Research ArticleIMMUNOLOGY

Chemokine receptor CXCR3 is required for lethal brain pathology but not pathogen clearance during cryptococcal meningoencephalitis

See allHide authors and affiliations

Science Advances  17 Jun 2020:
Vol. 6, no. 25, eaba2502
DOI: 10.1126/sciadv.aba2502


Cryptococcal meningoencephalitis (CM) is the major cause of infection-related neurological death, typically seen in immunocompromised patients. However, T cell–driven inflammatory response has been increasingly implicated in lethal central nervous system (CNS) immunopathology in human patients and murine models. Here, we report marked up-regulation of the chemokine receptor CXCR3 axis in human patients and mice with CM. CXCR3 deletion in mice improves survival, diminishes neurological deficits, and limits neuronal damage without suppressing fungal clearance. CD4+ T cell accumulation and TH1 skewing are reduced in the CNS but not spleens of infected CXCR3−/− mice. Adoptive transfer of WT, but not CXCR3−/− CD4+ T cells, into CXCR3−/− mice phenocopies the pathology of infected WT mice. Collectively, we found that CXCR3+CD4+ T cells drive lethal CNS pathology but are not required for fungal clearance during CM. The CXCR3 pathway shows potential as a therapeutic target or for biomarker discovery to limit CNS inflammatory damages.


Fungal pathogens can invade the central nervous system (CNS), resulting in high mortality rates (>50%) and risk of persistent neurological sequelae (1, 2). Our understanding of the relationship between antifungal immunity in the CNS and associated CNS damage is limited. Cryptococcus neoformans is the most common fungal cause of lethal cryptococcal meningoencephalitis (CM) with 181,000 deaths each year worldwide (3). The high prevalence of the cryptococcal disease in immune-deficient human patients (such as HIV/AIDS) supported the view that uncontrolled fungal invasion because of immunodeficiency is a major factor leading to disease pathology. Unexpectedly, recent studies have shown that strong anticryptococcal inflammatory responses are linked to severe neurological sequelae and mortality in a subset of patients with CM (48). Patients with AIDS with treated CM who undergo combination antiretroviral therapy can experience acute neurological deterioration caused by a paradoxical cryptococcal immune reconstitution inflammatory syndrome (c-IRIS) (9, 10). Clinical deterioration in previously healthy individuals with an aggravated anticryptococcal immune response has been termed postinfectious inflammatory response syndrome (PIIRS) (5, 1113).

Current treatments for patients with CM, particularly those with c-IRIS or PIIRS complications, are frequently ineffective (11, 14). This is due, at least in part, to a limited understanding of how dysregulated CNS inflammation causes neurological pathology during CM, as well as a lack of reliable biomarkers to separate patients with CM with inadequate versus exaggerated neuroinflammatory responses (11, 13, 15). The vast majority of mechanistic studies on host defense in cryptococcal infections use murine pulmonary infection models, in which T helper 1 (TH1)–polarized T cell–mediated responses were shown to be protective and to limit disease pathology (1621). The signature TH1 cytokines, interferon-γ (IFN-γ) and tumor necrosis factor–α (TNF-α), activate fungicidal responses in pulmonary effector phagocytes (monocytes, macrophages, and monocyte-derived dendritic cells), marked by the production of nitric oxide (NO) via inducible NO synthase (iNOS) (2224). However, the immune responses in the CNS and their consequences in human patients and murine models of CM appear to differ from those in the lungs during cryptococcal infections (25, 26). For example, enhanced CNS inflammation in patients with c-IRIS or PIIRS, which is often characterized by highly activated T cells in the cerebrospinal fluid (CSF), correlates with neurological damage (11). Patients with c-IRIS and PIIRS show elevated cytokine expressions such as IFN-γ and interleukin-6 (IL-6), and disease deterioration can be ameliorated by treatment with corticosteroids (11, 27, 28). However, PIIRS appear to lack an effective activation of myeloid cells in the brain (5, 11). We recently developed a model of CM/PIIRS in which TH1-polarized CD4+ T cells drive lethal immunopathology, reminiscent of findings in human patients with c-IRIS or PIIRS (29, 30). Because T cells are essential for both fungal clearance and the development of pathology, it is crucial to dissociate protective from pathological pathways within the infected CNS to help stratify patient subpopulations at risk and to define potential therapeutic targets in CM and other persistent CNS infections.

C-X-C motif chemokine 10 (CXCL10), also known as IFN-γ–induced protein 10, plays an important role in recruiting activated T cells into sites of tissue inflammation (31). CXCL10 is prominently up-regulated in the CSF of human patients with c-IRIS or PIIRS (11, 32, 33). Its receptor counterpart, CXCR3, is highly expressed on effector TH1 cells and plays a vital role in T cell trafficking, spatial distribution, and function, especially during type 1 inflammatory responses (3437). CXCR3 and its ligands have been implicated in facilitating neurological disease, including stroke (38), Alzheimer’s disease (39), and multiple sclerosis (40), as well as tissue damage that occurs during CNS infection with murine Plasmodium species (4143), Trypanosoma brucei (44), and HIV (45). These results suggest a potent but diverse role of the CXCR3 axis in neurological diseases. However, the role of CXCR3/CXCL10 axis in host defense and immunopathology during CM has yet to be investigated.

In this study, we found the CXCR3 pathway to be highly up-regulated in the CSF of patients with CM/PIIRS and in a murine CM model. CXCR3-dependent chemotaxis was found to be critical for pathological T cell accumulation in the brains of mice with CM, driving CNS tissue damage and neurological disability. However, CXCR3 was dispensable for intracerebral fungal control. Our results identify the CXCR3 axis as a potential therapeutic target and its blockade could be explored as a method to limit immunopathology without impairing fungal clearance during CM, especially in patients with c-IRIS or PIIRS.


