Cancer Cell Biology
Downregulation of tissue factor by RNA interference in human melanoma LOX-L cells reduces pulmonary metastasis in nude mice
Version of Record online: 18 AUG 2004
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 112, Issue 6, pages 994–1002, 20 December 2004
How to Cite
Wang, X., Wang, M., Amarzguioui, M., Liu, F., Fodstad, Ø. and Prydz, H. (2004), Downregulation of tissue factor by RNA interference in human melanoma LOX-L cells reduces pulmonary metastasis in nude mice. Int. J. Cancer, 112: 994–1002. doi: 10.1002/ijc.20527
- Issue online: 12 OCT 2004
- Version of Record online: 18 AUG 2004
- Manuscript Accepted: 23 JUN 2004
- Manuscript Received: 16 JAN 2004
- Research Council of Norway
- Norwegian Cancer Society
- short hairpin RNA;
- knockdown of tissue factor;
- anchorage-dependent cell proliferation;
- lung tumors;
- cDNA microarray analysis
Tissue factor (TF) is the membrane receptor of the serine protease coagulation factor VIIa (FVIIa). Formation of the TF/FVIIa complex initiates the coagulation cascade. We used short hairpin RNA (shRNA)-mediated RNA interference to knock down TF expression in the human metastatic melanoma cell line LOX-L. After transfection with the shRNA construct, 3 stable clones with significantly downregulated TF expression were established. They exhibited decreased proliferation in vitro as determined by 14C thymidine incorporation and soft agar assay. The in vivo metastatic potential was assessed in an experimental pulmonary metastasis model in which cells from different clones were injected into the tail vein of nude mice. The incidence of pulmonary tumors was significantly lower in mice receiving shRNA-expressing cells (33% ± 15%) than in control mice injected with wild-type cells or cells stably transfected with empty expression vector (90% ± 10%). The mice injected with TF-downregulated cells had markedly longer survival time (69 ± 17 days) compared to the control mice (35.6 ± 5 days; p = 0.03). Thus, reduction of TF levels in LOX-L cells significantly delayed and reduced lung tumor formation. As a first step in elucidating the molecular basis for this effect, we compared the global gene expression profile in TF-downregulated cells and control cells by using cDNA microarray analysis. Forty-four known human genes were found to be significantly upregulated (> 2-fold; p < 0.05) and 228 genes significantly downregulated (≥ 3-fold; p < 0.05) in TF-downregulated cells compared to control cells. The differentially expressed genes encode proteins functioning in transcription, translation, cell communication and cell growth/death. The results provide a basis for investigating molecular mechanisms underlying the effects of TF on the metastatic capacity of LOX-L melanoma cells. © 2004 Wiley-Liss, Inc.
Tissue factor (TF), a transmembrane protein, initiates the blood coagulation cascade by complexing with its serine protease ligand factor VII(a) (FVIIa), ultimately leading to thrombin generation, fibrin deposition and platelet aggregation. TF is aberrantly expressed in many solid tumors and tumor cell lines1, 2, 3, 4 with higher expression levels in malignant tumor cells compared to their nonmalignant counterpart. Subsequent results suggested that TF participates in the tumor growth and dissemination of various cancers.5 Clinically, TF expression strongly correlates with tumor progression.6, 7, 8, 9 TF can promote tumor progression via coagulation-dependent or -independent mechanisms.10, 11 Activation of coagulation may trap tumor cells in a fibrin-platelet clot and allow local tumor growth, which is critically important for tumor dissemination and metastasis.12, 13 Moreover, several products of the coagulation system, including thrombin and fibrin, can stimulate angiogenesis,13, 14, 15 which is necessary for tumor growth and metastasis. Independent of the downstream coagulation factors, FVIIa binding to TF can promote tumor-associated angiogenesis by upregulating angiogenic factors VEGF and IL-8.16, 17, 18 TF/FVIIa signaling resulting from the proteolytic activity of the complex is required for the upregulation of these angiogenic factors19, 20, 21 and for TF-dependent metastasis.22 TF/FVIIa signaling has received widespread interest10, 23, 24, 25, 26 after the discovery of TF/FVIIa-induced calcium signals.27 FVIIa/TF signaling, independent of the coagulation cascade, results in the upregulation of cytokines, growth factors22, 24 and induction of antiapoptosis,28 which may contribute to tumor growth. Blocking TF function by administration of TF antibody or inactivated FVIIa has resulted in the reduced metastasis in murine metastasis models.29, 30 We have in this study used RNA interference specifically to knock down TF expression and investigated the effects on the metastatic capacity of human malignant melanoma cells.
