•Soybean cyst nematodes (Heterodera glycines) produce secreted effector proteins that function as peptide mimics of plant CLAVATA3/ESR (CLE)-like peptides probably involved in the developmental reprogramming of root cells to form specialized feeding cells called syncytia.
•The site of action and mechanism of delivery of CLE effectors to host plant cells by the nematode, however, have not been established. In this study, immunologic, genetic and biochemical approaches were used to reveal the localization and site of action of H. glycines-secreted CLE proteins in planta.
•We present evidence indicating that the nematode CLE propeptides are delivered to the cytoplasm of syncytial cells, but ultimately function in the apoplast, consistent with their proposed role as ligand mimics of plant CLE peptides. We determined that the nematode 12-amino-acid CLE motif peptide is not sufficient for biological activity in vivo, pointing to an important role for sequences upstream of the CLE motif in function.
•Genetic and biochemical analysis confirmed the requirement of the variable domain in planta for host-specific recognition and revealed a novel role in trafficking cytoplasmically delivered CLEs to the apoplast in order to function as ligand mimics.
Heterodera glycines (soybean cyst nematode) are microscopic roundworms that enter plant roots as motile second-stage juveniles and subsequently feed, swell and molt three additional times (J3-adult) as sedentary obligate endoparasites. This evolutionarily advanced form of plant parasitism involves the modulation of host cell developmental pathways to form an enlarged, multinucleate feeding site, called a syncytium, which serves as the permanent food source for the sedentary parasitic stages of this obligate biotroph. After migrating into the roots of host plants, H. glycines selects a pericycle or an endodermal cell near the vasculature to initiate a syncytium (Hussey & Grundler, 1998). Current hypotheses suggest that the nematode uses its hollow, protrusible stylet for ingestion and to deliver esophageal gland cell-expressed effector proteins into the selected root cell (Davis et al., 2008). Syncytia are then formed through progressive cell wall dissolution and the coalescence of neighboring cell protoplasm (Endo, 1964). As cellular contents are incorporated into the syncytium, signaling reprograms the syncytial protoplasm to take on the characteristics of meristematic cells – dense cytoplasm, small vacuoles, prominent nuclei, high metabolic rate and persistence in a dedifferentiated state until nematode feeding stops (Hussey & Grundler, 1998). Nematode secreted effector proteins are believed to provide the signals required for the formation and maintenance of the syncytium (Davis et al., 2008).
Cyst nematodes are unique phytopathogens that produce proteins with functional similarity to plant CLAVATA3/ESR (CLE)-like proteins in their dorsal esophageal gland cell. These nematode CLE proteins presumably function as secreted effector proteins. We have shown previously that the overexpression of an H. glycines CLE gene (Wang et al., 2005; Davis, 2009) and, more recently, Globodera rostochiensis (potato cyst nematode) CLE genes (Lu et al., 2009) causes phenotypes similar to that of plant CLE gene overexpression in Arabidopsis (Arabidopsis thaliana). Furthermore, nematode CLE overexpression can complement the phenotype of an Arabidopsis clavata3 (clv3) mutant (Wang et al., 2005). These data confirm that nematode CLEs function as molecular ligand mimics of endogenous plant CLE peptides, most probably to facilitate the formation and/or maintenance of the syncytium during a successful parasitic association within host roots (Mitchum et al., 2008). Consistent with this hypothesis, dsRNA soaking of cyst nematodes and ingestion of host-derived RNAi to cyst nematode CLE genes have been shown to decrease nematode parasitic success (Bakhetia et al., 2007; Patel et al., 2008).
CLE proteins, identified from both monocot and dicot plant species (Cock & McCormick, 2001; Oelkers et al., 2008), play important roles in growth and development, including the maintenance of the stem cell pools in shoot, floral and root meristems (Jun et al., 2008). Plant CLEs belong to gene families (Jun et al., 2008) encoding small proteins with N-terminal signal peptides (SPs), diverse variable domains (VDs) and either a single or multiple conserved C-terminal CLE domain(s) (Cock & McCormick, 2001; Kinoshita et al., 2007; Oelkers et al., 2008). The N-terminal SP targets plant CLEs through the secretory pathway to function in the extracellular space (Rojo et al., 2002). The role of the VD, however, remains unclear, and is thought to be dispensable for plant CLE function (Fiers et al., 2006; Ni & Clark, 2006), whereas the conserved CLE domain is required for plant CLE function (Fiers et al., 2006). Recent studies have indicated that proteolytic processing of full-length CLE proteins into a 12- or 13-amino-acid CLE motif peptide is crucial for biological activity (Ohyama et al., 2008, 2009), but the nature and regulation of this processing are not yet understood. In addition, plant CLE peptides are post-translationally modified by hydroxylation and glycosylation (Ohyama et al., 2009). The modified CLE peptides serve as ligands for receptor binding to mediate plant signaling. Direct physical interactions between Arabidopsis CLE peptides and the extracellular leucine-rich repeat (LRR) domains of the corresponding receptors have recently been demonstrated (Hirakawa et al., 2008; Ogawa et al., 2008).
