Host cytosolic RNA sensing pathway promotes T Lymphocyte-mediated mycobacterial killing in macrophages

Mycobacterial infection leads to activation of the RIG-I/MAVS/TBK1 RNA sensing pathway in macrophages but the consequences of this activation remains poorly defined. In this study, we determined that activation of this RNA sensing pathway stimulates ICAM-1 expression in M.avium-infected macrophage through the inhibition of the E3 ubiquitin ligase CRL4COP1/DET1. CRL4 when active targets the transcription factor ETV5 for degradation by the ubiquitin–proteasome system. In the absence of the ETV5 transcription factor, ICAM-1 expression is significantly decreased. The M.avium-induced ICAM-1 production is required for the formation of immune synapse between infected macrophages and antigen-specific CD4+ T lymphocytes, and is essential for CD4+ T lymphocyte-mediated mycobacterial killing in vitro and in mice. This study demonstrates a previously undefined mechanism by which a host cytosolic RNA sensing pathway contributes to the interplay between mycobacteria infected macrophages and antigen-specific T lymphocytes.

Introduction Non-tuberculous mycobacteria (NTM) are opportunistic pathogens, predominantly causing pulmonary infections in susceptible populations including the elderly and in patients receiving immunosuppressive drugs, or with pre-existing conditions such as cystic fibrosis, chronic obstructive pulmonary disease (COPD) or bronchiectasis. The most commonly isolated NTM species are Mycobacterium avium complex (MAC) (M.avium and Mycobacterium intracellulare) and Mycobacterium abscessus complex (MABSC) (M. abscessus, M. massiliense and M. bolletii), which account for >90% of the total cases reported [1]. NTM infections are increasing in the U.S., Japan and many European countries. Much of this growth is due to increased infections of older Caucasian and Asian women who have no known underlying lung disease [2]. Despite their increased clinical importance we still have a limited knowledge of M.avium pathogenesis and host immunity.
In this study, we investigated the function of the cytosolic RNA sensing pathway during M. avium infection in vitro and in vivo. Host cytosolic RNA sensing pathways play a central role in detecting foreigner DNA and RNA in virally infected cells [3]. Recently, it was found that Listeria monocytogenes and Mycobacterium tuberculosis (M.tb) release their RNAs into host cells through the bacterial SecA2 secretion system and induce RIG-I/MAVS/TBK1-dependent type I IFN production [4,5]. However, whether the activation of the RIG-I/MAVS RNA sensor pathway stimulates other cellular responses following a bacterial infection remains undefined. During viral infections induction of the RIG-I/MAVS pathway has been shown to drive production of IL-1β and IL-18 [6], and promote release of cytochrome c from the mitochondria in a Bax-dependent manner leading to apoptosis of the infected cell [7]. These and other studies have demonstrated the importance of the RNA sensing pathway in promoting a protective host response to viral infections. However, the importance of RNA sensing pathways in the context of a bacterial infection is less clear and evidence points to both a protective and detrimental effect [5,8]. In support of the former we have identified a RIG-I/MAVS-dependent production of the Intercellular Adhesion Molecule 1 (ICAM-1) in macrophages following an M. avium infection.
ICAM-1 promotes cell-cell interaction by serving as the ligand for the leukocyte adhesion protein LFA-1 and Mac-1 [9]. It is also important in formation of immune synapse between T cells and antigen presenting cells (APCs) [10]. Expression of ICAM-1 is required to control an M. avium infection [11]. However, these studies were carried out using ICAM-1 -/mice so ICAM-1's role in T cell-APC interaction during a mycobacterial infection has not been defined. Moreover, what regulates ICAM-1 expression during a mycobacterial infection remains unclear. Previous studies have shown activation of signaling pathways such as PI3K/ Akt and the MAPKs result in activation of the transcription factors AP-1 and NF-ƙB driving ICAM-1 expression [12]. Initiation of these pathways can be induced by engagement of receptors such as TNFR1, TLR4 and EGFR, among others [13]. In the present study we identified a previously unknown role for the transcription factor ETV5 in regulating ICAM-1 expression. We also found that ICAM-1 expression was essential for CD4 + T-mediated M.avium killing within infected macrophages and mice. Our study sheds new light on the host cytosolic RNA sensing pathway in controlling an NTM infection and identified a previously undefined role for the RIG-I/MAVS/TBK1 RNA sensing pathway in regulating the immune synapse between macrophage and CD4 + T cells, and subsequent macrophage activation by CD4 + T cells.

