Rice variety Huangxiuzan was grown in the experimental station of Plant Protection Institute, Guangdong Academy of Agricultural Science, Guangdong Province, China. The experiment field was protected by a metal net, and there was an automatic water-spraying system underneath the net and above the field, this system sprayed mistlike water for 30 seconds per hour from 10 o’clock in the morning to 4 o’clock in the afternoon to increase humidity in the atmosphere during the infection period from artificial inoculation of rice plants with U. virens to emergence of the smut balls on rice panicles. PSB (potato and sucrose broth) medium was used to culture U. virens strain for inoculation. U. virens was cultured at 28˚C for 10 days, the whole culture was stirred three times in a stirrer, each for 15 sec, to break down hyphae to small pieces, the resultant mixture of hyphae and spores were used to inoculate rice panicles. We used a syringe with needle to inoculate rice plants at booting stage, each plant was injected with 2 - 4 ml of inoculum until the liquid flooded out from the top of leaf sheath. Smut balls appeared in rice panicles about 20 days after inoculation.

U. virens was cultured in PSB medium at 28˚C for 8 days, the culture was filtered with three-layer of lens papers to exclude hyphae, U. virens conidia were collected and adjusted to the concentration of 1 × 10⁷ conidia per milliliter. Panicles at booting stage were peeled out and rinsed in the filter-treated conidia solution for 10 minutes. After the treatment, the panicles were cultured in 1/2 MS medium for 5 days, spikelets were used to observe conidium germination and examine the early infection process of U. virens.

U. virens-infected spikelets were opened carefully under a dissection microscope, the processes of hyphae infection of the anthers and stigma were observed under stereomicroscope (Zeiss Stemi SV11, Carl Zeiss，Germany) and corresponding images were taken. Anthers and stigma from in vitro U. virens-infected spikelets were pretreated and mounted to a metal stub (10 mm in diameter). The specimens were sputter-coated with gold particles (approximately 30 nm in thickness) and examined with a scanning electron microscope (JSM-6360 LV, JEOL Ltd, Japan).

Smut balls collected from the field were first fixed in FAA (Formaldehyde acetic acid ethanol) solution for 24 hours. After several times of dehydration and rehydration, the specimens were embedded in epoxy resin and cut with a glass knife into semi-thin sections. The semi-thin sections were carried out aniline blue staining according to the procedure described by Hood and Shew [23] with minor modification. The stained sections were placed on glass slide and examined by means of epifluorescence microscopy (Zeiss Axioplan 2, Carl Zeiss, Germany), and images were record by Zeiss AxioCam.

To detect ROS production on the rice flower organs during U. virens infection, spikelets from infected panicles and uninfected control panicles were stained with 3, 3’- Diaminobenzidine tetrahydrochloride solution (1 mg of DAB-HCl per 1mL of water) for 3 h at room temperature. The samples were destained in ethanol:chloroform (4:1) and then kept in the dark in 60% glycerol until examination by light microscopy.

Callose deposition was monitored by potassium hydroxide-aniline blue staining. Briefly, the materials were vacuum infiltrated in 1 M potassium hydroxide for 2 h at room temperature, followed by staining with 0.05% (w/v) aniline blue in water. The stained materials were placed in glass slides and examined by means of epifluorescence microscopy (Leica DMRBE, Switzerland).

We used freshly cultured U. virens to inoculate rice panicles at the late booting stage, the concentration of inoculum was 10⁷ conidia per milliliter (Figure 1(a)). At day 5 after infection, hyphae emerged and enclosed rice spikelets (Figure 1(b)). Usually, flower-infecting fungi develop hyphae on the exterior surface of florets and grow toward the openings. Rice spikelets covered by U. virens hyphae were opened up to examine the infection process inside the spikelet (Figure 1(c)). Hyphae entered florets through the aperture of the apex of glumes, firstly occupied the upper space of the flower organs and formed a three-dimensional network (Figure 1(d)), then grew downward and intertwined the stamens and the bifurcate feathery stigma (Figure 1(e)). We observed large amounts of hyphae on the surface of anthers and plumose stigma (Figures 1(f), (g)). Growing downward, hyphae filled the whole inside space of the floret and enclosed the ovary (Figure 1(h)). This observation is in agreement with the findings of Ashizawa et al. [24] in that they found U. virens invaded rice spikelets through the small gap of the spikelet apex.

