Abstract
Toxoplasma gondii and Hammondia hammondi are closely-related coccidian intracellular parasites that differ in their ability to cause disease in animal and (likely) humans. The role of the host response in these phenotypic differences is not known and to address this we performed a transcriptomic analysis of a monocyte cell line (THP-1) infected with these two parasite species. The pathways altered by infection were shared between species ~95% the time, but the magnitude of the host response to H. hammondi was significantly higher compared to T. gondii. Accompanying this divergent host response was an equally divergent impact on the cell cycle of the host cell. In contrast to T. gondii, H. hammondi infection induces cell cycle arrest via pathways linked to DNA-damage responses and cellular senescence and robust secretion of multiple chemokines that are known to be a part of the senescence associated secretory phenotype (SASP). Remarkably T. gondii-conditioned media can suppress the SASP response during H. hammondi infection, and this suppression is accompanied by a corresponding increase in the replication rate of H. hammondi in recipient cells. Taken together our data suggest that T. gondii manipulation of the host cell cycle provides a novel mechanism to avoid stress and/or DNA-damage induced responses by the host cell, and that this ability has a direct impact on parasite replication rate both within the host cell as well as in bystander cells.
Introduction
The complex interaction between the intracellular pathogen Toxoplasma gondii and its host cell can lead to extensive changes in the host transcriptome [1–3]. These changes can have dramatic effects on host phenotypes including those responsible for the regulation of metabolism, apoptosis and innate immune signaling [4–6]. A successful infection in T. gondii results in rapid replication of infecting parasites as tachyzoites, dissemination to a variety of tissues, followed by the clearance of actively replicating stages and encystment of bradyzoite stages in muscle and neuronal tissues. Therefore a balance exists for T. gondii to expand its numbers during the acute phase such that a sufficient number of parasites can make it to the chronic, cyst stage and therefore be transmitted to the next host.
It has long been known that T. gondii infection dramatically alters the transcriptional landscape in the host cell [4], and many of these changes can be directly linked to the secretion of specific effectors directly into the host cell [7,8], and T. gondii is likely to harbor hundreds of such effectors [9–12]. Rhoptry-derived proteins such as ROP16 regulate important transcription factors such as STAT3 [13] and STAT6 [14], while dense granule proteins such as GRA15 [15], GRA16 [16] and GRA24 [17] modulate host NF-κB, p53 and p38 MAP kinase host pathways, respectively. GRA18 has recently been shown to modulate the Wnt signaling pathway in mouse cells, owing to its ability to stabilize the transcription factor β-catenin [18], and GRA25 induces CCL2 in human foreskin fibroblasts [19]. In addition, parasite effector TgIST represses the interferon (IFN)-γ response by recruiting the Mi-2/NuRD repressor complex and subsequently blocking STAT1-related IFNγ-stimulated transcription [20,21]. Many dense granule effectors including GRA16, GRA18 and GRA24 require the presence of a protein complex on the parasitophorous vacuole membrane (PVM) consisting of at least three proteins named MYR1, 2 and 3 [22,23]. This complex has also been shown recently to be critical for the induction of the chemokine CCL22 in human placental cells [24]. Importantly T. gondii has no impact on CCL22 production by human foreskin fibroblasts (HFFs) [24], further demonstrating that responses to T. gondii infection can vary dramatically across cell types. Taken together it is clear that multiple means of modulating the host cell environment have evolved in T. gondii and the sum total of these effects is an essential component of its compatibility with its many hosts. Hammondia hammondi is the closest extant relative of T. gondii, and was first discovered in the feces of a cat in Iowa, USA in 1975 [25]. Unlike T. gondii, H. hammondi has a restricted natural intermediate host range, having been known to infect rats, mice, goat and roe deer in the wild [26–28], and in the laboratory non-human primates can be infected but not birds [26–30]. H. hammondi shares >99% of its ~8000 genes in near perfect synteny with T. gondii and expresses functional orthologs of key T. gondii effectors such as ROP18 and ROP5 [31–33]. This is important from an epidemiological and evolutionary perspective, as H. hammondi is not known to cause clinical disease in any naturally infected intermediate (human and other animals) or definitive host [33]. In addition, in tissue culture H. hammondi replicates for a short period of time and then enters into a unique terminally differentiated bradyzoite state that is a) unable to be subcultured in vitro, b) incapable of infecting an intermediate mouse host and c) only capable of infecting a feline host [29]. Paradoxically, despite this natural (and possible pre-determined) bradyzoite developmental program and expression of canonical bradyzoite genes during tachyzoite-like replication, H. hammondi cannot be induced to form bradyzoites using stresses like high pH medium that robustly induce cyst formation in T. gondii [34]. While it is not yet known if H. hammondi is capable of infecting humans (and there are no tests as yet capable of distinguishing these closely-related species serologically), H. hammondi and T. gondii oocysts are known to co-circulate, suggesting that humans may encounter this parasite [34,35].
