Emergence of a Novel Chimeric Gene Underlying Grain Number in Rice

Abstract

Grain number is an important factor in determining grain production of rice (Oryza sativa L.). The molecular genetic basis for grain number is complex. Discovering new genes involved in regulating rice grain number increases our knowledge regarding its molecular mechanisms and aids breeding programs. Here, we identified GRAINS NUMBER 2 (GN2), a novel gene that is responsible for rice grain number, from “Yuanjiang” common wild rice (O. rufipogon Griff.). Transgenic plants overexpressing GN2 showed less grain number, reduced plant height, and later heading date than control plants. Interestingly, GN2 arose through the insertion of a 1094-bp sequence from LOC_Os02g45150 into the third exon of LOC_Os02g56630, and the inserted sequence recruited its nearby sequence to generate the chimeric GN2. The gene structure and expression pattern of GN2 were distinct from those of LOC_Os02g45150 and LOC_Os02g56630. Sequence analysis showed that GN2 may be generated in the natural population of Yuanjiang common wild rice. In this study, we identified a novel functional chimeric gene and also provided information regarding the molecular mechanisms regulating rice grain number.

AS a widely consumed staple food in the world, rice (Oryza sativa L.) is an important calorie source for humans. With the daily increase in the world’s population and the decrease in potential available farmlands, effectively increasing rice grain production has become a growing challenge (Khush 1999; Zhang 2007). Rice grain production is thought to be determined by three major factors: panicle per plant, grain weight and grain number (Xing and Zhang 2010). Compared with the other two major factors, rice grain number exhibits a wider phenotypic variation, and is regarded as a main selection and improvement target (Yamagishi et al. 2002). Consequently, currently cultivated rice lines display dramatically increased grain number compared with its wild progenitor, the common wild rice (O. rufipogon Griff.) (Sun et al. 2001; Kovach et al. 2007).

Rice grain number is a complex agronomic trait impacted by many factors, including panicle architecture, and the initiation and outgrowth of branches and spikelets (Xing and Zhang 2010; Wang and Li 2011). Recently, a few quantitative trait loci (QTL) and genes regulating rice grain number have been identified. Among which DEP1, DEP2, LP, SP1, PAP2, and sped1-D alter panicle architecture (Huang et al. 2009; Li et al. 2009, 2010, 2011; Zhou et al. 2009; Kobayashi et al. 2010; Jiang et al. 2014), whereas Gn1a, LAX1, SPA1, TAW1, DST, and FZP control the initiation and outgrowth of branches and spikelets (M. Komatsu et al. 2003; K. Komatsu et al. 2003; Ashikari et al. 2005; Li et al. 2013; Yoshida et al. 2013). In addition, some genes, such as Ghd7, Ghd8, PROG1, IPA1, FUWA, PAY1, and An1, underlying rice grain number show pleiotropic effects in many significant agronomic- or domestication-related traits, including plant height, heading date, plant architecture, awn habit, and grain size (Jin et al. 2008; Tan et al. 2008; Xue et al. 2008; Jiao et al. 2010; Miura et al. 2010; Yan et al. 2011; Luo et al. 2013; Chen et al. 2015; Zhao et al. 2015).

Chimeric genes originate from multiple parental loci, and they have been reported in a variety of organisms (Long et al. 2003). Chimeric genes have the potential to evolve novel functions different from those of the parental loci, and provide the opportunity to study gene function and evolution (Long et al. 2003). Some chimeric genes generate novel functions and influence phenotypes. For example, sphinx determines male courtship in Drosophila (Wang et al. 2002), CYPA-TRIM5 affects HIV1-resistance in primates (Sayah et al. 2004), CHANNEL11/12 regulates multiple pathogen-resistance responses in Arabidopsis (Yoshioka et al. 2006), BS1 takes part in the normal reproductive development of maize (Jin and Bennetzen 1989; Elrouby and Bureau 2010), and chimeric ATP6 and Rfo lead to male sterility in rice and radish, respectively (Kadowaki et al. 1990; Brown et al. 2003; Z. Wang et al. 2006). In the rice genome, chimeric genes have also been widely detected (W. Wang et al. 2006). Systematic analysis demonstrated that chimeric genes account for a high percentage of the total newly evolved genes (W. Wang et al. 2006; Fan et al. 2008; Zhang et al. 2013). However, the phenotypic effects of most rice chimeric genes have not been well-investigated.