RNA interference (RNAi) is a process of sequence-specific posttranscriptional gene silencing triggered by double-stranded RNA. This phenomenon has been developed into a powerful technology to elucidate mammalian gene function by delivery of small interfering RNA (siRNA), duplexes of 19-21-nt RNA with 2-nt 3′ overhangs, into cells.31 Variable transfection efficiency and the transient nature of silencing by exogenously delivered siRNA limit the applicability of chemically synthesized siRNA in functional genomic studies.32 Stable RNAi can be achieved in mammalian cells through RNA polymerase III-dependent expression of so-called short hairpin RNAs (shRNA) from integrated DNA constructs in stably transfected cells.33, 34 The transcribed shRNA is thought to be processed into siRNA-like molecules intracellularly. Studies from different cell types and against various genes have demonstrated the efficacy of endogenously expressed shRNA.33, 35, 36, 37
In order to explore the potential role and possible mechanism of TF in melanoma tumor progression, we have employed the RNA interference technology to knock down TF expression in LOX-L metastatic melanoma cells. The phenotypic changes resulting from the reduction of TF expression were studied both in vivo and in vitro. We show that TF downregulation reduced formation of pulmonary metastasis and prolonged mouse survival. Global gene expression analysis revealed changes in gene expression upon TF knockdown, which may explain the mechanisms whereby TF promotes the melanoma progression.
MATERIAL AND METHODS
The human metastatic melanoma LOX-L cell line was cultured in RPMI-1640 medium (Gibco, Gaithersburg, MD) supplemented with 10% inactivated fetal calf serum (FCS). The LOX-L cell line has a nearly normal diploid karyotype and resembles its original tumor characteristics during the establishment of the cell line.38
Cloning of short hairpin RNA construct
A cDNA construct (pZeo-U6-hTFsh) encoding an shRNA (Fig. 1a) targeting hTF16739 was cloned downstream of a U6 polymerase promoter. The plasmid (pTZ-U6-hp) containing the U6 promoter and Pol III terminator sequences was a kind gift from Dr. John Rossi (Beckman Research Institute, City of Hope, CA). Briefly, the vector pTZ-U6-hp was cut with SalI, blunted with mung bean nuclease, cut again with XbaI, then dephosphorylated with calf intestinal phosphatase. Linearized vector was purified by QiaEXII (Qiagen, Chatsworth, CA). Two partially overlapping oligos were annealed and the recessed ends were filled in with Klenow DNA polymerase. This yielded a specific dsDNA with sequence containing the siRNA against TF (5′-GCGCTTCAGGCACTACAAA-3′) separated by a short spacer from its reversed complementary sequence. After heat inactivation, the Klenow-extended cassette was cut with XbaI and phosphorylated with polynucleotide kinase. The cassette was then purified and ligated into the vector fragment. Correct inserts of transformants were confirmed by sequencing. For stable selection, the entire promoter-shDNA-terminator cassette was subcloned as a 0.36 kb BamHI fragment into a pZeoSV-derived vector (Invitrogen, La Jolla, CA) devoid of SV40 sequences, producing the plasmid pZeo-U6-hTFsh.
Transfection and selection of stable transfectants
LOX-L cells were transfected at 40–50% confluency in 36 mm wells with 1.0 μg pZeo-U6-hTFsh or empty vector at a 3:1 (v/w) ratio of Lipofectamine 2000 to DNA as recommended by the manufacturer. Complexes (200 μl) were diluted with 800 μl nonsupplemented medium before addition to washed cells. Cells were incubated with complexes for 5 hr before replacement with full medium. Selection with 200 μg/ml Zeocin (Cayla, Toulouse, France) in full medium was initiated 2 days after transfection. Selection medium was replaced the day after and then every second day. Single colonies were isolated and maintained at 200 μg/ml Zeocin. The colonies were characterized and used in subsequent experiments.
