Trophozoites of differently pathogenic E. histolytica clones release EVs similar in size and morphology

To assess particle release from the different parasite clones and size distribution of these particles, EVs were isolated from E. histolytica-conditioned medium using differential ultracentrifugation (Fig 1A) and subsequently subjected to nanoparticle tracking analysis (NTA) (Fig 1B–E). The majority of released particles ranged in size between 80 and 400 nm (Fig 1B and 1C). EVs released by clones EhA1 and EhB2 did not exhibit significant differences in terms of modal size (Fig 1D) or abundance (Fig 1E). The heterogeneity of vesicle sizes determined by NTA could also be visualized by transmission electron microscopy (TEM) (Fig 1F). Using immunogold labeling, we detected the presence of E. histolytica galactose/N-acetylgalactosamine (Gal/GalNAc) lectin and lipopeptidophosphoglycan (LPPG) on the surface of both EhA1 and EhB2 EVs, thus verifying the amebic origin of the visualized particles (Fig 1F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Physical characterization of E. histolytica EVs.

(A) Workflow used for the isolation of EVs from E. histolytica-conditioned medium. Differential centrifugation of supernatants was performed at increasing speeds for 15 min each to clear the samples of cellular debris. EVs were pelleted by ultracentrifugation at 100,000 g for 1 h and washed once with filtered PBS prior to further use. Created in BioRender. (B, C) Nanoparticle tracking measurements of EhA1 EVs (B) and EhB2 EVs (C) (data averaged from 5 videos per sample, overlay of multiple independent measurements, n = 12–13). (D) Comparison of the modal particle size of EhA1 EVs and EhB2 EVs as determined by nanoparticle tracking (n = 9–11, unpaired t test, ns = not significant). (E) Comparison of the particle concentration of EhA1 and EhB2 EV samples in a set of standardized experiments (n = 3-4, unpaired t test, ns = not significant). (F) Transmission electron microscopic visualization of EhEVs subjected to immunogold labeling with rabbit anti-galactose/N-acetylgalactosamine (Gal/GalNAc) lectin or mouse anti-lipopeptidophosphoglycan (LPPG) primary and gold-conjugated secondary antibodies. Arrowheads indicate sites of labeling on individual EVs. Shown are representative images of one sample for EhA1 and EhB2 each.

https://doi.org/10.1371/journal.pntd.0012997.g001

EhA1 and EhB2 EVs contain proteins associated with virulence and exhibit differences between the clones that are distinct from those in trophozoites

To characterize the protein composition of EhA1 and EhB2 EVs and compare them to the protein composition of whole trophozoites, we performed liquid chromatography-mass spectrometry. Together, a total of 889 proteins were detected in the EhA1 and EhB2 EV proteomes (Fig 2A). 852 different proteins were identified in a minimum of 2 out of 3 EhA1 EV samples and 703 proteins in 2 out of 3 EhB2 EV samples (Fig 2A and 2C). 666 proteins were common to both EV proteomes (Fig 2A), of which only 1, a hypothetical protein with unknown function, was significantly differently abundant (fold change ≥ |2|, FDR p < 0.05, s0 = 0.5) (Fig 2A and S1 Table). 78 proteins were completely unique to EhA1, while 7 proteins were exclusive to EhB2 EVs (Fig 2A and S1 Table), with 29 and 4 of these, respectively, listed in AmoebaDB [23] as hypothetical or predicted proteins (S1 Table). Molecular function gene ontology (GO) term analysis of the unique EhA1 EV proteins revealed an association of these proteins with transporter and GTPase signaling activity (S2 Table), consistent with the exclusive presence of multiple GTPases in this EV proteome (S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Comparison of the E. histolytica vesicle and trophozoite proteomes.

(A, B) Quantitative comparison of EhA1 and EhB2 EV (A) and trophozoite (B) proteomes. Shown are proteins significantly differentially abundant between two proteomes (FDR p < 0.05, s0 = 0.5, fold change ≥ |2|) (yellow) as well as proteins present in only one of both proteomes (blue/ green). Proteins were considered to be a part of the respective proteome if they were present in at least 2/3 samples. Hence, we differentiated between proteins completely unique to one proteome (absent from all samples of the other proteome) and proteins belonging to only one proteome but not unique (present in 1/3 samples of the other proteome). (C) Overview of the total number of proteins, hypothetical proteins, proteins with a transmembrane (TM) domain or signal peptide in each of the analyzed proteomes based on annotations in AmoebaDB release 68 [23]. (D) Schematic overview of the E. histolytica EV proteome with selected proteins of interest. Created in BioRender. (E) Venn diagram depicting the number of proteins differentially abundant in EhA1 compared with EhB2 EVs, as well as proteins differentially abundant in EhA1 compared with EhB2 trophozoites and amount of proteins common to both datasets. (F) Selected molecular function GO terms associated with proteins enriched or depleted in EV proteomes compared with trophozoite proteomes, based on statistical overrepresentation test performed with Panther knowledgebase [27,28].

https://doi.org/10.1371/journal.pntd.0012997.g002

With regard to trophozoite proteomes, 2711 proteins were detected in EhA1 and 2496 in EhB2 (Fig 2B and 2C and S4 Table). A total of 2376 of these were common to trophozoites of both clones, of which 312 significantly differed in their abundance between EhA1 and EhB2 (fold change ≥ |2|, Fig 2B and S4 Table). 214 proteins were detected uniquely in EhA1 and 53 proteins uniquely in EhB2 trophozoite proteomes (Fig 2B and S4 Table).

