Introduction

Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis (TB), has caused disease and death in humans for centuries (WHO, 2023). Mtb primarily infects macrophages in the lung (Cohen et al., 2018; Wolf et al., 2007) wherein the bacterium relies on host derived fatty acids and cholesterol for synthesis of its lipid rich cell wall, and to produce energy and virulence factors (Peyron et al., 2008; Russell et al., 2009; Singh et al., 2012; Daniel et al., 2011; Muñoz-Elías and McKinney, 2005; Pandey and Sassetti, 2008; Brzostek et al., 2009). Within the lung microenvironment, resident alveolar macrophages preferentially oxidize fatty acids and are more permissive to Mtb growth while recruited interstitial macrophages are more glycolytic and restrictive of Mtb replication (Huang et al., 2018; Pisu et al., 2020; Pisu et al., 2021). Globally, Mtb infection modifies macrophage metabolism in a manner that enhances its survival. Mtb infected macrophages shift their mitochondrial substrate preference to exogenous fatty acids which drives the formation of foamy macrophages that are laden with cytosolic lipid droplets (Peyron et al., 2008; Russell et al., 2009; Cumming et al., 2018; Singh et al., 2012; Podinovskaia et al., 2013). Foamy macrophages are found in abundance in the central and inner layers of granulomas, a common histopathological feature of human TB (Russell et al., 2009; Kim et al., 2010). Interference with key regulators of lipid homeostasis, such as the miR-33 and the transcription factors peroxisome proliferator-activated receptor α (PPARα) and PPAR-γ enhances macrophage control of Mtb (Kim et al., 2017; Almeida et al., 2009; Ouimet et al., 2016). Moreover, compounds which modulate lipid metabolism such as the anti-diabetic drug metformin and some cholesterol lowering drugs are under investigation for host directed therapeutics (HDTs) against Mtb (Parihar et al., 2014; Singhal et al., 2014). Although the dependence of intracellular Mtb on host fatty acids and cholesterol is well documented (Wilburn et al., 2018), the impact of specific aspects of macrophage lipid metabolism on the bacteria remains opaque. In Mtb infected foamy macrophages, bacteria containing phagosomes are found in close apposition to intracellular lipid droplets (Peyron et al., 2008). It is believed that the bacterial induction of a foamy macrophage phenotype in host cells results in a steady supply of lipids that addresses the bacteria’s nutritional requirements (Peyron et al., 2008; Russell et al., 2009; Singh et al., 2012; Daniel et al., 2011). In fact, intracellular Mtb has been shown to import fatty acids from host lipid droplet derived triacylglycerols (Daniel et al., 2011). However, other studies indicate that macrophage lipid droplet formation in response to Mtb infection can lead to the induction of a protective, anti-microbial response (Knight et al., 2018). Mtb appears unable to acquire host lipids when lipid droplets are induced by stimulation with interferon gamma (IFN-γ) (Knight et al., 2018). Moreover, there is some evidence that lipid droplets can be sites for the production of host protective pro-inflammatory eicosanoids (Knight et al., 2018; Daniel et al., 2011). Lipid droplets can also act as innate immune hubs against intracellular bacterial pathogens by clustering anti-bacterial proteins (Bosch et al., 2020). Inhibition of macrophage fatty acid oxidation by knocking out mitochondrial carnitine palmitoyl transferase 2 (CPT2) or using chemical inhibitors of CPT2 also restrict intracellular growth of Mtb (Chandra et al., 2020; Huang et al., 2018). These data demonstrate that modulation of the different stages in lipid metabolism inside Mtb infected macrophages can result in opposing outcomes.

We carried out a candidate-based, CRISPR mediated knockout of lipid import and metabolism genes in macrophages to determine their roles in intracellular growth of Mtb. By targeting genes involved in fatty acid import, sequestration, and metabolism in Hoxb8 derived conditionally immortalized murine macrophages (Kiritsy et al., 2021), we show that impairing lipid homeostasis in macrophages at different steps in the process negatively impacts the growth of intracellular Mtb, albeit to differing degrees. The impact on Mtb growth in the mutant macrophages was mediated through different mechanisms despite some common anti-microbial effectors. Mtb infected macrophages deficient in the import of long chain fatty acids increased production of pro-inflammatory markers such as interleukin 1β (IL-1β). In contrast, ablation of lipid droplet biogenesis and fatty acid oxidation increased production of reactive oxygen species and limited the bacteria’s access to nutrients. We also found that suppression of Mtb growth in macrophages that are unable to produce lipid droplets could not be rescued by exogeneous addition of fatty acids indicating that this is not purely nutritional restriction. Our data indicate that interference of lipid metabolism in macrophages leads to suppression of Mtb growth via multiple routes.

