SPONTANEOUS BACTERIAL MUTATIONS TO RESISTANCE TO ANTIBACTERIAL AGENTS S. E. LURIA Bacterial genetics differs from the genetics of higher organisms chiefly because of lack of informa- tion about the mechanism of orderly transmission of characters from generation to generation, owing to the absence of well-established processes of con- jugation and to the infant state of bacterial cytol- 0,~. Our only way of approach is the study of bac- terial mutability, for which the field of bacterial variation offers a wide range of material. One of the first tasks we face is that of establishing how closely mutations in bacteria are comparable to the so- called gene mutations in higher organisms. Geneti- cists as well as bacteriologists have often thought that the genetic system of bacteria may be less orderly; in particular, that properties may be trans- mitted at fission by more or less random distribu- tion of substances present in the protoplasm, with- out special mechanisms for equational division. If a case could be made, however, for similarity of the processes of mutation in bacteria and in higher organisms-that is, for the existence of discrete, gene-like hereditary units in bacteria- then these organisms might prove to be invaluable material not only for the study of physiological genetics, but also for an attack on the problems of gene structure and mutability (4, 5). The case would necessarily rest on analogies, some of which may be listed as follows: (1) per- manence of the mutated character; (2) "spontane- ous" occurrence, independent of specific environ- mental stimuli; (3) definite rate of occurrence; (4) independence of different mutations in the same organism; (5) inducibility by agents known to produce mutations in higher organisms; (6) rever- sibility, showing that the mutation is not neces- sarily a loss, caused by unequal division or some similar process; (7) physiological effects, fumish- ing proof that bacterial mutations can bring about the same type of metabolic changes known to be produced by gene mutations in higher organisms. One difficulty of bacterial genetics is that of recognizing a given mutational step, since the same character may be affected by a number of different mutations, which we cannot attribute to different genetic loci by classical genetic tests. Hence the necessity of identifying each mutation by as many of its phenotypic effects as possible. Another peculiarity of bacterial genetics is that we generally do not deal with the properties of single individuals, but only with those of large "clones" (colonies, single-cell cultures), each de- rived from one individual whose characters must be retraced from those of the clone. We must there. fore take into account intraclonal variability. Cases of high mutation frequency are generally unsuitable for the study of the mutation process, since they are likely to be balanced by reverse mutations; our visible clones will then contain mutant and re verted types in genetic equilibrium. These con- siderations have been discussed in greater detail by Delbriick (8). More convenient are cases with low mutation frequency, since these need not necessarily be balanced by either reversion or adverse selection. In these cases the mutants can be detected, however, only if we can select them out in a suitable environ- ment. Mutants capable of growth in media "in- complete" for the wild type (22, 19) are in this category. Other such mutations are those producing increased resistance to some destructive agent. We shall discuss in this paper the latter group of mutations, on the basis of work on bacterial re- sistance to bacteriophages, antibiotics, drugs, and radiation (18, II, 1, 2, 9, 24, 21). ORIGIN OF RESISTANT VARIANTS The resistant variants are generally isolated as colonies which appear after the parent strain has been plated in the presence of a destructive agent. The question then arises: were the cells that gave these colonies resistant before the plating, or have they acquired hereditary resistance as a response to the action of the antibacterial agent? Unappealing as the second alternative may be to classical genetic thought, the question is by no means idle. Con- siderable evidence has been interpreted as support- ing the idea that environmental conditions, in particular nutritional or inhibitory stimuli, may alter the metabolic machinery of the bacterial cell, not only temporarily, but in a permanent hereditary way (14, 13). It may be conceivable that a small fraction of the cells exposed to an antibacterial agent respond to its action with a specific change in properties, and that this change is somehow transmitted to the offspring. Mutational origin of the resistant variants, in- dependent of the action of the destructive agent, can only be proved either by direct isolation of resistant substrains before the treatment (7)) or by establishing the presence within sensitive cultures of "clones" of resistant bacteria, each clone stemming from a mutation that occurred during the growth of the cultures prior to the test for resistance (IS). If