Biochemical and Expression Analyses of the Rice Cinnamoyl-CoA Reductase Gene Family

Cinnamoyl-CoA reductase (CCR) is the first committed enzyme in the monolignol pathway for lignin biosynthesis and catalyzes the conversion of hydroxycinnamoyl-CoAs into hydroxycinnamaldehydes. In the rice genome, 33 genes are annotated as CCR and CCR-like genes, collectively called OsCCRs. To elucidate the functions of OsCCRs, their phylogenetic relationships, expression patterns at the transcription levels and biochemical characteristics were thoroughly analyzed. Of the 33 OsCCRs, 24 of them encoded polypeptides of lengths similar to those of previously identified plant CCRs. The other nine OsCCRs had much shorter peptide lengths. Phylogenetic tree and sequence similarities suggested OsCCR4, 5, 17, 18, 19, 20, and 21 as likely candidates for functional CCRs in rice. To elucidate biochemical functions, OsCCR1, 5, 17, 19, 20, 21, and 26 were heterologously expressed in Escherichia coli and the resulting recombinant OsCCRs were purified to apparent homogeneity. Activity assays of the recombinant OsCCRs with hydroxycinnamoyl-CoAs revealed that OsCCR17, 19, 20, and 21 were biochemically active CCRs, in which the NAD(P)-binding and NADP-specificity motifs as well as the CCR signature motif were fully conserved. The kinetic parameters of enzyme reactions revealed that feruloyl-CoA, a precursor for the guaiacyl (G)-unit of lignin, is the most preferred substrate of OsCCR20 and 21. This result is consistent with a high content (about 70%) of G-units in rice lignins. Phylogenetic analysis revealed that OsCCR19 and 20 were grouped with other plant CCRs involved in developmental lignification, whereas OsCCR17 and 21 were closely related to stress-responsible CCRs identified from other plant species. In agreement with the phylogenetic analysis, expression analysis demonstrated that OsCCR20 was constitutively expressed throughout the developmental stages of rice, showing particularly high expression levels in actively lignifying tissues, such as roots and stems. These results suggest that OsCCR20 is primarily involved in developmental deposition of lignins in secondary cell walls. As expected, the expressions of OsCCR17 and 21 were induced in response to biotic and abiotic stresses, such as Magnaporthe grisea and Xanthomonas oryzae pv. oryzae (Xoo) infections, UV-irradiation and high salinity, suggesting that these genes play a role in defense-related processes in rice.


INTRODUCTION
Plants are constantly confronted by both biotic and abiotic stresses, leading to significant reductions in their productivity (Strange and Scott, 2005;Vinocur and Altman, 2005;Oerke, 2006;Chakraborty and Newton, 2011). Abiotic stresses, including drought, salinity, and extreme temperature, are the primary factors in crop loss and can reduce the average yields of major crop plants by more than 50% (Boyer, 1982;Oerke, 2006). Biotic stresses, such as infection by pathogens, can cause serious reduction of cereal production (Strange and Scott, 2005;Oerke, 2006;Chakraborty and Newton, 2011). It has been reported that actual losses of worldwide rice production due to biotic stresses, in the period of 2001-2003, comprised an estimated 37.4% of the total attainable production (Oerke, 2006). To cope with biotic and abiotic stresses, plants have developed a wide array of defense mechanisms such as the fortification of cell walls, production of phytoalexins, and accumulation of reactive oxygen species (Moura et al., 2010;Ahuja et al., 2012;Großkinsky et al., 2012;Miedes et al., 2014;Rejeb et al., 2014).
Lignin is complex aromatic polymer primarily composed of p-hydroxyphenyl (H)-, G-and syringyl (S)-units derived from monolignols. Lignin is predominantly deposited in the secondary cell walls of xylem and fiber cells and makes the cell walls rigid and impervious (Campbell and Sederoff, 1996;Donaldson, 2001;Bonawitz and Chapple, 2010;Vanholme et al., 2010). The lignified secondary cell walls are important for the water conduction and mechanical support of vascular plants, and serve as a physical barrier against pathogens and herbivores (Campbell and Sederoff, 1996;Donaldson, 2001;Vanholme et al., 2010;Miedes et al., 2014). In addition to developmental deposition, the synthesis and deposition of lignin-related phenolics are induced in response to biotic and abiotic stresses (Moura et al., 2010;Hamann, 2012;Miedes et al., 2014).
In the MSU Rice Genome Annotation Project (RGAP) database, we found 33 genes annotated as CCR and CCR-like genes, collectively called OsCCRs. The gene expression profiles of different developmental stages, organs and stress conditions, and the activity of enzyme toward hydroxycinnamoyl-CoA substrates were examined for the functional characterization of OsCCRs in rice. An activity assay of recombinant OsCCR proteins revealed that OsCCR17, 19, 20, and 21 were biochemically functional CCRs in rice. Expression and phylogenetic analyses were performed to elucidate the physiological role of OsCCRs, and suggested that OsCCR19 and 20 are primarily involved in developmental lignification, while OsCCR17 and 21 likely play a role in defense responses.

