Dynamic cytoskeleton rearrangements in giant cells and syncytia of nematode-infected roots

Authors

  • Janice De Almeida Engler,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology (VIB), Ghent University, B-9052 Gent, Belgium,
    Search for more papers by this author
  • Kris Van Poucke,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology (VIB), Ghent University, B-9052 Gent, Belgium,
    Search for more papers by this author
  • Mansour Karimi,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology (VIB), Ghent University, B-9052 Gent, Belgium,
    Search for more papers by this author
  • Ruth De Groodt,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology (VIB), Ghent University, B-9052 Gent, Belgium,
    Search for more papers by this author
  • Greetje Gheysen,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology (VIB), Ghent University, B-9052 Gent, Belgium,
    2. Vakgroep Moleculaire Biotechnologie, Faculteit Landbouwkundige en Toegepaste Wetenschappen, Universiteit Gent, B-9000 Gent, Belgium, and
    Search for more papers by this author
  • Gilbert Engler,

    Corresponding author
    1. Laboratoire Associé de l'Institut National de la Recherche Agronomique (France), Universiteit Gent, B-9000 Gent, Belgium
    Search for more papers by this author
  • Godelieve Gheysen

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology (VIB), Ghent University, B-9052 Gent, Belgium,
    2. Vakgroep Moleculaire Biotechnologie, Faculteit Landbouwkundige en Toegepaste Wetenschappen, Universiteit Gent, B-9000 Gent, Belgium, and
    Search for more papers by this author

For correspondence (fax +33 493678955; e-mail gengler@antibes.inra.fr).
Present address: Institute National de la Recherche Agronomique, UMR Interactions Plantes-Microorganismes et Santé Végétale, B.P. 167, F-06903 Sophia Antipolis, France.

Summary

Giant cells induced by root knot nematodes and syncytia caused by cyst nematodes are large multinucleated feeding cells containing a dense cytoplasm generated during a complex host–parasite association in plant roots. To find out whether cytoskeleton changes occurred during feeding cell development, transcriptional activity of actin (ACT) and tubulin genes and organization of the ACT filaments and of the microtubules (MTs) were analyzed in situ. The importance of changes in the cytoskeleton architecture for the proper initiation and development of galls and syncytia was demonstrated by perturbing the cytoskeleton with chemical inhibitors. The expression levels of cytoskeletal components, such as tubulins and ACTs, are proposed to be upregulated to allow the assembly of a new cytoskeleton in expanding feeding cells. However, MTs and ACT filaments failed to properly organize and appeared partially depolymerized throughout feeding site development. Both the actin and tubulin cytoskeletons were strongly disrupted in syncytia and mitotic figures were never observed. In contrast, in giant cells, an ACT and cortical MT cytokeleton, although disturbed, was still visible. In addition, a functional mitotic apparatus was present that contained multiple large spindles and arrested phragmoplasts, but no pre-prophase bands. Chemical stabilization of the microtubular cytoskeleton with taxol blocked feeding site development. On the other hand, when the ACT or MT cytoskeleton of feeding cells was depolymerized by cytochalasin D or oryzalin, nematodes could complete their life cycle. Our data suggest that the cytoskeleton rearrangements and depolymerization induced by parasitic nematodes may be essential for a successful feeding process.

Introduction

Sedentary endoparasitic nematodes are widely spread parasites of crop plants that unfavorably influence the plant's growth and yield. Juvenile nematodes penetrate into the roots and migrate to reach cells in the vascular cylinder that will be induced to become feeding cells. The root knot nematodes (Meloidogyne sp.) induce galls that contain giant cells, which become multinucleate via a series of karyokinesis events without cytokinesis (Huang and Maggenti, 1969). On the other hand, cyst nematodes (such as Heterodera schachtii) induce syncytia that become multinucleated via fusion of an initial feeding cell with neighboring dividing cells. Fully developed feeding cells are dramatically enlarged in size and highly polyploid (Huang, 1985; Wiggers et al., 1990). In contrast, neighboring cells are smaller and their DNA content seems to remain mostly diploid (J. de Almeida Engler, unpublished data). During development of feeding sites, both giant cells and syncytia build up a dense cytoplasm and the central vacuole is replaced by numerous small ones. Feeding cells always develop adjacent to xylem elements where cell wall ingrowths expand, suggesting the withdrawal of solutes from the xylem. Mechanisms by which nematodes induce feeding cell formation and expansion are not yet understood, but in vivo observations (Wyss, 1992) indicate that products secreted by the nematode during parasitism (reviewed by Davis et al., 2000) are most probably involved in the formation of feeding sites.

Plants are uniquely adapted to respond to environmental signals for survival. Many of these signals result in fundamental changes in cell shape and cytoplasmic organization, which depend on the cytoskeleton (Gibbon, 2001). The cytoskeleton is a network of interconnected fibrous protein polymers that gives structural stability to the cytoplasm. It plays a role in many processes, such as anchoring and trafficking organelles and proteins (Vale, 1987), nuclear movement (Chytilova et al., 2000), cytoplasm streaming (Kamiya, 1981), cell wall synthesis (Goddard et al., 1994), cell morphogenesis, division (Nakaseko and Yanagida, 2001), and differentiation (Seagull and Falconer, 1991).

Elements of the cytoskeleton are reorganized adaptively when plants perceive various external stimuli (Nick et al., 1991), such as wounding (Hush and Overall, 1992; La Claire, 1989), hormone treatment (Roberts et al., 1985), symbiosis (Timmers et al., 1998, 1999), viral transport (Heinlein et al., 1995), and pathogen attack (Grant and Mansfield, 1999), and seem to be directly involved in plant defense mechanisms (reviewed in Kobayashi and Kobayashi, 2000). The cytoskeleton is dynamic because it rearranges continually its three components: actin (ACT) microfilaments (MFs), microtubules (MTs), and intermediate filaments (Bretscher, 1991). ACT monomers self-assemble in long polymers named filamentous actin (F-actin), and tubulin proteins polymerize forming MTs. Until now, there is only suggestive evidence of the presence of intermediate filament proteins in plants (Baskin, 2000) and no orthologs have been found in the Arabidopsis thaliana genome (Assaad, 2001).

