1. Introduction
Pancreatic cancer, with pancreatic ductal adenocarcinoma (PDAC) being the most common subtype, presents one of the top tumor types with unmet needs and only minimal advances in treatment outcome in recent years [
1,
2]. PDAC is the fourth leading cancer in tumor-related deaths and despite extensive research advances, the 5-year survival is still poor, with 7% [
2]. Tumor resection is the only curative regimen, however, only less than 20% of the patients profit from it. The challenging high therapy resistance of PDAC arises from various properties, such as a highly desmoplasmic microenvironment, genetic mutations and late diagnosis at already advanced or metastasized stages [
2,
3].
Due to the delicate localization of the pancreas close to the gastrointestinal tract, radiotherapy with photons is frequently used for metastatic PDAC or palliatively [
2]. With great technical advances, proton beam therapy is, however, only applied at specific clinics worldwide. Based on its physical properties, treatment with proton beam radiotherapy confers better sparing of organs at risk and is currently discussed as a preferable treatment modality in comparison to standard photon radiotherapy [
4]. It is the more optimal dose-depth profile of the proton beam that provides a higher precision. This is characterized by a low entrance dose and an increase at the stop, the so-called Bragg peak. The Bragg peak can be spread out (spread-out Bragg peak, (SOBP)) by combining various beams, each with a different initial energy, over the tumor volume and positioning them precisely [
5,
6].
Despite the first promising clinical results obtained for proton beam therapy of other tumor entities, little is known for PDAC. It further remains open whether proton irradiation induces similar or different modifications to the DNA and to signal transduction events and whether such differences might be exploitable for combinatorial therapies with biologicals [
7,
8]. In general, combining photon and proton irradiation with DNA repair inhibitors appears promising and may be superior to targeting of receptor tyrosine kinases (RTK) or cytoplasmic protein kinases, as the DNA remains as most important target in the context of cell survival after exposure with ionizing radiation.
To comparatively elucidate the efficacy of proton versus photon irradiation and to identify novel potential therapeutic targets for these two radiation types in PDAC, we cultured PDAC cell lines under more physiological three-dimensional (3D), matrix-based conditions and examined their tumoroid formation capacity and phosphoproteome modifications. Our results indicate a cell-line-dependent higher efficacy for protons over photons regarding 3D PDAC tumoroid growth. Combination treatments with RTK, cytoplasmic protein kinase and DNA repair inhibitors showed an inhibitor- and cell-line-specific cytotoxic and radiosensitizing potential, which presented as equipotent for photons and protons and with a strong non-homologous endjoining (NHEJ) dependency in our PDAC cell line panel.
3. Discussion
Given the higher precision accompanied by optimized sparing of normal tissue, proton beam irradiation is considered more favorable for PDAC patients than photon irradiation. However, supportive large clinical datasets, as well as systematic preclinical insights, are lacking. To address this point, we conducted a preclinical study in a panel of PDAC cell lines grown in 3D extracellular matrix with and without molecular-targeted pretreatments. Our study revealed an equal or higher efficacy of low-LET protons over photons in terms of reducing PDAC tumoroid formation. Moreover, we showed a greater extent of phosphoproteome alterations upon proton irradiation compared with photon irradiation. The targeting of proteins identified in the phosphoproteome uniquely altered by protons or photons failed to mediate marked radiation type-specific radiosensitization. Instead, inhibition of DNA repair proteins acting in NHEJ revealed a strong radiosensitizing potential independent of the radiation type.
These observations are in line with comparative survival analysis of proton versus photon irradiation in other tumor entities such as glioma stem cells [
9,
10] or lung cancer cells [
11]. Slightly different results, however, were found in cells cultured under 2D conditions [
12]. The higher efficacy of protons over photons reported in such studies might have been caused by differences in cell and nuclear size and chromatin organization; both parameters have been reported by us earlier for photon irradiation [
13]. Intensive discussion about the RBE as a clinically relevant parameter requires in vivo growth conditions to be determined. Generally, the RBE for protons is considered to be 1.1, but recent studies have already demonstrated its variability [
11,
12,
14,
15], which is also reflected in the PDAC cell line panel presented here.