Results

Knockout of fatty acid import and metabolism genes restricts Mtb growth in macrophages

To apply a holistic approach to assessing the role(s) of fatty acid metabolism on the intracellular growth of Mtb, we used a CRISPR genetic approach to knockout genes involved in lipid import (CD36, FATP1), lipid droplet formation (PLIN2) and fatty acid oxidation (CPT1A, CPT2) in mouse primary macrophages (Figure 1A). Deletion of CD36 or CPT2 from mouse macrophages has been shown to impair intracellular growth of Mtb (Hawkes et al., 2010; Chandra et al., 2020). But the role of specialized long chain fatty acid transporters (FATP1-6) on Mtb growth in macrophages is uncharacterized. FATP1 and FATP4 are the most abundant fatty acid transporter isoforms in macrophages (Nishiyama et al., 2018). PLIN2, or adipophilin, is known to be required for lipid droplet formation (Paul et al., 2008; Larigauderie et al., 2004). Five isoforms of mammalian perilipins (PLIN) are involved in lipid droplet biogenesis amongst which PLIN2 is the dominant isoform expressed in macrophages (Knight et al., 2018). However, macrophages derived from PLIN2^-/- mice show no defects in production of lipid droplets nor do they impair intracellular growth of Mtb (Knight et al., 2018). We targeted each of these genes with at least 2 sgRNAs in Hoxb8 Cas9⁺ conditionally immortalized myeloid progenitors (Kiritsy et al., 2021) to generate a panel of mutants that were deficient in the following candidates of interest; FATP1^-/-, PLIN2^-/-, CD36^-/-, CPT1A^-/- and CPT2^-/-. Each individual sgRNA achieved >85% CRISPR-mediated deletion efficiency for all the 5 genes, as analyzed by the Inference for CRISPR Edits (ICE) tool (Conant et al., 2022) (Supplementary file 1). We verified the protein knockout phenotypes by flow cytometry and western blot analysis of differentiated macrophages derived from the CRISPR deleted Hoxb8 myeloid precursors (Figure 1 – figure supplement 1).

Figure 1.

To confirm certain knockout phenotypes functionally, we checked lipid droplet biogenesis in PLIN2^-/- macrophages in comparison to macrophages transduced with a non-targeting scramble sgRNA by confocal microscopy of BODIPY stained cells. Cells were cultured for 24 hours in the presence of exogenous oleate to enhance the formation of lipid droplets (Listenberger and Brown, 2007). We observed a complete absence of lipid droplet formation in PLIN2^-/- macrophages as compared to controls (Figure 1 - figure supplement 2A). This is contrary to previous observations in macrophages derived from PLIN2 knockout mice which were reported to have no defect in lipid droplet formation (Knight et al., 2018). We also assessed the ability of CPT2^-/- macrophages to oxidize fatty acids using the Agilent Seahorse XF Palmitate Oxidation Stress Test. Scrambled sgRNA and CPT2^-/- macrophages were cultured in substrate limiting conditions and supplied with either bovine serum albumin (BSA) or BSA conjugated palmitate. As shown in Figure 1 - figure supplement 2B, control macrophages were able to utilize and oxidize palmitate in substrate limiting conditions indicated by a significant increase in oxygen consumption rates (OCRs) in contrast to cells supplied with BSA alone. Addition of the CPT1A inhibitor, etomoxir, inhibited the cell’s ability to use palmitate in these conditions (Figure 1 - figure supplement 2B). CPT2 knockout in CPT2^-/- macrophages impaired the cell’s ability to oxidize palmitate to a degree comparable to etomoxir treatment as evidenced by baseline OCRs when compared to scrambled sgRNA control (Figure 1 - figure supplement 2C).

We then assessed the different knockout mutant macrophages in their ability to support the intracellular growth of Mtb. We infected macrophages with Mtb Erdman at a multiplicity of infection (MOI) of 0.4 and assessed intracellular bacterial growth rates by counting colony forming units (CFUs) 5 days post infection. All the 5 mutant macrophages significantly impaired Mtb growth rates when compared to scrambled sgRNA as assessed by CFUs counts on day 5 (Figure 1B). PLIN2^-/-, FATP1^-/- and CPT2^-/- macrophages displayed the strongest growth restriction phenotypes while CD36^-/-macrophages had a moderate, but significant, impact on Mtb growth. In parallel, we quantified intracellular bacteria on day 0, 3 hours post infection (Figure 1C), to ascertain that subsequent differences on day 5 were not due to any disparity in initial bacterial uptake. The moderate growth restriction phenotypes of CD36^-/- macrophages were consistent with previous findings which reported similar impact on Mtb and M. marinum growth in macrophages derived from CD36^-/- mice (Hawkes et al., 2010). Impaired growth of Mtb in CPT1A^-/- and CPT2^-/- macrophages is also consistent with previous reports that genetic and chemical inhibition of fatty acid oxidation is detrimental to the growth of Mtb within macrophages (Huang et al., 2018; Chandra et al., 2020).