In this study, we identified a novel chimeric gene, GRAINS NUMBER 2 (GN2), which is responsible for rice grain number, as well as plant height and heading date, on the long arm of chromosome 2 in Yuanjiang common wild rice. We found that a 1094-bp sequence from LOC_Os02g45150 was inserted into the third exon of LOC_Os02g56630, and that the inserted sequence recruited its nearby sequence to generate the chimeric GN2. GN2 was generated in the natural population of Yuanjiang common wild rice. The discovery of GN2 could be useful in elucidating the molecular mechanisms underlying rice grain number and in uncovering information regarding gene emergence.

Materials and Methods

Plant materials

“YIL19” is a rice introgression line derived from a set of 106 introgression lines developed previously (Tan et al. 2007). This set of introgression lines was derived from a cross between an accession of Chinese common wild rice (Yuanjiang, O. rufipogon Griff.) and a high-yield indica cultivar (O. sativa L.) “Teqing” followed by four generations of selfing and three generations of backcrossing. Yuanjiang common wild rice was collected from Yuanjiang Country, Yunnan Province, China.

Map-based cloning

An F₂ population containing 190 individual plants derived from the cross between YIL19 and Teqing was constructed for QTL analysis. QTL analysis was carried out by interval analysis using Map Manager QTXb17 (Manly et al. 2001). An F₂ segregating population containing 4512 plants (including the previous 190 F₂ plants) was used for the genetic analysis and fine-mapping. The F₃ families, derived from the F₂ recombinant individuals, were planted to confirm the phenotypes of the F₂ recombinant individuals. Some molecular markers used for map-based cloning were published markers (McCouch et al. 2002). The newly developed molecular markers used in this study are listed in Supplemental Material, Table S4.

Gene prediction and annotation

The annotation of putative genes in the fine-mapping region were obtained from the Rice Genome Annotation Project (RGAP, MSU V7; http://rice.plantbiology.msu.edu/downloads.shtml). We also annotated genes according to the sequence information of YIL19 and Teqing from the Rice Genome Automated Annotation System (http://RiceGAAS.dna.affrc.go.jp/).

Detection of the possible transcripts by strand-specific reverse transcription-polymerase chain reaction (RT-PCR) analysis

To detect the possible transcript generated by the 1094-bp DNA fragment insertion in the locus of LOC_Os02g56630, 22 pairs of specific primers covering a 3-kb genomic region surrounding the insertion site were designed to analyze the complementary DNA (cDNA) of YIL19 and Teqing. The RT-PCR products were confirmed by gel electrophoresis and sequence analyses.

Real-time quantitative PCR (RT-qPCR) and rapid amplification of cDNA ends (RACE) analysis

Total RNA was extracted from various samples, including 7-day-old seedlings, roots, leaves, leaf sheathes, tiller bases, shoot apical meristems, young panicles shorter than 1 cm, young panicles longer than 1 cm and shorter than 5 cm, young panicles longer than 5 cm, mature panicles, and spikelet hulls, using TRIzol reagent (Invitrogen, Carlsbad, CA, http://www.lifetechnologies.com) and was purified using an RNeasy Micro Kit (QIAGEN, Valencia, CA, https://www.qiagen.com). For RT-qPCR analyses, first-strand cDNA was synthesized using an oligo(dT)18 primer (TaKaRa, http://www.takara.com) and SuperScript III Reverse Transcriptase (Invitrogen) from 2 µg of total RNA. The expression levels of GN2 and other genes were analyzed using a CFX96 Real Time System (Bio-Rad, Hercules, CA, http://www.bio-rad.com), and the rice Ubiquitin gene (LOC_Os03g13170) served as the control. Each set of experiments was repeated three times. 5’- and 3’-RACE were carried out using a 5’-Full RACE Kit with tobacco acid pyrophosphatase (TAP) and a 3’-Full RACE Core Set with PrimeScript RTase (TaKaRa) according to the manufacturer’s instructions.

