Fluorescence microscopy was performed as described in Garvetto et al. (2023). In brief, the samples were observed using a Nikon Eclipse Ti2-E (Nikon, Japan) microscope equipped with an Andor Zyla 5.5sCMOS monochrome camera (Andor Technology, United Kingdom) using Nikon CFI Plan-Fluor 40 × /0.75 NA and 60 × /0.85 NA objectives. The excitation wavelengths for Hoechst 33342 and FISH probes were 365 and 490 nm, respectively. Negative controls without the probe but with hybridization buffer and Hoechst 33342 were included. Overlays of the different channels (DIC, channel for Hoechst, and channel for the FISH probe) and measurements were conducted using the NIS Elements software AR 5.21.03 (Nikon, Japan).

The root galls of Brassica rapa subsp. pekinensis and the roots of healthy control plants were rinsed with tap water. The samples were preselected and screened under the microscope for infections. Transmission electron microscopy was performed as described in Garvetto et al. (2023). Selected samples were chemically fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer containing 10% sucrose at 4°C for 1 h. They were washed with cacodylate buffer and post-fixed with 1% osmium tetroxide in 0.05 M cacodylate buffer for 1 h at 4°C. This was followed by another washing with cacodylate buffer. After dehydration with an increasing acetone series, the samples were embedded in Embed 812 resin. A diamond knife (Diatome, Switzerland) and an Ultracut UCT (Leica, Austria) were used to cut cross-sections of uninfected control roots and infected root galls. The samples were mounted on grids and stained with lead citrate. A Libra 120 energy filter transmission electron microscope (Zeiss, Germany) equipped with a TRS 2 × 2k high-speed camera (Tröndle, Germany) and ImageSP software (Tröndle, Germany) was used for imaging.

Flow cytometry was performed as described in Suda et al. (2007) with some modifications explained in detail in the Supplementary Methods. The used standards (Bellis perennis for Brassica rapa subsp. pekinensis and Solanum pseudocapsicum for Ectocarpus siliculosus) were used because of their similar but not overlapping genome size with the used material.

Three publicly available RNA-seq datasets were analyzed to examine the cell cycle-related genes in phytomyxid-infected hosts. The first was from Brassica oleracea subsp. gongylodes infected with Plasmodiophora brassicae [(Ciaghi et al., 2019); BioProject: PRJEB26435], the second from Brassica rapa subsp. pekinensis [(Jia et al., 2017); BioProject: PRJNA322393], and the third from Ectocarpus siliculosus Ec32m (strain CCAP 1310/4) infected with Maullinia ectocarpii [strain CCAP 1538/1; (Garvetto et al., 2023); BioProject: PRJNA878940]. The inferred proteomes were searched for cell cycle-related genes using identity thresholds of above 50% of peptides and above 80% for transcripts. Keyword searches based on gene models from E. siliculosus to B. oleracea and vice versa were used to identify additional potential cell cycle-related homologous. Additionally, log₂fold change values were extracted (Supplementary Tables 2–4). The most important genes involved in endoreduplication are summarized in Supplementary Table 1 and compared to the literature. A more detailed description of the pipeline can be found in the Supplementary Methods.

R studio vs. R.4.3.1 was used to analyze the nuclear measurement data. The Brown–Mood median test (accounting for data not normally distributed and unequal variances, from package “coin”) was used as a non-parametric alternative to the Student's t-test to assess whether the differences between the medians were significant. Additionally, a Wilcox–Mann–Whitney test (“rstatix” package) was used to assess whether the distributions of the nuclear areas were significantly different. Violin plots were used to display the differences between the distributions of the nuclear areas of infected and uninfected host cells using the package “ggstatsplots” (Patil, 2021).

The nuclei of Brassica rapa subsp. pekinensis roots from healthy plants, not infected with P. brassicae, were oval to round in shape and showed little variation in overall size and shape (Figures 1A, 2A, A′, Supplementary Figure 1A). The median nuclear area of uninfected plant cells was 19.72 μm² [standard deviation (SD) = 8.5, n = 65] with a minimum of 8.76 μm² and a maximum nuclear area of 47.54 μm² (Figure 3A, Supplementary Figure 5). The median nuclear area of B. rapa cells colonized by P. brassicae was 55.97 μm² (SD = 48.9, n = 65), with a minimum of 14.03 μm² and a maximum of 278.93 μm² (Figure 3A). The difference between the nuclear area of cells from infected and uninfected plants was highly significant (based on the Brown–Mood median test, p = 2.2e-16), and the nuclear area of cells from P. brassicae-infected plants was 2.8 times bigger based on the median size. While the nuclei of non-colonized root cells in healthy plants were similar in size and round to oval in shape, the nuclei of cells colonized by P. brassicae from infected plants varied in size and had a convex, bulged appearance (Figures 1A, B, 2A–D, A′-D′, Supplementary Figures 1A, B). During the colonization of cortical cells of the plant, the plasmodium of P. brassicae was gradually growing and occupying more and more space within the host cell. The colonized host cells were increasingly hypertrophied, and their nuclei became larger over time (Figures 2B, B′, C, C′). When the colonized cell was completely filled with resting spores of the parasite (Figures 2D, D′), the host nucleus disappeared.

