrophic alphaproteobacterium, undergoes an asymmetric cell division, generating two distinct daughter cells, a motile swarmer cell, possessing a single polar flagellum, and a nonmotile, stalk-bearing cell (1, 12, 25) (Fig. (Fig.1).1). The progeny swarmer and stalked cell types differ in their global programs of gene expression and their capacity to reinitiate chromosomal DNA replication. Stalked cell progeny reinitiate chromosome replication almost immediately following cell division (40, 41). DNA replication in swarmer cells is repressed for a defined period of time (40, 41), during which the swarmer cells are motile and chemotactically competent. The swarmer cell phase ends with a loss of the flagellum and chemotaxis apparatus, synthesis of the hold-fast and stalk, and the initiation of DNA replication (Fig. (Fig.11).
In Caulobacter, asymmetry is established through the differential polar localization of two-component signaling proteins (22, 32, 44). Phosphotransfer signaling events catalyzed by these proteins control core cell cycle regulators such as the global transcription factor CtrA. The phosphorylated form of this essential response regulator has a critical role in repressing the initiation of DNA replication in the swarmer cell. CtrA is degraded by a programmed proteolysis that commences upon the onset of differentiation into a stalked cell, thus relieving the imposed block on the initiation of DNA replication (Fig. (Fig.1)1) (2, 16, 17, 51, 52). As the cell cycle progresses, CtrA is resynthesized and phosphorylated (8, 19, 49, 50) where, in the predivisional cell, it serves to regulate the transcription of a large number genes, including those required for stalk and flagellar biogenesis (23, 24) (Fig. (Fig.1).1). Experimental evidence, as well as computational analysis, demonstrates that this cell cycle engine operates as a closed loop that self-adjusts for transient changes in the levels of critical regulatory proteins and reaction rates (2, 54), suggesting that key events in the cell cycle appear to be little influenced by changes in the environment.
The obligatory asymmetric cell division exhibited by Caulobacter is a regulatory strategy designed to cope with a scarcity of critical nutrients in an oligotrophic environment (46). The motile and chemotactic swarmer cell can be viewed as a cell type committed to foraging for nutrients. Under typical culture conditions, the swarmer cell phase last for a relatively invariant period of time, appearing to be enslaved by an internal timing mechanism. However, since differentiation from swarmer cell into the sessile stalked cell is irreversible, it is likely that environmental factors such as the availability of sufficient nutrients for cell reproduction would be a major influence in governing the decision to abandon motility and adopt a sessile lifestyle. Caulobacter cells are found in freshwater environments often containing micromolar concentrations of macronutrients such as carbon, nitrogen, and phosphorus. In situ measurements of growth rates in natural settings indicate that Caulobacter cells can divide several times a day depending on the physical conditions (47). Since these cells are so well adapted to life in oligotrophic environments, this presents a major experimental hurdle in understanding how environmental cues regulate swarmer cell differentiation and the cell cycle. Previous experiments have shown that suspending isolated swarmer cells in medium lacking either a carbon or nitrogen source blocks differentiation into a stalked cell (7, 16, 27). These types of experiments, however, do not mimic conditions that actively growing Caulobacter cells would encounter in the environment. An alternative strategy could involve decreasing the concentration of key nutrients in the culture medium to levels (i.e., submicromolar) where they would be limiting for an oligotrophic organism. Again, this is not experimentally feasible, since the low quantities of nutrients present would be almost instantaneously consumed by the cells, thus resulting in a cessation of growth. Open, continuous flow culture systems such as nutrient-limited chemostats offer an experimental advantage over batch cultures when examining the effect of nutrient limitation on cell physiology and growth rate (35, 39). These culture systems, in which the increase in culture volume by addition of medium containing the limiting nutrient(s) is offset by overflow, contain extremely low levels of the limiting nutrient. The culture can be maintained in a state of perpetual growth at fixed generation times that can be experimentally manipulated by adjusting the rate of medium flow.
In the experiments described here, we tested the effect of imposed nitrogen or carbon limitation on gene expression and population dynamics of slowly growing populations of Caulobacter cells in continuous chemostat cultures. These continuous cultures were operated with imposed generation times comparable to measured in situ growth rates. Our experiments indicate that nitrogen availability, but not carbon limitation, results in an extension of the swarmer cell life span, suggesting that relative nutrient abundance influences core cell cycle regulatory networks, which, in turn, delays the initiation of an internal swarmer cell differentiation pathway.
Transcriptional profiling experiments showed that at least 62 transcripts were induced in Caulobacter cells grown under ammonium limitation. In many bacteria, changes in gene expression in response to nitrogen limitation requires RNA polymerase containing σ54, encoded by the rpoN gene (26, 38). Interestingly, the induction of most of the mRNA transcripts in Caulobacter cells grown under nitrogen-limitation did not require this highly conserved nitrogen regulatory pathway. Surprisingly, rpoN was required for repression of ribosomal protein gene expression under nitrogen limitation, suggesting that conserved rpoN-dependent nitrogen regulatory circuits in oligotrophs may be designed to globally regulate transcription and translation in response to relative nitrogen abundance. Growth under carbon limitation resulted in the induction of a number of transcripts predicted to be involved in acquiring carbon from plant-derived sources such as lignin and terpenes. This pattern of regulated gene expression is in striking contrast to most copiotrophic organisms which induce genes required for the utilization of a distinctly different spectrum of carbon sources (i.e., simple sugars, starches, and fatty acids). These observations suggest that bacteria optimally compete for resources through evolved genetic networks that respond uniquely to their natural environments.
