SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Bacteria comprehensively reorganize their global gene expression when faced with starvation. The alarmone ppGpp facilitates this massive response by co-ordinating the downregulation of genes of the translation apparatus, and the induction of biosynthetic genes and the general stress response. Such a large reorientation requires the activities of multiple regulators, yet the regulatory network downstream of ppGpp remains poorly defined. Transcription profiling during isoleucine depletion, which leads to gradual starvation (over > 100 min), allowed us to identify genes that required ppGpp, Lrp and RpoS for their induction and to deduce the regulon response times. Although the Lrp and RpoS regulons required ppGpp for their activation, they were not induced simultaneously. The data suggest that metabolic genes, i.e. those of the Lrp regulon, require only a low level of ppGpp for their induction. In contrast, the RpoS regulon was induced only when high levels of ppGpp accumulated. We tested several predictions of a model that explains how bacteria allocate transcriptional resources between metabolism and stress response by discretely tuning two regulatory circuits to different levels of ppGpp. The emergent regulatory structure insures that stress survival circuits are only triggered if homeostatic metabolic networks fail to compensate for environmental deficiencies.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Escherichia coli cells attempt to compensate for nutritional deficiencies in their environment by activating endogenous biosynthetic pathways (Bremer and Dennis, 1996). Should conditions deteriorate to the point these biosynthetic pathways can no longer remedy the situation, the cells transition into stationary phase, a physiological state oriented towards protection of cellular structures and long-term survival (Hengge-Aronis, 1996). Thus, as the nutritional quality of the environment diminishes, the cells must properly allocate resources between biosynthetic and stress/survival functions. The structure of the transcription network used by cells to balance these processes across this ‘feast to famine’ gradient remains incompletely understood.

In most bacteria, growth arrest prompts a restructuring of global transcription patterns known as the stringent response (Cashel et al., 1996; Potrykus and Cashel, 2008). The alarmone ppGpp is the arbiter of the stringent response and lies at the apex of the network that governs global gene expression in response to nutrient limitation (Cashel et al., 1996). This assertion rests on the observation that cells lacking ppGpp exhibit profoundly altered global gene expression patterns during carbon and amino acid starvation (Traxler et al., 2006; 2008; Durfee et al., 2008). While these studies suggest that ppGpp controls one of the largest transcription networks in the bacterial cell, lingering questions regarding the mechanisms underlying regulation by ppGpp and the sheer size of the ppGpp regulon have led to the omission of ppGpp from all large-scale, computational transcriptional network analyses to date. Before such an all-encompassing model can be built, the functional structure of the network must be defined.

Most experiments designed to examine the stringent response/stationary phase physiology utilize conditions of ‘feast or famine’. However, the ppGpp level is inversely proportional to the balanced growth rate, i.e. a condition that supports a suboptimal growth rate leads to an elevated, but not maximal level of ppGpp (Lazzarini et al., 1971; Sokawa et al., 1975; Ryals et al., 1982; Sarubbi et al., 1988). This suggests that the stringent response is rheostatic rather than ‘on or off’. As the ppGpp level serves as a general indicator of the nutritional status of the cell, it is logical that other regulators might key their activity to this signal. Indeed, global expression profiles from cells during the stringent response suggest that a multitude of other regulators are involved, depending on the type and severity of stress encountered (Traxler et al., 2006; 2008; Durfee et al., 2008). In our previous investigations designed to examine the physiological extent of the ppGpp-mediated stringent response to isoleucine starvation, we observed that many genes known to be regulated by the alternative sigma factor RpoS and the DNA-binding protein Lrp were induced (Traxler et al., 2008). Thus, as a step to understanding the larger architecture of the stringent response, here we examine the Lrp and RpoS networks as subordinate components of the ppGpp regulon and consider the interplay between these two networks at the systems level.

The primary synthase of ppGpp is RelA, which catalyses the production of ppGpp in response to amino acid starvation (Cashel et al., 1996; Wendrich et al., 2002). A secondary ppGpp synthase, SpoT, produces ppGpp in response to diverse stresses including carbon (Xiao et al., 1991), iron (Vinella et al., 2005) and fatty acid starvation (Battesti and Bouveret, 2006). SpoT also contains ppGpp hydrolase activity, and thus plays a crucial role in regulating the overall level of ppGpp (Sarubbi et al., 1988; Xiao et al., 1991). ppGpp binds directly to RNA polymerase (RNAP) with the help of the RNAP-binding protein DksA (Artsimovitch et al., 2004; Haugen et al., 2008). ppGpp and DksA compromise the ability of RNAP to form a productive open complex at intrinsically unstable promoters (e.g. ribosomal RNA promoters), via a mechanism that is not completely understood (Haugen et al., 2008; Vrentas et al., 2008; Rutherford et al., 2009). Conversely, ppGpp and DksA have been shown to directly stimulate transcription of amino acid biosynthesis genes, which have a much longer open complex half-life (Paul et al., 2005). Thus it has been suggested that when ppGpp accumulates, RNAP is liberated from rRNA promoters and becomes available for transcription of diverse promoters across the genome (Barker et al., 2001). However, more recent work has shown that elevating the levels of σ70-bound RNAP actually results in enhanced transcription of genes of the translation apparatus, rather than enhanced transcription of stress and metabolic genes (Gummesson et al., 2009). Based on these results, Gummesson et al. propose that ppGpp reduces the size of the free σ70-bound RNAP pool. The effects of ppGpp on the transcriptional apparatus and promoter choice remain an active area of investigation.

At the transcriptional level, we and others have found the stringent response to include the downregulation of diverse types of macromolecular synthesis (protein, DNA, RNA, fatty acids, etc.), a broad-scale restructuring of intermediary metabolism (including amino acid biosynthesis), as well as induction of regulons, controlled by Lrp and RpoS (Durfee et al., 2008; Traxler et al., 2008; Aberg et al., 2009). The global transcription factor Lrp (leucine responsive protein) regulates a large number of genes involved in amino acid biosynthesis and transport (Tani et al., 2002; Cho et al., 2008). A recent systems-level analysis of the Lrp network found that the different outputs regulated by Lrp encompass several coherent physiological states that balance amino acid uptake, degradation and biosynthesis (Cho et al., 2008). Thus, to the extent that Lrp regulates induction of amino acid biosynthesis/metabolism genes, its role is complementary to that of ppGpp during the stringent response to amino acid starvation. RpoS is the mediator of the general stress response in E. coli (Loewen and Hengge-Aronis, 1994) and is known to control > 100 genes in response to diverse conditions including starvation, oxidative and osmotic stresses (Lacour and Landini, 2004; Weber et al., 2005). The connections between ppGpp and RpoS are several-fold: (i) ppGpp is required for increased transcription of rpoS during entry into stationary phase (Lange et al., 1995), (ii) ppGpp facilitates competition of alternative sigma factors (including RpoS) with the housekeeping sigma (σ70) for core RNAP binding (Jishage et al., 2002) and (iii) ppGpp is required for increased transcription of the anti-adaptor protein gene iraP, which results in inhibited proteolysis of RpoS (Bougdour and Gottesman, 2007).

