C57BL/6 and BALB/c mice were purchased from Taconic Biosciences (Rensselaer, NY, United States) and were approximately 8 weeks old. Ovalbumin (OVA) transgenic TCR mice (OT-1) and perforin deficient mice (Prf1 ^−/−) were purchased from the Jackson Laboratory (Bar Harbor, ME, United States). All mice were used in protocols approved by the Institutional Animal Care and Use Committees at the respective institutions.

CRS was induced in BALB/c mice by ConA injection (20 mg/kg, IV). Animals were dosed with ruxolitinib (30 or 60 mg/kg, PO), anti-IL-1R mAb (25 mg/kg, IP) or anti-IL-6R mAb (25 mg/kg, IP) 60 min before (prophylactic) or 30 min after (therapeutic) CRS induction. Two hours after ConA injection, mice were sacrificed and serum was collected for cytokine measurement.

For the secondary HLH model, C57BL/6 mice were injected IP with 50 μg of CpG DNA on days 0, 2, 4, 6, and 8. Starting on day 5, mice were treated with vehicle (PO, BID), ruxolitinib (60 mg/kg, PO, BID), anti-IL-1R mAb (25 mg/kg, IP, QD), or anti-IL-6R mAb (25 mg/kg, IP, QD). In the primary (genetic) model of HLH, Prf1 ^−/− mice were IP infected with 2 × 10⁵ PFU LCMV Armstrong and treated with control or ruxolitinib chow (2 g/kg) starting on day 4 post-infection (Meyer et al., 2020).

Splenic T-cells were isolated using a pan T-cell isolation kit and an autoMACS Pro Separator (Miltenyi Biotec, Bergisch Gladback, Germany). Approximately 2 × 10⁶ cells were lysed in 500 µl Trizol (Invitrogen, Carlsbad, CA, United States) in gentleMACS M tubes. RNA was purified using a Trizol Plus RNA purification kit (Invitrogen). 100 ng of RNA was hybridized with the nCounter mouse immunology panel codeset (NanoString Technologies, Inc. Seattle, WA, United States) for 18 h. The cartridges were run on an nCounter SPRINT profiler (NanoString Technologies, Inc.). Data were analyzed using nSolver 4.0 Advanced Analysis software. p values were adjusted using the Benjamini-Hochberg method.

Splenocytes were incubated with CFSE to permit measurement of proliferation. T-cells were activated with Dynabeads (Thermo Fisher Scientific) at a 3:1 ratio, resuspended at a density of 0.5 × 10⁶ cells/mL in 24-well plates, and treated with ruxolitinib at various concentrations. The plates were incubated for 7 days, and proliferation was determined by flow cytometry.

Splenocytes from OT-1 or C57BL/6 were resuspended at 5 × 10⁶ cells/mL in complete RPMI, 20 IU IL-2, anti-CD3 (5 μg/ml, plate bound) and anti-CD28 (1 μg/ml) antibodies, and increasing ruxolitinib concentrations. After 3–5 days, OT-1 cells were collected and stained overnight with an anti-CD107a antibody. The cultures were then incubated with OVA⁺ EG-7 tumor cells for 5 h at 37°C. Immediately following stimulation, cultures were washed once, surface stained with directly conjugated antibodies against CD3 and CD8, and analyzed by flow cytometry.

Bone marrow was collected from C57BL/6 mice, and bone marrow cells were cultured in RPMI medium supplemented with 10% fetal bovine serum and 10 ng/mL M-CSF. On day 6, cells were treated with varying ruxolitinib concentrations and then incubated with 2.5 ng/ml lipopolysaccharide (LPS) on day 7. On day 8, cytokines were measured from supernatants. For the in vivo model, C57BL/6 mice were prophylactically dosed with vehicle, ruxolitinib (60 mg/kg, PO), anti-IL-1R mAb (25 mg/kg, IP), or anti-IL-6R mAb (25 mg/kg, IP). Mice were then challenged with LPS (5 µg per animal). Two hours after LPS injection, mice were euthanized, and a peritoneal lavage was performed.

