PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of moloncolLink to Publisher's site
 
Mol Oncol. 2015 January; 9(1): 17–31.
Published online 2014 July 24. doi:  10.1016/j.molonc.2014.07.010
PMCID: PMC5528683

Aldehyde dehydrogenase 1A3 influences breast cancer progression via differential retinoic acid signaling

Abstract

Aldehyde dehydrogenase (ALDH) 1A enzymes produce retinoic acid (RA), a transcription induction molecule. To investigate if ALDH1A1 or ALDH1A3‐mediated RA signaling has an active role in breast cancer tumorigenesis, we performed gene expression and tumor xenograft studies. Analysis of breast patient tumors revealed that high levels of ALDH1A3 correlated with expression of RA‐inducible genes with retinoic acid response elements (RAREs), poorer patient survival and triple‐negative breast cancers. This suggests a potential link between ALDH1A3 expression and RA signaling especially in aggressive and/or triple‐negative breast cancers. In MDA‐MB‐231, MDA‐MB‐468 and MDA‐MB‐435 cells, ALDH1A3 and RA increased expression of RA‐inducible genes. Interestingly, ALDH1A3 had opposing effects in tumor xenografts, increasing tumor growth and metastasis of MDA‐MB‐231 and MDA‐MB‐435 cells, but decreasing tumor growth of MDA‐MB‐468 cells. Exogenous RA replaced ALDH1A3 in inducing the same opposing tumor growth and metastasis effects, suggesting that ALDH1A3 mediates these effects by promoting RA signaling. Genome expression analysis revealed that ALDH1A3 induced largely divergent gene expression in MDA‐MB‐231 and MDA‐MB‐468 cells which likely resulted in the opposing tumor growth effects. Treatment with DNA methylation inhibitor 5‐aza‐2'deoxycytidine restored uniform RA‐inducibility of RARE‐containing HOXA1 and MUC4 in MDA‐MB‐231 and MDA‐MB‐468 cells, suggesting that differences in epigenetic modifications contribute to differential ALDH1A3/RA‐induced gene expression in breast cancer. In summary, ALDH1A3 induces differential RA signaling in breast cancer cells which affects the rate of breast cancer progression.

Keywords: ALDH1A3, ALDH1A1, Breast cancer, Triple-negative, Retinoic acid

Highlights

  • RA‐inducible gene expression in breast cancer is dependent on ALDH1A3 expression.
  • Triple‐negative breast cancer patients have higher expression of ALDH1A3.
  • ALDH1A3 determines breast tumor progression by inducing RA signaling.
  • Opposing tumor growth effects of ALDH1A3/RA is due to differential gene expression.
  • DNA methylation contributes to differential gene expression induced by ALDH1A3/RA.

Abbreviations

ALDH
aldehyde dehydrogenase
ATRA
all‐trans‐retinoic acid
AZA
5‐aza‐2'deoxycytidine
CSCs
cancer stem cells
HOXA1
homeobox transcription factor A1
MSP
methylation‐specific PCR
MUC4
mucin 4
PPAR/β/δ
peroxisome proliferator activated receptor beta/delta
PPRE
peroxisome proliferator response element
RA
retinoic acid
RAR
retinoic acid receptor
RARβ
retinoic acid receptor beta
RARRES1
retinoic acid receptor responder 1
RXR
retinoid X receptor
RARE
retinoic acid response element

1. Introduction

Aldehyde dehydrogenases (ALDHs) are a family of evolutionarily conserved enzymes (19 isoforms expressed in humans) responsible for oxidizing aldehydes to carboxylic acids (Black and Vasiliou, 2009; Marchitti et al., 2008). Aldehydes are generated by metabolic processes (e.g. lipid perioxidation) and their detoxification is mediated by the ALDHs. A few isoforms (ALDH1A1, ALDH1A2, ALDH1A3) also function in retinoic acid (RA) cell signaling by oxidizing vitamin A metabolite retinal to RA (Black and Vasiliou, 2009; Penzes et al., 1997; Rexer et al., 2001; Zhao et al., 1996).

Once produced in the cytoplasm, RA translocates to the nucleus where it activates nuclear receptors, retinoic acid receptors (RARs) and retinoid X receptors (RXRs) and regulates expression of genes with retinoic acid response element (RARE) sequences in their promoters (McGrane, 2007). In 2002, Balmer and Blomhoff compiled a list of over 500 genes whose expression is upregulated or downregulated by RA (Balmer and Blomhoff, 2002). Therefore, RA can modulate a wide variety of biological processes, including differentiation, apoptosis, cell cycle arrest and cell proliferation (Balmer and Blomhoff, 2002; McGrane, 2007; Tang and Gudas, 2011). One proposed mechanism for RA‐induced cell proliferation versus cell cycle arrest is predominant activation of alternative nuclear hormone receptors peroxisome proliferator activated receptor beta/delta (PPAR/β/δ) over RARs. Dominant activation of RAR/RXR and transcription of genes with RAREs typically leads to growth suppression. In contrast, dominant activation of PPAR/β/δ/RXR and transcription of genes with peroxisome proliferator response elements (PPREs) leads to cell proliferation (Schug et al., 2007, 2008). An additional alternative mechanism for the differential growth effects of RA is epigenetic silencing of RA‐inducible tumor suppressors (Tang and Gudas, 2011). Therefore, the cellular context in which RA signaling occurs has a major effect on the cellular outcomes induced by RA.

Of the RA‐producing ALDH1A enzymes, increasing evidence suggests a potential role for ALDH1A1 and ALDH1A3 in cancer. First, increased expression of both ALDH1A1 and ALDH1A3 (but not ALDH1A2) has been reported in Aldefluor‐positive‐identified cancer stem cell (CSC) populations of breast cancer and melanoma (Ginestier et al., 2007; Luo et al., 2012; Marcato et al., 2011). ALDH1A1 expression is often associated with worse prognosis in cancers, such as breast, prostate and lung (Charafe‐Jauffret et al., 2010; Ginestier et al., 2007; Khoury et al., 2012; Li et al., 2010; Li et al., 2012; Morimoto et al., 2009; Neumeister et al., 2010). Additionally, ALDH1A1 expression may contribute to chemotherapeutic resistance in cancer (Moreb et al., 2007; Sladek et al., 2002); however, it has not yet been functionally linked to cancer progression. High ALDH1A3 levels have also been associated with more aggressive forms of breast, glioblastoma, gall bladder cancer and pancreatic cancer (Jia et al., 2013; Mao et al., 2013; Marcato, et al., 2011; Yang et al., 2013; Zhang et al., 2013). Importantly, there is some evidence suggesting that ALDH1A3 may have functional relevance in cancer progression, with ALDH1A3 knockdown in a melanoma cell line resulting in reduced tumor growth (Luo et al., 2012). Similarly, ALDH1A3 knockdown in mesenchymal glioblastoma cells reduced their growth (Mao et al., 2013).

Therefore, we hypothesized that ALDH1A1 and/or ALDH1A3 may have a functional role in breast cancer. Furthermore, since these enzymes are known to produce RA, we wondered if RA signaling in breast cancer is dependent upon expression of either ALDH1A enzyme. Analysis of patient tumor gene expression data revealed that ALDH1A3 expression correlates significantly with expression of RA‐inducible genes, poorer survival and triple‐negative breast cancers. In breast cancer cell line experiments, expression of RA‐inducible genes was dependent upon ALDH1A3 expression and ALDH1A3/RA affected tumor growth and metastasis. Interestingly, depending on the breast cancer cell line, ALDH1A3/RA induced either tumor progression or suppression effects, which at least in part, likely depend upon differential epigenetic modifications of RA‐inducible genes. In summary, these findings link expression of ALDH1A3 with differential RA signaling and breast cancer progression.

