β8 Integrin Mediates Pancreatic Cancer Cell Radiochemoresistance

Pancreatic ductal adenocarcinoma (PDAC) is one of the five most lethal malignancies in the world. Although the 5-year overall survival rate is about 15% to 20% for resectable patients (1, 2), most patients presenting at late stage with a treatment-refractory disease survive significantly less. Neoadjuvant chemotherapy [e.g., gemcitabine, nab-paclitaxel, FOLFIRINOX (folinic acid, 5-fluorouracil, irinotecan, and oxaliplatin)] is administered to patients whose tumors seem irresectable or borderline resectable. Alternatives such as radiotherapy and biologicals were neither systematically evaluated in large clinical trials nor showed great advantages over standard of care yet (2, 3).

One putative cancer target in PDAC is the PDAC cell/extracellular matrix (ECM) interaction as PDACs are stroma-rich tumors (4). This stroma consists of numerous ECM components such as collagens, laminins, fibronectin, and hyaluronic acid (5). Cells interact with ECM components via cell adhesion receptors that coalesce with receptor tyrosine kinases, adapter and signaling molecules to form focal adhesion complexes. These membranous multiprotein complexes are essentially coregulating key cell functions such as survival, cell death, proliferation, metastasis, and therapeutic resistance (6, 7). In previous work, we and others have documented the radiochemosensitizing potential of integrin and focal adhesion protein targeting, as the major family facilitating cell/ECM interactions. Examples for preclinically identified novel targets are β1 integrin (8), αvβ3 integrin (9), αvβ6 integrin (10), FHL2 (11), APPL1 and 2 (12), Caveolin 1 (8, 13), as well as small integrin-binding ligand N-linked glycoproteins (called SIBLINGs) or secreted protein acidic and rich in cysteine (called SPARC) families (14). Particularly, the anti-integrin approaches were exploited for molecular imaging (15), whereas biologicals inhibiting focal adhesion proteins (FAP) have not found their way into the clinic. Owing to the critical function of focal adhesions for prosurvival signaling in normal and cancer cells, we established a high-throughput esiRNA-based screening for more physiologic three-dimensional (3D) cell cultures to more systematically characterize the role of FAPs for PDAC radiochemoresistance. Intriguingly, we identified β8 integrin as top druggable target eliciting radiosensitization of PDAC cells.

β8 integrin, a 769-amino acid containing type I transmembrane protein, consists of a large extracellular domain, a VWFA domain, four cysteine-rich repeats, and a short cytosolic domain (16, 17). Recent studies demonstrated that, unlike other β integrins, the cytosolic tail of β8 integrin shares no apparent homology and does not directly influence cell adhesion. This suggests β8 integrin signaling to be distinct from other β integrins (16). A connection between β8 and αV integrin as well as the TGFβ signaling cascade has been reported (18). Others exhibited β8 integrin involvement in liver cancer resistance to gefitinib (19), interactions with EphB4 receptor (20), as well as dependency of differentiation and radiosensitivity on β8 integrin in glioma cells (21).

Based on the fact that PDAC is highly therapy refractory, there appears an imbalance in survival and death mechanisms per se and under treatment. Partly, the therapy resistance found in PDAC arises from autophagy (22). Autophagy is a highly conserved catabolic process involving the formation of double-membraned vesicles known as autophagosomes that engulf cellular proteins and organelles for delivery to the lysosome (23). The impact of autophagy seems tissue- and cancer type dependent. In general, autophagy has two opposing functions. One is cytoprotective, eliciting therapeutic resistance; the other one is cytotoxic, inducing autophagic cell death. Recent studies have shown autophagy as a prosurvival and resistance mechanism against chemotherapy treatment in PDAC (24, 25).

In the present study, we explored the function of β8 integrin in PDAC therapy resistance and unraveled parts of its contributing molecular mechanism. In PDAC cell lines and PDAC primary cell cultures grown in 3D lrECM as well as in clinical samples from PDAC patients, we observed that β8 integrin inhibition sensitizes PDAC cells to X-rays and gemcitabine and localizes perinuclearly with the golgi apparatus. Furthermore, our data demonstrate that ionizing radiation induces a microtubule-dependent perinuclear-to-cytoplasmic shift of β8 integrin and changes in the composition of the β8 integrin interactome to transport, catalysis, and binding upon irradiation. Parts of the protein interactome of β8 integrin facilitate a connection to autophagy, which is diminished in the absence of β8 integrin.

