Trends in Genetics 2: 307–309 (1986)
Phenotypic consequences of environmentally induced changes in plant DNA

C. A. Cullis


Rapid genomic changes accompanied by heritable phenotypic changes can be induced in flax by environmental stress. The DNA changes are confined to a specific subset of the genome and may affect the phenotype by virtue of their position in the genome rather than by their specific nucleotide sequence.

The genetic behaviour observed in plants can only be considered appropriately in the context of the organization of their life cycle. The higher plant can be considered as an assemblage of multiple competing units (meristems), each of which is capable of reproduction. Control over which of the units dominates the developmental form is exercised through apical dominance. The release of alternative units from this dominance usually follows damage to the dominant meristem. Thus in plants (and colonial animals) there is the possibility that since there is no clear separation of the soma and germ line, genetic variation arising through a mutation in any somatic cell has the potential to be transmitted to the next generation. The consequences of this form of life strategy have been reviewed recently1,2.

It is now becoming clear that the plant genome is in a dynamic state and that a range of phenomena can be responsible for rapid changes. These phenomena include the activity of transposable elements, and amplification and deletion events. There is also evidence that the processes which cause the restructuring of the genome can be activated under certain conditions, while remaining quiescent at other times. Conditions that appear to increase the rate of variation include disruption of the chromosomes3, the formation of particular F1 hybrids4 and the external environment in which growth takes place5.

Environmental induction of heritable changes in flax

One of the most extensively studied systems is that in which heritable changes can be induced by the environment in certain flax varieties5,6. The initial observations were made by Durrant, who grew the flax variety Stormont Cirrus in media containing the nutrients nitrogen, potassium and phosphorus in a number of different concentrations and combinations. Flax is normally an inbreeder, so the individuals should be genetically homogeneous and phenotypically very similar if not identical. The first generation of growth under the different nutrient regimes resulted in plants that were different from one another; this was not surprising, since growth and development are obviously affected by the environment. However, a surprising event was observed in the subsequent generations, when seeds obtained from self-fertilized plants grown under various nutrient regimes were compared in a common environment. The progeny of plants from any one treatment were similar, as expected, but differed from the progeny of plants from other nutrient regimes. These altered characteristics were passed to subsequent generations and could be stably inherited, irrespective of the nutrient regimes applied in these following generations. Thus, in a single generation, there was an alteration which was heritable and could not be reversed by restoration of the original conditions.

The genetically different types derived from these experiments were termed genotrophs and could differ markedly from one another. Examples of the extreme types produced are shown in Fig. 1. In addition to the differences in height and weight, the latter being the character initially used by Durrant to differentiate the genotrophs7, the large and small types differed from one another in a number of other characters. Among these were the total nuclear DNA content as determined by Feulgen staining8, the number of hairs on the false septa of the seed capsules9 and the isozyme bands of peroxidase activity10,11. In the latter two characters the changes occurring involved the change from a dominant to a recessive condition in a single generation and each occurred in a different genotroph. Thus the hair number changed from about 60 in the original line to zero in the large genotroph (hairy dominant to hairless), while in the small genotroph the isozyme band pattern observed for the peroxidase activity in roots was recessive to that found in the original line. Thus dominant to recessive changes could be found in both extreme forms after the induction of changes.

Fig. 1. Fourth generation of two extreme flax phenotypes induced by fertilizer treatments. Reprinted, with permission, from Ref. 5.

Genomic changes

The DNA changes occurring in the genotrophs have also been characterized. The difference in the amount of total nuclear DNA between the extreme large and small genotrophs can be up to 16% (Ref. 8). The variation appeared to be mainly in the highly repetitive and intermediately repetitive fractions of the genome12. The flax genome has been extensively characterized and representatives from all the highly repeated sequence families have been cloned12. It was found that variation could occur in all but one of these families and that each family could vary independently. Included in the variable repeated sequence families were the DNAs coding for the 25S, 18S and 5S ribosomal RNAs.

There also appeared to be some heterogeneity within a family of repeated sequences with respect to the variation. Thus for three families, one of which encodes the 5S RNA, a specific subset could be identified that was particularly labile, and whenever variation occurred it took place in that subset12-14. One example is shown in Fig. 2, where the total amount of the sequence differs twofold between the lines being compared. The four bands arrowed are reduced in the low DNA line and each time a reduction occurs in this sequence it is these four bands that are preferentially reduced.

None of the intermediately repetitive sequences that vary have been identified, although some dramatic amplifications can be observed. One of these is shown in Fig. 3. The arrowed bands are present in one of the lines and absent from the parent line. The sequence present in these bands is not from any of the highly repeated sequence families and appears to represent a significant amplification of a sequence that was previously intermediately repeated or present at low copy number.

How does environmental stress induce genetic variation?

Most of the comparisons have been made between lines in which stable changes have occurred. Analysis of total nuclear DNA or ribosomal DNA during growth of the original line showed that the changes occurred in response to stressful conditions8,15. Thus, under these conditions the plants appeared to be mosaics, with the cells in different parts of the plant having both qualitative and quantitative differences in DNA. The nuclear composition of the progeny then depended on the extent of variation which had occurred prior to flowering of the parent plant grown under inducing conditions.

Fig. 2. Equal amounts of DNA from a high DNA line (a, the original line) and a low DNA line (b, a small genotroph, C35) were digested with the restriction enzyme TaqI, separated on a 2% agarose gel and blotted onto Hybond nylon membrane. The filter was hybridized with labelled plasmid pDC7 (a highly repetitive sequence making up 3.2% of the flax genome12). The bands arrowed are underrepresented in the low DNA line. The sizes are given in kilobase pairs.

