Michael Cashel, MD, PhD, Principal Investigator
Rajendran Harinarayanan, PhD, Postdoctoral Fellow
Katarzyna Potrykus, PhD, Postdoctoral Fellow
William Whalen, PhD, Research Associate
Helen Murphy, MS, Microbiologist

Ultimately, we wish to understand how nutrient availability coordinates bacterial genomic expression. Our efforts focus on the roles of (p)ppGpp, which are two regulatory analogues of GTP and GDP, respectively, with ribose 3′ pyrophosphates. Remarkably, nutrient limitation elevates levels of the (p)ppGpp nucleotides whether starvation is for sources of amino acids, phosphate, nitrogen, carbon, or iron. Because eliminating (p)ppGpp can abolish regulation during starvation and lead to difficulties in surviving starvation, we know that (p)ppGpp have signaling roles. Conversely, elevating (p)ppGpp without starvation can mimic aspects of starvation. Our recent work has led to a structural understanding of the catalytic domain interactions for (p)ppGpp synthesis and hydrolysis as well as to an appreciation of new modes of regulation of rRNA transcription initiation. This year, we generalized our finding of transcription regulation by the interplay of a class of RNA polymerase (RNAP) secondary channel--interacting proteins (GreA, GreB, and DksA). In addition, in an attempt to define rigorously the roles of (p)ppGpp in cellular metabolism and physiology, we successfully recovered synthetic lethals among random insertions and multicopy library screens for suppressors of the ensuing phenotypes.

Transcription regulation of ribosomal RNA promoters by (p)ppGpp

Potrykus, Murphy, Vinella, D' Ari, Cashel

Over the past year, we published studies describing an exploration of our early discovery (Brown et al., J Bacteriol 2002;184:4455) that regulation of transcription by (p)ppGpp requires a protein called DksA. An early observation made by others showed that the DksA protein functions as a co-factor for (p)ppGpp-RNA polymerase (RNAP) interactions. The alpha-helical coiled-coil finger of DksA is now believed to penetrate the secondary channel of RNAP, where it amplifies regulatory effects by stabilizing (p)ppGpp binding near the catalytic center. We asked if other secondary channel-binding proteins that share structural features with DksA can alter RNAP regulation through mutually competitive interactions with either DksA or RNAP. Two such known structural homologues are the transcription elongation factors GreA and GreB. Both GreA/B relieve arrested transcription by cleaving nascent RNA chains that have backtracked within RNAP so as to locate beyond the reach of the RNAP catalytic center. We used a reporter for a ribosomal RNA (rRNA) promoter regulated by (p)ppGpp together with cells bearing specific combinations of greA/B- or dksA-deleted alleles or overexpressed proteins and found that cellular regulation was indeed sensitive to balanced ratios of GreA:DksA and, to a lesser extent, of GreB:DksA. With a purified transcription system, we discovered that, surprisingly, GreA provoked a conformational change of initiating RNAP-DNA open complexes that was abolished at a later step in initiation by DksA. Our observations revealed a role in initiation for a transcription factor previously thought to function exclusively for elongation. We conclude that studies with a specific promoter can support the hypothesis of competition between secondary channel--interacting proteins.

