A novel nanobody-heavy chain antibody against Angiopoietin-like protein 3 reduces plasma lipids and relieves nonalcoholic fatty liver disease

Background Nonalcoholic fatty liver disease (NAFLD) is a metabolic disease mainly on account of hypercholesterolemia and may progress to cirrhosis and hepatocellular carcinoma. The discovery of effective therapy for NAFLD is an essential unmet need. Angiopoietin-like protein 3 (ANGPTL3), a critical lipid metabolism regulator, resulted in increased blood lipids and was elevated in NAFLD. Here, we developed a nanobody-heavy chain antibody (VHH-Fc) to inhibit ANGPTL3 for NAFLD treatment. Results In this study, we retrieved an anti-ANGPTL3 VHH and Fc fusion protein, C44-Fc, which exhibited high affinities to ANGPTL3 proteins and rescued ANGPLT3-mediated inhibition of lipoprotein lipase (LPL) activity. The C44-Fc bound a distinctive epitope within ANGPTL3 when compared with the approved evinacumab, and showed higher expression yield. Meanwhile, C44-Fc had significant reduction of the triglyceride (~ 44.2%), total cholesterol (~ 36.6%) and LDL-cholesterol (~ 54.4%) in hypercholesterolemic mice and ameliorated hepatic lipid accumulation and liver injury in NAFLD mice model. Conclusions We discovered a VHH-Fc fusion protein with high affinity to ANGPTL3, strong stability and also alleviated the progression of NAFLD, which might offer a promising therapy for NAFLD. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s12951-022-01456-z.

Background Hypercholesterolemia, accompanied by increased total cholesterol (TC) including low-density lipoprotein (LDL), shows positive correlation with nonalcoholic fatty liver disease (NAFLD) [1,2]. As one of the major metabolic organs, liver is closely regulated in response to hypercholesterolemia, undergoes steatosis and then evolves into NAFLD [3]. NAFLD is the most prevalent liver disease and characterized by abnormal hepatic triglyceride (TG) and cholesterol accumulations. Its pathology ranges from nonalcoholic steatohepatitis to cirrhosis and even progresses to hepatocellular carcinoma [4]. Although there have been advances in lipidlowering therapy including peroxisome proliferation activating receptors (PPAR)-α/γ agonists and statins, NAFLD still afflicted 1.7 billion population worldwide [5][6][7]. Therefore, the novel therapeutics for NAFLD are imminently craved.
Angiopoietin-like protein 3 (ANGPTL3), as a lipometabolic regulator, was a secretory protein and primarily expressed in the liver [8]. It could be cleaved into a coiledcoil domain (CCD) and a fibrinogen-like domain (FLD). The CCD and the full-length of ANGPTL3 all could inhibit the enzyme activity of lipoprotein lipase (LPL) as well as endothelial lipase (EL) [8][9][10][11]. The individuals with ANGPTL3-function deficiency existed natively with low plasma TG and TC levels and displayed no clinical risks [12,13]. It had been well documented that the high-cholesterol diet could elevate ANGPTL3 expression levels as a consequence of liver X receptors (LXRs) activation, which aggravated hypertriglyceridemia [14,15]. Besides, ANGPTL3 had increased expression levels in the liver of NAFLD patients, which resulted in elevated circulating ANGPTL3, indicating ANGPTL3 may be a critical target in NAFLD [16]. The ANGPTL3 inhibitors could efficiently alleviate plasma TG and TC levels in hypercholesterolemic patients and relieve the progress of atherosclerosis [17,18].
Up to now, three diverse blockade tactics against ANGPTL3 have been validated, including antisense oligonucleotide (ASO, IONIS-ANGPTL3-LRx), the CRISPR/Cas9 gene knockout technology and the monoclonal antibody [19][20][21]. Unfortunately, some of them have severe limitations and their druggability remain controversial. As was reported previously, the ASObased drugs were easily degradable and the CRISPR/ Cas9 gene knockout technology had off-target effects and ethical limitations [22,23]. In addition to conventional antibody, nanobody is a camelids single domain antibody (VHH) at the nanometer scale, which has various advantages over the classical immunoglobulin gamma (IgG) [24][25][26]. It could be stably expressed with high yield in mammalian cell, yeast and even bacteria [27,28]. It has efficient penetration even through the blood-brain barrier due to nanometer scale, superior thermostability and chemostability, preeminent specificity and binding-ability to antigen [29][30][31][32][33][34]. Lately, the first nanobody drug, a bivalent VHHs fusion protein (caplacizumab) was approved to treat thrombotic thrombocytopenic purpura by the European Commission [35]. Thus, VHH fusion protein blocking ANGPTL3 may be a promising treatment for hypercholesterolemic and NAFLD.
In this manuscript, we aim to discover the therapeutic effect of VHHs fusion protein against ANGPTL3 on NAFLD. After immunizing an alpaca with human ANGPTL3 (S17-K170) (hANGPTL3), we retrieved the VHH binding to the CCD of ANGPTL3 with high affinity, which were then fused to human IgG1-Fc to establish the fusion protein. Our results indicated that the VHH-Fc fusion protein could block ANGPTL3-mediated suppression of LPL activity in vitro, reduce serum lipid levels and ameliorated hepatic lipid accumulation and liver injury in NAFLD mice (Fig. 1).