M.avium activates the host cytosolic RIG-I/MAVS/TBK1/IRF3/IRF7 RNA sensing pathway
Our previous study showed that M.tb release mycobacterial RNA into the cytosol of infected macrophages via a mycobacterial SecA2-dependent pathway. Mycobacterial species including M.avium also express a SecA2 secretion system. The M.avium SecA2 shares 93% homology to M.tb SecA2. To evaluate whether M.avium also release their RNA into host cells and induce type I IFN production, we initially infected mouse bone marrow-derived macrophages (BMMs) with M.avium strains 104 and 2151. As shown in Fig 1A, M.avium significantly induced IFN-β production at 24 and 72 hr post-infection. An IFN-β mRNA expression peak was detected at 8 hr post-infection (Fig1B), which was similar to our finding with M.tb [5]. M.avium RNAs in the cytosol of infected macrophages were detected by quantitative RT-PCR. As seen in Fig 1C, four M.avium transcripts, mce1B, polA, rpoC and MAV4436, were detected in M.avium-infected but not uninfected macrophages. However, we did not detect 16S ribosomal RNA in the macrophage cytosol suggesting that the detected RNA was not due to contaminating lysed M.avium ( Fig 1C). To determine the link between M.avium RNA and type I IFN production, we measured the IFN-β production in WT and Mavs -/-BMMs following an M.avium infection. As shown in Fig 1D and 1E, the M.avium-induced IFN-β production was significantly diminished in Mavs -/relative to WT BMMs. Similar results were detected in M.avium-infected BMMs pretreated with RIG-I, TBK1, IRF3 or IRF7 siRNA when compared to cell pretreated with control siRNAs (Fig 1F and 1G, and S1 Fig). IRF3 and IRF7 are two critical transcription factors driving IFN-β expression in the context of viral and M.tb infection [3,5] and their activation and translocation into the nucleus of host cells requires phosphorylation by TBK1. As seen in Fig 1H, M.avium infection induced IRF3 and IRF7 nuclear translocation in WT but not Mavs -/-BMMs at 24 hr post-infection. As observed with an M.tb infection, IRF7 expression was induced by M.avium in WT BMMs. However, in contrast to M.tb, M.avium infection led to IRF3 nuclear translocation via a MAVS-dependent pathway. As predicted, MAVS was required for TBK1 phosphorylation in M. avium-infected BMMs ( Fig 1I).

The MAVS-dependent RNA sensing pathway is required for host defense against M.avium infection in mice
To further evaluate the host cytosolic RNA sensing pathway we intratracheally infected WT and Mavs -/-C57BL/6 mice with wild-type M.avium and measured mycobacterial survival and tissue histopathology at 0, 1, 2, 3, 5, 8 and 12 weeks post-infection. As shown in Fig 2A, MAVS deficiency resulted in a 0.5 to 1.0 log 10 increase in bacterial burden in the lung at 3, 5, 8 and 12 weeks post-infection when compared to WT infected mice. A similar difference in M.avium burden was observed in mouse spleen at 5, 8 and 12 weeks post-infection ( Fig 2B). Consistent with M.avium survival results, the lungs from M.avium-infected Mavs -/mice showed greater histopathological changes and immune cell infiltration at 3, 5, 8 and 12 weeks post-infection (Fig 2C and 2D). Flow cytometry analysis at 3 weeks post-infection showed the lungs of M. avium-infected Mavs -/mice to have a higher concentration of macrophages and T cells when compared to M. avium-infected WT mice (Fig 2E). No difference in the number of DCs and neutrophils was detected in the lung between M.avium-infected WT and Mavs -/mice. Similar to the results seen in BMMs, diminished IFN-β production was observed in the serum of M. avium-infected MAVS -/relative to infected WT mice (Fig 2F).