we define mutation rate as the probability that a E 130 1 BACTERIAL bacterium will mutate in a given interval of time, the frequency of mutations will be proportional to the number of bacteria present. Each mutation gives Origin to a "clone" of mutants; the earlier the mu- tation OCCUTS, the larger the clone derived from it will be at the moment of the test. If the. mutants sow at the same rate as the normal bacterra, clones of a certain "size" will be twice as frequent as those of double that size. The expectation should be, therefore, that, of the mutants present in a culture at any one time, equal numbers stem from muta- tions that occurred in each of the previous genera- tions. This situation, however, is never realized in actual cases. In any actual culture, the likelihood MUTATIONS 131 same culture. A detailed analysis of these considera- tions has been given by Luria and Delbriick (18). These expectations can best be tested in the case of resistance to bacteriophages. The phage- resistant variants are easily detected and counted in a mixture with sensitive cells; they neither grow the phage nor adsorb it, and their resistance is a permanent new character of the strain. As an example, Table 1 (from 18) shows the results of the enumeration of bacteria resistant to phage Tl in platings from similar cultures of Escherkhia co.%, strain B. The presence and type of fluctuations in the frequency of mutants provide strong confirmation of the mutational origin of re- TABLE 1. DISTRIBUTION OF THE NUMBERS OF BACTERIA RESISTANT TO PHAGE Tl IN SERIES OF SIMILAR CULTURES OF SENSITIVE BACTE~UA Experiment No. Number of cultures Volume of cultures, cc. Volume of samples, cc. 22 23 100 87 2: 1; * Resistant Bacteria 0 : i Al Resistant Number of' Bacteria Cultures 0 29 1 3 17 4 610 5 : 3 11-20 21-30 i 11-20 6 21-50 51-100 : 51-100 : 101-200 101-200 201-500 0 201-500 i 501-1000 1 5014000 0 Average per sample Variance (corrected for sampling) Average per culture Bacteria per culture %t 28.6 6431 40.48 28.6 2.8 x 10" 2.4 X 10' * Cultures in synthetic medium. of occurrence of early mutations is small, because not enough individuals are present. When the popu- lation reaches a size where mutations begin to occur, the frequency of their occurrence will be subject to tremendous fluctuations owing to their small aver- age number. In a large number of trials (series of similar cultures), therefore, we must expect to find COrresponding large fluctuations of the frequency of mutants; the presence of these fluctuations is a direct consequence of the clonal distribution of the mutants. When the proportion of variants in a series of similar cultures shows such large fluctua- tions! we may interpret them as proof of a clonal distnbution, hence of a mutational origin of the wiants. Such fluctuations among similar cultures would not be expected if resistance were actively produced hy the antibacterial agent. The fluctuations should not be greater than those among samples from the sistance. A refinement of this type of analysis would be afforded by a comparison of the actual distribu- tion of the number of variant individuals with a distribution calculated from the assumptions of the mutation hypothesis. Mathematical difficulties have been encountered, however, in attempts to calculate the theoretical distribution. It is important to notice in Table 1 (Exp. 23, in which the whole contents of the cultures were tested) the presence of large numbers of cultures containing one mutant only. This suggests that the spontaneous mutations may be phenotypically ex- pressed without a lag longer than one cell genera- tion. The "fluctuation" test has been' applied to a variety of cases-resistance of EC Goli B to several phages (II) and to radiation (24) ;` resistance of Staphylococcus 3 13 to penicillii (9)) and to sulfa- thiazole (II)-with results of the same type, fndicat- 132 S. E. ZURIA hg that in all these cases resistance arises as the're- sult of mutations occurring before the test. The data of Lewis (16) on Escherkhk cold mutabile show that similar fluctuations between cultures were pres- ent in the numbers of lactose-fermenting cells. Large fluctuations in the time of appearance of fermenta- tion variants were also present in experiments of Kristensen (15). The mutational origin of bacterial variants is well supported by all these observations. The type of test described above shows only that the variants were already a genetically modified type before the test. It is possible that in some cases the specific environment, besides selecting out the mu- tants, may also act by rendering phenotypical a new character otherwise masked by some cytoplas- mic effect. In a sense, this may be considered to be true of mutations affecting the production of adap- tive enzymes. In the case of mutations to resistance to antibacterial agents, similar instances might be difficult to detect. MUTATION RATES In work on bacteriophage resistance, Luria and Delbriick (18) have chosen as a definition of the mutation rate the probability of mutation per bac- terium per