Epigenetic Allelic States of a Maize Transcriptional Regulatory Locus Exhibit Overdominant Gene Action

Corresponding author: Vicki L. Chandler, Department of Plant Sciences, 303 Forbes Hall, University of Arizona, Tucson, AZ 85721., chandler{at}ag.arizona.edu (E-mail).

Communicating editor: J. A. BIRCHLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Using alleles of the maize purple plant locus (pl), which encodes a transcriptional regulator of anthocyanin pigment synthesis, we describe a case of single-locus heterosis, or overdominance, where the heterozygote displays a phenotype that is greater than either homozygote. The Pl-Rhoades (Pl-Rh) allele is subject to epigenetic changes in gene expression, resulting in quantitatively distinct expression states. Allelic states with low-expression levels, designated Pl'-mahogany (Pl'-mah), are dominant to the high-expression state of Pl-Rh. Pl'-mah states retain low-expression levels in subsequent generations when homozygous or heterozygous with Pl-Rh. However, Pl'-mah alleles frequently exhibit higher expression levels when heterozygous with other pl alleles; illustrating an overdominant allelic relationship. Higher expression levels are also observed when Pl'-mah is hemizygous. These results suggest that persistent allelic interactions between Pl'-mah and Pl-Rh are required to maintain the low-expression state and that other pl alleles are missing sequences required for this interaction. The Pl-Rh state can be sexually transmitted from Pl'-mah/pl heterozygotes, but not from Pl'-mah hemizygotes, suggesting that fixation of the high-expression state may involve synapsis. The existence of such allele-dependent regulatory mechanisms implicates a novel importance of allele polymorphisms in the genesis and maintenance of genetic variation.

HERITABLE changes in gene activity can, in some cases, be influenced by allelic interactions. Several examples of such interactions at loci affecting pigment production have been described in Zea mays (HOLLICK et al. 1997 ). The pl gene encodes a myb-like transcriptional regulator of the anthocyanin biosynthetic pathway, making the amount of visual pigment produced an excellent indicator of pl gene expression (COCCIOLONE and CONE 1993 ; PATTERSON 1993 ). The Pl-Rhoades (Pl-Rh) allele is highly expressed, producing strong pigment in most tissues of the juvenile and adult maize plant. However, the allele is unstable and subject to paramutation, a process in which one allele promotes a heritable alteration in the expression of another allele in the heterozygote (HOLLICK et al. 1995 , HOLLICK et al. 1997 ). At various frequencies (typically ~16% in our W23 inbred stocks), lighter pigmented progeny arise from crosses between Pl-Rh homozygous plants. This reduction in pigment is easily quantified in anther tissues, where we have defined a graded series of anther color scores (ACS; HOLLICK et al. 1995 ). Collectively, these lighter pigmented variants represent a continuum from virtually no color in the anthers (ACS of 1) to slightly less than that of Pl-Rh (ACS of 7). These reduced pigment phenotypes are due to the heritable alteration of Pl-Rh to states expressing lower levels of pl RNA (PATTERSON 1993 ) that are collectively designated Pl'-mahogany (Pl'-mah) (HOLLICK et al. 1995 ).

The behavior of different Pl'-mah states is distinct in subsequent crosses with homozygous Pl-Rh plants. Pl'-mah states with virtually no color (ACS of 1–2) are meiotically very stable and have a strong ability to change Pl-Rh alleles into Pl'-mah; in crosses of Pl'-mah plants to Pl-Rh homozygotes, only weakly pigmented phenotypes of Pl'-mah plants (ACS of 1–4) are recovered (HOLLICK et al. 1995 ). Pl'-mah states with intermediate levels of pigment (ACS of 3 or 4) frequently change to lower expression states, but changes to higher expression states (ACS of 5–7) have not been observed. The intermediate states are nonetheless capable of changing Pl-Rh to Pl'-mah with 100% efficiency. Pl'-mah states that confer slightly less color than Pl-Rh (ACS of 5 or 6) are metastable; they can change to lower expressed Pl'-mah states (ACS of 1–4), or they can change to fully expressed Pl-Rh states (ACS 7) that are unable to change Pl-Rh to Pl'-mah in subsequent generations. Although the molecular basis underlying the distinct expression states is not known, current evidence favors an epigenetic mechanism, perhaps involving heritable alterations in chromatin structure (HOLLICK et al. 1997 ).

