SATB1 expression is correlated with β-catenin associated epithelial–mesenchymal transition in colorectal cancer

ABSTRACT

SATB1, a global gene regulator, has been implicated in the growth and metastasis of multiple cancers, including colorectal cancer. While the understanding about the role of SATB1 in CRC remains limited. The aim of our study is to investigate the expression of SATB1 in CRC, and the relationship between SATB1 expression pattern and clinicopathological variables. A further aim is to analyze the correlation between SATB1 expression and epithelial–mesenchymal transition in CRC. Immunohistochemical expression of SATB1, β-catenin, E-cadherin, CK20, Vimentin, SMA, and desmin were assessed in a cohort of 200 patients using tissue microarrays. SATB1 was expressed in 133 (66.5%) CRC primary lesions, 14 (28%) adjacent colorectal mucosa specimens, and 60 (75%) corresponding lymph node metastases. The expression level of SATB1 was significantly higher in lymph node metastases than in CRC primary lesions and normal mucosa (P = 0.000). High expression of SATB1 in CRC was strongly correlated with poor differentiation of tumor tissues (P = 0.000). High expression of SATB1 was significantly correlated with aberrant expression of β-catenin (P = 0.0005), low expression of E-cadherin (P = 0.000) and CK20 (P = 0.000) and with high expression of Vimentin (P = 0.001). No SMA or desmin protein was expressed in the CRC cells. Our results suggested that high expression of SATB1 is significantly correlated with poor differentiation of CRC. SATB1 might promote the epithelial–mesenchymal transition by increasing the aberrant expression of β-catenin.

Abbreviations

SATB1

special AT-rich sequence-binding protein 1

CRC

colorectal cancer

CK20

cytokeratin 20

EMT

epithelial–mesenchymal transition

Introduction

Colorectal cancer (CRC) is the third most common cancer in the world,¹ and the fourth most deadly cancer (after lung, liver and stomach cancer).² The incidence rates of CRC continue to increase in economically transitioning countries. In addition, the latest statistical data showed that the incidence rate has increased among young adults and decreased among older adults.³ Although various modifiable risk factors for CRC have been found and rapid progress has been made in surgical techniques and gastrointestinal endoscopy, the molecular mechanisms remain poorly understood. Therefore, a more comprehensive understanding about the genetic and epigenetic alterations that drive the development and progression of CRC is sorely required.

Special AT-rich sequence-binding protein 1 (SATB1) is a nuclear matrix attachment region binding protein that acts as a transcription factor and epigenetic regulator by recruiting related proteins such as Mi-2, HDAC1, and ACF1, thereby remodelling the chromatin structure and regulating multiple genes.⁴ SATB1 was previously thought to play a vital role in T-cell development and differentiation.^5,6 Recently, however, increasing evidence has indicated that altered SATB1 expression is closely associated with malignant tumors. In addition, SATB1 was initially found to have a role in promoting growth and metastasis in breast cancer.⁷ It has subsequently been found to correlate with poor prognosis in several other forms of tumor, such as liver cancer,⁸ prostate cancer,⁹ ovarian cancer,¹⁰ gastric cancer,¹¹ and renal cell carcinoma.¹² As such, SATB1 is considered to be a new type of key oncogene regulator. In several current investigations of SATB1 expression in CRC, the data showed that high expression of SATB1 promotes the growth and metastasis of CRC.^13,14 Another study suggested that loss of special SATB1 is a predictor of poor survival in patients with CRC.¹⁵ No final conclusions have been reached on this matter yet, and the understanding about the underlying molecular mechanisms was limited.

