Acyl-CoA:Diacylglycerol Acyltransferase 1 Expression Level in the Hematopoietic Compartment Impacts Inflammation in the Vascular Plaques of Atherosclerotic Mice

The final step of triacylglycerol synthesis is catalyzed by acyl-CoA:diacylglycerol acyltransferases (DGATs). We have previously shown that ApoE-/-Dgat1-/- mice are protected from developing atherosclerosis in association with reduced foam cell formation. However, the role of DGAT1, specifically in myeloid and other hematopoietic cell types, in determining this protective phenotype is unknown. To address this question, we reconstituted the bone marrow of irradiated Ldlr–/–mice with that from wild-type (WT→ Ldlr–/–) and Dgat1–/–(Dgat1–/–→ Ldlr–/–) donor mice. We noted that DGAT1 in the hematopoietic compartment exerts a sex-specific effect on systemic cholesterol homeostasis. However, both male and female Dgat1–/–→ Ldlr–/–mice had higher circulating neutrophil and lower lymphocyte counts than control mice, suggestive of a classical inflammatory phenotype. Moreover, specifically examining the aortae of these mice revealed that Dgat1–/–→ Ldlr–/–mice have atherosclerotic plaques with increased macrophage content. This increase was coupled to a reduced plaque collagen content, leading to a reduced collagen-to-macrophage ratio. Together, these findings point to a difference in the inflammatory contribution to plaque composition between Dgat1–/–→ Ldlr–/–and control mice. By contrast, DGAT1 deficiency did not affect the transcriptional responses of cultured macrophages to lipoprotein treatment in vitro, suggesting that the alterations seen in the plaques of Dgat1–/–→ Ldlr–/–mice in vivo do not reflect a cell intrinsic effect of DGAT1 in macrophages. We conclude that although DGAT1 in the hematopoietic compartment does not impact the overall lipid content of atherosclerotic plaques, it exerts reciprocal effects on inflammation and fibrosis, two processes that control plaque vulnerability.

Exploring how DGAT1 deficiency reduces adiposity has highlighted its role in dietary fat assimilation [10][11][12][13][14][15]. For example, although reducing TG synthesis capacity globally might be expected to increase the systemic and peripheral tissue levels of glycerolipid precursors [DGs and free fatty acids (FFAs)] or metabolites (e.g. ceramides), this does not occur in Dgat1 -/mice [6]. Rather, Dgat1 -/mice have reduced intestinal lipid absorption and an impaired ability of enterocytes to form TGs from dietary FFAs [12,16]. Moreover, expressing DGAT1 exclusively in the intestine is sufficient to reverse the protection against obesity and hepatic steatosis, normally seen in Dgat1 -/mice fed a diet high in fat [16]. Together, these findings support the concept that the net impact of DGAT1 deficiency on lipid homeostasis is predominantly due to DGAT1 deficiency in enterocytes [13,16].
In addition to its role in controlling TG homeostasis, DGAT1 may also play a role in atherosclerosis, a disease usually linked to dysregulated cholesterol homeostasis. Dgat1 -/mice bred onto an apolipoprotein E (ApoE)-deficient background have reduced atherosclerotic plaque formation [17]. This reduced lesion size is associated with decreased cholesterol uptake and absorption by the intestine, reduced plasma TG and cholesterol concentrations, and increased cholesterol efflux from macrophages [17]. One possible mechanism for the atheroprotection seen in ApoE -/-Dgat1 -/mice may, at least in part, be akin to the one that drives the metabolic protection seen in Dgat1 -/mice. Specifically, both models may benefit from reduced rates of intestinal lipid uptake and absorption and a suppression of systemic lipid accumulation.
However, while atherosclerosis is traditionally associated with dysfunctional lipid metabolism, considerable evidence has accumulated to implicate vascular inflammation as a key component in atherogenic plaque progression, with activated macrophages residing in the vascular wall acting as key players. Specifically, macrophages engorged with intracellular cholesteryl esters (CEs), or "foam cells", accumulate in atherosclerotic plaques [18]. Foam cells undergo a form of inflammatory activation, a pathological hallmark of atherogenesis.
With this in mind, it is intriguing to note that, unlike the phenotype of Dgat1 -/mice, DGAT1 deficiency that is limited to tissues and cells other than the intestine has deleterious consequences. In particular, DGAT1 deficiency increases the vulnerability of isolated skeletal muscle to lipotoxicity [19] and potentiates a remarkable exaggeration in the inflammatory response of cultured macrophages to pro-inflammatory lipids [20].
Thus, whereas global DGAT1 deficiency may protect against both plaque development and plaque inflammation in atherosclerosis-prone mice, DGAT1 deficiency specifically in the myeloid compartment might paradoxically worsen atherosclerosis. To test this, we reconstituted the bone marrow of low-density lipoprotein receptor (Ldlr -/-) mice with that of WT or Dgat1 -/mice. Our studies show that DGAT1 expression in macrophages has no impact on the lipid content of atherosclerotic plaques developing in response to the consumption of a diet rich in saturated fat and cholesterol, but instead regulates the extent to which plaques exhibit markers of inflammation and fibrosis, two processes that determine plaque vulnerability.