2. Materials and methods

2.1. Cohort of patients and gene expression microarray analysis

176 treatment‐naïve primary breast cancer samples were obtained through the CBCF Tumor Bank and used for gene expression microarray. All 176 frozen dissected tumor samples had been banked by board certified breast cancer pathologists who also performed quality assurance reports (histological and cellular composition quantifications of the tumors) of the clinical tissue samples directly adjacent to (en face of) the banked tissue samples. The clinical tissues were oriented in the formalin fixed paraffin embedded (FFPE) blocks so that when the planes of the clinical tissues are sectioned, they are continuous with the banked samples. The clinical tissues were subjected to routine fixation and hematoxylin and eosin (H&E) staining. Cancer cells, stromal cells and normal cells were differentiated from each other based on two main parameters: (1) location, with cancer cells forming clusters, stromal cells usually flowing between the clusters and normal cells forming glands and ducts, and (2) cytology, with cancer cells having a similar morphology, and resembling low or high‐grade epithelial‐like cells in places outside of the normal ductal system. Samples were selected upfront for majority malignant cellularity (most had more than 70%). Average percent of area with cells in our microarray tumor tissue samples is 72% and average percent of tumor cells is 93.38% (the average non‐cancer cell component is less than 7%).

Patient material and clinical information was collected under Research Ethics Board Protocol ETH‐02‐86‐17. Patients received standardized guideline‐based chemo‐ and hormonal‐therapies. The 176 patients consisted of 88 patients who experienced early relapse (<5 years after the initial treatment) and 88 patients who had not relapsed as of September 30, 2009. The patient data is summarized in Supplementary Table 1. The median follow‐up time for surviving patients was 4.5 years. RNA isolation from frozen human breast tumor samples, gene microarray analysis (accession GSE22820) and data processing were as previously described (Liu et al., 2011a). The mRNA levels were estimated based on normalized gene microarray signal intensity. The cut‐off values to define “high” and “low” levels for each gene were determined with receiver operating characteristic (ROC) curve analysis.

2.2. Cell lines

All cell lines were cultured in Dulbecco's Modified Eagle's Medium (DMEM), 10% fetal bovine serum (FBS, Invitrogen) and were obtained from the American Type Culture Collection (ATCC). DDC Medical authenticated the cell lines by short tandem repeat (STR) profiling at 17 loci and verified them to be mycoplasma‐negative (most recently performed within six months of manuscript submission). We utilized the same ALDH isoform knockdowns generated and validated in our prior publication (ALDH1A3 shRNA 1, ALDH1A3 shRNA 3 and ALDH1A1 shRNA 1) (Marcato et al., 2011). The overexpression clones were generated using retroviral vector pMSCVpuro with either ALDH1A1 or ALDH1A3 coding sequences inserted. Retroviral supernatants were generated as previously described (Marcato et al., 2011) and were applied to cultured MDA‐MB‐468, MDA‐MB‐231 or MDA‐MB‐435 cells. Stable transfectants were selected with puromycin (Sigma Aldrich).

2.3. Quantitative PCR

RNA extracted with TRIzol and Purelink RNA kit (Invitrogen) was reverse transcribed with Superscript II reverse transcriptase kit (Invitrogen) as per manufacturer's instructions. GoTaq RT‐PCR kit (Promega) with gene‐specific primers (Supplementary Table 2) was used as per manufacturer's instructions. Standard curves were generated to incorporate primer efficiencies to calculate the relative level of mRNA compared to internal reference gene GAPDH. For RA treatment, 100 nM all‐trans‐retinoic acid (ATRA) was cultured with the cells for 18 h. For 5‐aza‐2'‐deoxycytidine (AZA) treatment, 1uM AZA was added for 3 days and 100 nM ATRA was added in the final 18 h.

2.4. RARE reporter assay

RARE luciferase reporter gene plasmid (Promega) was transfected into MDA‐MB‐231 cells using lipofectamine® 2000 reagent (Invitrogen) as per manufacturer's instructions. To RA treatment wells, 100 nM ATRA was cultured with the cells 24 h post transfection. Forty‐eight h post transfection, luciferase activity was quantified as per the manufacturer's instructions using a GloMax®‐20/20 luminometer (Promega).

2.5. Mouse studies

All experiments followed guidelines set by the Canadian Council on Animal Care and were performed according to a protocol approved by Dalhousie University Committee on Laboratory Animals (protocol 10–139). Eight to ten week old NOD/SCID female mice were orthotopically injected in mammary fat pad four with 2 × 106 cells of MDA‐MB‐231, MDA‐MB‐468 or MDA‐MB‐435 cells (knockdown and overexpression clones). Injected cells were mixed in 1:1 ratio with high concentration matrigel (BD bioscience). Alternatively, one week prior to injecting cells in the mice, 15 mg 90‐day slow‐release ATRA pellets (Innovative Research of America) were surgically implanted subcutaneously. Primary tumor growth was quantified (length × width × height/2) and the lungs were harvested, fixed, paraffin embedded and 5 μM thin sections generated for metastasis visualization by H&E staining. Metastasis quantification of each lung was performed blinded using a standardized grid imposed on Axiocam HRC. Color images were captured of at least two random H&E stained thin sections per tissue.

2.6. Invasion assay

Equal numbers of cells were seeded in serum‐free DMEM in transwells with 8 μM pores (BD Biosciences), with or without matrigel coating, with the lower well containing DMEM with 10% FBS. After 18 h, the invaded/migrated cells present on the underside of the transwell membrane were H&E stained and visualized by microscopy. Percentage of invaded cells was calculated as per manufacturer's instructions (mean number of cells invading through matrigel membrane divided by the mean number of cells migrating through uncoated membrane).

2.7. Gene expression microarray

Genome expression microarray analysis was performed by Ambry Genetics. RNA samples with spike‐in controls were converted to cDNA and transcribed for amplification and incorporation of labeled CTP (cRNA). Purified cRNA samples were fragmented and hybridized to an Agilent SurePrint G3 8 × 60K Gene Expression microarray v2 slide and scanned with an Agilent G2565CA High Resolution Scanner. Data was processed with Agilent's Feature Extraction software version 11.0.1.1 using the protocol GE1_1100_Jul11 and 8 × 60K grid file 039494_D_F_20120628. Samples were normalized using the percentile shift algorithm with no baseline transformation (microarray data available at accession GSE49322, MSCV = MDA‐MB‐231 vector control, 1A3 = MDA‐MB‐231 ALDH1A3 overexpression, SMP = MDA‐MB‐468 shRNA control, 61 = MDA‐MB‐468 ALDH1A3 shRNA 1).

2.8. Statistical analysis

All statistical analyses for the preclinical data were performed with GraphPad Prism Version 4. Unpaired or paired t‐tests were performed when two comparisons were made. ANOVA (one‐way analysis of variance or repeated measures) was performed followed by post‐tests Dunnett or Bonferroni (specified in the figure legends), when multiple comparisons were made. For all patient tumor data, statistical analyses were performed using MedCalc version 12.4.0.0 (MedCalc Software). Gene expression microarray signal intensity values of each gene analyzed were log‐transformed to better fit the normal distribution assumption. Student's t‐test was employed to test the significance of the associations between the mRNA levels of tested genes and clinicopathologic parameters. Significance of ALDH1A3, ALDH1A1 and RARRES1 mRNA levels on patient survival was analyzed using Kaplan–Meier log‐rank test. p values are represented as follows: * <0.05, ** <0.01, *** <0.001.

3. Results

3.1. ALDH1A3 expression in patient tumors correlates with expression of RA‐inducible RARβ and RARRES1, patient survival and triple‐negative breast cancers

To determine if expression of the ALDH1A enzymes correlates with expression of RA target genes in breast cancer we examined the levels of ALDH1A1, ALDH1A2 and ALDH1A3 in 176 breast cancer patient tumors (clinical data summarized in Supplementary Table 1, accession GSE22820) with the expression of previously described RA‐inducible genes containing RAREs in their promoters. Patient tumors with high ALDH1A3 levels had significantly higher levels of RARRES1 and RARβ (Figure 1A, B). Patient tumors with high ALDH1A1 levels had significantly higher levels of RARβ and CYP26A (Figure 1C). This suggests that both ALDH1A3 and ALDH1A1 may regulate the expression of particular RA‐inducible genes in patient breast tumors. It is notable that ALDH1A2 was expressed below background threshold levels in the breast cancer patient samples (accession GSE22820), not allowing for analyses with this isoform. However, the low levels of ALDH1A2 suggest it does not play an important role in breast cancer.

