Lipin-1 restrains macrophage lipid synthesis to promote inflammation resolution

Abstract

Macrophages are critical to maintaining and restoring tissue homeostasis during inflammation. The lipid metabolic state of macrophages influences their function and polarization, which is crucial to the resolution of inflammation. The contribution of lipid synthesis to proinflammatory macrophage responses is well understood. However, how lipid synthesis regulates proresolving macrophage responses needs to be better understood. Lipin-1 is a phosphatidic acid phosphatase with a transcriptional coregulatory activity that regulates lipid metabolism. We previously demonstrated that lipin-1 supports proresolving macrophage responses, and here, myeloid-associated lipin-1 is required for inflammation resolution, yet how lipin-1–regulated cellular mechanisms promote macrophage proresolution responses is unknown. We demonstrated that the loss of lipin-1 in macrophages led to increased free fatty acid, neutral lipid, and ceramide content and increased phosphorylation of acetyl-CoA carboxylase. The inhibition of the first step of lipid synthesis, the transport of citrate from the mitochondria, reduced lipid content and restored efferocytosis and inflammation resolution in lipin-1^mKO mice and macrophages. Our findings suggest macrophage-associated lipin-1 restrains lipid synthesis, promoting proresolving macrophage function in response to proresolving stimuli.

Introduction

Acute inflammation in response to noxious stimuli, often a beneficial response, must be resolved.¹ The resolution of inflammation is an active process that restores tissues to their normal function and prevents the harmful effects of unresolved inflammation.² An impairment in the resolution of inflammation is implicated in the pathophysiology of cardiometabolic and chronic inflammation–driven diseases.³ Macrophages are critical to initiating and resolving inflammatory responses.^4,5 To effectively carry out both inflammation and inflammation resolution, macrophages must polarize toward distinct phenotypes characterized by drastic transcriptional, translational, and metabolic changes.^6,7 Dysregulation in one or more of these pathways can significantly impact the macrophages' ability to function appropriately. In addition, the ability of macrophages to contribute to both the initiation and resolution of inflammation makes them key immune cells in the progression and resolution of cardiovascular diseases.⁸

Proinflammatory macrophages quickly upregulate glycolysis to meet energy demands and carry out inflammatory or antimicrobial responses such as cytokine production and phagocytosis.^9–11 Though glycolysis-derived pyruvate feeds into the tricarboxylic acid (TCA) cycle in proinflammatory macrophages, the deactivation of succinate dehydrogenase and isocitrate dehydrogenase disrupts oxidative metabolism, leading to the accumulation of citrate.^11,12 Citrate can be exported from the mitochondria by the mitochondrial citrate carrier (CIC)/citrate transport protein (CTP) to generate oxaloacetate-derived NAPDH for reactive oxygen species production and acetyl-CoA for de novo fatty acid (FA) biosynthesis.^11–13 Inhibition of the succinate dehydrogenase (SDH) complex leads to a buildup of succinate, promoting inflammatory cytokine production.^12,14,15 Because the TCA cycle generates electron carriers and the SDH is part of complex II of the electron transport chain, a dysfunctional TCA cycle impairs oxidative phosphorylation. Proresolving macrophages initially use glycolysis before switching to FA oxidation for metabolic needs. Though glycolysis is dispensable,¹⁶ it is suggested that glycolysis can fuel FA biosynthesis to provide substrates for FA oxidation in proresolving macrophages.¹² Unlike proinflammatory macrophages, proresolving macrophages have a functional TCA cycle that supports oxidative phosphorylation^11,12

Free FA metabolism can regulate macrophage polarization and effector functions.¹⁷ Accumulation of FA–derived phospholipids and triglycerides is a feature of lipopolysaccharide-activated macrophages as de novo FA synthesis supports proinflammatory macrophage activation, recruitment, phagocytosis, and production of inflammatory cytokines.^17,18 The contribution of lipid metabolism to proresolution responses is less well defined and more controversial. Interleukin (IL)-4 stimulation initially leads to increased β-oxidation but eventually requires sterol regulatory element binding proteins (SREBP)–mediated de novo lipid synthesis to restore lipid pools.^19,20 The contribution of β-oxidation to proresolving macrophage function is controversial. While β-oxidation does not appear to be required for canonical M2 gene expression,²¹ β-oxidation may promote proresolving macrophage effector functions and disease resolution.^22–24 Specifically, the enhancement of β-oxidation and oxidative phosphorylation enhanced the uptake of necroptotic cells.²³ Also, the catabolism of apoptotic cell (AC)–derived lipids orchestrates the production of IL-10, which promotes tissue repair.²² Consistent with the contribution of β-oxidation to proresolving macrophage function and disease resolution, impaired macrophage β-oxidation is associated with the progression of metabolic diseases such as hepatic steatosis, insulin resistance, atherosclerosis, and myocardial infarction.^22,25,26 Altogether, one must wonder about the upstream molecular events that decide the fate of FAs and how their use aligns with proresolving macrophage responses.

Lipin-1 is a phosphatidic acid phosphatase that converts phosphatidic acid into diacylglycerol.²⁷ Mutations in the gene encoding lipin-1 (LPIN1) lead to several metabolic syndromes in mice²⁸ and severe, sometimes fatal, episodic rhabdomyolysis in people.^29–31 Several studies have also reported that single nucleotide polymorphisms in the lipin-1 gene are associated with dyslipidemia, diabetes, and metabolic syndrome in humans.^32–34 Loss of lipin-1 also results in cardiac lipotoxicity,^35,36 mitochondria dysfunction,³⁵ insulin resistance,³⁷ and atherosclerosis.³⁸ In macrophages, the enzymatic activity of lipin-1 supports a proinflammatory response.^39,40 Lipin-1 also has an enzymatically independent transcriptional coregulator activity in which lipin-1 binds to transcription factors, such as peroxisome proliferator-activated receptors and SREBPs, to regulate their activity and subsequent gene expression.⁴¹ In nonmyeloid cells, lipin-1 augmentation of peroxisome proliferator–activated receptors activity upregulates FA utilization while suppressing lipid synthesis by inhibiting SREBP.^41,42 Altogether, during processes that require lipid catabolism, lipin-1 has been shown to suppress FA synthesis and upregulate lipid catabolism preferentially.^24,41,43 Previously, we demonstrated that lipin-1 nonenzymatic activity promotes proresolving macrophage polarization, β-oxidation, and efferocytosis.^24,44 We have also shown that in response to proresolving stimuli, loss of lipin-1 leads to a buildup of metabolites that may contribute to lipid synthesis.²⁴ These data suggest that lipin-1 potentially regulates lipid metabolism to align with macrophage function. However, the aspect of lipin-1–regulated lipid metabolism that supports proresolving macrophage function is unknown.

In this study, we have used transgenic mouse models lacking the enzymatic activity of lipin-1 or both activities of lipin-1 in myeloid cells to report that lipin-1 nonenzymatic activity promotes disease resolution and proresolving macrophage responses by restraining FA biosynthesis.

Material and methods

Mice

All mice studies were approved by LSU Health Shreveport Institutional Animal Care and Use Committee (#P22-014). Mice were housed and cared for according to the National Institutes of Health guidelines for the care and use of laboratory animals. Mice lacking lipin-1 enzymatic activity from myeloid cells (lipin-1^mEnzyKO) were generated as previously reported.⁴⁰ Briefly, mice with exons 3 and 4 of the Lpin1 gene flanked by LoxP sites (genetic background: C57BL/6J and SV129; generously provided by B.N.F. and R. Chrast) were crossed with C57BL/6J LysM-Cre transgenic mice purchased from the Jackson Laboratory. Mice entirely lacking lipin-1 from myeloid cells (lipin-1^mKO) were generated by crossing mice with exon 7 of the Lpin1 gene flanked by LoxP sites (genetic background: C57BL/6J and SV129; generously provided by B.N.F)⁴⁵ with C57BL/6J LysM-Cre transgenic mice purchased from the Jackson Laboratory. Age-matched lipin-1 flox/flox littermate mice were used as control mice. Mice were genotyped to detect Cre insertion using primer combinations based on Jackson Laboratory protocols and the following primers: AAGGAGGGACTTGGAGGATG, common forward; GTCACTCACTGCTCCCCTGT, wild-type (WT) reverse; ACCGGTAATGCAGGCAAAT, mutant reverse.

