Altered expression of glycan genes in cancers induced by epigenetic silencing and tumor hypoxia: Clues in the ongoing search for new tumor markers

Abstract

(Cancer Sci 2010; 101: 586–593)

The glycan molecules that preferentially appear in cancers are clinically utilized as serum tumor markers. The exact reason, however, why glycans are useful as tumor markers remain elusive. Here, we will summarize lessons learned from well‐established cancer‐associated glycans, and propose strategies to develop new cancer markers. Our recent results on cancer‐associated glycans, sialyl Lewis A and sialyl Lewis X, indicated that the repressed transcription of some glycan genes by epigenetic silencing during early carcinogenesis, and the transcriptional induction of some other glycan genes by tumor hypoxia accompanying cancer progression at locally advanced stages, are two major factors determining cancer‐associated glycan expression. Multiple genes are involved in glycan synthesis, and epigenetic silencing of a part of such genes leads to accumulation of glycans having truncated incomplete structures, which are readily detected by specific antibodies. Glycans are very unique and advantageous as marker molecules because they are capable of reflecting epigenetic silencing in their structures. Transcriptional induction of some glycan genes by tumor hypoxia at the later stages produces further glycan modifications, such as an unusual increase of the N‐glycolyl sialic acid residues in the glycan molecules. The entire process of malignant transformation thus creates abnormal glycans, whose structures reveal the effects of both epigenetic silencing and tumor hypoxia. The second advantage of a glycan marker over a proteinous marker is that they can reflect the plurality of genetic anomalies in a singular molecule, as it is synthesized by the cooperative action of multiple genes. Glycans are sometimes covalently bound to well‐known cancer‐associated proteins, such as CD44v, and this eventually contributes to a high cancer specificity and functional relevancy in cancer progression.

Cell surface glycans undergo drastic changes during malignant transformation.^{( 1} , ² , ³ , ⁴ , ⁵ , ⁶ , ^{7 )} The glycan molecules preferentially appearing on cancer cells are utilized in cancer diagnosis as serum tumor markers such as CA19‐9.^{( 8 )} CA19‐9 serum diagnosis is performed approximately 10 000 000 times every year for patients with cancers in Japan, and is one of the most frequently used serum tumor markers.^{( 8 )} Recently great advances have been made in therapeutic modalities for advanced cancers. Accordingly, serum tumor markers have become increasingly important clinically, because they are quite useful for monitoring recurrence after surgery and evaluating the effectiveness of chemotherapy and radiotherapy; albeit their use is limited for early diagnosis of small cancers. There is an urgent need to develop novel serum markers for cancers having spectra not yet covered by currently available marker molecules. Most serum markers currently in use are carbohydrate epitopes in their biochemical nature. For instance, the biochemical entity of the tumor marker CA19‐9 is a glycan epitope called sialyl Lewis A. The exact reason, however, why glycans are so useful as tumor markers still remains elusive. Here, we summarize the lessons learned from established cancer‐associated glycans, and propose a novel strategy to develop new glycan markers for cancers.

Comparison of Glycans on Normal and Malignant Cells

In the initial stages of study of cancer‐associated glycans, it has generally been assumed that the transcription of some genes involved in their synthesis must be somehow enhanced in cancer cells compared to normal cells, and this should be the reason for preferential expression of such glycans in cancers. Many papers have been published in which researchers looked for the genes involved in glycan synthesis that exhibit an increased transcription in cancers.^{( 9} , ¹⁰ , ¹¹ , ¹² , ^{13 )}

