The prognostic significance of bromodomain PHD-finger transcription factor in colorectal carcinoma and association with vimentin and E-cadherin

Abstract

Purpose

Bromodomain PHD-finger transcription factor (BPTF) is a chromatin-mediated regulation of transcription factor, playing an important role in embryogenesis and differentiation. Epithelial-mesenchymal transition (EMT) has a pivotal role in colorectal cancer (CRC) progression, sharing the similar characteristic with BPTF. Therefore, the aim of this study was to examine the expression and clinical value of BPTF and the correlation with EMT markers in patients with CRC.

Methods

Real-time PCR and Western blot were performed to evaluate the mRNA and protein expression levels of BPTF in 20 pairs of fresh-frozen CRC and non-tumor adjacent tissues (NATs). The expressions of BPTF, vimentin and E-cadherin were examined by immunohistochemical staining in 105 cases of paraffin-embedded primary CRC specimens. In addition, the clinicopathological significance and the prognostic value of BPTF, vimentin and E-cadherin expression were further determined. Then, the correlation of BPTF with vimentin and E-cadherin was also explored.

Results

BPTF mRNA and protein expression were significantly overexpressed in CRC tissues when compared with paired NATs (P < 0.01). The expression levels of BPTF and vimentin in CRC paraffin-embedded specimens were significantly higher than the expression in NATs (P < 0.01), while the expressions of E-cadherin in tumors were obviously lower than in NATs (P < 0.01). Tumors with high expression of BPTF and vimentin, as well as low expression of E-cadherin, were significantly correlated with various adverse clinicopathological factors (P < 0.05). The CRC patients with a high BPTF or vimentin expression had shorter overall survival than those with lower expression (P < 0.05). Furthermore, univariate analysis and multivariate analysis showed that high BPTF expression was an independent prognostic factor for patients with CRC. The last and more interesting Spearman rank correlation analysis and microscopic observation found that the expression of BPTF obviously correlated with the expression of EMT markers vimentin and E-cadherin.

Conclusion

Our results strongly suggested that the high BPTF expression was significantly correlated with tumor progression and may be a potent prognostic marker of CRC. Moreover, BPTF expression was significantly associated with EMT markers vimentin and E-cadherin.

Electronic supplementary material

The online version of this article (doi:10.1007/s00432-015-1937-y) contains supplementary material, which is available to authorized users.

Introduction

Colorectal cancer (CRC) is the third most common malignancies worldwide and ranked as the fourth of global cancer-related mortality (Brenner et al. 2014). Presently, the overall incidence of CRC is declined in the developed countries, but, unfortunately, it remains very high in the East Asia (Jemal et al. 2011). Metastasis results in 80 % cancer-related mortality in CRC patients. Thus, it is of great importance to dig out the mechanism of invasion and metastasis in CRC, and develop novel-targeted therapies.

Epithelial–mesenchymal transition (EMT), first defined by Elizabeth Hay in 1968, is a biological process by which epithelial cells can downregulate epithelial characteristics and acquire mesenchymal characteristics (Kalluri and Weinberg 2009). During the process, epithelial cells lose their tight adhesion junctions and apical–basal polarity, reorganize their cytoskeleton, undergo cell shape change from cobblestone to spindle-like morphology and increase the motility and potential of migratory, invasion and metastasis (Thiery et al. 2009). E-cadherin is a classical marker of epithelial cell, which retains the cell–cell contacts. The decreased expression of E-cadherin and increased expression of vimentin are referred as the landmark of EMT (Acloque et al. 2009; Lim and Thiery 2012). Now, it is widely accepted that EMT is involved in the development of embryo and organ, wound tissue repair, tumor invasiveness and metastasis. EMT further plays an important role in many cancers, such as breast cancer, lung cancer, pancreatic cancer, skin squamous cell carcinoma and CRC. It is showed that the EMT process is correlated with cancer progression and prognosis at the invasive frontier in CRC (Bastid 2012). The regulatory mechanisms of EMT include the transcriptional and translational machinery, expression of noncoding RNAs, alternative splicing and protein stability. Thereinto, the transcriptional factors, such as SNAIL, TWIST and zinc-finger E-box-binding (ZEB) 1/2 and so on, have been well exploited and display most important functions in modulating EMT (Lamouille et al. 2014). Therefore, a better understanding of the underneath regulatory mechanism of transcriptional factor is important to unveil the mystery of EMT.

