Portugaliae Acta Biologica - 1949
A Cytophysiological Theory of the Gene, Gene Mutation
and Position Effect
J. A. Serra
Secção de Zoologia-Anthropologia, Faculdade de Ciências. Universidade, Coimbra
3. POSITION EFFECTS AND THE EFFECTS OF REARRANGEMENTS
TYPES OF POSITION EFFECTS: PURE AND MIMIC
As position effects and the effects of rearrangements are of great importance for an understanding of the nature of the gene, it will be necessary to recall here the chief facts on them and their explanations. Position effect was first admitted in the case of Bar of Drosophila melanogaster (Sturtevant): it is well known that Double-Bar, corresponding to the presence of two Bar "factors" contiguously in the same chromosome, causes a greater reduction of the number of facetes than two separate Bar "factors", each in an homologous chromosome. In fact, Bar is a small tandem direct repeat and Double-Bar a triplication instead of a duplication, and their action in principle could perhaps be explained as being a consequence of "genetic unbalance", that is of an excess of genes located in the duplicated and triplicated 16A section of the X chromosome. This explanation, however, seems not to hold true, or at least not to hold necessarily, as a Bar phenotype may also be caused by rearrangements other than a duplication. In effect, a Bar-like phenotype may be induced by several rearrangements involving the section 16A, for instance insertion of this section between 3E1.2 and 3.4, towards the extreme of the X (see Sutton 1943) or inversions with a break in the neighbourhood of 16A and the other end in the chromocenter. A variable baroid effect has been observed also in a case of reversion with salivary chromosomes showing no detectable change (B263-49, Sutton 1943). In this case probably there is a deficiency or an inversion so small that their detection is very difficult or impossible with the actually existing means. Such small inversions have been described as probable in the case of lzs (Oliver and Green 1944) and svrpoi (Goldschmidt and Hannah 1944) and it must be expected that they have occurred also in many other cases in which the salivaries appear normal.
From these observations it follows, as Muller (1936) has already concluded, that usually if not always, Bar is a position effect and the "genetic unbalance", if it exerts some action at all, is little important for the chief phenotypic effect. Though it has been pointed out that there is a Bar locus, situated at bands 16A1.2 of the X (Sutton 1943), it seems that really no such conclusion can be drawn, at least if the mutant gene or the phenotypic effect of the corresponding deficiency is accepted as the relevant criterion for the demonstration that a locus exists (of course, this being followed by cross-over tests and map localization). In effect, among a series of 14 changes from + to Bar in experiments by several authors (Volotov, Muller, Bridges, Demerec, Hoover, Steinberg, Bishop, Bate, Sutton and others -- see table in Sutton 1943) none could be attributed to mutation and only the reverse has been observed, that is changes from B or BB to B+ with no detectable alteration in the salivaries, which have been obtained with relative frequency. It is important that deficiencies of the 16A section have been obtained, byt they do not give the Bar phenotype. On the basis of the postulate that Bar is localized at a certain salivary band, to explain all the apparently conflicting data on Bar and the rearragements producing it or a reversion to + it was necessary to put forward the hypothesis that the phenotypic Bar effect is due to an "interaction" between certain loci along the chromosome and the locus Bar, this latter being probably represented by 16A1.2 (Sutton 1943). It seems, however, that this "interaction" should be viewed as a particular case of the more general explanation of position effects by interaction of gene-products, whose criticism will be presented below and, as in the general case, such an explanation is not satisfactory. The sole conclusion which fits the facts is simply of a factual order, namely that the Bar effect may be caused by rearrangements involving the 16A section, some of them giving the effect others not, according to the region which now comes to lie next to 16A; on the other hand, the B+ locus seems to be no more than an abstraction, as no B+ > B gene mutation is known.
