Functions of Extracellular Vesicles in Immunity and Virulence

TECHNIQUES FOR EV ISOLATION

Given that EVs are heterogeneous in size and molecular composition and thus in density and charge, it is useful to discuss the different isolation methods, which may influence the nature of EVs. The standard technique and most widely used is differential ultracentrifugation with consecutive steps of low centrifugal forces (g) to discard cellular debris and high-speed forces (i.e. 40,000g and 100,000g) to collect EVs based on density (Li et al., 2017; Rutter and Innes, 2017). Differential ultracentrifugation is used as a standalone technique but more often combined with density gradient centrifugation (i.e. Suc and Optiprep) to reduce contaminants and thus purify refined EVs (Sunkara et al., 2016; Li et al., 2017; Rutter et al., 2017). This isolation technique was used to purify EVs from diverse cultured plant bacteria and fungi (Table 1), plant cells, and from plant extracellular fluids collected from leaves and seedlings (Table 2). Applying differential ultracentrifugation with or without density gradient centrifugation depends on the grade of desired purity and the purpose of EV isolation, i.e. proteomic and transcriptomic profiling. The choice of starting material and quantity should be carefully considered when purifying EVs. For example, medium should be collected from bacterial cultures in the exponential versus the stationary growth phase to reduce the amount of debris from dying cells, and extracellular plant (apoplastic) fluids should be used instead of plant fluids obtained by mechanical disruption of tissues as these will contain a mixture of extracellular and intracellular vesicles (Pérez-Bermúdez et al., 2017). EVs can also be isolated based on size by membrane filtration, i.e. size-exclusion chromatography, which was used to isolate EVs from plant-infecting Xylella fastidiosa subsp. pauca 9a5c (Santiago et al., 2016). Alternatively, EVs can be isolated using immunoaffinity capture and advanced imaging flow cytometry (Li et al., 2017; Mastoridis et al., 2018). However, the latter two approaches have not been described for the isolation of EVs from plants and plant-interacting microbes (Tables 1 and 2), particularly because they depend on suitable EV biomarkers.

TECHNIQUES FOR EV VISUALIZATION

Proper detection of EVs represents a major challenge. Initial discoveries of EVs in the interaction between plants and microbes were from transmission electron microscopy (TEM). EVs were observed to concentrate at bacterial and fungal infection sites, for example upon attempted hyphal penetration of Blumeria graminis f. sp. hordei into barley epidermal cells, and successful invasion of Golovinomyces orontii and Botrytis cinerea hyphae into Arabidopsis (Arabidopsis thaliana) leaf cells (An et al., 2006b; Micali et al., 2011; Solé et al., 2015; Cai et al., 2018b). TEM analysis also revealed plant MVBs fused with the PM (An et al., 2006b; Cai et al., 2018b), which suggested that EVs are released by MVB-mediated secretion. Consistent with this pathway of unconventional secretion, the EV proteome is enriched in proteins lacking a predicted signal peptide (Rutter and Innes, 2017).

In addition, most information on EV morphology and size is derived from TEM measurements, which reveal EVs as spheres ranging in size from 10 to 400 nm in plant-interacting bacteria and from 10 to 300 nm in plant fluids (Tables 1 and 2). Immunogold electron microscopy and cryo-electron microscopy provide additional information on EV contents and morphology (Wang et al., 2010; Prado et al., 2014; Solé et al., 2015; Sunkara et al., 2016). Whereas electron microscopy-based methods provide information on the purity of isolated EVs and whether the vesicles are intact, they require sample fixation and dehydration. This affects the true morphology and size of EVs, meaning that EVs could appear much smaller than that under hydrated conditions. Dynamic light scattering and nanoparticle tracking analysis measure EVs directly in solution and are therefore more likely to determine the true size of EVs and the size distribution of a population of vesicles (Sunkara et al., 2016). For example, the average diameter of EVs from Arabidopsis apoplastic fluids is around 12 nm (P100 fraction) and 100 nm (P40 fraction), as measured by TEM (Rutter and Innes, 2017). Yet, dynamic light scattering analysis revealed EV diameters ranging from 10 to 17 nm (P100 fraction) and 50 to 300 nm (P40 fraction), with the most abundant vesicles around 150 nm (Rutter and Innes, 2017). Moreover, nanoparticle tracking analysis is suited to quantify EVs (Ionescu et al., 2014). Atomic force microscopy also detects EVs in solution and was used to characterize X. fastidiosa ssp. pauca 9a5c EVs (Santiago et al., 2016).