Taking into consideration the therapeutic exploitability of the hallmarks of cancer represented by altered prosurvival signal transduction and the associated opportunities for treatment personalization [
16,
17], we undertook a comparative broad-spectrum phosphoproteome analysis of the response to proton versus photon irradiation 1 h post-irradiation. To identify common targets in PDAC cells, the most sensitive and most resistant cell lines to photon irradiation were chosen, revealing only a minor overlap of protein phospho-site modifications. The obtained results largely defeated our aim to find uniquely altered proteins to exploit them as sensitizers in PDAC cell lines towards either photon or proton irradiation. Several reasons might be causative, like the chosen early snapshot at 1 h after irradiation, the limited number of proteins on the array or the limited number of cell models tested. Future, more systematic examinations are warranted.
Nevertheless, we were able to discover three druggable targets from the array. Targeting the ER-α, unique for photon irradiation, and Chk1, for both photon and proton irradiation, resulted in moderate but significant radiosensitization irrespective of the radiation type. The ineffectiveness of trastuzumab-mediated HER2 deactivation in PDAC cell models observed in our study is in line with both preclinical and clinical findings but has not been addressed in combination with radiotherapy to date [
18,
19]. As trastuzumab cytotoxicity has been connected to immune cells to induce antibody-dependent, cell-mediated cytotoxicity in PDAC cells [
20], a second, non-antibody-based small molecule inhibitor, lapatinib, was applied prior to photon or proton irradiation. We hypothesize the higher efficacy of lapatinib, as compared with trastuzumab, to originate from its inhibitory spectrum against HER2 and EGFR. Concerning tamoxifen, our study is the first to identify a radiosensitizing potential towards both photon and proton irradiation in PDAC cells. For other cancer cell models, the role of ER-α is already known in the radiation response upon photon irradiation and attributed to its interactions with DNA repair proteins [
21]. With regard to Chk1 inhibition, Vance et al. showed a sensitization of pancreatic cancer cell lines towards photon irradiation [
22]. Likewise, Chk1 deactivation elicited radiochemosensitizing effects for photons together with gemcitabine [
23]. The body of literature for combined treatment with protons and biologicals is limited. Only one report outlined a higher degree of sensitization towards protons than to photons after Chk1 inhibition in triple-negative breast cancer MDA-MB-231 cells [
24]. Conversely, our results, especially the ones obtained in MiaPaCa-2 cells, indicate a higher efficacy of photons over protons.
As alternative approach, we widened the inhibitor spectrum, including candidates that are either validated molecular targets for other cancer types or are considered as potential cancer targets [
25]. Beyond the targets identified from our phosphoproteome array, a screening with a panel of signal transduction and DNA repair inhibitors provided clear evidence for a superiority of DNA repair inhibitors over inhibitors for receptor tyrosine kinase or cytoplasmic protein kinases. In contrast to others demonstrating afatinib, a pan-ErbB inhibitor, to radiosensitize PDAC cell lines [
26], we found no effect after specific ErbB2/HER2 targeting. This, however, is in line with the findings of Kimple et al. [
27]. PI3K inhibition, still under debate as potential cancer target [
28], did sensitize PDAC cells to photon and proton irradiation [
29]. Although the phosphoproteome array encompassed a part of these proteins such as DNA-PK, EGFR, MEK1, MDM2, p38 MAPK and VEGFR1/2, these proteins neither fulfilled our definition in terms of a ≥30% decrease or ≥50% increase in phosphorylation, nor did they overlap in the two tested cell lines. Yet, inhibition of these proteins, except for MDM2 and p38 MAPK, resulted in a cell-line-dependent and similar sensitization to photon and proton irradiation. These findings propose that the function of a protein in the radiation survival response upon photons and protons is not necessarily reflected by the detectable radiation-induced changes in phosphorylation or activity. Future studies are warranted, which analyze time points after irradiation beyond our 1 h post-irradiation snapshot, presented here.
