Inhibition of receptor tyrosine kinase signalling by small molecule agonist of T-cell protein tyrosine phosphatase

Antibodies against TCPTP (mAb CF4, Calbiochem; 3E2 from M. Tremblay, McGill University, Canada), EGFR phosphotyrosine 1068 (Cell Signaling Technology), PDGFRβ phosphotyrosine 1021 (Santa Cruz), and α-tubulin (Hybridoma Bank) were used. HRP-conjugated second-stage reagents were used, as appropriate. siRNA against TCPTP has previously been shown to be specific to TCPTP and to have identical effects to another TCPTP-targeting oligo [6]. TCPTP siRNA was from Ambion and the All Stars negative control siRNA from Qiagen. Peptides containing the cytoplasmic tails of the α1 or α2 integrins, the integrin-α1 tail fused to the 11-amino acid -long TAT peptide, and scrambled TAT (scrTAT, peptide with the TAT sequence fused to a scramble sequence of the α1-tail amino acids [6]). peptides were synthesized by Innovagen. EGF was purchased from Sigma, and human recombinant PDGF-BB from Cell Signaling Technology. Recombinant TCPTP, its mutants, and the constitutively active TC37 were produced and purified as GST-fusion proteins in Escherichia coli and cleaved from the GST using PreScission according to the manufacturer's instructions (BD Biosciences). DiFMUP (6,8-difluoro-4-methylumbelliferyl phosphate) was purchased from Molecular Probes, and Cell Titer Blue from Promega. Of the small molecule libraries, Spectrum 2000 was from Microsource Discovery Systems, LOPAC (Library of Pharmacologically Active Compounds) 1280 from Sigma, ChemDiv from ChemDiv Inc., ChemBridge from ChemBridge Corporation, and Tripos from Tripos International. Spermidine trihydrochloride, Mitoxantrone, Ruthenium red, and MDL-26,630-trihydrochloride were also from Sigma. Compounds N21 and F12 were no longer available from the library provider and could not be tested further.

HeLa cells (ATCC) were maintained in DMEM, 10% FBS, 2 mM L-glutamine, 5% CO2. Primary HUVECs were freshly isolated as described [26] and cultured in EBM2 medium. Only HUVECs passaged ≤ three times were used in this study. They were transfected with siRNA duplexes (100 nM) by nucleofection using Amaxa. Nucleofected HUVECs were harvested for RNA extraction 4 days post-transfection and TCPTP levels were analysed with Taqman qRT-PCR to study transfection efficiency. TCPTP wt (EFM7+/+) and knockout (EFM4-/-) immortalised mouse embryonic fibroblasts (kindly provided by M. Tremblay) were cultured in DMEM, 10% FBS, 2 mM L-glutamine, 5 ug/ml Plasmocin, 5% CO2.

Serum-starved HeLa cells or mouse embryonic fibroblasts were left untreated on plastic, stimulated with 10 μM spermidine for 1 h, or in addition treated with 50 ng/ml EGF or PDGF for 5 or 15 min. Cells were lysed in Laemmli's sample buffer and resolved on SDS-PAGE gels for western blot analysis. Western blot bands were quantified using digital image analysis and only non-saturated blots were used. The bands to compare in a blot were enclosed with equal size boxes and pixel intensity was quantified using GeneGenius Bioimaging system and Genetools software (Syngene). Automatic background correction was used.

96-well streptavidin plate (Costar) wells were incubated with TBS containing 2.5 μM α1-cytoplasmic tail (biotin-WKIGFFKRPLKKKMEK) or buffer alone at +4°C for 3 h. Non-specific binding was blocked with 2% BSA/TBS-Tween containing 10% FBS overnight at +4°C. Purified recombinant TC45 (cleaved from GST-TC45 as described in [6]) alone or in the presence of indicated concentrations of small molecules, α1-cyt or α2-cyt peptide (α1 cytoplasmic specific sequence lacking the conserved domain RPLKKKMEK or α2 cytoplasmic domain WKLGFFKRKYEMTKNPDEIDETTELSS) was added to the wells and incubated at room temperature for 1 hour. After extensive washing, TC45 protein binding to integrin tail sequences was detected using anti-TCPTP Ab (CF4) and standard horseradish peroxidase -based detection.