Standard protein assays can determine total EV protein contents, such as that from bacterial EVs (Chowdhury and Jagannadham, 2013; Mendes et al., 2016), but may not reflect true protein amounts due to the vesicle’s nature. Most commonly, EVs are characterized by immunoblotting and immunogold electron microscopy based on biomarkers initially identified in proteomic profiling being enriched in EVs (Sidhu et al., 2008; Matsumoto et al., 2012; Chowdhury and Jagannadham, 2013; Regente et al., 2017; Rutter and Innes, 2017; Cai et al., 2018b). However, this is limited as very few biomarkers have been described for EVs of plants and plant-interacting microbes (Tables 1 and 2). For plant-derived EVs, biomarkers include tretaspanin8 (TET8) and molecular components related to their biogenesis, i.e. the penetration1 (PEN1) syntaxin and the exocyst subunit Exo70E2 in plant-derived EVs (Table 2). Components related to the outer membrane (i.e. XadA1) and secreted proteins (i.e. LesA and Ax21) are used as biomarkers for EVs from plant-interacting bacteria (Table 1). Notably, although most plant-derived EVs have a diameter between 10 and 300 nm (Table 2), which is considerably small for detection by standard light-based microscopy, green fluorescence protein (GFP) tagging of biomarkers enabled the direct visualization of EVs using confocal laser scanning microscopy upon EV isolation and in planta (Rutter and Innes, 2017; Cai et al., 2018b). To which extent this reflects the observation of single vesicles versus aggregates and/or the intrinsic brightness of genetically encoded and highly expressed GFP-fused biomarkers is unknown. In addition, dyes labeling EV lipids, such as FM4-64 and DiO, were used to visualize plant-derived EVs (Regente et al., 2017; Rutter and Innes, 2017; Cai et al., 2018b). As EV composition reflects that of their producing donor cells, these vesicles can incorporate nucleic acids, i.e. in the form of small interfering (s)RNAs, exemplified by micro (mi)RNA822 that was revealed by RT-PCR in plant-derived EVs positive for TET8 (Cai et al., 2018b). Given that there is no single biomarker that can uniquely identify EVs, multiple techniques are best to isolate and characterize EVs.

