Liver-selective γ-secretase inhibition ameliorates diet-induced hepatic steatosis, dyslipidemia and atherosclerosis

Abstract

Hepatic γ-secretase regulates low-density lipoprotein receptor (LDLR) cleavage and degradation, affecting clearance of plasma triglyceride (TG)-rich lipoproteins (TRLs). In this study, we investigated whether γ-secretase inhibition modulates risk of Western (high-fat/sucrose and high-cholesterol)-type diet (WTD)-induced hepatic steatosis, dyslipidemia and atherosclerosis. We evaluated liver and plasma lipids in WTD-fed mice with hepatocyte-specific ablation of the non-redundant γ-secretase-targeting subunit Nicastrin (L-Ncsf). In parallel, we investigated the effect of liver-selective Ncst antisense oligonucleotides (ASO) on lipid metabolism and atherosclerosis in wildtype (WT) and ApoE knockout (ApoE^−/−/) mice fed normal chow or WTD. WTD-fed L-Ncst and Ncst ASO-treated WT mice showed reduced total cholesterol and LDL-cholesterol (LDL-C), as well as reduced hepatic lipid content as compared to Cre- and control ASO-treated WT mice. Treatment of WTD-fed ApoE^−/− mice with Ncst ASO markedly lowered total and LDL cholesterol, hepatic TG and attenuated atherosclerotic lesions in the aorta, as compared to control ASO-treated mice. L-Ncst and Ncst ASO similarly showed reduced plasma glucose as compared to control mice. In conclusion, inhibition of hepatic γ-secretase reduces plasma glucose, and attenuates WTD-induced dyslipidemia, hepatic fat accumulation and atherosclerosis, suggesting potential pleiotropic application for diet-induced metabolic dysfunction.

1. Introduction

The “Westernization” of diet towards consumption of highly palatable, fat, sucrose and cholesterol-rich foods contributes to obesity-related dyslipidemia, defined as elevations in LDL cholesterol (LDL-C) and plasma triglycerides [1]. Dyslipidemia in turn predisposes to cardiovascular disease (CVD) [2–4]. With numerous studies that confirm the hypothesis that reduced LDL-C reduces CVD risk, LDL-C lowering with statins and other agents are mainstay medications for patients with dyslipidemia. Nevertheless, statin efficacy remains incomplete [5–7], at least partially attributable to tolerability, leading to continued need for novel dyslipidemia therapeutic targets [8–10].

Genome-wide association studies (GWAS) and exome-wide association screens have identified genetic loci with population-level association with lipid traits [11–14], which may provide mechanistic insights into development of dyslipidemia and atherosclerosis. One of these studies uncovered variants near the γ-secretase subunit, Nicastrin that were independently associated with HDL and LDL cholesterol levels [14]. The γ-secretase catalyzes intramembrane proteolysis of type 1 transmembrane proteins [15], and is a multi-protein complex containing of redundant catalytic (Presenilin1 or 2) and regulatory (Aph-1a or -1b) subunits as well as unique targeting (Nicastrin; Ncst) and enhancer (Pen2) components. Although γ-secretase is promiscuous, some of the best-studied γ-secretase targets are Alzheimer’s Precursor Protein (APP) and the Notch receptors. As such, small molecule γ-secretase inhibitors (GSI) have been evaluated for dementia, cancer and most recently, metabolic indications [8,16–19], although use is associated with uncertain efficacy as well as gastrointestinal side effects [20–22].

