Paracrine Placental Growth Factor Signaling in Response to Ionizing Radiation Is p53-Dependent and Contributes to Radioresistance

Radiotherapy (RT) is widely used alone or as part of a combined treatment modality with surgery and chemo-/immunotherapy for the management of solid tumors (1–5). In addition to DNA damage and genome instability, ionizing radiation (IR) also leads to stress responses in the irradiated tumor cells by activating signal transduction pathways and inducing secretion of several auto- and paracrine factors (6, 7). We recently performed an exhaustive semiquantitative dot blot secretome analysis from irradiated tumor cells and identified among others that placental growth factor (PlGF) is secreted from tumor cells in response to irradiation. A secondary analysis revealed that multiple of these factors, but not PlGF, are substrates of the matrix metalloproteinase ADAM17 and that irradiation enhances its sheddase activity in a dose-dependent way (8).

PlGF is an N-glycosylated, homodimeric protein and belongs to the VEGF family, with 53% similarity to the platelet-derived growth factor (PDGF)-like region of VEGF (9). The PlGF gene contains seven exons and due to alternative splicing exists in four isoforms consisting of 131, 152, 203, and 224 amino acids (10, 11). Under normal conditions, PlGF expression is low in organs such as heart, skeletal muscle, and lungs (12–14). Binding of PlGF to its receptor VEGFR1 leads to phosphorylation of the receptor, activation of downstream pro-angiogenic signaling pathways, and a crosstalk between VEGFR1 and VEGFR2 (15–18). In comparison to VEGF-A, PlGF is not extensively investigated and its regulation and role in response to irradiation is poorly understood. PlGF is highly expressed in the placenta, however, its deletion does not affect embryonic development in mice (16). Interestingly, although PlGF-deficient mice develop normally, lack of PlGF results in reduced tumor vascularization and tumor growth (16, 17).

Several studies demonstrate the contribution of PlGF to tumor angiogenesis under pathologic conditions. Tumor stage, metastasis, and poor overall survival correlate to increased PlGF levels in different tumors entities (19–24). Of note, inhibition of VEGF and its receptors leads to increased PlGF levels, probably contributing to escape and resistance against these treatment modalities (24–27). On the preclinical level as investigated in several mouse models, blocking of PlGF by RNA interference, neutralizing antibodies, or gene silencing resulted in decreased angiogenesis, reduced tumor growth and dissemination (16, 24, 28–32). Interestingly, although blockage of PlGF normalized tumor vessels in some tumor models (28, 33), tumor vessel normalization was also observed in PlGF overexpressing tumors (34).

Expression of PlGF is regulated by several transcription factors, however contradictory results exist on the regulation of PlGF by the tumor suppressor p53 and the hypoxia-inducible factor HIF1α (9, 16, 18, 29, 35–37). Furthermore, almost no data exist on the regulation of PlGF in response to irradiation (38). Here we investigate expression and secretion of PlGF across multiple tumor cell lines in response to irradiation and its role for tumor angiogenesis in response to radiotherapy.

Human lung adenocarcinoma cells (A549, H460, H358, H292, H125, HCT116; ATCC) were cultured in RPMI1640 media (Gibco; ref. no. 22409-015), medulloblastoma cells (D341, D425, DAOY, UW228; kindly provided by Martin Baumgartner, Children's Hospital Zurich) in improved MEM media (Richter's modification; Gibco) and head and neck squamous cell carcinoma cells (FaDu; ATCC) in DMEM media (Gibco). All media were supplemented with 10% (v/v) FCS (Gibco; ref. no. 10270-106), 1% (v/v) penicillin–streptomycin (Gibco), and 1% (v/v) Glutamax (Gibco) and cells were kept at 37°C in 5% CO₂. HUVECs, endothelial cell growth medium (ECGM) and supplementary growth factors were purchased from Promocell. Human primary lung fibroblasts, fibroblast basal medium, and fibroblast growth kit were purchased from ATCC. All cells were regularly tested for mycoplasma infection (MycoAlert Mycoplasma Detection Kit, #LT07-118; Lonza). Cells were pretreated with Pifithrin-α (S2929), MI-773 (S7649), BAY87–2243 (S7309; all purchased from Selleckchem) in given concentrations 1 to 2 hours prior to irradiation with an Xstrahl 200 kV X-ray unit at 1 Gy/min or an RS-2000 225 kV irradiator at 4.2 Gy/min (Rad Source). Hypoxia experiments were performed with cells irradiated and/or treated with pharmacologic agents followed by incubation at 1% O₂.

