Activity-based Protein Profiling Implicates Urokinase Activation as a Key Step in Human Fibrosarcoma Intravasation*

Entry of malignant cells into the vasculature (i.e. intravasation) requires proteolytic remodeling of the extracellular matrix so that tumor cells may pass through the local stroma and penetrate the vessel wall. The circulatory system then provides a means of transporting tumor cells to distant sites where they extravasate and establish metastatic lesions. This study utilizes activity-based protein profiling to compare the active serine hydrolase repertoire in high intravasating (HT-hi/diss) and low intravasating (HT-lo/diss) variants of the human fibrosarcoma HT-1080 cell line to determine which enzyme(s) play a role in intravasation. Activity-based protein profiling revealed multiple serine hydrolases with altered activity between HT-hi/diss and HT-lo/diss cells, with the largest difference being the activity of urokinase-type plasminogen activator (uPA). Levels of inactive uPA zymogen were similar between the two cell variants, but only HT-hi/diss conditioned medium contained active uPA, suggesting that uPA activation may contribute to the enhanced intravasation of HT-hi/diss cells. To analyze the role of uPA activity specifically in the process of intravasation, we grafted cells from the two HT-1080 variants onto the chorioallantoic membrane of chick embryos and measured levels of tumor cell intravasation in the distal chorioallantoic membrane using quantitative human-specific Alu PCR. Inhibition of uPA activity with natural (plasminogen activator inhibitor-1) or synthetic (amiloride) inhibitors diminished HT-hi/diss Matrigel invasion in vitro and intravasation and metastasis in vivo. Additionally, treatment of HT-lo/diss tumors with exogenous active uPA increased the number of intravasated cells in vivo. These results indicate that active uPA promotes tumor cell intravasation and that uPA activation appears to be a key step in tumor progression.

Proteolytic enzymes play a critical role in tumor cell dissemination (1)(2)(3). They are responsible in part for remodeling of the extracellular matrix (ECM), 2 thus allowing tumor cells to invade the surrounding tissue, enter the vasculature, and establish metastatic foci at distant sites. Therefore, analyzing and understanding the protease activities induced by malignant cells may provide valuable insight into the metastatic process.
Activity-based protein profiling (ABPP) is a chemical proteomic technique that uses active site-directed probes linked to a reporter group to measure the activity levels of individual enzymes of a given class in a complex mixture. ABPP combined with mass spectroscopy can identify enzymes that have altered activity levels between experimental groups, such as metastatic and nonmetastatic tumor cells. Since proteins are detected based on activity rather than abundance, ABPP can distinguish between active enzymes and their inactive zymogen or their inhibitor-bound forms (4 -7), thereby providing an advantage compared with other methods of profiling. In this regard, profiling methods such as RNA microarrays are limited, because they only measure changes in mRNA pools that are not necessarily reflected at the protein level, and protein profiling methods, such as differential twodimensional gel electrophoresis, only measure protein quantity, disregarding some types of post-translational modification and the presence of endogenous inhibitors. In contrast, ABPP identifies molecules with altered levels of activity, thereby indicating prime candidates for enzymes that regulate the specific phenotype being studied. The ABPP system also allows for screening of specific inhibitors by adding the putative inhibitor prior to labeling the protein sample with the activitybased probe. A reduction in the amount of probe bound to the enzyme would demonstrate that the candidate inhibitor specifically blocked or reduced the activity of the given enzyme. That inhibitor then could be used to validate the specific role of the active enzyme in functional model systems.
ABPP has been used to screen various tumorigenic cell lines to identify active enzyme profile differences between aggressive and nonaggressive cell types (8,9), but the biological significance of these differences has yet to be determined. We describe herein the use of an ABPP probe designed to detect the serine hydrolase family of enzymes in a complex mixture of proteins, such as cell lysates or conditioned media. The serine hydrolase family contains several proteolytic enzymes implicated in tumor progression, including urokinase-type plasminogen activator (uPA). uPA plays an important role in remodeling of the ECM by converting plasminogen into plasmin, which facilitates tumor cell invasion by degrading various components of the ECM and by activating other protease zymogens. The role of uPA in tumor growth, invasion, and metastasis has been widely studied and is the subject of several reviews (10 -12), yet there are still aspects of the role of uPA in the metastatic cascade, including intravasation, that are not fully understood.
