Abstract
Structural studies showed that tumours induced by Southern rice black-streaked dwarf virus (SRBSDV; genus Fijivirus, family Reoviridae) were highly organized, modified phloem, composed of sclerenchyma, vessels, hyperplastic phloem parenchyma and sieve elements (SEs). Only parenchyma and SEs were invaded by the virus. There was a special region that consisted exclusively of SEs without the usual companion cells and a new flexible type of intercellular gateway was observed on all SE–SE interfaces in this region. These flexible gateways significantly increased the intercellular contacts and thus enhanced potential symplastic transport in the tumour. Flexible gateways were structurally similar to compressed plasmodesmata but were able to accommodate complete SRBSDV virions (~80nm diameter). Virions were also found in sieve-pore gateways, providing strong evidence for the movement of a virus with large virions within phloem tissue and suggesting that the unusual neovascularization of plant virus-induced tumours facilitated virus spread. A working model for the spread of tumour-inducing reoviruses in plants is presented.
Introduction
Reoviruses have 10–12 segments of genomic double-stranded RNA (dsRNA) and constitute a large family of viruses with members that naturally infect fungi, plants, and animals (Attoui et al., 2012). In humans, reovirus infection is frequent, but most cases are mild or subclinical; however, most plant reoviruses induce cell hyperplasia and seriously perturb normal development of their host plants, as first reported for wound tumour phytoreovirus (WTV), which is serologically related to human reoviruses (Black et al., 1946; Streissle and Maramorosch, 1963). It is interesting that these plant reoviruses multiply in their insect vectors but appear to be innocuous in them, similar to their counterparts in humans (Attoui et al., 2012).
Southern rice black-streaked dwarf virus (SRBSDV) is a destructive viral pathogen of rice (Oryza sativa L.) and other grain crops in China and Vietnam (Zhang et al., 2008; Hoang et al., 2011). It is a novel putative member of the genus Fijivirus, one of three plant-infecting genera in the family Reoviridae (Attoui et al., 2012). Similar to other plant reoviruses, the virus has a large icosahedral virion with a double shell approximately 75–85nm in diameter, containing multiple segments of genomic dsRNA and structural proteins, and it induces cell hyperplasia and characteristic cytopathology, including viroplasms, and crystalline and tubular structures within the cytoplasm of infected cells (Zhang et al., 2008; Wang et al., 2010; Hoang et al., 2011; Attoui et al., 2012; Li et al., 2013). The viral proteins P7-1 and P5-1/P6/P9-1 of SRBSDV are involved in the formation of tubular structures and viroplasms, respectively (Liu et al., 2011; Li et al., 2013). Typically, all known plant-infecting fijiviruses are limited to phloem tissue within host plants and induce hypertrophy of phloem leading to vein swellings, galls, or tumours (Attoui et al., 2012). SRBSDV induces obvious tumours on the stems of rice plants (Wu et al., 2013), a model monocot and staple food crop plant. Thus the virus–host interaction system is particularly interesting as a model for understanding the unique properties of reoviruses and the fundamental principles of virus-induced tumourigenesis in plants.
The local and systemic movement of plant viruses within their plant hosts involves passage across an extracellular matrix (ECM) and cell wall that is much more rigid than the ECM of animal cells. This movement is achieved by using or modifying the various natural connections between plant cells. Plants have unique and sophisticated cell wall-spanning membrane channels, plasmodesmata, that mediate the intercellular trafficking of signalling molecules or macromolecules (Lucas et al., 2009). Plasmodesmata consist of a desmotubule, a central axial membranous element composed of endoplasmic reticulum and cytoskeleton and used for symplastic connection, and a cytoplasmic sleeve, a space with soluble-phase continuity located between the plasma membrane and desmotubule and used for transportation (Blackman and Overall, 1998; Lucas et al., 2009; White and Barton, 2011). The size exclusion limit (SEL) of plasmodesmata varies and can be regulated by protein interactions or callose (β-1,3-glucan) turnover (Zavaliev et al., 2011). However, the maximum SEL is less than that required to transport the trimer green fluorescent protein (GFP) complex (Lucas et al., 2009). Only in extremis (within vasculature), are plasmodesmata structurally and functionally specialized into phloem sieve pores or pore–plasmodesma units (PPUs) with larger openings on the sieve element (SE) or companion cell interfaces, respectively (Sjolund, 1997; Oparka and Turgeon, 1999; van Bel et al., 2002; van Bel, 2003; Xie and Hong, 2011). Many plant viruses transport their viral ribonucleoprotein complexes or virions between plant cells by forming virion-containing tubules or by increasing the SEL of plasmodesmata with viral movement proteins (Oparka and Turgeon, 1999; Ritzenthaler and Hofmann, 2007; Niehl and Heinlein, 2011). Within their insect vectors, plant reoviruses induce virion-containing tubules composed of a viral nonstructural protein to traffic between insect cells (Chen et al., 2012). However, the plant reovirus non-tubular structural proteins, such as the P6 of Rice dwarf virus (RDV) and Rice ragged stunt virus (RRSV), complement intercellular movement of movement-defective Potato virus X (PVX) or Tobacco mosaic virus (TMV) in Nicotiana benthamiana (Li et al., 2004; Wu et al., 2010), suggesting that the mode of reoviral movement in plant hosts involves different proteins to those used to construct a conduit for intercellular movement in their insect (vector) hosts (Liu et al., 2011; Chen et al., 2012). The mechanism of reovirus movement within plants remains largely unknown but their virions are much too large to pass through plasmodesmata.
We now report ultrastructural and cytobiochemical studies of SRBSDV-induced tumours that provide some insight into plant virus-induced tumours and the mode of reovirus spread within plants.
Materials and methods
Virus source and antibodies
SRBSDV-infected rice plants were collected from Zhejiang province, China, identified by RT-PCR and sequence analysis (Zhang et al., 2008), and maintained in a greenhouse. Antibodies against the viral capsid protein P10 of SRBSDV were prepared as previously described (Li et al., 2013). Cytochemical labelling studies used antibodies against actin (Agrisera, Sweden), callose (Biosupplies, Australia) and β-1,3-glucanase (Agrisera, Sweden), and 5nm IgG–gold conjugates (Sigma, St Louis, MO) as a secondary antibody. An enzyme–gold probe against cellulase was made and kindly gifted by Prof. D.W. Hu (Xu et al., 2000).
