Allele-Specific, Bidirectional Silencing of an Alcohol Dehydrogenase Gene in Different Organs of Interspecific Diploid Cotton Hybrids

Interspecific hybridization is a common and important process that generates phenotypic novelty and can lead to hybrid speciation as well as to changes in gene expression. Using two different interspecific cotton (Gossypium) diploid hybrids, we show organ-specific, bidirectional allelic silencing at a heterozygous Adh locus, with alternate alleles being silenced in leaves and many floral organs, respectively. These results show that developmental regulation of gene expression is changed immediately upon hybridization between diploid species, possibly due to epigenetic factors or regulatory mismatch.

H YBRIDIZATION between different species is prevalent in plants and is thought to play an important role in adaptive evolution (Arnold 1997(Arnold , 2004Cronn and Wendel 2004;Hegarty and Hiscock 2005;Rieseberg 1997). Interspecific hybridization may lead to chromosomal rearrangements (Rieseberg et al. 1996;Shaked et al. 2001), transposable element mobilization (Liu and Wendel 2000;Shan et al. 2005), DNA methylation changes (Salmon et al. 2005), and a broad array of new and sometimes transgressive phenotypes (Rieseberg et al. 1999(Rieseberg et al. , 2003 and ecological innovations (Rieseberg et al. 2003). Hybridization between two species can result in changes in gene expression. Up-and downregulation of expression has been shown in triploid Senecio hybrids, in diploid and triploid maize hybrids, and in diploid wheat hybrids compared with their parents (Wu et al. 2003;Auger et al. 2005;. Hybridization also can affect allelic expression. Allele-specific differences in expression levels of nonimprinted genes have been documented in F 1 hybrids of mouse (Cowles et al. 2002), Drosophila (Wittkopp et al. 2004), and Saccharomyces cerevisiae (Ronald et al. 2005). In plants, expression levels of alleles (of nonimprinted genes) have been explored in intraspecific hybrid lines but not in interspecific diploid hybrids. In maize diploid hybrids, several genes displayed unequal expression in heterozygotes or silencing of one allele (Guo et al. 2003(Guo et al. , 2004, whereas two other genes exhibited allelic expression level differences that varied by organ type (Guo et al. 2004).
In allopolyploid hybrids expression levels of homeologous gene pairs (i.e., genes duplicated by polyploidy) relative to each other have been found to be variable by organ type (Adams et al. , 2004. The most extreme form of organ-specific partitioning of expression between homeologs is exemplified by the alcohol dehydrogenase gene AdhA in cotton, where there is silencing of one homeolog in some organs and of the other homeolog in other organs, suggesting subfunctionalization . The organ-specific partitioning of gene expression could be caused by hybridization, chromosome doubling, or both; diploid hybrids provide a means to separate these factors. To determine if organ-specific alteration in the expression of individual alleles can occur in response to interspecific hybridization between two diploid species, without chromosome doubling, we assayed expression of AdhA in several organs of two different diploid F 1 hybrids in the cotton genus (Gossypium). Two F 1 hybrids ( Figure 1A) were synthesized by pollinating emasculated flowers of the maternal species with pollen from the paternal species and growing the resulting seed to generate F 1 hybrid individuals. The success of the crosses was verified by morphology and allozymes. To assay gene expression, we used a combination of RT-PCR and SSCP-cDNA approaches, as described Cronn and Adams 2003).
Bidirectional, organ-specific allelic silencing: As shown in Figure 1, interspecific diploid hybrids display allelespecific expression changes at the first generation. In the 1 Gossypium trilobum 3 G. gossypioides F 1 hybrid, only the G. gossypioides (paternal) AdhA allele was expressed in petals, stamens, and stigmas and styles ( Figure 1B), although the genomic DNA control demonstrates that both alleles are present. AdhA is expressed in the equivalent organs of the G. trilobum parent ( Figure 1D). These results indicate that silencing of one allele can occur in the first generation after hybridization. In the G. raimondii 3 G. gossypioides hybrid, expression of the two AdhA alleles ranged from approximately equal in the petals to expression of one allele or the other in different organs ( Figure 1C, Table 1). The G. raimondii allele was silenced in the bracts (epicalyx), stigmas and styles, and ovules, whereas the G. gossypioides allele was silenced in leaves. AdhA is expressed in the equivalent organs of the G. raimondii and G. gossypioides parents ( Figure 1D). This organ-specific, reciprocal silencing of alternative alleles at a single locus in F 1 hybrids is analogous to the previously reported silencing of homeologous AdhA genes in natural and synthetic allopolyploids . A key distinction between the present phenomenon and that reported earlier is that in the present case the reciprocal silencing affects alternative alleles at a heterozygous locus, as opposed to duplicated genes residing in different, coresident genomes of a polyploid nucleus.