Bioplex Biomarker Cancer Panel assay (32 biomarkers) was performed with undiluted conditioned medium samples according to the manufacturer protocol (Bio-Rad). Briefly, beads covalently coupled to a capture antibody directed against specific biomarker were incubated with standards or samples (supernatant derived from non-irradiated and irradiated A549 cells). After a washing step, beads are incubated with biotinylated detection antibodies. The unbound biotinylated antibodies were washed away and the beads were incubated with a fluorescence reporter streptavidin–phycoerythrin conjugate (SA-PE), forming the detection complex. The fluorescence of each bead with bound SA-PE was measured when the beads were passed through the two laser bioplex reader. The data are presented as median fluorescence intensity (MFI) and concentration (pg/mL) using a Bioplex Manager software, where concentration is proportional to MFI of the reporter.

Secreted PlGF concentration was detected in filtered conditioned media (CM) at the indicated time points after irradiation using a PlGF DuoSet ELISA Kit according to manufacturer's guidelines (R&D Systems). The absorbance was determined with the plate reader EL808 Ultra Microplate Reader (Bio-Tek Instruments, Inc.) at 450 nm excitation and 570 nm emission, and normalized to live cell count using EVE Automatic cell counter (NanoEnTek).

Sample mRNA expression was determined at 4 and 24 hours after irradiation. Cell lysate collection and RNA isolation were performed using an RNeasy Mini Kit (Qiagen) according to manufacturer's guidelines. RNA was reverse-transcribed using High Capacity cDNA Reverse Transcription Kit (Applied Bioscience) and cDNA was amplified using SYBR Green Master Mix (Roche) with the following primers: (5′-3′): GAPDH forward: AACGGATTTGGTCGTATTGGGC; GAPDH reverse: TTGATTTTGGAGGGATCTCG; PlGF forward: TGTCACCATGCAGCTCCTAA; PlGF reverse: AGCATCGCCGCACCTTTC; p53 forward: CCCTTCCCAGAAAACCTA; p53 reverse: CTCCGTCATGTCCTGTGA. qRT-PCR was performed on LightCycler 480 (Roche).

Transfection was performed by using reverse transfection with Lipofectamine RNAiMAX (Invitrogen) in antibiotic-free medium. Cells were reseeded into 6-well plates with fresh medium supplemented with antibiotics 24 hours thereafter. siLuc-RNA was synthesized by Microsynth with the following sequence: (5′-3′): CGUACGCGGAAUACUUCGATT. sip53-RNA was purchased as a pool of four siRNAs (Dharmacon). Irradiation was performed 48 hours after siRNA transfection.

Whole cell extracts were collected using Laemmli buffer and samples were boiled for 5 minutes at 95°C. 40 to 50 μg of protein were separated by SDS-PAGE and transferred to PVDF membranes. The membrane was then blocked in 5% nonfat milk and incubated with primary antibodies [rabbit polyclonal anti-p53 (GTX102965; GeneTex), rabbit monoclonal anti-p21 (2947S; Cell Signaling Technology), rabbit monoclonal anti-HIF1α (D2U3T; Cell Signaling Technology), mouse monoclonal anti-β-actin (A1978; Sigma)] at 4°C overnight. After three washings, HRP conjugated secondary antibodies [mouse anti-rabbit (SC-2357; Santa Cruz Biotechnology) or sheep anti-mouse (NA931V; GE Healthcare)] were added for 1 hour at room temperature followed by additional washing steps. The membrane was developed with the ECL system (Amersham Bioscience) according to manufacturer's protocol in Fusion FX (Vilber).

