Proc. Nat. Acad. Sci. USA Vol. 73, No. 3, pp. 887-890, March 1976 Genetics Significance of constitutive integrase synthesis (lysogeny/phage lambda/prophage excision/haploidization of bacterial duplications) ALLAN CAMPBELL Department of Biological Sciences, Stanford University, Stanford, California 94305 Contributed by Allan Campbell, December 18, 1975
ABSTRACT One conceivable function for constitutive in- Direct evidence for differential control of the int and xis tegrase formation by lambda prophage is to stabilize the in- genes is scanty. Weisberg and Gottesman (5) showed -that serted state by catalyzing reinsertion of prophages that are when transiently derepressed lysogens were allowed to rees- accidentally excised. As this hypothesis implies a dynamic equilibrium between inserted and noninserted DNA, the ex- tablish repression, excisionase function decayed much more istence of such an equilibrium is explored. By examining the rapidly than integrase did. This explains why insertion is fa- frequency with which prophages appear in an initially unoc- vored in recently repressed cells. Whether the differential cupied chromosomal site of a lysogenic bacterium in which stability they observed was purely metabolic or whether the prophage attachment site is duplicated, the off-rate is es- continued transcription contributed to their result has not timated as less than 10-2 per generation for wild-type X, and less than 4 X 10-4 for N mutants of X. From the rate of inte- been determined. grase-catalyzed haploidization of certain partial diploid At that time, the only known manner in which int and xis strains, the rate of spontaneous integrase activity is estimat- could be transcribed was from the major leftward promoter ed as 3 X 10-3 per generation. From these values I conclude PL. Transcription from this promoter is repressed by the that constitutive integrase will not appreciably stabilize the products of genes cI and cro, and extended beyond early inserted state by virtue of its known activity. termination sites by the protein product of gene N (6). The int and xis genes should be transcribed coordinately from The defining characteristic of a temperate virus is the ability PL, and this transcription should cease entirely in a lysogenic to establish its genome as a permanent resident of an infect- cell, in which repression by cI product is well established. ed cell and its descendants. One method of accomplishing Shimada et al. (7, 8) demonstrated the presence within X this end is by inserting the viral genome into the continuity prophage of a weak promoter (pI), not shut off by repressor, of the host chromosome. This is the strategy employed by that lies near the int and xis genes. This promoter was de- bacteriophage lambda infecting Escherichia coli. The act of tected initially by transcriptional readthrough from a X pro- insertion includes breaking and rejoining of host and viral phage into adjacent genes of the host. Viral mutants (int-c) DNA at specific sites on both partners. This special recombi- selected for elevated rates of transcriptional readthrough national event is catalyzed by a viral enzyme, integrase (1). also express int at a high rate in repressed lysogens. The So long as the viral genome remains latent, it is passively int-c mutations appear to lie within the xis gene, and drasti- replicated as part of the host chromosome. As it is already cally reduce the efficiency of excision. The gene order with- inserted, there is no obvious need for further integrase. The in the affected region is only viral protein known to be essential for stabilization of the lysogenic condition is the repressor (product of the cI int (xis, pr) tL N PL gene, ref. 2). The repressor prevents transcription of most other viral genes, by binding to specific sites on the viral where the order of xis and pI is not known (pi may lie within DNA. In order for the latent provirus ever to generate infec- xis) and transcription from the major leftward promoter PL tious particles, repressor must be destroyed, and the inser- is extendable beyond the termination site tL by the action of tion process must be reversed so that the phage DNA is cut the gene N product. out of the bacterial chromosome. One possible explanation for the above result would be Excision of lambda DNA from the chromosome, like in- that PI is potentially a strong, rather than a weak, promoter, sertion, requires integrase. It also requires a second viral that is turned on, perhaps briefly, at some stage in normal product, called excisionase. The genetic determinants of in- infection; and that int-c mutations relieve this promoter of tegrase and excisionase have been mapped, and lie next to its dependence on whatever factors are needed to activate it. each other, close to the insertion site (3). The product of a On this view, the small amount of constitutive integrase third viral gene (hen) enhances the rate of insertion but not manifested in vivo (9), as well as the transcriptional read- of excision (4). through from wild-type prophages, represents the basal level The fact that insertion and excision have different genetic without activation. requirements raises the possibility of differential control of of the promoter the two functions. At least, one plausible reason that the virus might elect to impose different catalytic requirements Is there an equilibrium between non-inserted DNA on these two reactions is so that it can regulate the direction, and prophage DNA? as well as the rate, of the process. We might then expect that An alternative hypothesis is that the integrase in an estab- insertion would be favored in infected cells about to become lished lysogen stabilizes the lysogenic state by catalyzing the repressed, whereas excision would be more efficient in a reinsertion of any prophages that might accidentally be ex- derepressed lysogenic cell. cised during bacterial growth-as the result, for example, of a momentary derepression of transcription from PL. Superfi- Abbreviations: Gal, phenotype for galactose utilization; Bio, inde- cially, this might seem the most economical interpretation, pendence of exogenous biotin. inasmuch as it invokes no undemonstrated properties either 887 Downloaded by guest on September 25, 2021 888 Genetics: Campbell Proc. Nat. Acad. Sci. USA 73 (1976)
