Mice

All animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Michigan. Wild-type (WT) C57/BL6 mice (stock # 000664) from Jackson Laboratory were used for PSC isolation and orthotopic transplantation experiments at 8–12 weeks of age, including male and female mice. For orthotopic transplantation, 7.5 × 104 mT3 (provided by Dr. David A. Tuveson) [36] or 7.5 × 104 4662 cells (provided by Robert H. Vonderheide) [37] derived from primary PDAC in KrasLSL-G12D/+; Trp53LSL-R172H/+; Pdx1-Cre (KPC) mice of a C57/BL6 genetic background, were resuspended as a 30 µl suspension of 50% Matrigel (#356231, Corning) in PBS and injected into the pancreas. At 4 weeks post-transplantation, mice received an intraperitoneal injection with 60 mg/kg of Hypoxyprobe (pimonidazole hydrochloride, Hypoxyprobe, Inc) and were sacrificed 1.5–2 h later for flow cytometry analysis.

Cell lines and cell culture

PSCs were isolated from WT mice by enzymatic digestion of pancreatic tissue and subsequent density gradient centrifugation as previously described [8, 38]. Primary PSC lines between passages 2 and 4 were used for all experiments. The FB1 CAF line was generated from an iKras* p53* mouse [39] by fluorescence-activated cell sorting of PDGFRα+;EPCAM− cells. The mT3 (provided by Dr. David A. Tuveson) [36] and 4662 (provided by Robert H. Vonderheide) [37] PDAC cell lines were derived from primary murine KPC PDAC. FB1, mT3, and 4662 cell lines were cultured no more than 20–25 passages. All cells were passaged in DMEM with 10% FBS and 1% penicillin/streptomycin (Thermo Fisher). For 3D cocultures, PSCs were seeded in Matrigel (#356231, Corning) in a transwell insert (#662610, Greiner Bio-One) and cultured with PDAC tumor organoids in the lower compartment of the 24-well plate in DMEM containing 5% FBS and 1% penicillin/streptomycin (Thermo Fisher). For IL1ɑ neutralization experiments, cocultures were treated with 3 µg/ml IL1ɑ-neutralizing antibody (#MAB4001, R&D Systems) or isotype control antibody (#400902, BioLegend) for 72 h. For IL1R1 neutralization experiments, cocultures were treated with 0.5 µg/ml IL1R1-neutralizing antibody (#PA5-47937, Invitrogen) or isotype control antibody (#AB108C, R&D Systems) for 72 h. Cell line authentication for FB1 and mT3 was not performed. The 4662 cells were authenticated by the Research Animal Diagnostic Laboratory (RADIL) at the University of Missouri. Mycoplasma testing (MycoAlert Detection Kit, Lonza) was performed monthly.

Lentiviral-mediated shRNA transduction

PDAC cell lines were transduced with lentivirus containing shRNA plasmids at optimized viral titers. Stable cell lines were established post-puromycin selection. The following shRNA plasmids were used: pGIPZ Scrambled shRNA (#RH4346, Horizon), pGIPZ HIF1α shRNA (#RMM4431-200404026, Horizon), pLKO.1 Scrambled shRNA (#1864, Addgene), pLKO.1 HIF2α shRNA (#TRCN0000082307, Sigma).

Quantitative RT-PCR

Total RNA was isolated from cells using the RNeasy mini kit (#74104, Qiagen). cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit (#4368814, Applied Biosystems). PCR reactions were performed using SYBR Green PCR reagents (#A25742, Applied Biosystems) mixed with indicated cDNAs and primers (primer sequences are listed in Table S1) in a QuantStudio Real-Time PCR system (Applied Biosystems). Expression levels were normalized by 18S rRNA.

Immunofluorescence

Tissues were fixed in 4% paraformaldehyde/PBS (4 °C, overnight) and processed for paraffin embedding. For immunofluorescence, slides were boiled for 20 min in 10 mM sodium citrate (pH 6.0) for antigen retrieval and blocked with 5% serum/0.3% Triton X-100 for 1 h. Sections were incubated with FITC-conjugated Hypoxyprobe-1-MAb1 (4.3.11.3, #FITC-Mab, 1:500, Hypoxyprobe, Inc) and Alexa Fluor 594-conjugated PDPN antibody (8.1.1, #127414, 1:250) diluted in 1% BSA/0.3% Triton X-100 overnight at 4 °C. Slides were counterstained with DAPI (Invitrogen) and mounted in Prolong Gold antifade reagent (Invitrogen). Fluorescence images were acquired using an Olympus IX73 microscope.

