Copyright � 2006 by the Genetics Society of AmericaDOI: 10.1534/genetics.105.047894

Genomewide Nonadditive Gene Regulation in Arabidopsis Allotetraploids

Jianlin Wang,* Lu Tian,* Hyeon-Se Lee,* Ning E. Wei,*,† Hongmei Jiang,‡

Brian Watson,§ Andreas Madlung,§,1 Thomas C. Osborn,**R. W. Doerge,‡ Luca Comai§ and Z. Jeffrey Chen*,2

*Genetics Program and Department of Soil and Crop Sciences, Texas A&M University, College Station, Texas 77843-2474, †Department ofComputer Science, Texas A&M University, College Station, Texas 77843, ‡Department of Statistics, Purdue University, West Lafayette,

Indiana 47906, §Department of Biology, University of Washington, Seattle, Washington 98195-5325 and **Department ofAgronomy, University of Wisconsin, Madison, Wisconsin 53706

Manuscript received July 7, 2005Accepted for publication September 19, 2005

ABSTRACT

Polyploidy has occurred throughout the evolutionary history of all eukaryotes and is extremely commonin plants. Reunification of the evolutionarily divergent genomes in allopolyploids creates regulatoryincompatibilities that must be reconciled. Here we report genomewide gene expression analysis ofArabidopsis synthetic allotetraploids, using spotted 70-mer oligo-gene microarrays. We detected .15%transcriptome divergence between the progenitors, and 2105 and 1818 genes were highly expressed inArabidopsis thaliana and A. arenosa, respectively. Approximately 5.2% (1362) and 5.6% (1469) genes dis-played expression divergence from the midparent value (MPV) in two independently derived synthetic allo-tetraploids, suggesting nonadditive gene regulation following interspecific hybridization. Remarkably, themajority of nonadditively expressed genes in the allotetraploids also display expression changes betweenthe parents, indicating that transcriptome divergence is reconciled during allopolyploid formation.Moreover, .65% of the nonadditively expressed genes in the allotetraploids are repressed, and .94% ofthe repressed genes in the allotetraploids match the genes that are expressed at higher levels in A. thalianathan in A. arenosa, consistent with the silencing of A. thaliana rRNA genes subjected to nucleolar dom-inance and with overall suppression of the A. thaliana phenotype in the synthetic allotetraploids andnatural A. suecica. The nonadditive gene regulation is involved in various biological pathways, and thechanges in gene expression are developmentally regulated. In contrast to the small effects of genomedoubling on gene regulation in autotetraploids, the combination of two divergent genomes in allo-tetraploids by interspecific hybridization induces genomewide nonadditive gene regulation, providing amolecular basis for de novo variation and allopolyploid evolution.

WHOLE-genome duplication may occur via auto-polyploidization by multiplying a single genome

or via allopolyploidization by combining two or moredivergent genomes (Grant 1971; Stebbins 1971). Thecommon occurrence of allopolyploidy in many plant(Stebbins 1971; Masterson 1994) and some animal(Becak and Kobashi 2004) species in nature suggestsan evolutionary advantage of allopolyploids over theirprogenitors and implicates allopolyploidy as a rapidspeciation process (Soltis and Soltis 2000; Wendel

2000). The combination of homeologous chromosomesfrom divergent species not only promotes functional di-vergence of duplicate genes (Adams et al. 2003; Blancand Wolfe 2004), but also generates heterozygosityand novel interactions leading to genetic and pheno-typic variability and heterosis (Ramsey and Schemske

1998; Soltis and Soltis 2000; Wendel 2000; Osborn

et al. 2003) that are stably maintained in the disomicallopolyploids. The data document rapid changes, suchas de novo phenotypic variation, transposon activation,nucleolar dominance, gene loss and silencing, and sub-functionalization (Song et al. 1995; Chen and Pikaard1997a; Pikaard 1999; Comai et al. 2000; Wendel 2000;Ozkan et al. 2001; Kashkush et al. 2002, 2003; Adamset al. 2003, 2004; He et al. 2003; Osborn et al. 2003; Wang

et al. 2004) in allopolyploids, which are caused by mech-anisms involving dosage compensation, regulatory incom-patibility, genetic alteration, and epigenetic modifications(Osborn et al. 2003). Evidently, polyploidy is a prominentand pervasive force in plant evolution (Soltis and Soltis2000; Wendel 2000), in contrast to the notion thatpolyploidy has contributed little to progressive evolution(Stebbins 1971).

Despite the general importance and increased in-terest in understanding the mechanisms and evolutionof polyploidy (Soltis and Soltis 2000; Wendel 2000;Wolfe 2001; Osborn et al. 2003), little is known aboutgenomewide effects on the expression of progenitors’

1Present address: Department of Biology, University of Puget Sound,Tacoma, WA 98416.

2Corresponding author: Institute for Cellular and Molecular Biology,University of Texas, 1 University Station, A-4800, Austin, TX 78714-0159.E-mail: zjchen@mail.utexas.edu

Genetics 172: 507–517 ( January 2006)

genes between the diverged genomes in nascent allo-polyploids. We have produced Arabidopsis allotetra-ploids using interspecific hybridization between twotetraploid species, Arabidopsis thaliana (Ler) and A. arenosa(Comai et al. 2000; Chen et al. 2004; Wang et al. 2004),and tested the consequences of interspecific hybridiza-tion on gene expression during early stages of allotetra-ploid formation. We report the first comprehensiveanalysis of transcriptome divergence between the pro-genitors and their allotetraploid lineages. Approximately3900 genes (�15%) were differentially expressed betweenA. thaliana and A. arenosa. The majority of nonadditivelyexpressed genes in the synthetic allotetraploids dis-played expression divergence between the parents andwere involved in various biological pathways, which mayprovide a molecular basis of de novo variation for theselection and adaptation of new allopolyploid species.

MATERIALS AND METHODS

Plant materials and RNA samples: Plant materials includedA. thaliana isogenic autotetraploids (At4, accession no.CS3900) and diploid (At2, Ler), A. arenosa (Aa, accession no.CS3901), and synthetic allotetraploid lines (Allo733 and 738)(accession no. CS3895–3896). These plant materials weregenerated as previously described (Comai et al. 2000; Wang

et al. 2004). All plants were grown in a growth chamber at22� and under 16 hr of light per day at the University ofWashington with two biological replications. Leaves werecollected from 20 plants in each biological replication priorto bolting (with seven to eight rosette leaves) in each line tominimize developmental variation among species and bulkedfor DNA and RNA analyses (Madlung et al. 2002; Wang et al.2004). Another 20 plants were grown until flowering, andflower buds were harvested when the first flower bloomed inindividual plants.