Complementation test

The constructs pUbi::GN2.1 and pUbi::GN2.2 harbored the full-length sequence of GN2.1 and GN2.2 cDNA, respectively, under the control of a maize Ubiquitin promoter. These constructs were introduced into Agrobacterium tumefaciens strain EHA105 and were subsequently transferred into Teqing. Two independent lines per construct were used for further phenotypic evaluations.

Detection of the presence of GN2 in rice germplasm

We detected the presence of GN2 by two methods. First, the nucleotide BLAST search was performed using the genomic sequence and full-length cDNA sequence of GN2 against the corresponding databases. Second, a pair of diagnostic PCR primers flanking the 1094-bp DNA fragment was designed to detect the presence of GN2 in 236 rice germplasm accessions (Table S7). If GN2 is present, then the expected PCR amplification band size between the diagnostic primers is ∼2 kb, while if GN2 is absent, then the expected PCR amplification band size is ∼1 kb.

Protoplast transfection assays for regulatory sequence analysis

The sequence P1 contains the ancestral sequence of LOC_Os02g56630, including a ∼2-kb genomic sequence upstream of GN2, and P2 contains the upstream sequence of GN2 from the insert sequence of LOC_Os02g45150. All of the sequences were amplified by PCR from YIL19 genomic DNA and then inserted in the frame of pGreenII 0800-LUC (Hellens et al. 2005). Protoplasts were prepared from 7-day-old seedlings of Teqing. After protoplast preparation and transfection, firefly luciferase (LUC) and Renilla luciferase (REN) activities were measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, http://www.promega.com). Relative LUC activity was calculated by normalizing against the REN activity following the manufacturer’s instructions. In the experiments, 10 µg of each of the constructs was used, and all of the experiments were repeated three times with similar results.

Data availability

The authors state that all data necessary for confirming the conclusions presented in this article are represented fully within the article. Plant materials in this study are available upon request.

Results

YIL19 shows less grain number and pleiotropic phenotypes

An introgression line, YIL19, which originated from a set of introgression lines constructed using Yuanjiang common wild rice as the donor parent and the indica variety Teqing as the recurrent parent, exhibited less grain number compared with its recurrent parent (Figure 1, A–C). The investigation of panicle traits showed that the grain number per panicle for YIL19 and Teqing was 85.7 and 155.3, respectively (Figure 1D). Significant differences (P < 0.01) were detected in grain number per panicle, panicle length, mean length of primary branches, number of secondary branches, and grains on secondary branches between YIL19 and Teqing, while no significant difference was present in the number of primary branches (Figure 1, E–I and Figure S1). An evaluation of other important agronomic traits between YIL19 and Teqing indicated that there were significant differences (P < 0.01) in plant height, days to heading, panicle number per plant, and grain yield per plant; however, no significant differences were detected in grain weight or spikelet fertility (Figure 1, J and K and Table S1). In addition, all of the internodes of the stems were shortened in the YIL19 plants compared with the Teqing plants (Figure 1C).

Figure 1.

Phenotypes of YIL19 and Teqing. (A) Phenotypes of YIL19 and Teqing plants at the maturity stage. (B) The panicle structures of YIL19 and Teqing. The white arrows represents the place where a secondary branch emergences. Scale bar = 5 cm. (C) Stem structures of YIL19 and Teqing. On the left is the length of each internode in Teqing, and on the right is YIL19. (D–K) Comparison of grain number per panicle, panicle length, number of main panicles, mean length of primary branches, secondary branch number, grain number of secondary branches, plant height, and heading date between YIL19 and Teqing. All of the data above are given as mean ± SE (n = 20). (D–K) ** P < 0.01, determined by the Student’s t-test.

Detecting and fine-mapping GN2

YIL19 contained four O. rufipogon chromosomal segments, located on chromosomes 1, 2, 6, and 10 (Figure 2A). The grain number of the F₁ hybrid derived from the cross between YIL19 and Teqing was similar to that of YIL19 (Figure 2B). An F₂ population containing 190 individuals was genotyped using 14 simple sequence repeat markers distributed on four O. rufipogon introgression regions and evaluated for grain number on the main panicle. QTL analysis revealed that there was a QTL between markers RM5300 and RM6312 on the long arm of chromosome 2 with a LOD score of 12.24 that explained 26% of the phenotypic variance (Table S2 and Table S5). The O. rufipogon-derived allele contributed a decreasing effect on grain number. We referred to this locus as GN2. No QTL underlying rice grain number was detected in the other introgression regions, suggesting that GN2 was responsible for the less grain number phenotype of YIL19 (Figure 2C). Using 4512 F₂ individuals, the location of GN2 was narrowed down to an interval of 47 kb between RM3535 and the newly developed molecular marker R4 (Figure 2C, Figure S2, and Table S6). There were six predicted genes in this fine-mapping region based on the “Nipponbare” genome (Figure 2, D and E and Table S3).

