Increased Retinoic Acid Receptor-β₄ Correlates In vivo with Reduced Retinoic Acid Receptor-β₂ in Esophageal Squamous Cell Carcinoma

Different retinoic acid receptor-β (RAR-β) isoforms seem to have contrasting biological effects in human carcinogenesis. Both in vitro and in vivo data indicate that RAR-β₂ expression is frequently lost or reduced (and transfecting RAR-β₂ suppresses growth and promotes apoptosis) in various cancer cells and tissues, whereas RAR-β₄ expression is increased in several cancer cell lines. To clarify the effects of different RAR-β isoforms in esophageal carcinogenesis, we used real-time quantitative reverse transcription-PCR to assess in vivo RAR-β mRNA levels in specimens of normal and malignant human esophageal tissue, comparing these levels with each other and the expressions of other genes. RAR-β₂ mRNA expression was significantly reduced (i.e., lower in cancer than normal tissue) in 67% (18 of 27, P = 0.001) and RAR-β₄ mRNA was increased in 52% (14 of 27, P = 0.054) of our esophageal cancer cases. The expressions of RAR-β₁, chicken ovalbumin upstream promoter-transcription factor-I (COUP-TFI), COUP-TFII, and peroxisome proliferator-activated receptor-γ (PPAR-γ) mRNA were reduced, whereas epidermal growth factor receptor and cyclin D1 expressions were increased in tumor compared with in normal tissues. Reduced RAR-β₂ expression correlated with increased RAR-β₄ expression (P = 0.002) and with the suppression of COUP-TFI and COUP-TFII (P = 0.050 and 0.023, respectively) in tumor samples. These are the first in vivo expression patterns of RAR-β₂ and RAR-β₄ reported in humans or animals and support the in vitro data on these isoforms and their contrasting biological effects in human carcinogenesis.

Esophageal cancer is an important health problem with a poor prognosis and high incidence in many parts of the world (1-3). Although relatively less common in the United States than in other countries, esophageal cancer was accounted for an estimated 14,250 new cases and 13,300 deaths (the seventh leading cause of cancer death in U.S. men) in 2004 (4). The most common histologic types of primary esophageal cancer are squamous cell carcinoma (SCC) and adenocarcinoma (1-3). Tobacco smoking is a significant risk factor for both histologic types of esophageal cancer, whereas gastroesophageal reflux leading to Barrett's esophagus is more commonly associated with adenocarcinoma (1-3, 5-11).

Multistep esophageal carcinogenesis involves many genetic and epigenetic alterations (1, 2). A number of studies have suggested that the loss of RAR-β₂ expression may contribute to esophageal carcinogenesis. We previously showed that RAR-β₂ expression is progressively lost in dysplasia and SCC of the esophagus (12-14). Other groups also recently reported the loss of RAR-β₂ expression and hypermethylation of RAR-β₂ promoter in esophageal cancer (15-20). Our earlier studies have further shown that transfection of RAR-β₂ decreased cell growth and colony formation and induced apoptosis in esophageal cancer cells (21); that the induction of RAR-β₂ decreased epidermal growth factor receptor (EGFR) and cyclooxygenase-2 (COX-2) expression; and that antisense RAR-β₂ transfection increased EGFR and COX-2 expression (21),⁵

which are frequently induced during esophageal carcinogenesis (1). Benzo(a)pyrene diol epoxide (a carcinogen present in tobacco smoke) and tumor-promoting bile acids suppressed RAR-β₂ expression in esophageal cancer cell lines (22). Another recent study showed that the cancer preventive agent (−)-epigallocatechin-3-gallate, which is extracted from green tea, caused a concentration- and time-dependent reversal of hypermethylation of RAR-β₂, resulting in its reexpression (19).

