Physiol Rev Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


Physiol. Rev. 79: 181-213, 1999;
0031-9333/99 $15.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow An erratum has been published
Right arrow An erratum has been published
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by VESTWEBER, D.
Right arrow Articles by BLANKS, J. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by VESTWEBER, D.
Right arrow Articles by BLANKS, J. E.

PHYSIOLOGICAL REVIEWS   Vol. 79 No. 1 January 1999, pp. 181-213
Copyright ©1999 by the American Physiological Society

Mechanisms That Regulate the Function of the Selectins and Their Ligands

DIETMAR VESTWEBER AND JAMES E. BLANKS

Institute of Cell Biology, Center of Molecular Biology of Inflammation, University of Münster, Münster, Germany

I. INTRODUCTION
II. SELECTINS AS ROLLING RECEPTORS AND INITIATORS OF LEUKOCYTE ENTRY INTO TISSUE
III. CELLULAR MECHANISMS OF SELECTIN REGULATION
    A. Various Mechanisms for the Regulation of Selectin Expression
    B. Physicochemical and Biophysical Parameters of Selectin-Ligand Interactions
    C. How the Cell Surface Distribution of a Selectin Affects Its Function
IV. SELECTIN LIGANDS: CARBOHYDRATE MOIETIES THAT ARE PRESENTED ON A SELECTED NUMBER OF CARRIER MOLECULES
    A. Glycolipids as Binding Partners for the Selectins
    B. High-Affinity Glycoprotein Ligands of the Selectins
V. STRUCTURAL DETERMINANTS AND REGULATION OF SELECTIN LIGAND GLYCOSYLATION
    A. Oligosaccharides That Bind to the Selectins
    B. Regulation of Selectin Ligand Expression by Glycosyltransferases
VI. SELECTINS AND THEIR LIGANDS AS SIGNALING RECEPTORS
VII. CONCLUSIONS AND FUTURE DIRECTIONS
REFERENCES

    ABSTRACT
Top
References

Vestweber, Dietmar, and James E. Blanks. Mechanisms That Regulate the Function of the Selectins and Their Ligands. Physiol. Rev. 79: 181-213, 1999. --- Selectins are a family of three cell adhesion molecules (L-, E-, and P-selectin) specialized in capturing leukocytes from the bloodstream to the blood vessel wall. This initial cell contact is followed by the selectin-mediated rolling of leukocytes on the endothelial cell surface. This represents the first step in a cascade of molecular interactions that lead to leukocyte extravasation, enabling the processes of lymphocyte recirculation and leukocyte migration into inflamed tissue. The central importance of the selectins in these processes has been well documented in vivo by the use of adhesion-blocking antibodies as well as by studies on selectin gene-deficient mice. This review focuses on the molecular mechanisms that regulate expression and function(s) of the selectins and their ligands. Cell-surface expression of the selectins is regulated by a variety of different mechanisms. The selectins bind to carbohydrate structures on glycoproteins, glycolipids, and proteoglycans. Glycoproteins are the most likely candidates for physiologically relevant ligands. Only a few glycoproteins are appropriately glycosylated to allow strong binding to the selectins. Recently, more knowledge about the structure and the regulated expression of some of the carbohydrates on these ligands necessary for selectin binding has been accumulated. For at least one of these ligands, the physiological function is now well established. A novel and exciting aspect is the signaling function of the selectins and their ligands. Especially in the last two years, convincing data have been published supporting the idea that selectins and glycoprotein ligands of the selectins participate in the activation of leukocyte integrins.

    I. INTRODUCTION
Top
Next
References

The migration of leukocytes from the blood vessel into inflamed tissue is the central step in the process of inflammation. Binding of leukocytes to the blood vessel wall is strictly controlled by a complex cascade of molecular interactions between the leukocyte and the endothelial cell layer, mediated by cell adhesion molecules and leukocyte-activating factors (56, 306) (Fig. 1). These molecules allow leukocytes to recognize sites of extravasation, where they attach to and migrate across the endothelial barrier. Emigration of leukocytes from the blood is initiated by the capture of leukocytes from the bloodstream followed by their rolling along the endothelial cell surface. This process is mediated by the selectins, a special family of three cell adhesion molecules.


View larger version (39K):
[in this window]
[in a new window]
 
FIG. 1.   Entry of leukocytes into tissue is controlled by a cascade of multiple molecular interactions. Initial tethering of leukocytes to the endothelial cell surface is mediated by selectins. This enables leukocytes to roll along the blood vessel wall and to sense activating factors such as chemokines that are deposited on the endothelial cell surface. This leads to activation of leukocyte integrins that bind to members of the immunoglobulin superfamily and mediate firm adhesion, a prerequisite for directed migration of leukocytes on the endothelial cell surface. An increasing number of recent reports suggest that in addition to chemokines, selectins are also directly involved in integrin activation.