Animals and diets
Dgat1 -/- [4], WT, and Ldlr− /-(Jackson Laboratory, Bar Harbor, ME) were on a C57BL/6J background (>10 generations). Animals were housed in a clean barrier facility with regular 12-hour light-dark cycling and had unlimited access to water and either of two standard chow diets [5053 PicoLab (Purina, St. Louis, MO) or Altromin (Lage, Germany)]. All mouse experiments were approved by the Institutional Animal Care and Use Committees (IACUC) at the University of California San Francisco or Medical University of Graz followed either NIH guidelines (AN111420-01C) or those of the Austrian Ministry for Science, Research, and Economy, Division of Genetic Engineering and Animal Experiments (Vienna, Austria) (BMWF-66.010/0057-II/3b/2011) as appropriate.

Bone marrow transplantation
Eight week-old male and female Ldlr− /mice were irradiated twice with 5 Gy 3 hours apart 1 day before injection. Bone marrow cells were harvested from age-and sex-matched WT and Dgat1 -/donor mice, re-suspended in PBS containing 0.1% FA-free BSA, and intravenously injected (1-1.2 × 10 6 cells) into the irradiated Ldlr− /recipient mice. Recipient mice were then treated with antibiotics (Polymyxin B and Neomycin) in water for 4 weeks. Genotyping from blood was performed to confirm bone marrow transplantation efficiency. At 16 weeks of age, the mice were challenged with a Western-type diet (WTD; TD.88137; Harlan-Teklad, IN; containing 21.2% fat, and 0.2% cholesterol by weight) for additional 13 weeks (females) or 19 weeks (males).

Body composition
Body composition was analyzed by dual energy X-ray absorptiometry (DEXA) with a PIXI-mus2 scanner (GE Healthcare Lunar, Madison, WI).

Complete blood cell count
One hundred μl of blood was collected into EDTA-coated tubes (Sarstedt AG, Nuembrecht, Germany). A complete blood cell count was performed using a Hemavet 850 (Drew Scientific Group, Waterbury, CT).

