The DNA methylation landscape of the root‐knot nematode‐induced pseudo‐organ, the gall, in Arabidopsis, is dynamic, contrasting over time, and critically important for successful parasitism

Summary Root‐knot nematodes (RKNs) induce giant cells (GCs) within galls which are characterized by large‐scale gene repression at early stages. However, the epigenetic mechanism(s) underlying gene silencing is (are) still poorly characterized. DNA methylation in Arabidopsis galls induced by Meloidogyne javanica was studied at crucial infection stages (3 d post‐infection (dpi) and 14 dpi) using enzymatic, cytological, and sequencing approaches. DNA methyltransferase mutants (met1, cmt2, cmt3, cmt2/3, drm1/2, ddc) and a DNA demethylase mutant (ros1), were analyzed for RKN resistance/tolerance, and galls were characterized by confocal microscopy and RNA‐seq. Early galls were hypermethylated, and the GCs were found to be the major contributors to this hypermethylation, consistent with the very high degree of gene repression they exhibit. By contrast, medium/late galls showed no global increase in DNA methylation compared to uninfected roots, but exhibited large‐scale redistribution of differentially methylated regions (DMRs). In line with these findings, it was also shown that DNA methylation and demethylation mutants showed impaired nematode reproduction and gall/GC‐development. Moreover, siRNAs that were exclusively present in early galls accumulated at hypermethylated DMRs, overlapping mostly with retrotransposons in the CHG/CG contexts that might be involved in their repression, contributing to their stability/genome integrity. Promoter/gene methylation correlated with differentially expressed genes encoding proteins with basic cell functions. Both mechanisms are consistent with reprogramming host tissues for gall/GC formation. In conclusion, RNA‐directed DNA methylation (RdDM; DRM2/1) pathways, maintenance methyltransferases (MET1/CMT3) and demethylation (ROS1) appear to be prominent mechanisms driving a dynamic regulation of the epigenetic landscape during RKN infection.


Introduction
Plant-parasitic nematodes cause severe losses in agriculture (Singh et al., 2015). The endo-parasitic root-knot nematodes (RKN; Meloidogyne spp.) are among the major contributors. Root-knot nematodes penetrate host roots, delivering effectors into the vascular cylinder and inducing the development of a new organ (several giant cells; GCs) housed within a gall, from which they feed (Escobar et al., 2011. The multinucleated GCs undergo mitosis with incomplete cytokinesis . Several studies have reported observations of generalized gene repression in Arabidopsis and tomato galls, particularly in isolated GCs at early infection stages (Jammes et al., 2005;Fuller et al., 2007;Barcala et al., 2010;Portillo et al., 2013). However, the underlying mechanisms are poorly understood.
sRNAs induce either post-transcriptional gene silencing (PTGS) by targeting complementary mRNAs for degradation/translational repression in the cytoplasm, or transcriptional gene silencing (TGS) by repressive epigenetic modifications, such as DNA methylation and histone modifications, to homologous regions of the genome (Ruiz-Ferrer & Voinnet, 2009;Deleris et al., 2016). In plants, the major siRNA-mediated epigenetic pathway is RNA-directed DNA methylation (RdDM; Matzke et al., 2015;Cuerda-Gil & Slotkin, 2016). Plant DNA methylation occurs in all cytosine contexts: CG, CHG and CHH (where H is A, C or T). The establishment of de novo DNA methylation is catalyzed by DOMAINS REARRANGED METHYLASE 2 (DRM2) via the RdDM pathway (Ye et al., 2016), while maintenance of DNA methylation is mainly mediated by METHYL-TRANSFERASE1 (MET1), CHROMOMETHYLASE 3 (CMT3) and DRM2 or CMT2 in the CG, CHG and CHH methylation contexts, respectively (Zhang et al., 2018). Plant genomes also encode DNA glycosylases/lyases that remove cytosine methylation (Ortega-Galisteo et al., 2008), including REPRESSOR OF SILENCING 1 (ROS1; Gong et al., 2002;Zhu et al., 2007). Changes in DNA methylation were reported in soybean-cyst nematode and Arabidopsis-cyst nematode interactions, and in galls formed by Meloidogyne graminicola in rice (Rambani et al., 2015;Hewezi et al., 2017;reviewed in Hewezi, 2020;Atighi et al., 2020;respectively). However, a role for DNA methylation and its interplay with the dynamic regulation of sRNAs in the control of gene expression in galls of dicotyledonous species has not been described.
In this study, we analyzed the differential methylome of galls induced by Meloidogyne javanica at two critical infection time points (3 dpi and 14 dpi) from early gall formation to medium/ late infection stages, showing a dynamic regulation of the epigenetic landscape. Immunolocalization of methylation marks also showed the predominant contribution of the GCs at 3 dpi, among different cell types within the gall. The functions of several methylases involved in the RdDM pathway (such as DRM2), as well as those involved in methylation maintenance (CMT3, CMT2, MET1), together with DNA demethylase ROS1, are discussed in the context of gall formation. Our results strongly suggest that DNA methylation mediated by DRM2/1 through RdDM pathways and by maintenance methylases may contribute to preservation of genome integrity, while dramatic reprograming processes drive GC and gall formation.
Reproductive parameters were determined in soil-grown plants: number of females, number of egg masses produced, and number of total eggs per line (Olmo et al., 2020), in at least seven independent biological replicates per line (Student's t-test; P < 0.05). Giant cell phenotyping was carried out as described by Cabrera et al. (2018a).

RNA and DNA extraction and purification
Collection of Arabidopsis plant material and total RNA and genomic DNA extraction from the same samples were performed as described by Silva et al. (2019). Further details are provided in Methods S1.