The antibodies against β1 integrin (WB 1:1,000, Abcam, ab179471), β8 integrin (IF 1:200, WB 1:1,000; Abcam, ab80673, Rb, recognizes aa 614-663 (this antibody was generally used in this study); Abnova, H00003696-M01, Mo, recognizes aa 392–503 (this antibody was used only where specifically indicated), GM130 (IF 1:100, WB 1:1,000, BD, 610822), MEK1/2 (WB 1:1,000, Cell Signaling Technology, 4694s), γH2AX (WB 1:1,000, Cell Signaling Technology, 9718s), αV integrin (IF 1:250, Novus, NB100-2618), APPL2 (IF 1:100, Sigma-Aldrich, SAB1400605), Caveolin 1 (IF 1:100, BD, 610407), LC3B (IF 1:500, Sigma-Aldrich, SAB4200361), β-actin (WB 1:10,000, Sigma-Aldrich, A5441), HRP-conjugated secondary antibody (GE, Rb NXA931, Mo NXA931), Alexa Fluor 488– or Alexa Fluor 594–conjugated secondary antibodies (1:500, Life Technologies) were purchased and used as indicated.

PDAC cell lines BxPC3, MiaPaCa2, Panc-1, and Patu8902 were purchased from ATCC; Capan-1, patient-derived primary cell lines (PacaDD119, PacaDD137, PacaDD159) and COLO357 cells were a kind gift from Chr. Pilarsky (TU Dresden). Origin and stability of the cells were routinely monitored by short-tandem repeat analysis (microsatellites). Established cell lines were cultured in Dulbecco's modified Eagle medium (DMEM, Gibco, 61965026) with 10% fetal calf serum (FCS, PAN, A15-101) and 1% nonessential amino acids (Gibco, 11140050). Patient-derived primary cell cultures (PacaDD119, PacaDD137, and PacaDD159) and COLO357 cells were cultured in DMEM with 33% K-SFM (Gibco, 17005034) and 13% FCS. All cells were maintained at 37°C with 8.5% CO₂ at pH 7.4. In all experiments, asynchronously growing cells were used. All cells were tested negative for Mycoplasma by using a mycoplasma detection kit from Minerva biolabs (VenorGeM OneStep).

Human β8 integrin generated by PCR-based amplification from cDNA ORF Clone, purchased from Sino Biological (HG10367-M) with specific primers (pAcGFPC1-ITGB8-XhoI-F and pACGFPC1-ITGB8-KpnI-R). Constructs were flanked with XhoI and KpnI restriction sites and inserted into the XhoI and KpnI sites of pAcGFP-C1 (Clontech).

Cells were plated onto uncoated 35-mm dishes with a 0.17-mm glass bottom (MatTek) and allowed to reach 60% to 70% confluency. pAcGFPC1-ITGB8 (β8/AcGFP) and pAcGFPC1 empty vector (control, data not shown) were introduced into the cells using Lipofectamine LTX (Invitrogen, 15338100) according to the manufacture's protocol. Briefly, cells were incubated in 250 μL OptiMEM with a DNA-mix of 250 μL (containing plasmid DNA of 1–2 μg/μL and 10 μL Lipofectamine LTX with 15 μL with PLUS Reagent in 240 μL OptiMEM). Transfection media were removed after 4 hours, and cells were further incubated in fresh medium.

Cells (250,000) were seeded per well of 6-well plate, esiRNAs or nonspecific RLUC esiRNA of 1 μg/mL final concentrations (Eupheria Biotech) were transfected by using oligofectamine (Invitrogen, 12252011) according to the manufacturer's protocol. Knockdown efficiency of β8 integrins was measured by Western blotting.

Irradiation was delivered at room temperature using 2, 4, or 6 Gy single doses of 200-kV X-rays (Yxlon Y.TU 320; Yxlon; dose rate ∼1.3 Gy/minute at 20 mA) filtered with 0.5-mm Cu as published (11). The absorbed dose was measured using a Duplex dosimeter (PTW).

