A single‐nucleus multiome profiling reveals the regulatory programs driving plant nematode–induced syncytium
Single-nucleus transcriptome profiles of Heterodera schachtii–induced syncytium on Arabidopsis roots.
Single-nucleus transcriptome profiles of Heterodera schachtii–induced syncytium on Arabidopsis roots.

Plant-parasitic cyst nematodes (Heterodera spp.) are among the most destructive agricultural pests, hijacking host roots to form a specialized, multinucleated feeding structure, the syncytium, through successive, genetically programmed cell-to-cell fusion events. After penetrating and migrating within root cells, the infective nematode juveniles target a compatible cortical or pericycle cell, often adjacent to the protoxylem pole, known as the initial feeding cell (Golinowski et al., 1996). Guided by nematode-secreted effectors (Mejias et al., 2019; Rutter et al., 2022; Bali & Gleason, 2024), the initial feeding cell undergoes profound morphological and transcriptional reprogramming, switching from a differentiated root cell to a mitotically arrested, endoreduplicated nucleus-rich hub (Nvyss et al., 1984; de Almeida Engler & Gheysen, 2013). Surrounding cortical, endodermal, and pericycle cells are then recruited and transformed into fusion-competent partners (Jones & Northcote, 1972; Sobczak & Golinowski, 2008); as they lose their distinct cell identities and merge cytoplasmically, the syncytium expands to encompass hundreds of cells (Hewezi & Baum, 2017; Matuszkiewicz & Sobczak, 2023). The outcome is a novel, metabolically hyperactive plant organ optimized for nematode feeding.
Beyond its agricultural importance, the cyst nematode–induced syncytium is an unrivaled in vivo model for somatic cell fusion, a process fundamental to eukaryotic development and physiology, from fertilization and myogenesis to tissue repair and tumor biology (Hernández & Podbilewicz, 2017; Hewezi & Baum, 2017; Matuszkiewicz & Sobczak, 2023). While syncytia across systems share conserved fusion machinery, molecular dissection has been limited by a lack of tractable multicellular contexts. Cyst nematode syncytia recapitulate these core mechanisms and uniquely enable study of how somatic cell-to-cell fusion is coordinated with transcriptome reprogramming, epigenetic regulation, organelle biogenesis, and metabolic adjustment. At the cellular level, syncytium development involves synchronized activation of endoreduplication, cell wall modifications, metabolism activation, and selective autolysis (Kyndt et al., 2013; Matuszkiewicz & Sobczak, 2023). Endoreduplication increases nuclear DNA content, generating polyploid nuclei that drive high transcriptional output (de Almeida Engler & Gheysen, 2013). Concurrently, plant cell walls are selectively modified through the activity of expansins, cellulases, and pectinases, facilitating fusion events and cytoplasmic connectivity (Bohlmann & Sobczak, 2014). Vacuolar breakdown and selective autophagy free up membrane and space for organelle redistribution, while enhanced mitochondrial biogenesis and proliferation of the endoplasmic reticulum fuel the elevated metabolic demands (Sobczak & Golinowski, 2008; Matuszkiewicz & Sobczak, 2023). Together, these processes convert multiple root cells into a coordinated metabolic hub.
Concomitant with these structural changes, syncytium development involves extensive reprogramming of host transcriptional and epigenetic landscapes. Genome-wide expression profiling of syncytial cells isolated by micro-aspiration or laser microdissection has revealed that genes related to cell cycle regulation, hormone signaling, macromolecular biosynthesis, cell wall modification, and nutrient transport are differentially regulated within the feeding site (Ithal et al., 2007; Szakasits et al., 2009; Liu & Mitchum, 2025). Transcriptome analyses have also identified transcription factors and noncoding RNAs uniquely enriched in syncytial cells, suggesting that the cyst nematodes manipulate both canonical and specialized regulatory pathways to reshape root cellular identity (Hewezi et al., 2012; Piya et al., 2017; Willig et al., 2022; Liu & Mitchum, 2024). In addition, cyst nematode infection actively alters the host epigenome during syncytium development (Rambani et al., 2015, 2020a,b; Hewezi et al., 2017; Vijayapalani et al., 2018; Hewezi, 2020; Bennett et al., 2022, 2023).
Despite extensive characterization of syncytium formation and development, critical gaps remain in our understanding of how the fully formed feeding organ sustains nematode nutrition and host–pathogen interactions. For instance, while it is known that cortex, endodermis, pericycle, and protoxylem-associated cells all contribute nuclei to the mature syncytium (Sobczak & Golinowski, 2008; Kyndt et al., 2013; Rodiuc et al., 2014; Ohtsu et al., 2017), it remains unclear whether these lineages assume specialized roles to optimize feeding site function. In addition, the degree of functional heterogeneity, whether subsets of syncytial nuclei specialize in specific molecular functions, remains elusive. In this regard, recent advances in single-nucleus RNA sequencing (snRNA-seq) and single-nucleus ATAC sequencing (snATAC-seq) (Dorrity et al., 2021; Farmer et al., 2021; Feng et al., 2022; Cao et al., 2023; Bonarota et al., 2025; Huo et al., 2025; Wang et al., 2025; Yan et al., 2025; Zadegan et al., 2025) offer unprecedented resolution to dissect the syncytium's molecular complexity. For example, by profiling nuclear transcriptomes in situ, snRNA-seq can distinguish these lineage-derived nuclear ‘modules’ based on their expression, while complementary snATAC-seq maps will identify the cis-regulatory landscapes and transcription factors that sustain each module's identity. Similarly, the functional heterogeneity of syncytium nuclei can be dissected into distinct nuclear ‘states’ by clustering snRNA-seq data, while parallel chromatin-accessibility maps will pinpoint epigenetic regulators that maintain those states. Together, these single-nucleus approaches offer the resolution and mechanistic insight needed to dissect how distinct syncytial nuclei collaborate epigenetically and transcriptionally to uphold the complex, multinucleated architecture and metabolic vigor of the fully developed syncytium.
In this study, we used snRNA-seq and snATAC-seq approaches to examine, respectively, the transcriptome and regulatory networks of fully differentiated syncytia induced by the beet cyst nematode Heterodera schachtii on Arabidopsis roots. Our analysis revealed pronounced diversity in gene expression and regulatory landscapes among syncytial cell populations, demonstrating that syncytium is likely a functionally compartmentalized structure rather than a homogeneous entity. This functional heterogeneity may enable coordinated yet differentiated activities within the multinucleated syncytium.
Arabidopsis thaliana L. (wild-type Col-0) was used in this study. Homozygous T-DNA insertional mutants of bZIP44 (AT1G75390; SALK_202170C), TGA7 (bZIP) (AT1G77920; GK-314D02 and GK-434F04), CAMTA1 (AT5G09410; SALK_008187C and SALK_108806C), and BZR1 (AT4G18890; SALK_006525C and SALK_017577C) were obtained from the Arabidopsis Biological Resource Center.
