INTRODUCTION

Top Introduction References

It is clear from the endless stream of sequencing information now flooding the databases of the world that apparently disparate bacterial species have many features in common, especially within gram-positive or gram-negative lineages. The Functional Genomics of Gram-Positive Microorganisms meeting held in San Diego, Calif., 24 to 28 June 2001, was organized to bring together investigators working with related, in this case gram-positive, bacterial species under the premise that what has been learned about gene regulation, expression, and function from decades of molecular genetic and biochemical studies of the more common species would rapidly advance the understanding of less-well-studied members of the lineage. From all the comments received, this comingling of investigators and species under a common rubric was a successful strategy.

The meeting saw a surfeit of seemingly similar sounding suffixes made by splicing and ligating suffixes to truncated nouns. Aside from the emerging usage of transcriptome to describe the mRNA population of a cell and proteome as the protein population, the attendees were treated to septasome, proteins making up the septation apparatus (S. A. Robson, University of Connecticut Health Center); secretome, components of the machinery for protein secretion (J. Kok, University of Groningen); cellulosome, a multienzyme complex for cellulose degradation (T. Dror, Technion); and fluxome, the array of metabolic fluxes for all reactions in an organism, (U. Sauer, ETH Institute of Biotechnology). One can subdivide these categories by adding adjectives, e.g., stress proteome (A. Hartke, University of Caen), or prefixes, phosphoproteome (S. Seror, University of Paris Sud). It occurred to these reviewers that a hierarchical structure of "omes" could be developed and portions of each placed into either a factome or an artifactome, both being metastable interchangeable states.

Because the use of microarrays is a new technology leading to the rapid recognition of the role of regulatory proteins and the definition of transcriptomes, this meeting review may be devoted disproportionately to these new results. Many of the interesting and significant presentations could not be described in this review. Some of the associated papers appear in this issue (1-9).

	GENOMES

Without genome sequencing you can't have functional genomics, and while there were more than enough slides of lists of open reading frame categories to satisfy the truly addicted, the presenters went out of their way to point out the unique genes of these genomes that make each organism fit in its ecological niche. Joe Ferretti (University of Oklahoma) described the Streptococcus pyogenes genome and emphasized that there were more than 40 virulence-associated genes, CAMP factor, hyaluronidase, etc., that make this organism such a formidable pathogen. It was also noted that the genome contained four resident bacteriophage genomes, each carrying superantigen-like genes, illustrating the potential for horizontal gene transfer. In a similar vein, Frank Kunst (Institut Pasteur) remarked that the unpublished Listeria monocytogenes genome is proving to be very much like that of Bacillus subtilis without a spore but with a large number of surface proteins that may help it to interact with a wide range of eukaryotic cell types. In fact, the crystal structure of the internalins A and B revealed a concave interaction surface common to these abundant surface molecules which was proposed to be responsible for specific recognition of host binding partners during infection (D. W. Heinz, GBF, Braunschweig). One surprise in the Enterococcus faecalis genome (L. C. Banerjei, The Institute for Genomic Research [TIGR]) was the finding of a large number of PTS (phosphoenolpyruvate:sugar phosphotransferase) systems. These should allow it to transport and metabolize a wide range of carbohydrates, although it seems to be devoid of many amino acid biosynthetic pathways. Enterococci are specialists. There must be a wide variety of carbohydrates in the gut and a surfeit of amino acids. Similarly, the Lactobacillus plantarum genome was found to be packed with PTS systems, and this organism was shown to metabolize at least 24 different carbohydrates (R. van Kranenburg, Wageningen Center). This organism also counts the gastrointestinal tract as one of its ecological niches.

The genome of Lactococcus lactis, an important bacterium in cheese manufacture, was sequenced, and it revealed a large number of insertion elements. These occupy about 2% of the genome and include six prophages (S. D. Ehrlich, INRA). Bacillus anthracis, the mammalian pathogen, appears to be essentially the same as Bacillus thuringiensis, the insect pathogen, and Bacillus cereus in its basic chromosomal genes (T. Reed, TIGR). Pathogenicity is plasmid determined in the first two strains.

