Cancer-associated fibroblasts (CAF) are major contributors to pancreatic ductal adenocarcinoma (PDAC) progression through protumor signaling and the generation of fibrosis, the latter of which creates a physical barrier to drugs. CAF inhibition is thus an ideal component of any therapeutic approach for PDAC. SLC7A11 is a cystine transporter that has been identified as a potential therapeutic target in PDAC cells. However, no prior study has evaluated the role of SLC7A11 in PDAC tumor stroma and its prognostic significance. Here we show that high expression of SLC7A11 in human PDAC tumor stroma, but not tumor cells, is independently prognostic of poorer overall survival. Orthogonal approaches showed that PDAC-derived CAFs are highly dependent on SLC7A11 for cystine uptake and glutathione synthesis and that SLC7A11 inhibition significantly decreases CAF proliferation, reduces their resistance to oxidative stress, and inhibits their ability to remodel collagen and support PDAC cell growth. Importantly, specific ablation of SLC7A11 from the tumor compartment of transgenic mouse PDAC tumors did not affect tumor growth, suggesting the stroma can substantially influence PDAC tumor response to SLC7A11 inhibition. In a mouse orthotopic PDAC model utilizing human PDAC cells and CAFs, stable knockdown of SLC7A11 was required in both cell types to reduce tumor growth, metastatic spread, and intratumoral fibrosis, demonstrating the importance of targeting SLC7A11 in both compartments. Finally, treatment with a nanoparticle gene-silencing drug against SLC7A11, developed by our laboratory, reduced PDAC tumor growth, incidence of metastases, CAF activation, and fibrosis in orthotopic PDAC tumors. Overall, these findings identify an important role of SLC7A11 in PDAC-derived CAFs in supporting tumor growth.
This study demonstrates that SLC7A11 in PDAC stromal cells is important for the tumor-promoting activity of CAFs and validates a clinically translatable nanomedicine for therapeutic SLC7A11 inhibition in PDAC.
Pancreatic ductal adenocarcinoma (PDAC) is a lethal malignancy, with a 5-year survival rate of <9% (1). A major reason for this poor prognosis is the drug-refractory nature of PDAC caused by inherent chemoresistance and physical barriers to drug delivery. The dense fibrotic microenvironment drives these mechanisms (2–4). Fibrosis distorts the tumor vasculature physically hindering drug access and creates a harsh hypoxic and nutrient-deprived microenvironment, which promotes a more metastatic and chemoresistant mesenchymal phenotype (2–4). The architects of the PDAC microenvironment are cancer-associated fibroblasts (CAF; refs. 2–4). CAFs are activated by signals released from PDAC cells and hypoxia, resulting in a self-perpetuating loop of excessive extracellular matrix (ECM) protein deposition that creates fibrosis (2–4). CAFs also reciprocate prosurvival signaling to PDAC cells, thus promoting PDAC cell survival and epithelial-to-mesenchymal transition (2–4). This makes stromal remodeling and inhibition of CAF activity an important consideration for PDAC therapies.
CAFs and PDAC cells share an oxygen/nutrient poor microenvironment. PDAC cells have altered their metabolism to survive and proliferate in this stressful microenvironment (5). These alterations lead to metabolic addictions that can be therapeutically exploited. A potential target is the Xc− amino acid antiporter, which imports cystine into the cell, in exchange for glutamate (6–8). Xc− is a heterodimer of solute carrier 3A2 (SLC3A2; membrane anchor) and solute carrier 7A11 (SLC7A11, also known as xCT; amino acid transporter; refs. 6–8). This transporter sits at the crux of multiple metabolic activities necessary for cancer cell survival, including protein synthesis and redox regulation. First, cystine transported by SLC7A11 is reduced to cysteine, which is an irreplaceable component of proteins, which is required for disulphide bond formation. Second, cysteine is also the rate-limiting amino acid in the synthesis of the potent antioxidant glutathione (GSH; ref. 9). GSH is important in PDAC cell survival as KRAS-driven metabolic changes, protumor signaling, and microenvironment-driven hypoxia increase otherwise lethal intracellular oxidative stress (10).
SLC7A11 has been identified as a prognostic factor and potential therapeutic target in a number of cancers (11–16). In PDAC, Lo and colleagues (17) demonstrated that SLC7A11 was upregulated in PDAC cells under oxidative stress and cystine deprivation in vitro, and SLC7A11 inhibition (sulfasalazine, SSZ) significantly reduced subcutaneous PDAC tumor growth (18). Since then, studies have demonstrated the autophagic regulation (19) and therapeutic potential of inhibiting or genetically ablating SLC7A11 in PDAC cells (20–24). While promising, a key limitation of these studies was that they ignored the role of SLC7A11 in CAFs or the impact of CAFs on PDAC cell sensitivity to SLC7A11 inhibition. This is a critical gap in our knowledge for therapeutic inhibition of SLC7A11 in PDAC, given the prominent role CAFs play in PDAC cell survival and drug sensitivity and their potential to help overcome nutrient deficiencies (25, 26).
We hypothesized that SLC7A11 inhibition in CAFs had the potential to directly inhibit a key cellular target in PDAC and to break protumor interactions between CAFs and PDAC cells. We demonstrate that high stromal expression of SLC7A11 predicts poorer patient survival, and that SLC7A11 inhibition in CAFs reduces their proliferation, antioxidant capacity, and growth of three-dimensional (3D) spheroid cocultures with PDAC cells. Importantly, we show that stable knockdown of SLC7A11 in CAFs reduces tumor incidence, growth, and fibrosis, and demonstrate the therapeutic potential of inhibiting SLC7A11 expression in PDAC tumors using a gene silencing nanomedicine.
Materials and Methods
Cell isolation and culture
Commercial human PDAC cells (MiaPaCa-2, Panc-1, AsPC1, HPAFII; ATCC) were cultured as described previously (27–29). The TKCC5 human pancreatic cancer cells were established from patient-derived xenografts (PDX) and cultured as described previously (30). PDAC cell purity was confirmed by short tandem repeat profiling (CellBank Australia). Human pancreatic ductal epithelial (HPDE) cells (a gift from Ming Tsao, Ontario Cancer Institute, Toronto, Ontario, Canada) were cultured as described previously (31). All studies involving the use of human-derived fibroblasts and CAFs were approved by the UNSW Sydney human ethics committee (approvals: HC14039, HC180973, HREC13/023) and the German Technical University of Munich human ethics committee (approval: 5510/12). All patients provided written informed consent. Quiescent human pancreatic fibroblasts, activated by culture on plastic, were isolated from patients with benign pancreatic conditions as described previously (32). Human CAFs were isolated from PDAC tumor tissue by explant culture and used within 12 passages as described previously (33, 34). All CAFs were cultured in Iscove's modified Dulbecco medium, 10% FBS, 4 mmol/L l-glutamine. The purity of CAFs was assessed by positive immunostaining for glial fibrillary acidic protein (GFAP) and alpha-smooth muscle actin (αSMA) and negative immunostaining for cytokeratin, as described previously (32). Cells were maintained in a humidified incubator (37°C, 5% CO2) and tested negative for Mycoplasma [MycoAlert Mycoplasma Detection kit (Lonza)] during fortnightly testing throughout study.
Immortalization of human PDAC CAFs and establishment of human PDAC and CAF cells expressing SLC7A11 shRNA
Human patient-derived CAFs (passage 9) were immortalized by lentiviral delivery of a human telomerase construct (GenTarget, catalog no. LVP1131-RP). Cells were maintained in puromycin and red fluorescent protein–positive cells sorted on a BD FACS Aria II. We confirmed hTERT-CAFs had comparable proliferation to parent cells. MiaPaCa-2 cells and hTERT-immortalized CAFs were then transduced with lentiviral scramble (control)-shRNA, SLC7A11-shRNA sequence 1 constructs (Origene, catalog no. TL309282). Transduced cells were maintained in puromycin and GFP-positive cells sorted on a BD FACS Melody. Western blot analysis confirmed SLC7A11 knockdown. All CAF shRNA cell lines were revalidated as per patient-derived CAFs and used within 50 passages of immortalization.
