Intestine-specific DGAT1 deficiency improves atherosclerosis in apolipoprotein E knockout mice by reducing systemic cholesterol burden

Background and aims Acyl-CoA:diacylglycerol acyltransferase 1 (DGAT1) is the rate-limiting enzyme catalyzing the final step of triglyceride synthesis by esterifying a diglyceride with a fatty acid. We have previously shown that apolipoprotein E-knockout (ApoE −/−) mice lacking Dgat1 have reduced intestinal cholesterol absorption and potentiated macrophage cholesterol efflux, and consequently, exhibit attenuated atherogenesis. However, he-matopoietic Dgat1 deficiency lacked beneficial effects on atherosclerosis. Due to our recent results on the critical role of intestinal Dgat1 in murine cholesterol homeostasis, we delineated whether intestinal Dgat1 deficiency regulates atherogenesis in mice. Methods We generated intestine-specific Dgat1 −/− mice on the ApoE −/− background (iDgat1 −/− ApoE −/−) and determined cholesterol homeostasis and atherosclerosis development. Results When fed a Western-type diet, iDgat1 −/− ApoE −/− mice exhibited a substantial decrease in fasting plasma cholesterol content in ApoB-containing lipoproteins. Although lipid absorption was delayed, iDgat1 −/− ApoE −/− mice had reduced acute and fractional cholesterol absorption coupled with an elevated fecal caloric loss. In line, increased appearance of i.v. administered [3H]cholesterol in duodena and stool of iDgat1 −/− ApoE −/− animals suggested potentiated cholesterol elimination. Atherosclerotic lesions were markedly smaller with beneficial alterations in plaque composition as evidenced by reduced macrophage infiltration and necrotic core size despite unaltered collagen content, indicating improved plaque stability. Conclusions Disruption of Dgat1 activity solely in the small intestine of ApoE −/− mice strongly decreased plasma cholesterol levels by abrogating the assimilation of dietary cholesterol, partly by reduced absorption and increased excretion. Consequently, the reduced cholesterol burden significantly attenuated atherogenesis and improved the lesion phenotype in iDgat1 −/− ApoE −/− mice.

Importantly, negative aspects of systemic Dgat1 deficiency such as alopecia and mammary gland atrophy are absent in iDgat1 −/− mice (unpublished observations) and upon pharmacological inhibition of Dgat1 [12,13]. Expression of the enzyme solely in the intestine is capable to reverse the phenotype of global Dgat1 deficiency [14], underlining the importance of intestinal Dgat1 expression. Dgat1 −/− mice on an apolipoprotein E (ApoE) −/− background are protected from atherogenesis due to reduced dietary cholesterol uptake and absorption in the small intestine, increased macrophage cholesterol efflux, and diminished foam cell formation [11]. Dgat1 −/− macrophages, however, are associated with a pro-inflammatory phenotype [15]. Thus, specific deletion of Dgat1 in hematopoietic cells aggravates atherogen-esis, causing the formation of destabilized, pro-inflammatory atherosclerotic lesions in low-density lipoprotein (LDL) receptor-deficient mice [16]. We therefore hypothesized that loss of Dgat1 activity specifically in the small intestine is key to the atheroprotection of Dgat1 −/− ApoE −/− mice. Our results demonstrate that iDgat1 −/− ApoE −/− mice phenocopy systemic Dgat1 disruption [9,11] with reduction in the total cholesterol burden, resulting in substantial attenuation of atherosclerosis in iDgat1 −/− ApoE −/− mice.

Plasma lipid parameters and lipoprotein profiles
Plasma lipid parameters and lipoprotein profiles after separation by fast-protein liquid chromatography were determined as described [17].

Cholesterol absorption studies
To determine acute cholesterol absorption, mice fed a WTD for 11 weeks were fasted for 4 h and gavaged with 100 μl corn oil containing 2 μCi [ 3 H]cholesterol (ARC Inc, St. Louis, MO) and 0.25% cholesterol. Blood was collected every hour post-gavage and plasma was isolated by centrifugation at 3,500 rpm for 10 min at 4 °C. Four hours post-gavage, animals were sacrificed and tissues were isolated, lyophilized overnight, and lysed in 1 M NaOH. Radioactivity in plasma and tissues was determined by liquid scintillation counting.

