Silencing of a metaphase I-specific gene results in a phenotype similar to that of the Pairing homeologous 1 (Ph1) gene mutations

Significance

Maintaining diploid-like pairing behavior is essential for a polyploid to establish as a new species. The Pairing homeologous 1 (Ph1) gene, regulating such behavior in polyploid wheat, was identified in 1958, but its molecular function remained elusive. The present communication reports identification of the candidate Ph1 (C-Ph1) gene that is expressed exclusively during meiotic metaphase I, whose silencing resulted in formation of multivalents like the Ph1 gene mutations. Although the C-Ph1 gene has three homoeologous copies, the 5B copy has diverged in sequence from the other two copies. Heterologous gene silencing of the Arabidopsis homologue of the C-Ph1 gene also confirmed its function. Molecular characterization of this gene will make it possible to develop precise alien introgression strategies.

Abstract

Although studied extensively since 1958, the molecular mode of action of the Pairing homeologous 1 (Ph1) gene is still unknown. In polyploid wheat, the diploid-like chromosome pairing is principally controlled by the Ph1 gene via preventing homeologous chromosome pairing (HECP). Here, we report a candidate Ph1 gene (C-Ph1) present in the Ph1 locus, transient as well as stable silencing of which resulted in a phenotype characteristic of the Ph1 gene mutants, including HECP, multivalent formation, and disrupted chromosome alignment on the metaphase I (MI) plate. Despite a highly conserved DNA sequence, the C-Ph1 gene homeologues showed a dramatically different structure and expression pattern, with only the 5B copy showing MI-specific expression, further supporting our claim for the Ph1 gene. In agreement with the previous reports about the Ph1 gene, the predicted protein of the 5A copy of the C-Ph1 gene is truncated, and thus perhaps less effective. The 5D copy is expressed around the onset of meiosis; thus, it may function during the earlier stages of chromosome pairing. Along with alternate splicing, the predicted protein of the 5B copy is different from the protein of the other two copies because of an insertion. These structural and expression differences among the homeologues concurred with the previous observations about Ph1 gene function. Stable RNAi silencing of the wheat gene in Arabidopsis showed multivalents and centromere clustering during meiosis I.

The Pairing homeologous 1 (Ph1) gene was discovered in 1958 based on the observation that plants lacking wheat chromosome 5B exhibit homeologous pairing (1, 2). Lack of the gene results in multivalents during metaphase I (MI) of meiosis, resulting in partial sterility. Conversely, six doses of the gene in the triisosomic line of chromosome 5BL resulted in interlocking of the bivalents and reduced chiasmata frequency even among homologs, along with rare multivalents (3). Several other genes promoting or suppressing homeologous chromosome pairing (HECP) have also been reported (4, 5), although their effect is difficult to measure in the presence of the Ph1 gene (6). Ph1-like genes were also reported in other sexually propagating polyploids, including Avena sativa, Festuca arundinacea, Brassica napus, Gossypium hirsutum, and Gossypium barbadense, as well as in some diploids, including Lolium perenne, Lolium multiflorum, and Lolium rigidum (7–11).

Ph1 gene mutants in tetraploid (ph1c) (12, 13) and hexaploid (ph1b) (14) wheat were shown to be interstitial deletions involving an ∼0.84-μm region and an ∼1.05-μm region around the gene, respectively (15, 16) (SI Appendix, Fig. S1). Physical mapping localized the gene to an ∼2.5-Mb chromosomal region referred to as “Ph1 gene region,” bracketed by the distal breakpoint of ph1c deletion on the distal end and the breakpoint of deletion line 5BL-1 on the proximal end (16) (SI Appendix, Fig. S1). Various marker enrichment efforts identified nine markers for the region (17). Detailed microsynteny analyses and comparative mapping identified a 450-kb region of rice chromosome 9 (17). The corresponding rice region contained 91 genes. The major objective of the present study is to identify the gene(s) responsible for the Ph1 gene-like function using the available mapping information.

Results

Identification of the Candidate Gene.

The following criteria were used to select the potential Ph1 gene candidates from the 91 genes present in the 450-kb rice region: (i) genes expressed during meiosis and (ii) genes involved in chromatin reorganization, microtubule attachment, DNA binding, as well as acetyl- and methyltransferase activity. This detailed bioinformatics analysis identified 26 genes for further characterization. Virus-induced gene silencing (VIGS) was optimized for wheat meiosis using the disrupted meiotic cDNA1 (DMC1) gene, lack of which results in mostly univalents at MI. VIGS of TaDMC1 (Triticum aestivum DMC1) with an antisense construct resulted in an average of 37.2 univalents and 2.4 bivalents (18).