Mtb infected macrophages with impaired fatty acid import and metabolism display altered mitochondrial metabolism and elevated glycolysis

Impairment of fatty acids metabolism by FATP1 knockout in macrophages rewires their substrate bias from fatty acids to glucose (Johnson et al., 2016). We reasoned that deletion of genes required for downstream processing of lipids (Figure 1A) could also reprogram macrophages and increase glycolysis which could, in part, explain bacterial growth restriction. We analyzed the metabolic states of 3 knockout macrophages (FATP1^-/-, PLIN2^-/- and CPT2^-/-) in uninfected and Mtb infected conditions by monitoring oxygen consumption rates (OCR) and extracellular acidification rates (ECARs) using the Agilent Mito and Glucose Stress Test kits. All 3 mutant uninfected macrophages displayed reduced mitochondrial respiration as evidenced by lower basal and spare respiratory capacity (SRC) when compared to scrambled sgRNA controls (Figure 2 - figure supplement 1A, 1B). Mtb infection proportionally reduced basal and SRC rates across all the mutant macrophages and scrambled controls (Figure 2A, 2B) when compared to uninfected macrophages, which is consistent with previous findings (Cumming et al., 2018). PLIN2^-/- macrophages displayed the most marked reduction in mitochondrial activity in both uninfected and infected conditions, while FATP1^-/- macrophages were the least affected (Figure 2A, Figure 2 - figure supplement 1A). As reported previously (Johnson et al., 2016), uninfected FATP1^-/- macrophages were more glycolytically active with higher basal and spare glycolytic capacity (SGC) compared to scrambled controls (Figure 2 - figure supplement 1C, 1D). Uninfected PLIN2^-/- and CPT2^-/- macrophages were also more glycolytically active, but to a greater degree than FATP1^-/- (Figure 2 - figure supplement 1C, 1D). Mtb infection increased the glycolytic capacity of all the 3 mutant macrophages (Figure 2C, 2D). Overall, PLIN2^-/- macrophages exhibited the highest glycolytic capacity (Figure 2C, Figure 2 - figure supplement 1D). Our data indicates that impairment of fatty acid metabolism at different steps significantly impacts mitochondrial respiration and reprograms cells towards glycolysis. Increased glycolytic flux in macrophages has been linked to the control of intracellular Mtb growth (Gleeson et al., 2016; Shi et al., 2015). Metabolic realignment as a consequence of interference with lipid homeostasis, which results in enhanced glycolysis may contribute to Mtb growth restriction in these mutant macrophages.

Figure 2.

Knockout of lipid import and metabolism genes in macrophages activates AMPK and stabilizes HIF-1α

Mtb infection is known to induce increased glycolysis or the “Warburg effect” in macrophages, mouse lungs and human TB granulomas (Shi et al., 2015; Gleeson et al., 2016; Belton et al., 2016). Several studies have demonstrated that the Warburg effect is mediated by the master transcription factor hypoxia-inducible factor-1 (HIF-1) (Courtnay et al., 2015). During Mtb infection, HIF-1 is activated by production of reactive oxygen species (ROS), TCA cycle intermediates and hypoxia in the cellular microenvironments as a consequence of altered metabolic activities and increased immune cell functions (Li et al., 2024; Shi et al., 2015; Gleeson et al., 2016; Belton et al., 2016). We assessed HIF-1 stability in the 3 mutant macrophage lineages (FATP1^-/-, PLIN2^-/- and CPT2^-/-) by monitoring total HIF-1α protein levels by western blot, having confirmed that they were all more glycolytically active than the scrambled controls (Figure 2, Figure 2 - figure supplement 1). Indeed, all the 3 mutant macrophages displayed significantly higher amounts of total HIF-1α when compared to scrambled controls in uninfected conditions and after Mtb infection for 48 hours (Figure 3A, 3B). We also checked the phosphorylation status of the adenosine monophosphate kinase (AMPK), a master regulator of cell energy homeostasis (Garcia and Shaw, 2017), in the mutant macrophages since Seahorse flux analyses indicated that they had impaired mitochondrial activities (Figure 2, Figure 2 - figure supplement 1). Western blot analysis revealed that in both Mtb infected and uninfected conditions, impaired fatty acid metabolism correlated with moderate activation of AMPK as indicated by increased AMPK phosphorylation (Figure 3C, 3D). These data point to a metabolic reprogramming of cells through activation of HIF-1α and AMPK to promote glycolysis. In energetically stressed cellular environments, activated AMPK promotes catabolic processes such as autophagy to maintain nutrient supply and energy homeostasis (Garcia and Shaw, 2017). Autophagy is also an innate immune defense mechanism against intracellular Mtb in macrophages (Gutierrez et al., 2004). We examined the levels of autophagic flux in the mutant macrophages by monitoring LC3I to LC3II conversion by western blot and by qPCR analysis of selected autophagy genes (AMBRA1, ATG7, MAP1LC3B and ULK1). We observed a moderate increase in autophagic flux by western blot in uninfected FATP1^-/- and CPT2^-/- macrophages, which was amplified upon infection with Mtb (Figure 2 - figure supplement 2A, 2B). Our qPCR analysis revealed that the 4 autophagy genes were upregulated in both Mtb infected and uninfected conditions in all the 3 mutants (Figure 2 - figure supplement 2C, 2D). These data suggest that impaired fatty acid import and metabolism in macrophages could be restricting Mtb growth by promoting autophagy. These data agree with previous observations that inhibition of fatty acid oxidation enhances macrophage xenophagic activity whichleads to improved control of Mtb (Chandra et al., 2020).

