Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The splicing FK506-binding protein-51 isoform plays a role in glioblastoma resistance through programmed cell death ligand-1 expression regulation

The splicing FK506-binding protein-51 isoform plays a role in glioblastoma resistance through... Gliomas aberrantly express programmed cell death ligand-1 (PD-L1), which has a pivotal role in immunoevasion. The splicing isoform of FKBP5, termed FKBP51s, is a PD-L1 foldase, assisting the immune checkpoint molecule in maturation and expression on the plasma membrane. The concept that PD-L1 supports tumor-intrinsic properties is increasingly emerging. The aim of the present work was to confirm the pro-tumoral effect of PD-L1 on human glioma cell survival, stemness capacity and resistance, and to address the issue of whether, by targeting its foldase either chemically or by silencing, the aggressive tumor features could be attenuated. PD-L1-depleted glioma cells have a reduced threshold for apoptosis, while PD-L1 forced expression increases resistance. Similar results were obtained with FKBP51s modulation. The ability of PD-L1 to counteract cell death was hampered by FKBP51s silencing. PD-L1 expression was particularly high in glioma cells with a cancer-stem-cell profile. Moreover, PD-L1 sustained the spheroid formation capability of glioma cells. Targeting of FKBP51s by small-interfering RNA (siRNA) or the specific inhibitor SAFit2, reduced the number of formed spheroids, along with PD-L1 expression. Finally, in an orthotopic mouse model of glioblastoma, daily treatment with SAFit2 significantly reduced tumor PD-L1 expression, and tumor growth. In treated mice, caspase-3 activation and reduced vimentin expression were observed in excised tumors. In conclusion, targeting of FKBP51s hampers PD-L1 and its pro-tumoral properties, thereby affecting the self-renewal and growth capacities of glioblastoma cells in vitro and in vivo. Introduction primary brain cancer, for which no contemporary treat- Glioblastoma multiforme (GBM) is the most common, ments are curative. Gliomas, indeed, express several co- but also the most dangerous and aggressive form of inhibitory molecules that negatively regulate immune 1–4 system functions , among which programmed cell death ligand-1 (PD-L1) is one of the most represented and 3,4 Correspondence: Bernard Rogister (Bernard.Rogister@uliege.be) or Maria studied . The interaction of PD-L1 with its counter- Fiammetta Romano (mariafiammetta.romano@unina.it) receptor programmed cell death-1 (PD-1) expressed by Dipartimento di Medicina Molecolare e Biotecnologie Mediche, Università di activated T lymphocytes, inhibits T-cell activation and Napoli Federico II, Napoli, Italy GIGA-Neurosciences, Faculté de Médecine, Liège Université de Liège, Liège, proliferation, thus contributing to cancer immune eva- Belgium sion . So far, anti-PD-1 treatment has shown a therapeutic Full list of author information is available at the end of the article. efficacy limited to selected subsets of glioma patients . These authors contributed equally: Paolo D’Arrigo, Marina Digregorio Edited by A. Rufini © The Author(s) 2019 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Official journal of the Cell Death Differentiation Association 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 2 of 12 Several studies are in progress, especially exploring the function, i.e., SAFit2, which produced similar results to combination of anti-PD-1 antibodies with chemotherapies those found in FKBP51s silencing . or targeted therapies . It is worth noting that PD-L1 not The present study confirms that PD-L1 exerts impor- only binds to PD-1 but also competes with CD28-to- tant tumor-intrinsic properties in GBM. In particular, we CD80 binding , suggesting that a PD-1 blockade may show that PD-L1 sustains cell survival, resistance and sometimes be insufficient to prevent signals emanating stemness capability. PD-L1 expression was highest in from PD-L1 receptors. GBM cancer-initiating cells, due to a post-transcriptional Recent studies show that, in addition to its immuno- regulatory effect involving FKBP51s. Targeting of modulatory function, PD-L1 conveys a survival signal to FKBP51s by gene silencing or via the selective inhibitor the cancer cell and sustains tumor aggressiveness . SAFit2, downmodulated PD-L1 expression and inhibited Azuma et al. found that PD-L1, upon ligand stimulation, spheroid formation when GBM cancer-initiating cells acts as a receptor that transduces pro-survival signals that were cultured under non-adherent conditions. In an enhance the threshold for apoptosis induced by both orthotopic GBM mouse model, SAFit2 showed an anti- immune effectors and proapoptotic drugs. Zheng et al. tumor effect, as assessed by reductions in tumor volumes, found that PD-L1 sustains the self-renewal capability of caspase activation and attenuated expression levels of PD- malignant melanoma-initiating cells. Most recently, Qiu L1 and the mesenchymal marker vimentin. et al. demonstrated that PD-L1 binds to Ras and acti- vates the Ras/Erk pathway. Via this mechanism, PD-L1 Results promotes the epithelial-to-mesenchymal transition PD-L1 promotes apoptosis resistance (EMT) and GBM cell migration. We investigated the effect of PD-L1 silencing on GBM Akt supports aberrant PD-L1 expression in glioma cells, cell survival. To this end, we used two human GBM cell through a translational mechanism favouring the assem- lines previously found to highly express PD-L1 and bly of a polyribosome complex that allows better entrance FKBP51s, namely, D54MG and U251MG cells . For PD- of the PD-L1 transcript into polysomes . PD-L1 glyco- L1 downmodulation, cells were treated with specific siR- sylation is an essential element influencing protein stabi- NAs targeting PD-L1 or its co-chaperone FKBP51s. lity . Most recently, in glioma, the splicing isoform of the Twenty-four hours after transfection, some of the cells FK506-binding protein-51 (FKBP51s) was found to were harvested for lysate preparation. After a further 24 h, serve as a PD-L1 co-chaperone, assisting in protein gly- the remaining cells were collected for cell-death mea- cosylation surements via PI incorporation. Figure 1a shows a western . PD-L1 post-translational modifications have been thoroughly reviewed by Hsu et al. . According to blot assay of lysates obtained from human D54MG cells Jiang et al. and the Oncomine database (www.oncomine. treated with three different FKBP51s siRNAs and a spe- org), FKBP51 is among the top 10% of the most highly cific PD-L1 siRNA (siPD-L1). Two of the three siRNAs UTR1 expressed genes in GBM. The FKBP51 protein structure were designed on the 3′-UTR (siFKBP51s and UTR2 includes a C-terminal TPR three-tandem-repeat domain siFKBP51s ) and another (siFKBP51) on the coding responsible for protein-protein interaction, and two N- region. The PD-L1 signals at ≈50 kDa are those of mature terminal FK domains, of which the one with most N (glycosylated) forms and those under 37 kDa correspond terminals exhibits peptidyl-prolyl cis-trans isomerase to the naive protein (Fig. 1a). SiFKBP51s and 17,18 UTR1 (PPIase) activity . FKBP51s is generated by alternative siFKBP51s appeared to downmodulate FKBP51s 13 UTR2 splicing of FKBP5 pre-mRNA , which causes a frameshift more efficiently than siFKBP51s . Expression of PD- UTR1 with a premature stop codon, leading to a distinct C- L1 was also decreased by siFKBP51s and siFKBP51s , terminus, compared to the canonical isoform. FKBP51s in comparison to the control cells (NSRNA or none). The retains the PPIase activity but loses the TPR domain. An procaspase-7 level was decreased by the same siRNAs, IHC study on 29 GBM specimens showed that FKBP51s is with a cleaved fragment at ≈20 kDa also observable, broadly expressed in this tumor, albeit with different consistent with apoptosis activation (Fig. 1a). Measure- proportion/intensity scores, with nuclear and/or cyto- ment of hypodiploid cells confirmed that PD-L1 down- plasmic localization . Biochemical evaluation of glioma modulation, like FKBP51s silencing, produced cell death cell lines showed that naive PD-L1 is complexed with (Fig. 1a). The effect of different siRNAs on both FKBP51s FKBP51s in the endoplasmic reticulum, whereas the gly- and PD-L1, was also assessed in U251MG, as shown in the cosylated form was detected in the Golgi apparatus . Supplementary Information (Fig. S1). Since siPD-L1 and FKBP51s knockdown severely reduced the level of gly- siFKBP51s appeared to be the most effective for target cosylated PD-L1, whether constitutively expressed or downmodulation, these siRNAs were used in subsequent induced by ionizing radiation . The essential role of the experiments. Human U251MG cells showed similar PPIase activity in PD-L1 protein maturation was con- results to those obtained with D54MG cells (Fig. 1b). Both firmed by use of a selective inhibitor of this catalytic PD-L1 and FKBP51s silencing decreased PD-L1 Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 3 of 12 Fig. 1 PD-L1 regulates glioma cell apoptosis. a Analysis by western blot assay of PD-L1 and FKBP51s expression levels in D54MG cells, silenced for UTR1 UTR2 FKBP51s, using different siRNAs (siFKBP51s , siFKBP51s , and siFKBP51s) or PD-L1. The blot also shows caspase-7 levels recognized in its inactive (procaspase) and active forms. On the bottom of the panel, means and standard deviations of cell-death values of D54MG cells silenced for FKBP51s and PD-L1, are shown. Columns indicate the percentage of hypodiploid cells (N = 4). b Western blot assay of PD-L1 and FKBP51s levels in U251MG cells silenced for FKBP51s and PD-L1. On the bottom of the panel, means and standard deviations of cell-death values of U251MG cells silenced for FKBP51s and PD-L1, are shown. Columns indicate the percentage of hypodiploid cells (N = 4). c Graphical representation of means and standard deviations of cell-death values in PD-L1-depleted D54MG and U251MG cells. Twenty-four hours after transfection with PD-L1 siRNA, cells were treated with etoposide. They were then harvested after a further 24 h and analyzed for PI incorporation via flow cytometry. Columns indicate the percentage of hypodiploid cells (N = 3). d Flow-cytometric histograms of PD-L1 expression in D54MG and U251MG cells, cultured for 6 h with etoposide. On the right, a graphical representation is given of values (mean and standard deviation) from three different experiments. e Western blot assay of PD-L1 levels in GBM cells cultured with etoposide. f Measurement by qPCR of FKBP51s transcript level in cells treated or not treated with etoposide, for 6 h (N = 4). g Western blot assay of exogenous PD-L1 levels in D54MG and U251MG cells cultured in the absence or the presence of etoposide. h Graphical representation of means and standard deviations of cell-death values in PD-L1 D54MG and U251MG glioma cells. Twenty-four hours after transfection with EV or PD-L1-GFP, cells were treated with etoposide. They were then harvested after a further 24 h and analyzed for PI incorporation via flow cytometry. Columns indicate the percentage of hypodiploid cells. Each experiment was performed at least three times and in triplicate expression levels, but only FKBP51s siRNA decreased the of the two factors appeared to further increase cell death, FKBP51s expression level. Activation of caspase-3 was in comparison with the single treatment. This result registered in U251MG cells using flow cytometry (Sup- suggested