LncRNA TUG1 regulates ApoM to promote atherosclerosis progression through miR‐92a/FXR1 axis

Abstract This study aims to explore the possible mechanism of TUG1 regulating ApoM in AS. To this end, expression levels of TUG1 and ApoM were measured in high fat dieted C57BL/6J mice, normal dieted C57BL/6J mice, ob/ob mice and db/db mice. LV‐TUG1 or sh‐TUG1 was injected into C57BL/6J mice before isolating peritoneal macrophages to measure cholesterol efflux (CE) and expression levels of ABCA1, ABCG1 and SR‐BI. Meanwhile, CE in RAW264.7 cells was also measured after cell transfection. Dual luciferase reporter assay and anti‐AGO2 RIP were applied to verify the relationship among TUG1, FXR1 and miR‐92a. Total cholesterol (TC), triglyceride (TG), low‐density lipoprotein cholesterin (LDL‐C), high‐density lipoprotein cholesterol (HDL‐C) as well as expressions of inflammatory cytokines (TNF‐α, IL‐1β and IL‐6) in plasma were measured. Knock‐down or expressed TUG1, FXR1 or miR‐92a in NCTC 1469 cells or in ApoE−/− AS mice to determine the alteration on ApoM and plaque size. TUG1 was highly expressed while ApoM was down‐regulated in high fat dieted C57BL/6J mice, b/ob and db/db mice. Overexpression of TUG1 could reduce the expression of ApoM, ABCA1 and ABCG1 in addition to slowing down CE rate. Reversed expression pattern was found in cells with knock‐down of TUG1. TUG1 can compete with FXR1 to bind miR‐92a. FXR1 negatively target ApoM. Overexpression of TUG1 in ApoE−/− mice can increase plaque size and enhance macrophage contents accordingly. TUG1 can inhibit ApoM in both liver tissues and plasma to inhibit CE through regulating miR‐92a/ FXR1 axis. TUG1 is a promising target for AS treatment.

haemorrhage, rupture and calcification. 3,4 Nowadays, most applicable therapeutic strategies are preventive and mainly focus on attenuating the formation of thrombus and improving blood lipid profile, yet no treatment can directly target the atherosclerotic lesion. 5,6 The emergence of next-generation sequencing facilitates the elucidation of genetic perspective on AS development and progression; therefore, gene target therapy has boosted as a potential field for AS patients. 7 Long non-coding RNAs (lncRNAs) have been implied to be involved in many human diseases, including cancers and AS. 8 Moreover, several lncRNAs are emerged as novel molecular biomarkers for early diagnosis, potential therapeutic targets and prognosis of AS. [9][10][11] Taurine up-regulated gene 1 (TUG1) was initially identified as a transcript up-regulated by taurine, whose aberrant expression has been found in several cancers, including non-small cell lung cancer, bladder cancer and osteosarcoma. [12][13][14][15] Additionally, overexpression of TUG1 was reported to hinder the attenuation of tanshinol on oxidized low-density lipoprotein (ox-LDL)-induced endothelial cell apoptosis. 16 Apolipoprotein (apo) M is a typical lipocalin for the lipid sphingosine-1-phosphate (S1P). ApoM, by delivering S1P to the S1P (1) receptor on endothelial cells, affects high-density lipoprotein (HDL) metabolism to exhibit anti-atherosclerotic functions, such as protection against oxidation and regulation of cholesterol efflux (CE). 17,18 Furthermore, ApoM is an indispensable constitute for HDL and preβ-HDL formation to enhance CE. 19 Macrophage reverse cholesterol transport (RCT) is a protective mechanism of AS which can inhibit accumulation of excessive cholesterol in macrophages, while CE, mediated by three membrane proteins, including ABCA1, ABCG1 and SR-BI, is the key point for RCT. 20,21 As far as we known, TUG1 is associated with AS, but the mechanism of TUG1 on AS progression are not fully addressed. Considering the protective role of ApoM in AS, the possibility of TUG1 regulating ApoM in AS also needs to be identified.
MicroRNAs (miRs) have vital roles to play in CVDs, including hypertension and cardiovascular remodelling. 22 miR-92a has been emerged as a potential biomarker in AS as miR-92a silence could promote the expressions of angiogenesis factors. 23 In current study, we found TUG1 can bind miR-92a to regulate FXR1. Previous data proved FXR1 can negatively regulate ApoM. 24 Therefore, we hypothesize that TUG1 regulates ApoM expression mainly through miR-92a/FXR1 axis to promote AS progression. Collected evidence in this study supported that TUG1 in liver tissues could compete with FXR1 to regulate miR-92a expression, which results in the down-regulation of ApoM in both liver tissues and in plasma, consequently leading to inhibition of RCT and deterioration of AS.

| Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
Total RNAs from mouse liver tissues or macrophages were isolated using TRIzol (Invitrogen) and subjected to reverse transcript using a reverse transcript kit (TaKaRa) according to the instructions indicated in the kit.
LightCycler 480 (Roche) was used for mRNA detection. The conditions were set based on the instructions on the PCR kit (SYBR Green Mix; Roche Diagnostics). The PCR amplification was conducted based on following conditions: 95℃ for 10 seconds, 45 cycles of 95℃ for 5 seconds, 60℃ for 10 seconds and 72℃ for 10 seconds, and extension at 72℃ for 5 minutes. Each reaction for PCR was performed in triplicate. U6 or GAPDH was used as internal control. The mRNA expressions were calculated based on 2-ΔΔCt method. 25

| Western blot
Mouse liver tissues or macrophages were lysed using RIPA lysis buffer (beyotime) to obtain the protein sample. After the protein concentration was measured using a BCA kit (beyotime), certain volume of protein was mixed with loading buffer (beyotime) for boiling water bath for 3 minutes for degeneration. The proteins were subjected to electrophoresis at 80 V for 30 minutes and then at 120 V for 1-2 hours. Membrane was transferred at a current of 300 mA for 60 minutes and then washed in washing buffer for 1-2 minutes before blocking at room temperature for 1 hou or at 4℃ for overnight.

| CE rate
Peritoneal macrophages were isolated from mice in each group. RAW264.7 cells were treated with 50 mg/L ox-LDL. After cell transfection of LV-TUG1 or sh-TUG1 for 12 hours, the culture medium for cell culture was removed and replaced with DMEM culture medium containing 0.2% BSA. Cells were then incubated with 1 mCi/L

| Dual luciferase reporter assay
The binding sites of TUG1 and miR-92a, and that of miR-92a and FXR1 were predicted by starBase (http://starb ase.sysu.edu.cn/). The wide and mutant sequences were designed accordingly and named as wt-TUG1, mut-TUG1, wt-FXR1 or mut-FXR1. The sequences inserted into pGL3-Basic and co-transfected with miR-92a mimic (30 nM) or its negative control into HEK293T cells. The firefly luciferase activity and Renilla luciferase activity in each group were measured. The Renilla luciferase activity was used as internal control. Luciferase activity = Firefly luciferase activity/Renilla luciferase activity.

| RIP
Cells in each group was collected and washed in PBS for twice before centrifuged at 1500 rpm for 5 minutes. Then, certain volume of RIP lysate was added to fully mix with cells. Beads were re-suspended.
Then, 5 µg of Ago2 antibody (ab32381, 1:50; Abcam) was added while cells in negative control group were incubated with IgG antibody at room temperature for 30 minutes. After that, supernatant was abandoned. Cells were vibrated with 500 µL of RIP Wash Buffer to remove the supernatant for twice. Cells were then added with 500 µL of RIP Wash Buffer for vibration and maintained on ice for further use. The supernatant in magnetic bead tube was removed and 900 µL of RIP Immunoprecipitation Buffer was added in each tube. Cell lysis was centrifuged at 14,000 rpm, 4℃ for 10 minutes. Then, 100 µL of supernatant was added into the bead-antibody complex to make the final volume of 1 mL for incubation at 4℃ overnight. After transient centrifugation, the supernatant was removed and then 500 µL of RIP Wash Buffer was added for vibration with the supernatant removed.
The complex in centrifuge tube was washed for six times. RNA purification: 150 µL of Proteinase K Buffer was added to re-suspend the bead-antibody complex at 55℃ for 30 minutes with the supernatant removed. The RNA was extracted for qRT-PCR.

| Dissection of aorta for oil red O staining
Mice were fasted overnight before being sacrificed. Mice were anesthetized by injection of pentobarbital sodium (50 mg/kg). The chest of TA B L E 1 Primer sequences for genes in quantitative reverse transcription polymerase chain reaction

Name of primer Sequences
mouse was opened and the aorta was isolated. The aorta was washed before fixation using 4% paraformaldehyde for 24 hours. The aorta was dissected and fixed using needles with adventitia removed. The aorta was stained in oil red O staining for 5 hours and differentiated under 60% isopropanol. The differentiation liquid shall be replaced for several times till the AS plaques were in red while arterial wall in white.
The aortic root was also made into slices for oil red O staining.

