CUL4A contributes to the biology of basal-like breast tumors through modulation of cell growth and antitumor immune response

The CUL4A E3 ubiquitin ligase is involved in the regulation of many cellular processes and its amplification and/or overexpression has been observed in breast cancer. The 13q34 amplification, which is associated with the basal-like breast cancer subtype, has been proposed as one of the mechanism behind CUL4A up-regulation. However, the specific contribution of CUL4A to the biology of basal-like breast tumors has not yet been elucidated. In this work, by using cellular models of basal phenotype, we show the inhibitory effect of CUL4A silencing in the proliferation and growth of breast cancer cells both, in vitro and in vivo. We also demonstrate the transforming capacity of CUL4A exogenous overexpression in the 184B5 human mammary epithelial cells in vitro. Our results suggest a synergistic effect between CUL4A high levels and the activation of the RAS pathway in the tumorigenesis of basal-like breast cancer tumors. In addition, by using a proteomics approach we have defined novel candidate proteins and pathways that might mediate the oncogenic effect of CUL4A. In particular, we report a putative role of CUL4A in bypassing the immune system in breast cancer through the down-regulation of several molecules involved in the immune surveillance. These findings provide insight into the oncogenic properties of CUL4A in basal-like breast cancer and highlight the therapeutic opportunities to target CUL4A.


INTRODUCTION
The CUL4A E3 ubiquitin ligase belongs to the family of cullin proteins and is a component of the cullin-RING ligase 4A complex [1]. This complex plays a crucial role in the regulation of both stability and degradation of a huge spectrum of cellular proteins through the ubiquitin-proteasome system [1,2]. CUL4A confers substrate specificity and it is involved in the control of cell cycle, DNA replication and DNA repair depending on the nature of ubiquitinated proteins [3][4][5].
CUL4A has been found amplified and/or overexpressed in different types of carcinomas and a putative oncogenic role has been proposed for this gene [6][7][8][9]. Breast cancer patients with strong expression of CUL4 show shorter overall and disease-free survival [6,10] which has led to the proposal that CUL4A might confer an aggressive behavior in this malignancy. The functional in vitro and in vivo characterization of the putative oncogenic role of CUL4A has been previously studied in prostate cancer and mesothelioma [7,8]. In breast cancer, a recent functional analysis of CUL4A has revealed a pivotal role of the gene www.impactjournals.com/oncotarget in regulating the metastatic behavior of breast cancer cells [11].
In previous studies we showed that the overexpression of CUL4A was associated with markers of proliferation and tumor aggressiveness in familiar and sporadic primary breast tumors [12]. In addition we found that, although not exclusively, the 13q34 amplification would be one of the mechanisms triggering CUL4A overexpression in breast cancer. Importantly, we observed that the 13q34 amplification presents increased frequency in BRCA1-defective tumors and seems to be associated with primary breast tumors of basal-like phenotype [12,13]. Since basal-like breast cancers are characterized by lack of effective targeted therapy and poor clinical outcome [14], the elucidation of the specific role of CUL4A in the carcinogenic process of this tumor subtype is of major interest.
In this study we have evaluated in vitro and in vivo the functional significance of CUL4A down-regulation and up-regulation in basal-like breast cancer models that exhibit features resembling those found in primary breast tumors. Our results support the role of CUL4A as a driver oncogene at the 13q34 amplification in basal-like breast cancer. Also, suggest an implication of CUL4A overexpression in cancer aggressiveness and progression rather than a major function as a transformation-initiating event. In addition, our proteomics-derived data provide evidence for novel molecules and pathways that might be involved in mediating the tumorigenic process orchestrated by CUL4A.

