The data of clinical profiles and gene expression matrix of each normal and tumor individuals in TCGA database and GTEx database were processed from UCSC XENA database (https://xenabrowser.net/datapages/). Thereinto, the gene expression of 33 cases of cancer was analyzed through integration of normal tissue data from the GTEx database and tumor tissue data from the TCGA database. Moreover, 18 tumors and paired adjacent non-cancerous tissues were taken from the TCGA database for gene expression analysis. Data on pan-cancer immune infiltrating cells scores were obtained from the timer database (https://cistrome.shinyapps.io/timer/) and data on the expression of each tumor cell line were downloaded from CCLE database (https://portals.broadinstitute.org/). A PHF14 mRNA expression plot was constructed in tissue and cells with the Human Protein Atlas (HPA) database (version: 20.1) (https://www.proteinatlas.org/).

An edgeR software was adopted to analyze the differences in PHF14 expression levels in normal tissues, tumor tissues and paired non-cancerous tissues. Wilcoxon rank sum test was applied to analyze PHF14 expression level in different normal tissues and different tumor cell lines. Box plots were plotted by R package ggplot. UCSC XENA database (https://xenabrowser.net/datapages/) contributed clinicopathological features and RNA-sequencing (RNA-seq) data of these 33 cancers, namely, adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma (CESC), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), lymphoid neoplasm diffuse large B cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma (GBM), brain lower grade glioma (LGG), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), acute myeloid leukemia (LAML), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), mesothelioma (MESO), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), pheochromocytoma and paraganglioma (PCPG), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), sarcoma (SARC), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), testicular germ cell tumors (TGCT), thyroid carcinoma (THCA), thymoma (THYM), uterine corpus endometrial carcinoma (UCEC), uterine carcinosarcoma (UCS) and uveal melanoma (UVM). Log2 conversion was performed on all expression data. Nevertheless, MESO and uveal UVM have no normal or highly normal tissue deficiency in this module.

PHF14 was entered into the “CPTAC (Clinical proteomic tumor analysis consortium) analysis” module with the permission of the UALCAN portal (http://ualcan.path.uab.edu/analysis-prot.html), and the total protein expression level of PHF14 between normal tissues and primary tumors of TCGA project were compared. There were 10 tumor datasets identified in this study (breast cancer, clear cell RCC, colon cancer, GBM, HNSC, LIHC, LUAD, ovarian cancer, PAAD and UCEC). Finally, PHF14 expression violin plots was determined by the ‘Pathological Stage Plot’ module of GEPIA2 for different pathological stages (stage I, II, III, IV and V) of different tumors in TCGA. The violin plots were converted into forlog2 [Transcripts per million (TPM) +1] expression data.

To further explore how gene expression of PHF14 affects all TCGA tumors’ survival and prognosis, the “Survival Analysis” module of the Gene Expression Profile Interactive Analysis version 2 (GEPIA2) (http://gepia2.cancer-pku.cn/) was applied to generate the overall survival (OS) and disease-free survival (DFS) and Kaplan–Meier (K-M) plots of PHF14 across all tumors from the TCGA database. Also, UCSC Xena Browser was adopted to analyze the progression-free survival (PFS) of the TCGA Pan-Cancer datasets (version: 2018–09–13) (https://xenabrowser.net/) (settings: cutoff-low: 50% cutoff-high value: 50%). Hypothesis test was performed using log-rank tests.

In this study, PHF14 genetic alterations in cancer were studied with cBioPortal (version: 3.6.20) (https://www.cbioportal.org/). We input “PHF14” in the Quick Selection section of the cBioPortal web to explore the genetic alteration characteristics of PHF14 in TCGA Pan Cancer Atlas Studies. Next, a mutation site plot of PHF14 was generated by using the “Mutations” module, including the alteration frequency results, CNA (Copy number alteration), mutation type, and structural variants. Finally, the differences in the overall survival as well as disease-free and progression-free survival among UCEC cancer patients with or without PHF14 genetic alteration were analyzed using the “comparison/survival” module. Survival data were visualized using Kaplan-Meier curves.

