9.1.1.1 The Selectins

The selectins are a family of three C-type (Ca²⁺-dependent) lectins expressed exclusively by bone marrow-derived cells and endothelial cells: L-selectin (expressed on Leukocytes and HSPCs), E-selectin (expressed on Endothelial cells), and P-selectin (expressed on Platelets and endothelial cells). Due to their expression on the endothelium, E-selectin and P-selectin are often referred to as the “vascular selectins” whereas L-selectin is termed the “leukocyte selectin.” These molecules are type I transmembrane glycoproteins sharing a similar topology which comprises an amino-terminal lectin-like domain, an epidermal growth factor (EGF)-like domain, a variable number of consensus repeats (CRs; also known as complement regulatory proteins or “sushi” domains), a single membrane spanning domain, and a carboxy-terminal cytoplasmic domain (Figure 9.2A) [4,6]. The lectin domain primarily mediates ligand binding with minimal contribution provided by the EGF-like domain [19–22]. The number (and hence length) of sushi domains facilitates the recognition between selectins and their ligands as it allows a selectin to extend beyond the negatively charged cloud of the glycocalyx. The selectins bind to specialized carbohydrate determinants (Figure 9.2B), comprising sialofucosylations containing an α(2,3)-linked sialic acid substitution on galactose, and an α(1,3)-linked fucose modification on N-acetylglucosamine, prototypically displayed as the terminal tetrasaccharide sialyl Lewis x (sLe^x; or also to its isomer sLe^a) [16,23]. These determinants, also known as “CD15s,” may be displayed on either a protein scaffold (i.e., a glycoprotein) or a lipid scaffold (i.e., a glycolipid), and are recognized by mAbs such as CSLEX-1, KM93, and HECA-452 [24,25]. Although additional structural modifications, principally involving sulfation, increase the binding affinity of P-selectin and L-selectin to sLe^x, no known modifications are needed for optimal binding of E-selectin [26,27]. These characteristics of engaging rapidly and with high tensile strength to their ligands help to make the selectins the most important initiators of adhesion under flow. A brief introduction to each of the selectins is provided below.

L-selectin (CD62L) is expressed constitutively by all myeloid cells, virtually all B cells and naïve T cells, a subpopulation of memory T cells, NK cells, and both early and mature hematopoietic cells in the bone marrow. Unlike the other selectins, however, L-selectin is not expressed on endothelial cells [28]. Functionally, L-selectin is critical for the homing of naïve lymphocytes to high endothelial venules (HEVs) of secondary lymphoid organs [3,27]. L-selectin-dependent lymphocyte rolling on HEV requires localization of L-selectin to the tips of the microvilli characteristic of the surfaces of lymphocytes and other leukocytes, allowing for optimal leukocyte–endothelial interactions. Mapping of L-selectin domains by mAbs has revealed that the NH₂ terminal nine amino acids are critical for ligand binding. The EGF-like and the two short sushi domains maintain the spatial conformation and are also important for ligand binding [27]. HEV-borne L-selectin ligands are collectively referred to as peripheral node addressins (PNAds) and include the glycoproteins GlyCAM-1 (glycosylation-dependent cell adhesion molecule-1), CD34, MAdCAM-1 (mucosal vascular addressin cell adhesion molecule-1), podocalyxin, Sgp200, endomucin, and endoglycan. All of these PNAds have mucin-like domains that carry O-glycans [27]. Although the primary role of L-selectin is in promoting lymphocyte homing, its expression by circulating neutrophils may facilitate the secondary tethering of these cells to an already rolling neutrophil at an inflammatory site (secondary tethering) [29,30]; this, however, may be a rare event of which the physiological significance has been questioned [31,32]. If L-selectin is to recognize them, these ligands must be modified by various GTs to give specific sialylated, fucosylated, and sulfated oligosaccharides.

P-selectin (CD62P) is a glycoprotein that is constitutively expressed within secretory granules of endothelial cells (Weibel–Palade bodies) and in α-granules of platelets. Upon activation by inflammatory mediators (histamine, thrombin, LPS), secretory vesicles fuse rapidly with the plasma membrane allowing expression of P-selectin on the surface of the platelet or endothelial cells within minutes [6]. Thus, P-selectin is often involved in early leukocyte recruitment during an inflammatory response. P-selectin is also constitutively expressed at low levels on the endothelium of thymus [33], lung and choroid plexus microvessels [34], bone marrow microvasculature [35,36], in postcapillary venules in skin [37], and on peritoneal macrophages [38]. The transcription of P-selectin can be induced by cytokines such as interleukin (IL)-4 and IL-13 [39,40]. Once expressed on the surface of endothelial cells, P-selectin is rapidly internalized by endocytosis [41]. In addition to mediating the homing of immune cells to inflammatory sites, P-selectin-mediated interactions are also involved in controlling T-cell progenitor (circulating thymic progenitor, CTP) migration from the bone marrow to the site of T-cell development, the thymus [33].

It is important to note that differences appear to exist between the responses of human compared with mouse endothelium to inflammatory cytokines. Whereas IL-1 and tumor necrosis factor-alpha (TNF-α) are potent inducers of P-selectin expression in mouse endothelium, these cytokines only induce the expression of E-selectin (and not P-selectin) in human endothelium [42]. Additionally, experiments in transgenic mice comparing human and mouse P-selectin genes have demonstrated that following injection of TNF-α, the expression of human P-selectin mRNA decreases whereas mouse P-selectin mRNA increases [43]. This suggests that E-selectin may be the more dominant selectin in human inflammation and disease and brings into question the extrapolation to human disease of the emphasis on P-selectin ligand interactions related to mouse models.