PRESENCE OF EVs IN PLANT-INTERACTING MICROBES

Some of the best-characterized EVs from microbes are those produced by Gram-negative bacterial pathogens that interact with humans. The vesicles assist in intermicrobial and host-microbe interactions, i.e. biofilm formation, surface attachment, and immunomodulation of host cells. The interaction outcome is determined through the delivery of molecules at higher concentration and over longer distances while being protected in a membrane sphere (MacDonald and Kuehn, 2012). By contrast, surprisingly little is known about EVs from plant-interacting microbes, although the presence of EVs has been observed in cultured phytopathogens and in plant samples infected with microbes (Table 1). Since the 1980s, EVs were observed by electron microscopy in cultures of Erwinia amylovora and E. carotovora, necrotrophic pathogens of the Enterobacteriaceae that cause fire blight in fruits of Rosaceae and bacterial soft rot in a wide range of vegetables, respectively (Laurent et al., 1987; Yaganza et al., 2004). Several hemibiotrophic pathogens of the Xanthomonadaceae release EVs in culture and during plant infection as determined by biochemical purification and electron microscopy (Table 1): (1) cultured Xanthomonas campestris pv campestris (strains 33913 and B100), the causal agent of black rot disease in crucifers, releases lipopolysaccharide (LPS)-positive EVs from the outer membrane (Sidhu et al., 2008; Bahar et al., 2016); (2) EVs from X. campestris pv vesicatoria strain 85-10, which causes bacterial leaf spot on tomatoes (Solanum lycopersicum) and pepper (Capsicum spp.), have been observed in bacterial cultures and during the infectious process in pepper leaves (Solé et al., 2015); (3) EVs have been purified from cultured X. oryzae pv oryzae strain PXO99, the causal agent of bacterial blight in rice (Oryza sativa; Bahar et al., 2016) and (4) cultured X. citri ssp. citri strain 306, which causes citrus canker (Ionescu et al., 2014); (5) EVs were observed in cultures and extracellular fluids from plants infected with Xylella fastidiosa ssp. fastidiosa Temecula 1, responsible for Pierce’s disease in grapevine (Vitis vinifera; Matsumoto et al., 2012; Ionescu et al., 2014; Nascimento et al., 2016); and (6) in cultured X. fastidiosa ssp. pauca (strains 9a5c and J1a12), causing citrus variegated chlorosis (Mendes et al., 2016; Santiago et al., 2016). Pathovars of the hemibiotroph Pseudomonas syringae, which infect Arabidopsis and tomato causing bacterial speck disease, were also reported to release EVs (Chowdhury and Jagannadham, 2013; Bahar et al., 2016; E. Stigliano, K. Rybak, M. Janda, S. Robatzek unpublished data), as well as Acidovorax citrulli M6, a biotrophic pathogen that causes seedling blight and bacterial fruit blotch in cucurbits (Bahar et al., 2016). The fact that plant-interacting Gram-negative bacteria produce EVs, which were also observed in planta during infection (Solé et al., 2015), suggests that these EVs are involved in cross kingdom communication between bacteria and plant cells (Fig. 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Schematic representation of EVs functioning in plant-microbe interactions. In brief, EVs released from microbes can exert immunogenic and immune-suppressive activities, potentially both inside plant cells and in the extracellular space, i.e. apoplast and EHMx, and also could play roles in microbe-microbe interactions. Plant-produced EVs function in antimicrobial immunity, i.e. delivering sRNAs and papillae formation. Details are explained in the text. Dashed lines and question marks refer to possible but not yet demonstrated functions. Blue coloring indicates roles of EVs that promote plant immunity; red coloring indicates roles of EVs in the potential promotion of plant susceptibility. PRR, Pattern recognition receptor; PTI, pattern-triggered immunity; EHM, extrahaustorial membrane; EHMx, extrahaustorial matrix.

Whether other plant-interacting bacteria such as pathogenic Agrobacteria, Ralstonia, and beneficial Rhizobia produce EVs is unclear. Similarly, EVs have not been observed from the many plant-interacting fungi and oomycetes, including pathogens accounting for serious economic losses in major crops. However, it was shown that both Phytophthora infestans and Magnaporthe oryzae discharge virulence-associated proteins that function inside plant cells, so-called cytoplasmic effectors, via unconventional protein secretion (Wang et al., 2010; Giraldo et al., 2013; Liu et al., 2014). Given that EVs participate in unconventional protein secretion and that yeast and other fungi associated with human diseases have been shown to release EVs (Miura and Ueda, 2018), it is plausible to speculate that plant-interacting oomycetes and fungi produce EVs during the infectious process. Indeed, we recently detected EVs produced from germinating spores of cultured M. oryzae (P. Carranca Nunes Rosa, E. Stigliano, N. Talbot, S. Robatzek, unpublished data). Supporting evidence of EVs released from plant-infecting fungi has been obtained by TEM analysis. In a susceptible interaction, numerous EVs are present in the extrahaustorial matrix, the space at the interface of the plant cell and the invading fungal haustorium, which remain of unknown origin (Micali et al., 2011; Fig. 1). Since abundant MVBs were observed in haustoria, it suggests that G. orontii could release EVs into the extrahaustorial matrix (Micali et al., 2011). Another insight into the potential presence of fungal EVs comes from the finding that necrotrophic B. cinerea produces sRNAs that silence plant genes for promoting pathogen infection (Weiberg et al., 2013; Wang et al., 2016, 2017), a process that could involve EVs for delivery of the sRNAs.

IMMUNE MODULATION BY MICROBIAL EVS

It has been recognized that microbes produce EVs to aid infection success when colonizing a host. X. fastidiosa ssp. fastidiosa Temecula 1 releases EVs to mediate bacterial detachment from plant cells, thus enabling movement along and between xylem vessels (Ionescu et al., 2014). EV production in X. fastidiosa ssp. fastidiosa Temecula 1 is suppressed by the diffusible signal factor, an unsaturated fatty acid that accumulates as bacterial numbers increase. Consequently, ΔrpfF mutants lacking diffusible signal factors exhibit hypervesiculation and are more virulent compared with wild-type strains (Newman et al., 2004). In this case, EVs represent a mechanism of regulating bacterial-to-plant cell attachment to promote explorative colonization.