Plant Growth and Materials
Sterilized seeds of wild-type rice plants (O. sativa L. spp. Japonica cv. Dongjin) were germinated on Murashige and Skoog (MS) medium (Duchefa, Harlem, Netherlands) in a growth chamber with a 12 h photoperiod and temperature of 28 • C. Ten-day old seedlings were transferred to soil and grown in a greenhouse at 28 • C during the day and 20 • C at night. Stem and leaf samples were collected from 10-week-old rice plants, and panicle samples were collected from 14-week-old rice plants. Root and shoot samples were collected from 10-day old rice seedlings.

Multiple Sequence Alignments and Phylogenetic Analysis of OsCCRs
Deduced protein sequences of OsCCRs and functional CCRs identified from other plant species were retrieved from the MSU RGAP database (http://rice.plantbiology.msu.edu/, Kawahara et al., 2013) and the National Center for Biotechnological Information (https://www.ncbi.nlm.nih.gov/) database, respectively. Multiple amino acid sequence alignment was performed with Clustal-W ( Thompson et al., 1994), and a phylogenetic analysis was conducted with MEGA ver. 6 (Tamura et al., 2013) using the neighbor-joining method.

Cloning of OsCCRs
Total RNA was isolated from 8-week-old rice leaves with RNAiso (Takara, Shiga, Japan). The first cDNA was synthesized using the total RNA and SuPrimeScript RT premix with an oligo dT primer (GeNet Bio, Daejeon, Korea). Cloning primers for OsCCR genes were designed according to the sequences in the MSU RGAP database. The amplification primers and polymerase chain reaction (PCR) conditions are provided in Supplementary Table 1. PCR was performed using Solg TM Pfu DNA Polymerase (SolGent, Daejeon, Korea). The resulting PCR products were subcloned into the pGEM TM -T Easy vector (Promega, Madison, WI, USA) or pJET 1.2 blunt cloning vector (Thermo Scientific, Carlsbad, CA, USA). After sequence confirmation, each OsCCR gene was cut out with the appropriate restriction enzymes and inserted into the pET28a(+) vector (Novagen, Madison, WI, USA). The resulting OsCCR/pET28a(+) constructs were individually transformed into E. coli BL21(DE3) cells for heterologous expression of OsCCRs.

Expression and Purification of Recombinant OsCCRs
The E. coli transformants harboring the OsCCR/pET28a(+) construct were grown at 37 • C until an OD 600 of ∼0.6 in LB medium containing kanamycin (25 µg/mL) was achieved. At that point, 0.1 mM isopropyl β-D-thiogalactopyranoside (IPTG) was added in the culture for induction. After additional incubation at 18 or 25 • C for 16 h, the cells were harvested by centrifugation (5,000 g for 15 min). Cell pellets were resuspended in phosphatebuffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4, 2 mM KH 2 PO 4 ) supplemented with lysozyme (1 mg/mL) and phenylmethylsulfonyl fluoride (1 mM). The resuspended cells were sonicated on ice, and the crude protein extracts were obtained by centrifugation (15,900 g for 20 min, 4 • C). The crude protein samples were mixed with Ni-NTA Agarose beads (Qiagen, Hilden, Germany) and incubated at 4 • C for 2 h with agitation. The mixtures were packed into a chromatography column and washed three times with a five-column volume of 20 mM imidazole in Tris buffer (50 mM Tris, pH 8.0, 300 mM NaCl). The recombinant OsCCRs were eluted with 50-100 mM imidazole in Tris buffer. The eluted proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Enzymatic Synthesis of Hydroxycinnamoyl-CoAs
For the OsCCR activity assay, hydroxycinnamoyl-CoAs were synthesized by the method described by Beuerle and Pichersky (2002). Arabidopsis 4-coumarate:CoA ligase 1 (At4CL1) was cloned from A. thaliana cDNA, and the resulting gene was inserted into pET28a(+) vector (Supplementary Table 1) (Stuible and Kombrink, 2001). The recombinant At4CL1 was expressed in E. coli and purified with Ni-NTA Agarose beads according to the methods described above. To synthesize the hydroxycinnamoyl-CoA esters, 3.3 mg hydroxycinnamic acid (p-coumaric, ferulic, or sinapic acids), 2 mg CoA, and 6.9 mg ATP were dissolved into a total volume of 10 mL of 50 mM Tris-HCl pH 7.5 buffer containing 2.5 mM MgCl 2 . The reaction was initiated by the addition of 0.25 mg purified At4CL1. After a 5 h incubation at room temperature with agitation, 6.9 mg ATP, 2 mg CoA, and 0.25 mg purified At4CL1 enzyme were added to the reaction mixture, and the incubation was continued at room temperature for 12 h. Ammonium acetate (0.4 g) was added to the mixture to halt the reaction. Hydroxycinnamoyl-CoA esters were purified using Sep-Pak R Vac tC 18 cartridge (Waters, Milford, MA, USA) preconditioned with consecutive washes of MeOH, H 2 O, and 4% ammonium acetate solution (five column-volumes each). The reaction mixture was loaded on the preconditioned cartridge, and the column was rinsed with 4% ammonium acetate solution. The hydroxycinnamoyl-CoA esters were eluted with H 2 O. Fractions containing the hydroxycinnamoyl-CoA esters were identified by UV/Vis spectra recorded using a V-550 UV/Vis-spectrophotometer (Jasco, Tokyo, Japan), and the purified products were lyophilized for storage.

