Deletion of Inflammasome Components Is Not Sufficient To Prevent Fatal Inflammation in Models of Familial Hemophagocytic Lymphohistiocytosis

Hemophagocytic lymphohistiocytosis (HLH) is a severe disease caused by abnormal activation and proliferation of macrophages with an increased phagocytic activity (1). The main clinical and biochemical features of HLH include nonremitting high fever, hepato-splenomegaly, cytopenia, hypertriglyceridemia, and hyperferritinemia. This inflammatory syndrome is also characterized by the presence of numerous well-differentiated macrophages phagocytosing RBCs (hemophagocytosis) in the bone marrow (2).

Known primary causes of HLH are genetic defects of cytotoxicity (familial HLH [FHL]) or mutations of the inflammasome receptor and adaptor NLRC4 (1, 3, 4). However, HLH can also occur secondary to infections, especially by viruses of the herpes family, malignancies such as non-Hodgkin’s lymphoma, as well as autoinflammatory diseases such as systemic-onset juvenile idiopathic arthritis (sJIA) and its adult form, Still’s disease (5). Early diagnosis of HLH and aggressive treatment are mandatory to prevent fatal evolution. Despite improved treatment of HLH, its prognosis remains severe with ∼50% mortality. First line therapy protocols typically include steroids associated to etoposide (6). A better knowledge of immune mechanisms causing HLH are therefore crucially needed to improve treatment.

FHL is caused by mutations affecting the cytolytic effector protein perforin (PRF1), or proteins involved in the molecular machinery required for the biogenesis or exocytosis of PRF1-containing vesicles into the immune synapse. Thus, mutations in PRF1 are responsible for FHL type 2, UNC13D for FHL type 3, STX11 for FHL type 4, and STXBP2 for FHL type 5. FHL can also be presented with hypopigmentation as in Griscelli syndrome type 2 (RAB27A) and Chédiak–Higashi syndrome (LYST) that also involve proteins associated with vesicle trafficking (7). Our understanding of the immunopathological mechanisms responsible for HLH development has gained from the availability of several murine models of FHL. Cytotoxic-deficient mice (such as Prf1, Rab27a, or Unc13d- or Lyst-deficient mice) develop an HLH-like syndrome postinfection with viruses like lymphocytic choriomeningitis virus (LCMV) (8) or mouse CMV (MCMV) (9). In the absence of cytotoxic activity, an accumulation of APCs that continuously activate CTLs was reported in Prf1^−/− mice. Hypersecretion of IL-2, TNF-α, and IFN-γ by CTLs is linked to failed disengagement of Prf1^−/− NK cells or CTLs from their targets, causing a prolonged activating synapse (10). The hyperactivated CTLs thus secrete high levels of IFN-γ, which appears critical for the development of HLH-like symptoms as shown using blocking Abs (8). Indeed, prolonged systemic exposure to physiologically relevant levels of IFN-γ is sufficient to cause acute cytopenias and hemophagocytosis by macrophages, which is the hallmark of this syndrome (11). IFN-γ acts directly on macrophages in vivo to alter endocytosis and phagocytosis and provoke blood cell uptake. Moreover, IFN-γ induces production of IL-6 and other cytokines by multiple cell types (10), an effect that can be blocked by JAK1/2 inhibitors (12). The crucial role of CTLs for HLH-like development has been highlighted by experiments in which depletion of CTLs, but not NK cells, prevents HLH-like manifestations (8). Nevertheless, NK cells appear to exert a certain control on overly activated cells as the presence of PRF1-competent NK cells in an otherwise PRF1-deficient host diminishes both overactivation of CTL and macrophages and ameliorates survival upon infection with LCMV (13).