Except for about 4–6% of the MI cells that usually show aberrant chromosome pairing including multivalents, 21 bivalents were observed in the WT wheat cultivars (cv.) Chinese Spring (CS) and Bobwhite (BW) (SI Appendix, Tables S1 and S2). In the ph1b mutant, about 60% of the cells showed the aberrant chromosome pairing, with an average of 1.29 multivalents and 1.93 univalents per cell (SI Appendix, Table S1). The higher number of univalents observed in the ph1b mutant may be due to the combined effect of other genes present in the ∼1.05-μm chromosomal region deleted in the mutant line. At least two genes (LOC_Os09g30310 and LOC_ Os09g31310) have been identified in the deleted part of ph1b, silencing of which resulted in two to eight univalents. About 8% of the ph1b cells showed bivalent interlocking. The ph1b (SI Appendix, Fig. S2) and other Ph1 mutant and deletion lines show relatively normal chromosome alignment on the MI plate, and chromosome clumping is usually not observed.

VIGS screening of the 26 candidates (17) identified a gene that we designated as candidate Ph1 (C-Ph1; LOC_Os9g30320, wheat expressed sequence tag (EST) homolog BE498862) (SI Appendix, Fig. S2), silencing of which showed chromosome pairing behavior characteristic of the Ph1 mutant and deletion lines. Compared with almost all bivalents in 91% of the MI cells of the negative control [VIGS with pγ.MCS, carrying the sequence-matching multiple cloning site (MCS) of a plasmid], two of the five plants inoculated with the hairpin construct pγ.C-Ph1hp2 (Fig. 1 and SI Appendix, Table S3) showed multivalents and higher order pairing in 70.3% of the cells. In addition to multivalents, the MI chromosomes showed severe clustering and disrupted alignment on the MI plate. In comparison, only about 9% of the MI cells of MCS-inoculated plants showed misalignment and multivalents (Fig. 1).

Fig. 1.

Chromosomal pairing (CP) and C-Ph1 gene expression analysis in the VIGS and RNAi-silenced plants. (A) Chromosome spreads of PMCs at MI from FES and MCS controls and C-Ph1 silenced plants VIGS-5 and VIGS-7. CP analysis shows (i) the percentage of cells exhibiting aberrant pairing and the total number of cells analyzed (given in parenthesis) and (ii) the average number of multivalents per cell, with the range given in parentheses. EXP denotes the transcript expression levels (%) in the C-Ph1 silenced plants relative to the FES control, as observed by quantitative real-time PCR analysis. (B) Chromosome spreads of PMCs of control (BW) and C-Ph1 silenced RNAi plants indicating CP and EXP. (Scale bar, 5 μm.) (SI Appendix, Tables S2 and S3.)

Replicated VIGS experiments with the hairpin and antisense constructs showed a similar phenotype in the silenced plants. The construct pγ.C-Ph1hp1 showed the silencing phenotype in two of the 20 plants compared with seven of 20 plants for the pγ.C-Ph1as construct. The two plants with the pγ.C-Ph1hp2 construct showed multivalents and chromosome clustering in 72.7% (plant VIGS-7; Fig. 1) and 68% (plant VIGS-5; Fig. 1) of the cells, respectively, with an average number of only 6.93 and 9.08 bivalents, respectively. Only plant VIGS-7 showed bivalent interlocking in 5% of the cells, whereas no bivalent interlocking was observed in any of the control plants. Measured by quantitative real-time PCR analysis, expression of the gene in plant VIGS-7 was only 21.83% of expression of the gene in control plants compared with 54.56% in plant VIGS-5 (SI Appendix, Fig. S3). The expression of the gene in the remaining three plants that did not show any aberration in chromosome pairing ranged from 89–92% of the control plants (SI Appendix, Fig. S3).

