High Prevalence of Fatty Acid Synthase Expression in Colorectal Cancers in Middle Eastern Patients and Its Potential Role as a Therapeutic Target

OBJECTIVES:Many human epithelial cancers, particularly those with a poor prognosis, express high levels of fatty acid synthase (FASN), a key metabolic enzyme linked to synthesis of membrane phospholipids in cancer cells. Overexpression of FASN is linked with activation of the phosphatidylinositol-3′-kinase (PI3 K)/AKT pathway. However, the role of FASN in colorectal cancer (CRC) has not been fully elucidated. We investigated the expression of FASN and determined its functional association with the PI3/AKT pathway in CRC.METHODS:Expression of FASN and its associated targets were studied by immunohistochemistry on 448 CRC tumors in a tissue microarray (TMA) format. Analysis of apoptosis and cell cycle was evaluated in vitro using CRC cell lines by flow cytometry and DNA fragmentation assays. Protein expression was determined by immunohistochemistry and western blotting. In vivo xenograft studies were performed using CRC cell lines and NUDE mice.RESULTS:Correlation of FASN with various clinicopathological parameters on 448 CRC samples was assessed. Activated AKT was found in 294/409 (71.9%) of CRC and was associated with FASN overexpression. FASN expression was observed in 27.1% (109/403) of Middle Eastern CRC. Additionally, FASN expression was significantly more common in tumors characterized by microsatellite instability (MSI) than in those characterized by microsatellite stability (MSS) (P<0.01). Our in vitro data using HCT-15, an MSI CRC cell line, showed a better apoptotic response after inhibition of FASN activity as compared with Colo-320, an MSS CRC cell line. Finally, treatment of HCT-15 cell line xenografts with C-75 resulted in growth inhibition of tumors in NUDE mice via downregulation of FASN and AKT activity.CONCLUSIONS:These data identify FASN as a potential biomarker and a novel therapeutic target in distinct molecular subtypes of CRC.


INTRODUCTION
Colorectal cancer (CRC) still has a high morbidity and mortality. It is now established that the subtype of CRC characterized by microsatellite instability (MSI) follows a pathway di erent from that characterized by microsatellite stability (MSS) (1,2) . e clinicopathological characteristics are becoming clear, but speci c genetic and therapeutic di erences remain to be fully de ned. Signi cant improvements have been made in the management of this disease mainly through the introduction of adjuvant chemotherapy agents like urouracil and oxiplatin (3) . More recently, advances in understanding of tumor biology have led to the development of targeted therapies (4) , allowing progress in the treatment of colorectal cancer (5) .
One of the promising targets for therapeutic intervention is fatty acid synthase (FASN), which is a multifunctional enzyme that catalyses the terminal steps in the synthesis of long chain saturated fatty acid. In normal cells, energy balance is physiologically important in FASN regulation and high carbohydrate / low fat diets upregulate FASN (6) .
A wide variety of tumors and their precursor lesions undergo exacerbated de novo biosynthesis of fatty acids irrespective of the level of circulating lipids. Neoplastic lipogenesis is re ected by signi cantly increased activity and coordinate expression of several lipogenic enzymes such as FASN that is upregulated in most solid tumors (7 -11) . FASN appears to provide a selective proliferative advantage as its overexpression showed to correlate with poor prognosis in breast and prostate cancers and is found elevated in the blood of cancer patients (12 -14) . Furthermore, inhibition of FASN activity is selectively cytotoxic to cancer cells in vitro and in vivo (15,16) .
Upregulation of FASN expression in cancer cells has been linked to phosphatidylinositol-3-kinase (PI3K) / AKT signaling pathway (17 -20) . Activation of PI3-kinase pathway recruits a number of signaling proteins including protein kinase B (also known as AKT). On recruitment, AKT becomes phosphorylated / activated and exerts its anti-apoptotic activity through phosphorylation of downstream targets such as Bad, FOXO, and GSK3 (21,22) .
Interestingly, PI3K / AKT has been shown to modulate the activity of SKP2, an ubiquitin ligase that degrade the p27kip1 by proteasomes (23) . In addition, PI3K pathway has been shown to be capable of negatively regulating FASN-induced cell death (24,25) . erefore, we sought to assess activation of PI3K / AKT pathway in CRC, whether this activation correlates with increased FASN expression and relation to other clinicopathological parameters in a large cohort of Saudi CRC using tissue microarray (TMA) technology. We next examined the e ect of C-75, a synthetic slow binding inhibitor of FASN activity, on CRC cell lines both in vivo and in vitro . All together, our ndings strongly suggest that a tight functional association between FASN and AKT is taking place in a subset of CRC and that FASN expression can be a useful biomarker in this subset of CRC.