According to this interpretation, Bar is a pure position effect without, at least at present, a known equivalent gene mutant. Other cases are probably Hairy wing (1, 0.0 +/-), duplication of 1B1.2, as also Abruptex and Confluens (1, 3 +/-), Star (2, 1.3) and some others see also below. Very probably also belong here a series of other position effects with inhibitory or enhancer action upon mutant loci, and many other effects of a physiological or metrical character with phenotypes not conspicuously different from the standard. A case of this type is the Pale translocation, between the right limbs of the 2 and 3 chromosomes, which modifies the wc and wa eye and also the testes' and malphigian tubes' colour, that is a factor localized in another chromosome.
The majority of the known position effects, however, are of another kind, giving the same or a similar phenotype as a mutant locus near one of the breaks of the rearrangement and may also be called mimic position effects. This class includes in Drosophila melanogaster one hundred or more of known cases (Goldschmidt 1944) and probably their number will be increased in the future. To it belong many of the multiple alleles of mutant loci which, if an analysis of the salivaries was not possible, would be taken as gene mutations. Cases of this type of position effects are well known for the tip of the X chromosome, where several yellows, scutes, achaetes, whites, etc. have been found to be rearrangement effects. Other examples are Plum2, which produces a brownish variegated eye colour, inseparable from inversion In(2LR)Pm2 with one break near bw, and the effects of several other inversions (see Bridges and Brehme 1944). A case of a mimic position effect is also known with certitude in Oenothera (Catcheside 1947) and others very probably exist in maize, in the mouse and other species.
From the known facts about this group of position effects it is to be presumed that in cases of multiple allelic series with a relatively great number of factors (say, about half a dozen or so) probably some of these factors are no more than position effects. This rule may break down in several cases, for instance perhaps in the w and vg series of Dros. melanogaster, whose majority of mutants seem to be gene mutations. However, even in these cases several mutants are position effects; for instance, several of the wm (mottled) are known to correspond to insertions of the region containing the w locus in or near the chromocenter and it may rightly be assumed that all of the wm alleles belong to this class. The case is somewhat different with several vg alleles, which are known to be deficiencies (for instance vgH, vgC, vgD, vgI) and whose effects may be due, therefore, to the loss of the loci involved, but may or may not be caused also by position effects; a position effect of the vg series is probably vg2, which carried Cy inversions (refs. Bridges and Brehme 1944).
These constitute one of the most interesting types of position effects. In what concerns the phenotype they provoke are of the mimic class, but in many cases they show an additional characteristic, mottling or variegation. After the first case of variegation was explained as an effect of the "wild" allele being relocated in the chromocenter in the salivaries (Schultz 1936) many other cases of mottling have been demonstrated to have this causality. The variegation appears in heterozygotes, where the mutant loci corresponding to genes contained in the segments inserted near the heterochromatin may manifest in some patches of tissue, while adjacent ones remain of the wild type or of an intermediary phenotype between the + allele and the extreme recessive mutant. However, not always does the neighbourhood of heterochromatin near + alleles causes variegation: sometimes there is a uniform phenotype like the recessive. It seems also that, conversely, variegation may also appear in cases where an insertion near the heterochromatin seems not to have taken place (though in these cases it is not yet proved that intercalary heterochromatin has not played a role see for instance the case of rst3 (Kaufman 1943). Independently of heterochromatin neighbourhood, variegation may also be brought about: 1) by a cycle of breakage and reunion, as for instance in maize (McClintock 1941, 1944) and in the case of mustard induced mutations in Drosophila (Auerbach 1944); and 2) by an "ambiental" action, either of the rest of the genotype or of the external ambient itself, over a gene which works near a threshold.