Bromodomain PHD-finger transcription factor (BPTF), the largest subunit of nucleosome remodeling factor (NURF), locates at chromosome 17q24.3 and takes part in transcriptional regulation and chromatin remodeling (Wysocka et al. 2006). The completed sequence of human BPTF encodes a predicted protein of 2781 amino acids, which contains typical features of bromodomain, two PHD fingers and an extensive glutamine-rich acidic domain (Jones et al. 2000). Several studies have identified that BPTF is involved in the biological processes of development and morphogenesis and differentiation (Landry et al. 2008; Goller et al. 2008; Mulder et al. 2012). BPTF mutation leads to deficient growth and differentiation, exhibits morphological defects in the gastrula stage and causes embryonic lethality eventually. Histological analysis showed that BPTF mutants had apparent defects in anterior–posterior (AP) axis, and distal visceral endoderm (DVE) cells migrated toward the prospective anterior to form the anterior visceral endoderm (AVE), suggesting the essential role of BPTF in embryonic cell migration, which is similar to the function of EMT process in embryonic development. In addition, BPTF plays a regulatory role in multiple genes, e.g., SMAD, TGF-β and BMP2 at the transcriptional level, which have a vital function in EMT and tumor (Landry et al. 2008). Meanwhile, chromosome arm 17q aberrancy was further found in various cancers, e.g., lung, prostate cancer and neuroblastoma. Therefore, until recently, the characteristics of BPTF gene and its function in EMT are considered to be linked to tumor progression based on the above findings. A study discovered that BPTF promoted the malignant transformation of human embryonic lung fibroblasts and bladder cancer cells growth (Buganim et al. 2010; Kim et al. 2013). According to the previous studies, BPTF may play an important role in promoting the development and progression of tumor through EMT process. However, there are very few researches focus on the role of BPTF in CRC. Therefore, the purpose of this study is to explore the significance of BPTF in CRC and thus develop the novel-targeted therapies for CRC patients.

In this study, we discovered that abnormal high expression of BPTF in CRC tissues correlated with the negative clinicopathological features and poor prognosis, and had a significant relation with EMT markers vimentin and E-cadherin, which reveal a novel role of BPTF maybe a constitutive oncogene of the EMT-inducing transcription factors in CRC.

Materials and methods

Patients and tissue specimens

This study was approved by the Research Ethics Committee of Affiliated Nanhua Hospital of University of South China, China. Written informed consent was obtained from all patients. All specimens were handled and made anonymous according to the ethical and legal standards.

A total of 105 primary CRC specimens were obtained from patients who underwent CRC radical resection at the Department of General Surgery, Affiliated Nanhua Hospital of University of South China (Hunan, PR China) from January 2005 to December 2009. None of the patients had received chemotherapy or radiotherapy before surgery. These patients included 65 (61.9 %) males and 40 (38.1 %) females with median age of 52 years (range 19–85). Among these patients, 20 matched fresh CRC specimens and non-tumor adjacent tissues (NATs) were selectively employed for real-time quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and Western blot analysis. The diagnosis for each patient was confirmed by histopathology. The pathologic tumor–node–metastasis (TNM) status of all CRC was assessed according to the criteria of the seventh edition of American Joint Committee on Cancer/International Union Against Cancer TNM classification system.

Reagents and antibodies

The primary antibodies included rabbit polyclonal antihuman BPTF (sc-98404, Santa Cruz Biotechnology), mouse monoclonal antihuman vimentin (sc-66001, Santa Cruz Biotechnology) and mouse monoclonal antihuman E-cadherin (sc-8426, Santa Cruz Biotechnology) antibodies used for Western blot and immunohistochemistry detection, and mouse monoclonal antihuman β-actin antibody (sc-47778, Santa Cruz) and homologous secondary antibody (Zhongshan Golden Bridge Biotechnology) used for Western blot analysis. The PV-9000 Polymer Detection System, DAB and hematoxylin (Zhongshan Golden Bridge Biotechnology) were used for immunohistochemical staining according to the manufacturer’s recommendations.

