Available online at www.sciencedirect.com ScienceDirect Journal of Nutritional Biochemistry 64 (2019) 218 – 227 A high-corn-oil diet strongly stimulates mammary carcinogenesis, while a high-extra- virgin-olive-oil diet has a weak effect, through changes in metabolism, immune system function and proliferation/apoptosis pathways☆,☆☆ Raquel Escrich a , Irmgard Costa b , Montserrat Moreno a , Marta Cubedo c , Elena Vela a , Eduard Escrich a,⁎, Raquel Moral a,⁎ a Department of Cell Biology, Physiology and Immunology, Physiology Unit, Faculty of Medicine, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain b Department of Pathology, Corporació Parc Taulí-UDIAT, 08208 Sabadell, Barcelona, Spain c Department of Statistics, Universitat de Barcelona, 08028 Barcelona, Spain Received 24 January 2018; received in revised form 24 October 2018; accepted 9 November 2018 Abstract Breast cancer is the most common malignancy in women worldwide, and dietary lipids are important environmental factors influencing its etiology. We have investigated the effects, and the mechanisms associated, of high-fat diets on 7,12-dimethylbenz(a)anthracene-induced rat mammary tumors. Animals were fed a low-fat, a high-corn-oil (HCO) or a high-extra-virgin-olive-oil (HOO) diet from weaning or after induction. The HCO diet had a clear stimulating effect on mammary carcinogenesis, especially when dietary intervention started after induction, whereas the tumors from HOO diet groups exhibited clinical and morphological characteristics similar to those from low-fat controls. Transcriptomic and further protein and immunohistochemical analyses of tumors also indicated different modulatory effects of high-fat diets affecting relevant biological functions: metabolism, immunosurveillance and proliferation/apoptosis pathways. Thus, the results suggested different metabolic adaptations with increased glycolysis by effect of HOO diet. Moreover, leukocyte tumor infiltration and inflammation mediators showed increased cytotoxic T cells and decreased TGFβ1 expression by the HOO diet, while the HCO one increased arginase expression and IL-1α plasma levels. Furthermore, the study of proteins controlling proliferation/apoptosis pathways (Sema3A, Stat5, Smad1, Casp3) suggested an increase in proliferation by the HCO diet and an increase of apoptosis by the diet rich in olive oil. In conclusion, the HCO diet clearly stimulated mammary carcinogenesis, especially in the promotion phase, and induced molecular changes suggesting increased tumor proliferation/apoptosis balance and a proinflammatory microenvironment. The HOO diet, despite being high fat, had a weaker effect on tumorigenesis probably related to metabolic adaptations, enhanced immunosurveillance and increased apoptosis. © 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Keywords: Dietary lipids; Breast cancer; Olive oil; Transcriptome; Immune system; Apoptosis; Metabolism 1. Introduction environmental factors, diet being one of the most important [2]. Although there are some controversial data, epidemiological and Breast cancer was the most common cancer overall in 2015, with especially experimental data have shown the effects of dietary lipids 2.4 million cases worldwide, and a leading cause of women death [1]. on breast cancer risk [2–5]. Thus, their influence on malignant This neoplasia is multifactorial, and there is substantial contribution of transformation depends on the quantity and type of fat and timing Abbreviations: DMBA, 7,12-dimethylbenz(a)anthracene; EVOO, extra-virgin olive oil; FA, fatty acids; GO, gene ontology; HCO, high corn oil; HOO, high extra- virgin olive oil; LF, low fat; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids ☆ Funding: This work was supported by grants from “Plan Nacional de I+D+I” (AGL2006-07691); “Fundación Patrimonio Comunal Olivarero (FPCO)” (FPCO2008-165.396; FPCO2013-CF611.084); “Agencia para el Aceite de Oliva del Ministerio de Agricultura, Alimentación y Medio Ambiente” (AAO2008-165.471); “Organización Interprofesional del Aceite de Oliva Español (OIAOE)” (OIP2009-CD165.646), “Departaments de Salut i d'Agricultura, Alimentació i Acció Rural de la Generalitat de Catalunya” (GC2010-165.000), FPCO and OIAOE (FPCO-OIP2016-CF614.087). ☆☆ Author disclosures: no conflicts of interest. ⁎ Corresponding authors at: Department of Cell Biology, Physiology and Immunology, Physiology Unit, Faculty of Medicine, Avinguda de Can Domènech, Universitat Autònoma de Barcelona, 08193 Bellaterra (Cerdanyola del Vallès), Barcelona, Spain. Tel.: +34 93 581 13 41. E-mail addresses: m.fmm@uab.cat (R. Escrich), icosta@tauli.cat (I. Costa), gr.mecm@uab.cat (M. Moreno), mcubedo@ub.edu (M. Cubedo), Elena.Vela@uab.cat (E. Vela), Eduard.Escrich@uab.cat (E. Escrich), Raquel.Moral@uab.cat (R. Moral). https://doi.org/10.1016/j.jnutbio.2018.11.001 0955-2863/© 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 219 of dietary intervention, among other factors. In this regard, experi- the carcinogenesis. From weaning, animals were fed a low-fat diet mental data have revealed that, in general, n-6 polyunsaturated fatty (group LF, n=60), a high-corn-oil diet (group HCO, n=20) or a high- acids (PUFA) and animal saturated fatty acids clearly promote breast extra-virgin-olive-oil diet (group HOO, n=20) (Fig. 1). At 53 days of cancer, whereas n-3 PUFA would inhibit tumor growth and metastasis age, mammary tumors were induced by oral gavage with a single dose [3,4]. In relation to monounsaturated fatty acids (MUFA), evidence is of 5 mg of 7,12-dimethylbenz(a)anthracene (DMBA) (Sigma-Aldrich, still inconclusive, with data ranging from protective to weak St. Louis, MO, USA) dissolved in corn oil. After induction, 40 rats from promoting effects on experimental breast carcinogenesis [3,4]. LF group were changed to high-fat diets (groups LF-HCO and LF-HOO, Epidemiological data indicate that women in the Mediterranean n=20 each). Body weights were measured weekly throughout the European countries have lower rates of breast cancer than in other assay, and nose-to-anus length was measured at days 36, 51, 100 and Western countries [5,6]. Mediterranean diet is characterized by the the end of the assay to calculate body mass index (g/cm2). On days consumption of olive oil (rich in the MUFA oleic acid) as the main 236–256, rats were euthanized by decapitation. Plasma was obtained source of fat [7]. and maintained at −80°C until used. Tumors were removed and Although there is increasing experimental evidence of the measured, a portion of each was fixed in 4% buffered formalin, and the influence of diet on the development of neoplastic diseases, fewer rest was flash-frozen and stored at −80°C. Only confirmed mammary data support that dietary factors may influence not only the clinical adenocarcinomas have been included in this study. manifestation of the disease but also the histopathologic degree of malignancy, which has important prognostic value [8]. Moreover, the 2.3. Characterization of tumor malignant phenotype and tumor growth molecular mechanisms by which dietary lipids may influence mammary gland transformation are not well elucidated, but modifi- Tumor phenotype was characterized determining the degree of cations in hormonal status, structure and function of cell membranes, morphological aggressiveness by applying the modified Scarff–Bloom– signaling pathways or gene expression