An Intronic Risk SNP rs12454712 for Central Obesity Acts As an Allele-Specific Enhancer To Regulate BCL2 Expression

Obesity is a medical condition characterized by abnormal fat accumulation that may impair health (1). While overall obesity confers a threat to individual health, depot-specific accumulation of fat is also of great importance in determining this threat. Central obesity has especially been recognized as a risk factor for cardiometabolic diseases independent of overall BMI (2,3). An easily accessible measure of central obesity is the waist-to-hip ratio (WHR). There is strong evidence for genetic predisposition to WHR, with an estimated 30–60% of the risk explained by heritable factors (4,5). Finding its genetic susceptibility loci and understanding the molecular mechanisms might offer new insights into appropriate therapies.

Genome-wide association studies (GWAS) have successfully identified >300 genetic loci for central obesity (5,6). Some of these loci are located in coding or untranslated regions of genes with important functions, such as BDNF (6), LPL (6,7), and SH2B1 (5,6). However, some loci are located in noncoding regions, prompting a great challenge in illustrating the molecular mechanism linking these loci with the development of obesity. For example, the single nucleotide polymorphism (SNP) rs12454712 was identified to be associated with WHR adjusted for BMI (WHRadjBMI) initially in a large-scale GWAS (5) and further replicated in another two studies (6,8). SNP rs12454712 is located in the intronic region of BCL2, which encodes an integral outer mitochondrial membrane protein that blocks the apoptotic death of cells (9). During the development of obesity, the expansion of adipose tissue results in the activation of adipocyte apoptosis (10), which is a key initial event that contributes to adipose tissue macrophage infiltration, obesity-associated insulin resistance, and hepatic steatosis (11). Inhibition of adipocyte apoptosis is believed as a potential therapeutic strategy for the treatment of obesity-related metabolic disorders (11). It was reported that upregulation of BCL2 could lower the apoptosis sensitivity of human adipocytes (12). In addition, a previous study showed that BCL2 was significantly decreased in obese subjects compared with lean subjects in adipose tissues and that inhibition of BCL2 was linked to the development of insulin resistance (13). Therefore, BCL2 may constitute a possible target for the specific regulation of adipocyte apoptosis (12). However, whether rs12454712 could affect the expression of BCL2 is still unknown.

Recently, with the help of epigenomic data released by the Encyclopedia of DNA Elements (ENCODE) (14) and the Roadmap Epigenomics Project (15), researchers began to realize that some noncoding variants might constitute putative regulatory elements. The target genes of intronic SNPs might be distal genes rather than its located genes. For example, an obesity-risk SNP located in the intron of FTO was found to be involved in adipocyte browning through regulating the function of IRX3 and IRX5 rather than FTO (16). Interestingly, genes within 1 megabase of rs12454712 also include PHLPP1 (464 kilobase [kb]), whose expression in obese individuals has also been reported to be associated with insulin resistance (17). Therefore, it is extremely interesting to find out the target gene of this loci and investigate downstream regulatory mechanisms.

In this study, we hypothesized that the intronic SNP rs12454712 associated with WHRadjBMI might act as a regulatory element to influence the expression level of the target gene. To test our hypothesis, we conducted a series of analyses using data from sources including epigenomic annotation, promoter capture Hi-C, and expression quantitative trait locus (eQTL). Functional experiments were also performed for further validation. We demonstrated that the intronic SNP rs12454712 acts as an enhancer to regulate the expression of BCL2. Our findings illustrate the molecular mechanisms behind the GWAS signals at this loci for central obesity, which would be a potential and promising target for developing appropriate therapies.

We checked the association signals at 18q21.33 on WHRadjBMI by using a previously published summary data set (6). As the largest GWAS meta-analysis on WHRadjBMI so far, it includes up to 697,734 individuals from the UK Biobank and the Genetic Investigation of Anthropometric Traits (GIANT) data. Lead SNPs attaining P < 5 × 10⁻⁸ were firstly selected. Loci were defined as 500 kb upstream and downstream of the lead SNPs. Where loci overlapped, they were merged as a single locus. In each locus, using linkage disequilibrium (LD) information of 50,000 unrelated White British participants randomly selected from the UK Biobank, we considered SNPs in LD (r² > 0.8) with the lead association signal as potential causal variants. LDBlockShow (18) was used to visualize the LD heat map and GWAS association plot.