E-selectin (CD62E) is a glycosylated glycoprotein that recognizes diverse groups of glycoconjugates on hematopoietic and cancer cells. Unlike P-selectin, E-selectin is not presynthesized in endothelial cells but, rather, is transcriptionally regulated by mediators including TNF-α, IL-1β, IFN-γ, and LPS. However, exceptions to this rule exist including the observations that vascular beds within the skin [37,44] and bone marrow [36,45] constitutively express E-selectin. Expression can occur as early as 2 h after stimulation and decline within 24 h [46]. Despite this delayed expression (as compared to P-selectin), E-selectin overlaps with P-selectin temporarily enhancing leukocyte recruitment during an inflammatory response and functionally leading to a dramatic decrease in the rolling velocity (slow rolling), and enhancing the probability of adhesion [47] and the subsequent steps of the cascade (including integrin activation) [48]. Even following the downregulation of P-selectin, E-selectin continues to enhance leukocyte–endothelial interactions leading to slower rolling velocities and improved adhesion [49]. When the stimulus wanes, E-selectin becomes internalized and degraded in lysosomes [50]. Physiologically, E-selectin-mediated interactions are involved in controlling leukocyte recruitment to injury and inflammatory sites [51], steady-state, tissue-specific homing in cutaneous tropism of skin-homing T cells [52,53], HSPC entry into bone marrow [45,52,53], as well as metastasis of CTCs.

9.1.2.1 E-Selectin Binds Specific Selectin Ligands on HSPCs to Guide them to the Bone Marrow

The vascular endothelial cells within the bone marrow constitutively express E-selectin and vascular cell adhesion molecule-1 (VCAM-1)[36]. Numerous studies in mice, including intravital microscopy studies, have demonstrated the importance of E-selectin, VCAM-1, and stromal cell-derived factor 1 (SDF-1) in mediating the entry of cells in flow (such as HSPCs and immune cells found in the blood) to the bone marrow [54–56]. These findings have established that homing to bone marrow involves a multistep cascade initiated by E-selectin binding to its ligands on the HSPCs to mediate tethering and rolling, followed by SDF-1 binding to CXCR4 (on HSPC) leading to activation of VLA-4 (on HSPC), resulting in firm adhesion mediated by VLA-4 binding to VCAM-1 and subsequent transmigration. The Step 1 effectors of this cascade, E-selectin and its ligands, play a fundamental role in the recruitment of circulating cells to the bone marrow as demonstrated by many studies and reviewed by Sackstein [16].

Studies in both humans and mice indicate that E-selectin is constitutively expressed on marrow endothelial cells [36,54,57], and intravital studies have revealed that HSPC migration to marrow occurs at specialized microvascular beds expressing E-selectin [54]. These and several other independent lines of evidence [45,58–62] have highlighted a central role for E-selectin-dependent interactions in HSPC recruitment to marrow. E-selectin-dependent binding of HSPC to marrow microvessels, either concurrent with transmigration or subsequently within “vascular niches,” could affect hematopoietic activity and with differential effects depending on the target glycoprotein ligand(s). Importantly, it is known that ligation of P-selectin glycoprotein ligand 1 (PSGL-1) can suppress hematopoiesis in mouse [63] and human [64] HSPCs. This fact, together with evidence that non-PSGL-1 and E-selectin ligands can also trigger apoptosis and growth inhibition of HSPCs from human and mouse [64], highlights a broader role for E-selectin receptor/ligand interactions in hematopoiesis. Indeed, recent work by Winkler et al. demonstrated that E-selectin is a crucial component of the vascular HSC niche that actively induces hematopoietic stem cell (HSC) proliferation and if E-selectin is removed from the equation, HSC quiescence is maintained [65]. This suggests that in the absence of a specific E-selectin-mediated proliferative signal, greater numbers of HSCs remain dormant.

Control of this process may be through the expression upon “metabolic activation” [66] of a specific E-selectin ligand with an ability to bind E-selectin and induce the observed proliferation and commitment.

9.1.2.2 E-Selectin/E-Selectin Ligand Interactions Involved in CTC Metastasis, Regulation, and Maintenance

Metastasis is the culmination of an elaborate cascade of events in which a cancer cell breaks away from the primary tumor, enters blood vessels, and travels through the body until it finally extravasates through the vessels of a distant organ to establish a secondary colony. In order to successfully establish a metastatic clone at a distant site, any CTCs must survive the targeted abuse present in the blood stream that may lead to induction of apoptosis or necrosis as well as avoid elimination by immune cells. CTCs that possess these enhanced survival capabilities (and have overcome this “microevolutionary” process) will generate metastatic colonies in distant organs and even “self-seed” [67] the primary site with more hostile tumor cells. Research focused on elucidating the molecular mediators used by CTCs to adhere to the endothelial cells lining vessels at a target site, i.e., bone marrow, may lead to therapies that can be used to dissuade metastasis.