CCR Activity Assay and Determination of Kinetic Parameters
OsCCR activity was measured according to the methods of Lüderitz and Grisebach (1981). The reaction mixture consisted of 0.1 mM NADPH, 30 µM hydroxycinnamoyl-CoA, and 5 µg of purified recombinant OsCCR protein in 100 mM sodium/potassium phosphate buffer (pH 6.25) to a total volume of 500 µL. The enzyme reactions were carried out at 30 • C. The reaction was initiated by an addition of recombinant OsCCR protein, and decreases in A 366 were monitored for 10 min by a Cary 300 Bio UV/Vis-spectrophotometer (Varian, Mulgrave, Victoria, Australia). For determination of K M and V max , the substrates were used at concentrations of 5-50 µM. K M and V max were determined by extrapolation from Lineweaver-Burk plots. The enzyme assays were carried out in quadruplicate and the result represented the mean ± standard deviation.

UV and Salt Treatment
Wild-type Dongjin rice plants were grown in a greenhouse for 8 weeks after germination. UV-C treatment of rice plants were performed using the methods described by Park et al. (2013). UV-treated rice leaves were collected 1, 24, and 48 h after UV treatment.
To treat salt stress, rice seedlings were hydroponically grown on MS medium (Duchefa), and 10 day-old rice seedlings were treated with 250 mM NaCl. After 1, 3, 6, 12, and 24 h salt treatments, rice seedlings were collected for the analysis of OsCCR expression.

Analysis of OsCCR Gene Expression
The public transcriptomic analysis data of OsCCR genes in various rice developmental stages as well as under biotic [M. grisea, Xoo, and X. oryzae pv. oryzicolar (Xoc) infections] and abiotic stresses (drought, salt and cold) were downloaded from the Genevestigator plant biology database (https:// genevestigator.com/gv/doc/intro_plant.jsp) (Hruz et al., 2008). Microarray data of UV-C treated rice were obtained from the transcriptomic analysis conducted by Park et al. (2013). The genes that changed more than two-fold, with a p < 0.05, were identified as being differentially expressed genes. The normalized data was uploaded and heatmap expression patterns were generated using the Multi Experiment Viewer program (http://mev.tm4.org/#/welcome).

RNA Isolation and Quantitative Real-Time PCR Analysis
Total RNA extraction from rice samples and cDNA synthesis were accomplished using the methods described above. Quantitative real-time PCR (qRT-PCR) was performed using a Prime Q-Mastermix (GeNet Bio, Daejeon, Korea) on a Rotor-Gene Q instrument system (Qiagen). For normalization of transcript levels, rice ubiquitin5 (UBQ5) gene (Os01g22490) was used as a reference gene, which expresses stably in rice (Jain et al., 2006). The Ct method was applied to calculate expression levels (Choi et al., 2014). To ensure primer specificity, we used the data when the melting curve showed a single peak. Primers for qRT-PCR are listed in Supplementary Table 2. The primer sequences for OsCCR19, 20, 21, and UBQ5 were followed by Koshiba et al. (2013). To assess the expression of OsCCR17, 19, 20, and 21 in different tissues and stress conditions, qRT-PCR analysis was performed on triplicated biological samples, and each sample was analyzed twice for technical replicate. The results represent the mean ± standard deviation. One-way ANOVA and Tukey's HSD post-hoc test for qRT-PCR data were performed and significant differences (p < 0.05) were represented with the letters a and b. All statistical analysis was carried out using SPSS statistics.

