Antithrombin, a serpin family protease inhibitor crucial to hemostasis, acquires antiangiogenic properties on undergoing conformational alterations induced by limited proteolysis or elevated temperature. To better understand the biochemical mechanisms underlying antithrombin antiangiogenic activity, we did genome-wide expression profiling, coupled with quantitative reverse transcription-PCR, Northern blot, and Western blot analyses, to characterize the gene expression patterns that are induced by antiangiogenic antithrombin in cultured primary human umbilical vein endothelial cells. Overall, 35 genes with significantly increased expression and 93 genes with significantly reduced expression (≥2-fold changes) due to antiangiogenic antithrombin treatment were identified. More than half of the down-regulated genes have well-established proangiogenic functions in endothelial cells, including cell-surface and matrix proteoglycans (e.g., perlecan, biglycan, and syndecans 1 and 3) and mitogenesis-related signaling proteins (e.g., mitogen-activated protein kinase 3, signal transducers and activators of transcription 2, 3, and 6, and early growth response factor 1). In contrast, most up-regulated genes (e.g., caspase-3, p21, tissue inhibitor of metalloproteinases 1, 2, and 3, and adenomatosis polyposis coli) are known for their antiangiogenic functions which include the promotion of cell apoptosis and cell cycle arrest and the inhibition of tumor growth and metastasis. These results show that the antiangiogenic activity of antithrombin is mediated at least in part by a global genetic reprogramming of endothelial cells and strongly implicate an endothelial cell ligand-receptor signaling mechanism in this reprogramming. (Cancer Res 2006; 66(10): 5047-55)

Angiogenesis, the formation of new capillaries from preexisting vasculature by the proliferation, migration, and differentiation of endothelial cells, is a fundamental process required for a number of physiologic and pathologic events (1). Under physiologic conditions, angiogenesis is a highly regulated phenomenon and is controlled by the balance between angiogenic stimulators and inhibitors. However, unregulated angiogenesis is observed under pathologic conditions such as tumor growth, diabetic retinopathy, and psoriasis. Angiogenesis plays an important role in tumor growth, invasion, and metastasis (2). Compared with conventional clinical approaches for tumor therapy, antiangiogenesis-mediated tumor therapy offers several unique advantages including decreased toxicity to the host, lack of drug resistance, and broad-spectrum efficacy against tumors of varied origins (3). Therefore, using natural endogenous angiogenesis inhibitors to cut off the nutrients necessary for cancer cell growth has become one of the most promising tumor therapy approaches.

Antithrombin is one of the most important endogenous regulators of blood coagulation (4). Recent studies have shown that conformationally altered cleaved and latent forms of antithrombin have a strong antiangiogenic activity, as shown by their abilities to inhibit growth factor–stimulated proliferation, migration, and capillary tube formation in cultured endothelial cells and to induce tumor regression in an in vivo mouse model (57). More interestingly, cleaved and latent forms of antithrombin were found to not only constitute major components of endogenous angiogenic inhibitors secreted by human primary pancreatic cancer cells, which were capable of blocking secondary tumor growth (8), but to also exhibit an inhibitory efficacy in human pancreatic cancer regression in mice comparable with other well-known potent antiangiogenic agents such as endostatin and TNP-470 (9). Previous studies have shown that altered forms of antithrombin exert their antiangiogenic effects by inducing cell apoptosis (6, 9), inhibiting focal adhesion kinase activation (6), causing cell cycle arrest (10), and down-regulating the expression of the proangiogenic heparan sulfate proteoglycan (HSPG), perlecan, in endothelial cells (10). However, the precise biochemical mechanisms underlying antithrombin antiangiogenic activity still remain unclear. Because angiogenesis is a complex physiologic process which is tightly controlled by the integration of a multitude of gene activities (1), we expected that antithrombin would produce its antiangiogenic effects by inducing a unique antiangiogenic signaling network manifested by global alterations in endothelial cell gene expression. To test this hypothesis and provide new insights into the biochemical mechanisms of antithrombin antiangiogenic action, we sought to identify the gene expression pattern induced by antiangiogenic antithrombin treatment of endothelial cells using cDNA-based transcriptional profiling, coupled with real-time quantitative reverse transcriptase PCR (RT-PCR), Northern blot, and Western blot analyses. This effort led to the identification of 128 genes of which the expression was altered ≥2-fold in antiangiogenic antithrombin–treated cells relative to native antithrombin–treated and nontreated cells as controls. Based on their functional similarities, 60% of the altered genes were clustered into five major groups of genes of which the activities are well-known regulators of angiogenesis. Our results show that antithrombin exerts its antiangiogenic effects in cultured human endothelial cells at least in part by down-regulating the cluster of genes expressing (i) proangiogenic heparan sulfate and other proteoglycans (e.g., perlecan, syndecan-1, syndecan-3, biglycan, and proteoglycan 4), which serve as coreceptors for growth factors [basic fibroblast growth factor/vascular endothelial growth factor (bFGF/VEGF)] and other cytokines, and (ii) mitogenic signaling molecules [e.g., early growth response factor 1, mitogen-activated protein (MAP) kinases, and signal transducers and activators of transcription (STAT) family proteins]. In addition, antiangiogenic antithrombin also induced the expression of other genes [e.g., adenomatosis polyposis coli (APC), caspase-3, TIMP family proteins, and p21] of which the activities are associated with tumor suppression, induction of apoptotic death, inhibition of tumor cell migration, and cell cycle arrest. Overall, our results provide an overview of an antiangiogenic signaling network induced in cultured primary endothelial cells by antithrombin, which contributes to the antiangiogenic activity of this serpin family protein.

Cell culture and treatment. Fresh human umbilical cords were obtained from normal-term deliveries at Stanford University Hospital. Primary endothelial cells were isolated from umbilical cords as previously described (11) with modifications (10). Seventy-five percent confluent passage-2 human umbilical vein endothelial cells (HUVEC) in Falcon T12.5 flasks were treated with native and cleaved forms of human plasma antithrombin (30 μg/mL) in the presence or absence of bFGF (10 ng/mL; R&D Systems, Minneapolis, MN) for 24 hours with 0.5% heat-inactivated fetal bovine serum and 1% antibiotics. These cells were used for microarray and real-time RT-PCR experiments. For Northern blot and Western blot experiments, primary HUVECs were obtained from Clonetics (San Diego, CA) and cells before passage 10 were treated with antithrombin as indicated for the microarray experiments.