3DHT-esiRNAS was performed in MiaPaCa2 cells in 96-well plate. The esiRNAs of 10 ng/μL final concentration were transfected using oligofectamine. The cells were subcultured 24 hours after transfection. Cell suspensions were mixed with laminin-rich extracellular matrix (IrECM, final conentration at 0.5 mg/mL) and seeded into a new 96-well plate precoated with 1% agarose. After 1-day incubation, 3D IrECM cultured cells were irradiated with 6 Gy X-rays. Grown tumoroids containing 50 cells or more were counted 8 days after irradiation. The percentage of tumoroid formation (N) is given by:

where n is the number of tumoroids in each well treated with target esiRNAs, n₀ is average of control samples without irradiation or with 6 Gy. The average of six data sets was collected from three independent screens.

In addition to the 3D high-throughput screen, as described above, 3D tumoroid formation was determined under identical growth conditions after irradiation (2–6 Gy X-rays) or gemcitabine treatment (for 24 hours; PBS as control) as published previously (26, 27). For gemcitabine, media were removed, cells repeatedly washed, and fresh media were added to allow tumoroid growth over 8 days. Each point on the survival curves represents the mean surviving fraction from at least three independent experiments.

Cells were lysed with modified RIPA buffer consisting of 50 mmol/L Tris-HCl (pH 7.4), 1% Nonidet-P40, 0.25% sodium deoxycholate, 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L NaVO₄, 2 mmol/L NaF (all Sigma-Aldrich), complete protease inhibitor cocktail (Roche) as described previously (27). Total protein amount was measured by BCA assay (Thermo Fisher Scientific). After SDS-PAGE and transfer of proteins onto nitrocellulose membranes (GE Healthcare), probing of specific proteins was accomplished using indicated primary antibodies and horseradish peroxidase–conjugated donkey anti-rabbit and sheep anti-mouse antibodies. Enhanced chemiluminescent reagent (GE Healthcare) was used for detection of proteins on X-ray films (GE Healthcare), and protein expression levels were measured by densitometry.

The Subcellular Protein Fractionation Kit for Cultured Cells (Calbiochem, ProteoExtract Subcellular Proteome Extraction Kit, 539790) was used according to the manufacturer's protocol. In brief, cell lysis and fractionation were conducted after 24-hour plating. Equal loading was ensured by total protein concentration measurement using the BCA assay (Pierce). Fractionation efficacy was confirmed with detection of β1 integrin for the membrane fraction, MEK1/2 for the cytosol fraction and γH2AX for the nuclear fraction.

In brief, cells were grown on coverslip or 35-mm dishes with a 0.17-mm glass bottom (MatTek), fixed with 3.7% PFA at room temperature for 15 minutes, permeabilized with 0.25 % Triton X-100 for 10 minutes, and washed with PBS as published (27). Then, cells were incubated with blocking buffer (1% BSA in PBS) at room temperature for 1 hour followed by incubation with primary and secondary antibodies in blocking buffer in the dark at room temperature for 1 hour. The cells were washed with PBS three times between each step. Then, the cover slides were mounted using ProLong Diamond Antifade Mountant with DAPI. Samples were stored in dark until microscopy.

To investigate the localization of β8 integrin and expression levels of LC3, samples were imaged by a Zeiss Axioscope1 epifluorescence microscope using a 40×/0.75 or 100×/1.25 oil objective. Images for colocalization analysis were acquired by a Zeiss LSM 510 meta microscope using 63×/1.2 water immersion objective and were analyzed by Pearson correlation analysis using Coloc 2 plugin of Fiji. Images for translocation and expression of β8 integrin were recorded by spinning disk confocal microscopy (Olympus IX83 microscope with Yokogawa CSU-X1 Confocal Spinning Disc unit, and iXon Ultra 897 EMCCD Camera) with 60×/1.2 NA water immersion objective. Z stacks with a step of 1 μm were acquired per image. Maximum intensity projection of z stacks was processed using Fiji. All the detected β8 integrin signals from the whole cell were analyzed. The cell size and nucleus size were measured on bright field images. Coordinates of each β8 integrin (Xn, Yn) were determined by the position of maximum intensity. Distance of each β8 integrin to nucleus center (D) is given by

where x0 and y0 are the local positions of the nucleus center. Relative β8 integrin to nucleus center distance (rel. D),