Total RNA was isolated using Trizol reagent (Invitrogen,San Diego) according to the manufacturer’s recommenda-tions. Each RNA sample was quantified by measuring 260/280ratios using a UV-spectrometer (GeneQuant pro; AmershamBiosciences, Arlington Heights, IL) and by agarose–formaldehydegel electrophoresis. Total RNA was subjected to mRNA isolation,using a Micro-FastTrack 2.0 mRNA isolation kit (Invitrogen).Equal amounts of mRNA from A. thaliana and A. arenosa weremixed as a midparent value to detect nonadditive gene ex-pression in the allotetraploids.Fluorescence in situ hybridization: Fluorescence in situ

hybridization (FISH) in anther meiocytes was performed usingA. thaliana- or A. arenosa-specific 180-bp centromeric repeats asprobes (Comai et al.2003). The chromosomal images in meioticcells were analyzed using a Zeiss Axiovert microscope.Analysis of spotted oligo-gene microarrays: Spotted Arabi-

dopsis 70-mer oligo-gene microarrays (microarray data areavailable at http://microarrayabc.tamu.edu/pub_data/26k/26kmicroarrayset.htm) using 26,090 annotated genes werecooperatively developed with QIAGEN (Valencia, CA) andOperon (Alameda, CA) (Lee et al. 2004; Tianet al. 2005; Wang

et al. 2005). The 70-mer oligo was designed from the 39-end ofeach annotated gene. Every feature was spotted once on eachslide. We used 500 ng of mRNA in each labeling reaction usingCy3- or Cy5-dCTP (Amersham Biosciences). The Cy3-dCTPreaction was mixed with the Cy5-dCTP reaction for onehybridization, and then an equal amount of RNA samples

was reversely labeled for another hybridization (supplementalFigure 1 at http://www.genetics.org/supplemental/). There-fore, two ‘‘identical’’ samples each containing an equal amountof Cy3- and Cy5-labeled cDNAs were hybridized with two slides,which constitutes one dye swap. The dye-swap experimentwas replicated using an independently isolated RNA sample.Each experiment contains four dye swaps (eight slide hy-bridizations) or two dye swaps per biological replication(supplemental Tables 1 and 2 at http://www.genetics.org/supplemental/) (Chen et al. 2004; Tian et al. 2005).

A total of 48 slide hybridizations were performed for sixexperimental comparisons to determine changes in expres-sion between the progenitors (At4 and Aa), Allo733 andmidparent value (MPV) (leaves), Allo738 and MPV (leaves),Allo733 and MPV (flower buds), Allo738 and MPV (flowerbuds), and At2 and At4 (leaves) (supplemental Table 2 athttp://www.genetics.org/supplemental/). Probe labeling, slidehybridization, and washing were performed as previouslydescribed (Tian et al. 2005). Raw data were collected usingGenepix Pro4.1 after the slides were scanned using Genepix4000B. The data were processed using a lowess function toremove nonlinear components and analyzed using a linearmodel (Lee et al. 2004). This linear model was employed topartition variation in the observed data relative to technicaland biological variation. Given that each feature is repre-sented once on an array, the linear model is

Yijklm ¼ m1Ai 1Dj 1Tk 1Gl 1AGil 1DGjl 1TGkl 1TDGjkl 1 eijklm ;

where m represents the overall mean effect; A, D, T, and Grepresent main fixed effects from the array, dye, treatment(e.g., RNA from two species), and gene, respectively; and i ¼1, . . . , 8, k ¼ 1, 2, j ¼ 1, 2, and l ¼ 1, . . . , 27,648 (including26,090 70-mer Arabidopsis oligos plus controls). The interactionterms AG, DG, TG, and TDG represent array-by-gene, dye-by-gene, treatment-by-gene, and treatment-by-dye-by-gene inter-actions, and eijklm denotes the random error and is used to testfor significance of main and interaction effects in the model.Due to confounding and/or aliasing issues involving the array,dye, and treatment terms, not all two-way interactions areincluded in the model. The model residuals are assumed tobe normally distributed with a common variance [i.e., eijklmiid N(0, s2)], unless evidence of variance nonconstancy isobserved. In such a case, a per gene variance is assumed [i.e.,eijklm independent N(0, sl

2)].We tested differential expression using significant differ-

ences in T1 TG terms for a particular gene (Black 2002) be-cause we are interested in changes in expression beyond theaverage treatment effect. The hypotheses that reflect whethera gene, g, has undergone differential expression betweentreatments, t and t9 (e.g., A. thaliana and A. arenosa) are

H0: Tt 1TGtg ¼ Tt9 1TGt9g

H1: Tt 1TGtg 6¼ Tt9 1TGt9g:

A standard t-test statistic is used for this comparison, basedon the normality assumption for the residuals. To control formultiple testing errors the false discovery rate (FDR) ofBenjamini and Hochberg (1995) was employed as it providesweak control of the familywise error rate (FWER) and controlsthe FDR below level a. The FDR is defined as the expectedproportion of incorrect rejections of H0, relative to the totalnumber of rejections. The significance level a ¼ 0.05 waschosen for these investigations. All analyses of variance mod-els were fit using standard statistical packages (SAS, R, andMatlab) (Moser et al. 1988; Ihaka and Gentleman 1996).

508 J. Wang et al.

As mentioned previously, the common variance assumptionwas used for all genes and per-gene variances for individualgenes to estimate the significant changes of gene expressionbetween the two treatments. The genes that were expresseddifferently at a statistically significant level (FDR, a ¼ 0.05)using a common variance had largefold changes, some ofwhich also had high standard deviations, whereas the genesthat were expressed significantly differently using a per-genevariance included those with smallfold changes, which may bedifficult to verify. We used a conservative approach andselected the genes that were expressed significantly differentlyunder both statistical tests.