Epithelial-to-mesenchymal transition (EMT) is a biological process of transformation from epithelial phenotype to mesenchymal phenotype. This process results in a spectrum of epithelial cellular changes, including cell skeleton reconstruction, loss of cell–cell adhesion, increased mobility, and the acquisition of a mesenchymal phenotype.¹⁶ Current data have shown that EMT might be a critical event in the invasion, progression, and metastasis of epithelial cancers,^17,18 and increasing evidence has confirmed that EMT-related molecular pathways might be involved in CRC development and progression.¹⁹

β-catenin protein is a central molecule in the Wnt/β-catenin pathway, which is thought to initiate EMT in CRC.²⁰ The translocation of β-catenin into the nucleus might be associated with T-cell factor to form a functional transcription factor that can mediate the transactivation of target genes involved in tumor progression, invasion, and metastasis.²¹

In this study, we investigated the expression pattern of SATB1 and its correlation with clinicopathological parameters in a Chinese CRC cohort. In addition, we explored the probable mechanism of SATB1 action in CRC. The expression of EMT-related proteins, such as E-cadherin, cytokeratin 20 (CK20), Vimentin, α-smooth muscle actin (SMA), and desmin, and the localization of β-catenin protein were further detected. The associations among SATB1 protein expression, β-catenin localization, and EMT were explored and discussed.

Materials and methods

Patient cohort and sample collection

All biopsies and the medical records of 200 CRC patients were obtained from Suzhou Hospital, affiliated to Nanjing Medical University, between January 2011 and December 2012. None of the patients received chemotherapy or radiation therapy prior to surgery. All of the tumors were re-evaluated histopathologically on haematoxylin- and eosin-stained slides. The study included 200 cases of CRC primary lesions, 50 cases of adjacent colorectal mucosa, and 80 cases of corresponding lymph node metastasis. Medical records and pathology characteristics were reviewed, and data regarding patient sex and age, tumor diameter, differentiation, depth of invasion, lymphatic metastasis, distant metastasis, and TNM stage were collected. This study was conducted in accordance with the regulations and with the approval of our institutional review board.

The cohort included 107 male and 93 female patients; the median age was 66 y (range, 35–90 years). In terms of histological differentiation, there were 42 cases of high-grade differentiation, 127 cases of moderate-grade differentiation, and 31 cases of low-grade differentiation. Lymphatic and distant metastases were observed in 101 and 11 cases, respectively. In terms of TNM staging, there were 19 cases of stage I, 78 cases of stage II, 92 cases of stage III, and 11 cases of stage IV. Eighty metastatic lymph node samples came from 101 patients with lymph node metastasis, including 11 cases of high-grade differentiation, 60 cases of moderate-grade differentiation, and 9 cases of low-grade differentiation. Among these 80 cases, 10 cases had distant metastasis.

Tissue microarray

Areas representative of cancer were marked on haematoxylin- and eosin-stained slides. Tissue microarrays (TMAs) were constructed from archived formalin-fixed and paraffin-embedded tissue blocks using a manual tissue arrayer. Two 1.0mm cores were taken from each tumor, adjacent colorectal mucosa, and carcinoma nests in the matching metastatic lymph node.

Immunohistochemistry

For immunohistochemical analysis, 4μm TMA sections were cut from representative paraffin blocks and boiled in 10 mM sodium citrate buffer following deparaffinization and dehydration. Then, the specimens were blocked using anti-SATB1 antibody (catalog ab92307, 1:50), anti-β-catenin antibody (Dako), anti-E-cadherin antibody (Dako), anti-CK20 antibody (PW31; Novocastra), anti-Vimentin antibody (V9; Zymed), anti-SMA antibody (1A4; Dako), and anti-desmin antibody (ZC18; Zymed) overnight at 4°C. The sections were subsequently incubated in secondary antibody labeled with streptavidin–biotin at room temperature for 30 minutes. Diaminobenzidine (3, 30-diaminobenzidine) was used for vizualization. Appropriate negative and positive controls were included in each slide run.