have been reported [3,4]. It has Richardson grading method adapted to rat mammary adenocarcinomas long been known that dietary lipids can modulate the expression of that we previously described [15]. This global histopathologic grade of metabolism genes in tissues like liver, but scarce evidence exists on adenocarcinomas adds together the differentiation pattern, nuclear their effects on cancer-related genes [9]. We have previously pleomorphism and mitotic index and scores 3–5 points (grade I or well demonstrated that the stimulating effect of a corn-oil-enriched diet differentiated), 6–8 points (grade II or moderately differentiated) and 9– on experimental mammary carcinogenesis can be mediated, at least in 11 points (grade III or poorly differentiated). Moreover, though most rat part, by transcriptomic changes affecting the expression of differen- mammary carcinomas are known to be nodular and very clearly bounded, tiation genes [10]. In this work, we aimed to investigate the effects of stromal response to the invasion was classified as absent (−) or as different high-fat diets, i.e., high in a seed oil (rich in n-6 PUFA) or in positive (+) when occasional prominent fibroblasts were seen around extra-virgin olive oil (EVOO, rich in oleic acid and bioactive neoplastic nests. Significant cellular and stromal reaction was classified as compounds) and administered from prepuberty or in adulthood, on ++ or +++ depending on the extent and intensity of the response, the clinical and anatomopathological characteristics of experimental latter when a truly desmoplastic reaction was present. Microscopic tumor mammary tumors. Moreover, to gain insight into the molecular necrosis was also evaluated as absent, focal or extensive. mechanisms underlying such effects, we investigated transcriptomic On the other hand, tumor growth was characterized with several changes in tumors suggesting changes in metabolism, immune system carcinogenesis parameters: latency (average time from carcinogen function and proliferation/apoptosis pathways. administration to first palpable malignant tumor), tumor incidence (percentage of rats bearing at least one tumor), tumor yield 2. Methods (cumulative total number of tumors) and total tumor volume (volume of all the tumors of the group). 2.1. Diets 2.4. RNA extraction and microarrays analysis Three semisynthetic diets were designed: a low-fat diet (LF, 3% corn oil w/w), a high-corn-oil-diet (HCO, 20% corn oil) and a high- Total RNA was isolated using the RNeasy Mini Kit (QiaGen, Hilden, olive-oil-diet (HOO, 3% corn oil+17% EVOO) (Supplementary Table 1). Germany) according to the manufacturer. RNA quality was analyzed Carbohydrates in the form of dextrose were 67.9% w/w in LF and 45.9% by Agilent Bioanalyzer 2100 and by ethidium-bromide-stained in the high-fat diets. They also contained protein in the form of casein agarose gel electrophoresis. (18% w/w in LF diet and 23% in the high-fat diets) and 5% w/w Six mammary adenocarcinomas of each group were chosen for cellulose, 5.9% w/w salt mixture and 0.24% w/w vitamin mixture. In whole-genome gene expression profiling. Briefly, GeneChip Rat Exon order to maintain the normal lipidic metabolism, they were 1.0 ST Array filters containing probesets for 850,000 exons were supplemented with methionine (0.51% w/w in LF and 0.66% in the obtained from Affymetrix (Santa Clara, CA, USA). Three micrograms of high-fat diets), choline (1800 mg/kg diet) and folic acid (5 mg/kg RNA (integrity number N8) was labeled with One-Cycle Target diet). The definition, preparation and suitability of the experimental Labeling and Control Reagent (Affymetrix), hybridized to chips diets were previously described [11–13], as well as the specific (Hybridization oven 640, Affymetrix) and scanned using GeneChip composition of the oils used [14]. Diets were prepared weekly and Scanner 3000 7G (Affymetrix) in the Microarrays Service from Vall stored under nitrogen in the dark at 4°C. d'Hebron Research Institute. Images obtained were processed with Microarray Analysis Suite 2.2. Animals and experimental design 5.0 (Affymetrix). Raw expression values were preprocessed using the RMA method [16]. Data were also nonspecific filtered to remove low- All animals received humane care under an institutionally signal and low-variability genes. These normalized and filtered values approved experimental animal protocol, following the legislation were the basis for all the analysis. Microarray data have been applicable in Spain. Female Sprague–Dawley Crl:SD rats were deposited at ArrayExpress under accession number E-MTAB-3541. purchased from Charles River Lab (L'Arbresle Cedex, France) at 23 The selection of differentially expressed genes between conditions days of age and maintained under standard conditions. Animals were was based on a linear model analysis [17]. Different statistics tests distributed according to the type and timing of dietary intervention, (fold change, t or P values) were used to classify genes into ranks thus diets influencing initiation and promotion, or only promotion of depending on its differential expression. In order to deal with the 220 R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 Fig. 1. Experimental design. Female Sprague–Dawley rats were fed the low-fat control diet (LF), the high-corn-oil diet from weaning (HCO) or from induction (LF-HCO), and the high- EVOO diet from weaning (HOO) or from induction (LF-HOO); n=20 animals/group. Animals were induced with 5 mg of DMBA at 53 days of age and euthanized at 236–256 days. multiple testing issues, P values were adjusted to obtain strong control 1:2000) and anti-PCNA (Cell Signaling Technology, 1:5000). After over the false discovery rate using the Benjamini and Hochberg incubation with secondary antibody (Sigma-Aldrich) and Luminata method [18]. Forte Western HRP Substrate (EMD Millipore) luminogen, proteins We performed analysis of biological significance based on were visualized using the Chemidoc XRS+ hardware associated with overrepresentation tests aiming to establish if genes found as Image Lab Software 5.1-Beta (Bio-Rad). This Bio-Rad system normal- differentially expressed were concentrated in some Gene Ontology izes the densitometric values of the bands with the total protein (GO) categories. Enrichment tests were performed using PANTHER loaded in each line, correcting the load by comparing with the first (Protein ANalysis THrough Evolutionary Relationships) Classification lane of the blot [19]. Thus, in order for the blots to be comparable, we System and Genecodis and FatiGO from Babelomics tool, where used an “internal control” of pooled samples loaded in duplicate in all Fisher's Exact Test for 2×2 contingency tables and multiple test the blots. Original Western blots of proteins shown in the “Results” correction were applied to find significant overrepresentation of GO section are presented in Supplementary Fig. 1. terms. 