Seeds of wild-type Col-0 were surface-sterilized and sown in a randomized block design on 12-well Falcon tissue culture plates (BD Biosciences, San Jose, CA, USA) containing modified Knop's medium solidified with 0.8% Daishin agar (Research Products International Corp. Mount Prospect, IL, USA). Plants were grown at 24°C under a 16 h : 8 h, light : dark cycle. Ten-day-old seedlings were inoculated with c. 200 surface-sterilized second-stage juveniles (J2s) of Heterodera schachtii per plant. At 10 d post-inoculation (dpi), syncytia containing a single actively feeding nematode (swollen to the J3 stage) were identified under a stereoscope and carefully excised. Cuts were made at the outer borders of each syncytium to minimize contamination from adjacent tissues, so the collected root fragments were composed predominantly of syncytial tissues. Two biological replicates were prepared, each comprising at least 300 syncytia. At the same time, adjacent root regions next to each collected syncytium were sampled as paired controls, following previously described methods (Ozdemir et al., 2024; Zadegan et al., 2025). Biological samples corresponding to each treatment (syncytia or paired controls) were combined and processed for nuclei isolation and library construction.
Nuclei isolation and snRNA-seq library were constructed as recently described (Zadegan et al., 2025). In brief, frozen tissues were homogenized in Honda buffer using a gentleMACS Dissociator at 4°C to release intact nuclei. The homogenate was filtered through a 70-μm strainer, centrifuged at 700 g for 10 min at 4°C, and the pellet gently resuspended in 500 μl Honda buffer followed by filtration through a 30-μm strainer. Multiome libraries were generated using the Chromium Next GEM Single Cell Multiome ATAC + Gene Expression Reagent Bundle following the manufacturer's protocol (Manual part CG000338, Rev F; 10× Genomics). Nuclei suspensions were loaded on the Chromium X with a targeted recovery of 10 000 nuclei. Preliminary spike-in sequencing was conducted to estimate library complexity, and samples were then pooled to obtain comparable read counts across samples. Library quality and concentration were assessed by Agilent TapeStation 4200 and qPCR. Each library type (multiome gene expression and multiome chromatin accessibility) was sequenced on one lane of a NovaSeq × 10B 300 flow cell.
The analysis of snRNA-seq data was performed as described previously (Zadegan et al., 2025). Briefly, raw snRNA-seq data were processed using the Cell Ranger pipeline (v.7.2.0) developed by 10× Genomics. This pipeline was used for read alignment, barcode processing, and gene quantification to generate gene-nucleus count matrices from the sequencing output. Sequencing reads were aligned to the reference genome of Arabidopsis thaliana (TAIR10), and the corresponding genome sequence and annotation files were obtained from The Arabidopsis Information Resource (TAIR; www.arabidopsis.org). Following initial processing, the gene-nucleus count matrices generated by Cell Ranger were used for all downstream analyses, including quality control, normalization, and subsequent identification of differentially expressed genes (DEGs).
Ribosomal gene counts were removed, and low-quality nuclei with fewer than 200 detected genes as well as genes detected in < 20 nuclei were removed using the scanpy package (Wolf et al., 2018). Doublets were identified and filtered out using DoubletFinder (v.2.0.2) (McGinnis et al., 2019). SoupX (v.1.5.0) was similarly used to remove ambient mRNA (Young & Behjati, 2020). The obtained count matrix was then used for identifying cluster DEGs. Normalization of read counts was performed using Scran's pooling-based size factor estimation method (Lun et al., 2016), and the R package scry was utilized to identify highly deviant genes (Street et al., 2022). PCA was conducted using the scanpy package, followed by visualization using non-linear dimensionality reduction methods, t-SNE, and UMAP. The HarmonyPy package (https://github.com/slowkow/harmonypy) was used to integrate control and syncytium samples. Clustering was conducted for the integrated dataset using the Leiden algorithm with a resolution parameter set to 1.5. Cluster-specific DEGs were determined using scanpy.tl.filter_rank_genes_groups, applying thresholds of 0.2 for both min_in_group_fraction and max_out_group_fraction. Cell-type annotation of the identified clusters was done using Arabidopsis known marker genes (Denyer et al., 2019; Ryu et al., 2019; Farmer et al., 2021; Shahan et al., 2022), and DEGs in each cluster were identified.
The snATAC-seq data from syncytium and control samples were processed using the Cell Ranger ATAC (v.2.1.0) count pipeline, using the Arabidopsis TAIR10 reference genome. Peak-by-cell count matrices were imported into a Muon (Bredikhin et al., 2022) object for downstream analysis in Python. Detection and removal of doublets were conducted using scDblFinder (v.1.18.0). Low-quality nuclei were excluded based on the following thresholds: only nuclei with between 1500 and 10 000 fragment counts, and at least 750 peaks, and a calculated nucleosome signal of < 2 were retained. Overlapping or adjacent peaks within 50 bp were merged using pybedtools to account for minor shifts in peak positions.
Integration of snRNA-seq and snATAC-seq data, and transfer of cell type annotations from the snRNA-seq dataset to the scATAC-seq data, were conducted using a latent space-based label transfer approach in the scVI framework (v.1.1.2). A gene activity matrix was computed by mapping peaks to their corresponding genes and summing the accessibility counts for each gene. Peak-to-gene assignments were based on Cell Ranger ATAC's default annotation, which maps peaks within ± 2 kb of the transcription start site to the corresponding gene. Gene expression and gene activity matrices were restricted to shared genes and concatenated into a single AnnData object. We used scVI to create a shared latent representation of the combined snRNA-seq and snATAC-seq dataset. The model was trained with a negative binomial likelihood and a latent dimensionality of 30. Cell type annotations were then transferred from RNA to ATAC nuclei using a k-nearest neighbor approach, and UMAP coordinates corresponding to ATAC nuclei were then extracted from the joint embedding for downstream visualization and analysis. Differentially accessible peaks for each cluster were identified using the atac.tl.rank_peaks_groups function (adjusted P < 0.05). Average accessibility of identified peaks in each cluster was computed for syncytium and control nuclei in each cluster.
To validate the robustness of cell-type annotation transfer between snRNA-seq and snATAC-seq data, we assessed concordance between transcriptomic and chromatin accessibility profiles using physical barcode pairings from the original multiome dataset. A total of 5844 physically paired nuclei were identified and independently annotated based on snRNA-seq and snATAC-seq data. Nuclei assigned as ‘Unknown’ in either modality were excluded to reduce ambiguity associated with low-confidence annotations, resulting in 2928 confidently annotated paired nuclei. Concordance was quantified by comparing RNA- and ATAC-derived cell identities at the broad lineage level. For this analysis, fine-resolution clusters were consolidated into six major root lineages, including epidermis, cortex, endodermis, stele, phloem, and xylem.
