Human microsporidian pathogen Encephalitozoon intestinalis impinges on enterocyte membrane trafficking and signaling

Microsporidia are a phylum of eukaryotic obligate intracellular pathogens of animals and man. They form environmentally resistant spores transmissible to new hosts through food or water (Cali and Takvorian, 2004, 2014; Fayer and Santin-Duran, 2014). Phylogenetic analysis suggests that microsporidia are closely related to Fungi and are closely related to the Cryptomycota (Keeling et al., 2014; Han and Weiss, 2017). Over 200 genera and 1400 species of microsporidia have been described, and they are found in every major animal group including humans (Becnel et al., 2014).

In humans, microsporidiosis occurs worldwide. The prevalence and geographic distribution of infection are variable due to diagnostic methods and population demographics (Bednarska et al., 2014; Didier and Weiss, 2006; Fayer and Santin-Duran, 2014). Nine genera, containing 17 species, have been identified to cause microsporidiosis in humans (Fayer and Santin-Duran, 2014; Han et al., 2020). Gastrointestinal infection is the most common presenting symptom and this is usually due to either Enterocytozoon bieneusi or Encephalitozoon intestinalis. Enterocytozoon bieneusi is the cause of most human infections and it infects the biliary tract and small intestine, causing chronic diarrhea (Desportes et al., 1985; Orenstein et al., 1990). Unfortunately, it has not been successfully propagated in cell culture or laboratory animals including immune-suppressed mice (Visvesvara, 2002), greatly limiting its study.

Encephalitozoon intestinalis, formerly known as Septata intestinalis (Cali et al., 1993; Hartskeerl et al., 1995) can cause intestinal infection with associated dissemination to other organs. Encephalitozoon intestinalis infects both immune competent and immune-compromised individuals. Interestingly, E. intestinalis infection was found in 30% of patients with Crohn's disease (Andreu-Ballester et al., 2013). Encephalitozoon intestinalis infects enterocytes, its primary target cell type, causing diarrhea. In immune-compromised individuals, especially in those with advanced AIDS, E. intestinalis infection can lead to severe enteritis, disseminating infections in kidney and in hepatobiliary tract (Cali et al., 1993; El Fakhry et al., 2001; Hamamci et al., 2015; Raynaud et al., 1998; van Gool et al., 1997). It can also be responsible for eye and liver infections (Didier and Weiss, 2006). While Enterocytozoon bieneusi has not been propagated continuously in culture, E. intestinalis, like other Encephalitozoon species, grows in some mammalian cell lines, such as RK13, but continuous culture in human enterocyte has not been established.

Enterocytes are the mature intestinal epithelial cells that line the small intestinal and colonic mucosa. They are critically important for nutrient absorption and their barrier function. As intestinal epithelium is constantly exposed to chemicals, pathogens and assorted damaging materials, mature epithelial cells are sloughed off and undergo constant renewal from stem cells that undergo a cycle of differentiation. One identifying feature of differentiated enterocytes is the presence of apical arrays of microvilli that form brush borders. The nuclei of these polarized enterocytes are located in the basal portion of the cells (Crawley et al., 2014). Caco2 cells are a human enterocyte-like colonic epithelial cell line that undergoes differentiation and polarization (Gao and Kaestner, 2010; Knowles et al., 2015). Two initial studies of E. intestinalis utilizing Caco2 cells described the spore–host cell surface interaction and factors related to spore germination (Foucault and Drancourt, 2000; Leitch et al., 2005). Unfortunately, neither study provided ultrastructural images of the parasite or documented its life cycle within Caco2 cells or interaction with the host organelles.

There has been a general lack of understanding of human–microsporidial molecular interactions. The majority of observations on microsporidial–host molecular relationships have been demonstrated in the nematode, Caenorhabditis elegans (Botts et al., 2016; Kuo et al., 2018; Reddy et al., 2019), insects and zebrafish (Troemel, 2011). Watson et al. (2015), reported on the evolution of microsporidial gene expression of Trachipleistophora hominis grown in rabbit kidney cells (RK-13). They reported that T. hominis has ∼30% more genes than the ‘small genome’ microsporidia, such as Encephalitozoon. They also observed an overall reduction of gene activity of infected cells, except for genes that increase ATP production and other metabolites utilized by the parasite (Watson et al., 2015).