The known isoforms of RAR-β in mice are β₁, β₂, β₃, and β₄ and in humans are β₁, β₂, and β₄ (23, 24). RAR-β₂ is the most abundant isoform and the major retinoic acid-inducible form; therefore, the term RAR-β in the literature usually refers to the RAR-β₂ isoform. RAR-β₁ is a fetal isoform and may be a master developmental gene in humans although it is also expressed in small cell lung cancer (25). A recent study showed that RAR-β₁ has unique tumor suppressor activity that could not be entirely replaced by overexpressing RAR-β₂ and inhibiting RAR-β₄ expression (26). RAR-β₂ has tumor suppressor activity and is frequently lost in various human cancers (27). RAR-β₄ is generated by alternative splicing from the same primary transcripts as those generating RAR-β₂ and is initiated by the non-AUG codon CUG. The amino acid sequence of RAR-β₄ in regions B to F is identical to that of the regions of RAR-β₂. The A region of RAR-β₄ is much shorter (four amino acids long) than in RAR-β₂ (28). Therefore, RAR-β₄ may act as a dominant-negative form of RAR-β₂. Indeed, overexpression of RAR-β₄ seems to predispose lung and mammary tissues to hyperplasia and neoplasia (29-31). However, there are no previous reports of RAR-β₄ expression in esophageal carcinogenesis.

These previous in vitro studies of RAR-β₂ and RAR-β₄ suggest that the two RAR-β isoforms have contrasting biological effects in human cancer development, but these effects have not been assessed in vivo. The biological relevance of these effects could involve other genes, including chicken ovalbumin upstream promoter-transcription factors (COUP-TF), cyclin D1, and peroxisome proliferator-activated receptor-γ (PPAR-γ; refs. 32-35). Therefore, we quantitatively assessed RAR-β₂ and RAR-β₄ expressions in vivo, including assessments of their biological relevance in malignant and normal esophageal tissues.

We obtained 27 fresh specimens of esophageal SCC and 27 matched fresh specimens of distant normal squamous mucosae from Taishan Medical School, China. All specimens came from patients diagnosed with SCC and who had no presurgical therapy. All surgical samples were placed immediately into liquid nitrogen and stored at −80°C.

We analyzed expression of RAR-β₁, RAR-β₂, RAR-β₄, COUP-TFI, COUP-TFII, PPAR-γ, EGFR, and cyclin D1 using real-time quantitative reverse transcription-polymerase chain reaction (qRT-PCR), which was done in the core facility of University of Texas-Houston Medical School by using 7700 Prism real-time PCR machine (Perkin-Elmer/ABI, Norwalk, CT). Briefly, total RNA from normal and esophageal SCC tissue samples was isolated using Tri-reagent (Molecular Research Center, Cincinnati, OH), treated with RNase-free DNase (Roche Biochemicals, Indianapolis, IN), and diluted in 30 ng/μL RNA according to the standard method of the core facility. The primers used for detection of gene expression are listed in Table 1. The reverse transcription and PCR reactions were run in 96-well plates. The reaction mixture of the reverse transcription contained 2 μL extracted cellular RNA, 500 μmol/L of each deoxynucleotide triphosphate, 200 to 300 nmol/L of reverse primer, 1× first-strand buffer, 10 mmol/L DTT, and 10 units of Superscript II reverse transcriptase (Invitrogen-Life Technologies, Carlsbad, CA) in a volume of 10 μL. The reaction mixture was incubated at 50°C for 30 minutes followed by heat inactivation at 72°C for 10 minutes. Following first-strand synthesis, all of the 10 μL RT reaction mixture was used for subsequent PCR amplification by adding 40 μL PCR master mix to the same wells. The PCR reaction mixture included 1× PCR buffer, 200 μmol/L of each deoxynucleotide triphosphate, and 1.25 units Taq polymerase (Roche) in a final volume of 50 μL. Depending on the target transcript, the optimal concentration of MgCl₂, primers, and the fluorogenic Taqman probes were also added into the mixture. Each sample was measured in triplicate. The data are presented as mean of the percentage of β-actin for each RNA sample.

