Infection of Ustilaginoidea virens intercepts rice seed formation but activates grain-filling-related genes.

Rice false smut has become an increasingly serious disease in rice (Oryza sativa L.) production worldwide. The typical feature of this disease is that the fungal pathogen Ustilaginoidea virens (Uv) specifically infects rice flower and forms false smut ball, the ustiloxin-containing ball-like fungal colony, of which the size is usually several times larger than that of a mature rice seed. However, the underlying mechanisms of Uv-rice interaction are poorly understood. Here, we applied time-course microscopic and transcriptional approaches to investigate rice responses to Uv infection. The results demonstrated that the flower-opening process and expression of associated transcription factors, including ARF6 and ARF8, were inhibited in Uv-infected spikelets. The ovaries in infected spikelets were interrupted in fertilization and thus were unable to set seeds. However, a number of grain-filling-related genes, including seed storage protein genes, starch anabolism genes and endosperm-specific transcription factors (RISBZ1 and RPBF), were highly transcribed as if the ovaries were fertilized. In addition, critical defense-related genes like NPR1 and PR1 were downregulated by Uv infection. Our data imply that Uv may hijack host nutrient reservoir by activation of the grain-filling network because of growth and formation of false smut balls.

INTRODUCTION

Rice false smut is a destructive grain disease in rice production throughout the world. The typical symptom is the formation of a false smut ball that is attributable to the growth of a white fungal mass in a spikelet, protruding out from the gap between the palea and the lemma, and eventually forming a ball‐like colony, which produces numerous yellow or greenish‐black chlamydospores and sometimes is covered by sclerotia (Guo et al. 2012). False smut not only reduces grain yield but also affects grain quality. Moreover, the pathogen produces ustiloxins with antimitotic activities poisonous to both human and animals (Koiso et al. 1994; Nakamura et al. 1994). In recent years, the disease has been reported with increasing frequency in rice‐producing areas worldwide (Rush et al. 2000; Atia 2004; Tsuda et al. 2006; Brooks et al. 2009; Ladhalakshmi et al. 2012). The increasing disease incidence is experientially attributed to heavy application of chemical fertilizers and widespread planting of high‐yielding hybrid rice, and the epidemics of the disease are associated with rainy weather at rice booting stage (Wang et al. 2004; Fan et al. 2014).

The causative pathogen is an ascomycete fungus, which possesses an anamorphic state named Ustilaginoidea virens (Cooke) Tak. (Uv) and a teleomorphic state called Villosiclava virens (Tanaka et al. 2008). Chlamydospores and sclerotia of Uv are considered as the primary sources of infection, and both can produce conidia as direct inoculum to attack rice spikelets at late booting stage (Fan et al. 2010; Guo et al. 2012). It has been reported that Uv infects rice roots at the seedling stage (Ikegami 1962; Schroud and TeBeest 2005), and coleoptiles at the germination stage (Zheng et al. 2009). Most recently, we have found that under wet conditions Uv could epiphytically colonize leaf surfaces of paddy field weeds and abiotic surfaces such as Parafilm and cellophane (Fan et al. 2014). This finding implies that during the season when developing rice spikelets are not available, Uv may undergo a epiphytic stage producing a large number of conidia, which at rice booting stage could enter the rice sheath with rainwater and infect the developing spikelets (Fan et al. 2014).

A general infection process of Uv has been reported that Uv conidia land and germinate on the outer surface of rice spikelets, and the hyphae extend into the inner space of spikelets through the small gap between the lemma and the palea and invade/cover the floral organs (Ashizawa et al. 2012). Further observations reveal that the primary infection sites of Uv hyphae are the upper parts of the three stamen filaments between the ovary and the lodicules, and that the hyphae intercellularly extend along the filament base without killing the host cells. The ovary cannot be infected, and the hyphae fail to extend to the pedicels and stems of the panicles (Hu et al. 2013b; Tang et al. 2013). Transcriptome analysis of rice spikelets infected with Uv at asymptomatic stages reveals that the smut pathogen largely modulates genes involved in defense responses and gene regulation, and that a number of genes are uniquely responsive to Uv infection, such as genes specifically expressing during pollen development (Chao et al. 2014).

Interestingly, Uv hyphae not only infect flower organs of rice but also those of barley; however, no ball‐like colonies could be formed in the latter (Hu et al. 2013a). Thus, the formation of false smut ball must be a result of the specific interaction between Uv and rice flower. As false smut ball is the only visible symptom of the smut disease, deciphering the Uv‐rice interaction leading to smut ball formation is of particular importance to fully understand and effectively control the disease. In this work, microscopic and transcriptional analyses were combined to investigate rice responses to Uv infection. The data imply that Uv may hijack rice nutrient reservoir systems to successfully colonize rice floral organs and form false smut balls.

RESULTS

Observation of Uv infection process in rice spikelets

According to field observations and previous reports on the infection process of Uv, no clear symptoms could be seen on Uv‐infected rice plants until the macroscopic fungal mass emerges at approximately 15 d post‐inoculation (dpi), and not all the spikelets from a panicle could be infected by Uv (Ashizawa et al. 2012; Tang et al. 2013). In addition, the appearing time of false smut balls is variable among different Uv isolates on different rice accessions. To determine the accurate time of sampling spikelets that are discernable to be infected by Uv for transcriptional analysis, we repeated examination of the infection process of Uv towards the formation of mature false smut balls in our experimental conditions.

Developing panicles of a susceptible rice accession (i.e. Pujiang 6) were inoculated with the pathogen isolate Uv‐10 at rice booting stage. Ustilaginoidea virens hyphae were detected on the outer surface of spikelets at 1 dpi (Figure 1A). Hyphae extended rapidly and formed membrane‐like mycelia around the trichomes and at the gap between the palea and lemma at 2–9 dpi (Figure 1B–D), and were detected on the surface of the inner organs such as anthers, filaments, and lodicules at 9 dpi (Figure 1E). After its colonization on the floral organs, Uv started massive growth inside spikelets, and macroscopic white mycelia covered all floral organs at 17 dpi (Figure 1G, H). A few infected spikelets (<10%) had damaged anthers and floral organs including anthers were alive in most (>90%) infected spikelets at 17 dpi. Spikelets with damaged organs were excluded in subsequent gene expression studies. Fungal mass continued to expand and protruded out from the gap between the palea and lemma at approximately 20 dpi (Figure 1I). The Uv‐infected rice flowers never opened and the ovaries remained small (Figure 1H, I). These observations suggest that the pollination and fertilization of flowers are interrupted by Uv infection and the infected spikelets can be easily discernable at 17 dpi. Eventually, false smut balls covered with yellow and greenish‐black chlamydospores emerged at approximately 23 and 30 dpi, respectively (Figure 1J, K).

