Nobel Lecture in Physiology or Medicine 1946: The Production of Mutations Muller by Hermann J. Muller Lyrics
The Production of MutationsNobel Lecture, delivered by Hermann J. Muller on December 12, 1946
If as Darwin maintained the adaptiveness of living things results from natural selection, rather than from a teleological tendency in the process of variation itself, then heritable variations must, under most conditions, occur in numerous directions, so as to give a wide range of choice for the selective process. Such a state of affairs seems, however, in more or less contradiction to the commonly held idea, to which Darwin also gave some credence, that heritable variations of given kinds tend to be produced, in a fairly regular way, by given kinds of external conditions. For then we are again confronted with the difficulty, how is it that the "right kinds" of variations (i.e. the adaptive ones) manage to arise in response to the "right kinds" of conditions (i.e. those they are adapted to)? Moreover, the de Vriesian notion of mutations does not help us in this connection. On that view, there are sudden jumps, going all the way from one "elementary species" to another, and involving radical changes in numerous characters at once, and there are relatively few different jumps to choose between. This obviously would fail to explain how, through such coarse steps, the body could have come to be so remarkably streamlined in its internal and external organization, or, in other words, so thoroughly adaptive.
The older selectionists, thinking in terms of chemical reactions on a molar scale when they thought in terms of chemistry at all, did not realize sufficiently the ultramicroscopic randomness of the processes causing inherited variations. The earliest mutationists failed, in addition, to appreciate the qualitative and quantitative multiplicity of mutations. It was not long, however, before the results of Baur on Antirrhinum and of Morgan on Drosophila, supplemented by scattered observations on other forms, gave evidence of the occurrence of numerous Mendelizing mutations, many of them small ones, in varied directions, and they showed no discoverable relation between the type of mutation and the type of environment or condition of living under which it arose. These observations, then, came closer to the statistical requirements for a process of evolution which has its basis in accidents. In what sense, however, could the events be regarded as accidental? Were they perhaps expressions of veiled forces working in a more regular manner? It was more than ever evident that further investigation of the manner of occurrence of mutations was called for.
If the mutations were really non-teleological, with no relation between type of environment and type of change, and above all no adaptive relation, and if they were of as numerous types as the theory of natural selection would demand, then the great majority of the changes should be harmful in their effects, just as any alterations made blindly in a complicated apparatus are usually detrimental to its proper functioning, and many of the larger changes should even be totally incompatible with the functioning of the whole, or, as we say, lethal. That is, strange as it may seem at first sight, we should expect most mutations to be disadvantageous if the theory of natural selection is correct. We should also expect these mainly disadvantageous changes to be highly diversified in their genetic basis.
To get exact evidence on these points required the elaboration of special genetic methods, adapted to the recognition of mutations that ordinarily escape detection: (1) lethals, (2) changes with but small visible effects, and (3) changes without any externally visible effects but influencing the viability more or less unfavorably. It would take us too far afield to explain these techniques here. Suffice it to say that they made use of the principle according to which a chromosome is, as we say, "marked", by having had inserted into it to begin with one or more known mutant genes with conspicuous visible effects, to differentiate it from the homologous chromosome. An individual with two such differentiated chromosomes, when appropriately bred, will then be expected to give two groups of visibly different offspring, holding certain expected ratios to one another. If, however, a lethal mutation has occurred in one of the two chromosomes, its existence will be made evident by the absence of the corresponding expected group of offspring. Similarly, a mutated gene with invisible but somewhat detrimental action, though not fully lethal, will be recognized by the fact that the corresponding group of offspring are found in smaller numbers than expected. And a gene with a very small visible effect, that might be overlooked in a single individual, will have a greatly increased chance of being seen because the given group of offspring as a whole will tend to be distinguished in this regard from the corresponding group derived from a non-mutant.
In this way, it was possible in the first tests of this kind, which Altenburg and the writer conducted, partly in collaboration, in 1918-19, to get definite evidence in Drosophila that the lethal mutations greatly outnumbered those with visible effects, and that among the latter the types having an obscure manifestation were more numerous than the definite conspicuous ones used in ordinary genetic work. Visible or not, the great majority had lowered viability. Tests of their genetic basis, using the newly found facts of linkage, showed them to be most varied in their locus in the chromosomes, and it could be calculated by a simple extrapolative process that there must be at least hundreds, and probably thousands, of different kinds arising in the course of spontaneous mutation. In work done much later, employing induced mutations, it was also shown (in independent experiments both of the present writer and Kerkis, and of Timotéeff and his co-workers, done in 1934) that "invisible" mutations which by reason of one or another physiological change lower viability without being fully lethal form the most abundant group of any detected thus far, being at least 2 to 3 times as numerous as the complete lethals. No doubt there are in addition very many, perhaps even more, with effects too small to have been detected at all by our rather crude methods. It is among these that we should be most apt to find those rare accidents which, under given conditions or in given combinations with others, may happen to have some adaptive value. Tests of Timotéeff, however, have shown that even a few of the more conspicuous visible mutations do in certain combinations give an advantage in laboratory breeding.
Because of the nature of the test whereby it is detected - the absence of an entire group of offspring bearing certain conspicuous expected characters - a lethal is surer of being detected, and detected by any observer, than is the inconspicuous or invisible, merely detrimental, mutation. Fortunately, there are relatively few borderline cases, of nearly but not quite completely lethal genes. It was this objectivity of recognition, combined with the fact that they were so much more numerous than conspicuous visible mutations, that made it feasible for lethals to be used as an index of mutation frequency, even though they suffer from the disadvantage of requiring the breeding of an individual, rather than its mere inspection, for the recognition that it carries a lethal. In the earliest published work, we (Altenburg and the author) attempted not only to find a quantitative value for the "normal" mutation frequency, but also to determine whether a certain condition, which we considered of special interest, affected the mutation frequency. The plan was ultimately to use the method as a general one for studying the effects of various conditions. The condition chosen for the first experiment was temperature, and the results, verified by later work of the writer's, indicated that a rise of temperature, within limits normal to the organism, produced an increase of mutation frequency of about the amount to be expected if mutations were, in essentials, orthodox chemical reactions.
On this view, however, single mutations correspond with individual molecular changes, and an extended series of mutations, in a great number of identical genes in a population, spread out over thousands of years, is what corresponds with the course of an ordinary chemical reaction that takes place in a whole collection of molecules in a test tube in the course of a fraction of a second or a few seconds. For the individual gene, in its biological setting, is far more stable than the ordinary chemical molecule is, when the latter is exposed to a reagent in the laboratory. Thus, mutations, when taken collectively, should be subject to the statistical laws applying to mass reactions, but the individual mutation, corresponding to a change in one molecule, should be subject to the vicissitudes of ultramicroscopic or atomic events, and the apparition of a mutant individual represents an enormous amplification of such a phenomenon. This is a principle which gives the clue to the fact, which otherwise seems opposed to a rational, scientific and molarly deterministic point of view, that differences in external conditions or conditions of living do not appear to affect the occurrence of mutations, while on the other hand, even in a normal and sensibly constant environment, mutations of varied kinds do occur. It is also in harmony with our finding, of about the same time, that when a mutation takes place in a given gene, the other gene of identical type present nearby in the same cell usually remains unaffected, though it must of course have been subjected to the same macroscopic physico-chemical conditions. On this conception, then, the mutations ordinarily result from submicroscopic accidents, that is, from caprices of thermal agitation, that occur on a molecular and submolecular scale. More recently Delbrück and Timoféeff, in more extended work on temperature, have shown that the amount of increase in mutation frequency with rising temperature is not merely that of an ordinary test-tube chemical reaction, but in fact corresponds closely with that larger rise to be expected of a reaction as slow in absolute time rate (i.e. with as small a proportion of molecular changes per unit of time) as the observed mutation frequency shows this reaction to be, and this quantitative correspondence helps to confirm the entire conception.
Now this inference concerning the non-molar nature of the individual mutation process, which sets it in so different a class from most other grossly observable chemical changes in nature, led naturally to the expectation that some of the "point effects" brought about by high-energy radiation like X-rays would also work to produce alterations in the hereditary material. For if even the relatively mild events of thermal agitation can, some of them, have such consequences, surely the energetically far more potent point changes caused by powerful radiation should succeed. And, as a matter of fact, our trials of X-rays, carried out with the same kind of genetic methods as previously used for temperature, proved that such radiation is extremely effective, and inordinately more so than a mere temperature rise, since by this method it was possible to obtain, by a half-hour's treatment, over a hundred times as many mutations in a group of treated germ cells as would have occurred in them spontaneously in the course of a whole generation. These mutations too were found ordinarily to occur pointwise and randomly, in one gene at a time, without affecting an identical gene that might be present nearby in a homologous chromosome.
