Targeted therapy is effective in many tumor types including lung cancer, the leading cause of cancer mortality. Paradigm defining examples are targeted therapies directed against non-small cell lung cancer (NSCLC) subtypes with oncogenic alterations in EGFR, ALK and KRAS. The success of targeted therapy is limited by drug-tolerant tumor cells which withstand and adapt to treatment and comprise the residual disease state that is typical during treatment with clinical targeted therapies. Here, we integrate studies in patient-derived and immunocompetent lung cancer models and clinical specimens obtained from patients on targeted therapy to uncover a focal adhesion kinase (FAK)-YAP signaling axis that promotes residual disease during oncogenic EGFR-, ALK-, and KRAS-targeted therapies. FAK-YAP signaling inhibition combined with the primary targeted therapy suppressed residual drug-tolerant cells and enhanced tumor responses. This study unveils a FAK-YAP signaling module that promotes residual disease in lung cancer and mechanism-based therapeutic strategies to improve tumor response.

A focal adhesion kinase-YAP signaling axis drives drug tolerant persister cells and residual disease in lung cancer

The NIH LINCS L1000 database contains gene expression data from cultured human cells that were treated with small molecule and genetic perturbagens. Level 4 data was sourced from the Gene Expression Omnibus Series GSE70138. Expression data was restricted to small-molecule perturbagens and overlapped with the residual disease signature (N = 83 genes). Using a previously published computational pipeline66,67 (65, 66), a score for each signature-drug pair was determined using a non-parametric rank-based method that is similar to the Kolmogorov–Smirnov test statistic, where negative scores indicate genes in the ranked drug profile are oppositely regulated in the ranked signature. For the upregulated NSCLC residual disease signature, drug-gene expression profiles were chosen that produce the greatest significant negative score. P-values for drug profiles were determined by comparing their scores to a distribution of random scores and adjusted with the false discovery rate (FDR; Benjamini-Hochberg, a = 0.05) method.

LINCS L1000 concordance score

TCGA LUAD data was retrieved (https://portal.gdc.cancer.gov/projects/TCGA-LUAD). Samples were grouped according to their expression quantile (High > 0.75; 0.25 ≤ Medium ≥ 0.75; Low < 0.25) for the sum gene expression of the corresponding signature. Correlation of signature expression to patient overall survival was assessed. For the latter, data were processed in R with survival (v.3.2) using default arguments for the single outcome, single-event type and depicted via Kaplan-Meier (log-rank test) curves.

Survival analysis of YAP gene signature

Single-cell sequencing data were derived from previously published work7 and filtered to limit analysis to malignant lung epithelial cells only (N cells: TN = 621, RD = 484, PD = 138). Differentially expressed gene sets for YAP activation (YAP-5SA-UP), persistence (PC9, H3122, H358 and overlapping genes across all cell line models [overlap persisters]), and patient treatment timepoint (TN, RD, PD) were compared to the patient gene expression dataset via permutation analysis (R package, GSALightning, v.1.1.7). Subject classes were assigned to every single cell based on the corresponding time point, where RD = “RD” and TN or PD = “nonRD”. Each gene was tested for significance with unpaired t-tests and the mean across genes was used to calculate the gene set statistics. Multiple testing correction was done via Benjamini-Hochberg. The union of significantly upregulated genes from the YAP-5SA_UP gene set, significantly upregulated genes across persister models, and the RD treatment timepoint were then used to define the YAP gene signature (adjusted p-value < 0.05). Other (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2021.10.23.465573; this version posted October 24, 2021. The copyright holder for this preprint 32 signatures (cell cycle, FAK) were determined from relevant literature7,31. Signature expression for single-cell data was processed and plotted in R with ggplot2 (v.3.3.3) and Seurat (v.3.2.2).

