PDGFA-associated protein 1 is a novel target of c-Myc and contributes to colorectal cancer initiation and progression

2.1 Human tissues

Surgically resected CRC tissues were collected from the Department of Gastrointestinal Surgery, Xijing Hospital, which is affiliated with the Fourth Military Medical University (Xi'an, Shaanxi, China). Patients with more than one primary tumor were excluded from the study. All eligible patients were diagnosed between December 1, 2010, and April 25, 2017. All the cases had been followed up annually and the last date of follow-up was July 2019. The staging classification was based on the tumor, node, metastasis (TNM) staging by AJCC, 8th edition. The resected tissue was divided into two parts. One part was stored at −80°C, and the other part was formalin-fixed and paraffin-embedded. Experienced pathologists from the Xijing Hospital reviewed the tissue sections. All participants provided written informed consent, and the study was approved by the Hospital Ethics Committee (KY20213194-1). CRC tissue arrays were obtained from Outdo Biotech (Shanghai, China).

2.2 Mice

Male mice with a C57BL/6 background weighing approximately 22 g were used in this study. Pdap1^fl/fl mice were generated by Biocytogen (Beijing, China). As shown in Supplementary Figure S1A, the loxP fragments were inserted into introns 2 and 4. Exons 3 and 4 were conditionally removed using Cre recombinase. Southern blot analysis was performed during F1 mouse screening to exclude the possibility of random integration (Supplementary Figure S1B-C). Villin-Cre mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Immunodeficient nude mice (BALB/c, 6-8 weeks old) were obtained from Beijing HFK Bioscience (Beijing, China). All animal protocols were approved by the Animal Care and Welfare Committee of the Fourth Military Medical University. All the mice used in this study were housed in a specific pathogen-free facility.

2.3 Live cell microscopy

Live cell fluorescence imaging was performed as described previously [18]. Briefly, the reporter plasmid pCCC-TagRFP (PCNA-cb, Chromotek, Planegg-Martinsried, Germany) was used to obtain stable cell lines via random plasmid integration. A total of 5000 cells were seeded in 96-well imaging plates (Corning, NY, USA). Images were acquired at 37°C at 35 min intervals using Cytation 5 (BioTek Instruments, Winooski, VT, USA) 20 × objective and analyzed using Gen5 (Agilent, Palo Alto, CA, USA) and ImageJ (National Institutes of Health, Bethesda, MD, USA). Median nuclear red fluorescent protein (RFP) intensity was used as a surrogate measurement for DNA replication.

2.4 Dextran sulfate sodium (DSS)-induced colitis

Male mice, weighing approximately 22 g, were used in this study. A 2.5% solution of DSS (MP Biomedicals, Santa Ana, CA, USA) in drinking water was administered for the first five days, followed by regular drinking water for another five days. The mice were monitored daily for body weight. On day 10, they were euthanized, and colon lengths were measured.

2.5 Azoxymethane (AOM)/DSS-induced CAC

Male mice, weighing approximately 22 g, were used in this study. They were treated with an intraperitoneal injection of a single dose of AOM (12.5 mg/kg, MP Biomedicals). After three days, a 2% solution of DSS in drinking water was administered for five days, followed by regular drinking water for another two weeks. The DSS treatment was repeated for two additional cycles. On day 120, the mice were euthanized and the colon tumors were dissected for further analysis.

2.6 Analysis of intestinal permeability and BrdU incorporation

Fluorescein isothiocyanate (FITC)-conjugated dextran with an average molecular weight of 40000 (Merck, Kenilworth, NJ, USA) was dissolved at a concentration of 100 mg/mL in PBS. BrdU (Beyotime, Shanghai, China) was dissolved at a concentration of 20 mg/mL in PBS. FITC was administered to each mouse (500 mg/kg) via oral gavage. One hour later, the mice were intraperitoneally injected with BrdU (100 mg/kg). After 3 h, mice were anesthetized, and 300 μL of blood per mouse was collected. The concentration of FITC-dextran was determined using Cytation 5 with a GFP filter cube. Serum from untreated mice was used to determine the background.