While not unexpected, this entire group of bacteria and B. subtilis were clearly derived from a common ancestral bacterium. Most of the genes for housekeeping functions are related, whereas the pathogenicity factors are unique. If Enterococcus species, for example, lack the pathways to make many amino acids, was the progenitor strain devoid of these pathways? If so, did species like B. subtilis acquire them from elsewhere or are the ecological niches that enterococci occupy so amino acid rich that these pathways became redundant? Regardless, the availability of this sequence information is allowing rapid assessment of an organism's metabolic capabilities and identification of potential targets for the development of new anti-infectives (D. McDevitt, GlaxoSmithKline).

	MICROARRAYS: THE NEW -GALACTOSIDASE

It is clear that microarrays are at the vanguard of the analysis of the transcriptional consequences of regulatory protein action and have supplanted lacZ fusions for this purpose. E. Ferrari (1) kicked off reports of studies of this type by describing their experience in the analysis of the ScoC regulator, a MarR class regulator, in B. subtilis. ScoC was found to regulate directly or indirectly more than 500 genes, either by repression or activation. More importantly, he presented both the good, bad, and ugly points of microarray analysis, which set the tone for subsequent presentations. Sigma B is an alternative sigma factor whose activity is regulated by stress-sensing mechanisms in B. subtilis and, from genomic analysis, in Staphylococcus species but not Streptococcus pneumoniae (A. Chastanet, Institut Pasteur; T. Msadek, Institut Pasteur). C. Price (University of California) showed that Sigma B controls transcription of at least 127 genes, including those for transporters, drug efflux pumps, and products for carbon metabolism and other basic metabolic processes. Oxidative stress induced by the addition of hydrogen peroxide led to the induction of 70 genes and repression of 70 genes, while superoxide stress (paraquat addition) resulted in induction of 140 genes and repression of 160 genes (C. Scharf, Greisfwald University). Competence is an important metabolic state for gram-positive species in which the proteins for genetic transformation are induced under control of the ComK regulator. D. Dubnau (Public Health Research Institute) showed that B. subtilis ComK affects transcription of an additional 180 genes beyond what was previously known. These are organized in 30 operons. B. subtilis's growth at alkaline pH induces expression of at least 50 genes, including those for transporters and an Na⁺/H⁺ antiporter (T. Wiegert, University of Bayreuth). Some of these genes are part of the Sigma W regulon. Amino acid supplementation of minimal media down regulated about 100 genes for amino acid biosynthetic pathways, sporulation, and competence and of unknown function (U. Mäder, Greifswald University). At this point there are no comparisons between data to determine if different stresses induce many of the same genes or if each is unique, i.e., is there a pattern to all this.

	GENE REGULATION BY TWO-COMPONENT SYSTEMS

Two-component signal transduction systems are believed to function as a link between environmental or metabolic signals leading to gene activation and repression. Until this meeting, it was never fully appreciated how global their effects on gene transcription could be. Using an inducible promoter to drive response regulator transcription, K. Kobayashi et al. (6) analyzed 24 two-component systems by microarray analyses and showed that the number of genes regulated varied from as few as 4 for the YcbM/YcbL system to 128 for the DegS/DegU system. In most cases it is not possible to ascertain a central theme for the genes that are regulated by any two-component system. In those two-component systems regulating substantial numbers of genes there is a disproportionate predominance of unknown genes (e.g., 75% of unknown genes for the YdbG/YdbF system). This would be consistent with the notion that two-component systems regulate adaptability to various environmental conditions and that a large share of genes of unknown function are dedicated to this role. In another study of the DegS/DegU system using a degU deletion, about 80 genes were found to be regulated (E. Guédon, Institut Pasteur). Aside from the mathematical problems in interpreting microarrays, differences in the numbers of genes identified result from a variety of methodology differences.

The ciaR/ciaH two-component system of S. pneumoniae was analyzed by FISH (fluorescent in situ hybridization), fishing out DNA fragments that bound to CiaR, and CHIPS microarrays (T. Mascher, University of Kaiserslautern). At least 72 genes were differentially expressed in a ciaR mutant, most of which were related to biosynthesis of cell wall polymers and competence genes.

