- Email: [email protected]
Too late for the midwife toad
COMMENT
Stress, variability and Hsp90
Outlook
Too late for the midwife toad stress, variability and Hsp90 In the 1940s and 1950s, Waddington put forward his theories of canalization and genetic assimilation. These provided a genetic basis to account for the inheritance of some apparently ‘acquired’ characters. Rutherford and Lindquist have now provided a molecular framework for these theories. Their results are also relevant to observations from the 1950s concerning homozygosity and variability, with a bearing on current views concerning the use of inbred strains. istory may be circular, but the history of science is helical: it repeats itself, but each time at a deeper level. Rutherford and Lindquist1 have recently published a superb example of this truism. They showed that when the Drosophila heat shock protein Hsp90 was compromised, either genetically (by mutation) or environmentally (by high temperatures or drug treatment), genetic variation that had been silenced by the buffering action of Hsp90 became expressed and, after selection, continued to be expressed even when Hsp90 was restored to its normal state.
H
Signal-transduction chaperones The major heat shock protein Hsp90 lies at the interface of several developmental pathways, acting as a chaperone for unstable signalling proteins (e.g. steroid hormone receptors, cyclin-dependent kinases) and keeping them in check until they become stabilized by conformational changes associated with signal transduction. Rutherford and Lindquist found that when the function of the Hsp90 in Drosophila was impaired either genetically, using mutants, or environmentally, using the drug geldanamycin, abnormalities developed in many adult structures (e.g. wings, legs, eyes). The effects were exacerbated by high or low temperatures. Specific abnormalities depended on the genetic background and so were unlikely to be due to new mutations. After a few generations of selection they became independent of the level of Hsp90. Rutherford and Lindquist concluded that the abnormalities represent cryptic genetic variation that would normally be silenced by the buffering capacity of Hsp90; when this buffering is compromised, the variant genes are expressed, and can be enriched by selection to a point at which the abnormalities manifest despite the restoration of Hsp90 function (Fig. 1).
will be regained. If, however, the deviation is too great, homeostatic mechanisms will no longer be effective, development will be switched to a new channel, and an altered phenotype will result. The consequences of disrupting canalization were illustrated in Waddington’s ‘crossveinless’ experiment3. Drosophila subjected to a high temperature expressed a specific wing character: after a few generations of selection and repeated heat shock, the character continued to be expressed even when the temperature was maintained at its normal level. Waddington asserted that most examples of the so-called inheritance of acquired characters could be explained by this process of ‘genetic assimilation’. By this term he implied that some trait apparently acquired in response to an environmental factor, such as heat shock (in fact due to cryptic genetic variation made manifest by the altered environment), had become ‘assimilated’ into the expressed genetic programme through enrichment due to selection. Until the altered conditions exposed the underlying variation, selection could not act. The crossveinless character was not, of course, an adaptive response to the heat shock; nonetheless, ‘if an animal is subjected to unusual circumstances to which it can react in an adaptive manner, the development of the adaptive character might itself become so far canalized that it continued to appear even when the conditions returned to the previous norm’3. An example with tragic consequences is the work of Kammerer, discussed in detail by Koestler4. Kammerer’s claim that the ‘acquired’ pigmented thumb pads of midwife toads subsequently became heritable (a situation exactly analogous to Waddington’s crossveinless example) was attacked so relentlessly by Bateson that Kammerer eventually committed suicide.