Endothelial cells expressing E-selectin, such as the bone marrow or activated inflamed vessels, are able to capture CTCs displaying E-selectin ligands on their surface, and initiate the adhesion/metastasis cascade that leads to invasion of the tumor cell (Figure 9.1). These ligands have also been implicated in cancer stem cell regulation and maintenance relating to epithelial-to-mesenchymal transition (EMT) and other events necessary for metastatic growth. E-selectin-dependent interactions have been established in promoting metastasis of many cancers including breast, pancreatic, colon, and prostate [68–72]. sLe^x/a epitopes are significantly upregulated on cancer cells compared with normal cells and this may play a role in promoting CTC adhesion during metastasis [68,72,73]. This increased expression of selectin ligands is due, in part, to altered regulation of the GTs responsible for creating the sLe^x/a epitope [74,75]. Thus, it is necessary to identify not only the glycoproteins and glycolipids presenting sialo-fucosylated glycans but also the GTs responsible in the upregulated expression of selectin ligands in order to target their downregulation. The role of E-selectin ligands and relevant GTs will be discussed in the following section.

9.1.3.1 GTs Involved in Creation of E-Selectin Ligands

For the most part, the enzymes responsible for forming E-selectin ligands are similar to those responsible for both L-selectin and P-selectin ligand formation in that they require sLe^x structures. However, unlike L- and P-selectin ligands, the sLe^x cap can occur on a number of different glycans from N-glycans to glycolipids. In addition, analysis of FX^null mice (mice with a null mutation of the FX locus that encodes an enzyme within the de novo pathway for GDP-fucose synthesis) by Lowe’s group suggests that E-selectin ligands are much more sensitive than P-selectin ligands to loss of fucose, this further highlighting the importance of fucosylation to the formation of functional E-selectin ligands [81].

9.1.3.2 FTs and E-Selectin Ligand Formation

α1,3-fucosylation is imperative for E-selectin ligand activity and a number of human fucosyltransferases (FTs) have been identified that are able to modify terminal lactosamines. These include FT-III, FT-IV, FT-V, FT-VI, FT-VII, and FT-IX [82,83]. FT-III to FT-VI are all able to modify either sialylated or unsialylated type 2 lactosamines resulting in sLe^x or Le^x (Figure 9.2) structures, respectively, while FT-III (and FT-V) is also able to modify type 1 lactosamines resulting in sLe^a or Le^a.

Analyses of mice with targeted mutations in FT genes reveal that only two of these enzymes are required to form functional selectin counter receptor activity in leukocytes. These studies all support the observation that FT-VII and FT-IV cooperate in the synthesis of fucosylated glycans required for all leukocyte selectin ligand activity [84–86]. Specifically FT-VII^null mice have leukocytosis and their blood leukocytes have reduced E- and P-selectin ligand activity that effectively compromises the recruitment of leukocytes to inflammatory sites in in vivo mouse models of inflammation, such as the thioglycollate-induced peritonitis model [85] and the contact hypersensitivity model (CHS) [84,86]. However, residual recruitment is still evident (particularly with myeloid cells) and is due to FT-IV activity. CHS response studies performed in mice deficient in FT-IV, FT-VII, or both enzymes revealed that, although lymphocyte (Th1 and Tc1) recruitment was absent in FT-VII^null animals, both enzymes needed to be knocked out to inhibit the recruitment of myeloid cells. A subtle decrease in the myeloid cell CHS response was seen in FT-IV^null mice whereas lymphocyte recruitment was unaffected. Overall, these observations support the hypothesis that FT-VII dependent fucosylation of selectin ligands accounts for most of the selectin-dependent adhesion in leukocytes especially within the lymphocyte fraction. The dependence of selectin ligand formation on FT-VII is also paralleled in studies with human T lymphocytes [87–89]. Indeed, studies suggest that FT-IV acts preferentially on glycolipids whereas FT-VII acts on glycoproteins [90] and, interestingly, gangliosides are able to mediate tethering and rolling [91–93] on endothelium. However, glycoproteins are better at facilitating firm adhesion under physiological flow conditions [93] which could also be related to signaling through the glycoprotein selectin ligands to activate integrins directly [17]. Recent studies focused on assessing knockdowns of FT-IV and -VII in human cells suggest that an additional α1,3-fucosyltransferase, FT-IX, is important in decorating E-selectin ligands on human leukocytes but not mouse [94,95]. This is significant since the selectin ligands, particularly those that bind E-selectin, vary between different leukocyte cell populations and among species [96].

9.1.3.3 ST3GalTs and E-Selectin Ligands

α2,3-sialic acid linkages to terminal galactose residues are crucial in the formation of sLe^x epitopes. The contribution of sialylation to the formation of selectin ligands is well documented as sialidase treatment of leukocytes under static conditions in vitro completely abolishes binding of P-selectin and E-selectin to selectin ligands [97]. This linkage is catalyzed by the activity of the sialyltransferases (STs) ST3GalT-I, -II, -III, -IV, -V, and -VI. Only ST3GalT-III, -IV, and -VI have been shown to sialylate type 2 lactosamines and ST3GalT-IV and ST3GalT-VI are implicated in E-selectin (and P-selectin) ligand formation [97–100]. In vivo inflammatory studies of ST3GalT-IV^null mice reveal a mild but significant defect in selectin ligand function, including a mild reduction in E-selectin-dependent rolling, an increase in E-selectin-dependent rolling velocity, a decrease in L-selectin-dependent rolling during inflammation (but no effect on L-selectin-dependent rolling on HEVs [101]), and no contribution to P-selectin ligands [97,102]. Since sialylation is so crucial to selectin binding but yet the phenotype of the ST3GalT-IV^null is so mild, it is suggested that other STs, particularly ST3GalT-VI, contribute to selectin ligand formation. With the generation of the ST3GalT-VI^null and ST3GalT-IV/ST3GalT-VI double knockout mouse, the role of these STs appears to be critical for the generation of functional selectin ligands, especially in the case of the formation of P-selectin ligands (ST3GalT-VI is involved in E-selectin-mediated capturing but not in E-selectin-dependent rolling) [100]. Implications for the involvement of additional STs involved in selectin ligand synthesis and function are still warranted, especially for E-selectin and L-selectin.