Construction of the VHHs phage library
After immunizing an alpaca for four rounds of hANGPTL3 (S17-K170), the serological ELISA was performed with 200 ng of hANGPTL3 (S17-K170) antigen/ well. As shown in Additional file 1: Fig. S1, the titer of serological antibody specific to the hANGPTL3 (S17-K170) was 1:64,000. The immune VHH library was developed subsequently and the library size was approximate 5 × 10 9 by counting the number of transformants of E. coli TG1 with ten-fold serial dilution. Furthermore, the VHH genes insertion rate of library were measured by PCR, which was determined to be 97.91% through randomly picking 48 clones (Additional file 1: Fig. S2). After three rounds of bio-panning using hANGPTL3 (S17-220P)-His6, 33 candidate clones specifically binding with hANGPTL3 were identified (Fig. 2a). The 33 positive clones were sequenced and their CDR3 domains were analyzed, and we obtained three nanobodies (C27, C44 and C46) with unique CDR3 domains.

Expression of VHH-Fc and affinity test
To extend the half-life of nanobody, C44 was fused to the Fc domain of human IgG1 to construct the alpaca-human chimeric antibody C44-Fc (Fig. 3a). The C44-Fc was then expressed in ExpiCHO cells using a commercial pTT5 vector. The SDS-PAGE results in Fig. 3b demonstrated Fig. 1 Schematic presentation of the whole experiment. The whole experimental procedure includes immunization with ANGPTL3, B-cells separation, VHH library construction, phage-display to select ANGPTL3-specific VHH, sanger sequencing, VHH and VHH-Fc fusion proteins expression, affinity determination, in vitro LPL assay and pharmacodynamic evaluation in hypercholesterolemic and NAFLD mice model . c-e The affinity of C27-His6, C44-His6 and C46-His6 binding to hANGPTL3 (S17-P220)-His6. f The assay for hANGPTL3-binding epitopes of the C44-His6 was performed against evinacumab by the dual antibody sandwich ELISA. The horizontal and vertical axes represent the concentrations of the C44-His6 and the OD450 value that the C44-Fc fusion protein was appropriately produced with high purity (78 kDa in non-reduced PAGE and 39 kDa in reduced PAGE). As both the CCD and the full-length of ANGPTL3 proteins could inhibit the LPL activity, we also investigated the affinity of C44-His6 and C44-Fc to the human or mouse ANGPTL3 domains as well as CCD fragments. C44-Fc had excellent affinity to mANGPTL3 (S17-T455)-His10 at 0.1600 nM, hANGPTL3 (S17-P220)-His6 at 0.2842 nM, hANGPTL3 (S17-E460)-His10 at 0.4440 nM, mANGPTL3 (S17-T206)-His6 at 1.180 nM and (Fig. 3c-m), displaying better affinities comparing with the monomeric VHH C44-His6. Compared with evinacumab, C44-Fc displayed ~ 3 times higher affinities to the CCD fragment of hANGPTL3 and mANGPTL3, but a lower affinity to the CCD fragment of mANGPTL3. As a result, C44-Fc could crossly bind to human and mouse ANGPTL3 proteins as well as CCD fragment at high affinities.