Functional categories of up- and downregulated genes wereclassified using PENDANT (http://mips.gsf.de/proj/thal/db/index.html) and compared using Venn diagrams. Thenonadditively expressed genes (identities) were mapped tooligonucleotide sequences using Perl scripts. The oligos weremapped to genomic coordinates using high (red) and low(blue) gradients corresponding to gene densities. Verticallines above and below the chromosomes showed up- anddownregulation, respectively, and the length was proportionalto the logarithmfold changes in differential gene expression.

RT–PCR, qRT–PCR, and SSCP analyses: Approximately 5mg of total RNA was treated with DNase I, and first-strandcDNA was synthesized using reverse transcriptase (RT) Super-script II (Invitrogen) according to the manufacturer’s recom-mendations. An aliquot (1/100) of cDNA was used as templatein quantitative (or real-time) RT–PCR (qRT–PCR), single-strand conformation polymorphism (SSCP) (Adams et al.2003), and cleaved amplified polymorphism sequence (CAPS)analyses (Wang et al. 2004). qRT–PCR was performed in anABI7500 machine (ABI Biosystems, Columbia, MD), using theprimers (supplemental Table 5 at http://www.genetics.org/supplemental/) and SYBER green dye method as previouslydescribed (Lee et al. 2004), except that ACT2 was used as acontrol to estimate the relative expression levels of the genestested. The expression levels were converted to log-ratios (sup-plemental Table 4 at http://www.genetics.org/supplemental/)in comparison with the microarray data. For SSCP and CAPSanalyses, the primers were from A. thaliana loci and used toamplify both A. thaliana and A. arenosa loci. The PCR reactionswere performed using one cycle of 94� for 2 min followedby 25–30 cycles of amplification at 94� for 30 sec, 53� for30 sec, and 72� for 90 sec. The amplified products were di-gested by a restriction enzyme and subjected to agarose gelelectrophoresis (CAPS analysis) or denatured in a loadingbuffer and resolved in a 0.53 mutation detection enhance-ment (MDE) gel (SSCP analysis). The images were cap-tured, and band intensities were quantified using a FujifilmPhosphorimager.

RESULTS

Genetically stable allotetraploids resembled the A.arenosa parent: Allotetraploids can be formed through acombination of unreduced gametes or interspecifichybridization between diploid species followed bychromosome doubling (Grant 1971; Stebbins 1971).To study early events of gene regulation in syntheticallopolyploids, we created independent allotetraploidlineages (Allo733 and Allo738) by pollinating A. thali-ana autotetraploids (At4) with A. arenosa tetraploids(Aa) (Comai et al. 2000; Wang et al. 2004) (Figure 1),which appear to have the same number of genes at thesame ploidy levels (Comai et al. 2000). Heterozygosity

in the allotetraploid progeny was minimized by self-pollination for five generations. The chromosomenumbers and parental origins were verified in the firstand fourth generations, using FISH (Comai et al. 2003)and informative microsatellite markers (data not shown).Without exception, each line possessed five pairs of chro-mosomes fromA. thaliana and eight pairs fromA. arenosa(Figure 1) (Comai et al. 2000, 2003). The morphology ofthese plants varied between lineages, coincident withrapid genetic and epigenetic changes observed in newallopolyploids (Comai et al. 2000; Madlung et al. 2002;Wang et al. 2004). Many allotetraploid lineages resem-bled the A. arenosa parent and A. suecica, a natural al-lotetraploid (Pikaard 1999; Comai et al. 2000; Madlung

et al. 2002). These morphological characteristics includelong leaves, tall stature, many branches, deeply serratedrosette leaves, and large rosettes and flowers. The data

Figure 1.—(A) Production of stable synthetic allotetra-ploids (Allo733 and 738). A self-fertile A. thaliana autotetra-ploid (Ler, At4) was pollinated with a natural A. arenosatetraploid (Aa). Multiple independent allotetraploids in S1were self-pollinated by single-seed descent to the S5 genera-tion. Allo733 and Allo738 resembled A. arenosa and naturalA. suecica (Comai et al. 2000). Fluorescence in situ hybridiza-tion (FISH) analysis indicates that two sets of centromeres inAllo733 are derived from A. thaliana (At4) and A. arenosa(Aa), respectively. Bar, 5 mm. Allo, allotetraploid.

Gene Regulation in Arabidopsis Allotetraploids 509

indicate that A. arenosa appear to be morphologicallydominant over A. thaliana in the allotetraploids. Theflower colors varied from pink (like A. arenosa) in theearly generation (S1), to a mixture of pink and whiteflowers in the intermediate generations (S2–4), to whitecolors in the late generation (S5), suggesting the notionof stochastic and rapid changes in gene expression(Comai et al. 2000; Wang et al. 2004).

Transcriptome divergence between the progenitors:To determine the molecular basis of phenotypic differ-ences, we analyzed transcriptome changes in the pro-genitors, using spotted oligo-gene microarrays designedfrom A. thaliana annotated genes (Tian et al. 2005).Microarray data from four dye-swap experiments (i.e.,two dye swaps per biological replication) (supplementalFigure 1 at http://www.genetics.org/supplemental/) wereanalyzed using a linear model (Lee et al. 2004) and theresults were adjusted for multiple comparisons (Tianet al. 2005). Unless otherwise noted, we selected thedifferentially expressed genes that were statistically sig-nificant under both common and per-gene variances(Figure 2, Table 1).

We characterized transcriptome differences betweenA. thaliana and A. arenosa that diverged �5.8 MYA(Koch et al. 2000). We found that 3923 (�15%) geneswere differentially expressed between the progenitors,of which 2105 (�8%) and 1818 (�7%) were expressedat significantly higher levels in A. thaliana andA. arenosa,respectively (Figure 3A, Table 1). The differentiallyexpressed genes represented as much as �43% ofthe transcriptome, using a per-gene variance analysis(Table 1), indicating a wide range of gene expressiondifferences between the two species, which is reminis-cent of the .50% of transcriptome changes in Dro-sophila species that diverged �2.5 MYA (Ranz et al.2003). Among 11,199 differentially expressed genes,5232 (47%) genes were expressed at a higher level inA. thaliana than in A. arenosa, whereas 5967 (53%) geneswere expressed at a higher level in A. arenosa than inA. thaliana. In a separate study using Affymetrix chips,Schmid et al. detected several hundred genes thatwere expressed more than twofold differently between

A. thaliana Col and Ler ecotypes (Schmid et al. 2003).Although the two arrays employ different analyticaltools, it appears that the gene expression differencesdetected between species are much greater than thosebetween ecotypes.