Staining evaluation

TMAs were judged separately by 2 expert pathologists who were blinded to the clinicopathologic information. The estimated fractions of cells with nuclear SATB1 expression were denoted as 0 (0–1%), 1 (2–25%), 2 (26–50%), 3 (51–75%), and 4 (>75%). Nuclear intensity was scored as follows: 0 = negative, 1 = weak, 2 = moderate, and 3 = strong. A combined nuclear score (NS) was constructed by multiplying fraction and intensity. A final score <3 was considered negative, a score of 3–9 was considered intermediate expression, and a score ≥ 9 was considered high expression. CK20 was expressed on the cell membrane; the standard for evaluation was the same as that for SATB1. Normal expression of β-catenin was defined as membranous staining as seen in the normal colorectal mucosa. The expression of β-catenin in tumor cells was evaluated from cell membrane, cytoplasmic, and nuclear staining separately. If more than 70% of the cancer cells exhibited membrane staining, it was considered normal expression; otherwise, it was considered absent of membranous staining. If more than 10% of the cancer cells exhibited cytoplasmic or nucleus positive staining, it was considered ectopic expression. Lack of membranous staining and ectopic expression in the cytoplasm/nucleus were defined as aberrant expression. E-cadherin was normal membranous staining. A combined score was constructed by multiplying the proportion of positive cells and intensity of staining. The proportions of positive expression in tumor cells were evaluated as follows: 0 (0–10%), 1 (11–50%), 2 (51–75%), and 3 (76–100%). The scoring criteria for staining intensity were 0 (no staining), 1 (light yellow staining), 2 (yellow staining), and 3 (brown staining). A final score <3 was defined as negative and a score ≥3 was defined as positive. Vimentin was normally expressed in the cytoplasm of the mesenchymal cell; positive staining was brown or yellow. If there was positive staining in CRC epithelial cells, regardless of the number, it was defined as positive; otherwise, it was negative. SMA and desmin were cytoplasm expression; the standard for evaluation was the same as that of vimentin.

Statistical analysis

All statistical analyses were carried out using PASW Statistics 18 (PSS Inc., Chicago, IL). The association between SATB1 expression and different clinicopathological variables was analyzed using a 2-tailed Pearson's χ² test. Correlations among the expression levels of SATB1, β-catenin, and E-cadherin were evaluated by Spearman's bivariate correlation test. The association between SATB1 expression and Vimentin expression was analyzed with a χ² test. P < 0.05 was considered statistically significant.

Results

Expression of SATB1 protein increased in CRC prime lesions and lymph node metastases

To investigate the expression and localization of SATB1 protein in CRC tissues, 200 specimens of CRC primary lesions, 50 specimens of normal colorectal mucosa, and 80 specimens of corresponding lymph node metastasis were inspected using immunohistochemistry. SATB1 protein was negatively expressed in normal colorectal mucosa (Fig. 1A), and was located mainly in the nuclei of cancer cells (Fig. 1B); some tissues exhibited combined nuclear and cytoplasmic expression (Fig. 1C). The positive expression ratio of SATB1 protein in lymph node metastases (60/80, 75%) was significantly higher than in primary lesions (133/200, 66.5%) and normal colorectal mucosa (14/50, 28%) (P = 0.000).

Figure 1.

The expression of SATB1, E-cadherin, CK20, Vimentin and β-catenin in matched normal colorectal mucosa, primary CRC lesions and lymph node metastasis (Magnification, 200×). (A) Negative expression of SATB1 in normal colorectal mucosa. (B) SATB1 moderately expressed in the nuclear of primary CRC tissue. (C) Combined nuclear and cytoplasmic expression of SATB1 in CRC lymph node metastasis. (D) Strongly positive membrane expression of E-cadherin in normal colorectal mucosa. (E) Membrane expression of E-cadherin in CRC tissue. (F) E-cadherin expressed in cytoplasm instead of membrane in lymph node metastasis. (G) CK20 positively expressed on the membrane of normal colorectal mucosa. (H) CK20 expression was lost in this specimen of primary CRC. While the residual normal glandula still showed membrane expression. (I) Negative expression of CK20 in CRC lymph node metastasis. (J) Vimentin was negatively expressed in epithelial cells of normal colorectal mucosa. (K) Vimentin was expressed in mesenchymal cells and several cancer cells in CRC tissue. (L) Vimentin was expressed in mesenchymal cells and several cancer cells in lymph node metastasis. (M) Membrane expression of β-catenin in normal colorectal mucosa. (N) In addition to membrane expression, combined nuclear and cytoplasmic expression of β-catenin can also be detected in primary CRC tissue. (O) Combined nuclear and cytoplasmic expression of β-catenin in CRC lymph node metastasis.

