Warning Signs From the Crypt: Aberrant Protein Glycosylation Marks Opportunities for Early Colorectal Cancer Detection

Abstract

Colorectal cancer (CRC) remains a leading cause of cancer-related deaths despite being the most preventable and treatable forms of cancer when caught early through screening. There is an unmet need for novel screening approaches with improved accuracy, less invasiveness, and reduced costs. In recent years, evidence has accumulated around particular biological events that happen during the adenoma-to-carcinoma transition, especially focusing on precancerous immune responses in the colonic crypt. Protein glycosylation plays a central role in driving those responses, and recently, numerous reports have been published on how aberrant protein glycosylation both in colonic tissue and on circulating glycoproteins reflects these precancerous developments. The complex field of glycosylation, which exceeds complexity of proteins by several orders of magnitude, can now be studied primarily because of the availability of new high-throughput technologies such as mass spectrometry and artificial intelligence-powered data processing. This has now opened new avenues for studying novel biomarkers for CRC screening. This review summarizes the early events taking place from the normal colon mucosa toward adenoma and adenocarcinoma formation and associated critical protein glycosylation phenomena, both on the tissue level and in the circulation. These insights will help establish an understanding in the interpretation of novel CRC detection modalities that involve high-throughput glycomics.

INTRODUCTION

Despite worldwide population-based screening efforts, colorectal cancer (CRC) remains a global public health issue with an estimated 1.9 million new cases and 935,000 deaths worldwide (1,2). Nearly 80% arise from adenomatous polyps, progressing slowly from a small-to-large adenoma, showing dysplastic changes that transform to carcinoma over many years (3). Successful CRC screening programs identify individuals for colonoscopy to remove precancerous polyps or diagnose CRC (Table 1). Predictions have estimated the number of new CRC cases to increase to 3.2 million in 2040, primarily driven by a shifting lifestyle factor (4,5). Therefore, there is an unmet need for novel screening approaches with improved accuracy, less invasiveness and patient discomfort, lower risk of complications, reduced costs, and improved access.

Table 1.

Performance characteristics of recommended colorectal cancer screening modalities

Screening modality	Sensitivity	Specificity
Direct visualization tests
Colonoscopy	CRC: 0.92–0.99 (99) AA: 0.89–0.95 (100–102)	CRC: 0.90 (99) AA: 0.89 (100–102)
CT colonography	CRC: 0.86–1.0 (103) AA:0.67–0.94 (100,102,104–108)	CRC: Not reported (103) AA: 0.86–0.98 (100,102,104–108)
Flexible sigmoidoscopy (within reach)	CRC: 0.92–0.99 (99) AA: 0.90–0.92 (99)	CRC: 0.92 (99) AA: not reported (99)
Stool-based tests
gFOBT	CRC: 0.5–0.75 (109,110) AA: 0.06–0.17 (109,110)	CRC: 0.96–0.98 (109,110) AA: 0.96–0.99 (109,110)
FIT	CRC: 0.74 (110–118) AA: 0.23 (110–118)	CRC: 0.94 (110–118) AA: 0.96 (110–118)
sDNA-FIT	CRC: 0.93 (115,118–120) AA: 0.43 (115,118–120)	CRC: 0.85 (115,118–120) AA: 0.89 (115,118–120)

AA, advanced adenomas; CEA, carcinoembryonic antigen; CT, colonography; CRC, colorectal cancer; FIT, fecal immunochemical test; gFOBT, guaiac-based fecal occult blood test; sDNA, stool DNA.

Studies are underway to evaluate minimally invasive blood tests (Table 2; see Supplementary Table 1, http://links.lww.com/CTG/A940), and most fall short for screening purposes because of low sensitivity and/or specificity. Liquid biopsy is a novel category aiming to detect circulating cancer, including circulating tumor cells, circulating tumor DNA, and cell-free DNA (6). Currently, there are 35 liquid biopsy CRC trials ongoing to help establish their clinical utility (7); however, a potential problem is the need for sufficient tumor volume, which challenges its suitability for screening purposes. Developing a blood-based screening test targeting precancerous events would be an appealing approach.

Table 2.