In this study, PHF14 expression in tumor tissue and adjacent normal tissue of different types of cancers was examined with the “Immune-Gene” module of Tumor Immune Estimation Resource version 2 (TIMER2) (http://timer.cistrome.org/) was adopted to examine PHF14 expression in tumor tissues of various type of cancers and adjacent normal tissues in the TCGA project and also, the immune cells of cancer-associated fibroblasts (CAFs) were included for analysis. Subsequently, the degree of tumor immune infiltration was assessed with part or all of the EPIC, MCPCOUNTER, and TIDE algorithms. After purity adjustment, the P-values were determined through Spearman rank correlation test. Data were visualized using heatmaps and scatterplots.

To analyze how PHF14 and immune checkpoint genes correlate with each other, data of 33 normal and tumor tissues in the TCGA database were obtained from the Genomic Data Commons (GDC) data portal. These immune checkpoint genes were extracted and their correlation with PHF14 expression was calculated through analysis of Spearman correlation analysis heat maps between multiple tumor immune checkpoint genes and PHF14.

The STRING tool (https://string-db.org/) was adopted to construct a Homo Sapiens PHF14 co-expression network. At the same time, meet the following parameter requirements: 1) Fill in “50” in the maximum number of interactors; 2) Fill in “low confidence (0.150)” in the minimum required interaction score; 3) Fill in “co-expression” in the active interaction sources; and 4) Fill in “evidence” in the meaning of network edge. Finally, a co-expression network of 50 genes co-expressed with PHF14 was yielded. Subsequently, the visualization of the PPI network was realized by Cytoscape software. Furthermore, a combination of 50 genes co-expressed with PHF14 was enriched for KEGG (Kyoto Encyclopedia of Genes and Genomes) and GO (Gene Ontology) pathway analysis with the ClusterProfiler package to gain further insight into the biology and molecular features of these genes.

Gene set enrichment analysis (GSEA) is commonly used as an analytical method in comparing the expression status of genes and predefined gene sets with a particular biological process, cellular component or molecular function, so as to explore whether they are statistically significant to a certain extent. The threshold for GSEA was set at |NES| > 1, p-value <0.05, FDR <0.25, and if pathways met the sub-conditions, it was assumed that they are significantly enriched.

The Human Protein Atlas (https://www.proteinatlas.org) is a website that contains immunohistochemistry-based expression data for near 20 highly common kinds of cancers (Asplund et al., 2012). The user can identify the tumor types of specific protein expression patterns, these proteins are differentially expressed in certain types of cancer. In this study, the protein expression of PHF14 in human normal and KIRC tissues, human normal and PAAD tissues, and human normal and LIHC tissues were directly compared using immunohistochemical images from this website.

The human kidney renal clear cell carcinoma cell lines Caki-2 and 786-O, human pancreatic duct epithelial cell line hTERT-HPNE, human pancreatic cancer cell lines ASPC-1 and PANC-1, and human liver hepatocellular carcinoma cell lines MHCC-97h and Huh-7 were preserved and cultured in Dulbecco’s m o d ified Eagle’s medium (DMEM)-high glucose (Gibco, United States) containing 10% Fetal Bovine Serum (FBS, Gibco, United States) and 1% penicillin/streptomycin. The both human renal tubular epithelial cell line HK-2 and human normal liver cell line LO2 were preserved and cultured in RPMI-1640 (Gibco, United States) containing 10% FBS and 1% penicillin/streptomycin. Before the following experiments, all cells were kept under standard adherent conditions of 37°C, 5% CO₂ and humidified atmosphere.

All cells (2 × 10^–6 cells/well) were inoculated into 6-well plates and cultured for 48 h. Total RNA was isolated using TRIzol reagent (Invitrogen Carlsbad, CA, United States). For RT-PCR, single-stranded cDNA was synthesized from 1 µg of total RNA using reverse transcriptase (TaKaRa Biotechnology, Otsu, Japan). In addition, primers were designed against the following human sequences: PHF14(forward:AGTGCTCGGAATGTGACCAG,reverse:CCATCCGTAGCCTGTCTGTT),GAPDH (forward:GGAAGCTTGTCATCAATGGAAATC,reverse:TGATGACCCTTTTGGCTCCC). Real-time PCR was performed using SYBR^® Premix ExTaq™ II (Tli RNaseH Plus) (TaKaRa Biotechnology) and results were detected using an ABI 7500 Sequencing Detection System (Applied Biosystems, Foster City, CA, United States). The thermal cycling conditions were 95°C for 30 s and then 40 cycles of 95°C for 5 s and 60°C for 34 s.