Arabidopsis has eight functional highly homologous ACT genes (Meagher et al., 1999). ACT2 is the most abundant ACT mRNA in the vegetative stage (An et al., 1996). ACT7 and ATC8 are expressed in most vegetative tissues, preferentially in young tissues, and ACT7 is also involved in hormone-induced cell proliferation (Kandasamy et al., 2001; McDowell et al., 1996a). The five other functional ACT genes are mostly expressed in reproductive organs (McDowell et al., 1996b).

MTs consist of heterodimers of α- and β-tubulins and are mainly generated in the cell cortex and around the nuclear envelope (MT-organizing centers; Hepler et al., 1993; Lloyd, 1991). The cortical, interphase array of MTs is involved in alignment of cellulose along the cell wall, guiding the direction of cell expansion. In a mitotic cycle, three successive MT arrays are formed: the pre-prophase band (PPB) that predicts the cell division plane, the mitotic spindle, and the centrifugally growing phragmoplast that directs wall-forming vesicles to the site where a new cell plate is formed (Goddard et al., 1994). In Arabidopsis, six α-tubulin (Kopczak et al., 1992; Lloyd, 1991) and nine β-tubulin genes (Oppenheimer et al., 1988; Snustad et al., 1992) have been identified. At least two γ-tubulin homologs are present in Arabidopsis (Liu et al., 1994). Other tubulins, such as δ-, ε-, ζ-, and η-tubulins, are seemingly not ubiquitously distributed in eukaryotic organisms (reviewed by Dutcher, 2001).

Our aim was to get a picture on how the ACT and MT cytoskeletons are rearranged in the large feeding cells induced by two types of nematodes. To find out whether and how such changes occur, we followed different approaches. We monitored the transcription activity of ACT2 and ACT7 as representatives of the ACT gene family; mRNA expression analysis was performed for the three major plant tubulins (α, β, and γ). Immunolocalization experiments were carried out with antibodies against ACT and α-, β-, and γ-tubulins. As Arabidopsis is a model rather than being a natural host for nematode infection, we also analyzed galls in pea. Finally, the MT and ACT cytoskeleton organization in giant cells and syncytia was studied in vivo by using green-fluorescent protein (GFP)-labeled ACT and MTs.

To determine whether the nematode infection process depends on the rearrangement of MTs and/or ACT filaments, we tested two cytoskeleton inhibitors: paclitaxel (taxol) that enhances self-assembly of MTs and makes them resistant to depolymerization (Morejohn and Fosket, 1984) and the ACT-inhibiting chemical cytochalasin D that caps the MF end and inhibits globular ACT addition and loss (Sampath and Pollard, 1991), resulting in a fragmented ACT cytoskeleton.

Results

Promoter activity of the ACT2 and ACT7 genes

Arabidopsis lines that expressed the gus gene from the ACT2 and ACT7 promoters were analyzed by GUS assays of uninfected and infected seedlings at different time points after root knot or cyst nematode inoculation. In uninfected roots, both promoters were active in the vascular cylinder (Figure 1a,f). The promoter activity of ACT2 was stronger than that of ATC7 in the root meristem, but that of the latter was stronger in the elongation or differentiation zone (Figure 1f). For the two ACT–gus lines, infection with Meloidogyne incognita or H. schachtii resulted in strongly GUS-stained galls and syncytia, respectively, early after nematode infection (already 1–24 h after inoculation) with a peak approximately 4–5 days after inoculation (DAI). At this stage, GUS stains were more intense in feeding sites than in uninfected vascular root cells, an indication that the promoters of ACT2 (Figure 1b,d) and ACT7 (Figure 1g,i) had been induced. At later time points (15–20 DAI), the promoters of both ACT genes remained active and somewhat higher for ACT7 (Figure 1c,e,h,j). In the two lines, vascular tissues near galls and syncytia often showed stronger GUS staining than more distal tissues, demonstrating that nematode infection affects not only the feeding sites but also the flanking vascular cylinder.

Figure 1.

Activity of ACT2- and ACT7-promoter–gus fusions in uninfected and nematode-infected Arabidopsis roots.

(a) ACT2–gus expression in an uninfected root.

(b, c) ACT2–gus expression in galls 5 and 15 DAI, respectively.

(d, e) ACT2–gus expression in a syncytium 5 and 15 DAI, respectively.

(f) ACT7–gus expression in an uninfected root.

(g, h) ACT7–gus expression in galls 5 and 15 DAI, respectively.

(i, j) ACT7–gus expression in a syncytium 5 and 15 DAI, respectively.

EZ, elongation zone; G, gall; n, nematode; RM, root meristem; S, syncytium. Bar = 100 µm.

GFP analysis and immunofluorescence microscopy of ACT filaments in feeding cells

To investigate in vivo the ACT cytoskeleton in nematode feeding sites, Arabidopsis seedlings containing the GFP-mTn construct (Kost et al., 1998) were infected with cyst (Figure 2b,c) or root knot (Figure 2f,g) nematodes. During migration, nematodes were followed microscopically until the stylet was visualized within a vascular root cell. Early after infection, F-actin was seen within a diffuse GFP fluorescence in feeding sites close to the nematode (Figure 2b,f). During syncytium expansion only diffuse and amorphous ACT fluorescence was observed (Figure 2c). In contrast, throughout gall development, apparently fragmented ACT filaments remained present within a diffuse fluorescence in the cytoplasm of giant cells (Figure 2g). Confocal observations were confirmed by immunostaining on sections of infected roots of Arabidopsis (Figure 2e,j) and pea (Figure 2h,i) with an anti-ACT antibody. Table 1 summarizes the organization of the ACT cytoskeleton in different cell types.

Figure 2.

Projections of confocal images of GFP-mTn in uninfected root, syncytia, and galls of Arabidopsis, and immunofluorescence detection of ACT in nematode-infected Arabidopsis (e,j) and pea (h,i) roots.