The observation that genes of the Lrp and RpoS regulons appear to require ppGpp for their induction prompted us to consider how these regulatory networks are integrated as components of the stringent response. We used whole-genome microarrays to experimentally determine the extent and timing of gene expression controlled by ppGpp, Lrp and RpoS in response to isoleucine starvation. Our results suggest a model in which genes of the Lrp regulon have a lower threshold for induction by ppGpp, while RpoS-dependent genes require a relatively high level of ppGpp for their induction. Further experimentation correlating promoter activity of representative Lrp and RpoS regulon reporter genes with ppGpp accumulation patterns validated several aspects of our proposed model. We interpret these results to mean that at a systems level, nutritional genes can readily respond when ppGpp is at basal or only slightly elevated levels (signifying suboptimal growth conditions) while the general stress response is only fully developed under more severe growth limitation (which leads to high levels of ppGpp accumulation). Our model illustrates how a single signal molecule, ppGpp, can drive an independent feed-back loop (controlled by Lrp) coupled in parallel with a feed-forward loop (RpoS circuit) to appropriately balance basic biosynthetic and survival processes across the feast to famine gradient.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Overall strategy

To investigate the regulatory architecture of the stringent response, we developed an experimental system based on starvation for isoleucine, which could be equally applied to all strains used here, including the multi-auxotrophic ppGpp0 strain. This system has been described in detail elsewhere (Traxler et al., 2008). Briefly, E. coli K-12 strains cannot grow in the absence of isoleucine when valine is present due to a frame-shift mutation in the ilvG gene, which renders the encoded protein (valine insensitive acetohydroxy-acid synthase II) inactive (Lawther et al., 1981). Feedback inhibition by valine of the two other acetohydroxy-acid synthase enzymes shuts down isoleucine biosynthesis. Thus, isoleucine depletion in the presence of excess valine serves as an experimental model system for study of gradual amino acid limitation, to the point of starvation, and the elicitation of the stringent response.

Relative contributions of Lrp and RpoS in response to isoleucine starvation

We previously analysed WT and ppGpp0 strains and found the ppGpp-dependent response to isoleucine starvation to include a number of genes in the Lrp and RpoS regulons (Traxler et al., 2008). To elucidate the role of Lrp and RpoS during the stringent response, we grew strains lacking these transcription factors in isoleucine-limiting medium (Fig. 1). This MOPS medium includes glucose (0.2%) as a carbon source and all 20 amino acids in the amounts described in Wanner et al. (1977), with the exception that isoleucine was present at a low starting concentration of 60 µM. Under these conditions, the cultures exited logarithmic growth around OD 0.3, ultimately achieving a final OD of 0.6–0.7. RNA was harvested after the cells had transitioned into stationary phase when isoleucine was exhausted (arrows, Fig. 1). The transcriptomes of the mutant strains were analysed using Affymetrix GeneChips.

image

Figure 1. Growth curves for isoleucine starved cultures. Cultures were grown as described in Experimental procedures. Arrows indicate time points when RNA was harvested for microarray analysis.

Download figure to PowerPoint

For our analysis, we considered expression profiles from all four strains: WT, ppGpp0, Lrp- and RpoS-. Transcriptomes from each of these four strains starved for isoleucine were compared with the transcriptome of the WT strain during logarithmic growth under isoleucine-replete conditions. Using a simple twofold criterion (Wren and Conway, 2006), we categorized genes induced in the WT in response to isoleucine starvation according to their dependence on ppGpp, Lrp or RpoS for induction (Fig. 2A). For example, a gene whose expression was > twofold lower in both the ppGpp0 and Lrp- strains would be placed in Venn category 2.1 in Fig. 2A. This analysis leads to several important conclusions: (i) the majority of genes induced in response to isoleucine starvation require ppGpp for their normal induction (365 required it vs. 167 that did not), (ii) Lrp controls a smaller subset of genes than RpoS (39 controlled by Lrp vs. 133 controlled by RpoS, 11 were controlled by both) and (iii) the great majority of genes in both the Lrp and RpoS regulons also require ppGpp for their full induction (153 required ppGpp, 8 did not). These results suggest a regulatory scheme in which ppGpp serves as the apex regulator, with Lrp and RpoS functioning to control discrete subsets of the larger ppGpp regulon. A list of genes in each category is available in Table S1.

image

Figure 2. Contributions of Lrp and RpoS-mediated gene induction to the ppGpp-dependent stringent response. Only genes that were induced > twofold in the WT were considered for the analysis (532 total). A. Venn diagram of overlapping regulons. Numbers of genes in each field are shown in red. Each field is labelled 1.1–3.1 for identification in B–E. Genes were classified as dependent upon a given regulator if expression was > twofold lower in the mutant strain. ‘Unaffected’ indicates genes that were similarly induced in all strains. ‘Other’ indicates genes that were expressed > twofold higher than the WT in one or more of the mutant strains. B. Heat map of genes that require ppGpp and Lrp for their induction. Gene names are shown to the right and strains are shown at bottom. Colour legend for Log2 expression is shown below. C. Genes that require ppGpp, RpoS and Lrp for induction. D. Expression of the lrp and rpoS genes in array experiments. E. Genes that require ppGpp and RpoS for their induction.

Download figure to PowerPoint

Physiological roles of Lrp and RpoS

Our data suggest clear physiological roles for Lrp and RpoS based on the content of their respective regulons, and these data correlate well with other studies that examined global regulation by these regulators (Tani et al., 2002; Weber et al., 2005; Cho et al., 2008). Under isoleucine starvation, Lrp was required for induction of many metabolic genes (Fig. 2B) including several involved in amino acid metabolism, such as leucine biosynthesis (leu genes), threonine biosynthesis (thrA and thrB), alanine metabolism (dadAX) and serine metabolism (serA). Lrp was also required for induction of the glyoxylate shunt genes aceA and aceB, as well as the malic enzyme (maeB). Taken together, these data suggest that Lrp, along with ppGpp, plays a role in co-ordinating the metabolic response to isoleucine starvation.

RpoS was required for the induction of various genes that have roles in preparing for long-term survival in stationary phase (Fig. 2E). These included systems for protecting the cell against oxidative stress (sodC, dps, wrbA), osmotic stress (several osm genes, treA, otsA, otsB) and metabolic genes known to be RpoS-dependent (poxB, talA and fbaB). Thus the RpoS-dependent general stress response is initiated in response to isoleucine starvation. The Lrp and RpoS regulons are known to overlap (Weber et al., 2005) and we found that 11 genes required both Lrp and RpoS for their induction (Fig. 2C); this group was mostly comprised of genes involved in acid tolerance (gadA, gadB, hdeA, hdeB, hdeD, xasA, gabP). From these results, we conclude that the Lrp and RpoS networks serve specific functions within the larger stringent response to isoleucine starvation: Lrp works in conjunction with ppGpp to induce genes involved in amino acid metabolism (with the presumed function of providing depleted amino acids) while RpoS, together with ppGpp, readies the cell for survival under prolonged starvation conditions.

One possible way that ppGpp might influence the Lrp and RpoS regulons is by directly regulating the expression of the lrp and rpoS genes. Under the isoleucine starvation conditions tested here, the WT did not significantly induce lrp expression (Fig. 2D). Based on this result, we conclude that induction of the Lrp regulon does not require robust induction of lrp at the transcriptional level, but rather relies on signalling through existing Lrp protein. The difference in lrp expression between the WT and ppGpp0 strain was < twofold (Fig. 2D); thus, under these conditions, we conclude that the failure of the ppGpp strain to induce the Lrp regulon was not likely due to impaired transcription of the lrp gene. In contrast, while both the WT and ppGpp0 strain induced rpoS > twofold, rpoS transcription was much lower in the ppGpp0 strain (2.2-fold induced vs. 6.7-fold induction in the WT). Thus, impaired induction of the RpoS regulon in the ppGpp0 strain can be accounted for, at least in part, by poor induction of rpoS at the transcriptional level.