Colorectal cancer blood-based liquid biopsy biomarkers

Biomarker target	Sensitivity	Specificity
Circulating tumor cells (CTCs)a
EpCAM enrichment	CRC: 0.86 (121) AA: 0.71 (121)	CRC: 0.95 (121) AA: 0.85 (121)
Circulating tumor DNA (ctDNA)
cg10673833 (MYO1-G)	CRC: 0.84–0.87 (122,123) AA: 0.33 (123)	CRC: 0.90–0.95 (122,123) AA: 0.67 (123)
VIM	CRC: 0.59–0.84 (124) AA: 0.45 (124)	CRC: 0.93–0.95 (124) AA: 0.95 (124)
BCAT1/IKZF1	CRC: 0.66 (125) AA: 0.06 (125)	CRC: 0.94–0.95 (125) AA: 0.94–0.95 (125)
Cell-free DNA (cfDNA)
SEPT9 (mSEPT9)	CRC: 0.37–0.88 (126) AA: 0.23 (126)	CRC: 0.77–0.99 (126) AA: 0.91 (126)
11 methylation markersb	CRC: 0.84 (127) AA: 0.77 (127)	CRC: 0.86 (127) AA: 0.83 (127)
SFRP1/SFRP2/SDC2/PRIMA1	CRC: 0.92 (128) AA: 0.89 (128)	CRC: 0.97 (128) AA: 0.87 (128)
Circulating microRNA (miRNA) markers
miR-760	CRC: 0.80 (129) AA: 0.70 (129)	CRC: 0.70 (129) AA: 0.62 (129)
miR-601/miR-760	CRC: 0.83 (129) AA: 0.72 (129)	CRC: 0.69 (129) AA: 0.62 (129)
miR-29a/miR-92a	CRC: 0.83 (130) AA: 0.73 (130)	CRC: 0.85 (130) AA: 0.80 (130)
Exosomal long noncoding RNA (lncRNAs) markers
UCA1	CRC: 1.0 (131)	CRC: 0.43 (131)
TUG1/UCA1	CRC: 0.93 (131)	CRC: 0.64 (131)
SNHG11	CRC: 0.93 (132)	CRC: 0.71 (132)
ZFAS1/SNHG11/LINC00909/LINC00654	CRC: 0.92 (132)	CRC: 0.83 (132)
Exosomal circular RNA (circRNA) markers
circPanel (hsa_circ_0001900/hsa_circ_0001178/hsa_circ_0005927)	CRC: 0.68–0.73 (133) AA: 0.84 (133)	CRC: 0.83–0.90 (133) AA: 0.70 (133)
hsa_circ_0007534	CRC: 0.92 (134)	CRC: 0.52 (134)
hsa-circ-0004771	CRC: 0.81 (135)	CRC: 0.83 (135)
Messenger RNA (mRNA markers)
hTERT	CRC: 0.98 (136)	CRC: 0.64 (136)
CK-19/CK-20/CEA/REG4/uPA/TIAM1 mRNA	CRC: 0.89 (137)	CRC: 0.88 (137)
CEA/CK20/survivin	CRC: 0.61 (138)	CRC: 0.76 (138)
Exosomal proteins markers
QSOX1	CRC: 0.74–0.88 (139)	CRC: 0.83–0.94 (139)
CPNE3	CRC: 0.68 (140)	CRC: 0.84 (140)
HSP60	CRC: 0.63 (141)	CRC: 0.95 (141)
Other protein and cytological markers
CEA	CRC: 0.11 (142) AA: 0.05 (142)	CRC: 0.98 (142) AA: 0.98 (142)
CA 19-9	CRC: 0.08 (142) AA: 0.03 (142)	CRC: 0.98 (142) AA: 0.98 (142)
TDGF 1/Cripto 1/CEA/ECM	CRC: 0.79–0.91 (143) AA: >0.60 (143)	CRC: 0.81–0.87 (143) AA: not reported (143)
Combination of biological targets
CECs/cfDNA	CRC: 1.0 (144) AA: 0.73–0.83 (144)	CRC: 0.89–0.91 (144) AA: 0.87–0.91 (144)
cfDNA and protein immunoassays	CRC: 0.91 (145) AA: 0.41 (145)	CRC: 0.94 (145) AA: 0.90 (145)
cfDNA/ctDNA	CRC: 0.90–0.96 (146–148) AA: 0.20 (149)	CRC: 0.94–0.97 (146–148) AA: not reported (149)