The relative amount of proteins with a transmembrane domain was higher in EV proteomes (31.10% in EhA1, 33.85% in EhB2) than in trophozoite proteomes (13.32% in EhA1, 13.3% in EhB2) (Fig 2C). Similarly, EV proteomes were enriched in proteins with a signal peptide (27.11% in EhA1, 29.16% in EhB2) compared to their trophozoite counterparts (9.15% in EhA1, 10.10% in EhB2) (Fig 2C). Proteins present in EV proteomes included those with a function in EV biogenesis, such as endosomal sorting complex required for transport (ESCRT) proteins [24] and tetraspanins (TSPAN1, TSPAN4, TSPAN12) [25] (Fig 2D and S5 Table). Additionally, proteins involved in signal transduction, heat shock, oxidative stress response and virulence were detected (Fig 2D and S1 Table). To determine how similar E. histolytica EV proteomes were to those of other organisms, we analyzed the presence of amebic orthologs for the top 100 EV proteins according to Vesiclepedia [26] in our data. We identified amebic orthologs for 63 of the top 100 EV proteins, 44 of which were present in EhA1 EVs and 42 in EhB2 EVs (S6 Table).

To assess whether differences between EhA1 and EhB2 trophozoites were conserved in their corresponding EVs, we compared significantly different proteins between EV and trophozoite proteomes. For this, we considered all proteins significant that were differentially abundant with fold change ≥ |2| and FDR p < 0.05 or completely unique to one clone, amounting to 86 proteins of interest for EV proteomes and 579 for trophozoite proteomes. We found that only 15 of these differentially abundant proteins were shared (Fig 2E). Notably, of the 71 proteins different between EhA1 and EhB2 EV proteomes but not trophozoite proteomes, 15 were unique to EVs (S1 Table). Regarding the remaining 56 proteins, this result may indicate a selectivity for protein packaging into EVs that differs between the two clones.

A statistical overrepresentation test (Panther knowledgebase [27,28]) revealed that proteins depleted in both EhA1 and EhB2 EVs compared to their cells of origin were associated with the nucleus and organelle lumen, and have functions in DNA binding and catalytic activity acting on a nucleic acid (Figs 2F and S2 and S7 Table). Proteins enriched in EVs of both clones were predominantly membrane-associated and involved in hydrolase or pyrophosphatase activity, GTP binding and GTPase activity (Figs 2F and S2 and S7 Table).

Taken together, E. histolytica EVs share key proteins with EVs of other organisms, including typical EV markers and key biogenesis proteins such as tetraspanins. While EhA1 and EhB2 EVs showed distinct protein compositions, these differences largely did not reflect the proteins differentially abundant between EhA1 and EhB2 trophozoites.

E. histolytica EVs contain previously unidentified miRNAs

To analyze the micro RNAs (miRNAs) present in EhA1 and EhB2 EVs and evaluate their potential in the mediation of gene expression when transferred between cells, we isolated total EV RNA and performed miRNA sequencing. Out of all sequence reads, between 5,848,946 and 20,900,954 unique sequences per sample were detected. Curiously, none of the 199 previously described E. histolytica miRNAs were present in the sequencing data, with the exception of low amounts of Ehi-miR-4 [29]. Moreover, we only detected 26 out of 140,943 small RNA (sRNA) sequences published by Zhang et al. [30], most of them in very low abundance (S8 Table). Hence, to identify the miRNAs sequenced in our EV samples, we performed de novo miRNA prediction using BrumiR algorithm [31]. 1016 different miRNA sequences were identified overall, of which 167 were present in at least 2 out of 3 samples of either EhA1 or EhB2 with a minimum of 5 counts (S9 Table). We defined these 167 miRNA sequences as our EV miRNA dataset and, following the nomenclature established by Mar-Aguilar et al. [29], named them Ehi-miR-200 to Ehi-miR-366. All identified sequences were unique and did not exhibit identity with previously described E. histolytica sRNAs or miRNAs [29,30,32]. All 167 miRNAs were present in EhA1 EVs, while only 165 were detected in EhB2 EVs (Fig 3B and S9 Table). Out of the remaining 2, the difference in miRNA expression between EhA1 and EhB2 EVs was statistically significant (fold change ≥ |2|, FDR p < 0.05) only for Ehi-miR-200 (Fig 3F and S9 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Analysis of novel miRNAs in E. histolytica EVs.

(A) Comparison of the number of miRNAs detected in EhA1 compared with EhB2 EVs, based on de novo miRNA prediction using BrumiR algorithm version 3.0 [31] (n = 3 for each clone). (B–E) Quantitative analysis of potential targets of novel E. histolytica miRNAs in the Homo sapiens and E. histolytica genome, predicted using miRanda algorithm version 3.3a [33,34]. Shown are the amount of potential targets per miRNA in the H. sapiens (B) and E. histolytica (C) genome, as well as the amount of miRNAs potentially binding to the same target (D and E). (F) Normalized expression values of the two miRNAs detected only in EhA1 and not EhB2 EVs (EDGE test, *** FDR p < 0.001, n = 3). (G) Molecular function GO terms associated with identified miR-200 targets in the human genome (analysis performed with shinyGO version 80 [37], shown are the top 20 GO terms). (H) Amount of annotated versus hypothetical or unspecified predicted targets of miR-200 in the E. histolytica genome (according to AmoebaDB version 68 [23]). (I) Molecular function GO terms associated with miR-200 targets in the E. histolytica genome with an annotated function (analysis performed with shinyGO version 80).

