Temperate bacteriophages can enter into a long-term and mutually beneficial association with their bacterial hosts. This association is called lysogeny; it is a major mechanism of horizontal gene transfer between bacterial families, and temperate phages constitute a reservoir of genes, such as those encoding virulence factors, that allow their hosts to survive environmental changes. To establish and maintain lysogeny, an infecting phage must carefully control the expression of its own genes so that host viability is unimpaired; at the same time, the phage must synchronize its own replication with that of the host. We report on mechanisms that some phages use to resolve each of these problems.

Mechanism and Control of Transcription Termination
King, Sen, Sloan
The ability of organisms to adjust to environmental stress and to choose between alternative pathways of development requires precise control of gene expression. Transcription, the first step in gene expression, is catalyzed by RNA polymerase (RNAP), a target of many regulators. RNAP is a large, multi-subunit enzyme whose core is structurally and functionally conserved in all kingdoms of life. After the enzyme initiates transcription, it continues to elongate the transcript until it reaches a terminator site. At this point, the enzyme, template, and transcript frequently dissociate from each other such that elongation ceases. Terminators insulate contiguous groups of genes from each other and thus allow independent control of the transcription of each group. In addition, antiterminators can control the efficiency of termination and, hence, the expression of genes downstream of terminators. Termination helps silence virus gene expression in lysogens, and antitermination is used to express these genes during normal virus growth. The molecular mechanisms of both processes remain poorly understood. We are intensively studying an antitermination mechanism found in phage HK022, a virus that parasitizes E. coli and uses the host RNAP to express its genes.

We previously showed that phage HK022 uses an antitermination mechanism that differs from any previously described. Each of the left and right transcription units of the HK022 chromosome contains an antiterminator site, putL and putR, respectively. A nascent put transcript binds to the RNAP molecule that catalyzed its synthesis and remains associated with it as the enzyme moves down the template. Association with put RNA modifies the enzyme so that it resists termination at downstream terminators. No other protein factor is required for the binding to occur, other RNAP molecules in the same cell are unaffected, and there is no apparent terminator specificity. Antitermination promotes the expression of many genes that are required for virus growth. This particular elongation control pathway appears simpler than others found to date, and its simplicity makes it an attractive target for deeper analysis. We are investigating what is unique about the antiterminator RNA and how it changes the properties of RNAP.

Computer simulations, enzymatic probing, comparison of putL to putR, and analysis of put mutants have all provided clues about the details of put RNA structure and their importance for antitermination. Our evidence suggests that the newly synthesized put transcript folds into a structure consisting of two stem-loops separated by an unpaired base. Fig. 26 summarizes the results and conclusions of extensive structure-function studies. We have identified a large number of bases and base pairs that are important for antitermination, potentially allowing us to identify new antiterminator sites in other organisms and to understand the role of different regions of the put structure.

FIGURE 26

Mutational analysis of put

Characterization of RNAP mutants that prevent phage growth by reducing antitermination has provided clues about how the enzyme is modified. Such mutations alter a zinc binding domain (the Zn finger) located near the amino terminus of the b' subunit. The Zn finger contains two pairs of invariant cysteines flanking a moderately well conserved segment of 13 amino acids that is rich in basic residues. We replaced each of the basic residues with alanine and determined the effects of the substitutions on termination, antitermination, and cell viability (Fig. 27). All the mutants were defective in put-mediated antitermination, yet the severity of the defect depended on the mutant and the sequence of the upstream stem-loop of put RNA. Some, but not all, mutants distinguished between put variants that differed in this region.

FIGURE 27

Phenotypes of Zn finger mutants

Such allele specificity suggests that the Zn finger interacts directly and specifically with put RNA in order to anchor the transcript to the translocating enzyme. In fact, the observation that at least one of the mutations (Y75N) prevents association of put RNA with RNAP supports this role of the Zn finger. All the mutants with replacements of the basic residues were viable, and those mutant enzymes subjected to testing transcribed and terminated normally in vitro on a template that lacked a put site. By contrast, replacement of the invariant cysteines with histidine or alanine prevented cell growth unless a second, functional copy of the gene was also present.

Evolution of Insertion Site Specificity
Dorgai, Rutkai
Many temperate bacteriophages lysogenize their hosts by inserting their chromosomes into specific sites (attB or bacterial attachment sites) in the host chromosome, thus ensuring coordinate replication of phage and host chromosomes. Insertion requires integrase, which is a virus-encoded protein that catalyzes recombination between phage and bacterial attachment sites. Bacterial chromosomes contain many attBs, each specific to one or a few phages. Although different integrases share a common catalytic mechanism and are related to each other in structure, proteins encoded by different phages typically recombine only their cognate sets of attachment sites. The existing relationships and distribution of integrases and attachment sites imply that phages occasionally change their insertion specificity over an evolutionary time scale. Alteration of insertion specificity cannot occur in a single step because it requires alterations of both integrase and the multiple integrase binding regions within the attachment sites. In addition, efficient insertion requires sequence identity within the “overlap region,” a short DNA segment that lies between the points of DNA strand exchange in the attachment sites; in fact, the attachment sites of different phages typically have different overlap regions. We are using two closely related but functionally distinct sets of integrases and attachment sites as a tool to understand how changes in insertion specificity might occur.

We showed that the site specificity of phage l integrase can be changed to that of phage HK022 by the replacement of five l residues with their HK022 counterparts. By themselves, two of the replacements relax specificity: the double mutant efficiently recombines both l and HK022 attachment sites. The other three replacements restrict specificity: the triple mutant recombines neither l nor HK022 attachment sites but does recombine mutant l sites that can also be recombined by HK022 integrase. To understand how the mutations alter specificity, we asked if they affected recombination of sites other than those of wild-type l and HK022. When phage l infect a cell that lacks attB, lysogeny is infrequent, but, when lysogeny does occur, integrase usually catalyzes insertion of the phage chromosome into one of 10 to 20 secondary attachment sites. These sites are equivalent to attB mutants. Phage containing either or both of the relaxing int mutations lysogenized a host lacking attB with about the same frequency as wild type. One of the relaxed mutants differed from and the other was similar to wild type in its preferences for particular secondary sites. We conclude that specificity relaxation by these mutations is sequence-dependent and narrow in scope. The three restricting int mutations, which we tested as a triple mutant, strongly reduced the frequency of insertion into all secondary sites; therefore, specificity restriction is broad in scope. Among the few insertion sites that we found, one differed from any used by wild-type phage or the relaxed mutants. Phage that contained all five mutations inserted into secondary sites about twice as frequently as wild type and had new preferences for individual sites. The preferences of the quintuple mutant were not a simple sum of the component mutations, suggesting that interaction among the protein domains affected by the replacements is an important element of specificity alteration.

We used our findings to refine a model for the evolution of phage insertion specificity. The initial step is phage insertion into a secondary attachment site whose overlap region differs from that of the wild-type attB. Abnormal excision of this phage from the bacterial chromosome replaces phage DNA with adjacent bacterial DNA and incorporates the new overlap region sequence into the phage attachment site. We propose that this phage variant is adapted to insert efficiently into the secondary host site by a series of relaxing and restricting int mutations. According to this model, new attachment sites evolve from sequences that already have some activity as integrase substrates, an attractive feature not found in other models. In addition, the new phage attachment site will be located at one end of the bacterial DNA replacement, where it is usually found in existing phages.