Tissue preparation
Fresh tumours of infected rice stem were cut into 1×3mm pieces for fixation and dehydration as previously described (Li et al., 2013). For structural analysis, samples were fixed in 2.5% glutaraldehyde overnight and then in 1% osmium for 2h, embedded in epoxy Spurr resin. The resin was polymerized by heating at 70°C for 24h. For immuno-gold labeling, samples were fixed in a mixture of 4% paraformaldehyde and 1% glutaraldehyde for 2h and embedded in Lowicryl ‘K4M’ resin. The resin was polymerized by ultraviolet irradiation at –20°C for 72h and at room temperature for 48h.
Negative staining of virions
Crude sap from tumours was placed on a formvar-covered copper grid for 1min, fixed with 2.5% glutaraldehyde for 30 s, washed in water and then negatively stained using 2% phosphotungstic acid (PTA, pH 6.7) for 1min. After drying for 15min, virions were observed under transmission electron microscopy (TEM).
High-pressure freezing and freeze substitution
Fresh tumours of infected rice were cut into 1mm slices, immersed in the cryoprotectant 1-hexadecene, placed in a Leica HPM-100 high-pressure freezer for cryo-immobilizing and stored in liquid nitrogen. The frozen samples were transferred to a Leica EMAFS freeze-substitution apparatus for substitution with osmium-acetone (1% osmium). Substitution was done for 8h at each of –90°C, –60°C, and –30°C followed by pure acetone at 0°C for 1h. Samples were then permeated with Epon 812 resin using a gradient of 30% for 3h, 50% for 5h, 70% for 8–12h, and 100% for 3h. The resin was polymerized by heating at 37°C for 12h, 45°C for 12h, and 60°C for 24h.
Immuno-gold labelling
Ultra-thin sections (70nm) were cut with glass knives and placed on formvar-covered nickel grids. Immuno-gold labelling of protein was carried out as described previously (Li et al., 2013). Pre-immune serum was used as a negative control. Immuno-gold labelling of the cell wall-associated polysaccharides was done on sections of osmium fixed samples. Protocols and controls were the same as for protein immuno-gold labelling. The grids were then stained with uranyl acetate for 15min and lead citrate for 15min. Observations were made using an H-7650 transmission electron microscope (Hitachi, Ibaraki, Japan) at 80kV and photographed with a Gatan 830 CCD camera.
Quantitative study of SE organization
Measurements were made from three tissue types (virus-free stem phloem, SRBSDV-infected non-tumour phloem, and tumour phloem) to estimate the proportions of phloem cells that were SE cells and the proportions of SE cell perimeters that adjoined other SE cells. 30 sections per tissue type were examined under TEM at ×700; total phloem cell numbers were counted and their cell wall lengths were measured for total length and the length of the SE–SE interface. A total of 7588 phloem cells were counted and cell wall lengths of 1836 SE cells were measured. Measurements were made using the software of the pathology image digital processing system ‘JD801’ (JEDA, Nanjing, China). Summaries and statistical analysis were done using Microsoft® Excel.
Quantitative study of gateway frequency
The frequency of different gateways on cell wall interfaces was analysed as number per cell wall length, similarly to previous studies of plasmodesmata frequency (Evert et al., 1996; Botha and Cross, 1997). Gateways were identified as flexible gateways, pores, or plasmodesmata by their exclusive characteristics of a central module, a large opening, and desmotubule, respectively. Gateways with complex structures (especially branched plasmodesmata) were counted as more than one depending on the number of branches. Sections were placed on 50 mesh copper grids and photographs were taken at two different magnifications, ×1000 for measuring cell wall length and ×30 000 for distinguishing the types of gateway. The regions photographed were chosen randomly on the SE–SE interfaces of virus-free phloem, SRBSDV-infected non-tumour phloem, and of tumours. As a control the plasmodesmata frequency in leaf mesophyll cells of virus-free and SRBSDV-infected samples was also counted. Gateway numbers from ten sections were treated as a sample and at least 30 samples of each tissue type were taken. A total of 7501 gateways were counted including 3434 SE–SE wall gateways and 4067 mesophyll plasmodesmata. Length measurements and statistical analyses were done as described above.
Electron tomography
For ET reconstruction, thick sections (about 200nm) were cut and placed on carbon-covered copper grids. A random deposit of colloidal gold particles was applied to the grids to be fiducial markers for facilitating tomography alignments. Data were collected from a tilting specimen stage (single tilt) over an angular range of ±60° at 2° intervals in a JEM-2010 electron microscope (JEOL, Tokyo, Japan) under ×12 000 magnification at 200kV accelerating voltage. Eight series of images were recorded by a 2.7k × 4k bottom-mounted Gatan 832 CCD camera (Gatan, CA, USA) with an image step size of 1.09nm per pixel. The defocus value was within a range of 5 to 10 μm. Alignment, reconstruction, and dimension measurements were processed by the ‘IMOD’ program (Kremer et al., 1996) following the software manual to generate tomograms as the 3D structural model. These structural models were then manually segmented, the surface rendered by command ‘3dmod’ (Kremer et al., 1996), and visualized using the ‘Chimera’ program (Pettersen et al., 2004).
Results
Cell and tissue arrangements of SRBSDV-induced tumours
The most typical symptoms in SRBSDV-infected rice plants are white waxy tumours of various lengths along the veins of culms (Fig. 1a). In sections, these tumours were found to be enlarged versions of the vascular bundles normally present in the epidermis (Fig. 1b). The relative positions of the tissues in the two types of bundle are similar: sclerenchyma surrounds the bundle and the transporting xylem in the adaxial part characteristically has two large pitted vessels and one small annular vessel. Phloem forms the abaxial part of the vascular bundle and is greatly enlarged by hyperplasia in the tumours.
Morphology and anatomy of SRBSDV-induced tumours. (a) Waxy tumours protruding along the veins of an infected rice stem. (b) Section through tumour showing that the swelling was induced by hyperplasia of the vascular bundles located in the epidermis (E-VB). The relative positions of sclerenchyma and pitted vessels were identical in both types of vascular bundle. Enlargements of the boxed areas are shown in the two right panels. (c) Cell types within the tumour as observed by TEM. PP, phloem parenchyma; SE, sieve element; V, vessel. (d, e) Plastids (P) and virus-like particles (arrow) observed in SEs of virus-infected tumour phloem. (f) SE cells recognized by several prominent features: sieve plate (SP); SE-specific plastid (P); and SE-specific endoplasmic reticulum (SER) with stacked and anastomosing morphology. Note the arrangement of SEs without adjacent nucleate companion cells. (g–i) Longitudinal (g) and transverse (h, i) sections demonstrating the interlaced arrangement of SEs and companion cells (CCs) in virus-free rice phloem (g, h) and in the non-tumour phloem of SRBSDV-infected rice stems (i). Bars: (b) 150 μm; (c, f–i) 5 μm; (d) 1 μm; (e) 200nm.
