The UPR sensor IRE1α promotes dendritic cell responses to control Toxoplasma gondii infection

Abstract

The unfolded protein response (UPR) has emerged as a central regulator of immune cell responses in several pathologic contexts including infections. However, how intracellular residing pathogens modulate the UPR in dendritic cells (DCs) and thereby affect T cell‐mediated immunity remains uncharacterized. Here, we demonstrate that infection of DCs with Toxoplasma gondii (T. gondii) triggers a unique UPR signature hallmarked by the MyD88‐dependent activation of the IRE1α pathway and the inhibition of the ATF6 pathway. Induction of XBP1s controls pro‐inflammatory cytokine secretion in infected DCs, while IRE1α promotes MHCI antigen presentation of secreted parasite antigens. In mice, infection leads to a specific activation of the IRE1α pathway, which is restricted to the cDC1 subset. Mice deficient for IRE1α and XBP1 in DCs display a severe susceptibility to T. gondii and succumb during the acute phase of the infection. This early mortality is correlated with increased parasite burden and a defect in splenic T‐cell responses. Thus, we identify the IRE1α/XBP1s branch of the UPR as a key regulator of host defense upon T. gondii infection.

Infection of dendritic cells with Toxoplasma gondii triggers the IRE1α pathway in a MyD88‐dependent manner, a process required to promote T cell responses and to control parasite dissemination in infected mice.

Introduction

Many chronic infections are caused by intracellular microorganisms, which actively manipulate host responses to escape innate immune defenses. There is growing evidence that activation of appropriate antimicrobial activities of infected immune cells relies both on the detection of microbial compounds by pattern recognition receptors (PRRs) and on efficient sensing of variations in cellular homeostasis induced by the intracellular pathogen. An essential stress‐sensing pathway is the unfolded protein response (UPR). In metazoans, the UPR is activated by the coordinated action of three endoplasmic reticulum (ER) transmembrane sensors: inositol‐requiring enzyme 1α (IRE1α), PKR‐like ER Kinase (PERK), and activating transcription factors 6α/β (ATF6α/β). The downstream transcriptional programs of the UPR are primarily directed toward restoring proteostasis by increasing ER capacity as well as decreasing protein load, through transient translational inhibition and the process of ER‐associated peptide degradation (ERAD) (Walter & Ron, 2011). IRE1α, the most conserved UPR sensor, contains a serine/threonine kinase domain and an endoribonuclease domain. Upon sensing unfolded proteins, IRE1α oligomerizes, is activated by autophosphorylation and uses its endoribonuclease activity to excise an intron from the transcription factor XBP1 (X‐box binding protein). This unconventional splicing reaction of the Xbp1 mRNA generates the active transcription factor XBP1 spliced (XBP1s) (Yoshida et al, 2001). In addition to splicing XBP1, upon prolonged or severe stress, IRE1α degrades mRNAs in proximity to the ER in a process termed regulated IRE1‐dependent decay (RIDD) of mRNA (Hollien et al, 2009). XBP1s and ATF6α up‐regulate the expression of genes involved in increasing the protein‐folding capacity of the ER (chaperones) as well as ER expansion through phospholipid synthesis (Sriburi et al, 2007; Lee et al, 2008; Shoulders et al, 2013). However, upon irremediable ER stress, the UPR switches to a pro‐apoptotic mode by PERK‐mediated activation of CHOP expression. During the last decade, a growing number of studies have demonstrated an intricate molecular crosstalk between the immune and UPR pathways, opening new perspectives to treat infections and chronic inflammatory diseases by UPR modulators (Grootjans et al, 2016; Osorio et al, 2018; Reverendo et al, 2019). Notably, TLR stimulation of macrophages and dendritic cells (DCs) contributes to XBP1s and CHOP induction that directly activate the transcription of genes encoding pro‐inflammatory cytokines, such as IL‐6, TNF‐α, IFN‐β, and IL‐23 (Martinon et al, 2010; Cubillos‐Ruiz et al, 2015; Márquez et al, 2017; Smith, 2018; Mogilenko et al, 2019). The IRE1α/XBP1s arm of the UPR is also a key regulator of antigen presentation in DCs via partially characterized mechanisms involving XBP1s‐mediated modulation of lipid synthesis (Cubillos‐Ruiz et al, 2015) and RIDD‐mediated regulation of the antigen presentation machinery (Osorio et al, 2014, 2018; Medel et al, 2018). Several pathogens interfere with the function of the host ER as part of their life cycle and can selectively activate or repress distinct arms of the UPR (Celli & Tsolis, 2015; Galluzzi et al, 2017; Smith, 2018). UPR activation can be beneficial to the host by enabling the establishment of an appropriate innate response directed against the invading pathogen (Celli & Tsolis, 2015; Galluzzi et al, 2017; Smith, 2018). However, intracellular pathogens can also selectively modulate UPR pathways to promote their own survival and growth.

Toxoplasma gondii (T. gondii) can be considered as an opportunist parasite as severe symptoms develop in immuno‐suppressed patients. However, infection of immuno‐competent individuals by type II strains is characterized by bradyzoite‐containing cyst development in the brain, leading to life‐long chronic infection (Blanchard et al, 2015). Toxoplasma gondii is a very successful intracellular parasite able to invade any nucleated cells, including immune cells, in which the parasite rapidly replicates within a vacuole. To enable nutrient retrieval from the host and efficient escape from cellular defenses, the parasite secrete numerous effectors that modulate host cell signaling pathways (Hakimi et al, 2017). In addition, parasite effectors anchored at the limiting membrane of the vacuole promote recruitment of host organelles, such as mitochondria (Pernas et al, 2014, 2018), the Golgi apparatus (Deffieu et al, 2019) and the ER (Sinai et al, 1997; Goldszmid et al, 2009). The impact of T. gondii‐induced ER association with the parasitophorous vacuole (PV) on ER homeostasis has not been investigated. Here, we demonstrate that infection of DCs by the parasite induces a specific UPR signature, mainly characterized by a MyD88‐dependent activation of the IRE1α /XBP1s pathway. In addition, the presence of live parasites actively modulates specific branches of the UPR, in contrast to cell stimulated with heat‐killed parasites or soluble T. gondii antigens. Induction of XBP1s leads to enhanced IL‐6 and IL‐12 secretion, while IRE1α promotes MHCI antigen presentation independently of XBP1s induction. Moreover, activation of the IRE1α/XBP1s pathway in splenic DCs is required to promote the establishment of an efficient T‐cell response and to control parasite dissemination in infected mice.

Results

Toxoplasma gondii induces IRE1α/XBP1s activation in BMDCs

To monitor whether T. gondii infection activates the ER stress response, we infected bone marrow‐derived DCs (BMDCs) with type II Pru parasites for 16 h and analyzed IRE1α, PERK, and ATF6 pathway activation by measuring the expression of known downstream targets. As a positive control, we stimulated cells with tunicamycin (TN), a potent pharmacologic ER stress inducer. In comparison with TN, T. gondii infection led to only partial induction of specific UPR branches (Fig 1A and Appendix Fig S1). Notably, infection did not induce a classic ER stress signature as illustrated by the lack of induction of the ER chaperones Hsp90b1 (encoding GRP94) and Sec61a1 and the slight induction of Hspa5 (encoding BIP) compared to TN stimulation (Fig 1A and B and Appendix Fig S1). On the contrary, the IRE1α branch was induced as illustrated by the increased expression level of Ern1 (encoding IRE1α) and Xbp1s and its target genes Dnajb9 (encoding ERdj4) and Sec24d (Fig 1A and B and Appendix Fig S1). RIDD was not activated, as demonstrated by a lack of degradation of the prototypical RIDD target Bloc1s1 mRNA upon T. gondii infection (Fig 1B; Hollien et al, 2009; Bright et al, 2015). Western blot analysis of T. gondii‐infected BMDCs revealed an increasing expression in IRE1α levels with time and the induction of the spliced version of XBP1, which reached its maximal level at 4 h (Fig 1C). Sustained activation of IRE1α and transient increase in XBP1s expression was confirmed in so‐called Flt3/Notch‐DCs, an alternative system to generate cDC1s in vitro (Kirkling et al, 2018; Fig 1D). The PERK branch appeared not activated, neither at mRNA neither at protein level. PERK downstream effectors ATF4 and CHOP (encoded by Ddit3) were not induced, in contrast to TN stimulation (Fig 1A and B and Appendix Fig S1). In line with this, phosphorylation of PERK and its target eIF2α appeared decreased compared to steady‐state levels, and CHOP levels remained undetectable in infected BMDCs and Flt3/Notch‐DCs (Fig 1C and D). As reported before, splenic DCs display high level of p‐eIF2α at steady state suggesting that control of protein translation is essential to regulate DC homeostasis (Clavarino et al, 2012). GADD34 interacts with the phosphatase protein 1 (PP1) to dephosphorylate eIF2α to alleviate the protein translation block (Novoa et al, 2001; Clavarino et al, 2012). Ppp1r15a (encoding GADD34) gene expression was strongly increased in T. gondii‐infected BMDCs, consistent with a presumable role of GADD34 in the inhibition of eIF2α phosphorylation (Fig 1B). Finally, T. gondii infection caused a strong down‐regulation of the Atf6 gene and its target gene Herpud1 (Fig 1A and B), suggesting that the parasite actively modulates the ATF6 pathway in DCs, as proposed in a previous study (Yamamoto et al, 2011). In summary, infection of T. gondii leads to a mixed outcome on UPR signaling branches with a down‐regulation of the ATF6 branch and an activation of the IRE1/XBP1 branch.

