Gramd2+ alveolar type I cells can give rise to Kras-induced lung adenocarcinoma

Lung adenocarcinoma (LUAD) is the most common subtype of lung cancer and arises in the distal lung. LUAD encompasses several pathologic subtypes, including solid, lepidic, papillary, micropapillary, acinar and mixed subtypes, each with differing clinical outcomes or biological behavior. Specifically, lepidic histology is associated with improved survival relative to other LUAD histologies. Yet, the molecular and cellular underpinnings of these subtypes are largely unknown. Understanding which cell populations in the distal lung contribute to LUAD could provide insights for the marked heterogeneity in LUAD pathologic features, clinical presentation and responses to therapy.Recent studies have seen tumor-derived epithelial cells with an AT1 transcriptomic cell signature, suggesting AT1 cells may contribute to a subset of LUAD cases. We tested the ability of AT1 cells to give rise to LUAD by inducing KrasG12D, a known oncogenic driver in human LUAD. Activation of KrasG12D in Gram-domain containing 2 (Gramd2)+ AT1 cells gave rise to multiple LUAD lesions, primarily of papillary histology. In contrast, activation of KrasG12D in Sftpc+ AT2 cells resulted in LUAD lesions of lepidic histology. Immunohistochemistry established Gramd2:KrasG12D lesions were of primary lung origin and not metastatic events. Spatial transcriptomic profiling revealed distinct pathway alterations occurring within Gramd2- and Sftpc-derived LUAD. Immunofluorescence confirmed differences observed in the Spatial transcriptomic analysis in expression patterns and distribution of cell-specific markers between cell of origin, while universal upregulation of the Krt8 intermediate cell state marker was observed. Our results are consistent with Gramd2+ AT1 cells serving as a putative cell of origin for LUAD and suggest that LUAD may be a collection of adenocarcinomas that share a common location within the distal lung, but which arise from different cells of origin. Spatial transcriptomic profiling was performed at the Molecular Genomics Core, a part of the Norris Comprehensive Cancer Center and under manufacturer’s instructions. Briefly, three biological replicates (one block per mouse) of Gramd2:KrasG12D and Sftpc:KrasG12D underwent initial 10μM sectioning at the Tissue Pathology Core, part of the Norris Comprehensive Cancer Center. Sections were H&E stained (as above) and underwent pathology review. Regions of Interest (ROIs) then underwent sample preparation including test slide sample sequencing using the Visium Tissue Section Test Slide (PN-2000460, 10X Genomics, Dublin, CA, USA). H&E overlapping spatial transcriptomic regions was generated as part of the Cytassist pipeline. RNA quality assessment of test slides was determined by first extracting RNA using the Qiagen RNeasy FFPE kit (#73504, Qiagen, Hilden, Germany) and measuring RNA concentration using a Qubit fluorometer (#Q33238, ThermoFisher Scientific, Waltham, MA, USA). To determine The percentage of total fragments >200nt in length was determined by running the samples using a 4200 Tapestation (#G2991AA, Agilent, Santa Clara, CA). Once samples were determined to be of high enough quality to continue, 5μM FFPE sections were then placed in a 42C water bath and once rehydrated adhered to the Visium Spatial Gene Expression Slide (PN-2000233, 10X Genomics). Subsequently, samples were dried at 42C for 3hrs in a desiccation chamber and underwent deparrafinization using Qiagen Deparaffinization Solution (Cat # 19093) at 60C for 2 hr. Subsequently H&E staining (#MHS16, #HT110116 Millipore Sigma, Burlington MA, USA) was performed according to Visium technical parameters (CG000409) and images were captured with a Zeiss Axioscan2 microscope using a 10x objective. Decrosslinking was performed according to 10X standard protocol (CG000407) and immediately hybridized to the Visium Mouse Transcriptome probe set V1.0, which contained 20,551 genes targeted by 20,873 probes. Post-probe extension, sequencing library construction was performed using unique sample indices using the Dual Index Kit TS, Set A (PN-1000251) for Illumina-compatible sequencing. Paired-end sequencing (2x100) was performed on an Illumina instrument. Sequencing data was pre-processed using the Space Ranger pipelines (10x Genomics, https://cloud.10xgenomics.com/cloud-analysis). BCL data was demultiplexed and converted to FASTQ format using the spaceranger mkfastq pipeline. The spaceranger count pipeline was then used for read alignment to mouse genome reference mm10, Unique Molecular Identifier (UMI) counting, and generation of feature-spot matrices corresponding to the microscopic tissue image. The pipeline also performs automatic tissue detection and fiducial alignment from the image.