CA19‐9 is utilized as a serum marker mainly for cancers of digestive organs, especially in the pancreas and biliary tract. Expression of CA19‐9, the sialyl Lewis A glycan, is significantly increased in cancer tissues compared to non‐malignant epithelial cells, when studied by routine immunohistochemical techniques (Fig. 1). In contrast, a glycan epitope very similar to sialyl Lewis A, namely disialyl Lewis A, turned out to be preferentially expressed in normal epithelial cells compared to cancer cells (Fig. 1).^{( 14} , ¹⁵ , ^{16 )} This paradoxical finding has cast suspicion on the validity of the old notion that enhanced transcription of some genes for glycan synthesis is the major cause for cancer‐specific expression of sialyl Lewis A, because essentially all genes involved in the synthesis of sialyl Lewis A are predicted to be the same as those involved in the synthesis of disialyl Lewis A, except a gene for an α2‐6 sialyltransferase. Instead, this finding suggests that the reduced expression of disialyl Lewis A determinant upon malignant transformation could be the reason for induction of sialyl Lewis A expression in cancers.^{( 16 )}

Figure 1.

 Examples of malignant transformation‐associated glycan transition of epithelial cells. (a) Sialyl Lewis A glycan transition upon malignant transformation. Cancer cells preferentially express the cancer‐associated glycan, sialyl Lewis A, whereas non‐malignant epithelial cells express the normal glycan, disialyl Lewis A, suggesting an impairment of βGlcNAc: α2‐6 sialylation by epigenetic silencing occurring during malignant transformation. Adapted from Miyazaki et al., 16 with permission. (b) Sialyl Lewis X glycan transition upon malignant transformation. Cancer cells preferentially express the cancer‐associated glycan sialyl Lewis X, while non‐malignant epithelial cells express the normal glycan, sialyl 6‐sulfo Lewis X, suggesting an impairment of sulfation at C6 position of βGlcNAc by epigenetic silencing occurring during malignant transformation. Adapted from Izawa et al., 20 with permission. Typical distribution patterns shown were obtained by immunohistochemical staining using specific anti‐glycan antibodies of consecutive sections prepared from colon cancer tissues. Ca, cancer cells; N, non‐malignant epithelial cells.

Another glycan epitope, called sialyl Lewis X, is also utilized as useful serum tumor marker for cancers in the lung, breast, and ovary under the common names of SLX (sialyl SSEA‐1), CSLEX, or NCC‐ST‐439, covered by national medical insurance in Japan.^{( 17} , ¹⁸ , ^{19 )} The sialyl Lewis X glycan is also preferentially expressed in cancer cells compared to non‐malignant epithelial cells. Later, a glycan epitope that is very similar to sialyl Lewis X, namely sialyl 6‐sulfo Lewis X, was noticed to be preferentially expressed in normal epithelial cells compared to cancer cells (Fig. 1).^{( 20 )} Again, essentially all genes involved in the synthesis of sialyl Lewis X are predicted to be the same as those involved in the synthesis of sialyl 6‐sulfo Lewis X, except for the genes engaged in its sulfation. Thus this finding supports the idea that the reduced expression of sialyl 6‐sulfo Lewis X causes induction of sialyl Lewis X expression in cancers.^{( 20 )}

Epigenetic Silencing as an Initial Glycan Transition Mechanism in Cancers

It is notable that both glycans preferentially expressed in normal epithelial cells, disialyl Lewis A and sialyl 6‐sulfo Lewis X, have structures more complex than those of the glycans carried by cancer cells, sialyl Lewis A and sialyl Lewis X. This finding suggests that the expression of complex glycan structures is in some way impaired upon malignant transformation. This is compatible to the classical scheme of the cancer‐associated alteration of glycan structures, which is known as the “incomplete synthesis” theory.^{( 4} , ²¹ , ^{22 )} The structural studies on glycolipids in cancer tissues in the main indicate that malignant cells tend to accumulate glycans having simpler structures than those found in normal cells.

The structural difference between sialyl Lewis A and disialyl Lewis A is the presence of one extra sialic acid residue attached to the C‐6 position of βGlcNAc in the carbohydrate structure of the latter glycan epitope (Fig. 1). This implies that α2‐6 sialylation at the GlcNAc moiety was somehow impaired in cancer cells compared to non‐malignant cells. Later it was shown that transcription of the sialyltransferase gene responsible for α2‐6 sialylation of the βGlcNAc moiety is significantly decreased in cancers compared to normal cells due to epigenetic silencing (Fig. 2a).^{( 16 )}

Figure 2.