To quantify the expression of tissue factor antigen in the stable transfectants, the Imubind Tissue Factor ELISA kit (American Diagnostica, Greenwich, CT) was used according to the recommendation of the manufacturer. Every sample was analyzed in duplicate and TF antigen level was normalized to total cellular protein. The results are presented as percent of control.
14C thymidine incorporation
LOX-L cells were seeded at 2 × 105 in 6-well plates in triplicates and incubated for 20 hr; 2 ml of fresh medium containing 14C thymidine (1 μCi/ml; Amersham, Buckinghamshire, U.K.) was then added into each well. After 22 hr, the incorporation of the radiolabel into DNA was determined. Briefly, cells were washed with ice-cold PBS, then extracted with ice-cold 5% trichloroacetic acid (TCA). The precipitates were then incubated for 20 min on ice in 0.5 ml of 1 M NaOH and solubilized by repeated pippetting. Cell extract (400 μl) was transferred into scintillation vials containing 4 ml of scintillation cocktail (Packard, Meriden, CT) to determine the tritium incorporation by liquid scintillation counting. Protein concentration was measured using the Bio-Rad DC kit (Bio-Rad, Richmond, CA). Thymidine uptake was calculated as [total counts (cpm)]/[total protein amount (μg)]. LOX-L cells transfected with empty vector served as controls.
Soft agar assay
The soft agar assay was performed as described by Courtenay et al.40 LOX-L cells (1,000 counted cells) for the stably transfected clones C10, C12 and LOX-vector contained in 0.2 ml complete RPMI-1640 (10% FCS + RPMI) were added into separate 10 ml Falcon tubes each containing 0.2 ml Rowett rat blood (1–8 dilution) and 0.6 ml of 0.5% agar (Difco Laboratories, Detroit, MI) in complete RPMI. Cells were incubated at 37°C with 5% O2, 5% CO2 and 90% N2. One week later, 1 ml of complete RPMI medium was replenished and cells were cultured for an additional 11 days. Clones were counted under a Zeiss stereo microscope. A separate cluster with more than 50 cells was regarded as a clone. Colony formation in the TF-reduced clones was presented as the percentage of that in control cells. The experiment was performed twice and in triplicate.
Northern blot analysis
mRNA was extracted using the Dynal oligo (dT)25 beads. Procedures for RNA hybridization and signal quantification were as described.18 The signal of TF was normalized to that of GAPDH. Full-length cDNAs of human TF and GAPDH were used to generate probes for hybridization.
Cells were injected into GBNIH or NCR immunodeficient nude mice bred at the nude rodent facility of the Norwegian Radium Hospital. Mice used in these experiments were 6–8 weeks old, weighing 15–25 g. Regularly, 5 mice were included in each of 5 groups. Mice in each group received cells of one stably transfected LOX-L clone (C9, C10, C12, vector-transfected) or untransfected wild-type controls. Single cell suspensions were injected into the tail vein of each mouse. After injection, the general health condition of the mice was inspected daily. In the first experiment, all mice were killed by dislocation of the neck and checked for metastasis when the first mice developed respiratory symptoms after injection of 5 × 105 cells. In subsequent experiments, after injection of 106 cells in 0.1–0.2 ml, the animals were inspected daily and survival time was recorded. The animals were sacrificed once they developed respiratory symptoms attributable to lung metastasis.38 At the 94th day after injection, all the remaining mice were sacrificed and checked for metastasis. All procedures and experiments involving animals were approved by the National Animal Research Authority and carried out according to the European Convention for the Protection of Vertebrates Used for Scientific Purposes.
RNA extraction and cDNA microarray hybridization
Total RNA was extracted by the standard phenol:chloroform extraction method. Fluorescence-labeled cDNA probes were prepared from 20 μg of total RNA by using the FairPlay microarray labeling kit (Stratagene, La Jolla, CA) according to the manufacturer's recommendation. This kit separates generation of amino allyl-modified cDNA and CY-dye coupling into 2 steps, allowing a more even dye coupling efficiency between different dyes. A human cDNA microarray consisting of 15,000 cDNA spots [including Lucidea Universal ScoreCard (Amersham) as controls and 12,940 unique genes] were produced at the Norwegian cDNA microarray consortium (Norwegian Radium Hospital, http://www.med.uio.no/dnr/microarray). Hybridized slides were scanned with an Axon Instruments GenePix 4000B scanner and the scanned images were analyzed with the software program GenePix Pro 4.0 (Axon, Union City, CA).
