Introduction: The colorectal serrated pathway involves precursor lesions known as sessile serrated lesions (SSL) and traditional serrated adenomas (TSA). Mutations in BRAF or KRAS are crucial early events in this pathway. Additional genetic and epigenetic changes contribute to the progression of these lesions into high-grade lesions and, eventually, invasive carcinoma. Methods: We employed digital spatial profiling to investigate the transcriptional changes associated with SSL and TSA. The genes identified are confirmed by immunohistochemical (IHC) staining. Colorectal cancer (CRC) cell lines with CEACAM6 overexpression and knockdown were established to study the roles of CEACAM6 on tumorigenesis of CRC. Results: Ten genes were upregulated in SSL and TSA, and seven were upregulated in both types of lesions. IHC staining confirmed overexpression of CEACAM6, LCN2, KRT19, and lysozyme in SSL and TSA. CEACAM6 expression is an early event in the serrated pathway but a late event in the conventional pathway. Using cell line models, we confirmed that CEACAM6 promotes CRC cells’ proliferation, migration, and invasion abilities. Conclusion: These results highlight that the transcriptional changes in the early stages of tumorigenesis exhibit relative uniformity. Identifying these early events may hold significant promise in elucidating the mechanisms behind tumor initiation.

Colorectal cancer (CRC) stands as one of the most prevalent malignancies globally, ranking fourth and third in terms of both male cancer-related deaths and incidences and third and second among female cancer cases [1]. The pathogenesis of CRC arises from a progressive accumulation of genetic and epigenetic alterations. The development of CRC comprises various pathways, influenced by etiology, driver gene mutations, and the sequence of mutations. These diverse pathways lead to distinct subtypes of CRC, each characterized by unique clinical features, prognosis, and responses to treatment. The two most prevalent pathways in colorectal carcinogenesis are the conventional and serrated pathways.

The conventional pathway, also known as the adenoma-carcinoma sequence, is primarily characterized by adenomas, such as tubular or tubulovillous adenomas, serving as precursor lesions. This pathway is the most common route to CRC [2]. The initial step involves biallelic APC mutations, which activate the Wnt pathway. Subsequent mutations in KRAS, CDKN2A, TP53, and chromosomal instability lead to the progression from adenoma to invasive adenocarcinoma [3].

Around 15–30% of CRC cases originate from serrated lesions, identifiable by their “serrated” appearance in the glandular crypts of precursor lesions [4]. In the serrated pathway, the initial step involves the activation of the mitogen-activated protein kinase (MAPK) pathway through mutations in KRAS or BRAF [4]. The precursor lesions in the serrated pathway include hyperplastic polyp (HP), sessile serrated lesion (SSL), and traditional serrated adenoma (TSA) [5]. HPs are typically small and benign findings, more commonly seen in the distal colon. SSLs, on the other hand, are predominantly found in the proximal colon and tend to be larger. Molecularly, both HPs and SSLs harbor BRAF mutations [6]. While SSLs may appear benign morphologically, some progress to sessile serrated adenomas with dysplasia [7, 8]. The crucial driving mechanism behind this progression is the methylation of the MLH1 promoter, leading to MLH1 loss and DNA mismatch repair deficiency, resulting in a high mutation rate and, ultimately, the development of CRC.

Another less common serrated lesion is TSA. Grouped initially with SSL under the category of “serrated adenoma” in the 1990s, recent evidence has highlighted distinct molecular and phenotypic characteristics that set TSA apart from SSLs [9‒11]. Although most TSA cases harbor BRAF mutations, similar to SSL, a third of cases exhibit KRAS mutations [9, 10, 12]. TSA with BRAF mutations often contains foci of SSL, suggesting a potential origin from SSL [9]. In addition to MAPK pathway activation, TSA is characterized by Wnt pathway activation, primarily driven by inactivating mutations in RNF43 or RSPO2/3 fusions [12, 13]. The overexpression of RSPO3 coincides with the transition from SSL to TSA [14], highlighting Wnt pathway activation as the driving force behind the progression from SSL to TSA.

Identifying early events in tumorigenesis plays a crucial role in understanding cancer pathogenesis and developing preventive strategies. While early genetic events in serrated lesions have been identified, changes in gene expression levels remain largely unexplored. Challenges such as the small size of the lesions, the scarcity of fresh frozen tissue specimens, and the presence of mixed tumor and non-tumor cells have hindered in-depth transcriptome analysis of early tumorous lesions. To address this, we applied Digital Spatial Profiling (DSP), a novel technique platform capable of high-level multiplexing signal detection with precise spatial and temporal resolution in formalin-fixed paraffin-embedded (FFPE) tissue [15]. The DSP assay relies on antibodies or RNA probes coupled to photocleavable oligonucleotide tags. After hybridizing the probes to tissue sections, specific cell types labeled with immunofluorescent markers release the oligonucleotide tags in discrete tissue regions through ultraviolet exposure. These released tags can be quantified using an nCounter assay or next-generation sequencing (NGS), and the counts are mapped back to their tissue location, providing a spatially resolved digital profile of analyte abundance. Our study employed DSP to identify genes with differential expression in normal colonic epithelial cells, SSL, and TSA, corroborating our findings through immunohistochemical (IHC) staining and functional studies.

Specimen

For this study, we utilized FFPE colonoscopic polypectomy or resection specimens diagnosed between 2018 and 2022 at National Taiwan University Hospital. A gastrointestinal pathology expert (Y.-M.J.) reviewed these sections to confirm the diagnoses and identify specific lesion areas.

Digital Spatial Profiling

For the DSP procedure, 5 μm sections of FFPE tissue were first subjected to baking at 60°C for 30 min. The slides were then deparaffinized using CitriSolv™ Hybrid Solvent and Clearing Agent (Decon Labs, King of Prussia, PA, USA) and rehydrated. Antigen retrieval was accomplished by incubating the slides in Tris-EDTA solution (pH = 9) and boiling them to 99°C in a Tinto-Retriever pressure cooker (Bio-SB, Santa Barbara, CA, USA). The samples were further treated with 1 μg/mL proteinase K (Thermo Fisher Scientific, Waltham, MA, USA) for 15 min at 37°C and fixed in neutral-buffered formalin for 10 min.