2.7 His pull-down and mass spectrometry

The precooled CRC tissues were ground using a TissueLyser II (QIAGEN, Venlo, Netherlands) at 28 Hz for 5 min and then lysed with lysis buffer. His pull-down assays were performed using Pierce™ His Protein Interaction Pull-Down kit (Pierce, Rockford, IL, USA), where His₆-tagged PDAP1 proteins were the bait, and the lysates from the fresh CRC tissues described above were the prey. The eluted proteins were identified by mass spectrometry. An LTQ-Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) was used to obtain tandem mass spectra of the tryptic peptides. The resulting MS/MS spectra were analyzed using a Thermo Scientific Proteome Discoverer (version 2.5; Thermo Fisher Scientific) for putative protein identification.

2.8 Chromatin immunoprecipitation (ChIP) assay

The binding of c-Myc to the PDAP1 promoter was measured using a ChIP assay as previously described [19]. HCT-8 cells were fixed with 1% formaldehyde and treated with 1 mol/L glycine. After washing with ice-cold PBS, cells were scraped off and lysed with 1% SDS lysis buffer supplemented with a protease inhibitor cocktail on ice. The lysed cells were sonicated to shear the chromatin DNA to an optimal size of approximately 200 bp-1000 bp. Sheared chromatin DNA was immunoprecipitated using antibodies against c-Myc (Santa Cruz Biotechnology, Dallas, TX, USA). Equal amounts of IgG (Santa Cruz Biotechnology) served as negative controls. The immunoprecipitate was then incubated with protein A/G magnetic beads (Millipore, Boston, MA, USA) and pulled down on a magnetic stand. After washing, reverse crosslinking was performed in 5 mol/L NaCl at 65°C overnight. Contaminating RNA was cleaned with ribonuclease A, and the protein was digested with proteinase K. Finally, the sheared DNA recovered from reverse crosslinking was extracted using a DNA extraction kit for further analysis with quantitative polymerase chain reaction (qPCR). The same amount of sheared DNA without antibody precipitation after reverse crosslinking was used as the input control.

2.9 Clonogenic assay

Clonogenic assays were performed as previously described [20]. Briefly, cells were collected and inoculated into 6-well plates. The cells were cultured for 2 weeks until visible clones emerged and fresh medium was replaced every 4 days. After being washed twice with PBS, the cells were fixed and stained with 0.2% crystal violet (Beyotime) in 95% ethanol at 25°C for 30 min. The plates were washed several times with PBS to remove excess dye. The cell clones in each well were counted.

2.10 RNA sequencing

RNA sequencing was performed as previously described [21]. Sequence libraries were generated and sequenced by CapitalBio Technology (Beijing, China). The clean reads were aligned to the reference genome and the processed reads from each sample were aligned using HISAT. Cuffdiff was used to analyze the differentially expressed genes between the samples.

2.11 PDX mouse model

PDX mouse models were obtained from BEIJING IDMO Co. (Beijing, China). For expansion, 2 × 10⁶ cells in 0.1 mL RPMI 1640 medium were subcutaneously inoculated into the right posterior flank of nude mice. Two weeks after tumor cell implantation, mice were randomly allocated and administered in vivo-optimized small interfering RNAs (siRNAs). After four rounds of treatment, the mice were euthanized and the tumors were removed for analysis.

2.12 Statistical analysis

Quantitative results are presented as the mean ± standard deviation, and individual data points are plotted. All statistical analyses were performed for each panel. Differences were compared using two-tailed Student's t-test, as indicated in the figure legends. Spearman's R was used to determine the correlation between the relative expression levels of the genes or proteins. All reported P-values were two-tailed, and P < 0.05 was considered significant.

Detailed methods are available in the Supplemental materials.