Generalizing promoter regulation by (p)ppGpp

Harinarayanan, Vinella, Potrykus, Murphy, Whalen, Cashel

We next tested the generality of the hypothesis that the balance of GreA/B:DksA is important for (p)ppGpp regulation. Long ago, we found that E. coli deleted for its capacity to form (p)ppGpp simultaneously gains nutritional requirements for seven amino acids (ILVTFHS) on dropout plates that are short of a full complement by a single amino acid. The nutritional requirements can be abolished by suppressor mutations in RNAP rpoB, rpoC, or rpoD subunit genes that allow prototrophy, i.e., growth on glucose minimal medium completely lacking amino acids. We now understand that these suppressing RNAP alleles alter RNAP in a manner that mimics the effects of the presence of (p)ppGpp, even though the structural locations of the missense lesions themselves are, unexpectedly, not in the secondary channel and distant from the catalytic center. It is noteworthy that the (p)ppGpp-dependence of amino acid nutritional requirements reflects positive regulation of transcription by (p)ppGpp rather than the negative effects displayed by the rRNA promoter. This year, we found that reversal of the ILVTFHS requirements can be achieved by overproducing GreA and, to a lesser extent, GreB. Overproduction of GreA is dramatically amplified when the dksA gene is deleted. This surprising constellation of observations generalizes our hypothesis that regulation is attributable to a balance between GreA/B and DksA but requires a more extensive set of promoters. A plausible alternative explanation is that these amino acid-biosynthetic transcripts simply contain GreA/B-sensitive pause sites. If so, the absence of DksA together with the presence of excess greA/B might simply potentiate readthrough of arrested or paused transcripts of specific amino acid--biosynthetic genes.

However, evidence from mutants reveals that the interactions leading to prototrophy do not involve the known GreA/B functions in transcription elongation. Factor GreA with mutant residues at the tip of the coiled-coil finger is known to be unable to relieve arrested transcription during elongation. We tested such mutants and found that their ability to restore prototrophy is much stronger than that of native GreA. We conclude that GreA suppressor activity probably does not operate by effects on transcription arrest. Instead, suppressor activity reflects a regulatory activity opposite to the GreA effects on rRNA transcription mentioned above. Future genetic studies will aim at defining the features of specific amino acid--biosynthetic gene transcripts that result in the (p)ppGpp-deficient cell phenotype as well as the GreA domains responsible for suppressing poly-amino acid auxotrophy.

We also found that the balance of GreA/B and DksA influences the degree of growth inhibition accompanying artificial induction of (p)ppGpp without starvation, which we can quantitatively define as growth sensitivity to (p)ppGpp. Among single deletions of greA, greB, or dksA, a modest two-fold elevation of sensitivity was found only for a dksA deletion. Tests of the three possible double deletions revealed a further 50-fold enhancement only with the dksA greB combination. Again, an imbalance among GreA, GreB, and DksA seems to be the key to regulation, in this instance, of growth sensitivity to (p)ppGpp (see below for discussion of the operative mechanism).

Steady-state growth rate control of stable RNA/DNA ratios by (p)ppGpp

Murphy, Cashel

As mentioned above, E. coli cells lacking (p)ppGpp require the amino acids ILVTFHS as defined by the failure to grow on plates when a single amino acid is omitted from the full complement of 20. In liquid media, such strains grow only to low yields even when all amino acids are present, presumably owing to selective utilization and exhaustion of one or another amino acid. We found that high yields require very high levels of serine. We also established the minimum number of amino acids needed for rapid steady-state growth with normal yields. Slow growth requires nine amino acids while fast growth requires three more. Achieving prolonged steady-state growth at varied rates now allows us to assess the effects of a (p)ppGpp deficiency on growth rate control of RNA, DNA, and protein accumulation when the use of each nutrient to support growth is limited not by its concentration but only by the inherent ability of the cell to use nutrients. For wild-type cells containing (p)ppGpp, RNA/DNA ratios are known to approach 10.0 in fast-growing cells (25-minute doubling time) in contrast to ratios of 3.0 for cells doubling every 100 minutes; protein levels parallel those of DNA, which varies about 2.5-fold. We can document that slow-growing (p)ppGpp-deficient cells display RNA, DNA, and protein contents characteristic of rapidly growing cells regardless of growth rate. We made similar observations when (p)ppGpp is present but DksA is deleted. This important finding is the first indication that a redundant mechanism for steady-state growth rate control does not exist when the (p)ppGpp--DksA regulation system is missing or inoperative. When fast-growing wild-type strains exhaust their nutrients and enter a stationary growth phase, the content of RNA, DNA, and protein quickly adjusts to lower values. Such adjustment does not occur in (p)ppGpp-deficient strains. Instead, RNA/DNA ratios remain at the highest unregulated value as if the (p)ppGpp-deficient cells were unaware that they are starving. Our observations that these adjustments in RNA content occur despite normal RNA/protein ratios also reveal that the absence of ppGpp does not alter the efficiency of overall protein synthesis per ribosome.