Isolation and culture of KPC transgenic mouse PDAC and CAF cells
Mouse KPC PDAC cells were isolated from tumors from the KPC genetically engineered mouse model (KRAS/P53-mutated) and cultured as described previously (35). KPC CAFs were isolated as described previously (33, 34) and validated by immunocytochemistry for GFAP and αSMA. KPC CAFs were cultured as per human CAFs (33, 34).
Quantitative real-time PCR
Quantitative real-time PCR (qPCR) was performed using Power SYBR green kit (Thermo Fisher Scientific) run on an Applied Biosystems ViiA7 platform. Primers were obtained from Qiagen: SLC7A11 (catalog no. QT00002674); 18S ribosomal RNA (housekeeping gene, catalog no. QT00199367); SLC3A2 (catalog no. QT00085897); β2-microglobulin (housekeeping gene, catalog no. QT00199367). Concentrations were calculated on the basis of ΔCt standard curves and standardized to β2-microglobulin or 18S ribosomal RNA.
Western blot analysis for SLC7A11
Protein lysates were prepared, quantified, electrophoresed on a 10% SDS-PAGE gel and transferred to nitrocellulose membranes as described previously (27), except samples were not boiled before electrophoresis. Membranes were probed with primary and secondary antibodies detailed in Supplementary Table S1. Blots were visualized and bands quantified as described previously (27).
Validation of SLC7A11 antibodies
Serial sections of formalin-fixed, paraffin-embedded human PDAC tumor tissue were obtained through the Australian Pancreatic Cancer Genome Initiative (APGI). All studies involving the use of human specimens were approved by the UNSW Sydney human ethics committee (approvals: HC14039, HC180973, HREC13/023). All patients provided written informed consent through the APGI. Antigen retrieval was performed as described previously (27–29). Tissue sections were probed with antibodies detailed in Supplementary Table S1, then Vectastain ABC kit (Vector Laboratories). 3,3′ diaminobenzidine (DAB) was used as the substrate and hematoxylin as counterstain. The Cell Signaling antibody used for this study was validated as detailed in ref. 36. We showed similar staining patterns in PDAC tissue sections using three independent antibodies (Supplementary Fig. S1A). Positive control brain tissue (Supplementary Fig. S1B) showed abundant SLC7A11 staining and negative control skin tissue (Supplementary Fig. S1B) showed no SLC7A11 staining, consistent with expression in the human protein atlas. Specificity was also confirmed by its ability to detect specific SLC7A11 gene silencing (2 siRNAs and shRNA) in Western blot analysis (Supplementary Fig. S1C–S1F).
Immunofluorescence for SLC7A11 and αSMA colocalization
Formalin-fixed, paraffin-embedded human PDAC tumor tissue was obtained through the APGI. Antigen retrieval was performed as described previously (27–29). Tissue sections were then stained with primary and secondary antibodies detailed in Supplementary Table S1. Tissues were mounted using Prolong Gold Antifade (Thermo Fisher Scientific, catalog no. P36931) and imaged (Zeiss 900 confocal microscope).
Correlation of SLC7A11 expression in human PDAC specimens with overall survival
Formalin-fixed, paraffin-embedded human PDAC tissue microarrays (TMA) were obtained through the APGI (International Cancer Genome Consortium Cohort; patient demographics in Supplementary Table S2). All patients provided written informed consent. TMA rehydration and blocking for IHC was performed as described previously (27–29). TMAs were probed with primary and secondary antibodies as in Supplementary Table S1 followed by Vectastain ABC Kit (Vector Laboratories), DAB, and hematoxylin. Staining intensity in tumor and stromal compartments was scored using independent four-point scales (0–3) by two independent scorers, based on intensity in ≥75% of each compartment (normal acinar/ductal cells excluded). A consensus score was obtained for each core. For each set of three cores per patient, the highest tumor and stroma scores were used. Scores of 0–1 = SLC7A11low; Scores of 2–3 = SLC7A11high. Scores were then correlated with overall survival using a Kaplan–Meier survival curve (see statistical analyses). Patients deceased due to other causes/still alive were censored. Non-PDAC tumors and 2 patients that lacked tumor in all 3 cores were excluded.
Normalized SLC7A11 expression values (expression array) were from the PACA-AU cohort through the ICGC data portal (available from https://dcc.icgc.org/projects/PACA-AU). Non-PDAC patients were excluded (total PDAC patients = 242). Values were broken into tertiles (low ≤1.98, medium = 1.98–2.52, high ≥2.53) and correlated with overall survival using a Kaplan–Meier survival curve (see statistical analyses). For comparison of SLC7A11 expression in mouse inflammatory-like CAFs (iCAF), myofibroblast CAFs (myCAF), and quiescent pancreatic fibroblasts, we used normalized expression data from Ohlund and colleagues (37).
CAFs and PDAC cells were transfected with 100 nmol/L smartpool On-target plus human SLC7A11 siRNA pool (Dharmacon, catalog no. L-007612–01), human SLC7A11 single sequence siRNA [Dharmacon, catalog no. LU-007612–01 (part of J-007612–01)], mouse SLC7A11-siRNA (Dharmacon, catalog no. L-047420–01) or nonsilencing siRNA (ns-siRNA; Dharmacon, catalog no. D-001810–01–50)] as described previously (27–29).
Preparation of drugs
Sulfasalazine (SSZ, Sigma Aldrich, catalog no. S0883–10G) was dissolved in 0.1 mol/L NaOH, PBS, pH 7.5, and filtered (0.22 μm). Erastin (Cayman Chemical, catalog no. 17754) and Ferrostatin-1 (Sigma-Aldrich, catalog no. SML0583) were dissolved in DMSO. N-acetyl-l-cysteine (NAC; Sigma-Aldrich, catalog no. A9165) was dissolved in MilliQ H2O and filtered (0.22 μm).
Cell viability assays
Cell proliferation or viability was measured by Trypan blue exclusion and cell counting kit 8 (CCK8) assays as described previously (27–29) at: (i) 72 hours posttransfection with siRNA; (ii) 48 hours posttreatment with 0.2–0.4 mmol/L SSZ; (iii) 24–48 hours posttreatment with 40 μmol/L erastin; (iv) 72 hours postseeding of shRNA-stable lines.
(i) 0–60 μmol/L tert-butyl hydroperoxide (tBHP; oxidative stress inducer; Sigma Aldrich, catalog no. 458139) in the final 24 hours of assays; (ii) 0–66 μmol/L 2-mercaptoethanol (2-ME) in parallel with SSZ; (iii) 2 μmol/L Ferrostatin in the final 24 hours of assays; (iv) low nutrient conditions for the final 24 hours of assays: serum-free Plasmax, no serum, 2 mmol/L glutamine (refer to https://ximbio.com/reagent/156371/plasmaxsuptmsup-cell-culture- medium for comparison of Plasmax and IMDM).
Cell-cycle, cell death, and senescence assays
Adhered CAFs (i) 72 hours posttransfection with siRNA, (ii) 48 hours postseeding of stable shRNA CAFs, (iii) 9 hours posttreatment with 40 μmol/L erastin or 0.2% DMSO (control) were assayed for ferroptosis using Glutathione Peroxidase Assay Kit (Abcam, catalog no. ab102530).
Cell-cycle assay (72 hours posttransfection)
Cell-cycle analysis was performed by propidium iodide (PI) staining and flow cytometry on a BD LSR Fortessa, as described previously (29).
Cell senescence assay (72 hours posttransfection or SSZ treatment)
Senescence was assessed using a senescence β-galactosidase Cell Staining Kit (Cell Signaling Technologies, catalog no. 9860).
Measurement of autophagy (72 hours posttransfection)
Western blot analysis was performed for Light Chain 3 isoform B (LC3B), to check for conversion from cytosolic LC3BI to autophagosome-bound LC3BII. Western blot analysis was performed as per SLC7A11 procedure, except samples were boiled. Western blots were probed with antibodies described in Supplementary Table S1 then visualized and quantified as per SLC7A11 Western blot analysis.
Apoptosis assay (72 hours posttransfection, 24 hours post-tBHP)
Adherent and floating cells were collected and total apoptosis detected using an Annexin V-PE/7AAD kit (BD Biosciences, catalog no. 559763) according to the manufacturer's instructions and analyzed on a BD LSR Fortessa.