A variety of inflammatory mediators such as chemokines (18) or platelet-activating factor (PAF) (383), presented on the endothelial cell surface, are recognized by the leukocytes after initial contact. This leads to the activation of leukocyte integrins, which support stable cell attachment and enable leukocyte migration on the endothelial cell surface. Finally, the leukocyte transmigrates through the endothelial cell layer and the underlying basal membrane and enters into the tissue.

In contrast to most other cell adhesion phenomena, especially those during embryonal development, the recruitment of leukocytes from the flowing bloodstream is a very rapid process that requires special mechanisms for the establishment of cell contacts. The selectins represent a class of cell adhesion molecules that is specialized for this purpose. Their distribution is restricted to the leukocyte-vascular system. In contrast to the vast majority of most other cell adhesion molecules, the selectins function as lectins, binding carbohydrate ligands. The individual members of the selectins are designated by prefixes, which were chosen according to the cell type where the molecules were first identified: L-selectin is expressed on most types of leukocytes, E-selectin is expressed on activated endothelium, and P-selectin was first found in storage granules of platelets and is also expressed by endothelial cells.

L- and E-selectin were found as antigens for cell adhesion blocking antibodies. L-selectin was first defined by the monoclonal antibody (MAb) MEL14, which blocked the binding of lymphocytes to high endothelial venules (HEV) in mouse lymph nodes, a process called lymphocyte homing (104). Parallel to these studies, it was found that carbohydrate determinants were important for lymphocyte-endothelial interactions during lymphocyte recirculation (317, 318). Later, a mannose-6-phosphate-rich polysaccharide was shown to block the binding of lymphocytes to lymph node HEV (378). Furthermore, this polysaccharide was shown to bind to the MEL14-defined antigen, later named L-selectin. Cloning revealed that L-selectin does indeed carry an NH2-terminal lectin domain with homology to Ca2+-dependent mammalian lectins (188, 294).

E-selectin was identified by MAb that had been raised against cytokine-activated human endothelial cells and blocked the binding of neutrophils (31, 267). Cloning of this selectin revealed the close relatedness to L-selectin (32).

P-selectin was originally found as a membrane protein of storage granules in human platelets (141, 221). Cloning of P-selectin revealed the selectin nature of this molecule (150) and stimulated experiments that demonstrated the ability of this molecule to mediate neutrophil binding to platelets (127, 184) and to endothelial cells (107).

The extracellular part of all selectins is composed of three different types of protein domains also found in proteins of very diverse function (Fig. 2). The NH2 terminus of each selectin is formed by a 120-amino acid domain that shares some features with the lectin domain of the C-type animal lectins (88). This domain is followed by a sequence of ~35-40 amino acids similar to a repeat structure, which was first found in epidermal growth factor (EGF). The six cysteines in this element are located at equivalent positions in so-called "EGF repeats" of several proteins. A truncated, recombinant form of human E-selectin containing only the lectin domain and the EGF repeat has been crystallized, and the three-dimensional structure was determined (119).


View larger version (14K):
[in this window]
[in a new window]
 
FIG. 2.   Structural organization of selectins. Selectins are composed of an NH2-terminal lectin domain, a single epidermal growth factor (EGF)-type repeat, and various numbers of consensus repeats or so-called complement binding domains, which share sequence homology with a domain structure often found in proteins with complement binding activity. Proteins have a single transmembrane region and a short cytoplasmic tail. E- and P-selectins have different numbers of complement binding domains in different species. [Modified from Huang et al. (142).]

The single EGF element that is found in each selectin is followed by a varying number of repetitive elements, each ~60 amino acids long, which resemble protein motives found in complement regulatory proteins. The specific function of these so-called "complement binding" (CB) elements is yet undefined. It was shown that truncating increasing numbers of these domains impaired the efficiency with which P-selectin could support rolling of leukocytes (260), suggesting that the CB domains are important to extend P-selectin a sufficient length from the plasma membrane. Four of the six cysteine residues in these repetitive elements are conserved in the complement-related proteins. The size variation of the three selectins is due to the different numbers of CB domains (Fig. 2). Although L-selectin has two such domains in human, mouse, and rat, the number of CB domains in E- and P-selectin varies between different species. Human, mouse, and dog E-selectin has six such domains, rabbit and rat E-selectin has five, and bovine and pig E-selectin has four. Human P-selectin has nine CB domains; mouse, rat, and sheep P-selectin has eight; and bovine P-selectin has six. All three selectins are anchored in the membrane by a single transmembrane region that is followed by a short cytoplasmic tail consisting of only 17 amino acids for L-selectin and 32 and 35 for human E- and P-selectin, respectively. The functional analysis of the structural organization of the selectins has been reviewed (142).

On the basis of in vivo studies in various species with adhesion blocking antiselectin antibodies and of studies of mice deficient in the selectin genes, the important role of the selectins for the rolling of leukocytes on the blood vessel wall and for lymphocyte recirculation as well as leukocyte entry into inflamed tissue has been well established in numerous reports. The studies on selectin gene-deficient mice have been recently reviewed (51, 102) as well as the function of the selectins as rolling receptors (200), in lymphocyte homing (126), in lung inflammation (360), and in ischemia-reperfusion injury (333).