Correlation of clinicopathological features with expression of SATB1 protein in CRC primary lesions

The relationship between SATB1 protein expression and CRC clinicopathological parameters was summarised in Table 1. Thirty-seven (18.5%) cases exhibited high levels of SATB1 expression, 96 (48.0%) cases exhibited intermediate levels, and 67 (33.5%) cases exhibited negative SATB1 expression. Our data demonstrated that high expression of SATB1 protein was significantly associated with tumor differentiation (P = 0.000). No significant associations were found between SATB1 protein expression and other clinicopathological characteristics, such as sex, age, tumor diameter, invasion depth, lymph node metastasis, and distant metastasis (P > 0.05).

Table 1.

Correlation between SATB1 expression in CRC primary lesions and clinicopathological varibles.

		SATB1 expression
Clinicopathological factors	No.	High(%)	Intermediate(%)	Negative(%)	P-value
Sex					0.209
Male	107	15 (14.0%)	55 (51.4%)	37 (34.6%)
Female	93	22 (23.7%)	41 (44.1%)	30 (32.2%)
Age (years)					0.783
<60	60	12 (20.0%)	30 (50.0%)	18 (30.0%)
≥60	140	25 (17.9%)	66 (47.1%)	49 (35.0%)
Tumor diameter					0.705
≤5 cm	125	21 (16.8%)	62 (49.6%)	42 (33.6%)
>5cm	75	16 (21.3%)	34 (45.3%)	25 (33.4%)
Tumor differentiation					0.000
Well	42	1 (2.4%)	3 (7.1%)	38 (90.5%)
Moderate	127	23 (18.1%)	77 (60.6%)	27 (21.3%)
Poor	31	13 (41.9%)	16 (51.6%)	2 (6.5%)
Depth of invasion					0.841
Mucosa and muscularis	26	5 (19.2%)	13 (50.0%)	8 (30.8%)
Serosa	70	11 (15.7%)	37 (52.9%)	22 (31.4%)
Beyond the serosa	104	21 (20.2%)	46 (44.2%)	37 (35.6%)
Lymph node metastasis					0.350
Positive	101	20 (19.8%)	52 (51.5%)	29 (28.7%)
Negative	99	17 (17.2%)	44 (44.4%)	38 (38.4%)
Distant metastasis					0.586
Positive	11	1 (9.0%)	5 (45.5%)	5 (45.5%)
Negative	189	36 (19.0%)	91 (48.2%)	62(32.8%)
TNM					0.576
I+II	97	17 (17.5%)	44 (45.4%)	36 (37.1%)
III+IV	103	20 (19.4%)	52 (50.5%)	31 (30.1%)

Correlation of clinicopathological features with expression of SATB1 protein in lymph node metastases

Among the 200 patients in this study, 101 were diagnosed as positive for lymph node metastasis. Matched lymph node metastases were obtained from 80 of those patients. The relationship between SATB1 protein expression in lymph node metastases and CRC clinicopathological parameters was summarised in Table 2. Twenty-seven (33.8%) patients exhibited high levels of SATB1 expression (Fig. 1C), 33 (41.2%) patients exhibited intermediate levels, and 20 (25.0%) patients exhibited negative SATB1 expression. In those metastatic lymph node specimens, SATB1 protein expression was detected in all 10 cases of poor differential tumor tissues. The expression level of SATB1 was significantly higher in poorly differentiated tumors (100%, 10/10) than in well-moderate differentiated tumors (70%, 49/70) (P = 0.031). No significant associations were found between SATB1 protein expression and sex, age, tumor diameter, invasion depth, or distant metastasis (P > 0.05).

Table 2.

Correlation between SATB1 expression in tumor nests of lymph nodes and clinicopathological varibles.