As a sensitivity check and to assess our cross-modal integration strategy, weighted nearest-neighbor (WNN) analysis was conducted on 5844 nuclei with paired ATAC and RNA profiles passing quality control. Before multimodal integration, RNA and ATAC profiles were analyzed separately. RNA counts were normalized, and dimensionality reduction was conducted using PCA on 3000 highly variable features. For ATAC counts, term frequency-inverse document frequency (TF-IDF) normalization and latent semantic indexing (LSI) were used for normalization and dimensionality reduction, respectively. The resulting RNA PCA and ATAC LSI representations were then used to construct modality-specific nearest-neighbor graphs. The graphs obtained were integrated using WNN analysis to generate a joint multimodal neighbor graph. Leiden clustering (resolution = 1.0) and UMAP visualization were subsequently performed. To evaluate whether nuclei grouped within the same WNN clusters received concordant RNA-derived and ATAC-transferred annotations, cell type labels derived from RNA reference mapping and ATAC labels obtained after label transfer were assigned to the corresponding nuclei in the WNN embedding. The percentage of nuclei assigned to broad cell-type annotation was then calculated for each WNN cluster.
To determine whether chromatin accessibility changes in syncytium cells are associated with transcriptional activation, we focused on clusters 0, 10, 18, and 22, which were enriched for syncytium ATAC-seq peaks. For each cluster, differentially accessible peaks and DEGs were identified by comparing syncytium nuclei with control nuclei (adjusted P < 0.05). Genes associated with cluster-specific accessible peaks were then assigned, and their overlap with significantly upregulated genes was determined to identify syncytium genes potentially regulated by increased chromatin accessibility.
To identify cis-regulatory elements associated with syncytium-upregulated genes, motif enrichment analysis was performed on the snATAC-seq peak regions from clusters 0, 10, 18, and 22, which were enriched for syncytium accessible peaks and linked to upregulated genes. Analysis was conducted using the homer software package (findMotifsGenome.pl) (Heinz et al., 2010). Genomic sequences corresponding to these peak regions were analyzed against the Arabidopsis reference genome (TAIR10). Motifs showing at least twofold enrichment with P < 0.001 were considered significantly enriched and selected as candidate cis-regulatory elements. Enriched motifs were further annotated by comparison with known Arabidopsis transcription factor binding sites to identify potential regulatory factors active in syncytial cells.
Seeds of promoter:GUS reporter lines for AT3G03050 (Bernal et al., 2008), AT5G15070 (Zhu et al., 2019), AT4G03210 (Becnel et al., 2006), and AT4G11280 (Tsuchisaka & Theologis, 2004) were planted on 12-well plates (BD Biosciences). Ten-day-old seedlings were inoculated with c. 200 surface-sterilized J2s of H. schachtii per plant and assayed histochemically for GUS activity at 5, 10, and 15 dpi. GUS activity was conducted as described by Jefferson et al. (1987). For infected roots, GUS staining was monitored at 30-min intervals and evaluated within 8 h of the reaction to detect localized GUS activity at nematode feeding sites while minimizing background staining. Because GUS activity in uninoculated control tissues was weak or undetectable within this period, control tissues were incubated overnight (c. 16 h) to reveal basal promoter activity.
Seeds of the mutant lines along the wild-type Col-0 were surface-sterilized and grown in a randomized block design on 12-well Falcon tissue culture plates with at least 10 replicates per genotype. Plates were maintained at 24°C under a 16 h : 8 h, light : dark cycle as described above. Ten-day-old seedlings were inoculated with c. 200 surface-sterilized J2s of H. schachtii per plant. Inoculated plants were maintained under the same growth conditions for an additional 3 wk, after which adult female nematodes were counted. Mean values of mutant lines were compared with wild-type Col-0 using a t-test in SAS, and differences were considered statistically significant at P < 0.05.
Arabidopsis seeds of the mutant lines and wild-type Col-0 were germinated on modified Knop's medium, and 10-d-old seedlings were inoculated with c. 200 surface-sterilized J2s of H. schachtii. At least 20 nematode-associated syncytia were randomly selected from each line, photographed, and measured as previously described (Hewezi et al., 2012). For root length measurements, mutant and wild-type Col-0 plants were grown vertically on modified Knop's medium for 12 d, and the primary root length of at least 10 plants per line was measured from the crown to the root tip. Statistically significant differences between mutant lines and wild-type Col-0 plants were determined using a t-test (P < 0.05).
We generated single-nucleus multiome libraries from root segments containing Heterodera schachtii–induced syncytia on Arabidopsis roots at 10 dpi, which reflect fully differentiating feeding cells. At the same time point, c. 1 cm of root tissues adjacent to each collected syncytium and lacking visible syncytial characteristics were also collected and used as adjacent nonsyncytium root tissues (ART) control. We selected 10-d-old, fully developed syncytia for our analysis because previous transcriptomic profiling showed that gene expression patterns are already established by day 5, making the 10-dpi a representative stage of mature syncytium (Szakasits et al., 2009; Liu & Mitchum, 2025). Single-nucleus multiome sequencing provides a powerful approach to simultaneously assess gene expression (RNA-seq) and chromatin accessibility (ATAC-seq) in individual nuclei. However, after data processing and filtering, the number of nuclei with both high-quality RNA and ATAC data was low, with 336 nuclei from ART control and 581 nuclei for syncytium-containing root regions. Given the potential biological noise introduced by low coverage, we opted to analyze gene expression and chromatin accessibility independently to maximize the biological insights gained from each modality. By analyzing each modality independently, we retained a larger number of high-quality nuclei, improving the resolution and robustness of our findings.
A total of 6693 and 3788 high-quality nuclei from ART control and syncytium samples, respectively, were analyzed (Supporting Information Table S1). The median number of expressed genes per qualified nucleus was 688 in ART control roots and 638 in syncytium-containing root regions (Table S1). After read-count normalization, nuclei from both syncytium and ART control samples were clustered using the Uniform Manifold Approximation Projection (UMAP) method. The UMAP plot showed that nuclei isolated from ART control (orange dots) and syncytium-associated tissues (blue dots) have some overlapping yet distinct distributions (Fig. 1a). This separation suggests that syncytium-associated tissues undergo defined transcriptional reprogramming while retaining some shared gene expression profiles with ART control cells. Using previously confirmed cell-type specific marker genes (Table S2; Fig. S1), a total of 33 distinct cell clusters were identified and annotated (Fig. 1b). Of these, six clusters (10, 12, 18, 28, 30, and 32) contained a particularly high number of nuclei isolated from syncytium-containing root regions and were therefore considered syncytium-enriched clusters, although these clusters may represent cell populations associated with or enriched in syncytium-containing regions rather than exclusively syncytial cells (Fig. 1c; Table S3). These clusters were annotated as phloem, pericycle, xylem, cortex, and unknown (Fig. 1b). Clusters 16, 5, and 19 contained a significant proportion (50–70%) of nuclei from syncytium-containing root segments, though they are not exclusively syncytium-dominated (Fig. 1c; Table S3). The mixed composition of these ‘syncytium-associated clusters’ may represent cells with overlapping transcriptional signatures between ART control and syncytium-containing root regions.

Single-nucleus transcriptome profiles of Heterodera schachtii–induced syncytium on Arabidopsis roots. (a, b) Uniform Manifold Approximation and Projection (UMAP) plots of nuclei isolated from syncytium tissues and nonsyncytium adjacent root tissues (ART) control at 10 d post infection with H. schachtii. (a) Nuclei are colored based on tissue identity: control (orange dots) and syncytium (blue dots). (b) Nuclei are colored according to unsupervised clustering and assigned cell types. (c) Bar graph showing the percentage of syncytium nuclei relative to ART control nuclei across the 33 identified clusters.
