Oxidized Low-Density Lipoprotein Accumulation in Macrophages Impairs Lipopolysaccharide-Induced Activation of AKT2, ATP Citrate Lyase, Acetyl–Coenzyme A Production, and Inflammatory Gene H3K27 Acetylation

Abstract The accumulation of lipid and the formation of macrophage foam cells is a hallmark of atherosclerosis, a chronic inflammatory disease. To better understand the role of macrophage lipid accumulation in inflammation during atherogenesis, we studied early molecular events that follow the accumulation of oxidized low-density lipoprotein (oxLDL) in cultured mouse macrophages. We previously showed that oxLDL accumulation downregulates the inflammatory response in conjunction with downregulation of late-phase glycolysis. In this study, we show that within hours after LPS stimulation, macrophages with accumulated oxLDL maintain early-phase glycolysis but selectively downregulate activation of AKT2, one of three AKT isoforms. The inhibition of AKT2 activation reduced LPS-induced ATP citrate lyase activation, acetyl-CoA production, and acetylation of histone 3 lysine 27 (H3K27ac) in certain inflammatory gene promoters. In contrast to oxLDL, multiple early LPS-induced signaling pathways were inhibited in macrophages with accumulated cholesterol, including TBK1, AKT1, AKT2, MAPK, and NF-κB, and early-phase glycolysis. The selective inhibition of LPS-induced AKT2 activation was dependent on the generation of mitochondrial oxygen radicals during the accumulation of oxLDL in macrophages prior to LPS stimulation. This is consistent with increased oxidative phosphorylation, fatty acid synthesis, and oxidation pathways found by comparative transcriptomic analyses of oxLDL-loaded versus control macrophages. Our study shows a functional connection between oxLDL accumulation, inactivation of AKT2, and the inhibition of certain inflammatory genes through epigenetic changes that occur soon after LPS stimulation, independent of early-phase glycolysis.


INTRODUCTION
Atherosclerosis is a chronic inflammatory disease that underlies ischemic conditions such as myocardial infarction and stroke.The accumulation of cholesterol-rich plaques in the intimal layer of large or medium-size arteries is a characteristic feature of atherosclerosis, and macrophage foam cells constitute a major component of these atherosclerotic plaques.In general, foam cells are formed through intracellular accumulation of modified forms of low-density lipoprotein (LDL) particles, including oxidized LDL (oxLDL) (1).oxLDL can be taken up by macrophages (Mus) through scavenger receptors, such as CD36, Msr1, and SRA (2,3).Following uptake, oxLDL particles are then processed in lysosomes, and esterified cholesterol is stored in lipid droplets (4).We have previously shown that the accumulation of oxLDL or cholesterol in cultured Mus impaired their inflammatory gene expression in response to TLR agonists (5).This unexpected observation was supported by other studies including recent transcriptomics analyses of atheromas from mice and humans (6,7) suggesting that dampening of inflammatory gene expression in foam cells is an adaptation in atherosclerosis plaques.
Past studies from the immunometabolism field have comprehensively demonstrated that metabolic reprogramming from oxidative phosphorylation to glycolysis is critical for orchestrating the inflammatory response in myeloid cells (8).Specifically, myeloid cells undergo early and late phases of glycolytic reprogramming to support their inflammatory functions (9,10).The LPS-induced early phase of glycolysis (prior to 4 h) is dependent on early TLR4 signaling, whereas the late phase (after 4 h) is dependent on the stabilization and transcriptional activity of HIF-1a for the expression of IL-1b and other genes associated with inflammation and glycolysis (11,12).Recently, our study showed that oxLDL accumulation suppresses late-phase glycolysis and dampens macrophage inflammatory gene expression (13).However, how oxLDL accumulation regulates the early phase of glycolysis in Mus remains to be determined.
One of the early TLR4 signaling events is the activation of the phosphoinositide-dependent serine/threonine protein kinase AKT, also known as protein kinase B. Classical activation of AKT depends on high-affinity binding of its pleckstrin homology domain to the membrane-anchored phosphatidylinositol (3,4,5)trisphosphate (PIP 3 ), which is generated from phosphatidylinositol (4,5)-bisphosphate (PIP 2 ) by class I PI3Ks (14,15).Membranebound AKT is activated once it is fully phosphorylated at Thr 308 in the kinase domain by phophoinositol-dependent kinase 1, and at Ser 473 in the C-terminal domain by mTOR complex 2 (16,17).Phosphatase and tensin homolog (PTEN) is a phosphatase that dephosphorylates PIP 3 into PIP 2 , and consequently suppresses AKT activation (18).Conversely, activation of the PI3K/AKT pathway can induce PTEN ubiquitination and degradation (19).However, in the context of TLR4 signaling, AKT has been shown to be activated by TBK1/IKKe, two closely related kinases downstream of various TLRs (9,20).LPS-induced AKT activation promotes the translocation of hexokinase II to the mitochondrial membrane, which enhances glucose retention (9).LPS-induced AKT activation also activates ATP citrate lyase (ACLY), which converts the TCA cycle metabolite citrate into acetyl-CoA, thereby promoting fatty acid synthesis and histone acetylation (10).
To date, three AKT isoforms have been identified, and as revealed by genetic deficient models, each isoform plays nonredundant roles in the progression of atherosclerosis.For instance, AKT1 deficiency in Akt1 À/À Apoe À/À mice led to severe atherosclerosis that was attributed to reduced levels of NO and endothelial cell viability in atherosclerosissusceptible areas (21).In contrast, hematopoietic AKT2 deficiency in Ldlr À/À mice resulted in smaller plaques with less inflammation and promoted M2 Mu polarization, suggesting a proinflammatory function for AKT2 in Mus (22).Finally, AKT3 deficiency in hypercholesterolemic Apoe À/À mice or in the hematopoietic compartment led to increased foam cell and lesion formation that was associated with increased modified LDL uptake and cholesterol esterification (23).In summary, the functions of AKT1 and AKT3 are atheroprotective, whereas AKT2 promotes inflammation and lesion formation.
In this study, we show that oxLDL accumulation in Mus impaired LPS-induced activation of AKT2 and ACLY, the production of acetyl-CoA, and the acetylation of histone 3 lysine 27 (H3K27) in inflammatory gene promoters, all within 3 h of LPS stimulation.The inhibition of AKT2 activity was dependent on oxLDL-induced mitochondrial oxidative stress, and the effects of oxLDL appeared to be uncoupled from TLR4-mediated signaling and induction of early-phase glycolysis.These finding are distinct from cholesterol accumulation in Mus, which inhibited multiple LPS-induced signaling pathways and early-phase glycolysis.