Intestinal lipid concentrations
After 4 h of fasting, male mice fed a WTD for 9 weeks were gavaged with 100 μl corn oil. Mice were sacrificed 4 h post-gavage and lipid concentrations in duodenum, jejunum, and ileum were estimated [18]. Histologic sections of the small intestine and oil red O (ORO) staining were performed as described [19].

Chylomicron size measurement
Chylomicron (CM) size was measured in male mice fed a WTD for 9-12 weeks. Four hourfasted mice were intraperitoneally (i.p.) injected with 1 g poloxamer 407 (P407)/kg body weight. Thirty minutes post-injection, mice were gavaged with 100 μl corn oil containing 0.25% cholesterol, and blood was taken after 90 min. CMs in plasma were isolated and analyzed as described [9]. Briefly, 400 μl of pooled plasma were mixed with 900 μl PBS containing 4 M KBr, carefully overlaid with 0.9% NaCl in 5.5 ml Quick-Seal centrifugation tubes (Beckman Coulter, Brea, CA), and centrifuged at 416,000g for 45 min. Chylomicron size in the top layer was measured by Zetasizer Nano ZS (Malvern Instruments Ltd., Malvern, UK).

Cholesterol excretion
Cholesterol excretion was determined in male mice fed a WTD for 11 weeks as previously described [18]. Briefly, 6 h-fasted mice were intravenously (i.v.) injected with 200 μl native human LDL (2.75 mg cholesterol/ml PBS and 2 μCi [ 3 H]cholesterol; ARC Inc, St. Louis, MO). Animals had free access to food and water for the following 20 h, after which they were fasted for 4 h and sacrificed. Fecal samples and tissues were lyophilized and fecal lipids were extracted as described [18]. Extracts were re-dissolved in 200 μl of 2% Triton X-100 in dH 2 O and mixed with 1 ml methanol. Bile was isolated by centrifugation of complete gall bladders at 10,000 rpm for 10 min. Tissues were lysed in 1 ml of 1 M NaOH and radioactivity in tissues, fecal extracts, and bile was deter-mined by liquid scintillation counting.

De novo lipid synthesis
De novo lipid synthesis was measured in mice fed a WTD for 8 weeks and fasted for 4 h prior to the experiment as described [18]. Briefly, the animals were injected i.p. with 5 μCi [14C]acetic acid in PBS and sacri-ficed after 1 h. Livers were lyophilized, pulverized, and lipids from 100 mg of pulverized livers were extracted with chloroform:methanol (2:1) for 2 h. Extracts were dried under a stream of N 2 , re-dissolved in chlo-roform, and separated by thin-layer chromatography (n-hexane:dieth-ylether:acetic acid; 70:30:1, v:v:v). Radioactivity in bands corresponding to specific lipid classes was determined by liquid scin-tillation counting.

Bomb calorimetry
Mice fed a WTD for 8 weeks were single-housed to collect feces during one week. Feces were lyophilized overnight, pulverized, and 1 g of powder was compressed into a pellet. The pellets were combusted in an adiabatic oxygen bomb calorimeter C200 (IKA Analysentechnik, Staufen, Germany) at 20 atm and excess of oxygen to ensure total oxidation of any carbon-based compounds to CO 2 . The combustion heat was determined based on prior calibration with benzoic acid [20].

RNA isolation and quantitative real-time PCR analysis
Male mice fed a WTD for 9 weeks were fasted for 4 h prior to an oral gavage with 100 μl corn oil containing 0.25% cholesterol. The animals were sacrificed 4 h post-gavage and duodenum, jejunum, and ileum were isolated. RNA isolation and quantitative real-time PCR analyses were performed as previously described [18]. Primer sequences are listed in Supplementary Table S1.