In addition to the individual gene changes, radiation also produced rearrangements of parts of chromosomes. As our later work (including that with co-workers, especially Raychaudhuri and Pontecorvo) has shown, these latter were caused in the first place by breakages of the chromosomes, followed afterwards by attachments occurring between the adhesive broken ends, that joined them in a different order than before. The two or more breaks involved in such a rearrangement may be far apart, caused by independent hits, and thus result in what we call a gross structural change. Such changes are of various kinds, depending upon just where the breaks are and just which broken ends become attached to which. But, though the effects of the individual "hits" are rather narrowly localized, it is not uncommon for two breaks to be produced at nearby points by what amounts to one local change (or at any rate one localized group of changes) whose influence becomes somewhat spread out. By the rejoining, in a new order, of broken ends resulting from two such nearby breaks, a minute change of sequence of the genes is brought about. More usually, the small piece between the two breaks becomes lost (a "deficiency"), but sometimes it becomes inverted, or even becomes transferred into a totally different position, made available by a separate hit.
Both earlier and later work by collaborators (Oliver, Hanson, etc.) showed definitely that the frequency of the gene mutations is directly and simply proportional to the dose of irradiation applied, and this despite the wave-length used, whether X- or gamma- or even beta-rays, and despite the timing of the irradiation. These facts have since been established with great exactitude and detail, more especially by Timoféeff and his co-workers. In our more recent work with Raychaudhuri (1939, 1940) these principles have been extended to total doses as low as 400 r, and rates as low as 0.01 r per minute, with gamma rays. They leave, we believe, no escape from the conclusion that there is no threshold dose, and that the individual mutations result from individual "hits", producing genetic effects in their immediate neighborhood. Whether these so-called "hits" are the individual ionizations, or may even be the activations that occur at lower energy levels, or whether, at the other end of the scale, they require the clustering of ionizations that occurs at the termini of electron tracks and of their side branches (as Lea and Fano point out might be the case), is as yet undecided. But in any case they are, even when microscopically considered, what we have termed "point mutations", as they involve only disturbances on an ultramicroscopically localized scale. And whether or not they are to occur at any particular point is entirely a matter of accident, using this term in the sense in which it is employed in the mathematics of statistics.
Naturally, other agents than photons which produce effects of this kind must also produce mutations, as has been shown by students and collaborators working under Altenburg in Houston for alpha rays (Ward, 1935) and for neutrons (Nagai and Lecher, 1937), and extended in regard to the quantitative relations concerned by Zimmer and others working with Timoféeff (1936, 1937, 1938), and by others. Moreover, as Altenburg (1930, 1935) showed, even the smaller quantum changes induced by ultraviolet exert this effect on the genes. They cause, however, only a relatively small amount of rearrangement of chromosome parts (Muller and Mackenzie, 1939) and, in fact, they also tend to inhibit such rearrangement, as Swanson (1944), followed by Kaufmann and Hollaender (1944 et seq.), has found. Since the effective ultraviolet hits are in the form of randomly scattered single-atom changes in the purines and pyrimidines of the chromosomes, rather than in groups of atom changes, it seems likely that clusters of ionizations are not necessary for the gene mutation effects, at any rate, although we cannot be sure of this until the relation of mutation frequency to dosage is better known for this agent.
Inasmuch as the changes brought about in the genes by radiation must certainly be of an accidental nature, unpremeditated, ateleological, without reference to the value of the end result for the organism or its descendants, it is of interest to compare them with the so-called spontaneous or natural mutations. For in the radiation mutations we have a yardstick of what really random changes should be. Now it is found in Drosophila that the radiation - induced mutations of the genes (we exclude here the demonstrable chromosome rearrangements) are in every respect which has been investigated of the same essential nature as those arising naturally in the laboratory or field. They usually occur in one gene without affecting an identical one nearby. They are distributed similarly in the chromosomes. The effects, similarly, may be large or small, and there is a similar ratio of fully lethal to so-called visible gene mutations. That is, the radiation mutations of the genes do not give evidence of being more deleterious. And when one concentrates attention upon given genes one finds that a whole series of different forms, or alleles, may be produced, of a similar and in many cases sensibly identical nature in the two cases. In fact, every natural mutation, when searched for long enough, is found to be producible also by radiation. Moreover, under any given condition of living tried, without radiation, the effects appear as scattered as when radiation is applied, even though of much lower frequency. All this surely means then, does it not, that the natural mutations have in truth no innate tendency to be adaptive, nor even to be different, as a whole group, under some natural conditions than under others? In other words, they cannot be determinate in a molar sense, but must themselves be caused by the ultramicroscopic accidents of the molecular and submolecular motions, i.e. by the individual quantum exchanges of thermal agitation, taking this word in a broad sense. The only escape from this would be to suppose that they are caused by the radiation present in nature, resulting from natural radioactive substances and cosmic rays, but a little calculation (by Mott-Smith and the writer, 1930, corroborated by others) has shown that this radiation is quite inadequate in amount to account for the majority of mutations occurring in most organisms.
But to say that most natural mutations are the results of the quantum exchanges of thermal agitation, and, further, that a given energy level must be reached to produce them, does not, as some authors have seemed to imply, mean that the physico-chemical conditions in and around the organism, other than temperature, have no influence upon their chance of occurrence. That such circumstances may play a decided role was early evident from the studies of spontaneous mutation frequency, when it was found (1921, reported 1928) that the frequency in one experiment, with one genetic stock, might be ten times as high as in another, with another stock. And more recently (1945) we have found that, in different portions of the natural life cycle of the same individual, the mutation frequency may be very different. Finally, in the work of Auerbach and Robson (1941-46), with mustard gas and related substances, it has been proved that these chemicals may induce mutations at as high a frequency as a heavy dose of X-rays. In all these cases, however, the effects are similarly scattered at random, individually uncontrolled, and similarly non-adaptive.
It should also be noted in this connection that the genes are not under all conditions equally vulnerable to the mutating effects of X-rays themselves. Genes in the condensed chromosomes of spermatozoa, for example, appear to be changed more easily than those in the more usual "resting" stages. We have mentioned that, as Swanson has shown, ultraviolet exerts besides its own mutating effect an inhibition on the process of chromosome breakage, or at any rate on that of reunion of the broken parts in a new viable order, while infrared, in Hollaender's and Kaufmann's recent experiments, has a contrary action. And Stadler, in his great work on the production of mutations in cereals, started independently of our own, has obtained evidence that in this material X-radiation in the doses used is unable to produce a sensible rise in the gene mutation frequency, though numerous chromosome breakages do arise, leading to both gross and minute rearrangements of chromosome parts. Either the genes are more resistant in this material to permanent changes by X-rays, as compared with their responsiveness to thermal agitation, or a break or loss must usually be produced by X-rays along with the gene change. The milder ultraviolet quanta, on the other hand, do produce gene mutations like the natural ones in these plants.
Such variations in effectiveness are, I believe, to have been expected. They do not shake our conclusion as to the accidental, quantum character of the event which usually initiates a gene mutation. But they give rise to the hope that, through further study of them, more may be learned concerning the nature of the mutation process, as well as of the genetic material that undergoes the changes.
No one can answer the question whether some special means may not be found whereby, through the application of molar influences, such as specific antibodies, individual genes could be changed to order. Certainly the search for such influences, and for increasing control of things on a microscopic and submicroscopic scale as well, must be carried further. But there is as yet no good evidence that anything of the sort has been done artificially, or that it occurs naturally. Even if possible, there could be no generalized method of control of gene composition without far greater knowledge than we now have of the intimate chemical structure and the mode of working of the most complicated and diverse substances that exist, namely, nucleoproteins, proteins in general, and enzymes. The works of Sumner, Northrop, and Stanley, together with those of other protein chemists, point the way in this direction, but everyone will agree that it is a long and devious system of roads which is beginning here.
It is true that some cases are known of mutable genes which change selectively in response to special conditions. Such cases may be very informative in shedding light on gene structure, but we have as yet no indication that the alterations of these genes, which in the great majority of instances known are abnormal genes, have anything in common with ordinary natural mutations. It is also true that cases are known among bacteria and viruses of the induction of particular kinds of hereditary changes by application of particular substances, but here the substances applied are in each case the same as those whose presence is later found to have been induced, and so there is every reason to infer that they have in fact become implanted in some way, that is, that we do not really have a specifically induced mutation.
So far, then, we have no means, or prospect of means, of inducing given mutations at will in normal material, though the production of mutations in abundance at random may be regarded as a first step along such a path, if there is to be such a path. So long as we cannot direct mutations, then, selection is indispensable, and progress in the hereditary constitution of a living thing can be made only with the aid of a most thoroughgoing selection of the mutations that occur since, being non-adaptive except by accident, an overwhelming majority is always harmful. For a sensible advance, usually a considerable number of rare steps must be accumulated in the painful selective process. By far the most of these are individually small steps, but, as species and race crossings have shown, there may be a few large distinctive steps that have been, as Huxley terms it, "buffered", by small changes that readjust the organism to them. Not only is this accumulation of many rare, mainly tiny changes the chief means of artificial animal and plant improvement, but it is, even more, the way in which natural evolution has occurred, under the guidance of natural selection. Thus the Darwinian theory becomes implemented, and freed from the accretions of directed variation and of Lamarckism that once encumbered it.
