Our goal is to understand how global patterns of bacterial gene expression are coordinated by nutrient availability. Complex networks integrate synthesis of macromolecules and regulate expression of the genomic repertoire, thereby providing a model of regulatory interactions between a cell and its environment. We focus on the roles of two regulatory nucleotides found in most bacteria and recently in plants. The analogs we study are related to GTP and GDP but differ by virtue of a pyrophosphate ester on the ribose 3'-hydroxyl group and are abbreviated as (p)ppGpp. Nutrient limitation elevates basal (p)ppGpp levels, and nutrient sufficiency restores low basal levels. This mechanism operates when bacteria are starved of amino acids, phosphate, nitrogen, or energy sources. Regulatory roles are assigned to (p)ppGpp because eliminating (p)ppGpp can abolish regulation while artificially elevating (p)ppGpp (without nutrient limitation) mimics many regulatory effects of starvation. Responses to (p)ppGpp are an integral element in adaptive responses to starvation, ensuring survival by induction of the stationary phase–specific sigma factor (RpoS) of RNA polymerase and by facilitating its effects on transcription. We therefore wish to understand how nutrient limitation leads to changes in (p)ppGpp levels and how responses to (p)ppGpp operate. Global regulatory mechanisms present in bacteria but not in animals can be exploited to develop new microbial antibiotics.

Physical Mapping of RelA Binding to Ribosomes
Glaser, Dahlberg, Murphy, Cashel
Amino acid starvation is well known to activate synthesis of (p)ppGpp on bacterial ribosomes by the RelA protein; activation occurs because codon-specified uncharged tRNA binds to ribosomal acceptor (A) sites instead of the cognate aminoacylated tRNA for protein synthesis. We wish to understand the molecular details of the interaction between RelA, the ribosome, and the requirement for mRNA-specified uncharged tRNA binding.

Last year, Gad Glaser discovered differences in ribosomal RNA (rRNA) susceptibility to dimethylsulfate (DMS) modification associated with RelA binding. This footprint of specific base modifications was measured for less than half of all rRNA sequences. We also localized the ribosomal binding domain of the RelA protein as the C-terminal region of about 160 amino acids without additional contributions from the remainder of the 744–amino acid protein, which includes the domain catalyzing (p)ppGpp synthesis. This year, we have extended the primer sets to complete the mapping over the entire 16S and 23S rRNA molecules. The more extensive mapping did not appreciably extend our initial estimates of modified domains beyond what was initially discovered, namely, near the CCA end of bound tRNA and the decoding region where A site binds to tRNA anticodons paired with mRNA codons. However, the more precise maps do allow better localization of RelA binding sites to the neighborhood of four (B2b, B3, B5, and B6) of the RNA-RNA helical bridges that, from ribosome structural considerations, are thought to stabilize association of 30S and 50S ribosomal subunits. RelA binding to ribosomes usually increases the sensitivity of rRNA bases to DMS rather than to protect them from chemical modification. This is consistent with the notion that RelA binding opens RNA helices and thereby destabilizes RNA-RNA helical bridges. The structural changes seem so major that one would expect ribosomes with bound RelA not to be competent for peptide bond formation. However, optimization of RelA binding conditions reveals that binding occurs when ribosomes are inactive for protein synthesis. First, the presence or absence of mRNA and tRNA does not alter the binding we observe, although mRNA and tRNA are required for activation of (p)ppGpp synthesis. Second, the ammonium ion concentrations for optimal binding are below the level (100 mM) needed for the “Zamir/Elson” conformational transition, which is known to inactivate peptidyl transferase reversibly. Finally, ribosome-dependent (p)ppGpp synthesis occurs under optimal conditions for RelA binding (in the presence of mRNA and uncharged tRNA) despite a riboso-mal conformation with inactive peptidyl transferase. These considerations clearly do not support the long-held view that RelA is associated with functional ribosomes until it is activated by uncharged tRNA binding to A sites, leading to catalysis of a round of (p)ppGpp synthesis which, in turn, leads to dissociation of uncharged tRNA and perhaps also release of the RelA protein. The binding of RelA and the synthesis of (p)ppGpp on a ribosome that is idling for lack of appropriately charged tRNA may instead be mechanistically integrated with failure to form a peptide bond. There is a clear discrepancy between our current observations and what was considered to be true a short time ago.

Ribosomal Affinity Tag to Immobilize Functional Ribosomes on a Solid Support for Stepwise Studies of Protein Synthesis
Canna, Murphy, Cashel
The work just mentioned highlights the need to facilitate functional analysis of the synchronized steps in protein synthesis with respect to binding (or release) of uncharged tRNA as well as RelA protein. The success of existing approaches with immobilized RNA polymerase for addressing a formally analogous problem for RNA, rather than protein, polymerization led us to embark on an attempt to immobilize functional ribosomes as an initial goal. In view of the large number potential of uses of such a construct, it is perhaps surprising that laboratories primarily concerned with protein synthesis have not already attempted the construct.

We have chosen to construct a double-affinity (HisX6 + HA) tag on the 50S ribosomal L1 protein because this protein is among the most surface-exposed ribosomal proteins. In addition, it is distant from the peptidyl transferase catalytic center and distant from RelA binding sites and is reported to have been derivatized on its Cterminus with a GFP tag without compromising ribosomal function as judged by cellular viability. Sequences for the affinity tags were synthesized with (or without) a nine-residue spacer. Scott Canna recombined the sequences into the last codon of rplA by using “recombineering” techniques recently developed in Don Court’s laboratory (NCI). The ensuing viable chromo-somal constructs were sequenced and genetically mapped by transduction to the appropriate chromosomal locus. Initial studies suggest that nickel-agarose affinity purification of affinity-tagged ribosomes can be achieved whether or not the linker arm is present, although yields are low. Interestingly, Western analyses suggest that anti-HA antibody is reactive only with ribosomes bearing the linker arm, whereas the anti-HisX6 antibody binds regardless of the presence of the linker.

Regulation of (p)ppGpp by the SpoT Protein
Gal, Murphy, Cashel; in collaboration with Hogg
Regulation of (p)ppGpp levels by RelA responds only to amino acid availability. In contrast, regulation of (p)ppGpp by limiting sources of energy, phosphate, or nitrogen is mediated by mechanisms modulating SpoT-catalyzed (p)ppGpp synthetic and/or degradation activities. Interestingly, the SpoT protein does not bind to ribosomes despite broad homology with RelA. We have now constructed a viable, two-residue deletion mutant of SpoT that completely inactivates (p)ppGpp degradation activity without affecting (p)ppGpp synthesis activity. The design of this mutant arose from considerations of Tanis Hogg’s crystal structure determination. The mutant strain can now be exploited to assess regulation of synthesis activity unobscured by effects on degradation. Conserved domain searches from Eugene Koonin’s laboratory (NCBI) have revealed potential clues to SpoT regulation. The SpoT protein has a TGS domain (threonyl-tRNA synthetase and GTPase), followed by an ACT domain (shared by aspartokinase, chorismate mutase, and TyrA) that is deduced to be capable of binding low molecular–weight ligands. Jozsef Gal has used appropriate portions of the SpoT protein as bait to screen an E. coli library with a commercial bacterial two-hybrid system. Multiple isolates of two genes have passed screening tests for false positives; one is known to be involved in iron uptake while the function of the other is unknown. We are currently mapping the interacting regions of the candidate proteins and asking whether chromosomal deletions of the genes have a SpoT-dependent (p)ppGpp regulatory phenotype.