 Clinical application of sialyl Lewis A/disialyl Lewis A glycan transition. (a) Real‐time RT‐PCR analyzes of transcription of the βGlcNAc: α2‐6 sialyltransferase gene responsible for sialyl Lewis A/disialyl Lewis A glycan transition in colon cancer tissues. Note that its transcription is significantly reduced in cancer cells in patients at relatively early as well as advanced stages. RNA samples were prepared from the cancer tissues (Ca) and non‐malignant colonic mucosa (N) of the same patients. (b) Ratio of serum concentrations of sialyl Lewis A and disialyl Lewis A glycans in patients with various disorders. Note that the ratio is high in sera from patients with malignant disorders, while it remains low in sera from patients with benign disorders, providing information for differential diagnosis of malignant and benign disorders. Adapted from Itai et al., 14 and Miyazaki et al., 16 with permission.

The disialyl Lewis A glycan, as well as sialyl Lewis A, is released from cells, and is detectable in sera of patients using appropriate specific antibodies. As one might readily expect, the levels of disialyl Lewis A glycan are elevated in the sera of patients with benign disorders, since the glycan is preferentially expressed in non‐malignant cells, and is useful as a marker for tissue injuries occurring in benign diseases. While the sera of patients with cancers also exhibit moderately elevated levels of the normal glycan, as the growth of cancers is accompanied by the destruction of the surrounding non‐malignant tissues, the ratio of sialyl Lewis A glycan to disialyl Lewis A glycan provides precious information for the differential diagnosis of malignant and non‐malignant diseases, because the ratio is high in malignancy, yet low in non‐malignant disorders (Fig. 2b).^{( 8} , ¹⁴ , ^{15 )}

Similarly, the structural difference between sialyl Lewis X and sialyl 6‐sulfo Lewis X involves the addition of one sulfate residue to the C‐6 position of GlcNAc in the carbohydrate structure of the latter glycan epitope (Figure 1), suggesting an impairment of 6‐sulfation at the GlcNAc moiety in cancer cells. Recently it was found that the transcription of several genes involved in glycan sulfation is repressed in cancer cells but not in non‐malignant epithelial cells, including those for 6‐sulfotransferase, PAPS synthase, and sulfate transporter.^{( 23 )} Reduced transcription of the gene for a sulfate transporter, DTDST, was suggested to play a dominant role in the decreased expression of sialyl 6‐sulfo Lewis X and induction of sialyl Lewis X expression in colon cancers.

Glycan Markers are Capable of Reflecting Epigenetic Silencing

Epigenetic silencing of genes occurs even in a very early stage of carcinogenesis, and is known to be one of the key characteristics of cancer cells.^{( 24} , ²⁵ , ²⁶ , ^{27 )} With genes encoding proteins, however, epigenetic silencing causes a decrease or the extinction of their expression, and applying the detection of its decrease for serum marker analyses is no easy matter. Accordingly, methods for detection of epigenetic silencing mostly require extraction of nucleic acids from cell or tissue samples, which is less convenient than serum diagnosis.

In this context it is notable that serum glycan markers can reflect epigenetic silencing. In the case of glycans, epigenetic silencing of part of the genes involved in normal glycan synthesis leads to synthesis of cancer‐associated glycans having truncated incomplete structures, which have distinct antigenic epitopes readily detected by specific reagents such as monoclonal antibodies. This is because a glycan is synthesized by cooperative action of multiple genes, whereas a protein is usually encoded by a single gene. The distinct capability of reflecting epigenetic silencing gives glycans a significant advantage over proteins when they are applied to tumor markers.