Nuclei in healthy E. siliculosus (Ec32m) cultures had a uniform size and shape (Figures 1C, 2E, E′, Supplementary Figure 1C). The median nuclear area of E. siliculosus cells was 11.03 μm² (sd=2.5, n=171) with a minimum of 6.8 μm² and a maximum of 25.88 μm² (Figure 3B, Supplementary Figure 5). The median nuclear area of E. siliculosus cells colonized by M. ectocarpii was 13.34 μm² (SD = 4.8, n = 171) with a minimum of 6.78 μm² and a maximum of 32.65 μm² (Figure 3B). Based on the median sizes, the nuclei of colonized cells were 1.2 times bigger than in unaffected cells, and the difference was highly significant (based on the Brown–Mood median test, p = 2.15e-11). Unaffected brown algae cells had spherical to ellipsoidal nuclei with little variation in size, while nuclei of colonized cells showed a high variation in size and a higher variation in shape (Figures 1C, D, 2E–H, E′-H′, Supplementary Figures 1C, D). The nuclei of infected brown algae gradually increased in size as the parasite plasmodium developed. Cells colonized by a young plasmodium retained a normal-sized nucleus (Figures 2F, F′), while cells with a mature plasmodium exhibited enlarged nuclei (Figures 2G, G′). When the plasmodium differentiated into zoospores and filled the entire host cell, the host nucleus began to disappear (Figures 2H, H′).

We observed both, plasmodia and resting spores of Maullinia braseltonii in the tissue between the cortex and the medulla of infected Durvillaea blades. The median nuclear area of healthy Durvillaea incurvata cells was 6.21 μm² (SD = 2.4, n = 30) with a minimum of 2.12 μm² and a maximum of 13.43 μm² (Figure 3C, Supplementary Figure 5). The nuclei of D. incurvata cells colonized by M. braseltonii appeared enlarged (Figure 1F, Supplementary Figure 1F′). The median nuclear area of infected Durvillaea cells was 17.51 μm² (SD = 5.8, n = 30) with a minimum of 9.68 μm² and a maximum of 32.15 μm² (Figure 3C). The difference between host nuclei in colonized and unaffected cells was highly significant (based on the Brown–Mood median test, p = 7.551e-13). The nuclei of colonized cells were 2.8 times bigger than the nuclei of unaffected cells (based on the median of the nuclear area).

The TEM images confirmed that the host nuclei of infected B. rapa were not apoptotic, but that the increase in size was because of endoreduplication. The TEM images showed that the host nucleus of the colonized B. rapa cell had an intact membrane without holes as would be expected in case of apoptosis. Nucleoli and heterochromatin were present, indicating an active host nucleus (Supplementary Figures 2, 3).

Flow cytometry analysis of infected plant and algal material (together with healthy controls) supported the findings of the nuclear measurements and microscopy. B. rapa infected with P. brassicae showed ploidy levels of 8C and sometimes a few 16C nuclei were detected (Supplementary Figure 4B), while the roots of control plants had ploidy levels of 4C and rarely 8C (Supplementary Figure 4A). In the healthy E. siliculosus samples, only one ploidy level was detected (Supplementary Figure 4C). In the infected E. siliculosus samples, different ploidy levels could be detected; however, the interpretation of the peaks should be approached with caution, as the peaks are often not well separated and ambiguous (Supplementary Figure 4D).

By querying available RNA-seq datasets, we could find genetic signatures pointing toward the induction of endocycle-related processes in Brassica oleracea subsp. gongylodes and B. rapa subsp. pekinensis infected with P. brassicae (data from Ciaghi et al., 2019; Jia et al., 2017). In those datasets, the transcripts linked to the switch from the mitotic cell cycle to the endocycle (as described for A. thaliana by Olszak et al., 2019; Supplementary Tables 1–3). The changes involved the upregulation of the cell cycle switch protein CCS52A1 and the upregulation of genes for the progression from G1 to S phase. In contrast, no definitive pattern was identified for the G2/M transition (Supplementary Tables 1–3).

