When innate immune cells such as macrophages are challenged with environmental stresses or infection by pathogens, they trigger the rapid assembly of multi-protein complexes called inflammasomes that are responsible for initiating pro-inflammatory responses and a form of cell death termed pyroptosis. We describe here the identification of an intracellular trigger of NLRP3-mediated inflammatory signaling, IL-1β production and pyroptosis in primed murine bone marrow-derived macrophages that is mediated by the disruption of glycolytic flux. This signal results from a drop of NADH levels and induction of mitochondrial ROS production and can be rescued by addition of products that restore NADH production. This signal is also important for host-cell response to the intracellular pathogen Salmonella typhimurium, which can disrupt metabolism by uptake of host-cell glucose. These results reveal an important inflammatory signaling network used by immune cells to sense metabolic dysfunction or infection by intracellular pathogens.https://doi.org/10.7554/eLife.13663.001
Cells of the innate immune system, such as macrophages, are the body’s first line of defense against infection. Macrophages can sense a wide variety of danger signals associated with the presence of infectious microbes, and some of these signals cause macrophages to form protein complexes called inflammasomes inside the cell. Inflammasomes produce molecules that stimulate inflammation and trigger the death of the macrophage. This attracts other immune cells to the infection site to help combat the source of danger.
Inflammasome complexes form around an activated receptor molecule called NLRP3. NLRP3 is activated by a range of danger signals, including those produced by Salmonella bacteria. However, the sequence of events that leads to NLRP3 activation is still not well understood.
Sanman et al. have now identified a small molecule that unexpectedly causes the formation of inflammasomes via NLRP3 and so triggers the death of macrophages. Further investigation revealed that this molecule disrupts glycolysis, a process macrophages use to produce energy. The energy imbalance caused by disrupting glycolysis triggers a stress response in macrophages, which ultimately activates the NLRP3 receptor and hence the inflammasome.
Sanman et al. then found that Salmonella bacteria also activate the inflammasome by disrupting glycolysis when they invade macrophages. This occurs because the bacteria use up the macrophage’s supply of glycolysis precursor molecules. Replenishing the macrophage with products of glycolysis restored partial energy production and prevented the inflammasome from being activated.
Overall, Sanman et al. have identified a previously unknown trigger of inflammation and cell death in macrophages whereby cells can respond to infectious bacteria by sensing a change in energy levels. A next step will be to define the signaling molecules that activate NLRP3 to trigger the construction of the inflammasome. Sanman et al. also hope to uncover other infections and diseases where changes in energy balance might trigger inflammation and cell death.https://doi.org/10.7554/eLife.13663.002
Inflammation is an immunological process required for an organized response to infection, injury, and stress. Because excessive inflammation can be damaging, its initiation is highly regulated. Innate immune cells such as macrophages have evolved sensors of pathogens and homeostatic perturbations which, when activated, induce an immune response (Medzhitov, 2008). Amongst these sensors are Nod-like receptors (NLRs), which are activated in response to a diverse set of pathogen-associated molecular patterns (PAMPs) and danger-associated molecular patterns (DAMPs). Activated NLR proteins recruit and facilitate activation of the protease caspase-1 either directly, through caspase activation and recruitment domain (CARD) interactions, or indirectly, through the adaptor apoptosis-associated speck-like protein containing a CARD (ASC; also known as Pycard). The resulting macromolecular complex is referred to as the inflammasome (Lamkanfi and Dixit, 2014). The inactive precursor of the cytokine interleukin-1β (pro-IL-1β) is also recruited to the inflammasome complex, where proteolysis by caspase-1 induces activation and secretion of the bioactive cytokine, further promoting inflammation. In addition to cytokine maturation, inflammasome formation and caspase activation are associated with a pro-inflammatory form of cell death termed pyroptosis (Fink and Cookson, 2006). This form of cell death results in lytic release of cytosolic contents and other pro-inflammatory factors such as interleukin-1α and high-mobility group protein B1 (HMGB1), which are potent inducers of inflammation (Medzhitov, 2008; Croker et al., 2014).
Diverse activation signals have been reported as triggers of NLR signaling. For example, the NLR AIM2 is activated by cytosolic double-stranded DNA (Lamkanfi and Dixit, 2014; Fernandes-Alnemri et al., 2009; Hornung et al., 2009; Bürckstümmer et al., 2009), a structural feature associated with infections with pathogens and not found in healthy host cells (Fink and Cookson, 2006; Hornung et al., 2009; Jones et al., 2010). The NLR NLRP3 is a sensor of a wide variety of PAMPs and DAMPs but the unifying mechanism of its disparate activators is not understood (Sutterwala et al., 2014). Furthermore, while the NLRP3 signaling pathway can be activated by a variety of both gram-positive and gram-negative bacteria, the mechanism by which these pathogens induce inflammasome signaling through this receptor is often unclear (Storek and Monack, 2015). Specifically, effective defense against Salmonella typhimurium (S. typhimurium) requires NLRP3 (Broz et al., 2010), yet the mechanism by which the pathogen activates this pathway remains unknown.
Here, we report a small molecule, GB111-NH2, that induces NLRP3 inflammasome formation, caspase-1 activation, IL-1β secretion, and pyroptotic cell death in bone marrow-derived macrophages (BMDM). Using chemical proteomics, we identify the glycolytic enzymes GAPDH and α-enolase as the phenotypically relevant targets of this molecule. Facilitating TCA metabolism downstream of glycolysis by addition of pyruvate or succinate blocked the effects of the compound. We find that S. typhimurium infection, like direct chemical inhibition of the glycolytic enzymes, reduced glycolytic flux and that restoring metabolism downstream of glycolysis also prevented S. typhimurium-induced inflammasome formation, IL-1β secretion, and pyroptosis. We find that glycolytic disruption induced by either the small molecules or S. typhimurium infection impaired NADH production, resulting in the formation of mitochondrial ROS that were essential for NLRP3 inflammasome activation. Therefore, disruption of glycolytic flux is a biologically relevant trigger of NLRP3 inflammasome activation that is mediated by mitochondrial redox changes, revealing a mechanistic link between cellular metabolism and initiation of inflammation.
While screening peptide-based compounds for their effects on inflammasome signaling, we identified one compound, GB111-NH2 (Blum et al., 2005; Verdoes et al., 2012) (Figure 1A), that was sufficient to induce caspase-1 activation in LPS-primed bone marrow-derived macrophages. We measured caspase-1 activation by monitoring conversion of procaspase-1 to the mature p10 form by Western blot and, in parallel, by labeling BMDM with the caspase-1-selective activity-based probe (ABP), AWP28 (Puri et al., 2012) (Figure 1B). In addition to producing active caspase-1, we found that GB111-NH2-treated BMDMs secreted the cytokine IL-1β in a dose-dependent manner (Figure 1C). Western blot analysis confirmed that secreted IL-1β was primarily the bioactive p17 form (Figure 1—figure supplement 1) that is generated by active caspase-1.
By fluorescence microscopy, we observed formation of foci containing the inflammasome adaptor ASC and active caspase-1 in GB111-NH2-treated BMDMs (Figure 1D). Formation of these foci was dependent on NLRP3 and ASC but not caspase-1, caspase-11, NLRC4, or AIM2 (Figure 1E–F). We observed that GB111-NH2 induced a similar level of IL-1β secretion as the NLRP3 stimuli ATP and nigericin (Figure 1G) and that the absence of NLRP3 completely abrogated IL-1β secretion induced by GB111-NH2 treatment. The absence of other NLRs, specifically NLRC4 and AIM2, had no effect on IL-1β secretion (Figure 1H). Taken together, these data indicate that GB111-NH2 induces caspase-1 activation and IL-1β secretion solely through the NLRP3 inflammasome, acting as an activating ‘Signal II’ for the canonical NLRP3 pathway (Lamkanfi and Dixit, 2014).
In order for ‘Signal II’ to activate the NLRP3 inflammasome, BMDM must first be primed by a ‘Signal I’ such as LPS. LPS priming induces NF-κB-dependent transcription of pro-inflammatory genes such as IL-1β and inflammasome-independent secretion of pro-inflammatory cytokines such as IL-6 and TNF-α (Lamkanfi and Dixit, 2014). We measured lysate protein levels by Western blotting and supernatant cytokine levels by ELISA in BMDM treated as in previously described experiments; first primed for 3 hr with LPS and then treated for 2 hr with GB111-NH2. We observed the appearance of pro-IL-1β upon LPS priming (Figure 1I) but there was no effect of GB111-NH2 on either IL-1β protein levels in BMDM that had received LPS priming (Figure 1C, Figure 1G). In addition, IL-6 secretion decreased with increasing dose of GB111-NH2 and TNF-α secretion was unaffected by GB111-NH2 (Figure 1—figure supplement 2). Therefore, GB111-NH2 does not have a direct effect on Signal I, but functions predominantly as a Signal II for the NLRP3 inflammasome.
Macrophages containing active inflammasome complexes often rapidly die by a pro-inflammatory process called pyroptosis (Fink and Cookson, 2006). We observed features of this form of cell death in GB111-NH2-treated BMDM, including release of the intracellular enzyme lactate dehydrogenase (LDH) (Figure 1J), and foci of caspase-1 activity in propidium iodide (PI) and Annexin V (AnnV) positive cells (Figure 1K). These data confirm that GB111-NH2 is a small molecule activator of the NLRP3 inflammasome that also triggers pyroptotic cell death.
Given that GB111-NH2 is chemically distinct from other known activators of NLRP3 and easily modifiable, we wanted to use it as a tool to identify protein targets that are involved in triggering this pro-inflammatory response. To accomplish this, we first conducted a small structure-activity relationship (SAR) study in which we synthesized a series of analogs of GB111-NH2 to identify compounds that could be used for affinity isolation of labeled targets. We identified a number of modifications to the primary compound scaffold that resulted in loss of activity (Figure 2A), suggesting that the effects of the parent compound are likely dictated by affinity to specific protein targets. Importantly, our SAR efforts identified both an inactive analog (GB-IA) as well as an azide-containing analog (az-GB) that retained the inflammasome-activating ability of GB111-NH2 (Figure 2A, Figure 2—figure supplement 1). We used this azide analog as a probe to identify potential protein targets using Click chemistry to attach a fluorescent tag (Figure 2—figure supplement 1) or affinity tag (biotin) to labeled target proteins. The choice of this probe does limit identification to covalent binding partners. However, because removal of the acyloxymethylketone (AOMK) electrophile from GB111-NH2 (compound 2) resulted in loss of the ability of the compound to induce IL-1β secretion, we reasoned that the compound likely acted through covalent modification of its relevant targets.
We conducted a proteomic study in which we pre-treated BMDMs with either active or inactive analogs of GB111-NH2, labeled with the az-GB probe, reacted the resulting lysates with alkyne-biotin and identified affinity isolated targets using multidimensional protein identification technology (MudPIT) (Weerapana et al., 2007) (Figure 2B). By using active and inactive compounds in our pretreatment (GB111-NH2 and GB-IA, respectively), we could identify labeled proteins that were lost by pretreatment with the active compound but not the inactive control. Employing this strategy, we obtained a short list of potentially relevant binding partners (Supplementary files 1–3). Interestingly, this list included proteins critical to cellular metabolism and homeostatic maintenance.
To determine which of the potential protein targets of GB111-NH2 were responsible for its’ phenotypic effects, we first tested whether reported selective inhibitors of several of the identified targets mimicked the effects of GB111-NH2. Based on our target list (Supplementary file 3), we selected the compounds Atpenin A5 (Miyadera et al., 2003) (AA5; inhibitor of succinate dehydrogenase in the TCA cycle), 6-aminonicotinamide (Street et al., 1997) (6AN; inhibitor of 6-phosphogluconate dehydrogenase in the pentose phosphate pathway), koningic acid (Endo et al., 1985) (KA; inhibitor of GAPDH in glycolysis) and ENOblock (Jung et al., 2013) (EB; inhibitor of α-enolase in glycolysis). We tested these compounds for their effects on LPS-primed BMDM and found that only inhibitors of the glycolytic enzymes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and α-enolase induced IL-1β secretion, inflammasome formation, and caspase-1 processing (Figure 3A–C). We chose concentrations of each inhibitor based on literature descriptions of concentrations at which targets should be completely inhibited (Miyadera et al., 2003; Street et al., 1997; Endo et al., 1985; Jung et al., 2013). Mass spectrometry results demonstrated that GB111-NH2 also bound to lysosomal cysteine cathepsins (Supplementary file 2), though not in the expected pattern. To rule out a mode of action based on cathepsin inhibition, we tested the established cathepsin inhibitors leupeptin, E-64d, and Ca074Me and found that they did not induce caspase-1 activation in BMDM (Figure 3—figure supplement 1). These data indicate that the phenotypically relevant targets of GB111-NH2 are the glycolytic enzymes GAPDH and α-enolase.
