Genetic Ablation of the Fatty Acid–Binding Protein FABP5 Suppresses HER2-Induced Mammary Tumorigenesis

The intracellular lipid-binding protein (iLBP) termed fatty acid–binding protein 5 [fatty acid–binding protein (FABP5), also called keratinocyte FABP5, eFABP, cFABP, or mal1] is upregulated in multiple human cancers, including prostate (1–4), head and neck (5), bladder (6), and esophageal (7) cancers as well as in astrocytic glioma (8), glioblastoma (9), and squamous cell carcinoma (10, 11). Similarly, FABP5 is highly upregulated in the FVB/N-Tg(MMTVneu)202Mul/J (MMTV-neu) transgenic mouse model of breast cancer in which Neu/Erb-B2/Her2, encoding for an EGF receptor (EGFR) tyrosine kinase, is specifically overexpressed in the mammary epithelium, resulting in spontaneous development of mammary tumors in 100% of female mice (12). The mechanisms that underlie the upregulation of FABP5 in various cancers are incompletely understood. However, we previously showed that, in mammary carcinoma cells, FABP5 expression is controlled by EGFRs through activation of NF-κB, which directly targets the FABP5 promoter (13). Considering that the EGFR gene HER2/ERBB2/neu is amplified in a significant fraction of human breast cancers, these observation provide at least a partial explanation for upregulation of FABP5 in breast tumors as well as in mammary tumors that arise in the MMTV-neu mouse model. High expression level of FABP5 is correlated with poor patient survival in some cancers (9, 14), and it was shown that the protein displays pro-proliferative and antiapoptotic activities (13, 15–18). Available information thus suggests that FABP5 may play an important role in promoting tumor development in some cancers.

The iLBPs are members of a family of 15 small (∼15 kDa) soluble proteins that include FABPs, which associate with long-chain fatty acids and various other lipophilic compounds, and proteins that specifically bind retinoids, including cellular retinol-binding proteins (CRBP I-IV) and cellular retinoic acid (RA)–binding proteins (CRABP-I, II; ref. 19). It has been long believed that iLBPs serve a general role in solubilizing and protecting their hydrophobic and often labile ligands in the aqueous milieu of the cytosol. However, it is becoming increasingly clear that these proteins also have specific functions through which they regulate the biologic activities of their ligands (reviewed in ref. 20). The precise nature of the functions of most iLBPs remains to be elucidated, but it has been established that several of them cooperate with ligand-activated transcription factors of the nuclear hormone receptor family. It was thus shown that CRABP-II, FABP1, FABP4, and FABP5 transport specific ligands from the cytosol to the nucleus, where they directly deliver them to the nuclear receptors RAR, PPARα, PPARγ, and PPARβ/δ (PPARδ), respectively (21–26). These binding proteins thus facilitate the ligation and enhance the transcriptional activity of their cognate receptors. Interestingly, like FABP5, its cognate receptor PPARδ displays distinct pro-oncogenic activities (27–33). It has been suggested that the receptor exerts such activities in part by decreasing the expression of the tumor suppressor PTEN (32) and by direct induction of the survival factors ILK and PDPK1 (28, 32), the angiogenic factor VEGFα (27, 34), and the pro-oncogenic adipokine ANGPTL4 (33, 35). Hence, the oncogenic activities of FABP5 may stem from its ability to enhance the transcriptional activity of its cognate receptor. Indeed, it was reported that downregulation of either FABP5 or PPARδ inhibits the growth of the highly malignant prostate cancer PC3M cells (18) and abolishes the ability of the EGFR ligand heregulin β1 to enhance proliferation of MCF-7 mammary carcinoma cells (13).

PPARδ displays a broad ligand selectivity and can be activated by a variety of hydrophobic compounds including unsaturated long-chain fatty acids (36). It has also been reported that retinoic acid, the transcriptionally active metabolite of vitamin A classically known to activate the nuclear receptors RARs, serves as a high affinity natural agonist for PPARδ (16, 17, 37). Like its cognate receptor, FABP5 can associate with multiple types of lipophilic compounds. Interestingly, however, in support of the notion that FABP5 and PPARδ cooperate in regulating the transcriptional activity of shared ligands, the binding protein is mobilized to the nucleus only in specific response to ligands that activate this receptor. For example, while retinoic acid, the synthetic PPARδ agonist GW0742, the saturated fatty acid stearate, and synthetic ligands that selectively activate other PPAR isotypes all bind to FABP5 with similar affinity, only the first two compounds activate the nuclear localization of the protein.