Cloning and Heterologous Expression of OsCCRs
To elucidate biochemical functions of OsCCRs, we attempted to clone the likely functional candidates (OsCCR4,5,17,18,19,20, and 21) from wild type rice plants. The cDNAs of OsCCR5, 17, 19, 20, and 21 were successfully cloned from rice leaves. Despite many attempts, the cDNAs of OsCCR4 and 18 could not be cloned from rice, which was likely a result of very low expression levels of these genes throughout all developmental stages (Supplementary Figure 3). OsCCR1 and 26 were also cloned to examine their CCR activity. Heterologous expressions of the His-tagged OsCCR proteins were attempted under various growth temperatures and IPTG concentrations. OsCCR1, 5, 19, 20, 21, and 26 were successfully expressed as soluble protein in E. coli by 0.1 mM IPTG at an induction temperature of 25 • C. Only limited amounts of OsCCR17 soluble proteins were expressed at 18 and 25 • C, with most expressed proteins being in an insoluble form. Recombinant OsCCRs were purified with Ni 2+ affinity chromatography to apparent homogeneity (Figure 4). The purified OsCCR proteins exhibited molecular masses of 40.5-46.4 kDa on SDS-PAGE, which agreed well with their theoretical molecular masses (Figure 4 and Table 1).

CCR Activity and Kinetic Parameters of the Recombinant OsCCRs
To investigate the enzymatic properties of OsCCRs, the activities of recombinant OsCCRs were assayed using p-coumaroyl-, feruloyl-, and sinapoyl-CoAs, precursors for the H-, G-, and Sunits of lignin, respectively. OsCCR17, 19, 20, and 21 showed the reductase activity to the examined substrates (Supplementary Table 4). In these OsCCRs, the NAD(P)-binding and catalytic motifs were fully conserved (Figure 3). OsCCR1, 5, and 26 showed no detectable activity toward the hydroxycinnamoyl-CoA substrate (Supplementary Table 4). In OsCCR1 and OsCCR26, the signature NWYCY motif essential for CCR activity was replaced by NLYCC and KWYPV, respectively (Figure 3). Although OsCCR5 contained a fully conserved catalytic motif, it had no detectable activity toward the hydroxycinnamoyl-CoA substrate. This was likely caused by a polymorphism in the corresponding residue of H208, which is important in substrate binding as identified by a functional analysis of PtoCCRs (P. tomentosa CCRs) (Supplementary Figure 1) (Chao et al., 2017). The polymorphism of H208 to A, R, M, V, K, L, M, and P residues are found in OsCCRs. This likely occurred by various duplication and retention events in CCR gene family during the evolution (Barakat et al., 2011). In OsCCR5, H208 was replaced by an R residue (Supplementary Figure 1). Of the likely functional OsCCRs, OsCCR4, and 18 had well-conserved NWYCY motif (Figure 3). OsCCR4, however, featured an H208R replacement similar to that observed in OsCCR5 (Supplementary Figure 1). OsCCR18 included one amino acid insertion (RNPDDAAK) in the NADP specificity motif [R(X) 5 K]. In functional CCRs, this motif includes five amino acids between R and K residues and forms a short loop. The R and K residues form salt bridges with the phosphate in NADPH. This motif is important in distinguishing CCR from NAD(H)-dependent short-chain dehydrogenase/reductases (SDRs) (Figure 3) (Pan et al., 2014;Chao et al., 2017). Chao et al. (2017) suggested that mutations of this motif cause the loss of enzyme activity of PtoCCR8.   5,19,20,21,and 26 were expressed in E. coli as a soluble form. The recombinant proteins were purified by Ni 2+ -affinity chromatography. M, Molecular weight marker; 1, OsCCR1; 2, OsCCR5, 3, OsCCR19; 4, OsCCR20; 5, OsCCR21; 6, OsCCR26.
Therefore, we speculate that neither OsCCR4 nor 18 have any CCR activity.
To elucidate the enzymatic properties of OsCCRs, the kinetic parameters of the recombinant OsCCR19, 20, and 21 catalyzed reactions were determined toward the hydroxycinnamoyl-CoA substrate ( Table 2). Although OsCCR17 displayed enzyme activity, the amount of purified proteins from the E. coli culture was too small for kinetic analysis. The K M -values of OsCCR20 for p-coumaroyl-, feruloyl-, and sinapoyl-CoA were 24.08, 15.71, and 23.34 µM, respectively ( Table 2). The k cat /K Mvalues of OsCCR20 for feruloyl-CoA (1.41 µM −1 min −1 ) was about five-fold higher than those for p-coumaroyl-and sinapoyl-CoAs (0.32 and 0.24 µM −1 min −1 , respectively), indicating that it has a greater catalytic efficiency toward feruloyl-CoA than toward the other substrates ( Table 2). The K M -values of OsCCR21 for p-coumaroyl-, feruloyl-and sinapoyl-CoA were 16.36, 2.70, and 10.20 µM, respectively, indicating that OsCCR21 has a higher substrate affinity toward feruloyl-CoA than the other substrates ( Table 2). The k cat /K M -values also revealed a greater catalytic efficiency of OsCCR21 toward feruloyl-CoA (0.77 µM −1 min −1 ) than toward p-coumaroyl-or sinapoyl-CoAs (0.08 and 0.07 µM −1 min −1 , respectively). This result indicates that among three hydroxycinnamoyl-CoA substrates, both OsCCR20 and 21 have substrate preferences for feruloyl-CoA. The substrate preferences of both OsCCR20 and 21, with the strongest preference being toward feruloyl-CoA, is consistent with the lignin composition of rice. Gui et al. (2011) reported that rice lignin is composed of 70, 20, and 10% of G-, S-, and H-units, respectively. We also analyzed the lignin contents in stems of the Dongjin rice cultivar used in this study and found that the lignin composition was 62, 35, and 3% of G-, S-, and H-units, respectively (Supplementary Table 5  ( Table 2). OsCCR19 showed similar catalytic efficiency toward p-coumaroyl-, feruloyl-, and sinapoyl-CoAs with the k cat /K Mvalues of 0.60, 0.43, and 0.55 µM −1 min −1 , respectively ( Table 2).

