Orientia tsutsugamushi selectively stimulates the C-type lectin receptor Mincle and type 1-skewed proinflammatory immune responses

Orientia tsutsugamushi is an obligately intracellular bacterium and the etiological agent of scrub typhus. The lung is a major target organ of infection, displaying type 1-skewed proinflammatory responses. Lung injury and acute respiratory distress syndrome are common complications of severe scrub typhus; yet, their underlying mechanisms remain unclear. In this study, we investigated whether the C-type lectin receptor (CLR) Mincle contributes to immune recognition and dysregulation. Following lethal infection in mice, we performed pulmonary differential expression analysis with NanoString. Of 671 genes examined, we found 312 significantly expressed genes at the terminal phase of disease. Mincle (Clec4e) was among the top 5 greatest up-regulated genes, accompanied with its signaling partners, type 1-skewing chemokines (Cxcr3, Ccr5, and their ligands), as well as Il27. To validate the role of Mincle in scrub typhus, we exposed murine bone marrow-derived macrophages (MΦ) to live or inactivated O. tsutsugamushi and analyzed a panel of CLRs and proinflammatory markers via qRT-PCR. We found that while heat-killed bacteria stimulated transitory Mincle expression, live bacteria generated a robust response in MΦ, which was validated by indirect immunofluorescence and western blot. Notably, infection had limited impact on other tested CLRs or TLRs. Sustained proinflammatory gene expression in MΦ (Cxcl9, Ccl2, Ccl5, Nos2, Il27) was induced by live, but not inactivated, bacteria; infected Mincle-/- MΦ significantly reduced proinflammatory responses compared with WT cells. Together, this study provides the first evidence for a selective expression of Mincle in sensing O. tsutsugamushi and suggests a potential role of Mincle- and IL-27-related pathways in host responses to severe infection. Additionally, it provides novel insight into innate immune recognition of this poorly studied bacterium.

Introduction Scrub typhus is a vector-borne febrile illness caused by the obligately intracellular bacterium, Orientia tsutsugamushi. Transmitted via the bite of a larval Leptotrombidium mite (commonly known as chigger), this bacterium infects approximately 1 million people per year in an Asia-Pacific region housing over one-third of the world's population, termed the "tsutsugamushi triangle" [1]. Recent reports have indicated the presence of scrub typhus in areas previously thought free of the disease, including South America and Africa [2,3]. The lung is a major target organ of infection, and mild interstitial pneumonia predominates in self-limiting or appropriately treated cases [1]. However, if left untreated, disease may progress to severe lung damage and acute respiratory distress syndrome in up to 25% of cases [1,4]. While the facets underpinning progression from mild to severe disease remain ill-defined, both bacterial and host factors are speculated to play major roles.
The bacterial factors responsible for scrub typhus pathogenesis remain elusive. O. tsutsugamushi is remarkably unique in that over 40% of its genome consists of repeated DNA sequences, including transposable and massively amplified integrative and conjugated elements, as well as short repetitive sequences [5,6]. These elements have rendered attempts at genetic manipulation unsuccessful, precluding functional genomic studies and the generation of fluorescently tagged "trackable" bacteria [7]. Additionally, this bacterium lacks lipopolysaccharide and conventional peptidoglycan structures, which is a distinguishing feature from the closely related Rickettsia genus [8]. O. tsutsugamushi preferentially replicates within endothelial cells, monocytes, macrophages (MF), and dendritic cells [9,10]. After gaining entrance to host cells via the phagosome or endosome, the bacteria escape to the cytoplasm for replication. Following replication, bacteria disseminate by budding from the host cell cytoplasmic membrane [11]. Endothelial and MF responses to infection have been characterized and include detection of proinflammatory cytokines (IL-1β, TNF-α, IL-8) and activation of transcriptional factor NF-κB [12,13]. The MF response to O. tsutsugamushi has garnered research interest both in terms of being an innate immune responder and as a host cell to infection. In-vitro studies with human primary monocytes/MF have indicated the generation of antivirus-like immune responses shortly after exposure to O. tsutsugamushi, as well as a skewed proinflammatory (M1) phenotype in MF [14]. The biological roles of tissue-specific MF subsets in scrub typhus cases remain unexplored, although studies examining human tissue samples have been reported [10,15].
Animal models of scrub typhus, which mimic pathology observed in human patients, have shed light on possible mechanisms of pathogenesis and tissue-specific immune responses. Our recent studies with lethal O. tsutsugamushi infection in mice have revealed strong, type 1-skewed immune responses in the lung, spleen, liver, kidney, and brain tissues [16,17]. Since the lungs harbor the greatest bacterial burden throughout the course of infection, we recently examined the activation status of MF, neutrophils, and endothelial cells in the lungs of infected mice [18]. We observed a significant influx of monocytes/MFs on day 6 post-infection (onset of disease) and day 10 post-infection (prior to host death), with nearly the entire MF population (~97% cells) skewed towards the M1 phenotype at the terminal phase of disease (day 10), implying that M1 MFs confer ineffective protection and could play a role in tissue injury [18]. In-vitro experiments with murine bone marrow-derived MFs revealed restricted O. tsutsugamushi growth in M1-polarized MFs, but unrestricted growth in naïve and M2-polarized MFs [18]. Collectively, these studies alluded to the double-edged sword of M1 MFs in scrub typhus, suggesting that while M1 polarization contributes to controlling O. tsutsugamushi infection, the dysregulation of this response could lead to indiscriminate tissue damage. However, the initial driving factors for M1 MF polarization remain unknown.
Pathogen pattern recognition receptors (PRRs), including Toll-like receptors (TLRs), nucleotide-binding oligomerization domain-like receptors, retinoic acid-inducible gene-1 like receptors, and C-type lectin receptors (CLRs), are crucial for initiating and shaping immune responses to infection. CLRs orchestrate inflammatory responses via an immunoreceptor tyrosine-based inhibitory motif (ITIM) or immunoreceptor tyrosine-based activation motif (ITAM) in its own cytoplasmic tail, or through coupling with ITIM or ITAM-bearing signaling partners [19][20][21]. Mincle (macrophage inducible C-type lectin; Clec4e or Clecsf9) is expressed mostly on myeloid cells and has been studied extensively for its contribution to MF activation and skewing innate and adaptive responses. Mincle can bind pathogen-associated glycolipids, as well as host damage-associated molecular patterns (DAMPs) such as cholesterol crystals (in humans) and SAP130 (in humans and mice) [22][23][24]. Since Mincle does not harbor an ITIM or ITAM, it relies on Fc receptor gamma (Fcgr; FcγR) to initiate immune signaling cascades [24]. While the functional significance of Mincle activation is variable in different host-pathogen interactions or diseases, Mincle signaling is known to lead to proinflammatory cytokine production, M1 MF phenotype, and a type 1-or Th17-favoring cytokine milieu [23,25,26]. To date, CLRs have been virtually unstudied in the context of obligately intracellular pathogens, and the PRR profile responding to O. tsutsugamushi remains poorly characterized.
In this study, we tested whether CLRs, particularly Mincle, could contribute to the dysregulated type-1 response to O. tsutsugamushi. First, we utilized a lethal infection model in C57BL/ 6 mice to show that Mincle and its signaling partners, Fcgr, were highly differentially expressed in the lung tissues during scrub typhus, with the greatest mRNA and protein levels towards D10 (the terminal phase of disease prior to host death). The Mincle and Fcgr levels positively correlated with the expression levels of proinflammatory cytokines/chemokines and M1 polarization markers. Furthermore, we utilized bone marrow-derived MF from wild-type (WT) and Mincle-deficient (Mincle -/-) mice to demonstrate a selective activation of Mincle and an M1 transcriptional profile, but not other CLRs or type 2 markers, via interaction with live bacteria. To our knowledge, this is the first report defining a CLR response to O. tsutsugamushi infection.

