Metallothionein 3-Zinc Axis Suppresses Caspase-11 Inflammasome Activation and Impairs Antibacterial Immunity

Non-canonical inflammasome activation by mouse caspase-11 (or human CASPASE-4/5) is crucial for the clearance of certain gram-negative bacterial infections, but can lead to severe inflammatory damage. Factors that promote non-canonical inflammasome activation are well recognized, but less is known about the mechanisms underlying its negative regulation. Herein, we identify that the caspase-11 inflammasome in mouse and human macrophages (Mϕ) is negatively controlled by the zinc (Zn2+) regulating protein, metallothionein 3 (MT3). Upon challenge with intracellular lipopolysaccharide (iLPS), Mϕ increased MT3 expression that curtailed the activation of caspase-11 and its downstream targets caspase-1 and interleukin (IL)-1β. Mechanistically, MT3 increased intramacrophage Zn2+ to downmodulate the TRIF-IRF3-STAT1 axis that is prerequisite for caspase-11 effector function. In vivo, MT3 suppressed activation of the caspase-11 inflammasome, while caspase-11 and MT3 synergized in impairing antibacterial immunity. The present study identifies an important yin-yang relationship between the non-canonical inflammasome and MT3 in controlling inflammation and immunity to gram-negative bacteria.


INTRODUCTION
Gram-negative bacteria that cause more than 30% of the total healthcare-associated infections worldwide remain a global concern of morbidity and mortality (1). Assembly of inflammasome complexes during bacterial pathogenesis drives robust inflammation and shapes antibacterial immune responses. The non-canonical inflammasome is activated when innate immune cells such as Mf sense bacterial ligands in the cytosol (2). LPS from bacterial cell walls enters the cytosol during bacterial escape from vacuoles or via rupture of outer membrane vesicles (OMV) (3). iLPS directly binds caspase-11, triggering the non-canonical inflammasome cascade, followed by activation of pro-caspase-1 to caspase-1, processing of pro-IL-1b to mature IL-1b and pyroptosis, a lytic form of programmed cell death. Activation of gasdermin D (GSDMD) by caspase-11 and caspase-1, leads to pore formation on the cell membrane facilitating the exit of IL-1b from Mf (4). Unrestricted activation of this inflammatory cascade is a major underlying cause of tissue damage and sepsis-associated mortality (2). Thus, it is crucial to thoroughly understand the molecular cues that guard against excessive activation of the caspase-11 inflammasome. Although factors that promote noncanonical inflammasome activation have been extensively studied, less is known about the mechanisms that negatively regulate it.
MTs are Zn 2+ regulating proteins induced by endogenous and exogenous stimuli including cytokines, infection, oxidative stress and heavy metals (5,6). Intracellular availability of total Zn 2+ , exchangeable Zn 2+ and Zn 2+ redistribution among proteins is tightly regulated by MTs (7). Mice have 4 MT isoforms (MT1-4), whereas more than 16 MT isoforms are present in humans (8). Our knowledge on the role of the MT family in immune responses largely emerges from studies on MT1 and MT2. We and others have shown that MT1 and MT2 promote antifungal and antibacterial immunity in Mf primarily via Zn 2+ sequestration (9,10). MT3, on the other hand, suppresses manifestation of proinflammatory phenotypic and metabolic changes and impairs antifungal immunity in vitro in Mf and in vivo. Mf expression of MT3 is inducible by IL-4 stimulation. In these cells, MT3 increases intracellular free-Zn 2+ , promotes Zn 2+ uptake by a prototypic intracellular pathogen and favors microbial survival (11). Thus, MTs and their ability to regulate Zn 2+ homeostasis intricately ties Mf inflammatory responses to antimicrobial defense.
Zn 2+ is indispensable in many biochemical processes due to its role in structural and catalytic functions of enzymes and macromolecules. Zn 2+ excess or deficiency compromises the development and function of immune cells including monocytes and Mf, leading to increased risk of infection (12). Zn 2+ also has a profound role as a signaling ion attributable to the transient changes in intracellular exchangeable Zn 2+ . LPS triggers increased Zn 2+ import in leukocytes, monocytes and Mf (13,14). In peripheral blood mononuclear cells (PBMCs) and human Mf (hMf), LPS-induced Zn 2+ influx promotes IL-1b production (14). In contrast, exogenous exposure to Zn 2+ in human monocytes may reduce IL-1b production due to inhibition of nucleotide phosphodiesterases (15). Thus, the effects of Zn 2+ on signaling and cytokine production are context and Zn 2+ concentration-dependent. Changes in ion flux, specifically K + and Clegress and intracellular mobilization of Ca 2+ underlie canonical nod-like receptor pyrin domain containing-3 (NLRP3) inflammasome activation (16). Intriguingly, Zn 2+ exerts disparate effects on activation of this cascade. Long-term Zn 2+ depletion disrupts lysosomal integrity leading to increased activation of the canonical NLRP3 inflammasome (17). On the other hand, shortterm chelation of Zn 2+ attenuates the canonical pathway due to impaired function of the pannexin-1 receptor (18).
The importance of Zn 2+ regulation by MTs in the non-canonical inflammasome pathway remain unexplored. Given the suppressive role of MT3 in Mf inflammatory responses (19), we hypothesized that MT3 negatively regulates the highly inflammatory caspase-11 activation cascade. Bioinformatics analysis predicted the involvement of MT3 in regulating non-canonical inflammasomeassociated pathways. Using a combination of protein-protein interaction network analysis, immunological and massspectrometric approaches, we demonstrate that triggering caspase-11 activation results in a profound, gradual increase in the Mf Zn 2+ pool mediated by MT3. The increase in Zn 2+ attenuates signaling via toll/interleukin-1 receptor (TIR) domain containing adaptor-inducing interferon (IFN)b -interferon regulatory factor 3 -signal transducer and activator of transcription factor 1 (TRIF-IRF3-STAT1), a pathway that is prerequisite for caspase-11 inflammasome activation (20). Zn 2+ deficiency augments, whereas Zn 2+ supplementation suppresses the non-canonical inflammasome in Mf. Using whole-body Mt3 -/and myeloid-MT3-deficient mice, we elucidate that MT3 blunts noncanonical inflammasome activation in vitro and in vivo upon challenge with iLPS or gram-negative bacteria but not grampositive bacteria. Importantly, this function of MT3 is conserved in hMf. Although caspase-11 and MT3 form a negative regulatory loop, we find that these two molecules synergize in compromising antibacterial immunity. Our data uncover a previously unknown yin-yang relationship whereby the MT3-Zn 2+ axis exerts a brake on non-canonical inflammasome activation but the functions of MT3 and caspase-11 converge in crippling immunity to invading bacteria (see Graphical Abstract).

