Exogenous Stimulation of Type I Interferon Protects Mice with Chronic Granulomatous Disease from Aspergillosis through Early Recruitment of Host-Protective Neutrophils into the Lung

ABSTRACT Invasive aspergillosis (IA) remains the primary cause of morbidity and mortality in chronic granulomatous disease (CGD) patients, often due to infection by Aspergillus species refractory to antifungals. This motivates the search for alternative treatments, including immunotherapy. We investigated the effect of exogenous type I interferon (IFN) activation on the outcome of IA caused by three Aspergillus species, A. fumigatus, A. nidulans, and A. tanneri, in CGD mice. The animals were treated with poly(I):poly(C) carboxymethyl cellulose poly-l-lysine (PICLC), a mimetic of double-stranded RNA, 24 h preinfection and postinfection. The survival rates and lung fungal burdens were markedly improved by PICLC immunotherapy in animals infected with any one of the three Aspergillus species. While protection from IA was remarkable, PICLC induction of type I IFN in the lungs surged 24 h posttreatment and returned to baseline levels by 48 h, suggesting that PICLC altered early events in protection against IA. Immunophenotyping of recruited leukocytes and histopathological examination of tissue sections showed that PICLC induced similar cellular infiltrates as those in untreated-infected mice, in both cases dominated by monocytic cells and neutrophils. However, the PICLC immunotherapy resulted in a marked earlier recruitment of the leukocytes. Unlike with conidia, infection with A. nidulans germlings reduced the protective effect of PICLC immunotherapy. Additionally, antibody depletion of neutrophils totally reversed the protection, suggesting that neutrophils are crucial for PICLC-mediated protection. Together, these data show that prophylactic PICLC immunotherapy prerecruits these cells, enabling them to attack the conidia and thus resulting in a profound protection from IA.

(PICLC), chronic granulomatous disease (CGD), conidia, neutrophils, phagocyte recruitment, type I IFN I nvasive aspergillosis (IA) is one of the most serious fungal diseases that immunocompromised patients encounter throughout the world (1,2). The annual global burden of IA is estimated to be more than 300,000 with a high fatality rate (30 to 95%) (3). The patient populations at the highest risk for IA are those with prolonged neutropenia from intensive myeloablative chemotherapy and those with primary immune deficiency such as chronic granulomatous disease (CGD) (4,5). CGD is a rare genetic disorder of the NADPH-oxidase system, in which phagocytes are defective in generating the microbicidal reactive oxygen species (ROS) and their metabolites (4), which are required for host defense against a variety of pathogens. As a result, CGD patients are susceptible to recurrent, life-threatening bacterial and fungal infections (1,5,6). The primary etiologic agent of IA in both neutropenic and CGD patient populations is Aspergillus fumigatus, a species generally susceptible to triazoles and amphotericin B, which are the recommended antifungals for first-line therapy. It is noteworthy that occurrences of infection caused by azole-resistant A. fumigatus mutants have been increasingly reported (7). In contrast to patients with secondary immunosuppression, CGD patients are also presented with less well characterized pathogenic Aspergillus species such as A. nidulans (8), A. pseudoviridinutans (9), A. tanneri (10), A. udagawae (11), and A. calidoustus (12). These species are almost exclusively isolated from CGD patients, and all are more resistant to currently available antifungals than A. fumigatus (S. Seyedmousavi, M. S. Lionakis, M. Parta, S. W. Peterson, K. J. Kwon-Chung, submitted for publication). Generally, infections in CGD patients caused by these species are chronic and require higher doses of antifungals or longer durations of therapy or result in total therapy failure (8,10,11). The predilection of these non-fumigatus Aspergillus species for CGD patients is not yet understood.
Despite the availability of antifungal prophylaxis for CGD patients (13), infections by Aspergillus species are nonetheless the leading cause of mortality in these patients (14,15). Thus, there is a clear need for alternative therapeutic modalities, especially to combat infections by species refractory to currently available antifungal therapy. One such strategy is immunotherapy, which aims to enhance or restore antifungal immunity in patients (16,17). One such antifungal immunotherapy in CGD is type II interferon (gamma interferon [IFN-␥]) administration, which promotes a protective phenotype against aspergillosis (18,19). While IFN-␥ administration is generally beneficial, recent CGD cases of A. tanneri proved fatal even after combined IFN-␥ and antifungal treatment (10), suggesting that alternative immunotherapies are needed.