Next, tissue sections were subjected to hybridization with the GeoMx® Cancer Transcriptome Atlas panel, which includes photocleavable oligonucleotide-labeled probes targeting 1,823 oncology RNA targets. This hybridization occurred at 37°C overnight. Following hybridization, excess off-target probes were removed by washing, and the slides were counterstained using SYTO13 (GeoMx Nuclear Stain Morphology kit at 1:10) and GeoMx Solid Tumor TME Morphology kit (with 1:40 anti-panCK and 1:40 anti-CD45) for 2 h at room temperature. After staining, the slides were loaded onto the GeoMx DSP instrument, which was scanned to generate digital fluorescent images.

Regions of interest (ROI) were selected by a pathologist (Y.-M.J.), and the spatially indexed oligonucleotide barcodes from Pan-CK-positive cells within these ROIs were released through photocleavage. These released barcodes were then collected using microcapillary aspiration and transferred into a 96-well plate.

Subsequently, libraries were prepared via polymerase chain reaction (PCR) amplification using GeoMx Seq Code PCR Master Mix to add Illumina adapter sequences and unique dual-sample indices, following the manufacturer’s instructions. The libraries were then subjected to pair-end sequencing using an Illumina NextSeq 550 system. The GeoMx NGS Pipeline converted the resulting FASTQ sequencing files into digital count conversion files. Gene count and pathway analysis were done using Nanostring GeoMx Data Analysis Suite software (v2.2).

IHC Staining

IHC staining was conducted using the Ventana BenchMark XT autostainer (Ventana, Oro Valley, AZ, USA). Tissue sections underwent dewaxing and rehydration before antigen retrieval, achieved by incubating slides in a CC1 solution at 100°C for 40 min. Subsequently, the slides were exposed to primary antibodies at optimal dilutions for 18 min at room temperature. Following this, a polymer-HRP reagent (OptiView DAB IHC detection kit, Ventana) was applied, and peroxidase activity was visualized using a solution of diamino-benzidine tetrahydroxychloride. Hematoxylin was used to counterstain the sections. Immunoreactivity was evaluated by a single pathologist (Y.-M.J.), with results categorized as “negative” for absent staining or “positive” for cytoplasmic staining in tumor cells. The semiquantitative assessment of positive expression involved four categories: “−” (no stained cells), (“1+” less than 10% of cells stained), “2+” (10%–50% of cells stained), and “3+” (more than 50% of cells stained). Details of the antibodies and their dilutions can be found in online supplementary Table S1 (for all online suppl. material, see https://doi.org/10.1159/000539612).

Cell Culture

HCT116 cells were cultured in McCoy’s 5A Medium (Gibco, Thermo Fisher Scientific, Waltham, MA, USA)., while HEK293T, SW480, and Caco2 cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco). All media were supplemented with 10% fetal bovine serum (FBS)(Gibco), 100 U/mL penicillin, 100 μg/mL streptomycin, nonessential amino acids, and 1 mM sodium pyruvate. The culture medium for Caco2 cells also included human holo-transferrin (0.01 mg/mL, Sigma-Aldrich, St. Louis, MO, USA). Cells were maintained at 37°C in a humidified atmosphere containing 5% CO2 and 95% air.

RNA Interference

To explore the impact of CEACAM6 on colorectal tumorigenesis, we employed a strategy involving the downregulation of CEACAM6 expression in CRC cell lines using small hairpin RNA (shRNA). These shRNA constructs were designed within lentiviral vectors and were provided by the RNAi Consortium, distributed through the RNA Technology Platform and Gene Manipulation Core of Academia Sinica (Taipei, Taiwan). Two specific shRNA clones were utilized, each targeting CEACAM6 with the following sequences: CEACAM6-1: 5′- CCT​GCA​CAG​TAC​TCT​TGG​TTT-3′ and CEACAM6-2: 5′-CCC​AGA​ATC​GTA​TTG​GTT​ACA-3′. ShRNA vectors (shRFP or shLacZ) lacking significant homology with known human genome sequences were included as negative controls.

293T cells were transfected with 4 μg of pLKO in conjunction with 0.4 μg of envelope plasmid pMD.G and 3.6 μg of packaging plasmid pCMVΔR8.91 to generate lentiviral particles. Viral particles were collected from the culture medium 40 and 64 h post-transfection. To establish CEACAM6-knockdown cells, CRC cell lines were exposed to the virus-containing medium for 48 h. Subsequently, the medium was replaced with fresh medium supplemented with 2 μg/mL puromycin every 3 days for 2 weeks to select for drug-resistant cells.

Cell Proliferation Assay

To assess the influence of CEACAM6 knockdown on cell survival and proliferation, we conducted the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. In each experiment, 2.5 × 103 cells were seeded in individual wells of a 96-well plate, with six wells per group. At specified time points, MTT (2 mg/mL in PBS) was added to the wells, which were then incubated for 2 h. The resulting product of the color reaction, MTT formazan, was extracted using dimethyl sulfoxide, and the absorbance was measured at 570 nm.

Soft Agar Assay

To assess the clonogenic potential of cells, we employed an anchorage-independent cell proliferation assay known as the soft agar assay. In this procedure, we prepared two agar solutions: one with a 3.5% agar concentration and the other with 7% agar, both dissolved in ddH2O. These agar solutions were subjected to microwave heating and maintained in a water bath at 42°C. Subsequently, a two-layered agar plate was created by blending 2 × DMEM with an equal volume of melted agar, which was then promptly added to a culture dish. For the bottom layer, we introduced 2 mL of culture medium containing 0.7% agar into a 35 mm culture dish, allowing it to solidify at room temperature for 1 h. We mixed 5 × 103 cells with 0.35% agar in 2 × DMEM for the top layer and layered this mixture over the 0.7% agar base. The cells were cultured for a period of 8–12 weeks. After carefully removing the culture medium from the top of the soft agar-containing cell colonies, we added 300 μL of the 0.1% p-iodonitrotetrazolium violet (Sigma-Aldrich) solution to each well. Subsequently, the plates were incubated for 2 h at 37°C. Once the staining solution was removed, photographs were captured using a light microscope (×40 magnification). The numbers of colonies larger than 30 μm were counted. Each experiment was conducted in triplicate.

