CRBP is underexpressed in 24% (95% confidence interval = 12.5%-36.5%) of human breast carcinomas, implying a link between cellular vitamin A homeostasis and breast cancer. We hypothesize that the loss of CRBP restricts the effects of endogenous vitamin A on breast epithelial cells.

We showed earlier that cellular retinol-binding protein (CRBP) expression is downregulated in a subset of human breast cancers. We have now investigated the outcome of ectopic CRBP expression in MTSV1-7 cells, a SV40 T antigen-transformed human breast epithelial cell line devoid of endogenous CRBP expression. We found that: (i) CRBP did not inhibit adherent cell growth but suppressed foci formation in post-con¯uent cultures and colony formation in soft agar; (ii) this e ect was due to CRBP inhibition of cell survival, as demonstrated by viability and TUNEL assays of cells in soft-agar or plated on polyHEMA-coated dishes; (iii) CRBP inhibited protein kinase B/Akt activation in cells in suspension but not in adherent cells and the CRBP suppression of anchorage-independent growth was mimicked by cell treatment with the phosphatidylinositol-3 kinase (PI3K) inhibitor LY294002; (iv) CRBP enhanced retinyl ester formation and storage but did not regulate retinoic acid synthesis or retinoic acid receptor activity. Ectopic CRBP-mediated inhibition of anchorage-independent cell survival and colony formation in the absence of signi®cantly altered responses to either retinol or retinoic acid was also documented in T47D human breast cancer cells. In conclusion, the data suggest two novel and linked CRBP functions in mammary epithelial cells: inhibition of the PI3K/Akt survival pathway and suppression of anchorage-independent growth. Oncogene (2001) 20, 7413 ± 7419.Keywords: FABP; soft agar; anoikis; retinol; retinoic acid CRBP is a 15 kDa cytosolic protein that binds retinol with high a nity and is postulated to regulate the intracellular fate of retinol. For instance, CRBP is known to regulate liver retinol esteri®cation and storage (Ghyselinck et al., 1999) and may also regulate retinol oxidation to retinoic acid in target tissues (Napoli, 1999;Yamamoto et al., 1998). CRBP is widely expressed in epithelial tissues, whereas its homologs, CRBPII and two novel CRBPs have more restricted tissue distributions (Vogel et al., 2000;Folli et al., 2001). CRBPs are the only retinol-binding members of the large family of fatty-acid binding proteins (FABPs; reviewed in Glatz and van der Vusse (1996).We recently reported that human breast carcinogenesis is associated with the loss of CRBP expression in roughly one fourth of breast cancer cases . To begin to investigate the possible signi®cance of this ®nding, we reconstituted CRBP expression in CRBP-negative MTSV1-7 cells and assayed the e ect of this intervention on cell growth. MTSV1-7 (abbreviated MTS) is a human breast ductal epithelial cell line that was immortalized by introduction of the SV40 T antigen (Bartek et al., 1991) and that upon continued culture in our laboratory acquired the ability to form foci and colonies in soft agar (YS Kuppumbatti et al., in preparation). Stable re-expression of CRBP in MTS cells ( Figure 1a) had no e ect on exponential monolayer growth (Figure 1b) but blocked the formation of foci in post-con¯uent cultures ( Figure 1c). Moreover, CRBP re-express...

Human breast cancer cell lines expressing the estrogen receptor a (ERa), all-trans-retinoic acid (ATRA) receptor a (RARa) and cellular retinoic acid binding protein II (CRABPII) genes are sensitive to ATRA-mediated growth inhibition. To study the relationship among ERa, RARa and CRABPII expression, the protein levels of each member were compared in five breast cancer cell lines (T47D, MCF-7, ZR-75-1, Hs587 T and MDA-MB-231 cells) and two immortalized nontumorigenic breast epithelial cell lines (MTSV1.7 and MCF-10A). ERa, RARa and CRABPII proteins were detected in T47D, MCF-7 and ZR-75-1 cells but not in other tested cell lines. RARa and CRABPII proteins were either reduced or undetectable in T47D/C4:2W and MCF-7/ADR cells with lost expression of ERa. Estradiol increased and antiestrogens (tamoxifen and ICI 164,384) downregulated the expression of both RARa and CRABPII proteins in T47D and MCF-7 cells. RARa antagonist Ro-41-5253 inhibited CRABPII expression, but not RARa expression in estradiol-treated T47D and MCF-7 cells. Suppression of ERa by small interfering RNA (siRNA) reduced RARa and CRABPII gene expression and siRNA suppression of RARa reduced CRABPII expression while having no effect on ERa in T47D cells. Transient transfection of either RARa or ERa expression vectors increased CRABPII expression in MDA-MB-231 cells but only RARa, not ERa, activated hCRABPII promoter reporter. These results indicate that there is a gene activation pathway in which ERa drives RARa transcription and RARa drives CRABPII transcription in ERa-positive human breast cancer cells.

We investigated the mechanism of retinoic acid receptor (RAR) b2 gene silencing in breast cancer cells. Transfection experiments indicated that MCF-7 cells transactivate an exogenous b2 promoter (71470/+156) to the same extent as MTSV1.7 breast epithelial cells, which express endogenous RARb2. This was true even in the context of replicated chromatin, suggesting a cisacting rather than a trans-acting defect. Cytosine methylation, a cis-acting DNA modi®cation, has been implicated in RARb2 silencing in cancer cells. Upon bisul®te genomic sequencing, we found that 3 CpG sites in the b2 RARE region were variably methylated in MCF-7 cells but were not methylated in MTSV1.7 cells or in 2 MDA-MB-231 subclones that di ered in RARb2 expression (high in clone A2, low in clone A4). However, the 5'-UTR region was hypermethylated in clone A4 relative to clone A2 cells. Following 5-azacytidine treatment, RA and trichostatin A markedly induced RARb2 expression in MCF-7 cells but not in MDA-MB-231 clone A4 cells. A b2 RARE reporter construct in which the methylation-susceptible cytosines in the sense strand were replaced by thymine displayed marked loss of activity in a replicated chromatin-dependent manner. We conclude that cytosine methylation contributes to RARb2 gene silencing in MCF-7 cells and that methylation of the RARE region may be particularly important.

Downregulation of the cellular retinol-binding protein-I (CRBP-I) occurs in breast and other human cancers, but its significance is not well understood. Recently, we showed that restoration of CRBP-I expression in transformed MTSV1-7 breast epithelial cells increased retinoic receptor activity, inhibited anoikis, promoted acinar differentiation and inhibited tumorigenicity, suggesting that CRBP-I suppresses tumor progression. However, the mechanism underlying these effects of CRBP-I was not elucidated. Here we demonstrate, using genetic and pharmacological approaches, that CRBP-I inhibits, in a retinoic acid receptor-dependent manner, the PI3K/Akt survival pathway. Inhibition of PI3K/Akt was necessary and sufficient to explain the antitumor effects of CRBP-I and was mediated by decreased p85 regulatory and p110 catalytic subunit heterodimerization. We present evidence consistent with the idea that this effect is due to CRBP-I inhibition of p85 phosphorylation at Y688. To our knowledge, this is the first demonstration of PI3K regulation at the level of p85-p110 heterodimerization. These findings lead us to hypothesize that CRBP-I downregulation in cancer promotes tumor progression through inhibition of retinoic acid receptor activity and derepression of PI3K/Akt signaling via a novel mechanism.