Figure 1. The IRE1α/ XBP1s pathway is activated in BMDCs infected with T. gondii .

• A, B
  BMDCs were infected with Pru parasites (at a MOI of 4 parasites/BMDC) or treated with tunicamycin (TN) for 16 h, and the expression of UPR target genes was analyzed by RT‐qPCR. The results are expressed as (Log₂) fold change (A) or relative mRNA level (B) compared to non‐infected (NI) conditions. IRE1α/XBP1s pathway: Ern1 (IRE1α), Xbp1s, Dnajb9 (ERdj4), Sec24D, Bloc1s1, and Sec61a1; PERK/ATF4 pathway: Atf4, Ddit3 (CHOP), and Ppp1r15a (GADD34); ATF6 pathway: Atf6 and Herpud1; classical ER chaperons: Hspa5 (BIP) and Hsp90b1 (GRP94). Results are normalized to the housekeeping gene Gapdh. Unpaired Student's t‐test, ns: P > 0.05; **P < 0.01; ***P < 0.001; mean ± SEM (n = 2–5 independent experiments depending of the studied gene, 3 mice/independent experiment).
• C, D
  Immunoblot analysis of IRE1α, PERK, Phospho‐PERK (detected by the presence of a second upper band after migration on a 6% acrylamide gel), Phospho‐eIF2α, CHOP, and XBP1s protein expression in non‐infected BMDCs, BMDCs treated with TN and Pru‐infected BMDCs (C) and in non‐infected and Pru‐infected Flt3/Notch‐DCs (D) for the indicated times. GAPDH and tubulin were used as loading controls. The same blots are shown in Fig EV3B and C.

Source data are available online for this figure.

Induction of XBP1s is under the control of TLR signaling

UPR activation and downstream cytokine responses in DCs and macrophages were demonstrated to be under the control of PRR signaling (Martinon et al, 2010; Smith, 2018). Despite the established role of TLR11/12 in parasite detection in infected mice (Andrade et al, 2013; Koblanski et al, 2013; Hou et al, 2011; Yarovisnki et al, 2005), the innate sensing mechanisms of parasites residing into a intracellular vacuole are not fully elucidated (Poncet et al, 2019) and presumably involve distinct signals depending on early parasite attachment to host cells and subsequent partial vacuolar lysis leading to antigen escape into the host cytosol (Fisch et al, 2020; Kongsomboonvech et al, 2020; Sardhina‐Silva et al, 2019; Lee et al, 2015). Thus, to establish the role of innate sensing in UPR activation, we used BMDCs differentiated from mice lacking the adaptor MyD88 (MyD88KO mice) (Deguine & Barton, 2014). This adaptor is essential to induce signaling downstream of all TLRs except for TLR3, which relies on the TRIF adaptor.

First, we confirmed that T. gondii stimulates the secretion of IL‐6 and IL‐12 over the time course of infection (Fig EV1A). In addition, and in contrast to the more virulent type I parasites (“RH” strain), which kill their host prematurely and do not establish latent infections in laboratory mice, type II parasites (“Pru” strain) specifically induce the expression and secretion of IL‐23 (Fig EV1B) at a late time point of infection (Fig EV1A), suggesting that distinct innate signaling pathways are involved in the detection of these parasite strains. Lack of MyD88 did not impact on parasite invasion as illustrated by a similar rate of infection between WT and MyD88KO BMDCs (Fig 2A), although we did note a slight alteration in parasite growth (Fig 2B). As expected, the absence of TLR stimulation led to a drastic loss of pro‐inflammatory cytokine production upon infection (Fig 2C), as revealed by a block in IL‐6, IL‐12, and IL‐23 production. Loss of MyD88 strongly impacted the UPR response in T. gondii‐infected BMDCs. The T. gondii‐mediated increase in Xbp1s and Dnajb9 expression levels was no longer observed, and we even noted a down‐regulation of their basal expression level upon infection (Fig 2D). This was also observed for Ddit3 (encoding CHOP) expression levels when comparing infected versus non‐infected conditions (Fig 2D). These results therefore indicate that two separate pathways appear to modulate the host UPR response upon infection with T. gondii. TLR signaling is required to induce the IRE1α/XBP1s pathway. In the absence of TLR signaling, it becomes obvious that a parasite‐dependent pathway causes the down‐regulation of all UPR branches. Of note, Ppp1r15a induction (GADD34) was not dependent on MyD88, suggesting that a distinct MyD88‐ and PERK‐independent signaling pathway induces the expression of this gene during parasite replication (Fig 2D). In line, a previous study demonstrated that GADD34 expression is induced in a IRF3/7‐dependent manner after poly:IC stimulation of DCs and controls cytokine production (Clavarino et al, 2012; Dalet et al, 2017). Thus, parasite RNA detection following vacuolar lysis could be responsible for GADD34 induction in type II infected DCs.

Figure EV1. Induction of cytokine production from infected BMDCs.

1. IL‐6, IL‐12, and IL‐23 cytokine levels from non‐infected BMDCs (NI) and BMDCs infected with Pru parasites for 6 h and 16 h measured by ELISA. Unpaired Student’s t‐test, *P < 0.05; **P < 0.01; ***P < 0.001; mean ± SEM (n = 4‐6 mice pooled from 2 independent experiments).
2. IL‐6, IL‐12, and IL‐23 cytokine levels from non‐infected BMDCs (NI) and type I RH or type II Pru‐infected BMDCs measured by ELISA. ANOVA Tukey’s multiple comparisons, ns: P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001; mean ± SEM (n = 4‐6 mice pooled from 2 independent experiments).

Figure 2. UPR and cytokine responses are under the control of MyD88 protein in BMDCs infected with T. gondii .

1. Mean percentage of Pru parasite infection of WT and MyD88KO BMDCs; mean ± SEM (n = 3 mice, 1 representative experiment from 2 independent experiments is shown). Mann–Whitney U‐test, ns: P > 0.05.
2. Mean percentage of parasitophorous vacuoles containing 1, 2, 4, or 8 parasites/vacuole (p/v) in WT and MyD88 KO BMDCs after 16 h of infection with Pru parasites. Mean values ± SEM n = 3 independent experiments (biological replicates). Unpaired Student's t‐test, ***P < 0.001.
3. IL‐6, IL‐12, and IL‐23 cytokine levels from WT and MyD88KO BMDCs non‐infected (NI) and infected with Pru parasites for 16 h. Two‐way ANOVA Tukey’s multiple comparisons, ****P < 0.0001; mean ± SEM (n = 4–6 mice pooled from 2 independent experiments).
4. mRNA relative expression level of Xbp1s, Dnajb9 (ERdj4), Ddit3 (CHOP), and Ppp1r15a (GADD34) from WT and MyD88KO BMDCs infected with Pru parasites for 16 h compared to non‐infected (NI) conditions. Results are normalized to the housekeeping gene Gapdh. Two‐way ANOVA Tukey’s multiple comparisons, ns: P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001; mean ± SEM (n = 4–5 mice pooled from 2 independent experiments).

Our results therefore indicate that induction of the IRE1α/XBP1 signaling branch by T. gondii infection relies on TLR‐dependent triggering, confirming earlier studies (Martinon et al, 2010).

Toxoplasma gondii triggers the association of host ER with the parasitophorous vacuole

Toxoplasma gondii actively modulates host cell responses by secreting numerous parasite effectors that highjack host signaling pathways and activate host compartment recruitment at the PV. Notably, host ER (hER) components are recruited in a Sec22b‐dependent manner at the PV, a process required to stimulate cross‐presentation of soluble secreted OVA antigens by transgenic OVA‐expressing parasites (Goldszmid et al, 2009; Cebrian et al, 2011). Whether this process impacts on ER homeostasis has never been investigated, nor whether this is needed for triggering of the UPR. First, we quantified hER recruitment in type II Pru parasite‐infected BMDCs. Electron microscopy analysis showed that almost all the examined PVs (n = 116/119) display hER membranes in close association with the PV delimiting membrane (Fig EV2A). Staining of the hER marker KDEL by immunofluorescence confirmed this association (Appendix Fig S2A). In addition, this process was also detected in infected human fibroblasts, in which KDEL‐positive tubular‐like structures were also visualized inside the PV space in close proximity to parasites (Appendix Fig S2B). Moreover, we examined hER recruitment in BMDCs incubated with heat‐killed (HK) parasites that are presumably internalized by phagocytosis or macropinocytosis and do not secrete parasite effectors to modulate host responses. In sharp contrast to live parasite‐containing vacuoles, HK parasite‐containing vacuoles were never found positive for hER recruitment at their limiting membrane (Fig EV2B). This suggests that hER association with the PV is not specific to immune cells and may be triggered by live parasites to support their survival or replication.

Figure EV2. hER is recruited at the PV of infected BMDCs.