Since multiple T. gondii effectors modulate host transcriptional regulation, and to date many of these have been found to be functionally conserved between T. gondii and H. hammondi [31,36], we sought to test the hypothesis that differences in the response of the host cell to T. gondii and H. hammondi could be linked to phenotypic differences between these species, including replication rate and pathogenesis. To do this we performed the first thorough analysis of the cell-autonomous host response to H. hammondi and compared it to T. gondii. We found that the majority of infection-induced changes in the host cell were conserved between these two species, but also that the magnitude of these changes in H. hammondi-infected cells was much larger (often by many-fold) than in T. gondii-infected cells. We confirmed that this effect is manifested at the protein level for a subset of chemokines, and that a similar effect could be observed in vivo during mouse infections. In addition, we also identified a small, but important, subset of host cell pathways that were altered during T. gondii infection but were either unaltered or even inverted in H. hammondi-infected cells. The central theme in these differentially regulated pathways was a role in the host cell cycle, which T. gondii is well known to manipulate [37,38]. These data and follow up experiments described here suggest that a) T. gondii and H. hammondi have distinct effects on the host cell cycle, b) T. gondii infected cells produce significantly less chemokine in response to infection compared to H. hammondi infected cells, c) H. hammondi-infected cells closely resemble cells undergoing senescence and display something akin to a senescence-associated secretory phenotype (SASP), and d) all of these effects serve to increase the replication rate of T. gondii. Overall this work has linked cell cycle manipulation by T. gondii as a global mechanism of transcriptional suppression, and determined that this is yet another unique adaptation of T. gondii compared to its near relatives. Remarkably, T. gondii infection leads to paracrine signals that markedly increase the sensitivity of neighboring cells to parasite infection.
Results
H. hammondi sporozoites induce a more potent differential gene expression in THP-1 cells as compared to T. gondii during acute infection
To compare the impact of T. gondii and H. hammondi infection on the transcriptional host response during infection, we performed RNA-sequencing (seq) on THP-1 cells (a monocyte cell line) infected with representatives of three major lineages of T. gondii and two H. hammondi isolates. In all cases (unless specified), we used parasites excysted from oocysts that were grown for 24 hours on HFFs.
We first used principal component analysis (PCA) to look broadly at differences and similarities in the host transcriptome across strains and species [39]. According to this analysis the first PC (PC1, 68% variance) indicated a clear separation between parasite species, while the second (PC2, 20% variance) encompassed differences between strains of each species. These data indicate that THP-1 cells infected with all three T. gondii types (TgGT1, TgME49 and TgVEG) are more similar to each other than to THP-1 cells infected with either strain of H. hammondi (HhEth1 and HhAmer; Fig. 1A). Infection status (Mock vs. Infected) also was distributed across PC1, in that mock-infected cells clustered together within PC1 regardless of strain or species (Fig. 1A). Importantly the two H. hammondi strains were much further separated from all mock-infected samples along this PC, suggesting 1) a distinct response by the host cells to this species and/or 2) more potent induction of transcriptional changes by the parasite. PC2 showed a clear separation between TgGT1-infected cells and all other samples, including T. gondii and H. hammondi. While it was clear from PCA and other analyses (below) that TgGT1 infection induced extensive changes in transcript abundance compared to mock-infection (Figs. 1A, C and H), the TgGT1-infected cells were clearly the most distinct T. gondii strain along PC2. Therefore we have excluded TgGT1 infection in some of the downstream analysis. Hierarchical clustering of the distances between samples also further confirmed the differential expression profiles seen in THP-1 cells infected with H. hammondi as compared to T. gondii (Fig. 1B).
We compared host gene expression of parasite-infected cells to mock-infected cells for all types/strains using DESeq2 [40] with the thresholds of log2 fold-change ≥ 1 or ≤ −1 with padjuster (adj) < 0.01 and identified host transcripts that were of higher or lower abundance in response to infection (Table S1). The most dramatic outcome of this experiment was the striking difference in the host transcriptional response to H. hammondi compared to T. gondii, regardless of strain. Specifically, HhEth1 and HhAmer-infected THP-1 cells had significant changes in ~27% and ~32% of the 15452 queried transcripts, respectively (Fig. 1I and Table S1), while in T. gondii this ranged from ~4% for TgGT1 to ~7% for TgVEG (Fig. 1H and Table S1). MA-plots showing average expression vs. log fold-change of each gene, further illustrate the stark contrast in the extent of changes in transcript abundance that are induced by T. gondii and H. hammondi (Figs. 1C-E vs. F-G). When we compared sets of differentially expressed genes across strains of each species, we identified type and/or strain-specific gene expression profiles (Figs. 1H and I). It is also interesting to note that more strain type-specific changes in transcript abundance were observed among the T. gondii strains compared to HhEth1 and HhAmer infections (Figs. 1H vs. I).
Hyperinduction of chemokines by H. hammondi compared to T. gondii is recapitulated in vivo
While differences in host responses by T. gondii and H. hammondi were readily apparent in vitro, little is known about differences in the host response to these parasites in vivo. Therefore we infected mice intraperitoneally with parasites of each species and collected peritoneal lavage supernatants and peritoneal cells. Some of the highly differentially expressed genes by T. gondii and H. hammondi in vitro infection were the host cytokine genes (Table S1). Therefore we analyzed cytokine transcripts by reverse transcriptase (RT)-qPCR and measured cytokine levels in the supernatants by ELISA (Fig. 2). Since Il12 and Ifnγ are cytokines that were induced in response to T. gondii infection and Il12 plays an important role in activating Ifnγ during T. gondii infection [41–43], we quantified mouse Il12p40 and Ifnγ in peritoneal cell supernatants taken from mice infected with TgVEG or HhAmer at various time points to ensure that mice were infected (Figs. 2SF and G). Since H. hammondi replicates ~4X slower than T. gondii in vitro [34], we also monitored relative replication rates and parasite loads using species-specific RT-qPCR for the parasite GRA1 gene. Our data from the RT-qPCR showed that at any given time point, H. hammondi parasite burden was significantly lower than T. gondii (Sidak’s multiple comparisons test, **p<0.01, ***p<0.001 and ****p<0.0001 as compared to the respective time points), indicating that H. hammondi also replicates more slowly than T. gondii in vivo (Fig. 2SE). We also observed that H. hammondi sporozoites have a lower viability post-excystation as evidence by GRA1 levels at 30 min post-infection (Fig. 2SE; Sidak’s multiple comparisons test **p<0.01, ***p<0.001 and ****p<0.0001).