Ultrathin sections of hyperplasia regions within tumours always contained three cell types (Fig. 1c): SE (about 28%), phloem parenchyma (about 49%), and vessels (about 23%). SEs had no nucleus or vacuole but had stacked specific endoplasmic reticulum attached to the cell wall, the porous sieve plate and SE-specific plastids (Fig. 1f). In SEs of SRBSDV-induced tumours, plastids and double layered virus-like particles were also readily observed (Fig. 1d–e). Phloem parenchyma was characterized by its nucleus, cytoplasm, and vacuole. Vessels had discontinuous cell wall thickening and no cellular content.
The staggered pattern of SE and companion cells was always observed in both longitudinal (Fig. 1g) and transverse sections of normal phloem (Fig. 1h), and in the non-tumour regions of SRBSDV-infected rice stems (Fig. 1i). Each SE was connected directly with at least one companion cell and there were similar numbers of the two cell types (data not shown), in full agreement with the ontogenetic concept of the complex, namely that the two cells are derived from the unequal division of a ‘phloem mother cell’ (Oparka and Turgeon, 1999; Evert, 2007). By contrast, this interlaced arrangement was absent from the SRBSDV-induced tumour phloem and the SEs were aggregated into a special region. (Fig. 1c, f). Although nucleate parenchyma cells were observed in tumour sections, these cells were always located outside a region of exclusive SEs and could not be regarded as companion cells.
Quantification of cell numbers and their types of interface confirmed the anatomical evidence. In the tumour phloem, there were more cells, a larger proportion of SEs, and more frequent SE–SE interfaces than in the normal phloem of virus-free stems or in the phloem from virus-infected but non-tumour stems (Table 1). When the 30 samples were split into two groups based on tumour size, there was a slightly larger proportion of SEs (29.4% ± 1.65%) in the larger tumours than in the smaller ones (26.4% ± 0.98%), suggesting that the proportion of SEs increased as the tumours grew.
Patterns of SEs in different types of rice phloem
| Virus-free stem phloem | SRBSDV-infected non- tumour phloem | SRBSDV-infected tumour phloem | |
|---|---|---|---|
| Total cell number in phloem | 56.6 (±6.42)*,a | 59.1 (±6.29)a | 137.2 (±29.62)b |
| Number of SEs in phloem | 11.1 (±2.16)a | 11.6 (±2.09)a | 38.5 (±10.86)b |
| Percentage of SEs in phloem | 19.5% (±1.84%)a | 19.5% (±1.56%)a | 27.7% (±2.02%)b |
| SE–SE interface length per SE perimeter | 43.7% (±1.10%)a | 44.1% (±1.16%)a | 77.3% (±2.35%)b |
| Virus-free stem phloem | SRBSDV-infected non- tumour phloem | SRBSDV-infected tumour phloem | |
|---|---|---|---|
| Total cell number in phloem | 56.6 (±6.42)*,a | 59.1 (±6.29)a | 137.2 (±29.62)b |
| Number of SEs in phloem | 11.1 (±2.16)a | 11.6 (±2.09)a | 38.5 (±10.86)b |
| Percentage of SEs in phloem | 19.5% (±1.84%)a | 19.5% (±1.56%)a | 27.7% (±2.02%)b |
| SE–SE interface length per SE perimeter | 43.7% (±1.10%)a | 44.1% (±1.16%)a | 77.3% (±2.35%)b |
* Values are means (±SD) taken from measurements of all cells in 30 replicate sections. a,b Within rows, values followed by the same letter do not differ significantly in a Student’s t-test (P ≤ 0.01).
Patterns of SEs in different types of rice phloem
| Virus-free stem phloem | SRBSDV-infected non- tumour phloem | SRBSDV-infected tumour phloem | |
|---|---|---|---|
| Total cell number in phloem | 56.6 (±6.42)*,a | 59.1 (±6.29)a | 137.2 (±29.62)b |
| Number of SEs in phloem | 11.1 (±2.16)a | 11.6 (±2.09)a | 38.5 (±10.86)b |
| Percentage of SEs in phloem | 19.5% (±1.84%)a | 19.5% (±1.56%)a | 27.7% (±2.02%)b |
| SE–SE interface length per SE perimeter | 43.7% (±1.10%)a | 44.1% (±1.16%)a | 77.3% (±2.35%)b |
| Virus-free stem phloem | SRBSDV-infected non- tumour phloem | SRBSDV-infected tumour phloem | |
|---|---|---|---|
| Total cell number in phloem | 56.6 (±6.42)*,a | 59.1 (±6.29)a | 137.2 (±29.62)b |
| Number of SEs in phloem | 11.1 (±2.16)a | 11.6 (±2.09)a | 38.5 (±10.86)b |
| Percentage of SEs in phloem | 19.5% (±1.84%)a | 19.5% (±1.56%)a | 27.7% (±2.02%)b |
| SE–SE interface length per SE perimeter | 43.7% (±1.10%)a | 44.1% (±1.16%)a | 77.3% (±2.35%)b |
* Values are means (±SD) taken from measurements of all cells in 30 replicate sections. a,b Within rows, values followed by the same letter do not differ significantly in a Student’s t-test (P ≤ 0.01).
Gateways on the SE–SE interfaces of tumours are associated with virion movement
Two kinds of gateway were observed by TEM on the SE–SE interfaces of the special SE region in virus-induced tumours. The typical sieve-plate pores could be distinguished by their large openings and their arrangement on a sieve plate separated by condensed osmiophilic material. There was no fibrous cell wall material of moderate electron density on these plates. A second type of gateway had a flexible membrane tube 20nm in diameter crossing the cell wall (black arrow in Fig. 3a), surrounded by osmiophobic material. We designate this second type of gateway as a ‘flexible gateway’. Surprisingly, both kinds of gateway were found to be associated with the large virions of SRBSDV (see below).