Regulation of uPA activity is a complex process with various factors influencing uPA at several levels. uPA is translated as an inactive zymogen, pro-uPA, which requires proteolytic cleavage to become active. The efficiency of this cleavage is regulated by the abundance and activity state of molecules, such as plasmin and the uPA receptor (uPAR). In addition, uPA activity can be restricted by the endogenous serine protease inhibitors PAI-1 and PAI-2, which have multiple activation states of their own. Plasmin is believed to be the main activator of uPA, with uPA binding to uPAR improving the kinetics of the activation reaction (13). Other molecules, such as matriptase (14,15), cathepsin B (16), and cathepsin L (17), also have been shown to activate uPA in vitro. The role of the urokinase plasminogen activation system (uPAS) was analyzed in numerous studies of metastasis via uPA overexpression (18), inhibition of uPA with function-blocking antibodies (19), down-regulation of uPAR expression (20 -23), inhibition of the uPAR⅐uPA complex formation (24,25), and generation of uPA-deficient mice (26 -28). Together, the results of these studies indicate that the uPAS is involved in the progression of tumor metastasis.
Within the metastatic cascade, tumor cell intravasation remains one of the least studied steps. Intravasation, the process by which tumor cells enter into the vasculature, is one of the rate-limiting and earliest events in metastasis. The chorioallantoic membrane (CAM) assay is a valuable in vivo model for studying intravasation, because it recapitulates the steps of metastasis in a short period of time and is uniquely suited for quantitative measurements of intravasation (29 -31). Tumor cells are grafted onto the highly vascularized CAM of a developing chick embryo through a window cut in the egg shell. The grafted tumor cells form a primary tumor from which aggressive cell types may proceed to intravasate into the CAM vasculature and establish metastatic lesions in distant organs of the chick embryo. Numbers of intravasated cells can be quantitatively measured in a distant portion of the CAM by speciesspecific Alu PCR, since the Alu repeats found in high abundance in the human genome are absent the chick genome (29,31).
To analyze the role of serine proteases in intravasation, we utilized a high disseminating (HT-hi/diss) and a low disseminating (HT-lo/diss) variant of the HT-1080 human fibrosarcoma cell line recently generated by in vivo selection (32). Both HT-1080 variants give rise to primary tumors of approximately the same size, yet only HT-hi/diss cells efficiently intravasate into the CAM vasculature (32). In this study, the ABPP system was used to identify candidate enzymes involved in tumor cell intravasation by directly comparing the activity profiles of HT-hi/ diss and HT-lo/diss cells with a fluorophosphonate (FP) ABPP probe designed to detect active serine hydrolases. Several bands of activity on the ABPP gel differed between HT-hi/diss and HT-lo/diss cells, with the largest difference being the activity level of uPA. Despite producing abundant amounts of pro-uPA, the HT-lo/diss variant fails to activate uPA. In contrast, HT-hi/diss cells are characterized by high levels of active uPA and a highly metastatic phenotype, indicating that uPA activation may be a contributory factor in the intravasation ability of tumor cells. To study the role of active uPA in tumor cell intravasation, the ABPP system was further utilized to search for a synthetic inhibitor that directly and specifically inhibits uPA enzymatic activity. Inhibition of uPA activity with a low dose of amiloride (10 M) significantly reduced the rate of intravasation and metastasis of HT-hi/diss cells but had little effect on primary tumor growth. Conversely, the addition of a small amount of exogenous active uPA to HT-lo/diss cells increased their intravasation in vivo. These findings highlight the efficiency of ABPP in analyzing the role of serine proteases in tumor progression and demonstrate that uPA activity is a key determinant of tumor cell intravasation.

EXPERIMENTAL PROCEDURES
CAM Intravasation Assay-Fertilized White Leghorn eggs (SPAFAS, Storrs, CT) were incubated in a rotary incubator at 38°C with 60% humidity for 10 days. On day 10, the CAM was dropped by grinding away a small portion of the shell in the vascularized area of the CAM and applying a mild suction to a hole drilled into the air sac of the egg. A small window was then cut in the eggshell above the dropped CAM to allow for the grafting of tumor cells onto the CAM and for future treatments. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) containing 10% fetal calf serum (Atlas Biological) and Gentamicin (Invitrogen) (D-10), harvested by trypsinization, washed with serum-containing medium, and resuspended at 2 ϫ 10 7 cells/ml in serum-free DMEM. 1 ϫ 10 6 tumor cells were added to the dropped CAM of each embryo. The windows were subsequently sealed, and the eggs were placed in a stationary incubator for an additional 5 days. For topical treatments, amiloride (Calbiochem), PAI-1 (a generous gift from Dr. David Luskotoff), or active uPA (Chemicon) was applied directly to the CAM in a volume of 100 l at the indicated time points. Aprotinin (Calbiochem) was injected intravenously at a concentration of 1 mM in a volume of 100 l on days 2 and 4. Aprotinin, active uPA, and PAI-1 were dissolved in PBS, whereas amiloride was dissolved in 20% Me 2 SO (Sigma). On day 5 or 6 the tumors were excised, and a portion of the CAM most distal to the area of the grafted cells, the lung, and a portion of the liver were harvested. The numbers of human tumor cells present in the distal CAM, liver, or lung were measured by quantitative Alu PCR representing intravasated (distal CAM) and metastatic cells (liver and lung) (31).