The best-studied Arabidopsis CLE protein, CLV3, functions to control the balance between shoot apical meristem (SAM) cell proliferation and differentiation (Fletcher et al., 1999). Overexpression of CLV3 causes constitutive repression of the WUSCHEL (WUS) gene (Brand et al., 2000), which encodes a putative homeodomain transcription factor that promotes stem cell formation and maintenance (Laux et al., 1996). Suppression of WUS leads to the differentiation of stem cells and premature termination of shoot and floral meristems, similar to the wuschel (wus) mutant phenotype (Laux et al., 1996). clv3 mutants have enlarged shoot and floral meristems because of the uncontrolled proliferation of stem cells as a result of an expanded WUS expression domain (Clark et al., 1995; Fletcher et al., 1999). Overexpression of a number of Arabidopsis CLE genes has been shown to result in wus phenotypes similar to CLV3 overexpression, indicating functional conservation among CLE family members when expressed in the appropriate temporal and spatial context (Strabala et al., 2006).
Two different H. glycines cDNA sequences coding for CLE-like proteins have been reported, both of which are exclusively expressed in the dorsal esophageal gland cell of H. glycines. The first reported sequence was cDNA clone HgSYV46 (Wang et al., 2001), which is identical to cDNA clone 2B10 identified in a separate study (Gao et al., 2003). The second reported sequence was cDNA clone 4G12 (Gao et al., 2003). The 4G12 sequence was subsequently characterized (Wang et al., 2005), but mistakenly reported as the HgSYV46 sequence (see Corrigendum, Davis, 2009). From this point on, we refer to these two nematode CLE sequences simply as HgCLE1 (formerly HgSYV46 and 2B10) and HgCLE2 (formerly 4G12). The study presented here details the cellular targeting and functional domains of the HgCLE proteins in a comparative analysis. We present evidence to indicate that nematode CLE proteins are delivered to the cytoplasm of syncytial cells, but function in the extracellular space, consistent with their proposed role as ligand mimics of plant CLE peptides. Our results also demonstrate that both the VD and CLE motifs are required for their function in planta. Furthermore, we determined that, unlike plant CLEs, the nematode CLE VD functions in the host-specific recognition of nematode CLE proteins and is sufficient to traffic cytoplasmically delivered CLEs to the apoplast.
Materials and Methods
Nematode and plant material
Soybean cyst nematode (Heterodera glycines Ichinohe, 1952) inbred lines PA3 (HG type 0) and TN19 (HG type 1–7) were maintained on soybean [Glycine max (L.) Merr.]. Nematode eggs were extracted and hatched according to Wang et al. (2007). Parasitic life stages of the soybean cyst nematode were prepared as described previously (Goellner et al., 2000). Arabidopsis thaliana (L.) Heynh. ecotype Columbia (Col-0) was used in this study. The clv3-2 mutant (Fletcher et al., 1999) used in complementation studies was obtained from the ABRC stock center and Landsberg erecta (Ler) was used for comparison. Arabidopsis was transformed using the floral dip method (Clough & Bent, 1998). Soybean (G. max) cv. Williams 82 (W82) was used in this study.
Genomic DNA analysis
Nematode genomic DNA was extracted as described in Patel et al. (2009). HgCLE genomic sequences were amplified using primers corresponding to the untranslated region (UTR) sequences of the HgCLE1 cDNA and cloned into the pCR®4-TOPO vector (Invitrogen) for sequencing.
RNA isolation and quantitative real-time reverse transcription-polymerase chain reaction (RT-PCR)
Nematode and soybean mRNA were extracted using the Dynabeads® mRNA DIRECT™ Micro Kit (Invitrogen). Arabidopsis total RNA was extracted with the RNeasy® Plant Mini Kit (Qiagen). Contaminating DNA was removed using the DNA-free™ Kit (Ambion). cDNA was prepared using the First Strand cDNA Synthesis Kit for RT-PCR (Roche). Quantitative real-time PCR (qPCR) was carried out as described previously (Wang et al., 2007). The H. glycines GAPDH gene, Arabidopsis GAPDH gene and soybean SUBI-3 gene were used as internal controls, and the data were analyzed as described previously (Wang et al., 2007). For the nematode developmental expression study, a second set of experiments was conducted on an independent set of nematode life stages and the same pattern was observed.
Binary vector pMD1 (Tai et al., 1999) was used to generate cauliflower mosaic virus (CaMV) 35S overexpression constructs. To generate 2X35S overexpression constructs, genes were first cloned into pCGT-35S behind a double 35S promoter and then subcloned into binary vector pAKK1467B which has both green fluorescent protein (GFP) and BASTA selectable markers. pTA7002 (Aoyama & Chua, 1997) was use to create dexamethasone (DEX)-inducible constructs. The megaprimer PCR method (Sarkar & Sommer, 1990) was used for site-directed mutagenesis, and fusion PCR was used to generate constructs carrying chimeric sequences from different genes. Plasmids and oligonucleotides are described in Supporting Information Table S2.
Seedlings showing SAM termination were characterized as the severe wus phenotype. Seedlings with normal SAM development, but exhibiting defects in floral meristem development (no carpels and decreased stamen number) at later stages, were characterized as the weak wus phenotype. For the analysis of root growth, 6-d-old GFP-positive T1 transformants were identified using a Leica fluorescent stereoscope, transferred to square plates and grown vertically.