The spontaneous instability of Pl-Rh, and the exclusive transmission of Pl'-mah alleles from heterozygotes, suggests that an open-pollinated population of homozygous Pl-Rh plants would change to exclusively Pl'-mah plants within several generations. The persistence of the highly expressed Pl-Rh allele in culture is undoubtedly due in large part to the stringent artificial selection imposed by maize geneticists. Additionally, the maintenance of the Pl-Rh state in maize stocks may be facilitated by pl allele polymorphisms or other genetic factors affecting the stability of the Pl-Rh or Pl'-mah states. Herein, we show that the stability of the Pl'-mah state is sensitive to allelic interactions and that Pl'-mah can change back to Pl-Rh when it is heterozygous with polymorphic pl alleles. In heterozygous combination, two weakly expressed pl alleles produce more than additive pl gene activity, thus illustrating an example of overdominant gene action. Potential implications of such allelic interactions to studies on quantitative genetic variation are discussed.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic stocks:
Pl-Rh was maintained in the W23 inbred background and in four other genetic backgrounds of mixed parentage. The original Pl-Rh sm stock derives from an unknown genetic background (HOLLICK et al. 1995 ). Pl'-mah alleles arose spontaneously in three different genetic backgrounds and were subsequently propagated by self- or sib crossing. Self- or sib crossing propagated pl-CO159 in a CO159 x Tx303 recombinant inbred line. The TB-6Lc stock was provided by the Maize Genetics Cooperation Stock Center, accession no. 614C (Urbana, IL). All stocks used in these experiments were also homozygous for a functional allele of the r locus required for anther pigmentation, r-r.

Genetic crosses:
Hand pollinations were performed for all experiments. For the cosegregation experiment shown in Figure 2, three of the six families used Pl'-mah sm/Pl'-mah sm individuals as pistillate parents while the other three used Pl-Rh sm/pl-CO159 Sm individuals as pistillate parents. Sm/sm individuals have yellow silks, whereas sm/sm individuals have distinct salmon-colored silks. All heterozygotes contain at least one copy of the P-rr allele, which is required to score the sm/sm phenotype. All the Pl'-mah sm/Pl'-mah sm parental individuals used in this experiment were siblings from the same ear. The Pl-Rh sm/pl-CO159 Sm individuals derive from three closely related ears. Silk color was scored ~2 days after silk emergence from the husks.

View larger version (63K): In this window In a new window Download PPT slide   Figure 1. Epigenetic behavior of Pl'-mah. (A) Crosses between Pl'-mah (Pl'-mah/Pl-Rh) and Pl-Rh (Pl-Rh/Pl-Rh) individuals exclusively produce progeny with pigment phenotypes of Pl'-mah plants. Percentages of progeny with distinct anther pigment phenotypes (ACS) are indicated for 678 plants derived from the seeds of 14 ears (data from HOLLICK et al. 1995 ). (B) Crosses between pl-CO159 (pl-CO159/pl-CO159) and Pl'-mah (Pl'-mah/Pl'-mah) individuals produce more progeny with greater anther pigmentation than either parent. Percentages of progeny with distinct anther pigment phenotypes are indicated for 41 plants derived from the seeds of two ears.

View larger version (20K): In this window In a new window Download PPT slide   Figure 2. Segregation analysis shows that a pl allele exhibits overdominant gene action. (A) Schematic diagram of parental and segregant chromosomes with the relevant genetic markers. (B) Histogram representing pl gene expression among the segregant progeny derived from six ears. Open bars, sm/sm individuals; solid bars, Sm/sm individuals.