Induction of type I IFN is another immune therapy strategy studied for eventual use against cancer and/or infectious disease. The role of type I IFN is well documented in host defense against viral infection (20), but it has also been demonstrated to affect the outcome of infections with bacteria (21), parasites (22,23), and fungi (24)(25)(26)(27). The type I IFN family is a multigene cytokine family that encodes 13 partially homologous IFN-␣ subtypes in humans (14 in mice), a single IFN-␤, and several less well defined singlegene products (IFN-, IFN-, IFN-, IFN-, IFN-␦, and IFN-) (25,26). During fungal infection, type I IFN-mediated signaling exerted a protective role against Candida albicans and Cryptococcus neoformans in mice (24,27,28), while it had a deleterious effect against Histoplasma capsulatum (29) and Candida glabrata (30). Poly(I:C) and its stabilized derivative [Hiltonol; poly(I:C) condensed with poly-L-lysine and carboxymethyl cellulose (PICLC)] are synthetic mimics of the viral double-stranded RNA (dsRNA) designed to stimulate prolonged and high-level production of type I IFNs (31)(32)(33)(34). PICLC is currently being investigated in phase II clinical studies in patients with metastatic solid tumors: sarcomas, melanoma, squamous cell skin cancer, and head and neck cancers (clinicaltrials.gov, NCT01984892) (35).
In the current study, we investigated the effect of an elevated type I IFN environment induced by PICLC immunotherapy in murine aspergillosis caused by A. fumigatus and two non-fumigatus Aspergillus species, A. tanneri and A. nidulans, uniquely associated with especially problematic IA in CGD patients (8,10). PICLC immunotherapy protected mice from all three species by reducing the growth of fungi in the lung. Analysis of immune cells in the lung indicated that PICLC immunotherapy recruited significantly higher numbers of neutrophils and other myeloid leukocytes earlier than in the untreated control mice.

RESULTS
Immunotherapy with PICLC protects CGD mice against A. fumigatus infection. Since exogenous induction of type I IFN has shown positive results for the experimental treatment of cancer (35) and some infectious diseases (24), we resolved to investigate the efficacy of exogenous induction of type I IFN by PICLC in the setting of IA caused by Aspergillus species which are refractory to antifungal treatment in CGD patients. PICLC immunotherapy significantly improved survival compared to the nontreated controls in mice infected with A. fumigatus (Fig. 1A) and resulted in significant suppression of the fungal burden as measured by lung fungal DNA load throughout the experimental period (Fig. 1B). In histopathology sections taken 3 days postinfection, small lesions were visible throughout the lung in both groups of infected animals. However, fewer infectious foci were observed in the animals which received PICLC immunotherapy (Fig. 1C).
PICLC immunotherapy protects CGD mice against A. nidulans infection. To expand these results to non-fumigatus Aspergillus species, identical PICLC immunother- Type I Interferon and Aspergillosis ® apy was tested in A. nidulans infection. PICLC immunotherapy was also protective in this model, as 80% of the mice in the control group (phosphate-buffered saline [PBS] treated) succumbed to IA within 28 days postinfection, whereas no deaths were observed in the PICLC immunotherapy group during the same period ( Fig. 2A). Fungal DNA loads were substantially higher in the lungs of the control group mice than the PICLC-dosed mice, suggesting that growth of A. nidulans in host tissue was hampered by PICLC immunotherapy (Fig. 2B). While fewer lung infectious foci were observed 3 days postinfection in PICLC immunotherapy mice than in control mice, individual foci were similarly sized and contained similar germinated fungal elements ( Fig. 2C and D, top row). By 15 days postinfection, foci had matured into granulomas in both PICLC immunotherapy and control mice, but fungal elements were noticeably smaller and fewer in lung granulomas of PICLC immunotherapy mice than in infected-untreated controls (Fig. 2D, bottom row).

PICLC immunotherapy protects CGD mice against A. tanneri infection.