In Silico and qRT-PCR Analyses of OsCCR Gene Expression
Expression of OsCCRs were investigated with the microarray data obtained from the Genevestigator database. Some OsCCRs (OsCCR3,6,7,8,16,19,20,21,22, and 26) displayed a high level of expression throughout all developmental stages including germination, seedling, tillering, stem elongation, booting, heading, flowering, milk, and dough stages (Supplementary Figure 3). Of these constitutively expressed OsCCRs, OsCCR19, 20, and 21 were found to encode biochemically active CCRs ( Table 2 and Supplementary Table 4), suggesting that these genes likely play a physiological role in rice. The OsCCR17 gene encoding enzymatically active CCR was expressed only during the early growth stages (Supplementary Figure 3). Expression of OsCCR17, 19, 20, and 21 in different developmental stages and tissues of rice were also examined by qRT-PCR analysis. Similar with the microarray data, OsCCR20 and 21 were expressed in all examined stages and tissues, and OsCCR17 was expressed primarily in rice seedling shoots and roots (Figure 5). Although both OsCCR20 and 21 were constitutively expressed in the examined rice tissues, the OsCCR20 expressions were much higher in actively lignifying organs, such as roots and stems than those of OsCCR21. qRT-PCR analysis showed that expression of OsCCR19 was very low in most examined rice tissues (Figure 5). The phylogenetic analysis revealed that OsCCR19 and 20 was closely related to functional CCRs, including ZmCCR1, PvCCR1, LpCCR, and HvCCR. These CCRs may participate in developmental lignin deposition in secondary cell walls (Larsen, 2004a,b;Escamilla-Treviño et al., 2010;Tamasloukht et al., 2011). This evidence suggests that OsCCR20 acts as a functional rice CCR and involved in developmental lignification. Expression profiles of OsCCRs were also altered by biotic and abiotic stresses. Transcriptomic analysis showed that expression of OsCCRs was significantly induced during abiotic stress conditions, such as exposure to cold (OsCCR1, 2, 6, 21, 23, and FIGURE 5 | Quantitative real-time PCR analysis of OsCCR17, 19, 20, and OsCCR21 gene expression in rice seedlings and different organs. Root and shoot samples were collected from 10-day old rice seedlings. Ten-week old rice plants yielded leaf, leaf sheath and stem samples. Panicles were obtained from 14-week old rice plants. An ubiquitin gene (OsUBQ5) was amplified using specific primers and used as an internal control. Expression levels of each OsCCR gene are presented as the relative expression compared to the OsUBQ5 mRNA level. qRT-PCR analysis was performed on the triplicated biological samples. The results represent the mean ± standard deviation. 33), drought (OsCCR3, 7, and 15), and high salinity (OsCCR3, 7, 17, and 18) (Supplementary Figure 4A). Under abiotic stress conditions, the biochemically functional genes OsCCR21 and 17 were induced by cold and salt stresses, respectively. In our previous microarray data of UV-treated rice leaves (Park et al., 2013), OsCCR1, 3, 17, 20, 21, and 23 were found to be up-regulated in response to UV-irradiation (Supplementary Figure 4B). In silico analysis of public microarray data also showed that the expression of several OsCCRs was up-regulated by biotic stresses, such as M. grisea, Xoo, and Xoc infections. The expression of OsCCR1, 2, 3, 5, 17, 18, 20, and 21 were induced by M. grisea infection. The expression of functional OsCCR17, 20, and 21 were induced by both UV-irradiation and M. grisea infection. Infection with Xoo stimulated the expression of OsCCR20 and 21, and the Xoc infection induced OsCCR21 expression (Supplementary Figure 5). Among these stressinducible OsCCRs, expression of OsCCR17 and 21 was frequently observed to be stimulated by multiple abiotic stresses. To confirm the stress-inducible expression of OsCCRs, qRT-PCR analysis was performed with UV-treated rice leaves and salt-treated rice seedlings (Figure 6). The expression level of OsCCR17 and 21 in UV-treated rice leaves increased about 70-and 10-fold compared to those of the non-treated control an hour after UV treatment, respectively ( Figure 6A). The expressions of OsCCR17 and 21 were also significantly increased by salt treatment compared to a control, which received a mock treatment (Figure 6B). The qRT-PCR analysis showed that the transcript levels of OsCCR19 and 20 were not significantly changed by both stress conditions (Figure 6). These results suggest that OsCCR17 and 21 are most likely involved in the stress responses of rice.