Native and cleaved forms of antithrombin. Human antithrombin was purified from outdated plasma as described (12). The reactive loop cleaved form of antithrombin was obtained by treatment with human neutrophil elastase (Athens Research and Technology, Athens, GA) followed by removal of the elastase by heparin-agarose chromatography as described (10).

RNA preparation and cDNA microarray analysis. RNA isolation from treated and untreated endothelial cells and cDNA microarray analysis were done essentially as previously described (10). The only modification was that genes that were at least 2.0-fold induced or 2.0-fold repressed in cleaved antithrombin–treated cells compared with native antithrombin–treated cells in more than one of three microarray chip replicates were considered to be significantly altered in their expression. After hybridization with Cy3-labeled test cDNA and Cy5-labeled reference cDNA, the fluorescence signals from each gene array chip were scanned. The data were retrieved as Cy3/Cy5 fluorescence intensity ratios after normalization by setting the average log fluorescence ratio for all array elements on the chip equal to zero using the Stanford Microarray Database software (13). For UniGene clusters represented by multiple arrayed elements, the mean fluorescence ratios (for all elements representing the same UniGene cluster) were reported. Genes of which the expression was significantly altered in cleaved antithrombin–treated cells relative to native antithrombin–treated cells based on the selection criterion were filtered and the normalized fluorescence ratios observed for these genes in multiple chips were averaged for antithrombin-treated and untreated cell conditions. Genes were manually grouped by their functions into gene clusters using the human gene database.4

Quantitative RT-PCR analysis. Contaminating genomic DNA was removed from isolated RNA by DNase I treatment (Ambion, Austin, TX). First-strand cDNA was reverse transcribed from total RNA using the SuperScript First Strand Synthesis System in the RT-PCR Kit (Invitrogen, Carlsbad, CA) and was used at a final assay concentration of 0.1 ng/μL (in a 25 μL reaction volume). Real-time quantitation of mRNA was done in triplicate using a QuantiTect SYBR Green PCR kit and the ABI Prism 7000HT Sequence Detection System (Applied Biosystems, Foster City, CA) with 95% efficiency. In addition to profiling the mRNA target sequences in cleaved antithrombin–treated, native antithrombin–treated, and untreated control endothelial cells, mRNAs of β-actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were also profiled as controls. Forward and reverse primers (Supplementary Table S1) were designed based on previous publications or by using ABIs Primer Express software. For each single-well amplification reaction, a threshold cycle (Ct) was observed in the exponential phase of amplification and the quantitation of the test mRNA expression level was done relative to GAPDH mRNA levels measured in a separate reaction by determining the difference in threshold cycles between the test and GAPDH mRNAs (ΔCt = Cttest − CtGAPDH). The relative mRNA level was then calculated from the formula (1.95)ΔCt, where the factor 1.95 reflects the 95% amplification efficiency during each PCR cycle. Expression levels in antithrombin-treated cells were normalized to the levels measured in untreated cells.

Slot Northern blot analysis. cDNA clones encoding the various genes which were analyzed by slot Northern blots were purchased from American Type Culture Collection (Manassas, VA) or, in the case of p21WAF1/Cip1, provided by Dr. Guy Adami (University of Illinois at Chicago, Chicago, IL). DIG-labeled probes for each selected gene in Northern-blot analyses were synthesized by employing a DIG-Primer Labeling PCR kit from Roche Applied Science (Indianapolis, IN) along with the primers listed in Supplementary Table S1 and the cDNA clones as template. Total RNA (2 μg) prepared from antithrombin-treated or control HUVECs after 24 hours was applied to a nylon membrane and hybridization and washing were done under high-stringency conditions according to the protocol of the manufacturer for the DIG-labeled slot Northern blot analysis system (Roche Applied Science). GAPDH mRNA levels were determined simultaneously in each sample as an internal loading control.

Western blot analysis. Preparation of endothelial cell lysates and Western blot analyses were done using procedures previously described (7, 10). Lysates from treated or untreated endothelial cells (50 μg total protein) were electrophoresed in a 4% to 12% gradient SDS/polyacrylamide gel and transferred to an Immobilon P membrane. After blocking nonspecific binding sites, blots were incubated with appropriate antibodies [all from Santa Cruz Biotechnology (Santa Cruz, CA) except for perlecan (Zymed Laboratories, San Francisco, CA) and β-actin (Sigma, St. Louis, MO)]. Blots were then exposed to horseradish peroxidase–conjugated secondary antibodies and visualized by the enhanced chemiluminescence system from Amersham (Piscataway, NJ).

Morphologic detection of cellular apoptosis. Sixty to seventy percent confluent HUVECs in 12-well plates were incubated for 24 hours with 30 μg/mL of antithrombin in 0.5% bovine calf serum-F-12K (Invitrogen). Cells were harvested and resuspended in PBS containing 5% glycerol and 0.1 mol/L NaCl. The cells were dried onto slides, fixed with acetone/methanol (1:1), and the cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; 500 ng/mL). Apoptotic cells appearing as fractured nuclei were counted in random fields by fluorescence microscopy (×200 magnification, at least 10 fields per sample) and photographed.