Bone marrow–derived macrophage generation

Mice femurs were harvested and flushed with unsupplemented Dulbecco's Modified Eagle Medium (DMEM) (Gibco; 11965-092) to isolate bone marrow. Bone marrow cells were purified through a series of centrifugations. Red blood cells were removed by lysing bone marrow cells with ammonium chloride-potassium carbonate lysis (1× ACK lysis buffer), followed by filtration. Purified bone marrow cells were then resuspended and incubated for 7 d in nontreated T175 tissue culture flasks at 37 °C and 5% CO₂ in bone marrow–derived macrophage (BMDM) differentiation media composed of KnockOut DMEM (Gibco; 10829-018), 30% L-cell conditioned medium (generated from L929 cells; ATCC; CCL-1), 1 mM sodium pyruvate (Corning; 25-000-CI), 2 mM GlutaMAX (Gibco; 35050-061), 100 U/mL penicillin-streptomycin (Gibco; 15140-122) 0.2% sodium bicarbonate (Gibco; 25080-094) and 20% fetal bovine serum (FBS) (Atlas Biologicals; EF-0500-A). To harvest macrophages, after cells reach 80% confluency, BMDM differentiating media was removed, cells were washed with Dulbecco’s phosphate-buffered saline without calcium and magnesium (DPBS) (14190-144), then subsequently incubated in 11 mM pH 7.6 EDTA (Fisher Scientific; BP120-1) in DPBS (Corning; 21-031-CV) to detach cells. BMDMs were then placed in D10 media containing DMEM (Gibco; 11965-092), 10% FBS (ATLAS Biologicals; EF-0500-A), 2 mM GlutaMAX (Gibco; 35050-061), 100 U/mL penicillin-streptomycin (Gibco; 15140-122), and 1 mM sodium pyruvate (Corning; 25-000-CI)

L-Cell conditioned medium

The murine fibroblast cell line L929 (ATCC; CCL-1) was grown in RPMI 1640 medium (Gibco; 11875-093) supplemented with 10% FBS (ATLAS Biologicals; EF-0500-A), 2 mM GlutaMAX (Gibco; 35050-061), 1 mM sodium pyruvate (Corning; 25-000-CI), and 100 U/mL penicillin-streptomycin (Gibco; 15140-122). Briefly, 3.75 × 10⁵ L929 cells were seeded in a T225 tissue culture flask with 75 mL supplemented RPMI media. Flasks were incubated for 12 to 14 d at 37 °C until cells were 100% confluent and formed a cobblestone appearance. The medium was collected, cleared of cell debris by centrifugation at 1,600 rpm, filtered (0.2 µm), stored at 4 °C for 24 h, 0 °C for 24 h, and finally at −80 °C until use.

AC generation

Jurkat cells (ATCC; VTIB-152) cultured in RPMI 1640 medium (Gibco; 11875-093) supplemented with 10% FBS (ATLAS Biologicals; EF-0500-A), 2 mM GlutaMAX (Gibco; 35050-061), 10 mM HEPES buffer solution (Gibco; 15630-080), and 100 U/mL penicillin-streptomycin (Gibco; 15140-122) were centrifuged to remove culture media, resuspended in phosphate-buffered saline (PBS). A total of 10 mL of Jurkat cells (1–1.5 × 10⁶ cells/mL) were seeded in 100 mm × 15 mm Petri dishes (Thermo Fisher Scientific; FB0875712) exposed to ultraviolet (UV) radiation for 5 min in the UV Clave Ultraviolet Chamber (Benchmark Scientific). PBS containing Jurkat cells was collected, and 6 mL of 11 mM EDTA (pH 7.6) was added to the Petri dishes for 10 min to detach the adherent cells and added to the initial pool of UV-radiated cells. Cells were centrifuged at 1,600 rpm for 5 min, resuspended in supplemented (0% FBS) DMEM, and incubated for 2 h at 37 °C before downstream experimental use.

Efferocytosis assay

A total of 1.25 × 10⁵ BMDMs from lipin-1^mKO mice and littermate control mice were seeded on 12-mm coverslips (Fisher Scientific; 12541002), allowed to sit for 3 h, and treated with recombinant mouse IL-4 (Leinco Technologies; I-207) at a 40 ng/mL concentration for 6 h. Following IL-4 stimulation, macrophages were challenged with CFSE (BioLegend; 423801)-labeled ACs at a 4:1 (AC/macrophage) ratio for 45 min. Unbound ACs were washed off with PBS, and cells on coverslips were fixed with 10% neutral buffered formalin (VWR; 16004-128) for 1 h. Fixing was stopped through a series of PBS washes. Cells on coverslips were stained with DAPI (1:36,400; Invitrogen; D3571) for 15 min, mounted on slides, and imaged on the BZ-X810 Keyence microscope (Keyence Corporation of America) at 20×. The number of macrophages was counted with the Keyence analyzer software version 1.1.2.4, and the number of engulfed ACs was counted with the ImageJ version 1.54h cell counter (National Institutes of Health). For inhibitor studies, cells were pretreated for 12 h with 10 µm Citrate Transport Protein Inhibitor (CTPi) (Calbiochem; 475877-5MG), 10 µm C75 (Sigma-Aldrich; C5490-5MG), or 10 µm CPD9 (Calbiochem; 5343350001), or for 2 h with 10 µm Myriocin (MedChemExpress; HY-N6798) before 6-h cotreatment with 40 ng/mL IL-4 and respective inhibitor. In the citrate complementation experiments, the experimental setup was modified from.⁴⁶ BMDMs were pretreated with increasing concentrations (1 mM, 3 mM, 6 mM) of sodium citrate tribasic dihydrate (Sigma-Aldrich; 71402-250G) for 2 h before 6-h cotreatment with 40 ng/mL IL-4 and sodium citrate tribasic at respective concentrations.

Nile red assay

A total of 1.75 × 10⁵ BMDMs from lipin-1^mKO mice and littermate control mice were seeded on 12-mm coverslips (Fisher Scientific; 12541002) or in tissue culture–treated 24-well plates, allowed to sit for 3 h, and treated with recombinant mouse IL-4 (Leinco Technologies; I-207) for 4 h. For inhibitor studies, cells were pretreated as aforementioned. After IL-4 stimulation, cells were fixed with 10% neutral buffered formalin (16004-128) for 1 h. Following PBS washes, cells were stained with 0.5 µg/mL Nile red (Invitrogen; N1142) for 10 min, washed, and stained with DAPI (1:36,400 dilution; D3571) for 15 min. Cells were imaged on the BZ-X810 Keyence microscope. Macrophage numbers were counted using the BZ-X800_Analyzer (Keyence Corporation of America), and Nile red fluorescent intensity was quantified using cellSens Dimension 1.16 (Olympus).

In vivo peritonitis

Lipin-1^mKO mice, lipin-1^mEnzyKO mice, and littermate control mice were injected with 0.5 cm³ (0.1 mg/mouse) of 0.2 mg/mL zymosan solution (zymosan A from Saccharomyces cerevisiae; Sigma-Aldrich; Z4250). After 4, 12, 24, and 48 h, 5 mL of sterile, cold fluorescence-activated cell sorting (FACS) wash buffer (1% bovine serum albumin and 0.1% sodium azide in PBS) was injected intraperitoneally into the mice and collected back as the peritoneal lavage. Cells in the peritoneal lavage were counted and prepared for flow cytometry staining. In detail, 5 × 10⁵ peritoneal cells were blocked with CD16/CD32 (1:200) for 25 min at 4 °C. Cells were treated with LIVE/DEAD Aqua (Invitrogen; L34957) according to the manufacturer’s instructions, followed by incubation with AF700-conjugated anti-CD45.2 (1:2000) (109821, clone 104; BioLegend), PECy7-conjugated anti-CD11b (1:4000) (25-0112-81, clone M1/70; eBioscience), PECy5-conjugated anti-F4/80 (1:400) (15-4801-80, clone BM8; Invitrogen), and FITC-conjugated anti-Ly6G (1:800) (551460, clone 1A8; BD Biosciences) for 30 min in the dark at 4 °C. Excess antibodies were washed off from cells with FACS wash and centrifugation at 1,600 rpm for 5 min. Cells were resuspended in 500 µL wash, and the immune cell population was quantified using a NovoCyte Quanteon 4025 flow cytometer. Appropriate fluorescence-minus-one control mice were used to identify positive populations, and Compensation control mice (Comp Bead; Invitrogen; 01-2222-42) were included in the experimental setup to exclude spectral overlap. For inhibitor studies, mice were injected with 50 mg/kg of CTPI-2 (MedChemExpress; 68003-38-3) at 12 h and 18 h post–zymosan injection, and intraperitoneal lavage was collected over a time course and analyzed for immune cell population and distribution via flow cytometry. Data analysis was performed using NovoExpress (ACEA Biosciences).

