INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Septation is the bacterial equivalent of cytokinesis and involves the regulated formation of a specialized cell wall at the midpoint of the dividing cell during vegetative growth. In Escherichia coli, loss of septation causes the formation of multinucleate cell filaments and eventual death. FtsZ is a GTPase and is the most abundant of several cell division proteins whose activity is essential for septation (for reviews, see references 29 and 51). FtsZ appears to guide septal invagination. Its subcellular distribution cycles with growth, coalescing at the division site during cytokinesis and then dispersing into the cytosol afterward (33, 40). This bimodal property could reflect a cyclic progression of FtsZ through active and inactive states. Much evidence supports the idea that the cellular activity of FtsZ depends on a dynamic self-polymerization reaction involving changes in localized concentration and a GTPase cycle akin to the tubulin proteins of eukaryotic cells (4, 6, 11, 35, 44, 60, 65). The thermosensitive allele ftsZ84 has been the subject of numerous investigations into the nature of septation since its initial description more than 2 decades ago (47). Genetic, biochemical, and microscopic studies of ftsZ84 strains have helped to generate a model that assigns a structural, and perhaps kinetic, role to FtsZ throughout cytokinesis, as well as a regulatory role for the commencement of septation. The process of discovery and characterization of additional factors affecting FtsZ continue to refine our understanding of its function in the cell cycle.

Purified FtsZ84 protein has reduced GTPase activity at elevated temperature, and cell filamentation is the consequence of this defect in vivo. This filamentation is relieved by growth media of high osmotic strength, a phenomenon originally called "salt repair" (47). The mechanism of salt repair for ftsZ84 is not understood. Conditional lethality imparted by ftsZ84 is genetically suppressed by increased dosages of some genes. Most notably, when present in multiple copies, the ftsZ84 allele suppresses itself, verifying that lethality owes to its reduced activity (39). Three extragenic loci have been characterized as suppressors of ftsZ84 when present in high dosages: the gene encoding the transcription factor SdiA (66) and two genes for regulators of capsular polysaccharide synthesis, rcsB and rcsF (20, 21). The connection between capsule gene regulation and septation is not clear. Too much FtsZ activity also disrupts normal septation (13, 67). Interestingly, this condition of FtsZ excess suppresses filamentation caused by abnormally high expression of another cell division gene, ftsA, which suggests the necessity of a dosage equilibrium between FtsA and FtsZ for their proper functioning (10, 12, 13).

The ftsZ and ftsA genes and another cell division gene, ftsQ, form a complex operon which comprises the distal end of a cluster of 16 aligned and overlapping genes concerning cell division and cell wall metabolism (3). The ftsQAZ operon contains several promoters, and there is genetic evidence for the influence of factors at some of these promoters. The majority of transcripts containing ftsZ appear to originate from promoters located 5' of the operon (19, 69). One of these promoters (Qp2) is up-regulated by SdiA (66), and another (Qp1) is positively regulated by the stationary growth phase sigma factor ^s (2, 56). Also contributing to FtsZ expression are four internal promoters (Ap, Zp2, Zp3, and Zp4) located within the ftsQ and ftsA coding regions (30) and antisense RNAs that are complementary to the 5' region of ftsZ (14, 50, 58). The activities of some of these promoters vary inversely with the growth rate (2), and the levels of ftsZ transcripts appear to oscillate with the cell cycle (19). The short-lived nucleotide guanosine tetraphosphate has recently received attention as a possible effector of cell division (61-63). Guanosine tetraphosphate and the related compound guanosine pentaphosphate, together referred to herein simply as ppGpp for brevity, function as an alarmone system and are believed to integrate cellular responses to various forms of nutrient stress (8). Natural synthesis of ppGpp in Escherichia coli occurs exclusively from the RelA and SpoT proteins, and the principal contribution is made by RelA. SpoT normally degrades ppGpp but has synthetic activity under certain conditions (27, 68). In the classic stringent response to the presence of uncharged tRNA, ppGpp increases the fidelity of protein translation (36). ppGpp also globally regulates gene expression by affecting initiation or pausing of RNA transcription and can have a positive or negative effect, depending on the targeted promoter. The identification of mutant ⁷⁰, , and ' subunits of RNA polymerase that mimic the ppGpp-induced state supports the idea that this signal exerts an important effect at the level of transcription (8, 26).