This review gives an overview of recent studies on the interactions of selectins with their ligands and on the regulation and the function(s) of these molecules. Earlier reviews summarize previous work (161, 186, 222, 280, 331, 348).

    II. SELECTINS AS ROLLING RECEPTORS AND INITIATORS OF LEUKOCYTE ENTRY INTO TISSUE
Top
Previous
Next
References

L-selectin was the first of the selectins that was shown to be important for the entry of leukocytes into tissue. On the basis of the in vivo inhibitory effect of the MAb MEL14 on lymphocyte homing into peripheral lymph nodes of the mouse, L-selectin was defined as a lymphocyte-homing receptor (104). L-selectin was also the first selectin that was shown to be involved in the migration of neutrophils into inflamed tissue, again based on the inhibitory effect of the MAb MEL14 on neutrophil migration into inflamed skin (198) and into inflamed peritoneum of the mouse (156). Another class of leukocyte adhesion molecules that is important for leukocyte extravasation is the leukocyte integrins, a group of three integrins sharing the same beta 2-chain (316). It was soon shown that activation of neutrophils was accompanied with the downregulation of L-selectin and the upregulation of one of the leukocyte integrins, alpha Mbeta 2 or Mac-1 (171). This suggested that L-selectin may act before the beta 2-integrin in the process of adhesion. Indeed, in two different animal models, it was soon shown that L-selectin mediates leukocyte rolling, the first interaction between leukocytes and the blood vessel wall. A recombinant fusion protein carrying the extracellular part of mouse L-selectin and the Fc part of human IgG1 (L-selectin-Ig) blocked leukocyte rolling in rat mesenteric venules (202). Similarly, the MAb DREG 200 against human L-selectin (172), when injected into rabbits, inhibited the rolling of leukocytes in vivo while an anti-beta 2-integrin antibody did not interfere with the rolling process but blocked the subsequent firm attachment of leukocytes to the venular endothelium (351). This work established a two-step model for leukocyte adhesion to endothelial cells under flow conditions in vivo, with the selectin mediating the rolling process and the beta 2-integrin acting subsequently. In very elegant and well-defined in vitro experiments, it was demonstrated that P-selectin, but not intercellular adhesion molecule-1 (ICAM-1; the major endothelial ligand for beta 2-integrins), could support rolling of neutrophils under flow conditions on a lipid bilayer containing the purified proteins (193). In contrast, a static incubation of the cells with the lipid bilayer resulted in a cell binding, which was 100 times more shear resistant if ICAM-1 was incorporated into the bilayer, than when P-selectin was incorporated.

Since these initial studies, numerous reports have clearly established and confirmed that all three selectins are involved in leukocyte rolling in vivo and the initiation of physical leukocyte endothelial interactions. E-selectin was soon shown to support rolling of neutrophils in in vitro adhesion assays under flow conditions (1, 165, 194). Both endothelial selectins, E- and P-selectin, were demonstrated to function as rolling receptors in vivo (83, 254). Rolling of leukocytes on the endothelial selectins was not restricted to neutrophils but was also demonstrated for bovine gamma /delta T cells (154) and for human alpha /beta and gamma /delta T cells (81).

Apart from lymphocytes and neutrophils, other leukocytes utilize the selectins as rolling receptors. Monocytes were found to roll via L-selectin and P-selectin on tumor necrosis factor (TNF)-activated endothelial cells in vitro (214). Eosinophils were shown to roll via L-selectin (309), but not on E-selectin (308). However, all three selectins were reported to be involved in eosinophil recruitment in vivo (135). Most leukocytes that can bind to P-selectin can also bind to E-selectin. However, mouse bone marrow-derived mast cells were found to roll on P-selectin (307) but not on E-selectin (312). In addition to the selectins, alpha 4-integrins on lymphocytes can also support rolling under physiological flow (29), and it was suggested that the integrin alpha 4beta 7 would function in L-selectin-mediated rolling as well as in alpha Lbeta 2-mediated firm adhesion, forming a "bridge" between both steps (19).

As can be expected from the in vivo rolling data, all three selectins are involved in the entry of leukocytes into tissue. The homing of lymphocytes into lymph nodes is the only process that is exclusively mediated by L-selectin. E- and P-selectins have not been reported to be involved in this process. Entry of lymphocytes as well as neutrophils into inflamed tissue is mediated by all three selectins. This was shown by antibody-blocking studies for L-selectin in the mouse, as mentioned above (156, 198), and for E-selectin in a peritonitis model in rat with an anti-human E-selectin antibody reported to be cross-reactive with rat E-selectin (238). Antibodies against each of the three mouse selectins were shown to block neutrophil infiltration into chemically inflamed mouse peritoneum, although blocking of L- and P-selectin was more efficient than blocking of E-selectin (38). The involvement of E-selectin in neutrophil-mediated damage of lung endothelium during acute airway inflammation could be demonstrated with anti-E-selectin antibodies in rat (238) and in monkeys. A protective effect of an anti-human P-selectin MAb against cobra venom factor induced pulmonary injury in rats could also be demonstrated (238).