The inflammasome (14) is an innate immune platform activated in response to danger signals and infections, leading to pro–Caspase-1 maturation. Caspase-1 is an inflammatory caspase leading to the processing and release of pro–IL-1β and pro–IL-18. Canonical Caspase-1 activation takes place within a multiprotein complex, the inflammasome, which includes a sensor and an adaptor, ASC. Several inflammasomes involving different stimulus and sensor pairs have been described. The NLRP1, NLRP3, NLRC4, and AIM2 inflammasomes respectively sense bacterial proteolytic toxins, membrane stress, bacterial Type III secretion system and cytosolic flagellin, and both exogenous and endogenous cytosolic DNA. Two recent studies reported de novo missense mutations affecting the nucleotide-binding domain of the inflammasome component NLRC4, causing familial early-onset systemic autoinflammation and macrophage activation syndrome (called NLRC4-MAS) (3, 4). These mutations induce constitutive NLRC4 activation and subsequent production of IL-1β and IL-18. These studies suggest that a macrophage-intrinsic defect can drive the HLH phenotype in the absence of a primary cytotoxic defect, thus providing a new paradigm for the pathogenesis of HLH. This hypothesis is supported by other works showing that repeated injections of TLR9 ligands in T cell–deficient mice may lead to HLH-like syndrome (15). Monocytes from NLRC4-MAS patients constitutively produce high amounts of IL-1β and especially IL-18 during ex vivo culture. How the inflammasome and these cytokines drive the typical symptoms of HLH, however, remain unsolved. One possibility is that IL-1β and/or IL-18 may act on other cell types to induce various cytokines such as IFN-γ (16). IL-18 induces considerable IFN-γ production by NK cells and T cells when combined with IL-12 (17). Several articles also suggested that NK cell cytotoxicity could be impaired in patients with autoinflammatory syndromes (18–20), perhaps as a result of negative regulation by the chronic inflammatory reaction.

Circulating levels of IL-18 are very high in patients with primary cytotoxic defects or secondary forms of HLH, and IL-18 levels correlate with the various HLH markers (21). In the context of cytotoxic defects, these data suggest that IL-18 is produced secondary to the lack of control of virus infections and IFN-γ production by CTLs, which may promote the establishment of a vicious circle by further enhancing IFN-γ production by NK and CTLs. This hypothesis is supported by the previous observation that in a mouse model of FHL3, loss of MyD88 (the adaptor downstream of IL-1R and IL-18R among others) conferred protection from HLH upon LCMV infection, without affecting the anti–LCMV T cell response (22). Although blocking IL-18 using IL-18 binding protein (IL-18BP), a natural inhibitor of IL-18 function, is not sufficient to prevent HLH development in mouse models of FHL, it alleviates some of the inflammatory biological parameters (23). Moreover, in inflammatory diseases with secondary HLH, a few case reports also indicate an efficacy of anti-IL1 (24) or IL-18BP (25) treatment, thus suggesting a pathological role of IL-1 family members in this inflammatory condition. Whether several IL-1 members may synergize to induce IFN-γ production by NK and CTLs is not clear. Previous studies showed that both IL-1α and IL-1β could induce IFN-γ production by murine NK cells when combined with IL-12 (26). This synergistic effect was not as pronounced as in the case of IL-12 and IL-18 association, but it could contribute to the excessive IFN-γ production observed during HLH.

Altogether, genetic studies show that the development of HLH can have at least two different causes: a defect in cytotoxic function or a deregulated inflammasome activation, which suggests a functional link between both events. Our working hypothesis was that inflammasome activation could be a key mechanism of HLH, driving the overt inflammation observed during this syndrome and that it could be direct (e.g., patients with NLRC4 mutations) or indirect (due to deregulated cytokine secretion following hyperactivation of CTLs in FHL patients). To test this hypothesis, we used mouse models of HLH in which inflammasome activation was prevented by genetic means.

This study was carried out in strict accordance with the French recommendation in the Guide for the Ethical Evaluation of Experiments Using Laboratory Animals and the European guidelines 86/609/CEE. Procedures including animals were approved by local ethics review board under ENS2015-014 agreement. C57B6J mice were obtained at Charles River Laboratories (L’Arbresle, France). asc^−/− mice were obtained at Vishva Dixit Laboratory (Genentech, San Francisco, CA). casp1^−/− mice were obtained from Denise Monack’s Laboratory (Stanford University, Stanford, CA). prf1^−/− mice were obtained from The Jackson Laboratory (Bar Harbor, ME). All mouse strains were maintained in our animal facility. prf1^−/− mice were crossed with asc^−/− and casp1^−/− mice to generate prf1^−/−asc^−/− and prf1^−/−casp1^−/− mice.

LCMV-Armstrong titers were determined by plaque assay on the Vero cell line and propagated on the BHK cell line. MCMV-Smith was propagated by infecting 3-wk-old BALB/c mice with 50 PFU of MCMV cultured until passage two on primary BALB/c mouse whole-fetus cells. Briefly, salivary glands were collected on ice in 3% FCS DMEM, homogenized with an organ dissociator, and centrifuged at 800 × g for 5 min at 4°C. Supernatant was collected and stored at −80°C until use. Titers were determined by plaque assay using NIH-3t3 cells.

