Deficiency of lung-specific claudin-18 leads to aggravated infection with Cryptococcus deneoformans through dysregulation of the microenvironment in lungs

Cryptococcus deneoformans is an opportunistic fungal pathogen that infects the lungs via airborne transmission and frequently causes fatal meningoencephalitis. Claudins (Cldns), a family of proteins with 27 members found in mammals, form the tight junctions within epithelial cell sheets. Cldn-4 and 18 are highly expressed in airway tissues, yet the roles of these claudins in respiratory infections have not been clarified. In the present study, we analyzed the roles of Cldn-4 and lung-specific Cldn-18 (luCldn-18) in host defense against C. deneoformans infection. luCldn-18-deficient mice exhibited increased susceptibility to pulmonary infection, while Cldn-4-deficient mice had normal fungal clearance. In luCldn-18-deficient mice, production of cytokines including IFN-γ was significantly decreased compared to wild-type mice, although infiltration of inflammatory cells including CD4+ T cells into the alveolar space was significantly increased. In addition, luCldn-18 deficiency led to high K+ ion concentrations in bronchoalveolar lavage fluids and also to alveolus acidification. The fungal replication was significantly enhanced both in acidic culture conditions and in the alveolar spaces of luCldn-18-deficient mice, compared with physiological pH conditions and those of wild-type mice, respectively. These results suggest that luCldn-18 may affect the clinical course of cryptococcal infection indirectly through dysregulation of the alveolar space microenvironment.

Mice. Male or female mice at 6 to 8 weeks of age and 16 to 24 g of weight were used in the analyses. Mice were randomly allocated to the various experimental groups. All mice were kept under specific pathogen-free conditions at the Institute for Animal Experimentation, Tohoku University Graduate School of Medicine; the Osaka University Laboratory of Biological Science, Graduate School of Frontier Biosciences and Medicine; or RIKEN Kobe branch. The conditions of the breeding room were as previously described 11,30-35 . Inoculation with Cryptococcus deneoformans. A

Preparation of bronchoalveolar lavage fluids and lung interstitial homogenates.
Mice were sacrificed at various timepoints after infection. Bronchoalveolar lavage fluids (BALFs) were prepared as previously described 34,35 . After the BALFs were collected, lung homogenates were prepared as previously described 11,30-33 . Enumeration of viable C. deneoformans. Evaluation of the fungal burdens in the lungs and brains were carried out as previously described 11,[30][31][32][33] . Briefly, the lungs and brains were homogenized separately in 5 ml (lungs) or 1 ml (brains) of distilled water through a stainless-steel mesh at room temperature. In some experiments, BALFs and lung homogenates after BAL were analyzed separately in 3 ml and 5 ml, respectively, of distilled water. These samples were diluted, inoculated on PDA plates, and cultured before the resulting colonies were counted.
Histological examination. Histopathological specimens of the lungs were obtained as previously described 11,[30][31][32][33][34][35]  Cytokine assay. BALFs were collected in 1 ml PBS. After BAL, the lungs were excised and then homogenized separately in 5 ml PBS through a stainless-steel mesh. After centrifugation, the supernatants were collected and stored at − 70 °C before use. Concentrations of each cytokine in BALF and lung homogenates were measured using an ELISA kit as previously described 11,[30][31][32][33][34][35]  www.nature.com/scientificreports/ Measurement of ion concentration and pH in BALFs. BALFs were collected in 1 ml saline from 8-week-old mice under deep anesthesia. Na + and K + ion concentrations in BALFs were directly measured using Na + and K + compact ion meters (Horiba, Kyoto, Japan) according to the manufacturer's instructions. Cl − ion concentrations in BALFs diluted with MQ water were measured using the QuantiChrom Chloride Assay Kit (BioAssay Systems, Hayward, CA, USA). The electrical potentials were measured using the LAQAtwin compact pH meter (Horiba, Kyoto, Japan) according to the manufacturer's instructions. A standard curve was prepared from the electrical potential of saline adjusted to pH 4, 7, and 9 with HCl or NaOH. The pH values of the BALF samples were calculated from this standard curve after the measured values were corrected for the electrical potential of saline.