To compensate for this growth difference (which is an intrinsic property of these parasites species), we set up an independent experiment where we normalized cytokine transcript and protein level to GRA1 transcript level, a reasonable proxy for parasite burden [34]. After infecting mice with 20,000 freshly excysted TgVEG or HhAmer sporozoites, we found that, as expected, H. hammondi showed a significantly lower parasite burden over the course of the experiment compared to T. gondii (Fig. 2SH, Sidak’s multiple comparisons test **p<0.01 and ****p<0.0001). However, when we normalized cytokine transcript levels (specifically Cxcl10 and Ccl22) in peritoneal cells to parasite numbers we found that H. hammondi infection resulted in significantly higher levels of parasite-normalized transcript at 30 min, 20 and 48 h post-infection for Cxcl10 and 20 h post-infection for Ccl22 (Figs. 2A and B; Sidak’s multiple comparisons test *p<0.05, **p<0.01, *** p<0.001 and **** p<0.0001). When we performed the same analysis on parasite burden-normalized Cxcl10 and Ccl22 protein levels in the peritoneal lavage fluid, we also found that H. hammondi-infected mice had significantly higher parasite-normalized chemokine levels compared to T. gondii (Figs. 2C and D; Tukey’s multiple comparisons test; **p<0.01and **** p<0.0001). These data indicate that the host response to H. hammondi is much more robust in vivo compared to that in response to T. gondii when parasite burden is taken into account.
H. hammondi-infected cells had more robust changes in transcription compared to T. gondii, despite extensive overlap in the altered pathways
From PCA and DESeq2 analysis of THP-1 cells infected with parasites, it is clear that H. hammondi-infected cells display a more robust transcriptional change than T. gondii-infected cells (Fig. 1). However this could occur by differential induction of distinct sets of genes, and/or via differences in the overall magnitude of the induction of the same gene sets. To assess this we used Pre-ranked Gene Set Enrichment Analysis (GSEA; [44,45]) on each transcriptional profile using the curated “Hallmark” gene sets database. Overall we identified 47 gene sets that were significantly enriched in cells infected with T. gondii and/or H. hammondi (FDR-q value<0.05), and 41 of these were shared gene sets enriched (positively or negatively) in infected cells compared to mock-treated cells (Figs. 3A and B, Table S2). Interestingly, H. hammondi infections induced quantitatively higher enrichment of these shared transcriptional changes. For example, for IFNγ response all T. gondii and H. hammondi strains significantly altered this gene set during infection, but H. hammondi Eth1 and Amer had normalized enrichment scores (NES) of 11.1883 and 11.3855 respectively while TgGT1, TgME49 and TgVEG-infected cells had and NES of 8.0227, 8.5538 and 10.0764 respectively (Fig. 3A, Table S2). This is further illustrated by cluster analysis of a subset of the IFNγ response gene set which includes multiple IRF and chemokine genes (Fig. 3C, Table S3 and Fig. 3Sc (top panels)). Using Ingenuity Pathway Core Analysis (IPA), we also identified 280 canonical pathways significantly enriched in T. gondii and/or H. hammondi infections (Benjamin-Hochberg adjusted p <0.01, Table S4). These pathways include dendritic cell maturation, IL-1 signaling, IL-6 signaling and interferon signaling (Fig. 3SaA and B). Our IPA analysis reaffirm our GSEA results that while H. hammondi infection alters many of the same canonical pathways and gene sets in THP-1 cells as does infection with T. gondii, the magnitude of the effect is greater (e.g., dendritic cell maturation had z-scores of 6.38 and 6.20 for HhAmer and HhEth1 respectively; 5.57, 4.69 and 4.09 for TgGT1, TgME49 and TgVEG respectively; Fig 3SaA, Table S4). This could be due to 1) T. gondii suppression of these host responses and/or 2) direct induction of a more dramatic host response during infection by H. hammondi. Since IFNγ-signaling is critical for controlling T. gondii infection [41–43] we examined all members of the interferon signaling pathway and identified a subset of them (BAK1, 1FITM1, 1FITM2, JAK1, JAK2 and PSMB8) that were only differentially expressed in THP-1 cells infected with H. hammondi (Fig. 3SbA-C).
As for GSEA, a handful of T. gondii species-specific pathways identified were Type II diabetes mellitus signaling and CD27 signaling in lymphocytes and H. hammondi-specific immune-related pathways such as GM-CSF and TGF-β signaling pathways (Fig. 3Sa, Table S4).