Differential induction times of the Lrp and RpoS regulons

Having delineated the Lrp and RpoS regulons induced during isoleucine starvation, we sought to determine the timing of these components of the stringent response. In this context, the timing of induction serves as an output of the activity of each regulator and thus can provide information about the operational regulatory network. In a 12-point microarray time-course experiment to examine the behaviour of the Lrp and RpoS regulons in WT cells during logarithmic growth and as they starved for isoleucine (compare heat maps in Fig. 3 with growth curve in Fig. 4), several trends were readily apparent. First, many genes of the Lrp regulon, including those involved in amino acid biosynthesis/metabolism were induced very early (OD ∼0.3) in the response to isoleucine limitation, before the isoleucine was exhausted and while the cells were still in logarithmic growth phase. In contrast, we found that the RpoS regulon was induced later than the Lrp regulon. To quantify these potential differences in timing, we undertook a systems-level analysis.

image

Figure 3. Time-course (WT) heat maps of the Lrp and RpoS regulons as cells starved for isoleucine. Numbers 2.1, 2.3 and 3.1 refer to Venn fields in Fig. 2. Numbers below each heat maps refer to the time points identified in Fig. 4.

Download figure to PowerPoint

image

Figure 4. WT growth curve, average regulon response times and ppGpp accumulation as cells starved for isoleucine. Numbered time points indicate times of RNA sampling for array data shown in Fig. 3. Expression of each regulon shown in Fig. 3 was averaged at each time point and plotted. Sigmoidal regression curves were plotted for each regulon and the response time (time of ½ maximal induction) is marked as a solid vertical line.

Download figure to PowerPoint

The response time of a gene (defined as the time at which expression is half-maximal) offers a temporal measure of gene induction (Mangan and Alon, 2003; Alon, 2007). To compare the relative response times of the Lrp and RpoS regulons, we averaged the expression levels of all the genes in each regulon at each time point. We then plotted a sigmoidal regression curve for the averaged regulon expression values vs. time (Fig. 4). We found that the Lrp regulon was induced ∼45 min before the RpoS regulon. Genes that required both Lrp and RpoS for their induction had an intermediate induction time, ∼30 min after the Lrp regulon and ∼15 min before the RpoS regulon. Because the response of both the Lrp and RpoS regulons to isoleucine starvation requires ppGpp [see Fig. 2 and (Traxler et al., 2008)], these data suggest a regulatory and physiological hierarchy in which the initial reaction to isoleucine limitation entails the induction of genes involved in amino acid biosynthesis (mediated by Lrp); while the RpoS-mediated general stress response is not induced until later, after the growth rate begins to slow.

These observations lead to a central question: How can a single signal molecule, ppGpp, differentially dictate the response times of these two genetic networks? We suggest that this question can be considered in terms of ppGpp level and regulon output. We observed previously that ppGpp accumulated over ∼100 min under our experimental conditions, proportional to the extent of isoleucine limitation (Traxler et al., 2008), which is a relatively long time compared with most experimental conditions that provoke robust ppGpp accumulation, usually within 10–15 min (Lazzarini et al., 1971; Ryals et al., 1982; VanBogelen et al., 1987). The slow ppGpp accumulation associated with growth-dependent exhaustion of isoleucine suggests that the cells experienced a gradual decline in environmental quality, a situation that might approximate a continuum from feast to famine. We considered the accumulation of ppGpp relative to the induction times of the Lrp and RpoS regulons (Fig. 4), and noted that the induction time of the Lrp regulon corresponded to a lower level of ppGpp (< 100 pmol ml−1 OD−1), while the RpoS regulon was induced when the level of ppGpp was higher (∼400 pmol ml−1 OD−1). As the limit of ppGpp detection in our assay was 100 pmol ml−1 OD−1, we cannot state with certainty the lower boundary of the response. Nevertheless, the data suggest a model in which the Lrp and RpoS regulons require different amounts of ppGpp for their induction (Fig. 5).

image

Figure 5. Model of regulatory architecture within the stringent response under three different environmental conditions. Arrows indicate activation, flat ends represent repression. Green lines indicate active vs. inactive (black) pathways. This figure is available in colour online. A. When nutrients are plentiful, both the metabolic Lrp regulon and the stress survival RpoS regulon are inactive. ppGpp accumulation is prevented by exogenously available nutrients (i.e. amino acids). B. When an amino acid becomes limiting, ppGpp accumulates. Signalling through a regulatory/sensory protein such as Lrp works in conjunction with a low level of ppGpp (x) to activate transcription of amino acid biosynthetic genes. Endogenously produced amino acids cause the ppGpp level to decline, inactivating the general stress response [which requires a high level of ppGpp (y)] and allowing growth to continue. C. When the environment no longer supports growth, biosynthetic pathways are incapable of producing necessary metabolites. ppGpp accumulates to a high level, fully activating the general stress response.

Download figure to PowerPoint

Altering the ppGpp accumulation rate predictably impacts the response times of the Lrp and RpoS regulons

Based on the observations above, we sought to test if the Lrp and RpoS regulons required different threshold levels of ppGpp for their induction. A key prediction that follows from this model is that if the two regulons require different ppGpp levels for their induction, then altering the rate of ppGpp accumulation should impact the relative response times of the two regulons. For example, we observed that when ppGpp accumulated over ∼100 min, the difference in response times between the Lrp and RpoS regulons was ∼45 min. We reasoned that if we could accelerate the rate of ppGpp accumulation, then the difference in the response times between the Lrp and RpoS networks would be reduced.

To test this hypothesis, we developed an experimental system that allowed us to measure gene induction during rapid ppGpp accumulation. To monitor gene induction of the Lrp and RpoS regulons, we chose two representative reporter genes from each regulon. For the Lrp regulon we chose the leuL and dadA promoters; genes under control of both of these promoters were influenced by an lrp mutation in our data sets, and are previously known to be controlled by Lrp (Lin et al., 1992; Zhi et al., 1999). To represent the RpoS regulon, we selected the promoters for the yahO and wrbA genes. Both of these genes are strongly induced upon entry into stationary phase, are RpoS-dependent in our array experiments and known to be controlled by RpoS (Yang et al., 1993; Ibanez-Ruiz et al., 2000; Lacour and Landini, 2004). These four promoter regions were cloned into pUA66, a very low copy plasmid that has been used extensively for similar genetic analyses, upstream of a fast-folding GFP allele (Zaslaver et al., 2004; 2006; Rhodius et al., 2006). Thus, we monitored GFP fluorescence as an indicator of the respective promoter activities and hence Lrp and RpoS regulon induction. This set-up allowed us to grow the four reporter strains in parallel 50 ml cultures that facilitated experimental manipulation.

