Well-tolerated Hv1 pharmacologic inhibition hampers the inflammatory activation profile of microglia. Experiments were performed both on HAPI cells and primary microglia cultures, which both express Hv1 (Fig. S6). We first sought to characterize the main response properties of HAPI cells upon activation by LPS/IFNγ and to determine whether any of these responses would be altered by the presence of an Hv1 antagonist. HAPI cells were first exposed to increasing concentrations of LPS (0.25-5 mg/ml) in the presence of 0.1 mg/ml IFNγ for 24 hours. We note that concentrations higher than 0.5 mg/ml of LPS (with 0.1 mg/ml IFNγ) reliably increased all inflammatory activation parameters measured (Fig. S1), consistent with an M1 phenotype. We next established the maximal concentration of an Hv1 antagonist these cells could tolerate without observable cell death (Fig. S2). Cell death was quantified using MTS and LDH assays during both basal and activated conditions (1 mg/ml LPS/0.1 mg/ml IFNγ) 24 hours after exposure to increasing concentrations (5–20 µm) of ClGBI, an Hv1 blocker. ClGBI is highly efficacious (~ 90% channel block) and the most potent (EC50 = 1 µM) of the guanidine-derived antagonists of the proton channel49. We observed that 10 µM of the blocker was well-tolerated without any observable cell death in both non-activated and activated HAPI cells (Fig. S2 A-D). Next, using the MT-luciferase assay, we monitored the long-term viability of non-activated and activated HAPI cells that had been exposed to vehicle or 10 µM ClGBI for 24 h. Interestingly, no apparent long-term impact of Hv1 antagonism was observed in non-activated or activated cells after 72 hours after treatment. In the next set of experiments, we use the same approach in primary microglia cultures obtained from postnatal mouse brain. For these studies, we first evaluated the sensitivity of these cells to the presence of increasing concentrations of ClGBI, similar to the characterization performed in HAPI cells (Fig. S3). We found that the primary microglia cultures could withstand 10-fold less the concentration of the antagonist when compared to our cell line, showing as clear decrease in viability when exposed to ClGBI concentrations higher than 1 µM for 24 hours, both under basal and activated (1.0 mg/ml LPS/0.1 mg/ml IFN) conditions (Fig. S3A-D). Additionally, we assessed the effect of treatments on the long-term viability of microglia cultures. We found that after 4 weeks, the viability of activated microglia decreases by 50%. Remarkably, co-administration of 1 µM ClGBI restores viability by 25% (Fig. S3E). Given these results, we chose to utilize 1 mg/ml/0.1 mg/ml IFNγ to activate both immortalized and primary microglial cells as well to utilize 10 µM ClGBI for HAPI cells and 1 µM ClGBI for primary microglia cultures to antagonize Hv1 in subsequent studies.
As our initial data suggested an increase in ROS production (S1B) after activation of HAPI cells, we quantified the protein expression levels of the main reactive species-producing enzymes during HAPI cell activation, namely NOX2 (Fig. 1A and S7A) and NOS2 (Fig. 1B and S7B). We found that following HAPI cell activation NOX2 protein was increased nearly two-fold (Fig. 1C). Importantly, the presence of ClGBI during activation attenuated this response although incubation with ClGBI alone had a small stimulatory effect (Fig. 1C). In the case of NOS2, activation produced a dramatic 60-fold increase in protein expression, while Hv1 inhibition during activation reduced NOS2 expression to approximately half this level (Fig. 1D). ROS production itself was next evaluated utilizing DhE fluorescence, which detects superoxide and hydrogen peroxide (Fig. 1E). We observed a substantial increase in fluorescence in activated cells, which was significantly mitigated by ClGBI (Fig. 1F). Given that Hv1 antagonism reduced the expression of both NOX2 and NOS2 in activated cells, we hypothesized that the ClGBI-mediated decrease in reactive species production was due, in part, to a reduction in NOX and NOS2 enzymatic activity. Indeed, quantification of enzymatic activity revealed that while NOX activity increased threefold in response to activation, Hv1 inhibition completely suppressed this response (Fig. 1G and S4A). Additionally, activated cells showed a 20-fold increase in NOS2 activity, a phenomenon that was partially blocked by Hv1 inhibition (Fig. 1H). We next quantified phagocytic activity to establish whether this important activation parameter could also be regulated by Hv1 antagonism. After activation, HAPI cells more than doubled their phagocytic activity, while co-administration of ClGBI effectively limited this increase and returned it near baseline levels (Fig. 1I and S5B). Finally, we measured the extracellular concentration of zinc ([Zn2+]e) since macrophage response to LPS is associated with changes in zinc homeostasis77. We observed a significant increase in [Zn2+]e, secondary to the activation of HAPI cells, which was not modified by Hv1 antagonism (Fig. 1J). In sum, our data strongly suggest that HAPI microglial cells show a complete, general inflammatory pattern following activation. Furthermore, our data prove that Hv1 antagonism during activation can effectively hamper the extent of reactive oxygen and nitrogen species generated by limiting their enzymatic production. Importantly, while activation leads to enhance phagocytic activity, Hv1 antagonism in activated microglia dampens but does not eliminate this critical function of these cells.