Boyden Chamber Migration and Invasion Assay

For the Boyden chamber migration and invasion assay, we utilized transwell inserts with 8 μm pore size in 24-well plates (Nuclepore, Pleasanton, CA, USA). Specifically, we seeded 1 × 105 cells in 100 μL of serum-free culture medium into the upper chamber, while the lower chamber received 500 μL of culture medium containing 10% FBS. After a 16-h incubation period, the cells were fixed in 4% paraformaldehyde for 10 min and then stained with 4′-6-diamidino-2-phenylindole (DAPI). Cells on the upper side of the filters were removed using cotton-tipped swabs, and the filters were subsequently rinsed with phosphate-buffered saline (PBS). Cells on the underside of the filters were visualized and quantified using a fluorescence microscope. The cell invasion assay followed a similar protocol to the migration assay, with the key difference being that the filters were coated with Matrigel (30 μg; Collaborative Biomedical Products, Bedford, MA, USA), and the incubation time was extended to 20 h.

Real-Time PCR

To assess the mRNA levels of the target gene, we conducted a real-time PCR assay employing SYBR green technology. This analysis was carried out using the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA). As a reference for quantifying RNA, we utilized glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a commonly employed housekeeping gene. Details of the specific primers employed can be found in online supplementary Table S2. In brief, within a 20-μL PCR reaction mixture, 1 μL of complementary DNA template was combined with 10 μL of 2× Power SYBR® PCR master mix (Applied Biosystems), 200 nm of paired primers, and distilled water. The PCR amplification process commenced with an initial incubation at 50°C for 2 min, followed by denaturation at 95°C for 10 min. Subsequently, it underwent 40 cycles, which consisted of denaturation at 95°C for 15 s and annealing at 60°C for 1 min. Following each run, we analyzed the melting curves to confirm the amplification of a single PCR product only. To determine the relative expression of the target genes within each sample, we utilized the following formula: relative mRNA expression = 2−ΔCt × 256. ΔCt = Cttarget gene − CtGAPDH.

Identification of Differentially Expressed Genes in Colonic Lesions Using DSP Technology

To investigate the progressive changes from normal colonic mucosa to SSL and further to TSA, we analyzed 8 mucosectomy specimens containing TSA, adjacent SSL as a precursor lesion, and normal colonic mucosa, employing Nanostring’s GeoMx DSP technology. To unravel the transcriptomic alterations, we utilized the GeoMx® Cancer Transcriptome Atlas panel, comprising over 1,800 oligonucleotide probes targeting genes associated with cancer biology. Sections were subjected to immunostaining with cytokeratin to highlight epithelial cells. One or two ROIs were selected for normal colonic mucosa, SSL, and TSA in each sample. After photocleavage, the probes within cytokeratin-positive cells in each ROI were collected and subjected to analysis via NGS. Nine ROIs from normal mucosa, 12 from SSL, and 7 from TSA passed quality control and were included for further analysis. The experimental procedure and representative images of HE and cytokeratin immunostaining used for DSP analysis are displayed in Figure 1a and b, respectively.

Fig. 1.

Application of the DSP technique for discovering gene expression alterations in serrated pathway lesions. a The DSP workflow with NGS readout is illustrated. It involves a combination of high-plex photocleavable oligo-linked RNA probes (1) and fluorescence-conjugated cell markers (2) for tissue section staining. (3) Scanning of slides followed by the selection of ROIs. (4) Ultraviolet light exposure to release oligonucleotides from specific cell types in the selected ROIs. (5) Collection of photo-released oligonucleotides via a microcapillary tube into a microtiter plate. (6) Unique indexing and quantifying spatially resolved pools of photocleaved oligonucleotides during library preparation using NGS. b Depiction of DSP and H and E images on a slide containing normal colonic mucosa, TSA, and its precursor SSL. ROIs are labeled as follows: ROI-1: TSA, ROI-3: SSL, ROI-4: normal colonic mucosa. In the DSP image, epithelial cells and immune cells are labeled with fluorescent antibodies to PanCK (green) and CD45 (blue), while nuclei are stained with DAPI. c Volcano plots illustrating differential gene expression between normal colonic mucosa and SSL, as well as normal colonic mucosa and TSA. The x-axis represents the log2 change in gene expression between lesion types, while the y-axis displays the −log10 p value. Genes upregulated are shown in red dots, while downregulated genes are displayed in blue dots. d A three-dimensional principal component analysis plot showing normal mucosa ROIs’ clustering and SSL ROIs’ intermixing with TSA ROIs. Different tissue types are color-coded: gray for normal colonic mucosa, blue for SSL, and red for TSA. e The top 20 enriched pathways in SSL and TSA are displayed on the x-axis as log10 p values.

Fig. 1.

Application of the DSP technique for discovering gene expression alterations in serrated pathway lesions. a The DSP workflow with NGS readout is illustrated. It involves a combination of high-plex photocleavable oligo-linked RNA probes (1) and fluorescence-conjugated cell markers (2) for tissue section staining. (3) Scanning of slides followed by the selection of ROIs. (4) Ultraviolet light exposure to release oligonucleotides from specific cell types in the selected ROIs. (5) Collection of photo-released oligonucleotides via a microcapillary tube into a microtiter plate. (6) Unique indexing and quantifying spatially resolved pools of photocleaved oligonucleotides during library preparation using NGS. b Depiction of DSP and H and E images on a slide containing normal colonic mucosa, TSA, and its precursor SSL. ROIs are labeled as follows: ROI-1: TSA, ROI-3: SSL, ROI-4: normal colonic mucosa. In the DSP image, epithelial cells and immune cells are labeled with fluorescent antibodies to PanCK (green) and CD45 (blue), while nuclei are stained with DAPI. c Volcano plots illustrating differential gene expression between normal colonic mucosa and SSL, as well as normal colonic mucosa and TSA. The x-axis represents the log2 change in gene expression between lesion types, while the y-axis displays the −log10 p value. Genes upregulated are shown in red dots, while downregulated genes are displayed in blue dots. d A three-dimensional principal component analysis plot showing normal mucosa ROIs’ clustering and SSL ROIs’ intermixing with TSA ROIs. Different tissue types are color-coded: gray for normal colonic mucosa, blue for SSL, and red for TSA. e The top 20 enriched pathways in SSL and TSA are displayed on the x-axis as log10 p values.