We recently identified that GSI treatment in mice lowered plasma triglyceride-rich lipoproteins (TRLs) by increasing hepatocyte LDL-receptor (LDLR) stability [23], consistent with a role of LDLR in TRL clearance, but LDLR is well documented to regulate LDL-C uptake [24,25] and thus propensity to atherosclerosis [26]. As our prior work was performed in “HDL” models, we could not conclude whether γ-secretase inhibition affects LDL-C clearance in animal models with elevated non-HDL cholesterol. To address this question, we designed a set of experiments to explore whether hepatic γ-secretase inhibition increases LDL-C clearance and reduces the risk of atherosclerosis. Here we show that hepatocyte-specific γ-secretase knockout (albumin-Cre: Nicastrin^flox/flox, henceforth, L-Ncst) mice are less susceptible to Western-type diet (WTD)-induced abnormalities in plasma glucose, total/LDL-C and hepatic steatosis. Similarly, liver-selective Ncst antisense oligonucleotide (ASO) treatment reduced plasma glucose, total and LDL-C and liver lipids in WTD-fed wildtype (WT) mice, as well as aortic plaque development in WTD-fed ApoE^−/− mice, without intestinal metaplasia associated with GSI treatment. In sum, these data support the therapeutic potential of liver-directed γ-secretase inhibition for treatment of multiple diet-induced comorbidities.

2. Materials and methods

2.1. Animals

We crossed albumin-Cre and Nicastrin^flox/flox mice [18,23], both on a homogeneous C57BL/6 background, to generate albumin-Cre; Nicastrin^flox/flox (L-Ncst) mice. C57BL/6 wildtype (strain #000664) or ApoE-deficient mice (strain #002052) were purchased from Jackson Labs. All mice were housed 3-5 animals per cages, with a 12 hr light/dark cycle, in a temperature-controlled environment and were fed normal chow (Purina Mills 5053) or high-fat/sucrose and high-cholesterol (40% fat, 43% carbohydrate and 0.15% cholesterol; Research Diets Inc, D12079B) Western-type diet (WTD). All animal experiments were approved by the Columbia University Institutional Animal Care and Utilization Committee.

2.2. ASO studies

Control and Ncst ASOs were synthesized by lonis Pharmaceuticals [23]. Eight-week-old male C57BL/6 mice fed normal chow or WTD were injected i.p. at 25mg per kg body weight once weekly for indicated times prior to sacrifice.

2.3. Metabolic analyses

We measured blood tail vein glucose by glucose meter (Bayer). We measured triglyceride (ThermoFisher), cholesterol (ThermoFisher), and LDL-cholesterol (Crystal Chem) using colorimetric assays according to the manufacturer’s protocol. We extracted hepatic lipids by the Folch method [27,28] and measured lipids as above.

2.4. Antibodies and western blots

We performed immunoblots of 3 to 7 samples randomly chosen from each experimental cohort with antibodies against LDLR (ab30532) and ApoB (ab20737) from Abcam; ApoE (K23100R) from BioDesign; ApoAI (K23500R) from Meridian Life Science and α-tubulin (T5168) from Sigma-Aldrich.

2.5. Quantitative RT-qPCR

We isolated RNA with TRIzoI (Invitrogen) and synthesized cDNA with High-Capacity cDNA Reverse Transcription kit (Applied Biosystems), followed by quantitative reverse transcriptase PCR with Power SYBR Green PCR master mix (Applied Biosystems) in a CFX96 Real-Time PCR detection system (Bio-Rad).

2.6. Immunohistochemistry

We fixed liver, aortic sinus and small intestine in 4% paraformaldehyde, then incubated in 30% sucrose and embedded with OCT. We stained slides from frozen sections with hematoxylin and eosin (H&E), Oil Red O (ORO) or Periodic acid-Schiff (PAS). We took representative pictures with a light microscope coupled to an AxioCam Camera.

2.7. Quantification and statistical analysis

To assess statistical significance, we performed an unpaired two-tailed t test for comparison of 2 groups, or ANOVA followed by unpaired two-tailed t or Turkey tests for studies involving multiple groups. All data shown as mean ± SEM. Sample size and statistical details can be found in Figure legends.