The wild-type C. crescentus strain used in the present study is a synchronizable derivative (NA1000) of CB15. This strain does not produce holdfast material and thus does not adhere to the walls or other structures in the continuous culture vessel. Mutant strains used were all isogenic derivatives of NA1000. PV415 contains a chromosomally integrated YFP-CpaE fusion (58). SC1055 contains a Tn5 insertion in rpoN (6) and JG3403 possesses a deletion in ntrY (CC1742). In order to construct JG3403, a 7.3-kb EcoRI-XhoI fragment containing the entire C. crescentus ntr operon was generated by PCR using Pfx platinum DNA polymerase (Invitrogen) was subcloned into pBluescript KS, generating pJEZ20. A 3.7-kb HindIII-XhoI fragment from this plasmid, which contains ntrC, ntrY, and ntrX was subcloned into the sacB counterselection vector, pNPTS139 (M. R. K. Alley, unpublished data) to create pJEZ21. pJEZ21 was digested with SalI and then religated to create a 400-bp internal deletion in ntrY (pJEZ22). pJEZ22 was introduced into NA1000 cells by conjugation and the wild-type ntrY gene was replaced by the deleted allele using sacB counterselection as described previously (13). The deletion was verified by PCR (data not shown). Transcriptional reporter fusions were constructed in placZ/290 (15) using approximately 900-bp, PCR-generated fragments containing the upstream untranslated region extending 3′ into the coding regions of the glnB, glnK, and gltD operons. These plasmids were introduced into C. crescentus cells by conjugation using Escherichia coli helper strain S17-1 (53).
Caulobacter cells were grown at 31°C in peptone-yeast extract (PYE) (46) or M2 minimal medium with glucose (21) either alone or supplemented with one or more of the following: kanamycin (25 μg ml−1), tetracycline (2.0 μg ml−1), and nalidixic acid (20 μg ml−1). E. coli strains were grown at 37°C in Luria-Bertani (LB) medium (34) either alone or supplemented with kanamycin (50 μg ml−1) or tetracycline (12.5 μg ml−1).
Cells were grown in BioFlo (New Brunswick) continuous culture devices outfitted with Masterflex L/S (Cole-Parmer) peristaltic pumps. The medium in the culture vessel (350 ml) was agitated with a mechanical impeller at 300 rpm, aerated, and maintained at a growth temperature of 30°C. The base growth medium was a modified minimal M2 medium (21). For ammonium-limited cultures, the NH4Cl in the medium was decreased from 9.3 mM to 1.86 mM. In batch culture, with this concentration of NH4Cl, the yield of cells is limited by available nitrogen, reaching an optical density at 600 nm (OD600) of 0.5 (compared to 1.85 in medium containing 9.3 mM NH4Cl). This same nitrogen concentration (1.86 mM) was used when either nitrate or glutamate were used as nitrogen sources. For glucose-limited cultures, the final glucose concentration in the medium was decreased from 11 to 2.2 mM. In batch culture, with this concentration of glucose, the yield of cells is limited by available glucose to an OD600 of 0.5 versus 1.85 in medium containing 11 mM glucose. In order to prevent precipitated calcium phosphate from blocking the pump tubing, the final CaCl2 concentration was decreased from 0.5 to 0.1 mM. This lower CaCl2 concentration had no effect on the final yield of cells in batch cultures (data not shown). The cultures were started by filling the culture vessel with medium and inoculating them with 5 ml of an overnight (18-h) culture of cells grown in complex PYE medium. All measurements (OD, sampling for microscopy, gene expression, etc.) were conducted on cultures that had reached a steady state (i.e., operated for at least for 7 to 10 complete culture volume changes following flow rate adjustments). In order to visualize stalks, cells were harvested from the chemostat culture vessel and allowed to dry on glass slides. The dried cell sample was stained with a simple mordant from a modified flagellar stain containing 4% tannic acid (final concentration) added to a saturated solution of KAl(SO4)2-mercuric chloride (2.5:1) (14). The mordant was applied to the dried cell sample and allowed to incubate at room temperature for 6 min, the slides were washed with water, dried, and examined by bright-field microscopy. The microscopic images were captured with a video camera, and the proportion of nonstalked cells in the population was determined by counting 1,000 to 2,700 cells for each time point from three different cultures. As an alternative method for determining the number of cells in the swarmer cell phase, NA1000 cells containing a YFP-CpaE fusion grown in nutrient-limited chemostats were harvested, mixed with a fluorescent membrane stain (FM4-64; 0.1 mg/ml), placed on polylysine-coated slides, and examined for polar YFP fluorescence. The percentage of cells possessing polar YFP-CpaE foci to those where no foci were visible was determined on captured images by counting 500 to 1,000 cells for each time point from three different cultures. The concentration of ammonium and glucose in chemostat culture supernatants was determined enzymatically using commercially available kits (Roche). For the experiments presented here, at all flow rates tested, both ammonium and glucose concentrations were at, or below, the detectable limits (approximately 3.5 × 10−6 M for ammonium and 2 × 10−6 M for glucose).
Expression of lacZ transcriptional reporters in unsynchronized cultures was measured as previously described (30). All quantitative measurements of β-galactosidase activity were determined in triplicate, on three separately grown cultures. Cell cycle experiments were performed as described previously (37). Immunoblots were preformed essentially as described in Towbin et al. (55) and were analyzed using anti-CtrA (50) and anti-flagellin (31) antisera.
In order to determine mRNA levels from chemostat-grown cells, the cells were harvested from the continuous culture vessel by using a syringe and dispensed into 1.5-ml aliquots in microfuge tubes. These were centrifuged for 45 s at 13,000 × g at room temperature, the supernatants were removed by aspiration, and the pellets were flash-frozen in liquid nitrogen. For quantitative real-time PCR (qRT-PCR), total RNA was isolated by using an RNeasy miniprep kit (Qiagen). cDNA was synthesized from 0.1 μg of total RNA by using random hexamer primers, and the synthesized cDNA was amplified (200-bp product for each assayed transcript) by using a SuperScript III Platinum qRT-PCR kit (Invitrogen). The measured threshold cycle (CT) values of the assayed transcripts were normalized by subtracting the CT for the transcript encoding the Caulobacter surface array protein (rsaA), which was also assayed at each indicated time point. The resulting ΔCT value derived from the fastest flow rate samples (i.e., 1.6 ml min−1) was used as a reference to calculate the relative level of transcript(s) for each time point (i.e., ΔΔCT).