9.1.3.4 C2GnTs and Functional E-Selectin Ligands

C2GnT (core 2 β(1,6)-N-acetylglucosaminyltransferase) isoenzymes (C2GnT-I, -II, -III) create the core 2 O-glycan branch [103] by adding GlcNAc to the Galβ1,3GalNAc core 1 structure expressed on serine or threonine residues. Capping of the core 2 branch with a sLe^x moiety creates the P-selectin counter receptor. Analyses of C2GnT-I^null mice, however, show that it is the isoenzyme essential for PSGL-1 modifications on leukocytes [104–106]. Despite the essential role of C2GnT-I in P-selectin ligand formation, C2GnT-I^null mice possess a relatively mild phenotype, showing only a partial reduction in selectin ligands and essentially no effect on lymphocyte homing [104], reflecting the possible redundancy of the C2GnT isoenzymes. It has been suggested that C2GnT-II and/or C2GnT-III contribute to selectin ligand formation and cell trafficking [107,108]. C2GnT-I and C2GnT-III are expressed primarily by lymphocytes while C2GnT-II is associated with goblet cell mucin production in the intestinal epithelium. A single knockout of any of these enzymes and double knockout of C2GnT-I and C2GnT-III had no effect on the trafficking of lymphocytes in an inflammatory mouse model of the large intestine. However, mice deficient in C2GnT-I and C2GnT-II or all 3 C2GnT enzymes were not able to overcome a helminth infection, suggesting that C2GnT-dependent modifications on lymphocytes may play a role in migration of lymphocytes (to the large intestine) [108].

In the context of E-selectin ligands, a number of studies indicate that cells deficient in C2GnT-I are able to bind to E-selectin, although at a slightly reduced level compared with wild-type cells [104,107,109]. These studies reveal that 75–80% of E-selectin binding is maintained in the absence of C2GnT-I, suggesting that either the isoenzymes are compensating for the loss of C2GnT-I or that other E-selectin ligands (besides PSGL-1) are dominant mediators of this interaction.

9.2.3.1 Glycans and Their Importance in Mediating Engraftment

Once HSPCs have used their newly acquired homing molecules to enter the bone marrow, successful transplantation then requires engraftment of the stem cells within their particular niche. Implications for the activity of GTs depend not only on improving the homing of therapeutic cells to their niche but also on the successful engraftment of these cells. Normally HSCs are located within specific regions in the bone marrow that dictate the behavior of these cells. The complex interplay of cells and molecules helps to regulate HSPC quiescence, division, and differentiation. Various cell types have been identified that contribute to the HSPC niche, including various mesenchymal stem cells (CD146⁺ perivascular MSCs [164], nestin⁺ MSCs [165], and leptin receptor⁺ perivascular MSCs [166]), CXCL12 abundant reticular (CAR) cells, and MSC-derived osteoblast lineage cells [167–169]. In addition, endothelial cells form an overlapping HSC niche [166,170]. These niches also appear to be further regulated by local sympathetic nerves [171]. A model has been proposed whereby the “osteoblastic” niche (endosteal surface of bone) harbors and maintains HSCs in a quiescent state, whereas the “vascular” niche (marrow vessels) promotes HSC proliferation and differentiation [65,172,173]. However, many recent studies have challenged this model [174] stating that the osteoblastic niche is actually extensively vascularized [55,175], and that since the most potent and quiescent HSCs prefer to reside in hypoxic [176,177], poorly vascularized regions, these may not reside in the “osteoblastic” niche of the bone marrow. A particularly interesting study has demonstrated that E-selectin, expressed on around 20% of bone marrow endothelial cells (BMECs), directly promotes HSC proliferation and, in the absence of this selectin, HSC quiescence and self-renewal potential increases [65]. These findings suggest that, in addition to directing the migration of HSPCs to the bone marrow, E-selectin—through a currently undefined ligand—also plays a key role in regulating HSC engraftment and fate. Currently, quite which E-selectin ligand(s) on the HSC mediate this proliferation remains unknown as studies focused on knocking down the key E-selectin ligands identified to date do not present the same HSC phenotype [65,96]. Further studies are warranted to identify these ligands.

