Apolipoprotein M promotes cholesterol uptake and efflux from mouse macrophages

Apolipoprotein M (ApoM) exhibits various anti‐atherosclerotic functions as a component of high‐density lipoprotein (HDL) particles. Scavenger receptor class B type I (SR‐BI) is a classic HDL receptor that mediates selective cholesterol uptake and enhances the efflux of cellular cholesterol to HDL. However, the effect of ApoM on cholesterol transport in macrophages remains unclear. In this study, we identified for the first time that ApoM is expressed in mouse macrophages and is involved in cholesterol uptake, similar to SR‐BI. NBD‐cholesterol uptake and efflux in cells were characterized using fluorescence spectrophotometry. The uptake ratios of cholesterol by macrophages from ApoM −/− SR‐BI−/− mice were significantly lower than those from ApoM+/+SR‐BI −/− and ApoM −/− SR‐BI+/+ mice. Real‐time fluorescence quantitative PCR was used to analyze the expression of cholesterol transport‐related genes involved in cholesterol uptake. ApoM‐enriched HDL (ApoM+HDL) facilitated more cholesterol efflux from murine macrophage Ana‐1 cells than ApoM‐free HDL (ApoM−HDL). However, recombinant human ApoM protein inhibited the ability of ApoM−HDL to induce cholesterol efflux. In conclusion, ApoM promotes cholesterol uptake and efflux in mouse macrophages. A better understanding of ApoM function may lead to the development of novel therapeutic strategies for treating atherosclerotic diseases.

Apolipoprotein M (ApoM) exhibits various anti-atherosclerotic functions as a component of high-density lipoprotein (HDL) particles. Scavenger receptor class B type I (SR-BI) is a classic HDL receptor that mediates selective cholesterol uptake and enhances the efflux of cellular cholesterol to HDL. However, the effect of ApoM on cholesterol transport in macrophages remains unclear. In this study, we identified for the first time that ApoM is expressed in mouse macrophages and is involved in cholesterol uptake, similar to SR-BI. NBD-cholesterol uptake and efflux in cells were characterized using fluorescence spectrophotometry. The uptake ratios of cholesterol by macrophages from ApoM À/À SR-BI À/À mice were significantly lower than those from ApoM +/+ SR-BI À/À and ApoM À/À SR-BI +/+ mice. Real-time fluorescence quantitative PCR was used to analyze the expression of cholesterol transport-related genes involved in cholesterol uptake. ApoM-enriched HDL (ApoM + HDL) facilitated more cholesterol efflux from murine macrophage Ana-1 cells than ApoM-free HDL (ApoM À HDL). However, recombinant human ApoM protein inhibited the ability of ApoM À HDL to induce cholesterol efflux. In conclusion, ApoM promotes cholesterol uptake and efflux in mouse macrophages. A better understanding of ApoM function may lead to the development of novel therapeutic strategies for treating atherosclerotic diseases.
Scavenger receptor class B type I (SR-BI), the first cell surface protein to be characterized as an HDL receptor [9], is abundant in the liver and macrophages. Several experiments have demonstrated that HDL induces the efflux of cholesterol from foam cells via SR-BI [10,11]. Macrophage SR-BI is critical to atheroprotective effects because macrophages are the most abundant cell type in atherosclerotic lesions, and macrophage SR-BI expression affects cholesterol homeostasis and inflammation [12]. Pagler et al demonstrated that SR-BI is a receptor that facilitates the uptake of HDL particles. Internalization and subsequent resecretion of HDL are important for the efflux of cellular cholesterol [13]. Fuentes et al. showed that insulin promotes cholesterol uptake by SR-BI-dependent mechanisms in a model of the human intestinal epithelium [14]. However, the role of SR-BI in mediating cholesterol efflux from mouse macrophages is controversial [15,16]. Therefore, the purpose of the present study was to ascertain the roles of primary murine macrophage ApoM and SR-BI in cholesterol uptake and efflux.
Macrophage accumulation within the subendothelium or neointima constitutes one of the first steps in atherogenesis. The key to this process is the scavenging of excess cholesterol by macrophages that take up and store cholesterol until it can be mobilized from these cells by components of the HDL system. Therefore, in our study, we chose Ana-1 murine macrophage cells and four types of primary murine macrophages with different genotypes as research models. ApoM is a component of HDL, and SR-BI is the HDL receptor. Deletion of either the ApoM or SR-BI gene leads to a disturbance in cholesterol homeostasis. However, the impact of knockout of both ApoM and SR-BI on cholesterol transport in murine macrophages is not clear. Therefore, we crossed the ApoM and SR-BI single-gene knockout mice for generations to generate ApoM À/À SR-BI À/À double-knockout mice (DKO). The cholesterol uptake capacity of macrophages from different genotypes was further compared.
ATP-binding cassette transporter A1 (ABCA1) and ATP-binding cassette transporter G1 (ABCG1), two key cholesterol transporters in macrophages, function in mediating cellular cholesterol efflux and contributing to RCT [17,18]. ABCA1 facilitates intracellular phospholipids and cholesterol efflux to lipid-free ApoA-I [19], whereas ABCG1 mediates intracellular oxysterol and cholesterol efflux to HDL [20]. Kober et al. suggested that ABCG1, SR-BI, and ApoM are centrally involved in cholesterol efflux from porcine brain capillary endothelial cells [8]. However, the relationship between the ATP-binding cassette transporter and ApoM or SR-BI in macrophages has not yet been elucidated.
The present study aimed to clarify the influence of ApoM and/or SR-BI gene deletion on macrophage cholesterol uptake. Subsequently, we compared the macrophage cholesterol efflux capacity to identify whether there was any difference between ApoM + HDL and ApoM À HDL. This better understanding of ApoM function may yield novel therapeutic strategies for treating atherosclerotic diseases.

