The Single-Cell Challenge Cancer is not a uniform mass of identical cells. Within a single tumor, different cells carry different mutations, different epigenetic states, and different patterns of gene expression. Understanding how these three layers interact within individual cells - not just averaged across millions - requires fundamentally new technology.
scTrio-seq Technology Single-cell triple omics sequencing (scTrio-seq) was developed to simultaneously capture the copy number variation (CNV) landscape, the DNA methylation state, and the transcriptome from the same individual cell. This co-measurement eliminates the ambiguity that arises when comparing omics layers from different cells in the same tumor.
Application to Liver Cancer The technique was applied to 25 hepatocellular carcinoma (HCC) cells isolated from a patient's tumor. This allowed the researchers to ask how genetic diversity (different CNV profiles), epigenetic diversity (different methylation patterns), and transcriptomic diversity (different gene expression patterns) relate to each other within an individual cancer.
Why HCC Hepatocellular carcinoma is the most common primary liver cancer and arises in a background of chronic inflammation and fibrosis. Tumor heterogeneity in HCC is clinically relevant because it influences treatment resistance and metastatic potential. Single-cell multi-omics can reveal the subclonal architecture underlying this heterogeneity.
Cell Isolation and Lysis Individual HCC tumor cells were isolated by fluorescence-activated cell sorting and lysed under conditions that preserve both nucleic acids. The protocol uses a modified single-cell whole-genome amplification approach that retains allele-specific information for both genomic DNA and bisulfite-converted methylated DNA.
Simultaneous Capture After cell lysis, the genomic DNA fraction is split: one portion undergoes bisulfite conversion for methylation analysis while another portion undergoes regular amplification for CNV analysis. mRNA is captured using oligo-dT primers before the DNA is processed, ensuring that all three molecular layers come from the same cell.
Bioinformatic Integration Aligned sequencing reads from the same cell were integrated computationally to generate a per-cell profile combining CNV calls, CpG methylation levels at hundreds of thousands of sites, and gene expression quantification. This three-dimensional profile enabled simultaneous clustering by any or all omics layers.
Validation Approach The scTrio-seq data were validated by comparing bulk tumor profiles generated from the same patient tissue to the aggregate of single-cell profiles, confirming that the single-cell approach accurately captured the dominant features of the tumor without systematic biases introduced by the amplification steps.
Subpopulation Discovery Unsupervised clustering of the 25 HCC cells using CNV profiles identified two distinct subpopulations - subpopulation I (approximately 15 cells) and subpopulation II (approximately 10 cells) - with different patterns of chromosomal gains and losses. The same two subpopulations were recovered when clustering was performed on methylation profiles alone, confirming that genetic and epigenetic heterogeneity are concordant.
Transcriptomic Confirmation When the same cells were clustered using transcriptomic data, the two subpopulations identified by CNV and methylation were again recapitulated, demonstrating that all three molecular layers carry concordant information about subclonal identity within this tumor.
CNV-Methylation Independence While CNV and methylation profiles both separated the two subpopulations, the correlation between a specific gene's copy number change and its CpG methylation level was weak. This suggests that somatic copy number alterations do not directly remodel the local DNA methylation landscape - the two layers are epigenetically independent.
CNV-Expression Coupling In contrast, CNV changes showed a proportional relationship with gene expression in the same cell - regions with copy number gain tended to have higher transcript levels, and regions with copy number loss tended to have lower expression. This coupling was consistent across both subpopulations.
Invasion-Associated Genes Transcriptomic analysis showed that subpopulation I cells express higher levels of genes associated with epithelial-to-mesenchymal transition and cell migration. These include upregulation of matrix metalloproteinases and downregulation of cell-cell adhesion molecules, consistent with a more invasive phenotype.
Complement and Coagulation Pathway Suppression One of the most striking transcriptomic differences is the downregulation in subpopulation I of genes in the complement and coagulation cascades. Many of these genes - including components of the classical complement pathway - are normally highly expressed in hepatocytes, and their suppression in cancer cells may help tumors evade complement-mediated immune surveillance.
Immune Evasion Implications Loss of complement pathway gene expression in subpopulation I could reduce the ability of the immune system to tag these cells for destruction. This epigenetically distinct subpopulation may therefore be more likely to survive in the host, metastasize, and resist immunotherapy.
Methylation of Complement Genes The reduced expression of complement genes in subpopulation I was associated with higher promoter methylation at these loci, providing a direct link between the epigenetic divergence of the subpopulations and the transcriptomic differences. This demonstrates how DNA methylation can silence immune-surveillance pathways in cancer.
Resolving Cause and Effect When separate cells are profiled for different molecular layers, observed correlations between, say, methylation and expression may reflect different cell types rather than true mechanistic connections. scTrio-seq eliminates this ambiguity by measuring all three layers in the same cell, enabling causal inferences within individual cells.
Independence of CNV and Methylation The finding that copy number changes do not directly remodel DNA methylation at the same loci challenges a simple model in which genomic damage cascades directly into epigenomic change. The two layers evolve at least partially independently, implying that epigenetic reprogramming in cancer is not merely a passive consequence of genomic instability.
Subclonal Structure and Treatment Understanding that a tumor contains subpopulations with distinct epigenetic and invasive properties has direct clinical implications. Treating only the dominant clone may leave behind resistant subpopulations. Single-cell multi-omics can identify which subpopulation is most likely to drive recurrence.
Scalability The 25-cell dataset analyzed here is sufficient to establish proof of concept, but larger cohorts of cells and patients will be needed to generalize these findings. Future technological improvements reducing the cost and increasing throughput of scTrio-seq will make population-scale multi-omics studies feasible.
Expanding to More Patients Profiling single cells from multiple HCC tumors representing different etiologies (HBV, HCV, alcohol, NAFLD) and stages would reveal whether the subpopulation structure and omics-layer relationships observed here are universal features of HCC or specific to this individual tumor.
Adding Chromatin Accessibility Future extensions of the triple-omics approach could incorporate ATAC-seq data to measure chromatin accessibility as a fourth layer, providing information about transcription factor binding and regulatory element activity in the same cell alongside CNV, methylation, and expression.
Lineage Tracing Combining scTrio-seq with CRISPR-based lineage recording systems could allow researchers to track which cells give rise to which subpopulations over time, connecting the static multi-omics snapshot to dynamic clonal evolution.
Clinical Translation If distinct methylation or CNV profiles in single cells correlate with clinical outcomes such as metastasis or therapeutic response, scTrio-seq on small biopsy samples could eventually guide personalized treatment decisions by identifying the dominant subclonal architecture at diagnosis.