YAP signature in scRNA seq data of patient samples

Formalin-fixed paraffin-embedded (FFPE) tumor blocks were cut at 4-micron thickness and mounted as sections on positively charged histology slides. Immunohistochemistry staining was performed as described previously65. In brief, slides were deparaffinized in xylene, rehydrated and epitope retrieval was induced in a histology pressure cooker using pH 6.1 citrate buffer (Dako Denmark A/S, S2369). After endogenous peroxidase and protein block, slides were incubated with primary antibody solution overnight at 4 °C. Then slides were incubated with secondary antibody for 30 minutes (EnVision Dual Link Labelled Polymer HRP, Agilent K4065), stained using 3,3-DAB, and counterstained with hematoxylin. Slides were dehydrated and mounted before digitization using an Aperio AT2 Slide Scanner (Leica Biosystems) at a 20X objective. Quantification of nuclear YAP levels was performed via the Aperio Image Scope digital pathology software using the nuclear quantification algorithm.

Immunohistochemistry

Humanized xenograft models were kindly established by the laboratory of Jack Roth at MD Anderson Cancer Center as described previously42. All animal use was conducted in accordance with the guidelines of the Animal (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2021.10.23.465573; this version posted October 24, 2021. The copyright holder for this preprint 30 Care and Use Committee of MD Anderson Cancer Center. In brief, female 3-to-4-week-old NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice, which are suitable for the engraftment of human hematopoietic cells ), were housed in microisolator cages under specific pathogen-free conditions in a dedicated humanized mice room in the animal facility at The University of Texas MD Anderson Cancer Center. Mice were given autoclaved acidified water and fed a special diet (Uniprim diet). Human umbilical cord blood units were obtained from MD Anderson Cord Blood Bank under an IRB-approved protocol. Fresh cord blood units were delivered within 24 h of harvest and were HLA typed immediately at MD Anderson HLA-typing core facility. Cord blood was diluted to a ratio of 1:3 with phosphate-buffered saline, and mononuclear cells were isolated by using density-gradient centrifugation on Ficoll medium. CD34+ HSPCs were isolated using a direct CD34+ MicroBead kit (Miltenyi Biotec). NSG mice were irradiated with 200 cGy using a 137Cs gamma irradiator. Over 90% pure freshly isolated CD34+ HSPCs were injected intravenously, 24 h after irradiation, at a density of 1 to 2 × 105 CD34+ cells/ mouse. All Hu-NSG mice were verified for humanization before tumor implantation. For PC9-parental, PC9-WT (YAP overexpression), PC9-5SA (hyperactive YAP) cell lines, 5-7x106 cells were injected subcutaneously 8-weeks post humanization of mice. Another 3-4 weeks post tumor cells implantation in humanized mice and when tumor sizes reached 200mm3, animals were randomized into treatment and no-treatment groups based on tumor size and donor HLA type. Five mice per group from multiple umbilical cord blood donors were used. Mice were treated with vehicle or osimertinib (5 mg/kg) orally 5 days a week for consecutive 3 weeks. For immune analysis, erythrocytes in the peripheral blood were lysed with ACK lysis buffer (Fisher Scientific). Single-cell suspensions were prepared. Several 10-color flow cytometry panels were used for immune profiling of both innate and adaptive immune populations in humanized mice and for evaluating immune response after treatment. All samples were run on Attune NxT flow cytometer (Thermo fisher), and data were analyzed by Flow Jo and Kaluza software packages.

