Inhibition of the mitochondrial pyruvate carrier simultaneously mitigates hyperinflammation and hyperglycemia in COVID-19

The relationship between diabetes and COVID-19 is bi-directional: while individuals with diabetes and high blood glucose (hyperglycemia) are predisposed to severe COVID-19, SARS-CoV-2 infection can also cause hyperglycemia and exacerbate underlying metabolic syndrome. Therefore, interventions capable of breaking the network of SARS-CoV-2 infection, hyperglycemia, and hyper-inflammation, all factors that drive COVID-19 pathophysiology, are urgently needed. Here, we show that genetic ablation or pharmacological inhibition of mitochondrial pyruvate carrier (MPC) attenuates severe disease following influenza or SARS-CoV-2 pneumonia. MPC inhibition using a second-generation insulin sensitizer, MSDC-0602 K (MSDC), dampened pulmonary inflammation and promoted lung recovery, while concurrently reducing blood glucose levels and hyperlipidemia following viral pneumonia in obese mice. Mechanistically, MPC inhibition enhanced mitochondrial fitness and destabilized HIF-1α, leading to dampened virus-induced inflammatory responses in both murine and human lung macrophages. We further showed that MSDC enhanced responses to nirmatrelvir (the antiviral component of Paxlovid) to provide high levels of protection against severe host disease development following SARS-CoV-2 infection and suppressed cellular inflammation in human COVID-19 lung autopsies, demonstrating its translational potential for treating severe COVID-19. Collectively, we uncover a metabolic pathway that simultaneously modulates pulmonary inflammation, tissue recovery, and host metabolic health, presenting a synergistic therapeutic strategy to treat severe COVID-19, particularly in patients with underlying metabolic disease.


INTRODUCTION
Despite the development of many vaccines and highly successful vaccination campaigns, respiratory viruses such as influenza and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) continue to present a substantial public health burden (1). The ever-present threat of respiratory viral infections is a result of nonsterilizing immunity induced by vaccination (2,3) as well as the constant emergence of new viral variants (1,4). Respiratory viral infections are particularly dangerous to individuals with underlying metabolic syndrome, most notably insulin resistance associated with obesity and diabetes. Hyperglycemia (high blood sugar levels) is common in hospitalized patients with coronavirus disease 2019 (COVID- 19) and is strongly associated with worse outcomes after SARS-CoV-2 infection (5,6). Conversely, SARS-CoV-2 promotes insulin resistance and beta cell dysfunction, inducing hyperglycemia in a significant proportion of patients without history of metabolic disease (7)(8)(9)(10).
Immunological features of COVID-19 are linked to exuberant inflammation in the respiratory tract, driven by profound immune dysregulation (11)(12)(13)(14). Thus, it is expected that therapeutics targeting hyperinflammation have shown potential in ameliorating severe COVID-19 (15). The anti-inflammatory steroidal agent dexamethasone remains the most effective treatment for patients with severe COVID-19 with hypoxemia (16), with the restrictive therapeutic window of antiviral drugs in the early stage of infection (17,18). However, the efficacy of dexamethasone and other anti-inflammatory treatments [such as tocilizumab; anti-interleukin-6 (IL-6)] is still limited (18). Furthermore, dexamethasone treatment often results in hyperglycemia, complicating its use in patients with underlying metabolic disease-a population exhibiting increased risk of severe COVID-19 complications (19,20). Thus, there is a clinical need to develop interventions capable of simultaneously mitigating hyperinflammation and hyperglycemia to address two distinct processes dysregulated in COVID-19 patients with metabolic syndrome.
Pyruvate is the end-product of glycolysis that can be either reduced to lactate in the cytosol or used as a fuel for oxidative metabolism in the mitochondria. Glycolysis is well established to have a critical role in macrophage activation and inflammatory responses (21). However, the contribution of pyruvate oxidation in the tricarboxylic acid (TCA) cycle to macrophage function and inflammation is not well understood. Pyruvate is imported into the mitochondria via the mitochondrial pyruvate carrier (MPC), a protein known to maintain TCA cycle flux (22). Pyruvate fuels the TCA cycle to support anabolic metabolism and energy production. Pyruvate conversion to acetyl-coenzyme A (CoA) supports acetyl-CoA-dependent reactions, including fatty acid oxidation and protein acetylation. MPC-mediated pyruvate metabolism has emerged as an important player in the physiological and pathophysiological processes underlying type 2 diabetes (23,24). In this report, we found that inhibition of MPC activity dampened exaggerated pulmonary inflammation and concurrently promoted host metabolic health, thereby diminishing host morbidity and mortality after influenza or SARS-CoV-2 infection in obese mice. Furthermore, MPC inhibition decreased cellular inflammation in human COVID-19 lung autopsy tissue. MPC inhibition synergized with the antiviral component of Paxlovid, nirmatrelvir, to markedly lower host mortality and lung injury in SARS-CoV-2-infected obese mice. Our results suggest that the MPC inhibitor MSDC-0602K (MSDC), a second-generation insulin sensitizer with an outstanding safety profile, has potential as a therapeutic for treating severe COVID-19, particularly in patients with underlying metabolic disease.

