Second draft (G 335), 11/9/97

Written by Hongwei Qian

Genotype-Environment Interaction (G X E Interaction)

G X E interaction and its biological significance

        G (X E interaction mean that a specific difference of environment does not have the same
effects on a different genotype. A specific difference environment may have a greater effects on
some genotypes than others. In other words, genotype A might be superior to genotype B in
environment X, but inferior in environment Y (Falconer and Mackay, 1996). Historically,
Falconer (1952) first suggested that a character or a trait expressed in two environments can be
recognized as two characters which are genetically correlated. That is, the degree of a given trait
expressed in each environment is variable. The extent of the expression for a given trait reflects its
states in a given environment. For example, they can view body weight in one environment and
body weight in another environment as two genetically correlated states. Some G X E interactions
are due to difference of sensitivity of a genotype is called as its "reaction norm" (Schmalhausen,
1994). The alteration of a phenotype of an individual organism in response to environmental
conditions is known as phenotypic plasticity (Gause, 1947). The phenotypic plasticity reflects the
degree to which the phenotypic expression of a genotype varies under different environmental
conditions (Sultan, 1987).
        Bradshaw (1965) noted that the plasticity of a trait is under genetic control, and it has
evolved in some cases in relation to specific environmental influences. Phenotypic plasticity is the
differential response of a given genotype to dissimilar environment (Rawson, 1991). The
evolution of favored plasticity is both the process of disruption selection in space and time which
stabilize selection on different genotypes in a given environment and strong directional selection on
a trait for which genetic variation is not available. G X E interaction may play a crucial role in
promoting phenotype evolution. If a relative performance and rank order of a genotype depend on
its interaction with a specific environment, no single genotype is supposed to be of ubiquitously
better performance in all environments. Selection will favor those genotypes which have a higher
fitness in each environmental situation if the environment is at a low gene flow. So G X E
interaction is important in the maintenance of Polygonia variances. When gene flow is higher, G X
E interaction may not only promote but also constrain the evolution of plasticity. Variation among
some genotypes in phenotypic plasticity represents G X E interaction (Rawson, 1991). G X E
interaction affects the evolution of the mean phenotype in a specific environment. The extent of G
X E interaction also influences both the equilibrium genetic variability and evolutionary dynamics
of the additive genetic variance and covariance (Via, 1987).

Analysis of G X E interactions

        A standard method is to raise various genotypes in a set of environment and then compare
genotypes with respect to their patterns of phenotypic response (phenotypic plasticity). The G X E
interaction can then be analyzed through a non quantitative method in which response is plotted
against each particular environment, or through a quantitative method which basically relies on the
analysis-of-variance (ANOVA) approach, with the analysis of the coefficient of variation (CV)
(Schlichting, 1986).         Analyzes of a phenotypic variance underlie the estimating G X E interaction. Based on
original ANOVA, several analytic models have also been established. Among them, the regression
method is the most favored, because it divided the interaction into several parts, thus allowing
further to investigate how much association between varietal environmental differences and general
plasticity (Freeman, 1973). The amount of variation is measured and expressed as the variance.
The phenotypic variance is the sum of the variance of genotype values, the environmental variance,
and G X E interaction variance. Two-way ANOVA can be used to estimate the variance between
the specific environments, and variance attributable to G X E interaction. If there is no G X E
interaction, the best genotype in one environment will be the best in all. However, if the interaction
is not present, then particular genotypes must be better in particular environments (Falconer and
Mackay, 1996).
        On the other hand, analysis of genetic correlation is also being applied to decide G X E
interaction. Genetic correlations due to pleiotropy show the influence of a common set of genes on
two characters. If many same genes influence a character in two environments, it is expected that
the genetic correlation between the phenotype in two environments to be highly either positive or
large and negative due to antagenistic pleiotropy. If different genes determine the phenotype in each
environment, the genetic correlation across environments which is much less than one suggests the
existence of G X E interaction (Via, 1984).