Stress as a destabilizing factor Canalization Thus an apparently acquired character (‘acquired’, for example, by drug treatment, or temperature stress) has become inherited, providing a mechanism that (in Rutherford and Lindquist’s intriguing phrase) ‘assists the process of evolutionary change in response to the environment’. Rutherford and Lindquist point out that the underlying principle of developmental homeostasis was embodied in Waddington’s 1942 model of the canalization of development2. ‘Under the influence of natural selection, development tends to become canalized so that more-or-less normal organs and tissues are produced even in the face of slight abnormalities of the genotype or of the external environment’3. Thus, developmental homeostasis ensures that when development is moving down a particular channel, small deviations will be compensated for and the central channel 0168-9525/99/$ – see front matter © 1999 Elsevier Science All rights reserved. PII: S0168-9525(99)01732-1
How widespread is the effect that Rutherford and Lindquist demonstrated for Hsp90? Is there perhaps a plethora of proteins acting as chaperones for signal-transduction elements, buffering the effects of underlying genetic variation? Could this be the explanation of the heritable changes that Durrant5 induced in flax plants by different fertilizer treatments? When size was doubled or halved, the plants bred true (Fig. 2), but those that were less affected retained their plasticity. Perhaps the plants that reacted more drastically to the fertilizer treatments were expressing cryptic genetic variation that shifted development into a new channel, with a new target for the homeostatic buffering effect of signal-transduction chaperones. Similar results have been obtained more recently6, with tobacco as well as with flax, involving specific DNA alterations as well as phenotypic changes. TIG May 1999, volume 15, No. 5
Anne McLaren a.mclaren@ welc.cam.ac.uk Wellcome/CRC Institute, Tennis Court Road, Cambridge, UK CB2 1QR. 169
Outlook
COMMENT
Stress, variability and Hsp90
FIGURE 1. Exposing cryptic genetic variation Normal level of Hsp90
(a)
Lowered level of Hsp90
(b)
Deformed fore-leg
Deformed eye
(c)
Wild-type Drosophila
Wild-type Drosophila
Deformed wing
Deformed fore-leg
Normal level of Hsp90 restored
Deformed eye
Deformed fore-leg
Deformed wing
Wild-type Drosophila
Deformed eye
Deformed wing
(a) When heat-shock protein Hsp90 is expressed at the normal level, wild-type Drosophila with a normal phenotype develop (indicated by only the wild-type fly being in the white area). Numerous cryptic variations are suppressed by the buffering action of Hsp90 (grey area). (b) When the level of Hsp90 is reduced by gene targetting, or by drug treatment, or by heat treatment, cryptic variations are no longer suppressed, and mutant flies develop (white area). These mutants can be subject to selection. (c) After several generations of selection (in this case, for flies with deformed legs), mutant flies develop even when Hsp90 is restored to its normal level (white area). Many cryptic variations are again suppressed. Development has been shifted into a new pathway, that is, a change of canalization, in Waddington’s terminology.
There are many examples, often included under the blanket term ‘epigenetic inheritance’, of altered genetic or environmental conditions resulting in heritable changes in gene expression. Nuclear transplantation in early mouse embryos produced changes in the expression of genes encoding olfactory marker protein and major urinary proteins: these changes, some of which occasionally manifested even in controls, were transmitted to subsequent generations7. Evidently the new expression patterns, apparently ‘acquired’ in response to the altered cytoplasmic environment, were in reality due to cryptic genetic variation, manifested after a shift in canalization and affording a new target for homeostasis. Eighteen generations of selection for tameness in foxes8 resulted not
FIGURE 2. Heritable changes caused by environmental stress
only in heritable variations in the normal reproductive pattern, but also in the appearance of new morphological characters (tail curling, spotting, pigmentation changes). These were postulated by Belyaev to be due to the activation of ‘dormant’ alleles, activated by the stress of destabilizing selection. Similar examples were cited by Darwin9. In flour beetles, Wade and his colleagues10 were able to show that the destabilizing effect of distant hybridization (deformed appendages, changes in sex ratio and size) was uncovering genetic variation that pre-existed within species, but was concealed by homeostasis. ‘The lens of hybridization’, they concluded, ‘magnifies segregating genetic differences within species’10. These examples attest to the generality of the phenomenon that plant and animal breeders have exploited for centuries, and that Rutherford and Lindquist have for the first time explained in molecular terms – namely that genetic or environmental stress can produce heritable developmental changes.
Phenotypic uniformity
Fourth generation plants of two extreme types of flax induced by different fertilizer treatments in the parental generation. (Adapted from Ref. 21.)
170
TIG May 1999, volume 15, No. 5
An additional and more subtle aspect of the destabilization of heredity, one that could perhaps involve similar molecular mechanisms, relates to homeostasis and the link between phenotypic uniformity and heterosis or ‘hybrid vigour’; that is, the tendency for first-generation crosses to exceed both parental strains with respect to quantitative characters, in particular those associated with fitness. Livesay11 crossed three strains of rats: the F1 hybrid rats were larger than the parent strains, and they were also substantially more uniform in weight. The F2 generation, which was of course genetically more variable than the F1, was less uniform in weight than the F1, but still more uniform than the parental generation. McLaren and Michie12 found that two inbred lines of mice were more than three times as variable in their response to a narcotic drug as the F1 cross between the strains. They were also more variable than a randomly bred strain of mice, in spite of the genetic uniformity of the inbreds. It seems that the departure from the genetic norm that inbreeding and homozygosity represent might act as a stress factor, compromising the buffering effect of signal-transduction chaperones and resulting in increased phenotypic variability.