RIG-I/MAVS/TBK1-dependent and IRF3/7-independent RNA sensing pathway regulates ICAM-1 Production in M.avium-infected host cells
MAVS deficiency attenuated the host defense against an M.avium mouse infection. This result is in contrast to what we observed in M.tb-infected mice where M.tb burden was significantly lower in Mavs -/infected mice when compared to WT [5]. To better understand the MAVSdependent mycobacterial killing, we investigated the serum cytokine profile in WT and Mavs -/ avium-induced IFN-β production was determined in BMMs at 24 hr post-infection following a 48 hr pretreated with negative control (Control), RIG-I, TBK1, IRF3 or IRF7 siRNA. (G) Similar to (F), but M.avium-induced IFN-β production was measured by ELISA. (H) Western Blot analysis for IRF3 and IRF7 in whole cell lysate (WCL) and nuclear fraction (Nucleus) of WT and Mavs -/-BMMs at 24 hr post-infection. (I) Similar to (H), but TBK1 phosphorylation (Ser172) was analyzed in WCL. β-Actin and Histone H3 served as loading control for WCL and nuclear fraction, respectively. Data shown in (A-G) are the mean ±SD (n = 3 wells per group), and all data are representative of three independent experiments (biological replicates). For (D) and (F), the data was expressed as fold change relative to M.avium-infected WT (D) or control (F) BMMs. GAPDH served as loading control in qRT-PCR. � P < 0.05, �� P < 0.01, and ��� P < 0.001 by Student's t-test (two-tailed).   Fig 3A and 3B, we identified 15 proteins that were differential produced in M.avium-infected WT and Mavs -/mice at 3 weeks post-infection. The production of 6 proteins, IGFBP-1, IGFBP-2, LDL R, P-Selectin, PCSK9 and FGF acidic, increased in Mavs -/vs. WT mice. Nine proteins, Angiopoietin-1, Chitinase 3-like 1, Chemerin, Complement Factor D, C-Reactive Protein, DPPIV, Gas6, ICAM-1 and TNFRSF11B, were downregulated in Mavs -/vs. WT mice. As M.avium is an intracellular bacterial pathogen and primarily infects alveolar macrophages, we tested the expression of these 9 downregulated proteins in uninfected and M.avium-infected BMMs. Among those tested, we only detected 5 mRNA (Chitinase 3-like 1, Complement Factor D, DPPIV, ICAM-1 and TNFRSF11B) by qRT-PCR, and all showed an M.avium-inducible expression in WT BMMs. ICAM-1 was the only protein whose expression was decreased in Mavs -/-BMMs compared to WT cells ( Fig  3C), ICAM-1 is a cell surface glycoprotein expressed on endothelial cells and immune cells but can also be detected extracellularly [14]. Based on the qRT-PCR results we analyzed the abundance of surface ICAM-1 on macrophages using flow cytometry. As seen in Fig 3D, compared to uninfected BMMs, M.avium infection significantly increased ICAM-1 surface expression in WT but not Mavs -/-BMMs. A similar result was observed in alveolar macrophages isolated from WT and Mavs -/mice at 3 weeks post M. avium infection ( Fig 3E). In contrast, a comparable level of ICAM-1 was detected in DCs, neutrophils, CD4 + and CD8 + T cells in the lung of M. avium-infected WT and Mavs -/mice (S2A Fig). Interestingly, RIG-I-/TBK1-but not IRF3/ 7-knockdown diminished ICAM-1 production in M.avium-infected BMMs (Fig 3F), suggesting an undefined downstream pathway mediating ICAM-1 production in host cells.

RIG-I/MAVS/TBK1-dependent RNA sensing pathway mediates ETV5 stability in BMMs
As MAVS-dependent production of ICAM-1 in M. avium-infected BMM did not require IRF3 or IRF7 we analyzed the promoter region in both human and mouse ICAM-1 gene to identify conserved transcription factor-binding sites (Fig 4A), We found NF-κB, Ets-2, ETV5, Ap-2, Sp-1 and Stat1 binding sites [15] and hypothesized that MAVS deficiency alters the activation of one or more of these factors induced upon an M.avium infection. To test this hypothesis we investigated the nuclear translocation of these 6 transcription factors in WT and Mavs -/ -BMMs following an M.avium infection. Only ETV5 nuclear levels were significantly lower in Mavs -/compared to WT BMMs ( Fig 4B). A similar result was detected in the WCL. Moreover, ETV5 abundance in the nucleus and WCL was also diminished in M. avium-infected BMMs that were pretreated with RIG-I-or TBK1-specific siRNA relative to cells receiving control siRNA ( Fig 4C). However, qRT-PCR analysis showed a comparable level of ETV5 mRNA in M. avium-infected WT and Mavs -/-BMMs (Fig 4D), suggesting limited ETV5 translation or diminished protein stability in Mavs -/-BMMs. Recently, it was found that ETV5 can be degraded through the cullin-RING ubiquitin ligase CRL4 COP1/DET1 -meidated ubiquitination in mouse alveolar type II cells [16]. This degradation was shown to be inhibited by MLN4924, an inhibitor of CRL4 COP1/DET1 complex [17]. To test whether CRL4 COP1/DET1 -meidated ubiquitination was involved in the diminished ETV5 expression observed in M.avium-infected Mavs -/-BMMs, we inactivated the CRL4 COP1/DET1 enzyme complex using MLN4924. As shown in Fig 4E, inhibiting CRL4 COP1/DET1 rescued ETV5 protein production in Mavs -/ -BMMs while having no effect on EVT5 mRNA abundance (Fig 4F).
To further define ETV5's role in M.avium-induced ICAM-1 expression, we evaluated the ICAM-1 production in WT BMMs pretreated with ETV5-specific siRNA. As seen in Fig