Stable RNAi silencing of the C-Ph1 gene was accomplished by transforming wheat cultivar BW with the hairpin RNAi construct pHellsgate8 1-1 involving 200 bp of the gene segment. Seven of the 54 confirmed T₀ (plants after regeneration) transgenic plants (Materials and Methods) were randomly selected for meiotic chromosome pairing analysis. Compared with the negative control (BW), reduction in the gene expression among these seven plants ranged from 7–83% (SI Appendix, Fig. S4). Transcriptional suppression between 22% and 44% appears to be necessary to induce HECP. Two of the seven plants (RNAi-7 and RNAi-3) that exhibited 7% and 22% transcript suppression, respectively, showed relatively normal chromosome pairing, with only 6.2% and 14.2% of the MI cells, respectively, showing aberrant chromosome pairing. RNAi-3 showed slight, although nonsignificant, multivalents, along with misalignment. Multivalents were not observed in RNAi-7, although slight misalignment was observed similar to that of the negative control, which showed misalignment in only 6% of the cells (Fig. 2). These two plants were fully fertile.

Fig. 2.

Cytogenetic analysis showing different levels of C-Ph1 gene silencing in RNAi plants compared with ph1b. Chromosome spreads of PMCs from ph1b, BW (negative control), and four RNAi plants showing different levels of gene silencing. Expression denotes the normalized transcript expression levels (%) relative to BW, using the delta-delta threshold cycle (Ct) method, observed by quantitative real-time PCR analysis. Aberrant pairing (%) denotes the percentage of cells exhibiting aberrant chromosome pairing. “Multivalents/cell” denotes the average number of multivalents per cell, and the range is given in parentheses. Chromosome (Chr.) clustering and misalignment phenotype are represented by “+” and “−,” where (+) indicates increased severity levels and (−) indicates decreased severity levels. (Scale bar, 5 μm.)

RNAi-5 with a 44% reduction in gene expression was the most interesting finding because it showed a chromosome pairing phenotype similar to that seen in the ph1b mutant (Fig. 2). Multivalents were observed in 22% of the cells (Fig. 2). The number of bivalents in this plant was 12, compared with 16 in the ph1b mutant (SI Appendix, Tables S2 and S4). As is the case for the ph1 mutants, RNAi-5 was partially fertile. Misalignment and chromosome clustering were not commonly seen in this plant. The reduction in gene expression in the remaining four plants ranged from 50–83% (SI Appendix, Tables S2 and S3). The number of bivalents in these four plants ranged from 10.6 to 14.3, compared with 19.7 in the negative control. The number of MI cells showing multivalents, chromosome clustering, and misalignment ranged from 71.7–90% (Fig. 1 and SI Appendix, Table S3). The plant (RNAi-6) showing a 50% reduction in gene expression exhibited a more severe phenotype than RNAi-5, with additional chromosome clumping and disrupted alignment on the MI plate. This plant exhibited aberrant chromosome pairing in 71.7% of the cells (SI Appendix, Table S3). There was a significant increase in the levels of chromosome clumping and misalignment along the MI plate with a further 31.64% reduction in gene expression. RNAi-4 showed the maximum level of silencing (Figs. 1 and 2). Bivalent interlocking, which was not observed in the negative control, was present in 31.2% of the MI cells of these four transgenic plants. Overall, chromosome pairing aberration was very severe in these four plants compared with that seen in the ph1b mutant (Figs. 1 and 2). These four plants were completely sterile but exhibited no other phenotypic abnormality.

Structure of the C-Ph1 Gene.

Detailed bioinformatics and sequence analyses revealed three genomic and cDNA copies of the gene, one on each of the three wheat group 5 chromosomes (Fig. 3). At the DNA level, the three homeologues from cv. CS were 90% similar and the differences were due to several structural changes, including deletions and insertions (Fig. 3). The 5B copy of the gene sequence showed two insertions of 46 bp and 14 bp present 5 bp apart in exon II (referred to as a 60-bp insertion) (Fig. 3). A deletion of 29 bp was observed 80 bp upstream of the two insertions. There were two 5D-specific deletions of 12 bp and 15 bp present in exon II. Additional differences among the homeologues were a 13-bp deletion in the 5D copy present 11 bp downstream of the exon–intron junction and two insertions in the 5A copy: a 7-bp insertion in the intron and a 6-bp insertion in exon II (Fig. 3). The 5′ UTR of the 5B copy was 96.8% similar to that of the 5D copy and 94.3% similar to that of the 5A copy. A similar comparison between the 5A and 5D copies showed 92.6% similarity. The 3′ UTRs of the 5B and 5D copies were 92.6% similar, and the differences were mainly due to an 11-bp insertion in the 5B copy, along with six homeologous sequence variants (HSVs). The 3′ UTR of the 5A copy did not match with either the 5B or the 5D copy mainly due to a major deletion/insertion. Of the HSVs among the three gene copies, 26.7% were synonymous and 73.3% were nonsynonymous. Overall, the percentage of nonsynonymous bases in the gene was 77.8%, with the remaining 22.3% being synonymous.