Quantification of global 5-methylcytosine (5-mC)
To compare levels of global DNA methylation in galls induced by M. javanica in Arabidopsis and tomato with those in their corresponding RCs, the MethylFlash TM Global DNA Methylation (5-mC) ELISA Easy Kit was used (EpiGentek, New York, NY, USA) with c. 100 ng of DNA for each of the three independent biological samples for Arabidopsis (see the previous section). In tomato, a total of seven independent experiments were pooled in four independent biological samples per treatment. Each independent biological replicate contained at least 50 galls and > 30 RCs (DNA was extracted according to the procedure described by Silva et al., 2019), together with the negative and positive DNA controls provided in the kit. The 5-mC percentage was calculated according to the manufacturer's instructions.

Immunofluorescence and confocal microscopy
To localize 5-mC in GCs within gall sections, we performed immunofluorescence assays on resin sections (Testillano et al., 2013), followed by confocal analyses. 5-methylcytosine immunofluorescence was quantified using FIJI v.2.0.0-rc69/1.52n (Schindelin et al., 2012). The maximum projections (n ≥ 20 optical sections) of the 4 0 ,6-diamidino-2-phenylindole (DAPI) channel were used to define the Regions of Interest (ROIs) corresponding to the individual nuclei, and the maximum projections of the 5-mC (green) channel were used to measure fluorescence intensity values within ROIs. Only the ROIs presenting a maximum value of 85 (green channel) were considered. The corrected fluorescence intensity of each ROI was calculated as follows: Integrated density -(Area of selected nuclei 9 Mean fluorescence of background readings (adapted from Gavet & Pines, 2010;McCloy et al., 2014)). The average of the background readings was obtained using the mean fluorescence of two squares drawn in the background of each image. Significant differences were calculated using the Kruskal-Wallis test followed by Dunn's post hoc test (P < 0.05).
High-throughput MethylC-sequencing (MethylC-seq) library preparation and data processing The six independent MethylC-seq libraries, three for galls and three for RC at 3 dpi (see Table S2 for further details), were prepared using the NEXTflex ® Bisulfite-Seq Library Prep Kit (Bioo Scientific, Austin, TX, USA), indexed for post-sequencing demultiplexing, pooled in equimolar amounts and sequenced in one lane of an Illumina ® HiSeq 4000 PE100 platform (San Diego, CA, USA) and Illumina ® HiSeq X PE150 (RC2 and RC3; San Diego, CA, USA) by AllGenetics & Biology SL, A Coruña, Spain.
The six independent MethylC-seq libraries, three for galls and three for RC at 14 dpi (see Table S2 for further details), were prepared using the EZ DNA Methylation-Gold Kit (Zymo Research, Freiburg, Germany) following the library protocol Accel-NGS Methyl-Seq DNA Library Kit for Illumina Platforms by Macrogen Inc. (Seoul, South Korea) and sequenced using an Illumina ® HiSeq X PE150.
MethylC-seq data were processed using the pipeline available at https://github.com/seb-mueller/snakemake-bisulfite (commit f12bb6) in GitHub (GEO accession numbers GSE155853, GSE156025). To identify differentially methylated regions (DMRs), we used METHYLKIT v.0.99.2 (Akalin et al., 2012) with the adjusted P-value (q-value) < 0.05. We defined regions in the form of adjacent non-overlapping windows of 200 base pairs (bp; referred to hereafter as 'bins'). Bins were annotated as those overlapping with genes, promoters, or transposable elements (TEs). 'Promoter/TE' hereafter refers to a DMR that overlaps a gene promoter and a TE (as specified in Araport11, 06/2016; Cheng et al., 2017). Promoters are considered regions from 1000 bp upstream to 200 bp downstream of the transcription start point. For each region, the methylated and unmethylated cytosines were counted (separately for each library/region/methylation context) and tested for differential methylation through a logistic regression approach using the 'MN' overdispersion correction as implemented in METHYLKIT. Several DMRs were validated (Methods S1; Fig. S1d).
High-throughput RNA-sequencing (RNA-seq) library preparation and data processing Illumina's TruSeq Stranded mRNA Library Prep Kit was used to prepare the six independent libraries (three for galls and three for uninfected controls). Samples were dual-indexed for postsequencing demultiplexing, pooled in equimolar amounts and sequenced in a fraction of an Illumina ® HiSeq X PE150 by AllGenetics & Biology SL, A Coruña, Spain. A description of the DEGs acquired in TAIR10 and additional details are provided in Table S3.

Results
Meloidogyne javanica induces DNA hypermethylation during early establishment in Arabidopsis, but not at medium/late stages Based on the very high degree of gene repression observed in GCs (Barcala et al., 2010), coupled with the differential abundance of siRNAs at early stages of gall formation Ruiz-Ferrer et al., 2018), we hypothesized that RdDM may play an important role in RKN feeding site formation. Therefore, we quantified the global 5-mC DNA methylation induced by M. javanica in Arabidopsis at two infection stages, 3 and 14 dpi, which showed contrasting patterns of methylation, using two independent techniques: an ELISA-based assay, and bisulfite sequencing/ MethylC-seq. The ELISA-based approach identified significant differences in global DNA methylation at 3 dpi ( Fig. 1a) compared to RCs, but not at 14 dpi (Fig. 1a). MethylC-sequencing analysis of the same DNA showed a similar trend (P = 0.07; Fig. 1b; Table S2). The average degree of methylation in the CG, CHG, and CHH contexts was higher in 3 dpi galls than in RCs ( Fig. 1c

Research
New Phytologist respectively. The only significant difference between galls and RCs was found in the CHH context, although this context had the lowest level of DNA methylation (P = 0.03; Fig. 1c). By contrast, at 14 dpi, gall and RC global methylation levels were similar in all three methylation contexts ( Fig. 1d; Table S2d). These data support the idea that there is hypermethylation in early galls, but there are no differences at 14 dpi galls, both compared with its correspondent RCs.