CXCR3 and CXCL10 are highly up-regulated in the brains of mice and human during CM

We examined the expression of 21 chemokines and cytokines in the brains of mice during murine CM induced by experimental intravenous infections with C. neoformans 52D in C57BL/6 mice. CXCL10, a CXCR3 ligand, was the most highly expressed chemokine measured on the protein levels in brain homogenates (Fig. 1A). Levels of CXCL10 started to increase at 14 days post-infection (dpi) (Fig. 1B) and peaked at 21 dpi (Fig 1B). CXCL9, another CXCR3 ligand, was also up-regulated in the brain (Fig 1C) but at a much lower level compared with CXCL10. The expression of other chemokines such as CCL2 (C-C Motif Chemokine Ligand 2), CCL5, CXCL1, and CXCL13 in the brain during CM (Fig. 1A), while substantial, did not reach the level of CXCL10. Immunofluorescence microscopy revealed astrocytes [glial fibrillary acidic protein (GFAP) antibody], rather than CD11b+ myeloid cells, as the major source of CXCL10 during CM (Fig. 1D).

Fig. 1 CXCR3 and CXCL10 were highly up-regulated in the brains of mice during CM.

C57BL/6 mice were infected with 106 colony-forming units (CFU) of C. neoformans 52D via retro-orbital intravenous inoculation to induce CM. (A) Expression of 21 chemokines and cytokines were evaluated using cytometric bead assays in whole-brain homogenates at day 21 after infection. MIP-1a, macrophage inflammatory protein 1a; MDC, macrophage-derived chemokine; GM-CSF, granulocyte-macrophage colony-stimulating factor. BLC, B-cell-attracting chemokine 1; IP-10, interferon gamma-induced protein 10; KC, keratinocyte chemoattractant; LIX, lipopolysaccharide-induced CXC chemokine; MCP-1, monocyte chemoattractant protein-1; MIG, monokine induced by interferon-γ; TARC, thymus- and activation-regulated chemokine. (B and C) The kinetics of CXCL10 and CXCL9 production were evaluated using cytometric bead assays in whole-brain homogenates at the indicated time points after infection. (D) Immunofluorescent analysis of mouse brain stained with antibody to GFAP+ astrocytes, CD11b, and CXCL10 at 21 dpi. DAPI, 4′,6-diamidino-2-phenylindole. Note that stained GFAP and CXCL10 signal colocalized in the merged picture. Scale bars, 25 μm. (E) The kinetics of CD4+ T cell recruitment was evaluated using flow cytometry at the indicated time points after infection. (F) Levels of Cxcr3 mRNA were measured by reverse transcription quantitative polymerase chain reaction (RT-qPCR) in whole-brain homogenates at indicated time after infection. (G and H) The expression of CXCR3 on the surface of CD4+ T cells and CD8+ T cells was measured by flow cytometry. Bar graphs showed the total number of CD4+ T cells and CD8+ T cells expressing CXCR3 in the brains. Results represent means ± SEM from one of three matched experiments (n = 5 mice for each time point). (I) CXCR3 was highly expressed by CSF T cells in the brains of human patients with CM. CSF of human patients was collected, and CXCR3 surface expression levels on CD4+ T cells and CD8+ T cells were analyzed by flow cytometry. The representative flow plot showed the gating strategy and up-regulated expression of CXCR3 on the surface of CD4+ T cells and CD8+ T cells in the CSF of four patients with CM. SSC, side scatter.

Concurrent with the expression of CXCL10, the total number of CD4+ T cells in the brain during CM started to sharply increase after 14 dpi and reached a plateau at 21 dpi (Fig 1E). Cxcr3 mRNA expression in the brains of mice and CXCR3+ cell numbers by flow cytometry all corroborated in demonstrating marked influx of CXCR3+CD4+ and CXCR3+CD8+ T cells into the brain (Fig. 1, F and H) peaking at 21 to 35 dpi. Total numbers of CXCR3+CD4+ T cells and CXCR3+CD8+ T cells substantially increased in the infected brain at 21 dpi, which synchronized with the most pronounced up-regulation of CXCL10 in the brain during CM (Fig. 1, B and C). Thus, CXCR3 and its chemokines are highly up-regulated in the brain during murine CM, suggesting that the CXCR3 axis is an important mechanism for T cell recruitment into the CNS during C. neoformans infection. In parallel, human studies with patients with CM/PIIRS showed that CXCL10 is one of the most highly expressed chemokines in the CSF (11). To complement these findings, we evaluated CXCR3 expression by T cells in CSF of four previously healthy human patients with CM/PIIRS. Both CD4+ T cells and CD8+ T cells in the CSF of four human patients with PIIRS robustly expressed CXCR3 (Fig. 1I). The frequency of CD4+ T cells and CD8+ T cells producing CXCR3 is 89.9 ± 3.4% and 90.1 ± 2.2%, respectively (n = 4). Thus, both human and animal studies support a strong up-regulation of the CXCR3 axis in CM/PIIRS.

T cell recruitment to the brain does not require a breach of the blood-brain barrier

To determine whether a concurrent blood-brain barrier (BBB) breach contributed to T cell accumulation, we used an Evans blue intravenous injection assay. Unexpectedly, T cell accumulation in brains during CM was not linked to BBB damage, as we detected only minor Evans blue dye leakage into brain tissue at 35 dpi (fig. S1A). Thus, BBB breakdown only occurred after the initial wave of T cell infiltration (35 dpi) and was limited.