Mouse infection and tissue collection
Female C57BL/6J mice were purchased Jackson Lab or Envigo RMS, Inc. Mincle -/mice on the C57BL/6J background were kindly provided by Dr. Christine Wells (University of Melbourne, Melbourne, Australia) [27]. Animals were bred and maintained under specific pathogen-free conditions. Animals were infected at 8-12 weeks of age and performed in UTMB ABSL3 facilities in the Galveston National Laboratory, and subsequent tissue processing or analysis was performed in BSL3 or BSL2 facilities, respectively. Procedures were approved by the Institutional Biosafety Committee, in accordance with Guidelines for Biosafety in Microbiological and Biomedical Laboratories. The Karp strain of O. tsutsugamushi (OtK) was utilized for all infections. Groups of 5 animals were intravenously infected with the same bacterial stock prepared from Vero cell infection, as described previously [18,28]. Mice were inoculated with a lethal dose of infection (~1.325 x 10 6 viable bacteria, as determined via focus-forming assay) or PBS and monitored daily for weight loss and signs of disease. Lung and brain samples were collected at 2, 6, and 9-10 days post-infection with mock infected animals serving as controls and inactivated for immediate or subsequent analyses. Data shown is representative of three independent repeats.

Infection of mouse bone marrow-derived macrophages (MF)
Bone marrow cells were collected from the tibia and femur of WT or Mincle -/mice and treated with red blood cell lysis buffer (Sigma Aldrich). MF were generated by incubating bone marrow cells at 37˚C with 40 ng/ml M-CSF (Biolegend, San Diego, CA) in complete RPMI 1640 medium (Gibco), as described before [18]. Cell medium was replenished at day 3, and cells were collected at day 7. After collection and quantification, 5 x 10 5 viable cells were seeded into 12-or 24-well plates and allowed to adhere overnight prior to infection. For TNFα-treated MFs, recombinant mouse TNFα (Biolegend) was added 30 min prior to infection at a final concentration of 25 ng/mL [29]. Bacteria were added at a multiplicity of infection (MOI) of 2, 5, or 10 and centrifuged at 2,000 RPM for 5 min to synchronize infection. For experiments utilizing heat-killed bacteria, bacterial stocks were incubated at 56˚C for 30 min [30] and used at a MOI equivalent of 10.

Infection of mouse bone marrow-derived neutrophils
Bone marrow cells were harvested from femur and tibia of naïve mice and treated with red blood cell lysis buffer (Sigma Aldrich). Neutrophils were prepared by using anti-Ly6G magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany); the purity of CD11b + Ly6G + neutrophils was 96%. Cells were seeded in 24-well plates and incubated for 1 hr at 37˚C with 5% CO 2 . The infection dose was 10 MOI; the Vero cell culture supernatant was used as a mock control.
Staining of lung tissues was performed as in our previous reports [18,28]. Briefly, 8-μm frozen sections were taken from mock or lethally infected animals at day 2, 6, and 10. Sections were blocked and incubated with rabbit anti-OtK serum (1:1000) and rat anti-Mincle mAb (1:50, MBL International). Bound antibodies were visualized with Alexa Fluor 488-conjugated anti-rabbit Fab2 and Alexa Fluor 555-conjugated anti-rat IgG (1:50, clones 4412S and 4417S, respectively, Cell Signaling Technology). All samples were counterstained with DAPI (1:5000, Sigma-Aldrich). Staining with secondary antibodies and primary antibodies alone served as negative controls. Slides were imaged at the UTMB Optical Microscopy Core by using the Zeiss LSM 880 confocal microscope (405, 488, and 561 excitation lasers). Acquisition settings were identical among the experimental groups and representative images are presented from each time point.