COMMENT
Stress, variability and Hsp90
The developmental stability conferred by heterozygosity means that F1 hybrids are always likely to be phenotypically more uniform than inbred strains; they are also more costly to produce. Whether or not inbred mice are less or more uniform phenotypically than genetically heterogeneous, randomly bred strains (which are relatively cheap to produce) will depend on the trait under study, the degree to which it is related to the organism’s fitness12, and the extent to which it is affected by environmental influences during development. The assumption, still often made, that inbred strains are suitable for bioassays and drug testing because they are uniform, cannot be sustained13–15.
Genetic and environmental norms For most animal species, inbreeding represents a departure from the genetical situation to which the species is adapted. As Waddington2 pointed out, using as an example the relative uniformity of the wild-type compared with the variability of any mutant strain, it is this departure from the genetic norm that challenges the effectiveness of canalization. But organisms are adapted not only to genetic but also to particular environmental situations. It is, therefore, not surprising that adverse conditions of rearing lead not only to a decrease in vigour, but also to an increase in variability, at least for traits correlated with fitness. For example, the body weights of young mice were found to be strikingly more variable when reared in uniformly hot (288C) or uniformly cold (58C) environments than in the more temperate conditions (218C) to which the species has presumably been adapted by natural selection16. The same is true for plants: Went (1953) showed that, for peas and tomatoes, optimal growing conditions were associated with minimal phenotypic variability17. Waddington’s view of the link between heterozygosity and phenotypic uniformity, namely that departure from the genetic norm by inbreeding disrupts homeostasis and hence canalization, is essentially based on evolutionary and developmental considerations and can readily be extended to explain the link between environmental conditions and variability. A more biochemical view of the role of heterozygosity was well expressed by Robertson and Reeve18, in reporting their findings in Drosophila. For six highly inbred lines and six crosses between them, the F1 variance was consistently less than the mean variance of
References 1 Rutherford, S.L. and Lindquist, L. (1998) Hsp90 as a capacitor for morphological evolution. Nature 396, 336–342 2 Waddington, C.H. (1942) Canalization of development and the inheritance of acquired characters. Nature 150, 563–565 3 Waddington, C.H. (1953) Genetic assimilation of an acquired character. Evolution 7, 118–126 4 Koestler, A. (1971) The Case of the Midwife Toad. Hutchinson 5 Durrant, A. (1971) Induction and growth of flax genotrophs. Heredity 27, 277–298 6 Schneeberger, R.G. and Cullis, C.A. (1991) Specific DNA alterations associated with the environmental induction of heritable changes in flax. Genetics 128, 619–630 7 Roemer, I. et al. (1997) Epigenetic inheritance in the mouse. Curr. Biol. 7, 277–280 8 Belyaev, D.K. (1979) Destabilizing selection as a factor in domestication. J. Hered. 70, 301–308 9 Darwin, C.R. (1868) The Variation of Animals and Plants under Domestication. John Murray 10 Wade, M.J. et al. (1997) Genetic variation segregating in natural population of Tribolium castaneum affecting traits observed in hybrids with T. freemani. Genetics 147, 1235–1247
Outlook
the parental strains, and declined progressively with the number of pairs of chromosomes that were rendered heterozygous. They concluded: ‘The more heterozygous individuals will carry a greater diversity of alleles, and these are likely to endow them with a greater biochemical versatility in development. This will lead to heterosis, because of more efficient use of materials available in the environment, and also to a reduced sensitivity to environmental variations, since there will be more ways of overcoming the obstacles which such variations put in the way of normal development’. The ‘biochemical versatility’ hypothesis of hybrid vigour and hybrid uniformity is attractive, and became widely accepted19,20. It suffers, however, from a lack of supporting biochemical evidence, and fails totally to account for the effect of environmental conditions on phenotypic variability.