Close modal

Using criteria of a 2-fold change and p < 0.05 for inclusion, we identified 10 genes that were upregulated in SSL compared to normal colonic mucosa and 1 gene that was downregulated (Fig. 1c). Similarly, we found 10 genes upregulated in TSA compared to normal colonic mucosa, and 2 genes that were downregulated (Fig. 1c). Notably, among the genes upregulated in serrated lesions, 7 were elevated in both SSL and TSA (ANXA1, SERPIN5, CEACAM6, lipocalin-2 (LCN2), COL17A1, CD55, LYZ). In contrast, ADH1/2/3 were downregulated in both SSL and TSA. No differentially expressed genes were identified between SSL and TSA. These results suggest that these two serrated lesions share a similar transcriptome.

We employed principal component analysis to examine whether the expression profiles could distinguish different disease entities and reveal general similarities and consistencies within the same disease entity (Fig. 1d). This analysis demonstrated that ROIs from normal mucosa clustered together. However, the expression profiles could not separate TSA from SSL. These results indicate that significant changes in gene expression occur between normal mucosa and serrated lesions rather than between SSL and TSA. The expression profiles of different serrated lesions were highly similar, suggesting that early lesions in tumor formation exhibit significant homogeneity, in contrast to the high heterogeneity observed in malignant tumors.

Identification of the Pathways Activated in SSL and TSA

We employed Gene Set Enrichment Analysis to determine the pathways activated in SSL and TSA) (Fig. 1e). Our analysis revealed upregulation of inflammation and angiogenesis pathways in both SSL and TSA. Given that SSL and TSA result from mutations in MAPK pathway genes, we also observed significant MAPK pathway signaling in both lesions. Notably, while mutations activating the Wnt pathway are commonly found in TSA [12‒14], we did not detect Wnt pathway activation in TSA.

Confirmation of Upregulation of CEACAM6, LCN2, KRT19, and Lysozyme in Serrated Lesions through Immunohistochemistry

Among the upregulated genes, we selected CEACAM6, LCN2, KRT19, and lysozyme for further validation due to their consistent upregulation in serrated lesions (Fig. 2a) and the availability of specific antibodies. CAMCAM6 and KRT19 exhibited weak staining in normal colonic mucosa, primarily in the surface cells. LCN2 and lysozyme staining were negative in normal colonic mucosa. In contrast, HP, SSL, and TSA specimens displayed diffuse and strong CEACAM6, KRT19, and lysozyme staining in nearly all lesions. Immunostaining for LCN2 showed more heterogeneity, with 2+ to 3+ expression detected in 77.4% of HP, 51.9% of SSL, and 93.4% of TSA (Fig. 2b, c).

Fig. 2.

Validation of DSP findings through IHC staining. a The bar graph presents elevated CEACAM6, LCN2, KRT19, and LYZ expression levels in SSL and TSA compared to the colonic mucosa, as determined by the DSP assay. Significance levels are as follows: * for p < 0.05, ** for p < 0.01, *** for p < 0.001. The y-axis represents the normalized gene count obtained from the NGS readout. b Representative IHC staining images for these four proteins in normal colonic mucosa, HP, SSL, and TSA. Original magnification: ×100. c Stacked bar plots illustrate the distribution of IHC scores for the four markers in normal colonic mucosa, HP, SSL, and TSA (n = 30 for each). Scoring criteria are as follows: 0 for no stained cells, 1+ for less than 10% of cells stained, 2+ for 10%–50% of cells stained, and 3+ for more than 50% of cells stained.

Fig. 2.

Validation of DSP findings through IHC staining. a The bar graph presents elevated CEACAM6, LCN2, KRT19, and LYZ expression levels in SSL and TSA compared to the colonic mucosa, as determined by the DSP assay. Significance levels are as follows: * for p < 0.05, ** for p < 0.01, *** for p < 0.001. The y-axis represents the normalized gene count obtained from the NGS readout. b Representative IHC staining images for these four proteins in normal colonic mucosa, HP, SSL, and TSA. Original magnification: ×100. c Stacked bar plots illustrate the distribution of IHC scores for the four markers in normal colonic mucosa, HP, SSL, and TSA (n = 30 for each). Scoring criteria are as follows: 0 for no stained cells, 1+ for less than 10% of cells stained, 2+ for 10%–50% of cells stained, and 3+ for more than 50% of cells stained.

Close modal

Immunostaining of CEACAM6 in Conventional Pathway Colorectal Lesions

Among the four targets under analysis, CEACAM6 has previously been reported as a potentially oncogenic protein [16, 17]. Therefore, we conducted immunostaining on lesions from the conventional pathway in the colorectum to compare gene expression differences between serrated and conventional pathways. As illustrated in Figure 3a and b, most tubular adenomas (TAs) exhibited negative or weak staining for CEACAM6. In contrast, nearly all TAs with high-grade dysplasia (TA_HD) and adenocarcinomas displayed intense CEACAM6 staining. These findings suggest that CEACAM6 overexpression is an early occurrence in serrated pathway lesions but a later event in the conventional pathway.

Fig. 3.

CEACAM6 expression occurs late in the conventional adenoma-carcinoma pathway. a Representative IHC staining of CEACAM6 in TA, TA_HD, and AD. b Stacked bar plot illustrating the distribution of IHC scores for CEACAM6 staining in normal colonic mucosa, SSL, TSA, TA, TA_HD, and AD. Data are based on n = 30 samples for each tissue type. c Analysis of the RNA-Seq dataset GEO76987 demonstrates the overexpression of CEACAM6, LCN2, and LYZ in HP and SSL but not in TA. LYZ exhibits overexpression in SSL but not in HP or TA. AD, adenocarcinoma.

Fig. 3.

CEACAM6 expression occurs late in the conventional adenoma-carcinoma pathway. a Representative IHC staining of CEACAM6 in TA, TA_HD, and AD. b Stacked bar plot illustrating the distribution of IHC scores for CEACAM6 staining in normal colonic mucosa, SSL, TSA, TA, TA_HD, and AD. Data are based on n = 30 samples for each tissue type. c Analysis of the RNA-Seq dataset GEO76987 demonstrates the overexpression of CEACAM6, LCN2, and LYZ in HP and SSL but not in TA. LYZ exhibits overexpression in SSL but not in HP or TA. AD, adenocarcinoma.

