Dual Fatty Acid Synthase and HER2 Signaling Blockade Shows Marked Antitumor Activity against Breast Cancer Models Resistant to Anti-HER2 Drugs

Blocking the enzyme Fatty Acid Synthase (FASN) leads to apoptosis of HER2-positive breast carcinoma cells. The hypothesis is that blocking FASN, in combination with anti-HER2 signaling agents, would be an effective antitumor strategy in preclinical HER2+ breast cancer models of trastuzumab and lapatinib resistance. We developed and molecularly characterized in vitro HER2+ models of resistance to trastuzumab (SKTR), lapatinib (SKLR) and both (SKLTR). The cellular interactions of combining anti-FASN polyphenolic compounds (EGCG and the synthetic G28UCM) with anti-HER2 signaling drugs (trastuzumab plus pertuzumab and temsirolimus) were analyzed. Tumor growth inhibition after treatment with EGCG, pertuzumab, temsirolimus or the combination was evaluated in two in vivo orthoxenopatients: one derived from a HER2+ patient and another from a patient who relapsed on trastuzumab and lapatinib-based therapy. SKTR, SKLR and SKLTR showed hyperactivation of EGFR and p-ERK1/2 and PI3KCA mutations. Dual-resistant cells (SKLTR) also showed hyperactivation of HER4 and recovered levels of p-AKT compared with mono-resistant cells. mTOR, p-mTOR and FASN expression remained stable in SKTR, SKLR and SKLTR. In vitro, anti-FASN compounds plus pertuzumab showed synergistic interactions in lapatinib- and dual- resistant cells and improved the results of pertuzumab plus trastuzumab co-treatment. FASN inhibitors combined with temsirolimus displayed the strongest synergistic interactions in resistant cells. In vivo, both orthoxenopatients showed strong response to the antitumor activity of the combination of EGCG with pertuzumab or temsirolimus, without signs of toxicity. We showed that the simultaneous blockade of FASN and HER2 pathways is effective in cells and in breast cancer models refractory to anti-HER2 therapies.


Introduction
The human epidermal growth factor receptor 2 (HER2) is amplified or overexpressed in * 20% of human breast carcinomas and is associated with a more aggressive phenotype and worse prognosis [1].
HER receptors family is composed of four closely related tyrosine kinase (TK) receptors: HER1 (EGFR), HER2, HER3, and HER4. Dimerization of HER receptors, induced by ligand binding or receptor overexpression in the case of HER2, leads to the recruitment of several adaptor proteins that mediate the activation of downstream signaling pathways [2,3]. Among them, the phosphoinositide 3-kinase (PI3K)/protein kinase B (PKB/AKT)/mammalian target of rapamycin (mTOR) protein and the mitogen activated protein kinases (MAPK or ERK1/2) pathways promote cell proliferation, transformation, and survival [4,5].
HER2-overexpressing tumors are sensitive to monoclonal antibodies (mAb) and small-molecule TK inhibitors (TKI) that interfere with HER2 function and signaling [6][7][8]. Trastuzumab, a humanized mAb directed against the extracellular domain of the receptor, was the first approved therapy for the treatment of HER2-positive (HER2+) breast cancer. Despite the considerable clinical benefit provided, a large fraction of HER2+ tumors display primary or acquired resistance to trastuzumab [9]. Lapatinib, a small-molecule TKI targeting the intracellular tyrosine kinase domain of EGFR and HER2, was found to improve time to progression in HER2 breast cancer patients who had progressed to tratuzumab [7]. Lapatinib is administered alone or in combination with trastuzumab to abolish the activation of HER2-downstream pathway. But unfortunately, some tumors develop lapatinib resistance and also resistance against the combination of both drugs [10]. The molecular mechanisms leading to trastuzumab and lapatinib resistance has been extensively studied [11]. These include for example in vivo conversion of HER2+ to HER2-carcinoma after neoadjuvant trastuzumab [12], predominance of the constitutively active HER2 form (p95 HER2 ) [8], overexpression or hyperactivation of other HER family receptors or its ligands [13], amplification of the PI3K/AKT/mTOR pathway by loss of phosphatase and tensin homolog (PTEN) [14], gain-of-function mutation in PI3KCA (encoding the PI3K catalytic isoform p110α) [15] and AKT mutations or amplifications [16].
Fatty acid synthase (FASN) is a homodimeric multienzymatic protein that catalyzes de novo synthesis of long-chain fatty acids [17]. Blocking FASN activity causes in vitro and in vivo anticancer activity in several overexpressing FASN human carcinomas [18,19]. The proposed oncogenic properties of FASN seem to be the result of an increased activation of HER2 and its dowstream related PI3K/AKT/mTOR and MAPK signaling pathways [18][19][20]. FASN can also inhibit the intrinsic pathway of apoptosis [21], may also contribute to modulation of the membrane lipid rafts that anchor HER2 [22] and has been recently proposed as a direct target of p53 family members, including p63 and p73 [23]. In the past, FASN inhibitors with antitumour activity have been limited by either cross-activation of β-oxidation, which produces in vivo anorexia and body weight loss [24,25], or low potency [26,27]. We have developed new polyphenolic anti-FASN compounds that exhibit in vitro and in vivo anticancer activity improving the antitumor efficacy and the toxic effects of classical FASN inhibitors, in HER2+ breast cancer cells and mouse models [19,28,29]. Among of them, G28UCM has shown a strong antitumor effect, alone or in combination with anti-HER drugs, in HER2+ breast cancer cells and on breast cancer cells resistant to trastuzumab [29].