Plasma lipid parameters
Circulating lipid concentrations were determined from plasma of mice fasted for 10 hours using spectrophotometric kits (TG, Thermo Scientific, Middletown, VA; total cholesterol (TC) and free cholesterol (FC), WAKO chemicals, Richmond, VA). Cholesteryl ester (CE) levels were calculated as the difference between TC and FC. cavity, bi-valved in a Y-formation, and stored in 10% methanol-free formalin until staining. The cranial two-thirds of the heart were fixed with 10% methanol-free formalin for 24 hours and then stored in 30% sucrose. Prior to sectioning, the hearts were transferred into Tissue Tek (Sakura Finetek, Torrance, CA), and the aortic roots were serially cryosectioned (8 μm) 24 hours later at -20°C.
Oil red O (ORO) staining of aortae and aortic valve sections ORO (0.5 g) was dissolved in 100 ml isopropanol under constant stirring for 24 hours, mixed at a 3:2 ratio with dH 2 O for 10 minutes to create a working solution, which was then filtered and used freshly.
Bi-valved aortae were washed with PBS for 1 min, placed in 70% isopropanol for 5 min, and then stained with the ORO working solution for 20 minutes, after which they were dipped in 70% isopropanol to remove excessive ORO. Stained aortae were pinned out and imaged using a stereo zoom microscope (Leica MZ FL III equipped with Leica DC500 camera).
Aortic root sections were fixed in 10% methanol-free formalin for 10 minutes, placed in 70% isopropanol for 5 more minutes, and then placed in ORO working solution for 15 minutes. Sections were then dipped in 70% isopropanol, counterstained with hematoxylin for 5 minutes, and washed in 0.1% NaHCO 3 . Sections were air dried and mounted with glycerol-gelatin aqueous slide mounting medium (Sigma-Aldrich, St. Louis, MO). Mean lesion area (μm 2 ) was calculated from consecutive ORO-stained sections. Images were taken with a Leica DM 4000B microscope equipped with a Leica DFC 500 camera. Plaque areas were quantitated using ImageJ software.
Immunostaining with monoclonal antibody to macrophages-2 (MoMa-2) Aortic root sections were fixed for 15 minutes in 10% methanol-free formalin and washed three times with PBS. Sections were then blocked for 30 minutes to protect against endogenous tissue peroxidase activity (Signet Covance, Prineceton, NJ), washed twice with PBS, and then blocked with Ultra V protein block (Lab Vision, Fremont, CA) for 7 minutes. The sections were next incubated in Ultra V protein block solution (1:4 in PBS) containing monoclonal antibody to macrophages-2 (MoMa-2, 1:600; Acris, Hiddenhausen, Germany) for 1 hour at RT, and then overnight at 4°C.
The sections were next washed twice with PBS, incubated for 3 hours with an HRP-coupled polyclonal rabbit anti-rat secondary antibody (1:100, Dako Denmark A/S, Glostrup, Denmark), and washed three times with PBS. AEC substrate (Vector Laboratories, Burlingame, CA) was added until reaction development and then washed away 3 times with H 2 O. Sections were counterstained with hematoxylin for 5 min, washed in 0.1% NaHCO 3 , and mounted with glycerol-gelatin aqueous slide mounting medium (Sigma-Aldrich, St. Louis, MO). Images were taken with a Leica DM 4000B microscope equipped with a Leica DFC 500 camera. The intensity and extent of MoMa-2 staining of three consecutive sections per mouse was quantified using ImageJ software. The sections were then stained with Masson's Trichrome (Sigma-Aldrich, St. Louis, MO) according to manufacturer's instructions, washed with dH 2 O, and dehydrated (90, 95, and 100% ethanol/xylene). Sections were fixed with Cytoseal-60 mounting medium (Thermo Scientific, Kalamazoo, MI). Images were taken with a ScanScope T3 whole slide scanner (Aperio Technologies, Bristol, UK). Both the blue-stained collagen-rich areas and the acellular necrotic core areas of vascular plaques from three consecutive sections per mouse were quantitated with ImageJ software.

Isolation and cultivation of peritoneal macrophages
Mice were injected i.p. with 2.5 ml of 3% thioglycollate broth. Their peritoneal cavities were flushed 72 hours later with 10 ml PBS containing 1 mM EDTA (PBS/EDTA). The peritoneal cells were centrifuged and pelleted cells were plated in DMEM (Gibco, Invitrogen, Carlsbad, CA) containing 10% lipoprotein-deficient serum (LPDS) and 1% penicillin/streptomycin (P/S) for 2-3 hours. The plated cells were then washed twice with pre-warmed PBS, and the adherent cells (macrophages) were cultured in DMEM containing 25 mM glucose, 4 mM glutamine, 1 mM pyruvate, 10% LPDS, and 1% P/S for 24 hours. These cells were washed twice with prewarmed PBS and cultured for additional 24 hours either in the absence or presence of VLDL, acetylated (ac)LDL (both 100 μg protein/ml medium), or LPS as a control (10 ng/ml medium).