In-vitro culture of C. deneoformans. C. deneoformans was cultured on PDA plates for two to three days before use. The yeast cells were suspended in RPMI 1640 medium with 20 mM HEPES and without NaHCO 3 (Sigma-Aldrich, St. Louis, MO, USA), prepared at 1 × 10 4 cells/ml, and cultured at 37 °C for 24 h at different pH values (6.8 and 7.3). The number of organisms was counted using a hemocytometer. The yeast cells were concentrated onto microscope slides using a cytospin, stained with PAS (Muto Pure Chemicals, Tokyo, Japan), and observed under a microscope. The budding ratio of each PAS-stained specimen was estimated as the number of budding yeast cells out of 200 cells.
Cell division assay. C. deneoformans was cultured on PDA plates for two to three days before use. The yeast cells were stained with 50 μM carboxyfluorescein succinimidyl ester (CFSE) for 30 min in 10 ml PBS at 30 °C, and washed three times in PBS to remove excess CFSE. CFSE-labeled C. deneoformans was used for in vitro culture and in vivo infection by the method described above. The mean fluorescent intensity (MFI) of CFSE in the cultured yeast cells or CD45 − cells in BALFs after infection was analyzed using a BD FACS Canto II flow cytometer.
Statistical analysis. Data were analyzed using JMP Pro 11.2.0 software (SAS Institute Japan, Tokyo, Japan).
Data are expressed as mean ± SD. Differences between groups were examined for statistical significance using Welch's t-test. A p value of less than 0.05 was considered significant.

Effect of claudin deficiencies on host protection and inflammatory response against cryptococcal infection.
To explore the effects of Cldn-4 and luCldn-18 deficiency on host defense against cryptococcal infection, Cldn-4-deficient mice, luCldn-18-deficient mice, and control mice for each claudin-deficient genotype were infected with C. deneoformans, and the growth of this fungal pathogen in the lungs and brains was examined. After cryptococcal infection, mouse body weights were not lower in Cldn-4 −/− mice compared with Cldn-4 +/+ mice (Fig. 1a). In addition, the fungal burdens in the lungs were not significantly different between Cldn-4 +/+ and Cldn-4 −/− mice on day 14 post-infection (Fig. 1b). In contrast, the lung and brain burdens of yeast cells were higher in luCldn-18 −/− mice than in luCldn-18 +/+ mice on days 14 and 28 post-infection ( Fig. 1c-e).
We also conducted a histological analysis to determine how luCldn-18 deficiency affected the inflammatory response in the lungs after cryptococcal infection. Defective alveolar formation and increased alveolar macrophage counts were observed in uninfected luCldn-18 −/− mice compared to luCldn-18 +/+ mice (Fig. 2a). In addition, massive multiplication of yeast cells with poor granulomatous responses was observed in luCldn-18 −/− mice, whereas luCldn-18 +/+ mice showed markedly fewer yeast cells with granulomatous responses (Fig. 2b, c).
To examine the localization of the fungus in the lungs, we compared the numbers of live C. deneoformans colonies in BALFs (alveolar space) and lung homogenate after BAL (interstitial space) between luCldn-18 +/+ and luCldn-18 −/− mice. The fungal burdens in both BALFs and lung tissue after BAL were greater in luCldn-18 −/− mice than in luCldn-18 +/+ mice on days 1, 3, 7, and 14 post-infection ( Fig. 3a, b). In particular, the increase in the fungal burden in the alveolar space of luCldn-18 −/− mice was remarkable on days 3, 7, and 14 post-infection (Fig. 3c).

