Full-length cDNAs were isolated for the RPK1 from O. rufipogon (accession IRGC105491, designated as Oryza rufipogon receptor-like protein kinases 1, OrufRPK1; Genbank accession number GI: 380710170) and O. sativa ssp. indica cv. MR219 (named as Oryza sativa ssp. indica cv. MR219 receptor-like protein kinases 1, OsI219RPK1; Genbank accession number GI: 380710172). The coding regions of the RPK1 genes from the two rice species were identical in length (2055 bp) whilst the 5′ UTR of OrufRPK1 and OsI219RPK1 were 404 bp and 103 bp respectively and the 3′ UTR were 386 bp and 418 bp in length. A putative polyadenylation signal site was found in the presumptive 3′ UTR region of each cDNA. Alignment of each cDNA with the corresponding genomic sequences found that all exon-intron boundaries obey the canonical splice donor-acceptor rule (GT-AG). The length of the intron in OrufRPK1, OsI219RPK1, OsI9311RPK1 and OsJNipponRPK1 was found to be varied: 3752 bp, 3372 bp, 3345 bp and 3357 bp, respectively (Table 1). Analysis of the coding regions revealed the first start codon (ATG) in exon 1 and the first stop (TGA) located at exon 2. Both OrufRPK1 and OsI219RPK1 were found to share same genomic location of the ATG and TGA sites. Analysis of structural properties of the RPK1 proteins using SMART program [25,26] suggests that all of the four RPK1 sequences contain a predicted signal peptide, 6 extracellular LRR repeats, a transmembrane domain and a cytoplasmic serine/threonine kinase domain (Figures 1 and 2).

Southern-blot hybridization of rice genomic DNA using a 320 bp 3′ UTR fragment as probe indicated that only a single RPK1 copy exists in the genome of O. rufipogon and O. sativa ssp. indica cv. MR219 (Figure 3). As expected from restriction site analysis of the sequences, EcoRI fragments of approximately 3.3 kb and 2.9 kb hybridized with the 3′ UTR probe in the wild and cultivated genomes, respectively, consistent with the different intron sizes predicted from the sequence analysis. In marked contrast, the analysis revealed only weak signals for BamHI and HindIII restriction fragments, which were all larger than 8 kb in size (data not shown).

Multiple sequence alignment of the RPK1 coding and amino acid sequence from the O. rufipogon, O. sativa ssp. indica cv. MR219, O. sativa ssp. indica cv. 9311 and O. sativa ssp. japonica cv. Nipponbare identified a panel of SNPs (Figures 1 and 2). From a total of 12 SNPs identified within the exonic regions, six specifically defined the O. rufipogon allele, while 4, 1 and 1 SNPs characterized the indica (MR219), indica (9331) and japonica, respectively (Figure 2). Further characterization of the degree of genotypic variation between different rice species revealed a major insertion of a 392 bp segment at genomic sequence positions 2731 to 3122 bp in the O. rufipogon OrufRPK1. Additionally, 22 unique InDels were detected in the wild rice counterpart (data not shown). Among the 12 SNPs detected, four were found to involve pyrimidine substitutions (C and T bases), purine substitutions (A and G bases) occurred at three sites, and the remaining SNPs involved base substitutions between the purine and pyrimidine groups (Figure 2). Five SNPs incurred non-synonymous substitutions, with three and two amino acid substitutions detected within an LRR motif and the kinase domain, respectively.

Using the OrufRPK1 protein sequence as the search string for identification of orthologous sequences, a total of 11 orthologous RPK1 gene sequences were identified from the Rice Kinase Database [27,28], Rice Genome Annotation Project [29,30] and Gramene Database [31,32] as shown in Table 2. Phylogenetic analysis revealed two well-supported groups (designated as group I and group II) of orthologous RPK1 genes (Figure 3). The OrufRPK1, OsI219RPK1, OsI9311RPK1 and OsJNipponRPK1 grouped together with BD_RLL1, SB_RLL1, Zm_RLL1, Zm_RLL2 and OsJ_RLL1 in group II (Figure 4). The group II could be subdivided into three branches, with the OrufRPK1 grouped along with OsI219RPK1 and OsJNipponRPK1 in a strongly supported (100%) cluster. In addition, the topology of the maximum likelihood method generated from these sequences is in agreement with the previously reported classification grouping of most of the orthologous RPK1 genes to the LRR III families of RLK genes [33].