that reduced levels of PD-L1 could act in plementary Information, Fig. S2). We, then, investigated concert with the chemotherapeutic compound to enhance the effect of PD-L1 silencing on etoposide-induced cell its cytotoxicity (Fig. 1c). Using flow cytometry, we found death. Silencing of PD-L1 appeared to exert a cytotoxic that both cell lines, when cultured with etoposide for 6 h, effect similar to that of etoposide. However, combination had increased levels of PD-L1, compared to the same Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 4 of 12 untreated cells (Fig. 1d). As expected, western blot ana- FKBP51s hampered the pro-survival effect of ectopic PD- lysis confirmed the increase in the mature PD-L1 signals L1 (Fig. 2d). The cell-death modulatory effect of FKBP51s at 50 kDa (Fig. 1e) and showed an increased expression of was also assessed in U87MG cell lines (Supplementary FKBP51s in etoposide-treated cells. These results sug- Information, Fig. S5), as well as using SAFit2 (Supple- gested that etoposide induced FKBP5 mRNA splicing, mentary Information, Fig. S6). which was confirmed at the transcription level (Fig. 1f). Ectopic expression of PD-L1 (Fig. 1g), significantly PD-L1 expression increases with glioma aggressiveness reduced etoposide-induced cell death (Fig. 1h). Taken and affects spheroid formation in culture together, these results suggest that PD-L1 is a resistance The role of PD-L1 in the apoptosis resistance of GBM factor for GBM cells. Further confirmation of the PD-L1 cells, together with its previous involvement in EMT , pro-survival effect was obtained with the other two GBM supports the concept that this immune checkpoint is a cell lines, namely, U87MG and SF767 (see Supplementary hallmark of GBM malignancy. To address whether PD-L1 Information, Figs. S3 and S4 respectively). Interestingly, in expression increased with the aggressiveness of GBM SF767 cells, which show deficient PD-L1 levels , PD-L1 cells, we used two GBM cell lines previously established silencing did not produce an apoptosis-enhancing effect, by Kroonen et al. from the tumor mass (TM-GBM) and whereas PD-L1 ectopic expression reduced etoposide the subventricular zone (SVZ-GBM) of an orthotopic cytotoxicity (Supplementary Information, Fig. S4). GBM model. TM-GBM and SVZ-GBM cell lines, char- acterized by an increasing level of aggressiveness from FKBP51s promotes apoptosis resistance TM to SVZ, were derived from the human U87MG cell Since FKBP51s deprivation reduces PD-L1 expression , line. SVZ-GBM cells are particularly rich in GBM cancer we investigated whether FKBP51s affected GBM cell stem cells and have a high tumorigenic potential, as death using RNA silencing. The effect of protein down- assessed by mouse reinjection or by their capacity to expression was assessed using western blot analysis (Fig. 2a). grow in spheres when cultured under non-adherent Cell death was assessed in the absence or the presence of conditions . We confirmed the high tumorigenicity of etoposide, via PI incorporation. Figure 2a shows repre- SVZ-GBM cells showing increased levels of p-Akt, along sentative flow-cytometric histograms of hypodiploid with its substrate p-S6K1 (Fig. 3a) and the stemness events and a graphic representation of the means and marker CD133 (Fig. 3b), in comparison to TM-GBM cells. standard deviations of cell-death values obtained in three SVZ-GBM also showed increased PD-L1 expression in independent experiments, each performed in triplicate. contrast to TM-GBM cells, as assessed by western blot FKBP51s silencing in D54MG cells produced 20.0 ± 4.0% analysis (Fig. 3b) and flow cytometry (Fig. 3c). Surpris- cell death, while treatment of cells with a non-silencing ingly, PD-L1 mRNA levels appeared to be higher in the oligo (NSRNA) produced only 2.3 ± 0.5% cell death (p = TM-GBM cells than in the SVZ-GBM cells (Fig. 3d). In 0.01) (Fig. 2a). Similarly, in etoposide-treated cells, line with Goffart et al. , we found that SVZ-GBM cells FKBP51s siRNA and NSRNA produced 28.0 ± 4.0% and have an increased capacity to form spheroids in com- 20.0 ± 0.1% cell death respectively (p = 0.01) (Fig. 2a). parison with TM-GBM cells (Fig. 3e). As expected, FKBP51s-silenced U251MG produced an average of spheroids (+) expressed Sox-2 levels remarkably higher 14.3 ± 4.0% cell death, while NSRNA produced only 3.3 ± than those in the corresponding adherent cells (-) 0.5% cell death (p = 0.03) (Fig. 2a). In etoposide cultures, (Fig. 3e). Expression of PD-L1 resulted especially in an FKBP51s siRNA and NSRNA produced 55.0 ± 4.1% and increase in spheroids formed by TM-GBM cells, in 46.7 ± 1.1% cell death respectively (Fig. 2a) (p = 0.05). To comparison with attached TM-GBM cells. Although only confirm the role of FKBP51s in the resistance of GBM a slight increase in PD-L1 expression was observed in cells, we overexpressed FKBP51s and analyzed cell death SVZ-GBM cells growing in spheres (Fig. 3f), the finding using EV cells for comparison. In line with the FKBP51s that PD-L1 silencing impaired the capacity to form knockdown sensitizing effect, we registered a significant spheroids in both TM- and SVZ-GBM cells (Fig. 3g) decrease in etoposide-induced cell death in D54MG and highlighted the importance of PD-L1 in spheroid forma- U251MG cells transfected with FKBP51s vector, com- tion. Taken together, our results indicate that PD-L1 is a pared with EV cells (p = 0.03 and p = 0.004 for D54MG necessary but not a sufficient condition for allowing the and U251MG, respectively) (Fig. 2b). To investigate GBM cells to grow in spheres. whether PD-L1 requires FKBP51s to counteract chemotherapy-induced cell death, we overexpressed PD- Targeting of FKBP51s downregulates PD-L1 and hampers L1 under conditions of FKBP51s depletion. To this end, GBM malignancy we co-transfected PD-L1 and FKBP51s siRNA or NSRNA, Since the increased expression of PD-L1 in SVZ-GBM in D54MG and U251MG cells (Fig. 2c), and measured cell cells, in comparison to TM-GBM cells, appeared to be death in etoposide cultures (Fig. 2d). The silencing of related to a post-transcriptional mechanism, we investigated Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 5 of 12 Fig. 2 FKBP51s regulates glioma cell death. a Western blot assay of FKBP51s levels in D54MG and in U251MG silenced for FKBP51s. Cells were treated with FKBP51s siRNA or NSRNA for 24 h. Then, some of the cells were harvested for lysate preparation and some cells were treated with etoposide. After a further 24 h, cell death was measured. Representative flow-cytometric histograms of PI incorporation are shown along with graphical representations of means and standard deviations of cell-death values (N = 3). b Western blot assays of exogenous FKBP51s levels and graphical representations of means and standard deviations of cell-death values in D54MG and U251MG cells transfected with FKBP51s with EV- or FKBP51s-carrying plasmids (N = 3). Cells were treated with etoposide at 24 h from transfection. After a further 24 h, cell death was measured by flow cytometry (N = 3). c Western blot assays of the exogenous PD-L1-GFP level in D54MG and U251MG cells, silenced or not silenced for FKBP51s. Silencing of FKBP51s produced a decrease in PD-L1-GFP level in both cell lines. d D54MG and U251MG cells transfected with EV, PD-L1-GFP+ NSRNA and PD-L1-GFP+ FKBP51s siRNA, were treated with etoposide. After 24 h, cell death was measured using flow cytometry. FKBP51s depletion reduced the antiapoptotic effect of exogenous PD-L1. Each experiment was performed at least three times and in triplicate the effect of FKBP51s silencing on the modulation of PD-L1 the effect of ectopic PD-L1-GFP in promoting spheroid levels in both TM-GBM and SVZ-GBM cells. As shown in formation (Fig. 4d, e). A western blot validated the impact the western blot in Fig. 4a, the silencing of FKBP51s of FKBP51s silencing on protein levels (Fig. 4f). We then downmodulated PD-L1 levels in both cell lines. This result investigated the effect of SAFit2 on PD-L1 expression and was confirmed using flow cytometry (Fig. 4b). FKBP51s spheroid formation in TM-GBM and SVZ-GBM cells. PD- downmodulation impaired spheroid formation in both TM- L1 expression, as assessed by flow cytometry, showed a GBM and SVZ-GBM cells (Fig. 4c). Interestingly, SVZ- significant decrease on the plasma membrane in SAFit2 GBM appeared to be more sensitive than TM-GBM to the cultures (Fig. 5a). Like FKBP51s siRNA, SAFit2 impaired targeting of FKBP51s, according to the PD-L1 down- spheroid formation in both TM-GBM and SVZ-GBM cells expression, which was more powerful in SVZ-GBM than in (Fig. 5b). To address whether SAFit2 could affect tumor TM-GBM cells (Fig. 4b). FKBP51s silencing counteracted growth, we assessed Ki67 expression and cell counts in TM- Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 6 of 12 Fig. 3 PD-L1 sustains GBM capacity to form spheroids. a Immunoblot of p-Akt and p-S6K1 in TM-GBM and SVZ-GBM cells. Levels of phospho- enzymes were higher in SVZ-GBM cells than in TM-GBM cells. b Expression of PD-L1 and CD133 in TM-GBM and SVZ-GBM cells. Levels of both proteins were higher in SVZ-GBM cells than in TM-GBM cells. c Flow cytometry measurements of PD-L1 expression in TM-GBM and SVZ-GBM cells. Graph columns represent mean fluorescence intensities with related p-values (N = 3). A representative histogram is shown in overlay. d RT-qPCR results for mRNA levels of PD-L1 in TM-GBM and SVZ-GBM cells. Graph columns represent normalized quantification relative to TM-GBM cells (N = 3). e Representative images of formed spheroids taken using an optical microscope and spheroid counts in TM-GBM and SVZ-GBM cultures (scale bar represents 100 μm). Graph columns represent spheroid numbers relative to TM-GBM cells (N = 3). A western blot of Sox-2 levels shows increased levels of the stemness marker in the spheroids (+), in comparison with adherent cells (−). f Quantification using flow cytometry of PD-L1 levels, in TM-GBM-formed spheroids and SVZ-GBM-formed spheroids. Graph columns represent MFI (N = 3). g Spheroid formation assay with TM-GBM and SVZ-GBM cells silenced or not silenced for PD-L1. Graph columns represent the relative spheroid numbers after a 4-day culture. NSRNA-treated cells were used as the reference sample. Each experiment was performed at least three times and in triplicate GBM and SVZ-GBM cell cultures (Fig. 5c, d). On SAFit2 PD-L1 expression (Fig. 6d). Finally, in-brain slices, the treatment, both cell lines showed a significant decrease in percentage of tumor cells found to be positive for the Ki67 expression (Fig. 5c) and cell counts (Fig. 5d). To expression of the mesenchymal marker vimentin, was still address the in vivo effect of SAFit2, U87MG cells (carrying reduced in mice treated with SAFit2 (Supplementary GFP and the luciferase reporter gene) were injected into the Information, Fig. S6). right striatum of 32 nude mouse brains. Two weeks later, 16 mice were treated daily with SAFit2, while 16 mice were Discussion treated intraperitoneally with a vehicle designed to pass A growing body of literature supports the concept that through the brain-blood barrier, via intraperitoneal injec- PD-L1 has intrinsic pro-tumoral aspects that are inde- tion. After 2 weeks of treatment (corresponding to 4 weeks pendent of its immunoinhibitory functions . Moreover, from cell xenotransplantation) the mice were sacrificed, and several recent studies report that expression levels of PD- the brains obtained. A reduced tumor mass