Microarray analyses of the complex agr and sar systems of S. aureus revealed that both effectors influence the transcription of many more genes than previously thought and should be classified as global transcriptional regulators (4). Since these systems are known to be virulence regulators it may be time to view virulence as one feature of a cellular response.

	CONCLUSIONS AND CAVEATS FROM MICROARRAY ANALYSES

The transcriptome analyses with microarrays revealed that a large number of genes are transcriptionally regulated up or down by any regulatory protein tested and that many of the products of these genes don't seem to be related to the putative functional role of the regulator. The largest documented effect was the study of ScoC regulation, which showed that 560 genes were differentially expressed between wild-type and scoC strains (1). These include genes for amino acid biosynthesis, nucleotide metabolism, transport, motility, sporulation, and virtually everything else. A scoC mutant is a very different organism from it's parent. Many, if not most, of these differences are attributable to ScoC regulating other regulators regulating other regulators. This cascade effect also may explain the large number of genes controlled by the Agr and Sar regulators of S. aureus (4). Thus comparing two strains, one of which is mutant, will very likely identify a substantial number of indirectly regulated genes. One way around this problem is to design the experimental conditions such that the regulatory gene of interest can be rapidly switched on or off and mRNA levels can be assayed shortly after the switch. Heat shock-induced genes can be turned on by this method, and microarray studies could be carried out in a matter of minutes after induction. A large number of transcripts were increased or decreased by heat shock, and several regulatory mechanisms accounted for this effect. Furthermore, substantial differences were found between time points separated by only a few minutes (5).

There seems to be no doubt that previous investigations into the effects of regulatory genes have seriously underestimated the number of genes whose transcription is modified by mutation of the regulator. It is clear that regulatory circuits are interconnected and show considerable overlap. It is not that this wasn't already appreciated; the microarrays just drove the point home. Several questions arise from this realization. Do bacteria have a rationale for the subset of genes turned on, even if it's several hundred genes, by a given "global" regulator? The answer is probably yes, in a loose sort of way, but the large number of regulated genes of unknown function make this conclusion tentative. Changes in complex phenotypes such as sporulation or virulence in any given mutant are likely to be multifactorial. Experience from many years of research on the regulation of sporulation has shown that this cellular event is affected by regulatory proteins with global effects. Sporulation is regulated up or down only because it is a cellular process and subject to the balance of many individual regulators. For example, sporulation is greatly enhanced in a scoC strain, but what combination of the 560 genes ScoC regulates is responsible for this phenotype remains obscure. Similarly, the dependence of virulence in S. aureus on the Agr and Sar regulators (among others) is the result of global regulation, but many of the genes regulated have little, if anything, to do with virulence.

With any technique it takes time to sort out sources of artifacts and learn the right controls. E. Ferrari listed many of the problems encountered by his group (1). These include problems of reproducibility because of temperature or medium variations that seem minor to us but not to the bacteria that are extremely sensitive biosensors. Probably the largest unknown is whether mRNA levels directly correlate to protein levels. There are many reasons why this may not be the case. J. Helmann (Cornell University) pointed out that some genes may be influenced by proximity effects due to inefficient termination and may appear to be induced or repressed but not actually under control of the regulator being tested. Statistical analysis of the mountains of data that emerge from these studies leads to questions of what level of induction or expression is meaningful, i.e., 2×, 3×, or 4× over the noninduced state. Finally, cultures are mixed populations of cells in various stages of the cell cycle, some growing and some entering stationary phase in most medium conditions. Microarrays can't distinguish what portion of the population is producing the observed mRNA signal. Thus 2× changes overall may be 4× or 8× or more in a minor part of the population. The only certainty is that the scientific literature is going to fill up with this information and, verbum sapienti, the reader should keep an open mind about any conclusion reached.