A new strategy for finding novel glycan markers for cancers may be to widely screen the genes involved in glycan synthesis that are repressed by epigenetic silencing upon malignant transformation, and to deduce from the results any possible changes that may occur in the structures of normal glycans. Several genes involved in glycan synthesis are known to be inactivated by epigenetic silencing in cancers. For instance, a gene for heparan sulfate 3‐O‐sulfotransferase, 3‐OST‐2, is known to exhibit a marked DNA methylation, and is utilized as a marker gene for DNA methylation in cancers.^{( 28} , ²⁹ , ^{30 )} The glycosyltransferase genes for synthesis of ABO and Sd^a blood group substances are also known to frequently display significant DNA methylation in cancers.^{( 31} , ^{32 )} Not DNA methylation but histone modification is proposed to figure heavily in the silencing of the DTDST gene in colon cancers.^{( 33 )} Silencing occurring at core‐1 elongation leads to the appearance of Tn/sialyl Tn glycans, for which only genetic, but not epigenetic, silencing due to mutations in a unique chaperon gene has been so far reported.^{( 34 )}

Physiological Functions of Normal Complex Glycans

Epigenetic silencing is known to occur not only in cancers but also in some non‐malignant diseases such as inflammation. Such benign disorders sometimes exhibit epigenetic silencing of a set of genes very similar to that silenced in cancers, and this limits the validity of the above strategy for finding novel tumor‐specific markers.

The complex glycans preferentially expressed on normal epithelial cells were shown to serve as specific ligands for immunosuppressive receptors carried by tissue macrophages, natural killer (NK) and CD8⁺ T cells. The disialyl Lewis A glycan was shown to react to siglec‐7 and ‐9, while sialyl 6‐sulfo Lewis X glycan is known to be recognized by siglec‐7.^{( 16} , ^{35 )} So far, only normal glycans serve as ligands for siglecs, and cancer‐associated glycans such as sialyl Lewis A and sialyl Lewis X do not bind to these siglecs. One of the physiological functions of normal glycans seems to be the protection of epithelial cells from excess activation of immune cells in mucous membranes, as siglec‐7 and ‐9 have ITIM motives in their cytoplasmic domains, which inhibit signal transduction elicited by a variety of immunological stimuli. Loss of such immuno‐inhibitory normal glycans through epigenetic silencing will alter immune homeostasis of the mucous membranes, and favors progression of cancers, as chronic inflammation is known to facilitate malignant transformation of epithelial cells. Many immune cells expressing siglec‐7 and ‐9 are shown to be present in normal mucosal membranes.^{( 35 )}

Interestingly, epigenetic silencing of the DTDST gene facilitates proliferation of cancer cells. We recently showed that induction of DTDST transcription in the cells transfected with tetracycline‐inducible vectors significantly suppresses, while interruption of its transcription markedly enhances, proliferation of cancer cells.^{( 33 )} Transcription of a gene tends to be constitutively repressed in cancers, when its epigenetic silencing is advantageous for promoting cancer progression. Silencing of such genes is expected to be preserved more frequently in cancers than in non‐malignant disorders such as inflammation, something well worth considering in the ongoing search for novel glycan markers.

Further Glycan Alteration Introduced by Tumor Hypoxia in Locally Advanced Stages

Because of the uncontrolled growth of cancer cells, hypoxic areas appear in expanding tumor nests at the locally advanced stage of cancers.^{( 36} , ³⁷ , ^{38 )} Tumor hypoxia further accelerates abnormal glycan expression in cancer cells due to transcriptional induction of a set of genes involved in glycan synthesis by hypoxia (Fig. 3).^{( 39} , ^{40 )} For instance, genes for some sialyltransferases, fucosyltransferases, and sugar transporters are induced by hypoxia, which are involved in the synthesis of sialyl Lewis A and sialyl Lewis X glycans.^{( 40 )} This leads to a marked increase in expression of these glycan epitopes in cancers (Fig. 3).

Figure 3.