Close modal

Using RNA sequencing (RNA-Seq) data from an external dataset (GEO76987), which includes 41 specimens of normal colonic mucosa, 10 HPs, 21 SSLs, and 10 TAs, we confirmed the overexpression of CEACAM6, LCN2, and KRT19 in HPs and SSLs but not in TAs. LYZ was found to be overexpressed in SSLs but not in HPs or TAs. The discrepancy of LYZ expression in HP between our result and theirs may be due to the difference in the method used. We used IHC staining, which is more sensitive than RNA-Seq in identifying gene expression in tumor cells, especially for small lesions with abundant non-neoplastic cells. These results indicate that these genes are overexpressed explicitly in precursor lesions developed from the serrated pathway (Fig. 3c).

The Overexpression of Genes Identified by DSP Is Not Attributable to MAPK Pathway Activation

The defining genetic feature of serrated pathway lesions is the activation of the MAPK pathway, often driven by KRAS or BRAF mutations. Consequently, we initially hypothesized that the upregulation of genes identified through DSP might result from MAPK pathway activation. To investigate this hypothesis, we treated CRC cell lines (HCT116, SW480, and Caco2) with the MEK inhibitor trametinib. Western blot analysis confirmed a reduction in MEK and ERK phosphorylation levels (Fig. 4a). The impact of trametinib treatment on the expression levels of target genes was assessed through real-time PCR (Fig. 4b–d). Surprisingly, only ITGB4 consistently displayed downregulation upon trametinib treatment. Additionally, we compared the expression levels of CEACAM6 and LCN2 in KRAS-mutated CRC, BRAF-mutated CRC, and CRC without KRAS or BRAF mutations, finding that the expression levels of CEACAM6 and LCN2 were not elevated in KRAS- or BRAF-mutated CRCs (Fig. 4e, f). These results suggest that the overexpression of the genes identified by DSP is not directly linked to MAPK pathway activation.

Fig. 4.

Overexpression of genes identified by DSP is independent of MAPK pathway activation. a CRC cell lines were treated with the MEK1/2 inhibitor trametinib (100 nm) for 24 h. Western blotting reveals reduced phosphorylation levels of MEK and ERK following trametinib treatment. b–d Real-time PCR analysis of the impact of trametinib on mRNA expression levels of target genes. Statistically significant differences (*p < 0.05, **p < 0.01, ***p < 0.001) between trametinib-treated cells and DMSO control are indicated above the bars. Data are presented as mean ± standard error of the mean (SEM) for three technical replicates. e, f CEACAM6 and LCN2 expression analysis in KRAS-mutated CRC, BRAF-mutated CRC, and CRC without KRAS or BRAF mutations.

Fig. 4.

Overexpression of genes identified by DSP is independent of MAPK pathway activation. a CRC cell lines were treated with the MEK1/2 inhibitor trametinib (100 nm) for 24 h. Western blotting reveals reduced phosphorylation levels of MEK and ERK following trametinib treatment. b–d Real-time PCR analysis of the impact of trametinib on mRNA expression levels of target genes. Statistically significant differences (*p < 0.05, **p < 0.01, ***p < 0.001) between trametinib-treated cells and DMSO control are indicated above the bars. Data are presented as mean ± standard error of the mean (SEM) for three technical replicates. e, f CEACAM6 and LCN2 expression analysis in KRAS-mutated CRC, BRAF-mutated CRC, and CRC without KRAS or BRAF mutations.

Close modal

CEACAM6 Promotes Proliferation and Invasion Abilities of CRC

To explore the functional role of CEACAM6 in CRC tumorigenesis, we knocked down CEACAM6 expression using shRNA in HCT116, SW480, and Caco2 cells. Real-time PCR confirmed the downregulation of CEACAM6 expression (Fig. 5a–c). Silencing CEACAM6 expression inhibited the proliferation of CRC cell lines (Fig. 5d–f). A soft agar assay also revealed that silencing CEACAM6 expression hindered anchorage-independent cell growth (Fig. 5g–i). In modified Boyden chamber assays, silencing CEACAM6 inhibited the migration and invasion abilities of HCT116 and Caco2 cells (Fig. 5j, l), while the migration and invasion abilities of SW480 cells were unaffected by CEACAM6 knockdown (Fig. 5k).

Fig. 5.

Silencing of CEACAM6 suppresses proliferation, colony formation, migration, and invasion of CRC cells. a–c The efficiency of shRNA-mediated knockdown of CEACAM6 in HCT116, SW480, and Caco2 cells was verified by real-time PCR. d–f Silencing of CEACAM6 results in reduced proliferation of CRC cell lines. g–i Soft agar assay demonstrates that silencing of CEACAM6 inhibits anchorage-independent growth of CRC cells. j–l Silencing of CEACAM6 diminishes migration and invasion abilities of HCT116 and Caco2 cells. “ns” indicates no statistical significance, while *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 5.

Silencing of CEACAM6 suppresses proliferation, colony formation, migration, and invasion of CRC cells. a–c The efficiency of shRNA-mediated knockdown of CEACAM6 in HCT116, SW480, and Caco2 cells was verified by real-time PCR. d–f Silencing of CEACAM6 results in reduced proliferation of CRC cell lines. g–i Soft agar assay demonstrates that silencing of CEACAM6 inhibits anchorage-independent growth of CRC cells. j–l Silencing of CEACAM6 diminishes migration and invasion abilities of HCT116 and Caco2 cells. “ns” indicates no statistical significance, while *p < 0.05, **p < 0.01, ***p < 0.001.

Close modal

To further investigate the roles of CEACAM6 in CRC tumorigenesis, we overexpressed CEACAM6 in CRC cell lines via lentiviral transduction. Western blotting confirmed the overexpression of CEACAM6 (Fig. 6a–c). Although the overexpression of CEACAM6 had little to no effect on proliferation in both anchorage-dependent and -independent conditions (Fig. 6d–i), it notably promoted the migration and invasion abilities of all three CRC cell lines (Fig. 6j–l). These findings suggest that CEACAM6 plays a pivotal role in CRC by impacting clonogenicity, growth, migration, and invasion.

Fig. 6.