PHF14-specific siRNA were purchased from GenePharma (Shanghai, China). The corresponding target sequence of siRNA are shown as following: forward: CCA​GUA​ACA​CUA​ACG​GAA​ATT, reverse: UUU​CCG​UUA​GUG​UUA​CUG​GTT. PHF14-specific siRNA was represented by siPHF14 and negative control group by siNC. Lipofectamine 2000 reagent (Cat. No. 11668019, Invitrogen) was used to deliver siRNA and siNC into the human liver hepatocellular carcinoma cell lines MHCC-97h and human pancreatic cancer cell lines PANC-1. The final concentration of 20 μM siRNA and 1 mg/mL Lipofectamine 2000 reagent were diluted with OptiMEM (Gibco) before transfection. The culture medium was changed into DMEM containing 10% FBS and 1% penicillin/streptomycin after 4–6 h in carbon dioxide incubator at 37°C. Two days or 3 days after transfection, the expression level of PHF14 was detected by qRT-PCR to assess the validation of the knockdown. According to the manufacturer’s instructions, the cell viability was assessed by measuring 3-(4,5-dimethylthiazol-2-yl)-2,5-di-phenyltetrazolium bromide (Nacalai Tesque, Kyoto, Japan) dye absorbance (MTT assay) at 1,2, 3, and 4 days after siRNA transfection.

EdU Apollo 567 Cell Tracking Kit (Rib-bio, Guangzhou, China, Cat# C10310-1) was used to evaluate the proliferation of MHCC-97h and PANC-1 cells. PHF14 knock-down cells and negative control cells (1 × 10⁴/well) were inoculated in 96-well plates and incubated overnight at 37°C. Then 5-ethynyl-20-deoxyuridine (EdU, 200 μM) was added and incubated for 2 h at 37°C. Cells were cold fixed with 4% paraformaldehyde for 20 min, washed three times with PBS, and then treated with 0.5% Triton X-100 at room temperature for 10 min. Then, after washing with PBS for 3 times, each well was incubated with 100 μL Apollo reagent for 30 min. Finally, the nuclei were stained with Hoechest 33,342 for 5 min. The percentage of EdU-positive cells was calculated based on counts from 500 cells in three independent experiments.

The invasion potential of PANC-1 and MHCC-97h cells was evaluated by Transwell-Matrigel system. The culture upper inserts were coated with Matrigel (BD Matrigel, United States, Cat# 356234). The cells were resuspended in serum-free DMEM medium, and the cell density was adjusted to 1-10*10⁵/mL. 200μL cell suspension was added to each transwell upper chamber (24 wells, 8 mm pore size; BD Biosciences, United States, Cat# 3428), and DMEM medium containing 10%FBS was added to each lower chamber. After incubation for 48 h, the cells and Matrigel in the upper chamber were removed with a cotton swab and the cells adhering to the lower membrane of the inserts were fixed in ice-cold methanol at 4°C and stained with 1% crystal violet. Quantification of cell invasion was expressed as the mean count of stained cells in 5 random fields of each membrane under light microscope (×20objective lens). The migration ability of PANC-1 and MHCC-97h cells was analyzed by cell scratch assay. The cells were cultured in serum-free DMEM medium and performed scratch assay with pipette tip. The cell migration was compared and observed after 6, 24, and 48 h. All the experiments were performed in triplicates.

An annexin V-fluorescein isothiocyanate (FITC) apoptosis detection kit (BD Biosciences, San Jose, CA, United States) was used to measure apoptotic cells. PANC-1 and MHCC-97h cells were treated with siNC(Negative control-lentivirus) and siPHF14 (PHF14-siRNA-lentivirus) for 48 h, then collected and resuspended with 500 μL 1X binding buffer. Then, 5 μL of annexin V-FITC and 10 μL of propidium iodide (PI) were added. After gentle mixing, incubate at room temperature and away from light for 5 min. Finally, the stained cells were analyzed using a flow cytometer (BD Biosciences, San Jose, CA, United States).