MAVS deficiency attenuates CD4 + T Cell-mediated mycobacterial killing in BMMs
To understand the role of host cytosolic RNA sensing pathway in the antimycobacterial response, we investigated M.avium replication in Mavs -/-BMMs and RIG-I, TBK1, IRF3 or IRF7-knockdown BMMs. As seen in S3A and S3B Fig, a minimal increase in mycobacterial burden was detected in Mavs -/-BMMs as well as RIG-I, TBK1, IRF3 or IRF7-knockdown BMMs when compared to WT or control siRNA-treated cells at 72 hr post infection. This suggest that the host cytosolic RNA sensing pathway plays a limited role in the antimycobacterial response. In addition to its role as an adhesion molecule for cell-cell interaction, ICAM-1 is also an essential component of the immunological synapse between antigen-presentation cells and activated lymphocytes [9]. CD4 + T cell-mediated cellular immune response plays a critical role in controlling mycobacterial infections [18]. Therefore, we hypothesized that MAVSdependent ICAM-1 production in M.avium-infected macrophages could facilitated CD4 + T cell-mediated mycobacterial killing in host cells. To test our hypothesis, we isolated CD4 + T cells from M.avium infected mice and cocultured these T cells with M.avium-infected BMMs. As seen in Fig 5A, we observed significantly fewer CD4 + T cells associated with M.aviuminfected Mavs -/compared to WT infected BMMs. To determine if this difference in T cell-BMM interaction results in changes in M.avium survival, we lysed the infected BMMs following coculture and quantified the bacterial burden ( Fig 5B). As shown in Fig 5C, M.avium burden declined by~1.5 log10 in WT BMMs cocultured with CD4 + T cells (cell ratio, 1:2) isolated from M.avium-infected WT mice compared to BMMs alone. In contrast, CD4 + T cell-mediated mycobacterial killing was absent when these CD4 + T cells were cocultured with M.aviuminfected Mavs -/cells ( Fig 5D). Limited effect on M.avium survival was observed when CD8 + T cell isolated from M.avium-infected WT mice, were added to M.avium infected WT BMMs (S3C Fig). CD4 + or CD8 + T cells from naïve WT mice had no effect on M.avium survival in host cells (S3D Fig). To investigate whether CD4 + T cell-mediated mycobacterial killing relies on cell-cell contact, we assessed M. avium burden in BMMs cocultured with CD4 + T cells from M. avium-infected WT mice in a transwell system. As shown in Fig 5E, CD4 + T cells failed to activate antimycobacterial response in the absence of cell-cell contact. To better understand the role of the cytosolic RNA sensing pathway in host defense against M.avium infection in vivo, we compared the capability of CD4 + T cells isolated from M.avium-infected WT and Mavs -/mice at 3 weeks post infection to stimulate mycobacterial killing when cocultured with M.avium-infected WT BMMs. As seen in Fig 5F, CD4 + T cells from the two mouse strains had similar antimicrobial activity. In parallel, a comparable level of IFN-γ was detected in CD4 + and CD8 + T cells isolated from M. avium-infected WT and Mavs -/mouse lung (S2B and S2C Fig). This suggest that MAVS is not required for activation of Th 1 CD4 + T cell in M. avium-infected mice. In contrast, using the macrophage-CD4 + T cell coculture system we found that MAVS deficiency in host macrophages attenuated IFN-γ production by the CD4 + T cells (Fig 5G). Similarly, less T cell proliferation was detected when CD4 + T cells, isolated from infected mice, were cocultured with M.avium-infected Mavs -/cells compared to CD4 + T cells cocultured with infected WT cells ( Fig 5H).

PLOS PATHOGENS
RNA sensing pathway in T cell mediated killing of mycobacteria Following coculture with activated antigen-specific CD4 + T cells, M.tb-infected macrophages are known to produce antimicrobial reactive oxygen and nitrogen species (ROS and RNS) via NADPH oxidase (NOX2/gp91phox) and inducible nitric oxide synthase (iNOS) respectively [19]. However, little is known about ROS and NO production in the context of NTM infections. Since MAVS deficiency in M.avium infected BMMs resulted in a loss of CD4 + T cell-mediated killing of M.avium, we investigated if the production of ROS and RNS was altered in M.avium-infected Mavs -/compared to WT BMMs following cocultured with CD4 + T cells isolated from M.avium-infected WT mice. In the presence of CD4 + T cells, the level of ROS ( Fig 5I) and iNOS ( Fig 5J) was significantly higher in WT compared to Mavs -/-M. avium-infected BMMs. However, no significant difference in ROS and iNOS production was detected between M.avium-infected WT and Mavs -/-BMMs in the absence of CD4 + T cells.