MT3 Suppresses Activation of the Non-Canonical Inflammasome In Vitro
MT3 attenuates cell death in neuronal and glial cells, but the precise underlying mechanisms are not fully understood (21). As non-canonical inflammasome activation leads to pyroptotic cell death, we investigated if MT3 effector function is related to caspase-11 activation in Mf. We explored whether MT3 is involved in inflammatory and cell-death processes using functional enrichment analysis. We assessed protein-protein interaction networks of MT3 in Mus musculus and Homo sapiens using the STRING database (22). The MT3 interaction partners significantly enriched 15 mouse and 43 human gene ontology categories for biological processes (GO BP) related to programmed cell death (PCD), LPS responses, signaling via TRIF, IL-1 and type-I IFN, cytokine responses and several immune processes. A complete network of MT3 interactions and GO BP categories is in Table 1, Supplementary Table S1, Supplementary Files S1 and S2. PCD and LPS responses are linked to inflammasome activation and more specifically, TRIF and type-I IFN signaling are tied to the non-canonical inflammasome pathway. A lack of TRIF signaling ablates noncanonical inflammasome activation in response to iLPS without impacting Mf response to canonical NLRP3 triggers such as ATP and nigericin (20). Thus, our bioinformatics analysis together with our previously reported role for MT3 in suppressing proinflammatory responses in Mf led us to investigate whether MT3 negatively regulates the noncanonical inflammasome pathway.
In Mf, exposure to iLPS triggers caspase-11 activation and cell death by pyroptosis (2). We directly assessed if MT3 regulates non-canonical inflammasome activation using WT and Mt3 -/mice. BMDMf were exposed to iLPS or vehicle control and time-dependent changes in the gene expression of Mt1, Mt2 and Mt3 were examined. Mt3 expression increased gradually from 1 hour (h) and peaked at 48h in WT BMDMf challenged with iLPS ( Figure 1A). The expression of Mt1 and Mt2 peaked at 6h, but receded over time in both WT and Mt3 -/-BMDMf (Supplementary Figures S1A, B). To determine whether exogenous LPS had an effect on MT3, we stimulated WT BMDMf with extracellular LPS (exLPS) for 48h. While iLPS increased Mt3 expression by 10-fold, the fold increase observed with exLPS challenge was much lower ( Figure 1B).

MT3 Represses CASPASE-4 Activation and Antibacterial Resistance in Human Mf
Gram-negative bacteria activate the non-canonical inflammasome via CASPASE-4 in hMf (26). We investigated whether human MT3, similar to mouse MT3, suppressed noncanonical inflammasome activation and antibacterial immunity. Human monocyte-derived Mf obtained from PBMCs were transfected with scramble siRNA or MT3 siRNA followed by transfection with iLPS. To assess siRNA specificity, we analyzed the expression of MT3 and MT2A genes. MT3, but not MT2A expression was silenced in MT3 siRNA transfected hMf ( Figure 2A). MT3 deficiency resulted in elevated activation of CASPASE-4 and heightened release of IL-1b from hMf ( Figures 2B, C). We previously showed that a lack of MT3 increased resistance of mouse BMDMf to Escherichia coli (19). We therefore queried whether silencing MT3 in hMf impaired bacterial clearance in vitro. hMf treated with scramble siRNA or MT3 siRNA were infected with E. coli K12 for 24h. MT3deficient hMf exerted a sharp decline in intracellular bacterial survival compared to control hMf ( Figure 2D).

MT3 Dampens Antibacterial Resistance and Caspase-11 Inflammasome Activation
In Vitro and In Vivo We examined whether MT3 regulated antibacterial immunity and non-canonical inflammasome activation in vitro and in vivo.
WT and Mt3 -/-BMDMf were infected with E. coli. After 24h, bacterial survival was reduced in Mt3 -/-BMDMf ( Figure 2E). Next, we infected WT and Mt3 -/mice in vivo intraperitoneally (i.p.) with E. coli for 6h. Compared to WT mice, MT3 deficiency bolstered bacterial elimination from the blood and moderately improved bacterial clearance in the kidney and peritoneal lavage ( Figure 2F). Caspase-11, GSDMD (N-terminal) and caspase-1 activation were heightened in kidney homogenates of infected Mt3 -/mice compared to infected WT mice ( Figure 2G). The decrease in pro-GSDMD of Mt3 -/mice may be explained by increased conversion of pro-to active-GSDMD form ( Figure 2G). IL-1b in the peritoneal lavage was significantly elevated (p<0.01) and serum IL-1b exhibited a trend towards increase in infected Mt3 -/mice compared to WT controls ( Figure 2H). To determine if this response is consistent upon LPS challenge in vivo, we primed mice i.p. with poly(I:C) for 6h and challenged them i.p. with ultrapure LPS. After 18h, IL-1b was elevated in the peritoneal lavage and serum of Mt3 -/mice compared to WT mice ( Figure 2I). We further queried the impact of MT3 on LPS-induced sepsis. WT and Mt3 -/mice were challenged with ultrapure LPS (20 mg/kg) and assayed for weight loss, murine sepsis scores (MSS) as reported previously (27) and survival. MT3 deficiency resulted in greater weight loss and increased sepsis scores, but both genotypes similarly succumbed to septic shock (Supplementary Figures S2A-C).
We then investigated whether MT3 increased susceptibility to other gram-negative bacteria. WT and Mt3 -/mice were infected intranasally (i.n.) with a virulent, heavily encapsulated strain of Klebsiella pneumoniae (KP2 2-70). MT3 deficiency significantly (p<0.05) improved K. pneumoniae clearance in the spleen, but no changes were observed in the lung and kidney ( Figure 2J). Grampositive bacteria activate caspase-11 via the NLRP6 inflammasome (28). We determined whether the increased non-canonical inflammasome activation and antibacterial resistance observed in Mt3 -/mice extended to gram-positive bacterial infection. WT and Mt3 -/mice were challenged subcutaneously (s.q.) with a clinical isolate of Group-A-Streptococcus GAS5448 (29). After 72h, Mt3 -/mice manifested significantly (p<0.05) reduced GAS burden in the kidney and spleen compared to WT mice. Bacterial CFUs in the blood exhibited a similar trend (Supplementary Figure S2D). Importantly, although Mt3 -/mice exhibited higher activation of caspase-1 and IL-1b, the levels of active caspase-11 and procaspase-11 were diminished in GAS-infected Mt3 -/mice compared to WT mice (Supplementary Figure S2E). Thus, although MT3 compromises resistance to both gram-negative and gram-positive bacteria, it specifically suppresses noncanonical inflammasome signaling in response to gram-negative microbial triggers.
Mt3 -/mice were treated i.p. with MCC950 (NLRP3 inhibitor) followed by infection i.p. with E. coli. Treatment with MCC950 reduced IL-1b levels and sharply blunted antibacterial resistance in the blood and kidney of Mt3 -/mice compared to vehicletreated controls. Bacterial burdens were also elevated in WT mice by NLRP3 inhibition ( Figure 3C). Collectively, these data demonstrate that MT3 negatively controls activation of the non-canonical inflammasome and that both MT3 and caspase-11 cripple resistance to bacterial infection.