Presence of SRBSDV virions within sieve-plate pores
Under TEM, spherical particles with a clear double-layer structure (80nm in diameter) were observed within sieve-plate pores of tumour SEs in both transverse (Fig. 2a) and longitudinal sections (Fig. 2b). The particles were composed of an electron-dense core (50nm in diameter) and an outer layer, and were shown to be SRBSDV virions by specific immuno-gold labelling with antibodies against the coat protein P10 of SRBSDV (Fig. 2c). Electron tomography reconstruction provided further details of the sieve plate (Fig. 2d) and of virion association with the pores (Fig. 2e–g). A cellular membrane (arrow in Fig. 2e–h) was attached to the margin of the pore rather than packaging the virions closely (Fig. 2h). In 3D view, the pore, with size measured as 200–400nm (Fig. 2h), appeared to be large enough for the double-layered virions to pass with freedom. There was no obvious difference in size between the pores in healthy phloem (average 281.3nm, n = 57) and SRBSDV-induced tumours (average 273.2nm, n = 52).
Distribution of SRBSDV virions in the pore gateways of sieve plates. (a,b) Transverse and longitudinal sections of virion distribution in the sieve-plate pore gateway. (c) Immuno-gold labelling of SRBSDV-infected rice phloem using antiserum against RBSDV P10. (d) Electron tomography reconstruction demonstrating the structure of a sieve plate by slicing at a thickness of 132nm. (e–g) Electron tomography reconstruction demonstrating the structural details of an open pore. Slices of the 3D tomogram observed at thicknesses of 70, 140, and 182nm. Note cell membrane (arrow) attached to the inner wall of the pore. (h) Electron tomography reconstruction demonstrating that the membrane (arrow) was not attached to the virion. Bars: (a–c) 500nm; (d–h) 100nm.
Presence of SRBSDV virions within the flexible gateway
Virus-like particles were also associated with the flexible gateway (Fig. 3a, white arrow). This type of gateway is clearly different to the sieve-plate pore and appears to be an adjustable channel. In the absence of virions, the flexible gateway appeared to be a tightly folded tube but, unlike plasmodesmata, it was able to accommodate the 80nm virus-like particle apparently without the tubular structure that occurs in the Cowpea mosaic virus (CPMV)-modified plasmodesmata gateway (van Lent et al., 1990). All parts of the gateway were filled with osmiophobic cell wall material, leaving only the exact space for accommodating a virion and suggesting that the gateway is very elastic (Fig. 4a–b).
Distribution of SRBSDV virions in the flexible gateway. (a) On the SE–SE interface of SRBSDV-infected rice phloem, double-layer spherical virus-like particles (VLP, white arrow) were present in the flexible gateway (black arrow) within the osmiophobic cell wall material. (b) The presence of VLPs in the flexible gateway in samples prepared by cryo-immobilization and freeze substitution. (c) Morphology of the multi-layer SRBSDV virion demonstrated by negative staining. (d) Section of double-layer SRBSDV virions in vacuole. (e) Section of single-layer SRBSDV virions in a viroplasm. (f) Immuno-gold labelling of VLPs in a flexible gateway using antiserum against RBSDV P10. (g–m) Electron tomography reconstruction revealing VLPs embedded in the flexible gateway. The 2D projection is shown in (g). Slices of the 3D tomogram at thicknesses of 85, 97, 104, and 119nm from the XY dimension are shown in (h–k), which reveal particle embedding. Embedding of the virion within the cell wall from the YZ and XZ dimensions is shown in (l–m). (n–o) Virions embedded in the cell wall. Note membrane-like structure (arrowhead) packaged around the particles. CW, cell wall; P, SE-specific plastid. Bars, 100nm.
Locations of the flexible and pore gateways. (a, b) Flexible gateway without virion on the non-sieve plate wall of the SE–SE interface (a) and on sieve plate (b). (c–d) Flexible gateway (black arrow) with virion (white arrowhead) on sieve plate. (e–f) Virion in large open pores at the non-sieve plate wall of lateral SE–SE interface. SE, sieve element; SP, sieve plate. Bars, 100nm.
Negatively stained SRBSDV virions appear multilayered (Fig. 3c). In TEM of ultra-thin sections of plant tumours, immature SRBSDV virions appeared as osmiophilic core particles 50nm in diameter, distributed in the viroplasm of phloem parenchyma cells (Fig. 3e), which are amorphous electron-dense cytoplasmic inclusions and are thought to be structures for viral genomic replication and particle assembly (Li et al., 2013). Mature SRBSDV double-layered virions (Fig. 3d) were usually observed in the cytoplasm or vacuole and were similar in size and appearance to the structures observed in flexible gateways. Immuno-gold labelling using antibodies against SRBSDV P10 confirmed that structures in the flexible gateway were indeed SRBSDV virions (Fig. 3f).
To explore the association of virions with the gateway and exclude the possibility of misinterpretation from conventional TEM (McEwen and Marko, 2001), a 3D electron tomography reconstruction was implemented. Digital slicing of the 3D structural model showed that virions were embedded within the flexible gateway and that there was no tubule-like structure present. From the XY dimension, this embedding was clearly demonstrated by the varied diameter of the particle at different digital slices (Fig. 3h–k). From the other two dimensions (YZ and XZ), the double-layer virions were clearly seen embedded within the cell wall (Fig. 3l–m). Virions in the gateway were always packaged by a membrane-like structure (Fig. 3j and n–o; arrowheads) which occasionally contained more than one virion (Fig. 3o). The membrane-like structure was attached to the inner wall of the gateway.
To eliminate any possibility of misinterpretation induced by the chemical fixation used in TEM sample preparation (Radford et al., 1998), a sample was prepared for examination by high-pressure freezing. The virion-flexible gateway complex was also observed on the SE–SE wall interface of these samples (Fig. 3b), and there were no obvious structural differences between this sample and the chemically fixed one.
Location and frequency of the flexible and pore gateways
Although the flexible gateways were principally distributed on the non-sieve plate SE–SE interface wall (Fig. 4a), some were also observed on the sieve plate itself (Fig. 4b), and virions could be observed within them (Fig. 4c–d, white arrowhead). Pore gateways were not confined to the sieve plate but were also present on the other SE–SE wall interfaces and were open and large enough to accommodate virions (Fig. 4e–f).