We performed differential gene expression analysis in all clusters and identified cluster-specifically differentially upregulated and downregulated genes (Tables S4 and S5). Representative examples of top cluster-specific genes are provided in Fig. S2. For syncytium-enriched clusters (10, 12, 18, 28, 30, and 32), a total of 1599 DEGs were identified. Among these genes, 867 (739 upregulated and 128 downregulated) were identified as preferentially regulated in the syncytium-enriched clusters and were therefore classified as syncytium-regulated genes (Table S6). UMAP plots showing the expression profiles of two representative genes for each syncytium-enriched cluster are shown in Fig. 2(a–f). For syncytium-associated clusters (5, 16, and 19), 859 DEGs were identified, of which 465 genes (392 upregulated and 73 downregulated) were preferentially regulated in these three clusters (Table S7). Interestingly, a set of 142 DEGs were common to syncytium-enriched and syncytium-associated clusters (Fig. 2g), which may provide insights into the common cellular processes between these two cluster sets.

Functional annotation of syncytium differentially expressed genes in Arabidopsis. (a–f) Uniform Manifold Approximation and Projection (UMAP) plots illustrating the expression profiles of two representative genes from each syncytium-enriched cluster: Clusters 10 (a), 12 (b), 18 (c), 28 (d), 30 (e), and 32 (f). (g) Venn diagram demonstrating unique and common genes between syncytium-enriched and syncytium-associated clusters. (h) Gene Ontology (GO) enrichment analysis of differentially expressed genes from syncytium-enriched and syncytium-associated clusters. Significantly enriched GO terms (adjusted P < 0.05) were identified using Panther v.19.0 with Fisher's exact test and Bonferroni correction for multiple testing. The color bar indicates the adjusted P values of enriched GO terms.
Enrichment analysis of gene ontology (GO) terms revealed both overlapping and distinct biological processes between syncytium-enriched and syncytium-associated clusters (Fig. 2h). In syncytium-enriched clusters, upregulated genes were enriched in diverse metabolic processes, including inositol, oligosaccharide, and carboxylic acid metabolism as well as processes related to syncytium maintenance, vascular and transmembrane transport, and multicellular organism development. These clusters also showed a significant enrichment for genes involved in hormonal responses and various stress-related pathways, including oxidative stress. By contrast, upregulated genes in syncytium-associated clusters showed overrepresentation of genes involved in the biosynthesis of lignin and very long-chain fatty acids, which are critical for cell wall reinforcement, in addition to genes involved in wounding responses and biotic stimuli. Notably, both gene lists shared significant enrichment for genes related to stress responses and phenylpropanoid biosynthesis (Fig. 2h). Enrichment analysis of downregulated genes in syncytium-enriched clusters showed enrichment for genes involved in glutamine metabolism and reactive nitrogen species, whereas syncytium-associated clusters showed enrichment only for glutamine metabolism (Fig. 2h).
We used promoter : GUS reporters to validate and localize syncytium expression inferred from the snRNA-seq data. Transgenic lines carrying promoter : GUS constructs for AT3G03050 and AT5G15070 (cluster 10), AT4G03210 (cluster 18), and AT4G11280 (cluster 16) were inoculated with H. schachtii and assayed histochemically for GUS activity at 5, 10, and 15 dpi to capture different stages of syncytium development. GUS activity in noninfected lines was determined after overnight (c. 16 h) incubation to reveal basal promoter activity (Fig. S3), as staining was weak or undetectable after 8 h. In infected roots, GUS staining was monitored within 8 h of the reaction to detect localized activity at nematode feeding sites while minimizing background staining. In infected plants, strong GUS staining was observed in syncytia for all four genes at 5 and 10 dpi (Fig. 3a–d), consistent with localized promoter activity in nematode feeding sites. At 15 dpi, however, GUS staining was no longer detected for AT3G03050, AT5G15070, and AT4G03210, while AT4G11280 exhibited only weak staining in the syncytium (Fig. 3d). Together, these results validate the single-nucleus expression profiles and indicate that the transcription of many syncytium-associated genes is developmentally regulated and declines as the feeding site matures.

Promoter activity of selected syncytium-expressed genes during Heterodera schachtii infection. (a–d) Histochemical β-glucuronidase (GUS) staining of transgenic Arabidopsis lines carrying promoter:GUS constructs for AT3G03050 (a) and AT5G15070 (b) from cluster 10, AT4G03210 (c) from cluster 18, and AT4G11280 (d) from cluster 16. Roots were inoculated with H. schachtii juveniles and assayed for GUS activity at 5, 10, and 15 d post-infection (dpi). GUS staining was monitored at 30-min intervals, and staining patterns were evaluated within 8 h to detect localized GUS activity at nematode feeding sites while minimizing background staining. For each promoter construct and time point, six independent infected plants were analyzed, and all exhibited the representative GUS staining pattern is shown. Bar, 800 μm.
Multinucleated cells in animal systems are known to exhibit striking compartmentalization and functional heterogeneity during differentiation and development (Kim et al., 2020; Petrany et al., 2020). Thus, we tested this possibility by conducting dimensionality reduction analysis of syncytium-enriched and syncytium-associated clusters. Interestingly, syncytium-enriched and syncytium-associated clusters showed unique transcriptional signatures (Fig. 4a), suggesting specialized roles or states within the broader cellular network. We next performed pseudotime trajectory inference using Palantir, which produced two branched trajectories emerging from a common origin near cluster 18 (Fig. 4b–e). Nuclei from clusters 5, 10, and 12 occupy intermediate positions along the inferred axis, while clusters 28 and 19 map to divergent arms (Fig. 4c,e), implying distinct functional states. The distribution of nuclei from the syncytium-enriched and syncytium-associated clusters along the pseudotime axis, visualized by violin plots, positioned cluster 18 at one end of the transcriptional continuum, clusters 5, 10, and 12 at intermediate positions, and clusters 28 and 19 along distinct branches (Fig. 4f). Consistent with the trajectory analysis, this pattern points to the presence of transcriptionally distinct programs within or between the mature syncytia rather than discrete temporal developmental stages, although we cannot distinguish nuclear heterogeneity within individual syncytia from variation among syncytia or biological samples.

Pseudotime trajectory inference reveals distinct functional states within the Arabidopsis thaliana syncytium. (a) Uniform Manifold Approximation and Projection (UMAP) plot showing syncytium-enriched and syncytium-associated clusters used for trajectory analysis. Nuclei are color-coded according to their assigned clusters. (b–e) Pseudotime trajectory analysis showing two transcriptionally distinct states emerging from a common origin at cluster 18. One trajectory advanced into cluster 28 (b, c), and the alternative trajectory terminates at cluster 19 (d, e). Nuclei specific to each trajectory path are highlighted in blue in (c, e). (f) Violin plot showing pseudotime distribution of nuclei from syncytium-enriched and syncytium-associated clusters, as inferred by Palantir. Cluster 18 marks the origin of the trajectory, with nuclei progressing toward distinct terminal states represented by clusters 28 and 19.