Using HPA, GTEx, and FANTOM5 (Functional Annotation of Mammalian Genomes) datasets, PHF14 was shown to be enriched in thymus and ovary, and highly expressed in cerebral cortex and cerebellum among other cancers (Figure 1A; Supplementary Figure S1). In addition, PHF14 was also highly expressed in neuronal cells and glial cells based on single-cell RNA-seq (Figure 1B). According to our data, we found a lower PHF14 expression in terms of its tissue-specific and cell type specificity.

From an unpaired perspective (Figure 1C), significant increases in PHF14 expression levels were observed in tumor specimen tissues of adrenocortical carcinoma (ACC) (p < 0.01), cervical squamous cell carcinoma (CESC) (p < 0.05), bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), lymphoid neoplasm diffuse large B cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma (GBM), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), acute myeloid leukemia (LAML), brain lower grade glioma (LGG), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), testicular germ cell tumors (TGCT), thymoma (THYM), uterine corpus endometrial carcinoma (UCEC) and uterine carcinosarcoma (UCS) as compared to adjacent normal tissues (p < 0.001). However, PHF14 expression levels was lower in ovarian serous cystadenocarcinoma (OV) and thyroid carcinoma (THCA) tumor tissues than in adjacent normal tissues (p < 0.001). Moreover, PHF14 expression levels in kidney chromophobe (KICH) and pheochromocytoma and paraganglioma (PCPG) tumor tissues were also not distinctly different from those of the adjacent normal tissues (ns: p > 0.05). From a pairing point of view (Figure 1D), compared with the adjacent normal tissues, PHF14 expression levels were elevated markedly in the tumor tissues of bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PRAD), stomach adenocarcinoma (STAD) (p < 0.001), cholangiocarcinoma (CHOL), kidney renal papillary cell carcinoma (KIRP), thyroid carcinoma (THCA) (p < 0.01) and rectum adenocarcinoma (READ) (p < 0.05). However, higher levels of PHF14 expression were observed in adjacent normal tissues than that in esophageal carcinoma (ESCA) and kidney chromophobe (KICH) tumor tissues (p < 0.01). It was also observed that PHF14 expression levels was not statistically significant between the adjacent normal tissues and the pancreatic adenocarcinoma (PAAD) and uterine corpus endometrial carcinoma (UCEC) tumor tissues.

To clarify PH14 protein expression levels in various tumors, CPTAC dataset protein expression was analyzed. In Figure 1E, compared with adjacent normal tissues, the total protein expression level of PHF14 was higher in primary tumor tissues of breast cancer, colon cancer, kidney renal clear cell carcinoma (clear cell RCC), lung adenocarcinoma (LUAD), uterine corpus endometrial carcinoma (UCEC), glioblastoma (GBM), head and neck squamous cell carcinoma (HNSC), liver hepatocellular carcinoma (LIHC), pancreatic adenocarcinoma (PAAD) (p < 0.001) and ovarian cancer (p < 0.05). In addition, the relationship between PHF14 expression level and different tumor pathological stages was investigated using the GEPIA2. In Figures 2A–F, PHF14 expression levels in BRCA, COAD, KICH, SKCM, LIHC and STAD tumors at different pathological stages were significantly different (p < 0.05).

The cancer cases included in this study were sorted into a low PHF14 expression group and a high PHF14 expression group based on PHF14 expression. The survival prognosis between the two groups in different cancers was assessed and compared using the TCGA and GEO datasets. As can be seen from Figure 3A, high PHF14 expression was relevant to poor OS (overall survival) prognosis of patients with ACC (p = 0.0073), COAD (p = 0.044), LGG (p = 0.00015), SARC (p = 0.047) and UVM(p = 0.038).