Transgenic hairy roots
Soybean hairy roots were generated from cotyledons according to Wang et al. (2007), except that 238 μg ml−1 timentin (GlaxoSmithKline, Research Triangle Park, NC, USA) was used instead of carbenicillin. DEX (1 μM) (Sigma) was used for transgene induction. Root length was measured for 7 d before and after induction.
The affinity-purified anti-HgCLE peptide antibody was raised in rabbit against the synthetic peptide RLSPSGPDPHHH by Bethyl Labs, Inc. (Montgomery, TX, USA). For the immunofluorescence and immunolocalization experiments, Alexafluor 488 and 568 goat anti-rabbit secondary antibodies were used (Molecular Probes, Eugene, OR, USA). For Western blots, goat anti-rabbit alkaline phosphatase-conjugated secondary antibodies were used (Sigma-Aldrich).
Total protein was extracted from H. glycines PA3 nematodes 7 d post-inoculation (dpi) as described previously (Goellner et al., 2000). Samples were separated on 15% tris-tricine gels (Schagger & von Jagow, 1987). Gels were blotted to Immun-blot 0.2 μm polyvinylidene difluoride (PVDF) membranes with alkaline blotting as described previously (Cruz-Garcia et al., 2005). Membranes were immunostained with anti-HgCLE peptide antibody at a 1 : 5000 dilution and developed as described previously (Murfett et al., 1996).
In planta immunolocalization
Soybean roots were infected and grown as described previously (Ithal et al., 2007a). Roots were taken down at 5 dpi, infected root tissue sections (c. 1 cm) were cut from the seedling and fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.2, for 3 h under vacuum. Root sections were processed as described previously (Goellner et al., 2001).
Enzyme-linked immunosorbent assay (ELISA)
Secreted peptides were extracted from Arabidopsis whole-plant submerged cultures according to Ohyama et al. (2008), except that the purified peptides were dissolved in 1.0% formic acid instead of 1.0% trifluoroacetic acid and diluted 40 times for ELISA. More details are given in Methods S1.
Immunofluorescence images of nematode specimens were acquired with an Olympus IX 70 inverted microscope equipped with DIC/Nomarski optics and a × 20/0.7 UPlanApo objective lens. Immunofluorescence images of soybean root sections were acquired with an LSM 510 META NLO confocal system mounted on an Axiovert 200M inverted microscope frame equipped with DIC/Nomarski optics and a × 20/0.75 Plan-Apochromat objective lens (Zeiss).
The fragment-based modeling software ROSETTA (Bonneau et al., 2002) was used to model nematode and plant CLE VDs. ROSETTA employs a fragment-based modeling method that can be applied when low sequence identity between a target sequence and any template structure prevents the use of a homology modeling approach (Bonneau et al., 2002). It has been shown to successfully predict structures of small proteins and protein domains. For each target sequence of VD, an ensemble of c. 10 000 solutions was obtained, and the best scoring solution was selected as the predicted structure. In addition, for each target sequence, the consensus secondary structure of 50 best scoring models was determined.
Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: H. glycines CLE1 mRNA (AF273728), H. glycines CLE1 genomic DNA (FJ503004), H. glycines CLE2 genomic DNA (FJ503005), H. glycines CLE2 mRNA (AF473827), H. glycines GAPDH (CA940100), G. max Ubiquitin (D28123), A. thaliana GAPDH (NM_111283), A. thaliana CLV3 gene (NM_128283).
HgCLE1 and HgCLE2 differ by 12 amino acids in the VD
An alignment of HgCLE1 and HgCLE2 amino acid sequences revealed identical SPs and identical CLE motifs (Fig. 1a). The 12-amino-acid polymorphisms identified between HgCLE1 and HgCLE2 reside in the VD (Fig. 1a). The genomic structures of HgCLE1 and HgCLE2 were determined and consist of four exons and three introns. The three introns delineate sequences coding for distinct domains within the HgCLE proteins. Intron 1 delineates SP from VD; intron 2 splits VD into a highly conserved VdI and a highly variable VdII; and intron 3 separates VD from the 12-amino-acid CLE motif (Fig. 1a).
qPCR using gene-specific primers revealed similar expression of HgCLE1 and HgCLE2 during nematode parasitism. Expression was low in nonfeeding life stages, including eggs, preparasitic second-stage juveniles (ppJ2) and adult males. By contrast, expression was strongly upregulated when nematodes entered the root, from the onset of syncytium formation by parasitic second-stage juveniles (pJ2) through the J3–J4 molts of sedentary life stages that become adult females (Fig. 1b), suggesting an important role for HgCLEs in syncytium induction and maintenance.