Rice panicles were collected from U. virens infected plants at day 3 after infection, the panicles were immersed in a solution of 3, 3’-diaminobenzidine tetrahydrochloride (DAB-HCl) for 3 hours at room temperature, after destained in ethanol:chloroform solution, rice flower organs were observed under microscope to exam-

ine ROS accumulation. DAB forms a brown polymerization product in the presence of ROS. Flower organs from U. virens-infected panicles showed enhanced brown oxidized DAB precipitates (Figures 2(a)-(c)), in contrast, the flower organs from the control plants did not exhibit brown color (Figures 2(d)-(f)). We cut the ovaries from both the U. virens-infected and the control plants and examined the DAB oxidation under microscope, the ovaries from U. virens-infected plants displayed more DAB oxidation products than those from uninfected plants (Figures 2(a), (d)), especially, the base of the infected ovary showed the highest level of oxidation, indicating more ROS produced at this site. Probably, the base of the ovary is the place where the fungus obtains water and nutrients from host plant.

The stigma from U. virens-infected plants showed enhanced accumulation of ROS as evidenced by the brown DAB oxidation, whereas the stigma of the control plants did not show brown color (Figures 2(b), (e)). In addi-

tion, the anthers from U. virens-infected plants had higher levels of DAB oxidation than those of the uninfected plants (Figures 2(c), (f)), indicating that U. virens infection induced high levels of ROS accumulation at the places it invaded.

Callose plays important roles in response to multiple biotic and abiotic stresses, and can be induced upon the recognition of pathogen infection. We collected rice panicles at day 5 after U. virens inoculation to clarify callose deposition. Flower organs from U. virens infected plants were first recorded under bright light of microscope (Figures 2(g), (j) and (m)), and then stained with aniline blue to observe callose deposition. The infected rice flower including anthers, stigma and ovary, showed bright aniline blue staining (Figure 2(h)), while the control flower did not exhibit staining, thus, was barely seen under microscope (Figure 2(i)). Many bright aniline blue staining spots were scattered on the bifurcate plumose stigma of the infected rice panicles (Figure 2(k)), in contrast, no bright spots were observed in the stigma from the uninfected plants (Figure 2(l)). To investigate whether callose deposited in the ovary, we cut ovaries longitudinally and stained with aniline blue, the section of the ovary from infected plants showed a bright circle at the position of integuments surrounding the ovule (Figure 2(n)), on the contrary, the ovary from the control plants did not show bright circle (Figure 2(o)), indicating that U. virens infection can induce callose deposition in the ovary at the place around the ovule.

Booting stage is the critical time for U. virens infection. In transverse sections of anthers from U. virens infected plants, the pollen mother cells (PMC), both the cell wall and the contents, were slightly stained by aniline blue (Figure 2(p)), in contrast, the PMC from the control plants were heavily stained by aniline blue, indicating large amounts of callose accumulated (Figures 2(q), (s)). Importantly, the PMC of the infected anthers were in a state of plasmolysis (Figure 2(r)), which was probably caused by the metabolites secreted from U. virens.

U. virens secretes compounds including ustiloxins into media, and ustiloxins are toxic to humans and livestock. To verify the effects of secreted compounds on rice pollen development, we used the supernatant of old-cultured U. virens to treat rice pollen grains to investigate the germination rate. Rice pollen grains were first treated with germination solution (15% sucrose, 20 mg/L borate and 40 mg/L CaCl₂), after treatment for 10 min, about 75% of pollen grains germinated (Figure 3(a)), whereas pollen grain germination was completely inhibited by the treatment of the supernatant (Figure 3(b)). To explore whether or not the supernatant is able to inhibit rice pollen tube elongation, pollen grains were first treated for 6 min in the germination solution, the germinated pollen

grains were selected and the pollen tube length was recorded (Figure 3(c)). At the same time, some of the selected pollen grains were kept in the germination solution, and others were treated with the supernatant mixture (5 volumes of the supernatant and 1 volume of germination solution), the length of pollen tube was recorded after incubation in the solutions for 9 min. The pollen tube continued to elongate when pollen grains were kept in the germination solution (Figure 3(d)), whereas pollen tube elongation was completely inhibited if the germinated pollens were kept in the supernatant mixture (Figure 3(e)). These data demonstrated that the supernatant of U. virens culture can inhibit rice pollen germination and pollen tube elongation, this inhibition is probably due to the ustiloxins secreted by U. virens, although we could not rule out the possibility that other compounds existed in the supernatant also have inhibition ability.