3.1 Augmented PDAP1 in CRC correlates with poor survival

To identify the potential key players in CRC initiation and progression, we reanalyzed the single-cell RNA sequencing data of 63689 cells from 23 primary CRC and 10 matched normal mucosa samples (GSE132465) [22], and the proteome of 146 CRC tissues with paired remote normal tissues [23]. Of the 2307 differentially expressed genes (DEGs) with a fold change ≥ 1.5 and adjusted P < 0.05 between 17469 tumor cells and 1070 normal epithelial cells (Supplementary Figure S2A), 104 DEGs exhibiting highly similar expression patterns at the proteomic level were selected for further investigation of their effects on overall survival (OS) using the Gene Expression Omnibus (GEO) database (Supplementary Figure S2B). Of the 54 genes with significant effects on OS (P < 0.01), PDAP1 was brought to our attention because its regulation is obscured by bulk genomic analysis using The Cancer Genome Atlas (TCGA) data and its biological function remains almost unknown, especially in cancer (Supplementary Figure S2C). Immunohistochemistry was performed to examine PDAP1 expression in two independent cohorts of CRC biopsies (Supplementary Table S1 and Supplementary Table S2). The positive expression rates of PDAP1 in CRC tissues were 83.57% (234/280) and 96.00% (72/75), respectively. In paired paracancerous tissues, the positive expression rate of PDAP1 were 23.93% (67/280) and 25.33% (19/75), respectively. The expression of PDAP1 in cancer tissues was significantly higher than that in cancer-adjacent tissues (Supplementary Figure S3A-B). Furthermore, we determined the effects of PDAP1 expression on the prognosis of patients with CRC. We found that the survival of patients with high PDAP1 expression was worse than that of patients with low PDAP1 expression (Supplementary Figure S3C).

3.2 PDAP1 promotes CRC progression

To determine the role of PDAP1 in CRC progression, we evaluated PDAP1 expression in several CRC cell lines (Supplementary Figure S4A) and established stable PDAP1 knockdown (shPDAP1) cell lines (Supplementary Figure S4B-C). The clonogenic assay showed that the number of colonies derived from PDAP1 knockdown cells was significantly reduced (Supplementary Figure S4D). We observed that PDAP1 knockdown decreased cell migration (Supplementary Figure S4E), invasion (Supplementary Figure S4F), and proliferation (Supplementary Figure S4G). Cell cycle analysis showed that PDAP1 knockdown arrested the cell division cycle during the S/G2 phase transition (Supplementary Figure S4H). To monitor DNA replication in living cells, cells were transfected with a marker for DNA replication (proliferating cell nuclear antigen [PCNA]-cb) (Supplementary Figure S4I). We found that PDAP1 knockdown cells displayed weak and prolonged PCNA-cb foci compared to control cells (Supplementary Figure S4J), indicating that completion of DNA replication was delayed by PDAP1 knockdown. In sharp contrast, PDAP1 knockdown had little effect on sphere formation (Supplementary Figure S5A) and apoptosis (Supplementary Figure S5B). Furthermore, cells overexpressing PDAP1 showed enhanced colony formation (Supplementary Figure S5C), migration (Supplementary Figure S5D), invasion (Supplementary Figure S5E), and proliferation (Supplementary Figure S5F).

To test the tumorigenic potential of PDAP1 in vivo, we performed a xenograft assay using immunodeficient mice. Tumor progression was significantly decreased in mice injected with PDAP1 knockdown cells (Supplementary Figure S5G-I). Furthermore, we used a tail vein injection assay to investigate whether the metastatic ability of tumor cells was influenced by PDAP1 in vivo. Sixty days after injection, PDAP1 knockdown cells formed significantly less metastatic nodules in the lungs than the vector control cells (Supplementary Figure S5J-K). These results suggested that colonic PDAP1 may function in tumor initiation or as a neoplastic factor that contributes to CRC progression.

3.3 Pdap1 deficiency impairs mucosal restitution and retards tumor initiation and growth

As PDAP1 is highly conserved between humans and mice (Supplementary Figure S6A), we generated Villin-Cre;Pdap1^fl/fl mice with intestinal epithelial-specific knockout of Pdap1, which appeared to develop normally (Supplementary Figure S6B). In the DSS-induced colitis model, Villin-Cre;Pdap1^fl/fl mice appeared to be markedly more sensitive to the injurious effects of DSS. Weight loss and contraction in colon length were more pronounced in Villin-Cre;Pdap1^fl/fl mice (Figure 1A,B). Accordingly, Villin-Cre;Pdap1^fl/fl mice showed increased dextran permeability, strongly suggesting epithelial barrier dysfunction (Figure 1C). Flow cytometry analysis showed that DSS-treated Villin-Cre;Pdap1^fl/fl mice exhibited increases in neutrophils, macrophages, and CD4⁺ T cell infiltration (Figure 1D) and different patterns of chemokine and cytokine expression when compared to the control mice (Figure 1E). Histological examination confirmed the presence of multiple erosions and intense inflammatory changes, including crypt abscesses. Although the control mice exposed to DSS showed evidence of mucosal erosions, most of these were small and exhibited features typical of mucosal healing, with complete re-epithelialization of most lesions. In contrast, Villin-Cre;Pdap1^fl/fl mice exposed to DSS lacked evidence of re-epithelialization (Figure 1F). Consistent with these results, BrdU and TUNEL staining confirmed that the lack of PDAP1 resulted in impaired epithelial proliferation and increased apoptosis (Figure 1F). Furthermore, RNA sequencing (Figure 1G) and gene set enrichment analysis (GSEA) showed that gene set DNA replication, cell adhesion molecules, cytokine-cytokine receptor interaction, cell cycle, focal adhesion, ECM receptor interaction, chemokine signaling pathway, and mitogen-activated protein kinase (MAPK) signaling pathway were upregulated in the control mice (Figure 1H). These results suggest that intestinal knockout of Pdap1 impaired the defense of the intestinal mucosa in mice.