Myeloid MT3 Orchestrates Negative Control of the Non-Canonical Inflammasome
To affirm that the effects on caspase-11 inflammasome activation observed in the Mt3 -/mice were dependent on myeloid-MT3, we (C) Western blots of pro-caspase-11, active-caspase-11, pro-GSDMD, active-GSDMD (p31), pro-caspase-1, active-caspase-1, pro-IL1b and active-IL-1b in whole kidney homogenates of mice infected as above, n = 3-5 per group, two-tailed t-test. (D) Bacterial CFUs in kidney and whole blood of Lys2Cre and Lys2Cre Mt3 fl/fl mice infected i.p. with E. coli (1 X10 9 CFUs/mouse) for 6h, n = 3-5 per group, two-tailed t-test, data are mean ± SEM. **p < 0.01, ***p < 0.001. MT3 Exerts a Brake on the TRIF-IRF3-STAT1 Axis to Curtail Caspase-11 Signaling Signaling via the TRIF pathway is crucial for caspase-11 activation and synergistic engagement of the NLRP3 inflammasome leading to activation of caspase-1 and IL-1b (20). Downstream of TRIF, IRF3 and IRF7 induce IFNb production that activates STAT1 signaling and promotes transcription of inflammasome components including caspase-11 and guanylate binding proteins (GBPs). GBP2 and GBP5 facilitate LPS release into the cytosol from intracellular vacuoles containing bacteria (39,40). Our functional enrichment data based on protein-protein interaction network analyses revealed a potential involvement of MT3 in LPS, TRIF, type-I IFN and IL-1 signaling ( Table 1, Supplementary Table S1 and Supplementary Files S1, S2). We further examined our published RNA-seq data (NCBI SRA: PRJNA533616) to determine differentially expressed genes by comparing resting WT and Mt3 -/-BMDMf (19). The derived list of differentially expressed genes significantly enriched 12 GO BP categories directly related to cytokine and chemokine signaling and regulation of inflammatory responses based on DAVID functional enrichment analysis ( Figure 5A) (41). These analyses suggested that MT3 deficiency perturbed the expression of immune-related genes even at the resting state. We reported that a lack of MT3 augments IFNg responsiveness (19). Herein, from our RNA-seq analysis, we identified 20 genes related to IFN-signaling that were upregulated in resting Mt3 -/-BMDMf compared to resting WT BMDMf (p adj <0.05) ( Figures 5B, C). Among these, Isg15, Mx1 and Ifit (Ifit1bl1, Ifit3, Ifit3b, Ifit2, Ifit1, Ifit1bl2) family of genes are known targets of type-I IFNs (42)(43)(44)(45). These observations led us to posit that MT3 regulated the cellular response to LPS challenge by modulating the TRIF-IRF3-STAT1 axis upstream of non-canonical inflammasome activation. LPS engages the TRIF-IRF3-STAT1 axis via toll-like receptor 4 (TLR4) signaling in Mf. To address this hypothesis, we challenged WT and Mt3 -/-BMDMf with iLPS or vehicle and examined activation of the TRIF-IRF3-STAT1 pathway. Mf lacking MT3 exerted increased activation of phospho-IRF3 (pIRF3), pSTAT1, GBP2 and GBP5 ( Figure 5D). TRIF protein levels were unaltered by MT3 deficiency ( Figure 5E). Type-I IFN signaling is required for activation of the caspase-11 inflammasome cascade by gram-negative bacteria (20). As MT3 deficiency augmented the expression of genes involved in IFN signaling ( Figures 5B, C), we blocked the interferon-a/b receptor (IFNAR) 1 using a monoclonal antibody prior to iLPS challenge in WT and Mt3 -/-Mf. IFNAR1 blockade resulted in decreased pro-caspase-11 (p43 subunit). Total STAT1 and pro-caspase-1 (p38 subunit) were not greatly affected (Supplementary Figure S5). We found robust attenuation of pSTAT1, active-caspase-11 and active-caspase-1, but secretion of active-IL-1b in both WT and Mt3 -/-Mf was increased upon blockade of IFNAR1 signaling. This finding corresponded with high pro-IL-1b levels in Mf treated with the IFNAR1 antibody (Supplementary Figure S5). These data indicate that although IFNAR1 signaling is required for fueling the non-canonical inflammasome cascade and activation of caspase-1, pro-IL-1b and its activation are suppressed by IFNAR1. We queried if MT3 exerted a brake on TRIF signaling to downmodulate non-canonical inflammasome activation. WT and Mt3 -/-BMDMf were treated with scramble or Ticam1 (gene encoding TRIF) siRNA, and challenged with iLPS ( Figure 5F). Ticam1 silencing reversed the effects of MT3 deficiency resulting in a sharp reduction in caspase-11 and IL-1b activation in Mf ( Figure 5F). These data reveal a central role for MT3 in attenuating the crosstalk between TRIF signaling and the caspase-11 activation cascade.

Zn 2+ Flux by MT3 Drives Suppression of the Non-Canonical Inflammasome in Mf
MTs are master regulators of intracellular Zn 2+ availability and distribution (46,47). We determined if negative control of the non-canonical inflammasome by MT3 was Zn 2+ -dependent. First, we systematically assessed Zn 2+ changes in WT and Mt3 -/-BMDMf upon challenge with iLPS over time using SEC-ICP-MS. Activation of the non-canonical inflammasome in WT Mf was associated with profound changes in the intracellular Zn 2+ pool. iLPS exposure led to a gradual increase in total Zn 2+ largely associated with the chromatogram peak(s) between 18-21 min. that we previously identified as MTs ( Figures 6A, B) (9,11). The time-dependent elevation in Zn 2+ corresponded with kinetics of Mt3 induction ( Figures 1A and 6A, B). Mf lacking MT3 failed to elevate total Zn 2+ and MT-associated Zn 2+ in response to iLPS ( Figure 6B). In contrast to the increase in Zn 2+ pool observed in WT Mf challenged with iLPS, resting Mt3 -/-Mf harbored higher Zn 2+ content that reduced over time post iLPS challenge ( Figure 6B). These data indicate that MT3 drives an elevation in intracellular Zn 2 + in Mf during non-canonical inflammasome activation.
Zn 2+ chelation in human monocytes increases IRF3 activation (48). We reasoned that if the effects of MT3 were Zn 2+ dependent, altering the intracellular Zn 2+ concentration will at least in part reverse the heightened non-canonical inflammasome signaling observed in Mt3 -/cells. To test this postulate, we exposed WT and Mt3 -/-BMDMf to increasing amounts of ZnSO 4 and challenged them with iLPS in vitro. Exogenous ZnSO 4 supplementation remarkably reduced the ability of Mt3 -/-Mf to respond to iLPS. pIRF3, pSTAT1 and activation of caspase-11 were reduced in ZnSO 4 -supplemented Mt3 -/-Mf (Supplementary Figure S6A). A similar effect of Zn 2+ was also observed in WT Mf (Supplementary Figure S6A). We investigated if exposing WT Mf to a Zn 2+ -deficient environment would mimic the effects MT3 deficiency on the non-canonical inflammasome. WT BMDMf were cultured in Zn 2+ -sufficient or Zn 2+ -deficient Opti-MEM media prior to iLPS exposure. WT Mf exposed to a Zn 2+ -deficient milieu manifested higher pIRF3 and caspase-11 activation accompanied by increased activation and release of IL-1b ( Figure 6C). The amount of TRIF, pro-caspase-11 and pro-IL-1b were not affected by Zn 2+ deficiency ( Figures 6C and Supplementary Figure S6B).
Next, we directly addressed whether Zn 2+ is required for the suppressive function of MT3 on the caspase-11 inflammasome. We first overexpressed the MT3 gene in   66 Zn 2+ isotope ( Figure 6D). These data confirm that transfection of the MT3-Zn 2+ complexes resulted in an increase in intracellular 66 Zn 2+ in Mf. In parallel, we isolated cell lysates and supernatants proteins from these Mf to determine whether apo-MT3 or the MT3-Zn 2+ complexes modulated the non-canonical inflammasome pathway. Transfection of MT3-4Zn 2+ and MT3-6Zn 2+ but not apo-MT3, dampened pIRF3, active caspase-11 and active IL-1b in response to iLPS ( Figure 6E). TRIF levels were unaffected by MT3 transfection ( Figure 6E). Pro-caspase-11 and pro-IL-1b were modestly diminished by MT3-4Zn 2+ and MT3-6Zn 2+ exposure, but these changes were not significant ( Figure 6E). The effect of MT3-6Zn 2+ was more profound than that of MT3-4Zn 2+ indicating that a higher Zn 2+ saturation on MT3 corresponded with a stronger suppressive effect on the non-canonical inflammasome ( Figure 6E). Taken together, these findings reveal a previously undescribed interplay between the non-canonical inflammasome and its negative regulator, whereby the MT3-Zn 2+ axis suppresses caspase-11 inflammasome, but the two molecules concur in compromising immunological fitness of the host during bacterial pathogenesis.