Genetic ablation or pharmacological inhibition of MPC function improves disease outcome after influenza or SARS-CoV-2 infection
Excessive lung macrophage activation can initiate and contribute to unrestrained inflammation through release of various proinflammatory mediators that lead to the recruitment of inflammatory immune cells to the lung after respiratory viral infection such as SARS-CoV-2 (12,25,26). Glycolysis is known to support macrophage activation and inflammation, but the importance of glucose oxidation via pyruvate translocation into the mitochondria to macrophage inflammatory responses in vivo is understudied (Fig. 1A). We found that mice bearing myeloid-specific deletion of the mitochondrial pyruvate transporter 2 (MPC2 ΔLyz2 ), which forms the heteromeric MPC complex with MPC1, had decreased host morbidity and mortality after sublethal or lethal doses of H1N1 influenza A virus (IAV) infection (Fig. 1, B and C). MPC2 ΔLyz2 mouse lungs showed comparable viral titers compared to wild-type (WT) littermates but exhibited greatly diminished inflammatory gene expression in the lungs and decreased proinflammatory cytokine levels in bronchoalveolar lavage (BAL) fluid at 4 days post infection (d.p.i.) ( Fig. 1D and fig. S1, A to C). MPC2 ΔLyz2 mice also showed lowered accumulation of inflammatory monocytes (Ly6C + ) and neutrophils in the lung at 4 d.p.i., which are major contributors of pulmonary immune pathology after respiratory viral infection (fig. S1, D and E) (27)(28)(29). In addition, lung histologic tissue inflammation, disrupted alveolar areas, and total BAL protein content were all reduced in MPC2 Δlyz2 mice at 14 d.p.i. (fig. S1, F and G), suggesting that MPC depletion mitigates lung injury in response to IAV infection. Furthermore, alveolar type II cell (ATII)-specific genes (such as Abca3 and Sftpb) were enhanced in MPC2 Δlyz2 mice at 14 d.p.i.
( fig. S1H), suggesting that MPC depletion promotes tissue recovery. These data together suggest that genetic ablation of MPC function in myeloid cells reduces the severity of diseases after IAV pneumonia and that glucose-fueled mitochondrial metabolism likely supports pulmonary inflammation.
To determine whether pharmacological targeting of MPC could lead to attenuated viral pathology in the respiratory tract, we infected WT mice with IAV and then treated the mice with MSDC, which is an MPC inhibitor and also a second-generation insulin-sensitizing thiazolidinedione (TZD) that has a superior safety profile compared with first-generation TZDs (Fig. 1E) (24). Intraperitoneal administration of MSDC from 1 day after IAV infection resulted in decreased host morbidity and mortality without affecting the kinetics of viral replication (Fig. 1, F and G, and fig. S1, I and J). MSDC-treated mice exhibited decreased BAL cytokine levels, diminished monocyte and neutrophil infiltration (fig. S1, K and L), and reduced lung inflammatory responses at 4 d.p.i. (Fig. 1H and  fig. S2, A to E). MSDC treatment also enhanced inflammation resolution, decreased lung damage, and promoted ATII cell regeneration at 14 d.p.i. (Fig. 1I and fig. S2, F to H). We did not observe a significant reduction in the magnitude of adaptive T cell response in MSDC-treated animals at 6 d.p.i., indicating that its mode of action is primarily to dampen exuberant innate inflammation without interfering in antiviral adaptive immunity necessary for preventing viral dissemination and persistence (fig. S2, I to K).
Similar to IAV infection, SARS-CoV-2 infection results in profound inflammation in the lower respiratory tract (11). Because MSDC reduced pulmonary inflammation in IAV-infected hosts, we tested whether MSDC could dampen SARS-CoV-2-induced lung inflammation and host diseases. We challenged WT mice with the mouse-adapted SARS-CoV-2 strain MA10, which induces acute lung damage and pneumonia in mice similar to human COVID-19 (30), and then treated mice with vehicle or MSDC (Fig. 1J). MSDC treatment enhanced host weight recovery (Fig. 1K) and diminished pathological changes in the lungs at 8 d.p.i. (Fig. 1L and fig. S2L). This was accompanied by reduced accumulation of inflammatory monocytes and neutrophils as well as proinflammatory cytokine levels in the BAL, without altering viral titers (Fig. 1, M and N, and fig. S2, M and N). ATII cell loss is a prominent feature of COVID-19 (31), and ATII regeneration is vital for lung recovery from viral pneumonia (32). We found that MSDC treatment promoted ATII cell recovery in SARS-CoV-2-infected lungs at 8 d.p.i. (fig. S2, O and P). Collectively, these data suggest that MSDC ameliorates pulmonary inflammation and promotes tissue recovery after SARS-CoV-2 infection.
Next, we determined whether MSDC delivered at 1 or 2 days after SARS-CoV-2 infection would still be effective. In WT mice, MSDC treatment from 1 d.p.i. decreased morbidity, and the efficacy diminished with treatment starting from 2 d.p.i. (fig. S2, Q and R). These data suggest that the drug would be more efficacious if delivered early after viral infection, although the time "day 1" here would likely be analogous to several days after infection in humans because of the delivery of a large amount of virus directly into the lung in this model. To determine whether myeloid deficiency of MPC ameliorates host morbidity, we infected MPC2 Δlyz2 mice with SARS-CoV-2 MA10 and determined host weight loss after infection. Consistent with MSDC-treated mice, myeloid MPC-deficient mice showed diminished host weight loss (Fig. 1O). SARS-CoV-2 infection shows an age-dependent increase in disease severity (33,34). Parallel results were found in mice that became progressively more susceptible to mouse-adapted SARS-CoV-2 infection correlating with aging (30,35). We evaluated the efficacy of MSDC in aged (13-to 14-month-old) female C57BL/6 mice regarding SARS-CoV-2 infection ( fig. S3A). In aged mice infected with SARS-CoV-2 MA10, MSDC treatment significantly reduced host mortality, with concomitant effects on weight loss ( fig. S3, B and C). In addition, MSDC diminished pathological changes in the lungs of surviving mice at 10 d.p.i. (fig. S3D). These results showed that MSDC ameliorated SARS-CoV-2induced disease in older mice.

