How can we specifically target cancer cells in colorectal cancer? And how do we sort these cancer cells from all the background cells in a biopsy, tissue sample, or single-cell dataset? One common approach is to find reliable and unique tumor markers for colorectal cancer cells.
Tumor markers and/or tumor-specific antigens are a big deal in cancer research. There are plenty of tumor markers for colorectal cancer (CRC) being suggested lately, but the real challenge is validating them. We need to know if the overexpression is specific enough, if these markers are translated into functional proteins, if those proteins are actually expressed on the cell surface, and whether the overexpression shows up consistently across different studies and datasets.
In this mini project, we’ve rounded up some of the most promising tumor markers for CRC and put them to the cross-omics test using our C-DIAM Multi-Omics Studio. So, which ones came out on top? Let’s dive in and find out!
Why do we need a list of reliable tumor markers for colorectal cancer?
We need reliable tumor markers for CRC because they play a critical role in improving cancer research, detection and treatment. In the lab, these markers help scientists pinpoint cancer populations more easily, like quickly identifying the cancer/malignant cells in a single-cell dataset for comparison with other cell types. Clinically, tumor markers can make early detection more accurate, catching the cancer when it’s easier to treat. They’re also super useful for predicting how aggressive a cancer might be, which helps doctors create personalized treatment plans. On top of that, tumor markers can be good candidates for modern therapies like antibody-drug conjugates (ADCs) and bispecific antibodies (BsAbs).
What are some tumor markers for CRC that have been suggested?
Several genes have been proposed as potential tumor markers for CRC. Here are a few that keep popping up in studies:
ERBB2, FGFR1, CD24, CEACAM family, EphA2
These genes are often amplified in CRC, resulting in overexpression of their corresponding surface proteins, involved in processes like cell signaling, adhesion, and cancer progression. ERBB2 (also known as HER2) is a well-known, less heterogeneity oncogene and an emerging diagnosis protein in CRC and has been linked to poor prognosis (Venturini et al., 2024). FGFR1 (Fibroblast Growth Factor Receptor 1) oncogene plays roles in driving cellular proliferation (Zhao et al., 2024). CD24 is a biomarker for cancer stem cell (CSC), which is the tumor-initiating subset of cells, significantly contributes to the formation and promotion of tumor cells’ aggressiveness (Yang et al., 2023 ), while the CEACAM family plays a role in cell adhesion and tumor metastasis (Thomas et al., 2023). EphA2 upregulation in CRC influences the uncontrolled proliferation, supporting tumor’s aggressiveness as well as migration by decreasing the EMT attachments (Xiao et al., 2020) making it an attractive target for CRC therapies.
MYC, HNF4A, ASCL2, LCN2, ID1, KRAS, HDACs Family, GPRC5A, TM4SF1
These genes are oncogenes or upregulated in malignant CRC cells, though they do not directly encode cell surface receptors. For example, c-MYC oncogenes, which is reportedly upregulated in 70%-80% of CRC cases, is a master regulatory gene for cell proliferation and cell cycle arrest process (Sipos et al., 2016).The dysregulated expression of HNF4A is recognized as a key factor influencing cancer cell proliferation, apoptosis, invasion, dedifferentiation, and metastasis (Lv et al., 2021), while ASCL2 is a crucial marker of colorectal cancer stem cells and plays a vital role in preserving the characteristics of colon cancer stem or precursor cells (Wang et al., 2021). LCN2 is linked to inflammation, cell’s proliferation coupling with metastasis (Kim et al., 2017), and KRAS mutations are common in CRC, driving uncontrolled cell growth (Jasmine et al., 2024). GPRC5A (Zhang et al., 2017), and TM4SF1 (Tang et al., 2020) contribute to tumor growth and migration in general.
CCND2, CDK8
These cell cycle-related genes are upregulated in CRC but are not strictly cancer-specific. CCND2 (Cyclin D2) which is a downstream member of the JAK2/STAT3/CCND2 axis, takes part in disrupting the integrity of cell cycle checkpoint at the G1/S phase by blocking the tumor suppressor retinoblastoma protein (Park et al., 2019) to allow the mutated cancer stem cells sustain their survival. CDK8 (Cyclin-Dependent Kinase 8) plays dual effects on tumorigenesis: regulates cell cycle progression and impacts several oncogenic pathways (Liang et al., 2018). These genes are promising as supportive biomarkers or therapeutic targets when combined with other markers.