The GAPDH and α-enolase inhibitors KA and EB failed to induce inflammasome formation in cells that lack Pycard or Nlrp3 (Figure 3D), induced IL-1β secretion in a dose-dependent manner that was also NLRP3-dependent (Figure 3E, Figure 3—figure supplement 2), and had no effect on pro-IL-1β or NLRP3 levels in LPS-primed BMDM (Figure 3A, Figure 3F). These data demonstrate that structurally dissimilar inhibitors of either GAPDH or α-enolase activate the canonical NLRP3 inflammasome pathway similarly to GB111-NH2.
To further confirm the targets of GB111-NH2, we measured the ability of the parent compound and its analogs to covalently bind and inhibit the activity of the identified glycolytic enzyme targets. To test if our compounds covalently bound to GAPDH and α-enolase, we incubated recombinant GAPDH and α-enolase with our az-GB probe, used Click chemistry to attach TAMRA-alkyne to the az-GB probe, and analyzed reaction mixtures by fluorescent scanning of SDS-PAGE gels. We observed probe labeling of both GAPDH and α-enolase, indicating that the az-GB probe covalently binds to both enzymes (Figure 4A–B, Figure 4—figure supplement 1). Pretreatment with GB111-NH2 blocked az-GB binding to both GAPDH and α-enolase in a dose-dependent manner, indicating that both compounds bind to similar sites on their enzyme targets. The GB111-NH2 analog containing a more reactive phenoxymethylketone electrophile (compound 6), which also induced IL-1β secretion more potently in cells than GB111-NH2, also blocked az-GB binding to both GAPDH and α-enolase at lower concentrations compared to GB111-NH2. Importantly, the analogs that did not induce IL-1β secretion in BMDM, compound 2 (which lacks the AOMK electrophile) and GB-IA (which lacks the carboxybenzyl cap of GB111-NH2), did not compete for az-GB binding to GAPDH and α-enolase. We did observe some labeling of GAPDH and α-enolase by TAMRA-alkyne independent of the az-GB probe, which is potentially due to the ability of alkynes to function as cysteine electrophiles (Ekkebus et al., 2013).
az-GB probe binding to both enzymes was blocked by the cysteine-alkylating compound N-ethylmaleimide (NEM), and by KA and EB (for GAPDH and α-enolase, respectively), suggesting that binding was dependent on enzyme activity and was mediated by reaction with key reactive cysteines (Figure 4—figure supplement 1). To further investigate the proposed covalent interaction of GB111-NH2 with reactive cysteine residues in GAPDH and α-enolase, we performed competition studies with the general cysteine reative probe iodoacetamide fluorescein (IAF). IAF labeled both GAPDH and α-enolase, consistent with previous reactive cysteine profiling data demonstrating that the catalytic Cys 152 of GAPDH is highly reactive and Cys 388, an active-site proximal cysteine of α-enolase, is also reactive (Weerapana et al., 2010). NEM potently blocked IAF labeling, confirming that IAF was reacting with cysteine residues in GAPDH and α-enolase. Importantly, GB111-NH2 also competed for IAF labeling, indicating that it covalently binds these same reactive cysteines (Figure 4C–D).
Modification of the active-site cysteine of GAPDH and the active site-proximal Cys 388 of α-enolase have both been shown to potently impair enzyme activity (Kato et al., 1992; Ishii and Uchida, 2004). To confirm that binding of our compounds to these enzymes also inhibits enzyme activity, we performed substrate assays with recombinant GAPDH and α-enolase and found that GB111-NH2 dose-dependently inhibited turnover of the respective substrates, glyceraldehyde-3-phosphate and 2-phosphoglycerate (Figure 4E). GB-IA did not significantly inhibit GAPDH but did exhibit modest inhibitory activity towards α-enolase. GB111-NH2 also showed time-dependent inhibition of GAPDH and α-enolase activity (Figure 4F), suggesting that it is acting as an irreversible inhibitor (Singh et al., 2011). Taken together, these data indicate that GB111-NH2 binds covalently to reactive cysteine residues in both GAPDH and α-enolase and that binding to these cysteine residues inhibits enzyme activity.
Because GAPDH activity was recently shown to determine flux through aerobic glycolysis (Pietzke et al., 2014), we hypothesized that our compounds were activating NLRP3 by disrupting glycolytic flux. Productive glycolysis results in conversion of NAD+ to NADH, secretion of lactate, and ATP production (Figure 5A). To first test the hypothesis that GB111-NH2, KA, and EB block glycolytic flux we measured the ratio of NAD+/NADH, lactate production, and intracellular ATP concentration in inhibitor-treated BMDM. LPS stimulation, which up-regulates glycolysis in macrophages (Krawczyk et al., 2010), resulted in an increase in NADH levels (demonstrated by a decrease in NAD+/NADH ratio) and an increase in lactate secretion (Figure 5B–D). GB111-NH2 treatment completely blocked the lactate and NADH production induced by LPS stimulation, indicating that it directly impaired LPS-induced glycolytic flux (Figure 5B–D). ATP production was also significantly impaired in LPS-primed BMDM upon GB111-NH2 treatment (Figure 5E). Notably, treatment with the NLRP3 activator nigericin did not reduce NADH levels (Figure 5B, Figure 5—figure supplement 1), indicating that the metabolic disruption that we observed with GB111-NH2 is not a general feature of inflammasome activation and cell death. The GAPDH and α-enolase inhibitors KA and EB also affected metrics of glycolytic flux (Figure 5B–D). Finally, GB111-NH2 suppressed the increase in extracellular acidification rate (ECAR) induced by glucose stimulation (Figure 5F). 2DG, a glycolytic inhibitor that targets hexokinase, did not dramatically impair glycolytic flux (as measured by NADH production and lactate secretion) and did not induce inflammasome formation (Figure 5—figure supplement 2). This indicates that severe limitation of glycolytic flux is required to activate the NLRP3 inflammasome. Furthermore, these results are in accord with recent studies showing that inhibiting GAPDH, and not enzymes in upper glycolysis, is flux-limiting in highly glycolytic cells (Shestov et al., 2014). Inhibitors of the TCA cycle and pentose-phosphate pathways (AA5 and 6-AN, respectively), carbohydrate metabolism pathways that are closely tied to glycolysis, also had no effect on lactate secretion or NADH production in LPS-primed BMDM and did not induce IL-1β secretion (Figure 3A, Figure 5C).
We hypothesized that, due to the dependence of macrophages on glycolytic metabolism (Tavakoli et al., 2013), disruption of this pathway would create a metabolic signal that is responsible for activating NLRP3. We hypothesized that supplementation of downstream metabolites of glycolysis would restore partial metabolic function and block the NLRP3-activating signal. When we cultured GB111-NH2-treated BMDMs with cell-permeable versions of the terminal metabolite of glycolysis, pyruvate, or the TCA cycle metabolite succinate, we observed a dramatic reduction in the number of inflammasome foci that formed (Figure 6A). A structurally related metabolite, lactate, which does not fuel the TCA cycle, did not reduce GB111-NH2-induced NLRP3 inflammasome formation (Figure 6A). Doubling the media concentration of L-glutamine, a metabolite that can be converted into succinate via anaplerosis (Tannahill et al., 2013), significantly reduced the number of inflammasomes that formed. Complete removal of L-glutamine from media sensitized BMDM to GB111-NH2-induced inflammasome formation (Figure 6B). Taken together, this indicates that levels of glycolytic products that fuel downstream metabolism mediate inflammasome induction in response to glycolytic disruption.
In addition to preventing inflammasome formation, supplementation of the glycolytic product pyruvate resulted in significant reductions in caspase-1 activation, IL-1β secretion, and cell death induced by GB111-NH2. Pyruvate supplementation had no effect on inflammasome signaling induced by the NLRP3 activators ATP and nigericin (Figure 6C–E), indicating that pyruvate does not impair NLRP3 inflammasome signaling by a nonspecific mechanism. Pyruvate treatment also blocked inflammasome formation induced by KA and EB (Figure 6F) and restored NADH and ATP production in the treated cells (Figure 6G–H).
We hypothesized that changes in the NAD+/NADH ratio or a drop in ATP concentration could serve as a secondary signal that connects glycolytic disruption to NLRP3 inflammasome formation. To test whether either of these signals is important, we manipulated the NAD+/NADH and ATP levels downstream of glycolysis by chemically blocking specific components of the TCA cycle and oxidative phosphorylation. We first treated LPS-primed BMDM with GB111-NH2 and pyruvate to block glycolysis and stimulate downstream metabolism. We then added the succinate dehydrogenase (TCA cycle enzyme) inhibitor AA5 (Miyadera et al., 2003) to reduce NADH levels, the Complex I inhibitor rotenone to increase NADH levels and reduce ATP production, or the ATP synthase inhibitor Oligomycin A to only inhibit ATP synthesis by oxidative phosphorylation (Figure 7A–B). We found that AA5 addition partially reversed the protection conferred by pyruvate, as demonstrated by an increase in the number of inflammasome complexes. Rotenone treatment suppressed inflammasome formation more than pyruvate alone. Oligomycin A induced a small but statistically insignificant increase in the number of inflammasomes that formed (Figure 7C). The number of inflammasomes positively correlated with a drop in NADH production (an increase in NAD+/NADH ratio) (Figure 7D), while ATP concentration exhibited no correlation with the numbers of inflammasomes (Figure 7—figure supplement 1). Interestingly, rotenone treatment was sufficient to completely abrogate inflammasome formation induced by GB111-NH2 (Figure 7E), conditions under which we also observed a significant decrease in the NAD+/NADH ratio (Figure 7F). These data suggest that the inability to produce NADH, and not ATP, is predictive of NLRP3 inflammasome formation upon glycolytic disruption by GB111-NH2. It should be noted, however, that the α-enolase inhibitor EB did not induce a significant NAD+/NADH ratio defect (Figure 5), suggesting that either EB induces inflammasome activation through a distinct mechanism from GB111-NH2 and KA, or that there are additional universal signals responsible for inflammasome activation downstream of glycolytic disruption.
Mitochondrial ROS and K+ efflux are proposed to be unifying signals preceding NLRP3 inflammasome formation (Tschopp and Schroder, 2010; Muñoz-Planillo et al., 2013). Therefore, we wanted to determine whether either of these signals is relevant to NLRP3 inflammasome activation induced by disruption of glycolysis. We stained BMDM with MitoSOX, a dye that reports accumulation of mitochondrial ROS, and observed that GB111-NH2 induced an increase in cellular MitoSOX fluorescence that was abrogated by addition of pyruvate (Figure 7G). We also found that the ROS scavenger 4-hydroxyTEMPO (4-HT) prevented GB111-NH2-induced caspase-1 cleavage and activation (Figure 7H). Addition of extracellular K+, in contrast, did not reduce the number of inflammasome foci in GB111-NH2-treated BMDMs (Figure 7I) or impair GB111-NH2-induced cell death (Figure 7—figure supplement 2), indicating that mitochondrial ROS, but not K+ efflux, is required for GB111-NH2-induced NLRP3 activation and pyroptosis.
We and others have shown that the intracellular pathogen Salmonella typhimurium (S. typhimurium) requires glucose and its own glycolytic enzymes for intracellular replication (Bowden et al., 2009; 2014). In addition, host defense against S. typhimurium requires the NLRC4 and NLRP3 inflammasomes (Broz et al., 2010). While it is clear that S. typhimurium flagellin and type 3 secretion system proteins activate NLRC4 via NAIP proteins (Zhao et al., 2011), the mechanism by which NLRP3 is activated is not well understood. We hypothesized that S. typhimurium infection may stimulate NLRP3 through disruption of host cell metabolic pathways by co-opting cellular resources during intracellular replication. To test this notion, we infected naïve BMDM with S. typhimurium grown to stationary phase (conditions that lead to NLRP3-dependent inflammasome activation [Broz et al., 2010]). In this infection model, inflammasome complexes begin forming at ~11 hr post-infection and progressively accumulate. We confirmed that infection with either wildtype S. typhimurium or S. typhimurium lacking the SPI-1 secretion system (which genetically limits S. typhimurium to activate NLRP3) induced inflammasome formation (Figure 8A–B). In addition, the percentage of cells with ASC foci was similar in magnitude to GB111-NH2 and alum treatment but lower than nigericin, ATP, or log phase S. typhimurium stimulation (Figure 8C). Using this infection model, we assessed the extent to which intracellular S. typhimurium utilize host-cell glucose by culturing infected BMDM with the fluorescent glucose analog 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose (2-NBDG). We lysed infected BMDM at 5 hr post-infection (while bacteria are still intracellular), harvested the bacterial fraction of BMDM lysates, and measured 2-NBDG fluorescence. We observed fluorescent signal that was dependent on infection in 2-NBDG-treated macrophages, indicating that the bacterial fraction took up the fluorescent glucose analog from the host cell (Figure 8D). We also fixed uninfected and infected 2-NBDG-treated BMDM and analyzed the pattern of 2-NBDG fluorescence by confocal microscopy. We observed the presence of strongly 2-NBDG fluorescent punctae that resemble S. typhimurium and colocalize with Hoechst stain (which stains both host cell and bacterial DNA) in infected BMDM. Furthermore, we quantified the portions of cytosol that were not Hoechst-positive and observed a significant decrease in 2-NDBG fluorescence in the cytosol of infected BMDM compared to uninfected BMDM (Figure 8E). These data indicate that intracellular S. typhimurium derive glucose from host macrophages and reduce host glucose availability.
