Antisense oligonucleotides targeting hepatic angiotensinogen reduce atherosclerosis and liver steatosis in hypercholesterolemic mice

Hepatocyte-derived angiotensinogen (AGT) is the precursor of angiotensin II (AngII). We determined the effects of hepatocyte-specific (N-acetylgalactosamine-conjugated) antisense oligonucleotides targeting AGT (GalNAc AGT ASO) on AngII-mediated blood pressure (BP) regulation and atherosclerosis and compared its effects with losartan, an AngII type 1 (AT1) receptor blocker, in hypercholesterolemic mice. Eight-week-old male low-density lipoprotein (LDL) receptor deficient mice were administered vehicle or GalNAc AGT ASO (1, 2.5, or 5 mg/kg) subcutaneously beginning 2 weeks before the initiation of Western diet feeding. All mice were fed Western diet for 12 weeks. Their systolic BP was monitored by the tail-cuff technique, and the atherosclerotic lesion area was measured by an en face method. Although the effects of all 3 doses of GalNAc AGT ASO on plasma AGT concentrations were similar, GalNAc AGT ASO reduced BP and atherosclerotic lesion size in a dose-dependent manner. Subsequently, we compared the effects of GalNAc AGT ASO (5 mg/kg) with losartan (15 mg/kg/day). Compared to losartan, GalNAc AGT ASO led to more profound increases in plasma renin and reduction in BP but had similar effects on atherosclerosis. Remarkably, GalNAc AGT ASO also reduced liver steatosis, which was not observed in losartan-treated mice. In conclusion, the BP increase and atherosclerosis development in hypercholesterolemic mice are dependent on AngII generated from hepatic AGT. Deleting hepatic AGT improves diet-induced liver steatosis, and this occurs in an AT1 receptor-independent manner.

While the AngII-dependent functions of AGT have been well-documented [1,9] , the function of the des(AngI) AGT portion of AGT has not been extensively studied. Recently, the biological function of des(AngI)AGT has been investigated by infecting hepAGT −/− mice with an adeno-associated virus (AAV) encoding des(AngI)AGT. The expression of des(AngI)AGT in hepAGT −/− mice increased Western diet-induced body weight gain and liver steatosis but had no effects on BP and atherosclerosis [8] . These data support the notion that the effects of AGT deletion on BP and atherosclerosis are AngII-dependent, whereas des(AngI)AGT has effects on metabolic disorders in an AngII-independent manner.
Although hepatocytes are the major source of plasma AGT, its synthesis has also been reported at non-hepatic sites [10][11][12][13] . The degree to which non-hepatic sources of AGT contribute to the above effects remains unknown. Studies have shown that adipocyte-or macrophage-derived AGT deficiency has marginal or no effect on BP and atherosclerosis [8,10,11] . To verify the importance of hepatic AGT from a pharmacological approach, we targeted hepatocyte AGT mRNA, making use of N-acetylgalactosamine (GalNAc)-conjugated AGT ASO [14] , in LDL receptor −/− mice fed a Western diet for 12 weeks. A comparison was made versus losartan, a classic AT1 receptor blocker, to distinguish AngII-dependent and independent effects.