physiological time unit (generation). This detition is satisfactory as long as it can be proved that mutations occur only during growth of the bacteria, and that their frequency is propor- tional to the physiological rather than to the as- tronomical time. These conditions appeared to be fulfilled in our experiments on phage resistance (mutation rate independent of generation time in different media, no increase in mutants after maxi- mum growth). Several methods have been proposed for estimat- ing the mutation rate. Since mutations occur in- dependently of each other, when there are only a few they will be distributed among different cultures in a Poisson distribution, with the probability of any given number of mutattons occurring in one culture depending only on the average number of muta- tions per culture. The average number of mutations per culture, and hence the mutation rate, can be calculated from the number of cultures having no mutants and from the number of bacteria per culture. If. in a series of C simiir cultures, each containing N bacteria, a fraction C& proves to contain no mutant (the whole contents of the cultures must obviously be tested), the mutation rate, a, can be obtained: (1) In 2 x In (G/C) a=- N This method, although it requires careful plan- ning of the experiments, has the advantage of not introducing the actual numbers of mutants and thus avoiding complications due to possiile selec- tion. Another method that allows direct measurement of mutation rates without making use of the num. hers of mutants consists in applying the antibac. terial agent (phage) to bacteria growing on solid medium without disturbing the formation of micro. colonies (20). One colony appears for each mute- tion that has occurred. The difficulty here lies b precise estimation of the number of bacteria pres- ent at the moment of the test. Mutation rate can also be calculated from the actual number of mutants, estimating from this the number of clones present. One must take into so count that in cultures of limited size mutations are likely to occur only during the last part of their growth. If, for example, the mutation rate is 10-T, mutations are not likely to occur, in actual Cultures, until the population reaches a size of the order of 10' cells. Introducing the necessary corrections, the mutation rate can be estimated from the average number, r, of mutants from C similar cultures: aNC (2) r=EhrF A diagram for the use of this formula has been given by Luria and Delbriick (18). Mutation rates calculated by formula (2) are likely to be in- accurate because of the fluctuations in frequency of occurrence of the early mutations; in particular, they will be too high whenever one or more of the unlikely early mutations happens to occur. Impor- tant corrections should be introduced if the growth rates of normal and mutant types are different. Altogether, our present methods of estimating mu- tation rates are far from satisfactory. Some values of mutation rates calculated with formulas (1) and (2) are given as examples in Table 2. These values are subject to the further limitation that all mutations giving the same pheno- type for the character investigated are lumped to- gether, and that in some cases it is likely that not all mutants present in a sample are detected. The orders of magnitude, however, are probably correct. Even for the same wild-type strains, mutation rates for different mutations to resistance vary over a wide range. That they are generally very small is to be expected; in searching for stable resistance to destructive agents, we particularly select for rare mutations, since if stable mutants occurred very frequently they would generally have displaced the parent type, unless checked by strong adverse selection. Much higher mutation rates may he found for mutants that also show high rates of re- version. INDEPENDENT MUTATIONS It has already been mentioned that the same by terium can undergo different mutations. This 1s clearly seen in the case of resistance to bacteria- phages. Working with Escherichia co& B and bsc- teriophages Tl-T7, one finds a series of bacterial mutations, each producing resistance to one, two, BACTERIAL MUTATIONS 133 0r more phages. Indicating a mutant from B that is ,esistant to phage n by the symbol B/n, we have ior instance the mutants: B/l; B/l, 5; B/6; B/3, ) 7. From each of these mutants one can obtain &cr mutants with additional resistance to other @ages: B/l/6; B/6/3, 4, 7, etc. The successive mutations are clearly independent. 