Mouse strains
Eight-to 12-wk-old mice were used.C57BL/6J (strain no.000664), B6.129P2-Lyz2 tm1(cre)Ifo /J (strain no.004781), and B6.129S4-Pten tm1Hwu /J mice (strain no.006440) were purchased from The Jackson Laboratory.Lyz2-Cre:Pten fl/fl mice were generated by backcrossing Lyz2-Cre into Pten fl/fl mice.Breeding for experiments consisted of crosses between Cre-positive and Cre-negative Pten fl/fl mice.All mice were maintained in a pathogen-free, temperature-regulated environment with a 12-h light/12-h dark cycle and were fed a normal chow diet (16 kcal % fat).All studies were performed under the approval of Animal User Protocols by the Animal Care Committee at the University Health Network according to the guidelines of the Canadian Council on Animal Care.Littermates were used for experiments (see below).

Thioglycolate-elicited peritoneal Mu isolation
For each experiment, one or two litters of male and female wild-type C57BL/6J mice 812 wk of age (up to 10 mice in total) were injected i.p. with 1 ml of 4% aged thioglycolate (Thermo Fisher Scientific, no.211716) and peritoneal Mus (PMus) were harvested after 4 d by lavage with cold PBS containing 2% FBS.Cells were pooled, counted, divided into four experimental groups, and cultured (37 C, 5% CO 2 ) in DMEM supplemented with 10% FBS, 2 mM L-glutamine, and 10,000 U/ml penicillin/ streptomycin.Nonadherent cells were washed away and adherent PMus were used in experiments after 18 h.

Bone marrowderived Mu generation
Bone marrowderived Mus (BMDMus) were obtained from Cre 1 and Cre À Pten fl/fl mice.Sex-matched littermates of up to five mice per experiment (Cre 1 and Cre À ) were used.Mice were euthanized in a CO 2 chamber and bone marrow cells were isolated from leg bones.Cells were cultured (37 C, 5% CO 2 ) in RPMI 1640 supplemented with 10% FBS, 2 mM L-glutamine, 10,000 U/ml penicillin/streptomycin, and 40 ng/ml M-CSF (Pepro-Tech, no.AF-315-02) for 7 d.Cells were counted and replated for experiments.Male and female mice were used in different experiments.The results were comparable and therefore the data were combined.

Quantification of metabolites
To quantify acetyl-CoA, an acetyl-CoA assay kit (Sigma-Aldrich, no.MAK039A) was used.In brief, 1 × 10 6 PMus were plated in each well of a 24-well plate per experimental condition, after which 150 ml of acetyl-CoA assay buffer was first added per well followed by mechanical scraping of cells.Cell lysates were then used to quantify the amount of acetyl-CoA by following the manufacturers instructions.To quantify intracellular lactate, an L-lactate assay kit (Cayman Chemical, no.700510) was used.In brief, 2 × 10 6 PMus were plated in each well of a 12-well plate per experimental condition.After treatment, cells were washed with cold PBS three times, and lactate assay buffer was added to each well followed by mechanical scraping of cells.Cell lysates were then used to quantify the amount of intracellular lactate by following the manufacturers instructions.To quantify extracellular lactate, an L-lactate assay kit I (Eton Bioscience, no.1200012002) was used.In brief, 2 × 10 6 PMus were plated in each well of a 12-well plate per experimental condition.To reduce background, PMus were cultured in phenol redfree Opti-MEM (Thermo Fisher Scientific, no.11058021) during treatment.After treatment, media were filtered using 10-kDa filtered columns (Abcam, no.ab93349).Filtered media were then used to quantify the amount of extracellular lactate by following the manufacturers instructions.
Bulk RNA sequencing library preparation Total RNA was extracted from 2 × 10 6 PMus using an RNeasy mini kit (Qiagen, no.74104).To eliminate genomic DNA, 20 U of Roche RNase-free DNase I (Roche, no.4716728001) was added in the eluant (100 ml), followed by 20 min of incubation at 37 C. Samples were then passed through RNeasy columns.Total RNA was quantified by a NanoDrop ND-1000 spectrophotometer, and 750 ng was used to enrich mRNA using an NEBNext poly(A) mRNA magnetic isolation module (New England Biolabs, no.E7490).Library construction followed immediately using an NEBNext Ultra II directional RNA library prep kit (New England Biolabs, no.E7760) and NEBNext multiple oligonucleotides for Illumina (New England Biolabs, nos.E7335, E7500, E7710, E7730).Libraries were PCR amplified for 10 cycles, quantified by MiSeq Nano v2, and sequenced on an Illumina NovaSeq S1 flow cell with a 150-bp paired-end run to obtain $30 million reads per sample.The complete RNA sequencing (RNA-seq) dataset has been deposited to the GEO repository (GSE239696; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE239696).