C44-Fc shown strong stability and inhibitory effect on ANGPTL3
At first, the molecular size variations of recombinant C44-Fc protein were analyzed by SEC-HPLC. From the Fig. 4a, the C44-Fc had 100 percent of the main peak purity without aggregation or fragment and its molecular size was smaller than evinacumab. Thereafter, the yield and stability of C44-Fc were further analyzed. In the industry-standard ExpiCHO cell system, the fusion proteins C44-Fc reached its expression level of 500 mg/L, which was roughly twice as much as evinacumab (Fig. 4b). We applied several conditions of induced degradation to test biophysical stability by SEC-HPLC, including low pH (~ 5.0), high pH (~ 9.0), high temperature of 40 °C and freeze-thaw. As shown in Fig. 4c-f, all of the main peak were over 97 percentages, indicating that the stability of C44-Fc protein was strong and comparable to evinacumab.
We further studied the ability of C44-Fc to block ANGPTL3-mediated suppression of LPL activity by a cell-free LPL assay in vitro. The C44-Fc effectively neutralized the inhibition of LPL activity inducing by four different ANGPTL3 proteins (including CDD and fulllength of human and mouse ANGPTL3) with IC 50 values from 1.6 to 5.4 nM ( Fig. 5 and Table 1). Because C44-Fc bound to human and mouse ANGPTL3 proteins with high affinities and efficiently blocked their inhibitory effect, we employed C57BL/6 mice to assess the following pharmacodynamics study in vivo.

C44-Fc administration lowered serum lipids levels in hypercholesterolemic mice
C57BL/6 mice were fed with a high-fat/high-cholesterol and cholate diet (HF/HCCD) for 4 weeks to induce hypercholesterolemic model and assess the pharmacodynamics of C44-Fc in vivo. Mice were then grouped randomly and treated with a single dose of isotype control antibody (control Ab, 25 mg/kg), C44-Fc (10 mg/kg or 25 mg/kg), or evinacumab (25 mg/kg). Blood samples were acquired on days 0, 1, 4, 7 and 12 after antibody administration. C44-Fc and evinacumab evoked rapid downward trends in serum TG, TC, and LDL-C levels. On day 4, C44-Fc respectively decreased the TG and TC level by 44.2% and 36.6% when compared with isotype controls (Fig. 6a-d). While evinacumab showed similar reduction effects in TG and TC level (44.9% and 38.9%, respectively). During day 4 ~ 7 post antibody administration, LDL-C levels in C44-Fc and evinacumab-treated mice were significantly reduced over 54.4% (Fig. 6e, f ). Furthermore, the hypercholesterolemic mice receiving a single dose of C44-Fc or evinacumab still exhibited considerably lower TC level on day 12 (the end point of in vivo assay).
We further confirmed the LPL activity in hypercholesterolemic mice after C44-Fc treatment. Our results showed that the effect of C44-Fc on the serum lipids reduction was closely related with the upregulated activity levels of plasma LPL activity after intraperitoneally injecting heparin (Fig. 6g). The facilitation of TG clearance triggered by C44-Fc was further assessed by fat tolerance tests. After injection of intralipid, plasma TG of mice treated with control Ab had a remarkable increase which was significantly attenuated by C44-Fc (Fig. 6h), indicating C44-Fc administration remarkably improved fat tolerance in HF/HCCD-fed mice.