Two separate sgRNAs targeting different regions of the third exon of the PGF gene (target sequence 1: GAATCTGCACTGTGTGCCGG; target sequence 2: CGTGTCCGAGTACCCCAGCG) were selected a using freely available online tool (CRISPOR). Likewise two separate sgRNAs targeting different regions of the third exon of the TP53 gene (target sequence 1: GAATCTGCACTGTGTGCCGG; target sequence 2: CGTGTCCGAGTACCCCAGCG) were selected. SgRNA oligos were synthesized by Microsynth. The single-stranded oligomers were annealed and ligated into GFP and Cas9 expressing pLentiCRISPR-EGFP plasmid (#75159; Addgene) using the Golden Gate assembly cloning strategy. The ligation mix was transformed into competent bacteria by heat shock. Clones were identified by colony-PCR using an U6 promoter specific forward primer (5′-GACTATCATATGCTTACCGT-3′) and the reverse sgRNA oligo as reverse primer. Positive colonies were inoculated for mini-prep (Sigma) culture. HEK293T (ATCC) cells were used for production of lentiviral constructs according to guidelines from Cellecta. Filtered lentiviral supernatant was used to transfect the target cells (D341). To prepare cells for flow cytometry, live cells were harvested, resuspended in PBS with 2% FBS, and filtered using a 40 μm cell strainer (BD Falcon). Cell sorting was performed with a FACSAriaTM III Cytometry System (BD Biosciences). Sorted cells were further subcloned through limited dilution and sequenced before further experiments. Each subclone was named after their respective well number, that is, control cell lines C4, C6, and knockout (ko) cell lines B5, B7, G7, and F8.

For the transwell migration assay, 24-well transwell units (6.5 mm) with 1 μm pore size PET membranes (Greiner Bio-One; 662610) were used according to the manufacturer's instructions. Briefly, 3 × 10⁵ attracting cells (D341 PlGF-wt or PlGF-ko) were plated into the lower chamber of the transwell containing 1,000 μL ECGM supplemented with 1% FBS and without growth factors and allowed to attach for a minimum of 6 hours. The plate was sham irradiated or irradiated with 5 Gy. Next, 3 × 10⁴ endothelial cells in 200 μL of the same ECGM medium as above were seeded into the upper chamber of the transwell inserts and allowed to attach for a minimum of 6 hours. Thereafter, the inserts were irradiated with 5 Gy and immediately placed on the wells harboring the attracting cells. The co-culture was maintained at 37°C in 5% CO₂ for 48 hours. For quantification, cells from the upper side of the insert were scraped away with a cotton swap and inserts were then fixed in methanol/acetic acid (75%/25%, v/v), dried, and stained with DAPI (1:25,000) in 99% MetOH. Fluorescent microscopy pictures were taken (Leica 7000DT) and the migrated cells were counted manually in at least three images at 20× magnification per insert.

D341 PlGF control and PlGF-ko subclones (4 × 10⁶ cells in 150 μL PBS) were injected subcutaneously in the back of 8-week-old, female athymic CD1 nude mice (Charles River). Tumor volumes were determined with caliper according to the formula (L × l²)/2. Treatment started when tumors reached a volume of 200 mm³ ± 20%. Tumors were sham-irradiated or irradiated with a single dose of 5 or 10 Gy using an image-guided small animal radiotherapy platform (Precision X-Ray, X-Rad SmART) 225 kV unit with a dose rate of 3 Gy/min, equipped with a cone beam CT (CBCT) scanner. Precise irradiation plans were designed with the corresponding SmARTPlan software. Radiotherapy was applied with two opposing fields. Animals were kept under 3% isoflurane anesthesia for imaging and treatment. Eight days after treatment, tumors were harvested and immediately fixed in formalin. All in vivo experiments were performed according to guidelines for the welfare and use of animals of the Veterinäramt Kanton.