Having established the inflammatory activation pattern of immortalized microglia cells and the extent to which Hv1 inhibition modifies key components of this phenotype, we next aimed to establish whether primarily microglial cells responded similarly. We compared the expression levels of NOX2 (Fig. 2A and S7C) and NOS2 (Fig. 2B and S7D) in primary mouse microglial cultures at rest and following activation, both with and without Hv1 inhibition. We observed that ClGBI (1µM) alone could enhance NOX2 expression in non-activated microglia, and that activation could further enhance this process (Fig. 2C). Strikingly, activation in the presence of the Hv1 blocker depressed NOX2 expression levels well below control levels (Fig. 2C). Similar to the results obtained with HAPI cells, activation promoted a dramatic ~ 40-fold increase in NOS2 expression in primary microglial cells, a phenomenon that was attenuated by Hv1 inhibition (Fig. 2D). The changes in expression of these two proteins were paralleled by enhanced DhE staining in activated cells, with a concomitant decrease in staining in ClGBI-treated activated cells (Fig. 2E; quantified in Fig. 2F). NOX activity, NOS activity and phagocytosis followed a similar pattern to that observed in HAPI cells, with increased activity upon activation, as well as inhibition by Hv1 antagonism (Fig. 2G-I). Once again, activated cells retained phagocytotic capabilities near what is normally observed by non-activated microglia in the presence of ClGBI, (Fig. 2I). Activated microglia presented a similar increase in [Zn2+]e to that observed in HAPI cells, while of greater magnitude. Upon activation, [Zn2+]e concentration increased nearly three-fold, and this response was not attenuated by ClGBI treatment (Fig. 2J). Our findings suggest that ROS production and enhanced phagocytic activity are dependent on Hv1 activity and that its inhibition can have an important influence on their regulation. On the other hand, zinc release appears to be an activation parameter independent of Hv1 activity. Most importantly, the activation profile of HAPI cells appears to robustly reproduce what is normally observed in primary microglial cells making them a valuable tool in the study of brain inflammatory processes.
Hv1 block prevents proton extrusion and acidifies the intracellular pH of activated microglia. As Hv1 promotes proton extrusion in response to intracellular pH acidification, we next determined the impact of ClGBI on the intracellular (pHi) and extracellular (pHe) pH of HAPI cells, using BCFL-AM, an intracellular fluorescent indicator. First, we evaluated different concentrations of ClGBI (Fig. S4). Interestingly, we did not observe changes in pHi or pHe at any concentration of ClGBI tested (2.5–10 µM) on non-activated HAPI cells (Fig. S4A, B). However, following activation, a significant decrease in pHi, as well as an increase in pHe, was observed with 7.5–10 µM ClGBI (Fig. 3A, B; Fig. S4 C, D). These results indicate that activation leads to pronounced intracellular acidification and that Hv1 inhibition intensifies this effect. Moreover, this activation-induced intracellular acidification is associated with a decrease in proton extrusion, which can be measured by a change in pHe. As such, both pHi and pHe can be modulated by Hv1 inhibition in activated cells. These findings are consistent with the known function of the proton channel during microglial activation31,78 and are nicely recapitulated in our HAPI cell preparation. Finally, we confirmed that the activation and ClGBI-mediated changes in pHi and pHe observed in the cell line were also present in the primary microglial cells, which was indeed the case (Fig. 3C, D), further demonstrating the utility of HAPI cells as an excellent microglial cell line model.