Figure 2.

Fine-mapping and gene candidates for GN2. (A) The introgression condition of YIL19. The boxes represent rice chromosomes: solid boxes represent the chromosome segment from Yuanjiang common wild rice and the open boxes represent the segments from Teqing. (B) Grain number of YIL19 and Teqing, and their heterosis. (C) Primary mapping of GN2 between markers RM318 and RM6312. R represents the number of individuals in a population. (D) Fine-mapping of GN2 harboring a 47-kb region of AP005303. (E) The putative candidate genes for GN2 based on the annotation of Nipponbare. The open boxes represent the exons of genes based on the annotated information. (F) The gene structures of LOC_Os02g56630 in YIL19 and Teqing. The black box indicates the exon of LOC_Os02g56630; the black lines indicate introns and intergenic regions, while the red box indicates the inserted DNA segment of LOC_Os02g56630 in YIL19. (G) The mRNA splices of new transcripts. The boxes indicate the exons of the new transcripts, while different types of stripes and dots in the boxes represent the sequences having different origins. (H) The gene structure of the new transcripts. The boxes indicate the exons of new transcripts, while black lines among the boxes are the introns and intergenic sequences. The direct repeat sequences are marked by TGCATAAGCA, while the polyadenylation sites are marked by AATAAA. (I) The alternative splicing style of the new transcripts. The sequence in yellow is the exon of the new transcripts; the sequence in red represents the splicing site. The black arrows are pointing to the splicing site. The 3-bp deletion sequence in GN2.2 is underlined with a black line. Chr, chromosome; mRNA, messenger RNA.

The newly detected transcript in YIL19 is the major GN2 candidate

A sequence analysis of the six predicted genes revealed that a 1094-bp DNA fragment insertion was present at position +2275 on exon 3 of LOC_Os02g56630 in the YIL19 genome, when compared with the sequence of Teqing (Figure 2F), while no such large variation occurred in the other five genes. Furthermore, this 1094-bp DNA fragment insertion cosegregated with the less grain number phenotype. The annotation analysis also suggested that the 1094-bp DNA fragment insertion in YIL19 resulted in the premature termination of LOC_Os02g56630 messenger RNA (mRNA) translation. Therefore, LOC_Os02g56630, with three exons and two introns, which encodes an OsWAK (Wall-associated kinase) receptor-like protein kinase of 765 amino acids, appears to be a GN2 candidate. However, the expression of LOC_Os02g56630 cannot be detected in all of the tested tissues collected, which span the whole rice life cycle, in either YIL19 or Teqing. Therefore, we speculated that other unpredicted transcripts in this region may influence the rice grain number. To detect the possible transcripts, we designed 22 pairs of primers covering a ∼3-kb genomic region surrounding the insert site to conduct the strand-specific RT-PCR analysis. A ∼200-bp transcript spanning the inserted sequence and the third exon of LOC_Os02g56630 was detected in YIL19 (Figure S3), while no transcript was detected from the corresponding locus of Teqing. To determine the full length of the transcript, 5′- and 3′-RACE was conducted. We identified two unique transcripts, GN2.1 and GN2.2. Sequence analysis revealed that GN2.1 was 497 bp in length, while GN2.2 was 494 bp, with only a 3-bp deletion difference from GN2.1 (Figure 2G and Figure S4), suggesting that GN2.1 and GN2.2 were transcript variants of the same gene, which has two exons and one intron (Figure 2H). Further analyses demonstrated that the 3-bp deletion in GN2.2 occurs at position +22 of the cDNA sequence, and is caused by an alternative recognition of the RNA splice junction “AG” at the 5′-end of the second exon (Figure 2I). In addition, GN2.1 and GN2.2 showed a chimeric structure composed of 95 and 92 bp, respectively, from the 1094-bp inserted sequence and 402 bp from LOC_Os02g56630 (Figure 2H and Figure S5).