1. Electron microscopy images showing the recruitment of hER at the limiting membrane of vacuole containing live Pru parasites. Right images (scale bars = 500 nm): zoom of the region indicated by a white frame on the left images (scale bars = 1 μm).
2. Pru Heat‐killed (HK) parasite‐containing vacuoles do not recruit hER. Left image: scale bar = 2 μm; Right image: scale bar = 1 μm.

hER recruitment is not required to induce the UPR

In order to determine whether recruitment of the hER to the PV was needed for induction of the UPR, we monitored activation of the UPR and downstream cytokine responses upon infection of BMDCs by HK T. gondii. Similar to live parasites, we found that Xbp1s and its downstream target Dnajb9 (coding for ERdj4) were induced in BMDCs incubated with HK parasites both at the mRNA and protein levels (Fig EV3A–C). However, in sharp contrast to live parasites, the expression of the UPR‐induced ER chaperones Hspa5 (encoding BIP), Hsp90b1 (encoding by GRP94) and Sec61a1 was up‐regulated as well as the PERK and ATF6 pathways (Fig EV3A). Indeed, HK parasites internalization induced the expression of the Atf4 gene and its target genes Ppp1r15a (GADD34) and Ddit3 (CHOP) as well as Atf6 (Fig EV3A). However, despite increased in Ddit3 mRNA levels, we still failed to visualize protein expression in HK parasite‐stimulated BMDCs (Fig EV3B). This result correlates with previous studies showing that a TRIF‐dependent pathway triggered upon TLR stimulation attenuates CHOP translation (Woo et al, 2009, 2012). In addition, similar to HK parasites, stimulation of BMDCs with an extract of soluble T. gondii antigens (STAgs)—which presumably only triggers TLRs—also resulted in the activation of the three branches of the UPR (Fig EV3D and E).

Figure EV3. The three pathways of the UPR response are activated in BMDCs incubated with Pru HK parasites and STAgs.

• A
  BMDCs were incubated with Pru heat‐killed (HK) parasites for 16 h, and the expression of UPR target genes was analyzed by RT‐qPCR. The results are expressed as relative mRNA level compared to non‐infected conditions. Results are normalized to the housekeeping gene Gapdh. Unpaired Student’s t‐test, *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; mean ± SEM (n = 2‐5 independent experiments depending of the studied gene, 3 biological replicates (mice) / experiment).
• B, C
  Immunoblot analysis of IRE1α, phospho‐eIF2a, CHOP, and XBP1s protein expression in non‐infected BMDCs, BMDCs treated with tunicamycin (TN), BMDCs incubated with Pru live parasites and BMDCs incubated with Pru HK parasites for the indicated times. GAPDH and tubulin were used as loading controls. The same blots are shown in Fig 1C, but the HK lanes have been removed in Fig 1C.
• D
  Immunoblot analysis of IRE1α and XBP1s in non‐infected BMDCs (NI), BMDCs infected with Pru parasites for 16 h and treated with soluble T. gondii antigen extracts (STAgs) for 16 h. Tubulin was used as loading control.
• E
  BMDCs were incubated with STAgs for 16 h, and the expression of UPR target genes was analyzed by RT‐qPCR. The results are expressed as relative mRNA level compared to non‐infected conditions. Results are normalized to the housekeeping gene gapdh. Unpaired Student’s t‐test, ns: P > 0.05; *P < 0.05; **P < 0.01; mean ± SEM (n = 4 mice).

Source data are available online for this figure.

Therefore, our results on HK parasite and STAg‐stimulated BMDCs support the hypothesis that triggering of the IRE1α/XBP1s pathway in infected BMDCs is not dependent on hER recruitment to the PV, but rather relies on TLR‐dependent signaling.

The presence of replicating parasites modulates the UPR

To further explore a potential modulation of the UPR by replicating parasites, we sorted infected versus non‐infected bystander BMDCs from a mixed population of infected cells, which included 45% of non‐infected cells (Fig 3A) and monitored UPR induction. At 16 h post‐infection, non‐infected bystander BMDCs displayed a significantly higher induction of Xbp1s and Dnajb9 expression compared to infected BMDCs (Fig 3B). These data therefore confirm that parasite invasion/replication is not absolutely required to activate the IRE1α/XBP1s pathway. Furthermore, they also show that the presence of live parasites correlates with a down‐regulation of XBP1s induction triggered by TLR stimulation, in agreement with the observed repression of Xbp1s monitored in infected MyD88 KO cells compared to non‐infected conditions (Fig 2D). Of note, the down‐regulation of Xbp1s, Dnajb9, and Ddit3 expression in MyD88 KO‐infected cells was not observed in cells infected with HK parasites (Fig EV4A), further strengthening the hypothesis that secretion of parasite effectors and/or triggering of other signaling pathways upon invasion by live parasites leads to a dampening of UPR host responses. The inhibitory effect seemed to be specific to Xbp1s and its downstream target genes since Ern1 (encoding IRE1α) expression was comparable in infected and bystander cells (Fig 3B), in line with what we monitored on WB (Fig 1C and D). Similarly to the modulation of the IRE1α branch, we also observed an increased expression of the PERK downstream target genes Atf4 and Ddit3 in bystander non‐infected BMDCs compared to infected BMDCs (Fig 3B). Finally, the down‐regulation of the ATF6 target gene Herpud‐1 was no longer observed in bystander‐infected cells (Fig 3B), suggesting an active down‐regulation of the ATF6 branch by parasite derived effector molecules (Fig 3B).

Figure 3. Replicating parasites modulate the UPR.

1. BMDCs (CD11c⁺) were incubated with Pru Tomato parasites for 16 h. Infected cells (red gating) and non‐infected bystander cells (green gating) were sorted‐purified and further analyzed.
2. The expression of UPR target genes was analyzed by RT‐qPCR in infected and non‐infected bystander cells compared to non‐infected BMDCs (NI). The results are expressed as relative mRNA level compared to non‐infected conditions. Results are normalized to the housekeeping gene Gapdh. Unpaired Student's t‐test, ns: P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; mean ± SEM (n = 6 mice pooled from 2 independent experiments).

Figure EV4. UPR and cytokine responses are under the control of MyD88 in BMDCs incubated with Heat‐Killed parasites.

1. mRNA relative expression level of Xbp1s, Dnajb9 (ERdj4), Ddit3 (CHOP), and Ppp1r15a (GADD34) from WT and MyD88KO BMDCs incubated with heat‐killed (HK) parasites for 16 h compared to non‐infected (NI) condition. Results are normalized to the housekeeping gene Gapdh. Two‐way ANOVA Tukey’s multiple comparisons, ns: P > 0.05; ****P < 0.0001; mean ± SEM (=4‐5 mice pooled from 2 independent experiments).
2. IL‐6, IL‐12, and IL‐23 cytokine levels from WT and MyD88 KO BMDCs non‐infected (NI), infected with Pru parasites or incubated with Pru HK parasites for 16 h. Two‐way ANOVA Tukey’s multiple comparisons, ****P < 0.0001; mean ± SEM (n = 4–6 mice pooled 2 independent experiments combined).

In conclusion, the presence of live parasites induces a down‐regulation of the UPR triggered by TLR‐mediated sensing of parasite compounds.

The IRE1α/XBP1s pathway regulates cytokine response in infected BMDCs

The UPR transcription factors XBP1s and CHOP have been previously demonstrated to regulate cytokine response in TLR‐stimulated macrophages (Martinon et al, 2010; Smith, 2018) and DCs (Márquez et al, 2017; Mogilenko et al, 2019). Therefore, we investigated whether T. gondii‐mediated XBP1s activation impacts on cytokine secretion in infected BMDCs.

To establish the function of XBP1s in cytokine production, we used Xbp1 ^flox/floxCre‐DC mice lacking XBP1 specifically in the CD11c‐positive cells (hereafter called XBP1ΔDC) (Osorio et al, 2014; Tavernier et al, 2017). Mice with Xbp1^flox/flox alleles but no expression of Cre (hereafter called XBP1^fl/fl) served as controls for XBP1∆DC mice (Osorio et al, 2014; Tavernier et al, 2017). We found that XBP1ΔDC and control BMDCs display a similar rate of infection (Fig 4A and B) and parasite replication (Fig 4C). However, deletion of XBP1 in T. gondii‐infected BMDCs correlated with a reduced gene expression and protein secretion of IL‐6 and IL‐12 but not of IL‐23 when compared to control cells (Fig 4D and E). Of note, the decrease in cytokine production was similar at the mRNA and protein levels strongly supporting that XBP1s regulate cytokine expression at the mRNA level. To investigate whether IRE1α also contributes to the regulation of cytokine secretion, we used XBP1ΔDC IRE1α^truncDC mice. These mice were obtained by crossing XBP1ΔDC mice with Ern1 ^fl/fl mice that bear loxP sites flanking exon 20 and 21 of the Ern1 gene (encoding IRE1α) (Tavernier et al, 2017). DCs from XBP1ΔDC IRE1^truncDC mice are deficient for XBP1 and harbor a low expressed truncated IRE1α isoform with impaired endonuclease activity. Of note, as previously described (Medel et al, 2018), GM‐CSF differentiated BMDCs display strongly reduced but not fully abrogated IRE1α expression (Fig EV5A). We did not measure further decrease in IL‐6, IL‐12, or IL‐23 secretion in XBP1ΔDC IRE1^truncDCs indicating that T. gondii‐induced XBP1s induction selectively contributes to cytokine production in infected BMDCs upon activation of IRE1α (Fig 4D and E). Importantly, the defect in cytokine production monitored in XBP1ΔDC and XBP1ΔIRE1^truncDC BMDCs was not a result of cell death induced upon infection or impaired differentiation as illustrated by a similar percentage of CD11c⁺/MHCII⁺ positive cells (Fig EV5B). Additionally, XBP1ΔDC and XBP1ΔIRE1^truncDC BMDCs displayed similar surface expression levels of the co‐stimulatory molecules CD80 and CD86 upon LPS stimulation (Fig EV5C) indicating that deletion of XBP1 and IRE1α did not impact on BMDCs activation upon immunogenic stimulation. Moreover, similar to BMDCs invaded by live parasites, XBP1ΔDC BMDCs incubated with HK parasites displayed a drastic inhibition in the gene expression and protein secretion of IL‐6 and IL‐12 (Fig EV5D and E), further supporting the hypothesis that the IRE1α/XBP1s pathway is triggered in a TLR‐dependent manner to regulate cytokine production. Of note, in contrast to IL‐6 and IL‐12, IL‐23 was weakly induced by HK parasites, suggesting that the presence of live parasites is required to trigger production of this cytokine (Fig EV6). However, the production of all cytokines induced by HK parasites was fully dependent on TLR/MyD88 signaling (Fig EV4B).