was calculated L1 positively correlate with glioma grades. More precisely, in mice treated with SAFit2 compared to tumors from mice in GBM, these levels are much higher than in grade II and treated with the vehicle (Fig. 6a). The result was confirmed grade III gliomas . by a reduced luminescence intensity recorded in the living In this paper, consistent with the finding of Azuma mice before the sacrifice (Fig. 6b). The reduced tumor et al. that PD-L1 is a ubiquitous antiapoptotic receptor, growth was accompanied by the presence of active caspase- we show that PD-L1 is an essential element in GBM 3 in xenografts from the SAFit2-treated mice (Fig. 6c). In resistance to cell death. A previous paper also reported accordance with in vitro findings, SAFit2 reduced tumor that PD-L1 overexpression, as a result of the low miR-34a Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 7 of 12 Fig. 4 FKBP51s regulates PD-L1 expression and spheroid formation in TM-GBM and SVZ-GBM cells. a Immunoblot of PD-L1 and FKBP51s levels in cells silenced for FKBP51s. Canonical FKBP51 confirmed the specificity of the silencing. b Flow cytometry analysis of PD-L1 expression in cells, silenced or not silenced for FKBP51s. Graph columns represent MFI using non-silenced cells as the reference sample (N = 3). c Spheroid assay with cells from TM-GBM and SVZ-GBM cells, silenced or not silenced for FKBP51s. Representative images of formed spheroids are shown. Graph columns represent the relative spheroid numbers, after a 4-day culture, using NSRNA-treated cells as the reference sample (N = 3). d Spheroid assay with TM-GBM cells and SVZ-GBM cells transfected with EV, PD-L1-GFP+ NSRNA, and PD-L1-GFP+ FKBP51s siRNA. FKBP51s depletion reduced the spheroid formation stimulated by exogenous PD-L1 (scale bar represents 250μm) (N = 3). e Fluorescent microscopy visualization of ectopic PD-L1-GFP in the spheroid assay (scale bar represents 50 μm). FKBP51s depletion reduced spheroid numbers and the extent of fluorescence. f Western blot assay of FKBP51s levels in FKBP51s-silenced spheroids. Each experiment was performed at least three times and in triplicate 21 14 level in U87 glioma cells, induced paclitaxel resistance . this co-chaperone in PD-L1 expression . FKBP51s In the present paper, we confirm the resistance effect of appeared to significantly affect the spheroid formation PD-L1 in four different glioma cell lines, namely, U87MG, ability of GBM cancer-initiating cells, along with PD-L1. D54MG, U251MG and SF767. Moreover, in accordance Collectively, these findings, in addition to supporting the with Zheng et al. , who demonstrated a role for PD-L1 in pro-oncogenic role of PD-L1 in glioma, open up the sustaining the self-renewal capability of malignant possibility of targeting PD-L1 by acting on its co- melanoma-initiating cells, we show that PD-L1 expression chaperone FKBP51s. SAfit2 is, so far, the most well- is increased in GBM cancer-initiating cells and has a characterized FKBP51 ligand available . After 2 weeks of supporting role in spheroid formation. Finally, in accor- intraperitoneal treatment with SAFit2, assessment of dance with Qiu et al. , who demonstrated the promoting GFP + luc + tumor volumes in-brain slices showed a role of PD-L1 in GBM-cell migration, we found increased significant decrease in SAFit2-treated mice, in compar- expression of PD-L1 in GBM cells that had migrated to ison with vehicle-treated control mice. Consistent with the subventricular zone of the brain, compared with PD- this finding, reduced luminescence intensity was regis- L1 expression in the nigrostriatal tumor mass. When the tered in SAFit2-treated tumors in living mice before levels of PD-L1 were downmodulated by FKBP51s silen- sacrifice. The presence of active caspase-3 in tumor cing, GBM-cell sensitivity to both spontaneous and xenografts from SAFit2-treated animals suggested that chemotherapy-induced death was increased. Moreover, the reduction in tumor volumes can be ascribed, at least when the levels of PD-L1 were upregulated by an ectopic in part, to apoptosis induction. A decreased percentage of expression, a pro-survival effect on GBM cells was tumor cells expressing PD-L1 and vimentin, according to observed. This outcome was counteracted by FKBP51s histological examinations, suggested that in treated mice silencing, which is consistent with the essential role for the tumor had a reduced aggressiveness. Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 8 of 12 Fig. 5 SAFit2 decreases PD-L1 expression and spheroid formation and impairs in vitro growth of TM-GBM and SVZ-GBM cells. a Cytometry analysis of PD-L1 expression in cells treated with SAFit2 for 12 h. Graph columns represent mean fluorescence intensities using cells with Dimethyl sulfoxide (DMSO) as the reference sample (N = 3). b Spheroid assay with cells treated with SAFit2 or the vehicle. Representative pictures of formed spheroids are shown along graph columns for the relative spheroid number measured after a 4-day culture. DMSO-treated cells were used as the reference sample (N = 3). c Ki67 expression measured by immunofluorescence in cells treated with SAFit2 or the vehicle (scale bar represents 25 μm). Graph columns represent the percentage of Ki67-positive cells, relative to the control (N = 6). d Kinetics of counts of TM-GBM and SVZ-GBM cells treated with SAFit2 or the vehicle. SAFit2 significantly reduced cell counts (N = 4) In conclusion, in line with a previous study demon- DDK-tagged and a True-ORF-GFP-tagged expression vec- strating that in vivo knockdown of PD-L1 in nude mice tors were used (OriGene Technologies, Rockville, MD, completely abolished GBM xenograft formation USA), which carried the complementary DNA (cDNA) of , our finding shows that manipulation of PD-L1, through che- the human FKBP5 transcript variant 4 and CD274, mical inhibition of its co-chaperone FKBP51s, activates respectively. Control cells were transfected with the related apoptosis and mitigates the aggressive features of GBM empty vector (EV). FKBP51s silencing was performed using xenografts in nude mice. Our study contributes some short-interfering oligoribonucleotides. The siFKBP51 was a information to the emerging concept that there is a strict mix of two siRNAs designed on the coding region; UTR1 UTR2 interplay between cancer immunoediting and resistance siFKBP51s and FKBP51s were designed on 3′- and suggests that SAFit2 may be a beneficial neoadjuvant UTR . The non-silencing RNA (NSRNA) and PD-L1 strategy for the management of this tumor. siRNA were from Qiagen (siPD-L1) (Valencia, CA, USA) or Novus Biologicals (siPD-L1.2) (Littleton, CO). U87MG cells Methods and materials used for xenograft implantation were previously transduced Cell culture and transfection with lentiviral vectors allowing the dual expression of eGFP Human glioma cell lines D54MG, U251MG, SF767MG and luciferase as previously described . In experiments were obtained and cultured as described .U87MG from with etoposide, the drug was added 24 h after transfection at American Type Culture Collection (ATCC) were cultured aconcentration of 20 μM. For experiments with SAFit2 , in Dulbecco’sModified Eagle Medium (DMEM) containing which was provided by Prof. Felix Hausch laboratory, a 10% FBS, 200 mM glutamine, and 100 U/mL stock solution (60 μM in DMSO) wasusedat1:1000dilu- penicillin–streptomycin. TM-GBM cells and SVZ-GBM tion in DMEM-F12 media; for control cells, DMSO has cells were U87MG cells obtained by microdissection of diluted accordingly. xenografted mice brain , from, respectively, the tumor mass and the subventricular zone, and seeded in cultures . Western blot For FKBP51s and PD-L1 knockin and knockdown, cells Whole-cell lysates were homogenized in modified RIPA were transfected using the K2 Transfection System (Bion- buffer and assayed by immunoblot as previously 14 14 tex, Munich, Germany), as previously described and in described . The primary antibodies against FKBP51 accordance with the manufacturer’s recommendations. For (rabbit polyclonal; Novus Biological), FKBP51s and overexpressing FKBP51s and PD-L1 a True-ORF-Myc- CD274/PD-L1 (rabbit polyclonal; Novus Biological) were Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 9 of 12 Fig. 6 SAFit2 impairs GBM growth in vivo. a Graphic representation of tumor volumes formed in 16 mice treated daily with vehicle and 16 mice treated with SAFit2. b Luciferase assay, performed on mice treated or not treated with SAFit2, 2 weeks after in-brain injection with GFP+ luc+ U87MG cells. Three mice from each group are shown before starting the treatment and after 10 days (left). Graph columns represent mean values of signal intensities from the two groups, before and after treatment (right) (N = 16). c Expression of active caspase-3 by immunofluorescence on brain slices (scale bar represents 50 μm). Box plots represent the values of GFP+ cleaved Casp3+ cells measured using ImageJ software, from seven mice treated with vehicle and seven mice treated with SAFit2. d Expression of PD-L1 determined by immunofluorescence on brain slices (scale bar represents 50 μm). Box plots represent the values of GFP+ PD-L1+ cells measured using ImageJ software, from nine mice treated with vehicle and nine mice treated with SAFit2 used diluted 1:2500. CD133 (rabbit polyclonal; Abcam; 1:5000. Anti β-Actin-Peroxidase (mouse monoclonal; Cambridge, UK) was used diluted 1:1000. A further Sigma-Adrich) was used diluted 1:10000. Anti-phospho- antibody Pdcd-1L1 (rabbit polyclonal, Santa Cruz Bio- Akt (Ser473), Akt1/2/3, phosphor-S6 kinase, G3PDH technology; Santa Cruz, CA, USA) was used for PD-L1 and Sox-2 (rabbit monoclonal; Cell Signaling, Danvers, detection at the 1:1000 dilution. Antibody against M2- MA, USA) were used diluted 1:1000. Anti-p70S6 Flag, caspase-7 and γ-Tubulin (mouse monoclonal; kinase (rabbit polyclonal; Santa Cruz) was used Sigma-Aldrich, St. Louis, MO, USA) were used diluted diluted 1:500. Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 10 of 12 Flow cytometry contrast, and color balance of the images were adjusted in Expression of PD-L1 was assessed using anti-B7H1- Photoshop CS2 (Adobe Systems, San Jose, California, phycoerythrin (PE) (R&D Systems, Minneapolis, MN, USA). This adjustment was applied to every pixel in USA) at a concentration of 0.05 μg/ml. A PE-conjugated each image. control Ig isotype was used to assess non-specific binding. Briefly, cells were harvested and incubated with the Quantitative PCR (qPCR) antibodies mentioned above for 30 min in the dark at 4 °C, Total RNA was extracted from cells by Trizol (Invitro- washed and then analyzed with a BD Accuri™ C6 Cyt- gen, Carlsbad, CA, USA). Each RNA was used for cDNA ometer (BD Biosciences, New Jersey, USA). Cleaved synthesis with iScript Reverse Transcription (Bio-Rad, caspase-3 immunostaining was performed as previously CA, USA). Relative gene expression was quantified by 23 27 described . Apoptosis was assessed by two methods: qPCR with 2–ΔCt comparative method , using the propidium iodide (PI) incorporation to analyze the DNA SsoAdvancedTM SYBR Green Supermix (Bio-Rad) and content in permeabilized cells , and the annexin V- specific qPCR primers. Amplification of FKBP51s was binding assay in double staining with PI, according to performed as previously described . Oligo sequences Vermes et al. . Cells were analyzed with a BD Accuri™ C6 used for PD-L1 mRNA quantitation along with co- Cytometer (BD, Becton Dickinson). amplified housekeeping genes are: Fw-PD-L1 5′-GCTTTTCAATGTGACCAGCA-3′ Microscopy Rev-PD-L1 5′-TGGCTCCCAGCCTTACCAAG-3′ Before the immunostaining, cells were seeded for 3 h on Fw-18S 5′-CGATGCGGCGGCGTTATTC-3′ coverslips coated with polyornithine (0.1 mg/ml), then, Rev-18S 5′-TCTGTCAATCCTGTCCGTGTCC-3′ after washing with PBS the cells were fixed in 4% paraf- Fw-GAPDH 5′-GGACTCATGACCACAGTCCAT-3′ ormaldehyde (PFA) for 15 min. Brain coronal sections or Rev-GAPDH 5′-GTTCAGCTCAGGGATGACCTT-3′ cells on coverslips were permeabilized and unspecific Fw-β-Actin 5′-CGAGGCCCAGAGCAAGAGAG-3′ binding sites were blocked for an hour at room tem- Rev-β-Actin 5′-CGGTTGGCCTTAGGGTTCAG-3′ perature using a 10% donkey serum and 0.2% Triton X- 100 PBS solution. Tissue sections or coverslipped cells Spheroid formation assay were then incubated overnight at 4 °C with primary TM- and SVZ-GBM cells were cultured in DMEM/F12 antibodies diluted in PBS containing 0.1% of donkey serum-free medium containing B-27-Supplement without serum and 0.1% of Triton X-100, followed by tetra- vitamin A (Life Technologies; Carlsbad, CA, USA) and methylrhodamine (TRITC)-, fluorescein isothiocyanate supplemented with recombinant epidermal growth factor (FITC)-conjugated, or cyanine 5 (Cy5) secondary anti- and fibroblast growth factor 2 (EGF, 20 ng/mL and FGF-2, bodies (Jackson Immunoresearch Laboratories; Cam- 10 ng/mL; Preprotech, London, UK). After 4 days in cul- bridge, UK) diluted 1:500. Anti Ki-67 (BD Biosciences, ture, the number of spheroids was counted using an Becton Dickinson) was used diluted 1:500. Anti-Vimentin optical microscope. (rabbit monoclonal IgG, Cell Signaling, Danvers, MA, USA) was used diluted 1:400. Anti-PD-L1 (mouse Proliferation assay monoclonal, Abcam; Cambridge, UK) was used diluted TM- and SVZ-GBM cells were daily treated with 60 nM 1:200. Anti-Active Caspase-3 (Affinity-purified rabbit IgG, SAFit2 or opportunely diluted DMSO. Every 2 days, cells Promega, Madison, Wisconsin, USA) was diluted 1:250. were harvested, and cell number was estimated by the Anti-GFP (chicken polyclonal, Abcam; Cambridge, UK) Countess II Automated Cell Counter (Life Technologies). was used diluted 1:500. Counterstaining with DAPI After a 4-day treatment, cells were seeded on a coverslip (D9542, Sigma-Aldrich, St. Louis, MO, USA) was per- and processed for Ki67 immunostaining as previously formed and slides coverslipped with Fluoromount-G described. (00–4958–02, Invitrogen, Carlsbad, CA, USA). Images were acquired with OLYMPUS FV1000 confocal micro- Animal studies scopy. For PD-L1 and active caspase-3 expression, fluor- P40 female immune-deficient mice (Crl:NU-Foxn1nu) escence intensity was quantified according to “corrected were obtained from Charles River Laboratories (Charles total cryosection fluorescence” (CTCF) . Quantification River Laboratories®, Wilmington, UK) and handled of vimentin expression in-brain slices was performed as according to the Animal Ethical Committee of the Uni- followed: mean fluorescence intensity (MFI) was calcu- versity of Liège. All animals were cared for in accordance lated for each GFP+ cell stained with anti-vimentin along with the Declaration of Helsinki and the guidelines of the with GFP+ cell stained with control antibody. Positivity Belgium Ministry of Agriculture, in agreement with was assigned when vimentin MFI was at least four-fold the European Commission Laboratory Animal Care and higher than control MFI. When required, the brightness, Use Regulation (86/609/CEE, CE of J nL358, 18 December Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 11 of 12 Publisher’s note 1986). For intracranial transplantation, mice were anes- Springer Nature remains neutral with regard to jurisdictional claims in thetized with an intraperitoneal injection of ketamine published maps and institutional affiliations. (50 mg/mL, Pfizer, Bruxelles, Belgium) and xylazine The online version of this article (https://doi.org/10.1038/s41420-019-0216-0) (Sedativum 2%, Bayer, Bruxelles, Belgium) solution (V/V). contains supplementary material, which is available to authorized users. The cranium was exposed and a small hole was drilled 2.5 mm lateral and 0.5 mm anterior to the bregma with a Received: 13 July 2019 Revised: 1 August 2019 Accepted: 24 August 2019 size 34 Dremel Inverted Cone Burr. Mice were positioned in a stereotactic frame and 50000 GFP+ luc+ U87MG cells in 2 µl PBS were injected into the right striatum References through a 27-gauge needle over 1 min at 3 mm below the 1. D’Arrigo, P. et al. Manipulation of the immune system for cancer defeat: a dura mater. The incision was closed with 3 M Vetbond focus on the T cell inhibitory checkpoint molecules. Curr. Med. Chem. (2018). https://doi.org/0.2174/0929867325666181106114421, (Epub ahead Tissue Adhesive (Fisher Scientific, Hampton, NH, USA). of print). For mice treatment, with SAFit2, this was performed as 2. Wainwright, D. A. et al. Durable therapeutic efficacy utilizing combinatorial previously described . Before obtaining brains, mice were blockade against IDO, CTLA-4, and PD-L1 in mice with brain tumors. Clin. Cancer Res. 20,5290–5301 (2014). anaesthetized with an injection of Nembutal (Pento- 3. Berghoff, A. S. et al. Programmed death ligand 1 expression and tumor- barbital 60 mg/mL; Ceva Sante Animal, Libourne, France) infiltrating lymphocytes in glioblastoma. Neuro. Oncol. 17,1064–1075 (2015). followed by an intracardiac perfusion with a NaCl 0.9% 4. Nduom, E. K. et al. PD-L1 expression and prognostic impact in glioblastoma. Neuro Oncol. 18,195–205 (2016). solution (VWR International, Radnor, PA, USA) plus 4% 5. Parry,R.V. etal. CTLA-4 andPD-1receptors inhibit T-cell activation by distinct PFA at 4 °C (4,3 g/L NaOH, 40 g/L PFA, 18.8 g/L mechanisms. Mol. Cell. Biol. 25,9543–9553 (2005). NaH2PO4). Then after, brains were collected, postfixed in 6. Chen,R.Q., Liu, F.,Qiu,X.Y.&Chen,X.Q. The prognostic andtherapeutic value of pd-l1 in glioma. Front. Pharmacol. 9, 1503 (2018). 4% PFA and cryoprotected overnight in a 20% PBS/ 7. Butte,M.J., Keir, M. E., Phamduy, T. B., Sharpe, A. H. & Freeman,G.J. Pro- sucrose solution. Brains were frozen at −20 °C in a 2- grammed death-1 ligand 1 interacts specifically with the B7-1 costimulatory methylbutane solution (Sigma-Adrich) and cut into 14 μm molecule to inhibit T cell responses. Immunity 27,111–122 (2007). 8. Azuma, T. et al. B7-H1 is a ubiquitous antiapoptotic receptor on cancer cells. thick coronal sections using a cryostat. The tumor volume Blood 111,3635–3643 (2008). was calculated on GFP+ areas using ImageJ and, accord- 9. Zheng, F. et al. PD-L1 promotes self-renewal and tumorigenicity of malignant ing to Kim et al. , the formula (Volume = 0.5 × length × melanoma initiating cells. Biomed. Res. Int. 2017,1–8 (2017). 10. Qiu, X. Y. et al. PD-L1 confers glioblastoma multiforme malignancy via Ras width × height). For bioluminescence imaging: immuno- binding and Ras/Erk/EMT activation. Biochim. Biophys. Acta Mol. Basis Dis. 1864, deficient nude mice bearing intracranial xenografts were 1754–1769 (2018). injected intraperitoneally with D-luciferin (150 mg/kg, 11. Parsa, A. T. et al. Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in glioma. Nat. Med. 13,84–88 (2007). Sigma-Adrich). After anesthesia using 2.5% isoflurane, 12. Li, C. W. et al. Glycosylation and stabilization of programmed death ligand-1 mice were imaged with a charge-coupled device camera- suppresses T-cell activity. Nat. Commun. 7, 12632 (2016). based bioluminescence imaging system (IVIS 50, Xeno- 13. Romano, S. et al. Immunomodulatory pathways regulate expression of a spliced FKBP51 isoform in lymphocytes of melanoma patients. Pigment Cell gen; exposure time 1–30 s, binning 8, field of view 12, f/ Melanoma Res. 28,442–452 (2015). stop 1, open filter). 14. D’Arrigo, P. et al. A regulatory role for the co-chaperone FKBP51s in PD-L1 expression in glioma. Oncotarget 8,68291–68304 (2017). 15. Hsu, J. M., Li, C. W., Lai, Y. J. & Hung, M. C. Posttranslational modifications Statistical analysis of PD-L1 and their applications in cancer therapy. Cancer Res. 78, Student’s t-test was used to analyze differences between 6349–6353 (2018). means of values. A p-value of ≤0.05 was considered sta- 16. Jiang, W. et al. FK506 binding protein mediates glioma cell growth and sensitivity to rapamycin treatment by regulating NF-kappaB signaling path- tistically significant. way. Neoplasia 10, 235–243 (2008). 17. Romano, S. et al. The emerging role of large immunophilin FK506 binding protein51 incancer. Curr.Med.Chem. 18,5424–5429 (2011). Acknowledgements 18. Romano, S., D’Angelillo, A., Staibano, S., Ilardi, G. & Romano, M. F. FK506- We thank Regione Campania SATIN grant 2018–2020; the STAR L1-Junior binding protein 51 is a possible novel tumoral marker. Cell Death Dis. 1,e55 Principal Investigator Grants 2018 (funded by Unina and CSP). We are also (2010). grateful to the Cardiovascular Service srl for continually supporting our 19. Kroonen, J. et al. Human glioblastoma-initiating cells invade specifically the research. Regione Campania SATIN grant 2018–2020; the STAR-Junior Principal subventricular zones and olfactory bulbs of mice after striatal injection. Int. J. Investigator Grants 2018. Cancer 129, 574–585 (2011). 20. Goffart,N.etal. Adult mouse subventricular zones stimulate glioblastoma stem cells specific invasion through CXCL12/CXCR4 signaling. Neuro Oncol. 17, Author details 81–94 (2015). Dipartimento di Medicina Molecolare e Biotecnologie Mediche, Università di 21. Wang, Y. & Wang, L. miR-34a attenuates glioma cells progression and che- Napoli Federico II, Napoli, Italy. GIGA-Neurosciences, Faculté de Médecine, moresistance via targeting PD-L1. Biotechno. Lett. 39,1485–1492 (2017). Liège Université de Liège, Liège, Belgium. Technical University Darmstadt 22. Gaali, S. et al. Selective inhibitors of the FK506-binding protein 51 by induced Institute of Organic Chemistry and Biochemistry, Darmstadt, Germany fit. Nat. Chem. Biol. 11,33–37 (2015). 23. Giordano, A. et al. Tirofiban counteracts endothelial cell apoptosis through the VEGF/VEGFR2/pAkt axis. Vasc. Pharmacol. 80,67–74 (2016). 24. Romano, S. et al. Role of FK506 binding protein 51 [FKBP51] in the control of Conflict of interest apoptosis of irradiated melanoma cells. Cell Death Differ. 17, 145–157 (2010). The authors declare that they have no conflict of interest. Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 12 of 12 25. Vermes, I.,Haanen, C.,Steffens-Nakken,H.& Reutelingsperger,C. A novelassay 27. Schmittgen, T. D. & Livak, K. J. Analyzing real-time PCR data by the com- for apoptosis. Flow cytometric detection of phosphatidylserine expression on parative C(T) method. Nat. Protoc. 3,1101–1108 (2008). early apoptotic cells using fluorescein labelled Annexin V. J. Immunol. Methods 28. Kim, W. et al. Real-time imaging of glioblastoma using bioluminescence in a 184,39–51 (1995). U-87 MG xenograft model mouse. J. Korean Soc. Appl. Biol. Chem. 58,243–248 26. Lin, D. et al. Tenomodulin is essential for prevention of adipocyte accumu- (2015). lation and fibrovascular scar formation during early tendon healing. Cell Death Dis. 8, e3116 (2017). Official journal of the Cell Death Differentiation Association http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cell Death Discovery Springer Journals

The splicing FK506-binding protein-51 isoform plays a role in glioblastoma resistance through programmed cell death ligand-1 expression regulation

Download PDF
12 pages

Loading next page...