Summary statistics including mean and SEs were computed to characterize the distribution of expression for each of the eight genes analyzed. To compare the gene expression between normal and tumor tissues for each patient, numbers of patients with higher expression in normal, equal expression between normal and tumor, and higher expression in tumor were reported. Wilcoxon signed-rank test with continuity correction was applied to compare the differential expression between normal and tumor. In addition, symmetrized percent change (SPC), defined as (T − N) / (T + N) × 100%, was computed to measure the amount of change between the normal (N) and tumor (T) samples. When gene expression in both N and T are zero, SPC is defined as 0. SPC is a robust statistic for measuring change because it is bounded between −100% and 100% and can be computed when either one or both measures are zero (36). Fisher's exact test was computed to compare the association of SPC between any two genes. All tests were done with two-sided 5% type I error rate.

Table 2 shows gene expressions in normal and SCC tissue specimens for the eight genes of interest (RAR-β₁, RAR-β₂, RAR-β₄, COUP-TFI, COUP-TFII, PPAR-γ, EGFR, and cyclin D1) by qRT-PCR. The mean and SE of gene expression in normal and tumor tissues as well as the counts of patients with higher expression in normal, equal expression between normal and tumor, and higher expression in tumor tissues were given. The comparisons of the differential gene expression between normal and tumor tissues were done by the Wilcoxon signed-rank test with continuity correction. Our major findings involved the expressions of RAR-β₂ and RAR-β₄, COUP-TFI, and COUP-TFII. Table 2 shows that the expressions of RAR-β₂, COUP-TFI, and COUP-TFII were significantly lower in esophageal SCC tissue specimens (versus normal specimens), whereas the expression of RAR-β₄ was marginally higher in tumor versus normal tissue. Table 3 shows the associations of expression changes (between normal and tumor) for pairs of genes with statistically significant results based on the Fisher's exact test. We found a significant inverse correlation between the expressions of RAR-β₂ and RAR-β₄ (P = 0.002, Fisher's exact test). Twenty patients had decreased or unchanged RAR-β₂ in tumor (versus in normal tissue), 14 (70%) of which had increased RAR-β₄. All 14 patients with increased RAR-β₄ in tumor had decreased (n = 12) or unchanged (n = 2) RAR-β₂ in tumor (versus normal tissue).

Figure 1 shows the SPC of RAR-β₄ versus RAR-β₂. Most of the patients fell into the quadrant with increased RAR-β₄ and decreased RAR-β₂. There were no cases with increased expressions of both RAR-β₂ and RAR-β₄. In a sensitivity analysis with a SPC of 10% as the cutoff for defining increased RAR-β₄ and decreased RAR-β₂, the inverse association between RAR-β₂ and RAR-β₄ remained statistically significant (P = 0.046).

We also found that COUP-TFI and COUP-TFII expressions were positively correlated with RAR-β₂ expression (P = 0.05 and 0.023, respectively), whereas the expressions of COUP-TFI and RAR-β₄ were negatively correlated (P = 0.033; Table 3) and of COUP-TFII and RAR-β₄ were nonsignificantly negatively associated (P = 0.23).

We also assessed the expressions of other genes with a potential role in esophageal cancer development. Table 2 shows that the expressions of RAR-β₁ and PPAR-γ were significantly lower in esophageal SCC tissue specimens (versus normal specimens), whereas cyclin D1 expression was significantly higher and EGFR expression was marginally higher in tumor versus normal tissue. There was a marginally significant positive association between EGFR and cyclin D1 expressions (e.g., both increased in 12, both decreased or unchanged in seven tumors versus normal specimens; P = 0.057).