RNA isolation and quantitative real-time PCR analysis
Macrophage RNA was isolated using TriFast™ reagent according to the manufacturer's protocol (Peqlab, Erlangen, Germany). One μg of total RNA was reverse transcribed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA). Quantitative real-time PCR (qPCR) was performed on a Roche LightCycler 480 (Roche Diagnostics, Palo Alto, CA) using the QuantifastTM SYBR1 Green PCR kit (Qiagen, Hilden, Germany). mRNA levels were analyzed in duplicate and normalized to those of the reference gene hypoxanthine phosphoribosyltransferase (Hprt). Fold-changes and associated statistical parameters were determined by the 2 -ΔΔCT method. The following primer sequences were used: Arg1:

Statistics
Statistical analyses were performed using GraphPad Prism 5.0 software. Significances were determined using the Student's unpaired t-test and the Welch test (in cases of unequal variances) for two-group comparisons, and ANOVA followed by Bonferroni correction for multiple group comparisons. Data are presented as mean values ± SEM. Significance levels were set at p < 0.05 ( Ã ), p 0.01 ( ÃÃ ) and p 0.001 ( ÃÃÃ ).

Results
Bone marrow reconstitution of WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice We investigated the consequences of hematopoietic DGAT1 deficiency on atherogenesis by transplanting Ldlr -/mice with bone marrow cells from age-and sex-matched WT and Dgat1 -/donor mice. PCR analyses of the genomic DNA from peripheral blood cells of transplant recipient mice following recovery confirmed both successful irradiation and the expected genetic chimerism (S1 Fig). Specifically, the donor-derived circulating leukocytes in recipient mice expressed Ldlr (S1A Fig), whereas Dgat1 mRNA in these leukocytes was appropriately present or absent depending on the donor genotype (S1B Fig). We placed both male and female transplant recipient mice on a Western-type high-fat diet (WTD). Since female mice, at least on an ApoE -/background, develop atherosclerosis more rapidly than males [21], we fed female transplant recipients a WTD for 13 weeks and male recipients a WTD for 19 weeks.
Comparable body weights and composition of WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice Dgat1 -/-! Ldlr -/mice fed a WTD gained a similar amount of weight as controls over both 13 weeks ( Fig 1A) and 19 weeks (Fig 1B). Body composition was analyzed one week before the extent of atherosclerosis was analyzed. Dual energy X-ray absorptiometry (DEXA) revealed comparable total body fat (Fig 1C), percent body fat (Fig 1D), and lean mass ( Fig 1E) between WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice. The weights of the gonadal fat pads ( Fig 1F) and livers ( Fig 1G) were also comparable between both genotypes. Dgat1 -/-! Ldlr -/mice have sex-specific alterations in cholesterol homeostasis WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice had similar fasting plasma TG concentrations, regardless of sex, after consuming a WTD for 13 weeks (females) or 19 weeks (males). By contrast, deleting DGAT1 from the hematopoietic compartment had a sex-specific impact on systemic cholesterol levels. Female Dgat1 -/-! Ldlr -/mice fed a WTD (13 weeks) had lower plasma TC concentrations than controls, whereas male Dgat1 -/-! Ldlr -/mice fed the same diet (19 weeks) had increased plasma TC concentrations (Table 1). Moreover, the effect of hematopoietic DGAT1 deficiency on plasma TC in either sex was coincident with a significant and directionally concordant effect on plasma CE concentration (Table 1). Therefore DGAT1 deficiency among hematopoietic cell types impacts cholesterol homeostasis by altering circulating CE levels, an effect the directionality of which is determined by sex.
Altered relative distribution of white blood cell populations in Dgat1 -/-! Ldlr -/mice Since only the hematopoietic system lacked DGAT1 in our mouse model, we monitored the impact of this deficiency on the numbers of white blood cells (WBCs) and their specific subsets   Plasma parameters measured after a 10-hour fast in WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice (n = 9-11) chronically fed a WTD. Data are means ± SEM. *, p< 0.05 in the circulation. The total number of circulating WBCs in both male and female mice was comparable between WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice fed a WTD (Fig 2A), as were circulating levels of monocytes, eosinophils, and basophils (Fig 2B-2D). Lymphocyte frequencies, however, were reduced in both female (trend) and male Dgat1 -/-! Ldlr -/mice fed a WTD (Fig 2E). By contrast, circulating neutrophil counts were increased in both female (trend) and male Dgat1 -/-! Ldlr -/mice ( Fig 2F). These findings are consistent with the presence of classical systemic inflammation in Dgat1 -/-! Ldlr -/animals.