Bulk RNA-seq differential expression and pathway analysis
The DESeq2 R/Bioconductor package was used to perform hierarchical cluster analysis, principal component analysis, and to determine differential expression between Mus with and without oxLDL.Differential expression was assessed using the Wald test with a BenjaminiHochberg correction; genes with an adjusted p value of <0.1 were considered significantly differentially expressed.EnhancedVolcano (version 1.11.3) was used to create volcano plots.Genes were preranked using the DESeq2 output (Àlog 10 (p value) × sign(log fold change)) for enrichment analysis.Pathway enrichment was performed using gene set enrichment analysis software (version 4.1.0)from the Broad Institute (https:// software.broadinstitute.org/cancer/software/gsea/wiki/index.php/Main_Page).

Statistical analysis
All of the statistical details of experiments can be found in the figure legends.In brief, all figures show pooled data from independent experiments.All experiments were repeated at least three times.The number of biological replicates is listed as the n value.Statistical analyses were performed using Prism software (9.2.0) unless otherwise specified in the figure legends.

Both oxLDL and cholesterol accumulation in PMus impair LPS-induced AKT signaling, whereas early glycolysis is inhibited only by cholesterol
Prior to investigating the effects of oxLDL and cholesterol accumulation on LPS-induced AKT signaling and early glycolytic reprogramming in PMus, we confirmed that LPS stimulation of PMus activates AKT and that inhibition of AKT with MK-2206, an allosteric pan-AKT inhibitor, reduces the uptake of 2-NBDG, a fluorescent glucose analog (Supplemental Fig. 1A).Having established that LPS signaling induces AKT-dependent glucose uptake, we investigated AKT activation in PMus with accumulated oxLDL or cholesterol.We observed diminished phosphorylation of AKT residues Ser 473 and Thr 308 (Fig. 1A, 1B), although in some experiments, inhibition of AKT phosphorylation was observed at earlier time points.These observations demonstrate that LPS-induced AKT activation is suppressed in PMus with accumulated oxLDL or cholesterol.In support of this conclusion, the phosphorylation of known AKT targets, such as PRAS40 and GSK3a/b, was also suppressed in PMus cultured with oxLDL or cholesterol (Supplemental Fig. 1B, 1C).We next used the Seahorse bioanalyzer to measure the extracellular acidification rate (ECAR) and evaluate early glycolysis.Glycolysis stress tests were performed using control or lipid-loaded PMfs that were unstimulated or stimulated for 3 h with LPS.As expected, LPS stimulation of control PMus (without accumulated lipid) increased the ECAR upon the addition of glucose and oligomycin (Fig. 1C, 1D, left panels), consistent with an early glycolytic response.When PMfs were not stimulated with LPS, the ECAR was elevated relative to controls when PMfs with accumulated oxLDL, but not cholesterol, were assayed (Fig. 1C, 1D, middle panels).Three hours after LPS stimulation, the ECAR was reduced relative to control when PMfs with accumulated cholesterol were assayed (Fig. 1D, right panel), but it was comparable to control when PMfs with accumulated oxLDL were assayed (Fig. 1C, right panel).Consistent with the ECAR, extracellular lactate, a metabolite generated by glycolysis, increased 3 h after LPS stimulation in control PMus (Fig. 1E, 1F).Notably, extracellular lactate was comparable when PMus with accumulated oxLDL were stimulated with LPS for 3 h (Fig. 1E), but it was significantly reduced in cultures of PMus with accumulated cholesterol (Fig. 1F), consistent with the Seahorse data.Intracellular levels of lactate in PMus with accumulated cholesterol followed a similar pattern to extracellular levels (Supplemental Fig. 1D).The Seahorse and lactate data were also supported by reduced accumulation of the glucose analog 2-NBDG following LPS stimulation of PMus with accumulated cholesterol (Fig. 1G).In contrast to the 3 h time point, extracellular lactate was reduced when PMus with accumulated oxLDL were stimulated with LPS for 6 h (Supplemental Fig. 1E), consistent with our previous study that focused on HIF-1adependent glycolysis (13).Collectively, we demonstrated that lipid accumulation in PMus impairs LPS-induced AKT signaling; however, LPS-induced early glycolysis remains intact in PMus with accumulated oxLDL but is suppressed in cells with accumulated cholesterol.oxLDL accumulation in PMus selectively inhibits LPS-induced AKT2 and ACLY activation Similar to lactate, acetyl-CoA is a product of glycolysis.We determined acetyl-CoA levels at 3 h after LPS stimulation and found reduced abundance in PMus with accumulated oxLDL (Fig. 2A).This result suggests that a pathway for acetyl-CoA generation other than glycolysis is inhibited in PMus with accumulated oxLDL.One possibility is ACLY, an enzyme that functions in de novo lipogenesis by converting citrate to acetyl-CoA (24).We thus evaluated an established marker of ACLY activationthe phosphorylation of Ser 455 , which is a residue that is phosphorylated by AKT (25).Following LPS stimulation, ACLY Ser 455 phosphorylation was significantly lower in PMus with accumulated oxLDL (Fig. 2B).This implies that the inhibition of AKT signaling in cells with accumulated oxLDL inhibits ACLY activation and production of acetyl-CoA.
To gain further insights into differences in PMu biology in the setting of oxLDL versus cholesterol accumulation, we evaluated the activation of the three AKT isoforms.LPS stimulation of control PMus (without lipid accumulation) increased the activation of AKT1 (phosphorylation of Ser 473 ) and AKT2 (phosphorylation of Ser 474 ) in a time-dependent manner, but not of AKT3 (phosphorylation of Ser 472 ) (Fig. 2C).In PMus with accumulated oxLDL, AKT1 and AKT3 activation was comparable to control, but AKT2 activation was significantly reduced (Fig. 2C).In contrast, cholesterol accumulation in PMus significantly reduced both AKT1 and AKT2 activation but not AKT3 (Fig. 2C).oxLDL accumulation selectively inhibited AKT2 activation in the RAW264.7 murine Mu cell line, although the overall kinetics of AKT activation after LPS stimulation were distinct in these cells (Supplemental Fig. 2A).AKT2 activation in PMus induced by other TLR agonists, including CpG-ODN (TLR9 agonist) and Pam 3 CSK 4 (TLR1/2 agonist), was also selectively inhibited by oxLDL accumulation (Supplemental Fig. 2B).