4 strain that has undergone a certain mutation will &rally give the same mutations to resistance to zther phages as the parent wild-type strain. Within the limits of precision of the available methods of measurement, the rates for the same mutation in the parent strain and in any of its mutants appear could decide by what mechanism resistance to un- related phages is produced in one mutational step. The answer to this question might be of the greatest importance, also, in relation to the mechanism of bacteriophage action. A possible interpretation has been suggested by E. H. Anderson's work on the nutritional require- ments of some phage-resistant mutants (I, 2). Anderson found that a number of B/l mutant strains from Escherichia coli B differed from the wild type also in being unable to grow in a mini- mal synthetic medium. The metabolic disturbance was identified as inability to synthesize Ltrypto- TABLE 2. MUTATION RATES Organism Escherichia coli B Staphylococcus aureus 313 = -- Mutation Reference Resistance to phage Tl Resistance to phage Tl Resistance to phage T3 Resistance to phage T6 Resistance to radiation Resistance to penicillin, one-step Resistance to sulfathiazole, one-step to be equal. For example, the mutations B + B/l, and B/6 -+ B/6/1 occur with the same frequency (11). The same holds true for other types of mutation as well. The mutant B/r resistant to radiation (24) has the same pattern of mutability to phage re- sistance as the wild-type B. These results provide evidence of the independ- ence of different mutations in bacteria, and agree with the hypothesis of the existence of discrete, mutable, gene-like units. MULTIPLE EFFECTS OF MUTATIONS In cases of resistance to drugs and antibiotics, the resistance is often very specific; cross-resistance is limited to related substances. In the case of phage resistance, however, multiple resistance to several phages may appear as one mutational step. The phages to which resistance can be produced by one mutation are often unrelated as judged by their size, structure, serological specificity, and growth characteristics (8). It is easy to prove, on the basis of the mutation rates, that the multiple resistance does not result from chance occurrence of two or more independent mutations in the same bacterial line. Moreover, resistance to the same group of phages can be reached in one or two mutational steps. It is clear, therefore, that tests of resistance to a limited number of phages are not sufficient to characterize the genotypes of a group of bacterial mutants. It would be an important step forward if we Mutation Rate per Bacterium per Generation 40-7 -10-g phane, which therefore became a required nutrilite for these strains. Moreover, these strains exhibited another disturbance in their nitrogen metabolism, reflected by their inability to utilize inorganic nitro- gen in the absence of at least one of a series of amino acids. Additional inability to synthesize the amino acid proline was found by Anderson in a strain B/1/3, 4, 7. Other phage-resistant strains that fail to grow in a minimal medium have also been en- countered. Anderson (1) interpreted his findings as indicat- ing the existence of common steps in the chains of reactions leading to the synthesis of an essential metabolite and of substances needed for phage sensitivity, the mutation blocking one of the en- zymatic reactions. As an extension of this inter- pretation, it was suggested that multiple resistance to various phages acquired in one mutational step results from the blocking of a reaction common to the syntheses of substances needed for growth or adsorption of the phages. Common steps could oc- cur in the chains of reactions leading to sensitivity to unrelated phages. Different mutations producing resistance to more phages would probably produce blocks at different levels of the series of reactions. Since bacteria resistant to a certain phage are often sensitive to some mutants from that phage, the "blocks" in the chains of synthetic reactions are likely to alter, rather than suppress completely, the synthesis of substances necessary for phage sensitivity. 134 S. E. LURIA It was thought (17) that the mutational changes might simply involve slight alterations in the con- figuration of the bacterial receptors for phage ad- sorption, which were supposed to react with the phage by simple coming together of complementary surface structures. Mutational alteration of the surface receptors could be brought about either by a primary change in the receptor itself or by an alteration in the structure of a gene-like center act- ing as template for the receptor, without involve- ment of any long reaction chain. The association of specific metabolic deficiencies with phage resist- ance seems to contradict this idea. Moreover, the fmding of T. F. Anderson (3) that certain amino phage-resistant mutant of the usual type. Unfor- tunately, the relation between the two types of resistance was not thoroughly analyzed. The in- dications were, however, that phage resistance in the acridme-resistant strain was caused by a change in an acridine-sensitive mechanism involved in phage growth. Interestingly enough, both antiphage action of the acridines and phage resistance of the acridine-resistant strain could be antagonized by the addition of ribose nucleic acid. The idea that the usual type of phage resistance is due to a block in a chain of synthetic reactions would be convincingly proved if it were possible to restore sensitivity to a resistant mutant by supply- TABLE 3. ONE-STEP MUTANTS Phenotype Wild type Frequent mutants a b Ii ; Complex