In the in vitro phosphatase assays with purified protein, full-length purified TCPTP (TC45, see above) was incubated in phosphatase reaction buffer (25 mM Hepes, 50 mM NaCl, 1 mM DTT) in the presence or absence of synthetic integrin cytoplasmic tail peptides or the novel small molecule TCPTP activators as indicated. The samples were assayed for phosphatase activity in triplicate using DiFMUP (6,8-difluoro-4-methylumbelliferyl phosphate; Molecular Probes, Eugene, OR) as a substrate. Sample with only reaction buffer and DiFMUP (blank) was used as a control.

Sprouting assays from spheroids were performed as previously described [9], and based on the method described earlier [27]. Briefly, HUVECs were divided into round-bottomed, non-treated 96-wells (Greiner Bio-One), 3000 cells per well, in EBM2 medium with 0.25% methyl cellulose (Sigma). After overnight incubation at +37°C 5 spheroids per treatment were pooled and resuspended in spheroid medium (40% FBS, 0,5% methyl cellulose, plain medium). Collagen gel (20 mM Hepes, 1 × DMEM (Sigma), PureCol collagen (Vitrogen)) was added 1:1 to cell suspension and the mix transferred to 48-wells. After polymerisation, medium +/- VEGF 50 ng/ml, 200 nM ScrTAT or α1TAT-peptide, and 10 uM spermidine were added. Cumulative length of the sprouts around spheroids was quantified using 10× magnification after 24 h incubation.

Five small molecule libraries were screened for TCPTP activators. The library compounds were transferred to 384-well assay plates with a Microlab Star automated liquid handling workstation (Hamilton Robotics) fitted with 100-nl 96-channel pintool (V&P Scientific Inc., San Diego, CA). HTS mode TCPTP dispensation was carried out using a Multidrop Combi dispenser (Thermo Fisher Scientific, Waltham, MA), and the plates were incubated for 10 min. at RT. Background fluorescence was measured during the TCPTP incubation. DiFMUP was added using Multidrop and the plates were incubated for 10 min. at RT. Urea was added to stop the reactions and fluorescence was measured with EnVision 2101 multilabel plate reader (Wallac Oy, PerkinElmer Life Sciences and Analytical Sciences, Turku, Finland). Screening data was analysed and the results normalised using the B-score method, which decreases assay and instrumentation specific variation between datapoints (Brideau et al. 2003) implemented in R, a software environment for statistical computing. (R Development Core Team 2008). B-score hits were compared with background measurement results. Hits in wells having high background signal were discarded as false positive.

TCPTP wt and knockout mouse embryonic fibroblasts were applied on 384-wells, 1500 cells in 35 μl medium (DMEM, 10% FBS, 2 mM L-glutamine) per well. Seven different compound dilutions were prepared in DMSO of Spermidine, Mitoxantrone, Ruthenium Red and MDL. Compounds were added in medium to wells, with 6 replicates of each dilution. The resulting end concentrations of the compounds in the wells were 100 nM, 1 μM, 10 μM, 30 μM, 100 μM, 300 μM and 1 mM. Sample containing only DMSO was used as a negative control. Plates were incubated for 72 h at 37°C, 5% CO₂. Cell Titer Blue was added to wells and incubated for a further 4 h after which fluorescence at 560 nm was measured with Envision. Data analysis was performed using GraphPad Prism software (GraphPad Software, USA).