Genomewide nonadditive gene regulation in the allo-tetraploids: To determine how transcriptome divergence

Figure 2.—Logarithm fold change vs. per-gene standarddeviation in a replicated experiment containing four dyeswaps. The hybridization probes were cDNAs from Allo733and two parents, A. thaliana tetraploid and A. arenosa. Thedata were analyzed using a linear model as previously de-scribed (Tian et al. 2005). Green, black, and red dots indicatethe pools of significant genes selected by multiple comparisontests (false discovery rate, FDR, a¼ 0.05) using a per-gene var-iance, a common variance, and the intersection of the two,respectively. Statistical significance for extremely smallfoldchanges was detected for two features replicated 6 and 49times within each microarray slide, indicating the power ofreplication in microarray experiments.

TABLE 1

The number of differentially expressed genes detected using a common variance and/or a per-gene variance

False discovery rate (FDR) (a ¼ 0.05)

Experiment Common variance Per-gene variance Shared 61.5-fold (from per-gene variance)

At4 vs. Aa 4,363 11,199 3,923 4,476Allo733 vs. MPV 1,708 8,377 1,362 1,792Allo738 vs. MPV 1,856 9,875 1,469 2,358

At4, A. thaliana autotetraploid; Aa, A. arenosa tetraploid; MPV, midparent value; Shared, shared data sets ofthe statistically significant genes detected using both common variance and per-gene variance. The last columnindicates the number of significant genes using an arbitrary cut for fold change (61.5) from the genes selectedon the basis of per-gene variance.

510 J. Wang et al.

contributes to genetic and morphological variation inallotetraploids, we compared mRNA abundance in an al-lotetraploid with the MPV (an equal mixture of RNAsfrom two parents) (supplemental Figure 1 at http://www.genetics.org/supplemental/). Violating the nullhypothesis for no gene expression difference betweenthe allotetraploid and midparent value suggests thata gene(s) is nonadditively expressed; however, we can-not detect the situation where silencing of a locus iscompensated by increased expression of its homeolo-gous locus. Thus, microarray analysis may underestimatethe number of genes that are differentially expressedbetween an allotetraploid line and the parents. Wediscovered that 1362 (�5.2%) and 1469 (�5.6%) geneswere expressed nonadditively in Allo733 and Allo738,respectively (Table 1). When a per-gene variance wasused, the nonadditively expressed genes accounted for�32% (8377, Allo733) and �38% (9875, Allo738) of thetranscriptome. The data suggest that orthologous genesin allopolyploids are frequently expressed in a non-additive fashion.

If the regulatory changes inherited from the parentsdetermine species divergence, the genes displayingspecies-specific expression patterns may be modulatedin the allotetraploids. Indeed, among the 2011 genesthat were nonadditively regulated in two allotetraploids,1377 (�68%) genes were included in those that weredifferentially expressed between the parents (Figure3B), which are significantly different from a randomdistribution of nonadditively expressed genes (�15%,x2 ¼ 1180.5, P # 0.00001). Among them, 820 (�41%)genes were common to both allotetraploids (Allo733and Allo738), whereas 649 (�32%) and 542 (�27%)genes were unique to All733 and Allo738, respectively,indicating general and specific effects of allopolyploidformation on gene regulation in the independentlyderived allotetraploids. The 820 nonadditively expressedgenes in both allotetraploids (Allos) were randomlydistributed across the genome and displayed no obvi-

ous chromosomal regions susceptible to allopolyploidy-dependent gene regulation (Figure 3C).Progenitor-biased gene repression in the allotetra-

ploids: We analyzed direction of change and parentalorigin of nonadditively expressed genes. Among them,1038 (�76%) and 952 (�65%) genes were downregu-lated in Allo733 and Allo738, respectively (Figure 4A),suggesting that repression is a mode of nonadditivegene regulation in synthetic allotetraploids. We dividedthe repressed genes into three categories on the basis oftheir expression patterns in the parents. First, 838(�99%) and 611 (�94%) genes that showed higherlevels of expression in A. thaliana than in A. arenosa wererepressed in Allo733 and Allo738, respectively (Figure4B), which coincides with the silencing ofA. thaliana butnot of A. arenosa rRNA genes (Chen et al. 1998; Pikaard1999) and the overall suppression of A. thaliana pheno-type in new allotetraploids and in natural A. suecica.Second, 90 (�35%) and 159 (�50%) genes that wereexpressed at higher levels in A. arenosa than in A.thaliana were downregulated in Allo733 and All738,respectively (Figure 4C). Third, 110 (�42%) and 182(36%) genes that were equally expressed in A. thalianaautotetraploid and in A. arenosa were repressed inAllo733 and Allo738, respectively (Figure 4D). Therewas no bias toward gene repression in the last twocategories. The data demonstrate that the genes morehighly expressed in A. thaliana autotetraploids than inA. arenosa are subject to orchestrated repression in thesynthetic allotetraploids.Nonadditive gene regulation in various biological

pathways: According to 15 functional classifications of820 nonadditively expressed genes detected in bothallotetraploids (Figure 5A), the percentages of genes inthe hormonal regulation and cell defense and agingcategories were 150–175% of those in the same catego-ries classified using all annotated genes in Arabidopsis(Figure 5B), suggesting that these genes are particularlysusceptible to expression changes in response to the

Figure 3.—Transcriptome di-vergence and nonadditive geneexpression between allotetraploidsand their progenitors. (A) Theproportion of transcriptome thatwas highly expressed in A. thaliana(At4),A. arenosa (Aa), or equally ex-pressed (both). (B) Venn diagramshowing the number of genes withexpression divergence between theprogenitors (blue) and betweenAllo733 (red) or Allo738 (green)and the midparent value (MPV, sup-plemental Figure 1 at http://www.genetics.org/supplemental/). (C)Chromosomal distribution of the820 genes displaying nonadditiveexpression in both allotetraploids(see text).