Animals
Male C57BL/6N mice with different genotypes but of identical age and body weight were used in this study (8-10 weeks old, 23-25 g). Wild-type (WT) mice were obtained from the Center for Experimental Animals of Soochow University (Suzhou, China). ApoM-deficient (ApoM À/À ) mice with C57BL/6N background were generated by homologous recombination with the help of the Model Animal Research Center of Nanjing University (Jiangsu, China) as previously described [21]. Chimeric male mice were identified using PCR and were bred with C57BL/6N female mice to generate F1 offsprings. The F1 offsprings were crossed for at least two to three generations until they were ApoM À/À homozygous. SR-BI-deficient (SR-BI À/À ) mice with C57BL/6N background were generously provided by Nanjing Medical University. To obtain ApoM/SR-BI DKO, male homozygous ApoM +/+ SR-BI À/À mice were crossbred with female homozygous ApoM À/À SR-BI +/+ mice. Littermates from that generation (ApoM +/À SR-BI +/À ) were then backcrossed to obtain ApoM À/À SR-BI À/À mice. Because of the low reproduction rate in DKO mice, they were fed a 1% probucol chow diet to reduce blood lipids during the reproductive period. The fertility rate of mice increased significantly after medication. The genotypes of offsprings and laboratory mice were analyzed by PCR, as previously described [22,23]. All mice were maintained under a 12:12-h light-dark cycle in a temperature-controlled facility (25°C) with access to standard chow. All animal procedures were approved by the Animal Care and Use Committee of Soochow University, Suzhou, China (Permit Number: SYXK(Su)2017-0043).