For transcriptional profiling by microarray analysis, RNA was extracted by using RNeasy kits (Qiagen) according to the manufacturer's instructions and subsequently reverse transcribed into cDNA, as described previously (23). Labeled cDNAs were combined, mixed with Agilent hybridization buffer, and competitively hybridized to custom-designed Agilent microarrays according to the manufacturer's instructions (Agilent). The custom-designed arrays contain approximately 10 probes per gene, ratios obtained for probes corresponding to the same gene were averaged, and a single value was reported. Independent samples for rpoN::Tn5 cells grown in carbon-limited and nitrogen-limited conditions were also prepared for global transcriptional analysis by using custom, spotted 50-mer arrays, as described previously (3). A comparison of data from the two microarray platforms is provided in Table S1 in the supplemental material.
Motility in the Caulobacter swarmer cell phase functions in the dispersal of progeny cells, thus limiting competition for essential nutrients among sibling cells. Environmental conditions such as the availability of key nutrients may therefore be expected to regulate the swarmer to stalked cell transition. One hallmark of this differentiation sequence is the proteolysis of flagellar components such as flagellin proteins and an accompanying loss of the polar flagellum. We found that synchronized swarmer cells suspended in nitrogen- or carbon-free medium retained both the 27- and 25-kDa flagellins indicative of a block in the swarmer cell differentiation (Fig. (Fig.2).2). However, CtrA proteolysis, which also occurs upon swarmer cell differentiation, was readily observed after 20 min of incubation whether or not the medium contained carbon or nitrogen (Fig. (Fig.2).2). These results are similar to those of previous experiments (16, 27) that showed a modest, but significant, immediate reduction (ca. 30%) in CtrA pools when swarmer cells were suspended in medium lacking either ammonium or glucose. On the whole, these observations suggest that the absence of nitrogen or carbon in the medium blocks some key events in swarmer cell differentiation such as proteolysis of the flagellins, but not others, as indicated by the correct timing of CtrA proteolysis.
Experiments such as those described above do not adequately address whether growth under nutrient limitation influences development since suspending cells in medium lacking a macronutrient such as carbon or nitrogen is likely to elicit a starvation response. After growth in conventional minimal or complex medium, isolated swarmer cells differentiate into stalked cells within approximately 30 min, a period that is apparently governed by an internal timing mechanism. We examined the growth dynamics in continuous culture (chemostat) under conditions of either carbon or nitrogen limitation to determine whether the start of this internal timer could be regulated by nutrient availability in growing cultures.
One condition for the maintenance of a steady-state biomass concentration in a chemostat is the assumption that each progeny cell arising from a given cell division event possesses the capacity to complete a subsequent cell division within the same period of time (Fig. (Fig.3A)3A) (35, 39). Simply stated, the reproductive rate will be constant for all progeny at a fixed imposed flow rate of medium containing the limiting nutrient. Although this is true for the majority of bacteria that have been cultivated in a chemostat, it is not the case for Caulobacter (48). Since the newborn swarmer cell possesses an extended G1 period compared to the progeny stalked cell, it cannot reinitiate the cell cycle immediately after division. These differing progeny each have their own unique cell cycle, with the progeny swarmer cell cycle being ~1.33 times longer than the progeny stalked cell cycle in nutrient replete medium (Fig. (Fig.3A)3A) (48). Therefore, the likelihood that a swarmer cell will exit the culture vessel before it has divided into progeny cells will be a function of the flow rate and the time required to differentiate into a reproductive form (i.e., a stalked cell). We exploited this phenomenon to address how nutrient limitation influences the lifetime of the swarmer cell in an actively growing population. We hypothesized that at imposed growth rates approaching the maximal possible doubling time, we would observe an accompanying decline in the steady-state biomass or wash-out as the medium flow rate increases. The flow rate(s) at which this loss of biomass is observed should have a direct relationship to the length of the swarmer cell cycle under specific nutrient limitation: the longer the swarmer cell cycle, the greater the probability that washout will occur at lower dilution rates. In order to assess changes in the composition of the cell population in the continuous culture vessel, we used a Caulobacter strain (NA1000) that does not synthesize a holdfast. This was necessary to ensure that stalked cells, like swarmer cells, did not adhere to the culture vessel and could be washed out by medium flow.
When cultivated under carbon limitation (Fig. (Fig.3B),3B), the concentration of cells in the culture vessel exhibited two distinct steady-state phases when an increasing dilution rate was imposed. In the first phase, at imposed doubling times (tD) ranging from 18 to 3.5 h (equivalent to a flow rate ranging from 0.05 to 0.2 h−1), the biomass remained at a relatively constant concentration, a behavior typical of most species of bacteria when grown under chemostat conditions. However, when the dilution rate was further increased, there was a steady decline in the biomass. At each higher dilution rate there was a proportional decrease in the steady-state concentration of cells in the vessel. This result is not unexpected since at higher flow rates there exists a statistically higher probability that the nonreproducing swarmer cells will exit the growth vessel before generating progeny cells, since their cell cycle should be at least 1.33 times longer than that of stalked cells. Therefore, our data indicate that carbon limitation does not significantly influence the timing of swarmer cell differentiation under these culture conditions since biomass loss was observed only at generation times approaching those of the theoretical progeny swarmer cell maximal doubling time (equivalent to a tD = 2 h).