Spectral flow cytometric metabolic and inflammation profiling

After zymosan injection, peritoneal lavage was collected after 6 d, as described previously. If required, the cells were treated with ACK lysis buffer to lyse the red blood cells, washed, and counted. The 27-color/29-parameter panel to look at the metabolic and inflammatory markers was adapted from.⁴⁷ Antibodies used for spectral flow cytometry are listed in Table 1. Abcam lightning-link kits were used to conjugate the intracellular targets. For each experiment, 2 × 10⁶ cells per sample were seeded in the 96-well V-bottom plate (Corning; 3894). Cells were washed with DPBS (21-031-CV) and first stained with Zombie Ultraviolet fixable viability dye (1:1,000; BioLegend; 423107) for 30 min at room temperature to exclude the dead cells. Fc receptors were then blocked by incubating cells in TruStain Fcx plus Fc block (1:100; BioLegend; 156604) in FACS buffer (DPBS containing 2% FBS and 2 mM EDTA) at 4 °C for 10 min. The surface staining was carried out in FACS buffer for 30 min at 4 °C using anti-mouse antibodies against CD11c, CD206, CX3CR1, CD36, and CD19. Following surface staining, cells were fixed with eBioscience Foxp3 fixation/permeabilization staining kit (Invitrogen; 00-5523-00) for 30 min at 4 °C, permeabilized in 1× permeabilization buffer overnight, and Fc-blocked for 15 min at 4 °C. Intracellular staining was done in 1× permeabilization buffer for 1 h at 4 °C using anti-mouse antibodies against Glut1, PKM, succinate dehydrogenase (SDHA), CPT1A, ACC1, cytochrome C (CytC), and G6PD. After intracellular staining, cells were washed twice with permeabilization buffer and once with FACS buffer. The remaining surface targets were stained in FACS buffer for 30 min at 4 °C (surface stain 2, Table S1). Brilliant stain buffer plus (BD Biosciences; 566385) and monocyte blocker (BioLegend; 426103) were added to the staining buffer whenever required. Single-color (using both cells and ultracomp beads), unstained, and fluorescence-minus-one control were prepared for each staining experiment. Finally, cells were suspended in 200 mL of FACS buffer and acquired on Bigfoot 5-laser Spectral Cell Sorter (Invitrogen). Acquired samples were unmixed using Bigfoot Sasquatch software and analyzed with FlowJo software v10.9 (TreeStar).

High-dimensional analysis was done using the OMIQ software from Dotmatics. Initial gating was done using FlowJo; this involved selecting leukocytes, removing doublets, and removing dead cells. The cell population of interest was identified and gated as CD45⁺ CD11b⁺ F4/80⁺ (CD3⁻ CD19⁻ Ly6G⁻). This gated cell population was downsampled to 90,000 cells per sample. After downsampling, FCS files were exported and uploaded to the OMIQ platform. First, the parameters were scaled (scaling type = Arcsinh, cofactor = 400, minimum = −800, maximum = 100,000). Uniform Manifold Approximation and Projection was run using the default settings (Euclidean distance function, nearest neighbors = 15, and minimum distance = 0.4). It was followed by phenograph clustering using the Euclidean distance function and K = 100. OMIQ was also used to calculate the median fluorescent intensity. Graphs were made in Prism v10.2.0 (GraphPad Software).

Lipidomics assay

A total of 1 × 10⁶ BMDMs from lipin-1^mKO mice and littermate control mice were seeded in 6-well tissue culture–treated plates. Three technical replicates were included per individual experiment. Macrophages were incubated for 3 h and subsequently stimulated with 40 ng/mL IL-4 for 4 h. After IL-4 stimulation, culture media was aspirated, and each well was washed with 1 mL of room temperature 0.9% sodium chloride solution (B. Braun Medical; R5201-01). Saline solution was aspirated, and 400 µL of ice-cold liquid chromatography–mass spectrometry (LC-MS) grade methanol (Infinity Lab Ultrapure; Agilent Technologies; 5191-4497) was added to each well and allowed to sit for 5 min, after which 400 µL of LC-MS grade water (Pierce Water; Thermo Fisher Scientific; 51140) was added. Cells were scraped and transferred to a labeled polypropylene microfuge tube (Agilent Technologies; tube: 5191-8150, caps: 5191-8151). Samples were stored at −80 °C before shipment to the Metabolomics core at the Rocky Mountain National Lab.

A total of 400 µL of ice-cold LC-grade chloroform (Fisher Chemical; C607-4) was added to each sample. Samples were shaken for 30 min at 4 °C and centrifuged at 16,000 × g for 20 min. A total of 400 µL of the bottom (organic) layer was collected and dried in a Savant SpeedVac SPD130 (Thermo Fisher Scientific). Lipids were resuspended in 1 mL of 5 µg/mL butylated hydroxytoluene in 6:1 isopropanol:methanol (isopropanol [Fisher Chemical; A461-4], methanol [Fisher Chemical; A456-4], butylated hydroxytoluene [MP Biomedicals; 0210116280]).

Bulk lipids were analyzed as previously described.⁴⁸ Samples were separated using a Shimadzu Nexera LC-20ADXR HPLC and a Waters XBridge Amide column (3.5 µm, 3 mm × 100 mm). Lipids were separated by headgroup with a 12-min binary gradient from 100% 5 mM ammonium acetate, 5% water in acetonitrile (pH_apparent 8.4) to 95% 5 mM ammonium acetate, 50% water in acetonitrile (pH_apparent 8.0) (water [Fisher Chemical; W64], acetonitrile [Fisher Chemical^; A996-4], ammonium acetate [Fisher Chemical; A11450]). Lipids were detected using a Sciex 6500+ QTRAP mass spectrometer with polarity flipping and scheduled MRMs.

All signals were integrated using MultiQuant Software 3.0.3 (Sciex). Signals with greater than 50% missing values or an intensity of <3,000 units were discarded, and the remaining missing values were replaced with the group average of 3,000 units. All signals with a quality control coefficient of variance >40% were discarded. A total of 731 individual lipid species remained in the final dataset. Data was the total sum normalized before analysis. Single and multivariate analysis was performed in MarkerView Software 1.3.1 (Sciex).

FFA assay

A total of 3 × 10⁻⁶ BMDMs were seeded on 6 well plates, allowed to sit for 2 h, and were pretreated with increasing concentrations (1 mM, 3 mM, 6 mM) of sodium citrate tribasic dihydrate (Sigma-Aldrich; 71402–250G) for 2 h before 4-h cotreatment with 40 ng/mL IL-4 concentrations. Cells were collected in a tube, and free fatty acid (FFA) estimation was performed as described by the kit (Sigma-Aldrich; MAK466-1KT).

Mitochondrial bioenergetics

A total of 1 × 10⁵ BMDMs were seeded in Agilent Seahorse XF24 Cell Culture Microplate (Agilent Technologies; 100777-004) and treated as required by experiments. After treatment, culture media was aspirated and replaced with warm Seahorse assay media containing Agilent Seahorse XF DMEM Medium, pH 7.4 (103575-100); 1 mM pyruvate (Agilent; 103578-100); 2 mM GlutaMAX (Gibco; 35050-061); and 10 mM glucose solution (Agilent; 103577-100). Cells were incubated in a no-CO₂ incubator for 45 min, and the oxygen consumption rate was measured by Agilent Seahorse XFe24 Analyzer (Agilent Technologies). Concentrations of the mitochondrial stress assay inhibitors are as follows: 1 µM oligomycin (Sigma-Aldrich; 75351-5MG), 2 µM FCCP (Sigma-Aldrich; SML2959-1ML), 1 µM rotenone (Sigma-Aldrich; R8875-1G), and antimycin (Sigma-Aldrich; A8674-25MG).

Western blotting

Cells were lysed with 1× NuPAGE lysis buffer containing 1× Halt protease inhibitor (Thermo Fisher Scientific; 18612793 + 9), 1× phosphatase inhibitor cocktail 2 (Sigma-Aldrich; P5726-5ML), 1× phosphatase inhibitor 3 cocktail (Sigma-Aldrich; P0044-5ML), and 100 mM dithiothreitol (VWR; 0281-25G). Pierce 660 nm Protein Assay (Thermo Fisher Scientific; 22660) was used to determine the protein concentration of sonicated and heat-denatured protein samples. 10/20 µg of protein sample was loaded in the wells of precasted 4% to 12% polyacrylamide NuPAGE Novex gel (Invitrogen; NP0322BOX) submerged in 1× MOPs running buffer. Protein bands were separated for 48 min at 200 volts and 400 mAMPs. Subsequently, protein bands within the gel were transferred on an Immobilon-FL transfer membrane (Millipore; IPFL00010) for 45 min at 20 volts and 400 mAMPs. Membranes were blocked with PBS intercept buffer (Li-cor; 927-70001) for 1 h at room temperature. Membranes were incubated overnight with appropriate primary antibodies. Following primary antibody incubation, membranes were washed trice with 1× Tris-buffered saline with Tween 20 (TBST) before incubation with secondary antibody in 1× TBST containing 5% nonfat dry milk Omniblok (AmericanBio; AB10109-01000) and 0.01% sodium dodecyl sulfate (Thermo Fisher Scientific; 28365). Membranes were washed trice with 1× TBST before 1 min incubation with ImmunoCruz Western blotting luminol reagent (Santa Cruz Biotechnology; sc-2048). The Amersham Imager 680 (GE Healthcare Bio-Sciences) was used to capture protein bands, and the ImageQuant TL 8.1 (Cytiva) was used to quantify protein bands. For histone, immunoblot 5 to 15 µg of protein sample was loaded in the wells (equally) in 15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis tris-glycine gels and run for 30 min at 200 V. Proteins were then transferred to nitrocellulose membranes (GVS NitroBind; EP2HY00010) for 1 h at 90 V. Blocking was performed using Odyssey Intercept blocking buffer (LI-COR; 92760001) for 1 h at room temperature. Primary antibody incubations were also performed using the Odyssey Intercept blocking buffer. After washing, membranes were incubated with a 1:15,000 dilution of LI-COR secondary antibodies. Membranes were washed four times in 1× TBST. Membranes were stored in 1× Tris-buffered saline, imaged using an Odyssey infrared imaging system, and quantified using Bio-Rad Image Lab.