We annotated epigenetic regulatory features for SNPs of interest using chromatin immunoprecipitation (ChIP) sequencing data from the ENCODE (14) and Roadmap (15) project, including enhancer markers (H3K27ac and H3k4me1) in adipose/adipocytes. All data were visualized using the WashU Epigenome Browser (https://epigenomegateway.wustl.edu/).

The eQTL results from Genotype-Tissue Expression (GTEx v6p) database (19) in subcutaneous and visceral omentum adipose were obtained to explore the association between independent obesity SNPs genotype and expression of genes within 1 megabase region. For significant eQTL results, we further tested whether the GWAS and eQTL signals were colocalized by using Bayesian colocalization (20).

Promoter capture Hi-C data on human adipocytes and other obesity-related tissue/cells (e.g., dorsolateral prefrontal cortex, hippocampus, liver, psoas, spleen, and GM12878) were obtained from previously published studies (21–23). BEDTools (24) was used to extract our prioritized SNPs and genes within the same pair of Hi-C interaction regions.

The effect of rs12454712 on transcription factor (TF) binding motifs was analyzed using the MEME Suite toolkit (25) using motifs collected from three public databases: JASPAR (26), Homo sapiens Comprehensive Model Collection (HOCOMOCO) (27), and SwissRegulon (28).

Simpson-Golabi-Behmel syndrome (SGBS) cells were cultured in DMEM/F12 medium supplemented with 10% FBS (Invitrogen, Waltham, MA), penicillin (100 units/mL), and streptomycin (0.1 mg/mL) in 5% CO₂ at 37°C.

The 3T3-L1 murine preadipocytes (ATCC) were differentiated into adipocytes as described previously (29). Briefly, 2 days after confluence (day 0), cells were treated with 10% FBS-DMEM with 5 μg/mL insulin (Sigma-Aldrich), 1 μmol/L dexamethasone (Sigma-Aldrich), and 0.5 mmol/L 1-methyl-3-isobutyl-xanthine. After 2 days, cells were switched to DMEM with 10% FBS and 1 μg/mL insulin for up to 11 days, with a medium change every 2 days. RNA was extracted on day 0, 2, 4, 6, 8, and 11 to measure the expression levels of Bcl2 and adipogenic differentiation marker genes (Adipoq, C/Ebp-α, and Lpl). The primers are listed in Supplementary Table 1.

A 1,200-base pair (bp) putative enhancer fragment containing rs12454712, a 1,500-bp promoter fragment surrounding the BCL2 transcription start site, was amplified from genomic DNA and cloned into a pGL3-Basic vector. Site-directed mutagenesis was performed by using the QuickChange II Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA). The primers are listed in Supplementary Table 1. The constructed plasmids were transfected into SGBS cells by using the ViaFect Transfection Reagent (Promega, Fitchburg, MA). The phRL plasmid containing Renilla luciferase (Promega) was simultaneously transfected into the cells as an internal control. After 48 h of transfection, the luciferase activity with Renilla luciferase vector phRL as the internal reference luciferase signals was detected by using the Dual-Luciferase Reporter Assay System (Promega).

We used the clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 technology to delete the enhancer region containing rs12454712, as described previously (30). The single guide RNA (sgRNA) was designed by using the platform maintained by the Zhang Laboratory at the Broad Institute. The indel frequency of candidate sgRNAs was detected by T7 endonuclease I assay after cotransfected into human embryonic kidney (HEK) 293T cells with Cas9-expression plasmid (42876; Addgene) using the ViaFect Transfection Reagent (Promega). The best pair of sgRNAs were selected and cloned into self-constructed dual-sgRNA plasmid, cotransfected into SGBS cells with lentiCRISPR v2 (52961; Addgene). After selection with puromycin (3 μg/mL), the cells were harvested for DNA and RNA extraction. The primers are listed in Supplementary Table 1.

The ZNF329 shRNAs (ZNF329-sh1 and ZNF329-sh1) and negative control (NC) shRNA (ZNF329-shNC) were constructed into pcDNA3.1 vector (V79020; Invitrogen). Human ZNF329 cDNA was cloned into the pcDNA3.1 by PCR to generate the overexpression plasmid. After construction, 2 μg of each plasmid were independently transfected into SGBS cells by using the ViaFect Transfection Reagent (Promega). After 48 h of transfection, RNA was isolated to detect the mRNA expression. Moreover, ZNF329 shRNA plasmids were cotransfected with the luciferase plasmid (rs12454712-C allele or rs12454712-T allele enhancer with BCL2 promoter) into SGBS cells for luciferase reporter assay. The measurement of luciferase activity is the same as described in the Dual-Luciferase Reporter Assay section. The primers are shown in Supplementary Table 1.