A. tanneri is a novel species noted for its high baseline resistance to antifungal drugs (10), isolated from two fatal cases of IA in CGD patients. While infection kinetics were different between A. tanneri and the two abovementioned Aspergillus species, PICLC immunotherapy was nonetheless protective against A. tanneri. As shown in Fig. 3A, all mice in the control group succumbed to IA in 90 days, whereas 100% of the mice dosed with PICLC were still alive at 90 days postinfection. Fungal DNA loads were also substantially higher in the lungs of the control group mice than in the PICLC-treated mice (Fig. 3B). Histology confirmed the presence of granulomas at 30 days postinfection in the lungs of control group mice, but fewer and smaller lesions were observed in the PICLC-dosed mice (Fig. 3C). Thus, PICLC immunotherapy was protective against multiple Aspergillus species.
Type I IFN is briefly induced by PICLC in the lungs of CGD mice. To confirm that PICLC was able to induce type I IFN in CGD mice, IFN protein was measured following Type I Interferon and Aspergillosis ® a single dose of PICLC. As shown in Fig. 4, there was significant expression (P value Ͻ 0.05) of IFN-␣ and IFN-␤ from 6 to 30 hours postinjection, which then returned to near baseline by 48 h postinjection. These data indicate that CGD mice mount a type I IFN response to PICLC exposure and that this response lasts for approximately 2 days following injection.
PICLC immunotherapy induces significant leukocyte recruitment in the lungs. Since the IFN response to PICLC lasts approximately 2 days following injection (Fig. 4) coupled with effective protection from IA (Figs. 1 to 3), we hypothesized that PICLC immunotherapy results in some immunologic change that impacts the early events of Aspergillus infections. To characterize the effects of PICLC on these early immune events, leukocytes were isolated from the lungs of A. nidulans-infected or control mice with or without PICLC dosing at the time of infection (0 days postinfection) or 1 and 3 days postinfection. Full leukocyte recruitment into the lungs was monitored utilizing a fluorescein isothiocyanate (FITC)-conjugated CD45-specific antibody injected into the mice intravenously 3 to 5 min prior to euthanasia. This technique marks vascular cells with FITC by flow cytometry. In samples taken 24 h after a single dose of PICLC and before infection (day 0), a significant influx of immune cells was observed compared to control samples (Fig. 5A, 0 days postinfection). Histopathological analysis of similarly dosed samples shows that these recruited leukocytes are especially notable as infiltrates accumulating near airways ( Fig. 6A and C) and as newly recruited cells near endothelium (Fig. 6B). It should be noted that this time point represents the condition of the lungs at the time of infection in the PICLC immunotherapy-receiving animals. Immunophenotyping of the lung leukocyte populations showed that PICLC results in the recruitment of several immune cell types. Monocyte-derived myeloid cells and neutrophils dominated this infiltrate, but increased numbers of blood monocytes, NK cells, and plasmacytoid dendritic cells were also observed in response to PICLC (Fig. 5B).
Interestingly, the numbers of recruited CD45 ϩ leukocytes and of most individual immune cell types were converging in the lungs of infected mice either with or without PICLC immunotherapy at 1 and 3 days postinfection ( Fig. 5A and B). This is consistent with similar periendothelial accumulation of recruited cells (Fig. 6B) in the two mouse groups. Thus, A. nidulans infection resulted in a robust recruitment of immune cells between 24 and 72 hours postinfection in the mice not receiving immunotherapy, but these mice were unable to control the fungus and eventually succumbed to the infection (Fig. 2). While PICLC-dosed mice recruit essentially the same types of cells, the timing of the recruitment was earlier and correlated with protection (Fig. 2). Importantly, Aspergillus species go through a gradual morphological change in this period from dormant spores (conidia) to developing hyphae. Thus, we hypothesized that this early recruitment allows immune cells to act on conidia, which are more susceptible to the antifungal activities mustered by CGD hosts (36).