The expression patterns of OrufRPK1 and OsI219RPK1 were analyzed in various tissues from five different developmental stages (Figure 5). During the reproductive booting phase, RPK1 expression in the panicle tissue of both species was significantly higher than at other developmental stages. A relatively lower expression level was observed in MR219 when compared to the wild rice (relative quantitative value 50, as compared to 81 in O. rufipogon, Figure 5). Overall, OrufRPK1 transcripts had higher expression levels across all developmental stages. After the booting stage, gene expression gradually decreased until the flowering stage. The expression of OrufRPK1 of the milk grain stage coincides with the onset of the ripening phase. Statistical analysis using two-way ANOVA with rescaled normalized expression level from the RPK1 of O. rufipogon and O. sativa ssp. indica cv. MR219 collected at different rice development stages, revealed significant differences (p < 0.05) between species and between stages within each species.

The three dimensional structure of OrufRPK1, OsI219RPK1, OsI9311RPK1 and OsJNipponRPK1 proteins were modeled using Robetta webserver (Figure 6 and Figure S1, Supplementary) [34]. The modeled OrufRPK1 protein possesses two main domains, the LRR domain and kinase domain (Figure 6). Structural assessment by Ramachandran plot predicted the structure to have 83.4% of residues in the core region and the rest of the residues within allowed regions. The predicted three dimensional structure of OrufRPK1 LRR domain contains similar unique features with the comparative modeling template, polygalacturonase-inhibiting protein, which contains four LxxLxLxxN motifs (Figure 6b). The secondary structure of the LRR domain of the OrufRPK1 protein is distributed as Helixes (α + 3/10) and β-strands (β1 and β2), together comprising of 35% of the structure with the remainder forming turns and bends, which are responsible for the curving nature of LRR proteins (Figure 6b) [35]. Figure 6c shows several highly conserved residues within the kinase domain of the OrufRPK1 protein, including a Rossmann motif, protein acceptor and activation segment.

Plant LRR-RLKs are known as cell surface reporters which regulate developmental and defence related signal transduction pathways. In rice, a model which represents monocots, 380 LRR-RLK encoding sequences have been identified in the genome [36], almost twice the number found in Arabidopsis. However, only a small fraction of the rice LRR-RLK members have been functionally investigated or molecularly characterized. The LRR-RPK-encoded OrufRPK1 gene characterized in the present study was partially mapped in previous work using an ab initio gene prediction method [23]. However, high-accuracy identification of promoter, transcriptional start site and terminal exons remains difficult for ab initio approaches. Here, we employed direct mapping of the sequenced OrufRPK1 cDNA to its genomic sequence and performed comparative analysis with orthologs in cultivated rice genomes.

Seeds of O. rufipogon accession IRGC105491 and O. sativa ssp. indica cv. MR219 were obtained from the Malaysian Agricultural Research and Development Institute (MARDI) Rice Genebank, Seberang Perai, Malaysia. All plants were grown under 12 h light/12 h dark at 28 °C and 75% relative humidity in a glasshouse. Leaves and whole plants (for 8 day old seedlings), panicle (at booting, heading and flowering) and grain (milk grain stage) were collected for RNA extraction.

Total RNA was isolated during the vegetative (leaf and whole plant), reproduction (booting, heading and flowering panicle) and ripening (milk grain) phases [51]. Concentration of total RNA was measured by spectrophotometer (Eppendorf, Germany). The integrity of total RNA extracted was confirmed by 1% (w/v) agarose gel electrophoresis. Only samples with 260/280 nm values between 1.9 and 2.1 and 260/230 nm values greater than 2.0 were used for cDNA synthesis by High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems, USA) according to the manufacturer’s instructions.

FirstChoice^® RLM-RACE Kit (Ambion, USA) was used to amplify the full length cDNA of the OrufRPK1 sequence. The rapid amplification of 5′ cDNA ends reactions were performed with 10 μg total RNA. The RNA processing and 5′ adaptor ligation for cDNA synthesis was performed according to the manufacturer’s instructions. The outer 5′ RLM-RACE PCR was performed using 5′ RACE gene-specific outer primer: 5′-AACTGCAGCGGATACGACCACAC-3′ and 5′ RACE outer primer; while the inner 5′ RLM-RACE PCR was performed with a 5′ RACE gene-specific inner primer: 5′-TGCAGCGGATACGACCACACT-3′ and a 5′ RACE inner Primer. For the rapid amplification of 3′ cDNA ends, 1 μg total RNA was used. The 3′ adaptor ligation for cDNA synthesis was performed according to the manufacturer’s instructions. The outer 3′ RLM-RACE PCR was performed with a 3′ RACE gene-specific outer primer: 5′-AGGTTCTTTGGCTGCCTGTATG-3′ and a 3′ RACE outer primer; while the inner 3′ RLM-RACE PCR was performed with a 3′ RACE gene-specific inner primer: 5′-TCATCGTCTGCCCCCTTTGTG-3′ and a 3′ RACE inner primer. The 5′ and 3′ RLM-RACE PCRs were performed using AccuPrime™ GC-Rich DNA Polymerase (Invitrogen, California). PCR products were analyzed by 1% agarose gel electrophoresis, and then cloned using a CloneJET™ PCR Cloning Kit (Fermentas, Lithuania).

