Modeling hypercholesterolemia and vascular lipid accumulation in LDL receptor mutant zebrafish[S]

Elevated plasma LDL cholesterol is the dominant risk factor for the development of atherosclerosis and cardiovascular disease. Deficiency in the LDL receptor (LDLR) is a major cause of familial hypercholesterolemia in humans, and the LDLR knockout mouse is a major animal model of atherosclerosis. Here we report the generation and characterization of an ldlr mutant zebrafish as a new animal model to study hypercholesterolemia and vascular lipid accumulation, an early event in the development of human atherosclerosis. The ldlr mutant zebrafish were characterized by activated SREBP-2 pathway and developed moderate hypercholesterolemia when fed a normal diet. However, a short-term, 5-day feeding of ldlr mutant larvae with a high-cholesterol diet (HCD) resulted in exacerbated hypercholesterolemia and accumulation of vascular lipid deposits. Lomitapide, an inhibitor of apoB lipoprotein secretion, but not the antioxidant probucol, significantly reduced accumulation of vascular lipid deposits in HCD-fed ldlr mutant larvae. Furthermore, ldlr mutants were defective in hepatic clearance of lipopolysaccharides, resulting in reduced survival. Taken together, our data suggest that the ldlr knockout zebrafish is a versatile model for studying the function of the LDL receptor, hypercholesterolemia, and related vascular pathology in the context of early atherosclerosis.—

but their use for drug screening is limited because expensive, time-consuming, and labor-intensive procedures are involved in the analysis of atherosclerosis. In contrast, the use of zebrafish is cost-effective because of large progeny numbers and low maintenance costs. Transparent zebrafish larvae are suitable for live imaging, and a quick staining protocol for lipids and easy genetic manipulation allow for effective screening. Recently, new zebrafish models emerged to investigate aspects of lipoprotein metabolism, lipid abnormalities, and atherosclerosis (13)(14)(15). Importantly, genes involved in lipid and lipoprotein metabolism, including APOB, APOA family, APOC2, LPL, CETP, and LDLR, are conserved between humans and zebrafish (16)(17)(18).
Particularly, Ldlr, which is responsible for LDL uptake in liver, has been reported to be functionally conserved in zebrafish. Transient knockdown of ldlr with antisense morpholino oligonucleotides leads to higher LDL-C levels and vascular lipid deposits after feeding a high-cholesterol diet (HCD) (19). To further improve robustness and simplify the experimental procedure, we generated a genetic zebrafish model in which an ldlr mutation achieved by using the CRISPR/Cas9 approach resulted in hypercholesterolemia and robust vascular lipid accumulation following a very short, 5-day HCD feeding. Our data provide evidence that the ldlr mutant zebrafish is a versatile model to study hypercholesterolemia and related vascular pathology.

Zebrafish maintenance and feeding
Adult zebrafish of the AB strain were maintained at 28°C on a 14 h light/10 h dark cycle and fed brine shrimp twice a day. Zebrafish larvae were fed Golden Pearls (100-to 200-m size from Brine Shrimp Direct) twice a day, starting from 4.5 days postfertilization (dpf). For HCD feeding, 4% weight per weight (w/w) cholesterol (Sigma, 362794) were added to Golden Pearls, as reported (14,20). For drug treatment experiments, lomitapide (Cayman, 10009610) or probucol (Cayman, Catalog no. 15043), together with cholesterol, were dissolved in diethyl ether (Spectrum, E1010) and thoroughly mixed with Golden Pearls. After diethyl ether evaporation, content of lomitapide was 0.14% w/w, and probucol 0.05% w/w. For visualization of vascular lipid deposits, the same method was used to supplement Golden Pearls with 1 g/g of a fluorescent cholesteryl ester analog (cholesteryl BODIPY 576/589-C11, Invitrogen, C12681), as previously described (13). All animal studies were approved by the University of California, San Diego institutional animal care and use committee.