ChIP assays were performed in HEK293T cells by using the SimpleChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology), as described previously (31). A total of 50 μg of sonicated chromatin was incubated with 3 μg of ZNF329 antibody (sc-377455; Santa Cruz Biotechnology, Dallas, TX), the histone H3 rabbit antibody (positive control), or normal rabbit IgG (negative control). Protein-DNA complex was precipitated with agarose beads and reversely cross-linked by 5 mol/L NaCl and proteinase K. The DNA fragments were purified using DNA purification spin columns for subsequent quantitative PCR (qPCR) with allele-specific primers listed in Supplementary Table 1.

DNA was isolated using the TIANGEN Genomic DNA Extraction Kit (catalog no. DP304; TIANGEN Biotech, Beijing, China). RNA was isolated with TRIzol reagent (Invitrogen), and 5 mg of total RNA per reaction was used to synthesize the cDNA with the SuperScripts II First-Strand cDNA Synthesis Kit (Invitrogen). Real-time qPCR was performed with the QuantiTect SYBR Green PCR Kit (QIAGEN, Dusseldorf, Germany) by the CFX Connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA). Samples were tested in a 96-well format in triplicate. For experiments in the SGBS cells, GAPDH was used as an endogenous control.

Antibodies against ZNF329 and BCL2 were purchased from Santa Cruz Biotechnology. Antibody against β-actin was purchased from Abcam (Cambridge, U.K.). Cells were lysed with Western and immunoprecipitation lysis buffers (Beyotime, Shanghai, China). The lysates were centrifuged at 13,000g for 6 min at 4°C. The supernatants were collected, and protein concentrations were determined with a BCA Protein Assay kit (Pierce, Rockford, IL). Equal aliquots of protein samples were loaded onto 10% SDS-PAGE gels and transferred to nitrocellulose membranes. Membranes were blocked with 5% nonfat milk for 1 h and then incubated with primary antibodies at 4°C overnight, after which they were incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. Western blots were developed using an enhanced chemiluminescence Western blotting detection kit (Pierce). The corresponding images were captured by the ChemiDoc XRS+ system (Bio-Rad).

We constructed shRNAs (Bcl2-sh1, Bcl2-sh2, and Bcl2-shNC) as described previously (32) and cloned them into a lentiviral backbone vector. Then we transfected the lentiviral vector (1.5 μg each), pVSVG, and psPAX2 into HEK293T cells using 12 μL Lipofectamine 2000 to package the shRNA lentivirus. For viral transduction, the 3T3-L1 cells were cultured with lentivirus-containing medium with 8 mg/mL polybrene (Sigma-Aldrich). Cells were screened by medium containing 1 mg/mL puromycin (Sigma-Aldrich) 24 h after transduction. Total RNA was extracted and real-time qPCR was used to confirm whether Bcl2 was knocked down successfully. Then the cells were up to adipogenic differentiation as mentioned above. RNA was extracted on day 0, 2, 4, 6, 8, and 11. Real-time qPCR was performed to measure the expression levels of Bcl2, Adipoq, C/Ebp-α, and Lpl. The 18s rRNA was used as endogenous control. The primers are shown in Supplementary Table 1.

For Bcl2 knockdown 3T3-L1 cells, we detected cell apoptosis during adipogenic differentiation on day 4, 8, and 11. The TUNEL reaction was performed using the one-step TUNEL apoptosis assay kit (C1089; Beyotime). The cells were incubated with 5-bromo-4-chloro-3-indolyl-phosphate substrate labeled with Cy-3 for 40 min at 37°C. Slides were observed under the microscope (magnification ×200). Fluorescence density was assessed using ImageJ (33), and the results were presented in mean optical density (integrated optical density/area).

The mRNA expression levels were calculated with the 2^−ΔΔCt method and displayed as the mean ± SD. The unpaired two-tailed Student t test was used to calculate the P value. All analytical data were obtained from three independent experiments, and each experiment was performed in triplicate.