Murine and human lung macrophages are prominent targets of MSDC
Single-cell RNA sequencing (scRNA-seq) analysis revealed that lungs isolated from MSDC-treated mice exhibited relatively higher proportions of lung structural and resident immune cells, including alveolar epithelial cells, endothelial cells, and alveolar macrophages (AMs), but diminished infiltrating immune cells such as neutrophils, monocytes, proliferating T cells, and plasmacytoid dendritic cells after IAV infection at 4 d.p.i. (Fig. 2, A and B, and fig. S4, A and B). Consistently, MSDC-treated lungs showed enrichment of gene sets associated with wound healing, epithelial regeneration, and fatty acid oxidation while demonstrating downregulation of inflammation-associated gene sets ( Fig. 2C and fig.  S4, C to E). Monocyte-and macrophage-mediated inflammatory responses are considered a major driver of respiratory viral pathogenesis (12,26). We delineated eight subsets of monocyte and macrophage populations in the infected lungs, and MSDC treatment diminished the presence of Ly6C + inflammatory monocytes, monocyte-derived macrophages (MdMs), and inflammatory AM subsets ( Fig. 2D and fig. S4F). MSDC markedly inhibited the expression of a large number of inflammatory associated genes/sets within the total AM population but had less prominent inhibitory effects on the inflammatory responses of monocytes or MdMs ( Fig. 2E and fig. S4, G to J). Reverse transcription polymerase chain reaction (RT-PCR) analysis of inflammatory gene expression by sorted AMs, CD11b + macrophages, and monocytes further confirmed that MSDC exhibited marked anti-inflammatory effects on AMs but with more modest effects on CD11b + macrophages and monocytes at 4 d.p.i. (Fig. 2F and fig. S5A). MSDC also had modest effects in suppressing bone marrow-derived macrophage (BMDM) inflammation after polyinosine-polycytidylic acid (Poly IC) stimulation in vitro compared with those of AMs ( fig. S5B). RNA-seq analysis found that MSDC promoted fatty acid degradation, oxidative phosphorylation (OXPHOS), and peroxisome proliferator-activated receptor (PPAR) signaling and inhibited genes associated with interferon signaling, cytokine responses, and monocyte chemotaxis in AMs with or without Poly IC treatment ( Fig. 2G  and fig. S5, C to E), indicating that MSDC inhibition of MPC function profoundly altered the balance of antiviral versus proinflammatory AM status. MSDC also exhibited marked antiinflammatory effects on human AMs but showed relatively moderate immunoinhibitory effects on blood monocytes or MdMs (Fig. 2H). MSDC inhibitory effects on AM inflammatory responses were abrogated in MPC deficiency, confirming that MSDC inhibits AM inflammation via MPCs ( fig. S5F). Although not responding to MSDC treatment, MPC-deficient AMs showed slightly higher baseline expression of Il6 compared with MSDC-treated AMs ( fig.  S5F), indicating a potential off-target effect of MSDC in this setting. Together, these data indicate that lung macrophages, but not circulating monocytes, are the prominent targets of MSDC during respiratory viral infection, consistent with a recent finding that MPC function is dispensable for BMDM inflammatory responses after lipopolysaccharide stimulation (36).

MPC inhibition by MSDC selectively reduces HIF-1α levels in lung macrophages
To explore the molecular mechanisms by which MPC inhibition by MSDC suppressed lung macrophage inflammatory responses, we performed Western blot analysis probing hypoxia-inducible factor-1α (HIF-1α), nuclear factor κB (NF-κB), and signal transducer and activator of transcription 1 (STAT1) activation, which are known to promote macrophage inflammatory responses (37,38), after Poly IC treatment. Poly IC stimulation led to the accumulation of HIF-1α protein and higher NF-κB p65 and STAT1 phosphorylation in AMs, and MSDC treatment inhibited HIF-1α levels but not NF-κB and STAT1 activation (Fig. 3, A and B). In contrast, MSDC did not interfere with HIF-1α expression, p65 phosphorylation, or STAT1 activation in BMDM, consistent with its moderate effects on BMDMs (Fig. 3C). MSDC did not affect Hif1a mRNA levels, and MPC2 deficiency in AMs recapitulated the MSDC effects on HIF-1α protein levels ( fig. S6, A and B). In addition, MSDC treatment decreased HIF-1α levels in AMs at day 4 after IAV infection in vivo (Fig. 3D). Consistent with the diminished HIF-1α protein levels after MSDC treatment, MSDC treatment in Poly IC-stimulated AMs showed diminished enrichment of hypoxia-associated genes (Fig. 3E). MSDC suppressed HIF-1α accumulation in human primary AMs after Poly IC treatment (Fig. 3F). Furthermore, in vitro treatment with molidustat and roxadustat, two HIF-1α stabilizers that promote HIF-1α accumulation in AMs, abrogated the suppressive effects of MSDC on AM inflammatory responses ( Fig. 3G and fig. S6C).
UK5099 is a well-known MPC small-molecule inhibitor (39), although a recent study has also identified off-target effects (36). We examined AM inflammation and HIF-1α expression in the presence of UK5099. MSDC and UK5099 both reduced Poly IC-induced inflammatory gene expression and HIF-1α levels in AMs in vitro (fig. S6, D and E). To this end, our data from MSDC-treated MPC2-deficient AMs confirmed that the effects of MSDC on macrophage HIF-1α expression and that inflammatory responses are largely dependent on MPC function ( fig. S6B). Together, these data suggest that MSDC inhibits MPC function and promotes HIF-1α instability, thereby inhibiting lung macrophage inflammatory responses after viral stimuli.