Looking forwards, looking back Could it be that Hsp90-like signal-transduction chaperones play a central role in homeostasis, and function best under those genetic and environmental conditions to which the species is adapted? Major loss of function owing to mutation or drug treatment will expose sufficient of the underlying genetic variation to produce visible abnormalities; lesser impairment of function due to suboptimal genetic or environmental conditions could expose just enough of the underlying genetic variation to increase phenotypic variability of many quantitative characters. Alternatively, minor impairment of function might simply reduce the buffering action against ‘developmental noise’ caused by random micro-environmental effects, with little or no genetic basis (a possibility considered by Rutherford and Lindquist, but ruled out in the context of major impairment of Hsp90 function by their experimental evidence). More than two thirds of the following references predate 1960. They describe experimental findings that tend to be overlooked today because they do not fit comfortably into the conventional framework of molecular development. The work of Rutherford and Lindquist points to ways in which this framework could usefully be extended.
Acknowledgement I am grateful to the Wellcome Trust for financial support.
11 Livesay, E.A. (1930) An experimental study of hybrid vigor or heterosis in rats. Genetics 15, 17–54 12 McLaren, A. and Michie, D. (1956) Variability of response in experimental animals: A comparison of the reactions of inbred, F1 hybrid and random bred mice to a narcotic drug. J. Genet. 54, 440–455 13 Michie, D. (1955) Towards uniformity in experimental animals. Coll. Pap. Lab. Anim. Bur. 3, 37–47 14 McLaren, A. and Michie, D. (1954) Are inbred strains suitable for bioassay? Nature 173, 686–688 15 Biggers, J.D. and Claringbold, P.J. (1954) Why use inbred lines? Nature 174, 596 16 Ashoub, M.R. et al. (1958) The effect of the environment on phenotypic variability. Proc. R. Soc. London Ser. B 148, 192–203 17 Went, F.W. (1953) Gene action in relation to growth and development. I. Phenotypic variability. Proc. Natl. Acad. Sci. U. S. A. 39, 839–848 18 Robertson, F.W. and Reeve, E.C.R. (1952) Heterozygosity, environmental variation and heterosis. Nature 170, 286 19 Mather, K. (1953) Genetic control of stability in development. Heredity 7, 297–336 20 Lerner, I.M. (1954) Genetic Homeostasis. Oliver and Boyd 21 Durrant, A. (1962) The environmental induction of heritable change in Linum. Heredity 17, 27–61
TIG May 1999, volume 15, No. 5
171
Stress, variability and Hsp90
Outlook
Too late for the midwife toad stress, variability and Hsp90 In the 1940s and 1950s, Waddington put forward his theories of canalization and genetic assimilation. These provided a genetic basis to account for the inheritance of some apparently ‘acquired’ characters. Rutherford and Lindquist have now provided a molecular framework for these theories. Their results are also relevant to observations from the 1950s concerning homozygosity and variability, with a bearing on current views concerning the use of inbred strains. istory may be circular, but the history of science is helical: it repeats itself, but each time at a deeper level. Rutherford and Lindquist1 have recently published a superb example of this truism. They showed that when the Drosophila heat shock protein Hsp90 was compromised, either genetically (by mutation) or environmentally (by high temperatures or drug treatment), genetic variation that had been silenced by the buffering action of Hsp90 became expressed and, after selection, continued to be expressed even when Hsp90 was restored to its normal state.
H
Signal-transduction chaperones The major heat shock protein Hsp90 lies at the interface of several developmental pathways, acting as a chaperone for unstable signalling proteins (e.g. steroid hormone receptors, cyclin-dependent kinases) and keeping them in check until they become stabilized by conformational changes associated with signal transduction. Rutherford and Lindquist found that when the function of the Hsp90 in Drosophila was impaired either genetically, using mutants, or environmentally, using the drug geldanamycin, abnormalities developed in many adult structures (e.g. wings, legs, eyes). The effects were exacerbated by high or low temperatures. Specific abnormalities depended on the genetic background and so were unlikely to be due to new mutations. After a few generations of selection they became independent of the level of Hsp90. Rutherford and Lindquist concluded that the abnormalities represent cryptic genetic variation that would normally be silenced by the buffering capacity of Hsp90; when this buffering is compromised, the variant genes are expressed, and can be enriched by selection to a point at which the abnormalities manifest despite the restoration of Hsp90 function (Fig. 1).