Protein Preparation and ABPP-HT-hi/diss and HT-lo/diss cells, generated from the HT-1080 human fibrosarcoma cell line (32), were grown to 80% confluence in D-10 and then switched to serum-free DMEM plus Gentamicin for 48 h, after which conditioned medium was collected and cells were harvested. Conditioned medium from MDA-MB-231mfp cells, generated from in vivo passage of MDA-MB-231 cells (33), was produced in a similar fashion, with the cells grown in L15 medium from ATCC with 10% fetal calf serum, glutamine, and Hepes. Cells were lysed by Dounce homogenization and sonication in 50 mM Tris-HCl, pH 8.0 (Buffer 1), followed by centrifugation at 100,000 ϫ g for 60 min. The resulting supernatant was collected as the soluble fraction, and the pellet was resuspended in Buffer 1 and analyzed as the insoluble fraction. Conditioned medium was centrifuged at 2,400 ϫ g for 5 min, precipitated with ammonium sulfate (56.1 g/100 ml), resuspended in 50 mM Tris-HCl, pH 7.5 (Buffer 2), and desalted over a PD-10 column (Amersham Biosciences). The PD-10 eluate was then concentrated ϳ2ϫ with a Centricon 10 filter (Millipore Corp.).
Proteome Labeling-Protein fractions were labeled at a concentration of 1 mg/ml in their respective 50 mM Tris buffer with a rhodaminecoupled FP probe, 4 M FP-PEG for conditioned medium proteins and 2 M FP-TAM for soluble and insoluble proteins. Following labeling for 60 min at room temperature, an aliquot of each proteome was deglycosylated with PNGase F according to the manufacturer's instructions (New England Biolabs). Inhibition assays were performed by preincubation of a sample with the indicated inhibitor for 15 min prior to the addition of the FP probe. The labeling reaction was stopped by the addition of an equal volume of 2ϫ reducing SDS-PAGE loading buffer and boiling for 8 min. Labeled proteins were resolved by SDS-PAGE and then visualized and quantitated in gel with a Hitachi FMBio IIe flatbed fluorescence scanner (MiraiBio).
Identification of FP-labeled Active Proteins-Protein fractions were labeled as previously described with the exception of using a trifunctional probe, containing both a rhodamine tag and a biotin group attached to the FP-reactive group. Labeled proteins were enriched by an avidin-based precipitation and separated by SDS-PAGE. Bands of interest were excised and digested with trypsin. Protein identification was accomplished by analyzing the tryptic digest with microcapillary liquid chromatography-electrospray tandem MS (LC-MS/MS), using a 1100 high pressure liquid chromatograph (Agilent, Palo Alto, CA) coupled to a Finnigan LTQ Deca MS (Thermo Finnigan, San Jose, CA). The resulting data were used to search a public data base for protein identification.

uPA Activation, a Key Step in Tumor Cell Intravasation
Matrigel Assay-Invasion assays were performed using 6.5-mm diameter Transwells (Costar) with the 8-m pore membranes coated with Matrigel (34). Briefly, Matrigel (BD Biosciences) was diluted 1:25-1:30 with PBS, and 50 l was placed on the membrane of the insert and allowed to dry overnight at room temperature under a laminar flow. Matrigel was reconstituted by incubation of the inserts in PBS at 37°C. Cells from a nonconfluent plate were detached with trypsin/EDTA, washed in D-10, and resuspended in serum-free DMEM. A total of 1-2 ϫ 10 5 cells in 0.1 ml were placed into the inner chamber of the Transwell, and the outer chamber was filled with 0.25 ml of DMEM, 5% fetal calf serum. Amiloride or PAI-1 was added to the both inner and outer chambers at the indicated concentrations. Following incubation for 18 -24 h, the invaded cells were detached from the undersurface of the membrane and bottom of the individual Transwells and counted using a hemocytometer.