Genome-wide expression profiling. To identify target genes of which the expression is altered by antiangiogenic antithrombin treatment of endothelial cells, we profiled the expression of genes in primary HUVECs after 24-hour treatment with 30 μg/mL of native antithrombin, cleaved antithrombin, or buffer alone using cDNA-based microarray analysis. Expression levels in cells treated with antiangiogenically active cleaved antithrombin which differed from the expression levels in cells treated with the antiangiogenically inactive native form of antithrombin by a factor ≥2.0 in at least two of three replicate microarray chips were considered to be up-regulated, whereas average ratios of ≤0.5 were considered to be down-regulated. We found a total of 128 genes of which the expression was altered based on this criterion. For most of these genes (∼85%), native antithrombin treatment of cells insignificantly altered expression relative to untreated control cells (<1.5-fold), indicating that the majority of affected genes were only responsive to the antiangiogenically active form of antithrombin and not to the inactive form. The majority of the genes affected by cleaved antithrombin are represented by five groups of gene clusters of which the activities are clearly implicated in angiogenesis control (Fig. 1). The expression level of each selected gene relative to a universal mRNA control is indicated for the three treatments by a gradient of color from red (low expression) to green (high expression). Interestingly, most of the up-regulated genes are known to play a negative role in the regulation of angiogenesis. For example, vitronectin has been shown to be a necessary matrix molecule for mediating antithrombin antiangiogenic function (14). p21WAF1/Cip1 is well known for its cell cycle arrest activity (15) and the up-regulation of this gene is in keeping with our previous finding that antiangiogenic antithrombin inhibits the G1-to-S phase cell cycle transition in bFGF-stimulated HUVECs (10). The APC protein and tissue inhibitor of metalloproteinase (TIMP) family proteins (TIMP-1, TIMP-2, and TIMP-3) are known inhibitors of tumor metastasis (16, 17) and caspase-3 is an important mediator of cell apoptosis (18). The majority of genes with altered expression were down-regulated, with 93 of the total 128 affected genes showing a decreased expression of 2- to 4-fold in the cleaved antithrombin–treated cells versus the native antithrombin–treated ones. Most of those genes (>80%) are in gene clusters of which the functions are involved in cell adhesion, communication, migration, mitogenic signaling, and cell cycle control. Among those down-regulated genes, the group of genes encoding cell-surface and matrix proteoglycans and mitogenic signaling proteins particularly stand out as they represent primarily proangiogenic genes. The former group include the HSPG family members, perlecan (HSPG2; refs. 1921), syndecan-1 (22), and syndecan-3 (23), as well as other proteoglycans such as biglycan and proteoglycan 4. The latter group of mitogenic signaling molecules include early growth response factor-1 (EGR1; ref. 24), immediate early response protein-3 (IER3/IEX-1; ref. 25), fibroblast growth factor-20 (26), signal transducer and activator of transcription (STAT) family proteins 2, 3, and 6 (27), and MAP kinase 3 (28). Supplementary Table S2 identifies the 128 genes of which the expression was found to be altered by treatment of cultured HUVECs with antiangiogenic antithrombin compared with native antithrombin or control treatments of cells based on our selection criterion. Average relative expression changes <2-fold for some genes in Fig. 1 and Supplementary Table S2 reflect the fact that the differential expression in one of the three chips did not meet the 2-fold criterion.

Figure 1.

Differential expression of five groups of angiogenesis-related genes in cleaved and native antithrombin–treated HUVECs. Five groups of angiogenesis-related genes, of which the expression in cleaved antithrombin–treated HUVECs was minimally 2-fold altered from the expression in native antithrombin–treated cells in at least two of three replicate microarray chips, are tabulated from Supplementary Table S2. The expression levels of each gene in untreated, native antithrombin–, and cleaved antithrombin–treated cells relative to a universal mRNA control are depicted for each gene by the intensity of color ranging from red (low expression) to green (high expression) as indicated by the color key.

Figure 1.

Differential expression of five groups of angiogenesis-related genes in cleaved and native antithrombin–treated HUVECs. Five groups of angiogenesis-related genes, of which the expression in cleaved antithrombin–treated HUVECs was minimally 2-fold altered from the expression in native antithrombin–treated cells in at least two of three replicate microarray chips, are tabulated from Supplementary Table S2. The expression levels of each gene in untreated, native antithrombin–, and cleaved antithrombin–treated cells relative to a universal mRNA control are depicted for each gene by the intensity of color ranging from red (low expression) to green (high expression) as indicated by the color key.

Close modal

Validation of microarray results by Northern blot and Western blot analyses. For an independent confirmation of the microarray results, we did slot Northern blot analyses of the RNA isolated from cultured HUVECs untreated or treated with native and cleaved antithrombins as in the microarray analyses. This was done using DIG-labeled cDNA probes specifically designed for certain selected genes of which the expression was differentially regulated by cleaved and native antithrombins and of which the functions are clearly involved in the regulation of angiogenesis (Fig. 2). The mRNA levels of eight genes found to be up-regulated by microarray analysis of HUVECs treated with cleaved versus native antithrombin after 24 hours were also significantly elevated by slot Northern blot analysis of these genes. Moreover, the mRNA levels of nine genes shown by microarray analysis to be down-regulated in cleaved antithrombin–treated versus native antithrombin–treated cells were also depressed in the slot Northern blots. To determine whether the protein products of the genes observed to be differentially expressed in cleaved antithrombin–treated versus native antithrombin–treated HUVECs showed a parallel alteration in expression level, we conducted Western blot analyses of the protein products of the affected HUVEC genes in the microarray and Northern blot analyses. The protein levels of p21 WAF1/Cip1, TIMP-1, caspase-3, vitronectin, and apolipoprotein A-1 were significantly elevated whereas the levels of STAT-3, syndecan-1, EGR1, biglycan, perlecan, and MAP kinase 3 were all significantly decreased in cleaved antithrombin–treated cells relative to native antithrombin– or control-treated cells, in agreement with the microarray and Northern blot analyses (Fig. 3). However, the protein levels for TIMP-3, APC, and syndecan-3 were not detectably changed among cells undergoing cleaved antithrombin versus native antithrombin or control treatments. As an internal control, β-actin protein levels remained constant over the three treatment conditions. The final protein products of TIMP-3, APC, and syndecan-3 may thus be under additional levels of regulatory control because the mRNA levels of these genes clearly differed in cleaved antithrombin–treated versus native antithrombin–treated cells.

Figure 2.

Slot Northern blot analysis of differentially expressed genes in cleaved antithrombin–treated versus native antithrombin–treated HUVECs. Total RNA from HUVECs after 24-hour treatment with buffer (CT), native antithrombin (ATN), or cleaved antithrombin (ATC) was subjected to slot Northern-blot analysis using individual labeled probes specific for the indicated gene sequences. GAPDH expression was used as an internal loading control.

Figure 2.

Slot Northern blot analysis of differentially expressed genes in cleaved antithrombin–treated versus native antithrombin–treated HUVECs. Total RNA from HUVECs after 24-hour treatment with buffer (CT), native antithrombin (ATN), or cleaved antithrombin (ATC) was subjected to slot Northern-blot analysis using individual labeled probes specific for the indicated gene sequences. GAPDH expression was used as an internal loading control.

Close modal
Figure 3.

Western blot analysis of differentially expressed genes in cleaved antithrombin–treated versus native antithrombin–treated HUVECs. Fifty micrograms of whole lysate proteins from treated or control HUVECs were subjected to 4% to 12% gradient SDS-PAGE, transferred to Immobilon P membranes, and individual proteins were detected using specific antibodies. β-Actin protein levels were determined as an internal loading control.

Figure 3.

Western blot analysis of differentially expressed genes in cleaved antithrombin–treated versus native antithrombin–treated HUVECs. Fifty micrograms of whole lysate proteins from treated or control HUVECs were subjected to 4% to 12% gradient SDS-PAGE, transferred to Immobilon P membranes, and individual proteins were detected using specific antibodies. β-Actin protein levels were determined as an internal loading control.