Complementation test of GN2

To verify whether the novel transcripts resulted in the less grain number phenotype of YIL19, we performed a transgenic analysis. A construct overexpressing the novel transcript GN2.1, driven by a maize Ubiquitin promoter, was introduced into Teqing plants. Twelve overexpressing transgenic lines (designated as GN2.1-Teqing) were obtained (Figure 3, A and B). A RT-qPCR analysis showed that these transgenic lines overexpressed GN2.1 transcripts compared with the control lines (Figure 3C). All of the GN2.1-Teqing transgenic lines displayed less grain number than control lines. Significant differences (P < 0.01) were detected in grains per panicle, panicle length, average length of primary branches, and number of secondary branches, compared with the control lines (Figure 3, A, B, and D–G). Additionally, GN2.1-Teqing transgenic lines showed reduced plant height and delayed heading date compared with their control lines (Figure 3, H and I).

Figure 3.

Complementation test of GN2. (A) Phenotype comparison between the GN2-overexpressing transgenic plants and control plants transformed with an empty plasmid. (B) Panicle comparison between the GN2-overexpressing transgenic plants and control plants transformed with an empty plasmid. Oe-GN2 here represents the GN2.1-overexpressing transgenic line. Scale bar = 5 cm. (C) The relative expression levels of GN2 in different GN2-overexpressing transgenic plants. CL1 represents the Teqing control line with an empty plasmid used for analysis. pOE4 and pOE6 represent GN2.1-overexpressing transgenic lines used for analysis. (D–I) Comparisons of grain number per panicle, panicle length, average length of primary branches, number of secondary branches, plant height, and heading date among GN2-overexpressing transgenic lines and their control lines. All of the data above are presented as mean ± SE (n = 20). ** P < 0.01, determined by the Student’s t-test.

We also developed a construct overexpressing the GN2.2 transcript and introduced it into Teqing (designated as GN2.2-Teqing). Ten independent transgenic lines were obtained. Similar phenotypes were observed in the transgenic plants containing these two different constructs (Figure S6). Thus, the novel gene, with two alternative transcripts, corresponded to GN2, and influenced the grain number and other plant agronomic-related traits in rice.

Expression pattern of GN2

To detect the expression pattern of GN2, we surveyed its expression profiles in YIL19 using the RNA from tissues spanning the whole life cycle, including 7-day-old seedlings, roots, leaves, leaf sheathes, tiller bases, shoot apical meristems, young panicles from different stages, mature panicles, and spikelet hulls. GN2 was constitutively expressed in all of the surveyed tissues. However, the relative expression level of GN2 appeared to be low in the tested tissues (Figure 4A).

Figure 4.

The spatial expression pattern of GN2. (A) The spatial expression pattern of GN2. Values are means ± SD (B) The occurrence frequencies of different GN2 transcripts. The dot indicates the 3-bp deletion in GN2.2. L, leaf; LS, leaf sheath; P1, young panicle shorter than 1 cm; P2, young panicle longer than 1 cm and shorter than 5 cm; P3, young panicle longer than 5 cm; P4 mature panicle; R, root; S, stem; SAM, shoot apical meristem; Se, seedling; SH, spikelet hull; TB, tiller base.

To determine the occurrence ratio of the two independent transcripts, we randomly sequenced 20 GN2 clones. Eleven clones expressed GN2.1, and the remaining nine clones expressed GN2.2, suggesting that GN2.1 and GN2.2 have similar expression levels (Figure 4B).

GN2 may influence rice grain number using a distinct pathway from known grain number regulators

Many genes associated with grain number have been identified in rice (Xing and Zhang 2010). To understand whether these genes are regulated indirectly by GN2, we carried out a RT-qPCR analysis to investigate their expression profiles. The expression levels of most of the tested genes did not show significant differences between YIL19 and Teqing in young panicles (Figure S7). Interestingly, the expression level of TAW1, and its related genes OsMADS22 and OsMADS55, showed a higher expression level in young panicles of YIL19 compared with in Teqing (Figure S7). However, a previous study revealed that the higher expression level of TAW1 led to an increase in the number of secondary branches (Yoshida et al. 2013). Thus, GN2 may influence rice grain number through a novel pathway distinct from those genes reported previously.