The length of SE–SE interface was greater in tumour phloem than in virus-free stem phloem (1.231mm/0.998mm) and the total numbers of gateways (flexible and pore) was also larger (1853/1581). Of the 1853 gateways in infected tumour phloem, 89 (4.8%) were observed to have SRBSDV virions inside. There were no differences in plasmodesmata gateway frequencies between virus-free mesophyll cells and those that were SRBSDV-infected (Table 2). There was also a remarkable difference in the distribution pattern of gateways: in virus-free stem phloem, almost all gateways were concentrated on the sieve plate (99.7%, 1576/1581) but in tumours the majority of gateways were present on non-sieve plate SE–SE interfaces (61.6%, 1142/1853), where they were distributed in several independent sites or areas (similar to sieve areas). Thus the frequency of the sites increased and the degree of gateway concentration declined when compared to virus-free stem phloem (Table 2). The increased frequency of gateways generated in tumour phloem provides more potential connectivity and a more dispersed pattern of gateway distribution on SE–SE interfaces, potentially enhancing symplastic transport.
Frequency of cell wall gateways on different types of cell interfaces
| SE–SE interface of virus-free phloem | SE–SE interface of infected non-tumour | SE–SE interface of infected tumour with virions | Mesophyll interface of virus-free leaf | Mesophyll interface of infected leaf without virions | |
|---|---|---|---|---|---|
| Length of cell wall interface (µm) | 1.0 (±0.04)a | 1.0 (±0.04) | 3.0 (±0.22) | 3.5 (±0.21) | 3.5 (±0.24) |
| Number of gateway sites per lengthb | 3.7 (±0.65) | 3.7 (±0.65) | 14.4 (±1.20) | 10.4 (±1.44) | 10.6 (±1.53) |
| Number of flexible and pore gateways on sieve plate per length | 51.0 (±5.25) | 58.3 (±6.67) | 8.1 (±1.13) | 0 (–) | 0 (–) |
| Number of flexible and pore gateways on normal SE wall per length | 0.16 (±0.37) | 0.14 (±0.35) | 12.9 (±1.19) | 0 (–) | 0 (–) |
| Degree of gateway concentration within gateway site | 14.2 (±1.77) | 16.2 (±2.65) | 1.5 (±0.09) | 1.9 (±0.17) | 1.9 (±0.18) |
| Number of plasmodesmata per length | 0 (–) | 0 (–) | 0 (–) | 19.3 (±1.18) | 19.6(±1.10) |
| SE–SE interface of virus-free phloem | SE–SE interface of infected non-tumour | SE–SE interface of infected tumour with virions | Mesophyll interface of virus-free leaf | Mesophyll interface of infected leaf without virions | |
|---|---|---|---|---|---|
| Length of cell wall interface (µm) | 1.0 (±0.04)a | 1.0 (±0.04) | 3.0 (±0.22) | 3.5 (±0.21) | 3.5 (±0.24) |
| Number of gateway sites per lengthb | 3.7 (±0.65) | 3.7 (±0.65) | 14.4 (±1.20) | 10.4 (±1.44) | 10.6 (±1.53) |
| Number of flexible and pore gateways on sieve plate per length | 51.0 (±5.25) | 58.3 (±6.67) | 8.1 (±1.13) | 0 (–) | 0 (–) |
| Number of flexible and pore gateways on normal SE wall per length | 0.16 (±0.37) | 0.14 (±0.35) | 12.9 (±1.19) | 0 (–) | 0 (–) |
| Degree of gateway concentration within gateway site | 14.2 (±1.77) | 16.2 (±2.65) | 1.5 (±0.09) | 1.9 (±0.17) | 1.9 (±0.18) |
| Number of plasmodesmata per length | 0 (–) | 0 (–) | 0 (–) | 19.3 (±1.18) | 19.6(±1.10) |
a Values are means (±SD) taken from measurements of all cells in 30 replicate samples each of 10 sections. b Gateway site is the tentative name for a small region of the cell wall where one or more gateways are clustered. It provided a parameter to measure the efficiency of transportation in this cell wall region. In phloem it is a sieve plate containing a cluster of pore gateways. In a tumour it is a small cell wall region containing one or several flexible/pore gateways. In mesophyll it is a small cell wall region containing one or more plasmodesmata.
Frequency of cell wall gateways on different types of cell interfaces
| SE–SE interface of virus-free phloem | SE–SE interface of infected non-tumour | SE–SE interface of infected tumour with virions | Mesophyll interface of virus-free leaf | Mesophyll interface of infected leaf without virions | |
|---|---|---|---|---|---|
| Length of cell wall interface (µm) | 1.0 (±0.04)a | 1.0 (±0.04) | 3.0 (±0.22) | 3.5 (±0.21) | 3.5 (±0.24) |
| Number of gateway sites per lengthb | 3.7 (±0.65) | 3.7 (±0.65) | 14.4 (±1.20) | 10.4 (±1.44) | 10.6 (±1.53) |
| Number of flexible and pore gateways on sieve plate per length | 51.0 (±5.25) | 58.3 (±6.67) | 8.1 (±1.13) | 0 (–) | 0 (–) |
| Number of flexible and pore gateways on normal SE wall per length | 0.16 (±0.37) | 0.14 (±0.35) | 12.9 (±1.19) | 0 (–) | 0 (–) |
| Degree of gateway concentration within gateway site | 14.2 (±1.77) | 16.2 (±2.65) | 1.5 (±0.09) | 1.9 (±0.17) | 1.9 (±0.18) |
| Number of plasmodesmata per length | 0 (–) | 0 (–) | 0 (–) | 19.3 (±1.18) | 19.6(±1.10) |
| SE–SE interface of virus-free phloem | SE–SE interface of infected non-tumour | SE–SE interface of infected tumour with virions | Mesophyll interface of virus-free leaf | Mesophyll interface of infected leaf without virions | |
|---|---|---|---|---|---|
| Length of cell wall interface (µm) | 1.0 (±0.04)a | 1.0 (±0.04) | 3.0 (±0.22) | 3.5 (±0.21) | 3.5 (±0.24) |
| Number of gateway sites per lengthb | 3.7 (±0.65) | 3.7 (±0.65) | 14.4 (±1.20) | 10.4 (±1.44) | 10.6 (±1.53) |
| Number of flexible and pore gateways on sieve plate per length | 51.0 (±5.25) | 58.3 (±6.67) | 8.1 (±1.13) | 0 (–) | 0 (–) |
| Number of flexible and pore gateways on normal SE wall per length | 0.16 (±0.37) | 0.14 (±0.35) | 12.9 (±1.19) | 0 (–) | 0 (–) |
| Degree of gateway concentration within gateway site | 14.2 (±1.77) | 16.2 (±2.65) | 1.5 (±0.09) | 1.9 (±0.17) | 1.9 (±0.18) |
| Number of plasmodesmata per length | 0 (–) | 0 (–) | 0 (–) | 19.3 (±1.18) | 19.6(±1.10) |
a Values are means (±SD) taken from measurements of all cells in 30 replicate samples each of 10 sections. b Gateway site is the tentative name for a small region of the cell wall where one or more gateways are clustered. It provided a parameter to measure the efficiency of transportation in this cell wall region. In phloem it is a sieve plate containing a cluster of pore gateways. In a tumour it is a small cell wall region containing one or several flexible/pore gateways. In mesophyll it is a small cell wall region containing one or more plasmodesmata.