Cell culture
The peritoneal macrophages were harvested from the peritoneal cavity of mice, as previously described [24]. Briefly, mice were injected intraperitoneally with 2 mL of 3% thioglycollate broth to recruiting a larger number of macrophages. After 3 days, the mice were euthanized, and the peritoneal cavities were washed with 5 mL of cold PBS. After a gentle massage of the abdominal wall, the peritoneal fluid containing macrophages was collected. After centrifugation for 5 min at 180 g and 4°C, the cells were seeded into 48-well plates (4 9 10 5 cells/well) or 6-well plates in RPMI 1640 medium supplemented with 10% FBS. The cells were allowed to adhere to the substrate by culturing for 2 h at 37°C. Non-adherent cells were removed by gently washing twice with warm PBS. Adherent cells were selected as macrophages and were maintained in culture for further use. The murine macrophage cell line Ana-1 was purchased from the Shanghai Cell Bank of the Chinese Academy of Sciences, and the cells were cultured in regular tissue culture dishes in RPMI 1640 medium supplemented with 10% FBS at 37°C in a humidified incubator with 5% CO 2 , and 95% air. The Ana-1 cell line was established from bone marrow cells of C57BL/6 (H-2 b ) mice infected with J2 recombinant retrovirus for immortalization [25].

Wright staining and flow cytometry
The morphology of peritoneal macrophages was observed by microscopy after Wright staining, according to the manufacturer's instructions. Quantification of murine peritoneal macrophages was performed by flow cytometry. Cell suspensions were processed for staining of macrophage surface markers, including F4/80 and CD11b. Data were collected using the BD FACSCanto II (BD Biosciences, San Jose, CA, USA) and analyzed using the FLOWJO software (Tree Star Inc. San Carlos, CA, USA).

Phenotypic analysis of peritoneal macrophages
WT mice were induced with thioglycollate broth as described above. Control group was peritoneal injection with the same amount of normal sodium. Peritoneal macrophages from control group exposure to LPS (100 ngÁmL À1 ) or IL-4 (20 ngÁmL À1 ) overnight for M1 or M2 polarization. The mRNA expressions of M1 marker iNOS and M2 marker Arg-1 were examined by means of qRT-PCR.

Identification of PCR amplification products and capillary western immunoassay
The PCR product of ApoM was validated by agarose gel electrophoresis and sequencing at Sangon Biotech Co., Ltd. Capillary western immunoassay (Wes) was performed in protein lysates from Ana-1 cell, peritoneal macrophages, liver, and kidney. Briefly, protein lysates were analyzed on a Wes system (ProteinSimple, Santa Clare, CA, USA) using a 12-230 kDa separation module (Bio-Techne Ltd., Abingdon, UK). Level of ApoM (1 : 75) was normalized using the reference protein b-actin (1 : 50). The peaks were analyzed using Compass software (ProteinSimple, Santa Clare, CA, USA). Detailed procedures are as described in the cited reference [26].

Uptake of NBD-cholesterol by Ana-1 cells and mouse peritoneal macrophages
The NBD fluorophore can be attached to carbon-22 or carbon-25 on the side chains of cholesterol. The excitation and emission wavelength of NBD-cholesterol are 469 and 537 nm, respectively. Different concentrations of NBDcholesterol and NBD (2.5-10 lM) were prepared before the experiments. Briefly, 2 mg of NBD-cholesterol (molecular weight, 494.63) was dissolved in 4 mL of 37°C absolute ethanol to obtain a concentration of 0.5 mgÁmL À1 (1 9 10 3 lM) NBD-cholesterol. The dissolution of NBD was similar to that of NBD-cholesterol. These solutions were mixed slowly into RPMI 1640 medium without phenol red to three final concentrations of 2.5, 5, and 10 lM. Ana-1 cells were cultured as described above and seeded into 48-well plates at a concentration of 2 9 10 5 cells/well. medium supplemented with NBD-cholesterol (2.5-10 lM) at 37°C. A similar experiment was performed using NBD. Then, the cells and culture supernatants were collected at 0, 0.5, 1, 2, 4, 6, 8, and 12 h. Peritoneal macrophages from the mice of four different genotypes were seeded into 48well plates (2 9 10 5 cells/well). After incubation for 24 h, the peritoneal macrophages were loaded with 10 lM NBD-cholesterol. Then, the cells and culture supernatants were collected at 1, 2, 4, 6, and 12 h. At each time point, the cells were washed twice with warm PBS and lysed with 1% Triton X-100. The QuantiFluor TM -ST Handheld Fluorometer (Promega, Madison, WI, USA) was used to measure the fluorescence intensity of the supernatant and cell lysate. The uptake ratio was calculated as the percentage of NBD-cholesterol in cells at a given time relative to the total NBD-cholesterol content in the supernatant and cells.