Statistical analyses were performed using the two-tailed Student's t-test. All results are expressed as the mean ± SEM. Following P-values were used to show statistical significance: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

We performed a high-throughput screen to identify small molecule compounds capable of activating TC45. Five small molecule libraries, Microsource Spectrum with 2000 biologically active and structurally diverse compounds, LOPAC1280 with 1280 pharmaceutically active compounds, Tripos with 6000 compounds, ChemDiv with 25000 compounds and ChemBridge with 30000 compounds were screened in an in vitro phosphatase assay. The assay was based on incubation of purified recombinant TC45 with a fluorescent phosphatase substrate in the presence or absence of small molecule compounds. The compounds were printed on 384-well plates. Compound interference was taken into account by measuring background fluorescence from plates which contained the compounds and the reaction mix, but no TC45. TC45 activity assay was initiated by adding purified phosphatase to the wells and allowing dephosphorylation of phosphatase substrate DiFMUP to proceed for 10 minutes. The dephosphorylation reaction was stopped with urea and the fluorescence measured using a multilabel plate reader (Fig. 1A). Due to the autofluorescence of some compounds, the true hits were revealed by comparing the background and assay fluorescence values (Fig. 1B). After removal of false positives, 213 putative TC45 activators were found in the primary screen (Table 1). Interestingly, many potential inhibitors (data points with a negative B-score) of TC45 were also identified in the primary screen, however these were not followed further in this study. A secondary screen confirmed that six of the compounds activated TC45 in a concentration-dependent manner (Fig. 2A). The molecules capable of activating TC45 in vitro were spermidine trihydrochloride (spermidine), mitoxantrone, ruthenium red, MDL-26,630-trihydrochloride (indicated hereafter as MDL), N21 (C₁₅H₁₃N₅) and F12 (C₃₀H₃₈N₄O₂), the chemical structures of which are shown in Fig. 1C. Out of these mitoxantrone was most potent showing on average a 3.3-fold activation of TC45 which was comparable with the 2.4-fold activation by α1-cyt, a known activator of TC45 [6]. (Fig. 2B).

Bild vergrößern

Bild vergrößern

More detailed information on the molecular mechanism governing activation of TC45 is needed to facilitate rational drug design of activating molecules. Thus, we investigated the mechanism behind the small molecule -mediated TC45 activation. Based on our previous results, integrin α1 cytoplasmic tail residues 1164-1179 interact with the amino-terminal part of TC45 activating it by alleviating the proposed autoregulatory interaction between the C- and N-terminus of the protein [6]. Since α1 cytoplasmic tail is positively charged (RPLKKKMEK) and also the majority of the identified small molecular activators of TC45 carry positively charged amine-groups, we investigated whether the compounds would compete with α1-cyt for binding of TC45. Purified recombinant TC45 interacts directly with biotinylated α1-cyt peptide and the interaction is sensitive to competition with 11 μM α1-peptide but not α2-peptide (Fig. 2C). Interestingly, spermidine and mitoxantrone competed with the solid-phase bound α1-cyt in a concentration-dependent manner (Fig. 2D), suggesting that these compounds unlike MDL and Ruthenium Red activate TCPTP by binding to the same site as the integrin cytoplasmic tail.

The majority of TCPTP activators identified here are previously described molecules with known cellular targets other than TC45. Therefore, it was important to characterize whether they influence cell behaviour in a TC45-dependent manner. TCPTP is a known negative regulator of many mitogenic signalling pathways and acute silencing of TCPTP induces cell proliferation in cancer cells and alterations in signalling pathways [9, 10, 28, 29]. To study the TC45-dependency of these compounds, we tested four of the compounds for their effect on cell proliferation in TC45 knockout and wt mouse embryonic fibroblasts. The cells were incubated with the compounds at indicated concentrations for 72 h and live cells were detected. Strikingly, spermidine displayed specificity towards TC45, since TC45 knockout cells were 43-fold more resistant to the compound than wt cells, suggesting that the presence of TCPTP makes the wt cells more sensitive to the drug regarding proliferation (Fig. 3). Since spermidine can be oxidized in serum-containing culture [30], we cannot rule out the possibility that the spermidine effect is in fact a function of a spermidine-derived compound. The other three compounds did not show specificity towards TCPTP and either inhibited cell proliferation equally well in both cell types (ruthenium red and mitoxantrone) or had no significant effect on proliferation (MDL) at the investigated concentrations.