 Schematic illustration of hypoxia‐induced cancer progression and its effects on glycan expression. (a) Hypoxia‐resistant cancer clones, which acquired enhanced cell mobility and cell‐adhesive glycan expression, propagate in hypoxic area of cancer nests. They eventually occupy the entire cancer cell nests, and undergo vigorous vascular infiltration. This process accompanies acceleration of abnormal glycan production in cancer cells through transcriptional induction of genes involved in glycan synthesis. (b) Enhanced expression of cancer‐associated glycans, sialyl Lewis A/X, by hypoxia in cultured cancer cells. (c) Examples of genes involved in glycan synthesis showing hypoxia‐induced transcription. (d) Real‐time RT‐PCR analyzes of transcription of hypoxia‐dependent glycan genes in colon cancer tissues. The transcription of glycan genes, which showed a clear induction by hypoxia in experiments in vitro, is elevated also in actual in vivo cancer tissues prepared from surgical specimens. Note that their transcription tends to increase preferentially in patients at relatively advanced stages, because the acquisition of hypoxia resistance is a relatively late event in cancer progression. The RNA samples were prepared from the cancer tissues (Ca) and non‐malignant colonic mucosa (N) of the same patients. (b–d) Adapted from Koike et al., 40 with permission.

Some qualitative changes of glycans are also induced by tumor hypoxia. A part of the sialic acid residues in glycans is replaced by N‐glycolyl sialic acid in cancers, while normal glycans usually carry N‐acetyl sialic acid residues. This turned out to be due to induction of a gene for sialic acid transporter, Sialin, in cancers by tumor hypoxia.^{( 41 )} As sialic acid residues are usually located at the outermost terminus of a glycan, this alteration produces a significant change in antigenic epitope, and the glycans carrying N‐glycolyl sialic acid is detectable with specific reagents. The successful generation of antibodies that can discriminate between N‐glycolyl and N‐acetyl sialic acid residues has already been reported for several ganglioside epitopes.^{( 42} , ⁴³ , ^{44 )}

In locally advanced tumors, cancer cells must cope with the hypoxic environment to survive and proliferate, and some cancer cell clones having hypoxia‐resistant characteristics appear through accumulation of genetic anomalies. Such cancer cells have a growth advantage over other cancer cell clones, and will ultimately occupy whole cancer cell nests (Fig. 3). It is postulated that tumors may not grow more than 3 mm in diameter, if the cancer cells fail to acquire hypoxia resistance. The common feature of hypoxia‐resistant cancer cells is sustained activation of HIF, the transcription factor that induces expression of genes required to adapt to, or cope with, the hypoxic environment. HIF also induces transcription of several genes involved in carbohydrate synthesis, and accelerates abnormal glycan expression. The abnormal glycans thus appearing serve as stable markers for advanced cancer cells, because the hypoxia‐resistant cancer cells have an irreversibly sustained HIF activity irrespective of actual oxygen tension in their environment. For instance, such cancer cells will frequently express N‐glycolyl sialyl Lewis A and N‐glycolyl sialyl Lewis X (Fig. 4). It can be proposed as a strategy for developing novel glycan markers for cancers to widely screen the genes induced by tumor hypoxia, and to deduce from the results any possible glycans that could exhibit enhanced expression in cancers.

Figure 4.

 A scheme illustrating the stepwise accumulation of structural anomalies in cell surface glycans during carcinogenesis and cancer progression. Normal epithelial cells express a glycan disialyl Lewis A having a complex carbohydrate structure. Epigenetic silencing of glycan genes occurring during carcinogenesis leads to accumulation of glycans having a truncated incomplete structure, sialyl Lewis A. Acquisition of hypoxia resistance in locally advanced cancers confers a marked increase of its expression and eventually results in the appearance of N‐glycolyl sialyl Lewis A having an abnormal sialic acid residue in its structure, which is expected to be a good marker for hypoxia‐resistant cancer cells. On the other hand, ischemic changes occurring in non‐malignant epithelial cells will lead to an increase of the normal glycan, disialyl Lewis A, and will facilitate its modification with N‐glycolyl sialic acid, that is an accumulation of N‐glycolyl disialyl Lewis A glycan, which may serve as a marker for benign ischemic diseases.