HgCLE propeptides are delivered to host cells
The HgCLE1 prepropeptide is predicted to be a 14.51-kDa, 139-amino-acid protein containing a putative 22-amino-acid N-terminal SP. The propeptide is presumably the secreted form of HgCLE1 and is predicted to be 12.07 kDa. Both forms of this protein are basic with isoelectric points of 10.84 and 10.60, respectively. The HgCLE2 prepropeptide is predicted to be a 14.46-kDa, 138-amino-acid protein containing a putative 22-amino-acid N-terminal SP. The propeptide is presumably the secreted form of HgCLE2 and is predicted to be 12.02 kDa. Both forms of this protein are basic with isoelectric points of 10.92 and 10.62, respectively. We generated and validated an affinity-purified anti-HgCLE peptide antibody that labeled proteins in the dorsal gland lobe and along the nematode effector secretion route (i.e. cellular extension and ampulla) of parasitic specimens (Fig. 2a). Immunoblots probed with the anti-HgCLE peptide antibody, but not with the pre-immune sera, specifically identified two protein bands in the predicted size range for HgCLEs from extracts of H. glycines parasitic stages at 7 dpi: a faint, higher molecular weight band and a dark, lower molecular weight band (arrowheads, Fig. 2b). The upper band may represent remnants of the HgCLE prepropeptides and the lower band the HgCLE propeptides available for secretion. The alkaline blotting conditions used, which have been shown to enhance the transfer of highly basic proteins (Cruz-Garcia et al., 2005), greatly enhanced the visualization of these bands in parasitic nematode extracts, indicating that the proteins identified in the 7 dpi extracts behave chemically like the alkaline HgCLE proteins. The slight difference in size with that predicted for HgCLEs may be an indication that the protein is post-translationally modified, which has been reported for plant CLE proteins (Kondo et al., 2006; Ohyama et al., 2009). Detection of the HgCLE propeptide from extracts of parasitic stages on immunoblots indicates that HgCLEs are unlikely to be processed further within the nematode.
In planta immunolocalization localizes HgCLEs in syncytia cytoplasm
To determine the site of secretion of HgCLEs by the nematode, we conducted immunolocalization studies on sectioned soybean roots infected with H. glycines at 5 dpi (Fig. 3). Figure 3(a–c)shows HgCLE localization in the cytoplasm of a syncytium. Staining with anti-HgCLE preimmune and anti-HgCLE peptide antibody in serial sections through the same syncytium (Fig. S1a–c, d–f) shows that the preimmune serum does not label syncytial contents, whereas the anti-HgCLE peptide antibody does label syncytial contents. No specific HgCLE labeling was observed in uninfected root tissue (Fig. S1g–i) or within the cells surrounding syncytia (Fig. S1d–f). We found direct evidence of dorsal gland labeling in nematodes associated with HgCLE-labeled syncytia (Fig. 3d–l). Furthermore, HgCLE labeling was observed in the ampulla of a nematode associated with an HgCLE-labeled syncytium (Fig. 3j–l). These data indicate that CLE proteins originating in the dorsal esophageal gland cell are probably delivered via the stylet directly to the host cell cytoplasm.
The HgCLE CLE motif is required but not sufficient for function in planta
Previous studies of plant CLEs have shown that the CLE motif is absolutely required for function, because, without the CLE peptide, there is no ligand to trigger signaling cascades (Fiers et al., 2006; Kondo et al., 2006). In a previous study, we have shown that the overexpression of full-length HgCLE2 in Arabidopsis causes above-ground wus phenotypes and short roots (Wang et al., 2005; Davis, 2009), demonstrating that Arabidopsis can be used as a tool to study nematode CLE protein activity in a nonhost. To test for a requirement of the nematode CLE motif, HgCLE2 was expressed in Arabidopsis without the CLE motif (Fig. 4). No obvious wus phenotypes were observed in transgenic plants overexpressing HgCLE2ΔCLE (Fig. 4). Like the plant CLEs (Fiers et al., 2006), deletion of the CLE motif abolishes nematode CLE function in planta, indicating that this motif is essential and most probably acts as a plant CLE ligand mimic.
To test whether the CLE motif is sufficient for function in planta, plants expressing the 12-amino-acid CLE motif in the cytoplasm or apoplast (by including the native nematode SP) were monitored for phenotypes (Fig. 4). No visible phenotypes were observed. These results demonstrate a necessary role for other regions of the HgCLE2 protein for function in planta.
HgCLE2 functions without SP in planta
In a previous study, the expression of HgCLE2 with the native nematode SP in Arabidopsis caused wus phenotypes similar to plant CLEs (Wang et al., 2005). Thus, it was assumed that the nematode SP was recognized in planta and targeted the nematode CLE peptide to the apoplast to function as a ligand mimic (Rojo et al., 2002). Here, we tested directly the functionality of the nematode SP in planta by replacing the SP of AtCLV3 with the HgCLE2 SP to generate AtCLV3HgCLESP (Fig. 5a). For comparison, constructs carrying AtCLV3, either with or without its SP (AtCLV3ΔSP) sequence, were included (Fig. 5a). Multiple independent primary Arabidopsis transformants were characterized. Ninety-five per cent of the primary transformants expressing AtCLV3 with its native SP displayed characteristic wus phenotypes (Laux et al., 1996), 93% of which were severe (Fig. 5b), consistent with previously published gain-of-function phenotypes for AtCLV3 (Brand et al., 2000; Strabala et al., 2006). As reported previously (Rojo et al., 2002), no phenotypes were observed in AtCLV3ΔSP overexpression lines (Fig. 5b), highlighting the extracellular function of plant CLEs. However, 90% of primary transformants expressing AtCLV3HgCLESP displayed wus phenotypes, 31% of which were severe and 59% were weak (Fig. 5b). From these results, we concluded that the HgCLE SP is functional in Arabidopsis and therefore could be used as a tool to target proteins to the apoplast for the studies of HgCLE protein activity described here. These data also validated our earlier hypothesis that HgCLE2 can function in the apoplast like plant CLEs.