Animals
Male LDL receptor deficient mice (LDL receptor −/− , strain # 002207) were purchased from The Jackson Laboratory (Table S1). Eight-week-old male LDL receptor −/− mice (Table S2) were subcutaneously injected with phosphate-buffered saline (PBS; vehicle) or GalNAc AGT ASO (1, 2.5, or 5 mg/kg) beginning 2 weeks before Western diet feeding (diet # TD.88137, Envigo). Week −2 represents the start of subcutaneous injection of either PBS or GalNAc AGT ASO, while week 0 represents the 1 st week of Western diet feeding ( Figure  1A). On week −2, the mice were injected with vehicle or any of the 3 doses of GalNAc AGT ASO on days 1, 3, and 5, and then the 1 st day of week −1. Two weeks after the initiation of subcutaneous injections (week 0), the mice were fed Western diet for 12 weeks, while vehicle or GalNAc AGT ASO was injected once every week. GalNAc AGT ASO was provided by Ionis Pharmaceuticals Inc. (Carlsbad, CA, USA).
We then tested the rapidity of GalNAc AGT ASO reduction in plasma AGT concentrations. Male C57BL/6J mice (~8 weeks old) were injected with GalNAc AGT ASO 10 mg/kg once (Table S3), and their plasma samples were collected sequentially for 10 days. Subsequently, GalNAc AGT ASO and losartan were compared. Eight-week-old male LDL receptor −/− mice were randomly assigned to three groups (Table S4). The mice in Group 1 (vehicle) were subcutaneously injected with PBS (on days 1 and 3, then once every week) and implanted a pump to deliver water subcutaneously; mice in Group 2 (GalNAc AGT ASO) were subcutaneously injected with GalNAc AGT ASO (5 mg/kg on days 1 and 3, then once every week) and implanted a pump to deliver water subcutaneously; mice in Group 3 (losartan) were subcutaneously injected with PBS (on days 1 and 3, then once every week) and implanted a pump to deliver losartan (15 mg/kg/day; Cat # 61188-100 mg, Millipore Sigma). The mini osmotic pump ALZET Model 2006 (Durect Corp.) was used to deliver water or losartan subcutaneously. The first mini osmotic pumps were replaced by second ones after 6 weeks of infusion. Week −1 represents the start of subcutaneous injection, while week 0 represents the first week of Western diet feeding. On week −1, the mice were injected with PBS or GalNAc AGT ASO on days 1 and 3, and then once each week. Mini osmotic pumps were implanted on day 5 of week −1 to deliver either water or losartan. One week after the initiation of subcutaneous injections (week 0), the mice were fed Western diet for 12 weeks.
All animal experiments (Table S5 and S6) reported in this article were performed with the approval of the University of Kentucky Institutional Animal Care and Use Committee (IACUC protocol number 2018-2968 or 2015-2050).

Systolic blood pressure
Systolic BP was measured in conscious mice using a non-invasive tail-cuff system (BP-2000, Visitech Systems) following our standard protocol [15] and was calculated based on 20 measurements per mouse per day for 3 consecutive days. The mean systolic BP of each mouse from the 3-day measurements was used for data analysis.

Plasma profiles
Mouse blood was collected in the presence of ethylenediaminetetraacetic acid (EDTA; final concentration: 1.8 mg/mL) through submandibular bleeding during the study. At termination, blood was collected through cardiac bleeding through the right ventricle. Plasma AGT concentrations were measured using a mouse AGT enzyme-linked immunosorbent assay (ELISA) kit (Cat # 245718, Abcam), which detects both intact AGT and des(AngI)AGT, thereby measuring total AGT. Plasma renin concentrations in mice were measured by an enzyme-kinetic assay using an AngI ELISA kit (Cat #: IB59131, IBL-America) after being incubated with exogenous recombinant mouse AGT for 1 h.
Plasma alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were measured on the Olympus AU400 clinical analyzer (Olympus, Center Valley, PA) using L-Type ALT.J2 and L-Type AST.J2 reagents and calibrators (Fujifilm Healthcare) in Ionis in a blinded manner.

Quantification of atherosclerosis
After euthanasia, mouse aortas were dissected and fixed in 10% neutrally buffered formalin overnight. Subsequently, adventitial tissues were removed, and the intimal surface was exposed by a longitudinal cut and pinned on a black rubber surface. Images of en face aortas were taken using a digital camera (Nikon Digital Sight DS-Ri1), with a ruler for calibration.
An en face method was used to measure atherosclerotic lesions on the intimal surface of the aorta in accord with the American Heart Association (AHA) statement and as detailed in our standard protocol [16,17] . Atherosclerotic lesions were traced manually from the ascending aorta to the proximal part of the descending thoracic aorta (1 mm distal from the orifice of the left subclavian artery) using Nikon NIS-Elements software (NIS-Elements AR 5.11.00.) under a dissecting microscope.