9.3.1.1 Selectin Ligands and Metastasis

To date, many selectin ligands are known to be expressed on carcinoma cells, including PSGL-1 [182,225], CD44v [181,185,226–228], podocalyxin-like protein (PCLP) [229], death receptor-3 (DR-3) [186], gangliosides [93], mucin 1 (MUC1) [180,230,231], CD43 [232,233], and Mac-2BP [234] (Figure 9.1). PSGL-1 is the most characterized selectin ligand at the cellular, molecular, and functional levels in addition to being the only ligand currently known to be able to bind all three selectins. PSGL-1 is a type 1 transmembrane sLe^x/a-bearing mucin-like glycoprotein of 402 or 412 amino acids [235,236]. Moore et al. [237] first identified PSGL-1 as a P-selectin ligand on myeloid cells including neutrophils and HL-60 cells. PSGL-1 is a disulfide homodimer with apparent molecular weights of 240 kD (dimer) and 120 kD (monomer) on SDS-PAGE. The extracellular domain is rich in serine, threonine, and proline, and consists of 15 or 16 decameric repeats extended linearly due to the presence of multiple serine/threonine O-glycosylated sites that straighten the backbone of mucin proteins [238]. Three N-terminal tyrosine amino acids at positions 5, 7, and 10 are arranged in a consensus sequence that favors tyrosine sulfation. The cytoplasmic domain of PSGL-1 has 69 amino acids with signaling functionality but with no consensus sequence for phosphorylation [235]. PSGL-1 is expressed on hematopoietic cells and cancer cells. CD34⁺ HSPCs and myeloid cells at different states of maturation (except erythrocytes and megakaryocytes) and all peripheral blood leukocytes express PSGL-1 [239,240]. Expression of PSGL-1 on platelets has been reported by Frenette et al., although remains debated [241]. PSGL-1 is also expressed by leukemic cells [242] and carcinoma cells such as colon [225] and prostate cancer cells [182]. PSGL-1 binds to E-selectin in a different way than to L-selectin and P-selectin. Both sialylation and α1,3-fucosylation, that appear on core 2 O-glycan chains, are required for E-selectin binding but sulfation (of either sLe^x moiety or tyrosine) is not required [243]. The binding to E-selectin of PSGL-1 expressed by CTCs implies a functional role in bone tropism of metastatic cancer cells. Dimitroff et al. demonstrated that PSGL-1, expressed on the surface of the bone-metastatic cell line MDA-PCa-2b, supports rolling on BMECs. This rolling was abolished by using neutralizing antibodies against E-selectin, suggesting that E-selectin/E-selectin ligand interactions mediate this adhesion. Neuraminidase treatment of the MDA-PCa-2b cells also impaired rolling indicating the importance of sialic acid in mediating this interaction and subsequent rolling. Performing a Western blot analysis on the membrane fraction of MDA-PCa-2b for the expression of sLex (detected by HECA-452 mAb) revealed reactivity with CD44 in addition to PSGL-1 [182,183].

Many reports have characterized the binding of CD44 to E-selectin on HSCs, leukocytes, and CTCs. CD44 is known to direct the recruitment of lymphocytes toward inflammatory sites through its interaction with matrix HA (hyaluronin) and endothelial cell expressed E-selectin, and a similar interaction is thought to direct the metastasis of many cancer types [80]. The E-selectin binding form of CD44 is known as HCELL. HCELL is expressed on the surface of HSPCs, human leukemia cells and has recently been located on colon cancer and breast cancer cells. Unlike mucin-like glycoproteins, CD44 bears sLe^x/a structures on N-linked glycan branches [181,185,226–228,244].

MUC1 is a large heavily glycosylated transmembrane protein with an extracellular region comprising a variable number of tandem repeats that include 20 conserved amino acids (HGVTSAPDTRPAPGSTAPPA) each with five potential sites of O-glycosylation (as underlined). In normal cells, MUC1 is expressed on apical surfaces and glycosylated with core 2 branched O-glycans and lactosamine extensions, whereas in cancer cells MUC1 is expressed on the whole surface and decorated with aberrantly truncated O-glycosylated carbohydrate antigens. MUC1 expressed on breast cancer cells tends to present sialylated and unsialylated core 1 structures instead of core 2 structures. It has long been known that ST expression and activity is higher in breast cancer cells than in normal breast tissue. Thus, the exposure of truncated glycan structures may be due to sialylation of core 1 GalNAc or lactosamine structures leading to a blockade of further extension and expression of T and Tn antigens [245]. The MUC1 interaction with E-selectin was first described in colon cancer cells by Zhang et al. [231]; MUC1 from a colon cancer cell line or colon cancer patient serum was shown to be able to inhibit the binding of leukemic cells to E-selectin-expressing cells. MUC1, expressed by colon cancer cells, was reported to bind E-selectin better than CD43 from the same cells under flow conditions [230]. A recent study by Geng et al. [180] used E-selectin and ICAM-coated microtubes as a model for microvascular endothelium to show that the MUC1-expressing metastatic breast cancer cell line, ZR-75-1, was able to roll and adhere to the walls of the microtube under flow conditions, so implicating MUC1 as a ligand for E-selectin on breast cancer metastatic cells. CD24, another mucin-like glycoprotein, was also identified as an E-selectin and P-selectin ligand in breast cancer cell lines (KS and MCF-7) [246,247].

More recently, a number of novel ligands have been identified as E-selectin ligands on cancer cells including PCLP on colon carcinoma cells [229], Mac-2BP on breast cancer cells [234], and DR-3 on colon cancer [186]. In addition to these glycoproteins, gangliosides have also been shown to play a role as E-selectin ligands in breast cancer cell lines [93]. Mac-2BP is a highly glycosylated protein whose expression is correlated with cancer metastasis. Shirure et al. were able to immunoprecipitate E-selectin ligands from metastatic breast cancer (ZR-75-1) lysate using recombinant E-selectin and identified Mac-2BP as a novel ligand through MS analysis. Gene silencing of Mac-2BP significantly reduced the binding of ZR-75-1 cells to E-selectin-expressing human umbilical vein endothelial cell (HUVEC) cells in flow assays [234]. All these studies highlighted the role of selectin/selectin ligand interaction in supporting metastasis. Since it is the glycan decorations, primarily the expression of sLe^x/a on these proteins (and lipids), that mediate this interaction of the CTC with the endothelial E-selectin, the contribution of GTs to migration and metastasis-supporting programs such as EMT needs to be better understood.