Another benefit of EV production to pathogens is the secretion of virulence factors that provide protection against the degradative environment of extracellular plant fluids. Preliminary proteomic analysis identified that EVs of cultured X. campestris pv campestris strain B100 are enriched in virulence determinants such as cellulase and xylosidase (two cell-wall-degrading enzymes), components of the type-III-system and the secreted proteins (Sidhu et al., 2008). X. campestris pv vesicatoria strain 85-10 produces EVs that contain degradative enzymes, i.e. a putative protease and xylanase, and type-II-secreted virulence-associated proteins (Solé et al., 2015). Similarly, the type-II-secreted lipase/esterase LesA is present in EVs of cultured X. fastidiosa ssp. fastidiosa Temecula 1 (Nascimento et al., 2016). Consistent with a function as a virulence factor, ΔlesA mutants display reduced infection rates in grapevine. Preliminary proteomic analysis of EVs from cultured P. syringae pv tomato T1 identified type-III-secreted effectors, i.e. AvrA1 and HopI1, known to suppress plant immune responses (Chowdhury and Jagannadham, 2013). Interestingly, we observed that EVs purified from P. syringae pv tomato DC3000 cultures have immune-suppressing activities such that the vesicles impair prototypic pattern-triggered responses and full antibacterial immunity (E. Stigliano, K. Rybak, M. Janda, and S. Robatzek, unpublished data; Fig. 1). Although evidence for EVs in the secretion of virulence factors is growing and the presence of EVs has been observed during the infectious process in leaves (Solé et al., 2015), it is unknown whether microbial EVs discharge their cargo to the plant’s extracellular space and/or have the ability to interact with plant cells to deliver cargo into the cytosol (Fig. 1). It was also noted that the production and composition of EVs differed depending on culture conditions (Sidhu et al., 2008; Ionescu et al., 2014). This is in agreement with EV production representing an envelope stress response of Gram-negative bacteria, i.e. as experienced during host infection, so that the cargo of EVs is regulated (MacDonald and Kuehn, 2012). Therefore, it is possible that the EV cargo differs significantly when examined from extracellular fluids of infected plants versus that of cultured bacteria.

Abundant components of EVs are LPS and elongation factor Tu (EF-Tu), as shown for EVs of X. campestris pv campestris, P. syringae pv tomato T1, and X. oryzae pv oryzae, respectively (Sidhu et al., 2008; Chowdhury and Jagannadham, 2013; Bahar et al., 2016). Both LPS and EF-Tu represent microbe-associated molecular patterns (MAMPs), which activate pattern-triggered immunity (PTI) upon recognition by cognate plant-encoded immune receptors, i.e. the EF-Tu receptor (EFR; Boutrot and Zipfel, 2017). In agreement with the presence of MAMPs at EVs, the vesicles provoked prototypic PTI responses in plants. EVs from X. campestris pv campestris strain 33913 stimulated the production of reactive oxygen species (ROS), ion release, and expression of defense genes in Arabidopsis (Bahar et al., 2016). Representing a prototypic PTI response, the ROS burst induced by EVs was dependent on EFR and revealed that this immune response is triggered by EF-Tu present at EVs (Bahar et al., 2016). Since EF-Tu was detected in both cell-free supernatant and EV preparations, it indicates that this MAMP is likely present at the outside of EVs (Bahar et al., 2016).

Unlike the ROS burst, EV-induced defense gene expression was neither mediated by EFR nor by the immune receptors that recognize the bacterial MAMPs flagellin and peptidoglycan, as well as a MAMP specific to Xanthomonas (Bahar et al., 2016; Boutrot and Zipfel, 2017). However, EV-induced defense gene expression was partially dependent on BRI1-associated kinase1 (BAK1) and suppressor of BAK1-interacting receptor1 (SOBIR1) kinase (Bahar et al., 2016). Both BAK1 and SOBIR1 are essential coreceptors common to several PTI signaling pathways including EFR-mediated immunity (Boutrot and Zipfel, 2017). This suggests that EVs from X. campestris pv campestris contain several MAMPs, including yet unidentified patterns, which could be recognized by BAK1- and SOBIR1-dependent immune receptors. It is also possible that EV-induced defense gene expression involves recognition of LPS present at the vesicles, since BAK1 and SOBIR1 have not been implicated in perception of LPS (Sidhu et al., 2008; Boutrot and Zipfel, 2017). Interestingly, Proteinase K treatments, which would remove proteins from the outside of EVs, i.e. EF-Tu, improved the immunogenic activity of the EVs (Bahar et al., 2016). This suggests that nonproteinaceous MAMPs (i.e. LPS) could become more exposed from EVs, resulting in a stronger defense gene induction.