Real-time PCR

Total RNA was extracted from frozen tumor specimens using Trizol reagent (Invitrogen) according to the instructions. Real-time PCR was performed using the SYBR Green Real-time PCR Master Mix (Toyobo) as described. The primers of BPTF were as follows: forward, 5′- TTCTCCCACAGACAGAGGTG-3′ and reverse, 5′-ACGTGATGACAATGGACTGG-3′; β-actin was used as a control using the following primers: forward, 5′-ACTCGTCATACTCCTGCT-3′ and reverse, 5′-GAAACTACCTTCAACTCC-3′. The results were analyzed using the 2^−ΔΔCt method with the formula: ΔΔCt-(CtCRC-Ctβ-actin)-(CtNAT-Ctβ-actin).

Western blot analysis

Total proteins were extracted and separated by 10 % sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto polyvinylidene fluoride membranes (Millipore). The blotted membranes were incubated antihuman BPTF antibody (1:1000 dilution), antihuman E-cadherin antibody (1:2000 dilution), antihuman vimentin antibody (1:2000 dilution) and then an appropriate secondary antibody (1:2000 dilution). β-actin protein determined by its antibody (1:5000 dilution) was used as a loading control. The Western blot band density was measured by ImageJ software, which should repeat three times (Carvajal-Vergara et al. 2010).

Immunohistochemistry

The tissue specimens were fixed in 10 % formalin and embedded in paraffin; 4-mm sections were cut and placed on silane-coated slides for immunohistochemical studies. Part of the specimens was stained with H&E and microscopically examined to confirm the diagnosis. The paraffin sections were dewaxed and pretreated in 0.01 M sodium citrate buffer (pH 6.0) for 15 min at 95 °C for tissue antigen retrieval. These sections were then incubated with 3 % hydrogen peroxide for 20 min at room temperature to block endogenous peroxidase. After rinse three times for 2 min with PBS, 50 µl of 10 % goat serum was added as blocking liquid, and incubation was conducted for 15 min under room temperature. Then, the serum was removed, and 50 µl per slide of diluted primary antibody (BPTF, 1:200 dilution; E-cadherin, 1:300 dilution; vimentin, 1:200 dilution) was added to each section and incubated overnight at 4°. The sections were developed according to manufacturer’s recommendations (PV-9000, Zhongshan Golden Bridge Biotechnology) and counterstained with hematoxylin, and ethanol dehydration was conducted by grade followed by addition of xylene to make the sections transparent. Finally, mounting was performed with neutral balsam.

The staining intensity of BPTF, vimentin and E-cadherin was evaluated in a 4-step scale (0, no staining;1+, weak intensity; 2+, moderate intensity; and 3+, strongest intensity). The fraction of stained cells was scored according to the following criteria: score 0 (0–10 % positive cancer cells), score 1 (11–50 % positive cancer cells), score 2 (51–80 % positive cancer cells) and score 3 (>80 % positive cancer cells). The results of staining intensity and extent gave an overall staining score (Fan et al. 2013). Immunostaining was evaluated by two pathologists simultaneously on a multi-headed microscope without knowledge of the patients’ clinicopathological features and the clinical follow-up data.

Patients follow-up and prognostic study

The follow-up period was from December 2009 to 31 December, 2013. The patients were evaluated at 3-month interval in the first year and twice-a-year in the next years by means of home visits, telephone calls and outpatient consultations to determine their status after surgery and ascertain whether patients had tumor recurrence and metastasis, required resurgery and/or chemotherapy, or had died. Only deaths from the disease being studied were counted. Patients who died from some other cause were not included in the calculation. Serum CEA, CT or MRI, and electronic colonoscopic examination were performed to find the possibility of relapse. Finally, we obtained data of 105 patients with valid follow-up results (excluding the patients who died within 4 weeks after surgery); 47 patients were dead, and 37 patients had recurrence. Follow-up ranged from 1 to 81 months, with an average follow-up period of 43.2 weeks.

Statistical analysis

All data were analyzed using the statistical software SPSS, version 17.0 for Windows (SPSS, Chicago, III). The Fisher’s exact test was used for statistical analysis of categorical data, whereas independent t tests were used for continuous data. Survival curves were constructed using the Kaplan–Meier method and evaluated using the log-rank test. The Cox proportional hazards regression model was established to identify factors which were independently associated with the overall survival of CRC patients. P values <0.05 were considered statistically significant.