DNA methylation preferentially targets transposable elements, particularly retrotransposons, in early galls
To identify DMRs of the genome that might be important for the interaction between Arabidopsis and RKNs at early and mediumlate infection stages, the Arabidopsis genome was divided into non-overlapping regions of 200 bp. A total of 775 DMRs (minimum methylation difference 15%; Table S5) were identified at 3 dpi. Most DMRs were in the CHG context (199, 468, and 108 DMRs in the CG, CHG and CHH contexts, respectively; Fig. 2a). Consistent with our previous observations, the vast majority (c. 90%) of DMRs, in all contexts, were hypermethylated in early galls (Figs 1, 2a). Further validation shows some representative DMRs matching five transposons belonging to different superfamilies (i.e. GYPSY, COPIA and LINE), a DNA transposon and a b-ZIP transcription factor, all showing similar hypermethylation tendencies in 3 dpi galls with respect to their corresponding RCs (q-value < 0.05) in both of the independent MethylC-seq analyses (Fig. S1). Similarly, DMRs identified at 14 dpi (minimum methylation difference 15%; 603) were also predominant in the CHG context (65, 450 and 88 for CG, CHG and CHH, respectively). In contrast to the findings at 3 dpi, c. 80% of DMRs, in all contexts, were hypomethylated in 14 dpi galls ( Fig. 2b; Table S5). Since there is no overall global change in the degree of methylation at 14 dpi ( Fig. 1), this finding suggests a remodeling of the methylation landscape during the infection. Strikingly, at 3 dpi, the genomic elements which overlapped with most DMRs in all three cytosine contexts were transposable elements (TEs; 64%); far fewer DMRs overlapped with genes (8%) promoters (3%) and promoter/TEs (6%; Fig. 2c; Table S5). By contrast, a lower proportion of DMRs overlapped with TEs at 14 dpi, and the proportion of DMRs overlapping promoter/TEs was similar (38% and 26%; respectively; Fig. 2d; Table S5). Each Arabidopsis chromosome was categorized into two different regions: pericentromeric (PCs) and the rest of the chromosome arms (ARs; Ruiz-Ferrer et al., 2018). At 3 dpi the vast majority of DMRs in promoters, genes and TEs were hypermethylated, independent of their position in the ARs or PCs (Fig. 2e). By contrast, by 14 dpi a large-scale rearrangement of methylation is evidenced by the opposite pattern: the vast majority of DMRs in promoters, genes and TEs were hypomethylated, The results also show that at 3 dpi, while promoter and gene DMRs are located at ARs on the Arabidopsis genome with predominant CHH and CG contexts, respectively (Fig. 3a,c), DMRs overlapping TEs were mostly located at PC regions with a predominant CHG context (Fig. 3e). Interestingly, the pattern was also unequivocally different at 14 dpi, as DMRs overlapping TEs were mostly located in ARs and the predominant context in TEs, promoters and genes was CHG (Fig. 3b,d,f). We further classified DMRs overlapping TEs ('Only TEs') and those overlapping TEs and promoters ('Promoters/TEs'), as well as those overlapping two different TE types: class I (retrotransposons that function via intermediate RNAs and reverse transcriptase) and class II (DNA-TEs that function via a DNA intermediate and transposase). Interestingly, at 3 dpi, DMRs in 'Only TEs' were predominantly hypermethylated retrotransposons ( Fig. 4a) in the GYPSY superfamily (Fig. 4e). By contrast, the predominant DMRs in 'Promoters/TEs' were DNA transposons (Fig. 4c) in the HELITRON and MUTATOR superfamilies (Fig. 4g). Furthermore, the proportion of DMRs matching retrotransposons (55.6%) is much higher than the proportion of retrotransposons (25.4%) among the TEs identified within the Arabidopsis genome in TAIR (Table S6). Similarly, the GYPSY superfamily represents 13.4% of TEs within the genome but 40% of TEs in DMRs at 3 dpi (Table S6), clearly indicating preferential changes in DNA methylation at 3 dpi ( Fig. 4). At 14 dpi, the pattern is different (preferential towards DNA transposons vs retrotransposons at 3 dpi, in HELITRON and MUTATOR; Fig. 4b,d,f,h) and 86% of TEs overlap with DNA transposons, compared to 74.5% in the Arabidopsis genome (Table S6). All these data point to a dynamic and contrasting DNA methylation landscape during infection.

New Phytologist
Functional analysis of the main Arabidopsis DNA methylases and demethylase ROS1 confirm the relevance of methylation dynamics during Meloidogyne javanica infection To confirm a functional role of DNA methylation/demethylation in galls, we performed infection tests in Arabidopsis mutant lines in which methyltransferase and DNA demethylase ROS1 activities were compromised. Infection was performed according to Olmo et al. (2017). We found a significant decrease in the infection rates in the DNA methyltransferase mutants (cmt3, dr-m1/drm2, ddc and met1) and in the DNA demethylase ros1 mutant compared to the Col-0 wild-type control (P < 0.05; Fig. 5a,b). By contrast, no significant differences were encountered for cmt2 or cmt2/cmt3 (Fig. 5c,d). Additionally, the gall diameter in all DNA methyltransferase mutants (except cmt2 and cmt2/cmt3) and ROS1-glycosylase was significantly smaller than in galls formed in Col-0 (P < 0.05; Fig. 5e). The most noticeable reduction was found in the met1-3 mutant (147 AE 5.0 lm), followed by ros1, as compared with Col-0 (215 AE 4.9 lm; Fig. 5e). These results are also consistent with the infection and reproduction parameters for soil-grown plants, in which all mutants tested showed significant differences ( Fig. 5f,g). Hence, reproductive parameters also confirmed the impact of these genes on the ability of nematodes to complete their life cycle ( Fig. 5f,g). Interestingly, GCs in all mutants were smaller compared to Col-0, except for cmt2 ( Fig. 6a-k; Video S1-S8), with marked differences particularly evident in met1 (P < 0.01; Fig. 6b,i; Video S7) and drm1/drm2 (Fig. 6d,j; Video S8). These results, together with the evidence presented in the previous sections, highlight the crucial role not only of the de novo DRM2/1 methylation mediated by sRNAs (RdDM) and direct methylation maintenance (MET1, CMT3), but also of the demethylation mediated by ROS1 activity during M. javanica infection, in line with the DNA methylation dynamics observed at early and late infection stages (Figs 1-4).