Mammary tumors that arise in MMTV-Erbb2 mice display profound retinoic acid -resistance. Hence, while retinoic acid exerts anticarcinogenic activities and is currently used in treatment of various human cancers (38), the hormone promotes tumor development in MMTV-Erbb2 mice (17). It has been suggested that this paradoxical response stems from the marked upregulation of FABP5 upon tumorigenesis in these mice, i.e., that the increased expression of the binding protein results in diversion of retinoic acid to PPARδ and thus in induction of procarcinogenic PPARδ target genes (17).

Here, the involvement of FABP5 in regulating oncogenic properties and mammary tumor development was investigated using cultured cells as well as in the mouse model of breast cancer FVB-Tg(MMTV-Erbb2)NK1Mul/J (MMTV-Erbb2) in which an activated form of Erbb2 is specifically expressed in mammary tissue (39). Similarly to MMTV-Neu mice in which the wild-type Neu gene is amplified, 100% of MMTV-Erbb2 female mice develop mammary tumors. MMTV-Erbb2 mice were crossed with FABP5-null animals to generate mouse models expressing varying levels of FABP5 and development of tumors in these mice was monitored. The observations indicate that FABP5 enhances the ability of PPARδ to promote oncogenic properties in cultured cells. Moreover, the data show that ablation of FABP5 inhibits the transcriptional activity of PPARδ and dramatically suppresses Erbb2 signaling and tumorigenesis in vivo.

All-trans-RA was purchased from Calbiochem, GW0742 was obtained from Toronto research Chemicals, and 4-[(E)-2-(5,6,7,8-Tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl)-1-propenyl]benzoic acid TTNPB was from Sigma-Aldrich. The PPARδ antagonist PT-S58 was synthesized using the protocol described in (40).

Human normal mammary epithelial cells (HMEC) were cultured in HMEC ready medium (Invitrogen Life Sciences). NaF cells, originated from MMTV-neu tumors, were cultured in Dulbecco's Modified Eagle Medium supplemented with 10% FBS (Invitrogen Life Sciences). 3T3 fibroblasts were passaged fewer than 6 to 7 times and never grown to confluence. Cell lines stably overexpressing FABP5 or with reduced FABP5 expression were generated by transfecting a vector encoding FABP5 or harboring FABP5 short hairpin RNA (FABP5shRNA), respectively, using PolyFect transfection reagent (Qiagen). Cells overexpressing FABP5 or FABP5shRNA were selected using G418 (10 mg/mL) and puromycin (10 mg/mL), respectively. Individual colonies were pooled and assessed.

A total of 1,000 cells were plated in quadruplicates in a 96-well plate and treated with ligand for 4 days. Cells were incubated with MTT reagent (5 mg/mL in PBS; Sigma-Aldrich) until formazan crystals formed. Crystals were dissolved in 4 mmol/L HCl in isopropanol, and absorbance at 590 nm was measured using a microplate reader.

Binding assays were carried out by fluorescence titrations. FABP5 was bacterially expressed and purified, and the equilibrium dissociation constants (K_d) that characterize its interactions with different fatty acids were measured by fluorescence competition assays. The method entails two steps. In the first step, K_d for the association of the protein with the fluorescent fatty acid probe anilinonaphtalene-8-sulphonic acid (ANS) was measured. Protein (1 μmol/L) was titrated with ANS from a concentrated solution in ethanol. Ligand binding was monitored by following the increase in the fluorescence of the ligand upon binding to the protein, and K_d for the association of ANS with FABP5 was computed from titration curves. K_ds for binding of nonfluorescent ligands were then measured by monitoring their ability to compete with ANS for binding to the protein. FABP5 was precomplexed with ANS at 1:1 molar ratio and titrated with the different fatty acids, the binding of which was reflected by a decrease in probe fluorescence. K_ds were extracted from the EC₅₀ of the competition curve and the measured K_d for ANS. Analyses were carried out using Origin 8 software (MicroCal Software Inc.).