Figure 3.

Exogenous oleate fails to rescue the Mtb icl1-deficient mutant in FATP1^-/-, PLIN2^-/- and CPT2^-/- macrophages

The mycobacterial isocitrate lyase (icl1) acts as an isocitrate lyase in the glyoxylate shunt and as a methyl-isocitrate lyase in the methyl-citrate cycle (MCC) (Gould et al., 2006; McKinney et al., 2000). Mtb uses the MCC to convert propionyl CoA originating from the breakdown of cholesterol rings and β-oxidation of odd chain fatty acids into succinate and pyruvate, which are eventually assimilated into the TCA cycle (Muñoz-Elías et al., 2006; Griffin et al., 2012). The buildup of propionyl CoA is toxic to Mtb and the bacteria relies on the MCC together with the incorporation of propionyl CoA to methyl-branched lipids in the cell wall as an internal detoxification system (Muñoz-Elías et al., 2006; Savvi et al., 2008). Mtb propionyl CoA toxicity is, in part, due to a cellular imbalance between propionyl CoA and acetyl CoA as an accumulation of the former or paucity of the latter results in the propionyl CoA mediated inhibition of pyruvate dehydrogenase (Lee et al., 2013). Consequently, Mtb icl1 deficient mutants (Mtb Δicl1) are unable to grow in media supplemented with cholesterol or propionate, or intracellularly in macrophages (Lee et al., 2013). However, this growth inhibition could be rescued both in culture and in macrophages by exogenous supply of acetate or even chain fatty acids which can be oxidized to acetyl-CoA (Lee et al., 2013). We took advantage of this metabolic knowledge to assess whether exogenous addition of the even chain fatty acid oleate can rescue the intracellular growth of Mtb Δicl1 mutants in our CRISPR knockout macrophages. Scrambled controls, FATP1^-/-, PLIN2^-/- and CPT2^-/- macrophages in normal macrophage media or media supplemented with oleate were infected with an Mtb Δicl1 strain expressing mCherry at MOI 5. Bacterial growth measured by mCherry expression was recorded 5 days post infection. Consistent with previous observations (Lee et al., 2013), the Mtb Δicl1 mutant failed to replicate in both mutant and scramble macrophages that were grown in normal macrophage media as evidenced by baseline mCherry fluorescence (Figure 4A). Oleate supplementation successfully rescued the Mtb Δicl1 mutant in scrambled control macrophages. However, the growth restriction of the Mtb Δicl1 strain could not be alleviated by exogenous addition of oleate to the mutant macrophages (FATP1^-/-, PLIN2^-/- and CPT2^-/-) (Figure 4A). These data suggest that impaired import (FATP1^-/-), sequestration (PLIN2^-/-) or β-oxidation of fatty acids (CPT2^-/-) blocks Mtb’s ability to access and use cellular lipids.

Figure 4.

Oleate supplementation in macrophages induces the formation of lipid droplets and we were able to confirm the inability to produce lipid droplets in PLIN2^-/- macrophages using this approach (Figure 1 - figure supplement 2A). As an indirect measure to track the fate of supplemented oleate in the mutant macrophages, we monitored lipid droplet biogenesis in FATP1^-/- and CPT2^-/- macrophages to check if the inability to rescue the Mtb Δicl1 impaired growth phenotypes in these mutant macrophages could be possibly related to disruptions in lipid droplet biogenesis. Confocal analysis of BODIPY stained cells upon oleate supplementation revealed that FATP1^-/- macrophages also fail to generate lipid droplets (Figure 4B). In contrast, CPT2^-/- macrophages produced more and larger lipid droplets in comparison to scrambled controls (Figure 4B). These data suggest that the inhibition of Mtb growth in these mutant macrophages is not merely through limitation of access to fatty acid nutrients.

Dual RNA sequencing to identify host and bacterial determinants of Mtb restriction in mutant macrophage lineages