DNA methylation within the early galls is predominant in the giant cells
Galls are pseudo-organs formed by diverse cell types including GCs, and, in Arabidopsis and S. lycopersicum, gene expression of the isolated GCs is characterized by widespread gene repression (Barcala et al., 2010;Portillo et al., 2013). In order to focus on the epigenetic changes that correlate with the timing of general gene repression in GCs (Barcala et al., 2010) and may contribute to early gall/GC development, we focused on the 3 dpi stage for further analyses.

New Phytologist
To determine the relative contributions of different cell types within galls (i.e. the GCs and surrounding cells) to the global hypermethylation pattern observed in 3 dpi galls, we exploited the technical tractability of tomato (morphology preservation is poor in early galls in Arabidopsis). We chose to analyze the spatial distribution of DNA methylation using immunolocalization of 5-mC in early tomato galls as they also showed significant global hypermethylation compared to uninfected roots (Fig. 7a). The corrected immunofluorescence signal (see the Materials and Methods section; 'Immunofluorescence and confocal microscopy' sub-section) emitted by GCs, the proliferative cells surrounding the GCs (PLCs), and the nuclei of vascular cylinder cells from the uninfected root segments (VCs) were compared. Significant differences were observed between GCs and VCs, GCs and PLCs, and between PLCs and VCs nuclei (Fig. 7b), and these differences can be distinguished in the representative images shown in Fig. 7(c) (compare panel (c) with (d) and (c 00 ) with (b 00 )). The higher degree of fluorescence observed in GCs compared to PLCs and VCs suggests a predominant contribution of GCs to the general hypermethylation described by ELISA in tomato galls (Fig. 7a). Taken together, both the timing (Fig. 1) and location (Fig. 7), of DNA methylation in galls is consistent with the downregulation of gene expression in early developing GCs.

Meloidogyne javanica induces chromosome region methylation patterns matching sRNA distribution during early infection
To analyze whether sRNAs may drive the hypermethylation found in early galls, we compared the DMR data with available sRNA-seq data for equivalent biological samples (3 dpi Arabidopsis-M. javanica galls; Cabrera et al., 2016). Interestingly, most of the siRNAs that are exclusively present in galls and not in control roots (eGall-siRNAs) overlapped with DMRs matching TEs, followed by those targeting DMRs in promoters and genes (Table S4). This indicates a correlation between the differential accumulation of siRNA in galls, particularly those that are gall-distinctive (eGall; siRNAs only present in galls), and gall DMRs. The predominant eGall-siRNA size in TEs, promoters and genes was 24-nt, followed by 22-nt ( Fig. 8a; Table S4), that are usually products of DCL3 and DCL2, respectively. Neither promoters nor genes seem to be regulated by DCL4-dependent 21-nt siRNAs (Table S4; Fig. 8a).
Interestingly, the distribution pattern of eGall-siRNAs matched the methylation distribution in the CHG and CG contexts along the five chromosomes, was concentrated at PC regions, and correlated with the predominant retrotransposon locations in the Arabidopsis genome ( Fig. 8b; Simon et al., 2015), also matching the DMR distribution in retrotransposons at 3 dpi (Figs 2, 3, 4). However, those DMRs methylated in the CHH context extend along the chromosomes (Fig. 8b) and match the DMR distribution at promoters in the ARs (Figs 2, 3,  4). Taken together, these data indicate that the accumulation of eGall-siRNAs in retrotransposons could be related to their DNA hypermethylation profiles, and is clearly in agreement with the finding that several retrotransposon superfamilies which are wellknown targets of the RdDM pathways (i.e. ATCOPIA48 and ATHILA2; Zhang et al., 2006), are downregulated at 3 dpi in Arabidopsis galls (Ruiz-Ferrer et al., 2018; Fig. S1a). In addition, in galls at 14 dpi the expression of both retrotransposons significantly increased respect to galls at 3 dpi (Fig. S1b), indicating a dynamic in the repression/activation of major families of retrotransposons. Furthermore, differences in expression between the 3 dpi galls and uninfected tissue were not significant in the dmr1/ 2 mutant background (Fig. S1c).