Three mouse mutants have been generated that is each deficient in one of the selectin genes (Table 1). Lymphocyte homing was significantly reduced in L-selectin-deficient mice (12, 311). Similarly leukocyte rolling and peritoneal emigration of neutrophils in response to thioglycolate were reduced (12, 201). L-selectin deficiency also affected the successful execution of an immune response (58, 332, 374). P-selectin-deficient mice showed reduced neutrophil emigration in chemically inflamed peritoneum especially at early time points, 1 and 2 h after stimulation (53, 218). In contrast to P- and L-selectin mutants, E-selectin null mutants unexpectedly have no obvious abnormalities of the inflammatory response (52, 183). A more detailed analysis of E-selectin null mutants revealed a subtle defect in these mice: the slow-rolling granulocytes (~5 µm/s) were missing in these mice (182). Severe defects were observed when P-selectin was blocked in these animals by antibodies. This led to a strong reduction of neutrophil emigration into inflamed peritoneum and of edema formation in a delayed type hypersensitivity (DTH) model at late time points when anti-P-selectin antibodies had no effect in wild-type animals (183). These findings suggest that E-selectin and P-selectin share overlapping functions.

 
View this table:
[in this window] [in a new window]
 
TABLE 1.   Defects of selectin-deficient mice

Interesting results were obtained with double deficient mice. Because all three selectin genes are closely linked in a gene cluster covering ~300 kb on chromosome 1 (362), mice deficient in several selectins cannot simply be generated by breeding single mutant strains. Despite this difficulty, double deficient mice, lacking E- and P-selectin, were generated (52, 100). In contrast to the single-mutant mice, double-mutant mice displayed an increased susceptibility to bacterial infections, with the majority of the animals developing chronic inflammatory lesions of the oral mucosa and skin. Interestingly, neutrophil accumulation in Streptococcus pneumoniae-stimulated peritoneum was completely blocked 4 h after instillation, whereas the number of emigrated neutrophils was normal compared with wild type at 24 h after stimulation, arguing for other adhesive mechanisms (e.g., L-selectin) mediating neutrophil emigration at later time points.

Mice double deficient in P-selectin and ICAM-1, in contrast to single P-selectin mutants and single ICAM-1 mutants, showed complete absence of surgically induced rolling in cremaster venules for at least 2 h (181). This effect was not seen if P-selectin-deficient mice were treated with an anti-ICAM-1 antibody. It is not known how the lack of ICAM-1 could affect leukocyte rolling. Early emigration of neutrophils into S. pneumoniae-stimulated peritoneum was completely blocked at 2-4 h after stimulation in contrast to only partial effects in single mutant mice (53). Surprisingly, neutrophil accumulation in the alveolar space after intratracheal instillation of S. pneumoniae was not significantly inhibited in P-selectin/ICAM-1 double-mutant mice (53). Not in all organs are the selectins important for leukocyte-endothelium interactions. For the liver, it was shown that leukocytes activated by the chemoattractant peptide formyl-methionyl-leucyl-phenylalanine (FMLP) adhered to the wall of sinusoids in E/P-selectin double-deficient mice as well as in wild-type mice, even when L-selectin was blocked by antibodies (373). Rolling and adhesion were completely blocked in these animals in cremaster venules.

Numerous reports have demonstrated that the selectins are involved in ischemia/reperfusion injury, as reviewed in Reference 333. Antibodies against P-selectin significantly protected attenuated myocardial necrosis in a feline model of myocardial ischemia-reperfusion (365). A similar protective effect was seen when the selectin binding oligosaccharide sialyl Lewis X (see sect. VA) was administered (50). Similar results were obtained in a rat myocardial ischemia-reperfusion model (337). In a very careful study, the rolling and adhesion of leukocytes was analyzed in postischemic mesenteric venules of cats (176). The results demonstrated that antibodies against L-selectin and P-selectin as well as the polysaccharide fucoidin blocked reperfusion-induced leukocyte rolling. However, rolling needed to be blocked by >90% to achieve reasonable (~50%) attenuation in leukocyte adhesion in postischemic venules.

A human genetic disease was described, causing, in addition to other defects, a markedly reduced ability of neutrophils to adhere to endothelium, recurrent episodes of bacterial infection, and localized cellulitis without pus formation (92). This disease is believed to be based on defect(s) in fucose metabolism (92). No sialyl Lewisx is found in these patients, which is a fucose-containing tetrasaccharide known to bind to all three selectins (see sect. VA). Although this tetrasaccharide is not necessarily a physiological ligand for the selectins, it is now well established that the physiological selectin ligands contain alpha -(1,3)-fucose as an essential structural element (217) (see sect. VA). Indeed, neutrophils of the patients do not bind to E- or P-selectin-expressing endothelial cells in vitro (92, 264) and do not roll in venules under shear force (349).

