Dipeptidyl Peptidases as Survival Factors in Ewing Sarcoma Family of Tumors

Ewing sarcoma family of tumors (ESFT) is a group of aggressive pediatric malignancies driven by the EWS-FLI1 fusion protein, an aberrant transcription factor up-regulating specific target genes, such as neuropeptide Y (NPY) and its Y1 and Y5 receptors (Y5Rs). Previously, we have shown that both exogenous NPY and endogenous NPY stimulate ESFT cell death via its Y1 and Y5Rs. Here, we demonstrate that this effect is prevented by dipeptidyl peptidases (DPPs), which cleave NPY to its shorter form, NPY3–36, not active at Y1Rs. We have shown that NPY-induced cell death can be abolished by overexpression of DPPs and enhanced by their down-regulation. Both NPY treatment and DPP blockade activated the same cell death pathway mediated by poly(ADP-ribose) polymerase (PARP-1) and apoptosis-inducing factor (AIF). Moreover, the decrease in cell survival induced by DPP inhibition was blocked by Y1 and Y5R antagonists, confirming its dependence on endogenous NPY. Interestingly, similar levels of NPY-driven cell death were achieved by blocking membrane DPPIV and cytosolic DPP8 and DPP9. Thus, this is the first evidence of these intracellular DPPs cleaving releasable peptides, such as NPY, in live cells. In contrast, another membrane DPP, fibroblast activation protein (FAP), did not affect NPY actions. In conclusion, DPPs act as survival factors for ESFT cells and protect them from cell death induced by endogenous NPY. This is the first demonstration that intracellular DPPs are involved in regulation of ESFT growth and may become potential therapeutic targets for these tumors.

. All of these enzymes have been implicated in regulation of growth and metastases of many tumors (15,18,19). The increasing interest in these potential therapeutic targets led to the development of numerous DPP inhibitors. For example, a broad range DPP inhibitor, PT-100, has already been tested in clinical trials for the treatment of various cancers (20,21). Surprisingly, however, this compound was introduced to clinics without knowledge of the exact mechanisms of its actions, which ultimately led to the failure of these trials. Thus, given the role of DPPs in regulation of tumor growth, but also the complex nature of their actions, elucidation of the mechanism underlying their effects is essential for further clinical trials. This is particularly important for intracellular DPP8 and DPP9 because their ability to cleave secretory peptides, which are known DPPIV substrates, has been shown only in cell extracts but never documented in intact cells. Thus, as of now, their natural substrates have not been identified. Taking into account the known growth-inhibitory effect of NPY in ESFT and its interactions with DPPs, we sought to determine the role of both membrane and intracellular DPPs in regulation of ESFT growth and survival, as well as assess their value as therapeutic targets in these tumors.
Cell Culture-Human ESFT cell line, SK-N-MC, was obtained from ATCC (Manassas, VA) and cultured in Eagle's minimum essential medium with 10% FBS. Other cell lines were obtained and cultured as reported previously (23).
Cell Viability Assay-The cells cultured in 96-well plates were put into 0.25% FBS media, and 24 h later, they were treated with NPY (10 Ϫ7 M), DPP inhibitors (10 Ϫ5 M), and NPY receptor antagonists (10 Ϫ7 M). The concentrations were determined based on dose-response curves or previous studies (4,13). Cell viability was measured 48 h later using the MTS-based Cell-Titer 96AQueous One solution cell proliferation assay (Promega, Madison, WI).
siRNA Transfection-Predesigned siRNAs for NPY, DPPIV, DPP8, DPP9, FAP, and negative control were purchased from Applied Biosystems (Foster City, CA). Cells were transfected with 30 nM siRNA using the TransIT-TKO reagent (Mirus Bio, Madison, WI). The efficiency of inhibition was tested by real time RT-PCR, Western blot, neuropeptide Y enzyme immunoassay (Bachem), or DPP activity assay. For the survival assay, the transfected cells were cultured for 96 h in 1% FBS medium with or without NPY or Y1 and Y5R antagonists (10 Ϫ7 M), and then cell viability was assessed as above.
Real Time RT-PCR-RNA from cultured cells was isolated using the High Pure RNA Isolation Kit (Roche Applied Science) and from tissues using TRI reagent (Sigma). cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad Laboratories) and amplified using the iCycler iQ detection system (Bio-Rad Laboratories), TaqMan universal PCR master mix, and predesigned primers and fluorescein-labeled probes (Applied Biosystems). The results were calculated by the comparative C T method using ␤-actin as a reference gene.