BxPC3 cells grown in a 96-well plate were treated with colchicine, paclitaxel, chloroquine, and gemcitabine with serial dilution. After 3 days, cells were fixed with trichloroacetic acid. After washing with tap water, 50 μL of 0.04% (wt/vol) SRB solution was added to each well and incubated for 1 hour. Then, plates were rinsed four times with 1% (vol/vol) acetic acid to remove unbound dye. Tris base solution (pH 10.5; 100 μL of a 10 mmol/L solution) was added to solubilize the protein-bound dye. Measurement of absorbance was accomplished at 510 nm in a microplate reader (Tecan).

Patient material was examined according to the ethic approval (EK 378092017; dated February 1, 2018) provided by the ethic committee of the Technische Universität Dresden. The tissue used was provided by the tissue bank for tumor and normal tissues of the UCC/NCT Dresden and used in accordance with the regulations of the tissue bank. For immunofluorescence analysis, paraffin sections were deparaffinized in xylene and rehydrated. Antigen retrieval was performed in 10 mmol/L citric acid, pH 6.0 at 98°C for 15 minutes, and sections were stained with hematoxylin and eosin or antibodies against β8 integrin. For tyramide-based immunofluorescence detection, the TSA-kit (Thermo Fisher Scientific) was used according to the manufacturer's instructions. Sections were mounted using ProLong Gold Antifade Mountant with DAPI (Thermo Fisher Scientific) for nuclear counterstaining. Images were acquired using an Axioscope1 plus fluorescence microscope (Zeiss).

Immunoprecipitation was performed as previously published (26). In brief, 1 mg of whole-cell lysates (Cell Lysis Buffer, Cell Signaling Technology) was incubated with specific antibodies at 4°C for 1 hour. Beads were washed with Cell Lysis Buffer (Cell Signaling Technology) and added to lysates. After overnight incubation at 4°C, immunoprecipitates were washed, mixed with sample buffer, and loaded on SDS gels.

To identify β8 integrin interacting proteins, mass spectrometric analysis was carried out at the Max Planck Institute of Molecular Cell Biology and Genetics Mass Spectrometry (MPI-CBG MS) Facility (Dresden, Germany) and performed as published (28). In brief, immunoprecipitates were separated by gel electrophoresis, in-gel digested with trypsin and peptides recovered from the gel matrix analyzed on a LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) coupled on-line to the Ultima3000 LC system (Dionex) via a TriVersa robotic ion source (Advion BioScience). PANTHER (Protein ANalysis THrough Evolutionary Relationships) classification system (http://www.pantherdb.org/) was used to categorize interactomes of β8 integrin according to their functions. Connections between the identified β8 integrin interactome and autophagy were analyzed by the Autophagy Regulatory Network (http://arn.elte.hu/).

All results represent mean ± standard deviation (SD) of at least three independent experiments. Unpaired, two-sided Student t test was performed with Microsoft Excel. A P value of less than 0.05 was considered statistically significant.

The role of FAPs in PDAC therapy resistance has not been comprehensively unraveled. By means of a 3D high-throughput RNA interference (3D-HTP-RNAi) screen (Fig. 1A; Supplementary Table S1), we sought to identify novel potential FAP candidates whose depletion potently diminishes PDAC tumoroid forming ability under basal (Fig. 1B) or irradiation conditions (Fig. 1C). We found basal tumoroid forming ability to be significantly (P < 0.05) compromised upon depletion of, for example, Synemin, JUB, ITGB5, CRK, GAB1, and Caveolin 1 relative to controls (Fig. 1B). Upon 6 Gy X-rays, we surprisingly found none of the FAP depletions to result in a significantly enhanced radiosensitivity of MiaPaCa2 cells (Fig. 1D). We proceeded with the top 10 candidates for further evaluation based on the following criteria (Supplementary Fig. S1): (i) novelty in the context of radiosensitivity, (ii) current knowledge about the candidates and, (iii) higher RNA expression levels in PDAC compared with normal tissue according to the Oncomine database. Our analyses favored β8 integrin as a novel and potent FAP candidate whose depletion enhanced, although not significantly, PDAC cell radiosensitivity in our screen.