We measured glycolytic flux in infected BMDM to assess the effect of limited glucose availability on the host macrophages. Importantly, as observed for GB111-NH2 treatment, we observed reduced production of NADH and lactate (Figure 8F–H) that correlated with the multiplicity of infection and magnitude of inflammasome formation in host cells (Figure 8A, Figure 8F–H). These metabolic defects appeared on a similar timescale as initiation of NLRP3 inflammasome formation, suggesting that infection with S. typhimurium has a direct effect on glycolytic flux in host cells.
Consistent with our findings using glycolytic inhibitors, we also observed that supplementation of cells with the glycolytic end product pyruvate significantly reduced inflammasome formation, IL-1β secretion, and cell death induced by S. typhimurium infection (Figure 9A–D). Pyruvate was effective at blocking inflammasome formation in BMDMs infected with both wildtype S. typhimurium and S. typhimurium defective for the SPI-1 secretion system (Figure 9—figure supplement 1). Notably, pyruvate did not completely block ASC focus formation, IL-1β secretion, and death induced by S. typhimurium infection, which could be due to compensatory activation of other inflammasomes such as the non-canonical caspase-11 inflammasome (Broz et al., 2012), or because S. typhimurium could also partially co-opt host pyruvate.
We did not observe inflammasome focus formation in Nlrp3 -/- BMDMs upon infection with stationary phase S. typhimurium. In contrast, Nlrc4 -/- BMDMs had a similar number of inflammasome foci upon infection as wildtype macrophages (Figure 9E). Pyruvate did not prevent inflammasome formation or cell death induced by infection with S. typhimurium in log phase growth (Figure 9—figure supplement 2), an infection model that activates the NLRC4 inflammasome. We also verified that pyruvate was not blocking inflammasome formation by inhibition of bacterial replication using both an intracellular replication reporter plasmid (Helaine et al., 2010) and by monitoring bacterial replication by microscopy (Figure 9F–G). Reporter plasmid expression over the course of the intracellular replication assay indicates that intracellular S. typhimurium are viable in BMDM cultured in DMEM with or without pyruvate (Helaine et al., 2010). In vitro replication assays demonstrated that S. typhimurium grew at a similar rate in minimal media with glucose or pyruvate as a carbon source (Figure 9H), further indicating that pyruvate supplementation affects host cell recognition of intracellular bacteria rather than bacterial dynamics. Importantly, we found that the NAD+ consumption rate increased upon pyruvate treatment (Figure 9I), indicating induction of productive metabolism downstream of glycolysis in infected BMDMs. Taken together, these data indicate that glycolytic perturbation is a mechanism by which innate immune cells sense altered homeostasis during S. typhimurium infection and induce a pro-inflammatory response via NLRP3 inflammasome formation and pyroptotic cell death.
The inflammasome is a multiprotein complex that forms in response to various pathogen- and danger-associated signals. Formation of the inflammasome leads to processing and secretion of pro-inflammatory cytokines to activate the immune system (Lamkanfi and Dixit, 2014; Biswas and Mantovani, 2012). While inflammasome formation and pyroptotic cell death are critical for fighting infection and also contribute to inflammation in diseases including type II diabetes, obesity, and atherosclerosis (Kuemmerle-Deschner et al., 2011; Wen et al., 2011; 2012), the signals that trigger caspase-1 activation remain poorly understood. In this study, we used a small molecule, GB111-NH2, to identify two glycolytic enzymes that regulate inflammasome formation. When functionally blocked, innate immune cells sense metabolic perturbation as a danger signal, resulting in inflammasome formation, caspase-1 activation, and cytokine secretion. Our results using this molecule and other established inhibitors of these enzymes suggest that disrupting glycolytic flux serves as a trigger for inflammation and cell death in macrophages. Disturbance of glycolytic flux by the intracellular pathogen S. typhimurium similarly results in inflammasome formation and pyroptotic cell death in an effort to clear the pathogen. Restoration of metabolism downstream of glycolytic disruption by GB111-NH2 or S. typhimurium infection was sufficient to abrogate the inflammasome response by restoring NADH production and preventing mitochondrial ROS production.
Though the enzymes and metabolites involved in glycolysis are well established, the specific mechanisms that limit glycolytic flux are not well understood. The irreversible reactions within glycolysis, catalyzed by hexokinase, phosphofructokinase, and pyruvate kinase, were historically thought to be rate limiting. However, recent metabolite flux analyses have also suggested that flux through GAPDH, the enzyme separating upper and lower glycolysis, is rate-limiting under nutrient-rich conditions (Shestov et al., 2014). Under similar conditions, we also observed that targeting GAPDH or enzymes in lower glycolysis is sufficient to disrupt glycolytic flux and activate the NLRP3 inflammasome. Conversely, targeting entry into glycolysis using 2DG was not sufficient to disrupt glycolysis in highly glycolytic cells, as has previously been reported (Pietzke et al., 2014), or to activate NLRP3. This reaffirms previous observations that the pre-existing metabolic state of the cell and the point of intervention are equally important for effectively limiting glycolytic flux (Sengupta et al., 2013). Disruption of glycolytic flux led to an NAD+/NADH imbalance and induced mitochondrial ROS accumulation, which has been shown here and elsewhere (Tschopp and Schroder, 2010; Zhou et al., 2010) to activate the NLRP3 inflammasome (Figure 10).
During infection with S. typhimurium, inflammasome activation is an especially important mechanism of host response because, though it kills the host cell, it initiates inflammatory signals that activate the immune system and combat infection (Storek and Monack, 2015). Two inflammasome complexes, NLRP3 and NLRC4, are required to fully combat infection (Broz et al., 2010). The NLRC4 inflammasome responds to a clear pathogen-associated molecular pattern presented by S. typhimurium—cytosolic flagellin and type 3 secretion system components (Zhao et al., 2011). Here, we provide evidence that NLRP3 activation results in response to another effect of S. typhimurium colonization of the host cell, namely disruption of host cell glycolytic metabolism. This could explain a recent study showing that mutants of S. typhimurium defective for the TCA cycle enzyme aconitase induce a more rapid NLRP3-dependent immune response in vivo (Wynosky-Dolfi et al., 2014). We reason that aconitase deficiency would force S. typhimurium to rely even more heavily on glycolysis to survive within the host. These S. typhimurium mutants would likely disrupt cellular glycolysis more quickly and thus activate NLRP3 more rapidly. It is also interesting that, in long-term models of S. typhimurium infection, the bacteria preferentially resides within alternatively activated or ‘M2’ macrophages, which primarily utilize oxidative metabolism rather than glycolysis (Eisele et al., 2013). Thus, the macrophages in which Salmonella survives the longest are those in which host metabolic pathways are minimally perturbed, enabling prolonged infection without invoking an inflammasome response.
These findings additionally shed light on recent work highlighting the connection between metabolic changes and immune system activation (Blatnik et al., 2008; Chawla et al., 2011; Young et al., 1984). For example, metabolic sensing by immune cells has been recently shown to drive NLRP3-dependent IL-1β release and inflammation in diseases ranging from type II diabetes and obesity to Muckle-Wells syndrome (Kuemmerle-Deschner et al., 2011; Strowig et al., 2012), though the specific mechanisms driving macrophage and NLRP3 activation in these diseases have remained unclear. We speculate that, since glucose metabolism is often impaired in these diseases, glycolytic impairment may be the mechanism driving NLRP3-dependent inflammation. Restoring glycolysis or downstream TCA cycle metabolism through supplementation with specific metabolites or activation of glycolytic enzymes could be therapeutically useful for dampening inflammation and associated immunopathology.
In summary, our results suggest that inhibition of glycolysis creates a unique metabolic state that activates the NLRP3 inflammasome. They suggest that innate immune cells sense perturbed metabolite production and flux through the glycolytic pathway, in turn activating NLRP3 to initiate inflammatory responses. Inhibitors of flux-limiting enzymes and S. typhimurium effectively limited glycolysis through distinct mechanisms, each resulting in NLRP3-mediated inflammasome formation and pyroptosis. Glycolytic disruption may be a broadly relevant mechanism of NLRP3 activation triggered in response to metabolic parasitism by microbes. Moreover, this pathway may also provide novel avenues for treating diseases in which NLRP3-driven inflammation results in pathology.
See below for synthesis and characterization of GB111-NH2 and analogs. NMR spectra were recorded on a Varian 400 MHz (400/100) or a Varian Inova 500 MHz (500/126 MHz) equipped with a pulsed field gradient accessory. Chemical shifts (∂) are reported in parts per million (ppm) downfield from tetramethylsilane and are reference to the residual protium signal in the NMR solvents. Data are reported as follows: chemical shift, multiplicity (s=singlet, d=doublet, t=triplet, m=multiplet and q=quartet), coupling constant (J) in Hertz (Hz) and integration. E64d (Enzo Lifesciences, Farmingdale NY), leupeptin (Sigma Aldrich, St. Louis MO), CA074Me (EMD Millipore, Hayward CA), LPS from E. coli 0111:B4 (Sigma Aldrich), 6-aminonicotinamide (Santa Cruz Biotech, Santa Cruz CA), Imject Alum (Pierce Biotechnology, Rockford IL), Atpenin A5 (Santa Cruz Biotech), N-ethylmaleimide (Sigma), rotenone (Sigma), oligomycin A (Cayman Chemical, Ann Arbor MI), nigericin (Cayman Chemical), MitoSOX (Life Technologies, Carlsbad CA), ATP (Sigma), and koningic acid (Adipogen, Switzerland) were purchased from commercial sources, dissolved in vendor-recommended solvents, and used without further purification. ENOblock (Jung et al., 2013) was a generous gift from Dr. Darren Williams.
Strains used in this study were Salmonella typhimurium NCTC 12023 and ATCC SL1344. Bacteria were grown in LB at 37°C with aeration and supplemented with 0.2% arabinose if needed to induce expression of fluorescent proteins.
Mice lacking Pycard, Nlrp3, Nlrc4, and Aim2 have been previously described (Jones et al., 2010; Broz et al., 2010; Kayagaki et al., 2011; Mariathasan et al., 2004). Mice were maintained following guidelines approved by the Stanford University School of Medicine Administrative Panel on Laboratory Animal Care.
BMDM were isolated by culturing mouse bone marrow in DMEM with 2 mM L-glutamine, 10% FBS, and 10 ng/mL recombinant mouse M-CSF (eBioscience, San Diego CA) for 5 days in petri dishes. After 5 days, the cell monolayer was washed several times with sterile PBS to remove cell debris and then the BMDM harvested using CellStripper (Corning CellGro, Manassas VA). BMDM were then plated for experiments, frozen, or cultured for up to a week. One day prior to treatment, cells were seeded in 6 well plates at a density of 1-2x106 cells/well (or 2x105 cells/well of 24-well dish, or 3x104 cells/well of 96-well dish). C57BL/6 SV40-immortalized macrophages were cultured in RPMI with 10% FBS and 2 mM L-glutamine and were a generous gift from Petr Broz.
In this study, biological replicates indicate replicates of the same experiment conducted upon separately seeded cultures on separate days. Technical replicates indicate separate measurements made on cells seeded on the same day and treated simultaneously. The number of biological replicates is indicated in the figure legends and was generally n=3. For plate reader-based assays, experiments were generally conducted in technical triplicate as recommended by assay manufacturers. For microscopy experiments, at least four fields of view were generally analyzed – covering the four quadrants of the cover slip. Within each quadrant, a field was chosen at random using the DAPI channel (to simply find a region that contained cells). Each field of view was counted as a technical replicate because it was a separate measurement of a singly seeded culture. When ascertaining whether differences between samples were statistically significant, an unpaired, two-tailed t test was used. This makes the assumptions that the two samples under analysis were approximately normally distributed and had equal variances. p<0.05 was considered significant. Because measurements were taken within linear range of the detection method (i.e., below saturation and above noise for absorbance-based plate reader assay, within linear range of detector for flow cytometry measurements), etc, technical replicates should be normally distributed around the mean.
