Illustration of auto- and allopolyploids. For simplicity, only one pair of homologous chromosomes (either in red or blue) is shown in a diploid. (a) Formation of an autotetraploid by doubling a basic set of chromosomes. A triploid (not shown) can be formed by hybridization between a diploid and an autotetraploid. (b) Formation of an allotetraploid by interspecific hybridization followed by chromosome doubling. (c) Formation of an allopolyploid by fusion of unreduced gametes in two diploid species. (d) Formation of an allotetraploid by interspecific hybridization between two autotetraploid species. (e) An allotetraploid (left) may become a segmental allotetraploid (right) if homoeologous chromosomes contain some homologous chromosomal segments.

Hypotheses to explain changes in genome structure and function in polyploids. (a) Genetic and epigenetic changes in polyploid genomes. Genetic changes include sequence mutations and chromosomal instabilities (deletion/insertion, translocation, transposition, etc.). Epigenetic changes may occur at transcriptional and post-transcriptional levels. Transcriptional regulation is associated with the formation of heterochromatin (black segments in red chromosomes) and euchromatin (irregularly-shaped long arms of blue chromosomes), leading to gene silencing or activation. Chromatin modifications [through multiple mechanisms including RNA interference (RNAi)] may transcriptionally silence or activate mobile elements, protein-coding genes, and rRNA genes that are responsible for novel variation in allopolyploids. Post-transcriptional regulation involves RNAi, RNA processing, and stability. AA: diploid genome; AAAA or BBBB: autotetraploid; AABB: allotetraploid. Blue segments in red chromosomes or red segments in blue chromosomes in an allotetraploid indicate nonreciprocal exchanges between homoeologous chromosomes. White segments in the red chromosomes indicate a deletion. (b) A dosage-dependent () or “rheostat” () model suggests additive effects of gene expression in a diploid (upper panel) and tetraploid (lower panel). The duplicate loci in the tetraploid provide additional levels of controls for gene expression. The levels of gene expression are shown as “off” (0) and by gray arrows (1, 2, and 3), and the maximum level (2 in diploid and 4 in tetraploid) is shown by green arrows. This model may explain dominant effects of hybrid vigor but does not explain overdominant performance or novel variation in allopolyploids. (c) A regulatory compatibility model suggests that regulatory factors (proteins) produced from orthologous genes generate incompatible heterologous products (heterodimers between red squares and blue circles). Alternatively, heterologous proteins may perform better than homologs (homodimers between red squares or blue circles), which may explain overdominant effects and hybrid vigor. (d) Hypotheses for testing additive and nonadditive gene regulation in Arabidopsis allotetraploids. The null hypothesis (Ho) is that gene expression levels in an allotetraploid (Allo) are equal to the sum of two progenitors, A. thaliana autotetraploid (At4) and A. arenosa (Aa). Typical seedling leaves in At4, Aa, and Allo are shown. Ha: alternative hypothesis.

Production of Arabidopsis allotetraploids. (a) Schematic karyotypes of F₁ nascent allotetraploids and their progenitors. Seedlings of the two progenitors, A. thaliana autotetraploid (At4) and A. arenosa tetraploid (Aa), are shown. (b) Phenotypic variation was infrequently observed in the F₁ allotetraploids but was very common in the segregating populations (F₂) (A–D). A and D: A. arenosa-like, B: A. thaliana-like, and C: intermediate. (c) Independent lineages of allotetraploids are produced by selfing multiple F₁ lines. Allopolyploids self-pollinate spontaneously (or can be manually self-pollinated), and progressive inbreeding occurs in advanced generations. Karyotypes in a meiotic cell of a resynthesized allotetraploid (Allo733 in the fifth generation) show 5 pairs of A. thaliana centromeres (At Cen, red) and 8 pairs of A. arenosa centromeres (Aa Cen, green) () (reproduced with kind copyright permission of the Genetics Society of America). Seedling of Allo733 is shown. All plants were photographed when they were 4–5 weeks old. The size bars indicate 3 cm.

(a) Phenotypic variation of the plants after five generations of selfing. The plants include two parents, A. thaliana autotetraploid (At4) and A. arenosa (Aa), six allotetraploids (S–1 to S–6), and a natural allotetraploid, A. suecica (As). Allotetraploids have the same chromosome number ( and data not shown) but display phenotypic and flowering-time variation. Some plants (S5–5 and S5–6) resemble A. suecica, whereas others (S5–1 to S5–4) show novel phenotypes. Differential expression of parental genes may contribute to the phenotypic variation of Arabidopsis allotetraploids. (b) Evidence for epigenetic silencing in resynthesized allopolyploids. A. suecica (As, white flower) is the naturally occurring allotetraploid derived from A. thaliana (not shown) and A. arenosa (Aa), which have white and pink flowers, respectively. The flower colors in the third generation of resynthesized allotetraploids (S3–1, 2, and 3) segregate from all white (S3–1) to all pink (S3–3). Variegated flower colors in the same inflorescence branch (S3–2) indicate epigenetic regulation of genes controlling flower color pathways. The same-size bars are 50 mm in (a), except for At4, and 5 mm in (b), except for As.