Total RNA was extracted using TRIzol (Molecular Research Center). cDNA was generated using GeneAmp RNA PCR (Applied Biosystems). Quantitative real-time PCR (qRT-PCR) was carried out using TaqMan chemistry and Assays-on-Demand probes (Applied Biosystems). 18s rRNA (4352930) was used for normalization. Relative expression was calculated as 2^−ΔΔCT for cell culture experiments and as 2^−ΔCT for tumors.

Assays were conducted using 8-μm pore size Corning Transwell migration plates following the manufacturer's protocol. A total of 10⁴ cells/cm² of NIH3T3 fibroblasts were plated. Cells in the top chamber were allowed to migrate for 12 hours, the top portion of the insert was wiped with a cotton swab, and inserts were fixed in 4% formalin for 30 minutes and then washed twice with PBS. Cells were then stained with 0.5% crystal violet for 10 to 15 minutes, washed twice with PBS, visualized, and counted.

NIH3T3 fibroblast cells were grown to 100% confluence and scratched using a 10-μL pipette tip. Cells were washed extensively with PBS to remove floating cells, and images were taken at initial 24 hours after scratching.

Assays were conducted using BD bioscience invasion assay following the manufacturer's protocol. Briefly, 5 × 10⁴ cells/mL were placed in a 24-well plate containing Matrigel and grown for 24 hours in normal growth media. Medium was aspirated and excess cells removed using a cotton swab. Membranes were then stained using the Diff-Quik staining Kit (IMBC Inc), excised, and mounted on a microscope slide and cells were counted.

Antigen retrieval was done by boiling the slides in 10 mmol/L citric buffer (pH 6.0, 15 minutes). Apoptosis and cell proliferation were examined by immunohistochemical analyses using rabbit polyclonal antibody recognizing cleaved caspase-3 (Cell Signaling Technology, no. 9661; dilution 1:200), and Ki67 (Novocastra Laboratories, no. NCL-Ki67p; dilution 1:500), respectively. Primary antibodies were incubated with deparaffinized sections (overnight at 4°C). The sections were incubated (30 minutes) with biotinylated secondary antibody and detected with EnVision System-HRP (DAB; Dako) kit.

Total protein was extracted from cells or tumor samples using radioimmunoprecipitation assay buffer (25 mmol/L Tris-HCl, pH 7.2, 150 mmol/L NaCl, 1% NP40, 1% sodium deoxycholate, and 0.1% SDS). Proteins were resolved by SDS-PAGE and transferred onto a nitrocellulose membrane. Membranes were incubated with primary antibodies, washed, and incubated with HRP-conjugated antibodies. Proteins were detected by exposure to enhanced chemiluminescence (ECL) and visualized on XR-B x-ray film. Band intensities were quantified using ImageJ 1.40g software (Wayen Rasband). The following antibodies were used: FABP5 antibodies were obtained from R&D Systems (catalog no. AF3077); antibodies against AKT (catalog no. 9272), phospho-AKT (Thr308, catalog no. 4056), ERK1 (catalog no. 4372), phospho-ERK 1/2 (Thr202/Tyr204, catalog no. 9106), and PTEN (catalog no. 9559) were from Cell Signaling Technology. β-Tubulin antibodies were from Santa Cruz Biotechnology (catalog no. sc-9104), Connexin 26 was purchased from Zymed (catalog no. 13-8100), and antibodies against PDPK1 were from BD Biosciences (catalog no. 61107).

Zebrafish (strain AB/TL) were bred and maintained under standard conditions at 28°C. Morphologic features were used to determine the stage of the embryos in hours (hpf) or days (dpf) post fertilization. Embryos used for injection were raised in ×1 E3 medium in the presence of 200 μmol/L 1-phenyl-2-thiourea (PTU; Sigma) to inhibit pigmentation.