CUL4A is overexpressed in breast cancer cell lines
We examined the expression of CUL4A in 11 human breast cancer cell lines and two non-transformed human mammary epithelial cell lines by quantitative PCR (qPCR) and Western blot (WB). In both analysis the nontransformed cells (HBL100 and 184B5) were within the lower relative expression levels indicating that CUL4A is overall up-regulated in breast cancer cell lines compared with normal mammary epithelial cells ( Figure 1A and B). CUL4A expression varied across the breast cancer cell lines. MDAMB157 with extra DNA copies at 13q34 was found among those cell lines exhibiting the highest CUL4A relative expression values. Also, HCC1937 and MDAMB436 cells, with BRCA1-impairment, were found in the upper limit of the expression range. With few exceptions cell lines ordered according to their CUL4A relative expression by qPCR matched the order obtained by WB ( Figure 1A and B), with both data set exhibiting a high correlation (ρ=0.714, P=0.004). Since the 13q34 amplification and/or CUL4A overexpression is more frequent in BRCA1-associated and basal-like primary breast tumors [12] we selected the BRCA1-null HCC1937 and the 13q34 amplified MDAMB157 cell lines to further characterize CUL4A function in breast cancer progression. Both cell lines are of basal-like subtype [15,16] and as shown by our experiments express high levels of CUL4A. In addition, we included the basal-type and low CUL4Aexpressing 184B5 mammary epithelial cells as a nontumorigenic model to assess the implication of CUL4A overexpression in the initial steps of the transformation process.

CUL4A down-regulation inhibits breast cancer tumorigenesis in vitro and in vivo
To directly assess the contribution of endogenous CUL4A overexpression on the transformed phenotype of human breast cancer cells, we examined the effects of knocking down CUL4A in the HCC1937 and MDAMB157 breast cancer cell lines where CUL4A is amplified and/or overexpressed. Cells transduced with a CUL4A-specific shRNA construct (SH4) (see Methods) presented reduced CUL4A protein levels by at least 90% compared with control cells (transduced with scrambled vector) (Figure 2A). The CUL4A-knocked-down cells showed decreased proliferation compared with control  Figure 2B). In agreement, a reduced S-phase population was observed in both CUL4A knocked-down cells compared with controls ( Figure  2C). In addition, HCC1937-and MDAMB157-CUL4Asilenced cells showed reduced ability to form colonies in anchorage-dependent conditions relative to controls ( Figure 2D). Proliferation was also inhibited in CUL4Aknocked-down HCC1937 cells when grown in anchorageindependent conditions ( Figure 2E). Modifications in anchorage-independent growth could not be evaluated in the MDAMB157 model as these cells do not grow in softagar. The 184B5-CUL4A-silenced cells ( Figure 3A) did not show significant changes in proliferation ( Figure 3B) and colony formation abilities ( Figure 3C). This would be consistent with a specific dependence of the CUL4A function acquired in CUL4A-overexpressing cancer cells. In agreement with our in vitro results, orthotopical inoculation of CUL4A-silenced cells in immunodeficient mice generated tumors with significantly slower growth rate than that of Scramble-derived tumors ( Figure 4A). At the end point of these in vivo experiments the average volume of tumors induced by CUL4A-silenced cells was half the average volume exhibited by tumors produced by control cells ( Figure 4B).

CUL4A overexpression promotes in vitro transformation of human mammary epithelial cells
To elucidate the role played by CUL4A in the initial steps of the carcinogenic process the 184B5 immortalized non-transformed human mammary breast cells were transduced with a CUL4A expression vector. Cells with exogenous CUL4A expression showed effective increase in CUL4A protein levels compared with cells infected with empty vector ( Figure 5A) and showed increased viability with respect to control cells in cell-based luminescence assays ( Figure 5B). Although CUL4A overexpression also induced an S-phase population increase relative to control ( Figure 5C) differences did not reach statistically Cells were fixed and stained with crystal violet. Then crystal violet was solubilized and readouts (optical density at 590 nm) were obtained. Representative example of three-independent experiments is shown. Data are expressed as mean ± SD of three replicates. (F) CUL4Aoverexpressing and control cells were grown in soft agar to evaluate anchorage-independent growth ability Graphics represents the average number of colonies ± SD. **P<0.01, *P<0.05. www.impactjournals.com/oncotarget significance. However, when we specifically compared the subpopulation of polyploid cells we found significantly higher BrdU incorporation in the CUL4A-overexpressing cells compared to control cells (3-fold increase, P<0.05). These results suggest that CUL4A overexpression might enhance proliferation through promotion of S-phase entry. In particular, significant increase of poliploid cells in S-phase suggests that CUL4A expression might contribute to tetraploid checkpoint failure mostly leading to propagation of poliploid cells. Consistently, cell cycle analysis revealed that 184B5 cells transduced with the CUL4A vector exhibited a significant increase of polyploid cells compared to those transduced with the empty vector ( Figure 5D). We next performed colony formation assays and observed a significant enhancement of the capability of the CUL4A up-regulated cells to form colonies both, in anchorage-dependent ( Figure 5E) and anchorageindependent conditions ( Figure 5F). Altogether these results support the CUL4A contribution to malignant transformation of normal breast epithelial cells. Despite these in vitro evidences overexpression of CUL4A did not generate 184B5-derived tumors in orthotopically xenografted mice.