mutants f f Fl ;i;s T2 T6 T4 * T3 T7 T7h T7h' rrypto- phane lequire ment - + - - + i kneration Time, Minutes, in Broth at 37" C. 19-20 19-20 19-20 19-20 25-26 25-26 29-30 29-30 28-29 28-29 40 * The braces indicate serological relationship between phages. acids may act as necessary cofactors for phage adsorption indicates that the latter process is likely to involve some rather complex enzymatic activity We found recently that some of the B/3, 4, 7 of the phage particle on the bacterial surface. mutants, besides not adsorbing phages T3, T4, and T7, also show a partial inability to grow phages T2 and T6. They adsorb these phages like normal wild-type bacteria, but 80-90s of the infected bacteria fail to liberate any phage at all. Similar observations were also made by Hershey (personal communication). This phenomenon, which needs further investigation, indicates a relation between adsorption capacity and ability to grow phages, and supports the idea that both phenomena in- volve chains of reactions variously linked. Additional evidence for relating phage resistance to specific changes in enzymatic reactions is offered by recent results of Fitzgerald and his collaborators (22). It was found that some acridines exerted a specific inhibition on phage growth in de bacterial cell, pnd that a variant resistant to one of the acrldmes was almost as resistant to phage as a ing its medium with the intermediate metabolite whose synthesis is supposedly blocked, and whose Some additional information on multiple resist- absence is assumed to be responsible for the phage ance has been derived from a category of rare mu- tants with unusual combinations of resistance to resistance. Such cases have not yet been found. various phages (11). We have recently isolated a number of these "complex" mutants, and attempted a further characterization of their phenotypes, com- paring them with those of the more common my- tams. This was done by taking into account sensl- tivity to phages Tl-T7 and to some of their host- range mutants (active on some of the resistant bacterial mutants), nutritional requirements, and growth rates, which will be considered later. The complex mutants were isolated by plating large numbers of sensitive bacteria in the presence of suitable mixtures of phages. Most of them occur with mutation rates of the order of lo-lo. The re- sults of some of these tests are summarized in Table 3. We see that certain basic groupings of characters BACTERIAL MUTATIONS 135 awear in the phenotypes of both simple and com- plex mutants: resistance to phages Tl and TS; to ohages T3, T4, and T7; resistance to phage Tl and tryptophane requirement. All of the complex mu- tants incapable of growing in minimal medium were found to have the same tryptophane require- ment as the B/l mutants, although some of them mro\v more poorly in the presence of the same imount of the amino acid. The coupling of resistance to phage T2 and to @age T6, as present in the complex mutants, is not shown by any of our simple mutants. This seems due to the fact that all the frequent mutants isolated from our B strain in the presence of phage T2 prove sensitive to this phage after growth in phage-free media, owing to some as yet unclear mechanism of FIG. 1. Chains of synthetic reactions needed to account for the coupling of resistance in the mutants of Table 3. The symbol X in each case indicates the block corresponding to the mutant phenotype indicated by the adjacent letter. The broken lines indicate the coupling of resistance to T3, T4, T7 with failure to grow T2 and Td. partial reversion to sensitivity. Thus most of the B/Z, 6 strains show an apparent B/6 phentoype (phenotype c, Table 3). The groups of characters shown by the "simple" mutants appear in the complex mutants, associated in a number of ways but not in all possible ways. For example, resistance to phages T3, T4, T7 was found associated either with resistance to Tl and TS (phenotype h) or with resistance to Tl and tryptophane requirement (phenotype g), or with resistance to T2 and T6 (phenotype i). Resistance to phages Tl and TS was not found associated with Qptophane requirement. Some of the complex phenotypes are clearly a superposition of two of the simpler ones: phenotype g is like phenotypes a + f; phenotype h, like b + f. Other complex mutants differ from the sum of two simple ones in some character, particularly in sensitivity to various mutants from T7. It may be possible to account for these various combinations of characters, according to Ander- son's scheme, on the basis of blocks or changes in common steps of different chains of synthetic re- actions, assuming that each phenotype corresponds to a different synthetic block; the situation, as shown in the diagram (Fig. l), becomes quite com- plicated. One must assume that there are a number of different reactions common to the chains leading to sensitivity to certain groups of phages, and that some of these reactions are shared by different chains of reactions leading to sensitivity to other phages. Further complications are revealed by