Measurement of cystine uptake
Nontransfected CAFs (SSZ experiments) or CAFs, 48 hours posttransfection with siRNA, were seeded at 15,000 cells/well into Wallac isoplate 96-well plates. The following day, cystine uptake was assayed immediately (knockdown experiments) or 2–3 hours posttreatment with SSZ (SSZ maintained during assay). Cystine uptake was measured as described previously (38).
Assessing glutathione synthesis, oxidative stress, and glutamate efflux
Measurement of intracellular glutathione
Intracellular glutathione was measured in whole-cell lysates [prepared: (i) 72 hours posttransfection with siRNA; (ii) 72 hours postseeding of shRNA-stable lines; (iii) after 48 hours 0.25 mmol/L SSZ; (iv) 16 hours post 0.25 mmol/L SSZ with or without 1 mmol/L NAC] using ApoGSH Glutathione Colorimetric Assay (BioVision, catalog no. K261–100) or Glutathione Assay Kit (Cayman Chemical, 703002) according to the manufacturer's instructions.
Detection of intracellular ROS using CellROX
Seventy-four hours posttransfection or after addition of 0.4 mmol/L SSZ, CAFs were incubated with 325 μmol/L tBHP for 1 hour. CAFs were then stained with CellROX green reagent (Life Technologies, C10444) according to manufacturer's instructions. Fluorescence was analyzed on a BD LSRFortessa.
Glutamate efflux assay
Seventy-four hours posttransfection, CAFs were washed and cultured in M2 buffer (Sigma Aldrich) for 4 hours. Glutamate was measured in cell supernatant using Amplex Red Glutamic Acid/Glutamate Oxidase Assay Kit (Thermo Fisher Scientific; catalog no. A12221). Fluorescence measured using an EnSpire Multimode Plate Reader (PerkinElmer) at 571/586nm (excitation/emission).
Metabolomics for intracellular cystine and glutamate
To assess glutamate:cystine exchange, immortalized CAFs expressing shRNA were cultured in glutamine-free IMDM (Sigma), 10% dialyzed FBS, 4 mmol/L 13C5-glutamine (Sigma) for 24 hours. Cells were washed twice and medium replaced with complete culture medium, 10% dialyzed FBS. Cell extracts were harvested by scraping into cold lysis medium (methanol: acetonitrile: MilliQ water at 50:30:20 v/v) and debris pelleted at ≥14,000 × g for 15 minutes at 4°C. The supernatant was then used for LC/MS analysis as described previously (39). Metabolites were predetermined with in-house standards (20–40 μmol/L, HPLC-grade, Sigma). Labeling patterns were analyzed by Xcalibur (v4.1, Thermo Fisher Scientific; RRID:SCR_014593) and mass of isotopomers corrected against naturally occurring stable carbons.
3D coculture models
Spheroid outgrowth assay
MiaPaCa-2 PDAC cells and patient-derived CAFs were transfected with siRNA. After 24 hours, 2,500 of each was coseeded in suspension, spheroids formed over 24 hours under normal culture conditions, then transferred to 0.33% soft-agarose medium (20% FBS, 4% l-glutamine, 0.33% agarose in low-glucose DMEM:F12 medium) and overlayed on 50 μL preset 0.5% soft-agarose medium in a 96-well round-bottom plate
Spheroid growth assay
PDAC cells and immortalized CAFs expressing shRNA were seeded into low adherence round-bottom 96-well plates (2,500 PDAC cells + 7,500 CAFs). After 24-hour culture to allow spheroid formation, supernatant on spheroids was replaced with 1:1 complete culture medium:Corning Matrigel (In Vitro Technologies, catalog no. 354230) then returned to culture. Daily brightfield photos were taken on an inverted light microscope and spheroid growth/outgrowth quantified using ImageJ (RRID:SCR_003070).
Matrix contractility assay
Organotypic assays were adapted from published protocols (40). A total of 0.3–1 × 105 CAFs (n = 4, 24 hours posttransfection) were embedded in 1.25 mL of rat-tail collagen I. Once polymerized, fibroblast-collagen matrices were allowed to contract in IMDM, 10% FBS, 4 mmol/L l-glutamine, 1% penicillin/streptomycin for six days with images taken at days 2–6. Collagen plug area was quantified from brightfield images. At day 6, collagen matrices were fixed, paraffin-embedded, and sectioned for Second Harmonic Generation (SHG) analysis as described previously (40, 41). SHG signal was acquired (line average:16) and intensity/gray-level cooccurrence matrix (GLCM) analysis performed in Matlab (Mathworks; RRID:SCR_001622) as described previously (40, 41). Paraffin-embedded samples were cut into 4-μm sections and stained with 0.1% picrosirius red (Polysciences) according to manufacturer's instructions. Polarized light imaging and intensity measurement was performed as described previously (40, 41).
Transgenic pancreatic cancer mouse models and genetic ablation of SLC7A11
Transgenic mouse studies were approved by the local ethical review committee at University of Glasgow according to UK Home Office regulations (licence: 70/8646). KC (Kras-mutated) and KPC (Kras- and p53-mutated) mouse models [alleles used: Pdx-1-promoter, lox-stop-lox-KrasG12D/+ allele, and lox-stop-lox-Trp53R172H/+ ± Slc7a11fl/fl (IMPC (MGI:1347355); ref. 42] were genotyped by Transnetyx. In these models, expression of mutated Kras, P53, and conditional deletion of SLC7A11 are driven by Pdx1-Cre, meaning that SLC7A11 is lost only from the epithelial cells of the pancreas.
Pancreatic intraepithelial neoplasia scoring
Slc7a11+/+ (KC) and Slc7a11fl/fl (KC with conditional Slc7a11 knockout) mice were sampled at 70 days old and pancreatic intraepithelial neoplasia (PanIN) scored from whole hematoxylin and eosin (H&E) sections (normalized to mm2).
KPC and KPC Slc7a11fl/fl mice were monitored thrice weekly and sampled when exhibiting clinical signs of PDAC (abdominal swelling, jaundice, hunching, piloerection and weight loss). RNA in situ hybridization (ISH) was performed on formalin-fixed KPC tumor sections. RNA ISH (RNAscope) was performed according to the manufacturer's protocol (ACD RNAscope 2.0 High Definition–Brown) for Slc7a11 (Basecope probe targets floxed exon 3). Western blot analysis was performed as above, using antibodies in Supplementary Table S1.
Star nanoparticle synthesis
Star nanoparticles (Star 3) were synthesized as described previously (43).
Subcutaneous and orthotopic pancreatic cancer mouse models
All mouse studies were approved by the UNSW Sydney Animal Care and Ethics Committee (approval: ACEC 16/25B, ACEC 19/3A).
TKCC5 PDAC cells were coimplanted with immortalized CAFs expressing shRNA (106 each) into the flank of 8-week-old female BalbC nude mice in 1:1 mix of PBS and Matrigel. Tumor growth was monitored by caliper measurement.
Orthotopic model (stable shRNA cell lines)
Luciferase-expressing MiaPaCa-2 cells (27) and immortalized CAFs expressing shRNA (as per spheroids) were coimplanted (106 each) into the pancreas tail of 8-week-old female BalbC nude mice, as described previously (27, 34). Tumor volume was measured weekly by ultrasound on a VisualSonics Vevo 3100 from week 4 postimplant. Mice that had no measurable tumor at week 4 were excluded (1 exclusion).
Orthotopic model (therapeutic knockdown)
Orthotopic tumors were established as per stable shRNA models.
Star 3-siRNA gene-silencing efficiency study
Six weeks postsurgery, mice were treated with 3 mg/kg control-siRNA (antisense: 5′-GAACUUCAGGGUCAGCUUGCCG) or SLC7A11-siRNA (antisense: 5′-AGACCCAAUAAGUUUGCCG) complexed to Star 3, intravenously once daily for three days.