Western blot antibodies

Primary antibodies used for western blotting were phospho-ACC (D7D11; Cell Signaling Technology), ACC (3662S; Cell Signaling Technology), P-AMPKalpha (T172; Cell Signaling Technology), AMPKalpha (2532S; Cell Signaling Technology), lipin-1 (D2W9G; Cell Signaling Technology), CIC/CPT/SLC25AI (15235-I-AP; Proteintech), AceCS1 (D19C6; Cell Signaling Technology), fatty acid synthase (FAS) (C20G5; Cell Signaling Technology), GAPDH (14C10; Cell Signaling Technology), actin (A2066-2ML; Sigma-Aldrich), pan H3 (05-928; Sigma-Aldrich), pan H4 (05-858; Sigma-Aldrich), acetyl H3 (06-599; EMD Millipore), and acetyl H4 (06-866; Sigma-Aldrich). The working stock of all primary antibodies was made at 1:1,000 dilution, except GAPDH and actin, which were made at 1:5,000 and 1:2,000 dilution, respectively. Secondary antibodies used were goat anti-rabbit IgG (111-035-003; Jackson ImmunoResearch; 1:2,000) and goat anti-mouse (610-1319; Rockland; 1:2,000).

Statistical analysis

GraphPad Prism 9 (GraphPad Software) was used for statistical analyses. We carried out a test for normal distribution using the D'Agostino and Pearson test and Shapiro-Wilk test. For comparison between two datasets that passed the normality test, we used Student’s t test analysis. A 1-way analysis of variance was used to analyze grouped datasets, while a 2-way analysis of variance was used to analyze grouped data sets with >2 technical replicates averaged as single values. Data were presented as mean ± SEM, and statistical significance was assigned at P < 0.05. Details of the statistical analysis of individual experiments are included in the figure legend.

Results

Myeloid-associated lipin-1 promotes inflammation resolution

We have previously demonstrated that lipin-1 contributes to macrophage efferocytosis,²⁴ and deletion of myeloid-associated lipin-1 delayed full skin wound closure⁴⁴ and increased atherosclerosis progression.³⁸ Collectively, our previous studies suggest that macrophage-associated lipin-1 is involved in inflammation resolution. We used a zymosan-induced peritonitis inflammation model (Fig. 1A) to investigate the contribution of myeloid-associated lipin-1 to the resolution of inflammation. Mice lacking lipin-1 in myeloid cells (lipin-1^mKO) and their littermate control mice were intraperitoneally injected with 0.1 mg zymosan to induce peritonitis. Clearance of neutrophils after the onset of zymosan-induced inflammation is a marker for inflammation resolution.⁵ We quantified neutrophils (CD45⁺ CD11b⁺ Ly6g⁺) and monocytes/macrophages (CD45⁺ CD11b⁺ Ly6g⁻ F4/80⁺) numbers at 4, 12, 24, and 48 h by collecting peritoneal cells by lavage, staining them, and performing flow cytometry (Fig. S1A). These time points correspond to initial acute recruitment (4 h and 12 h) and the subsequent clearance (24 and 48 h) of polymorphonuclear leukocytes (PMNs) that align with inflammation and inflammation resolution.^49,50 We observed a rapid influx of leukocytes after 4 h of injection (Fig. 1B–D), with the number of neutrophils peaking at 12 h after injection (Fig. 1C). There was a delay in the clearance of neutrophils within the lipin-1^mKO mice compared with WT mice, suggesting a defect in inflammation resolution with 10-h increase in resolution index compared with WT mice (Fig. 1D). We observed no significant difference in the number of recruited monocytes/macrophages, suggesting that the difference in PMN numbers is due to a defect in the ability of lipin-1^mKO macrophages to clear PMNs, rather than a defect in monocyte recruitment in lipin-1^mKO mice, or an increased number of macrophages in WT mice (Fig. 1E). We further examined if a difference in the number of PMNs was observed in lipin-1 enzymatic knockout (KO) mice. In contrast, lipin-1 enzymatic KO mice had fewer PMNs than control mice at 24 h post–zymosan injection (Fig. S1B–D), suggesting that lipin-1 contribution to inflammation resolution is independent of the enzymatic activity.

Figure 1.

Loss of myeloid lipin-1 delays inflammation resolution. (A) Lipin-1^mKO mice and littermate control mice were subjected to zymosan challenge (0.1 mg/mouse). PMNs and macrophages were quantified from the peritoneal cavity by flow cytometry. (B) Total number of cells isolated from the peritoneal cavity. (C) Total number of PMNs isolated from the peritoneal cavity. Resolution interval (Ri) was determined. (D) Total number of macrophages isolated from the peritoneal cavity. The illustration in panel A was created using BioRender.com. A minimum of 3 mice per group per time point were analyzed. Values are mean ± SEM. Unpaired 2-tailed t tests were performed between groups at each time point. *P ≤ 0.05. MΦ, macrophage.

Lipin-1 promotes a proresolution metabolic phenotype and attenuates inflammation in vivo

The metabolic status of macrophages is critical to their ability to perform proper efferocytosis, which is crucial for inflammation resolution.^22,24,50 We have previously demonstrated in vitro that the loss of lipin-1 increased glycolysis, broke the TCA cycle with increased citrate and isocitrate, and reduced oxidative phosphorylation.²⁴ We sought to determine if similar metabolic alterations were seen in vivo. We chose to investigate macrophages at 6 d post–zymosan injection, as this corresponds to a postresolution phase that is enriched in proresolving macrophages, minimizing confounding issues of ongoing inflammatory responses, and a time point in which we previously demonstrated that peritoneal macrophages (pMACs) are defective for efferocytosis.⁵¹ We performed a high-dimensional, flow cytometric analysis of immune cell populations and metabolic targets that correlated with the metabolic state of the cells isolated from the peritoneal cavity during zymosan-induced peritonitis.⁴⁷ We isolated cells from the peritoneal cavity by lavage.⁴⁷ See Table S1 for antibody targets. No significant change in macrophage (Lin⁻ CD11b⁺ F4/80⁺ CD45⁺) percentage or number was observed (Fig. 2A). In lipin-1^mKO pMACs, there was a significant increase in the median fluorescence intensity (MFI) of Glut1 (Fig. 2B), suggesting a potential glycolysis increase supported by our previous study.²⁴ We also report a significant decrease in MFI of CytC and SDHA, components of the electron transport chain and TCA cycle (Fig. 2B) in lipin-1–deficient pMACs (Fig. 2B). Low levels of CytC and SDHA suggest reduced mitochondria content, impaired mitochondria function, and a defective TCA cycle. These data indicate a reduction in oxidative metabolic capacity in these macrophages, a phenotype we previously observed in lipin-1 KO macrophages in vitro.²⁴ We observe a significant decrease in Siglec F and an increase in Ly6C and major histocompatibility complex class II (MHCII) staining, suggesting a more proinflammatory state of macrophages from lipin-1^mKO mice (Fig. 2C).

Figure 2.

Lipin-1 promotes proresolution phenotype in vivo. Lipin-1^mKO mice and littermate control mice were challenged with 0.1 mg zymosan, and 6 d later, peritoneal cells were isolated by lavage. Isolated cells were stained for flow cytometric analysis. (A) The number and ratio of CD11b⁺ F4/80⁺ lin⁻ cells in the peritoneal cavity. (B, C) MFI of metabolic and inflammatory markers of CD11b⁺ F4/80⁺ lin⁻ cells. (D) Uniform Manifold Approximation and Projection and phenograph clustering of macrophages and those enriched in WT (orange) and lipin-1^mKO (purple) mice using metabolic markers. (E) Heat map of percent population changes of phenoclusters between WT and lipin-1^mKO mice and min-max heat map of metabolic and inflammatory markers from identified phenoclusters. (F) The number and ratio of CD11b⁺ F4/80⁺ lin⁻ cells by phenocluster type. (G, H) MFI of metabolic and inflammatory markers of CD11b⁺ F4/80⁺ lin⁻ cells from phenocluster 6 in WT mice (orange) and phenoclusters 9, 4, and 2 from lipin-1^mKO (purple) mice. n = 3 mice per group. Dots represent individual mice; lines are means. Unpaired 2-tailed t test was performed between groups to define significance *P ≤ 0.05. MΦ, macrophage.