BMDM were primed with 100 ng/mL LPS-EK (Invivogen, San Diego CA) or LPS 0111:B4 (Sigma) in DMEM for 3 hr before addition of inflammasome activating agents. GB111-NH2 was added to LPS-primed BMDM at 10 μM (unless otherwise indicated) for inflammasome activation. The canonical NLRP3 activators ATP and nigericin were added to LPS-primed BMDM at 5 mM and 12.5 μM, respectively, typically for 1 hr. Alum (Pierce) was used at a concentration of 100 μg/mL for 5 hr.
For stationary phase infections, S. typhimurium grown to stationary phase (typically overnight culture in LB) were centrifuged onto BMDM for 10min at 500g. After 1 hr, medium was switched to DMEM with 100 μg/mL gentamicin sulfate (Sigma) to kill extracellular bacteria. After 1 hr, cells were washed with plain DMEM and then incubated in DMEM with 10 μg/mL gentamicin sulfate for the remainder of the infection. For log phase infections, S. typhimurium in log phase growth (typically a 4 hr subculture of a 1:50 dilution of an overnight culture) were centrifuged onto BMDM for 10min at 500g in antibiotic-free DMEM. Unless otherwise stated, samples were analyzed after 1 hr of log phase infection.
Probes were diluted to the desired final concentration (1 μM for AWP28, 250 nM for BMV109) from a 1000x stock solution in DMSO directly in the media of the cell monolayer. Cells were labeled for the final hour of treatment at 37°C prior to sample preparation and analysis. For gel labeling experiments, labeled cell monolayers were washed in PBS and lysed directly with 50 μL sample buffer. For harvested supernatants, following treatment the supernatant was removed and proteins precipitated by adding 4 equivalents of cold acetone. Samples were incubated in acetone overnight at -20°C, then proteins pelleted by centrifugation for 5 min at 2000 rpm. Acetone was aspirated and protein pellets dried for 30 min at 37°C before addition of sample buffer. Samples were resolved by SDS-PAGE and visualized on a Typhoon flatbed fluorescent laser scanner (GE Healthcare, United Kingdom).
Following separation of samples by SDS-PAGE and transfer to 0.2 μM nitrocellulose resin (BioRad, Hercules CA), the following antibodies were used. For cell lysates: anti-caspase-1 p10 (1:200, Santa Cruz Biotechnology #514), anti-HSP-90 (1:1000, BD Biosciences, San Jose CA), anti-NLRP3 (1:500, R&D Systems, Minneapolis MN), anti-α-tubulin (1:10000, Sigma), anti-GAPDH (1:1000, Santa Cruz Biotechnology C-9), anti-α-enolase (1:1000, Cell Signaling Technology, Danvers MA). For cell supernatant: anti-IL-1β (1:200, Cell Signaling Technology). HRP-conjugated α-mouse and α-rabbit secondary antibodies were from GE Healthcare.
BMDMs were seeded in triplicate in 96-well plates at a density of 3x104 cells/well. Following treatment, the supernatant was removed and IL-1β, IL-6, or TNF-α release was measured using a Mouse IL-1β, IL-6, or TNF-α READY-SET-GO ELISA kits (eBioscience) according to the manufacturer’s instructions.
BMDMs were seeds in triplicate in 96-well plates at a density of 3x104 cells/well. Following treatment, the supernatant was removed, and the cells were lysed with 2% Triton-X-100 in D-PBS. The lysate was diluted in culture media to the original volume. LDH release was calculated as supernatant LDH activity/total LDH activity using the CytoTox 96 Assay (Promega, Madison WI).
BMDMs were seeded on poly-L-lysine coated glass coverslips in 24 well plates at a density of 2x105 cells/well. Following treatment and labeling with AWP28 (1 μM for final hour of treatment), the cell monolayer was rinsed 3x with warm D-PBS and then fixed with 4% paraformaldehyde in PBS for 15 min at 37°C. The cells were washed with PBS and incubated with anti-ASC (1:200, Santa Cruz Biotechnology N-15) primary antibody in blocking buffer (3% BSA, 0.1% saponin, 0.02% sodium azide in PBS) for 30 min. The cells were washed 3x with PBS and incubated with Alexa 647 or Alexa 594-conjugated secondary antibody (both 1:1000, Invitrogen, Carlsbad CA) for 30min. The cells were washed with D-PBS, mounted in Vectashield with DAPI (Vector Labs, Burlingame CA), and imaged on a Zeiss LSM700 confocal microscope. Snapshots of fields were taken at random (at least 4 fields/condition using a 10x or 20x air objective, typically ~2000 cells/condition). Nuclei were counted using the ITCN plug-in in ImageJ and inflammasome (ASC and/or AWP28 positive) foci were counted using the ‘Analyze Particles’ function in ImageJ after automated thresholding. Replicates indicate cells plated and treated on separate days. For Annexin V and propidium iodide staining, AWP28 labeled cells on coverslips were washed with Annexin V binding buffer (10 mM HEPES pH 7.4, 150 mM NaCl, 2.5 mM CaCl2) and then incubated with 1 μg/mL propidium iodide (ImmunoChemistry, Bloomington MN) and 1:50 Alexa 647 conjugated Annexin V (Invitrogen) in Annexin V-binding buffer on ice for 30 min. Cells were washed with Annexin V-binding buffer and mounted in Vectashield (Vector Labs) for immediate imaging.
C57BL/6 BMDMs were seeded onto 15 cm dishes (2x107 cells/dish). The number of dishes per condition was calculated such that approximately 3 milligrams of protein were yielded per condition. The competition experiment took place as follows: For condition 1, BMDMs were incubated with 100 ng/mL LPS for 3 hr, after which 50 μM az-GB from 100x DMSO stock was added to culture media for 2 hr. For condition 2, BMDMs were incubated with 100 ng/mL LPS for 2 hr. 10 μM GB111-NH2 from 1000x DMSO stock was added to the culture media for 1 hr, after which 50 μM az-GB from 100x DMSO stock was added to culture media for 2 hr. For condition 3, BMDMs were incubated with 100 ng/mL LPS for 2 hr. 50 μM GB-IA from 1000x DMSO stock was added to the culture media for 1 hr, after which 50 μM az-GB from 100x DMSO stock was added to the culture media for 2 hr. For condition 4, BMDMs were incubated with 100 ng/mL LPS for 3 hr, after which vehicle was added for 2 hr. After treatment, all cells were lifted from tissue culture dishes using CellStripper (Corning Cellgro), pelleted at 1000 rpm for 5min, washed once with PBS, and lysed on ice in D-PBS containing 1% NP-40 and 0.1% SDS. Cellular debris was pelleted by centrifugation at 14000 rpm for 15 min at 4°C. The supernatant was removed and protein concentration determined by BCA Assay (Pierce). Protein concentrations were then normalized to 2 mg/mL in PBS with 1% SDS.
Protein samples (>3 mg/condition) then underwent click chemistry. Biotin azide was added to 10 μM final concentration, fresh TCEP (Sigma) to 1 mM, TBTA (Sigma) to 100 uM, and CuSO4 to 1 mM. The samples were allowed to react at room temperature for 3 hr. Proteins were then precipitated using 5 volumes -20°C acetone. After 2 hr, protein precipitates were pelleted. The pellets were washed 4x with -20°C acetone, air dried, and resuspended in PBS with 1.2% SDS. These solutions were incubated with 100 μL streptavidin-agarose beads (Thermo Scientific) at 4°C for 16 hrs. The solutions were then incubated at room temperature for 2.5 hr. The beads were washed with 0.2% SDS/PBS (5 mL), PBS (3 x 5 mL), and water (3 x 5 mL). The beads were pelleted by centrifugation (1400 x g, 3 min) between washes.
The washed beads were suspended in 6 M urea/PBS (500 μL) and 10 mM dithiothreitol (DTT) (from 20X stock in water) and placed in a 65°C heat block for 15 min. Iodoacetamide (20 mM, from 50X stock in water) was then added and the samples were placed in the dark and allowed to react at room temperature for 30 min. Following reduction and alkylation, the beads were pelleted by centrifugation (1400 x g, 3 min) and resuspended in 200 μL of 2 M urea/PBS, 1 mM CaCl2 (100X stock in water), and trypsin (2 μg). The digestion was allowed to proceed overnight at 37°C. The peptide digests were separated from the beads using a Micro Bio-Spin column (BioRad). The beads were washed with water (2 x 50 μL) and the washes were combined with the eluted peptides. Formic acid (15 μL) was added to the samples. These tryptic digests were stored at -20°C until mass spectrometry analysis.
LC-MS analysis was performed on an LTQ Orbitrap Discovery mass spectrometer (ThermoFisher, Waltham MA) coupled to an Agilent 1200 series HPLC. Digests were pressure loaded onto a 250 μm fused silica desalting column packed with 4 cm of Aqua C18 reverse phase resin (Phenomenex, Torrance CA). The peptides were eluted onto a biphasic column (100 μm fused silica with a 5 μm tip, packed with 10 cm C18 and 3 cm Partisphere strong cation exchange resin (SCX, Whatman, United Kingdom) using a gradient 5–100% Buffer B in Buffer A (Buffer A: 95% water, 5% acetonitrile, 0.1% formic acid; Buffer B: 20% water, 80% acetonitrile, 0.1% formic acid). The peptides were eluted from the SCX onto the C18 resin and into the mass spectrometer following the four salt steps outlined in Weerapana et al., 2007. The flow rate through the column was set to ~0.25 μL/min and the spray voltage was set to 2.75 kV. One full MS scan (400–1800 MW) was followed by 8 data dependent scans of the nth most intense ions with dynamic exclusion enabled.
The generated tandem MS data was searched using the SEQEST algorithm against the human UNIPROT database. A static modification of +57 on Cys was specified to account for iodoacetamide alkylation. The SEQUEST output files generated from the digests were filtered using DTASelect 2.0 to generate a list of protein hits with a peptide false-discovery rate of <5%.
When comparing results from Conditions 1–4, spectral counts were first normalized based on the spectral counts of the four endogenously biotinylated mammalian proteins, pyruvate carboxylase, 3-methylcrotonyl CoA carboxylase, propionyl CoA carboxylase, and acetyl CoA carboxylase (Chandler and Ballard, 1985). Condition 4 determined 'background' levels of reactivity with alkyne-biotin. Candidate proteins were those with >30 spectral counts in condition 1, >80% competition by GB111-NH2 for az-GB binding in condition 2, and less than 50% competition by GB-IA for az-GB binding in condition 3. Pearson correlation between enrichment in different samples and expected enrichment was calculated for confidence in hit proteins.
Recombinant GAPDH (ScienCell, Carlsbad CA), α-enolase (BioVision, Milpitas CA) were diluted into assay buffer (50 mM Tris-HCl pH 7.4, 1.5 mM MgCl2) and incubated with inhibitor or vehicle for 30 min at 37°C. After this, az-GB (50 μM) was added for 2 hr at 37°C. TAMRA-alkyne was then added under previously described Click reaction conditions (Child et al., 2013) to visualize az-GB-labeled protein. Reaction mixtures were separated by SDS-PAGE and visualized on Typhoon scanner.
Recombinant GAPDH (0.02 units) was incubated in GAPDH Assay Buffer (ScienCell) for 30 min at 37°C in the presence of inhibitor or vehicle. This mixture was then added to Assay buffer, which contains 6.7 mM phosphoglyceric acid, 3.3 mM L-cysteine, 117 μM β-NADH, 1.13 mM ATP, and 0.05 U 3-phosphoglycerate kinase in 150 μL. A340, representing conversion of β-NADH to NAD+, was measured every minute for 30 min by plate reader (SpectraMax M5, Molecular Devices, Sunnyvale CA). Percentage inhibition was calculated as: (treatment △A340/vehicle △A340)x100.
Approximately 0.013 units of recombinant α-enolase (MyBioSource.com) were incubated in assay buffer (50 mM Tris-HCl pH 7.4, 1.5 mM MgCl2) for 30 min at 37°C in the presence of inhibitor or vehicle. Phosphoenolpyruvate (Sigma) was added to a final concentration of 1.5 mM. A240, representing conversion of phosphoenolpyruvate to 2-phosphoglycerate, was measured every minute for 30 min by plate reader. Percentage inhibition was calculated as: (treatment △A240/vehicle △A240)x100.
BMDM were plated in 96-well dishes at 50 k cells/well. The next day, cells were treated with chemical compound or infected with Salmonella typhimurium. Plates were centrifuged at 500g for 5 min at room temperature, after which culture medium was aspirated and 100 μL lysis buffer (Cayman Chemical) added to each well. Plates were nutated at room temperature for 30 min and then centrifuged at 1000 g for 10 min at 4°C. Supernatants were transferred to wells of a new plate, and 100 μL NAD+/NADH reaction solution (Cayman Chemical) was added to each well. After 1.5 hr, A450 was measured.