Mass Spectrometry-Conditioned media collected after 24 h of culture were subjected to ultrafiltration at 37°C and 2900 rpm using 30-kDa cutoff filters. The ultrafiltrate contained ϳ7 mg/dl protein plus peptides, which include NPY 1-36 and NPY  . These were then quantified using multiple reaction mode monitoring. The multiple reaction monitoring transition for NPY   DPPIV mRNA Transfection-pGEM4Z plasmid encoding rat DPPIV cDNA (25) was linearized with HindIII restriction enzyme and used as a template for the in vitro transcription reaction performed using the mMESSAGE mMACHINE SP6 kit (Applied Biosystems). The Xenopus elongation factor 1␣ mRNA served as a control mRNA. SK-N-MC cells plated into 96-well plates were transfected with 2 ng/l DPPIV or control mRNA using Lipofectamine 2000 (Invitrogen). 18 h after transfection, the cells were assayed for DPP activity and treated in 2.5% FBS medium with NPY or Y1 and Y5R antagonists (10 Ϫ7 M). 48 h later, cell viability was assessed as above. For the cotransfection experiments, DPPIV mRNA was combined with 30 nM negative control siRNA or DPPIV siRNA (Applied Biosystems) and transfected as above.
Nuclear Extract Isolation and Western Blot-ESFT cells were treated with NPY (10 Ϫ7 M) with or without Y1 and Y5R antagonists (10 Ϫ7 M) in 0.25% FBS media. 1 or 8 h after treatment, the nuclear extracts were isolated using the NE-PER nuclear and cytoplasmic extraction kit (Thermo Scientific). SK-ES cells were transfected with the desired siRNA, and 24 h after transfection, they were treated in 1% FBS media with Y1 and Y5R antagonist (10 Ϫ7 and 10 Ϫ8 M, respectively). Approximately 54 h after transfection, the nuclear extracts were isolated, as above. The Western blot on nuclear extracts was performed using rabbit polyclonal anti-apoptosis-inducing factor (AIF) antibody (Cell Signaling Technology, Inc., Beverly, MA), whereas cytosolic fraction was used for detection of poly(ADP-ribose) (PAR) with rabbit polyclonal antibody (BD Pharmingen). Immunoblotting with rabbit polyclonal antibodies against DPPIV, DPP8, DPP9 (Abcam, Cambridge, MA), cleaved PAR polymerase-1 (PARP-1; Cell Signaling Technology), and mouse monoclonal anti-FAP antibody (Abcam) was performed on whole cell extracts. Mouse monoclonal anti-␤-actin antibody (Sigma) was used as a control. Densitometry was performed using the NIH Scion Image software (Scion Corp., Frederick, MD).
Colony Formation on Soft Agar-SK-ES cells were resuspended in 0.3% agar (2 ϫ 10 4 cells/ml) and overlaid onto 0.5% agar in 6-well plates in triplicates. Once the agar solidified, the medium with the desired treatments was added and changed daily for 5 days. The colonies were stained 2 weeks later using 0.005% crystal violet for 1 h at 37°C, and the number of colonies was counted using Image J.
Nude Mice Xenograft Model-7-10-week-old nude mice (Taconic, Hudson, NY) were subcutaneously injected into their right flank with 2 ϫ 10 6 of SK-ES cells suspended in 0.1 ml of Matrigel (BD Biosciences). 5 days after tumor cell inoculation, the daily treatment with NPY (10 Ϫ7 M) with or without P32/98 (10 Ϫ5 M), administered as local injection (ϳ1 cm from the tumor) of 100 l solution in saline or with saline alone, was started. Tumor size was measured periodically, and volume was calculated by the formula: 0.44 ϫ length ϫ width 2 (26). The SK-N-MC xenograft experiment and TUNEL staining were described previously (4).
Statistical Analysis-Statistical analysis was performed using the SigmaStat software. One-way repeated measure analysis of variance with post hoc t test (p Ͻ 0.05) using Dunnett's method was used for data comparison and analysis. Data are presented as mean Ϯ S.E. for the indicated numbers of repetitions. For the analysis of the SK-ES xenografts, the log-transformed tumor volumes were compared using linear mixed models to account for potential experimental effect and repeated measures of each animal.