With or without variegation, it is typical of heterochromatin effects that they cause a kind of inactivation of the near-by inserted + alleles, which allows that the recessives manifest in heterozygotes; this is called the Dubinin effect, discovered by this author for cubitus interruptus (ci) in Drosophila. The case of ci has just recently been again extensively investigated and the results obtained on it were valued in connection with a general interpretation of the factors (Stern 1943, 1946, 1948 and refs.). The rearrangments with an action upon ci involve the translocation of pieces of the small right limb of chromosome 4 to other chromosomes. As the very little 4 chromosome has also chromocenter heterochromatin, insertion in the limbs of other chromosomes generally corresponds to the transference of the ci locus to more euchromatic "ambients". Besides the well known Dubinin effect, in which R+ rearrangements involving a break near the ci+ allele of a wild (in relation to ci) 4 chromosome cause a weakening of the + allele which allows a more or less marked manifestation of ci in the heterozygote, Stern and collaborators have described also a type of effect in which R+ causes in R+/ci heterozygotes a more extreme or exagerated ci phenotype (greater interruption of the cubital vein) than the homozygote ci/ci, while in hemizygotes the same rearrangements do not show any effect except a weak minute one. At the same time, the viable R+/R+ homozygotes (some are inviable) in one case are normal in the other like ci/ci. It seems, therefore, that the R+ rearrangements have in themselves only a weak or even nul effect, while in combination with ci serve as enhancers of this latter. By this property, the R+ rearrangements are almost intermediary between pure position effects without any visible action by themselves, and they simply mimic position effects. It seems that considered from this point of view, of serving as enhancers of ci by having in themselves only a small effect, the case of the Rci+ rearrangements has nothing which is not encountered also in the two other types of position effects.
In conclusion, it may be stated that the effects brought about by insertion of euchromatic parts near chromocenter heterochromatin, or conversely, by relocation of regions normally near the chromocenter into euchromatic ones, have the properties of the other two types and, as a particularity, they may show mottling or variegation. It is this latter characteristic which allows [us] to interpret the causality of these effects.
For a valid interpretation of the position effects the following facts are particularly relevant.
1) Position effect is not a feature special to the Diptera. It is perfectly comprehensible that the great bulk of facts about it were obtained in Drosophila, since for the recognition of a position effect a concomitant detailed knowledge of chromosome structure is imperative. However, facts about it in other species are slowly being discovered. An example of a proved position effect has been described in Oenothera blandina (Catcheside 1947). A variegated effect of the locus P accompanying a reciprocal translocation has been demonstrated without any doubt to be a position effect, as the phenotype returned to normal when the chromosomes reverted to their original constitution. The locus S, situated not far from P, shows also a similar position effect. In maize it seems likely that some of the mutants are no more than position effects, particularly in the case of the Pr pr pr endosperm with a translocation involving the chromosome with Pr (Jones 1944), and also in maize several translocations were found to have an influence upon metrical or quantitative characters (Roberts 1942). Some of the biochemical mutants of Neurospora crassa have been found to be associated with translocations (McClintock 1945) and it seems probable that many more will in future be recognised to be linked with rearrangements, some of which give position effects. Finally, in the mouse the brachyury factors T, Fu, Ki, t0 and t1, which suppress crossing-over in a segment about 8 units (Dunn and Caspari 1945). It seems a reasonable extrapolation to conclude that facts like these will come to light in future in ever increasing number, at the measure that the cytogenetics of several species will advance to a stage, as far as possible, like that of Drosophila melanogaster.
2) A valid theory must explain not only the mimic as also the pure position effects; that is, position effects can not simply be taken as corresponding to the effect of known mutants and instead the possibility of their causing a different and "new" (in the sense of not known as a gene mutant) phenotype must also be fully explained. Of course, it may be supposed that any apparently "new" effect is only a consequence of our lack of knowledge on the mutants of a certain chromosome region, but this kind of argument seems impossible to hold true in cases so detailedly studied as that of Bar in Drosophila.
3) In the case of mimic position effects the mutant phenotype must realize in a manner similar to that of mutant alleles, and therefore the same phenogenetic action must be exerted. To them applies all we said above on the unitary action of factors. Pure type effects seem also to have an unitary action, as was said above in connection with the case of Bar. This unitary action results from the fact that hereditary factors, independently of their nature, must act in ontogenesis either through the action of a special biocatalyst which is produced in a reaction system, or by means of modifications in the balance of an enzymatic system, as will be discussed again below, when dealing with gene physiology. In what concerns position effects, for instance the probable case of the R+ rearrangements reported by Stern (1948) it seems that they do not present any problem which has not previously appeared in the phenogenetics of orthodox mendelian factors. A modifyer in general is a factor which acts upon a reaction system only after the chief factor has switched ontogenesis into the appropriated reaction chain or, in the case of biochemical characters, which influences a reaction contributing to the end phases of the synthesis of a product, v. g the latest phases of pigment formation in an Insect eye, after the simple chromogen was oxidized. For instance, in the case of the Rci+ rearrangements with a modifyer action in R+/ci heterozygotes it could perhaps be admitted that they act after ci has caused a partial obliteration of the system of lacunes which in the inflated wing sac seems to be the precursor of L4 (data on wing phenogenesis in Waddington 1941, refs. Serra 1949). As well in mimic as in pure and modifyer types, the phenogenic action of position effects seems to have nothing of special in relation to that of mutants admitted to be orthodox mendelian factors.