Our immunolocalization data (Fig. 3) suggested that HgCLE proteins are delivered directly to the host cell cytoplasm by the nematode. Therefore, we reasoned that HgCLEs might also function in the cytoplasm of host plant cells. To test this, we generated transgenic Arabidopsis expressing HgCLE2 without the nematode SP (Fig. 5a). Transgenic Arabidopsis expressing the full-length HgCLE2 was included for comparison (Fig. 5a). Seventy-two per cent of the primary transformants expressing full-length HgCLE2 (with SP) exhibited wus phenotypes, including premature termination of the shoot and floral meristems (Fig. 5b), consistent with our previously published data (Wang et al., 2005; Davis, 2009). In addition to the above-ground wus phenotypes, we observed consumption of the primary root apical meristem leading to a short-root phenotype in HgCLE2-expressing plants (Fig. 5c,d). These phenotypes are consistent with gain-of-function phenotypes displayed in Arabidopsis overexpressing a subset of CLEs that cause dwarf wus-like above-ground phenotypes, and stunted root phenotypes (Strabala et al., 2006). Surprisingly, unlike plants expressing AtCLV3ΔSP, Arabidopsis primary transformants overexpressing HgCLE2 without SP also showed a phenotype: 85% displayed above-ground wus phenotypes (Fig. 5b) and 77% displayed short roots (Fig. 5c,d). This surprising result was confirmed by complementation. The AtCLV3 promoter (PCLV3) (Brand et al., 2002) was utilized to express HgCLE2 with or without SP in the appropriate temporal and spatial context in a clv3-2 (strong allele) mutant background (Fletcher et al., 1999). HgCLE2 with and without the SP both partially complemented the clv3-2 mutant phenotype (Fig. S2). Thus, both forms of HgCLE2 (with and without SP) function in CLE signaling in planta.
The finding that HgCLE2ΔSP is functional in Arabidopsis indicated that, unlike plant CLEs, nematode CLEs may function in both the cytoplasm and apoplast. Alternatively, the HgCLE2 VD might function as a noncanonical secretion signal, targeting CLEs to the apoplast. These possibilities were investigated further.
HgCLE1 does not function in Arabidopsis
In contrast with HgCLE2-expressing plants, no phenotypes were observed in plants expressing HgCLE1 or HgCLE1ΔSP (Fig. 5b,d). This observation is particularly remarkable when considering that HgCLE1 and HgCLE2 contain identical 12-amino-acid CLE motifs (Fig. 1a). qPCR confirmed that the expression levels of HgCLE1 were equal or greater than the expression level of HgCLE2 lines (Fig. S3), indicating that the HgCLE1 transgene is expressed to sufficient levels. The finding that HgCLE1 is nonfunctional in Arabidopsis indicated that VD might play a role in recognition specificity and was investigated further.
Structural modeling of the HgCLE VDs predicts differences from plant CLEs
The finding that HgCLE2, but not AtCLV3, can function in planta without SP, and the functional differences between HgCLE1 and HgCLE2 in Arabidopsis, led us to compare the VDs using computational models of their structures. Structural models of the VDs revealed that, although the nematode and plant CLE VDs consist exclusively of α-helical substructures (the consensus secondary structure includes five helices), VD of nematode CLEs can be clearly divided into two corresponding structural domains (Fig. 6a,b; VD models of HgCLE1 and HgCLE2 were similar; therefore, only HgCLE2 is shown). The two structural domains of HgCLE2 correspond to VdI (amino acids 23–90) and VdII (amino acids 91–126), consistent with the position of intron 2 (Fig. 1a), whereas the VD of AtCLV3 is likely to consist of only one (Fig. 6a,b). In addition, the VDs of the plant and nematode CLE proteins are likely to be structurally diverse. One difference is the presence of long loop regions between the helices in VdI of HgCLE2 (Fig. 6b). In addition, the sequence corresponding to VdII of HgCLE2 is predicted to consist of a 24-amino-acid helix which is absent in the predicted structure of AtCLV3 (Fig. 6a). These structural differences may reflect functional diversification of the nematode and plant CLE VDs.
VD is required for HgCLE function in the apoplast
To clarify whether HgCLE2 functions in the cytoplasm, or whether the VD is involved in targeting HgCLE2 to the apoplast to function, a deletion analysis was conducted. Overexpression of an HgCLE2 VdI deletion (HgCLE2Δ23–90; Fig. 7a) produced wus phenotypes in Arabidopsis (Fig. 7b). Thus, HgCLE2 VdII with the CLE motif is sufficient for HgCLE2 signaling when targeted to the apoplast by SP in Arabidopsis. However, deletion of both SP and VdI (HgCLE2Δ1–90; Fig. 7a) abolished HgCLE2 signaling (Fig. 7b). Therefore, cytoplasmic VdII with the CLE motif is not sufficient for CLE signaling. These results, together with the observed activity of HgCLE2ΔSP (Fig. 5), confirmed that the site of action of HgCLE2 is in the apoplast, not the cytoplasm, and the sequence in VdI is required for the targeting of HgCLE2 to the apoplast. To examine the role of VdII in processing of the CLE peptide in the apoplast, all but six amino acids of VdII preceding the CLE motif were deleted (HgCLE2Δ23–120; Fig. 7a). This partial deletion of VdII abolished the activity of the protein (Fig. 7b). Therefore, VdII is required for HgCLE2 function when targeted to the apoplast, but the six amino acids preceding the CLE motif are not sufficient for HgCLE2 peptide cleavage and signaling.