To investigate the routes of U. virens infection and analyze the infection effects on rice flower development, we performed semi-thin sectioning of smut balls. The crosssection showed the stigma with two branches were enclosed by hyphae (Figure 4(a)). High magnification showed hyphae could not breach the epidermal cells of the stigma (Figure 4(b)), even the stigma were densely covered by hyphse. In transverse section of smut balls, mycelia compactly packed the stigma and deformed the papillae. Hyphae entered the intercellular spaces of stigma via the thin-walled cells of papillae, they only breached at the sites of the plumose part of the stigma, but could not reach the vascular bundles (Figure 4(c)).

In transverse section of the anthers embedded in the mycelia of smut balls, four locules of an anther were still in shape and visible, but the contents were decomposed and replaced with hyphae (Figure 4(d)). High magnification showed the trace of PMC, but the contents were decomposed and replaced with hyphae (Figure 4(e)). In transverse section of anther filaments embedded in smut balls, the filaments were filled by hyphae but the boundaries between the filaments and the enclosed mycelia were still clear since the hyphae inside the filaments had different staining extent from those surrounding the filaments (Figure 4(f)). Anther filaments consist of epidermis, transmitting and vascular-like tissues. The vascular-like tissue is not a real vascular bundle in comparison to that of the pedicel, thus, it was vulnerable to U. virens infection (Figure 4(g)). In high magnification, we were able to verify that the invaded hyphae were well-organized inside the filaments, they decomposed the contents of the cells but did not disturb the original cell organization (Figure 4(h)). These findings are consistent with the results of Tang et al. [11].

Rice ovary contains an ovule consisting of the nucellus and an embryo sac containing an egg cell and a central cell. The nucellus is covered by inner and outer integuments, each typically with two layers of cells (Figure 5(a)), and these two layers of the integuments will be absorbed after fertilization.

We did serial cross-sectioning of the smut balls at different positions, in transverse section of mid-region of the ovary persisted in the smut ball and wrapped by mycelia, hyphae were observed in the ovary. The intruded hyphae digested the integuments and grew around the nucellus at the position of integuments without infecting the ovular vascular cells (Figure 5(b)). Although residing the inside of ovary, hyphae did not spread out, they invaded neither the nucellar epidermis nor the inner epidermis of the ovary (Figure 5(b)). On the other hand, the hyphae that wrapped the outside of ovary could not invade the epidermis because we did not observe hyphae existed in the epidermal cells of the ovary, and in high magnification the boundaries between hyphae and rice flower tissues were obviously clear (Figure 5(c)). This result is contradictory to the findings of Tang et al. [11], it is probably caused by different cross-sectioning positions of the ovary in the smut ball.

Rice spikelet is supported by the pedicel that connects to the main stem of the panicle. Transverse section of the pedicel from the control rice plant showed normal vascular bundle (phloem and xylem) and parenchyma (Figure 5(d)). We reasoned that if U. virens caused a systemic disease in rice, at least we would observe traces of fungal infection in the pedicel of the spikelet which was

infected by U. virens and formed smut ball, in this case, some changes to the cells of the pedicel should occur, in other words, hyphae would cause the xylem or phloem tissues of the infected rachilla to become at least partially dysfunctional, and the traces of hyphae should be observed. We carried out cross-sectioning of the pedicel of the spikelet with smut ball at the position just right below the rudimentary glumes. In transverse section, the phloem, xylem and the parenchyma cells were clearly observed and well organized, we did not observe any trace of hyphae in either the vascular tissues or the thin-walled parenchyma cells (Figure 5(e)). Furthermore, there were no differences in cell organization and the state of vascular bundle in the pedicels from the infected and uninfected spikelets. These results indicated that U. virens was not able to go through the pedicel from the infected spikelets to the adjacent spikelets, thus, rice false smut is not a systemic disease.