G X E interactions in plants and animals

        G X E interactions are well documented in biological studies of plants and animals. Mather
and Jinks (1982) observed that ten inbred lines of the tobacco plant
Nicotiana rusticagrown in
each of eight specific environments resulted in the difference in the date of sowing and in the
densities of planting. An analysis of variances showed that the differences between the genotypes
(line) and between the environments was significant and that was a significant G X E interaction.
Evans and Hughes (1961) reported that the variety of morphological and physiological plasticity is
related to the variation of light availability. Plants with "shade" phenotypes have relatively low
growth rates than "sun" plant under high light conditions and vice versa. That is, plants from
shaded population acted better in low light conditions than plants from populations exposed to full
sun, and the reverse situations occurred in high light. The study of genetic variation in larval
fitness, traits in pest species,
liriomyza sativae, has shown the genetic correlation of
development time across hosts, but not pupal. It has also been shown that the existence of G X E
interaction in development time when sib lings were reared on two crops, cowpea and tomato (Via,
Plantago lanceolata grown in four treatments showed marked changes in the correlation
structure of traits, such as growth, vegetativeand reproductive characters. In one population, the correlation between reproductive effect and
total weight ranged from 0.80 to -0.39. This heterogeneity among treatments is statistically
significant for five of eight populations (Antonovics et al., 1982). G X E interactions were also
found for density effects on plants and animals (Gupter and Lewonti, 1982; Show, 1986), nutrient
levels on plants (Zuberi and Gale, 1976), and temperature variation on animal (Coyne and
Beecham, 1987).