In this study, we have investigated the anticancer activity of the classical FASN inhibitor epigallocathequin-3-gallate (EGCG) and G28UCM, as single agents or in combination with pertuzumab and temsirolimus, in our developed trastuzumab (SKTR), lapatinib (SKLR) and trastuzumab plus lapatinib (SKTLR) resistant HER2+ breast cancer models. In addition, we analyzed the antitumor activity of EGCG, alone or in combination, in two in vivo xenografts: one HER2+ patient and another from a HER2+ patient who fail to respond to trastuzumab and lapatinib therapies.

Materials and Methods
Cell culture and development of long-term resistant breast cancer cells SKBr3 (SK) breast carcinoma cells were obtained from Eucellbank (University of Barcelona) [30]. SKBr3 cells were routinely grown in McCoy's (Gibco) supplemented with 10% FBS (HyClone Laboratories), 1% L-glutamine, 1% sodium pyruvate, 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco). Trastuzumab-resistant SK cells (SKTR) were developed by exposing SK cells continuously to trastuzumab (Herceptin, Hoffmann-La Roche Pharma), starting with 1μM concentration for three months of exposure and increasing the concentration up to 2 μM for a 12 months period, as we previously described [29]. Thus, cells resistant to trastuzumab were maintained in 2 μM trastuzumab, a concentration at which SK parental cells were not viable. To develop lapatinib-resistant cells (SKLR), SK cells were treated for one month with an initial dose of 1.5 μM of lapatinib (GW572016; Tykerb, GlaxoSmithKline) and after one month the dose of lapatinib was increased up to 3 μM for 12 months as we described [29], a concentration at which SK parental cells were not viable. To develop lapatinib plus trastuzumab resistant cells (SKLTR), SKLR were co-cultured with lapatinib 3 μM and trastuzumab 1μM and after one month in culture the dose of trastuzumab was increased up to 2 μM. Cells were co-cultured with lapatinib and trastuzumab for 12 months. SKLTR cells were mantained with 3 μM of lapatinib and 2 μM of trastuzumab. Trastuzumab, lapatinib and trastuzumab plus lapatinib resistance was confirmed by dose-response studies using the standard colorimetric MTT assay as we describe in S1 File. Cell line authentication was performed with STR analysis in an external laboratory (Genetica DNA Laboratories) (S2 File). Parental and resistant cells shared 100% STR profile with SKBr3 cell line.

Genetic analysis of PI3K mutations
DNA was extracted from SK, SKTR, SKLR and SKLTR cells following commercial protocols (QIAamp DNA blood Mini kit, Qiagen). Subsequently, polymerase chain reaction (PCR) was used to amplify the PI3K gene (NM_006218) (NCBI-National Center for Biotechnology Information). PCR products were purified using ExoSAP-IT (Isogen Life Science), and the analysis of the exonic and intron-exon regions was performed forward/reverse by direct sequencing (Genetic Analyzer 3130XL, Applied Biosystems).

Quantitative real-time PCR analysis of HER ligands
Parental and resistant cells were washed with PBS and trypsinized. Total-RNA from each sample was isolated using RNeasy mini kit (Qiagen). RNA was reverse-transcribed into complementary DNA (cDNA) using High Capacity cDNA Archive Kit (Applied Biosystems). HER ligands expression (EGF, TGF-α, AR, BTC, EREG, NRG-1 and HB-EGF) was quantified by real-time PCR using a pre-designed, gene-specific TaqMan probe and primer sets (TaqMan Gene Expression assays, Applied Biosystems). Quantitative PCR was performed using TaqMan One-Step Universal Master Mix (Applied Biosystems) and 7300 Real-Time PCR system (Applied Biosystems). All samples were tested in triplicate. Relative quantification of the mRNA level (μg/ml) of HER ligands was carried out. Then, mRNA level was normalized to the housekeeping gene TATA box binding (TBP) protein.

