Distinct patterns of ALDH1A1 expression predict metastasis and poor outcome of colorectal carcinoma

Abstract

Purpose: Aldehyde dehydrogenase 1A1 (ALDH1A1) has been proposed as a candidate biomarker for colorectal carcinoma (CRC). However, the heterogeneity of its expression makes it difficult to predict the outcome of CRC. The aim of this study was to evaluate the diagnostic and prognostic value of this molecule in CRC. Methods and Results: In this study, we examined ALDH1A1 expression by immunohistochemistry including 406 cases of primary CRC with corresponding adjacent mucosa, with confirmation of real-time PCR and Western blotting. We found that the expression patterns of ALDH1A1 were heterogeneous in the CRC and corresponding adjacent tissues. We defined the ratio of ALDH1A1 level in adjacent mucosa to that in tumor tissues as R_A/C and found that the capabilities of tumor invasion and metastasis in the tumors with R_A/C < 1 were significantly higher than those with R_A/C ≥ 1. Follow-up data showed the worse prognoses in the CRC patients with R_A/C < 1. For understanding the underlying mechanism, the localization of β-catenin was detected in the CRC tissues with different patterns of ALDH1A1 expression from 221 patients and β-catenin was found preferentially expressed in cell nuclei of the tumors with R_A/C < 1 and ALDH1A1^high expression of HT29 cell line, indicating that nuclear translocation of β-catenin might contribute to the increased potentials of invasion and metastasis. Conclusion: Our results indicate that R_A/C is a novel biomarker to reflect the distinct expression patterns of ALDH1A1 for predicting metastasis and prognosis of CRC.

Introduction

Colorectal cancer (CRC) is one of the most common malignancies with annual increases by approximately 4.2%. The patients with advanced CRC have poor outcomes, and about 93% of them survived for shorter than 5 years [1]. Although studies have showed several molecules related to metastasis and prognosis of this cancer [2-5], sensitive, reliable and practical indicators are still unavailable for surgical pathology.

Cancer stem cells (CSCs) have been recognized as a special subpopulation of cancer cells responsible for metastasis and resistance to therapies [6-9]. Therefore, the markers of CSCs have been considered as candidate predictors for metastasis and prognosis. Recently, several reports showed that aldehyde dehydrogenase 1A1 (ALDH1A1), a member of ALDH1 family, may be a biomarker of CSCs and can be used as a prognostic predictor of many cancers, including CRC [10,12]. However, the conclusions for the prognostic significance of ALDH1A1 expression in CRC tissues and corresponding adjacent mucosa remain controversial [12-16].

In the present study, we examined the patterns and levels of ALDH1A1 expression in human CRC specimens to define practical criteria for evaluating metastasis and prognosis in CRC and for better understanding the role of this functional CSC marker in CRC.

Materials and methods

Tissue samples and patient characteristics

CRC tissues were surgically obtained from 299 patients from Southwest Hospital, Third Military Medical University between 2002 and 2012, 37 patients from Tongji Hospital, Central China University of Science and Technology, Wuhan between 2010 and 2011, 70 patients from Southern Hospital, Southern Medical University, Guangzhou between 2005 and 2006. Tumor size was determined as the maximum tumor diameter, and lymph node metastasis was histologically diagnosed. All of the patients had sufficient samples for immunohistochemistry and follow-up data. Twelve pairs of fresh CRC and adjacent normal colorectal mucosa specimens were selected from Southwest Hospital. Twenty cases of normal human colorectal mucosa were obtained from colonoscopy of patients without CRC from Southwest Hospital. Both specimens were fixed in 10% buffered formalin and embedded in paraffin for preparation of 4 μm sections. Written informed consent was obtained from all patients. The Institute Research Medical Ethics Committee of Third Military Medical University granted approval for this study.

Immunohistochemistry and semi-quantitative analysis

Primary antibodies used in this study included mouse monoclonal anti-human ALDH1A1 (1:1000 dilution, BD Pharmingen, CA), and mouse monoclonal anti-human β-catenin (1:200, Santa Cruz Biotech, CA). Immunostaining was performed using the Envision System with diaminobenzidine (Dako, Glostrup, Denmark) as previously described. The expression levels were semi-quantitatively scored by two independent pathologists using a visual grading system based on staining intensity and number of positive cells [17]. The percentage of positive cells was evaluated quantitatively. We defined more than 5% to 25% positive cells as 1 score, scored as 2 for staining of 26% to 50%, and 3 scores for staining of 51% to 75%, and 4 scores for staining of > 75% of the cancer cells evaluated. Intensity of the cells was graded as follows: 0 score for no signal; 1 score for weak; 2 scores for moderate; and 3 scores for strong staining. A total ‘staining score’ of 0-12 was gained from multiplying the scores of percentage of positive cells by the scores of intensity of positive cells [11].