MPC inhibition promotes mitochondrial fitness and diminishes HIF-1α-stabilizing metabolites
Viral stimuli have been demonstrated to inhibit macrophage mitochondrial metabolism, facilitating macrophage-mediated inflammatory responses (34,40). MSDC treatment enhanced maximal oxygen consumption rate (OCR) and respiratory reserve in AMs, but not BMDMs, after stimulation with Poly IC (Fig. 4A and fig.  S7A). Consistent with these observations, fewer depolarized mitochondria were seen in MSDC-treated AMs but not BMDM populations ( Fig. 4B and fig. S7B). In addition, MSDC treatment increased respiratory reserve and mitochondrial fitness in AMs at 4 days after Pyruvate is oxidized in the TCA cycle after its translocation into mitochondria (22). We therefore measured TCA metabolites after in vitro Poly IC stimulation in the presence or absence of MSDC. Poly IC greatly promoted the accumulation of succinate in AMs and BMDMs, which was diminished in the presence of MSDC in AMs but not BMDMs ( Fig. 4E and fig. S7G). High succinate/α-ketoglutarate (α-KG) ratio is an indication of reduced complex II activity of the electron transport train. MSDC treatment also reduced the succinate/α-KG ratio in AMs but not BMDMs, consistent with improved mitochondrial respiration ( Fig. 4F and fig. S7H). HIF-1α protein can be stabilized by both succinate and acetylation (42,43). We measured acetyl-CoA levels and found that MSDC reduced acetyl-CoA accumulation in AMs, but not BMDMs (Fig. 4G). Reduced acetyl-CoA levels could affect gene expression by reducing histone acetylation; however, MSDC treatment did not markedly suppress the acetylation of total H3 or H3K27 ( fig. S7I). Rather, diminished acetyl-CoA in AMs was associated with decreased HIF-1α acetylation (Fig. 4H) (42,43). We next sought to determine whether exogenous succinate or acetyl-CoA could promote AM HIF-1α levels and promote their inflammatory responses after MSDC treatment. We treated Poly IC-and/or MSDC-stimulated AMs with sodium acetate (SDA) or dimethyl succinate (DMS) to boost intracellular acetyl-CoA and succinate, respectively. Exogenous SDA or DMS treatment promoted HIF-1α accumulation in AMs and enhanced Tnf and Ccl2 expression even in the presence of MSDC (Fig. 4, I and J), indicating that diminished succinate and/or acetyl-CoA levels likely contribute to decreased HIF-1α after MSDC treatment in AMs.
Previous studies in T cells and other cell types have found that inhibition of pyruvate translocation led to increased glutamine and lipid incorporation into the mitochondria (44). Next, we sought to investigate whether increased mitochondrial respiration in AMs after disruption of pyruvate metabolism by MSDC could be due to the increased utilization of fatty acid or glutamine oxidation. We performed Seahorse assays in AMs in the presence of glutaminase inhibitor BPTES and/or carnitine palmitoyltransferase 1 inhibitor etomoxir in the context of MPC inhibition in vitro. MSDC treatment increased OXPHOS and mitochondrial fitness compared with vehicle treatment; however, single and particularly the combined treatment of BPTES and/or etomoxir inhibited OXPHOS and mitochondrial fitness compared with MSDC alone-treated cells ( fig. S8). These data suggest that the increased mitochondrial fitness in AMs after MPC inhibition is likely due to the compensatory effects of glutamate and/or fatty acid oxidation.
Previously, we have shown that genetic HIF-1α deficiency in AMs or inhibition of HIF-1α stability by LW6 administration in vivo diminished lung inflammation after IAV infection (38). Similarly, we found that LW6 treatment decreased host morbidity and inflammation after SARS-CoV-2 infection ( fig. S9, A to F), suggesting that the increased HIF-1α expression is a primary driver of pulmonary inflammatory responses. HIF-1α is known to promote IL-1β production in macrophages, and IL-1β release is a major contributor of pulmonary inflammation during COVID-19 (42,45,46). Next, we sought to determine whether inhibition of HIF-1α-dependent IL-1β contributed to decreased macrophage inflammation by MSDC. We treated WT or Il1b-deficient AMs with MSDC or LW6 in the presence or absence of Poly IC. MSDC or LW6 diminished proinflammatory cytokine expression in both WT and Il1b-deficient AMs after Poly IC stimulation ( fig. S9G), suggesting that the decreased inflammatory capacity of AMs by MSDC or HIF-1α inhibitor treatment is independent of IL-1β in vitro. Consistently, blockade of IL-1β did not significantly decrease host morbidity, as seen in MSDC-or LW6-treated mice in vivo ( fig. S9H). Therefore, we conclude that the diminished macrophage inflammation after MSDC treatment is unlikely to be due to diminished IL-1β production after treatment. Together, these data suggest that MPC inhibition by MSDC improves mitochondrial metabolism and diminishes the accumulation of metabolites capable of stabilizing HIF-1α, thereby suppressing the inflammatory responses of lung macrophages after respiratory viral infection.

MSDC promotes metabolic health and concurrently suppresses pulmonary hyperinflammation
Respiratory virus infections are particularly dangerous to people who have underlying metabolic syndromes, most notably insulin resistance with obesity and diabetes (5,47). This paradigm is observed in models where, compared with lean mice, obese mice also showed increased morbidity and mortality after IAV or SARS-CoV-2 infection (48)(49)(50). Because MSDC has been found safe and effective in lowering glycemia and liver steatosis (24) and in dampening IAVinduced inflammation in lean host (Fig. 1), we tested the therapeutic efficacy of MSDC in ameliorating IAV pneumonia in obese mouse models (Fig. 5A). High-fat diet-induced obese (DIO) mice showed higher levels of blood glucose and total cholesterol at 5 d.p.i. compared with lean mice, whereas MSDC administration improved glucose tolerance and lowered total cholesterol in the blood compared with vehicle-treated DIO mice (Fig. 5, B and C). Furthermore, MSDC-treated mice had decreased cytokine levels in the BAL and a marked reduction in expression of multiple proinflammatory genes in the lung ( Fig. 5D and fig. S10, A  Emerging evidence has suggested that obesity predisposes hosts to severe COVID-19 after SARS-CoV-2 infection (5,6,47). Our data also indicated that DIO mice showed increased host mortality compared with lean mice after SARS-CoV-2 infection ( fig. S12A). To this end, we examined the therapeutic potential of MSDC in mitigating severe diseases after SARS-CoV-2 MA10 infection in obese mice (Fig. 5F). MSDC-treated DIO mice had lower levels of proinflammatory cytokines, which was accompanied by decreased numbers of inflammatory monocytes and neutrophils in the BAL at 5 d.p.i. (Fig. 5G and fig. S12, B and C), suggesting that MPC inhibition dampened SARS-CoV-2-induced pulmonary inflammation. MSDC administration also decreased glucose and cholesterol levels in the blood at 5 d.p.i. (Fig. 5H), indicating that MSDC ameliorated metabolic conditions in obese mice after SARS-CoV-2 infection. In addition, MSDC treatment lowered lung inflammatory cytokine expression and BAL protein levels and reduced disrupted alveolar areas by lung histopathological analysis without affecting viral gene expression (Fig. 5I and fig. S12, D and E). MSDCtreated lungs showed increased ATII gene expression and elevated levels of ATII cell presence at 5 d.p.i. (Fig. 5J and fig. S12F), which suggest that MSDC can potently enhance lung recovery and regeneration after SARS-CoV-2 infection in obese hosts. Consequently, SARS-CoV-2-induced lethality was partially abrogated after MSDC treatment at 6 hours or 1 d.p.i., whereas most of the vehicle-treated DIO mice had succumbed to SARS-CoV-2 infection (Fig. 5K and fig. S12G). These data suggest that MPC inhibition by MSDC diminishes pulmonary hyperinflammation while concurrently promoting metabolic health after respiratory viral pneumonia in hosts with underlying metabolic conditions.