T cell recruitment to the brain in CM is largely CXCR3 dependent

We next compared CD4+ T cell accumulation in the brains of infected wild-type (WT) and CXCR3−/− mice. The frequency of intracerebral CD4+ T cells rose sharply in WT mice at day 21 and was sustained through day 35 (Fig. 2A). The number of intracerebral CD4+ T cells at 21 dpi was at least 80% lower in CXCR3−/− mice compared with their WT counterparts (Fig. 2A). While CNS T cell accumulation occurred in CXCR3−/− mice between days 21 and 35, it never reached the level observed in WT mice. CD8+ T cell recruitment to the CNS was also reduced in CXCR3−/− mice compared to WT mice at 21 and 35 dpi (Fig. 2B). Immunofluorescence microscopy revealed that CD4+ T cells were preferentially located at the margins of cryptococcal growth areas (Fig. 2C). T cell accumulation in these areas was profoundly attenuated in the infected CXCR3−/− mice (Fig. 2C). Furthermore, CD4+ T cells in infected CXCR3−/− mice were more diffused, and their average distance to the nearest cryptococcal yeast (143.60 ± 23.56 μm) was greater than that of WT CD4+ T cells (28.33 ± 4.92 μm) at day 21 histological sections (fig. S1B). Thus, our data demonstrate the CXCR3 axis plays a critical role in T cell migration into sites of cryptococcal infection in the brain.

Fig. 2 T cell recruitment to the brain was reduced in CXCR3−/− mice during CM.

(A and B) The total cell number of CD4+ T cells and CD8+ T cells in the whole brain was evaluated by flow cytometry. Results represent means ± SEM from one of three matched experiments. n = 5. *P < 0.05 and **P < 0.01. (C) Immunofluorescent analysis of mouse brain stained with antibody to CD4 at 21 dpi. Note that more CD4+ T cells surround the cryptococcal lesions in WT control mice than CXCR3−/− mice. Scale bars, 40 μm.

CXCR3−/− mice are resistant to CM-associated pathology

Recent clinical and animal studies indicate that the host inflammatory response is, in part, responsible for neurological deficits in patients with IRIS and PIIRS (4, 7). We sought to determine whether the CXCR3 pathway contributes to the pathogenesis of CM. WT mice with CM succumbed to infection between days 21 and 35 with a mortality rate of 60 to 80% (Fig. 3A). In contrast, CXCR3−/− mice were significantly protected from CM, with a mortality rate of only 30 to 40% (Fig. 3A). Neurological signs were significantly mild in CXCR3−/− mice at 21 dpi (Fig. 3B), as measured by the murine coma and behavioral scale (MCBS) (29). Levels of fungal burden in the brains of WT and CXCR3−/− mice were comparable through 35 dpi, indicating that fungal clearance was not dependent on the CXCR3 pathway (Fig. 3C). These results demonstrate that CXCR3 deficiency is protective against CNS injury in our CM model and does not compromise fungal clearance. Histological analyses showed that C. neoformans form “Swiss cheese”–like foci in both WT and CXCR3−/− mice (fig. S1C). Inflammatory cells accumulated at the margins of these foci in the brain parenchyma of infected WT mice; however, this intense cellular inflammation was largely absent in the CXCR3−/− mice (fig. S1C).

Fig. 3 CXCR3−/− mice were markedly protected from CM.

CXCR3−/− mice and control mice were infected with C. neoformans 52D to induce CM. (A) Survival was monitored through day 35 after infection (n = 10). (B) Overall health and neurological status were assessed using a modified MCBS. ****P < 0.0001. (C) Fungal burdens were measured in whole-brain homogenates at the indicated time points after infection. (D) Immunofluorescent analysis of mouse brain stained with antibody to β-III tubulin neurons and caspase-3 at 21 dpi. (E) Immunofluorescent analysis of mouse brain stained with antibody to CD4 and caspase-3 at 21 dpi. Results represent means ± SEM from one of three matched experiments (n = 5 for each time point).

We next stained CNS sections for cleaved caspase-3, a marker of apoptosis, as well as β-III tubulin, a neuronal marker, and CD4. In infected WT mice, β-III tubulin staining identified bundles of neurons located around Swiss cheese foci, indicating neuron displacement by fungal growth. The cleaved caspase-3 signal was mainly detectable in neurons at 21 dpi (Fig. 3D). CD4+ T cells were aggregated adjacent to the damaged neurons (Fig. 3E). In CXCR3−/− mice, we also noted Swiss cheese foci and evidence of neuronal displacement; however, the level of cleaved caspase-3 signal was minimal at 21 dpi, and there were no CD4+ T cells adjacent to the displaced neurons (Fig. 3, D and E). Thus, our data demonstrate that the CXCR3 axis is required for T cell accumulation, as well as the induction of apoptotic signals and histopathological alterations of neurons at the sites proximal to the cryptococcal lesion during CM.