		SATB1 expression
Clinicopathological factors	No.	High(%)	Intermediate(%)	Negative(%)	P-value
Sex					0.089
Male	38	12 (31.6%)	20 (52.6%)	6 (15.8%)
Female	42	15 (35.7%)	13 (31.0%)	14 (33.3%)
Age (years)					0.826
<60	26	10 (38.5%)	10 (38.5%)	6 (23.0%)
≥60	54	17 (31.5%)	23 (42.6%)	14 (25.9%)
Tumor diameter					0.758
≤5 cm	55	20 (36.4%)	22 (40.0%)	13 (23.6%)
>5cm	25	7 (28.0%)	11 (44.0%)	7 (28.0%)
Tumor differentiation					0.031
Well	10	5 (50.0%)	2 (20.0%)	3 (30.0%)
Moderated	60	15 (25.0%)	27 (45.0%)	18 (30.0%)
Poor	10	7 (70.0%)	3 (30.0%)	0 (0.00%)
Depth of invasion					0.438
Mucosa and muscularis	3	1 (33.3%)	2 (66.7%)	0 (0.00%)
Serosa	18	4 (22.2%)	7 (38.9%)	7 (38.9%)
Beyond the serosa	59	22 (37.3%)	24 (40.7%)	13 (22.0%)
Distant metastasis					0.490
Positive	10	3 (30.0%)	3 (30.0%)	4 (40.0%)
Negative	70	24 (34.3%)	30 (42.9%)	16 (22.8%)

Expression of EMT markers and their correlations with SATB1 and β-catenin expression in CRC

All of the colorectal mucosa specimens in this study exhibited purely membranous expression of E-cadherin (Fig. 1D) (100%, 50/50). In contrast, membranous expression of E-cadherin was remarkably lower in the CRC cases (46.5%, 93/200). CK20 was expressed in the upper portions of the normal epithelial glands (Fig. 1G); the positive expression ratio was 100%. In the CRC lesions, CK20 was detected in only 77.5% (155/200) of the specimens; 22.5% (45/200) of the CRC tissues lacked CK20 expression (Fig. 1H). Vimentin is typically expressed in the cytoplasm of mesenchymal cells. In our study, no positive Vimentin staining was detected in the epithelial cells of the colorectal mucosa; however, sporadic expression of vimentin was detected in the cancer cells. SMA protein was expressed in the smooth muscle cells in tumor stroma and in vascular smooth muscle cells. No SMA was detected in the epithelial cells. Similarly, desmin was expressed in the smooth muscle cells, but not in cancer cells, regardless of the clinicopathologic characteristics. In SATB1 high-expression specimens, 78.4% (29/37) of the cases exhibited a deficiency of E-cadherin membrane expression. In the SATB1 intermediate expression group, 54.2% (52/96) of the cases lost E-cadherin expression; in the SATB1 negative expression group, only 38.8% (26/67) of the cases lacked E-cadherin expression. In contrast, positive expression of Vimentin was significantly higher in the SATB1 high expression group (24.3%, 9/37) than in the intermediate expression group (6.3%, 6/96) and negative expression group (3.0%, 2/67). Table 3 summarized the correlations between SATB1 expression and EMT markers in CRC. A significant association was identified between high expression of SATB1 and decreased membranous expression of E-cadherin (Spearman's correlation coefficient, −0.348; P = 0.000). And SATB1 expression was negatively correlated with CK20 expression (Spearman's correlation coefficient, −0.390; P = 0.000). In addition, high expression of SATB1 was significantly associated with Vimentin expression (P = 0.001). Table 4 summarized the correlations between β-catenin expression and EMT markers in CRC. Data showed that in β-catenin aberrant-expression specimens, 67.0% (77/115) of the cases lost E-cadherin membrane expression, significantly higher than β-catenin normal-expression specimens (30/85, 35.3%) (P = 0.000). The positive expression ratio of CK20 in β-catenin aberrant-expression specimens (45/115, 39.1%) was notably lower than in β-catenin normal-expression samples (51/85, 60.0%) (P = 0.013). And the aberrant-expression of β-catenin was significantly associated with Vimentin expression (P = 0.030).