AKT2 inhibition reduces LPS-induced ACLY activation and acetyl-CoA, but not early glycolysis
The selective inhibitory effect of oxLDL accumulation on AKT2 activation is of interest due to its established role in promoting inflammation in Mus (22).A relatively specific AKT2 inhibitor was employed to explore the potential role of AKT2 in inducing the early glycolytic response and ACLY activation in LPS-stimulated PMus.Initially, we determined concentrations that selectively inhibited AKT2 activation in PMus 1 h after LPS stimulation (Fig. 2D).We then performed glycolysis stress tests using the Seahorse bioanalyzer to assess whether AKT2 regulates early glycolysis 3 h after LPS stimulation.We found that the AKT2 inhibitor (used at 3 mM) increased the ECAR when added to unstimulated PMus and did not reduce the ECAR when added to LPS-stimulated PMus (Fig. 2E).However, the AKT2 inhibitor reduced LPS-induced ACLY activation in a timedependent manner (Fig. 2F) and reduced the abundance of acetyl-CoA (Fig. 2G).This phenotype closely resembled PMus with accumulated oxLDL (Figs.1C, 2A, 2B).
oxLDL accumulation selectively and irreversibly blocks AKT2 activation, whereas cholesterol accumulation inhibits LPS-induced TBK1 activation Recent studies have shown that TBK1/IKKe activate AKT and regulate AKT-dependent glycolytic reprogramming upon LPS stimulation of myeloid cells (9).We therefore evaluated the phosphorylation of TBK1 on Thr 172 in LPS-stimulated PMus and found that LPS-induced TBK1 phosphorylation was comparable in cells with accumulated oxLDL; however, it was reduced in cells with accumulated cholesterol (Fig. 3A).These data suggest that signaling upstream of AKT is inhibited in PMus with accumulated cholesterol but not in cells with accumulated oxLDL.Nevertheless, oxLDL selectively inhibited AKT2 activation (Fig. 2C).
The activation of AKT isoforms requires localization to the plasma membrane through binding of an AKT pleckstrin homology domain to PIP 3 .Plasma membrane localization of AKT can be increased by inhibiting PTEN, which converts PIP 3 to PIP 2 by dephosphorylating the 3 0 phosphate.We obtained bisperoxovanadium (HOpic), a potent PTEN inhibitor (PTENi) and performed a dose-response experiment to determine that 10 mM is the optimal concentration (Supplemental Fig. 3A).After PMus accumulated oxLDL or cholesterol, cells were treated with PTENi for 1 h prior to LPS stimulation.In PMus with accumulated oxLDL, LPS-induced activation of both AKT2 and ACLY remained suppressed when cells were treated with PTENi (Fig. 3B).Treatment with PTENi rescued LPS-induced activation of both AKT and ACLY, which was inhibited by cholesterol accumulation (Fig. 3C).
Collectively, we demonstrated that the accumulation of oxLDL in PMus does not inhibit LPS-induced TBK1 activation; nevertheless, AKT2 and ACLY activation is suppressed and cannot be rescued by inhibition of PTEN.This suggests that the impaired activation of AKT2 is unlikely due to defective upstream kinase signaling cascades, as oxLDL in PMus does not block most TLR4-induced signaling pathways, including TBK1 (Fig. 3A), MAPK, and NF-kB (5).In contrast, the accumulation of cholesterol in PMus inhibits LPS-induced TBK1 activation (Fig. 3A), which accounts for suppressed AKT and ACLY activation, and the restoration of their inhibition by inhibiting PTEN (Fig. 3C).This suggests that cholesterol loading interferes with phosphorylation cascades upstream of AKT2.