Triglyceride and total cholesterol measurements and lipoprotein analysis
Blood was collected from adult zebrafish, 4 h after feeding, through tail amputation and diluted 1:200 in PBS. Five dpf larvae were gently homogenized in PBS with pestle. After centrifugation, supernatants were collected and referred to as "homogenate." Triglyceride (TG) and total cholesterol levels in diluted plasma or larvae homogenate were measured with kits according to manufacturer's protocol (Biovision, Triglyceride Quantification Kit, K622-100; Cholesterol Quantification Kit, K623-100). Lipoprotein fractions were assessed using native agarose gel electrophoresis (Helena Laboratories, 3045) as we previously described (13).

Fast protein liquid chromatography lipoprotein profile
A total of 50 to 60 l of pooled plasma from 20 to 30 adult WT, ldlr mutant, and apoc2 mutant zebrafish were loaded onto a Superose 6 PC 3.2/30 column (GE Healthcare Life Science, 17-0673-01), and total cholesterol and TG levels were determined in each fraction (250 l), collected at a flow speed of 0.5 ml/min.

Oil Red O and BODIPY staining
For Oil Red O (ORO) staining, embryos or larvae were fixed in 4% paraformaldehyde for 2 h, washed three times in PBS, incubated in 0.3% ORO solution for 2 h, and then washed with PBS before imaging. For BODIPY staining, live larvae were immersed in E3 water (5.0 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 , 0.33 mM MaSO 4 ) containing 0.1 g/ml BODIPY 505/515 (Invitrogen, D-3921) for 1 h in the dark and then rinsed with E3 water before imaging.

LPS treatment
Native Escherichia coli O111:B4 lipopolysaccharide (LPS) (Invitrogen, L4130), Pseudomonas aeruginosa LPS (Invitrogen, L9143), and a fluorescent conjugate of LPS (Alexa Fluor 568-LPS/ Alexa-LPS, Invitrogen, L23352) were dissolved in PBS to a stock concentration of 20 mg/ml. For intravenous LPS injection, 4.5 dpf embryos were laterally aligned in 0.5% low melting temperature agarose, and 5 nl of Alexa-LPS (1 mg/ml) were injected into the circulation through the cardinal vein using a Femtojet microinjector (Eppendorf). For the LPS survival challenge, 5 dpf larvae were immersed in water with different concentrations of E. coli LPS and P. aeruginosa LPS (100-1,000 g/ml). Larvae survival, assessed by detectable heartbeat) was determined every hour after the immersion.

Imaging of live embryos or larvae
For in vivo microscopy, anesthetized embryos or larvae were mounted in 0.5% low melting temperature agarose (Sigma) containing tricaine (0.02%, Sigma) in 50 mm glass-bottom dishes (MatTek). Images were captured with a Leica CTR5000 or a BZ9000 Keyence fluorescent microscope. To quantify vascular lipid deposits, we captured a z-stack of 15 images in the trunk/tail region with a ×20 objective (BZ9000 Keyence). After overlaying images in the z-stack, using the BZ-X analyzer software (Keyence), vascular lipid deposits were manually counted in five consecutive vascular segments posterior to the anus.

Generation of an ldlr mutant with the CRISPR/Cas9 system
To create a loss-of-function ldlr mutant in zebrafish, we chose to use the CRISPR/Cas9 system (21). In zebrafish, there are two ldlr orthologs: ldlra (NCBI gene ID: 387529) and ldlrb (NCBI gene ID: 393460), with ldlra showing the higher degree of conservation with the human LDLR gene. The protein sequences between zebrafish Ldlra and human LDLR are conserved with an identity of 51.7% (supplemental Fig. S1), and the zebrafish Ldlra is predicted to have the conserved ligand-binding domain, epidermal growth factor precursor homology domain, and transmembrane domain. Hereinafter in this article, we designate ldlra as the ldlr gene in zebrafish. To disrupt Ldlr protein function with CRISPR/Cas9, we targeted the 5-prime of the ldlr coding region. Five ldlr genomic targets were chosen, and their corresponding gRNAs were synthesized. With T7 endonuclease digestion assay, we found one target, which is localized in exon 5, that was efficiently edited after Cas9/gRNA injection. The F0 founder fish were raised and outcrossed with WT to obtain F1 generation. F1 were further outcrossed with WT to obtain F2 heterozygotes carrying the same mutation (Fig. 1A). In F1 generation, we screened the ldlr mutants and found one mutant line in which a fragment of 10nt was deleted, leading to a prestop codon (Fig. 1B). The truncated protein caused by the 10nt deletion and prestop codon in the ldlr mutant lost the whole C-terminal region, which is required for normal LDLR protein function (Fig. 1C). It is reported that frameshift indels may lead to in-frame exon skipping and alternative splicing (22). To check whether this 10nt deletion in exon 5 affected the splicing of the ldlr transcript, we designed primers for exon 3 and exon 7 to include exon 5 and the two neighboring exons. Semiquantitative PCR results from adult liver indicated decreased ldlr expression but no alternative splicing in the ldlr mutant (Fig. 1D), validating the predicted prestop codon. Therefore, this mutant line with the 10nt deletion was defined as our loss-of-function ldlr mutant and used in the following studies.