Next, we performed an Oncomine data (https://www.oncomine.org) analysis and found a significant, 3.1-fold upregulation of β8 integrin mRNA expression level in PDAC as compared with normal pancreas (Fig. 2A). Furthermore, an analysis using OncoLnc. (http://www.oncolnc.org) for analysis of The Cancer Genome Atlas database of the PDAC patient cohort suggested high β8 integrin expression levels result in shorter survival of PDAC patient as compared with lower expression levels (Fig. 2B; ref. 29). Then, we determined the protein expression level of β8 integrin in six established PDAC cell lines (BxPC3, Capan1, Colo357, Mia PaCa2, Panc1, and Patu8902) and three human PDAC patient primary cell cultures (PacaDD119, PacaDD137, and PacaDD159) grown in IrECM 3D (Fig. 2C and D). The expression of β8 integrin varied in a cell line–dependent manner (Fig. 2C and D). To characterize the subcellular location of β8 integrin, immunofluorescence staining was performed in a panel of human pancreatic cancer cell lines indicating β8 integrin to be located in the perinuclear area in PDAC cells (Fig. 3A, B, and D). In contrast, a clear β8 integrin accumulation at the plasma membrane was observable in glioblastoma cell lines (Fig. 3C and E). Next, we examined cytosol, membrane, and nucleus protein fractions revealing β8 integrin localized in cytosol and membranes, in contrast to our β1 integrin control found only in the membrane fraction (Supplementary Fig. S2A and S2B). Intriguingly, confirmatory data for perinuclear localization in situ are presented by β8 integrin staining in resection specimen of human PDAC (Fig. 3F and G). To further address this abnormal localization, we screened the International Cancer Genome Consortium (ICGC; icgc.org) database for mutations in the ITGB8 gene (https://icgc.org/ZyH). In 52 of 472 donors, 83 mutations are found of which 67 are in introns and 9 in exons. The COSMIC database provided no ITGB8 gene mutations in any of the cell lines used in the presented study (https://cancer.sanger.ac.uk/cosmic/gene/samples?all_data=&coords=AA%3AAA&dr=&end=770&gd=&id=5149&ln=ITGB8&seqlen=770&sn=pancreas&src=gene&start=1#positive). These findings demonstrate a β8 integrin expression in PDAC cells without functionally critical gene mutations and, as compared with other integrin subunits, an unusual subcellular localization that warrants further investigation.

To confirm that β8 integrin plays an essential role in PDAC radiochemoresistance, we effectively silenced β8 integrin in a panel of PDAC cell lines (Fig. 4A and B), detected unaltered tumoroid growth (Fig. 4C and E) but observed a highly significant decrease in tumoroid growth upon irradiation relative to controls (Fig. 4D and E; Supplementary Fig. S3A–S3D). In addition, β8 integrin–depleted PDAC cell lines (BxPC3 and Patu8902) exhibited enhanced chemosensitivity to gemcitabine (Fig. 4F; Supplementary Fig. S3E). Collectively, our data suggest a critical function in PDAC cell survival after X-ray irradiation and gemcitabine treatment.

To define the interacting proteins likely to be causative for the perinuclear localization of β8 integrin, sequential immunoprecipitation-mass spectrometry was used. First, we found 133 proteins bound to β8 integrin in unirradiated cells (Fig. 5A; Supplementary Table S2). Intriguingly, the interactions changed dramatically at 2 hours after 6-Gy X-rays, revealing 637 interacting proteins (Fig. 5A), of which only 32 proteins were present under both untreated and irradiation conditions. This tremendous change in the interactome stimulated us to perform a GO analysis based on the protein molecular functions according to the PANTHER classification system (see the categorization of proteins in Supplementary Table S2). We calculated the increased rate between the protein numbers in unirradiated versus 6-Gy–irradiated cells. The number of interacting proteins increased in different categories as follows: transporter activity by 95%, antioxidant activity by 100%, catalytic activity by approximately 84%, signal transducer activity by approximately 83%, structural molecule activity by approximately 84%, receptor activity by approximately 64%, binding by approximately 68%, and translation regulator activity by approximately 86%, respectively (Fig. 5B). Considering both factors, the absolute number of proteins and their relative increase before and after irradiation, we reckoned that the function of β8 integrin undergoes a critical shift toward transport/binding and catalysis.