It is probable that, in a state of nature, most species have a not very much (though somewhat) lower frequency of gene mutation than would be most advantageous for them, in consideration of the degree of rigor of the natural selection that occurs in the given species. A much higher frequency would probably lead to faster genetic degenerative processes than the existing selection could well cope with. But, under conditions of artificial breeding, where selection can be made more effective, a higher mutation frequency can for a time at least be tolerated in some cases, and larger mutations also can be nursed through to the point where they become suitably buffered. Here it may become of practical use to apply X-rays, ultraviolet, or other means of inducing mutations, as Gustafsson especially has demonstrated for X-rays. This will be especially true in species which naturally undergo much inbreeding, or in which there is a well-expressed haploid phase, or a considerable haploid portion of the genotype, for under these circumstances many of the spontaneous mutations that might otherwise have accumulated in the population and that could be brought to light by inbreeding, will have become eliminated before being found, and the natural mutation rate itself will be lower.
We have above largely confined ourselves to considering the relation of the production of gene mutations to the problems of the general method of evolution, including that of the nature of hereditary variation, because this has been, historically, the main line of approach to the subject of artificial mutations. It was from the first evident, however, that the production of mutations would, as we once stated, provide us with tools of the greatest nicety, where-with to dissect piece by piece the physiological, embryological, and bio-chemical structure of the organism and to analyze its workings. Already with natural mutations, such works as those of Bonnevie, Grueneberg, Scott-Moncrief, Ephrussi, and Beadle, etc., have shown how the intensive tracing of the effects, and interrelations of effects, of just one or a few mutations, can lead to a deeper understanding of the complex processes whereby the genes operate to produce the organism. But there are thousands of genes, and it is desirable to be able to choose them for study in an orderly fashion as we proceed with our dissection process. For this purpose we have thought that it would often be advantageous to produce mutations artificially in abundance, so as then to take our pick of those more suited for successive steps in our analysis. The work of Beadle and his co-workers on Neurospora in recent years, followed by similar work of Malin and Fries and of others, has brilliantly shown the applicability of this method for studies of the paths of bio-chemical synthesis of amino acids, vitamins, purines, and pyrimidines. And yet, in a sense, the surface of the subject as a whole has barely been scratched, and we may look forward with confidence to the combination of this technique with that of tracer substances and with all the other techniques of bio-chemistry, physiology and experimental embryology, for the increasing unravelling of that surpassingly intricate tangle of processes of which the living organism is constituted. There is not time, however, to go further into this subject here.
For we cannot neglect here a brief outline of another phase of the artificial mutation work, more specifically of interest to geneticists: that is, the further analysis of the properties of the chromosomes and their parts, gained chiefly from studies in which parts have been removed, added, or rearranged. We have already spoken, in passing, of the studies of the mechanism of such structural change, in which a relatively simple general scheme lying at the basis of all such alterations has emerged: namely, breakage first, followed by adhesion of broken ends. It was early evident that by the use of such rearranged chromosomes additional proof of the physical validity of the linkage maps could be obtained, and this was done (Muller and Painter, 1929 et seq.). Furthermore, it has been possible to throw light on problems of crossing-over, as in the demonstration (Muller, Stone, and Offermann, 1930 et seq.) that to whatever position the centromere is moved, it causes a strong inhibition of crossing-over, the strength of which gradually diminishes with distance. Moreover, the same proves to be true of any point of discontinuity in pairing, caused by heterozygosity in regard to a structural change. Such studies on crossing-over, and on the pairing forces that affect segregation, are still capable of considerable extension.
We must remember, in speaking of the centromere and other apparently distinctive chromosome parts, that we have no right to infer them to be autonomous, locally determined structures, dependent on the genes of the regions in which they are seen to lie, before observations have been made that show the effects of removing or displacing those regions. Therefore, it has in the main been necessary to wait for the study of induced inversions, deletions and translocations of chromosomes, before the inference could be secure that the centromere is, in most instances, such an autonomous organelle, dependent upon a gene or genes in its immediate neighborhood (but not in all instances in its neighborhood, as Rhoades has recently shown in a special strain of maize). Similarly, it has been possible to show (despite some contrary claims, the validity or invalidity of which cannot be discussed here) that the free end of the chromosome, or telomere, constitutes in much material a locally determined distinctive structure.
By a combined genetic and cytological analysis of various cases of break-age and rearrangement of parts, in work done in collaboration with Painter, Stone, Prokofyeva, Gershenson, and others, it was found that there are distinctive, largely locally determined regions of the chromosomes, usually most markedly developed near the centromeres, which we at first called "inactive" but which are now usually referred to as "heterochromatic". These were also found independently in purely cytological studies by Heitz. It would be fascinating to enter here into a discussion of the remarkable peculiarities which the cytogenetic studies have shown these regions to have the evidence of repetition of more or less similar parts, of a tendency to conjugation between the differently placed parts, of distinctive cytological appearance correlated with whether or not such conjugation occurs, of inordinately high tendency to structural change, of strong influence of certain of their genes upon segregation, etc. - and then to go on to discuss hypotheses of their evolutionary origin and their functions. This would unfortunately take us too far afield. We must however insist upon one point - as it is not yet generally enough recognized - namely, that the evidence is very strong that what in the Drosophila chromosome, as seen at mitosis, is called "the heterochromatic region", is simply a large temporary body of accessory, non-genie nucleoprotein, produced under the influence of one or two particular genes from among the dozen or more that constitute the whole heterochromatic region, as detected by genetic analysis and by the chromosome as seen at the resting stage (as in the salivary gland). And it is not these conspicuous non-genie blocks which are responsible for the other known peculiarities of the heterochromatin, above mentioned - the function of the blocks is still undetermined. In other words, the so-called "heterochromatin" with which the cytologist deals in studying mitotic chromosomes is a quite different thing from, although in the neighborhood of, the heterochromatin proper having the above described complex of properties. More-over, it has been possible to show (Sutton-Gersh in collaboration with the author, unpublished) that the conspicuous nucleoli often associated with the heterochromatin are produced under the influence of still other autonomous genes in it, that are separate from those for the mitotically visible blocks.
One of the most interesting findings which has come out of the study of Drosophila chromosomes that underwent rearrangement of parts as a result of irradiation has been the generalization of the existence of the phenomenon known as "position effect". This effect was first found by Sturtevant (1925) in the case of the spontaneous mutant known as Bar eye, but it was not known to what extent the effect might be a special one until numerous rearrangements could be studied. The term position effect implies that the functioning of a gene is to a certain extent at least dependent upon what other genes lie in its neighborhood. There is now adequate evidence that this is a general principle applying to very many if not all of the genes in Drosophila, and that their functioning can be qualitatively as well as quantitatively conditioned by the character of the genes in their vicinity, some of the genes having much more effect than others and different genes working in different ways and to different extents.
It is possible that, as Sturtevant (ibid.) suggested, the position effect is caused by the interaction between gene products in the vicinity of the genes producing them, assuming that such products are more concentrated there and under such circumstances tend to react more with one another than when dispersed. However, the interpretation which we favor is that the functioning of the gene is affected by its shape and that this, in turn, varies with the strength and nature of synaptic forces acting on the region of the chromosome in which it lies. These might consist of forces directly exerted on the gene by other genes, whether allelic or not (Muller, 1935), or they might be resultants of the state of spiralization, etc., of the chromosome region, circumstances which in their turn are in part dependent on synaptic forces (Ephrussi and Sutton, 1944 et seq.). This interpretation, in either of its variants, would explain why position effects are so much more general in Drosophila, an organism in which the synaptic forces are known to operate strongly even in somatic cells, than in other organisms tested, in which such forces are much weaker or absent in somatic cells. It would also fit in with the findings (Muller, 1935) that the heterochromatic regions tend to have especially strong, extensive, and distinctive kinds of position effects, effects varying in degree with the total amount of heterochromatin present in a cell, as well as with vacillating embryological factors. For these genetic findings are in conformity with the cytological effects of heterochromatin, observed first by Prokofyeva, on the degree of extension, synaptic properties, etc., of euchromatin in its neighborhood, effects which she showed to be subject to similar vacillations, that are correlated with the variations in the phrenotypically observed position effects. Recent observations, both by Ephrussi and Sutton (ibid.), undertaken jointly with the author, and by Stern (1944, 1946), also seem to point in this direction, for they show an influence, on the position effects exhibited by given parts, of the arrangement of homologous chromosome parts. If this interpretation based on gene shape should hold, it would open up a new angle of attack on the structure and method of functioning of the gene, perhaps ultimately relating it to nucleoprotein composition and properties.