To explore the biological processes underlying the observed transcriptional continua, we performed cluster-specific functional annotation and GO term enrichment analysis using cluster-specifically expressed genes (Fig. S4). Cluster 18 was enriched for ‘anatomical structure morphogenesis’, with elevated expression of genes implicated in cell-wall remodeling, including, for instance, those encoding pectate lyases, pectin methylesterase 3, expansin, xyloglucan endotransglucosylase, and endo-1,4-beta-glucanase whose functions in syncytium differentiation and development were previously reported (Bohlmann & Sobczak, 2014; Zhang et al., 2017). Clusters 5, 10, and 12 together were marked with enrichment for genes involved in inositol metabolic process, organic acid metabolic process, transport, and response to stress (Fig. S4). Clusters 19 and 28 were similarly enriched in genes involved in phenylpropanoid metabolic process, secondary metabolic process, suberin biosynthetic process, and long-chain fatty-acyl-CoA metabolic process (Fig. S4). Taken together, these enrichments are consistent with distinct gene expression patterns among syncytium-associated nuclei at 10 dpi and may reflect functional differences among nuclei within or between syncytia, rather than discrete temporal stages of development.
After filtering and quality control, we obtained 6605 and 2435 high-quality snATAC-seq nuclei from ART control and syncytium samples, with an average of 88 990 and 91 370 read pairs per nucleus, respectively (Table S8). We then integrated snRNA-seq and snATAC-seq (Fig. 5a) to enhance the interpretation of chromatin accessibility data by leveraging the well-defined cell type annotations from snRNA-seq. Since snATAC-seq alone lacks direct gene expression information, we used joint clustering to align ATAC-seq profiles with snRNA-seq-derived cell types (Fig. 5b). This approach allowed us to transfer cell type annotations from snRNA-seq to snATAC-seq clusters (Fig. 5c), enabling a more accurate and biologically meaningful annotation of chromatin accessibility patterns across syncytial cell populations. UMAP visualization of annotated snATAC-seq clusters revealed distinct chromatin accessibility profiles across different cell populations (Fig. 5c; Table S9). The clustering pattern suggests clear epigenetic differences, with clusters 0, 1, 7, 10, 18, 22, 25, and 32 containing nuclei from syncytium-containing root tissues separated from the more heterogeneous clusters representing mixed root cell populations (Fig. 5c; Table S9). The well-defined separation of these clusters points to a unique chromatin accessibility landscape, suggesting specialized regulatory programs that distinguish syncytial cells from other root cell populations. However, the more dispersed nature of the remaining root cell clusters suggests a broader spectrum of chromatin states, likely reflecting diverse cell types or transitional states within root tissues.

Integration and annotation of snATAC-seq data using snRNA-seq defined cell types in Arabidopsis thaliana. (a) Uniform Manifold Approximation and Projection (UMAP) projection of the combined snRNA-seq and snATAC-seq datasets. (b) UMAP of the joint clustering, in which snATAC-seq profiles are aligned to snRNA-seq-defined clusters. (c) UMAP of snATAC-seq clusters colored by the cell type labels transferred from snRNA-seq. (d) Impact of chromatin accessibility on gene expression in syncytium. Heatmap of syncytium-associated snATAC-seq clusters showing counts of nonsyncytium adjacent root tissues (ART) control vs syncytial nuclei, numbers of peaks with increased accessibility in syncytium vs ART control, numbers of accessible peaks associated with syncytium-upregulated genes, and numbers of syncytium-upregulated genes associated with accessible peaks. Color intensity reflects relative values. Note that the number of accessible peaks associated with syncytium-upregulated genes and the number of syncytium-upregulated genes associated with accessible peaks differ across clusters because a single gene may be linked to multiple peaks, and a single peak may be linked to multiple genes.
As snRNA-seq and snATAC-seq datasets were analyzed independently, and snATAC-seq nuclei were annotated by label transfer from the snRNA-seq modality, we further evaluated the robustness of this integration using physically paired nuclei. We identified 2928 confidently annotated paired nuclei and compared RNA- and ATAC-derived labels at the broad lineage level. The two modalities showed 62.74% concordance overall (Fig. S5a,b), with particularly strong agreement across major epidermal, cortical, and endodermal states. The lower concordance for xylem likely reflects the limited number of xylem nuclei available for cross-modal validation (87 nuclei), of which only 10 (11.49%) were classified as xylem by ATAC-seq, resulting in an inherently less robust estimate (Fig. S5a,b).
To further evaluate the consistency of cell-type organization across the two modalities and determine if the native pairing of RNA and ATAC profiles resulted in a multimodal organization similar to that obtained from our cross-modal analysis, we performed WNN analysis on 5844 nuclei with paired ATAC and RNA profiles. Joint analysis of RNA and ATAC modalities resulted in distinct multimodal neighborhoods (Fig. S6a). Mean modality weights for RNA and ATAC in the WNN representation were 0.581 and 0.419, respectively, indicating the balanced contributions from the two modalities. Cell type labels derived from RNA reference mapping and ATAC labels obtained after label transfer were assigned to the corresponding nuclei in the WNN embedding (Fig. S6b,c). These results showed that nuclei with similar broad RNA-based cell type annotations and ATAC-transferred annotations are mostly clustered together. Nuclei with an ‘Unknown’ annotation occupied distinct regions of WNN embedding rather than being randomly distributed (Fig. S6b,c). The percentages of nuclei assigned to each broad cell-type annotation derived from RNA or ATAC for each WNN cluster were calculated (Fig. S6d,e). The results revealed broad consistency between the joint multimodal structure and the cell-type organization identified in our original analysis.
We next determined differentially accessible peaks in each cluster using all 33 clusters as a background for comparison (Table S10). Chromatin accessibility, measured by the number of ATAC-seq peaks, varied widely across clusters. Interestingly, among the eight syncytium-enriched clusters with separated chromatin accessibility signature (Fig. 5c), clusters 0, 1, 7, 10, 18, and 22 contained substantial numbers of differentially accessible peaks, strongly suggesting these clusters capture the core syncytium accessibility program (Tables S9, S10). No peaks were detected in the remaining two clusters (25 and 32), and this is most likely due to the low number of cells in these two clusters. Notably, a separate set of clusters that contain few or no syncytium cells (clusters 8, 9, 11, 13, 17, 19, 24, and 27) nonetheless contained very large peak sets, implying that those accessibility signatures arise from ART control-associated or other cell-type programs rather than syncytium-related changes (Tables S9, S10).