Immunohistologic endpoints were analyzed on formalin-fixed paraffin-embedded (FFPE) 4 μm tissue sections derived from PlGF control or PlGF ko-derived tumor xenografts for hematoxyline and eosine (H&E), smooth muscle actin (SMA; 1:50; M0851; Dako), and CD31 (1:10; M0823; Dako). Images were taken on a wide-field Nikon Eclipse TI microscope (Minato). Amount of vessels were counted in at least 10 different fields in each xenograft.

Cellular experiments were performed at least three times and statistical analysis was performed using GraphPad Prism v.8. Data were tested for normality using the Shapiro–Wilk test. Tumor volumes over time were evaluated by using linear regression and the slopes of the best linear fit line were statistically analyzed. In all cases, two groups are compared by a two-side t test (or Mann–Whitney test for data not following the normal distribution) and multiple groups are compared by one-way ANOVA (or Kruskal–Wallis test for data not following the normal distribution), followed by Fisher LSD test in the case of planned comparisons and the Holm–Sidak correction otherwise. For all experiments data are represented as mean ± SD, *P < 0.05, **P < 0.01, and ***P < 0.001. P values are indicated for P < 0.05.

Secretome analysis by semiquantitative antibody array screening previously indicated that PlGF is secreted in response to irradiation (8). Additional bioplex-based quantification of 32 major cancer biomarkers in the supernatant from irradiated A549 lung adenocarcinoma cells now revealed that PlGF is most strongly increased among these cancer biomarkers in response to irradiation (Fig. 1A). To analyze irradiation-induced PlGF-expression in more detail and across multiple tumor entities (medulloblastoma, lung adenocarcinoma, colon carcinoma; D341, DAOY, H125, A549, HCT116), tumor cells were irradiated, supernatants collected at multiple time points after irradiation and analyzed for PlGF secretion by ELISA. IR-enhanced PlGF-secretion was identified to different extents in the CM from these cell lines in a time- and IR-dose-dependent manner (Fig. 1B and C for the medulloblastoma cell lines D341 and DAOY, and Supplementary Figs. S1B–S1D for the cell lines derived from other tumor entities). Similar to the original bioplex analysis, PlGF secretion was also most strongly enhanced in the D341 medulloblastoma cell line relative to other cancer biomarkers determined (Supplementary Fig. S1A).

Enhanced secretion in response to irradiation might be related to IR-induced PGF-transcription. Upregulation of PlGF mRNA was observed in several cancer cell lines as early as 4 hours after irradiation (D341, A549), whereas only minimally or only observed in a delayed way in other cell lines 24 hours after irradiation (DAOY; Fig. 2A–F; see Supplementary Fig. S2 for other cell lines, H292, H460, H358, FaDu, UW228, H125, HCT116, HUVEC, HPLFB). On the basis of a literature search on the mutational status of Kras and p53 in these cell lines (Supplementary Table S1), this differential IR-induced expression pattern of PlGF could be correlated with the mutational status of the tumor suppressor and transcription factor p53.

Interestingly, cells expressing wild-type p53 demonstrated IR-induced PlGF transcription as early as 4 hours after irradiation, which correlated with early PlGF secretion. In contrast cells expressing mutant p53 did not show any changes in the PlGF mRNA level at this early time point after irradiation, suggesting that p53 regulates immediate IR-induced PlGF expression. Of note, delayed expression of PlGF in the p53 mutated cells in response to irradiation correlated with delayed PlGF secretion into the CM of the respective cells (48-hour time point; see above Fig. 1). Due to their high PlGF expression and secretion levels, and that treatment of medulloblastoma patients includes radiotherapy (39), further experiments were primarily performed with p53 wild-type D341 cells.