Gene Regulation in Arabidopsis Allotetraploids 511

perturbation resulting from intergenomic interactionsin the allotetraploids. Many genes involved in theethylene biosynthesis pathway were repressed in oneor two allotetraploids (Figure 6A and supplementalTable 3 at http://www.genetics.org/supplemental/),which may induce expression changes in ethylene-responsive genes involved in a wide range of develop-mental processes and fitness responses, including seedgermination, leaf and flower senescence, fruit ripening,

programmed cell death, and biotic and abiotic stressresponses (Guo and Ecker 2004). Of the 97 HSPs inArabidopsis (Arabidopsis Genome Initiative 2000),33 displayed expression differences from the midparentvalue (Figure 6B). Thirty-one HSPs that were highlyexpressed in A. thaliana were repressed, which mayreflect ‘‘buffering’’ effects (Queitsch et al. 2002) onpathway redundancy. Notably, fewer than expectedtransposons altered expression in the allotetraploids(Figure 5B), although some might be included in theunclassified category. This appears to be inconsistentwith B. McClintock’s notion of ‘‘genomic shock’’(McClintock 1984) but we note that these allopoly-ploid lineages represent the survivors among theoriginal F1 products (Comai et al. 2000) and in the lategeneration (S5).

Developmental and parental contributions to non-additive gene regulation: We tested whether nonaddi-tive gene regulation in synthetic allotetraploids issensitive to developmental changes by comparing thegene expression divergence detected in leaves and flow-ers. Allo733 displayed 1355 genes nonadditively ex-pressed in flower buds, of which 175 (�7%) were alsodetected in the leaves (Figure 7A), and 1180 were non-additively expressed only in flowers. Little overlap ofthe genes detected between leaves and flowers sug-gests a developmental role in nonadditive gene regula-tion in the allopolyploids in a manner reminiscent ofdevelopmental derepression of silenced rRNA genes(Chen and Pikaard 1997b) and subfunctionalization ofsome duplicate genes (Adams et al. 2003). It is notablethat gene expression changes may occur during thetransition from vegetative to reproductive development,but appear to be consistent within a tissue type (e.g.,rosette leaves) (Chen and Pikaard 1997b). Comparedto Allo738, fewer nonadditively expressed genes weredetected in the flower buds in Allo733 (supplemen-tal Table 1 at http://www.genetics.org/supplemental/),which may reflect developmental variation amongallotetraploid lineages (Comai et al. 2000; Madlung

et al. 2002).We verified expression patterns of 11 nonadditively

expressed genes using qRT–PCR analysis (Figure 7B).Six were repressed and 5 were upregulated in theallotetraploids, consistent with the microarray data(supplemental Table 5 at http://www.genetics.org/supplemental/). Five genes (WRKY, BCB, HSP90, PDF,andLRR) that were expressed at higher levels inA. thalianathan in A. arenosa (At4 . Aa) were repressed in theallotetraploids. Three of four genes (FLC, PORa, andPORb) that displayed higher expression levels in A.arenosa than in A. thaliana (Aa . At4) were upregulated,and one (CYC) was repressed in the allotetraploids. Twogenes (CHI and SPP) that were equally expressed in theparents (At4 ¼ Aa) were upregulated in the allotetra-ploids. Using locus-specific SSCP or CAPS assays, weanalyzed the contribution of A. thaliana and A. arenosa

Figure 4.—Downregulation of A. thaliana genes in the syn-thetic allotetraploids. (A) Distribution of nonadditively ex-pressed genes detected in each allotetraploid (Allo733 orAllo738) or both allotetraploids (Allos). (B) The nonaddi-tively expressed genes in each allotetraploid matched thegenes that were highly expressed in the A. thaliana autotetra-ploid. (C) The nonadditively expressed genes in each allote-traploid matched the genes that were highly expressed in A.arenosa. (D) The nonadditively expressed genes matched thegenes that were equally expressed in both parents. The per-centages of downregulated genes are indicated above the col-umns in each histogram.

512 J. Wang et al.

loci to the nonadditive gene regulation (Figure 7C). ForWRKY, BCB, and COL2, both A. thaliana and A. arenosaloci were repressed in the allotetraploids, whereasHSP17.6b repression was due to the A. thaliana locus.The repression of several COLs and upregulation of FLCmay correlate with late flowering in the allotetraploids.Similarly, upregulation of PORb and SPP in the allote-traploids was related to A. thaliana loci, whereas upre-gulation of CHI was caused by the A. arenosa locus. Thedata suggest both cis-regulatory and trans-acting effects(Wittkopp et al. 2004) on nonadditive gene regulationin the allotetraploids. Furthermore, upregulation ofSPP encoding starch phosphorylase and of PORa andPORb encoding protochlorophyllide oxidoreductases inthe photosynthetic pathway may lead to vigorous growthin the allotetraploids.

Autopolyploidization does not induce genomewidenonadditive gene regulation: To determine whethernonadditive gene regulation is affected by genomedosage, we analyzed transcriptome differences betweenthe A. thaliana diploid (At2) and the isogenic autote-traploid (At4) (supplemental Table 1 at http://www.

genetics.org/supplemental/). Only 88 genes were ex-pressed significantly differently between the diploidand the autotetraploid, which is reminiscent of thedosage-dependent regulation of a dozen genes as ob-served in yeast autoploids (Galitski et al. 1999). Theresults suggest that doubling the same genome in auto-polyploids has much smaller effects on gene regulationthan combining the divergent genomes in allopoly-ploids. However, allopolyploidy effects may not be assimple as the sum of ‘‘hybridization’’ and ‘‘genome dou-bling’’ (see discussion).

DISCUSSION

Effects of autopolyploidization and allopolyploidiza-tion on gene regulation: Polyploidy effects on generegulation may be caused by genome doubling and/orintergenomic interactions. Autopolyploidization indu-ces gene expression changes in response to the increasein genome dosage (Birchler 2001). Only 12 and 88genes, respectively, respond to autoploidy changes inyeast (Galitski et al. 1999) and Arabidopsis, suggesting

Figure 5.—Classification ofnonadditively expressed genesdetected in synthetic Arabidop-sis allotetraploids. (A) The 820genes detected in both Allo733and Allo738 lines (Allos) wereclassified into 15 functionalcategories using the PEDANTanalysis system (http://mips.gsf.de/proj/thal/db/index.html)(ArabidopsisGenome Initiative2000). (B) The percentages of thegenes in each functional categorydetected in Allo733, Allo738, orboth (Allos). The relative ratiosin the y-axis were estimated us-ing the percentage of genesdetected in each functional cat-egory in an Allo line divided bythe percentage of all �26,000annotated genes in the Arabi-dopsis genome (ArabidopsisGenome Initiative 2000). Thepercentage of the genes de-tected in the Allo line equal tothat of all genes in the whole ge-nome is shown as 100% (dashedline).