Splenocytes from OT-1 mice were incubated in the presence of 2 µg/ml of the ovalbumin peptide, SIINFEKL, for 3 days. During this time, OVA-expressing EG-7 cells were transfected with a pGL3 luciferase plasmid (Promega, E1751) using lipofectamine 2000 (ThermoFisher, 11,668,030) according to manufacturer’s instructions. After 3 days, OT-1 cells were mixed with EG-7 target cells in a 5:1 ratio and incubated at 37° for 5 h. Following incubation, 50 µL of Bright-Glo luciferase reagent (Promega, E2610) was added to the cultures and fluorescence was measured by plate reader.

Data are reported as mean + SEM in the relevant figures. Differences between groups were analyzed by nonparametric Mann-Whitney test. Statistical analysis for multiple groups was performed by Kruskal-Wallis with Dunn’s post hoc test for nonparametric data sets or analysis of variance with Holm-Sidak’s test for parametric results. All tests were performed using GraphPad Prism (GraphPad Software Inc., San Diego, CA, United States).

Cytokine production is the hallmark of CSS, and the majority of the implicated cytokines signal through the JAK/STAT pathway (Albeituni et al., 2019). We therefore examined whether JAK1/2 inhibition with ruxolitinib at doses that mimic clinically achievable human JAK/STAT target inhibition could be used as a therapeutic modality to dampen hyperinflammation without inducing broad immune suppression. First, using the potent T-cell mitogen ConA, we studied an in vivo model characterized by broad inflammatory cytokine release and lymphocyte proliferation (Gantner et al., 1995). Similar to individuals experiencing hyperinflammatory syndromes, mice receiving ConA have elevated serum levels of multiple inflammatory cytokines, as well as physiological changes, such as fever, malaise, hypotension, hypoxia, capillary leak, multiorgan toxicity, and potentially death. To study the effect of ruxolitinib in this model, animals were therapeutically dosed with 30 or 60 mg/kg ruxolitinib 30 min post ConA challenge. Two hours after ConA injection animals were sacrificed and serum was collected for cytokine analysis. When compared with vehicle-treated animals, ruxolitinib significantly reduced systemic levels of many CSS-implicated cytokines that signal through the JAK/STAT pathway (e.g., IL-6, IL-12, and IFN-γ) in a dose-dependent manner (Figure 1A). Importantly, other JAK-dependent cytokines did not show a significant reduction (e.g., IL-4 and IL-5), indicating that ruxolitinib treatment downregulated the exaggerated cytokine milieu, but did not induce broad immunosuppression. Ruxolitinib did not have a significant effect on cytokines independent of the JAK signaling pathway (e.g., IL-1β, Figure 1A).

Beyond inhibiting the JAK/STAT pathway, the use of monoclonal antibodies (mAb) such as those targeting IL-1 or IL-6 receptors has been explored as a treatment strategy against CSS (Mehta et al., 2020a). To understand these mechanisms further, some animals were dosed therapeutically with blocking mAb against IL-1R or IL-6R 30 min after ConA challenge. Interestingly, neither anti-IL-1R nor anti-IL-6R treatment was able to significantly reduce cytokine levels in our model (Figure 1A), indicating that inhibiting the JAK/STAT pathway effectively targets multiple inflammatory cytokines and clearly differentiates from strategies that block individual mediators (Meletiadis et al., 2020). While higher plasma levels of IL-6 and/or IL-1 can be attributed to the antibodies blocking the receptor, it is unclear why anti-IL-1R and anti-IL-6R failed to reduce other cytokines levels (Nishimoto et al., 2008).