CXCR3 gene deletion prevents the development of a profoundly skewed TH1 response in infected brains

Patients with IRIS and PIIRS show a dominant TH1 T cell polarization with elevated levels of intrathecal IFN-γ (9, 11). Therefore, we next sought to determine whether the CXCR3 pathway contributes to such strong TH1 polarization in the C. neoformans–infected CNS. CXCR3 deletion profoundly ablated the robust induction of IFN-γ in the brains of CM mice (Fig. 4A). Both the absolute number and frequency of CD4+ T cells producing IFN-γ decreased in CXCR3−/− mice compared to WT mice (Fig. 4B). Last, mean fluorescence intensity among IFN-γ–producing CD4+ T cells was also significantly reduced in CXCR3−/− mice compared to that of WT mice, indicating that even the TH1-differentiated cells recruited to the CNS of CXCR3−/− mice produced less IFN-γ relative to the WT mice. We did not observe a significant change in TH2 and TH17 transcriptional factors, Gata-3 (GATA binding protein 3) and Rorc (RAR Related Orphan Receptor C) mRNA expression by reverse transcription quantitative polymerase chain reaction (RT-qPCR) (fig. S2A). However, we did find an increased frequency in forkhead box P3–positive (Foxp3+) regulatory T cells (Fig. 4C), which could account for improved regulation of the immune response in the CNS. We found decreased absolute number but not frequency of IFN-γ–producing CD8+ T cells in the brains of CXCR3−/− mice and WT mice (Fig. 4D), suggesting that the CXCR3 axis affects CD8+ T cell accumulation but not their IFN-γ production.

Fig. 4 Type 1 immune response was impaired in the brains of CXCR3−/− mice during CM.

(A) Levels of IFN-γ mRNA were measured by RT-qPCR in whole-brain homogenates at 21 dpi. (B and C) Frequencies and mean fluorescence intensity (MFI) of IFN-γ and Foxp3 by CD4+ T cells isolated in whole brains were measured by intracellular flow cytometry. (D) Expression of IFN-γ by CD8+ T cells isolated in whole brains was measured by intracellular flow cytometry. The numerical data on individual flow plots in (B to D) represent the frequency as mean ± SEM. (E) The total cell number of Ly6C+CD11b+ cells in the whole brain at 21 dpi was evaluated by flow cytometry. (F and G) iNOS production by microglia and recruited MoCs was evaluated by intracellular flow cytometry. Results represent means ± SEM from one of three matched experiments. n = 5 for each time point. **P < 0.01 and ***P < 0.001.

The TH1/IFN-γ response is known to up-regulate the expression of iNOS by myeloid cells. Consistent with diminished T cell responses, we found that the recruitment of CD11b+Ly6C+ monocyte-derived cells (MoCs) was reduced in CXCR3−/− mice by more than 50% compared to WT mice (Fig. 4E). Furthermore, mRNA expressions of type 1 response genes Nos2, Cd80, and Cd86 were significantly lower in the infected brain homogenate of CXCR3−/− mice than that of WT mice (fig. S2B). However, we did not detect differences in expression levels of type 2 response genes such as Arg1, Gal3, and Ym2 between CXCR3−/− and WT CM mice (fig. S2B), indicating that myeloid responses, albeit less vigorous, have not become M2 polarized. To further evaluate iNOS protein expression in the specific myeloid cell subsets, we used intracellular flow cytometric analysis (Fig. 4, F and G). Results revealed that MoCs, rather than microglia, demonstrated the most robust expression of iNOS. While both myeloid subsets showed decreased iNOS expression in CXCR3−/− mice compared to the WT mice, 30% of MoCs in CXCR3−/− still showed iNOS expression, which, together with the preserved fungal clearance, suggests that myeloid cell M1 polarization could occur to a sufficient level to support their fungicidal function even in the absence of CXCR3 signaling. Together, these data suggest that the CXCR3 pathway promotes excessive type 1 responses likely linked to detrimental clinical outcomes during cryptococcal CNS infections.

CXCR3 deletion has not altered the systemic anticryptococcal immunity

We next asked whether the effect of the CXCR3 pathway was brain restricted. CXCR3 deletion did not affect cell numbers of total leukocytes, CD4+ T cells, CD8+ T cells, or CD11b+Ly6C+ MoCs in the infected spleen at 21 dpi (Fig. 5A). Furthermore, both CD4+ and CD8+ T cells expressed similar levels of IFN-γ between WT and CXCR3−/− mice (Fig. 5, B and C). Consistent with this data, we also found no significant difference in the fungal burden in spleens of infected WT and CXCR3−/− mice (Fig. 5D). Thus, in contrast with its major role in T cell migration and TH1 activation in the CNS, the CXCR3 axis does not profoundly affect the systemic antimicrobial response to cryptococcal infection.

Fig. 5 CXCR3−/− mice did not have impaired immune responses in spleens during CME.

(A) Total cell number of CD45+ cells, CD4+ T cells, CD8+ T cells, and CD11b+Ly6C+ MoCs in the spleens at 21 dpi was evaluated by flow cytometry. (B and C) The expression of IFN-γ by CD4+ T cells and CD8+ T cells isolated in the spleens was measured by intracellular flow cytometry. (D) Fungal burdens were measured in spleens at 21 dpi. Results represent means ± SEM from one of three matched experiments. n = 5 for each time point. ns, not significant.