Nanostring gene expression profiling
Lung samples were stored in RNALater (Ambion, Austin, TX) until extraction was performed. Total RNA was extracted from mock (day 0) or lethally infected lungs collected at day 2, 6, and 10, as well as mock or lethally infected brain tissues at day 10, by using the RNeasy RNA Isolation kit (Qiagen). Total RNA samples (2 mice per group, 200 ng per sample in ribonucleasefree water) were processed at the Baylor College of Medicine Genomic and RNA Expression Profiling Core (Houston, TX). Gene expression profiling was performed by using the nCounter platform and two Nanostring kits: Mouse Immunology Panel comprising 561 genes and 14 housekeeping genes; Mouse Inflammation_v2 Panel comprising 254 genes and 6 housekeeping genes (NanoString Technologies, Seattle, WA). Results from the two kits were pooled after gene expression was normalized to housekeeping gene expression and analyzed following the manufacturer's instructions by using the nSolver Software Version 4 and Advanced Analysis Version 2.0 (NanoString Technologies).

Quantitative reverse transcriptase PCR (qRT-PCR)
To determine host gene expression, mouse lung tissues, MF, and neutrophil cultures were collected in RNALater or Trizol (Ambion) and incubated at 4˚C overnight for inactivation. Total RNA was extracted via RNeasy mini kit (Qiagen), and cDNA was synthesized utilizing iScript cDNA kit (Bio-Rad Laboratories, Hercules, CA). qRT-PCR assays were performed using iTaq SYBR Green Supermix (Bio-Rad) on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad). The assay included: denaturing at 95˚C for 3 min followed with 40 cycles of: 10s at 95˚C and 30s at 60˚C. To check specificity of amplification, melt curve analysis was performed. Transcript abundance was calculated utilizing the 2 -ΔΔCT method and normalized to glyceraldehyde-4-phosphate dehydrogenase (GAPDH). Primers used in qRT-PCR analysis are listed in S1 Table.

Bacterial load analysis (qPCR)
To determine bacterial loads, MF DNA were collected at 4, 24, and 48 hr after infection via a DNeasy kit (Qiagen) and used for qPCR, as described previously [16]. Bacterial loads were normalized to total nanogram (ng) of DNA per μL for each sample. Data are expressed as copy number of 47-kDa gene per ng of DNA. The copy number of the 47-kDa gene was determined by using known concentrations of a control plasmid harboring a single-copy insert of the gene. Sample gene copy numbers were then determined by a serial dilution of the control plasmid.

Flow cytometry
Bone marrow-derived MF (3 x 10 6 ) were cultivated as described above, aliquoted into 50 mL conical tubes (Falcon), and allowed to rest for 1 hr at 37˚C. Cells were then infected with CFSE-labeled O. tsutsugamushi. CFSE-labeling was performed as previously described [31]. Briefly, CFSE (Invitrogen) was mixed with bacterial stocks at a 1:1000 ratio and incubated in dark for 10 min at 4˚C. The reaction was quenched by adding complete RPMI, centrifuged at 20,000 x g for 10 min, and washed twice with PBS prior to addition to MF cultures. At 4 hr of infection, cells were collected and divided equally for surface vs. intracellular staining as previously described [18]. Cells were stained with rat anti-Mincle mAb (MBL International), Alexa-Fluor594-conjugated chicken-anti-rat IgG (Molecular Probes Inc, Eugene, OR), and fixed in 2% paraformaldehyde overnight at 4˚C prior to analysis. Data were collected by a BD LSRFortessa (Becton Dickinson, San Jose, CA) and analyzed by using FlowJo software version 10.7.2 (Becton Dickinson).

Western blot
Proteins from lung tissues and MF were extracted with RIPA lysis buffer (Cell Signaling Technology) and quantified with BCA Protein Assay kit (Thermo Fisher Scientific). Samples were stored at -80˚C until processing. Thawed cell lysates were heated for 10 min at~105˚C in Laemmle buffer (Bio-Rad) containing 2-β-mercaptoethanol, loaded into 4-20% SDS-PAGE gel (Bio-Rad) then transferred onto polyvinylidene difluoride membranes (Bio-Rad). After blocking, membranes were incubated with anti-Mincle (1:500, MBL International) and anti-β-actin (1:2000, Cell Signaling Technology) and anti-rabbit/goat secondary antibodies. Pierce ECL Western Blotting substrate (Thermo Fisher Scientific) was subsequently added to the membranes and light emission was captured using Amersham Imager 680 (GE Healthcare Lifesciences, Upssala, Sweden). Quantification of band intensity was performed by using ImageJ.

Statistical analysis
Gene expression profiling data were presented graphically as mean ± standard error of the mean (SEM) and utilized the Benjamini-Yekutieli procedure to test for significance, yielding adjusted (adj.) p-values. Data thereafter were analyzed using GraphPad Prism software and presented as mean ± SEM. Differences between control and treatment groups were analyzed using one-way ANOVA with Dunnett's multiple comparisons. Statistically significant values are denoted as � p < 0.05, �� p < 0.01, ��� p < 0.001, and ���� p < 0.0001, respectively.