HLH, the prototypical CSS (Meyer et al., 2020), is a rare and often fatal immune disorder characterized by overwhelming activation of T-cell and macrophage compartments and an autocrine loop of proinflammatory cytokines, resulting in unremitting fever, tissue damage and multiorgan failure. To explore the efficacy of oral ruxolitinib at doses which mimic achievable human JAK/STAT target inhibition in primary (i.e., genetic) HLH, we utilized the well-established LCMV infection model of Prf1 ^−/− mice (Albeituni et al., 2019; Das et al., 2016). To define the therapeutic efficacy of JAK inhibition, Prf1^−/− mice were treated with phosphate-buffered saline or infected with 2 × 10⁵ PFU LCMV Armstrong (Figure 1B). Beginning on day 4 post-infection, animals were fed with either control or ruxolitinib chow (2 g/kg) to achieve a JAK target inhibition comparable to 60 mg/kg PO, BID ruxolitinib (Incyte Corporation, Wilmington, DE, United States). On day 9 post LCMV infection, coinciding with the peak of T-cell antiviral immune response, animals were euthanized and organs harvested. Consistent with the published literature (Das et al., 2016), LCMV-infected animals developed anemia, splenomegaly and thrombocytopenia (Figures 1C–F and data not shown); however, following treatment with ruxolitinib, these parameters completely normalized (Figures 1C–F). Importantly, ruxolitinib reduced circulating levels of cytokines implicated in the hyperimmune response, such as G-CSF, IFN-γ, IL-10 and TNF-α (Figure 1G). Furthermore, we also detected a significant reduction in both the percentage and total numbers of IFNγ⁺TNFα⁻ and IFNγ⁺TNFα⁺ CD44⁺CD62L⁻CD8⁺ effector memory T-cells in mice treated with ruxolitinib chow (Supplementary Material Figure S1).

We previously showed that ruxolitinib treatment reduced symptoms in a mouse model of secondary HLH (Das et al., 2016; Albeituni et al., 2019), where hyperinflammation is triggered by immune events as opposed to genetic conditions. Next, we queried whether orally administered doses of ruxolitinib that mimic clinically achievable human JAK/STAT target inhibition would effectively reduce hyperinflammation. As shown in Figure 2A, when compared with vehicle-treated animals, mice receiving 60 mg/kg ruxolitinib had significantly reduced splenomegaly, which coincided with lower levels of pro-inflammatory cytokines KC/GRO (IL-8 functional analogue in rodents) and TNF-α in plasma (Figure 2B) and spleen (data not shown).

Considering that both IL-1 and IL-6 are critical cytokines in CSS, tocilizumab and anakinra have been explored as potential treatments for patients with severe HLH (Dufranc et al., 2020; Mehta et al., 2020a). In our experimental model, treatment with either anti-IL-1R or anti-IL-6R mAb failed to significantly reduce splenomegaly or cytokine levels and indeed dramatically increased plasma levels of IL-6 (Figures 2A,B). These data suggest that blockade of the JAK/STAT pathway may differentiate in efficacy compared with targeting individual cytokines such as IL-1 or IL-6.

To characterize the immunological pathway changes that underpin ruxolitinib-mediated efficacy in the murine secondary HLH model, transcriptome profiling using RNA isolated from splenic T-cells was performed. Principle component analysis revealed clustering of the expression profiles of T-cells from vehicle and anti-IL-6R−treated mice, while ruxolitinib-treated animals clustered separately (Figure 2C). This analysis confirmed that ruxolitinib, but not anti-IL-6R, treatment induced discrete changes in gene expression in HLH mice. Differential expression analysis further revealed that ruxolitinib downregulated many proinflammatory genes, including those involved in lymphocyte activation, cytokine signaling and chemokine signaling (Figure 2D). This analysis resulted in a list of 27 genes that we then analyzed for putative upstream regulators of gene expression (Figure 2E). Consistent with the known physiological mechanism of action of ruxolitinib (Spoerl et al., 2014; Verstovsek et al., 2017; Gozzetti et al., 2020; Zeiser et al., 2020), we confirmed that genes implicated in the response to IL-1 and IFN, as well as several T-cell trafficking chemokines, were the top gene regulators inhibited in ruxolitinib-treated mice.

Because activated T-cells play an important antitumor role, the effect of ruxolitinib on T-cell expansion and effector function was studied. As a model for antigen-specific proliferation and cytolytic activity, we used the well-established OT-1 system, where lymphocytes express a transgenic TCR VαVβ5 specific for the OVA257–264 peptide (SIINFEKL) and restricted to H-2Kb. Antigen-specific T-cells were obtained from OT-1 spleens and expanded in vitro with anti-CD3/CD28 antibodies. As seen in Figure 3A, only concentrations of ruxolitinib higher than the in vitro IC₅₀ (in the single-digit nanomolar range (Quintás-Cardama et al., 2010)) result in a reduction in proliferation, as measured with CFSE staining by flow cytometry (Quah et al., 2007). Importantly, ruxolitinib induced a significant and concentration-dependent reduction on the levels of several inflammatory cytokines implicated in CSS hyperactivity (e.g., IFN-γ and IL-10, with non-significant trends in IL-6 and TNF-α) (Figure 3B).