Figure 1

Infection process of

(A, B) Uv‐10 isolate was inoculated into sheaths of Pujiang 6 at late booting stage. Environmental scanning electron microscope (ESEM) and digital camera were used to monitor the entire infection process of the pathogen. Ustilaginoidea virens hyphae were evident on the outer surface of spikelets at 1 d post‐inoculation (dpi) (A) and formed membrane‐like mycelia at 2 dpi (B).

(C, D) Mycelia were observed around the gap between the palea and lemma at 9 dpi.

(E, F) At 9 dpi, the pathogen extended to the floral organs inside spikelets, such as filaments (fi), anthers (an), and lodicules (lo).

(G, H) The pathogen continued to grow and covered most of the floral organs with macroscopic white mycelia at 17 dpi.

(I) At approximately 20 dpi, the fungal mass protruded out from the gap between palea and lemma.

(J, K) Yellow and greenish‐black false smut balls were found at approximately 23 dpi (J) and 30 dpi (K), respectively.

Arrows and arrowheads indicate Uv mycelia and rice smut balls formed by the fungus, respectively. Stars represent the greenish ovaries in Uv‐infected spikelets. Scale bars = 100 µm (A, F), 200 µm (B, D, E), or 500 µm (C, G).

As demonstrated above, the infection process of Uv was very long and included at least two quite different phases. Phase 1: growth on the outer surface of rice spikelets; this phase could last for more than 1 week (under our experimental conditions). Phase 2: Uv hyphae extended into the inner space of spikelets and infected inner floral organs; this phase directly determined the formation of false smut balls. In this study, false smut balls at very young age were found at approximately 17 dpi. Thus, this time point was suitable for investigating gene expression profiling related to smut ball formation.

Expression stability analysis of rice housekeeping genes during Uv‐rice interaction

Normalization with a transcriptionally stable reference gene is important for accurate evaluation of gene expression. Housekeeping genes such as glyceraldehydes‐3‐phosphate dehydrogenase (GAPDH), Actin, and Ubiquitin (Ubi) are commonly used as reference genes. However, these genes are not always suitable for transcriptional analysis, because their expression levels could vary under different conditions and in different tissues (Lee et al. 2002; Czechowski et al. 2005). The transcription stability of a set of housekeeping genes in rice have been assessed in different tissues/organs under various developmental stages and stresses (Jain et al. 2006; Li et al. 2010), but not in samples related to Uv‐rice interaction.

To identify a gene whose expression can be mostly stable and thus suitable as an internal reference in our transcriptional analysis, we used geNorm version 3.5 to analyze the expression of six frequently used rice reference genes in eight rice spikelet samples before and after Uv infection (see Materials and Methods). As depicted by geNorm, the lowest M value represents the most stable in transcription. When considering all eight samples (Uv‐infected and uninfected samples at four developmental stages of rice spikelets) together, OsGAPDH and OsUbc were the most stable with the lowest M values, while OsTubβ was the least stable in expression (Figure 2A). When only four uninfected samples were included for analysis, OsGAPDH and OsUbc were the most stable in transcription. When four Uv‐infected samples were taken into account, OsUbi and OsGAPDH displayed the most stable expression (Figure 2B, C). Altogether, among the six tested reference genes, OsGAPDH showed the most stable expression in different developmental stages of rice spikelets before and after Uv infection. Therefore, OsGAPDH was selected as the reference gene for evaluating relative expression of rice genes involved in Uv‐rice interaction in subsequent experiments.

Figure 2

Expression stability analysis of six rice housekeeping genes by geNorm

(A) Data analysis was performed on eight samples including Pujiang 6 spikelets at four different developmental stages (B) and Uv‐infected spikelets at corresponding stages (C) (see Materials and Methods). The lower value of the average expression stability (M), the more stable the housekeeping gene is.

Suppression of putative flower‐opening genes in rice spikelets upon Uv infection

As demonstrated in Figure 1H, I, Uv‐infected rice flowers never opened. We speculated that expression of genes related to flower opening may be repressed. It has been reported that Arabidopsis auxin response factor (ARF)6, ARF8, and MYB21 play important roles in flower development and opening (Nagpal et al. 2005; Mandaokar et al. 2006). Rice genes Os02t0164900‐01, Os06t0677800‐01, and Os11t0684000‐01 were highly homologous to AtARF6, AtARF8, and AtMYB21, respectively, with identities ranging 58%–74% (Figure S1). Temporal expression profiles of the three rice homologs were examined across different stages of spikelet development before and after Uv infection. In mock‐inoculated samples, transcripts of OsARF6 (Os02t0164900‐01) accumulated along with spikelet development before pollination (i.e. 1–10 dpi), and the accumulation decreased at 17 dpi (4–6 d after pollination). A similar expression pattern was observed for OsARF8 (Os06t0677800‐01) and OsMYB21 (Os11t0684000‐01) (Figure 3). Upon Uv infection, the transcriptional levels of OsARF6 and OsARF8 were suppressed approximately 2‐fold at 10 and 17 dpi compared with the mock‐inoculated controls at corresponding stages, although upregulation was detected at 1 dpi (Figure 3A, B). OsMYB21 was repressed 2.3‐fold at 17 dpi, though no obvious difference was detected at 10 dpi (Figure 3C).

Figure 3

Transcriptional analysis of flower opening‐related genes in response to

Expression changes of flower opening‐related genes Os02t0164900‐01 OsARF6 (A), Os06t0677800‐01OsARF8 (B), and Os11t0684000‐01 OsMYB21 (C) in mock‐inoculated (Os‐CK) and Uv‐infected (Os‐Inf) rice spikelets at 1, 5, 10, and 17 d post‐inoculation (dpi).

The rice GAPDH was used as a reference gene. The C T value in mock‐inoculated spikelet sample at 1 dpi was set as a calibrator for each gene. Student's t‐test was performed to determine the significance of difference between Os‐CK and Os‐Inf at each time point (*P < 0.05, **P < 0.01). Data were means ± SD of three biological replicates. Similar results were obtained in at least two individual experiments.