4) For an explanation of position effect it is also important to consider the phenomenon of variegation caused by the neighbourhood of heterochromatin. This variability of the position effect resembles the also variable effect of heterochromatic regions in what concerns their charge of peripheric or matricial nucleoproteins, that is of the so called chromatin part of heterocromatic regions, during interkinesis and pro- and telophase.
The first and simpler hypothesis to explain the position effect was that, concomitantly with the rearrangement, a gene mutation occurred at the affected locus. The many cases of reversion to the original phenotype at the same time that the chromosomes reverted also to their original structure disposed of this hypothesis. Other hypotheses which endeavour to explain position effect fail under two headings (Ephrussi and Sutton 1944): kinetic or of the diffusion type, and structural or of alterations in the physico-chemical and cytological states of the genes or the chromosomes. For a discussion of all the hypotheses, it is important to consider the distance of the breaks at which position effect manifests. In Drosophila, position effects may reach about 50 salivary bands (Demerec 1940), which for a chromosome of 800-1000 bands corresponds to about 1/16-1/20 of the chromosome length. In a mitotic X chromosome with about a 2µ length and 2/3 of euchromatin, this would correspond to about 4/60=.07µ or 700A. As the chromosome thread is coiled in a helix, the distance over the chromonema shall be greater by a factor of 3 or 4, and so the position effect could perhaps manifest in the interkynetic chromonema over distances of about 2,500A. However, more usually the distance reached by position effect is of about 10 bands or less and thus corresponds to an interkinetic chromonema length of 250A or less, or 80-100A by the shortest way.
These hypotheses are based on an idea of Muller, supposing an interaction of gene-products immediately after they are produced, along the chromosome. An improved version due to Waddington (1939) admitted that the interaction takes place by means of the two neighbouring genes reacting with a common gene substrate. Ephrussi and Sutton (1944) have discussed these hypotheses from a quantitative point of view and shown that practically at distances of 10g (in which g is the radius of the gene) an interaction could no more be effective. In an X chromosome of about 2µ length, of which 0.6µ in the proximal heterochromatic part, the length of the euchromatic chromonema, supposing that it must be multiplied by 3 or 4 to account for spiralisation, is about 5µ. Admitting that there are 200 genes in the X, each gene should correspond to about 0.025µ if all the chromonema was formed of genes. Assuming that only one half is of genic material, then each gene should correspond to about 0.012µ or 102A. Ten times this length is less than 1/2 of the distance attained by the far reaching position effects and therefore these effects could not be explained by this hypothesis.
A similar, though perhaps less representative, conclusion is reached by considering that the gene corresponds in mean to about 4-5 bands and that at the distance of 50 bands a low concentration of gene product, of the order of 0.1 of that existing in the immediate neighbourhood of the gene, should prevail and therefore a strong interaction would be impossible. Usually the length of the gene is given as much less than the length we admit here; for instance, Ephrussi and Sutton (1944) suppose that the "active part" of the gene has a radius of 5A and in the interpretation of the action of radiations upon mutation the radius of the gene is also habitually found to be of this order of magnitude. As we have already seen aboe and will be discussed at length in a later section (see chapter 7) this conclusion must not be valid and the length of the gene on the chromonema must be considered greater than these results at first sight suggest.