Close modal

Validation of microarray results by quantitative real-time RT-PCR. Because of potential variations in the dynamic expression patterns of endothelial cell genes affected by antiangiogenic antithrombin, we wished to verify whether the 24-hour treatment of HUVECs with antithrombin was appropriate to reveal the maximal changes in gene expression. Eleven genes with established roles in the regulation of angiogenesis were therefore chosen for real-time quantitative RT-PCR analysis. Gene-specific primers (Supplementary Table S1) were used to measure the mRNA levels of these genes at varying times of treatment of HUVECs with native or cleaved antithrombins (0.5, 2, 6, 12, 24, and 48 hours). mRNA expression levels over this time course were determined relative to GAPDH and β-actin controls. The results clearly show substantial differential alterations of the mRNA levels of the 11 genes in a time-dependent manner for cleaved antithrombin–treated versus native antithrombin–treated HUVECs, with the maximal changes of gene expression occurring for most genes at 24 hours (Fig. 4). Exceptions were p21WAF1/Cip1 of which the maximal induction by cleaved antithrombin occurred relatively early, ∼6 hours, following treatment and vitronectin, which was induced relatively late, its expression peaking at 48 hours. Interestingly, syndecan-1 gene expression was dramatically induced in both cleaved antithrombin–treated and native antithrombin–treated HUVECs over the initial 12-hour period and became down-regulated in cleaved antithrombin–treated versus native antithrombin–treated cells only at the 24- and 48-hour time points. Overall, the kinetic analysis of gene expression for the 11 selected genes confirmed the reliability of the HUVEC gene expression patterns revealed by microarray analysis after 24-hour treatment of cells with antithrombin.

Figure 4.

Kinetics of expression of selected genes in antithrombin-treated and untreated HUVECs. Real-time quantitative RT-PCR analysis was done on total RNA samples from HUVECs treated with 30 μg/mL cleaved antithrombin, native antithrombin, or untreated control for 0.5, 2, 6, 12, 24, and 48 hours. Columns, average expression levels in quadruplicate samples from cleaved antithrombin–treated cells (black columns) and from native antithrombin–treated cells (white columns), normalized to the expression levels from control untreated cells (gray columns); bars, SD.

Figure 4.

Kinetics of expression of selected genes in antithrombin-treated and untreated HUVECs. Real-time quantitative RT-PCR analysis was done on total RNA samples from HUVECs treated with 30 μg/mL cleaved antithrombin, native antithrombin, or untreated control for 0.5, 2, 6, 12, 24, and 48 hours. Columns, average expression levels in quadruplicate samples from cleaved antithrombin–treated cells (black columns) and from native antithrombin–treated cells (white columns), normalized to the expression levels from control untreated cells (gray columns); bars, SD.

Close modal

Cleaved antithrombin–induced apoptosis in HUVECs. Previous studies have shown that antiangiogenic forms of antithrombin induce apoptotic cell death in cultured bovine capillary endothelial cells (6) and in tumors implanted in mice (9). Interestingly, our microarray, slot Northern blot, Western blot, and quantitative real-time RT-PCR analyses all showed that caspase-3 gene expression was induced >2-fold in HUVECs treated with cleaved antithrombin versus native antithrombin or control for 24 hours. To further confirm the apoptotic effects of cleaved antithrombin on endothelial cells, we quantitated apoptosis in antithrombin-treated and untreated HUVECs. HUVECs treated with cleaved antithrombin for 24 hours showed >2-fold increase in apoptotic death relative to control-treated cells (Fig. 5A and B). Moreover, cleavage of poly(ADP-ribose) polymerase, a landmark of cell apoptosis (29), was elevated in cleaved antithrombin–treated cells relative to controls (Fig. 5C). Because other apoptosis-related genes may have escaped detection in our microarray analysis, we determined the mRNA levels of four representative genes including those in the Bcl family (i.e., p53, Bcl-2, Max, and Bax; ref. 30) by quantitative real-time RT-PCR analysis. The results (Fig. 5D) show that cleaved antithrombin produced no significant alterations in Bcl-2 and p53 gene transcription relative to controls but induced the expression of the proapoptotic genes, Max and Bax, by 1.5- to 2.5-fold.

Figure 5.

Antiangiogenic antithrombin induces endothelial cell apoptosis. A, primary HUVECs were cultured in serum-reduced medium in the presence or absence of 30 μg/mL of native or cleaved antithrombin for 24 hours. The apoptotic bodies of antithrombin-treated HUVECs were detected by fluorescence microscopy after staining with DAPI. B, columns, mean percentage of apoptotic cells in HUVECs subjected to control, native antithrombin, or cleaved antithrombin treatments, quantified by counting at least four fields of cells; bars, SD. *, P < 0.01. C, cleavage of poly(ADP-ribose) polymerase (PARP) in HUVECs treated for 24 hours as in (A and B) assessed by Western blot analysis. D, real-time quantitative RT-PCR analysis of mRNA expression of apoptosis-related genes in HUVECs treated with buffer (blank columns), native antithrombin (gray columns), and cleaved antithrombin (black columns). Columns, mean of three assays; bars, SD. *, P < 0.05.

Figure 5.

Antiangiogenic antithrombin induces endothelial cell apoptosis. A, primary HUVECs were cultured in serum-reduced medium in the presence or absence of 30 μg/mL of native or cleaved antithrombin for 24 hours. The apoptotic bodies of antithrombin-treated HUVECs were detected by fluorescence microscopy after staining with DAPI. B, columns, mean percentage of apoptotic cells in HUVECs subjected to control, native antithrombin, or cleaved antithrombin treatments, quantified by counting at least four fields of cells; bars, SD. *, P < 0.01. C, cleavage of poly(ADP-ribose) polymerase (PARP) in HUVECs treated for 24 hours as in (A and B) assessed by Western blot analysis. D, real-time quantitative RT-PCR analysis of mRNA expression of apoptosis-related genes in HUVECs treated with buffer (blank columns), native antithrombin (gray columns), and cleaved antithrombin (black columns). Columns, mean of three assays; bars, SD. *, P < 0.05.