https://doi.org/10.1371/journal.pntd.0012997.g003

To elucidate the potential of these miRNAs to regulate gene expression, we performed target gene prediction using miRanda algorithm [33,34]. Since EVs are not only relevant in the communication between a parasite and its host, but also in inter-parasite communication, we performed target gene prediction for both the human genome (as the natural host) and the E. histolytica genome. Individual miRNAs were found to have anywhere between 350 and 10,350 potential targets in the human genome (Fig 3B) and between 5 and 1355 potential targets in the E. histolytica genome (Fig 3C). For most putative targets in the human genome, less than 20 different miRNAs exhibited the capacity to bind, while in some instances, more than 50 different miRNAs were detected to possibly bind to the same target (Fig 3D). In the E. histolytica genome, most identified targets were associated with less than 10 different miRNAs (Fig 3E). Considering that all putative targets of all 167 miRNAs covered practically the entire human and E. histolytica genomes, we subsequently focused the target analysis on individual miRNAs of interest (S10 Table). For Ehi-miR-200, which was significantly differentially expressed between EhA1 and EhB2 EVs, a total of 2480 targets in 1227 different genes were identified in the human genome (S10 Table). This corresponds to about 6% of all human protein-coding genes [35,36]. Molecular function GO terms associated with these targets include ‘calmodulin binding’, ‘guanyl-nucleotide exchange factor activity’ and others (Fig 3G) that indicate a potential of Ehi-miR-200 to mediate intracellular signaling cascades in host target cells.

In summary, we here identified novel E. histolytica miRNAs present in EVs that have the potential to regulate gene expression in host and amebic target cells. Two of these miRNAs differed in their expression between EhA1 and EhB2 EVs.

Primary monocytes internalize and are activated by E. histolytica EVs

To track cellular uptake of EVs by murine bone marrow-derived monocytes, we labeled EhA1 and EhB2 EVs with the dye BCECF,AM and evaluated fluorescence in stimulated cells via flow cytometry (Figs 4A and S3B). The majority of male- and female-derived monocytes were positive for BCECF following stimulation with labeled EhA1 or EhB2 EVs (Fig 4B and 4C). Both the percentage of positive cells and the median fluorescence intensity (MFI) were higher in EV-stimulated cells compared with mock controls despite some fluorescence detected also in these controls (Fig 4B–D). These results indicate a specific absorption of EVs by monocytes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Cellular uptake of E. histolytica EVs and activation of primary monocytes.

(A) Graphical depiction of the workflow used for the determination of EV uptake by primary monocytes. Bone-marrow derived monocytes from male (shown as dots in graphs) and female (shown as triangles) mice were stimulated for 30 min with 0.5 µg EVs labeled with BCECF or corresponding volumes of controls and cellular uptake was quantified by spectral flow cytometry. Created in BioRender. (B) Dot plots of representative samples depicting flow cytometric identification of BCECF-positive cells following stimulation with mock control, EhA1 EVs or EhB2 EVs labeled with BCECF. (C) Percent BCECF⁺ monocytes after stimulation based on gates in (B). (D) Median fluorescence intensity (MFI) of BCECF in single cells after stimulation. MFIs were normalized to monocytes stimulated with unlabeled EVs for each independent experiment. (Kruskal-Wallis test with Dunn‘s multiple comparisons test, *p < 0.05, **p < 0.01, n = 6 from 3 independent experiments). (E) Dot plots of representative samples depicting surface expression of the activation marker CD38 on Ly6C^hi monocytes upon stimulation. Monocytes were stimulated in vitro for 24 h with 1000 EVs/cell of EhA1 or EhB2 EVs or corresponding volume mock control and subsequently analyzed by flow cytometry. (F) Percent CD38⁺ Ly6C^hi monocytes following EV stimulation. To control for the effect of protein denaturation on stimulatory capacity, EV and control samples were heat inactivated (h.i.) at 95°C for 10 min prior to stimulation. (G) Percent CD38⁺ Ly6C^lo monocytes following EV stimulation. (One-way ANOVA with Šídák‘s multiple comparisons test, *p < 0.05, **** p < 0.0001, n = 6).

https://doi.org/10.1371/journal.pntd.0012997.g004

We next examined the effect of EV stimulation on the presence of the monocyte activation marker CD38 and C-C chemokine receptor type 2 (CCR2) on the cell surface using flow cytometry. CD38 is an enzyme catalyzing the synthesis of the second messenger adenosine diphosphate ribose (ADPR) and its cyclic form cADPR and thus mediates multiple functions during cell activation through calcium signaling [38,39]. CCR2 is a receptor for the chemokine CCL2, which is released during inflammation [40]. As such, CCR2 is crucial for monocyte egress from the bone marrow and recruitment to sites of infection [41].

Here, we combined data from male- and female-derived monocytes as no significant differences between the sexes could be detected. We applied the differentiation into classical pro-inflammatory monocytes expressing high amounts of Lymphocyte antigen 6 C (Ly6C^hi monocytes) and non-classical Ly6C^lo monocytes, which are considered reparative monocytes [42,43] (Figs 4F, 4G and S4). The majority of isolated monocytes across all experiments were Ly6C^hi monocytes (53.9% - 90.7% of CD11b⁺ cells) and only a minority Ly6C^lo monocytes (5.1% - 33.2% of CD11b⁺ cells) (S4C and S4D Fig). EhA1 EVs and EhB2 EVs both significantly increased the percentage of CD38-positive Ly6C^hi monocytes (median 18.7% (EhA1), 20.65% (EhB2)) compared with mock controls (median 3.8%) (Fig 4E and 4F) without affecting the viability of stimulated cells (S6B Fig). The same was observed for Ly6C^lo monocytes, albeit to a lesser extent (median 7.1% (EhA1) and 7.4% (EhB2) CD38⁺ cells versus 1.6% (mock control)) (Fig 4G). Heat inactivation of EVs, performed as a control for protein denaturation, led to a slight reduction in stimulatory capacity for EhB2 EVs, but not EhA1 EVs (Fig 4F and 4G). Our results on CD38 suggest that EVs released by both E. histolytica clones induce activation of monocytes. Although EV stimulation did not induce a change in the relative amount of CCR2⁺ cells, a slight increase in the MFI for CCR2 could be observed on Ly6C^hi, but not Ly6C^lo monocytes (S4E–H Fig). This finding hints at a stronger potential for migration of pro-inflammatory monocytes upon contact with E. histolytica EVs.