NaF cells stably expressing control shRNA or FABP5shRNA were trypsinized, washed, and stained with the fluorescent cell membrane dye CM-DiO (Vibrant, Invitrogen). Labeled cells were resuspended in PBS (150 cells/4 nL) containing 0.5% phenol red in PBS (Sigma). Cells were then injected into the yolk sac of 2-dpf dechorionated zebrafish embryos, using a manual micro-injector (Pressure-pulsed micro injector Model PV830; World Precision Instruments) according to the method described by Marques and colleagues (43). Fish with fluorescent cells outside the implantation area 2 hours after injection were excluded from further analysis. All other fish were maintained at 28°C for 4 days in ×1 E3 medium containing PTU and analyzed. Microscopy was carried out using a Zeiss AxioCam MR camera attached to a Zeiss Axioplan 2 fluorescence microscope (GFP filter set: excitation 488, emission 530; Samrock; Imaging Core Facility of the Genetics Department at Case Western Reserve University) at a magnification of ×5 to ×40. Experiments were repeated four times with 10 to 15 fish per condition.

Mice were housed according to ARC protocol and Institutional Animal Care and Use Committee regulations. Mice were maintained on a 12 hours light and dark cycle on a normal chow diet. The mice had access to water and diet ad libitum. FVB/N-Tg(MMTV-Erbb2)1Led (Jackson Laboratories, stock no. 005038) mice, homozygous for the Erbb2 transgene were crossed with C57BL/6-FABP5^−/− mice, generously provided by Gökhan S. Hotamisligil (Harvard School of Public Health, Boston, MA; ref. 41). F1 mice were bred with FABP5^+/− females to yield MMTV-Erbb2^+/−/FABP5^+/+ (M5^+/+), MMTV-Erbb2^+/−/FABP5^+/− (M5^+/−), and MMTV-Erbb2^+/−/FABP5^−/− (M5^−/−) littermates. All experiments were conducted using virgin females. Tumor development was monitored by measuring the length and width with calipers and tumor volume calculated as length × width²/2. Statistical analyses of tumor growth curves were conducted on the natural log of tumor volumes in the two groups and compared with the wild type with unpaired t test using Origin 8.5.1 software.

Similarly to the marked upregulation of FABP5 that accompanies mammary tumor development in the MMTV-neu mouse model of breast cancer (17), examination of expression array profiles of human breast tumors deposited in the Oncomine™ DataBase (https://www.oncomine.org/) indicated that the expression level of FABP5 was elevated in human breast tumors in two independent studies (Fig. 1A). In accordance with these results, analysis of FABP5 mRNA levels in cDNA arrays of human breast cancer samples (OriGene) showed that FABP5 is markedly upregulated in early-stage breast cancer and remains high through disease progression (Fig. 1B).

Two types of cells were used to examine whether FABP5 is involved in regulating mammary carcinoma cell growth: NaF cells, a cell line generated from mammary tumors that arise in MMTV-neu mice, which expresses a high level of FABP5, and normal HMEC, which display a low expression level of the protein (HMEC, Fig. 2A). An NaF cell line in which FABP5 expression is stably reduced was generated (Fig. 2B). Reducing the expression of FABP5 markedly suppressed the proliferation of these cells (Fig. 2C). Conversely, ectopic overexpression of FABP5 in HMEC (Fig. 2D) facilitated their growth (Fig. 2E). The effect of FABP5 on cell growth was further assessed using 3T3 fibroblasts, cells that have long been used as a model for investigating processes underlying oncogenic transformation (42). Two 3T3 clones that overexpress different levels of FABP5 were generated (Fig. 2F), and their rates of proliferation were measured using MTT assays. As was the case with HMEC, overexpression of FABP5 facilitated the proliferation of 3T3 cells and did so in a dose-responsive fashion (Fig. 2G).

The effect of reducing the expression of FABP5 in NaF cells on their metastatic potential was evaluated in vivo using a recently developed zebrafish embryo model of cell invasion (43, 44). Embryos were raised in medium containing 1-phenyl-2-thiourea (PTU) to inhibit pigmentation. NaF cells or counterparts with reduced expression of FABP5 (Fig. 2B) were stained with the fluorescent cell membrane dye CM-DiO (Vibrant, Invitrogen) and injected into the yolk sac of 2-dpf dechorionated zebrafish embryos (see Materials and Methods for details). Fish with fluorescent cells outside the implantation area 2 hours after injection were excluded from further analysis. Fish were maintained for 4 days before fluorescence microscopy imaging (Fig. 2H). Control NaF cells efficiently exited the yolk sac, and many of them were found along the embryo body (Fig. 2H, top). In contrast, cells that expressed a low level of FABP5 remained in the yolk sac and very few were observed in the embryo tail (Fig. 2H, bottom). Hence, expression of FABP5 markedly enhanced the metastatic potential of the cells. Further attesting to an involvement of FABP5 in promoting oncogenic properties, “wound healing” assays showed that ectopic overexpression of FABP5 in 3T3 fibroblasts facilitated the wound closure behavior of the cells and did so in a dose-dependent manner (Fig. 2I).