To investigate the impact of chromatin accessibility on gene expression in syncytium, we first identified DEGs and ATAC peaks within four syncytium-enriched clusters annotated as unknown (cluster 0), phloem (cluster 10), xylem (cluster 18), and cortex (cluster 22) (Tables S11, S12). Clusters 1 and 7 were excluded because they are annotated as epidermis, a nonsyncytium cell type. We next linked significantly highly accessible peaks in syncytial cells with upregulated genes in each cluster. This analysis resulted in the identification of a total of 4264 peaks with increased accessibility in syncytium compared to ART control (Fig. 5d). Of these, 1408 significantly highly accessible peaks in syncytial cells were found to be linked to 1414 upregulated genes (Fig. 5d; Table S13). Cluster 10 is strongly syncytium-enriched (403 vs 57 control nuclei) and exhibits extensive chromatin opening (2518 differentially accessible peaks). Approximately 43% of these peaks (1093) are linked to 1115 transcriptionally upregulated genes, consistent with tight coupling between chromatin accessibility and transcriptional activation in this state. Cluster 18 shows a similar pattern with 41% of the accessible peaks linked to upregulated genes in syncytium. Cluster 0 displays more moderate chromatin accessibility (1088 peaks) with only c. 21% of peaks (226) linked to 210 upregulated genes. By contrast, cluster 22, despite exhibiting the highest number of differentially accessible peaks (3007), shows minimal gene linkage (2.6%) (Fig. 5d), suggesting widespread chromatin remodeling that is largely uncoupled from immediate transcriptional activation and may reflect regulatory priming or distal enhancer activation. Representative accessible regions and their associated gene expression patterns are shown in Fig. 6(a–j). Given that the 1408 ATAC-seq peaks identified across the four clusters originated exclusively from syncytium-containing root tissues, genes significantly upregulated and associated with these regions likely represent syncytium-enriched transcriptional activation events, highlighting the central role of chromatin state in shaping the syncytium transcriptome.

Promoter accessibility correlates with gene upregulation in Arabidopsis syncytium. (a–e) Representative genome-browser snapshots showing aggregate snATAC-seq signals in syncytium vs nonsyncytium adjacent root tissues (ART) controls at the promoters of AT2G05440 (a), AT5G65700 (b), AT1G26730 (c), AT4G36360 (d), and AT5G15970 (e). Chromatin regions that are significantly more accessible in the syncytium than control are indicated by red dotted lines. Tracks are normalized for library size. (f–j) Normalized gene expression from the snRNA-seq dataset for the corresponding genes: AT2G05440 (f), AT5G65700 (g), AT1G26730 (h), AT4G36360 (i), and AT5G15970 (j). Higher promoter accessibility in the syncytium is accompanied by increased transcript levels for these loci. (k) Examples of cis-regulatory elements most significantly enriched within snATAC peak regions associated with syncytium-upregulated genes from clusters 0, 10, 18, and 22. Each panel shows the motif logo, the assigned transcription factor family, and P value.
To gain mechanistic insight into transcriptional regulation in syncytial cells, we scanned peak regions associated with syncytium-upregulated genes in clusters 0, 10, 18, and 22 for enrichment of cis-regulatory elements to identify putative regulators of cell-type-specific transcription. The analysis revealed that cis-elements for bZIP, AP2, bHLH, and TCP factors are the most common in peak regions associated with upregulated genes (Fig. 6k; Table S14), suggesting a key role of these families of transcription factors in regulating gene expression in syncytium. The results also revealed overlapping as well as distinct patterns of cis element enrichment across the analyzed clusters. For example, 19 cis regulatory elements were similarly enriched among the peak regions identified in clusters 0, 10, and 18 (Table S14). Of these, 11 are binding elements for bZIP transcription factors. In addition, 23, 27, 3, and 2 regulatory elements were found to be uniquely enriched in clusters 0, 10, 18, and 22, respectively. The uniquely enriched cis-elements in each cluster revealed clear, cluster-specific transcription factor family signatures, suggesting distinct regulatory programs. Cluster 0 was strongly dominated by AP2 family motifs, with additional enrichment of bHLH, MYB, and ZF motifs, indicating a prominent role for AP2-related regulators. Cluster 18 showed enrichment for specific bHLH, MYB, and ZF motifs, whereas cluster 22 was characterized exclusively by MYB motifs, suggesting a more specialized regulatory module (Fig. 6k; Table S14).
Among the enriched cis binding elements identified above, a subset has been previously validated as binding motifs for 71 transcription factors spanning multiple families (Table S15). To assess the potential role of these transcription factors in regulating syncytium-associated gene expression, we screened syncytium upregulated genes from clusters 0, 10, 18, and 22 (Table S11) for overlap with the 71 transcription factors. Interestingly, 44 transcription factors overlapped. These 44 factors belong to the AP2 (11), bZIP (11), TCP (7), MYB (3), BZR/BES (3), bHLH (2), FAR1 (2), CAMTA (1), Trihelix (1), BASIC PENTACYSTEINE (1), CCT (1), and zinc finger (1) families (Table S16), suggesting key regulatory functions in reprogramming the syncytium transcriptome.
Homozygous T-DNA insertional mutants of four syncytium-upregulated transcription factors linked to ATAC-seq peaks were identified and used in nematode susceptibility assays to determine their impacts on plant susceptibility to H. schachtii. This included mutants of the transcription factors bZIP44 (AT1G75390; SALK_202170C), TGA7 (bZIP) (AT1G77920; GK-314D02 and GK-434F04), CAMTA1 (AT5G09410; SALK_008187C and SALK_108806C), and BZR1 (AT4G18890; SALK_006525C and SALK_017577C). Interestingly, all mutants displayed reduced nematode susceptibility relative to the wild-type Col-0 (Fig. 7a).

Reduced nematode susceptibility and syncytium size in Arabidopsis T-DNA insertion mutants for candidate syncytium regulatory factors. (a) Heterodera schachtii susceptibility was assessed in homozygous T-DNA insertion mutants for four transcription regulators identified as upregulated in syncytia, including bZIP44 (SALK_202170C), TGA7 (bZIP) (GK-314D02 and GK-434F04), CAMTA1 (SALK_008187C and SALK_108806C), and BZR1 (SALK_006525C and SALK_017577C). Nematode susceptibility levels were determined by counting the number of females per plant at 3 wk post-inoculation. Bars represent mean ± SE (n = 10). Statistical significance differences from wild-type Col-0 plants were determined using t test (P < 0.05) and are indicated by asterisks. Similar data were obtained from two independent experiments. (b) Mature syncytium size was measured at 10 d post-inoculation (dpi) in wild-type Col-0 and the indicated mutant lines. All mutant lines developed significantly smaller syncytia than wild-type plants, with reductions ranging from 23 to 40%. Bars represent mean of at least 20 syncytia ± SE. Statistical significance differences relative to wild-type Col-0 were determined using t-test (P < 0.05) and are indicated by asterisks.
To determine whether the reduced susceptibility could be attributed to differences in root growth, we next measured primary root length in all mutant lines. Root length was comparable to that of wild-type plants for all mutants except SALK_108806C, which exhibited a modest but statistically significant increase in root length (Fig. S7). We then examined whether these transcription factors contribute to syncytium development by measuring the size of mature syncytia at 10 dpi. All mutant lines produced significantly smaller syncytia than wild-type plants, with reductions ranging from 23 to 40% (Fig. 7b). Together, these findings indicate that transcription factors acting on enriched cis-regulatory elements within chromatin peaks are likely to function as central regulators of the transcriptional reprogramming required for syncytium development and nematode susceptibility.