Overexpression of CEACAM6 enhances migration and invasion of CRC cells. a–c Confirmation of overexpression of CEACAM6 in CRC cells following lentiviral transduction by western blotting. d–f Overexpression of CEACAM6 enhances the proliferation of SW480 cells but does not affect HCT116 and Caco2 cell proliferation. g–i Overexpression of CEACAM6 does not impact the colony-forming abilities of CRC cells in the soft agar assay. j–l Overexpression of CEACAM6 promotes migration and invasion of CRC cells. “ns” indicates no statistical significance, while *p < 0.05, **p < 0.01.

Fig. 6.

Overexpression of CEACAM6 enhances migration and invasion of CRC cells. a–c Confirmation of overexpression of CEACAM6 in CRC cells following lentiviral transduction by western blotting. d–f Overexpression of CEACAM6 enhances the proliferation of SW480 cells but does not affect HCT116 and Caco2 cell proliferation. g–i Overexpression of CEACAM6 does not impact the colony-forming abilities of CRC cells in the soft agar assay. j–l Overexpression of CEACAM6 promotes migration and invasion of CRC cells. “ns” indicates no statistical significance, while *p < 0.05, **p < 0.01.

Close modal

In this study, we utilized DSP for the first time to discern alterations in gene expression within the initial stages of colorectal tumorigenesis via the serrated pathway. Within a pool of 1,823 oncology-related RNA targets, we pinpointed 10 genes exhibiting elevated expression in SSL and another 10 genes showing increased expression in TSA. Subsequently, we selected four genes for further scrutiny through IHC staining, and all four genes were confirmed. These findings underscore the reliability of DSP as a method for detecting gene expression modifications in minute lesions, particularly in tissues with intricate microenvironments such as colorectal mucosa.

An intriguing revelation is that, in contrast to the pronounced heterogeneity observed in advanced tumors, the transcriptional changes during the early stages of tumorigenesis display a relatively uniform pattern. IHC staining demonstrates that nearly all SSL and TSA specimens express all four proteins examined via this method. These results are also corroborated by RNA-Seq data obtained from an external dataset (see Fig. 3c). This outcome is not entirely unexpected, given that fewer genetic alterations and chromosomal stability characterize early lesions. The consistent expression patterns observed here offer promising avenues for intervening to thwart tumor formation.

The serrated pathway is hallmarked by the abnormal activation of the MAPK pathway, often attributed to mutations in KRAS or BRAF. Thus, the upregulated genes identified through DSP might result from heightened MAPK pathway activity. However, when we exposed CRC cell lines to the MEK inhibitor trametinib, it failed to diminish the expression of the upregulated genes in SSL or TSA. These outcomes suggest that while mutations in key driver genes remain pivotal for the initiation of tumors, epigenetic changes and alterations in RNA or protein expression levels may wield significant influence in the progression of these tumors. Uncovering these early expression events is challenging in cell line models alone. Hence, it is imperative to delve into clinical samples derived from precursor lesions of tumors to elucidate these initial molecular events.

The human carcinoembryonic antigen (CEA)-related cell adhesion molecule (CEACAM) family comprises 12 proteins that are either anchored to cell membranes via a glycosylphosphatidylinositol linkage or through the transmembrane domain [18]. These CEACAM proteins participate in various cellular functions, including cell proliferation, the inhibition of apoptosis and anoikis, and the progression and metastasis of cancer [18, 19]. CEACAM5, also known as CEA, is frequently overexpressed in CRC and serves as a serum marker for CRC in routine clinical practice [20]. CEACAM6 has been reported to be overexpressed in a wide range of cancer types, including CRC [21, 22]. Because of the frequent expression of CEACAM6 in cancer, the possible roles of CEACAM6 as a diagnostic or prognostic biomarker were explored. However, the sensitivity of CEACAM6 as a serum marker for CRC is only 34.5% [23]. The lack of sensitivity is probably due to the expression of CEACAM6 in nontumorous cells, such as neutrophils and alveolar cells of the lung [24, 25]. CEACAM6 expression is associated with higher stage and independently predicts poor overall survival in patients with CRC [26, 27]. CEACAM6 overexpression promotes tumor cell proliferation through the ERK and AKT pathways [28] and enhances tumor invasion by activating the Src/FAK pathway while inhibiting anoikis [22]. The widespread expression of CEACAM6 in CRC and its precursor lesions make it an ideal target for CRC treatment and prevention. NEO-201, a humanized monoclonal antibody targeting CEACAM5 and CEACAM6, has demonstrated safety and tolerability in a phase 1 clinical trial [29]. A phase 1/2 trial of NEO-201 combined with pembrolizumab is underway (ClinicalTrials.gov Identifier: NCT03476681).

LCN2 is a glycoprotein belonging to the lipocalin superfamily and plays pivotal roles in various physiological functions, including antioxidant, antibacterial, and iron transport activities [30]. LCN2 is overexpressed in multiple malignancies and facilitates tumor invasion and metastasis [31, 32]. Elevated LCN2 levels are indicative of poor prognosis in CRC patients [33]. Interestingly, LCN2 expression in colon mucosa is inversely correlated with cancer development in patients with ulcerative colitis [34]. The precise roles of LCN2 in different pathways of colorectal tumorigenesis require further investigation.

We observed widespread expression of lysozyme in both SSL and TSA. Lysozyme is an antimicrobial protein typically found in Paneth cells, rarely present in normal colonic epithelium. The unexpected diffuse presence of lysozyme in SSL and TSA suggests an abnormal differentiation of tumor cells. Interestingly, LCN2, another antibacterial protein, has been reported to be associated with these lesions, and our pathway analysis revealed an upregulation of phagosome and immune pathways in SSL and TSA. The significance of these elevated immune-related pathways in tumorigenesis remains unclear.

One significant limitation of this study is our use of the GeoMx® Cancer Transcriptome Atlas panel, which includes only 1,823 probes targeting genes related to cancer biology. The company has also developed a whole transcriptome panel, providing a more comprehensive view of gene expression alterations in SSL and TSA. Additionally, beyond tumor cells, DSP can be employed to assess changes in the tumor microenvironment, an aspect not addressed in this study. Future research could focus on environmental cells and the mutual interactions between tumor cells and these environmental components.