Figure 1

ALDH1A3 and ALDH1A1 expression in breast cancer patient tumors is associated with expression of RA‐inducible genes and patient outcomes. Correlation of the mRNA levels of ALDH1A3 (A and B) and ALDH1A1 (C) with RA‐inducible genes in 176‐patient ...

Given the predominant correlation of ALDH1A3 with RARRES1 in patient tumors, we included RARRES1 in our next set of analyses where we examined the expression of these genes with patient outcomes (Figure 1D). Patients with high ALDH1A3 had significantly poorer survival than patients with low ALDH1A3. RARRES1 expression correlated even more significantly with patient survival. Notably, patients with high levels of both ALDH1A3 and RARRES1 in their tumors (double high) had the worst survival (Figure 1D). To determine the contribution of each gene in poorer survival, we performed a multivariate survival analysis (Figure 1E). Although the overall regression model is statistically significant (p = 0.0011), the regression component of ALDH1A3 as a covariate is not (p = 0.2059). This indicates that the contribution of ALDH1A3 in poorer patient survival is not independent of RARRES1. ALDH1A1 expression correlated with poorer survival; however, the trend was not statistically significant (Figure 1D).

Furthermore, ALDH1A3 and RARRES1 (but not ALDH1A1) correlated with more aggressive triple‐negative (ER, PR, HER2) breast cancers (Table 1). We also noted that RARRES1 expression correlated with higher Ki67 positivity and grade, which may be related to its highly significant association with triple‐negative breast cancers. In contrast, ALDH1A1 correlated with increased local metastasis but lower Ki67 positivity and grade.

Table 1

Clinicopathologic associations of ALDH1A1, ALDH1A3 and RARRES1 mRNA expression levels in breast cancer.

Taken together, the patient tumor data suggests two possibilities: 1) the associated poorer survival of patients with high ALDH1A3 (and/or RARRES1) may be a consequence of these genes being highly expressed in more aggressive triple‐negative breast cancers, or 2) the rate of breast cancer progression may be influenced by ALDH1A3 expression and increased RA signaling.

To discern between these two possibilities we first tested if expression of these RA‐inducible genes in breast cancer is dependent upon ALDH1A3 expression. For this purpose, we employed commonly used triple‐negative cell lines, which we previously characterized to have low and high endogenous levels of the enzymes (Marcato et al., 2011). We used our previously constructed and validated ALDH1A1 and ALDH1A3 knockdowns (Marcato et al., 2011) and newly generated overexpression clones (Supplementary Figure 1).

3.2. ALDH1A3 and RA induce expression of RARβ and RARRES1 in MDA‐MB‐231, MDA‐MB‐435 and MDA‐MB‐468 cells

In MDA‐MB‐231 cells (that express low levels of ALDH1A1 and ALDH1A3 (Marcato et al., 2011)), overexpression of ALDH1A3 (but not ALDH1A1) induced significant expression of RARE‐bearing genes RARβ and RARRES1 (Figure 2A). This is consistent with the patient tumor data where high ALDH1A3 levels correlated with high levels of both these genes in patient tumors. As expected, knockdown of ALDH1A1 and ALDH1A3, which are expressed at low levels in MDA‐MB‐231 cells (Marcato et al., 2011), did not affect expression of the RA‐inducible genes (Figure 2B). Addition of all‐trans‐retinoic acid (ATRA), which bypasses the need for ALDH‐mediated retinoid processing, increased expression of RARβ and RARRES1 in MDA‐MB‐231 cells and confirmed their RA‐dependent expression (Figure 2C).

Figure 2

ALDH1A3 and RA induce expression of RA–inducible genes in the breast cancer cell lines. Relative mRNA expression levels of RA‐inducible genes quantified by QPCR, in MDA‐MB‐231 and MDA‐MB‐468 ALDH1A overexpression ...

Unsurprisingly, in MDA‐MB‐468 cells, with high endogenous levels of ALDH1A1 and ALDH1A3 (Marcato et al., 2011), overexpression did not increase expression of the RA‐inducible genes (Figure 2A). However, knockdown of ALDH1A3 resulted in significantly decreased RARβ and RARRES1 expression (Figure 2B). ALDH1A1 knockdown did not have a significant effect on expression of the RA‐inducible genes. Addition of ATRA to MDA‐MB‐468 ALDH1A3 knockdown shRNA 1 and 3 cells increased expression of RARβ and RARRES1 (Figure 2D). Similarly, knockdown of ALDH1A3 in MDA‐MB‐435 cells (which possess high levels of endogenous ALDH1A3 (Marcato et al., 2011)) significantly decreased RARβ and RARRES1 expression (Supplementary Figure 2). It is noteworthy that the tissue‐of‐origin of MDA‐MB‐435 cells is controversial; however, increasing evidence suggests they are triple‐negative breast cancer cells as originally represented (Bachmeier et al., 2008; Chambers, 2009; Hollestelle and Schutte, 2009; Nerlich and Bachmeier, 2013).

To confirm that ALDH1A3/RA activates RAR/RXR in the cells, we performed a reporter assay. ALDH1A3 overexpression and ATRA induced luciferase activity in a RARE reporter assay performed with MDA‐MB‐231 cells (Figure 2E). Taken together, our results are consistent with the notion that ALDH1A3 induces expression of RA‐inducible genes (increases RA signaling capacity) in the breast cancer cells.

3.3. High levels of ALDH1A3 and RA increase tumor progression of MDA‐MB‐231 and MDA‐MB‐435 cells

Next, we assessed if ALDH1A3 or ALDH1A1 contribute to tumor growth and/or metastasis in breast cancer. The low endogenous expression of ALDH1A1 and ALDH1A3 in the highly tumorigenic/metastatic MDA‐MB‐231 cells (Marcato et al., 2011) suggests that neither enzyme is likely a critical tumorigenicity factor in the breast cancer cell line. Unsurprisingly, ALDH1A1 or ALDH1A3 knockdown did not significantly affect tumor growth of MDA‐MB‐231 cells (Figure 3A; tumor weights, Supplementary Figure 3). However, ALDH1A3 overexpression (but not ALDH1A1) in MDA‐MB‐231 increased tumor growth significantly (Figure 3A; tumor weights, Supplementary Figure 3). Furthermore, mice bearing the larger MDA‐MB‐231 overexpressing ALDH1A3 tumors also had significantly increased metastasis (Figure 3B; images of lung metastasis, Supplementary Figure 4). To investigate if the increased metastasis was due to increased seeding by larger tumors, or at least in part due to increased metastatic capacity of ALDH1A3‐overexpressing MDA‐MB‐231 cells, we first performed a correlation analysis of the individual mouse tumor weights versus percentage lung metastasis (Supplementary Figure 5A). Tumor size correlated positively with metastasis for MDA‐MB‐231 vector control cells, but did not for MDA‐MB‐231 ALDH1A3 overexpression cells. Furthermore, when the lung metastasis is divided by the tumor weight, the lung metastasis of mice implanted with ALDH1A3 overexpression cells is significantly greater than the lung metastasis of mice implanted with vector control cells (Supplementary Figure 5B). This suggests that the increased metastasis of mice bearing MDA‐MB‐231 ALDH1A3 overexpression tumors is not just a consequence of larger tumor size. To investigate this further, we performed a transwell invasion assay. ALDH1A3 overexpression increased invasion of MDA‐MB‐231 cells, suggesting that ALDH1A3‐overexpressing cells have increased metastatic capacity (Figure 3C). Similarly, in MDA‐MB‐435 cells with high endogenous ALDH1A3 levels (Marcato et al., 2011), ALDH1A3 knockdown reduced their tumor growth significantly (Figure 3D; tumor weights, Supplementary Figure 3).

Figure 3

ALDH1A3 expression and RA treatment increases tumor progression of MDA‐MB‐231 and MDA‐MB‐435 cells. (A) ALDH1A3 overexpression increases tumor volume in mice injected with MDA‐MB‐231 cells (n = 11–13), ...