Functional and evolutionary significance: The functional consequences of organ-specific allelic silencing are unknown, as is the scope of the phenomenon. In the present example, there are no differences in the inferred AdhA amino acid sequences among the parents and hybrids surveyed, so one might suggest that the phenomenon has little relevance from an evolutionary perspective. We can, however, envision evolutionarily relevant dimensions to the observations reported here.
One key question would appear to be whether the expression pattern segregates in the F 2 and subsequent generations. These populations were not available for this study, but in principle an F 2 progeny could be generated for the reportedly fertile (Brown and Menzel 1952;Menzel and Brown 1954) G. raimondii 3 G. gossypioides hybrid (the other hybrid is sterile). To the Lanes are numbered in pairs as follows: 1, D5 bract; 2, D5 stamen; 3, D5 stigma and style; 4, D5 ovule; 5, D6 leaf; 6, D8 stamen (repeat); 7, D8 petal; 8, D8 stamen; and 9, D8 stigma and style. ''m'' indicates size marker. Plant materials: G. gossypioides (accession HC8065), G. raimondii (Wendel laboratory stock), and G. trilobum (accession D8-1) were crossed as described in the text and as shown in A. The same perennial G. raimondii and G. gossypioides individuals used to make the hybrids and a plant from the accession of G. trilobum were used for RNA extractions. Two progeny plants derived from the same cross were used as biological replicates for each of the diploid hybrids. All plants were grown under the same conditions in the greenhouse at Iowa State University. Methods: RNA was extracted from the plants (flowers were collected between 10:00 am and noon on the day of flower opening), the quality was checked by electrophoresis on agarose gels, and the concentration was estimated with a spectrophotometer as previously described . The RNA was treated with DNase to remove residual DNA and then it was reverse transcribed also as previously described . PCRs were performed with cDNA templates, using primers that amplify both alleles, as described in Adams et al. (2003). Primer sequences were 59-GGTGAGG GTGTGAATGATTTG-39 and 59-GTACATATCCCACAGCTAA TGG-39. RT-PCR products were separated by SSCP-cDNA ; gels included 2% urea and were run at 4°. Bands were quantified from dried gels by phosphorimaging using ImageQuant software (Molecular Dynamics, Sunnyvale, CA). RT-PCR products showing one allele on the SSCP gels were directly sequenced to verify the presence of only one allele. extent that organ-specific silencing of alleles is controlled by cis-regulatory elements (Wittkopp et al. 2004), heterozygotes in subsequent generations are expected to display the bidirectional silencing reported here, while parental patterns should be restored in both classes of parental homozygotes. When extended across the genome to include hundreds to perhaps thousands of independently assorting genes, each exhibiting the type of regulatory effects shown here, one can imagine that populations derived from the initial hybridization event will exhibit a remarkable diversity of combined expression patterns. Thus, one potential basis of the morphological, physiological, and ecological novelty that characterizes hybrid populations and species may be the cumulative effect of developmentally regulated, reciprocal allelic expression biases.
Organ-specific partitioning of aggregate ancestral expression patterns, as observed for AdhA in cotton polyploids , may be considered a form of subfunctionalization. In an analogous fashion, the reciprocal silencing of AdhA alleles reported here in the cotton F 1 hybrids raises the possibility of instantaneous subfunctionalization that could lead to preservation of both parental alleles in subsequent generations. Thus, an additional evolutionarily relevant consequence of this process, to the extent that it operates, would be the preservation of higher levels of intrapopulation allelic diversity than would otherwise occur.
Possible causal factors: From a mechanistic perspective, what might be the cause of allelic silencing in F 1 hybrids in an organ-specific manner? At present no data specifically address this question, but clearly epigenetic factors, such as changes in DNA cytosine methylation, histone modifications (including deacetylation and methylation), or small RNAs may play a role. Another possibility is that sequence variation in the cis-regulatory elements of the alleles might cause differential binding of regulatory factors, causing different stoichiometric biases and regulatory mismatches that vary by organ and tissue (Birchler et al. 2005;Veitia 2005).
We thank Rich Cronn for making the G. raimondii 3 G. gossypioides F 1 hybrid. This research was supported by grants from the U.S. Department of Agriculture and the National Science Foundation to J. F. Wendel. Numbers represent transcript proportions derived from each parental allele, as determined by phosphorimaging. Each pair of rows represents a biological replicate (rep). Allelic expression was not assayed in some organs of the D8 3 D6 hybrid.