G X E interactions in cancer etiology

        It has been wide agreement that most cancer results from man-made and natural
environmental exposure such as tobacco smoke; chemical pollutants in air, water food, drug;
radiation; dietary constituents; and infectious agents (virus) (Tomatia et al., 1987). The
carcinogenesis is known as a multistage process in which an orderly progression of the cell
through four distinct stages: initiation by exposure to genotoxic agents, promotion by agents that
stimulated the initiatal cell to proliferate and expand colonially to form a benign tumor, conversion
by genotoxic agent that further stimulate malignancy to growth and subsequent metastasises
(Harris, 1985). Cancer risk from environment not only depends on chemicals such as a
carcinogen itself, but also strongly on many other factors, including genetics, age, ethnicity, sex,
immune function, pre-existing disease, and level of nutrition (Venitt, 1994).
        Genetics modulates the cancer risk among individuals through individuals' response to
environmental exposures. This is the case of genotype-environment interactions in cancer
One of examples is the cancer risk is associated with the variation of a carcinogen metabolism due
to genetic polymorphism of related enzymes. The genetic traits that substantially control the
metabolic activation or detoxification of carcinogenic chemicals to their DNA - damaging
intermediates seem important determinants of risk (Nebert et al.; Poulsen et al. 1993). The
difference of genetic traits dramatically varies among individuals in their ability to metabolize
endogenous and exogenous carcinogens (Taylor, 1990).
        The superfamily of cytochrome (CYP) P450 consists of Phase I and Phase II enzymes.
Phase I enzymes catalyze the oxidative metabolism of most endogenous chemicals (e.g., steroids,
fatty acids) and exogenous chemicals from environmental contaminants such as PAHs and
aromatic amines. While this housekeeping process converts chemicals to water-soluble, readily
excretable forms, it also creates high-energy electrophilic intermediates which directly interact with
DNA. Most genotoxic chemical carcinogens are not intrinsically reactive but require this metabolic
conversion to final DNA-damaging intermediates. In contrast to the phase I P450-activating
enzymes, phaseII enzymes ( epoxied hydrolase, glutathione s-transferase, N-acetytransferase and
sulfortransferase) generally detoxify carcinogenic metabolites by conjugating them with
glucuronide, glutathione, or sulfate to produce hydrophilic products that are readily excreted
(Miller et al., 1978; Harris, 1991). The balance between phase I and phase II enzymes determine
the molecular dose of carcinogens, thereby substantially influencing cancer risk.
        The evidence that variations in chemical metabolism cause an increased cancer risks
through genotype-environment interactions have been accumulated in several common sites such as
lung, bladder et al. Many of P450 genes are known to exist in variant forms or polymorphism that
has differing activities. For P-450 phase I gene, it has CYA1A1, CYP1A2, CYP2D6 and CYP3A4
etc. (Nebert, 1991). P450 CYP1A1 catalyzes the oxygenation of PAHs such as BP, thereby
producing certain metabolites that are highly reactive with DNA. CYP1A1 is induced by cigarette
smoke and environmental chemicals, such as dioxin and PAAs (Nebert et al., 1991; Whyatt et al.
1995). Individuals vary widely in the extent to which CYP1A1 induced by such exposures, with
10% of the Caucasian population being highly inducible (Nebert et al., 1991). The level of
CYP1A1 inducibility varies by 20-fold in human liver (Schweikl et al., 1993), while activity of the
CYP1A1 enzyme differ by more than 50-fold in human lung tissues (Petruzzelli et al., 1988). A
number of molecular epidemiologic studies have convincingly associated increased risk of lung
cancer in smokers with high CYP1A1 inducibility (Kellermann et al., 1973; Koun et al., 1982).         In Japanese, a threefold increase in lung cancer risk was correlated with a polymorphism in
the CYP1A1 gene, called Msp I (120), with the greatest lung cancer risk in light smokers. The
Msp I polymorphism is itself linked to an exon 7 mutation in the CYP1A1 gene, resulting in an
altered protein having the amino acid valine in place of isoleucine. Experimentally, the valine-type
protein is almost twice as active in metabolizing BP as the isoleucine form. It implicate that this
variant affects risk (Nakachi et al., 1991; Kawajlri et al., 1993). Cigarette smokers of one to two
packs per day with the exon 7 (Ile-Val) genotype had significantly ( more than two-fold) higher
levels of PAH-DNA adducts in white blood cells than smokers without the CYP1A1 variant,
suggesting that this genetic trait may increase cancer risk by facilitating PAH activation and binding
to DNA (Mooney et al., 1995).
        The cytochrome CYP2D6, another CYP Phase I enzyme, is important in the metabolism of
a variety of drugs including antidepressants, beta blockers etc. Although a major role in catabolism
of genotoxic substances has not been established, CYP2D6 has been noted to activate 4-
(methylnitrosoamino)-1-(3-pyridyl)-1-butanone, a nitroso ketone in tobacco smoke. Unlike
CYP1A1, CYP2D6 is not inducible, but individuals vary greatly in their capacity. The poor
metabolite phenotype have two mutant alleles, neither of which expresses normal CYP2D6.
Heterozygotes for mutant alleles are intermediate metabolizers, and homozygous wild-type
individuals are extensive metabolizers. A summary analysis of seven independent studies revealed
an aggregate odds ratio estimate of 2.3 comparing lung cancer risk among extensive or intermediate
metabolite versus risk among poor metabolizers (Amos et al., 1992).
        The similar situations were found in CYP Phase II enzymes. Glutathione-s-Transferases
(GST) is one of the P-450 phase II enzymes. Generally, expression of GSTs varies by tissue and
is also inducible. GSTM1, one of GST, detoxifies a number of reactive, electrophilic substance,
including the carcinogens PAHs, ethylene oxide, and styrene. Approximately 50 % Caucasians
have a deletion in the GSTM1 (Seidegard et al., 1990). An study indicates that persons with the
null genotype had little risk of bladder cancer in the absence of environmental exposure. With
smoking as much as 25% of all bladder cancer may be attributable to the GST null genotype.
Furthermore, risk increased by almost twofold with any exposure to tobacco smoke and by sixfold
in heavy smokers (Æ 50 pack-years) compared with unexposed persons without deletion (Bell et
al., 1993).
        A second phase II enzyme, N-acetyltransferase 2 ( NAT2), is a noninducible liver enzyme
that detoxifies a wide range drugs and carcinogenic aromatic amines via N-acetylation (Hein,
1988). Based on the drug metabolic rate, NAT2 has three variants, including slow , intermediate,
and fast form (Butler et al., 1986). 50%-60% of Caucasians and 30%-40% African American are
slow acetylators, (Bell et al., 1993; Yu et al., 1994). One study indicate that NAT2 slow
acetylators are at higher risk of bladder cancer due to environmental exposure such as 2-
naphthylamine and 4-ABP (Vineis et al., 1995).
  A recent study (Ambrosone et al., 1995) indicates that women who smoke and have a slow
acetylators may be eight times more likely to develop breast cancer than women who do not
smoke. While NAT2 is involved in detoxification of aromatic amine bladder carcinogens, it
apparently plays an opposite role in colon cancer through the metabolic activation of the
carcinogenic heterocyclic amines. Thus, persons who are NAT2 rapid acetylators appear to be at
increased risk of colon cancer(Kadlubar et al., 1992).