Cells were treated and lysates harvested as for the NAD+/NADH assay. After this, NAD+ was decomposed by heating at 60°C for 30 min. Then, reaction solution was added and after 1.5 hr, A450 was measured.
BMDM were plated in 96-well dishes at 50 k cells/well. The next day, cells were treated with chemical compound or infected with Salmonella typhimurium in phenol red-free DMEM. Plates were centrifuged at 500 g for 5 min at room temperature, after which 50 μL of supernatant/well was transferred to a new 96-well dish. Lactate reaction solution (50 μL; Eton Biosciences, San Diego CA) was added. After 30 min, the reaction was quenched with 50 μL/well of 0.5M acetic acid and A490 was measured.
BMDM were plated in opaque-walled 96-well dishes at 50 k cells/well. The next day, the cells were treated with chemical compounds in 100 μL well volume. After 1 hr of treatment at 37°C, the plate was brought to room temperature for 30 min as per manufacturer’s instructions (Promega – CellTiter Glo). ATP reaction mixture was added directly to wells (100 μL/well) and plate was nutated for 2 min to lyse cells. Plate was allowed to stabilize for 10-15 min at room temperature, after which luminescence was read by plate reader (1 s integration time/well).
For ECAR measurements, BMDM were analyzed using a Seahorse XF96 Analyzer. On the day prior to the assay 8x103 BMDM were plated per well of a 96-well Seahorse Analyzer plate. The next day, cells were washed with and then immersed in 180 mL Assay Medium (RPMI at pH 7.4 with 2 mM L-glutamine and without HEPES or sodium bicarbonate). Cells were incubated in a CO2-free incubator for 1 hr at 37°C. At initiation of assay, the plate was loaded into the Seahorse Analyzer, allowed to equilibrate, and compounds injected in Assay Medium with fresh glucose. ECAR was measured for 2 hr after compound injection. Cells were stained with Hoechst and counted after conclusion of assay. Measured ECAR values were normalized to cell number and averaged across each condition.
The fluorescent glucose analog 2-NBDG (Cayman Chemical; Abs/Em 465/540 nm) was used to monitor glucose uptake by both infected and uninfected BMDM. BMDM were infected with SL1344 Salmonella typhimurium grown to stationary phase at an MOI of 100:1. After 1 hr, media was changed to DMEM with high gentamicin (100 μg/mL) to kill extracellular bacteria. After 1 hr, BMDM were washed with plain DMEM and then incubated in DMEM with low gentamicin and 10 μM 2-NBDG. For microscopy analysis, cells were fixed and mounted in Vectashield with DAPI after 4 hr of infection. 2-NDBG was imaged using ‘FITC’ absorption/emission settings in ZenBlack software on a Zeiss LSM700 microscope. Quantification of average cytosolic 2-NBDG fluorescence was done using ImageJ software. In the uninfected condition, cytosol was identified as 2-NBDG (+) areas proximal to nuclei. In the infected condition, cytosol was identified as areas proximal to nuclei that were not Salmonella typhimurium (+). 4 fields per sample were quantified and the average and standard deviation of average cytosolic 2-NDBG fluorescence measurements reported. For measurement of 2-NBDG uptake into S. typhimurium, after 7 hr of infection BMDM were lysed in 0.1% Triton-X-100 in PBS for 10 min. Lysates were centrifuged at 5000 g/10 min/4°C. Supernatant was aspirated and the resulting bacterial pellet resuspended in PBS, transferred to an opaque 96-well plate, and measured in triplicate on a plate reader at Abs/Em 465/540 nm.
Salmonella typhimurium (strain 12023) expressing a replication plasmid were grown overnight in LB containing 0.2% arabinose. BMDM were plated in 12-well dishes at 500 k cells/well and infected with Salmonella typhimurium strain NCTC 12,023 at MOI 25:1. At 12, 16, and 24 hr post-infection, BMDM were lysed and bacterial samples analyzed by flow cytometry. Generations of bacteria were calculated as previously described by Helaine et al. For in vitro growth curves, S. typhimurium were grown in MgM-MES minimal media supplemented with 2 mM glucose, 2 mM pyruvate, or vehicle (ddH2O). OD600 was measured at various timepoints after inoculation of culture.
BMDM were plate in 12-well dishes, primed for 3 hr with 100 ng/mL LPS, and then stimulated in the presence or absence of pyruvate. BMDM were labeled for the last 15min of treatment with 2.5 μM MitoSOX Red (Life Technologies), collected, centrifuged for 5min at 2000 rpm at 4°C, then resuspending in ice cold PBS with 0.5% BSA and analyzed by flow cytometry (488 nm excitation, PE channel collection for MitoSOX Red). >25,000 cells were analyzed per condition.
LPS-primed BMDM were treated with NLRP3-activating compound in Ringer’s buffer with varying concentrations of K+. Osmolarity was kept constant by varying NaCl concentration accordingly.
Peptide carboxylic acid (1eq), was stirred with isobutyl chloroformate (1.1 eq) and N-methyl morpholine (1.2 eq) in anhydrous THF in a bath of dry ice/isopropanol for 1 hr, after which a solution of CH2N2 (approximately 1.7 eq, freshly generated from diazald) was added. The mixture was stirred in dry ice/isopropanol for 1 hr, and then brought to room temperature and stirred for 3 hr. The reaction was quenched with 1:1 concentrated HCl:HOAc (v:v). Ethyl acetate was added to the crude reaction mixture and the organic layer was washed with H2O, saturated NaHCO3, and brine. The organic layers were pooled and dried with MgSO4, and concentrated in vacuo to yield crude chloromethylketone.
Rink resin (1g, 0.59 mmol) was taken up in DMF and deprotected in 20% piperidine in DMF for 45 min at room temperature. The resin was washed with DMF. Fmoc-Lys(Boc)-OH (829 mg, 3 eq, 1.77 mmol), HOBt (239 mg, 3 eq, 1.77 mmol), and DIC (277 μL, 3 eq, 1.77 mmol) were added and the reaction mixture nutated for four hours. The resin was washed with DCM and DMF and the Fmoc group removed by incubation with 20% piperidine in DMF for 45 min. The resin was washed with DMF and Z-Phe-OH (530 mg, 3 eq, 1.77 mmol), HOBt (239 mg, 3eq, 1.77 mmol), and DIC (277 μL, 3 eq, 1.77 mmol) were added and the reaction mixture nutated overnight at room temperature. The resin was washed with DCM and DMF. The product NR-GB111 was cleaved from the Rink resin using 95% TFA, 2.5% triisopropylsilane, and 2.5% H2O for 30 min. The crude was purified by HPLC (reverse phase C18 column, CH3CN/H2O 0.1% TFA, 5:95 to 80:20 over 9 column volumes (CVs) Pure fractions were lyophilized and 5.55 mg (0.013 mmol, 2.2% yield) NR-GB111 (3) were afforded as a white powder.
1H NMR (500 MHz, CD3OD) δ 7.36 – 7.19 (m, 10H), 5.03 (q, J = 12.6 Hz, 2H), 4.38 – 4.27 (m, 2H), 3.08 (dd, J = 13.7, 6.5 Hz, 1H), 2.92 (dd, J = 13.7, 8.6 Hz, 1H), 2.86 (t, J = 7.6 Hz, 2H), 1.93 – 1.79 (m, 1H), 1.69 – 1.53 (m, 3H), 1.47 – 1.32 (m, 2H).
HRMS (ES+): [M+H+]+ calculated for C23H30N4O4 expected mass 427.2345 found 427.2345. LCMS (ES+): retention time 5.57 min.
Chlorotrityl resin (900 mg, 1.134 mmol, 1 eq) was swelled in anhydrous DCM. Fmoc-Lys(Boc)-OH (798 mg, 1.701 mmol, 1.5 eq) and DIPEA (402 μL, 2.31 mmol, 2 eq) were added and the reaction mixture nutated for 3 hr at room temperature. 500 μL anhydrous methanol was added for 30 min. The resin was washed with DCM, DMF, and then resin loading measured (0.531 mmol). The Fmoc group was removed by nutating the resin in 5% DEA in DMF for 30 min at room temperature. The resin was washed with DMF and Fmoc-Phe-OH (617 mg, 1.593 mmol, 3 eq), HOBt (215 mg, 1.593 mmol, 3 eq), and DIC (249 μL, 1.593 mmol, 3 eq) were added and the reaction mixture nutated for 2 hr at room temperature. The resin was washed with DCM and DMF and the Fmoc group removed by nutating in 5% DEA in DMF for 30 min. The resin was washed with DCM and DMF and 4-pentynoic acid (156 mg, 1.593 mmol, 3eq), HOBt (215 mg, 1.593 mmol, 3 eq), and DIC (249 μL, 1.593 mmol, 3 eq) were added and the reaction mixture nutated overnight at room temperature. Intermediate 9 was cleaved from resin using 1% TFA in DCM for 15 min. Concentration with toluene in vacuo yielded a white crystalline solid. The crude was purified by HPLC (reverse phase C18 column, CH3CN/H2O 0.1% TFA, 10:90 to 80:20 over 9 CVs. Pure fractions were lyophilized and 160 mg (0.428 mmol, 80.6% yield) Intermediate 9 were afforded as a white powder.
Carboxylic acid 9 (127 mg, 0.34 mmol was converted to the chloromethylketone using the procedure described above. The crude material was purified by flash column chromatography (20% ethyl acetate in hexane -> 60% ethyl acetate in hexane), and pure fractions pooled to yield 25.6 mg (0.06 mmol, 19% yield) of white crystalline solid.
Intermediate 10 (25.6 mg, 0.05 mmol, 1 eq) was converted to the AOMK following the general procedure. The crude was purified by HPLC (reverse phase C18 column, CH3CN/H2O 0.1% TFA, 20:80 to 60:40 in x column volumes). Pure fractions were pooled and lyophilized. The lyophilized fractions were taken up in 50% TFA in DCM and stirred for 1 hr at room temperature. The reaction was concentrated with toluene in vacuo to yield 4.68 mg (9 μmol, 5.6% yield) of white crystalline solid, GB-IA (4).
1H NMR (400 MHz, CD3OD/CDCl3 1/1) δ 7.32 – 7.24 (m, 4H), 7.23 – 7.15 (m, 2H), 7.06 – 7.01 (m, 2H), 4.61 – 4.41 (m, 4H), 3.10 (dd, J = 13.6, 8.4 Hz, 1H), 3.00 (dd, J = 13.6, 7.4 Hz, 1H), 2.89 (t, J = 7.4 Hz, 2H), 2.44 – 2.39 (m, 4H), 2.35 (s, 6H), 2.16 (t, J = 2.2 Hz, 1H), 2.01 – 1.84 (m, 1H), 1.72 – 1.53 (m, 3H), 1.51 – 1.34 (m, 2H).
HRMS (ES+): [M+H+]+ calculated for C30H37N3O5 expected mass 520.2811 found 520.2797. LCMS (ES+): retention time 6.55 min.
Intermediate 7 (200 mg, 0.38 mmol, 1 eq) was converted to the chloromethyl ketone as described in the general procedure above. The crude was purified by flash column chromatography (20% ethyl acetate in hexane ->60% ethyl acetate in hexane), and pure fractions pooled to yield 150 mg (0.27 mmol, 70% yield) of white crystalline solid.
Intermediate 8 (30 mg, 0.05 mmol) was converted to the acyloxymethylketone as described above in the general procedure. The crude was purified by HPLC (reverse phase C18 column, CH3CN/H2O 0.1% TFA, 20:80 to 60:40 over 25 min, 15 mL per minute. Pure fractions were lyophilized. Lyophilized fractions were taken up in 50% TFA in DCM and stirred for 30 min, after which the reaction mixture was concentrated with toluene in vacuo to yield 14.5 mg (25.26 μmol, 51%) GB111-NH2 as a white powder. Refer to Patent US2007/36725 A1 for previous synthetic scheme of Intermediates 7 and 8 and GB111-NH2 and compound characterization.
GB111-NH2 (1) (4.58 mg, 8.81 μmol,1 eq) was dissolved in anhydrous DCM. Triethylamine (1.35 μL, 9.69 μmol, 1.1 eq) was added and the reaction mixture stirred for 5 min before the addition of acetyl chloride (0.94 μL, 13.21 μmol, 1.5 eq). The mixture was stirred at room temperature for 30 min and then concentrated in vacuo. The crude was taken up in DMSO and purified by HPLC (reverse phase C18 column, CH3CN/H2O 0.1% TFA, 20:80 to 50:50 over column volumes. Pure fractions were lyophilized to yield 0.45 mg (0.73 μmol, 8.2% yield) of white crystalline solid, ac-GB111 (5).