When Caulobacter cells were cultured under ammonium limitation, a distinctly different relationship between biomass and dilution rate was observed (Fig. (Fig.4A).4A). At imposed dilution rates that were very low (corresponding to a tD > 13 h), the biomass remained relatively high (OD600 of 0.35 to 0.48) (phase I). In contrast, when the dilution rate was raised to greater than 0.05 h−1, the biomass rapidly decreased in concentration. We describe this as a “quasi-washout,” since it was incomplete, with the reduced biomass concentration exhibiting an apparent new steady state (phase II). The quasi-washout occurred with a sharp transition between flow rates that were equivalent to tD = 11 to 13 h and was attributable to a decrease in the absolute cell number since the culture contained 4.3 ± 0.7 × 108 cells ml−1 in phase I (OD600 = 0.47) and 1.9 ± 0.4 × 108 cells ml−1 in phase II (OD600 = 0.2). This new steady state (phase II) in cell concentration was constant over a wide range of flow rates (0.06 to 0.3 h−1). When the flow rate was again decreased from these relatively high rates to less than 0.05 h−1, the biomass concentration in the culture vessel increased back to the high phase I level (Fig. (Fig.4B).4B). This phenomenon, with the culture exhibiting high cell densities at low flow rates and lower cell densities at high flow rates, was reproducible and could be observed many times in the same culture by sequentially raising and then slowing the imposed flow rate (Fig. (Fig.4B).4B). This indicates that the two different steady-state culture phases observed here are not likely a consequence of mutations arising in the population. In addition, this effect was also observed when glutamate or nitrate was used as the limiting nitrogen source (not shown).
One possibility is that the quasi-washout of cells observed here under higher flow rates was attributable to the loss of nonreproducing swarmer cells, the implication being that under conditions of low concentrations of available ammonium, swarmer cell differentiation required an extended period of time. In order to test this idea, we determined the relative proportion of swarmer and stalked cells in the continuous culture vessel over a range of different flow rates. We utilized two different methods to enumerate swarmer and stalked cell types. First, the proportion of nonstalked cells in the population was determined by light microscopy of cells stained with a simple mordant (see Materials and Methods). At low flow rates of 0.02 and 0.034 h−1, the culture contained 43 and 36% swarmer cells, respectively (Fig. (Fig.5).5). When the flow rate was increased (0.05 h−1 or greater), the proportion of swarmer cells in culture decreased, ranging from 21% to as low as 10% swarmer cells. We also determined the relative proportion of swarmer cells by quantifying the number of cells with polarly localized YFP-CpaE foci. CpaE is a pilus assembly protein that transiently localizes to the pole of swarmer cells and disappears by the time the swarmer cell differentiates into a stalked cell (58). Cultures grown at a low flow rate (0.034 h−1) had proportionally more cells possessing polarly localized YFP-CpaE (25%) than did cultures maintained at higher flow rates, where the proportion of YFP-CpaE-containing cells ranged from 7 to 17% (Fig. (Fig.5).5). Thus, the loss of biomass observed when the culture flow rate was increased to ≥0.5 h−1 is likely attributable to a decrease in the relative number of swarmer cells. This loss of swarmer cells only occurred under ammonium limitation as there was no significant change in the number of swarmer cells in glucose-limited cultures (see Fig. S1 in the supplemental material).
These observations suggest that a maximum steady-state biomass concentration can only be maintained in ammonium-limited continuous cultures when the flow rate is correspondingly low, thus permitting a substantial fraction of swarmer cells in the culture vessel adequate time to reproduce. In log-phase, nutrient-replete batch cultures, the swarmer cell phase lasts for ca. 25% of the cell cycle (i.e., 30 min at tD = 2 h). If ammonium availability were not affecting the length of swarmer life span, the swarmer cell phase in a chemostat culture at an imposed tD = 13.86 h should be approximately 3.46 h (i.e., 25% of 13.86 h). However, since the number of swarmer cells in the culture vessel sharply decreases at this flow rate, the estimated swarmer cell life time must be at least, and is more than likely greater than, 13 h.
In typical chemostat cultures maintained under steady-state conditions, the absolute biomass concentration is dictated by the concentration of the limiting nutrient. As the limiting nutrient is immediately consumed by the biomass, its effective concentration is near zero. One consequence of the quasi-washout phenomenon observed here is that the biomass concentration in phase II (see Fig. Fig.4A)4A) is approximately one-half of what would be expected with the concentration of ammonium in the culture reservoir. Therefore, the ammonium concentration in the culture vessel at phase II may be greater than zero. In order to test this possibility, we assayed the ammonium concentration in continuous culture supernatants sampled over a range of differing flow rates. Interestingly, the ammonium concentration in the medium at all flow rates was less than can be accurately measured via an enzyme-based assay (~3.5 × 10−6 M) (data not shown), suggesting that Caulobacter cells, even when growing at low culture densities, can efficiently take up most of the available ammonium. One possibility is that the increased flux of available ammonia during phase II allows the doubling cell mass to offset its loss through dilution and overflow, resulting in a relatively lower biomass concentration over a wide range of increasing flow rates (i.e., from 0.05 to 0.3 h−1). To examine this possibility, we assayed the expression levels of glnK mRNA over a range of increasing flow rates. This conserved gene encodes a PII family protein, members of which sense cellular nitrogen status and control a diverse array of physiological processes (26). Importantly, experiments in E. coli have demonstrated that glnK is expressed to high levels in cells only under conditions of nitrogen deficiency (5). We found that glnK mRNA levels were highest under low flow rates (0.034 h−1) and declined sharply in cells cultured at increasing flow rates (Fig. (Fig.6).6). This finding is consistent with the idea that cells cultured at imposed higher flow rates under nitrogen limitation (i.e., at phase II, Fig. Fig.3)3) have greater available ammonium for growth.