Bild vergrößern

TC45 negatively and site-selectively regulates PDGFRβ phosphorylation [10]. To investigate whether spermidine could attenuate PDGFRβ signalling in a TC45-dependent manner, we used TC45 knockout and wt mouse embryonic fibroblasts. We studied the phosphorylation levels of PDGFRβ Tyr 1021 in spermidine-treated cells in the presence or absence of PDGF-BB, after overnight starvation. In line with previous data [10], we observed that PDGF-BB induced 45 ± 4% higher phosphorylation of PDGFRβ in TC45 knockout cells compared to wt cells (Fig. 4A, B). Importantly, spermidine was able to attenuate PDGF-induced PDGFRβ phosphorylation by 42 ± 6% in TC45 wt cells whereas no inhibition was detected in TC45 negative cells (Fig. 4A, C). Therefore, activation of TC45 with spermidine results in TC45-dependent negative regulation of a well-characterized TC45 target in cells.

Bild vergrößern

Previously we have shown that collagen-induced activation of TC45 by integrin α1β1 attenuates EGFR phosphorylation at multiple sites [6]. Since our results with spermidine implied that it can function as a TC45 activator in mouse cells, we tested if spermidine would have an effect on EGFR phosphorylation in human cancer (HeLa) cells. Treatment of the cells with spermidine was sufficient to significantly attenuate EGFR phosphorylation in cells (Fig. 4D, E). The observed attenuation was similar to the one achieved with TC45 activation by cell-membrane permeable TAT-α1-cyt peptide in this cell line [6]. Taken together, these data indicate that spermidine-induced activation of TC45 can be used to inhibit signalling of TC45 target RTKs like PDGFRβ and EGFR in human and mouse cells.

Recently we showed that VEGFR2 is under the negative regulation of TC45 and α1-integrin in endothelial cells (Mattila et al. 2008). We demonstrated that activation of TC45 with cell-membrane permeable α1-TAT peptide significantly reduced the length of the VEGF-induced sprouts in three-dimensional cultures of HUVECs in a TC45-dependent manner [9]. Here we tested the effect of spermidine on VEGF-induced sprout formation in HUVEC spheroids (Korff, Augustin 1999). Also in this model spermidine and α1-TAT but not scrTAT (a control peptide with TAT-sequence fused to a scramble sequence containing the α1 amino acids in random order) were able to significantly inhibit VEGF-induced sprouting in HUVECs (Fig. 5A, B). The basal sprouting detected in the absence of VEGF was not affected by spermidine (Fig. 5A, B). Importantly, the ability of spermidine to inhibit VEGF-induced endothelial sprouting was at least partially dependent on TC45. We used a previously characterized TC45-specific siRNA that shows identical effects to another independent siRNA oligo and does not influence levels of other PTPs like SHP-2 [6, 9]. Silencing of TC45 (Fig. 5D) increased VEGF-induced sprouting in TC45-silenced cells compared to control cells by 35 ± 7% (Fig. 5C). Furthermore, in TC45-silenced cells spermidine inhibited VEGF-induced sprouting by 43 ± 5% compared to the 89 ± 3% inhibition by spermidine in control cells. The somewhat limited effects induced by the TCPTP siRNA in these experiments are most likely due to moderate TCPTP silencing achieved in HUVEC (Fig. 5D). These results demonstrate that in human primary endothelial cells VEGF-induced sprouting is inhibited by spermidine and that TC45 expression is required for efficient spermidine-mediated inhibition of VEGF signalling.

Bild vergrößern