Effect of luCldn-18 deficiency on the immune response after infection with C. deneoformans. During infection with various microorganisms, an inflammatory response is induced, increasing total
protein concentrations and inflammatory cell counts 36 . The next series of experiments examined the effect of luCldn-18 deficiency on the immune response to C. deneoformans. Total protein concentrations and cell counts in BALFs were significantly greater in luCldn-18 −/− mice than in luCldn-18 +/+ mice on days 0, 1, 3, 7, and 14 post-infection (Fig. 4a, b).
The distributions of various cell types in the BALF were analyzed using flow cytometry on day 3 post-infection (when cell counts peaked). The numbers of neutrophils, alveolar macrophages, inflammatory monocytes, NK, CD4 + T, CD8 + T, and NKT cells were significantly greater in luCldn-18 −/− mice than in luCldn-18 +/+ mice (Fig. 4c). Host defense against cryptococcal infection is largely regulated by the balance between the Th1 and Th2 immune responses [2][3][4][5][6][7] . Our previous study showed that IL-17A is involved in the negative regulation of local host defenses against C. deneoformans infection through suppression of the Th1 response 11 . Here, therefore, we examined the effect of luCldn-18 deficiency on cytokine production in response to C. deneoformans infection by measuring cytokine concentrations in BALFs and lung homogenate after BAL on days 3, 7, and 14 post-infection. Production of IFN-γ in BALF on days 3 and 7 and in lung homogenate after BAL on day 14 were significantly lower in luCldn-18 −/− mice than in luCldn-18 +/+ mice (Fig. 4d, e). Similarly, production of IL-4 and IL-13 in BALF on day 14 and in the lungs on days 3 and 14 were significantly lower in luCldn-18 −/− mice than in luCldn-18 +/+ mice www.nature.com/scientificreports/ ( Fig. 4d, e). In addition, production of IL-17A in BALF on day 7 and in the lungs on day 14 were significantly lower in luCldn-18 −/− mice than in luCldn-18 +/+ mice (Fig. 4d, e).

Effect of luCldn-18 deficiency on homeostasis of the lung microenvironment. luCldn-18 is
involved in alveolar fluid homeostasis by regulating solute and ion permeability between alveolar epithelial cells 27 . The next series of experiments examined the effect of luCldn-18 deficiency on the homeostasis of the lung microenvironment. In a steady state, K + ion concentration in BALFs was significantly higher in luCldn-18 −/− mice than in luCldn-18 +/+ mice, while pH was significantly lower (Fig. 5a). Airway surface liquid pH can be altered by disease and lung inflammation, in part because airway inflammation promotes acidification 37 . Alveolus acidification was observed after cryptococcal infection in both genotypes (Fig. 5b). In addition, alveolar pH remained significantly lower in luCldn-18 −/− mice than in luCldn-18 +/+ mice until at least seven days after infection (Fig. 5b).
Effect of luCldn-18 deficiency on fungal growth. Cryptococcal replication has been reported to be enhanced at acidic pH as opposed to physiological pH 38,39 . In this study, similarly, the numbers of yeast cells and the budding rate after in vitro culture were significantly greater at BALF's pH condition of luCldn-18 −/− mice (pH 6.8) than in those of luCldn-18 +/+ mice (pH 7.3) (Fig. 6a, b). In addition, in the cell division assay using CFSE, CFSE labeling of yeast cells after in vitro culture, which is correlated with cell propagation, was significantly lower at pH 6.8 than at pH 7.3 (Fig. 6c, d). In our in vivo infection experiment, likewise, CFSE expression on CD45 − cells in BALFs after cryptococcal infection was significantly lower in luCldn-18 −/− mice than in luCldn-18 +/+ mice (Fig. 6e, f), suggesting that cryptococcal replication was increased at the acidic pH that occurs in luCldn-18 −/− mice as opposed to the physiological pH that occurs in luCldn-18 +/+ mice.