Cellular nitrogen status in bacteria is sensed by a highly conserved signal transduction pathway (26, 38). The expression of genes involved in scavenging ammonium (as well as other nitrogen sources) from the environment such as glutamine synthetase (GlnA) and glutamate synthase (GltD) is increased when the cell senses a low-nitrogen environment. In general, the indicator of this state is the relative internal concentrations of 2-oxoglutarate and glutamine. The cellular concentration of these effectors is sensed through the covalent modification of a family of regulatory proteins known as PII (GlnB and GlnK) which, either directly or indirectly, influence the activity of the NtrB sensor histidine kinase, the ammonium transporter (AmtB), and glutamine synthetase (GlnA) (4, 9, 20, 45, 57). In enteric bacteria, NtrB regulates the expression of nitrogen acquisition genes by influencing the phosphorylation state of the response regulator, NtrC, a σ54 transcriptional activator. The Caulobacter genome contains a conserved gene cluster predicted to encode nitrogen regulatory proteins including NtrB, NtrC, and another two-component regulatory pair, the predicted NtrY sensor kinase, and NtrX, another σ54-dependent transcription factor (Fig. (Fig.7A).7A). These latter two proteins are likely to be involved in responding to nitrogen availability based on their conserved co-occurrence and proximity to ntrB and ntrC in other bacterial genomes. In order to determine whether these trans-acting factors influenced swarmer cell differentiation under nitrogen limitation, we examined the population dynamics of a ΔntrY and rpoN::Tn5 (encoding the σ54 subunit of RNA polymerase) (6) mutant strains in continuous cultures grown under nitrogen limitation. Both mutant strains grew and wild-type cells under imposed ammonium limitation. In addition, the response of both ΔntrY and rpoN::Tn5 mutant strains to increasing flow rate was similar to that of wild-type cells with the biomass concentration exhibiting a quasi-washout at flow rates of ≥0.05 h−1 (Fig. (Fig.7B7B and see Fig. S2 in the supplemental material). Our results suggest that these conserved nitrogen regulatory proteins are probably not involved in controlling swarmer cell differentiation and, surprisingly, are not required for growth under nitrogen limitation.
A strain containing an rpoN::Tn5 mutation would possess reduced levels of both NtrC- and NtrX-regulated gene expression since both require the σ54-containing RNA polymerase in order to activate transcription. Since rpoN::Tn5 mutant cells grew as well under nitrogen limitation as wild-type cells, we wanted to determine whether rpoN and its cognate transcription factors were required for the induction of nitrogen regulatory or nitrogen acquisition genes in Caulobacter. Previous experiments demonstrated that glutamate synthase activity was induced in cells grown in minimal medium compared to those cultured in complex medium (11). In order to determine whether rpoN was required for this induction and to explore its requirement for the expression of nitrogen regulatory genes, we constructed transcriptional reporter fusions of glnK, glnB, and gltD to a promoterless lacZ gene. We then compared β-galactosidase activity in wild-type and rpoN::Tn5 cells containing these fusions, grown in complex (PYE), and minimal medium (M2G) (Fig. (Fig.8).8). As expected, all three fusions exhibited a significant increase in promoter activity in cells grown in M2G compared to complex PYE medium. The observed increase in promoter activity was almost identical in magnitude in rpoN::Tn5 cells, indicating that rpoN and conserved nitrogen regulatory signaling pathways do not regulate the expression of these genes in Caulobacter.
We next determined the global pattern of gene expression in Caulobacter cells grown in continuous culture under nitrogen and carbon limitation. In order to accomplish this, we harvested cells from nitrogen- and carbon-limited chemostats operated at a flow rate of 0.034 h−1 (equivalent to tD = 20 h) and compared the relative abundance of mRNAs present under each condition using DNA microarray analysis. This analysis was chosen over a comparison of nutrient-limited to nutrient-replete cells (i.e., chemostat- versus batch-grown cultures) in order to negate the effect of differences in growth rate on global gene expression. Over 62 transcripts were induced at least twofold in cells grown under nitrogen limitation compared to those grown under carbon limitation. The genes encoding these transcripts fell into several distinct classes including those predicted to be involved in nitrogen acquisition, the regulation of gene expression, polysaccharide biosynthesis, and motility (Table (Table1)1) . The relatively highest induced transcripts under nitrogen limitation were those predicted to encode 21 nitrogen assimilation genes. These included genes involved in the transport and assimilation of nitrate such as a nitrate ABC transporter, subunits of both nitrate and nitrate reductases, the nasT gene, predicted to encode a transcriptional antiterminator protein containing a response regulator domain, and glnA, encoding glutamine synthetase. Other nitrogen acquisition genes expressed at high levels are predicted to encode proteins involved in the assimilation of organic nitrogen compounds, such as an enzyme involved in urea catabolism (urea carboxylase [CC1829] and urea carboxylase-associated proteins 1 and 2 [CC1828 and CC1827]), a taurine (2-aminoethanesulfonic acid) ABC transporter (CC1831, CC1832, and CC1833), and an amine oxidase of unknown substrate specificity (CC2793). Transcripts predicted to encode three TonB-dependent receptor transport proteins (CC1093, CC1750, and CC3436) were also increased in nitrogen- versus carbon-limited cells.
Of the remaining 41 transcripts exhibiting relatively increased abundance in nitrogen- versus carbon-limited cells, 23 encode proteins of unknown function. The products of the remaining 18 transcripts are predicted to be involved in a number of diverse functions in the cell. Notably, the expression of genes involved in motility, including two different flagellin and three chemotaxis genes were increased in nitrogen-limited cells. Since these genes are expressed under cell cycle control, this increased abundance may be attributable to a greater proportion of swarmer cells present in the culture at this low flow rate. When Caulobacter cells were harvested from nitrogen-limited chemostat cultures, we noticed that after centrifugation, the cell pellet was “fluffy” in constituency, suggesting that the cells were secreting a polysaccharidelike material. Consistent with this idea, nitrogen-limited cells expressed a gene predicted to be involved in exopolysaccharide assembly (CC2384) and another encoding a phosphoglucomutase family protein (CC2264). This observation suggests that capsular polysaccharide synthesis in Caulobacter is stimulated when the cells are confronted with an environment that is relatively rich in carbon and poor in available nitrogen.