MSDC diminishes cellular inflammation in COVID-19 lung autopsies and enhances response to antiviral therapy
We next sought to further explore the translational potential of MSDC in treating COVID-19, particularly in patients with metabolic conditions. Infection of macrophages by SARS-CoV-2 has emerged as an important contributor to COVID-19-associated inflammation (45,51). We thus examined whether MSDC could diminish human lung macrophage inflammatory responses after SARS-CoV-2 infection. SARS-CoV-2 infection caused markedly elevated inflammatory gene expression in AMs isolated from two of three healthy donors, but MSDC treatment markedly inhibited a large number of inflammatory genes up-regulated after SARS-CoV-2 infection in AMs (Fig. 6A and fig. S13, A and B). We next examined whether MSDC could dampen lung inflammatory responses in patients with COVID-19 to determine its potential as a treatment for severe COVID-19. To this end, we incubated total lung cells from seven deceased COVID-19 patients with or without MSDC and determined TNF, CCL2, CCL4, and CXCL10 expression in the lung (Fig. 6B and fig. S13C). MSDC treatment inhibited the expression of these inflammatory genes in cells from the lungs of patients with COVID-19 ( Fig. 6B and fig. S13C), demonstrating the potential of MSDC in treating exuberant pulmonary inflammation in severe COVID-19.
Last, because many patients at risk for severe disease from COVID-19 are now likely to be treated with antiviral drugs such as Paxlovid, we examined whether MSDC could work with antiviral therapies to provide an added level of protection against severe COVID-19 in this setting. We treated SARS-CoV-2 MA10-infected DIO mice with nirmatrelvir (the antiviral component of Paxlovid) at 6 hours after infection in the absence or presence of MSDC (Fig. 6C). MSDC plus nirmatrelvir treatment protected the majority of mice from death after SARS-CoV-2 infection in obese mice, whereas nirmatrelvir alone appeared less efficacious (Fig. 6C). Furthermore, lungs from mice treated with MSDC plus nirmatrelvir showed less tissue inflammation and alveolar disruption compared with mice treated with nirmatrelvir alone (Fig. 6D), indicating that MSDC in combination with antiviral therapy can mitigate severe COVID-19 in at-risk hosts. Together, these data support a potential role for MSDC in the treatment of COVID-19 in patients with metabolic conditions, particularly when combined with antiviral therapy such as nirmatrelvir.