Another use to which the process of breakage and rearrangement of chromosome parts by irradiation has been put is for the study of the effects of adding and of subtracting small pieces of chromosomes, in order to determine the relation of gene dosage to gene expression. In this way, it has been found out (1) that most normal genes are, even in single dose, near the upper limit of their effectiveness, and (2) that most mutant genes have a final effect qualitatively similar to but quantitatively less than that of their allelic normal genes. The dominance of normal genes over their mutant alleles, then, turns out in most instances to be a special case of the principle that one dose of a normal gene usually produces nearly though not quite as much effect as two doses. This in turn is best understood as resulting from a long course of selection of the normal gene and its modifiers for stability of expression, when under the influence of environmental and genetic conditions which would affect the gene's operation quantitatively, i.e. in a manner similar to that of dosage changes. This does not mean that selection has specifically worked to produce dominance of the normal gene over its alleles, however, because (3) not all mutant genes behave merely like weaker normal genes, and (4) those which the dosage tests show to produce qualitatively different effects from the normal genes seem oftener to escape from the principle of being dominated over by the normals, just as would be expected on our hypothesis.
Among the further results of gene dosage studies carried out by the use of chromosome fragments produced by irradiation, attention should be especially called to the findings coming under the head of "dosage compensation". These have shown: (1) that, when the dosage of virtually all genes in the X-chromosome except a given one is held constant, the expression of that one is usually so very nearly the same when present in one dose as in two that no difference in the character can ordinarily be seen, and (2) that nevertheless this invisible difference has been so important for the organism that, in the course of the past natural selection, a system of modifying genes, called compensators, has been established, having the function of making the effects of the one and two doses normally present in the two respective sexes much more nearly equal still, when these dosage differences in the given genes are present simultaneously with those in all the other X-chromosomal genes. Each gene seems to have acquired a different system of compensators, the interrelations of all together being extremely complicated. This then gives evidence from a new angle of the meticulousness of natural selection, of the very precise adaptiveness of the characters existing in a species, and of the final grade of a character having ordinarily become established through the accumulation of numerous small mutations having very complex functional relations with one another. It is in line with our previous thesis of evolution through the selection of multitudinous tiny accidental changes.
When attention is concentrated on a given very circumscribed region of a chromosome, by a comparison of various induced rearrangements all of which have a point of breakage within that region, other facts come to light, bearing on the problems of chromosome and gene divisibility. By means of special genetic methods, which cannot be detailed here, evidence has been obtained that the breaks in any such limited region tend to occur at specific points, giving indication that discrete units or segments lie between these points, and thus arguing against the idea of the chromosome being a continuum and in favor of its genes corresponding to physical entities rather than merely to concepts arbitrarily set up for the convenience of geneticists. We are also enabled in this way to make estimates of the probable number of genes in the chromosome, as well as to get maximally limiting figures for their size. These estimates agree as closely as could have been expected with those based on previous genetic work, using entirely different methods, although not with the estimates based on the "sensitive volume" hypothesis.
Another finding made in studies of cases having a small fragment of chromosome moved, as a result of irradiation, to another position, was that individuals are frequently able to survive and reproduce even when they have the given chromosome part present in its original position as well as in the new position. In fact, it was in work of this kind that the effect of extra doses of genes was determined. Now, in some of these cases stocks could even be obtained which were homozygous for the duplicated piece as well as for the original piece. This led to the idea that duplications of chromosome material might in this manner have become established in the previous course of evolution. When in the analysis of a limited region of the X-chromosome, including the locus of the so-called "scute" effect, it was found that there are in fact, within the normal X-chromosome, two genes of closely related effect ("achaete" and "scute") very close or adjacent to one another, it became evident that this was in all probability an example of the above postulated occurrence. This then showed the way, and apparently the main if not the only way (aside from the far rarer phenomena of polyploidy and "tetrasomy"), by which the number of genes has become increased during the course of evolution. By a curious coincidence, Bridges was at the same time making his studies of salivary chromosomes and finding direct cytological evidence for the existence of such "repeats", as he called them, in the normal chromosomes, and he interpreted these in the same manner. In the twelve years since that time, various other clear cases of the same kind have been demonstrated. Thus, increase in gene number, brought about by the duplication of small parts of chromosomes, more usually in positions near their original ones, must be set down as one of the major processes in evolution, in addition to the mutations in the individual genes. By itself, this process would not be of great importance, but it becomes important because, by allowing gene mutations to come afterwards that differentiate the genes in one position from the originally identical ones in the other position, the number of different kinds of genes is increased and so the germ plasm, and with it the processes of development and the organism as a whole, are eventually enabled to grow more complex.
Rearrangements of chromosome parts which do not lead to an increase in gene number can of course also occur in evolution, although it is unlikely that their role is so fundamental. By producing such changes in the laboratory it has been possible to find out a good deal more about what types can arise, and what their properties are. Various inferences can then be drawn concerning the viability and fertility that the different types would have, under varied genetic circumstances, and whether they would tend to become eliminated or to accumulate in a population of a given type. Some of them can be shown to have, under given conditions, an evolutionary survival value, both by aiding in the process of genetic isolation and in other ways, as by affecting heterosis. In this manner, evolutionary inferences have been drawn which have later been confirmed by comparison of the chromosome differences actually existing between related races, subspecies, and species.
Probably of greater ultimate interest will be the results of studies of gene mutations occurring at individual loci. Radiation mutations are frequent enough to lend themselves to comparisons of the potentialities of different loci, although not nearly enough has yet been done along these lines. Similarly, a comparison of the different mutations which can occur at the same locus can lead to very important results, especially since it has been shown that the different alleles may have very complex relationships to one another, so as even, in some cases, to reconstitute the normal type when they are crossed together. The way in which genes change as a result of successive mutations remains to be gone into at much greater length. So too does the question of changes in gene mutability, brought about by gene mutation itself.
The further the analysis of the genetic effects of irradiation, particularly of the breakage and rearrangement of chromosome parts, has gone, the more does our conviction grow that a large proportion if not the great majority of the somatic effects of irradiation that have been observed by medical men and by students of embryology, regeneration, and general biology, arise secondarily as consequences of genetic effects produced in the somatic cells. The usefulness of this interpretation has been shown in recent studies of Koller, dealing with improved methods of irradiation of mammalian carcinoma. This is too large a subject to digress upon here, but it is to be noted that it has been the analyses based in the first place on genetic and cytogenetic studies of the reproductive cells, as shown by subsequent generations, which are thus helping to clear the way for an understanding of the mechanism by which radiation acts in inhibiting growth, in causing sterilization, in producing necrosis and burns, in causing recession of malignant tissue, and perhaps also, on occasion at least, in inducing the initiation of such tissue.
During the war years, a curious confirmation of the correctness of the above inference regarding the nature of the somatic effects of irradiation has come to light. While working with mustard gas in Edinburgh, J. M. Robson was struck with the remarkable similarity between the somatic effects of this agent and those produced by X-ray and radium irradiation. This led him to wonder whether perhaps mustard gas might produce genetic changes of essentially the same kind as those known to be brought about by radiation. Comprehensive experiments were thereupon undertaken by C. Auerbach, working in collaboration with Robson, and (as mentioned on p. 160) she succeeded in showing that in fact this substance does produce mutations, both in the individual genes and by breakage and rearrangement of chromosome parts, such as X-rays and radium do, and in similar abundance. Other substances of the same general group were then found to have a similar effect. This constitutes the first decided break in the chemical attack on mutation. The fact that these findings were made as a direct result of the above inference, when so many previous attempts to produce mutations by chemical means had failed, appears to provide strong evidence that these peculiar somatic effects are in truth consequences of the more underlying ones which, when occurring in the germ cells, are analyzed by the geneticist in his breeding tests. There are, however, some very interesting differences between the nature of the genetic effects of irradiation and of these chemicals, which we cannot go into here, but which give promise of allowing an extension of the genetic and somatic analyses.
We see then that production of mutations by radiation is a method, capable of being turned in various directions, both for the analysis of the germ plasm itself, and of the organism which is in a sense an outgrowth of that germ plasm. It is to be hoped that it may also, in certain fields, prove of increasing practical use in plant and animal improvement, in the service of man. So far as direct practical application in man himself is concerned, however, we are as yet a long way from practicing any intentional selection over our own germ plasm, although like most species we are already encumbered by countless undesirable mutations, from which no individual is immune. In this situation we can, however, draw the practical lesson, from the fact of the great majority of mutations being undesirable, that their further random production in ourselves should so far as possible be rigorously avoided. As we can infer with certainty from experiments on lower organisms that all high-energy radiation must produce such mutations in man, it becomes an obligation for radiologists - though one far too little observed as yet in most countries - to insist that the simple precautions are taken which are necessary for shielding the gonads, whenever people are exposed to such radiation, either in industry or in medical practice. And, with the coming increasing use of atomic energy, even for peace-time purposes, the problem will become very important of insuring that the human germ plasm - the all-important material of which we are the temporary custodians - is effectively protected from this additional and potent source of permanent contamination.