Proliferation Assay-8.5 ϫ 10 3 HT-hi/diss cells per well were seeded in 24-well plates in D-10 supplemented with the increasing concentrations of amiloride, from 10 M to 1 mM. Culture medium containing the appropriate concentration of amiloride was exchanged on a daily basis. Each day, at least four wells for each amiloride concentration were harvested by trypsinization and counted with a hemocytometer after staining for viable cells with trypan blue.
Western Blotting-10 g of concentrated conditioned media were resolved on a 10% SDS-polyacrylamide gel and transferred to Immobilon-P polyvinylidene difluoride membrane (Millipore). After transfer, the membrane was blocked with 5% milk in PBS containing 0.1% Tween 20 (PBS-T) for 1 h and probed overnight with American Diagnostica 398 anti-uPA antibody. Membranes were then washed three times with PBS-T, 5 min each, and probed with a horseradish peroxidase-conjugated secondary antibody in 5% milk-PBS-T for 1 h, followed by four washes in PBS-T, once for 15 min and three times for 5 min each. Immunoreactive bands were visualized with SuperSignal West Pico Chemiluminescent Substrate (Pierce) and Blue Bio Film (Denville Scientific) and then imaged with an Alpha Imager 2000 (Alpha Innotech).
Statistical Analysis and Data Presentation-Results are expressed as mean values Ϯ S.E., given as a percentage change compared with vehicle control (100%) values. The data were analyzed by Student's t test (Microsoft Excel) to determine statistical significance, expressed as a p value. p values less than 0.05 were considered significant.

Serine Hydrolase Activity Profiling of HT-hi/diss and HT-lo/diss
Cells-To establish the role of serine proteases in tumor cell intravasation, we treated embryos carrying HT-hi/diss tumors with the broad spectrum serine protease inhibitor, aprotinin. Aprotinin treatment had no effect on tumor weight but reduced the number of intravasated cells in the distal CAM by 51% ( p Ͻ 0.05) (Fig. 1), indicating a contributory role of serine proteases in the intravasation process of HT-hi/diss cells.
To determine which specific serine protease(s) are involved in HThi/diss intravasation, we profiled the serine protease activity of HT-hi/ diss cells and HT-lo/diss cells with the ABPP system using an FP probe that specifically binds to enzymatically active members of the serine hydrolase family. Proteins in the isolated subfractions of HT-hi/diss and HT-lo/diss cultured cells were labeled for ABPP analysis and resolved on an SDS-polyacrylamide gel after a portion of each sample was deglycoslylated by PNGase F. Analysis of the ABPP gel identified four areas of serine protease activity that significantly differed between the HT-hi/ diss and HT-lo/diss variants, indicated by the yellow arrows in Fig. 2A. Two areas of altered activity were observed in the conditioned media, a 50-kDa band with activity only in HT-lo/diss conditioned medium and a 33/28-kDa zone (without or with deglycosylation, respectively) only active in HT-hi/diss conditioned medium ( Fig. 2A, lanes 1-4). In addition, the soluble fraction from HT-hi/diss cells contained a band of increased activity at 34 kDa ( Fig. 2A, lanes 5-8), whereas the HT-lo/diss insoluble fraction contained a more active band at 37 kDa ( Fig. 2A, lanes 9 -12).
ABPP Identifies uPA Activity as a Distinguishing Characteristic of HT-hi/diss Cells-The largest observed difference in serine protease activity between the HT-hi/diss and HT-lo/diss variants detected by the FP probe was a secreted protein that, under reducing conditions, migrated as a broad band at 33 kDa without PNGase F treatment and as a 28-kDa band after deglycosylation by PNGase F (Fig. 2A, lanes 3 and 4,  respectively). An unrelated PNGase F-insensitive band at 28 kDa was present in both HT-hi/diss and HT-lo/diss conditioned medium, thereby masking the magnitude of the 28-kDa activity difference in the PNGase F-treated lanes. The broad zone of activity at 33 kDa, characteristic of HT-hi/diss conditioned medium, was absent in HT-lo/diss cells and therefore was chosen as the initial target for further analysis.