EVs of differently pathogenic E. histolytica clones induce a pro-inflammatory gene expression profile in stimulated monocytes

After having shown that E. histolytica EVs can activate monocytes from male and female mice, we investigated the impact of EV stimulation on their gene expression profile by RNA sequencing and again combined data from male- and female-derived cells for this analysis. A total of 39 genes were significantly differentially expressed (fold change ≥ |2|, Bonferroni p < 0.05) between EhA1 EV- and mock control-stimulated monocytes, all of which were upregulated in EV-stimulated cells (S11 Table and Fig 5A and 5B). In response to EhB2 EV stimulation, 53 genes were upregulated in monocytes in comparison with mock controls, while 4 genes were downregulated (S11 Table and Fig 5A and 5C). No statistically significant differences in gene expression were detected between EhA1 EV- and EhB2 EV-stimulated cells (S11 Table). Genes induced by EV stimulation included those encoding for cytokines Tnf, Cxcl2, Ccl3, Ccl4, Ccl5 and Cxcl10 (S11 Table and Fig 5A–C). In accordance with this, most of the top Molecular function GO terms associated with genes upregulated upon EV stimulation were related to chemokine receptor binding, chemokine and chemoattractant activity (Fig 5D and 5E). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis revealed that upregulated genes were involved in key immune signaling pathways, such as NF-κB signaling, TNF signaling and IL-17 signaling, among others (S5A and S5B Fig). Along with the induction of cytokine-encoding genes, several type I interferon-stimulated genes (ISGs) were upregulated upon EV stimulation, including Icam1, Gbp5, Ifit1, Rsad2 and Oasl1 (S11 Table and Figs 5A–C and S5). Furthermore, genes induced by EV stimulation included Clec4e encoding for the Mincle receptor, an important initiator of pro-inflammatory immune responses associated with phagocytosis [44], Sod2, involved in the processing of reactive oxygen species (ROS) and thus protection of cells from oxidative stress [45], and Cd40, which encodes a co-stimulatory molecule involved in the production of pro-inflammatory cytokines [46,47] (S11 Table and Fig 5A–C). The most significantly upregulated gene upon both EhA1 and EhB2 EV stimulation was Acod1 (S11 Table and Fig 5B and 5C), encoding for Aconitate decarboxylase 1, an enzyme involved in the mediation of immune responses to inflammatory stimuli through the production of the metabolite itaconate [48]. The interferon regulatory factor Irf4 was among the few genes downregulated upon EV stimulation (S11 Table and Figs 5A, 5C and S5D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Transcriptional profile of EV-stimulated monocytes.

Male and female bone marrow-derived monocytes were stimulated for 8 h in vitro with 1000 EVs/cell or equal volume mock control. mRNA expression levels were subsequently analyzed using RNA sequencing. (A) Heatmap depicting expression levels of genes significantly differentially expressed between EhA1 EV and/or EhB2 EV-stimulated monocytes and mock controls (fold change ≥ |2|, Bonferroni-corrected p < 0.05, n = 4). (B, C) Volcano plots depicting relative gene expression between EhA1 EVs (B) or EhB2 EVs (C) and mock controls. Genes significantly upregulated in EV-stimulated monocytes are depicted in red, whereas genes significantly downregulated are depicted in blue (log fold change ≥ |1|, Bonferroni-corrected p < 0.05, n = 4). Selected genes of interest are labeled. (D, E) Molecular function GO term analysis of significantly upregulated genes in EhA1 EV (D) or EhB2 EV (E)-stimulated monocytes compared with mock controls (analysis performed with shinyGO version 80 [37], shown are the top 20 GO terms). (F, G, H, I) qPCR analysis of selected genes of interest (based on A–C) in EV-stimulated and control male- (dots) and female-derived (triangles) monocytes. Rps9 was used as calibrator. ΔCq values were normalized to the median Cq value of pre-stimulation controls. (One-way ANOVA with Dunnett‘s or Šídák‘s multiple comparisons test, *p < 0.05, n = 5-6).

https://doi.org/10.1371/journal.pntd.0012997.g005

We validated the sequencing results using quantitative PCR (qPCR) for selected genes, showing increased expression of Tnf, Cxcl2, Ccl5 and Oasl1 upon both EhA1 and EhB2 EV stimulation compared with mock controls (Fig 5F–I). Interestingly, expression of Cxcl2 was strongly increased in all monocytes cultured during stimulation, including mock controls, compared with pre-stimulation samples (Fig 5G, note the logarithmic scale).

These results show that stimulation of monocytes with E. histolytica-derived EVs led to the induction of a predominantly pro-inflammatory immune response on mRNA level, primarily characterized by upregulation of chemokine-encoding genes. Despite differences in pathogenicity of the amebic clones, EVs released by EhA1 and EhB2 did not trigger significantly different immune responses in this context.