will be regained. If, however, the deviation is too great, homeostatic mechanisms will no longer be effective, development will be switched to a new channel, and an altered phenotype will result. The consequences of disrupting canalization were illustrated in Waddington’s ‘crossveinless’ experiment3. Drosophila subjected to a high temperature expressed a specific wing character: after a few generations of selection and repeated heat shock, the character continued to be expressed even when the temperature was maintained at its normal level. Waddington asserted that most examples of the so-called inheritance of acquired characters could be explained by this process of ‘genetic assimilation’. By this term he implied that some trait apparently acquired in response to an environmental factor, such as heat shock (in fact due to cryptic genetic variation made manifest by the altered environment), had become ‘assimilated’ into the expressed genetic programme through enrichment due to selection. Until the altered conditions exposed the underlying variation, selection could not act. The crossveinless character was not, of course, an adaptive response to the heat shock; nonetheless, ‘if an animal is subjected to unusual circumstances to which it can react in an adaptive manner, the development of the adaptive character might itself become so far canalized that it continued to appear even when the conditions returned to the previous norm’3. An example with tragic consequences is the work of Kammerer, discussed in detail by Koestler4. Kammerer’s claim that the ‘acquired’ pigmented thumb pads of midwife toads subsequently became heritable (a situation exactly analogous to Waddington’s crossveinless example) was attacked so relentlessly by Bateson that Kammerer eventually committed suicide.
Stress as a destabilizing factor Canalization Thus an apparently acquired character (‘acquired’, for example, by drug treatment, or temperature stress) has become inherited, providing a mechanism that (in Rutherford and Lindquist’s intriguing phrase) ‘assists the process of evolutionary change in response to the environment’. Rutherford and Lindquist point out that the underlying principle of developmental homeostasis was embodied in Waddington’s 1942 model of the canalization of development2. ‘Under the influence of natural selection, development tends to become canalized so that more-or-less normal organs and tissues are produced even in the face of slight abnormalities of the genotype or of the external environment’3. Thus, developmental homeostasis ensures that when development is moving down a particular channel, small deviations will be compensated for and the central channel 0168-9525/99/$ – see front matter © 1999 Elsevier Science All rights reserved. PII: S0168-9525(99)01732-1
How widespread is the effect that Rutherford and Lindquist demonstrated for Hsp90? Is there perhaps a plethora of proteins acting as chaperones for signal-transduction elements, buffering the effects of underlying genetic variation? Could this be the explanation of the heritable changes that Durrant5 induced in flax plants by different fertilizer treatments? When size was doubled or halved, the plants bred true (Fig. 2), but those that were less affected retained their plasticity. Perhaps the plants that reacted more drastically to the fertilizer treatments were expressing cryptic genetic variation that shifted development into a new channel, with a new target for the homeostatic buffering effect of signal-transduction chaperones. Similar results have been obtained more recently6, with tobacco as well as with flax, involving specific DNA alterations as well as phenotypic changes. TIG May 1999, volume 15, No. 5
Anne McLaren a.mclaren@ welc.cam.ac.uk Wellcome/CRC Institute, Tennis Court Road, Cambridge, UK CB2 1QR. 169
Outlook
COMMENT
Stress, variability and Hsp90
FIGURE 1. Exposing cryptic genetic variation Normal level of Hsp90
(a)
Lowered level of Hsp90
(b)
Deformed fore-leg
Deformed eye
(c)
Wild-type Drosophila
Wild-type Drosophila
Deformed wing
Deformed fore-leg
Normal level of Hsp90 restored
Deformed eye
Deformed fore-leg
Deformed wing
Wild-type Drosophila
Deformed eye
Deformed wing
(a) When heat-shock protein Hsp90 is expressed at the normal level, wild-type Drosophila with a normal phenotype develop (indicated by only the wild-type fly being in the white area). Numerous cryptic variations are suppressed by the buffering action of Hsp90 (grey area). (b) When the level of Hsp90 is reduced by gene targetting, or by drug treatment, or by heat treatment, cryptic variations are no longer suppressed, and mutant flies develop (white area). These mutants can be subject to selection. (c) After several generations of selection (in this case, for flies with deformed legs), mutant flies develop even when Hsp90 is restored to its normal level (white area). Many cryptic variations are again suppressed. Development has been shifted into a new pathway, that is, a change of canalization, in Waddington’s terminology.