HgCLE2 VD is sufficient to target CLEs to the apoplast
It has been shown that AtCLV3 localization to the apoplast is required for its function (Rojo et al., 2002; Hirakawa et al., 2008). Consequently, the expression of AtCLV3ΔSP is nonfunctional in Arabidopsis (Rojo et al., 2002; Fig. 5a,b). Our deletion studies indicated that the HgCLE2 VD probably plays a role in targeting nematode CLEs from the cytoplasm to the apoplast of host cells. To test this possibility directly, the HgCLE2 VD was fused to the CLV3 CLE motif (HgCLE2VD-CLECLV3; Fig. 7a). Remarkably, 100% of Arabidopsis plants expressing HgCLE2VD-CLECLV3 displayed severe wus phenotypes (Fig. 7b). These data demonstrated that the VD of HgCLE2 can properly target the AtCLV3 CLE motif to the apoplast to activate the CLV signaling pathway.
As GFP-tagged HgCLE proteins were nonfunctional in planta, we could not reliably use transient subcellular localization assays as an independent approach to our genetic analyses to confirm VD targeting of CLE peptides to the extracellular space. Therefore, a biochemical approach was utilized to confirm the secretion of CLE peptides to the apoplast by the HgCLE2 VD. It has been reported that peptides secreted to the apoplast of Arabidopsis plants can diffuse into the medium of whole-plant submerged cultures (Ohyama et al., 2008). Therefore, CLE peptides extracted from the media of whole-plant submerged cultures of plants overexpressing HgCLE2ΔSP were quantified by an indirect ELISA using the anti-HgCLE peptide antibody. Whole-plant submerged cultures of plants overexpressing HgCLE2 were included as positive controls. Wild-type plants and several HgCLE2Δ1–90 (HgCLE2 without SP and VdI sequences; Fig. 7a) overexpression lines were included as negative controls. CLE peptide levels accumulated several hundred times higher in HgCLE2 and HgCLE2ΔSP overexpression lines relative to the negative controls (Fig. 7c). qPCR confirmed equal or higher transgene expression levels in the HgCLE2Δ1–90 lines relative to HgCLE2ΔSP lines (Fig. 7c). These results clearly demonstrated that HgCLE2 VD is sufficient for targeting the CLE peptide to the apoplast to function as a ligand mimic.
VD confers host-specific recognition specificity in planta
The observed lack of function of HgCLE1 in Arabidopsis implicated the VD in conferring recognition specificity in planta. To test this possibility, the Arabidopsis CLV3 VD was replaced with VD of either HgCLE1 or HgCLE2 (Fig. 8a). The constructs were transformed into Arabidopsis to generate multiple independent lines for phenotypic characterization. An Arabidopsis CLV3 overexpression construct minus the extension sequence (AtCLV3ΔEXT) was used for comparison (Fig. 8a). It has been shown that the CLV3 extension sequence is not required for CLV3 function (Fiers et al., 2006). Significantly, 100% of the transformants expressing AtCLV3HgCLE2VD exhibited severe wus phenotypes, similar to AtCLV3ΔEXT (Fig. 8a). By contrast, only 24% of the AtCLV3HgCLE1VD-expressing plants exhibited wus phenotypes and 23% of these were very weak (Fig. 8d). To further determine which amino acids within the H. glycines VD were required for recognition specificity, the five tandem amino acids differing between the VDs of HgCLE1 and HgCLE2 were swapped (Fig. 8b). Remarkably, swapping these amino acids abolished the in planta function of HgCLE2ΔSP, whereas it led to a gain of function for HgCLE1ΔSP (Fig. 8d). Site-directed mutagenesis was used to individually substitute HgCLE2 amino acids at positions 121–125 with the corresponding amino acid from HgCLE1 (Fig. 8c). All constructs were transformed into Arabidopsis for phenotypic characterization. Of the five amino acid substitutions, only S125P resulted in a significant reduction in the observed wus phenotypes (38%; Fig. 8d) when compared with the HgCLE2ΔSP construct (85%; Fig. 8b). Together, these data clearly demonstrated that the VD confers recognition specificity to nematode CLEs in planta and this specificity is mediated, in large part, by the amino acids immediately N-terminal of the CLE motif in VdII.
We reasoned that the observed lack of phenotypes for Arabidopsis expressing HgCLE1 might be related to the host-specific recognition of nematode CLE proteins via the VD. In general, cyst nematodes have evolved specific host preferences, with the host range of H. glycines limited to legumes and some winter annual weeds. Arabidopsis is not a host for H. glycines. What specifies host range and whether there is a genetic basis for this host specificity are not known. To test this possibility, DEX-inducible (Aoyama & Chua, 1997) HgCLE1 and HgCLE2 constructs were made with and without the native SP and transformed into soybean, a host for H. glycines, to generate transgenic hairy roots for phenotypic analysis. HgCLE1 and HgCLE2, with or without the SP, were functional, indicating that the VD can also traffic CLE peptides to the apoplast in soybean. A strong correlation was observed between HgCLE1 or HgCLE2 expression levels and root meristem consumption phenotypes in high-expressing lines (Fig. 9). Root growth reduction > 38% was never observed in any of the control or low-expressing lines (n =24). By contrast, root growth terminated in several HgCLE1 lines exhibiting high levels of expression following DEX treatment (Fig. 9a). Unlike in Arabidopsis, both HgCLE1 and HgCLE2 can function as ligand mimics in soybean. These data demonstrated that the recognition of specific nematode CLEs occurs in soybean, a host for H. glycines, but not in Arabidopsis, a nonhost plant for H. glycines.