Humanized mouse model

All animal experiments were conducted under UCSF IACUC-approved animal protocol no. AN187306-01B. H1975 tumor xenografts were generated by injection of one million cells in a 50/50 suspension of matrigel/PBS into 6- to 8-wk-old female SCID mice. Once the tumors grew to an average size of ~200 mm3, mice were randomized and treated with vehicle (2% HPMC E-50, 0.5% Tween-80 in 50 mM Sodium Citrate Buffer, pH 4.0), 5 mg/kg osimertinib q.d., 50 mg/kg VS-4718 b.i.d., 3 mg/kg VT104 q.d. as well as combinations of osimertinib with VS-4718 or VT104. No substantial toxicity was observed in mice treated with either combination regimen incorporating FAK or TEAD inhibitors by assessment of body weight (Supplementary Fig. 8c) and general animal well-being. EGFR-mutant TH021 and ALK fusion-positive LG0812 PDX, tumors were propagated into 6- to 8-wk-old female SCID mice. Once the tumors grew to an average size of ~400 mm3, mice were randomized and treated with vehicle, 10 mg/kg osimertinib q.d. (TH021) or 6 mg/kg alectinib q.d. (LG0812). For the EGFR-mutant TH021 PDX model, combinatorial treatment effects were assessed upon treatment with vehicle, 5 mg/kg osimertinib q.d., 3 mg/kg VT104 q.d. as well as combinations of osimertinib with VT104. Tumor volume was assessed regularly. At the treatment endpoint, tumors were halved and harvested in ice-cold PBS. One-half of tumors were incubated in 10% neutral-buffered formalin for 24-72 hours and then stored in 70% Ethanol until embedding in paraffin blocks and subsequent analysis by immunohistochemistry. The other half was snap-frozen for later analysis by RNA sequencing.

Subcutaneous xenograft and PDX experiments

Organoid cultures from NSCLC specimens were established as previously described62,63. Organoid cultures were embedded in Reduced Growth Factor Basement Membrane Extract, Type 2 (BME2) matrix (Thermo Fisher Scientific) and maintained in Dulbecco’s Modified Eagle’s Medium/Ham’s nutrient mixture F12 (DMEM/F-12) GlutaMAX supplement, supplemented with 100 U/mL penicillin/streptomycin, 10 mM HEPES, 25 nM hRspondin, 1x B27, 5 mM Nicotinamide, 1.25 mM N-Acetylcysteine, 500 nM A-8301, 500 nM SB202190, 50 µg/mL Primocin, 100 ng/mL hNoggin, 100 ng/mL hFGF-10, and 25 ng/mL hFGF-763,64. Mutational profiling of organoid cultures was performed by whole-exome sequencing. Signaling alterations upon osimertinib treatment in EGFR-mutant organoids were evaluated by treating single suspensions for 2 hours and subsequent Western blot analysis. Drug sensitivity was analyzed by 3D CellTiter-Glo assay (Promega). In brief, single cells suspensions were prepared by TrypLE digestions, and cells (7.5x103 / well) were seeded in BME2 on clear-bottom 96-well plates (Corning). After seven days in culture, newly formed organoids were treated with indicated drug concentrations in reduced growth factor media. Five days after treatment initiation, the viability of cells was assessed. Persister cell generation of organoids was performed as outlined, seeding NSCLC organoid cells (1.8x105 / well) embedded in BME2 in a 6-well plate format. After three days in culture, organoids were treated with 0.1% DMSO or 1 µM osimertinib. DMSO-treated control cells were harvested three days after treatment. For persisters, the drug was replenished every 3 days, and cells were harvested post 11 days on treatment. The engagement of YAP signature genes was evaluated by RNA sequencing and gene set enrichment analysis.

NSCLC organoid cultures

DNA was extracted from snap-frozen cell pellets using the DNeasy Blood & Tissue kit (Qiagen). DNA quality was assessed by automated electrophoresis using the High Sensitivity DNA Kit and an Agilent 2100 BioAnalyzer (Agilent Technologies, Inc.). DNA was quantified using the Qubit dsDNA HS Assay kit and a Qubit 2.0 fluorometer (Thermo Fisher Scientific). Library preparation and paired-end 150bp (PE150, Illumina) DNA sequencing were performed by Novogene (Novogene Corporation, Sacramento, USA). Pair-end fastq files were mapped to the hg19 genome and mutation calling using the SeqMule pipeline61. The vcf files were annotated using ANNOVAR software at a high-performance computing cluster (UCSF Helen Diller Comprehensive Cancer Center). Further analysis of annotated variants was conducted under the RStudio/R environment.