	NEW VISTAS IN REGULATION

The regulation of biosynthetic pathways is, in many cases, greatly different in gram-positive and gram-negative bacteria. Charles Yanofsky (Stanford University) showed that the regulation of the tryptophan biosynthetic pathway is a case in point. Transcription of the trp genes in B. subtilis is regulated through an attenuation mechanism mediated by the TRAP protein, which binds trp mRNA. Crystal structure studies revealed that TRAP is assembled into a ring of 11 subunits that binds mRNA around its outside perimeter through GAG/UAG repeats (P. Gollnick, State University of New York, Buffalo). The phosphodiester backbone was located on the outside of the ring, with only the bases making contact with protein. As if tryptophan and TRAP weren't enough regulation for the trp operon, A. Valbuzzi (Stanford University) discovered a protein, YczA, induced by uncharged tRNA^Trp, that binds directly to TRAP and prevents TRAP from binding leader RNA or releases already bound TRAP from the leader RNA. The seeming complexity with which bacterial cells regulate biosynthetic pathways has reached a new level. In a similar vein, PyrR of B. subtilis regulates the pyrimidine operon by attenuation in response to either UTP or UMP. PyrR is also a uracil phosphoribosyl transferase, but mutations were found that indicate neither activity is dependent on the other (R. L. Switzer, University of Illinois).

Gram-positive and gram-negative bacteria also differ significantly in the regulatory strategies they use to control metabolic pathway gene expression. Regulation of carbon and nitrogen metabolism, for example, are well-studied processes in both B. subtilis and Escherichia coli, and the different schemes used are representative of the great diversity these organisms have evolved in coping with the same framework of bacterial physiology. This conference saw a major advancement in our understanding of the mechanisms regulating key steps in the B. subtilis response to nutrient availability. Susan Fisher (Boston University) has elegantly unraveled the nature of the nitrogen signal that allows B. subtilis cells to adapt to growth in nitrogen-limited conditions. The key regulatory protein of this system is glutamine synthetase; it was found to transduce the nitrogen signal by directly regulating the DNA binding activity of the TnrA transcriptional factor, a positive activator of genes for nitrogen scavenging proteins and a repressor of glutamate synthase. Interaction of feedback-inhibited glutamine synthetase (nitrogen sufficient conditions) with TnrA was shown to inhibit TnrA DNA binding in vitro.

The effector molecule regulating the activity of another key protein involved in sensing nutrient availability has been revealed by the work on CodY. This is a highly conserved protein in low G+C gram-positive bacteria that controls expression of several genes involved in nitrogen metabolism, competence, and acetate metabolism by repressing transcription in conditions of nutrient-rich growth. The signal that indicates nutrient excess or limitation and thereby regulates CodY activity was shown to be GTP (A. L. Sonenshein, Tufts University). Thus, the transient decrease in the GDP and GTP pools in B. subtilis cells undergoing the transition from rich to poor growth conditions (exponential versus stationary growth) provides the signal sensed by CodY. As a GTP-binding protein, CodY responds to growth in excess nutrients by repressing many stationary-phase genes. Repression is relieved when the level of GTP decreases under slow-growth conditions. Thus, at least part of the role of guanine nucleotide fluctuation in regulating gene expression in transition from active growth to stationary phase, described decades ago by the late Ernst Freese, has been solved.

Another distinctive key regulator in gram-positive metabolism is the bifunctional HPr kinase/phosphatase (HPrK/P). HPrK/P provides tight control of expression of many genes and operons submitted to carbon catabolite control by modulating the level of P-Ser-HPr in complex with the CcpA transcriptional regulator protein. HPrK/P phosphorylates the HPr protein at Ser46 in conditions of high catabolite levels, while it dephosphorylates P-Ser46-HPr in the absence of catabolites. Resolution of the three-dimensional crystal structure of HPrK/P from Lactobacillus casei revealed a hexameric organization, with each subunit containing an ATP-binding domain (Walker motif A) and a putative HPr-binding domain. The structure, the first of a bacterial Ser/Thr protein kinase, establishes that HPrK/P is unrelated to eukaryotic Ser/Thr protein kinases; rather it belongs to the P-loop-containing family of nucleotide binding proteins (S. Poncet, INRA). The interpretations of extensive biochemical and genetic analyses of HPrK/P mutant forms (S. Poncet, INRA; A. Galinier, LCB-CNRS) are now greatly facilitated and will soon provide a clear understanding of the molecular mechanisms underlying this fundamental cellular function that allows gram-positive bacteria to adapt to a wide range of growth conditions.