2.7. Immunohistochemistry and quantification of immunoreactivity 2.5. Reverse transcription and real-time PCR Formalin-fixed and paraffin-embedded tumor sections were Two micrograms of total RNA was reverse-transcribed using the deparaffinized, rehydrated and stained using the Vectastain Elite High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; ABC system (Vector Laboratories Inc.). Heat-induced antigen retrieval Foster City, CA, USA). Twenty-five nanograms of cDNA was amplified was performed using EDTA (for CD3 detection) or citrate buffer (for by real-time PCR with TaqMan methodology in the iCycler iQ Real- CD8 and CD68 detection). For staining of immune cells, anti-CD3 Time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA). (Abcam), anti-CD8 (Millipore) and anti-CD68 (Abcam) antibodies Specific Gene TaqMan Assays for Scd, Fabp3, Notch4, Pfkl, Il18, Il2Rg, were used at a 1:100 dilution. Sema3a, Jak2, Smad1, Stat5a, Cdkn2b, Arg1, Nos2, TGFβ1 and Hprt as For quantification of CD3 and CD8 immunostaining, 10 photo- the housekeeping control gene were obtained from Applied Biosys- graphs per tumor at ×400 magnification were taken, and CD-3 and tems. PCRs were performed at 10 min at 95°C followed by 15 s at 95°C CD8-positive cells were manually counted. The total number of cells and 60 s at 60°C for 40 cycles. Cycle thresholds (Ct) for each sample was quantified by Image J software, and results were expressed as the were obtained and 2−ΔCt calculated. percentage of positive cells with respect to the total number of cells in the sample. For CD68, due to the difficulty to define individual positive 2.6. Protein extraction and Western blot cells, a semiquantitative score system was used by two independent investigators, classifying the staining as low, medium or high. Tumor samples were homogenized in a buffer containing 50 mM Tris–HCl pH 7.2, 250 mM sucrose, 2 mM EDTA, 1 mM EGTA, 5 mM MgCl2, 50 μM NaF, 100 μM Na3VO4, 10 μl/ml protease inhibitor cocktail 2.8. Circulating levels of cytokines (Sigma-Aldrich), 10 mM β-mercaptoethanol and 1% Triton X-100. Protein extracts (10–20 μg) were subjected to SDS-PAGE on a 7.5% Determination of plasma levels of 27 cytokines was performed acrylamide using the Mini-Protean TGX Stain Free Gels and trans- with the LuminexMAP technology (Rat Cytokine/Chemokine Magnet- ferred to PVDF membrane with Trans-Blot Turbo Transfer Sytstem ic Bead Panel; Milliplex MAP Kit, Millipore) by immunoassay on the (Bio-Rad). Primary antibodies and dilutions used were anti-GLUT1 surface of fluorescent-coated magnetic beads following manufactur- (Cell Signaling Technology, 1:1000), anti-PFKL (Cell Signaling Tech- er's instructions. Cytokines analyzed were: GCSF, GMCSF (=CSF2), nology, 1:500), anti-ALDOA (Cell Signaling Technology, 1:1000), anti- leptin, EGF, VEGF, IL1α, IL1β, IL2, IL4, IL5, IL6, IL10, IL12 p70, IL13, NOTCH4 (Abcam, Cambridge, 1:5000), anti-Semaphorin 3A (Abcam, IL17α, IL18, IFNγ, MCP1 (=Ccl2), MIP1α (=Ccl3), RANTES (=Ccl5), 1:3000), anti-SMAD 1/5/9 (Abcam, 1:1000), anti-Phospho-SMAD 1/5/ eotaxin (=Ccl11), GRO-KC(=Cxcl1), MIP2 (=Cxcl2), LIX (=Cxcl5), 9 (Cell Signaling Technology, 1:500), anti-STAT5A/B (Cell Signaling IP10 (=Cxcl10), fractalkine (=Cx3cl1) and TNFα. Samples were run Technology) (1:3000), anti-CASP3 (Cell Signaling Technology, in duplicate. R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 221 2.9. Hormone levels group and 23.5% from LF-HCO showed intense reaction) and tumor necrosis (in 10.6% of those from LF-HCO was extensive). Tumors from Plasma levels of 17-β-estradiol and progesterone were determined LF-HCO also displayed the highest mitotic index (Fig. 2c). using Estradiol Concentration ELISA Kit and Progesterone Concentra- The results of carcinogenesis parameters are detailed in Table 1. tion ELISA Kit according to the manufacturer's directions (GenWay Latency time was shorter in high-corn-oil groups, although differences Biotech, Inc.). Plasma prolactin levels were measured by radioimmu- did not reach statistical significance. Tumor incidence was maximum noassay using 125I-prolactin (NEN) as the tracer, rat prolactin (rat PRL- (100%) in both HCO diet groups, significantly higher than control and RP-3) as the standard and an antibody rose against rat prolactin (anti- HOO. Tumor yield was higher in the corn-oil-enriched diet groups and rPRL-S-9) (Dr. AF Parlow, NIDDK National Hormone and Peptide LF-HOO versus the control. HCO, with the largest number of tumors, Program, CA, USA). was also statistically different to HOO. Finally, total tumor volume was higher in all high-fat diet groups, especially in LF-HCO. This 2.10. Metabolic parameters quantitative parameter cannot be statistically analyzed since there is one single value per group. Nonfasting plasma glucose and insulin levels were obtained at the end of the assay from animals fed ad libitum as an appropriate and 3.2. High-fat diets differentially modify the gene expression profile of the acceptable method based on rodent habits, the stress caused by fasting experimental tumors and the interference with other parameters [20]. Plasma levels of glucose were determined using an automated clinical analyzer in the In order to find genes modulated by dietary lipids, we compared Biochemistry and Molecular Genetics, Biomedical Diagnostic Center gene expression profiles of tumors from high-fat diet groups with (CDB), Hospital Clínic. Insulin levels were determined with an insulin those from control group. Lists of differentially expressed sequences enzyme-linked immunosorbent assay kit from Millipore. were further compared to identify genes modulated in common (co- modulated by effect of the high-fat diets) or unique to each specific 2.11. Statistical analysis group. Venn diagram in Fig. 3a depicts the number of overlapping and specific genes. In general, there were a low number of common genes. Statistical analyses for histopathological, clinical and molecular LF-HCO was the group with the largest number of specific genes. analyses other than microarrays were performed using SPSS software Analyses of biological significance (functional clustering) of all the 17.0. We used parametric or no-parametric statistics depending on the genes found as differentially expressed in each group are shown in distribution of each variable studied, determined by the Kolmogorov– Supplementary Table 3 (GO level 3). Some functional categories were Smirnov test, and the equality of variances, determined by Levene's significantly overrepresented in the clusters of down-modulated genes test. Body weight throughout the assay was analyzed using nonlinear in several groups: metabolism, response to stimuli or proliferation and mixed models. Parametric quantitative data were analyzed with cell cycle. Moreover, immune system categories were specifically ANOVA followed by Tukey's test and depicted as mean+S.E.M. overrepresented among the down-modulated genes in LF-HCO group. Nonparametric quantitative data were analyzed with Mann– Functional analysis was also performed dividing the clusters of genes Whitney U test and depicted as median, and mean+S.E.M. is in commonly modulated and specific for each group. Only genes specific presented in Supplementary Fig. 2. The χ2 test was used for the for LF-HCO and HOO groups showed significantly enriched categories. In qualitative data. The level of significance was established at Pb.05. particular, the genes down-modulated in LF-HCO group were mainly related to the immune system, response to stress and apoptosis (Fig. 3b– 3. Results c), whereas the up-modulated genes had a role in metabolism (Fig. 3d– e). Metabolism, cell homeostasis and proliferation were overrepresent- 3.1. High-fat diets differentially influence tumor malignant phenotype ed in the up-modulated genes in HOO group (Fig. 3f–g). and tumor growth 