Flow cytometry

Single-cell suspensions from mouse tissues were prepared as previously described [20]. Tumor tissues were cut in half, and one half was minced and processed for flow cytometry analysis. Cells were stained in PBS/0.5% FBS/2 mM EDTA with the following fluorochrome-conjugated antibodies: BV421-conjugated anti-Ly6C (HK1.4, #128031, 1:100), PerCP-Cy5.5-conjugated anti-CD45 (30-F11, #103132, 1:200), PE-conjugated anti-EPCAM (G8.8, #118205, 1:200), PE-conjugated anti-CD31 (390, #102407, 1:200), PE-Cy7-conjugated anti-PDPN (8.1.1, #127411, 1:100), APC-Cy7-conjugated anti-MHCII (M5/114.15.2, #107628, 1:300) (from BioLegend); FITC-conjugated Hypoxyprobe-1-MAb1 (4.3.11.3, #FITC-Mab, 1:200) (from Hypoxyprobe, Inc). The viability marker Zombie Aqua was purchased from BioLegend (#423102). Flow cytometry was performed on a ZE5 Cell Analyzer (Bio-Rad), and data were analyzed using FlowJo software.

ELISA

For ELISA of media, 3D cocultures were grown under 21% O2 or 1% O2 for 72 h. Media were collected, spun down, and assayed using the manufacturer’s protocol. ELISA assays used were IL1ɑ (#433404, BioLegend), CXCL1 (#EMCXCL1, Invitrogen), IL6 (#DY406-05, R&D Systems), and LIF (#445104, BioLegend).

RNA-seq and data analysis

Total RNA was isolated from cells using the RNeasy mini kit (#74104, Qiagen). Libraries were constructed using NEB polyA RNA ultra II and subsequently subjected to 150 cycles of sequencing on NovaSeq-6000 (Illumina). Adapters were trimmed using Cutadapt (v2.3). FastQC (v0.11.8) was used to ensure the quality of data. Reads were mapped to the mouse genome (GRCm38) using STAR (v2.6.1b) and assigned count estimates to genes with RSEM (v1.3.1). Alignment options followed ENCODE standards for RNA-seq. FastQC was used in an additional post-alignment step to ensure that only high-quality data were used for expression quantitation and differential expression. Differential gene expression analysis was performed using DESeq2, using a negative binomial generalized linear model (thresholds: linear fold change >1.5 or <−1.5, Benjamini–Hochberg FDR (Padj) < 0.05). GSEA was performed using GSEA 4.1.0.

Single-cell RNA-seq analysis

Human single-cell RNA-seq (scRNA-seq) data were previously published in [23], and fibroblasts were annotated in ref. [24]. Both raw and processed data are available at the NIH dbGaP database (accession #phs002071.v1.p1; [23]), with full clinical annotation. Downstream analysis was performed using Seurat V4.0.3 [40]. Hypoxia signature scoring was performed using Seurat’s “AddModuleScore” function.

Western blot analysis

Cells were lysed with 10 mmol/L Tris at pH 7.5, 150 mmol/L NaCl, 5 mmol/L EDTA, 0.1% SDS, and protease/phosphatase inhibitor cocktail (#78440, Thermo Fisher). Cell lysates were separated by SDS-PAGE, transferred to nitrocellulose membranes, blotted with primary antibodies overnight at 4 °C, and detected using horseradish peroxidase-conjugated secondary antibodies followed by exposure to chemiluminescence reagents (#PI34580, Thermo Fisher). The following antibodies were used: rabbit anti-HIF1α (#10006421, 1:500, Cayman), goat anti-HIF2α (#AF2997, 1 µg/ml, R&D Systems), mouse anti-beta actin (#MA1-91399, 1:50,000, Invitrogen), HRP-linked anti-rabbit IgG (#7074, 1:10,000, Cell Signaling), HRP-linked anti-mouse IgG (#7076, 1:60,000, Cell Signaling), and HRP-linked anti-goat IgG (#705035147, 1:20,000, Jackson ImmunoResearch).

Statistical analysis

Data were analyzed using GraphPad Prism 7 software. Statistical tests with normally distributed variables included two-tailed student’s t-test and two-way ANOVA. D’Agostino and Pearson test and/or Shapiro–Wilk test was used to test the normality of sample distribution. When variables were not normally distributed, we performed nonparametric Mann–Whitney test. Bonferroni correction was applied for multiple comparisons. P-value < 0.05 was considered statistically significant. No statistical method was used to predetermine sample sizes, experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment.