ANRV329-GE41-01 ARI 2 November 2007 12:22 by Annual Reviews on 12/13/07. For personal use only. Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org ANRV329-GE41-01 ARI 2 November 2007 12:22

Phage Integration and Chromosome Structure. A Personal History

Allan Campbell

Department of Biological Sciences, , Stanford, California 94305-5020; email: [email protected]

Annu. Rev. Genet. 2007. 41:1–11 Key Words First published online as a Review in Advance on lysogeny, , conditional lethals, deletion mapping April 27, 2007 Abstract by Annual Reviews on 12/13/07. For personal use only. The Annual Review of Genetics is online at http://genet.annualreviews.org In 1962, I proposed a model for integration of λ into the This article’s doi: bacterial chromosome. The model postulated two steps (i ) circular- 10.1146/annurev.genet.41.110306.130240 ization of the linear DNA molecule that had been injected into the Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org Copyright c 2007 by Annual Reviews. cell from the phage particle; (ii ) reciprocal recombination between All rights reserved phage and bacterial DNA at specific sites on both partners. This 0066-4197/07/1201-0001$20.00 resulted in a cyclic permutation of gene order going from phage to prophage. This contrasted with integration models current at the time, which postulated that the prophage was not inserted into the continuity of the chromosome but rather laterally attached or synapsed with it. This chapter summarizes some of the steps leading up to the model including especially the genetic characterization of specialized transducing phages (λgal ) by recombinational rescue of conditionally lethal mutations. The serendipitous discovery of the conditional lethals is also described.

1 ANRV329-GE41-01 ARI 2 November 2007 12:22

PRELUDE what I was getting into. I had no special incli- This chapter covers part of my career around nation for research. I liked learning and en- joyed being a student and taking classes. I’ve gal: a cluster of 1960 and the questions that occupied my at- genes whose tention at that time. A brief summary of my read many accounts from scientists (including products are needed previous life indicates where I was coming some of my friends) who were motivated by to metabolize the from. At the end, I’ll say something about the ambition for research success from an early sugar galactose subsequent 40 years. age. Others of us just stumbled into research In high school I was inclined toward cre- careers. ative writing, but I thought a career in science My favorite professors at Berkeley had all should provide more secure employment. It been research-oriented, but that was not what was a fortunate decision, because I realized in had drawn me to them. I cared very little about retrospect that I have virtually no talent for how much they had accomplished. What I effective original writing. My best subject in liked was that they seemed to have thought school was mathematics. As a potential scien- deeply about the basic issues in their fields. It tist, I suffered from an inability to do much took me much longer to appreciate fully the of anything right in the laboratory. But I also linkage between depth of understanding and felt that experimental science kept me in touch hands-on involvement in research. with reality, whereas mathematics invited me In Urbana the Spiegelman lab was next to to dwell within a dream world of abstractions. the Luria lab. Salva’s interests were turning So in 1946 I enrolled as a chemistry major toward temperate phages, and Joe Bertani, at the University of California. My basic atti- then a research associate in Salva’s lab, was tudes toward scientific thinking were strongly performing many of the pioneering experi- shaped by the Berkeley chemists. Probably ments on lysogeny. I finished my thesis with my favorite professor was William Giauque, Sol in 1953, then started my first academic ap- who won a Nobel Prize for low temperature pointment as Instructor in Bacteriology at the chemistry during my senior year. I had never University of Michigan Medical School (an encountered anyone so completely immersed appointment interrupted almost immediately in science. He seemed a worthy role model by two years of military service). At Michigan, (and not because of his Nobel). Outside of I started out working on yeast (a continua- the Chemistry Department, my imagination tion of my doctoral research), but spent the was stimulated by the biochemically oriented summer of 1956 with Joe Bertani in Pasadena

by Annual Reviews on 12/13/07. For personal use only. microbiologists—especially Mike Doudoroff (where he had moved to the Cal Tech phage and Roger Stanier. Inspired by them, I de- group). I arrived with a plan for some ex- cided in my senior year to apply to grad- periments on restriction/modification (which uate school in . Roger recom- produced some very preliminary results), but Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org mended several places, and also directed then became engrossed with specialized trans- me toward C.B. van Niel’s course at the duction of the gal operon by phage λ. Hopkins Marine Station that summer (1950). van Niel became a second role model. I then became a graduate student of Sol Spiegelman’s SPECIALIZED TRANSDUCTION at the University of Illinois. Before I arrived, At that time, Jean Weigle had joined the Cal I.C. Gunsalus and Salva Luria had been re- Tech , but spent his summers cruited to Illinois, which then arguably had in Geneva (where he held a professorship the best bacteriology department in the world in physics). When I arrived in Pasadena for (at least for basic, nonmedical bacteriology.) the summer, he had left behind some pre- Graduate school application was much less liminary notes that Joe thought I might fol- organized than now, and I had little idea of low up on. Specialized transduction had been

2 Campbell ANRV329-GE41-01 ARI 2 November 2007 12:22

discovered by Larry Morse (16), but Jean tle fiefdoms. The tradition of arts colleges wanted to look into it further (19). I was un- comes from clerical institutions, which of aware at the time that he had also encouraged course had their own discipline but adapted SOS mutagenesis: in Geneva to investigate the more readily to a democratic structure. induction of same system. Eventually, Werner and I came, I was not offered any position in arts mutations by action quite independently, to very similar conclu- colleges at that time, but Miloslav De- of the products of sions. Two such parallel studies are about the merec, Director of the Carnegie Institu- some of the bacterial right number to provide corroboration but tion Department of Genetics at Cold Spring genes inducible by DNA damage avoid needless duplication. Harbor, wanted a one-year appointment to On returning to Michigan, I postponed replace George Streisinger, who was spend- further work on yeast (for 50 years now) and ing the year abroad. Cold Spring Harbor dedicated myself full time to the study of spe- was a marvelous place to do research, cialized transduction. I had a small NIH grant free from other obligations, and the senior to support the yeast work, so I informed NIH people—Demerec, Barbara McClintock and in my annual renewal that I was now work- Al Hershey—provided inspiration and moral ing mainly on phage. I am eternally grateful support. to have started my career at a time when such The initial task I set myself that year was flexibility was routine. to find out more precisely which segment of My initial work, paralleled by that of Arber the λ genome was missing from λgal. I knew (3), examined the effect of multiplicity of that the h gene was gone and that some other infection on the frequency and nature of genes were present. Dale Kaiser had made a the transductants (5)—something that nei- good map of λ, but the number of available ther Morse nor Weigle had done. The re- genetic markers was very limited. So I decided sults indicated that transduction was carried to make more. out by chimeric phages that had lost some of Jean Weigle had discovered that, if both their own genes while acquiring the nearby λ and are exposed separately gal genes. The phage DNA that was miss- to UV light and then the λ is plated on the ing included the h gene, which was in the irradiated host, a substantial fraction of the middle of the phage genetic map. The find- observed plaques are mutant. The mecha- ing heightened my consciousness that most nism was unknown. One favored hypothesis previous students of lysogeny had treated the at the time was recombinational rescue from

by Annual Reviews on 12/13/07. For personal use only. prophage as an indivisible unit (analogous to a postulated prophage homolog in the host, the concept of the gene that prevailed in the but it is now understood as inducible SOS first half of the twentieth century), whereas it mutagenesis arising during radiochemical re- might sometimes be more relevant to try trac- pair (9). My intention was to collect small Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org ing the origins of the individual genes of the plaque mutants, map them by phage crosses, prophage. then see whether their wild-type alleles could Two years in the Michigan Medical School be rescued from λgal. was enough for me, and I resigned the fol- lowing year, never again to serve on a med- ical school faculty. Although many medi- CONDITIONAL LETHALS cal schools have become more egalitarian I did not get that far, because of a chance than they were in 1957, the basic system discovery. I was routinely plating phage on is hierarchical. Historically, medical schools strain C600, where λ made nice large plaques. grew up around hospitals, which needed However, one day I ran out of C600 plating a chain of command from chief surgeon bacteria, so I replated my latest mutants through scrub nurses. Department chairmen on W3350, the gal strain used as recipient frequently served for life, ruling their own lit- in transduction. Both C600 and W3350

www.annualreviews.org • Phage Integration and Chromosome Structure 3 ANRV329-GE41-01 ARI 2 November 2007 12:22