dif- ferences in sensitivity to various mutants of the T7h group, indicating either a greater multiplicity of pathways, or different alterations of the same reaction. Each of the numerous reactions assumed in order to explain multiple resistance to such a small number of phages must be capable of being altered by a nonlethal mutation. We shouuld also assume that the rare frequency of occurrence of all the more complex mutations is purely coincidental. One way out of some of these difhculties would be to suppose that some or all of the very complex mutations are actually combinations of two or more simple mutations occurring simultaneously. Such simultaneous occurrence might be brought about by some type of mass rearrangement of the genetic material of the bacterial cell, possibly comparable to that responsible for chromosomal rearrangements in higher organisms. Altogether, we feel that the problem of multiple effects of a bacterial mutation cannot yet be in- terpreted satisfactorily by any single hypothesis. Recognition of a larger number of complex mutants, their characterization by as many traits as possible, study of the physiology of the changes involved, and work on reverse mutations should help clarify the still obscure situation. Cases of simultaneous acquisition of several growth requirements may offer good material for this study. QU.~NTITATIVE RESISTANCE In studying the acquisition of resistance to anti- biotics (penicillin, 9) and drugs (sulfathiazole, 21), it has been found that resistance to increasing con- centrations of an antibacterial agent is built up stepwise by accumulation of successive mutations, each contributing further resistance, until strains are obtained that for all practical purposes can be considered as completely resistant. Mutants isolated by the selective action of a certain concentration of an agent are often resistant to much higher con- centrations. Supposed correspondence of the re- sistance level with the concentration used in the "training" to an antibacterial agent has been con- sidered (13) as one proof of the active induction of resistance. The actual lack of such correspondence S. E. LURIA is corroborating evidence of its spontaneous, muta- tional origin. The successive mutations to resistance might represent either similar changes in a series of gene-lihe units, or successive changes in the same unit, or independent genetic changes each affecting one of the metabolic processes on which the anti- bacterial agent may act. A study of sulfonamide resistance in Staphylo; coccus (21) indicates that the last alternative is probably correct. A number of different mutations are involved in acquisition of thii resistance, most of the recognized ones occurring at comparable rates (between 1CF' and N-lo). These mutations csn be differentiated first of all by the degree of resistance produced, expressed by the drug con- centration withstood by the mutant. Even more important is the fact that different mutants with increased sulfonamide resistance differ in other cor- related characters, particularly in the production of an extraceMar sulfonamide antagonist appar- ently identical with paminobenzoic acid. Some of the resistant mutants release during growth up to 30 times more p.a.b. than the parent strain, while others do not produce, any increased amount of it. The same mutation affecting p.a.b. production may occur at the same rate either in the parent strain or in mutants with already increased resistance but without increased p.a.b. production. The simplest explanation is that sulfonamides act as inhibitors of a number of metabolic reactions (whether all involving p.a.b. or not is immaterial for our problem). Successive mutations to resistance produce changes in different enzymatic mechanisms, the residual level of sensitivity being determined by the most sensitive mechanism left. In the case of penicillin resistance (9)) the oc- currence of mutations producing different degrees of resistance indicates a similar situation. That penicillin affects a number of different enzyme sys- tems of the bacterial cell is indirectly indicated by data of Treffers (23) on growth inhibition brought about by associations of penicillm with each of a number of metabolic poisons in noninhibitory EUIlOUIItS. The acquisition of resistance to drugs and anti- biotics by mutation in bacteria seems, therefore, to fall in line with the idea of genetic determination of single enzyme reactions by individual genes. It is interesting to note that, in the case of resistance to radiation in E. co23 B, Within (24) found only one mutational step increasing resistance. All at- tempts to select mutants having higher resistance failed. It is likely that only one of the synthetic mechanisms responsible for sensitivity to radiation can be modified by a nonlethal genetic change. The increased production of p.a.b. by sulfona- mide-resistant mutants of Staphylococcus exempli- fies a metabolic change increasing the extracellular production of a metabolite. P.a.b. is not a required metabolite, and is normally produced by the