Star 3-siRNA therapeutic study
Four weeks postsurgery, mice were randomized on the basis luminescence as described previously (27), then treated with 3 mg/kg Star 3+siRNA, intravenously daily for the first three days, followed by twice weekly for 4 weeks. Mice were cotreated intravenously with 10 mg/kg Abraxane (Specialised Therapeutics Australia) or 90 mg/kg human albumin (control), once weekly for 4 weeks. At endpoints, mice were humanely euthanized and organs/tumors harvested. Tumor volume was calculated by caliper measurement with operator blinded to treatment. Tumor fragments were 4% paraformaldehyde-fixed and paraffin embedded, or frozen in Tissue-Tek Optimal Cutting Temperature Compound (OCT, VWR International). Metastases were detected by ex vivo luminescence (>600 counts) and confirmed by H&E as described previously (27).
Measurement of collagen content and organization in pancreatic tumor sections (orthotopic and transgenic PDAC models)
Picrosirius red staining and measurement of total collagen content and fibril density (polarized light analysis)
Five-micron–thick sections of paraffin-embedded tumors were stained with 0.1% picrosirius red and counterstained with methyl green or hematoxylin. Collagen content was calculated from representative regions (percent of area picrosirius red positive/region excluding necrosis, average tumor coverage = 13%) using ImageJ (RRID:SCR_003070) or Qupath, then averaged for each tumor. Polarized light analysis was performed on the same regions, as described previously (40, 41).
SHG analysis of collagen organization and directionality
SHG signal was acquired in OCT-embedded tumor sections (15 μm) and maximum intensity projections used for GLCM and orientation analyses [three representative regions (775 μm × 775 μm)/tumor] as described previously (40, 44, 45).
IHC for SLC7A11, αSMA, and CD31 in mouse PDAC tumors
IHC was performed in paraformaldehyde-fixed and paraffin-embedded tumor sections, as described previously (27–29) using antibodies in Supplementary Table S1. SLC7A11 staining intensity, αSMA area coverage, and CD31-positive blood vessels (open vs. closed manual count) were analyzed in representative images (excluding necrosis) using color deconvolution in ImageJ (RRID:SCR_003070), then averaged per tumor. SLC7A11 staining intensity was calculated from three images/tumor (average tumor coverage = 10%). Optical density (staining intensity) = log[maximum intensity of pixels/average intensity of pixels]. αSMA representative regions average tumor coverage was 21%. CD31-positive representative regions' average tumor coverage was 55%. Observers were blinded to treatments.
3D human PDAC tumor explant model
Human PDAC tumor specimens were obtained from patients undergoing pancreaticoduodenectomy at Prince of Wales Hospital or Prince of Wales Private Hospital (Randwick, New South Wales, Australia). All patients provided written informed consent through the Health Science Alliance Biobank, all work was approved by UNSW human ethics (HC180973), and all experiments were performed in accordance with the relevant regulations. Patient-derived explants (1–8 mm3) were established and cultured for 12 days with daily medium change, as described previously (46). Star 3 nanoparticles (60 μg) complexed to siRNA were added to the reservoir as described previously (46) on days 0, 3, 6, 9. Alternatively, 400 μmol/L SSZ or vehicle was added to the reservoir on days 0, 3, 6, and 9. Patient-derived explants were treated with 10 μmol/L BrdU substrate (BD Biosciences, catalog no. 550891) for 24 hours prior to fixation (day 12), as described previously (46).
Statistical comparisons were performed using two-tailed Student t test (two groups) or ANOVA (≥3 groups; post hoc tests: Dunn, Sidak multiple comparison) using GraphPad Prism (RRID:SCR_002798). Comparisons of univariate time to event (survival) were performed using the log-rank test and HRs calculated from the Cox proportional hazards (PH) model. Multivariate associations between variables and time to event were contained from PH regression and survival curves calculated using the method of Kaplan–Meier (KM). Where tumor and stroma scores correlated with outcome, baseline variables associated with predicting scores were examined by multivariate logistic regression. Tumor grade was excluded from multivariate analyses as it did not correlate with overall survival in our cohort due to the low percentage of grade 3–4 tumors (13% of cohort). Survival analyses were performed using Analysis of Censored and Correlated Data (ACCoRD; RRID:SCR_009015) V6.4 Boffin. A P value ≤0.05 was considered statistically significant.
All data generated or analyzed during this study are included in this published article (and its Supplementary Information) and can be made available upon reasonable request. Expression array data for the PACA-AU cohort are publicly available through the ICGC data portal (https://dcc.icgc.org/projects/PACA-AU). More specific PDAC patient cohort data can be obtained through the Australian Pancreatic Cancer Genome Initiative. We are willing to share CAF cells, which would require appropriate human ethics approvals and an MTA with our laboratory.
High stromal SLC7A11 expression in human PDAC specimens predicted poor overall survival
Results showed that SLC7A11 and its partner SLC3A2 mRNA levels are upregulated in PDAC patient-derived CAFs compared with patient-derived noncancerous pancreatic fibroblasts (Fig. 1A). Similar results were obtained when we analyzed Ohlund and colleagues data (37), which showed SLC7A11 mRNA expression was increased in iCAF (2.7-fold) and myCAF (1.6-fold) subpopulations versus quiescent pancreatic stellate cells (Supplementary Table S3). We confirmed that all human PDAC CAFs expressed SLC7A11 protein (Fig. 1B). SLC7A11 protein levels in human CAFs (5/6 CAFs tested) were comparable with PDAC cells with the highest SLC7A11 expression (HPAFII, ASPC1; Fig. 1B). SLC7A11 protein levels were higher in PDAC cells derived from metastatic sites (HPAFII/AsPC1) relative to those from primary tumor sites (MiaPaCa2/Panc-1; Fig. 1B). Coimmunofluorescence staining for SLC7A11 and αSMA (CAF marker) in human PDAC tumors demonstrated abundant SLC7A11 protein in αSMA-positive CAFs (Fig. 1C). We next scored SLC7A11 protein expression, as determined by IHC, in tumor and stromal compartments of human PDAC tissue microarrays (Fig. 1D; independent scoring scales used for tumor and stroma) and correlated with overall patient survival (Fig. 1E–G). Fifty-eight percent of patients were Tumorhigh (Fig. 1E) and 47% were Stromahigh (Fig. 1F). SLC7A11 expression in the PDAC tumor compartment alone did not predict patient survival (Fig. 1E). Similar results were obtained when we analyzed the ICGC publicly available mRNA data (Supplementary Fig. S1G; ref. 47). Importantly, in a multivariate logistic regression, high SLC7A11 in the stroma was independently prognostic of poorer overall survival (Fig. 1F, P = 0.041, HR = 1.45; Supplementary Table S4), when adjusted for vascular invasion. In addition, we identified a subgroup of patients (TumorlowStromahigh) that had significantly poorer overall survival compared with all other score combinations (Fig. 1G and H).
Inhibition of SLC7A11 in CAFs reduced cell proliferation and metabolically rewired CAFs
To assess SLC7A11 function in CAFs, we used siRNA and two pharmacologic inhibitors (SSZ, erastin). SLC7A11-siRNA reduced mRNA (Supplementary Fig. S1H) and protein levels (Supplementary Fig. S1C and S1D) compared with control siRNA. Inhibition of SLC7A11 by siRNA, SSZ, or erastin significantly decreased CAF proliferation and viability (Fig. 2A–C) including under low nutrient conditions (30.6% ± 13.8% reduction in live cells relative to ns-siRNA; P ≤ 0.05, n = 3). Importantly, SLC7A11 knockdown in CAFs inhibited proliferation in both SLC7A11low and SLC7A11high CAFs (Supplementary Fig. S2A), indicating SLC7A11 is functionally essential in CAFs regardless of expression level. 2-mercaptoethanol (2-ME), which facilitates bypass of xCT for cysteine (45), rescued CAF growth in the presence of SSZ (Fig. 2B). This indicated that SSZ-induced growth arrest of CAFs was due to cystine starvation. Indeed, treatment of CAFs with SLC7A11-siRNA or SSZ significantly reduced cystine uptake (Fig. 2D and E) and intracellular glutathione (Fig. 2F and G), relative to controls. The reduction in glutathione was rescued by addition of N-acetyl-cysteine (NAC; Fig. 2G), which provides an SLC7A11-independent source of cysteine. Stable knockdown of SLC7A11 (Supplementary Fig. S1E and S1F) in hTERT-immortalized CAFs using shRNA similarly decreased intracellular cystine and increased glutamate retention, consistent with SLC7A11 inhibition (Supplementary Fig. S2B and S2C).