To determine regulation of PlGF expression and secretion by p53, D341 and DAOY medulloblastoma cells were treated with increasing concentrations of the MDM2-inhibitor MI-773. MDM2 downregulates the transcriptional activity of p53 by direct binding to the tumor suppressor p53, its ubiquitination and subsequently inducing its proteasomal degradation (40). Thereby, pharmacologic disturbance of the MDM2–p53 interaction might lead to p53 stabilization and increased transcriptional activity independent of irradiation. Cellular treatment with MI-773 rapidly induced PlGF expression in the D341 but not in the DAOY cells and resulted in enhanced PlGF levels in the CM of D341 cells as determined 24 and 48 hours after treatment start (Fig. 3A–D). MI-773-induced PlGF expression and secretion were confirmed with p53-wild-type (wt) HCT116 colon carcinoma cells (Supplementary Figs. S3A and S3B).

Putative regulation of PlGF by p53 was more specifically investigated using p53-directed siRNA. p53 mRNA and protein levels were downregulated on siRNA treatment for up to 96 hours. Likewise, expression of the well-known p53 downstream target p21 was also downregulated (Supplementary Figs. S3C and S3D). Tumor cells were irradiated 48 hours after siRNA transfection, and PlGF secretion was determined in the CM 48 hours after irradiation. PlGF levels were strongly enhanced in response to irradiation in siLuc-pretreated cells, but basal and irradiation-induced PlGF levels were significantly decreased in CM of sip53-pretreated cells (Fig. 3E). Eventually, D341 p53-ko cells were generated. IR-induced PlGF expression and secretion was completely abrogated in p53-ko D341 tumor cells (Fig. 3F and G).

MDM2 has additional binding targets besides p53 and thereby MI-773 might reactivate also other transcriptional activities besides p53 leading to enhanced PlGF expression. Therefore, PlGF expression and secretion was also analyzed in siRNA-p53-downregulated and D341 p53-ko cells on treatment with MI-773 for 4 and 24 hours. Neither expression nor secretion of PlGF were enhanced by MI-773 in these cells (Fig. 4A–C; see also Supplementary Figs. S3D–S3H for colon carcinoma cells).

To analyze dependence on p53 transcriptional activity, additional experiments were performed with the p53-inhibitor Pifithrin (PFT), which prevents nuclear translocation of p53 and transcriptional activation of downstream genes (41, 42). Prior to irradiation, cells were preincubated with Pifithrin for 2 hours and the PlGF level determined 24 hours thereafter. Similar to the experiments performed with the MDM2-inhibitor and p53-directed siRNA, decreased PlGF levels in Pifithrin-pretreated cells indicate p53-mediated PlGF regulation in response to irradiation (Fig. 4D and E; see also Supplementary Fig. S3I for colon carcinoma cells). Overall, these complementary experiments strongly demonstrate that p53 is a major transcriptional regulator for PlGF expression.

To determine whether p53 truly binds to the promoter region of PlGF, we performed additional ChIP-qPCR assay. Unfortunately, we could not identify a p53 binding site on the PGF promoter using p53-antibodies and PlGF site-specific primers mentioned by Rashi-Elkeles and colleagues (38).

Both p53 and the hypoxia inducible transcription factor-1 (HIF1) regulate cellular transcriptional responses to hypoxia and to ionizing radiation. At the same time, an intriguing interplay exists between p53 and HIF1, including transregulation and competition for limiting cofactors (43). Of note, contradictory results exist on the expression of PlGF under hypoxic stress conditions, which might be cell type and genetic background dependent (35–37). To investigate p53-dependent and IR-induced PlGF secretion under hypoxic conditions, D341 medulloblastoma cells were irradiated followed by incubation under normoxic and hypoxic (1% O₂) conditions. In addition, cells were pre-incubated with increasing concentrations of BAY 87–2243 (0, 10, and 50 nmol/L) 1 hour prior to IR. BAY87–2243 is a potent inhibitor for hypoxia-induced, HIF1α- and HIF2α-mediated gene activation (44). Hypoxia minimally enhanced basal and IR-induced PlGF secretion, and cellular pretreatment with BAY 87–2243 did not interfere with PlGF secretion neither under normoxic nor hypoxic conditions (Fig. 5A). PlGF secretion was also determined in CM derived from D431 cells transfected with p53-oriented siRNA and incubated under hypoxic conditions (1% O₂). Similar to the experiments performed under normoxic condition (see above), downregulation of p53 by p53-directed siRNA strongly decreased irradiation-induced PlGF secretion also under hypoxic condition (Fig. 5B). Basal PlGF secretion was only minimally increased under hypoxic conditions in comparison to normoxic conditions in both D341-p53-wt and D341-p53-ko cells. Irradiation increased PlGF secretion only from D341 p53-wt cells under hypoxic conditions, which was not affected by cellular pretreatment with the HIF1 inhibitor BAY 87–2243 (Fig. 5C). Detection of HIF1α in non-irradiated and irradiated cells by Western blotting demonstrated hypoxia-upregulated HIF1α under hypoxic conditions in both D341-p53-wt and D341-p53-ko cells, which was absent in cells pretreated with BAY 87–2243 (Fig. 5D). These results strongly corroborate that regulation of PlGF secretion under normoxic and hypoxic conditions is primarily p53- and not HIF-signaling dependent.