Histological analyses of aortas, aortic valves, and intestinal sections
Female mice fed a WTD for 9 weeks were anesthetized by i.p. in-jection of sodiumpentobarbital (200 mg/kg body weight). The heart was perfused with 1 mM EDTA in PBS and fixed by perfusion with 10% methanol-free formalin for 15 min. Aortic adventitial adipose tissue was removed, aortic arch and thoracic aorta were excised from the thoracic cavity and bi-valved in a Y-formation. Hearts were fixed with 10% methanol-free formalin for 24 h and stored in 30% sucrose. One day before sectioning, hearts were transferred to Neg-50™ frozen section medium (Richard-Allan Scientific, Kalamazoo, MI). Seven micrometers-thick aortic root serial sections were cut at 20 °C using a cryostat-microtome (HM 560 Cryo-Star; Microm International GmbH, Walldorf, Germany).
Staining of aortas and aortic valve sections with ORO was performed as previously described [16,17]. Aortas were imaged on a black dis-secting wax using a stereo zoom microscope (Olympus SZX12, Tokyo, Japan) equipped with a CCD camera (Olympus DP21, Tokyo, Japan). Aortic valve and intestinal sections were imaged with ScanScope T3 whole slide scanner (AperioTechnologies, Bristol, UK). Plaque areas were quantitated using ImageJ software. MoMa-2 and Masson's tri-chrome staining, imaging, and quantitation were performed as described [16,17].

Statistics
Statistical analyses were performed using GraphPad Prism 5.1 soft-ware. Statistically significant differences were determined by Student's unpaired t-test with Welch correction (in case of unequal variances) for two group comparisons. Multiple group comparisons were calculated by two-way ANOVA followed by Bonferroni correction. Data represent mean values ± SD with the following grades of statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001.   1D) and associated with reduced plasma ApoB abundance ( Supplementary Fig. 2), whereas high-density lipoprotein cholesterol was slightly increased in iDgat1 −/− ApoE −/− mice (Fig. 1D). TG levels remained unchanged between the genotypes upon WTD feeding (Table 1 and Fig. 1E). inhibition of peripheral lipolysis with P407, followed by an oral lipid bolus, resulted in reduced plasma TG and TC concentrations (Fig. 2E). Presuming that CM have a spherical structure, the observed reduction in CM diameter corresponds to an average reduction of 33.5% in the particle volume.  (Fig. 3A), we observed increased radioactivity in duodena and feces of iDgat1 −/− ApoE −/− mice (37% and 60%, respectively) (Fig. 3B). Unaltered abundance of the tracer in the bile (Fig. 3C) argued against a potentiated bioconversion of cholesterol into bile acids and/or its secretion via the hepatobiliary pathway. Except decreased Abcg5, hepatic mRNA expression of genes involved in bile acid synthesis as well as cholesterol and bile acid secretion were comparable between iDgat1 −/− ApoE −/− and ApoE −/− mice (Fig.   3D).  (Fig. 3E) and

Reduced cholesterol uptake and absorption in iDgat1
mRNA expression of genes involved in hepatic lipoprotein uptake, cholesterol uptake and synthesis, and CE synthesis were unaltered (Fig. 3F).
To circumvent the influence of increased gastric retention [9,21] (Fig. 2A) 3H) argues in favor of a mild malabsorption in these animals.

Reduced atherosclerotic plaque size and lesion inflammation in iDgat1 −/− ApoE −/− mice
Finally, we assessed atherosclerosis after 9 weeks of WTD feeding. Plaque formation in the aortic arch (Fig. 4A) and thoracic aorta (Fig. 4B) of iDgat1 −/− ApoE −/− mice was drastically reduced by 56%, as evidenced by ORO staining and quantification of neutral lipid accumulation in the aortas. Atherosclerotic lesion size in the aortic valve sections was decreased by 26% (Fig. 5A). The plaque size reduction was attributable to decreased macrophage abundance (Fig. 5B), whereas collagen con-tent was unaltered (Fig. 5C). Reduced macrophage infiltration into the lesions resulted in 50% reduction of the necrotic core area in the plaques of iDgat1 −/− ApoE −/− mice, as evidenced by decreased distribution of acellular compartments within the lesions (Fig. 5D). Plaque stability, assessed as the ratio of collagen (stabilizing compartment) to necrotic core (inflammatory compartment), was 1.9-fold higher in iDgat1 −/− ApoE −/− compared to control animals (Fig. 5E). Scoring of atherosclerotic lesions according to Whitman et al. [22] revealed that all le-sions from ApoE −/− mice reached stage V, characterized by a highly organized and well-established fibrous cap, lipid-rich core, and pro-nounced necrotic core. In contrast, only 44.5% of lesions from iDgat1 −/− ApoE −/− mice reached stage V, 44.5% were at stage IV of development with less pronounced collagen infiltration into the lesions and poorly defined necrotic core, and 11% were at stage III with plaques consisting almost entirely of foam cells (Fig. 5F).
Taken together, our results demonstrate that in a murine model of atherosclerosis, Dgat1 deficiency exclusively in enterocytes is sufficient to prevent diet-induced hypercholesterolemia without major effects on fasting TG homeostasis. Reduced circulating cholesterol concentrations in ApoB-containing lipoproteins render iDgat1 −/− ApoE −/− mice less susceptible to atherosclerosis development.