Moreover, we checked a small selection of putative proteins related to transport (golgi apparatus reported by GM130; APPL2 and Caveolin 1) and stress (mitochondria) and published interaction partners (αV integrins; ref. 18). In contrast to mitochondria, the previously reported interacting integrin αv, APPL2, and Caveolin 1, only GM130 colocalized to β8 integrin as indicative from the calculated Pearson correlation (0.55 ± 0.04; Fig. 5C and D; Supplementary Fig. S4).

As another possible explanation for the abnormal β8 integrin localization, genetic mutations were evaluated in ITGAV. The ICGC database revealed 54 of 472 donors with mutations (https://icgc.org/Zyr). In total, 78 gene mutations can be found, of which 60 are located in introns and 3 are categorized missense. Similar to ITGB8, no mutations were recognized in any of the used PDAC cell lines using the COSMIC database (https://cancer.sanger.ac.uk/cosmic/gene/samples?all_data=&coords=AA%3AAA&dr=&end=1049&gd=&id=5530&ln=ITGAV&seqlen=1049&sn=pancreas&src=gene&start=1#positive). OncoLnc. (http://www.oncolnc.org) based analysis of the PDAC patient cohort indicated high αV integrin expression to significantly correlate with shorter survival of PDAC patient as compared with lower expression levels (http://www.oncolnc.org/kaplan/?lower=10&upper=90&cancer=PAAD&gene_id=3685&raw=ITGAV&species=mRNA) (Supplementary Fig. S5).

To contextually connect the β8 integrin interactome data with β8 integrin behavior upon irradiation, we next conducted immunofluorescence staining followed by analysis of positive particle intensity and the distance from the cell center to the particles. We correlated this by multiplying with a correction factor, which was determined by measuring the ratio of cell size to nucleus size. Under normal conditions, β8 integrin was located in the perinuclear area. However, upon irradiation, the distance of β8 integrin particles increased from center to nucleus, indicating a movement of β8 integrin–positive particles from the perinuclear area to the cell membrane. This dynamic was accompanied by a significant, approximately 4-fold elevation of β8 integrin expression at already 2 hours as well as 24 hours after genotoxic stress induced by 6-Gy X-ray irradiation (Fig. 6A and B). Furthermore, both average and maximum distances as measures of perinuclear-to-cytoplasmic translocation were significant at 24 hours but not 2 hours after irradiation (Fig. 6A, C, and D).

The association of the β8 integrin interactome proteins led us to speculate that the microtubule system is involved in the radiogenic β8 integrin perinuclear-to-cytoplasmic shift. Therefore, we pretreated cells with the microtubule assembly inhibitor colchicine and found a loss of the significance change in average and maximum distance of β8 integrin (Fig. 6E–G). This finding suggests microtubule dependence of the radiogenic perinuclear-to-cytoplasmic translocation of β8 integrin.

Based on our observations that PDAC cells are radiochemosensitized by β8 integrin targeting and β8 integrin protein interaction partners are associated not only with transporter activity and binding but also with catalytic activity, we hypothesized β8 integrin to function, for example, in autophagy. Hence, candidate proteins involved in autophagy were identified in the list of interacting proteins (Fig. 7A). Although electron transfer flavoprotein alpha subunit (ETFA), thymopoietin (TMPO), and glycoprotein P43 (RBMX) dispersed from β8 integrin at 2 hours after 6-Gy X-rays (Fig. 7A), DnaJ homolog subfamily C member 7 (DNAJC7), aldehyde dehydrogenase 9 family members A1 (ALDH9A1) already present under nonirradiated conditions showed an enhanced binding to β8 integrin at 2 hours after irradiation. Intriguingly, we also found newly connecting proteins such as MAPK8, PRKAR1A, SAFB2, PRKAG1, DNAJB1, SMC1A, MYH4, RHEB, FKBP1A, RPA2, HADHA, HACD3, and DIP2B categorized as proteins regulating autophagy according to the PANTHER classification system. Collectively, our interactome findings suggest β8 integrin to contribute to the regulation of autophagy after exposure to X-rays in PDAC cells.