RESULTS
ESFT Cell Lines Vary in Response to NPY-Previously, we have shown that NPY stimulates the death of SK-N-MC ESFT cells (4,13). However, not all ESFT cell lines are equally responsive to NPY. Although NPY treatment reduced the number of viable SK-N-MC cells with maximum effect at concentration 10 Ϫ7 M, no statistically significant effect was observed in SK-ES cells (Fig. 1A). In line with this, in the panel of nine ESFT cell lines, only three responded to NPY (10 Ϫ7 M) (Fig. 1B).
Similar differences were observed in response to endogenous NPY. In both SK-N-MC and SK-ES cell lines, two different NPY siRNAs exerted comparable inhibitory effects on expression and release of the tumor cell-derived NPY (supplemental Fig.  1). In SK-N-MC cells, this reduction in NPY levels led to a significant increase in the number of viable cells, suggesting that under basal conditions, the endogenous peptide decreases cell survival (Fig. 1C). The effect of NPY siRNA was reversed by treatment with exogenous NPY, which confirmed its specificity. In contrast, NPY siRNA in SK-ES cells had no effect on cell survival (Fig. 1C). Thus, in the cells non-responsive to exogenous NPY, the endogenous peptide also does not affect cell viability.
ESFT Cell Lines Not Responsive to NPY Release High Levels of NPY 3-36 -Because ESFT cell lines did not vary significantly in Y1 and Y5R expression (data not shown), we compared by mass spectrometry the forms of NPY released by ESFT cells in responsive and non-responsive cells. In SK-N-MC conditioned medium, NPY was detected mostly in its intact form, NPY 1-36 . In contrast, ϳ50% of NPY released by SK-ES cells was in its cleaved form, NPY  , inactive at Y1Rs (Fig. 2A). The relevance of these findings was confirmed by the fact that in SK-N-MC cells, the decrease in cell viability was achieved with NPY 1-36 , but not with NPY 3-36 (Fig. 2B). Because NPY 3-36 is a product of DPPIV cleavage, the above results suggested that the lack of the responsiveness to NPY in some ESFT cells might result from elevated DPPIV-like activity.
ESFT Cell Lines Vary in Expression and Activity of DPPs-As suggested by previous experiments, we evaluated the expression and activity of DPPIV and its analogs in nine investigated ESFT cell lines. Average DPPIV mRNA and activity levels were significantly higher in the cells non-responsive to NPY (5838, A4573, ES-925, SK-ES, TC-32, and TC-71), as compared with those responsive to the peptide (MMHES1, SK-N-MC, and RDES) (Fig. 2C). Moreover, the non-responsive cells had elevated DPP8/9 activity despite a lack of differences in mRNA levels of these cytosolic DPPs (Fig. 2C). FAP mRNA levels were highly variable, and no significant difference between responsive and nonresponsive cells was observed. In addition, under our culture conditions, no FAP activity was detected in ESFT cells.
DPPIV Overexpression Abolishes the Effect of NPY in Responsive Cells-To prove that the observed differences in DPP levels regulate the response of ESFT cells to NPY, SK-N-MC cells were transfected with DPPIV mRNA, which led to more than a 6-fold increase in the protease activity (Fig. 3A). As a result, the cells overexpressing DPPIV lost their original responsiveness to exogenous NPY, whereas both non-transfected cells and cells transfected with control mRNA maintained their Y1/Y5R-dependent responses. Co-transfection of DPPIV mRNA with DPPIV siRNA restored the response to NPY, confirming the specificity of the observed effects (Fig. 3A).