To identify the protein responsible for the 33/28-kDa activity, HT-hi/ diss conditioned medium was enriched for proteins that bind the serine hydrolase probe by labeling the proteins with a biotin-tagged FP probe and precipitating the labeled proteins with avidin-conjugated beads, followed by LC-MS/MS analysis. The MS/MS spectra and peptide coverage map clearly identified the 33/28-kDa band as uPA (Fig. 2B). Fifteen peptides and their corresponding amino acid sequences, determined by MS/MS, matched the uPA sequence, giving 31% coverage of the uPA B-chain. Moreover, 14 of the 15 peptides that matched uPA are not found in any other protein in the data base. Therefore, identification of the 33/28-kDa band as uPA was unambiguous.

uPA Activation, a Key Step in Tumor Cell Intravasation
Activation of Pro-uPA Occurs in HT-hi/diss Cells but Not in HT-lo/ diss Cells-Analysis of uPA in HT-hi/diss and HT-lo/diss conditioned media by Western blotting under nonreducing conditions demonstrated that both cell variants produce uPA zymogen. In fact, the 50-kDa band representing the inactive single chain zymogen and the disulfidelinked two-chain active protein was slightly more abundant in HT-lo/ diss cells, compared with HT-hi/diss cells (Fig. 2C). However, Western blotting under reducing conditions confirmed that only HT-hi/diss conditioned medium contained active uPA, as demonstrated by the presence of the 33-kDa B-chain band (Fig. 2C). Therefore, the Western blot analysis allowed us to conclude that the lack of uPA activity in HT-lo/diss conditioned medium on the ABPP gel was truly due to an absence of active uPA rather than inhibition of uPA activity by endogenous inhibitors, such as PAI-1, the presence of which could have prevented ABPP labeling. Importantly, these results demonstrate for the first time the ability of the ABPP system to identify an activity difference of an enzyme whose protein levels were shown to be equal. Thus, the Western blot analysis not only confirms the finding obtained by ABPP

uPA Activation, a Key Step in Tumor Cell Intravasation
but exemplifies the usefulness of the chemical-based activity assay for identifying enzymatic activity differences in proteins that have active and inactive forms which are not easily distinguished by standard gel electrophoretic procedures.
For additional confirmation that uPA was responsible for the bands of activity at 33/28 kDa in HT-hi/diss conditioned medium on the ABPP gel, the natural uPA inhibitor PAI-1 was added 15 min prior to labeling with the FP probe. The addition of 0.4 M PAI-1 specifically blocked the activity of the 33-kDa band and abrogated the increased intensity of the 28-kDa band (Fig. 2D, lanes 6 and 8), supporting the identification of these bands as uPA. It should also be noted that activity of the 50-kDa band, present only in HT-lo/diss conditioned medium, was not inhibited by the addition of PAI-1 (Fig. 2D, lanes 2 and 4), indicating that uPA was not responsible for the serine protease activity manifested at 50 kDa.
uPA Activity Regulates Intravasation of the HT-1080 Cell Variants-To determine the in vivo functional relevance of the uPA activity difference detected with ABPP, we inhibited uPA activity in HT-hi/diss cells and also added exogenous active uPA to HT-lo/diss cells. Treatment of the embryos with 20 g of PAI-1, applied topically on days 2 and 4 after grafting of HT-hi/diss cells, decreased the number of intravasated cells by more than 50% (Fig. 3A). Conversely, the addition of 0.1 g of purified active uPA to embryos with developing HT-lo/diss tumors increased intravasation levels 2.2-fold (p Ͻ 0.05) (Fig. 3B). It is noteworthy that neither treatment affected primary tumor growth. These results demonstrate that activation of uPA is a significant event in the intravasation process of HT-hi/diss cells, and the lack of active uPA in HT-lo/ diss cells may be a determining factor in their inability to enter the vasculature.
ABPP Analysis Identifies Amiloride as a Specific Inhibitor of uPA-To further investigate the role of active uPA in tumor cell intravasation, we employed ABPP to identify a specific synthetic inhibitor of uPA that, in contrast to PAI-1, was readily available and could be easily applied to tumor-bearing embryos. We took advantage of the nature of the ABPP system to examine two commercially available inhibitors of uPA activity, amiloride and uPA STOP, for cross-inhibition of other similar serine proteases. The MDA-MB-231mfp (33) cell line was used as an additional source of conditioned medium so that the inhibitors could be tested against closely related serine proteases, such as tPA, whose in vitro activity levels were not detectable by ABPP in HT-hi/diss cells. The single chain zymogen of tPA is enzymatically active and therefore was detected on the ABPP gel at 60/50 kDa, whereas the B-chain of active tPA was resolved at 35/32 kDa (without or with deglycosylation, respectively) (Fig. 4, lanes 11 and 12). Treatment of HT-hi/diss conditioned medium with amiloride at 0.1 or 1 mM concentrations inhibited uPA activity by 96 and 97%, respectively (Fig. 4, lanes 3-6), but amiloride did not affect the activity of any other serine protease detected by the probe in either HT-hi/diss (Fig. 4, lanes 3-6) or MDA-MB-231mfp (Fig. 4,  lanes 13-16) conditioned media. Importantly, amiloride did not inhibit tPA, present in MDA-MB-231mfp conditioned medium (Fig. 4, lanes  13-16). In contrast, 10 M uPA STOP inhibited the activity of both uPA (97% in HT-hi/diss conditioned medium) and tPA (64% in MDA-MB-231mfp conditioned medium) (Fig. 4, lanes 7-10 and lanes 17-20,  respectively).