Until now, no mutation has been reported that suppresses the high-temperature lethality of ftsZ84 on Luria-Bertani (LB) medium. Several chromosomal mutations are identified here that restore septation and high-temperature growth on LB medium to strains having the ftsZ84 mutation. Four of these suppressors are loss-of-function alleles for different genes that regulate general nitrogen assimilation and glutamine metabolism during nitrogen-restricted growth. The data presented here indicate that the likely mechanism of suppression by these mutations involves RelA-dependent synthesis of the nutrient starvation signal ppGpp in response to glutamine limitation. This introduces evidence associating ppGpp with glutamine regulation on LB medium.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains, plasmids, bacteriophages, and growth media. The bacterial strains, plasmids, and bacteriophages used in this study are listed in Table 1. The ftsZ84 strains used in this study have the genetic background of W3110 (lac)U169 (Court laboratory strain WJW45), except as specifically noted otherwise. Parental strain BSP610 and its derivatives were constructed by P1vir transductions using standard procedures (34) in the following steps. To make BSP610, the ftsZ84 allele of strain JFL100 was first linked to a leu::Tn10 marker and then cotransduced into WJW45 by selection for tetracycline resistance. Heat-sensitive transductants were then converted back to Leu⁺ prototrophy by using a P1 lysate of WJW45 as the donor and selection on minimal-salts medium plus glucose medium, with subsequent confirmation of heat sensitivity and tetracycline sensitivity. Mutations of glnA, glnD, glnE, glnF, glnG, and gltB were introduced into BSP610 and congenic FtsZ⁺ parental strain WJW45 by P1vir transductions from phage stocks prepared on donor strains containing the given mutation, employing selection for the drug resistance marker associated with each mutation. Unmarked mutations in glnL, glnB, and gltBDF were introduced, respectively, by transductions into the glnA, glyA, and glnF derivatives of BSP610 and WJW45 by selection for prototrophic conversion of these linked mutations.

Thermosensitivity tests. While the thermosensitivity of ftsZ84 is known to vary among laboratories, often requiring LB medium without added sodium chloride, we have found that our standard LB medium (10 g of Bacto Tryptone, 5 g of NaCl, 8 g of yeast extract; sterilized via autoclave) consistently produces filamentation and lethality at 42°C for ftsZ84 strains in our W3110 genetic background. We observed that other ftsZ84 strains of different genetic backgrounds were less sensitive on this LB medium, and so all heat sensitivity tests were performed on BSP610 and its derivatives. Heat sensitivities and filamentation were assayed by monitoring colony formation on LB plates and by microscopic inspection of suspended colonies or liquid LB cultures.

Assays for ftsQAZ operon expression. Reporter constructs containing ftsQAZ fusions to lacZ were introduced into tester strains either on low-copy-number plasmids (9) or in single copy on lambda prophages (Table 1). Phages BBP133 and BBP134 contain 342 nucleotides 5' of the ftsZ start codon, which includes promoter Zp2, along with the first 11 codons of ftsZ joined to lacZ in protein and operon fusion configurations, respectively. To make these, DNA was amplified from the chromosome by using the oligonucleotide primers gttgacgaattcaagcttcgccaacaaggggttaaacatcac and actttaggatccgcgtcattggtaagttccattggttcaaac and ULTma DNA polymerase (Perkin Elmer) and then digested with EcoRI and BamHI and cloned into plasmids pRS414 or pRS415 (55), making plasmids pBP133 and pBP134, respectively. The absence of potential errors acquired during cloning was confirmed by direct sequencing of plasmids with the ABI Prism DNA Sequencing kit (Perkin Elmer) and a 373A DNA sequencer (Perkin Elmer). DNA sequences were assembled and analyzed by using Sequencher 3.0 (Gene Codes Corp.) and the Genetics Computer Group (Madison, Wis.) package, version 8.0. Each fusion was recombined from a plasmid onto BDC531 to make phages BBP133 and BBP134 and used to infect BSP610, and prophage lysogens were screened for unit copy number as previously described (42). These strains were used as parental strains for all related strain constructions. -Galactosidase activities were assayed by method B of Miller (34) with the following modification. Cultures were harvested by twofold dilution into ice-cold stop solution comprised of Z buffer plus 60-µg/ml chloramphenicol and 0.04% sodium azide.