AA, advanced adenomas; BCAT1, branched-chain amino acid transaminase 1; CA, carbohydrate antigen; CEA, carcinoembryonic antigen; CK, cytokeratin; CPNE3, exosomal Copine III; CRC, colorectal cancer; DNA, desoxyribonucleic acid; ECM, extracellular matrix; EpCAM, epithelial cell adhesion molecule; HSP60, heat-shock protein 60; hTERT, human telomerase reverse transcriptase; IKZF1, IKAROS, family zinc finger 1; LINC00, long intergenic noncoding RNA; PRIMA1, Proline-rich membrane anchor 1; REG4, regenerating islet-derived family, member 4; RNA, ribonucleic acid; QSOX1, quiescin sulfhydryl oxidase 1; SDC2, Syndecan 2; SEPT9, septin 9; SFRP, secreted frizzled-related protein; SNHG11, small nucleolar RNA, host; TDGF-1, teratocarcinoma-derived growth factor 1; TIAM1, T-cell lymphoma invasion and metastasis 1; TUG1, taurine upregulated gene 1; UCA1, urothelial cancer-associated 1; uPA, plasminogen activator urokinase; VIM, vimentin; ZFAS1, zinc finger antisense 1.

^a Outcomes for CTC-based studies in CRC are frequently reported as detection rates, overall survival, and progression-free survival.

^b This panel contains cg00310855, cg01857475, cg01922936, cg11320449, cg11407741, cg11596863, cg15020425, cg22329423, cg24733262, cg25300584, cg26337020.

Key drivers for loss of colonic homeostasis and subsequent neoplasia formation include dysbiosis, (epi)genetic changes, and lack of proper immune surveillance intercalated with glycosylation changes. Aberrant protein glycosylation is a universal feature in various steps of malignant transformation and CRC development (8). In this review, we will summarize the sequence of events from normal colon mucosa to carcinoma and its correlations with glycoproteomic changes that might serve as potential biomarkers for CRC.

THE ADENOMA-TO-CARCINOMA SEQUENCE

(Epi)genetic changes during adenoma-to-carcinoma progression

Chromosomal instability, microsatellite instability, and serrated neoplasia are 3 mechanisms that drive tumorigenesis in the adenoma-to-carcinoma pathway (Figure 1) (9). Conventional adenomas stem from biallelic mutations in the tumor suppressor gene adenomatous polyposis coli (APC) in the chromosomal instability pathway but can also occur due to hypermethylation of the APC promoter (9). Mutated APC triggers the activation of the WNT signaling pathway (10,11), as a result the oncoprotein, β-catenin, encoded by the CTNNB1 gene, accumulates in the cytosol because of ineffective phosphorylation (11). β-catenin translocates to the nucleus (11). Inactivation of APC leads to activation of the proto-oncogene, Kirsten rat sarcoma viral homolog (KRAS) (12). β-catenin is aided by additional signaling pathways mediated by KRAS to drive gene transcription responsible for tumor growth and invasion (13). KRAS activates the downstream Raf-MEK-ERK pathway and PI3K/AKT signaling through mTOR (14). The loss of heterozygosity at chromosome 18q can lead to loss of function of tumor suppressor genes in this location including SMAD2/4 which are part of transforming growth factor-β (TGF-β) pathway (15). Loss of TP53 occurs late in tumorigenesis and can coincide with the transition of a large adenoma to adenocarcinoma (12).

Figure 1.

Genetic alterations and signaling pathways associated with the polyposis to the colorectal cancer sequence. The 2 main pathways involved in the normal colonic epithelium to colorectal carcinoma pathway are the classic tubular adenomas to colorectal carcinoma pathway (upper half) and the serrated polyps to serrated colorectal cancer pathway (lower half). Adapted from Figure 3, Kuipers et al (159) and Figure 2, Davies et al (160). Created with BioRender.com. AKT, protein kinase B; APC, adenomatous polyposis coli; BRAF, v-raf murine sarcoma viral oncogene homolog B1; CIN, chromosomal instability; CTNNB1, catenin-β1; KRAS, Kirsten rat sarcoma viral oncogene homolog; MAPK, mitogen-activated protein kinase; MSI, microsatellite instability; mTOR, mammalian target of rapamycin; NRAS, neuroblastoma ras viral oncogene homolog; P53, tumor protein 53; PI3K, phosphatidylinositol 3‐kinase; PI3KCA, phosphatidylinositol-4; 5-bisphosphate 3-kinase catalytic subunit-α; TGFβ, transforming growth factor-β; TGFBR2, TGFβ receptor 2; TP53, tumor protein 53 gene; WNT, Wingless-related integration site.