Quantitative reverse transcription pcr (QRT-PCR)

Tumor tissues and adjacent mucosa from 12 cases of CRC (Southwest Hospital, June 2012) were used for analysis of ALDH1A1 mRNA with qRT-PCR using SYBR^® PrimeScript™ PCR kit (TaKaRa, Japan). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control. The sequences of primers, the product sizes and the annealing temperatures are shown in (Table S1).

Western blotting

The samples used in qRT-PCR were also analyzed by western blotting using NE-PER Nuclear/Cytosol Extraction Kit (Thermo Scientific, PA). Primary antibodies used in this study included rabbit polyclonal anti-human ALDH1A1 (1:1000, BD Pharmingen, San Diego, CA), mouse monoclonal anti-human β-catenin (1:2000, Santa Cruz Biotech, CA), rabbit polyclonal anti-human β-actin (1:1000, Beyotime, Haimen, China), and rabbit polyclonal anti-human Lamin B1 (1:1000, Abcam, Cambridge, British). After the application of horseradish peroxidase-labeled secondary antibodies, chemiluminescence (SuperSignal West Pico, Pierce, PA) was quantified using Image Quant 5.0.

Flow cytometry

An ALDEFLUOR kit (Stem Cell Technologies, Canada) was used to isolate tumor cells with ALDH1A1A1 enzymatic activity according to the manufacturer’s instruction. Briefly, colorectal carcinoma cells were suspended in ALDEFLUOR assay buffer containing ALDH1A1 substrate, add BODIPY amino acetaldehyde (BAAA) and incubated for 30 min at 37°C. Dimethylaminobenzaldehyde (DEAB) that is a specific ALDH1A1 inhibitor was used as a negative control.

Tumorigenicity assay

Colorectal carcinoma cells were injected subcutaneously into the axilla of 5-week-old NOD-SCID mice with 1 × 10⁵ cells per mice. At the end of sixth week, mice were euthanized and xenograft tumors were removed and made section.

Statistical analysis

SPSS (v17.0) software package was used for statistical analysis of the correlations between ALDH1A1 expression and clinicopathological criteria in the study. R_A/C was defined as a ratio of ALDH1A1 levels in adjacent mucosa to those in tumor tissues. Survival curves were estimated using Kaplan-Meier product-limit method and log-rank test was used. Univariate and multivariate survival analyses were performed using a Cox proportional-hazards model. Results were considered statistically significant at P < 0.05.

The prognosis of patients with CRC positive for the expression of ALDH1A1 was calculated using the unadjusted odds ratio (OR) with the corresponding 95% CI according the overall survival (OS) in cases and controls. Other clinicopathological factors were sorted into several subgroups: sex, age, site, grade (WHO), pT status, pN status, pM status, and clinical stage. Fixed and random effects models were used to calculate a pooled OR. The statistical significance of the pooled OR was evaluated with Z test and P values, and P < 0.05 was considered statistically significant. Heterogeneity across studies was evaluated by applying a Q test. In this approach, the Q value is defined as identical to the effect size of the I² value. A random effects model was used when the I² value for heterogeneity test was > 50%; otherwise, a fixed effects model was used. Begg’s rank correlation method and Egger’s weighted regression method were used to assess publication bias (P < 0.05 was considered statistically significant). All statistical tests for this meta-analysis were performed using Review Manager 4.2 software.