Impact of gall and giant cell DNA methylation on gene expression at early infection stages
We performed RNA-seq on 3 dpi galls (Table S3) from the same biological samples as those used for MethylC-seq to ensure consistent comparisons between methylation patterns and gall transcriptomics, increasing the chances of finding reliable correlations (Silva et al., 2019). The reliability of our RNA-seq analysis was validated as it was consistent with the findings of a previous analysis of a gall microarray at the same infection stage (3 dpi; Barcala et al., 2010) that is, 84.5% of the DEGs in the 3-dpi gall microarray were common with the RNA-seq analysis (Fig. S2a). Additionally, a comparison with the DEGs from micro-dissected GCs (3 dpi; Barcala et al., 2010) showed that around half (48.3%) of the total DEGs in GCs are common to both analyses (Fig. S2b), indicating that a high proportion of transcripts from GCs in the total population of gall RNAs were detected by RNA-seq. The fact that 46% of the characteristic transcripts that peak in different cell-cycle stages (Menges et al., 2002) are DEGs in galls at 3 dpi provides further evidence of the large contribution of the early GC transcripts to the gall RNA-seq transcriptome. The largest proportion of upregulated genes were characteristic of M phase, followed by G1 phase. Consequently, M phase upregulated transcripts were overrepresented (v 2 test with P < 0.05), in accordance with the active cell divisions taking place in the early stages of GC differentiation (Table S7).
To analyze the putative impact of selective gall DNA methylation in gene expression, we identified the gall DEGs and GC DEGs (Barcala et al., 2010) within the genes and promoters that overlapped with DMRs (minimum methylation difference 15%; Table S5). Parameters for the cutoff of DEGs in the RNA-seq were set up at an adjusted P-value < 0.05 (Table S3b) to mimic as much as possible the criteria used in the GC microarray. Of those DMRs overlapping with a gene (102 DMRs; 99 different genes), 23 were DEGs, most of which were differentially methylated in the CG context and hypermethylated (Tables 1, S10). Of these 23 DEGs, 26.1% mapped siRNAs, with a high abundance of eGall-siRNAs (83.3%), which were either 24-nt or 22-nt in length, relative to those that are distinctive of control roots (16.7%; 1; Table S4).
Among the DMRs overlapping gene promoters (100 DMRs; 100 different genes), 33 corresponded to DEGs (Table 2). They were found to be preferentially methylated in the CHH context and were hypermethylated (Tables 2, S10). Interestingly, 19 out of the 33 DMRs matched transposons, and most were DNA-TEs (class II transposons). Of these 33 DEGs, 24.4% were targeted by siRNAs, with a higher abundance of eGall-siRNAs (50%) and with equal distributions of those that are 24-nt and 22-nt in length (Table S4).
Gene ontology enrichment analysis of the genes in the correlation list (Tables 1, 2), showed that 'RNA' and 'Protein' were the predominant categories (Fig. S4). Interestingly, in the 'RNA' category, the vast majority of the genes encoded transcription factors. Furthermore, there were several genes encoding proteins with basic cellular functions described in other biological systems, such as CHR19, a chromatin remodeler, the replication protein A (RPA2B), and a histone deacetylase (HDT2), among others (Tables 1, 2). Four DEGs in GCs that were not common to the gall RNA-seq results, also matched DMRs; three were hypermethylated and downregulated (Table 3).

Discussion
Transcriptomic analyses of Arabidopsis infected with Meloidogyne spp. indicated a very high degree of gene repression at early stages of infection, particularly in GCs, that is conserved in tomato (Barcala et al., 2010;Portillo et al., 2013). These results are in agreement with the finding that, several microRNAs that act by repressing plant genes encoding transcription factors were induced in galls (i.e. miR390/ARF3, miR159/MYB33 and miR172/TOE1) at different stages of the RKN infection Medina et al., 2017, D ıaz-Manzano et al., 2018respectively). Additionally, retrotransposons from several superfamilies were dramatically repressed in early galls (Ruiz-Ferrer et al., 2018). Moreover, the differential accumulation of sRNAs in early galls, such as rasiRNAs, as compared to control tissues strongly suggests the participation of RdDM pathways during nematode establishment Ruiz-Ferrer et al., 2018). However, no information is available regarding the

The relevance of methylation dynamics during Meloidogyne javanica infection
Meloidogyne javanica induces generalized hypermethylation during early gall development in tomato and Arabidopsis, as assessed here using two independent methods (3 dpi ; Figs 1,  7). Several studies reported methylation changes during pathogen attack in plants, such as bacterial infection (Dowen et al., 2012;Yu et al., 2013), fungal infection (Le et al., 2014;L opez S anchez et al., 2016) and infection by syncytia-forming nematodes, Heterodera spp. (reviewed in Hewezi, 2020). However, this is the first report of gall-localized hypermethylation during a plant-nematode interaction. By contrast, studies in Heterodera spp. and in the RKN M. graminicola in monocotyledonous species (rice; Atighi et al., 2020) showed general hypomethylation at early and mid stages of infection. Differences in the evolutionary origins of sedentary endo-parasitism, parasite life cycles, and experimental conditions could explain the observed divergences, or maybe different types of nematodes elicit divergent epigenomic strategies that may also depend on the host plant. Conversely, similarities in the transcriptomic profiles were observed between GCs and crown-galls formed by Agrobacterium tumefaciens (Barcala et al., 2010), showing generalized hypermethylation (Gohlke et al., 2013) and demonstrating that there are parallels between the response to RKN infection and other plant-pathogen interactions. Together, these findings suggests that different epigenomic strategies are used by different pathogen species during the interaction with their hosts. The methylation context more abundant in the DMRs (CG) is highlighted. The context and difference in methylation and corresponding gene expression (log 2 FC) are indicated, as well as the transposable elements (TEs) and their superfamilies present in the same bin, ID and gene description. Blue indicates DEGs that were found to be 'GC distinctive' when compared with the gall DEGs by Barcala et al. (2010). Red represents induced (and hypermethylated, compared to uninfected control roots) genes and green represents repressed (and hypomethylated) genes.