(a) Uninfected cells in the vascular tissue with F-actin.

(b) Syncytium (1 h after infection) with diffuse and F-actin.

(c) Syncytium 3 DAI with mainly diffuse ACT fluorescence compared to NC.

(d) Cells neighboring a syncytium.

(e) Syncytium 5 DAI with a diffuse ACT stain.

(f) Initiating gall (3 h after infection); based on an interference microscopy image, a line was drawn where the nematode is located within the root.

(g) Young giant cells (3 DAI) with a diffuse ACT stain compared to cells neighboring giant cells with F-actin. White arrowhead points to remaining F-actin.

(h) Pea gall 8 DAI with F-actin and bundled ACT within the cytoplasm of giant cells.

(i) Pea gall 4 DAI with ACT bundles throughout the giant cell cortex. White and yellow arrowheads point to cytoplasmic and cortical ACT bundles, and mitotic nuclei, respectively.

(j) Arabidopsis gall 10 DAI.

ACT is visualized in red and 4′,6-diamidino-2-phenylindole-stained nuclei in blue. *Giant cell; IG, initiating gall; IGC, initiating giant cell; ISC, initial syncytial cell; n, nematode; NC, neighboring cells; S, syncytium. Bars = 20 µm (a,c,d,f,g), 10 µm (b), 50 µm (e,h–j).

Table 1.  Cytoskeleton organization in uninfected and infected Arabidopsis root cells
InfectionCell typeTimeACTMTs
  1. NC, neighboring cells; S, syncytia; GC, giant cells; HAI, hours after inoculation; DAI, days after inoculation.

Uninfected (above root tip)Vascular tissue Filaments mainly longitudinally oriented (Figure 2a)Obliquely oriented (data not shown)
Cortex Filaments mainly randomly oriented (Figure 2a)Transversely to obliquely oriented (not shown)
Epidermis Filaments mainly randomly oriented (not shown)Mainly transversely oriented (not shown)
Cyst nematodeNC Filaments randomly oriented (Figure 2d)Obliquely to randomly oriented MTs and mitotic cells (Figure 4d,e)
S1–3 HAIFilaments within diffuse fluorescence (Figure 2b)Obliquely to randomly oriented MTs (not shown)
S1–3 DAIDiffuse and amorphous fluorescence (Figure 2c)Dense arrays mostly obliquely oriented within a diffuse fluorescence (Figure 4e)
S5–20 DAIDiffuse and amorphous label (Figure 2e)Diffuse and amorphous label (Figure 4i)
Root knot nematodeNC Filaments mainly randomly oriented (Figure 2g)Randomly oriented and fine MTs and mitotic cells (Figure 4k,r)
GC3 HAIDense ACT fluorescence close to nematode (Figure 2f)
Filaments within diffuse fluorescence (not shown)
Fragmented MT arrays randomly oriented within a diffuse fluorescence (Figure 4f″)
GC1 DAIFilaments within diffuse fluorescence (Figure 2g)Fragmented MT arrays randomly oriented within a diffuse fluorescence (not shown)
 GC3–10 DAIGiant cell cortex: thick cables longitudinally oriented alternating with parallel and randomly oriented ones (Figure 2h)Giant cell cortex: few cortical MTs (Figure 4h)
   Giant cell cytoplasm: thin ACT bundle segments randomly oriented within an amorphous ACT staining (Figure 2i)Giant cell cytoplasm: amorphous fluorescent label (Figure 4l) and presence of multiple spindles and phragmoplasts (Figure 4r,s)

α-, β-, and γ-tubulin gene expression

Sections of uninfected and nematode-infected roots at different stages of development were hybridized in situ. In uninfected roots, the α- and γ-tubulin mRNA levels were higher in developing vascular tissues than in differentiated cortex and epidermis cells (Figure 3a,b). At all stages analyzed (1–20 DAI), a clear homogeneous hybridization signal of the α- and β-tubulin genes was detected in syncytia and in dividing neighboring cells that would be incorporated later on (Figure 3c,d). However, hybridization with a γ-tubulin antisense probe showed a stronger signal in neighboring cells than in the syncytial cell (Figure 3e). In galls, α-, β-, and γ-tubulins were highly expressed in giant cells and neighboring cells (1–10 DAI; Figure 3g–i). At later stages of gall development, expression of the three tubulins decreased but persisted until 25 DAI. No hybridization signal was seen with sense probes (Figure 3f,j).

Figure 3.

α-, β-, and γ-tubulin mRNAs in uninfected and nematode-infected roots.

(a, b) Uninfected roots hybridized with antisense γ- and α-tubulin probes, respectively.

(c–e) Syncytia 10 DAI hybridized with antisense α-, β-, and γ-tubulin probes, respectively.

(f) Syncytium 10 DAI hybridized with a sense α-tubulin probe.

(g–i) Galls 7 DAI hybridized with antisense α-, β-, and γ-tubulin probes, respectively.

(j) Gall 7 DAI hybridized with a sense α-tubulin probe.

Red dots represent hybridization signals. G, gall; n, nematode; NC, neighboring cells; S, syncytium. Bar = 50 µm.

GFP analysis and immunofluorescence microscopy of MTs in feeding cells

Transgenic seedlings that produced the GFP–MT-binding domain (MBD) protein had a wild-type organization of the MT cytoskeleton and could thus be used to follow in vivo cytoskeletal changes within the feeding sites. The juvenile and the initiating feeding sites could be visualized with differential interference contrast optics (Figure 4a, initial syncytium cell). During in vivo observations of infected roots at low magnification, both galls and syncytia were easily viewed by the intense GFP fluorescence (Figure 4b,c). During syncytium development, the three mitotic arrays were detected in several mitotic neighboring cells (PPBs, spindles, and phragmoplasts; arrowheads in Figure 4d) that at later stages would be incorporated into the expanding syncytia. No mitotic MT arrays were visible during syncytial cell development. During root knot nematode penetration, most of the MTs in epidermal and cortical cells maintained the transverse-to-oblique orientation normally observed in these cells at the root elongation zone (Figure 4f,f′). During gall and syncytium development, an amorphous fluorescence was observed within the cytoplasm of feeding cells (Figure 4e,f″,h). Data of MT organization during the infection process by the juvenile nematode are summarized in Table 1. Immunolocalization (Figure 4i–s) of α- and β-tubulins in Arabidopsis roots infected with both nematodes confirmed in vivo observations.