Figure 4. The IRE1α/XBP1s pathway modulates cytokine responses in infected BMDCs and cross‐presentation of soluble secreted parasite antigens.

1. Dotplots showing the percentage of CD11c⁺/MHCII^hi cells in Pru Tomato SAG1‐OVA parasite‐infected XBP1^fl/fl, XBP1ΔDC, XBP1^fl/flIRE1^fl/fl, and XBP1ΔDC IRE1^truncDC BMDCs (upper panels) and of Tomato‐positive cells (percentage of infected BMDCs: lower panels).
2. Mean percentage ± SEM of infection is indicated (n = 6 mice pooled from 2 independent experiments). Tukey's ANOVA one‐way test, ns: P > 0.05.
3. Mean percentage of parasitophorous vacuoles containing 1, 2, 4, or 8 parasites/vacuole (p/vac) in XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs. Tukey's ANOVA one‐way test, ns: P > 0.05. Mean values ± SEM. n = 3 independent experiments (biological replicates).
4. IL‐6, IL‐12, and IL‐23 secreted cytokine levels from Pru parasite‐infected XBP1^fl/fl, XBP1ΔDC, XBP1^fl/flIRE1^fl/fl, and XBP1ΔDC IRE1^truncDC BMDCs measured by ELISA. Two‐way ANOVA Tukey’s multiple comparisons, ns: P > 0.05; ***P < 0.001; ****P < 0.0001; mean ± SEM (n = 6 mice pooled from 2 independent experiments).
5. mRNA relative expression level of Il‐6, Il‐12, and Il‐23 from XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC and XBP1ΔDC BMDCs infected with Pru parasites for 16 h compared to non‐infected (NI) conditions. Results are normalized to the housekeeping gene Gapdh. Two‐way ANOVA Tukey’s multiple comparisons, ns: P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; mean ± SEM (n = 3–5 mice pooled from 2 independent experiments).
6. H‐2K^b‐SKL8 MHC class I presentation by XBP1^fl/flIRE1^fl/fl treated or not with 4µ8C, XBP1ΔDC IRE1^truncDC and XBP1ΔDC BMDCs infected with Pru Tomato SAG1‐OVA at the indicated ratios, assessed with the B3Z T‐cell hybridoma. Unpaired Student's t‐test, *P < 0.05; **P < 0.01; ***P < 0.001; mean ± SEM (n = 3 mice, one representative experiment from 2 independent experiments).
7. Percentage of Pru Tomato SAG1‐OVA infected BMDCs at the indicated ratios of parasites:BMDC. Bar graphs depict mean ± SEM (n = 3 mice, one representative experiment from 2 independent experiments). Tukey's ANOVA one‐way test, ns (P > 0.05) when p values are not identicated.
8. Exogenous MHCI presentation of synthetic SKL8 peptide pulsed on XBP1^fl/flIRE1^fl/fl treated or not with 4µ8C, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs assessed with the B3Z T‐cell hybridoma. Bar graphs depict mean ± SEM (n = 3 mice, one representative experiment from 2 independent experiments). Unpaired Student's t‐test, *P < 0.05, ns (P > 0.05) when P values are not identicated.
9. Graph depicts the expression (indicated as the mean fluorescence intensity) of H‐2K^b by Pru Tomato SAG1‐OVA infected XBP1^fl/fl IRE1^fl/fl treated or not with 4μ8C, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs. Bar graphs depict mean ± SEM (n = 3 mice, one representative experiment from 2 independent experiments). Tukey's ANOVA one‐way test, ns: P > 0.05.

Figure EV5. IRE1α/XBP1s depletion does not impact on BMDC differentiation and activation but alters cytokine production by HK parasite‐stimulated BMDCs.

1. Immunoblot analysis of IRE1α and XBP1s protein expression in XBP1^fl/flIRE1^fl/fl, XBP1ΔDC, and XBP1ΔDC IRE1^truncDC BMDCs treated or not with tunicamycin (TN) to induce ER stress. Please note that there is never 100% excision of the floxed alleles in GM‐CSF BMDCs cultures, as can be noted from the strongly reduced but not fully abrogated expression of IRE1α and truncated IRE1α remaining visible in the XBP1ΔDC IRE1^truncDC BMDCs line.
2. Dotplots showing the percentage of live cells (live/dead marker, upper panels) and of cells expressing CD11c and MHCII molecules (lower panels) in Pru‐infected XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs. Graph depicts the mean percentage of live cells and CD11c⁺ MHCII ^high cells in XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs non‐infected (NI), infected with Pru parasites, incubated with Pru HK parasites or treated with LPS for 16 h. Bar graphs depict mean ± SEM (n = 3 mice, one representative experiment from 2 independent experiments). Tukey's ANOVA one‐way test, ns: P > 0.05.
3. Staining profiles of LPS‐treated XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs indicating the surface expression of CD80 and CD86 molecules. Graph depicts the quantification (indicated as the mean fluorescence intensity) of CD80 and CD86 expression in LPS‐treated XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs. Bar graphs depict mean ± SEM (n = 3 mice, one representative experiment from 2 independent experiments). Tukey's ANOVA one‐way test, ns: P > 0.05.
4. IL‐6, IL‐12, and IL‐23 cytokine levels from XBP1^fl/fl, XBP1ΔDC, XBP1^fl/flIRE1^fl/fl, and XBP1ΔDC IRE1^truncDC BMDCs infected with Pru parasites or incubated with Pru heat‐killed (HK) parasites for 16 h compared to non‐infected (NI) BMDCs. Two‐way ANOVA Tukey’s multiple comparisons test, ns: P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001; mean ± SEM (n = 3–6 mice pooled from 2 independent experiments).
5. mRNA relative expression of Il‐6, Il‐12, and Il‐23 from XBP1^fl/flIRE1^fl/fl, XBP1ΔDC IRE1^truncDC, and XBP1ΔDC BMDCs infected with Pru parasites or incubated with Pru heat‐killed (HK) parasites for 16 h compared to non‐infected (NI) BMDCs. Results are normalized to the housekeeping gene Gapdh. Two‐way ANOVA Tukey’s multiple comparisons test, ns P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001; mean ± SEM (n = 3–6 mice pooled from 2 independent experiments).

Collectively, these data demonstrate that upon T. gondii infection of BMDCs, IL‐6 and IL‐12 secretion is partially under the control of IRE1α/XBP1s signaling activated via MyD88. In contrast, IL‐23 production is IRE1α/XBP1s‐independent, requires live parasite invasion, and is only partially dependent on MyD88 signaling.

The IRE1α/XBP1s pathway regulates cross‐presentation of soluble secreted parasite antigens

Several studies reported the implication of the IRE1α branch of the UPR in regulating MHCI antigen presentation in DCs both at steady‐state and in activated conditions (Osorio et al, 2018). Activation of the IRE1α/XBP1s axis was found to have beneficial or detrimental outcomes on T‐cell activation depending on the nature of the stimulation and the antigen. Notably, in the context of tumors, reinforcing XBP1s expression ameliorates cross‐presentation and priming of CD8⁺ T cells (Zhang et al, 2015). Similarly, the inhibition of IRE1α endoribonuclease activity in cDC1 leads to reduced soluble melanoma antigen cross‐presentation to CD8⁺ T cells but has no impact on MHCII antigen presentation (Zhang et al, 2015). So far, the implication of the IRE1α/XBP1s pathway in MHCI antigen presentation has not been addressed in the context of DCs infected by intracellular pathogens. To address this question, we infected BMDCs with type II parasites expressing the soluble model antigen OVA, which is secreted in the vacuolar space of the PV. MHCI presentation of this antigen has been previously demonstrated to depend on hER component recruitment at the PV, notably TAP and the translocon Sec61 (Gubbels et al, 2005; Goldszmid et al, 2009; Cebrian et al, 2011). We used LacZ‐inducible reporter CD8 T‐cell hybridomas (B3Z) that specifically respond to H‐2K^b‐SIINFEKL complexes to assess secreted OVA presentation by control XBP1^fl/fl IRE1^fl/fl, XBP1ΔDC, and XBP1ΔDC IRE1α^truncDC BMDCs. Since XBP1ΔDC IRE1α^truncDC may display residual IRE1α activity (Fig EV5A), we additionally treated control cells with 4μ8C, a selective and potent inhibitor of the IRE1α endoribonuclease domain as illustrated by the complete inhibition of Xbp1s induction and its target Dnajb9 upon infection or treatment with tunicamycin (Appendix Fig S3A and Appendix Fig S3B). 4μ8C treatment also inhibited IL‐6 and IL‐12 secretion (but not IL‐23 production) induced upon infection (Appendix Fig S3C) similarly to what we observed in XBP1ΔDC and XBP1ΔDC IRE1α^truncDC BMDCs (Fig 4D).