We collected data sets from the Gene Expression Omnibus database to investigate the relationship between BCL2 and obesity. The data we used are subcutaneous adipose tissue of 770 men who are part of the Metabolic Syndrome in Men (METSIM) study (GSE70353), needle biopsy subcutaneous adipose tissues of 9 lean and 9 obese subjects (GSE12050), subcutaneous adipose tissue in 8 lean and 7 obese children (GSE29718), and pooled inguinal fat of male C57BL/6J mice exhibiting high or low weight gain after 4 weeks on a diet high in saturated fat. In addition, gene expression data of omental and subcutaneous adipose tissue samples obtained from insulin-sensitive obese and insulin-resistant obese patients (GSE20950) were also used to see whether BCL2 is correlated with insulin-resistance.

We used the Mouse Genome Database (MGD) (34), which is a community model organism genetic and genome resource for the laboratory mouse, to check the function of BCL2.

The data sets generated during the current study are available from the corresponding author upon reasonable request.

For WHRadjBMI, there was only one significant association signal (rs12454712, P = 1.33 × 10⁻²⁰) (Fig. 1A). With the threshold of r² > 0.8, no SNP was found to be in LD with rs12454712 in the European population (Fig. 1B), indicating the high potential functionality of rs12454712.

To evaluate the functionality of SNPs, we used the epigenetic profiles in adipose to annotate rs12454712. We found that rs12454712 was located within the strong epigenetic enhancer markers (H3K27ac and H3k4me1) (Fig. 2A). Therefore, these analyses prioritized rs12454712 as a functional independent variant for further experimental validation.

rs12454712 is located in the intronic region of BCL2. We performed cis-eQTL analysis using the data of adipose tissues from the GTEx project to identify its target genes. Of the 20 surrounding genes detected, we found that the minor allele C of rs12454712 was exclusively and significantly associated with increased expression of BCL2 (P = 9.03 × 10⁻⁴, β = 0.13) (Fig. 2B) after multiple testing correction in subcutaneous adipose. Strong colocalization evidence between GWAS and BCL2 eQTL signals was also detected (posterior probabilities for hypothesis 4 was equal to 0.80). Similarly, individuals with the CC genotype of rs12454712 also showed higher BCL2 expression levels than individuals with TT genotypes in six other tissues (Supplementary Fig. 1).

Because rs12454712 is located ∼140 kb away from the promoter of BCL2, we explored the potential long-range chromatin interactions between rs12454712 and nearby genes within 1 megabase region using capture Hi-C data on human adipocytes (22). The results showed that rs12454712 is located in a region interacting with another region containing the promoter of BCL2 only (chr18:60845031–60848302 interact with chr18:60982875–60990419, Capture HiC Analysis of Genomic Organisation [CHiCAGO] score = 5.49). In addition, capture Hi-C data from six other tissues/cells also support the interaction between rs12454712 and the promoter of BCL2 (Fig. 2C). Together, these data indicated the potential long-range regulation on BCL2 for rs12454712.

Genotyping assay revealed that the SGBS cell line is a homozygous CC genotype for rs12454712. To directly validate the long-range regulation between rs12454712 and BCL2, we deleted a 181-bp enhancer region containing rs12454712 using CRISPR/Cas9 in SGBS cells. As shown in Fig. 3A, significantly decreased BCL2 expression (P < 0.001) was detected in enhancer-deleted cells (knockout) compared with the normal cells (wild-type), indicating that BCL2 was the target gene underlying distal enhancer-promoter regulation (Fig. 3A). Next, to experimentally validate the allelic regulation between rs12454712 and BCL2, we compared the regulatory activity of genomic fragments containing different genotypes of rs12454712 by using dual-luciferase reporter assays in SGBS cells. We found that both alleles of rs12454712 could reinforce the expression of the pGL-3 basic promoter and BCL2 (P < 0.001) (Fig. 3B and C). Particularly, the rs12454712-C allele had significantly enhanced effect compared with the rs12454712-T allele (P < 0.01), which is consistent with the eQTL analysis results.

We next explored the functional mechanism underlying rs12454712 as the strong allele-specific enhancer on BCL2. The motif analysis showed that ZNF329 could specifically bind to rs12454712-C (Fig. 4A). To validate the allelic binding affinity of ZNF329 on rs12454712, we performed an allele-specific ChIP-qPCR assay in rs12454712-CT 293T cells (Fig. 4B). We found that ZNF329 was preferentially recruited by the rs12454712-C allele compared with rs12454712-T allele (P < 0.001) (Fig. 4C).