Figure 4.

Confocal projections of GFP-MBD fluorescence in syncytia and in galls of Arabidopsis and immunofluorescence detection of α- and β-tubulins in nematode-infected Arabidopsis (i,j) and pea roots (k–s). The GFP-MBD fusion is under the control of the 35S promoter (b,c,f) or the WRKY promoter (d,e,g,h); in the immunolocalization experiments MTs are labeled green, 4′,6-diamidino-2-phenylindole -stained nuclei fluoresce blue, and red is autofluorescence of the root tissue.

(a) Differential interference contrast image of a nematode and the initial syncytial cell.

(b) Gall 3 DAI.

(c) Syncytium 3 DAI.

(d) Cells neighboring a syncytium. White arrowheads point to mitotic cells.

(e) Syncytium 3 DAI showing an initiating syncytium and the fusing neighboring cells.

(f) Root knot nematodes penetrating the roots.

(f′) Detail of the epidermal and cortical cells.

(f″) Young giant cells 1 DAI.

(g) Cells neighboring giant cells of a gall 5 DAI. White arrowheads point to phragmoplasts and neighboring cells with random MT orientation.

(h) Giant cells 5 DAI without clear tubular structures.

(i) Syncytium of Arabidopsis 7 DAI.

(j) Gall of Arabidopsis 7 DAI.

(k) Cells neighboring the giant cells in pea.

(l) Giant cell of pea 4 DAI containing synchronous mitotic nuclei; yellow arrowhead points to a phragmoplast.

(m) Spindle in a giant cell of pea.

(n) Spindle in a cell neighboring a giant cell of pea.

(o) Phragmoplast of a pea giant cell 4 DAI.

(p, q) Phragmoplasts of cells neighboring giant cells of pea.

(r) Giant cell of pea 4 DAI; white arrowheads point to two spindles within a single giant cell. Note the cortical MTs in a neighboring cell.

(s) Giant cell of pea 4 DAI; white and yellow arrowheads point to four phragmoplasts within a single giant cell and to two apparently fusing nuclei, respectively.

*Giant cells; C, cortex, Ep, epidermis; G, gall, ISC, initial syncytial cell; n, nematode; NC, neighboring cells; S, syncytium. Bars = 25 µm (a,e,f), 20 µm (i,j,m–s), 50 µm (c,d,g,h,k), and 100 µm (b,l).

In galls, the MT cytoskeleton was also studied by immunocytochemistry in pea because mitotic figures in giant cells occur more often in pea than in Arabidopsis; therefore, it is a better system to analyze whether PPBs, spindles, and/or phragmoplasts were present in mitotic giant cells.

In contrast to the apparently unstructured cytoplasmic MT cytoskeleton, a mitotic cytoskeleton containing spindles (Figure 4m,r) and phragmoplasts (Figure 4o,s) was clearly seen in giant cells, but PPBs were never observed. Spindles in giant cells were larger and often altered in shape (arrowheads in Figure 4m,r) when compared with spindles of neighboring cells (Figure 4n). Also phragmoplasts were thicker than in the neighboring cells (Figure 4o–q) and had not always their typical double-banded shape. Phragmoplasts were never observed close to the cell wall (Figure 4). Two mitotic nuclei (yellow arrows in Figure 4s) that apparently fused confirmed previous observations of abnormal division of genetic material in giant cells (Starr, 1993). Frequently, tubulins were strongly stained around the mitotic nuclei and the cell cortex of giant cells, suggesting that MTs might still interconnect the host cell nucleus to the cell cortex through the dense cytoplasm (Figure 4l). Phragmoplasts of neighboring cells were often asymmetrically positioned, explaining why diverse cell shapes were observed in gall cells surrounding the giant cells (Figure 4q).

Cytoskeleton inhibitor treatments

Infected roots were treated with the cytoskeleton inhibitors taxol and cytochalasin D to investigate whether MT stabilization or ACT depolymerization would affect nematode infection and/or feeding site development. Incubation of juveniles with the inhibitors (up to 20 µm taxol and up to 2 µm cytochalasin) did not interfere with nematode viability and reproduction. Lower concentrations of the inhibitors were used (1 µm of taxol and 0.5 µm of cytochalasin) for longer incubations (up to 40 DAI) to minimize possible long-term effects on nematode development. Long treatments with low concentrations of inhibitors allowed treated roots to grow and nematodes to develop and reproduce. In vivo GFP and immunolocalization analyses of seedlings treated with taxol, cytochalasin D, or oryzalin (data not shown) revealed thicker MTs, degraded ACT cytoskeleton and degraded MTs, respectively, when compared with untreated roots.

Nematodes were able to infect the roots of Arabidopsis plantlets pre-treated on media with high concentrations of taxol or cytochalasin D. However, especially in taxol-treated plantlets, the rate of feeding site initiation was low and no nematodes matured after 40 days (39 days at low taxol concentration). Control experiments showed that when roots were transferred to medium without inhibitors after the initial incubation on inhibitor-containing medium, the expected rate of nematode maturation (approximately 70%) was observed. Table 2 summarizes how treatments with taxol or cytochalasin D were performed (for details, see Experimental procedures), and Figure 5 presents the results on nematode reproduction after long incubations.

Table 2.  Treatments with cytoskeleton inhibitors (taxol or cytochalasin D) on Arabidopsis seedlings and time points
Set-upPeriod of consecutive treatments with
No inhibitorInhibitor at high concentrationInhibitor at low concentrationNo inhibitor
  1. HBI, hours before inoculation; DAI, days after inoculation; h, hour; –, not used in the set-up.