CUL4A overexpression collaborates with H-Ras in the transformation of 184B5 cells
In order to continue deciphering CUL4A role in the transformation of human mammary epithelial cells we evaluated CUL4A cooperation with known oncogenes. To this end by using a retroviral system we further modified the CUL4A-overexpressing and control (transduced with the empty vector) 184B5 cells to stably express c-Myc or the mutant active H-Ras (H-Ras-V12).
Overexpression of c-Myc or H-Ras-V12 was confirmed by western blot ( Figure 6A) and transforming capacities were evaluated by growing cells in soft agar. Cells with exogenous expression of H-Ras-V12 that concomitantly overexpressed CUL4A presented a statistically significant increase in the number of colonies also presented higher number of colonies than control cells (cells transduced with empty vector and c-Myc) however, differences did not reach significance ( Figure 6B). The magnitude of the relative increment in colony number in 184B5 cells that co-expressed H-Ras-V12 and CUL4A was significantly larger (4 fold versus control) than the effect induced in cells that overexpressed CUL4A alone (2 fold versus control) ( Figure 6B).

Proteome expression profiling analysis reveals several mediators of CUL4A activity
To better understand the molecular consequences associated with the modulation of CUL4A expression, we examined changes in protein expression of our CUL4Amodified cellular models. To this end, CUL4A-silenced and overexpressed cells and their corresponding controls (cells transduced with scrambled shRNA and empty vector, respectively) were analyzed using a quantitative proteomic approach. We confidently identified and quantified 5,087 proteins (Supplementary Table S4) from which 764 showed differential expression in at least one of the three experiments conducted (Supplementary Figure S1) (see Materials and Methods). As expected, and technically validating our approach, CUL4A was one of the top-ranking differentially expressed proteins being down-regulated in silenced cells and up-regulated in overexpressed cells ( Figure 7A). We then defined a subset of 64 differentially expressed proteins that showed opposite behavior between the CUL4A-overexpressing cells and at least one of the CUL4A knocked down models. Proteins that did not show regulation or with no available data in the CUL4A-overexpressing cells but that showed same behavior in both CUL4A silenced cells were also included in the list. Among the 64 differentially expressed proteins we found a large variety of protein classes ( Figure  7B) including nucleic acid binding proteins (e.g. HMGA1, PDCD4, MAX), hydrolases (e.g. DDX60, SETX, MX1) and transporters (e.g. TAP1, ABCF3, SLC39A9).
To get further insights into the functional roles of such proteomic changes, we performed enrichment analysis of the 64 proteins by using Ingenuity Pathway Analysis (IPA) which showed a significant overrepresentation of functions such as "cell cycle", "cellular growth and proliferation", "cellular movement" or "antigen presentation" and of canonical pathways such as "interferon signaling" (Benjamini-Hochberg test, P<0.05). Interestingly, several known or putative oncogenes showed up-regulation upon CUL4A overexpression in the 184B5 cells and exhibited down-regulation in the CUL4Asilenced models ( Figure 7A). Among them, SSRP1 and SEPT2 that participate in cell cycle progression [17,18] or MYCBP, HMGA1 and MT2A involved in cell growth and proliferation and previously linked to tumor progression [19][20][21]. Also, replication-dependent histones such as HIST1H4A and HIST2H3A were up-regulated in CUL4A-overexpressing cells and down-regulated in at least one of the silencing models. Conversely, a number of putative or known tumor suppressors were down-regulated in CUL4A-overexpressing cells and/or overexpressed in CUL4A knocked down cells ( Figure  7A). Among them, the Programmed cell death protein 4 (PDCD4) that blocks cell cycle progression and cell proliferation and the anti-tumoral activity-related protein CRABP2 that is involved in cellular growth inhibition [22,23]. In addition a set of 15 proteins that participate in the antitumor immune response such as the tumor suppressors PSMB9, ERAP1 and TAP1 or the interferon-stimulated genes IFIT1 and MX1 were found down-regulated in CUL4A-overexpresing cells and up-regulated in CUL4Asilencing models ( Figure 7A). Given the novelty of a possible relation between the CUL4A E3 ubiquitin ligase and immune response modulation in breast cancer we interrogated IPA to specifically evaluate the potential link between the ubiquitin degradation system and the differentially expressed immune system-related proteins from our list. Remarkably, IPA generated a protein network showing multiple connections between this set of proteins and ubiquitin, which would be consistent with a putative role of CUL4A as modulator of immune system processes ( Figure 7C).