Quantitation of FtsZ protein. The steady-state level of FtsZ84 protein in various strains was measured by Western blot assay done by standard procedures (23) and the protocols specified by the manufacturers of the electrophoretic and immunologic reagents. Overnight LB cultures were refreshed by 200-fold dilution and grown at 30°C to an optical density at 600 nm of 0.2 and then shifted to 42°C and grown for 2.5 h. Aliquots were collected at equivalent points in their respective growth curves, and volumes were normalized by optical density (5 ml at an optical density at 600 nm of 0.2). Cells were collected by centrifugation, washed in stop solution (described above), frozen immediately on dry ice, and then stored frozen until later use. To prepare extracts, cell pellets were thawed in 0.5 ml of buffer (50 mM sodium phosphate [pH 7.4], 10 µl of 10-mg/ml phenylmethylsulfonyl fluoride [Boehringer Mannheim]), mixed gently, passed through another freeze-thaw cycle, and then disrupted by sonication. Total protein was estimated via protein assay (Bio-Rad), and then cell extracts were normalized among each other by total protein concentration. Extracts were serially diluted into sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer and boiled briefly. Several dilutions of extract from each strain were tested. Proteins were fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose (Schleicher & Schuell Inc.) or Immobilon (Millipore) for immunoblot analysis. FtsZ was detected with anti-FtsZ serum and compared to purified FtsZ (gifts from the Rothfield laboratory) by using a peroxide-labeled secondary antibody and a LumiGlo chemiluminescent substrate kit (Kierkegaard & Perry). Band densities on X-Omat AR autoradiograph films (Kodak) were quantified with an LKB Ultroscan XL enhancer laser densitometer (Pharmacia LKB Biotechnologies). To check for consistency of measurements, total proteins were compared to estimates made with bicinchoninic acid and Coomassie Plus Protein Assay reagents (Pierce), and antigen detection was compared to immunoblots analyzed with a Storm 820 Phosphorimage Analyzer and the Vistra Systems Western Blot kit (Molecular Dynamics). Units representing the percent maximum signal of the desired protein band were compiled for FtsZ and an internal standard, NusA, detected with anti-NusA serum (59). FtsZ values from several measurements were normalized against their respective NusA values, adjusted for their respective loading volumes and exposure times, and finally summed to calculate an average FtsZ value for each strain.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Suppression correlates with the Ntr phenotype. While exploring the functions of newly identified genes of the rpoN operon (41), we discovered that mutations which disrupt the rpoN gene (41) allowed ftsZ84 strains to grow on LB agar at 42°C. Examination of these suppressor mutations by complementation tests showed that plasmid or lambda clones that replenish only the wild-type rpoN gene could abolish suppression (Table 2). Thus, loss of RNA polymerase sigma factor ⁵⁴ appeared to be the causative defect for suppression. From this information, the mechanism of ftsZ84 suppression could be direct or indirect; attributable either directly to an inability to induce promoters under ⁵⁴ control or indirectly to the physiological state of diminished glutamine metabolism, since the loss of ⁵⁴ activity creates both deficiencies. The relevance of each effect for suppression was tested as described next.

Suppression of ftsZ84 by Ntr mutants is blocked by externally added glutamine. It is known that LB medium contains no free glutamine due to deamidation that occurs during autoclaving (49). We confirmed this absence by physiological HPLC (16a). glnA mutants presumably acquire glutamine from glutamine-containing oligopeptides in the medium. The hypothesis of glutamine limitation for the Ntr mutants was first tested by enriching LB medium separately with each of the 20 standard amino acids. The added presence of glutamine reversed suppression by all four mutations, as exemplified for the glnF mutation in Table 4. These mutants were not identically sensitive, however, since the glnF strain required an amount of glutamine (0.25% [wt/vol]) five times as great as that required by the other mutants to completely restore the high-temperature lethality of ftsZ84. No other amino acid fully blocked suppression, although leucine, glycine, and methionine produced mild effects (Table 4). Since a return to glutamine sufficiency reversed suppression in all of the suppressor mutants, the physiological state created by glutamine limitation seemed to be important for their suppression of ftsZ84 heat sensitivity.