T. gondii and H. hammondi-infected cells have transcriptional profiles indicating distinct cell cycle states
Only two gene sets, Myc targets v1 and v2, were significantly enriched in all T. gondii-infected cells regardless of strain type and either negatively enriched or not significantly enriched H. hammondi-infected cells (Figs. 3A and B, 3Sc (bottom panels) and 4A). Given the fact that T. gondii is known to induce c-Myc translocation into the nucleus and also induces S phase transition and ultimately G2/M cell cycle arrest in human host cells [37,46], we wanted to determine if the lack of induction in the Myc targets v1 and v2 gene sets in H. hammondi-infected cells reflected an inability of this species to manipulate cell cycle gene expression and ultimately progression through the cell cycle. We therefore looked specifically at transcript levels for a number of cell cycle related genes in T. gondii- and H. hammondi-infected cells and found multiple E2F-family related genes were of lower abundance in H. hammondi-infected cells compared to T. gondii-infected cells (e.g., E2F1, E2F2 and MCMs 4, 5, 6 and 7; Fig. 4B and Table S3). This observation is consistent with the lower absolute NES values (for negative enrichment) in H. hammondi infection compared to T. gondii strains for the G2M Checkpoint and E2F Targets gene sets (Fig. 3A, Table S2), and suggests that H. hammondi-infected cells may be in a different cell cycle state. Consistent with this idea, we found that that H. hammondi-infected cells had significantly higher transcript abundance for GADD45A, B and G (genes involved in DNA damage-induced growth arrest) and CDKN1A and CDKN1C compared to T. gondii-infected cells (Fig. 4B), suggesting that H. hammondi-infected cells may have a phenotype consistent with DNA damage-induced cell cycle arrest. A role for activation of DNA-damage-mediated pathways is also supported by IPA, in which we identified 38 H. hammondi-specific pathways including DNA damage-induced 14-3-3σ signaling which contains a number of DNA damage-induced cell cycle checkpoint proteins [47,48] (Fig. 3Sb and Table S4; highlighted in blue). Combined with the differential MYC targets gene sets that were suppressed during H. hammondi infection compared to T. gondii (Figs. 3A and 4A), these data suggest that T. gondii and H. hammondi-infections may have a divergent impact on the cell cycle of the host cells that they infect. While T. gondii induces progression through S phase and into G2/M, H. hammondi infection may lead to cell cycle arrest (in either G1/S or G2/M) via DNA damage-related stress responses [49].
Upstream regulator analysis reveals differential regulation of cell cycle-related transcription factors by H. hammondi infections
Using IPA upstream regulator analysis, we identified cascades of upstream transcription regulators that might be responsible for driving the observed transcriptional changes after infection with T. gondii and H. hammondi. We shortlisted upstream transcription regulators in response to H. hammondi or T. gondii infection (compared to mock-infected THP-1 cells) using z scores ≥ 2 for activation and ≤ −2 for inhibition (Fisher’s exact test; overlap p <0.01). From the analysis we identified type/strain-specific and common transcription regulators activated or inhibited in T. gondii and/or H. hammondi infections (Figs. 4C-E, Table S5). Common upstream transcription regulators include many well-known signaling mediators involved in T. gondii infection (e.g. multiple IRFs and STATs including IRF1, 3 and 5 and STAT1) as well as the NFκB p65-encoding gene RELA (Fig. 4C).
While most upstream regulators were shared between T. gondii and H. hammondi-infected cells (which was consistent with the number of infection-altered pathways shared between them at the transcriptional level), we did identify FOXM1 as a significant T. gondii-specific upstream transcription regulator (average Z-score in T. gondii of 2.31 compared to −0.79 in H. hammondi), and CDKN2A as being much more highly enriched in H. hammondi-infected cells (average Z-score in T. gondii of 1.96 compared to 5.08 in H. hammondi; Figs. 4D and E). FOXM1 is a component of the of the FOXM1-MMB complex which is active during G2/M and promotes transcription of E2F target genes, while CDKN2A (and CDKN1A as described above) are both involved in DNA-damage or stress-induced cell cycle arrest via CDK inhibition [50–53]. This provides further clues that H. hammondi might be regulating cell cycle progression differently than T. gondii during infection, and could be doing this in manner very similar to P53 mediated cell cycle arrest (Fig. 4F).
H. hammondi and induced cytokine secretion T. gondii-infected THP-1 cells are in distinct phases of the cell cycle
We analyzed cell cycle progression of sporozoite-infected cells using flow cytometry. Consistent with the literature [37,46], a TgVEG-infected THP-1 cells showed a distinct cell cycle profile compared to uninfected cells, with a small but evident subpopulation of cells that were in G2/M (Fig. 5A). In contrast, HhAmer-infected cells looked much more similar to uninfected THP-1 cells with respect to the cell cycle (Fig. 5A). In a separate experiment we compared RH88-(type I T. gondii tachyzoites) infected THP-1 cells to THP-1 cells infected with HhAmer sporozoites, and again found that in contrast to T. gondii-infected THP-1 cells which exhibited a prominent G2/M peak, HhAmer-infected cells had a host cell cycle profile that lacked this peak and was much more similar to uninfected cells (Fig. 5B). All supernatants were assayed for CXCL10 secretion to verify H. hammondi infection (Fig. 5SA).
Among the gene sets that were significantly enriched in H. hammondi-infected THP-1 cells compared to T. gondii was the Friedman Senescence Up gene set [54] with an NES of 3.3942 (Fig. 5C), suggesting that H. hammondi-infected cells may have some similarities to senescent cells which would not only explain differences in the host cell cycle but also the increased production of inflammatory cytokines via the Senescence-Associated Secretory Phenotype (SASP) [55]. This hypothesis was further supported by IPA® analyses showing significant enrichment of genes in the DNA damage-14-3-3σ signaling pathway (Fig. 3Sa), the higher expression of CDKN1A, increased CDKN2A signaling along with reduced FOXM1 signaling (Fig 4). One of many defining characteristics of senescent cells is the production of lysosomal senescence-associated β-galactosidase (gal) [56–62] and β-gal enzyme activity can be detected both in cells and supernatants [63]. We evaluated this in parasite-infected THP-1 cells and found no significant increases in β-gal activity in THP-1 cells infected with either parasite species, even though β-gal activity did increase in Phleomycin-treated THP-1 cells (Fig. 5SB). However, we did find that supernatants from H. hammondi-infected THP-1 cells contained significantly higher levels of IP10/CXCL10 [64,65], RANTES/CCL5 [65–67], and MIP-1a/CCL3 [68,69], which are also well known SASP factors (Tukey’s multiple comparisons test *p <0.05), and this correlated with a similarly robust increase in transcript abundance for these chemokines (Figs. 5D-E and Table S6).