The genetic and molecular mechanisms of the endocycle in brown algae are not fully known, but they are thought to be conserved in eukaryotes (Bothwell J. H. et al., 2010). Therefore, we analyzed the publicly available RNA-seq dataset of E. siliculosus infected with M. ectocarpii (Garvetto et al., 2023). Note that the term Ectsi refers to transcripts of Ectocarpus siliculosus. Transcripts of Ectsi FZR1 (or CDH1/CCS52), a homolog to the positive endocycle regulator CCS52A in plants, were upregulated in infected E. siliculosus (Ec32m) (Supplementary Table 1). Wee1 (the plant homolog WEE1 is important for endocycle onset in specific tissues in plants) was downregulated in infected algae (Supplementary Table 1). Transcripts of the anaphase-promoting complex/cyclosome (APC/C), which is activated through CCS52A, were both up and downregulated (Supplementary Table 4). The expression of genes responsible for regulating the G2-M transition during mitosis was found to be downregulated in infected algae when compared to uninfected hosts. Ectsi CDKA2/CDKB, whose diatom homolog is important for the G2-M transition, was downregulated in infected brown algae (Supplementary Table 1). Both A-type cyclins and B-type cyclins were downregulated in infected algae (Supplementary Table 4). CDKA1, important for the G1/S transition (i.e., DNA replication), was upregulated in M. ectocarpii-infected E. siliculosus (Supplementary Table 1). There was an upregulation of certain transcripts of D-type cyclins, whose plant homologs are involved in the G1/S transition, while others were downregulated. Additionally, E2F, a positive transcriptional regulator of the G1/S transition, was upregulated in infected algae, while RBR, its inactivator, was downregulated (Supplementary Tables 1, 4).

The predicted proteomes of M. ectocarpii and P. brassicae were filtered for putative effectors and those were further filtered by the COG category “cell cycle” (D), to identify putative effectors interacting with the cell cycle regulation of their host to actively induce endoreduplication (Supplementary Tables 5, 6). Based on high expression levels (represented by TPM normalized counts) and cell cycle-related annotation of the transcripts we identified two potential effectors in P. brassicae; a mitotic checkpoint protein (BUB3) and a serine threonine kinase (AURKA) and in the M. ectocarpii transcriptome a putative Anaphase complex subunit 10 (ANAPC10) and a MOB kinase activator (MOB1).

We demonstrate local endoreduplication in plant and stramenopile host cells colonized by phytomyxids, hinting at a central role of this altered physiological state of the host cells for the parasite (Figure 4). During phytomyxid infection, endoreduplication plays a crucial role in creating hypertrophied cells and consequently allows the parasite to use more space and create a nutrient sink for itself. Both the sporangial phase (Maullinia ectocarpii) and the sporogenic phase (P. brassicae and M. braseltonii) show endoreduplication, but the effect appears stronger during the sporogenic phase. Endoreduplication is an important mechanism involved in plant growth and development (Joubès and Chevalier, 2000; Lee et al., 2009), and changes to the plant endocycle are involved in the successful colonization of plants by biotrophic pathogens and mutualists (Carotenuto et al., 2019; De Almeida Engler et al., 2012). An increase in the size of host nuclei and cells is an indicator of active endoreduplication because DNA in the nucleus is multiplied, but the cells do not divide (Carotenuto et al., 2019; Sugimoto-Shirasu and Roberts, 2003). Measurements of the size of the nuclei, analysis of the ploidy of cells, and transcriptome data of infected material support our hypothesis that endoreduplication is induced by phytomyxids in their host and that this process is conserved in plants and brown algae. According to the presented findings, host endoreduplication is linked to the growth of all Phytomyxea, likely inducing an energy sink in the host that causes the energy transfer from the host to the parasite. This expands previous findings in P. brassicae to other Phytomyxea (Malinowski et al., 2019; Olszak et al., 2019).

