Copyright 0 1989 by the Genetics Society of America Perspectives Anecdotal, Historical and Critical Commentaries on Genetics Edited by James F. Crow and William F. Dove REPLICAPLATINGANDINDIRECTSELECTIONOFBACTERIAL MUTANTS: ISOLATIONOFPREADAPTIVEMUTANTSINBACTERIABYSIBSELECTION ISTORICALLY and conceptually, the two about genetic recombination in bacteria (LEDERBERG H themes in these titles (LEDERBERGand LEDER- 1987). BERG 1952; CAVALLI-SFORZA and LEDERBERG 1956) By 1950 it was no longer controversial among ge- are complexly intertwined. Replica plating is a homely neticists. Puce periodic selection, the monotonic in- methodology to ease the technical burdens of screen- crease of mutant number with time in long-term che- ing large numbers of bacterial colonies for mutants to mostat cultures observed by NOVICK and SZILARD use in genetic analysis. This was sometimes easy. For (1950) sealed lacunae in the 1943 experiments. (Oth- example, phage or drug resistance could be readily erwise, one could have entertained some labored hy- obtained with positive selection using phage or a drug, pothetical alternatives like uncontrolled environmen- respectively. But many other kinds of mutants were tal variation from tube to tube, or a subset of that: desired, and replica plating facilitated their discovery. temporal fluctuations within a culture in its overall Furthermore, the broader philosophical question re- propensity to turn resistant on exposure to a selective mained open, whether the resistance was a preadap- agent.) Nevertheless, many, perhaps most, readers of tive or a postadaptive change: did it emerge by natural the 1943 article did not understand its abstruse math- selection of rare preexisting mutants or did the agent ematical argument and would respond either with somehow induce the heritable change? I elaborate uncritical acceptance or uncritical rejection. Notable first on this second broader question and its resolution for the latter, the last holdout, was the prestigious Sir by indirect (sib) selection. CYRIL HINSHELWOOD, President of the Royal Society LURIA and DELBR~~CK (1943) introduced a carefully of London, who denounced every assertion of genes thought out quantitative kinetics into the study of in bacteria in favor of an extended reaction network bacterial mutation. The jackpot theory (LURIA 1984) model of biological continuity and adaptivity (HIN- was a consequence of clonal expansion. Spontaneous SHELWOOD 1946; DEAN and HINSHELWOOD 1957). As mutations, in a culture started from a small inoculum, inappropriate as these network models have proven would occur rarely among the few cells present at to be for the fundamental elements of genetic struc- early stages of exponential growth. But when this ture, they have been revived for the explanation of happened they would have a disproportionate number developmental switches, where one gene product rein- of mutant offspring. Hence, while the distribution of forces its own synthesis and represses an alternative mutational events should follow a Poisson distributior. gene’s and vice versa (DELBR~~CK 1949; PTASHNE over a series of similar cultures, the distribution of 1986). mutant cells would show occasional jackpots. Uncited Meanwhile, bacterial genetics had grown to the by and perhaps unknown to LURIA and DELBR~CK, point of needing ever larger libraries of mutants, most YANG and BRUCE WHITE (1934) had previously noted of which bore biochemical defects and were not read- such fluctuations and remarked that they signified the ily amenable to positive selection. Today we apply spontaneous occurrence of “rough” mutants in Shi- many ingenious tricks to this game (VINOPAL 1987). gella. In 1948, the penicillin method (LEDERBERG and ZIN- Nevertheless, the theoretical rigor of LURIA and DER 1948; DAVIS 1948) partly answered this need by DELBR~JCK’Sanalysis and its concurrence with the selectively killing cells actively growing in minimal quantitative data were for many the decisive stimulus medium. Penicillin could exert positive selection in to look seriously at mutation in bacteria, to think favor of auxotrophic mutants. But many bacterial about a bacterial genetics. In my own development, strains were relatively recalcitrant to the feeble mu- this paper was one of the key impulses to inquire tagens than available, and even after penicillin selec- Genetrcs121: 395-399 (March, 1989) Perspectives 397 roosters are selected by indices of milk and egg pro- pointed out, this view taken to an extreme contradicts duction by their sisters and daughters (LUSH 1945). the understanding of the gene as a material substance At a time when the genetic basis of altruism is in (we would now say as DNA). We must have an open question (MICHOD 1982), we can reflex on a system mind about evolutionary specializations where meta- whereby replica-platefuls of clones, sensitive and re- bolic alterations can target the DNA itself. This might sistant alike, were relegated to the autoclave once they sometimes lead to postadaptations, that is, adaptive had given the locational cues for their indirectly se- genetic changes specifically