Mitochondrial oxidative stress in response to oxLDL accumulation inactivates AKT2
In the absence of LPS stimulation, treatment of control PMus with the PTENi increased the activation of AKT1 and AKT2 but not AKT3 (Supplemental Fig. 3B).PTENi-induced activation of AKT2, but not AKT1, was inhibited in PMus with accumulated oxLDL (Supplemental Fig. 3B).These observations together with the differential effects of PTENi in LPS-stimulated cells with accumulated oxLDL versus cholesterol (Fig. 3B, 3C) suggested the possibility that AKT2 is intrinsically deactivated by oxLDL loading.
Past research has shown that the activation of AKT2 can be regulated by oxidative stress (26,27).Because we have previously showed that oxLDL accumulation in PMus induces oxidative stress (13), this suggests the possibility that oxLDL-induced ROS deactivate AKT2.To test this, we first explored the potential source of ROS by comparing the transcriptomes of PMus with or without oxLDL accumulation with RNA-seq.Principal component analysis showed that oxLDL did not induce wide-scale transcriptome alterations, yet the expression of certain metabolic genes such as CD36 was induced significantly (Supplemental Fig. 3C, 3D).Hallmark pathway analyses revealed that oxLDL accumulation significantly altered intracellular metabolic processes, quantification of p-ACLY (Ser 455 ) and total ACLY in a LPS time course (0-3 h).PMus with or without oxLDL accumulation were used.For each time point, p-ACLY values were normalized to corresponding total ACLY and the 0 h LPS time point in the ÀoxLDL group (assigned a value of 1, n 5 5).including upregulation of oxidative phosphorylation, adipogenesis, and fatty acid metabolism and downregulation of cholesterol homeostasis and bile acid metabolism (Fig. 4A).Of particular interest is the increase in fatty acid oxidation, as it has been associated with impaired inflammatory responses in Mus (28).The breakdown of lipid droplets in lysosomes releases free fatty acids.We stained PMus with LysoTracker and found evidence of activated lysosomal function upon accumulation of oxLDL (Fig. 4B).We also measured by qPCR increased expression of genes that regulate fatty acid oxidation, such as Cd36, Lipa, Cpt1a, and Acaa2 (Fig. 4C).Because fatty acid oxidation can fuel the TCA cycle and generate energy in the mitochondria, we hypothesized that oxLDL-induced ROS are generated in mitochondria.Indeed, staining with MitoSOX showed increased mitochondrial ROS in PMus with accumulated oxLDL (Fig. 4D).Therefore, Mito-TEMPO, a mitochondrialtargeted antioxidant, was added 1 h prior to culture of PMus with oxLDL.This treatment with Mito-TEMPO did not alter oxLDL accumulation in PMus (Supplemental Fig. 4A) but it reduced mitochondrial ROS in PMus with accumulated oxLDL (Fig. 4E).Mito-TEMPO was then used to determine whether mitochondrial-derived ROS underlie the deactivation of AKT2 in PMus with accumulated oxLDL.Unfortunately, Mito-TEMPO inhibited LPS-induced activation of AKT2 (Supplemental Fig. 4B); therefore, we used PTENi to activate AKT2 and AKT1 (Supplemental Fig. 3B) and found that PTENi-induced activation of AKT2 and ACLY was restored by Mito-TEMPO in PMus with accumulated oxLDL (Fig. 4F).This finding is consistent with the possibility that generation of mitochondrial ROS in response to oxLDL accumulation modifies and inactivates AKT2.oxLDL accumulation in PMus suppresses inflammatory gene expression 3 h after LPS stimulation, and cholesterol accumulation inhibits LPS-induced signaling Previously, we showed that oxLDL accumulation in PMus does not inhibit LPS-induced MAPK or NF-kB signaling (5).We now show that the LPS-induced early glycolytic response also remains intact in PMus with accumulated oxLDL (Fig. 1).Considering this, we asked whether inflammatory gene expression is reduced at the 3 h time point after LPS stimulation.Indeed, we found the mRNA expression of many inflammatory genes was reduced significantly (Fig. 5A).For comparison, we also evaluated the effects of cholesterol accumulation and found a similar phenotype (Fig. 5B).Because we showed that cholesterol accumulation in PMus inhibits LPS-induced TBK1, AKT1, AKT2, and ACLY activation (Figs.13), we assessed whether MAPK and NF-kB signaling, which are critical for the induction of many inflammatory genes, is also inhibited by cholesterol.We found that LPS-induced phosphorylation of ERK1/2 and p65 (RelA) is suppressed in PMus with accumulated cholesterol, and nuclear translocation of p65 appears to be reduced at the 3 and 6 h time points (Fig. 5C).These data suggest that cholesterol accumulation in Mus inhibits multiple proximal signaling pathways triggered by TLR4 stimulation, including the MAPK, NF-kB, and the PI3K/AKT pathways, and this likely contributes to suppression of inflammatory gene expression.In contrast, oxLDL accumulation in PMus selectively inhibits AKT2 and ACLY but does not inhibit LPS-induced NF-kB or MAPK signaling or early glycolysis.Because inflammatory gene expression by PMus with accumulated oxLDL is suppressed 3 h after LPS stimulation (Fig. 5A), this implicates AKT2 and ACLY signaling in the inhibition of inflammation at this early time point.
oxLDL accumulation in PMus suppresses LPS-induced acetylation of H3K27 AKT contributes to the regulation of LPS-induced proinflammatory gene transcription through activation of ACLY and production of acetyl-CoA, which is required for protein acetylation, including K27 of histone H3 (H3K27ac) (10).Earlier, we showed that the accumulation of oxLDL in PMus selectively suppresses activation of AKT2, ACLY, and the production of acetyl-CoA (Fig. 2).We therefore performed ChIP experiments to investigate whether the accumulation of oxLDL suppresses H3K27ac in LPS-stimulated PMus.The initial time course showed that within 12 h LPS induces H3K27ac at promoters of proinflammatory genes but not Hprt, a constitutively expressed gene (Fig. 6A).Subsequent ChIP experiments demonstrated that oxLDL accumulation in PMus suppresses H3K27ac within promoters of several proinflammatory genes at 1 h (Fig. 6B) after LPS stimulation.
To investigate a causal relationship between the reduction in AKT2 activation by oxLDL accumulation and the reduction in H3K27ac of the proinflammatory genes in LPS-stimulated PMfs, we took advantage of BMDMus from mice with myeloid cell deletion of Pten (Lyz2-Cre:Pten fl/fl ), which express reduced levels of Pten mRNA compared with cells obtained from to Pten wild-type (Pten fl/fl ) littermates (Supplemental Fig. 4C).The uptake of DiI-labeled oxLDL was comparable in the two genotypes (Supplemental Fig. 4D); however, oxLDL accumulation in Lyz2-Cre:Pten fl/fl BMDMus did not inhibit LPS-induced activation of AKT2 and ACLY, in contrast to Pten fl/fl BMDMus (Supplemental Fig. 4E).As in wild-type PMus (Fig. 6B), oxLDL accumulation caused a reduction in H3K27ac in Pten fl/fl BMDMus 3 h after LPS stimulation (Fig. 6C).In contrast, oxLDL accumulation in Lyz2-Cre:Pten fl/fl BMDMus did not suppress H3K27ac within promoters of proinflammatory genes (Fig. 6D).ChIP experiments showed that treatment of PMus with the AKT2i suppressed H3K27ac (Fig. 6E), and this phenocopied the effects of oxLDL accumulation in wild-type Mus (Fig. 6B, 6C).