Another human genetic disease, called leukocyte adhesion deficiency (LAD), is due to the lack of functional integrin beta 2-chains (CD18), essential for neutrophil extravasation into sites of inflammation. Such patients suffer from life-threatening infections (11). In analogy to this disease, the deficiency described by Etzioni et al. (93) has been named LAD II. This defect demonstrates the importance and the essential role of carbohydrate recognition, probably via the selectins, for host defense mechanisms and inflammation.

    III. CELLULAR MECHANISMS OF SELECTIN REGULATION
Top
Previous
Next
References

A. Various Mechanisms for the Regulation of Selectin Expression

Because contact formation between most leukocytes and endothelium is initiated by the selectins, the regulation of their presence on the cell surface is important for the control of leukocyte extravasation. The two endothelial selectins, E- and P-selectin, are absent from the cell surface of nonactivated endothelial cells and become induced upon exposure of the endothelium to various proinflammatory stimuli. This ensures that these selectins are only present on endothelium in inflamed tissues. L-selectin, in contrast, is constitutively expressed on leukocytes, which is in agreement with its special function in the continuous process of lymphocyte homing. The function of L-selectin in the initiation of leukocyte-endothelial contacts in inflamed areas is controlled by the regulated appearance of its ligand(s).

E-selectin is induced by cytokines such as TNF-alpha or interleukin (IL)-1beta and by lipopolysaccharide (LPS) as was first found for human umbilical vein endothelial cells (HUVEC) (31, 267). Induction occurred on the transcriptional level, and within 3-4 h after stimulation, maximal levels of E-selectin protein are expressed at the cell surface (32). Basal levels are reached again after 16-24 h, in contrast to other cytokine-inducible adhesion molecules such as ICAM-1 and vascular cell adhesion molecule-1 (VCAM-1). A similar mechanism and similar kinetics of the regulation of mouse E-selectin were found on mouse endothelioma cells (124, 363).

The 5'-flanking regions of human E-selectin were cloned and sequenced (66), and the regulatory elements of the gene were studied intensively. The results were comprehensively summarized in a recent excellent review (219). Some of the most important reports are mentioned here. Four regulatory elements were found in the human E-selectin promoter of which three are NFkappa B binding sites (199, 232, 289, 367, 368) and one is an ATF-binding element (167, 368). Although the NFkappa B elements are not sufficient for the cytokine-stimulated induction of E-selectin transcription (367), they are necessary, since proteasome inhibitors, which block the degradation of Ikappa B and thereby block the activation of NFkappa B also block transcriptional activation of E-selectin (276). In addition to the NFkappa B elements, the activating transcription factor (ATF) element is involved in cytokine-stimulated expression of E-selectin as well (167, 368). Stimulation with TNF-alpha activates two signaling pathways, NFkappa B and the kinases c-Jun NH2-terminal kinase (JNK1) and p38, which are both required for maximal expression of E-selectin (277).

In addition to TNF-alpha and IL-1beta , several other stimuli were found to activate expression of E-selectin. Interleukin-10 was shown to induce transcription of E-selectin in cultured human endothelial cells (353). On human dermal microvascular endothelial cells, induction was as efficient as with IL-1beta , whereas induction on HUVEC was less efficient. Similar strong expression was seen at 4 and 24 h after stimulation, while baseline levels were reached again at 48 h. Interleukin-3 induced E-selectin with the same kinetics as IL-1beta , but the amount of E-selectin was roughly one-half that induced by IL-1beta (45). Oncostatin M, a cytokine belonging to the IL-6 family, was found to stimulate E-selectin expression with similar kinetics as TNF-alpha (231). In addition to LPS from gram-negative bacteria, lipoteichoic acid from gram-positive bacteria was also found to induce E-selectin expression (168). Immune complexes were shown to stimulate E-selectin expression via the heat-labile complement component C1q (213), although it is not known whether this is a direct or indirect effect.

The stimulation of E-selectin expression can be suppressed by various mediators. Inhibition of E-selectin expression can be achieved with IL-4 (334). Interleukin-4 induced suppression of TNF-alpha -stimulated E-selectin expression is mediated by STAT6, which antagonizes the binding of NFkappa B (23). Furthermore, glucocorticoids (70), transforming growth factor-beta (TGF-beta ) (105), and elevation of cAMP (110, 268) can counteract cytokine-induced expression of E-selectin. The effect of glucocorticoids is mediated by affecting NFkappa B and not by interfering with ATF or c-Jun (46). The dual cyclooxygenase/lipoxygenase inhibitor tepoxaline was recently shown to suppress LPS-induced E-selectin expression and to block neutrophil migration into inflamed murine skin in vivo (382).