Different Glycans are Induced in Normal Cells and Cancers Under Hypoxia

HIF induces gene transcription also in non‐malignant cells under a hypoxic condition, such as ischemic disorders. In hypoxic non‐malignant tissues, HIF induces a set of genes very similar to that induced in cancers, and this would limit the validity of the above strategy. If production of a protein is induced in hypoxic cancer cells, the same protein would be produced also by non‐malignant cells under hypoxic conditions, and such a protein marker cannot be tumor‐specific.

However, different glycans are clearly induced by hypoxia in cancer cells and non‐malignant tissues. Genes for some sialyltransferase, fucosyltransferase, and sugar transporters are induced by hypoxia, and this leads to enhanced expression of sialyl Lewis A and sialyl Lewis X in cancer cells. In contrast, hypoxia‐induced transcription of the same set of genes will lead to enhanced expression of disialyl Lewis A and sialyl 6‐sulfo Lewis X in non‐malignant cells, because these genes are commonly involved in synthesis of both normal ‐ as well as cancer‐associated glycans.

Likewise, hypoxia‐induced Sialin transcription will lead to accumulation of N‐glycolyl sialyl Lewis A and N‐glycolyl sialyl Lewis X in cancer cells, while ischemic diseases will induce an accumulation of N‐glycolyl disialyl Lewis A and N‐glycolyl sialyl 6‐sulfo Lewis X in non‐malignant cells (Fig. 4). Thus, distinct sets of glycans are expected to result from hypoxia in cancer and normal cells, respectively and this could well be another advantage of glycans as tumor marker molecules over proteins.

Again, the background for this advantage is that a glycan is synthesized through cooperation of multiple genes, and therefore is capable of reflecting multiple genetic anomalies, whereas a protein is usually encoded by a single gene. An increase of single glycan, such as N‐glycolyl disialyl Lewis A, reflects multiple genetic events occurring in cancer cells, that is epigenetic silencing of a certain gene, and hypoxia‐induced transcription of some other genes. The capability of glycans to simultaneously reflect multiple gene anomalies in a single molecule contributes to the higher cancer specificity of glycans when applied as tumor markers. This principle could also be taken into consideration in the search for novel glycan markers. A glycan that simultaneously reflects the effects of both epigenetic silencing and tumor hypoxia in a single molecule could be a much better tumor marker.

Pathobiological Relevance of Hypoxia‐Induced Glycans

HIF‐induced genes confer on cancer cells the ability to adapt to hypoxia mainly by: (i) inducing tumor angiogenesis; (ii) shifting the intracellular carbohydrate metabolism to anaerobic glycolysis; and (iii) enhancing cellular mobility to escape the adverse environment. The sialyl Lewis A and sialyl Lewis X glycans are known to serve as ligands for E‐selectin expressed on vascular endothelial cells.^{( 45} , ^{46 )} Enhanced interaction of cancer cells with endothelial cells facilitates tumor angiogenesis and hematogenous metastasis.^{( 45} , ⁴⁶ , ^{47 )} This gives growth advantages to cancer cells expressing sialyl Lewis A and sialyl Lewis X glycans under hypoxic conditions.