Table 3.

Correlation between SATB1 expression and E-cadherin, CK20 and Vimentin.

		E-cadherin (n, %)			CK20 (n, %)				Vimentin (n, %)
SATB1	n	positive	negative	P	high	intermediate	negative	P	positive	negative	P
high	37	8 (21.6)	29 (78.4)	0.001	13 (35.1)	11 (29.8)	13 (35.1)	0.006	9 (24.3)	28 (75.7)	0.001
intermediate	96	44 (45.8)	52 (54.2)		39 (40.6)	33 (34.4)	24 (25.0)		6 (6.3)	90 (93.7)
negative	67	41 (61.2)	26 (38.8)		44 (65.7)	14 (20.9)	9 (13.4)		2 (3.0)	65 (97.0)

Table 4.

Correlation between β-catenin expression and E-cadherin, CK20 and Vimentin.

		E-cadherin (n, %)			CK20 (n, %)				Vimentin (n, %)
β-catenin	n	positive	negative	P	high	intermediate	negative	P	positive	negative	P
normal	85	55 (64.7)	30 (35.3)	0.000	51 (60.0)	18 (21.2)	16 (18.8)	0.013	3 (3.5)	82 (96.5)	0.030
aberrant	115	38 (33.0)	77 (67.0)		45 (39.1)	40 (34.8)	30 (26.1)		14 (12.2)	101 (87.8)

Expression of β-catenin and its correlation with SATB1 expression in CRC

Strong membranous β-catenin expression was observed in 100% (50/50) of normal colorectal mucosa, located in the cell membranes of epithelial cells at the cell-to-cell borders (Fig. 1M). However, tumor cells exhibited a weaker expression level of β-catenin. Membranous expression in CRC tissues was down-regulated; cytoplasmic and nuclear expression were detected (Fig. 1N); and to a considerable extent, the expression of β-catenin was deficient. In the cohort, 85 (42.5%) specimens exhibited normal β-catenin expression, located in the membrane, whereas most of the specimens exhibited an abnormal localization of β-catenin (57.5%, 115/200). Fig. 2 details the expression pattern of β-catenin in CRC tumor cells, observed with immunohistochemistry. It can be seen that β-catenin was located in the membrane in 59.7% (40/67) of specimens without SATB1 expression, which is higher than in the intermediate (38.5%, 37/96) and high expression group (21.6%, 8/37). Furthermore, the occurrence rate of aberrant localization of β-catenin in the SATB1 high expression group (78.4%, 29/37), including cytoplasm overexpression, nuclear accumulation, and expression loss, is greater than in the SATB1 intermediate expression group and SATB1 negative expression group (61.5%, 59/96 and 40.3%, 27/67, respectively). The correlation analysis revealed that high level of SATB1 expression was significantly positively correlated with a reduced level of membrane-bound β-catenin in CRC cells (P = 0.0005) (Table 5). Fig. 3 presented the expression of β-catenin, E-cadherin, CK20, and Vimentin in 2 representative cases of SATB1 high expression and negative expression respectively at the same view.

Figure 2.

Intracellular localization of β-catenin in colorectal cancer tissue, with different degree of SATB1 expression.

Table 5.

Correlation between SATB1 expression and β-catenin.

		β-catenin (n,%)
SATB1 expression	n	normal	aberrant	χ2	P
high	37	8 (21.6)	29 (78.4)	15.328	0.0005
intermediate	96	37 (38.5)	59 (61.5)
negative	67	40 (59.7)	27 (40.3)

Figure 3.

Expression of β-catenin, E-cadherin, CK20, and Vimentin at the same view in 2 representative cases of CRC primary lesions (Magnification, 400×). (A-E) Case1 is of SATB1 negative expression. In this case, β-catenin located on the membrane of cancer cells, E-cadherin and CK20 also normally expressed on the cell membrane, while none Vimentin expression can be detected in CRC cancer cells. (F-J) Case2 is of SATB1 high expression. In this specimen, β-catenin showed weakly cytoplasmic expression instead of membrane expression. E-cadherin and CK20 also lost membrane expression. While Vimentin was positively expressed in CRC cancer cells.