DISCUSSION
Human CASPASE-4 or mouse caspase-11 are inflammatory caspases that drive cell death via pyroptosis. These caspases directly recognize bacterial LPS in the cytosol resulting in CASPASE-4 or caspase-11 auto-processing and synergistic activation of the NLRP3 inflammasome that culminates in caspase-1 activation, processing and release of IL-1b and IL-18 (2,32). While the non-canonical inflammasome boosts host immunological fitness to some bacterial infections, heightened activation of this cascade poses the danger of tissue injury and organ failure. Thus far, negative regulation of IFNb production by prostaglandin E2, immunity-related GTPases M clade, cyclicadenosine monophosphate, and low dose oxidized phospholipid oxPAPC have been shown to thwart activation of the noncanonical inflammasome (49)(50)(51)(52). The role of MTs in regulating inflammasome activation pathways has largely been unknown. Herein, we identify a previously undescribed function of MT3 in curtailing the highly inflammatory non-canonical inflammasome activation cascade via Zn 2+ regulation. We demonstrate that while MT3 orchestrates negative regulation of the caspase-11 inflammasome, the combined presence of MT3 and caspase-11 blunts resistance to E. coli infection in vivo. These studies illuminate a central role for the MT3-Zn 2+ axis in shaping the intricate balance between host antibacterial immunity and unrestrained inflammation.
MT1 and MT2 are ubiquitously expressed and can be induced by infection (53)(54)(55)(56). Initial studies on MT3 revealed tissuerestricted expression with high levels predominantly found in the brain tissue where it inhibits neuronal cell death (21,57). The immunological functions of MT3, particularly in the innate arm have only recently been investigated (11,19,21). We reported that Mt3 is inducible by the pro-resolving cytokines IL-4 and IL-13 in Mf. One inducer of Mt3 expression is STAT6 signaling, and this MT is crucial in shaping the phenotypic and metabolic attributes of Mf stimulated with type-2 cytokines (11,19). Studies on MTs in response to exogenous LPS stimulation have largely focused on MT1 and MT2. Monocytes and Mf induce MT1 and MT2 upon extracellular LPS exposure (58,59). We found that iLPS challenge also induced Mt1 and Mt2 in Mf, although their expression receded to baseline over time. In contrast, Mt3 expression gradually increased as non-canonical inflammasome activation progressed. Subversion of inflammatory responses in Mf by MT3 and the delayed expression pattern in response to a non-canonical inflammasome trigger are reminiscent of waning inflammation after the initial peak of inflammasome activation has subsided (19). In line with this hypothesis, protein interaction network analysis predicted the involvement of MT3 in cellular responses related to non-canonical inflammasome activation. Mf lacking MT3 exerted robust activation of caspase-11, caspase-1 and IL-1b. Similar to our observations with mouse MT3, a lack of human MT3 exacerbated the activation of CASPASE4 and IL-1b in hMf. Human and mouse MT3 proteins that share 86% identity thus have consistent roles that culminate in negative regulation of the non-canonical inflammasome cascade in Mf (60). As noncanonical inflammasome activation progressed, MT3 guarded against its unrestrained activation to avert potential inflammatory damage. Together, these observations reveal a pivotal role for MT3 in curtailing the vigor of the caspase-11 activation cascade. Although Mt3 -/mice exerted higher sepsis scores and weight loss, they succumbed to septic shock similar to WT controls, suggesting that a threshold level of caspase-11 activation may be sufficient to promote sepsis-associated mortality.
In vivo, LPS released from OMV of gram-negative bacteria triggers caspase-11 activation (2, 3). Myeloid-MT3 contributed to averting excessive activation of caspase-11 and synergistic activation of the canonical inflammasome in response to gram-negative microbial triggers. MT3 compromised antibacterial resistance to E. coli and K. pneumoniae, but this was not due to its suppressive action on the caspase-11 inflammasome in vivo. Instead, Mt3 -/and Casp-11 -/mice manifested improved antibacterial immunity, an effect that was further augmented when Mt3 -/mice lacked caspase-11. The synergism between MT3 and caspase-11 may result from independent or combined effects of MT3 and caspase-11 in vivo. Of note, the activation of caspase-1 and caspase-8 in the absence of MT3 and caspase-11 reveal that canonical inflammasome activation was operational and both caspase-1 and caspase-8 may contribute to IL-1b activation in vivo (38).
Mf utilize Zn 2+ deprivation and Zn 2+ intoxication mechanisms as strategies for antimicrobial defense (56,61,62). We previously showed that ablation of MT3 in Mf augments immunity to Histoplasma capsulatum as well as E. coli. The increased antimicrobial resistance in Mt3 -/-Mf is at least partially attributable to a decrease in the Mf exchangeable Zn 2+ pool and exaggerated IFNg responsiveness (11,19). Myeloid and non-myeloid cells may together contribute to bacterial elimination in whole-body MT3-deficient mice. Nonetheless, the augmented bacterial elimination observed in Lys2Cre Mt3 fl/fl mice indicates that myeloid-MT3 contributes to the suppression of antibacterial defenses in vivo. The finding that MT3 deficiency dually bolstered inflammasome activation and antibacterial immunity underpins a role for this protein in suppressing the emergence of a proinflammatory phenotype in Mf. Our data unveil a unique crosstalk between caspase-11 and its negative regulator, whereby although MT3 keeps caspase-11 activation under control, the two synergistically compromise host immunological fitness to gram-negative bacterial infection.
Caspase-11 activation can have opposing effects on clearance of different bacteria. It improves resistance to Burkholderia thailandensis, B. pseudomallei, Brucella abortus, and Legionella pneumophila but may compromise immunity to B. cenocepacia, Salmonella typhimurium, E. coli, Shigella flexneri, K. pneumoniae and gram-positive infections including Streptococcus pyogenes, Staphylococcus aureus and Listeria monocytogenes (28, 31-33, 35, 63-69). Lipoteichoic acid from gram-positive bacteria engages the caspase-11 inflammasome via NLRP6 (28). Likewise, GAS infection led to caspase-11 activation in vivo. Although MT3 exerted disparate effects on caspase-11 activation in gram-positive and gram-negative infections, caspase-1 and IL-1b activation was suppressed by MT3 in both infection settings. In the context of GAS infection, both the host and the pathogen contribute to canonical inflammasome activation. Surface and secreted GAS virulence factor emm, and the streptococcal pyrogenic exotoxin B (SpeB) proteins act as second signals to activate caspase-1 signaling (70)(71)(72). Our data do not exclude the role of pathogen-derived factors in contributing to the increased canonical inflammasome activation observed in Mt3 -/mice infected with GAS. Nonetheless, our findings indicate that MT3 exerted a suppressive effect on the canonical caspase-1 pathway activated by gram-positive bacteria, while sparing negative regulation of the upstream non-canonical inflammasome activation in vivo.
The TRIF pathway is a central node in activation of the caspase-11 inflammasome in response to gram-negative infection (20). Targeting TRIF, but not IFNAR1, completely reversed the inflammatory cascade, including IL-1b activation. Blockade of IFNAR1 attenuated downstream activation of STAT1, caspase-11 and caspase-1, but both pro-IL-1b and active IL-1b levels were dramatically enhanced. This finding contrasts with the previously reported requirement of both TRIF and IFNAR1 in this pathway (20). Although that study utilized BMDMf from IFNAR1 -/mice and we used an anti-IFNAR1 monoclonal antibody, both approaches resulted in attenuation of targets downstream of IFNAR1. Emerging evidence points to an indirect inhibitory effect of type-I IFNs on inflammasome activation by decreasing pro-IL-1b transcription via IL-10 or 25-hydroxycholesterol (73,74). The subdued activation of caspase-1 and heightened active IL-1b levels suggests that IL-1b activation occurs via a caspase-1 independent pathway when IFNAR is blocked. Interfering with IFNAR1 signaling can therefore subdue activation of critical inflammasome components including caspase-11 and caspase-1 but sustain IL-1b production and activation.
The crucial function of Zn 2+ as a signaling molecule is well documented (13,(75)(76)(77). Changes in plasma and cellular Zn 2+ levels regulate the production of various cytokines via NF-kB signaling (78,79). Specifically, Zn 2+ deficiency in humans increases the production of tumor necrosis factor (TNF)a and IL-1b by LPStreated PBMCs ex vivo, whereas Zn 2+ supplementation reduces it (80,81). Therefore, a shift from physiological Zn 2+ concentrations at the systemic or cellular level can impact proinflammatory cytokine responses and inflammation. To our knowledge, modulation of Mf Zn 2+ homeostasis during non-canonical inflammasome activation has not been previously demonstrated. Our data show that caspase-11 activation is accompanied by a gradual expansion of the intracellular Zn 2+ pool driven by MT3. Zn 2+ deficiency did not augment pro-forms of caspase-11 (p43 and p38), but specifically increased their activation. Zn 2+ diminishes signaling via IRF3 by limiting its nuclear localization (48). Accordingly, MT3 interfered with signaling via the TRIF-IRF3-STAT1 axis by shaping the Mf Zn 2+ pool. The MT3-Zn 2+ axis dampened IRF3 phosphorylation and downstream mediators without impacting TRIF levels.
Although we cannot rule out the direct effect of Zn 2+ on inflammasome components downstream of IRF3, the suppressive action of MT3 on the non-canonical inflammasome was Zn 2+ dependent. Our data demonstrate that by manipulating the Mf Zn 2+ milieu, caspase-11 activation can either be triggered or averted. This finding has important implications in defining a role for Zn 2+ in subverting caspase-11 driven hyperinflammation. Developing therapeutic strategies that temper activation of the caspase-11/4 inflammasome have garnered tremendous interest to alleviate endotoxemia. In light of this, the MT3-Zn axis emerges as a fresh and vital candidate that guards the vigor of a caspase-11 fueled inflammatory response. In the context of gram-negative bacterial pathogenesis, our data indicate that strategies aimed at combined targeting of MT3 and the caspase-11 inflammasome may be more beneficial in infection control than targeting caspase-11 alone.
Taken together, our studies illuminate a double-edged phenomenon in inflammasome regulation whereby the MT3-Zn 2+ axis is a sentinel of the caspase-11 inflammasome but MT3 and the non-canonical inflammasome function in concert to compromise host antibacterial resistance.