Failure of ovary fertilization in rice spikelets infected with Uv

Interruption of rice flower opening could lead to non‐pollination and infertility of ovaries. As expected, the ovaries inside the false smut balls remained small and could not be stained by KI‐I2 (Figures 1I, 4A), indicating no endosperm development and starch accumulation. By contrast, ovaries from the uninfected spikelets expanded and were stained black‐blue by KI‐I2, indicating successful pollination and endosperm development (Figure 4A).

Figure 4

Examination of ovary development and expression patterns of grain‐filling‐related genes in rice spikelets upon

(A) KI‐I2 staining of ovaries from mock‐inoculated and Uv‐infected rice spikelets at 17 d post‐inoculation (dpi).

(B) Western blot analysis of seed storage protein accumulation in Uv‐infected rice, using antibodies against GluA‐2 and 11‐S seed protein. Lane 1, Uv‐infected rice spikelets at 17 dpi; lane 2, rice spikelets with false smut balls at 23 dpi; lane 3, mature rice seeds as positive control.

(C) Expression changes of OsGlutln3 (Os02t0249000‐01), OsGlutln4 (Os10t0400200‐01), OsPromln2 (Os05t0332000‐01), OsSP11S (Os03t0336100‐01), OsSSI (Os06t0160700‐01), OsSSIIa (Os06t0229800‐01), OsSSIIIa (Os08t0191433‐00), OsAGPL2 (Os01t0633100‐01), OsAGPS2b (Os08t0345800‐01), OsRISBZ1 (Os07t0182000‐01), and OsRPBF (Os02t0252400‐01) in mock‐inoculated and Uv‐infected spikelets at 17 dpi. The C T value in the ovary sample excised from Uv‐infected spikelets at 17 dpi was set as a calibrator for each gene.

(D) Time‐course expression profiles of OsGlutln1 (Os02t0248800‐01), OsGlutln2 (Os02t0453600‐01), and OsPromln1 (Os07t0206400‐01) across different stages of Uv infection. Note that no transcripts were detected in the control and infected samples at 1 and 10 dpi, and in the ovaries excised from Uv‐infected spikelets at 17 dpi. Thus, the C T value in mock‐inoculated spikelet sample at 5 dpi was set as a calibrator for each gene.

For quantitative real‐time polymerase chain reaction experiments in (C) and (D), the rice GAPDH was used as a reference gene. Data were means ± SD of three biological replicates. Student's t‐test was performed to determine the significance of difference between mock‐inoculated and Uv‐infected rice spikelets at 17 dpi (*P < 0.05, **P < 0.01). Similar results were obtained in at least two individual experiments.

To determine whether seed storage proteins (SSPs) accumulated in ovaries from Uv‐infected rice spikelets, immunoblot analysis was carried out using two antibodies against glutelin type‐A2 (Os10t0400200‐01) and 11‐S plant SSP (Os03t0336100‐01), respectively. As displayed in Figure 4B, no obvious accumulation of glutelin and 11‐S seed protein were detected in Uv‐infected rice spikelets at 17 dpi and in false smut balls at 23 dpi. Together with the findings of KI‐I2 experiment, our data indicate that ovaries in Uv‐infected spikelets are unfertilized and in turn do not accumulate grain‐filling substances such as starch and SSPs.

Activation of grain‐filling‐related genes in rice spikelets infected with Uv

Many pale‐looking infected spikelets were found at approximately 17 dpi, which is attributed to the white fungal mass inside spikelets, and were distinguishable from non‐infected green spikelets (Figure 1H). To investigate the influence of Uv infection on the transcription profile of rice spikelets, we performed RNA‐Seq analysis on pale‐looking infected spikelets at 17 dpi, with spikelets from comparable positions of mock‐inoculated panicles as controls. Illumina RNA‐Seq generated 26 and 28 million clean reads from mock‐inoculated and Uv‐infected spikelets, respectively. Differentially expressed genes (DEGs) were identified using the reads per kilobase transcriptome per million mapped reads (RPKM) data under the criteria of false discovery rate (FDR) of 0.001 or less and absolute log2‐fold change of 1 or more. As a result, 4,109 (12.3%) out of 33,417 detected rice genes were differentially expressed upon Uv infection, of which 1,590 genes were upregulated and 2,519 were downregulated (Table S1).

Surprisingly, a large number of grain‐filling‐related genes were coordinately upregulated in Uv‐infected spikelets, compared with mock‐inoculated controls. For instance, a set of genes (32) encoding rice SSPs, such as prolamin, glutelin, and globulin, were greatly induced upon Uv infection (Table 1). Critical starch metabolism genes, such as those encoding large and small subunits of ADP‐glucose pyrophosphorylase (AGPase), starch synthase, and branching enzyme, also accumulated more transcripts in Uv‐infected spikelets than in the controls. In addition, two transcription factors, OsRISBZ1 and OsRPBF which have been demonstrated to regulate starch and storage protein synthesis in rice seed (Onodera et al. 2001; Yamamoto et al. 2006; Kawakatsu et al. 2009), displayed elevated transcriptional levels in spikelets infected with Uv (Table 1). Expression data in the database of RicePLEX revealed that the above grain‐filling‐related genes were specifically or inductively expressed in rice maturing seeds but showed only detectable or much lower levels in other organs and in spikelets before pollination (Figure S2).

Table 1

Expression changes of rice grain‐filling‐related genes in response to Ustilaginoidea virens (Uv) infection