Growth inhibition and dose-response studies
Parental and resistant cells were plated out at a density of 5 x 10 3 cells/100 μL/well in 96-well microtitre plates. Following overnight cell adherence fresh medium along with the corresponding concentrations of HER2 inhibitors (trastuzumab and pertuzumab [2C4, Perjeta, Genentech]), FASN inhibitors (EGCG [Sigma] and 1,3-bis((3,4,5-thilhydroxybenzoil)oxy) naphthalene (G28UCM) synthesized as we previously described [19]) or mTOR inhibitor (temsirolimus; CCI-779, Torisel, Pfizer) was added to the cultures. Pertuzumab (5 μg/ml) was combined with trastuzumab (20 μM) or FASN inhibitors (60 μM of EGCG or 5 μM of G28UCM) for 5 days. For temsirolimus drug-combination experiments cells were treated for 2 days with a dose curve concentration of EGCG (5-300 μM) or G28UCM (0.1-15 μM) plus fixed concentrations of temsirolimus (0.05, 0.1, 0.5 and 1 μM). Same treatments were assessed in monotherapy. Following treatment, media was replaced by drug-free medium (100 μL/well) containing MTT (3,4,5-dimethylthiazol-2-yl-2,5-diphenyltetrazolium bromide, Sigma) solution, and incubation was prolonged for 3 h at 37°C. Formazan crystals formed by metabolically viable cells were dissolved in DMSO (100 μL/well) and absorbance was determined at 570 nm in a multi-well plate reader (Model Anthos Labtec 2010 1.7). Using control OD values (C) and test OD values (T), % of cell proliferation inhibition (% cpi) was calculated from the equation, 100 -[(T x 100) / C]. Data presented are from three separate wells per assay and the assay was performed at least three times. Combinatorial effects were evaluated using the ratio of % cpi produced by each drug alone vs % cpi produced by drug combination (% cpi drug / % cpi combination), average of both ratios was calculated to know the effect of combination compared with both compounds alone. Interactions of G28UCM and EGCG with temsirolimus were also evaluated by the isobologram method as we previously published [29]. Briefly, the concentration of one agent producing a 30% inhibitory effect is plotted on the horizontal axis, and the concentration of another agent producing the same degree of effect is plotted on the vertical axis; a straight line joining these two points represents zero interaction (addition) between two agents. The experimental isoeffect points were the concentrations (expressed relative to the IC 30 concentrations) of the two agents that when combined kill 30% of the cells. When the experimental isoeffect points felt below that line, combination effect of the 2 drugs was considered to be supra-additive or synergistic, whereas antagonism occurs if the experimental isoeffect points lie above it. Within the designed assay range, a set of isoeffect points was generated because there were multiple FASN inhibitors and anti-target agent concentrations that achieved the same isoeffect. A quantitative index of these interactions was provided by the equation Ix = (A/a) + (B/b), where, for this study, a and b represent the respective concentrations of FASN inhibitors (EGCG or G28UCM) and anti-mTOR agent (temsirolimus) required to produce a fixed level of inhibition (IC 30 ) when administered alone, and A and B represent the concentrations required for the same effect when the drugs were administered in combination, and Ix represents an index of drug interaction (interaction index). Ix values of <1 indicate synergism, a value of 1 represents additivism, and values of >1 indicate antagonism. For all estimations of Ix, we used only isobolos where intercept data for both axes were available.

In vivo studies: human breast tumor PDX (patient-derived xenografts) experiments
Tumor chunks from HER2+ breast cancer patient and HER2+ patient who relapsed after trastuzumab and lapatinib-based treatment were orthotopically implanted into both inguinal cleared mammary fat pads of NOD/SCID (Harlan Laboratories, Inc.) or NSG (NOD/SCID; IL2Rγ−/−) mice (Charles River Laboratories). When tumors reached 30-60 mm 3 , animals were randomized into treatment-groups. Each group received a single intraperitoneal (i.p.) injection (maximum of 0.2 mL) of control (vehicle alone 3d/wk), 30 mg/kg EGCG 3d/wk, 30 mg/kg pertuzumab 1d/w, 10 mg/kg temsirolimus 1d/w, or combination of EGCG + pertuzumab and EGCG + temsirolimus. Tumor xenografts were measured with calipers and tumor volumes were determined using the formula: (π/6 x (v1 x v2 x v2)), where v1 represents the largest tumor diameter, and v2 the smallest one. Body weight was registered 2-3 d/wk. At the end of the experiment, animals were weighed and then euthanized using CO 2 inhalation. Tumors and serum were stored at -80°C. Lung, heart, liver, kidneys and tumor were fixed with formalin (S1 File). Apoptosis in control and treated tumors was analyzed by fluorescent TUNEL assay. Briefly, DNA fragmentation by terminal deoxynucleotidyl transferase-mediated dUTP-biotin nickend labeling (TUNEL) was performed according to the manufacturer's instructions (in situ cell death detection, Roche). Nuclei contrast was performed using fluorescent DAPI staining (4 0 -6 0 -Diamidino-2-phenylindole, Sigma). Pictures shown are representative of two samples per each treatment group.

Ethics Statement
Experiments were conducted in accordance with guidelines on animal care and use established by Biomedical Research Institute of Bellvitge (IDIBELL) Institutional Animal Care and Scientific Committee (protocol EST-FOR-070.03). All mice were maintained in a specific pathogenfree AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care International) International accredited facility in accordance with Spanish and European regulations with controlled light/dark cycle, temperature, and humidity. As approved by the abovementioned committee, all surgery was performed under inhaled isoflurane anesthesia, burprenorphine was administered as analgesic to mice after surgery, all mice were euthanized with CO 2 asphyxiation at the end of the experiment or when tumors reached 1 cm of diameter, and all efforts were made to minimize suffering.