Amiloride Inhibits uPA-mediated Invasion, Intravasation, and Metastasis of HT-hi/diss Cells-We further analyzed the functional role of uPA, employing amiloride as a specific inhibitor of uPA activity. To determine the amiloride concentrations that would not affect tumor growth, cell proliferation assays were performed with HT-hi/diss cells in the presence of increasing concentrations of the inhibitor. Amiloride concentrations up to 30 M demonstrated no significant inhibition of HT-hi/diss proliferation (data not shown). Therefore, amiloride was used at concentrations of 30 M or less in our in vitro and in vivo experiments.
Initial studies on the potential effectiveness of amiloride as an inhibitor of uPA activity in a complex cellular setting were conducted using an in vitro Matrigel invasion assay. HT-hi/diss cells were treated with amiloride at 10 and 30 M, resulting in a significant (up to 69%, p Ͻ 0.0001) and dose-dependent inhibition of cell invasion (Fig. 5). Similarly, treatment of HT-hi/diss cells with the natural uPA inhibitor PAI-1 at concentrations of 40 and 100 nM resulted in a significant (up to 65%, p Ͻ 0.0005) and dosedependent inhibition of cell invasion (Fig. 5), confirming that inhibition of uPA activity reduces HT-hi/diss invasion in vitro.
Having established amiloride as an effective inhibitor of uPA-mediated cell invasion in vitro, we tested the effect of uPA inhibition on tumor cell intravasation and metastasis in vivo. Treatment of developing HT-hi/diss CAM tumors with 3 and 10 M amiloride showed a dose-dependent reduction of HT-hi/diss cell intravasation into the chick vasculature and metastasis to the liver. Whereas the primary tumor weight was only slightly reduced upon treatment with 10 M amiloride, the number of intravasated and metastatic HT-hi/diss cells was reduced by 55% (p Ͻ 0.001) and 54% (p Ͻ 0.005), respectively (Fig.  6). These findings indicate that inhibition of active uPA in vivo effec-

DISCUSSION
Understanding all steps of the metastatic cascade is crucial for effective cancer intervention, since it is the metastatic lesions, not the primary tumor, that often prove to be lethal (35). One of the earliest and rate-limiting steps of metastasis is intravasation, the process by which tumor cells enter into the vasculature. Once tumor cells have intravasated, the circulatory system carries the cells to distant sites, where they exit the vasculature and establish metastatic lesions. Being prevented from intravasating, tumor cells would be confined to the site of the primary tumor and would become a more focused target for cancer therapy.
