Epistasis: Gene Interaction and Phenotype Effects

In the first decade of the twentieth century, British geneticists William Bateson and R. C. Punnett conducted research showing that the shape of the comb in chickens was caused by the interaction between two different genes. Bateson and Punnett were aware of the fact that different varieties of chickens possess distinctive combs. For instance, Wyandottes have a "rose" comb, Brahmas have a "pea" comb, and Leghorns have a "single" comb. When Bateson and Punnett crossed a Wynadotte chicken with a Brahma chicken, all of the F₁ progeny had a new type of comb, which the duo termed a "walnut" comb. In this case, neither the rose comb of the Wyandotte nor the pea comb of the Brahma appeared to be dominant, because the F₁ offspring had their own unique phenotype. Moreover, when two of these F₁ progeny were crossed with each other, some of the members of the resulting F₂ generation had walnut combs, some had rose combs, some had pea combs, and some had a single comb, like that seen in Leghorns (Figure 1). Because the four comb shapes appeared in a 9:3:3:1 ratio (i.e., nine walnut chickens per every three rose chickens per every three pea chickens per every one single-comb chicken), it seemed that two different genes must play a role in comb shape.

Through continued research, Bateson and Punnett deduced that Wyandotte (rose-combed) chickens must have the genotype RRpp, while Brahma chickens must have the genotype rrPP. A cross between a Wyandotte and a Brahma would yield offspring that all had the RrPp genotype, which manifested as the walnut-comb phenotype. Indeed, any chicken with at least one rose-comb allele (R) and one pea-comb allele (P) would have a walnut comb. Thus, when two F₁ walnut chickens were crossed, the resulting F₂ generation would yield rose-comb chickens (R_pp), pea-comb chickens (rrP_), and walnut-comb chickens (R_P_), as well as chickens with a new, fourth phenotype—the single-comb phenotype. Based on the process of elimination, it could be assumed that these single-comb chickens had the rrpp genotype (Bateson & Punnett, 1905; 1906; 1908).

Bateson and Punnett next performed a set of experiments in peas that also showed gene-gene interaction. The duo opted to use peas because it is relatively easy to perform hybrid crosses with these plants, and they chose to focus on the characteristic of flower color.

Bateson and Punnett began by crossing two varieties of pea, each of which was pure-breeding for white flowers. This cross yielded an F₁ generation in which all progeny had purple flowers. Next, two F₁ plants were crossed to create the F₂ generation. In this generation, Bateson and Punnett counted a total of 382 purple-flowered plants and 269 white-flowered plants. The ratio of purple flowers to white flowers was thus 9.4:6.6, or approximately 9:7.

What could explain this variation from Mendelian ratios? Bateson set out to answer this question in a 1909 report, in which he first proposed what he called the ability of one "allelomorphic pair" (pair of gene alleles) to mask the affects of the alleles for another gene. To rephrase this in terms of Bateson and Punnett's pea experiment, it seemed that two recessive alleles at one flower locus could mask the effects of the alleles at the other flower locus. Let's designate the first locus as the C locus, and the second as the P locus. If Bateson's theory held true, it meant that any flower with the cc genotype would be white, no matter what alleles were present at its P locus. Similarly, any flower with the pp genotype would also be white, no matter what alleles were present at its C locus. Bateson later used the word "epistasis," which translates as "standing upon," to define the masking action of one gene by another. (Since then, scientists have come to understand that genes can interact in more ways than just masking.)

Many years after Bateson first described this 9:7 phenotypic ratio in pea plants, researchers were finally able to determine the two genes responsible for it (Dooner et al., 1991). These genes control flower color by controlling pea plant biochemistry, in particular that related to pigment compounds called anthocyanins. In peas, there is a two-step chemical reaction that forms anthocyanins; gene C is responsible for the first step, and gene P is responsible for the second (Figure 2). If either step is nonfunctional, then no purple pigment is produced, and the affected pea plant bears only white flowers. The dominant C and P alleles code for functional steps in anthocyanin production, whereas the recessive c and p alleles code for nonfunctional steps. Thus, if two recessive alleles occur for either gene, white flowers will result.

Table 2 shows in detail how the 9:7 ratio is a modification of phenotypic but not genotypic Mendelian ratios. Note that the C and P genes independently assort, and remember that the presence of a recessive genotype at one locus (i.e., cc or pp) masks the effects of the alleles at the other locus. Note also that there are nine combinations of alleles in the F₁ generation that feature at least one dominant C and one dominant P allele, which would yield a purple flower phenotype (indicated within the table by purple shading). Conversely, there are seven combinations that result in either a cc or a pp, which would yield the white flower phenotype-hence, the 9:7 ratio of purple to white flowers.