In addition to soluble factors, leukocytes contacting the endothelial cell surface can modulate the expression of E-selectin. Coincubation of HUVEC with human blood monocytes induced E-selectin expression and prolonged E-selectin expression for more than 24 h (274). Cell contact was needed for this effect, and antibodies against TNF-alpha were partially inhibitory (247). T cells from Leishmania-infected mice or from mice sensitized to the contact allergen trinitrochlorbenzene stimulated E-selectin when cocultured with mouse endothelioma cells (326). Again, cell contact was necessary, but anti-TNF antibodies could not block the effect. It is possible that the CD40 ligand was partially responsible, since binding of a soluble recombinant form of CD40 ligand to CD40 on endothelial cells leads to the induction of E-selectin (140, 166, 379).

P-selectin is inducible by two different mechanisms. It is stored in granules inside of platelets (alpha -granules) and endothelial cells (Weibel-Palade bodies) and can rapidly be mobilized to the cell surface of endothelial cells within minutes (within seconds in platelets) upon stimulation with histamine or thrombin or with pharmacological compounds such as Ca2+ ionophores or phorbol esters (107, 130). Expression is maximal at ~5-10 min after stimulation, and the protein is rapidly cleared from the cell surface within the next 30-60 min by endocytosis. Both endothelial selectins are rapidly internalized, but only P-selectin molecules can be recycled from endosomes into the trans-Golgi network, where they are targeted to Weibel-Palade bodies (325). Aside from this pathway, a considerable amount of the internalized P-selectin molecules (121) and all of the endocytosed E-selectin molecules (325) are delivered from endosomes into lysosomes. Endocytosis occurs via clathrin-coated pits (292). The cytoplasmic tail of P-selectin is responsible for endocytosis and intracellular targeting (82, 121, 292). The transmembrane domain of P-selectin enhances targeting into storage granula as was shown with E-selectin/P-selectin fusion proteins (98). P-selectin as well as E-selectin contain Tyr residues in their cytoplasmic tails, which were thought to be putative internalization signals. However, a clearly defined internalization signal has yet to be defined. The tyrosine residue in the cytoplasmic tail of E-selectin is not necessary for endocytosis (63), and it was shown that residues throughout the cytoplasmic domain of P-selectin affect the internalization efficiency (292).

A second regulation mechanism for P-selectin is similar to the one observed for E-selectin. Tumor necrosis factor-alpha was initially found to stimulate the transcript level and protein level of P-selectin in mouse and bovine endothelial cells with similar kinetics as that of E-selectin (124, 286, 363). This stimulation of P-selectin synthesis could be confirmed in vivo for the mouse (118) and the rat (16). Studies of a cytokine-induced meningitis model in wild-type mice and mice deficient in P-selectin or for both endothelial selectins revealed that cytokine-induced E- and P-selectin cooperatively contributed to meningitis and leukocyte accumulation in the cerebrospinal fluid (329). However, in experimental autoimmune encephalomyelitis, E- and P-selectin were not induced on blood-brain barrier forming endothelium and, consequently, were not involved in the infiltration of inflammatory cells (90).

In HUVEC, P-selectin expression could neither be stimulated by LPS nor by TNF-alpha or IL-1beta (377). Instead, IL-4 and oncostatin M were found to induce P-selectin transcription and protein expression, which lasted 72 h (377). Interleukin-4 or oncostatin M stimulated P-selectin expression more slowly than TNF-alpha in the mouse system. Interestingly, oncostatin M was also reported to stimulate P-selectin transport from storage granules to the cell surface (231). Together with the effect on E-selectin transcription (231) (peak at 4 h), oncostatin M induces a tripartite increase of leukocyte adhesion, based first on P-selectin, then E-selectin, and then P-selectin again (223). In addition to oncostatin M, IL-3 may increase similar mechanisms, since it also leads to an immediate upregulation of P-selectin (169), an E-selectin-dependent delayed adhesion (45), and a very small increase in P-selectin on human endothelial cells over a period of days (169).

Both endothelial selectins also seem to be constitutively expressed in certain tissues. Using immunohistochemistry on sections of human hematopoetic tissue showed constitutive expression of E-selectin on endothelium of such organs (290). Similarly, noninflamed skin venules support significant rolling interactions that are mediated in part by P-selectin (248, 375), indicating that some blood vessels do not require inflammatory stimuli for P-selectin expression.

A very interesting novel endothelial adhesion mechanism for neutrophils was recently found to be inducible on HUVEC by IL-1 (151). This mechanism is probably a lectin, since it depends on sialic acid on the neutrophil cell surface. Most importantly, the expression kinetics of this mechanism are different from those of the endothelial selectins, since the novel adhesion activity is maximally expressed only at 24 h after stimulation. A mechanism that may be similar was identified on bovine endothelial cells. This adhesion activity was maximally induced 24 h after stimulation, supported lymphocyte and neutrophil rolling, and could be blocked with a MAb against a 110- to 120-kDa glycoprotein (157). A soluble form of this endothelial protein binds to lymphocytes, and this binding could be blocked by EDTA and O-sialoglycoprotease, but not by neuraminidase treatment of the target cells. Thus this novel adhesion molecule could be a novel, lectinlike, and cytokine-inducible adhesion molecule on endothelium.