Despite the overt human pathologies associated with microsporidia in general and, more specifically, E. intestinalis infection, it remains unknown how these pathogens impact enterocytes at the molecular and ultrastructural level. Studies of the complete developmental cycle of E. intestinalis, its utilization of host organelles and subsequent parasite-induced pathologies in a human enterocyte model should enhance our understanding of how microsporidia, such as E. intestinalis, alter enterocyte function and hijack host organelles for propagation and dissemination. Herein, we report the development of a system for the continuous culture of E. intestinalis in Caco2 cells. In this system, we demonstrate that E. intestinalis infection of Caco2 cells recapitulated the in vivo life cycle of E. intestinalis. Additionally, we identified two types of spore development – typical environmental spores that leave the host, and autoinfective spores that immediately germinate, spreading infection to neighboring cells. Furthermore, we found that infection altered the host cell mitochondria, nucleus and brush-boarder structures. Genome-wide untargeted transcriptomic analysis demonstrated that E. intestinalis infection had an impact on cell signaling networks related to energy, metabolism and membrane trafficking.

To gain mechanistic insights into E. intestinalis development and induced pathology in its natural human host cell targets, Caco2 cells were grown to confluence in culture flasks or on cover glasses, prior to infection with E. intestinalis. Visualization through fluorescence staining for the actin cytoskeleton, as an indicator for the monolayer, was performed by incubation of fixed uninfected Caco2 and E. intestinalis-infected cells with -halloidin–Alexa Fluor 488. Spores were identified by staining with Calcofluor White 2MR, which stains the chitin that is present in the spore wall. Uninfected Caco2 cells displayed a typical confluent monolayer arrangement and chitin was not present (Fig. 1A).

Transmission electron microscopy (TEM) examination of uninfected cells revealed enterocytes in various stages of differentiation. Immature cells were small, rounded and contained a large oval nucleus surrounded by a thin layer of cytoplasm (Fig. 1B). Cells undergoing differentiation displayed a morphological initiation of polarization. They were elongated, with the nucleus positioned towards the basal part of the cells, while showing some microvilli on the apical surface (Fig. 1C). Fully differentiated cells were elongated with organized microvilli that formed a well-defined apical brush border (Fig. 1D).

At 2 weeks after E. intestinalis inoculation, infected Caco2 monolayers were vacuolated and numerous cells were found to contain clusters of spores in vacuoles identified by Calcofluor White 2MR staining (red, Fig. 1E). Close examination detected intracellular scattered spores (white arrowhead, Fig. 1G) as well as spores in the extracellular space close to the plasmalemma (open arrowheads, Fig. 1F,G).

The in vivo infection cycle of E. intestinalis in human enterocytes (Fig. 1H) starts with spores infecting cells and forming a parasitophorous vacuole (PV) inside the host cell cytoplasm, where the parasite proliferates, then undergoes sporogonic development, ending in production of new infective spores (Cali and Takvorian, 2014).

TEM examination of infected cells revealed one or multiple parasites in the PV within the host cell cytoplasm. We noted that infected cells demonstrated a curved or crescent-shaped host nuclei with a sporoplasm or early PV tightly nested against the invaginated portion of the nucleus (Fig. 1I,J). The E. intestinalis PV contained parasites at different developmental stages with proliferative, sporogonic and mature spores present (Fig. 1K). Some spores, while inside the PV, extruded their polar tubes, which often exited the PV and on occasion, the infected host cell. This indicates that two spore populations develop within the cell, one reflecting an autoinfective feature and the other for environmental dissemination outside the host (Fig. 1L). Encephalitozoon intestinalis infection in vitro in these enterocyte-like human Caco2 cells contained mature PVs with all the developmental and sporogonic stages of E. intestinalis, recapturing the in vivo life cycle as illustrated in Fig. 1H.

To understand the impact of E. intestinalis infection on host cells at the molecular level, we performed bulk RNA sequencing (RNAseq) analysis on uninfected (n=4) and infected Caco2 cell cultures (n=3), harvested in two independent experiments. Principal component analysis (PCA) revealed distinct clustering of the E. intestinalis-infected samples (red) from uninfected cells (blue) (Fig. 2A) and a volcano plot demonstrated the distribution of 17,836 genes by fold changes and P-values (Fig. 2B). Genes showing the most dramatic increase were involved in proliferation, such as CD44 (Lv et al., 2016), NPTX1 (Peng et al., 2018) and MYEOV (Moss et al., 2006). In contrast, genes showing the most significant reduction were involved in anti-proliferation, including KNG1 (Xu et al., 2018), CUBN (Gremel et al., 2017) and CYP2C9, suggesting that E. intestinalis infection stimulated host cell proliferation. Differential gene expression and heatmap analysis demonstrated that 269 genes increased by at least 1.5-fold and that 461 genes decreased by at least 1.5-fold, compared to levels in uninfected cells (P<0.05) (Fig. 2C).