Patient selection and TMA construction
In total, 448 patients with CRC diagnosed between 1990 and 2006 with available formalin xed para n embedded tumor tissues were selected from King Faisal Specialist Hospital and Research Centre. All patients diagnosed with colorectal cancer during this period were randomly selected and only those patients were excluded where formalin xed para n embedded tissues were not available. All samples were analyzed in TMA format. TMA construction was performed as described earlier (26) . ree 0.6 mm cores of CRC were arrayed from each case. All clinical data and long-term follow-up was provided by colorectal unit, Department of Surgery, King Faisal Specialist Hospital and Research Centre. e institutional review board of King Faisal Specialist Hospital and Research Center approved the study.

Immunohistochemistry
TMA slides were processed and stained manually. e streptavidin-biotin peroxidase technique with diaminobenzidine as chromogen was applied. For antigen retrieval, Dako Target Retrieval Solution pH 9.0 was used. e list of primary antibodies used are listed in Supplementary Table 1 . Immunohistochemistry scoring was done in a blinded manner and TMA spots were scored under a 20 × 0.70 objective on an Olympus BX 51 microscope (Olympus America Inc., Center Valley, PA).
FASN expression was categorized as negative (no or weak expression), moderate (1 + ), or strong (2 + ) as has been reported earlier (11) . Adipose tissue served as internal positive control for FASN expression. p-AKT scoring was done as described earlier (27,28) . Cases were considered positive if 50 % or more of tumor cells were stained positive for p27kip1 and SKP2. is cuto was chosen based on prior analysis on some of the markers and similar cuto used by others (29) . Statistics e so ware used for statistical analysis was statview 5.0 (SAS Institute Inc., NC). Survival curves were constructed by Kaplan -Meier method and multivariate analysis by Cox regression; P values < 0.05 were considered signi cant. Two-sided tests were used throughout all the analyses.

Apoptotic assay
Cell lines were treated with and without C-75 for 24 and 48 h, and apoptosis was measured by annexin V / PI staining and DNA laddering using a 1.5 % agarose gel as described earlier (30 -32) .

Cell lysis and immunoblotting
Cells were treated with C-75 as described in the legends and lysed as described earlier (33) . Proteins were immunoblotted with di erent antibodies and visualized by the enhanced chemiluminescence (Amersham, Piscataway, NJ) method.

Detection of Bax conformational changes
is assay was performed as described earlier (31) . Brie y, a er treatment with indicated reagents for indicated time points, cells were harvested, washed with PBS, and lysed with Chaps lysis bu er (10 m M HEPES (pH 7.4), 150 m M NaCl, 1 % Chaps). Immunoblotting was performed using Bax polyclonal antibody.

Measurement of mitochondrial potential and cytochrome c release
A er treatment of CRC cell lines with C-75 for 48 h, mitochondrial membrane potential was measured using JC1 dye and release of cytochrome c was analyzed using immunoblotting of mitochondrial and cytosolic protein fractions as described earlier (32,34) .
Gene silencing using siRNA FASN siRNA, AKT1 siRNA, AKT2 siRNA, and scrambled control siRNA were purchased from Qiagen (Valencia, CA). Cells were transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) for 6 h a er which the lipid and siRNA complex was removed and fresh growth medium was added. Cells were lysed 48 h a er transfection, and speci c protein levels were determined by western blot analysis with speci c antibodies.