DISCUSSION
Although plant CCRs comprise a large gene family, only a small number of CCR genes have been reported to encode biochemically active CCRs for the biosynthesis of lignin and defense-related phenolic compounds (Lauvergeat et al., 2001;Costa et al., 2003;Escamilla-Treviño et al., 2010;Barakat et al., 2011). Xu et al. (2009) suggested that the expansion of lignin biosynthetic gene families was rapidly occurred after divergence of monocots and dicots at 120 million years ago. The large gene family of CCRs in plants was suggested to occur by various duplication and retention events during the evolution and indeed, 67% of rice CCR and CCR-like genes were located on duplicated chromosome regions (Barakat et al., 2011). A large member of CCR gene family in plants has also been supposed to because of their substrate diversity (Xu et al., 2009). All previously characterized CCRs showed similar peptide lengths ranging from 332 to 374 amino acids in A. thaliana, wheat, sorghum, switchgrass, and E. gunnii (Lacombe et al., 1997;Lauvergeat et al., 2001;Ma, 2007;Escamilla-Treviño et al., 2010;Li et al., 2016). Twenty-four OsCCRs had peptide lengths similar to known CCRs (Table 1). These OsCCRs showed high homology to wellconserved NAD(P)-binding and catalytic motifs of functional CCRs (Supplementary Figure 1 and Table 1). As a member of the mammalian 3β-HSD/plant DFR superfamily, CCRs share the NAD(P)-binding domain with DFRs. CCRs, however, have the distinct catalytic motifs with signature NWYCYGK sequence different from DFRs (Lacombe et al., 1997;Escamilla-Treviño et al., 2010;Barakat et al., 2011;Chao et al., 2017). Of OsCCRs with appropriate peptide lengths, OsCCR19 and 20 exhibited the fully conserved catalytic motif, and OsCCR4, 5, 17, 18, and 21 had the signature motif with one amino acid variation (G to A) (Figure 3). The G to A variation in the CCR catalytic motifs has been frequently found in other active CCRs, such as PvCCR2a and ZmCCR2 (Figure 3) (Pichon et al., 1998;Escamilla-Treviño et al., 2010;Li et al., 2016). Indeed, our biochemical assays confirmed that OsCCR17, 19, 20 and 21 had CCR activity to hydroxycinnamoyl-CoAs ( Table 2 and Supplementary Table 4). OsCCR21. An ubiquitin gene (OsUBQ5) was amplified using specific primers and used as an internal control. Expression level of each OsCCR gene was presented as the relative expression compared to the OsUBQ5 mRNA level. qRT-PCR analysis was performed on the triplicated biological samples. One-way ANOVA and Tukey's HSD post-hoc test for qRT-PCR data were performed and significant differences (p < 0.05) were represented with the letters a and b. The results represent the mean ± standard deviation.
OsCCR1 and 26, with two and four mismatches in this motif, respectively, had no CCR activity (Supplementary Table 4). This evidence suggests that the NWYCY(G/A)K sequence is crucial for CCR activity. An activity assay also revealed that OsCCR5 had no CCR activity, although it contained the signature catalytic motif. A recent study demonstrated that H208 in PtoCCRs is indispensable for substrate binding, and is conserved in the functional CCRs from other plant species (Supplementary Figure 1) (Chao et al., 2017). In OsCCR5, H208 was replaced by R, which likely caused the loss of its CCR activity. Like OsCCR5, OsCCR4 had the H208R replacement. In addition to the NAD(P)-binding motif, the NADP-specific R(X) 5 K motif was identified by structural analysis of M. truncatula CCR2 and petunia CCR1. This NADP-specificity motif is a key structure distinguishing CCRs from NAD(H)-dependent SDRs (Pan et al., 2014). This motif was well-conserved in the active OsCCRs (Figure 3). OsCCR18 showed one amino acid insertion in the NADP specificity motif (Figure 3). Although no activity assay was performed, for these reasons, we speculate that OsCCR4 and 18 had no CCR activity. Altogether, this evidence suggested that of the 33 OsCCRs studied here, OsCCR17, 19, 20, and 21 may encode biochemically functional CCRs in rice. In addition, a previous study reported that the enzyme activity of OsCCR1 is activated by the small GTPase OsRac1 that controls defenserelated lignin synthesis (Kawasaki et al., 2006).