cDNA microarray data analysis
Control spots of Lucidea Universal ScoreCard (Amersham) were printed throughout the slide. The test and control mRNA samples were separately spiked with calibration control mRNA and labeled in the reverse transcription reaction. As an internal control, control mRNA spikes will specifically hybridize to the control spots. The data were analyzed using BASE (http://pubgeneserver.uio.no/base/).41 Briefly, spots with a signal-to-noise ratio below 2 for any channel were regarded as unreliable spots and not included in the data analysis. Global median ratio normalization was applied to all slides and the sample-to-control ratio for each gene was calculated. One-tailed Student's t-test was performed on all 4 microarray data sets and p-values less than 0.05 were regarded as significant. Only genes with a p-value < 0.05 were considered for further analysis. The clone IDs for these genes were translated into gene symbol according to the Gene Ontology Consortium,42 which subsequently served as the input to GoMiner. GoMiner was used as a gene clustering tool based on gene function.43
The results are presented as mean ± SD. Data were analyzed by using the Student's 2-tailed t-test. Differences were considered significant at p < 0.05.
Evaluation of short hairpin RNA efficacy and stable suppression of TF expression
In mammalian cells, delivery of siRNA can specifically and efficiently silence gene expression. We have previously identified a site within human TF that is highly susceptible to siRNA.39 This site was now targeted with a short hairpin RNA with the same configuration that was reported by Brummelkamp et al.37 to be effective (Fig. 1a). The efficacy of the hTF-specific shRNA-encoding construct (hTFsh) was evaluated by transient cotransfections in HaCaT cells. These experiments demonstrated 80–85% inhibition of transient hTF-dependent luciferase expression by hTFsh compared to the control constructs coding for irrelevant shRNAs (data not shown). This level of inhibition was similar to that attained by the corresponding chemically synthesized siRNA in both HaCaT and LOX-L cells.39 The construct hTFsh was then endogenously expressed in LOX-L cells from a Zeocin selection vector. Stably transfected colonies were isolated and evaluated for TF antigen expression by ELISA. The TF antigen level was normalized to total protein. Normalized TF antigen levels were substantially reduced in 3 clones (C9, C10 and C12; Fig. 1b) compared to levels in polyclonal control cells transfected with empty vector (LOX-vector, 5461 ± 5 pg/mg protein). To control for effects related to the chromosomal integration site of the plasmid, all 3 stable cell lines were used in later experiments. Four human transcripts shared sequence similarity (up to 50 %) with our selected target sequence human TF, but no differential expression was observed (data not shown). Therefore, we conclude that there is no substantial off-target primary effect of shRNA in our study.
Reduced proliferation in hTFsh-expressing LOX-L cells
The 3 stable clones with reduced TF expression (LOX-C9, LOX-C10 and LOX-C12) grew more slowly in culture than both parental and LOX-vector cells. The proliferation rate was therefore investigated by measuring 14C thymidine incorporation (Fig. 2a). LOX-vector control cells exhibited indistinguishable thymidine incorporation from the parental LOX-L cells, indicating that the process of stable selection did not alter the proliferation rate of the cells. The hTFsh-expressing clones, however, exhibited modest but significantly reduced rates of proliferation (about 15% reduction; p < 0.05) as measured by thymidine incorporation. This suggested that reduction of TF expression reduced the rate of DNA synthesis in LOX-L cells. One of the phenotypes of cells having undergone malignant transformation is their ability to grow with anchorage independence in soft agar assays. We used a soft agar assay to examine whether reduction of TF expression correlated with a reduction of the anchorage–independent growth potential of LOX-L cells. The TF-downregulated LOX-L clones demonstrated a 70–80% reduction in colony formation compared to the vector-transfected control cells (Fig. 2b). These data thus suggested that specific reduction of TF expression attenuated both anchorage-dependent and -independent growth potential in LOX-L cells.