3.3. Validation of gene expression profiles and protein expression analyses Diagnosis of mammary lesions revealed a high percentage of malignant tumors among all the neoplasias collected (Fig. 2a). Thus, We selected for real-time PCR validation genes regulated in the percentages of adenocarcinomas in relation to all mammary common (Scd, Fabp3 and Notch4) or specific for each condition and tumors (benign and malignant) were: control LF 88.5% (46/52), HCO related to overrepresented functions. Gene expression results were 84.8% (100/118), LF-HCO 79.8% (87/109), HOO 80.6% (58/72) and LF- similar by arrays and real-time PCR when we compared the same six HOO 81.2% (82/101) (Supplementary Table 2). samples in each group (Supplementary Fig. 3). When real-time PCR In order to analyze the tumor aggressiveness, we studied the analyses were extended to 17–22 samples per group, a high variability differentiation pattern, nuclear pleomorphism and mitotic activity among tumors was observed. index of adenocarcinomas. Applying all these parameters, we scored Extended analysis is detailed in Fig. 4. In relation to the common the modified Scarff–Bloom–Richardson overall grade adapted to rat genes, Scd was down-modulated in high-fat diet groups, significantly mammary carcinomas [15]. Groups of animals fed with high-corn-oil in EVOO groups, while Fabp3 and Notch4 were significantly up- diet presented the most aggressive tumors, with significant differ- modulated in HCO and EVOO groups. Another metabolism gene, ences in comparison to control and LF-HOO groups (Fig. 2b). Thus, phosphofructokinase (Pfkl), was up-modulated in LF-HCO and HOO while more than 57% of the carcinomas from control showed low groups. Expression of genes related to immune system showed great aggressive grade (scores 3–5), fewer tumors in the high-corn-oil-fed variability with few significant differences between groups. The groups exhibited such grade (23% in HCO and 20.1% in LF-HCO), apoptosis-related gene Sema3a was significantly up-modulated in whereas groups fed the EVOO diet showed intermediate values (36.5% HOO group. Regarding the genes with a role in proliferation and cell in HOO and 41.7% in LF-HOO). On the contrary, the percentage of high- cycle control, we found differences in the expression of Jak2 (up- grade aggressive tumors (scores 9–11) in groups fed the corn-oil- modulated in HCO), Smad1 (down-modulated in high-EVOO diet enriched diet was 33% (HCO) and 39% (LF-HCO), in contrast to the groups) and Cdkn2b (up-modulated in LF-HOO), while a not 12.8% in the control group, and also the intermediate percentages of significant decrease in Stat5a levels was also observed in EVOO groups. HOO (21.2%) and LF-HOO (24%). Moreover, tumors from HCO diet In order to get insight into the overrepresented functions, expression groups showed the highest stromal reaction (24.7% tumors from HCO of metabolism and proliferation/apoptosis proteins was analyzed by 222 R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 Fig. 2. Effects of high-fat diets on the histopathological features of DMBA-induced mammary adenocarcinomas. (a) Representative images of adenocarcinomas from the different experimental groups (200× magnification). (b) Distribution of mammary adenocarcinomas of each group according to the modified Scarff–Bloom–Richardson index, which includes differentiation pattern, nuclear pleomorphism and mitotic index (solid line: Pb.05, dashed line: Pb.1, χ2 test). (c) Stromal reaction (− absent, + mild, ++ intense, +++ truly desmoplastic), necrosis and number of mitosis in 10 high-power fields (400× magnification); solid line: Pb.05, nonparametric Mann–Whitney U test. n=46 (LF), 100 (HCO), 87 (LF- HCO), 58 (HOO) and 82 (LF-HOO). Western blot. In relation to metabolism, validation of the glycolytic 3.4. High-fat diets differentially modulate tumor infiltration of cytotoxic enzyme phosphofruktokinase (PFKL) was performed, and the analysis lymphocytes and immunosuppressive mechanisms was extended to other proteins whose mRNA was found in the microarrays analysis as modulated by the high-fat diets: Glut1 (Slc2a1) Since no clear trend was observed in the expression of immune and Aldolase A (AldoA). Results showed an increase of Glut1 and PFKL system genes, to further investigate the role of immunological levels in tumors from HOO diet groups compared with those from HCO mechanisms on the effects of the high-fat diets in this experimental diet groups, while no differences in AldoA were found (Fig. 5a). model, we first evaluated tumor infiltration of leukocytes (Fig. 6a). Notch4, Sema3a, Stat5A/B and Smad1 as well as Caspase 3 and Infiltration of T lymphocytes (CD3-positive cells) was found to be PCNA were also investigated. No differences in Notch4 and PCNA similar between the different experimental groups. However, in- were observed (results not shown). The 90-kDa full-length isoform creased cytotoxic T lymphocytes (CD8-positive cells) infiltrate was of Sema3a was barely detected, but the 65-kDa cleaved fragment was observed in the LF-HOO group compared to the LF group. On the other decreased, in tumors from EVOO groups. No differences were hand, tumor-associated macrophages (CD68-positive cells) infiltra- observed in the total levels of Smad1/5/9 and StatA/B, but the tion was not influenced by the high-fat diets. Since these macrophages activated form of Smad (phopho-Smad 1/5/9) was increased in HCO and other populations such as myeloid-derived suppressor cells can versus HOO. On the other hand, total Caspase 3 (Casp3) and its suppress T cell proliferation and activity through production of activated 17-kDa cleaved form were significantly increased in the LF- arginase (Arg1), inducible nitric oxide synthase (Nos2) and trans- HOO group (Fig. 5b). forming growth factor β 1 (Tgfβ1), gene expression of these molecules Table 1 Carcinogenesis parameters LF HCO LF-HCO HOO LF-HOO Latency time (days, median) 97 71.5 68 89 89 Incidence (% of tumor-bearing animals) 80 (16/20)a 100 (20/20)b 100 (20/20)b 75 (15/20)a 85 (17/20)ab Tumor yield (total number of tumors) 46a 100b 87bc 58ac 82bc Total tumor volume (cm3) 69.3 105.0 188.2 121.9 100.6 Values within a row with different superscript letters are significantly different by Mann–Whitney U test (Pb.05). R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 223 Fig. 3. Effects of high-fat diets on gene expression profiles. (a): Number of sequences differentially expressed (Benjamini and Hochberg false discovery rate method) in adenocarcinomas (n=6/group) from animals fed the high-fat diets compared to control. Sequences differentially expressed were then compared between high-fat diet groups. Venn diagram depicts the number of specific and overlapping genes. Blue numbers indicate down-regulated genes, and red numbers indicate up-regulated genes. Italic numbers in parentheses indicate not annotated sequences (unknown genes). Numbers inside boxes highlight the lists of genes where overrepresented functional categories have been found. (b, d, f) Pie charts represent GO term for biological functions in genes down-regulated in LF-HCO (b), up-regulated in LF-HCO (d) and up-regulated in HOO (f) through PANTHER Classification System. Bold underlined terms indicate significantly overrepresented functions (Fisher's Exact Test). (c, e, g) GO term enrichment for biological functions in genes down-regulated in LF-HCO (c), up-regulated in LF-HCO (e) and up-regulated in HOO (g) through FatiGO-Babelomics software (Fisher's exact test). was evaluated in total extracts of tumor samples from the experi- induction (group LF-HCO). Body mass index was also higher in the mental groups. No differences were found in Nos2. The LF-HOO group HCO group at the end of the assay (Supplementary Fig. 4). The HOO showed lower expression of Arg1compared to both HCO groups and diet did not change body weight or mass. lower expression of Tgfβ1 compared to all other groups (Fig. 6b). In addition, there were no significant differences between groups in the metabolic parameters plasma glucose or insulin (Supplemen- 3.5. Dietary lipids influence circulating levels of inflammatory cytokines tary Fig. 5). There were also no significant differences in the plasma levels of the hormones estradiol, progesterone or prolactin (Supple- Inflammation is a well-known player on the cancer process; thus, mentary Fig. 6). circulating levels of different inflammation-related cytokines were evaluated by immunoassay. We first carried out a screening in a panel 4. Discussion of 27 analytes in 6 samples per group from which we selected 7 cytokines for further analyses (MCP1=Ccl2, IL18, GRO-KC=Cxcl1, In this work, we have observed a different influence of high-fat VEGF, fraktalkine=Cx3cl1, IL1α and leptin; n=17–19 per group). diets, depending on the type of fat and timing of dietary intervention, Results are shown in Fig. 6c. IL1α plasma levels were higher in the LF- on anatomopathological and clinical characteristics of experimental HCO group compared with the other high-fat diet groups. Leptin levels mammary tumors. The diet rich in n-6 PUFA, especially when were increased in all the high-fat diet groups in comparison to the administered after puberty onwards, clearly stimulated mammary low-fat control and especially in HCO which also had higher levels carcinogenesis and promoted tumors to a high degree of morpholog- than the LF-HOO group. No statistically significant differences were ical malignancy. The diet rich in EVOO oil exerted a weaker influence, found in the other cytokines. and tumors developed were more similar to those from low-fat controls. Gene expression profiles of tumors and validation analysis 3.6. Dietary lipids effects in other systemic factors also indicated different effects of the high-fat diets, likely being one of the molecular mechanisms of their influence on mammary carcino- Body weight throughout the study, as well as the body weight genesis and suggesting a differential influence of these diets on basic change compared to control, was increased in the rats fed with the cellular processes: metabolism, immune system function and prolif- HCO diet from weaning but not if dietary intervention started after eration/apoptosis pathways. Fig. 4. Real-time PCR analysis of genes found in the microarrays as modulated by the high-fat diets. Data represent relative gene expression (median) in the different experimental groups (n=17–22/group). Solid lines connecting groups indicate statistically significant differences (Pb.05); dashed lines indicate differences close to significance (Pb.1), Mann– Whitney U test. 224 R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 Fig. 5. Effects of high-fat diets on the expression of proteins related with metabolism and with proliferation/apoptosis pathways. (a) Representative Western blots and relative protein levels of glycolytic enzymes (PFKL, AldoA) and the glucose transporter Glut1 in the different experimental groups. (b) Representative Western blots and relative levels of proteins related to proliferation and apoptosis pathways. Bars represent medians. Solid lines connecting groups indicate statistically significant differences (Pb.05); dashed lines indicate differences close to significance (Pb.1), Mann–Whitney U test, n=8–20/group. ic refers to an internal control of pooled samples. Histopathologic evaluation of the mammary lesions developed in differential effect that this diet may exert depending on the timing of the DMBA-induced animals showed the predominance of malignant dietary exposure could be related to several factors. On the one hand, adenocarcinomas and confirmed their similarity with human tumors the administration of this diet from weaning advanced the puberty [15]. Benign and malignant tumors arise from different epithelial onset [14], thus modifying the window of susceptibility of the structures, and benign lesions are highly influenced by hormones such mammary gland [21] and therefore the progression of the carcino- as prolactin [21]. Although these high-fat diets have shown to genesis. Puberty is a particularly sensitive life stage in which dietary influence the puberty onset [14], which would affect the number of factors can have a great influence [14,22]. On the other hand, we found differentiated structures at the moment of induction, there were no an effect of this diet on the initiation of carcinogenesis through the significant differences between groups in the percentage of the modulation of detoxification enzymes [23,24]. In addition, the HCO malignant tumors in relation to all mammary tumors, as well as no diet has a clear promoter effect, already observed in previous works differences in hormone levels by effect of high-fat diets. [8,13,14,25,26] and in other experimental models [27–29]. The We characterized the tumor malignant phenotype determining the combination of all these effects would produce a greater initiation of tumor growth and the degree of morphological aggressiveness by cancer in the HCO group than in the LF-HCO, but of slower progression applying the modified Scarff–Bloom–Richardson grading method that (more tumors but with a slight increase in the latency time and of we previously described [15]. Results revealed that adenocarcinomas smaller volume, number of mitosis and necrosis). from animals fed with the high-corn-oil diet, especially from LF-HCO The effects of the high-fat diets through the caloric density should group, displayed morphologic characteristics of high malignancy, i.e., also be taken into consideration. Animals fed with the HCO diet from high pattern and nuclear grade, mitotic activity, stromal reaction and weaning did gain significant body weight and mass in the long term, necrotic areas, as well the overall histopathologic grade. Likewise, the which is in accordance with the highest leptin levels in the HCO group. study of clinical parameters of the carcinogenesis showed a clear Also interesting are the consequences that such diet may have to the stimulatory effect of the high-corn-oil diet since HCO and LF-HCO metabolic conditions associated to obesity. In this regard, there were groups had the earliest onset of tumor appearance and the highest no significant differences in plasma glucose or insulin levels between incidence and total malignant mammary tumors. However, the high- groups despite the increased body weight in HCO group. Several EVOO diet exerted a weaker stimulating effect on mammary known risk factors (such as early sexual maturation or obesity) can be carcinogenesis since tumors from animals fed such diet showed underlying mechanisms of the high-fat diet effects of tumorigenesis. intermediate values between control and the high-corn-oil groups in Nevertheless, the HCO diet had a clear promoting effect on the