 
/lp/springer-journals/the-splicing-fk506-binding-protein-51-isoform-plays-a-role-in-0RyaqBg9Yp

References (30)

Publisher
Springer Journals
Copyright
Copyright © 2019 by The Author(s)
Subject
Life Sciences; Life Sciences, general; Biochemistry, general; Cell Biology; Stem Cells; Apoptosis; Cell Cycle Analysis
eISSN
2058-7716
DOI
10.1038/s41420-019-0216-0
Publisher site
See Article on Publisher Site

Abstract

Gliomas aberrantly express programmed cell death ligand-1 (PD-L1), which has a pivotal role in immunoevasion. The splicing isoform of FKBP5, termed FKBP51s, is a PD-L1 foldase, assisting the immune checkpoint molecule in maturation and expression on the plasma membrane. The concept that PD-L1 supports tumor-intrinsic properties is increasingly emerging. The aim of the present work was to confirm the pro-tumoral effect of PD-L1 on human glioma cell survival, stemness capacity and resistance, and to address the issue of whether, by targeting its foldase either chemically or by silencing, the aggressive tumor features could be attenuated. PD-L1-depleted glioma cells have a reduced threshold for apoptosis, while PD-L1 forced expression increases resistance. Similar results were obtained with FKBP51s modulation. The ability of PD-L1 to counteract cell death was hampered by FKBP51s silencing. PD-L1 expression was particularly high in glioma cells with a cancer-stem-cell profile. Moreover, PD-L1 sustained the spheroid formation capability of glioma cells. Targeting of FKBP51s by small-interfering RNA (siRNA) or the specific inhibitor SAFit2, reduced the number of formed spheroids, along with PD-L1 expression. Finally, in an orthotopic mouse model of glioblastoma, daily treatment with SAFit2 significantly reduced tumor PD-L1 expression, and tumor growth. In treated mice, caspase-3 activation and reduced vimentin expression were observed in excised tumors. In conclusion, targeting of FKBP51s hampers PD-L1 and its pro-tumoral properties, thereby affecting the self-renewal and growth capacities of glioblastoma cells in vitro and in vivo. Introduction primary brain cancer, for which no contemporary treat- Glioblastoma multiforme (GBM) is the most common, ments are curative. Gliomas, indeed, express several co- but also the most dangerous and aggressive form of inhibitory molecules that negatively regulate immune 1–4 system functions , among which programmed cell death ligand-1 (PD-L1) is one of the most represented and 3,4 Correspondence: Bernard Rogister (Bernard.Rogister@uliege.be) or Maria studied . The interaction of PD-L1 with its counter- Fiammetta Romano (mariafiammetta.romano@unina.it) receptor programmed cell death-1 (PD-1) expressed by Dipartimento di Medicina Molecolare e Biotecnologie Mediche, Università di activated T lymphocytes, inhibits T-cell activation and Napoli Federico II, Napoli, Italy GIGA-Neurosciences, Faculté de Médecine, Liège Université de Liège, Liège, proliferation, thus contributing to cancer immune eva- Belgium sion . So far, anti-PD-1 treatment has shown a therapeutic Full list of author information is available at the end of the article. efficacy limited to selected subsets of glioma patients . These authors contributed equally: Paolo D’Arrigo, Marina Digregorio Edited by A. Rufini © The Author(s) 2019 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Official journal of the Cell Death Differentiation Association 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 2 of 12 Several studies are in progress, especially exploring the function, i.e., SAFit2, which produced similar results to combination of anti-PD-1 antibodies with chemotherapies those found in FKBP51s silencing . or targeted therapies . It is worth noting that PD-L1 not The present study confirms that PD-L1 exerts impor- only binds to PD-1 but also competes with CD28-to- tant tumor-intrinsic properties in GBM. In particular, we CD80 binding , suggesting that a PD-1 blockade may show that PD-L1 sustains cell survival, resistance and sometimes be insufficient to prevent signals emanating stemness capability. PD-L1 expression was highest in from PD-L1 receptors. GBM cancer-initiating cells, due to a post-transcriptional Recent studies show that, in addition to its immuno- regulatory effect involving FKBP51s. Targeting of modulatory function, PD-L1 conveys a survival signal to FKBP51s by gene silencing or via the selective inhibitor the cancer cell and sustains tumor aggressiveness . SAFit2, downmodulated PD-L1 expression and inhibited Azuma et al. found that PD-L1, upon ligand stimulation, spheroid formation when GBM cancer-initiating cells acts as a receptor that transduces pro-survival signals that were cultured under non-adherent conditions. In an enhance the threshold for apoptosis induced by both orthotopic GBM mouse model, SAFit2 showed an anti- immune effectors and proapoptotic drugs. Zheng et al. tumor effect, as assessed by reductions in tumor volumes, found that PD-L1 sustains the self-renewal capability of caspase activation and attenuated expression levels of PD- malignant melanoma-initiating cells. Most recently, Qiu L1 and the mesenchymal marker vimentin. et al. demonstrated that PD-L1 binds to Ras and acti- vates the Ras/Erk pathway. Via this mechanism, PD-L1 Results promotes the epithelial-to-mesenchymal transition PD-L1 promotes apoptosis resistance (EMT) and GBM cell migration. We investigated the effect of PD-L1 silencing on GBM Akt supports aberrant PD-L1 expression in glioma cells, cell survival. To this end, we used two human GBM cell through a translational mechanism favouring the assem- lines previously found to highly express PD-L1 and bly of a polyribosome complex that allows better entrance FKBP51s, namely, D54MG and U251MG cells . For PD- of the PD-L1 transcript into polysomes . PD-L1 glyco- L1 downmodulation, cells were treated with specific siR- sylation is an essential element influencing protein stabi- NAs targeting PD-L1 or its co-chaperone FKBP51s. lity . Most recently, in glioma, the splicing isoform of the Twenty-four hours after transfection, some of the cells FK506-binding protein-51 (FKBP51s) was found to were harvested for lysate preparation. After a further 24 h, serve as a PD-L1 co-chaperone, assisting in protein gly- the remaining cells were collected for cell-death mea- cosylation surements via PI incorporation. Figure 1a shows a western . PD-L1 post-translational modifications have been thoroughly reviewed by Hsu et al. . According to blot assay of lysates obtained from human D54MG cells Jiang et al. and the Oncomine database (www.oncomine. treated with three different FKBP51s siRNAs and a spe- org), FKBP51 is among the top 10% of the most highly cific PD-L1 siRNA (siPD-L1). Two of the three siRNAs UTR1 expressed genes in GBM. The FKBP51 protein structure were designed on the 3′-UTR (siFKBP51s and UTR2 includes a C-terminal TPR three-tandem-repeat domain siFKBP51s ) and another (siFKBP51) on the coding responsible for protein-protein interaction, and two N- region. The PD-L1 signals at ≈50 kDa are those of mature terminal FK domains, of which the one with most N (glycosylated) forms and those under 37 kDa correspond terminals exhibits peptidyl-prolyl cis-trans isomerase to the naive protein (Fig. 1a). SiFKBP51s and 17,18 UTR1 (PPIase) activity . FKBP51s is generated by alternative siFKBP51s appeared to downmodulate FKBP51s 13 UTR2 splicing of FKBP5 pre-mRNA , which causes a frameshift more efficiently than siFKBP51s . Expression of PD- UTR1 with a premature stop codon, leading to a distinct C- L1 was also decreased by siFKBP51s and siFKBP51s , terminus, compared to the canonical isoform. FKBP51s in comparison to the control cells (NSRNA or none). The retains the PPIase activity but loses the TPR domain. An procaspase-7 level was decreased by the same siRNAs, IHC study on 29 GBM specimens showed that FKBP51s is with a cleaved fragment at ≈20 kDa also observable, broadly expressed in this tumor, albeit with different consistent with apoptosis activation (Fig. 1a). Measure- proportion/intensity scores, with nuclear and/or cyto- ment of hypodiploid cells confirmed that PD-L1 down- plasmic localization . Biochemical evaluation of glioma modulation, like FKBP51s silencing, produced cell death cell lines showed that naive PD-L1 is complexed with (Fig. 1a). The effect of different siRNAs on both FKBP51s FKBP51s in the endoplasmic reticulum, whereas the gly- and PD-L1, was also assessed in U251MG, as shown in the cosylated form was detected in the Golgi apparatus . Supplementary Information (Fig. S1). Since siPD-L1 and FKBP51s knockdown severely reduced the level of gly- siFKBP51s appeared to be the most effective for target cosylated PD-L1, whether constitutively expressed or downmodulation, these siRNAs were used in subsequent induced by ionizing radiation . The essential role of the experiments. Human U251MG cells showed similar PPIase activity in PD-L1 protein maturation was con- results to those obtained with D54MG cells (Fig. 1b). Both firmed by use of a selective inhibitor of this catalytic PD-L1 and FKBP51s silencing decreased PD-L1 Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 3 of 12 Fig. 1 PD-L1 regulates glioma cell apoptosis. a Analysis by western blot assay of PD-L1 and FKBP51s expression levels in D54MG cells, silenced for UTR1 UTR2 FKBP51s, using different siRNAs (siFKBP51s , siFKBP51s , and siFKBP51s) or PD-L1. The blot also shows caspase-7 levels recognized in its inactive (procaspase) and active forms. On the bottom of the panel, means and standard deviations of cell-death values of D54MG cells silenced for FKBP51s and PD-L1, are shown. Columns indicate the percentage of hypodiploid cells (N = 4). b Western blot assay of PD-L1 and FKBP51s levels in U251MG cells silenced for FKBP51s and PD-L1. On the bottom of the panel, means and standard deviations of cell-death values of U251MG cells silenced for FKBP51s and PD-L1, are shown. Columns indicate the percentage of hypodiploid cells (N = 4). c Graphical representation of means and standard deviations of cell-death values in PD-L1-depleted D54MG and U251MG cells. Twenty-four hours after transfection with PD-L1 siRNA, cells were treated with etoposide. They were then harvested after a further 24 h and analyzed for PI incorporation via flow cytometry. Columns indicate the percentage of hypodiploid cells (N = 3). d Flow-cytometric histograms of PD-L1 expression in D54MG and U251MG cells, cultured for 6 h with etoposide. On the right, a graphical representation is given of values (mean and standard deviation) from three different experiments. e Western blot assay of PD-L1 levels in GBM cells cultured with etoposide. f Measurement by qPCR of FKBP51s transcript level in cells treated or not treated with etoposide, for 6 h (N = 4). g Western blot assay of exogenous PD-L1 levels in D54MG and U251MG cells cultured in the absence or the presence of etoposide. h Graphical representation of means and standard deviations of cell-death values in PD-L1 D54MG and U251MG glioma cells. Twenty-four hours after transfection with EV or PD-L1-GFP, cells were treated with etoposide. They were then harvested after a further 24 h and analyzed for PI incorporation via flow cytometry. Columns indicate the percentage of hypodiploid cells. Each experiment was performed at least three times and in triplicate expression levels, but only FKBP51s siRNA decreased the of the two factors appeared to further increase cell death, FKBP51s expression level. Activation of caspase-3 