We then measured the ability of OT-1 antigen-specific lymphocytes to kill target cells after being activated in the presence of ruxolitinib. OT-1 splenocytes activated in the presence of increasing doses of ruxolitinib only saw a reduction in lytic activity at the highest concentrations of ruxolitinib (Figure 3C). Next, we measured the capacity of OT-1 antigen-specific lymphocytes to degranulate in response to OVA-expressing EG-7 tumor cells. As seen in Figure 3D, even T-cells expanded at the higher ruxolitinib concentrations (300 nM) were able to upregulate CD107a in response to the OVA antigen (Betts et al., 2003), indicating that ruxolitinib does not have an adverse effect on the capacity of antigen-specific T-cells to degranulate upon tumor cell encountering. As a negative control, C57BL/6 splenocytes did not degranulate when cocultured with OVA-expressing EG-7 tumor cells (Figure 3D). Therefore, while ruxolitinib treatment reduced proliferation rates and cytokine production of lymphocytes, it did not affect the lytic activity.

Along with T-cells, macrophages are the main cellular subset implicated in hyperinflammatory conditions (Giavridis et al., 2018; Norelli et al., 2018; Crayne et al., 2019). We therefore investigated the effect of JAK1/2 inhibition on cytokine production by macrophages. First, murine bone marrow–derived macrophages were expanded in vitro with M-CSF, and ruxolitinib was added to the cultures on day 6. Treatment with ruxolitinib reduced IL-6 production following stimulation with LPS in a concentration-dependent manner, indicating that the pharmacologic activity of ruxolitinib in reducing production of inflammatory cytokines is not exclusive to T-cells (Figure 4A). We also observed a non-significant trend toward reduction in IL-12p70 and TNF-α (Figure 4A).

We then queried whether ruxolitinib, anti-IL-1R, or anti-IL-6R mAbs would effectively reduce in vivo cytokine production by peritoneal macrophages. Mice were prophylactically treated with vehicle, ruxolitinib (PO), anti-IL-1R (IP), or anti-IL-6R (IP) for 3 days before receiving an intraperitoneal LPS injection. Two hours after LPS injection, cytokines were measured from peritoneal lavage fluid. Only ruxolitinib-treated animals showed a significant reduction in IL-6 levels (Figure 4B), as well as a trend toward IL-12 reduction (data not shown). The inability of anti-IL-6R to reduce IL-6 levels is consistent with clinical studies that have noted an increase in serum IL-6 levels in patients with Castleman disease (Nishimoto et al., 2008), CAR-T cells (Quintás-Cardama et al., 2010) or COVID-19 (Luo et al., 2020) treated with tocilizumab.

To explore the translational validity of the mouse studies, T-cells were obtained from freshly isolated human PBMCs of healthy adults. Following activation with anti-CD3/CD28−coated beads (Richman et al., 2018) and treatment with ruxolitinib, concentrations of several inflammatory cytokines were measured after 7 days of culture. As seen in Figure 4C, when compared with DMSO control, concentrations of ruxolitinib associated with the cellular IC₅₀ significantly reduced levels of key CSS cytokines (IL-6, IL-10, IL-13, TNF-α and IFN−γ).

Taken together, our preclinical studies using doses of ruxolitinib that mimic clinically achievable JAK/STAT target inhibition demonstrate profound downregulation of the proinflammatory cytokine milieu implicated in CRS and CSS pathophysiology. These experimental findings may have implications for the treatment of primary and secondary HLH (Jianguo et al., 2020), as well as other hyperinflammatory conditions (Capochiani et al., 2020; Del Valle et al., 2020; Karki et al., 2020; Yeleswaram et al., 2020).