Mice were infected i.p. with 1 × 10⁵ PFU of LCMV-Armstrong or 5 × 10⁴ PFU of MCMV-Smith. Mice were weighted and scored for pain indicators daily. For LCMV, the spleen and liver were harvested 10, 16, and 21 d postinfection. For MCMV, organs were harvested 6 d postinfection. The spleen and liver were separated into two parts for flow cytometry analysis and RNA extraction. For cytokine, hemoglobin, and platelet level determination, blood was collected by retro-orbital sinus rupture.

Blood was collected in 0.5 M EDTA, and RBCs were lysed with ACK Lysing Buffer. Mononuclear cells were isolated from the spleen, and the liver was injected with complete RPMI 1640 5% FCS 0.4 mg/ml Collagenase IV (Serlabo Technologies, Vedène, France) and 0.1 mg/ml DNase I (Roche), cut into small pieces, and incubated at 37°C with 150 rpm shaking for 30 min. After incubation, organs were homogenized, and cells from the liver were purified using Percoll (GE Healthcare) density gradient separation. Virus-specific CD3⁺CD8⁺ lymphocytes were identified using H-2 Db LCMV NP396 396–404 (FQPQNGQFI)–BV421 or H-2 Db MCMV M45 985–993 (HGIRNASFI)–BV421 (NIH Tetramer Core Facility) tetramer staining for 30 min at 4°C followed by CD3-FITC (145-2C11), CD8–APC-eFluor780 (53-6.7), and NK1.1-APC (PK136) staining. Inflammatory macrophages were identified as Lin–F4/80⁺ CD11b⁺CD64⁺LY6C^hi. Flow cytometry was carried out on a FACS Fortessa (Becton Dickinson). Data were analyzed using FlowJo (V10; Tree Star)

Platelets, hemoglobin, and neutrophil levels were assessed on a XN-10 analyzer (Sysmex). IFN-y (DuoSet, R&D), IL1β (Quantikine, R&D), IL-18 (MBL), IL-6 (DuoSet, R&D), and ferritin (Abnova) levels in serum were measured by ELISA. Triglyceride levels were dosed using enzymatic method (Abbott Architect C16). Serum aspartate aminotransferase (ASAT) and alanine aminotransferase (ALAT) were measured by enzymatic method (Abbott Architect).

Spleen and liver pieces were immediately put in TRIzol on ice, homogenized using Precellys (Bertin Technologies), and kept at −80°C until analysis. Relative LCMV levels were obtained by RT-qPCR (Reagents) using the following primer pairs: LCMV; RT-Ms-LCMV 5′-CATTCACCTGGACTTTGTCAGACTC-3′ (forward) and RT-Ms-LCMV 5′-GCAACTGCTGTGTTCCCGAAAC-3′ (reverse).

Data were analyzed with GraphPad Prism 6 software. Survival and HLH incidence curves were analyzed by using the Mantel-Cox test. Other analyses were performed by using one-way ANOVAs. Differences were considered to be statistically significant when p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).