Fig. 3.

Structural differences among the C-Ph1 gene homeologues in hexaploid wheat. Nucleotide sequences of cloned CS C-Ph1-5B, its splice variant (C-Ph1-5B^alt), and C-Ph1-5D and C-Ph1-5A copies were aligned to each other; the differences are drawn to scale (1 = 1 nucleotide). The symbols ▲ and ▼ represent deletions and insertions in the sequences, respectively. Insertions and deletions were determined by majority consensus rule. The shaded region in C-Ph1-5B and C-Ph1-5B^alt represents a corresponding region similar to the C-Ph1-5D and C-Ph1-5A sequences; the nucleotide sequence is not translated as a protein (predicted) but forms a part of the UTR. The colored bars below C-Ph1-5B represent VIGS and RNAi oligos, denoted by as (antisense), hp1 (hairpin 1), hp2 (hairpin 2), and RNAi, respectively. The gray dots at the end of C-Ph1-5A exon II represent deletion/insertion not present in the C-Ph1-5B and C-Ph1-5D sequences.

The genomic copy of 5B was the largest (954 bp) compared with the 5D (883 bp) and 5A (539 bp) copies. Excluding insertions and deletions, the 5B genomic copy is 95% similar to that of the 5D copy. The similar proportions are 94.4% for the comparison between 5B and 5A and 85.9% for the comparison between 5A and 5D. Overall, the genomic copies of the three homeologues shared 41% DNA sequence similarity.

Among the three gene copies, 5A produced the smallest transcript (excluding 5′ and 3′ UTRs) of 420 bp by splicing an intron of 120 bp, whereas 5D produced a transcript of 783 bp by splicing an intron of 100 bp (Fig. 3). The 5B copy of the gene showed signs of alternate splicing to produce transcripts of 954 bp and 763 bp. The difference between the two variants was two introns of 113 bp and 78 bp, which were spliced to generate the 763-bp version and were retained to generate what turned out to be the largest transcript among all gene copies (Fig. 3). The 5B copy with the larger transcript generated a smaller protein than its alternate form due to retention of the intron around the 60-bp addition, which contained an in-frame stop codon (Fig. 3). Predicted proteins from either of the two splice variants from the 5B copy were smaller than proteins produced by the 5D copy but larger than proteins produced by the 5A copy.

Predicted proteins from the 5B copy of the gene were 204 aa and 221 aa in length compared with 174 aa from the 5A copy and 260 aa from the 5D copy. The most conserved part of the gene corresponding to exon I was not present in either of the predicted proteins from the 5B copy but was present in the predicted proteins from the 5A and 5D copies. The two proteins representing the 5B copy of the gene were 91% similar, with the only difference being an additional 17 aa on the N terminus in the larger version. Not counting the 86-aa deletion in 5A created by a premature stop codon, the predicted proteins of the three copies were 82% similar. Considering all deletions and insertions however, the 5A copy protein is only 25–30% and 46.4% similar to the 5B copy and 5D copy proteins, respectively. Similarly, the 5D copy and the 5B copy proteins were 68–74% similar. Likewise, the predicted 3D structures of the four proteins were significantly different, indicating functional divergence of the homoeologues after allopolyploidization (SI Appendix, Fig. S5).

C-Ph1 Expression Pattern.

Quantitative real-time expression analysis representing the cumulative expression of all homeologues showed that the gene is primarily expressed during the postflower initiation stages, although significant expression was observed in the roots as well (Fig. 4). Maximum expression of the gene was during meiosis I; however, the flag leaf of the plants undergoing meiosis also showed significant expression. Essentially no expression was observed in mature pollen grains or the subsequent seed development stages.

Fig. 4.

C-Ph1 gene expression pattern in various tissues and substaged meiotic anthers. (A) Chromosome spreads of PMCs from CS denoting various stages of meiosis. (Scale bar, 5 μm.) (B) Quantitative expression analysis using gene-specific primers (SI Appendix, SI Text) in the root (R), leaf (L), flag leaf (FL), 3- to 5-cm spike (I to MI), and 6- to 8-cm spike (MII to T), as well as at anthesis (AN) and 5 d postanthesis (5DPA). (C) Quantitative expression analysis at interphase (I), prophase I (P), late prophase I and metaphase I (M), anaphase I (A), dyad (D), and tetrad (T). The y axis in B and C denotes the normalized mRNA levels using the delta-delta Ct method. (D) Tissue- and stage-specific expression of homeologues analyzed by single-strand conformation polymorphism (SSCP) analysis (SI Appendix, SI Text). (E) Meiotic stage-specific expression of homeologues in the different substages of meiosis, as mentioned in C.