Expression analysis of cholesterol transportrelated genes
The mRNA levels of cholesterol transport-related genes were analyzed by real-time quantitative PCR. Peritoneal macrophages from the mice of four different genotypes were loaded with or without 10 lM NBD-cholesterol for 6 h. Then, the total RNA from peritoneal macrophages was isolated according to the manufacturer's protocol using the RNAeasy TM Animal RNA Isolation Kit with Spin Column (Beyotime Biotechnology, Shanghai, China). Firststrand cDNA was synthesized from 2 lg of total RNA using the RevertAid TM First-Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA). PRIMER PREMIER 5.0 software (Premier Biosoft, Palo Alto, CA, USA) was used to design primers and probes (Table 1). Quantitative gene expression analysis of Abca1, Abcg1, ApoA-I, ApoE, and the reference gene gapdh was performed by RT-qPCR on the LightCycler480 Ò II system (Roche, Basel, Switzerland).

Gene
Primer / Probe were collected after 4 h of induction. The efflux ratio was calculated as the percentage of NBD-cholesterol released into the medium relative to the total NBD-cholesterol content in the medium and cells. As a control, Ana-1 cells were incubated with basic medium to obtain values for 'background' efflux.

Statistical analysis
Each experiment is representative of at least three independent experiments. Data are shown as the mean AE SD. Statistical analysis was performed using two-way ANOVA followed by Bonferroni test or one-way ANOVA test. All calculations were performed using GRAPHPAD PRISM software (GraphPad Software, Inc., San Diego, CA, USA). Statistical significance was set at P < 0.05.

Results
Phenotype identification of thioglycollateinduced peritoneal murine macrophages Figure 1A shows the purity of mouse peritoneal macrophages by Wright staining. As shown in Fig. 1A (b), predominant cells are round or oval macrophages with deeply stained nuclei and little cytoplasm. Occasionally, a few spindle fibroblasts with loose chromatin and abundant cytoplasm were observed in the field (Fig. 1A(a)).
The original image is shown in Fig. S1. CD11b and F4/80 are macrophage surface markers. Flow cytometric analysis revealed that the percentage of F4/80 + CD11b + cells was 86.8 % (Fig. 1B). Macrophages are typically divided into two groups based on whether they are classically (pro-inflammatory, M1) or alternatively activated (antiinflammatory, M2). High levels of cellular induced nitric oxide synthase (iNOS) and type-1 arginase (Arg-1) are often observed in M1-and M2-polarized macrophages. Thus, we quantified cellular iNOS and Arg-1 by RT-qPCR to evaluate macrophage polarization. Peritoneal murine macrophages derived from WT mice were divided into four groups and cultured for 12 h: normal saline group, thioglycolate-elicited group, LPS-elicited group (M1, 100 ngÁmL À1 ), and IL-4-elicited group (M2, 20 ngÁmL À1 ). As shown in Fig. 1C,D, the expressions of iNOS and Arg-1 were higher than those in the other groups after LPS and IL-4 stimulation. However, compared with the normal saline group, thioglycollateinduced macrophages did not polarize.

Expression of ApoM in macrophages, liver, and kidney
We characterized the expression of ApoM in Ana-1 cells, peritoneal macrophages, hepatocytes, and kidney cells from WT mice. Fig. 2A shows the mRNA level of ApoM in Ana-1, peritoneal macrophages, liver, and kidney. As shown in the RT-qPCR analysis of mRNA expression levels, despite the much lower expression level compared with the liver and kidney, ApoM was expressed in both Ana-1 and macrophages. The expression level of ApoM in peritoneal macrophages was~2.5 times higher than that in Ana-1 cells. Fig. 2B confirmed the protein level of ApoM by capillary western immunoassay. ApoM protein levels in macrophage, Ana-1, kidney, and liver detected by western blot based on capillary electrophoresis technology and full-length blots are presented in Fig. S2. Agarose gel electrophoresis and sequencing were subsequently performed, and the results confirmed that the PCR products were mouse ApoM (Fig. 2C,D). The original image of agarose gel electrophoresis is shown in Fig. S3.