Essential functions generated by (p)ppGpp deficiency

Harinarayanan, Cashel

We devised a genetic screen for synthetic lethals to identify genes or pathways that perform essential functions specifically in the absence of (p)ppGpp during growth on complex media (Luria broth, LB). We screened about 20 of 20,000 random insertions as positives that were backcrossed, sequenced, and complemented by a wild-type gene copy. We found that insertions in the three genes proC, tktA, and aceE were repeatedly lethal only when (p)ppGpp was absent.

proC.

We found that mutants in other proine-biosynthetic (proA, proB) pathway genes mimic the proC defect as if the effect results from the absence of proline synthesis and is not a peculiarity of the ProC protein itself. Although LB contains enough proline to allow growth of pro mutants when (p)ppGpp is present, additional proline (200 μg/ml) is needed in the absence of (p)ppGpp, especially at high temperatures. Others have observed that high intracellular proline levels can exert a chaperonin effect on misfolded proteins, suggesting that the absence of (p)ppGpp might enhance protein misfolding. We studied the effects of single as well as combined deletions of dksA, greA, and greB. Again, we found profound regulatory effects when high levels of GreA are juxtaposed with low or absent levels of DksA. We transformed the proC synthetic lethal strain with a multicopy random fragment library from E. coli, selecting viable cells with the expectation that we might identify chaperone genes. We found no chaperone genes, even though we detected the expected fragments containing proC itself. Instead, we found that two single gene fragments had multicopy suppressor activity. The fragments encode either adenine PRPP pyrophosphoryl transferase or GGDEF di-cyclic GMP synthetase. A third suppressing fragment encoded three genes: the hok-sok pair of toxin-antitoxin genes and the nhaR sodium-proton pump gene.

tktA.

tktA and tktB encode isozymes catalyzing a key enzymatic step in the pentose phosphate shunt needed to make erythrose-4-phosphate. Deleting tktA and tktB (but not either one alone) in an otherwise wild-type strain makes the strain dependent on products requiring erythrose-4-phosphate for their synthesis, such as aromatic amino acids and pantothenic acid, a vitamin involved in acetyl CoA synthesis. When (p)ppGpp is absent, deletion of tktB has no effect on growth or on a tryptophan requirement, whereas a deletion of tktA (or tktAB) generates both phenotypes. We verified our conclusion that TktB expression is dependent on (p)ppGpp with tktB promoter::lacZ reporter fusions. We do not know why TktA is lethal in the absence of (p)ppGpp but not in a tatAB deletion. One explanation could lie in the absence of catalytic activity and shunt intermediates. However, mutant growth on gluconate suggests otherwise. Alternatively, lethality could be related to a role for TktA as a major determinant of chromosomal supercoil domains, as recently shown by others, particularly given that (p)ppGpp transcription regulation frequently involves promoters whose activity is supercoil-dependent.

aceE.

aceE encodes a key subunit in pyruvate dehydrogenase, which is needed to form acetyl CoA. Supplementation of LB or amino acid--rich glucose minimal medium with acetate reverses the synthetic lethality of this and other pyruvate dehydrogenase subunit mutants. We deduced that a (p)ppGpp-deficient strain is unable to synthesize either acetate or acetyl CoA. The three enzymatic sources of acetyl CoA are pyruvate via ace, acetate via acs, and acetate with pta via acetylphosphate and ackA. We also found that mutations in ace, ackA, and pta were synthetic lethals on glucose amino acid media, with lethality in each case reversed by acetate supplementation. We are testing our deduction that acs expression (or activity) is (p)ppGpp-dependent.

COLLABORATORS

Richard D' Ari, PhD, Institut Jacques Monod, CNRS, Université Paris 7, Paris, France
Daniel Vinella, PhD, Institute Jacques Monod, CNRS, Université Paris 7, Paris, France

For further information, contact mcashel@nih.gov.