Fig. 3. Ethidium bromide stained gel of HaeIII digested DNAs from Stormont Cirrus (a) and a plant regenerated from callus derived from Stormont Cirrus (b) showing the appearance of three bands in the regenerated plant DNA which are absent in the original line under these conditions. These bands do not hybridize with any of the highly repeated families isolated from Stormont Cirrus. The sizes are in kilobase pairs.

Observations of plants growing under inducing conditions suggest that selection may be involved in the changes. Plants grow suboptimally under the most effective inducing conditions although all the individuals usually survive and contribute to the next generation. After some time of growth under inducing conditions one part of the plant starts to grow more rapidly and usually contributes maximally to the next generation. This can be a side shoot as shown in Fig. 4; a single shoot has grown out vigorously and all the seeds produced from the plant came from this shoot. The limited amount of material produced on such plants means that it has not yet been possible to demonstrate directly any DNA changes correlated with the increased vigour of the shoot.

Comparison with other plants

Is flax unique in its response to stress by altering its genome? It is well documented that the passage of plant cells through a cycle of tissue culture and regeneration can result in heritable changes. When flax tissue was cultured and plants regenerated, phenotypic and DNA changes were observed16. The changes observed in the regenerated families were similar to those found in the genotrophs derived from previous studies on whole plants grown in different environments. The qualitative changes in the DNA were also in the same subset of variable sequences delineated in the genotroph comparisons. Thus it is highly likely that these two sets of treatments cause the DNA variability by the same mechanisms.

Fig. 4. A Stormont Cirrus plant growing under inducing conditions (in an inert medium watered with 0.1 × Hoaglands solution every 2 weeks6, otherwise watered with tap water). Note the vigorous side shoot (arrowed) which grew out and flowered before the main stem. Although the main stem eventually flowered no seed was set from these flowers and so all the seed for the next generation were from the side shoot. The progeny from this plant were similar to the small genotroph.

However, it remains to be determined whether or not the somaclonal variation observed in other species is also the result of a similar mechanism. Two other examples of phenotypic changes in response to the environment have been described, in Nicotiana rustica17 and in pea18. In pea, the changes were not as stable as those observed for flax. It may be that stable changes are observed in flax because after an initial period of variation at high frequency the plant loses the ability to respond to further stress. Thus the prevailing phenotype remains and can be characterized. In other plant systems the ability to respond to stresses may be more constant and so any stably inherited induced variation is obscured by fluctuations occurring in successive generations.

No connection has been demonstrated so far between the changes in any one DNA sequence and a phenotypic character for the flax genotrophs. However, since the families of highly repetitive sequences in flax are all arranged in long tandem arrays, the deletion and amplification events found must affect these blocks.

Thus it is possible that the phenotypic changes are a consequence of alterations in gene control brought about by changes in chromosome structure that are mediated by these changes in blocks of tandem arrays (blocks of heterochromatin?). There are precedents for the alteration of gene expression and development by blocks of heterochromatin. One example is the pleiotropic effects associated with distal X heterochromatin in Drosophila19. Another example is the effect of telomeric heterochromatin on seed development in triticale20.


Much is still unknown about the system in flax, perhaps most importantly which gene(s) are responsible for controlling the responses and generating the genomic variation and how the environmental signal triggering such variation is perceived. In addition we still have to learn how widespread the phenomenon is within the plant kingdom and what its relationship is to other mechanisms that cause genomic changes, such as wide species crosses and stresses or shocks which activate transposable elements. Are all or many of these responses controlled by the same genetic system and therefore different manifestations of the same phenomenon, or are there many varied mechanisms by which the plant genome can be modified to generate new variants?


  1. Walbot, V. and Cullis, C. A. (1985) Annu. Rev. Plant Physiol. 36, 367-396
  2. Walbot, V. (1985) Trends Genet. 1, 165-169
  3. Dellaporta, S. L. and Chomet, P. S. (1985) in Genetic Flux in Plants (Plant Gene Research Vol. 2) (Hohn, B. and Dennis, E., eds), pp. 169-216, Springer-Verlag
  4. Gerstel, D. U. and Burns, J. A. (1976) Genetica 46, 139-153
  5. Durrant, A. (1962) Heredity 17, 27-61
  6. Cullis, C. A. (1981) Heredity 46, 286-297
  7. Durrant, A. (1959) New Sci 6, 293-296
  8. Evans, G. M., Durrant, A. and Rees, H. (1966) Nature 212, 687-689
  9. Durrant, A. and Nicholas, D. B. (1970) Heredity 25, 513-527
  10. Cullis, C. A. and Kolodynska, K. (1975) Biochem. Genet. 13, 687-697
  11. Fieldes, M. A. and Tyson, H. (1973) Can. J. Genet. Cytol. 15, 731-744
  12. Cullis, C. A. and Cleary, W. (1986) Can. J. Genet. Cytol. 28, 252-259
  13. Goldsbrough, P. B., Ellis, 1. H. N. and Cullis, C. A. (1981) Nucleic Acids Res. 9, 5895-5904
  14. Cullis, C. A. Am. Nat. (in press)
  15. Cullis, C. A. and Chariton, L. (1981) Plant. Sci. Lett. 20, 213-217
  16. Cullis, C. A. and Cleary, W. (1986) Can. J. Genet. Cytol. 28, 247-251
  17. Hill, J. (1965) Nature 207, 732-734
  18. Highkin, H. R. (1958) Am. J. Bot. 45, 626-631
  19. Hilliker, H. J. and Appels, R. (1982) Chromosoma (Berl.) 86, 469-490
  20. Gustafson, J. P. and Bennett, M. D. (1982) Can. J. Genet. Cytol. 24, 83-92

C. A. Cullis is at the Department of Biology, Case Western Reserve University, Cleveland, Ohio 44106, USA.