Structure and components of the central module in the flexible gateway
The flexible and pore gateways both allow passage of the SRBSDV virion, which is 80nm in diameter and much larger than the size exclusion limit of normal plasmodesmata. However, in its structural features, the flexible gateway is more similar to plasmodesmata than to the pores. The central module of the flexible gateway resembles that of the plasmodesmata (desmotubule) and is absent from the open pore gateway (Fig. 2e, h). These central modules were therefore compared structurally and their components examined by immuno-gold labelling. Because there are no authentic plasmodesmata in SEs, we used plasmodesmata at the interface of uninfected phloem parenchyma for comparison (Fig. 5a). The flexible gateways used for comparison were both wall-located and sieve plate-located ones (Fig. 5b, e) and those with virions were also included to study any effects on the central module (Fig. 5d). Electron tomography reconstructions (Fig. 5c, f) showed that the 3D structure of the central module is a thin osmiophilic core. A membrane-like structure was closely attached to the inner cell wall surface of the flexible gateway with a small sleeve-like distance (5nm) around the central module. The central modules had similar structural details, regardless of their gateway locations or the presence of virions (Fig. 5b, d, e; arrowhead) except that the central module disappeared in the vicinity of a virion (Fig. 5d). The flexible gateway (20nm in diameter) is thinner than the plasmodesmata (35–40nm) in both its central module and the sleeve-like distance around the module making flexible gateways appear like compressed plasmodesmata. While the central module of the flexible gateway was sometimes discontinuous and did not always extend across the whole cell wall, the desmotubule of plasmodesmata was always continuous across the cell wall. These characteristics make the flexible structure appear to be somewhat degraded.
Structure and components of the central pith module in the flexible gateway. (a) Structure of plasmodesmata desmotubule at the interface between two phloem parenchyma, shown as a 2D projection (left box) and as a slice of electron tomography (ET) reconstruction (right). (b) Structure of the flexible gateway on non-sieve plate wall, shown as a 2D projection (left box) and as a slice of ET reconstruction (right). The central pith module is marked with a black arrowhead. (c) Detailed structure of the central pith module of flexible gateway on a non-sieve plate wall, shown in three different dimensions (XY, YZ, and XZ). The same position in each dimension is marked with a cross. (d) Structure of flexible gateway with virion on non-sieve plate wall shown by a slice of ET reconstruction. Note that the central pith module (black arrowhead) disappears around the virion-involving site, but is retained elsewhere. (e) Structure of flexible gateway on the sieve plate and its central pith module (black arrowhead), shown as a 2D projection (left box) and as a slice of ET reconstruction (right). (f) Detailed structure of the central pith module of flexible gateway on the sieve plate, shown at three different dimensions (XY, YZ, and XZ). The same position in each dimension is marked with a cross. (g) Immuno-gold labelling of the central pith module (arrowhead) of flexible gateway using an antibody against actin. (h) Immuno-gold labelling of plasmodesmata in virus-free phloem using an antibody against actin as a control. (i–k) Modified flexible gateway with gulf-like openings at the terminus, demonstrated by slices of ET reconstruction. The central pith module (black arrowhead) extends from the gateway into the cytoplasm. (l) The gulf-like opening at the terminus of flexible gateway large enough to trap a virion. CW, cell wall; Cyto, cytoplasm; SP, sieve plate. Bars: (a–h) 100nm; (i–l) 50nm.
The central module of plasmodesmata (desmotubule) is composed of a cytoskeleton bundle formed from actin (Blackman and Overall, 1998; White and Barton, 2011). Immuno-gold labelling using an antibody against actin showed specific labelling on the central module of the flexible gateway but not on other parts of the cell wall (Fig. 5g), a pattern similar to that seen with the plasmodesmata desmotubule (Fig. 5h). Thus both gateways appear to have an actin-associated cytoskeleton as the central module.
In addition, some flexible gateways were observed with open terminal structures (Fig. 5i–l) which resemble the pore gateway. Observations of the sliced electron tomography reconstruction showed that these openings were large enough for virion trapping (Fig. 5l), and the central module could be seen extending from these open ends of the gateway into the cytoplasm (Fig. 5i–k, arrowhead).
Modification of cell wall components at the gateways
Under TEM, the most prominent feature of the flexible gateway is the osmiophobic cell wall material which lacks any fibril-like structure (Fig. 3a). This feature could result from alterations in the cell wall components of this region. Callose (β-1,3-glucan) deposition is responsible for rapid cell wall adjustments (Zavaliev et al., 2011), while cellulose (β-1,4-D-glucan) is the main cell wall constituent. Gold labelling was therefore done using an antiserum against β-1-3-glucan and a cellulose-gold probe (Picard et al., 2000). Labelling for callose was absent in the parts of the cell wall harbouring fibril-like structures (Fig. 6a), but was observed in the osmiophobic region around gateways on the cell wall of SEs (Fig. 6b, c). The osmiophobic region could be divided into two parts on the basis of electron density. The gold particles labelling for callose were mostly present in the grey part, which was generally located at the periphery of the non-labelled translucent part (Fig. 6d–g). In the flexible gateway without virions, large amounts of colloid gold particles filled the whole gateway, regardless of their location on the sieve plate (Fig. 6c) or non-sieve plate wall (Fig. 6g). In the open pore gateway, large amounts of gold particles were precisely located on the margin of the pore (Fig. 6d). The results suggested that there was no callose within the open pore. Patterns of callose deposition pattern in the pore gateways where virions were present (Fig. 6e) were similar to those where no virions were observed and the same was true of the flexible gateways (Fig. 6f, g). By contrast, only a handful of gold particles were occasionally observed on the neck of plasmodesmata (Fig. 6h). An electron-lucent osmiophobic region was also observed in the vicinity of plasmodesmata, but no gold particles were seen in this region (Fig. 6h).