Effects of reduced TF expression on tumor growth and metastasis
The effect of reduced TF expression on the metastatic potential of LOX-L cells was examined in a mouse pulmonary metastasis model using immunodeficient nude mice. The incidence of pulmonary tumors appearing is generally considered to express the metastatic potential of the injected cells. In the first experiment, cells from the 3 clones with downregulated TF expression as well as parental and vector-transfected control cells were injected intravenously and evaluated for their metastatic potential. All mice were killed and examined for metastasis at the time when the first mouse developed respiratory symptoms after injection. Only 1 of 16 mice receiving TF-downregulated cells developed lung tumor colonies, while 4 out of 9 mice receiving control cells had lung tumors 34 days after injection. The lungs of mice in the control groups often contained confluent tumors. In contrast, the only lung with tumors from the TF downregulation cells developed separate and much smaller tumors. These data clearly demonstrated that knockdown of TF expression in LOX-L cells delayed and reduced their tendency to form pulmonary tumors.
Next, we addressed the question of the degree of delay in pulmonary tumor formation in TF-downregulated LOX-L cells. After injection, the animals were inspected daily and sacrificed for examination of metastasis in all organs once they developed respiratory symptoms. On the 94th day after injection, all the remaining mice were sacrificed and checked for tumor formation. They proved to be free from visible tumors in all organs, indicating that the visual evaluation of respiratory symptoms in the whole animal correlated well with the incidence of tumor metastasis in mice. In total, 45 mice were evaluated in 2 independent experiments. The incidence of tumors (Fig. 3) demonstrated that 33% ± 15% of mice injected with TF-downregulated cells had tumors, whereas the incidence was 90% ± 10% in mice injected with control cells (both LOX-WT and LOX-vector). There was no significant difference in tumor formation between parental and vector-transfected cells. The survival time for each mouse with tumor formation is described (Table I). The mice injected with the TF-downregulated cells had significantly longer survival time (69 ± 17 days) compared to mice given control cells (35.6 ± 5 days; p = 0.03).
|Cell line||Number of mice with tumors to number of injected mice||Symptom appearance (days after injection)||Average survival time (days, mean ± SD)|
|2/5||32||31.4 ± 1.4|
|1/10||36||38 ± 5.6|
|1/10||52||70 ± 18.9|
|C10||1/10||46||58.7 ± 11|
|C12-B||1/5||37||51.6 ± 21|
Interestingly, 6 months after the establishment of stable clones, TF expression in the previously low-expressing clone C12 reverted to the level as seen in LOX-vector cells (Fig. 4). Concomitant with this reversion in TF expression, the cells (C12-B) now exhibited an increased metastatic potential compared to the low TF-expressing C12 cells in terms of the incidence of metastasis and survival time (Fig. 3, Table I). These data therefore suggest that reduction of TF expression in tumor cells retarded tumor formation and prolonged the survival time of mice, whereas restoration of TF expression returned the metastatic potential to normal. Thus, TF expression in LOX-L melanoma cells correlates with their tendency for pulmonary metastasis.
Identification of differentially expressed genes by TF repression
To elucidate the mechanisms whereby downregulation of TF expression decreased the metastatic potential, we assessed the change of global gene expression using cDNA microarray profiling of stably transfected cells with normal (LOX-vector) and downregulated (C9, C10) TF levels. RNA from clones C9 and C10 was separately hybridized to the reference RNA obtained from LOX-vector control cells. Two independent experiments were performed in duplicate, producing 4 data sets. Student's t-test was implemented to each gene in all data sets to identify differentially expressed genes. In TF-downregulated cells, 228 genes were significantly downregulated 3-fold (p < 0.05), among them 25 genes at p < 0.001. Forty-four genes were significantly upregulated more than 2-fold (p < 0.05), among them 23 genes at p < 0.01. The upregulated and most significantly downregulated genes are listed in Tables II and III, respectively. The genes that were upregulated in TF-downregulated cells have functions in regulation of apoptosis or cell death, transport and cytoskeleton organization (Table II).