Furthermore, an AOM/DSS model was established to evaluate the role of Pdap1 in the initiation and development of CAC. Consistent with the results from human CRC tissues, PDAP1 expression was elevated in CAC tissues compared to that in the normal mucosa (Figure 2A). There was greater body weight loss in Villin-Cre;Pdap1^fl/fl mice (Figure 2B), which was consistent with that observed in DSS-induced colitis. Moreover, we found that the number and size of tumors were significantly decreased in Villin-Cre;Pdap1^fl/fl mice compared with control mice (Figure 2C-E). Flow cytometric analysis showed that tumors from Villin-Cre;Pdap1^fl/fl mice exhibited decreased macrophage infiltration (Figure 2F). Histological examination confirmed the presence of colonic carcinoma and BrdU staining confirmed that the absence of PDAP1 resulted in impaired cell proliferation (Figure 2G). These results suggested that intestinal Pdap1 loss decreased CAC initiation and development.

3.4 PDAP1 promotes FRA-1 expression

To explore the mechanisms by which PDAP1 promotes CRC progression, we used gene expression profiling in three PDAP1 knockdown CRC cell lines (Figure 3A). Concordant with its effects on DNA replication in the S phase, GSEA revealed a positive association between the PDAP1 profile and DNA replication signature (Figure 3B). Of the 913 DEGs identified, FOSL1, which encodes Fos-related antigen 1 (FRA-1), was a unique differentially expressed gene in all three PDAP1 knockdown cell lines compared to paired control cells (Figure 3C). The downregulation of FOSL1 by PDAP1 knockdown was confirmed by qPCR (Figure 3D) and western blotting (Figure 3E). The correlation between PDAP1 and FOSL1 was further validated using GEO datasets (Supplementary Figure S7A). We also found that overexpression of PDAP1 in CRC cells increased FOSL1 expression at both the mRNA (Supplementary Figure S7B) and protein (Supplementary Figure S7C) levels. Further analysis showed that PDAP1 mildly affected the turnover (Supplementary Figure S7D), proteasome-mediated degradation (Supplementary Figure S7E), and cellular localization (Figure 3F and Supplementary Figure S7F) of FRA-1. Furthermore, we analyzed PDAP1 and FRA-1 expression in serial sections of clinical CRC tissues (Figure 3G) and found that FRA-1 expression was significantly correlated with PDAP1 expression (ρ = 0.481, P = 0.00003, Supplementary Table S3). In addition, we confirmed the regulatory role of PDAP1 in FRA-1 expression in a panel of hepatocellular carcinoma cells and found that overexpression of PDAP1 led to increased FRA-1 expression in all the checked cells (Supplementary Figure S7G). Using TCGA data, we found a positive correlation between PDAP1 and FOSL1 in many cancers and normal tissues (Supplementary Table S4). These results indicated that PDAP1 promotes FRA-1 expression in CRC.

3.5 PDAP1 promotes FRA-1 expression regardless of its phosphorylation or sumoylation status

PDAP1 can be phosphorylated by casein kinase II [11], and numerous phosphoproteomic studies have identified several phosphorylation modification sites in multiple tissues [24]. We selected the eight most frequently phosphorylated sites and mutated these residues to alanine. As shown in Supplementary Figure S8A, these mutations did not significantly affect FRA-1 expression. Furthermore, PDAP1 was predicted to undergo sumoylation modification [25]. We mutated the potential sumoylation sites to alanine and found that these mutations had little effect on FRA-1 expression (Supplementary Figure S8B). These results indicate that PDAP1-induced FRA-1 expression was not dependent on its phosphorylation or sumoylation modifications.