Dependence of ppGpp for suppression of ftsZ84. Because ppGpp is a global regulator induced by nutrient restriction, we tested whether suppression is affected by changes in the level of ppGpp. This association was first explored by artificially increasing ppGpp with plasmid pALS10. The RelA protein produced from this plasmid is unregulated and gratuitously expresses ppGpp in direct relation to its transcription from a tac promoter (53). Induction of RelA from this plasmid allowed high-temperature growth of all of the ftsZ84 strains tested (Table 5). This indicated that all types of suppression may resemble or be identical to effects caused by the induction of ppGpp.

Suppression of ftsZ84 by a high salt concentration also depends on RelA function. The ftsZ84 relA strain was used to explore the necessity of ppGpp for a different but familiar kind of suppression, i.e., that of salt repair (47). The presence of additional NaCl in the LB medium suppressed the filamentation of our parental ftsZ84 strain but not that of its derivative which lacks RelA activity (Table 6). Furthermore, a requirement for RelA activity was also observed for suppression by (NH₄)₂SO₄ and NH₄Cl salts, indicating that the connection between suppression and relA is not limited to sodium salts (52). It is known that high sodium chloride levels are associated with accumulation of ppGpp (24).

Suppression does not involve some other known effectors of ftsZ. To explore the possible involvement of well-known effectors of the FtsZ protein, ftsZ84 strains were constructed that combined either glnL, glnD, or glnF with a mutation of sdiA, sulA, rcsB, rcsF, or rpoS. Perhaps of most relevance is our finding that suppression by glnL is unaffected by a null mutation of rpoS since strain BSP690 (ftsZ84 glnL rpoS) grew at 42°C as well as parental strain BSP666 (ftsZ84 glnL). Importantly, loss of the ^s transcription factor did not alter suppression by a ⁷⁰ (rpoD*) mutation (strain BSP815) or by a high salt concentration (data not shown). The added disruption of either sdiA, sulA, rcsB, or rcsF had no effect on the suppression of ftsZ84 by glnF. A potential involvement of the leucine regulatory protein Lrp was tested because we had seen a slight effect of added leucine on suppression (Table 4). Since an lrp glnF ftsZ84 strain grew well at all temperatures and an lrp ftsZ84 strain was still heat sensitive, Lrp appears to be unimportant for this kind of suppression by ppGpp on LB medium. Surprisingly, however, the ftsZ84 lrp strain acquired a heat-sensitive phenotype on M63 minimal salts plus glucose medium where none exists for the parental ftsZ84 (Lrp⁺) strain.

Suppression causes changes in transcription of the ftsZ operon. Since filamentation caused by the ftsZ84 allele is known to be suppressed by its own overexpression, we tested whether the glnF mutation or artificially high RelA activity could alter patterns of transcription within the ftsQAZ operon. Transcription of two sets of ftsQAZ operon promoters was measured by using reporter plasmids developed and tested elsewhere (66). These low-copy-number plasmids separately place lacZ under the transcriptional control of ftsZ promoters Zp2, Zp3, and Zp4 (herein called the pZ promoters) or ftsQ promoters Qp1 and Qp2 (herein called the pQ promoters). Contrary to what we expected, the activity of the pQ set of promoters was two- to threefold lower in the glnF strain than in the wild-type strain (Fig. 1A). glnD, glnG, and glnL mutants produced similar effects (data not shown). The effects on ftsZ operon transcription caused by artificially elevated levels of ppGpp were measured by using a compatible plasmid that overproduces RelA' as the genetic condition of suppression. As was seen for the Ntr mutations, constitutively high RelA activity also reduced transcription from the pQ promoters about twofold (Fig. 1B). Plasmids that separately contained lacZ fusions to either Qp1 or Qp2 (pCX39 and pCX40, respectively) were tested, and both had decreased -galactosidase activity in the glnF mutant compared to the wild type (data not shown). In contrast, no consistent change in activity of the pZ set of promoters was observed that was common to both conditions of suppression (data not shown). The activities of protein and operon fusions to the Zp2 promoter carried on BBP133 and BBP134, respectively, did not differ between the wild-type and glnF genetic backgrounds (data not shown). This suggested that any change in FtsZ activity caused by suppression is not due to alterations at the level of ftsZ translation and is not observable by using an isolated pZ promoter. Thus, induction of RelA activity correlates with a two- or threefold decrease in the activity of the Qp1 and Qp2 promoters.