We found that inhibition of IRE1α by 4μ8C results in a pronounced decrease in SIINFEKL (SKL8) peptide presentation by infected cells (Fig 4F). A significant inhibition of SKL8 peptide presentation was also observed in XBP1ΔDC IRE1α^truncDC BMDCs (Fig 4F), although less pronounced, which may reflect the partial deletion of IRE1α in these cells. Interestingly, XBP1ΔDC‐infected BMDCs were not inhibited in SKL8 peptide presentation demonstrating that this process is not dependent on XBP1s induction (Fig 4F). The defect in SKL8 peptide presentation observed in XBP1ΔDC IRE1α^truncDC was not due to altered parasite infection (Fig 4G) or to reduced MHCI surface expression (Fig 4H and I), also supported by similar exogenous presentation of the SKL8 peptide in all studied conditions (Fig 4H). In conclusion, our data indicate that IRE1α but not XBP1s contributes to CD8⁺ T‐cell priming by T. gondii secreted antigens in BMDCs.

The IRE1α pathway is stimulated in splenic cDC1 during infection

To assess whether the IRE1α pathway is activated during T. gondii infection of mice, we used ERAI reporter mice (Iwawaki et al, 2004). These mice express a partial sequence of human XBP1 fused to Venus fluorescent protein (VenusFP) that includes the sites at which IRE1α splices XBP1 (Iwawaki et al, 2004). Thus, detection of the VenusFP signal in cells allows reporting of IRE1α activation. Our previous findings demonstrated a preferential induction of the IRE1α arm of the UPR in basal‐state splenic cDC1 over cDC2 and pDC (Osorio et al, 2014; Tavernier et al, 2017). ERAI and control mice were infected with a low dose of Pru parasites, and the different subpopulations of splenic DCs were analyzed 6 days post‐infection by flow cytometry.

In line with our previous findings, a VenusFP signal was detected at steady state in cDC1s (CD26⁺/CD11c^hi/MHCII^hi/XCR1⁺/CD172a⁻) but not in cDC2s (CD26⁺/CD11c^hi/MHCII^hi/XCR1⁻/CD172a⁺) (Fig 5A–C). Upon infection, the VenusFP signal greatly increased in the cDC1 subset while it remained unchanged in the cDC2 population (Fig 5A–C) indicating that IRE1α activation and subsequent XBP1s expression is restricted to cDC1s. We also did not monitor substantial XBP1s induction in splenic neutrophils, monocytes, B, and T cells suggesting that the UPR contributes to the regulation of cDC1 functions upon T. gondii infection (Fig 5D and S4A). To further confirm these results, we sorted splenic cDC1s and cDC2s from 6 days WT‐infected mice and examined UPR induction by quantifying the mRNA levels of different UPR targets. Consistent with the results obtained in ERAI mice, a strong up‐regulation of Ern1 (IRE1α) and Xbp1s transcripts were monitored only in cDC1s but not in cDC2s. The expression of the PERK target Ddit3 (CHOP) was not induced in none of the cDC sub‐types (Fig 5E and F). Importantly, similar to observations made in previous studies (John et al, 2009), the percentage of dTomato‐positive infected cells was very low in the spleen of ERAI infected mice suggesting that parasite invasion and replication may not be required to trigger the massive activation of the IRE1α pathway monitored in cDC1s (Appendix Fig S4B). This hypothesis correlates with our in vitro findings indicating that induction of XBP1s in BMDCs is dependent on innate sensing of particulate antigens and does not require parasite invasion.

Figure 5. The IRE1α/XBP1s pathway is activated in splenic cDC1 of mice infected with T. gondii .

• A, B
  Expression of VenusFP by splenic cDC1 (CD26⁺CD11c⁺ MHCII⁺XCR1⁺ CD172a⁻, upper panels) and cDC2 (CD26⁺CD11c⁺ MHCII⁺XCR1⁻ CD172a⁺, lower panel) from WT (A) and ERAI (B) mice non‐infected (NI) or infected with Pru (6 days post‐infection (dpi)) reporting on IRE1α pathway activation.
• C
  Graph depicts the mean fluorescence intensity measured for each condition in ERAI mice (6 dpi). Two‐way ANOVA Tukey’s multiple comparisons, ns: P > 0.05; ***P < 0.001; ****P < 0.0001; mean ± SEM (n = 5 mice per genotype and condition, one representative experiment from 2 independent experiments is shown).
• D
  Graph depicts the expression (indicated as the mean fluorescence intensity) of VenusFP by splenic monocytes, neutrophils, B, and T cells from Pru‐infected or non‐infected (NI) ERAI mice (6 dpi). Bar graphs depict mean ± S.E.M (n = 5 mice per genotype and condition, one representative experiment from 2 independent experiments is shown). Unpaired Student's t‐test, ns: P > 0.05.
• E, F
  RT‐qPCR analysis of the indicated UPR genes in splenic cDC1 (E) and cDC2 (F) sorted‐purified 6 dpi from WT mice, non‐infected (NI) or infected with Pru parasites. Results were normalized to the housekeeping gene Gapdh and are expressed as relative mRNA level compared to the NI condition. Bar graphs depict mean ± SEM Unpaired Student's t‐test, ns: P > 0.05; *P < 0.05; **P < 0.01; mean ± SEM (n = 3–5 mice pooled from 2 independent experiments).

IRE1α/XBP1s pathway activation plays an essential protective role against T. gondii infection

To determine the contribution of DC‐specific activation of IRE1α in the resistance to T. gondii infection, we assessed the survival rate of control XBP1^fl/fl IRE1α^fl/fl mice compared to XBP1ΔDC IRE1α^truncDC mice injected intraperitoneally with Pru parasites. Our data showed that XBP1ΔDC IRE1α^truncDC mice were highly susceptible to infection when compared to control mice and all succumbed between 9 and 12 days post‐infection (Fig 6A). This result indicates that activation of the IRE1α pathway in DCs plays an essential protective role against T. gondii acute infection. To gain insights into the processes leading to XBP1ΔDC IRE1α^truncDC mice early mortality, we isolated splenocytes and cells from the peritoneal exudate (PEC) of infected mice and analyzed parasite load by qPCR. We found a significant increase in parasite burden in both organs in XBP1ΔDC IRE1α^truncDC mice compared to control mice, indicating an altered parasite control (Fig 6B). We did not detect any significant changes in Il‐12 expression in the spleen of infected XBP1ΔDC IRE1α^truncDC mice, presumably due to the contribution of other splenic cell types, notably cDC2s and monocytes, in the secretion of this cytokine (Fig 6C). However, a reduced Ifn‐g and Tnfα expression was monitored in XBP1ΔDC IRE1α^truncDC mice compared to control mice suggesting a defect in T‐cell activation that could be caused by a cDC functional alteration in antigen presentation (John et al, 2009; Lee et al, 2015). Accordingly, while in control mice, CD4 T and CD8 T cell showed increased proliferation upon infection, this expansion was not observed in XBP1ΔDC IRE1α^truncDC mice (Fig 6D and E). We also observed a defect in CD8 T‐cell activation as noted by the decreased number of IFNγ‐expressing CD8 T cells but not of IFNγ‐expressing CD4 T cells, suggesting an alteration of cross‐presentation by cDC1s (Fig 6D and E). Finally, we noticed that the number and frequency of splenic cDC1s and cDC2s both increased upon infection of WT mice (Fig 6F). On the contrary, XBP1ΔDC IRE1α^truncDC mice, which exhibit a similar number of the two cDC subsets at steady‐state compared to control mice, do not show increased cDC1 number and frequency upon infection (Fig 6F). This defect was not observed for cDC2s, consistent with a specific activation of the IRE1α pathway in cDC1s (Fig 5A and B) and a defect in CD8 T‐cell activation (Fig 6E). These data therefore suggest that the IRE1α pathway may not only regulate cross‐presentation to CD8 T cells but also promote the expansion of the cDC1 subset triggered upon infection.

Figure 6. The IRE1α pathway protects mice from acute infection and promotes T‐cell responses.