The microsatellite instability (MSI) pathway leads to the formation of the hypermutable phenotype characterized by DNA base pair mutations (9). Epigenetic disruptions including MLH1 hypermethylation and germline mutations in the DNA mismatch repair gene form the MSI phenotype (9). The MSI pathway can predict a favorable outcome and is seen in all patients with Lynch syndrome (12). The MSI phenotype can lead to aberrant methylation of the CpG dinucleotides leading to CpG island methylator phenotype. An APC or BRAF mutation can be the initiating event of adenoma formation in the MSI pathway (12). TGF-β receptor 2 gene mutation causes cell proliferation and is mutated in more than 90% of MSI colorectal tumors (9). There are a plethora of genes disrupted by MSI, including those that regulate proliferation, DNA repair, and apoptosis (9).

An important feature of the serrated pathway is the activation of V600E in the v-Raf murine sarcoma viral oncogene homolog B1 (BRAF) and subsequent MAPK pathway activation (9). Once acquiring a BRAF mutation, there are 2 ways serrated tumors form. One way is typically through sessile serrated adenomas when the pathway converges with the MSI pathway, which results in the MSI-H phenotype (10). Another way serrated tumors can form is through mutation of TP53, which leads to activation of other pathways including WNT, TGF-β, and epithelial-to-mesenchymal transition (9).

Role of host immunity in the adenoma-to-carcinoma sequence

The progression from normal tissue to carcinoma is accompanied by a series of phenotypic and functional changes of immune cells at the site of the lesion. CRC develops slowly from single crypt lesions to adenoma and carcinoma through a multistep process that is both controlled and shaped by the immune system (Figure 2). Transformed epithelial cells release danger signals, a group of molecules that are chemically unrelated but promote sterile inflammation. These damage-associated molecular patterns include proteins (e.g., high-mobility group box 1 protein, S100 proteins, and heat shock proteins) that are typically found inside the cell but are released under stress conditions or cell death (16). Damage-associated molecular patterns lead to the recruitment of immune cells that can eliminate damaged cells and induce regression of lesions (immunosurveillance) (17,18). However, premalignant cells develop diverse strategies to evade host control while increasing in size and progressing to a high degree of dysplasia. This state of relative dormancy (equilibrium), in which cancer expansion is kept at bay by the immune system, may extend for a very long time (19). A progressive shift of the immune landscape from an antitumor state to a protumor state, during which immune cells undergo marked phenotypic and functional changes, facilitates the progression of premalignant lesions to cancer (17,18). In this escape phase, the premalignant cells acquire the ability to escape immune control, undergo uncontrolled invasive growth, and develop into CRC.

Figure 2.

Mechanisms of tumor initiation and promotion in colorectal cancer influence the pool of secreted proteins. Tumor-initiating events that transform normal intestinal epithelial cells by spontaneous mutations, environmental mutagens, and bacteria. Cytokines produced by innate and adaptive immune cells, stromal compartment, and tumor cells exhibit pleiotropic roles, which depend on the developmental stage of the tumor. Cytokines and other factors produced in the tumor might enter circulation and influence the composition of glycoproteins secreted by the liver and by circulating blood cells. Created with BioRender.com.

The adenoma microenvironment contains immune cells that engage in complex interactions with premalignant cells and exhibit both antitumor and protumor functions (19–21). These immune cells change in number and function in the transition from adenoma to carcinoma. For instance, an increased population of T lymphocytes, including CD8+ cytotoxic T lymphocytes (CTLs) and CD4+ T helper cells, has been reported in adenoma tissues, mostly distributed in the stromal region and infiltrated into the adenomatous epithelium (22–24). CTLs can recognize and eliminate transformed cells, whereas CD4+ T helper cells play critical roles in regulating host immune response and promoting CTL function (25). Strikingly, CD4+ T cells progressively increase across the normal mucosa-adenoma-carcinoma sequence, whereas CTLs increase in premalignant adenoma and decrease in the transition to CRC (22,26). The antitumor capacity of CTLs can be further affected by exhaustion and by suppression driven by regulatory T cells (Tregs). T-cell exhaustion is a dysfunctional state that occurs in response to chronic antigen stimulation and is characterized by reduced cytokine production and increased expression of inhibitory receptors; it is detected only in CRC and not in the premalignant state (27). Similarly, Tregs are significantly increased in adenoma tissues compared with control tissues and are even higher in CRC tissues (28,29).