To further investigate auto- and paracrine effects of IR-induced secretion of PlGF, D341 PlGF-ko cells were generated using two different exon 3-targeting single guide RNAs of the PGF gene and cloned into a GFP and Cas9 expressing pLentiCRISPR plasmid. Sequencing of the targeting region of exon 3 in our cell clones confirmed successful PGF-gene targeting (Supplementary Fig. S4A). IR-induced PlGF secretion was determined in the cell supernatants derived from a control cell clone and a selected knockout cell clone 24 and 48 hours after irradiation. PlGF was undetectable in the supernatants of PlGF-ko cell lines and IR-enhanced PlGF levels were only determined in the supernatants of the control cell line (PlGF-wt; Supplementary Fig. S4B). To investigate a paracrine effect of PlGF, we performed a Boyden chamber migration assay with irradiated human umbilical vein endothelial cells (HUVEC) seeded in the upper chamber migrating towards non-irradiated and irradiated D341 PlGF-wt and D341 PlGF-ko cells, respectively. No quantitative difference in HUVEC migration towards unirradiated PlGF-wt versus unirradiated PlGF-ko cells could be determined. However, significantly increased migration of HUVECs towards irradiated PlGF-wt cells versus irradiated PlGF-ko cell could be observed (Fig. 6A).

Prior to in vivo experiments, proliferative activity and clonogenicity of PlGF-wt and PlGF-ko cell clones were determined and revealed only minor and PlGF-independent differential readouts (Supplementary Figs. S4C and S4D).

To determine a role of IR-regulated PlGF expression on the endothelial compartment in vivo, medulloblastoma xenografts were developed from subcutaneously-injected PlGFwt and PlGF-ko tumor cells. Overall, 83% to 87% of mice injected with PlGF-wt cells and 52% to 58% of mice injected with PlGF-ko cells developed tumors. Days to reach the starting volume ranged from 31 to 37 days for PlGF-wt and 36 to 50 days for PlGF-ko cells, respectively. Tumors were irradiated with increasing doses of IR (0, 5, and 10) and mice were sacrificed 8 days following irradiation. Tumor sections were stained for H&E, CD31, and SMA and the amount of vessels were counted in at least 10 fields for each tumor. Microvessel densities (MVD; CD31-staining) in PlGF-wt and PlGF-ko medulloblastoma xenografts were comparable, and MVD only minimally changed in PlGF-wt tumor xenografts in response to irradiation. Interestingly though, a decrease in MVD was observed in tumor xenografts derived from PlGF-ko-medulloblastoma cells irradiated with 5 Gy and MVD was strongly reduced on irradiation with 10 Gy (Fig. 6B). A quantitative similar outcome was determined on smooth muscle actin (SMA) staining as an indicator of pericyte coverage and vessel functionality (Fig. 6C). Representative staining of tumor sections are shown in Fig. 6D.