DISCUSSION
Lung macrophage populations are heterogeneous immune sentinel cells, critical for antiviral innate immunity and tissue recovery after respiratory viral infections (52). Conversely, macrophage-derived inflammatory and/or injurious mediators also contribute to excessive pulmonary inflammation and collateral tissue injury after viral infections (12,38,53). The aberrant activation of monocytes or resident macrophages is considered a key driver of virus-induced inflammation in severe COVID-19 (12, 51, 54). However, specific Quantification of proSP-C + cell number was performed using at least 10 random fields (10×) of alveolar space per mouse lung. (K) Host mortality was monitored. Representative (I and J) or pooled data (B, D, E, and K) from at least two independent experiments. Data are presented as means ± SEM. *P < 0.05; **P < 0.01. The P value was determined by log-rank test (E and K), one-way ANOVA (B), or two-tailed Student's t test (C, D, and G to J). pathways and/or mediators that can be targeted to dampen exuberant macrophage-mediated lung inflammation without compromising beneficial antiviral immunity remain largely unknown. Here, we showed that mitochondrial pyruvate translocation is selectively required for detrimental lung macrophage-mediated inflammatory responses after viral infections, including SARS-CoV-2. Moreover, this pathway can be safely targeted to improve outcomes after severe viral pneumonia using the second-generation TZD MSDC ( fig. S14).
AMs are among the first immune cells responding to viral infections. The inflammatory mediators produced by AMs not only directly contribute to pulmonary inflammation but also could indirectly augment the overall levels of inflammation by recruiting inflammatory immune cells such as monocytes and neutrophils after infection (25,38,55). Thus, the inhibition of AM inflammatory responses by MSDC has the potential to directly and/or indirectly dampen the pathogenic inflammation after respiratory viral infection. In addition, we did observe modest inhibitory effects of MSDC on MdM and monocyte inflammatory responses in vivo. Thus, MSDC may function to inhibit the inflammatory responses of both resident and recruited macrophages to mitigate severe disease development after viral infection.
Whereas glycolysis is well established to be involved in inflammatory responses for both monocytes and macrophages (56), MPCdependent pyruvate oxidation appears to be preferentially required for pulmonary macrophage-mediated inflammation by stabilizing HIF-1α. This distinct feature of MPC-dependent inflammatory activity in lung macrophages may be particularly meaningful and exploited to dampen pulmonary inflammation. Unlike corticosteroid treatment such as dexamethasone that induces a systemic anti-inflammatory state, often resulting in secondary infections and complications (19,57), the immunosuppressive effect of the MPC inhibitor may be limited to the respiratory tract. Inhibition of pyruvate flux into mitochondria can lead to compensatory metabolic reactions, including increased fatty acid and/or glutamine oxidation in T and other cell types (44), which is concordant with our transcriptomic analysis and increased mitochondrial respiration. Fatty acid and glutamine oxidative metabolism is known to promote pro- recovery M2-like macrophages (58), consistent with the observation that MSDC treatment enhanced lung recovery and regeneration after viral clearance. Thus, MSDC may also serve as a pro-reparative therapeutic in the clinic by augmenting lung tissue recovery (such as ATII cell replenishment) after COVID-19 lung injury.
Growing evidence indicates that SARS-CoV-2 induces mitochondrial dysfunction in immune cells. Acute SARS-CoV-2 infection resulted in rapid mitochondrial dysfunction in both CD4 and CD8 T cells, which compromised T cell functionality, contributing to suppressed T cell immune responses to viral infection (59). Patients with SARS-CoV-2 infection displayed depolarized mitochondria and abnormal mitochondrial ultrastructure in monocytes, which was correlated with enhanced inflammatory responses (60). Recently, targeted transcriptome analysis also revealed impairment of mitochondrial OXPHOS and antioxidant gene expression in autopsy samples, which was associated with enhanced HIF-1α stabilization (61,62). Thus, means that can promote mitochondrial metabolic fitness may be promising for developing a novel therapeutic avenue for COVID-19. In line with this notion, our study also showed compromised mitochondrial respiration and increased HIF-1α expression in AMs after viral infection in vivo (40), and the inhibition of pyruvate metabolism by MSDC enhanced mitochondrial OXPHOS and fitness, which was associated with the reduction of proinflammatory cytokines.
Obesity and/or diabetes greatly increases the risk of severe disease after IAV or SARS-CoV-2 infections (5,63). Correction of insulin resistance and hyper-glycemia using insulin sensitizer drugs has been suggested for the management of diabetic patients with COVID-19 (5). Metformin, the most prescribed antidiabetic drug, has been suggested as a repurposed drug for COVID-19 because of its anti-inflammatory properties (64,65). Nevertheless, metformin failed to provide significant clinical benefits in randomized placebocontrolled clinical trials (66,67), and the efficacy of other antidiabetic drugs such as glucagon-like peptide 1 receptor agonists (GLP1-RAs) also remains controversial (68). Therefore, interventions capable of circumventing the deadly cycle of viral infection, hyperglycemia, and hyperinflammation are critical for improving treatment of severe COVID-19 patients with a history of metabolic disease. First-generation TZDs, including pioglitazone and rosiglitazone, are effective in treating type 2 diabetes (69) but induce considerable side effects, resulting in their restriction and removal from the market in many countries. MSDC is a second-generation insulin sensitizer, maintaining all the pharmacological benefits of first-generation TZDs without the potential for edema and exhibiting an outstanding safety profile as per a recent multicenter, doubleblinded, randomized controlled trial (24). Our data also show that MSDC treatment is effective when combined with a current standard-of-care antiviral therapy.
We have shown that MSDC can simultaneously dampen hyperglycemia and hyperinflammation in obese hosts. Hyperglycemia per se predisposes the host to more severe disease development after viral infection, including SARS-CoV-2. Therefore, we have not assessed the relative contribution of hyperglycemia versus macrophage inflammatory activities in driving the severity of respiratory viral infection during obesity. Future experiments using MPC myeloid conditional deficient mice in the obesity setting would be ideal to study this question. In addition, our data showed that inhibition of glutaminase and/or carnitine palmitoyltransferase 1 disrupted elevated mitochondrial metabolism in response to MSDC treatment, suggesting that the increased fitness and respiration of mitochondria after MPC inhibition is likely due to the increased glutamine and/or fatty acid oxidation. However, such a model cannot be definitively established without analyzing the metabolic flux using specific isotope tracers. Unfortunately, such tracing experiments remain impractical because they require a large number of primary tissue macrophages. Nevertheless, our data have uncovered a metabolic pathway that concurrently modulates macrophage inflammation, lung recovery, and host metabolic health and suggest a potentially viable therapeutic agent that may be combined with existing antiviral agents to treat severe COVID-19 in patients with underlying metabolic disease.

Study design
The aim of this study was to determine the therapeutic potential of MPC inhibitor MSDC in treating severe viral infection, including SARS-CoV-2 infection, in normal and obese hosts. We examined the efficacy of MSDC in dampening host diseases and promoting metabolic health after influenza virus and mouse-adapted SARS-CoV-2 MA10 virus infection in lean and obese mouse models. scRNA-seq, bulk RNA-seq, metabolic analysis, and flow cytometry were used to uncover the cellular, molecular, and metabolic mechanisms by which MSDC regulates lung macrophage inflammatory process after influenza or SARS-CoV-2 infection. Last, we tested the potential roles of MSDC in regulating SARS-CoV-2-induced inflammation in humans by culturing SARS-CoV-2-infected primary human lung macrophages and human COVID-19 lung autopsy samples in vitro with or without MSDC. Viral infections in mice were ended upon mouse sacrifice at indicated days after infection. In general, experiments were conducted in replicates, and the number of mice used in the studies is included in figure legends.

Ethics and biosafety
The study involving human participants was reviewed and approved by Mayo Clinic Institutional Review Boards (IRB no. 19-012187). All animal experiments were performed in animal housing facilities at the Mayo Clinic (Rochester, MN) or University of Virginia (UVA; Charlottesville, VA). Sex-and age-matched adult mice of both sexes unless otherwise specified were used in the experiments. The animal experiments were approved by the Mayo Clinic or UVA Institutional Animal Care and Use Committees. All work with SARS-CoV-2 infection was approved for use under Animal Biosafety Level 3/Biosafety Level 3 (ABSL3/BSL3) conditions and was performed with approved standard operating procedures and safety conditions by the UVA Institutional Review Board.

Mouse and infection
WT C57BL/6 (catalog no. 000664), Lyz2-cre (catalog no. 004781), and Mpc2 fl/fl (catalog no. 032118) mice were purchased from the Jackson Laboratory and bred in-house. Mpc2 ΔLyz2 mice were generated by crossing Mpc2 fl/fl mice with Lyz2-cre mice. High-fat DIO male mice on C57BL/6 background (with 60% kcal fat chow for 20 weeks) (catalog no. 380050) and age/sex-matched control mice (catalog no. 380056) were purchased from the Jackson Laboratory and bred in-house for another 2 weeks with 60% kcal fat chow (Research Diets, catalog no. D12492) or normal chow before experiments. All mice were housed in a specific pathogen-free environment. For host morbidity experiments after regular-dose IAV infection, the influenza A/PR8/34 strain was diluted in fetal bovine serum (FBS)-free Dulbecco's Modified Eagle's Medium on ice and inoculated in anesthetized mice through the intranasal route. For host mortality experiments after high dose (2.5-fold of the sublethal dose, lethal)-IAV infection, the outcome was determined on the basis of the humane endpoint (more than 30% weight loss or moribund) or deaths before humane sacrifice as described before (38). For SARS-CoV-2 MA10 infection, mice were infected with 10 5 plaque-forming units (PFU) (for 9-to 12-weekold WT C57BL/6 mice), 8 × 10 4 PFU (for aged C57BL/6 mice), or 10 4 PFU (for 26-to 28-week-old DIO mice) of MA10 intranasally under anesthesia. Body weight was monitored daily for virus-infected mice.