From Nobel Lectures, Physiology or Medicine 1942-1962, Elsevier Publishing Company, Amsterdam, 1964
If as Darwin maintained the adaptiveness of living things results from natural selection, rather than from a teleological tendency in the process of variation itself, then heritable variations must, under most conditions, occur in numerous directions, so as to give a wide range of choice for the selective process. Such a state of affairs seems, however, in more or less contradiction to the commonly held idea, to which Darwin also gave some credence, that heritable variations of given kinds tend to be produced, in a fairly regular way, by given kinds of external conditions. For then we are again confronted with the difficulty, how is it that the "right kinds" of variations (i.e. the adaptive ones) manage to arise in response to the "right kinds" of conditions (i.e. those they are adapted to)? Moreover, the de Vriesian notion of mutations does not help us in this connection. On that view, there are sudden jumps, going all the way from one "elementary species" to another, and involving radical changes in numerous characters at once, and there are relatively few different jumps to choose between. This obviously would fail to explain how, through such coarse steps, the body could have come to be so remarkably streamlined in its internal and external organization, or, in other words, so thoroughly adaptive.
The older selectionists, thinking in terms of chemical reactions on a molar scale when they thought in terms of chemistry at all, did not realize sufficiently the ultramicroscopic randomness of the processes causing inherited variations. The earliest mutationists failed, in addition, to appreciate the qualitative and quantitative multiplicity of mutations. It was not long, however, before the results of Baur on Antirrhinum and of Morgan on Drosophila, supplemented by scattered observations on other forms, gave evidence of the occurrence of numerous Mendelizing mutations, many of them small ones, in varied directions, and they showed no discoverable relation between the type of mutation and the type of environment or condition of living under which it arose. These observations, then, came closer to the statistical requirements for a process of evolution which has its basis in accidents. In what sense, however, could the events be regarded as accidental? Were they perhaps expressions of veiled forces working in a more regular manner? It was more than ever evident that further investigation of the manner of occurrence of mutations was called for.
If the mutations were really non-teleological, with no relation between type of environment and type of change, and above all no adaptive relation, and if they were of as numerous types as the theory of natural selection would demand, then the great majority of the changes should be harmful in their effects, just as any alterations made blindly in a complicated apparatus are usually detrimental to its proper functioning, and many of the larger changes should even be totally incompatible with the functioning of the whole, or, as we say, lethal. That is, strange as it may seem at first sight, we should expect most mutations to be disadvantageous if the theory of natural selection is correct. We should also expect these mainly disadvantageous changes to be highly diversified in their genetic basis.
To get exact evidence on these points required the elaboration of special genetic methods, adapted to the recognition of mutations that ordinarily escape detection: (1) lethals, (2) changes with but small visible effects, and (3) changes without any externally visible effects but influencing the viability more or less unfavorably. It would take us too far afield to explain these techniques here. Suffice it to say that they made use of the principle according to which a chromosome is, as we say, "marked", by having had inserted into it to begin with one or more known mutant genes with conspicuous visible effects, to differentiate it from the homologous chromosome. An individual with two such differentiated chromosomes, when appropriately bred, will then be expected to give two groups of visibly different offspring, holding certain expected ratios to one another. If, however, a lethal mutation has occurred in one of the two chromosomes, its existence will be made evident by the absence of the corresponding expected group of offspring. Similarly, a mutated gene with invisible but somewhat detrimental action, though not fully lethal, will be recognized by the fact that the corresponding group of offspring are found in smaller numbers than expected. And a gene with a very small visible effect, that might be overlooked in a single individual, will have a greatly increased chance of being seen because the given group of offspring as a whole will tend to be distinguished in this regard from the corresponding group derived from a non-mutant.
In this way, it was possible in the first tests of this kind, which Altenburg and the writer conducted, partly in collaboration, in 1918-19, to get definite evidence in Drosophila that the lethal mutations greatly outnumbered those with visible effects, and that among the latter the types having an obscure manifestation were more numerous than the definite conspicuous ones used in ordinary genetic work. Visible or not, the great majority had lowered viability. Tests of their genetic basis, using the newly found facts of linkage, showed them to be most varied in their locus in the chromosomes, and it could be calculated by a simple extrapolative process that there must be at least hundreds, and probably thousands, of different kinds arising in the course of spontaneous mutation. In work done much later, employing induced mutations, it was also shown (in independent experiments both of the present writer and Kerkis, and of Timotéeff and his co-workers, done in 1934) that "invisible" mutations which by reason of one or another physiological change lower viability without being fully lethal form the most abundant group of any detected thus far, being at least 2 to 3 times as numerous as the complete lethals. No doubt there are in addition very many, perhaps even more, with effects too small to have been detected at all by our rather crude methods. It is among these that we should be most apt to find those rare accidents which, under given conditions or in given combinations with others, may happen to have some adaptive value. Tests of Timotéeff, however, have shown that even a few of the more conspicuous visible mutations do in certain combinations give an advantage in laboratory breeding.
Because of the nature of the test whereby it is detected - the absence of an entire group of offspring bearing certain conspicuous expected characters - a lethal is surer of being detected, and detected by any observer, than is the inconspicuous or invisible, merely detrimental, mutation. Fortunately, there are relatively few borderline cases, of nearly but not quite completely lethal genes. It was this objectivity of recognition, combined with the fact that they were so much more numerous than conspicuous visible mutations, that made it feasible for lethals to be used as an index of mutation frequency, even though they suffer from the disadvantage of requiring the breeding of an individual, rather than its mere inspection, for the recognition that it carries a lethal. In the earliest published work, we (Altenburg and the author) attempted not only to find a quantitative value for the "normal" mutation frequency, but also to determine whether a certain condition, which we considered of special interest, affected the mutation frequency. The plan was ultimately to use the method as a general one for studying the effects of various conditions. The condition chosen for the first experiment was temperature, and the results, verified by later work of the writer's, indicated that a rise of temperature, within limits normal to the organism, produced an increase of mutation frequency of about the amount to be expected if mutations were, in essentials, orthodox chemical reactions.
On this view, however, single mutations correspond with individual molecular changes, and an extended series of mutations, in a great number of identical genes in a population, spread out over thousands of years, is what corresponds with the course of an ordinary chemical reaction that takes place in a whole collection of molecules in a test tube in the course of a fraction of a second or a few seconds. For the individual gene, in its biological setting, is far more stable than the ordinary chemical molecule is, when the latter is exposed to a reagent in the laboratory. Thus, mutations, when taken collectively, should be subject to the statistical laws applying to mass reactions, but the individual mutation, corresponding to a change in one molecule, should be subject to the vicissitudes of ultramicroscopic or atomic events, and the apparition of a mutant individual represents an enormous amplification of such a phenomenon. This is a principle which gives the clue to the fact, which otherwise seems opposed to a rational, scientific and molarly deterministic point of view, that differences in external conditions or conditions of living do not appear to affect the occurrence of mutations, while on the other hand, even in a normal and sensibly constant environment, mutations of varied kinds do occur. It is also in harmony with our finding, of about the same time, that when a mutation takes place in a given gene, the other gene of identical type present nearby in the same cell usually remains unaffected, though it must of course have been subjected to the same macroscopic physico-chemical conditions. On this conception, then, the mutations ordinarily result from submicroscopic accidents, that is, from caprices of thermal agitation, that occur on a molecular and submolecular scale. More recently Delbrück and Timoféeff, in more extended work on temperature, have shown that the amount of increase in mutation frequency with rising temperature is not merely that of an ordinary test-tube chemical reaction, but in fact corresponds closely with that larger rise to be expected of a reaction as slow in absolute time rate (i.e. with as small a proportion of molecular changes per unit of time) as the observed mutation frequency shows this reaction to be, and this quantitative correspondence helps to confirm the entire conception.
Now this inference concerning the non-molar nature of the individual mutation process, which sets it in so different a class from most other grossly observable chemical changes in nature, led naturally to the expectation that some of the "point effects" brought about by high-energy radiation like X-rays would also work to produce alterations in the hereditary material. For if even the relatively mild events of thermal agitation can, some of them, have such consequences, surely the energetically far more potent point changes caused by powerful radiation should succeed. And, as a matter of fact, our trials of X-rays, carried out with the same kind of genetic methods as previously used for temperature, proved that such radiation is extremely effective, and inordinately more so than a mere temperature rise, since by this method it was possible to obtain, by a half-hour's treatment, over a hundred times as many mutations in a group of treated germ cells as would have occurred in them spontaneously in the course of a whole generation. These mutations too were found ordinarily to occur pointwise and randomly, in one gene at a time, without affecting an identical gene that might be present nearby in a homologous chromosome.