Next, we investigated whether β8 integrin per se or together with the microtubule system affects LC3B expression as key readout for autophagy. In addition to the microtubule assembly inhibitor colchicine, we pretreated cells with the microtubule stabilizing agent paclitaxel and the autophagy inhibitor chloroquine (Supplementary Fig. S6A–S6C; ref. 30). Unirradiated esiRNA control cells showed nonsignificant elevation in LC3B intensity under colchicine and chloroquine (Fig. 7B and C; Supplementary Fig. S7), whereas paclitaxel led to a slight reduction of LC3B intensity (Fig. 7B). In contrast, β8 integrin depletion elicited a significant decline in LC3B fluorescence intensity under exposure to all three agents, including PBS, relative to esiRNA controls (Fig. 7B). Upon irradiation, the intensity of LC3B was not significantly increased in esiRNA control cells independent of PBS or a specific agent relative to unirradiated esiRNA controls (Fig. 7B). Similar to nonirradiated cells, β8 integrin–depleted and irradiated cells showed a significant reduction in LC3B intensity (Fig. 7B).

To determine whether the conditions tested for LC3B fluorescence intensity affect PDAC cell survival, cells were depleted of β8 integrin without and with colchicine, paclitaxel or the autophagy inhibitor chloroquine. Intriguingly, 3D lrECM PDAC tumoroid formation remained unaffected upon treatment (Fig. 7D), whereas irradiated and treated PDAC cell cultures under removal of β8 integrin showed significant lower tumoroid formation. This effect was not significantly different between the two microtubule-inhibiting agents, but significantly lower for chloroquine (Fig. 7E). The salient findings are that β8 integrin regulates LC3B expression in a microtubule-independent manner and that the interplay between β8 integrin and autophagy measured by LC3B intensity becomes functionally effective for cell survival only upon stress as induced here by X-ray irradiation.

Integrin-mediated cell/ECM interactions fundamentally coregulate cancer cell resistance to therapy. Owing to the massive stroma in PDAC, we conducted a high-throughput knockdown screen in three-dimensional (3D), ECM-based cell cultures to identify novel focal adhesion protein targets. Here, we show that β8 integrin (i) is overexpressed in PDAC relative to normal pancreas, (ii) is critically involved in radiochemoresistance in PDAC cells, (iii) colocalizes with the golgi apparatus perinuclearly in PDAC cells and in resection specimen from PDAC patients, (iv) shows a microtubule-dependent perinuclear-to-cytoplasmic shift and strong changes in its proteomic interactome regarding the cell functions transport, catalysis, and binding upon radiogenic genotoxic injury, and (v) connects with parts of its interactome to autophagy.

β8 integrin does not show an apparent homology to other β integrin subunits in its cytosolic tail, questioning its role in adhesion to ECM and downstream signaling (16, 17). Nonetheless, β8 integrin is overexpressed in various cancer types originating from brain (21), lung (31), prostate (20), and gastrointestinal tract (19, 33) as well as pancreas as shown here. Intriguingly, although other β integrins are located in the cell membrane, β8 integrin is found in the perinuclear region in 3D IrECM grown PDAC cultures as well as in resection specimen from PDAC patients. Our characterization of this perinuclear localization revealed colocalization with the golgi apparatus marker GM130 but neither with the reported integrin αV to form one of the primary receptors for latent-TGFβ1 and latent-TGFβ3 (18) nor mitochondria, APPL2, or Caveolin 1. Genetic mutations in ITGB8 as well as its reported binding partner ITGAV were not detected in regions inevitably entailing dysfunctionality in the protein using the ICGC and COSMIC databases. Hence, the underlying mechanism possibly involving abnormal function of golgi and endoplasmic reticulum proteins requires further exploration.

In addition, β8 integrin is documented to impact on cells from different tumor types. In HepG2 liver cancer cells, β8 integrin contributes to resistance to gefitinib via multidrug transporters and apoptosis regulators (19). It interacts with EphB4 receptor in prostate cancer cells (20) and controls differentiation and radiosensitivity of glioma cells (21). Moreover, β8 integrin has central roles in promoting glioma initiation in vitro and in vivo and its inhibition reduces glioma cell self-renewal ability, stemness, and migration ability (21, 33). These observations are in line with our data demonstrating that β8 integrin targeting sensitizes PDAC cells grown under 3D lrECM-based conditions to irradiation and chemotherapy.