1H NMR (400 MHz, CD3OD/CDCl3 1/1) δ 7.33 – 7.10 (m, 11H), 6.99 (d, J = 7.4 Hz, 2H), 5.00 – 4.98 (m, 2H), 4.65 (s, 2H), 4.42 (dd, J = 11.3, 6.3 Hz, 2H), 3.17 – 3.09 (m, 1H), 3.09 – 3.00 (m, 2H), 2.97 – 2.88 (m, 1H), 2.32 (s, 6H), 1.86 (s, 3H), 1.62 – 1.51 (m, 1H), 1.44 – 1.34 (m, 3H), 1.31 – 1.22 (m, 2H).
HRMS (ES+): [M+H+]+ calculated for C35H41N3O7 expected mass 616.3023 found 616.3017. LCMS (ES+): retention time 8.08 min.
GB111-NH2 (1) (2.2 mg, 3.83 μmol, 1 eq) was dissolved in anhydrous methanol. K2CO3 (1.68 mg, 12.2 μmol, 3 eq), imidazole-1-sulfonyl azide HCl (Goddard-Borger and Stick, 2007) (0.9 mg, 5.2 μmol, 1.36 eq), and Cu(II)SO4 pentahydrate (0.0034 mg, 0.014 mmol, 0.003 eq) were added and the reaction mixture was stirred overnight at room temperature. The reaction mixture was concentrated in vacuo. The crude was taken up in DMSO and purified by HPLC (reverse phase C18 column, CH3CN/H2O 0.1% TFA, 20:80 to 60:40 over column volumes. Pure fractions were lyophilized to yield 1.77 mg (2.95 μmol, 77% yield) of white crystalline solid, az-GB (6).
1H NMR (500 MHz, CD3OD/CDCl3 1/1) δ 7.37 – 7.19 (m, 11H), 7.06 (d, J = 7.6 Hz, 2H), 5.08 (s, 2H), 4.71 – 4.60 (m, 2H), 4.51 – 4.44 (m, 2H), 3.25 (t, J = 6.8 Hz, 2H), 3.11 (dd, J = 13.6, 7.5 Hz, 1H), 2.99 (dd, J = 13.6, 7.4 Hz, 1H), 2.39 (s, J = 6.3 Hz, 6H), 1.99 – 1.87 (m, 1H), 1.68 – 1.51 (m, 3H), 1.51 – 1.32 (m, 2H).
HRMS (ES+): [M+H+]+ calculated for C33H37N5O6 expected mass 600.2822 found 600.2818. LCMS (ES+): retention time 8.90 min.
Potassium fluoride (15.56 mg, 0.27 mmol, 3 eq) and 2,3,5,6-tetrafluorophenol (16.3 mg, 0.1 mmol, 1.1 eq) were added to DMF and the reaction mixture stirred at 80°C for 10 min. Intermediate 10 (50.41 mg, 0.09 mmol, 1 eq) was taken up in DMF and added to the reaction mixture. This mixture was stirred for 2 hr at 80°C then concentrated in vacuo. The crude was taken up in DCM and purified by flash column chromatography (hexane -> 55% ethyl acetate in hexane). Pure fractions were pooled and concentrated in vacuo. This product was taken up in 50% TFA in DCM and stirred for 30 min, after which it was concentrated with toluene in vacuo to yield 35.3 mg GB111-PMK (2) (0.06 mmol, 65% yield) as a white crystalline solid.
1H NMR (500 MHz, cd3od) δ 7.35 – 7.19 (m, 10H), 7.16 – 7.06 (m, J = 14.4, 8.7, 5.3 Hz, 1H), 5.09 – 4.94 (m, 2H), 4.81 – 4.68 (m, 2H), 4.55 – 4.44 (m, 1H), 4.39 – 4.32 (m, 1H), 3.09 – 2.91 (m, 2H), 2.85 (t, J = 7.6 Hz, 2H), 1.95 – 1.74 (m, J = 40.2 Hz, 1H), 1.68 – 1.49 (m, 3H), 1.48 – 1.34 (m, 2H).
HRMS (ES+): [M+H+]+ calculated for C30H31F4N3O5 expected mass 590.2278 found 590.2278. LCMS (ES+): retention time 7.04 min.
Dynamic imaging of protease activity with fluorescently quenched activity-based probesNature Chemical Biology 1:203–209.https://doi.org/10.1038/nchembio728
Redundant roles for inflammasome receptors NLRP3 and NLRC4 in host defense against SalmonellaThe Journal of Experimental Medicine 207:1745–1755.https://doi.org/10.1084/jem.20100257
Distribution and degradation of biotin-containing carboxylases in human cell linesThe Biochemical Journal 232:385–393.https://doi.org/10.1042/bj2320385
Macrophage-mediated inflammation in metabolic diseaseNature Reviews. Immunology 11:738–749.https://doi.org/10.1038/nri3071
Small-molecule inhibition of a depalmitoylase enhances Toxoplasma host-cell invasionNature Chemical Biology 9:651–656.https://doi.org/10.1038/nchembio.1315
Specific inhibition of glyceraldehyde-3-phosphate dehydrogenase by koningic acid (heptelidic acid)The Journal of Antibiotics 38:920–925.https://doi.org/10.7164/antibiotics.38.920
On terminal alkynes that can react with active-site cysteine nucleophiles in proteasesJournal of the American Chemical Society 135:2867–2870.https://doi.org/10.1021/ja309802n
Dynamics of intracellular bacterial replication at the single cell levelProceedings of the National Academy of Sciences of the United States of America 107:3746–3751.https://doi.org/10.1073/pnas.1000041107
Induction of reversible cysteine-targeted protein oxidation by an endogenous electrophile 15-deoxy-delta12,14-prostaglandin J2Chemical Research in Toxicology 17:1313–1322.https://doi.org/10.1021/tx049860+
Absent in melanoma 2 is required for innate immune recognition of Francisella tularensisProceedings of the National Academy of Sciences of the United States of America 107:9771–9776.https://doi.org/10.1073/pnas.1003738107
Koningic acid (heptelidic acid) inhibition of glyceraldehyde-3-phosphate dehydrogenases from various sourcesBiochimica Et Biophysica Acta (BBA) - Protein Structure and Molecular Enzymology 1120:113–116.https://doi.org/10.1016/0167-4838(92)90431-C
Atpenins, potent and specific inhibitors of mitochondrial complex II (succinate-ubiquinone oxidoreductase)Proceedings of the National Academy of Sciences of the United States of America 100:473–477.https://doi.org/10.1073/pnas.0237315100
Caspase-1 activity is required to bypass macrophage apoptosis upon Salmonella infectionNature Chemical Biology 8:745–747.https://doi.org/10.1038/nchembio.1023
Quantitative determinants of aerobic glycolysis identify flux through the enzyme GAPDH as a limiting stepeLife, 3, 10.7554/eLife.03342.
Bacterial recognition pathways that lead to inflammasome activationImmunological Reviews 265:112–129.https://doi.org/10.1111/imr.12289
Quantitation of metabolic and radiobiological effects of 6-aminonicotinamide in RIF-1 tumor cells in vitroCancer Research 57:3956–3962.
Bioenergetic profiles diverge during macrophage polarization: implications for the interpretation of 18F-FDG PET imaging of atherosclerosisJournal of Nuclear Medicine : Official Publication, Society of Nuclear Medicine 54:1661–1667.https://doi.org/10.2967/jnumed.112.119099
NLRP3 inflammasome activation: The convergence of multiple signalling pathways on ROS production?Nature Reviews. Immunology 10:210–215.https://doi.org/10.1038/nri2725
Oxidative metabolism enables Salmonella evasion of the NLRP3 inflammasomeThe Journal of Experimental Medicine 211:653–668.https://doi.org/10.1084/jem.20130627
Benjamin F CravattReviewing Editor; The Scripps Research Institute, United States
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your work entitled "Disruption of glycolytic flux is a signal for NLRP3 inflammasome formation and pyroptosis" for consideration by eLife. Your article has been favorably evaluated by Charles Sawyers (Senior editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
The manuscript by Sanman and colleagues describes the discovery that disruption of glycolysis, either by chemical probes or pathogen infection, activates the NLRP3 inflammasome of macrophages. The initial findings were made using a small-molecule phenotypic screen, which furnished a covalent ligand that blocks GAPDH and enolase activity. This compound, as well as other GAPDH and enolase inhibitors, promoted NLRP3 inflammasome formation, and this effect was blocked by treating cells with the glycolytic end product pyruvate. Further studies indicate that reductions in NADH are important for promoting NLRP3 inflammasome activation. Similar effects were also observed in macrophages infected with Salmonella typhimurium, which appears to impair macrophage glycolysis by consuming host glucose.
The reviewers agreed that your manuscript reports novel findings of potentially sufficient general interest for eventual publication at eLife. However, two major concerns were raised that must be addressed with additional experiments.
1) Both Reviewers 1 and 2 expressed concern about the target identification experiments for GB111-NH2. More evidence to support how GB111-NH2 interacts with GAPDH and enolase-1 to apparently inhibit these proteins, for instance, would benefit the manuscript. Likewise, the use of additional inactive control probes could be helpful (e.g., probes that are structurally related to GB111-NH2, but not reactive, which would support that a covalent mechanism of target binding is required for the observed pharmacological effects).
2) A more thorough characterization of the effects of GB111-NH2 and comparison compounds on the metabolism and inflammatory responses (e.g., IL-1β secretion) of cells would strengthen the paper, as noted in more detail by Reviewer 3.
1) The target identification and validation for GB111-NH2 was difficult for this Reviewer to follow. More details on how the final candidate target list (Supplementary file 3) was generated would be helpful (e.g., what filter were used to designate these proteins as targets over other proteins identified in their proteomic experiments). Also, the data in Figure 4A are confusing – this Reviewer had difficulty noting any substantive change in az-GB labeling of GAPDH in the GB111-NH2 competition lane. Likewise, why is there a background signal in the DMSO lanes for both enzymes? Perhaps this relates to non-specific reactivity of the TAMRA-alkyne tag with the proteins, since click reactions run in the orientation used by the authors are known to exhibit some background reactivity. This point should be clarified. The authors should also consider performing a quantitative, concentration-dependent analysis of GB111-NH2 blockade of az-GB labeling of GAPDH and Eno1, which might better convince readers that these chemical probes are competitively binding/reacting with the same site on each enzyme.
2) Somewhat related to point 1, the authors state in the subsection “Inhibition of glycolytic enzymes activates the NLRP3 inflammasome and induces 139 pyroptosis” that both GAPDH and enolase have "key catalytic site cysteine residues". This Reviewer knows that GAPDH uses a conserved cysteine for catalysis, but was not aware that enolase depends on cysteine residues for activity. Could the authors clarify this statement as relates to enolase and cite relevant papers showing the importance of one or more active-site cysteines in enolase. This is important, since the authors do not appear to directly identify the site of GB111-NH2 reactivity in either enzyme.
Although not explicitly stated in the manuscript, the method used to identify cellular targets of GB111-NH2 is heavily biased towards the identification of proteins that are covalently modified by the small molecule. Therefore, potential cellular targets making non-covalent interactions with GB111-NH2 are likely missed. This should be stated clearly in the text, and is a limitation of the analysis.
The cellular effects of GB111-NH2 are compared against those of published GADPH and α-enolase inhibitors. It is unclear to me if these compounds are selective and properly validated- the data in the references I can find is quite limited. Would be good to clarify this with additional data or text.
The 'inactive compound' still appears to be active against α-enolase (only 2-3 fold less inhibition at 10 µM). Is dual inhibition of GAPDH and α-enolase required for inflammatory signaling or is it only inhibition of GAPDH that is relevant? If we assume that the published ENOblock compound is selective for α-enolase, then this should not be the case.
Does GB111-NH2 react with unique sites on GAPDH/α-enolase? The SAR data shown suggests that it does, however there is no mutational analysis of the proteins to confirm. Furthermore, do the published inhibitors used in the study bind/react with similar sites?
Finally, mapping the binding site of GB111-NH2 would potentially enable identification of resistance mutations, which could be used to convincingly show that GAPDH/α-enolase are indeed the phenotypically relevant targets of the compound.
Figure 5 – measurement of NAD/NADH levels – LPS/GB111 vs. LPS/Nigericin and showing that nigericin does not behave like GB111 in terms of altering NAD/NADH ratios – this is a useful control but are kinetics/extent of inflammasome activation in GB111 vs. nigericin treated cells the same under these conditions? Measurement of LDH release or PI uptake by these cells and showing that they are equivalent is important for this interpretation to be correct. LPS + Enoblock does not seem to do the same thing as LPS + GB111 or LPS + KA (Figure 5C and D), suggesting that blocking α-enolase alone, while inducing inflammasome activation (by IL-1β secretion) does not do so in the same way as GB111.