Close modal

Effect of bFGF on the gene expression profile of HUVECs treated with cleaved antithrombin. Because previous studies showed the antiangiogenic effects of conformationally altered forms of antithrombin mostly in growth factor (VEGF/bFGF)–stimulated endothelial cells (57, 10), we determined the effects of bFGF on the differential expression of genes in HUVECs treated with cleaved antithrombin versus native antithrombin or buffer using the microarray technique. Based on our 2.0-fold change selection criterion, we found that cleaved antithrombin resulted in an altered expression of 136 genes relative to native antithrombin treatment of cells stimulated with bFGF (data not shown). However, only 15 of these genes were identical to those differentially expressed in unstimulated HUVECs (Table 1). bFGF enhanced or produced a similar differential expression of 10 genes in cleaved antithrombin–treated versus native antithrombin–treated cells including the angiogenesis-related genes, APC, caspase-3, CDCA3, EGR1, HSPG2, and IGFBP1. There were two genes of which the altered expression in cleaved antithrombin–treated versus native antithrombin–treated cells was attenuated in the presence of bFGF. Interestingly, bFGF reversed the effects of cleaved antithrombin on the expression of three genes which were mostly not angiogenesis related. These results indicate that bFGF modulates the effects of antiangiogenic antithrombin on endothelial cell gene expression and results in more pronounced effects on several angiogenesis-related genes.

Table 1.

Effects of bFGF on the changes in HUVEC gene expression induced by cleaved antithrombin

Gene symbolCluster IDFold change
−bFGF+bFGF
Enhanced or unchanged    
    APC Hs.158932 2.3 3.1 
    APOA1 Hs.93194 2.0 5.7 
    CASP3 Hs.141125 2.5 3.3 
    CDCA3 Hs.524216 0.49 0.25 
    CYP111B1 Hs.184927 0.24 0.24 
    EGR1 Hs.326035 0.41 0.23 
    GPR143 Hs.74124 0.48 0.29 
    HSPG2 Hs.555874 0.44 0.24 
    IGFBP1 Hs.401316 0.59 0.22 
    NYD-SP18 Hs.131098 0.43 0.20 
Attenuated    
    FCER1G Hs.433300 0.30 0.46 
    NIN Hs.310429 0.45 0.50 
Reversed    
    HIPK2 Hs.397465 0.25 6.62 
    IGLL1 Hs.348935 0.28 4.18 
    PCSK2 Hs.315186 2.17 0.15 
Gene symbolCluster IDFold change
−bFGF+bFGF
Enhanced or unchanged    
    APC Hs.158932 2.3 3.1 
    APOA1 Hs.93194 2.0 5.7 
    CASP3 Hs.141125 2.5 3.3 
    CDCA3 Hs.524216 0.49 0.25 
    CYP111B1 Hs.184927 0.24 0.24 
    EGR1 Hs.326035 0.41 0.23 
    GPR143 Hs.74124 0.48 0.29 
    HSPG2 Hs.555874 0.44 0.24 
    IGFBP1 Hs.401316 0.59 0.22 
    NYD-SP18 Hs.131098 0.43 0.20 
Attenuated    
    FCER1G Hs.433300 0.30 0.46 
    NIN Hs.310429 0.45 0.50 
Reversed    
    HIPK2 Hs.397465 0.25 6.62 
    IGLL1 Hs.348935 0.28 4.18 
    PCSK2 Hs.315186 2.17 0.15 

The present study was designed to identify target genes affected by treatment of primary HUVECs with antiangiogenic antithrombin and thereby to reveal how endothelial cell gene expression is reprogrammed to orchestrate the antiangiogenic response. Our previous report showed that antiangiogenic cleaved and latent forms of antithrombin significantly suppressed the expression of the proangiogenic HSPG, perlecan, in HUVECs independent of whether the cells were stimulated with growth factor (10). This clearly showed that the effects of antiangiogenic antithrombin were not simply mediated by blocking growth factor effects on endothelial cells but involved changes in endothelial cell gene expression presumably requiring signaling through a receptor. Our current study more firmly establishes that the antiangiogenic cleaved form of antithrombin significantly alters the expression not only of the perlecan gene but also of numerous other genes in endothelial cells, strongly implicating an endothelial cell ligand-receptor signaling mechanism in mediating these global changes in gene expression. This idea is supported by our finding that the majority of changes in gene expression induced by cleaved antithrombin involved the down-regulation of proangiogenic genes and the up-regulation of antiangiogenic genes, in keeping with the antiangiogenic activity of the protein. Our selection criterion for affected genes required that minimally 2-fold expression changes in at least two microarray chips be observed in cleaved antithrombin–treated cells relative to cells treated with native antithrombin. Native antithrombin was considered the best reference state because this form of the protein lacks antiangiogenic activity and endothelial cells are exposed to ∼100 μg/mL levels of the protein in circulating blood under normal physiologic conditions (4). For those genes of which the expression was significantly altered by cleaved antithrombin treatment of HUVECs, native antithrombin treatment typically showed marginal changes from untreated cells, consistent with the effects being specific for the conformationally altered form of the serpin. This sets antithrombin apart from other antiangiogenic serpins of which the antiangiogenic activity is not conformation dependent. Selected genes of which the expression was altered in the microarray experiments by cleaved antithrombin treatment and which have established roles in angiogenesis were confirmed to have altered expression by Northern blot and real-time RT-PCR analyses and, in most cases, also altered protein levels by Western blot analyses. Cleaved antithrombin thus seems to induce an antiangiogenic signaling network in endothelial cells resembling in its overall effect that induced by the angiogenesis inhibitor, endostatin (31), although the specific genes affected differ for the two proteins. Although the effects of the antiangiogenic latent form of antithrombin on endothelial cell gene expression were not tested in the present study, it should be noted that cleaved and latent antithrombin forms have previously been shown to display indistinguishable antiangiogenic activities in proliferation, migration, capillary-like tube formation, bFGF signaling, and perlecan gene expression assays of angiogenesis (6, 7, 10).