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Activity of the Qp1 and Qp2 promoters in ftsZ84 heat-sensitive and heat-resistant strains. Transcription as measured by -galactosidase activity of plasmid pCX32 (66) is plotted as a function of growth at 32°C, beginning from a culture density of 107 cells/ml and growth with selection for plasmids. (A) Suppression by glnF. Symbols: , BSP610 (ftsZ84); , BSP623 (ftsZ84 glnF). (B) Suppression by multicopy relA'. Symbols: , BSP610/pGB2 (ftsZ84/vector); , BSP610/pHM675 (ftsZ84/relA'). We noted that the strains containing plasmid pHM675 grew slowly, and this may reflect the toxic effect of high RelA' concentrations observed by others (53).

Suppression correlates with increased levels of FtsZ84 protein. To estimate possible changes in the amount of FtsZ protein that may accompany suppression, the steady-state level of FtsZ84 protein in these strains was measured by immunoblot analysis of whole-cell extracts. The amount of FtsZ84 protein was measured relative to that of an internal standard, NusA. A weighted average among several different loadings and measurements was calculated so as to minimize variations due to sample handling and antigen detection and finally compared between strains. The amount of FtsZ84 relative to the total amount of protein was greater in both of the suppressed strains than in their respective parental strains (Table 7). The glnF mutant strain had 3.9 times more FtsZ84 than the heat-sensitive parent, and the multicopy relA' strain had 3.4 times more FtsZ84 protein. It appears, then, that the operative effect of suppression by ppGpp is an increase in the steady-state amount of FtsZ84 protein.

E. coli requires ppGpp to grow on LB medium in the absence of a functional Ntr system. While investigating the role of RelA in the suppression of filamentation, we found that our ftsZ84 strains that lack ppGpp and genes for nitrogen regulation (Ntr) do not grow in LB medium. To further explore this effect, we constructed Ntr ftsZ⁺ derivatives of strains having different levels of ppGpp and tested them for growth on various media. The growth of a relA glnF double mutant strain (BSP816) on LB medium, was severely affected compared to that of its relA⁺ parental strain (BSP396), while a relA glnD strain (BSP810) was completely nonviable on LB medium (Table 8). Furthermore, the ppGpp⁰ state rendered both Ntr strains (BSP821 and BSP822) fully nonviable on LB medium. The lethality of this LB effect was relieved genetically by the rpoD* mutations (Table 8) or physiologically by addition of glutamine to the medium (Table 8). Therefore, ppGpp appears to be critical for fulfillment of the glutamine requirement of Ntr cells grown on LB medium. This introduces the possibility of a ppGpp-dependent survival mechanism for making glutamine available from LB medium that functions independently of ⁵⁴-dependent nitrogen regulation. Interestingly, these relA Ntr (ftsZ⁺) strains filamented soon after being transferred from LBQ medium to plain LB medium. However, neither septation nor survival was restored by artificially increasing ftsZ expression (data not shown). This suggests that this effect is not identical to the ppGpp effect on ftsZ under investigation or may involve an earlier stage of septation. In contrast to the glnF and glnD derivatives, a glnA relA strain, which is simply defective for glutamine synthesis (Gln Ntr⁺), grew well on LB medium. Since the glnA strain is not stressed on LB medium, as surmised from its inability to suppress ftsZ84, while Ntr mutants are, it stands to reason that the Ntr system functions in LB medium to provide glutamine in spite of the conventional understanding that the Ntr system is not active in this regard during growth in LB medium (46). From our data, we propose that both the Ntr regulon and an undefined ppGpp-dependent pathway facilitate the supplying of glutamine from a source(s) in LB medium other than de novo synthesis or the transport of free glutamine, which is absent from standard LB medium.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

This study demonstrates that an aspect of septation directed by FtsZ is influenced positively by nutritional stress and that the mechanism of this effect likely depends on the synthesis of ppGpp by RelA. We propose that ppGpp may effect suppression of FtsZ84 by increasing the total activity of the protein through an altered pattern of transcription, and we offer the following arguments in support of this hypothesis.