anatomopathological and clinical parameters. carcinogenesis in normal-weight animals (group LF-HCO). The stimulating effect that n-6 PUFA-enriched diet exerted on On the other hand, few studies have investigated the effects of mammary carcinogenesis was stronger during the promotion phase, EVOO on mammary carcinogenesis. We have previously observed given that tumors showed characteristics of higher malignancy when from a weak protective [3,13,25] to a weak promoting [14,26] effect of the dietary intervention started after induction (group LF-HCO). The the high-EVOO diet on experimental breast cancer. In any case, all the R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 225 Fig. 6. Effects of high-fat diets on immune system function. (a) Representative images (magnification ×400) and quantification of CD3, CD8 and CD68 immunostaining in the different experimental groups, n=18–20/group. (b) Gene expression of the immunosuppressive mediators arginase 1 (Arg1), nitric oxide synthase 2 (Nos2) and transforming growth factor β 1 (Tgfβ1), n=16–31/group. (c) Circulating levels of inflammation-related cytokines in plasma, n=17–19/group. Bars represent medians, except for CD68. For CCL2, IL18, CXCL1, VEGF and CX3CL1, dots represent medians and whiskers 25th–75th percentiles. Solid lines connecting groups indicate statistically significant differences (Pb.05), Mann–Whitney U test. studies indicated that the experimental tumors from animals fed with influence of these high-fat diets on the gene expression profile of this diet, despite being influenced by a high intake of total fat, showed mammary adenocarcinomas. Although the number of differentially clinical and histopathological characteristics of low aggressiveness, expressed genes was low, probably due to the intrinsic high similar to those from control group. Considering the unspecific variability among tumors, transcriptomic analysis showed different stimulating effect on carcinogenesis through caloric density [30] and effects of corn oil and EVOO. Biological significance analysis and that both experimental high-fat diets were isocaloric, the results functional clustering of differentially expressed genes revealed suggest a beneficial effect of the EVOO that partially counteracts the different enriched categories depending on the type of diet admin- deleterious effect of an excessive intake of fat. Differences observed istered. The high-n-6-PUFA diet down-modulated genes related to among the assays can be related to the different timing of the dietary cell death and immune response, whereas the high-EVOO diet up- intervention, as well as to distinct varieties of oil used, since it has been modulated genes with a role in cell death and proliferation. Moreover, reported that the influence of olive oil on carcinogenesis depends on when we divided clusters of genes in commonly modulated the proportions of oleic acid and minor bioactive components [31]. (influenced by both kinds of high-fat diets) or specific for each Several molecular mechanisms have been proposed in relation to group, among the genes only down-modulated in LF-HCO, apoptosis the effects of dietary fat on carcinogenesis. There is evidence that and immune system were also significantly overrepresented dietary fatty acids (FA) modulate biological functions through the categories. Metabolic processes were also enriched categories by regulation of gene expression [9,32]. Hence, we analyzed the both high-fat diets. 226 R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 Further expression analyses by real-time PCR have also suggested exhibit increased staining for CD68, the marker of macrophages, but metabolic adaptation to the increased intake of lipids in all high-fat differences between groups were not statistically significant. Never- diets groups, up-regulating the expression of the fatty acid binding theless, other studied markers [42] did suggest differences in the protein Fabp3 and down-regulating that of the desaturase enzyme immunosuppressive context of tumors by effect of the dietary Scd, especially in the EVOO groups. Extensive evidence has been intervention, resulting in higher levels of arginase in HCO groups, published regarding metabolic dysfunction of cancer cells, but there while in the LF-HOO group, Tgfβ1 was down-modulated. Moreover, are little data about how a high intake of FA probably determines the plasma levels of the proinflammatory and protumorigenic [43] metabolic adaptation of tumor cells. Transformed cells have a highly cytokine IL-1α and those of Leptin were increased in HCO group in increased need for FA, and it has been mostly considered that the main comparison with animals fed the high-EVOO diet. Taken together, source of FA in tumor cells was de novo synthesis [33]. However, our these results suggest that there is a modulation of the immune system expression results point out the importance of the extracellular source by dietary intervention that would be one of the mechanisms on the of lipids. Besides the adaptation to the high amount of unsaturated FA, differential effects of these high-fat diets on the tumor progression. our results indicate other metabolic modifications that depend on the Finally, results of microarrays profiling and validation by real-time PCR type of FA administered. Expression and Western blot analyses highlighted the molecular heterogeneity among tumors and within a suggested enhanced glycolytic pathways by effect of the high-EVOO single tumor. Beyond the limitations of each technique, both methodol- diet in comparison with the high-corn-oil diet, rising PFKL levels (a ogies showed concordant results in the same samples, while intertumor key point control of glycolytic pathway) and the glucose transporter variability was manifested when analyses were extended to a higher Glut1. Nevertheless, recently, the intratumoral compartmentalization number of samples. Thus, each tumor has a unique and also heteroge- of metabolic features and the distinct relevance if enzymes are neous molecular context with its single combination of genomic and expressed by cancer epithelial cells or cancer-associated fibroblasts epigenomic features, where environmental factors may exert subtle but [34] have been pointed out. Further studies are needed to elucidate the relevant influence. Taking all the molecular results together and significance of the effects of high-fat diets in the context of considering the effects of these diets on the anatomopathological and microenvironment and dynamic interplay among cell types. clinical characteristics of the mammary tumors, a scenario is plausible in On the other hand, several genes related to proliferation/apoptosis which tumors from animals fed the high-EVOO diet show modulation of pathways were detected as differentially modulated by the high-fat pathways conducting to Casp3 mediated apoptosis, perhaps enhanced by diets, although the variability among tumors resulted in subtle Sema3a activity (which could also participate in immunosurveillance). On differences, which were later analyzed at the protein level. Sema3a the other hand, a high intake of corn oil would induce more active protein has been related to apoptosis, migration and regulation of the proliferation pathways, e.g., increasing the active form of Smad1, and a tumor microenvironment including the immune cells [35]. While the more inflammatory