C44-Fc administration relieved HF/HCCD-induced hepatic steatosis
The therapeutic effect of C44-Fc on NAFLD was investigated in HF/HCCD-fed mice. After 8 weeks after HF/ HCCD feeding, C57BL/6 mice showed the marked Then mice were continuously dosed with isotype control Ab (25 mg/kg), C44-Fc (25 mg/kg), or evinacumab (25 mg/kg) once a week for 6 weeks (Fig. 7a). Blood samples were collected after 4-h fasting on day 4 after each injection. Compared with the chow-diet mice, HF/ HCCD mice had increase in liver weight (~ 1.7-fold), liver TG contents (~ 1.6-fold) and hepatic lipid accumulation, indicating the pathological changes of NAFLD in HF/ HCCD mice [36,37]. The treatments with C44-Fc (red lines) or evinacumab (blue lines) decreased plasma TG, TC and LDL-C markedly and sustainably in comparison to isotype control Ab (Fig. 7b-e). After the first injection of C44-Fc, a significant decline trends in TG (from 0.95 to 0.59 mmol/L), TC (from 9.2 to 5.9 mmol/L) and As shown in Fig. 8b, c, the liver overweight, liver coloration and morphology of HF/HCCD mice were significantly ameliorated after administration of C44-Fc for 6 weeks. Furthermore, our results also demonstrated that HF/HCCD feeding led to a hepatic lipid accumulation as well as increased level of ALT and AST, which were distinctly ameliorated by weekly dosage of C44-Fc (Fig. 8di). Fairly, 25 mg/kg of C44-Fc treatment had no significant influence on body weight and TG contents of epididymal fat or heart, insulin tolerance test (ITT) and glucose tolerance test (GTT) (Additional file 1: Figs. S3 and S4). Collectively, these findings suggested that C44-Fc fusion protein treatment efficiently reduced the serum lipids levels and relieved the hepatic steatosis induced by HF/ HCCD feeding.