Animals and xenograft study
Six-week-old NUDE mice were obtained from Jackson Laboratories (Maine) and maintained in a pathogen-free animal facility at least 1 week before use. All animal studies were done in accordance with institutional guidelines. For xenogra study, mice were inoculated subcutaneously into the right abdominal quadrant with 5 × 10 6 HCT-15 or Colo-320 cells in 200 μ l PBS. A er 1 week, mice were randomly assigned into three groups: two groups receiving two doses of C-75 (10 and 20 mg / kg) and remaining one group receiving 0.9 % saline. e body weight and tumor volume of each mouse was monitored weekly. e tumor volume was measured as described earlier (35) . A er 4 weeks treatment, mice were killed and individual tumors were weighed, then snap-frozen in liquid nitrogen for storage.

FASN expression and its correlation with p-AKT and other clinicopathological parameters
Elevated levels of FASN expression have been reported in a variety of solid tumors (7,20,36) . We, therefore, rst sought to study the expression of FASN and its association with di erent clinicopathological parameters in 448 cases of CRC in a TMA format. High levels of FASN expression were seen in 27.1 % of the colorectal carcinomas ( Figure 1 ). e incidence of expression for other immunohistochemistry markers is presented in Table 1 .
As shown in Table 1 , FASN overexpression was not associated with age, gender, tumor site, American Joint Cancer Committee stage, and tumor di erentiation. However, FASN overexpression was signi cantly associated with overexpression of p-AKT ( P < 0.01) and Ki-67 ( P < 0.01). FASN overexpression was also signi cantly associated with overexpression of SKP2 ( P = 0.02). FASN overexpression was signi cantly more common in the MSI group as compared with MSS group ( P < 0.01). FASN expression was signi cantly more common in the colorectal carcinoma subset with PI3 K mutation as compared with colorectal carcinomas without PI3K mutations ( P = 0.03). ere was no signi cant association between expression of FASN and overall survival ( P = 0.21). Finally, FASN overexpression was found to be higher in Middle Eastern CRC as compared with the Western data (11) .

C-75 causes apoptosis in a dose-dependent manner in CRC cell lines
Our clinical data showed high level of FASN expression in CRC, and it has been shown earlier that FASN overexpression provides a selective advantage to tumor cells by promoting proliferation and inhibiting apoptosis (7,8,14) . Interestingly, FASN inhibition has been shown to be more e ective in inhibiting cell viability and initiating apoptosis (7,8,14) . erefore, we investigated whether C-75 treatment of CRC cell lines induced cell-cycle arrest or apoptosis in MSI and MSS cell lines. We selected two CRC cell lines, HCT-15, an MSI and Colo-320, an MSS. Both cell lines were treated with C-75 for 24 and 48 h and cell-cycle fractions were determined. As shown in Figure 2a , a er 24 h treatment, there was substantial increase in the G2 / M population in both cell lines. G2 / M population increased from 16.16 ± 3.2 % in untreated sample to 18.93 ± 4.1 % at 25 μ M treatment and 39.6 ± 3.9 a er 50 μ M treatment of C75 in Colo-320 cell line and from 11.01 ± 3.1 % to 13.33 ± 4.2 % and 36.17 ± 3.6 % in HCT-15 cell line. Interestingly, a er 48 h treatment, the sub-G1 population of cells increased in both the cell lines, from 3.23 ± 0.9 % in untreated sample to 16.16 ± 3.1 % at 25 μ M and 27.92 ± 5.0 % at 50 μ M in Colo-320 cell line and from 6.10 ± 1.9 % to 15.95 ± 3.2 % and 59.95 ± 6.1 % in HCT-15 cell line ( Figure  2b ). is increase in sub-G1 population was accompanied by loss in G0 / G1, S, and G2 / M phase in treated cells. We also used annexin V / PI dual staining and DNA laddering for con rmation of C-75-induced apoptosis in CRC cells. Cells were treated with 25 and 50 μ M C-75 for 48 h and apoptosis was measured by annexin V / PI dual staining. As shown in Figure 2c , treatment of MSS Colo-320 cells with 25 and 50 μ M induced 42 ± 6.2 % and 53 ± 5.8 % apoptosis, respectively. Interestingly, the response to C-75 treatment in MSI HCT-15 cell line was more pronounced at 50 μ M dosage 34 ± 4.5 % and 87 ± 10.3 % . Similar results were obtained in two additional MSS cell lines, Caco-2 and SW-480, and two MSI cell lines, DLD-1 and HCT-116. As shown in Figure 2c (lower panel), the response of C-75 treatment was more pronounced in MSI cell lines as compared with MSS cell lines. We analyzed DNA fragmentation, which is another hallmark of apoptosis. As shown in Figure 2d , C-75 caused a dose-dependent DNA fragmentation in Colo-320 and HCT-15 cell lines. Finally, we averaged the percentage apoptosis of all three MSS cell lines and compared them with the PATHOLOGY Fatty Acid Synthase-Mediated Signaling in Colorectal Carcinoma average of three MSI cell lines a er treatment with C-75. As shown in Figure 2e , there was a statistically signi cant di erence in the response of MSI cell lines as compared with MSS cell lines ( P < 0.01). ese data suggest that even though all the cell lines respond to C-75 inhibition, the apoptotic response is signi cantly higher in MSI cell lines.