Addition of downstream metabolites to GB111-treated cells prevented inflammasome activation providing a nice test of the authors' hypothesis. A prediction is that these metabolites should also reverse alteration of the NAD/NADH ratio by GB111. Is this the case? If the NAD/NADH ratio is the thing being sensed by the cell as a trigger of inflammasome activation, is it possible to directly alter these ratios in primed cells to mimic the effect of GB111, or conversely to override the effect of GB111?
Figure 7 – lack of inhibition of GB111-induced inflammasome activation by extracellular potassium is quite interesting, but it would be important to have an additional readout besides just% of cells with ASC foci – LDH release or IL-1β secretion – 5% is quite low, and although the background tends to be very low as well, it isn't much above noise, so having an additional way of confirming this is important.
Figure 9 – the levels of IL-1β secretion that are being measured in Figure 9B are low relative to what is typically seen. While there are other data that support the conclusion that disruption of glycolysis by Salmonella induces NLRP3 inflammasome previous literature (for example Broz et al. 2014) suggests that there should be quite a bit more IL-1β secretion under these conditions. Moreover, while it is fair to say that pyruvate reduces Salmonella-induced inflammasome activation, a drop from 80 to 40 pg/mL IL-1β or a drop from 30% to 25% LDH does not seem exceptionally convincing – since all of the inflammasome activation that Salmonella is inducing under these conditions depends on NLRP3, what accounts for the pyruvate-independent inflammasome activation in response to Salmonella? Presumably this is due to caspase-11. At the very least this discrepancy between the effect of pyruvate on Salmonella-induced inflammasome activation and GB111 induced inflammasome activation should be discussed. A further test of whether Salmonella induces NLRP3 inflammasome activation by altering the NAD/NADH ratio is whether adding pyruvate or glutamine to Salmonella-infected cells prevents this NLRP3 inflammasome activation.
Figure 1—figure supplement 1 – this is a poor quality western blot that should really be cleaned up for publication purposes. The relevant band is hard to distinguish from the background in the western blot. Even though this is a supplemental figure it should still be done to the same standard as a main text figure.
Figure 1—Figure supplement 2 – the title of this figure is that GB111 doesn't impair secretion of inflammasome-independent pro-inflammatory cytokines, but it has this effect only for TNF – it DOES inhibit IL-6 in a dose-dependent way. The explanation is likely related to the difference in kinetics of IL-6 and TNF transcription – since the priming was only done for 3 hours, it is likely that at higher concentrations of GB111 more of the cells are dying more quickly, leading to the appearance of an effect on IL-6, but not TNF since TNF is a primary response gene and much more rapidly expressed. This is testable with a longer period of LPS priming. This should either be done or the title of the figure legend should be changed to appropriately reflect the data.https://doi.org/10.7554/eLife.13663.039
The reviewers agreed that your manuscript reports novel findings of potentially sufficient general interest for eventual publication at eLife. However, two major concerns were raised that must be addressed with additional experiments. 1) Both Reviewers 1 and 2 expressed concern about the target identification experiments for GB111-NH2. More evidence to support how GB111-NH2 interacts with GAPDH and enolase-1 to apparently inhibit these proteins, for instance, would benefit the manuscript. Likewise, the use of additional inactive control probes could be helpful (e.g., probes that are structurally related to GB111-NH2, but not reactive, which would support that a covalent mechanism of target binding is required for the observed pharmacological effects).We conducted several additional studies to further demonstrate how GB111-NH2 interacts with both GAPDH and α-enolase and added data from these studies to the manuscript.
We agree that the GAPDH labeling by az-GB was initially hard to interpret (revised manuscript Figure 4—figure supplement 1). We optimized buffer conditions and now observe much cleaner labeling. Using these conditions, we performed a dose-response pretreatment of GB111-NH2 before labeling with the az-GB probe. We observe that GB111-NH2 inhibits az-GB binding to GAPDH and α-enolase in a dose-dependent manner, indicating that it likely binds to a discrete site on both enzymes. Compound 6, which contains a more reactive electrophile than GB111-NH2 and more potently induces IL-1β secretion in BMDM (Figure 2), also competed with az-GB for binding to both GAPDH and α-enolase. Importantly, this compound showed competition at lower concentrations than GB111-NH2, consistent with it having greater potency for the two targets. Furthermore, as suggested by the reviewers, we more carefully evaluated the two negative control compound. This includes compound 2 which lacks the electrophile and GB-IA which is identical to GB111 but lacks a carboxybenzyl cap. Both of these compounds did not induce IL-1β release and we show now also do not compete for az-GB binding to GAPDH or α–enolase (revised manuscript Figure 4A-B). Collectively, these data strengthen our correlation between in vitro target-binding and cellular effects.
Our initial observation in our original manuscript that the cysteine alkylating compound N-ethylmaleimide blocked az-GB binding to GAPDH and α-enolase (revised manuscript Figure 4—figure supplement 1), suggested that GB111-NH2 covalently binds to reactive cysteine residues on GAPDH and α-enolase. To further explore this mechanism of compound binding, we performed labeling experiments on purified GAPDH and α-enolase with iodoacetamide fluorescein (IAF), which labels reactive cysteine residues. Our collaborator in this study has previously shown (Weerapana et al., Nature 2010) that iodoacetamide preferentially labels the active site Cys 152 of GAPDH and Cys 388 of α-enolase. Notably, chemical modification of Cys 388 is deleterious to α-enolase activity (Ishii et al., Chem Res Toxicol 2004). Our new data (revised manuscript Figure 4C-D) shows that IAF labels both GAPDH and α-enolase and is competed by both N-ethylmaleimide and by GB111-NH2. Together, these results indicate that GB111-NH2 binds to the same active site cysteines as IA and support a mechanism in which GB111-NH2 function by blocking activity of these enzyme targets by direct binding to active site cysteine residues.
To further support a covalent mechanism of target binding, we conducted experiments in which we pretreated GAPDH and α-enolase with GB111-NH2 for increasing periods of time. Longer pretreatment resulted in increased inhibition of both GAPDH and α-enolase (revised manuscript Figure 4F), which indicates that GB111-NH2 is a covalent inhibitor, again consistent with a mechanism in which inhibition results from covalent modification of the active site cysteines.
Taken together, these data support the hypothesis that GB111-NH2 covalently binds to reactive cysteine residues on both GAPDH and α-enolase and identifies target cysteines that GB111-NH2 may bind. Furthermore, these data strengthen the correlation between the binding of our GB111-NH2 analogs to GAPDH and α-enolase and their ability to induce IL-1β secretion in cells.
Finally, while we agree rigorous analysis of target binding is important for any small molecule, we would like to point out that the major diving focus of this paper is not to identify a new class of selective inhibitors of GAPDH or α-enolase. Rather, we use these tools to uncover a previously unknown pathway for activation of pyroptosis that is relevant to pathogen infection. By showing this result using many different classes of small molecule and pathogen stimuli we feel that we have made the point very clearly. Finally, by rescuing the phenotype induced by both the small molecule stimuli as well as the pathogen using downstream components of glycolysis, we confirm that changes in glycolytic flux are in fact responsible for the induction of pyroptosis, thus making extensive characterization of all possible targets of the original small molecule relatively unimportant to the conclusions of the paper.
2) A more thorough characterization of the effects of GB111-NH2 and comparison compounds on the metabolism and inflammatory responses (e.g., IL-1β secretion) of cells would strengthen the paper, as noted in more detail by Reviewer 3.As suggested by the reviewers, we identified a concentration of nigericin (1 μM) which has a similar timecourse of inflammasome formation/pyroptosis to 10 μM GB111-NH2. We analyzed NADH levels (alongside LDH release and IL-1β secretion) and found that NADH levels are reduced by treatment with GB111-NH2, but that the nigericin controls do not show this reduction in NADH levels (see Figure 5—figure supplement 2). We think that these data more convincingly indicate that changes in metabolism observed upon GB111-NH2 treatment are indeed due to its proposed mechanism of action rather than a side effect of inflammasome formation and/or cell death.
We agree with the Reviewer’s conclusion that LPS+Enoblock has a somewhat different metabolic signature from LPS+GB111 and LPS+KA, specifically that NAD+/NADH ratio is not altered in BMDM by Enoblock treatment. Therefore, it is possible that a) all glycolytic inhibitors converge on a different secondary signal than NAD+/NADH that creates mitochondrial ROS and activates the inflammasome or, b) Enoblock activates the inflammasome through a distinct mechanism. We addressed these possibilities in the text as they are important avenues for future studies.
Reviewer 1: 1) The target identification and validation for GB111-NH2 was difficult for this Reviewer to follow. More details on how the final candidate target list (Supplementary file 3) was generated would be helpful (e.g., what filter were used to designate these proteins as targets over other proteins identified in their proteomic experiments). Also, the data in Figure 4A are confusing – this Reviewer had difficulty noting any substantive change in az-GB labeling of GAPDH in the GB111-NH2 competition lane. Likewise, why is there a background signal in the DMSO lanes for both enzymes? Perhaps this relates to non-specific reactivity of the TAMRA-alkyne tag with the proteins, since click reactions run in the orientation used by the authors are known to exhibit some background reactivity. This point should be clarified. The authors should also consider performing a quantitative, concentration-dependent analysis of GB111-NH2 blockade of az-GB labeling of GAPDH and Eno1, which might better convince readers that these chemical probes are competitively binding/reacting with the same site on each enzyme.The Reviewer is correct regarding the background signal in the DMSO lanes – we believe that the background labeling we see is nonspecific because alkynes can function as cysteine-targeting electrophiles (Ekkebus et al., JACS 2013) and both GAPDH and α-enolase contain reactive cysteine residues. We changed the buffer to one in which nonspecific interaction of the TAMRA-alkyne with α-enolase and GAPDH is reduced to allow clearer interpretation of our results (Figure 4A, C). We will address this in the text to clarify for the readers.
We did experiments looking at competition of GB111-NH2 and analogs for az-GB binding to both GAPDH and α-enolase over several concentrations, as the reviewer suggested. We observe dose-dependent inhibition of az-GB binding to both GAPDH and α-enolase by GB111-NH2. Compound 6, which contains a more reactive electrophile than GB111-NH2 and more potently induces IL-1β secretion in BMDM (Figure 2), also competed az-GB binding to both GAPDH and α-enolase more potently than GB111-NH2. Importantly, both the GB111-NH2 analog that lacks an electrophile (compound 2) and the GB111 analog that lacks a carboxybenzyl cap (GB-IA) (both of which did not induce IL-1β secretion in BMDM) did not compete for az-GB binding to GAPDH and α–enolase (revised manuscript Figure 4A-B).
In addition, we conducted experiments where we pretreated GAPDH and α-enolase with GB111-NH2 for increasing amounts of time and then measured activity. Longer pretreatment time resulted in less enzyme activity (Figure 4F), which is a characteristic of covalent inhibitors.
2) Somewhat related to point 1, the authors state in the subsection “Inhibition of glycolytic enzymes activates the NLRP3 inflammasome and induces 139 pyroptosis” that both GAPDH and enolase have "key catalytic site cysteine residues". This Reviewer knows that GAPDH uses a conserved cysteine for catalysis, but was not aware that enolase depends on cysteine residues for activity. Could the authors clarify this statement as relates to enolase and cite relevant papers showing the importance of one or more active-site cysteines in enolase. This is important, since the authors do not appear to directly identify the site of GB111-NH2 reactivity in either enzyme.We agree that this statement is a bit confusing and will clarify both here and in the main text of the paper. As the Reviewer notes, GAPDH has an active site cysteine. Our collaborator (Weerapana et al., Nature 2010) has identified a reactive cysteine in α-enolase that can be labeled with iodoacetamide. Based on the crystal structure of human α-enolase (Kang et al., Acta Crystallogr Sect D 2008), this cysteine, Cys 388 is close to the entrance to the enzyme’s substrate-binding site. Furthermore, it has previously been shown that modification of this cysteine residue results in a loss of activity (Ishii et al., Chem Res Toxicol 2004).
We did an experiment to confirm that GAPDH and α-enolase contain reactive cysteines by labeling with the reactive cysteine probe iodoacetamide-fluorescein (IAF). Both label with IAF and labeling can be blocked by incubation with the cysteine-alkylating compound N-ethylmaleimide (NEM). We observe that GB111-NH2 pretreatment results in reduced iodoacetamide fluorescein labeling (Figure 4C-D), indicating that GB111-NH2 binds to these reactive cysteines.