are E. coli K-12 derivatives, bearing various wrote back something like “Great minds move laboratory-induced mutations. in the same tracks.” He too had been isolating To my dismay, there were no plaques on thermosensitive T4 mutants to complement the plates the next morning. But checking the amber work (8). things out showed that the result was real: Determination of phage gene functions Some of my small plaque mutants really did was not my primary goal. Luckily, many other not plate on W3350. Primed by Benzer’s rII workers became interested in λ, and the ge- work, I realized that these “host-dependent” netic hierarchy of λ development was eluci- mutants might be very useful in determining dated. Toyounger investigators, I recommend the gene content of λgal. I’ll finish up with sharing stocks freely. Not only is science ad- what became of mutants, then come back to vanced that way, but the donor of stocks can their use with λgal. keep abreast of what’s going on in the field. The mutations were widely distributed I also recommend free sharing of ideas, as across the λ map and seemed to belong to with Bob Edgar and me in 1959. Our 1959 many “cistrons,” in Benzer’s lexicon. Benzer’s conversation was very fruitful both for us and work was invaluable to me because it pro- for other workers in the field. Before appear- vided a simple operational protocol for classi- ing to seek the last word on that exchange, I re- fying my mutants by recombination and com- alize in retrospect that neither of us was paying plementation. In 1959, after a year at Cold full attention to what the other said. For exam- Spring Harbor and a subsequent year in Paris, ple, I clearly remember mentioning the work I stopped by Cold Spring Harbor on the way of Horowitz & Leupold (13) on thermosensi- to my new academic post in Rochester. I en- tive mutants, whereas Bob seems later to have countered Bob Edgar, who updated me on rediscovered the concept without that prior some Cal Tech T4 results of Dick Epstein’s. knowledge. On the other hand, I can’t really Epstein had isolated T4 mutants, which he remember what Bob thought at the time about called amber, that plated on a K-12 strain but the mechanism of amber suppression, which not on E. coli B. Like my λ mutants, they later proved to be through compensating er- were widely distributed around the map and rors between the cell’s RNA dictionary and defined many complementation groups. Fur- nonsense mutations in phage genes (18). Bob thermore, Epstein’s mutants plated on certain was probably thinking along those lines, but K-12 derivatives such as C600 and not on oth- I tended to block out hypotheses that seemed

by Annual Reviews on 12/13/07. For personal use only. ers, such as W3350. They plated only on those unnecessary for reaching the goals under dis- strains where my mutants (which I had sent to cussion. I could imagine a number of mecha- Jean Weigle at Cal Tech)also plated. Bob was nisms for genetic suppression that would af- enthusiastic about spending the rest of his life fect some, but not all, alleles of many genes; Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org identifying all the essential functions of T4. and discovering which one was responsible It didn’t work out quite that way. In a few in the case at hand did not seem like a high years, the Cal Tech group had classified a priority. number of the functions (10) and many other Amber mutants have of course become one labs were getting into the act. No technical of the workhorses of molecular biology. I’ve breakthrough belongs to its discoverers for since read numerous second-hand accounts very long. Following my conversation with about their discovery and use. Some of these Bob, I isolated many more amber mutants, read as though someone had started with a and also some thermosensitive mutants, ver- knowledge of missense suppression, then used ified that the amber mutants were respond- it to design a mutant screen. In fact, the ing to a suppressor mutation in permissive whole program could go on before the mech- strains like C600, and wrote up the results anism was thoroughly understood. Certainly (6). When I sent the manuscript to Bob, he neither Epstein nor I started out to design