Advance and Perspectives

        Quantitative genetics is an exciting research area. Its theory and analytic system is still
developing............. (I am still working on this point).
        The quantitative genetics combined with molecular biology described above provide new
evidence that environmental factors are major contributors to human cancer and that their risks are
strongly influence by genetic and acquired susceptibility. That is, genotype-environment interaction
play a crucial role in human cancer etiology. It indicates substantial individual variability in biologicresponse to carcinogens and suggest that those with predisposing genetic traits are likely to have
greater risk from selected exposure tan other members of the population.
        Besides the genes related to metabolism and detoxification the cancer susceptibility gene
may also include and those associated with DNA repair, cell cycle , signaling molecular receptor
ligand. However, the role of the interaction of those genotypes with environmental factors in
human cancer is still less known. Therefore, the future studies on genotype-environment
interaction in human cancer should emphases on the identification of the cancer susceptibility
genes mentioned above. This will further provide major new biological insights into the regulation
of cell growth and maintenance of orderly cell division, thus eventually provide new approaches to
cancer detection and prevention.


Ambrosone CB, Freudenheim JL, Graham S, Marshall JR, Vene JE, Brasure JR (1995):
        Cytochrome P450 A1 and gluathione S-transferase (M1) genetic polymorphism and
        postmenopausal breast cancer risk. Cancer Res 55:3483-3485.
Amos CI, Rubin LA (1995): Major gene analysis for disease and disorder of complex etiology.
        Exp Clin Immunogener. 12:141-155.
Antonovics J, Primack RB (1982):Experimental ecological genetics in
plantago. VI. The
        demogarphy of seeding transplants of
P. lanceolata. J Ecol 70:55-75.
Brawshaw AD (1965): Evolutionary significance of phenotypic plasticity in plants. Adv Genet
Bell DA, Taylor JA, Butler MA, Stephens EA, Wiest J, Brubaker LH (1993): Genotype/phenotype
        discordance for arylamine N-acetyltransferase (NAT2) reveals a new slow-acetylator allele
        common in African-Americans. Carcinogenesis 14:1689-1692.
Bell DA, Taylor JA, Paulson DF, Robertson CN, Mohler JL, Lucier GW (1993): Genetic risk and
        carcinogen exposure: a common inherited defect of the carcinogen-metabolism gene
        glutathione S-transferase M1 (GSTM1) that increase susceptibility to bladder cancer. J Natl
        Cancer Inst 85:1159-1164.
Butler MA, Iwasaki M, Guengerich P,Kadlubar FF (1986): Human cytochrome P-450 PA
        (P450IA2), the phenacetin-deethylase, is primarily responsible for the hepatic 3-
        demethylation of caffeine and the N-oxidation of carcinogenic arylamine. Proc Ntal Acad
        Sci USA 86:7696-7700.
Coyne JA, Beecham E (1987): Heritability of two morphological characters within and among
        natural populations of
Drosophila melanogaster. Genetics 117:727-737.
Evans GC, Hughes AP (1961): Plant growth and the aerial environment. I. Effect of artificial
        shading on
Impatiens parviflora. New phytologist 60:150-180.
Falconer DS (1952): The problem of environment and selection. Amer Natur 86:293-298.
Falconer DS, Mackay TFC (1996): "Introduction to Quantitative Genetics." Longman P.132-133.
Freeman GH (1973): Statistical methods for the analysis of genotype-environment interactions.
        Heredity 31:339-354.
Gause GF (1947): Problems of evolution. Trans Coon Acad Sci 37:17-68.
Gupta AG, Lewontin RC (1982): A study of the reaction norms in natural populations of
        Drosophila pseudo obscura. Evolution 36:934-948.
Harris CC (1985): Future directions in the use of DNA adducts as internal dosimeters for
        monitoring human exposure to environmental mutagens and carcinogens. Environ Health
        Perspect 62:185-191.
Harris CC (1991): Chemical and physical carcinogenesis: advances and perspectives for the 1990s.
        Cancer Res (18 suppl) 51:5023-5044.
Hein DW (1988): Acerylator genotype and arylamine-induced carcinogenesis. Biochim Biophys
        Acta 948:37-66.Kadlubar FF, Butler MA, Kaderlik KR, Chou HC, Lang NP (1992): Polymorphism for aromatic
        amine metabolism in humans: relevance for human carcinogenesis. Environ Health
        Perspect 98:69-74.