We first confirmed that the Lrp regulon was induced before the RpoS regulon during slow ppGpp accumulation (Fig. 6A). We found that the Lrp-dependent promoters were induced an average of 26 min before the RpoS-dependent promoters. While this 26-min difference is shorter than the 45 min whole-regulon measurements in our array data, it represents a statistically robust difference in response times between the two regulons (P < 0.001). In our normal isoleucine starvation regime, the cells deplete isoleucine from the medium until growth can no longer be supported. To abruptly trigger isoleucine starvation (and hence accelerate ppGpp accumulation), we grew the four reporter strains in medium with glucose and all 20 amino acids in excess. When the cultures reached an OD of ∼0.3–0.4 we rapidly collected the cells on glass fibre filters and immediately resuspended them in medium containing glucose, 19 amino acids, and no isoleucine. Under these conditions we found that ppGpp accumulates to ∼800 pmol ml−1 OD−1 in ∼10 min (Fig. 6B). Abrupt isoleucine starvation reduced the average difference in the response times between Lrp-dependent and RpoS-dependent genes to a statistically insignificant 6.6 min (P = 0.224). Thus, one prediction of our proposed model appears to hold true: response times of Lrp- and RpoS-dependent gene expression vary in accordance with the ppGpp accumulation rate.

image

Figure 6. Differences in response times of the Lrp and RpoS regulons are reduced when ppGpp accumulates rapidly. A. Fluorescence of GFP reporter fusions to RpoS-dependent (PyahO and PwrbA) and Lrp-dependent promoters (PleuL and PdadA) under conditions wherein culture growth results in exhaustion of exogenous isoleucine. ppGpp accumulates slowly (in > 100 min) under such a scenario. Response times are marked as vertical coloured lines. T0 = time of inoculation. B. Fluorescence from the same promoters as in A except that cells were grown in media replete with all 20 amino acids, collected on filters at T0, and resuspended in medium with all amino acids except isoleucine. ppGpp accumulates rapidly and the differences in response times between promoters are reduced. C. Response times of each promoter from A. Blue lines indicate the average response time of promoters from each regulon. D. Response times from promoters from B. Dotted blue lines indicate that the difference in regulon response times is statistically insignificant.

Download figure to PowerPoint

Integration of the Lrp and RpoS networks within the stringent response

In considering other predictions implied by the proposed model (Fig. 5), we noted features that may serve to partition expression of the Lrp and RpoS regulons under a range of physiological conditions. Dependence of the Lrp and RpoS networks on ppGpp constitutes two different types of physiological/regulatory motifs. The Lrp side of the model contains a physiological feedback loop. Under amino acid limiting conditions, the series of events in this feedback loop begins with ppGpp accumulation. A relatively low level of ppGpp and Lrp (activated by a mechanism that is not understood) together allow for induction of the Lrp regulon. The physiological result of Lrp regulon induction is biosynthesis of the missing amino acids. Newly synthesized amino acids lead to a reduction in the ppGpp level. The RpoS side of the model contains a coherent feed-forward loop in which increased ppGpp prompts RpoS induction (at the transcriptional level) and stabilization (via IraP). RpoS and ppGpp then work together (via enhanced alternative sigma factor competition) to induce transcription of the RpoS regulon. A critical element of this model is that as the Lrp regulon might indirectly influence ppGpp levels, Lrp activity has the potential to indirectly modulate the induction of the RpoS regulon. This important connection within the stringent response offers a directly testable prediction, which we sought to investigate experimentally.

If the model is correct, then Lrp activity should play a role in setting the level of ppGpp under growth conditions that require de novo synthesis of certain amino acids. However, Lrp is a pleiotropic regulator and efforts to study it have long been frustrated by the lack of a straightforward phenotype (Lin et al., 1992). For example, one of the few known growth phenotypes of Lrp- mutants is marginally slower growth in glucose minimal medium (Lin et al., 1992; Tani et al., 2002). Thus, while Lrp is clearly implicated in regulation of many amino acid biosynthetic genes, Δlrp strains are not amino acid auxotrophs. These considerations led us to design an experimental regime in which starvation for all amino acids could be abruptly triggered. The system we devised begins with cells in balanced growth in MOPS medium containing glucose plus all amino acids in abundance. Rapid collection and resuspension of the cells in glucose minimal medium disturbs the homeostasis previously maintained by the cells (i.e. the needs met by the originally available exogenous amino acids are suddenly unmet). The onset of amino acid starvation would be expected to stimulate robust, transient ppGpp accumulation. Proper induction of the Lrp regulon (along with other metabolic pathways) should eventually lead to endogenous biosynthesis of amino acids. In this framework, Lrp allows the cell to efficiently achieve a new homeostatic balance, resulting in the lowering of the ppGpp level and the resumption of growth. Accordingly, the RpoS regulon, which would be induced as ppGpp reaches high levels, should also return to a low level of expression as growth resumes. The model also suggests that under these amino acid starvation conditions, a strain lacking Lrp would also rapidly accumulate ppGpp. However, Δlrp cells would be impacted in their ability to biosynthesize some amino acids and would thus continue to maintain a high level of ppGpp.

To examine the role of Lrp in influencing the ppGpp level in response to abrupt amino acid starvation, we grew 50 ml cultures in minimal glucose medium containing double the concentration of 18 amino acids (tyrosine and isoleucine were also included at normal levels). At an OD of ∼0.3–0.4 cells were rapidly collected on glass fibre filters and resuspended in glucose minimal medium (Fig. 7). In the WT, this treatment triggered a growth lag of ∼160 min. In contrast, growth was arrested in the Lrp mutant (outgrowth occurred only after many hours, i.e. overnight). Under these conditions, the nutritional downshift resulted in a very high level of ppGpp accumulation in both strains, peaking at 30 min post-filtration at ∼2400 pmol ml−1 OD−1. This high level presumably reflects starvation for all 20 amino acids, as opposed to a single amino acid in our previously described experiments (Figs 4 and 6). In the WT, growth resumed as the ppGpp dropped to a level below ∼400 pmol ml−1 OD−1, which ultimately fell below 100 pmol ml−1 OD−1. In contrast, in the Lrp- mutant the ppGpp level declined only to ∼700 pmol ml−1 OD−1, a level comparable with that reached by cells starved for isoleucine (Fig. 4). This level was maintained for at least 270 min after amino acid starvation was induced. Thus, in accordance with the proposed model (Fig. 5), we observe that under conditions that lead to amino acid starvation, Lrp activity can influence the ppGpp level. Moreover, the proposed model not only suggested a role for Lrp in modulating ppGpp level, but also led to discovery of a new phenotype for Lrp- cells, i.e. that they fail to recover in a timely manner from abrupt, comprehensive amino acid starvation.

image

Figure 7. Lrp activity influences ppGpp level. A. Growth curves of WT (black) and Δlrp (blue) strains grown in MOPS medium with 2× the normal amount of 18 amino acids (tyrosine and isoleucine were included at normal levels). Cells were collected on filters and resuspended in MOPS minimal medium at T0. The difference in culture density following filtration is due to Lrp cells adhering more readily to glass fibre filters than the WT. B. ppGpp kinetics of WT and Δlrp strains. In the WT, ppGpp accumulated rapidly after filtration and remained high until growth resumed around 150 min. ppGpp declined to ∼700 pmol ml−1 OD−1 in the Δlrp strain, which did not resume growth for many hours. This figure is available in colour online.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

In this report, we examined the Lrp and RpoS transcription networks as components of the stringent response. We experimentally determined the Lrp and RpoS regulons induced in response to isoleucine starvation and found broad regulatory overlap between these gene systems and the ppGpp regulon (Fig. 2). Next we elucidated the timing of induction of the Lrp and RpoS networks in a microarray time-course obtained as cells gradually starved for isoleucine (Fig. 3). We found that the Lrp regulon was induced before the RpoS regulon (Fig. 4). The response times of the two networks corresponded to different threshold levels of ppGpp accumulation (Fig. 4), suggesting a model in which the Lrp regulon requires a relatively low level of ppGpp for its induction compared with the RpoS regulon (Fig. 5). Based on this model, we predicted that accelerating the ppGpp accumulation rate would reduce the difference in the response times of these two regulons. Using gene reporters for the Lrp and RpoS regulons, we found that when ppGpp accumulates rapidly (in < 15 min as opposed to 100 min), the difference in response times of the two regulons was irresolvable (Fig. 6). Finally, our proposed model postulates a feed-back loop wherein Lrp activity indirectly regulates the ppGpp level by controlling expression of amino acid biosynthesis genes and hence eliminating amino acid deprivation. In support of this hypothesis, we observed that an Lrp mutant maintained an abnormally high level of ppGpp after a downshift from medium replete with amino acids to glucose minimal medium (Fig. 7). Taken together, these data suggest a framework for understanding the regulatory structure that governs gene expression during the stringent response.