4 Campbell ANRV329-GE41-01 ARI 2 November 2007 12:22

such a mutant screen. As indicated earlier, I for elements like λ and F that can be either just happened upon the λ mutants; whereas autononous or chromosome associated). This the first T4 ambers came from an attempt to gave me a reason to spend more time thinking F: a conjugative verify an erroneous hypothesis derived from about the literature in the field and whether I (one whose radiobiology. really understood it. bearer can act as Review writing is sometimes undervalued genetic donor in by the research community. Deans and pro- bacterial MAPPING OF λgals motion committees, for example, may view it conjugation) After that long digression, I return to my as a means to expand one’s bibliography with- own use of the mutants from 1958 onward. out accomplishing anything. And indeed some In his deletion mapping of T2rII, Seymour reviews fit that description. But it can also pro- Benzer had defind the methodology. No te- vide a time for fresh creative thinking if one dious conventional crosses to see what part tries to frame a satisfactory scheme to explain of the λ genome was present in λgal; simply the available facts in an area wider than one’s by spotting a mutant lysate and a lysogen for own research. So I approached the topic with the defective λgal on a lawn of the nonper- an open mind, hoping I would learn some- missive host W3350, irradiating lightly to lift thing. The most important ideas in the review prophage immunity and stimulate recombina- developed as I was writing. tion, I could test whether the wild-type allele One question that I re-examined was the of the mutant was present in λgal. attachment of episomes to chromosomes. The The results were clearcut. In my first λgal, model then favored, proposed by Jacob and some genes were present and others absent. Wollman, was that (although they This was true of every λgal. But each λgal had could be mapped among chromosomal mark- a unique gene content. All λgal were missing ers) were not inserted into the chromosome a “core region” including the h locus (and the but were instead synapsed with them because sites within the corresponding J gene defined of substantial homology. On this idea, λgal by conditional lethals) and some adjacent loci. should arise by a double crossover that uti- Thus each λgal had lost a connected segment lizes λ homology on both sides of gal. of the λ chromosome. So λgals could be used One paper that had appeared shortly be- to construct a deletion map of λ, similar to fore my review challenged this concept. Calef Benzer’s deletion map of rII. The unique- & Licciardello (4) crossed lysogenic strains λ

by Annual Reviews on 12/13/07. For personal use only. ness of each gal was corroborated by phys- genetically marked in both prophage and ical studies (20) showing that each λgal had chromosome. Their results fit the hypothesis a different buoyant density, which implied a that λ prophage was inserted into the chro- different DNA content. mosome, between gal and trp, but required Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org I’ll return to the origin of λgal shortly, I the gratuitous assumption that the gene or- was unable to square the results with the mod- der in the prophage had changed from the els that were then most popular, but I didn’t phage order (m5 co mi) to co mi m5. Play- have a better one in mind. ing with the possibilities, I noted that, if the phage genome were to circularize and then to insert by reciprocal crossing over between THE INTEGRATION MODEL m5 and co, their result would be obtained. At The “better model” resulted from a different about the same time, Frank Stahl mailed me a activity. Around 1960, Ernst Caspari (then the manuscript called “Some Circular Thoughts chair of my Department) invited me to write about Chromosomes,” which postulated, for a review for “Advances in Genetics.” I chose rather arcane reasons, that all chromosomes to write on “Episomes” (a term that had re- in bacteria, including phage chromosomes, cently been proposed by Jacob and Wollman should be circular.

www.annualreviews.org • Phage Integration and Chromosome Structure 5 ANRV329-GE41-01 ARI 2 November 2007 12:22

Reassessing, I finally concluded that Jacob Later on, Sankar Adhya, in my lab, selected and Wollman had arrived at their picture chlorate-resistant mutants of a normal λ lyso- of lysogeny by opposing a model (insertion) gen and mapped deletions penetrating the bio: a cluster of genes whose that made very specific predictions against a prophage (2). The chlorate resistance stems products are needed loosely defined model of lateral attachment from deletion of genes whose products cat- for biosynthesis of or synapsis with many more degrees of free- alyze steps in molybdopterin synthesis, two the vitamin biotin dom. Then, any result that contradicted the clusters of which, then called chlD and chlA, PI: a simplest version of the insertion model was bracket λ prophage. able to package taken as evidence for lateral attachment. To All these mapping experiments were facili- bacterial DNA and my mind, insertion (even with the gratuitious tated by the availability of conditionally lethal deliver it to other bacteria assumption of circularization) was still sim- mutants, which I had furnished to Rothman pler than lateral attachment. Insertion would and Franklin. Where the Calef/Licciardello then take place by a crossover between two experiments used three phage genetic markers circles that had some homology (very little, as and looked at various recombinant prophages it later turned out) at the crossover site. I also by examining plaque morphology, with the postulated that Jacob and Wollman’s other conditional lethals the genotype at each prototypic episome—the fertility plasmid F— prophage locus could be assessed by a simple was circular when autonomous and integrated spot test. into the chromosome by crossing over. Some years later, Enrico Calef called my attention to a cross between lysogenic strains published by Jacob and Wollman several IMMEDIATE AFTERMATH years before Calef/Licciardello, which could After the review appeared, it received a mixed be read as supporting the Calef/Licciardello reaction. The most sympathetic came from map; i.e., if you looked at the most com- those like Luria, who regarded the model as mon genetic classes and assumed they were beautiful but wrong (because of the Jacob/ single crossovers, the Calef/Licciardello map Wollman evidence.) Franc¸ois Jacob read the fell out. Table 1 shows the data. The crit- manuscript before publication, and provided ical classes are those showing recombina- detailed and encouraging comments. On the tion within the prophage, where the singles other hand, I kept hearing reports that Eli (S2 and S3) outnumber the triples. In the Wollman had a bacterial gene (I presume it same experiment, Jacob and Wollman also