MAVS deficiency attenuates CD4 + T Cell-mediated mycobacterial killing in mice
Our data indicates that MAVS deficiency impairs macrophage-T cell immune synapse formation upon in vitro coculture. However, a comparable level of IFN-γ was observed in CD4 + T cells (S2B Fig) and the lungs (S2C Fig) of M. avium infected WT and MAVS -/mice. We hypothesized that MAVS play a key role in CD4+ T cell-mediated stimulation of macrophages to kill intracellular M.avium but not CD4 + T cell activation in vivo. To test our hypothesis, we established a triple-cell population coculture system (Fig 6A), in which CD4 + T cells were isolated from WT mice at 3 weeks post M.avium infection. These T cells were then cocultured with WT BMMs infected with gentamicin-sensitive (Gen S ) M.avium, followed by coculture with WT or Mavs -/-BMMs that were infected with gentamicin-resistant (Gen R ) M.avium strains. The Gen R and Gen S bacterial burden in the BMMs were quantified. A difference in Gen R M.avium burden between WT and Mavs -/-BMMs in coculture would indicate a role for MAVS in CD4 + T cell-mediated stimulation of macrophages to kill intracellular M.avium. As shown in Fig 6B, in the presence of CD4 + T cells isolated from M.avium-infected WT mice, Gen R M.avium burden in WT BMMs was significantly lower compared to Mavs -/-BMMs. In contrast, in the absence of CD4 + T cells, Gen R M.avium had a similar survival rate in WT and Mavs -/-BMMs. Similarly, Gen S and Gen R M.avium showed a comparable survival in WT BMMs (Fig 6C). CD4 + T cells from M.avium-infected WT mice did kill Gen S M.avium in WT BMMs (Fig 6C).
In the triple-cell coculture wells, a similar level of IFN-γ and T cell proliferation was seen regardless of whether Gen R M.avium-infected WT or Mavs -/-BMMs were present, suggesting an effective activation of CD4 + T cells by Gen S M.avium-infected WT BMMs (Fig 6D and 6E).
Our results indicate an important role for the host cytosolic RNA sensing pathway in the CD4 + T cell-mediated antimycobacterial response. To evaluate the engagement of MAVS in M.avium-infected macrophages in vivo, we performed an alveolar macrophage (AM) adoptive transfer experiment where WT or Mavs -/-AMs, which were infected with Gen R M.avium, were intratracheally administered to WT mice that had been previously infected with the Gen S M.avium for 3 weeks as described in Fig  , and all data are representative of three independent experiments. In qRT-PCR, GAPDH served as loading control and data was expressed as fold change relative to uninfected WT cells or control siRNA-treated uninfected WT cells. n.s., not statistically significant and � P < 0.05 by Student's ttest (two-tailed). https://doi.org/10.1371/journal.ppat.1008569.g004

PLOS PATHOGENS
RNA sensing pathway in T cell mediated killing of mycobacteria infected WT but not Mavs -/-AMs were injected into mice previously infected with Gen S M. avium (Fig 6G). AM adoptive transfer had no effect on the Gen S M.avium burden in mouse lung. In macrophage culture, Gen R M. avium had a comparable survival rate in WT and Mavs -/-AMs (Fig 6H). Additionally, neither the transfer of WT or Mavs -/-AMs affected the IFN-γ production in the lung of Gen S M.avium-infected WT mice (Fig 6I).

M.avium-induced ICAM-1 production contributes to CD4 + T Cellmediated mycobacterial killing in BMMs and mice
Since ICAM-1 is necessary for the formation of immune synapse between APC and T cells 9 , we hypothesized that the impaired mycobacterial killing in Mavs -/-BMMs was due to diminished ICAM-1 production in infected host cells. To test this hypothesis we first defined the surface expression of additional proteins required for the APC-T cell synapse in M.aviuminfected BMMs. MHCII, CD80, CD86 and CD40 showed similar abundance on the plasma membrane of M.avium-infected WT and Mavs -/-BMMs (S2D Fig and Fig 3D). To further test our hypothesis, we infected WT and Mavs -/-BMMs with M.avium in the presence of MLN4924 before cells were cocultured with CD4 + T cells isolated from M.avium-infected WT mice. To avoid the potential interference of MLN4924 on CD4 + T cells, MLN4924 was not included in macrophage-T cell coculture system. MLN4924 restored the immune synapsis between CD4 + T cells and M.avium-infected Mavs -/-BMMs but had no effect on immune synapsis formation when using ICAM-1 -/-BMMs (Fig 7A). MLN4924 also rescued the antimycobacterial activity in M.avium-infected Mavs -/-BMMs cocultured with CD4 + T cells to levels comparable to what was observed in M.avium-infected WT BMMs ( Fig 7B). As predicted, CD4 + T cell failed to control M.avium infection in ICAM-1 -/-BMMs in comparison with M. avium-infected WT BMMs (Fig 7C). Consistent with MAVS deficiency, loss of ICAM-1 expression on infected BMMs resulted in attenuated CD4 + T cell activation characterized as defined by decreased IFN-γ production (Fig 7D). In addition, loss of ICAM-1 expression also impaired ROS (Fig 7E) and iNOS (Fig 7F) production by M.avium-infected BMMs following coculture with antigen-specific and activated CD4 + T cells. Finally, ICAM-1 -/-AMs also showed limited control of a M. avium infection compared to WT AMs when cells were transferred to M. avium-infected WT mice (Fig 7G and S4B Fig).

Discussion
Our previous study demonstrated that M.tb releases mycobacterial RNAs into the cytosol of host cells and stimulates the cytosolic RIG-1/MAVS/TBK1/IRF7 RNA sensing pathway and production of type I IFNs 5 . A similar result has been observed in L. monocytogenes [4]. In this , and all data are representative of at least three independent experiments. In (E-G), the ratio between BMMs and CD4 + T cells was 1:2. n.s., not statistically significant, � P < 0.05, �� P < 0.01 and ��� P < 0.001 by Student's t-test (two-tailed). https://doi.org/10.1371/journal.ppat.1008569.g005