Hypercholesterolemia and activation of hepatic SREBP-2 pathway in adult ldlr mutants
It is well established that mutations in the LDLR C-terminal region that disrupt LDLR anchoring or processing cause FH in human patients (4). To test whether truncated zebrafish Ldlr protein affects cholesterol homeostasis, we measured plasma lipid levels of the progenies from ldlr heterozygotes in-cross. Consistent with human and mouse studies, adult zebrafish homozygous ldlr mutants had significantly higher total cholesterol levels than did WT or ldlr heterozygote siblings when fed normal diet ( Fig. 2A). Plasma triglyceride levels tended to be higher in homozygous ldlr mutants but were not significantly different from those in WT or heterozygotes (Fig. 2B). In agreement, neutral lipid staining of plasma separated on native gel suggested higher LDL, but lower VLDL in ldlr mutants, when compared with WT (Fig. 2C). Fast protein liquid chromatography data indicated that ldlr mutants had a much higher cholesterol levels in the IDL/LDL fractions. In contrast, apoc2 mutants, reported in our earlier work (13) and used here as a control, showed higher cholesterol levels in the VLDL fraction (Fig. 2D). The TG levels were mildly increased in the LDL fraction in ldlr mutants but dramatically increased in the VLDL fraction in apoc2 mutants, as was expected (Fig. 2E).
The LDLR is required for extracellular cholesterol uptake and regulation of the SREBP-2 pathway in the liver, the major organ responsible for plasma cholesterol regulation (3,8). We next isolated liver from adult males and extracted total RNA for qPCR. The ldlr mRNA levels were reduced by 75% in ldlr mutants, likely due to nonsensemediated mRNA decay (23). Interestingly, although srebp2 expression itself was not significantly changed, expression of hmgcr, the gene encoding HMG-CoA reductase and a major target gene of SREBP-2, was increased by as much as 13-fold (Fig. 2F).

Hyperlipidemia and activation of SREBP-2 pathway in ldlr mutant larvae
We next tested whether the ldlr mutant larvae had hyperlipidemia. Compared with WT, 5 dpf ldlr mutant larvae showed stronger vascular ORO and BODIPY staining (Fig. 3A, B), indicating elevated neutral lipid levels in the circulation. In agreement, total cholesterol and triglycerides in larvae homogenates were significantly higher in ldlr mutants than in WT (Fig. 3C, D). To test SREBP-2 activation, we extracted total RNA for qPCR from 4 dpf WT and ldlr mutant larvae. There were no significant changes in srebp1 or srepb2 mRNA expression. However, there was a 4-fold increase in hmgcr expression and a 2-fold increase in fatty acid synthase (fasn) expression in ldlr mutants. Interestingly, expression of pcsk9, another SREBP-2 target, was not changed in ldlr mutants (Fig. 3E).