phenotype. full-length mRNA was up-regulated in the HOO group, a proteolytic In conclusion, this work reports several effects of high-fat diets on 65-kDa isoform was significantly decreased. Although disparate data experimental breast carcinogenesis depending on the type and time of have been published, this isoform was originally reported as a dietary intervention, showing a clear stimulating influence of corn-oil- fragment with reduced functional activity [36]. Thus, the interpreta- enriched diet which is manifested on the morphological and clinical tion of the increased mRNA but decreased cleaved protein by effect of characteristics of tumors. The high-EVOO diet, despite having the EVOO is unclear but suggests modifications in apoptosis and/or unspecific promoting effect that all high-fat diets exert, seemed to immunosurveillance. In relation to proliferation genes, different have some beneficial influence eliciting a progression of tumors more trends have also been observed in expression analysis in concordance similar to that of the controls. Different and complex molecular with the clinical and morphological features of tumors. Smad1 mRNA mechanisms would drive such effects, including modifications in gene (a promoter of proliferation and migration [37]) was decreased in both expression profiles and changes in metabolic pathways, alteration in high-olive-oil groups. While little differences were observed in immunosurveillance and modifications in proliferation/apoptosis Smad1/5/9 protein levels, the active protein (phosphoSMAD1/5/9) balance. Thus, these results point out the relevance that dietetic was significantly increased in HCO compared to HOO, which was in habits from childhood may have on the susceptibility to malignant accordance with the higher degree of malignancy of tumors from HCO transformation and the progression of the disease. group. Some differences were also observed in the total levels of Supplementary data to this article can be found online at https:// Stat5A/B proteins, but they did not reach significance. We further doi.org/10.1016/j.jnutbio.2018.11.001. determined key proteins in apoptosis (Caspase 3 and its activated cleaved form), finding increased levels of activated Caspase 3 in tumors from LF-HOO which suggested an increase in apoptosis in such Acknowledgments tumors. In relation to immune system, it has been reported that the This work was supported by grants from “Plan Nacional I+D+I” expression of immune genes reflects leukocytes infiltration in tumor (AGL2006-07691), “Fundación Patrimonio Comunal Olivarero tissue [38]. However, although microarrays indicated down- (FPCO)” (FPCO2008-165.396; FPCO2013-CF611.084), “Agencia para modulation of Il18 and Il2Rg in LF-HCO group, extended real-time el Aceite de Oliva del Ministerio de Agricultura, Alimentación y Medio PCR analysis did not confirm such regulation. Since it is likely that Ambiente” (AAO2008-165.471), “Organización Interprofesional del subtle modifications in leukocyte infiltration may not be detected in Aceite de Oliva Español (OIAOE)” (OIP2009-165.646), “Departaments whole tumor extracts, immunohistochemical analyses were conduct- de Salut i d'Agricultura, Alimentació i Acció Rural de la Generalitat de ed. Thus, we observed an increase in the infiltration of cytotoxic T Catalunya” (GC2010-165.000), FPCO and OIAOE (FPCO-OIP2016- lymphocytes (CD8) in tumors from LF-HOO group. Other authors have CF614.087). The sponsors had no role in the study design, data also suggested an effect of an EVOO diet on production of CD8 collection and analysis and interpretation of the results. The authors lymphocytes and their infiltration into the colon mucosa of colon- wish to thank Dr. Montserrat Solanas, Marta Corral and Laia Llucià tumor-bearing rats [39], and immunomodulatory effects on B from the Department of Cell Biology, Physiology and Immunology, lymphocytes [40] and natural killer cells [41]. In addition, we also Universitat Autònoma de Barcelona, Barcelona (Spain), for their evaluated tumor-infiltrating macrophages, a cell population that plays collaboration in this work and Dr. Wladimiro Jiménez from Biochem- a key role in the generation and maintenance of an immunosuppres- istry and Molecular Genetics, Biomedical Diagnostic Center (CDB), sive microenvironment within the tumor. The groups fed the HCO diet Hospital Clínic, for the determination of plasma glucose. R. Escrich et al. / Journal of Nutritional Biochemistry 64 (2019) 218–227 227 References xenobiotic-metabolizing enzymes in the liver and in the mammary gland in a rat chemically induced breast cancer model. Eur J Cancer Prev 2015;24:215–22. https://doi.org/10.1097/CEJ.0000000000000032. [1] Global Burden of Disease Cancer CollaborationFitzmaurice C, Allen C, Barber RM, [24] Manzanares MA, de Miguel C, Ruiz de Villa MC, Santella RM, Escrich E, Solanas M. Barregard L, Bhutta ZA, et al. Global, regional, and national cancer incidence, Dietary lipids differentially modulate the initiation of experimental breast mortality, years of life lost, years lived with disability, and disability-adjusted life- carcinogenesis through their influence on hepatic xenobiotic metabolism and years for 32 cancer groups, 1990 to 2015: a systematic analysis for the global DNA damage in the mammary gland. J Nutr Biochem 2017;43:68–77. https://doi. burden of disease study. JAMA Oncol 2017;3:524–48. https://doi.org/10.1001/ org/10.1016/j.jnutbio.2017.01.016. jamaoncol.2016.5688. [25] Solanas M, Grau L, Moral R, Vela E, Escrich R, Escrih E. Dietary olive oil and corn oil [2] World Cancer Research Fund/American Institute for Cancer Research. Food, differentially affect experimental breast cancer through distinct modulation of the nutrition, physical activity, and the prevention of cancer: a global perspective. p21Ras signaling and the proliferation-apoptosis balance. Carcinogenesis 2010; Washington DC: American Institute for Cancer Research; 2007. 31:871–9. https://doi.org/10.1093/carcin/bgp243. [3] Escrich E, Solanas M, Moral R. Olive oil, and other dietary lipids, in cancer: [26] Moral R, Escrich R, Solanas M, Vela E, Ruiz de Villa MC, Escrich E. Diets high in corn experimental approaches. In: Quiles JL, Ramírez-Tortosa MC, Yaqoob P, editors. oil or extra-virgin olive oil differentially modify the gene expression profile of the Olive oil and health. Oxford: CABI Publishing; 2006. p. 317–74. mammary gland and influence experimental breast cancer susceptibility. Eur J [4] Escrich E, Solanas M, Moral R, Escrich R. Modulatory effects and molecular Nutr 2016;55:1397–409. https://doi.org/10.1007/s00394-015-0958-2. mechanisms of olive oil and other dietary lipids in breast cancer. Curr Pharm Des [27] MacLennan M, Ma DW. Role of dietary fatty acids in mammary gland 2011;17:813–30. https://doi.org/10.2174/138161211795428902. development and breast cancer. Breast Cancer Res 2010;12(211). https://doi. [5] Buckland G, Travier N, Cottet V, González CA, Luján-Barroso L, Agudo A, et al. org/10.1186/bcr2646. Adherence to the Mediterranean diet and risk of breast cancer in the European [28] De Assis S, Warri A, Cruz MI, Laja O, Tian Y, Zhang B, et al. High-fat or ethinyl- prospective investigation into cancer and nutrition cohort study. Int J Cancer oestradiol intake during pregnancy increases mammary cancer risk in several 2013;132:2918–27. https://doi.org/10.1002/ijc.27958. generations of offspring. Nat Commun 2012;3(1053). https://doi.org/10.1038/ [6] Buckland G, Gonzalez CA. The role of olive oil in disease prevention: a focus on the ncomms2058. recent epidemiological evidence from cohort studies and dietary intervention [29] Bartsch H, Naiir J, Owen RW. Dietary polyunsaturated fatty acids and cancers of trials. Br J Nutr 2015;113(Suppl. 