For the segmental aneuploidy experiment in Figure 3, normal diploid individuals (Pl'-mah/Pl'-mah) were crossed as female with pollen from Pl-Rh individuals that carried two TB-6Lc B-A chromosomes (hyperploids) in which 90% of 6L is now attached to the supernumerary B chromosome. Over four successive growing seasons, progeny plants were scored for anther color on the 1–7 scale. Using a x50 magnification pocket microscope, these plants were further classified, using frequency of pollen grain abortion as an indicator of 6L ploidy; hypoploids, euploids, and hyperploids have 50, 25, and <10% aborted pollen, respectively. For the initial crosses, hyperploid plants were identified by their low levels of pollen abortion, and these were confirmed to be hyperploids by crosses with recessive pl testers. Progeny of these testcrosses were primarily hyperploid (low pollen abortion) and hypoploid (~50% pollen abortion) individuals, and in 10/10 cases, the hyperploid individuals had a pigment phenotype like that of Pl-Rh plants. Following exposure of these B-A chromosomes to Pl'-mah, they now conferred phenotypes like that of Pl'-mah plants and they were fully capable of causing naive Pl-Rh alleles to change to Pl'-mah in subsequent crosses. Although the original hyperploid parents also have a normal chromosome 6 that carries a recessive, neutral pl allele, experimentally this chromosome is only rarely transmitted through a hyperploid male. Because meiosis will result in the segregation of a B-A chromosome with the normal chromosome, such pollen grains have a duplication of 6L genetic material. The failure to transmit the normal chromosome is presumed to reflect a selective growth, or fertilization, disadvantage for such pollen grains (BECKETT 1991 ). We found no evidence in any subsequent generations that the B-A chromosomes carried neutral pl alleles. However, because low levels of recombination are expected to place a neutral pl allele onto a B-A chromosome, it is possible that a few of the hyperploids represented in Figure 3 are (Pl'-mah/pl/pl).

View larger version (21K): In this window In a new window Download PPT slide   Figure 3. Hemizygosity leads to increased pl gene expression. (A) Schematic diagram of the chromosome 6L dosage series generated through the use of B-A translocation chromosomes. The B chromosome segments are enlarged relative to the normal chromosome 6 regions. (B) Histogram representing pl gene expression among hyperploid (open bars), euploid (crosshatched bars), and hypoploid (solid bars) individuals. Data represent individuals derived from two ears scored over four growing seasons. Higher frequencies of hyperploid vs. hypoploid individuals are typically observed with B chromosomes (BECKETT 1991 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Pl'-mah alleles exhibit higher expression levels when heterozygous with other pl alleles:
Most pl alleles are "neutral" with respect to paramutation; they are unable to cause Pl-Rh to change to Pl'-mah, and their expression levels or their ability to paramutate Pl-Rh is unaffected by exposure to Pl'-mah. Previously we showed that when the Pl'-mah allele is heterozygous with the neutral pl allele, pl-W23, the Pl'-mah allele exhibits higher expression levels (HOLLICK et al. 1995 ). In some cases, a fully expressed Pl-Rh state was observed, and this state was sexually transmitted from such Pl'-mah/pl-W23 heterozygotes (Table 1). However, in such crosses we could not distinguish between allele interactions leading to instability or the involvement of nonallelic genetic factors. To examine this further, we have determined the stability of Pl'-mah when heterozygous with other neutral alleles in distinct genetic backgrounds (Table 1). One particular pl allele (pl-CO159) produces no detectable pigment or RNA and is thought to be nonfunctional (CONE et al. 1993 ). Most pl-CO159/Pl'-mah heterozygotes showed increased pigment levels relative to their parents, and nearly 10% of these had Pl-Rh-like levels of pigment (Figure 1B, Table 1). Crosses of these Pl-Rh-like plants with a Pl-Rh/Pl-Rh tester resulted in 47/52 progeny with the phenotype of Pl-Rh plants. Although the frequency of the Pl'-mah exceptions is the same as that typically seen spontaneously among Pl-Rh homozygotes, it is possible that the exceptions represent the transmission of some Pl'-mah gametes. These results confirm that Pl'-mah had changed to Pl-Rh in ~10% of the Pl'-mah/pl-CO159 heterozygotes, and this change was, for the most part, heritable. We have also observed and confirmed these changes when Pl'-mah is heterozygous with one other neutral pl allele, pl-606B (Table 1). All three neutral pl alleles tested have distinct pigment phenotypes, and restriction fragment analyses demonstrate that both pl-CO159 and pl-W23 are structurally polymorphic from each other and from Pl-Rh (CONE et al. 1993 ; K. CONE, J. HOLLICK, and V. CHANDLER, unpublished results). The DNA structure of pl-606B has not yet been determined.