Little is known about the underlying mechanisms how β8 integrin controls cancer cell survival. We, therefore, characterized the β8 integrin proteomic interactome and found pronounced changes in its composition upon radiogenic genotoxic injury. The deduced categorization of interacting proteins from this analysis revealed that β8 integrin engages in transporter activity, antioxidant activity, catalytic activity, signal transducer activity, structural molecule activity, receptor activity, binding, and translation regulator activity in genotoxically injured cells. This wide spectrum of functions stimulated us to focus on transport and catalysis.

Upon genotoxic stress, we found β8 integrin to undergo a perinuclear-to-cytoplasmic translocation that seems cytoprotective, as a siRNA-mediated β8 integrin depletion resulted in elevated cell death and radiosensitization. Mechanistically, the cytoprotective process depended on functional microtubules. Disturbing microtubule assembly by the microtubule inhibitor colchicine strongly reduced the perinuclear-to-cytoplasmic translocation and induced radiosensitization. Which proteins from our β8 integrin interactome profile are responsible for these actions require further investigation.

The interactome also included proteins involved in autophagy. Autophagy is a regulated catabolic pathway to degrade cellular organelles and macromolecules. In malignant tumors, autophagy can either function as a prosurvival response to stress such as starvation, hypoxia, and chemotherapy and radiotherapy that is thought to mediate resistance to anticancer therapies or as an antisurvival response (23, 34). Intriguingly, β8 integrin seems to be a critical autophagy regulator, and autophagy is a cytoprotective mechanism in PDAC. These findings are in line with published work describing high levels of basal autophagy and macropinocytosis arising from intense metabolic rewiring (35). Although the various proteins found in the β8 integrin interactome also require further attention to unravel the underlying mechanism, some are already well-known determinants of the autophagy process such as RHEB (36, 37), DNAJB1 (38), FKBP1A (39), and HADHA (40). Importantly, we observed increasing interactions of these autophagy-associated proteins with β8 integrin upon radiogenic genotoxic stress. Accordingly, the β8 integrin expression in PDAC cells seems connected with autophagy detected by LC3 intensity. LC3 intensity was clearly dependent on β8 integrin expression but neither significantly altered by the two microtubule-inhibiting agents, colchicine and paclitaxel, nor the late-phase autophagy inhibitor chloroquine. Concerning tumoroid formation, it was interesting to exhibit unchanged tumoroid forming capacity in β8 integrin–expressing cells that were pretreated with these three agents. In contrast, radiosensitization occurred in dependence of β8 integrin depletion without strong impact of microtubule or autophagy inhibitors.

Taken together, β8 integrin is overexpressed in PDAC and plays an essential role in PDAC cell radiochemosensitivity. Based on its translocation and regulation of autophagy upon radiogenic genotoxic injury, β8 integrin seems to be critically involved in prosurvival mechanisms. The β8 integrin–driven molecular circuitry and prosurvival signaling networks as well as its druggability warrant further investigations to identify the potential of β8 integrin as a potential cancer target.

D. Aust is a consultant/advisory board member for Roche Pharma and Astra-Zeneca. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Jin, W.-C. Lee, N. Cordes

Development of methodology: S. Jin, W.-C. Lee, N. Cordes

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Jin, W.-C. Lee, D. Aust, C. Pilarsky

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Jin, W.-C. Lee, D. Aust, C. Pilarsky, N. Cordes

Writing, review, and/or revision of the manuscript: S. Jin, W.-C. Lee, D. Aust, N. Cordes

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Jin, W.-C. Lee

Study supervision: N. Cordes

The authors thank Inga Lange for excellent technical assistance and all the members of the Core Facility Cellular Imaging (CFCI) at Faculty of Medicine Carl Gustav Carus, Technische Universität Dresden for technical assistance. The tissue used was provided by the “Tumor- und Normalgewebebank des UCC Dresden” and used in accordance with the regulations of the tissue bank (and the vote of the ethics committee from the Technische Universität Dresden). This project has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 642623.

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.