Gene IDa	Gene description	Log2‐fold change (Os‐Inf/Os‐CK)b	FDRc
Seed storage protein
Os02t0248800‐01	Similar to glutelin type‐B2 precursor	2.67	0.00E + 00
Os02t0453600‐01	Similar to glutelin	2.60	0.00E + 00
Os03t0427300‐01	Glutelin type‐A III precursor	2.54	0.00E + 00
Os02t0249000‐01	Glutelin, seed storage protein	2.49	0.00E + 00
Os02t0268100‐01	Similar to glutelin (fragment)	2.17	0.00E + 00
Os02t0268300‐00	Similar to glutelin (fragment)	2.11	0.00E + 00
Os10t0400200‐01	Glutelin type II precursor	2.06	0.00E + 00
Os01t0761800‐00	Similar to glutelin type‐A3	1.92	2.50E − 01
Os02t0242600‐01	Similar to glutelin	1.88	0.00E + 00
Os02t0249600‐01	Similar to glutelin	1.73	8.22E − 65
Os01t0762500‐00	Glutelin subunit mRNA	1.68	0.00E + 00
Os02t0456100‐01	Similar to glutelin	1.30	6.39E − 29
Os03t0188500‐01	Glutelin family protein	1.27	2.51E − 81
Os12t0472500‐01	Glutelin family protein	1.16	1.71E − 02
Os07t0219300‐01	Prolamin precursor (13 kDa prolamin)	9.79	5.20E − 02
Os05t0332000‐01	Similar to prolamin precursor	4.94	1.32E − 38
Os05t0331532‐01	Similar to prolamin	3.98	0.00E + 00
Os05t0329200‐00	Similar to prolamin	3.92	2.93E − 04
Os05t0330600‐00	Similar to prolamin	3.62	0.00E + 00
Os05t0331800‐01	Similar to prolamin	2.98	2.53E − 136
Os05t0331366‐00	Similar to prolamin	2.85	0.00E + 00
Os05t0328466‐00	Similar to prolamin	2.66	2.52E − 02
Os05t0328800‐00	Prolamin 7	2.66	1.09E − 03
Os05t0328333‐00	Similar to prolamin	2.51	2.25E − 09
Os06t0507100‐01	Similar to prolamin	2.00	2.56E − 43
Os07t0206400‐01	13 kDa prolamin precursor	1.84	0.00E + 00
Os07t0220000‐01	Similar to prolamin	1.76	0.00E + 00
Os07t0219400‐01	Prolamin precursor	1.74	0.00E + 00
Os07t0206500‐00	13 kDa prolamin precursor	1.53	0.00E + 00
Os12t0269200‐01	Similar to prolamin precursor	1.13	0.00E + 00
Os03t0336100‐01	11‐S plant seed storage protein family protein	1.39	1.21E − 208
Os05t0499100‐01	26 kDa globulin (alpha‐globulin)	1.14	0.00E + 00
Starch metabolism d
Os01t0633100‐01	AGPase large subunit	1.03	0.00E + 00
Os08t0345800‐01	AGPase small subunit	2.32	5.55E − 210
Os06t0160700‐01	Starch synthase I	−0.26	4.02E − 02
Os06t0229800‐01	Starch synthase IIa	1.73	6.39E − 86
Os08t0191433‐00	Starch synthase IIIa	1.46	8.23E − 173
Os06t0726400‐01	Branching enzyme I	1.17	3.24E − 75
Os02t0528200‐01	Branching enzyme IIb	1.10	0.00E + 00
Os08t0520900‐00	Isoamylase1	0.91	2.78E − 163
Os06t0133000‐01	Granule‐bound starch synthase I	1.88	0.00E + 00
Transcription factor
Os07t0182000‐01	bZIP transcription factor	1.29	6.47E − 60
Os02t0252400‐01	DOF zinc finger transcription factor	1.29	1.90E − 82

Genes selected for quantitative real‐time polymerase chain reaction analysis are underlined and presented in Figure 4.

Gene accession number for rice gene in the Rice Annotation Project Database (http://rapdb.dna.affrc.go.jp/).

The log2‐fold change of gene expression in Uv‐infected rice spikelets versus mock‐inoculated rice samples from RNA‐Seq experiment.

False discovery rate (FDR) as calculated according to Benjamini and Yekutieli (2001).

These starch metabolism genes have been implicated in grain filling during rice endosperm development (Zhou et al. 2013).

Using quantitative real‐time polymerase chain reaction (qPCR), we validated the expression patterns of four SSPs, five starch metabolism genes, and two transcription factors (OsRISBZ1 and OsRPBF) in the mock‐inoculated and Uv‐infected spikelets at 17 dpi. In addition, their expression in the ovaries excised from the infected spikelets was also examined. As shown in Figure 4, these genes were expressed at comparable or higher levels in Uv‐infected (non‐pollinated) spikelets than in the mock‐inoculated controls (4–6 d after pollination), except for OsSSI and OsRPBF. On the contrary, much lower expression levels were detected in the ovaries excised from infected spikelets; the amount of reduction varied 446–65,536‐fold for the SSPs, 2.6–82‐fold for the starch metabolism genes, and 28–111‐fold for the two transcription factors. Both RNA‐Seq and qPCR data demonstrated that grain‐filling‐related genes were transcriptionally activated in infected non‐pollinated spikelets at 17 dpi. Furthermore, time‐course transcriptional analysis confirmed that three selected SSPs, including OsGlutln1 (Os02t0248800‐01), OsGlutln2 (Os02t0453600‐01), and OsProlmn1 (Os06t0507100‐01) displayed high expression levels in both the mock‐inoculated and Uv‐infected spikelets at 17 dpi, with more transcripts in the latter (Figure 4D). By contrast, their transcripts failed to be detected at 1 and 10 dpi (data not shown) and were barely detected at 5 dpi (Figure 4D). Additionally, OsGlutln1, OsGlutln2, and OsProlmn1 were not expressed in the ovaries excised from the infected spikelets at 17 dpi (data not shown). These data imply that Uv infection may mimic pollination and fertilization to trigger expression of grain‐filling genes in unfertilized rice spikelets, and this activation does not occur in the ovaries.

Expression changes of rice defense/stress‐related genes in response to Uv infection

Many defense‐related genes have been reported to be critical in plant innate immunity. RNA‐Seq data demonstrated that a set of defense genes were coordinately downregulated in rice spikelets upon Uv infection (Table 2). Two homologs of the NPR1 gene, which is a key regulator of salicylic acid‐mediated disease resistance in Arabidopsis (Cao et al. 1997), displayed a 2–4‐fold reduction in Uv‐infected rice spikelets compared with those in the mock‐inoculated controls (Table 2). Five PR1 homologs were differentially expressed, three of which were remarkably downregulated (Table 2). The expression of three CNGC homologs involved in the plant pathogen response signaling cascade (Ma and Berkowitz 2011) were also suppressed (Table 2). MIN7 and RIN4 are key regulators in effector‐triggered immunity in Arabidopsis (Mackey et al. 2002, 2003; Nomura et al. 2006). Their homologs in rice were downregulated over 4‐fold upon Uv infection (Table 2). COI1 and JAZ proteins are two critical factors with opposite roles in the jasmonic acid (JA) signal transduction pathway (Wasternack and Hause 2013). In Uv‐infected rice spikelets, the expression levels of COI1 and JAZ homologs were decreased and increased, respectively, suggesting repression of JA signaling (Table 2). In addition, two rice DEGs associated with induced systemic resistance showed reduced transcriptional levels (Table 2).