	CELL DIVISION

The last decade has seen a significant expansion of cytological studies with prokaryotes. These have stimulated our thinking about the organization of the bacterial cell. The use of immunofluorescence microscopy and green fluorescent protein(s) has made it possible to greatly increase the sensitivity with which protein localization studies could be carried out and has permitted the visualization of position and movement of specific proteins in living cells. In the session dedicated to "Cell Division," recent findings were presented that provide an understanding of bacterial cell morphogenesis and its coordination with cell cycle progression and differentiation.

Rut Carballido-López (Oxford University) showed the first evidence that bacteria possess a cytoskeleton-like structure that is the determinant of cell shape. The Mbl protein was found to form a helical filamentous structure lying close to the cell surface and running the length of the cell. The filamentous nature was suggestive of a polymerization event and, along with its (limited) sequence similarity to actin, raised the possibility that the Mbl protein may be a functional homolog of the eukaryotic actin proteins. It is interesting that homologs of the mrl gene can be found only in a wide range of eubacteria, all related by having a nonspherical cell shape; i.e., the gene is absent from species with spherical shapes.

How a cell coordinates chromosome replication with cell division remains a major mystery. Since formation of the so-called Z ring by the FtsZ protein is the earliest event in cell division, it is likely that coordination of chromosome replication with cell division occurs at this stage. By means of synchronous cell populations, obtained by using thymine-requiring outgrowing B. subtilis spores, and immunofluorescence microscopy, a correlation between the ability to block mid-Z-ring assembly at a midcell nucleation site and entry into the elongation phase of replication was found (L. Harry, University of Sydney). This suggests that a replication-mediated checkpoint control is set up at chromosome replication initiation or soon after entry into the elongation phase. This must block Z-ring formation at midcell and is subsequently relieved late in the round of replication. This replication checkpoint would be the third factor, in addition to the Min system and nucleoid occlusion, involved in midcell Z-ring assembly, and it would ensure the coordination between replication and division.

The quest for the holy grail of cell division, i.e., how cell division is controlled, continued with F. J. Gueiros-Filho (Harvard University) describing the identification of yet another protein involved in the process, the product of the yshA gene, whose overexpression counteracts the deleterious effect of MinD overexpression. YshA is likely to be a novel component of the cytokinetic ring that promotes Z-ring formation; however, its mechanism of action remains to be elucidated.

A genomic approach applied to these same issues was presented by workers in other laboratories. Valerie Vagner (Oxford University) showed the results of a DNA microarray analysis aimed at identifying proteins involved in asymmetric division and chromosome segregation in the early stages of B. subtilis sporulation. Starting from the assumption that the Sigma H transcription factor, together with Spo0A, is involved in the expression of the proteins constituting the segregation machinery, DNA arrays of wild-type and sigH strains were analyzed and compared. The results indicated that 20% of the genes of the genome were affected by the absence of sigH. These involved functions such as initiation of DNA replication (induced), or nucleic acid metabolism, DNA repair, respiration, and carbohydrate and lipid metabolism, etc. (repressed). For the genes that seemed to affect sporulation and/or asymmetric division and chromosome segregation, further studies will clarify whether their effect is direct or indirect. A different approach was taken by the group of Naotake Ogasawara (Nara Institute) as they sought to identify proteins associated with the SMC (structural maintenance of chromosome) protein in B. subtilis. The SMC family of proteins is conserved in eukaryotes, archaea, and bacteria, and it is known to play a key role in chromosome organization and segregation. SMC protein is generally active as a subunit of multiprotein complexes. By means of sedimentation studies and by exploiting the European and Japanese functional analysis projects, genes encoding potential members of the B. subtilis SMC complex have been identified as four genes of unknown function belonging to the group of essential genes. One wonders how many important functions would have been found by now if the list of essential genes had been made public by these consortia.