Induction of defense genes was observed for EVs from three other Gram-negative bacteria, X. oryzae pv oryzae, P. syringae pv tomato DC3000, and A. citrulli M6 (Bahar et al., 2016; E. Stigliano, K. Rybak, M. Janda and S. Robatzek, unpublished data). The ability of EVs to induce plant immune responses demonstrates their immunogenic potential. In addition, the activation of the immune response involves EFR, BAK1, and SOBIR1, which are cell-surface-localized receptors (Boutrot and Zipfel, 2017), thus suggesting that EVs interact with plant cells (Fig. 1). It is worth noting that the level of defense gene expression was less induced upon stimulation with EVs from P. syringae pv tomato DC3000 compared to that with EVs from X. campestris pv campestris, X. oryzae pv oryzae, and A. citrulli, although P. syringae pv tomato DC3000 produced large amounts of EVs (Bahar et al., 2016). Since EVs contain immunogenic determinants and virulence-associated proteins (Chowdhury and Jagannadham, 2013), EVs generate an immunomodulatory response in plant cells that does not directly mimic the effect of purified MAMPs or transgenic expression of an effector.

Bacterial EVs also perform functions in intermicrobial interactions (Fig. 1). EVs of X. fastidiosa ssp. pauca contain XfYgiT, a component of the toxin-antitoxin system known to regulate biofilm formation and be involved in the survival of X. fastidiosa ssp. fastidiosa Temecula 1 (Merfa et al., 2016; Santiago et al., 2016). Moreover, when microbial communities colonize a host, such as in the context of the plant microbiota (Durán et al., 2018), EVs could represent a mechanism by which the donor microbe influences the competitor microbe’s ability to adapt to the host environment (Barrett et al., 2011; Hammerschmidt et al., 2014), i.e. through promoting cell lysis (MacDonald and Kuehn, 2012). This way, intermicrobial competition in the root microbiota could involve EVs, observed as direct antifungal activities and redundant determinant of bacterial root commensals that protects plants against the detrimental root-derived filamentous eukaryotes (Durán et al., 2018). However, whether bacterial plant commensals produce EVs and whether the vesicles interact with other microbes of the microbiota and the host plant is not yet known.

PLANT DEFENSE BY EVs

Plant-derived EVs play also important roles in cell-to-cell communication with microbes. Several studies have shown that immune stress stimulates EV secretion from plant cells. Both infection with P. syringae pv tomato DC3000 bacteria and activation of immune signaling in response to the defense hormone salicylic acid increases EV abundance in Arabidopsis extracellular fluids (Rutter and Innes, 2017). An obvious function of EVs at the plant-pathogen interface is to serve as mechanical protection against invading pathogens (Fig. 1). Indeed, plant resistance to pathogen entry is linked to the polarized immune response (Kwon et al., 2008a). This pathway functions by concentrating EVs in the extracellular space beneath attempted sites of pathogen penetration, resulting in the formation of so-called papillae (Kwon et al., 2008a). When the pathway is altered by mutation in genes affecting vesicle fusion and secretion, pathogen entry is increased (Collins et al., 2003; Kwon et al., 2008b; Nielsen et al., 2012). Consistent with serving as mechanical protection, papillae contain callose, which is a β-1,3-glucan cell wall polymer (Xu and Mendgen, 1994; Meyer et al., 2009) that contributes to resisting pathogen entry (Assaad et al., 2004; Nielsen et al., 2012). Given that callose-containing MVBs have been observed in infected cells (Xu and Mendgen, 1994), EVs could be involved in the delivery of papillary callose. In agreement, the callose synthase GSL5/PMR4 is found at papillae together with the PEN1 syntaxin, a biomarker of EVs (Meyer et al., 2009; Rutter and Innes, 2017). This polarized immune response pathway is conserved in divergent plants—barley (Hordeum vulgare), bean (Phaseolus vulgaris), and Arabidopsis—and functions in a similar fashion against diverse fungal pathogens, i.e. Blumeria graminis f. sp. hordei, Erysiphe graminis, and Uromyces vignae (Zeyen and Bushnell, 1979; Xu and Mendgen, 1994; An et al., 2006a, 2006b).