• A
  Survival curves of XBP1^fl/flIRE1^fl/fl and XBP1ΔDC IRE1^truncDC mice infected intraperitoneally with 500 Pru parasites. Log‐rank (Mantel–Cox) test, P = 0.0004; n = 10 mice per genotype.
• B
  Number of parasite genomes per spleen and PEC from XBP1^fl/flIRE1^fl/fl and XBP1ΔDC IRE1^truncDC infected intraperitoneally with 500 Pru parasites 6 days post‐infection. Unpaired Student's t‐test **P < 0.01; mean ± SEM (n = 2 independent experiments combined).
• C
  mRNA relative expression of Il‐6, Il‐12, Ifn‐g, and Tnf‐a by splenocytes from XBP1^fl/flIRE1^fl/fl and XBP1ΔDC IRE1^truncDC mice infected or not (NI) with 500 Pru parasites for 6 days. Results are normalized to the housekeeping gene Gapdh. Two‐way ANOVA Tukey’s multiple comparisons, ns: P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001; mean ± SEM (n = 2 independent experiments combined).
• D, E
  Quantification of the number of total (left histogram) and activated (right histogram) (IFNγ‐positive) CD4⁺ T cells (D) and CD8⁺ T cells (E) in the spleen of control mice (XBP1^fl/flIRE1^fl/fl) and XBP1ΔDC IRE1^truncDC infected for 6 days. Two‐way ANOVA Tukey's multiple compare, ns: P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; mean ± SEM (n = 2 independent experiments combined).
• F
  Frequency and number of cDC2 (left) and cDC1 (right) in the spleen of control mice and XBP1ΔDC IRE1^truncDC infected for 6 days. Two‐way ANOVA Tukey's multiple compare, ns: P > 0.05; ****P < 0.0001; mean ± SEM (n = 2 independent experiments combined).

Discussion

Very few studies addressed the impact of UPR signaling in DCs in response to infection by intracellular pathogens. Our results demonstrate that infection of BMDCs by T. gondii specifically activated the IRE1α/XBP1s branch, while the PERK pathway was not activated and the ATF6 branch even repressed. Experiments comparing infected versus non‐infected bystander BMDCs or heat‐killed versus live parasites revealed that triggering of the UPR does not require active invasion with live parasites, but relies on TLR‐mediated sensing of parasite compounds (Fig 7A). On the other hand, secretion of parasite effectors and/or triggering of other signaling pathways upon invasion by live parasites lead to a dampening of the UPR, which is especially prominent for the ATF6 and PERK branch (Fig 7B).

Figure 7. Modulation of the UPR and downstream responses in T. gondii‐infected BMDCs.

1. After incubation of BMDCs with heat‐killed parasites or STAg, TLRs are stimulated and induce transcriptional activation of the three pathways of the UPR. The IRE1α/XBP1s pathway stimulates the secretion of IL‐6 and IL‐12.
2. In BMDCs infected with live Pru parasites for 16 h, the IRE1α/XBP1s pathway is activated in a MyD88‐dependent manner, while the PERK pathway is not induced and the ATF6 pathway is repressed. XBP1s stimulates the secretion of IL‐6 and IL‐12 and IRE1α promotes MHCI presentation of secreted parasite antigens. The secretion of IL‐23 is independent of XBP1s and CHOP and only partially depends on MyD88. Our data also indicate that live parasites down‐regulate the induction of XBP1s and CHOP induced by TLRs. Blue arrows indicate pathways putatively induced or down‐regulated by parasite effectors.

Manipulation of the UPR has been well documented during viral infections (Smith, 2018). For instance, the chikungunya virus shuts‐down the PERK pathway to sustain eIF2α‐mediated translation of viral proteins and to avoid CHOP‐induced apoptosis of the host cell, but activates the IRE1α and ATF6 pathways to enhance ER capacity (Rathore et al, 2013). Similarly, we found that the presence of live parasites led to a down‐regulation of the UPR triggered by TLR stimulation. Along these lines, it has been reported that TgROP18, a parasite effector anchored at the limiting membrane of the vacuole phosphorylates ATF6β, leading to its degradation via the proteasome (Yamamoto et al, 2011). Our data indicate that the ATF6 pathway is also repressed at the transcriptional level in infected BMDCs by yet unknown mechanisms. Also, ATF4 and CHOP, downstream mediators of the PERK pathway, were not induced in infected BMDCs. TLR stimulation has been previously shown to correlate with a TRIF‐dependent attenuation of CHOP translation as we also noted in HK or STAg‐stimulated parasites (Woo et al, 2009, 2012). However, infection with live parasites led to repression of CHOP already at the transcriptional level, which became apparent upon comparison of WT versus MyD88 KO cells. This process may allow the infected host cell to avoid CHOP‐induced apoptosis.

The mechanisms by which infection by live parasites modulates the UPR remain to be elucidated. To escape cellular defenses, T. gondii secretes numerous parasite effectors that modulate host cell signaling pathways (Hakimi et al, 2017). In addition, parasite effectors anchored at the limiting membrane of the vacuole promote recruitment of host organelles, most likely to stimulate host lipid uptake required to endure the successive divisions of this fast replicating parasite. T. gondii stimulates the recruitment of mitochondria (Pernas et al, 2014, 2018), fragmented Golgi (Deffieu et al, 2019), endosomes (Romano et al, 2017), and the hER (Sinai et al, 1997; Goldszmid et al, 2009). It is likely that sensing mechanisms detect changes in organelle structure and functions, which result in the stimulation of specific downstream responses. In support of this hypothesis, the transcriptional response of cells invaded by TgMAF‐1‐deficient parasites, which are unable to recruit mitochondria at the PV, revealed major changes in cytokine responses (Pernas et al, 2014). By contrast to mitochondria and Golgi vesicles, the effectors that stimulate hER association with the PVM are not identified. A previous work reported the interaction of TgGRA3 with calcium modulating ligand (CAMLG), a type II transmembrane protein of the ER, but demonstration of a functional role of TgGRA3 in this process is lacking (Kim et al, 2008). As previously observed by others, we detected hER association with the PV in fibroblasts and BMDCs infected with type II Pru parasites but this process was not required to induce the UPR. However, hER recruitment by the parasite may contribute to the modulation of the UPR pathways induced by TLR stimulation. Thus, identification of the parasite effectors contributing to hER recruitment at the PV would allow exploring its impact on the regulation of hER functions, the possible modulation of the UPR and downstream innate immune responses. In general, further characterization of the induced UPR pathways in DCs infected with parasites deficient in the secretion of parasite effectors (Franco et al, 2016; Naor et al, 2018; Panas et al, 2019) will also help to identify the mechanisms by which T. gondii modulates the UPR.

Also in vivo, infection with live parasites appeared not essential to trigger the UPR in DCs, as non‐infected bystander rather than infected DCs showed activation of the IRE1α branch. Indeed, while activation of IRE1α in splenic cDC1 was high at day 6 post‐infection, the level of infected cells at that moment was very low, although parasite DNA was still clearly detected by qPCR. This is consistent with the study of John et al suggesting that cross‐presentation by bystander DCs rather than infected DCs is an important route of antigen presentation during toxoplasmosis (John et al, 2009). Along these lines, it has been demonstrated that during influenza virus infection, efficient cross‐presenting cDC1 are non‐infected due to their ability to efficiently restrict virus replication by an IFNβ‐mediated induction of cellular defenses (Helft et al, 2012). Thus, also for T. gondii infection, detection of infected cells by flow cytometry may not clearly distinguish the cells that have internalized particulate antigens from the ones that have efficiently restricted parasite replication. Activation of the IRE1α branch was observed specifically in splenic cDC1s but not in cDC2s upon T. gondii infection. This result may correlate with the predominant expression of TLR11/12 in cDC1 over cDC2 (Dalod et al, 2014; Heng et al, 2008) and extends earlier findings in steady state showing preferential activation of the IRE1 branch in cDC1s (Osorio et al, 2014; Tavernier et al, 2017).

Activation of the IRE1α pathway upon T. gondii infection regulates key DC functions required for optimal host defenses (Fig 7A and B). Induction of XBP1s promoted pro‐inflammatory cytokine production at the mRNA level in a cytokine specific manner. IL‐6 and IL‐12p40 appeared XBP1‐dependent, while IL‐23p19 was not. These data are in line with earlier reports revealing that macrophage stimulation by TLR2 and TLR4 agonists activates the IRE1α pathway via a TRAF6‐NOX2 axis (Martinon et al, 2010). IRE1α‐mediated activation of XBP1s promoted sustained production of inflammatory mediators and XBP1 deficiency resulted in an increased bacterial burden in mice infected with the TLR2‐activating intracellular pathogen Francisella tularensis (Martinon et al, 2010). In our hands, the absence of MyD88 led to a complete abrogation of T. gondii‐induced XBP1 activation. Still, it remains possible that on top, alternative sensing mechanisms are involved. Along these lines, a recent study demonstrated that T. gondii infection activates IRE1α through calcium release from the ER. As described previously (Urra et al, 2018), IRE1α oligomerization and recruitment of filamin A promoted migration of T. gondii‐infected MEF and BMDCs (Augusto et al, 2020). While this study did not investigate activation of additional UPR branches in T. gondii‐infected BMDCs, these findings are consistent with our results showing a sustained activation of IRE1α in infected BMDCs (Augusto et al, 2020).