New Phytologist
However, no global differences in DNA methylation were encountered in medium/late galls relative to the control roots (14 dpi; Fig. 1). The contrast between early and late infection stages suggests a dynamic DNA methylation landscape during the infection. Accordingly, the distribution of differentially methylated regions (DMRs) was markedly different between the infection stages (Figs 2-4). Most DMRs were hypermethylated at 3 dpi and overlapped TEs, mostly retrotransposons located at pericentromeric regions of the chromosomes with a predominance of the GYPSY superfamily (Figs 2-4). However, at 14 dpi, most DMRs were hypomethylated and overlapped DNA transposons located at the chromosome arms with a predominance of the HELITRON superfamily (Figs 2-4). Since there is no overall global change in the amount of 5-mC-methylation in 14 dpi galls The methylation context more abundant in the DMRs (CHH) is highlighted. The context and difference in methylation and corresponding gene expression (log 2 FC) are indicated, as well as the transposable elements (TEs) and their superfamilies present in the same bin, ID and gene description. Blue indicates DEGs that were found to be 'GC distinctive' when compared with the gall DEGs by Barcala et al. (2010). Red represents induced (and hypermethylated, compared to uninfected control roots) genes and green represents repressed (and hypomethylated) genes.
New Phytologist (2022) 236: 1888-1907 www.newphytologist.com compared to RCs, these findings strongly suggest active remodeling of the methylation landscape from early establishment to the medium/late stage of infection. Consequently, main methylase and demethylase functions were required for adequate gall and GC formation, as well as for nematode reproduction in Arabidopsis (Figs 5, 6). Thus, mutants of methylases involved in the RdDM pathway (DRM2) and methylation maintenance (CMT3, MET1), and a mutant of DNA demethylase ROS1, showed either reproduction and/or infection parameters, and gall size/GC size severely compromised, but this was not observed in CMT2 during early infection (Figs 5, 6). These results also highlight the crucial and dynamic role of direct maintenance of methylation in the CG (mainly MET1-mediated), and CHG (mainly CMT3-mediated) contexts (Pikaard & Scheid, 2014), but perhaps a limited role for CMT2 maintenance of CHH methylation at 3 dpi. In line with this, the decrease in the infection parameters in drm1/drm2 with respect to the control line does not correspond to the reduction in the reproduction parameters (around 30% and 60-70%, respectively; Fig. 5). This may suggest that the relative contribution of different methylases/ demethylases could change over the course of the infection. MET1 is specifically recruited to hemi-methylated cytosines after DNA replication to methylate the newly synthesized unmethylated strand (Pikaard & Scheid, 2014), which should be essential during early GC formation due to an enhanced S phase as a consequence of repeated mitosis (de Almeida Engler & Gheysen, 2013), in accordance with the overrepresentation of Mphase-induced genes observed at 3 dpi in the RNA-seq analysis (Table S7). The absence of a clear phenotype in cmt2 galls reinforces the role of the DRM1/2 RdDM pathways during the early stages of M. javanica infection, not only in terms of CHG and CG methylation, but perhaps as a relevant mechanism of CHH methylation (Figs 5, 8;Pikaard & Scheid, 2014). This is further supported by the fact that genes encoding proteins involved in de novo methylation and maintenance, DNA-methyltransferases (MET1, CMT2 and CMT3), ARGONAUTE 4 (AGO 4), DRM1, the RNA polymerase PolIV/PolV subunits (NRPD2/ NRPE2, NRPD4/NRPE4) and INVOLVED IN DE NOVO 2 (IDN2), among others, were induced in galls at 3 dpi (Fig. S3). Generalized hypermethylation, particularly in the CHH context, as observed in 3 dpi galls, has been described during somatic embryogenesis in soybean after auxin induction (Ji et al., 2019). In accordance with this result, in this study, the greatest differences in methylation between galls and control tissue were observed in the CHH context at 3 dpi ( Fig. 1), consistent with previous studies that demonstrate an auxin maxima within early galls (Hutangura et al., 1999;Karczmarek et al., 2004;Absmanner et al., 2013;Cabrera et al., 2015;Kyndt et al., 2016;Olmo et al., 2020). Gall formation is a de novo post-embryonic organogenesis process that shares characteristics with other developmental processes, such as lateral-root initiation or callus formation, where auxins are the main orchestrating hormones (Olmo et al., 2020). The roles of various proteins involved in DNA methylation in the control of auxin biosynthesis and transport were recently described elsewhere (Forgione et al., 2019;Mateo-Bonmat ı et al., 2019). In this context, late-developing galls (14 dpi) show low auxin responses (Karczmarek et al., 2004), in agreement with the lack of differences in the global methylation observed between galls and uninfected roots at this stage of infection (Fig. 1). Methylation changes observed in early galls might, therefore, be at least partially connected to enhanced auxin signaling pathways, although further research should be performed to clearly elucidate its role.

DNA methylation within the early galls is predominant in the giant cells, matching sRNA distribution
We detected a selective hypermethylation in GCs using immunofluorescence targeted to 5-mC, but not in the surrounding cells of the galls, which suggests that the methylation state of GCs makes a key contribution to that of the general gall hypermethylation at 3 dpi (Fig. 7). This constitutes a novel finding that might be related to the dramatic morphological and transcriptomic changes suffered by the plant in order to provide nourishing cells that support nematode development . Some of these changes are reflected in the RNA-seq analysis of galls at 3 dpi, where 'GC-distinctive' transcripts defined in Barcala et al. (2010) are highly represented (Fig. S2) and Mphase transcripts are overrepresented (Table S7), as expected from cells that undergo repeated mitosis with partial cytokinesis (reviewed in de Almeida Engler et al., 2015). Enhanced methylation due to endoreduplication within the nuclei at 2-7 dpi can be  Barcala et al., 2010) overlapping differentially methylated regions (DMRs; methylation difference > 15%) described in this study.

Genomic region
Gene ID Description The context and difference in methylation and corresponding gene expression (log 2 FC) are indicated, as well as the transposable elements (TEs) and their superfamilies present in the same bin, ID and gene description. Red represents induced (and hypermethylated, compared to uninfected control roots) genes and green represents repressed (and hypomethylated) genes.