Immunogold labelling to detect callose in gateways. (a) No labelling for callose in the part of the cell wall harbouring fibril-like structures. (b,c) Substantial labelling for callose in the osmiophobic region around the gateways on the cell wall of SEs. (d) Substantial labelling on the osmiophobic material distributed on the margin of the mature pore gateway. (e) Substantial labelling on the osmiophobic cell wall of pore gateway with virion. (f) Substantial labelling on the osmiophobic region of flexible gateway with virion. (g) Substantial labelling on the osmiophobic region of flexible gateway without virion. (h) Small amount of labelling on the neck region of normal plasmodesmata. (i, j) Substantial labelling against β-1,3-glucanase on the osmiophobic region of flexible gateway with virion. The central module of the flexible gateway is marked with a black arrowhead. Bars: (b,i) 0.5 μm; (a, c–h, j) 200nm.
β-1,3-glucanase is the enzyme responsible for callose degradation (Zavaliev et al., 2011). In experiments using immuno-gold labelling for this enzyme, colloidal gold particles were observed at the periphery of the osmiophobic region of the flexible gateway (Fig. 6j, arrowhead), but were absent on normal fibrillar cell wall (Fig. 6j). No labelling was found on plasmodesmata or on the well-formed sieve-pore gateway. It is therefore possible that both callose deposition and degradation occur in the osmiophobic region of the flexible gateway.
The cellulose cell wall component of the flexible and pore gateways also underwent severe modification. Cellulose is the main component of normal cell wall (with moderate electron density), and there was a large amount of specific gold labelling for cellulose in the normal cell wall control (Fig. 7a). Cellulase labelling could also be observed on plasmodesmata (Fig. 7b), but not on the osmiophobic region of the flexible gateway regardless of the presence of virions (Fig. 7c, d). Similar labelling results were observed on the osmiophobic region of the sieve-pore gateway (Fig. 7e).
Cellulose modification of the flexible gateway revealed by cellulase immuno-gold labelling. (a, b) Specific labelling against cellulase on normal cell wall (a) and plasmodesmata (b) (controls). (c, d) Absence of labelling on flexible gateways without (c) and with (d) a virion, although the adjacent cell wall is labelled. (e) Absence of labelling on pore gateways with and without virions. The pore gateway is indicated by a black arrow. Virion is indicated with a white arrowhead. Bars, 2 μm.
Taken together, these results indicate that during the biogenesis of the flexible and pore gateways in virus-induced tumours, the amount of cellulose on the adjacent cell wall decreases while the amount of callose increases. This suggests that the turnover of callose may be a key regulator for the formation of these gateways.
Discussion
Cytopathology and microenvironment of SRBSDV-induced tumours
Our previous (Zhang et al., 2008; Hoang et al., 2011) and present studies have shown that characteristic reoviral cytopathological structures are found in the cytoplasm of phloem parenchyma cells in infected hosts. These include crystalline arrays of particles, tubular structures enclosing viral particles, amorphous viroplasms containing complete and incomplete virus particles, and thin fibrillar material. The frequent viroplasms, which appear to have no limiting membrane and contain much fibrillar material, are sometimes surrounded by small groups of viral particles. Only mature virions occur outside the viroplasms and within crystals or tubular structures, suggesting that the phloem parenchyma is the main cell type in which virus multiplies. In this study, some free particles were found in SEs in addition to phloem parenchyma cells, while other cell types, including sclerenchyma and xylem, were virus-free and appeared to be normal. This is to be expected if the virus is phloem-limited (Attoui et al., 2012). Although plant tumours have sometimes been regarded as unorganized tissue, it is clear that plant reovirus-induced tumours are more highly organized although SEs are not associated into the bundles seen in healthy phloem. Because tumours are not induced by the feeding of the insect vector in the absence of virus, we assume that these changes are induced by one or more viral proteins.
The flexible gateway in virus-induced tumours
In higher eukaryotes, life depends on intercellular communication and exchange by traffic of molecules or molecular complexes through direct cytoplasmic bridges (such as plasmodesmata in plants and membrane nanotubes in mammals) (Oparka, 2005; Sherer and Mothes, 2008). However, most known viruses hijack the host machinery, including these intercellular cytoplasmic bridges, for their own purposes (Sherer and Mothes, 2008; Lee and Lu, 2011). In mammals, recent studies have shown that retroviruses including murine leukemia virus and HIV-1 can drive the formation of new membrane nanotubes, which become conduits for intercellular trafficking of viruses (Sherer et al., 2007; Sherer and Mothes, 2008). In plants, it has been well established that plasmodesmata, a kind of membrane-lined cytoplasmic bridge between non-vascular cells, are often commandeered by plant viruses and undergo radical modification with viral movement proteins or movement protein-formed tubules for the movement of viral ribonucleoprotein complexes or virions (Carrington et al., 1996; Lee and Lu, 2011). By contrast, the mechanism governing virus movement through the SE–SE interfaces in the plant phloem vascular system is far from being understood, but it is believed that viruses are passively transported through the SE–SE interfaces by mass flow from source to sink tissues. Sieve pores, located on the SE–SE interfaces in the phloem, are usually ~200–400nm in diameter. Because these are much bigger cytoplasmic bridges than plasmodesmata (an average diameter of only ~33nm), they are thought to be large enough for most viruses to traffic, although there has been little experimental evidence (Oparka and Turgeon, 1999). Here we provide strong evidence that a reovirus moves in the form of intact virions through the sieve pores. However, it appears not to be passive transport by mass flow because tumours have generally been regarded as a strong metabolite sink tissue for their host plants (Ullrich and Aloni, 2000). The direction and speed of virus movement might be different from those of photoassimilates and remain to be intensively studied, but the reovirus-induced tumour provides an excellent and unique model system.
Interestingly, SRBSDV can not only hijack the sieve-pore gateways, but also drive the formation of new flexible cytoplasmic bridges for trafficking its virions across the SE–SE interfaces of the phloem. These flexible gateways are obviously different in ultrastructure and location to all well-known intercellular gateways, including plasmodesmata, sieve-plate pores, and PPUs (a type of special gateway that is always only branched on the companion-cell side of the SE–companion cell interface). The flexible gateway is intermediate in size between plasmodesmata and sieve-plate pores. It resembles plasmodesmata in having a central membrane-lined channel and shares with the sieve pore gateway a callose deposition pattern and the potential capacity for virion transport. It would be interesting to study the way in which reovirus infection reprogrammes the development of cytoplasmic bridges to gain a better insight into the differentiation of intercellular gateways.