PlGF-wt and PlGF-ko cells displayed comparable proliferative activity and radiosensitivity in vitro (see above). Interestingly, the determination of a short-time efficacy-oriented endpoint in vivo revealed statistically significantly enhanced tumor shrinkage of PlGF-ko-tumors in comparison with PlGF-wt tumors in response to the same dose of irradiation (Fig. 7). Furthermore three of five PlGF-ko tumors (60%) in each treatment group (5 and 10 Gy) showed complete tumor regression in comparison with none of PlGF-wt tumors in the group of mice treated with 5 Gy and only one PlGF-wt tumor in the group of mice treated with 10 Gy of IR. These results demonstrate enhanced radiosensitivity of PlGF-ko tumors in comparison with PlGF-wt tumors and suggest that the vasculature protective effect of tumor-derived PlGF co-determines tumor radiosensitivity in vivo.

Overall, these results suggest that tumor-derived PlGF has a tumor vasculature protective effect and thereby co-determines tumor radiosensitivity in vivo.

Anti-angiogenic agents are mainly directed against the VEGF/VEGFR signaling pathway. However, targeting the VEGF/VEGFR pathway eventually leads to resistance and to an angiogenic switch to other growth factors, such as PlGF (29). The expression level of placental growth factor is increased in several cancer types and correlates with poor survival, cancer progression, and resistance to therapy (19–22). Patients with metastatic colorectal cancer have shown increased PlGF levels after combined treatment with bevacizumab, chemotherapy, and radiotherapy (45, 46). Furthermore, increased PlGF expression has been linked to increased vessel number, size, and permeability (24) and targeting of PlGF demonstrated reduced tumor vascularization in VEGFR-inhibitor-resistant tumor models (29). However, clinical studies with PlGF neutralizing antibodies did not show improved survival in combination with bevacizumab in previously treated patients with glioblastoma and therefore the strategy to target PlGF was discontinued, despite minimal normal tissue toxicities (47–49).

Here, we investigated PlGF expression and secretion in multiple cancer cell lines with differential genetic backgrounds in response to increasing doses of IR. Early, treatment-induced PlGF expression could be directly linked to the p53-status of the irradiated cells. Other transcription factors, could play a role in PlGF regulation in cells with mutated p53 and could be relevant for delayed expression in response to irradiation.

The PlGF promoter contains binding sites for multiple transcription factors including p53. It has four NF-κB, five metal transcription factor 1 (MTF1), three Sp1, one BF2 binding sites, and a predicted hypoxia responsive element (HRE) in its promoter. However, all the binding sites were identified in different cell types (HEK, immortalized mouse embryonic fibroblasts, HeLa cells) and under different conditions (37, 50, 51). First indications that p53 might regulate PlGF expression derive from a wide-scale transcriptome study in normal human B-lymphoblastoid cells (TK6 cells), investigating the transcriptional activity of p53 in response to ionizing radiation. Binding of p53 to the PlGF-promoter was also confirmed by ChIP assay but has not been further investigated beyond these transcriptional studies (38). Despite several attempts we could not perform successful ChIP assays demonstrating direct binding of p53 to the promoter site of PlGF in the medulloblastoma. This might be due to technical limitations or the intrinsic challenge to detect direct promoter binding of p53 due to the multiple layers of different transcriptional regulation of p53 downstream targets in response to stress (52). Nevertheless, our expression and secretion studies using different p53-interfering agents and performed in multiple cell lines clearly demonstrate that p53 plays a major role in PlGF-regulation in response to ionizing radiation.

Of note, cellular treatment with the MDM2 inhibitor MI-773 stabilized p53 and subsequently resulted in increased PlGF expression and secretion with similar kinetics to IR-induced PlGF expression and secretion in these p53-wt tumor cells. The PGF promoter has additional binding sites for other transcription factors than p53, which could also be affected by MDM2-inhibitor. Control experiments with the MDM2-inhibitor with concomitant downregulation of p53 and in p53-ko cells indicated, which p53 is the major transcription factor relevant for IR-mediated regulation of PGF in p53-wt tumor cells. IR-induced and p53-mediated secretion of PlGF corresponds to the categorization of PlGF into the class of senescence-associated secretory phenotype (SASP)-related cytokines (53). However, SASP was not further investigated in this report as several of the investigated cell lines do not undergo senescence in response to irradiation.