Untargeted gene set enrichment analysis (GSEA) revealed that E. intestinalis infection significantly activated host gene networks denoted as (i) synaptic and presynaptic membrane formation; (ii) clathrin adaptor complex and vesicle coat; (iii) the phagocytic cup and extracellular organelles; (iv) mitochondrial respiratory chain; (v) SNARE complex; (vi) cell cortex, midbody and centriolar structural components; and (vii) AMPA glutamate receptor complex (Fig. 2D). This suggests that E. intestinalis strongly promoted the formation of host–pathogen interfacing membranes and host mitochondrial respiration. Encephalitozoon intestinalis infection suppressed gene networks that were responsible for (i) the nuclear membrane components; (ii) the formation of mature cellular membrane structures, including brush border, lamellar body and apical membrane; (iii) endo-lysosomal functions, including various proton pumps, membrane and lumen components, and (iv) the mechanistic target of rapamycin (mTOR) complex (Fig. 2E). These genome-wide transcriptomic analyses suggested that E. intestinalis rewired host cell mitochondrial energy production and membrane trafficking in favor of parasitic development. Interestingly, the gene angiotensin converting enzyme 2 (ACE2) is downregulated in Caco2 cells infected with E. intestinalis. The protein encoded by this gene has more recently received attention because of its role as the receptor binding partner for the S-protein on the SARS-CoV2 virus responsible for the 2020 pandemic (Bourgonje et al., 2020).

Our RNA-seq data suggested that E. intestinalis infection has a significant effect on host mitochondrial function. Consequently, we sought to determine by TEM whether there was an observable morphological feature that could be correlated with this observation during infection. We found that the mitochondria of cells containing parasites at the early developmental stage appeared to be recruited by the parasites and accumulated around the sporoplasm (Fig. 3A) and mitochondria tightly abutted the newly formed PV membrane adjacent to proliferative parasites, which were attached to the inner portion of the PV membrane (Fig. 3B). During the stages of parasite proliferative, some mitochondria had alternating electron dense and lucent structure that appeared to connect to the proliferative parasite membrane (Fig. 3C), while others developed thin tubular structures interfacing with the PV membrane (Fig. 3D). Prolonged infection and interaction with parasites induced a number of pronounced morphological changes in mitochondria, such as aberrant geometrically shaped cristae membranes, loss of cristae and severe swelling (Fig. 3B–D). Approximately 70% of mitochondria in infected cells exhibited these abnormalities (Fig. 3E), suggesting that host mitochondrial machinery was being modified by infection.

To determine what molecular modifications were exerted by E. intestinalis on host mitochondria, we interrogated mitochondria-related gene sets that underwent significant changes in infected cells. First, there was a systemic elevation of genes contributing to the intrinsic components of mitochondrial membrane (Fig. 3F,I). The elevated IMMT (An et al., 2012), CHCHD3 (Darshi et al., 2011), CHCHD6 (An et al., 2012) and CHCHD10 (Genin et al., 2016) genes all encode scaffolding proteins with known involvement in the maintenance of mitochondrial cristae morphology (Fig. 3I). There was increased PISD, which encodes a protein that normally catalyzes the conversion of phosphatidylserine to phosphatidylethanolamine on the inner mitochondrial membrane, while AGK, which was also increased, works with the TIMM22 complex to assemble proteins in the mitochondrial membrane.

In addition to mitochondrial membrane components, the genes encoding TIMM17A, TIMM17B (Bomer et al., 1996; Moro et al., 1999), TIMM23 (Bauer et al., 1999; Yamamoto et al., 2002) and SLC25A4 (Willis et al., 2018), which are involved in the maintenance and assembly of protein complexes necessary for ATP production, were also increased in infected cells (Fig. 3I). As microsporidia are known to use host ATP, consistent with an overt attempt to elevate mitochondrial ATP production, genes contributing to the mitochondrial respiratory chain complex I (also known as type I NADH dehydrogenase), were increased upon infection with the parasites (Fig. 3G,H,J). The complex is responsible for the transfer of electrons from NADH to coenzyme Q10 and the transport of protons across inner mitochondrial membrane. The elevated complex I gene network is exemplified by the increase seen in the NDUFC genes, which are responsible for the formation of the core catalytic subunit of complex I (Fig. 3H,J), NDUFV1, which encodes another complex I-forming oxidoreductase on the matrix side of the mitochondria, and mitochondrial nuclear-destabilizing factors (MT-NDFs), which are believed to be critical for ATP production (Sazanov, 2015). The chaperone-encoding gene PARK7, which helps stabilize complex I (Junn et al., 2009), was also increased in infected cells.