induced by an environ- lected cousins. mental stress. As specialized evolutionary develop- As satisfying as this demonstration was, I lamented ments, one does not expect them to be a routine that it depended on a new technology (that of replica occurrence. They have been hard to find and authen- plating). An Euclidean bent has always led me to seek ticate, with one generic exception: lysogenizing vi- for proofs that came closer to the use of nothing more ruses typically confer immunity to the lytic function than a straight edge and a compass. Furthermore, it of the virus, a subset of lysogenic inductions (HERSHEY was difficult to quantitate the progress of indirect 197 1; PTASHNE 1986). The frequency of “processed selection on agar plates. It was difficult to maintain pseudogenes” in eukaryotic genomes is particular tes- rigorous control over the proportion of cells that were timony to a history of RNA insertions (see pp. 448ff. transferred during replica plating or the proportion in DARNELL,LODISH and BALTIMORE 1986).With site- that could be plucked by going back to the source directed mutagenesis we know today how to expose, plate. There was no reason, I thought, that the same or even directly use, DNA sequences so as to achieve methodology could not be used with an array of test mutation by intelligent design. tubes taking the place of the agar plate, with pipettes To turn to less specific responses, mutational storms transferring well defined volumes from tube to tube hypothetically related to insertional transposons may in place of the velvet transfer. This was the burden of also be mediated by RNA in train of environmental CAVALLI~FORZA and LEDERBERG(1956): during one stress (MCDONALD 1983). In addition, contemporary of several visits to our laboratory, CAVALLI took an with the development of replica plating, my own interest in following the quantitative aspects of selec- laboratory had concluded that UV mutagenesis was tion using that methodology. If one starts, say, with probably a secondary physiological response during 100 tubes so adjusted that a mutational event will recovery and DNA repair (LEDERBERGet al. 1951). have occurred in only one of them, one can find out We know now how environmental DNA damage can which tube that was by testing a sample of each for evoke the “SOS” response (WALKER 1987) leading to drug resistance. Discarding the other 99 tubes would several categories of genetic instability. One could in principle then give a lOO-fold enrichment of the argue that UV resistance was a postadaptive mutation, proportion of mutant cells within that tube compared but this is to ignore the broad-ranging nonspecificity to the overall population. Then one knows what di- of the mutations that occur during the SOS response. lution to make of the culture in that tube for a second It is too early to answer many questions that have cycle of inocula and can recycle from there with been raised during the past few months about posta- progressive enrichment. CAVALLI found that this pro- daptive mutational responses claimed to occur in gly- cedure worked very well, although this progress of cosidase-deficient Escherichia coli mutants incubated selection was often not as rapid as predicted by the in the presence of lactose or salicin (CAIRNS, OVER- first-order theory. This could be explained by the BAUGH and MILLER 1988; CAIRNS et al. 1988; HALL observed growth lags of the resistant mutants and thus 1988). There are many pitfalls in the exclusion of offers no great problem. artefacts in mutagenesis experiments (LEDERBERG Replica plating has grown into a major industry, its 1948) and each has entrapped unwary investigators in progeny including the various blots around the com- the course of microbial genetics. Such artefacts aside, pass-Northern, Southern, etc.-as well as its direct cells starved for carbon but receiving a trickle of application to a number of microbiological problems. nutrient through spontaneous or allospecific enzy- Sib selection by serial dilution has been used in several matic hydrolysis of the substrate are in a metabolic genetic engineering enterprises where a special pro- state that requires critical examination. As a reducing ducing clone is a needle in the haystack but its prod- sugar, unmetabolized lactose might well be expected ucts can be sensitively detected (KEDES et al. 1975; to react with the DNA of such mutants (LEE 1987). NAGATA et al. 1980). Even if it should be verified that lactose is a mutagen A 1989 perspective on postadaptive mutation: The for lac- mutants, its specificity, i.e., lactose at the lac concept of the gene as immutable in the course of locus us. salicin at the bgl locus in a common genetic hereditary transmission (MORGAN 1926) was an ideal- background, should be corroborated before further ization that played a constructive role in the emer- evolutionary speculation. gence of Mendelian genetics. As HALDANE (1949) Specific DNA alterations are achieving higher cred- 398 J. Lederberg ibility in a role in epigenesis, in reaction against long- preadaptive mutants in bacteria by sib selection.