Increased vascular lipid accumulation in ldlr mutant larvae following short-term high-cholesterol feeding
We and others used zebrafish larvae to monitor vascular lipid accumulation in vivo (13,14,19,20). In our previous studies, WT embryos were fed HCD for 2 weeks before imaging vascular lipid deposits (14). Considering the hypercholesterolemia in nonfed ldlr mutant larvae, we tested whether an even shorter-term HCD feeding would induce vascular lipid accumulation. WT and ldlr mutants were fed control diet or HCD starting at 4.5 dpf. After a 5-day feeding, 9 dpf live larvae were immobilized in low-melting temperature agarose and imaged, as described (14,20). When fed control diet, ldlr mutant larvae showed a moderate but significant increase in vascular lipid deposits in comparison with WT. When fed an HCD, ldlr mutant larvae displayed a dramatic increase in lipid deposits. In contrast, WT larvae did not accumulate vascular lipid deposits within this short time frame of HCD challenge (Fig. 4A, B). There were no apparent morphological or body-size changes in either WT or ldlr mutant larvae following HCD feeding (supplemental Fig. S2).
Our results indicate that the ldlr loss-of-function mutation has resulted in an increased susceptibility to a dietary cholesterol challenge, closely resembling the phenotypes observed in Ldlr / mice (11,12). To evaluate whether the new animal model in which ldlr mutant zebrafish are subjected to short-term (5 days) HCD feeding can be useful for drug screening, we tested effects of probucol, an antioxidant, and lomitapide, an MTP inhibitor, on vascular lipid accumulation. Both probucol and lomitapide have been shown to exert antioxidant and MTP inhibitor properties, respectively, in zebrafish (20,24). As was expected, lomitapide decreased the plasma lipid levels in ldlr mutants, as assessed by ORO staining (Fig. 5A), by blocking their absorption in the intestine. Vascular lipid deposits were not decreased by probucol treatment but were significantly decreased by the treatment with lomitapide (Fig. 5B, C), suggesting that lipid levels, but not lipid oxidation, play a dominant role in early vascular lipid accumulation event in loss-of-function ldlr mutant larvae.

Decreased hepatic clearance of LPS in ldlr mutant larvae
LPS are found in the outer membrane of gram-negative bacteria and elicit strong immune response in animals, in extreme cases resulting in septic shock (25). Previous studies have shown that hepatocytes are responsible for plasma LPS clearance, and this process is dependent on LDLR and its regulator, PCSK9 (26,27). We tested these hypotheses using our new ldlr mutant zebrafish. At 4.5 dpf, WT and ldlr larvae were injected with red fluorescent Alexa-LPS. After 2 days, fluorescence signals were detected in both liver and vasculature (Fig. 6A). A ratio of Alexa-LPS fluorescence intensity in the liver to that in the vasculature was used as an index for hepatic LPS uptake. The ldlr mutant larvae had a decreased hepatic uptake of Alexa-LPS in comparison with WT (Fig. 6B). To test whether decreased hepatic clearance of LPS affects tolerance of zebrafish larvae to LPS challenge, we tested larvae survival following LPS exposure. Consistent with a previous report (28), we found that P. aeruginosa LPS induced more larval death than did the same doses of E. coli LPS (supplemental Fig. S3). The ldlr mutant larvae exposed to P. aeruginosa LPS had a lower survival than did WT (Fig. 6C). These results suggest that Ldlr-mediated hepatic clearance of LPS is important for zebrafish survival.