In vivo, macrophages are heterogeneous in metabolic and inflammatory phenotypes.⁴⁷ As such, we examined this heterogeneity by dimensional reduction and clustering of concatenated samples of Lin^- CD11b+ F4/80⁺ CD45 cells, concentrating on metabolic markers identified in Table S1. A total of 20 clusters were defined, and clusters that predominantly contained WT or KO were identified (Fig. 2D–F) and used for further analysis. Cluster 6 primarily comprised WT macrophages, while clusters 9, 4, and 2 predominantly comprised lipin-1 KO macrophages (Fig. 2D–F). We want to emphasize that macrophages from clusters 6, 9, 4, and 2 are present in both WT and KO mice. However, to make the data easier to understand, we are showing the MFIs for WT mice in cluster 6 and the MFIs for KO mice in clusters 9, 4, and 2. This approach highlights the differences in the phenotypes of these clusters. We chose this method because the MFI values for clusters 6, 9, 4, and 2 macrophages are the same in both WT and KO mice. Analysis of these clusters for their expression of metabolic markers showed that loss of lipin-1 leads to an increase in a population of macrophages with reduced in CPT1A protein, a reduction in surface expression of CD36, a scavenger receptor known to mediate AC clearance, and a significant decrease in CytC and SDHA, which are consistent with the global pMAC staining (Fig. 2G). CPT1a control mice the first committed step of β-oxidation, suggesting these macrophages may have reduced β-oxidation capacity, a phenotype that we observed in vitro.²⁴ Additionally, the decreased CytC and SDHA would indicate a broken TCA cycle and could lead to increased citrate and isocitrate, observations we have seen of in vitro lipin-1 KO macrophages stimulated with IL-4.²⁴ In sum, these metabolic phenotypes of macrophages in vivo align well with our previous in vitro data.

We also looked at the inflammatory profile of the pMACs within these clusters. Our data suggest that there were lipin-1^mKO mice were enriched with macrophages with a more myeloid-derived inflammatory phenotype with an observed increase in MHCII, PDL2, Ly6C, CX3CR1, and CD64, the high-affinity IgG FC receptor (Fig. 2H). An increase in CX3CR1 and Ly6C, known to be high in bone marrow monocytes,^43,44 suggests ongoing inflammation and a high number of inflammatory macrophages recently generated from recruited inflammatory monocytes (Fig. 2H). Lipin-1–dependent increase a macrophage population with high surface expression of PDL2, which inhibits T cell proliferation and cytokine production,^52,53 suggests that lipin-1 promotes inflammation resolution and further supports the notion of ongoing inflammation in the lipin-1^mKO mice (Fig. 2H). In addition, the increase in macrophages with more MHCII expression suggests not only unresolved inflammation, but also defective efferocytosis, as efficient efferocytosis prevents the antigen presentation of efferocytotic cargo peptides, and one such mechanism is via the degradation of efferocytotic cargo peptides in such a manner that they become unsuitable for antigen presentation by MHCII.^3,54

The result from the lipin-1^mEnzyKO suggests that these phenotypic changes in metabolism and inflammatory profile are independent of the enzymatic activity of lipin-1 as we observe no differences in their metabolic and inflammatory markers staining compared with WT (Fig. S2). Our current data align with what we have previously demonstrated in vitro, suggesting that the metabolic defects observed in lipin-1 deficient macrophages are likely responsible for the defects in inflammation resolution.

Lipin-1 regulates the production and channeling of lipids

Lipin-1 regulates lipid synthesis in several other cell types,^41,42 and both our previous²⁴ and current data (Fig. 2G) suggest that lipin-1 restrains lipid synthesis in macrophages in response to proresolving stimuli. We had previously demonstrated elevated citrate,²⁴ a precursor for FA synthesis, in lipin-1–deficient macrophages. We investigated the distribution and composition of lipids in IL-4–stimulated BMDMs from lipin-1^mKO mice and littermate control mice by LC tandem MS. Principal component analysis from our lipidomics assay showed a distinct difference in the lipid profile of WT and lipin-1^mKO BMDMs under unstimulated conditions that were largely maintained or exacerbated when stimulated with IL-4 (Fig. 3A). When lipid species are grouped by family, we observed alterations in the lipid composition of lipin-1 KO macrophages (Fig. 3B–E). There was an increase in FA levels at both baseline and in response to IL-4 in the lipin-1 KO macrophages (Fig. 3B, C, and E). Though our data do not demonstrate an IL-4–dependent increase in FA levels in macrophages deficient in lipin-1, the observed IL-4–dependent increase in the levels of triacylglycerol (TAG), diacylglycerol (DAG), cholesterol esters, and several phospholipids (Fig. 3B, C, and E) supports that excess FAs accumulating in response to IL-4 in lipin-1 KO BMDMs are being channeled for storage. We also confirmed this buildup of lipid storage species by Nile red staining (Fig. 3F). In addition to lipid storage in triglycerides, sphingolipid biosynthesis responds to excess FFAs. The buildup of bioactive sphingolipids such as ceramides induces cellular dysfunction and impairs proresolving macrophage function.^55–58 Our lipidomic analysis of macrophages deficient in lipin-1 shows a trending rise in total ceramides and an IL-4–dependent increase in dihydroceramide, suggesting an increase in de novo sphingolipid biosynthesis (Fig. 3B–E).

Figure 3.

Loss of lipin-1 results in an increase in FAs, neutral lipids, and ceramides. Lipids harvested from IL-4–stimulated (40 ng/mL, 4 h) BMDMs from lipin-1^mKO mice and littermate control mice were processed via LC-MS for lipidomics analysis. (A) Principal component analysis showing the spatial distribution of lipids within each condition. (B) Heat map of most changed lipid species relative to untreated cells (n = 6). A 3% false discovery rate correction of Student’s t test analysis between fold change adjusted IL-4–stimulated WT and KO values was used to populate lipids in panel B. (C) Heat map of lipid species as a family (n = 6). Lipids that are represented by family in panel C include all respective species that passed the 3,000-unit cutoff, regardless of statistical significance. (D) Quantified signals of lipid species from the ceramide and dihydroceramide family (n = 6). (E) Quantification of lipid families in panel C. (F) Nile red staining images of IL-4–stimulated (40 ng/mL, 4 h) BMDMs. Heatmaps were made by a log2 fold change analysis from WT untreated mice. Significance was determined by 1-way analysis of variance. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.001. Cer, ceramide; DihCer, dihydroceramide; GlcCer, glucosylceramide; LPC, lysophosphatidylcholine; LPE, lysophosphatidylethanolamine; LPG, lysophosphatidylglycerol; LPI, lysophosphatidylinositol; LPS, lysophosphatidylserine; MAG, monoacylglycerol; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PE O, ether-linked phosphatidylethanolamine; PE P, phosphatidylethanolamine plasmalogen; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SM, sphingomyelin; UNT, untreated.

Lipin-1 restrains FA synthesis in response to proresolving stimuli

The buildup of FAs and neutral lipids observed in IL-4–stimulated lipin-1 KO macrophages might result from increased lipid uptake, reduced lipid catabolism, or de novo FA synthesis (Fig. 4A). Though we have previously shown that macrophage lipid uptake is independent of lipin-1,²⁴ the loss of lipin-1 could be causing an increase in the expression of enzymes involved in FA biosynthesis or activation of FAs. We quantified the amount of these enzymes using Western blot analysis. We observe no difference in cytoplasmic acetyl-CoA synthetase (AceCS1) protein levels at baseline and in response to IL-4 (Fig. 4B). Citrate is the first metabolite in FFA biosynthesis, where it is transported from the mitochondria by the CTP or CIC (Fig. 4A). We observe no difference in the protein levels of CTP, and we also do not see differences in the protein levels of FAS, which catalyzes palmitate biosynthesis from malonyl-CoA (Fig. 4B). These observations suggest that lipid biosynthesis may be upregulated without significantly increasing enzyme abundance, potentially by post-translational regulatory mechanism.

Figure 4.

Lipin-1 regulates FA biosynthesis at the ACC checkpoint during IL-4 stimulation. (A) Illustration of the de novo FA biosynthesis pathway (B, C) Protein was isolated from IL-4–stimulated (40 ng/mL, 4 h) BMDMs from lipin-1^mKO mice and littermate control mice, and protein abundance of FA biosynthesis enzymes and effectors was quantified by Western blot analysis (n > 3). (D) Representative Western blot images of IL-4–stimulated lipin-1^mKO mice and littermate control BMDMs, quantified protein abundance of histone proteins, and post-translational modifications (n = 4). The illustration in panel A was created using BioRender.com. Bars represent SEM. Significance was determined by 1-way analysis of variance (fold change) and paired Student’s t test (ratio). *P ≤ 0.05; **P ≤ 0.01. UNT, untreated.