Results

BPTF expression was significantly elevated in human fresh-frozen CRC tissues

The expression levels of BPTF mRNA and protein in the tumor and paired NAT samples from 20 patients examined by real-time PCR and Western blot analysis were shown in Figs. 1 and 2. Most tumor tissues from CRC patients (17/20; 85.0 %) showed higher levels of BPTF mRNA compared with the corresponding NATs (mean ± SD, 2.73 ± 0.29 vs. 0.89 ± 0.17; P < 0.001) (Fig. 1). Consistent with mRNA expression, the Western blot results showed that the expression of BPTF protein in CRC tissues was also significantly higher than that in the corresponding NATs (P < 0.01) (Fig. 2).

Fig. 1.

High expression of BPTF mRNA in CRC tissues. a Differential expression levels of BPTF mRNA in 20 paired CRC samples of tumoral and corresponding NATs. b The expression levels of BPTF mRNA in CRC tissues were significantly higher than those in NATs (P < 0.001). NAT non-tumor adjacent tissues

Fig. 2.

High expression of BPTF protein in CRC tissues. a Representative figures of BPTF protein expression in 20 paired CRC samples of tumoral and corresponding NATs. b The expression levels of BPTF protein in CRC tissues were significantly higher than those in NATs (P < 0.001)

Aberrant expression of BPTF, E-cadherin, vimentin and in CRC

One hundred and five paraffin-embedded samples from CRC patients were used to determine the expression of BPTF, E-cadherin and vimentin by immunohistochemistry. The protein expression pattern was categorized into two groups: high expression (IHC level 2–3) versus low expression (IHC level 0–1). Of the 105 samples, 71 samples (67.6 %) exhibited high expression level of BPTF staining (P < 0.001) and 59 samples (56.2 %) exhibited high expression level of vimentin staining (P < 0.001; Table 1). In contrast, E-cadherin showed low expression in tumor tissues, and 78 samples (74.3 %) exhibited low expression level of membrane and cytoplasm staining in 105 tumors (P < 0.001; Table 1). Representative samples of the three protein expressions in the tumor tissues are shown in Fig. 3. BPTF was identified positive in nucleus and cytoplasm of cells (Fig. 3a–d); E-cadherin was found negative in tumor cells (Fig. 4a, c); and vimentin was found major positive in cytoplasm of cells (Fig. 4b, d). More interestingly, we found BPTF-positive expression in the poor-differentiated CRC was more intense than the well-differentiated CRC (Fig. 4; Table 2).

Table 1.

Immunohistochemistry analysis of BPTF, vimentin and E-cadherin expression in CRC tissues and corresponding NATs

	BPTF expression		P value
	High (n, %)	Low (n, %)
BPTF
Tumor	71 (67.6)	34 (32.4)
NAT	4 (3.8)	101 (96.2)	<0.01
Vimentin
Tumor	59 (56.2)	46 (43.8)
NAT	3 (2.9)	102 (97.1)	<0.01
E-cadherin
Tumor	27 (25.7)	78 (74.3)
NAT	98 (93.3)	7 (6.7)	<0.01

BPTF bromodomain PHD-finger transcription factor, CRC colorectal carcinoma, NAT non-tumor adjacent tissues

Significant results (P < 0.05) are given in bold

Fig. 3.

Representative photomicrographs of BPTF, E-cadherin and vimentin expression in CRC a, d BPTF-positive expression in CRC is shown. b, e E-cadherin-negative expression in CRC is shown. c, f Vimentin-positive expression in CRC is shown. Magnification 100×(a–c), 400×(d–f)

Fig. 4.

Representative photomicrographs of BPTF expression in CRC. a, c BPTF-positive expression in well-differentiated CRC is shown. Magnification 100×(a), 400×(c). b, d BPTF-positive expression in poor-differentiated CRC is shown. Magnification 100×(b), 400×(d)

Table 2.

