Abstract
1 Introduction
Accurate in vitro models of the patient specific tumor immune microenvironment (TIME) and robust analysis techniques enabling in-depth dissection of these models are indispensable for the development of next generation immunotherapy. Imaging-based and sequencing-based analysis of the surgically dissected patient tissues has enabled the deconvolution of not only the cell types that constitute the TIME but also their cell-state heterogeneity.[1-4] However, they can only provide the snapshots rather than the dynamics of TIME, i.e., the subtle changes of TIME in response to treatments. Simplified in vitro models representing the in vivo tumor micro-niche, retaining the local TIME, and facilitating the modulation of drug treatment are desired. Patient-derived organoids (PDOs) have been widely accepted as a valuable in vitro tumor model that faithfully recapitulates the histological and molecular characteristics of original tumors.[5-9] Depending on the methods for organoid generation and the period of in vitro culture, tumor organoids can preserve the immune components of the parental tissues.[10, 11] For example, the PDOs established using mechanical crushing methods and cultured either in a gel matrix[12] or an air-liquid interface[13] contained tumor infiltrating lymphocytes and responded to immune checkpoint blockade (ICB).[12, 13] However, whether they recapitulate the inter- and intra- patient heterogeneity of local TIME remains to be proved. More importantly, limited by the number of cells, current analysis methods for PDOs are mainly phenotypic- or imaging-based, impeding the fully dissection and wide application of PDOs as a model of patient TIME.
Single-cell RNA sequencing (scRNA-seq) is an indispensable tool to dissect the cellular diversity within a complex biological system. However, current sequencing-based techniques are not applicable for deciphering the TIME in an organoid owing to the disruption of the spatial distribution of immune cells. For example, the commonly used scRNA-seq techniques, such as 10x Chromium, stochastically capture and encode single cells, leading to the requirement of a minimum of 104–105 starting cells with a considerable cell loss.[14] The efficient processing of every single cell from each individual organoid remains challenging. Although ample single cells can be digested from PDOs and pooled for processing, the link of single cells with their specific microenvironment within individual PDO will be disrupted (i.e., immune cells in individual organoids are not traceable). Actually, the retrospection of single cells to their parental organoids is extremely important, as it would enable the single-cell data analysis under the supervision of phenotypic responses of organoids.
In order to probe the immune microenvironment in local tumor micro-niches, we established primary lung cancer organoids (pLCOs) from dissected tumor tissues of lung cancer patients. Owing to the nature of the mechanical processing methods, clusters of tumor cells were isolated and the infiltrating immune cells were retained. We proved that the pLCOs represent the general and patient-specific features of local TIME in tumor parenchyma. To perform single-cell analysis on individual pLCOs, we developed the function-associated scRNA-seq (FascRNA-seq) platform, with which the phenotypic changes of pLCOs under ICB treatment were evaluated, and in the meanwhile, the transcriptomes of all the cells from each pLCO were sequenced and correlated to the phenotypic data (Figure 1A). Subsequent bioinformatic analysis was conducted with phenotype supervision, which certified that the anti-PD-1 (αPD-1) induced cytotoxic effect was mediated by CD8+ T cells with patient-specific activation patterns. Our methods can be used for characterizing the local tumor micro-niche, identifying the tumor-reactive T cells, and potentially facilitating the development of novel immunotherapy strategies.