|Gene symbol||Gene description||Clone ID||p-value||Fold change|
|CCT4||Chaperonin containing t-complex polypeptide 1 subunit 4||897880||0.03||2.77|
|TMSB10||Thymosin beta 10||840788||<0.01||2.41|
|TPM2||Tropomyosin 2 beta||740620||<0.01||2.27|
|SLC7A5||Cationic amino acid transporter||755578||<0.01||3.99|
|AQP4||Aquaporin 4 C2 isoform||279172||<0.01||2.41|
|SLC3A2||Solute carrier family 3||856454||0.02||2.16|
|Apoptosis and cell death|
|DXS1357E||B-cell receptor-associated protein 31||877832||<0.01||4.36|
|MFGE8||Milk fat globule-EGF factor 8 protein||785744||<0.01||3.26|
|PSMB5||Proteasome beta 5 subunit||1460110||0.02||3.25|
|PSMD2||Proteasome 26S non-ATPase subunit 2||809992||<0.01||2.78|
|PDCD61P||Programmed cell death 6 interacting protein||242955||<0.01||2.43|
|PDCD1||Programmed cell death 1 precursor||2464548||0.02||2.36|
|GRIM19||Cell death-regulatory protein||321389||<0.01||2.15|
|STK38L||Serine/threonine kinase 38 like||307304||0.03||2.23|
|SIP2-28||Calcium and integrin binding 1||841679||<0.01||2.21|
|SEC4L||Member of RAS oncogene family||81315||0.03||2.06|
|Regulation of transcription|
|MYC||p67 myc protein||417226||0.03||3.95|
|FOSL1||Fos related antigen 1||110503||<0.01||2.47|
|HCNGP||Transcriptional regulator protein||838662||0.02||2.38|
|ATF4||Activating transcription factor 4||949971||<0.01||2.28|
|CDK4||Cyclin-dependent kinase 4||842806||<0.01||3.15|
|HSPA1L||Heat shock 70 kda protein 1-like||50615||0.04||3.04|
|RPL32||Ribosomal protein L32||32684||0.03||2.88|
|RPL10||Ribosomal protein L10||66686||0.02||2.65|
|TNFRSF5||Tumor necrosis factor receptor superfamily member 5 precursor||261519||<0.01||2.46|
|AKAP12||A-kinase anchor proteins||784772||0.02||2.41|
|HF1||Complement factor H1||84750||<0.01||2.41|
|RAGE||Renal tumor antigen||289666||<0.01||2.31|
|MAT2A||Methionine adenosyltransferase II, alpha||79502||0.03||2.22|
|PDAP1||PDGFA-associated protein 1||824426||<0.01||2.15|
|NME1||Nonmetastatic cells 1 protein||845363||0.03||2.12|
|PTP4A1||Protein tyrosine phosphatase type IVA, member 1||42739||0.04||2.08|
|PMS2||Postmeiotic segregation increased 2||116906||0.05||2.03|
|GABARAP||GABA(A) receptor-associated protein||810741||0.04||2.01|
|Gene symbol||Gene description||Clone ID||Fold change|
|Cell growth and maintenance|
|COX17||Cytochrome c oxidase copper chaperone||489823||8.23|
|NDUFB4||NADH dehydrogenase (ubiquinone) 1 beta subcomplex||450896||7.14|
|CDKN3||Cdk inhibitor 3 (CDK2-associated dual specificity phosphatase)||700792||4.17|
|CEP3||CDC42 effector protein 3 (rho GTPase binding)||683059||3.87|
|Protein synthesis and catabolism|
|RPL39||Ribosomal protein L39||244911||13.06|
|SUI1||Translation initiation factor||1607039||5.64|
|REN||Renin beta chain||813402||5.12|
|MRPL35||Mitochondrial ribosomal protein L35||490147||5.10|
|Transcription and RNA processing|
|LSM3||U6 snRNA-associated Sm-like protein (MDS017)||796176||5.94|
|POLR2C||DNA directed RNA polymerase II polypeptide C||770391||5.23|
|KOC1||IGF-II mRNA-binding protein 3||429494||4.87|
|PLAUR||Plasminogen activator, urokinase receptor||810017||6.57|
|ARHG||Ras homolog gene family, member G (rho G)||159118||4.18|
|CDR2||cerebellar degeneration-related protein||366067||3.24|
|SNX14||Sorting nexin 14||284792||4.87|
|PP15||Nuclear transport factor 2 (NTF-2)||299388||4.60|
|SSR3||Signal sequence receptor gamma subunit||795353||3.81|
|PIN||Protein inhibitor of neuronal nitric oxide synthase||853938||7.05|
|CTNNA1||Catenin (cadherin-associated protein) alpha 1||897164||4.13|
|EBRP||Emopamil binding related protein||416240||4.11|
|AHCYL1||S-adenosylhomocysteine hydrolase-like 1||809758||4.05|