In conclusion, the bacterial cell continues to resist our efforts to understand how it accomplishes such precise coordination of all of its functions in cell division.

	SECRETION

Gram-positive bacilli (e.g., B. subtilis, B. licheniformis, and B. amyloliquefaciens) are particularly renowned for their high capacity to secrete proteins into the environment. This property has been exploited commercially for several years but not without major limitations when it comes to production of proteins from gram-negative organisms or eukaryotes. The sequencing of the B. subtilis genome made possible a series of studies aimed at a greater understanding of the mechanism of protein secretion in this organism. Presentations from several laboratories provided a view of current knowledge and major recent achievements on what is now called "the secretome."

Predictions of the composition of the secretome, which includes components of the pathway for protein transport and secreted proteins, were presented (H. Tjalsma, University of Groningen). These studies suggested the presence of four distinct pathways for protein export from the cytoplasm and predicted that approximately 300 proteins were exported. In addition to the major "Sec" system, three additional pathways, the twin-arginine translocation or "Tat" pathway, the type IV prepilin-like export pathway, and the ATP-binding cassette (ABC) transporters, act most likely as "special-purpose" pathways. These are probably involved in transport of a limited number of proteins.

A more detailed view of the Tat system was provided by a collaboration that carried out a genetic and proteomic analysis of this export pathway (J. D. H. Jongbloed, University of Groningen). Tat pathways are present in many but not all eubacteria, in some archaea, in a few plant mitochondria, and in chloroplasts. Tat pathways are distinct from the Sec pathways by the specific recognition of signal peptides containing a double arginine (R-R) motif and by the ability to transport rapidly or tightly folded proteins that the Sec system would not transport.

A proteomic approach was used to identify extracellular proteins (H. Antelmann, Greifswald University). Interestingly, an investigation of substrate specificity by the four type I signal peptidases (SPases), SipS, -T, -V, and -U, indicated that, with the exception of the transmembrane protein YfnI, which was found to be less processed in mutants lacking both SipT and SipV, no significant difference in the protein secretion patterns was observed. This confirmed previous observations suggesting large overlapping substrate specificities among B. subtilis SPases. This is in contrast with the specificity observed for the unique type II SPase, LspA, which is required for the processing of prelipoproteins and whose absence results in a clear change of the protein secretion pattern.

A mechanistic view of protein trafficking through the membrane was given by Kunio Yamane (University of Tsukuba). Studies on the interaction between the proteins of the Sec system (that translocate the exported proteins across the cytoplasmic membrane) and the signal recognition particle (SRP) system (that binds to signal peptides emerging from the ribosome and is targeted to the membrane by the FtsY GTPase) have shown the ability of SecA to form a complex with the secretion-specific targeting protein Ffh both in vivo and in vitro. Furthermore, the complex was shown to bind precursor proteins, with the binding being dependent upon an active Ffh protein.

Perhaps a more direct contribution to the improvement of heterologous protein production and secretion by bacilli came from the presentation by Vesa Pekka Kontinen (Public Health Institute, Finland), who reported an improvement in the processing of precursor proteins and increased secretion levels in strains carrying knockout mutations of the clpP and/or clpX genes, encoding the ClpP protease and the ClpX substrate-binding component of the Clp complex, respectively. Of course, the growth impairment caused by either clpX or clpP deletions may constitute a problem in the exploitation of this observation! On the subject of potential tools for improvement of heterologous protein secretion in B. subtilis, it is clear that the three Bdb thiol-disulfide oxidoreductase proteins identified in B. subtilis have a major role in ensuring the correct maturation of heterologous proteins requiring disulfide bond formation (R. Dorenbos, University of Groningen). It doesn't appear that human antibodies will be secreted from bacteria any time in the near future.