Elevations of the above mitochondrial membrane and respiratory chain components reflect a de novo biogenesis of mitochondria in infected cells, possibly an attempt to replace parasite-damaged mitochondrial organelles or the parasite increasing host energy supply. Two genes encoding master regulators of mitophagy, AFG3L2 and PINK1, were both activated in infected cells (Fig. 3I). During mitophagy, damaged mitochondria are selectively degraded by autophagosomes. The cleavage of PINK1 by AFG3L2 initiates a signaling cascade driving mitophagy. In line with this, FIS1, another critical regulator of membrane fusion with lysosome during mitophagy (Xian et al., 2019), was also increased in infected cells (Fig. 3I). Staining for mitochondrial reactive oxygen species (ROS) by using CM-H₂Xros (a Mitotracker) suggested a two-fold elevation of mitochondrial counts in infected cells (Fig. 3K). Thus, there was a strong tendency for E. intestinalis to recruit, hijack and stimulate the reproduction of host mitochondria.

The parasites appeared to also affect mitochondrial function via increasing membrane permeability and transport across mitochondria. This was exemplified by elevated levels of genes encoding SPG7 (Fig. 3I), a metalloprotease regulating mitochondrial permeability (Hurst et al., 2019), ABCB6, which binds to heme and porphyrins to facilitate ATP uptake into mitochondria (Krishnamurthy et al., 2006), and MPC1, which regulates pyruvate transport into the mitochondria (Koh et al., 2018) (Fig. 3I). In addition, BAK1 that promotes mitochondrial outer membrane permeabilization was also increased.

Furthermore, TEM-revealed recruitment of mitochondria by E. intestinalis was also reflected at the molecular level. Encephalitozoon intestinalis appeared to activate genes encoding RHOT1, which aids the anterograde transport of mitochondria (Grossmann et al., 2020), and MGARP, which facilitates kinesin-dependent mitochondrial trafficking along microtubules in infected cells (Li et al., 2009) (Fig. 3I). These data support that E. intestinalis robustly alters host mitochondrial biogenesis and ATP production.

One distinct TEM feature of E. intestinalis-infected cells was the presence of crescent- or horseshoe-shaped host nuclei abutted against an established PV containing developing organisms. Early infections, as indicated by the presence of a sporoplasm or early proliferative stages, abutted or adjacent to the host nucleus, had limited indentation (Fig. 4A). The parasite membrane did not appear to touch the host nuclear envelop (Fig. 4A). As the infecting stage advanced to the formation of a PV membrane, the host perinuclear membrane began to form an obvious indentation, and the PV membrane appeared to closely interface to the host nucleus (Fig. 4B). As the organisms (meronts) proliferate, they remain attached to the PV membrane, portions of which were then abutted to the host nucleus with nuclear pores often visible in this area (Fig. 4C). An indicator of microsporidial karyokinesis, is the presence of an electron-dense area of the parasite nuclear membrane, identified as a nuclear plaque (NP), which acts as a microtubule organizer, similar to a centriole (Fig. 4D). The presence of both meronts attached to the PV membrane and sporogonic stages free in the PV, indicates asynchronous development. At the point when organisms formed PV membrane, we found some PVs were almost completely engulfed by the host cell nuclei (Fig. 4D). A measurement of nuclear indentation at early and mature stages indicated a progression of this process (Fig. 4E).

These morphological events were also reflected at the molecular level. GSEA analysis illustrated a significant alteration of the nuclear outer membrane gene set in infected cells [P<0.05; normalized enrichment score (NES) of −1.37; Fig. 4F,H). Representative genes that were highly reduced by E. intestinalis included CDCC155 (Morimoto et al., 2012), and SYNE1, SYNE2, SYNE3 and SYNE4 (Horn et al., 2013; Morgan et al., 2011; Stewart-Hutchinson et al., 2008; Zhang et al., 2005), which interact with spectrins to act as mediators between the nuclear lamina and cytoskeleton (Fig. 4G,I). SNCA, a molecular chaperone specifically involved in the folding of synaptic fusion components of SNARE complex, and CLMN, a paralog of dystonin involved in spectrin and actin crosslinking and scaffolding were also reduced (Burré et al., 2010; Young et al., 2003). In addition, the E. intestinalis-induced nuclear indentation was consistent with altered SYNE genes, which belong to the linker of nucleoskeleton and cytoskeleton (LINC) complex, which controls the cytoskeletal rearrangement of perinuclear and nuclear space (Zhang et al., 2009). Three other genes in this functional set included LTC4S, for biosynthesis of leukotrienes (Welsch et al., 1994), RETSAT, for conversion of all-trans-retinol to all-trans-13,14,dehydroretinol (Moise et al., 2004), and PSEN1, which is responsible for cleavage of Notch receptors and APP (Fraering et al., 2004).