Studies of SE–SE cell wall gateways have usually focused on the development of sieve pore gateways from plasmodesmata (Sjolund, 1997; van Bel et al., 2002). Callose deposition and degradation at the gateway site generates an opening that facilitates unimpeded transport (Esau and Thorsch, 1985; Lee and Lu, 2011; Xie and Hong, 2011). Both cellulose evacuation and callose degradation lead to cell wall disappearance during the formation of the pore gateway (Xie and Hong, 2011). Our results suggest that similar processes may be involved in the formation of the flexible gateways in SRBSDV-infected rice phloem.
SRBSDV induces the de novo formation of a SE–SE region: hyperplasia or programmed cell death?
Long-distance transport in plants in the phloem is accomplished inside the sieve tube, formed from a series of SEs. Typically, SEs are terminally enucleated and differentiated cells for unimpeded transport, dependent on their adjacent companion cells (Esau, 1969; Sjolund, 1997; Oparka and Turgeon, 1999). However, in the reovirus-induced tumours, a region composed exclusively of SEs was induced. The only report of SEs lacking companion cells is of a few cells in the late metaphloem of grass leaves (Evert, 2007) but, frustratingly, information on their ontogeny and role remains very limited (Botha, 2013). Many more SEs were found in the phloem of the tumours studied here, suggesting that the SEs were derived from the division of phloem–parenchyma cells or the differentiation of companion cells. In the tumours, there was hyperplasia of both SEs and phloem–parenchyma, and both were invaded by virus while the mature SEs underwent drastic changes, including modification of cell wall structures and selective autolysis of organelles, similar in many ways to programmed cell death (Sjolund, 1997). Thus the virus creates an ecological habitat suitable for its own multiplication by inducing hyperplasia of phloem–parenchyma and drives a number of symplastic bridges suitable for its own movement by directing the selective autolysis of organelles during SE differentiation.
A model for the movement of tumour-inducing reoviruses in plants
Observation of virions in cell wall gateways has generally been considered to provide sufficient evidence for viral movement, such as CPMV generating the virion-packaged tubule in mesophyll plasmodesmata (Ritzenthaler and Hofmann, 2007; Niehl and Heinlein, 2011), CRLV virions in the SE–companion cell PPU for long-distance movement in phloem (Oparka and Turgeon, 1999), and Rice yellow mottle virus (RYMV) in pit membranes for long-distance movement between xylem cells (Opalka et al., 1998). In our study, we provide the first evidence that SRBSDV is transported as virions through sieve pores and flexible gateways on the SE–SE interfaces in the SE region of the tumour. SRBSDV virions were found in the flexible and sieve pore gateways of both sieve plate and non-sieve plate SE–SE interfaces, and not in mesophyll plasmodesmata. Unlike the plasmodesma–virion–tubule pattern, there were no tubules formed from viral movement protein to accommodate the large virions. The distinctive features of these gateways may explain why most plant reoviruses are phloem-limited and how they create their own ecological habit within the host.
Callose is the best-studied universal key regulator of plasmodesmata (Ueki and Citovsky, 2011) and plays multifunctional roles in defence against virus infection (Iglesias and Meins, 2000; Bucher et al., 2001; Lee and Lu, 2011; Luna et al., 2011; Zavaliev et al., 2011), sieve pore formation, and wounding response (Xie and Hong, 2011). The involvement of SRBSDV virions in the flexible and sieve pore gateways was associated with local callose deposition, suggesting that callose turnover might also contribute to the formation and control of these gateways for virus transport on SE–SE interfaces in tumours. Callose deposition would block virus movement directly, while its removal by β-1,3-glucanase would allow the pore to open for virus trafficking.
We propose a working model to describe the movement of SRBSDV in the specific SE region of the tumour (Fig. 8a). SRBSDV infection induces the formation of a specific SE region where cells have many flexible and sieve-pore gateways on all interfaces, which would allow for free movement of virions throughout the region. The flexible and sieve pore gateway hyperplasia on SE–SE interfaces resembles the mechanism of plasmodesmata–pore gateway development on sieve plates (Lee and Lu, 2011) and suggests that the flexible gateways may also be highly modified plasmodesmata. The remodelling of the cell wall to generate these gateways involves callose accumulation and cellulose evacuation, resembling the cell wall remodelling pattern that happens on sieve plates (Xie and Hong, 2011).
Proposed model for the cell-to-cell movement of SBRSDV in tumours. SE arrangement in healthy phloem (left) and virus-infected tumour (right). In healthy phloem, small numbers of SEs and companion cells are interlaced in a vascular bundle, while in SRBSDV-infected phloem, hyperplasia of SEs has produced a special SE region (tumour) without companion cells, increasing the number of SE–SE interfaces. This benefits the cell-to-cell movement of SRBSDV virions, which mostly occurs in the SE–SE interface. In the tumour region, more flexible/pore gateways are generated for facilitating the movement of large-sized SRBSDV virions. These gateways occur not only in the direction of mass flow but in all orientations. (b) Proposed structural model of flexible/pore gateway. On the left is a tight or semi-open flexible gateway. A cytoskeleton of actin is retained as the central module in some part of the gateway. A rapidly responding callose component fills over the gateway while constitutive cell wall cellulose is absent. The movement of SRBSDV virions in such a crowded flexible gateway is impeded. On the right is an open pore gateway. There is no actin cytoskeleton or callose in the centre of the gateway and SRBSDV virions are able to move freely. β-1,3-glucanase plays an important role in this development by degrading callose. CW, cell wall; SE, sieve element.
Funding
This study was financially supported by National Science and Technology Support Program (2012BAD19B03), the State Basic Research Program of China (2010CB126203), the Special Fund for Agro-scientific Research in the Public Interest of China (201003031), the Zhejiang Provincial Natural Science Foundation of China (LQ13C010003), and the Department of Science and Technology, Zhejiang Provincial Government (2009R50032).
Acknowledgements
We were grateful to Profs Jian Hong and Dongwei Hu, Zhejiang University, for guiding observations and providing cellulose-gold probe, respectively. We thank Prof. M.J. Adams, Rothamsted Research, Harpenden, UK for help in correcting the English of the manuscript.
References
Comments