The only known specific activity of FABP5 is the delivery of ligands to the nuclear receptor PPARδ, an activity through which the binding protein facilitates the ligation and thus enhances the activation of the receptor (13, 18, 26). The observations that FABP5 affects cellular behavior in the absence of added ligands likely reflect the presence of endogenous agonists for the receptor. Notably, the ligand selectivity of PPARδ is quite broad, and the receptor can be activated by multiple endogenous compounds, including retinoic acid and various unsaturated long-chain fatty acids. As cells were cultured in 5% (3T3) or 10% (NaF, HMEC) FBS, which contains approximately 0.05 or approximately 0.1 μmol/L retinol, the precursor for retinoic acid, as well as long-chain fatty acids, the presence of these compounds accounts for the activity of the FABP5/PPARδ pathway in the absence of added ligands. To examine the effect of defined exogenously added compounds, the association of FABP5 with several ligands was assessed. The protein was expressed in Escherichia coli and purified (Fig. 3A), and the equilibrium dissociation constants K_d that characterize its association with ligands were measured by fluorescence competition assays using the fluorescent probe ANS. The method entails two steps (45). In the first step, K_d for association of the protein with ANS was measured by fluorescence titrations. Titration with ANS resulted in a saturable increase in fluorescence (Fig. 3B), and analysis of titration curves (46) yielded a K_d of 70 ± 6.4 nmol/L. K_ds for nonfluorescent ligands were then measured by monitoring their ability to compete with the probe for binding. FABP5 was precomplexed with ANS and titrated with a nonfluorescent ligand the binding of which was reflected by a decrease in fluorescence upon displacement of the probe from the protein. Association of FABP5 with the synthetic PPARδ-selective ligand GW0742, the natural PPARδ ligand retinoic acid (16, 37), and the pan-RAR ligand TTNPB was thus examined (Fig. 3C). K_ds for GW0742 and retinoic acid, extracted from the EC₅₀ of competition curves and the measured K_d for the probe, were found to be 38.4 ± 8.2 and 42.3 ± 6.4 nmol/L, respectively. While displaying a high affinity toward these PPARδ ligands, FABP5 was not associated with TTNPB (Fig. 3C), showing the specific common ligand selectivity of the binding protein and its cognate receptor.

3T3 fibroblasts with varying expression levels of FABP5 (Fig. 2F) were used to further examine whether PPARδ is involved in the oncogenic activities of FABP5. In accordance with the reports that this receptor promotes cell growth and survival (27–30), treatment with the selective PPARδ agonist GW0742 facilitated proliferation of 3T3 cells (Fig. 4A). The ligand also enhanced the ability of the cells to migrate through a membrane in transwell migration assays (Fig. 4B) and to translocate through a Matrigel matrix in invasion assays (Fig. 4C). Indicating that 3T3 cells contain endogenous ligands for PPARδ, increasing the expression of FABP5 per se enhanced cell proliferation, migration, and invasion and did so in a dose-dependent manner. Increasing the expression of FABP5 markedly potentiated the ability of ectopically added GW0742 to induce proliferation, migration, and invasion (Fig. 4A–C). In keeping with the reports that retinoic acid serves as a ligand for PPARδ and that, like other PPARδ ligands, it is delivered to the receptor by FABP5 (16, 37), retinoic acid mirrored the effects of the synthetic PPARδ ligand in enhancing proliferation, migration, and invasion of 3T3 cells, and overexpression of FABP5 enhanced the effects in a dose-dependent fashion (Fig. 4D–F). Unlike retinoic acid and GW0742, the synthetic pan-RAR ligand TTNPB had little effect on these parameters of cell behavior (Fig. 4D–F), substantiating that the pro-oncogenic activity of retinoic acid in these cells is mediated by FABP5 and its cognate receptor PPARδ and not by the classical retinoic acid receptor RAR.