Gene Regulation in Arabidopsis Allotetraploids 513

that increasing genome dosage affects a small subset ofgenes. During autopolyploidization, mechanisms suchas dosage compensation (Birchler 2001) are respon-sible for maintaining expression patterns of the genesexcept those associated with the large size of polyploidcells (Galitski et al. 1999). For the majority of genesstudied in maize, their expression levels are dependenton the dosage of chromosomes or chromosome arms(Guo et al. 1996; Auger et al. 2005).

The dramatic changes in nonadditive gene regulationobserved in the allotetraploids may be induced byinterspecific hybridization. Our data suggest that 15%of the transcriptome diverged between A. thaliana andA. arenosa, which accounted for 68% of nonadditivelyexpressed genes (2011) in the synthetic allotetraploids.In addition to 820 genes that changed expression inboth allotetraploids, 649 and 542 genes were unique toAllo733 and Allo738, respectively. These genes maycorrelate with specific changes in individual allotetra-ploids and facilitate selection and adaptation of newallopolyploid species in response to environmental cuesand developmental changes. Indeed, nonadditive gene

regulation is developmentally regulated, which maylead to subfunctionalization of duplicate genes (Lynchand Force 2000; Adams et al. 2003) in different organsor tissues (Chen and Pikaard 1997b). Transposons areunderrepresented in the genes that display expressionchanges in the allotetraploids. It is likely that manytransposons are not included in the annotated genes formicroarray analysis. Alternatively, the effects of genomicshock (McClintock 1984) may be ‘‘settled’’ in theselfing progeny (S5). Finally, dosage-dependent generegulation (Birchler 2001; Auger et al. 2005) mayaccount for part of the gene expression changes in theallotetraploids. Indeed, 51% (45/88) and 32% (28/88)of the genes that display expression divergence betweenA. thaliana diploids and isogenic autotetraploids werealso expressed nonadditively in two allotetraploids.

There is a possibility that 70-mer oligos designed fromA. thaliana may not hybridize well to the A. arenosagenes, although we have shown that 192 A. thalianaoligos hybridized equally well to A. arenosa and Brassicagenes (Lee et al. 2004), probably because of.95% genicsequence identity between A. thaliana and A. arenosa

Figure 6.—Nonadditive gene regulation oc-curs in various pathways. (A) Progenitor-dependentrepression of the genes involved in the ethylenebiosynthesis pathway in Arabidopsis allotetra-ploids. Each number in parentheses below an en-zyme or molecule in the pathways indicates thefold change for the expression of a gene, homo-log (h), or putative (p) homolog detected by mi-croarray analysis. Red, green, blue, and purplecolors indicate that gene expression differencesare detected in both allotetraploids (Allo733and -738), in Allo738, in Allo733, and betweenthe two parents, respectively. (B) Repression ofheat-shock protein (HSP) genes in Arabidopsisallotetraploids. Thirty-one out of 33 HSPs were re-pressed in each allotetraploid (Allo733 or -738).The length and directions of vertical bars indi-cate logarithm fold changes in up- (above theline) or downregulation (below the line) of theHSPs relative to the midparent in the allotetra-ploid lines.

514 J. Wang et al.

(Lee and Chen 2001) and .85% of it between A.thaliana and B. oleracea (Cavell et al. 1998). As a result,the sequence divergence may also contribute to the dif-ference in gene expression detected betweenA. thalianaand A. arenosa.

Insights into nonadditive gene regulation in thesynthetic allotetraploids: In Arabidopsis allotetraploids,the progenitor-dependent gene regulation is not re-stricted to rDNA loci subjected to nucleolar dominance(Pikaard 1999) but occurs at a genomewide scale invarious biological pathways. The available data suggestthat the expression of orthologous genes during evolu-tion and speciation is not purely neutral. Selection andadaptation over evolutionary time may promote di-vergence of regulatory elements and/or transcriptionfactors and regulatory proteins. The competition be-tween the diverged regulatory pathways may determinenonadditive gene regulation in allopolyploids of Arabi-dopsis (Wang et al. 2004), cotton (Adams et al. 2004),Senecio (Hegarty et al. 2005), and wheat (Kashkushet al. 2002; He et al. 2003), in interspecific hybrids(Wittkopp et al. 2004) in Drosophila and intraspecifichybrids in maize (Guo et al. 2004; Auger et al. 2005),and in sex-dependent gene regulation in Drosophila(Ranz et al. 2003; Gibson et al. 2004). It is notable thatoutcrossing in A. arenosa and inbreeding in A. thalianamay accelerate their divergence during evolution. Eachprogenitor might have evolved specific regulatory sys-tems affecting rDNA and other loci, perhaps via con-certed evolution (Coen et al. 1982), and the interactions

between these diverged regulatory systems in allo-polyploids may trigger repression of the A. thaliana-‘‘specific’’ genes and of the rDNA loci (Chen andPikaard 1997a; Pikaard 1999; Wang et al. 2004). Al-though the underlying mechanisms for preferentialrepression of A. thaliana genes are yet to be determined,sudden reunification of divergent genomes may inducegenome instability (Madlung et al. 2002; Wang et al.2004) and changes in chromatin structure and RNA-mediated processes (Osborn et al. 2003; Chen et al.2004). Hybrid- or allopolyploidy-induced incompatibil-ities may be overcome by gene expression modulationthrough chromatin modifications, transcription factorssuch as Myb (Barbash et al. 2003), and/or RNA in-terference. Interestingly, nonadditive gene regulationin the allotetraploids depends largely on expressiondivergence between the parents. Thus, hybrids derivedfrom distantly related species may induce a high level ofgene expression changes in a nonadditive fashion,providing molecular bases of hybrid vigor (Birchleret al. 2003) and of novel variation in the allotetraploidprogeny (Comai et al. 2000; Wang et al. 2004). Further-more, the stochastic establishment of nonadditive generegulation in newly synthesized allotetraploids (Wang

et al. 2004) may increase the potential for fitness andselective adaptation. In contrast to the lethality andsterility observed in interspecific hybrids (Barbash et al.2003), nonadditive gene expression changes may bemaintained and transmitted in meiotically stable allo-polyploids, providing a mechanism for de novo variation