Table 2

Expression changes of rice defense/stress‐related genes in response to Ustilaginoidea virens (Uv) infection

Gene IDa	Homolog name	Gene description	Log2‐fold change (Os‐Inf/Os‐CK)b	FDRc
Plant–pathogen interaction
Os01t0194300‐01	NPR1	Similar to NPR1	−1.96 (−0.73 ± 0.11)	1.04E‐79
Os11t0141900‐00	NPR1	Ankyrin domain containing protein	−1.58	4.12E‐04
Os05t0595000‐01	PR1	Allergen V5/Tpx‐1 related family protein	−7.78 (−5.64 ± 0.38)	0.00E + 00
Os12t0633400‐01	PR1	Similar to Pathogenesis‐related protein PR‐1 precursor	−6.11 (−5.07 ± 0.54)	1.02E‐121
Os01t0382400‐01	PR1	Similar to Pathogenesis‐related protein PRB1‐2 precursor	−1.88 (−0.82 ± 0.28)	5.05E‐157
Os10t0191300‐01	PR1	Similar to PR‐1a pathogenesis related protein (Hv‐1a) precursor	1.43	4.75E‐05
Os01t0382000‐01	PR1	Similar to Pathogenesis‐related protein PRB1‐2 precursor	2.77 (2.45 ± 0.37)	1.36E‐32
Os03t0646300‐01	CNGC	Similar to Cyclic nucleotide‐gated channel A (Fragment)	−10.30	2.88E‐04
Os06t0250600‐01	CNGC	Ankyrin domain containing protein	−1.78 (−1.02 ± 0.10)	1.96E‐71
Os01t0788500‐01	CNGC	Disease resistance protein domain containing protein	−1.06	7.02E‐04
Os03t0265500‐01	COI1	Similar to Coronatine‐insensitive 1	−2.30 (−2.54 ± 0.14)	4.55E‐08
Os05t0449500‐01	COI1	Similar to Coronatine‐insensitive 1	−1.48	1.08E‐79
Os01t0853400‐01	COI1	Coronatine‐insensitive 1	−1.07	1.62E‐28
Os10t0391400‐01	JAZ	Tify domain containing protein	3.47 (3.95 ± 0.15)	2.98E‐15
Os04t0117300‐01	MIN7	Similar to H0207B04.10 protein	−2.52 (−2.08 ± 0.26)	8.54E‐05
Os07t0564750‐00	MIN7	Conserved hypothetical protein	−2.10	3.82E‐05
Os04t0620600‐01	RIN4	Similar to NOI protein	−4.47 (−3.68 ± 0.28)	1.25E‐61
Os06t0636100‐00	RIN4	Hypothetical conserved gene	−3.66	8.67E‐14
Induced systemic resistance
Os03t0243900‐01	TLP	Similar to Thaumatin‐like protein	−1.83	1.90E‐27
Os05t0586200‐01	GH3‐5/JAR1	GH3 auxin‐responsive promoter family protein	−1.29	1.13E‐46
Symbiont immune response
Os05t0156500‐01	DnaJ	Similar to Apobec‐1 binding protein 2	1.07 (1.18 ± 0.30)	1.23E‐22
Os08t0278900‐01	SDF2‐like	Stromal‐derived factor‐2, MIR domain containing protein	1.08	3.09E‐07
Os08t0440500‐01	SDF2‐like	Stromal‐derived factor‐2, MIR domain containing protein	1.29 (1.40 ± 0.31)	1.36E‐26
Os05t0156401‐00	Unknown	Hypothetical gene	1.17	1.40E‐08
Dehydration‐responsive‐element‐binding protein
Os01t0885900‐00	TINY	Similar to transcriptional factor TINY	12.41	1.32E − 09
Os02t0676800‐01	DREB1E	Similar to dehydration‐responsive‐element‐binding protein 1E	6.25 (8.03 ± 0.43)	1.50E − 131
Os06t0127100‐01	DREB1C	Dehydration‐responsive‐element‐binding protein 1C	5.43	9.07E − 70
Os08t0545500‐00	DREB1J	Similar to dehydration‐responsive‐element‐binding protein 1J	5.38	1.14E − 12
Os09t0522100‐00	DREB1H	Similar to dehydration‐responsive‐element‐binding protein 1H	3.41	7.23E − 23
Os09t0522000‐01	DREB1B	Similar to dehydration‐responsive‐element‐binding protein 1B	2.58	2.66E − 30
Os08t0521600‐01	DREB2C	Similar to dehydration‐responsive‐element‐binding protein 2C	2.29	5.61E − 112
Os02t0797100‐00	DREB2B	Similar to dehydration‐responsive‐element‐binding protein 2B	−1.31	4.16E − 08

Genes selected for quantitative real‐time polymerase chain reaction (PCR) analysis are underlined.

Gene accession number for rice gene in the Rice Annotation Project Database (http://rapdb.dna.affrc.go.jp/).

The log2‐fold change of gene expression in Uv‐infected rice spikelets versus mock‐inoculated rice samples from RNA‐Seq experiment. Data in parentheses represent expression changes (mean ± SD) determined by quantitative real‐time PCR.

False discovery rate (FDR) as calculated according to Benjamini and Yekutieli (2001).

By contrast, four DEGs such as genes encoding SDF2‐like and DnaJ proteins were coordinately upregulated (Table 2). These genes were assigned to the biological process of host innate immune response to pathogen‐associated molecular pattern (PAMP) of symbiont and activation of tolerogenic host immune pathway could promote host tolerance (Detournay et al. 2012), which implies that the Uv‐rice interaction may have some symbiotic feature at a certain infection stage. The dehydration responsive element binding protein (DREB) family functions in plant response to stresses; for example, DREB2C confers tolerance to oxidative stress (Hwang et al. 2012). Upon Uv infection, eight DREB family genes were differentially expressed, seven of which displayed obvious upregulations. For instance, the transcriptional levels of TINY, DREB1E, DREB1C, and DREB1J were highly induced by more than 40‐fold in Uv‐infected rice spikelets, which may lead to elevated stress tolerance in infected plants (Table 2).