Increased gene activity cosegregates with the presence of a neutral pl allele:
As several different genetic backgrounds were used in these experiments, it seemed likely that the neutral pl alleles facilitated the instability of Pl'-mah. However, it was formally possible that other loci, rather than the neutral alleles, were responsible. As such, we tested whether the ability of Pl'-mah to change to higher expression states cosegregated with a neutral allele between individuals from single crosses. A linked morphological marker (salmon silks, sm), which is 10 cM distal to pl on chromosome 6, was used to indicate the pl allele combination. Homozygous recessive sm individuals have salmon-colored silks, while Sm/sm heterozygotes have yellow silks. Anther color scores were determined for the plants derived from six ears produced by the crossing scheme illustrated in Figure 2A. Scoring for the sm marker allowed classification by pl allele based on linkage to sm. The sm/sm (Pl-Rh/Pl'-mah) class had lower ACS and no potential Pl-Rh plants relative to the Sm/sm (pl-CO159/Pl'-mah) class (Figure 2B). All six families gave the same result; a higher frequency of individuals with increased pl gene expression cosegregated with the Sm allele. The average ACS values between the sm/sm and Sm/sm classes were significantly different from one another (P 0.01; two-sample z-test). These results indicate that the ability to cause increased gene expression of the Pl'-mah allele is genetically linked to the homologous region of chromosome 6 containing a neutral pl allele. Furthermore, the observation that most heterozygotes (Pl'-mah/pl-CO159) have higher gene expression levels than either homozygote represents an overdominant allelic relationship. Overdominance is defined as an allelic relationship whereby the heterozygote displays a phenotype exceeding that of either homozygote.

Hemizygosity of Pl'-mah leads to increased gene expression:
Two simple explanations for this overdominant relationship are that either the neutral allele actively promotes the instability of the Pl'-mah state, or allelic interaction between Pl'-mah and Pl-Rh are required to maintain the reduced expression states. To address these possibilities, we used B-A translocation stocks to create a 6L dosage series with zero, one, or two doses of Pl-Rh combined with one dose of Pl'-mah and then examined the anther pigment phenotypes of the resulting progeny. Pl'-mah/Pl'-mah individuals were crossed with pollen from an individual that carried B-A translocations consisting of most of chromosome 6L, carrying Pl-Rh, linked to the supernumerary B chromosome centromere (see MATERIALS AND METHODS). Due to a high frequency of B centromere nondisjunction during the second mitotic division of the male gametophyte (ROMAN 1947 ), progeny are produced that are hypoploid (Pl'-mah/-) or hyperploid (Pl'-mah/Pl-Rh/Pl-Rh). Euploid progeny (Pl'-mah/Pl-Rh) are also produced when nondisjunction fails to occur. Figure 3A illustrates the relevant chromosomes in these different individuals. These three classes are distinguishable by different percentages of pollen abortion, which can be scored with a hand-held pocket microscope. Anthers of progeny from the above-mentioned cross were scored on the 1–7 ACS scale, and the amount of pollen abortion was then used to discriminate among the various 6L classes.

More plants in the hemizygous (hypoploid) class had higher expression levels than did plants in either the euploid or hyperploid classes (Figure 3B). The average ACS value of the hypoploid class was significantly different from the average ACS values of both the hyperploid (P 0.01; two-sample z-test) and euploid (P = 0.01; two-sample z-test) classes. In addition, the plants with the highest expression levels were Pl'-mah/- hemizygotes. The increase in expression observed in the hemizygote is unlikely to be due simply to gene dosage. If gene expression were inversely proportional to gene dosage, one would predict that the euploid class (Pl'-mah/Pl-Rh) would have pigment values intermediate to those seen in the hypoploid and hyperploid classes, which is not observed. The average ACS score of the euploid class (2.3 ± 0.72; 95% confidence interval is ± 2 SE, calculated with d.f. = 8) was not statistically different from the average ACS score of the hyperploid class (2.55 ± 0.28; 95% confidence interval is ± 2 SE). However, there are examples where dosage compensation upregulates some hemizygous genes (GUO and BIRCHLER 1994 ). We think this is unlikely in our examples because dosage compensation usually results in 2- to 3-fold changes in expression, and the phenotype of the Pl'-mah hemizygote suggests >10-fold increase in expression (PATTERSON 1993 ). In addition, we would expect to see increased expression in every hemizygous plant, which is not observed.