Figure 7.—Developmen-tal and parental contribu-tions to nonadditive geneexpression. (A) Venn dia-grams of the genes thatdisplayed nonadditive ex-pression in leaves andflower buds in Allo733.Only 175 of 2542 genesshowed overlap betweenleaves and flower buds.(B) Verification of 11 genesdetected in microarrays byqRT–PCR. The gene ex-pression levels in the pa-rents were higher in A.thaliana (At . Aa), higherin A. arenosa (Aa . At4),and equal (At4 ¼ Aa). (C)SSCP and CAPS analysesshowing parental contribu-tions to nonadditive generegulation in the allotetra-ploids. The genes stud-ied in B and C are in thefunctional classifications ofstress (HSP or HSP90 and

HSP17.6b); cell cycle, defense, and aging (CYC, CHI, LRR, PDF, and WRKY); metabolism and energy (BCB, PORa, PORb, andSPP); and flower development (COL2 and FLC). The restriction enzymes used in CAPS analysis are indicated in parentheses.

Gene Regulation in Arabidopsis Allotetraploids 515

and evolutionary opportunities for selection and adap-tation of new allopolyploid species.

We thank James A. Birchler, Gary E. Hart, Robert A. Martienssen,J. Chris Pires, Douglas E. Soltis, Jennifer Tate, and Jonathan F. Wendelfor critical suggestions. This work was supported by a grant from theNational Science Foundation Plant Genome Research Program(DBI0077774). Work in the Chen lab is supported in part by a grantfrom the National Institutes of Health (GM067015). The authorsdeclare that they have no financial conflict of interest.

LITERATURE CITED

Adams, K. L., R. Cronn, R. Percifield and J. F. Wendel, 2003 Genesduplicated by polyploidy show unequal contributions to the tran-scriptome and organ-specific reciprocal silencing. Proc. Natl.Acad. Sci. USA 100: 4649–4654.

Adams, K. L., R. Percifield and J. F. Wendel, 2004 Organ-specificsilencing of duplicated genes in a newly synthesized cotton allo-tetraploid. Genetics 168: 2217–2226.

Arabidopsis Genome Initiative, 2000 Analysis of the genome se-quence of the flowering plant Arabidopsis thaliana. Nature 408:796–815.

Auger, D. L., A. D. Gray, T. S. Ream, A. Kato, E. H. Coe, Jr. et al.,2005 Nonadditive gene expression in diploid and triploid hy-brids of maize. Genetics 169: 389–397.

Barbash, D. A., D. F. Siino, A. M. Taroneand J. Roote, 2003 A rap-idly evolving MYB-related protein causes species isolation in Dro-sophila. Proc. Natl. Acad. Sci. USA 100: 5302–5307.

Becak, M. L., and L. S. Kobashi, 2004 Evolution by polyploidy andgene regulation in Anura. Genet. Mol. Res. 3: 195–212.

Benjamini, Y., and Y. Hochberg, 1995 Controlling the false discov-ery rate: a practical and powerful approach to multiple testing.J. R. Stat. Soc. Ser. B 57: 289–300.

Birchler, J. A., 2001 Dosage-dependent gene regulation in multi-cellular eukaryotes: implications for dosage compensation, aneu-ploid syndromes, and quantitative traits. Dev. Biol. 234: 275–288.

Birchler, J. A., D. L. Auger and N. C. Riddle, 2003 In search of themolecular basis of heterosis. Plant Cell 15: 2236–2239.

Black, M. A., 2002 Statistical issues in the design and analysis ofspotted microarray experiments. Ph.D. Thesis, Department ofStatistics, Purdue University, West Lafayette, IN.

Blanc, G., and K. H. Wolfe, 2004 Functional divergence of dupli-cated genes formed by polyploidy during Arabidopsis evolution.Plant Cell 16: 1679–1691.

Cavell, A. C., D. J. Lydiate, I. A. Parkin, C. Dean and M. Trick,1998 Collinearity between a 30-centimorgan segment of Arabi-dopsis thaliana chromosome 4 and duplicated regions within theBrassica napus genome. Genome 41: 62–69.

Chen, Z. J., and C. S. Pikaard, 1997a Epigenetic silencing of RNApolymerase I transcription: a role for DNA methylation and his-tone modification in nucleolar dominance. Genes Dev. 11: 2124–2136.

Chen, Z. J., and C. S. Pikaard, 1997b Transcriptional analysis of nu-cleolar dominance in polyploid plants: biased expression/silenc-ing of progenitor rRNA genes is developmentally regulated inBrassica. Proc. Natl. Acad. Sci. USA 94: 3442–3447.

Chen, Z. J., L. Comai and C. S. Pikaard, 1998 Gene dosage and sto-chastic effects determine the severity and direction of uniparen-tal ribosomal RNA gene silencing (nucleolar dominance)in Arabidopsis allopolyploids. Proc. Natl. Acad. Sci. USA 95:14891–14896.

Chen, Z. J., J. Wang, L. Tian, H.-S. Lee, J. J. Wang et al., 2004 Thedevelopment of an Arabidopsis model system for genome-wideanalysis of polyploidy effects. Biol. J. Linn. Soc. 82: 689–700.

Coen, E. S., J. M. Thoday and G. Dover, 1982 Rate of turnover ofstructural variants in the rDNA gene family of Drosophila mela-nogaster. Nature 295: 564–568.

Comai, L., A. P. Tyagi, K. Winter, R. Holmes-Davis, S. H. Reynoldset al., 2000 Phenotypic instability and rapid gene silencing innewly formed Arabidopsis allotetraploids. Plant Cell 12: 1551–1568.

Comai, L., A. P. Tyagi and M. A. Lysak, 2003 FISH analysis of mei-osis in Arabidopsis allopolyploids. Chromosome Res. 11: 217–226.

Galitski, T., A. J. Saldanha, C. A. Styles, E. S. Lander and G. R.Fink, 1999 Ploidy regulation of gene expression. Science 285:251–254.

Gibson, G., R. Riley-Berger, L. Harshman, A. Kopp, S. Vacha et al.,2004 Extensive sex-specific nonadditivity of gene expression inDrosophila melanogaster. Genetics 167: 1791–1799.