Reagents and Resources
Reagents and resources can be found in Table 2.

Mice
All mice used in this study were on the C57BL/6 background. WT and Casp4 tm1Yuan /J (Casp-11 -/-) mice were acquired from the Jackson Laboratory. Lys2Cre mice were kindly provided by Dr. George S. Deepe. Jr. (University of Cincinnati). Mt3 -/-(exon 3 deleted) and Mt3 fl/fl mice were generated using clustered regularly interspaced short palindromic repeats (CRISPR) by the Transgenic Animal and Genome Editing Core facility at the Cincinnati Children's Hospital Medical Center (CCHMC). In mice, the Mt gene cluster is located on chromosome 8. The Mt3   The methods for the design of sgRNAs, donor oligos and the production of Mt3 fl/fl (loxp sites surrounding exon 3 of the murine Mt3 gene) animals were as described previously (82). The sgRNAs were selected according to the on-and off-target scores from the CRISPR design web tool (http://genome-engineering.org) as well as CRISPOR (http://crispor.tefor.net) (83). The selected sgRNA target sequences were cloned, according to the published method (84), into the pX458 vector (addgene #48138) that was modified by us to contain an optimized sgRNA scaffold (85) and a Cas9-2A-GFP. Their editing activity were validated by the T7E1 assay in mouse mK4 cells (86), compared side-by-side with Tet2 sgRNA that was known to work in mouse embryos efficiently (87). Validated sgRNA was transcribed in vitro using the MEGAshorscript T7 kit (ThermoFisher) and purified by the MEGAclear Kit (ThermoFisher), and stored at -80˚C. To prepare the injection mix, we incubated sgRNA and Cas9 protein (ThermoFisher) at 37˚C for 5 mins. to form the ribonucleoprotein complex and then added the donor oligos to it. The initial attempt was to insert both loxP sites simultaneously via piezo-driven cytoplasmic injection (88) of 100 ng/ul Cas9 protein, 50 ng/ul 5' sgRNA, 50 ng/ul 3' sgRNA, 50 ng/ul 5' donor, and 50 ng/ul 3' donor into fertilized eggs. Injected eggs were transferred into the oviductal ampulla of pseudo-pregnant CD-1 females on the same day. Pups were born and genotyped by PCR and Sanger sequencing. However, only 3' loxP-containing mice were obtained from this attempt. After breeding them to homozygosity for the 3' loxP, a new set of 5' sgRNA and the donor oligo was designed and injected into the zygotes with the mix containing 150 ng/ul Cas9 protein, 75 ng/ul 5' sgRNA, and 100 ng/ul ssDNA donor oligo. Injected eggs were transferred into the oviductal ampulla of pseudopregnant CD-1 females on the same day. Pups were born and genotyped by PCR and Sanger sequencing. Founder mice carrying both 5' and 3' loxP sites in cis were finally obtained. Animals were housed in a controlled environment with a 12-h light/12-h dark cycle, with free access to water and a standard chow diet. All animal procedures were carried out in accordance with the Institutional Animal Care and Use Committee-approved protocol of Cincinnati Children's Hospital and Medical Center.
Mf Culture hMf were prepared from peripheral blood mononuclear cells (PBMCs). Briefly, human blood obtained from the Hoxworth Blood Center, University of Cincinnati was diluted (1:2) with calcium-and magnesium-free 1X Dulbecco's phosphatebuffered saline (DPBS) and inverted gently to mix. Ficoll-Paque (10 ml for the total volume of 40 ml diluted blood) was layered at the bottom of the tube. The tubes were centrifuged at 400 X g for 30 mins. without break at 20˚C. The PBMC interface was transferred to sterile tubes and washed three times using 40 ml of DPBS containing 2mM EDTA and centrifuged at 120 X g for 10 mins. without break at 4˚C. A final wash was performed with DPBS (without EDTA). Isolated PBMCs were resuspended in complete RPMI 1640 medium. PBMCs (5 X 10 6 ) were plated in 24 well plates containing complete RPMI medium. After 24h, adherent monocytes were washed three times with DPBS and plated in complete RPMI 1640 medium (Corning ® ) containing 10 ng/ml macrophagecolony stimulating factor (M-CSF), 10%FBS, 10 mg/ml gentamycin sulfate (Alkali Scientifi c Inc.) and 2mercaptoethanol. Cells were differentiated by exposure to human recombinant M-CSF on days 0, 2 and 4. After 6 days, hMf were washed with DPBS prior to use for the experiment. Mouse BMDMf were prepared by differentiating bone marrow cells in complete RPMI 1640 medium containing 10 ng/ml mouse M-CSF, 10% fetal bovine serum (FBS) (HyClone Laboratories, Utah), gentamycin sulfate (10 mg/ml) and 2mercaptoethanol. BMDMf were fed on days 0, 2 with complete RPMI 1640 medium containing 10 ng/ml M-CSF and were supplemented on day 4 with M-CSF. After 6 days, adherent Mf were harvested by washing with DPBS followed by trypsinization and centrifuged at 1600 rpm for 5 mins. at 4˚C. Mf were washed again with DPBS at 1600 rpm for 5 mins. at 4˚C and counted under the microscope. Dead cells were excluded from enumeration using Trypan Blue stain. BMDMf (0.5 X 10 6 in 24 well or 1 X 10 6 in 12 well plate) were seeded.