Knockdown of DPPs Restores the Effect of NPY in Non-responsive Cells-To determine whether down-regulation of DPP expression in cells not responding to NPY will render them responsive, SK-ES cells were transfected with siRNAs targeting particular DPPs. Real time RT-PCR analysis confirmed significant and specific knockdown of selected DPPs (supplemental Fig. 2), which resulted in a decrease in DPP protein levels and overall reduction of DPP activity with DPPIV, DPP8, and DPP9 siRNA transfections (Fig. 3B). Consistent with the lack of FAP activity in ESFT cells, FAP siRNA had no effect on overall DPP activity despite the effective inhibition of enzyme expression at the mRNA and protein levels (supplemental Fig. 2, Fig. 3B). In agreement with changes observed in DPP activity, transfections with DPPIV, DPP8, and DPP9 siRNAs, but not FAP siRNA, significantly decreased the number of viable SK-ES cells (Fig.  3C). Furthermore, all of these effects were blocked by NPY receptor antagonists: Y1R antagonist alone or in combination with Y5R antagonist. Thus, the effect of DPP siRNAs on SK-ES cell survival appears to be dependent on endogenous NPY. The level of cell death induced by DPP siRNAs was not increased by exogenous NPY, suggesting that the autocrine peptide is sufficient to saturate the receptors. The fact that the Y1R antagonist was equally effective alone, as it was in combination with the Y5R antagonist, indicated that blocking Y1Rs is sufficient to prevent the NPY effect in ESFT cells. Thus, although both Y1R and Y5R seem to contribute to NPY actions (4), activation of Y1Rs is indispensable for NPY-induced cell death in ESFT cells. This observation further confirmed the crucial role of DPPs, which convert NPY to its Y1R-inactive form, in preventing this effect.
Both NPY and DPP siRNAs Stimulate Cell Death via AIF-dependent Pathway-Previously, we have shown that the growthinhibitory effect of NPY in SK-N-MC depends on cell death, as shown by increased TUNEL staining both in vitro and in vivo (4). However, activation of the classical apoptotic pathway (caspase 3/7) was modest and observed late after NPY treatment. Instead, we observed molecular events consistent with AIF-mediated caspase-independent cell death ( Fig. 4A) (27)(28)(29)(30). In NPY-treated SK-N-MC cells, the elevated activity of PARP-1, measured by formation of PAR polymers, was detected 1 h after treatment and followed by an increase in nuclear levels of AIF 57 (Fig. 4B). This effect was blocked by Y1/Y5R antagonists (Fig. 4B).
In SK-ES cells, similar increases in PAR and AIF 57 levels were observed upon transfection with DPPIV, DPP8, and DPP9 siRNA, but not FAP siRNA (Fig. 4C). However, treatments with the same siRNAs in the presence of the Y1/Y5R antagonists did not cause a significant increase in AIF levels (Fig. 4D). Thus, changes in PARP-1 activity and nuclear levels of AIF 57 mimic the effects of NPY and DPP siRNAs on ESFT cell survival and suggest caspase-independent programmed cell death as their main mechanism.
Selective DPP Inhibitors Mimic the Effects of DPP siRNAs-Although the majority of DPPIV actions are due to its enzymatic activity, some of these result from its binding to other regulatory proteins and extracellular matrix (31). Thus, we used selective DPP inhibitors to confirm that the effects observed with DPP siRNAs depend on changes in DPP activity. To this end, we chose three representative cell lines with various levels of NPY release and DPP activity (Fig. 5A).
In SK-ES cells with high DPP activity and high NPY release (Fig. 5A), broad range DPP inhibitor, P32/98, as well as DPPIVselective (UG92) and DPP8/9-selective (UG93) inhibitors significantly decreased SK-ES cell viability to levels consistent with those achieved by NPY treatment (ϳ30% inhibition) (supplemental Fig. 3, Fig. 5B). As observed with DPP siRNAs, these effects were blocked by Y1R antagonist, but not enhanced with exogenous NPY. On the other hand, NPY did not exert a signif-icant inhibitory effect. A similar pattern was observed with PARP-1 activity. Although some increase in PAR polymer formation was observed 1 h after treatment with NPY, its levels were higher upon P32/98 administration (Fig. 5B). These results are consistent with the high levels of endogenous NPY and high DPP activity present in SK-ES cells.
In SK-N-MC cells with low NPY release and low DPP activity (Fig. 5A), exogenous NPY significantly decreased the number of viable cells (Fig. 5C). In contrast, broad range DPP inhibitor, P32/98, as well as selective DPPIV inhibitor, UG92, had no effect on cell survival, independent of the presence of exogenous NPY. These results were consistent with PARP-1 activation observed after treatment with NPY, but not with P32/98 alone (Fig. 5C). Selective DPP8/9 inhibitor, UG93, exerted a modest (ϳ15%) inhibitory effect even without exogenous NPY that was blocked by Y1R antagonist. Moreover, inhibition of DPP8/9 augmented the effect of exogenous NPY up to 40%, when UG93 and NPY were applied in combination. To confirm that the observed effects of DPP inhibitors are NPY-dependent, we used ES925 cells, which have high DPP activity but no detectable NPY release (Fig. 5A). Indeed, in these cells, neither NPY nor DPP inhibitors alone exerted an effect on cell viability (Fig. 5D). However, the Y1R-dependent decrease in cell survival was observed when DPP inhibitors were given in combination with exogenous NPY. Consistently, PARP-1 activation was detected only upon treatment with both NPY and P32/98. Thus, the presence of NPY is necessary for DPP inhibitor-induced cell death.