Release of pro-inflammatory cytokines is enhanced by EV stimulation and stronger in male- than female-derived monocytes

Several studies have previously underlined the important role of cytokine signaling in ALA immunopathology [13,14,18]. Moreover, monocytes from male mice were shown to release higher amounts of TNF and CXCL1 during hepatic amebiasis compared with their female counterparts, contributing to the differing disease burdens observed between the sexes [18]. To assess cytokine release in response to E. histolytica EVs, enzyme-linked immunosorbent assay (ELISA) and a bead-based multiplex assay (LEGENDplex) were performed on supernatants of stimulated monocytes. Here, we investigated monocytes from male and female mice separately. We found that EhA1 and EhB2 EV stimulation resulted in marked increases in the release of IL-12p40, CXCL1, IL-6, TNF, IL-1β and CCL3 by monocytes in comparison with mock controls (Figs 6A and S6A–K). This was observed in monocytes derived from both sexes. However, for IL-12p40, CXCL1 and IL-1β, the effect was stronger in monocytes isolated from males than from females (Figs 6A and S6). Interestingly, heat inactivation of EhB2 EVs led to a partial ablation of the observed effect, which was not observed upon heat inactivation of EhA1 EVs (Figs 6A and S6A–K). Despite the fact that CCL2 is a main mediator of ALA immunopathology in mice [13] and is known to be expressed in the vast majority of Ly6C^hi monocytes during ALA [18], we could not detect any CCL2 release by EV-stimulated monocytes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Release of cytokines and myeloperoxidase upon EV stimulation.

Bone marrow-derived monocytes from male and female mice were stimulated for 24 h in vitro with 1000 EVs/cell or equal volume mock control. Supernatants of stimulated cells were analyzed by ELISA or flow cytometry-based multiplex cytokine assay (LEGENDplex). To control for the effect of protein denaturation on stimulatory capacity, EV and control samples were heat inactivated (h.i.) at 95°C for 10 min prior to stimulation. (A) Heatmap depicting median fluorescence intensity values for cytokines in supernatants of monocytes stimulated with EVs or controls as determined by LEGENDplex (IL-12p40, CXCL1, IL-6, TNF, IL-1β) or cytokine concentration as determined by ELISA (CCL3) (n = 3-6, depicted is the median of all samples of one condition). (B, C) Myeloperoxidase (MPO) concentration in supernatants of stimulated monocytes (B), bone marrow-derived neutrophils (C) and peripheral neutrophils (D) (isolated from spleen and blood, stimulated in the same manner as monocytes) as determined by ELISA. (One-way ANOVA with Šídák‘s multiple comparisons test, ** p < 0.01, **** p < 0.0001, n = 3-7).

https://doi.org/10.1371/journal.pntd.0012997.g006

EhA1 EVs but not EhB2 EVs trigger increased myeloperoxidase release by monocytes and neutrophils

In addition to investigating cytokine release, we also analyzed release of the granular enzyme MPO in response to E. histolytica EVs. MPO is released from azurophilic granules during degranulation and extracellular trap release and catalyzes the formation of microbicidal mediators, thus contributing to the innate immune response to pathogen invasion [49,50]. MPO released by host cells has been previously shown to kill E. histolytica trophozoites by using parasite-derived hydrogen peroxide for the formation of cytotoxic hypochlorous acid [51].

Interestingly, MPO concentrations were higher in supernatants of EhA1 EV-stimulated monocytes of both sexes than in mock controls, but not in supernatants of EhB2 EV-stimulated cells (Fig 6B). Since neutrophils are known to be a major source of MPO in circulation [50] and also contribute to ALA, we decided to follow up on this intriguing discovery by stimulating neutrophils with E. histolytica EVs. When using bone marrow-derived neutrophils, no significant differences in MPO release between EV- and mock-stimulated cells could be detected (Fig 6C). However, when we isolated neutrophils from spleen and blood (peripheral neutrophils) and subjected them to EV stimulation, EhA1 EVs triggered significantly increased MPO levels, while EhB2 EVs did not (Fig 6D).

Ethics statement

All mice used were bred in the animal facility of the Bernhard Nocht Institute for Tropical Medicine and kept in individually ventilated cages under pathogen-free conditions with a day/night cycle of 12 h, 21°C RT and 50 – 60% humidity. For harvesting of organs, C57BL/6J mice aged 9 – 13 weeks were euthanized by CO₂ overdose followed by cervical dislocation or cardiac puncture and blood withdrawal in accordance with German animal protection laws. Organ harvest was approved by the Animal Welfare Officer of the BNITM with acknowledgement of the Authority for Justice and Consumer Protection, Veterinary Affairs, Hamburg, Germany under the permission file numbers T-008 and T-011. Harvested organs were used for experiments immediately after euthanasia. Mice of both sexes were used for all experiments.