Suppression of heat sensitivity on LB medium by loss of nitrogen control, i.e., the Ntr phenotypic condition, correlates with an increased steady-state concentration of the FtsZ84 protein (Table 7). This concurs with other findings that the cellular amount of FtsZ is normally rate limiting (38, 67) and that an increase of as little as a 1.5-fold can suppress heat-sensitive filamentation of ftsZ84 (66). It is worth noting here that suppression of filamentation by ppGpp appears to be allele specific since the Ntr mutants did not abate heat sensitivity caused by the ftsZ26 allele, which cannot suppress itself by increased copy number (5) or by the ftsA12 mutation (data not shown). We postulate that the increase in FtsZ protein that accompanies loss of nitrogen control probably involves an elevated basal level of the nutrient stress signal ppGpp for the following reasons. The glnL mutant requires RelA to suppress ftsZ84 (Table 5), an abundance of glutamine annuls suppression by the nitrogen control mutations (Tables 2 and 4), and genetic conditions that overproduce RelA or mimic the ppGpp-induced state (rpoD*) suppress ftsZ84 (Table 5). The requirement of RelA for salt repair shows separately that the relief of ftsZ84 heat sensitivity by ppGpp operates there as well (Table 6).

A reasonable mechanism for RelA-dependent suppression would invoke changes in the transcription of ftsZ due to the known effect of ppGpp on RNA polymerase (8, 26). Indeed, the restoration of high-temperature growth of ftsZ84 strains by SdiA correlates with altered transcription and increased FtsZ protein levels (66). However, the mechanism of SdiA suppression is not similar to that described here since ppGpp decreases expression at both promoters Qp1 and Qp2 (Fig. 1), while SdiA activates the Qp2 promoter. Furthermore, SdiA is not needed for suppression by ppGpp. We note that there is no measured effect of ppGpp on transcription from the downstream pZ promoters or on translation of FtsZ.

We offer two possible scenarios of suppression that might accommodate the decrease in promoter activity upstream of ftsZ84. If the pZ promoters adjacent to ftsZ function independently of the pQ promoters upstream, then suppression might be the consequence of an increase in the ratio of FtsZ to FtsA and/or FtsQ. This scenario alone, however, does not explain why the absolute amount of FtsZ84 protein increases upon suppression. Alternatively, the phenomenon of promoter occlusion may account for the effect of decreased promoter activity, as well as that of increased FtsZ84 protein levels. Promoter occlusion is the inhibitory effect that elongating RNA polymerase molecules have upon transcription initiation at relatively weaker promoters downstream. This was originally suggested with respect to the activities of multiple trp operon promoters (25) and substantiated with respect to the lambda p_L promoter and nearby pGal promoters (1). Promoter occlusion is a well-known phenomenon often called transcription interference at eukaryotic promoters (18, 43). By comparison, then, the predominant transcription arising early in the ftsQAZ operon may inhibit the activity of promoters nearer to ftsZ. Thus, downstream promoters may be activated by reducing promoter activity upstream. A corollary of this model is that translation is more efficient from mRNA originating at the downstream promoter (1). The predicted activation of the reporter-linked promoters on prophages BBP133 and BBP134 would not be detectable in these experiments since they do not contain the upstream promoters in cis. Plasmids carrying reporter fusions that do contain both sets of promoters in cis (68) are not stable under the genetic conditions tested here. Perhaps such constructs adversely affect the cell by introducing extra copies of ftsQ and ftsA. If promoter occlusion explains ftsZ84 suppression by decreased activity of promoters Qp1 and Qp2, then how can it agree with SdiA suppression by increased activity of promoter Qp2? We predict that although activation of Qp2 would further occlude pZ, the increase in total ftsZ translation from an increased amount of the longer transcripts must allow sufficient FtsZ expression. These models do not necessarily exclude one another and, in fact, there are probably many ways to activate FtsZ. For example, a cis-dependent model for regulation of the activities of ftsQAZ operon promoters based on DNA-looping and putative multiple operator sites has also been proposed (15). Early on, we had considered the possibility of a direct effect by ppGpp on the GTPase activity of FtsZ, but this idea was dismissed by the finding that the rpoD* mutation suppresses ftsZ84 in the complete absence of ppGpp (Table 5). Thus, it appears that the essence of suppression by ppGpp may be an increase in the total activity of FtsZ84 protein through transcriptional regulation. These data, however, do not distinguish between a direct and an indirect effect of ppGpp on ftsZ transcription.