As bacteria encounter declining nutrient availability they must judiciously allocate cellular resources. The model presented in Fig. 5 considers how cells might partition their gene expression (and hence manage major cellular processes) under three conditions along the feast to famine gradient. In nutrient-rich environments the ppGpp level is low and both biosynthetic (e.g. amino acid biosynthetic pathways) and stress survival mechanisms are kept at a minimal level of activity. Under intermediate nutrient limitation, or ‘hunger’ conditions (such as those found in minimal medium) a slightly elevated level of ppGpp is maintained. The data presented here suggest that biosynthetic processes require only a relatively low level of ppGpp for their induction; with this sensitivity being theoretically mediated through promoter structure and/or signalling through other regulatory proteins. We hypothesize for example that ppGpp reprogrammed RNAP and Lrp act simultaneously at a given promoter, although we cannot rule out the possibility that synergy between the two might be temporally separated. Under these hunger conditions, de novo biosynthesis of amino acids resets ppGpp at a relatively low level, a new homeostatic balance is achieved, and growth can continue. Should an essential nutrient be exhausted, starvation occurs, resulting in high levels of ppGpp. Only when ppGpp reaches a high level is the general stress response initiated and the cell transitions into survival mode. Thus, biosynthetic and stress responses are tuned to require different levels of ppGpp, which in turn signify the overall physiological state of the cell.

What threshold levels are required for different components of the stringent response to function? This question must be considered in light of ppGpp levels associated with different physiological conditions. In quickly growing cells (∼30 min doubling time, with glucose and casamino acids) the basal level of ppGpp has been measured at ∼20–40 pmols OD−1 (Lazzarini et al., 1971; Sokawa et al., 1975; Ryals et al., 1982). This basal level of ppGpp ranges up to ∼80 pmols OD−1 for slower growing cells (∼200 min doubling−1, with alanine as the sole carbon and energy source) (Lazzarini et al., 1971; Ryals et al., 1982). In contrast, conditions that lead to growth arrest often prompt ppGpp levels in excess of ∼800 pmols OD−1[(Lazzarini et al., 1971; Ryals et al., 1982; VanBogelen et al., 1987), and this report]. Our ppGpp measurements under balanced growth conditions are in general agreement with these published levels (resumed growth of the WT in minimal medium, Fig. 7 and data not shown); however, our HPLC method is not sensitive enough to reliably distinguish small differences in ppGpp associated with different rates of rapid, balanced growth.

The overarching conclusion drawn from these measurements is that the range of basal ppGpp levels found during growth, even slow growth, is small compared with ppGpp levels found during growth arrest. This distinction is important as it implies that even the relatively low level of ppGpp found during slow growth (e.g. < 80 pmols OD−1) is sufficient to allow for induction of biosynthetic genes (which must be induced in order to grow in minimal medium). We do not contend that the ppGpp levels observed at the times of Lrp and RpoS regulon induction represent the exact ppGpp levels required for induction of the respective regulons. Rather, we suggest that low basal levels of ppGpp present during slow growth [20–80 pmols OD−1 as shown by (Lazzarini et al., 1971; Ryals et al., 1982)] are sufficient to allow induction of the Lrp regulon. Accordingly, we observed that induction of the Lrp regulon in response to isoleucine limitation occurred while the cells were still actively growing, not during transition to stationary phase (Fig. 4). Thus, we expect that the timing of induction of the Lrp regulon observed here is determined primarily by signalling through Lrp (though the specific signal remains unclear) and that the ppGpp level is sufficient for the Lrp-dependent response to occur. One possibility is that Lrp serves to recruit RNAP and/or maintain RNAP at the promoter, while ppGpp expedites transition to the transcription initiation complex as described previously (Paul et al., 2005).

From our measurements we conclude that ppGpp levels in excess of 400 pmols ml−1 OD−1 correspond to growth arrest. Specifically, this conclusion is based on our observations that cells starving for isoleucine exhibit little growth after this level of ppGpp is reached (Fig. 4), and that growth of the WT resumed after ppGpp dropped below this level (Fig. 7). This level of ppGpp also correlated with the response time of the RpoS regulon (Figs 4 and 6). Thus, the ‘relatively high’ level of ppGpp required for full RpoS regulon induction is likely to be > 400 pmol ml−1 OD−1. The experiments described here do not imply a mechanism for how the requirement for a higher amount of ppGpp might be built into the RpoS signalling pathway. We hypothesize that the manifold levels of control exerted by ppGpp on RpoS, including inhibition of proteolysis via IraP (Bougdour and Gottesman, 2007), and/or sigma factor competition (Jishage et al., 2002) may play a role in defining the threshold level of ppGpp required for robust RpoS activity (considered in detail below). Because ppGpp is also implicated in control of RpoS at a transcriptional level (Lange et al., 1995), the ppGpp/RpoS signalling pathway can be considered a coherent feed-forward loop (i.e. ppGpp is required for rpoS transcription and for induction of the RpoS regulon). Such feed-forward loops can introduce a time delay between stimulus and response and insulate systems from short-lived spikes in stimulus (noise) (Mangan and Alon, 2003).

Competing models seek to explain the global redistribution of RNAP by ppGpp. In the affinity model, RNAP liberated from stable RNA synthesis by the action of ppGpp results in an increase in free RNAP, thus allowing for increased transcription of amino acid biosynthetic or stress genes, which compete more effectively for the larger pool of free RNAP (Zhou and Jin, 1998; Barker et al., 2001). In contrast, a recently proposed modified saturation model contends that the result of ppGpp accumulation is a net decrease in the size of the free RNAP (Eσ70) pool because of increased competitiveness of alternative sigma factors for core RNAP (Gummesson et al., 2009). The overall result is decreased transcription of unsaturated stable RNA promoters, while the ability of saturated amino acid biosynthesis and stress response promoters across the genome to be activated presumably remains unchanged (or at least is not negatively impacted by ppGpp accumulation). Previously we observed the ppGpp0 strain induces just as many genes as the WT upon isoleucine starvation, although with radical differences in which genes were induced (Traxler et al., 2008), a finding that is inconsistent with predictions of the affinity model. Thus we hypothesized that the rather slow onset of isoleucine starvation relieved the constraints implied by the affinity model (Traxler et al., 2008). More consistent with our results is the saturation model advanced by Gummesson and co-workers who examined expression of ribosome component genes vs. stress and amino acid biosynthetic genes. The model we present here further parses expression within the latter group, in that we suggest that for effective induction of stress genes (RpoS-dependent) and amino acid biosynthetic genes requires different levels of ppGpp accumulation.