Using CXCL10 secretion as a sentinel chemokine for SASP, as expected we found that H. hammondi (HhAmer)-infected cells produced significantly more CXCL10 compared to mock-infected cells (2600 pg/105 cells; Fig. 6A, left; Tukey’s multiple comparisons test, ****p<0.0001), while supernatants from TgVEG-infected cells did not contain significantly higher CXCL10 compared to mock-infected cells (Fig. 6A; left). When Transwell® inserts were used to separate parasites from THP-1 cells, CXCL10 induction by H. hammondi was completely abrogated (Fig. 6A), and heat-killed parasites of both species also failed to induce CXCL10 (Fig. 6B).
The SASP can be transferred to bystander cells by paracrine signaling [70], and to test this we used Transwell® inserts to separate H. hammondi-infected THP-1 cells from bystander and quantified CXCL10 induction by qPCR. We found that bystander cells exposed to HhAmer and host excretory/secretory products (ESPs) from the Transwell® insert expressed significantly higher CXCL10 transcript as compared to bystander cells of the Transwell® insert of the mock-infected THP-1 cells (by ~10-fold; Fig. 6C; Tukey’s multiple comparisons test, *p<0.01, ***p<0.001 and ****p<0.0001). RT-qPCR for H. hammondi GRA1 transcript confirmed the absence of parasites in the bystander cells (Table S7), and shows that CXCL10 transcript induction was due exclusively to a paracrine effect from infected THP-1 cells from the upper chamber of the Transwell® insert. CXCL10 was detected in supernatants in both conditions, since this chemokine could freely diffuse after its secretion. In contrast to CXCL10, we did not observe any increase in transcript abundance for CCL22 in bystander cells (Fig. 6D). CCL22 is also induced by T. gondii and H. hammondi infection [24], but is not a typical constituent of the SASP [71]. Taken together this provides further evidence for as SASP-like response in H. hammondi-infected, but not T. gondii-infected, THP-1 cells.
T. gondii- conditioned medium suppresses H. hammondi-induced cytokine secretion in THP-1 cells and increases replication rates of H. hammondi in primary human fibroblasts
Next, we speculated that T. gondii might be capable of suppressing the secretory response by H. hammondi. We exposed THP-1 cells to TgVEG sporozoites, or to 4 h conditioned medium from TgVEG-infected THP-1 cells, for 4 h prior to infection with H. hammondi. Interestingly we found that TgVEG sporozoites and TgVEG-conditioned media both suppress H. hammondi-mediated cytokine secretion (Figs. 7A and B). However, when THP-1 cells were pre-infected with H. hammondi sporozoites, T. gondii-mediated cytokine suppressive effect was no longer observed (Fig. 7A).
Our previous work found that H. hammondi replicates ~4-8-fold slower than T. gondii [34], and given the divergent effects on the host cell cycle we hypothesized that T. gondii infection may result in a more permissive host cell state, and that this may alter the replication rate of H. hammondi. To address this we infected HFF cells that were pre-treated with TgVEG/THP-1 conditioned medium (collected 4 h post THP-1 infection) with freshly excysted HhAmer sporozoites and quantified parasite vacuole number and size. Remarkably, we observed a significant increase in H. hammondi vacuole size in pre-conditioned HFFs compared to untreated HFFs in two separate experiments performed with completely distinct parasite batches and reagents (Figs. 7C and 7S; *p<0.05, unpaired t test comparing means of vacuole size of the coverslips; means of unconditioned vs. conditioned well are 2.141±0.1479 and 2.869±0.2169 respectively). In particular we saw a decrease of vacuoles with single parasite and increase of two, four and eight parasites per vacuole in the TgVEG/THP-1 conditioned medium (Fig. 7C). While there was variation in the impact of TgVEG-conditioned media on H. hammondi replication (data in Fig. 7S show a much more dramatic increase in vacuole size compared to data in Fig 7C), these data clearly show that T. gondi-infected host cells produce (a) factor(s) that can render bystander cells more permissive to parasite replication (at least for H. hammondi), and that the replication rate of H. hammondi can at least be partially determined by the condition of the host cell.
Discussion
It is well known that T. gondii tachyzoites activate a potent host immune response. Despite that, once inside a host cell, T. gondii is able to modulate, survive and even evade immune responses, all of which lead to replication and ultimate dissemination to distant host tissues. This has largely been attributed to the powerful strategy that T. gondii employs to co-opt host gene expression and to protect itself from host proteins designed to destroy these parasite-containing vacuole [15–17,72,73]. In contrast little is known about the impact of H. hammondi on the host cell, and whether its comparatively “avirulent” lifestyle can be attributed at all to differences in the host response. Many of the unique aspects of H. hammondi biology, including its comparatively slower replication rate, spontaneous (and terminal) cyst formation and inability to be lethal even in IFNγ knockout mice [29,34,74], suggest that it is engaged in an inflexible developmental program once sporozoites are released from the oocyst. In the extreme, then, one could hypothesize that the nature and magnitude of the host response may be completely irrelevant to the outcome of infection with H. hammondi. However our data provide strong evidence to the contrary.