To “pinpoint” the gene expression pattern during various meiotic stages, one of the three anthers from each floret was used for the meiotic chromosome analysis and the remaining two were used for real-time gene expression analyses (primer details are provided in SI Appendix, SI Text and Table S4). All three anthers from a single wheat floret are known to be at a developmentally identical stage (19). Chromosome spreads of pollen mother cells (PMCs) from wheat cv. CS denoting the various stages of meiosis are shown in Fig. 4A. Expression of the gene increased 13-fold in the transition from prophase I to late prophase I, followed by a further increase of about 26-fold during MI (Fig. 4C). Relative to MI, expression dropped by 34-fold during anaphase I, followed by a further drop of 6.4-fold during the dyad stage (Fig. 4C). Maximum expression of the gene was observed during MI. Surprisingly, there was a 16.5-fold increase in gene expression during the tetrad stage, suggesting additional functions of the gene.

Expression analysis of each of the gene copies individually by single-strand conformation polymorphism analysis revealed that the three copies of the gene have dramatically different expression patterns. With the exception of roots, where almost all copies showed expression, the 5B copy specifically expressed in a 3- to 5-cm long spike, which, in CS, contains meiotically dividing cells (Fig. 4D). Expression of the 5B copy dropped significantly in the 6- to 8-cm spike that contains cells at the meiosis II stage. Essentially no expression was observed for the copy at the mature anther stage (Fig. 4D). The expression pattern of the 5D copy was very different from that of the 5B copy. Unlike 5B, the 5D copy showed a low level of expression in the leaves and the 3- to 5-cm spike as well as during anthesis or 5 d postanthesis (Fig. 4D). The 5A copy was expressed predominantly during meiosis II and showed very little expression in the 3- to 5-cm spike, suggesting its role in cytokinesis and/or gametophyte development.

In the substaged meiotic anthers, the 5B copy was specifically expressed during MI and a low level of expression was also seen during anaphase I. No expression was observed for the copy in the subsequent meiotic stages (Fig. 4E). Unlike 5B, the 5A copy was expressed during the anaphase I, dyad, and tetrad stages. Compared with the other two copies, the 5D copy showed significantly higher expression during interphase. A significant amount of expression of the 5D copy was also observed during prophase I and MI (Fig. 4E).

Additionally, expression of the 5B copy was analyzed in the wheat homeologous group 5 nullisomic-tetrasomic (NT) lines, ph1 mutants, and a series of 5BL deletion lines during meiosis (SI Appendix, Fig. S6). The results provide an additional line of evidence to confirm that the newly identified C-Ph1 gene in this study corresponds to the Ph1 locus.

Gene Orthologs in Other Plants.

Structurally conserved orthologs of the C-Ph1 gene were observed in all studied monocots, including rice, barley, maize, and Brachypodium. The rice ortholog (LOC_Os09g30320) maps on chromosome 9, the Brachypodium ortholog (Bradi4g33300) maps on chromosome 4, and the maize ortholog (GRMZM2G078779) maps on chromosome 7. The gene showed a typical pattern of three exons and two introns in all these species except maize, which had two additional splice variants. In addition to the conserved BURP domain (named based on four typical members, BNM2, USP, RD22, and PG1-β), a 45-bp first exon was observed, followed by a 94- to 138-bp intron and a 114- to 138-bp second exon. The size of second intron in rice was 599 bp, which is comparatively larger than 97 bp, 119 bp, and 122 bp in maize, Brachypodium, and barley, respectively. The third exon was between 596 and 644 bp in size and contained BURP domain-related sequences. The transcribed part of the barley copy showed 90% similarity with the 5B copy of the C-Ph1 gene, whereas the other species had 71–72% similarity. Structurally, the putative C-Ph1 ortholog from several diploids, including barley, maize, Brachypodium, and rice, resembled the 5D copy more closely, suggesting it to be the conserved and ancestral version of the gene.