Control24 hInfection up to 40 DAI
2 HBI2 hInfection up to 40 DAI
24 HBI24 hInfection up to 40 DAI
3 DAIInfection up to 3 DAI24 hUp to 40 DAI
14 DAIInfection up to 14 DAI24 hUp to 40 DAI
Figure 5.

Effect of cytoskeleton inhibitors applied at different time points before or after inoculation on the reproduction (%) of cyst and root knot nematodes.

For every treatment, a total of 50 plants were infected each with 50 nematodes. Percentages of cyst and root knot nematodes that were able to mature under taxol (black) or cytochalasin D (gray) treatments (up to 40 DAI) were estimated by counting how many nematodes completed their life cycle compared to the number of juveniles that infected the Arabidopsis roots. DAI, days after inoculation; HBI, hours before inoculation.

When seedlings were transferred to taxol 3 DAI with the cyst nematode, syncytium development was arrested and, as a result, no females matured (Figure 6a). Microscopical observations showed that nematodes could develop from juveniles into living females that did not mature. By GFP analysis of taxol-treated and cytochalasin D-treated syncytia (5 DAI), stabilized MT arrays (Figure 6b) and a depolymerized ACT cytoskeleton (Figure 6d) were seen, respectively. In tissue sections of taxol-treated infected roots (40 DAI), syncytia were significantly smaller than untreated ones (compare Figure 6c with 6g), whereas in those treated with cytochalasin D, the syncytia were small, expanded irregularly, and had a dense cytoplasm (Figure 6e). Untreated syncytia were larger (Figure 6g) than those treated with taxol or cytochalasin D (Figure 6c,e, respectively). As shown in Figures 5 and 6, a fraction of the nematodes was able to complete their life cycle under inhibitor treatment.

Figure 6.

Microscopic analysis of Arabidopsis roots infected with cyst and root knot nematodes treated and untreated with cytoskeleton inhibitors.

(a) Infected root with cyst nematode 40 DAI, of which the last 37 days were on taxol-containing medium.

(b) Confocal image of GFP-MBD fluorescence in a syncytium 5 DAI, of which the last 24 h were on taxol-containing medium.

(c) Section through a syncytium 40 DAI, of which the last 37 days were on taxol-containing medium.

(d) Confocal image of GFP-mTn fluorescence in a syncytium 5 DAI, of which the last 24 h were on cytochalasin D-containing medium.

(e) Section through a syncytium 40 DAI, of which the last 37 days were on cytochalasin D-containing medium.

(f) Infected root with cyst nematode 40 DAI, of which the last 37 days were on cytochalasin D-containing medium.

(g) Section through a syncytium 40 DAI.

(h) Infected root with root knot nematode 40 DAI, of which the last 37 days were on taxol-containing medium.

(i) Confocal image of GFP-MBD in a gall 5 DAI, of which the last 24 h were on taxol-containing medium.

(i′) Same image visualized by differential interference contrast microscopy.

(j) Section through a gall 40 DAI, of which the last 37 days were on taxol-containing medium.

(k) Confocal image of GFP-mTn fluorescence in a gall 5 DAI, of which the last 24 h on cytochalasin D-containing medium.

(l) Section through a gall 40 DAI, of which the last 37 days on cytochalasin D-containing medium.

(m) Infected root with root knot nematode 40 DAI, of which the last 37 days on cytochalasin D-containing medium.

(n) Section through a gall 40 DAI.

*Giant cell; e, eggs; em, egg mass; G, gall; n, nematode; NC, neighboring cells; S, syncytium. Bars = 50 µm (b–e,g,i,k), 100 µm (f,h,j,l,n), and 150 µm (a,m).

Taxol treatment of seedlings, early after infection with the root knot nematodes, arrested gall development (Figures 5 and 6h). GFP analysis of galls confirmed the stabilization of MT in young giant cells (Figure 6i), which often had a less dense cytoplasm (Figure 6j) than untreated galls (Figure 6n). GFP analysis of cytochalasin D-treated galls revealed a diffuse fluorescence resulting from the depolymerized ACT cytoskeleton (Figure 6k). During cytochalasin D treatment, several females matured and produced fertile eggs although the galls were smaller (Figure 6m). Sectioned galls showed that giant cells were indeed small and contained a smooth and weakly stained cytoplasm (Figure 6l). Untreated galls were larger (Figure 6n) than those treated with taxol or cytochalasin D.

When plantlets infected with nematodes were transferred to taxol at later time points (14 DAI), the size of the feeding sites was similar to that of untreated ones. Although juveniles could molt and become females (or males), their maturation was delayed or they were unable to complete their life cycle (Figure 5). In contrast, treatment with cytochalasin D from 14 DAI on resulted in normal galls and syncytia, and almost normal numbers of mature females with fertile eggs (Figure 5). During taxol or cytochalasin D treatments, nematodes did not stain with potassium permanganate, and feeding sites showed no sign of senescence, demonstrating that nematodes were alive. Eggs within egg masses and within cysts collected from infected treated plants hatched normally into infective stage-2 juveniles.

Discussion

The plant cytoskeleton plays an active role in directing changes in cell morphology and architecture (Cyr, 1994). The cytoskeleton is expected to be reorganized in giant cells and syncytia induced by two types of plant parasitic nematodes. Surprisingly, in plant cells infected by nematodes, the cytoskeleton was disorganized and partially depolymerized.

Increased promoter activity of the ACT2 and ACT7 genes as well as disturbed organization of the ACT cytoskeleton are observed in feeding cells

Both ACT2 and ACT7 genes are transcriptionally active throughout gall and syncytium development. Both promoters, but especially ACT7, respond to wounding, hormones (McDowell et al., 1996a) and cell proliferation (Kandasamy et al., 2001; McDowell et al., 1996b). Nematodes perforate root cells to feed, feeding cell development involves increased hormone levels (reviewed by Gheysen and Fenoll, 2002) and mitosis occurs during feeding site formation. A homolog of the Arabidopsis ACT7 gene is also induced during a compatible plant–fungus interaction (Jin et al., 1999). Therefore, plant cellular responses promoted by external stimuli, such as fungi or nematodes, might result in altered expression of ACT mRNAs.