Statistical analysis
Data were analyzed by Student t test when comparing two groups or ANOVA using a Bonferrony test as post-test when comparing more than 2 groups. Statistical significant levels were p < 0.05 (denoted as Ã ), p < 0.01 (denoted as ÃÃ ) and p < 0.001 (denoted as ÃÃÃ ). p-value is shown in results when significance is reached (p < 0.05). All data are means ± standard error (SE). All observations were confirmed by at least three independent experiments.

Characterization of trastuzumab (SKTR), lapatinib (SKLR) and lapatinib plus trastuzumab-resistant (SKLTR) breast cancer cells
As preclinical models of acquired resistance to anti-HER2 drugs, we developed a panel of resistant HER2+ breast cancer cells (SK) with long-term (12 months) and high drug concentration exposure of trastuzumab (SKTR), lapatinib (SKLR) and lapatinib plus trastuzumab (SKLTR) (S1 Fig), following Nahta R et al. methodology [31]. To elucidate molecular mechanisms regarding acquired resistance in our developed resistant cells (SKTR, SKLR and SKLTR) we first examined HER2 gene amplification by fluorescence in situ hybridation. The ratio of the average HER2 gene copy number to the average CEP17 gene copy number in SK was 3.53 and in SKTR, SKLR and SKLTR was 6.32, 3.99 and 4.84, respectively ( Fig 1A). These results showed that resistant cells possess HER2 amplification, similar as parental cells.
We next analyzed changes in HER family protein receptors and their downstream proteins related to PI3K/AKT/mTOR and MAPK/ERK1/2 pathways. As shown in Fig 1B, trastuzumab  cells (SKTR) had a significant increase in EGFR and phosphorylated EGFR proteins and to a lesser extent in p-ERK1/2 and AKT, with a noticeably decreased p-HER2 protein and reduced levels of HER3, p-HER3 and p-AKT compared to SK control cells. Cells resistant to lapatinib (SKLR) showed a great increase in p-EGFR and p-HER2 proteins and a slightly increase in p-ERK1/2, whereas levels of HER3, p-HER3 and p-AKT were decreased compared to SK control cells. Lapatinib plus trastuzumab-resistant cells (SKLTR) showed increased expression of phosphorylated forms of HER2, EGFR, HER4 and AKT compared to its control lapatinib-resistant cells (SKLR). Interestingly, no significant changes in mTOR and p-mTOR protein levels were observed in any resistant cells compared to SK cells. Regarding FASN, which transcription and translation is mediated by HER2 signaling pathway [29,32], protein expression levels showed no changes in resistant cells.
Mutation of PI3K is another described mechanism of resistance to anti-HER2 treatments. We found low incidence of activating mutations in the p110α catalytic subunit of PI3K (PIK3CA) in all resistant cells. We detected the PIK3CA_E545K mutation in SKTR cells and the PIK3CA_E542K mutation in SKLR and its derivative SKLTR (Fig 1C).
Changes in crosstalk between receptors of HER family prompted us to investigate the mRNA expression profile of several HER activating ligands by real-time PCR in resistant and parental cells (Fig 1D). HB-EGF (heparin-binding EGF-like growth factor) mRNA expression was significantly up-regulated (more than 3-fold; p-value: 0.012) in trastuzumab resistant cells (SKTR) compared to SK control cells. In contrast, AR (amphiregulin; p-value: 0.024) was down-regulated, and TGF-α (transforming growth factor-α) and BTC (beta-cellulin) expression was slightly down-regulated in SKTR versus SK cells. Otherwise, SKLR cells up-regulated EGF (epidermal growth factor) (1.5 folds compared with SK; p-value: 0.045), TGF-α (nearly 3 folds; p-value: 0.044), EREG (epiregulin) (4.5 folds) and HB-EGF (more than 2.5 folds; p-value: 0.014), but down-regulated AR and BTC expression in about 0.5 folds, comparing with SK. Double-resistant (SKLTR) cells showed a great up-regulation in almost all ligands (EGF in more than 6 folds, AR in almost 5 folds (p-value: 0.011), BTC and EREG in 3 folds and HB-EGF almost 2 folds), only TGF-α was slightly down-regulated in 0.5 folds, comparing with SK cells.
Since some studies reported that HER2 mediates tumor growth and metastasis [33,34], and several molecular changes (that could alter tumor aggressiveness) occurred on HER2 and its downstream pathways, we conducted experiments to evaluate such hallmark by measuring cell invasion and adhesion to extracellular matrix (opening metastatic event) capacities of our developed resistant cells (Fig 1E). Moreover, several types of resistance such as some chemotherapeutic drugs and multidrug resistance (combining several natural chemotherapeutic drugs) have been associated with cancer invasion and metastasis [35,36]. Hence, we wanted to know if mechanisms of resistance in our anti-HER2 resistant models also turned to a more aggressive phenotype of tumor cells. SKTR cells showed a huge significantly increased capacity to adhere to extracellular matrix proteins: fibronectin (2.9 fold-change; p-value: 0.002), collagen I (3.3 fold-change; p-value: 0.013), collagen IV (4.4 fold-change; p-value: 0.043), laminin I (5.8 fold-change; p-value: 0.001) and a slightly increase in fibrinogen adherence (1.3 fold-change) compared with SK parental cells. SKLR cells had a remarkable increased adherence to fibronectin (1.5 fold-change; p-value: 0.031), collagen I (2 fold-change; p-value: 0.003) and collagen IV (2.3 fold-change; p-value: 0.005) and to a lesser extent to laminin I (1.4 fold-change) and fibrinogen (1.1 fold-change) compared with SK cells. SKLTR showed a significant increased capacity to adhere to fibronectin (1.6 fold-change; p-value: 0.011) and collagen I (1.2 fold-change; pvalue: 0.030), a slight adherence to laminin I (1.4 fold-change) and fibrinogen (1.3 foldchange), whereas adhesion to collagen IV was almost unchanged, compared to SK. Regarding the invasion capacity, both SKTR and SKLR cells showed relevant increased capacity than SK cells [1.4 and 1.6 (p-value: 0.011) folds more, respectively]. Conversely, SKLTR cells showed half invasion capacity (0.5 folds; p-value: 0.025) than SK cells (Fig 1E).