of pi or of integrase. I will argue here that this apparently An upper limit for the rate of equilibration between simple explanation is in fact inadequate and does not make insertion sites sense except perhaps in conjunction with additional unde- A large body of experience with strains carrying phages X (7, monstrated assumptions. 8) or P2 (10, 12) inserted at sites other than the preferred If integrase stabilizes the inserted state by virtue of its one argues that, at least at these secondary sites, insertion is known activity-namely, by catalyzing insertion-it follows stable. For X at its preferred site, the relevant data come logically that, in a lysogenic culture, there must be some sort from strains carrying a duplication of the site and adjacent of dynamic equilibrium between inserted and excised DNA. DNA. My results have accumulated in the course of analyz- DNA cannot be reinserted unless it is first excised. We can ing a strain which does not carry an F' element, but what is describe the equilibrium by the equation apparently a tandem duplication of a small segment of chro- mosomal DNA including the X insertion site attX (13). The Int strain was heterozygous for bacterial markers within the Noninserted DNA 4 Prophage DNA I Int + Xi9 duplicated segment, so that one prophage attachment site is genetically linked to a gal+ gene and the other to a gal-. The equation is not intended to represent a complete Many different lysogenic derivatives of this strain were chemical description of the reaction, which probably con- synthesized. Some of these were lysogenic at one of the two sumes ATP as well as DNA (1). Leftward and rightward ar- attachment sites, whereas others carried genetically distin- rows both represent effectively irreversible reactions of guishable phages at the two sites. To determine the location which the macromolecules appearing in the equation are of a prophage within a given strain, haploid gal- segregants among the products and reactants, respectively. This is im- were examined for their prophage content. Given the origi- plied by the different catalytic requirements of the two. We nal gene order are speaking of a flux equilibrium, not a true thermodynam- ic one. gal+ attxi bio gal attX2 bio Most measurements of the state of viral DNA in lysogenic bacteria tell us only that we almost always find phage DNA a haploid gal- derivative can arise by a single internal cross- inserted rather than free, and that spontaneous prophage loss over to the left of gal. Except for rare multiple crossovers, all is rare. This indicates that the rate of the rightward reaction gal- segregants will, therefore, carry whatever prophages (on-rate) greatly exceeds the leftward rate (off-rate); but were located at attx2 and not those located at attxi. does not define the absolute magnitudes of on- and off-rates. There is no independent evidence for the gene order or Inasmuch as several authors (e.g., refs. 8-10) have made the rarity of multiple crossovers other than the fact that al- statements that might be taken to imply such an equilibri- most all haploid gal- derivatives from any one strain are um, it seems worthwhile to ask whether any relevant infor- identical in prophage content and (where the bio genes were mation is available. also marked genetically) at the bio locus (13, 14). But this In considering the off-rate, we may note first that it has fact in itself precludes frequent migration of prophages be- two components, only one of which relates to the present tween attXi and attX2 during cellular growth. Detailed ex- discussion: (a) Excision takes place in those cells in which amination of the data allows us to place an upper limit on spontaneous phage production is commencing. (b) Excision the possible exchange rate. might also take place in cells that survive and continue to In order to assure an independent origin of different gal- multiply. The first class is irrelevant because reinsertion that segregants, each lysogenic strain was analyzed in the fol- takes place in cells that are about to lyse anyhow cannot sta- lowing manner: a colony that was recognizably diploid (ap- bilize the lysogenic state. We need only concern ourselves pearing variegated on galactose indicator agar) was re- with surviving cells. streaked, and several diploid daughter colonies were picked One way to explore the dynamics of the transition be- and streaked again. From each of these later streakings, one tween inserted and noninserted DNA is to infect lysogenic or two gal- colonies were tested for their prophage content. cells with genetically marked phage and see how often the Thus, analyses of each strain included data from several dif- superinfecting genome replaces the prophage. The low rate ferent gal- colonies. of such prophage substitution (10, 11) argues against a very In the particular data summarized in Table 1, row 1, I rapid equilibrium. The force of the argument is diminished scored six different gal- colonies for each isolate (two from by the technical possibility that the "noninserted DNA" each of three diploid daughter colonies). Out of 256 strains formed by an infecting phage genome is not precisely studied, there were 232 cases in which all six colonies were equivalent to that formed by excision. identical. In 22 others, five were alike and one different. In The data that seem most free from limitations of this sort two others, four were alike and two different. In only one of come from studies of bacterial strains in which more than these two cases did the two minority type segregants come one attachment site for the prophage is present in the bacte- from the same diploid daughter colony. rial genome. It is a commonplace observation, for example, If we consider the 256 original diploid colonies that were that a bacterium harboring an Fgal factor that includes the restreaked to pick three diploid subcolonies from each, each attachment site for X can be lysogenized either at the chro- such colony contains 107_108 cells, and therefore its develop- mosomal site or at the F' site. The mere fact that such a ment entails somewhat more than 20 cell generations. In a strain can be kept in the laboratory with the prophage re- total of 256 X 3 X 20 generations, there was only one case in maining in the same site among all cells examined implies which there might have been an internal change in pro- that the prophage is not continually popping in and out of phage arrangement before isolation of daughter colonies. the chromosome. If that were the case, half of the reinser- Bearing in mind that half of all possible reinsertions will tions should take place at the initially free site, and an equi- restore the original configuration, an estimated upper limit librium mixture of cells with prophages at the two positions for the exchange rate between inserted and non-inserted would soon be achieved. prophage would then be 2/15,360, or about 1 X 10-4 per Downloaded by guest on September 25, 2021 Genetics: Campbell Proc. Nat. Acad. Sci. USA 73 (1976) 889
Table 1. Prophage exchange rates calculated from haploidization results Individuals Possible changes Calculated Type of lysogen analyzed in location exchange rate* Lysogens of Nam mutantst 768 1 <4 X 10-4 Lysogens of other am mutants gal+ (XAam 109)/gal-(-)t 20 0 gal+ (-)/galF (XBam2)t 6 0 Total 26 0