Our results show that several genes clustered in the cell matrix protein group, including perlecan (HSPG2), syndecan-1, syndecan-3, biglycan, proteoglycan-4, and α5-laminin, were all down-regulated by >2-fold. It has been clear that most of the molecular events associated with tumor growth, neovascularization, and metastasis are influenced by interactions between cancer cells and their extracellular matrix components (32). HSPGs such as perlecan, syndecan-1, and syndecan-3 can act both as reservoirs for growth factors and as coreceptors for ligand binding and subsequent intracellular signaling (33). Therefore, reductions of these HSPG molecules on endothelial cells have been considered as an efficient approach to control cell growth, invasion, angiogenesis, and tumor progression (34). Intriguingly, the matrix HSPG, agrin, was found to be up-regulated by antiangiogenic antithrombin, suggesting that this HSPG may not be proangiogenic. The α5-laminins are prominent basement membrane components which promote bFGF- and VEGF-induced endothelial cell growth via ligation of αvβ3 integrins (35). An indirect inhibitory effect on the expression of other endothelial cell adhesion molecules is also expected based on our observation that the IκBε inhibitor of NF-κB, which controls NF-κB-dependent expression of adhesion molecules, is up-regulated by antiangiogenic antithrombin (Supplementary Table S2; ref. 36). Interestingly, our microarray results show that certain matrix molecules such as vitronectin were up-regulated >2-fold in cleaved antithrombin–treated versus native antithrombin–treated cells. A previous report has shown that vitronectin is essential for cleaved antithrombin to express its antiangiogenic function in vivo (14). It has been suggested that vitronectin forms a complex with antiangiogenic antithrombin either directly or indirectly through the binding of both proteins to heparin, which then exerts an antiangiogenic effect by blocking endothelial cell integrins required for growth. Integrin blockade underlies the mechanism of antiangiogenic action of several other natural angiogenesis inhibitors (37). Together, the alterations in the expression of endothelial cell matrix molecules induced by antiangiogenic antithrombin may contribute to the reported attenuating effects of latent antithrombin on focal adhesion kinase activation (6).

Antiangiogenic forms of antithrombin have been shown to induce endothelial cell apoptotic death in vitro (6) and in vivo (9), consistent with our current microarray data that the expression of caspase-3 and several other proapoptotic molecules is significantly induced in cleaved antithrombin–treated cells. Furthermore, our quantitative real-time RT-PCR analysis also indicated that the ratio of Bcl-2/Bax in endothelial cells treated with cleaved antithrombin was decreased, which is a well-known indicator of apoptosis (30). Previous studies have shown that numerous angiogenesis inhibitors, including several serpins, exert their antiagiogenic effects partly by inducing endothelial cell apoptosis (3841).

We previously showed that antiangiogenic forms of antithrombin inhibit growth factor–stimulated cell proliferation by arresting the cell cycle at the G1-S phase (10). The present study shows that the antiangiogenic cleaved form of antithrombin acts directly to inhibit cell growth by down-regulating cell cycle regulatory proteins which promote growth and by up-regulating proteins which inhibit growth (42). The antiangiogenic activity of endostatin is similarly mediated by altering the expression of cell cycle control proteins in endothelial cells to favor a quiescent state (31).

The cytokine-induced oligomerization of receptor subunits and engagement of an intracellular signaling machinery is a common feature by which cytokines such as bFGF and VEGF induce mitogenic effects in target cells (43). Cytokine growth factors may signal through several parallel MAP kinase pathways to induce cytokine-specific gene transcription patterns or through receptor-associated Janus kinases, which phosphorylate STAT family transcription factors, to induce gene transcription (44, 45). Compared with normal cells and tissues, constitutively activated MAP kinases and/or STATs have been detected in a wide variety of human cancer cell lines and primary tumors. It is clear that MAP kinases and STATs, particularly STAT-3, play an important role in growth factor–mediated angiogenesis (27, 43, 46). Our microarray results, partially confirmed by real-time quantitative RT-PCR, Northern-blot, and/or Western blot analysis, show that 26 mitogenesis-related signaling molecule genes were significantly down-regulated in endothelial cells following cleaved antithrombin treatment. Among these down-regulated genes, the transcriptional regulators EGR1 and IER3/IEX-1 are downstream targets of extracellular signal–regulated protein kinase 1/2, the protein kinase that is activated by mitogens like bFGF/VEGF (25, 47). The biological function of EGR1 has been closely connected with the development of human cancers because overexpression of EGR1 in a majority of human prostate cancers has been observed (48). Moreover, tumor progression in transgenic mouse models of prostate cancer was reported to be significantly impaired when EGR1 was lacking (49). The down-regulation of EGR1 gene expression in antiangiogenic antithrombin–treated cells was fully confirmed at the mRNA and protein levels and this down-regulation was shown to be further enhanced by bFGF stimulation of HUVECs (Table 1). Interestingly, EGR1 gene expression is significantly suppressed in pigment epithelium–derived factor–treated cells (50), another potent antiangiogenic serpin. Two down-regulated mitogenic transcription factors of the ets family, SPDEF and ELK1, are downstream targets of MAP kinase pathways and found to be up-regulated in cancer (51, 52). Together, these results suggest the importance of down-regulating the MAP kinase– and STAT-dependent signaling pathways to mediate the antiangiogenic activity of antithrombin.

The matrix metalloproteinases (MMP) play an important role in tumor invasion by functioning to degrade the macromolecules of the extracellular matrix. Abundant evidence indicates that controlling the activity of MMPs by maintaining their zymogen state (pro-MMPs) or inactivating the active MMPs is an effective approach to control angiogenesis and tumor metastasis (53). The naturally occurring inhibitors, TIMPs, are important regulators of the activities of MMPs in normal and disease processes (17). Thus far, four distinct TIMPs (TIMP-1, TIMP-2, TIMP-3, and TIMP-4) have been isolated, cloned, and characterized in several species. Interestingly, cleaved antithrombin was found to induce the expression of three TIMPs >2-fold in HUVECs and the induction of TIMP-1 and TIMP-3 was confirmed at the mRNA level and, in the case of TIMP-1, also at the protein level. Because the activities of MMPs are strictly regulated by TIMPs, these results may at least partly explain the previous in vivo observations that antiangiogenic forms of antithrombin induce significant tumor regression in mice (5, 6, 8, 9). The control of tumor metastasis in certain cancers by inducing endogenous TIMP expression or delivering exogenous TIMPs has become a promising cancer therapy approach (54). Our data thus support the potential efficacy of antiangiogenic antithrombin as an antitumor drug.