TEM observations suggested that the normal apical brush borders developed in mature Caco2 cells (Fig. 5A,B), and that these were disrupted in E. intestinalis-infected cells (Fig. 5C,D). Consistent with the TEM results, GSEA analysis revealed a global suppression of the cellular transcriptome for brush border morphogenesis (Fig. 5E,F). Reduction in the levels of CDHR2 and CDHR5, which are responsible for the formation of inter-microvillus links (Crawley et al., 2014; Goldberg et al., 2000), was particularly consistent with the collapsed brush border structures in E. intestinalis-infected cells (Fig. 5C,D). A majority of the other genes with reduced levels encode proteins involved in transport of solutes (Hayashi and Yamashita, 2012; Schlingmann et al., 2002), hormones (Mahon et al., 2002) and nutrients (Drover et al., 2005; Lee et al., 1993; Qiu et al., 2006) across the apical membrane of mature enterocytes. Among them, CUBN encodes a brush-border localized receptor for vitamin B12. These data, under the context of a significantly reduced gene signature for apical plasma membrane (P<0.05; NES 1.53) suggested that E. intestinalis globally inhibited the apical membrane trafficking process in Caco2 cells (Fig. 5F,G).

In addition to apical transport, the reductions in the levels of gene encoding multiple adaptor proteins, including AP2A2 and AP2M1, which regulate clathrin coat vesicles and vacuolar ATPase-dependent acidification of endolysosome (Aguet et al., 2013; Kadlecova et al., 2017), suggested a concomitant suppression of endocytic process by the parasites. Indeed, GSEA showed a significant reduction of the gene set related to the structure or function of endolysosome (Fig. 5H). Interestingly, several endosome-dependent pathogen pattern recognition receptor genes, such TLR3, TLR7 and TLR9, were also suppressed by E. intestinalis. Of note, and consistent with a suppressed endocytic pathway, our extensive TEM analysis failed to detect distinct lysosomes or autophagosomes in infected cells. Using a Lysotracker (Green DND-26), a pH-dependent fluorescent probe, we found reduced lysosomal staining events in infected cells (Fig. 5I). These data collectively suggest that E. intestinalis not only hijacks and extracts mitochondrial energy through PV membrane fusion but appears to globally block host cell membrane trafficking and lysosomal function, presumably for its intracellular reproduction.

Because the top genes upregulated and downregulated by fold-change are involved in cell growth and suppression, respectively, we ran GSEA against the MSidDB hallmarks gene sets (Tables S1 and S2). Indeed, the Wnt/β-catenin along with Myc and late estrogen response gene sets are significantly enriched in E. intestinalis-infected Caco2 cells (Fig. 6A–C). For example, the abundance of these following specific transcripts were increased in infected cells (please see Table S3 for a full list of differentially expressed genes). GNAI1 is a GTPase responsible for the inhibition of adenyl cyclase and plays a role in centrosome regulation during mitosis (Cho and Kehrl, 2007). HEY1 is a transcription factor which was previously shown to be upregulated in colorectal cancer and associated with a poorer prognosis (Candy et al., 2013). Another transcription factor gene, histone deacetylase 11 (HDAC1) encodes a protein that is normally expressed in a tissue-specific manner and is found to be upregulated in several human cancer cell lines (Gao et al., 2002). Finally, JAG1, which encodes a well-characterized Notch1 and Notch2 receptor ligand and which was previously shown to be increased in APC^min intestinal tumors (Guilmeau et al., 2010). Another Notch ligand, Delta-like1 (DLL1) is also upregulated (Table S3) and is known to be highly expressed in goblet cells in the intestinal epithelial tissue (Akiyama et al., 2010).