To address the role of inflammasome activation in FHL, we crossed prf1^−/− mice with asc^−/− (also known as pycard^−/−) or casp1/11^−/− (hereinafter referred to as casp1^−/−) mice to generate prf1^−/−asc^−/− mice or prf1^−/−casp1^−/− mice. Asc deletion abolishes the activation of most inflammasomes, whereas casp1 deletion abolishes activation of all canonical inflammasomes (including inflammasomes not strictly dependent on ASC, such as NLRC4-dependent inflammasomes). Groups of prf1^−/−, prf1^−/−asc^−/−, prf1^−/−casp1^−/−, and control mice (wild type, asc^−/−, or casp1^−/−) were infected with LCMV-Armstrong. As previously reported, most prf1^−/− mice succumbed to infection within 8–20 d (Fig. 1A), which correlated with pronounced weight loss (Fig. 1B) and a lack of virus control (Fig. 1C) compared with control mice of either genotype. Similar observations were made for prf1^−/−asc^−/− and prf1^−/−casp1^−/− mice, despite a consistently delayed death and reduced weight loss for prf1^−/−casp1^−/− mice (Fig. 1A, 1B). We monitored several biological parameters associated with HLH, including blood platelet and neutrophil counts, spleen weight, blood hemoglobin, triglycerides, ferritin, and hepatic transaminases (ALAT and ASAT). For all these parameters, prf1^−/−, prf1^−/−asc^−/−, and prf1^−/−casp1^−/− mice grouped together, aside from control C57BL/6, asc^−/−, and casp1^−/− mice (Fig. 1D–J), which correlated with the overall survival. In particular, platelet and neutrophil counts and hemoglobin levels dropped postinfection, whereas triglycerides, ferritin, spleen weight, and hepatic enzymes notably increased. To correlate these biological parameters with immune mechanisms, we measured the level of serum cytokines IFN-γ and IL-18. Serum levels of IFN-γ, known to be a critical factor in the development of HLH, were significantly elevated in all three prf1^−/− murine models compared with control groups, thus demonstrating that IFN-γ production in this HLH model is independent of inflammasome activation (Fig. 1K). As expected, IL-18 levels were strongly reduced in mice deficient of Caspase-1 or ASC, but still were consistently higher than background levels (Fig. 1L). Serum IL-1β was undetectable in all mice analyzed (data not shown). The LCMV-specific CD8 T cell response was also monitored using LCMV nucleoprotein (NP)–specific MHC tetramers. In all three prf1^−/− murine models, the percentage of NP-specific CD8 T cells was decreased on day 10 and especially on day 16 postinfection compared with control groups, in both the spleen and liver (Fig. 1M). A similar trend was observed for the percentage of NK cells in both the spleen and liver (Fig. 1N). No significant difference was noted between prf1^−/−, prf1^−/−asc^−/−, and prf1^−/−casp1^−/− groups for either NP-specific CD8 T cells or NK cells. We then quantified the macrophage infiltration in the liver of infected animals. Inflammatory macrophages were defined as Ly6C⁺ CD64⁺ (Fig. 1O), as previously reported (27). All PRF1-deficient mice had a strong infiltrate of inflammatory macrophages 10 d and especially 16 d postinfection, irrespective of the inflammasome status, in both the spleen and liver. prf1^−/−asc^−/− mice displayed a significantly increased macrophage infiltration compared with prf1^−/− and prf1^−/−casp1^−/− mice.

Thus, we concluded that invalidation of inflammasome components is not sufficient to prevent fatal outcome in this FHL mouse model, which correlates with similar inflammatory parameters in prf1^−/−, prf1^−/−asc^−/−, and prf1^−/−casp1^−/− mice on days 10 and 16 postinfection. Yet, as noted above, the prf1^−/−casp1^−/− mice displayed a delayed death compared with prf1^−/− mice, occurring in half of the mice at late stages of infection between days 16 and 25. In an attempt to identify the causes of this delayed death, we characterized infection and inflammation parameters on day 21 postinfection in all prf1^−/− mouse groups. As shown in Fig. 2A, the virus was still detectable in the blood at this stage and was increased in prf1^−/− mice compared with prf1^−/−asc^−/− and prf1^−/−casp1^−/− mice. Hemoglobin and platelet levels were very low in all mouse groups (Fig. 2B, 2C). Liver damage was very severe in all groups as measured by ASAT and ALAT levels, with a trend for increased inflammation in inflammasome-deficient groups (Fig. 2D, 2E). Interestingly, IFN-γ and IL-18 (Fig. 2F, 2G) were much decreased compared with earlier time points, IL-18 production being dependent on both Caspase-1 and ASC. The percentage of LCMV-specific CD8 T cells was increased in the liver, but not in the spleen, of inflammasome-deficient groups compared with controls (Fig. 2H). Finally, the percentage of activated macrophages was very high, reaching more than 50% of macrophages found in both the spleens and livers of all mouse groups (Fig. 2I). This analysis suggests that late inflammation in this FHL model is driven by activated macrophages that ultimately lead to multiorgan failure and fatal outcome, regardless of the inflammasome status.