Discussion
ApoE −/− mice globally lacking Dgat1 activity are protected from atherosclerosis by at least two mechanisms: (i) reduced macrophage foam cell formation and increased cholesterol efflux, and (ii) reduced uptake of dietary cholesterol [11]. Loss of Dgat1 specifically in the he-matopoietic cells, however, resulted in unaltered atherosclerotic lesion size, but increased intraplaque inflammation independent of plasma cholesterol concentrations [16]. Here, we demonstrate that Dgat1 deficiency solely in enterocytes is sufficient to protect ApoE −/−  [11] and iDgat1 −/− ApoE −/− mice upon WTD feeding is likely the consequence of the resistance to diet-induced obesity caused by ApoE deficiency [23].
Accordingly, unaltered body weights and gonadal adipose tissue mass ( Supplementary Fig.  3), which is the main source of fatty acids as substrate for VLDL synthesis during fasting [24], caused comparable circulating TG concentrations between fasted iDgat1 −/− ApoE −/− and control mice.
Dgat1 −/− mice accumulate neutral lipids in the intestine 2 h after an oral lipid load [25], but not after overnight fasting [9]. Thus, in contrast to the fasting state, postprandial homeostasis of neutral lipids seems to be strongly affected by intestinal Dgat1 deficiency. iDgat1 −/− ApoE −/− mice accumulated neutral lipids in the proximal parts of the intestine 4 h after an oral fat bolus, in line with elevated TG content in Dgat1 −/− jejuna 2 h, but not 6 h after an oil gavage [26]. Transient lipid accu-mulation in the proximal intestinal parts together with delayed post-prandial TG excursions and rapid enterocyte turnover may contribute to elevated fecal caloric and cholesterol loss. Indeed, increased fecal loss of fatty acids [9,21] and cholesterol [9] was observed in iDgat1 −/− mice and in wild-type mice treated with a specific Dgat1 inhibitor [9], but, interestingly, not in Dgat1 −/− mice [4,25].