Discussion
NAFLD is a metabolic disease and positively associated with hypercholesterolemia, which is increasingly acknowledged as the key factor of excessive lipid uptake and lipotoxicity in liver, resulting in pathologic insult [3]. It is projected to develop into cirrhosis with increasing risk of hepatocellular carcinoma and may be the main inducement of liver transplantation [4]. Despite its prevalence and seriousness, NAFLD still lacks of effective therapeutics [38]. ANGPTL3 is a key lipid metabolism regulator leading to high levels of plasma TG, TC and LDL-C [10]. In this study, we investigated whether inhibiting ANGPTL3 by VHH-Fc was an effective approach for NAFLD treatment. Herein, we developed a VHH-Fc fusion protein against ANGPTL3, which significantly ameliorated hepatic lipid accumulation and liver injury in HF/HCCD-induced NAFLD mice through reducing serum lipid levels of TG, TC and LDL-C. Although NAFLD could be treated by reducing blood lipids and blood glucose, anti-fibrosis or anti-inflammatory in animal models, it still affects approximately 25% of the population worldwide [39]. Until now, the PPARα/γ agonist saroglitazar was exclusively approved for the treatment of NAFLD [7]. Unfortunately, the PPAR-α/γ agonist had been reported to increase about 2-4% of body weight and had adverse reactions including peripheral oedema in probably 5% of treated patients and fatal heart failure in approximately 11% of treated patients [40]. The classical lipids-reducing agents statins were not applicable to NAFLD as for its intolerance or side effects [41]. Thus, safe and efficient targets for NAFLD are urgently needed. ANGPTL3, as an inhibitor of LPL and EL, was recently revealed elevated expression levels in liver of NAFLD patients and contributed to increased circulating ANGPTL3 [16]. Interestingly, neither the homozygous loss-of-function of ANGPTL3 nor the inhibitors of ANGPTL3 did generate adverse effects with preeminent safety and tolerability, but leaded to very low TG, VLDL, LDL and non-esterified fatty acids (NEFAs) [42,43]. Our results suggested ANGPTL3 inhibitor C44-Fc significantly reduced TG (44.2%), TC (36.6%) and LDL-C (54.4%). More importantly, multiple administrations of C44-Fc significantly reduced 18.3% of liver weight, ameliorated liver coloration and morphology, improved hepatic lipid accumulation and protected HF/ HCCD-related liver injury. All these results indicated that blocking ANGPTL3 was a novel therapeutic approaches for NAFLD with satisfactory efficacy and safety (in Additional file 1: Fig. S5).
Nanobody and its fusion protein, which are promising therapeutic agents, display excellent binding affinity and stability. Besides, they could be expressed in yeast with high yield and low cost [44]. Recently, a bivalent VHHs fusion protein, caplacizumab, had been approved by the FDA for thrombotic thrombocytopenic purpura treatment, indicating the druggability of nanobody and its fusion protein [35,45,46]. The development and characterization of VHHs that specifically block ANGPTL3 have not been reported so far. Hence, we established an immune phage-display VHHs library after immunizing alpaca with hANGPTL3 (S17-K170) and obtained 3 specific VHH candidates which had different CDR3 sequences from the library. Finally, we retrieved C44 clone for in-depth study that had the highest affinity to the CCD of hANGPTL3 at 0.6026 nM.
The VHHs have prominently short circulation half-life due to their small size of 15 kDa. Multiple methods have been applied to increase their half-life period in vivo, such as binding to human serum albumin, fused to IgG-Fc and PEGylated [47,48]. Meanwhile, previous studies revealed that bivalent nanobody had better affinity than monovalent modality. In particular, the bivalent form via Fc fusion showed above 10 times higher affinity and stronger therapeutic effects than monovalent in vitro and in vivo, also emerged longer half-life up to 15 days when compared with monovalent (~ 30 min) and tandem bivalent VHHs (~ 60 min) [44,49]. In present study, C44 was fused to human IgG1 Fc domain to construct the bivalent VHH-Fc fusion protein. The affinities of C44-Fc to ANGPTL3 proteins were about 2.1-11.0 times higher than the individual C44. The pharmacodynamics study of single administration demonstrated that the C44-Fc fusion protein had a long half-life in vivo and the durable lipid-reducing effect lasted up to 12 days.
When compared to the conventional antibody evinacumab, C44-Fc fusion protein had similar ability of rescuing ANGPTL3-mediated suppression of LPL activity and the lipid-reducing effect as well as relieving function in NAFLD. However, the yield of camelidae antibody C44-Fc in ExpiCHO cells reached ~ 500 mg/L, which was twice as much as evinacumab We also investigated the stability of C44-Fc at pH 5, pH 9, 40 °C for 28 days and 6 cycles of freeze-thaw. The results proved that C44-Fc fusion protein showed satisfying stability with no obvious degradation or aggregation at these conditions, indicating that C44-Fc might decrease the expenditure of production, storage and transportation in the future development as a therapeutic antibody.
Consistent with these results, C44-Fc, as an ANGPTL3 inhibitor, showed remarkable potency in lowering lipid and therapeutic effect on NAFLD in mice model, and was likely to be a prospective therapeutic agent for hypercholesterolemic and NAFLD patients. It also might be a latent treatment in other metabolic disorders such as cardiovascular diseases. Nonetheless, further research such as humanization, pharmacodynamics study in cynomolgus monkeys and optimization of expression system, should be conducted to develop it into a therapeutic agent.

Conclusions
In this work, we demonstrated that blocking ANGPTL3 by VHH-Fc fusion protein could relieve hepatic lipid accumulation and liver injury in HF/HCCD-induced NAFLD mice. We developed a novel VHH-Fc fusion protein (C44-Fc) that showed superior characteristics in affinity, yield and stability. We further illustrated its potency in inhibiting ANGPTL3-mediated LPL activity in vitro, lipid-reducing function in hypercholesterolemic mice, as well as ameliorating efficacy in NAFLD mice, indicating that the C44-Fc could be a potent therapeutic candidate for NAFLD.