We validated previous findings that both SSZ and erastin significantly decreased MiaPaCa-2 PDAC cell proliferation (Supplementary Fig. S2D and S2E) and SSZ inhibition can be rescued by 2-ME (18). In contrast to results in PDAC cells and CAFs, SLC7A11 knockdown (Supplementary Fig. S2F) had minimal effect on the viability of nontumor HPDE cells (Supplementary Fig. S2G).
We next assessed intracellular reactive oxygen species (ROS; oxidative stress) and found that SLC7A11 knockdown using siRNA in CAFs had no effect on intracellular ROS in the absence of stress, but significantly increased intracellular ROS in the presence of stress (tBHP; Fig. 2H; tBHP increased mitochondrial ROS, Supplementary Fig. S3A), suggesting decreased antioxidant capacity. In contrast, SSZ treatment alone increased intracellular ROS in CAFs, to levels where tBHP treatment had no significant additive effect on intracellular oxidative stress (Fig. 2I). Note that the lack of an increase in intracellular ROS with tBHP alone is due to the short incubation time (1 hour) for this assay. SLC7A11 knockdown in CAFs had no effect on glutamate secretion (Supplementary Fig. S3B).
Inhibition of SLC7A11 increased sensitivity to oxidant stress and ferroptosis
We confirmed SLC7A11 knockdown was maintained in the presence of oxidative stress (Supplementary Fig. S3C) and observed that oxidative stress increased SLC7A11 protein expression in cells treated with control siRNA (Supplementary Fig. S3C). Knockdown of SLC7A11 using siRNAs in CAFs sensitized them to external oxidant stress (tert-butyl hydroperoxide, tBHP) by decreasing viability (Fig. 2J) and increasing apoptosis (Fig. 2K; Supplementary Fig. S3D). Previous studies have shown that SLC7A11 inhibition in PDAC cells induces oxidative stress–induced cell death (ferroptosis; refs. 21, 23, 48). We observed that inhibition of SLC7A11 with erastin reduced glutathione peroxidase activity indicative of ferroptosis in CAFs (Fig. 2L). Importantly, the ferroptosis inhibitor ferrostatin rescued CAFs from the antiproliferative effects of erastin confirming ferroptosis (Fig. 2M). Likewise, stable SLC7A11 knockdown in CAFs decreased cell viability in the presence of oxidant stress (Fig. 2N) and increased ferroptosis (i.e., decreased glutathione peroxidase activity, Fig. 2O).
SLC7A11 inhibition increased senescence of CAFs in the absence of stress
Given SLC7A11-siRNA alone had no effect on apoptosis, we explored other antiproliferative mechanisms. We showed that SLC7A11-siRNA had no effect on autophagy (Supplementary Fig. S3E), but increased CAFs in S-phase of cell cycle (Supplementary Fig. S3F, suggesting hindered S-phase progression). SLC7A11 knockdown or SSZ treatment in CAFs also significantly increased senescence (Fig. 2P and Q). Results showed that SLC7A11 knockdown in CAFs induced senescence and in the presence of additional oxidative stress compromised CAF survival.
SLC7A11 inhibition decreased CAF and PDAC coculture spheroid growth in vitro
To determine whether SLC7A11 inhibition in CAFs affected their ability to support PDAC cell growth, we performed 3D coculture assays [spheroid outgrowth (Fig. 3A) and spheroid growth assays (Fig. 3B)]. Transient knockdown of SLC7A11 in either CAFs, PDAC cells, or both significantly reduced spheroid outgrowth (Fig. 3C). Importantly, knockdown of SLC7A11 in CAFs alone or in both cell types was more effective at inhibiting spheroid outgrowth than SLC7A11 knockdown in PDAC cells alone (Fig. 3C). Using a stable knockdown approach in MiaPaCa2 PDAC cells and immortalized CAFs, in a 3D Matrigel-embedded spheroid assay, we observed similar results (Fig. 3D). Except SLC7A11-shRNA in MiaPaCa-2 PDAC cells alone had no effect on spheroid growth. In contrast, SLC7A11-shRNA in CAFs alone or in both PDAC cells and CAFs reduced spheroid size (relative to start) at endpoint by 40% (Fig. 3D). We repeated the assay using the CAF shRNA lines above combined with additional PDAC cells (Panc-1, AsPC1, TKCC5) and observed a comparable reduction in spheroid growth relative to CAF(control-shRNA) spheroids (Fig. 3E–G).
SLC7A11 inhibition decreased collagen remodeling in vitro
SLC7A11 knockdown in CAFs significantly reduced contraction of collagen plugs over 6 days (Fig. 3H; higher contraction = greater remodeling). Brightfield analysis of picrosirius red–stained collagen plugs demonstrated that collagen plugs remodeled by CAFs transfected with SLC7A11-siRNA had decreased collagen relative to controls (Fig. 3I). Polarized light analysis (measures density of collagen fibrils) showed that plugs remodeled by CAFs transfected with SLC7A11-siRNA decreased overall birefringent fibrils (Fig. 3J) and high/medium density fibrils, and increased low density fibrils, relative to controls (Fig. 3K). This was confirmed by SHG analysis of fibrillar collagen (Fig. 3L). Fibril organization was also assessed by Gray-Level Co-Occurrence Matrix (GLCM) analysis of SHG images but showed no significant difference between ns-siRNA and SLC7A11-siRNA groups (Fig. 3L).
Genetic ablation of SLC7A11 in PDAC cells only had no effect on tumor growth in genetically engineered mouse models
Results above suggested that the presence of CAFs would likely influence the effect of an SLC7A11 inhibition approach in vivo. We assessed the impact of genetic ablation of SLC7A11 driven by an epithelial-specific promoter (Slc7a11fl/fl, does not affect CAFs), in two transgenic mouse models of PDAC [KC and KPC (42)]. Slc7a11 knockout was confirmed by RNA ISH and Western blot analysis (Fig. 4A–B). We observed no significant difference in PanIN formation between control and Slc7a11fl/fl KC mice (Fig. 4C). Similarly, in KPC mice we observed no difference in survival (Fig. 4D) or intratumoral αSMA-positive CAFs (Fig. 4E), but did note a significant decrease in intratumoral collagen in KPC Slc7a11fl/fl mice, relative to controls (Fig. 4F). In vitro, we observed higher basal expression of SLC7A11 was observed in KPC PDAC cells relative to CAFs (Supplementary Fig. S4A). Consistent with in vivo results, transient SLC7A11 knockdown in vitro (Supplementary Fig. S4B and S4C) had no effect on KPC PDAC cell proliferation (Fig. 4G), but significantly reduced proliferation of KPC CAFs (Fig. 4H).
Stable knockdown of SLC7A11 in both human PDAC cells and CAFs is required to reduce tumor growth, metastatic spread, and fibrosis
We performed a subcutaneous model to assess the effect of SLC7A11 knockdown in hTERT-immortalized CAFs in combination with a patient-derived PDAC cell line (TKCC5). Stable SLC7A11 knockdown in CAFs substantially reduced tumor incidence (Fig. 5A). To further delineate the contribution of PDAC cells and CAFs to the antitumor effects of SLC7A11 knockdown, we orthotopically implanted MiaPaCa-2 PDAC cells and immortalized CAFs expressing control-shRNA or SLC7A11-shRNA into the pancreas of host mice. SLC7A11 knockdown in either PDAC cells alone, CAFs alone or both reduced PDAC tumor growth (Fig. 5B). SLC7A11 knockdown in PDAC cells alone or in both cell types significantly reduced metastasis (Fig. 5C). SLC7A11 knockdown in CAFs alone or in both cell types significantly reduced intratumoral collagen (fibrosis; Fig. 5D). In addition, SLC7A11 knockdown in either PDAC cells or CAFs alone trended toward decreased fibrosis (Fig. 5D) and metastases (Fig. 5C), respectively, although this was not significant.