The kinetic hypotheses may also be criticised from a qualitative point of view. According to all that is known about nuclear physiology, the synthetic activities are chiefly an attribute of the resting, and not of the dividing, nucleus. The distances between far located genes may be very short in the interkinetic nucleus, for instance if two chromonemata of non-homologous chromosomes come to lie side by side during the resting stage (Serra 1944). This would bring about a variability of the gene effects imcompatible with the strict causality which is always observed in what respects the phenogenesis of gene mutations and rearrangements.
It may be concluded that both quantitatively and qualitatively the kinetic hypotheses do not explain the known facts.
These include several explanations at successive levels of organisation, molecular and structural. At the molecular level, it could be supposed (Muller) that a different gene composition could act by modifying the chemical bonds in the aggregate comprising a group of genes. Under this simple form the hypothesis is untenable; chemical valence bonds, for instance those of the covalent type, reach distances of less than 3A and usually Van der Waals forces only a little more (3-4.5A). These are distances very much lower than the 2,500A reached by some position effects.
Among the forces active at a physico-chemical level only coulomb ones may act at distances of this order of magnitude. It could be assumed that a number of dipole moments developed on each gene locus and the distribution of these moments in interacting pieces of the chromonema that produce position effects should be such that greater coulomb fields would result, causing a distorsion of the action of the genes. Supposing that each dipole moment has a mean strength of 2x10-18 e. s. u., which is found in many organic molecules, and acts at distances of about 4A like Van der Waals forces, for an effective work of about 8 kcal/mole, like that which is necessary to separate to an infinite distance two particles linked by cohesion forces (Meyer and Mark 1930) and a distance of about 2,500A, it would be necessary according to the formula E=m/r3 (E, work, m, charge, r distance) and the particles had m of about 2.5x108 polarisable groups, which, although a very high one, does not seem an absolutely impossible number for a big particle of protein, or perhaps of nucleoprotein, but is impossible for a single molecule.
The greater difficulty with this hypothesis, however, lies in the number of supplementary assumptions which it necessarily implies. In effect, it is not clear why some genes interact and others not, since all must have not very different numbers of polarisable groups and therefore the interacting forces should not be very different from gene to gene. In order that they could interact due to dipole moments it would also be necessary either that they could assume special complementary configurations, or that they had the possibility of changing of "position" in order to obtain a maximum of polarisation; both these assumptions seem incompatible with the fixed position of the two interacting genes on the fibrous chromonema. As the position effects are practically independent of the temperature (except mottling, for which a special explanation is valid) the chief part of the dipole moments should not correspond to orientation polarisation and this would also exclude the hypothesis of the genes changing of "position" or structure in order to obtain the maximum dipole effects*. In view of these facts, at present it seems that the hypothesis of an interaction of genes by means of forces of the coulomb type does not give a satisfactory explanation of the position effects.
Another structural explanation was developed by Ephrussi and Sutton (1944), already mentioned in a simpler form by Sturtevant (1925) and by Muller (1941), and is based upon the idea that following breakage and rearrangement the pairing of the homologous segments was impaired, so that a deformation of the affected segment could result. As Ephrussi and Sutton stress, this hypothesis requires structural heterozygoty of the segment in question in order that position effect may manifest. It is known, however, that several position effects manifest also in homozygous and hemizygous condition, for instance Bar and several euchromatic rearrangements in the y-sc and w regions of Drosophila. Although the case of Bar could yet be brought in accord with the hypothesis by considering that the two duplications of a same chromosome tend to pair and so cause deformation, and that in the case of rearrangements involving heterochromatin changes the pairing condition of this latter could also be invoked with the same finality (Ephrussi and Sutton) the effects of small euchromatic rearrangements are not accounted for by the hypothesis. The case of Star and asteroid in Dros. melanogaster is also against this interpretation. S and ast. which are a reverse repeat, located contiguously in the doublet 21E1.2 (left arm of chromosome 2) give rise to different phenotypes if carried by the same or by homologous chromosomes: that is, S ast/++ is different from S+/+ast (Lewis 1945, see also Catcheside 1947). If the pairing-deformation was true both should give about the same phenotype.