In addition to the individual gene changes, radiation also produced rearrangements of parts of chromosomes. As our later work (including that with co-workers, especially Raychaudhuri and Pontecorvo) has shown, these latter were caused in the first place by breakages of the chromosomes, followed afterwards by attachments occurring between the adhesive broken ends, that joined them in a different order than before. The two or more breaks involved in such a rearrangement may be far apart, caused by independent hits, and thus result in what we call a gross structural change. Such changes are of various kinds, depending upon just where the breaks are and just which broken ends become attached to which. But, though the effects of the individual "hits" are rather narrowly localized, it is not uncommon for two breaks to be produced at nearby points by what amounts to one local change (or at any rate one localized group of changes) whose influence becomes somewhat spread out. By the rejoining, in a new order, of broken ends resulting from two such nearby breaks, a minute change of sequence of the genes is brought about. More usually, the small piece between the two breaks becomes lost (a "deficiency"), but sometimes it becomes inverted, or even becomes transferred into a totally different position, made available by a separate hit.
Both earlier and later work by collaborators (Oliver, Hanson, etc.) showed definitely that the frequency of the gene mutations is directly and simply proportional to the dose of irradiation applied, and this despite the wave-length used, whether X- or gamma- or even beta-rays, and despite the timing of the irradiation. These facts have since been established with great exactitude and detail, more especially by Timoféeff and his co-workers. In our more recent work with Raychaudhuri (1939, 1940) these principles have been extended to total doses as low as 400 r, and rates as low as 0.01 r per minute, with gamma rays. They leave, we believe, no escape from the conclusion that there is no threshold dose, and that the individual mutations result from individual "hits", producing genetic effects in their immediate neighborhood. Whether these so-called "hits" are the individual ionizations, or may even be the activations that occur at lower energy levels, or whether, at the other end of the scale, they require the clustering of ionizations that occurs at the termini of electron tracks and of their side branches (as Lea and Fano point out might be the case), is as yet undecided. But in any case they are, even when microscopically considered, what we have termed "point mutations", as they involve only disturbances on an ultramicroscopically localized scale. And whether or not they are to occur at any particular point is entirely a matter of accident, using this term in the sense in which it is employed in the mathematics of statistics.
Naturally, other agents than photons which produce effects of this kind must also produce mutations, as has been shown by students and collaborators working under Altenburg in Houston for alpha rays (Ward, 1935) and for neutrons (Nagai and Lecher, 1937), and extended in regard to the quantitative relations concerned by Zimmer and others working with Timoféeff (1936, 1937, 1938), and by others. Moreover, as Altenburg (1930, 1935) showed, even the smaller quantum changes induced by ultraviolet exert this effect on the genes. They cause, however, only a relatively small amount of rearrangement of chromosome parts (Muller and Mackenzie, 1939) and, in fact, they also tend to inhibit such rearrangement, as Swanson (1944), followed by Kaufmann and Hollaender (1944 et seq.), has found. Since the effective ultraviolet hits are in the form of randomly scattered single-atom changes in the purines and pyrimidines of the chromosomes, rather than in groups of atom changes, it seems likely that clusters of ionizations are not necessary for the gene mutation effects, at any rate, although we cannot be sure of this until the relation of mutation frequency to dosage is better known for this agent.
Inasmuch as the changes brought about in the genes by radiation must certainly be of an accidental nature, unpremeditated, ateleological, without reference to the value of the end result for the organism or its descendants, it is of interest to compare them with the so-called spontaneous or natural mutations. For in the radiation mutations we have a yardstick of what really random changes should be. Now it is found in Drosophila that the radiation - induced mutations of the genes (we exclude here the demonstrable chromosome rearrangements) are in every respect which has been investigated of the same essential nature as those arising naturally in the laboratory or field. They usually occur in one gene without affecting an identical one nearby. They are distributed similarly in the chromosomes. The effects, similarly, may be large or small, and there is a similar ratio of fully lethal to so-called visible gene mutations. That is, the radiation mutations of the genes do not give evidence of being more deleterious. And when one concentrates attention upon given genes one finds that a whole series of different forms, or alleles, may be produced, of a similar and in many cases sensibly identical nature in the two cases. In fact, every natural mutation, when searched for long enough, is found to be producible also by radiation. Moreover, under any given condition of living tried, without radiation, the effects appear as scattered as when radiation is applied, even though of much lower frequency. All this surely means then, does it not, that the natural mutations have in truth no innate tendency to be adaptive, nor even to be different, as a whole group, under some natural conditions than under others? In other words, they cannot be determinate in a molar sense, but must themselves be caused by the ultramicroscopic accidents of the molecular and submolecular motions, i.e. by the individual quantum exchanges of thermal agitation, taking this word in a broad sense. The only escape from this would be to suppose that they are caused by the radiation present in nature, resulting from natural radioactive substances and cosmic rays, but a little calculation (by Mott-Smith and the writer, 1930, corroborated by others) has shown that this radiation is quite inadequate in amount to account for the majority of mutations occurring in most organisms.
But to say that most natural mutations are the results of the quantum exchanges of thermal agitation, and, further, that a given energy level must be reached to produce them, does not, as some authors have seemed to imply, mean that the physico-chemical conditions in and around the organism, other than temperature, have no influence upon their chance of occurrence. That such circumstances may play a decided role was early evident from the studies of spontaneous mutation frequency, when it was found (1921, reported 1928) that the frequency in one experiment, with one genetic stock, might be ten times as high as in another, with another stock. And more recently (1945) we have found that, in different portions of the natural life cycle of the same individual, the mutation frequency may be very different. Finally, in the work of Auerbach and Robson (1941-46), with mustard gas and related substances, it has been proved that these chemicals may induce mutations at as high a frequency as a heavy dose of X-rays. In all these cases, however, the effects are similarly scattered at random, individually uncontrolled, and similarly non-adaptive.
It should also be noted in this connection that the genes are not under all conditions equally vulnerable to the mutating effects of X-rays themselves. Genes in the condensed chromosomes of spermatozoa, for example, appear to be changed more easily than those in the more usual "resting" stages. We have mentioned that, as Swanson has shown, ultraviolet exerts besides its own mutating effect an inhibition on the process of chromosome breakage, or at any rate on that of reunion of the broken parts in a new viable order, while infrared, in Hollaender's and Kaufmann's recent experiments, has a contrary action. And Stadler, in his great work on the production of mutations in cereals, started independently of our own, has obtained evidence that in this material X-radiation in the doses used is unable to produce a sensible rise in the gene mutation frequency, though numerous chromosome breakages do arise, leading to both gross and minute rearrangements of chromosome parts. Either the genes are more resistant in this material to permanent changes by X-rays, as compared with their responsiveness to thermal agitation, or a break or loss must usually be produced by X-rays along with the gene change. The milder ultraviolet quanta, on the other hand, do produce gene mutations like the natural ones in these plants.
Such variations in effectiveness are, I believe, to have been expected. They do not shake our conclusion as to the accidental, quantum character of the event which usually initiates a gene mutation. But they give rise to the hope that, through further study of them, more may be learned concerning the nature of the mutation process, as well as of the genetic material that undergoes the changes.
No one can answer the question whether some special means may not be found whereby, through the application of molar influences, such as specific antibodies, individual genes could be changed to order. Certainly the search for such influences, and for increasing control of things on a microscopic and submicroscopic scale as well, must be carried further. But there is as yet no good evidence that anything of the sort has been done artificially, or that it occurs naturally. Even if possible, there could be no generalized method of control of gene composition without far greater knowledge than we now have of the intimate chemical structure and the mode of working of the most complicated and diverse substances that exist, namely, nucleoproteins, proteins in general, and enzymes. The works of Sumner, Northrop, and Stanley, together with those of other protein chemists, point the way in this direction, but everyone will agree that it is a long and devious system of roads which is beginning here.
It is true that some cases are known of mutable genes which change selectively in response to special conditions. Such cases may be very informative in shedding light on gene structure, but we have as yet no indication that the alterations of these genes, which in the great majority of instances known are abnormal genes, have anything in common with ordinary natural mutations. It is also true that cases are known among bacteria and viruses of the induction of particular kinds of hereditary changes by application of particular substances, but here the substances applied are in each case the same as those whose presence is later found to have been induced, and so there is every reason to infer that they have in fact become implanted in some way, that is, that we do not really have a specifically induced mutation.
So far, then, we have no means, or prospect of means, of inducing given mutations at will in normal material, though the production of mutations in abundance at random may be regarded as a first step along such a path, if there is to be such a path. So long as we cannot direct mutations, then, selection is indispensable, and progress in the hereditary constitution of a living thing can be made only with the aid of a most thoroughgoing selection of the mutations that occur since, being non-adaptive except by accident, an overwhelming majority is always harmful. For a sensible advance, usually a considerable number of rare steps must be accumulated in the painful selective process. By far the most of these are individually small steps, but, as species and race crossings have shown, there may be a few large distinctive steps that have been, as Huxley terms it, "buffered", by small changes that readjust the organism to them. Not only is this accumulation of many rare, mainly tiny changes the chief means of artificial animal and plant improvement, but it is, even more, the way in which natural evolution has occurred, under the guidance of natural selection. Thus the Darwinian theory becomes implemented, and freed from the accretions of directed variation and of Lamarckism that once encumbered it.