organ- km. It is possible, on the one hand, that p.ab. ac- cumulates because of a block in a sulfonamide-sensi- tive reaction that utiliies p.a.b. Provided an altema- tive metabolic pathway for growth were available, such a block would be expected to produce sulfona- mide resistance. It is possible, on the other hand, that an actual increase in p.a.b. synthesis is caused by the mutation. That positive biochemical mutations may be responsible for resistance to antibacterial drugs seems indicated by the increased formation of pantothenate by C. diptctheriae in the course of its becoming pantoyltaurine-resistant and independent of pantothenate for growth (20). GROWTH OF T= RESISTANT MUTANTS The resistant mutants often show definite dif- ferences in growth characteristics from the sensitive wild-type parents, which can be used.ss additional criteria for identification of a given mutational step. In the case of phage-resistant mutants, some mu- tations are correlated with definite changes in growth rate (generation time). Some of the data for the growth of one-step mutants in nutrient broth at 37O C. are included in Table 3. The growth rates of the mutants range from the same value as for wild-type to values more than twice as great. These differences again indicate that the mutations produce profound changes in the metabolic pro- cesses of the bacterial cell. It was found that when successive mutations are accumulated in the same bacterium the growth rate is always determined by the mutation that by itself gives the slowest growth. We can see in Table 3 that those complex mu- tants that appear to be the superposition of two simple mutant types also have the same generation time as the slower simple mutant type included. This agrees with the suggestion that they actually originate by combined mutations. Others of the most complex mutants have a much slower division time, probably indicating a more profound altera- tion of their synthetic abilities. The effects on growth of mutations to resistance are not liited to changes of generation time in the logarithmic phase. All phases of growth may be affected Among mutants resistant to phages or to penicillin, we found some having longer or shorter lag phases, others with higher or lower maximum titers than the wild-type strain. Of a large number of mutants whose growth was measured under comparable conditions, a majority grew less well than their wild-type parents. Only s few seemed to grow as well or better. This seems to agree with the general principle that most muts- tions arising in well-established wild-types produce less favorable phenotypes. Complications may arise, however, when one at- tempts to predict the growth of parent and mutant strains in mixtwes on the basis of their growth in isolated cultures. Production of diffusible growth BACTERIAL MUTATIOiVS 137 $-&itOrS 01 StimUhtS, Or diffeRIltid Kh!S Of utilixstion of some nutrilite, may greatly alter the picture. For instance, we found that the viable counts of a penicillm-resistant mutant in mixed cuiture with the parent strain began to fall long before the end of the growth phase for separate cultures of either strain. Other instances of interac- tions of this type have also been encountered. In different phases of the life of a bacterial culture either the parent type or its mutants may be favored, and only detailed study of individual cases can clarify the most complex situations. The indica- tions are, however, that many cases of bacterial variation, including some of the most complex types of so-called life cycles, can be explained in terms of the simplest hypothesis-mutation followed by selection within the resulting mixtures of pheno- types (6). SUMMARY Bacterial mutations to resistance have provided favorable material for the quantitative study of bacterial mutability. The mutations discussed in this paper present suggestive analogies with gene mutations in higher organisms, although the. similar- ity of the genetic systems involved can only be considered, for the time being, as a useful working hypothesis. The problems of determination of mu- tation rates, of identification of single mutation steps in strains with different genotypes, and of multiple effects of mutations have been discussed. 1. 2. 3. 4. 5. 6, 7. REFERENCES ANDERSON, E. H. Incidence of metabolic changes among virus resistant mutants of a bacterial strain. Proc. Nat. Acad. Sci. 30: 397-403. 1944. ANDERSON, E. H. Growth requirements of virus resistant mutants of Escherichiu coli strain B. Proc. Nat. Acad. Sci. 32: 120-128. 1946. ANDERSON, T. F. The role of tryptophane in the adsorp- tion of two bacterial viruses on their host, Esckerichiu coli. J. Ceil. Comp. Physiol. 25: 14-26. 1945. AVERY, 0. T., MACLEOD, C. M. and MCCARTY, M. Studies on the chemical nature of the substance inducing transformation of pneumococcai types. Induction of transformation by a desoxyribonudeic acid fraction iso- lated from Pneumococcus Type HI. J. Exp. Med. 79: 137-1.58. 