A unique pair of cell lines, a highly intravasating (HT-hi/diss) and a low intravasating (HT-lo/diss) variant, were generated by in vivo selection from the heterogeneous HT-1080 cell line to investigate tumor cell intravasation (32). These two HT-1080 cell variants generate primary tumors of similar size in the embryonic chicken CAM assay, yet only HT-hi/diss cells exhibit the ability to intravasate and metastasize. However, in an experimental metastasis model where cells are injected intravenously, both cell variants colonize the distal CAM and metastasize to secondary organs at equal rates (32), indicating that it is specifically the process of intravasation that distinguishes the two cell variants. These results suggest that alternate compilations of enzymes may be required for the individual invasive steps of metastasis (i.e. intravasation and extravasation), highlighting the importance of studying the individual steps of the metastatic cascade. The focus of this study was to identify enzymes specifically involved in the seldom studied . Specificity analysis of two commercially available uPA inhibitors, amiloride and uPA STOP, using the ABPP system. 50 g of protein from HT-hi/diss or MDA-MB-231mfp conditioned media were treated with the indicated concentrations of uPA STOP or amiloride 15 min prior to the labeling reaction, or were left nontreated (No Inh.). Deglycosylation by PNGase F was preformed on each of the fractions, which were then electrophoresed next to their nontreated counterparts. Following treatment and FP labeling, the samples were resolved on a reducing 10% SDS-polyacrylamide gel. Active uPA, detected at 33/28 kDa in conditioned medium from HT-hi/diss cells (yellow arrows, lanes 1 and 2), was inhibited by amiloride and uPA STOP, indicated by the lack of corresponding 33/28-kDa uPA bands in lanes 3-10. In conditioned medium from MDA-MB-231 mfp cells, the tPA zymogen detected at 60/50 kDa (blue arrows, lanes 11 and 12) and the B-chain of active tPA detected at 35/32 kDa (green arrows, lanes 11 and 12), were not inhibited by 1 mM amiloride (lanes [13][14][15][16]. In contrast, both tPA zymogen and active tPA were inhibited by 10 M uPA STOP (lanes 17-20). uPA Activation, a Key Step in Tumor Cell Intravasation process of intravasation, by comparing the enzymatic activity of HT-hi/diss and HT-lo/diss cells in vitro. Analysis of enzymatic activity from primary tumors is possible (33), but the interference from host/stromal enzymes and the inability to detect secreted proteins deterred a direct analysis of primary tumors from the CAM assay in this study.
ABPP analysis with a fluorophosphonate probe that detects active serine hydrolases revealed uPA activity as a distinguishing characteristic between HT-hi/diss and HT-lo/diss cells. Logically, this difference in uPA activity could account for the striking difference in intravasation capability between HT-hi/diss and HT-lo/diss cells. In order to intravasate, tumor cells must degrade the surrounding ECM so that they can escape from the primary tumor, invade the local stroma, and ultimately enter the vasculature. Numerous proteolytic enzymes are involved in remodeling the ECM during intravasation, and the uPAS regulates some of these enzymes (e.g. plasmin/plasminogen). The uPAS has been implicated in the metastatic process in previous studies, typically by decreasing uPAR expression or interfering with the uPA/uPAR interaction to decrease uPA activity (20 -23, 25, 36). The concept of HT-hi/diss fibrosarcoma cells directly providing the uPA that promotes intravasation is supported by the finding that T241 (murine fibrosarcoma) cells also produce large amounts of functional uPA in uPA knock-out mice (37).
In this study, involvement of the uPAS in HT-hi/diss intravasation was first indicated by the inhibition of HT-hi/diss intravasation with a broad range inhibitor, aprotinin. Aprotinin is a general serine protease inhibitor that inhibits various extracellular proteases, including trypsin, chymotrypsin, plasmin, and kallikreins (38). Although aprotinin does not directly inhibit uPA, it does effectively inhibit plasmin, which is a major activator of uPA. In addition, plasmin itself is generated from plasminogen by active uPA in a positive feedback mechanism. Therefore, whereas inhibition of intravasation with aprotinin cannot be directly linked to uPA, these results do implicate a serine protease pathway(s), possibly plasmin and therefore the uPAS, in the regulation of intravasation.
Inhibition of intravasation with low microgram amounts of PAI-1, the natural serine protease inhibitor (serpin) of uPA, provides more convincing evidence that active uPA is involved in the increased intravasation capacity of HT-hi/diss cells (Fig. 3A). However, PAI-1 is not an ideal inhibitor for anti-metastatic treatment, since, paradoxically, high levels of PAI-1 have been shown to promote tumor progression by inducing tumor-associated angiogenesis and cellular motility, presumably through a mechanism independent of its protease inhibitor activity (39 -41). These data are supported by the correlation of high PAI-1 levels with a poor prognosis for survival in a variety of tumors (42,43). Therefore, we sought out an alternate inhibitor of uPA to confirm the role of uPA activity in intravasation.
We took advantage of the ABPP system to screen the enzyme specificity of two synthetic inhibitors of uPA, amiloride, and uPA STOP. The activity-based nature of ABPP and its gel-based detection system make it a useful tool for screening the functionality and specificity of potential inhibitors. ABPP can be used to test inhibition of any enzyme detected by one of the chemical probes, alleviating the need for developing an enzymatic activity assay for each enzyme. Another benefit of the ABPP inhibitor screening method is that cross-inhibition of enzymes with similar binding sites can be examined in the same assay. MDA-MB-231mfp cells were used as an alternate source of conditioned medium to test serine hydrolases whose activity could not be detected in conditioned medium from HT-hi/diss cells in vitro but may play a role in intravasation in vivo. The inhibitor screen indicated that amiloride specifically inhibited uPA activity and did not interfere with the activity of any other enzyme detected by the FP probe in proteins secreted by HT-hi/diss or MDA-MB-231mfp cells. In contrast to amiloride, uPA STOP inhibited tPA activity in MDA-MB-231mfp conditioned medium as well as uPA activity in HT-hi/diss cells. These results indicated amiloride was the preferable uPA inhibitor for future studies. Amiloride has been used previously as an effective inhibitor of uPA activity in a number of metastasis and angiogenesis studies (44 -49), but to our knowledge, the effect of uPA inhibition by amiloride on the process of intravasation has not been investigated.