Studied using the available microarray-based expression data, the maize ortholog of the C-Ph1 gene showed the highest expression in the meiotic tassel and anthers containing the PMC meiotic stages (20). The developing endosperm and kernel also showed a significant level of expression, but no other developmental stage showed any expression of the gene. Similarly, the rice ortholog of the gene, LOC_Os09g30320, also showed the highest expression in the heading panicle and stamens containing meiotic tissues, and no expression was observed in any other developmental stages (21). The barley ortholog was identified from ESTs derived from immature male inflorescences (22), although a detailed expression pattern of the gene in barley is not yet known.

Initially, DNA sequence analysis and a domain/motif search identified At5g25610 as the putative Arabidopsis ortholog of Os09g30320 (17). At5g25610 encoded a BURP domain containing protein with a putative function in the dehydration stress response, because it encodes dehydration-induced protein RD22 (17), thus making it a less likely candidate in Arabidopsis. Sequence comparison at the DNA or protein level identified another ortholog of the wheat gene in Arabidopsis, although with poor conservation. At1g78100 mRNA matched perfectly with the 22 bp of the wheat RNAi fragment. The gene expresses during anthesis and various other developmental stages, thus making it a likely candidate for the gene ortholog. Poor sequence conservation among orthologs has been well documented for many other meiotic genes.

Gene Function in Arabidopsis.

With the assumption of functional conservation of the gene between Arabidopsis and wheat due to conserved catalytic motifs, we performed RNAi-based silencing of the Arabidopsis ortholog. The resistant transgenic lines, along with WT Columbia (Col-8), were analyzed for meiotic chromosome pairing analysis, and the results are shown in Fig. 5 and summarized in SI Appendix, Tables S5 and S6.

Fig. 5.

Multivalent formation in the RNAi-silenced Arabidopsis plants. Each image is a flat projection across the entire nucleus. Chromosomes were counterstained with DAPI (red); the centromeric probe was labeled with cyanine-5 (green). (A–C) Normal meiosis progression from leptotene to late pachytene, leading to formation of five bivalents in the WT. Centromere coupling during leptotene (D) and multivalent formation in zygotene (E) (Inset shows quadrivalent formation) in the gene silenced plants. Centromere coupling in pachytene, leading to formation of two clusters of centromeres (F) instead of five pairs in the WT (C). (Scale bar, 5 μm.) (SI Appendix, Tables S5 and S6.)

Twenty-five cells were imaged and analyzed for chromosome pairing defects during early stages of leptotene to pachytene. In all of the cells analyzed, the WT Arabidopsis showed five bivalents (Fig. 5C) in contrast to an average of 3.05 bivalents in the silenced plants. On average, 0.9 quadrivalents and 0.05 hexavalents per meiotic cell were observed in the RNAi plants, whereas no quadrivalent or hexavalent was observed in the WT plants (SI Appendix, Table S5). The RNAi plants showed multiple associations in the form of centromere coupling in all of the analyzed cells at the leptotene stage (Fig. 5 and SI Appendix, Table S6), whereas no such centromere coupling was observed in the WT. The centromere coupling led to the formation of multivalents during zygotene and pachytene stages. The silenced plants showed multivalent formation in 90–95% of the analyzed cells in the zygotene and pachytene stages (Fig. 5 and SI Appendix, Table S6). The WT was normal and showed no such chromosomal aberrations.

Cell Division Cycle 2 Gene Is Not a Good Candidate for the Ph1 Gene.

The Cell division cycle 2 (Cdc2-4) gene is present in chromosome deletion lines 5BL-1, 5BL-3, and 5BL-8 (SI Appendix, Fig. S7), which are known to lack the Ph1 gene, because their chromosome pairing matched with that of the ph1b and other Ph1 gene mutants (16, 23). VIGS analysis with a 96-bp antisense construct targeting the Cdc2-4 gene showed normal chromosome pairing at meiotic MI (SI Appendix, Fig. S8 and Table S7). The VIGS plants showed an average of 20.9 bivalents compared with 21 bivalents in the MCS and FES (abrasive agent used for inoculation) control plants (SI Appendix, SI Text and Fig. S8).