We performed RNA sequencing of both host and bacteria in Mtb infected mutant macrophages as a preliminary step in identification of pathways restricting bacterial growth. We infected scrambled controls, FATP1^-/-, PLIN2^-/- and CPT2^-/- macrophages with the Mtb smyc’::mCherry strain for 4 days and processed the samples for dual RNA sequencing as previously described (Simwela et al., 2024). Principal component analysis (PCA) of host transcriptomes revealed a clustering of all the 3 mutant macrophages away from scrambled controls (Figure 5A). Interestingly, there was a separation in transcriptional responses between the 3 mutant macrophages as CPT2^-/- and PLIN2^-/- macrophages clustered closer together and more distant from FATP1^-/- (Figure 5A). Overall, using an adjusted p-value < 0.05 and absolute log2 fold change > 1.2, we identified 900 genes which were differentially expressed (DE) in PLIN2^-/- macrophages (589 up, 311 down), 817 genes which were DE in FATP1^-/- macrophages (501 up, 315 down) and 189 genes which were DE in CPT2^-/- (124 up, 65 down) (Supplementary file 2, Figure 5B). Consistent with the PCA analysis, Venn diagram of the DE genes (Figure 5B) indicated divergent responses in the 3 mutant macrophage populations. We performed pathway enrichment analysis (Wu et al., 2021) of the DE genes to identify anti-microbial pathway candidates in the 3 mutant macrophages. We found that defects in fatty acid uptake in FATP1^-/- infected macrophages upregulated pro-inflammatory pathways involved in MAPK and ERK signaling and production of inflammatory cytokines (IFN-γ, interleukin-6, Interleukin-1α, β) (Figure 5D, Supplementary file 3). The pro-inflammatory signatures of the FATP1^-/- macrophages are consistent with previous observations that demonstrated that a deficiency in FATP1 exacerbated macrophage activation in vitro and in vivo (Johnson et al., 2016). FATP1 is a solute carrier family member (SLC27A1) and Mtb infected FATP1^-/- macrophages showed reduced expression of other solute carrier transporters (SLC) such as FATP4 (SLC27A4), GLUT1 (SLC2A1) and 8 SLC amino acid transporters (Figure 5 - figure supplement 1A, 1B, 2A, Supplementary file 2). Interestingly, Mtb infected FATP1^-/- macrophage transcriptomes exhibited upregulation of macrophage scavenger receptors (MSR1) and the ATP binding cassette transporter ABCC1 (Figure 5 - figure supplement 1A), both of which can independently transport fatty acids into cells (Vogel et al., 2022; Raggers et al., 1999).

Figure 5.

Meanwhile, inability generate lipid droplets in Mtb infected PLIN2^-/- macrophages led to upregulation in pathways involved in ribosomal biology, MHC class 1 antigen presentation, canonical glycolysis, ATP metabolic processes and type 1 interferon responses (Figure 5C, Supplementary file 3). In the downregulated PLIN2^-/- DE gene set, enriched pathways included those involved in production of pro-inflammatory cytokines; interleukin-6 and 8, IFN-γ and interleukin-1 (Figure 5 - figure supplement 2B, Supplementary file 3). Oxidative phosphorylation and processes involved in the respiratory chain electron transport were also significantly enriched in PLIN2^-/- downregulated genes. This suggests that Mtb infected PLIN2^-/- macrophages increase glycolytic flux and decrease mitochondrial activities, which is consistent with our metabolic flux analysis data (Figure 2). Unlike FATP1^-/- macrophages, PLIN2^-/- macrophages are, however, broadly anti-inflammatory as most pro-inflammatory genes were downregulated upon Mtb infection.

Further downstream in the lipid processing steps, inhibition of fatty acid oxidation in CPT2^-/- macrophages upregulated pathways involved in MHC class 1 antigen presentation, response to IFN-γ and interleukin-1 and T-cell mediated immunity (Figure 5 - figure supplement 3, Supplementary file 3). There was a limited overlap in enriched pathways in the upregulated genes between Mtb infected CPT2^-/- and FATP1^-/- macrophages such as those involved in the cellular responses to interleukin-1 and IFN-γ. However, many pathways over-represented in CPT2^-/- macrophages were common to PLIN2^-/- macrophages (Figure 5C, Figure 5 - figure supplement 3, Supplementary file 3). Similarly, both Mtb-infected PLIN2^-/- and CPT2^-/- macrophages were downregulated in expression of genes involved in oxidative phosphorylation (Supplementary file 3). We confirmed expression levels of key genes by qPCR analysis of IL-1β and the type 1 interferon (IFN-β) response during the early time points of infection. Indeed, 4 hours post infection, IL-1β and IFN-β were both upregulated in FATP1^-/- macrophages as compared to scrambled controls consistent with their pro-inflammatory phenotype (Figure 5 - figure supplement 4A, 4B). On the contrary, PLIN2^-/- macrophages downregulated IL-1β (Figure 5 - figure supplement 4B). These data indicate that macrophages respond quite divergently to the deletion of the different steps in fatty acid uptake, which implies that the intracellular pressures to which Mtb is exposed may also differ.

Oxidative stress and nutrient limitation are major stresses experienced by Mtb in PLIN2^-/- and CPT2^-/- macrophages