Even though the hypoploid class had many plants with a phenotype like that of Pl-Rh plants, it was unknown whether Pl'-mah had heritably changed to Pl-Rh. An additional cross was performed to determine whether or not the pl allele in the Pl'-mah/- hemizygotes was capable of causing paramutation. Three of the ACS 7 hypoploids and one of the ACS 6 hypoploids were crossed to homozygous Pl-Rh plants. Progeny with an ACS of 7 would be expected if Pl'-mah had changed to Pl-Rh in the Pl'-mah/- individuals. This was not observed; all of the progeny (95/95) from these crosses had phenotypes like that of Pl'-mah plants. This indicates that these hypoploids with strongly pigmented anthers transmitted only Pl'-mah alleles. Thus, when hemizygous, the expression levels of Pl'-mah increase to those of Pl-Rh in some plants, but these alleles are not transmitted in the Pl-Rh state, as they can still promote paramutation of naive Pl-Rh alleles. These results contrast with the Pl-Rh examples found among Pl'-mah/pl-neutral heterozygotes (Table 1), in which plants with an ACS 7 transmitted Pl-Rh.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Our results demonstrate that the stability of Pl'-mah depends on the other allele. Pl'-mah is very stable when homozygous or when heterozygous with Pl-Rh. In contrast, the expression of Pl'-mah frequently increases when it is heterozygous with a neutral allele or when hemizygous. The stronger expression phenotype of the Pl'-mah/pl-CO159 heterozygote when compared to the phenotype of either homozygote represents an example of overdominance or single-locus heterosis. The presence of pl allelic interactions results in a repression of gene activity, whereas the absence of allelic interactions results in increased gene activity.

Gene activity is increased when Pl'-mah is heterozygous with a neutral allele or when hemizygous:
While the expression of Pl'-mah frequently increases when it is heterozygous with a neutral allele or hemizygous, increases in Pl'-mah expression do not occur in every instance. One possibility is that the changes in state are stochastic and that absence of allelic interactions increases the frequency that a change to a higher expression state can occur but does not dictate that it will occur in every individual. It is also possible that changes occur at high frequency in all allele combinations, but in the absence of interacting DNA sequences, the higher expression states are more stable and, thus, more frequently observed. Additionally, or alternatively, there may be intrinsic differences between individual Pl'-mah alleles following meiotic segregation, or there may be modifier loci that play a role in stabilizing higher expression states.

The observations that gene expression is increased when Pl'-mah is hemizygous and when Pl'-mah is heterozygous with neutral pl alleles suggests that neutral alleles do not interact with Pl'-mah and that the low-expressing Pl'-mah state is destabilized by the absence of allelic interactions. Direct interactions between Pl'-mah alleles or between Pl'-mah and Pl-Rh alleles are thus presumed to favor the lower expression state. In this interpretation, neutral alleles are hypothesized to lack specific sequences that are found in Pl-Rh and Pl'-mah alleles. One prediction from these results is that fully expressed Pl-Rh alleles would be more stable, less likely to undergo spontaneous changes to Pl'-mah, when maintained in heterozygous association with a neutral allele. Long-term experiments are in progress to address this prediction.

When Pl'-mah is hemizygous, its expression levels increase to those of Pl-Rh in some plants, but these alleles are not transmitted in the Pl-Rh state as they can still promote paramutation of naive Pl-Rh alleles. These results contrast with the Pl-Rh examples found among Pl'-mah/pl-neutral heterozygotes, which transmitted Pl-Rh. Unlike the heterozygotes, the hypoploids do not have homologous-pairing partners. Thus, one very intriguing possibility is that fixation of the Pl-Rh state for meiotic transmission may depend on homologue synapsis during meiotic prophase. These results further suggest that the establishment of the Pl-Rh expression states may be mechanistically separable from the heritable maintenance of the expression states through meiosis.

The fact that none of the hyperploid individuals (Pl'-mah/Pl-Rh/Pl-Rh) had a phenotype like that of Pl-Rh plants (Figure 3B) demonstrates that paramutation can occur between multiple pl alleles within the span of a single generation. However, unlike diploid (Pl'-mah/Pl-Rh) plants from the related W23 inbred line, which all had anther color scores between 1 and 4 (HOLLICK et al. 1995 ), the hyperploid class did have a number of plants with anther color scores of 5 and 6. More euploid individuals will need to be scored to determine if these ACS 5 and 6 plants are due to two Pl-Rh alleles being less efficiently paramutated than one.