We next investigated if ALDH1A3‐mediated tumor growth and metastasis effects of MDA‐MB‐231 breast cancer cells are related to its RA‐production function. We hypothesized that if the tumor growth and metastasis effects induced by ALDH1A3 in MDA‐MB‐231 cells are due to RA production by ALDH1A3, an exogenous source of RA should also induce similar tumor growth and metastasis effects as ALDH1A3 overexpression in MDA‐MB‐231 cells. To test this hypothesis, we performed tumor xenotransplantation experiments in mice with or without implantation of a slow‐release ATRA pellet as an exogenous source of RA. In mice implanted with MDA‐MB‐231 cells, ATRA increased tumor growth and metastasis in the absence of ALDH1A3 overexpression (day 50, Figure 3E, F; tumor weights, Supplementary Figure 3; images of lung metastasis, Supplementary Figure 6).

It is noteworthy that ATRA treatment exacerbated tumor growth and metastasis beyond the effect of ALDH1A3 overexpression alone (day 57, Figure 3A, C; compared to day 43, Figure 3E, F). The increased metastatic burden and morbidity of ATRA treatment required earlier termination of the experiments. In comparing the mice treated with ATRA (Figure 3F), ALDH1A3 overexpression further augments metastasis, as mice implanted with ALDH1A3 overexpression cells and ATRA pellet had comparable metastatic burden to mice implanted with the ATRA pellet alone 7 days earlier (day 43, 33.2%; versus day 50, 34.1%). The average lung metastasis from these experiments is summarized in Figure 3G. Finally, mice bearing ALDH1A3 overexpression tumors (no ATRA treatment, Figure 3F, right) have an average of 0.5% metastatic lung surface area on day 43. Although 0.5% may seem insignificant, in terms of absolute cell numbers, 0.5% metastatic lung tissue likely represents a high number of cancer cells that have infiltrated the lungs by day 43. Given the doubling rates of cancer cells, it is reasonable that 0.5% metastatic cancer tissue in the lungs could grow to 59.3% if mice are sacrificed 14 days later (ALDH1A3 overexpression, day 57, Figure 3B).

3.4. ALDH1A3 and RA suppress tumor growth in MDA‐MB‐468 cells

To further investigate the effects of ALDH1A3 and RA on breast cancer cells we performed a similar experiment with MDA‐MB‐468 cells. As expected, ALDH1A1 and ALDH1A3 overexpression did not significantly affect the tumor growth of the implanted MDA‐MB‐468 cells with endogenous high ALDH1As levels (Marcato et al., 2011). In contrast to the effects seen in MDA‐MB‐231 and MDA‐MB‐435 cells, ALDH1A3 knockdown in MDA‐MB‐468 cells significantly increased tumor growth (Figure 4A; tumor weights, Supplementary Figure 7). A second shRNA ALDH1A3 knockdown (shRNA 3) also resulted in increased tumor growth of MDA‐MB‐468 cells, ruling out the possibility that this is due to off‐target effects (Supplementary Figure 7). We hypothesized that if ALDH1A3‐mediated tumor suppression in MDA‐MB‐468 cells is related to its RA signaling function, an exogenous source of RA should suppress the increased tumor growth observed in ALDH1A3 knockdown cells.

Figure 4

ALDH1A3 and RA suppress tumor growth of implanted MDA‐MB‐468 cells. (A) Tumor volume in mice injected with MDA‐MB‐468 vector control, ALDH1A1 overexpression and ALDH1A3 overexpression cells (n = 13–15), ...

To test this hypothesis, we performed tumor xenotransplantation experiments in mice with ATRA treatment. ATRA inhibited the increased tumor growth resulting from ALDH1A3 knockdown in the cells (Figure 4B; tumor weights, Supplementary Figure 7). Thus, RA compensates for the loss of ALDH1A3 in suppressing tumor growth of MDA‐MB‐468 cells.

3.5. ALDH1A3 and RA induce differential gene expression in MDA‐MB‐468 and MDA‐MB‐231 cells

We next wondered if the opposing tumor growth effects of ALDH1A3/RA in breast cancer cells depend upon which nuclear receptors are dominantly activated by RA (Schug et al., 2007, 2008). We hypothesized that the decreased tumor growth induced by ALDH1A3/RA in MDA‐MB‐468 cells is due to dominant activation of RAR/RXR and transcription of genes with RAREs, while the increased tumor growth induced by ALDH1A3/RA in MDA‐MB‐231 cells is due to dominant activation of PPARβ/δ/RXR and transcription of genes with PPREs.

We quantified the expression of several RA‐inducible genes with PPRE sequences previously used as indicators of predominant PPARβ/δ/RXR activation by RA (Schug et al., 2007, 2008) in MDA‐MB‐231 cells and MDA‐MB‐468 cells. However, ALDH1A3 failed to induce expression changes in these genes (Supplementary Figure 8). Furthermore, ALDH1A3 and RA failed to induce luciferase activity in a PPRE reporter assay (Supplementary Figure 8).

This raises two possibilities. First, it is possible other genes with PPREs are being induced by ALDH1A3/RA and that the PPRE reporter assay failed to reflect PPARβ/δ/RXR activation in MDA‐MB‐231 cells. Alternatively, differential expression of genes with RAREs or genes indirectly induced by RA signaling events may be responsible for the opposing tumor growth effects observed.

To discern between the two possibilities, we performed comparative genome expression analyses on MDA‐MB‐231 cells (with or without ALDH1A3 overexpression) and MDA‐MB‐468 cells (with or without ALDH1A3 knockdown). Increased ALDH1A3 expression upregulated 1286 and 1358 genes in MDA‐MB‐231 and MDA‐MB‐468 cells, respectively (Figure 5A, GSE49322, Supplementary File 1). Only 121 genes were upregulated in both cell lines. In validation of the microarray gene expression data we noted that RARβ and RARRES1 were among the 121 common upregulated genes. The downregulated genes are summarized in Supplementary Figure 9, GSE49322 and Supplementary File 2. This data demonstrates a large divergence in gene expression changes induced by ALDH1A3 in the two cell lines.

Figure 5

ALDH1A3 and RA induce differential gene expression in MDA‐MB‐231 and MDA‐MB‐468 cells. (A) Microarray genome expression analysis of MDA‐MB‐231 vector control cells versus ALDH1A3 overexpression cells and ...

Genome‐wide analyses by Lalevée et al., and Lemay and Hwang identified 3249 genes with RAREs and 1085 genes with PPREs, respectively (Lalevee et al., 2011; Lemay and Hwang, 2006). Among the gene list generated by Lalevée et al. (Lalevee et al., 2011), we identified 112 and 114 genes with RAREs upregulated by ALDH1A3 in MDA‐MB‐231 and MDA‐MB‐468 cells, respectively (Figure 5B). To determine if preference exists for certain variations of RARE sequences in the cell lines, the RARE sequences of the upregulated genes (Lalevee et al., 2011) were used to generate transcription factor binding site sequence logos (Crooks et al., 2004; Schneider and Stephens, 1990). This analysis revealed similar RARE sequence logos in both cell lines (Figure 5C), suggesting that the differential gene expression in MDA‐MB‐231 and MDA‐MB‐468 cells is independent of preferences for certain RARE sequences.

We identified 19 and 24 genes with PPREs upregulated by ALDH1A3 in MDA‐MB‐231 and MDA‐MB‐468 cells, respectively (Figure 5B) (Lemay and Hwang, 2006). In summary, these analyses suggest similar transcriptional induction of genes with RAREs and PPREs in MDA‐MB‐231 and MDA‐MB‐468 cells and that the opposing tumor growth effects induced by ALDH1A3 in the cells is likely independent of dominant RAR/RXR over PPARβ/δ/RXR activation (and vice‐versa) in the breast cancer cells.