3. Results

3.1. Hepatocyte-specific γ-secretase knockout mice show reduced plasma cholesterol and liver lipids

In chow fed-mice, we observed that γ-secretase inhibition stabilized LDLR, leading to lower TG but unchanged cholesterol [23]. We hypothesized that WTD-feeding, which provides sufficient dietary cholesterol to modestly increase LDL-C levels [29], might reveal differences in plasma cholesterol content in L-Ncst mice (Fig. 1A), As expected, WTD-fed L-Ncst mice showed reduced Nest expression as compared to Cre- (Nicastrin^flox/flox) controls (Fig. 1B), as well as lower fasting glucose, liver TG and cholesterol than Cre- controls (Fig. 1C–E), despite unchanged body weight (Fig. 1F). Plasma transaminases were unaffected (Fig. 1G), but as hypothesized, WTD-fed L-Ncst mice showed increased liver LDLR, leading to lower plasma cholesterol and LDL-C (Fig. 1H–J).

Fig. 1. Hepatocyte-specific γ-secretase inhibition ameliorates WTD-induced dyslipidemia.

A-J, Experimental outline (A), liver Ncst mRNA expression (B), blood glucose levels (C), liver TG (D) and cholesterol (E), body weight (F), AST (G), serum total (H) and LDL cholesterol (I), and liver LDLR levels (J) in WTD-fed Cre- and L-Ncst mice (n=7-8 mice per group). *P < 0.05, ***P < 0.001 as compared with Cre- control. All data are shown as the means ± SEM.

3.2. Ncst ASO reduces WTD-induced dyslipidemia and hepatic lipid accumulation

These data prompted us to evaluate whether pharmacologic γ-secretase inhibition can similarly protect from WTD-induced dyslipidemia and related metabolic toxicities. As GSI treatment show dose-limiting GI toxicity [21], we utilized a liver-selective γ-secretase antagonist (Ncst ASO) to bypass intestinal distribution [19,23]. Ncst ASO efficiently lowered liver Ncst expression in chow- or WTD-fed WT mice (Fig. 2A and B). Consistent with data from L-Ncst mice, Ncst ASO reduced plasma glucose in chow- and WTD-fed mice (Fig. 2C), and ameliorated WTD-induced increases in plasma cholesterol and LDL-C (Fig. 2D and E). Also similar to WTD-fed L-Ncst mice, WTD-fed Ncst ASO treatment reduced liver TG and cholesterol (Fig. 2F and G). These data suggest that liver-selective γ-secretase inhibition can have multiple metabolic benefits in WTD-fed mice.

Fig. 2. Ncst ASO attenuates diet-induced dyslipidemia.

A-G, Experimental outline (A), liver Ncst mRNA expression (B), blood glucose levels (C), serum total (D) and LDL cholesterol (E), liver TG (F) and cholesterol (G) in control ASO or Ncst ASO-treated mice fed chow or WTD (n=7-8 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001 vs control ASO mice. All data are shown as the means ± SEM.

3.3. Ncst ASO ameliorates dyslipidemia in ApoE^−/− mice

WTD feeding induces relatively mild dyslipidemia. To investigate the effects of γ-secretase inhibition in a model of severe dyslipidemia, we administrated control or Ncst ASO to WTD-fed ApoE^−/− mice (Fig. 3A). In the absence of this critical ligand for chylomicron and VLDL remnant clearance [30], ApoE^−/− mice show marked elevations in VLDL particles and progressive atherosclerosis [31,32]. Consistent with effects in WTD-fed WT mice, Ncst ASO-treated ApoE^−/− mice show reduced Ncst expression (Fig. 3B), blood glucose (Fig. 3C) and plasma cholesterol and LDL-C (Fig. 3D and E). Plasma in mice treated with Ncst ASO was visibly less lipemic (Fig. 3F), and liver lipid accumulation markedly attenuated (Fig. 3G–I).

Fig. 3. Ncst ASO reduces dyslipidemia and hepatic steatosis in ApoE^−/− mice.

A-I, Experimental outline (A), liver Ncst mRNA expression (B), blood glucose levels (C), serum (D) and LDL cholesterol (E), representative photographs of serum (F), liver TG (G) and cholesterol (H) and H&E or Oil-Red O-staining (I) in WTD-fed control or Ncst ASO-treated ApoE^−/− mice (n=7-8 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001 vs control ASO mice. All data are shown as the means ± SEM.