|TMEM5||Transmembrane protein 5||430007||3.47|
|MLF2||Myeloid leukemia factor 2||810743||3.11|
The most significantly downregulated genes (> 3-fold) are involved in a wide range of cellular processes such as cell communication, cell growth and maintenance, cell death, translation and transcription. Downregulation of transcription factors may represent a key element in the observed gene downregulation. At least 17 downregulated genes play roles in different steps of cell proliferation and some are related to apoptosis. Altered expression of signaling molecules may elicit multiple cellular responses. These signaling molecules include G-protein-coupled receptor, transmembrane receptor protein serine/threonine kinase, transmembrane receptor protein tyrosine kinase, transmembrane receptor protein tyrosine phosphatase, integrin-mediated signaling, mitogenic-activated protein kinases. The characteristics of the downregulated gene expression profile reflect the general inhibition of cell growth and protein synthesis, which may contribute in part to the reduced tendency of metastasis in mice.
The connection between blood coagulation and cancer was realized a century ago.44 Recently, the role of TF in cancer has received much attention, as TF is the major initiator of the coagulation cascade and a protease binding receptor. In this study, we have demonstrated that downregulation of TF by endogenous expression of a short hairpin RNA targeting TF resulted in reduced metastatic potential of human malignant melanoma cells in nude mice, demonstrating that TF is of importance in promoting tumor metastasis. We have investigated the effects of TF knockdown on global gene expression, identifying gene products that may contribute to the observed reduction in metastatic potential following the reduction of TF expression. Our results expand the previous work in murine models that have demonstrated the correlation of TF expression with metastasis.16, 30
The specific knockdown of a target gene without affecting unintended genes is of critical importance for investigating gene function. Recently, the specificity of siRNA- and shRNA-mediated RNA interference has been studied by genomewide expression profiling.45, 46, 47 The conclusions of these studies are somewhat conflicting. In the study observing substantial off-target activity, the majority of said activity appeared to be due to sequence similarity between siRNA and unintentionally targeted genes.47 We detected no differential expression of the genes having the highest similarity to our target sequence. Therefore, we conclude that there is no substantial off-target primary effect of shRNA in our study. It has been suggested that exogenous delivery of high doses of siRNA will cause nonspecific effects.46 Most of these effects may be due to the toxicity of the liposomes commonly used for transient transfection. These specific concerns therefore do not apply to endogenously expressed shRNA. In addition, compared to the endogenously expressed shRNA, a similar degree of TF mRNA reduction without toxicity was observed in LOX-L cells by transient transfection of the corresponding chemically synthesized siRNA against hTF, indicating that the endogenous expression level of shRNA is not overdosed and has no toxicity to cells. Another side effect recently reported for siRNA is induced activation of the interferon pathway.48, 49 We did not observe any differential expression among 15 interferon and interferon-induced genes. Therefore, the observed phenotypic changes in our system are most likely due to downregulation of the target TF gene.