Intriguingly, our conducted experiments revealed the efficacy of the selected inhibitors to be independent from the radiation type in 3D lrECM PDACcultures. Our data pinpoint two aspects: (i) the radiation type-unrelated strong dependence of PDAC cell survival on DNA repair; (ii) the great similarity in dependence on the same DNA repair machinery upon photon and proton irradiation. DNA double-strand breaks are repaired by two main DNA repair pathways, i.e., homologous recombination (HR) and NHEJ [
5,
8]. Our 3D lrECM PDAC cell culture panel showed clear a dependence on classical and alternative NHEJ, as indicated by DNA-PKcs and PARP inhibitors, in contrast to HR impairment using a Rad51 inhibitor. Likewise, Li et al. reported PDAC cell lines to strongly depend on DNA repair via NHEJ when irradiated with photons [
30]. In Panc-1 and KP4 cells, as well as in esophageal cancer cells, PARP inhibition resulted in sensitization towards proton irradiation [
31,
32]. In further studies in Ligase IV knockout mouse embryonic fibroblasts and DNA-PKcs-deficient glioblastoma cells, both resembling NHEJ deficiency, a similar radiosensitization to photons or protons was exhibited [
12]. We and others already reported radiosensitization by ATM inhibition in photon-irradiated 2D-cultured PDAC cells [
33,
34]. Here, we document the sensitizing potential of ATM inhibition in a panel of 3D-cultured, proton-irradiated PDAC cell lines. Intriguingly, the inhibition of the MRN complex, consisting of MRE11, Rad50 and NBS1 [
35], failed to mediate radiosensitization in contrast to ATM deactivation.
Concerning HR targeting, Colo357 cells were the only cell line strongly and marginally radiosensitized by Rad51 inhibition to photon and proton irradiation, respectively. Importantly, we were unable to confirm the reported greater cellular sensitivity towards protons under HR-deficiency in the here-presented 3D lrECM PDAC cell cultures [
31,
36,
37].
4. Materials and Methods
4.1. Antibodies
All used primary antibodies for Chk1, phospho-Chk1 S345, Estrogen Receptor-α, phospho-Estrogen Receptor-α S118, HER2/ErbB2 and phospho-HER2/ErbB2 Y1248 were purchased from Cell Signaling (Frankfurt, Germany). Secondary antibodies anti-mouse IgG, HRP conjugated and anti-rabbit IgG, HRP conjugated, were purchased from Pierce (Bonn, Germany).
4.2. Cell Culture
Human pancreatic ductal adenocarcinoma (PDAC) cell lines BxPC3, MiaPaCa2, Panc-1, and Patu8902 were purchased from the American Type Culture Collection (ATCC). Capan-1, and COLO357 cell lines were a kind gift from Chr. Pilarsky (University Erlangen-Nürnberg, Germany). The origin and stability of the cells were routinely monitored by short-tandem repeat analysis (microsatellites). All cell lines were cultured in Dulbecco’s modified Eagle medium (DMEM, Sigma-Aldrich, Taufkirchen, Germany) with 10% fetal calf serum (FCS, Sigma-Aldrich, Taufkirchen, Germany) and 1% non-essential amino acids (Sigma-Aldrich, Taufkirchen, Germany) at 37 °C with 8.5% CO2 at pH 7.4. In all experiments, asynchronously growing cells were used. All cells tested negative for Mycoplasma by using the mycoplasma detection kit Venor®GeM OneStep (Minerva Biolabs, Berlin, Germany).
4.3. 3D Tumoroid Formation Assay
PDAC cells were seeded for 3D tumoroid formation assay, as published [
38]. In brief, cells were embedded into lrECM (Matrigel™; BD, Heidelberg, Germany) at a concentration of 0.5 mg/mL in 96 well plates coated with 1% agarose. Twenty-four hours later, cells were irradiated with 2, 4 or 6 Gy of photons or protons or left unirradiated. In case of inhibitor treatment, cells were pretreated for 1 h before irradiation. After 24 h, inhibitors (except for Trastuzumab) were removed and replaced by fresh complete DMEM. After a cell line-dependent incubation period of 7–13 days, PDAC tumoroids consisting of a minimum of 50 cells were counted microscopically.
4.4. Radiation Exposure
4.4.1. Photon Irradiation
Cells were irradiated at room temperature using 2, 4, or 6 Gy single doses of 200-kVp X-rays (Yxlon Y.TU 320; Yxlon; dose rate ≈ 1.3 Gy/min at 20 mA) filtered with 0.5-mm Cu, as published [
39]. The absorbed dose was measured using a Semiflex ionization chamber (PTW Freiburg; Freiburg, Germany). Cells in tissue culture plates were irradiated horizontally both for tumoroid formation assays (96-well plates) and whole-cell lysates (24-well plates) as published [
39].