3.4. Ncst ASO ameliorates atherosclerosis in ApoE^−/− mice

As dyslipidemia predisposes to atherosclerosis [33], we tested the hypothesis that γ-secretase inhibition ameliorates arterial wall inflammation and lipid deposition. First, we measured the ApoB100/ApoAI ratio – a representation of balance between atherogenic and anti-atherogenic lipoprotein particles, and a strong risk factor for CVD [34,35] – in chow- and WTD-fed ApoE^−/− mice. We observed that the ApoB100/ApoAI ratio increased with WTD feeding, which was significantly mitigated with Ncst ASO treatment (Fig. 4A and B). Next, we turned our attention to the vessel wall. Immune cell infiltrate, as indicated by Cd45 and Cd68 expression in the aortic arch, was reduced in Ncst ASO-treated ApoE^−/− mice (Fig. 4C). Consistently, Ncst ASO reduced plaque size in the aortic root sinus as compared to control ASO (Fig. 4D and E), without the intestinal metaplasia that plague GSI treatments (Fig. 4F). Collectively, these data suggest that Ncst ASO safely and effectively ameliorates atherosclerosis in WTD-fed ApoE-deficient mice.

Fig. 4. Ncst ASO ameliorates atherosclerosis in ApoE^−/− mice.

A-F, Serum apolipoprotein levels (A), relative ApoB/ApoAI ratios (B) Cd45 and Cd68 mRNA expression in aortic arch (C), representative images of Oil-Red-stained section of aortic root sinus (D) with quantification (E), and Periodic acid-Schiff (PAS) staining of small intestine (F) in WTD-fed control Ncst ASO-treated ApoE^−/− mice (n=7-8 mice per group). *P < 0.05, **P < 0.01 as compared with the indicated control. All data are shown as the means ± SEM.

4. Discussion

The γ-secretase complex regulates intramembrane proteolysis of transmembrane proteins, leading to APP cleavage (to Amyloid-β peptides) and Notch activation [36]. But novel γ-secretase targets are still being discovered. For instance, we recently confirmed in silico predictions that LDLR is cleaved by γ-secretase [23,38], leading to lysosomal degradation and diet-dependent increase in plasma TRLs [23]. In chow and high-fat diet-fed mice, γ-secretase inhibition reduced plasma TG, but with minimal effect on LDL-C [23]. These data led to the hypothesis that models with increased non-HDL cholesterol would be necessary to uncover effects of γ-secretase on LDL-C and atherosclerosis. Indeed, we find that γ-secretase inhibition reduces LDL-C in WTD-fed WT, and more markedly so in ApoE^−/− mice.

In this work, we also observe that γ-secretase inhibition prevents WTD-induced hepatic lipid accumulation, a phenotype absent in chow or HFD-fed mice [23]. One possible explanation for this difference is that dietary cholesterol increases γ-secretase activity [39], which then decreases LDLR stability to drive hepatic lipogenesis and cholesterol biosynthesis [40]. This hypothesis requires further study, including development of enzymatic γ-secretase assays.

We have previously shown that small molecule GSIs ameliorate obesity-induced insulin resistance, mimicking hepatocyte-specific Notch loss-of-function mice [17,23]. As bioavailable GSIs have advanced through clinical trials for Alzheimer’s Disease and cancer, coupled with our work described here, we considered repurposing these medications for obesity-induced metabolic defects. The therapeutic potential of these drugs, however, is undermined by intestinal toxicity. GSI-induced goblet cell metaplasia may be mitigated by concomitant glucocorticoid administration [37], but this is a far less appealing strategy for metabolic as compared to neoplastic indications. Ncst ASO provides an excellent balance between efficacy and safety, simultaneously ameliorating diet-induced hyperglycemia, fatty liver, dyslipidemia and atherosclerosis, without intestinal or other adverse side effects.

Highlights.

• Hepatocyte-specific γ-secretase knockout animals are protected from WTD-induced dyslipidemia and hepatic lipid accumulation

• Ncst ASO decreases WTD-induced dyslipidemia and hepatic lipid accumulation

• Ncst ASO ameliorates dyslipidemia and atherosclerosis in WTD-fed ApoE^−/− mice