FASN-mediated regulation of PI3 / AKT signaling pathways in CRC cell lines
Our clinical data also suggests signi cant association between FASN and AKT; we, therefore, sought to determine the e ect of FASN inhibition on AKT signaling pathway. Using the CRC cell lines, we sought to determine the expression of FASN and p-AKT and their response to C-75 treatment. As shown in    To better understand the association between FASN and AKT, we performed transfection studies with siRNA against AKT. AKT-speci c siRNA transfection in HCT-15 cell line did not show any e ect on FASN expression but dephosphorylated AKT ( Figure 3c ). In addition, AKT inhibitor decreased AKT phosphorylation without a ecting the FASN protein level in both the cell lines ( Figure 3c ). AKT inhibitor treatment also caused about 50 % and 56 % apoptosis in Colo-320 and HCT-15 cell line ( Figure 3d )  mitochondrial membrane potential as measured by JC1 stained green uorescence depicting apoptotic cells. We then studied the release of cytochrome c from mitochondria into cytosole. As shown in Figure 4d , higher level of cytochrome c was measured in cytosolic and lower levels in mitochondrial fraction in both cell lines a er C-75 treatment. We then sought to determine whether C-75-induced release of cytochrome c is capable of activation of caspase-3 and PARP. Figure 5a shows that C-75 treatment resulted in the activation of caspase-3 and cleavage of PARP in Colo-320 and HCT-15 cells. In addition, pretreatment of CRC cells with 80 μ M z-VAD-fmk, a universal inhibitor of caspases, followed by C-75 treatment, abrogated apoptosis from 50 % to 12 % in Colo-320 cells and from 59 % to 9 % in HCT-15 cells as well as prevented caspase-3 and PARP activation induced by C-75 ( Figure 5b and c ), clearly indicating that caspases play a critical role in C-75-induced apoptosis in CRC cells.
upstream of AKT and AKT inhibition does not lead to change in FASN levels.

C-75 treatment of CRC cells induced apoptosis via mitochondrial pathway-caspase mediated in CRC cell lines
Next, we determined whether downregulation of FASN and p-AKT signaling involves mitochondria in CRC cell lines. C-75 treatment resulted in activation of caspase-8 leading to truncation of BID in both cell lines tested ( Figure 4a ), as inferred by decreased intensity of full-length BID band. As it is known that truncated BID plays a role in the activation of Bax, we examined Bax in response to C-75 treatment. As shown in Figure 4b , conformationally changed Bax was detected in both cell lines a er C-75 treatment for 24 h. We then tested the e ect of C-75 on mitochondrial membrane potentials in these cells. As shown in Figure 4c , treatment of both cell lines with C-75 resulted in loss of