L-selectin is constitutively expressed on myeloid cells and a large subset of lymphocytes (198). It can be downregulated at the transcriptional level during lymphocyte differentiation from a naive to memory cell phenotype. Mitogen stimulation of T lymphocytes leads to a transient increase of L-selectin on the cell surface paralleled by an increase in L-selectin mRNA and followed by a decrease in L-selectin transcription and L-selectin protein exposed on the cell surface over the next 7 days (160). Stimulation of L-selectin activity by qualitative changes in receptor activity was reported (304) but has not yet been verified in other reports.

L-selectin is involved in the process of lymphocyte recirculation as well as in the migration of neutrophils and lymphocytes into inflamed tissues. Induction of L-selectin-mediated adhesion in inflammatory processes is probably achieved by the upregulation of L-selectin ligands. On the basis of indirect evidence, yet unidentified ligands were upregulated by cytokine activation on human endothelial cells (39, 305). Furthermore, L-selectin is an important adhesion molecule in the so-called "secondary tethering" process that describes the rolling of leukocytes on blood vessel wall-associated leukocytes (20) (see sect. IVB). This process depends on L-selectin ligands on the endothelium-associated leukocytes, most likely P-selectin glycoprotein ligand-1 (PSGL-1) (357; see sect. IIIB1).

On lymphocytes as well as on neutrophils, cell activation causes rapid downregulation of L-selectin within minutes (171), by proteolytic activity cleaving L-selectin at an extracellular site proximal to the cell membrane (228). Proteolytic shedding occurs on neutrophils within 1-5 min and can be induced by a variety of chemoattractants and activating factors such as C5a, FMLP, leukotriene B4 , IL-8, TNF, granulocyte-macrophage colony stimulating factor (CSF), and calcium ionophores (122, 156, 171), but not by granulocyte CSF, macrophage CSF, IL-1, or interferon-gamma (122). Furthermore, L-selectin shedding is also stimulated by cross-linking L-selectin with immobilized MAb (257) and by incubating neutrophils with IL-1-activated HUVEC monolayers for 30 min (299).

Proteolytic shedding of L-selectin from the cell surface leaves an intact 6-kDa transmembrane cleavage fragment on the cell surface that can be detected with a serum against the cytoplasmic tail of L-selectin (158). The cleavage site was determined to be located between Lys-321 and Ser-322. Although the membrane proximal region of L-selectin was found to be essential for cleavage, extensive mutations of this region revealed an extremely relaxed sequence specificity surrounding the cleavage site (60, 227). L-selectin shedding is resistant to a large variety of protease inhibitors such as inhibitors of serine proteases, metalloproteases, aspartic proteases, and cysteine proteases. Finally, hydroxamic acid-based metalloprotease inhibitors were found to be able to block proteolytic shedding of L-selectin (13, 24, 95, 270), revealing that the proteolytic activity which led to L-selectin shedding was based on a metalloprotease.

As soon as L-selectin was found to be rapidly lost from the surface of leukocytes after activation, it was speculated that L-selectin shedding facilitates detachment of leukocytes from the endothelial cells as they start migrating through the endothelial cell layer. The identification of protease inhibitors that could block the shedding allowed the testing of the functional significance of this process in leukocyte rolling. In an elegant study, Walcheck et al. (356) showed that neutrophils, rolling on immobilized L-selectin ligands (purified peripheral lymph node addressins, PNAd, see sect. IVB2), rolled at considerably lower velocities when treated with a hydroxamic acid-based protease inhibitor. The neutrophil accumulation rate increased. These studies suggest that L-selectin shedding occurs within seconds after stimulation and that L-selectin shedding is an important determinant for rolling velocities. However, in more complex systems, a shedding-blocking inhibitor could not influence the rate of initial attachment, rolling velocity, or transendothelial migration of neutrophils incubated with TNF-activated HUVEC monolayers under flow conditions (5). Thus the means by which regulated shedding of L-selectin contributes to the physiological process of leukocyte extravasation has not yet been established in all detail. Interestingly, it was recently demonstrated that the intracellular association of calmodulin with L-selectin regulates L-selectin shedding (159).