Whole exome sequencing

A BFP-tagged barcode library (Addgene #85968) was lentivirally infected into isogenic EGFR-mutant PC9-C2 and H1975-B10 cells. Cells were sorted and serially titrated to allow for ~1000 unique lineages. After expansion, cells were subjected to 0.1% DMSO or 2 µM osimertinib treatment and frozen down at the indicated timepoints. Cells were thawed, hashed with TotalSeq A anti-human hashtag antibodies (Biolegend), and pooled for single cell RNA sequencing on the 10X chromium v3 platform (10x Genomics). Cell hash libraries were prepared as specified by Biolegend. Custom barcode amplification was performed by two rounds of PCR. Libraries were (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2021.10.23.465573; this version posted October 24, 2021. The copyright holder for this preprint 27 sequenced on the NovaSeq Illumina platform (Center for Advanced Technology, UCSF). After NGS sequencing, cells were called with 10X Cell Ranger pipeline and cell hashes were called using the scEasyMode package in Python. In addition, bulk genomic barcodes were prepared from the same time points used for single-cell RNA sequencing using the Quick Extract gDNA extraction protocol (Lucigen Corporation) and custom barcode amplification primers for NGS library preparation. A custom script for calling genomic barcodes mapping between single-cell genomic barcodes and bulk genomic barcodes collected from the same samples was used to assess population frequency and map onto single-cell transcriptomes. The diversity index was calculated as 1 - Sum_i (pi^2), where pi is the relative abundance of line i. The diversity index is at its maximum when all lineages are equally abundant and decreases if some lineages are enriched and others depleted. The index was scaled by the max possible index given the number of lineages which is max(Lineage diversity index) = 1 - n[(1/n)^2] = 1 - 1/n; n: number of lineages.

scRNA sequencing trajectory

RNA was extracted from snap-frozen tissue or cell pellets. For tissue samples, tissue was minced using a liquid nitrogen-cooled mortar and pestle before RNA extraction. RNA isolation was performed using the RNeasy Mini kit (Qiagen) including an on-column DNase I digestion. RNA quality was assessed by automated electrophoresis using the RNA 6000 Pico Kit and an Agilent 2100 BioAnalyzer (Agilent Technologues, Inc.). RNA was quantified using the Qubit RNA HS Assay Kit and a Qubit 2.0 fluorometer (Thermo Fisher Scientific). Library preparation and paired-end 150bp (PE150, Illumina) RNA sequencing was performed by Novogene (Novogene Corporation, Sacramento, USA). RNA-Seq reads were mapped to the hg19 reference genome using STAR (Spliced Transcripts Align to a Reference, v2.4.2a). The expression level of transcript per million (TPM) reads were quantified using RNA-Seq by Expectation-Maximization algorithm (RSEM v1.2.29). The quantified gene expressions of 26,334 transcripts (including coding genes and non-coding genes) were processed in R studio. Differentially expressed genes between tumor and normal samples were identified using the EdgeR algorithm. Gene set enrichment analysis was done using GSEA 4.0.1 software59,60.

RNA sequencing and gene set enrichment analysis

Cell survival and persister cell numbers upon genetic or pharmacologic perturbations were evaluated by counting cells using a Vi-CELL XR Cell Viability Analyzer (Beckman Coulter, Inc.). Response to escalating drug doses in stable transduced PC9 and H358 cells was analyzed by CellTiter-Glo assay (Promega). For the latter, cells (5x103/ well) were seeded in clear-bottom 96-well plates. After overnight incubation, cells were treated with escalating drug concentrations and harvested at day 5 post drug treatment.

