Single‐cell multi‐omics reveals tumor microenvironment factors underlying poor immunotherapy responses in ALK‐positive lung cancer

Lung cancer remains the leading cause of cancer death in 2024, with ∼80% being non-small cell lung cancer (NSCLC). Anaplastic lymphoma kinase (ALK) rearrangements occur in ∼5% of NSCLC cases, typically treated with ALK inhibitors, though resistance often develops [1]. Immunotherapy has been explored for advanced or resistant ALK-positive NSCLC, but immune checkpoint blockade (ICB) treatments have shown limited clinical benefits [1]. A comprehensive study of ALK-positive NSCLC tumor microenvironment (TME) is needed to understand immunotherapy limitations and improve treatment strategies. We generated and collected single-cell RNA sequencing (scRNA-seq) and single-cell Assay for Transposase Accessible Chromatin with high-throughput sequencing (scATAC-seq) datasets from lung adenocarcinoma (LUAD) patients with ALK rearrangements and wild type without major oncogenic drivers (WT) (Supplementary Table S1). By comparing TME, we aimed to identify features explaining poor immunotherapy responses (Figure 1A). After batch corrections for each of the RNA and ATAC profiles, we identified epithelial, stromal, and immune cells (Supplementary Figure S1A, Supplementary Table S2). From the immune compartment, we identified major cell types such as myeloid cells, T cells, natural killer (NK) cells, and B cells (Figure 1B, Supplementary Figure S1B, Supplementary Table S3). Integration of the RNA and ATAC profiles confirmed consistency between these two omics profiles (Supplementary Figure S1C-D). Compositional analysis revealed that ALK-positive samples showed an enrichment of innate immune cells (myeloid and NK cells) and depletion of adaptive immune cells (T and B cells) (Figure 1C, Supplementary Figure S1E-F), suggesting weak adaptive anti-tumoral responses in ALK-positive TME. Oncogenic mutations primarily impact epithelial cells, so we classified epithelial cell malignancy with the cell-classifier and copy number variations (Supplementary Figure S1G-H) and measured ALK and PD-L1 expression. Malignant cells expressing ALK or PD-L1 were predominantly from ALK-positive malignant cells (Supplementary Figure S1I). We observed that ALK-positive tumors had a higher malignant-to-normal epithelial cell ratio than WT tumors (Figure 1D). Additionally, malignant cells in ALK-positive tumors exhibited increased stemness, indicating greater developmental potential (Figure 1E). Top 100 upregulated genes in ALK-positive tumors were linked to aggressive cancer pathways like epithelial-mesenchymal transition (EMT) and hypoxia [2] (Figure 1F). To identify malignant subsets with higher progression potential, we constructed developmental trajectories (Supplementary Figure S1J). State 4, enriched with ALK-positive malignant cells, showed higher stemness and elevated stress, hypoxia, and EMT gene signatures (Supplementary Figure S1J-L, Supplementary Table S4). Survival analysis with LUAD patients in The Cancer Genome Atlas (TCGA) [3] showed worse outcomes for patients with higher expression of State 4 signature genes, while differentially expressed genes (DEGs) between ALK-positive and WT malignant cells did not predict survival (Supplementary Figure S1M). This suggested that ALK rearrangement leads to more aggressive and metastatic tumors and emergence of clinically unfavorable malignant sub-states. Cell-cell interaction analysis showed ALK-positive malignant cells had increased interactions with myeloid cells (Figure 1G). Among these myeloid cell subtypes, ALK-positive tumors had significantly higher proportions of tumor-associated macrophages (TAMs) (Supplementary Figure S2A). ALK-positive TAMs were linked to M2-like pro-tumoral functions like adipogenesis and lipid metabolism, while WT TAMs were associated with M1-like immune-related functions, including TNF-alpha signaling and interferon gamma response (Figure 1H). Sub-clustering of TAMs (Supplementary Figure S2B-C) and RNA velocity analysis revealed WT TAMs transitioned into anti-tumoral, inflammation-related states, whereas ALK-positive TAMs transitioned into pro-tumoral, metabolic states (Figure 1I-J, Supplementary Figure S2D), maintaining immunosuppressive TME in ALK-positive tumors. Depleted adaptive immunity in ALK-positive tumors suggested reduced anti-tumoral immune functions [4]. B cell analysis (Supplementary Figure S2E-F) revealed that memory B cells and plasma cells were more abundant in WT tumors (Figure 1K). Top DEGs of WT memory B cells were enriched in oxidative phosphorylation pathways, indicating functional activation and effector capabilities [5] (Supplementary Figure S2G). Reconstructed B cell receptor sequences showed a lower ratio of expanded to non-expanded memory B cells in ALK-positive tumors (Figure 1L), suggesting reduced humoral anti-tumoral responses and activation. B cells are key components of tertiary lymphoid structures (TLSs), vital for immunotherapy responses [6]. We found positive correlations between B cells and CD4+ T cells, with WT tumors having more B cells (Figure 1M). ALK-positive tumors showed reduced interactions among major components of TLS, suggesting fewer TLS-like structures (Figure 1N), contributing to inadequate anti-tumoral responses and unfavorable TME for immunotherapy. Further experimental validations would be beneficial to confirm the existence of TLS-like structures. For CD8+ T cells, crucial in ICB treatment, we identified subtypes using marker genes (Supplementary Figure S2H-I). DEG analysis showed tumor-reactive and exhaustion genes (CXCL13, ENTPD1, ITGAE) were enriched in WT tumors, while bystander-indicating NK receptors (KLRK1, KLRC2) [7] were enriched in ALK-positive tumors (Figure 1O). T cell signature analysis