Histology
At termination, a piece of each liver sample was fixed in paraformaldehyde (4% wt/vol) overnight and then embedded in paraffin. Five-micron sections were used for hematoxylin and eosin (H&E) staining. In addition, a piece of liver (fresh frozen) was embedded in optimal cutting temperature compound (OCT; Cat # 14-373-65, Fisher Scientific), sectioned using a cryostat (Leica CM 1850, Leica) at 10 μm/section, and stained with Oil Red O to visualize neutral lipid accumulation.

Statistical analysis
There were two types of data in this study: non-repeated measures after termination, and repeated measures during the study. Prism v9 (GraphPad Software Inc., La Jolla) was used for non-repeated measures, while R version 4.2.1 was used for repeated measures. Before analyzing non-repeated measures, normality and homogeneous variance assumptions were tested with Shapiro-Wilk and Brown-Forsythe tests, respectively. Since these assumptions were satisfied, all non-repeated data were analyzed using one-way analysis of variance (ANOVA) to compare means among three or more groups, followed by the Sidak post hoc test. For repeated measures, mixed-effect models with inverse-variance weights were used with random intercept and slope for time. A piecewise model was fitted to estimate separate slopes for distinct time points (−2 to 0 and 0 to 11 weeks) in plasma AGT concentrations. The mixed-effect models were run using the lme function in the nlme R package. Data of non-repeated measures were represented as individual data points and mean ± standard error of the mean (SEM). P < 0.05 or Bonferroni-corrected P < 0.05 was considered statistically significant.

N-acetylgalactosamine-conjugated antisense oligonucleotides targeting angiotensinogen reduced blood pressure, atherosclerosis, and Western diet-induced liver steatosis
In the vehicle group, plasma AGT remained unaltered ( Figure 1B) in mice fed either normal laboratory diet (from week −2 to week 0) or Western diet (from week 0 through week 12).
GalNAc AGT ASO significantly reduced plasma AGT concentrations in mice when they were fed normal laboratory diet; Western diet did not alter this outcome. All three doses of GalNAc AGT ASO yielded the same degree of AGT lowering, maximally reducing plasma AGT by ~90%. GalNAc AGT ASO reduced systolic BP in a dose-dependent manner versus vehicle ( Figure 1C), with the effect at 5 mg/kg being significantly larger than that at 1 mg/kg (P < 0.001). Similarly, a dose-dependent reduction in atherosclerotic lesion size was observed ( Figure 1D and E).
LDL receptor −/− mice fed a Western diet developed liver steatosis with increased liver weight and liver cholesterol and triglyceride content [8,[19][20][21][22] . The results obtained in this study (Figure 2A-C) confirmed this outcome. GalNAc AGT ASO administration reduced liver weight and liver total cholesterol and triglyceride content (Figures 2A-C) in a doseindependent manner. H&E staining and Oil Red O staining revealed diminished neutral lipid accumulation in mice administered GalNAc AGT ASO at all doses compared to the vehicle group ( Figure 2D).