Effects of hypoxia on cellular glucose metabolism are well analyzed, but it also influences galactose metabolism as exemplified by enhanced expression of genes involved in transport, phosphorylation, and transfer of galactose, and this seems to trigger drastic alteration of other monosaccharides including sialic acid.^{( 39} , ^{48 )} The influence of hypoxia on the metabolism of monosaccharides other than glucose, however, is not fully elucidated. Recently we found that metabolism of hyaluronan is markedly influenced by hypoxia.^{( 49 )} Hyaluronan serves as a specific ligand for CD44, and the cell adhesion mediated by the CD44/hyaluronan interaction is heavily involved in cancer cell motility.^{( 50} , ^{51 )}

Cell Adhesion Mediated by CD44 and Selectin: A Glycoproteomic Approach

Indeed, hypoxia simultaneously facilitates the activity of two cell adhesion systems, the selectin/glycan ligand interaction and the CD44/hyaluronan interaction, both known to be involved in hematogenous metastasis of cancers (Fig. 5). The capability of cancer cells to undergo hematogenous metastasis seems to be acquired through a “survival of the fittest” mechanism during the course of cancer progression in locally advanced hypoxic cancer nests.

Figure 5.

 Schematic illustration of dual functions of CD44v carrying selectin ligand glycans. The selectin ligand glycans, sialyl Lewis A and sialyl Lewis X, are carried by cell surface glycoproteins, and their interaction with vascular selectin facilitates hematogenous metastasis (a). CD44s and CD44v on cancer cells are known to bind equally to hyaluronan in the vascular bed, and this is also implicated in hematogenous metastasis (b,c). Expression of CD44v, but not CD44s, had been known to well correlate with hematogenous metastasis clinically, but the molecular basis for the correlation of CD44v with hematogenous metastasis remained elusive, because not much difference has been noted in the hyaluronan‐binding activities between CD44s and CD44v. The CD44v molecule carrying sialyl Lewis A and sialyl Lewis X glycans has dual functions in cell adhesion; it binds to hyaluronan and can also serve as a ligand for vascular selectins (d). This could explain the preferential clinical correlation of CD44v with the frequency of hematogenous metastasis, as CD44s has much fewer potential O‐glycosylation sites compared to CD44v. The fragments of CD44v carrying sialyl Lewis A and sialyl Lewis X glycans are released during cancer cell migration by the action of metalloproteinases (e), and are detectable in the sera of patients with cancers.

While normal leukocytes express a standard form of CD44 (CD44s), cancer cells are known to frequently express variant forms of CD44 (CD44v), which have additional variant domains encoded by various combinations of variant exons. There are many clinical statistics indicating that the expression of a certain variant form of CD44 on cancer cells well correlates with hematogenous metastasis in patients with cancers.^{( 52} , ⁵³ , ⁵⁴ , ^{55 )} But the hyaluronan‐binding activities do not differ much between CD44s and CD44v, and the molecular basis for statistical correlation of CD44v with hematogenous metastasis remains to be clarified.

The selectin ligand glycans, the sialyl Lewis A and sialyl Lewis X epitopes, are known to be carried by various cell surface molecules including glycoproteins and glycolipids. The major carriers had been assumed to be mucinous proteins such as MUC1 carrying the epitopes on their O‐glycans, while CD44 was also known to be one of the carrier proteins.

Recently it was reported that CD44v, but not CD44s, carries the sialyl Lewis X epitope, and this serves as a good ligand for selectin in human colon cancer LS174T cells.^{( 56} , ^{57 )} This is because the variant domain of CD44 is particularly rich in O‐glycan‐attaching sites, and subsequent studies indicated that the sialyl Lewis A epitope is also carried preferentially by CD44v compared to CD44s in some other cancer cells (Fig. 6a).^{( 58 )} CD44v can also carry non‐malignant glycans such as disialyl Lewis A and sialyl 6‐sulfo Lewis X, suggesting that the incomplete glycan synthesis due to epigenetic silencing also takes place on glycans carried by CD44v. Hypoxia‐induced enhancement of selectin ligand glycans was observed also with glycan side chains carried by CD44v (Fig. 6b).

Figure 6.