At first glance, our comparisons of the transcriptomes T. gondii and H. hammondi-infected host cells indicate that these parasites modulate the host cell using qualitatively similar strategies. Over 95% of the underlying altered pathways were shared between T. gondii and H. hammondi, and fell into categories like inflammation, apoptosis, cell growth, and metabolism that have long been known to be targeted by T. gondii [4–6] (Fig. 3). However when the magnitude of the response was taken into account (whether measured by normalized enrichment scores from GSEA or cytokine secretion), it is clear that H. hammondi-infected cells are responding much more robustly to infection compared to those infected with T. gondii. This effect, which is also recapitulated in an animal model of infection when differences in parasite burden is taken into account (Fig. 2), suggests either that a) H. hammondi has fewer countermeasures to counteract the host response than T. gondii or b) H. hammondi has a factor or factors that cause hyperactivation of the immune response (via secreted effectors, and/or production of pathogen- or damage-associated molecular patterns).
Intriguingly our results show that the most obvious differential host response in THP-1 cells infected with T. gondii and H. hammondi is the regulation of cell cycle (Figs. 3–5), a phenomenon long-described in T. gondii but whose role in determining infection outcome is unknown. Unlike T. gondii infection that actively induces host c-Myc [75], the MYC targets v1 and v2 gene sets were suppressed in THP-1 cells infected with H. hammondi (Figs. 3 and 4). T. gondii infection leads to an accumulation of cells in the G2/M phase [37,46] and in our data this was accompanied by increased transcript abundance of FOXM1-transcribed genes (Fig. 4; [50,52]). In contrast, H. hammondi leads to increased transcription of multiple GADD45 genes which block cell cycle progression as a part of the DNA damage response [76–78]. DNA damage can induce P53 activation and increase transcription of CDKN1A, which we observed only in H. hammondi-infected cells (Fig. 4D-F; [48]). DNA damage responses can have dramatic effects on cell biology that are independent of cell cycle arrest, and one of these is the Senescence Associated Secretory Phenotype (SASP; [65,70,71]). We observed increased production of multiple chemokines typically associated with the SASP, including CXCL10 which not only was uniquely present in H. hammondi-infected supernatants but was also activated in bystander cells via paracrine signaling (Figs. 5 and 6). We did not detect elevated β-gal activity in host cells infected with H. hammondi, cell types and different cellular conditions could also affect the reliability of any types of senescence markers [79]. It is very intriguing to postulate that the manipulation of the host cell cycle by T. gondii, and specifically its ability to induce S-phase transition and ultimately G2/M arrest, is a means to prevent the host cell from becoming senescent and activating the SASP in both the infected cell and ultimately in neighboring bystander cells. Disrupting the ability of the host cell to secrete cytokines like CXCL10 would certainly be advantageous given the importance of this chemokine to maintain T-cell populations capable of controlling T. gondii proliferation [41,80].
Similar to paracrine effects of the H. hammondi-induced SASP on bystander cells, T. gondii induces host- and infection-altering factors to be produced by host cells as well. H. hammondi has a markedly slow replication rate compared to T. gondii (shown in multiple studies including [34,74]), but this slow rate of division appears to be flexible depending upon the host cell environment. The ability to alter the replication of an intracellular parasite like H. hammondi using T. gondii-conditioned medium (Fig. 7) leads to a number of interesting biological questions about the nature and identity of the factors capable of controlling parasite replication. Supernatants from T. gondii-infected cells have been found to have a variety of effects on recipient cells that vary depending on the donor and recipient cell types (e.g., [81,82]). Although to date the factors themselves have not been identified, interaction between T. gondii and the host cells which reduced UHFR1 and cyclin B1 has been shown to be advantageous for T. gondii replication [46]. Our data provide further compelling evidence for their importance in determining the outcome of T. gondii infections. In addition to this interesting biology there are additional practical implications of this work with respect to the H. hammondi/T. gondii comparative system. The use of T. gondii-conditioned media may be a first step in further improving culture conditions for H. hammondi, with the ultimate goal of promoting long term in vitro culture which is not yet possible [29,74].
Materials and Methods
Cells
Human foreskin fibroblasts(HFFs) and human monocytes (THP-1) were maintained in cDMEM (100 U/ml penicillin/streptomycin, 100 μg/ml streptomycin, 2 mM L-glutamine, 10 % FBS, 3.7 g NaH2CO3/L, pH7.2; ThermoFisher Scientific) and cRPMI-1640 (100 U/ml penicillin/streptomycin and 10 % FBS; ThermoFisher Scientific) respectively. All cells were grown at 37 °C in 5 % CO2. One day prior to parasite infection, THP-1 cells growth media were replenished with 20 % (v/v) of new media (cRPMI). All THP-1 cell infections were performed in cDMEM.
Mice
Balb/c mice were purchase from Jackson Laboratory and were female aged 6-8 weeks. All animal experiments were approved by the local IACUC at the University of Pittsburgh (Protocol #18032113 and #092011), with euthanasia and anesthesia conducted according to AVMA guidelines. Anesthesia of the rodents were performed with CO2 followed by decapitation of the animals. Euthanasia was performed with isofluorine.
Parasites
Oocysts of Toxoplasma gondii (Tg) genotype I (GT1), II (ME49) and III (VEG) and Hammondia hammondi (Hh) American (Amer) and Ethiopian-1 (Eth1) isolates were harvested from cat feces 7-11 days after feeding mouse tissues (brain for T. gondii, leg muscle for H. hammondi) infected with parasites to 10-20 week old cat free of pathogens [83,84]. Unsporulated oocysts were isolated via sucrose floatation and placed at 4 °C in 2 % H2SO4 to encourage sporulation and for long-term storage.