Similarly, cytokine pathways responding to pathogens (IL2-STAT5 and IL6–JAK–STAT3 and ROS gene sets) were significantly enriched in E. intestinalis-infected Caco2 (Fig. 6D–F). Genes upregulated in E. intestinalis infection include colony-stimulating factor 2 (CSF2), better known as GM-CSF, which is a well-known cytokine secreted in response to a variety of bacteria and viruses (Hsu et al., 1995). Additional genes that were increased in infected cells include the following, as shown in Table S3. Lymphotoxin β, CD79b and CD83 are known to promote lymphocyte survival and activation (Cragg et al., 2002; Pinho et al., 2014). APLP1, is a membrane associated glycoprotein and has been previously shown to play a role in the dissemination of intestinal carcinoids (Arvidsson et al., 2008). GLIPR2 enhances type 1 interferon signaling in response to TLR4 (Zhou et al., 2016). Genes found in the ROS hallmarks GSEA are primarily involved in reductase activity in response to an upregulation of ROS including GLRX, GLRX2, TXNRD1 and GPX4. Additionally, the expression patterns of these genes have been shown to be dysregulated in studies focusing on colon cancer cell lines or tissue (Barrière et al., 2004; Cha and Kim, 2009; Peters et al., 2008).

Ingenuity pathway analysis (IPA) of our RNA-seq data was used to visualize members of the Wnt/β-catenin (Fig. 6G) and IL-6 (Fig. 6H) signaling pathways. A majority of the members in these pathways are significantly upregulated (red highlighted). Taken together, our GSEA and IPA analysis suggest that Caco2 cells respond to E. intestinalis infection by enhancing reparative and defensive signaling cascades.

E. intestinalis was first identified in human intestines of an HIV patient in 1993 (Cali et al., 1993). Early morphological study of E. intestinalis relied on the rabbit kidney RK13 cell line. Foucault and Drancourt (2000) reported that E. intestinalis uptake into Caco2 cells was by phagocytosis. Subsequently, Leitch et al. (2005) demonstrated that infection of these cells was by polar tube penetration with sporoplasm deposition, and suggested phagocytosis was not a significant factor for E. intestinalis infection. Both studies concentrated on initiation of infection. Consequently, neither of these reports described developmental stages, the life cycle, host cell–parasite interaction or pathology. Additionally, neither report provided ultrastructural images to support their position.

This current study utilized TEM to evaluate E. intestinalis infection in a Caco2 cell model and found that this method was able to capture most of its distinguishing in vivo life cycle features based on morphological analysis. Additionally, we demonstrate the presence of two types of spores – an environmental spore, which may exit the infected host in feces and eventually infect a new host, and auto-infective spores, which become activated and extrude their polar tube to spread infection inside the PV or upon release from a lysed cell. Auto-infective spores may re-infect the host cell or neighboring cells, propagating infections in the host organism (Nigrelli, 1946). Unlike E. bieneusi, which primarily infects enterocytes, E. intestinalis often disseminates from enterocytes to other intestinal cell types, the liver, kidney, urinary tract, respiratory tract and eyes. These autoinfective spores may be responsible for the ability of E. intestinalis to disseminate. Furthermore, this represents one of the first studies systematically dissecting the molecular targets of E. intestinalis in a physiologically relevant human cell model. Our data suggest that E. intestinalis blocks the major membrane transport processed carried out by mature enterocytes, while preferentially establishing perinuclear vacuoles that directly engage the host nucleus and hijack host mitochondria. Several specific cellular signaling cascades were activated by E. intestinalis, potentially for defensive and reparative mechanisms (Fig. 7).

Microsporidia have undergone a minimization of their genome and cellular components, with one feature being their lack of traditional mitochondria, and they have been considered ‘amitochondriate’. These organisms do, however, possess a mitochondrial remnant termed the ‘mitosome’, which retains the biosynthetic machinery for iron-sulfur clusters and some transport functions, but is no longer involved in ATP production (Williams et al., 2014). A major open question regarding microsporidia–host interaction has been how these organisms interface and exploit host machinery for propagation. Microsporidial dependence on host organelles has been exemplified by the recruitment of host mitochondria (Hacker et al., 2014; Scanlon et al., 2004) and ER to the PV (Ferguson and Lucocq, 2019). This utilization and modification of host organelles has been reported in other intracellular pathogens, for example. Coxsackie viral infections (Hsu et al., 2010) and Toxoplasma gondii (Weiss and Kim, 2020). Likewise, we previously showed that during the development of the microsporidian Anncaliia algerae in human HeLa cells, they repurpose host microtubules and Golgi for parasite development (Santiana et al., 2016).