3.6 PDAP1 promotes CRC progression via FRA-1

Because FRA-1 was regulated by PDAP1, we investigated whether FRA-1 mediated PDAP1-induced CRC progression. We found that overexpression of FRA-1 in PDAP1 knockdown cells (Figure 4A) reversed the effects of PDAP1 knockdown on colony formation (Figure 4B), cell invasion (Figure 4C), migration (Figure 4D), proliferation (Figure 4E), and cell cycle progression (Figure 4F). In vivo assays showed that the recovery expression of FRA-1 in PDAP1 knockdown cells restored the growth and metastatic abilities of PDAP1 knockdown cells (Figure 4G,H). These results indicated that PDAP1 facilitated migration, invasion, proliferation, cell cycle progression, and metastasis, mainly by stimulating FRA-1 expression.

3.7 PDAP1 interacts with EGFR and promotes EGFR signaling

To investigate the underlying mechanism by which PDAP1 induces FRA-1 expression, we analyzed the intracellular signaling using a phosphoprotein antibody array (Supplementary Figure S9A). Of 304 proteins involved in the 16 cellular pathways analyzed, the most significant effects of PDAP1 overexpression were to induce phosphorylation of MAPK/ERK kinase 1 (MEK 1) (Figure 5A) and regulate MAPK signaling (Figure 5B). These results were confirmed by western blotting against ERK1/2 phosphorylated isoforms (T202/Y204) compared to total ERK1/2 protein (Supplementary Figure S9B). Furthermore, we characterized the interacting partners of PDAP1 using a His pull-down assay and LC-MS/MS (Supplementary Table S5). Of the 38 potential interacting proteins identified, epidermal growth factor receptor (EGFR), which is closely linked to the Raf/MAPK/ERK1/2 signaling pathway, was selected for further analysis. We confirmed the interaction between PDAP1 and EGFR, using co-immunoprecipitation (Supplementary Figure S9C). Immunofluorescence staining showed that PDAP1 co-localized with EGFR along the cytosolic side of the cell membrane in both HCT-8 and human CRC tissues (Figure 5C and Supplementary Figure S9D). To determine which domain of EGFR was responsible for interacting with PDAP1, we truncated EGFR and fused a GFP tag to the carboxyl terminus (Figure 5D). As shown in Figure 5E, the intracellular juxtamembrane (JM) region of EGFR is indispensable for its interaction with PDAP1. Truncation analysis of PDAP1 based on its predicted structure (Supplementary Figure S9E) showed that the C-terminal domain of PDAP1 was responsible for its interaction with EGFR (Supplementary Figure S9F-G). We further analyzed the effect of PDAP1 on EGFR activation. We found that PDAP1 knockdown impaired the phosphorylation of EGFR at Tyr1173 and Tyr1086 and ERK1/2 at Thr202 and Tyr204 in EGF-treated CRC cells (Figure 5F and Supplementary Figure S9H). In contrary, overexpression of PDAP1 enhanced the phosphorylation of EGFR at Tyr1173 and Tyr1086 and ERK1/2 at Thr202 and Tyr204 in EGF-treated CRC cells (Figure 5G). These results suggested that PDAP1 interacts with the JM domain of EGFR and promoted EGFR/MAPK signaling.

3.8 PDAP1 promoted FRA-1 via EGFR/MAPK signaling

As previous results indicated transcriptional regulation of FRA-1 by PDAP1 (Figure 3D and Supplementary Figure S7B), we confirmed this regulation in HCT-8 cells by a luciferase reporter assay (Supplementary Figure S10A). Given that PDAP1 enhances EGFR-MAPK signaling (Figure 5) and Ras-MAPK signaling regulates FRA-1 in multiple contexts [26-28], we examined the involvement of EGFR-MAPK signaling in PDAP1-induced FRA-1 expression, by treating PDAP1-overexpressing cells with the phosphatidylinositol-3 kinase (PI3K) inhibitor ZSTK474 or MEK1 inhibitor GDC-0623. We found that PDAP1 overexpression enhanced the activation of protein kinase B (PKB) and ERK1/2; MEK1 inhibition, but not PI3K inhibition, eliminated the upregulation of FRA-1 induced by PDAP1 overexpression in both HCT-8 (Supplementary Figure S10B) and HCT116 (Supplementary Figure S10C) cells. Further analysis using the EGFR inhibitor PD153035 showed that EGFR inhibition eliminated the effects of PDAP1 overexpression on the activation of PKB and ERK1/2 and the upregulation of FRA-1 in a dose-dependent manner (Supplementary Figure S10D). These results suggested that PDAP1 induced FRA-1 expression via EGFR/MAPK signaling.