Results

Expression levels of ALDH1A1 in colorectal carcinoma and adjacent tissues

First, we analyzed the expression of ALDH1A1 in 20 cases of normal colorectal mucosa. To our surprise, ALDH1A1 displayed a heterogeneous pattern, ranging from no staining at all, to several positive cells distributed at the bottom of crypts where stem or progenitor cells are putatively located, or to uniform staining throughout the whole crypts, including the differentiated absorptive and secretory cells at higher crypt levels. Next, the expression pattern of ALDH1A1 was analyzed in CRC samples containing both the tumor tissue and the corresponding adjacent mucosa (Figure 1A-F). A significantly heterogeneity in staining patterns was also observed. One hundred and ninety-eight out of 406 (48.8%) adjacent mucosa samples were positively stained for ALDH1A1. The staining intensity was fairly variable, ranging from low (1.0 ± 0.0) to high (2.7 ± 0.5). The overall expression score was 4.0 ± 2.8. In tumor tissues, 55.4% cases (225/406) were ALDH1A1 positive. ALDH1A1 cytoplasmic staining was of variable intensity, ranging from 1.0 ± 0.0 to 2.5 ± 0.5, both between different cases, and within a single case. The overall expression score was 5.1 ± 2.9. We also found that even though some patients had a high expression of ALDH1A1 in their cancerous tissues, the expression of ALDH1A1 was fairly low in the corresponding adjacent tissue (32.3%, 131/406). On the contrary, in some patients with low or absent ALDH1A1 expression in their tumor tissue, ALDH1A1 expression was high in the adjacent colorectal mucosa (13.8%, 56/406). Only 16.0% (65/406) cases displayed no difference in ALDH1A1 expression between tumor tissue and the corresponding adjacent mucosa. These results derived from IHC staining were further verified in the fresh CRC samples from 12 patients by real time PCR and Western blotting (Figure 1G and 1H).

Figure 1.

Two distinct patterns of ALDH1A1 expression in colorectal carcinoma. IHC staining of ALDH1A1 in CRC tissue and corresponding adjacent colorectal mucosa (A-F). ALDH1A1 expression in CRC tissue and corresponding adjacent colorectal mucosa was detected by real-time PCR (G). ALDH1A1 expression in CRC tissue and corresponding adjacent colorectal mucosa was detected by Western blotting (H). The scale bar 25 μm in (A to F).

Thus, based on the relative expression levels of ALHD1A1 between the adjacent and the tumor tissue, which we termed R_A/C, CRC patients can be divided into two groups. It appears that the pattern of ALDH1A1 expression changes during the progression from normal colorectal mucosa to cancer and in some patients, as ALDH1A1 expression may be up-regulated (R_A/C < 1) or down-regulated (R_A/C ≥ 1) (Figure 1A-F). In order to evaluate R_A/C can be used as an independent prognostic predictor for CRC. We assessed the relationship be tween R_A/C patte rns and the clinicopathologic characteristics of ptien ts in our study com prising 406 cases of CRC patients. We observed that R_A/C ≥ 1 (vs. R_A/C < 1) cases were significantly associated with better differentiated tumors, lower Clinical stage, and fewer lymph node and distal metastases (Table 1). Thus, tumors in the R_A/C ≥ 1 group exhibit better differentiation and lower invasive and metastatic properties than tumors in R_A/C < 1 group. A univariate Cox proportional hazard regression model indicated that R_A/C correlated with the prognosis of CRC patients (221 patients). Patients in the R_A/C ≥1 group had a median survival of 57.1 months, whereas patients in the R_A/C < 1 group had a median survival of 35.4 months (Figure 2, Table 2). The R_A/C ratio, together with other prognostic clinicopathologic factors, was also examined in a multivariate model (221 patients) (Table 2). The results indicated a favorable prognostic value for R_A/C in terms of prolonged overall survival (OS) (HR 2.109, 95% CI: 1.315 to 3.381, P = 0.000).

Table 1.

Relationship between the expression of ALDH1A1 and pathological features

		All cases	ALDH1A1 protein expression		P valuea
			RA/C ≥ 1	RA/C < 1
Sex					0.989
	Male	237	139 (58.6%)	98 (41.4%)
	Female	169	99 (58.5%)	70 (41.5%)
Age					0.801
	< 56.3 yb	177	105 (59.3%)	72 (40.7%)
	≥ 56.3 y	229	133 (58.1%)	96 (41.9%)
Site					0.003
	Rectum	243	128 (52.7%)	115 (47.3%)
	Colon	163	110 (67.5%)	53 (32.5%)
Grade (WHO)					0.002
	G1	65	47 (72.3%)	18 (27.7%)
	G2	288	170 (59.1%)	118 (40.9%)
	G3	53	21 (39.6%)	32 (60.4%)
pT status					0.026
	1	8	5 (62.5%)	3 (37.5%)
	2	110	76 (69.1%)	34 (30.9%)
	3	253	142 (56.1%)	111 (43.9%)
	4	35	15 (42.9%)	20 (57.1%)
pN status					0.000
	0	265	189 (71.3%)	76 (28.7%)
	1	141	49 (34.8%)	92 (65.2%)
pM status					0.000
	pMX	345	221 (64.1%)	124 (35.9%)
	pM1	61	17 (27.9%)	44 (72.1%)
Clinical stage					0.000
	I	98	70 (71.4%)	28 (28.6%)
	II	144	109 (75.7%)	35 (24.3%)
	III	103	42 (40.8%)	61 (59.2%)
	IV	61	17 (27.9%)	44 (72.1%)

^aχ² test.