Immunization and library generation
Alpaca immunizations and the following construction of VHH library were performed as reported previously [50]. A female alpaca was subcutaneously injected with 500 μg of emulsified hANGPTL3(17-170)-mFc antigen (fused from human S17-K170 ANGPTL3 and mouse IgG1-Fc and expressed by our laboratory) on days 1, 21, 42, 63. Blood was collected at 7 days after the last immunizations for the isolation of peripheral blood lymphocytes (PBLs) by density gradient centrifugation using ficollpaque plus (Cytiva, USA, Cat# 17144002). The total RNA of PBLs was extracted by Trizol, followed by a RT-PCR for amplification of the coding sequences of VHHs. Next, the VHHs library was generated by electro-transforming the VHH genes and pHEN1 phagemid recombinant vector into competent E.coli TG1. Finally, the VHHs library was rescued by coinfection with M13KO7 helper phages (Thermo Scientific Ltd, USA, Cat# 18311019).

hANGPTL3-specific VHHs enrichment
Phages presenting VHHs with hANGPTL3-specific binding were enriched after three rounds of screening on hANGPTL3 (S17-220P)-His-immobilized (Novoprotein, China, Cat# Q9Y5C1) 96-well plates. The concentrations of antigen coated on every well for three rounds were decreased in the order of 5, 2, 1 μg per well and simultaneously a blank well was employed as a negative control. After washing with PBST for three times, all wells were blocked in a 37 °C constant temperature incubator for 1 h with PBS containing 3% (w/v) milk powder and washed with PBS for 3 times. Approximately 10 11 phages (diluted with PBS containing 3% (w/v) milk powder) were added to the antigen-coated plates and then put into a 37 °C constant temperature incubator for 1 h. After washing with PBST (containing 0.05% Tween 20) for 6 times in the first round and 8 times in the remaining two panning rounds, hANGPTL3-specific phages were retained and added into E. coli TG1culture to amplify at 30 °C with shaking and rescued with M13KO7 helper phages at 30 °C overnight. The purified phages were precipitated by PEG 8000/NaCl for the subsequent panning round. For each bio-panning, enrichment was performed as described above and was evaluated by the bacteria population which were infected with the purified phages and cultured on LB agar plates with ampicillin.

Indirect ELISA screening
ELISA was employed to evaluate the affinity of phagedisplayed VHH to ANGPTL3. Forty-eight individual bacteriophages were selected from the second and third bio-panning LB agar plates and rescued with M13KO7 helper phages at 30 °C overnight. Next, the viral supernatants containing VHHs were added into the plates coated with hANGPTL3 (S17-220P)-His protein, and PBS containing 3% (w/v) milk powder was used as a negative control. After washing with PBST (containing 0.05% Tween 20) for 6 times, the bound VHHs were determined by HRP-conjugated rabbit anti-M13 antibody (1:500, Abcam, USA) that could react with TMB (Beyotime, China, P0209) followed by the OD450 value testing. Finally, the bacteriophages with OD450 value five-fold higher than the negative control were regarded as positive clones and used for subsequent sequencing.

Nanobody expression and purification
The C27, C44 and C46 gene sequences with 6 × His tag at the amino terminus were synthesized and cloned into the pET21a expression vectors, that were then transformed into E. coli Origami B(DE3) for the nanobody expression. These E. coli Origami B(DE3) cells were inoculated into a LB medium with 100 µg/mL of ampicillin at 37 °C, 220 rpm. The cells were induced by 0.1 M IPTG (Sangon Biotech, China) when they reached the logarithmic growth phase and cultured at 220 rpm at 28 °C for 16 h. The nanobodies were harvested from the cytoplasm of E. coli Origami B(DE3) cells by ultrasonic degradation. Finally, the nanobodies were purified by HisTrap IMAC HP column (GE Healthcare, China).

Epitope binning test
The 100 ng/well of evinacumab was coated on the ELISA plate at 4 °C for 12 h. After washing the ELISA plate with PBST and blocking with 3% BSA, the 1 μg/well of hANGPTL3 (S17-K170)-Fc antigen was added for two hours at 25 °C. After washing the ELISA plate with PBST, the C44-His6 was then added to bind the other epitopes on the hANGPTL3 (S17-K170)-Fc for 2 h at 25 °C. After washing the ELISA plate with PBST for three times, the bound C44-His6 were determined by the His tag polyclonal antibody with HRP (1:10,000 dilution, Thermo Scientific Ltd, USA) that could react with TMB (Beyotime, China, P0209) followed by the OD450 value testing.