T. gondii and H. hammondi oocyst excystation
Sporulated oocysts were washed 3 X in Hank’s balanced salt solution (HBSS; Life Technologies) and treated with 10% bleach (in PBS) for 30 min with shaking at room temperature. Next, bleach was washed off using HBSS and parasites were added to 4 g sterile glass beads (180 μM; Sigma-Aldrich). Oocysts were vortexed on high speed for 15 s on/15 s off for a total duration of 2 min to disrupt the oocyst wall mechanically. Sporocysts were pelleted by centrifugation at 1,000 x g for 10 min. Pellet was resuspended in 5 ml of pre-warmed and freshly made, 0.22 μM filtered excystation media (0.1 g porcine trypsin (Sigma-Aldrich), 2 g Taurocholic Acid (Sigma-Aldrich) in 40 ml PBS, pH 7.5). Sporocysts were incubated in 37 °C water bath with 5 % CO2 for 45 min and syringe-lysed using a 25 and a 22 gauge needle. H. hammondi sporocysts were syringe-lysed with a 25 gauge needle. To quench the excystation media, 7 ml of cDMEM was added. Excysted parasites (sporozoites) were pelleted and resuspended in cDMEM and grown onto monolayers of HFFs for overnight at 37 °C in 5 % CO2. Freshly excysted sporozoites were used in in vivo and experiments analyzing induction of CXCL10 in Transwell®, heat-killed assays and parasite growth assays.
Human monocyte cell line infection
Prior to infection, THP-1 cells were seeded at 1 × 105 cells/well in 24-well plates in cDMEM. To prepare parasites for the infections, HFFs monolayers containing parasite sporozoites were scraped, syringe-lysed, and pelleted (in some cases, freshly excysted oocysts were used). After resuspending pellet in cDMEM, the parasite mixture was filtered through a 5 μM syringe-driven filter (Millipore). THP-1 cells were infected with either T. gondii or H. hammondi at multiplicity of infection (MOI) of 4, 2 or 1.6. Three biological replicates were made for each parasite infection. THP-1 cells were also mock-infected with parasite filtered through a 0.22 μM syringe-driven filter (Millipore).
RNA isolation
RNA was collected at 24 h post-infection from parasite- and mock-infected cells using the RNeasy Kit according to the manufacturer instructions (Qiagen). QIAShredder spin columns (Qiagen) were used to homogenize the samples prior to RNA extraction and contaminating DNA was degraded using RNase-free DNase (Qiagen). RNA was eluted in 50 μl of RNase-free water and gel electrophoresis and Nanadrop RNA quantification were performed to ensure RNA integrity. One mock and three biological infection replicates were done for all experiments. Total RNA samples were kept at −80 °C.
mRNA-sequencing and data processing
mRNA-sequencing libraries and Illumina next generation sequencing were performed at the Core Facility at the University of Pittsburgh. Integrity of the RNA was analyzed with Agilent 2100 Bioananalyzer and all purified RNA samples had RIN scores >9. RNA was sequenced using NextSeq 550 (Illumina) and were pooled and sequenced over four lanes. Strand-specific, 150 bp, single-end RNA-sequencing was performed. Read libraries were mapped to the human genome (Homo sapiens ensemble v81; hg38) and transcriptome (Homo sapiens ensemble v81; hg38) with default options on CLC Genomics Workbench v11.0. Fastq files have been deposited in the NCBI short read archive (Accession number: SRX3734421-8, and accession numbers pending for some of the Fastq files).
Differential expression analysis using DESeq2 and Pre-ranked Gene Set Enrichment Analysis
DESeq2 [40] was used to perform differential expression analysis of genes in THP-1 cells infected with T. gondii and H. hammondi. Raw reads (total gene reads; exported from CLC Genomics Workbench) with at least 1 read count in all samples were analyzed. Prior to analyzing differential gene expression, integrity of the data was examined using principal component analysis (PCA) and distances of all samples were calculated (embedded in the DESeq2 package). Genes were considered to be differentially expressed in THP-1 cells if the log2 fold-change was ≥ 1 or ≤ −1 and with a padj value (alpha) <0.01.
Pre-ranked Gene Set Enrichment Analysis (GSEA; [45]) was performed to compare gene sets that were enriched in THP-1 cells in relation to parasite infections. Ranked list calculating the fold-change difference (subtracting normalized log2 fold-change of infected host cells from mock infected host cells) was used for the analysis.
Ingenuity® Core Pathway Analysis
Core analysis from Ingenuity® Pathway Analysis (IPA; Qiagen) was used to examine biological relevance of the RNA-seq data. Canonical pathways and upstream regulators that are over-represented in T. gondii and H. hammondi infection were analyzed using log2 fold-change and padj value obtained in DESeq2 [40]. Only genes that were deemed significantly expressed were used in the analysis (log2 fold-change ≥ 1 or ≤ −1 and padj value < 0.01 for Infected vs. Mock). IPA default settings were used for the analysis. For both canonical pathway and upstream regulator analysis, pathways and genes with activation z scores ≥ 2 were deemed to be activated while activation z scores ≤ −2 were deemed to be inhibited for different analysis. Threshold for p value significant used in the analysis was <0.05 or −log(p-value) >1.3.
Mouse peritoneal cell collection
For mouse infections, 40,000 24 h parasite zoites or 20,000 freshly excysted sporozoites (TgVEG or HhAmer) in 200 μL of PBS were intraperitoneally injected into 7-9 week old mice. Mice were euthanized with CO2 according to IUCAC protocols and dissected. Peritoneal cells were collected in 3 ml PBS as described previously [85].
Multiplex cytokine analysis
THP-1 cells were infected with MOI 4 of T. gondii or H. hammondi (same infection set up as the RNA-seq experiments). THP-1 cells infected or mock-infected with parasites were pelleted at 1,000 x g for 10 min. Supernatant were collected and stored at −80 °C. Luminex was performed at the Luminex Core Facility University of Pittsburgh Cancer Institute. Each sample was measured (fluorescence intensity) in duplicate.