Extracellular vesicles

E. histolytica trophozoites of clones A1 and B2 [22] were cultured axenically under microaerophilic conditions in TY-I-S-33 medium [92] supplemented with 1% 100x Penicillin/Streptomycin antibiotic mixture (Capricorn Scientific GmbH) at 37°C. For EV isolation, trophozoites were washed with sodium phosphate-buffered saline (NaPBS; 6.7 mM NaHPO₄, 3.3 mM NaH₂PO₄, 140 mM NaCl, pH 7.2) and resuspended in EV-depleted TY-I-S-33 medium, which was obtained by ultracentrifugation at 100,000 g for 18 h to ensure absence of serum-derived EVs. 2x10⁵ trophozoites/well were then cultured on collagen-coated 6-well plates (5 µg/cm² collagen from calf skin (Bornstein & Traub type I, Sigma-Aldrich C3511)) in 5 ml EV-depleted medium for 24 h (EV uptake experiments) or 46 h (EV isolation for other stimulation experiments) under anaerobic conditions (Anaerocult A, Merck KGaA) at 37°C. Medium without trophozoites was co-incubated as mock control. Subsequently, supernatants were harvested and subjected to differential centrifugation performed according to a protocol modified from Mantel et al. [93] (15 min centrifugation each at 600 g, 1600 g, 3600 g and 10,000 g, supernatants processed for the next step and pellets discarded). EVs were pelleted in polypropylene tubes (open-top thinwall, Beckman Coulter 326823) by ultracentrifugation using Optima XE-90 centrifuge with SW 32 Ti swinging bucket rotor (Beckman Coulter) at 100,000 g and 4°C for 1 h (maximal acceleration, deceleration 5) and washed once with PBS (136.9 mM NaCl, 2.7 mM KCl, 1.8 mM KH₂PO₄, 10 mM Na₂HPO₄, 0.22 µm filtered) with the same settings. EV pellets were finally resuspended in 200 μl PBS, aliquoted and stored at – 80°C. For mock control samples, all steps were performed the same way. For isolation of total RNA from EVs, 500 μl QIAzol reagent (QIAGEN) were added to 100 μl of EV sample before storage. For subsequent stimulation experiments, EVs of four separate EV isolations were combined into EV pools to minimize batch effects on stimulation. EV pools were again aliquoted before storage at −80 °C. EV pool aliquots were only thawed once for use in stimulation experiments and not re-frozen to minimize freeze-thaw cycles.

Nanoparticle tracking analysis

Nanoparticle tracking was performed using Malvern Panalytical NanoSight LM14C with NTA 3.0 software according to the manual. EV samples were measured in a 1:100–1:300 dilution in 0.22 µm filtered PBS. A total of 900 frames were recorded over five separate 30 s measurements (camera level 16, 25 °C temperature control, detection threshold 6 for image processing).

Immunogold labeling for transmission electron microscopy

EVs were spun down onto glow-discharged, carbon and formvar coated nickel grids (Plano GmbH) by centrifugation at full speed in a table-top centrifuge for 15 min, washed once with PBS and 4 times with 0.05% glycine in PBS for 3 min each. Blocking of unspecific binding was performed by incubation of the grids in blocking buffer (0.5% bovine serum albumin in PBS) for 10 min. Grids were subsequently incubated in primary antibody diluted in blocking buffer for 24 h at 4°C. Antibodies used were rabbit anti-Gal/GalNAc lectin (170 kDa subunit) [94,95] (1:200 dilution) and mouse anti-LPPG [96] (1:100 dilution). In order to control for antibody specificity, controls without primary antibody were performed (S1 Fig). After incubation, the grids were washed with blocking buffer 6 times for 3 min, followed by incubation in colloidal gold conjugated secondary antibody diluted 1:100 in blocking buffer for 24 h at 4°C (goat anti-mouse colloidal gold-conjugated secondary antibody 12 nm, Dianova GmbH, and goat anti-rabbit colloidal gold-conjugated secondary antibody 10 nm, Science Services GmbH). Grids were again washed with blocking buffer 4 times for 3 min, followed by 2 washes with H₂O for 3 min. Sample fixation was performed using 2% glutaraldehyde for 5 min, followed by 2 washes with H₂O for 3 min each. Finally, grids were incubated with 2% aqueous uranyl acetate for 15 s, washed once with H₂O and dried at room temperature (RT). Sample imaging was performed using a Tecnai Spirit electron microscope at 80 kV.

Mass spectrometry and proteome analysis

For trophozoite proteomes, E. histolytica were seeded on collagen-coated 6-well plates and incubated for EV isolation. Trophozoites were harvested by resuspension in cold NaPBS, centrifuged at 400 g for 4 min, washed once with NaPBS and centrifuged as before, followed by storage of the pellet at – 80°C. EVs were processed for mass spectrometry (MS) as described above. The supernatant of the first ultracentrifugation step was used as negative control to determine proteins resulting from burst EVs or detached from the EV surface (S3 Table). Protein concentration of the samples was determined using Qubit 4 fluorometer (Thermo Fisher Scientific) according to the manual. 30 μg of protein for each sample were processed according to Hughes et al. [97] and Rappsilber et al. [98]. Tandem MS was performed with a Thermo Fisher Scientific Orbitrap Fusion (Q-OT-qIT) mass spectrometer. Raw MS data were processed using MaxQuant software (version 2.0.3.0) [99]. False discovery rate (FDR) was set to 1% for proteins and peptides and a minimum peptide length of seven amino acids was specified. Andromeda search engine in MaxQuant was used for spectra search against E. histolytica HM-1:IMSS annotated proteins (AmoebaDB release 56, https://amoebadb.org/amoeba/app [23]). The MaxQuant label-free algorithm was used for quantification [100]. Data annotation and statistical analysis of the MaxQuant output were performed with Perseus by MaxQuant [101]. Statistical comparison between two datasets was performed with Student’s t test in Perseus with FDR p-value set to 0.05 and s0 = 0.5. Proteins present in only 1 out of 3 samples of a dataset were excluded from the proteome for downstream analysis. For comparison of EV proteomes with trophozoite proteomes, statistical overrepresentation test was performed with Panther knowledgebase 17.0 using Fisher’s exact test with FDR-adjusted p value < 0.05 [27,28]. To identify amebic orthologs for the top 100 mammalian EV proteins from Vesiclepedia [26] (http://www.microvesicles.org/, version 5.1, accessed on 13 Aug 2024), human protein sequences were retrieved from UniProtKB (https://www.uniprot.org/) and orthologs identified through protein BLAST (blastp) against E. histolytica. For downstream analysis of proteomes and identification of transmembrane domains and signal peptides, AmoebaDB release 68 [23] (accessed on 16 Oct 2024) was used.