by Annual Reviews on 12/13/07. For personal use only. was bio) that he’d mapped between gal and scored some recombinants that were donor λ prophage but was not included in λgal (al- type for both flanking markers (gal and 21) though he could get λbio phages that do not and found that 127/135 of these were donor include gal )—a strong argument against the type for the prophage marker (noncrossover Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org model, at least as it applied to λgal. class), whereas even the pure recipient type New supporting evidence came from Wu prophage (double crossover) was very rare & Kaiser (21), who showed that λ DNA actu- 1/122). ally could circularize as postulated, and from So why didn’t Jacob and Wollman inter- Luria’s and Yanofsky’s labs. June Rothman pret their result that way? Their genetic acu- (then a student with Luria) did PI transduc- men is unquestionable. Or why didn’t any tions with marked λ prophages that supported of the rest of us who read their paper (ex- the gene order I had proposed and also placed cept for Calef ) do so? At the time, one thing the λ prophage between gal and bio (17). that was known for sure about the λ genome And Franklin et al., using a λ that had been was the order of genes, as determined from placed in an abnormal location near trp, per- phage crosses. The map was linear, m5 was formed a deletion mapping of the prophage at one end, co was in the middle, and mi was together with nearby bacterial markers (11). at the right. This was a rock on which to

6 Campbell ANRV329-GE41-01 ARI 2 November 2007 12:22

Table 1 Recombinational map from a cross between lysogenic strains of E. coli a Recombinant genotypeb Gal (1) Co (2) mi (3) m5 (4) 21 Crossover classc No. recombinants (D) D D D (R) S4 276 (D) R R R (R) S1 57 (D) D D R (R) S3 12 (D) R R D (R) T134 1 (D) D R D (R) T234 3 (D) R D R (R) T123 1 (D) R D D (R) T124 2 (D) D R R (R) S2 6

a + + + − s − − + + R Hfr donor was gal (λm5co mi )21 Sm ;F recipient was gal (λm5 co mi) 21 Sm + + (Sm is unlinked to the region shown). Selection was for Gal + SmR. Only gal 21 recombinants are shown. Data from (14). bGenes are shown in the order they occur on the lysogenic chromosome. D: donor type; R: recipient type. cS: single; T: triple; regions 1,2,3,4 shown in column 1.

build. A single result putting the markers of distance came from crosses between nonlyso- the prophage in a different order seemed in- gens. I agree very much with Wollman’s logic. adequate to show that the prophage was ac- The expectation that λgal should include all tually rearranged. So instead the assumption genes between gal and λ prophage was crit- was that the gene order was the same in the ical; a single exception could have killed the prophage as in the phage, but that recombi- model. I wrote Wollman about some later re- nation frequencies were distorted by interac- sults, such as Rothman’s, but I never received tions with the chromosome. a reply. One problem was that, in both the Calef/Licciardello and the Jacob/Wollman crosses, there were only three genetic markers ROOTS AND RAMIFICATIONS on the prophage. Since there was no expected So that’sthe story of one segment of my career. orientation, the result boiled down to the fact Some additional comments seem in order, re- that, of the three markers used, the wrong ferring to issues already mentioned. by Annual Reviews on 12/13/07. For personal use only. one was in the middle. How much more obvi- ous might the cyclic permutation have been had the prophage been marked at five or Relation to Chromosome Structure

Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org six sites instead of only three! This was one From the outset, many geneticists (including dividend from the use of conditional lethals. myself) found insertion more appealing than Thus, Rothman had prophages marked at 4 synapsis or sticking. But the more appealing or 5 sites. To do the equivalent with con- an idea is, the more wary one should be of data ventional markers required distinguishing 16 that seem to support it. It helps to understand Hfr: a bacterial or 32 phenotypic classes—difficult when the why the idea was so attractive. strain where the F phenotypes are limited to plaque morphology. The roots of my own preference for in- plasmid has inserted into the As to Wollman’s mapping of bio between sertion go to some fundamental concepts of chromosome, gal and λ prophage, I never saw published chromosome structure and how they relate to creating a strain that data. There were notes in the Comptes Rendus lysogeny. All my conditioning, mostly from gives a High referring to future notes that would docu- Luria, Bertani, and Lwoff, strongly empha- frequency of ment that conclusion. I presume that it was sized that prophages behaved in most re- recombinants in bacterial crosses based on 2 factor crosses, and that the gal-bio spects like parts of the bacterial chromosome.