NPDC1, TGFBI, TACSTD2, HES6, TSPAN13, REG4, TFF3, SERPINA1, CRIP2
These genes promote overproliferation, epithelial-mesenchymal transition (EMT), and a pro-inflammatory environment along the transformation of normal epithelial cells into adenomatous polyps and, eventually, aggressive tumors. TACSTD2 overexpression has been suggested as a reliable marker for the onset of precancerous lesions, enhancing epithelial cell proliferation, tumorigenesis, and sustaining an inflammatory tumor microenvironment via the Wnt/β-catenin pathway (Siskova et al., preprint, 2024). REG4 greatly modulates the adenoma transformation process not only by enhancing the proliferation and tumorigenesis of colonic epithelial cells via various pathways (Zheng et al., 2022) but also creating the sustainable inflamed tumor microenvironment via Wnt/β-catenin pathway (Hwang et al., 2020). SERPINA1 mediates CEBPB-induced promotion of neoplastic cell growth and migration via STAT3 signaling (Ma et al., 2024) and is considered a good biomarker to distinguish early adenoma polyps from normal tissue. TFF3 facilitates the-adenoma-to-carcinoma progression by inhibiting apoptosis, promoting proliferation, angiogenesis, and enabling neoplastic invasiveness (Yang et al., 2022).
What CRC tumor markers win our cross-omics test?
To validate these tumor markers, we used C-DIAM Multi-Omics Studio for meta-analysis across six datasets:
Dataset | Original study | Data type | Sources |
PDC000116 | CPTAC | Proteomics | C-DIAM public database |
GSE178341 | Pelka et al., 2021 | Single-cell RNA-seq | C-DIAM public database |
GSE166555 | Uhlitz et al., 2021 | Single-cell RNA-seq | C-DIAM public database |
EMTABB107 | Qian et al., 2020 | Single-cell RNA-seq | C-DIAM public database |
GSE146771 | Zhang et al., 2020 | Single-cell RNA-seq | C-DIAM public database |
TCGA_COAD | TCGA | Bulk RNA-seq | C-DIAM public database |
Datasets used for our cross-omics validation
Using C-DIAM MultipleDE pipeline, we summarized the DE genes from different studies and omics - comparing malignant epithelial cells against normal epithelial cells in single-cell datasets, and primary tumor tissue versus normal solid tissue in bulk RNA and proteomics datasets.
What were upregulated in at least three datasets? We’ve got a list: CEACAM6, EPHA2, LCN2, GPRC5A, TACSTD2, MYC, ASCL2, TM4SF1, HDAC2, HDAC3, CCND2, CDK8, NPDC1, TGFBI, HES6, REG4, and TFF3. Noticeably, the first five genes were up-regulated in the proteomics dataset (with FDR ≤ 0.05 and log2FC ≥ 0.585).
A summary of our validation steps in C-DIAM Multi-omics Studio. (A) Validating candidate markers across omics using MultipleDE pipeline; (B) Checking candidate markers in healthy tissue in GTEx; and (C) Validating marker association with morphology using VisiumHD CRC data (data source: 10x Genomics)
Validating short-listed markers in healthy colon tissue
We then validated the expression of these genes in healthy colon tissue using a GTEx dataset. This is important because markers with high expression in healthy tissue might not be ideal for distinguishing cancer cells from the surrounding background. Most genes had low expression in healthy tissue as expected, except for TFF3, which showed some outliers.
Validating short-listed markers on Visium HD
To further confirm the tumor relevance of these markers, we revisited the CRC Visium HD dataset by 10x Genomics. This allowed us to associate gene expression with morphological features and explore what genes can differentiate the tumor regions. Here’s what we’ve got:
Strong differentiation: CEACAM6, GPRC5A, MYC, HDAC2
Moderate differentiation: ASCL2, TACSTD2/ TROP2, TM4SF1 (tends to be at the boundary of tumor regions), CCND2, EPHA2
Low differentiation: LCN2, REG4, TGFBI, HES6, HDAC3, CDK8, NPDC1
Validating candidate tumor markers on Visium HD CRC data using C-DIAM Multi-omics Studio. (A) Tumor regions (red) were labeled based on morphological features and following Oliveira et al., 2024 (preprint), (B),(C),(D),(E),(F) Expression of CEACAM6, GPRC5A, MYC, TACSTD2, TGFBI.
Final verdict?
We were able to have a list of wins! See our full results here in this Application report.
Conclusion
This mini project shows how cross-omics validation can help pinpoint the best tumor markers for colorectal cancer. Using the C-DIAM Multi-Omics Studio, we integrated data from different omics layers, validated it across datasets, and discovered markers that could potentially detect colorectal cancer cells. While more research is needed to confirm their reliability, it’s exciting to see how multi-omics analysis and data-driven approaches are advancing cancer research!
Kommentare