Association between expression of BPTF, vimentin and E-cadherin and clinicopathological characteristics in CRC patients (n = 105)

Variables	BPTF (n)			Vimentin (n)			E-cadherin (n)
	High	Low	P value	High	Low	P value	High	Low	P value
Age (years)
≤60	43	21		37	27		14	50
>60	28	13	0.9060	22	19	0.6756	13	28	0.2608
Gender
Male	42	23		36	29		16	49
Female	29	11	0.4018	23	17	0.8320	11	29	0.7426
Tumor size (cm)
≤5	38	24		40	22		21	41
>5	33	10	0.0961	19	24	0.0390	6	37	0.0217
Histology grade
Low (G1–G2)	42	28		34	36		24	46
High (G3–G4)	29	6	0.0183	25	10	0.0261	3	32	0.0092
UICC stage
I/II	32	29		32	29		23	38
III/IV	39	5	0.0001	27	17	0.3642	4	40	0.0020
Nodal metastasis
Negative	36	25		28	33		15	46
Positive	35	9	0.0265	31	13	0.0124	12	32	0.7563
Recurrence
Negative	40	28		38	30		22	46
Positive	31	6	0.0090	21	16	0.9313	5	32	0.0349

UICC Union for International Cancer Control

Significant results (P < 0.05) are given in bold

These results indicate that BPTF and vimentin are upregulated, and E-cadherin is downregulated specifically in CRC tissues.

Association of BPTF, vimentin and E-cadherin expression with clinicopathological characteristics of CRC

Subsequently, we examined whether aberrant expression of BPTF, vimentin and E-cadherin is associated with clinicopathological characteristics in our CRC cohort. BPTF, vimentin and E-cadherin expressions were not correlated with the variables of gender and age (P > 0.05; Table 3). High BPTF expression was significantly correlated with the histology grade, UICC stage, nodal metastasis and recurrence of CRC (P < 0.05). High expression of vimentin was significantly associated with tumor size, histology grade and nodal metastasis of CRC (P < 0.05). Low expression of E-cadherin was significantly associated with tumor size, histology grade, UICC stage and recurrence of CRC (P < 0.05). These results showed that BPTF, vimentin and E-cadherin expression maybe correlated with the outcome of CRC patients.

Table 3.

Cox regression analyses of overall survival (OS) and BPTF, vimentin, E-cadherin expression levels as well as clinicopathological factors in patients with CRC

Variables	Cases (n)	Univariate analysis		Multivariate analysis
		HR (95 % CI)	P	HR (95 % CI)	P
Age
≤60	64	1		1
>60	41	1.073 (0.537–1.405)	0.496	1.113 (0.733–1.379)	0.594
Gender
Male	65	1		1
Female	40	0.733 (0.314–1.216)	0.638	0.905 (0.633–1.592)	0.273
Tumor size (cm)
≤5	62	1		1
>5	43	1.529 (0.934–2.881)	0.093	1.414 (0.732–1.987)	0.154
Histology grade
Low (G1–G2)	70	1		1
High (G3–G4)	35	1.475 (0.588–2.243)	0.208	1.319 (0.772–1.846)	0.351
UICC stage
I/II	61	1		1
III/IV	44	2.933 (1.384–5.962)	0.001	2.285 (1.414–4.737)	0.008
Nodal metastasis
Negative	61	1		1
Positive	44	2.538 (1.227–4.913)	0.005	2.712 (1.334–5.216)	0.003
Recurrence
Negative	68	1		1
Positive	37	3.476 (1.932–6.062)	0.001	2.937 (0.971–5.136)	0.001
BPTF expression
Low	34	1		1
High	71	1.974 (0.812–3.285)	0.011	1.836 (0.692–3.735)	0.018
Vimentin expression
Low	46	1		1
High	59	2.038 (1.142–3.795)	0.010	1.573 (0.792–2.133)	0.084
E-cadherin expression
High	27	1		1
Low	78	1.916 1.155–4.038)	0.023	1.633 (0.583–2.977)	0.059

CI confidence interval, HR hazard ratio

Significant results (P < 0.05) are given in bold

High BPTF expression and vimentin expression are associated with a poor survival outcome of CRC patients

We next analyzed 5-year overall survival (OS) rates using Kaplan–Meier estimates with log-rank tests. Survival curves indicated that high BPTF expression (P = 0.002) and high vimentin expression (P = 0.003) were significantly associated with a poor overall survival (Fig. 5a, c). However, no association was observed for E-cadherin expression (P = 0.073) with CRC patients’ overall survival (Fig. 5b). The results demonstrated that BPTF and vimentin expression could be prognostic markers of CRC patients.