The expression patterns for 13 of the above genes were further validated by qPCR. Similar trends and fold changes of gene expression were observed (Table 2). Moreover, the expression patterns of PR1, CNGC, and DREB1E homologs were examined in rice spikelets across different developmental stages before and after Uv infection. Time‐course transcriptional analysis showed that OsPR1#051, OsPR1#121, and OsPR1#012 accumulated transcripts in mock‐inoculated samples along with spikelet development before pollination (i.e. at 1, 5, and 10 dpi) and peaked at 10 dpi, then the accumulation began to decline after pollination (i.e., at 17 dpi), indicating their flower‐preferential expression patterns (Figure 5A–C). Note that the pollination of uninfected spikelets occurred at approximately 12 dpi. Upon Uv infection, their transcriptional levels were downregulated at 5, 10, and 17 dpi (Figure 5A–C). Remarkably, the expression of OsPR1#121 and OsPR1#051 were reduced by 33‐ and 50‐fold, respectively, compared with their corresponding controls at 17 dpi (Figure 5A, B). On the contrary, OsPR1#011 displayed upregulation during the spikelet development and further induction upon Uv infection at 10 and 17 dpi (Figure 5F). Distinct expression patterns between OsPR1#011 and the other three PR1s (i.e., OsPR1#051, OsPR1#121, and OsPR1#012) suggest that they play different roles in spikelet development and Uv‐rice interaction. The CNGC homolog Os03t0646300‐01 showed a similar expression profile with OsPR1#121 (Figure 5D). Compared with mock‐inoculated controls, the transcriptional level of the DREB1E homolog Os02t0676800‐01 was 4‐, 2‐, and 261‐fold higher at 1, 5, and 17 dpi, respectively, although approximately 2‐fold lower at 10 dpi (Figure 5E).

Figure 5

Time‐course transcriptional analysis of rice defense/stress‐related genes in response to

quantitative real‐time polymerase chain reaction was carried out to examine the expression changes of Os12t0633400‐01 OsPR1#121 (A), Os05t0595000‐01 OsPR1#051 (B), Os01t0382400‐01 OsPR1#012 (C), Os03t0646300‐01 OsCNGC1 (D), Os02t0676800‐01 OsDREB1E (E), and Os01t0382000‐01 OsPR1#011 (F) in mock‐inoculated (Os‐CK) and Uv‐infected (Os‐Inf) rice spikelets across different infection stages.

The rice GAPDH was used as a reference gene. The C T value in a mock‐inoculated spikelet sample at 1 d post‐inoculation (dpi) was set as a calibrator for each tested gene, except for OsPR1#011 where the C T value in Uv‐infected sample at 1 dpi was set as a calibrator. Data were means ± SD of three biological replicates. Student's t‐test was performed to determine the significance of difference between Os‐CK and Os‐Inf at each time point (*P < 0.05, **P < 0.01). Similar results were obtained in at least two individual experiments. ND, not detected.

DISCUSSION

Flower is an important sink organ with abundant nutrients allocated to it and can produce many kinds of nutrient‐rich secretions, thus serving as an excellent habitat for microorganisms. Meanwhile, risks exist for pathogens specializing in flower infection, due to the ephemeral nature of floral organs. For instance, flower production is highly seasonal in most plant species, and floral organs are prone to abscise upon completion of pollination or under stresses (Sun et al. 2004; Rogers 2006). These challenges require that flower‐infecting pathogens should have phenologies tightly coupled with those of their hosts, and have the ability to maintain flowers living long enough until they propagate offspring. Nevertheless, flower‐infecting fungi widely exist in nature and cause many economically important plant diseases, such as Fusarium head blight caused by Gibberella zeae in wheat and Ergot disease caused by Claviceps purpurea in rye (Ngugi and Scherm 2006). Anatomical, physiological, and/or molecular evidence has been demonstrated for fungi infecting ovaries (Ngugi and Scherm 2004; Tudzynski and Scheffer 2004). As an example, C. purpurea infects rye ovary by first penetrating the cuticle of stigmatic hairs, then growing directly towards the rachilla at the base of the ovary probably mimicking pollen tubes, and tapping the vascular tissue for acquiring nutrients (Tudzynski and Scheffer 2004). Ustilaginoidea virens displays a different organ specificity, that is, infecting stamen filaments exclusively (Tang et al. 2013). In the present work, our data suggest that Uv may colonize rice flowers and hijack the rice nutrient reservoir through modulating host defense responses, flower opening, and grain‐filling network.

Successful pathogens should have the abilities to evade or subvert host defense, to colonize host tissues/organs, and to propagate within the host and exit the host eventually. In this work, we found that Uv could suppress expression of rice defense‐related genes homologous to NPR1, PR1, CNGC, and AtMIN7 in Arabidopsis (Table 2; Figure 5). The Arabidopsis NPR1 protein is an important regulatory component in plant immunity, controlling the onset of systemic acquired resistance (SAR). Loss‐of‐function mutation in NPR1 leads to increased susceptibility to pathogens and little expression of PR genes (Cao et al. 1997). Expression of three rice NPR1 homologs have been identified to be induced by the rice bacterial blight pathogen Xanthomonas oryzae pv. oryzae and the blast fungus Magnaporthe grisea, and one (Os01t0194300‐01) of the homologs is proved to be the ortholog of Arabidopsis NPR1 (Yuan et al. 2007). PR1 genes are widely used as marker genes for SAR. There are 12 PR1 members in the rice genome, all of which are upregulated in compatible and/or incompatible rice‐blast fungus interactions (Mitsuhara et al. 2008). By contrast, the expression of the rice NPR1 ortholog (Os01t0194300‐01) and three PR1 genes preferentially expressed in rice flowers (Mitsuhara et al. 2008) were suppressed in Uv‐infected spikelets (Table 2; Figure 5), which suggests that the SAR pathway in rice spikelets may be suppressed upon Uv infection.

In plant defense response signaling cascades, Ca2+ elevation in cytosol is a critical early event. Plant CNGCs are channels for Ca2+ conductance across plasma membrane, and thus are important in activating downstream components of defense signaling (Ma and Berkowitz 2011). Mutation in the Arabidopsis CNGC2 leads to failure of hypersensitive response to Pseudomonas syringae DC3000 and impaired cytosolic Ca2+ elevation; meanwhile, PAMP‐induced nitric oxide (NO) is also impaired (Clough et al. 2000; Ali et al. 2007; Ma et al. 2009). In this work, we detected that the expression of three CNGC homologs was suppressed in rice infected with Uv, implying that Ca2+‐mediated defense responses may be repressed upon Uv infection. Arabidopsis AtMIN7, an interactor of effector protein HopM1 from P. syringae, plays an important role in cell wall‐associated defense. Knock‐out of AtMIN7 causes impairment of cell wall‐associated defense, characterized by reduction of polarized callose deposition in response to the P. syringae mutant delta CEL, a strain that is defective in suppressing callose deposition (Nomura et al. 2006). Here, we showed that during Uv‐rice interaction, two rice homologs of AtMIN7 were downregulated by more than 4‐fold (Table 2). In addition, the expression of a gene encoding putative callose synthase 1 catalytic subunit (Os02t0832400‐02) was also suppressed (Table S1). Suppression of extracellular cell wall‐associated host defense should be an effective strategy for Uv to subvert host surveillance system. This speculation is consistent with a recent report that Uv is an extracellular pathogen whose hyphae extend intercellularly in rice stamen filaments (Tang et al. 2013). Taken together, Uv may suppress multiple targets in rice immunity, presumably through secreted proteins and effectors (Zhang et al. 2014).