DPP Inhibitor, P32/98, Inhibits Colony Formation of SK-ES Cells-The effect of DPP inhibition on SK-ES cells was also tested using a colony formation assay in soft agar. As observed with the MTS assay, a broad range DPP inhibitor, P32/98, significantly decreased the number of SK-ES colonies. The effect of DPP inhibitor on colony formation was more pronounced than that observed under monolayer culture conditions (60% versus 25% inhibition, respectively). On the other hand, exogenous NPY had only a modest effect, which did not reach statistical significance (Fig. 6A). The growth-inhibitory effect of P32/98 was blocked by Y1R antagonist alone or in combination with Y5R antagonists (Fig. 6B). Therefore, the colony formation assay confirmed the results obtained in monolayer culture.
The Effect of NPY on ESFT Cell Survival Is Enhanced in Vivo-SK-N-MC and SK-ES xenografts seemed to recapitulate features of these cells cultured in vitro. Both NPY level and DPP activity in plasma of mice bearing SK-ES tumors were elevated, as compared with the plasma of animals with SK-N-MC xenografts (Fig. 7A). These systemic changes reflect a high release of NPY and shedding of DPPIV from SK-ES tumors.
Surprisingly, treatment with NPY alone resulted in significant inhibition of SK-ES xenograft growth (p ϭ 0.0016), which was associated with increased levels of cell death, as measured by DNA fragmentation (TUNEL) and PARP-1 cleavage, both of which are late events in AIF-mediated cell death (29,30) (Figs.  4A and 7, B-D). However, we were not able to achieve successful DPP inhibition in vivo with P32/98 despite testing different doses and routes of administration. Paradoxically, at the end of the experiment, DPP activity in P32/98-treated tumor tissues was significantly higher than in control, which was associated with elevated mRNA levels for all investigated DPPs (supplemental Fig. 4).

DISCUSSION
Recent microarray data identified NPY and its Y1 and Y5Rs as "Ewing sarcoma signature genes" up-regulated by the EWS-FLI1 fusion protein (3,6). However, a clinically relevant role of NPY in ESFT has not been elucidated. The cell death induced by endogenous NPY previously reported by us (4) seems paradoxical and raises the question of how ESFT cells are protected from its effect. The data presented here suggest that high levels of DPP expression in ESFT cells provide a survival mechanism for these cells. We have shown that not only membrane DPPIV, but also its intracellular homologs, DPP8 and DPP9, are actively FIGURE 4. NPY and DPP siRNAs induce caspase-independent cell death. A, caspase-independent programmed cell death is triggered by factors increasing intracellular calcium levels, which leads to formation of reactive oxygen species (ROS), DNA damage, and PARP-1 activation resulting in formation of PAR polymers. This, in turn, leads to further deregulation of mitochondrial calcium and cleavage of AIF to AIF 57, which translocates to the nucleus and induces DNA fragmentation. In the end stages of this process, PARP-1 cleavage is observed. B, SK-N-MC cells were treated with NPY with or without Y1R and Y5R antagonists (10 Ϫ7 M). 1 and 8 h after treatment, nuclear and cytosolic extracts were collected, and levels of PAR and AIF were assessed by Western blot. 1 h after treatment, the presence of PAR polymers in cytosolic extracts was observed, whereas at 8 h, an increase in the nuclear levels of AIF 57 was detected. This effect was blocked by Y1R and Y5R antagonists (10 Ϫ7 M). C and D, SK-ES cells were transfected with 30 nM negative control (NC) or DPP siRNAs with or without Y1R and Y5R antagonists (10 Ϫ7 M). 48 h later, cytosolic and nuclear extracts were collected, and levels of PAR polymers and AIF 57 were assessed, respectively. DPPIV, DPP8, and DPP9 siRNA transfections increased the levels of PAR and AIF 57 (C). No significant increases in nuclear levels of AIF 57, as measured by densitometry, were observed when SK-ES cells were transfected with DPP siRNAs in the presence of Y1R and Y5R antagonists (D). For B and D, the graphs represent average densities of AIF 57 normalized to ␤-actin from three independent experiments. Representative Western blots are shown. *, p Ͻ 0.05 as indicated (A) or as compared to negative control (NC) siRNA (D).
involved in preventing NPY-induced cell death. This is the first demonstration that these enzymes are involved in regulation of ESFT growth and the first evidence that the intracellular DPPs are able to cleave releasable peptides in intact cells.