MPC inhibitor MSDC treatment in vivo
MSDC was provided by Cirius Therapeutics, which was developed for nonalcoholic steatohepatitis. MSDC was dissolved in dimethyl sulfoxide (DMSO). For treatment of IAV-infected WT C57BL/6 lean mice or DIO mice, mice were administered by intraperitoneal (for C57BL/6 lean mice) or oral gavage (for DIO mice) daily with either 5% DMSO as vehicle or MSDC (30 mg/kg) in 200 μl of phosphate-buffered saline (PBS) from 1 to 8 d.p.i. unless otherwise specified. For treatment of SARS-CoV-2 MA10-infected WT C57BL/6 mice, aged C57BL/6 mice, or DIO mice, mice were administered by intraperitoneal (for C57BL/6 mice or aged C57BL/6 mice) or oral gavage (for DIO mice) daily with either 5% DMSO as vehicle or MSDC (30 mg/kg) in 200 μl of PBS from 6 hours after infection to 7 d.p.i. unless otherwise specified. Because DIO mice have greatly enhanced morbidity and mortality after infection, we reasoned that, for DIO mice, intraperitoneal injection of 200-μl liquid daily is a huge burden; therefore, we chose to use oral gavage. For treatment of SARS-CoV-2 MA10-infected DIO mice with nirmatrelvir, nirmatrelvir (PF-07321332) was purchased from MedChemExpress (catalog no. HY-138687). The compounds were dissolved in DMSO and formulated as 40 mg/ml in corn oil containing 10% DMSO. DIO mice were administered by oral gavage with either 10% DMSO as vehicle, nirmatrelvir (300 mg/ kg), or nirmatrelvir (300 mg/kg) plus MSDC (30 mg/kg). The treatment of nirmatrelvir or MSDC was initiated at 6 hours after MA10 infection and continued twice daily for a total of 3 days for nirmatrelvir or continued once daily for a total of 7 days for MSDC, respectively. For treatment of SARS-CoV-2 MA10-infected WT C57BL/6 mice, mice were administered by intraperitoneal injection with 400 μg of immunoglobulin G (IgG) (Bio X Cell, catalog no. BE0091) or anti-IL-1β (Bio X Cell, catalog no. BE0246) antibodies (Abs) at days 1 and 3 after infection. Mice were monitored for body weight changes. At indicated time points, a subset of mice was euthanized, and lung or BAL samples were collected for inflammation and titer analysis. Another subset of mice were euthanized, and half of each lung lobe was taken for histopathology and fixed in 10% phosphate-buffered formalin before paraffin embedding and sectioning, and half of lung lobe was taken for further analysis.

Human AM culture and treatment in vitro
For human AMs, we selected donors without a history of immunosuppression and chemo-or radiotherapies and free of inflammation or pulmonary infection. The study was reviewed and approved by the Institutional Review Board (IRB no. 19-012187) at the Mayo Clinic. All participants provided written informed consent before sample collection and subsequent analysis.
Human AMs were obtained from BAL of adult donors undergoing flexible bronchoscopy as described before (38). About 100 to 200 ml of saline were instilled in 20-ml aliquots until 60 ml of lavage fluid were obtained. The specimen was placed on ice and immediately hand-carried to the laboratory for cell isolation. AMs were purified by adherence for 2 hours in complete medium (RPMI 1640, 10% FBS, and 1% penicillin/streptomycin/glutamate) at 37°C and 5% CO 2 . The nonadherent cells were washed off with warm PBS. The remaining adherent cells were cultured in complete medium supplemented with recombinant human granulocytemacrophage colony-stimulating factor (GM-CSF) (50 ng/ml) (Bi-oLegend, catalog no. 572903) and M-CSF (BioLegend, catalog no. 574804). For Poly IC treatment, AMs were pretreated with DMSO (vehicle) or MSDC (10 μM) for 2 hours; then, cells were stimulated with or without Poly IC (5 μg/ml) for 24 hours and were analyzed by quantitative RT-PCR.
For SARS-CoV-2 infection, AMs were pretreated with DMSO (vehicle) or MSDC (10 μM) for 2 hours. Subsequently, cells were washed with cold PBS and challenged with or without 1 multiplicity of infection (MOI) of SARS-CoV-2 virus for 1 hour and then cultured in the presence of DMSO or MSDC (10 μM) for 48 hours. Cells were analyzed by quantitative RT-PCR or NanoString.

Human lung tissue specimens
Lung autopsy samples from seven individuals who died from COVID-19 were obtained from the Mayo Clinic Department of Pathology. Informed consent was obtained from relatives of study participants. Lung tissue specimens were obtained within 24 hours of autopsy and immediately processed for single-cell suspension. For lung cell treatment ex vivo, the cells were treated with DMSO (vehicle) or MSDC (10 μM) in complete medium supplemented with recombinant human GM-CSF and M-CSF (50 ng/ml) for 24 hours. Cells were analyzed by quantitative RT-PCR.

Glucose tolerance test
Glucose tolerance testing was performed as described before (70). Briefly, DIO mice were weighed and fasted overnight at 4 d.p.i. Then, the mice were injected intraperitoneally with D-glucose (1 g/kg) in 0.9% NaCl immediately after collecting blood from tail vein (T = 0) at 5 d.p.i. Subsequently, the blood was obtained at 0.5, 1, 2, and 4 hours after injection from tail vein. Blood glucose was measured with 2 μl of serum from each blood sample at indicated time points by colorimetric glucose assay kit (Abcam, catalog no. ab65333) according to the manufacturer's instructions.