Microsporidial infection puts a tremendous stress on host mitochondria. The observed elevation of mitochondrial outer membrane and respiratory chain complex genes in the current studies suggests a cellular attempt to supply new mitochondria. Interfaces between host mitochondria and E. intestinalis are expected to allow ATP to flow to the parasite for its reproduction. Hacker et al. (2014) conducted an extensive study of mitochondrial binding with E. cuniculi PVs in rabbit cells (RK-13). They demonstrated that ATP and other metabolites were transported to proliferative stages by mitochondrial–PV voltage-dependent anion-channels. E. hellem sporoplasm surface protein 1 (EhSSP1) has been found to interact with mitochondrial voltage-dependent anion channels (VDACs) at the PV interface (Han et al., 2019). The current study showed an enrichment of ATP transporting proteins at the host cell–PV interface. Watson et al. (2015) reported a similar upregulation of host genes for proteins involved with glycolysis and ATP production in T. hominis-infected RK-13 cells.

While we and others have reported PV interaction with host mitochondria for E. cuniculi (Hacker et al., 2014), we describe here a similar phenomenon with respect to parasitic engagement with the host nuclear outer membrane (Fig. 7). Our data suggest that E. intestinalis potentially exploits the SNARE machinery to gain direct access and interface with host mitochondrial and nuclear membranes. Mechanistically, our TEM analysis revealed that the fusion between the PV membrane and host nuclear outer membrane was potentially facilitated by an activated SNARE network in infected cells. The expression of numerous syntaxins that are normally found in different intracellular membrane compartments were promoted by E. intestinalis. Among the elevated syntaxins, syntaxin 2 is known to localize in the perinuclear region (Band and Kuismanen, 2005), syntaxin 11 is present in late endosomes and trans-Golgi network (Valdez et al., 1999), and syntaxin 8 recycles claudin-16 at tight junctions in epithelial cells (Ikari et al., 2014). Furthermore, BNIP, an ER protein responsible to promote apoptosis when defective vesicle fusion occurs, was also elevated in infected cells, suggesting that the perinuclear localization of the growing PV might be detected by the host cells as a defective vesicle fusion event. Alternatively, it may hijack the mitophagy membrane fusing machinery to establish a host mitochondrial–parasite membrane interface (Fig. 7). This speculation is supported by the finding of elevated TOMM20, which was shown to appear on surfaces of lysosomal-mitochondrial contact interfaces (Wong et al., 2018). These molecular insights provide a potential mechanistic basis for the frequently observed microsporidial protoplasmic extensions and multiple forms of tubular appendages that interface with host organelles (Cali et al., 1998; Takvorian and Cali, 1983).

Further experiments need to focus on how E. intestinalis avoids detection and degradation by host machinery while affecting the dynamics of cellular organelles through mitochondrial and lysosomal interactions. In addition, the parasitic proteins responsible for engaging the various cellular organelles need to be identified and characterized. While our study provides a detailed ultrastructural study and molecular characterization of physiologically relevant microsporidian infection, further advances in sequencing techniques, such as sequencing at the single-cell level, could both refine the pathways specific to microsporidian infected cells and map out the development of pathogenesis.

E. intestinalis was initially grown in rabbit kidney cells (RK-13; ATCC- CCL-37), that were cultured in minimum essential medium Eagle (MEM) containing 10% fetal bovine serum (FBS) penicillin-streptomycin (Thermo Fisher Scientific) at 37°C with 5% CO₂. The medium from RK-13 cultures infected with E. intestinalis was removed weekly during medium changes and spores were collected from this medium, purified by passing them through 5 μm size filter (Millex, Ireland) to remove host cells, concentrated by centrifugation (5000 g for 5 min), and stored in sterile distilled water at 4°C. Spores were also obtained from disrupted infected cells. The medium was removed and 1 ml of 0.25% (w/v) trypsin and 0.5% EDTA dissolved in phosphate-buffered saline (PBS) solution was added to the flask, covering the monolayer, for 5–10 min. Cells were released by streaming MEM from a pipette and discharging it onto the monolayer. The cells were collected and spores purified as described above.