We also analyzed transcriptomes from intracellular Mtb from scrambled controls, FATP1^-/-, CPT2^-/- and PLIN2^-/- macrophages in parallel with host transcriptomes in Figure 5A. Using an adjusted p-value of < 0.1 and an absolute log2 fold change > 1.4, 0 genes were DE in FATP1^-/- macrophages, 105 Mtb genes were DE in PLIN2^-/- macrophages (69 up, 36 down) and 10 genes were DE in CPT2^-/- macrophages (3 up, 7 down) (Figure 6A). Despite being restrictive to Mtb growth (Figure 1B) and appearing more pro-inflammatory (Figure 5D), FATP1^-/- macrophages did not elicit a detectable shift in the transcriptional response in Mtb, compared to control host cells. We speculate that pro-inflammatory responses in FATP1^-/- macrophages could be enough to restrict the growth of bacteria, but the resulting compensatory responses as evidenced by the upregulation of macrophage scavenger receptors (Figure 5 - figure supplement 1A) alleviate some of the stresses that a lack of fatty acid import could be duly exerting on the bacteria. PLIN2^-/- macrophages appeared to elicit the strongest transcriptional response from Mtb which is consistent with our CFU data (Figure 1B) as PLIN2^-/- macrophages exhibited the strongest growth restriction. Among the DE genes in Mtb from PLIN2^-/- macrophages, a significant number of upregulated genes are involved in nutrient assimilation (Figure 6B). Mtb in PLIN2^-/- macrophages upregulated CobU (Rv0254c) which is predicted to be involved in the bacteria’s cobalamin (Vitamin B₁₂) biosynthesis. Vitamin B₁₂ is an important cofactor for the activity of Mtb genes required for cholesterol and fatty acid utilization (Campos-Pardos et al., 2024; Savvi et al., 2008). Genes involved in de novo long chain fatty acid synthesis (AccE5, Rv281) (Bazet Lyonnet et al., 2014), cholesterol breakdown (HsaD, Rv3569c) (Lack et al., 2010), β-oxidation of fatty acids (EchA18, Rv3373; FadE22, Rv3061c) (Schnappinger et al., 2003), purine salvage (Apt, Rv2584c) (Warner et al., 2014) and tryptophan metabolism (Lott, 2020) were also upregulated in Mtb from PLIN2^-/- macrophages (Figure 6B). This metabolic re-alignment response is seen most frequently under nutrient limiting conditions (Huang et al., 2018; Pisu et al., 2020; Theriault et al., 2022). Mtb in PLIN2^-/- macrophages also appears to experience a significant level of other cellular stresses as genes involved in DNA synthesis and repair, general response to oxidative stress and PH survival in the phagosome (DnaN, Rv0002; RecF, Rv0003; DinF, Rv2836c; Rv3242c, Rv1264) were upregulated (Figure 6B). Amongst the down-regulated genes in Mtb in PLIN2^-/- macrophages, FurA (Rv1909c), a KatG repressor was the most significant (Figure 6B, Supplementary file 4). FurA downregulation derepresses the catalase peroxidase, KatG, which promotes Mtb survival in oxidative stress conditions (Zahrt et al., 2001). These data suggest that PLIN2^-/- macrophages could be, in part, restricting Mtb growth by increasing production of ROS. The data also suggest that, contrary to a previous report (Knight et al., 2018), blocking lipid droplet formation in host macrophages does place increased nutritional and oxidative stress on intracellular Mtb.

Figure 6.

CPT2^-/- macrophages elicited a modest shift in transcriptional response from Mtb, and the majority of the DE genes (8 out of 10) were also DE in PLIN2^-/- macrophages (Supplementary file 4). This in in agreement with the host transcription response as CPT2^-/- and PLIN2^-/- macrophages share similar candidate anti-bacterial responses (Figure 5A-C, Figure 5 - figure supplement 3). In common with the Mtb transcriptome response in PLIN2^-/- macrophages, upregulated genes in Mtb isolated from CPT2^-/- macrophages included those involved in response to oxidative stress (DinF, Rv2836c) (Supplementary file 4). To substantiate some of these pathways, we assessed the levels of total cellular ROS in FATP1^-/-, PLIN2^-/- and CPT2^-/- macrophages in both Mtb infected and uninfected conditions by staining the cells with the Invitrogen CellROX dye and confocal microscopy analysis of live stained cells. In both infected and uninfected conditions, all the 3 mutants displayed significantly elevated total cellular ROS when compared to scramble controls (Fig. 6 - figure supplement 1A, 1B).

Inhibitors of lipid metabolism block intracellular growth of Mtb in macrophages but not in broth culture

We next examined if pharmacological inhibitors would phenocopy the growth inhibition phenotypes we observed with specific gene deletions. Compounds which modulate lipid homeostasis are currently being investigated for HDT against TB, which is an area of considerable interest (Kim et al., 2020). Such compounds include metformin, a widely used anti-diabetic drug which activates AMPK, inhibits fatty acid synthesis and promotes β-oxidation of fatty acids (Fullerton et al., 2013; Singhal et al., 2014). Chemical inhibition of fatty acid β-oxidation is already known to promote macrophage control of Mtb (Chandra et al., 2020; Huang et al., 2018). We targeted macrophage lipid homeostasis with trimetazidine (TMZ), an inhibitor of β-oxidation of fatty acids, metformin, and an FATP1 inhibitor, FATP1 In (Matsufuji et al., 2013). We assessed the impact of these compounds on extracellular Mtb cultured in broth over 9 days in the presence of the inhibitors (DMSO, TMZ; 500 nM, Metformin; 2 mM, FATP1 In; 10 μM, Rifampicin; 0.5 μg/ml). None of the 3 lipid metabolism inhibitors had a measurable effect on Mtb growth in liquid culture media when compared to DMSO controls (Figure 7A). Treatment with rifampicin completely blocked bacterial growth under the same conditions (Figure 7A). We next infected scrambled sgRNA control macrophages with Mtb at MOI 0.5. Inhibitors were added to infected macrophages 3 hours post infection and bacterial CFUs were enumerated 4 days post treatment. Consistent with previous observations (Chandra et al., 2020; Singhal et al., 2014), TMZ and metformin significantly reduced bacterial loads in macrophages when compared to DMSO controls (Figure 7B). Similarly, FATP1 In also impacted the intracellular growth of Mtb (Figure 7B). The results provide independent data that both genetic and chemical modulation of fatty acid metabolism at different steps in the process negatively impacts intracellular growth of Mtb.