Our observation that an Lrp mutant had a sustained high level of ppGpp following an amino acid downshift illustrates another critical connection within the architecture of the stringent response (Fig. 7). This finding implies that ppGpp accumulation is subject to feed-back control resulting from the activity of biosynthetic enzymes via the endogenous amino acid pool. In this scenario, when exogenous amino acids first become limiting, ppGpp accumulates and works in conjunction with other regulators such as Lrp to activate transcription of biosynthetic genes. The resulting increase in endogenous amino acids lowers the ppGpp level until growth can resume. Because the ppGpp level also influences induction of the RpoS regulon, the feed-back loop proposed above describes a mechanism for homeostatic (metabolic) control of the general stress response. The model we propose attributes the increased ppGpp level to increased ppGpp synthesis in response to prolonged amino acid limitation in the Lrp mutant strain. It is possible that increased levels of ppGpp in the Lrp mutant could also result from inhibition of SpoT ppGpp hydrolase activity. We currently favour the model in Fig. 5 because: (i) Lrp is known to respond to amino acid availability and influence amino acid biosynthesis (Cho et al., 2008) and (ii) the available data do not explain how deletion of Lrp might influence hydrolytic activity of SpoT.

Several advantages of the proposed regulatory network structure are apparent. First, the low level of ppGpp required for induction of biosynthetic genes allows for rapid and flexible response to metabolic perturbations. Second, the high threshold level of ppGpp required for induction of the general stress response could work to buffer against erroneous activation of a large number of stationary phase genes in situations of transient or easily remediated nutritional stresses. Thus, only cells experiencing acute interruption of metabolic homeostasis develop a full-fledged general stress response. The discrete tuning of each of the proposed regulatory loops allows for multiple response systems to be tethered to a single indicator of cellular physiology. Moreover, as the ppGpp level required for robust induction of the general stress response is set appropriately high, a range of signalling through RpoS is likely possible across levels of ppGpp that still allow for active growth. Indeed such a range of induction of the RpoS regulon is observed in the published transcriptome data sets of cells growing at several different rates (Liu et al., 2005). The evidence presented in this report supplies a rationale for how bacterial cells utilize ppGpp to partition global gene expression for rapid growth, biosynthetic processes, and stress response across environmental conditions ranging from feast to famine. These findings also underscore the idea that the stringent response is not an all-or-nothing phenomenon, but is a rheostat that can be dialled up or down depending on the richness of the nutritional environment. At a fundamental level, the results illustrate how bacterial cells can utilize a single indicator of cellular physiological state (ppGpp) in combination with discretely calibrated regulatory systems to: (i) establish a new homeostatic balance or (ii) protect the cell in the event that homeostasis cannot be maintained during times of starvation.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Bacterial strains and growth conditions

All strains used in this study were derivatives of E. coli K-12 strain MG1655. A list of strains and plasmids used appears in Table 1. Mutant strains were constructed by using a modified version of the method described by Datsenko and Wanner (Datsenko and Wanner, 2000). The ΔrelAΔspoT (ppGpp0) and ΔrpoS strains have been used in previous studies (Traxler et al., 2006; 2008). Marker-less mutants were made by removal of antibiotic cassettes using surrounding FRT sites and confirmed by sequencing and PCR. ppGpp0 cultures were routinely checked for suppressor mutants by checking for growth on glucose minimal medium at the conclusion of experiments.

Table 1.  Strains used in this study.
StrainRelevant genotypeSource or derivation
MG1655 seqWTSequenced strain
MFT704ΔrelA::FRT, ΔspoT::FRTTraxler et al. (2008)
ΔrpoSΔrpoS::kanRTraxler et al. (2006)
MFT760Δlrp::FRTThis study
MFT720MG1655 (pUA66, PleuL::GFP)This study
MFT721MG1655 (pUA66, PdadA::GFP)This study
MFT722MG1655 (pUA66, PyahO::GFP)This study
MFT723MG1655 (pUA66, PwrbA::GFP)This study

For array studies the WT and isogenic mutants were cultured in a 2 l Biostat B fermenter (Braun Biotech) containing 1 l of morpholinepropanesulfonic acid (MOPS) medium (Neidhardt et al., 1974) with 2.0 g l−1 glucose and amino acids at the concentrations described in Wanner et al. (1977), with the exception that isoleucine was included at 60 µM instead of the usual 400 µM. The growth medium did not contain uracil, which has been shown to stimulate growth of E. coli MG1655, which has an rph frameshift mutation (Jensen, 1993). However, inclusion of uracil had no effect on logarithmic growth, growth arrest caused by isoleucine starvation or rescue of growth by addition of isoleucine (data not shown). The temperature was maintained at 37°C, and pH was kept constant at 7.4 by the addition of 1 M NaOH. The dissolved oxygen level was maintained above 40% of saturation by adjusting the agitation speeds in the range of 270–500 r.p.m. with fixed 1.5 l min−1 air flow. Growth was monitored as absorbance at 600 nm with a Beckman-Coulter DU 800 spectrophotometer.

For GFP isoleucine starvation experiments, strains were grown in 50 ml cultures in 500 ml flasks in medium as described above. Flasks were incubated with shaking at 250 r.p.m. at 37°C. Kanamycin was included at 25 µg ml−1 to maintain pUA66 derivatives. 50 ml cultures were inoculated to a calculated starting OD of 0.0015 from overnight 5 ml seed cultures (which were not isoleucine limited). To induce abrupt isoleucine starvation, 50 ml cultures were started in medium replete with isoleucine (400 µm). During log growth (OD 0.3–0.4), cells were rapidly harvested on a glass fibre filter atop a vacuum tower. Filters were immediately dropped into identical flasks with pre-warmed MOPS medium containing all amino acids except isoleucine. Filters were removed after 10 min of shaking incubation. For quantification of ppGpp in the Lrp- mutant (Fig. 7), strains were grown in MOPS medium with double the normal amount of all amino acids except tyrosine and isoleucine (which were provided at normal levels).

Nucleotide extraction and ppGpp quantification

Nucleotides were extracted as described, with minor modifications (Bochner and Ames, 1982). For fermenter cultures, 5 ml of culture was sampled directly into a 15 ml round-bottom tube containing 0.5 ml of 11 M formic acid. For 50 ml flask cultures, 1 ml of culture was pipeted into 1.5 ml eppendorf tubes containing 0.1 ml of 11 M formic acid. The sample was vigorously mixed and chilled on ice. 1 ml aliquots of this mixture were incubated at 0°C in an ice water bath for 45 min with periodic vortexing. These 1 ml samples were centrifuged at 4°C at 6000 r.p.m. for 5 min. The supernatant was then filtered through 0.2 µm filters and stored at −20° until HPLC analysis.

ppGpp was quantified by anion exchange HPLC using a Mono Q 5/50 GL column (GE Healthcare). Absorbance at 254 nm was used to detect eluted nucleotides. 250 µl of supernatant was injected under initial conditions of 95% 20 mM Tris (pH 8.0) and 5% 20 mM Tris + 1.5 M sodium formate (pH 8.0). This initial condition was maintained for 5 min. Over a period of 30 min, the level of sodium formate buffer was ramped up to 65%. ppGpp was identified as a peak that eluted at ∼28 min (or ∼52% 1.5 M sodium formate buffer). Samples were run in duplicate for at least two separate time-course experiments. Combined results for at least two experiments are shown. ppGpp standard was purchased from TriLink Biosciences. Standard curves established that the linear range of detection of ppGpp was 10 nM to 100 µM.