Antiangiogenic forms of antithrombin significantly inhibit growth factor–stimulated proangiogenic activities including cell proliferation, migration, capillary tube formation, and growth factor–induced signaling (57, 10). Whereas a comparable number of genes were found to be differentially expressed in bFGF-stimulated HUVECs treated with cleaved antithrombin versus native antithrombin, only ∼10% of these genes were identical to those differentially expressed in unstimulated HUVECs. Several of the angiogenesis-related genes, including APC, caspase-3, perlecan, and EGR-1, showed an even greater differential expression in cleaved antithrombin–treated versus native antithrombin–treated cells stimulated with bFGF than cells not stimulated. The remaining ∼90% of genes with significantly altered expression in unstimulated, but not stimulated, cells may have fallen below the threshold of our selection criterion for altered expression because of the up-regulation of a wide range of endothelial cell genes in control bFGF-stimulated cells. The effects of endostatin on endothelial cell gene expression also were determined in the absence of growth factor stimulation presumably to more clearly reveal its antiangiogenic signaling effects (31). That the differential effects of cleaved antithrombin versus native antithrombin on gene expression observed in unstimulated cells are likely to be important also in stimulated cells is suggested by our recent finding that antiangiogenic forms of antithrombin inhibit bFGF-induced proangiogenic effects by competing with bFGF for binding a common heparan sulfate coreceptor.5

5

W. Zhang, R. Swanson, Y. Xiong, S.T. Olson, submitted for publication.

The observed effects of cleaved antithrombin on gene expression in bFGF-stimulated HUVECs are thus likely to reflect a balance between reversing the effects of growth factor and the direct effects mediated through an HSPG receptor or coreceptor.

In summary, our microarray results show that the antiangiogenic cleaved form of antithrombin exerts its antiangiogenic function by globally altering gene expression in cultured HUVECs, mainly by suppressing proangiogenic genes, including cell-surface HSPGs and MAP kinase and STAT signaling molecules, and by inducing antiangiogenic genes, including tumor suppressors, proapoptotic, and cell cycle control proteins. The genetic reprogramming of endothelial cells by antiangiogenic antithrombin revealed in this study suggests that certain target genes are involved in mediating the antiangiogenic response. Further studies will be required to decipher how the protein products of these genes accomplish this function.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

Current address for Y-J. Chuang: Institute of Bioinformatics and Structural Biology, National Tsing Hua University, Taiwan, Republic of China.

Grant support: NIH grant HL-39888 (S.T. Olson).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Peter Gettins (University of Illinois at Chicago) for helpful comments on the manuscript.