Caco2 cells (ATCC-HTB-37) were grown in 25 cm² flasks in MEM containing 20% FBS, 1% penicillin-streptomycin (Thermo Fisher Scientific) at 37°C with 5% CO₂. Additionally, cells were also grown in two-well cover glass chamber wells. When confluent, several wells of cells were infected with 100 µl of ∼10⁶ spores/ml solution and maintained for 14 days. Uninfected and E. intestinalis-infected Caco2 cells were fixed with 4% paraformaldehyde and stained for F-actin with phalloidin–Alexa Fluor 488, and Calcofluor White 2MR (American Cyanamid Corp., Princeton, NJ) for detection of chitin with a Nikon T2000 fluorescence microscopy. Lysotracker (Green DND-26; Thermo Fisher Scientific, L7526) and Mitotracker (CM-H₂Xros; Thermo Fisher Scientific, M7513) were used according to manufacturer's recommendation to stain lysosomes and mitochondria, respectively. Fluorescent puncta events were counted per field for Mitotracker.

Four flasks of Caco2 cells were confluent after 7 days. Two flasks were maintained uninfected and two flasks were each infected with 200 µl of 10⁶ spores/ml solution and maintained for an additional 7 days. The cells of both uninfected and infected flasks were then split and added to four flasks. At 14 days post exposure, the infected cells and uninfected cells were both harvested and prepared for EM and RNA extraction, respectively.

For TEM, individual samples of infected and uninfected cells were placed in microfuge tubes. The tubes were then centrifuged at 5000 g for 30 s to obtain a pellet of cells as previously described (Cali et al., 2002; Takvorian et al., 2020). The resulting pellets were fixed for 20 min in room temperature (RT) 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer. After 20 min, the RT fixative was removed and fresh 4°C fixative was added and stored overnight. The pellet was rinsed in 0.1 M sodium cacodylate buffer and post-fixed in 1% buffered OsO₄ overnight. The post-fixed pellets were dehydrated through a graded series of ethanol, transitioned into propylene oxide, and embedded in Epon LX-112 resin. Thin sections were stained with uranyl acetate and lead citrate. The samples were observed with an FEI Tecnai 12 TEM (FEI, Hillsboro, OR) and images recorded with a OneView 16 Megapixel digital camera (Gatan, Pleasanton, CA) at the Rutgers, Newark Electron Microscopy Facility.

The Caco2 cells of both uninfected and infected flasks were split and added to four flasks at 7 days post exposure. At 14 days post exposure, the infected cells and uninfected cells were both harvested and prepared for RNA extraction. RNA isolation was conducted using RNeasy Mini Kit (Qiagen, Catalog No. 74104) according to the manufacturer's instructions. The quality of RNA was first checked using Agilent 2200 TapeStation System (Agilent Technologies, Santa Clara, CA). All the RNA samples had an RNA integrity number (RIN) of >9.0 and were used for subsequent processing. The maximum recommended concentration was used as an input and they were subjected to oligo(dT)-based mRNA isolation using an NEBNext poly(A) mRNA Magnetic Isolation Module (New England BioLabs, Ipswich, MA). Illumina compatible RNAseq libraries were then prepared using an NEBNext Ultra II RNA Library Prep Kit (New England BioLabs, Ipswich, MA). AMPureXP beads (Beckman Coulter, Brea, CA) were used to purify cDNA libraries. The quality and the size distribution of the libraries were determined by TapeStation and were quantified using Qubit Fluorometer (Invitrogen). Barcoded libraries were pooled at equimolar ratios and sequenced on Illumina NextSeq 500 platform (Illumina, San Diego, CA) with 1×75 configuration. Demultiplexing of raw data into Fastq files was undertaken using bcl2fastq (Illumina, San Diego, CA). Fastq files were analyzed using Partek^® software. Copyright, Partek Inc. Partek and all other Partek Inc. product or service names are registered trademarks or trademarks of Partek Inc., St Louis, MO, USA. For gene set enrichment analysis (GSEA) (Subramanian et al., 2005), the molecular signature database (MSigDB) GO Biological Process (http://www.gsea-msigdb.org/gsea/msigdb/collections.jsp) was used to generate enrichment plots and heatmaps to determine significance of differential expression between conditions. One thousand permutations were performed for each gene list tested; normalized enrichment scores (NES) and nominal P-values are reported for gene signatures. Heatmap Z-scores are centered and normalized around 1.5 standard deviations from the mean. Heatmaps generated by GSEA are on a relative min max scale. The nominal P-value of <0.05 was considered to indicate significant enrichment. The Wnt BetaCatenin and IL6 networks were generated through the use of IPA (QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis).

The authors thank Seema Husain (The Genomics Center) for helping with bulk RNA sequencing analysis.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.253757.reviewer-comments.pdf