To verify the microarray genome expression analysis we performed QPCR on some of the RARE‐containing genes. Among the list of ALDH1A3‐induced genes in MDA‐MB‐468 cells is homeobox transcription factor A1 (HOXA1, Supplementary File 1). HOXA1, a transcription factor with a RARE sequence, was previously shown to be inducible by RA (Gudas and Wagner, 2011; Wardwell‐Ozgo et al., 2014). HOXA1 expression is significantly reduced by ALDH1A3 knockdown and induced by ATRA in MDA‐MB‐468 cells but is undetectable in MDA‐MB‐231 cells (Figure 5D).

Mucin 4 (MUC4), a potential oncogene with a RARE, is inducible by RA and associated with triple‐negative breast cancer (Chaturvedi et al., 2008; Mukhopadhyay et al., 2013), and it was among the list of ALDH1A3 upregulated genes in MDA‐MB‐231 cells. We verified by QPCR that MUC4 is significantly induced by ALDH1A3 and RA in MDA‐MB‐231 cells, but not in MDA‐MB‐468 cells (Figure 5D).

We also verified that expression of RARE‐bearing eyes absent homolog 2 (EYA2) and carcinoembryonic antigen‐related cell adhesion molecule 6 (CEACAM6) is dependent upon ALDH1A3 and RA in MDA‐MB‐468 cells (Figure 5D), but not in MDA‐MB‐231 cells. Their expression was below detection in MDA‐MB‐231 cells. Therefore, neither high ALDH1A3 expression nor abundant RA necessarily results in expression of genes with RAREs in breast cancer cells. This suggests that additional factors play key roles in determining the inducibility of a RARE‐bearing gene by ALDH1A3/RA.

3.6. DNA methylation contributes to differential expression of ALDH1A3/RA inducible genes in breast cancer cells

We wondered if differential epigenetic modifications such as gene silencing by DNA hypermethylation contribute to the divergent gene expression profiles induced by ALDH1A3 in the cell lines. To test the theory we treated the cells with DNA methyltransferase inhibitor AZA. AZA treatment not only restored expression of HOXA1 in MDA‐MB‐231 cells, but also restored HOXA1 inducibility by ALDH1A3 (Figure 6A) and ATRA (Figure 6B). Therefore, upon AZA treatment, HOXA1 is similarly inducible by ALDH1A3 and RA in MDA‐MB‐231 and MDA‐MB‐468 cells. Methylation‐specific PCR (MSP) confirmed that HOXA1 is hypermethylated in MDA‐MB‐231 cells, as previously reported (Novak et al., 2006), and hypomethylated in MDA‐MB‐468 cells (Supplementary Figure 10).

Figure 6

ALDH1A3/RA‐inducibility of genes with RAREs in MDA‐MB‐231 and MDA‐MB‐468 cells depends up their methylation levels. HOXA1 expression in MDA‐MB‐231 vector control and ALDH1A3 overexpression cells ...

Similarly, AZA treatment of MDA‐MB‐468 increased MUC4 levels, which was also reduced in ALDH1A3 knockdown cells (Figure 6C) and increased by ATRA treatment (Figure 6D). This suggests that MUC4 is hypermethylated in MDA‐MB‐468 cells and when DNA methylation is inhibited with AZA treatment, MUC4 becomes similarly inducible by ALDH1A3 and RA in MDA‐MB‐468 and MDA‐MB‐231 cells. MSP confirmed that MUC4 is hypermethylated in MDA‐MB‐468 cells (Supplementary Figure 10). Additionally, in the 176 archived panel of breast cancer patient tumor samples, HOXA1 and MUC4 expression levels correlated significantly with levels of ALDH1A3 (Figure 6E), reinforcing the notion that expression of RA‐inducible genes (RA signaling) in breast cancer is tied to ALDH1A3 expression.

It is important to note that after AZA treatment, expression of EYA2 and CEACAM6 in MDA‐MB‐231 cells remained below detection. Therefore, while DNA methylation contributes to the differential gene expression induced by ALDH1A3/RA (e.g. HOXA1 and MUC4), it is not the only mechanism and other transcription regulatory mechanisms (e.g. histone modifications) likely contribute to differential expression of ALDH1A3/RA inducible genes in breast cancer.

4. Discussion

Our analysis of breast cancer patient tumors and breast cancer cell lines revealed that increased ALDH1A3 expression is associated with worse prognosis and triple‐negative breast cancers, and that it actively influences tumor progression via initiating RA signaling. RA has a controversial role in cancer and is generally thought of as cancer suppressor (Tang and Gudas, 2011). For example, RA‐based therapies are used successfully to treat acute promyelocytic leukemia, by inducing differentiation of leukemia cells into granulocytes (Huang et al., 1988). However, attempts to use retinoids and dietary precursors to treat other cancers, including breast cancer, have had limited success (Bryan et al., 2011; Chiesa et al., 2007; Singletary et al., 2002; The Alpha‐Tocopherol Beta Carotene Cancer Prevention Study Group, 1994). This has led to studies into the mechanism of RA‐resistance and RA‐sensitivity in breast cancer, which theoretically may allow the selective use of retinoid therapies in patients who are deemed most likely to respond (Guarnaccia, 2010; Liu et al., 2011a; Schug et al., 2008).

We found that the effects of RA and ALDH1A3 were opposing in the breast cancer cell lines: tumor‐promoting in MDA‐MB‐231 and MDA‐MB‐435 cells, but tumor‐suppressive in MDA‐MB‐468 cells. Furthermore, the pre‐existing level of ALDH1A3 in the cell lines is not correlative with their established tumorigenic potential. In 1991, Zhang et al. performed comparative tumor xenograft assays with the three cells lines and reported that MDA‐MB‐435 > MDA‐MD‐231 > MDA‐MB‐468 in tumorigenic and metastatic potential (Zhang et al., 1991). In our 2011 study, we reported that MDA‐MB‐468 > MDA‐MB‐435 > MDA‐MB‐231 in ALDH1A3 expression and Aldefluor activity (Marcato et al., 2011). First, the lack of correlation between tumorigenicity and ALDH1A3 levels illustrates that ALDH1A3 expression does not predict the tumorigenic/metastatic potential of breast cancer cells, nor is it a necessary factor for tumorigenicity. Second, considering the gene expression and tumor xenograft experiments performed here, ALDH1A3 induces RA signaling, but it does not determine the downstream effect of that RA signaling (promoting in MDA‐MB‐231 and MDA‐MB‐435 cells, and inhibiting in MDA‐MB‐468 cells). Instead, it is likely that downstream factors determine if ALDH1A3/RA promotes tumor growth or suppresses it (i.e. divergent ALDH1A3‐induced gene expression as seen in MDA‐MB‐468 and MDA‐MB‐231 cells, Figure 5). It is plausible that among the large pool of differentially expressed genes there are effectors (e.g. tumor suppressors and oncogenes) that cause the opposing tumor growth effects induced by ALDH1A3. Identifying the key ALDH1A3/RA‐inducible genes that promote or suppress tumor growth is necessary to decipher the mechanism of the opposing effects of ALDH1A3/RA in breast cancer.

The DNA methylation inhibition experiments indicate that differential methylation contributes to divergent gene expression induced by ALDH1A3/RA in MDA‐MB‐468 and MDA‐MB‐231 cells. These differentially methylated genes include HOXA1 and MUC4; however, we do not know if HOXA1 and MUC4 are effectors of ALDH1A3/RA‐mediated breast tumor suppression or progression. In addition to being hypermethylated in MDA‐MB‐231 cells, HOXA1 is often hypermethylated in cancer, suggesting a tumor‐suppressive function (Chung et al., 2011; Kim et al., 2009; Makiyama et al., 2005; Park et al., 2012; Pilato et al., 2013). In contrast, MUC4 is typically hypomethylated in cancers and its expression is associated with more aggressive cancer (Chaturvedi et al., 2008; Mukhopadhyay et al., 2013; Yonezawa et al., 2008; Yonezawa et al., 2011; Zhu et al., 2011). MUC4 knockdown in MDA‐MB‐231 cells reduced their tumorigenic and metastatic properties (Mukhopadhyay et al., 2013), suggesting MUC4 may represent a gene that contributes to ALDH1A3/RA‐mediated tumor growth and metastasis of MDA‐MB‐231 cells (MUC4 is hypermethylated in MDA‐MB‐468 cells). Taken together, this suggests one potential mechanism whereby ALDH1A3/RA contributes to breast cancer progression or suppression depends upon the balance of RA‐inducible tumor suppressors or oncogenes that are predominately expressed or epigenetically silenced (Supplementary Figure 11).