^bMean age.

Figure 2.

R_A/C predicts survival of CRC patients. Analysis by Kaplan-Meier product-limit method and log-rank test indicate that patients in the R_A/C ≥ 1 group (n = 148) had a median survival of 57.1 months, whereas patients in the R_A/C < 1 group (n = 73) had a median survival of 35.4 months (P < 0.001).

Table 2.

Univariate and multivariate analysis of different prognostic parameters

Variable	All cases	Univariate analysisa		Multivariate analysisb
		Mean survival (months)	p	HR (95% CI)	p
Sex			0.444
Male	131	48.1
Female	90	49.7
Age at surgery			0.058
< 56.5 yc	96	52.7
≥ 56.5 y	125	47.8
Tumor location			0.323
Rectum	99	52.0
Colon	122	48.1
WHO grade			0.000
G1-G2	191	52.3
G3	30	34.5
pT status			0.000		0.002
T1-T2	66	61.5		1.0
T3-T4	155	45.0		2.801 (1.450 to 5.11)
pN status			0.000		0.983
N0	156	55.5		1.0
N1-N2	65	36.7		0.993 (0.501 to 1.966)
pM status			0.000		0.069
pMX	175	56.2		1.0
pM1	46	25.9		1.898 (0.952 to 3.783)
Clinical stage			0.000		0.023
I-II	136	59.1		1.0
III-IV	85	35.2		2.807 (1.150 to 6.850)
Expression patterns			0.000		0.002
RA/C ≥ 1	148	57.1		1.0
RA/C < 1	73	35.4		2.109 (1.315 to 3.381)

^aLog-rank test;

^bCox regression model;

^cMean age.

Expression of β-catenin and localization in colorectal carcinoma

The aberrant activation of β-catenin signaling has been reported to be involved in a number of tumors, most notably CRC [8,18]. The nuclear translocation of β-catenin acts as a coactivator of T-cell and lymphoid enhancer (TCF-LEF) factors, resulting in the transcriptional activation of target genes [8,18] and profoundly regulating the invasion, metastasis and differentiation of CRC. In addition, β-catenin is used as a supplemental marker in the analysis to stratify the expression profiles of investigated genes in CRC patients, such as CD44, CD133, CD166, and ALDH1A1 [13,19-21]. Therefore, we investigated the relationship between R_A/C and the nuclear translocation of β-catenin. By immunohistochemistry staining (Figure 3A and 3B) of 249 cases from Southwest Hospital, and Wuhan, we observed that in the tumor tissue of patients from the R_A/C ≥ 1 group, β-catenin was predominantly located at the cell membrane, with nearly absent or weak staining in the cytoplasm and nucleus (Figure 3A and 3B). In contrast, the nuclear translocation and cytoplasmic staining of this molecule was evident in the tumor cells of patients in the R_A/C < 1 group (Figure 3A and 3B). The nuclear translocation rates in two groups were 36.0% (45/125, R_A/C < 1 group) and 16.9% (21/124, R_A/C ≥ 1 group), respectively (Figure 3C). Western blotting was carried out in 6 CRC samples (3 cases of R_A/C ≥ 1, 3 cases of R_A/C < 1) in order to verify the results from the IHC staining (Figure 3D). We found that β-catenin was positive in the nuclear protein extract from two of three CRC patients of the R_A/C < 1 group. In three cases of R_A/C ≥ 1, the staining of β-catenin was weak or even absent in the nuclear extracts (Figure 3D). These results indicate that there are significant differences in the nuclear translocation of β-catenin between the two groups of CRC patients as divided by R_A/C.

Figure 3.

The nuclear localized β-catenin is found more frequently in R_A/C < 1 group. Nuclear translocation of β-catenin was detected by immunohistochemistry (A, B). Quantitative analysis of nuclear translocation of β-catenin in R_A/C ≥ 1 group and R_A/C < 1 group (P = 0.001) (C). β-catenin in nuclear extracts was detected by Western blotting. β-catenin was positive in the nuclear protein extract from two of three CRC patients of the R_A/C < 1 group (cases 1, 2 and 3), but was weak or even absent in the nuclear extracts in three cases of R_A/C ≥ 1 (cases 4, 5 and 6) (D). The scale bar is 50 μm in (A and B).