| Immunohistochemistry
The aortic root slices (4 µm) were baked for 20 minutes before dewax using xylene and washing in distilled water. Slices were then washed in PBS for three times and added with 3% H 2 O 2 at room temperature for 10 minutes. PBS washing for three times before antigen repair. After that, slices were washed in PBS for three times and blocked with goat serum for 20 minutes at room temperature.
Primary antibody of MOMA-2 (ab33451, 1:50) was added for incubation at 4℃ overnight. PBS washing for three times before slices were incubated with secondary antibody for 1 hours. PBS washing for three times before colour development with DAB for 1-3 minutes. Then, slices were stained with haematoxylin, dehydrate and transparent and sealed.

| Measurement on plasma
Total cholesterol (TC), triglyceride (TG), low-density lipoprotein in plasma were correspondingly determined using detection kits (Nanjing Jiancheng Bioengineering Institute) based on instructions.

| Measurement on inflammatory cytokines
Inflammatory cytokines including TNF-α, IL-1β and IL-6 were measured using ELISA kit (R&D Systems) according to instructions.

| Statistical analysis
Data were analysed using GraphPad prism7 software. All data were expressed using mean ± standard deviation (x ± SD). Comparison between two groups was achieved through t test and data among groups was analysed using one-way analysis of variance with Dunnett's multiple comparisons test as post hoc test. P value of <0.05 was considered to have statistical significance.

| Up-regulation of TUG1 and down-regulation of ApoM in liver tissues of HFD dieted mice
The expression levels of TUG1 and ApoM in liver tissues were detected by qRT-PCR and Western blot. The detection showed that the expression of TUG1 in mice in HFD group is 1.52 times than that of ND group

| TUG1 competes with FXR1 to bind miR-92a
Starbase (http://starb ase.sysu.edu.cn/) software found the binding site of FXR1 with TUG1 and with miR-92a. Wide type (wt) and mutant type (mut) of binding site were designed for dual luciferase reporter assay ( Figure 5A). No difference was revealed in cells inserted with mut-TUG1 between miR-92a mimic group and mimic NC group while the luciferase activity in wt-TUG1 inserted cells in miR-92a

F I G U R E 5
The competitive relationship of TUG1 with FXR1 to bind miR-92a. Wide type and mutant type of the bind sites predicted by Starbase online software (A). Dual luciferase reporter assay verified the binding of TUG1 with miR-92a (B), and the target relationship of FXR1 with miR-92a (C), n = 3. After mouse liver NCTC 1469 cells were transfected with miR-92a mimic, miR-92a inhibitor or its negative control, qRT-PCR and Western blot were applied to measure the expression levels of miR-92a (D), TUG1 (E) and FXR1 (F, G), n = 3. RIP assay was performed to verify the enrichment of miR-92a, TUG1 and FXR1 in RNA-induced silencing complex (RISC) (H), n = 3. *, compared with mimic NC group, P < 0.05; **, compared with mimic NC group, P < 0.01; #, compared with inhibitor NC group, P < 0.05; ##, compared with inhibitor NC group, P < 0.01; &&&, compared with IgG group, P < 0.001 mimic group was much reduced when compared with that in mimic NC group ( Figure 5B, P < 0.05). As expected, cells inserted with mut-FXR1 showed no difference in luciferase activity between miR-92a mimic group and mimic NC group. However, cells inserted with wt-FXR1 in miR-92a mimic group reduced luciferase activity in comparison to that in mimic NC group ( Figure 5C, P < 0.05). Collectively, both TUG1 and FXR1 can directly bind miR-92a.
Then, we further verified the relationship among the TUG1, FXR1 and miR-92a in mouse NCTC 1469 cells. RT-PCR and Western blot after cell transfection found that miR-92a mimic transfection would evidently up-regulate miR-92a expression level ( Figure 5D, P < 0.05) and down-regulate both TUG1 ( Figure 5E, P < 0.05) and FXR1 ( Figure 5F,G, P < 0.05). Different expression pattern was found in NCTC 1469 cells after transfection of miR-92a inhibitor. Collectively, it was proved that miR-92a expression level was negatively associated with those of TUG1 and FXR1. Anti-AGO2 RIP assay was also performed to verify the enrichment of miR-92a, TUG1 and FXR1 in RNA-induced silencing complex (RISC). RIP results demonstrated that AGO2 group found expressions of miR-92a, TUG1 and FXR1 while IgG group barely found the existence of these three factors ( Figure 5H).