was in comparison with the single treatment. This result registered in U251MG cells using flow cytometry (Sup- suggested that reduced levels of PD-L1 could act in plementary Information, Fig. S2). We, then, investigated concert with the chemotherapeutic compound to enhance the effect of PD-L1 silencing on etoposide-induced cell its cytotoxicity (Fig. 1c). Using flow cytometry, we found death. Silencing of PD-L1 appeared to exert a cytotoxic that both cell lines, when cultured with etoposide for 6 h, effect similar to that of etoposide. However, combination had increased levels of PD-L1, compared to the same Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 4 of 12 untreated cells (Fig. 1d). As expected, western blot ana- FKBP51s hampered the pro-survival effect of ectopic PD- lysis confirmed the increase in the mature PD-L1 signals L1 (Fig. 2d). The cell-death modulatory effect of FKBP51s at 50 kDa (Fig. 1e) and showed an increased expression of was also assessed in U87MG cell lines (Supplementary FKBP51s in etoposide-treated cells. These results sug- Information, Fig. S5), as well as using SAFit2 (Supple- gested that etoposide induced FKBP5 mRNA splicing, mentary Information, Fig. S6). which was confirmed at the transcription level (Fig. 1f). Ectopic expression of PD-L1 (Fig. 1g), significantly PD-L1 expression increases with glioma aggressiveness reduced etoposide-induced cell death (Fig. 1h). Taken and affects spheroid formation in culture together, these results suggest that PD-L1 is a resistance The role of PD-L1 in the apoptosis resistance of GBM factor for GBM cells. Further confirmation of the PD-L1 cells, together with its previous involvement in EMT , pro-survival effect was obtained with the other two GBM supports the concept that this immune checkpoint is a cell lines, namely, U87MG and SF767 (see Supplementary hallmark of GBM malignancy. To address whether PD-L1 Information, Figs. S3 and S4 respectively). Interestingly, in expression increased with the aggressiveness of GBM SF767 cells, which show deficient PD-L1 levels , PD-L1 cells, we used two GBM cell lines previously established silencing did not produce an apoptosis-enhancing effect, by Kroonen et al. from the tumor mass (TM-GBM) and whereas PD-L1 ectopic expression reduced etoposide the subventricular zone (SVZ-GBM) of an orthotopic cytotoxicity (Supplementary Information, Fig. S4). GBM model. TM-GBM and SVZ-GBM cell lines, char- acterized by an increasing level of aggressiveness from FKBP51s promotes apoptosis resistance TM to SVZ, were derived from the human U87MG cell Since FKBP51s deprivation reduces PD-L1 expression , line. SVZ-GBM cells are particularly rich in GBM cancer we investigated whether FKBP51s affected GBM cell stem cells and have a high tumorigenic potential, as death using RNA silencing. The effect of protein down- assessed by mouse reinjection or by their capacity to expression was assessed using western blot analysis (Fig. 2a). grow in spheres when cultured under non-adherent Cell death was assessed in the absence or the presence of conditions . We confirmed the high tumorigenicity of etoposide, via PI incorporation. Figure 2a shows repre- SVZ-GBM cells showing increased levels of p-Akt, along sentative flow-cytometric histograms of hypodiploid with its substrate p-S6K1 (Fig. 3a) and the stemness events and a graphic representation of the means and marker CD133 (Fig. 3b), in comparison to TM-GBM cells. standard deviations of cell-death values obtained in three SVZ-GBM also showed increased PD-L1 expression in independent experiments, each performed in triplicate. contrast to TM-GBM cells, as assessed by western blot FKBP51s silencing in D54MG cells produced 20.0 ± 4.0% analysis (Fig. 3b) and flow cytometry (Fig. 3c). Surpris- cell death, while treatment of cells with a non-silencing ingly, PD-L1 mRNA levels appeared to be higher in the oligo (NSRNA) produced only 2.3 ± 0.5% cell death (p = TM-GBM cells than in the SVZ-GBM cells (Fig. 3d). In 0.01) (Fig. 2a). Similarly, in etoposide-treated cells, line with Goffart et al. , we found that SVZ-GBM cells FKBP51s siRNA and NSRNA produced 28.0 ± 4.0% and have an increased capacity to form spheroids in com- 20.0 ± 0.1% cell death respectively (p = 0.01) (Fig. 2a). parison with TM-GBM cells (Fig. 3e). As expected, FKBP51s-silenced U251MG produced an average of spheroids (+) expressed Sox-2 levels remarkably higher 14.3 ± 4.0% cell death, while NSRNA produced only 3.3 ± than those in the corresponding adherent cells (-) 0.5% cell death (p = 0.03) (Fig. 2a). In etoposide cultures, (Fig. 3e). Expression of PD-L1 resulted especially in an FKBP51s siRNA and NSRNA produced 55.0 ± 4.1% and increase in spheroids formed by TM-GBM cells, in 46.7 ± 1.1% cell death respectively (Fig. 2a) (p = 0.05). To comparison with attached TM-GBM cells. Although only confirm the role of FKBP51s in the resistance of GBM a slight increase in PD-L1 expression was observed in cells, we overexpressed FKBP51s and analyzed cell death SVZ-GBM cells growing in spheres (Fig. 3f), the finding using EV cells for comparison. In line with the FKBP51s that PD-L1 silencing impaired the capacity to form knockdown sensitizing effect, we registered a significant spheroids in both TM- and SVZ-GBM cells (Fig. 3g) decrease in etoposide-induced cell death in D54MG and highlighted the importance of PD-L1 in spheroid forma- U251MG cells transfected with FKBP51s vector, com- tion. Taken together, our results indicate that PD-L1 is a pared with EV cells (p = 0.03 and p = 0.004 for D54MG necessary but not a sufficient condition for allowing the and U251MG, respectively) (Fig. 2b). To investigate GBM cells to grow in spheres. whether PD-L1 requires FKBP51s to counteract chemotherapy-induced cell death, we overexpressed PD- Targeting of FKBP51s downregulates PD-L1 and hampers L1 under conditions of FKBP51s depletion. To this end, GBM malignancy we co-transfected PD-L1 and FKBP51s siRNA or NSRNA, Since the increased expression of PD-L1 in SVZ-GBM in D54MG and U251MG cells (Fig. 2c), and measured cell cells, in comparison to TM-GBM cells, appeared to be death in etoposide cultures (Fig. 2d). The silencing of related to a post-transcriptional mechanism, we investigated Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 5 of 12 Fig. 2 FKBP51s regulates glioma cell death. a Western blot assay of FKBP51s levels in D54MG and in U251MG silenced for FKBP51s. Cells were treated with FKBP51s siRNA or NSRNA for 24 h. Then, some of the cells were harvested for lysate preparation and some cells were treated with etoposide. After a further 24 h, cell death was measured. Representative flow-cytometric histograms of PI incorporation are shown along with graphical representations of means and standard deviations of cell-death values (N = 3). b Western blot assays of exogenous FKBP51s levels and graphical representations of means and standard deviations of cell-death values in D54MG and U251MG cells transfected with FKBP51s with EV- or FKBP51s-carrying plasmids (N = 3). Cells were treated with etoposide at 24 h from transfection. After a further 24 h, cell death was measured by flow cytometry (N = 3). c Western blot assays of the exogenous PD-L1-GFP level in D54MG and U251MG cells, silenced or not silenced for FKBP51s. Silencing of FKBP51s produced a decrease in PD-L1-GFP level in both cell lines. d D54MG and U251MG cells transfected with EV, PD-L1-GFP+ NSRNA and PD-L1-GFP+ FKBP51s siRNA, were treated with etoposide. After 24 h, cell death was measured using flow cytometry. FKBP51s depletion reduced the antiapoptotic effect of exogenous PD-L1. Each experiment was performed at least three times and in triplicate the effect of FKBP51s silencing on the modulation of PD-L1 the effect of ectopic PD-L1-GFP in promoting spheroid levels in both TM-GBM and SVZ-GBM cells. As shown in formation (Fig. 4d, e). A western blot validated the impact the western blot in Fig. 4a, the silencing of FKBP51s of FKBP51s silencing on protein levels (Fig. 4f). We then downmodulated PD-L1 levels in both cell lines. This result investigated the effect of SAFit2 on PD-L1 expression and was confirmed using flow cytometry (Fig. 4b). FKBP51s spheroid formation in TM-GBM and SVZ-GBM cells. PD- downmodulation impaired spheroid formation in both TM- L1 expression, as assessed by flow cytometry, showed a GBM and SVZ-GBM cells (Fig. 4c). Interestingly, SVZ- significant decrease on the plasma membrane in SAFit2 GBM appeared to be more sensitive than TM-GBM to the cultures (Fig. 5a). Like FKBP51s siRNA, SAFit2 impaired targeting of FKBP51s, according to the PD-L1 down- spheroid formation in both TM-GBM and SVZ-GBM cells expression, which was more powerful in SVZ-GBM than in (Fig. 5b). To address whether SAFit2 could affect tumor TM-GBM cells (Fig. 4b). FKBP51s silencing counteracted growth, we assessed Ki67 expression and cell counts in TM- Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 6 of 12 Fig. 3 PD-L1 sustains GBM capacity to form spheroids. a Immunoblot of p-Akt and p-S6K1 in TM-GBM and SVZ-GBM cells. Levels of phospho- enzymes were higher in SVZ-GBM cells than in TM-GBM cells. b Expression of PD-L1 and CD133 in TM-GBM and SVZ-GBM cells. Levels of both proteins were higher in SVZ-GBM cells than in TM-GBM cells. c Flow cytometry measurements of PD-L1 expression in TM-GBM and SVZ-GBM cells. Graph columns represent mean fluorescence intensities with related p-values (N = 3). A representative histogram is shown in overlay. d RT-qPCR results for mRNA levels of PD-L1 in TM-GBM and SVZ-GBM cells. Graph columns represent normalized quantification relative to TM-GBM cells (N = 3). e Representative images of formed spheroids taken using an optical microscope and spheroid counts in TM-GBM and SVZ-GBM cultures (scale bar represents 100 μm). Graph columns represent spheroid numbers relative to TM-GBM cells (N = 3). A western blot of Sox-2 levels shows increased levels of the stemness marker in the spheroids (+), in comparison with adherent cells (−). f Quantification using flow cytometry of PD-L1 levels, in TM-GBM-formed spheroids and SVZ-GBM-formed spheroids. Graph columns represent MFI (N = 3). g Spheroid formation assay with TM-GBM and SVZ-GBM cells silenced or not silenced for PD-L1. Graph columns represent the relative spheroid numbers after a 4-day culture. NSRNA-treated cells were used as the reference sample. Each experiment was performed at least three times and in triplicate GBM and SVZ-GBM cell cultures (Fig. 5c, d). On SAFit2 PD-L1 expression (Fig. 6d). Finally, in-brain slices, the treatment, both cell lines showed a significant decrease in percentage of tumor cells found to be positive for the Ki67 expression (Fig. 5c) and cell counts (Fig. 5d). To expression of the mesenchymal marker vimentin, was still address the in vivo effect of SAFit2, U87MG cells (carrying reduced in mice treated with SAFit2 (Supplementary GFP and the luciferase reporter gene) were injected into the Information, Fig. S6). right striatum of 32 nude mouse brains. Two weeks later, 16 mice were treated daily with SAFit2, while 16 mice were Discussion treated intraperitoneally with a vehicle designed to pass A growing body of literature supports the concept that through the brain-blood barrier, via intraperitoneal injec- PD-L1 has intrinsic pro-tumoral aspects that are inde- tion. After 2 weeks of treatment (corresponding to 4 weeks pendent of its immunoinhibitory functions . Moreover, from cell xenotransplantation) the mice were