NOX2 activity promotes intracellular acidification in activated HAPI cells. During microglial activation, ROS production by NOX2 results in a release of protons into the intracellular space. As such protons extrusion, particularly by Hv1, is necessary for maintaining electroneutrality. We investigated the impact of NOX2 activity on pHi using GSK279, a specific competitive inhibitor of NOX279. As previously noted, treatment with LPS/IFNγ led to a four-fold enhancement of NOX activity in HAPI cells. In contrast, activation in the continuous presence of GSK279 (1-100 µM) led to a concentration-dependent inhibition of NOX activity, reverting to near control activity at the highest concentration of the inhibitor (Fig. 4A). As NOX2 activity likely leads to the aforementioned intracellular acidification following activation, we confirmed that GSK279 similarly led to a concentration-dependent reversal of pHi changes observed in HAPI cells following treatment with LPS/IFNγ (Fig. 4B). Importantly, GSK279 could also effectively abrogate the enhanced acidification produced by Hv1 inhibition in activated cells (Fig. 4C). These results suggest not only that NOX2 is largely responsible for the intracellular acidification secondary to activation but that the functional coupling between Hv1 and NOX2 governs intracellular pH during the microglial inflammatory response. Moreover, the enhanced intracellular acidification as a result of Hv1 block in activated cells is the likely mechanism behind ClGBI-mediated inhibition of NOX2 activation80.
Pharmacological inhibition of Hv1 modifies the activation metabolic profile. As intracellular acidification is associated with intermediary metabolism changes81,82, we next measured critical parameters of metabolism under basal and activated conditions in our microglial cell line. First, we assessed intracellular lactate (lactatei) concentrations, since proinflammatory microglial activation is associated with increased anaerobic glycolysis82. Intracellular lactate increased almost four-fold during activation, compared to control, a phenomenon that was significantly attenuated by ClGBI treatment (Fig. 5A). Additionally, activation not only promoted lactate production but also its release into the extracellular space. After 24 hours of treatment with LPS/IFNγ, the concentration of extracellular lactate (lactatee) increased more than fifteen times baseline levels, which, again, was strongly inhibited by Hv1 block (Fig. 5B). These changes were also reflected by total lactate measurements (Fig. 5C). Subsequently, we investigated whether these metabolic changes would also be present in activated primary microglia cultures. The effects of activation and Hv1 blockade on the concentrations of lactatei, lactatee and total lactate seen in HAPI cells were again reproduced in this system (Fig. 5G-I). This rise in lactate production is indicative of a metabolic shift toward anaerobic glycolysis in activated cells which negatively impacts energy production through oxidative phosphorylation83.
An accurate indicative parameter of the energy status of cells is the oxidation state of NAD. A decrease in NAD+/NADH ratio is generally associated with oxidative stress while increases in ratios are an indicator of metabolic stress84. Activated HAPI cells showed a moderate decrease in NAD+ compared to non-activated cells, which could be completely prevented by Hv1 inhibition (Fig. 5D). Additionally, we found that activation dramatically decreased the concentration of NADH when compared to non-activated cells, an effect that was again mitigated by ClGBI (Fig. 5E). These changes in NAD+ and NADH led to a 4-fold increase in the NAD+/NADH ration upon activation, which was significantly attenuated by Hv1 antagonism (Fig. 5F). A nearly identical pattern of NAD+ and NADH changes were observed in our primary microglia cultures (Figs. 5G-L). These results suggest that activation promotes lactate production and NADH consumption, processes directly associated with increased anaerobic glycolysis. The fact that Hv1 antagonism effectively prevents these metabolic changes in both cellular models indicates that inhibition of this voltage-gated proton channel may promote the conservation of a more favorable energy status.