Dgat1 -/-! Ldlr -/mice develop pro-inflammatory atherosclerotic plaques
Next, we analyzed atherosclerotic plaque formation along the aortae and in the aortic roots of bone marrow-transplanted animals. En face staining revealed that Dgat1 -/-! Ldlr -/and control mice have comparable atherosclerotic plaque areas in both the thoracic aorta (Fig 3A and  3B) and the aortic arch (Fig 3A and 3C), irrespective of sex. Analyzing the neutral lipid content of aortic root sections showed that Dgat1 -/-! Ldlr -/and control mice have similarly sized plaques in the aortic valve region as well (Fig 4A). However, a more detailed analysis of atherosclerotic lesions in both genotypes of mice showed that lesional macrophage content in female Dgat1 -/-! Ldlr -/mice is greater than control, a finding also seen as a trend in male Dgat1 -/-! Ldlr -/mice (Fig 4B). This indicates that limiting the ability of hematopoietic cells to synthesize TG promotes innate inflammation in the vascular wall, which is where excess lipid deposition occurs in Ldlr -/mice. Moreover, lesional collagen content was reduced in male Dgat1 -/-! Ldlr -/mice and trended similarly in female Dgat1 -/-! Ldlr -/mice ( Fig 4C). Together, these data indicate that atherosclerotic lesions in the aortic roots of Dgat1 -/-! Ldlr -/mice may be less stable than in controls. This concept was supported by our finding that aortic root plaques from Dgat1 -/-! Ldlr -/mice have a reduced ratio of collagen to macrophage content, an important indicator of plaque instability (Fig 4D).
On the other hand, other mechanistic determinants of plaque stability were intact despite the loss of DGAT1 among macrophages. For example, the extent of intra-plaque necrosis, defined as acellular compartments within atherosclerotic lesions, was similar across genotypes, irrespective of sex ( Fig 4E). The ratio of plaque collagen to intra-plaque necrosis in aortic roots was also comparable between genotypes (Fig 4F). Similarly, minimal thicknesses of plaque fibrotic caps, which correlates with the risk of plaque rupture and intraluminal thrombosis, was comparable between Dgat1 -/-! Ldlr -/and control mice ( Fig 4G). Thus, plaque structures were largely similar between both genotypes, suggesting that the differences in plaque composition between these genotypes is limited to their inflammatory nature.

Macrophage phenotype is regulated independently of DGAT1 expression
Based on our in vivo findings, we sought to determine if altering DGAT1-dependent lipid handling by macrophages impacts inflammatory responsiveness at a cell-intrinsic level. Therefore, we investigated the effects of lipoprotein loading on the polarization state of elicited mouse peritoneal macrophages in vitro. As expected, treating these macrophages for 24 hours with LPS induced a classical inflammatory transcriptional response, and thus served as a positive control. On the other hand, the mRNA levels of genes marking both the classical M 1 (Ccl5, Itgax, Tnfa, Mcp1, Nos2, and Gro1) ( Fig 5A) and alternative M 2 (Arg1 and Mrc1) (Fig 5B) macrophage polarization states were comparable between Dgat1 -/and WT macrophages, whether untreated or loaded with VLDL, acLDL, or LPS. These results indicate that DGAT1 deficiency does not heighten the inflammatory responsiveness of macrophages to lipoprotein stress in this model, a distinction from the vulnerability to inflammatory activation that was reported for Dgat1 -/macrophages treated with free pro-inflammatory saturated FAs [20].

