Substantial expression of novel small open reading frames in Oryza sativa

Abstract

In our previous integrated study combining informatics and molecular biology analyses, we revealed that Arabidopsis small open reading frames (sORFs) predicted by computational analysis have biological functions in morphogenesis. Here, we report that sequences homologous to Arabidopsis sORFs are abundant in intergenic regions of the rice genome. These sequences represent a subset of non-protein-coding DNA, and some are transcribed into mRNA. These results indicate that many sORFs associated with morphogenesis are hidden in the genomes of crop species.

In plants, secreted peptides encoded by small coding genes play roles in development, self-incompatibility, pollen tube guidance, and defense responses against insect herbivores, pathogens, and environmental stresses.^1-6 Similar to plant hormones, these peptides function as signaling molecules that bind to receptor kinases. Peptide signaling is now recognized to be as important as signaling by classical plant hormones. Plant genomes encode many leucine-rich-repeat receptor kinases, but only a few of the ligand peptides for these receptors have been identified.^7,8 It is likely that many sequences encoding hormone-like peptides are hidden in plant genomes. Most hormone-like peptides have been identified in studies on Arabidopsis Genome Initiative (AGI) code genes.⁹ However, small coding genes (those encoding < 100 amino acids) tend to be missed when AGI code genes are annotated, because small coding sequences are often computed as false-positive predictions.¹⁰ To overcome the false-positive prediction problem, we developed computational approaches to predict small open reading frames (sORFs; 30–100 amino acids) with high coding potential, and identified 7,901 coding sORFs in intergenic regions of the Arabidopsis genome.¹¹ Of these, 2,099 sORFs showed significant expression in at least one condition or organ out of 16 organs and 17 environmental conditions.¹² Furthermore, approximately 10% (49/473) of the manually selected coding sORFs resulted in various morphological changes when overexpressed in transgenic Arabidopsis plants. These results indicated that sORFs predicted by computational analysis play significant roles in growth and morphogenesis in Arabidopsis.

CLE peptides, a representative class of peptide hormones, were first found in Arabidopsis, and function as key signaling molecules in development.¹³ Sequences homologous to those encoding Arabidopsis CLEs are conserved in Lotus, Oryza, and Zinnia. Indeed, treatment with CLE peptides induced cell differentiation and meristem maintenance in Lotus, Oryza, and Zinnia.^14-16 As well as sequences encoding CLE peptides, those encoding other peptide hormones (ROT4, RTFL/DVL, EPF/EPFL) are also conserved in higher plant species.^17-26 The conservation of coding sequences among multiple plant species provides good evidence for the functionality of these peptides.

The aim of this study was to explore the functionality of newly identified homologs of Arabidopsis sORFs in rice. We searched the rice genome for sequences homologous to 49 Arabidopsis sORFs associated with morphogenesis, and examined their RNA expression.

Rice Homologs of Arabidopsis sORFs

To identify sequences homologous to the Arabidopsis sORFs in rice, we conducted BLAST searches (E-value < 0.01) against the whole Oryza sativa genome. The amino acid sequences encoded by sORFs were aligned by CLUSTALW. Out of 49 AtsORFs, 35 AtsORFs had homologous sequences in the rice genome (IRGSP-1.0). Six of these sORFs are shown in Figure 1 as examples. Previous, we reported that transgenic Arabidopsis plants overexpressing AtsORF1189, 1411, 1626, 0359, 5537, and 2874 showed enlarged rosette leaves, altered leaf color, a seedling-lethal phenotype, small plant size, and altered leaf number, respectively. These results provided evidence for the biological functions of these sORFs.¹² The sequences of OsORF1189L and 5537L were very similar to those of AtsORF1189 and 5537, respectively (Fig. 1A), whereas the other rice sORFs homologs showed moderate similarity to their homologous Arabidopsis sORFs. However, when we considered the amino acid sequence similarities between rice and Arabidopsis sORFs, the sequences of OsORF1626L, 2874L, 0359L, and 1411L were very similar to their corresponding sequences in Arabidopsis (Fig. 1A). AtsORF0359 had eight cysteines in the coding sequence; this even number of cysteine residues is often found in functional cysteine-rich peptides such as LURE and EPF/EPFL.^2,24 However, an odd number of cysteine residues was conserved between OsORF0359L and AtsORF0359. The start codon of OsORF0359L should be identified to predict this functional sORF accurately.