To assess the transcriptional influence of rs12454712 on BCL2 expression mediated by ZNF329, we suppressed ZNF329 by two independent shRNAs (shRNA-1 and shRNA-2) in SGBS cells. Compared with the negative control shRNA-transfected cells, we detected significant decline of BCL2 expression in ZNF329 knockdown SGBS cells (P < 0.001) (Fig. 4D). In addition, the expression of BCL2 was significantly enhanced in SGBS cells overexpressing ZNF329 (P < 0.001) (Fig. 4E).

To validate the allele-specific regulatory role of ZNF329 for rs12454712, we cotransfected ZNF329 shRNA and the putative enhancer plasmid (rs12454712-C allele or rs12454712-T allele) with the BCL2 promoter into SGBS cells. As shown in Fig. 4F, compared with controls, the luciferase reporter assay detected significant decline of BCL2 expression with the C allele enhancer plasmid. Meanwhile, we detected no significant variation of BCL2 expression with the T allele enhancer plasmid. These results supported the potential allele-specific regulatory role of ZNF329 for rs12454712.

In the 3T3-L1 cells, we detected an induction of Bcl2 in the normal adipogenic differentiation process (Fig. 5A). To further investigate the function of Bcl2, we knocked down its expression in 3T3-L1 cells. As shown in Fig. 5B, transfection with Bcl2-sh1 or Bcl2-sh2 significantly reduced the Bcl2 expression compared with that in the Bcl2-shNC group. Meanwhile, during the adipogenic differentiation process, the expression levels of adipogenic differentiation marker genes (Adipoq, C/Ebp-α, and Lpl) were significantly downregulated in the Bcl2-sh1 and Bcl2-sh2 groups (Fig. 5C). These results suggested that knockdown of Bcl2 inhibited the adipogenic differentiation of 3T3-L1 cells. In addition, as shown in Fig. 5D, compared with the Bcl2-shNC group, the Bcl2-sh1 or Bcl2-sh2 group showed significantly increased cell apoptosis on day 4, 8, and 11 of the adipogenic differentiation process.

Pearson correlation analysis using gene expression data of subcutaneous adipose tissues from 770 men (GSE70353) showed that BCL2 expression was negatively correlated with WHR (R = −0.22, P = 9.20 × 10⁻¹⁰) and WHRadjBMI (R = −0.10, P = 5.77 × 10⁻³) (Supplementary Fig. 2A and B). Consistently, differential expression analysis in the other two data sets (GSE12050 and GSE29718) Supplementary Fig. 2C and D) showed that the expression level of BCL2 in lean controls was significantly higher than in obese subjects. The expression level of BCL2 in mice exhibiting low weight gain after 4 weeks on a diet high in saturated fat was significantly higher than the group with high weight gain (Supplementary Fig. 2E). In addition, when compared with the insulin-sensitive obese patients, the expression levels of BCL2 in both omental and subcutaneous adipose tissues of insulin-resistant obese patients were significantly lower (Supplementary Fig. 2F).

According to the information from MGD, because of premature death with dysfunction of kidney and immune system, global Bcl2 knockout (Bcl2^−/−) mice did not appear to have obvious obesity-related phenotypes (MGI Ref ID including J:73316, J:34326, J:15224, J:22391, and J:175730). However, mice contain knock-in mutations in BCL2 phosphorylation sites and showed decreased exercise endurance and impaired chronic exercise-mediated protection against glucose intolerance induced by a high-fat diet (J:181308).

The intronic SNP rs12454712 has been reported to be associated with WHRadjBMI in previous GWAS. Deciphering the underlying functional mechanisms is critical to translate GWAS findings into clinically useful information. Here, through the combination of bioinformatics analyses and various functional experiments, we demonstrated that rs12454712 acted as an allele-specific enhancer to modulate the expression of its located gene BCL2. We corroborated that the enhancer activity of rs12454712 was elevated by the TF ZNF329 preferentially binding to the rs12454712-C allele. Downregulation of Bcl2 in 3T3-L1 cells increased cell apoptosis and suppressed the expression of adipogenic differentiation marker genes. Taken together, our results highlighted the allele-specific functional role of rs12454712 regulating the expression level of BCL2 (Fig. 6).