Likewise, for the LCMV model, PRF1-dependent cytotoxicity is also essential to control MCMV, and prf1^−/− mice develop fatal HLH upon infection (9). We decided to use the MCMV infection model as well in our study because many FHL patients develop HLH following infection by human CMV (28). Moreover, infection of C57BL/6 mice with MCMV induces a strong innate immune response, culminating with the expansion of antiviral NK cells bearing the Ly49H receptor that binds to the viral m157 MHC-like molecule on infected cells (29). Previous articles have established the essential role of dendritic cells and IL-18 and IL-12 in inducing protective IFN-γ secretion by NK cells and in promoting the expansion of the Ly49H-positive subset (reviewed in Ref. 30). This observation prompted us to test the role of inflammasomes in HLH development in the MCMV model. prf1^−/−, prf1^−/−asc^−/−, prf1^−/−casp1^−/−, and control mice (wild type or casp1^−/−) were thus infected with MCMV-Smith at the 5 × 10⁴ PFU dose. As previously described (9), most prf1^−/− mice died following infection, within 4–8 d (Fig. 3A). This correlated with pronounced weight loss (Fig. 3B). Biological analyses on day 6 revealed thrombocytopenia and mild neutropenia (Fig. 3C, 3D). However, there was no hepatosplenomegaly (Fig. 3E), and blood hemoglobin levels (Fig. 3F) were normal in all PRF1-deficient animals. Hepatic enzymes were only elevated in inflammasome-deficient animals, either Caspase-1– or ASC-deficient, irrespective of PRF1 expression (Fig. 3G). When we compared serum cytokine levels in the different mouse groups, we noted that prf1^−/− mice had an increased level of IFN-γ at early time points compared with all other mouse groups (Fig. 3I). At this time point, the T cell response was still negligible, suggesting that early production of IFN-γ by NK cells is dependent on inflammasome-derived cytokines in the MCMV model of HLH. However, at later time points, the level of serum IFN-γ was highly increased in all PRF1-deficient groups compared with control groups, and at least 4-fold higher than at early time points (Fig. 3I). IL-18 levels progressively increased in prf1^−/− mice compared with all other genotypes. As expected, the deletion of asc or caspase-1 prevented this increase (Fig. 3H). As in the LCMV model, we also monitored the viral-specific CD8 T cell response, NK cell percentage, and inflammatory macrophage infiltration. As shown in Fig. 3J–M, a strong MCMV-specific CD8 T cell response was detected in all mouse groups in both the spleen and liver. The percentage of CD8 T cells, however, was decreased to the same extent in prf1^−/−, prf1^−/−asc^−/−, and prf1^−/−casp1^−/− mice compared with control mice in both organs. No major difference was observed for the percentage of NK cells between the different mouse groups in either organ (Fig. 3L). The percentage of inflammatory macrophages was low in the spleens of all mice groups but was increased in all PRF1-deficient mice groups compared with control mice in the liver, and the absence of ASC or Caspase-1 did not influence this manifestation (Fig. 3M).

Altogether, our data demonstrate that HLH induction is independent of inflammasome components in PRF1-deficient mice.

IL-18 is produced at exceptionally high levels in the serum of patients undergoing HLH (21). It was previously suggested that IL-18 and other IL-1 family members may play a role in HLH based on the following evidence: 1) disruption of MyD88-signaling prevents HLH development in a mouse model of FHL3, 2) IL-1 blockade with anakinra can successfully treat HLH in the context of sJIA (31), and 3) IL-18BP treatment alleviates some of the symptoms observed in the course of the disease in a mouse model of FHL type 2 (23). However, the latter treatment may not optimally block IL-18 activity in all tissue or cellular microenvironments, such as the immunological synapses (32). Besides, other inflammasome-dependent IL-1 family members may compensate IL-18 blockade to induce IFN-γ production by NK and CTLs. To circumvent this issue and further test the role of inflammasome-dependent cytokines in FHL, we blocked inflammasome activation by deleting asc or caspase-1 in prf1^−/− mice infected with LCMV or MCMV.