Discussion
Numerous reports have shown that DGAT1 deficiency and pharmacological inhibition have beneficial effects on whole body metabolism, lipid homeostasis, and longevity [4][5][6][7][8][9][10][11][12][13]. Our previous finding that global ApoE -/-Dgat1 -/mice have reduced atherosclerotic plaque formation [17] left us wondering whether amelioration of the atherosclerosis in this model is a consequence of reduced intestinal cholesterol uptake and absorption, resulting in lower plasma cholesterol levels, or alterations in macrophage function.
To address this, we adoptively transferred WT and Dgat1 -/hematopoietic cells, respectively, into irradiated Ldlr -/recipient mice. We chose Ldlr -/over ApoE -/mice as recipients since in contrast to Ldlr -/mice, transplantation of ApoE-expressing bone marrow into Apoe -/recipients is able to repair hyperlipidemia and atherosclerosis [22,23]. Our approach reveals that (A-C, right) Quantification of these images to measure plaque size, macrophage content, and collagen deposition, respectively. (D-F) Integrated quantification of data from A-C to determine collagen-to-macrophage ratios, necrotic core per plaque area, and collagen-to-necrotic core ratios, respectively. (G) Representative images of Masson's Trichrome-stained aortic root sections focused on plaque fibrotic caps, with quantification of minimal fibrotic cap thickness. All data represent means ± SEM of 12 aortic root sections for ORO and three aortic root sections in the area of maximal plaque size for MoMa-2-and Trichrome-staining per mouse after 13 (n = 9 females, ♀) and 19 weeks (n = 10-11 males, ♂) of WTD feeding. *, p < 0.05. doi:10.1371/journal.pone.0156364.g004 DGAT1 expression specifically in the hematopoietic compartment limits the inflammatory impact of WTD consumption on vascular plaques in the context of LDL receptor deficiency while having little if any effect on atherosclerosis per se or on other structural determinants of plaque composition. The effect of hematopoietic DGAT1 deficiency on increasing plaque macrophage content in the aortic root was mirrored by a relative systemic neutrophilia and lymphocytopenia, both markers of classical inflammation. Finally, it was interesting to note that mice lacking DGAT1 in immune cells had alterations in circulating CE and TC levels, indicative of a role for these cells in regulating systemic cholesterol homeostasis. DGAT1 has been well demonstrated to exert effects on energy balance and body weight. For example, both genetically deleting DGAT1 throughout the body in mice and pharmacologically inhibiting its activity systemically leads to a reduction in body weight and adiposity [4,12,13]. By contrast, overexpressing DGAT1 in the white adipose tissue of mice leads to increased adiposity and a propensity for diet-induced obesity [20]. Macrophages are immunological sentries and coordinators of tissue remodelling and homeostasis throughout the body that notably also express DGAT1. Here we assessed the effect of DGAT1 deficiency in the hematopoietic compartment, which gives rise to macrophages, on energy homeostasis, and found that it does not alter body weight gain, adiposity, or liver size in mice fed a WTD. This finding indicates that DGAT1 deficiency limited to immune and other hematopoietic cells is insufficient to alter whole body energy homeostasis. Our results mirror those seen in mice overexpressing DGAT1 specifically in macrophages, a model that also fails to impact body weight and adiposity [20].
In contrast to the lack of effect on fat storage, we found that Dgat1 -/-! Ldlr -/mice had alterations in systemic lipid homeostasis. Specifically, hematopoietic DGAT1 deficiency impacted fasting plasma TC and CE levels versus WT! Ldlr -/controls. The intriguing aspect of this finding, however, is that the effects were directionally opposite in male and female mice. Whether this discrepancy arises from sex-specific differences in how macrophages regulate cholesterol homeostasis or whether differences in the duration of the WTD (13 weeks in females, 19 weeks in males) reveal a temporally bi-phasic effect on cholesterol homeostasis in our model remains unclear. Regardless, it is important to point out that the changes we saw in systemic cholesterol levels reflect the capacity of DGAT1, expressed exclusively in the hematopoietic compartment, to determine systemic cholesterol levels. This finding will be followed up in future studies.