| TUG1/miR-92a/FXR1 axis regulates ApoM in NCTC 1469 cells
As we proved above that TUG1 in liver tissues of mice negatively regulate ApoM, but the regulatory role of miR-92a/FXR1 on ApoM remains to be determined. In this regards, we aim to explore the ef-

| D ISCUSS I ON
Data in this study demonstrate the promotive role of TUG1 in AS progression by inhibiting ApoM expression levels as well as blocking the RCT in macrophages mainly through miR-92a/FXR1 axis.
As a result, knock-down of TUG1 markedly protects against the progression of AS and promotes CE rate. Importantly, inhibition of TUG1 also increases the expression levels of ABCA1 and ABCG1 in C57BL/6J mice, both of which are critical genes involved in pathogenesis of AS. TUG1 can compete with FXR1 to bind with miR-92a.
Thus, our current works suggest that TUG1 can serve as a promising target for the prevention and treatment of AS.
High-density lipoprotein is a critical effector in RCT which facilitates the transport of cholesterol from cells in the vessel wall to the liver to maintain balance of cholesterol in arterial wall. 26 Meanwhile, HDL is one of the endogenous factors that restores the optimal endothelial function in vascular disease and plasma apolipoprotein M-containing HDL (ApoM + HDL) can activate the G protein-coupled sphingosine 1-phosphate (S1P) receptors to promote vascular barrier function. 27 The CE from macrophage foam cells requires the participation of both ABCA1 and preβ migrating HDL. 28 Evidence from previous study supported that ApoM are closely associated with HDL. 29 Furthermore, ApoM also has certain role to play in the formation of the pre-HDL particles to exert its anti-atherogenic functions. 30 It is well known that ApoM is a specific S1P chaperone, whose overexpression can contribute to the stimulation and formation of apoM/S1P complex in HDL by increasing the synthesis and secretion of its cargo, S1P. 31 In vitro experiments in current study found down-regulated expression level of ApoM in liver tissues and plasma of HFD mice. In addition to that, we found that the abnormal lipids metabolism also associated with the abnormal expression of TUG1, which was highly expressed in liver tissues of mouse. Those results are consistent with the promotive role of TUG1 in AS reported in previous studies. 13 After mice were injected with LV-TUG1, sh-TUG1 or negative control, qRT-PCR and Western blot were applied to measure the expression level of TUG1 and ApoM in liver tissues or in plasma. TUG1 in liver tissues determined by qRT-PCR (A), n = 6; mRNA and protein expression of ApoM in liver tissues were measured (B, C), n = 6; mRNA and protein expression of ApoM in plasma were measured (D, E), n = 6; qRT-PCR (F) and Western blot (G) were used to measure the expression of miR-92a and FXR1 in liver tissues of mice, n = 6; concentrations of TC, TG, HDL-C and LDL-C were detected by kits (H), n = 6; ELISA was utilized to detect the concentrations of TNF-α, IL-1β and IL-6 (I), n = 6; oil red O staining detected the plaque area in aorta (J) and arterial wall (K), n = 6. Immunohistochemistry detected the expression level of MOMA-2 in aorta root (L), n = 6. TC, cholesterol; TG, triglyceride; LDL-C, low-density lipoprotein cholesterin; HDL-C, high-density lipoprotein cholesterol; *, compared with LV-NC group, P < 0.05; **, compared with LV-NC group, P < 0.01; # compared with sh-NC group, P < 0.05; ##, compared with sh-NC group, P < 0.01 | 8847 YANG ANd LI knock-down plasmid. Interestingly, the data obtained that overexpression of TUG1 can inhibit miR-92a while promote FXR1 in liver tissue of mice, while reverse results were observed in mice transfected with sh-TUG1. TUG1 can compete with FXR1 for interaction with miR-92a. In addition to that, knowledge obtained from other study supported the negative regulatory of FXR1 on ApoM. 24  In conclusion, we have revealed that TUG1 is a potential target for AS. The mechanism of TUG1 deteriorating AS is through miR-92a/FXR1 axis to inhibit the expression of ApoM and inhibit its anti-atherogenic effect.

ACK N OWLED G EM ENT
Thanks for all the contributors and participants.

CO N FLI C T O F I NTE R E S T
The authors declare they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used or analysed during the current study are available from the corresponding author on reasonable request.