Acetyl-CoA carboxylase (ACC) conversion of acetyl-CoA to malonyl-CoA is a rate-limiting step of FA synthesis. ACC is negatively regulated by adenosine monophosphate–activated protein kinase (AMPK) phosphorylation at serine 79.⁵⁹ In response to IL-4, phosphorylation of AMPK and ACC is increased in WT BMDMs but not in lipin-1 KO BMDMs (Fig. 4C). We also observed similar ACC phosphorylation results when using ACs as an alternative proresolving stimulus (Fig. S3A). Lipin-1 KO BMDMs are missing both lipin-1 activities; we performed the same experiment using BMDMs from KO mice lacking enzymatic activity but with preserved transcriptional regulatory function to determine which lipin-1 activity was responsible for the observed phenotype. We observed an increase in ACC and AMPK phosphorylation in both WT and lipin-1 Enzymatic KO macrophages, suggesting that lipin-1 nonenzymatic activity is responsible for the decreased phosphorylation of ACC and AMPK in lipin-1^mKO macrophages (Fig. S3B).

Previously, we reported an IL-4–dependent decrease in acetyl-CoA in lipin-1 KO BMDMs.²⁴ Acetyl-CoA can be used in numerous processes, including de novo FA biosynthesis or epigenetic modifications, which contribute significantly to macrophage biology and function.¹⁷ We propose that the previously reported elevated citrate leads to increased production of acetyl-CoA, which is utilized for lipid synthesis. However, it is possible that acetyl-CoA is being used up for epigenetic modifications. We report no differences in histone acetylation between WT and lipin-1–deficient macrophages (Fig. 4D). This further supports a more causative effect of increased acetyl-CoA utilization for FA synthesis. Collectively, loss of lipin-1 does not necessarily lead to an upregulation of the FA biosynthesis pathway, but rather a failure to restrain FA synthesis at the CIC and ACC checkpoints.

Inhibition of the CIC restores efferocytic function

We previously demonstrated that lipin-1 was required for efficient efferocytosis²⁴ but only after macrophages encountered the first AC. IL-4 can enhance macrophage efferocytosis.⁴⁴ We have shown that lipin-1 contributes to numerous IL-4–mediated responses in macrophages, including increased β-oxidation, FA metabolism, and proresolving gene expression.^24,44 We aimed to investigate if IL-4–enhanced efferocytosis is also defective in lipin-1 KO BMDMs. After 6 h of IL-4 treatment, we observed a defect in AC uptake by lipin-1 KO BMDMs (Fig. 5A), consistent with our previous data in which lipin-1^mKO were defective in IL-4–enhanced phagocytosis of zymosan particles.⁴⁴ These results suggest that lipin-1 is required for augmenting efferocytic capacity in response to proresolving stimuli.

Figure 5.

Inhibition of the CIC and de novo ceramide biosynthesis restores efferocytic capacity in lipin-1^mko macrophages. (A) BMDMs from lipin-1^mKO mice and littermate control mice were stimulated with IL-4 for 6 h and then subsequently challenged with CFSE-labeled ACs. (B) Illustration of the de novo FA biosynthesis pathway and checkpoints of inhibition. (C) Lipin-1^mKO mice and littermate control BMDMs were treated with 10 µM inhibitors (FAS: C75; ACC: Cpd9; CTP: CTPi) and dimethyl sulfoxide vehicle (VEH) for 12 h and subsequently cotreated with 40 ng/mL IL-4 and 10 µM of respective inhibitors for 6 h. Images were taken at 20× and zoomed at 1.5× to generate representative images. The experiment was done twice, each time with 3 unique pairs of individual WT and full KO BMDMs (n = 6). At least 3 random images of each group were taken, quantified (C), and grouped to give individual dots. Individual experiments incorporated all inhibitors, so each inhibitor group had the same vehicle control group. (D–E) BMDMs from lipin-1^mKO mice and littermate control mice were treated with increasing concentrations of SCT (1 mM, 3 mM, 6 mM) for 2 h before cotreatment with IL-4 for 6 h (D) and 4 h (E). Subsequently, in vitro efferocytosis (D) and FFA estimation (E) were carried out. (F) In vitro efferocytosis with 10 µM Myriocin. Lipin-1^mKO mice and littermate control BMDMs were treated with Myriocin for 2 h and cotreated with Myriocin and IL-4 for 6 h before the AC challenge. The illustration in panel B was created using BioRender.com. Bars represent SEM. Significance was determined by 2-way analysis of variance. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.001. MYR, Myriocin; UNT, untreated.

We sought to determine whether alterations in FFA metabolism in lipin-1 KO BMDMs were responsible for inhibiting efferocytosis capacity. Acetyl-CoA is the precursor for all FFA synthesis (Fig. 4A). Acetyl-CoA is converted into malonyl-CoA by ACC1/2, and then malonyl-CoA and acetyl-CoA are used by FAS to generate palmitate (Fig. 4A). To fix the efferocytosis defect in lipin-1 KO macrophages, we attempted to inhibit FA synthesis using c75, a synthetic FAS inhibitor (Fig. 5B). However, inhibiting FAS did not reverse the efferocytosis defect in lipin-1 KO macrophages (Fig. 5C). Interestingly, FAS inhibition impaired the efferocytic capacity of WT macrophages, indicating that basal production of FAs is required for efferocytosis (Fig. 5C). Inhibition of ACC with Cpd9 showed similar trends in inhibiting efferocytosis in WT macrophages and a failure to restore efferocytic capacity in lipin-1 KO BMDMs (Fig. 5C). We previously demonstrated that loss of lipin-1 led to increased amounts of citrate in IL-4–stimulated BMDMs.²⁴ Citrate can activate and fuel the FA synthesis pathway; therefore, we inhibited the CTP/CIC to prevent citrate influx into the FA biosynthesis pathway. CIC inhibition restored efferocytosis capacity in lipin-1 KO BMDMs without impairing WT macrophages (Fig. 5C). ACC can use either acetate or citrate-derived acetyl-CoA for FA biosynthesis (Fig. 5B). Thus, inhibition of the citrate transporter with CTPi prevents the supply of citrate-derived acetyl-CoA while still allowing the cell to use other acetyl-CoA sources. On the other hand, ACC, or FAS inhibition, completely blocks FA synthesis, which is necessary for membrane remodeling during efferocytosis (Fig. 5B). We suggest that CTPi-mediated restoration of efferocytosis in lipin-1–deficient macrophages was due to the prevention of excessive citrate-mediated FA synthesis. Our data suggest that WT BMDMs do not have increased citrate-mediated FFA synthesis, leading to excessive FFAs. To further demonstrate that impaired efferocytosis in lipin-1 KO macrophages is due to excess citrate export that can be utilized for FA synthesis, we treated WT macrophages with increasing amounts of exogenous citrate that have been previously shown to augment proinflammatory responses in lipopolysaccharide-stimulated monocytes,⁴⁶ and increase lipid synthesis.⁶⁰ Our data show that exogenous citrate impairs the efferocytosis function of WT macrophages in a dose-dependent manner (Fig. 5D). Additionally, the addition of sodium citrate caused a dose-dependent increase in FFAs (Fig. 5E), which strongly suggests that excessive citrate-induced FA synthesis inhibits efferocytosis in lipin-1^mKO BMDMs.

Excess FFAs can be toxic to the cell, and when present in large amounts they are directed toward the de novo sphingolipid pathway.^61,62 This pathway leads to the production of ceramides, which impair the function of mitochondria, proresolving macrophages, and efferocytosis.^55,57,58 Our results indicate that the excessive synthesis of FAs is directed toward the de novo sphingolipid pathway. To address this, we inhibited serine palmitoyltransferase (SPT), the enzyme responsible for the first step of ceramide biosynthesis. SPT catalyzes the reaction between palmitoyl-CoA and L-serine to produce ceramides. Inhibiting SPT with myriocin restored the efferocytic capacity of lipin-1 KO BMDMs (Fig. 5F). Therefore, our data suggest that the buildup of ceramides due to increased FA synthesis is responsible for the impairment of efferocytosis in lipin-1 KO BMDMs.

CIC inhibition depletes neutral lipids without restoring mitochondria function

β-oxidation and oxidative metabolism have been attributed to promoting efferocytosis. In contrast, the contribution of the anabolic arm of lipid metabolism has been less studied.^22,24 Excessive FA-induced ceramide production can impair macrophage efferocytosis function.^57,58 Several studies have shown that ceramides inhibit mitochondrial function and oxidative phosphorylation,^55,63 therefore, we hypothesized that CIC inhibition in macrophages promoted mitochondrial function. Consistent with previous studies where inhibition of CIC impaired mitochondrial respiration,⁶⁴ we report that CIC inhibition failed to restore the mitochondrial respiratory capacity in lipin-1 KO macrophages (Fig. 6A). Because CIC inhibition restored efferocytosis in lipin-1 KO macrophages without restoring mitochondrial respiration, this suggests that mitochondrial defect in lipin-1 KO macrophages is upstream of lipid synthesis.