In summary, our utilization of USP technology has enabled us to identify genes that are overexpressed in SSL and TSA. Among them, we have confirmed the overexpression of CEACAM6 and verified its role in colorectal tumorigenesis. These findings shed light on the early molecular changes in colorectal tumorigenesis and offer potential targets for intervention.

We thank the First and Second Core Lab of National Taiwan University and National Taiwan University Hospital and the RNA Technology Platform and Gene Manipulation Core of Academia Sinica for providing technical support. We also thank Arkady Cheng for the English editing.

The Research Ethics Committee of the National Taiwan University Hospital approved this study (Approval No. 202105070RINA), and all research activities were conducted following the principles outlined in the Declaration of Helsinki. The National Taiwan University Hospital Research Ethics Committee waived the need for informed consent.

The authors have no conflict of interest.

This study is supported by a grant from Min-Sheng General Hospital to M.C.S. and Y.-M.J. by the Yushan Scholar Program by the Ministry of Education, Taiwan, (NTU-112V1402-5) to R.Y.J.H.

M.-C.S. and Y.-M.J. conceived and designed the experiments and provided financial support. C.-H.H., J.-R.L., and Y.-T.F. carried out the experiments. H.-Y.L. analyzed the bioinformatics data. R.Y.-J.H. and K.-C.C. provided technical support and equipment. M.-C.S., C.-H.H., and Y.-M.J. interpreted the data and wrote the manuscript. All authors discussed the results, commented on the paper, revised it, and provided final approval.

Additional Information

Min-Cheng Su and Ching-Hsiang Hsu contributed equally to this work.

Data are not publicly available due to ethical reasons. Further inquiries can be directed to the corresponding author.