Besides stimulating pro‐inflammatory cytokine production, we found that IRE1α, but not XBP1s, was required for optimal MHCI antigen presentation of secreted T. gondi antigens to CD8 T cells. We previously demonstrated that IRE1α‐mediated RIDD activity reduces MHCI antigen presentation of dead cell‐derived antigens by degrading key factors of the cross‐presentation machinery (Osorio et al, 2014). In contrast to these results, upon infection, we found a role for IRE1α in promoting MHCI antigen presentation. The mechanisms involved are not yet identified. It is possible that IRE1α regulates trafficking events impacting on vacuole fate, parasite antigen escape to the cytosol or on antigen processing. IRE1α induction may thereby facilitate parasite control by stimulating CD8 T‐cell priming, a process that favors preservation of the host and transmission of the parasites after establishment of a chronic infection. Consistent with this hypothesis, we found that mice lacking XBP1 and IRE1α failed to control the infection and succumbed during the acute phase. This high susceptibility was correlated with a defect in CD4 and CD8 T‐cell expansion as well as altered CD8 T‐cell activation in the spleen of infected mice. In line with the altered CD8 T‐cell proliferation and activation, IRE1α/XBP1‐deficient cells exhibit an impaired expansion of cDC1s but not cDC2s upon infection. The defect in CD4 T‐cell proliferation likely results from a role of cDC1s in assisting CD8⁺ T‐cell responses through distinct mechanisms, including a process whereby CD4⁺ T cells “license” cDC1 for CD8⁺ T‐cell priming via CD40 signaling (Smith et al, 2004). While our study revealed that the IRE1α pathway is essential to protect mice against an acute T. gondii infection, it is only a first step toward understanding the likely very complex process of IRE1α‐mediated regulation of innate and adaptive immunity against T. gondii. Future studies will aim to further characterize the mechanisms by which activation of IRE1α and XBP1s regulates cDC1 functions upon infection, including cell expansion and MHCI antigen presentation. It will be important to also explore the role of the IRE1α branch in regulating non‐lymphoid tissue cDC responses, notably peripheral migratory CD103⁺‐cDC1 after oral ingestion of cysts. For instance, XBP1s‐mediated regulation of IL‐12 production may be important at early stages of parasite invasion in the intestine to activate IFNγ secretion by NK cells. Thus, our study opens perspectives for the identification of novel strategies to boost immunity against T. gondii, notably CD8 T‐cell responses, by manipulating the UPR and thereby to restrict parasite spreading to the brain and establishment of cerebral chronic infection.

Materials and Methods

Mice

Wild‐type C57BL/6J (B6) mice were purchased from Janvier (France). C57BL/6 XBP1^fl/fl (Lee et al, 2008), C57BL/6 XBP1ΔDC (Osorio et al, 2014), C57BL/6 XBP1^fl/fl IRE1^fl/fl (Iwawaki et al, 2010), C57BL/6 XBP1ΔDC IRE1^truncDC (Tavernier et al, 2017), and ERAI mice (Iwawaki et al, 2004) were provided by S. Janssens (IRC/VIB, Ghent University). C57BL/6 XBP1^fl/fl IRE1^fl/fl and C57BL/6 XBP1ΔDC IRE1^truncDC mice were bred at the « Institut Pasteur de Lille » in pathogen‐free conditions. Experiments were carried out in accordance with the European regulations (86/609/EEC). All animal procedures were approved by the local Ethical Committee for Animal Experimentation registered by the “Comité National de Réflexion Éthique sur l’Expérimentation Animale” (CEEA‐75). All mice were housed and bred at the “Institut Pasteur de Lille” according to the protocol (APAFIS#12804‐2017122111233479v2) approved by the French Minister of Higher Education, Research and Innovation.

BMDC differentiation and parasite culture

Wild‐type C57BL/6, C57BL/6 XBP1^fl/fl, C57BL/6 XBP1ΔDC, C57BL/6 XBP1^fl/fl IRE1^fl/fl, C57BL/6 XBP1ΔDC IRE1^truncDC, and C57BL/6 MyD88^−/− (B. Lambrecht, IRC/VIB, Ghent University) mice from 8 to 10 weeks of age were used to obtain bone marrow dendritic cells (BMDCs) from the femur and tibia. BM cells were differentiated for 10 days in Petri dish with RPMI (Gibco™ by Life Technologies) supplemented with 10% fetal bovine serum (FBS, Gibco™ by Life Technologies), 1% penicillin–streptomycin (Gibco™ by Life Technologies), 1% sodium pyruvate (Gibco™ by Life Technologies), 0.2% 2‐mercaptoethanol (Gibco™ by Life Technologies), and 10% granulocyte‐macrophage colony‐stimulating factor. Flt3/Notch‐differentiated DC were generated as described in (Kirkling et al, 2018). Flt3 ligand was generated by the protein core facility of VIB Center for Inflammation Research (Ghent, Belgium).

Tachyzoites of Toxoplasma gondii Type I RH Luc GFP (J. Boothroyd; Stanford University of School Medicine), Type II Pru Luc GFP (J. Boothroyd; Stanford University of School Medicine), and Type II Pru Tomato SAG1‐Ova (Schaeffer et al, 2009) parasites were grown in vitro on confluent Human Foreskin Fibroblast (HFF) cells (CCD‐1112Sk (ATCC, CRL‐2429^TM)) using Dulbecco’s modified Eagles medium (Gibco™ by Life Technologies) supplemented with 10% fetal calf serum (Gibco™ by Life Technologies), and 1% penicillin–streptomycin (Gibco™ by Life Technologies). Prior to experiments, parasites were extracted from host cells by sequential passages through 17‐gauge and 26‐gauge needles followed by filtration with a 3‐µm polycarbonate membrane filter. Parasite cultures were regularly tested for mycoplasma using MycoAlert™ mycoplasma detection kit (Lonza). If not indicated otherwise, BMDCs were infected at a multiplicity of infection (MOI) of 4 parasites/BMDC. Heat‐Killed (HK) parasites were obtained after incubation of live parasites for 20 min at 56°C. Soluble T. gondii antigens (STAgs) were obtained after lysis of purified parasites by sonication in ice‐cold sterile water followed by centrifugation at 11,000 g for 30 min to remove cell debris. The solution was then adjusted to 1X PBS and filtered. Protein concentration was determined using the BCA kit. BMDCs were incubated for 16 h with STAgs at a final concentration of 10 µg/ml, with tunicamycin (TN) (SIGMA‐ALDRICH^®) at a final concentration of 1 μg/ml for 16 h, with lipopolysaccharide (LPS) (InvivoGen) at a final concentration of 100 ng/ml for 16 h and with the IRE1 inhibitor, 4µ8C (SIGMA‐ALDRICH^®) at a final concentration of 10 µM for 16 h. For in vivo infection, 500 parasites were injected intraperitoneally.

Immunofluorescence assays (IFA)

Adherent BMDCs on glass coverslips precoated with poly‐L‐lysine were infected with parasites at a MOI of 1 during 16 h before fixation with 1% paraformaldehyde for 20 min. After quenching with 50 mM NH₄Cl for 15min, the cells were permeabilized with 0.1% Triton for 3 min and incubated with mouse anti‐KDEL (10C3 Enzo^® Life Sciences) antibodies in 0.1% BSA‐PBS, followed by DAPI and goat anti‐mouse secondary antibodies conjugated to Alexa Fluor 594. Images were acquired using a Zeiss LSM880 confocal microscope.

Electron microscopy

BMDCs infected with T. gondii parasites at the MOI of 2 were fixed in 4% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M NaCacodylate buffer, pH 7.2, and centrifuged at 400 g. Low melting point agarose was used to keep the cells concentrated for further processing. Cells were fixed for 4 h at room temperature followed by fixation O/N at 4°C after replacing with fresh fixative. After washing in buffer, they were post‐fixed in 1% OsO₄ with 1.5% K3Fe(CN)6 in 0.1 M NaCacodylate buffer at room temperature for 1 h. After washing in ddH₂O, cells were subsequently dehydrated through a graded ethanol series, including a bulk staining with 1% uranyl acetate at the 50% ethanol step followed by embedding in Spurr’s resin. Ultrathin sections of a gold interference color were cut using an ultra‐microtome (Leica EM UC7), followed by a post‐staining in a Leica EM AC20 for 40 min in uranyl acetate at 20°C and for 10 min in lead stain at 20°C. Sections were collected on Formvar‐coated copper slot grids. Grids were viewed with a JEM 1400plus transmission electron microscope (JEOL, Tokyo, Japan) operating at 80 kV.

Flow cytometry

After FcBlock (BD Pharmingen^TM) incubation, cell suspensions were labeled with a dead cell marker (Aqua Live Dead Molecular Probes^® by Life Technologies^TM), anti‐CD11c APC (N418, BioLegend^®), anti‐I‐A/I‐E (MHCII) FITC (M5/114.15.2, BioLegend^®) or anti‐CD80 FITC (16‐10A1, BioLegend^®), or anti‐CD86 FITC (GL‐1, BioLegend^®) or anti‐H2K^b FITC (AF6‐88.5.5.3, BioLegend^®) in PBS. After cells fixation (Fixation buffer BioLegend^®), samples were analyzed using the Attune NxT^TM flow cytometer (Thermo Fisher SCIENTIFIC) and analyzed using the FlowJo10 software.