In this study, single-nucleus RNA-seq and chromatin-accessibility datasets were analyzed separately using workflows optimized for each data type, improving sensitivity for transcriptional and regulatory signals. Although joint processing with Cell Ranger ARC identified only 336 control and 581 syncytium-associated nuclei, modality-specific processing followed by barcode-based matching recovered 5844 paired nuclei for the integrative analysis, indicating that the low number of jointly called nuclei reflected differences in cell-calling between the joint and modality-specific workflows rather than the number of nuclei available for downstream analysis. Given the distinct biological and technical properties of RNA and chromatin accessibility, independent analysis not only uncovered higher-resolution patterns within each dataset but also produced cross-validated discoveries whose concordance supports their functional significance and ensures robust and biologically meaningful conclusions.
The identification of nine distinct syncytium-associated clusters through snRNA-seq, annotated as cortex, xylem, phloem, pericycle, endodermis, and unknown (Fig. 1b) is consistent with the known induction of syncytia from procambial, pericycle, cortical, and endodermal cells during nematode infection (Jones & Northcote, 1972; Sobczak & Golinowski, 2008). This finding also highlights a possible functional heterogeneity within or between syncytia as nuclei within the multinucleated feeding structure may retain some signatures of their original cell lineages, reflecting varied identity of the fused cells. An alternative possibility is that some nuclei detected in syncytium-containing root segments originate from neighboring cortex, xylem, phloem, pericycle, and endodermal cells that remain outside the syncytium and whose transcriptional states are altered by the developing feeding site. However, we favor the possibility that the distinct transcriptional states reflect functional heterogeneity within or between syncytia, with nuclei retaining signatures of their original cell lineages following incorporation into the multinucleated feeding structure. This interpretation is consistent with the known cellular origins of syncytia (Jones & Northcote, 1972; Sobczak & Golinowski, 2008) and the multiple cell-type-associated signatures detected in these regions, although the current data cannot definitively distinguish nuclei within the syncytium from those in adjacent cells. On the contrary, the ‘unknown’ clusters may represent novel syncytium-related transcriptional reprogramming where certain cells may lose their original identity once fused to the syncytium rather than simply fusing as predefined cell types. This functional heterogeneity may suggest that the syncytium functions as a coordinated, yet compartmentalized, network in which individual nuclei contribute specialized activities. Compartmentalization within or between syncytia likely ensures that different cellular processes are optimized and coordinated, contributing to the overall functionality of this large, multinucleated feeding structure. However, because our analysis was performed on nuclei isolated from syncytium-containing tissues at a single time point, these observations should be interpreted with caution. The observed heterogeneity may reflect functional specialization, asynchronous nuclear states, or differences in the timing of nuclear incorporation into the syncytium, rather than definitive evidence of stable functional compartmentalization. Similar to nematode-induced syncytium, recent single-cell and single-nucleus transcriptomic studies in animal systems have revealed that multinucleated cells exhibit striking compartmentalization and functional heterogeneity. In mammalian skeletal muscle, for instance, it has been shown that individual myonuclei within the multinucleated myofibers possess distinct transcriptional profiles (Kim et al., 2020; Petrany et al., 2020). Similarly, osteoclasts, the multinucleated cells responsible for bone resorption, have also been found to exhibit nuclear heterogeneity (Tsukasaki & Takayanagi, 2022; Wang et al., 2022). Collectively, these findings underscore that even within a shared cytoplasmic environment, nuclei in fused syncytial cells can assume distinct, compartmentalized functions. Although we cannot distinguish nuclear heterogeneity within individual syncytia from variation among syncytia or biological samples, further spatial and temporal validation is needed to determine whether the transcriptional heterogeneity reflects stable spatial compartmentalization or transient developmental and physiological states.
GO term enrichment analysis (Fig. 2h) indicates that while both syncytium-enriched and syncytium-associated clusters share stress response and phenylpropanoid biosynthesis pathways, syncytium-enriched clusters are uniquely characterized by diverse metabolic, developmental, hormonal, and oxidative stress processes, whereas syncytium-associated clusters are distinguished by cell-wall reinforcement (lignin and very long-chain fatty acid biosynthesis) and responses to wounding and biotic stimuli. These findings point to a mechanism in which transcriptional profiles of syncytium-enriched clusters drive core cellular reprogramming within or between the feeding cells, whereas syncytium-associated clusters reflect shared defensive and structural responses operating at the host tissue interface. Comparable transcriptional patterns have also been reported in host plants during root-knot nematode infection (Ozdemir et al., 2024; Zadegan et al., 2025).
Complementing the clustering and enrichment results, pseudotime inference delineates a transcriptional continuum (Figs 4, S4). Given the single-time-point nature of our dataset (10 dpi), the inferred pseudotime ordering should be interpreted as a computational representation of transcriptional similarity rather than direct evidence of developmental progression. One end of the continuum was characterized by strong expression of cell-wall remodeling enzymes, including pectate lyases, expansins, and xyloglucan endotransglucosylases, which have been implicated in syncytium structural modifications (Bohlmann & Sobczak, 2014). Intermediate transcriptional positions were enriched for genes involved in transport, stress responses, and inositol and organic acid metabolism, consistent with active metabolic and physiological functions that support syncytium maintenance and activity (Rodiuc et al., 2014). Nuclei occupying alternative branches of the continuum showed enrichment for phenylpropanoid metabolism, suberin biosynthesis, long-chain fatty-acyl-CoA metabolic processes, and pathways associated with cell-wall reinforcement and tissue specialization (Zhang et al., 2017). These transcriptional patterns suggest distinct nuclear states and potential functional specialization among syncytial nuclei as well as possible transcriptional similarities between syncytial and neighboring cells. Such compartmentalization could support the complex metabolic, developmental, and physiological functions of fully developed syncytium. However, it may be important to indicate that the inferred continuum may also reflect asynchronous nuclear maturation and differences in the timing of nuclear incorporation into the syncytium.
To better understand the transcriptional landscape of syncytial cells, we integrated ATAC-seq and gene expression data to identify putative regulatory programs and transcription factors associated with cell-type-specific transcription. Our joint clustering of snRNA-seq and snATAC-seq enabled accurate transfer of cell-type labels to chromatin accessibility profiles, revealing that ATAC syncytium-rich clusters formed distinct, well-delineated groups in UMAP space (Fig. 5c), indicative of a specialized epigenetic program. This is consistent with recent findings that distinct root cell types exhibit unique patterns of chromatin accessibility (Dorrity et al., 2021). Gene expression analysis of four ATAC syncytium-rich clusters (0, 10, 18, and 22) resulted in the identification of 7467 cluster-specific DEGs. Notably, this set includes a larger cohort of 7110 DEGs that were not captured by snRNA-seq analysis alone, thereby expanding the catalogue of genes associated with the feeding-cell program to 8300 genes (Table S17). Of these 8300 genes, 5352 (64%) overlapped with DEGs reported from syncytial cells using micro-aspiration (Szakasits et al., 2009) or laser capture microdissection (LCM) (Liu & Mitchum, 2025). The extensive overlap with previously reported datasets establishes a conserved core transcriptional program that defines syncytium identity and validates the robustness of our approach across independent profiling methods. By contrast, the nonoverlapping genes may represent a previously unrecognized set of syncytium-associated candidates uncovered by the single-nucleus approach, whose enhanced resolution substantially broadens the transcriptome of syncytial cells.