miRNA sequencing and analysis

Total RNA was isolated from EV samples in QIAzol (QIAGEN) using miRNeasy Mini kit (QIAGEN) according to the manufacturer’s instructions. RNA concentration was determined with RNA 6000 Pico Kit for 2100 Bioanalyzer (Agilent Technologies). Libraries for miRNA sequencing were constructed using NEXTFLEX small RNA-Seq Kit (PerkinElmer). Samples were sequenced at 27 – 132 million reads per sample with 50 bp read length using NovaSeq 6000 sequencing system with NovaSeq SP flow cell (Illumina). Raw data analysis and alignment to Zhang et al. [30] and Mar-Aguilar et al. [29] datasets was performed using CLC genomics workbench software version 24 (QIAGEN, https://digitalinsights.qiagen.com) using the Quantify miRNA tool version 1.3 with sequence length set to 15 – 40 nt or 15 – 30 nt, respectively. For this, Zhang et al. [30] reference data were accessed from NCBI gene expression omnibus (accession number GSE43668) and, for Mar-Aguilar et al. [29] dataset, a custom reference sequence list was created based on sequences listed in their supplementary information. De novo miRNA prediction from sequencing data was performed using BrumiR algorithm version 3.0 [31] with precursor clustering at 98% identity, followed by quantification and annotation using CLC genomics software. Out of the 3 samples of one clone, the largest dataset each was used for de novo miRNA prediction. miRNAs were considered true hits if they were present in at least 2 out of 3 samples with a minimum of 5 read counts. Target gene prediction was performed using miRanda algorithm (version 3.3a) [33,34] with Homo sapiens genome assembly GRCh38.p14 and E. histolytica HM-1:IMSS genome (AmoebaDB release 68 [23]) as reference. Interactions were filtered based on a pairing score of >150 and an energy score of ≤15 as described previously [102,103].

Isolation of bone marrow-derived monocytes

Bones were sterilized by incubation in 70% isopropanol for 2 min, cut open and flushed with Dulbecco’s PBS (DPBS, PAN-Biotech) using hypodermic needles to obtain bone marrow cells, which were separated using a 70 µm cell strainer. Monocytes were isolated from bone marrow cells using EasySep Mouse Monocyte Isolation Kit (StemCell Technologies) according to the manufacturer’s instructions. Isolation efficacy was controlled by flow cytometry (S3A Fig). For this, isolated cells were stained with live/dead fixable blue dye (1:1000, Thermo Fisher Scientific) to determine viability and subsequently labeled with anti-CD11b (M1/70, Alexa Fluor 488-conjugated, 1:400, BD Biosciences 557672), anti-Ly6C (HK1.4, APC-conjugated, 1:200, BioLegend 128016) and anti-Ly6G (1A8, PE-conjugated, 1:400, BioLegend 127608) antibodies in Fc blocking solution. Samples were measured at Cytek Aurora spectral flow cytometer or BD Biosciences accuri C6 flow cytometer and analysis performed with FlowJo software version 10.

Isolation of peripheral and bone marrow-derived neutrophils

For isolation of peripheral neutrophils, blood obtained by cardiac puncture was collected in EDTA-coated tubes and subjected to two rounds of erythrocyte lysis. Spleen cells were passed through a 70 µm cell strainer, washed with DPBS and subjected to one round of erythrocyte lysis. Immune cells from blood and spleen were then combined and passed through a 30 µm cell strainer prior to neutrophil isolation. Bone marrow cells were obtained as described for monocyte isolation above. Neutrophil isolation was performed using Neutrophil isolation kit (Miltenyi Biotec) according to the manufacturer’s instructions. Isolation efficacy was controlled by flow cytometry using the same antibodies used for control of monocyte isolation (S3C Fig). Samples were measured at BD Biosciences accuri C6 flow cytometer and analysis performed with FlowJo software version 10.

EV uptake

EVs were labeled during the EV isolation process after the first ultracentrifugation step with 10 µM 2’,7’-Bis-(2-Carboxyethyl)-5-(and-6)-Carboxyfluorescein, Acetoxymethyl Ester (BCECF,AM) (Thermo Fisher Scientific) for 30 min, followed by PBS washing and ultracentrifugation. BCECF,AM is non-fluorescent until it is internalized in cells (or EVs) and cleaved by cytosolic esterases, yielding the fluorescent BCECF. Protein concentrations were determined using Qubit 4 fluorometer (Thermo Fisher Scientific) according to the manual. 0.5 µg of EVs or the corresponding mock control volume were added to 1x10⁵ bone marrow-derived monocytes in cRPMI (RPMI supplemented with 10% charcoal-stripped FCS, 1% penicillin/streptomycin, 1% L-glutamine) and incubated for 30 min. Stimulated monocytes were washed once with DPBS and fixed using eBioscience Foxp3/ Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) according to the manufacturer’s instructions. Samples were measured at Cytek Aurora spectral flow cytometer and analysis performed with FlowJo software version 10.

EV stimulation

As quality control measure, EV pools were first tested for stimulatory capacity on crude bone marrow cells. 1×10⁶ cells were stimulated with 1000 EVs/cell (concentration determined by NTA) or equal volume mock control in cRPMI for 24 h, followed by detection of IL-6 in culture supernatants using BD OptEIA Mouse IL-6 ELISA (BD Biosciences). Only EV pools eliciting increased concentrations of IL-6 upon stimulation compared with mock controls were further used for experiments. 1x10⁵ bone marrow-derived monocytes/peripheral neutrophils or 5x10⁵ bone marrow-derived neutrophils were stimulated with EV pools at a concentration of 1000 EVs/cell in cRPMI for 8 h (for NGS) or 24 h (for flow cytometry analysis and cytokine assays). For heat inactivated (h.i.) controls, EV samples were incubated in a heating block for 10 min at 95 °C prior to use.