1944. BOMN, A., DELAUNAY, A., VENDRELY, R., and LEHOULT, Y. L'acide thymonu&ique polymer%, principe parais- sant susceptible de d6terminer la speciiicit6 skologiqua et l'equipment enxymatique des bact6ries. Signification pour la biochimie de i'h6r6diti. Experientia 1: 334-335. 194s. BRAUN, WERNZR. Dissociation in B*uce&z abortw: a demonstration of the role of inherent and environmental factors in bacterial variation. J. Bact. 51: 327-349. 1946. BURNET, F. M. Smooth-rough variation in bacteria in its relation to bacteriophage. J. Path. Bact. 32: 15-42. 1929. 8. DFJ,BR%X, M. Spontaneous mutations of bacteria. Ann. Missouri Bot. Garden 32: 223-233. 194% 9. Dxacxaxc, M. Production of Staphylococcus strains 10. 11. 12. 13. 14. 1.5. 16. 17. 18. 19. 20. 21. 22. 23. 24. resistant to various concentrations of peniciiiin. Proc. Nat. Acad. Sci. 31: 16-24. 194s. DEMEREC, M. Induced mutations and possible mecha- nisms of the transmission of heredity in Eschet-ichM coli. Proc. Nat. Acad. Sci. 32: 36-46. 1946. D-EC, M., and FANO, U. Bacteriophageksistant mutants in Ercherichia coli. Genetics 30: 119-136. 1945. FITZGERALD, R. J., and LEE, M. E. Studies on bacterial viruses. II. Observations on the mode of action of acridines in inhibiting lysis of virus-infected bacteria. J. Immunol. 52: 127-135. 1946. HINSEIELWOOD, C. N. Bacterial growth. Biol. Rev. 19: 150-163. 1944. KNIGHT, B. C. J. G. Bacterial nutrition. Material for a comparative physiology of bacteria. Spec. Rep. Ser. Med. Res. Coun. No. 210. 1936. KRISTENSEN, M. Recherches sur la fermentation muta- tive des bactkes. Acta path. microbial. stand. 17: 193- 231. 1940. LEWIS, I. M. Bacterial variation with special reference to behavior of some mutable strains of colon bacteria in synthetic medium. J. Bact. 28: 619-639. 1934. LURIA, S. E. Genetics of bacterium-bacterial virus reia- tionship. Ann. Missouri Bot. Garden 32: 235-242. 194.5. Luam, S. E., and DELBR~~CK, M. Mutations of bacteria from virus sensitivity to virus resistance. Genetics 28: 491-511. 1943. LWOFF, A., and AUDUKZAU, A. Sur une mutation de Morazella kuofi apte a se dCvelopper dams les milieux a I'acide succinique. Ann. Inst. Past. 67: 94-111. 1941. MCILWAIN, d. Nutritional studies of bacterial varia- tion. IL The derivation of drug resistant strains in the absence of any inhibitor. Brit. J. Exp. Path. 24: 212-217. 1943. OAKBERC, E. F., and LURIA, S. E. Mutations to sulfona- mide resistance in Staphylococcus uureus. Genetics, in Press. PARR, L. W. A new `mutation" in the coiiform group of bacteria. J. Hered. 29: 380384. 1938. Txx~xxas, H. P. The potentiation of penicillin or streptomycin action by certain enzyme inhibitors. J. Bact. 52: 502-.X& 1946. WITKIN, E. M. Inherited differences in sensitivity to radiation in Eschetichia coli. Proc. Nat. Acad. Sci. 32: 59-68. 1946. DISCUSSION KIDD: I wonder if Dr. Luria has seen a resem- blance between the complex races of phage-resistant bacteria with which he has been working and the strains of drug-fast trypanosomes which have inter- ested parasitologists since the time of Ehrlich. While the genesis of the drug-fast strains of trypanosomes has not perhaps been particularly well studied as judged by the criteria of critical quantitative genetics, the fact has long been known that they represent more or less stable heritable variations in microor- ganisms. Furthermore, the work of Warrington Yorke and his collaborators and that of Jan& and others has made it abundantly plain that the lethal effects of drugs on trypanosomes depend upon their fina- tion by the parasites: arsenical and acridine com- pounds, for example, are absorbed by strains of normal trypanosomes, which then succumb to their 138 S. E. LURIA effects; but the compounds are not absorbed by the . drug-fast strains, which in this way resist their ac- tion. From a study of the phenomena of drug re- sistance, moreover, Frank Hawking has distin- guished four different kinds of receptors in tryp- anosomes-those for arsenicals and acriflavine, for parafuchsin, for diamidine compounds, and for Bayer ZOS-and he has studied the relationships among them. It seems to me that the parasitologists have managed to acquire a noteworthy comprehension of the phenomena with which they were confronted- the practical implications for chemotherapy in- cluded-and it may become manifest that the prin- ciples involved are much the same as those with which Dr. Luria is dealing. SONNEBORN: The finding of a single mutant colony when an entire culture is plated out need not .necessarily indicate that the mutation was brought to phenotypic expression at once in the absence of further cell reproduction. Suppose the mutation had occurred several cell generations earlier, with a lag of some cell generations between mutation and phenotypic expression and with varia- tion among the progeny of the mutated cell in the number of cell generations preceding phenotypic ex- pression. In Paramecium, for example, precisely such variation is observed. Under such conditions a single descendant cell might resist phage and the others be destroyed. It would then be diicult to distinguish thii from cases of immediate phenotypic expression.