Mincle is highly transcribed in murine lung tissues during terminal infection
Studies with scrub typhus animal models have revealed exaggerated, type 1-skewed immune responses in O. tsutsugamushi-infected lung tissues [16,18,28,32]; however, these studies only examined a selected panel of cytokines/chemokines. In this study, we sought to gain a broader understanding of immune crosstalk by using differential expression analysis. After infection with a lethal dose of O. tsutsugamushi, murine lungs were collected at day 2 (D2, incubation period), day 6 (D6, disease onset), and day 10 (D10, severe stage prior to host death), as described in our previous report [16]. We performed differential expression analysis on 671 immunology-and inflammation-related genes via NanoString, using mock samples (D0) as the baseline. Of note, there were no statistically significant differences by D2 for any transcriptional targets (complete list in S2 Table). Such silent responses may not be surprising given the relatively slow replication rate of O. tsutsugamushi [33], as well as our previous PCR and ELISA findings of minimal activation of cytokines/chemokines at D2 [16,18]. For D6 samples, we found 221 genes exhibiting significant changes (adj. p < 0.05), among which 130 genes were upregulated and 91 genes were downregulated (complete list in S3 Table). Categorically, the top-20 most highly upregulated genes were dominated by those involved in type-1 responses (Cxcl9-11, Ifng, Il12rb1, Il27), monocyte/MF migration and activation (Ccl2, Ccl7, Ccl4), as well as interferon-stimulated genes (Socs1, Iigp1, Ifi204).
The greatest degree and spectrum of gene expression changes were found at D10 (complete list in S4 Table), with a total of 312 genes significantly differentially expressed (adj. p < 0.05). Of these 312 significant genes, 150 were upregulated, and 162 were downregulated (  with genes of interest marked in red). Fig 1B lists the top 20 upregulated genes (by fold change), including Th1-and CD8 T cell-recruiting/activating CXCR3 ligands (Cxcl9-11) and cytokines (Ifng, Tnf, Il12rb1, and Il27), as well as monocyte/MF recruiting/activating chemokines (Ccl2, Ccl7, Ccl4). Notably, Mincle was the 4 th greatest differentially expressed gene and the only PRR observed within the top 20. Genes exhibiting the greatest downregulation included complement components (C7, Vtn, Hc), the atypical chemokine receptor Ccrl1, and the aryl hydrocarbon receptor (Ahr) (S1A Fig). To determine functional relationships among differentially expressed genes of significance (adj. p < 0.05), we queried upregulated-and downregulated-genes to the String database (string-db.org) for pathway analysis via Reactome [34]. The number of genes differentially expressed within a given Reactome pathway were ranked and plotted (S1B Fig). On D10, there was a greater degree of changes in innate immune response markers (43 genes) than adaptive immune response markers (35 genes). Overall, these findings reveal a high degree of innate immune involvement at late, rather than early (D2), stages of disease, which was concurrent with a type 1-favoring cytokine/chemokine milieu. Furthermore, the abundantly high expression of Mincle highlights the need to better understand the contribution of CLRs in innate recognition of O. tsutsugamushi.

Mincle and signaling partners are increased in the lung and brain tissues during infection
Knowledge regarding PRR recognition of O. tsutsugamushi is limited. To date, only one study revealed TLR2 activation in-vivo and in-vitro [35], while the role of other classes of PRRs has been largely unexplored [33]. Since Mincle was one of the most highly transcribed genes on D10, we examined the differential expression of other CLRs during infection. Parsing our differential expression analysis, we observed a progressive increase in Mincle and Clec5a transcripts. By D10, Mincle expression was increased 36-fold and Clec5a exhibited a 7-fold increase compared with mock controls (Fig 2A and Table 1). It is known that unlike other CLRs, Mincle and Clec5a do not possess ITAMs/ITIMs and must interact with FcγRs or DAP10/DAP12, respectively, to initiate immune responses [23]. We therefore examined whether these signaling partners were also differentially expressed during infection (Fig 2B  and Table 1). As shown in Table 1, significant increases in the Fcgr1, Fcgr4, Fcer1g, Fcgr3, and Fcgr2b genes were consistently detected on D10, while no significant or relatively small (2-fold) increases were found for DAP10 or DAP12. Likewise, D10 brain tissues had significant increases in Mincle and Fcgr4 levels (440-517-folds), with no changes in DAP10 or a low-level increase of DAP12, further highlighting the important role of Mincle/FcγR signaling in immune responses in different organs. Regarding TLR involvement in the lung, we found a relatively low, but significant upregulation of Tlr1, Tlr2, and Tlr6 on D10, with no changes for Tlr4, and found a small, but significant reduction in Tlr5 (S2 Fig). Therefore, the activation of Mincle and Fcgr genes in infected tissues were highly consistent and selective.