By examining the changes in colonized host cells during the phytomyxid life cycle, it has been observed that the nucleus size of brown algae and plant cells increases following the growth of phytomyxid plasmodia (Figure 2). Endoreduplication is induced when a plasmodium colonizes a host cell and is maintained until the plasmodium differentiates into resting spores/zoospores (Figure 2), supporting the hypothesis that the induction of local endoreduplication is involved in generating a nutrient sink for phytomyxids (Malinowski et al., 2019). Cells that undergo endoreduplication show an increase in transcription and metabolic activity, making them an energy sink within the plant (Bourdon et al., 2012; Lang and Schnittger, 2020; Lee et al., 2009). This energy sink is exploited by Phytomyxea until they form resting spores (Figures 2D, H) when metabolic activity in the parasite is likely ceased. The nucleus is likely phagocytized by the phytomyxid shortly before the resting spores are differentiated (Garvetto et al., 2023). Exploiting of endoreduplication is therefore likely a common, and very basal feature of Phytomyxea–host interaction, as it is present in plants and brown algae.

Phytomyxea have two functionally different types of plasmodia: the sporangial (primary) plasmodia are formed during the initial phase of infection and can be found in main and alternative hosts (Neuhauser et al., 2014). Sporogenic (secondary) plasmodia are only found in main hosts (Neuhauser et al., 2011). The data we present indicate that both types of intracellular plasmodia trigger local endoreduplication in their respective host cells. Sporogenic plasmodia of P. brassicae and M. braseltonii showed 2.8 times enlarged nuclei in cells where multinucleate parasite plasmodia were present (Figures 1, 3). The sporangial plasmodia of M. ectocarpii induced a significant, yet smaller endoreduplication effect on the host than the closely related M. braseltonii or the plant colonizing P. brassicae (1.2 times enlarged nuclei, Figures 1, 3). Because of the limitations in obtaining infected material where both stages are present concurrently, we were unable to measure the effect of both sporangial and sporogenic plasmodia in the same host. Despite this limitation, different degrees of endoreduplication are linked to biological differences between the two plasmodial stages. The sporangial phase of the phytomyxid life cycle involves relatively short-lived and small plasmodia that colonize the root hairs (in plants) and filamentous thalli/gametophytes (in brown algae) of their host (Kageyama and Asano, 2009; Maier et al., 2000). In comparison, in P. brassicae, the completion of the sporogenic part of the life cycle takes ~20–40 days. The faster transition of sporangial plasmodia from infection to zoospore formation limits the number of endoreduplication cycles in the colonized host cell.

The sporogenic part of the phytomyxid life cycle is strongly linked to hypertrophy in the host, at the level of isolated cells, but also at the level of hypertrophied areas of tissue and macroscopic galls (Karling, 1968; Kolátková et al., 2021; Neuhauser et al., 2010). Infected cells in these galls are filled with large, multinucleate plasmodia with sometimes hundreds of nuclei, and it has been estimated that one large clubroot can contain billions of resting spores (Hwang et al., 2013; Liu et al., 2020). To produce such large amounts of spores and such large plasmodia, time and energy are needed for the parasite to grow, which is reflected in the longer duration of this part of the life cycle. It is unclear if the larger host nuclei in the sporogenic phase are due to a specialized interaction between phytomyxid and host (Figure 3); or if they are indirectly caused by the prolonged physical interaction period between plasmodia and host cells (Figure 4). Endoreduplication as a conserved feature of phytomyxid–host interaction could have different biological constraints or drivers. Larger plasmodia such as those seen during sporogenic growth can be interpreted as a result of the increased tolerance of specific tissues for endoreduplication (Bothwell J. H. et al., 2010; Garbary and Clarke, 2002). Tissues with greater tolerance for endoreduplication allow for larger, faster-growing cells, ultimately leading to more energy transfer and longer growth time for the parasite, which enables larger plasmodia formation.