High T-helper 17 (T_H17) cell activity contributes to the development of malignant lesions (30). T_H17 cells are the dominant source of interleukin-17A (IL-17A), a cytokine that stimulates proliferation, survival, and mobilization of stromal fibroblasts through the upregulation of the TGF-β receptor that enhances the expression of profibrotic genes. The density of IL-17A-positive T_H17 cells is increased in colorectal adenoma tissues (31,32). Given that TGF-β is also an upstream stimulator of T_H17 cell differentiation and increased expression levels of both TGF-β and IL-17A occur in patients with CRC, TGF-β may work in synergy with T_H17/IL-17A on the progression of CRC (33).

Myeloid cells are also found in premalignant lesions and change in number during the transition from normal tissue to carcinoma. Tumor infiltration by CD68⁺ macrophages increases progressively as colorectal adenomas arise, grow, and become dysplastic; this trend continues in the early stages of CRC (22). Progression from adenoma to cancer is accompanied by the emergence of mechanisms that suppress the generation, maturation, and function of dendritic cells to escape immune recognition and elimination, resulting in lower antigen presentation and lower numbers of these cells in both adenoma and CRC tissues (34).

In summary, the progression from normal tissue to adenoma and carcinoma is accompanied by a series of phenotypic and functional changes of the immune cells in the site of the lesion, supported by alterations in associated cytokine networks. We propose that cytokines, metabolites, and other factors produced in the lesion may travel to the periphery and exert global effects on the liver and circulating leukocytes.

ABERRANT GLYCOSYLATION DURING THE ADENOMA-TO-CARCINOMA SEQUENCE

Introduction to protein glycosylation

Protein glycosylation is the most abundant and complex post-translational modification common to all eukaryotic cells and involves an enzymatic process in which carbohydrate structures (glycans) are attached to proteins, mainly through nitrogen (N) and oxygen (O) linkages (35). It is estimated that 85% of serum proteins are glycosylated (36) and that specific glycosylation determines a wide range of physiological events, including protein folding and trafficking, cell-cell and cell-matrix interactions, cellular differentiations, and the immune response (37–42). Glycosylation significantly amplifies the proteome by producing diverse protein isoforms that demonstrate a variety of properties and functions (43–45). Figure 3 shows a schematic overview of N-linked and O-linked protein glycosylation.

Figure 3.

A schematic overview of N-linked and O-linked glycosylation. Glycosylation is a post-translational modification where sugar moieties are attached to proteins, affecting protein structure, conformation, and function. N-glycosylation and O-glycosylation are the 2 main types of glycosylation. Glycosylation changes have been associated with physiological and pathological events. The analysis helps identify novel biomarkers for diagnosis, prognosis, and disease monitoring. Created with BioRender.com.

Aberrant glycosylation is described as a hallmark of cancer and a critical feature during malignant transformation and tumor development (46,47). Aberrant N-glycan structures in cancer cells are associated with malignant transformation, tumor invasiveness, and metastatic disease (48–52). Importantly, aberrant glycosylation of cancer cells greatly influences the interaction between cancer cells and the immune system, particularly immunosurveillance and immunoediting. The aberrant glycosylation observed during cancer development includes increased branching of N-glycans, higher density of O-glycans, incomplete synthesis of glycans, neosynthesis, increased sialylation, and increased fucosylation.

Cellular and tissue glycosylation during CRC development

The dynamically changing glycoproteomic profiles in the colonic crypt result from the concordant enzymatic activity of glycosyltransferases and glycosidases, orchestrating the cellular transformations from normal mucosa to adenoma and ultimately to CRC. During this transition, predominantly N-linked and O-linked glycans are added to or removed from tissue proteins, thereby constantly changing the glycocalyx of cellular surfaces and mediating the local immune responses in the crypt. One example of an environmental trigger influencing local glycosylation is red meat consumption (53). This leads to aberrant cell surface glycosylation due to metabolic incorporation of the red meat-derived sialic acid (N-glycolyl neuraminic acid [Neu5Gc]) (54). Humans are unable to synthesize Neu5Gc because of a mutation in the Cmah gene (55,56); however, dietary intake of Neu5Gc from red meat leads to its incorporation in glycoconjugates (57,58). Such Neu5Gc-containing glycans are targeted by naturally circulating anti-Neu5Gc antibodies, which are present in all humans (59,60). This leads to an inflammatory response that has been shown to be associated with CRC risk in 2 large independent human cohorts (54,61,62).