Mincle predominates the CLR expression profile in infected lungs
Considering that Mincle and Clec5a were the only significantly changed CLRs in our gene profiling analysis, we used lung tissues for qRT-PCR analyses to validate the expression kinetics of these and other related CLRs that were not included in the NanoString kits (Clec6a, Clec7a, Clec4b1, Clec9a, Clec12a). We observed 13-fold upregulation of Mincle (p < 0.0001) and 4-fold upregulation of Clec5a (p < 0.05) at D9 (Fig 3A), corroborating our findings from differential expression analysis. We detected a 2-fold increase in Clec12a (p < 0.01), but a significant decrease in Clec7a (p < 0.001), Clec4b1 (p < 0.0001), and Clec9a (p < 0.001) at D9, further suggesting that Mincle was the preeminent CLR expressed in the lungs during infection (Fig 3A and S3 Fig). Additionally, we observed a 2.5-fold upregulation of Clec4d (also known as MCL, p < 0.0001) at D9. This finding was relevant, given that Mincle and MCL belong to the asialoglycoprotein receptor family (type II), forming a heterodimeric complex which stabilizes Mincle on the cell surface [20]. This complex formation has been shown to magnify Mincle signaling through the FcγRs in the context of dendritic cells treated with trehalose-6,6'-dimycolate, the Mincle ligand from Mycobacterium tuberculosis [36]. Examining the Fcgr expression profile at D9, we observed a 7-fold upregulation of Fcgr1 (p < 0.01), 5-fold upregulation of Fcgr4 (p < 0.0001), and approximately 3-fold increase in Fcer1g (p < 0.001, Fig 3B). Only Fcgr2b exhibited a decrease in expression (p < 0.001), which was contrary to our differential expression findings, while Fcgr3 exhibited no significant change. Together, our

Mincle protein levels are increased in the lungs and localized to infected regions
To examine Mincle protein expression and levels in lung tissues, we performed immunofluorescent staining and Western blot. Mincle-positive (red) cells were undetectable in mock

PLOS PATHOGENS
controls or weakly detected in D2 samples, but readily and intermittently detected at D6 and D10 of infection ( Fig 4A). Mincle staining was observed in cells surrounding areas with detectable O. tsutsugamushi antigens (green). We found Mincle-positive cells with no bacteria, as well as bacterium-carrying cells with no or very weak Mincle staining. Western blot on whole lung tissue homogenates revealed a low, but detectable level of Mincle at D6. However, the D10 samples consistently contained the highest (~11-fold) and significantly elevated levels of Mincle (p < 0.01, Fig 4B and 4C). Together, these data reveal the induction of Mincle at the translational level, most prominently at the severe stages of infection.

MF upregulation of Mincle and proinflammatory markers is dependent on O. tsutsugamushi doses and replication
We have reported that sustained recruitment and activation of M1-like MFs and neutrophils in inflamed lung tissues is a hallmark in murine models of lethal scrub typhus [18,28]. Since Mincle may be expressed in both MFs and neutrophils, we sought to examine the cellular sources of Mincle expression during O. tsutsugamushi infection. We first evaluated the Mincle transcription profile in infected neutrophils (MOI 10) and observed a small (~2-fold), but significant (p < 0.01), increase in Mincle transcription at 4 hr, which was completely diminished by 18 hr (S4 Fig). We then infected bone marrow-derived MFs with O. tsutsugamushi (MOI 5 or 10) and examined a panel of CLR and cytokine/chemokine transcripts at 2 to 48 hr of infection. Mincle transcription was significantly increased throughout the course of infection, with a peak increase of 17-fold (p < 0.001) at 4 hr for the high-dose group, but a peak increase of 12-fold (p < 0.01) at 48 hr for the low-dose group (Fig 6A). This infectious dose-dependent increase of Mincle, but not MDL-1, at 2 and 4 hr suggested a selective induction of Mincle by O. tsutsugamushi. It is known that Cxcl1, a neutrophil chemoattractant, is induced by Mincle activation [38]. Indeed, we found an infectious dose-dependent increase of Cxcl1 starting at 2 hr and peaking at 4 hr (80-fold increase in the high-dose group, p < 0.001). Together, our data indicate a selective upregulation of Mincle in response to O. tsutsugamushi infection that is most prominent in MFs.
The early induction of Mincle in response to live bacteria inspired us to further examine MF responses to heat-killed (HK) bacteria at a concentration equivalent of 10 MOI. We focused on two life cycle stages of O. tsutsugamushi: at 4 hr (when bacteria have just escaped from the endosome/phagosome to inhabit the cytoplasm) and at 24 hr (when bacteria are located near the perinuclear region for initial replication) [11,33]. While both inactivated and live bacteria induced a similar degree of Mincle upregulation at 4 hr, only live bacteria generated a sustained response at 24 hr (Fig 6B). With regard to inflammatory responses to HK and live bacteria, we observed only live O. tsutsugamushi generated significant and sustained expression of type 1-promoting cytokines/chemokines Il27, Nos2 (M1 MF marker), Ccl2 and Ccl5 (proinflammatory, M1-skewing chemokines), as well as Cxcl9. Concurrently, there was a significant reduction in Ahr and Mrc-1 (M2 activation markers) following exposure to live and inactivated bacteria. These results suggest both bacterial growth-dependent and -independent activation of Mincle and its related inflammatory responses in MFs.

Mincle protein expression increases in MFs exposed to O. tsutsugamushi
To confirm Mincle protein levels in infected MFs, we performed western blot and included LPS stimulation as a positive control. We found a marked increase in Mincle protein levels at 4 hr post-exposure to live or HK bacteria; however, at 24 hr, Mincle proteins were detected only in live bacterium-exposed MFs (Fig 7A and 7B), which were consistent with our PCR findings (Fig 6). We then examined the Mincle expression pattern in MF exposed to live or HK bacteria to determine whether signal colocalization would occur. Indirect immunofluorescent staining and confocal image analyses revealed comparable Mincle positivity (red) at 4 hr post-exposure to live or HK bacteria (green), on par with LPS stimulation (Fig 7C). While Mincle-positive staining was detected with or close to O. tsutsugamushi antigen-positive cells, there was limited evidence of Mincle-bacterium colocalization. At 24 hr, while Mincle staining was still readily detectable in MFs exposed to live

PLOS PATHOGENS
bacteria, only baseline levels of staining were found in MFs exposed to HK bacteria. Collectively, our data indicate that O. tsutsugamushi antigens are capable of stimulating Mincle protein expression in MFs, with the most pronounced response evident at early, rather than late, time points.