As noted above, in other bacteria the cellular nitrogen status is sensed and transduced through the family of PII proteins. The Caulobacter genome is predicted to encode three PII homologs, GlnB, GlnK, and CC0521, all of which exhibited significantly increased expression in nitrogen- versus carbon-limited cells. Three other regulatory proteins, the NtrC response regulator transcription factor, an extracellular sigma factor family gene (ECF) (CC0981), and an endoribonuclease (CC1089) were also expressed at relatively higher levels in nitrogen-limited cells. Since in many bacteria, the primary response to nitrogen limitation involves the expression of σ54-dependent genes, we also performed microarray analysis comparing nitrogen- and carbon-limited rpoN::Tn5 mutant cells (Fig. (Fig.7).7). In addition, we compared the transcript levels in rpoN::Tn5 cells and wild-type cells grown under carbon and nitrogen limitation (Table (Table11 and see also the supplemental data). Remarkably, of the 62 transcripts that were increased ≥2-fold in nitrogen-limited wild-type cells, only 8 were not induced in rpoN::Tn5 cells. These included two flagellin genes (fljN and fljO) that are known to require RNAP-σ54 for transcription, as well as another potential flagellin homolog (CC1462). The other genes, which have not been previously demonstrated to require rpoN for expression, included a predicted TonB-dependent receptor (CC1750), an extracellular protease (CC2610), and three encoded proteins of unknown function (CC0903, CC2695, and CC3710). These results suggest that, unlike most other well-studied bacteria, RNAP-σ54-dependent transcription does not regulate the majority of genes induced under nitrogen limitation in Caulobacter.
Nitrogen-limited cells also exhibited a significant decrease in the expression of genes encoding ribosomal proteins, other proteins involved in translation (i.e., initiation factor, IF-1, elongation factors, EF-Tu, EF-Ts, and FusA), and general transcription machinery such as RNA polymerase (rpoA and rpoZ) and the transcription antitermination factor, NusG (Fig. (Fig.9).9). Interestingly, rpoN mutant cells exhibited no difference in the expression of these transcripts, suggesting that, in Caulobacter, conserved nitrogen regulatory proteins are involved in controlling the transcription of genes required for basic cellular processes such as translation and transcription, perhaps in response to available cellular pools of carbon and nitrogen. This differs from the case of enteric bacteria, in which nitrogen-modulated σ54-dependent transcription is mainly confined to the regulation of genes involved in nitrogen acquisition.
The expression of at least 153 additional transcripts increased at least twofold in carbon- versus nitrogen-limited chemostat-grown cells. At least 48 of these transcripts could be assigned to encoding proteins involved in transport and catabolism of carbon substrates (Fig. 10A). Ten of the transporter genes were predicted to encode TonB-dependent receptors which may be involved in the uptake of polysaccharides. The upregulated catabolic genes were predicted to encode enzymes involved with the utilization of a diverse array of carbon substrates including fatty acids, polysaccharides, peptides, lignocellulose degradation products, and other small organic molecules (Table (Table2).2). In addition, several genes for central metabolic pathway enzymes and energy metabolism were increased in carbon-limited cells.
Notably, in this regard, RNAP-σ54 was also required for the induction of at least 112 transcripts in carbon- versus nitrogen-limited cells (Fig. 10B). Several of these genes are predicted to be involved in a variety of metabolic processes, including energy generation, lipid and carbohydrate metabolism, and amino acid biosynthesis. However, most of the σ54-dependent transcripts have no known function. Conversely, rpoN::Tn5 cells induced 55 genes that were not significantly increased in expression in wild-type cells, suggesting that the absence of rpoN-dependent transcription alters the global pattern of gene expression in response to glucose limitation. Transcripts encoding potential regulatory proteins that were also increased in carbon- versus nitrogen-limited cells included a transcriptional activator of the MerR family (CC0081), a predicted repressor (CC1356), a single domain response regulator (CC0284), and a response regulator DNA binding protein (CC1182), all of which were dependent on the presence of rpoN for induction. Likewise, rpoN was required for the induction of two other regulatory proteins, GsiB (CC1178) and DksA (CC2580), both of which have been described in other bacteria to play a role in response to stress or starvation. GsiB, for glucose starvation-induced protein (or general stress induced), is conserved throughout the alphaproteobacteria and has been shown to be expressed in Bacillus subtilis cells in response to glucose limitation (33, 36, 59); however, its function remains obscure. DksA has been demonstrated to function as a coregulator of transcription, along with (p)ppGpp in cells undergoing a stringent response (42, 43). A previous report showed that carbon starved Caulobacter cells elicited a stringent response (27), and thus it is possible that DksA may be a critical regulator of transcription in these carbon-limited, chemostat-grown cells.