Microarray analysis

Cells were sampled directly from the fermenter into an equal volume of ice-cold RNAlater (Ambion) and total RNA was extracted using Qiagen RNeasy Minikits with optional DNAse treatment steps. RNA was checked for integrity by gel electrophoresis and maintained in a 2:1 dilution of EtOH at −80°C until labelling. RNA was converted to cDNA by first-stranded synthesis using Superscript II (Invitrogen) and random hexamers, according to the manufacturer's specifications. The cDNA was fragmented and biotinylated (Enzo Kit, Roche Diagnostics) according to the Affymetrix prokaryotic labelling protocol.

The microarrays used for single time-point mutant analysis were custom built Affymetrix GeneChips containing probes for several prokaryotic genomes including E. coli K12 MG1655, E. coli O157:H7 EDL933, Bacteriodes thetaiotaomicron VPI-5482, Enterococcus faecalis V583, Salmonella typhimurium LT2 and Bacillus anthracis, as described previously (Traxler et al., 2008). Biotinylated samples were prepared according to the manufacturer's instructions and hybridized for 16 h at 60°C. Hybridized arrays were stained using Affymetrix protocol ProkGE_WS2v2-450 (for mutant analysis) and Mini_prok2v1-450 for E. coli 2.0 chips (wt time-course). Stained microarrays were scanned and the raw data files (.cel) were further analysed using RMA processing with quartile normalization (Irizarry et al., 2003). WT and mutant samples were duplicated biologically and technically (n = 2 arrays for each strain); r2 was > 0.95 for all replicates.

For the 12-point WT isoleucine starvation time-course (Figs 3 and 4), Affymetrix E. coli 2.0 genome arrays were used as directed by the manufacturer. Data shown are based on a single array for each time point across the time-course. Array data from two biological replicates sampled from rapidly growing WT cells grown in medium replete all amino acids were averaged to serve as the control values for comparison with the time-course arrays (r2 = 0.998 for control arrays). Three-parameter sigmoidal regressions were plotted through averaged data points collected for each strain to obtain the response times. Statistical analysis was done using SigmaPlot 8.0. We considered genes to be significantly induced or repressed if the absolute value of the expression ratio was > twofold (Wren and Conway, 2006). Hierarchical clustering algorithms were implemented in DecisionSite for Functional Genomics (Spotfire). The microarray data were deposited at GEO (GEO accession: GSE11087).