Uptake of NBD-cholesterol in Ana-1 cells
To ensure that the fluorescence intensity of NBD represents cholesterol content, we first ruled out the possibility of NBD uptake by cells. The results showed that the uptake of NBD by Ana-1 cells was independent of time and concentration (Fig. 3A); however, the uptake of NBD-cholesterol was concentration-dependent (Fig. 3B). In each group, the uptake ratio significantly increased after incubation with NBD-cholesterol for 30 min (P < 0.05). When the cells were incubated with 10 lM NBD-cholesterol, the uptake ratios increased over time in a certain range and plateaued at 6 h (Fig. 3B). There was no significant difference in the uptake ratios at 8 and 6 h (P > 0.05). The curve was most stable after 6 h, so we chose 10 lM NBDcholesterol as the optimal concentration for follow-up experiments.

Deficiency of ApoM and SR-BI impairs the uptake of NBD-cholesterol in peritoneal macrophages
Peritoneal macrophages from WT, ApoM À/À , SR-BI À/ À and DKO were used to study the uptake ratios of NBD-cholesterol. The data are presented in Fig. 4. The uptake ratios of all the groups plateaued at 6 h. There was no significant difference between the uptake ratios at 6 and 12 h (P > 0.05). However, at the plateau stage, NBD-cholesterol uptake by ApoM À/À mouse macrophages was significantly decreased compared to that by WT mouse macrophages (P < 0.05). Additionally, peritoneal macrophages from DKO mice also showed a lower uptake ratio than SR-BI À/À mice and ApoM À/À mice (P < 0.05).
In addition, we detected the mRNA levels of ABCA1, ABCG1, ApoA-I, and ApoE in these macrophages 6 h after incubation with or without NBDcholesterol. As shown in Fig. 5, the ABCA1 levels were lower in peritoneal macrophages from ApoM À/À , SR-BI À/À , and DKO mice than in those from WT mice. In the WT and ApoM À/À groups, the mRNA expression of ABCA1 and ApoE decreased significantly after cholesterol ingestion, while the expression of ApoA-I increased significantly. The expression of ABCG1 decreased only in the ApoM À/À group after cholesterol ingestion. Values are the mean AE SD (n = 3 per group and three independent experiments). Statistical analysis was performed using two-way ANOVA followed by Bonferroni test. *Significantly different from 2.5 lM NBD-Chol (P < 0.05). § significantly different from 5lM NBD-Chol (P < 0.05). uptake (%) = fluorescence intensities of the cell lysate / (fluorescence intensities of supernatant + fluorescence intensities of the cell lysate) 9 100%. Effects of ApoM on the efflux of NBD-cholesterol from Ana-1 cells Fig. 6A shows the effects of ApoM on cholesterol efflux. ApoM + HDL (20 lgÁmL À1 ) and ApoM À HDL (20 lgÁmL À1 ) enhanced cholesterol efflux from Ana-1 cells, and ApoM + HDL seemed to have a stronger effect on promoting cholesterol efflux than ApoM À HDL (P < 0.05). The efflux ratio of NBD-cholesterol increased over time, peaking at 4 h, and subsequently remained constant (Fig. 6A). There was no detectable difference in the cholesterol efflux-stimulating capacity between the control and rh-ApoM (10 lgÁmL À1 ). It was supposed that the concentration of rh-ApoM was too low, so we increased its concentration to 50 and 100 lgÁmL À1 . However, the result showed that increasing rh-ApoM concentration did not promote cholesterol efflux (Fig. 6B). Figure 7 showed that rh-ApoM (10 lgÁmL À1 ) did not change the ApoM + HDL-induced cholesterol efflux, but it reduced the effects of ApoM À HDL on cholesterol efflux (P < 0.05).