1
Carmeliet P. Angiogenesis in health and disease.
Nat Med
2003
;
9
:
653
–60.
2
Gastl G, Hermann T, Steurer M, et al. Angiogenesis as a target for tumor treatment.
Oncology
1997
;
54
:
177
–84.
3
Tandle A, Blazer DG III, Libutti SK. Antiangiogenic gene therapy of cancer: recent developments.
J Transl Med
2004
;
2
:
22
–41.
4
Björk I, Olson ST. Antithrombin, a bloody important serpin. In: Church FC, Cunningham DD, Ginsburg D, Hoffman M, Stone SR, Tollefsen DM, editors. Chemistry and biology of serpins. New York: Plenum Press; 1997. p. 17–33.
5
O'Reilly MS, Pirie-Shepherd S, Lane WS, Folkman J. Antiangiogenic activity of the cleaved conformation of the serpin antithrombin.
Science
1999
;
285
:
1926
–8.
6
Larsson H, Sjoblom T, Dixelius J, et al. Antiangiogenic effects of latent antithrombin through perturbed cell-matrix interactions and apoptosis of endothelial cells.
Cancer Res
2000
;
60
:
6723
–9.
7
Zhang W, Swanson R, Izaguirre G, Xiong Y, Lau LF, Olson ST. The heparin-binding site of antithrombin is crucial for antiangiogenic activity.
Blood
2005
;
106
:
1621
–8.
8
Kisker O, Onizuka S, Banyard J, et al. Generation of multiple angiogenesis inhibitors by human pancreatic cancer.
Cancer Res
2001
;
61
:
7298
–304.
9
Prox D, Becker C, Pirie-Shepherd SR, Celik I, Folkman J, Kisker O. Treatment of human pancreatic cancer in mice with angiogenic inhibitors.
World J Surg
2003
;
68
:
47
–54.
10
Zhang W, Chuang Y-J, Swanson R, et al. Antiangiogenic antithrombin down-regulates the expression of the proangiogenic heparan sulfate proteoglycan, perlecan, in endothelial cells.
Blood
2004
;
103
:
1185
–91.
11
Jaffe EA, Nachman RL, Becker CG, Minick CR. Culture of human endothelial cells derived from umbilical veins: identification by morphologic and immunologic criteria.
J Clin Invest
1973
;
52
:
2745
–56.
12
Olson ST, Björk I, Shore JD. Kinetic characterization of heparin-catalyzed and uncatalyzed inhibition of blood coagulation proteinases by antithrombin.
Methods Enzymol
1993
;
222
:
525
–60.
13
Sherlock G, Hernandez-Boussard T, Kasarskis A, et al. The Stanford Microarray Database.
Nucleic Acids Res
2001
;
29
:
152
–5.
14
Yi M, Sakai T, Fässler R, Ruoslahti E. Antiangiogenic proteins require plasma fibronectin or vitronectin for in vivo activity.
Proc Natl Acad Sci U S A
2003
;
100
:
11435
–8.
15
Bokoch GM. Biology of the p21-activated kinases.
Annu Rev Biochem
2003
;
72
:
743
–81.
16
Fodde R, Smits R, Clevers H. APC, signal transduction and genetic instability in colorectal cancer.
Nat Rev
2001
;
1
:
55
–67.
17
Brew K, Dinakarpandian D, Nagase H. Tissue inhibitors of metalloproteinases: evolution, structure and function.
Biochim Biophys Acta
2000
;
1477
:
267
–83.
18
Denault JB, Salvesen G. Caspases: keys in the ignition of cell death.
Chem Rev
2002
;
102
:
4489
–99.
19
Aviezer D, Hecht D, Safran M, Eislinger M, David G, Yayon A. Perlecan, basal lamina proteoglycan, promotes basic fibroblast growth factor-receptor binding, mitogenesis and angiogenesis.
Cell
1994
;
79
:
1005
–13.
20
Aviezer D, Iozzo RV, Noonan DM, Yayon A. Suppression of autocrine and paracrine functions of basic fibroblast growth factor by stable expression of perlecan antisense cDNA.
Mol Cell Biol
1997
;
17
:
1938
–46.
21
Sharma B, Handler M, Eichstetter I, Whitelock JM, Nugent MA, Iozzo RV. Antisense targeting of perlecan blocks tumor growth and angiogenesis in vivo.
J Clin Invest
1998
;
102
:
1599
–608.
22
Filla MS, Dam P, Rapraeger AC. The cell surface proteoglycan syndecan-1 mediates fibroblast growth factor-2 binding and activity.
J Cell Physiol
1998
;
174
:
310
–21.
23
Chernousov MA, Carey DJ. N-syndecan (syndecan 3) from neonatal rat brain binds basic fibroblast growth factor.
J Biol Chem
1993
;
268
:
16810
–4.
24
Fahmy RG, Dass CR, Sun L-Q, Chesterman CN, Khachigian LM. Transcription factor Egr-1 supports FGF-dependent angiogenesis during neovascularization and tumor growth.
Nat Med
2003
;
9
:
1026
–32.
25
Garcia J, Ye Y, Arranz V, Letourneux C, Pezeron G, Porteu F. IEX-1: a new ERK substrate involved in both ERK survival activity and ERK activation.
EMBO J
2002
;
21
:
5151
–63.
26
Chamorro MN, Schwartz DR, Vonica A, Brivanlou AH, Cho KR, Varmus HE. FGF-20 and DKK1 are transcriptional targets of β-catenin and FGF-20 is implicated in cancer and development.
EMBO J
2005
;
24
:
73
–84.
27
Bartoli M, Gu X, Tsai NT, et al. Vascular endothelial growth factor activates STAT proteins in aortic endothelial cells.
J Biol Chem
2000
;
275
:
33189
–92.
28
Whitmarsh AJ, Davis RJ. Transcription factor AP-1 regulation by mitogen-activated protein kinase signal transduction pathways.
J Mol Med
1996
;
74
:
589
–607.
29
Scovassi AI, Diederich M. Modulation of poly(ADP-robosylation) in apoptotic cells.
Biochem Pharmacol
2004
;
68
:
1041
–7.
30
Antonsson B, Martinou J-C. The Bcl-2 protein family.
Exp Cell Res
2000
;
256
:
50
–7.
31
Abdollahi A, Hahnfeldt P, Maercker C, et al. Endostatin's antiangiogenic signaling network.
Mol Cell
2004
;
13
:
649
–63.
32
Boudreau N, Bissell MJ. Extracellular matrix signaling: integration of form and function in normal and malignant cells.
Curr Opin Cell Biol
1998
;
10
:
640
–6.
33
Iozzo RV, SanAntonio JD. Heparan sulfate proteoglycans: heavy hitters in the angiogenesis arena.
J Clin Invest
2001
;
108
:
349
–55.
34
Sanderson RD. Heparan sulfate proteoglycans.
Semin Cell Dev Biol
2001
;
12
:
89
–98.
35
Genersch E, Ferletta M, Virtanen I, Haller H, Ekblom P. Integrin αvβ3 binding to human α5-laminins facilitates FGF-2 and VEGF-induced proliferation of human ECV304 carcinoma cells.
Eur J Cell Biol
2003
;
82
:
105
–17.
36
Spiecker M, Darius H, Liao JK. A functional role of IkB-α in endothelial cell activation.
J Immunol
2000
;
164
:
3316
–22.
37
Bix G, Iozzo RV. Matrix revolutions: “tails” of basement-membrane components with angiostatic functions.
Trends Cell Biol
2005
;
15
:
52
–60.
38
Huang H, Campbell SC, Nelius T, et al. α1-Antitrypsin inhibits angiogenesis and tumor growth.
Int J Cancer
2004
;
112
:
1042
–8.
39
Volpert OV, Zaichuk T, Zhou W, et al. Inducer-stimulated Fas targets activated endothelium for destruction by anti-angiogenic thrombospondin-1 and pigment epithelium-derived factor.
Nat Med
2002
;
8
:
349
–57.
40
Brakenhielm E, Veitonmaki N, Cao R, et al. Adiponectin-induced antiangiogenesis and antitumor activity involve caspase-mediated endothelial cell apoptosis.
Proc Natl Acad Sci U S A
2004
;
101
:
2476
–81.
41
Veitonmaki N, Cao R, Wu LH, et al. Endothelial cell surface ATP synthase-triggered caspase-apoptotic pathway is essential for K1-5-induced antiangiogenesis.
Cancer Res
2004
;
64
:
3679
–86.
42
Doree M, Galas S. The cyclin-dependent protein kinases and the control of cell division.
FASEB J
1994
;
8
:
1114
–21.
43
Klint P, Claesson-Welsh L. Signal transduction by fibroblast growth receptors.
Front Biosci
2000
;
275
:
24653
–60.
44
Seger R, Krebs EG. The MAPK signaling cascade.
FASEB J
1995
;
9
:
726
–35.
45
Karnitz LM, Abraham RT. Cytokine receptor signaling mechanisms.
Curr Opin Immunol
1995
;
7
:
320
–6.
46
Wei D, Lem X, Zheng L, et al. STAT3 activation regulates the expression of vascular endothelial growth factor and human pancreatic cancer angiogenesis and metastasis.
Oncogene
2003
;
22
:
319
–29.
47
Kaufmann K, Bach K, Thiel G. Extracellular signal-regulated protein kinases Erk1/Erk2 stimulate expression and biological activity of the transcriptional regulator EGR-1.
Biol Chem
2001
;
382
:
1077
–81.
48
Eid MA, Kumar MV, Iczkowski KA, Bostwick DG, Tindall DJ. Expression of early growth response genes in human prostate cancer.
Cancer Res
1998
;
58
:
2461
–8.
49
Abdulkadir SA, Qu Z, Garabedian E, et al. Impaired prostate tumorigenesis in EGR-1-deficient mice.
Nat Med
2001
;
7
:
101
–7.
50
Yabe T, Herbert JT, Takanohashi A, Schwartz JP. Treatment of cerebellar granule cell neurons with neurotrophic factor pigment epithelium-derived factor in vitro enhances expression of other neurotrophic factors as well as cytokines and chemokines.
J Neurosci Res
2004
;
77
:
642
–52.
51
Feldman RJ, Sementchenko VI, Gayed M, Fraig MM, Watson DK. Pedf expression in human breast cancer is correlated with invasive potential and altered gene expression.
Cancer Res
2003
;
63
:
4626
–31.
52
Yang S-H, Jeffrey E, Hay RT, Sharrocks AD. Dynamic interplay of the SUMO and ERK pathways in regulating Elk-1 transcriptional activity.
Mol Cell
2003
;
12
:
63
–74.
53
Brinckerhoff CE, Matrisian LM. Matrix metalloproteinases: a tail of a frog that became a prince.
Nat Rev
2002
;
3
:
207
–14.
54
Lu W, Zhou X, Hong B, Liu J, Yue Z. Suppression of invasion in human U87 glioma cells by adenovirus-mediated co-transfer of TIMP2 and PTEN genes.
Cancer Lett
2004
;
214
:
205
–13.

Supplementary data