Gene sequences were analyzed using GenScan [52,53]. The predicted peptide sequences were analyzed using Simple Modular Architecture Research Tool [25,26]. The kinase domains of OrufRPK1, OsI219RPK1, OsI9311RPK1 and OsJNipponRPK1 were identified by SMART analysis. The signal peptide was predicted using SignalP 4.0 Server [54,55]. The transmembrane helix was predicted using Transmembrane Protein Topology with a Hidden Markov Model (TMHMM) Server 2.0 [56,57]. The LRR domain and protein kinase domain were predicted using protein domain families (Pfam) database [58]. The sequences of protein kinases, LRK1 (GI: 8132685), CLAVATA1 (At1g75820), ERECTA (At2g26330) and TMKL1 (At3g24660), were aligned with OrufRPK1 (GI: 380710170), OsI219RPK1 (GI: 380710172), OsI9311RPK1 (GI: 218188631) and OsJNipponRPK1 (GI: 53792194) using ClustalW [59,60]. The characterization of the rice kinase was based on the presence or absence of conserved amino acids, arginine and aspartic acid (D), in kinase subdomain VIb and aspartic acid in kinase subdomain VII [61].

Southern hybridization analysis was used to confirm presence of single copy of the RPK1 gene in both wild and cultivated rice genomes. Total genomic DNA of genomic DNA from O. rufipogon and O. sativa cultivar MR219 were digested separately using EcoRI, BamHI and HindIII and fractionated in agarose gel. Digested DNA was Southern blotted to a nylon membrane, and hybridized with a 320 bp probe derived from the RPK1 3′ UTR region. All blotting procedures and immunological detection were carried out according to the DIG DNA labeling and detection Kit application manual (Roche, USA).

Primer Express 2.0 software (Applied Biosystems, USA) was used to design all primers and facilitate real time qRT-PCR analysis. Applied Biosystems 7300 Real Time PCR System was used to perform relative quantitative (Comparative CT Method) real time qRT-PCR experiments. A total of four replicates of each stage of each line were used for each gene specific primer set: OrufRPK1-F and OsI219RPK1-F: 5′-AGGTTCTTTGGCTGCCTGTATG-3′, OrufRPK1-R and OsI219RPK1-R: 5′-AGG TTCTTTGGCTGCCTGTATG-3′, and housekeeping gene primer sets [62]: eEF-1α-F: 5′-TTTCACTC TTGGTGTGAAGCAGAT-3′, eEF-1α-R: 5′-GACTTCCTTCACGATTTCATCGTAA-3′, UBQ5-F: 5′-ACCACTTCGACCGCCACTACT-3′, UBQ5-R: 5′-ACGCCTAAGCCTGCTGGTT-3′. Each 50 ng cDNA sample, was mixed with 1× SYBR Green PCR Master Mix (Applied Biosystems, USA), and 200 nM of each forward and reverse primer then added to PCR grade dH₂O to 25 μL in an 8 well optical reaction strip (Applied Biosystems, USA). Three replicates of a negative control with dH₂O as template were also included for each primer pair. The real time qRT-PCRs were performed under the following conditions: 1 cycle of 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and 1 min at 60 °C. geNORM v3.4 software (Primer-Design, UK) was used to analyze relative expression of each transcript in order to obtain more accurate normalization [63]. Data were analyzed by two-way repeated measures analysis of variance (ANOVA) with p < 0.05 considered significant using Prism 5 (GraphPad Software, USA). Rescaled normalized expression target and standard error of rescaled normalized expression target generated from geNORM v3.4 software (Primer-Design, UK) were used for statistical analyses.

OrufRPK1 amino acid sequence was used for homology searches against the Rice Kinase Database [27,28,37], Rice Genome Annotation Project database [29,30] and Gramene Database [31,32]. A total of 15 orthologous OrufRPK1 gene sequences were selected for phylogenetic analysis (Table 2). A maximum likelihood phylogenetic tree was constructed using MrBayes 3.2.0 program [64].

Homology modeling of OrufRPK1, OsI219RPK1, OsI9311RPK1 and OsJNipponRPK1 used a similarity based approach for modeling three dimensional proteins with Modeler Version 9v8 [65,66]. The LRR domain of polygalacturonase-inhibiting protein (1ogqA) [67] and kinase domain of interleukin-1 receptor-associated kinase 4 (2nruB) [68] templates were used for comparative modeling of the four predicted protein structures. These domains were defined using ROBETTA web server [34] which provides both ab initio and comparative models of protein domains [69]. Ginzu algorithm searches for homologs in various protein structures through PSI-BLAST searches were used to define domain boundaries for the given input sequence [70].