Discussion
TJs function as the primary structural component controlling paracellular permeability and as barriers in the spaces between the alveolar epithelial cells 40 , yet their contribution to respiratory infection has not been thoroughly analyzed. Cldn-4 and luCldn-18 are among the components of airway TJs 22,23 . In the present study, luCldn-18 deficiency increased susceptibility to pulmonary C. deneoformans infection, while Cldn-4 deficiency did not affect fungal clearance. TJs assemble through homotypic and heterotypic cis-and trans-interactions between claudins in paracellular localization 41,42 . In lung epithelial cells, Kage et al. found that Cldn-4 deficiency did not alter the expression of other Cldns (Cldn-3, 5, 7, and 18) or ion permselectivity at steady state 24 .
In alveolar epithelial cells, in contrast, Li et al. demonstrated that luCldn-18 deficiency increased expression of Cldn-3 and Cldn-4 in whole lungs and ion permeability at steady state 27 . In addition, impaired alveolarization was observed in luCldn-18-deficient mice 28 . In the current study, likewise, luCldn-18 deficiency led to altered expression of other Cldns (Cldn-3 and 5) in the lungs, with increased K + ion concentration, decreased pH in BALF, and impaired alveolar formation. These data suggest that luCldn-18 may have an important role in maintaining homeostasis of the microenvironment within alveolar spaces and that its impairment may increase susceptibility to cryptococcal infection. luCldn18 deficiency increased fungal burden not only in the lungs but also in the brain. The blood-brain barrier (BBB) protects the central nervous system (CNS) by restricting the passage of molecules and microorganisms 43 . It has been suggested that C. deneoformans may enter the blood or brain from the lung through a Trojan horse mechanism, crossing the BBB as a passenger inside host phagocytes 44 . The increased fungal dissemination into the brain in luCldn-18-deficient mice may be related to pressure changes due to the increased fungal burden in the lungs. Berndt et al. have reported that Cldn-5, 11, 12, and 25 are expressed in murine brain capillaries 45 , and some pathogens are known to infiltrate the CNS by regulating the expression www.nature.com/scientificreports/ of these Cldns 46 : human immunodeficiency virus type 1 (HIV-1), for example, has been shown to disrupt BBB integrity via modification of Cldn-5, thereby allowing HIV-1 to enter the brain 47 . Group B streptococcus disrupts the paracellular pathway of the BBB by downregulating claudin-5 via upregulation of Snail1 and enhances bacterial invasion into the CNS 48 . Yet the effect on fungal passage of the expression of these Cldns within the brain has not been analyzed. Since the present study did not analyze whether the expression of Cldns in the brain was altered in luCldn-18-deficient mice, further investigations are required to address the possibility that the passage of fungus, including phagocytosed fungus, might be enhanced in luCldn-18-deficient mice due to altered Cldn expression patterns in their brains.
In luCldn-18-deficient mice, the increased population of C. deneoformans was more remarkable in the alveolar space than in the lung interstitium. In addition, acidification in the alveolar spaces of luCldn-18-deficient mice was observed during infection with this fungal pathogen. In previous studies, stCldn-18-deficient mice have been found to have lower pH in the stomach due to gastric acid leakage, leading to gastritis and gastric tumorigenesis 18,49,50 . Alveolar macrophages, which constitute a sizable fraction of alveolar subphase fluids, are a substantial source of metabolic H + production under physiological conditions 37 . In the current study, the number of alveolar macrophages was significantly greater in luCldn-18 −/− mice than in luCldn-18 +/+ mice after cryptococcal infection, suggesting that acidification in the alveolar spaces of luCldn-18-deficient mice may be caused by increased production of metabolic H + by alveolar macrophages. In addition, the enhanced cell destruction of increased numbers of macrophages and airway epithelial cells may have induced high K + ion concentrations in alveolar subphase fluids, although our limited analysis of this matter prevents us from drawing firm conclusions.