GFP experiments

Growth conditions for GFP strains are described under ‘bacterial strains and growth conditions’. 100 µl samples for fluorescence reading were pipetted into transparent 96-well microplates (Corning) in triplicate. Wells contained 1 µl of 25 mg ml−1 chloramphenicol to quench protein synthesis. Fluorescence (485 nm excitation and 520 nm emission) and optical densities (600 nm) were read on a FLUOstar Optima fluorimeter (BMG Labtech). For each fluorescent strain, uninduced control values were obtained as the average fluorescence across the time of log growth (OD ∼0.2 to ∼1.0) in medium with all 20 amino acids available at non-limiting levels. Control values were based on two biologically replicated growth curves for each strain. Experimental fluorescence values were compared with these control values to yield Log2 expression ratios. Plots shown in Fig. 6 contain data from three experimental runs for each strain. Three-parameter sigmoidal regressions were plotted through all data points collected for each strain to obtain the response times. Statistical analysis was done using SigmaPlot 8.0.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The authors wish to thank Joe Grissom for invaluable bioinformatics support and Katherine Lemon for critical comments on the manuscript. This research was supported by Public Health Service Grant AI72401.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Aberg, A., Fernandez-Vazquez, J., Cabrer-Panes, J.D., Sanchez, A., and Balsalobre, C. (2009) Similar and divergent outcomes of ppGpp and DksA deficiencies on transcription in Escherichia coli. J Bacteriol 191: 32263236.
  • Alon, U. (2007) Network motifs: theory and experimental approaches. Nat Rev Genet 8: 450461.
  • Artsimovitch, I., Patlan, V., Sekine, S., Vassylyeva, M.N., Hosaka, T., Ochi, K., et al. (2004) Structural basis for transcription regulation by alarmone ppGpp. Cell 117: 299310.
  • Barker, M.M., Gaal, T., and Gourse, R.L. (2001) Mechanism of regulation of transcription initiation by ppGpp. II. Models for positive control based on properties of RNAP mutants and competition for RNAP. J Mol Biol 305: 689702.
  • Battesti, A., and Bouveret, E. (2006) Acyl carrier protein/SpoT interaction, the switch linking SpoT-dependent stress response to fatty acid metabolism. Mol Microbiol 62: 10481063.
  • Bochner, B.R., and Ames, B.N. (1982) Complete analysis of cellular nucleotides by two-dimensional thin layer chromatography. J Biol Chem 257: 97599769.
  • Bougdour, A., and Gottesman, S. (2007) ppGpp regulation of RpoS degradation via anti-adaptor protein IraP. Proc Natl Acad Sci USA 104: 1289612901.
  • Bremer, H., and Dennis, P.P. (1996) Modulation of chemical composition and other parameters of the cell by growth rate. In Escherichia coli and Salmonella: Cellular and Molecular Biology. Neidhardt, F.C., Curtiss, R., III, Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., et al. (eds). Washington, DC: ASM Press, pp. 15531569.
  • Cashel, M., Gentry, D.R., Hernandez, V.J., and Vinella, D. (1996) The stringent response. In Escherichia coli and Salmonella: Cellular and Molecular Biology. Neidhardt, F.C., Curtiss, R., III, Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., et al. (eds). Washington, DC: ASM Press, pp. 14581496.
  • Cho, B.K., Barrett, C.L., Knight, E.M., Park, Y.S., and Palsson, B.O. (2008) Genome-scale reconstruction of the Lrp regulatory network in Escherichia coli. Proc Natl Acad Sci USA 105: 1946219467.
  • Datsenko, K.A., and Wanner, B.L. (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97: 66406645.
  • Durfee, T., Hansen, A.M., Zhi, H., Blattner, F.R., and Jin, D.J. (2008) Transcription profiling of the stringent response in Escherichia coli. J Bacteriol 190: 10841096.
  • Gummesson, B., Magnusson, L.U., Lovmar, M., Kvint, K., Persson, O., Ballesteros, M., et al. (2009) Increased RNA polymerase availability directs resources towards growth at the expense of maintenance. EMBO J 28: 22092219.
  • Haugen, S.P., Ross, W., and Gourse, R.L. (2008) Advances in bacterial promoter recognition and its control by factors that do not bind DNA. Nat Rev Microbiol 6: 507519.
  • Hengge-Aronis, R. (1996) Regulation of gene expression during entry into stationary phase. In Escherichia coli and Salmonella: Cellular and Molecular Biology. Neidhardt, F.C., Curtiss, R., III, Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., et al. (eds). Washington, DC: ASM Press, pp. 14971512.
  • Ibanez-Ruiz, M., Robbe-Saule, V., Hermant, D., Labrude, S., and Norel, F. (2000) Identification of RpoS (sigma(S))-regulated genes in Salmonella enterica serovar typhimurium. J Bacteriol 182: 57495756.
  • Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Scherf, U., and Speed, T.P. (2003) Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4: 249264.
  • Jensen, K.F. (1993) The Escherichia coli K-12 ‘wild types’ W3110 and MG1655 have an rph frameshift mutation that leads to pyrimidine starvation due to low pyrE expression levels. J Bacteriol 175: 34013407.
  • Jishage, M., Kvint, K., Shingler, V., and Nystrom, T. (2002) Regulation of sigma factor competition by the alarmone ppGpp. Genes Dev 16: 12601270.
  • Lacour, S., and Landini, P. (2004) SigmaS-dependent gene expression at the onset of stationary phase in Escherichia coli: function of sigmaS-dependent genes and identification of their promoter sequences. J Bacteriol 186: 71867195.
  • Lange, R., Fischer, D., and Hengge-Aronis, R. (1995) Identification of transcriptional start sites and the role of ppGpp in the expression of rpoS, the structural gene for the sigma S subunit of RNA polymerase in Escherichia coli. J Bacteriol 177: 46764680.
  • Lawther, R.P., Calhoun, D.H., Adams, C.W., Hauser, C.A., Gray, J., and Hatfield, G.W. (1981) Molecular basis of valine resistance in Escherichia coli K-12. Proc Natl Acad Sci USA 78: 922925.
  • Lazzarini, R.A., Cashel, M., and Gallant, J. (1971) On the regulation of guanosine tetraphosphate levels in stringent and relaxed strains of Escherichia coli. J Biol Chem 246: 43814385.
  • Lin, R., D'Ari, R., and Newman, E.B. (1992) Lambda placMu insertions in genes of the leucine regulon: extension of the regulon to genes not regulated by leucine. J Bacteriol 174: 19481955.
  • Liu, M., Durfee, T., Cabrera, J.E., Zhao, K., Jin, D.J., and Blattner, F.R. (2005) Global transcriptional programs reveal a carbon source foraging strategy by Escherichia coli. J Biol Chem 280: 1592115927.
  • Loewen, P.C., and Hengge-Aronis, R. (1994) The role of the sigma factor sigma S (KatF) in bacterial global regulation. Annu Rev Microbiol 48: 5380.
  • Mangan, S., and Alon, U. (2003) Structure and function of the feed-forward loop network motif. Proc Natl Acad Sci USA 100: 1198011985.
  • Neidhardt, F.C., Bloch, P.L., and Smith, D.F. (1974) Culture medium for enterobacteria. J Bacteriol 119: 736747.
  • Paul, B.J., Berkmen, M.B., and Gourse, R.L. (2005) DksA potentiates direct activation of amino acid promoters by ppGpp. Proc Natl Acad Sci USA 102: 78237828.
  • Potrykus, K., and Cashel, M. (2008) (p)ppGpp: still magical? Annu Rev Microbiol 62: 3551.
  • Rhodius, V.A., Suh, W.C., Nonaka, G., West, J., and Gross, C.A. (2006) Conserved and variable functions of the sigmaE stress response in related genomes. PLoS Biol 4: e2.
  • Rutherford, S.T., Villers, C.L., Lee, J.H., Ross, W., and Gourse, R.L. (2009) Allosteric control of Escherichia coli rRNA promoter complexes by DksA. Genes Dev 23: 236248.
  • Ryals, J., Little, R., and Bremer, H. (1982) Control of rRNA and tRNA syntheses in Escherichia coli by guanosine tetraphosphate. J Bacteriol 151: 12611268.
  • Sarubbi, E., Rudd, K.E., and Cashel, M. (1988) Basal ppGpp level adjustment shown by new spoT mutants affect steady state growth rates and rrnA ribosomal promoter regulation in Escherichia coli. Mol Gen Genet 213: 214222.
  • Sokawa, Y., Sokawa, J., and Kaziro, Y. (1975) Regulation of stable RNA synthesis and ppGpp levels in growing cells of Escherichia coli. Cell 5: 6974.
  • Tani, T.H., Khodursky, A., Blumenthal, R.M., Brown, P.O., and Matthews, R.G. (2002) Adaptation to famine: a family of stationary-phase genes revealed by microarray analysis. Proc Natl Acad Sci USA 99: 1347113476.
  • Traxler, M.F., Chang, D.E., and Conway, T. (2006) Guanosine 3′,5′-bispyrophosphate coordinates global gene expression during glucose-lactose diauxie in Escherichia coli. Proc Natl Acad Sci USA 103: 23742379.
  • Traxler, M.F., Summers, S.M., Nguyen, H.T., Zacharia, V.M., Hightower, G.A., Smith, J.T., and Conway, T. (2008) The global, ppGpp-mediated stringent response to amino acid starvation in Escherichia coli. Mol Microbiol 68: 11281148.
  • VanBogelen, R.A., Kelley, P.M., and Neidhardt, F.C. (1987) Differential induction of heat shock, SOS, and oxidation stress regulons and accumulation of nucleotides in Escherichia coli. J Bacteriol 169: 2632.
  • Vinella, D., Albrecht, C., Cashel, M., and D'Ari, R. (2005) Iron limitation induces SpoT-dependent accumulation of ppGpp in Escherichia coli. Mol Microbiol 56: 958970.
  • Vrentas, C.E., Gaal, T., Berkmen, M.B., Rutherford, S.T., Haugen, S.P., Vassylyev, D.G., et al. (2008) Still looking for the magic spot: the crystallographically defined binding site for ppGpp on RNA polymerase is unlikely to be responsible for rRNA transcription regulation. J Mol Biol 377: 551564.
  • Wanner, B.L., Kodaira, R., and Neidhart, F.C. (1977) Physiological regulation of a decontrolled lac operon. J Bacteriol 130: 212222.
  • Weber, H., Polen, T., Heuveling, J., Wendisch, V.F., and Hengge, R. (2005) Genome-wide analysis of the general stress response network in Escherichia coli: sigmaS-dependent genes, promoters, and sigma factor selectivity. J Bacteriol 187: 15911603.
  • Wendrich, T.M., Blaha, G., Wilson, D.N., Marahiel, M.A., and Nierhaus, K.H. (2002) Dissection of the mechanism for the stringent factor RelA. Mol Cell 10: 779788.
  • Wren, J.D., and Conway, T. (2006) Meta-analysis of published transcriptional and translational fold changes reveals a preference for low-fold inductions. OMICS 10: 1527.
  • Xiao, H., Kalman, M., Ikehara, K., Zemel, S., Glaser, G., and Cashel, M. (1991) Residual guanosine 3′,5′-bispyrophosphate synthetic activity of relA null mutants can be eliminated by spoT null mutations. J Biol Chem 266: 59805990.
  • Yang, W., Ni, L., and Somerville, R.L. (1993) A stationary-phase protein of Escherichia coli that affects the mode of association between the trp repressor protein and operator-bearing DNA. Proc Natl Acad Sci USA 90: 57965800.
  • Zaslaver, A., Mayo, A.E., Rosenberg, R., Bashkin, P., Sberro, H., Tsalyuk, M., et al. (2004) Just-in-time transcription program in metabolic pathways. Nat Genet 36: 486491.
  • Zaslaver, A., Bren, A., Ronen, M., Itzkovitz, S., Kikoin, I., Shavit, S., et al. (2006) A comprehensive library of fluorescent transcriptional reporters for Escherichia coli. Nat Methods 3: 623628.
  • Zhi, J., Mathew, E., and Freundlich, M. (1999) Lrp binds to two regions in the dadAX promoter region of Escherichia coli to repress and activate transcription directly. Mol Microbiol 32: 2940.
  • Zhou, Y.N., and Jin, D.J. (1998) The rpoB mutants destabilizing initiation complexes at stringently controlled promoters behave like ‘stringent’ RNA polymerases in Escherichia coli. Proc Natl Acad Sci USA 95: 29082913.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
MMI_7498_sm_TableS1.pdf213KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.