DISCUSSION
In atherosclerotic lesions, Mus take up cholesterol as cholesteryl esters in modified lipoprotein particles.This is modeled in cell culture by use of oxLDL, and scavenger receptors are critical to the internalization of oxLDL.Consistent with our past studies (5,13), we have cultured Mus with oxLDL and in parallel with free cholesterol to overcome the controversies related to the differences in oxLDL preparations (29) and potential signaling downstream of scavenger receptors.In fact, this parallel approach enabled us to discover how oxLDL and cholesterol accumulation differentially suppressed Mu inflammatory responses, one of which is the differential suppression of phosphorylation cascades downstream of TLR stimulation.For instance, whereas cholesterol accumulation in Mus suppressed LPS-induced AKT1 and AKT2 signaling, oxLDL accumulation selectively impaired LPS-induced AKT2 signaling (Fig. 2C).We hypothesized that this difference is attributed to the impairment of phosphorylation cascades upstream of AKT signaling by cholesterol, but not oxLDL accumulation.Indeed, we found that the activation of TBK1, an upstream kinase of AKT (9,20), was suppressed in PMus with accumulated cholesterol, but not oxLDL (Fig. 3A).The notion that cholesterol, but not oxLDL, accumulation could interfere with early phosphorylation cascades was further reinforced by experiments showing that inhibition of PTEN prior to LPS stimulation restored AKT activation in PMus with accumulated cholesterol (Fig. 3C), but not AKT2 activation in PMus with accumulated oxLDL (Fig. 3B).Indeed, this differential response was also found in other major LPS-induced inflammatory phosphorylation events.As shown in Fig. 5C, although cholesterol accumulation in PMus impaired LPS-induced NF-kB and ERK activation, they remained unaffected in PMus with accumulated oxLDL, as was shown in our past study (5).Taken together, our data demonstrate that cholesterol and oxLDL accumulation have differential effects on suppressing signaling cascades.We speculate that the inhibitory effect of cholesterol accumulation is due to a significant disruption of the cholesterol composition within the plasma membrane and lipid rafts (30).In contrast, the selective inhibitory effect of oxLDL accumulation is caused by generation of mitochondrial ROS, which modifies and inhibits AKT2.
Because phosphorylation cascades regulate downstream biological functions, it is possible that Mus with accumulated cholesterol versus oxLDL would exhibit different biological responses to inflammatory stimuli.Although both cholesterol and oxLDL accumulation impaired LPS-induced activation of AKT (Fig. 1A, 1B), a critical kinase with pleiotropic functions including early glycolytic influx, the induction of early-phase glycolysis remains intact in PMus with accumulated oxLDL (Fig. 1C).We attribute this to the selective impairment of AKT2 signaling and intact AKT1 activation.This result demonstrates that AKT2 activation is dispensable for LPS-induced early-phase glycolysis and was confirmed by use of a selective AKT2 inhibitor (Fig. 2E).Although AKT2 activation is not essential for glycolysis, we found that it is required for the activation of ACLY and consequently the production of acetyl-CoA (Fig. 2A, 2B, 2F, 2G).Recent research has demonstrated the importance of ACLY-derived acetyl-CoA in the regulation of Mu inflammatory gene expression through epigenetics (10).Specifically, ACLY-derived acetyl-CoA is critical for regulating H3K27ac of LPS-induced proinflammatory gene promoters (10).In our study, we have refined this concept by providing evidence that AKT2 regulates ACLY activity, acetyl-CoA levels (Fig. 2F, 2G), and H3K27ac of proinflammatory genes (Fig. 6E).More importantly, in Mus with accumulated oxLDL where AKT2 activation is selectively inhibited, ACLY activity and acetyl-CoA levels (Fig. 2A, 2B) and H3K27ac of proinflammatory genes were impaired (Fig. 6B, 6C).Collectively, our data suggest that oxLDL-mediated inhibition of early inflammatory responses is regulated by AKT2 and ACLY, instead of glycolytic enzymes.An important question that remains to be answered is whether oxLDL-induced reduction of H3K27ac is directly linked to the suppression of proinflammatory gene expression (Fig. 5A).Our previous study has shown that the binding of the prototypic p65 (RelA) subunit of NF-kB to the promoters of key proinflammatory genes, such as Ccl5 and Il6, was impaired in Mus with accumulated oxLDL, and that blocking histone deacetylase activity partially restored the expression of these genes (5).Because histone acetylation regulates the accessibility of transcription factors binding to promoters of target genes, this raises the possibility that suppression of H3K27ac by oxLDL accumulation (Fig. 6B) impairs the binding of p65 within 3 h of LPS stimulation.
Apart from p65, other lipid sensing transcription factors, such as liver X receptors (LXRs), peroxisome proliferator-activated receptors (PPARs), and retinoid X receptors (RXRs) have been implicated in mediating the suppressive inflammatory responses found in lipid loaded Mus.For instance, cholesterol loading-induced desmosterol accumulation was shown to activate LXRs and suppress inflammation in Mu foam cells (31).These findings were subsequently confirmed in foam cells derived from atherosclerotic lesions, as deletion of LXRs in myeloid cells accelerated atherosclerosis (32).Similar to LXRs, in vitro studies have shown that agonists of PPARs could block LPS-induced Mu activation and secretion of inflammatory cytokines due to inhibition of inflammatory transcription factor functions (33).The possible antiatherogenic role of PPARs was then confirmed with in vivo atherosclerotic mouse models (34).Finally, agonists for RXRs have also been shown to be critical for mediating the suppressive inflammatory responses in Mus, especially in chronic inflammatory disease models, such as atherosclerosis (35).Because RXRs have been recently shown to regulate the maintenance and identity of large PMus (36), it also raises the possibility that RXRs could mediate the reduced inflammatory responses that we have observed in oxLDL-loaded PMus.
We noted that oxLDL accumulation in PTEN-deficient Mfs did not suppress LPS-induced AKT2 and ACLY activation (Supplemental Fig. 4E), unlike the effects of PTENi that was added 1 h before LPS treatment and after oxLDL-induced ROS production (Figs.3B, 4D).These observations support the notion that oxLDL/ROS-mediated inhibition of AKT2 occurs prior to the addition of PTENi.In contrast, PTEN-deficient Lyz2-Cre: PTEN fl/fl Mfs have constitutively increased PIP 3 and AKT activation and appear to be resistant to oxidative stress induced by oxLDL accumulation.
In addition to the reduction of ACLY activation, we have also observed that oxLDL loading of LPS-activated Mfs led to the suppression of mTORC1 activation downstream of AKT2 signaling (Supplemental Fig. 1C).Past research has shown in adipocytes that AKT2, specifically its phosphorylation on Ser 474 , is critical for insulin-mediated mTORC1 activation (37).This raises the possibility that oxLDL-mediated inhibition of AKT2 led to the suppression of mTORC1 signaling cascades, including protein synthesis.Because the mRNA of HIF-1a is particularly sensitive to mTORC1 activation, as it harbors a 5 0 terminal oligopyrimidine tract (38), oxLDL-mediated inhibition of AKT2 activity may thus reduce the translation of HIF-1a mRNA, and potentially contribute to the impaired HIF-1a expression observed in Mus with accumulated oxLDL during the late phase of glycolytic reprogramming (13).
oxLDL-induced ROS production plays a critical role in shaping how foam cells respond to inflammatory stimuli.For instance, although others have shown that oxLDL-induced ROS is a rapid NADPH oxidasedependent event resulting from inflammatory signaling (39), we found in this study, as well as previously (13), that oxLDL-induced ROS production is a relatively late event.This suggests that the production of ROS induced by oxLDL accumulation is a consequence of a metabolic adaptation in the mitochondria.Indeed, our transcriptomic data and qPCR analysis demonstrate an upregulation of fatty acid oxidation in PMus with accumulated oxLDL (Fig. 4A, 4C).This is significant not only because it demonstrates that cells with lipid accumulation acquire a new metabolic adaptation, but more importantly, ROS generated as part of this adaptation may modulate the response upon subsequent stimulation with inflammatory agonists.For instance, we have previously shown that oxLDL-induced ROS primed an enhanced LPS-induced NRF2-dependent antioxidative response that suppressed HIF-1adependent late-phase glycolysis and inflammation (13).Now we show that oxLDLinduced mitochondrial ROS suppress LPS-induced activation of AKT2 and ACLY and thus reduce H3K27ac of inflammatory genes.Taken together, our previous and current studies collectively demonstrated that the generation of ROS in Mus with accumulated oxLDL plays an important role in modulating epigenetic and transcriptional responses to inflammatory stimuli.
AKT2 genetic deficient models from in vitro and in vivo studies have established its role in mediating cellular metabolism and inflammation (15,22), yet how AKT2 can be regulated independently from other isoforms is not well understood.In recent years, the importance of cysteine 124 (C124) of AKT2 in regulating its catalytic function has been demonstrated (26,27).C124 of AKT2 is a highly conserved residue only found in AKT2, but not other isoforms (26).Due to its position within the linker region, and its highly reactive nature to oxidative radicals, it is postulated that ROS are involved in AKT2-specfic regulation.Indeed, several studies have now demonstrated that the oxidation of C124 is critical in regulating AKT2 activity in response to oxidative stress, thereby linking redox homeostasis with the regulation of phosphorylation cascades (26,27).In our study, inhibition of mitochondria-derived ROS during oxLDL accumulation reinstated AKT2 activation (Fig. 4F), thereby supporting the above notion that changes in the redox environment can induce AKT2-specifc regulations, and that the oxidation of C124 by oxLDL-induced ROS may explain how oxLDL can selectively inactivate AKT2 but not other isoforms.
In this study we used thioglycolate-elicited monocytederived PMus to model arterial intimal Mus in atherosclerotic lesions, which are derived primarily from blood monocytes.Key findings were reproduced in BMDMus and RAW264.7 cells.However, it is possible that in other Mus, such as tissue-resident Mus, intracellular lipid accumulation could elicit different responses upon exposure to inflammatory stimuli.
Taken together, our study shows that the accumulation of oxLDL or cholesterol in Mus impairs LPS-induced AKT signaling.Unlike cholesterol, oxLDL loading selectively impairs the activation and activity of AKT2, but not other isoforms, due to mitochondria-derived ROS.This impairment subsequently led to the reduction of ACLY activity, acetyl-CoA levels, and H3K27ac of proinflammatory genes, independent of LPS-induced early-phase glycolysis.