DISCUSSION
The overexpression of the CUL4A E3 ubiquitin ligase has been related to tumor aggressiveness and poor clinical outcome in breast cancer [6,10,24]. One of the mechanisms triggering CUL4A overexpression is the 13q34 amplification and this genomic aberration has been shown to be associated with breast tumors characterized by exhibiting BRCA1 impairment or a basal-like phenotype [12,13]. With these antecedents we were interested in studying the role of CUL4A in the carcinogenic process of the basal-like breast tumor subtype, which is aggressive in nature and lacks effective targeted therapies. Our results show that CUL4A would contribute to the tumorigenicity of basal-like breast cancers through the modulation of cell growth and anti-tumor immune response.
CUL4A silencing in CUL4A-overexpressing basal-like breast cancer cells induced a reduction of cell proliferation and colony formation in short-term assays and long-term soft agar assays and decreased the tumor growth rate of cells injected into mammary glands of nude mice. Importantly, we specifically demonstrated this effect in 13q34 amplification-bearing cells, which is biologically relevant given the association found between this amplification and primary breast tumors of basal-like phenotype [12,13]. We previously found an increased frequency of the 13q34 amplification in BRCA1 tumors [12] and in this study we also demonstrate the growth inhibitory effect of CUL4A silencing in BRCA1-deficient cells. Altogether our results support the role of CUL4A as target gene of the 13q34 amplification. We provide evidence that, rather than being a passenger alteration derived from the increased genomic instability shown by basal-like and BRCA1 primary breast tumors [14,25], CUL4A overexpression would confer selective advantage to tumor cells.
This dependence or ´addiction´ to CUL4A for maintenance of the malignant phenotype and cell survival of CUL4A-overexpressing cells would be similar to that observed in breast tumor cells that overexpress ERBB2 or C-MYC [26,27]. Hence, it may offer the opportunity to take advantage of CUL4A up-regulation in order to develop targeted therapies. In this regard our data suggest that the 13q34 amplification could be a robust biomarker for identification of up to 20% basal-like breast cancer patients [12,13] that might eventually benefit from CUL4A-targeted therapy. Interestingly, in previous studies we observed increased sensitivity to trabectedin in breast cancer cell lines exhibiting high levels of CUL4A in combination with a "BRCAness" status [28]. The potential therapeutic option of CUL4A-targeting in certain subtypes of breast cancer has also been discussed in a recent study reporting a fundamental role of CUL4A in regulating the metastatic behavior of breast cancer cells [11]. However, the authors did not provide data on the effect of CULA4 silencing in normal mammary cells. We showed that CUL4A silencing would not present detrimental consequences on the non-tumorigenic 184B5 mammary epithelial cells. These observations would be consistent with lack of CUL4A dependence in cells that do not overexpress the gene and might suggest low toxicity of eventual CUL4A-targeted inhibitory therapeutic approaches.
Although exogenous overexpression of CUL4A was sufficient to promote proliferation of 184B5 cells, CUL4A up-regulation alone was not enough to generate 184B5derived tumors in xenografted mice. This suggests that CUL4A overexpression, rather than an initiating event, would constitute a secondary oncogenic hit acquired and selected during tumoral evolution. Our results support a potential synergistic effect between CUL4A overexpression and the constitutive activation of the RAS mitogenic signal in the transformation of human mammary epithelial cells, which is consistent with the high frequency of RAS pathway activation in basal-like breast tumors [29,30].
To our knowledge, this is the first study that aims to define mediators of CUL4A transforming capacities in breast cancer using a proteomics approach. By using this methodology we propose novel molecules and pathways that might modulate the oncogenic effect of CUL4A in basal-like tumors. This can shed light on the understanding of the molecular events driven by CUL4A that may be explored in order to develop new therapeutic strategies.
Among the proteins that showed up-regulation upon induction of CUL4A exogenous expression were HMGA1, SSRP1 and MT2A. Interestingly, HMGA1, SSRP1 and MT2A overexpression has been reported in breast cancer [19,20,31] and HMGA1 was previously described as a master regulator of the tumor progression in triplenegative breast cancer cells [32]. These molecules may be good candidates to develop targeted therapies in CUL4Aoverexpressing tumors. In particular, SSRP1 is reported to be a prospective target for anti-cancer small molecules [31,33]. On the other hand, PDCD4 and CRABP2 protein levels were reduced with CUL4A overexpression. Consistently, low levels of PDCD4 in breast tumors have been associated with poor prognosis [34] and high CRABP2 levels have been shown to be indicative of longer overall survival in breast cancer patients [22].
For the first time, we report a putative role of CUL4A in bypassing the immune system in breast cancer through the down-regulation of several molecules involved in the anti-tumor immune surveillance. The overexpression of CUL4A induced a reduction of TAP1, IFIT1 and MX1 protein levels. Low or defective TAP1 in breast tumors predicts higher risks for developing metastasis [35] and down-regulation of the interferonstimulated genes IFIT1 and MX1 has been linked to immune evasion mechanisms and tumor progression [36][37][38][39]. Interestingly, E3 ligases are involved in the efficient regulation of the immune system [40][41][42] and alterations in the ubiquitination machinery/pathway are associated with immune deficiencies and cancer [43].
In summary, we propose CUL4A as an important contributor to the development and progression of basallike breast cancers. Our results indicate that CUL4A might play a central role in the oncogenic process by modulating the expression of a number of oncogenes and tumor suppressors. Interestingly, there is a growing number of E3 ubiquitin ligases being involved in cancer. For example, MDM2 and IAP are highly expressed in several human cancers [44,45] and inhibitors for these enzymes are being evaluated in clinical trials [43]. Thereby CUL4A would constitute a promising target for therapeutic intervention and our data reinforce its clinical value in basal-like breast cancers.