Due to the interplay between p53 and the HIF1, we hypothesized that other transcription factors, mainly HIFs, could be additional regulators for PlGF expression in irradiated tumor cells lacking p53, in particular under hypoxic conditions. Therefore PlGF secretion was determined in non-irradiated and irradiated p53-wt, p53-downregulated, and p53-ko tumor cells, respectively, both under normoxic and hypoxic conditions. Basal and IR-induced PlGF-secretion was comparable under normoxic and hypoxic conditions, and IR-enhanced PlGF secretion under hypoxic conditions was also strongly p53 dependent. HIF1-oriented inhibitory studies with BAY87–2243 confirmed that the transcription factor p53 and not HIFs is most relevant for IR-induced PlGF regulation. Our results relate to investigations performed by others, showing though increased PlGF expression under hypoxic conditions, but in a HIF1-independent way (50, 51, 54). In-depth mechanistic investigations are now required to identify additional, but so far unresolved mechanisms relevant for its minimal and delayed expression in tumor cells with a p53-mutated genetic background.

Tumor growth rate in the PlGF-ko medulloblastoma tumor xenografts was only slightly delayed, corresponding to previous reports on the influence of PlGF on basal tumor growth (16). However, PlGF-ko tumor xenografts were more radiosensitive with an increased rate of total remission in response to single high doses of IR. Reduced tumor growth and enhanced radiosensitivity, though, was not due to a differential proliferative activity and radiation response on the tumor cell level but might be linked to an enhanced secondary effect in response to irradiation on the tumor microenvironment (55) due to the lack of tumor cell-derived PlGF. Such PlGF-dependent, paracrine-mediated effects in response to irradiation could also be observed in vitro determining the migratory capacity of endothelial cells. Furthermore, the formation of the tumor vasculature was comparable in the growing PlGF-ko and PIGF-wt tumor cell-derived xenografts. However, the integrity of the vasculature was strongly reduced in the PlGF-ko tumor xenografts on irradiation and correlated with enhanced radiosensitivity of the PlGF-ko tumors in comparison to their wild-type counterparts.

Overall, these results indicate that PlGF plays a minor role for the formation of an intact vasculature during tumor growth, which might explain the lack of a therapeutic effect by PlGF-oriented inhibitory antibodies (47–49). On the other hand, PlGF has a strong tumor vasculature protective role in response to irradiation, which might be mediated by IR-induced secretion of PlGF from the irradiated tumor cells. Lack of this protective effect contributes to a microenvironment-mediated increase of tumor radiosensitivity. Thus, targeting of PlGF still represents an interesting rationale but might be more effective not as part of a single but as part of a combined treatment modality with radiotherapy, thereby overcoming a radiotherapy-induced treatment resistance mechanism, namely radiotherapy-induced PlGF secretion.

I. Telarovic reports grants from Swiss Academy of Medical Sciences during the conduct of the study. No disclosures were reported by the other authors.

T. Kazimova: Conceptualization, funding acquisition, investigation, methodology. F. Tschanz: Investigation, methodology. A. Sharma: Investigation. I. Telarovic: Data curation, validation. M. Wachtel: Methodology, writing–review and editing. G. Pedot: Investigation, methodology. B. Schäfer: Investigation, methodology, writing–review and editing. M. Pruschy: Conceptualization, formal analysis, supervision, funding acquisition, project administration, writing–review and editing.

This work was supported by grants from European Union's Horizon 2020 research and innovation program (Marie Sklodowska-Curie ITN RADIATE, 642623 to M. Pruschy), the Swiss National Science Foundation (172885 to M. Pruschy), and Hartmann Müller-Stiftung (2333 to T. Kazimova). The authors thank Department of Pathology, University Hospital Zurich, Institute of Anatomy and Center for Microscopy and Image Analysis, University of Zurich for their help with tumor sections. We also thank Martin Baumgartner for kindly providing the medulloblastoma cell lines.

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