O. tsutsugamushi stimulates Mincle expression in both infected and uninfected MFs
Since our IFA findings revealed Mincle expression in both infected and uninfected MFs, we sought to measure Mincle protein levels via using flow cytometric analysis. First, we observed increased frequency of Mincle + cells in WT MF at 4 hr after infection (5 MOI), with no changes in Mincle -/cells (Fig 8A), corroborating our IFA findings (Fig 7). Then, we utilized CFSE-labeled O. tsutsugamushi to categorize infected (CFSE + ) or uninfected (CFSE -) cells at 4 hr. We found that 57% of WT MF were infected (p < 0.0001), while 64% percent of Mincle -/cells were infected (p < 0.0001) (Fig 8B). Though small, the difference in percent infected Mincle -/cells and WT cells was significant (p < 0.05). The mean fluorescent intensity (MFI) of CFSE was also significantly (p < 0.001) greater in Mincle -/cells when compared with WT. Additionally, our analysis revealed that 6% percent of WT cells were Mincle + (p < 0.01) and 4.5% of cells were CFSE + Mincle + (p < 0.001) after infection, with comparable results between surface stained and permeabilized MF (Fig 8C). Together, our data indicate both infected and uninfected cells express Mincle.
To examine whether and how host factors contribute to the Mincle expression, we focused our studies on TNFα, because it is known to induce Mincle in murine MF [39] and is highly upregulated in lungs of lethally infected mice (Fig 1)