WT CD4+ T cells but not CXCR3−/− CD4+ T cells restored immune responses and pathology in brains of CXCR3−/− mice

Because CXCR3 is expressed on multiple cell subsets, we next assessed whether CXCR3+CD4+ T cells contributed directly to CNS immunopathology during CM. CD4+ T cells isolated from spleens of WT or CXCR3−/− mice 14 dpi were adoptively transferred into the infected CXCR3−/− recipients (Fig. 6A), and mice were analyzed at 21 dpi. Before adoptive transfer, CD4+ T cells isolated from the spleens of WT and CXCR3−/− mice showed a similar ratio of activated T cells (CD44highCD62Llow) to naïve T cells (CD44lowCD62Lhigh) (fig. S3A). CXCR3 expression was up-regulated in the activated CD4+ T cells but not in naïve CD4+ T cells from WT mice (fig. S3B). CXCR3−/− mice that received WT CD4+ T cells showed nearly the same numbers of CD4+ T cells in the CNS as the infected WT mice, whereas CXCR3−/− mice that received CXCR3−/− CD4+ T cells showed T cell numbers comparable to that in infected CXCR3−/− mice (Fig. 6B). Consistently, IFN-γ expression by CD4+ T cells was restored in CXCR3−/− mice that received WT CD4+ T cells but not in CXCR3−/− mice that received CXCR3−/− CD4+ T cells (Fig. 6C). CXCR3−/− mice that had received WT CD4+ T cells developed neurological symptoms similar to WT mice at 21 dpi, whereas CXCR3−/− mice that had received CXCR3−/− CD4+ T cells showed less severe CM symptoms (Fig. 6D). Furthermore, carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled CD4+ T cells can be detected in the brain of infected CXCR3−/− mice using fluorescent microscopy after adoptive transfer, confirming that WT CD4+ T cells are capable of migrating into the brain of CXCR3−/− mice during CM (fig. S3C). Together, our findings demonstrate that CXCR3 expression by CD4+ T cells plays a vital role in T cell recruitment in the brain and establishes a causal relationship between CXCR3+CD4+ T cells and the inflammatory pathogenesis of murine CM/PIIRS.

Fig. 6 WT CD4+ T cells but not CXCR3−/− CD4+ T cells restored immune responses and pathology in brains of CXCR3−/− mice.

(A) Schematic of the adoptive transfer experiments. CXCR3−/− mice received isolated CD4+ T cells from spleens of infected WT and CXCR3−/− mice at 14 dpi. (B) The total cell number of CD4+ T cells in the brains at 21 dpi was evaluated by flow cytometry in all indicated groups of mice. (C) Expression of IFN-γ by CD4+ T cells in brains was measured by intracellular flow cytometry. (D) Overall health and neurological status were assessed using MCBS. Results represent means ± SEM from one of two matched experiments. n = 5 for each time point. *P < 0.05 and **P < 0.01.


CNS cryptococcal infections are characterized by a high mortality rate (30 to 70%) in patients (5), and survivors often develop severe neurological sequelae requiring permanent care, particularly in patients with PIIRS and c-IRIS (46). Suboptimal treatment can be partly attributed to a lack of understanding of the mechanisms that underlie CNS damage in those infected individuals. Both clinical and animal research increasingly highlights the destructive impact of neuroinflammation in CM pathogenesis in the “apparently immunocompetent” hosts or those with immune restoration (6). Several pathways, such as expression of CXCL10/CXCR3, have been associated with the disease pathology in human studies, but further mechanistic studies in murine models are urgently needed to establish causality and identify targets for intervention.

The question of how to approach the treatment of CM, especially with PIIRS and IRIS, is highly problematic. On one hand, T cells and their TH1 polarization (IFN-γ) are needed for the immune defenses to control the fungus, suggesting that immune activation should be beneficial. On the other hand, inflammation and a strong TH1 response can inflict significant neuronal damage, which is thought to be irreversible in the CNS. To study host defenses within the CM brain, we recently established a murine model of CM. The infected mice developed CNS infiltration with highly TH1-polarized CD4+ T cells, high CSF pressure, neurological signs, and high mortality rates (29, 30), which recapitulate the pathological and clinical features of human CM patients with c-IRIS and PIIRS. Here, we corroborate high levels of IFN-γ, and CCL2 in the CNS during murine CM, in contrast to relatively low expression of CCL3, TNF-α, IL-12, IL-23, and IL-10, which also mirrors human findings (11) (25). Thus, our mouse CM model mimics not only the clinical course but also the chemokine/cytokine profiles of human patients with cryptococcal IRIS and PIIRS.

Because TH1 cells in our model not only are required for fungal clearance but also mediate the severe damage within the CNS (29), our next goal was to further dissect the major players in the host response to separate protective aspects from those that are predominantly pathogenic. We were motivated by studies that identified CXCL10 as one of the most highly expressed chemokines in the CSF of the patients with PIIRS and c-IRIS (11, 32) and conclusively found that two major components of the CXCR3 axis (CXCR3 and CXCL10) are highly up-regulated in both human and murine CM. Our results close the loop on those clinical associations, demonstrating that both outcomes of human and animal studies support a conserved role of the CXCR3/CXCL10 axis in the inflammatory process during CM.