Comparisons between N-acetylgalactosamine-conjugated antisense oligonucleotides targeting angiotensinogen and losartan
A single dose of GalNAc AGT ASO reduced plasma AGT maximally within 3 days after the injection ( Figure 3A). Therefore, when comparing GalNAc AGT ASO (5 mg/kg) and losartan, we began feeding the Western diet 1 week after the initiation of GalNAc AGT ASO ( Figure 3B). GalNAc AGT ASO significantly reduced plasma AGT concentrations, while losartan had no effect ( Figure 3C). A similar pattern was observed for hepatic AGT mRNA abundance ( Figure 3D). As expected, both drugs increased plasma renin concentrations and renal renin mRNA abundance ( Figure 4A and B), but the effects of GalNAc AGT ASO was greater than that of losartan. This coincided with the effect on systolic BP ( Figure  4C), which was lowered more significantly by GalNAc AGT ASO. In contrast, both drugs equivalently reduced atherosclerotic lesion areas ( Figure 4D and E).
Neither drug increased plasma ALT or AST concentrations (Supplementary File, Figure S1), indicating that GalNAc AGT ASO and losartan do not cause liver damage. Consistent with observations in the first study (Figure 2), GalNAc AGT ASO reduced liver weight and lipid accumulation in the liver ( Figure 5). Compared to the vehicle group, losartan did not affect liver weight, nor reduced the total cholesterol, triglycerides, or the severity of neutral lipid accumulation in the liver ( Figure 5).

Discussion
This study demonstrated that GalNAc AGT ASO reduced systolic BP and atherosclerosis in LDL receptor −/− mice fed a Western diet. The effects of GalNAc AGT ASO 5 mg/kg were more pronounced in reducing systolic BP than losartan (15 mg/kg/day). This is likely because this ASO dose induced a greater degree of renin-angiotensin blockade, reflected by the higher renin increase observed after administering GalNAc AGT ASO compared to losartan. Of note, the effects of both drugs on atherosclerosis were not different. Although high BP is an independent risk factor for atherosclerosis, there is a growing body of evidence suggesting that BP per se does not contribute to atherosclerosis [23] . The different magnitudes of suppressing AGT versus blocking AT1 receptor on BP and atherosclerosis in this study support this notion. Overall, this study unequivocally shows that BP regulation and atherosclerosis in hypercholesterolemic LDL receptor −/− mice are dependent on AngII generated from AGT of hepatic origin, thereby agreeing with the observations made in mice displaying genetic hepatocyte-specific AGT deficiency [8,24,25] .
Remarkably, GalNAc AGT ASO, but not losartan, reduced Western diet-induced liver steatosis, which was manifested as increased cholesterol and triglyceride content in the liver. This finding is consistent with our previous studies in hepAGT −/− mice [8,19] . These effects occurred in a dose-independent manner and likely represent the direct consequence of lowering AGT, which occurred following GalNAc AGT ASO administration. Indeed, the expression of des(AngI)AGT induced liver steatosis in hepatocyte-specific AGT deficient mice fed a Western diet [8] . Given the liver-specific activity of GalNAc AGT ASO [14] , we would expect non-hepatic AGT to have no effect on liver steatosis.
It is important to note that mice, unlike humans, display a very high turnover of AGT, reflected by lower plasma AGT concentrations than in humans. Hence, the des(AngI)AGT/ intact AGT ratio is higher in mice than in humans [26] . Although administration of losartan did not affect AGT and liver steatosis in the present study, earlier studies have reported suppression of liver steatosis in mice administered with RAS inhibitors [27][28][29] . This may reflect a different and larger degree of renin-angiotensin blockade, potentially lowering AGT [30] and/or affecting the des(AngI)AGT/intact AGT ratio. The molecular mechanism underlying the direct effect of des(AngI) AGT is unknown. It may involve interference with the protein kinase B (Akt)/mammalian target of rapamycin (mTOR)/sterol responsive element-binding protein 1c (SREBP-1c) pathway [19] , which is associated with liver steatosis and suppressed in mice with hepatocyte-specific deletion of AGT. It is also unknown whether intact AGT exerts similar effects. Studying this would require the AAV-induced expression of intact AGT in hepAGT −/− mice under complete renin-angiotensin blockade.

Conclusions
GalNAc AGT ASO not only reduces BP and atherosclerosis, but also improves diet-induced liver steatosis in mice. Future studies should investigate whether AGT suppression induces similar effects in humans and nonhuman primates.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.