	REGULATION OF SPORULATION

The regulation of sporulation in bacilli is an exceedingly complex developmental process responding to a plethora of regulating molecules, signal transduction systems, and sigma factors. It is well established that the initiation of sporulation, as defined by the transcription of sporulation-specific sigma factors, is dependent upon the Spo0A transcription factor, which is activated by phosphorylation through the phosphorelay signal transduction system. The major kinase activating (phosphorylating) the phosphorelay is KinA. While the signals activating KinA remain obscure, KinA was shown to bind ATP in one of the PAS domains making up its signal recognition domain (K. Stephenson, The Scripps Research Institute). Negative regulation of the phosphorelay occurs through a series of Rap phosphatases that are regulated by peptides generated by proteolysis of phr gene products. The Rap phosphatases were shown to contain tetratricopeptide elements which were proposed to bind the inhibitory peptide. A peptide-insensitive mutant of RapA was discovered to have the mutant residue in the proposed peptide binding site of this element modeled from known structures of tetratricopeptide-peptide complexes (M. Perego, The Scripps Research Institute).

Spo0A phosphorylation was proposed to result in a small conformational change in the N-terminal regulatory domain leading to dimerization of this domain as a prelude to gene activation (A. J. Wilkinson, University of York; I. Barák, Slovak Academy of Sciences). The crystal structure of the Spo0A C-terminal domain bound to its DNA target also revealed dimerization on the DNA in a head-to-tail configuration, which explains the oligomerization of Spo0A at its promoters (K. I. Varughese, The Scripps Research Institute). Crystal and nuclear magnetic resonance (NMR) structures continue to play a driving role in research to understand how the SpoIIAA anti-anti-sigma factor is regulated by phosphorylation (J. A. Brannigan, University of York; M. D. Yudkin, Oxford University).

One of the more mysterious regulators of sporulation is the E2 subunit of the pyruvate dehydrogenase complex. Its regulatory functions have been at least partially revealed by the discovery of an association of E2 with the delta subunit of RNA polymerase in genetic studies. Despite the fact that the role of delta itself is a long-term mystery, the data suggest that E2 and delta interact at the level of transcription of some genes required for sporulation (A. I. Aronson, Purdue University).

Unraveling the mechanism behind differential gene expression in the mother cell and forespore compartments has been a fascinating journey in cell biology spearheaded by the Losick laboratory. Several pieces of this puzzle fell into place at the meeting. One of the newest twists in this process was the realization that the position of a gene on the chromosome, i.e., origin proximal or distal, is very important for the correct timing and temporal order of expression of the many proteins involved (J. Dworkin, Harvard University; A. Hofmeister, University of California). What is spectacular about sporulation is the complexity of this seemingly simple system of differentiation and how the bacterial cell takes advantage of even the position of a gene on the chromosome to coordinate the events in this process. More of these position effects may now come to light. P. Piggot (Temple University) described a clever new system to study spatial compartmentalization of gene expression based on the sacB/sacY system which should make it easier to determine which genes are expressed where.

	PROTEIN STRUCTURE

Certainly some of the most satisfying presentations, from a long-term perspective, were those elucidating the structures of several regulatory proteins that have been the subject of numerous genetic and biochemical investigations. In addition to the structures of proteins in complex with their targets already mentioned, such as Spo0A and TRAP, the NMR structure of the AbrB global regulator revealed a novel DNA-binding motif named a "looped-hinged helix fold" (M. A. Strauch, University of Maryland). The structure of the phosphorylation-regulated transcriptional antiterminator LicT will certainly lead to an understanding at the atomic level of the mechanism by which this regulator functions (N. DeClerk, INRA, CNRS).

Richard Brennan (Oregon Health Sciences University) showed how different regulators of multidrug transporter genes, leading to resistance, recognize a wide variety of different drugs by simply binding in a hydrophobic packet. Structural studies are playing a key role in revealing how modular peptide synthetases function to generate structurally defined classes of natural oligopeptides (M. A. Marahiel, University of Marburg).

As more structures are determined and we learn of the mechanisms behind physiological phenomena, they do not seem to get any simpler. The elegant complexities of a bacterial cell are beginning to unfold, and it is evident that its regulation is more complex than ever imagined. The chemical engineers' dream of reducing bacterial metabolism and regulation to an equation will find it to be a complex equation whose understanding will take longer to solve than Fermat's Last Theorum.