3.9 c-Myc directly transactivates PDAP1 in CRC

To investigate the mechanism underlying the upregulation of PDAP1 in CRC cells, we first analyzed copy number alterations (CNA) in CRC using TCGA data. Analysis of the PDAP1 CNA distribution showed that amplification represented only a small proportion of the samples (Supplementary Figure S11A). Furthermore, when we quantified the PDAP1 gene copy number in DNA from 10 primary CRC tissues and paired adjacent mucosa we observed no significant difference (Supplementary Figure S11B-C). These results indicated that PDAP1 overexpression is not driven by PDAP1 CNA. Next, we constructed a series of 5’-truncated versions of the PDAP1 promoter and found that the promoter region between -254 bp and -1 bp was indispensable for basal transcription of PDAP1 in CRC cells (Supplementary Figure S12A and Figure 6A). Sequence analysis identified two c-Myc binding sites in the promoter between -254 bp and -1 bp (Figure 6B). ChIP-qPCR confirmed the direct binding of c-Myc and PDAP1 promoter (Figure 6C). Further analysis showed that mutation of the c-Myc binding sites, especially the first binding site, decreased reporter activity (Figure 6D), indicating that PDAP1 transcription is primarily regulated by c-Myc. These results were further confirmed by the overexpression of stable c-Myc (c-Myc^T58A) and shRNA-mediated knockdown of c-Myc in CRC cells, which resulted in increased and decreased PDAP1 expression, respectively (Figure 6E,F and Supplementary Figure S12B-C). In addition, inhibiting c-Myc with 10058-F4 blocked PDAP1 expression in a dose-dependent manner (Figure 6G-H and Supplementary Figure S12D). Gene expression analysis using TCGA data showed that PDAP1 expression was highly correlated with MYC expression in CRC (Supplementary Figure S12E). Consistently, the analysis of c-Myc and PDAP1 expression in 282 CRC tissues using immunohistochemistry (IHC) staining showed that they were significantly correlated (Figure 6I). These results indicated that c-Myc-mediated transactivation of PDAP1 contributes to its overexpression in CRC.

3.10 Depletion of PDAP1 reduced tumor growth in vivo

Having shown that PDAP1 promotes CRC progression, we then investigated the potential of targeting PDAP1 to prevent tumor growth in vivo. In a PDX model (Figure 7A), depletion of PDAP1 using in-vivo-optimized siRNAs targeting PDAP1 significantly reduced FRA-1 expression and tumor growth (Figure 7B-E). These results demonstrated that inhibition of PDAP1 inhibited FRA-1 expression and cancer development, highlighting its role as a potential therapeutic target.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

The study was approved by the Hospital Ethics Committee of Xijing Hospital (permit number: KY20213194-1). Written informed consent was obtained from all participants. All animal protocols were approved by the Animal Care and Welfare Committee of the Fourth Military Medical University (permit number: 20220861).

CONSENT FOR PUBLICATION

Not applicable.

AVAILABILITY OF DATA AND MATERIALS

The data that support the findings of this study are available from the corresponding author upon reasonable request.

COMPETING INTERESTS

The authors declare that they have no competing interests.

AUTHORS’ CONTRIBUTIONS

Hong-Yong Cui and Shi-Jie Wang conceived and designed the project. Wei Wei, Mei-Rui Qian, Ruo-Fei Tian and Xin Fu performed experiments and/or conducted data acquisition and analyses; Hong-Wei Li, Gang Nan, Ting Yang, Peng Lin, Jian-Hua Dou, Xi Chen, Yu-Meng Zhu, Bin Wang and Xiu-Xuan Sun contributed technical/reagents materials, analytic tools, and/or grant support; Hong-Yong Cui, Shi-Jie Wang, Jian-Li Jiang, Ling Li and Zhi-Nan Chen prepared, wrote, reviewed, and/or revised the manuscript.