Pleiotropy occurs when one allele has more than one phenotypic effect.
Some clear examples of this phenomenon are the gene S in Nicotiana tabacum, which affects several structures, including the shapes of leaves, flowers and capsules (Fig. 3.2), and the colour gene Cin onions, which not only regulates colour but also determines resistance to the smudge fungus through the production of specific phenolic acids. The allele dlin the tomato (Lycopersicon esculentum) reduces the number of hairs on stems, peduncles and stamens. It also produces separated anthers, which diminish levels of self-pollination (Rick, 1947). The fact that these single gene changes influ- ence so many traits argues strongly that the genome must be interconnected.
As we shall discuss more fully in Chapter 6, a number of pleiotropic genes were important in the early domestication of beans and maize. The fin gene in dry beans conditions the earliness of flowering, and has significant effects on node number on stems, pod number and the number of days from flowering to fruiting (Koinange et al., 1996). Several pleiotropic quanti- tative trait loci (QTL) have been identified in Zea, including: (i) teosinte glume architecture 1(tb1), which affects internode lengths, inflorescence sex and structure; (ii) teosinte branched 1 (te1), which also affects internode lengths and numbers, and inflorescence sex; and (iii) suppressor of sessile spikelets 1(sos1), which affects branching in the inflorescence and the pres- ence of single vs. paired spikelets in the ear (Doebley et al., 1995).
In epistasis, the allelic constitution at one locus affects the level of expression of alleles at another locus, again illustrating the interactive nature of the genome. For example, the presence of prussic acid in clover requires a dominant allele at both of two loci. Bulb colour in onion is also regulated by alleles at two loci; one locus determines whether the bulb will be coloured at all and a second locus determines whether it will be red or yellow. The pres-
ence of the pungent chemical capsaicin in hot peppers is determined by a dominant allele at one locus, while the degree of heat is regulated by a series of modifiers at several other loci.
Quantitative traits are commonly influenced by epistatic interactions.
These can be identified by plotting the trait values associated with each geno- type (Fig. 3.3). Suppose you have two loci regulating plant height, A and B, with two alleles each and you plot the values of the genotypes of BB, Bb and bb for each allelic substitution at the A locus. If there is no epistasis and no dominance, the relative values for each genotype will rely solely on the addi- tive combination of alleles at each locus, and the slopes of all the lines will equal 1. If there is no epistasis but there is complete dominance, the values for heterozygotes will equal those of one of the homozygotes and the trajectories of each line will level off at the same point. If there is epistasis, the alleles will interact in a more complex fashion, and the trajectories of each line will differ.
Fig. 3.2. Multiple effects of the gene Sin Nicotiana tabacum.(From Stebbins, 1959.)
Statistical analyses have been developed to partition levels of quantita- tive variation into additive, dominance and epistatic interactions, using analysis of variance techniques (Falconer and Mackay, 1996). In one of the simplest analyses, breeders measure complex genomic interactions by calcu- lating general and specific combining ability (Griffing, 1956). In this type of analysis, the breeder makes a series of crosses between a group of parents and compares their mean performance with that of their progeny. The mean performance of each line in crosses with all the other lines is called the gen- eral combining ability (GCA) and represents the additive component of vari- ance. The deviation of a particular individual cross from the average GCA of the two lines is called the specific combining ability (SCA) and represents intra- and interlocus interactions (Fig. 3.4). Standard analysis of variance techniques are used to calculate the relative importance of GCA and SCA (Gilbert, 1967). More complex crossing and statistical approaches are required to separate dominance and epistatic interactions.
Fig. 3.3. A demonstration of the effects of additive, dominant and epistatic interactions on a quantitative trait.
Classical statistical studies of quantitative variation have uncovered con- siderable evidence of epistasis (Falconer and Mackay, 1996), but the docu- mentation of intergenic interactions in QTL mapping studies has proved more elusive (Tanksley, 1993; Paterson, 1995; Kim and Rieseberg, 2001). A large part of the difficulty in identifying epistatic interactions with molecular markers deals with the statistical power represented by smaller population sizes, and limitations in the analysis of variance technique itself (Wade, 1992;
Doebley et al., 1995). In spite of these limitations, several plant studies have clearly documented epistasis. Yamamoto et al. (2000) found a significant interaction between two QTL (Hd2and Hd6) involved in photoperiod sensi- tivity in rice. Yu et al. (1997) identified 32 QTL associated with four yield traits in rice, and found almost all of them to interact significantly with at least one other QTL. Doebley’s group has discovered a significant epistatic interac- tion in maize between a QTL on chromosome arm 1L (QTL-1L) and another
Expected yields
Observed yields (lb/tree/year)
regression, b = 1
correlation, r = 0.945
r2 = 0.894
example, R × C observed: 14.4
expected: −2.55 + 4.94 + 14.19 = 16.58 SCA effect: 14.4 − 16.58 =−2.18 24
20
16
12
8
4
0
0 4 8 12 16 20 24
Fig. 3.4. Combining ability for yield in rubber, Hevea brasiliensis. The observed yield of progenies is plotted against the expected yields or general combining ability (GCA) of each line. The GCA is calculated as the mean performance of each line in crosses with all other lines. When a regression line is drawn through the various parental values, the deviations above and below the line represent specific combining ability (SCA). (Used with permission from N.W. Simmonds, © 1985, Principles of Crop Improvement, Longman, London.)
on chromosome arm 3L (QTL-3L), which both influence the number and length of the internodes in both the primary lateral branch and the inflores- cence (Doebley et al., 1995; Lukens and Doebley, 1999). Alleles at these loci derived from either maize or teosinte had the strongest phenotypic effect in their own species background, further signalling the complexity of genomic interactions in Zea mays. QTL-1L was determined to be the locus for tb1, and QTL-3L could be te1, which we have already described as being highly pleiotropic. Again, selection at these two loci would have had dramatic phe- notypic effects during the domestication process.