In vitro cell growth interactions between HER2 inhibitors, trastuzumab and pertuzumab, in SK, SKTR, SKLR and SKLTR SKTR, SKLR and SKLTR cells maintained HER2 protein expression levels similar to parental cells (Fig 1B). Therefore, we first checked the effects of the HER2-dimerization inhibitor, pertuzumab, in comparison with trastuzumab in resistant cells. As shown in

In vitro cell growth interactions between FASN inhibitors and pertuzumab in SK, SKTR, SKLR and SKLTR
Since targeting only HER2 was far from obtaining desired results in resistant models, we decided to explore dual targeting pharmacological strategies. Because FASN showed similar expression levels in parental and resistant cells (Fig 1B), this could be a candidate target to overcome anti-HER2 resistance. Thus, we conducted series of combinatory experiments to evaluate the inhibitory effect of EGCG and G28UCM alone and in combination with pertuzumab in SK, SKTR, SKLR and SKLTR cells. The natural anti-FASN compound EGCG had similar cytotoxic effect in parental and resistant cells. IC 50 values ranged from 206 ± 18.7 μM to 229 ± 29.4 μM in SK, SKTR, SKLR and SKLTR cells. G28UCM, the synthetic derivative of EGCG, improved the cytotoxic effect of EGCG in all cell lines. IC 50 value of G28UCM in parental and resistant cells ranged from 9 ± 1.5 μM to 19 ± 2.1 μM (S1 Table). In addition and according to our previous results in HER2+ parental cells, EGCG and G28UCM induced apoptosis (cleavage of PARP) in resistant cells (S2 and S3 Figs).
Results regarding pertuzumab combinatory experiments with EGCG and G28UCM are shown in Fig 3A. Pertuzumab (5 μg/ml) combined with anti-FASN compounds (60 μM of EGCG or 5 μM of G28UCM) increased cpi in parental and resistant cells. Ratios of cpi induced for treatments alone versus inhibition induced for co-treatment was less than 1 in all combinatory experiments. In SK cells, ratio of mono-treatments/combination was 0.34. In SK, SKLR and SKTR, ratio of mono-treatments/combination was 0.34, 0.82 and 0.62, respectively. In SKLTR, monotreatments/combination ratio was 0.31. In this case, pertuzumab plus EGCG combination significantly improved effects of each treatment alone (which is graphed as 1 in Fig 3; p-value compared with 1 : 0.036). G28UCM plus pertuzumab slightly improved EGCG plus pertuzumab inhibitory effects in SKTR cells (0.69 when combined with G28UCM compared with 0.82 in EGCG case) (Fig 3A).
Together, these data show that the co-exposure of the FASN inhibitors, EGCG and G28UCM, with pertuzumab in parental and resistant HER2+ breast cancer cells is more active than either of the drugs used as a single agent.

In vitro cell growth interactions between FASN inhibitors and temsirolimus in SK, SKTR, SKLR and SKLTR
In our resistant cells we showed changes in EGF family receptors expression and activation without changes in mTOR expression and activation, neither in FASN expression. Since resistant cells express similar levels of mTOR and FASN as same as parental cells, we tested the apoptotic (PARP cleavage) effect of inhibiting mTOR (temsirolimus) and FASN (EGCG) in combination with trastuzumab and/or lapatinib. Temsirolimus or EGCG did not recovered trastuzumab and/or lapatinib sensitivity in resistant cells (S3 Fig). Therefore we conducted experiments to evaluate the inhibitory effect of temsirolimus (mTOR inhibitor) alone and in combination with anti-FASN compounds (EGCG and G28UCM) in the developed resistant HER2+ models (SKTR, SKLR and SKLTR).