The synthetic PPARδ-selective antagonist PT-S58 (47) was then used to further examine the involvement of PPARδ in the pro-oncogenic activities of FABP5. In keeping with its function in enhancing the transcriptional activity of the receptor, overexpression of FABP5 increased the levels of two well-established PPARδ-target genes, Pdpk1 and Plin2 (Fig. 4G). Treatment with PT-S58 decreased the expression of both genes and abrogated the effect of FABP5 on their expression (Fig. 4G). Strikingly, PT-S58 completely abolished the ability of FABP5 to facilitate cell proliferation both in the absence and in the presence of added PPARδ ligands (Fig. 4H). Taken together, the data establish that pro-oncogenic activities of FABP5 are critically mediated by PPARδ and emanate from its ability to enhance the transcriptional activity of the receptor.

The MMTV-Erbb2 transgenic mouse model of breast cancer (39) was used to examine the effect of FABP5 expression on mammary tumor development in vivo. To generate MMTV-Erbb2 mouse models with varying expression levels of FABP5, MMTV-Erbb2 mice were crossed with FABP5-null mice (41) to obtain MMTV-Erbb2^+/−/FABP5^+/+ (M5^+/+), MMTV-Erbb2^+/−/FABP5^+/− (M5^+/−), and MMTV-Erbb2^+/−/FABP5^−/− (M5^−/−) mice. Monitoring tumor development in these mice showed that animals with reduced expression of FABP5 developed tumors at a dramatically slower rate than control MMTV-Erbb2^+/− mice and, notably, that the rate of tumor development was correlated with the expression level of the binding protein (Fig. 5A). Survival of the mice, defined as the age at which the first tumor to arise reached a volume of 1,200 mm³, was markedly longer in M5^−/− versus control mice, and M5^+/− mice displayed an intermediate survival rate (Fig. 5B). The differential tumor growth in these groups is exemplified by assessing tumor volumes at 59 weeks of age (Fig. 5C). At this age, all M5^+/+ had large tumors, whereas only one M5^−/− mouse had a discernible tumor, which was quite small. M5^+/+ mice also displayed a higher number of tumors per animal (Fig. 5D) as compared with mice with reduced expression of FABP5, with M5^+/− mice bearing an intermediate number of tumors per mouse. Strikingly, whereas tumors developed in 100% of mice in the control group, 3 of 15 M5^+/−, and 6 of 14 M5^−/− mice remained tumor free throughout the experiment (Fig. 5D).

Consistent with the behavior of cultured cells, immunostaining for the proliferation marker Ki67 showed that reduced expression of FABP5 was associated with a dose–responsive decrease in proliferation in tumors (Fig. 5E). On the other hand, the fraction of apoptotic cells, evaluated by immunostaining for cleaved caspase 3, was higher in mammary tumors of mice with reduced expression of FABP5 compared with the MMTV-Erbb2 controls (Fig. 5E).

It was previously reported that reducing the expression of FABP5 abolishes the ability of the EGFR ligand hertegulin β1 to enhance the proliferation of MCF-7 mammary carcinoma cells (13). EGFRs, for example, Erbb2, can promote cell growth through activation of a signaling network that includes pathways that are mediated by extracellular signal–regulated kinase (ERK) and phosphoinositide 3-kinase (PI3K) and its downstream effector the survival factor AKT (48, 49). Examination of the activation status of these pathways revealed that the reduced expression of FABP5 in M5^+/− and M5^−/− tumors resulted in decreased phosphorylation levels of both ERK (Fig. 6A) and AKT (Fig. 6B), suggesting that the oncogenic activity of the FABP5/PPARδ pathway in these tumors emanates at least in part through promotion of EGFR signaling. The reduced expression of FABP5 in M5^+/− and M5^−/− tumors also resulted in a dose-responsive upregulation of Connexin 26 (CNX26 Fig. 6C), a protein that was reported to function as a tumor suppressor in mammary carcinomas (50).

In accordance with the cooperation of FABP5 with PPARδ, the expression levels of two direct PPARδ target genes involved in mediating oncogenic activities of the receptor, the growth factor VEGFα (27, 34) and the survival factor PDPK1 (28, 29), were lower in tumors that arose in MMTV-Erbb2 mice with reduced expression of FABP5 as compared with tumors of control mice (Fig. 6D and E). Notably, the expression of these genes correlated in a dose-responsive manner with the expression of FABP5, showing an intermediate level in FABP5^+/− tumors. Further indicating that lack of FABP5 impairs the activity of PPARδ, PTEN, a tumor suppressor the expression of which is repressed by this receptor (28), was upregulated in FABP5^−/− tumors (Fig. 6E).