DISCUSSION
The goal of this work was to establish a genetic model of hypercholesterolemia in zebrafish. The ldlr mutant zebrafish developed in this study adds to the set of zebrafish models of lipid abnormalities and lipoprotein oxidation, which includes apoc2 mutant zebrafish that develop severe hypertriglyceridemia (13), loss and gain of function liver X receptor mutants (29,30), and hsp70:IK17-EGFP zebrafish that serve as a reporter for oxidation-specific epitopes (20), among others. We have previously been able to achieve hypercholesterolemia in WT zebrafish by feeding larvae an HCD for 2 weeks, which resulted in accumulation of vascular lipid deposits (14). This approach was further advanced by the introduction of a protocol in which an antisense morpholino oligonucleotide-mediated knockdown of ldlr allowed for shorter HCD feeding times to achieve meaningful hypercholesterolemia and vascular lipid accumulation (19). The introduction of our new ldlr mutant reduces the variability associated with morpholino oligonucleotide injections and feeding protocols and allows for achieving a robust and consistent phenotype.
The mutation we introduced into the zebrafish ldlra gene created a stop codon and resulted in a truncated, nonfunctional, and rapidly degraded ldlr transcript. The ldlr Fig. 5. Effects of probucol and lomitapide on hyperlipidemia and vascular lipid accumulation. A: ORO staining of 9 dpf larvae, following 5-day feeding with HCD, supplemented with probucol and lomitapide. Green arrows point to dorsal aorta, and yellow arrows point to intestine. Scale bar =500 m. B, C: Representative images and quantitative data for vascular lipid deposits. Scale bar = 50 m. Mean ± SEM (n = 8-10 in each group). ns, P > 0.05; **P < 0.01; ***P < 0.001 (Student's t-test). Fig. 6. Reduced hepatic uptake of LPS and survival of ldlr mutants upon LPS challenge. A: Representative images of 6.5 dpf WT and ldlr mutant larvae following Alexa-LPS injection (5 nl, 1 mg/ml) at 4.5 dpf. White and green arrows point to fluorescent signals in liver and the tail vasculature, respectively. Scale bars = 200 m. B: Quantification of the ratio of fluorescent intensity in liver to that in vasculature. Mean ± SEM (n = 6 for WT and n = 5 for ldlr mutant). **P < 0.01 (Student's t-test). C: Survival curves for 5 dpf WT and ldr mutant larvae following immersion in water with 500 m/ml P. aeruginosa LPS (n = 13 for WT and n = 18 for ldlr mutants). P < 0.05 (Log-rank Mantel-Cox test). mutants fed a normal diet had moderate hypercholesterolemia and activated the SREBP-2 pathway, as is evident from the increased expression of hmgcr. Remarkably, as short as a 5-day HCD feeding resulted in robust vascular lipid accumulation in ldlr mutants. Lomitapide, but not probucol, prevented HCD-induced accumulation of vascular lipid deposits. We propose that a protocol using 5-day HCD feeding of ldlr mutant larvae can be used for mechanistic studies, as well as for initial genetic or drug screening to identify new therapeutic targets.
We also suggest that the advantage of using zebrafish is in the close similarity between zebrafish and human lipoprotein metabolism, as well as the differences that increase zebrafish propensity to developing hypercholesterolemia. Unlike mice, zebrafish express cholesteryl ester transfer protein, which helps retain cholesterol esters in the circulation. In addition, though mouse ApoB proteins are mostly ApoB48, which causes a rapid hepatic cycling of VLDL and low-atherogenic LDL levels (31), zebrafish do not express the apoB mRNA editing enzyme APOBEC1 (apolipoprotein B mRNA editing enzyme 1), and the majority of zebrafish ApoB protein is likely ApoB100, as our earlier studies suggest (32). ApoB100-containing LDL and VLDL remnants have a longer half-life time in plasma, adding to their atherogenicity. Indeed, the lipoprotein profile of normal dietfed ldlr mutants shows a substantial IDL/LDL fraction, surpassing that seen in chow-fed Ldlr / mice. This may explain, in part, why ldlr zebrafish larvae are so sensitive to HCD feeding and accumulate vascular lipid deposits within a short-time frame.
Interestingly, we found that HMG-CoA reductase was consistently and dramatically upregulated in ldlr mutant. However, other srebp2 genes, including pcsk9, were not significantly changed. We propose that different target genes are regulated by SREBP-2 in a different manner, and it will be intriguing to explore the underlying mechanisms, which might provide new insights and suggest new approaches for differential targeting of genes regulating cholesterol homeostasis. Another interesting observation was the lack of the effect of probucol treatment on vascular lipid accumulation in 5-day HCD-fed ldlr mutants. This was in contrast to our previous results with 15-day HCD-fed WT zebrafish (20). This discrepancy may be explained, in part, by the different time scale of these experiments and disrupted hepatic uptake of LDL in ldlr mutants, which affects the LDL particle size and its vulnerability to oxidation.
In summary, this work introduces a new genetic model of hypercholesterolemia and early atherogenesis in which the ldlr zebrafish mutants subjected to a short, 5-day HCD feeding results in robust and consistent accumulation of vascular lipid deposits. We propose that this new animal model can be used for mechanistic studies and for the screening of new therapeutic targets and treatments.