RNA isolation and real-time quantitative PCR (qPCR)
Total RNA was extracted from three independent plates for each cell line (RNAeasy Kit, Quiagen, Valencia, CA) and converted to cDNA by using 500ng as template (High Capacity cDNA RT kit, Applied Biosystems, Foster City, CA). qPCR assays were designed for target gene (CUL4A) and endogenous control (β-ACTIN ) using the Roche Universal Probe Library Assay Design Centre web site (Roche Applied Science, Indianapolis, IN) (Supplementary Table S2). Reactions were performed in triplicate using the ABI Prism 7900HT Sequence Detection System according to the manufacturer's protocol (Applied Biosystems, Foster City, CA). Relative expression was determined using the qBase software that allows for PCR efficiency correction and implements normalization by endogenous genes [46].

Lentiviral constructs
Cell line models were modified for the expression of CUL4A. For gene down-regulation short hairpin RNAs (shRNA) for human CUL4A (NM_003589) were obtained from Sigma-Aldrich (MISSION shRNA Bacterial Glycerol Stock). Five different shRNA constructs were transduced, and the one providing better knockdown efficiency and non-citotoxic effects was selected (SH4, TRCN0000006530; Supplementary Table S3) to silence CUL4A expression in the HCC1937 and MDAMB157 cell lines. A MISSION Non-target shRNA vector (Scramble SHC002) was used as negative control. In order to upregulate the expression of CUL4A a human cDNA of the gene (ID CNIO: hAE5834) containing an HA tag was cloned into FG12-CMV vector (kindly provided by Dr. M. Soengas at CNIO) that expresses green fluorescent protein (GFP). FG12-CMV empty vector was used as negative control.
CUL4A-overexpressing cell models and controls were modified for the expression of H-Ras-V12 and c-MYC by using a retroviral system. The retroviral vectors pBabe-Neo-H-Ras-V12 and pBabe-Puro-c-Myc were kindly provided by Dr. Víctor Javier Sánchez-Arévalo at CNIO.