Gene silencing nanoparticles targeting SLC7A11 decreased orthotopic tumor growth, CAF activity, and fibrosis
To overcome the physical barrier of fibrosis and deliver SLC7A11-siRNA to PDAC mouse tumors, we used Star 3 nanoparticles (43). Star 3+SLC7A11-siRNA decreased SLC7A11 protein levels in orthotopic pancreatic tumors (Fig. 6A). The therapeutic efficacy of Star 3+SLC7A11-siRNA was then assessed using a therapeutic regimen on tumors randomized on the basis of luminescence (Supplementary Fig. S4D), with or without Abraxane (human albumin-bound paclitaxel; Fig. 6B). While Abraxane treatment did not affect tumor growth, SLC7A11 inhibition alone, or in combination, significantly decreased tumor growth (Fig. 6C). In addition, SLC7A11 knockdown alone led to the greatest reduction in the incidence of mice with metastasis (50% in control-siRNA + albumin; 37.5% in control-siRNA/SLC7A1-siRNA+Abraxane; 28.6% in SLC7A11-siRNA+albumin), but had no effect on the number of metastases per mouse (Fig. 6D). Furthermore, Star 3+SLC7A11-siRNA significantly decreased intratumoral αSMA-positive cells (Fig. 6E) and picrosirius red staining (fibrosis), relative to controls (Fig. 6F), although fibril density and organization were not significantly affected (Supplementary Fig. S4E–S4G). This resulted in an increase in the fraction of open CD31-positive blood vessels, relative to controls (Fig. 6G), suggesting normalization of intratumoral vasculature.
SLC7A11 inhibition in 3D human PDAC tumor explants decreased activated CAFs and proliferation
We silenced (Star 3+SLC7A11-siRNA) or inhibited (sulfasalazine) SLC7A11 in 3D human PDAC tumor explants obtained from surgical resections (Fig. 7A). This model retains 3D organization and interactions of PDAC cells and stromal cells for up to 12 days in a tissue culture dish (46). Knockdown of SLC7A11 significantly decreased activated CAFs, cytokeratin-positive tumor cells, and cell proliferation relative to controls (Fig. 7B–E). Using an orthogonal approach (SSZ), we observed a similar reduction in activated CAFs and cell proliferation relative to controls (Fig. 7F and G).
In this study, we showed for the first time that high SLC7A11 expression in the stroma of human PDAC tumors predicts poorer patient survival. A key finding of this study is the importance of SLC7A11 in regulating the growth and function of CAFs, and that inhibiting SLC7A11 in CAFs is essential to maximize the full therapeutic benefit of targeting SLC7A11 in PDAC.
High SLC7A11 expression has been reported to predict poorer survival in different tumor types (15, 49, 50). However, the role of SLC7A11 in PDAC is less clear. Maurer and colleagues (51) and more recently Badgley and colleagues (23) observed significantly higher levels of SLC7A11 mRNA in the PDAC epithelial compartment relative to stroma. We found that SLC7A11 mRNA expression did not correlate with overall survival, but that high SLC7A11 protein in the stroma (but not tumor), was prognostic of poorer patient survival (APGI ICGC cohort). Our results highlight that segregation of tumor and stroma is a critical consideration in any prognostic analysis for PDAC. Interestingly, we identified a subpopulation of patients (TumorSLC7A11lowStromaSLC7A11high) that had the poorest overall survival, which may suggest unique PDAC cell:CAF metabolic dependencies that drive more aggressive disease and requires further investigation. Our results highlight that despite the higher average expression of SLC7A11 in the tumor compartment, expression in the PDAC stroma may be functionally significant for disease progression. In addition, a retrospective analysis of gene expression array data from (37) showed that SLC7A11 expression was elevated in myCAFs and iCAFs, opening an avenue for future investigations into the immune modulatory role of SLC7A11 in iCAFs.
Functional inhibition (knockdown vs. drug) of SLC7A11 in CAFs reduced their proliferation and induced cell death including ferroptosis. Interestingly, both SLC7A11 knockdown and SSZ treatment in CAFs induced senescence, which may have been a response to amino acid deprivation and hindered protein synthesis. Indeed, Daher and colleagues (21) demonstrated that SLC7A11 knockout in PDAC cells induced an amino acid stress response. We confirmed that SLC7A11 inhibition significantly hindered cystine uptake and GSH. A key point of difference between siRNA-based and SSZ-based approaches was that SLC7A11 knockdown alone did not significantly increase oxidative stress in CAFs, whereas SSZ did. A potential explanation is that SSZ can decrease levels of additional enzymes and signaling involved in oxidative stress protection (52–54), which might induce oxidative stress more rapidly than knockdown. Our results highlighted the crucial dependence of CAFs on SLC7A11 for cystine uptake and GSH synthesis.
Stable SLC7A11 knockdown in immortalized CAFs combined with multiple PDAC lines reduced spheroid growth, implying that SLC7A11 targeting in CAFs may be beneficial regardless of tumor cell heterogeneity. The lack of effect observed with SLC7A11 knockdown in PDAC cells alone may have been due to their reliance on CAFs for spheroid formation and the 3-fold excess of CAFs, potentially placing more emphasis on the role of CAFs. This is implied by our spheroid outgrowth assay, whereby SLC7A11 knockdown in PDAC cells was equally effective to knockdown in CAFs with equal cell ratios. Given our immortalized CAFs had adapted to stable SLC7A11 knockdown (no antiproliferative effect), it is likely our observations are a consequence of reducing paracrine protumor signals from CAFs in addition to potentially reducing CAF survival under more stressful conditions than a 2D monoculture. Furthermore, SLC7A11 knockdown reduced CAF-mediated remodeling of 3D collagen in vitro, suggesting this approach might help overcome a PDAC drug delivery barrier.
The antifibrotic effect of conditional SLC7A11 knockout in KPC tumors, in the absence of a survival benefit, implied profibrogenic cross-talk between tumor and stromal cells had been disrupted. This was also reflected in in vitro proliferation assays. Our results are in contrast to prior KPC studies in which SLC7A11 inhibition or genetic ablation (22, 23) significantly reduced tumor growth. A key difference is that these prior studies did not selectively target PDAC cells making it possible that CAFs were affected. We can directly compare our results to those of Badgley and colleagues (23) who utilized KPC mice with deletion of SLC7A11 under temporal control of tamoxifen and driven from the Rosa26 locus, meaning that SLC7A11 is lost from tumor and stroma. In contrast, our model specifically ablated SLC7A11 in pancreatic epithelial cells. These differences support our hypothesis regarding the importance of SLC7A11 in the PDAC microenvironment. However, the difference may also be explained by selection of PDAC cells resistant to SLC7A11 inhibition from tumor initiation.
In an orthotopic PDAC mouse model, stable SLC7A11 knockdown in PDAC cells, immortalized CAFs or both reduced PDAC tumor growth, implying that CAFs and PDAC cells are equally important targets for SLC7A11 inhibition. Importantly, SLC7A11 knockdown in PDAC cells alone significantly reduced metastasis while knockdown in CAFs significantly decreased fibrosis, with the dual knockdown group exhibiting both. We also observed trends (although not significant) toward decreased fibrosis and metastases in the PDAC only and CAF only groups, respectively. The results again highlight that dual inhibition of PDAC cells and CAFs is important to maximize stromal remodeling and antimetastatic effects, but that inhibition of SLC7A11 in either compartment has the potential to interfere with protumor cross-talk. While the outcome is consistent with antitumor effects observed in prior studies (22, 23), our results appear to be at odds with our KPC model findings and in vitro spheroid results. These differences are likely explained by species and experimental approach. In contrast to human PDAC cells, SLC7A11 knockdown in KPC PDAC cells had no effect on proliferation, suggesting mouse KPC PDAC cells are more resistant to this approach. In our spheroid assay, as discussed above, we utilized an excess of and were dependent on CAFs for spheroid formation, potentially placing a heavier emphasis on the role of CAFs.