Grant, V., 1971 Plant Speciation. Columbia University Press,New York.

Guo, H., and J. R. Ecker, 2004 The ethylene signaling pathway: newinsights. Curr. Opin. Plant Biol. 7: 40–49.

Guo, M., D. Davis and J. A. Birchler, 1996 Dosage effects ongene expression in a maize ploidy series. Genetics 142: 1349–1355.

Guo, M., M. A. Rupe, C. Zinselmeier, J. Habben, B. A. Bowen et al.,2004 Allelic variation of gene expression in maize hybrids.Plant Cell 16: 1707–1716.

He, P., B. R. Friebe, B. S. Gill and J. M. Zhou, 2003 Allopolyploidyalters gene expression in the highly stable hexaploid wheat. PlantMol. Biol. 52: 401–414.

Hegarty, M. J., J. M. Jones, I. D. Wilson, G. L. Barker, J. A. Coghillet al., 2005 Development of anonymous cDNA microarrays tostudy changes to the Senecio floral transcriptome during hybridspeciation. Mol. Ecol. 14: 2493–2510.

Ihaka, R., and R. Gentleman, 1996 A language for data analysisand graphics. J. Comput. Graph. Stat. 5: 299–314.

Kashkush, K., M. Feldman and A. A. Levy, 2002 Gene loss, silenc-ing and activation in a newly synthesized wheat allotetraploid.Genetics 160: 1651–1659.

Kashkush, K., M. Feldman and A. A. Levy, 2003 Transcriptionalactivation of retrotransposons alters the expression of adjacentgenes in wheat. Nat. Genet. 33: 102–106.

Koch, M. A., B. Haubold and T. Mitchell-Olds, 2000 Compar-ative evolutionary analysis of chalcone synthase and alcohol de-hydrogenase loci in Arabidopsis, Arabis, and related genera(Brassicaceae). Mol. Biol. Evol. 17: 1483–1498.

Lee, H. S., and Z. J. Chen, 2001 Protein-coding genes are epigenet-ically regulated in Arabidopsis polyploids. Proc. Natl. Acad. Sci.USA 98: 6753–6758.

Lee, H. S., J. Wang, L. Tian, H. Jiang, M. A. Black et al., 2004 Sensitivityof 70-mer oligonucleotides and cDNAs for microarray analysis ofgene expression in Arabidopsis and Brasssica. Plant Biotechnol. J. 2:45–57.

Lynch, M., and A. Force, 2000 The probability of duplicategene preservation by subfunctionalization. Genetics 154: 459–473.

Madlung, A., R. W. Masuelli, B. Watson, S. H. Reynolds, J. Davison

et al., 2002 Remodeling of DNA methylation and phenotypic andtranscriptional changes in synthetic Arabidopsis allotetraploids.Plant Physiol. 129: 733–746.

Masterson, J., 1994 Stomatal size in fossil plants: evidence for poly-ploidy in majority of angiosperms. Science 264: 421–424.

McClintock, B., 1984 The significance of responses of the genometo challenge. Science 226: 792–801.

Moser, E. B., A. M. Saxton and J. P. Geaghan, 1988 Biological ap-plications of the SAS system: an overview. Comput. Appl. Biosci.4: 233–238.

Osborn, T. C., J. C. Pires, J. A. Birchler, D. L. Auger, Z. J. Chenet al., 2003 Understanding mechanisms of novel gene expres-sion in polyploids. Trends Genet. 19: 141–147.

Ozkan, H., A. A. Levy and M. Feldman, 2001 Allopolyploidy-induced rapid genome evolution in the wheat (Aegilops– Triticum)group. Plant Cell 13: 1735–1747.

Pikaard, C. S., 1999 Nucleolar dominance and silencing of tran-scription. Trends Plant Sci. 4: 478–483.

Queitsch, C., T. A. Sangster and S. Lindquist, 2002 Hsp90 as acapacitor of phenotypic variation. Nature 417: 618–624.

Ramsey, J., and D. W. Schemske, 1998 Pathways, mechanisms, andrates of polyploid formation in flowering plants. Annu. Rev. Ecol.Syst. 29: 467–501.

Ranz, J. M., C. I. Castillo-Davis, C. D. Meiklejohn and D. L.Hartl, 2003 Sex-dependent gene expression and evolutionof the Drosophila transcriptome. Science 300: 1742–1745.

Schmid, M., N. H. Uhlenhaut, F. Godard, M. Demar, R. Bressanet al., 2003 Dissection of floral induction pathways using globalexpression analysis. Development 130: 6001–6012.

516 J. Wang et al.

Soltis, P. S., and D. E. Soltis, 2000 The role of genetic and geno-mic attributes in the success of polyploids. Proc. Natl. Acad. Sci.USA 97: 7051–7057.

Song, K., P. Lu, K. Tang and T. C. Osborn, 1995 Rapid genomechange in synthetic polyploids of Brassica and its implicationsfor polyploid evolution. Proc. Natl. Acad. Sci. USA 92: 7719–7723.

Stebbins, G. L., 1971 Chromosomal Evolution in Higher Plants. EdwardArnold, London.

Tian, L., M. P. Fong, J. J. Wang, N. E. Wei, H. Jiang et al., 2005 Re-versible histone acetylation and deacetylation mediate genome-wide, promoter-dependent and locus-specific changes in geneexpression during plant development. Genetics 169: 337–345.

Wang, J., L. Tian, A. Madlung, H. S. Lee, M. Chen et al., 2004 Sto-chastic and epigenetic changes of gene expression in Arabidopsispolyploids. Genetics 167: 1961–1973.

Wang, J., J. J. Lee, L. Tian, H. S. Lee, M. Chen et al., 2005 Methodsfor genome-wide analysis of gene expression changes in poly-ploids. Methods Enzymol. 395: 570–596.

Wendel, J. F., 2000 Genome evolution in polyploids. Plant Mol.Biol. 42: 225–249.

Wittkopp, P. J., B. K. Haerum and A. G. Clark, 2004 Evolutionarychanges in cis and trans gene regulation. Nature 430: 85–88.

Wolfe, K. H., 2001 Yesterday’s polyploidization and the mystery ofdiploidization. Nat. Rev. Genet. 2: 333–341.

Communicating editor: G. Gibson

Gene Regulation in Arabidopsis Allotetraploids 517