Discussion

Since its discovery in 1958, various studies have implicated the Ph1 gene in many different meiotic processes. While studying the somatic association of chromosomes in premeiotic cells, the Ph1 gene was suggested to be involved in ensuring strict homologous pairing by suppressing premeiotic homeologous chromosome association (24). Careful analyses of the published data specifically implicates the 5D copy of the Ph1 gene in the initial chromosome pairing of both homologs and homeologues because asynapsis was observed in the absence of the 5D copy but not the 5A or the 5B copy (3, 25). The absence of chromosome 5A had essentially no effect on HECP. Although the 5B copy of the Ph1 gene was shown specifically to regulate diploid-like pairing (2, 26), various lines lacking either chromosome 5B, its long-arm, or the Ph1 locus all showed increased HECP (14, 16, 27). The function of the 5B copy in differentiating homologs from homeologues was further supported by the fact that four copies of either 5A or 5D were not able to restore normal chromosome pairing in the absence of 5B (3, 25). Multivalents and other types of higher order pairing observed in the absence of the 5B copy were not observed in the absence of the 5A copy and were not as robust in the absence of the 5D copy (3, 25). Along with asynapsis, lack of 5D exhibited frequent bivalent interlocking and rare multivalents. A minimum of four copies of the 5A copy were needed to compensate for the absence of the 5D copy, suggesting that the two copies share a common function, with the 5A copy having a weaker effect.

The C-Ph1 gene that we have identified in the present study explains the observations made on the Ph1 gene function. The expression and silencing data clearly suggest that the C-Ph1 gene has multiple functions during meiosis, each controlled by one or more copies of the gene. One of these functions is the initial pairing of both homologs and homeologues, as suggested by the higher expression level of the 5D copy during interphase and the gene silencing phenotype. Chromosome 5B was implicated in the specific function to differentiate homologous pairing from HECP as shown by the unique expression pattern of the 5B copy of the C-Ph1 gene, along with the gene silencing phenotype. The 39.7-fold increase in expression of the 5B copy between late prophase and MI coincided with the stages when this precise function takes place. Expression of the 5D copy during the MI stage suggests an additive function of the copy during MI. The 5A copy was expressed predominantly during meiosis II, suggesting its role in cytokinesis and/or gametophyte development.

In accordance with the interpretation that the 5A and 5D copies of the Ph1 gene share a common function, the predicted proteins of the 5A and 5D copies of the C-Ph1 gene are very similar, except that the 5A copy produces a truncated, and thus perhaps less effective, protein. The two proteins share a highly conserved motif corresponding to exon I, which is almost identical in the two homeologous gene copies but is absent in the 5B copy proteins. The presence of this highly conserved motif suggests unique function(s) for the two copies, including initial pairing of both homologs and homeologues. Alternatively, the unique function of the 5B copy may be due to the lack of this conserved motif along with an insertion of 60 bp that contains an in-frame stop codon, thus resulting in smaller proteins. The unique function(s) of the 5B protein(s) may also be due to its very specific expression pattern (mentioned above). The presence of the two 5B copy proteins resulting from alternate splicing suggests multiple functions of the 5B copy. The differences in structure and expression patterns among the three copies of the gene suggest neofunctionalization of the 5B copy, with at least one of the functions being different from that of the 5A or 5D copy. Sequence similarity with diploid species suggests the 5D copy to be the ancestral copy.

VIGS and RNAi-silenced plants showed all of the chromosome pairing aberrations observed in the Ph1 gene mutations along with some additional phenotypes, including chromosome clumping and disrupted alignment on the MI plate. In both the VIGS and RNAi plants, the severity of the chromosome pairing phenotype correlated well with the level of gene silencing. One of the RNAi plants (RNAi-5) with a 44% reduction in gene expression showed a chromosome pairing phenotype similar to that seen in the ph1 mutants and the lines lacking the Ph1 gene (Fig. 2). Characteristic of the ph1 mutants, multivalents were observed in this plant without any disruption in chromosome alignment on the MI plate or chromosome clumping (Fig. 2). These results suggest that about a 44% reduction in expression of the gene is needed to show the aberrant chromosome-pairing phenotype, similar to that observed in the Ph1 gene mutants. Previously, bivalent interlocking was observed in ph1b mutant and in the plants lacking the 5B or 5D copy, as well as in plants triisosomic for chromosome 5BL (3). We also observed bivalent interlocking in the C-Ph1–silenced plants. The bivalent interlocking is probably caused by pairing between distant homologs in the absence of the gene. Bivalent interlocking in the plants carrying a triple dose of chromosome 5BL is probably due to the dosage effect of Ph1 on the relative separation of homologs before meiosis (28). Bivalent interlocking could also be due to silencing of the gene triggered by the higher copy number (29).