Europe PMC Funders Author Manuscripts
Europe PMC Funders Author Manuscripts pattern of cholesterol transporters was identical to mice lacking intestinal Mttp [28,29], which are characterized by pronounced postprandial reduction in plasma TG and TC concentrations. Although the exact mechanism of regulation of cholesterol transporters by MTTP remains unclear, similar reductions in plasma cholesterol concentrations were observed upon pharmacologic MTTP inhibition in ApoE −/− mice, Zucker rats, Watanabeheritable hyperlipidemic rabbits, and human subjects (reviewed in Ref. [30]). Unaltered jejunal cholesterol content after [ 3 H]cholesterol gavage may merely be the consequence of decreased apical cholesterol uptake and reduced basolateral release of lipoprotein-bound cholesterol and/or retarded gastric emptying and slower appearance of chyme in the distal parts of the small intestine.
The diminished CM size may cause reduced circulating cholesterol levels in iDgat1 −/− [9] and iDgat1 −/− ApoE −/− mice. Alternatively, smaller CMs in Dgat1 −/− mice may result from defective generation of lumenal lipid droplets (LDs) in the endoplasmic reticulum and, eventually, mature lipoproteins [31]. Indeed, various expression levels of Dgat1 and Dgat2 in the murine small intestine point to the existence of several lipid pools with distinct fates within enterocytes. Notably, Dgat1 overexpression potentiates the formation of lumenal LDs without affecting size or numbers of CMs; Dgat2 overexpression increases cyto-solic TG content and CM number but not their size, whereas Dgat1 deficiency causes TG accumulation in cytosolic LDs and reduction in size and number of secreted CMs [31].
These results suggest that Dgat2 may participate in the initial ApoB lipidation and in directing TG overflow to cytosolic LDs, whereas Dgat1 participates in the formation of lumenal LDs, which eventually fuse with nascent CMs to generate mature, secretioncompetent lipoproteins [31]. It is tempting to speculate that a large proportion of dietaryderived CE within enterocytes is not directly incorporated into nascent CMs, but is rather routed to indirect CM maturation via lumenal LDs. Interestingly, Mttp has been shown to participate in the formation of lumenal LDs [32,33]. Thus, directing TG as Mttp substrate away from lumenal LDs may reduce Mttp transcription and, consequently, influence the expression of cholesterol transporters. VLDL particle size was also reduced in hepatocytespecific Dgat1 −/− mice [34], indicating that Dgat1 is critical for the size of ApoB-containing lipoproteins. Unaltered cholesterol content in VLDL of these mice is in line with differently regulated cholesterol incorporation into ApoB-containing lipoproteins of hepatocytes and enterocytes, since in enterocytes this process is largely dependent on TG incorporation into CMs [29]. inhibitor-treated wild-type mice [9], indicating that this route of cholesterol excretion may indeed be potentiated in iDgat1 −/− ApoE −/− animals. Unchanged radioactivity in the liver or bile of iDgat1 −/− ApoE −/− mice originating from i. v. injected [ 3 H]cholesterol together with unaltered gene expression of bile acid synthetizing and cholesterol and bile acid secreting enzymes argue against potentiated cholesterol excretion via the hepatobiliary pathway, in accordance with genetic ablation or pharmacologic inhibition of Dgat1 in mice [9].
The delay in gastric emptying and gut transit [9,21], however, may cause significant changes in spatiotemporal distribution of orally administered lipids and differently affect intestinal cholesterol pools in iDgat1 −/− ApoE −/− mice and controls. To circumvent any interference of delayed chyme transit on cholesterol absorption and to allow a full cycle of cholesterol uptake, absorption, biotransformation, and elimination, we measured fractional cholesterol absorption 72 h after the substrate bolus. Comparable with Dgat1 −/− , iDgat1 −/− , and Dgat1 inhibitor-treated wild-type mice [9], fractional cholesterol absorption was markedly reduced in iDgat1 −/− ApoE −/− mice, and almost identical to the one found in Dgat1 −/− ApoE −/− mice [11]. We conclude that the reduction of systemic cholesterol load and atherogenesis of iDgat1 −/− ApoE −/− mice are a consequence of decreased assimilation of dietary cholesterol and likely the concerted activity of reduced cholesterol absorption and increased cholesterol excretion.
Is DGAT1 a potential target to treat metabolic disorders, including cardiovascular disease? Despite promising data in mouse models, gastrointestinal side effects in patients treated with DGAT1 inhibitors [36,37] and case reports of DGAT1 mutations associated with congenital diarrhea, an inability to thrive, and even lethal outcomes [38][39][40], argue against DGAT1 inhibition as a therapeutic strategy. The reason probably lies in different enterocyte proteomes of humans and mice, especially with regard to TG-synthesizing enzymes [1,4,38,3,40,41]. Considering that the majority of beneficial metabolic characteristics observed in mice are due to Dgat1 deficiency in enterocytes, pharmacological manipulation of DGAT1 activity in humans is unlikely an ideal thera-peutic approach. Accumulating evidence points toward manipulation of cholesterol metabolism at the intestinal level as a potential novel target to diminish cholesterol burden and, consequently, to treat cardiovascular diseases [27,42,43]. The NPC1L1 inhibitor ezetimibe, which pre-vents cholesterol absorption in the small intestine, is widely used to reduce systemic cholesterol load. Recently, the MTTP inhibitor lomita-pide has been registered on the European and US markets for the treatment of unresponsive homozygous familial hypercholesterolemia, despite some limitations regarding its side effects [44]. Manipulating enterocyte TG metabolism to control plasma cholesterol concentrations, however, might represent an important field of future studies. Increased TICE by high-fat but not high-cholesterol feeding [45] suggests a regu-latory role of TG metabolism on cholesterol homeostasis in enterocytes. Targeting multiple pathways, which indirectly affect cholesterol balance (and particularly ApoB-containing lipoproteins), may be considered as an additional future therapy of hypercholesterolemia and cardiovascu-lar diseases.