A Th1-mediated immune response contributes toward the eradication of C. deneoformans through inducing macrophage activation and accelerating the formation of granuloma at the infection sites [6][7][8][9] , whereas the Th2-mediated immune response counteracts these host protective responses, worsening infection 6,7,10 . In the present study, the production of Th1-related cytokines such as IL-12p40 and IFN-γ was significantly decreased in luCldn-18 −/− mice compared to luCldn-18 +/+ mice after cryptococcal infection. Unexpectedly, however, Th2 cytokines such as IL-4 and IL-13, which suppress host defense against cryptococcal infection by inhibiting macrophage activation and enhancing mucin production 6,7,10,30 , were not increased but rather also decreased in luCldn-18 −/− mice compared to luCldn-18 +/+ mice. In addition, the synthesis of IL-17A, which is involved in the negative regulation of local host defense against C. deneoformans infection 11 , was significantly lower in luCldn-18 −/− mice than in luCldn-18 +/+ mice. Thus, the changes in production of host defense-related cytokines were not consistent with the impaired fungal elimination observed in luCldn-18 −/− mice. Similar inconsistent results were observed in the infiltration of inflammatory cells including CD4 + T cells into the alveolar spaces, which was drastically increased in luCldn-18 −/− mice compared to luCldn-18 +/+ mice. There are no reports of a single luCldn-18 deficiency leading to increasing paracellular cell transport. In the current study, luCldn-18 deficiency induced Cldn expression changes, but the expression of many TJ proteins, including occludin, was maintained. Therefore, the possibility of increased mechanical paracellular transport through alveolar epithelial cells is thought to be low. Previously, Leblebicioglu and co-workers reported that chemotactic responses of human peripheral blood polynuclear leukocytes were significantly increased under decreased pH conditions in in vitro cultures 51 . Lower pH induces activation of monocytes/macrophages and dendritic cells but suppresses T cell immune responses 36,52,53 . Higher levels of extracellular K + ions are reported to suppress the activation of T cell-mediated immune responses 54,55 , which is also observed under the luCldn-18-deficient condition in the current study. These findings suggest that acidification in the alveolar spaces of luCldn-18 −/− mice may promote  www.nature.com/scientificreports/ the infiltration of inflammatory cells into the alveolar space while also suppressing the T cell immune response to C. deneoformans through acidification and increased levels of K + ions.
In the current study, fungal replication of C. deneoformans was significantly enhanced both in acidic culture conditions in vitro and in the alveolar spaces of luCldn-18-deficient mice in vivo compared with physiological pH and control mice, respectively. Several investigators have demonstrated that acidic conditions enhance cryptococcal growth while alkaline conditions inhibit it 37 neoformans phenoloxidase activity, which has long been known as a virulence factor 58 . Collectively, these results suggest that acidification in the alveolar spaces of luCldn-18 −/− mice may not only suppress T cell activation but also enhance the growth of the fungal pathogen C. deneoformans.
Diseases in which the expression of luCldn-18 changes are not well known, such as diabetes mellitus, hematological diseases and collagen diseases, are known to be risk factors for cryptococcosis in non-HIV patients 59 . In addition, luCldn-18 may be impaired by inflammatory responses and tissue damage induced after cryptococcal infection, but the present study did not analyze whether the expression of luCldn-18 in the lungs was decreased after infection. Analysis of luCldn-18 gene mutations and expression in high-risk patients and luCldn-18 injury mechanisms involved in cryptococcal infection may be useful in finding therapies for infection with this pathogen.
In conclusion, the present study demonstrates that luCldn-18 deficiency probably suppresses T cell activation during cryptococcal infection through acidification and increased extracellular K + ion concentration in the alveolar spaces. As C. deneoformans has greater replication efficiency in lower pH conditions, the acidification of the alveolar spaces may increase the lung burden of this fungal pathogen in luCldn-18 −/− mice. Thus, luCldn-18 may be indirectly involved in limiting the progression of cryptococcal infection in the lungs through maintaining