Bioinformatics Analysis
The STRING database was used to review the protein-protein interaction networks of MT3 in Mus musculus and Homo sapiens (22). A full network analysis was conducted based on text mining, experiments, databases, co-expression, neighborhood, gene fusion and co-occurrence data with a minimum required interaction score of 0.4 and a maximum of 50 interactors in the first shell and 50 interactors in the second shell. Statistical significance of the enriched biological processes (GO BP categories) in the MT3 network was set with a false discovery rate (FDR) <0.05.
Identification of differentially expressed genes in resting WT compared to resting Mt3 -/-Mf was based on our previously published RNA-seq data (NCBI SRA: PRJNA533616) (19). The number of biological replicates used in the analysis was 3 per group. Genes differentially expressed with a fold change FC>2 and adjusted p value q<0.05 were considered significant. The Benjamini-Hochberg correction was used to adjust p values for multiple hypothesis testing. Differentially expressed genes in Mt3 -/-BMDMf compared to WT BMDMf with q<0.05 were queried using the functional annotation clustering tool DAVID to identify statistically enriched GO categories (using the GO terms, BP direct, CC direct and MF direct) in Mt3 -/-BMDMf compared to WT BMDMf (41). Significance of enrichment was set to FDR <0.05.

Gene Silencing
For gene silencing, Mf were transfected with the transfection complex (50 ml) of siRNA and TransIT-TKO ® (0.5%) transfection reagent (Mirus Bio ™ ) in 500 ml of complete RPMI 1640 medium without antibiotics as per the manufacturer's instructions. Concentration of siRNAs used for gene silencing were 100 nM each of the non-targeting pool (ON-TARGETplus ™ scramble siRNA), human MT3 (MT3 Silencer ® Pre-designed siRNA) and mouse Ticam1 (ON-TARGETplus SMARTpool). All siRNAs were purchased from Dharmacon (GE Healthcare). Both BMDMf and hMf were incubated with the siRNA containing transfection complexes for 24h in RPMI medium and washed prior to transfection with LPS in Opti-MEM medium. TRIF silencing was assessed by protein expression using Western blots. Human MT3 silencing was assessed by gene expression using qRT-PCR.

Preparation of Zn 2+ -Sufficient and Zn 2+ -Deficient Opti-MEM Media
Molecular biology grade chelex-100 resin (BioRad) was washed three times with metal free ddiH 2 O prior to use. To prepare Zn 2+ -deficient Opti-MEM medium, washed Chelex-100 resin (3 g per 100 ml) was added to Opti-MEM media and vigorously shaken for 1h on an orbital shaker at room temperature. After this time, media was filtered using a 0.22 mm filter and mixed with fresh washed chelex-100 resin and the same procedure was repeated for a total of 3 times to eliminate metals from Opti-MEM media. Chelex was removed from the media by a final filtration step. During each of these stages, an aliquot of the media was saved to monitor the efficiency of Ca 2+ , Mg 2+ , Mn 2+ , Co 2+ , Zn 2+ , Cu 2+ , Ni 2+ and Fe 2+ elimination by ICP-MS. The amount of Zn 2+ in Opti-MEM media was decreased by 95% by the above chelation method. To prepare Zn 2+ -sufficient media, chelexed Opti-MEM was reconstituted with Ca 2+ , Mg 2+ , Mn 2+ , Co 2+ , Cu 2+ and Zn 2+ at the original concentrations as measured by ICP-MS. To prepare Zn 2+ -deficient media, all measured elements except Zn 2+ were added to the chelexed Opti-MEM media at the original measured concentrations. Finally, the pH of Zn 2+ -sufficient and Zn 2+ -deficient Opti-MEM media was adjusted to 7.4 and filtered prior to use.

MT3 Overexpression and Purification
The pCMV6-Ac-GFP vector containing the mouse Mt3 gene (pCMV6-Ac-MT3-GFP) and empty pCMV6-Ac-GFP vectors were acquired from Origene and dissolved in nuclease-free sterile H2O. Plasmid DNA (5 ng) was added to 50 ml of thawed Novablue competent E. coli cells (EMD Millipore) and transformation was performed as per manufacturer's instructions. E. coli cells were serially diluted in S. O. C media (ThermoFisher Scientific) and plated onto Luria-Bertani (LB) plates with 50 mg/ml carbenicillin and grown for 24h at 37˚C. A single colony was isolated and inoculated in LB media containing carbenicillin and grown at 37˚C for 5h in a shaker. The culture was further amplified by passaging for another 24h. E. coli cells were then harvested by centrifugation at 2000 rpm for 10 mins. and plasmid was extracted using the EndoFree plasmid MAXI kit (Qiagen) as per the manufacturer's instructions. The plasmid was reconstituted in endotoxin-free TE buffer and OD readings obtained were in the range of 1.8-1.9. The resulting endotoxinfree plasmid DNA was set to a concentration of 1 mg/ml in filtersterilized EndoFree TE buffer and frozen into aliquots until further use.
Mt1 -/-Mt2 -/-BMDMf were transfected with pCMV6-Ac-GFP control vector or pCMV6-Ac-MT3-GFP vector using the LT1 transfection reagent (Mirus Bio) in RPMI media containing 10% serum without antibiotics as per the manufacturer's instructions. After 48h, BMDMf cultures were lysed with 250 ml of 0.1% SDS prepared in double-deionized (ddi) H 2 O for 20 min. on ice with intermittent mixing. Cell lysates were transferred to 0.22 mm filter tubes and centrifuged at 13000 rpm for 5 mins. Filtered cell lysates were subjected to SEC-ICP-MS to isolate the MT3 protein as described below.
Preparation of apo-MT3, 4Zn 2+ -MT3 or 6Zn 2+ -MT3 Cell lysates from above were analyzed by SEC-ICP-MS to detect the MT3-associated peak, followed by collection of the fraction of interest (18-21 mins.). The collected fraction was concentrated by freeze drying in Millrock lyophilizer (Millrock, NY). The concentrated fraction was treated with 1 g of Chelex X-100 resin to remove the divalent metals associated to the protein. The total MT concentration was calculated by the total sulfur concentration in the sample with a 1:20 stoichiometry. The sample was divided into 3 fractions, and each fraction was incubated for 2 h in 50 mM Tris-HCl with the appropriate concentration of 66 Zn 2+ nitrate to obtain an MT3-Zn 2+ saturation of 0, 4 or 6 Zn 2+ ions per MT3 molecule. After incubation, samples were filtered using a 3 kDa MWCO filter to remove the unbound Zn 2+ and reconstituted in 1X PBS.

Transfection of Apo, Zn 2+ -MT3 Complexes Into BMDMf
In a 24 well plate, 5 X 10 5 Mt3 -/-BMDMf were transfected with the transfection complex (10 ml) containing 500 ng of apo-MT3, 4Zn 2+ -MT3 or 6Zn 2+ -MT3 and Pro-Ject ™ (1.75 ml) protein transfection reagent (ThermoScientific-US) in 250 ml of 2% FBS containing antibiotic free RPMI 1640 medium. Control Mt3 -/-BMDMf were treated with Pro-Ject ™ alone. Cells were incubated with the transfection complexes for 3.5h and washed two times using HBSS prior to challenge with iLPS in Zn 2+ free Opti-MEM medium. At the experiment end point, cell lysates were either prepared for SEC-ICP-MS analysis or both cell lysates and supernatants were collected for the analysis of noncanonical inflammasome activation.