Likewise, in Figure 3B, high PHF14 expression was correlated to inferior prognosis of patients with ACC (p = 0.004), COAD (p = 0.0091), KIRC (p = 0.029), LGG (p = 0.021), LIHC (p = 0.023), LUAD (p = 0.047) and MESO (p = 0.027) from DFS (disease-free survival) analysis. However, low PHF14 gene expression was relevant to poor OS prognosis in TGCT patients (p = 0.035). Moreover, as indicated in Figures 3C–F, high PHF14 expression was relevant to poor progression-free survival in patients with KIRC (p = 0.0209), LGG (p = 0.0301), LUAD (p = 0.0157), and PAAD (p = 0.0422). From the above results, it can be observed that PHF14 expression levels are relevant to the prognosis of patients with pan-cancer, and variations can be observed in accordance with the type of cancer.

In this study, cBioPortal was adopted to investigate the genetic changes of PHF14 in various types of tumors in the TCGA dataset. It was demonstrated in Figure 4A that the highest alteration frequency (>10%) of the PHF14 gene appeared in UCEC tumor samples, predominantly expressed in the “mutation” type (>8%). In addition, the “mutation” type was the predominant type of genetic alteration in SKCM patients, accounting for approximately 4%. Notably, all types of PHF14 gene alterations (∼2% frequency) in LIHC patients were “mutation.” Moreover, all alterations of PHF14 gene in UCS, TGCT, THYM, KIRP, PCPG and PAAD patients were “amplification,” with this being the predominant type of genetic alteration in all TCGA tumor samples.

In Figure 4B, we can further observe the type, location and number of PHF14 genetic alterations. In addition, we also detected alterations in the K182Nfs*19/K182Mfs*20/P180L in two cases of STAD, seven cases of UCEC and one case of OV, and the major type of PHF14 gene mutation was missense mutation. Subsequently, how PHF14 gene mutation reacted with survival prognosis of patients with various types of tumors was examined in this study. As shown in Figures 4C–H, in BRCA and CESC cancer patients, the PHF14-unaltered group exhibited a favourable prognosis for DSS (disease-specific survival) (P-values were 1.55e-9 and 1.143e-3, respectively), OS (P-values were 6.087e-6 and 5.968e-5, respectively) and PFS (P-values were 1.191e-6 and 0.0141, respectively) as compared to the PHF14-altered group. The above results indicate that PHF14 expression in pan-cancer is relevant to PHF14 amplification and mutation, further suggesting that the genetic alterations of PFH14 are highly relevant to survival prognosis of different cancer patients.

Figure 5 presents how infiltration of cancer-associated fibroblasts (CAFs) interacts with PHF14 gene expression in different types of cancers in TCGA. A positive correlation between PHF14 expression and the level of infiltration of cancer-associated fibroblasts in CESC, COAD, HNSC, KIRC, LUAD, PAAD, READ and SKCM can be observed using the EPIC, MCPCOUNTER and TIDE algorithms. Additionally, TIDE algorithm was used to generate the above-mentioned tumor scatter plot data, as indicated in Figure 5B. The findings reveal that cancer-associated fibroblast immune infiltration is closely correlated with cancer occurrence, development or metastasis.

To explore the relationship between PHF14 expression and immune checkpoint gene expression in this study, an analysis of the expression data of 47 immune checkpoint genes commonly found in various tumors was performed and the findings are presented in Figure 6. As revealed in Figure 6A, PHF14 expression is in positive correlation with the expression levels of almost all immune checkpoint genes in KICH, KIRC, UVM and other tumors. The above results indicate that in some tumors, the regulation of immune checkpoint gene expression levels by PHF14 may affect tumor immunity.

As demonstrated in Figure 6B, we can find that the 5 common MMRs genes (EPCAM, MLH1, MSH2, MSH6 and PMS2) exhibited highly positive association with PHF14 expression in various types of TCGA tumors, suggesting that PHF14 could maintain the viability of tumor cells by up-regulating genes related to DNA mismatch repair. In addition, DNA methylation was produced as a result of DNA methyltransferases, which covalently binds a methyl group at the 5’ carbon position of cytosine, a CpG dinucleotide in the genome. As shown in Figure 6C, the heatmap data obtained from the TIMER2 online tool revealed that PHF14 and methyltransferase expression levels positively correlated with most TCGA tumor types, suggesting that PHF14 can mediate tumorigenesis and progression through regulating human pan-cancer epigenetic status.