Discussion
In this study, we mainly explored the role of ApoM in two aspects: the uptake and efflux of cholesterol in mouse macrophages. In the investigation of ApoM knockout mice, we found that similar to SR-BI, ApoM deficiency leads to impairment of cholesterol uptake in macrophages. ApoM + HDL facilitated more cholesterol efflux from Ana-1 cells than ApoM À HDL. Rh-ApoM alone did not affect the cholesterol efflux from macrophages. In contrast, when working together with ApoM À HDL, the ApoM À HDL-induced cholesterol efflux decreased. The formation of excess cholesterol deposits in macrophages is the pathological basis for the development of atherosclerosis [27]. RCT is the process by which excess cholesterol in peripheral tissues is transported to the liver via HDL [28]. The efflux of cholesterol from lipid-loaded macrophages is not only the initial step of RCT but also the limiting step [29]. Regulation of cholesterol homeostasis is highly likely to be an effective treatment for atherosclerosis; therefore, it is necessary to study the factors that influence cholesterol transport. The fluorescence intensity of NBD-cholesterol is proportional to the concentration of cholesterol [30,31]. When cells were treated with NBD fluorescent dye alone, almost no fluorescence (less than 1%) was observed in the cells; however, an obvious fluorescence gradient was detected in cells cultured with different concentrations of NBD-cholesterol. Therefore, we assumed that the NBD fluorescent dye was transported into cells along with cholesterol, and the fluorescence intensity detected in the following experiments represented the cholesterol content. In this study, the uptake ratio of 10 lM NBD-cholesterol by Ana-1 cells and peritoneal macrophages plateaued at 6 h which is consistent with the results of a previous study [32]. We chose 10 lM as the working concentration for better results in subsequent experiments, given that Ana-1 cells uptake cholesterol in a concentrationdependent manner. Interestingly, at low NBD-cholesterol concentration such as 2.5 lM, the uptake of cholesterol decreases over time. The balance between hydrolysis and esterification of cholesterol plays a key role in maintaining cholesterol homeostasis and preventing the generation of foam cells [33]. When cells Values are presented as the mean AE SD (n = 3-6 per group). Statistical analysis was performed using two-way ANOVA followed by Bonferroni test and one-way ANOVA test. *Significantly different from the control (P < 0.05), §Significantly different from ApoM À HDL (P < 0.05). ns not significantly different from the control (P > 0.05). efflux (%) = fluorescence intensities of supernatant / (fluorescence intensities of supernatant + fluorescence intensities of the cell lysate) 9 100%. are stimulated with low concentrations of cholesterol, the cholesterol intake is still within a range that the cells can manage. Cells process and metabolize cholesterol over time; therefore, the amount of cholesterol decreases. In the case of cholesterol overload, macrophages can be transformed into foam cells. Cholesterol accumulates in the cell beyond the metabolic capacity of the cell, and the amount of metabolized cholesterol is negligible compared to that taken in.
ApoM is a member of the lipocalin superfamily, which has different functions in lipid construction and transport [34]. It was reported that SNPs in the proximal promoter region of the ApoM gene are associated with dyslipidemia [35]. In vivo, the concentration of ApoM in the plasma is positively correlated with total cholesterol [36]. These studies suggest that ApoM might be associated with cholesterol metabolism.