We quantitatively assessed the expressions of RAR-β₂ and RAR-β₄ in vivo in malignant and normal human esophageal tissues. RAR-β₂ was significantly decreased, confirming our earlier in vivo and in vitro data in esophageal cancer (12-14). The biological significance of the RAR-β₂ reduction in this study is suggested by its correlations with increased RAR-β₄ and suppressed COUP-TFs. There was a highly statistically significant inverse correlation between the expressions of RAR-β₂ and RAR-β₄. It is interesting to note that each of the 14 patients with increased RAR-β₄ expression in tumor (versus normal tissue) had decreased or unchanged RAR-β₂, supporting alternative splicing of RAR-β₂ as the source of the increased RAR-β₄ (28). To our knowledge, no previous study has assessed the relationship between RAR-β₂ and RAR-β₄ in vivo in malignant and normal human tissues.

We found that COUP-TF1 and COUP-TFII expressions were significantly reduced and correlated with reduced RAR-β₂ expression in esophageal SCC versus normal tissues, which is consistent with previous in vitro findings (32, 33). COUP-TFs are the most-studied orphan receptors in the nuclear receptor superfamily and are important in the regulation of organogenesis, neurogenesis, cellular differentiation, and homeostasis (37). COUP-TFI and COUP-TFII are distinct genes that show a remarkably high degree of homology, and their expression patterns often overlap, suggesting that they may have redundant functions. Each factor, however, possesses its own distinct expression profile during development (37). Although their involvement in carcinogenesis is largely unknown, COUP-TFs have been shown to be expressed differentially in several tumor cell lines (37) and to be expressed in correlation with RAR-β₂ induction by retinoic acid in human breast, bladder, and lung cancer cell lines (32). Inhibition of COUP-TF expression by stable expression of COUP-TF antisense RNA in COUP-TF–positive bladder cancer cells repressed the effect of retinoic acid on RAR-β₂ expression, demonstrating that COUP-TF is required for induction of RAR-β₂, growth inhibition, and apoptosis by retinoic acid in human cancer cell lines (32). Another study from the same group showed that stable expression of COUP-TF in nur77-positive, retinoic acid–resistant lung cancer cells enhanced the inducibility of RAR-β₂ gene expression and growth inhibition by retinoic acid (33). Our in vivo data from the current study showed a statistically significant positive correlation between COUP-TFs and RAR-β₂ expression and an inverse correlation between COUP-TFII and RAR-β₄ expression.

We also assessed the expressions of other potentially important genes in human esophageal cancer (RAR-β₁, EGFR, cyclin D1, and PPAR-γ). Although not well studied to date, our finding of reduced RAR-β₁ expression in tumor (versus normal) tissue, which is consistent with its reported tumor suppressor activity (26), should be explored further in future studies. Previous data have shown that cyclin D1 is overexpressed in esophageal cancer (1), that EGFR and cyclin D1 are down-regulated by retinoic acid in other cancer sites (34), and that PPAR-γ expression is down-regulated in esophageal cancer and PPAR-γ ligands can suppress esophageal cancer growth in vitro (38, 39). PPAR-γ also has been reported to induce RAR-β₂ expression in lung cancer cell lines (35). In the present study, we found that the expressions of EGFR and cyclin D1 were increased and of PPAR-γ was decreased in tumor (versus normal) tissue. These expression patterns are consistent with other reports (1, 38).

In conclusion, this study reveals a provocative shifting pattern in vivo between RAR-β₂ and RAR-β₄ expressions in human esophageal carcinogenesis, supporting previous in vitro data of ours and other groups. We observed that these shifting RAR-β isoform patterns were significantly associated with changes in COUP-TF expressions. Future studies to clarify the molecular basis of the contrasting effects of RAR-β₂ and RAR-β₄ could lead to new approaches (e.g., for assessing the risk and prognosis of esophageal cancer). Such future studies also could have important implications for targeted approaches for preventing and treating this devastating disease.

We thank the staff of the core facility at University of Texas-Houston Medical School for performing real-time qRT-PCR and Kendall Morse of the Department of Clinical Cancer Prevention for editorial assistance.