confirmed these findings (Figure 1P, Supplementary Table S4). This suggested that ALK-positive "exhausted" CD8+ T cells became dysfunctional through non-canonical mechanisms, reducing their effectiveness during ICB treatment. Single-cell T cell receptor (TCR) sequencing showed fewer expanded ALK-positive CD8+ T cells with smaller TCR clonal sizes (Supplementary Figure S2J-K), indicating reduced T cell antigenicity. DEG analysis between expanded and non-expanded effector and exhausted CD8+ T cells revealed WT tumors had increased expression of exhaustion and tumor-reactivity genes, while the ALK-positive tumors showed decreased expression of these genes and increased NK receptor expression (Figure 1Q). This reaffirmed impaired tumor reactivity and more bystander-like CD8+ T cells in ALK-positive tumors. We analyzed gene regulatory changes in exhausted CD8+ T cell networks due to ALK rearrangement (Supplementary Figure S2L-M). In the WT network, T cell activation and exhaustion-related genes emerged as hubs with IFNG as a top hub gene, consistent with its role in effector T cell stimulation and reported deficiency in ALK-positive patients [1] (Supplementary Figure S2N). NK receptors like KLRK1 had high centrality in the ALK-positive network. We re-evaluated hub genes based on neighboring gene expression signatures. IFNG signature score was higher in exhausted T cells of WT tumors, while KLRK1 signature score was higher in exhausted T cells of ALK-positive tumors (Figure 1R). Since T cell exhaustion is also epigenetically regulated, we identified CD8+ T cell subtypes using scATAC-seq profiles and marker gene scores (Supplementary Figure S2O-P). Although transcription of exhaustion and tumor-reactivity genes was downregulated in ALK-positive CD8+ T cells, no clear depletion in chromatin accessibility was observed (Supplementary Figure S2Q), suggesting transcription factor (TF) binding caused these differences. Motif analysis showed BATF-related TFs, linked to CD8+ T cell effector functions [8, 9], were enriched in WT, while ETS1, important for NK cell differentiation [10], was enriched in ALK-positive cells (Figure 1S). We hypothesized that BATF is a key TF contributing to differences in T cell dysfunction. Using DEGs in CD8+ T cells between WT and ALK-positive tumors (see Figure 1O), we found WT DEGs had stronger BATF motif enrichment (Figure 1T) and all exhaustion/tumor-reactivity genes, except KRT86, had at least one peak with BATF motifs (Figure 1U). Interestingly, KRT86, the only exhaustion-related gene upregulated in expanded cells from both groups (see Figure 1Q), lacked BATF motifs. These findings suggested that differences in TF activities could lead to differences in gene expression. Additionally, WT exhausted T cells had more nearby peaks containing BATF motifs with higher co-accessibility, suggesting stronger co-regulation with nearby genomic regions in WT compared to the ALK-positive tumors (Figure 1V). In summary, this study provides key insights into the TME of ALK-positive lung cancer through single-cell multi-omics analysis, shedding light on poor immunotherapy responses. ALK-positive tumors displayed aggressive malignant phenotypes, enriched pro-tumoral myeloid cells, and depleted adaptive T and B cells. ALK-positive TAMs shifted away from inflammatory states, while CD8+ T cells showed reduced tumor reactivity and more bystander traits. We suggest that these CD8+ T cells may have distinct dysfunction mechanisms. Epigenetic analysis revealed depleted BATF motifs in ALK-positive CD8+ T cells, indicating altered TF activity. These findings highlight how ALK rearrangements drive an immunosuppressive TME, hindering effective immune responses, and suggest a need for strategies to reinvigorate adaptive immunity in ALK-positive lung cancer. Hye Ryun Kim, Seong Yong Park, and Insuk Lee conceived the study. Seungbyn Baek performed single-cell multi-omics data analysis under the supervision of Insuk Lee. Euijeong Sung assisted bioinformatic analysis. Gamin Kim contributed sample preparation. Hye Ryun Kim and Seong Yong Park organized clinical sample and data collections. Min Hee Hong and Chang Young Lee contributed to clinical sample collection. Hyo Sup Shim contributed to the pathological examination of tumor tissues. Insuk Lee and Hye Ryun Kim contributed to the financial and administrative support for this study. Seungbyn Baek, Hye Ryun Kim, and Insuk Lee wrote the manuscript. All authors read and approved the final manuscript. Not applicable. The authors declare that they have no conflicts of interest. This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation funded by the Ministry of Science and ICT (2021R1A2C2094629 and 2017M3A9E9072669 to Hye Ryun Kim, and 2018R1A5A2025079, 2022M3A9F3016364, and 2022R1A2C1092062 to Insuk Lee). The work was supported in part by Brain Korea 21 (BK21) FOUR program. This work was supported by the Technology Innovation Program (20022947) funded by the Ministry of Trade Industry & Energy (MOTIE, Korea). This work was supported by the Yonsei Fellow Program, funded by Lee Youn Jae. The studies were approved by the Institutional Review Board of Yonsei University Severance Hospital with IRB No 4-2018-1161-1. Written informed consent was obtained prior to enrollment and sample collection at Yonsei University Severance Hospital. The research conformed to the principles of the Helsinki Declaration. The single-cell RNA, TCR, and ATAC sequencing data generated in this study are deposited in the Gene Expression Omnibus database with accession number GSE274934 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?&acc=GSE274934). The data will be available upon publication. The remaining data are available within the Article or Supplementary Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.