Isolation of splenocytes

Spleens were isolated from mice and digested with PBS‐2% FBS containing 1 mg/ml of collagenase (SIGMA‐ALDRICH^®) and 2 µg/ml of DNase I (Roche) for 20 min at 37°C followed by lysis of red blood cells (Red Blood Cell Lysing Buffer Hybri‐Max^TM, SIGMA‐ALDRICH^®). The cells were then used to perform RT‐qPCR, cell labeling for flow cytometry or cell sorting.

Splenic DC and T‐cell analysis

After isolation, splenocytes were incubated with FcBlock (BD Pharmingen^TM). For DC analysis, splenocytes were labeled with a dead cell marker (Zombie Violet^TM Fixable Viability, BioLegend^®), anti‐CD45 APC‐Cy7 (30‐F11, BioLegend^®), anti‐CD3 PE‐Cy7 (17A2, BioLegend^®), anti‐CD11c APC (N418, BioLegend^®), anti‐I‐A/I‐E FITC (M5/114.15.2, BioLegend^®), and anti‐CD8α BV605 (53‐6.7, BioLegend^®) antibodies in PBS. CD8α^−/+ DCs were isolated by cell sorting using the BD FACSARIA^TM III with gating on forward scatter and side scatter/singlets/live cells/CD45⁺CD3⁻ CD11c⁺MHCII^hi and CD8α⁺ or CD8α⁻. For splenic T‐cell analysis, splenocytes were labeled with a dead cell marker (Zombie Violet^TM Fixable Viability, BioLegend^®), anti‐CD45 APC‐Cy7 (30‐F11, BioLegend^®), anti‐CD3 PE‐Cy7 (17A2, BioLegend^®), anti‐CD4 AF700 (GK1.5, BioLegend^®), and anti‐CD8 BV605 (53‐6.7, BioLegend^®) antibodies. The intracellular labeling of IFN‐γ was performed according to the manufacturer’s instructions using anti‐IFN‐γ PE antibodies (XMG1.2, BioLegend^®). Samples were analyzed using the Attune NxT^TM flow cytometer (Thermo Fisher SCIENTIFIC) and the FlowJo10 software.

Quantitative real‐time PCR (RT‐qPCR)

Non‐infected or infected BMDCs RNA was obtained using the RNeasy^® Mini kit (QIAGEN) and treated with DNase I (SIGMA‐ALDRICH^®). When specified, infected or bystander BMDCs from a mixed population were isolated by cell sorting with gating on forward scatter and side scatter/singlets/live cells/CD11c + MHCII⁺/±Pru Tomato Ova using the BD FACSARIA^TM III. Total cDNA was generated using the High‐Capacity cDNA Reverse Transcription kit (Applied Biosystems™ by life technologies) according to manufacturer’s instructions. RT‐qPCR was carried out on a QuantStudio 12K Flex Real‐Time PCR system (Applied Biosystems™ by life technologies) with SYBR^® Selected Master Mix (Life technologies) and primers (listed in Appendix Table S1). The expression of individual genes was normalized to the housekeeping gene GAPDH.

The RNA of CD8α⁺ DCs was isolated using the RNeasy^® Micro kit (QIAGEN) and treated with a DNase I. cDNA was generated using the SuperScript kit^TM VILO^TM cDNA Synthesis kit (Invitrogen by Thermo Fischer Scientific), and qPCR were performed as previously described.

Western blot

Non‐infected or infected BMDCs were incubated with lysis buffer (250mM NaCl, 20 mM Hepes, 1 mM EDTA, 1% NP‐40, 20 mM DTT, phosphatase inhibitor (PhosSTOP™ Roche), and protease inhibitor (PROTEOLOC™)). After centrifugation, supernatants were complemented with Laemmli 5X and total proteins of 2 × 10⁵ cells were subjected to electrophoresis in a 6 or 10% polyacrylamide gel. The proteins were transferred onto a nitrocellulose membrane (Amersham^TMProtran^TM 0.45μ NC) by a standard Western blot procedure. The membrane was blocked with 5% BSA in TBT‐T buffer (20mM Tris à pH 7.5, 150 mM NaCl, and 0.1% Tween 20) and probed with primary antibodies overnight at 4°C with the following antibodies IRE1α (14C10, Cell Signaling TECHNOLOGY^®), XBP1s (83418, Cell Signaling TECHNOLOGY^®), PERK (C33E10, Cell Signaling TECHNOLOGY^®), p‐eIF2α (Cell Signaling TECHNOLOGY^®), CHOP (B‐3, Santa Cruz^®), GAPDH (D16H11, Cell Signaling TECHNOLOGY^®), and β‐tubulin (Cell signaling TECHNOLOGY). Primary antibody incubation was followed by washing using the TBS‐T buffer prior to species‐specific secondary antibodies conjugated to HRP. The membranes were visualized using ECL Western blotting substrate (Pierce) or SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific™) on a ChemiDoc™ XRS⁺ (BIO‐RAD).

ELISA

ELISA experiments were conducted on non‐infected or infected BMDCs according to the manufacturer’s instructions using the IL‐6 Mouse Uncoated ELISA kit (Invitrogen by Thermo Fischer Scientific), IL‐12/IL‐23p40 (Total) Mouse Uncoated ELISA kit (Invitrogen by Thermo Fischer Scientific), and Mouse IL‐23 DuoSet^® ELISA (R&D SYSTEMS^®). Plates were analyzed on a Spark^® 10M (TECAN).

Antigen presentation assays

BMDCs were plated into flat‐bottom 96‐well and infected with serial dilution of tachyzoïtes. The rate of infected cells (Pru Tomato SAG1‐Ova detection) was determined by flow cytometry. After 6 h of infection, B3Z (Hosken & Bevan, 1990; Karttunen et al, 1992) reporter hybridomas were added. After 16 h of incubation with B3Z, presentation of SIINFEKL (SKL8) peptide by H‐2K^b was assessed by quantification of β‐galactosidase using the chromogenic substrate chlorophenol red‐β‐D‐galactopyranoside (CPRG, Roche). Absorbance was read at 595nm with a reference at 655nm with a Spark^® 10M (TECAN).

ERAI experiment

WT and ERAI mice were infected with 1000 parasites by intraperitoneal injection. Splenocytes were isolated 6 days post‐infection and labeled with Live/Dead eF506 (eBioscience™), anti‐CD45 BV605 (30‐F11, BD biosciences), anti‐Ly6C PerCP‐eF710 (HK1.4, eBioscience™), anti‐CD11b BUV395 (M1/70, BD biosciences), anti‐CD64 BV711 (X54‐5/7.1, BioLegend^®), anti‐MHCII APC‐eF780 (M5/114.15.2, eBioscience™), anti‐CD3e PE‐Cy5 (145‐2C11, eBioscience™), anti‐CD19 PE‐Cy5 (eBio1D3(1D3), eBioscience™), anti‐CD11c eF450 (N418, eBioscience™), anti‐CD26 APC, anti‐XCR1 BV650 (ZET, BioLegend^®), and anti‐CD172a PE‐Cy7 (P84, BioLegend^®). The cells were gated on: Non‐debris/Singlets/Live/CD45⁺/Gated out the monocytes (Ly6C^hi CD11b⁺⁺) and neutrophils (Ly6C^INT+ CD11b⁺)/Gated out macrophages (CD64⁺MHCII⁺)/Gated out T cells (CD3eCD19⁺MHCII⁻) and B cells (CD3eCD19⁺MHCII⁺)/CD26⁺CD11c⁺ MHCII⁺XCR1⁺ CD172a⁻ for cDC1s and CD26⁺CD11c⁺ MHCII⁺XCR1⁻ CD172a⁺ for cDC2s.

Parasite quantification by qPCR

DNA extractions from splenocytes or PEC were performed with the NucleoSpin Tissue kit (MACHEREY NAGEL). Individuals qPCR reactions were processed using TOX9 (5’‐AGGAGAGATATCAGGACTGTAG‐3’) and TOX11 (5’‐GCGTCGTCTCGTCTAGATCG‐3’) primers (Reischl et al, 2003) with SYBR^® Selected Master Mix (Life technologies). qPCR reactions were carried out on a QuantStudio 12K Flex Real‐Time PCR system (Applied Biosystems™ by life technologies). Total parasite number was estimated by using the average Cq for each reaction by comparison to a standard curve generated with defined numbers of parasites.

Statistical analysis

All data were analyzed with Graph Pad Prism 7.0 software (San Diego, California USA). Statistical analyses are performed with the unpaired Student’s t‐test (with Welch’s correction when variances differ significantly), Tukey's ANOVA one‐way test, Tukey's ANOVA two‐way test, and the log‐rank test.

Author contributions

AFP, VB, EH, SC, SR, LH, SMaré, and VP performed experiments. AFP, VB, EH, NB, SJ, and SMari conceived and designed the experiments and performed data analysis. NB and JK contributed reagents, materials, and analysis tools. AFP, VB, NB, SJ, and SMari wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Source Data for Expanded View and Appendix

Review Process File

Source Data for Figure 1

EMBO Reports (2021) 22: e49617

Data availability

No large primary datasets have been generated and deposited.