PLOS PATHOGENS
RNA sensing pathway in T cell mediated killing of mycobacteria study, we demonstrate that M.avium, like M.tb, activates the host cytosolic RIG-I/MAVS/ TBK1/IRF3/IRF7 RNA sensing pathway and stimulate type I IFN production in infected macrophages. This extends to in vivo as MAVS deficiency attenuates IFN-β production in mice  avium infection suggests that the RNA sensing pathway play significantly different roles in controlling these two mycobacterial infections in mice. We also found that the cytosolic RIG-I/MAVS/TBK1 but not IRF3/IRF7 RNA sensing pathway regulates ICAM-1 production in macrophages by inhibiting CRL4 COP1/DET1 -mediated ETV5 degradation in M. aviuminfected cells. Although the ICAM-I production is controlled by multiple transcription factors including NF-κB, Ets-2, ETV5, Ap-2, Sp-1 and Stat1 [15], our data indicates that ETV5 is a key transcription factor in M. avium-induced ICAM-1 expression in macrophages. The ICAM-1 production in M. avium-infected Mavs -/macrophages was restored by inhibition of the CRL4 COP1/DET1 enzyme complex.
The in vitro macrophage killing assay reveals that the RIG-I/MAVS/TBK1 RNA sensing pathway plays a limited role in controlling M.avium infection in BMMs. However, the mouse infection studies suggest that this pathway may be playing a larger role in vivo. CD4 + T cells are essential in protective immunity against M.tb and M.avium infection in animal models and patients 18 and their recruitment and/or activation may be dysregulated in the absence of RNA cytosolic RNA sensing pathway. Our findings that ICAM-1 expression is diminished in BMMs that lack RIG-I or MAVS expression following an M.avium infection suggest a link between RNA sensing pathways and an immune response to this bacterial infection. Multiple studies have shown ICAM-1 production is induced in host cells following various bacterial infections including M.tb, M.avium, S. pneumoniae and Salmonella [20][21][22][23][24], although the pathways leading to its expression were not defined in these studies. While the role of ICAM-1 during bacterial infections has been investigated using ICAM-1 -/mice, little is known about ICAM-1's function in promoting an antibacterial response within infected host cells [11,[23][24][25][26]. The engagement of ICAM-1 in APC-T cell interaction prompted us to test whether the host cytosolic RIG-I/MAVS/TBK1 RNA sensing pathway facilitates CD4 + T cell-mediated mycobacterial killing during an M.avium infection. Interestingly, M.avium-infected Mavs -/-BMMs fail to form immune synapse with CD4 + T cells isolated from M.avium-infected WT mice. Moreover, M.avium burden was significantly higher in Mavs -/vs WT macrophages cocultured with these CD4 + T cells. This defect in M.avium killing by Mavs -/macrophages was restored by stimulating ICAM-1 production through the inhibition of CRL4 COP1/DET1 . Consistent with the in vitro macrophage killing study, the adoptive transfer experiment using Mavs -/and ICAM-1 -/-AMs supports the importance of MAVS-dependent pathway and ICAM-1 in APC-T cell interaction during a mouse M.avium infection. While CD8 + T cells are also important in the immune response against M.tb infection, it may play a lesser role in immunity to M.avium at least in the context of a mouse infection [27]. Our results support this finding as CD8 + T cells isolated from M.avium-infected WT mice showed limited ability to kill M. avium when added to infected macrophages in a coculture system. Interestingly, except for ICAM-1, MAVS deficiency has no measurable effect on the cell surface expression of various other molecules involved in APC-CD4 + T cell synapse formation including MHC II, CD80, CD86 and CD40.

RNA sensing pathway in T cell mediated killing of mycobacteria
In the context of an NTM infection, the regulation of the various molecules that mediate adhesion between Ag-specific CD4 + T cells and infected macrophages remains poorly defined. In this study we observed that the RIG-I/MAVS/TBK1/ETV5/ICAM-1 macrophage response to a M. avium infection significantly increases the CD4 + T cell-mediated activation and the antimycobacterial response. Our data suggest that this MAVS-dependent antimicrobial response is not mediated by IFN-γ. Evidence to support this conclusion includes: 1) When cocultured with CD4 + T cells, the MAVS/ICAM-1-dependent control of the M. avium infection was only partially mimicked by blocking IFN-γ following incubation of activated CD4 + with infected BMM. 2) A similar percentage of IFN-γ-producing CD4 + T cells and a comparable level of IFN-γ production was detected in the lung of M. avium-infected WT and Mavs -/ mice. 3) CD4 + T cells isolated from M. avium-infected Mavs -/and WT mice have a comparable activity in stimulating antimycobacterial activity when added to M. avium-infected WT BMMs. In addition, a similar M. avium burden was detected in WT AMs that were transferred into WT and Mavs -/mice. Altogether, our data indicates that there is a cell-cell contact dependent mechanism by which CD4 + T cells promote M. avium killing in host cells. Mechanisms for the T cell mediated response could include macrophage activation by the CD40-CD40L axis and downstream effects such as iNOS expression and NO production [28,29]. However, we cannot completely rule out that the MAVS -/and ICAM -/infected macrophages may also be more refectory to cytokine activation.
In summary (Fig 8), our study reveals a previously undefined RIG-I/MAVS dependent mechanism for inducing ICAM-1 expression in macrophages following an M. avium infection and its role in CD4 + T cell-mediated mycobacterial killing. Additionally, in contrast to M.tb, we identified that activation of the RNA sensing pathway is beneficial to the host immune response to an M. avium infection. At present we lack a clear understanding as to for the differential response to activation of RIG-1/MAVS following the two different mycobacterial infections. To address this question additional studies are needed to determine which host factors are differentially regulated by activation of RIG-1/MAVS during an M.tb and M. avium infection including how ICAM-1 expression is induced during an M.tb infection.