Total cholesterol detection
Total cholesterol concentrations in each blood sample were measured with colorimetric cholesterol assay kit (Abcam, catalog no. ab65390) according to the manufacturer's instructions. Intraassay coefficients of variability (CVs) were 0.8, 0.8, and 0.7% at 85, 178, and 340 mg/dl, respectively.

Acetyl-CoA measurement
Mouse AMs or BMDMs were pretreated with DMSO (vehicle) or MSDC (10 μM) for 2 hours; then, cells were stimulated with Poly IC (5 μg/ml) for 24 hours. The concentration of acetyl-CoA was quantified by BioVision's PicoProbe Acetyl CoA Assay Kit (catalog no. MAK039) according to the protocols provided by the manufacturer.

Bulk RNA sequencing
Total RNA from lungs of IAV-infected mice and in vitro cultured AMs was used for bulk RNA-seq. After quality control, high-quality (Agilent Bioanalyzer RIN >7.0) total RNA was used to generate the RNA-seq library. cDNA synthesis, end-repair, A-base addition, and ligation of the Illumina indexed adapters were performed according to TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA). The concentration and size distribution of the completed libraries were determined using an Agilent Bioanalyzer DNA 1000 chip (Santa Clara, CA) and Qubit fluorometry (Invitrogen, Carlsbad, CA). Paired-end libraries were sequenced on Illumina HiSeq 4000 following Illumina's standard protocol using the Illumina cBot and HiSeq 3000/4000 PE Cluster Kit. Base calling was performed using Illumina's RTA software (version 2.5.2). Paired-end RNAseq reads were aligned to the mouse reference genome (GRCm38/ mm10) using RNA-seq spliced read mapper Tophat2 (v2.2.1). Preand post-alignment quality controls, gene-level raw read count, and normalized read count [i.e., fragments per kilobase per million reads (FPKM)] were performed using the RSeQC package (v2.3.6) with the National Center for Biotechnology Information (NCBI) mouse RefSeq gene model. Differential expression for each gene between various groups specified in the text was identified on basis of the results of DESeq2 Wald tests. For visualization, data were logarithmic-transformed, and genes that exhibited log 2 fold change values > 2 and log 10 P > 25 between compared groups were highlighted. For functional analysis, gene set enrichment analysis (GSEA) (71) was applied to identify enriched gene sets from MSigDB, using a weighted enrichment statistic and a log 2 ratio metric for ranking genes. The bulk RNA-seq was conducted once using multiple biological samples per group (as indicated in figures).
Single-cell RNA sequencing C57BL/6 WT mice were infected with~200 PFU of IAV and treated with vehicle or MSDC for 3 days. Lung cells were pooled from three individual mice from each group at 4 d.p.i. and subjected to scRNAseq analysis. Single-cell libraries were prepared using the Chromium Single Cell 5′ Reagent Kit (10x Genomics) following the manufacturer's instruction. Paired-end sequencing was performed using DNBSEQ-G400 in rapid-run mode. scRNA-seq data were aligned and quantified using 10x Genomics Cell Ranger Software Suite. Subsequently, the doublet cells were removed by the package "scDblFinder." Remaining cells were analyzed using "Seurat" (version 4.1.1). The following criteria were applied for quality control: gene number > 200; UMI count > 1000; and mitochondrial gene percentage < 5%. The workflow included normalization, dimension reduction, and clustering as well as identification of marker genes for clusters and differentially expressed genes. GSEA (71) analysis was based on the results of FindAllMarkers with the package clusterProfiler (72).

Metabolic analysis
Real-time OCR and ECAR of AMs or BMDM were measured using Seahorse XFp Analyzers (Seahorse Bioscience) (38). AMs or BMDMs (1 × 10 5 ) were seeded into each well of Seahorse XFp Cell Culture Miniplates and pretreated with DMSO (vehicle) or MSDC (10 μM) for 2 hours; then, cells were stimulated with or without Poly IC (5 μg/ml) overnight at 37°C and 5% CO 2 . On the following day, the cells were washed twice and incubated at 37°C for 1 hour in the absence of CO 2 in unbuffered assay medium (pH 7.4, Agilent Technologies) with 10 mM glucose for mitochondrial stress test (or without glucose for glycolytic stress test). OCR and ECAR were measured under basal conditions and after the addition of the following compounds: 1 μM oligomycin, 1.5 μM FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone), 0.5 μM rotenone + 0.5 μM antimycin, 10 mM glucose, and 50 mM 2-DG (2-deoxy-Dglucose) (all compounds obtained from Sigma-Aldrich) as indicated. Data were analyzed with Wave Desktop software version 2.6 (Agilent Technologies).

Measurement of mitochondrial mass
AMs or BMDMs (1 × 10 5 ) were seeded into 24-well plates and pretreated with DMSO (vehicle) or MSDC (10 μM) for 2 hours; then, cells were stimulated with or without Poly IC (5 μg/ml) overnight at 37°C and 5% CO 2 . On the following day, the cells were washed and rinsed and incubated with MitoTracker Deep Red (Invitrogen, catalog no. M22426) and MitoTracker Green (Invitrogen, catalog no. M7514) at 50 nM for 30 min at 37°C. Then, cells were washed twice with PBS and lifted off the plates for flow cytometry.

Metabolite analysis
For TCA analyte testing as described before (73)

Statistical analysis
Data are means ± SEM of values from individual mice (in vivo experiments). Unpaired two-tailed Student's t test (two-group comparison), one-way analysis of variance (ANOVA; multiple-group comparison), multiple t test (weight loss), or log-rank test (survival study) was used to determine statistical significance by GraphPad Prism software. We considered P < 0.05 to be significant. *P < 0.05; **P < 0.01.

Supplementary Materials
This PDF file includes: Materials and Methods Figs. S1 to S14 Table S1 Other Supplementary Material for this manuscript includes the following: Data file S1 MDAR Reproducibility Checklist View/request a protocol for this paper from Bio-protocol.