Protection by PICLC immunotherapy is significantly diminished in CGD mice infected with germinated conidia. To test the specificity of PICLC immunotherapy for the conidial stage of IA, we compared the differences between CGD mice infected with resting conidia and with conidia germinated for 5 h (germlings) of A. nidulans (Fig. 7A). We hypothesized that infection with germlings would bypass whatever susceptibility exists in the conidial stage and that early prerecruited leukocytes would be less efficient in eradicating germlings, rendering PICLC immunotherapy less protective. In Fig. 7B, CGD mice infected with the germinated A. nidulans conidia were only partially protected by PICLC immunotherapy, whereas mice infected with resting conidia were completely protected. This indicates that a significant portion of PICLC immunotherapeutic protection is attributable to the ability of prerecruited immune cells to suppress conidia.
Neutrophils are required for PICLC immunotherapy-mediated protection from IA. We observed significantly more neutrophils in PICLC immunotherapy mice than in control mice at experimental day 0 (after receiving one dose of PICLC but before infection), but no difference was detected between the treated and untreated infected mice at days 1 and 3 postinfection (Fig. 5B). Infiltration of neutrophils was also noted by histopathology, gathering on the parenchymal side of the endothelium, near airways, and in the alveoli of PICLC-dosed mice at day 0 ( Fig. 6C and D, green-circled  cells). As neutrophils have a well-established role in IA infection and were highly Type I Interferon and Aspergillosis ® recruited by PICLC, we probed the role of neutrophils in PICLC-mediated protection utilizing monoclonal antibody (MAb) depletion. The 1A8 monoclonal antibody, administered to the mice prior to infection, efficiently depleted neutrophils, resulting in a 90% reduction in recruited neutrophils 1 day postinfection (Fig. 8A). This depletion of neutrophils resulted in significantly higher day 3 fungal burdens than in those mice receiving PICLC immunotherapy alone (Fig. 8B). These data suggest that neutrophils in the lung may play a protective role in PICLC immunotherapy.

DISCUSSION
PICLC immunotherapy was effective in reducing mortality, pathology, and fungal loads in the CGD model of murine IA for all three Aspergillus species tested. Prophylactic PICLC immunotherapy recruited leukocytes so that immune cells were present in the lungs in significant numbers and localized specifically in the alveoli preceding the arrival of the infectious conidia. However, a similar set of leukocytes was recruited in mice infected with Aspergillus alone with a delayed kinetics compared to PICLC immunotherapy. While we have not excluded the importance of other types of immune cells, we observed neutrophils to be the critical cells for mediating this protection based on their presence in the alveoli subsequent to PICLC dosing and the loss of protection upon depletion of these cells. Several studies have demonstrated the critical role that neutrophils play in host defense against Aspergillus (reviewed in references 37 to 41). Our data support the hypothesis that PICLC immunotherapy induces very early recruitment of neutrophils, which destroy a significant fraction of the fungus before it can fully differentiate into hyphae.
We suggest that the earliest stages of Aspergillus infection are a window of opportunity for CGD hosts where efficient induction of immunity can limit infection because our results also indicated that the in vitro-differentiated germlings were more refractory to PICLC immunotherapy than conidia (Fig. 7B). While CGD neutrophils are defective in hyphal killing due to their deficiency in phagosomal oxidase activity, recent studies have shown that these cells do retain killing activity against Aspergillus conidia via nonoxidative pathways specifically involving CD11b/CD18 integrin and phosphatidylinositol 3-kinase (PI3K)-and lactoferrin-mediated iron depletion (36,40,42). However, the Aspergillus factors which contribute to this conidial susceptibility to nonoxidative killing remain uncharacterized. Additionally, the most susceptible stage of Aspergillus may not be conidia but rather the early stages of conidium germination, because lactoferrin-mediated iron depletion arrests the growth of Aspergillus at the germling stage but not at the initiation of germination (42). Ultimately, the factors that result in  Type I Interferon and Aspergillosis ® the susceptibility of Aspergillus conidia to nonoxidative killing remain an exciting avenue of future study, especially since manipulation of any putative susceptibility factors may itself become a possible therapeutic strategy for IA.