Mice
Wild-type C57BL/6 and Mavs -/mice have been described previously 5 . ICAM-1 -/mice on a C57BL/6 background were purchased from Jackson Laboratory (Stock No: 002867). All mice were housed and bread at the institutional animal facility under specific pathogen-free conditions. Mice were housed with standard housing in cages containing 3-5 mice per cage.

Ethic statement
The University of Notre Dame is accredited through the Animal Welfare Assurance (#A3093-01). All animal experiments were approved by the University of Notre Dame's Institutional Animal Care and Use Committees (IACUCs) (protocol approval number # 17-02-3630). Isoflurane was used as the mouse anesthesia and mice were killed by cervical dislocation following isoflurane inhalation.

RNA purification
To measure mycobacterial RNA in the cytosol of M.avium-infected macrophages, cytosolic RNA was purified as described previously 5 . For the measurement of host cytokine and ETV5 mRNA, BMMs were infected with M.avium at an MOI of 10 at 37˚C and 5% CO 2 for various time as indicated and then total cellular RNA was purified using Qiagen RNeasy Plus Mini Kit (Cat.No. 74136, Qiagen).

Whole-cell lysates and nuclear fraction preparation
BMMs were left uninfected or infected with M.avium 104 at an MOI of 10 for 24 hr at 37˚C and 5% CO2, and then washed with pre-cold PBS three time. Whole-cell lysates (WCL) and nuclear fraction were prepared as described previously [5].

Mouse infection assay
Mavs -/and wild-type C57BL/6 mice (8-10 weeks old, female) were intratracheally infected with M.avium 104 at a dose of 1X10 7 CFU as described previously [32]. M.avium input in the lung of mice was determined at 5 hr post-infection. At 1, 2, 3, 5, 8 and 12 weeks post-infection, mouse serum was obtained/prepared via cardiac puncture and BD Microtainer Serum Separator Tube. Mouse lungs and spleens were also harvested, homogenized, and plated onto Middlebrook 7H10 agar plates, and mycobacterial colonies were counted after 10-14 days of incubation at 37 o C. For pathological analysis, mouse lung sections were prepared and stained with H&E at the Histology Core Facility of University of Notre Dame, and histopathological score was evaluated as described previously [33]. M.avium infection was carried out in the biosafety level 2 laboratory. Isoflurane was used as the mouse anesthesia and mice were killed by cervical dislocation following isoflurane inhalation.

Mouse cytokine array analysis
Mouse blood was harvested by cardiac puncture at 3 weeks post M.avium infection and serum was prepared using BD Microtainer Serum Separator Tube as described above. The abundance of mouse cytokines in serum was analyzed using Proteome Profiler Mouse XL Cytokine Array (Cat. ARY028, R&D Systems) according to manufacturer's instruction. The intensity of the spots was determined by ImageJ software.

Survival assay of M.avium in mouse BMMs and AMs
Cells were infected with M.avium 104 at an MOI of 10 for 1 hr at 37˚C and 5% CO 2 , and then washed with complete medium thrice, and incubated for another 24 and 72 hr at 37˚C and 5% CO 2 . At the given time, cells were washed with pre-cold PBS thrice and lysed in 0.05% SDS. A series of dilution of cell lysates in PBS (1x) were added onto 7H10 agar plates supplemented with 10% (v/v) OADC and 0.2% glycerol. Plates were incubated at 37˚C for 10-15 days until counting.

M.avium survival assay in BMMs cocultured with T cell
Mouse BMMs were differentiated as described above in complete DMEM medium. Seven days after incubation, cells were washed and resuspended in BMM media and seeded into 12-well plates (1x10 5 cells/well) overnight at 37˚C and 5% CO 2 . BMMs were then infected with M.avium strains for 1 hr at an MOI of 10 at 37˚C and 5% CO 2 and subsequently washed with complete RPMI-1640 medium thrice. Infected cells were further incubated at 37˚C and 5% CO 2 overnight. These cells were used for the BMM-T cell coculture assay. Mouse CD4 + T cells and CD8 + T cells were isolated from the lung of uninfected and M.avium-infected WT or Mavs -/mice (3 weeks post-intratracheal infection as described above). T cells were isolated using MojoSort™ Mouse CD4 + T Cell Isolation Kit (Cat. 480005, Biolegend) and MojoSort™ and mouse CD8+ T Cell Isolation Kit (Cat. 480007, Biolegend). Purified CD4 + T cells and CD8 + T cells were added to the M.avium-infected BMMs culture and incubated for 1, 24 and 72 hr at 37˚C and 5% CO 2 . Cocultured cells were washed with pre-cold PBS thrice and then lysed in 0.05% SDS and spread on 7H10 agar plates containing 10% OADC. Plates were incubated at 37˚C for 10-15 days until counting and results were expressed as Log 10 .