SEC-ICP-MS-MS Analysis and Normalization of Data
SEC-ICP-MS-MS analysis of WT and Mt3 −/− BMDMf was performed as described previously (11). Mf were plated in Opti-MEM media, and either left untreated or transfected with 10 mg/ml LPS for 1, 24 and 48h. After this time, Mf were washed twice in HBSS and cells were lysed with 0.1% SDS on ice for 20 mins. Cell lysates were then centrifuged in 0.22 mm filter tubes at 13000 rpm for 10 mins. Filtered cell lysates were frozen at -80°C until further analysis by SEC-ICP-MS. 50-80 µl of cell lysates were injected to the HPLC-ICP-MS system according to protein concentration. To normalize the response of ICP-MS-MS signal from SEC separations on different days, 50 ml of 0.5 mg/ml carbonic anhydrase was injected into the liquid chromatography system, and area of Zn 2+ signals from samples was normalized to area of the carbonic anhydrase peak from each day. The absorbance of carbonic anhydrase at 280 nm was followed to ensure protein integrity.
The instrumentation consisted of an Agilent 1100 HPLC equipped with a degasser, a binary pump, a thermostated auto sampler, a column oven compartment and a diode array detector. For the Mf lysates, a TSK gel 3000SW gel filtration column (TSK Tokyo Japan) 7.8 × 300 mm, 10 mm particle size was used. The mobile phase was ammonium acetate pH 7.4, 0.05% MeOH at 0.5 ml/minute. The HPLC system was coupled to the ICP-MS-MS nebulizer via a short polyether ether ketone capillary of 0.17 mm internal diameter. An Agilent 7500ce ICP-MS system equipped with a micromist quartz nebulizer, a chilled double pass Scott spray chamber and a standard 2 mm insert quartz torch with shield torch was used for all experiments. The ICP-MS was operated by the Agilent Mass Hunter integrated chromatographic software in the helium collision mode as reported previously (11). The isotope dilution experiments were processed by exporting the chromatographic data to Origin (Origin labs, CA) and the signal, in the form of counts per second, was used to calculate the ratio of 66

ICP-MS-MS and SEC-ICP-MS-MS Quality Control to Avoid External Zn 2+ Contamination
All metal analysis experiments were performed using trace metal grade reagents with acid washed plastic vials. Reagent blanks were used to correct the background signal. The analysis was performed through a metal free encased auto sampler. The concentration of Zn 2+ in the blanks was always below 100 parts per trillion (ppt), the blank estimate concentration on the calibration curves was always below 50 ppt, while the detection limits were below 30 ppt.
For chromatographic analysis, the mobile phase was cleaned using a Chelex 100 resin, using the batch method. In brief, 3 g of Chelex-100 was added to a liter of mobile phase, stirred for 30 mins. and passed through a 0.45 mm membrane. This decreased the Zn 2+ concentration below 200 ppt (measured as total). By this method, the base line ICP-MS-MS 66 Zn 2+ signal was below 1000 counts per second, which represents sub-ppb levels. The SEC column was cleaned using 10 volumes of 0.2 M NaCl and equilibrated with the mobile phase, followed by injection of 50 ml of 2% HNO 3 three times to remove any accumulated Zn 2+ in the column. With this procedure, Zn 2+ distribution in the samples never deviated more than 10% compared to the theoretical natural Zn 2+ isotope distribution in the control Mf samples. Four blanks and four carbonic anhydrase standards were injected after the cleaning procedure for monitoring Zn 2+ signal by ICP-MS-MS to ensure optimal column performance. Typically, the column was cleaned every 30-40 samples.

LPS Treatment In Vivo
Thirteen-week-old WT and Mt3 -/mice were primed with i.p. injection of 10 mg/kg poly(I:C) for 6h followed by 2 mg/kg ultrapure LPS-B5 (In vivoGen) prepared from E. coli 055:K59 (B5) (i.p. injection) for 18h. At the experiment end point, blood was collected by cardiac puncture, allowed to clot, and centrifuged at 2000 rpm for 30 mins. at 4˚C to isolate serum. Serum was used to measure cytokines by enzyme-linked immunosorbent assay (ELISA).

In Vitro and In Vivo Infection With Gram-Negative Bacteria
For in vitro infection, E. coli (K12) was grown in LB broth at 37˚C overnight in an orbital shaker. The culture was pelleted, washed and resuspended with ice-cold DPBS. Optical Density (OD) of the culture was measured at 600nm using a spectrophotometer. To analyze E. coli burden in in vitro, hMf were transfected with scramble siRNA or MT3 siRNA as described above. hMf, WT and Mt3 -/-BMDMf were infected with a multiplicity of infection (MOI) of 25 E. coli: 1 hMf for 3.5h in Opti-MEM media. Mf were washed 3 times with 10 mg/ ml gentamycin sulfate containing DPBS to kill extracellular bacteria and incubated in Opti-MEM media with antibiotic for 24h. Mf were again washed 3 times with antibiotic-free DPBS, diH 2 O was added and cells were incubated for 30 min. to induce osmotic lysis. Cells were scraped and lysates were diluted in DPBS followed by plating on LB agar plates and incubated at 37˚C for 24h. Colonies were enumerated as above. Intracellular bacterial burden was represented as percent inhibition of bacterial growth in MT3-silenced hMf compared to scramble siRNA treated hMf and in Mt3 -/-BMDMf compared to WT BMDMf.
To analyze antibacterial immunity in vivo, 10 to12 week-old mice were used. Mice were infected with E. coli 1 X 10 9 CFUs via i.p. injection (300 µl/mouse) for 1h or 6h. To investigate the role of NLRP3 inflammasome in antibacterial immunity, mice were treated with 1 mg/mouse MCC950 (an inhibitor of NLRP3 inflammasome) via i.p. injection (100 µl/mouse) for 1h followed by E. coli 1 X 10 9 CFUs via i.p. injection (300 µl/mouse) for 6h. K. pneumoniae KP2 2-70, a virulent, heavily encapsulated gramnegative bacterial strain (89,90), was grown overnight in brain heart infusion (BHI) broth. The following morning bacteria were washed with DPBS, and administered at 4 x 10 4 CFUs in 50 µl per mouse by the i.n. to isoflurane-anesthetized mice for 48h. At the infection end point, blood was collected by cardiac puncture. A portion of the blood sample was acquired in anticoagulant (3% Na-citrate or EDTA) containing tubes to determine bacterial CFUs in blood. The remaining blood was allowed to clot, and centrifuged at 2000 rpm for 30 mins. at 4˚C to isolate serum. Peritoneal lavage was collected using ice-cold 10 ml DPBS. Kidney, lung, and spleen were collected after perfusion with 3 ml of DPBS, indicated organs and skin was rinsed in DPBS and ground with 5 ml DPBS using a glass grinder. Bacterial growth was measured in blood, peritoneal lavage, kidney, lung, skin and spleen samples. Serum and peritoneal lavage were used to measure cytokines by enzyme-linked immunosorbent assay (ELISA).

LPS-Induced Septic Shock
Septic shock was induced in mice via i.p. injection with ultrapure LPS-B5 (20 mg/kg) (In vivoGen) prepared from E. coli 055:K59 (B5). Mice were weighed and sepsis scores were determined at various intervals. The MSS scoring method assesses sepsis severity with scores ranging from 1-4 based on 7 parameters (appearance, consciousness, activity, stimulus (sounds/touch), eyes aspect, respiration rate and respiration quality) (27). Survival analysis was conducted using the Kaplan-Meier analysis method and log-rank (Mantel-Cox test) was used to determine statistical differences in survival.