Cell viability and persister cell numbers

Full-length YAP was amplified by PCR using forward primer 5’- TTTGACCTCCATAGAAGATTCTAGATGGAACAAAAACTCATCTC-3’ and reverse primer 5’-AGCGATCGCAGATCCTTCGCGGCCGCTATAACCATGTAAGAAAGCTTTC-3’ from pQCXIH expression constructs encoding myc-tagged YAP-WT (Addgene #33091), YAP-5SA (Addgene #33093), and YAP-S94A (Addgene #33094), respectively. After XbaI/NotI digestion, PCR products were cloned into the lentiviral pCDH-puro plasmid backbone and correct insertion was verified by Sanger sequencing. For lentivirus production, 293T cells were co-transfected with pCDH-YAP expression plasmids and lentiviral packaging plasmids pCMV-dR8.91 and pMD2.G using the TransIT-LT1 Transfection Reagent (Mirus Bio). Viral supernatant was harvested 72 hours after transfection. Target cells were infected and selected with 1 µg/mL puromycin. YAP overexpression and YAP nuclear localization upon treatment with 0.1% DMSO or 2 µM osimertinib in stable transduced PC9 cells was monitored by Western blot analysis and confocal microscopy.

YAP-5SA overexpression

CRISPR-mediated YAP and FAK knock-out cells were engineered by Synthego (Synthego Corporation, Redwood City, USA). Transient YAP silencing was achieved by knock-down using individual Dharmacon ON TARGETplus YAP1 siRNAs (siYAP#1: J-012200-08, siYAP#2: J-012200-07) compared to non-target control (D-001810-02). Knock-down of focal adhesion kinase signature genes EphB1 (siEPHB1: L-003121-00), FAK (siPTK2B: L-003165-00), and ACK1 (siTNK2: L-003102-01) as well as of LATS1 (L-004632-00) and LATS2 (L-003865-00) was induced using SMARTpool Dharmacon ON-TARGETplus siRNAs compared to non-target control (D-001810-10). Target cells were transiently transfected using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific). SiRNA-mediated knock-down was initiated either at the beginning of treatment with the targeted inhibitor (knock-down during persister cell generation) or when persisters cells were established after 7 days of treatment (knock-down at persister cell state). In both cases, knock-downs were (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2021.10.23.465573; this version posted October 24, 2021. The copyright holder for this preprint 25 repeated every three days and a consecutive number of three knock-downs was performed before harvest. Knock out and knock-down of the protein of interest were verified by Western blot analysis.

Knock-down and CRISPR knock-out

Primary antibodies were coupled to Dynabeads Protein G beads at a 1:5 ratio (antibody:beads, v/v) by constant rotation for 6 hours at 4 °C. Nuclear fractions of persister cells were prepared using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific) and kept on ice. Lysates were quantified using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Equal masses of proteins (300 µg) were added to antibody-coated beads and incubated by constant rotation overnight at 4 °C. Bead-coupled samples were washed and resuspended in 4x Laemmli buffer [277.8 mM Tris-HCl, pH 6.8, 44.4% (v/v) glycerol, 4.4% LDS, 0.02% bromophenol blue, supplemented with 10% (v/v) 2-Mercaptoethanol]. Beads were collected and samples were analyzed by Western blot analysis as outlined above.

Endogenous immunoprecipitation

Cells were seeded in 35mm glass-bottom dishes (MatTek Corporation). Proximity ligation assays were performed using the Duolink In Situ Red Starter Kit Mouse/Rabbit (Millipore Sigma). In brief, cells were washed carefully and fixed in 4% paraformaldehyde. Cells were then permeabilized in 0.1% Triton-X/PBS before blocking. Primary antibodies were added overnight at 4 °C. PLA probes were added for 1 hour at 37 °C before ligation and amplification. After washing and staining cell nuclei with DAPI solution (1:1000 in PBS, stock 1 mg/mL, Thermo Fisher Scientific), cells were imaged at a Yokogawa CSU22 spinning disk confocal microscope using a Plan Apo VC 60X/ 1.4 Oil objective (Nikon Imaging Center, UCSF). Image analysis was done via Fiji ImageJ software58 and PLA signals per nuclei were counted.

PLA assay

Oncological Sciences PhD Candidate. Metastasis research enthusiast. STEMinist. EDI. Science Comm.

Guild Members

Aldo Pourchet

Pavitra Viswanath

Oncology

Medicine

Community Guidelines

Have a question?