9.3.1.2 Perturbation in the Expression of GTs in Metastasis

It is widely accepted that abnormalities in glycosylation are closely associated with malignant transformation and progression [201–203]. Accordingly, highly metastatic carcinomas (e.g., colon, breast, and prostate) exhibit high levels of sLe^x/a expression [74,180–183,185,186,225]. sLe^x/a biosynthesis involves the concerted action of four GTs [68] and, hence, any dysregulation in GT expression and function has a direct impact on selectin-mediated metastasis. For instance, C2GnT is overexpressed in colorectal adenocarcinoma [248,249]. Lung carcinoma cells were found to acquire a colonizing phenotype upon FT-VII overexpression [188,189,250] and, in colon carcinoma, FucT-III gene silencing caused downregulation of sLe^x expression and subsequent inhibition of tumor proliferation and metastasis to the liver [251,252]. Matsuura et al. [74] investigated the expression of FTs and STs in breast cancer samples and cell lines and found that FT-VI and -III are significantly elevated, and that by expressing these FTs in breast cancer cells normally negative for sLe^x and sLe^a expression (MCF-7), rolling events are amplified on IL-1β activated HUVECs. One decade later, Barthel et al. reported similar results, i.e., elevated expression of FucT-III and -VI in the metastatic prostate cancer cell line, MDA-PCa-2b, compared with nonmetastatic cell lines. When PC3 cells, a prostate cancer cell line that is normally negative for E-selectin binding, were transfected with FucT-III, -VI or -VII, the cells gained sLe^x expression and the ability to roll on E-selectin-expressing endothelial cells and to metastasize to bone [71,201].

Moreover, it was demonstrated that the most highly expressed ST in patient samples of breast carcinoma was ST3Gal-III. This ST is responsible for sLe^a biosynthesis and its overexpression was associated with shorter overall survival [253,254]. Correspondingly, ST expression in metastasizing mammary tumors is higher compared with levels in nonmetastasizing tumors [255]. A recent report has indicated that induction of EMT, a prometastatic transformation program, was accompanied by elevation in levels of ST3Gal-I/-III/-IV and FT-III which are responsible for terminal capping of lactosamine units with sialic acid and fucose residues to form sLe^x/a moieties on colon cancer cells [209]. In combination, these studies indicate the fundamental role of GTs as modifiers of surface glycome features in supporting cancer metastasis and also reveal the potential significance of GT inhibitors as basic research and therapeutic tools.

9.3.2.1 GT Inhibitors and Primers

Natural GT inhibitors include papulacandin B [258], polyoxin D [259], tunicamycin [260], castanospermine [261], swainsonine [262], deoxymannojirimycin [263], and N-butyldeoxynojirimycin [264]. Natural inhibitors may cause extensive modifications in glycoconjugate structures such that it is difficult to study or control a specific interaction or enzyme; this introduces the necessity for the design and synthesis of selective agents. Synthetic GT inhibitors can be divided into two classes (refer to Figure 9.5). The first comprises substrate analogs which cause dead-end inhibition whereby a modified substrate is used to mimic one or more of the enzyme substrate(s) in order to block the enzyme or compete with natural substrates and terminate or delay glycan chain initiation or elongation. This class is further separated into three classes: donor substrate analogs (mimetics) of the donor sugar, nucleoside diphosphate, nucleoside monophosphate, or a diphosphate moiety. Because the mammalian system has only nine sugar nucleotide donors, the selectivity of GTs is thought to be associated with the identification of the acceptor rather than with the donor substrate. Within the group of donor analogs, diphosphate mimetics are thought to confer even less selectivity than sugar-modified donor analogs. In order to fulfill the selectivity requirement of GT inhibitors, acceptor mimetics have been developed. Acceptor substrate analogs compete with natural acceptors or block the active site of the enzyme to prevent the glycosyl transfer to the natural acceptor. Acceptor analogs usually have modifications at the hydroxyl group. The group may be deoxygenated, derivatized, or replaced with a halogen or other functionalities. Bisubstrate analogs have two covalently linked motifs mimicking both the acceptor and donor substrates. These analogs are also referred to as transition state analogs because they mimic the transition state of the reaction, and they are thought to be the most active and selective inhibitors of GTs. The second class includes GT primers and metabolic decoys. In contrast to the dead-end GT inhibitors, GT primers prime the enzyme activity using irrelevant substrates leading to the consumption of free active enzyme and the production of irrelevant products. Products of the priming reaction may be either inactive or metabolic decoys, or further modified with downstream enzymes to produce metabolic decoys capable of inhibiting a target pathway or ligand. GT synthetic inhibitors will be discussed in more detail in the following section. For simplicity, the first class will be designated as “inhibitors” and the second class as “primers.”

9.3.2.2 GT Substrate-Analogous Inhibitors

As this chapter is focused on selectin ligand prototype sLe^x/a, this section in turn will discuss inhibitors against the key GTs involved in its biosynthesis (Figure 9.2) and with which sLe^x/a expression and/or selectin ligand binding were reported to be affected: C2GnT, FT, and ST.