In vivo activity of FASN inhibitor C-75 against CRC cancer cells xenograft
Our observation that CRC cells exhibit enhanced sensitivity to FASN inhibitor-induced apoptosis in vitro suggests the potential for therapeutic responses to treatment of CRC with FASN inhibitor in in vitro . erefore, ability of C-75 to inhibit CRC tumor growth was examined with a mouse xenogra model of CRC cancer. NUDE mice were inoculated subcutaneously in the right abdominal quadrant with 5 million HCT-15 cells. Mice were then treated with either two doses of C-75 treatment group (10 and 20 mg / kg / dose), or vehicle DMSOtreated control groups ( n = 6). A er 4 weeks of treatment, mice were killed and tumors were collected. As shown in Figure 6a , C-75 treatment causes a time-dependent regression of HCT-15 xenogra tumors in mice as compared with vehicle-treated mice. As shown in Figure 6a , at rst week, tumor volume for vehicle-treated mice was 68.9 mm 3 , 39.4 mm 3 for mice treated with 10 mg / kg C75, and 54.7 mm 3 for 20 mg / kg treatment. Second week measurements were 240, 98.6, and 122.6 mm 3 , third week measurements were 729.1, 305, and 122.6 mm 3 , and a er 4 weeks of treatment, the tumor volume measurements were 1,210.7, 378.3, and 203.5 mm 3 in vehicle-treated mice, 10 mg / kg treatment, and 20 mg / kg treatment, respectively. e regression reached signi cance ( P < 0.05) at the end of third week of treatment by C-75. A signi cant reduction in tumor weight ( Figure 6b ) was observed in mice treated with C-75

DISCUSSION
In light of our earlier ndings that alteration of PI3K / AKT pathway is frequently seen in large number of primary CRC specimens and considering the role of PI3K / AKT signaling pathway in regulating FASN expression (2,17 -20) , we sought to explore the potential link between FASN overexpression and PI3K / AKT pathway alteration in a large series of CRC sample and to study the e ect of FASN inhibition on CRC both in vitro and in vivo . Our study showed a signi cant association between FASN and p-AKT expression, MSI status, Pik3ca oncogenic mutation, Ki67 and SKP2 in clinical samples of CRC. e close correlation between FASN expression and AKT activation in CRC suggests that FASN plays an important role in the pathogenesis of this molecular subtype of CRC. We have reported earlier that activating mutation of PIK3CA is also seen mainly in MSI subtype of CRC samples (2) . So the apparent association between FASN overexpression and ( P < 0.05) as the weight of the tumor decreased from 2.51 g in vehicle-treated mice to 1.01 g in mice treated with 10 mg / kg C75 and 0.84 g in mice treated with 20 mg / kg C-75. Additionally, images of tumor before and a er necropsy showed that C-75 treatment resulted in shrinkage of tumor size ( Figure  6c ). As shown in Figure 6d , the level of FASN and p-AKT proteins were markedly decreased in primary tumors of mice treated with C-75 as compared with vehicle-treated mice as detected by western blotting. Immunohistochemistry done on HCT-15 xenogra tumor sections showed a reduced level of FASN as well as p-AKT staining a er C-75 treatment ( Figure  6e ) that was statistically signi cant between untreated group vs. 20 mg / kg / dose C-75-treated group ( P = 0.04 and P < 0.01). Finally, we also examined the e ect of C-75 treatment in MSS Colo-320 cell line xenogra ( Supplementary Figure 4 ). Our data suggests that MSS xenogra s respond in a similar manner as MSI xenogra s but with a slower rate.    (11) . Although FASN expression has been shown to be a prognostic marker in certain cancer ( (38 -41) , FASN expression was not found to have a prognostic value in our study.
To further elucidate the pivotal role of FASN expression in determining the molecular characteristic of CRC, we conducted in vitro analysis on two cell lines that were screened earlier for mismatch repair de ciency and activating PIK3CA mutation, HCT-15 MSI, PIK3CA mutation + ve, and Colo-320 MSS, PIK3CA mutation − ve. Our in vitro studies further support the correlation between FASN expression and AKT activation in both MSI and MSS CRC cell lines. Interestingly, inhibition of FASN also substantially reduced levels of SKP2, an F-box protein essential for proteasome degradation of p27kip1. Consequently, we conclude that inhibition of FASN acts upstream of proteasome to control p27kip1 levels and ultimately blocks cell-cycle progression. e mechanistic connection between the FASN blockade and reduction in SKP2 might be mediated through AKT activation. Recently, it has been shown that AKT modulates the expression of FASN in a positive feedback manner in ovarian cancer cells (20) . Our pharmacological inhibition and gene silencing studies suggests that inhibition of AKT does not a ect the expression of FASN. On the other hand, C-75 treatment of CRC cell lines as well as gene silencing of FASN inactivated AKT activity in addition to downregulation of FASN expression in both the CRC cell lines. ese ndings suggest that FASN is an upstream e ector of AKT in CRC cell lines.
Our in vivo studies of the a ect of C-75 on HCT-15 and Colo-320 tumor growth in a murine xenogra model are consistent with results obtained with breast cancer, ovarian cancer, and mesothelioma xenogra models (14,20,42) . In addition to an overall signi cant inhibition of xenogra tumor growth, western blot analysis of tumors showed decreased AKT activity in most C-75-treated mice, concordant with reduced FASN protein level. us, FASN inhibition may have signi cant promise as a therapeutic approach in subset of CRC.
In conclusion, our ndings suggest that there is an overall increase in expression of FASN in CRC tumors of this region (11) . erefore, targeting FASN for treatment of CRC maybe more bene cial in CRC tumors. Furthermore, data presented here demonstrate a signi cant correlation between expression of FASN and active AKT in CRC and indicate that inhibition of PI3K / AKT signaling synergize the FASN inhibitors to induce apoptosis in CRC cell lines with constitutively active AKT. is may have signicant clinical implications. Elucidating the molecular link between FASN overexpression and AKT activation may be important as a biomarker and may serve as promising target for therapeutic intervention in this molecular distinct subtype of CRC.