Previous studies have shown that SR-BI plays an important role in facilitating cholesterol trafficking [37]. SR-BI deficiency impairs both lipid influx and efflux from cells [38], which was verified in our experiments. Our previous work showed that SR-BI deficiency also promotes ApoM expression [39] in liver cells, which indicates that they can compensate for each other in vivo. However, in vitro, increased ApoM might be independent of HDL-mediated cholesterol uptake [39]. To explore whether ApoM affected cholesterol uptake, we tested the cholesterol uptake ability of macrophages derived from ApoM knockout mice. The results showed that the uptake of NBD-cholesterol in peritoneal macrophages from ApoM À/À mice was lower than that in macrophages from WT mice. The same results were observed when comparing DKO mice to SR-BI À/À mice, indicating that the deletion of ApoM might decrease cholesterol uptake, similar to SR-BI deletion in peritoneal macrophages. Therefore, we believe that ApoM and SR-BI might have synergistic effects on cholesterol uptake. Previous studies have generally focused on the effects of ApoM on HDL, but our results demonstrate for the first time that ApoM expressed in macrophages has the same function as SR-BI in regulating cholesterol transport. On the one hand, this finding encouraged us to explore other physiological functions of ApoM in different cells and tissues. On the other hand, anti-atherosclerotic therapies should perhaps not simply focus on increasing or decreasing the concentration of cholesterol transport-related factors but also on maintaining intracellular cholesterol homeostasis. Cluster of differentiation 36 (CD36), a class B scavenger receptor similar to SR-BI, proposed to be a crucial molecule in cholesterol homeostasis in various mechanisms including absorption/reabsorption, synthesis, and transport of cholesterol and bile acids. CD36 plays a crucial role in foam cell formation and LDL uptake in macrophages [40]. Animal studies performed on CD36-deficient mice suggest that deficiency of CD36 prevents the development of atherosclerosis [41]. However, ApoM deficiency leads to lipid deposition and metabolic distress and promotes atherosclerosis [42]. Therefore, although ApoM affects macrophage cholesterol uptake like CD36, its function in vivo should be more complex.
At the same time, we detected the mRNA levels of cholesterol transport-related factors among different genotypes with or without NBD-cholesterol incubation. ABCA1 and ABCG1 govern the rate-limiting step for cholesterol efflux from peripheral cells. In our study, ABCA1 levels were lower in peritoneal macrophages from ApoM À/À , SR-BI À/À , and DKO mice than in those from WT mice, indicating that the knockout of ApoM or SR-BI could affect the expression of ABCA1. In the WT and ApoM À/À groups, the mRNA expression of ABCA1 and ApoE was decreased significantly after cholesterol ingestion, while the expression of ApoA-I was increased significantly. The hepatic ApoM expression level in ApoE-deficient mice was lower than that in WT mice. ApoE plays a role in accelerating the clearance of plasma ApoM [43]. Our results suggest that ApoM deficiency may also influence ApoE expression during macrophage cholesterol uptake. The expression of ABCG1 decreased only in the ApoM À/À group after cholesterol ingestion. This phenomenon is worthy of attention because cholesterol loading significantly decreases ABCA1 levels. It is well accepted that ABCA1-mediated cholesterol efflux is initiated by the interaction of lipid-and cholesterol-free ApoA-I particles with ABCA1 transporters in peripheral tissues [44]. The expression level of ABCA1 is positively correlated with the capacity to promote cellular cholesterol efflux [45]. Sparrow et al. [46] reported that Ac-LDL-loading increases ABCA1 mRNA levels in macrophages. Venkateswaran et al. [47] reported that the degree of induction of ABCA1 mRNA was clearly dependent on the type of exogenously added lipid. Acetylated lowdensity lipoprotein (Ac-LDL), oxidized low-density lipoprotein (ox-LDL), and 25-dihydroxycholesterol (25-OHC) were more potent inducers of ABCA1 mRNA levels than LDL. Our results seemed inconsistent with those of previous studies. One possible explanation is that no exogenous lipid-poor ApoA-I as a cholesterol receptor in the supernatant was formed, so intracellular cholesterol did not efflux outside the cell (it can be verified by the control group in Fig 6). The effect of free cholesterol on the level of ABCA1 in macrophages is not clear. Another explanation is that in the WT group, the rate of cholesterol influx becomes too high and mechanisms of cholesterol homeostasis regulation do not function properly, which can lead to endoplasmic reticulum (ER) stress initiation. One consequence of ER stress is the downregulation of ABCA1, which leads to foam cell formation [48,49]. From the current data, we can see that ApoM deletion affects the expression of cholesterol transport-related genes in the process of cholesterol uptake. Further studies are required to elucidate the underlying mechanism.