Plant CCR and CCR-like genes are composed of different numbers of exons and exon-intron structures. The A. thaliana CCR gene family has been suggested to have seven patterns of exon-intron structures (Barakat et al., 2011). The OsCCRs examined in this study also exhibited eight exon-intron patterns (Figure 1). Barakat et al. (2011) divided the exon-intron structures of PoptrCCRs into three patterns (Patterns 1-3) comprised of 4, 5, and 6 exons, respectively. In Pattern 2, the fourth exon is about two times longer than other exons. The length of the fourth exon in Pattern 2 is similar to the combined lengths of the fourth and fifth exons of Pattern 3 (Barakat et al., 2011). Most functional CCRs, such as AtCCR1, EuCCR, ZmCCR1, and SbCCR1, involved in developmental lignification are grouped into Pattern 2 (Figure 1) (Lacombe et al., 1997;Lauvergeat et al., 2001;Tamasloukht et al., 2011). Consistently, OsCCR19 and 20 were composed of five exons with the exon-intron structure of Pattern 2. Although OsCCR21, which encoded biochemically active CCR, had six exons, the exon-intron structure differed from that of Pattern 3. Rather, the exon-intron structure of OsCCR21 was more similar to Pattern 2 (Pattern 2-like), with the length of the fourth exon equaling that seen in Pattern 2 (Figure 1). SbCCR2-2 has been reported to also exhibit a Pattern 2-like exon-intron structure (Figure 1) (Li et al., 2016). The biochemically active OsCCR17 was composed of four exons with an exceptionally long fourth exon (Pattern 5) (Figure 1). ZmCCR2 showed the exon-intron structure of Pattern 5 (Pichon et al., 1998). AtCCR2 exhibited the exon-intron structure of Pattern 4, being composed of four exons (Figure 1) (Lauvergeat et al., 2001). Unlike CCR genes involved in developmental lignification, which were mostly grouped into Pattern 2, stressrelated CCRs such as AtCCR2, ZmCCR2, and SbCCR2-2 exhibited diverse exon-intron patterns.
Enzymatic properties of CCRs have been elucidated from many plants, and observed to reflect the lignin compositions of the source plant species (Piquemal et al., 1998;Ma and Tian, 2005;Tamasloukht et al., 2011). Lignins of gymnosperm wood are predominantly composed of G-units (Campbell and Sederoff, 1996;Donaldson, 2001;Vanholme et al., 2010). Unlike gymnosperm lignins, most angiosperm lignins are a mixture of G-and S-units (Donaldson, 2001;Vanholme et al., 2010). The proportion of H-units is variable within plant species and even between tissues in the same plant (Campbell and Sederoff, 1996;Vanholme et al., 2010). Wheat CCRs (TaCCR1 and 2) and switchgrass PvCCR1 have substrate preference for feruloyl-CoA, a precursor for the G-unit (Ma and Tian, 2005;Ma, 2007;Escamilla-Treviño et al., 2010). Similarly, OsCCR20 and 21 showed a preference for feruloyl-CoA over other CoA esters ( Table 2). This result agrees well with rice lignin compositions, which has a high G-unit content and a relatively small portion of S-and H-units (Gui et al., 2011 and Supplementary Table 5). Different from OsCCR20, OsCCR19 showed similar catalytic efficiency toward three examined substrates. It has been known that the substrate preferences of CCRs vary between CCRs from different plant species, even in isozymes from the same species (Goffner et al., 1994;Baltas et al., 2005;Li et al., 2005;Escamilla-Treviño et al., 2010;Tamasloukht et al., 2011).
During development, lignin is deposited in the thickened secondary cell walls. In addition, its synthesis can be induced by diverse biotic and abiotic stresses (Moura et al., 2010;Miedes et al., 2014). In A. thaliana, maize and switchgrass, CCR1 genes are related to lignin biosynthesis during development and CCR2 genes are involved in stress-related processes (Pichon et al., 1998;Lauvergeat et al., 2001;Escamilla-Treviño et al., 2010;Tamasloukht et al., 2011). Phylogenetic analysis of functional CCRs has revealed that constitutive CCRs involved in developmental lignification are grouped separately from CCRs implicated in defense-related processes (Figure 2 and Supplementary Figure 2) (Escamilla-Treviño et al., 2010;Li et al., 2016;Chao et al., 2017). OsCCR19 and 20 were closely related to constitutive CCRs, such as ZmCCR1, PvCCR1, and LpCCR (Figure 2). CCRs in this group have been observed to be highly expressed in actively lignifying tissues, including stems and roots (Pichon et al., 1998;Larsen, 2004a,b;Escamilla-Treviño et al., 2010). In silico transcriptomic analysis showed that OsCCR19 and 20 are constitutively expressed throughout all developmental stages of rice (Supplementary Figure 