CONFLICT OF INTEREST
Guarantor of the article: Khawla S. Al-Kuraya, MD, FCAP. Speci c author contributions: Shahab Uddin, PhD: designed the study, performed experiments, analyzed data, wrote the paper, and gave nal approval for submission of the manuscript in its current format. Azhar Hussain, MBBS: designed the study, performed experiments, analyzed data, helped in writing and proofreading the manuscript, and gave nal approval for submission of the manuscript in its current format. Maqbool Ahmed, PhD: performed experiments, analyzed the data, helped in dra ing the manuscript, and gave nal approval for submission of the manuscript in its current format. Jehad Abubaker, PhD: performed experiments, analyzed the data, and gave nal approval for submission of the manuscript in its current format. Nasser Al-Sanea, MD: contributed substantially to the acquisition and interpretation of clinical data, participated in revising and formulating the content of the manuscript and gave nal approval for submission of the manuscript in its current format. Alaa Abdul Jabbar, MD: contributed substantially to the acquisition and interpretation of clinical data, participated in revising the content of the manuscript, and gave nal approval for submission of the manuscript in its current format. Luai H Ashari, MD: contributed substantially to the acquisition and interpretation of clinical data, and gave nal approval for submission of the manuscript in its current format. Samar Alhomoud, MD: contributed substantially to the acquisition and interpretation of clinical data, participated in revising and formulating the content of the manuscript and gave nal approval for submission of the manuscript in its current format. Fouad Al-Dayel, MD: contributed substantially to the acquisition and interpretation of clinical data and gave nal approval for submission of the manuscript in its current format. Prashant Bavi, MD: designed the study, performed experiments, analyzed data, helped in writing the manuscript, performed statistical analysis, and gave nal approval to submit the manuscript in its present form. Khawla S. Al-Kuraya, MD, FCAP: designed, formulated, and lead the project, analyzed and interpreted data, wrote the manuscript, and gave nal approval. Financial support: None. ( is work was not supported by the National Institutes of Health, the Wellcome Trust, the Howard Hughes Medical Institute, or others.) Potential competing interests: None.