C2GnT inhibitors: C2GnT (β(1,3)-galactosyl-O-glycosyl-glycoprotein β(1,6)-N-acetylglucosaminyltransferase; EC 2.4.1.102) is an inverting GT that catalyzes the transfer of a GlcNAc residue from UDP-αGlcNAc to β(1,6) linkage on core 1 structure (Galβ1,3GalNAcα-O-pp) to form core 2 branching, one of the major mucin-type O-glycan core structures [265]. An inhibitor to C2GnT GT was developed based on a deoxygenated analog of the enzyme acceptor (Galβ1,3,6-deoxy-GalNAcα-O(CH₂)₈COOCH₃) which produced moderate inhibition compared with the corresponding acceptor (Galβ1,3GalNAcα-O(CH₂)₈COOCH₃) in vitro [266]. A recent study showed that 5-thio-GlcNAc could act as a donor substrate analog against the C2GnT enzyme. This inhibitor is active in vitro and the first to be reported as active in vivo. Although it caused a decrease in levels of cellular O-GlcNAc, the surface glycan profile did not change. In vivo activity is thought to be due to peracetylation of 5T-GlcNAc, which overcomes the problem of poor diffusion of charged analogs. The peracetylated prodrug is converted to the 5T-GlcNAc by the intracellular estrases and, in turn, is converted to UDP-5T-GlcNAc through the salvage pathway [267].

FUT inhibitors: α1,3-fucosyltransferase (FucT-IV, -V, -VI, -VII; EC 2.4.1.152) is an inverting enzyme that catalyzes the transfer of the fucosyl moiety from GDP-fucose to the 3-OH of group of N-acetyllactosamine to give a Lewis X trisaccharide. α1,3/4-fucosyltransferase (EC 2.4.1.65) transfers a fucose sugar from GDP-fucose to the 4-OH of Galβ1,3GlcNAc moiety to give a Lewis A trisaccharide. FTs can act on sialylated LacNAc to give sLe^x/a tetrasaccharides [268].

As for 5T-GlcNAc, the same research group developed 5-thio-fucose and evaluated its activity against sLe^x expression and selectin binding activity. Peracetylated 5T-Fuc was taken up by the relevant cells and converted through a salvage pathway into a sugar nucleotide donor analog, GDP-5T-Fuc that, in turn, inhibited the addition of fucose by the FT enzyme. 5T-Fuc was able to inhibit sLe^x expression in HepG2 cells and impaired adhesion of the cells to selectin-coated surfaces, and activated endothelial monolayer cells [269]. A cell-permeable acetylated analog of a fucose bearing a fluorine atom near the endocyclic oxygen, 2-fluoro-fucose (2F-Fuc), was reported to exhibit global inhibition of FTs. After being deacetylated intracellularly, this fluorinated analog is converted to the corresponding nucleotide sugar substrate GDP-2-fluoro-Fuc by the salvage pathway and is able to inhibit sLe^x formation in a human myeloid cell line and, subsequently, to hinder E-selectin- and P-selectin-mediated rolling on coated surfaces [270]. A novel study investigated the oral bioavailability of 2F-Fuc in a murine model. 2F-Fuc inhibited protein fucosylation, neutrophil sLe^x expression, and selectin-mediated adhesion during oral administration with no apparent toxicity in mice over the course of 3 weeks. Inhibitory activities were reversed upon discontinuing the drug administration. 2F-Fuc was also found to inhibit tumor outgrowth in colon and renal cancer metastasis mouse models after oral administration. This study supports the development and potential use of clinically effective GT inhibitors [179,180] in the treatment of cancer and prevention of metastasis.

ST inhibitors: The α2,3-sialyltransferase enzyme (ST3Gal) is a retaining enzyme that catalyzes the transfer of sialic acid from cytidine-5′-monophospho-N-acetylneuraminic acid (CMP-Neu5Ac) to the terminal Galβ1,4GlcNAc (EC 2.4.99.6) or Galβ1,3GlcNAc. Donor-analogous substrates were developed to mimic the common donor substrate, CMP-Neu5Ac, by replacing the anomeric carbon-linked oxygen atom with an ethyl bridge. This analog caused significant inhibition as it probably prevented the nucleophilic attack by the enzyme basic group [271]. Recently, using a cell-permeable acetylated analog of a sialic acid bearing a fluorine atom near the endocyclic oxygen, a global inhibition of ST activity was reported. After being deacetylated intracellularly, this fluorinated analog is converted to the corresponding nucleotide sugar substrate, CMP-2-fluoro-NeuAc, by the salvage pathway. This analog is then able to inhibit sLe^x formation in a human myeloid cell line as well as E-selectin- and P-selectin-mediated rolling on coated surfaces. It is thought that the observed inhibition is due to both enzyme and feedback inhibition of de novo synthesis of CMP-Neu5Ac by the salvage pathway and, to date, this is the only α2,3-sialyltransferase analog that has been found to produce inhibitory effects in vivo [270,272]. Acceptor-analogous inhibitors were reported to inhibit rat liver α2,3-sialyltransferase activity in vitro using a trisaccharide analog with a modified 3′-OH. The modifications include deoxygenation, fluorination, amination, or methoxy derivatization [273]. The 3′-deoxy lactosamine derivative was also reported, but with no inhibition activity in vitro [266]. Bisubstrate analogs containing the donor CMP-Neu5Ac mimetic and galactose, lactose or lactosamine acceptor have been developed and found to produce weak inhibitory effects against α2,3-sialyltransferase in the case of galactose- and lactose-containing acceptors, whereas they produced potent effects in the case of lactosamine-containing analogs [271,274].