While the means to establish a cause-and-effect relationship of a specific pathogenic pathway by human studies are limited, our CM mouse model allows dissection of the potential mechanistic links within the CNS immunological network. Our previous study revealed the pathological role of CD4+ T cells (29), but it is still not clear whether all CD4+ T cells are pathological and whether any specific pathway is involved in recruiting pathological T cells into the brain during CM. Here, we show that the CXCL10/CXCR3 pathway critically contributes to neuronal damage and mortality. CXCR3 deletion protected the C. neoformans–infected mice from mortality and brain injury, without significantly affecting fungal clearance from the CNS, which is an important prognostic indicator in human patients with CM. However, CXCR3 is required for the migration of pathogenic CD4+ T into murine CNS. Apart from T cell recruitment, CXCR3 also affects the TH polarization state of the recruited T cells. Adoptive transfer of WT CD4+ T cells reconstituted the CD4+ T cell accumulation in the CNS and neuropathology in infected CXCR3−/− mice. This is in contrast with the mice with complete CD4+ T cell depletion, in which we can delay mortality but cannot maintain effective control of fungal growth in the brain (29). These results suggest that CXCR3+CD4+ T cells might be particularly pathological and blocking this pathway may help reduce immunopathology while fungal growth can still be controlled. Because our model closely mimics symptomatology and pathology of human c-IRIS and PIIRS, including the significant accumulation of CXCR3+ T cells and expression of CXCR3 ligands, our study implies a conserved role of CXCR3 axis in the development of inflammatory CNS pathology in human disease. Another encouraging finding was that CXCR3 deletion does not affect T cell or myeloid cell responses in the spleen, suggesting that blocking the CXCR3 axis would have minimal effects on an effective systemic immune response to cryptococcosis that is necessary for microbiological control while ameliorating the more pathogenic aspects of antifungal immunity in c-IRIS or PIIRS. These results indicate a CNS-specific role of CXCR3 in CM, which could be attributed to the signals like CXCL10 produced in the infected brain. While this CXCL10/CXCR3 axis promotes T cell accumulation in the brain, it plays a minimal role in T cell function systemically. Our studies thus suggest that CXCL10 and CXCR3 should be further explored as (i) potential biomarkers to identify patients at increased risk of developing immune-mediated brain injury and (ii) potential therapeutic targets to reduce immune pathology using blocking antibodies or antagonists to the CXCL10/CXCR3 pathway, without affecting CNS fungal clearance and systemic antifungal immunity in CM patients, especially for those with a risk of inflammation-induced neurological sequelae.

The inflammatory response is thought to be a major factor contributing to the mortality or the development of persistent deficits in cognitive and neurologic function in a variety of refractory CNS infections after microbiological control, especially in those patients with c-IRIS and PIIRS (46, 47). The pathological potential of CXCR3+ T cells in the CNS during C. neoformans meningoencephalitis is supported by their implication in the pathogenesis of Alzheimer’s disease (39), multiple sclerosis (40), and CNS malarial and HIV infection (4145). Therefore, a variety of approaches such as CXCR3 knockout mice, blocking antibodies or specific antagonists toward the CXCR3 pathway in preclinical animal models, and gene expression in human tissues have been used to evaluate the potential of the CXCR3 pathway as therapeutic targets or biomarkers (48, 49). A phase 2 clinical trial found that an anti-CXCL10 antibody was safe and effective to treat rheumatoid arthritis; however, its efficacy in inflammatory bowel disease is encouraging only in specific patient populations (5052). Further attempts to define subgroups of patients based on their immunologic features might be beneficial (52). Cumulatively, these studies and our current data highlight the contribution of the CXCR3/CXCL10 axis to a spectrum of inflammatory diseases, including those affecting the CNS. The role of this pathway continues to be investigated in other neurological diseases in which CD4+ T cells and CD8+ T cells inflict damage, including viral encephalitis (53, 54), cerebral toxoplasmosis (55, 56), and cerebral tuberculosis (6, 57).

The pathogenesis of cryptococcal infection was predominantly viewed as the mere consequence of a weakened immune system (5). That view has been challenged by recent clinical studies and our previous reports showing that CD4+ T cells orchestrate lethal immune pathology, despite developing TH1 bias (previously seen as “protective only”), without suppressing fungal clearance (29). However, one major unanswered question in this area is whether pathways mediating fungal control and immune pathology can be separated from one another. We report here that the CXCR3 pathway mediates marked CNS pathology, but its deletion does not significantly affect the host’s ability to control fungal growth during CM. Our study supports the notion that the CXCR3 axis in CD4+ T cells could be a specific “switch” between detrimental or protective responses during CM. A related question addressed by our studies is whether all T cell recruitment to CNS is CXCR3 axis dependent. The first hint for additional mechanisms apart from CXCR3 axis was that not all T cells found in CM expressed CXCR3 and that other chemokines known to recruit T cells apart from CXCR3 ligands were expressed in the infected brain. Furthermore, we still detected some level of CD4+ T cell accumulation in the CNS of CXCR3−/− mice, and these T cells seem to be able to efficiently execute fungal control in the CM. These CD4+ cells still expressed IFN-γ, albeit in lower levels than those in the WT mice. Our results support that T cells recruited to CNS via CXCR3-independent mechanisms induce less acute immune response overlaid with the appropriate level of immunoregulation while being sufficient to control fungal clearance. Future studies are needed to address the pathways responsible for CXCR3-independent T cell migration and function during CM, as well as possible immunoregulatory circuits that help curb the excessive CNS inflammation in combination with antifungal drug treatments.

Expression of CXCL10 can be induced in a variety of cells, including astrocytes, brain endothelium, inflammatory monocytes, monocyte-derived dendritic cells, and neutrophils during different neuroinflammation conditions (42, 44). We found that astrocytes are one major cell type producing CXCL10 during CM. Because CXCR3+CD4+ T cells are central to the development of the CNS damage during CM, our data indicate that resident astrocytes can locally direct the host immune responses and immunopathology. Future studies are needed to determine how CXCL10 and other host factors orchestrate T cell migration and their specific localization during CM.

In summary, we have shown that the CXCR3 axis is strongly induced in both human patients and mice CM models. CXCR3 is the major factor leading to CD4+ T cell accumulation in the C. neoformans–infected brain. Furthermore, we found that CXCR3+CD4+ T cells substantially contribute to pathogenesis but are dispensable for fungal clearance during CM. The deletion of the CXCR3 has minimal effects on systemic anticryptococcal immunity. Our findings suggest that the CXCR3 pathway could represent a potential therapeutic target to limit immunopathology without affecting fungal clearance. Future experiments combining CXCR3 blockade with antifungal drugs are needed to test its efficacy in treating severe CM in murine models and in human patients who are “apparently immunocompetent” or immune restored. CXCR3 pathway could also become a biomarker to stratify patients at increased risk of developing immune-mediated brain injury and determine whether immune-boosting or immunosuppressive strategies are needed for their treatments.