ALDH1A1^high colorectal carcinoma cells properties

We examined the proportion of ALDH1A1^high cells contained in the colorectal carcinoma cell lines HT29 and SW480 by flow cytometry, and found that the percentages of ALDH1A1^high cells in HT29 were higher than those in SW480 (77.2% vs. 3.3%) (Figure 4A), which was further verified by Western-blotting (Figure 4F). Furthermore, the sorted ALDH1A1^high cells possessed stronger capabilities of invasion as compared with ALDH1A1^low cells in the xenograft tumor. The xenograft tumors derived from ALDH1A1^low cells had complete capsules relative to those from ALDH1A1^high cells of HT 29 and SW480 (Figure 4B-E). Meanwhile, we found that ALDH1A1 high expression cell line such as HT29 showed increased expression of β-catenin in the nucleus as compared to ALDH1A1 low expression cell line such as SW480 (Figure 4F). Finally, we found that high expression of ALDH1A1 in the lymph vessel tumor emboli and the invasive frontier area of lymph node metastasis of primary carcinoma of R_A/C < 1 (Figure 4G and 4H).

Figure 4.

Analysis of the invasive characterization of ALDH1A1^high cancer cells. Determination of the frequency of ALDH1A1^high cells in both HT29 and SW480 cell lines by flow cytometry (A). The xenograft tumor derived from ALDH1A1^low cells had intact capsule relative to ALDH1A1^high cells of HT 29 and SW480 (B-E). ALDH1A1 high expression HT29 cells showed increased expression of β-catenin in the nucleus relative to low expression of ALDH1A1 of SW480 (F). The ALDH1A1 staining intensity was stronger in lymph vessel tumor emboli and the invasive frontier than that in the center of lymphnode metastasis tissue (G, H). The scale bar is 50 μm in (B-H).

Discussion

More sensitive and reliable indicators are required for the diagnosis of CRC and the accurate prediction of its prognosis. In this study, we found that R_A/C, the ratio of ALDH1A1 level in adjacent mucosa to that in tumor tissues, was closely related to invasion, metastasis and prognosis. Higher frequency of nuclear translocation of β-catenin may contribute to the increased potentials of invasion and metastasis in R_A/C < 1 group. Our results provide a novel biomarker R_A/C for predicting metastasis and prognosis of CRC.

The expression of ALDH1A1 in normal crypts and colorectal carcinoma tissues has been previously investigated. Some researchers have indicated that cells with ALDH1A1 expression were sparse and limited to the bottom of the normal crypts, where the stem cells [10] or the proliferative cells reside [15]. During the progression from normal epithelium to mutant epithelium (adenomatous polyposis coli, APC), adenoma and CRC, the number of cells with high ALDH1A1 expression increased and became distributed further up the tissues [12]. However, other studies have suggested that this pattern may only represent one form of the ALDH1A1 expression pattern in normal, adjacent and malignant colon tissues. In normal mucosa or adjacent tissues, weakly ALDH1A1 positive cells could be found in nearly all crypts [22]. The cytoplasmic staining of ALDH1A1 was also observed in the cells located at the bottom of the crypts, as well as in many of the partially differentiated absorptive and secretory cells at higher crypt levels [14]. Christoph et al. even observed the nuclear expression of ALDH1A1 throughout all sections of the colon crypt [16]. These results indicate that ALDH1A1 expression in normal colorectal mucosa and the adjacent tissues is heterogeneous and does not cluster; instead, it is evenly distributed [12].

IHC staining of ALDH1A1 in CRC tissues also displayed a heterogeneous expression pattern, with differences in the rate and intensity of positivity [12,14,22]. The cytoplasmic expression pattern of ALDH1A1 ranges from no staining at all in some cases, to intense positive staining in others. A tissue microarray containing 1287 CRC samples was immunostained for ALDH1A1 by Lugli et al. They found that 987 cases (76.7%) were negative for cytoplasmic ALDH1A1 expression in tumor tissue. The overexpression of this protein, which was defined as > 25% of positive cells, was only observed in the remaining 300 cases (23.2%) [22]. Similar results were obtained by other groups, who showed that the expression of ALDH1A1 was negative in some cases, and positive in 30-90% of CRC samples [12,20]. Among these cases, the staining intensity was highly variable, ranging from no staining to low and strong. Only approximately 20-50% tumor cells were positive for cytoplasmic ALDH1A1 expression [12,20]. This heterogeneous pattern of ALDH1A1 expression was also observed in our study of 20 cases of normal colorectal mucosa and 65 cases of CRC and their corresponding adjacent tissues. More interesting, we found for the first time that, when the expression of ALDH1A1 was assessed in cancer tissues and their adjacent colorectal mucosa, ALDH1A1 was up-regulated (R_A/C < 1) in some cases but down-regulated (R_A/C ≥ 1) in others.