 CD44v fragments carrying sialyl Lewis A and sialyl Lewis X glycans in cultured supernatants of cancer cells and in clinical serum samples. (a) Western blotting results showing that CD44v is capable of carrying cancer‐associated glycan, sialyl Lewis A, as well as normal glycan, disialyl Lewis A. CD44 was immunoprecipitated using an anti‐CD44 antibody from culture supernatants after ionomycin treatment of cultured human colon cancer cell line SW1083 (parent), or the cells transfected with a gene for a βGlcNAc: α2‐6 sialyltransferase (transfectant), and analyzed for glycan expression by Western blotting using specific anti‐glycan antibodies. (b) Western blotting results showing that expression of sialyl Lewis X on the CD44v molecule is also enhanced by hypoxia. CD44 was immunoprecipitated using an anti‐CD44 antibody from culture supernatants of human colon cancer cell line, LS174T, cultured under normoxic or hypoxic (1% O₂) conditions. (c) Preliminary results on the levels of CD44v fragments carrying sialyl Lewis A or sialyl Lewis X glycans in sera of patients with cancers. The serum levels of CD44v fragments carrying sialyl Lewis A or sialyl Lewis X were determined by enzyme‐linked double‐determinant sandwich assays using a combination of immobilized anti‐CD44v as the catcher‐ and specific anti‐glycan antibodies as the tracer antibodies. (a,c) Adapted from Lim et al., 58 with permission.

These results imply that the CD44v molecule carrying selectin ligand glycans plays dual functions in cell adhesion (Fig. 5), serving as ligands for selectins through its O‐glycans attached to the variant domain, while contributing to cell adhesion through its classical binding to hyaluronan. It is notable that cell adhesion mediated by selectin‐glycan interaction induces rolling of cells on vascular beds at the physiological shear force of the normal bloodstream (at more than 1.0 dyne/cm²),^{( 59 )} while it does not confer the stopping of cells required for completion of the metastatic process. On the other hand, cell adhesion mediated by CD44/hyarulonan confers both rolling and stopping of cells, albeit at much lower shear stress (at less than 0.5 dyne/cm²),^{( 60 )} which is physiologically hardly relevant. A simultaneous activation of both cell adhesion systems by CD44v carrying selectin ligand glycans is expected to support rolling and stopping at physiological shear stress.

As CD44 molecules are released by the action of metalloproteinases such as ADAM‐10/‐17 upon cellular movements,^{( 50 )} the fragments of CD44v carrying selectin ligand glycans are released into the general blood circulation, and are detectable by heterogenous sandwich assays using appropriate antibodies as serum tumor markers (Fig. 6c).^{( 58 )}

Conclusion

Cancer is traditionally regarded as a primarily genetic disorder, while a glycan is not a direct product of genes. Paradoxically, quite a few antibodies have turned out to recognize glycans instead of proteins, when the monoclonal antibody approach was widely applied for systematic search of cancer‐specific molecules in the 1980s. It is still somewhat puzzling how glycans, which are only indirectly regulated by genes, can be good markers for cancers, which are primarily caused by genetic anomalies. The tentative answer to this question at this moment is that glycans are capable of reflecting multiple gene anomalies, a capacity that seems sometimes to fit them ideally for specific detection of cancer, which is now becoming accepted to be actually caused by an accumulation of multiple genetic and epigenetic anomalies. In addition, glycans are capable of reflecting even genetic silencing in a distinct way by production of glycan molecules having incomplete structures. From another perspective, some glycans have intriguing biological functions, and their functionally beneficial expression for cancer progression tends to be preserved, while those having effects adverse to cancer progression tend to disappear, in cancer cells.

Abbreviations

ADAM

A‐dismlegrin and metalloprotemase

DTDST

diastrophic dysplasia sulfate transporter

FUT

fucosyltransferase

GlcNAc

N‐acetylglucosamine

GLUT‐1

glucose transporter 1

HIF

hypoxia‐inducible factor

ITIM

immunoreceptor tyrosine‐based inhibition motif

PAPS

adenosine 3′‐phosphate 5′‐phosphosulphate

Siglec

sialic acid binding immunoglobulin (Ig)‐like lectins

VEGF

vascular endothelial growth factor