Recent studies show that chromatin accessibility and transcript abundance of nearby genes are not perfectly coupled (Dorrity et al., 2021; Marand et al., 2021; Zhang et al., 2021; Nobori et al., 2023). By integrating snATAC-seq and snRNA-seq, we directly mapped accessible regions to transcriptional outputs and thereby defined distinct regulatory modules across syncytial states. Clusters 10 and 18 display tight chromatin-transcription coupling consistent with active regulatory programs, whereas cluster 0 shows more modest coupling, reflecting a distinct regulatory configuration. By contrast, cluster 22 contains the largest number of differentially accessible regions (3007) but minimal gene linkage (78 peaks linked to 58 genes; 2.6%). This pattern likely reflects regulatory priming, activation of distal enhancers, or context-dependent accessibility that precedes detectable changes in steady-state transcripts rather than immediate promoter-driven upregulation. However, the extensive chromatin accessibility observed in this cluster may also be influenced by the unique biology of syncytial nuclei, which undergo extensive endoreduplication and become highly polyploid during syncytium development. These changes could promote broad chromatin relaxation associated with increased nuclear volume and altered chromatin organization, independent of immediate transcriptional activation. In addition, nematode infection is known to induce extensive host chromatin remodeling through effector-mediated manipulation of epigenetic pathways (Hewezi, 2020). Consequently, some accessible chromatin regions may reflect structural or epigenetic reprogramming rather than direct transcriptional output. Additional regulatory mechanisms, such as epigenetic modifications, may contribute to the regulation of genes that are not directly associated with accessible chromatin regions. In this context, previous studies have shown that DNA methylation and histone modifications play important roles in regulating gene expression in H. schachtii-induced syncytia (Hewezi et al., 2017; Bennett et al., 2022). Integrative analysis across these clusters further identified 1408 ATAC peaks that are significantly more accessible in syncytial nuclei and that map to 1414 syncytium-upregulated genes (Table S13). Because these peaks derive exclusively from nuclei isolated from syncytium-containing root tissues, the linked genes most likely represent syncytium-enriched transcriptional activation events that are enabled or facilitated by local chromatin opening.
The linked accessible chromatin and increased transcript levels at the 1414 peak regions in the syncytium nominate candidate cis-regulatory elements and the transcription factors that bind them as the likely factors that govern regulatory programs and cell-state specification in syncytial cells (Fig. 6k; Table S14). We identified significantly enriched cis-binding elements for 71 transcription factors, with the majority of these factors belonging to the AP2, bZIP, bHLH, and TCP families. Notably, our results highlight clear differences in transcriptional regulatory architecture among ATAC-seq clusters, with AP2-driven regulation predominating in cluster 0, bHLH- and bZIP-associated control in cluster 18, broad combinatorial regulation in cluster 10, and a focused MYB-mediated program in cluster 22. This diverse repertoire of enriched binding elements not only provides a mechanistic basis for functional specialization within the multinucleated structure but also suggests a complex regulatory network that may integrate developmental signals with stress or defense signaling pathways required for syncytium maintenance and function. Members of these families have been previously shown to regulate different aspects of syncytium formation and development, including ROS homeostasis (TCP9) (Willig et al., 2022), syncytium development (MYB59, bHLH25, and bHLH27) (Jin et al., 2011; Ali et al., 2014; Wiśniewska et al., 2021), callose deposition to strengthen defense at the syncytial interface (AP1 member RAP2.6) (Ali et al., 2013), defense response and metabolic pathways (MYB83) (Piya et al., 2017) and hormone signaling (the MYB-related HRS1) (Wiśniewska et al., 2022).
Notably, although we identified 71 transcription factors whose binding motifs were significantly enriched in the promoters of syncytium-upregulated genes, 44 were transcriptionally upregulated in syncytium. Motif enrichment highlights potential DNA-binding sites but does not imply that the corresponding transcription factor mRNAs are also increased. Some factors may be present at low transcript levels below our detection threshold or may be regulated posttranscriptionally through alternative splicing, protein–protein interaction or phosphorylation (Jin et al., 2011; Pireyre & Burow, 2015; Zhang et al., 2023; Hawk et al., 2024; Ozdemir et al., 2024). Moreover, cis regulatory elements are often recognized by multiple related family members (Franco-Zorrilla et al., 2014; Weirauch et al., 2014; O'Malley et al., 2016), and thus the identified cis elements may represent binding sites for closely related transcription factor family members whose expression may be upregulated in syncytium. In accordance with this suggestion, we observed upregulation of several bHLH family members whose canonical E-box binding sequence ‘CACGTG’ was significantly enriched, suggesting that motif enrichment may reflect activity of alternative paralogs. A similar interpretation may apply to regulatory elements bound by bZIP, MYB, AP2, and TCP proteins, where motif enrichment does not necessarily pinpoint the single cognate factor. Additionally, it may be important to mention that promoter regulatory elements represent only one layer of gene regulation, and distal enhancers, silencers, and the presence or absence of essential co-factors all contribute critically to functional gene regulation.
Importantly, mutant lines of syncytium upregulated transcription factors acting on enriched cis-regulatory elements within accessible chromatin regions exhibited an intermediate but consistent reduction in susceptibility to H. schachtii (Fig. 7a). This reduction was not associated with major differences in root growth, as all mutant lines displayed root lengths comparable to wild-type plants except for one CAMTA1 allele, which showed a modest increase in root length (Fig. S7). By contrast, all mutants of bZIP44, TGA7, CAMTA1, and BZR1 developed significantly smaller syncytia than wild-type plants (Fig. 7b), indicating that these transcription factors contribute to syncytium development in addition to promoting nematode susceptibility, despite the direct regulatory relationships between these validated transcription factors and their downstream targets remaining to be experimentally validated. The partial nature of this effect likely reflects functional redundancy among transcription factors and the involvement of additional regulatory components that collectively coordinate syncytial transcriptome programming. Together, these findings support a model in which transcription factors within accessible chromatin function as key positive regulators of the transcriptional reprogramming required for syncytium development and nematode parasitism, while also highlighting the complex and multilayered regulatory networks underlying host susceptibility.
This work was supported by funds from the University of Tennessee, Institute of Agriculture and USDA-Agricultural Research Service (5070-21000-044-00D).
None declared.
TH conceived the project, designed and supervised the experiments, contributed to the experimental works, analyzed data and wrote the manuscript. SBZ analyzed data and contributed to the experimental work. PL, CÖ and JHR contributed to the experimental work. HBK contributed to project conceptualization and manuscript editing. All authors contributed to the article and approved the submitted version.
All sequencing data of snRNA-seq and snATAC-seq libraries generated in this study have been submitted to the National Center for Biotechnology Information (NCBI), Gene Expression Omnibus under the accession number GSE320172.
References
The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.
Texto de New Phytologist tal como lo publicó, leído de su página y mostrado aquí con el formato de El Autómata. El original, en nph.onlinelibrary.wiley.com · Volver a la noticia de El Autómata