4.4.2. Proton Irradiation
Proton irradiation (low-LET of 3.7 keV/µm) was performed at the horizontal fixed-beam beam line in the experimental hall of the University Proton Therapy Dresden (UPTD). For 150 MeV protons, a dedicated beam shaping system consisting of a double-scattering device and a ridge filter provides a laterally extended 10 × 10 cm
2 proton field and a SOBP of 26.3 mm (90% dose plateau) in water. Cells in tissue culture plates were irradiated at room temperature using two different setting to assure mid-SOBP position. For tumoroid formation assays 96-well plates were placed perpendicular to beam axis (90°), whereas for whole-cell lysates 24-well plates were positioned at 42° relative to beam axis to avoid destruction of 3D structure [
40]. For absolute dosimetry, a Markus ionization chamber (PTW) readout by an Unidos dosemeter (PTW) at sample position was applied. Details of absolute dosimetry and beam control are given in Beyreuther and colleagues [
41].
4.4.3. Calculation Relative Biological Effectiveness (RBE)
The RBE values were calculated as follows (1):
The doses (D) were calculated as follows: data points were fitted in the linear-quadratic formula and the values for α and β were taken from the fitted curve. D was then solved at a SF of 50% from the following Formula (2):
4.5. Total Protein Extraction and Western Blotting
Whole-cell lysates, SDS-PAGE and Western blotting were performed as previously described in [
39]. In brief, cells cultured in 0.5 mg/mL Matrigel were irradiated after 24 h or left unirradiated. One hour after irradiation, whole-cell lysates were harvested with RIPA lysis buffer containing a protease inhibitor (complete protease inhibitor cocktail from Roche, Mannheim, Germany) and phosphatase inhibitors (Na3VO4 and NaF from Sigma-Aldrich, Taufkirchen, Germany). After incubation for 30 min on ice, cells were mechanically lysed using a syringe followed by centrifugation at 13,000×
g for 20 min to remove debris. The chemiluminescent detection was performed using ECL™ Prime Western Blotting System (Sigma-Aldrich). Densitometric analysis was carried out with ImageJ. In
Figure S7, the original uncropped images of Western Blot are displayed.
4.6. Inhibitors and Reagents
Cells were treated with pharmacological inhibitors for ATM (KU55933, Calbiochem, San Diego, CA, USA; 10 µM) Chk1 (Prexasertib, Selleckchem, Houston, TX, USA; 1 and 3 nM), DNA-PK (NU7026, Selleckchem; 10 µM), EGFR (Tarceva®, Roche, Basel, Switzerland; 10 µM), ER-α (Hydroxy-Tamoxifen, Sigma Aldrich; 10 µM), HER2 (Ontruzant®, MSD SHARP & DOHME GMBH, Haar, Germany; 2 µg/mL and Lapatinib, Selleckchem; 1 µM), MAPK (SB203580, Selleckchem; 10 µM), MDM2 (AMG232, Axon Medchem, Groningen, The Netherlands; 10 µM), MEK (PD98059, Selleckchem; 20 µM), MRNcomplex (Mirin, Sigma Aldrich; 10 µM), PI3K (LY294002, Selleckchem; 10 µM), PARP (Olaparib, Cell Signaling, Frankfurt a. M., Germany; 10 µM), Rad51 (B02, Axon Medchem; 10 µM) and VEGFR (Axitinib, Sigma Aldrich; 1 µM) with the indicated concentrations. Ethanol (for Hydroxy-Tamoxifen), IgG (for Ontruzant) and DMSO (for all other inhibitors) were used as controls.
4.7. Phosphoproteome Analysis
Colo357 and MiaPaCa-2 cells were cultured in 0.5 mg/mL lrECM and exposed to 6-Gy photon or proton irradiation after 4 days or unirradiated. To determine differences in the early events on the molecular level, here the DNA damage response upon proton and photon irradiation, whole cell lysates were harvested 1 hour post-treatment as previously published [
42]. The samples were transferred to Full Moon BioSystems Inc. on dry ice for conducting the Phospho Explorer Antibody Microarray. Proteins were biotin-labeled and put on preblocked microarray slides. Detection of total and phosphorylated proteins was carried out by the use of Cy3-conjugated streptavidin. The array consists of antibodies against 342 proteins and 606 phospho-sites. For analysis, protein phosphorylation was normalized to corresponding total protein expression. Proteins with phosphorylation site changes of at least a 30% decrease or 50% increase (arbitrary cut-off) were considered relevant and selected, and changes in both cell lines comparatively analyzed.
4.8. Statistics
Means ± standard deviation (SD) of at least three independent experiments were calculated. For statistical significance analysis of tumoroid formation capacity, two-sided Student’s t-test was performed with Excel (Microsoft) and a p value of less than 0.05 was considered statistically significant.