In Vivo Infection With Gram-Positive Bacteria
A representative M1T1 clonal Group-A-Streptococcus GAS5448 was used for subcutaneous infections (91). GAS was grown at 37°C under static conditions in Todd-Hewitt broth (BD, MD, USA) supplemented with 1.5% yeast extract (BD Biosciences, MD, USA) and in vivo infections were performed as described previously (29,92). WT and Mt3 -/mice (n=8/group) were used in this study as a model for subcutaneous GAS infections. One day prior to infection, the hair on the back of the mice was depilated (using Nair cream) and mice were infected subcutaneously with 0.1 ml of GAS suspension prepared in sterile DPBS (Ca 2+ /Mg 2+ free, low endotoxin, Mediatech, VA, USA, DPBS) (OD 600 adjusted to yield~1-5x10 (8) CFUs). Actual inoculum was determined by plating on trypticase soy agar containing 5% sheep blood (BD Biosciences, MD, USA). Mice were monitored twice daily for body weight, lesions, and mortality. To determine GAS dissemination and load, mice were humanely euthanized 72h post-infection. Blood was drawn through cardiac puncture; necrotic skin, kidney, and spleen were recovered aseptically and weighed. One ml of DPBS was added per 100 mg tissue and homogenized (Omni International, Marietta, GA) followed by plating of ten-fold dilutions on blood agar plates. GAS burden was calculated as colony-forming units (CFUs) per ml (blood) or per mg of tissue. The remaining homogenates were centrifuged for 15 mins. at 12,000 x g at 4°C, and supernatants were stored at -80°C for western blot analysis.

Gene Expression
RNA was isolated from Mf after elimination of genomic DNA using RNeasy Plus Mini kit (Qiagen) or QUICK-RNA ™ MINIPREP KIT (Thomas Scientific). cDNA was prepared using Reverse Transcription Systems Kit (Promega, WI) or rAmp First Strand cDNA Synthesis Flex Kit (Thomas Scientific). Taqman primer/probe sets (Applied Biosystems, CA) were used for real-time gene expression analysis using ABI Prism 7500. For time course analysis of expression of murine Mt genes, Mf were left unstimulated or stimulated with 2 mg/ml iLPS for 0h, 1h, 6h, 24h and 48h. Data are presented as fold change in gene expression normalized to unstimulated Mf at the 0h time point. To examine the effects of extracellular ultrapure LPS (exLPS) on murine Mt3 gene expression, BMDMf were left unstimulated or stimulated with 10 mg/ml exLPS. Data are presented as fold change in gene expression normalized to unstimulated Mf at the 48h time point. For MT2A gene expression analysis in MT3 silenced hMf, cells were treated with either MT3 siRNA or scramble siRNA as mentioned above. Data are presented as fold change in gene expression normalized to control siRNA treated hMf.
Hypoxanthine guanine phosphoribosyl transferase (Hprt) was used as an internal control to compare target gene expression.

Western Blotting
TRIF (Proteintech), pIRF3 (BIOSS), STAT1, pSTAT1 (Abcam), GBP2, GBP5 (Proteintech), caspase-11 (Abcam and eBioscience ™ ), CASPASE-4 (MBL), Gasdermin D (Proteintech and Cell Signaling Technologies), caspase-1 (AdipoGen Life Sciences), IL-1b (R&D Systems) and caspase-8 (Enzo Life Sciences) were assessed in kidney homogenates of E. coli infected mice. Cell lysates were prepared using Denaturing Cell Extraction Buffer (Invitrogen) containing protease & phosphatase inhibitor cocktail (ThermoScientific). Culture supernatants were frozen at -80˚C until use and processed using methanol-chloroform protein extraction method. Briefly, supernatants were mixed with equal volume of 100% ice-cold methanol and 0.25 times of the total volume of chloroform followed by gentle vortexing and centrifugation at 20,000 X g at 4˚C for 10 min. Upper-phase was discarded without disturbing inter-phase proteins. Ice-cold methanol (500 ml) was added to the tube, gently vortexed and centrifuged at 20,000 X g at 4˚C for 10 mins. Supernatants were discarded and pellet was dried at 37˚C for 3-5 mins. Urea (8 M, pH-8.0) was used to dissolve the pellet and extracted proteins were stored at -80˚C. Total cell lysates, supernatants proteins, cell lysates + supernatants and kidney homogenates were boiled in SDS-PAGE 1X sample buffer at 95˚C for 5 mins. Kidney homogenates were centrifuged at 20,000 X g at 4˚C for 15 mins. Supernatants were collected for protein analysis. Reduced proteins were run on 8%,10% or 12% SDS-PAGE gels and transferred on to 0.22 mm nitrocellulose membranes (GE Healthcare Life Sciences). Membranes were blocked using 5% skim milk in 1X Tris-buffered saline and 0.1% Tween 20 (1XTBST) and probed overnight with primary antibodies at 4˚C. Membranes were washed 3 times for 10 mins. each with 1XTBST and probed with corresponding HRP conjugated or IRDyes (LI-COR) secondary antibodies, washed and developed using BrightStar ™ Femto HRP Chemiluminescent 2-Component Substrate Kit (Alkali Scientific Inc. b-actin was used as an internal loading control. Western blots were analyzed using ImageJ software and densitometry data were normalized to b-actin.

ELISA
Human and mouse IL-1b (BioLegend) concentration in media supernatants, serum and in peritoneal lavage and mouse IL-1a (BioLegend) in media supernatants were determined using commercial ELISA kits according to the manufacturer's instructions.

Quantification and Statistical Analysis
Data were analyzed using Sigma plot or GraphPad Prism by oneway ANOVA for multiple comparisons using the indicated adhoc methods with at least 3 or more independent biological replicates. Where two groups were compared, two-tailed t-test was used. For in vivo infection, bacterial CFUs were logtransformed for statistical analysis. p values were calculated, *p < 0.05, **p < 0.01, ***p< 0.001; NS, not significant, ND, not detected.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Cincinnati and were conducted within the Department of Laboratory Animal Medicine accredited by American Association for Accreditation of Laboratory Animal Care (Frederick, MD). All animal experiments were conducted in accordance with Animal Welfare Act guidelines of the National Institutes of Health.

AUTHOR CONTRIBUTIONS
DC and KSV planned and conducted molecular and biochemical in vitro and in vivo experiments, generated Casp-11 -/-Mt3 -/and Lys2Cre Mt3 fl/fl mice, analyzed data, and wrote the manuscript. JG conducted in vivo infections with K. pneumoniae. AS, SN, and SM conducted in vivo experiments with GAS infection and analyzed data. AP assisted with bioinformatics analysis. JL conducted chromatographic and mass spectrometric analysis using ICP-MS and SEC-ICP-MS and MT3-Zn 2+ complex preparations, and analyzed data. KSV designed and supervised the project. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by a Junior Faculty Pilot Project Award, American Heart Association 19CDA34770022 Award, American Association of Immunology Careers in Immunology Fellowship Awards to KSV and NIAIDR01 AI106269-06 awarded in part to KSV. GAS studies were supported by the UND CoBRE Host-Pathogen Interactions Pilot Award -NIH/ NIGMS award P20GM113123, UND SMHS funds (SN), and UND VPRED Postdoctoral funding support (AS).

ACKNOWLEDGMENTS
We thank Transgenic Animal and Genome Editing Core at Cincinnati Children's Hospital Medical (CCHMC) Center for production of Mt3 fl/fl mice.