Ethics statement and human patients

All four patients were seen at the National Institutes of Health (NIH) Clinical Center, Bethesda, MD in 2018 to 2019, and informed consent was obtained under an Institutional Review Board–approved protocol (National Institute of Allergy and Infectious Diseases protocol 93-I-0106) for a prospective observational study examining the host genetics and immunology of cryptococcal disease in previously healthy, non–HIV-infected adults. A diagnosis of CM was defined as a positive latex agglutination cryptococcal antigen or the isolation of Cryptococcus in one or more CSF cultures, or both. The four patients analyzed for the present study had completed standard therapy of liposomal amphotericin B plus flucytosine and had negative CSF fungal cultures but underwent referral to the NIH after they developed refractory symptoms consistent with PIIRS, including declining mental status and cranial nerve defects.


C57BL/6 mice were obtained from the Jackson laboratory (Bar Harbor, ME). CXCR3−/− mice were bred and housed under specific pathogen–free conditions in the Animal Care Facility at the Veterans Affairs Ann Arbor Healthcare System. Both female and male mice were aged between 8 and 12 weeks at the time of infection and were humanely euthanized by CO2 inhalation at the time of data collection. For mortality studies, mice were euthanized when they lost 20% body weight, had persistent cranial swelling, and/or developed neurological symptoms. All experiments were approved by the Veterans Affairs Institutional Animal Care and Use Committee under protocol 1408-004 and were performed following NIH guidelines and the Guide for the Care and Use of Laboratory Animals.

C. neoformans

ATCC 24067 (American Type Culture Collection, Manassas, VA), C. neoformans 52D strain, was used to infect mice in this study. Cryptococcal strain was grown for 4 days in Sabouraud Dextrose Broth (Difco). Fungal cells were washed twice in phosphate-buffered saline (PBS), counted on a hemocytometer with trypan blue, and adjusted to a concentration of 5 × 106 per ml before infection. Mice were infected with 106 yeast (in 200 μl of PBS) via retro-orbital intravenous injection under inhaled isoflurane anesthesia. Serial dilutions of the C. neoformans suspension were plated on Sabouraud dextrose agar to confirm the number of viable fungi for the inoculum.

Murine coma and behavioral scale

The MCBS to assess the overall physical and neurological condition of infected mice was performed as previously described (29). Briefly, mice were scored on a scale of 0 to 3 for exploration, balance, gait, body posture, coat condition, grip strength, reflexes (body, neck, pinna, and footpad reflexes), and response to visual stimuli. Lower scores reflect more-pronounced symptoms.

Fungal burdens

To quantify total fungal burden on a per organ basis, brains and spleens samples were serially diluted with distilled water. Aliquots (10 μl) of each sample were then plated on Sabouraud dextrose in duplicates. Forty-eight hours after plating, total colony-forming units were calculated from the average of the duplicate counts.

Cytokine and gene expression

Cytokine protein levels in the brain homogenate were quantified using LEGENDplex cytometric bead assays (BioLegend) following the manufacturer’s instructions. Gene transcript levels were determined by qPCR. Total RNA from brain homogenate was achieved using the TRIzol reagent (Life Technologies), which were then converted to cDNA with QuantiTect Reverse Transcription Kit (Qiagen) according to the manufacturer’s instructions. Gene expression was quantified with SYBR Green amplification (Radiant Green master mix; Alkali Science) using a LightCycler 96 thermocycler (Roche). Relative gene expression was then analyzed using the 2-ΔΔCq method relative to housekeeping genes Gapdh and 18S ribosomal RNA.

Cell isolation

Mice were euthanized and then perfused with 10 ml of PBS to remove circulating red blood cells and leukocytes from the brain. The brains were aseptically removed, transferred to gentleMACS C tubes containing 5 ml of sterile complete RPMI 1640 (with 5% fetal bovine serum, 25 mM Hepes, GlutaMAX, penicillin-streptomycin, nonessential amino acids, sodium pyruvate, and β-mercaptoethanol). The tissue was minced and then processed on a gentleMACS homogenizer (Miltenyi Biotec). Small samples of the homogenate were collected immediately after processing for whole-brain fungal burden and RNA and cytokine measurements. The remaining homogenate was washed with RPMI 1640 and filtered through a 70-μm cell strainer. A discontinuous 30%/70% Percoll (GE Healthcare) gradient was used to remove cell debris, myelin, and neurons, and then, microglia and brain-infiltrating leukocytes were recovered from the 30%/70% Percoll interface. Isolated cells were washed twice with PBS to remove residual Percoll before use in assays. Total cell numbers were determined by counting live cells on a hemocytometer with trypan blue.

For splenocyte isolation, spleens were removed and then mechanically dispersed by using a 3-ml sterile syringe plunger to press through a 70-μm cell strainer (BD Falcon, Bedford, MA) in complete RPMI 1640 medium. The cell suspension was washed with PBS and centrifuged. Erythrocytes in the cell pellets were lysed by the addition of 5 ml of NH4Cl buffer [0.829% NH4Cl, 0.1% KHCO3, and 0.0372% Na2EDTA (pH 7.4)] for 5 min, followed by addition of a 10-fold excess of RPMI 1640 medium. After centrifugation, the splenocytes were saved for further use. For CSF cells from human patients, CSF was centrifuged; then, cells were washed and resuspended at a concentration of 106 per ml and were immediately proceeded to Fc block and antibody staining.