The molecular mechanism by which TF promotes tumor metastasis is not fully understood. In this study, we have performed a first step in attempts to elucidate the underlying molecular basis by comparing the global gene expression of the TF-deficient cells with normal expression LOX-L cells. Our findings suggest that a large subset of genes involved in cell growth, invasion and angiogenesis is downregulated in TF-deficient clones. Of special interest are 20 genes encoding ribosomal proteins as well as the eukaryotic translation elongation factor 1 alpha 1 (eef1a1) and translation initiator (sui1). This suggests that downregulation of TF affects the translation machinery, thereby slowing down protein synthesis. Overexpression of ribosomal proteins has been reported in transformed phenotypes, including melanoma,50 where they are needed for maintaining cell homeostasis and survival.51 In addition to ribosomal genes, the downregulated genes included (pro)oncogenes (myb, c-yes, vav2, src and ras family) and genes for growth factors and receptors (vegfc, ngfrap1 and pdgfra). This downregulation may account for decreased tumor formation, proliferation and angiogenesis. Proteolytic enzymes, such as matrix metalloproteinase (MMP) and members of the plasminogen activator-plasmin system, contribute to the tumor progression by degrading the extracellular matrix component. Downregulation of mmp19 and urokinase-type plasminogen activator receptor (plaur) genes was also observed, which might lead to decreased invasiveness. It is worth noting that some of these genes are either of importance in relaying TF/FVIIa signaling (yes, src, ras, pkc and MAP kinase)52, 53 or effectors of TF/FVIIa signaling (vegf and plaur).20, 21, 54
In concordance with the decreased tumor growth of the TF-downregulated cells in vivo, these stable clones also had reduced in vitro proliferation, as evidenced by thymidine uptake and reduced ability to grow in soft agar. This is in agreement with other studies where TF is identified as an immediate-early gene, whose expression correlated with the induction of cell growth.55 TF/FVIIa stimulated cell growth and increased the antiapoptotic ability of cells.28 These effects may associate with the activation of mitogenic and survival pathways induced by TF/FVIIa signaling.24, 28, 56 Although in some cell types TF is physically associated with cytoskeleton proteins and accumulates in the midbodies during cytokinesis, colocalizing with actin-rich regions,57, 58 the underlying role of TF in cytokinesis is yet unknown. Three out of 44 upregulated genes are cytoskeleton actin genes (tpm2, vim and tmsb10). The biologic cause and effect of this upregulation warrant further study. Compared to the control cells, TF-downregulated cells showed much more reduced growth potential in the soft agar assay than in the thymidine incorporation assay. These 2 assays measure different growth characteristics. Anchorage-independent growth is tested over a longer period (2–3 weeks) and in 3-dimensional matrix, thus collectively reflecting tumor growth potential in vivo. In contrast, the thymidine incorporation assay used reflects the rate of DNA synthesis during one cell cycle of logarithmic growth. As demonstrated in our microarray data, certain genes involved in degrading extracellular matrix are downregulated when TF is low, which may collectively reduce colony formation. On the other hand, several genes involved in apoptosis and cell death were upregulated in the TF-deficient clones. We detected, however, no typical apoptotic cleavage patterns (e.g., of caspase 3 or PARP) in Western blot (data not shown), indicating that cells with lower TF expression did not undergo significantly more apoptosis in culture.59, 60, 61
The proteolytic activity of TF/FVIIa is important for TF-dependent metastasis in a murine model. This was shown by injection of cells transfected with a mutated TF gene, in which FVIIa can no longer bind to TF.30 These cells displayed largely reduced metastasis compared to the cells transfected with the wild-type TF gene. Therefore, a reduced protease activity of tumor cells caused by the reduced TF protein expression may also contribute to the decreased metastatic potential in vivo. Since FXa, thrombin and fibrin can mediate tumor metastasis via different mechanisms,14, 62, 63, 64, 65, 66 reduced activation of downstream coagulation factors may also partly contribute to the effect of reduced tumor metastasis. However, previous work has demonstrated that independent of the coagulation cascade, TF/FVIIa signaling can induce expression of cytokines, growth factors, angiogenic factors and MMPs.18, 20, 21, 24 Downregulation of some of these genes has been revealed by microarray analysis in our study. Together, this implies that TF/FVIIa signaling itself is of importance in angiogenesis and metastasis.
In summary, we show that downregulation of TF in metastatic LOX-L cells resulted in reduced metastasis formation in immunodeficient mice. A first step in elucidating the molecular basis of this effect was to perform genomewide expression profiling. The combined alterations following downregulation of TF, leading to decreased cell growth, and putatively to loss of invasiveness and angiogenesis, may collectively contribute to the reduced metastatic ability of LOX-L cells. The results provide new evidence for the role of TF in inducing molecular and cellular characteristics promoting tumor progression.
The authors thank Dr. Rut Valgardsdottir and Dr. Antony Mullen for constructive suggestions. Supported by the Research Council of Norway and the Norwegian Cancer Society (to H.P.).
- 41BioArray Software Environment (BASE): a platform for comprehensive management and analysis of microarray data. Genome Biol 2002; 8: software 003., , , , , .