ALDH1A1 and ALDH1A3 both oxidize retinal to RA, and yet in our experiments only expression of ALDH1A3 in the breast cancer cells induced significant changes in expression of RA‐inducible genes, tumor growth and metastasis. However, mice injected with MDA‐MB‐468 cells overexpressing ALDH1A1 did have smaller tumors (not significant, Figure 4). This indicates a potential overlap in function between the ALDH isoforms. We observed a similar overlap in function of these two isoforms in the Aldefluor assay; however, ALDH1A3 overexpression increased the Aldefluor activity of MDA‐MB‐231 to a greater extent than ALDH1A1 overexpression (Supplementary Figure 1). This is congruent to our previous findings in which ALDH1A3 knockdown reduced the Aldefluor activity of breast cancer cell lines with high Aldefluor activity (e.g. MDA‐MB‐468 cells), but ALDH1A1 knockdown did not (Marcato et al., 2011). In breast cancer patient tumor samples, both ALDH1A1 and ALDH1A3 (and other ALDH isoforms) are expressed at greater levels in Aldefluor‐positive patient‐derived tumor cells (Ginestier et al., 2007; Marcato et al., 2011); however, there is a more consistent correlation with ALDH1A3 expression (Marcato et al., 2011). Furthermore, biochemical analyses performed on ALDH1A3 by Sima et al., suggest that the catalytic activity of ALDH1A3 for all‐trans retinal may be 10‐fold higher than that of ALDH1A1 (Gagnon et al., 2003; Sima et al., 2009), which could explain why only ALDH1A3 expression induced significant changes in our gene expression and tumor xenograft assays.

It is possible that ALDH1A1, ALDH1A3 and RA signaling in the non‐cancer cell component of the tumor microenvironment plays a role in breast cancer progression. ALDH1A1 expression has been reported in the stroma of patient breast tumors (Kang et al., 2014; Resetkova et al., 2010). Furthermore, a potential role for RA signaling in breast tumor stroma was illustrated using a spontaneous mammary tumor mouse model, where stromal RARβ promoted mammary tumorigenesis, angiogenesis and the recruitment of inflammatory cells and myofibroblasts (Liu et al., 2011b). Our gene expression analysis of patient breast tumors does not separate cancer cells from non‐cancer cells (which represent less than 7% of the average of the total cells). Therefore, we cannot exclude the possibility that part of the detected ALDH1A1, ALDH1A3 and RA‐inducible gene expression is arising from the non‐cancer cell component of the tumor. However, in our breast cancer cell line experiments, ALDH1A3 expression changes originated from the cancer cells and it was sufficient to induce expression of RA‐inducible genes and mediate effects on tumor growth and metastasis. Systemic RA had similar effects. Determining if ALDH1A1 or ALDH1A3 expression originating from the non‐cancer cells of breast tumor is sufficient to affect tumor growth and metastasis would provide further information into the potential role of RA signaling in the breast tumor microenvironment.

ATRA treatment of breast cancer cell lines HCC1954, SUM149 and SUM159, inhibited in vitro mammosphere formation suggesting that ATRA may induce differentiation of CSCs (Ginestier et al., 2009). It is unknown how ATRA would affect the mammosphere formation of MDA‐MB‐231, MDA‐MB‐468 and MDA‐MB‐435 cells. Extrapolation of the observed tumor growth effects leads us to speculate that ATRA would inhibit mammosphere formation of MDA‐MB‐468 cells, but may increase mammosphere formation of MDA‐MB‐231 and MDA‐MB‐435 cells. A differential effect on mammosphere formation could be explained by divergent ALDH1A3/ATRA‐induced gene expression. However, these in vitro assays may not reflect the potential effect of ATRA on the expansion of breast CSC and non‐CSC in patient tumors. For this purpose, studies with ATRA treatment of breast cancer patient tumor xenografts would be more informative and could reveal new insights into CSC biology, breast cancer progression and the development of novel therapeutics.

In conclusion, this study details two completely divergent responses of cell line xenografts to ALDH1A3/RA signaling, which is likely the result of differential gene expression. Several clinical trials have identified patients who exhibit some response to retinoid‐based therapies in combination with standard of care (Bryan et al., 2011; Budd et al., 1998; Sutton et al., 1997); while cancers of the remaining patients continue to progress. Development of a gene signature which predicts response to retinoid treatment may lead to the successful application of retinoid‐based therapies in select patients.

Conflicts of interests

The authors have no conflicts of interests to declare.

Supporting information

The following are the supplementary data related to this article:

Supplementary data

Supplementary data

Supplementary data

Acknowledgments

This work was completed with funding from following sources. Grant funding awarded from the Canadian Breast Cancer Foundation (CBCF) Atlantic to PWKL, the Dalhousie Medical Research Foundation to PM and CAG, the Canadian Institutes of Health Research (MOP‐130304) to PM, and the CBCF Prairies/NWT to RG.

1. 

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.molonc.2014.07.010.

Notes

Marcato Paola, Dean Cheryl A., Liu Rong-Zong, Coyle Krysta M., Bydoun Moamen, Wallace Melissa, Clements Derek, Turner Colin, Mathenge Edward G., Gujar Shashi A., Giacomantonio Carman A., Mackey John R., Godbout Roseline, Lee Patrick W.K., (2015), Aldehyde dehydrogenase 1A3 influences breast cancer progression via differential retinoic acid signaling, Molecular Oncology, 9, doi: 10.1016/j.molonc.2014.07.010.

Contributor Information

Paola Marcato, ac.lad@otacram.aloap.

Patrick W.K. Lee, ac.lad@eel.kcirtap.