The presence of an intense GFP-mTn fluorescence in initiating feeding sites indicates a high concentration of ACT protein in giant cells and syncytia. This observation is supported by data on promoter activity. In uninfected cells and cells neighboring the feeding sites, a clear F-actin organization was observed, and between these filaments no GFP fluorescence was detected. In contrast, during syncytium development, no filamentous organization, but rather a diffuse signal throughout the feeding cell was observed, indicating that the ACT cytoskeleton may undergo depolymerization. Although no F-actin can be seen, it cannot be excluded that because of the large amount of ACT protein in the dense cytoplasm, a remaining F-actin cytoskeleton may be obscured.

By immunostaining of giant cells at different developmental stages, randomly distributed F-actin was seen within an amorphous label in the cytoplasm. Abnormally thick ACT cables were longitudinally and transversely oriented mainly in the cell cortex. These data suggest that the ACT cytoskeleton in giant cells is only partially fragmented or depolymerized in contrast to syncytial cells that have no remaining filaments.

High transcriptional activity of tubulin genes as well as rearrangements of the microtubular cytoskeleton are observed in feeding cells

Messenger RNA levels of α-, β-, and γ-tubulin genes are higher in galls than in syncytia. Giant cells undergo several cycles of mitoses and for each mitotic event, a cytoskeleton must be assembled. High expression levels of α-, β-, and γ-tubulins in cells surrounding the giant cells might result from the high mitotic activity observed in galls. The reduced transcript levels (compared to galls) in syncytia confirm the idea of a lack of mitotic activity in this type of feeding cell. However, strong signals were also seen in cells that surround syncytia, which undergo division before being incorporated into the feeding cell. Transcriptional activation of the α-tubulin gene has been reported in other biotic interactions, such as ecto- and endomycorrhizas (Bonfante et al., 1996; Carnero Diaz et al., 1996).

By in vivo and immunocytochemical visualization of MT, an amorphous tubulin fluorescence was observed in giant cells and syncytial cytoplasm, suggesting a possible depolymerization of the MT cytoskeleton. These results question the need for the high transcription activity of tubulin genes. One possibility is that the new tubulins induced by nematodes are isotopes different from those found in uninfected root cells, which might change the MT structure and therefore result in an unstable cytoskeleton.

During the early phases of syncytium development, the spatial MT organization changed gradually and a different MT orientation was seen within each syncytium domain from mostly oblique to randomly oriented cortical arrays. The strong and diffuse fluorescence in the cytoplasm indicates that new microtubular proteins are synthesized and/or that MTs may be broken down. The reason why cells neighboring the syncytia have a normal microtubular organization may be that a syncytium behaves as an isolated symplastic domain with no functional plasmodesmata (Böckenhoff et al., 1996; Golinowski et al., 1996). Therefore, signals that cause MT disorganization in syncytia may not diffuse to neighboring cells until they fuse with the syncytial cell.

During giant cell development, the amorphous cytoplasmic MT cytoskeleton may result from an accumulation of newly synthesized monomers and/or the depolymerization of the existing cytoskeleton. Alterations in the MT arrays may affect or prevent the formation of a functional PPB, but it cannot be excluded that the PPB may be masked by the bright tubulin stain. Nevertheless, spindles and phragmoplasts were observed in giant cells, although in some mitotic events, their typical shape was disturbed when compared to mitotic neighboring cells. Giant cells are unable to accomplish cytokinesis, and their phragmoplasts seem not to further develop. Completion of cytokinesis requires the interaction of fusing secretory vesicles, dynamic MTs, ACT, and their associated proteins (Sylvester, 2000). Cell plate vesicles align in giant cells but fail to fuse and disperse, and cell division is arrested (Jones and Payne, 1978). The lack of correct guidance of the phragmoplast to its division site, resulting in arrest of cytokinesis in giant cells, may be the consequence of disrupted MT and ACT cytoskeletons.

Nematodes are phytopathogens that induce long-term cytoskeleton rearrangements

Nematodes are the first pathogens described that induce long-term rearrangements and fragmentation of the cytoskeleton during the infection process. We observed that the cytoskeleton is disturbed in the two types of feeding cells (giant cells and syncytia), and therefore this may be a common step in nematode feeding cell initiation and/or expansion.

During symbiotic interactions, a rapid and transient depolymerization of the cytoskeleton is essential for rhizobia penetration into root hairs and for nodule formation (Cárdenas et al., 1998; Timmers et al., 1998, 1999). Remodeling of the cytoskeleton occurs also in mycorrhizal associations (Genre and Bonfante, 1998; Uetake et al., 1997).

Among the earliest responses of plant cells to pathogen attack is the reorganization of the cytoskeleton (reviewed by Baluska et al., 2000). Furthermore, upon treatment with drugs that depolymerize the cytoskeleton, non-host plants become susceptible to fungal infection possibly by inhibition of defense responses (Kobayashi et al., 1997; Takemoto et al., 1999). Therefore, relaxation of the cytoskeleton as observed in nematode feeding sites may open the way for a successful infection.

To investigate whether cytoskeleton changes are a prerequisite for nematode infection, we analyzed their life cycle in seedlings treated with cytochalasin D (that destabilizes ACT) or taxol (that stabilizes MT). Prolonged treatments were carefully interpreted. Nematodes are protected by a strong and impermeable cuticle, which acts as a barrier against penetration of chemicals (Cox et al., 1984; Reddigari et al., 1986). Under the conditions tested, chemically treated juveniles were still infective, could molt, were able to differentiate into fertile females (or males) and were capable of completing their life cycle. Histological analysis shows that neither giant cells nor syncytia present any sign of deterioration as observed when nematodes are dead. In addition, our previous results have shown that short or long incubations with other cytoskeleton inhibitors do not affect nematode viability (de Almeida et al., 1999). Therefore, we believe that the inhibitors applied here have no direct effect on the nematode.