We have shown that NPY directly stimulates cell death only in ESFT cells with low DPP activity. In these cells, NPY is released mainly in its intact NPY 1-36 form, which can activate Y1R/Y5Rs. We have identified PARP-1-and AIF-dependent programmed cell death as the primary mechanism of NPY actions in ESFT cells. This observation is consistent with the fact that the majority of known ESFT cell lines, including these used in our study, lack a functional p53 pathway (32), which is necessary for intrinsic activation of caspase-dependent apoptosis. In such cells with impaired apoptotic pathways, AIF serves as a substitute mechanism of executing cell death (27,28). Moreover, PARP-1 and AIF activation is triggered by an  ). B, in SK-ES cells, which contain high levels of endogenous NPY and high DPP activity, exogenous NPY did not exert a growth-inhibitory effect, whereas DPP inhibitors alone significantly decreased cell viability. This effect was blocked by Y1R antagonist. C, in SK-N-MC cells, which present with low levels of endogenous NPY and low DPP activity, exogenous NPY decreased cell survival, whereas broad range and DPPIV-specific inhibitors had no effect. DPP8/9specific inhibitor, UG93, significantly decreased the number of viable cells and enhanced the effect of NPY alone, which was blocked by Y1R antagonist. D, in ES925 cells, which do not release NPY and possess high levels of DPPs, the decrease in cell survival was achieved only when exogenous NPY was applied with DPP inhibitors. Similar patterns of responses were observed with PAR polymer formation. *, p Ͻ 0.05 as compared to non-treated control; #, p Ͻ 0.05 as indicated.
increase in intracellular calcium, a classical signaling event induced by NPY in various cells, including SK-N-MC (27,33,34).
In the cells with high DPP activity, a significant amount of NPY is released in its NPY 3-36 form, which does not activate Y1Rs and, consequently, does not stimulate cell death. However, the effect of NPY can be restored by blocking DPPs with selective inhibitors or siRNAs. Interestingly, similar effects were observed for the membrane-bound DPPIV as for cytosolic DPP8 and DPP9. Although NPY is a well established substrate for DPPIV (17), the natural substrates of the cytosolic DPPs have not been identified. Both DPP8 and DPP9 have been shown to cleave NPY in cell extracts (22,35); however, the ability of cytosolic enzymes to process releasable peptides under physiological conditions is highly controversial. Recently, DPP8 has been shown to cleave certain chemokines in vitro and has been implicated in the inactivation of internalized chemokine-receptor complexes; however, no direct proof for the latter was provided (36). Here, we have shown that blocking DPP8 and DPP9 activates the same cell death pathway as NPY and that all of these effects can be blocked by NPY receptor antagonists. Thus, we provide the first evidence that NPY is a natural substrate of these intracellular DPPs. This phenomenon can be explained by the fact that NPY is present in various intracellular compartments, such as the cytosol, nucleoplasm, and mitochondria (37)(38)(39). Moreover, Y1Rs have been identified on the nuclear membrane, suggesting an intracellular mode for NPY actions (37). Thus, cytosolic DPPs may be potentially involved in regulating such actions of the peptide. Although the efficiency of NPY cleavage by DPP8 and DPP9 is ϳ2-fold lower than that of DPPIV (35), DPP8/9 accounts for ϳ80% of overall DPP activity in ESFT cell extracts (13). Due to this high representation of DPP8/9, these proteases may still significantly contribute to NPY cleavage despite their lower activity limited to the intracellular pool of the peptide. This was observed in SK-N-MC cells, in which DPP8/9-selective inhibitor significantly inhibited cell survival in an NPY-dependent manner, whereas the DPPIV-selective inhibitor had no effect.