2):S94–101. https://doi.org/10.1017/ the breast and colorectum: emerging evidence for their role as risk modifiers. S0007114514003936. Carcinogenesis 1999;20:2209–18. [7] Willett WC, Sacks F, Trichopoulou A, Drescher G, Ferro-Luzzi A, Helsing E, et al. [30] Guthrie N, Carroll KK. Specific versus non-specific effects of dietary fat on Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr carcinogenesis. Prog Lipid Res 1999;38:261–71. https://doi.org/10.1016/S0163- 1995;61:1402S–6S. 7827(99)00006-5. [8] Costa I, Moral R, Solanas M, Escrich E. High-fat corn oil diet promotes the [31] Cohen LA, Epstein M, Pittman B, Rivenson A. The influence of different varieties of development of high histologic grade rat DMBA-induced mammary adenocarci- olive oil on N-methylnitrosourea(NMU)-induced mammary tumorigenesis. nomas, while high olive oil diet does not. Breast Cancer Res Treat 2004;86: Anticancer Res 2000;20:2307–12. 225–35. https://doi.org/10.1023/B:BREA.0000036896.75548.0c. [32] Hammamieh R, Chakraborty N, Miller SA, Waddy E, Barmada M, Das R, et al. [9] Jump DB. Fatty acid regulation of gene transcription. Crit Rev Clin Lab Sci 2004;4: Differential effects of omega-3 and omega-6 fatty acids on gene expression in 41–78. https://doi.org/10.1080/10408360490278341. breast cancer cells. Breast Cancer Res Treat 2007;101:7–16. https://doi.org/10. [10] Escrich E, Moral R, García G, Costa I, Sánchez JA, Solanas M. Identification of novel 1007/s10549-006-9269-x. differentially expressed genes by the effect of a high-fat n-6 diet in experimental [33] Kinlaw WB, Baures PW, Lupien LE, Davis WL, Kuemmerle NB. Fatty acids and breast cancer. Mol Carcinog 2004;40:73–8. https://doi.org/10.1002/mc.20028. breast cancer: make them on site or have them delivered. J Cell Physiol 2016;231: [11] Escrich E, Solanas M, Segura R. Experimental diets for the study of lipid influence 2128–41. https://doi.org/10.1002/jcp.25332. on induced mammary carcinoma in rats: I — diet definition. In Vivo 1994;8: [34] Penkert J, Ripperger T, Schieck M, Schlegelberger B, Steinemann D, Illig T. On 1099–106. metabolic reprogramming and tumor biology: a comprehensive survey of [12] Escrich E, Solanas M, Ruiz de Villa MC, Ribalta T, Muntane J, Segura R. metabolism in breast cancer. Oncotarget 2016;7:67626–49. https://doi.org/10. Experimental diets for the study of lipid influence on induced mammary 18632/oncotarget.11759. carcinoma in rats: II — suitability of the diets. In Vivo 1994;8:1107–12. [35] Capparuccia L, Tamagnone L. Semaphorin signaling in cancer cells and in cells of [13] Solanas M, Hurtado A, Costa I, Moral R, Menéndez JA, Colomer R, et al. Effects of a the tumor microenvironment — two sides of a coin. J Cell Sci 2009;122:1723–36. high olive oil diet on the clinical behavior and histopathological features of rat https://doi.org/10.1242/jcs.030197. DMBA-induced mammary tumors compared with a high corn oil diet. Int J Oncol [36] Adams RH, Lohrum M, Klostermann A, Betz H, Puschel AW. The chemorepulsive 2002;21:745–53. https://doi.org/10.3892/ijo.21.4.745. activity of secreted semaphorins is regulated by furin-dependent proteolytic [14] Moral R, Escrich R, Solanas M, Vela E, Costa I, Ruiz de Villa MC, et al. Diets high in corn processing. EMBO J 1997;16:6077–86. https://doi.org/10.1093/emboj/16.20. oil or extra-virgin olive oil provided from weaning advance sexual maturation and 6077. differentially modify susceptibility to mammary carcinogenesis in female rats. Nutr [37] Liu IM, Schilling SH, Knouse KA, Choy L, Derynck R, Wang XF. TGFbeta-stimulated Cancer 2011;63:410–20. https://doi.org/10.1080/01635581.2011.535956. Smad1/5 phosphorylation requires the ALK5 L45 loop and mediates the pro- [15] Costa I, Solanas M, Escrich E. Histopathologic characterization of mammary migratory TGFbeta switch. EMBO J 2009;28:88–98. https://doi.org/10.1038/ neoplastic lesions induced with 7,12-dimethylbenz(α)antracene in the rat. A emboj.2008.266. comparative analysis with human breast tumours. Arch Pathol Lab Med 2002;126: [38] Dedeurwaerder S, Fuks F. DNA methylation markers for breast cancer prognosis: 915–27. https://doi.org/10.1043/0003-9985(2002)126b0915:HCOMNLN2.0.CO;2. unmasking the immune component. Oncoimmunology 2012;1:962–4. https:// [16] Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, et al. doi.org/10.4161/onci.19996. Exploration, normalization, and summaries of high density oligonucleotide array probe [39] Kossoy G, Madar Z, Ben-Hur H, Gal R, Stark A, Cohen O, et al. Transplacental effect level data. Biostatistics 2003;4:249–64. https://doi.org/10.1093/biostatistics/4.2.249. of a 15% olive-oil diet on functional activity of immune components in the spleen [17] Smyth GK. Linear models and empirical Bayes methods for assessing differential and colon tumors of rat offspring. Oncol Rep 2001;8:1045–9. https://doi.org/10. expression in microarray experiments. Stat Appl Genet Mol Biol 2004;3:3. https:// 3892/or.8.5.1045. doi.org/10.2202/1544-6115.1027. [40] Zusman I, Ben-Hur H, Budovsky A, Geva D, Gurevich P, Tendler Y, et al. [18] Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and Transplacental effects of maternal feeding with high fat diets on lipid exchange powerful approach to multiple testing. J R Stat Soc B 1995;1:289–300. and response of the splenic lymphoid system in mice offspring exposed to low [19] Gürtler A, Kunz N, Gomolka M, Hornhardt S, Friedl AA, McDonald K, et al. Stain- doses of carcinogen. Int J Mol Med 2000;6:337–43. https://doi.org/10.3892/ijmm. free technology as a normalization tool in Western blot analysis. Anal Biochem 6.3.337. 2013;433:105–11. https://doi.org/10.1016/j.ab.2012.10.010. [41] Leung KH, Ip MM. Effect of dietary polyunsaturated fat and 7,12-dimethylbenz [20] Jensen TL, Kiersgaard MK, Sørensen DB, Mikkelsen LF. Fasting of mice: a review. (a)-anthracene on rat splenic natural killer cells and prostaglandin E synthesis. Lab Anim 2013;47:225–40. https://doi.org/10.1177/0023677213501659. Cancer Immunol Immunother 1986;21:161–3. [21] Russo IH, Russo J. Mammary gland neoplasia in long-term rodent studies. Environ [42] Draghiciu O, Lubbers J, Nijman HW, Daemen T. Myeloid derived suppressor cells- Health Perspect 1996;104:938–67. an overview of combat strategies to increase immunotherapy efficacy. Oncoim- [22] Aupperlee MD, Zhao Y, Tan YS, Zhu Y, Langohr IM, Kirk EL, et al. Puberty-specific munology 2015;4:e954829. https://doi.org/10.4161/21624011.2014.954829. promotion of mammary tumorigenesis by a high animal fat diet. Breast Cancer Res [43] Tjomsland V, Spangeus A, Valila J, Sandstrom P, Borch K, Druid H, et al. Interleukin 2015;17:138. https://doi.org/10.1186/s13058-015-0646-4. 1alpha sustains the expression of inflammatory factors in human pancreatic [23] Manzanares MA, Solanas M, Moral R, Escrich R, Vela E, Costa I, et al. Dietary extra- cancer microenvironment by targeting cancer-associated fibroblasts. Neoplasia virgin olive oil and corn oil differentially modulate the mRNA expression of 2011;13:664–75. https://doi.org/10.1593/neo.11332.