There are many examples, often included under the blanket term ‘epigenetic inheritance’, of altered genetic or environmental conditions resulting in heritable changes in gene expression. Nuclear transplantation in early mouse embryos produced changes in the expression of genes encoding olfactory marker protein and major urinary proteins: these changes, some of which occasionally manifested even in controls, were transmitted to subsequent generations7. Evidently the new expression patterns, apparently ‘acquired’ in response to the altered cytoplasmic environment, were in reality due to cryptic genetic variation, manifested after a shift in canalization and affording a new target for homeostasis. Eighteen generations of selection for tameness in foxes8 resulted not
FIGURE 2. Heritable changes caused by environmental stress
only in heritable variations in the normal reproductive pattern, but also in the appearance of new morphological characters (tail curling, spotting, pigmentation changes). These were postulated by Belyaev to be due to the activation of ‘dormant’ alleles, activated by the stress of destabilizing selection. Similar examples were cited by Darwin9. In flour beetles, Wade and his colleagues10 were able to show that the destabilizing effect of distant hybridization (deformed appendages, changes in sex ratio and size) was uncovering genetic variation that pre-existed within species, but was concealed by homeostasis. ‘The lens of hybridization’, they concluded, ‘magnifies segregating genetic differences within species’10. These examples attest to the generality of the phenomenon that plant and animal breeders have exploited for centuries, and that Rutherford and Lindquist have for the first time explained in molecular terms – namely that genetic or environmental stress can produce heritable developmental changes.
Phenotypic uniformity
Fourth generation plants of two extreme types of flax induced by different fertilizer treatments in the parental generation. (Adapted from Ref. 21.)
170
TIG May 1999, volume 15, No. 5
An additional and more subtle aspect of the destabilization of heredity, one that could perhaps involve similar molecular mechanisms, relates to homeostasis and the link between phenotypic uniformity and heterosis or ‘hybrid vigour’; that is, the tendency for first-generation crosses to exceed both parental strains with respect to quantitative characters, in particular those associated with fitness. Livesay11 crossed three strains of rats: the F1 hybrid rats were larger than the parent strains, and they were also substantially more uniform in weight. The F2 generation, which was of course genetically more variable than the F1, was less uniform in weight than the F1, but still more uniform than the parental generation. McLaren and Michie12 found that two inbred lines of mice were more than three times as variable in their response to a narcotic drug as the F1 cross between the strains. They were also more variable than a randomly bred strain of mice, in spite of the genetic uniformity of the inbreds. It seems that the departure from the genetic norm that inbreeding and homozygosity represent might act as a stress factor, compromising the buffering effect of signal-transduction chaperones and resulting in increased phenotypic variability.
COMMENT
Stress, variability and Hsp90
The developmental stability conferred by heterozygosity means that F1 hybrids are always likely to be phenotypically more uniform than inbred strains; they are also more costly to produce. Whether or not inbred mice are less or more uniform phenotypically than genetically heterogeneous, randomly bred strains (which are relatively cheap to produce) will depend on the trait under study, the degree to which it is related to the organism’s fitness12, and the extent to which it is affected by environmental influences during development. The assumption, still often made, that inbred strains are suitable for bioassays and drug testing because they are uniform, cannot be sustained13–15.
Genetic and environmental norms For most animal species, inbreeding represents a departure from the genetical situation to which the species is adapted. As Waddington2 pointed out, using as an example the relative uniformity of the wild-type compared with the variability of any mutant strain, it is this departure from the genetic norm that challenges the effectiveness of canalization. But organisms are adapted not only to genetic but also to particular environmental situations. It is, therefore, not surprising that adverse conditions of rearing lead not only to a decrease in vigour, but also to an increase in variability, at least for traits correlated with fitness. For example, the body weights of young mice were found to be strikingly more variable when reared in uniformly hot (288C) or uniformly cold (58C) environments than in the more temperate conditions (218C) to which the species has presumably been adapted by natural selection16. The same is true for plants: Went (1953) showed that, for peas and tomatoes, optimal growing conditions were associated with minimal phenotypic variability17. Waddington’s view of the link between heterozygosity and phenotypic uniformity, namely that departure from the genetic norm by inbreeding disrupts homeostasis and hence canalization, is essentially based on evolutionary and developmental considerations and can readily be extended to explain the link between environmental conditions and variability. A more biochemical view of the role of heterozygosity was well expressed by Robertson and Reeve18, in reporting their findings in Drosophila. For six highly inbred lines and six crosses between them, the F1 variance was consistently less than the mean variance of