sacrificed, and several recent studies report that expression levels of PD- the brains obtained. A reduced tumor mass was calculated L1 positively correlate with glioma grades. More precisely, in mice treated with SAFit2 compared to tumors from mice in GBM, these levels are much higher than in grade II and treated with the vehicle (Fig. 6a). The result was confirmed grade III gliomas . by a reduced luminescence intensity recorded in the living In this paper, consistent with the finding of Azuma mice before the sacrifice (Fig. 6b). The reduced tumor et al. that PD-L1 is a ubiquitous antiapoptotic receptor, growth was accompanied by the presence of active caspase- we show that PD-L1 is an essential element in GBM 3 in xenografts from the SAFit2-treated mice (Fig. 6c). In resistance to cell death. A previous paper also reported accordance with in vitro findings, SAFit2 reduced tumor that PD-L1 overexpression, as a result of the low miR-34a Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 7 of 12 Fig. 4 FKBP51s regulates PD-L1 expression and spheroid formation in TM-GBM and SVZ-GBM cells. a Immunoblot of PD-L1 and FKBP51s levels in cells silenced for FKBP51s. Canonical FKBP51 confirmed the specificity of the silencing. b Flow cytometry analysis of PD-L1 expression in cells, silenced or not silenced for FKBP51s. Graph columns represent MFI using non-silenced cells as the reference sample (N = 3). c Spheroid assay with cells from TM-GBM and SVZ-GBM cells, silenced or not silenced for FKBP51s. Representative images of formed spheroids are shown. Graph columns represent the relative spheroid numbers, after a 4-day culture, using NSRNA-treated cells as the reference sample (N = 3). d Spheroid assay with TM-GBM cells and SVZ-GBM cells transfected with EV, PD-L1-GFP+ NSRNA, and PD-L1-GFP+ FKBP51s siRNA. FKBP51s depletion reduced the spheroid formation stimulated by exogenous PD-L1 (scale bar represents 250μm) (N = 3). e Fluorescent microscopy visualization of ectopic PD-L1-GFP in the spheroid assay (scale bar represents 50 μm). FKBP51s depletion reduced spheroid numbers and the extent of fluorescence. f Western blot assay of FKBP51s levels in FKBP51s-silenced spheroids. Each experiment was performed at least three times and in triplicate 21 14 level in U87 glioma cells, induced paclitaxel resistance . this co-chaperone in PD-L1 expression . FKBP51s In the present paper, we confirm the resistance effect of appeared to significantly affect the spheroid formation PD-L1 in four different glioma cell lines, namely, U87MG, ability of GBM cancer-initiating cells, along with PD-L1. D54MG, U251MG and SF767. Moreover, in accordance Collectively, these findings, in addition to supporting the with Zheng et al. , who demonstrated a role for PD-L1 in pro-oncogenic role of PD-L1 in glioma, open up the sustaining the self-renewal capability of malignant possibility of targeting PD-L1 by acting on its co- melanoma-initiating cells, we show that PD-L1 expression chaperone FKBP51s. SAfit2 is, so far, the most well- is increased in GBM cancer-initiating cells and has a characterized FKBP51 ligand available . After 2 weeks of supporting role in spheroid formation. Finally, in accor- intraperitoneal treatment with SAFit2, assessment of dance with Qiu et al. , who demonstrated the promoting GFP + luc + tumor volumes in-brain slices showed a role of PD-L1 in GBM-cell migration, we found increased significant decrease in SAFit2-treated mice, in compar- expression of PD-L1 in GBM cells that had migrated to ison with vehicle-treated control mice. Consistent with the subventricular zone of the brain, compared with PD- this finding, reduced luminescence intensity was regis- L1 expression in the nigrostriatal tumor mass. When the tered in SAFit2-treated tumors in living mice before levels of PD-L1 were downmodulated by FKBP51s silen- sacrifice. The presence of active caspase-3 in tumor cing, GBM-cell sensitivity to both spontaneous and xenografts from SAFit2-treated animals suggested that chemotherapy-induced death was increased. Moreover, the reduction in tumor volumes can be ascribed, at least when the levels of PD-L1 were upregulated by an ectopic in part, to apoptosis induction. A decreased percentage of expression, a pro-survival effect on GBM cells was tumor cells expressing PD-L1 and vimentin, according to observed. This outcome was counteracted by FKBP51s histological examinations, suggested that in treated mice silencing, which is consistent with the essential role for the tumor had a reduced aggressiveness. Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 8 of 12 Fig. 5 SAFit2 decreases PD-L1 expression and spheroid formation and impairs in vitro growth of TM-GBM and SVZ-GBM cells. a Cytometry analysis of PD-L1 expression in cells treated with SAFit2 for 12 h. Graph columns represent mean fluorescence intensities using cells with Dimethyl sulfoxide (DMSO) as the reference sample (N = 3). b Spheroid assay with cells treated with SAFit2 or the vehicle. Representative pictures of formed spheroids are shown along graph columns for the relative spheroid number measured after a 4-day culture. DMSO-treated cells were used as the reference sample (N = 3). c Ki67 expression measured by immunofluorescence in cells treated with SAFit2 or the vehicle (scale bar represents 25 μm). Graph columns represent the percentage of Ki67-positive cells, relative to the control (N = 6). d Kinetics of counts of TM-GBM and SVZ-GBM cells treated with SAFit2 or the vehicle. SAFit2 significantly reduced cell counts (N = 4) In conclusion, in line with a previous study demon- DDK-tagged and a True-ORF-GFP-tagged expression vec- strating that in vivo knockdown of PD-L1 in nude mice tors were used (OriGene Technologies, Rockville, MD, completely abolished GBM xenograft formation USA), which carried the complementary DNA (cDNA) of , our finding shows that manipulation of PD-L1, through che- the human FKBP5 transcript variant 4 and CD274, mical inhibition of its co-chaperone FKBP51s, activates respectively. Control cells were transfected with the related apoptosis and mitigates the aggressive features of GBM empty vector (EV). FKBP51s silencing was performed using xenografts in nude mice. Our study contributes some short-interfering oligoribonucleotides. The siFKBP51 was a information to the emerging concept that there is a strict mix of two siRNAs designed on the coding region; UTR1 UTR2 interplay between cancer immunoediting and resistance siFKBP51s and FKBP51s were designed on 3′- and suggests that SAFit2 may be a beneficial neoadjuvant UTR . The non-silencing RNA (NSRNA) and PD-L1 strategy for the management of this tumor. siRNA were from Qiagen (siPD-L1) (Valencia, CA, USA) or Novus Biologicals (siPD-L1.2) (Littleton, CO). U87MG cells Methods and materials used for xenograft implantation were previously transduced Cell culture and transfection with lentiviral vectors allowing the dual expression of eGFP Human glioma cell lines D54MG, U251MG, SF767MG and luciferase as previously described . In experiments were obtained and cultured as described .U87MG from with etoposide, the drug was added 24 h after transfection at American Type Culture Collection (ATCC) were cultured aconcentration of 20 μM. For experiments with SAFit2 , in Dulbecco’sModified Eagle Medium (DMEM) containing which was provided by Prof. Felix Hausch laboratory, a 10% FBS, 200 mM glutamine, and 100 U/mL stock solution (60 μM in DMSO) wasusedat1:1000dilu- penicillin–streptomycin. TM-GBM cells and SVZ-GBM tion in DMEM-F12 media; for control cells, DMSO has cells were U87MG cells obtained by microdissection of diluted accordingly. xenografted mice brain , from, respectively, the tumor mass and the subventricular zone, and seeded in cultures . Western blot For FKBP51s and PD-L1 knockin and knockdown, cells Whole-cell lysates were homogenized in modified RIPA were transfected using the K2 Transfection System (Bion- buffer and assayed by immunoblot as previously 14 14 tex, Munich, Germany), as previously described and in described . The primary antibodies against FKBP51 accordance with the manufacturer’s recommendations. For (rabbit polyclonal; Novus Biological), FKBP51s and overexpressing FKBP51s and PD-L1 a True-ORF-Myc- CD274/PD-L1 (rabbit polyclonal; Novus Biological) were Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 9 of 12 Fig. 6 SAFit2 impairs GBM growth in vivo. a Graphic representation of tumor volumes formed in 16 mice treated daily with vehicle and 16 mice treated with SAFit2. b Luciferase assay, performed on mice treated or not treated with SAFit2, 2 weeks after in-brain injection with GFP+ luc+ U87MG cells. Three mice from each group are shown before starting the treatment and after 10 days (left). Graph columns represent mean values of signal intensities from the two groups, before and after treatment (right) (N = 16). c Expression of active caspase-3 by immunofluorescence on brain slices (scale bar represents 50 μm). Box plots represent the values of GFP+ cleaved Casp3+ cells measured using ImageJ software, from seven mice treated with vehicle and seven mice treated with SAFit2. d Expression of PD-L1 determined by immunofluorescence on brain slices (scale bar represents 50 μm). Box plots represent the values of GFP+ PD-L1+ cells measured using ImageJ software, from nine mice treated with vehicle and nine mice treated with SAFit2 used diluted 1:2500. CD133 (rabbit polyclonal; Abcam; 1:5000. Anti β-Actin-Peroxidase (mouse monoclonal; Cambridge, UK) was used diluted 1:1000. A further Sigma-Adrich) was used diluted 1:10000. Anti-phospho- antibody Pdcd-1L1 (rabbit polyclonal, Santa Cruz Bio- Akt (Ser473), Akt1/2/3, phosphor-S6 kinase, G3PDH technology; Santa Cruz, CA, USA) was used for PD-L1 and Sox-2 (rabbit monoclonal; Cell Signaling, Danvers, detection at the 1:1000 dilution. Antibody against M2- MA, USA) were used diluted 1:1000. Anti-p70S6 Flag, caspase-7 and γ-Tubulin (mouse monoclonal; kinase (rabbit polyclonal; Santa Cruz) was used Sigma-Aldrich, St. Louis, MO, USA) were used diluted diluted 1:500. Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 10 of 12 Flow cytometry contrast, and color balance of the images were adjusted in Expression of PD-L1 was assessed using anti-B7H1- Photoshop CS2 (Adobe Systems, San Jose, California, phycoerythrin (PE) (R&D Systems, Minneapolis, MN, USA). This adjustment was applied to every pixel in USA) at a concentration of 0.05 μg/ml. A PE-conjugated each image. control Ig isotype was used to assess non-specific binding. Briefly, cells were harvested and incubated with the Quantitative PCR (qPCR) antibodies mentioned above for 30 min in the dark at 4 °C, Total RNA was extracted from cells by Trizol (Invitro- washed and then analyzed with a BD Accuri™ C6 Cyt- gen, Carlsbad, CA, USA). Each RNA was used for cDNA ometer (BD Biosciences, New Jersey, USA). Cleaved synthesis with iScript Reverse Transcription (Bio-Rad, caspase-3 immunostaining was performed as previously CA, USA). Relative gene expression was quantified by 23 27 described . Apoptosis was assessed by two methods: qPCR with 2–ΔCt comparative method , using the propidium iodide (PI) incorporation to analyze the DNA SsoAdvancedTM SYBR Green Supermix (Bio-Rad) and content in permeabilized cells , and the annexin V- specific qPCR primers. Amplification of FKBP51s was binding assay in double staining with PI, according to performed as previously described . Oligo sequences Vermes et al. . Cells were analyzed with a BD Accuri™ C6 used for PD-L1 mRNA quantitation along with co- Cytometer (BD, Becton Dickinson). amplified housekeeping genes are: Fw-PD-L1 5′-GCTTTTCAATGTGACCAGCA-3′ Microscopy Rev-PD-L1 5′-TGGCTCCCAGCCTTACCAAG-3′ Before the immunostaining, cells were seeded for 3 h on Fw-18S 5′-CGATGCGGCGGCGTTATTC-3′ coverslips coated with polyornithine (0.1 mg/ml), then, Rev-18S 5′-TCTGTCAATCCTGTCCGTGTCC-3′ after washing with PBS the cells were fixed in 4% paraf- Fw-GAPDH 5′-GGACTCATGACCACAGTCCAT-3′ ormaldehyde (PFA) for 15 min. Brain coronal sections or Rev-GAPDH 5′-GTTCAGCTCAGGGATGACCTT-3′ cells on coverslips were permeabilized and unspecific Fw-β-Actin 5′-CGAGGCCCAGAGCAAGAGAG-3′ binding sites were blocked for an hour at room tem- Rev-β-Actin 5′-CGGTTGGCCTTAGGGTTCAG-3′ perature using a 10% donkey serum and 0.2% Triton X- 100 PBS solution. Tissue sections or coverslipped cells Spheroid formation assay were then incubated overnight at 4 °C with primary TM- and SVZ-GBM cells were cultured in DMEM/F12 antibodies diluted in PBS containing 0.1% of donkey serum-free medium containing B-27-Supplement without serum and 0.1% of Triton X-100, followed by tetra- vitamin A (Life Technologies; Carlsbad, CA, USA) and methylrhodamine (TRITC)-, fluorescein isothiocyanate supplemented with recombinant epidermal growth factor (FITC)-conjugated, or cyanine 5 (Cy5) secondary anti- and fibroblast growth factor 2 (EGF, 20 ng/mL and FGF-2, bodies (Jackson Immunoresearch Laboratories; Cam- 10 ng/mL; Preprotech, London, UK). After 4 days in cul- bridge, UK) diluted 1:500. Anti Ki-67 (BD Biosciences, ture, the number of spheroids was counted using an Becton Dickinson) was used diluted 1:500. Anti-Vimentin optical microscope. (rabbit monoclonal IgG, Cell Signaling, Danvers, MA, USA) was used diluted 1:400. Anti-PD-L1 (mouse Proliferation assay monoclonal, Abcam; Cambridge, UK) was used diluted TM- and SVZ-GBM cells were daily treated with 60 nM 1:200. Anti-Active Caspase-3 (Affinity-purified rabbit IgG, SAFit2 or opportunely diluted DMSO. Every 2 days, cells Promega, Madison, Wisconsin, USA) was diluted 1:250. were harvested, and cell number was estimated by the Anti-GFP (chicken polyclonal, Abcam; Cambridge, UK) Countess II Automated Cell Counter (Life Technologies). was used diluted 1:500. Counterstaining with DAPI After a 4-day treatment, cells were seeded on a coverslip (D9542, Sigma-Aldrich, St. Louis, MO, USA) was per- and processed for Ki67 immunostaining as previously formed and slides coverslipped with Fluoromount-G described. (00–4958–02, Invitrogen, Carlsbad, CA, USA). Images were acquired with OLYMPUS FV1000 confocal micro- Animal studies scopy. For PD-L1 and active caspase-3 expression, fluor- P40 female immune-deficient mice (Crl:NU-Foxn1nu) escence intensity was quantified according to “corrected were obtained from Charles River Laboratories (Charles total cryosection fluorescence” (CTCF) . Quantification River Laboratories®, Wilmington, UK) and handled of vimentin expression in-brain slices was performed as according to the Animal Ethical Committee of the Uni- followed: mean fluorescence intensity (MFI) was calcu- versity of Liège. All animals were cared for in accordance lated for each GFP+ cell stained with anti-vimentin along with the Declaration of Helsinki and the guidelines of the with GFP+ cell stained with control antibody. Positivity Belgium Ministry of Agriculture, in agreement with was assigned when vimentin MFI was at least four-fold the European Commission Laboratory Animal Care and higher than control MFI. When required, the brightness, Use Regulation (86/609/CEE, CE of J nL358, 18 December Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 11 of 12 Publisher’s note 1986). For intracranial transplantation, mice were anes- Springer Nature remains neutral with regard to jurisdictional claims in thetized with an intraperitoneal injection of ketamine published maps and institutional affiliations. (50 mg/mL, Pfizer, Bruxelles, Belgium) and xylazine The online version of this article (https://doi.org/10.1038/s41420-019-0216-0) (Sedativum 2%, Bayer, Bruxelles, Belgium) solution (V/V). contains supplementary material, which is available to authorized users. The cranium was exposed and a small hole was drilled 2.5 mm lateral and 0.5 mm anterior to the bregma with a Received: 13 July 2019 Revised: 1 August 2019 Accepted: 24 August 2019 size 34 Dremel Inverted Cone Burr. Mice were positioned in a stereotactic frame and 50000 GFP+ luc+ U87MG cells in 2 µl PBS were injected into the right striatum References through a 27-gauge needle over 1 min at 3 mm below the 1. D’Arrigo, P. et al. Manipulation of the immune system for cancer defeat: a dura mater. The incision was closed with 3 M Vetbond focus on the T cell inhibitory checkpoint molecules. Curr. Med. Chem. (2018). https://doi.org/0.2174/0929867325666181106114421, (Epub ahead Tissue Adhesive (Fisher Scientific, Hampton, NH, USA). of print). For mice treatment, with SAFit2, this was performed as 2. Wainwright, D. A. et al. Durable therapeutic efficacy utilizing combinatorial previously described . Before obtaining brains, mice were blockade against IDO, CTLA-4, and PD-L1 in mice with brain tumors. Clin. Cancer Res. 20,5290–5301 (2014). anaesthetized with an injection of Nembutal (Pento- 3. Berghoff, A. S. et al. Programmed death ligand 1 expression and tumor- barbital 60 mg/mL; Ceva Sante Animal, Libourne, France) infiltrating lymphocytes in glioblastoma. Neuro. Oncol. 17,1064–1075 (2015). followed by an intracardiac perfusion with a NaCl 0.9% 4. Nduom, E. K. et al. PD-L1 expression and prognostic impact in glioblastoma. Neuro Oncol. 18,195–205 (2016). solution (VWR International, Radnor, PA, USA) plus 4% 5. Parry,R.V. etal. CTLA-4 andPD-1receptors inhibit T-cell activation by distinct PFA at 4 °C (4,3 g/L NaOH, 40 g/L PFA, 18.8 g/L mechanisms. Mol. Cell. Biol. 25,9543–9553 (2005). NaH2PO4). Then after, brains were collected, postfixed in 6. Chen,R.Q., Liu, F.,Qiu,X.Y.&Chen,X.Q. The prognostic andtherapeutic value of pd-l1 in glioma. Front. Pharmacol. 9, 1503 (2018). 4% PFA and cryoprotected overnight in a 20% PBS/ 7. Butte,M.J., Keir, M. E., Phamduy, T. B., Sharpe, A. H. & Freeman,G.J. Pro- sucrose solution. Brains were frozen at −20 °C in a 2- grammed death-1 ligand 1 interacts specifically with the B7-1 costimulatory methylbutane solution (Sigma-Adrich) and cut into 14 μm molecule to inhibit T cell responses. Immunity 27,111–122 (2007). 8. Azuma, T. et al. B7-H1 is a ubiquitous antiapoptotic receptor on cancer cells. thick coronal sections using a cryostat. The tumor volume Blood 111,3635–3643 (2008). was calculated on GFP+ areas using ImageJ and, accord- 9. Zheng, F. et al. PD-L1 promotes self-renewal and tumorigenicity of malignant ing to Kim et al. , the formula (Volume = 0.5 × length × melanoma initiating cells. Biomed. Res. Int. 2017,1–8 (2017). 10. Qiu, X. Y. et al. PD-L1 confers glioblastoma multiforme malignancy via Ras width × height). For bioluminescence imaging: immuno- binding and Ras/Erk/EMT activation. Biochim. Biophys. Acta Mol. Basis Dis. 1864, deficient nude mice bearing intracranial xenografts were 1754–1769 (2018). injected intraperitoneally with D-luciferin (150 mg/kg, 11. Parsa, A. T. et al. Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in glioma. Nat. Med. 13,84–88 (2007). Sigma-Adrich). After anesthesia using 2.5% isoflurane, 12. Li, C. W. et al. Glycosylation and stabilization of programmed death ligand-1 mice were imaged with a charge-coupled device camera- suppresses T-cell activity. Nat. Commun. 7, 12632 (2016). based bioluminescence imaging system (IVIS 50, Xeno- 13. Romano, S. et al. Immunomodulatory pathways regulate expression of a spliced FKBP51 isoform in lymphocytes of melanoma patients. Pigment Cell gen; exposure time 1–30 s, binning 8, field of view 12, f/ Melanoma Res. 28,442–452 (2015). stop 1, open filter). 14. D’Arrigo, P. et al. A regulatory role for the co-chaperone FKBP51s in PD-L1 expression in glioma. Oncotarget 8,68291–68304 (2017). 15. Hsu, J. M., Li, C. W., Lai, Y. J. & Hung, M. C. Posttranslational modifications Statistical analysis of PD-L1 and their applications in cancer therapy. Cancer Res. 78, Student’s t-test was used to analyze differences between 6349–6353 (2018). means of values. A p-value of ≤0.05 was considered sta- 16. Jiang, W. et al. FK506 binding protein mediates glioma cell growth and sensitivity to rapamycin treatment by regulating NF-kappaB signaling path- tistically significant. way. Neoplasia 10, 235–243 (2008). 17. Romano, S. et al. The emerging role of large immunophilin FK506 binding protein51 incancer. Curr.Med.Chem. 18,5424–5429 (2011). Acknowledgements 18. Romano, S., D’Angelillo, A., Staibano, S., Ilardi, G. & Romano, M. F. FK506- We thank Regione Campania SATIN grant 2018–2020; the STAR L1-Junior binding protein 51 is a possible novel tumoral marker. Cell Death Dis. 1,e55 Principal Investigator Grants 2018 (funded by Unina and CSP). We are also (2010). grateful to the Cardiovascular Service srl for continually supporting our 19. Kroonen, J. et al. Human glioblastoma-initiating cells invade specifically the research. Regione Campania SATIN grant 2018–2020; the STAR-Junior Principal subventricular zones and olfactory bulbs of mice after striatal injection. Int. J. Investigator Grants 2018. Cancer 129, 574–585 (2011). 20. Goffart,N.etal. Adult mouse subventricular zones stimulate glioblastoma stem cells specific invasion through CXCL12/CXCR4 signaling. Neuro Oncol. 17, Author details 81–94 (2015). Dipartimento di Medicina Molecolare e Biotecnologie Mediche, Università di 21. Wang, Y. & Wang, L. miR-34a attenuates glioma cells progression and che- Napoli Federico II, Napoli, Italy. GIGA-Neurosciences, Faculté de Médecine, moresistance via targeting PD-L1. Biotechno. Lett. 39,1485–1492 (2017). Liège Université de Liège, Liège, Belgium. Technical University Darmstadt 22. Gaali, S. et al. Selective inhibitors of the FK506-binding protein 51 by induced Institute of Organic Chemistry and Biochemistry, Darmstadt, Germany fit. Nat. Chem. Biol. 11,33–37 (2015). 23. Giordano, A. et al. Tirofiban counteracts endothelial cell apoptosis through the VEGF/VEGFR2/pAkt axis. Vasc. Pharmacol. 80,67–74 (2016). 24. Romano, S. et al. Role of FK506 binding protein 51 [FKBP51] in the control of Conflict of interest apoptosis of irradiated melanoma cells. Cell Death Differ. 17, 145–157 (2010). The authors declare that they have no conflict of interest. Official journal of the Cell Death Differentiation Association D’Arrigo et al. Cell Death Discovery (2019) 5:137 Page 12 of 12 25. Vermes, I.,Haanen, C.,Steffens-Nakken,H.& Reutelingsperger,C. A novelassay 27. Schmittgen, T. D. & Livak, K. J. Analyzing real-time PCR data by the com- for apoptosis. Flow cytometric detection of phosphatidylserine expression on parative C(T) method. Nat. Protoc. 3,1101–1108 (2008). early apoptotic cells using fluorescein labelled Annexin V. J. Immunol. Methods 28. Kim, W. et al. Real-time imaging of glioblastoma using bioluminescence in a 184,39–51 (1995). U-87 MG xenograft model mouse. J. Korean Soc. Appl. Biol. Chem. 58,243–248 26. Lin, D. et al. Tenomodulin is essential for prevention of adipocyte accumu- (2015). lation and fibrovascular scar formation during early tendon healing. Cell Death Dis. 8, e3116 (2017). Official journal of the Cell Death Differentiation Association

Journal

Cell Death Discovery – Springer Journals

Published: Sep 24, 2019

There are no references for this article.