Comparison to other allele-dependent regulatory mechanisms:
Paramutation interactions have been detailed for three of the four maize loci that encode transcriptional activators of the anthocyanin biosynthetic enzymes. While certain features are similar, each case of paramutation has unique properties, implicating either distinct molecular mechanisms or intrinsic differences in how alleles at a particular locus interface with common regulatory machinery (HOLLICK et al. 1995 , HOLLICK et al. 1997 ). Two of these three examples, pl and r, show an increased frequency of enhanced expression when heterozygous with neutral alleles or when hemizygous. The pigment levels of both the R-r allele and its weakly pigmented, paramutant derivative R-r', are heritably increased by exposure to either a neutral r allele or to a small chromosomal deficiency spanning the r locus (STYLES and BRINK 1968 ). These changes in r gene expression are meiotically heritable. By repeated exposure to a neutral allele or a small deficiency over several generations, R-r' will change back to the original R-r level of pigment. Changes of Pl'-mah to Pl-Rh can occur within the span of a single generation. In contrast to the r and pl examples, the paramutant B' allele does not change to higher expression states when carried over neutral b alleles (COE 1966 ; PATTERSON et al. 1995 ). While large numbers of plants hemizygous for B' have yet to be generated, no increases in B' expression levels have been observed among the 40 hemizygotes examined to date (V. CHANDLER, unpublished results). Thus, the B' state is much more stable than either the Pl'-mah or R-r' states.

Allele-dependent regulation is not unique to these examples of paramutation. Gene regulatory systems exist in Drosophila, where the hemizygote exhibits more gene expression than the homozygote. These are cases where proper gene repression is dependent upon chromosome-pairing interactions. In the classic example of bithorax-complex transvection (LEWIS 1985 ), proper developmental repression of the Ubx gene is dependent upon the ability to pair; hemizygosity leads to gene activity in ectopic positions within the developing fly. With certain zeste mutations, the activity of the white gene is reduced when the alleles can pair, yet this repression is relieved when pairing is disrupted or when the white allele is hemizygous (JACK and JUDD 1979 ). Additionally, several examples of pairing-sensitive DNA elements have been identified that facilitate Polycomb-mediated repression. These elements have the remarkable ability to confer silencing on transgene constructs when homozygous, yet these transgenes regain activity when hemizygous (KASSIS 1994 ; GINDHART and KAUFMAN 1995 ).

Additional examples of allele-influenced gene silencing are found among plant transgenes. In a recent study on uidA transgenes in tobacco (NAP et al. 1997 ), the activity of complex alleles (loci composed of multiple transgenes arranged in an unknown orientation) was observed to be enhanced when maintained in a hemizygous state. Results from several other studies have implicated a dosage-sensitive repression of transgene expression, suggesting that an RNA threshold exists beyond which homologous RNA molecules are rapidly degraded (JORGENSEN 1995 ; METZLAFF et al. 1997 ). Hemizygotes are less likely to exceed these thresholds and would therefore not experience a high frequency of silencing.

Epigenetic sources of heritable variation:
Our results demonstrate that the primary determinant of a quantitative trait (plant color) has alleles that exhibit overdominant gene action. Multiple mechanisms responsible for overdominant allelic interactions have been proposed and discussed, with most models focusing on interactions of gene products (CROW 1952 ). In the pl example, it is gene activity that is increased in the heterozygote. Some of the models for overdominance predict that a hemizygote would have the same phenotype as the heterozygote. While other models are inconsistent with this prediction, we are unaware of any arguments that exclude an overdominant relationship between two alleles, based on a similar phenotype in the hemizygote. The key feature of the pl example of overdominance is that allelic interactions are required to maintain repression of gene activity; the absence of such interactions results in increased gene activity.