Temsirolimus alone displayed a potent anti-proliferative effect in parental and resistant cells. IC 50 concentration ranged from 9 ± 0.9 μM to 11 ± 0.4 μM in resistant models (S1 Table). Temsirolimus (0.05, 0.1, 0.5 and 1 μM) combined with anti-FASN compounds (60 μM of EGCG or 5 μM of G28UCM) for 2 days increased cpi in parental and resistant cells. Mean ratios of cpi induced for each treatment alone versus inhibition induced for co-treatment is shown in Fig 3B EGCG plus temsirolimus ratio was similar in parental and resistant cells These results were confirmed by the isobologram method, using a series of isobologram transformations of multiple dose-response curves at an effect level of 30% (IC 30 ), an statistical analysis that we have used previously [29]. Simultaneous treatment of SK, SKTR, SKLR and SKLTR cells with EGCG and temsirolimus resulted in a strong synergistic interaction index (0.84 < Ix < 0.94). Combination of G28UCM plus temsirolimus had an enhanced synergistic interaction index in parental and resistant cells (0.36 < Ix < 0.58), shown in S1 Table. Effects of EGCG plus temsirolimus compared with mono-treatments and other combinations (EGCG plus pertuzumab and pertuzumab plus trastuzumab and/or lapatinib) on FASN and mTOR expression were also assessed by western blot analysis (see S4 Fig). No significant inhibition in FASN and mTOR protein levels was seen in any treatment except in the case of EGCG plus temsirolimus combination. EGCG plus temsirolimus completely abolished mTOR expression in SKTR, SKLR and SKLTR resistant cells. These data show that co-exposure of temsirolimus with FASN inhibitors, EGCG and G28UCM, display a more potent synergistic effect in HER2+ parental and resistant cells than either of the drugs used as a single treatment.
Antitumor activity of EGCG in combination with pertuzumab and temsirolimus in HER2+ sensitive and resistant patient derived xenografts To better recapitulate the clinical setting we extent our findings in vivo evaluating the antitumor activity of pertuzumab, temsirolimus and EGCG and the combination in a HER2+ PDX model (HER2-PDX) and in a trastuzumab plus lapatinib-resistant HER2+ PDX (HER2-PDXR) model. Both PDX models showed similar HER2, mTOR and FASN expression levels as the in vitro parental and resistant cellular models (S5 Fig). EGCG (30 mg/kg for 3d/w) and pertuzumab (30 mg/kg/once weekly), as single agents, reduced tumor growth in the HER2-PDX model after 24 days of treatment. Control animals achieved a median tumor growth of 461.0 ± 65.6 mm 3 whereas EGCG significantly reduced tumor growth to 247.6 ± 45.0 mm 3 (p-value: 0.017), and pertuzumab reduced to 301.0 ± 62.9 mm 3 . Interestingly, superior (and more rapid) tumor regression was achieved by dual FASN and HER2 blockade (87.2 ± 38.2 mm 3 ; p-value vsEGCG : 0.017 and p-value vsPertu : 0.010), compared with EGCG or pertuzumab as single agents (Fig 4A,  left panel). Despite the absence of complete tumor shrinkage, combinatorial treatment significantly reduced tumor growth in the HER2-PDX model. Under the same schedule, in the HER2-PDXR model EGCG (30 mg/kg for 3d/w) and pertuzumab (30 mg/kg/once weekly) also reduced tumor growth after 24 days of treatment (Fig 4A, right panel), but in vivo efficacy of the dual FASN and HER2 blockade was also superior (and more rapid) compared with EGCG and pertuzumab as a single agents. Compared with the control group (393.9 ± 95.5 mm 3 ), EGCG and pertuzumab decreased tumor growth to 285.9 ± 36.5 mm 3 and 310.4 ± 34.5 mm 3 , respectively. The combination of EGCG with pertuzumab significantly reduced tumor growth up to 177.64 ± 34.5 mm 3 (p-value vsEGCG : 0.030 and p-value vsPertu : 0.008).
Regarding mTOR and FASN inhibition in vivo, EGCG (30 mg/kg for 3d/w) reduced tumor growth in the HER2-PDX model after 21 days of treatment (Fig 5A, left panel). Control animals achieved a median tumor growth of 386.4 ± 66.7 mm 3 whereas EGCG median tumor growth was significantly reduced to 183.3 ± 15.1 mm 3 (p-value: 0.017). Despite the strong antitumor activity exhibited by temsirolimus when used as a single agent (18.0 ± 15.1 mm 3 ; pvalue: 0.000), its activity was little enhanced (day 21) by the addition of EGCG. Combination of temsirolimus with EGCG not only reduced tumor ratio of growth, but also achieved tumor shrinkage compared with the initial tumor volume (-8.2 ± 6.0 mm 3 ). In the trastuzumab plus lapatinib-resistant HER2-PDX model EGCG treatment decreased the median tumor growth (231.8 ± 38.4 mm 3 ) compared with control group (314.8 ± 81.1 mm 3 ) at the end of the experiment (Fig 5A, left panel). In the HER2-PDXR model, temsirolimus significantly decreased tumor growth when used as a single agent (114.3 ± 27.1 mm 3 ; p-value: 0.045), and its activity was enhanced by the addition of EGCG (94.9 ± 33.1 mm 3 ). These results show that temsirolimus could have a relevant effect in patients with HER2 breast cancer, even those that have progressed to anti-HER2 therapies, and combination with FASN-inhibitors could even assist temsirolimus to achieve tumor depletion.