Cytokine analysis using ELISA
Concentrations of the human pro-inflammatory cytokine CXCL10 and CCL22 and mouse Ifnγ, Il12p40, Cxcl10 and Ccl22 in mouse peritoneal cells were analyzed by ELISA according to the manufacturers instructions (Human and Mouse DuoSet ELISA respectively, R&D Systems). For relative chemokine concentration in comparison to parasite load, log2 values of the absolute concentration of chemokines was calculated and relative expression in relation to parasite burden was calculated using the 2−ΔΔCT method (ΔΔCT = log2 chemokine concentration – ΔCT GRA1).
Transwell® and heat-killed parasite infections
Freshly excysted TgVEG or HhAmer was added to THP-1 cells seeded in 24-well plates (control) or added onto Transwell® inserts (Corning). For heat-killed parasite infection, freshly excysted sporozoites were heat-killed at 95 °C for 5 min and cooled to room temperature before infecting THP-1 cells. Multiplicity of infection of 2 was used in these experiments. Supernatants were collected either by pelleting the cells or collected from the bottom chamber of the Transwell® setup.
Reverse transcriptase-quantitative PCR
RNA were extracted using RNeasy RNA extraction kit as above and according to manufacturer’s instructions (Qiagen). cDNA was reverse transcribed from 1 μg of RNA using SuperScript IV First-Strand synthesis system (ThermoFisher Scientific). RT-qPCR was performed using QuantStudio with SYBR Green (Applied Biosystems). The PCR mixture contained 1 X SYBR Green buffer (BioRad), 0.25 μl of forward and reverse primers (Table S8) and cDNA. Genes were amplified using the standard protocol (95 °C for 10 min and 40 cycles of 95 °C for 15 sec and 60 °C for 1 min). Data were acquired and analyzed using QuantStudio™ Design & Analysis Software (ThermoFisher Scientific) and data exported to Microsoft Excel for threshold values (CT) and 2−ΔΔCT for fold change analysis. GAPDH and Gapdh was used as the human and mouse reference gene respectively (control) for the 2−ΔΔCT analysis. For in vivo 2−ΔΔCT analysis, relative expression of target genes were normalized against ΔΔCT of mock-infected mice. For relative expression normalized by comparison with parasite load, we used GRA1 and as a reference gene for parasite burden. To do this, the ΔCT of target genes and GRA1 were first obtained (CT target gene or GRA1 – CT Gapdh). Then, ΔΔCT target gene was calculated (ΔCT target genes – ΔCT GRA1). Relative expression was finally calculated using the 2-ΔΔCT methods for respective target genes. Primers validation and melt curve analysis were performed to ensure integrity of the RT-qPCR. No RT and water controls were also included in every RT-qPCR.
Propidium iodide cell cycle analysis
Cell cycle profiles of THP-1 cells were analyzed with propidium iodide staining according to manufacturers instructions (ThermoFisher Scientific). THP-1 cells were infected with T. gondii or H. hammondi at an MOI of 2. For uninfected cells, THP-1 cells were seeded at the same time as the other cells in different treatment groups and an equal volume of media was added to the uninfected cells in place of parasite infection. At 20 h post-infection, cells were pelleted and washed 1 X with chilled PBS and fixed in chilled 80% ethanol at −20 °C for overnight. Cells were then washed with chilled PBS and rehydrated for 15 min with chilled PBS. Cells were stained with 2 μg/mL propidium iodide/RNaseA (ThermoFisher Scientific) solution for 15 min at room temperature with gentle agitation. Stained cells were analyzed immediately by flow cytometry using LSR II (BD Biosciences) at the United Flow Core at University of Pittsburgh or kept at −20 °C in the dark for no longer than 24 hr until flow cytometry analysis. Cell cycle profiles were analyzed with ModFit LT 5.0 (VSH).
β-galactosidase activity assay
β-galactosidase activity was analyzed in THP-1 cells and secreted proteins using the β-galactosidase assay kit (ThermoFisher) according to manufacturers instructions. Briefly, THP-1 cells were inoculated with T. gondii or H. hammondi, treated with 1 μg/mL of phleomycin in cDMEM. At 72 h post-infection, cells were pellected at 300 x g for 10 min and supernatant was collected for β-galactosidase activity assay. The cell pellets were washed 2 times with PBS and lyzed with IP lysis buffer (ThermoFisher) with Halt Protease Inhibitor Cocktail (ThermoFisher), before analyzing for β-galactosidase activity. β-galactosidase activity was determined after 1 h of incubation at 37 °C and absorbance was read at 420nm.
H. hammondi growth (vacuole size) under T. gondii/THP-1 conditioned medium
T. gondii Veg and THP-1 cells ESP was prepared in THP-1 cell infections as described above. Briefly THP-1 cells were infected with T. gondii VEG with an MOI of 2 for 4 h. HFF cells pre-seeded on coverslips were pre-conditioned with TgVeg/THP-1 cells ESP and freshly excysted Hh sporozoites were added into the same wells. At 72 h post-infection, coverslips were washed 2 times with PBS and fixed with 4% paraformaldehyde and stained with DAPI. The total number of visible vacuole and number of parasites per vacuole was counted from two coverslips for each infection conditions. Differences in the growth rates of the two conditions were determined by calculating the mean vacuole size of each coverslips.
Acknowledgments
The authors thank Cori Richards-Zawacki and Carolyn Coyne for sharing cells and equipment.
References
- 1.↵
- 2.
- 3.↵
- 4.↵
- 5.
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.
- 11.
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.
- 58.
- 59.
- 60.
- 61.
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.