www.annualreviews.org • Phage Integration and Chromosome Structure 7 ANRV329-GE41-01 ARI 2 November 2007 12:22

At least at a gross level, λ prophage could chromosomes. I doubt that many genomicists be mapped; and an observer unaware that of the younger generation will have the pa- the phage genome had entered the cell from tience to go through the whole tortuous path outside would see the prophage as just an- of genetic logic that I have presented here, other chromosomal segment. So I rejected when the same conclusions come out of sim- the notion that prophages were attached to ple sequencing. But we may still ask at this chromosomes in a different manner than one point: Suppose the synapsis model is right— chromosomal gene is connected to another; not for λ, but for some other proviruses. if prophage could add by synapsis, some of After all, diversity is the general rule in the genes already classifed as “chromosomal” living systems; usually every imaginable pos- must have added that way in the past. In prin- sibility eventually shows up somewhere. The ciple, I was open to the possibility that chro- question for the current molecular geneticist mosomes were more complex than it seemed. is then, if there are proviruses attached by After all, most chromosomal mapping had synapsis, would they be identified by current been performed in an era where genes could methods? be treated as beads on a chromosomal string. Such a provirus, being unconnected to the Maps were said to be linear, but the evidence chromosome by covalent bonds, would prob- spoke only to the unidimensionality of the ably be classified as a plasmid. There are of string. The internal structure of the beads be- course prophages such as P1 that are perpetu- gan to be accessible in the 1950s, culminat- ated as autonomous , unlinked to any ing in Benzer’s work on T4rII. Benzer’s re- chromosomal genes. But a synapsed prophage sults strongly indicated the linearity of short would show linkage to its chromosomal ho- segments of the genome, going down to intra- molog. Genomics has effectively redefined genic segments, but stopped short of integrat- “chromosome.” Classically, chromosomes are ing these short segments into the continuity visible structures containing RNA and pro- of the chromosome. tein as well as DNA. Most of that DNA com- In 1960, DNA chemistry was less advanced prises a continuous double helix. When DNA than it is today. Serious evidence-based pro- is isolated in preparation for sequencing, any posals were being floated that some cellular noncovalent bonds are ruptured so that only DNA was four-stranded rather than two- the covalent connections are recorded. But is stranded (7) and that chromosomes were con- there evidence at any level for DNA segments

by Annual Reviews on 12/13/07. For personal use only. structed from short DNA segments joined by (not necessarily prophages) that are associ- protein linkers (12). Much of that ambiguity ated with a specific part of the chromosome has passed with technical improvements and by noncovalent bonds? If they existed, would with sequencing projects that assemble DNA they have been found? Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org segments into complete molecules. Thus we I think they would have been found be- can now assert with strong confidence that the cause genetics would have put them on the genes of the E. coli chromosome lie on one chromosome and genomics would have ex- continuous circular molecule and that each cluded them from it. That assumes that the of the four Drosophila chromosomes contains elements in question are recognizable by some a single linear DNA molecule. These facts phenotypic effects. change our perspective but don’t completely Some eukaryotic viruses are chromosome- resolve all the older questions. associated. For example, the of Epstein Insertion of λ prophage is an established Barr Virus and Bovine Papillomavirus are at- fact. The molecular proof started with the tached to chromatin by specific viral proteins; sequencing of junction fragments by Landy when the gene encoding the protein is deleted, & Ross (15), who initially got the fragments viral genomes are no longer segregated regu- from λgal and λbio rather than from lysogenic larly at mitosis (1). However, the only known