Emergence of the chimeric GN2 gene

To trace the emergence of GN2, we conducted a BLAST search against the O. sativa L. ssp. japonica genome set and found that 1074 bp of the 1094-bp inserted DNA fragment had a very high DNA sequence identity (> 99%) with the sequence of LOC_Os02g45150 in Nipponbare. The remaining 20 bp is the direct repetitive sequence “TGCATAAGCA” of unknown origin that flanks the 1074-bp sequence. Further analysis showed that the 1074-bp sequence belongs to the partial 5′-untranslated region sequence (or partial 5′-untranslated region and intron according to another annotation version) of LOC_Os02g45150 on chromosome 2 (Figure 5, A and B). Interestingly, the 1074-bp sequence in LOC_Os02g45150 described above was absent in the genome of C4, the accession of Yuanjiang common wild rice that is the parent of the introgression line used in this study, suggesting that the inserted sequence may be “cut” from the locus of LOC_Os02g45150 and then “pasted” into the locus of LOC_Os02g56630 in Yuanjiang common wild rice (Figure 5C and Figure S9).

Figure 5.

Emergence process of GN2. (A) Initial gene structures of LOC_Os02g45150 and LOC_Os02g56630. The boxes indicate the exons of the genes; the dark blue box indicates the CDS region based on the annotation of Nipponbare, while the connecting line in dark blue represents the intron. The black connecting line and delimiter represent chromosome regions. (B) The movement of the DNA fragment from LOC_Os02g45150 to the putative exon of LOC_Os02g56630. The box and connecting line in red represent the “sheared” DNA fragment in LOC_Os02g45150, and the triangles in purple represent the “TGCATAAGCA” repeats. (C) The gene structures of LOC_Os02g45150 and LOC_Os02g56630 after the insertion event. (D) The emergence of the GN2 gene. The box and triangle in green represent the exons of GN2, and the connecting line in green represents the intron of GN2. The dotted line in black represents the ancestral gene structure of LOC_Os02g56630, and the dotted line in red represents the inserted genomic sequence. The red right angle arrow represents the TSS, and the AATAAA represents the polyadenylation site of GN2. CDS, coding sequence; Chr, chromosome; TSS, transcription start site.

Based on the GN2 gene construct, we inferred that the 1094-bp inserted sequence from the locus of LOC_Os02g45150 contributed the transcriptional start site, the first exon, the whole intron, and the partial second exon of GN2, and LOC_Os02g56630 provided the rest of the second exon and the polyadenylation [poly(A)] site (Figure 5D). Therefore, we concluded that GN2 is a chimeric gene produced by the fusion of the two sequences from different origins.

To analyze the possible transcriptional regulatory region of GN2, a transient expression assay was conducted. The inserted sequence from LOC_Os02g45150 had a weak regulatory ability, while the upstream sequence of GN2 from LOC_Os02g56630 did not, indicating that GN2 acquired its regulatory sequences from the 1094-bp inserted sequence (Figure S8).

A BLAST similarity search showed that GN2 is unique in the rice genome. It also revealed that GN2 has no homolog in any other organisms. We further analyzed the presence of GN2 in 236 rice accessions representing diverse germplasms, including both wild and cultivated species (Table S7). None of the tested accessions exhibited the presence of GN2, suggesting that GN2 may be newly generated in the genome of Yuanjiang common wild rice.

To detect the occurrence of GN2 in the Yuanjiang common wild rice population, we analyzed the presence of GN2 in 22 individuals from the local natural population of Yuanjiang common wild rice in Yunnan Province, China. We found that 12 individuals contained homozygous GN2, four individuals exhibited heterozygosity at the GN2 locus, and six individuals did not contain GN2, suggesting that GN2 has a high frequency in the local natural population of Yuanjiang common wild rice, although it is not completely fixed in this population.