To investigate whether rh-ApoM, ApoM + HDL, and ApoM À HDL have different effects on cholesterol efflux, Ana-1 cells loaded with NBD-cholesterol were incubated with the above three particles. The data showed that ApoM + HDL particles were more efficient than ApoM À HDL in stimulating cholesterol efflux from Ana-1 cells and reducing total levels of intracellular cholesterol, which is consistent with observations reported by Elsoe et al [5]. Kober reported that cholesterol efflux to HDL increased with increasing ApoM molecule content per HDL particles [8]. Their results and ours suggested that ApoM could increase the capacity of HDL to facilitate the transport of cholesterol to the extracellular acceptor.
However, different concentrations of rh-ApoM could not effectively promote cholesterol efflux from Ana-1 cells. This result suggests that rh-ApoM fails to promote cholesterol efflux while lacking HDL in vitro.
To determine whether rh-ApoM forms an effective complex with HDL to promote cholesterol efflux, we added rh-ApoM to the supernatant containing ApoM + HDL or ApoM À HDL to observe the effects of rh-ApoM on HDL-mediated cholesterol efflux. The results showed that rh-ApoM did not affect ApoM + HDL-induced cholesterol efflux, whereas it significantly decreased ApoM À HDL-induced cholesterol efflux. This indicates that in vitro, rh-ApoM cannot bind ApoM À HDL to form effective ApoM + HDL complexes. To explain our conjecture more clearly, we drew a simple sketch map to illustrate (Fig. 8). Since both of ApoM + HDL and ApoM À HDL contain ApoA-I, we hypothesized that the affinity of rh-ApoM for SR-BI was higher than that of ApoA-I but lower than that of ApoM + HDL (i.e., ApoM + HDL > rh-ApoM > ApoA-I). When macrophages were incubated with rh-ApoM and ApoM À HDL, although rh-ApoM might compete with ApoM À HDL in combination with part of SR-BI expressed on macrophages, it fails to promote cholesterol efflux while lacking HDL. Therefore, blocking the receptor that binds to ApoA-I on ApoM À HDL will reduce ApoM À HDL-induced cholesterol efflux. However, rh-ApoM could not compete with ApoM + HDL, so it did not affect ApoM + HDLmediated cholesterol efflux. If ApoM is compared to a locomotive, HDL can be considered as a carriage. Cellular cholesterol cannot be transported by the 'locomotive' without the 'carriage'. Taken together, these findings suggest that HDL promotes cholesterol efflux not only through the ApoA-I pathway but also via the ApoM pathway, which hypothetically involves SR-BI. Moreover, receptors on Ana-1 cells may have a stronger affinity for rh-ApoM than other ligands on ApoM À HDL, such as ApoA-I, which results in an impaired ability of ApoM À HDL to promote cholesterol efflux. As shown in Fig. 8, there is no evidence of a direct interaction between ApoM and SR-BI. However, this does not rule out the possibility that they work in concert in HDL-dependent cholesterol efflux, which will be appealing to investigate in future studies.
In summary, our data suggest the importance of ApoM in the uptake and efflux of cholesterol in mouse macrophages. ApoM deficiency leads to decreased cholesterol uptake by mouse peritoneal macrophages. ApoM + HDL promotes cellular cholesterol efflux more efficiently than ApoM À HDL does.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Wright staining of peritoneal macrophages from WT mice. Scale bar: 50 lm. Fig. S2. ApoM protein levels in macrophage, Ana-1, Kidney, and liver detected by western blot based on capillary electrophoresis technology. A pack with 25 capillary cartridges was used in this study. ApoM and b-actin protein levels were detected by 12-230 kDa pre-filled plates, respectively. The blots in Fig. 2B in main text are from blots which are framed.