It is probable that, in a state of nature, most species have a not very much (though somewhat) lower frequency of gene mutation than would be most advantageous for them, in consideration of the degree of rigor of the natural selection that occurs in the given species. A much higher frequency would probably lead to faster genetic degenerative processes than the existing selection could well cope with. But, under conditions of artificial breeding, where selection can be made more effective, a higher mutation frequency can for a time at least be tolerated in some cases, and larger mutations also can be nursed through to the point where they become suitably buffered. Here it may become of practical use to apply X-rays, ultraviolet, or other means of inducing mutations, as Gustafsson especially has demonstrated for X-rays. This will be especially true in species which naturally undergo much inbreeding, or in which there is a well-expressed haploid phase, or a considerable haploid portion of the genotype, for under these circumstances many of the spontaneous mutations that might otherwise have accumulated in the population and that could be brought to light by inbreeding, will have become eliminated before being found, and the natural mutation rate itself will be lower.
We have above largely confined ourselves to considering the relation of the production of gene mutations to the problems of the general method of evolution, including that of the nature of hereditary variation, because this has been, historically, the main line of approach to the subject of artificial mutations. It was from the first evident, however, that the production of mutations would, as we once stated, provide us with tools of the greatest nicety, where-with to dissect piece by piece the physiological, embryological, and bio-chemical structure of the organism and to analyze its workings. Already with natural mutations, such works as those of Bonnevie, Grueneberg, Scott-Moncrief, Ephrussi, and Beadle, etc., have shown how the intensive tracing of the effects, and interrelations of effects, of just one or a few mutations, can lead to a deeper understanding of the complex processes whereby the genes operate to produce the organism. But there are thousands of genes, and it is desirable to be able to choose them for study in an orderly fashion as we proceed with our dissection process. For this purpose we have thought that it would often be advantageous to produce mutations artificially in abundance, so as then to take our pick of those more suited for successive steps in our analysis. The work of Beadle and his co-workers on Neurospora in recent years, followed by similar work of Malin and Fries and of others, has brilliantly shown the applicability of this method for studies of the paths of bio-chemical synthesis of amino acids, vitamins, purines, and pyrimidines. And yet, in a sense, the surface of the subject as a whole has barely been scratched, and we may look forward with confidence to the combination of this technique with that of tracer substances and with all the other techniques of bio-chemistry, physiology and experimental embryology, for the increasing unravelling of that surpassingly intricate tangle of processes of which the living organism is constituted. There is not time, however, to go further into this subject here.
For we cannot neglect here a brief outline of another phase of the artificial mutation work, more specifically of interest to geneticists: that is, the further analysis of the properties of the chromosomes and their parts, gained chiefly from studies in which parts have been removed, added, or rearranged. We have already spoken, in passing, of the studies of the mechanism of such structural change, in which a relatively simple general scheme lying at the basis of all such alterations has emerged: namely, breakage first, followed by adhesion of broken ends. It was early evident that by the use of such rearranged chromosomes additional proof of the physical validity of the linkage maps could be obtained, and this was done (Muller and Painter, 1929 et seq.). Furthermore, it has been possible to throw light on problems of crossing-over, as in the demonstration (Muller, Stone, and Offermann, 1930 et seq.) that to whatever position the centromere is moved, it causes a strong inhibition of crossing-over, the strength of which gradually diminishes with distance. Moreover, the same proves to be true of any point of discontinuity in pairing, caused by heterozygosity in regard to a structural change. Such studies on crossing-over, and on the pairing forces that affect segregation, are still capable of considerable extension.
We must remember, in speaking of the centromere and other apparently distinctive chromosome parts, that we have no right to infer them to be autonomous, locally determined structures, dependent on the genes of the regions in which they are seen to lie, before observations have been made that show the effects of removing or displacing those regions. Therefore, it has in the main been necessary to wait for the study of induced inversions, deletions and translocations of chromosomes, before the inference could be secure that the centromere is, in most instances, such an autonomous organelle, dependent upon a gene or genes in its immediate neighborhood (but not in all instances in its neighborhood, as Rhoades has recently shown in a special strain of maize). Similarly, it has been possible to show (despite some contrary claims, the validity or invalidity of which cannot be discussed here) that the free end of the chromosome, or telomere, constitutes in much material a locally determined distinctive structure.
By a combined genetic and cytological analysis of various cases of break-age and rearrangement of parts, in work done in collaboration with Painter, Stone, Prokofyeva, Gershenson, and others, it was found that there are distinctive, largely locally determined regions of the chromosomes, usually most markedly developed near the centromeres, which we at first called "inactive" but which are now usually referred to as "heterochromatic". These were also found independently in purely cytological studies by Heitz. It would be fascinating to enter here into a discussion of the remarkable peculiarities which the cytogenetic studies have shown these regions to have the evidence of repetition of more or less similar parts, of a tendency to conjugation between the differently placed parts, of distinctive cytological appearance correlated with whether or not such conjugation occurs, of inordinately high tendency to structural change, of strong influence of certain of their genes upon segregation, etc. - and then to go on to discuss hypotheses of their evolutionary origin and their functions. This would unfortunately take us too far afield. We must however insist upon one point - as it is not yet generally enough recognized - namely, that the evidence is very strong that what in the Drosophila chromosome, as seen at mitosis, is called "the heterochromatic region", is simply a large temporary body of accessory, non-genie nucleoprotein, produced under the influence of one or two particular genes from among the dozen or more that constitute the whole heterochromatic region, as detected by genetic analysis and by the chromosome as seen at the resting stage (as in the salivary gland). And it is not these conspicuous non-genie blocks which are responsible for the other known peculiarities of the heterochromatin, above mentioned - the function of the blocks is still undetermined. In other words, the so-called "heterochromatin" with which the cytologist deals in studying mitotic chromosomes is a quite different thing from, although in the neighborhood of, the heterochromatin proper having the above described complex of properties. More-over, it has been possible to show (Sutton-Gersh in collaboration with the author, unpublished) that the conspicuous nucleoli often associated with the heterochromatin are produced under the influence of still other autonomous genes in it, that are separate from those for the mitotically visible blocks.
One of the most interesting findings which has come out of the study of Drosophila chromosomes that underwent rearrangement of parts as a result of irradiation has been the generalization of the existence of the phenomenon known as "position effect". This effect was first found by Sturtevant (1925) in the case of the spontaneous mutant known as Bar eye, but it was not known to what extent the effect might be a special one until numerous rearrangements could be studied. The term position effect implies that the functioning of a gene is to a certain extent at least dependent upon what other genes lie in its neighborhood. There is now adequate evidence that this is a general principle applying to very many if not all of the genes in Drosophila, and that their functioning can be qualitatively as well as quantitatively conditioned by the character of the genes in their vicinity, some of the genes having much more effect than others and different genes working in different ways and to different extents.
It is possible that, as Sturtevant (ibid.) suggested, the position effect is caused by the interaction between gene products in the vicinity of the genes producing them, assuming that such products are more concentrated there and under such circumstances tend to react more with one another than when dispersed. However, the interpretation which we favor is that the functioning of the gene is affected by its shape and that this, in turn, varies with the strength and nature of synaptic forces acting on the region of the chromosome in which it lies. These might consist of forces directly exerted on the gene by other genes, whether allelic or not (Muller, 1935), or they might be resultants of the state of spiralization, etc., of the chromosome region, circumstances which in their turn are in part dependent on synaptic forces (Ephrussi and Sutton, 1944 et seq.). This interpretation, in either of its variants, would explain why position effects are so much more general in Drosophila, an organism in which the synaptic forces are known to operate strongly even in somatic cells, than in other organisms tested, in which such forces are much weaker or absent in somatic cells. It would also fit in with the findings (Muller, 1935) that the heterochromatic regions tend to have especially strong, extensive, and distinctive kinds of position effects, effects varying in degree with the total amount of heterochromatin present in a cell, as well as with vacillating embryological factors. For these genetic findings are in conformity with the cytological effects of heterochromatin, observed first by Prokofyeva, on the degree of extension, synaptic properties, etc., of euchromatin in its neighborhood, effects which she showed to be subject to similar vacillations, that are correlated with the variations in the phrenotypically observed position effects. Recent observations, both by Ephrussi and Sutton (ibid.), undertaken jointly with the author, and by Stern (1944, 1946), also seem to point in this direction, for they show an influence, on the position effects exhibited by given parts, of the arrangement of homologous chromosome parts. If this interpretation based on gene shape should hold, it would open up a new angle of attack on the structure and method of functioning of the gene, perhaps ultimately relating it to nucleoprotein composition and properties.