Fig. 5.

Postoperative 5-year survival curves of patients with CRC. a The CRC patients with a high BPTF expression had shorter overall survival than those with a low BPTF expression (P = 0.002). b The CRC patients with a high or low E-cadherin expression had a similar survival (P = 0.073). c The CRC patients with a high vimentin expression had shorter overall survival compared with low vimentin expression (P = 0.003)

BPTF expression is an independent risk factor for survival of CRC patients

To investigate whether higher BPTF, vimentin and E-cadherin expression alone could predict poor outcome of CRC patients, we performed univariate and multivariate Cox proportional hazard analyses to examine the effect of BPTF, vimentin and E-cadherin expression (Table 3). The results showed that in the univariate analysis, UICC stage, nodal metastasis, recurrence, high BPTF and vimentin expression, and low E-cadherin affect the prognosis of CRC patients, and in the multivariate analysis, UICC stage, nodal metastasis, recurrence and high BPTF expression were statistically significant independent prognostic factors (P < 0.05) for survival (Table 3). These results indicated that higher BPTF expression is an independent risk factor for survival of CRC patients along with UICC stage, nodal metastasis and recurrence.

BPTF expression correlated with EMT markers vimentin and E-cadherin

Since BPTF may be an EMT inducer, could affect the expression of EMT landmark protein vimentin and E-cadherin, we hypothesized that correlation between BPTF and vimentin and E-cadherin expression may be tight. We next examined the correlations between BPTF and vimentin and E-cadherin expression in CRC tissues. The Spearman rank correlation analysis indicated that high expression of BPTF significantly correlated with evaluated vimentin expression (r = 0.827, P = 0.010) and reduced E-cadherin expression (r = −0.315, P = 0.012) (Table 4). What’s more important, we found BPTF, E-cadherin and vimentin expression in the similar lesion of the same CRC sample presented obviously correlation, where high BPTF expression showed low E-cadherin expression and high vimentin expression (Fig. 6). These data suggested that BPTF may correlate with EMT markers in CRC.

Table 4.

Correlation between BPTF expression and the expression of vimentin and E-cadherin in CRC patients by Spearman rank correlation analysis (n = 105)

Protein expression	BPTF (n)		r
	High	Low	P
Vimentin (n)
High	46	13	0.827
Low	25	21	0.010
E-cadherin (n)
High	13	14	−0.315
Low	58	20	0.012

Significant results (P < 0.05) are given in bold

Fig. 6.

Representative photomicrographs of BPTF, E-cadherin and vimentin expression in CRC BPTF-positive expression (a), E-cadherin-negative expression (b) and vimentin-positive expression (c) in the similar field of CRC

Discussion

Most of the human solid tumors including CRC are carcinomas that originate from epithelial cell types. In order to break away from neighboring cells to invade adjacent cell layers, dissemination and metastasis, these tumor cells must lose their cell–cell adhesions and acquire motility and invasiveness. Recent years, these series of changes are considered to be closely related with EMT program (Kalluri and Weinberg 2009). Some transcriptional factors (TFs) such as Snail, Slug, Twist, and Zeb1/2, involved in the functional loss of E-cadherin and gain of vimentin in an epithelial cell during EMT, were considered as the hallmark of EMT. These TFs were identified to take part in regulating the EMT program in various malignant tumors; thus, intervention of these TFs could suppress EMT and invasion–metastasis cascade in experiments (Yang and Weinberg 2008; Hanahan and Weinberg 2011; Sanchez-Tillo et al. 2012). However, all of the discovered TFs have shortcomings for completed inhibiting tumor progression in clinical trails. Therefore, further researches on EMT-TFs are needed to screen novel powerful TFs to modulate EMT in tumor.