Figure 6.

Inhibition of CIC reduced neutral lipids levels without restoring mitochondria respiratory capacity. (A) BMDMs isolated from lipin-1^mKO mice and littermate control mice were treated with CTPi for 12 h before cotreatment with CTPi and 40 ng/mL IL-4 for 4 h. Oxygen consumption rate (OCR) and mitochondrial function parameters were analyzed via Seahorse extracellular flux analyzer. Graphed data represent mean OCR with SEM (n = 4) (B) Lipin-1^mKO mice and littermate control BMDMs were treated with 10 µM inhibitors (FAS: C75; ACC: Cpd9; CTP: CTPi) and dimethyl sulfoxide vehicle (VEH) for 12 h and subsequently cotreated with 40 ng/mL IL-4 and 10 µM of respective inhibitors for 4 h before Nile red staining. Images were taken at 10× to generate representative images. The experiment was done twice, each time with 2 unique pairs of individual WT and full KO BMDMs (n = 4). At least 3 random images of each group were taken, quantified, and grouped to give individual dots. Bars represent SEM. Significance was determined by 1- and 2-way analysis of variance for Seahorse data and Nile red values, respectively. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. ATP, adenosine triphosphate; ns, not significant.

Because inhibiting the cytoplasmic citrate influx should reduce FA biosynthesis and downstream lipids, we treated WT and lipin-1 KO macrophages with lipid synthesis inhibitors and stained them for neutral lipids. As expected, we observed a significant decrease in neutral lipids in lipin-1 KO macrophages with CTP, C75, and Cpd9 inhibition (Fig. 6B). Even though WT macrophages start with reduced amounts of neutral lipids, similar trends were observed in WT macrophages except for C75, which only showed a decreasing trend in neutral lipid levels (Fig. 6B). Though we hypothesized that there would be fewer lipids with FAS and ACC inhibition, we acknowledge the limitation of Nile red staining, as it does not detect global lipids and cannot differentiate between intracellular and membrane-neutral lipid composition.

CIC activity is responsible for defects in inflammation resolution

We demonstrated that loss of lipin-1 impairs inflammation resolution. Because efficient efferocytosis is critical for the resolution of inflammation,⁵ and the inhibition of the CIC restored efferocytosis in macrophages deficient in lipin-1, we hypothesized that inhibition of the CIC in lipin-1^mKO mice could restore inflammation resolution. Similar to the experiment in Fig. 1A, we used the zymosan-induced peritonitis model. Following injection with zymosan, lipin-1^mKO mice and littermate control mice were injected with 50 mg/kg of CTPI-2, a more potent and in vivo suited second-generation CTP inhibitor at 12 and 18 h. Intraperitoneal lavage was collected at several different time points for flow cytometric analysis (Fig. 7A). CTP-2 restored inflammation resolution in lipin-1^mKO mice (Fig. 7B), consistent with the restoration of efferocytosis via CIC inhibition. Collectively, lipin-1 restrains FA synthesis to promote proresolving macrophage efferocytosis, which is required to resolve inflammation.

Figure 7.

Inhibition of the CIC restores inflammation resolution in lipin-1^mKO mice. (A) Lipin-1^mKO mice and littermate control mice were injected with 50 mg/kg CTP-2 at 12 h and 18 h post–zymosan injection. Intraperitoneal (I.P.) lavage was collected over a time course for flow cytometric analysis. (B) Time course flow cytometry analysis of I.P. lavage after acute (0.1 mg/mouse) zymosan injection. Illustrations in panels A and C were created using BioRender.com. Bars represent SEM. Significance was determined by 1-way analysis of variance. Ri, resolution interval.

Discussion

Unresolved inflammation promotes the progression of numerous disease conditions.^1–3 Macrophages facilitate disease resolution via efferocytosis and by secretion of anti-inflammatory mediators to restore tissue homeostasis.^22,65 The metabolism of macrophages influences their activation state and function; thus, to promote inflammation resolution, macrophages require alignment of metabolism with their activities.^6,7 Several aspects of lipid metabolism, including FA oxidation and biosynthesis, contribute to macrophage polarization and function^18,19,24,66 Lipin-1 is crucial to regulating lipid metabolism. We have previously demonstrated that lipin-1 promotes disease resolution, as loss of myeloid-associated lipin-1 delayed wound healing and worsened atherosclerosis.^38,44 In this study, we used lipin-1^mKO mice and the lipin-1^mEnzyKO mouse to investigate the contribution of lipin-1 to inflammation resolution and further understand how lipin-1 regulation of macrophage metabolism supports proresolving macrophage responses. We provide evidence that lipin-1 restrains FA synthesis to promote efferocytosis and inflammation resolution.

After an acute inflammatory response, macrophages must polarize into a proresolving phenotype to dampen the ongoing inflammation.⁵ To this effect, proresolving macrophages alter their metabolism to support inflammation resolution processes, such as producing anti-inflammatory mediators and efferocytosis.^3,6,7 In this study, we show that myeloid-associated lipin-1 promotes inflammation resolution by facilitating the clearance of neutrophils in a zymosan model of peritonitis. As macrophages must attain a metabolic profile that supports proresolving macrophage functions, we investigated the contribution of lipin-1 to the metabolic profile of macrophages during the resolution phase of inflammation. We report that loss of lipin-1 in pMACs promotes a metabolic profile similar to a proinflammatory macrophage rather than a proresolving macrophage.^11,12,17 We observed a population of macrophages that were more prevalent in lipin-1^mKO mice with decreased key metabolic markers such as CD36, CPT1A, CytC, and SDHA. The low levels of CPT1A, CytC, and SDHA in pMAC clusters unique to lipin-1^mKO mice suggest an impairment in the TCA cycle and mitochondria homeostasis. CD36 is a scavenger receptor known to recognize ACs.^67,68 CD36 and the platelet-activating factor receptor are proposed to promote AC recognition and engulfment.⁶⁸ Stimulation of macrophages with ACs was also shown to promote CD36 and platelet-activating factor receptor interaction at the macrophage plasma membrane with the possible coupled function of forming lipid rafts that facilitate efferocytosis.⁶⁸ Though Parks et al.⁶⁷ showed that the resolution of 24-h zymosan-induced peritonitis is independent of CD36 activity, loss of CD36 leads to a delay in inflammation resolution typified by a 7-d postinjury increase in neutrophils, ACs, and macrophage numbers in a murine model of bleomycin-induced lung injury.⁶⁷ We attribute the impairment in inflammation resolution in lipin-1^mKO mice to defects in proresolving macrophage functions. This observation is consistent with other studies correlating defects in proresolving macrophage functions to the progression of chronic inflammatory diseases such as atherosclerosis and myocardial infarction.^22,50,65 In addition, the altered metabolic profiles of lipin-1–deficient pMACs are consistent with our previous in vitro studies, in which we reported a dysfunctional TCA cycle and impaired mitochondrial function in macrophages deficient in lipin-1 when stimulated with IL-4.²⁴ Altogether, our data suggests that lipin-1 supports proper proresolving macrophage metabolism to promote inflammation resolution.

We investigated the lipid profile in macrophages in response to proresolving stimuli. Results from the lipidomics assay position lipin-1 as a significant regulator of macrophage lipid metabolism, as the loss of lipin-1 led to significant changes in the lipidomic profile of macrophages. The IL-4–dependent increase in neutral lipids and ceramides observed in macrophages deficient in lipin-1 suggests altered lipid channeling.^61,62 Upon saturation of lipid stores, FAs are channeled into the sphingolipid biosynthesis pathway to contribute to plasma membrane integrity and other cellular functions.⁶¹ However, owing to their bioactive and cytotoxic properties, ceramides are produced minimally as needed or used to make inert membrane sphingolipids.⁵⁶ Ceramides impair efferocytosis, reduce the respiratory chain efficiency, inhibit M2 polarization, and are highly enriched in atherosclerotic plaques.^{55,57,58,69,70} Notably, de novo synthesis of ceramides impairs efferocytosis by altering actin polymerization via inhibition of Rac1 plasma membrane recruitment, which decreases membrane ruffle formation.⁵⁸ Efferocytosis-induced Rac1 activation promotes the formation and stabilization of the phagocytic cup, which supports the uptake of multiple ACs.⁶⁵ Together with our data in which myriocin treatment restored efferocytosis in lipin-1 KO macrophages, we further provide evidence of the detrimental nature of ceramides on efferocytosis and inflammation resolution and the importance of lipin-1 in proper lipid channeling.