Previous first-generation of FASN inhibitors such as C75 and cerulenin have been limited by inducing severe body weight loss, which is thought to be related to a parallel stimulation of fatty acid oxidation by these inhibitors [24,37]. But we have previously reported that animals treated with EGCG didn't display neither change on body weight nor on hepatic, renal and haematological function serum markers compared to control animals [19]. In this study, animals treated with EGCG and also with pertuzumab and temsirolimus (alone or in combination) were weighed daily to evaluate in vivo body weight effect. With respect to control animals, we identified no significant changes on food and fluid intake or body weight after treatment with EGCG, pertuzumab and temsirolimus alone or in combination (Figs 4C and 5C).
Histological studies (Hematoxylin-Eosin) of liver, lung, kidney and heart showed no tissue structural abnormalities between control and treated animals in both HER2-PDX models (S6, S7 and S8 Figs).

Discussion
Despite the remarkable success of anti-HER2 therapies, patients with advanced HER2-positive breast cancer frequently display primary resistance and, in patients initially sensitive to these agents, acquired resistance may emerge over time [9,10]. To date, even several mechanisms of resistance to anti-HER2 agents are known, this clinical problem is not fully understood. Here, we have developed and characterized stable cell lines derived from the HER2-positive SKBr3 cells that are resistant to either trastuzumab (SKTR), lapatinib (SKLR) or both (SKLTR). Some molecular mechanisms of resistance in our developed anti-HER2 resistant models are consistent with the previously described [8][9][10][11][12][13][14][15][16]. One of the commonly described mechanisms of anti-HER2 therapies is the overexpression of other RTKs or their ligands. Thus, it has been reported that HER3 overexpression leads to HER2/HER3 heterodimer formation consequently activating the PI3K/AKT/mTOR pathway [6]. Conversely, our trastuzumab, lapatinib and trastuzumab plus lapatinib resistant models show a decrease in HER3 expression and activation, whereas an overexpression of EGFR and increased expression levels of EGFR (EGF and TGFα) and EGFR-HER4 (EREG and HB-EGF) ligands. This is consistent with several reports that show EGFR overexpression (and its ligands) in trastuzumab resistant SKBr3 cells and in xenograft models of acquired trastuzumab and lapatinib resistance [38][39][40]. Even more, after dual trastuzumab and lapatinib long-term exposure, our patented SKLTR [41] cells overexpressed HER4 besides EGFR, and increased the expression of EGFR (EGF and AR) and EGFR-HER4 (BTC, EREG and HB-EGF) ligands to overcome the anticancer effects of both anti-HER2 agents. In a clinical study, constitutive presence of HER4 has been directly related with sensitivity to anti-HER2 drugs in breast cancer [42] whereas in prostate cancer an increase of HER4 expression has been shown to be responsible of resistance to the EGFR inhibitor erlotinib [43]. Our findings highlight that HER4 overexpression and activation could be a new molecular mechanism of resistance to anti-HER2 therapies.
Changes in HER2 downstream proteins (such as loss of PTEN, PI3K mutations/hyperactivation, AKT overexpression and hyperactivation) have also been identified as resistant mechanisms to trastuzumab and lapatinib therapies [14][15][16]. In patients treated with trastuzumab, activating mutations of PI3K (PI3KCA) were associated with poor response and survival [44]. Eichhom PJ et al also reported PI3KCA mutations as responsible of lapatinib resistance [15]. In this study, we have shown that although PI3KCA mutations (PIK3CA_E545K mutation in SKTR cells and PIK3CA_E542K mutation in SKLR and SKLTR cells) have not enough incidence to show effects in AKT activation, they collaborate with RTKs changes and downstream loops in order to maintain PI3K/AKT/ mTOR pathway activation in trastuzumab, lapatinib and also in a trastuzumab plus lapatinib resistant cells. Lapatinib effects have been described to be mediated preferentially through the MAPK/ERK pathway through Ras overexpression or mutation [45]. Accordingly, in our study ERK1/2 overactivation is also shown to be another downstream change that leads to cell proliferation signaling of the resistant cells. No significant changes in mTOR and p-mTOR proteins were observed in our long-term resistant cells. Although the main pathway related to mTOR is the PI3K/AKT axis, mTOR is a downstream in which several signaling pathways converge. In addition, this pathways act as a complex network, and activation of one important effector can be accomplished by several emissaries. Even with decreased activation of AKT in our resistant models, maintenance of mTOR activation can be accomplished by direct signaling of PI3K to mTOR, bypassing AKT [46]. Also, it is described that MAPK inhibit the tuberous sclerosis complex (TSC1/TSC2), which in turn inhibits mTOR activation [47].