C. crescentus cells are found in oligotrophic freshwater environments containing nanomolar to micromolar concentrations of critical nutrients. The obligatory asymmetric cell division exhibited by these bacteria, generating a motile swarmer cell from a sessile stalked cell, is an evolutionary adaptation designed to both efficiently exploit and colonize these sparse environments. We devised an experimental strategy here that permitted the examination of the role of nutrient availability in governing the timing of the swarmer to stalked cell differentiation in growing cultures. We hypothesized that an open, continuous-flow culture system such as the chemostat used in the experiments described here could be used as an experimental system to determine whether the limitation a single nutrient would influence the temporal progression of swarmer cell development. In order to test this idea, we measured two different parameters, biomass concentration and the percentage of nonstalked cells, over a range of imposed flow rates. The prediction was that if cell reproduction was delayed by the limitation of a given nutrient, the biomass concentration in the culture vessel would decrease when the flow rate exceeded the culture generation time. Furthermore, if the delay in cell reproduction was attributable to an extended swarmer cell period, then the culture would contain a greater proportion of swarmer cells at lower flow rates (i.e., D < the generation time) where the biomass concentration was high. We found that cells in continuous culture under ammonium limitation exhibited this behavior, with a high swarmer cell and biomass concentration only achievable at relatively low flow rates (≤0.05 h−1). These results indicate that nitrogen-limited, chemostat-grown cells spend an extended period of time in the swarmer cell phase of the cell cycle.
We hypothesize that nitrogen limitation extends the swarmer cell lifetime by delaying the onset of a sequence of differentiation events, which when initiated by the correct combination of external environmental cues, sets the swarmer cell on a path to differentiate into a stalked cell within a fixed time period. In most published experiments, this differentiation time period is equivalent to approximately 30 min in nutrient-replete medium (at tD = 2 h). One possibility is that once this differentiation pathway is initiated, the transition from swarmer to stalked cell cannot be reversed. This is supported by the observation that the timing of swarmer cell differentiation, when cultivated in nutrient-replete medium, appears to be invariant. In addition, when swarmer cells are grown in rich medium and transferred to medium lacking either carbon or nitrogen, they appear to enter into the differentiation pathway (7, 16, 27) (Fig. (Fig.2)2) but do not complete the transition into a stalked cell. For example, under these starvation conditions, CtrA is degraded on time (Fig. (Fig.2),2), but motility continues and DNA replication does not initiate. In the case of carbon starvation, this is attributable to the absence of glucose in the medium triggering a stringent response and resulting in the degradation of the DNA replication initiator protein, DnaA (27). Thus, in these swarmer cells, the differentiation pathway is initiated as a consequence of growth in rich medium and an alternative regulatory pathway (i.e., stringent response) blocks entry into S phase.
The transcriptional profiling experiments of Caulobacter cells grown under nitrogen and carbon limitation have generated the first comprehensive picture of the global regulatory strategies used by an oligotroph when confronted with an environment limited in key macronutrients. The pattern of regulated gene expression shares some features in common with most copiotrophic organisms, but critical differences suggest that Caulobacter, and perhaps other oligotrophs, are “hard-wired” to deal distinctly with their natural environments. This idea is most clearly demonstrated in the expression pattern of genes encoding either nitrogen or carbon acquisition proteins. For example, genes of this class induced in nitrogen- versus carbon-limited cells include those required for the acquisition of two inorganic nitrogen sources (nitrate and ammonium) and two organic nitrogen sources (urea and taurine). Only five genes encoding transport systems were increased in expression (nitrate, taurine, and three TonB-dependent receptors). In contrast, chemostat-grown E. coli cells induced 28 different transporter genes in nitrogen- versus carbon-limited chemostat-grown cultures (18). These E. coli transcripts are predicted to encode transport systems involved in the uptake of a diverse array of nitrogen sources, including ammonia, amino acids, nucleosides, and polyamines. Similarly, E. coli cells grown in nitrogen-limited continuous culture also induce 29 genes predicted to encode amino acid biosynthetic enzymes (18), whereas nitrogen-limited Caulobacter cells only induced glutamine synthetase. Thus, the primary strategy used by both of these organisms, and probably by heterotrophic bacteria in general, is to induce genes required for scavenging nitrogen sources that are most likely found in their natural environments. An analogous pattern of regulation of gene expression exists in carbon-limited Caulobacter cells. Most of the carbon-limitation-induced genes that have a clear predicted function are required to degrade carbon compounds known to be abundant in oligotrophic freshwater environments, many of which would be derived from plant material. For example, these include 4-hydroxy-3-methoxycinnamaldehyde dehydrogenase (CC1849) and 3-(4-hydroxy-3-methoxyphenyl)-2-propenoyl-coenzyme A synthetase (CC3338), enzymes catalyzing two sequential steps in the degradation of lignin. Similarly, genes predicted to encode enzymes required for the oxidation of the monoterpene geraniol (CC1310 and CC1310), as well as an arabinosidase, are also elevated in expression under carbon limitation. Carbon-limited Caulobacter cells also induced the expression of genes encoding 10 different TonB-dependent receptors, which are outer membrane components of the TonB uptake system. In enteric bacteria, transporters of this class have been shown to facilitate the transport of chelated metals such as cobalamin. It is now believed that this class of transport system is utilized to promote the uptake of a diverse range of molecules with substrate specificity uniquely conferred by different TonB-dependent receptors. The C. crescentus genome is predicted to encode 68 different TonB-dependent receptors. One of these has been shown to be required for the uptake of maltodextrins (28), and another may be required for the uptake of N-acetylglucosamine-containing oligosaccharides (10), perhaps indicating that the 10 TonB receptor-encoding genes induced in carbon-limited cells are used to take up oligosaccharides and polysaccharides derived from plant material. Interestingly, carbon-limited cells induced the expression of genes that may be involved in the degradation of xenobiotics such as hexachlorohexane (CC0094) and nylon oligomers (CC1323), indicating that Caulobacter may play an important role in scavenging man-made pollutants from the environment.