Upon the export of citrate into the cytoplasm, a series of enzyme-catalyzed reactions led to the synthesis of FAs. Our data investigating the FA biosynthesis pathway suggests that the basal expression of enzymes involved in the FA biosynthesis pathway is sufficient to handle a high substrate influx because we do not observe significant differences in the abundance of CIC and FAS. As the notable increase in lipid species is in specific lipid families and not globally, we believe the differences in the lipidome profile in macrophages deficient in lipin-1 result from impaired lipid handling. Furthermore, we believe the increase in FA and other neutral lipids that we observe in lipin-1 KO macrophages is not due to persistent activation of de novo FA biosynthesis but possibly because of increased cytoplasmic citrate export and failure to restrain FA biosynthesis at key checkpoints, particularly at the rate-limiting step. This might explain why we see no difference in the protein levels of FAS. In addition, citrate is a known allosteric activator of ACC, and we have previously reported an increase in citrate in macrophages deficient in lipin-1. While one would expect the inverse outcome, that is, more buildup of lipids in WT macrophages, as WT macrophages retain the enzymatic activity of lipin-1 that can dephosphorylate phosphatidic acid to make DAG and TAG further downstream, there is the possibility that lipin-1 KO macrophages use monoacylglycerol acyltransferase and possibly, other novel enzymes to generate DAG. In addition, other studies have shown that loss of lipin-1 leads to increased lipid synthesis.^{35–37,71,72} Collectively, lipin-1 reduces FA synthesis as a proresolving mechanism that supports efferocytosis.

Lipid metabolism is self-regulating, and a negative correlation exists between lipid catabolism and anabolism.⁷³ Efficient β-oxidation has been shown to promote efferocytosis.^22,24 Thus, inhibiting FA synthesis may restore efferocytosis. We demonstrated that CIC inhibition restored efferocytic capacity in lipin-1 KO macrophages and reduced the amount of neutral lipids. Additionally, CIC inhibition restored inflammation resolution in lipin-1^mKO mice. CIC contributes to other inflammatory cardiometabolic disease as well, as CIC levels are high in nonalcoholic steatohepatitis livers,⁷⁴ and in a model of murine nonalcoholic fatty liver disease, CIC inhibition reverted hepatic steatohepatitis, normalized glucose intolerance, and reduced inflammation with subsequent induction of an anti-inflammatory response.⁷⁴ Inhibition of CIC also reduced liver messenger RNA and protein levels of ACC, SREBP, and FAS.⁷⁴ Consistent with our findings, Tan et al.⁷⁴ also showed that loss of CIC led to decreased liver fat, palmitate, monoglycerides, DAGs, and TAGs. Other studies have also reported the inflammatory role of the mitochondrial CIC and its contribution to altered lipid biosynthesis.^13,46,75 In macrophages, the CIC is required for inflammatory cytokine induction of nitric oxide and prostaglandins.¹³ CIC-mediated cytokine, nitric oxide, and reactive oxygen species production have also been implicated in the pathophysiology of COVID-19.⁷⁶ Thus, lipin-1–mediated inhibition of CIC activity to reduce FFA synthesis allows for proper efferocytic function.

Over the years, much emphasis has been placed on the role of β-oxidation in supporting proresolving macrophage function and efferocytosis. It is important to realize that β-oxidation and FA biosynthesis are interconnected and that lipid metabolism is self-regulating because FA synthesis inhibits β-oxidation, and β-oxidation exhausts substrates required for FA synthesis.⁷³ Consequently, studies that have reported impairment in efferocytosis due to pharmacological inhibition or genetic ablation of essential β-oxidation proteins need to be open to the possibility that inhibition of β-oxidation increases the acetyl-CoA pool that becomes available for lipid biosynthesis. This buildup in lipids, particularly ceramides, could then be responsible for inhibiting efferocytosis, which is usually attributed to β-oxidation impairment. At the same time, in studies in which β-oxidation agonists have been used to promote efferocytosis, the underlying mechanism could result from decreased acetyl-CoA pool available for lipid biosynthesis due to increased β-oxidation efficiency. Consistent with this notion, inhibiting the mitochondrial CIC restored efferocytosis function in lipin-1 KO macrophages without restoring mitochondria function. In addition, CIC inhibition in cancer cells decreases the respiratory chain complex-I activity and induces membrane potential destabilization, mitochondria fragmentation, and depletion by mitophagy.^64,77 Thus, cell-type differences could exist because CIC inhibition does not impair WT macrophage mitochondrial function. Another possibility is that, in contrast to WT macrophages in our study, a more noticeable reduction would be observed in cancer cells’ mitochondria respiration because cancer cells already express a very high amount of CIC activity to support oxidative metabolism.^64,77 The inability of ACC and C75 inhibition to restore efferocytosis supports the notion that a basal amount of lipid biosynthesis is required for efferocytosis.^20,68 While our study suggests an inhibitory role of FA synthesis during inflammation resolution, FA synthesis is critical to effectively mount an effective antimicrobial or proinflammatory response.¹⁸ Notably, the loss of ACC in macrophages impaired the ability to mount a proinflammatory response and phagocytic clearance of bacteria, which exacerbated the outcome of murine bacteria peritonitis.⁷⁸

While the findings in this study are sufficient to make inferences and draw conclusions, we understand that this study has limitations. Because of their low signal ratio, certain lipid species were undetectable or excluded in our lipidomics assay. We also understand that our in vivo studies were done in transgenic mice lacking full-length lipin-1 or the enzymatic domain of lipin-1 in myeloid cells. Thus, there might be other contributory aspects of other cells of myeloid origin. With this in mind, we demonstrate macrophage-specific differences throughout this study. While the efficacy of the mitochondrial CIC has been shown in cancer and nonalcoholic steatohepatitis studies.^64,74 Our future direction will be to investigate the effect of the CIC on the amelioration of atherosclerosis and other cardiometabolic diseases by employing pharmacological and genetic depletion/ablation experimental strategies.

In this study, we have shown the role of lipin-1 in restraining FA and resolving inflammation. Based on our previous work that lipin-1 promoted oxidative metabolism and efferocytosis independent of its enzymatic activity, it may be that lipin-1 restrains FA and resolves inflammation independent of enzymatic activity as well; however, future studies will need to be conducted.²⁴ Aside from its enzymatic and transcriptional regulatory activity, lipin-1 can be associated with other proteins and on membranes and diverse organelles^27,79,80 Lipin-1 could also have a scaffolding or tethering function,^27,80,81 to facilitate lipid handling and anaplerosis. This idea will also form the basis of our future investigations to delineate if a lipin-1 nonenzymatic activity is responsible for our current findings. In conclusion, the ability to restore efferocytosis and inflammation resolution independently of improved mitochondria function with CIC inhibition suggests that impaired mitochondria function in lipin-1 KO macrophages is upstream of altered lipid metabolism and possibly that the toxic downstream effect of FA biosynthesis is more crucial to the efferocytic process than efficient mitochondria function. Our findings strongly suggest that in response to proresolving stimuli, lipin-1 restrains FA synthesis, thus preventing the downstream production of ceramides that impair proresolving macrophage function and activities required for inflammation resolution (Fig. 7C).

Author contributions

Conceptualization: T.T.B., R.M.S., O.O.I., M.D.W. Data curation: T.T.B., R.M.S., O.O.I., E.H.N., S.B., B.S., M.D.W. Methodology: T.T.B., R.M.S., O.O.I., E.H.N., M.B., S.B., C.D., B.S., E.B., R.S.S., A.Y., M.D.W. Investigation: T.T.B., R.M.S., O.O.I., E.H.N., M.B., S.B., B.S., E.B., M.D.W. Formal analysis: T.T.B., R.M.S., O.O.I., E.H.N., B.S., M.D.W. Visualization: T.T.B., O.O.I., S.B., B.S., M.D.W. Validation: T.T.B., S.B., B.S., M.D.W. Supervision: M.D.W. Resources: C.M.B., R.S.S., B.N.F., A.Y., M.D.W. Writing – original draft: T.T.B., R.M.S., M.D.W. Writing – review and editing: T.T.B., R.M.S., E.H.N., C.D., B.S., C.M.B., B.N.F., R.S.S., A.Y., M.D.W. Project administration: M.D.W.

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

This work was supported by the following grants: National Institutes of Health grants R01HL163106 (M.D.W.), P20GM134974 (M.D.W., R.S.S., S.B.), R01HL131844 (M.D.W.), R00HL145131 (A.Y.), R01HL167758 (A.Y.), and R01HL119225 (B.N.F.); an Ike Muslow Predoctoral Fellowship Intramural Award (T.T.B.); and the Intramural Research Program of the National Institutes of Health National Institute of Allergy and Infectious Diseases (C.M.B.).

Conflicts of interest

The authors declare that they have no competing interests that could have created bias or influenced the authenticity of the research findings.

Data availability

All data needed to deduce conclusions from this research article are included in the main manuscript and supplemental materials.