Triple-cell population coculture assay
A two-cell population (BMM-T cells) coculture was first established as described above, Briefly, WT BMMs were initially infected with Gen S M.avium at MOI of 10 for 24 hr at 37˚C and 5% CO 2 and then cocultured with CD4 + T cells isolated from WT M.avium-infected C57B/6 mice (isolated from mice 3 weeks post-infection). 24 hr after the addition of CD4+ T cells to the infected macrophages this two-cell population was used for Triple-cell Population Coculture Assay. WT, Mavs -/and ICAM-1 -/-BMMs were infected with Gen R M.avium at an MOI of 10 for 1 hr at 37˚C and 5% CO 2 . Cells were then washed with complete medium thrice, and incubated for another 24 at 37˚C and 5% CO 2 . Subsequently, these infected WT, Mavs -/and ICAM-1 -/-BMMs were added into the two-cell population coculture at a ratio: Gen S M. avium-infected BMMs: Gen R M.avium-infected BMMs:T cells = 4:1:10. Mixed cells were incubated for another 1, 24 and 72 hr at 37˚C and 5% CO 2 . Cocultured cells were washed with precold PBS thrice and then lysed in 0.05% SDS. Gen R M.avium CFU was determined using 7H10 agar plates supplemented with gentamycin (200μg/ml) and 10% OADC. Total M.avium CFU was determined using 7H10 agar plates plus 10% OADC. Gen S M.avium CFU was calculated by the formula: total M.avium CFU-Gen R M.avium CFU.

Confocal microscopy analysis for macrophage-T cell synapse
C57BL/6, Mavs -/or ICAM-1 -/-BMMs were infected with an M.avium strain expressing tdTomato at an MOI of 10. Twenty-four hours post infection, infected cells were cocultured with CD4 + T cells isolated from WT M.avium-infected mice at a ratio (BMM:T cell) of 1:2. Twentyfour hours post incubation, cells were gently washed with precold PBS thrice, and fixed in 2% PFA for 15 min at RT, and then permeabilized in 0.1% Triton X-100 in PBS for 15 minutes at RT. Cells were blocked in PBS containing 2% FBS and further stained with Alexa Fluor™ 647 phalloidin (Cat. A22287, Invitrogen) and Alexa Fluor 488 anti-mouse CD4 Antibody (Cat. No. 100529, Biolegend), and analyzed by Nikon C2+ confocal laser scanning microscope. For quantitative analysis, at least 100 BMMs per condition were counted in multiple random areas of the slide. Data was normalized to the number of T cells conjugated to M.avium-infected WT BMMs for one biological sample, which was set to 100%.

Alveolar macrophage adoptive transfer assay
Mouse AMs were isolated from WT, Mavs -/or ICAM-1 -/mice (Female, 8 to 10 weeks old) as described above, and then infected with Gen R M.avium strains at an MOI of 10 for 24 hr at 37˚C and 5% CO 2 . Infected AMs were washed with pre-cold PBS thrice and then intratracheally injected into mice at a dose of 2x10 5 cells/mouse in 15 μl PBS. At the time of AM injection, recipient mice had been infected with Gen S M.avium for 3 weeks as described [32]. At day 1, 3 and 6 post injection, mouse lungs were harvested, homogenized, and spread on 7H10 plus 10% OADC with or without gentamycin (200μg/ml). Bacterial number was counted 10-15 days after incubation at 37˚C. For AM survival assay in mice after adoptive transfer, Gen R M. avium-infected AMs were labeled with Vybrant1 CFDA SE Cell Tracer Kit before intratracheal injection. Single mouse lung cells were prepared as described previously at day 1, 3 and 6 post injection 34 , and analyzed by BD LSRFortessa X-20 cytometer.

Intracellular ROS measurement
Mouse WT, Mavs -/or ICAM-1 -/-BMMs were infected with WT M.avium at an MOI of 10 for 24 hr at 37˚C and 5% CO 2 , and then cocultured with/without CD4 + T cells isolated from WT mice (Female, 8 to 10 weeks old) previously infected with WT M.avium for 3 weeks. Intracellular ROS concentration in infected BMMs was measured at various time points post coculture using Fluorometric Intracellular Ros Kit (Cat. MAK142-1KT, Sigma-Aldrich) following the manufacturer's instruction. The fluorescence intensity was measured using BioTek Synergy XL microplate reader.

Statistical analysis
Statistical analysis was performed to determine differences between groups by two-tailed Student's t-tests or Mann-Whitney U test using GraphPad Prism software (Version 5.04, Graphpad Software). P-value < 0.05 was considered significant.