FAP, like DPPIV, is a membrane protease. However, its substrate specificity varies significantly from DPPIV. Unlike other DPPs, FAP has endopeptidase activity, whereas its DPP activity is lower than that of DPPIV for most substrates (40,41). In the panel of ESFT cells, we did not detect significant activity of FAP. However, its catalytic efficiency for the hydrolysis of Gly-Pro dipeptide used in our assays is ϳ100-fold lower than observed for DPPIV (40,41). Moreover, the efficiency of FAP cleavage for dipeptides with tyrosine in the first position, as in NPY, is among the lowest observed for various P 2 -Pro1 combinations (41). Therefore, it is very unlikely that FAP plays a significant role in NPY processing.
Despite the fact that FAP did not seem to be involved in regulation of ESFT growth and survival, its role in these tumors cannot be underestimated. Although we did not detect a significant FAP-dependent DPP activity in non-confluent ESFT cells, it appeared to increase in confluent cells, whereas DPPIV activity decreased (data not shown). Thus, DPPIV may be important under conditions promoting cell proliferation, whereas FAP may be involved in other functions of ESFT cells, such as invasiveness and motility. Importantly, these actions of both FAP and DPPIV are often not dependent on their DPP activity but rather on their binding properties (31,42,43).
Actions of DPPs may be further altered in vivo due to their high expression in stromal cells, such as tumor-associated fibroblasts, endothelial cells, and immune cells (43)(44)(45)(46). For example, we have shown that DPPIV activity in cell extracts from TC-32 ESFT cells accounts for only 20% of total DPP activity, whereas in extracts from TC-32 xenograft tissue, DPPIV-dependent activity rises to 40% (13). Thus, the relative contribution of particular DPPs may change significantly in the tumor tissue, as compared with the isolated tumor cells. Moreover, in the tumor microenvironment, DPPs may acquire new functions dependent on interactions with host cells, extracellular matrix, and stroma-derived proteolytic substrates, such as regulation of cell invasiveness and anti-tumor immune response (19,21,45).
The importance of tumor environment and the clinical relevance of our data are further supported by the fact that the effect of NPY and DPP inhibitor was significantly enhanced in three-dimensional culture and in vivo. As shown previously, the relatively modest effect of NPY on SK-N-MC cell survival in vitro translated to a substantial inhibition of tumor growth in vivo (4). Similarly, in SK-ES xenografts, NPY inhibited tumor growth despite no significant effect on survival of these cells in culture. Thus, prolonged treatment with high doses of NPY in the context of the tumor microenvironment may partially overcome the protective effect of DPPs and lead to effective activation of Y1R/Y5R-mediated apoptosis. These results are consistent with our observations in neuroblastoma, where the effect of NPY on tumor growth was also much more dramatic in vivo than in vitro (47).
Taking into account the significant growth-inhibitory effects of NPY in vivo, the successful inhibition of DPP activity in combination with NPY treatment should lead to pronounced reduction of the tumor growth rate. Moreover, in vivo, DPP inhibitors should not only enhance the Y1R/Y5R-mediated cell death of tumor cells but also impair Y2R-mediated angiogenesis, normally favored by DPPs (4). Unfortunately, our attempts to use a broad range DPP inhibitor, P32/98, in vivo were not successful because prolonged treatment with P32/98 triggered up-regulation of the DPP expression and activity to the levels higher than in control. Therefore, careful pharmacodynamic assays will be important with clinical use of these agents against tumors.
Although we were not able to attain an effective DPP inhibition with P32/98, this may be achieved with more potent inhibitors. Recent years have brought a strong interest in DPPIV and its homologs, which led to the development of numerous highly selective and potent DPP inhibitors. For example, DPPIV-selective inhibitors, Sitagliptin and Vildagliptin, are already used for the treatment of diabetes (48), whereas the broad range DPP inhibitor, PT-100, has been tested in clinical trials for various types of cancer (20,21,49). Using such potent inhibitors in an appropriate treatment regime may allow overcoming or avoiding induction of DPPs observed in P32/98-treated tumors.
In summary, we have shown that both DPPIV and its cytosolic homologs, DPP8 and DPP9, serve as survival factors for ESFT cells, protecting them from NPY-induced cell death mediated by Y1R/Y5Rs. This is the first demonstration that intracellular DPPs are involved in regulation of ESFT growth and the first evidence that NPY is their natural substrate. These findings indicate that targeting DPPs in vivo may enhance the growth-inhibitory effect of NPY and become a new therapeutic strategy for ESFT. Moreover, given the pleiotropic actions of NPY and its processing enzymes, DPPs, continued explorations of this EWS-FLI1-driven pathway may also open other therapeutic opportunities for ESFT patients.