Our in vivo analyses confirmed that the inhibitors disturbed the organization of the cytoskeleton. By preventing cytoskeleton breakdown with taxol, young feeding sites fail to expand. Giant cells and syncytia contained less cytoplasm, indicating that a lack of nutrient supply might inhibit the juvenile's development with a low percentage of mature females as a result. When infected seedlings were transferred to taxol after feeding site formation (14 DAI), feeding cells expanded but few nematodes completed their life cycle. These observations suggest that stabilization of the plant microtubular cytoskeleton interferes with nematode maturation.

Breakdown of the ACT cytoskeleton by cytochalasin D treatment at early stages of infection resulted in full arrest of gall development. Syncytium development was less affected. This difference between both types of feeding sites may be explained by the fact that untreated giant cells have a more F-actin cytoskeleton than syncytia. On the other hand, at later developmental stages, when the ACT or MT (by oryzalin treatment; de Almeida et al., 1999) cytoskeletons were chemically fragmented, both root knot and cyst nematodes were able to complete their life cycle, in contrast to results seen when the MT cytoskeleton was stabilized.

MT and ACT cytoskeleton reorganization and depolymerization possibly facilitate nematode feeding

Our data suggest that nematode infection triggers changes in both MT and ACT cytoskeleton architecture in both types of feeding cells, which persist throughout the complete nematode's life cycle (up to 40 DAI). Transcriptional activation of ACT and tubulin genes suggests that more globular ACTs and tubulins may be needed in the large feeding cells. In addition, tubulin and ACT proteins may also serve as nutrient source for nematodes. Both the ACT and tubulin cytoskeletons seem disrupted in syncytia. In contrast, in giant cells, a functional mitotic apparatus and an ACT and cortical tubulin cytokeleton are still present but disturbed. Although it is unclear how these cytoskeletal changes in giant cells and in syncytia occur, they may be triggered by depolymerizing existing F-actin and MTs, and/or by preventing G-actin or tubulin monomers to polymerize.

Our observations suggest that both ACT and MT cytoskeletons are concomitantly affected in feeding cells. Indeed, MF and MT interact with each other structurally and functionally and are regulated by common mechanisms (reviewed by Goode et al., 2000). In plant cells, actin and tubulin co-localize, and their functional interactions have been reported (Collings et al., 1996).

The depolymerization of the MT and F-actin cytoskeletons after nematode infection indicates that these structural changes of the cytoskeleton may be caused by a direct response to yet not identified molecules injected by the nematode in feeding cells. Disruption of MTs may cause depolarized cell growth (reviewed by Kost et al., 1999) as observed in giant cells induced by nematodes. Moreover, because MTs are known to contribute to cytoplasm viscosity (Gross et al., 1993; Virgin, 1953), depolymerization of MT and also of F-actin may create a decrease in cytoplasm viscosity. Nematode feeding involves the retrieval of large cytoplasm volumes (Müller et al., 1981), and a degree of cytoskeleton fragmentation may facilitate the suction and ingestion of cytoplasm during the feeding process of nematodes. Stabilization of the MT cytoskeleton at later stages results in normally developed feeding sites, but leads to improper maturation of the infecting juveniles, possibly by perturbing nematode feeding. On the other hand, drug-induced depolymerization of the ACT or MTs (de Almeida et al., 1999) leads to almost normal maturation of the infecting nematode, which suggests that the breakdown of the cytoskeleton is probably a prerequisite to allow completion of the nematode's life cycle. In the future, it will be interesting to investigate whether nematode secretions contain proteins that could interact with ACT filaments or MTs and be responsible for the controlled breakdown of the plant cytoskeleton.

Experimental procedures

Plant growth conditions and nematode infections

Surface-sterilized seeds of A. thaliana (L.) Heyhn. C24 and transgenic plants harboring the ACT2- and ACT7-promoter–GUS fusions were germinated, grown, and infected as described by de Almeida Engler et al. (1999). Infected roots were harvested after 1, 3, 5, 10, 15, and 20 DAI with either the root knot nematode M. incognita or the cyst nematode H. schachtii for GUS assays and for paraffin or methacrylate embedding.

Histochemical GUS assays

Transgenic plants carrying the two different constructs, ACT2–GUS (An et al., 1996) and ACT7–GUS (McDowell et al., 1996a) were kind gifts of Richard Meagher (University of Georgia). Infected seedlings were followed under a stereomicroscope and GUS staining was performed at different time points starting from the first hour after inoculation, up to 20 DAI. GUS activity was monitored as described previously by de Almeida Engler et al. (1999). To prevent diffusion of the GUS precipitate, samples were fixed in 2.5% glutaraldehyde prior to the transfer to clearing solution (Beeckman and Engler, 1994). GUS staining of galls and syncytia was microscopically analyzed with Nomarski optics.

mRNA in situ hybridization

Galls and syncytia were dissected after 1, 5, 10, 15, 20, and 25 DAI and placed in fresh fixative (2.5% glutaraldehyde). Fixed feeding sites were dehydrated, embedded in paraffin, and sectioned.

Full-length α- and γ-tubulin cDNAs (Liu et al., 1994; Ludwig et al., 1987) and a genomic clone of β-tubulin (Oppenheimer et al., 1988), all kindly given by P. Snustad (University of Minnesota), were cloned in a pBluescript II KS vector (Promega, Madison, WI, USA) and used to generate 35S-labeled sense and antisense probes. Considering that α-, β-, and γ-tubulins are gene families, these probes will cross-hybridize with other members of their family. Probe synthesis, pre-hybridization, and hybridization steps were carried out according to de Almeida Engler et al. (2001). Images were taken with a digital Axiocam (Zeiss, Jena, Germany) with standard dark- and bright-field optics. Overlays were performed with adobe photoshop 5.

Immunostaining and microscopy of actins and tubulins

Infected roots of Arabidopsis cv. Columbia and Pisum sativum (pea) were fixed in 4% formaldehyde in 50 mm Pipes buffer (pH 6.9). Feeding sites were dehydrated and embedded in butyl-methylmethacrylate as described by Kronenberger et al. (1993).