Figure 7.

Discussion

It is clearly established that host derived fatty acids and cholesterol are important carbon sources for Mtb (Wilburn et al., 2018). However, the relationship between Mtb and the infected host cell lipid metabolism remains a subject of conjecture. In this report, we characterized the role of macrophage lipid metabolism on the intracellular growth of Mtb by a targeted CRISPR mediated knockout of host genes involved in fatty acid import, sequestration and catabolism. Macrophage fatty acid uptake is mediated by the scavenger receptor CD36 and specialized long chain fatty acid transporters, FATP1 and FATP4 (Deng et al., 2023). Earlier studies reported that a deficiency of CD36 enhances macrophage control of Mtb, albeit to a modest degree (Hawkes et al., 2010). Work from Hawkes et al.(Hawkes et al., 2010) indicated that the anti-microbial effectors in CD36 deficient macrophages were not due to bacterial uptake deficiencies, differences in the rate of Mtb induced host cell death, production of ROS or pro-inflammatory cytokines (Hawkes et al., 2010). We similarly observed a moderate Mtb growth restriction phenotype in our CRISPR generated CD36^-/- macrophages. However, a strong growth restriction of Mtb was observed when we knocked out the long chain fatty acid transporter, FATP1. FATP1^-/- macrophages displayed altered metabolism characterized by stabilization of HIF-1α, activated AMPK, increased glycolysis and reduced mitochondrial functions. Given that both CD36 and FATP1 perform similar functions, it is expected that they should be some degree of compensation between the transporters when either of the genes are deleted. Indeed, we found out that FATP1 knockout resulted in the upregulation of other lipid import transporters (MSR1, ABCC1). This would be consistent with the moderate anti-Mtb phenotypes in CD36^-/- macrophages which could easily be compensated by the presence of FATPs to alleviate the reduction in fatty acid supply experienced by intracellular Mtb. However, the FATP1^-/- macrophage phenotype appears to be more severe on Mtb and could be exacerbated by an elevated pro-inflammatory response as has been reported previously both in vitro and in vivo (Johnson et al., 2016).

After uptake into the cells, most fatty acids either undergo β-oxidation in the mitochondria to provide energy or are esterified with glycerol phosphate to form triacylglycerols that may be incorporated into lipid droplets in the endoplasmic reticulum (Deng et al., 2023). Over the last 2 decades, it has been believed that lipid droplets are a nutrient source for Mtb in macrophages (Russell et al., 2009; Daniel et al., 2011; Peyron et al., 2008; Singh et al., 2012). However, recent work indicates lipid droplets may serve as centers for the production of pro-inflammatory markers and anti-microbial peptides (Knight et al., 2018; Bosch et al., 2020). It has also been reported that bone marrow macrophages derived from PLIN2^-/- mice did not have defects in the formation of lipid droplets and supported robust intracellular Mtb replication (Knight et al., 2018). However, our mutant macrophages with myeloid specific knockout of PLIN2 are unable to form lipid droplets and are defective in supporting the growth of Mtb. The discrepancies with previous observations (Knight et al., 2018) could be a consequence of compensatory responses to PLIN2 knockout in whole mice which when performed at embryonic level would allow for sufficient time for the cells to recover by upregulating related PLIN isoforms. In fact, PLIN2^-/- macrophages displayed the strongest anti-Mtb phenotypes amongst our mutants exhibiting activated AMPK, increased glycolysis and autophagy and impaired mitochondrial functions. Mtb isolated from PLIN2^-/- macrophages displayed signatures of severe nutrient limitation and oxidative stress damage.

It has also been previously reported that chemical, genetic or miR33 mediated blockade of fatty acid β-oxidation in macrophages induces lipid droplet formation (Chandra et al., 2020; Ouimet et al., 2016) but that this enhanced lipid droplet formation does not correlate with improved intracellular Mtb growth (Chandra et al., 2020). We observed a similar phenotype as CPT2^-/- macrophages that generated larger and more abundant lipid droplets than scrambled control macrophages were still restrictive to Mtb growth. This implies that the presence or absence of lipid droplets does not in itself indicate whether a macrophage will support or restrict Mtb growth, and that the anti-microbial environment extends beyond simple nutrient availability.