Research
New Phytologist discarded, as mitosis is the predominant mechanism at early stages, but endoreduplication of nuclei starts later in the infection, a process that has only been directly analyzed in 40 dpi Arabidopsis galls (Vieira et al., 2014;de Souza et al., 2017).
Interestingly, at 3 dpi, 24-nt siRNAs, the most abundant eGall-siRNAs, followed by 22-nt siRNAs, accumulated at DMRs overlapping TEs, promoters and genes, (Fig. 8); DCL3 and DCL2 can produce 24-nt siRNAs and 22-nt siRNAs, respectively. DCL3 mediates TGS (24-nt siRNAs), but DCL2dependent siRNAs can mediate PTGS and TGS (22-nt siRNAs), catalyzed by DRM2/DRM1 (Zheng et al., 2007;Naumann et al., 2011;Du et al., 2015;McCue et al., 2015). In accordance with these findings, the double mutant drm1/2 was severely impaired in nematode infection and gall/GC development (Figs 5,6). DRM2 catalyzes methylation in all three cytosine contexts (CG, CHG, CHH) in the RdDM pathway (Pikaard & Scheid, 2014). Three d post-infection eGall-siRNA accumulation matched DMRs and hypermethylation in the CHG context, followed by the GC context, mainly in TEs at pericentromeric chromosome regions correlating with retrotransposon locations; by contrast, CHH hypermethylation was less frequent, and expanded along the chromosomes, matching predominantly the distribution of DMRs at promoters, as well as DNA transposons (Figs 2, 4, 8;Ruiz-Ferrer et al., 2018). Therefore, the severe phenotype of the galls of the double mutant drm1/2 confirms the large contribution of the de novo RdDM methylation pathway during gall formation, presumably with a major role in CHG and CG de novo methylation in retrotransposons. These findings are also in agreement with observations of Meloidogyne-resistant phenotypes resulting from mutations in crucial molecular components of the canonical and noncanonical RdDM pathways, such as DCL2, DCL3, DCL4 or RDR2, RDR6 (Ruiz-Ferrer et al., 2018).
Furthermore, there is a robust correlation between the hypermethylation, the preferential accumulation of eGall-siRNAs at retrotransposons (particularly of the GYPSY and COPIA superfamilies), and their repression in early galls (Figs 3, 4, 8, S1;Ruiz-Ferrer et al., 2018). By contrast, in galls at 14 dpi the expression of retrotransposons ATCOPIA48 and ATHILA2 significantly increased with respect to galls at 3 dpi, indicating that the repression/activation of major families of retrotransposons is dynamic during the infection and correlates with the changes in the global methylation status of galls, as the general hypermethylation described at 3 dpi disappeared at 14 dpi (Fig. 1). Moreover, it also matched the high number of hypermethylated DMRs in the COPIA and GYPSY retrotransposon families at 3 dpi (Fig. 4e), and the low number of hypermethylated DMRs in the same retrotransposons at 14 dpi (Fig. 4b,f). These results strongly suggest that this could be a strategy in the early developing GCs for the stabilization of TEs. Furthermore, the repression of ATCOPIA48 and ATHILA2 observed in Col-0 galls at 3 dpi was nearly abolished in the drm1/2 mutant background (Fig. S1), suggesting the participation of those methylases in the regulation of retrotransposons in galls. Silencing and stabilization of transposons are typical of plant egg cells and embryos, with the aim of increasing genome integrity in the offspring (Feng et al., 2010).
RdDM is also active during embryogenesis, with increased CHH methylation in the endosperm and young embryo (Chow et al., 2020). However, in Arabidopsis, TE silencing in the developing embryo is also assisted by endosperm-derived 24-nt sRNAs (Bouyer et al., 2017). Like GCs within the galls, rapidly differentiating cells from the columella within the root meristem exhibit hypermethylated TEs and an increased abundance of transcripts encoding RdDM pathway components ( Fig. S3; Kawakatsu et al., 2016;Ruiz-Ferrer et al., 2018). Thus, TE stabilization driven by RdDM could be relevant in the differentiation process that leads to gall/GC formation at early stages, where a reprogramming of gene expression occurs, while successive mitosis takes place predominantly in the GCs (de Almeida Engler & Gheysen, 2013; Table S7), consistent with the distinctive hypermethylation observed in GCs but not in the surrounding cells (Fig. 7). By contrast, in Arabidopsis plants infected with the cyst nematode Heterodera shachtii, a widespread hypomethylation was predominant and some hypermethylated regions were not clearly associated with siRNA abundance Hewezi, 2020). Additionally, a majority of TEs, particularly DNA-TEs, were induced, as observed under other environmental stresses . Therefore, the epigenomic changes observed in early-galls seem to be more strongly related to postembryogenic developmental processes that require rapid cell proliferation and differentiation than to those typical of defense responses in other plant-pathogen interactions.
However, we should not forget the contribution of the PTGS pathway, which is consistent with the overrepresentation of the 'microRNA, natural antisense' DEG categories in the MAPMAN data for galls at 3 dpi (Fig. S2) and with the defined role of several microRNAs during gall formation (i.e. miR172/TOE, miR390/ ARF3, miR159/MYB33) (reviewed in Cabrera et al., 2018b;Jaubert-Possamai et al., 2019). Thus, RKNs use several strategies at the transcriptional and post-transcriptional level to reprogram their host plant cells.
Additionally, 52 DEGs identified in the transcriptomic analysis at 3 dpi correlated with DMRs overlapping genes and promoters, a low number compared to the total number of DEGs (Tables 1, 2), a tendency also described by other authors (Rambani et al., 2015;Hewezi et al., 2017), while in other Meloidogyne-monocotyledonous species studies no correlation was obtained at 3 dpi (Atighi et al., 2020). Most of those genes encode proteins that participate in critical basic cell functions that could be also crucial for gall/GC development, such as those regulating transcription or those involved in plant developmental processes or functions related to epigenetic pathways; for example, HISTONE DEACETYLASE 2B (HDT2) was shown to be upregulated and functional during callus formation from leaf explants (Lee et al., 2016;Tables 1, 2). HDTs interact with MET1 to silence transposons, regulating DNA methylation (Liu et al., 2012). Nevertheless, a direct relationship between DNA hypermethylation and repression, or viceversa, was not observed for all genes. Methylation at promoter regions is usually associated with gene silencing , but it can also be associated with gene activation as it could favor the binding of transcriptional activators or prevent the binding of transcriptional repressors (Zhang et  Therefore, to assign the methylation state of a promoter/gene to its transcriptional pattern is a complex matter. Interestingly, most of the DEG promoters at 3 dpi were methylated in the CHH context, where DRM1/2-mediated methylation seems to be the most prominent driving mechanism, and is probably mediated by the accumulated 24/22-nt eGall-siRNAs, as the contribution of CMT2 is quite limited, judging by the lack of phenotype in galls in mutant plants (Figs 5, 6, 8).
In conclusion, we have shown that galls formed by M. javanica in Arabidopsis and tomato are characterized by generalized hypermethylation at early stages, to which GCs are the main contributors with respect to other cell types within the gall. The lack of differences in global methylation between medium/late galls and RCs, as well as the remarkably different locations of DMRs between the two infection stages under study here, strongly suggest a dynamic remodeling of the epigenetic landscape in terms of DNA methylation during infection. Functional assays with Arabidopsis mutant lines point to the dynamic participation of methylases and demethylases during early-medium/late infection. Moreover, DNA methylation patterns and the accumulation of eGall-siRNAs, as well as the retrotransposon repression described in early galls (3 dpi; Ruiz-Ferrer et al., 2018), are likely regulated by DNA methylation and mediated by DRM2/1 via RdDM pathways and by maintenance methylases (CMT3, MET1). This process may contribute to the TE stability and therefore genome integrity that should be required for the dramatic reprogramming processes accompanying cell differentiation, which are concurrent with repeated mitosis during GC formation. Interestingly, most of the DEGs matching DMRs overlapping promoters or genes encode proteins that participate in critical basic cell functions which seem pertinent to GC and gall development. Further research with loss of function mutants for these genes will shed some light on their putative roles during gall formation. Castilla-La Mancha Government to CE. The laboratory received support from UCLM intramural funds. ACS and AM-G were supported by a fellowship from the University of Castilla-La Mancha, co-funded by the European Social Fund, and ACS was supported by an EMBO Short-Term Fellowship during her short-term stay at the University of Cambridge. AM-G was supported by a fellowship from Fundaci on Tatiana P erez de Guzm an el Bueno. Work on plant-parasitic nematodes at the University of Cambridge was supported by DEFRA license 125034/359149/3 and funded by BBSRC grants BB/R011311/ 1, BB/N021908/1, and BB/S006397/1. CP is supported by an individual MCSA fellowship. Arabidopsis homozygous T-DNA insertion mutant lines cmt3-11, drm1-2 drm2-2 and drm1-2 drm2-2 cmt3-11 (ddc) were kindly supplied by Dr Steve Jacobsen (University of California, Los Angeles, CA, USA; T-DNAs SALK_031705, SALK_150863, and SALK_148381, respectively) and cmt2-3, cmt2-3/cmt3-11 by Prof. Ian Henderson (University of Cambridge, Cambridge, UK). The homozygous T-DNA insertion mutant met1-3 was kindly provided by Dr Angelique Deleris and Dr Lionel Navarro (Institut de Biologie de l' Ecole Normale Sup erieure, ESN, CNRS, Paris, France). The Arabidopsis T-DNA insertion line for ros1 (SALK_045303C) was obtained from the Salk Institute Genomic Analysis Laboratory (La Jolla, CA, USA) SiGnAL (http://signal.salk.edu/cgi-bin/ tdnaexpress). We are also grateful for the scientific support of Prof. Sir David Baulcombe (University of Cambridge, Cambridge, UK).