Discussion

In this study, we identified a novel gene, GN2, underlying rice grain number, using an introgression line, YIL19. GN2 exhibits pleiotropism, not only influencing grain number, but also affecting the plant height and heading date. Interestingly, GN2 is a novel chimeric gene that emerged through the shuffling of a 1094-bp genomic fragment from LOC_Os02g45150 to LOC_Os02g56630 on the same chromosome. We also found that GN2 is unique to the rice genome and newly emerged in the natural population of Yuanjiang common wild rice.

The insertion event produced new GN2 transcripts. Sequence analysis revealed that the longest predicted open reading frame in the GN2 transcript was 213-bp, encoding a 70-amino acid polypeptide. Therefore, we speculated that GN2 may exert its function as a novel polypeptide.

The panicle branch plays an important role in determining rice grain number (Wang and Li 2011). Both the branch meristem’s activities and the elongation of branches regulate rice grain number. Previous studies showed that a couple of genes, such as LAX1, Gn1a, and TAW1, influence rice grain number by regulating the number of branches (K. Komatsu et al. 2003; Ashikari et al. 2005; Yoshida et al. 2013). Several genes, such as SP1, control rice grain number by regulating the elongation of panicles, primary branches, and secondary branches (Li et al. 2010), while sped1-D influences the length of pedicels and secondary branches (Jiang et al. 2014). Unlike the genes described above, GN2 has a specific effect on the elongation of the primary branches. The shortened length of the primary branches leads to a lower number of secondary branches in YIL19. This suggested that GN2 regulates rice grain number in a manner distinct from those of known genes.

Chimeric genes are able to generate new functions that differ from those of parent sequences (Long et al. 2003). Although chimeric genes form a certain proportion of the rice genome, and many of them have expression profiles and undergo selection, their phenotypic effects are mostly unknown (W. Wang et al. 2006; Fan et al. 2008; Zhang et al. 2013). In this study, we verified that GN2 is a novel chimeric gene that exhibits a distinctive gene structure and expression pattern from its parental genes, and we also found that GN2 acquires novel functions and has a significant effect on morphological traits.

GN2 arose through the insertion of a 1094-bp sequence that may come from LOC_Os02g45150. Its emergence is consistent with the “cut and paste” DNA fragment shuffling model, having the direct repeat sequence “TGCATAAGCA” flanking the inserted region. RNA-mediated retroposition and DNA-mediated recombination are known to be two major mechanisms for chimeric gene emergence (Arguello et al. 2007; Kaessmann 2010). Most of the rice chimeric genes arose through one of the two mechanisms (Turcotte et al. 2001; Bennetzen 2005; W. Wang et al. 2006). Although we detected the flanking direct repeat, one of the hallmarks of retroposition, in GN2, we were unable to find other structural hallmarks of retroposition, such as the loss of intron(s) and the presence of poly(A) tracts. Furthermore, the repeat sequence “TGCATAAGCA” does not resemble any characterized transposon or retrotransposon, and this insertion event induced by the “TGCATAAGCA” repeat sequence has rarely been detected in the rice genome. Therefore, we speculated that the formation of the chimeric GN2 may not have occurred through the mechanisms of RNA-mediated retroposition or DNA-mediated recombination, but through a different approach of chimeric gene origination.

Comparative genomic studies revealed that the emergence of novel genes in the rice genome occurred at a higher frequency and more rapid pace than other organisms, such as primates, and the process of gene emergence appears to be ongoing (Ge et al. 1999; Vaughan et al. 2003; W. Wang et al. 2006). Interestingly, GN2 is only detected in the genome of Yuanjiang common wild rice. Furthermore, it exhibited a high frequency (up to 63%) in the natural population of Yuanjiang common wild rice. Hence, we speculated that GN2 had newly emerged in this natural population. The GN2 high frequency in the local natural population may be subject to positive selection, and thus GN2 may be functional.

Many newly emerged genes, which have evolved novel functions, are involved in development, reproduction, and stress responses in various organisms (Tautz and Domazet-Loso 2011; Chen et al. 2013). These novel genes have the potential to be used in the improvement of important agronomic traits and broaden germplasm resources for breeding. Furthermore, understanding the mechanism of gene emergence will shed light on artificial synthesis of functional genes in the future. The identification of the newly emerged GN2, which obviously influences agronomic traits, provides more information to elucidate the molecular mechanisms underlying rice grain number and new insight into general phenotypic evolution.

Supplementary Material

Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.116.188201/-/DC1.