Generation of stable cell lines
Lentiviral vectors were cotransfected with lentiviral packaging plasmids into 293FT cells (Invitrogen, Carlsbad, CA) in the presence of Fugene 6 (Promega, Madison, WI). The virus-containing supernatants were collected after 36 hours of transfection and filtered. Cells were infected and the stably transduced cells were selected either by adding medium containing 1 mg/mL of puromycin (CUL4A down-regulation) or by GFP sorting by flow cytometry (FACS Calibur; BD Biosciences, San Jose, CA) (CUL4A up-regulation). Modification of CUL4A levels was confirmed by q-PCR and Western blot. Retroviral vectors (pBabe-Neo-H-Ras-V12 or pBabe-Puro-c-Myc) were cotransfected with pCL-Ampho retroviral packaging plasmid into 293FT cells. Infection was carried out as explained above and the selection was performed by using neomycin or puromycin, respectively. www.impactjournals.com/oncotarget Cell viability assay Viability was measured by using the CellTiter-Glo Luminescent Cell Viability Assay from Promega, according to the manufacturer's recommendations. Cells were seeded on 96-well plates at densities of 1500 or 2500 cells per well. Luminescence was recorded using a plate reader (Wallac 1420 VICTOR2™, PerkinElmer, Waltham, MA).

Cell-cycle analysis
Cells were fixed in ethanol and incubated with 50 µg/ml propidium iodide and 10 mg/ml RNAse A (Sigma-Aldrich, St. Louis, MO) followed by analysis on a FACS Calibur flow cytometer (BD Biosciences). At least 10,000 single events were collected. All data were analyzed using FlowJo software (TreeStar, Oregon).

Colony formation assay
Cells were seeded at densities of 1,500 or 2500 cells per well in 6-well plates and incubated for 9 to 10 days. Medium was removed and cells were washed, fixed, and stained with 0.1% crystal violet. Cell density was assessed by solubilization of crystal violet with 15% acetic acid and absorbance of the solution was measured at 590 nm (Wallac 1420 VICTOR2™, PerkinElmer, Waltham, MA). 30,000 cells were embedded in 0.3% agar over a 0.6% base agar layer. Soft agar cultures were fed with growth media once a week and incubated during 3 weeks. To visualize colonies, cultures were stained with MTT and counted manually.

Orthotopic xenografts
Primary tumorigeneis in vivo was evaluated by orthotopicaly inoculating CUL4A-modified breast cell lines into the mammary glands of 6-month-old female adult athymic nude mice (Hsd:Athymic Nude-Foxn1 nu ). One million HCC1937 cells and two million MDAMB157 cells infected with the CUL4A-shRNA vector (or with the Scramble-shRNA) were inoculated. Tumor growth was monitored weekly with a caliper and tumor volumes were calculated using the formula volume= [(length x width 2 )/2]. The experiment was stopped, mice were sacrificed, and tumors were harvested when tumors in control animals reached approximately 1 cm 3 . Mean volume of tumors per group of animals (injected with CUL4A-silenced and control cells) was measured each week. Animal experimentation at the CNIO, Madrid, was performed according to protocols approved by the CNIO-ISCIII Ethics Committee for Research and Animal Welfare (CEIyBA).

Proteomic analysis
An iTRAQ proteomics approach was performed to identify proteins and pathways associated with CUL4A modulation. To this end we looked for differentially expressed proteins between each of the CUL4A-overexpressing cell line models (HCC1937 and MDAMB157 transduced with CUL4A specific shRNA, SH4) and their corresponding control cells (transduced with scrambled shRNA). We also looked for differentially expressed proteins between the CUL4A-overexpresing 184B5 cells (transduced with CUL4A cDNA) and control cells (transduced with empty vector). Cell pellets were extracted and proteins were digested using the FASP protocol [47]. Peptides were labeled with iTRAQ reagents and samples were pooled. Then, the complex mixture was subjected to IEF fractionation. The resulting fractions were separated by on-line nano-LC and analyzed by electrospray MS/MS using a LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, Bremen, Germany). Raw files were searched against SwissProt human database (release date: March 21, 2012; 20329 entries) using MASCOT [48] as a search engine through the Proteome Discoverer (Thermo Scientific) software. Peptides were filtered at 1% FDR using a concatenated target-decoy database. Proteins were considered differentially regulated using a cutoff > 1.25-fold. A more detailed description of methodology is provided in the Supplementary Information.

Statistical analyses
We used SPSS version 13 software to conduct the statistical analysis of our data. Two-tailed Student t tests were used for comparisons between different groups. Statistical analysis of the mean tumor volumes was done by U test of Mann-Whitney analysis. P< 0.05 and P< 0.01 were considered to be significant. Unless otherwise stated, the error bars represent standard deviation of the biologic triplicates.