References

  • Bachmeier B.E., Nerlich A.G., Mirisola V., Jochum M., Pfeffer U., 2008. Lineage infidelity and expression of melanocytic markers in human breast cancer. Int. J. Oncol. 33, 1011–1015. [PubMed]
  • Balmer J.E., Blomhoff R., 2002. Gene expression regulation by retinoic acid. J. Lipid Res. 43, 1773–1808. [PubMed]
  • Black W., Vasiliou V., 2009. The aldehyde dehydrogenase gene superfamily resource center. Hum. Genomics. 4, 136–142. [PubMed]
  • Bryan M., Pulte E.D., Toomey K.C., Pliner L., Pavlick A.C., Saunders T., Wieder R., 2011. A pilot phase i trial of all-trans retinoic acid (Vesanoid) and paclitaxel (Taxol) in patients with recurrent or metastatic breast cancer. Invest. New Drugs. 29, 1482–1487. [PubMed]
  • Budd G.T., Adamson P.C., Gupta M., Homayoun P., Sandstrom S.K., Murphy R.F., McLain D., Tuason L., Peereboom D., Bukowski R.M., Ganapathi R., 1998. Phase I/II trial of all-trans retinoic acid and tamoxifen in patients with advanced breast cancer. Clin. Cancer Res. 4, 635–642. [PubMed]
  • Chambers A.F., 2009. MDA-MB-435 and M14 cell lines: identical but not M14 melanoma?. Cancer Res. 69, 5292–5293. [PubMed]
  • Charafe-Jauffret E., Ginestier C., Iovino F., Tarpin C., Diebel M., Esterni B., Houvenaeghel G., Extra J.M., Bertucci F., Jacquemier J., Xerri L., Dontu G., Stassi G., Xiao Y., Barsky S.H., Birnbaum D., Viens P., Wicha M.S., 2010. Aldehyde dehydrogenase 1-positive cancer stem cells mediate metastasis and poor clinical outcome in inflammatory breast cancer. Clin. Cancer Res. 16, 45–55. [PubMed]
  • Chaturvedi P., Singh A.P., Batra S.K., 2008. Structure, evolution, and biology of the MUC4 mucin. FASEB J. 22, 966–981. [PubMed]
  • Chiesa M.D., Passalacqua R., Michiara M., Franciosi V., Di C.F., Bisagni G., Camisa R., Buti S., Tomasello G., Cocconi G., 2007. Tamoxifen vs Tamoxifen plus 13-cis-retinoic acid vs Tamoxifen plus Interferon alpha-2a as first-line endocrine treatments in advanced breast cancer: updated results of a phase II, prospective, randomised multicentre trial. Acta Biomed. 78, 204–209. [PubMed]
  • Chung J.H., Lee H.J., Kim B.H., Cho N.Y., Kang G.H., 2011. DNA methylation profile during multistage progression of pulmonary adenocarcinomas. Virchows Arch. 459, 201–211. [PubMed]
  • Crooks G.E., Hon G., Chandonia J.M., Brenner S.E., 2004. WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190. [PubMed]
  • Gagnon I., Duester G., Bhat P.V., 2003. Enzymatic characterization of recombinant mouse retinal dehydrogenase type 1. Biochem. Pharmacol. 65, 1685–1690. [PubMed]
  • Ginestier C., Hur M.H., Charafe-Jauffret E., Monville F., Dutcher J., Brown M., Jacquemier J., Viens P., Kleer C.G., Liu S., Schott A., Hayes D., Birnbaum D., Wicha M.S., Dontu G., 2007. ALDH1 is a marker of normal and malignant human mammary stem cells and a predictor of poor clinical outcome. Cell Stem Cell. 1, 555–567. [PubMed]
  • Ginestier C., Wicinski J., Cervera N., Monville F., Finetti P., Bertucci F., Wicha M.S., Birnbaum D., Charafe-Jauffret E., 2009. Retinoid signaling regulates breast cancer stem cell differentiation. Cell Cycle. 8, 3297–3302. [PubMed]
  • Guarnaccia V., 2010. Retinoic Acid and Breast Cancer: How to Improve the Therapeutic Effect on the Basis of Moleuclar Knowledge.
  • Gudas L.J., Wagner J.A., 2011. Retinoids regulate stem cell differentiation. J. Cell Physiol. 226, 322–330. [PubMed]
  • Hollestelle A., Schutte M., 2009. Comment Re: MDA-MB-435 and M14 cell lines: identical but not M14 melanoma?. Cancer Res. 69, 7893 [PubMed]
  • Huang M.E., Ye Y.C., Chen S.R., Chai J.R., Lu J.X., Zhoa L., Gu L.J., Wang Z.Y., 1988. Use of all-trans retinoic acid in the treatment of acute promyelocytic leukemia. Blood. 72, 567–572. [PubMed]
  • Jia J., Parikh H., Xiao W., Hoskins J.W., Pflicke H., Liu X., Collins I., Zhou W., Wang Z., Powell J., Thorgeirsson S.S., Rudloff U., Petersen G.M., Amundadottir L.T., 2013. An integrated transcriptome and epigenome analysis identifies a novel candidate gene for pancreatic cancer. BMC. Med. Genomics. 6, 33 [PubMed]
  • Kang E.J., Jung H., Woo O.H., Park K.H., Woo S.U., Yang D.S., Kim A.R., Lee J.B., Kim Y.H., Kim J.S., Seo J.H., 2014. Association of aldehyde dehydrogenase 1 expression and biologically aggressive features in breast cancer. Neoplasma. 61, 352–362. [PubMed]
  • Khoury T., Ademuyiwa F.O., Chandrasekhar R., Jabbour M., Deleo A., Ferrone S., Wang Y., Wang X., 2012. Aldehyde dehydrogenase 1A1 expression in breast cancer is associated with stage, triple negativity, and outcome to neoadjuvant chemotherapy. Mod. Pathol. 25, 388–397. [PubMed]
  • Kim B.H., Cho N.Y., Shin S.H., Kwon H.J., Jang J.J., Kang G.H., 2009. CpG island hypermethylation and repetitive DNA hypomethylation in premalignant lesion of extrahepatic cholangiocarcinoma. Virchows Arch. 455, 343–351. [PubMed]
  • Lalevee S., Anno Y.N., Chatagnon A., Samarut E., Poch O., Laudet V., Benoit G., Lecompte O., Rochette-Egly C., 2011. Genome-wide in silico identification of new conserved and functional retinoic acid receptor response elements (direct repeats separated by 5 bp). J. Biol. Chem. 286, 33322–33334. [PubMed]
  • Lemay D.G., Hwang D.H., 2006. Genome-wide identification of peroxisome proliferator response elements using integrated computational genomics. J. Lipid Res. 47, 1583–1587. [PubMed]
  • Li T., Su Y., Mei Y., Leng Q., Leng B., Liu Z., Stass S.A., Jiang F., 2010. ALDH1A1 is a marker for malignant prostate stem cells and predictor of prostate cancer patients' outcome. Lab Invest. 90, 234–244. [PubMed]
  • Li X., Wan L., Geng J., Wu C.L., Bai X., 2012. Aldehyde dehydrogenase 1A1 possesses stem-like properties and predicts lung cancer patient outcome. J. Thorac. Oncol. 7, 1235–1245. [PubMed]
  • Liu R.Z., Graham K., Glubrecht D.D., Germain D.R., Mackey J.R., Godbout R., 2011. Association of FABP5 expression with poor survival in triple-negative breast cancer: implication for retinoic acid therapy. Am. J. Pathol. 178, 997–1008. [PubMed]
  • Liu X., Nugoli M., Laferriere J., Saleh S.M., Rodrigue-Gervais I.G., Saleh M., Park M., Hallett M.T., Muller W.J., Giguere V., 2011. Stromal retinoic acid receptor beta promotes mammary gland tumorigenesis. Proc. Natl. Acad. Sci. U S A. A108, 774–779. [PubMed]
  • Luo Y., Dallaglio K., Chen Y., Robinson W.A., Robinson S.E., McCarter M.D., Wang J., Gonzalez R., Thompson D.C., Norris D.A., Roop D.R., Vasiliou V., Fujita M., 2012. ALDH1A isozymes are markers of human melanoma stem cells and potential therapeutic targets. Stem Cells. 30, 2100–2113. [PubMed]
  • Makiyama K., Hamada J., Takada M., Murakawa K., Takahashi Y., Tada M., Tamoto E., Shindo G., Matsunaga A., Teramoto K., Komuro K., Kondo S., Katoh H., Koike T., Moriuchi T., 2005. Aberrant expression of HOX genes in human invasive breast carcinoma. Oncol. Rep. 13, 673–679. [PubMed]
  • Mao P., Joshi K., Li J., Kim S.H., Li P., Santana-Santos L., Luthra S., Chandran U.R., Benos P.V., Smith L., Wang M., Hu B., Cheng S.Y., Sobol R.W., Nakano I., 2013. Mesenchymal glioma stem cells are maintained by activated glycolytic metabolism involving aldehyde dehydrogenase 1A3. Proc. Natl. Acad. Sci. U S A. A110, 8644–8649. [PubMed]