FIGURE 1 .
FIGURE 1.Both oxLDL and cholesterol accumulation in PMus impair LPS-induced AKT signaling, whereas early glycolysis is inhibited only by cholesterol.(A and B) Effect of oxLDL or cholesterol (Chol) accumulation on LPS-induced AKT activation.Representative immunoblots and quantification of p-AKT (Thr 308 ), p-AKT (Ser 473 ), and total AKT is shown in an LPS time course experiment using PMus with or without accumulated oxLDL (A, n 5 4) or Chol (B, n 5 4-6).For each time point, p-AKT values were normalized to the corresponding total AKT value and the 0 h LPS time point of (Continued)

FIGURE 2 .
FIGURE 2. oxLDL accumulation in PMus selectively inhibits LPS-induced AKT2 activation and reduces activation of ACLY and intracellular acetyl-CoA, and selective inhibition of AKT2 reproduces the effects of oxLDL accumulation on glycolysis, ACLY activation, and acetyl-CoA.(A) Intracellular acetyl-CoA in LPS-stimulated (3 h) PMus with or without oxLDL accumulation.The data are normalized to the without (À)oxLDL group (assigned a value of 1, n 5 5).(B) Effect of oxLDL accumulation on LPS-induced activation of ACLY.Representative immunoblots and (Continued)