Specific components of Arabidopsis EV cargo have recently been identified and shown to be involved in antimicrobial defense, including the glucosinolate transporters PEN3 and NTR1 as well as the myrosinase EPITHIOSPECIFIER MODIFIER1 (Rutter and Innes, 2017). Indole glucosinolate metabolism has been associated with innate immunity in response to diverse fungal pathogens as well as insects (Zhang et al., 2006; Bednarek and Osbourn, 2009; Hiruma et al., 2010; Sanchez-Vallet et al., 2010; Campe et al., 2016). PEN3 is also required for immunity against P. syringae pv tomato DC3000 bacteria (Xin et al., 2013). Given that the myrosinase-glucosinolate defense system is normally compartmentalized so that hydrolysis only occurs upon pathogen attack (Shirakawa and Hara-Nishimura, 2018), we can speculate that different types of EVs are released. In addition, proteins involved in immune signaling, i.e. BIR2, GRP7, RIN4, SOBIR1, and the polarized immune response pathway, i.e. PEN1, SYP122, SYP132, are cargoes of Arabidopsis EVs (Rutter and Innes, 2017). Their association with EVs could indicate that the vesicles play roles related to regulation of cellular homeostasis during immune signaling and regulation of EV release. Interestingly, EV abundance, but not protein composition, changed significantly when vesicles were purified from P. syringae pv tomato DC3000-infected leaves (Rutter and Innes, 2017), suggesting that a default secretory pathway is adopted for immune responses (Kwon et al., 2008b).

Recently, sRNAs present in EVs, including vesicles positive for TET8, were characterized, showing that EVs of Arabidopsis were able to mediate a response in fungal cells (Fig. 1) in an sRNA-dependent manner (Baldrich et al., 2018; Cai et al., 2018b). TAS1c-siR483, TAS2-siR453, IGN-siR1, and miRNA166 were enriched in EVs purified from extracellular fluids of Arabidopsis infected with B. cinerea compared with that in EVs from uninfected plants (Cai et al., 2018b). Pathogen infection therefore modifies the quality of EVs released by Arabidopsis in addition to their quantity (Rutter and Innes, 2017). The same sRNAs accumulated in fungal cells and were shown to silence B. cinerea target genes during infection (Cai et al., 2018b). Accumulation of sRNAs in fungal cells was lost in tet8 tet9 double mutants, suggesting that sRNA transfer to fungal cells involved these tretraspanins (Cai et al., 2018b). Moreover, sRNAs present in EVs were protected against degradative conditions, indicating their intraluminal localization. Consistent with a role in delivering Arabidopsis sRNAs, EVs, including vesicles positive for TET8 and TET9, concentrated at fungal infection sites and accumulated inside Botrytis cells (Fig. 1). In this context, evidence lines up that sRNAs shown to function in host-induced gene silencing, which is the transgenic expression of double stranded RNA in plants to silence genes in pests and pathogens (Knip et al., 2014), are loaded into EVs and thereby transferred to the infecting organism (Cai et al., 2018a). Notably, abundant sRNAs derived from the same mRNA precursors, such as TAS1c-siR483 and TAS2-siR453, were not enriched in EVs (Cai et al., 2018b), suggesting that sRNA loading of EVs involves a selective process. A recent study supports the selective sRNA loading of EVs, albeit not for TAS1- and TAS2-derived sRNAs but for a specific group of miRNAs (Baldrich et al., 2018). This study also discussed a role of EVs as waste disposal for tiny sRNAs (Baldrich et al., 2018). Since at least two types of EVs, PEN1- and TET8-positive, have been identified from plant extracellular fluids (Rutter and Innes, 2017; Cai et al., 2018b), it will be interesting to see whether PEN1- and TET8-positive EVs represent overlapping or distinct pools of EVs, i.e. related to their purification using 40,000g versus 100,000g (Rutter and Innes, 2017; Cai et al., 2018b), and whether this changes upon pathogen infection.

• Download figure
• Open in new tab
• Download powerpoint