  • Marcato P., Dean C.A., Pan D., Araslanova R., Gillis M., Joshi M., Helyer L., Pan L., Leidal A., Gujar S., Giacomantonio C.A., Lee P.W., 2011. Aldehyde dehydrogenase activity of breast cancer stem cells is primarily due to isoform ALDH1A3 and its expression is predictive of metastasis. Stem Cells. 29, 32–45. [PubMed]
  • Marchitti S.A., Brocker C., Stagos D., Vasiliou V., 2008. Non-P450 aldehyde oxidizing enzymes: the aldehyde dehydrogenase superfamily. Expert Opin. Drug Metab. Toxicol. 4, 697–720. [PubMed]
  • McGrane M.M., 2007. Vitamin A regulation of gene expression: molecular mechanism of a prototype gene. J. Nutr. Biochem. 18, 497–508. [PubMed]
  • Moreb J.S., Mohuczy D., Ostmark B., Zucali J.R., 2007. RNAi-mediated knockdown of aldehyde dehydrogenase class-1A1 and class-3A1 is specific and reveals that each contributes equally to the resistance against 4-hydroperoxycyclophosphamide. Cancer Chemother. Pharmacol. 59, 127–136. [PubMed]
  • Morimoto K., Kim S.J., Tanei T., Shimazu K., Tanji Y., Taguchi T., Tamaki Y., Terada N., Noguchi S., 2009. Stem cell marker aldehyde dehydrogenase 1-positive breast cancers are characterized by negative estrogen receptor, positive human epidermal growth factor receptor type 2, and high Ki67 expression. Cancer Sci. 100, 1062–1068. [PubMed]
  • Mukhopadhyay P., Lakshmanan I., Ponnusamy M.P., Chakraborty S., Jain M., Pai P., Smith L.M., Lele S.M., Batra S.K., 2013. MUC4 overexpression augments cell migration and metastasis through EGFR family proteins in triple negative breast cancer cells. PLoS One. 8, e54455 [PubMed]
  • Nerlich A.G., Bachmeier B.E., 2013. Density-dependent lineage instability of MDA-MB-435 breast cancer cells. Oncol. Lett. 5, 1370–1374. [PubMed]
  • Neumeister V., Agarwal S., Bordeaux J., Camp R.L., Rimm D.L., 2010. In situ identification of putative cancer stem cells by multiplexing ALDH1, CD44, and cytokeratin identifies breast cancer patients with poor prognosis. Am. J. Pathol. 176, 2131–2138. [PubMed]
  • Novak P., Jensen T., Oshiro M.M., Wozniak R.J., Nouzova M., Watts G.S., Klimecki W.T., Kim C., Futscher B.W., 2006. Epigenetic inactivation of the HOXA gene cluster in breast cancer. Cancer Res. 66, 10664–10670. [PubMed]
  • Park S.Y., Kwon H.J., Choi Y., Lee H.E., Kim S.W., Kim J.H., Kim I.A., Jung N., Cho N.Y., Kang G.H., 2012. Distinct patterns of promoter CpG island methylation of breast cancer subtypes are associated with stem cell phenotypes. Mod. Pathol. 25, 185–196. [PubMed]
  • Penzes P., Wang X., Napoli J.L., 1997. Enzymatic characteristics of retinal dehydrogenase type I expressed in Escherichia coli. Biochim. Biophys. Acta. 1342, 175–181. [PubMed]
  • Pilato B., Pinto R., De S.S., Lambo R., Paradiso A., Tommasi S., 2013. HOX gene methylation status analysis in patients with hereditary breast cancer. J. Hum. Genet. 58, 51–53. [PubMed]
  • Resetkova E., Reis-Filho J.S., Jain R.K., Mehta R., Thorat M.A., Nakshatri H., Badve S., 2010. Prognostic impact of ALDH1 in breast cancer: a story of stem cells and tumor microenvironment. Breast Cancer Res. Treat. 123, 97–108. [PubMed]
  • Rexer B.N., Zheng W.L., Ong D.E., 2001. Retinoic acid biosynthesis by normal human breast epithelium is via aldehyde dehydrogenase 6, absent in MCF-7 cells. Cancer Res. 61, 7065–7070. [PubMed]
  • Schneider T.D., Stephens R.M., 1990. Sequence logos: a new way to display consensus sequences. Nucleic Acids Res. 18, 6097–6100. [PubMed]
  • Schug T.T., Berry D.C., Shaw N.S., Travis S.N., Noy N., 2007. Opposing effects of retinoic acid on cell growth result from alternate activation of two different nuclear receptors. Cell. 129, 723–733. [PubMed]
  • Schug T.T., Berry D.C., Toshkov I.A., Cheng L., Nikitin A.Y., Noy N., 2008. Overcoming retinoic acid-resistance of mammary carcinomas by diverting retinoic acid from PPARbeta/delta to RAR. Proc. Natl. Acad. Sci. U S A. A105, 7546–7551. [PubMed]
  • Sima A., Parisotto M., Mader S., Bhat P.V., 2009. Kinetic characterization of recombinant mouse retinal dehydrogenase types 3 and 4 for retinal substrates. Biochim. Biophys. Acta. 1790, 1660–1664. [PubMed]
  • Singletary S.E., Atkinson E.N., Hoque A., Sneige N., Sahin A.A., Fritsche H.A., Lotan R., Lu T., Hittelman W.N., Bevers T.B., Stelling C.B., Lippman S.M., 2002. Phase II clinical trial of N-(4-Hydroxyphenyl)retinamide and tamoxifen administration before definitive surgery for breast neoplasia. Clin. Cancer Res. 8, 2835–2842. [PubMed]
  • Sladek N.E., Kollander R., Sreerama L., Kiang D.T., 2002. Cellular levels of aldehyde dehydrogenases (ALDH1A1 and ALDH3A1) as predictors of therapeutic responses to cyclophosphamide-based chemotherapy of breast cancer: a retrospective study. Rational individualization of oxazaphosphorine-based cancer chemotherapeutic regimens. Cancer Chemother. Pharmacol. 49, 309–321. [PubMed]
  • Sutton L.M., Warmuth M.A., Petros W.P., Winer E.P., 1997. Pharmacokinetics and clinical impact of all-trans retinoic acid in metastatic breast cancer: a phase II trial. Cancer Chemother. Pharmacol. 40, 335–341. [PubMed]
  • Tang X.H., Gudas L.J., 2011. Retinoids, retinoic acid receptors, and cancer. Annu. Rev. Pathol. 6, 345–364. [PubMed]
  • The Alpha-Tocopherol Beta Carotene Cancer Prevention Study Group 1994. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N. Engl. J. Med. 330, 1029–1035. [PubMed]
  • Wardwell-Ozgo J., Dogruluk T., Gifford A., Zhang Y., Heffernan T.P., van D.R., Creighton C.J., Chin L., Scott K.L., 2014. HOXA1 drives melanoma tumor growth and metastasis and elicits an invasion gene expression signature that prognosticates clinical outcome. Oncogene. 33, 1017–1026. [PubMed]
  • Yang Z.L., Yang L., Zou Q., Yuan Y., Li J., Liang L., Zeng G., Chen S., 2013. Positive ALDH1A3 and negative GPX3 expressions are biomarkers for poor prognosis of gallbladder cancer. Dis. Markers. 35, 163–172. [PubMed]
  • Yonezawa S., Goto M., Yamada N., Higashi M., Nomoto M., 2008. Expression profiles of MUC1, MUC2, and MUC4 mucins in human neoplasms and their relationship with biological behavior. Proteomics. 8, 3329–3341. [PubMed]
  • Yonezawa S., Higashi M., Yamada N., Yokoyama S., Kitamoto S., Kitajima S., Goto M., 2011. Mucins in human neoplasms: clinical pathology, gene expression and diagnostic application. Pathol. Int. 61, 697–716. [PubMed]
  • Zhang R.D., Fidler I.J., Price J.E., 1991. Relative malignant potential of human breast carcinoma cell lines established from pleural effusions and a brain metastasis. Invasion Metastasis. 11, 204–215. [PubMed]
  • Zhang W., Yan W., You G., Bao Z., Wang Y., Liu Y., You Y., Jiang T., 2013. Genome-wide DNA methylation profiling identifies ALDH1A3 promoter methylation as a prognostic predictor in G-CIMP-primary glioblastoma. Cancer Lett. 328, 120–125. [PubMed]
  • Zhao D., McCaffery P., Ivins K.J., Neve R.L., Hogan P., Chin W.W., Drager U.C., 1996. Molecular identification of a major retinoic-acid-synthesizing enzyme, a retinaldehyde-specific dehydrogenase. Eur. J. Biochem. 240, 15–22. [PubMed]
  • Zhu Y., Zhang J.J., Zhu R., Zhu Y., Liang W.B., Gao W.T., Yu J.B., Xu Z.K., Miao Y., 2011. The increase in the expression and hypomethylation of MUC4 gene with the progression of pancreatic ductal adenocarcinoma. Med. Oncol. 28, (Suppl. 1) S175–S184. [PubMed]

Articles from Molecular Oncology are provided here courtesy of Wiley-Blackwell