Our example of overdominance may be relevant to the debate concerning the poorly understood phenomenon of heterosis. The term "heterosis" describes the process(es) responsible for the superiority of traits in the F₁ over that observed in their parents. The underlying mechanisms responsible for this phenomenon are likely to be diverse, and specific hypotheses have been the subject of considerable debate for over 90 yr (SCHULL 1908 ; EAST 1936 ; SPRAGUE 1953 ; CROW 1993 ). Overdominance has been suggested as one cause of heterosis because the heterozygote would display a phenotype exceeding that of either homozygote (HULL 1945 ). We think it is important to consider the possibility that allele-dependent mechanisms of gene regulation could also contribute to heterosis. Interestingly, in discussing heterosis, SCHULL noted "that these differences [between uniting gametes] need not be Mendelian in their inheritance" (SCHULL 1948 ).

Quantitative traits are highly uniform among hybrid maize plants produced by intercrosses of elite inbred lines. In contrast, heterozygosity in the pl case leads to considerable variation rather than a uniform level of gene action. Due to the mixed genetic parentage of our material, it remains unclear whether the variation in our experiments is due to the segregation of additional genetic modifiers or reflects a stochastic variation solely determined by the pl locus itself.

Molecular genetic mapping of quantitative trait loci may be complicated by alleles subject to epigenetic regulatory mechanisms. In maize, molecular mapping experiments show that heterozygosity for certain chromosomal regions is often positively correlated with increased performance of specific traits (STUBER et al. 1992 ); although such relationships appear to be rare in rice (XIAO et al. 1995 ). However, because the molecular nature of these quantitative trait loci are mostly unknown, genetic and molecular mapping studies cannot always discriminate between apparent overdominance (cases where dominant complementation of deleterious recessive alleles at linked loci lead to increased phenotypic traits) and true overdominance (cases where single-gene heterozygosity leads to an increase in phenotypic traits). In generating the mapping populations themselves, the gene activity of some alleles may be heritably changed. In some cases, similar to the example of Pl'-mah changing to Pl-Rh in the Pl'-mah/pl-CO159 heterozygote, gene activity could be heritably increased, giving the impression that recombination has uncoupled genetic linkage from a deleterious recessive allele.

Allele-dependent regulatory systems also have important implications for the maintenance of allele polymorphisms. Epigenetic changes in pl gene activity are clearly influenced by specific allelic combinations. In the absence of strong selection, the highly expressed Pl-Rh state would be driven to extinction without the presence of neutral alleles. Thus, even if a given neutral allele is selected against when homozygous, it may have a selective advantage when heterozygous with other alleles. The combination of alleles subject to epigenetic changes together with neutral alleles that influence those changes provides a diverse and dynamic source of heritable variation.

	ACKNOWLEDGMENTS

We are grateful to the Maize Genetics Cooperation Stock Center (Urbana, Illinois) and to Karen Cone (University of Missouri) for generously providing germplasm. We thank William Tracy and Jane Dorweiler for insightful comments and critical review of this manuscript. This work was supported by grants from the American Cancer Society (NP875), the National Science Foundation (MCB-9603638) to V.L.C., a National Science Foundation Postdoctoral Research Fellowship in Plant Biology (BIR-9303601), and a National Research Initiative Competitive Grants Program/United States Department of Agriculture award (9701367) to J.B.H.

Manuscript received March 17, 1998; Accepted for publication July 10, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BECKETT, J. B., 1991 Cytogenetic, genetic and plant breeding applications of B-A translocations in maize, pp. 493–529 in Chromosome Engineering in Plants: Genetics, Breeding, Evolution, Vol IIA, edited by P. K. GUPTA and T. TSUCHIYA. Elsevier, Amsterdam.

COCCIOLONE, S. M. and K. C. CONE, 1993  Pl-Bh, an anthocyanin regulatory gene of maize that leads to variegated pigmentation. Genetics 135:575-588[Abstract].

COE, E. H., JR., 1966  The properties, origin, and mechanism of conversion-type inheritance at the B locus in maize. Genetics 53:1035-1063[Free Full Text].

CONE, K. C., S. M. COCCIOLONE, F. A. BURR, and B. BURR, 1993  Maize anthocyanin regulatory gene pl is a duplicate of c1 that functions in the plant. Plant Cell 5:1795-1805[Abstract].

CROW, J. F., 1952 Dominance and overdominance, pp. 282–297 in Heterosis, edited by J. W. GOWEN. Iowa State College Press, Ames, IA.