3). Our qRT-PCR analysis also revealed strong expressions of OsCCR20 in lignifying tissues such as roots and stems (Figure 5). Unlike the microarray data, OsCCR19 was rarely expressed in most examined tissues. The kinetic analysis also showed that OsCCR20 was more enzymatically efficient toward feruloyl-CoA, a precursor of the lignin G-unit ( Table 2). These results suggest that OsCCR20 primarily participates in developmental deposition of lignins in secondary cell wall. Lignification occurs prominently in differentiating xylem tissues and interfascicular fibers in stems and roots (Lacombe et al., 1997;Goujon et al., 2003;Tamasloukht et al., 2011). Functional CCRs involved in developmental lignification have been found to localize in these tissues. In situ hybridization of the CCR antisense probe shown that the CCR transcripts are localized in the differentiating xylem tissues of poplar stems (Lacombe et al., 1997). In Leucaena leucocephala seedlings, the CCR proteins are localized in the developing xylem tissues of stems and roots (Srivastava et al., 2015). Transient expression of SbCCR-GFP in tobacco leaves indicated that CCR proteins are localized in the cytoplasm (Li et al., 2016). Kawasaki et al. (2006) also reported that OsCCR1 is localized in the cytoplasm. Analysis of N-terminal sequence of OsCCRs using the SignalP tool (http://www.cbs.dtu.dk/services/ SignalP/) showed that all OsCCRs, except OsCCR27, have no signal sequence. The expression of CCRs in other groups are induced under various stress conditions (Lauvergeat et al., 2001;Fan et al., 2006;Escamilla-Treviño et al., 2010;Li et al., 2016). For instance, ZmCCR2 expression was highly induced by water deficit in the root elongation zone of maize (Fan et al., 2006). PvCCR2 was highly induced after the rust disease infection (Escamilla-Treviño et al., 2010), and SbCCR2-2 expression was stimulated by sorghum aphid infection (Li et al., 2016). Phylogenetic analysis indicated that OCCR17 and 21 were grouped with the stressinducible CCRs (Figure 2). Although transcriptomic analysis revealed constitutive expression of OsCCR21, its expression was strongly stimulated by infections of rice pathogens (M. grisea, Xoo and Xoc) (Supplementary Figure 5). Expression of OsCCR17 was also induced by M. grisea and Xoo infections. In addition, OsCCR17 and 21 expressions were strongly induced by abiotic stresses, such as cold, high salinity, and UV-irradiation (Supplementary Figure 4). Transcriptomic analysis of UV-treated rice has revealed that a set of phenylpropanoid and monolignol pathway genes are co-expressed immediately after UV-treatment, with the response involving biosynthesis of defense-related compounds such as phytoalexins (Park et al., 2013(Park et al., , 2014Cho and Lee, 2015). Our qRT-PCR analysis also observed strong induction of OsCCR17 and 21 in response to UV and salt-treatment ( Figure 5). These results strongly suggest that OsCCR17 and 21 likely participates in defense-related lignification and synthesis of phenolic compounds.

CONCLUSION
Expression patterns and biochemical properties of the rice CCR gene family were thoroughly analyzed in the present study. OsCCR17, 19, 20, and 21 were found to have NAD(P)-binding and NADP-specific motifs as well as the CCR signature motif. The recombinant OsCCR17, 19, 20, and 21 showed enzyme activity toward hydroxycinnamoyl-CoA substrates, indicating that these OsCCRs are biochemically functional CCRs in rice. Phylogenetic analysis revealed that OsCCR19 and 20 were closely related to other plant CCRs involved in developmental lignification. In silico transcriptomic analysis and qRT-PCR consistently demonstrated that OsCCR20 were constitutively expressed throughout all developmental stages of rice, with especially high expression levels in actively lignifying tissues such as roots, stems and panicles. These results suggest that OsCCR20 are primarily involved in the developmental deposition of lignins in secondary cell walls. Meanwhile, the expressions of OsCCR17 and 21 were induced in response to biotic and abiotic stresses, such as M. grisea and Xoo infections, UVirradiation and high salinity. OsCCR17 and 21 were also grouped with stress-responsible CCRs identified from other plant species. Therefore, we suggest that OsCCR17 and 21 play a role in defense-related processes of rice under biotic and abiotic stress conditions.