FIGURE 3 .
FIGURE 3. In contrast to the effects of cholesterol, oxLDL accumulation does not inhibit LPS-induced TBK1 activation, yet AKT2 and ACLY activation remain inhibited even when the localization of AKT2 to the plasma membrane is enhanced by inhibition of PTEN.(A) Effect of oxLDL or cholesterol (Chol) accumulation on LPS-induced activation of TBK1.Representative immunoblots and quantification of p-TBK1 (Thr 172 ) and total TBK1 in a LPS time course (0-3 h).PMus with or without oxLDL (n 5 4) or Chol (n 5 3) accumulation were used.For each time point, p-TBK1 values are normalized to corresponding total TBK1 and the 0 h LPS time point in the without (À)oxLDL/Chol groups (assigned a value of 1).(B and C) Effects of PTEN inhibitor (PTENi) on LPS-induced activation of AKT in PMus with accumulated oxLDL or Chol.Representative immunoblots and quantification of p-AKT2 (Ser 474 ) and total AKT2 (B), p-AKT (Ser 473 ) and total AKT (C), and p-ACLY (Ser 455 ) and total ACLY (B and C) in PMus with/without (6)oxLDL accumulation (B, n 5 3), 6Chol accumulation (C, n 5 5-8), 6PTENi and 6LPS stimulation (3 h) are as indicated.The mean 6 SEM is plotted in all graphs.Significant differences were determined by a two-way ANOVA (A) or one-way ANOVA (B and C) and a Bonferroni post hoc test.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.1, with; À, without.

FIGURE 4 .
FIGURE 4. The accumulation of oxLDL in PMus generates mitochondrial oxidative stress, which inactivates AKT2.(A) Hallmark pathway analysis of significantly upregulated and downregulated genes (p < 0.01) in oxLDL versus control PMus.Red arrows indicated fatty acid metabolism-related pathways.Blue arrows indicate cholesterol metabolism-related pathways.(B) LysoTracker staining in PMus with or without accumulated oxLDL.Representative fluorescence microscope images and quantification of LysoTracker mean fluorescence intensity (MFI) (n 5 15 fields, >3 independent experiments).(C) Heat map of fatty acid oxidation gene mRNA expression assessed by qPCR in PMus during sequential time points of oxLDL accumulation (n 5 3).(D and E) MitoSOX staining of PMus.(D) Staining at 0 or 18 h after accumulation of oxLDL.(E) Staining of cells with or without Mito-TEMPO (25 mM) treatment 1 h prior to culture with oxLDL for 18 h.Shown are representative fluorescence microscope images and quantification of MitoSOX MFI (n 5 50 fields, >3 independent experiments).(F) Effect of mitochondrial ROS suppression on the activation of AKT2 and ACLY in response to PTEN inhibition in PMus with or without oxLDL accumulation.Representative immunoblots and quantification of p-AKT2 (Ser 474 ), total AKT2, p-ACLY (Ser 455 ), and total ACLY in PMus with or without Mito-TEMPO (added 1 h prior to oxLDL), oxLDL accumulation (24-h), and PTENi (3-h treatment after culture with oxLDL) are as indicated.The p-AKT2 and p-ACLY values were normalized to the corresponding total AKT2 or total ACLY and the without (À)Mito-TEMPO ÀoxLDL groups (assigned a value of 1, n 5 3).The mean 6 SEM is plotted in all graphs.Significant differences were determined by an unpaired Student t test (B, D, and E) or a one-way ANOVA (C) or two-way ANOVA (F).*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.1, with; À, without.

FIGURE 5 .
FIGURE 5. oxLDL accumulation in PMus suppresses inflammatory gene expression at 3 h after LPS stimulation, and cholesterol accumulation inhibits LPS-induced signaling.(A and B) Analysis of proinflammatory gene mRNA expression by qPCR in LPS-stimulated (3 h) PMus with or without accumulated oxLDL (A, n 5 3-11) or cholesterol (Chol) (B, n 5 4-5).For each gene the data were normalized to cells without oxLDL or cholesterol (assigned a value of 1).(C) Representative immunoblots and quantification of p-ERK1/2 (Thr 202 /Tyr 204 ), total ERK, p-p65 (Ser 536 ), and total p65 in PMus with or without Chol accumulation after LPS stimulation (0-6 h).Values were normalized to the 0 h LPS time point in PMus without Chol (n 5 3).A representative immunoblot of nuclear and cytoplasmic (Cyto) fractions shows the post-LPS time course of p65 translocation to the nucleus in PMus with or without Chol accumulation.The mean 6 SEM is plotted in all graphs.Significant differences were determined by an unpaired Student t test (A and B) or a two-way ANOVA and a Bonferroni post hoc test (C).*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.1, with; À, without.

FIGURE 6 .
FIGURE 6. oxLDL accumulation in PMus suppresses LPS-induced H3K27ac.ChIP-PCR experiments measuring H3K27ac enrichment in promoters of inflammatory genes.(A) Time course after LPS stimulation of PMus.For each gene, the data were normalized to the 0 h LPS time point (assigned a value of 1, n 5 3).(B) PMus with or without oxLDL accumulation were assayed 1 h after LPS stimulation.For each gene, the data were normalized to the without (À)oxLDL group (assigned a value of 1, n 5 4).(C and D) Bone marrow-derived Mus from (C) Pten fl/fl and (D) Lyz2-Cre:Pten fl/fl mice with or without oxLDL accumulation were assayed 3 h after LPS stimulation.For each gene, the data were normalized to the ÀoxLDL group (assigned a value of 1, n 5 4-8).(E) Selective inhibition of AKT2 reproduces the H3K27ac phenotype of PMus with