It has been reported by different groups that a high expression of ALDH1A1 in tumor tissues is associated with poor outcomes in breast, lung, pancreatic, bladder, ovarian and prostate cancer [22]. However, the prognostic value of ALDH1A1 expression in CRC tissue remains controversial, despite of an association between the increased expression level of ALDH1A1 in CRC tissues and some clinical pathology parameters, such as tumor grade, stage and metastases [15,22,24]. For example, the results from the study by Lugli et al. in 1287 CRC cases indicated that the expression of ALDH1A1 in CRC cancer was not related to differences in survival time [23]. Increased ALDH1A1 showed a trend towards decreased OS without statistical significance [24]. Results from our group also indicated that the expression of ALDH1A1 in tumor tissue alone could not be used as a predictor for the prognosis of CRC patients (Data not shown). However, if the corresponding adjacent mucosa in addition to the tumor tissue was included, and CRC patients were divided into two groups based on their ALDH1A1 expression patterns, we could observed a significant difference in the pathology features and outcomes. Interestingly, Christoph et al. found that a small subset of colon cancer samples, which were positive for the nuclear expression of ALDH1A1, displayed a significantly reduced overall survival and disease-free survival as compared to patients without nuclear staining. Considering well-known prognostic markers in colorectal cancer, such as age, TNM status and treatment, theirs multivariate analysis using the Cox regression model confirmed that the nuclear expression of ALDH1A1 was a prognostic marker for overall survival in colon cancer [16]. Of note, we did not observe any samples that were positive for the nuclear expression of ALDH1A1, which may be due to an insufficient number of cases in our study.

A crucial observation of our study is the significant difference in the β-catenin nuclear translocation between the two groups of CRC patients as divided by R_A/C, which may be due to the cross-talk between the retinoic acid (RA) and the Wnt/β-catenin signaling pathways. ALDH1A1 is responsible for the synthesis of RA from retinal and is crucial in regulating RA signaling [25]. The lipophilic RA generated by ALDHs can function in a paracrine or an endocrine manner by binding to the RA receptor (RAR) and the retinoid x receptor (RXR) to initiate transcription of target genes. These genes thereby regulate a variety of biological processes such as proliferation, differentiation, and apoptosis. The activity of RA signaling is closely correlated with tumor grade, treatment response and prognosis of CRC patients [26]. The Wnt/β-catenin is another important pathway which can profoundly influence the origin, development and survival outcome of CRC. The nuclear translocation of β-catenin is one of the hallmark events required for the activation of this signaling pathway and is a significant predictor of poor survival in the majority of sporadic CRC cases. Interestingly, the cross-talk between these two pathways has been previously reported [27-33]. β-catenin pathway can positively or negatively influence the function of RA signaling, thereby affecting important biological and pathological processes in a variety of cancers, including CRC. Conversely, RA signaling also influences cell proliferation through the activation of growth-stimulatory signals mediated by Wnt/β-catenin signaling [32]. For example, RAR can regulate Wnt/β-catenin signaling positively or negatively, depending on the availability of retinoid ligands [35]. RAR has been found to inhibit gene transcription mediated by β-catenin [36,37] but be reversed after the administration of a RA agonist [35]. RXR agonists reduce the activation of transcription mediated by β-catenin and inhibit cell proliferation through a protein degradation mechanism. The inhibition of Wnt/β-catenin signaling by a TCF4 dominant negative construct can also decrease the activity of ALDH in breast cancer cells [31]. Our results further demonstrate the close interplay between these two signaling pathways, which may profoundly influence the pathologies of CRC and be used as a potential target for therapy.

In summary, our results indicate for the first time that the two distinct expression patterns of ALDH1A1 in CRC and its adjacent tissue are associated with different clinicopathological features and prognosis in CRC patients. Additional prospective studies are warranted to explore whether the nuclear translocation of β-catenin provides a molecular mechanism for these observations.

Disclosure of conflict of interest

None.