All RNAi and VIGS plants with gene silencing of more than 44% resulted in chromosome clustering and misalignment of chromosomes on the MI plate, in addition to the expected multivalent formation. This phenotype was not observed in any of the NT lines, probably because multiple copies of the gene perform the same function and loss of a copy in NT lines is compensated for by the other copies.

Our data suggest a plausible explanation of the above-mentioned observations. Firstly, expression of the 5B copy increases between late prophase and MI coinciding with the stages when centromere–microtubule interactions takes place. Secondly, transient VIGS as well as stable RNAi silencing of the C-Ph1 gene resulted in severe centromere clustering, along with disrupted alignment of chromosomes on the MI plate, suggesting a plausible role in centromere–mictrotubule interaction. This dramatic clustering and misalignment was not observed in the absence of any one of the three gene copies during previous studies, suggesting that two or more of the gene copies act in an additive manner to accomplish this very important function. Also, expression pattern of the 5B copy of the C-Ph1 gene closely coincides with that of the motor protein CENP-E (30), a kinetochore-associated protein involved in the sustained movement of chromosomes leading to proper alignment on the MI plate (31). Taken together, these and the observations of disrupted chromosome alignment on the MI plate in the VIGS and the RNAi plants suggest that either the C-Ph1 gene functions by regulating centromere–microtubule interaction as was previously suggested, or that this is one of the additional functions of the gene where two or more copies of the gene have the same function, in addition to regulating HECP.

Studied in the root-tip cells, chromosomes in ph1b mutant showed higher mitotic association of homeologues and hypersensitivity to colchicine compared to those in the normal CS, and disrupted centromere-microtubule association was suggested to be the cause (32–34). A low level of expression of the 5B copy of the C-Ph1 gene was observed in all mitotically dividing tissue, including roots and leaves, suggesting its role in mitotic cell division. This also supports the previous reports that Ph1 gene functions during mitotic and particularly premeiotic stages, affecting chromosomal movement towards the poles and, consequently, their arrangement in the nucleus. This may effect premeiotic association of homologous chromosomes and relative separation of homeologues, thus determining exclusive homologous pairing in wheat already before the commencement of synapsis (35–37).

Materials and Methods

Plant Material.

Plant material used in this study included WT hexaploid wheat (Triticum aestivum cv. CS and cv. BW), a CS mutant lacking the Ph1 locus (ph1b), wheat homeologous group 5 NT lines, and a series of 5BL deletion lines. Based on the efficient utilization of the cv. CS for chromosome squash preparations, it was selected as an ideal cultivar for VIGS. CS, NT, and deletion lines were used for mapping and cloning experiments. BW was used for RNAi experiments because it can be efficiently transformed using Agrobacterium-mediated gene transfer (38). The detailed growth conditions are given in SI Appendix, SI Text.

VIGS.

The preparation of vector constructs, transcription, and inoculation of viral RNAs has been described previously (18). On the basis of comparative sequence analysis, the unique gene region for the C-Ph1 homeologue on chromosome 5B and the Cdc2-4 gene was selected for silencing. The procedure is elaborated in SI Appendix, SI Text. FES buffer (abrasive agent used for inoculation) was used as a negative control, and the plasmid pγ.MCS [containing a 121-bp antisense fragment of the MCS (pBluescript K/S; Stratagene)] was used as a “virus-only” control to differentiate the effect of the target gene from that of the virus. For the experiment using an antisense construct, 10 plants were inoculated with pγ.MCS and four plants were inoculated with FES. Four CS plants were also used as a control. Similarly, for VIGS using a pγ.C-Ph1hp2 construct, one and three plants were inoculated with FES and pγ.MCS, respectively. Likewise, for the Cdc2-4 gene, five plants each were inoculated with pγ.MCS and FES.

To target the gene in PMCs, the flag leaf of the main tiller was inoculated at the boot stage by rubbing. Inoculated plants were lightly misted with water and covered with plastic bags for 16–18 h.

RNAi Genetic Transformation.

For RNAi-based silencing of the Arabidopsis ortholog, a 200-bp wheat RNAi construct was cloned in the pANDA35HK vector, driven by the 35S promoter and carrying a gene for hygromycin resistance. Details of this procedure are provided in SI Appendix, SI Text.

The RNAi construct for the stable wheat transformation was developed by amplifying a 200-bp target sequence from the C-Ph1 gene. Details of this procedure are provided in SI Appendix, SI Text.