Bone marrow transplantation exerts specific long-term effects on hematopoiesis in recipients [24]. Given this, we were encouraged to find that total WBC numbers were similar in mice receiving Dgat1 -/and WT bone marrow in our study, indicating that DGAT1 is not required for normal hematopoiesis. This finding corroborates what is already known about global Dgat1 -/mice, which do not have any gross hematopoietic defect. Given this, it was interesting to note that Dgat1 -/-! Ldlr -/mice in our study did have alterations in the relative abundance of specific circulating WBC populations following chronic consumption of a WTD. Both male and female mice tended to have elevated circulating neutrophil counts and concomitantly reduced lymphocyte counts. Since neutrophils have been implicated in promoting the pathogenesis of atherosclerosis [25] by modulating processes associated with the vulnerability and inflammatory properties of vascular plaques [26,27], it is reasonable to speculate that neutrophils may have influenced the plaque phenotype in Dgat1 -/-! Ldlr -/mice or that their systemic levels reflect a response to chronically increased plaque inflammation.
When quantified by monitoring overall neutral lipid content, WT! Ldlr -/and Dgat1 -/-! Ldlr -/mice had a similar aortic plaque distribution and area, regardless of sex. We found that aortic valve lesions developed more rapidly in female mice (13 weeks of WTD) than in males (19 weeks of WTD), and these sex-specific differences were not impacted by DGAT1 deficiency in the hematopoietic compartment.
Although it is used to track plaque growth, the neutral lipid content of atherosclerotic plaques may not be a clinically valuable pathological marker of plaque advancement. On the other hand, the number and polarization state of macrophages within atherosclerotic plaques have been implicated in mediating the advancement of these lesions [28]. Interestingly, the key effect of hematopoietic DGAT1 deletion on plaque structure in the female mice we studied here was to visibly increase the number of intra-plaque macrophages, a trend that was also seen in male mice.
Enzymes involved in lipid metabolism and handling within the hematopoietic compartment have been shown to be important in systemic lipid homeostasis and atherosclerosis [29][30][31][32][33][34][35]. However, because such enzymes have potentially pleiotropic roles in systemic metabolism, the results of deleting genes encoding them in mice often have surprising effects on atherosclerosis.
For example, lipoprotein lipase (LPL) deficiency in the bone marrow might be expected to impair intravascular TG hydrolysis. However, such mice do not have altered systemic TG levels, and unexpectedly display reduced systemic TC concentrations. Indeed, hematopoietic LPL deficiency is associated with reductions in both macrophage ApoE secretion andatherosclerosis [29]. Similarly, mice lacking the intracellular TG hydrolytic enzyme adipose triglyceride lipase (ATGL) in the bone marrow have reduced atherosclerotic progression, despite the prediction that ATGL-deficient macrophages would accumulate TG and thus aggravate foam cell formation and atherogenesis [36]. By contrast, macrophages deficient in peroxisome proliferator-activated receptor γ (PPARγ) are resistant to foam cell formation, yet display increased inflammatory potential and a concomitant vulnerability to atherogenesis [37].
It was thus unclear what impact hematopoietic DGAT1 deficiency would have on plaque inflammation in Dgat1 -/-! Ldlr -/mice, and our observation that these mice have an increased plaque macrophage burden has several potential mechanistic implications. First, it underscores the concept that enhancing the capacity of macrophages to stably store intracellular fatty acids (FAs) within TG is anti-inflammatory. It also suggests that increasing macrophage TG storage capacity limits the flux of intracellular FAs into pro-inflammatory species. These could include glycerolipid intermediates such as lysophospholipids and DGs, and lipids of alternate metabolic pathways, including sphingolipids, ceramides, and phospholipids.
We also investigated the inflammatory responsiveness of cultured Dgat1 -/macrophages. We have previously shown that directly exposing macrophages to saturated FAs drives Dgat1 -/macrophages to a pro-inflammatory, M 1 -like phenotype [20], which is associated with complicated, destabilized atherosclerotic lesions [38]. However, the mechanism by which saturated FAs exert their pro-inflammatory effects is an area of ongoing investigation, as is the nature of the enhanced inflammatory vulnerability of Dgat1 -/macrophages.