Together, overactivation of HER2 in SKLR and SKLTR cells and maintenance of mTOR, p-mTOR and FASN expression in all the resistant models provided the rationale to test combined FASN and HER2 or mTOR pathways blockade in this setting. We found that the simultaneous treatment of parental and resistant (SKTR, SKLR and SKLTR) cells with anti-FASN compounds (EGCG and the novel derivative G28UCM) plus pertuzumab improved the effects of each drug alone in SKLTR. But, inhibiting mTOR, the downstream target of HER2 pathway in combination with FASN inhibition resulted in a strong synergistic interaction in all parental and resistant cells. G28UCM, the novel FASN inhibitor, also improved combinatorial effect of EGCG, producing much more synergism between temsirolimus and G28UCM. We had previously shown that G28UCM improved EGCG effects, alone and in several combinatorial regiments with anti-HER2 drugs and chemotherapy, in parental and trastuzumab-or lapatinibresistant AU565 HER2 breast cancer cells [29]. Several studies have used mTOR inhibition to overcome resistance to HER2-targeted therapies [48] and it has already been assayed in women with trastuzumab-resistance [49]. It has also been shown the synergism between mTOR and FASN inhibition to induce cytotoxicity in ER/HER2-positive breast cancer cell lines [50]. These in vitro results support the rationale to test in vivo the antitumor efficacy of these agents in a combination regimen in tumors resistant to anti-HER2 therapies.
In previous preclinical studies conducted in nude mice bearing HER2 cells, we and others showed that EGCG displayed in vivo antitumor activity without decreasing food intake and induction of weight loss [24,37]. This is the first study that attempt to evaluate the in vivo *** (p 0.001). (b) Apoptosis, by TUNEL fluorescent assay, was perfomed in control (C), EGCG (E), temsirolimus (T) and combination (T+E) treated as in (A) tumors. Tumors were collected at the end of the experiment and fixed in formalin. Pictures are representative of two samples of each group. (c) Body weight of the mice treated as in (A). Data are expressed as body weight at the end of the experiment and boxes show the 25th to 75th percentiles, whereas whiskers extend to the 5th and 95th percentiles. doi:10.1371/journal.pone.0131241.g005 Dual FASN and HER2-Signaling Blockade in HER2+ Resistant Breast Cancer efficacy and feasibility of dual blockade of FASN and the HER2 signaling pathway in HER2-positive patient tumor samples (HER2-PDX) and in HER2 samples of a patient who relapsed after trastuzumab and lapatinib therapies (HER2-PDXR). Here, we report the validity of this approach clearly showing that the combination of EGCG with pertuzumab and temsirolimus resulted in synergistic reduction of HER2-PDX and HER2-PDXR tumors, without signs of toxicity (weight loss) related to in vivo antitumor efficacy experiments using anti-FASN compounds [24,28,29]. Reduction of tumor growth could be accomplished, in part, by an apoptotic event. Increase of apoptosis has been seen in parental and resistant tumor samples treated with EGCG, pertuzumab and temsirolimus. As synergism in tumor reduction after combinatorial treatments, apoptosis has also been synergistically increased when combining EGCG with pertuzumab or temsirolimus. These similar profiles in tumor growth reduction and apoptosis suggest that apoptosis is responsible for tumor growth inhibition. In fact, we had previously shown that EGCG produces apoptosis in vitro and in vivo [18,19,28,37]. Apoptosis is also consistent with other studies of pertuzumab in cells and mouse models [51,52]. Temsirolimus has been shown to produce apoptosis in a resistant oropharyngeal carcinoma cell line [53], colorectal cancer cells [54] and other cancers. But little, if any, apoptosis has been seen in different breast cancer cell lines treated with temsirolimus [55]. In this study, we demonstrate that tumor growth inhibition in HER2 breast cancer PDX (non-resistant and resistant) occurs by apoptotic event in tumoral cells, and this is consistent what have been found in other types of cancer. These findings, accordingly with those obtained in vitro, encourages us to think that combining FASN inhibitors with temsirolimus or pertuzumab could be one example of a potential combinatorial available in the clinical management of HER2-positive breast cancer patients who progressed to standard treatments.
In this study, we have developed novel mono-and dual-long term trastuzumab plus lapatinib resistant breast cancer models to find out new pharmacological strategies to overcome this setting. Then, we have showed in vitro and in vivo that the inhibition of FASN, alone or in combination with anti-HER2 signaling drugs (temsirolimus and pertuzumab), could have relevant clinical implications in patients who fail to respond to current therapies.
In summary, our findings provide a rationale for the preclinical development of inhibitors of FASN activity in combination with anti-HER2 signaling agents in breast cancer refractory to anti-HER2 therapies.  Table. Synergy analysis between FASN inhibitors and temsirolimus in parental and resistant cells. Drug cytotoxicity was calculated as the concentration of drug needed to produce 50% of cell death (IC 50 ) when parental SKBr3 (SK) or trastuzumab-, lapatinib-and trastuzumab plus lapatinib-resistant cells (SKTR, SKLR and SKLTR). Values represent the mean ± SE from at least three independent experiments performed in triplicate. The interaction index (Ix) for temsirolimus plus FASN inhibitors effect was calculated using isobologram analysis. The Ix parameter indicate whether the doses of the two drugs required to produce a given degree of cytotoxicity are greater than (Ix > 1 or antagonism) equal to (Ix = 1 or additivism) or less than (Ix < 1 or synergism) the doses that would be required if the effect of two agents were strictly synergic. Ix mean values ± SE for the two drug treatment were obtained from triplicate studies with different combination treatments and performed at least twice independently. Ã (p < 0.05), ÃÃ (p < 0.01) and ÃÃÃ (p < 0.001) indicate the level of statistical significance of the Ix compared with an Ix of 1.