We consider our results as having important implications for Aspergillus pathobiology. Following deposition of Aspergillus conidia into the lung, a fraction of these conidia are destroyed by the relatively numerous resident alveolar macrophages (AMs) (39,43,44) and the relatively scarce neutrophils that are resident in the lungs or are recruited in the first few hours of infection (36,38,42). However, some conidia escape these initial defenses (Fig. 9, i and ii). These conidia swell and germinate in response to environmental cues in the first 5 to 12 h of an infection (Fig. 9B). In humans and mice with normal immune function, neutrophils kill hyphal Aspergillus through ROS production (36,40), which stifles the infection subclinically (Fig. 9, iv). Neutrophils from CGD patients and CGD model mice, however, are unable to kill germinated Aspergillus (36,40,42). This allows the progressive growth of fully germinated vegetative hyphae (Fig. 9, v), which is associated with morbidity. PICLC prophylaxis alters this trajectory in CGD mice as immunotherapy recruits the immune cells, especially neutrophils, into alveoli, which intercept a significant fraction of conidia (Fig. 9, iii). The few remaining conidia in PICLC immunotherapy lungs do seem to germinate normally and generate similar infectious foci (Fig. 2C) but at a greatly reduced frequency per animal (Fig. 1C  and 2C), eventually resulting in reduced overall fungal loads (Fig. 9, vi) and a significantly better survival outcome of the PICLC-treated mice.
The mechanisms and shortcomings of IA immunotherapy elucidated in this study have important implications for the conception and development of future IA-targeted immunotherapies. These data and others (36,39,40,42) clearly separate early IA into two major immunologic stages roughly corresponding to the dominant fungal morphology. We suggest that effective immunotherapy for IA will require enhancing immunity during one, or both, of these stages, probably by addressing the immunologic deficiencies of each stage. During the initial stage of infection, Aspergillus conidia are killed via nonoxidative killing by macrophages and neutrophils in both healthy and CGD patients (36). A clear limitation of immunity during this stage is the relatively small population of phagocytes . Several alveoli are depicted, with each alveolus representing a condition examined in this work. In non-CGD lungs, a fraction of conidia are killed by alveolar macrophages, leaving some conidia to germinate (i), but upon recruitment of fully fungicidal neutrophils, the infection is contained in a reactive oxygen species-dependent manner (iv). In CGD lungs, initial stages proceed similarly (ii). These infectious foci recruit many neutrophils (v), but due to their deficiency in ROS production, these cells cannot contain the infection and contain many hyphae. PICLC prophylaxis alters the initial stages in CGD lungs in that the prerecruited neutrophils drastically reduce the initial infectious load at the conidial stage (iii). The few fungal elements that do survive and germinate in PICLC-treated CGD lungs do seem to develop into infectious foci (vi), but the overall infection is nonlethal. Seyedmousavi et al. ® present in unperturbed lungs (Figs. 5 and 6). Therefore, putative immunotherapeutic approaches which aim to act on this first stage need to result in the enhanced recruitment of neutrophils and perhaps other phagocytes capable of killing conidia. Our data predict that significant numbers of leukocytes need to be present within the first few hours of infection to be efficacious, so successful strategies which target this stage will probably require prophylactic dosing, such as demonstrated here by PICLC. Once the Aspergillus germinates into full hyphae, generation of reactive oxygen is required to kill the mold. So, while there are plenty of local neutrophils at this stage, CGD neutrophils are incapable of fungal control. This suggests that an immunomodulatory strategy that targets this later stage would need to enhance hyphal killing but not necessarily improve recruitment. An example of immunotherapy aimed at this later stage is recombinant IFN-␥ treatment (described clinically in reference 18), which improves activity of neutrophils against Aspergillus hyphae in vitro (45). While the evidence presented in this study strongly supports PICLC immunotherapy acting during the conidial stage, we cannot rule out some additional PICLC effects which act later at the hyphal stage. In fact, while histological analysis of infected lungs early after infection shows similarly composed infectious foci with and without PICLC immunotherapy (Fig. 2C), later time points show markedly less fungal material in granuloma cores from PICLC-dosed mice than from untreated groups (Fig. 2D). Furthermore, infection with germlings only partially reversed PICLC-mediated protection (Fig. 7). Together, these results suggest that PICLC immunotherapy induces some additional antihyphal activity in CGD mice, perhaps similar to that of IFN-␥ (45). Type I IFN has been shown to positively influence several potentially relevant neutrophil activities in other models such as direct cytotoxicity (46), neutrophil extracellular trap formation (47,48), and maximal neutrophil differentiation allowing full release of cytotoxic factors (47), reinforcing the plausibility of PICLC immunotherapy having some activity during the hyphal stage. Future studies are therefore necessary to test whether any PICLC-mediated antihyphal activities exist and whether they show efficacy against IA in vivo.