TNFα amplifies Mincle-dependent inflammation in MFs exposed to O. tsutsugamushi
Based on the observation that MFs treated with TNFα prior to O. tsutsugamushi infection exhibited pronounced Mincle staining intensity, we asked whether TNFα may modulate Mincle-related inflammation. Using qRT-PCR, we found that WT MF treated with TNFα 30 min prior to infection with O. tsutsugamushi (5 MOI) exhibited significantly greater Mincle expression (25-fold increase, p < 0.0001) at 4 hr after infection than cells receiving TNFα (20-fold increase, p < 0.001) or O. tsutsugamushi alone (8-fold increase, p < 0.01, Fig 9). We also observed similar trends for the key proinflammatory markers Cxcl1, Nos2, Il27, Cxcl9, and Cxcl10. Overall, Mincle -/-MF presented parallel trends for the inflammatory markers analyzed. However, we observed that the degree of Il27, Ccl2, and Cxcl10 transcripts generated in infected Mincle -/cells receiving TNFα pretreatment was less than half that of their WT comparators. The differences in Il27 and Cxcl10 production in infected Mincle -/and WT cells was even more pronounced by 24 hr post-infection (S6 Fig). Together, our data reveal that TNFα may promote and exacerbate Mincle-mediated inflammation. Firstly, the NanoString study described herein offered comprehensive profiles of important host immune response/inflammation genes in the lungs at different stages of infection. This data confirmed and greatly expanded previous animal studies that examined a limited number of cytokine/chemokine genes in visceral organs via PCR during O. tsutsugamushi infection [16,28,32]. Our finding of Il27 upregulation in inflamed lung tissues (Fig 1) was important, given our previous study revealing increased IL-27 protein via Bioplex in infected lungs [17]. The identification of Il27 and other immunologic pathways that were significantly increased surface vs. intracellular Mincle staining, showing percentages of Mincle + and CSFE + Mincle + cells. (D) Immunofluorescent assay detection of Mincle (red), O. tsutsugamushi (green), and DNA (DAPI, blue) in cells pretreated with TNFα (25 ng/mL for 30 min) prior to infection (scale bar = 10 μm for close-view images, 20 μm for all others). Unpaired t-test was used for statistical analysis. For comparison with mock controls within WT or Mincle -/cells � , p < 0.05; �� , p < 0.01; ���� , p < 0.0001. For comparison between WT and Mincle -/cells ##, p < 0.01, ###, p < 0.001.
https://doi.org/10.1371/journal.ppat.1009782.g008 Secondly, the PRR activation profile to O. tsutsugamushi is scarcely understood; our studies with mouse tissues and cultured MFs has suggested an important role for CLRs, but not TLRs, in response to this Gram-negative bacterium. One study has described TLR2 activation in O. tsutsugamushi-infected mice and cultured HEK293 cells [35] and others identified no role for TLR4 in mice during infection [40,41]. These reports were consistent with our observation of low TLR2 and no TLR4 induction in vivo (S2 Fig). For intracellular bacteria replicating in the cytosol, activation of cytoplasmic DNA sensors, such as STING and RIG-I, would likely occur; however, evidence for the activation of these sensors remains speculative [41]. Our discovery of two highly upregulated CLRs, Mincle and Clec5a, in lethally infected mice was novel and significant. Among these two CLRs, Mincle was likely the most important sensor for O. tsutsugamushi, because 1) it was the 4 th most upregulated gene in the lungs by the terminal phase of disease and the only PRR within the top-20 (Fig 1); 2) the timing and peak activation of Mincle coincided with the onset of disease and progression of lethal infection in the lungs (Figs 2 and 3) and brains (Table 1) of mice; and 3) signaling partners for Mincle (Fcgr1 and Fcgr4), but not those for Clec5a (DAP10 or DAP12), displayed a high degree of upregulation, implicating Mincle pathways as potentially playing a more compelling role during infection (Figs 2 and 3). This conclusion was further supported by the positive correlation between the peak upregulation of proinflammatory cytokines/chemokines through the terminal phase of disease (Fig 5).
At present, Mincle has mainly been characterized in the context of MF-pathogen interactions, especially for M1 polarization and proinflammatory responses to M. tuberculosis [25,42]. MFs play key roles in infection with O. tsutsugamushi and other closely related Rickettsia species [43][44][45][46]. This study, for the first time, revealed O. tsutsugamushi infection-triggered Mincle expression in M1-like MFs in an infectious dose-dependent manner (Fig 6). We were surprised at the capacity of inactivated O. tsutsugamushi to stimulate transient Mincle expression along with other proinflammatory or M1-promoting genes (Il27, Nos2, Tnf, Ccl5, Cxcl1), but not Ccl2, Ccl4, and Cxcl9 at 4 hr, while strong or significant upregulation was sustained in live bacteria-exposed cells at 24  Nevertheless, our cytokine signatures for WT cells largely agree with previous studies that explored cytokine profiles of human monocytes/MF responding to live O. tsutsugamushi infection [14,47]. Our findings are also relevant to the basic biology of O. tsutsugamushi, as the early phase (4 hr) represents the stage immediately after endosomal/phagosomal escape, whereas the later phase (24 hr) coincides with the start of bacterial replication [33]. The reduction in the magnitude of immune responses at 24 hr of infection (Fig 6) may be due to the activation of cellular regulatory molecules/pathways, or pathogen-driven mechanisms, as O. tsutsugamushi can actively modulate host immune responses during infection. Since O. tsutsugamushi bacteria replicate very slowly (with~10-hr doubling time in vitro) [48,49], the capacity of evading host immune responses is critical for their successful replication. For example, O. tsutsugamushi can actively thwart the NF-κB activation pathway by 24 hr of infection [29,50]. Regardless of the underlying mechanisms, our findings of rapid Mincle expression elicited by inactivated bacteria suggest the presence of potential bacterial ligands for Mincle recognition, which were preserved in our heat-inactivation procedure. This is important as it may allow HK bacteria to be explored further for the identification of putative Mincle ligands, precluding the technical and logistical challenges associated with performing such experiments in BSL-3 containment facilities.
Thirdly, we revealed a unique Mincle protein expression pattern in infected lung tissues and MFs (Figs 4, 7, and 8). Looking at the cellular expression profile for Mincle in the lungs, we observed Mincle staining in cells nearby O. tsutsugamushi antigen by D10 (Fig 4). This suggested that non-infected cells may be responding to bacterial components, or alternatively, to cytokines/chemokines. While we did not perform co-staining for specific cell phenotypes, we speculated that Mincle-positive cells were most likely infiltrating monocytes/MFs, based on our previous FACS analyses of lung-derived cells [18]. Our in-vitro studies and IFA staining of infected MF supported this notion and revealed an intriguing pattern of Mincle expression. Early in infection, comparable levels and patterns of Mincle proteins were detected in cells treated with either inactivated or live bacteria (Fig 7). Later in infection (24 hr), however, only cells infected with viable bacteria had detectable Mincle, though overall signal intensity was lower compared with the early timepoint. These IFA findings were in congruence with our qRT-PCR and WB data, revealing the importance of viable bacteria in generating sustained Mincle expression. The IFA findings of Mincle positivity in both bystander and infected cells after treatment with live or inactivated O. tsutsugamushi suggested to us that bacterial components may not be the sole source underlying our findings. This notion was further supported by our flow cytometric data, confirming that both infected and uninfected cells can express Mincle following exposure to live O. tsutsugamushi (Fig 8). Of note, though Mincle is is known to be anchored to the cytoplasmic membrane, we observed a diffuse cytoplasmic staining pattern in our infected or LPS-stimulated MFs (Fig 7), which were consistent with other reported IFA studies in MF-like RAW264.7 cells stimulated with LPS in-vitro [51] and in dermal dendritic cells in situ [52]. To confirm an intracellular pool of Mincle proteins, we analyzed the percent Mincle + cells from surface vs. intracellular stained MF via flow cytometry. We confirmed that while the majority of Mincle proteins were detected via surface staining, the percent Mincle + cells determined via intracellular staining was slightly greater (Fig 8).
While it remains unclear as to whether the minimal co-localization of Mincle and Oriential antigens in infected lung tissues and MFs was due to the unique biology of O. tsutsugamushi or technical issues related to antibodies used in this study, it was evident that Mincle is not required for O. tsutsugamushi entry or replication (Fig 8B and S5 Fig).
Finally, Mincle expression and related inflammation were likely initiated via multiple avenues during O. tsutsugamushi infection, including via bacteria-related components, cytokine/ chemokine production, and host DAMP release. This study confirms a previous report describing TNFα as a strong stimulus of Mincle expression [39]. Importantly, we revealed a significant, TNFα-mediated amplification of Mincle expression, as well as proinflammatory expression (Il27, Cxcl9, Cxcl10, Ccl2, Cxcl1, Nos2) in the context of O. tsutsugamushi infection (Fig 9). More importantly, we found that such TNFα-mediated enhancement during infection was 2-to 3-folds reduced in Mincle -/-MFs (Il27, Cxcl10, Ccl2), implying the mitigation of proinflammatory responses. Our data collectively indicate a prominent role of Mincle in magnifying inflammatory response in a TNFα-rich environment. This finding has important implications, since TNFα levels are known to correlate of disease severity in human scrub typhus [53], and that TNFα is highly upregulated in the lungs of lethally infected mice (Fig 1) [16]. Studies are ongoing to examine possible mitigation of proinflammation and acute tissue injury during infection in Mincle -/mice (plus anti-TNFα treatment). Future investigation should also be aimed at defining potential bacterial components that could induce Mincle expression. As an obligately intracellular pathogen, O. tsutsugamushi relies on the host cell as a source of nutrients [33]. Therefore, it is conceivable that enzymatic altering of host cell proteins or lipids during this process may, in turn, stimulate Mincle, in a similar manner to that described for Helicobacter pylori [26]. Identification of such a ligand, however, presents major challenges due to the lack of available tools for genetic manipulation of O. tsutsugamushi at the present time [7,11].
Based on our findings presented herein, we propose a hypothetic model for the contribution of Mincle and related pathways in immune responses to O. tsutsugamushi infection (Fig 10). MFs can sense invading or inactivated bacteria via inducing Mincle and FcγR expression at the initial stages of infection. MFs can also respond to released host DAMPs and TNFα, as intracellular bacterial growth progresses, via increased Mincle expression on the cell surface and in an intracellular pool. The activation of Mincle/FcγR promotes the expression of diverse proinflammatory cytokines/chemokines for potential recruitment and activation of phagocytes (CXCL1, CCL2, TNFα, iNOS), as well as Th1 and CD8 + T cells (CXCL9, CXCL10, IL27, TNFα). These M1-like MFs help control bacterial infection; however, excessive TNFα in the microenvironment can result in sustained Mincle expression and MF inflammatory responses. These innate responses may ultimately contribute to Th1/CD8-skewed responses and acute tissue damage, as we observed in the lungs and other organs [17,18,54]. This model suggests