Our findings collectively demonstrate that SLC7A11 knockdown in PDAC cells or CAFs has the potential to reduce tumor growth and fibrosis, and that the most effective therapeutic approach is to inhibit SLC7A11 in both. Our results also indicate that SLC7A11 inhibition might interfere with protumor cross-talk between PDAC cells and CAFs. While we cannot confirm as yet the central cross-talk mechanism(s) that are influenced, these may encompass: (i) reducing activated CAF numbers available to support tumor cells; (ii) directly or indirectly interfering with fibrosis production/remodeling by CAFs; (iii) inhibiting secretion of prosurvival signaling molecules; (iv) interfering with metabolic cross-talk that could help support survival of both cell types in the nutrient-poor tumor microenvironment. This will be the subject of future studies.
To complement prior work highlighting the potential of SLC7A11 inhibition as a therapeutic approach (22, 23), we again used an orthotopic model, with a defined premortality endpoint to allow for time-matched comparison of CAF activation, fibrosis, and tumor size. We used a polymeric nanoparticle (Star 3) that we developed to package SLC7A11-siRNA and overcome physical barriers to drug delivery in PDAC (43). Our nanoparticle accumulates in PDAC tumors without specific targeting, but is rapidly cleared from normal organs, minimizing the chance for off-target toxicity (43). Star 3+SLC7A11-siRNA decreased tumor growth and reduced the incidence of mice with metastases, although metastases per mouse were not significantly reduced. This dampened antimetastatic effect may reflect the more potent impact of stable SLC7A11 knockdown from tumor establishment. However, the overall lower frequency of metastases in our therapeutic model (due to biological variation), may have limited statistical power. Future studies will assess this in more metastatic models of the disease. We also observed decreased CAF activation and intratumoral collagen (fibrosis), as well as normalized tumor vasculature. It should be noted that our approach used human-specific SLC7A11-siRNA and would not have targeted mouse cells. Thus, observed effects may be an underestimate as mouse CAFs can be corecruited/activated in the model. Regardless, our results demonstrate the efficacy of Star 3+SLC7A11-siRNA against PDAC tumors and its potential to alleviate a drug delivery barrier. While we did not sensitize tumors to Abraxane in our study, the dosing schedule selected was suboptimal to test whether SLC7A11 inhibition could sensitize to lower amounts of Abraxane. Future studies will investigate chemosensitization to higher doses of Abraxane and other potential therapeutic combinations that increase oxidative stress.
There is also some debate around the role of fibrosis in PDAC progression, with some studies in transgenic mouse PDAC models suggesting the PDAC stroma restrains tumor growth (55–57). Limitations of these studies included off-target effects on vasculature, causing hypoxia that drives more aggressive disease, and interference in signaling pathways that may have affected stromal cells that restrain tumor growth. Moreover, genetic ablation studies from tumor development may not reflect a late-stage tumor's response to therapeutic inhibition. However, these studies highlight the need to specifically target stromal cell populations that support tumor growth. Our approach overcomes these limitations by targeting a key metabolic function in both CAFs and PDAC cells, and reduces rather than depletes fibrosis and activated CAFs.
Finally, we validated our in vivo findings using PDAC patient-derived 3D explant model developed by our lab that retains the extensive fibrosis, 3D spatial architecture and cellular/stromal interactions of the original patient's tumor (46). In multiple patient explants, SLC7A11 inhibition or knockdown significantly reduced tumor cell and activated CAF frequency, confirming our in vitro and in vivo findings. This study brings together over a decade of research into the therapeutic potential of SLC7A11 inhibition in PDAC. Taken together, our findings and those of previous studies have demonstrated that SLC7A11 inhibition in PDAC is a multipronged therapeutic approach that targets both tumor and stroma to exert its maximal therapeutic effect.
D. Goldstein reports grants from Amgen and grants from Cellgene outside the submitted work. O.J. Sansom reports grants from Novartis and Astra Zeneca; personal fees from Boehringer Ingelheim; other support from iOnctura, Aglios, and grants from Cancer Research Technology, a subsidiary of CRUK, outside the submitted work. M.V. Apte reports grants from National Health and Medical Research Council of Australia, US Department of Defense, and grants from CONCERT outside the submitted work. No disclosures were reported by the other authors.
G. Sharbeen: Conceptualization, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J.A. McCarroll: Conceptualization, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. A. Akerman: Formal analysis, investigation, visualization, writing–review and editing. C. Kopecky: Formal analysis, investigation, visualization, writing–review and editing. J. Youkhana: Formal analysis, supervision, investigation, visualization, writing–review and editing. J. Kokkinos: Conceptualization, formal analysis, supervision, investigation, visualization, writing–review and editing. J. Holst: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, writing–review and editing. C. Boyer: Conceptualization, resources, writing–review and editing. M. Erkan: Resources, funding acquisition, investigation, methodology, writing–review and editing. D. Goldstein: Conceptualization, resources, formal analysis, funding acquisition, investigation, methodology, writing–review and editing. P. Timpson: Formal analysis, funding acquisition, investigation, visualization, writing–review and editing. T.R. Cox: Formal analysis, funding acquisition, investigation, visualization, writing–review and editing. B.A. Pereira: Formal analysis, investigation, visualization, writing–review and editing. J.L. Chitty: Formal analysis, investigation, visualization, writing–review and editing. S.K. Fey: Formal analysis, investigation, visualization, methodology, writing–review and editing. A.K. Najumudeen: Formal analysis, investigation, visualization, methodology, writing–review and editing. A.D. Campbell: Formal analysis, investigation, visualization, writing–review and editing. O.J. Sansom: Conceptualization, formal analysis, funding acquisition, investigation, visualization, methodology, writing–review and editing. R.C. Ignacio: Investigation, visualization, writing–review and editing. S. Naim: Investigation, visualization, writing–review and editing. J. Liu: Investigation, writing–review and editing. N. Russia: Investigation, writing–review and editing. J. Lee: Investigation, writing–review and editing. A. Chou: Investigation, writing–review and editing. A. Johns: Resources, investigation, writing–review and editing. A.J. Gill: Conceptualization, resources, investigation, methodology, writing–review and editing. E. Gonzales-Aloy: Formal analysis, investigation, visualization, writing–review and editing. V. Gebski: Formal analysis, investigation, writing–review and editing. Y. Guan: Formal analysis, investigation, writing–review and editing. M. Pajic: Resources, investigation, methodology, writing–review and editing. N. Turner: Conceptualization, methodology, writing–review and editing. M.V. Apte: Resources, writing–review and editing. T.P. Davis: Conceptualization, resources, writing–review and editing. J.P. Morton: Conceptualization, resources, investigation, writing–review and editing. K.S. Haghighi: Conceptualization, resources, formal analysis, investigation, methodology, writing–review and editing. J. Kasparian: Resources, investigation, writing–review and editing. B.J. McLean: Resources, investigation, writing–review and editing. Y.F. Setargew: Resources, data curation, investigation, writing–review and editing. APGI: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. P.A. Phillips: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.