Overall, while the data presented here provide some insight into the generalized parameters for IA immunotherapies, we believe that the efficacy shown here by PICLC merits further development as a possible clinical prophylactic agent for patients at high risk of IA. While prophylactic regimens can be burdensome, several prophylactic strategies are standard practice in certain immunocompromised populations, including CGD patients who are at risk for IA. PICLC immunotherapy may also have some distinct advantages over existing antifungal strategies (summarized in reference 13). Prophylactic antifungal chemotherapy can be effective against susceptible organisms, but resistant organisms are unaffected, and despite the common prophylactic regimens, a significant clinical IA burden remains (14,15). Because immune therapies, such as PICLC, function by entirely different mechanisms, they can be effective even against antifungal-resistant organisms. As proof of this principle, here we show the efficacy of PICLC immunotherapy against two such azole-resistant species, A. nidulans and A. tanneri (Figs. 2 and 3). Furthermore, PICLC may be less costly than recombinant IFN-␥ (13), as polymer compounds are typically less expensive to manufacture than recombinant proteins. In some situations, type I IFN can lead to less immunopathology than IFN-␥ (49). Relatedly, long-term PICLC dosing was well tolerated in mice (24). Finally, we showed that PICLC immunotherapy is effective in the mouse A. tanneri IA model, suggesting that this treatment may be useful in preventing IA caused by Aspergillus species which are refractory to IFN-␥ treatment (10). Thus, while it is too early to speculate about the efficacy of PICLC immunotherapy compared to other prophylactic regimens, the data presented here demonstrate clear efficacy in the CGD model of IA and suggest further study of this promising immunotherapy.

MATERIALS AND METHODS
Fungal strains. Three Aspergillus species isolated from IA cases were used for in vitro and in vivo studies, including Aspergillus fumigatus B-5233 (50) and A. nidulans M24 and A. tanneri NIH 1004 (10), all three strains being isolated from CGD patients. The isolates were stored in 10% glycerol at Ϫ80°C and were revived on malt extract agar (MEA) for 5 to 7 days at 37°C. All isolates were freshly cultured on MEA for 5 to 7 days at 37°C before preparation of the inoculum.
Type I Interferon and Aspergillosis ® tific, Waltham, MA) and then stained using CD68-PerCP-Cy5.5 for the intracellular antigen. Analysis of stained cells was performed with a BD Fortessa flow cytometer using FACSDiva software (BD Biosciences, USA). Compensation settings were acquired using OneComp beads (eBioscience, Thermo Fisher Scientific, Waltham, MA) for the antibody channels and ArC amine-reactive beads (Molecular Probes, Eugene, OR) for the viability stain. Analysis of flow cytometry data utilized FlowJo software (BD Biosciences, USA). The gating strategy for flow cytometry analysis is shown in Fig. S1.
In vivo depletion of cell subsets. Neutrophils were depleted using anti-Ly6G monoclonal antibody (MAb) 1A8 (Bio X Cell, West Lebanon, NH). Antibodies were administered to groups of 5 CGD mice by intraperitoneal injection on days Ϫ6, Ϫ4, and Ϫ1 preinfection and day ϩ1 postinfection. Injections comprised 500 g per mouse in 200 l. Depletion efficiency was monitored using flow cytometry as described above.
Ethics statement. The Institutional Animal Care and Use Committee of the National Institute of Allergy and Infectious Diseases approved all animal studies (approval no. LCIM-5E). Studies were performed in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (55).
Statistical analysis. All data analyses were performed using GraphPad Prism, version 7 (GraphPad Software, San Diego, CA). Mortality data were analyzed by the log rank test. Student's t test and two-way analysis of variance (ANOVA) followed by multiple-comparison test were used to define whether there are significant differences between indicated samples. Statistical significance for comparisons was defined as a P value of Յ0.05 (two-tailed), which was reported as follows: nonsignificant (NS), P Ͼ 0.05; *, P Յ 0.05; **, P Յ 0.01; ***, P Յ 0.001.