Discussion

SATB1 is a nuclear matrix-associated protein that acts as a genome organizer by folding chromatin into loop domains and recruiting chromatin-modifying enzymes, thus orchestrating the overexpression or repression of multiple genes.^4,22 SATB1 was first reported in 1992 in thymocytes,²³ and it was shown to be involved in the proliferation, development, and differentiation of T cells.⁵ More recently, SATB1 has attracted considerable attention due to its role in the promotion of tumor growth and metastasis. A study by Han et al. demonstrated that SATB1 was expressed in aggressive breast cancer cells and its expression level had high prognostic significance, revealing a new mechanism of tumor progression.⁷

In the present study, we observed SATB1 protein expression in CRC primary lesions, corresponding lymph node metastases, and adjacent colorectal mucosa specimens by immunohistochemical analysis. Our results showed that SATB1 protein was expressed in 66.5% of colorectal cancer tissues, which is significantly higher than in normal colorectal mucosa (28%). This phenomenon has also been reported in other types of tumors, such as liver, prostate, ovarian, gastric, and kidney tumors.^8-12 We further analyzed the expression of SATB1 in corresponding lymph node metastases, and we found that SATB1 expression was remarkably higher in lymph node metastases than in CRC primary lesions. These results suggest that SATB1 plays a crucial role in promoting tumor growth, invasion, and metastasis.

Correlation analysis of SATB1 and clinicopathologic factors indicated that the expression of SATB1 was strongly correlated with poor differentiation, both in CRC tissues and in lymph node metastases. It is worth noting that in primary lesions, there was no SATB1 expression in the vast majority of well-differentiated specimens (90.5%, 38/42), while in lymph node metastases, only 30% of the specimens exhibited negative expression of SATB1. Although SATB1 expression was correlated with tumor differentiation in both primary lesions and metastatic foci, SATB1 expression in well-differentiated cancer nests displayed an apparent difference. This phenomenon is a reminder that characteristics of tumor cells undergo a great deal of change during the tumor metastasis process.

The transformation of normal colorectal mucosa to cancer is a multiple genetic and epigenetic process. During colorectal carcinogenesis, dysregulation of β-catenin is a fatal step.²⁴ Interestingly, Wnt/β-catenin signaling is crucial in T cell development and differentiation. Notani et al. demonstrated that SATB1 can recruit β-catenin and p300 acetyltransferase on the GATA-3 promoter, hence mediating Wnt/β-catenin signaling in T_H2 cells.²⁵ In order to explore whether the β-catenin pathway is involved in the SATB1-mediated regulation of differentiation and metastatic capacity of CRC, immunohistochemical staining was used to detect the expression of β-catenin. Different from normal colorectal mucosa epithelial cells, abnormal β-catenin expression, such as discontinued or absent membranous staining, with or without cytoplasmic staining and nuclear accumulation, was noted in colorectal cancer cells. In specimens with high SATB1 expression, up to 78.4% of cases showed abnormal β-catenin expression. In specimens without SATB1 expression, β-catenin localization was disordered in only 40.3% of cases. These data, combined with previous reports, indicate that SATB1 might play an important role in regulating aberrant β-catenin transposition.

The translocation of β-catenin from the cytomembrane to the nucleus or cytoplasm can trigger the expression of EMT-related proteins and pre-invasive factors.^26,27 EMT is a process by which epithelial cells transdifferentiate and obtain an invasive, mesenchymal, fibroblastic phenotype. EMT of tumor cells implicates a loss of polarity of epithelium cells, a decline in cell adhesion, and an enhanced ability to invade and metastasise. For example, the reduction in E-cadherin expression and upregulation of Vimentin expression suggest the occurrence of EMT. Therefore, we further detected the expression of EMT-related proteins, including epithelial morphology hallmarks E-cadherin and CK20, mesenchymal phenotype hallmark Vimentin, desmin, and SMA. To a large extent, we observed reduced E-cadherin expression, CK20 loss, and fragmentary expression of vimentin in cases with positive SATB1expression. The other 2 signs of mesenchymal cells—desmin and SMA—did not appear in those cancer cells. These results demonstrate the loss of epithelial morphology, although the acquisition of the mesenchymal phenotype is limited.