Reviewer 2: Although not explicitly stated in the manuscript, the method used to identify cellular targets of GB111-NH2 is heavily biased towards the identification of proteins that are covalently modified by the small molecule. Therefore, potential cellular targets making non-covalent interactions with GB111-NH2 are likely missed. This should be stated clearly in the text, and is a limitation of the analysis.This will be stated in the text. Based on data with our non-reactive analog of GB111-NH2, we did not think that non-covalent interactions are important for the phenotype under study, but we will state this clearly anyway.
The cellular effects of GB111-NH2 are compared against those of published GADPH and α-enolase inhibitors. It is unclear to me if these compounds are selective and properly validated- the data in the references I can find is quite limited. Would be good to clarify this with additional data or text.The relevant papers in which specificity characterization was reported have been cited. Specifically, specificity of KA was determined amongst enzymes in the glycolytic pathway in Endo et al., J Antibiotics 1985 (cited in our manuscript). Proteomic analysis has not been undertaken but its effects in different species correlate with the in vitrosensitivity of those species GAPDH isoforms (indirect genetic evidence of its effects). When EB was first reported (ACS Chem Biol 2013), the authors performed an affinity enrichment of all binding partners of EB. The proteins that they identified were all enolase isoforms (Jung et al., ACS Chem Biol 2013).
The 'inactive compound' still appears to be active against α-enolase (only 2-3 fold less inhibition at 10 µM). Is dual inhibition of GAPDH and α-enolase required for inflammatory signaling or is it only inhibition of GAPDH that is relevant? If we assume that the published ENOblock compound is selective for α-enolase, then this should not be the case.Our data suggest that inhibition of either GAPDH or α-enolase is capable of activating the inflammasome. GAPDH inhibition may drive the phenotype more than α-enolase inhibition, based on the observation that the metabolic signature of GB111-NH2 is more similar to KA. However, the increase in potency from GB111-NH2 relative to KA or EB alone may be due to a synergistic effect of blocking both. Therefore, we do not think that the in vitro activity that the ‘inactive compound’ exhibits towards -enolase is sufficient to cause it to have activity in cells, especially if α-enolase inhibition is less of a driver of the biological phenotype.
Does GB111-NH2 react with unique sites on GAPDH/α-enolase? The SAR data shown suggests that it does, howeverα
there is no mutational analysis of the proteins to confirm. Furthermore, do the published inhibitors used in the study bind/react with similar sites?KA binds to the catalytic cysteine of GAPDH. The binding site of EB on α-enolase has not been characterized. We did a competition labeling study of GAPDH and α-enolase with 10 μM iodoacetamide-fluorescein (IAF) (Figure 4C-D). According to proteomic data previously generated by the authors (Weerapana et al., Nature 2010), 10 μM iodoacetamide will predominantly label the active site Cys 152 of GAPDH and Cys 388 of α-enolase (a cysteine flanking the substrate-binding pocket of α-enolase). We see that IAF binding to both GAPDH and α-enolase is competed by GB111-NH2 pretreatment, indicating that GB111-NH2 also binds to these cysteines.
Finally, mapping the binding site of GB111-NH2 would potentially enable identification of resistance mutations, which could be used to convincingly show that GAPDH/α-enolase are indeed the phenotypically relevant targets of the compound.An interesting study, but beyond the scope of this paper. The feasibility of this study is also questionable because it has previously been shown that expression of GAPDH mutants that lack activity is cytotoxic (Yogalingam et al., JBC 2013), and our studies are being conducted in BMDM which do not live long enough in culture to generate resistance mutants.
Reviewer 3:Figure 5 – measurement of NAD/NADH levels – LPS/GB111 vs. LPS/Nigericin and showing that nigericin does not behave like GB111 in terms of altering NAD/NADH ratios – this is a useful control but are kinetics/extent of inflammasome activation in GB111 vs. nigericin treated cells the same under these conditions? Measurement of LDH release or PI uptake by these cells and showing that they are equivalent is important for this interpretation to be correct. LPS + Enoblock does not seem to do the same thing as LPS + GB111 or LPS + KA (Figure 5C and D), suggesting that blocking α-enolase alone, while inducing inflammasome activation (by IL-1β secretion) does not do so in the same way as GB111.We have done a more careful titration of nigericin to find a concentration at which the kinetics of inflammasome activation and cell death are similar to those in GB111-NH2-treated BMDM. At this concentration of nigericin (1 μM), we see a decrease in NADH production upon GB111-NH2 treatment but do not observe a decrease in NADH production in the nigericin-treated cells (Figure 5—figure supplement 1).
The α-enolase inhibitor does not have an identical metabolic signature to the GAPDH inhibitor or GB111-NH2 – though lactate production is impaired, NAD+/NADH is not significantly altered. This either indicates that the α-enolase inhibitor is working through a distinct mechanism from glycolytic inhibition or that NAD+/NADH, while predictive of inflammasome formation, is not the universal signal sensed downstream of glycolytic disruption. To prevent confusion for future studies, this will be directly discussed in the text.
Addition of downstream metabolites to GB111-treated cells prevented inflammasome activation providing a nice test of the authors' hypothesis. A prediction is that these metabolites should also reverse alteration of the NAD/NADH ratio by GB111. Is this the case? If the NAD/NADH ratio is the thing being sensed by the cell as a trigger of inflammasome activation, is it possible to directly alter these ratios in primed cells to mimic the effect of GB111, or conversely to override the effect of GB111?These metabolites do reverse the NAD+/NADH ratio defect observed upon GB111 treatment – see Figure 6G. We do override this by adding rotenone (see Figure 7E-F), where we see a reversal of the NAD+/NADH ratio caused by GB111 and then see complete suppression of inflammasome formation induced by GB111.
Figure 7 – lack of inhibition of GB111-induced inflammasome activation by extracellular potassium is quite interesting, but it would be important to have an additional readout besides just% of cells with ASC foci – LDH release or IL-1β secretion – 5% is quite low, and although the background tends to be very low as well, it isn't much above noise, so having an additional way of confirming this is important.We think that this is also an interesting point so we also looked at LDH release under these conditions. We also observe that high extracellular K+ blocks nigericin-induced LDH release but does not have a significant effect on GB111-induced LDH release (Figure 7—figure supplement 2).
Figure 9 – the levels of IL-1β secretion that are being measured in Figure 9B are low relative to what is typically seen. While there are other data that support the conclusion that disruption of glycolysis by Salmonella induces NLRP3 inflammasome previous literature (for example Broz et al.
2014) suggests that there should be quite a bit more IL-1β secretion under these conditions. Moreover, while it is fair to say that pyruvate reduces Salmonella-induced inflammasome activation, a drop from 80 to 40 pg/mL IL-1β or a drop from 30% to 25% LDH does not seem exceptionally convincing – since all of the inflammasome activation that Salmonella is inducing under these conditions depends on NLRP3, what accounts for the pyruvate-independent inflammasome activation in response to Salmonella? Presumably this is due to caspase-11. At the very least this discrepancy between the effect of pyruvate on Salmonella-induced inflammasome activation and GB111 induced inflammasome activation should be discussed. A further test of whether Salmonella induces NLRP3 inflammasome activation by altering the NAD/NADH ratio is whether adding pyruvate or glutamine to Salmonella-infected cells prevents this NLRP3 inflammasome activation.The difference could be due to caspase-11. Another explanation is that pyruvate supplementation is not completely blocking NLRP3 inflammasome formation because it is actively used – it is not a genetic block. However, it is true that pyruvate more effectively blocks GB111-NH2-induced NLRP3 inflammasome formation than Salmonella typhimurium-induced NLRP3 inflammasome formation, and we have added text to point this out and discuss both potential reasons for this difference. As a note, we do observe that cells infected with Salmonella, when supplemented with pyruvate, regain their ability to turn over NAD+ (Figure 9I).
Figure 1—figure supplement 1 – this is a poor quality western blot that should really be cleaned up for publication purposes. The relevant band is hard to distinguish from the background in the western blot. Even though this is a supplemental figure it should still be done to the same standard as a main text figure.While we agree that the western blot is not ideal, we have not be able to get any significantly better results with the antibody we have. We think that the results are still clear in that we consistently see the appearance of the processed IL-1β mature form in cells at concentrations of GB111-NH2 where we get induction of pyroptosis and accumulation of IL-1b signals by ELISA.
Figure 1—figure supplement 2 – the title of this figure is that GB111 doesn't impair secretion of inflammasome-independent pro-inflammatory cytokines, but it has this effect only for TNF – it DOES inhibit IL-6 in a dose-dependent way. The explanation is likely related to the difference in kinetics of IL-6 and TNF transcription – since the priming was only done for 3 hours, it is likely that at higher concentrations of GB111 more of the cells are dying more quickly, leading to the appearance of an effect on IL-6, but not TNF since TNF is a primary response gene and much more rapidly expressed. This is testable with a longer period of LPS priming. This should either be done or the title of the figure legend should be changed to appropriately reflect the data.
We agree with the reviewer’s explanation of the IL-6 data, given published time course data of IL-6 secretion after LPS priming (Bjorkbacka et al. Physiol. Genomics 2004). We changed the title of the figure legend accordingly.https://doi.org/10.7554/eLife.13663.040
- Matthew Bogyo
- Laura E Sanman
- Eranthie Weerapana
- Denise M Monack
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank the Bogyo lab for helpful discussions regarding the direction of this project, M Child and S Ewald for comments on the manuscript, M Spitzer for comments and flow cytometry assistance, A Ondrus, T-H Lee, A Bhaduri, and E LaGory for assay advice and reagents. This work was supported by the National Science Foundation under grant DGE-114747 (to LES) and by the National Institutes of Health under grants R01 EB005011 and R01 CA179253 (to MB) and AI063302 and AI065359 (to DMM). EW is grateful for financial support from the Smith Family Foundation, the Damon Runyon Cancer Research Foundation (DRR-18-12) and Boston College.
Animal experimentation: This work was approved under ABP protocol 1331 (Entitled Chemical probes to study host responses to bacterial pathogens) and APLAC protocol 18026. Primary cells were isolated from mouse bone marrow following strict accordance with the NIH guide for the care and use of laboratory animals. These protocols were reviewed and approved by the Environmental Health and Safety Department of Stanford University and the Institutional Animal Care and Use Committee of Stanford University, respectively.
- Benjamin F Cravatt, The Scripps Research Institute, United States
© 2016, Sanman et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Downloads (link to download the article as PDF)
Download citations (links to download the citations from this article in formats compatible with various reference manager tools)
Open citations (links to open the citations from this article in various online reference manager services)
A small fraction of HIV-1- infected humans develop broadly neutralizing antibodies (bNAbs) against HIV-1 that protect macaques from simian immunodeficiency HIV chimeric virus (SHIV). Similarly, a small number of macaques infected with SHIVs develop broadly neutralizing serologic activity, but less is known about the nature of simian antibodies. Here, we report on a monoclonal antibody, Ab1485, isolated from a macaque infected with SHIVAD8 that developed broadly neutralizing serologic activity targeting the V3-glycan region of HIV-1 Env. Ab1485 neutralizes 38.1% of HIV-1 isolates in a 42-pseudovirus panel with a geometric mean IC50 of 0.055 µg/mLl and SHIVAD8 with an IC50 of 0.028 µg/mLl. Ab1485 binds the V3-glycan epitope in a glycan-dependent manner. A 3.5 Å cryo-electron microscopy structure of Ab1485 in complex with a native-like SOSIP Env trimer showed conserved contacts with the N332gp120 glycan and gp120 GDIR peptide motif, but in a distinct Env-binding orientation relative to human V3/N332gp120 glycan-targeting bNAbs. Intravenous infusion of Ab1485 protected macaques from a high dose challenge with SHIVAD8. We conclude that macaques can develop bNAbs against the V3-glycan patch that resemble human V3-glycan bNAbs.
Bacteria of the genus Shigella cause shigellosis, a severe gastrointestinal disease that is a major cause of diarrhea-associated mortality in humans. Mice are highly resistant to Shigella and the lack of a tractable physiological model of shigellosis has impeded our understanding of this important human disease. Here we propose that the differential susceptibility of mice and humans to Shigella is due to mouse-specific activation of the NAIP–NLRC4 inflammasome. We find that NAIP–NLRC4-deficient mice are highly susceptible to oral Shigella infection and recapitulate the clinical features of human shigellosis. Although inflammasomes are generally thought to promote Shigella pathogenesis, we instead demonstrate that intestinal epithelial cell (IEC)-specific NAIP–NLRC4 activity is sufficient to protect mice from shigellosis. In addition to describing a new mouse model of shigellosis, our results suggest that the lack of an inflammasome response in IECs may help explain the susceptibility of humans to shigellosis.