1.
Baidoun
F
,
Elshiwy
K
,
Elkeraie
Y
,
Merjaneh
Z
,
Khoudari
G
,
Sarmini
MT
, et al
.
Colorectal cancer epidemiology: recent trends and impact on outcomes
.
Curr Drug Targets
.
2021
;
22
(
9
):
998
1009
.
2.
Goldstein
NS
.
Serrated pathway and APC (conventional)-type colorectal polyps: molecular-morphologic correlations, genetic pathways, and implications for classification
.
Am J Clin Pathol
.
2006
;
125
(
1
):
146
53
.
3.
Cho
KR
,
Vogelstein
B
.
Genetic alterations in the adenoma-carcinoma sequence
.
Cancer
.
1992
;
70
(
6 Suppl l
):
1727
31
.
4.
Patai
AV
,
Molnár
B
,
Tulassay
Z
,
Sipos
F
.
Serrated pathway: alternative route to colorectal cancer
.
World J Gastroenterol
.
2013
;
19
(
5
):
607
15
.
5.
Pai
RK
,
Bettington
M
,
Srivastava
A
,
Rosty
C
.
An update on the morphology and molecular pathology of serrated colorectal polyps and associated carcinomas
.
Mod Pathol
.
2019
;
32
(
10
):
1390
415
.
6.
O'Brien
MJ
,
Yang
S
,
Mack
C
,
Xu
H
,
Huang
CS
,
Mulcahy
E
, et al
.
Comparison of microsatellite instability, CpG island methylation phenotype, BRAF and KRAS status in serrated polyps and traditional adenomas indicates separate pathways to distinct colorectal carcinoma end points
.
Am J Surg Pathol
.
2006
;
30
(
12
):
1491
501
.
7.
Liu
C
,
Walker
NI
,
Leggett
BA
,
Whitehall
VL
,
Bettington
ML
,
Rosty
C
.
Sessile serrated adenomas with dysplasia: morphological patterns and correlations with MLH1 immunohistochemistry
.
Mod Pathol
.
2017
;
30
(
12
):
1728
38
.
8.
Longacre
TA
,
Fenoglio-Preiser
CM
.
Mixed hyperplastic adenomatous polyps/serrated adenomas. A distinct form of colorectal neoplasia
.
Am J Surg Pathol
.
1990
;
14
(
6
):
524
37
.
9.
Bettington
ML
,
Walker
NI
,
Rosty
C
,
Brown
IS
,
Clouston
AD
,
McKeone
DM
, et al
.
A clinicopathological and molecular analysis of 200 traditional serrated adenomas
.
Mod Pathol
.
2015
;
28
(
3
):
414
27
.
10.
Sekine
S
,
Yamashita
S
,
Tanabe
T
,
Hashimoto
T
,
Yoshida
H
,
Taniguchi
H
, et al
.
Frequent PTPRK-RSPO3 fusions and RNF43 mutations in colorectal traditional serrated adenoma
.
J Pathol
.
2016
;
239
(
2
):
133
8
.
11.
Torlakovic
EE
,
Gomez
JD
,
Driman
DK
,
Parfitt
JR
,
Wang
C
,
Benerjee
T
, et al
.
Sessile serrated adenoma (SSA) vs. traditional serrated adenoma (TSA)
.
Am J Surg Pathol
.
2008
;
32
(
1
):
21
9
.
12.
Tsai
JH
,
Liau
JY
,
Yuan
CT
,
Lin
YL
,
Tseng
LH
,
Cheng
ML
, et al
.
RNF43 is an early and specific mutated gene in the serrated pathway, with increased frequency in traditional serrated adenoma and its associated malignancy
.
Am J Surg Pathol
.
2016
;
40
(
10
):
1352
9
.
13.
Sekine
S
,
Yamashita
S
,
Tanabe
T
,
Hashimoto
T
,
Yoshida
H
,
Taniguchi
H
, et al
.
Frequent PTPRK-RSPO3 fusions and RNF43 mutations in colorectal traditional serrated adenoma
.
J Pathol
.
2016
;
239
(
2
):
133
8
.
14.
Hashimoto
T
,
Ogawa
R
,
Yoshida
H
,
Taniguchi
H
,
Kojima
M
,
Saito
Y
, et al
.
Acquisition of WNT pathway gene alterations coincides with the transition from precursor polyps to traditional serrated adenomas
.
Am J Surg Pathol
.
2019
;
43
(
1
):
132
9
.
15.
Merritt
CR
,
Ong
GT
,
Church
SE
,
Barker
K
,
Danaher
P
,
Geiss
G
, et al
.
Multiplex digital spatial profiling of proteins and RNA in fixed tissue
.
Nat Biotechnol
.
2020
;
38
(
5
):
586
99
.
16.
Tian
C
,
Zhang
B
,
Ge
C
.
Effect of CEACAM6 silencing on the biological behavior of human gallbladder cancer cells
.
Oncol Lett
.
2020
;
20
(
3
):
2677
88
.
17.
Zang
M
,
Zhang
B
,
Zhang
Y
,
Li
J
,
Su
L
,
Zhu
Z
, et al
.
CEACAM6 promotes gastric cancer invasion and metastasis by inducing epithelial-mesenchymal transition via PI3K/AKT signaling pathway
.
PLoS One
.
2014
;
9
(
11
):
e112908
.
18.
Beauchemin
N
,
Arabzadeh
A
.
Carcinoembryonic antigen-related cell adhesion molecules (CEACAMs) in cancer progression and metastasis
.
Cancer Metastasis Rev
.
2013
;
32
(
3–4
):
643
71
.
19.
Zhang
Y
,
Zang
M
,
Li
J
,
Ji
J
,
Zhang
J
,
Liu
X
, et al
.
CEACAM6 promotes tumor migration, invasion, and metastasis in gastric cancer
.
Acta Biochim Biophys Sin
.
2014
;
46
(
4
):
283
90
.
20.
Nikolaou
S
,
Qiu
S
,
Fiorentino
F
,
Rasheed
S
,
Tekkis
P
,
Kontovounisios
C
.
Systematic review of blood diagnostic markers in colorectal cancer
.
Tech Coloproctol
.
2018
;
22
(
7
):
481
98
.
21.
Jantscheff
P
,
Terracciano
L
,
Lowy
A
,
Glatz-Krieger
K
,
Grunert
F
,
Micheel
B
, et al
.
Expression of CEACAM6 in resectable colorectal cancer: a factor of independent prognostic significance
.
J Clin Oncol
.
2003
;
21
(
19
):
3638
46
.
22.
Kim
EY
,
Cha
YJ
,
Jeong
S
,
Chang
YS
.
Overexpression of CEACAM6 activates Src-FAK signaling and inhibits anoikis, through homophilic interactions in lung adenocarcinomas
.
Transl Oncol
.
2022
;
20
:
101402
.
23.
Kuroki
M
,
Matsushita
H
,
Matsumoto
H
,
Hirose
Y
,
Senba
T
,
Yamamoto
T
.
Nonspecific cross-reacting antigen-50/90 (NCA-50/90) as a new tumor marker
.
Anticancer Res
.
1999
;
19
(
6C
):
5599
606
.
24.
Kuroki
M
,
Yamanaka
T
,
Matsuo
Y
,
Oikawa
S
,
Nakazato
H
,
Matsuoka
Y
.
Immunochemical analysis of carcinoembryonic antigen (CEA)-related antigens differentially localized in intracellular granules of human neutrophils
.
Immunol Invest
.
1995
;
24
(
5
):
829
43
.
25.
Chapin
C
,
Bailey
NA
,
Gonzales
LW
,
Lee
JW
,
Gonzalez
RF
,
Ballard
PL
.
Distribution and surfactant association of carcinoembryonic cell adhesion molecule 6 in human lung
.
Am J Physiol Lung Cell Mol Physiol
.
2012
;
302
(
2
):
L216
25
.
26.
Jantscheff
P
,
Terracciano
L
,
Lowy
A
,
Glatz-Krieger
K
,
Grunert
F
,
Micheel
B
, et al
.
Expression of CEACAM6 in resectable colorectal cancer: a factor of independent prognostic significance
.
J Clin Oncol
.
2003
;
21
(
19
):
3638
46
.
27.
Kim
KS
,
Kim
JT
,
Lee
SJ
,
Kang
MA
,
Choe
IS
,
Kang
YH
, et al
.
Overexpression and clinical significance of carcinoembryonic antigen-related cell adhesion molecule 6 in colorectal cancer
.
Clin Chim Acta
.
2013
;
415
:
12
9
.
28.
Zhu
R
,
Ge
J
,
Ma
J
,
Zheng
J
.
Carcinoembryonic antigen related cell adhesion molecule 6 promotes the proliferation and migration of renal cancer cells through the ERK/AKT signaling pathway
.
Transl Androl Urol
.
2019
;
8
(
5
):
457
66
.
29.
Cole
CB
,
Morelli
MP
,
Fantini
M
,
Miettinen
M
,
Fetsch
P
,
Peer
C
, et al
.
First-in-human phase 1 clinical trial of anti-core 1 O-glycans targeting monoclonal antibody NEO-201 in treatment-refractory solid tumors
.
J Exp Clin Cancer Res
.
2023
;
42
(
1
):
76
.
30.
Asaf
S
,
Maqsood
F
,
Jalil
J
,
Sarfraz
Z
,
Sarfraz
A
,
Mustafa
S
, et al
.
Lipocalin 2-not only a biomarker: a study of current literature and systematic findings of ongoing clinical trials
.
Immunol Res
.
2023
;
71
(
3
):
287
313
.
31.
Ören
B
,
Urosevic
J
,
Mertens
C
,
Mora
J
,
Guiu
M
,
Gomis
RR
, et al
.
Tumour stroma-derived lipocalin-2 promotes breast cancer metastasis
.
J Pathol
.
2016
;
239
(
3
):
274
85
.
32.
Du
ZP
,
Wu
BL
,
Xie
YM
,
Zhang
YL
,
Liao
LD
,
Zhou
F
, et al
.
Lipocalin 2 promotes the migration and invasion of esophageal squamous cell carcinoma cells through a novel positive feedback loop
.
Biochim Biophys Acta
.
2015
;
1853
(
10 Pt A
):
2240
50
.
33.
Maier
HT
,
Aigner
F
,
Trenkwalder
B
,
Zitt
M
,
Vallant
N
,
Perathoner
A
, et al
.
Up-regulation of neutrophil gelatinase-associated lipocalin in colorectal cancer predicts poor patient survival
.
World J Surg
.
2014
;
38
(
8
):
2160
7
.
34.
Kou
F
,
Cheng
Y
,
Shi
L
,
Liu
J
,
Liu
Y
,
Shi
R
, et al
.
LCN2 as a potential diagnostic biomarker for ulcerative colitis-associated carcinogenesis related to disease duration
.
Front Oncol
.
2021
;
11
:
793760
.