8 Campbell ANRV329-GE41-01 ARI 2 November 2007 12:22

persistent eukaryotic viruses that map to a esis of λgals. I said I did not really understand specific chromosomal location are those, like it, and that I could only postulate some kind

retroviruses, that are inserted. of looping out followed by breakage and join- F : a derivative of F The location of F elements in a cell that ing of heterologous DNA. (I suspect that my that has incorporated contains a complete chromosome is also rel- actual words were much less coherent than a segment of evant. The F can certainly recombine with what I’ve just written.) Max then proclaimed bacterial DNA its chromosomal homolog, as evidenced by that the idea was “very ugly” and that nei- chromosome mobilization; and a strain where ther it nor the model should be discussed fur- the F has actually inserted seems a reason- ther. But of course if you look at Benzer’sdele- able intermediate in mobilization. However, tion mapping (which Max seemed to accept), I don’t know of strains where the integrat- the molecular mechanisms that generate ing crossover has been shown to take place at most deletions are equally obscure (“ugly?”). a specific site within the homologous region. Benzer did not make the tactical mistake of Both in insertion and excision, there must be trying to explain how deletions arise. Both a stage of synapsis preceding recombination, deletions and λgals are produced by events and if that stage is long, the F could spend that are much rarer than homologous recom- most of its time neither in the chromosome bination, so that crosses with conventional nor fully autonomous. markers can delineate the endpoints.

Experimental Basis of the Model Tandem Duplications as a Source For a few years after it was proposed, my of Instability model was sometimes described as “purely Another reaction came from Bill Hayes. Bill theoretical,” as opposed, I guess, to other wrote a highly (and deservedly) regarded models that are more directly data-based. book, The Genetics of Bacteria and Their Viruses, This did not bother me at the time. I myself whose definitive nature has not since been regarded all models as tentative and subject to equaled. He covered the insertion model in revision. But in retrospect, what was special the 1964 version of the first edition. (His about my model seems primarily a matter of viewpoint was modified substantially by the presentation. If one compares my results on second (1968) edition.) I felt honored to be λgal with, say, Benzer’s “deletion mapping” included. Despite some obvious skepticism

by Annual Reviews on 12/13/07. For personal use only. of rII mutants, both data sets can be used about whether λ and F are really inserted, in equivalent arguments; each λgal, like each he kept returning to the model to account deletion, representing a connected segment of for some of the facts, especially about Fs. At the lysogenic chromosome. It didn’t occur to least the model had explanatory value. But Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org me to make that argument at the time partly one of his assumptions surprised me; if he because the same review where I proposed doubted whether λ was really inserted, he had insertion was also the first place I had sug- no doubts about λgal. He thought it was not gested how λgal might originate. Both ideas inserted, for reasons too obvious to require were new to me and I still felt uncertain about discussion. them. As far as I can tell, Hayes’ view of λgal λgals result from heterologous breaking (which I also encountered in other bacterial and joining of prophage and bacterial DNA, geneticists at the time) stemmed from the fact creating one “novel joint” characteristic of that λgal could rather easily be lost from a each λgal. I still remember a conversation cell that carries it. Plasmids could be lost; in- with Max Delbruck on the subject. At first, he tegrated elements should be stable. My own seemed very interested in the insertion model, thinking was that, since a strain with an inte- but then at one point he asked about the gen- grated λgal has two copies of the gal operon

www.annualreviews.org • Phage Integration and Chromosome Structure 9 ANRV329-GE41-01 ARI 2 November 2007 12:22

in close proximity and in the same orien- the genetics and biochemistry in depth. tation, recombination between the two can These included Bob Weisberg,Howard Nash, loop out the λgal, leaving one copy of gal Kiyoshi Mizuuchi, and Art Landy, to mention behind. just a few. I have continued to work in the area, Such thinking was not common among looking especially at regulation. I’ve also ded- bacterial geneticists at the time. As a grad- icated some time to biotin biosynthesis, evo- uate student at Illinois, I was fortunate to lution of λ related phages, and genomics. In have taken a cytogenetics course from Mar- most of these areas, students and collaborators cus Rhoades, where we worked out many contributed heavily to my lab’s output. I won’t of the consequences of homologous recom- try to list them, because they’re not part of bination between and within chromosomes. the story I set out to tell here. At the moment, I mistakenly assumed a readership to whom my long-time collaborator Alice del Campillo this was, likewise, classroom stuff. I’ve some- Campbell and I are studying the evolution of times thought that my main contribution to the phage-coded enzymes needed for inser- prokaryotic genetics could be characterized tion and excision, and their regulation. simply as applying the principles of eukary- otic cytogenetics to bacteria. DEDICATION This chapter is dedicated to the memory AFTERWORD of Esther M. Lederberg (1922–2006) whose After 1965, λ insertion attracted the interest early work laid some of the ground for the of a host of talented investigators who probed author’s research described herein.