With these results, we analyzed the potential correlations among SATB1 expression, β-catenin localization, and expression of EMT markers in CRC, using Spearman's bivariate correlation test. Our study revealed that overexpression of SATB1 was closely related with aberrant localization of β-catenin, which might lead to reduced levels of membrane-bound β-catenin, accumulation of β-catenin in the cytoplasm, and nuclear migration of β-catenin. Overexpression of SATB1 was also negatively related with E-cadherin and CK20, whose epithelial cell membrane positions were weakened or even absent. In addition, the overexpression of SATB1 was significantly correlated with Vimentin expression. The results of our investigation strongly suggest that overexpression of SATB1 is likely to induce EMT via the Wnt/β-catenin pathway, thus influencing the differentiation, invasion, and metastasis of CRC. This results verified the hypothesis proposed by Meng WJ, that SATB1 may be involved in the development and progression of colorectal cancer in a Wnt/β-catenin signaling-dependent manner.²⁸

There are some debates regarding the role played by SATB1 in cancers. Most studies have indicated that overexpression of SATB1 is strongly associated with tumor differentiation, stage, and worsened survival rate. However, contradictory findings have been reported in studies of non-small cell lung cancers. Selinger et al. reported for the first time that loss of SATB1 was associated with poor prognosis in lung squamous cell carcinomas.²⁹ Al-Sohaily then reported that loss of SATB1 nuclear expression correlated with poor survival and a less favorable response to adjuvant chemotherapy in CRCs.¹⁵ Other researchers have suggested that SATB1 was not an independent prognostic factor, and was only significantly associated with poor prognoses in SATB2 negative tumors.¹³

As for the molecular mechanism that SATB1 played in CRC evolution process, several studies gave some clues. According to Zhang J et al.,³⁰ SATB1 expression level was strongly associated with tumor differentiation and stage in 80 cases of CRC, and was positively associated with the expression of various biological and genetic markers, including Cyclin D1, MMP-2, NF-kB, and PCNA. Fang et al. reported SATB1 over-expressed in 30 cases of CRC tissues. They studied the consequences of SATB1 overexpression in SW480, and showed that SATB1 could up-regulate MMP-2, MMP-9, cyclin D1 and vimentin, meanwhile E-cadherin was downregulated.³¹ Frömberg et al. described a correlation between SATB1 and several molecules in CRC cell line LS174T, including caspases, E-cadherin, slug, twist, β-catenin and MMP7.³²

Apart from these researches, our study consisted more clinical samples, especially collected the data of 80 cases of corresponding lymph node metastases, which made the results more comprehensive and credible. Furthermore, our study observed SATB1/E-catenin/EMT pathway in human tissues systematically. These findings are clearly distinct from previous analyses of SATB1 protein function, and provide a valuable insight into the potential role of SATB1 in regulating cell differentiation, migration, and invasion in CRC.

Moreover, many of the target genes of SATB1, such as bcl-2 and c-myc, can also be targeted by Wnt/β-catenin, suggesting a functional overlap between the Wnt/β-catenin pathway and SATB1. Whether SATB1 interacts with β-catenin and then recruits it to SATB1s genomic binding sites need to be explored further.

Taken together, our findings suggest the SATB1/β–catenin pathway as a target for CRC, driven by EMT. SATB1 might be involved in the development and progression of CRC in a SATB1/β-catenin-dependent manner. These data were obtained from the study of clinical specimens, in vitro experiments are needed in order to verify more detailed and exact molecular mechanisms.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

Supported by National Natural Science Foundation of China (81171391; XJ Zhou), and Suzhou Health Bureau Foundation (KJXW2013023; JH Lv).