VHHs-Fc expression and purification
The encoding sequence of C44 fused to human IgG1-Fc (with the hinge region and had mutated the first Cys into Ser) was synthesized and inserted into pTT5 vector. The DNA sequences encoding the variable regions of evinacumab were cloned and digested into corresponding regions of human IgG1 on pTT5 vector by EcoRI and NheI (Takara, Biotechnology, Otsu, Japan). The C44-Fc and evinacumab expression plasmids were transfected into ExpiCHO-S cells (Gibco, Thermo Scientific Ltd.) by CHO Transfection Kit and cultured in ExpiCHO expression media (Gibco, Thermo Scientific Ltd.) according to its Max Titer protocol. The culture supernatants were collected 12 days after transfection and purified by HiTrap MabSelect Prism A column (GE Healthcare).

Biophysical stability evaluation
SEC-HPLC was performed to assess the concentration of VHHs-Fc using an Agilent 1260 Infinity II SFC System with TOSOH TSKgel G3000WXL column (7.8 mm × 30 cm, 5 μm) at 1.0 mL/min flow rate of PBS. Each sample was injected with 100 μg of protein and measured by UV detection for 30 min at 280 nm and 37 °C.

Animal study
C57BL/6 mice (male, 5-6 weeks old) were obtained from Shanghai SIPPR-BK Laboratory Animal Co., Ltd. (Shanghai, China) and were kept under the standard specific pathogen-free (SPF) environment. Mice were supplied with high-fat/high-cholesterol and cholate diet (HF/HCCD, Teklad TD.90221, which contains 15.8% fat, 1.25% cholesterol and 0.5% sodium cholate) or chow diet after being acclimated to experiment environment for 1 week [51]. For single administration studies, C57BL/6 mice were fed with HF/HCCD for 4 weeks and then treated with subcutaneous injection of control Ab, C44-Fc or evinacumab. Blood samples were collected before antibody injection as a baseline and 1, 4, 7 and 12 days later by retro-orbital bleed after a 4 h fast. For the postheparin plasma LPL activity, mice were treated with heparin (100 U/kg) intraperitoneally injected in tail and 5 min later plasma samples were collected into EDTA-coated tubes which were preserved on ice. As for multiple administration, C57BL/6 mice were fed with HF/HCCD for 8 weeks and treated with subcutaneous injection of control Ab, C44-Fc or evinacumab weekly for 6 weeks. Blood samples were collected 4 days after each injection. At the end of the experiment, plasma lipoproteins were separated as described previously [52].

Metabolic analysis
Lipid tolerance test (LTT), insulin tolerance test (ITT), and glucose tolerance test (GTT) were performed 4 days after antibody injection (25 mg/kg). For LTT, C57BL/6 mice were injected intraperitoneally with 10 μg/kg body weight of intralipid (Baxter Healthcare Corporation, USA) after a 2-h fast. For ITT, mice were intraperitoneally injected with 1 IU insulin (Beyotime, Shanghai) per kg body weight after a 4-h fast. For GTT, mice were intraperitoneally injected with 1 g/kg body weight of glucose (Beyotime, Shanghai) after 6 h of fasting. All blood samples were acquired at indicated time points after C44-Fc administration.

Biochemistry analysis
All serum levels of TG, TC, LDL-C, alanine aminotransferase (ALT), aspartate aminotransferase (AST) and tissue lipid content were detected using commercial kits (Nanjing Jiancheng Bioengineering, Nanjing). The blood glucose levels were tested by blood glucose test strips (Roche, Germany).

Histopathological study
The liver samples were weighed and fixed with 4% paraformaldehyde. Subsequently, all liver samples were embedded with paraffin and stained with hematoxylin and eosin (H&E) or embedded in OTC and freezed for oil red O staining.

Data analyses
All data were presented as mean ± SEM. The significant difference between groups were operated using Student's t-test or one-way ANOVA and were indicated †, § or * as P < 0.05, † †, § § or ** as P < 0.01 and § § § or *** as P < 0.001.