Cyst nematodes produce secreted plant CLE peptide mimics probably involved in the redirection of CLE signaling pathways to form and maintain syncytial cells in host roots. HgCLEs are specifically and developmentally expressed in the dorsal gland of parasitic nematodes, and present in the nematode effector secretion route (i.e. cellular extension and ampulla) for delivery through the nematode stylet into host tissues (Wang et al., 2005; Fig. 2a). We have shown that H. glycines CLEs are highly upregulated in feeding life stages during a compatible interaction and expression is sustained through the adult female life stage (Fig. 1b), indicating that the continuous secretion of CLE proteins may be required to maintain the feeding cell in a dedifferentiated state.
Several lines of evidence suggest that the propeptide, not a processed CLE peptide, is delivered to host plant cells. For example, detection of the HgCLE propeptide on immunoblots of total protein isolated from parasitic life stages (Fig. 2b) suggests that the CLE proteins are unlikely to be processed in the nematode, other than SP cleavage. Although the processing mechanism of plant CLE peptides is still unclear, it appears to involve a complex process that may require specific sequence recognition. The observation of above-ground wus phenotypes and short roots in Arabidopsis and soybean overexpressing HgCLEs suggests that the nematode CLEs are being processed in the plant and indicates that the nematodes are able to utilize the processing machinery that is already in place in the host to produce a functional CLE peptide. It has been shown recently that the Pseudomonas syringae effector protein, AvrRPS4, which is delivered to the host cell cytoplasm, requires in planta processing to function (Sohn et al., 2009). Thus, it appears that pathogens representing diverse groups have evolved to rely on the use of cellular machinery of their hosts to promote disease.
Electron micrographs of root cells parasitized by the ring nematode, Criconemella, show that the plant cell wall is perforated by the nematode stylet which then invaginates the plasma membrane (PM) without breaching it (Hussey et al., 1992). However, at the stylet orifice, a small pore has been observed in PM (Hussey et al., 1992) and could simply allow for the delivery of effector proteins directly into the host cell cytoplasm (Hussey & Grundler, 1998). Technical constraints have made it difficult to determine whether nematodes specify delivery of effectors directly to the cytoplasm, or the apoplastic space; however, several lines of indirect evidence indicate that cyst nematode effector proteins are delivered to the cytoplasm of parasitized host cells. For example, some cyst nematode effectors have been inferred to enter plant cells where they are then targeted to host nuclei (Elling et al., 2007). Second, several cyst nematode effectors can exert their function when overexpressed in plants without conventional secretory leader sequences and have been shown to interact with nonsecretory host proteins (Hewezi et al., 2008; Patel et al., 2009; Rehman et al., 2009; Sacco et al., 2009). An example of this is the cyst nematode SPRYSEC protein, RBP-1, which has recently been shown to function intracellularly through the nucleotide binding site (NBS)-LRR protein GPA2, presumably after direct delivery to the host cell cytoplasm via the nematode stylet (Sacco et al., 2009). In addition, several other plant resistance genes to cyst nematodes encode NBS-LRR proteins with a predicted intracellular function, and it is plausible to assume that secreted nematode effectors are the requisite avirulence proteins (Williamson & Kumar, 2006).
Cyst nematode secreted effector proteins have been immunolocalized in planta during migration through root tissues (Wang et al., 1999; Goellner et al., 2001); however, detection of effector proteins within syncytial cells is lacking. Here, we show for the first time, by direct immunodetection, a nematode effector protein (CLE) within the cytoplasm of the syncytial feeding cell (Fig. 3). Although anti-HgCLE peptide antibody cross-reaction with native plant CLEs expressed in syncytial cells cannot be excluded, we believe that this is most likely not the case for several reasons. First, there is no specific anti-HgCLE labeling in uninfected root tissue sections or in cells surrounding syncytia (Fig. S1). We also provide evidence of in planta labeling of syncytia directly associated with nematodes expressing HgCLE in the dorsal gland and along the secretory route (Fig. 3g–l). Furthermore, there is no experimental evidence to date for the detection of plant CLEs in the cytoplasm of plant cells using immunodetection techniques. The only direct localization of a plant CLE was found in the apoplastic space of phloem cells (Hirakawa et al., 2008). It is presumed that plant CLEs are processed through normal secretory routes targeted by their SP, and would not accumulate in the cytoplasm. Finally, we searched the two most comprehensive microarray datasets available that describe the transcriptional profiles of syncytia in Arabidopsis (Szakasits et al., 2008) and soybean (Ithal et al., 2007b), and did not find any evidence of upregulation of plant CLE gene expression in cyst nematode feeding sites (Table S1). Taken together, we believe that we are detecting HgCLEs directly in the host syncytial cell cytoplasm.