Figure 1. Molecular features of rice homologs of functional Arabidopsis sORFs. (A) Comparison of amino acid sequences encoded by sORFs between rice and Arabidopsis. Homologous sequences to Arabidopsis sORFs in rice were identified by BLAST searches. Sequences were aligned using CLUSTALW. Asterisk (*) indicates positions with a single fully conserved residue. Colon (:) indicates conservation between groups with strongly similar properties (scoring > 0.5 in Gonnet PAM 250 matrix). Period (.) indicates conservation between groups with weakly similar properties (scoring = < 0.5 in Gonnet PAM 250 matrix). (B) Physical map of rice sORFs on rice genome. DNA sequences for rice sORFs were input into RiceGE database (http://signal.salk.edu/cgi-bin/RiceGE), and physical locations of rice sORFs were identified by BLAST searches. (C) Detection of transcribed rice sORFs. A mixture of cDNAs was synthesized from mRNAs from 40 different samples (22 different organs, and shoots and roots of plants in nine different environmental conditions). Asterisk (*) represents expected PCR product size.

Physical Map of Rice sORFs on Rice Genome

Next, we determined whether the rice sORFs homologs were annotated as coding genes in the rice genome database. Out of 35 sORFs, 22 sORFs were in intergenic regions of the rice genome that have no annotated genes. For example, OsORF1189L, 1626L, 0359L, and 1411L were mapped in intergenic regions (Fig. 1B). This result implies that many sORFs are hidden in the rice genome. The other 13 sORFs overlapped with annotated genes in the rice genome. For example, OsORF2874L and OsORF5537L overlapped with Os02 g56120 and Os06 g12030, respectively (Fig. 1B). Os02 g56120 is annotated as an auxin-responsive AUX/IAA family protein that is involved in IAA signal transduction. However, an orthologous gene has not been identified in the Arabidopsis genome in the PlantPAN database (http://plantpan.mbc.nctu.edu.tw). Of the 13 sORFs, 7 showed a similar trend, indicating that some Arabidopsis sORFs are simply annotated genes in Oryza.

Os06 g12030 encodes a putative glutaredoxin-related protein that functions as an oxidoreductase. In TAIR version 8, AtsORF5537 neighbors AT4G08550, which encodes a putative glutaredoxin-related protein. Therefore, sORF5537 is likely to be a part of a known gene in both Arabidopsis and Oryza. Indeed, the expression pattern of AT4G08550 was quite similar to that of AtsORF5537 in HanaDB-AT (http://evolver.psc.riken.jp/seiken). Of the 13 sORFs, 6 sORFs showed a similar trend to that of sORF5537, indicating that these 6 sORFs are likely to be part of known genes. These results show that comparative genomics analyses are useful for validating sORFs as independent transcriptional units.

Expression Analysis of Rice sORFs

Next, we examined whether the sORFs in rice were transcribed into mRNA. We conducted RT-PCR analyses using a mixture of cDNAs, which were synthesized and prepared from 40 different samples. Out of 35 rice sORFs, 29 were detected as PCR products of the predicted size. Six of these sORFs are shown in Figure 1C. Because 83% (29/35) of the analyzed sORFs were transcribed into RNA in Oryza,¹² it is likely that many other sORFs are present in the rice genome and are expressed. In future research, deep transcriptomics and proteomics analyses will be used to identify the transcriptional units of sORFs and to provide evidence of their encoded peptides, respectively.

Hanada K, Higuchi-Takeuchi M, Okamoto M, Yoshizumi T, Shimizu M, Nakaminami K, Nishi R, Ohashi C, Iida K, Tanaka M, et al. Small open reading frames associated with morphogenesis are hidden in plant genomes. Proc Natl Acad Sci U S A. 2013;110:2395–400. doi: 10.1073/pnas.1213958110.