Virtually nothing is known regarding how oligotrophic bacteria regulate gene expression under conditions of nutrient limitation. One surprising finding was the limited involvement of the conserved σ54-dependent NtrB/NtrC nitrogen regulatory system in the induction of nitrogen acquisition genes. Of the 68 transcripts that were induced, only eight appeared to require rpoN. In contrast, microarray experiments with E. coli that were aimed at identifying genes under the control of NtrC revealed 31 operons containing 86 genes that would be predicted to be induced under nitrogen limitation (61). These included the transport systems for 17 different nitrogenous compounds that are indicative of the diversity of nitrogen sources E. coli likely encounters in its ecological niches. Furthermore, Caulobacter rpoN mutants did not exhibit any discernible impairment of growth under nitrogen limitation, suggesting that they possess a distinct and independent regulatory network to deal with nitrogen limitation. Unlike enteric bacteria, Caulobacter possesses an additional sensor kinase/response regulator pair (NtrX/NtrY) that is thought be a component of the nitrogen regulon. It is as yet unclear whether these two proteins and NtrB/NtrC have some role in regulating nitrogen acquisition genes under different growth conditions, such as the utilization of alternative nitrogen sources. Since Caulobacter is an oligotroph, its alternative nitrogen sources are likely to be amino acids, purines, pyrimidines, and polyamines.
When bacteria are starved for either amino acids or carbon they induce the stringent response, which leads to the production of the nucleotide alarmone, (p)ppGpp. Enteric bacteria possess two different (p)ppGpp synthetases, RelA and SpoT, each of which are involved in responding to different metabolic stresses (29, 54). RelA interacts with stalled ribosomes, catalyzing (p)ppGpp synthesis in response to amino acid starvation, whereas SpoT, possibly through interactions with acyl carrier protein, synthesizes (p)ppGpp under conditions of carbon deprivation. Increased levels of (p)ppGpp result in global transcriptional responses, including an inhibition of stable RNA synthesis and a stimulation of transcription of a subset of amino acid biosynthetic genes and alternative sigma factor encoding genes (i.e., rpoS) (29, 54). Recent transcriptional profiling experiments of isoleucine-starved E. coli cells indicate that the global response to nutrient deprivation and increased levels of (p)ppGpp may be even more extensive, possibly involving alterations in the expression of hundreds of genes (56). In addition to regulating a global response to nutrient deprivation, the stringent response is thought to have a major role in response to changes in growth conditions by coordinating the rate of stable RNA synthesis and translation with that of growth rate.
It is a likely that the Caulobacter cells maintained at the slowest imposed growth conditions in the experiments presented here, have increased levels of (p)ppGpp and are undergoing a stringent response. In support of this idea, at least 39 of the genes downregulated in nitrogen- versus carbon-limited Caulobacter cells encode components of the translation machinery and proteins involved in transcription. This would be a hallmark of the stringent response, since decreases in the rate of stable RNA transcription results in a reduction in the transcription of ribosomal protein genes. Members of the alphaproteobacteria do not possess both relA and spoT but instead contain a single relA/spoT homolog (60). Previous experiments with Caulobacter have demonstrated that the relA/spoT homolog (i.e., spoT) is required for the synthesis of ppGpp in response to carbon deprivation (27). It is possible that the Caulobacter SpoT homolog also plays a role in synthesizing ppGpp in response to low growth rates. However, since we compared transcripts from nitrogen- and carbon-limited cells growing at identical imposed growth rates, induction of ppGpp synthesis cannot be the sole regulatory factor influencing the observed differences in the transcription of ribosomal protein genes. Since Caulobacter cells possess a sole ppGpp synthetase, we propose that additional nutrient-specific regulatory networks are used in response to substrate limitation at low growth rates. In the case of nitrogen-limited growth this additional regulatory network is likely to involve RNAP-σ54-dependent transcription, since rpoN mutants, unlike wild-type cells, exhibited no significant repression of genes encoding ribosomal proteins when grown under nitrogen limitation. In this case, in a simple scheme, it is possible that nitrogen status as transmitted through PII proteins influences either NtrC- or NtrX-activated transcription, thereby regulating the expression of ribosomal protein genes. Thus, a conserved network that senses cellular nitrogen status not only would regulate nitrogen acquisition genes, but also is co-opted to control essential cellular processes in response to nitrogen availability. Interestingly, 112 of the genes induced in carbon- versus nitrogen-limited cells also required rpoN for complete induction. This large number of genes, ca. 2.7% of the entire genome, mostly encode proteins predicted to be involved in metabolism of carbon substrates. It is possible that the conserved RNAP-σ54-dependent nitrogen regulatory system which regulates gene expression in response to the relative cellular pools of 2-oxoglutarate and glutamine may also be controlling the expression of carbon acquisition genes.
One experimentally challenging aspect in understanding Caulobacter development is determining how environmental cues are integrated within the operation of the cell cycle. Progression through the cell cycle is driven by successive oscillations in the abundance and activity of regulatory proteins such as CtrA, and the chromosome replication initiator protein, DnaA (25). Ultimately, the timing of these oscillations is influenced by the temporal and spatial regulation of key regulatory proteins which are activated by, and simultaneously control, critical landmark events in the cell cycle. Interestingly, many of these regulatory proteins are members of the large conserved family of bacterial two-component regulators. Typically, these consist of sensor histidine kinases that phosphorylate response regulator proteins, affecting gene expression, metabolism, or motility. The identity of the sensory inputs modulating the activity of these kinases remains experimentally elusive. Do these developmentally regulated kinases sense environmental cues, internal cues (i.e., cell division, DNA replication), or perhaps some blend of both? Previous experiments, as well as the experiments described here, indicate that the regulatory networks monitoring the availability of key nutrients may very well intersect with signal transduction pathways controlling cell cycle progression.
We thank Matthew Graf and Midge Llewellyn for helpful comments on the manuscript.
This study was supported by Public Health Service Grants GM048417 (to J.W.G.) and GM082899 (to M.T.L.) from the National Institutes of Health and a grant from the National Science Foundation (MCB-064133) to J.W.G.
Published ahead of print on 30 November 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.