References 1 Rutherford, S.L. and Lindquist, L. (1998) Hsp90 as a capacitor for morphological evolution. Nature 396, 336–342 2 Waddington, C.H. (1942) Canalization of development and the inheritance of acquired characters. Nature 150, 563–565 3 Waddington, C.H. (1953) Genetic assimilation of an acquired character. Evolution 7, 118–126 4 Koestler, A. (1971) The Case of the Midwife Toad. Hutchinson 5 Durrant, A. (1971) Induction and growth of flax genotrophs. Heredity 27, 277–298 6 Schneeberger, R.G. and Cullis, C.A. (1991) Specific DNA alterations associated with the environmental induction of heritable changes in flax. Genetics 128, 619–630 7 Roemer, I. et al. (1997) Epigenetic inheritance in the mouse. Curr. Biol. 7, 277–280 8 Belyaev, D.K. (1979) Destabilizing selection as a factor in domestication. J. Hered. 70, 301–308 9 Darwin, C.R. (1868) The Variation of Animals and Plants under Domestication. John Murray 10 Wade, M.J. et al. (1997) Genetic variation segregating in natural population of Tribolium castaneum affecting traits observed in hybrids with T. freemani. Genetics 147, 1235–1247
Outlook
the parental strains, and declined progressively with the number of pairs of chromosomes that were rendered heterozygous. They concluded: ‘The more heterozygous individuals will carry a greater diversity of alleles, and these are likely to endow them with a greater biochemical versatility in development. This will lead to heterosis, because of more efficient use of materials available in the environment, and also to a reduced sensitivity to environmental variations, since there will be more ways of overcoming the obstacles which such variations put in the way of normal development’. The ‘biochemical versatility’ hypothesis of hybrid vigour and hybrid uniformity is attractive, and became widely accepted19,20. It suffers, however, from a lack of supporting biochemical evidence, and fails totally to account for the effect of environmental conditions on phenotypic variability.
Looking forwards, looking back Could it be that Hsp90-like signal-transduction chaperones play a central role in homeostasis, and function best under those genetic and environmental conditions to which the species is adapted? Major loss of function owing to mutation or drug treatment will expose sufficient of the underlying genetic variation to produce visible abnormalities; lesser impairment of function due to suboptimal genetic or environmental conditions could expose just enough of the underlying genetic variation to increase phenotypic variability of many quantitative characters. Alternatively, minor impairment of function might simply reduce the buffering action against ‘developmental noise’ caused by random micro-environmental effects, with little or no genetic basis (a possibility considered by Rutherford and Lindquist, but ruled out in the context of major impairment of Hsp90 function by their experimental evidence). More than two thirds of the following references predate 1960. They describe experimental findings that tend to be overlooked today because they do not fit comfortably into the conventional framework of molecular development. The work of Rutherford and Lindquist points to ways in which this framework could usefully be extended.
Acknowledgement I am grateful to the Wellcome Trust for financial support.
11 Livesay, E.A. (1930) An experimental study of hybrid vigor or heterosis in rats. Genetics 15, 17–54 12 McLaren, A. and Michie, D. (1956) Variability of response in experimental animals: A comparison of the reactions of inbred, F1 hybrid and random bred mice to a narcotic drug. J. Genet. 54, 440–455 13 Michie, D. (1955) Towards uniformity in experimental animals. Coll. Pap. Lab. Anim. Bur. 3, 37–47 14 McLaren, A. and Michie, D. (1954) Are inbred strains suitable for bioassay? Nature 173, 686–688 15 Biggers, J.D. and Claringbold, P.J. (1954) Why use inbred lines? Nature 174, 596 16 Ashoub, M.R. et al. (1958) The effect of the environment on phenotypic variability. Proc. R. Soc. London Ser. B 148, 192–203 17 Went, F.W. (1953) Gene action in relation to growth and development. I. Phenotypic variability. Proc. Natl. Acad. Sci. U. S. A. 39, 839–848 18 Robertson, F.W. and Reeve, E.C.R. (1952) Heterozygosity, environmental variation and heterosis. Nature 170, 286 19 Mather, K. (1953) Genetic control of stability in development. Heredity 7, 297–336 20 Lerner, I.M. (1954) Genetic Homeostasis. Oliver and Boyd 21 Durrant, A. (1962) The environmental induction of heritable change in Linum. Heredity 17, 27–61
TIG May 1999, volume 15, No. 5
171