We observed that Bcl2 was upregulated in the adipogenic differentiation process of 3T3-L1 cells. Knocking down Bcl2 significantly increased the apoptosis of 3T3-L1 cells during adipogenic differentiation. Consistently, a previous study showed that knocking down the expression of BCL2 in SGBS cells led to a significant upregulation of apoptosis rates (12). In addition, we also observed that knocking down the expression of Bcl2 inhibited the expression of adipogenic differentiation marker genes (Adipoq, C/Ebp-α, and Lpl). Because mature adipocytes are less sensitive to apoptotic stimuli than preadipocytes (35), the regulation of Bcl2 on adipogenic differentiation may lower the apoptosis sensitivity of adipocytes. The importance of BCL2 in obesity is highlighted by the fact that we observed its expression levels in adipose tissue correlated with WHR and WHRadjBMI. Consistently, a previous study showed that BCL2 was significantly decreased in adipose tissues in obese subjects compared with lean subjects (13). We also observed that the expression of BCL2 was higher in mice with low weight gain after a diet high in saturated fat than in mice with high weight gain.

In addition, BCL2 in adipose tissues of insulin-resistant obese patients was significantly lower than in the insulin-sensitive obese patients, suggesting its involvement in the development of insulin resistance. Tinahones et al. (13) also reported that BCL2 inhibition in human adipose tissue was linked to the development of insulin resistance. Because adipocyte apoptosis is a key initial event that links adipose tissue inflammation and obesity-related metabolic disorders (10), inhibition of adipocyte apoptosis through targeting BCL2 may be a potential therapeutic strategy for treating obesity-related metabolic disorders. On the other hand, disruption of the BCL2–beclin-1 complex is crucial for stimulus (e.g., exercise or starvation)-induced autophagy (36). According to the information from MGD (37), mice with knock-in mutations in BCL2 phosphorylation sites were deficient in stimulus-induced autophagy, leading to decreased exercise endurance and impaired chronic exercise-mediated protection against glucose intolerance induced by a high-fat diet intolerance. Therefore, it is also possible that BCL2 is associated with obesity through its regulation of autophagy.

Our study revealed that TF ZNF329 could preferentially bind to the rs12454712-C allele to increase BCL2 expression. ZNF329 belongs to the zinc finger protein family, which is one of the most abundant groups of proteins and has a wide range of molecular functions (38). The relationship between ZNF329 and obesity is still unclear, but other zinc finger protein family members have been proved to be associated with obesity. For example, it was reported that ZNF638 could promote adipogenesis by acting as a transcriptional cofactor of CCAAT/enhancer-binding protein (39). Our results showed that overexpression of ZNF329 in SGBS cells could enhance the expression of BCL2, whereas ZNF329 knockdown resulted in a significant decline of BCL2 expression, supporting its involvement in the development of obesity.

Limitations of this study should be addressed. First, we used eQTL analysis to find the target gene of rs12454712. However, due to the small sample size of current eQTL data, unbalanced signals might exist between eQTLs and GWAS SNPs. We therefore further integrated Hi-C data analysis and functional assays to validate the findings of eQTL analysis. Second, it is worth noting that our regulatory model could not exclude the contribution of other genetic variants on BCL2 expression but instead illustrates the regulatory mechanism of rs12454712.

In summary, through combining various computational analyses and functional assays, we provide a mechanistic insight that rs12454712 acts as an allele-specific strong enhancer to regulate BCL2 expression mediated by ZNF329. Our approach could be used to investigate the functional mechanisms underlying susceptibility loci for other complex diseases.

Funding. This study is supported by the National Natural Science Foundation of China (32070588, 31871264, and 31970569), Natural Science Basic Research Program of Shaanxi (2021JC-02), Natural Science Foundation of Zhejiang Province (LWY20H060001), China Postdoctoral Science Foundation (2020M683454), Shaanxi Postdoctoral Science Foundation, and the Fundamental Research Funds for the Central Universities.

No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.-S.D. and D.-L.Z. wrote and edited the manuscript. S.-S.D. and F.J. collected and analyzed the data. S.-S.D. and Y.G. designed the study. D.-L.Z., X.-R.Z., M.Z., and X.-M.T. performed the functional experiments. Y.R. and J.-B.C. drew the figures. Z.F. and T.-L.Y. reviewed the manuscript. Y.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.