CROW, J. F., 1993 Mutation, mean fitness, and genetic load, pp. 3–42 in Oxford Surveys in Evolutionary Biology, Vol IX, edited by D. FUTUYMA and J. ANTONOVICS. Oxford University Press, London.

EAST, E. M., 1936  Heterosis. Genetics 21:375-397[Free Full Text].

GINDHART, J. G., JR. and T. C. KAUFMAN, 1995  Identification of Polycomb and trithorax group responsive elements in the regulatory region of the Drosophila homeotic gene Sex combs reduced. Genetics 139:797-814[Abstract].

GUO, M. and J. A. BIRCHLER, 1994  Trans-acting dosage effects on the expression of model gene systems in maize aneuploids. Science 266:1999-2002[Abstract/Free Full Text].

HOLLICK, J. B., G. I. PATTERSON, E. H. COE, JR., K. C. CONE, and V. L. CHANDLER, 1995  Allelic interactions heritably alter the activity of a metastable maize pl allele. Genetics 141:709-719[Abstract].

HOLLICK, J. B., J. E. DORWEILER, and V. L. CHANDLER, 1997  Paramutation and related allelic interactions. Trends Genet. 13(8):302-308[Medline].

HULL, F. G., 1945  Recurrent selection and specific combining ability in corn. J. Am. Soc. Agron. 37:134-145.

JACK, J. W. and B. H. JUDD, 1979  Allelic pairing and gene regulation: a model for the Zeste-white interaction in Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 76:1368-1372[Abstract/Free Full Text].

JORGENSEN, R. A., 1995  Cosuppression, flower color patterns and metastable gene expression states. Science 268:686-691[Abstract/Free Full Text].

KASSIS, J. A., 1994  Unusual properties of regulatory DNA from the Drosophila engrailed gene: three "pairing-sensitive" sites within a 1.6-kb region. Genetics 136:1025-1038[Abstract].

LEWIS, E. B., 1985  Regulation of the genes of the bithorax complex in Drosophila. Cold Spring Harbor Symp. Quant. Biol. 50:155-164[Abstract/Free Full Text].

METZLAFF, M., M. O'DELL, P. D. CLUSTER, and R. B. FLAVELL, 1997  RNA-mediated RNA degradation and chalcone synthase A silencing in Petunia. Cell 88:845-854[Medline].

NAP, J.-P., A. J. CONNER, L. MLYNAROVA, W. J. STIEKEMA, and R. C. JANSEN, 1997  Dissection of a synthesized quantitative trait to characterize transgene interactions. Genetics 147:315-320[Abstract].

PATTERSON, G. I., 1993 Paramutation: a directed, meioticaly heritable change in gene expression. Ph.D. thesis, Department of Biology, University of Oregon, Eugene, OR.

PATTERSON, G. I., K. M. KUBO, T. SHROYER, and V. L. CHANDLER, 1995  Sequences required for paramutation of the maize b regulatory gene map to a region containing the promoter and upstream sequences. Genetics 140:1389-1406[Abstract].

ROMAN, H., 1947  Mitotic nondisjunction in the case of interchanges involving the B-type chromosome in maize. Genetics 32:391-409[Free Full Text].

SCHULL, G. H., 1908  The composition of a field maize. Am. Breeders' Assoc. Rep. 4:296-301.

SCHULL, G. H., 1948  What is "heterosis"? Genetics 33:439-446[Free Full Text].

SPRAGUE, G. F., 1953 Heterosis, pp. 113–136 in Growth and Differentiation in Plants, edited by W. E. LOOMIS. Iowa State College Press, Ames, IA.

STUBER, C. W., S. E. LINCOLN, D. W. WOLFF, T. HELENTJARIS, and E. LANDER, 1992  Identification of genetic factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecular markers. Genetics 132:823-839[Abstract].

STYLES, E. D. and R. A. BRINK, 1968  The metastable nature of paramutable R alleles in maize. IV. Parallel enhancement of R action in heterozygotes with r and in hemizygotes. Genetics 61:801-811.

XIAO, J., J. LI, L. YUAN, and S. D. TANKSLEY, 1995  Dominance is the major genetic basis of heterosis in rice as revealed by QTL analysis using molecular markers. Genetics 140:745-754[Abstract].