CD4+ T cell enrichment and adoptive transfer

Spleens from mice at 14 dpi were harvested and digested as mentioned above. CD4+ T cells were isolated using an EasySep Mouse CD4+ T Cell Isolation Kit (Stemcell Technologies, Vancouver, Canada) according to the manufacturer’s instructions. Greater than 98% of the enriched cells were CD4+ cells, which was determined by flow cytometry. All infected mice at 14 and 18 dpi were given 2 million enriched CD4+ T cells via retro-orbital intravenous injection under inhaled isoflurane anesthesia. CellTrace CFSE Cell Proliferation Kit (Thermo Fisher Scientific) was used to label CD4+ T cells according to the manufacturer’s protocol.

Flow cytometry

Cells were stained with fixable LIVE/DEAD dye (Life Technologies), blocked with anti-CD16/32, and stained with antibodies for CD45 (30-F11), CD3 (145-2C11), CD4 (GK1.5), CD8 (53-6.7), CD11b (M1/70), CD11c (N418), CXCR3 (CXCR3-173), Ly6C (HK1.4), Ly6G (1A8), CD44 (IM7), CD62L (MEL-14), and/or major histocompatibility complex class II (M5/114.15.2). All antibodies were purchased from BioLegend. For IFN-γ production, the cells were stimulated for 6 hours with phorbol myristate acetate and ionomycin in the presence of brefeldin A and monensin for the final 4 hours. The cells were stained for extracellular markers and then fixed with fixation/permeabilization buffer, and intracellular staining for Foxp3 and IFN-γ was performed in permeabilization buffer. Fluorescence minus one controls were used for all experiments. Data were collected on either an LSR II cytometer (BD) or LSRFortessa (BD) and were analyzed using FlowJo (Tree Star). Microglia are defined as CD45intCD11b+, while MoCs are CD45hiCD11b+Ly6c+.

Confocal microscopy

The excised brain was freshly frozen in OCT (Optimal Cutting Temperature compound)–filled mounting molds (Sakura Finetek USA Inc., Torrance, CA). Frozen microtome sections (10 μm thick) were mounted on adhesive slides and fixed in ice-cold acetone for 10 min. Sections were washed in Dulbecco’s PBS (DPBS) for 10 min and then were blocked with LEAF Purified anti-mouse CD16/32 antibody (10 mg/ml; BioLegend) in DPBS for 30 min at room temperature. Primary antibodies were added at a 1:200 dilution in DPBS for Alexa Fluor 488 anti-mouse CD11b antibody (BioLegend), Alexa Fluor 594 anti-mouse CD4 antibody (BioLegend), Alexa Fluor 647 anti-mouse GFAP (BioLegend), anti-mouse CXCL10 antibody (R&D Systems), β-III tubulin (Abcam), and cleaved caspase-3 (Cell Signaling Technology). The secondary antibody was added at a 1:500 dilution for fluorescein isothiocyanate–donkey anti-goat immunoglobulin G (IgG) (Santa Cruz Biotechnology), Alexa Fluor 647–donkey anti-rabbit IgG (Invitrogen), or goat anti-mouse antibody (1:500) conjugated to Alexa Fluor 555. After washing in DPBS three times (3 min per wash), ProLong Gold Antifade Mountant with 4′,6-diamidino-2-phenylindole (Life Technologies) was used for mounting the coverslips. In control experiments, matching isotype antibodies (isotype and fluorochrome) were used instead of the primary antibodies. Samples were visualized using a KEYENCE microscope BZ-X800 series. Images were captured by the camera provided by the vendor. Images were captured by using a ×10 objective 0.45 numerical aperture (NA) Nikon or a ×60 1.4 NA Nikon. For ×60 magnification, three-dimensional stacks of 0.5 μm were taken. Images were processed by BZ-X800 Analyzer. Images with ×60 magnification were deconvolved for enhancing the clarity of the images. The C. neoformans–infected areas in the brain are defined by the presence of yeast cells, especially cryptococcal clusters forming the “Swiss cheese holes” within the brain parenchyma under phase-contrast microscopy before taking pictures with fluorescence microscopy.

Statistical analysis

Statistical analysis was performed using GraphPad Prism version 8 software with Student’s t test or analysis of variance (ANOVA) plus Bonferroni post hoc test for multiple comparisons. Asterisks on figures (in graphs or in the corners of flow cytometry plots) indicate statistical significance as follows: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We acknowledge the technical assistance of E. Xing, A. Young, H. Gong, A. Murine, P. Coit, and Z. Zhao. Funding: Funding was from the U.S. Department of Veterans Affairs (1I01BX000656 to M.A.O.), VA RCS Award (1IK6BX003615 to M.A.O.), the NIH (NHLBI, T32 HL007749 to L.M.N.), and the National Natural Science Foundation of China (31560706 to X.H.). P.R.W. was supported, in part, by the Intramural Research Program of the NIAID and NIH (AI001123 and AI001124) and extramural awards (U01 AI109657). Author contributions: Experiments were performed by J.X., L.M.N., A.G., J.L.K., C.H., W.E., and J.C.H. Experimental support and methods: X.H., M.I., R.L., and J.Z. Writing and revision or serious assistance to writing and revision: J.X., B.S., P.R.W., and M.A.O. Project supervision: M.A.O. Competing interests: The authors declare that they have no competing interest. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article