PLOS PATHOGENS
that besides Mincle -/mice, future investigation with targeted blockage of TNFα-or FcγR1/ FcγR4-mediated pathways will help define the interplay of TNFα/Mincle/FcγR signaling in the regulation of host-O. tsutsugamushi interactions and pathology of scrub typhus. Such studies would also help understand immune responses against other obligately intracellular pathogens, including Rickettsia, Anaplasma, Ehrlichia, and Chlamydia, for which virtually no information is yet available regarding the role of Mincle or other CLRs in pathogen recognition.
In summary, this study revealed new insights into the innate immune recognition of O. tsutsugamushi. Through comprehensive differential expression analysis of mouse lung and brain tissues, we provided the first evidence of Mincle involvement during O. tsutsugamushi infection. Our observations from in-vitro MF infection suggest a selective upregulation of Mincle that may rely on bacterial growth-dependent and -independent factors, as well as TNFα stimulation. To-date, CLRs represent a virtually unexplored realm of immunology for O. tsutsugamushi and other obligately intracellular bacteria. While potential bacterium-and/or hostderived ligands for Mincle/FcγR signaling remain unknown, it is conceivable that Mincle activation during infection may contribute to or drive an overzealous type 1 response in both innate and adaptive immune cells, leading to progressive tissue damage during O. tsutsugamushi infection. A better understanding of how Mincle activation contributes to immune dysregulation may aid the design of treatments for severe scrub typhus.
Supporting information S1 Table. Primer sequences for qRT-PCR analysis used within the study. Nanostring gene profiling analysis was performed on RNA isolated from lung tissue homogenates of lethally infected mice at D2, D6, and D10, respectively, and compared with the mock controls. We then parsed our data to examine expression of Tlr genes. Scale is Log2Fold change compared to mock. Graphs are shown as mean ± SEM. Significance was determined utilizing the Benjamini-Yekutieli test. � , p < 0.05, �� p < 0.01. (TIF)

S3 Fig. Expression of additional CLRs in the lungs of lethally infected mice.
Whole lung tissue homogenates from lethally infected mice were measured for expression of indicated CLRs via qRT-PCR. All data are presented relative to GAPDH values and shown as mean ± SEM. Three independent mouse infection experiments were performed with similar trends; representative data are shown. One-way ANOVA with Dunnett's multiple comparison test was used for statistical analysis, with mock samples used as a reference. � , p < 0.05, �� p < 0.01, ��� p < 0.001, ���� p < 0.0001. (TIF)

S4 Fig. Expression of CLRs infected neutrophils.
Bone marrow-derived neutrophils of C57BL/6J mice were exposed to live bacteria (10 MOI). mRNA levels of select CLRs were analyzed via qRT-PCR. Data are presented relative to GAPDH values and shown as mean ± SEM. Unpaired t-test was used for statistical analysis, with Mock samples used as a reference. � , p < 0.05, �� p < 0.01. (TIF)

S5 Fig. Proinflammatory signatures and bacterial loads in WT and Mincle -/-MF following infection.
Bone marrow-derived WT or Mincle -/-MF were exposed to live bacteria (MOI 5, or 10). qRT-PCR analyses of indicated genes at (A) 4 hr and (B) 24 hr post-infection are presented with data presented relative to GAPDH and shown as mean ± SEM. One-way ANOVA with Dunnett's multiple comparison test was performed for treatment groups within the WT or Mincle -/-MF background, respectively. Unpaired t-test was utilized for comparison between infected WT and Mincle -/-MFs. (C) Bacterial growth in infected MFs (MOI 2 or 5) was analyzed at 4, 24, and 48 hr. Bacterial loads were determined by qPCR. Data are presented as the copy number of O. tsutsugamushi 47-kDa gene copy per ng of DNA. Unpaired t-test was used for statistical analysis. � , p < 0.05; �� , p < 0.01; ��� , p < 0.001; ���� , p < 0.0001. (TIF)