In this study, PHF14 co-expressed genes were analyzed for pathway enrichment to explore the mechanism of PHF14 gene in tumor. An experimentally validated set of 50 available msh6 binding proteins was generated by using the STRING tool. The interaction network of these PHF14-binding proteins is clearly presented in Figure 7A. Furthermore, in Figure 7B, we combined 50 PHF14 co-expressed genes for CC, BP, MF and KEGG enrichment analysis. Enrichment analysis of KEGG pathway revealed that “mRNA surveillance pathway” may be involved in the pathogenesis of tumor PHF14. The results of MF and CC showed that “histone binding” and “nuclear chromatin” may be involved in the tumorigenesis and development mechanism of PHF14. Finally, the enrichment results of BP pathway revealed that, “RNA splicing,” “histone modification” and “chromatin remodeling” may all be involved in the pathogenesis of tumor PHF14. Furthermore, based on PHF14 expression levels, human pan-cancer samples were divided into high and low expression groups, and the enrichment of signaling pathways in KEGG and the markers of the two groups were analyzed by GSEA. As shown in Figures 7C–E, we have listed the top 3 most significantly enriched signaling pathways in the KEGG database, namely, KEGG_WNT_SIGNALING_PATHWAY (NES = 1.763; P. adjust = 0.043; FDR = 0.034), REACTOME_GPCR_LIGAND_BINDING (NES = 1.454; P. adjust = 0.043; FDR = 0.034), and WP_OSTEOBLAST_DIFFERENTIATION (NES = 1.944; P. adjust = 0.043; FDR = 0.034), which suggested that PHF14 may be involved in signaling pathways regulating tumor metabolism and immunity.

We tried to explore the protein expression patterns of PHF14 in KIRC, PAAD, and LIHC by the Human Protein Atlas. As shown in Figures 8A–C, PHF14 protein were highly expressed in tumor tissues (KIRC, PAAD, and LIHC) compared to normal tissues. After finding the difference in the expression of PHF14 protein in normal and tumor tissues, we cultured tumor cell lines and corresponding normal cell lines in vitro. As shown in Figures 8D–F, PCR results showed that the mRNA expression level of PHF14 in human kidney renal clear cell carcinoma cell lines (Caki-2 and 786-O) was higher than that in human renal tubular epithelial cell line (HK-2). Similarly, The mRNA expression of PHF14 in human pancreatic adenocarcinoma cell lines (ASPC-1 and PANC-1) and human liver hepatocellular carcinoma cell lines (MHCC-97h and Huh-7) was higher than that of human pancreatic duct epithelial cell line (hTERT-HPNE) and human normal liver cell line (LO2), respectively. Taken together, our results showed that transcriptional and proteinic expressions of PHF14 were over-expressed in patients with KIRC, PAAD and LIHC.

To determine whether defective expression of PHF14 had a functional role in PANC-1 and MHCC-97h cells, lentivirus siRNA was used to knockout PHF14 expression in cells (Figure 9A). Two or three days after transfection, the PHF14-knockdown led to a downregulation of cell growth as determined via the MTT assay (Figure 9B). Subsequently, EdU assay showed that the growth of PANC-1 and MHCC-97h cells decreased after PHF14 gene silencing (Figures 9C–E).

The results of the transwell assay revealed that invasion potential of PANC-1-siPHF14 and MHCC-97h-siPHF14 cells were decreased relative to control cells (siNC) in transwell based experiment (Figures 9F, G). 48 h after transfection, the flow cytometry showed that the percentage of apoptotic cells was higher (25.63% versus 11.59%, and 25.74% versus 9.79%) than the control cells in PANC-1 and MHCC-97h short interfering PHF14 cells (Figure 9H). Wound closure scratch was used to assess the effect of PHF14 on the migratory capacity of PANC-1 and MHCC-97h cells. In contrast to control cells, cells with PHF14-knockdown had a significantly lower percentage of gap closures in the scratch assay (Figure 9I).