BPTF, a chromatin-mediated regulation of transcription factor, was used to be considered to participate in the development of epiblast and visceral endoderm. Our findings first confirmed that BPTF expression was significantly overexpressed in CRC tumor tissues through real-time PCR, Western blot and immunohistochemistry, respectively. In addition, high expression of BPTF was correlated significantly with advanced tumor progression and clinicopathological features. The expression of BPTF tends to be elevated in CRC tissues with poor differentiation, advanced clinical stage, nodal metastasis and tumor relapse, suggesting that BPTF expression might be high clinical relevance of CRC progression. The impact of overexpression of BPTF on clinical outcome was assessed by Kaplan–Meier analysis. The OS rates of the high BPTF expression patients group were significantly lower than those in the low group. The results of univariate and multivariate analyses clearly showed that high BPTF expression was a statistically significant independent risk factor affecting the OS of CRC patients, suggesting that the elevated expression of BPTF could be considered as a useful survival predictor for CRC patients. In one of our previous study, we found that BPTF maybe a prognosis indicator for hepatocellular carcinoma (HCC) (Xiao et al. 2014). Therefore, we compare the BPTF expression in CRC and HCC, which showed there was no difference between the two cancers (Table S1). The Kaplan–Meier analysis and Cox regression analysis also indicated that high BPTF expression correlated poor survival of the two cancers (Fig S1; Table S2). These results showed that BPTF may be a pan-marker of gastrointestinal cancer for poor prognosis.

More significantly, we discovered the expression of EMT hallmark proteins E-cadherin and vimentin by IHC, simultaneously. These results demonstrated that the expression level of epithelial marker E-cadherin was obviously lower in CRC tumor tissues than paired NATs, while the expression level of mesenchymal marker vimentin was significantly higher in tumor tissues. Both of them are correlated with adverse clinicopathological features, such as tumor size and histology stage. However, E-cadherin is associated with clinical stage and relapse, and vimentin is related to the nodal metastasis. The diverse roles of these two proteins lead to their different functions in the EMT process of cancer (Alderton 2013; Kalluri 2009). In the survival analysis, it was showed that vimentin rather than E-cadherin could affect the 5-year OS of CRC patients, although neither of them was an independent risk factor for CRC. In the IHC study, photomicrograph presented the directed correlation of BPTF, E-cadherin and vimentin expression in the similar lesion visually. Spearman correlation analysis further confirmed this relationship. The above findings fully suggest that BPTF may be an important regulator of EMT process via elevating the expression of vimentin and decreasing the expression of E-cadherin.

However, besides our study, the potential mechanism of BPTF regulating vimentin and E-cadherin still has no reports. Some previous researches may shine a little light on the relationship between BPTF and EMT. Wysocka et al. (2006) found that the PHD finger of BPTF which was considered as a trimethyl-lysine-binding domain could highly specifically recognize the H3K4me3. Malouf et al. (2013) discovered that EMT was accompanied by a widespread gain in H3K4me3-mediated gene such as Twist1 activation. Furthermore, Wang et al. (2014) found that the recruitment of H3K4me3 to ZEB1 gene promoter could activate the ZEB1 expression to induce EMT. Both Twist1 and ZEB1 are referred as the crucial and widely accepted transcriptional factors of EMT with strong regulatory functions (Craene and Berx 2012). These findings hint that BPTF probably control EMT process through epigenetic modulating mechanism via the PHD domain. Mulder et al. (2012) also found that BPTF took part in the epidermal stem cell differentiation through acetylation or methylation. Recently, a series of studies discovered that BPTF played an important role in embryogenesis and differentiation, partly through the TGF-β/smad signaling, which also played a pivotal role in modulating EMT process (Landry et al. 2008; Goller et al. 2008). Some researches further indicated that BPTF had a cancer-promoting effect on the human lung embryonal-derived cells and various cancer cell lines (Buganim et al. 2010). Given that BPTF has multiple important domains such as PHD, DDT and bromodomain, which may target or interact with many EMT-related genes, BPTF can promote cancer progression by inducing EMT through transcriptional and epigenetic mechanisms.

Taken all together, our results suggested that high expression of BPTF was significantly correlated with tumor progression and could be used as a potent prognostic biomarker of CRC. Moreover, BPTF may promote invasion and metastasis via EMT process, and further precise mechanism is needed to unveil.

Electronic supplementary material

Below is the link to the electronic supplementary material.