9.3.2.3 GT Primers and Metabolic Decoys

Traditional GT inhibitors are analogs with alterations in specific functionalities that affect the thermodynamics and kinetics of the reaction. These analogs can still bind to the enzyme, but cannot undergo covalent modifications to give reaction products (dead-end inhibition). Conversely, priming activity is achieved when the modified analog binds the active enzyme and undergoes covalent modification to yield modified soluble glycan. Modified glycans can be inactive and thus inhibitory by only the consumption of the active enzyme in an extraneous pathway diverting glycan synthesis from endogenous glycoconjugates (see Figure 9.5). Otherwise, the modified glycan can be biologically active so that it acts as a metabolic decoy, or further modified by the biosynthesis pathway enzymes to a metabolic decoy that can inhibit the glycoconjugate formation or block an effector target. Many of the well-known inhibitors are inactive in vivo probably due to poor permeation into cells and subcellular compartments such as the Golgi apparatus where most glycan formation takes place. Primers can circumvent this problem by conjugating the substrate to a hydrophobic aglycone that enables passive diffusion through the lipophilic cell membrane without affecting the binding kinetics to the target enzyme. Conjugating specific aglycones may also facilitate the isolation of reaction products for further study or create probes for the study of active sites. Also, heterogeneity between cell types in glycosylation patterns can be studied through analysis of the primed products. The monosaccharide glycoside paranitrophenol α-GalNAc (pNp α-GalNAc) was used to inhibit mucin glycosylation through priming modified lactosamine formation and extension in colon cancer and lymphoid tumor cells [275,276]. PNp α-GalNAc is an acceptor analog for β1,3-galactosyltransferase that transfers a Gal residue from UDP-Gal to GalNAc to form core 1 structure. The core 1 structure (Gal β1,3GalNAc) acts as a molecular scaffold for mucin decorations with glycan side chains (i.e., core 2 branching, lactosamine extension, and sLe^x/a formation). PNp α-GalNAc could enter viable cells, probably due to the hydrophobic aryl group, and consume glycosylation machinery from mucin decoration toward the priming reaction. The aryl group was also used to isolate and secrete the priming reaction products Galβ1,3GalNAcα-pNp and Galβ1,3(Galβ1,4GlcNAcβ-6)GalNAcα-pNp, and to study the glycosylation patterns in these cells. Two independent groups who used GalNAcα-O-benzyl as a primer for such pathways [277–279] also reported similar results. Likewise, several hydrophobic glycosides of GlcNAc, GlcNAcα-O-benzyl, GlcNAcβ-O-benzyl, GlcNAcβ-O-phenyl, and GlcNAcβ-O-pnp could act as primers for polylactosamine chain synthesis [280]. More recently, fluorinated peracetylated GalNAc was reported to be incorporated in selectin ligands expressed by human promyelocytic leukemia cells (HL-60) suggesting a metabolic priming activity and inhibition of glycan chain elongation through the blockade of carbon 4 with a fluorine atom, so preventing the addition of either fucose or galactose sugars. Addition of 4F-GalNAc to the culture medium caused a significant decrease in sLe^x expression and leukocyte adhesion in vivo [281]. Another study confirmed the activity of 4F-GlcNAc in inhibiting sLe^x expression and selectin binding. However, this research showed that there was no incorporation of 4F-GlcNAc into the growing glycan chain, suggesting that the main mechanism of 4F-GlcNAc activity lies in interference with UDP-GlcNAc biosynthesis and not in the priming and incorporation of 4F-GlcNAc [282].

Disaccharide glycosides can also be used as primers for GTs. GlcNAcβ1,3Galβ-O-naphthalenemethanol (GlcNAcβ1,3Galβ-O-NM) and GlcNAcβ1,4Galβ-O-NM inhibited sLe^x expression in leukemia, colon cancer, and lung cancer models, and subsequently reduced E-selectin-dependent cell adhesion to activated endothelial cells [283–286]. Furthermore, it was shown that the treatment of colon and lung adenocarcinoma with the peracetylated disaccharide primer (Ac)₆GlcNAcβ1,3Galβ-O-NM reduced the potential for lung colonization in immunodeficient mice. Acetylated derivatives are thought to be more readily bioavailable to cells and, hence, more active in vivo as the polar hydroxyl groups are masked. Acetyl groups are then removed by the action of intracellular and membrane bound esterases [287]. 4-deoxy derivative of (Ac)₆GlcNAcβ1,3Galβ-O-NM acted as an acceptor-analogous substrate against β1,4-galactosyltransferase and inhibited experimental tumor metastasis by Lewis lung carcinoma in vivo [288].

The development of a diverse group of GT inhibitors as modulators of surface selectin glycoconjugates holds great promise for the treatment of metastatic tumors and inflammatory disorders. Although this field is several decades old, most of the inhibitors known lack activity in vivo due to their polar nature and underprivileged cell permeability. Recently, several reports have indicated that this problem can be avoided by using the prodrug technology whereby peracetylated derivatives show enhanced uptake. Peracetylated prodrugs are easily deacetylated intracellularly to give the active form of the drug. Another approach is to use bulky hydrophobic aglycones to confer hydrophobic properties on the inhibitor candidate. Nevertheless the development of GT inhibitors with acceptable bioavailability remains a goal in the development of clinically effective antimetastasis therapeutic agents. On the other hand, since both physiological and pathological cell players share the sLe^x decoration, increased efforts should be directed toward to the development of selective agents that target-specific GT isoforms that are differentially expressed in CTCs but not in normal immune cells.