Models for transcriptional and post-transcriptional regulation of nonadditive gene expression in allopolyploids. (a) Cis- and trans-acting effects on transcriptional regulation of nonadditive gene expression in allotetraploids. I. During evolution, changes in cis-regulatory elements may confer “strong/dominant” and “weak/recessive” loci. II. Combination of these loci in the same allotetraploid cells induces cis- and trans-acting effects. Because A. thaliana FRI (AtFRI) is nonfunctional, A. arenosa FRI (AaFRI) trans-activates A. thaliana FLC (AtFLC) and cis-regulates one of the A. arenosa FLC (AaFLC1) loci, leading to overdominance or very late flowering in resynthesized allotetraploids. The cis-regulation and trans-activation are maintained by histone acetylation (Ac in a small purple circle) and methylation (Me in a small orange circle). Functional AaFRI is in a large red circle, and nonfunctional A. thaliana FRI (atflc) is in a white small circle with a slash line. The straight and inclined red arrows indicate cis- and trans-acting effects, respectively, by AaFRI. FLC loci are in oval circles. Blue and green indicates active AtFLC1 and AaFLC1, respectively, and gray indicates silenced AaFLC2. The differences in the expression of three FLC loci may reflect cis-regulatory variation (). III. Cis- and trans-acting interactions and epigenetic regulation of FRI and FLC loci may facilitate selection for flowering-time variation in resynthesized allotetraploids (indicated by a circular wheel). IV. In A. suecica, additional selection forces may enhance the expression of AaFLC1 (color change from dark green in II to yellow green in IV) and related genes such as MAFs (), which facilitate adaptation to cold climate. This simple model may be generalized to explain a mechanism for regulatory interactions between divergent loci in other pathways in the allotetraploids (see text for details). (b) Nonadditive accumulation of mRNAs and small RNAs in allotetraploids (null hypothesis H_o: 2 + 1 = 3). There is evidence that 15–40% of the genes are expressed differently between two closely related species of Arabidopsis (), and a subset of those genes are subjected to nonadditive expression in the allotetraploids. Up- or downregulation of mRNAs may be caused by activation of genes originating in parent A, B, or both. We exclude the situation in which the expression of one locus is compensated by its corresponding homoeologous locus. If the nonadditively accumulated transcripts are small RNAs, such as microRNAs (miRNAs), heterochromatic and transposon-related siRNAs (ht-siRNAs), trans-acting siRNAs (ta-siRNAs), and natural anti-sense (nat-siRNAs), they typically negatively regulate the expression of their target genes. The red and blue wave lines indicate RNA transcripts or small RNAs produced in species A and B, respectively. We hypothetically assign two RNA molecules for species A and one RNA molecule for species B. Note that the number of RNAs in the nascent allotetraploids (F₁) is not equal to 3 (2 + 1), suggesting nonadditive regulation (dominance and overdominance if larger than 3 or repression if smaller than 3). Possible changes (dashed arrows) in RNA composition and accumulation (column X) may occur in self-pollinating progeny, which are indicated by orange and green wave lines, respectively. According to the hypothetical outcome, the origins of small RNA upregulation (red in 1–3) and repression (blue in 4–6) are shown in the last column.

Switching of self-incompatibility in resynthesized Arabidopsis allotetraploids. (a) Flower morphology of self-compatible A. suecica. (b) Self-incompatible A. arenosa and (c and d) two self-compatible synthetic allotetraploid lines. The altered flower organs (elongated stigma and relatively short stamens) in the synthetic allotetraploids (c and d) may contribute to a low fertility without manual pollination. Size bars (5 mm) are the same in (a) and (b) or in (c) and (d). (e) Diagram of cytoplasmic and nuclear interactions in resynthesized allotetraploids (Allos, 2n = 4x = 26). T or A and tt or aa denote cytoplasm and nuclear genomes, respectively. The cytoplasmic-nuclear genotypes for autotetraploid A. thaliana (At4) are Ttttt, tetraploid A. arenosa (Aa), Aaaaa; and allotetraploids, and either Taatt or Aaatt in reciprocal crosses. Blue and gray indicate A. thaliana and A. arenosa cytoplasms, whereas green, red, and yellow indicate nuclear genomes of A. thaliana, A. arenosa, and allotetraploids, respectively. A. thaliana is self-compatible (SC), whereas A. arenosa is self-incompatible (SI). The resynthesized allotetraploids (Allos) are SC. Fertile seeds are readily obtained from the cross using A. thaliana as a maternal parent and A. arenosa as a pollen donor, and the reciprocal cross is usually unsuccessful ().