Spectral flow cytometry for surface marker analysis

For the analysis of surface marker presence on 24 h stimulated monocytes using spectral flow cytometry, cells were washed with DPBS, stained with Zombie UV (1:1000, BioLegend) to determine viability and subsequently labeled with anti-CD11b (M1/70, BV510-conjugated, 1:400, BioLegend 101263), anti-Ly6C (HK1.4, PE-conjugated, 1:800, BioLegend 127608), anti-Ly6G (1A8, APC-conjugated, 1:400, BioLegend 127614), anti-CD38 (90, BV421-conjugated, 1:800, BioLegend 104445) and anti-CCR2 (SA203G11, PE-Cy7-conjugated, 1:100, BioLegend 150611) antibodies in Fc blocking solution. Labeled cells were then fixed using eBioscience Foxp3/ Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) according to the manufacturer’s instructions. Samples were measured at Cytek Aurora spectral flow cytometer and analysis performed with FlowJo software version 10.

RNA-sequencing and data analysis

Monocytes stimulated with EVs for 8 h were fixed in RLT buffer and RNA was isolated using RNeasy Mini kit (QIAGEN) according to the manufacturer’s instructions. RNA quality and concentration were assessed with RNA 6000 Pico kit for 2100 Bioanalyzer (Agilent Technologies). Only samples with a RNA integrity number (RIN) of >7.5 were processed for sequencing. Libraries were prepared using QIASeq Stranded mRNA Library kit (QIAGEN) and sequenced as 75 bp paired-end reads on Illumina NextSeq 550 system with NextSeq 500/550 Mid Output Kit v2.5 (150 cycles) (Illumina) at 5 – 6 million reads per sample. Raw data were mapped to Mus musculus reference genome GRCm39.112 and analyzed using CLC genomics Workbench software version 21 with RNA-Seq Analysis tool version 2.6 (QIAGEN). Genes that were not expressed in a minimum of 3 out of 4 samples with at least 20 read counts were excluded from downstream analyses. Volcano plots for the depiction of significantly differentially expressed genes were created with the European Galaxy Server (https://rna.usegalaxy.eu/) https://rna.usegalaxy.eu/ [104]. Heatmaps were created using Heatmapper (http://heatmapper.ca/) [105]. ShinyGO version 0.80 was used for GO term analysis (http://bioinformatics.sdstate.edu/go/) [37].

Quantitative PCR

RNA was isolated from 8 h stimulated monocytes and freshly isolated monocytes (pre-stimulation) as described above and transcribed into cDNA using Maxima First Strand cDNA Synthesis kit (Thermo Fisher Scientific) according to the manual. Primers used for qPCR were Ccl5: forward 5-GGACTCTGAGACAGCACATG-3, reverse 5-GCAGTGAGGATGATGGTGAG-3, Cxcl2: forward 5-AGTTTGCCTTGACCCTGAAG-3, reverse 5-GGTCAGTTAGCCTTGCCTTT-3, Oasl1: forward 5-TGACGGTCAGTTTGTAGCCAT-3, reverse 5-AAATTCTCCTGCCTCAGGAAC-3, Tnf: forward 5-TCTGTGAAAACGGAGCTGAG-3, reverse 5-GGAGCAGAGGTTCAGTGATG-3, Rps9: forward 5-GCTAGACGAGAAGGATCCCC-3, reverse 5-TTGCGGACCCTAATGTGACG-3 (custom DNA oligos, Eurofins Genomics). Annealing temperature was determined by gradient PCR using Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific) and LightCycler 96 (Roche) and set to 58°C. Primer pair efficiency was determined by qPCR on serial cDNA dilutions and calculated as described in Pfaffl et al. [106]. Gene expression in EV-stimulated monocytes was quantified according to the Pfaffl method [106], with Rps9 used as reference gene, and normalized to pre-stimulation control samples.

Measurement of cytokine and MPO concentration

CCL3 and MPO concentrations were detected in immune cell supernatants after 24 h stimulation using Mouse CCL3/MIP-1 alpha DuoSet ELISA and Mouse Myeloperoxidase DuoSet ELISA kits (both R&D systems) according to the manufacturer’s instructions. All other cytokines were assessed using Mouse M1 Macrophage LEGENDplex flow-based assay (CXCL1, IL-1β, IL-6, IL-12p40, IL-12p70, IL-18, IL-23, TNF; BioLegend) with BD LSRII flow cytometer (BD Biosciences). Only data for cytokines with concentrations above the lower limit of detection are shown.

Statistics

All data statistically analyzed by GraphPad prism version 9 (this excludes sequencing data) were tested for normal distribution using Shapiro-Wilk test. Statistical significance between two groups was then tested using Student’s t test in the case of NTA data. For datasets with multiple comparisons, statistical significance was tested using non-parametric Kruskal-Wallis test with Dunn’s multiple comparisons test for data that did not pass the normality test or with parametric one-way ANOVA tests for normally distributed data. Brown-Forsythe test was used to test for statistically different standard deviations (SDs). Datasets with no significant differences in SDs were analyzed by ordinary one-way ANOVA with Šídák‘s multiple comparisons test, whereas datasets with significantly different SDs were analyzed with Brown-Forsythe and Welch’s ANOVA with Dunnett’s multiple comparisons test. Parametric tests are indicated as ‘one-way ANOVA’ in figure legends. Significance levels correspond to * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.