Of note, IL-18 levels were not null in prf1^−/− asc^−/− or prf1^−/− casp1^−/− mice upon LCMV infection, suggesting that ASC or Caspase-1 is not completely essential for the production of these cytokines. The nature of the inflammasomes activated during LCMV or MCMV infection remains unclear; AIM2 at least is activated during MCMV infection (33). Moreover, other proteases may activate IL-1 family cytokines, as previously reviewed (34). Despite this limitation, our data clearly show that more than 80% of the total levels of serum IL-18 are reduced during MCMV or LCMV infection. In the case of MCMV, and as previously published (35), early production of IL-18 on day 2 was required for IFN-γ production, which is mediated by NK cells (36). However, at the peak of the inflammation (day 6 for MCMV and day 10 for LCMV), there was an absence of correlation between IL-18 and IFN-γ levels in infected mouse sera, and the invalidation of inflammasome components did not prevent fatal outcome in prf1^−/− mice. In prf1^−/− casp1^−/−, but not prf1^−/− asc^−/− mice, the kinetics of death following LCMV infection were slower than those of prf1^−/− mice. This was not due to a reduced inflammation in this group on day 10 and 16, and in particular not to reduced IFN-γ production. Because half of the prf1^−/− casp1^−/− mice died at late stages of infection, we compared inflammatory parameters in the different mouse groups on day 21. No major difference was noted between the mouse groups, suggesting that no specific reaction or feedback loop may occur in a late stage of infection in prf1^−/− casp1^−/− mice. Yet, caution should be taken when interpreting those data, because 90% of prf1^−/− were already dead on day 21, and the few mice surviving at this stage may be outliers (i.e., mice injected with slightly less virus). The mechanism of delayed lethality in the absence of Caspase-1 thus remains enigmatic. Caspase-1 is broadly expressed and has roles in cell biology beyond inflammation, such as apoptosis, for example. It can be speculated that multiorgan failure could be delayed by virtue of a decreased cell apoptosis.

Our data demonstrate that inflammasome-dependent IL-1 family cytokines are not essential drivers of HLH development in mouse models of FHL, and suggest that viral Ags are sufficient to induce IFN-γ production by T and NK cells at the peak of inflammation. Targeting activated T cells in this group of disease, as is the current practice, is therefore an appropriate option. This can be efficiently achieved using etoposide, which selectively ablates activated T cells (37). Moreover, a recent study showed that blocking Abs directed against ST2, the receptor for the inflammasome-independent IL-1 member IL-33, reduced FHL symptoms in prf1^−/− mice infected with LCMV by decreasing the number of activated T cells (38). However, in MCMV-infected mice, deletion of the ST2-encoding gene Il1rl1 had the opposite outcome, IL-33 signaling being crucial for regulatory T cell accumulation after MCMV infection (39). Targeting IL-33 may therefore not be appropriate in all HLH contexts.

Several articles have reported the characterization of gain-of-function activating mutations in NLRC4 (3, 4) or, more recently, in NLRP1 (40) or Pyrin (41) in patients with autoinflammatory diseases. In NLRC4-MAS, constitutive activation leads to pathogenic macrophage activation, uncontrolled production of active forms of IL-1β and IL-18, and increased pyroptosis. This macrophage-intrinsic defect may drive the HLH phenotype in the absence of a primary cytotoxic defect. A recent article further validated this point by showing that mice carrying a transgene expressing high levels of free IL-18 (i.e., not associated with IL-18BP) develop more severe HLH upon repeated TLR9 stimulation (42). Accordingly, the conditions of these patients improved upon IL-1 blockade (3). This treatment, however, was ineffective in the case of another NLRC4 mutated patient, the condition of the latter improving only after the introduction of IL-18BP as a neutralizing agent of free IL-18 (25). Thus, in the context of autoinflammatory diseases with proven inflammasome mutations, targeting IL-1 family cytokines is the treatment of choice. Targeting IFN-γ using Abs may also be effective, as shown in a transgenic mouse model overexpressing human IL-6 (43). However, this therapeutic option should be undertaken with caution in sJIA patients undergoing viral infection because IFN-γ is also a crucial antiviral cytokine that plays a protective role against intracellular pathogens, even if it depends on the genetic context (44).

In conclusion, we show in this study that reducing the production of IL-1 family cytokines via genetic suppression of inflammasome activation does not reduce IFN-γ levels and HLH development in PRF1-deficient mice. However, intriguing links between cytotoxicity and inflammasome activation remain, as best exemplified by the case of type 2 X-linked lymphoproliferative disease caused by XIAP loss-of-function mutations. Indeed, X-linked lymphoproliferative disease patients are highly susceptible to infection by herpesviruses such as EBV, similar to FHL patients (45). These patients present several autoinflammatory symptoms due to a role for XIAP in negative regulation of inflammasome activation (46). More in-depth studies will be required to address the nature of inflammasomes activated during HLH and how cytotoxicity, or lack thereof, may influence this activation.

We thank Dr. Marc Dalod and Dr. Elena Tomasello for providing the MCMV virus and for technical advice, Dr. Rafi Ahmed for providing the LCMV virus, and Dr. Vishva Dixit and Dr. Denise Monack for providing knockout mice. We also thank core facilities of the SFR Biosciences, in particular the flow cytometry and animal experimentation facilities.

The authors have no financial conflicts of interest.