Rice stamen filament elongates rapidly during anthesis and senesces shortly after flowering. As a stamen filament‐infecting fungus (Hu et al. 2013a; Tang et al. 2013), Uv should have the ability to complete successful colonization before flowering and/or to inhibit flowering. In this work, we found that the floral organs were covered by a mass of Uv mycelia before flowering time and the infected flowers never opened, indicating that Uv may inhibit rice flowering to maintain the colonization sites (Figure 1H). Arabidopsis miRNA miR167 and its targets AtARF6 and AtARF8 regulate stamen and gynoecium development in immature flowers. Double‐mutant plants of AtARF6 and AtARF8 and overexpressors of miR167a produce non‐opening flowers in Arabidopsis (Nagpal et al. 2005; Wu et al. 2006). The Arabidopsis mutant myb21/24 displays defects in floral development and flower opening (Mandaokar et al. 2006). Knocking‐down of EOBII, a Petunia × hybrida homolog of MYB21, prevents flowers entering anthesis (Colquhoun et al. 2011). In the present study, both qPCR and RNA‐Seq showed that rice OsARF6, OsARF8, and OsMYB21 were suppressed upon Uv infection (Figure 3; Table S1), suggesting that the pathogen may interfere with flower opening via modulating these transcription factors.

The process of rice grain filling involves a pronounced transportation event of nutrients, mainly sugars and amino acids; and starch and SSPs highly accumulate in the developing endosperm. Rice SSPs include glutelins, prolamins, globulins, and albumins. In mature seed, prolamins and glutelins account for 20–30% and 60–80% of total protein, respectively (Zhou et al. 2013). These proteins are specifically accumulated in maturing seed and mostly encoded by multigene families (Nie et al. 2013), which are highly expressed only after fertilization (Figure S2). Promoters of rice SSPs could drive endosperm‐specific expression of the β‐glucuronidase reporter gene, demonstrating their seed‐specific expression (Wu et al. 1998). Several genes involved in starch anabolism have been identified to be crucial for endosperm development and grain filling in rice (Zhou et al. 2013), and these genes also display seed‐specific/preferential expression patterns (Figure S2). Two transcription factors, the rice prolamin box binding factor RPBF and the basic leucine zipper factor RISBZ1, have been demonstrated to regulate the expression of rice SSPs and starch metabolism genes. Knocking‐down of the two regulators leads to reduced levels of SSPs and starch in rice seed (Onodera et al. 2001; Yamamoto et al. 2006; Kawakatsu et al. 2009). In addition, rice osbzip58 (osrisbz1) null mutants have decreased levels of total starch and amylose (Wang et al. 2013). A set of rice SSPs, starch synthesis genes, and the two seed‐specific transcription factors (i.e., RPBF and RISBZ1) were highly activated in Uv‐infected spikelets in which the ovaries were unfertilized (Table 1; Figure 4). It is indicated that Uv infection may activate the grain‐filling system in rice spikelets.

Acquiring nutrients from the host is critical for pathogen growth and propagation. Ustilaginoidea virens infects rice reproductive tissues and forms a ball‐shape colony, of which the size is usually several times larger than mature rice seed. It is reasonable that the smut pathogen needs to acquire large amounts of nutrients from rice. Our data suggest that Uv may have the ability to mimic fertilization in rice flowers, so that ample nutrients can be allocated to the colonization sites and hijacked by Uv for growth and smut ball formation. It should be noted that SSPs such as glutelin and 11‐S seed protein were not detected in the infected spikelets (Figure 4B), although their encoding genes were highly transcribed (Table 1). Whether or where starch and SSPs are ever synthesized in Uv‐infected spikelets needs to be further investigated. Also, it would be interesting to identify pathogen components that can upregulate the expression of grain‐filling‐related genes in rice, and dissect how these genes affect the formation of false smut balls.

MATERIALS AND METHODS

Plant material, pathogen isolate, and artificial inoculation

Pathogen isolate Uv‐10 was obtained via amerosporous purification from rice (Oryza sativa L.) false smut balls in Sichuan Province, China. Discs of Uv‐10 mycelia growing on potato dextrose agar were inoculated into potato sucrose broth (PSB), and cultured for 7 d at 28 °C and 120 rounds/min. A mixture of conidia and mycelia was either collected for RNA extraction, or blended and adjusted with fresh PSB to an appropriate concentration (i.e. the density of conidia reached 1 × 106/mL) as the inoculum.

The injection inoculation method was applied as described with minor modifications (Tang et al. 2013). Plants of a susceptible rice cultivar (Pujiang 6) were grown in plastic pots under natural conditions. At a late booting stage (∼1 week before heading), inoculum of Uv‐10 was injected into rice sheaths with a sterile syringe until the inoculum suspension dripped. At least 30 sheaths from six pots were inoculated with the pathogen suspension, and mock inoculation was carried out using PSB on another set of plants. Inoculated plants were kept at 25 °C with 85%–95% relative humidity (RH) for 5 d, and then transferred to 28 °C with 80% RH. Spikelet samples were collected at an interval of 1–3 dpi for subsequent experiments.

Microscopic observation

To monitor the infection process of Uv‐10 in rice spikelets, samples collected at multiple time points after inoculation were examined directly under an environmental scanning electron microscopy (ESEM) (FEI Quanta 450; FEI, Hillsboro, OR, USA) with the low vacuum model (70 Pa). Macroscopic images of the spikelets and false smut balls were acquired with a Canon EOS Rebel T2i digital camera (Canon, Tokyo, Japan).