The data presented above indicate that expression of FABP5 is upregulated early in breast tumor development and remains high throughout disease progression. The observations thus add another cancer to the list of human cancers in which tumorigenesis is accompanied by a high level of expression of this binding protein (1–11). The observations further show that reducing the expression of FABP5 in NaF mammary carcinoma cells dramatically reduces their rate of proliferation and metastatic potential. Furthermore, ectopic overexpression of FABP5 in normal mammary epithelium cells or in 3T3 fibroblasts enhanced their proliferation, and, in 3T3 cells, facilitated “wound closure” in scratch assays. Considering that the only known specific activity of FABP5 is the delivery of ligands to PPARδ (13, 18, 26), and that it was reported that PPARδ promotes cell migration, proliferation, and survival (27–32), the oncogenic activities of the binding protein are likely to stem from its ability to enhance the transcriptional activity of the receptor. In accordance, ectopic expression of FABP5 enhanced the ability of endogenous ligands and two exogenously added agonists, the synthetic PPARδ/δ-selective ligand GW0742 and the natural PPARδ ligand retinoic acid, to promote proliferation, migration, and invasion of 3T3 fibroblasts, and the activity was completely abolished in the presence of a PPARδ antagonist. To examine whether the FABP5/PPARδ pathway is involved in mammary tumor development in vivo, we generated MMTV-Erbb2 mouse models of breast cancer expressing varying levels of FABP5. Monitoring mammary tumor growth revealed that ablation of FABP5 markedly suppressed tumor development in these mice. Whereas tumors rapidly developed in all M5^+/+ mice, tumor growth was dramatically delayed in mice with reduced expression of FABP5 and 20% of M5^+/− mice, and 43% of M5^−/− mice remained tumor free throughout the 504 days of the experiment. In agreement with the report that FABP5 is a critical mediator of the pro-proliferative activities of EGFR signaling (13), the phosphorylation/activation status of two downstream effectors of HER2/Erbb2, the mitogenic kinase ERK and the survival factor AKT, was markedly reduced in M5^−/− tumors. In contrast, the level of the tumor suppressor CNX26, the expression of which decreases upon EGFR activation, was higher in tumors with reduced FABP5 levels. Finally, in accordance with the function of the binding protein in enhancing the transcriptional activity of PPARδ, expression of the PPARδ target genes PDPK1 and VEGFα was lower and that of the PPARδ-repressed tumor suppressor gene PTEN was higher in tumors with reduced FABP5 expression.

Taken together with previous reports, the findings described here show that FABP5 efficiently promotes tumor development and that it does so by activating PPARδ. It is noteworthy that PPARδ agonists are currently being tested in clinical trials for treatment of metabolic diseases (51). The observations that the FABP5/PPARδ pathway displays potent pro-oncogenic activities indicate that such agonists should be carefully assessed before clinical use. The present observations suggest that the pathway is a promising druggable target for counteracting oncogenic activities in the many human cancers in which the expression of FABP5 is upregulated. Hence, inhibitors of FABP5 may comprise a novel class of potent anticarcinogenic agents.

No potential conflicts of interest were disclosed.

Conception and design: L. Levi, D. Seachrist, N. Noy

Development of methodology: L. Levi, G. Lobo, D. Seachrist

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Levi, G. Lobo, M.K. Doud, J. von Lintig, G.P. Tochtrop

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Levi, M.K. Doud, J. von Lintig

Writing, review, and/or revision of the manuscript: L. Levi, M.K. Doud, G.P. Tochtrop, N. Noy

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Levi

Study supervision: L. Levi, N. Noy

The authors thank Gökhan S. Hotamisligil (Harvard School of Public Health) for sharing his FABP5^−/− mice and Philip Leder (Harvard Medical School) for the NaF mammary carcinoma cells.

This work was supported by NIH grant R01 DK060684 (N. Noy). M.K. Doud was supported by NIH training grant R25-CA148052.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.