Hv1 inhibition promotes an anti-inflammatory activation phenotype. As the microglial activation phenotype can be strongly influenced by metabolic changes24,85, such as those promoted by Hv1 inhibition, we next characterized the secretome profile of both non-activated and activated HAPI cells and primary microglia in the presence and absence of ClGBI. To do so, we quantified the concentration of cytokines in the culture medium of cells in the aforementioned conditions. Interestingly, ClGBI alone moderately increased the production of IL-1β in non-activated HAPI cells, when compared to control. IL-1β levels were further increased in activated cells. However, in this case, Hv1 block did not modify the response (Fig. 6A). In contrast, ClGBI alone did not influence IL-1β production in non-activated primary microglia cultures, but it did significantly reverse the increased production of this cytokine in activated cultures (Fig. 7A). Both HAPI and primary microglial cell activation produced a robust increase in IL-6 production, with ClGBI co-treatment markedly inhibiting its production (Fig. 6B and 7B). A similar pattern was observed with TNFα production, and its inhibition by ClGBI, in both HAPI (Fig. 6C) and microglial cells (Fig. 7C). These results confirmed that upon activation both our models increase production of well-known pro-inflammatory cytokines, which are largely suppressed by Hv1 antagonism.
Given these findings and our data demonstrating that Hv1 hampers inflammatory activation (Figs. 1 and 2), we examined whether Hv1 could also promote an anti-inflammatory phenotype. We quantified three cytokines directly associated with anti-inflammatory responses in microglia, namely IL-4, IL-10, and TGF-β. HAPI and microglial cell activation was accompanied by a moderate decrease of IL-4, which was normalized to control levels by ClGBI in both cell types (Fig. 6D and 7D). In contrast to these observations, the patterns of IL-10 production differed between HAPI (Fig. 6E) and primary microglial (Fig. 7E) cells. In both cases, production of this neuroprotective cytokine increased upon activation. We observed a nearly 50% decrease in IL-10 levels in HAPI cells treated with ClGBI alone as well as a similar decrease in activated cells exposed to the Hv1 antagonist. In contrast, we did not observe a decrease in IL-10 in non-activated or activated primary microglial cells treated with ClGBI. TGF-β, a potent inducer of anti-inflammatory polarization86, decreased following activation in both cell types. For this cytokine, we did not observe a significant rescue in production by ClGBI in neither HAPI nor microglial cells, although there was a trend towards recovery in the cell line model (Fig. 6F and 7F). These results suggest a complex effect of Hv1 inhibition on anti-inflammatory cytokine production in the cell line and primary microglia cultures. However, the general pattern observed is one in which an anti-inflammatory phenotype is maintained in activated cells when Hv1 is antagonized, favoring an IL-4 (and perhaps TGF-β)-favored neuroprotective profile in HAPI cells, and an IL-4/IL-10 profile in the mouse microglial cultures.
Finally, we examined the morphological changes associated with HAPI and microglial cell activation by means of Iba-1 labeling (Fig. 6G and 7G). Control non-activated HAPI cells generally presented a compact, uniform, and rounded soma, with few projections. This morphology appeared unaffected by the presence of ClGBI alone. After treatment with LPS/IFNγ, we observed an increase in the size of the soma, acquiring an irregular appearance. When activation occurred during Hv1 inhibition, cells exhibited both types of morphology, although the vast majority resembled non-activated, inactivated cells. This morphology activation pattern suggests the acquisition of an alternative or mixed phenotype87 under these conditions (Fig. 6G), which is consistent with their secretome profile. Similarly, to the cell line, ClGBI treatment in non-activated microglial cells did not change their morphology. However, after 24 hours of LPS/IFNγ treatment, we observed growth in the size of the soma and a reduction in the number of processes, correlating with a classic proinflammatory activation profile87. Treatment of activated microglia with ClGBI again resulted in a morphological profile more akin to the non-activated state (Fig. 7G).
Hv1 inhibition during activation reduces inflammatory neurotoxicity. The results presented thus far shown that Hv1 inhibition decreases the microglial production of reactive species and proinflammatory mediators, while promoting the release of anti-inflammatory cytokines. To determine the functional implications of these changes to the activation profile of microglia, we examined their neurotoxicity in the presence and absence of ClGBI. Rat cortical neuron cultures were exposed to conditioned medium from non-activated and activated HAPI and microglial cells treated with or without ClGBI. Twenty-four hours after exposure, neuronal viability was quantified by LDH and MTS assays. Cortical cultures exposed to medium harvested from activated HAPI (Fig. 8A), and microglial (Fig. 9A) cultures resulted in a significant loss of neuronal viability as evidenced by increased LDH release (Figs. 8A and 9A). Critically, we found that Hv1 antagonism with ClGBI during activation prevented the neurotoxicity of the conditioned medium (Figs. 8A and 9A). These findings were confirmed in the MTS assay (Fig. 8B and 9B). These data strongly suggest that in the presence of an Hv1 blocker, the secretion of neurotoxic factors by activated microglia is significantly attenuated.