Australian Pancreatic Cancer Genome Initiative (APGI): Garvan Institute of Medical Research Amber L. Johns1, Anthony J Gill1, 5, David K. Chang1, 22, Lorraine A. Chantrill1,8, Angela Chou1,5, Marina Pajic1, Angela Steinmann1, Mehreen Arshi1, Ali Drury1, Danielle Froio1, Ashleigh Parkin1, Paul Timpson1, David Hermann1. QIMR Berghofer Medical Research Institute Nicola Waddell2, John V. Pearson2, Ann-Marie Patch2, Katia Nones2, Felicity Newell2, Pamela Mukhopadhyay2, Venkateswar Addala2, Stephen Kazakoff2, Oliver Holmes2, Conrad Leonard2, Scott Wood2, Christina Xu2. University of Melbourne, Centre for Cancer Research Sean M. Grimmond3, Oliver Hofmann3. University of QLD, IMB Angelika Christ4, Tim Bruxner4. Royal North Shore Hospital Jaswinder S. Samra5, Jennifer Arena5, Nick Pavlakis5, Hilda A. High5, Anubhav Mittal5. Bankstown Hospital Ray Asghari6, Neil D. Merrett6, Darren Pavey6, Amitabha Das6. Liverpool Hospital Peter H. Cosman7, Kasim Ismail7, Chelsie O'Connnor7. St Vincent's Hospital Alina Stoita8, David Williams8, Allan Spigellman8. Westmead Hospital Vincent W. Lam9, Duncan McLeod9, Adnan M. Nagrial1,9, Judy Kirk9. Royal Prince Alfred Hospital, Chris O'Brien Lifehouse James G. Kench10, Peter Grimison10, Caroline L. Cooper10, Charbel Sandroussi10, Annabel Goodwin7,10. Prince of Wales Hospital R. Scott Mead1,11, Katherine Tucker11, Lesley Andrews11. Fremantle Hospital Michael Texler12, Cindy Forest12, Krishna P. Epari12, Mo Ballal12, David R. Fletcher12, Sanjay Mukhedkar12. St John of God Healthcare Nikolajs Zeps14, Maria Beilin14, Kynan Feeney14. Royal Adelaide Hospital Nan Q Nguyen15, Andrew R. Ruszkiewicz15, Chris Worthley15. Flinders Medical Centre John Chen16, Mark E. Brooke-Smith16, Virginia Papangelis16. Envoi Pathology Andrew D. Clouston17, Patrick Martin17. Princess Alexandria Hospital Andrew P. Barbour18, Thomas J. O'Rourke18, Jonathan W. Fawcett18, Kellee Slater18, Michael Hatzifotis18, Peter Hodgkinson18. Austin Hospital Mehrdad Nikfarjam19. Johns Hopkins Medical Institutes James R. Eshleman20, Ralph H. Hruban20, Christopher L. Wolfgang20, Mary Hodgin20. ARC-Net Centre for Applied Research on Cancer Aldo Scarpa21, Rita T. Lawlor21, Stefania Beghelli21, Vincenzo Corbo21, Maria Scardoni21, Claudio Bassi21. University of Glasgow Andrew V Biankin1, 22, Judith Dixon22, Craig Nourse22, Nigel B. Jamieson22. 1The Kinghorn Cancer Centre, Garvan Institute of Medical Research, 370 Victoria Street, Darlinghurst, Sydney, New South Wales 2010, Australia. 2QIMR Berghofer Medical Research Institute, 300 Herston Rd, Herston, Queensland 4006, Australia. 3University of Melbourne, Centre for Cancer Research, Victorian Comprehensive Cancer Centre, 305 Grattan Street, Melbourne, Victoria 3000, Australia. 4Institute for Molecular Bioscience, University of QLD, St Lucia, Queensland 4072, Australia. 5Royal North Shore Hospital, Westbourne Street, St Leonards, New South Wales 2065, Australia. 6Bankstown Hospital, Eldridge Road, Bankstown, New South Wales 2200, Australia. 7Liverpool Hospital, Elizabeth Street, Liverpool, New South Wales 2170, Australia. 8 St Vincent's Hospital, 390 Victoria Street, Darlinghurst, New South Wales, 2010 Australia. 9Westmead Hospital, Hawkesbury and Darcy Roads, Westmead, New South Wales 2145, Australia. 10Royal Prince Alfred Hospital, Missenden Road, Camperdown, New South Wales 2050, Australia. 11Prince of Wales Hospital, Barker Street, Randwick, New South Wales 2031, Australia. 12Fremantle Hospital, Alma Street, Fremantle, Western Australia 6959, Australia. 13Sir Charles Gairdner Hospital, Hospital Avenue, Nedlands, Western Australia 6009, Australia. 14St John of God Healthcare, 12 Salvado Road, Subiaco, Western Australia 6008, Australia. 15Royal Adelaide Hospital, North Terrace, Adelaide, South Australia 5000, Australia. 16Flinders Medical Centre, Flinders Drive, Bedford Park, South Australia 5042, Australia. 17Envoi Pathology, 1/49 Butterfield Street, Herston, Queensland 4006, Australia. 18Princess Alexandria Hospital, Cornwall Street & Ipswich Road, Woolloongabba, Queensland 4102, Australia. 19Austin Hospital, 145 Studley Road, Heidelberg, Victoria 3084, Australia. 20Johns Hopkins Medical Institute, 600 North Wolfe Street, Baltimore, Maryland 21287, USA. 21ARC-NET Center for Applied Research on Cancer, University of Verona, Via dell'Artigliere, 19 37129 Verona, Province of Verona, Italy. 22Wolfson Wohl Cancer Research Centre, Institute of Cancer Sciences, University of Glasgow, Garscube Estate, Switchback Road, Bearsden, Glasgow, Scotland G61 1BD, United Kingdom.
The authors would like to thank Dr. Andrea Nunez and Ms. Amanda Mawson for their assistance in mouse monitoring. They acknowledge the Flow Cytometry, Biomedical Imaging, and Biological Resource Imaging Facilities within the Mark Wainwright Analytical Centre at the University of New South Wales for their technical support. Biospecimens and clinical data for prognostic studies were provided by the APGI (www.pancreaticcancer.net.au), which is supported by an Avner Grant from PanKind, The Australian Pancreatic Cancer Foundation (www.pankind.org.au). Biospecimens and data used for 3D explants were obtained from the HSA Biobank, UNSW Biorepository, UNSW Sydney, Australia. The authors sincerely thank the patients who consented to donate their tumor samples for research. They would like to thank Dr. Carmel Quinn and Dr. Anusha Hettiaratchi of the HSA Biobank for their support in managing clinical samples and patient consent. The authors would also like to acknowledge their community consumers Mr. Gino Iori and Ms. Claire Harvey for their invaluable input on the project and grant applications.
This research was made possible by major funding from NHMRC project grant (P.A. Phillips, J.A. McCarroll, J.P. Morton, D. Goldstein, APP1144108) and an Avner Grant from PanKind, The Australian Pancreatic Cancer Foundation (Avner Innovation Grant (P.A. Phillips, J.A. McCarroll, D. Goldstein, J. Holst, J.P. Morton, T.P. Davis, and G. Sharbeen, APCF0050618; https://pankind.org.au). The following sources supported author contributions and research: NHMRC CDF-I (P.A. Phillips, APP1024896), NHMRC Ideas Grant (P.A. Phillips, J.A. McCarroll, G. Sharbeen, APP2002707), Cancer-Institute NSW ECF/CDFs (P.A. Phillips 08/ECF/1-37; G. Sharbeen, CDF181166; J.A. McCarroll, CDF102; T.R. Cox, CDF171105), Cancer Institute NSW Innovation Grant (P.A. Phillips, 09/RFG/2-58), Cancer Institute NSW ‘The Professor Rob Sutherland AO Make a Difference Award' (D. Goldstein, 2017/AWD002), Cancer Australia/Cancer Council (P.A. Phillips, J.A. McCarroll, and D. Goldstein, APP1126736), Cancer Australia/Kids Cancer Project (APP1184840 to J.A. McCarroll, P.A. Phillips, T.P. Davis), Translational Cancer Research Network and Australian Postgraduate Award Scholarships (A. Akerman), Australian Government Research Training Program Scholarship & UNSW Sydney Scientia PhD Scholarship (J. Kokkinos), Cure Cancer Australia (G. Sharbeen, APP1122758), Tour de Cure PhD Support Scholarship (J. Kokkinos, P.A. Phillips, D. Goldstein, RSP-011-18/19), Tour de Cure Established Research Grant (P.A. Phillips, J.A. McCarroll, D. Goldstein, UNSWR002; RSP-235-19/20), Tour de Cure Pioneering Research Grant (G. Sharbeen, P.A. Phillips, D. Goldstein UNSWR004; RSP-255-19/20), UNSW Interlude grant scheme (P.A. Phillips, RG210839), Cancer Research UK Core Funding and Grand Challenge grants (O.J. Sansom, A.D. Campbell, A.K. Najumudeen, and S. Fey, A17196, A21139, A25045), Pancreatic Cancer UK Future Leaders Academy (O.J. Sansom, S. Fey), NHRMC project grants (T.R. Cox and J.L. Chitty, APP1140125), NHMRC CDF-II (T.R. Cox, APP1158590), NHMRC Senior Research Fellowship (P. Timpson, APP1136974), Len Ainsworth Pancreatic Cancer Fellowship and support from Suttons, Cancer Council NSW (T.R. Cox and J.L. Chitty, RG19-09). We also acknowledge the generous philanthropic support of Mr Paul Dainty, Dr Marjorie O'Neil and Dr Keri Spooner.
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.