During malignant transformation, changing glycoproteomic crypt profiles specifically influence the host immune response. For instance, increased branching of N-glycans is used by transforming cells to escape immune recognition, instructing the creation of immunosuppressive networks through inhibition of interferon gamma (53). In addition, mutations in the guanosine diphosphate-mannose-4,6-dehydratase gene that is involved in the synthesis of fucosylated proteins have been identified in human CRC tissue (63). Cells deficient in GDP-mannose-4, 6-dehydratase, therefore lacking fucosylation, evade natural killer cell-induced apoptosis through the death receptors TNF-related apoptosis-inducing ligand (63). In a large clustered regularly interspaced short palindromic repeats screening, it was demonstrated that the enzyme B3GNT2, responsible for the synthesis of polylactosaminylated N-glycans, can glycosylate various cell surface receptors on tumor cells to prevent subsequent killing by the cytotoxic T cells (64). Furthermore, CRC cells express high levels of sialic acid-containing glycans that are known engagers of the immune-suppressing receptors called Siglecs (sialic acid-binding immunoglobulin-like lectins) (65). Siglec/sialoglycan interaction can compromise dendritic cell activation and antigen presentation, tumor-associated macrophage polarization, and CD8 T-cell antitumor effector functions (66–68). Not surprisingly, many studies have pointed to the blocking of Siglec/sialoglycan interaction as a new potential immune checkpoint blockade therapy during cancer (69). Finally, recent work has provided an insight into the mechanism underlying the low response of INF-y therapy in patients with CRC. Krug et al (70) have elucidated the role of bisecting N-glycans on INFyRa, demonstrating that a decrease in INFyRa bisection leads to receptor degradation and resistance to INF-y therapy, which could be salvaged by pharmacological reconstitution of bisection through administration of all-trans retinoic acid.

A wide range of observations have further solidified tissue glycosylation alterations to be a hallmark of CRC development, including increased branching of N-glycans, higher density of O-glycans, incomplete glycan synthesis, glycan neo-synthesis, increased sialylation, and increased fucosylation (8). A major N-glycan branching structure related to cancer formation is the complex β1,6-GlcNAc-branched N-glycan (51). The aberrant overexpression is catalyzed by the N-acetylglucosaminyltransferase V (GnT-V) enzyme, implicated in inducing pro-malignant, pro-invasive, and pro-metastatic cancer phenotypes (48,49,71). Interestingly, upregulation of GnT-V is implicated early in colon adenoma progression and is detectable in immunohistochemical staining of aberrant crypt loci, the earliest morphological lesions visible in colonic carcinogenesis (72). Mechanistically, it has been shown that GnT-V promotes colon cancer stem cell self-renewal and tumorigenesis potential through regulation of Wnt/β-catenin signaling (72). The WNT/beta-catenin pathway is also known to interact with the P13K/AKT pathway to promote CRC progression (73). The WNT pathway is also known to interact with the Notch pathway through Jagged1 (74) to promote stem cell renewal and differentiation. Similarly, the WNT pathway interacts with the TGF-beta (75), MAPK/ERK (76), Hedgehog (77), and ND-KappaB pathway (78), leading to various hallmarks of cancer. Removal of these GnT-V-synthesized glycans renders CRC cells more susceptible to the immune system and improves cancer immunotherapy (79). GnT-V has also been implicated in metastasis development by modifying proteins involved in extracellular matrix degradation, cell adhesion, and cell motility, helping the spread of metastatic cancer cells (80,81). One of the best described epitopes in metastatic disease are sialyl Lewis antigens, namely sialyl Lewis X and A, which are key in aiding tumor extravasation through the endothelium by interactions with E-selectin (82–84).

Changes in serum glycosylation during CRC development

The adenoma-to-carcinoma sequence is associated with dynamically changing glycosylation profiles in colonic crypts and associated host immune responses that, in turn, influence systemic immune responses, including glycosylated acute-phase proteins and plasma-derived antibodies. For purposes of detection, immunoassays for circulatory glycoproteins have been tried with mixed results, including carcinoembryonic antigen, CA19-9, and CA-125 (85,86). Although aiding in staging evaluation and monitoring after treatment, they lack sensitivity and specificity for proper screening. Table 3 summarizes the most relevant markers to date.

Table 3.