Author contributions
ACS and VR-F performed most of the experiments, data collection, and in silico analysis of genomic and transcriptomic data, and participated in the interpretation of the results, as well as in the writing of the manuscript. CP, SYM and SEvdA performed and guided ACS in the in silico analysis of genomic and transcriptomic data. MFA, AM-G, AG-R and PA-U performed some of the experiments. PST and EB participated in and guided some of the experiments. CF and SEvdA participated in the writing of the manuscript and interpretation of the results. CE planned, designed, and guided most of the research and participated in the interpretation of the results and in the writing of the manuscript.   Methods S1 Detailed description of extended methods.

Table S1
List of primers and Arabidopsis mutants used in this study.

Table S2
Summary of general data from MethylC-seq and processing, for galls induced by Meloidogyne javanica and controls at 3 d post-infection (dpi) and 14 dpi.

Table S3
Summary of general data from RNA-seq and processing, for galls induced by Meloidogyne javanica and controls at 3 d post-infection.

Table S4
Percentages and numbers of unique siRNAs matching genes, promoters and transposable elements (TEs) identified in DMRs (methylation difference > 15%) in galls at 3 d post-infection.

Table S7
Total numbers and percentages of DEGs in galls formed by Meloidogyne javanica in Arabidopsis at 3 d post-infection common with genes involved in different phases of the cell division cycle.

Table S9
Differentially expressed genes represented in the MAP-MAN category 'microRNAs, natural antisense etc', in galls at 3 d post-infection.
Table S10 Genes and promoters that overlapped a DMR (methylation difference > 15%) and were differentially expressed in galls at 3 d post-infection (adjusted P-value < 0.05).
Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.