Another use to which the process of breakage and rearrangement of chromosome parts by irradiation has been put is for the study of the effects of adding and of subtracting small pieces of chromosomes, in order to determine the relation of gene dosage to gene expression. In this way, it has been found out (1) that most normal genes are, even in single dose, near the upper limit of their effectiveness, and (2) that most mutant genes have a final effect qualitatively similar to but quantitatively less than that of their allelic normal genes. The dominance of normal genes over their mutant alleles, then, turns out in most instances to be a special case of the principle that one dose of a normal gene usually produces nearly though not quite as much effect as two doses. This in turn is best understood as resulting from a long course of selection of the normal gene and its modifiers for stability of expression, when under the influence of environmental and genetic conditions which would affect the gene's operation quantitatively, i.e. in a manner similar to that of dosage changes. This does not mean that selection has specifically worked to produce dominance of the normal gene over its alleles, however, because (3) not all mutant genes behave merely like weaker normal genes, and (4) those which the dosage tests show to produce qualitatively different effects from the normal genes seem oftener to escape from the principle of being dominated over by the normals, just as would be expected on our hypothesis.
Among the further results of gene dosage studies carried out by the use of chromosome fragments produced by irradiation, attention should be especially called to the findings coming under the head of "dosage compensation". These have shown: (1) that, when the dosage of virtually all genes in the X-chromosome except a given one is held constant, the expression of that one is usually so very nearly the same when present in one dose as in two that no difference in the character can ordinarily be seen, and (2) that nevertheless this invisible difference has been so important for the organism that, in the course of the past natural selection, a system of modifying genes, called compensators, has been established, having the function of making the effects of the one and two doses normally present in the two respective sexes much more nearly equal still, when these dosage differences in the given genes are present simultaneously with those in all the other X-chromosomal genes. Each gene seems to have acquired a different system of compensators, the interrelations of all together being extremely complicated. This then gives evidence from a new angle of the meticulousness of natural selection, of the very precise adaptiveness of the characters existing in a species, and of the final grade of a character having ordinarily become established through the accumulation of numerous small mutations having very complex functional relations with one another. It is in line with our previous thesis of evolution through the selection of multitudinous tiny accidental changes.
When attention is concentrated on a given very circumscribed region of a chromosome, by a comparison of various induced rearrangements all of which have a point of breakage within that region, other facts come to light, bearing on the problems of chromosome and gene divisibility. By means of special genetic methods, which cannot be detailed here, evidence has been obtained that the breaks in any such limited region tend to occur at specific points, giving indication that discrete units or segments lie between these points, and thus arguing against the idea of the chromosome being a continuum and in favor of its genes corresponding to physical entities rather than merely to concepts arbitrarily set up for the convenience of geneticists. We are also enabled in this way to make estimates of the probable number of genes in the chromosome, as well as to get maximally limiting figures for their size. These estimates agree as closely as could have been expected with those based on previous genetic work, using entirely different methods, although not with the estimates based on the "sensitive volume" hypothesis.
Another finding made in studies of cases having a small fragment of chromosome moved, as a result of irradiation, to another position, was that individuals are frequently able to survive and reproduce even when they have the given chromosome part present in its original position as well as in the new position. In fact, it was in work of this kind that the effect of extra doses of genes was determined. Now, in some of these cases stocks could even be obtained which were homozygous for the duplicated piece as well as for the original piece. This led to the idea that duplications of chromosome material might in this manner have become established in the previous course of evolution. When in the analysis of a limited region of the X-chromosome, including the locus of the so-called "scute" effect, it was found that there are in fact, within the normal X-chromosome, two genes of closely related effect ("achaete" and "scute") very close or adjacent to one another, it became evident that this was in all probability an example of the above postulated occurrence. This then showed the way, and apparently the main if not the only way (aside from the far rarer phenomena of polyploidy and "tetrasomy"), by which the number of genes has become increased during the course of evolution. By a curious coincidence, Bridges was at the same time making his studies of salivary chromosomes and finding direct cytological evidence for the existence of such "repeats", as he called them, in the normal chromosomes, and he interpreted these in the same manner. In the twelve years since that time, various other clear cases of the same kind have been demonstrated. Thus, increase in gene number, brought about by the duplication of small parts of chromosomes, more usually in positions near their original ones, must be set down as one of the major processes in evolution, in addition to the mutations in the individual genes. By itself, this process would not be of great importance, but it becomes important because, by allowing gene mutations to come afterwards that differentiate the genes in one position from the originally identical ones in the other position, the number of different kinds of genes is increased and so the germ plasm, and with it the processes of development and the organism as a whole, are eventually enabled to grow more complex.
Rearrangements of chromosome parts which do not lead to an increase in gene number can of course also occur in evolution, although it is unlikely that their role is so fundamental. By producing such changes in the laboratory it has been possible to find out a good deal more about what types can arise, and what their properties are. Various inferences can then be drawn concerning the viability and fertility that the different types would have, under varied genetic circumstances, and whether they would tend to become eliminated or to accumulate in a population of a given type. Some of them can be shown to have, under given conditions, an evolutionary survival value, both by aiding in the process of genetic isolation and in other ways, as by affecting heterosis. In this manner, evolutionary inferences have been drawn which have later been confirmed by comparison of the chromosome differences actually existing between related races, subspecies, and species.
Probably of greater ultimate interest will be the results of studies of gene mutations occurring at individual loci. Radiation mutations are frequent enough to lend themselves to comparisons of the potentialities of different loci, although not nearly enough has yet been done along these lines. Similarly, a comparison of the different mutations which can occur at the same locus can lead to very important results, especially since it has been shown that the different alleles may have very complex relationships to one another, so as even, in some cases, to reconstitute the normal type when they are crossed together. The way in which genes change as a result of successive mutations remains to be gone into at much greater length. So too does the question of changes in gene mutability, brought about by gene mutation itself.
The further the analysis of the genetic effects of irradiation, particularly of the breakage and rearrangement of chromosome parts, has gone, the more does our conviction grow that a large proportion if not the great majority of the somatic effects of irradiation that have been observed by medical men and by students of embryology, regeneration, and general biology, arise secondarily as consequences of genetic effects produced in the somatic cells. The usefulness of this interpretation has been shown in recent studies of Koller, dealing with improved methods of irradiation of mammalian carcinoma. This is too large a subject to digress upon here, but it is to be noted that it has been the analyses based in the first place on genetic and cytogenetic studies of the reproductive cells, as shown by subsequent generations, which are thus helping to clear the way for an understanding of the mechanism by which radiation acts in inhibiting growth, in causing sterilization, in producing necrosis and burns, in causing recession of malignant tissue, and perhaps also, on occasion at least, in inducing the initiation of such tissue.
During the war years, a curious confirmation of the correctness of the above inference regarding the nature of the somatic effects of irradiation has come to light. While working with mustard gas in Edinburgh, J. M. Robson was struck with the remarkable similarity between the somatic effects of this agent and those produced by X-ray and radium irradiation. This led him to wonder whether perhaps mustard gas might produce genetic changes of essentially the same kind as those known to be brought about by radiation. Comprehensive experiments were thereupon undertaken by C. Auerbach, working in collaboration with Robson, and (as mentioned on p. 160) she succeeded in showing that in fact this substance does produce mutations, both in the individual genes and by breakage and rearrangement of chromosome parts, such as X-rays and radium do, and in similar abundance. Other substances of the same general group were then found to have a similar effect. This constitutes the first decided break in the chemical attack on mutation. The fact that these findings were made as a direct result of the above inference, when so many previous attempts to produce mutations by chemical means had failed, appears to provide strong evidence that these peculiar somatic effects are in truth consequences of the more underlying ones which, when occurring in the germ cells, are analyzed by the geneticist in his breeding tests. There are, however, some very interesting differences between the nature of the genetic effects of irradiation and of these chemicals, which we cannot go into here, but which give promise of allowing an extension of the genetic and somatic analyses.
We see then that production of mutations by radiation is a method, capable of being turned in various directions, both for the analysis of the germ plasm itself, and of the organism which is in a sense an outgrowth of that germ plasm. It is to be hoped that it may also, in certain fields, prove of increasing practical use in plant and animal improvement, in the service of man. So far as direct practical application in man himself is concerned, however, we are as yet a long way from practicing any intentional selection over our own germ plasm, although like most species we are already encumbered by countless undesirable mutations, from which no individual is immune. In this situation we can, however, draw the practical lesson, from the fact of the great majority of mutations being undesirable, that their further random production in ourselves should so far as possible be rigorously avoided. As we can infer with certainty from experiments on lower organisms that all high-energy radiation must produce such mutations in man, it becomes an obligation for radiologists - though one far too little observed as yet in most countries - to insist that the simple precautions are taken which are necessary for shielding the gonads, whenever people are exposed to such radiation, either in industry or in medical practice. And, with the coming increasing use of atomic energy, even for peace-time purposes, the problem will become very important of insuring that the human germ plasm - the all-important material of which we are the temporary custodians - is effectively protected from this additional and potent source of permanent contamination.
From Nobel Lectures, Physiology or Medicine 1942-1962, Elsevier Publishing Company, Amsterdam, 1964