Of course, position effects must manifest in the nucleus of synthesis, in interkynesis. For the validity of the pairing-deformation hypothesis it would be necessary to admit that in the resting nucleus the forces of pairing continued to act. In the case of Diptera, with their somatic pairing, this could perhaps be yet admitted, though it seems that only with a great deal of improbability, but for other organisms this should be a pure assumption without any basis. Now, as position effects are known to occur not only in Diptera as also in other groups, the hypothesis of the pairing-deformation seems untenable.
While the explanations hitherto referred accept the particulate gene concept and indeed try to free this concept of the objections raized by the discovery of position effects, Goldschmidt presents a theory which endeavours to explain these effects at the same time that gene mutation, admitting that there is no particulate or individualized gene and that the unit is the chromosome, with its structure. The theory of Goldschmidt is the most strictly structural which has been presented as it postulates that, excluding "numeric" mutations of the polyploidy and heteroploidy types, all the happenings which provoke a genetic difference, that is gene mutations as well as position effects, are caused by a change of structure in a chromosome. Alterations in the structure of the chromosome are called gene mutations if they happen at a sub-microscopical level, or position effects when they can be demonstrated at the microscope. As we have already said above, it is only a step from this architectural theory to the concept of an hereditary continuum formed by all the chromosomes of the haploid set associated in a composite body, of the Ascaris megalocephala univalens type. Changes in the architecture of this chromosomic continuum should bring about alterations in the phenogenesis, causing a different phenotype: though there would be mutation and mutants, such concepts as particulate genes and wild type alleles should be only descriptive terms, or at most pragmatic concepts without any real basis.
The criticism of a part of this explanation of position effect must be made jointly with that of the concept of mutation and the denial of the unitary gene concept, and in this latter part it has already been attempted above. If there is no wild type allele, it is very difficult to explain why a mutation brings about the loss of one kind of enzyme, the phenotype resulting normal when the product formed by the action of the enzyme is experimentally furnished. For instance in Drosophila, if kynurenin is administered to v flies, their eyes may now form brown pigments; it seems unescapable to conclude that v+ is responsible for the oxidation of triptophane by means of an oxidase. Another most serious difficulty of this theory is the explanation of the covering effect of deficiencies by corresponding duplications located elsewhere in the chromosomes, which has been verified to hold true in a number of cases. In these cases the stucture, or better the architecture, of both the deficient and the duplication-carrying chromosomes has been altered and no phenotypic change of the mutant type, which should be expected if mutations were merely structural alterations, is apparent (excepting, obviously, eventual cases of accompanying position effects).
On the other hand, if position effect is no more than mutation at an higher level, the number of pure position effects, that is position effects without equivalent "point" or "gene" mutations, should be expected to be higher, as it is not understandable why should in many cases a mutation at an higher level mimic precisely a mutation which has occurred at a submicroscopic level on a near-by locus. It is possible to assume that in reality what is affected is a certain segment of chromosome, and that within this segment submicroscopic mutation and microscopically visible change of structure both produce the same effect; this, however, is the recognition that the segment exists, a unit along the chromosome, connected with a certain genetic activity and this in the limit is not different from a not too narrow unitary gene theory. Besides, if position effect is simply mutation at an higher level, how is to be explained the production of variegation in certain of them?
Another interesting point about the architectural theory is that by explaining position effect as a mutation and mutation as a position effect caused by structural changes, nothing is really explained: it would be necessary to explain why a change in the succession of bands or other structures along the chromosomes brings about a phenogenetic unitary effect, at least in biochemical characters, and very probably also in more complex morphological ones. Though this is not a typical circular argument, it really only deferres the explanation to another level.
Finally, another explanation which may yet be included in the structural theories, although more properly it could be called cytological, since the level at which it works is the cytological and not the molecular-structural, has been presented by Serra (1944) as a generalization of the interpretation of the rearrangements involving heterochromatin. Now we are convinced that this interpretation is not so generalizable as was postulated in our former work and that the reality is more complex. In the following chapter we present the development of this hypothesis, further elaborated to cover all kinds of known position effects.