The microscopic evidence of endoreduplication in infected host cells is supported by existing transcriptome data, suggesting that Phytomyxea induce endoreduplication via CCS52 and/or WEE1 kinase (Supplementary Table 1). In plants, the endocycle has been extensively studied and can be induced through different pathways including the activation of the APC/C and the consequent inactivation of mitotic cyclins through cell cycle switch proteins (CCS52) in leaves, trichomes, and roots (Cebolla et al., 1999; Heyman et al., 2017; Lammens et al., 2008); the activation of the WEE1 kinase in tomato fruit (Gonzalez et al., 2007); and the inhibition of CDKs by SIM/SMR proteins in leaves and trichomes (Kasili et al., 2010; Kumar et al., 2015). The molecular basis of endoreduplication in brown algae is less well understood (Bothwell J. H. et al., 2010; Garbary and Clarke, 2002). A variety of cell cycle switch proteins was upregulated in infected B. oleracea subsp. gongylodes and B. rapa subsp. pekinensis plants compared to the control plants (Supplementary Table 1) similar to a targeted study in A. thaliana (Olszak et al., 2019). A homolog of plant cell cycle switch proteins was differentially regulated in E. siliculosus infected with M. ectocarpii. The observed patterns are similar to findings of galls during biotrophic nematode infections (De Almeida Engler et al., 2012), or during arbuscular mycorrhiza symbiosis (Carotenuto et al., 2019), or in rhizobia-induced nodules in soybean (Fan et al., 2022). Transcriptional activation of the WEE1 kinase inhibits CDK activity, which subsequently induces endoreduplication (Gonzalez et al., 2007). The regulatory pathway involving the WEE1 kinase was also activated in plant hosts infected with phytomyxids (Supplementary Table 1) (Olszak et al., 2019). In M. ectocarpii-infected E. siliculosus, however, we detected a decrease in the expression of WEE1 homologous transcripts. Endoreduplication induced by WEE1 is very specific and appears to be restricted to specific tissues or organisms; for example, it controls endocycle onset in tomato fruit and maize endosperm, but not in A. thaliana leaves (Bhosale et al., 2019; De Veylder et al., 2011). However, the role of WEE1 has not been validated in brown algae and may differ. In mammals, for example, WEE1 regulates the cell cycle and is important during DNA damage checkpoints (Elbæk et al., 2020). WEE1 may also be involved in DNA checkpoint control in Arabidopsis during nematode infection (Cabral et al., 2020).

The mechanism behind the Phytomyxea-induced endocycle is still unknown (Malinowski et al., 2019; Olszak et al., 2019) but our findings allow for a more comprehensive debate. Phytomyxea can induce endoreduplication either actively via effector molecules that target the cell cycle machinery or passively through the mechanical stress/tension caused by the growing plasmodium (Toruño et al., 2016). Biotrophs, such as arbuscular mycorrhiza fungi, powdery mildew, nematodes, and bacteria including rhizobia are known to use effectors to manipulate the host cells (Goverse and Smant, 2014; Rafiqi et al., 2012). Effectors are key to understanding the pathogen–host interaction, therefore potential effectors of P. brassicae are well-studied (Muirhead and Pérez-López, 2022; Pérez-López et al., 2020; Rolfe et al., 2016). Our search for effector candidates that interact with cell cycle proteins identified two putative effectors in P. brassicae and M. ectocarpii, respectively. Those potential effector candidates together with the results of Pérez-López et al. (2020), who identified a P. brassicae cyclin as a putative effector, highlight the potential of effectors in manipulating the host cell cycle. This makes putative effectors targeting host processes related to the induction of the endocycle promising targets for future studies to better understand phytomyxid–host interaction.

The second viable hypothesis for the induction of endoreduplication is mechanic stress caused by the growth of the intracellular parasite, indirectly keeping the cells “locked” in the endocycle stage until the plasmodium produces spores. Cells in the root cortex undergo one or two rounds of endoreduplication during normal root growth (Bhosale et al., 2019). Physical activation of the endocycle has been suggested as the mechanism for biotrophic nematodes (de Almeida Engler and Gheysen, 2013). It is also discussed that P. brassicae induces wall stress and cell expansion in its host (Badstöber et al., 2020b), resulting in alterations in the cell cycle (Olszak et al., 2019). Endoreduplication is linked to cell growth and cell wall remodeling (Bhosale et al., 2019; Bhosale and Vissenberg, 2023), so the growth of the parasite could indirectly induce the host to undergo additional rounds of endocycling, with the resulting energy sink providing nutrients for the Phytomyxea. The regulation and induction of endoreduplication in brown algae are even less clear, as studies on the topic are scant (Bothwell J. H. et al., 2010; Garbary and Clarke, 2002). However, based on the findings discussed here, we hypothesize that similar active and passive processes could be involved in the induction of endoreduplication in all hosts of Phytomyxea.

Endoreduplication plays a role in cell growth, and ploidy level often correlates with cell size (Breuer et al., 2007; Chevalier et al., 2011; Sugimoto-Shirasu and Roberts, 2003). Increased size of the host cells provides space for the growth of the plasmodia, as demonstrated by galls formed in A. thaliana ccs52a1 mutants with deficient endoreduplication, which were found to be significantly smaller than those of wild-type A. thaliana (Olszak et al., 2019). Therefore, the increased metabolic activity of the host allows for higher energy transfer and combined with larger host cells provides more space and energy to be translated into the growth of the phytomyxid parasites.