Graphical abstract

   graphic file with name fx1.jpg
   [47]Open in a new tab

Highlights

     * •
       An scRNA-seq study reveals shared and distinct features of both
       sides of colon cancer
     * •
       The PS^high malignant epithelia shape the glucose metabolism
       reprogramming niche
     * •
       The IS^high malignant epithelia spatially localize in hypoxic
       regions
     * •
       IS meta-program predicts clinical response to ICI therapy in RCC
     __________________________________________________________________

   Liu et al. reveal shared and distinct features of left- and right-sided
   colon cancers by integrating single-cell transcriptome, spatial
   transcriptome, and large-scale omics analysis. PS^high and IS^high
   malignant epithelia drive distinct immunosuppressive patterns in two
   tumor types, leading to varying clinical outcomes and treatment
   responses in patients.

Introduction

   Colon cancer is a highly heterogeneous disease and a leading cause of
   cancer-related deaths worldwide.[48]^1 It is further categorized into
   left- and right-sided colon cancers based on the primary site, each
   having distinct clinical and molecular characteristics.[49]^2
   Right-sided colon cancer (RCC) is more commonly associated with
   microsatellite instability-high (MSI-H) and KRAS and BRAF mutations
   compared to left-sided colon cancer (LCC).[50]^3 Several studies have
   demonstrated that RCC tends to have poorer overall
   survival.[51]^4^,[52]^5 Tumor laterality also plays a role in the
   effectiveness of targeted therapies, particularly epidermal growth
   factor receptor inhibitors (EGFRis), in metastatic colon cancer. While
   EGFRis can effectively prolong survival in LCC, their efficacy is
   limited in RCC.[53]^6 Previous studies have described various
   histological and genetic differences related to intratumoral
   heterogeneity (ITH) in these two tumor types.[54]^7^,[55]^8^,[56]^9
   However, the associations of these features with clinical outcomes and
   treatment response in LCCs and RCCs are still unclarified.

   ITH is a fundamental property of tumors that is driven by genetics,
   epigenetics, and microenvironmental influences. It is central to
   treatment failure, metastasis, and other cancer phenotypes.[57]^10
   Single-cell RNA sequencing (scRNA-seq) efficiently enables the
   characterization of ITH and has seen a rapid expansion of its use
   across virtually all common cancer types.[58]^11 Recent scRNA-seq
   studies have introduced the concept of ITH “expression meta-programs,”
   which consist of sets of dozens of genes with coordinated variability
   in their expression across malignant cells, to characterize ITH within
   given tumors, such as melanoma and
   craniopharyngioma.[59]^12^,[60]^13^,[61]^14 Importantly, specific cell
   types within the tumor microenvironment (TME) have been found to have a
   causative relationship with cancer cell states. For example, a study
   demonstrated that a subset of cancer cells undergoing partial
   epithelial-to-mesenchymal transition (EMT) located at the leading edge
   of head and neck tumors interacted with cancer-associated fibroblasts,
   thereby facilitating invasion.[62]^15 In colorectal cancer,
   mismatch-repair-deficient and mismatch-repair-proficient tumors
   interacting cellular meta-programs were identified, which revealed the
   logic underlying spatially organized immune-malignant cell
   networks.[63]^16 However, in colon cancer, the phenotypic heterogeneity
   of malignant and non-malignant cells in the TME and their crosstalk
   that contribute to initiate, control, and maintain “sidedness” remain
   unclarified.

   Here, we characterized the TME of two different types of colon cancers
   with integrative analyses, including scRNA-seq, spatial transcriptome
   sequencing (ST-seq), bulk RNA sequencing, and multiplex fluorescence
   immunohistochemical staining (mIHC) analyses. We revealed multicellular
   interaction networks based on co-variation of gene meta-program
   activity in malignant epithelium across left- and right-sided tumors
   and imaged key molecules for predicted cell subsets and meta-programs
   to localize these interaction networks in matched tissues from affected
   individuals. Our study supports the theory that primary tumor
   localization is an option for appropriate therapy in colon cancer
   patients and provides potential therapeutic targets.

Results

A single-cell expression atlas between LCCs and RCCs

   A schematic of the study design and experimental procedures is depicted
   in [64]Figure 1A. The discovery and validation cohorts were explored in
   parallel for cell-type-specific subclustering analysis for reciprocal
   validation ([65]Table S1). Following rigorous quality control and
   filtering, we performed clustering analysis by using an unsupervised
   method and obtained cell subpopulations. According to the
   transcriptomic characteristics of the subpopulations, nine major cell
   types were identified ([66]Figures 1B and [67]S1A‒S1C).
   Cluster-specific marker genes were used to annotate the cell types
   based on previous studies ([68]Figures S2A‒S2D; [69]Table S2).[70]^17
   Global cell-type annotations for each patient with colon cancer were
   categorized according to the tumor side. All these cell types were
   shared among patients and between LCCs and RCCs, albeit at different
   proportions ([71]Figure 1C). In RCC, the relative abundance of myeloid
   cells and T lymphocytes increased in comparison with LCC, while that of
   B lymphocytes and plasma cells was decreased ([72]Figures 1D and
   [73]S2E‒S2H). Seven-plex immunohistochemistry (IHC) staining was
   further conducted to provide an overview of the multicellular
   ecosystems of left- and right-sided colon cancers ([74]Figure 1E).
   These results suggest that RCCs represent a multicellular ecosystem
   distinct from those of LCC.

Figure 1.

   [75]Figure 1
   [76]Open in a new tab

   Single-cell transcriptome atlas of LCCs and RCCs

   (A) Sample collection and analytic process of this study.

   (B) Uniform manifold approximation and projection (UMAP) plot of 93,146
   CD45^− cells and 96,807 CD45^+ cells, colored by cell type or cancer
   type.

   (C) Proportions of the major cell types in normal mucosa and colon
   tumor tissues in individual samples, grouped by left- (LCC) and
   right-sided colon cancer (RCC).

   (D) Heatmaps showing the odds ratios (ORs) of major cell types
   occurring in each tissue. OR > 1.5 indicates distribution preference.

   (E) Multiplex immunohistochemistry (mIHC) staining of EPCAM
   (epithelia), CD68 (myeloid cells), CD3 (T cells), VIM (stromal cells),
   CD79 (B cells), and MZB1 (plasma cells) in left- and right-sided tumor
   tissues (validation cohort 5).

Transcriptomic heterogeneity of malignant epithelium between LCCs and RCCs

   Malignant epithelia identified by inferring large-scale copy number
   variations (CNVs) using immune cells and stromal cells as references