Although Arabidopsis is not a host for H. glycines, our work demonstrated that the HgCLE2 protein can functionally mimic the Arabidopsis CLV3 peptide (Wang et al., 2005; this study). That is, regardless of whether or not Arabidopsis is a host, when the HgCLE2 protein is expressed in the same temporal and spatial pattern as CLV3, it is targeted, processed and perceived by receptors to produce the resultant wus phenotypes or restore CLV3 function in a mutant background. Similarly, many plant CLE family members can functionally replace CLV3 activity when expressed under the control of the CLV3 promoter, although this is not indicative of their true function in plant growth and development (Ni & Clark, 2006). Likewise, the wus phenotype does not necessarily reflect how the nematode gene product actually functions locally at the feeding site. However, the ability of the HgCLE2 protein to mimic CLV3 to produce wus phenotypes served as an excellent readout for protein activity that allowed us to determine the site of action of nematode CLEs in planta.
In this study, we determined that nematode CLEs function in the apoplast, consistent with a role in ligand mimicry of plant CLEs, but are first delivered directly to the cytoplasm of host cells. In other words, our data indicate that cyst nematodes do not need to deliver processed CLE peptides to the apoplast of the host cell in order to function as ligand mimics. Not all pathogen effectors are delivered directly to their functional destination. Therefore, a trafficking step is required before effectors can be recognized by host proteins (Panstruga & Dodds, 2009). Oomycete effectors delivered to the extrahaustorial matrix have been clearly shown to function in the host cytoplasm (Dou et al., 2008). Recent studies have shown that an RXLR-dEER domain at the N-terminus of the secreted proteins is necessary and sufficient for translocation from the apoplast to the cytoplasm in the absence of the pathogen, although the trafficking mechanism is not yet known (Dou et al., 2008). On the other hand, some secreted effectors reach the cytoplasm even though no RXLR-dEER domain or other conserved peptide motifs have been identified (Panstruga & Dodds, 2009). These data indicate that there are different mechanisms for trafficking effectors from the apoplast to the cytoplasm.
Here, we report a highly evolved novel effector protein trafficking mechanism utilized by a plant-parasitic nematode to deliver effector proteins secreted in host cell cytoplasm to the apoplast. Using a combination of genetic and biochemical approaches, we show that CLE proteins secreted by cyst nematodes can traffic from the cytoplasm to the apoplast of the plant cell in the absence of the pathogen, and that VD is required for this translocation (Fig. 7). Unlike plant CLEs, this function appears to be unique to nematode CLE proteins, but it is a function that would be a necessary evolution for ligand mimicry when nematode CLE proteins are delivered directly to the host cell cytoplasm. Our genetic analysis demonstrating that the expression of the CLE motif alone within the cytoplasm of host cells is not sufficient for activity in planta (Fig. 4) corroborates this result. The recent finding that plant CLEs are arabinosylated peptides and that these post-translational modifications enhance receptor binding (Ohyama et al., 2009) suggests that the delivery of nematode CLEs to the cytoplasm may in fact be necessary for the nematode CLEs to function as ligand mimics if further modification is required. Determining the mechanism of trafficking and whether this is unique to nematode CLE effector proteins, and whether there is a conserved motif or predicted secondary structure that is important for this trafficking, awaits further studies.
Significantly, the use of Arabidopsis also led us to the interesting finding that sequences in the VD also confer host-specific recognition of nematode CLE proteins in planta. The most likely possibility relates to an involvement of the VD in CLE peptide processing and/or receptor recognition. The identification of which factors determine the host range of pathogenic microbes is a continuing challenge in pathogenic microbiology, which relates to both the co-evolution of host susceptibility and pathogen virulence. The bacterium P. syringae is a host-specific pathogen, and its type III secretion system effectors appear to be key factors controlling its host range (Lindeberg et al., 2009). For pathogenic nematodes, little is known regarding the factors that determine host specificity and how host specificity is linked to virulence. The cyst nematode–plant interaction provides a unique pathosystem for exploring host-specific recognition processes. Our results revealed that the recognition of specific H. glycines CLE effectors occurs in the host plant soybean, but not in a nonhost plant Arabidopsis. It will be interesting to explore whether cyst nematode CLEs may be signaling different pathways in host vs nonhost plants to gain insight into whether nematode CLE gene evolution may be an underlying mechanism driving the specific adaptation of cyst nematodes to particular host plant species.
We thank Bob Heinz for maintaining nematode cultures, Jim Schoelz for assistance with ELISA, Esteban Fernandez and Stephanie Wise at the molecular cytology core for microscopy help, John Walker for critical review, and Melody Kroll for editorial assistance. We thank Rudiger Simon for the pBU14 binary vector, Walter Gassmann for the pMD1 binary vector, and Chris Taylor for the pCGT-35S and pAKK1467B vectors. This work was supported by the USDA-NRI Competitive Grants Program (grant nos 2007-35607-17790 and 2009-35302-05304 to M.G.M and X.W., and grant no. 2008-35302-18824 to T.J.B, E.L.D. and M.G.M), a USDA Special Grant (grant no. 2008-34113-19420) to M.G.M., an NSF CAREER Award (DBI-0845196) to D.K., and an MU Life Sciences Fellowship to A.R.