More recently, we have shown that the flux of intracellular FAs into the phospholipid compartment of macrophages mediates activation of the NLRP3 inflammasome [39]. This suggests that at least one way how DGAT1 deficiency contributes to a pro-inflammatory predisposition may be through limiting the flux of intracellular FAs into TG, and increasing their flux into phospholipids. Membrane phospholipids are important in the release of arachidonic acid, a precursor of eicosanoid synthesis [40]. Therefore, increased phospholipid flux in macrophages could affect systemic immune responsiveness by altering eicosanoid homeostasis.
Unlike in metabolic tissues such as the WAT, skeletal muscle, and liver, where the concentration of FAs and their metabolites may rise in the context of diet-induced obesity and exert pro-inflammatory effects on resident and infiltrating macrophages, lipid-activation of macrophages in atherosclerotic plaques may be driven more by locally deposited LDL, including species with pro-atherogenic modifications. Given this, it isinteresting that unlike the response we saw in macrophages treated with FFAs [20], the M 1 response of macrophages treated with either VLDL or acLDL was not impacted by the presence or absence of DGAT1. This finding suggests that the role DGAT1 plays in the manner by which macrophages handle FFAs taken up from their environment is different from the role it plays in macrophage handling of intracellular FAs during the course of lipoprotein uptake and metabolism.
It should be noted, however, that the current study used macrophages that were elicited into the peritoneum by thioglycollate, harvested and plated, and then studied in culture. Such elicited peritoneal macrophages change their polarization from more naïve M 0 -like state to a "preactivated" M 1 -like state. Therefore, these macrophages may have already been somewhat primed prior to lipoprotein or LPS treatment, and this could have limited our ability to measure the pro-inflammatory effects of VLDL or acLDL treatments in both WT and Dgat1 -/macrophages. On the other hand, LPS-treatment of peritoneal macrophages in our study robustly induced the transcription of M 1 genes, indicating that these cells are still amenable to further activation by a pro-inflammatory stimulus.
The collagen content and collagen-to-macrophage ratios of plaque lesions were reduced in Dgat1 -/-! Ldlr -/animals, whereas the thickness of the fibrotic cap, which protects the lesion from rupture and the necrotic core from leakage into arterial lumen, was comparable between both genotypes. These results indicate that at least some indices marking plaque instability are worsened when macrophages lack DGAT1 in the context of progressive atherosclerosis in mice. It will be important to discern what factors reduce collagen deposition in this model while sparing cap fibrotic potential. Both WT! Ldlr -/and Dgat1 -/-! Ldlr -/plaques contained a high proportion of intra-plaque necrosis, commonly found in advanced lesions [41]. This is not surprising, as plaque inflammation and downstream necrosis may have been potentiated in both groups of mice by exposure to ionizing irradiation in preparation for bone marrow transplantation [42][43][44].
In summary, we examined the contribution of macrophage and other hematopoietic DGAT1 expression on atherosclerosis by studying Ldlr -/mice reconstituted with bone marrow taken from Dgat1 -/mice. We found that the resulting Dgat1 -/-! Ldlr -/mice developed atherosclerotic lesions with increased macrophage content and showed a tradeoff between reduced plaque collagen content and greater inflammation. However, hematopoietic DGAT1 deficiency had no appreciable impact on overall plaque burden or other aspects of plaque structure (lipid content, necrosis, cap thickness).

Conclusion
Our data support the possibility that DGAT1 plays a dual role in atherosclerosis. On one hand, global DGAT1 deficiency protects against atherosclerosis [17]. On the other hand, we show here that DGAT1 in macrophages may specifically participate in limiting the development of plaque inflammation. Our findings strongly suggest that the former finding in ApoE -/-Dgat1 -/mice is due to DGAT1 deletion in another cell or tissue type. One possibility is that DGAT1 deficiency in the gut reduces that rate at which dietary TG and cholesterol is assimilated, thus limiting the availability of the lipid substrate for plaque development in ApoE -/-Dgat1 -/mice.