Table 2: Results of the Cross Between Two Pea Plants with Genotype CcPp

		Female Gametes
		CP	Cp	cP	cp
Male Gametes	CP	CCPP	CCPp	CcPP	CcPp
	Cp	CCPp	CCpp	CcPp	Ccpp
	cP	CcPP	CcPp	ccPP	ccPp
	cp	CcPp	Ccpp	ccPp	ccpp

Later researchers discovered that flower petal color can also be controlled by dominant epistatic forces. For instance, in the Primula plant, the pigment malvidin creates blue-colored flowers. Synthesis of malvidin is controlled by gene K, yet production of this pigment can be suppressed by gene D, which is found at completely different locus (Figure 3). In this case, the D allele is dominant to the K allele, so plants with the genotype KkDd will not produce malvidin because of the presence of the D allele.

So, if two plants with genotype KkDd are crossed with each other, what is the ratio of blue offspring to nonblue offspring? The results of such a cross are detailed in Table 3.

Table 3: Results of the Cross Between Two Primula Plants with Genotype KdDd

		Female Gametes
		KD	Kd	kD	kd
Male Gametes	KD	KKDD	KKDd	KkDD	KkDd
	Kd	KKDd	KKdd	KkDd	Kkdd
	kD	KkDD	KkDd	kkDD	kkDd
	kd	KkDd	Kkdd	kkDd	kkdd

In Table 3, all of the shaded boxes contain genotypes that indicate an absence of malvidin. In particular, the yellow boxes represent genotypes that feature at least one D allele, and the presence of the D allele suppresses the production of malvidin. Meanwhile, the light turquoise box represents a genotype with no malvidin suppression (dd), but also no malvidin production (kk). On the other hand, the three unshaded/white boxes are the only genotypes that permit the production of malvidin, which means that plants with these phenotypes bear blue flowers. The phenotypic ratio is therefore 13:3. This type of epistasis is sometimes called dominant suppression, because the deviation from 9:3:3:1 is caused by a single allele that produces a dominant phenotype, and the action of this allele is to suppress the expression of some other gene.

Genes don't always have to act in opposition to each other for an interaction to be epistatic, however. Sometimes, two genes that each have the same role in protein production can substitute for each other.

The mechanism by which wheat kernel color is determined is an example of duplicate gene action. In wheat, kernel color is dependent upon a biochemical reaction that converts a precursor substance into a pigment, and this reaction can be performed with the product of either gene A or gene B (Figure 4). Thus, having either an A allele or a B allele produces color in the kernel, but a lack of either allele will produce a white kernel that is devoid of color.

Table 4 depicts the results of a dihybrid cross between two plants with genotype AaBb. Note that among the offspring of this cross, the phenotype is nearly uniform.

Table 4: Results of the Cross Between Two Wheat Plants with Genotype AaBb

		Female Gametes
		AB	Ab	aB	ab
Male Gametes	AB	AABB	AABb	AaBB	AaBb
	Ab	AABb	AAbb	AaBb	Aabb
	aB	AaBB	AaBb	aaBB	aaBb
	ab	AaBb	Aabb	aaBb	aabb

In this cross, whenever a dominant allele is present at either locus, the biochemical conversion occurs, and a colored kernel results. Thus, only the double homozygous recessive genotype produces a phenotype with no color, and the resulting phenotypic ratio of color to noncolor is 15:1.

In the malvidin example in Table 3, gene D is epistatic to gene K because one allelic pair masks the expression of an allele at the second locus. In the case of duplicate gene action, as in Table 4, the outcome is less variable, but it is still derived from multiple gene interactions. Here, if wheat kernel color is controlled by genes A and B, then A is epistatic to allele B or allele b, and B is epistatic to allele A or allele a.

Yet another type of epistasis occurs when one gene interacts with another to modify—but not mask—a phenotype. For example, in horses, the extension gene determines whether an animal's coat color will be red or black; here, the dominant allele E produces black pigment in the coat, while the recessive allele e produces red pigment. All horses with genotype ee are therefore red, yet there are many different types of red horses. These differences exist because of the action of epistatic modifier genes.

One such modifier gene is called cream dilution. The cream dilution gene has two alleles: C^Cr and C. The C^Cr allele is semidominant; it dilutes red to yellow in the heterozygous state and red to pale cream in the homozygous state. On the other hand, the C allele has no diluting effect on coat color. Thus, horses with genotype eeCC are chestnut colored, and they have reddish-brown coats, tails, and manes. In contrast, horses with one copy of the C^CR allele (genotype eeCC^CR) are palomino (i.e., they have a gold coat with a white mane and tail), while horses with two copies of the C^CR allele (genotype ee C^CRC^CR) are cremello (i.e., basically white or cream colored).