In summary, our study shows that blocking macrophage’s ability to import, sequester or catabolize fatty acids chemically or by genetic knockout impairs Mtb intracellular growth. There are shared features between potential anti-microbial effectors in macrophages which lack the ability to import (FATP^-/-) or metabolize fatty acids (PLIN2^-/-, CPT2^-/-) such as increased glycolysis, stabilized HIF-1α, activated AMPK, enhanced autophagy and production of ROS. However, there are also intriguing points of divergence as FATP^-/- macrophages are more pro-inflammatory while PLIN2^-/- macrophages appear to be broadly anti-inflammatory. The routes to Mtb growth restriction in these mutant macrophages are clearly more complex than the bacteria’s inability to acquire nutrients. The data further emphasizes that targeting fatty acid homeostasis in macrophages at different steps in the process (uptake, storage and catabolism) is worthy of exploring in the development of new therapeutics against tuberculosis.

Materials and methods

All materials and methods are as described (Simwela et al., 2024) unless otherwise specified.

Flow cytometry and western blot analysis

Generation of CRISPR mutant Hoxb8 macrophages was carried out described previously (Simwela et al., 2024). Antibodies used for both western blot and flow cytometry were as follows: rat anti-Mouse CD36:Alexa Fluor®647 (Biorad, 10 μl/million cells), rabbit anti-PLIN2 (Proteintech, 1:1000), rabbit anti-FATP1 (Affinity biosciences, 1:1000), rabbit anti-CPT1A antibody (Proteintech, 1:1000), rabbit anti-CPT2 antibody (Proteintech, 1:1000), rabbit anti-HIF-1α antibody (Proteintech, 1:1000), rabbit anti-AMPKα(1:1000, Cell signalling Technology), rabbit anti-Phospho-AMPKα(1:500, Cell signalling Technology), rabbit anti-LC3B (1:1000, Cell signalling Technology) and rabbit anti-β Actin (1:1000, Cell Signalling Technology). For western blots, secondary antibodies used were anti-rabbit/mouse StarBright Blue 700 (1:2500, Biorad). Blots were developed and imaged as described previously (Simwela et al., 2024).

Staining for cellular lipid droplets

Macrophages monolayers in Ibidi eight-well chambers were supplemented with exogenous 400 μM oleate for 24 hours to induce the formation of lipid droplets (Listenberger and Brown, 2007). Cells were then fixed in 4% paraformaldehyde and stained with BODIPY™ 493/503 (Invitrogen, 1 μg/ml) in 150 mM sodium chloride. Stained cells were mounted with media containing DAPI and imaged using a Leica SP5 confocal microscope.

Seahorse XF palmitate oxidation stress test

A modified Seahorse mitochondrial stress test was used to measure macrophage’s ability to oxidize palmitate in substrate limiting conditions. 2 days before the assay, 1 x 10⁵ cells were plated in Seahorse cell culture mini plates. 1 day before the assay, macrophage media was replaced with the Seahorse substrate limited growth media (DMEM without pyruvate supplemented with 0.5 mM glucose, 1 mM glutamine, 1% FBS and 0.5 mM L-Carnitine). On the day of the assay, substrate limited media was replaced with assay media (DMEM without pyruvate supplemented with 2 mM glucose and 0.5 mM L-Carnitine). In selected treatment conditions, cells were either supplied with bovine serum albumin (BSA), BSA palmitate or BSA palmitate plus etomoxir (4 μM). Oxygen consumption rates (OCRs) were measured using the Mito Stress Test assay conditions as described previously (Simwela et al., 2024).

Rescue of the Mtb Δicl1 mutant in oleate supplemented media

The Mtb H37Rv Δicl1 mutant expressing mCherry (Lee et al., 2013) was used for the rescue experiments. The strain was maintained in 7H9 OADC broth as previously described (Simwela et al., 2024) in the presence of Kanamycin (25 μg/ml) and hygromycin (50 μg/ml). 24 hours before infection, macrophages were cultured in normal macrophage media or media supplemented with 400 μM oleate. Cells were then infected with the Mtb Δicl1 mutant at MOI 5. The bacterial mCherry signal was measured on day 0 and day 5 post infection on an Envision plate reader (PerkinElmer). Oleate was maintained throughout the experiment in the rescue assay conditions.

Measurement of total cellular ROS

Uninfected or Mtb infected macrophages monolayers in Ibidi eight-well chambers were stained with the CellROX Deep Red dye (Invitrogen) as per manufacturer’s staining protocol. Live stained cells were imaged on a Leica SP5 confocal microscope. Z-stacks were re-constructed in ImageJ from which mean fluorescence intensities (MFI) for individual cells were obtained.

Datasets

The RNA-seq data from the Dual RNA-seq analysis of infected mouse macrophages,542 which includes both macrophage and Mtb reads are available in GEO: GSE270571

Acknowledgements

We would like to thank Dr. Jen K. Grenier and Ann E. Tate from the Cornell BRC Transcriptional Regulation and Expression Facility for their help with the development of dual RNA-Seq protocols. This work was supported by grants from the National Institutes of Health (AI155319, AI162598, and OD032135), Bill and Melinda Gates Foundation and the Mueller Health Foundation to D.G.R. EJ was supported by T32AI007349.