Connections between ppGpp and septation have been suggested previously. Studies on the first ppGpp⁰ strains, which filament upon nutritional downshift (68), provided the first direct indication that ppGpp may normally stimulate septation. Gervais et al. mentioned unpublished data showing that multiple copies of relA suppress ftsZ84 (21). ppGpp has been shown to help confer resistance to the cell wall inhibitor mecillinam via several different mutations (62-64). In these examples, it is proposed that FtsZ levels could be especially limiting due to the abnormally wide diameter of these cells, and ppGpp may exert a suppressive effect by increasing the total amount of FtsZ. A similar explanation is given for suppression by ppGpp of heat-sensitive filamentation caused by a mutant of the  subunit of RNA polymerase (61). Perhaps this rpoB mutation affects the same step in transcription as the rpoD* mutants used here, except that it creates an RNA polymerase that is less responsive to ppGpp. Another rpoB mutant appears to resemble rpoD* and increases expression of ftsZ (7). The work reported here concurs with these findings and is the first direct evidence that RelA-mediated synthesis of ppGpp suppresses the ftsZ84 mutant on LB medium. Other indications of a role for ppGpp have come from its effect on growth rate control. It has been known for some time that E. coli cell size decreases under poorer growth conditions. In concordance with this, FtsZ expression has been shown to vary inversely with growth (2, 16, 48) and, by inference, with ppGpp, although a direct connection has not been shown until now. Thus, increased ppGpp correlates with increased FtsZ protein levels in E. coli cultures that enter stationary growth (2, 19, 56) or are otherwise under nutritional stress (54).

Altogether, the data presented here may elucidate some previously unexplained phenomena. One example is the suppression of ftsZ84 by multiple copies of the capsule regulators rcsB and rcsF (20, 21). Since suppression involved high-copy plasmids containing both an intact rcs regulator gene and its ⁵⁴ promoter, it is plausible that the plasmids may titrate a finite and constitutive supply of ⁵⁴ (41), thus leading to glutamine insufficiency, elevated ppGpp, and, hence suppression. The physiological suppression of ftsZ84 by high-osmotic-strength medium (Table 6) probably operates partly through a similar mechanism, since osmotic stress induces ppGpp (24). Interestingly, salt and ppGpp act oppositely on the isomerization reaction that converts the RNA polymerase holoenzyme into the elongating conformation at the rrn promoters of E. coli (22, 37). In this regard, it would be informative to measure the activities of ftsQAZ promoters as a function of medium osmolarity.

During this study, we found evidence that Ntr regulation is important for supplying glutamine in cells grown in LB medium since the absence of Ntr activity requires the nutrient stress factor ppGpp or enrichment of the LB medium with glutamine. This also reveals that ppGpp-dependent gene induction comprises an auxiliary path for supplying glutamine on LB medium. Peptide-bound glutamine can be less susceptible than free glutamine to nonenzymatic deamidation under harsh conditions (e.g., autoclaving) (49), and so the source of glutamine in LB made accessible by the ppGpp pathway may derive from the abundant supply of oligopeptides. Specific mutants which bypass the ppGpp requirement of this peptide transport and degradation pathway might be found by using combinations of Ntr glnA relA strains and selecting for growth on LB medium.

These findings interrelate nitrogen control and cell division by a global signal of nutrient stress, ppGpp, and begin to explain some of the previously observed difficulty and variability encountered in working with Ntr strains, as well as with ftsZ84 strains. While glnF, glnG, glnD, glnL, and rpoD* are the first mutations reported to suppress ftsZ84 on LB medium, the detailed mechanism of the ppGpp effect on FtsZ is not readily apparent and the promoter occlusion model proposed here awaits more direct experimental examination.