Sectioned feeding sites were treated with acetone for 20 min to remove the plastic. Primary and secondary antibodies were diluted 100- and 200-fold, respectively, in blocking solution (2% bovine serum albumin in 50 mm Pipes, pH 6.9). Sections were incubated with blocking solution for 20 min and then for 2 h at 37°C with the following primary antibodies: monoclonal mouse anti-ACT clone C4 (ICN, Irvine, CA, USA); a mixture (1 : 1) of the monoclonal anti-α-tubulin clone DM 1A (Sigma-Aldrich, St Louis, MO, USA) and anti-β-tubulin clone TUB 2.1 (Sigma-Aldrich); and monoclonal anti-γ-tubulin clone GTU-88 (Sigma-Aldrich) with the polyclonal anti-γ-tubulin, goat IgG1 isotype (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or with a polyclonal anti-γ-tubulin of Arabidopsis (kindly provided by Nelson Madeira, Ghent University). As controls, primary antibodies were omitted in some slides. Slides were washed twice in Pipes buffer and once with blocking solution. A 2-h incubation at 37°C was performed with the following secondary antibodies: Alexa 594 goat antimouse IgG and Alexa 488 goat antimouse IgG (Molecular Probes, Eugene, OR, USA) to visualize ACTs and tubulins, respectively. DNA was stained with 1 µg ml−1 4′,6-diamidino-2-phenylindole (Sigma-Aldrich) in Pipes buffer, and rinsed twice briefly in deionized water to remove salts. Slides were mounted in ProLong Antifade Kit (Molecular Probes), samples were observed with a microscope (Axioskop; Zeiss, Jena, Germany) equipped for epifluorescence microscopy, and images were collected with a digital camera (Axiocam; Zeiss).

GFP-MAP4 and GFP-mTn fusions and confocal analysis of feeding sites

A construct containing the 35S promoter fused to the N-terminal domain of the MT-binding domain of MAP4 to the GFP was kindly given by Richard Cyr (Pennsylvania State University). Because the 35S promoter is downregulated in nematode feeding cells, we also replaced the 35S promoter with the AtWRKY23 promoter, which is upregulated in nematode feeding sites (M. Karimi, unpublished data). Both 35S- and AtWRKY23-GFP-MBD constructs were introduced into Arabidopsis cultivar C24 via Agrobacterium tumefaciens transformation. Stably expressing lines were obtained, and homozygous progeny was used for nematode infection.

To visualize the ACT cytoskeleton in nematode feeding sites, we analyzed Arabidopsis plants that produced the chimeric GFP fused to the talin ACT-binding domain (GFP-mTn) under control of the 35S promoter (Kost et al., 1998).

For in vivo observations, sterile transgenic seeds were germinated in Gamborg B5 medium containing 2% sucrose and 0.3% phytagel poured in a Chambered Coverglass System (Laboratory-Tek, Christchurch, New Zealand). After germination, chambered dishes were kept inclined for roots to grow at the bottom, thus enabling imaging of infection sites. Freshly hatched juveniles were inoculated, and infected seedlings were observed under a stereomicroscope to identify initiating feeding sites. As nematodes must penetrate and migrate into the root until they reach a suitable feeding site, the time between root invasion and feeding site establishment may be 1 h to several days. Development of feeding cells was followed (starting from 1 h after inoculation) with a confocal laser scanning microscope (LSM 510; Zeiss) equipped with standard argon, UV, and He–Ne lasers. For a three-dimensional view, serial images collected at different focal planes were processed by simple stacking of selected images.

Taxol and cytochalasin D treatments

To test the possible toxicity of the cytoskeleton inhibitors, hatched juveniles of root knot and cyst nematodes were incubated for 3 days in medium with high to low concentrations of the inhibitors taxol (20–1 µm) and cytochalasin D (2.0–0.5 µm). The juveniles were then used to inoculate Arabidopsis seedlings. Nematodes were observed with Nomarski optics. Another fraction of the nematodes was stained with 0.5% KMnO4 to check nematode mortality (Jatala, 1975): dead nematodes stain amber to deep brown, whereas living ones do not. Arabidopsis seedlings were kept for 24 h in medium containing taxol and cytochalasin D at high concentrations and inoculated with nematodes to test whether they were able to infect roots under these conditions. Immunostaining of taxol- and cytochalasin D-treated feeding sites and GFP analysis of 35S-GFP-MBD, AtWRKY23-GFP-MBD and GFP-mTn confirmed the effectiveness of the chemical treatments.

Seedlings were pre-treated on medium with high concentration of the inhibitors for 2 or 24 h, infected with nematodes, and transferred to low inhibitor concentration for up to 40 DAI (time sufficient for nematodes to complete their cycle). As controls, similar experiments were performed, except that, after 24 h of inhibitor treatment, infected seedlings were kept for 40 days in medium without inhibitors. For the two other treatments, seedlings were transferred to medium with the inhibitors 3 and 14 DAI, kept for 24 h at high concentrations, and then transferred to medium containing low inhibitor concentrations for up to 40 DAI inoculation. Untreated seedlings were used as controls. After long treatments, nematodes within infected roots were stained with 0.5% KMnO4 to check viability. To control whether inhibitors affect nematode molting and development into fertile females, nematode growth was followed under a stereomicroscope up to 40 DAI. For cytological analysis, sections were double stained with Alcian Green Safranin and mounted with Depex (Sigma-Aldrich). Microscopy was performed with bright-field optics.

Acknowledgements

The authors thank Richard Meagher for the ACT2 and ACT7 transgenic seeds, Peter Snustad for the α-, β-, and γ-tubulin clones, Richard Cyr for the 35S-MDB-GFP construct, Nam-Hai Chua for the GFP-mTn construct, Patrick Hussey for recommending the anti-ACT antibody, undergraduate student Sven Wuyts with GUS assays, and Martine De Cock for help in preparing the manuscript.

Ancillary