Quantitative real‐time polymerase chain reaction

Using TRIzol reagent (Life Technologies, Carlsbad, CA, USA), total RNAs were isolated from mock‐inoculated and Uv‐inoculated rice spikelet samples at 1, 5, 10, and 17 dpi. The QuantiTect Reverse Transcription Kit (QIAGEN, Hilden, Germany) was used for reverse transcription of total RNAs following the manufacturer's instruction. Contaminated DNA was eliminated by the Kit‐supplied gDNA Wipeout Buffer. Quantitative real‐time polymerase chain reaction was carried out using a QuantiTect SYBR Green PCR Kit (Qiagen). Gene expression analysis was conducted by the comparative C T method 2−ΔΔCT according to Fan et al. (2012). Three biological replicates were used, each of which was extracted from mixed spikelets collected from at least six panicles; and at least two individual experiments were carried out to confirm the results. Primer sequences are included in Table S2.

geNorm version 3.5 software was employed to select a stable rice reference gene for qPCR analysis according to the manufacturer's instruction. In total, eight rice spikelet samples were used, including spikelets from mock‐inoculated and Uv‐inoculated panicles at 1, 5, 10, and 17 dpi, representing different stages of spikelet development, seed maturation, and Uv infection. Note that pollination occurred on mock‐inoculated panicles at approximately 12 dpi. Six housekeeping genes were included: GAPDH, Actin1, Ubi, beta‐tubulin (Tubβ), eukaryotic elongation factor 1‐alpha (eEF1α), and ubiquitin‐conjugating enzyme E2 (Ubc). Primers sequences for the last three genes were adopted from Jain et al. (2006).

KI‐I 2 staining

To determine the fertilization status of ovaries, KI‐I2 staining was performed on mock‐inoculated and Uv‐infected spikelets at 17 dpi. Briefly, ovaries were excised from control spikelets and early false smut balls, incubated in 75% ethanol for 1–3 h until the green color faded, rinsed with distilled water, and then stained in KI‐I2 solution for 0.5 h. After staining, the ovaries were rinsed with 75% ethanol and incubated in 100% ethanol for 0.5 h before examined.

Protein extraction and western blotting

Total protein was extracted as described (Kawakatsu et al. 2008). Briefly, spikelets and false smut balls were ground into fine powder with a mortar and pestle, and extracted with buffer containing 50 mmol/L Tris‐HCl (pH 6.8), 4% sodium dodecylsulfate (SDS), 8 mol/L urea, 5% 2‐mercaptoethanol, and 20% glycerol for 2 h at room temperature. Total protein was obtained from the supernatant by centrifugation at 20,817 g for 5 min. Proteins were separated by SDS polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes, and subjected to immune‐blot analysis using antibodies against glutelin type‐A2 (Os10t0400200‐01) and 11‐S plant SSP (Os03t0336100‐01), which were produced in rabbits with the synthetic peptides ANAYRISREEAQR and DERWEEKKKAAKQRK, respectively, and were kindly provided by Dr Guozhen Liu (Beijing Genomics Institute). An ECL kit (GE Healthcare, Pittsburgh, PA, USA) was used to detect signals.

RNA‐Seq analysis

Total RNAs were isolated from mock‐inoculated and Uv‐inoculated rice spikelet samples collected at 17 dpi. RNA quantity and quality were determined using a NanoDrop ND‐1000 spectrophotometer (NanoDrop Technologies, USA) and Agilent 2100 Bioanalyzer (NanoDrop Technologies, Wilmington, DE, USA). Three aliquots of rice RNA, each of which was extracted from mixed spikelets collected from at least six panicles, were equally pooled for deep transcriptome sequencing. cDNA synthesis, library preparation, and Illumina HiSeq 2000 sequencing were conducted at Beijing Genomics Institute (BGI, Shenzhen, China). For each rice sample, approximately 2 Gb of reads were generated.

After removing adaptor sequences and filtering low‐quality sequences, clean reads from rice samples were mapped to the rice reference genome (http://rapdb.dna.affrc.go.jp/download/irgsp1.html) using SOAPaligner/SOAP2 (Li et al. 2009). The RPKM method was used to calculate the normalized expression data of each rice and Uv transcript (Mortazavi et al. 2008). Differentially expressed genes were identified according to Audic and Claverie (1997), under the criteria of the absolute log2‐fold change of 1 or more and FDR of 0.001 or less (Benjamini and Yekutieli 2001).

Additional supporting information can be found in the online version of this article at the publisher's web‐site.

Figure S1. Sequence alignment of ARF6, ARF8 and MYB21 homologs from rice and Arabidopsis

Using DNAMAN v5.2.2 software with default parameters, amino acid sequences of OsARF6 (Os02t0164900‐01), OsARF8 (Os06t0677800‐01) and OsMYB21 (Os11t0684000‐01) from rice were aligned with their homologs from Arabidopsis (AT1G30330, AT5G37020, AT3G27810, respectively). Identical amino acids are shaded in black. The ARF6 (A), ARF8 (B) and MYB21 (C) homologs share high identities of 62%, 58% and 74%, respectively. Conserved motifs are underlined based on BLASTP analysis in NCBI (http://blast.st‐va.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastHome). B3_DNA, plant specific B3 DNA binding domain; Auxin_resp, a conserved region of auxin‐ responsive transcription factors; AUX_IAA, a conserved domain of the AUX/IAA family; SANT, ‘SWI3, ADA2, N‐CoR and TFIIIB’ DNA‐binding domain.

Click here for additional data file.

Figure S2. Expression profiling of rice grain‐filling‐related genes in various tissues/organs and at different stages of reproductive development

Global expression data for the grain‐filling‐related genes listed in Table 1 were retrieved from Affymetrix microarray experiments in the RicePLEX database (http://www.plexdb.org/modules/PD_browse/experiment_browser.php). Sample information was as following: Mature leaf; Young leaf; up to 0.5 mm, shoot apical meristem and rachis meristem (SAM); 0‐3 cm, floral transition and floral organ development (P1); 3‐10 cm, meiotic stage (P2 and P3); 10‐15 cm, young microspore stage (P4); 15‐22 cm, vacuolated pollen stage (P5); 22‐30 cm, mature pollen stage (P6); 0‐2 days post pollination (dap), early globular embryo (S1); 3‐4 dap, middle and late globular embryo (S2); 5‐10 dap, embryo morphogenesis (S3); 11‐20 dap, embryo maturation (S4); 21‐29 dap, dormancy and desiccation tolerance (S5).

Click here for additional data file.

Table S1. The complete list of differentially expressed genes from rice in response to Uv infection. (provided as an Excel file)

Click here for additional data file.

Table S2. Primers used in this study

Click here for additional data file.