SUMMARY POINTS 1. Bacteriophage λ is perpetuated in lysogenic cells as a prophage inserted into the chromosome. 2. The gene order of the inserted prophage is a cyclic permutation of that in the free phage particle. 3. Specialized tranducing phages, such as λgal played a critical role in demonstrating

by Annual Reviews on 12/13/07. For personal use only. insertion before direct sequencing of bacterial genomes became possible.

Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org DISCLOSURE STATEMENT The author is not aware of any biases that might be perceived as affecting the objectivity of this review.

LITERATURE CITED 1. Abroi A, Ilves I, Kivi S, Ustav M. 2004. Analysis of chromatin attachment and partitioning functions of bovine papillomavirus type I E-2 protein. J. Virol. 78:2100–13 2. Adhya S, Cleary P, Campbell A. 1968. A deletion analysis of prophage lambda and adjacent genetic regions. Proc. Natl. Acad. Sci. USA 61:956–62 3. Arber W. 1958. Transduction des caracteres` gal par le bacteriophage´ lambda. Arch. Sci. (Geneva) 11:259–358

10 Campbell ANRV329-GE41-01 ARI 2 November 2007 12:22

4. Calef E, Licciardello G. 1960. Recombination experiments on prophage host relationships. Virology 12:81–103 5. Campbell A. 1957. Transduction and segregation in Escherichia coli K12. Virology 4:366–84 6. Campbell A. 1961. Sensitive mutants of bacteriophage l. Virology 14:22–32 7. Cavalieri LF, Funston R, Rosenberg BH. 1961. Multi-stranded deoxyribonucleic acid as determined by x-irradiation. Nature 189:833–34 8. Edgar RS, Liclausis I. 1964. Temperature sensitive mutants of bacteriophage T4D: their isolation and genetic characterization. Genetics 49:649–62 9. Ennis DG, Fisher B, Demiston S, Mount DW. 1985. Dual role for Escherichia coli Rec A protein in SOS mutagenesis. Proc. Natl. Acad. Sci. USA 82:3325–29 10. Epstein RH, Bolle A, Steinberg CM, Kellenberger E, Boy dela Tour E, et al. 1963. Phys- iological studies of conditional lethal mutants of bacteriophage T4D. Cold Spring Harbor Symp. Quant. Biol. 18:375–92 11. Franklin NC, Dove WF, Yanofsky C. 1965. The linear insertion of a prophage into the chromosome of E. coli shown by deletion mapping. Biochem. Biophys. Res. Commun. 18:910– 23 12. Freese E. 1958. The arrangement of DNA in the chromosome. Cold Spring Harbor Symp. Quant. Biol. 23:13–18 13. Horowitz NH, Leupold U. 1951. Some recent studies bearing on the one gene-one enzyme hypothesis. Cold Spring Harbor Symp. Quant. Biol. 16:65–74 14. Jacob F, Wollman EL. 1957. Genetic aspects of lysogeny. In The Chemical Basis of Heredity, ed. WD McElroy, B Glass, pp. 468–98. Baltimore: Johns Hopkins Press 15. Landy A, Ross W. 1977. Viral integration and excision: structure of the lambda att sites. Science 197:1147–60 16. Morse M, Lederberg E, Lederberg J. 1956. Transduction in Escherichia coli K-12. Genetics 41:121–56 17. Rothman JL. 1965. Transduction studies on the relation between prophage and host chro- mosome. J. Mol. Biol. 12:892–912 18. Stretton AO, Brenner S. 1965. Molecular consequences of the amber mutation and its suppression. J. Mol. Biol. 12:456–65 19. Weigle JJ. 1957. Transduction by coliphage l of the galactose marker. Virology 4:14–25

by Annual Reviews on 12/13/07. For personal use only. 20. Weigle J, Meselson M, Paigen K. 1959. Density alterations associated with transducing ability in the bacteriophage lambda. J. Mol. Biol 1:379–86 21. Wu R, Kaiser AD. 1968. Structure and base sequence of the cohesive ends of bacteriophage lambda DNA. J. Mol. Biol. 35:523–37 Annu. Rev. Genet. 2007.41:1-11. Downloaded from arjournals.annualreviews.org

www.annualreviews.org • Phage Integration and Chromosome Structure 11