The specific inhibition of uPA with amiloride is also beneficial in that it directly inhibits active uPA and does not disrupt the uPA/uPAR interaction, which is important, since uPA/uPAR binding induces intracellular signaling cascades that may play an important role in the regulation of cellular growth and survival via the extracellular signal-regulated kinase (ERK)/mitogen-activated protein kinase (MAPK) pathway (50 -52). By directly inhibiting uPA activity, the results of this study indicate that it is the enzymatic activity of uPA that is crucial for intravasation. Inhibition of uPA activity significantly reduced the ability of HT-hi/diss cells to intravasate; conversely, the addition of small amounts of exogenous active uPA increased the intravasation ability of noninvasive HTlo/diss cells. The 2-3-fold increased intravasation of HT-lo/diss cells after treatment with active uPA is a noteworthy result, because it sug- uPA Activation, a Key Step in Tumor Cell Intravasation gests that the inability to activate uPA is an important factor in the nonmetastatic characteristic of HT-lo/diss cells. However, inhibition of uPA activity does not completely block intravasation. A more substantial inhibition of uPA-mediated intravasation may require a combination of approaches, including simultaneous inhibition of uPA activation and uPA activity. Therefore, elucidation of the uPA activation pathway functioning in HT-hi/diss cells, which is apparently deficient in HT-lo/ diss cells, merits further investigation.
Whereas we cannot conclusively determine if the active uPA added to HT-lo/diss tumors is interacting with the cells or acting solely as an extracellular enzyme, it is likely that the active uPA is generating plasmin, which then remodels the ECM, allowing for an increased number of HT-lo/diss cells to intravasate. This model is supported by the data presented in Fig. 1, which shows that aprotinin, a known inhibitor of plasmin, significantly decreases the intravasation ability of HT-hi/diss cells. Alternate enzymes and pathways may also play a role in the increased intravasation ability of HT-hi/diss cells, including the matrix metalloproteases (32) and possibly other serine proteases with altered activity, indicated by ABPP in this study.
On a broader scale of metastasis research, these results lend support to the theory that only a subset of tumor cells are metastatic (35,(53)(54)(55)(56)(57). HT-hi/diss and HT-lo/diss variants were isolated from the heterogeneous HT-1080 cell line, yet only HT-hi/diss cells possess the ability to intravasate and subsequently metastasize in a spontaneous metastasis model. Whereas we do not know what percentage of the HT-1080 cells exhibit the more invasive type, HT-hi/diss cells are significantly more aggressive than the heterogeneous parental cell line (32). The results of this study indicate that uPA activation and/or activity might create a microenvironment that is conducive to tumor cell intravasation. Therefore, a possible mechanism for the increased metastatic potential of a subset of tumor cells may be the ability to activate uPA, generating a signaling cascade, involving activation of ECM remodeling enzymes, which in turn would allow tumor cells to invade the local stroma and intravasate into nearby blood vessels.
The combination of ABPP analysis with the chick embryo CAM assay provides a proficient tool for identifying candidate enzymes that regulate metastasis and an efficient model for testing the role of these enzymes in the specific steps of the metastatic process by altering their activities in vivo.
In this study, ABPP analysis with a fluorophosphonate probe revealed multiple zones of activity that differed between HT-hi/diss and HT-lo/ diss cells. Due to the intrinsic characteristics of HT-hi/diss and HT-lo/ diss cells, the enzymes that are responsible for these activity differences are considered potential regulators of tumor cell intravasation. This study indicates that one of the enzymes with altered activity between the cell variants, uPA, does play a key role in the process of intravasation, providing strong validation of the ABPP method in identifying candidate enzymes involved in regulating intravasation. Future studies will focus on identifying the other enzymes with altered activity between HT-hi/diss and HT-lo/diss cells, indicated by ABPP analysis, to determine the role of these enzymes in the process of intravasation.