Immunoassays that measure glycosylation-related changes in CRC

Assay	GlycoEpitope	Changes in CRC	Clinical utility
CEA (150–152)	CD66 family of glycoproteins	Elevated in certain cases of CRC	Limited or no utility
CA19-9 (153–155)	Sialyl-LewisA	Elevated in metastatic CRC	Limited or no utility
CA-125 (156–158)	Heavily O-glycosylated MUC16	Unclear or mixed results	May be useful for prognostication

CEA, carcinoembryonic antigen; CRC, colorectal cancer; MUC16, Mucin-16.

Although glycosylation immunoassays have limitations, mass spectrometry (MS) has opened new avenues. Owing to the simplicity of acquiring blood, numerous studies have evaluated aberrant glycosylation, especially N-glycosylation as disease biomarkers (87). A study of the N-glycome in patients with CRC demonstrated a strong association with branched and poly-LacNAc elongated N-glycans, normalizing after treatment (88). Additional serum studies on N-glycan profiling demonstrated that increased fucosylation and sialylation on complement C3 and kininogen-1 may aid in CRC detection (89). Studying IgG N glycome profiles in CRC, differentially expressed levels of N-glycosylation were shown for normal individuals compared to those with CRC (90). Moreover, using matrix-assisted laser desorption/ionization-MS in patient sera, a positive correlation between multiantennary, sialylated N-glycans was shown during cancer progression while biantennary core-fucosylated N-glycans correlated negatively (91). Interestingly, the authors not only showed accuracy for rightly classifying CRC but also for advanced adenomas (AA), which could potentially introduce a paradigm shift for CRC screening. This was recently addressed in a study using ultra performance liquid chromatography and demonstrating distinct serum IgG N-glycan profiles for AA and CRC (92). Another group studying 34 serum glycans reported that 9 of both hybrid-type and multiantennary glycans were differentially expressed between CRC and AA (93). Not only glycosylated proteins but also glycan-binding proteins such as galectin-2, galectin-3, galectin-4, and galectin-8 are found to be elevated in the serum from patients with CRC. For instance, high levels of galectin-2 positively correlate with high CRC mortality (94). This study demonstrated that galectins can promote the adhesion of cancer cells to the blood vascular endothelium, therefore contributing to cancer metastases (94).

In summary, this suggests that glycosylation events associated with CRC development can be readily observed at the periphery through aberrant glycosylation observed on circulating glycoproteins. The question remains on how early does developing colonic neoplasia result in aberrant serum glycosylation profiles, mainly composed of liver-secreted glycoproteins and B-cell-derived antibodies. This will need to be addressed, but there seems to be sufficient evidence to justify larger prospective studies on the use of blood-based glycan profiles for the purpose of AA/CRC detection and screening. It also remains to be seen whether the glycoproteomic changes in circulation differ by the pathway of CRC development and/or by individual mutations present in these lesions. Nevertheless, it is an important pursuit particularly in the context of biomarkers and surrogate end points for personalized therapies.

CONCLUSION

To improve CRC screening, there is a need to develop biomarkers that are associated with precancerous and early cancerous events in the colonic crypt. We reviewed the early events taking place during the adenoma-to-carcinoma sequence and introduced critical protein glycosylation phenomena associated with the changing crypt microenvironment. The host immune response plays a central role in orchestrating the process from single crypt lesions to carcinoma with aberrant protein glycosylation being a key factor, both locally and on circulating glycoproteins. It is not until recently that the enormously complex field of glycosylation can be studied, primarily because of the availability of the appropriate tools (95). Machine learning models (e.g., Support Vector Machines and Random Forest) can decipher the vast amount of complex data coming from, for instance, liquid chromatography-MS (96–98). Large-scale prospective clinical studies are currently underway to evaluate the clinical validity of glycosylation markers and determine their clinical utility and become an additional tool in the armamentarium for CRC screening.

CONFLICTS OF INTEREST

Guarantor of the article: Daniel W. Hommes, MD, PhD.

Specific author contributions: D.C., D.W.H.: study planning; D.C., C.G., F.S., C.D., T.C., F.S., D.W.H.: data collection and interpretation; D.C., C.G., F.S., C.D., T.C., F.S., D.W.H.: manuscript drafting.

Financial support: None to report.

Potential competing interests: All authors are employed at InterVenn BioSciences, a life sciences company that specializes in glycomic analysis and biomarker development and holds multiple patents in this field.

Supplementary Material

ct9-14-e00592-s001.docx ^{(66.6KB, docx)}