Having demonstrated that Hv1 inhibition limits neurotoxicity associated with the release of pro-inflammatory mediators by activated microglia, we turned our focus to identifying the diffusible molecules that contribute to the observed neurotoxic profile in our models. As oxidative stress is closely related to inflammatory neurotoxicity88, we first assessed whether antioxidant exposure could increase neuronal viability using the MTS assay. Indeed, EUK134 (100 µM), a synthetic superoxide dismutase/catalase mimetic89, provided neuroprotection against toxicity associated with HAPI and microglial conditioned media exposure (Figs. 8C and 9C). Given that extracellular zinc is neurotoxic to our cortical cultures90 and that we demonstrate that activation in both HAPI cells and microglia causes an increase in extracellular Zn2+, we evaluated whether the extracellular zinc chelator ZX191 (5 µM) could provide some degree of neuroprotection to conditioned media exposure. Indeed, ZX1 treatment significantly increased neuronal viability in cortical cultures exposed to conditioned media from activated microglia (Fig. 8D and 9D). Finally, we used neutralizing antibodies to block the activity of IL-692 and TNFα93, neurotoxic cytokines released during activation in both our models. Surprisingly, neutralization of IL-6 (5 µg/ml) was not sufficient to reverse the neurotoxicity induced by the conditioned media from either activated HAPI (Fig. 8E) or microglial (Fig. 9E) cultures. In contrast, we found that a TNFα neutralizing antibody (5 µg/ml) significantly increases the viability in both cases (Fig. 8F and 9F). Of note, none of the neuroprotective actions of EUK134, ZX1 or TNFα neutralization were additive with the previously noted protective actions of ClGBI (Fig. 8C-F and 9C-F), suggesting that Hv1 block protects neurons, in part, by preventing sufficient pro-inflammatory cytokine production by activated microglial cells.
Hv1 inhibition promotes neuroprotection from microglia during excitotoxic damage. In the next set of experiments, we evaluated: i) whether co-cultures of microglia and cortical cells resulted in neurotoxicity following activation, ii) whether inducing excitotoxicity in neurons could itself result in microglial activation and further enhance neurotoxicity, and iii) whether Hv1 inhibition could restore neuronal viability following microglial activation and/or excitotoxic insults. In both our HAPI/cortical and microglia/cortical co-cultures we observed a 2-fold increase in LDH release following activation, indicating, once again, that microglial activation results in neurotoxicity (Fig. 10A, D). Importantly, in both cases, the presence of ClGBI abrogated the neurotoxicity (Figs. 10A, D). Since an elevated LDH could be reflective of the presence of both injured microglia (induced by dying neurons) and neurons, we took advantage of a luciferase viability assay in previously transfected neurons, which exclusively reflects neuronal viability in co-culture or mixed culture conditions60,63. Using this assay, we confirmed both a decrease in neuronal viability following microglial activation, as well as neuroprotection by Hv1 block (Fig. 10B, E).
We next explored whether microglia were protective against a canonical excitotoxic insult and whether Hv1 antagonism modified this response. We first exposed cortical cultures alone to the glutamate uptake inhibitor TBOA (75 µM), noting, as expected62,63, a substantial decrease in neuronal viability, as measured by the luciferase assay (Fig. 10C). In the absence of HAPI cells, the observed excitotoxicity was not influenced by the presence of the Hv1 blocker. In contrast, when neurons were in co-culture with non-activated HAPI or microglial cells, TBOA toxicity could be partially ameliorated by ClGBI (Figs. 10C and 10F), suggesting that neuronal excitotoxic injury, in and of itself, is sufficient to trigger microglial cell activation, as observed in other systems37. Moreover, we observed increased TBOA toxicity when neurons were exposed to previously activated HAPI or microglial cells, but only when ClGBI was absent during activation. However, even under these circumstances ClGBI was protective if added during the TBOA exposure (Figs. 10C and 10F, right two sets of bar graphs). This critically important result indicates that Hv1 block not only prevents activated microglia neurotoxicity but also, by reprogramming these cells, Hv1 antagonism can generate an overall neuroprotective environment.