Kawajlrl K, Nakachi K, Imai K, Watanabe J, Hayashi S. (1993): The CYP1A1 gene and cancer
        susceptibility. Crit Rev Oncol Hematol 14:77-87.
Kellermann G, Shaw CR, Luyten-Kellerman M (1973): Aryl hydrocarbon hydroxylase inducibility
        and bronchogenic carcinoma. N Engl J Med 289:934-937.
Koun RE, McKinney CE, Slomiany DJ, Snodgrass DR, Wray NP, McLemore TL (1982):
        Positive correlation between high aryl hydroxylase activity and primary lung cancer as
        analyzed in cryopreserved lymphocyte. Cancer Res 42:5030-5037.
Mather K, Jinks JL (1977): Introduction to Biometrical Genetics. Chapman and Hall
        London.Miller BC (1978): Some current perspectives on chemical carcinogenesis in human
        and experimental animals: Presidential Address. Cancer Res 38:1479-1496.
Mouney LA, Santella RM, Covey L, Jeffrey AM (1995): Decline in DNA damage and other
        biomarkers in peripheral blood following smoking cessation. Cancer Epidemiol
        Biomarkers Prev 4:627-634.
Nakachi K, Imai K, Hayashi S, Watanabe J, Kawajiri K (1991): Genetic susceptibility to
        cell carcinoma of the lung in relation to cigarette smoking dose. Cancer Res 51:5177-5180.
Nebert DW (1991): Role of genetics and drug metabolism in human cancer risk. Mutat Res
Petruzzelli S, Camus AM, Carrozzi L, Chelarducci L, Rindi M, Menconi G (1988): Long-lasting
        effects of tobacco smoking on pulmonary drug-metabolizing enzymes: a case-control study
        on lung cancer patients. Cancer Res 48:4695-4700.
Poulsen HE, Loft S, Wassermann L (1993): Cancer risk related to genetic polymorphism in
        carcinogen metabolism and DNA repair. Pharmacol Toxicol 72:93-103.
Rawson PD, Hilbish TJ (1991): Genotype-environment interaction for juvenile growth in the hard
mercenaria mercenaria (L). Evolution 45:1924-1935.
Schmalhausen II (1949): Factors of evolution: The theory of stabilizing selection. Blakiston,
        Philadelphia, PA.
Schweikl H, Taylor JA, Kitareewan S, Linko P, Nagorney D, Goldatein JA (1993): Expression of
        CYP1A1 and CYP1A2 genes in human liver. Pharmacogenetics 3:239-49.
Schlichting CD (1986): The evolution of phenotypic plasticity in plants. Ann Rev Ecol syst
Seidegard J, Pero RW, Markowitz MM, Roush G, Miller DG, Beattie EJ (1990): Isoenzyme(s) of
        glutathione transferase (class mu) as a marker for the susceptibility to lung cancer, a
        follow-study. Carcinogenesis 11:33-36.
Shaw RG (1986): Response to density in a wild population of the perennial herb
Salvia lyrata:
        Variation among families. Evolution 40:492-505.
Sultan SE (1987): Evolutionary implications of phenotypic plasticity in plants. Evol Biol 21:127-
Taylor JA (1990): Epidemiologic evidence of genetic susceptibility to cancer. In: Spatz L, Bloom
        AD, Paul NW, eds. Detection of Cancer Predisposition: Laboratory Approaches. White
        Plains, NY, P113-127.
Tomatia L, Aitio A. Wllbourn J, Shuker L (1987): Human carcinogens so far identified. Jpn J
        Cancer Res 78:887-898.
Via S (1984): The quantitative genetics of polyphagy in an inset herbivore. II. Genetic correlations
        in larval performance within and among host plants. Evolution 38:896-903.
Via S, Russellande (1987): Evolution of genetic variability in a spatially heterogeneous
        environment: effects of genotype-environment interaction. Genet Res 49:147-156.
Vineis P, Caporaso N (1995): Tobacco and cancer: epidemiology and the laboratory. Environ
        Health Perspect 101:156-160.Venitt S (1994): Mechanism of carcinogenesis and individual susceptibility to cancer. Clin Chem
Whyatt RM, Garte SJ, Cosma G, Bell DA, Jedrychowski W, Wahrendorf J (1995): CYP1A1
        messenger RNA levels in placental tissues as a biomarker of environmental exposure.
        Cancer Epidemiol Biomarkers Prev 4:147-153.
Yu MC, Skipper PL, Taghizadeh K, Tannenbaum SR, Chan KK, Henderson BE (1994):
        Acetylator phenotype, aminoblpheny-hemoglobin adduct levels, and bladder cancer risk in
        whith, black and Asian men in Los Angeles, California. J Natl Cancer Inst 86:712-716.
Zuberi MI and Gale JS (1976): Variation in wild populations of
pavaer dubium. Heredity
        36:359- 388.

Comments about this draft paper or any of the papers submitted for this class should be sent to the instructor, Dr. Garbutt, using the form that can be accessed through button above.

Back to the Population and Quantitative Genetics Page