Rapid shedding was also observed for other important cell surface molecules, such as TGF-beta , IL-6-receptor, angiotensin converting enzyme, the beta -amyloid precursor protein, TNF-alpha , and several other proteins. Interestingly, a Chinese hamster ovary (CHO) cell mutant defective in the ability to process pro-TGF-beta was also not able to shed L-selectin and the IL-6 receptor (13). Recently, the protease that cleaves the membrane-bound TNF-alpha precursor and releases mature TNF-alpha from the cell surface was identified and cloned as a disintegrin metalloproteinase (33, 236). This protease, called the TNF-alpha converting enzyme (TACE), can also be blocked by hydroxamic acid-based inhibitors, but unlike the L-selectin shedding enzyme, TACE can also be blocked by EDTA. Tumor necrosis factor-alpha converting enzyme is a protein of 85 kDa, contains a disintegrin domain, and is a member of the family of mammalian adamalysins or ADAM (371); TACE is certainly a good candidate for a protease that is involved in L-selectin shedding. It was found that thymocytes from mice deficient in the TACE gene failed to shed L-selectin; however, in a cell-free assay, TACE could not directly cleave L-selectin, possibly arguing for an indirect involvement of TACE in L-selectin shedding (298).

B. Physicochemical and Biophysical Parameters of Selectin-Ligand Interactions

The recruitment of leukocytes from the rapidly flowing bloodstream is a special form of contact formation between cells that requires special molecular mechanisms. The selectins seem to be ideally suited for this purpose. To support leukocyte rolling, selectins have been proposed to have rapid bond association (kon) and dissociation (koff) rate constants and special mechanical properties linking tensile forces and bond dissociation (78, 128, 193). It was often argued that the affinity of the selectins for their ligands does not need to be high. Indeed, selectins have been shown to bind synthetic oligosaccharides such as the tetrasaccharides sialyl Lewisx (sLex) [NeuAcalpha 2,3Galbeta 1,4 (Fucalpha 1,3)-GlcNAc] or its stereoisomer sialyl Lewisa (sLea) with very low affinities [dissociation constant (Kd) 0.1-5 mM] (40, 67, 99, 147, 244, 287). Other, more complex carbohydrate compounds such as the tetra-antenary N-linked carbohydrate with an unusual sialylated di-Lex on one arm (261) were estimated to bind with a 1,000 times higher affinity to E-selectin. Soluble recombinant forms of P-selectin (344) and of E-selectin (136) were reported to bind leukocytes with affinities of a Kd >= 1 µM. However, it was not completely ruled out that these measurements were possibly influenced by the presence of oligomeric forms of the E- and P-selectin molecules.

Very recently, affinity constant and binding kinetics were determined for the interaction of L-selectin with its soluble glycoprotein ligand glycosylation-dependent cell adhesion molecule-1 (GlyCAM-1) (245). With the use of surface plasmon resonance, it was shown that a soluble monomeric form of L-selectin binds to purified immobilized GlyCAM-1 with a Kd of 108 µM. L-selectin dissociates from GlyCAM-1 with a very fast dissociation rate constant of >= 10 s-1. The calculated association rate constant is >= 105 M-1·s-1. Similar studies with a soluble monomeric form of human P-selectin and isolated purified PSGL-1 from human neutrophils revealed a Kd of 200 nM, a kon of >= 7 × 106 M-1·s-1, and a koff of >= 1.5 s-1 (226). With the use of a 19-amino acid, sulfated PSGL-1 glycopeptide and a recombinant form of P-selectin consisting of the lectin and the EGF domain, a Kd of ~800 nM was determined (57; see Table 2).

 
View this table:
[in this window] [in a new window]
 
TABLE 2.   Binding parameters for the interactions of selectins with their ligands

In addition to fast dissociation rates, a high tensile strength of the selectin-ligand bonds was suggested to support the rolling function of the selectins (9). Measurements were performed in laminar flow chambers by visualizing tethering and release of neutrophils on lipid bilayers containing incorporated P-selectin at a density at which the cells did not roll but transiently adhered (tethered) to the support. The kinetics of these transient binding events (tethers) were analyzed. Because flow subjects the neutrophils to a shear force that increases the koff , the koff in the absence of an applied force ("intrinsic" koff) was estimated by extrapolation to zero flow rate. This intrinsic koff was determined for P-selectin as 1 s-1. The bond interaction distance was determined as 0.5 Å; this is the distance at which separation of selectin and ligand weakens their interaction (9). Similar measurements for L-selectin, using L-selectin expressing neutrophils flowing over substrates coated with L-selectin ligand (PNAd), revealed an intrinsic koff of ~7 s-1 (6), which is in very good agreement with the solution koff determinant for the L-selectin/GlyCAM-1 interaction, obtained by surface plasmon resonance measurements (245). L-selectin-mediated leukocyte rolling is clearly faster than rolling mediated by E- or P-selectin (153, 182, 272). These rolling kinetics are in excellent agreement with the ~10 times higher koff of L-selectin versus E- or P-selectin. The specialization of the selectins, cell rolling and tethering, was elegantly demonstrated when leukocyte rolling on immobilized antibodies against Lex and sLex was examined (61). In contrast to selectins, antibodies supported rolling only within a restricted range of site densities and wall shear stresses, outside of which firm adhesion or detachment occurred (61). On the basis of in vitro adhesion assays under flow, other adhesion molecules were shown to support leukocyte rolling, such as tenascin (65), very late antigen (VLA)-4/VCAM-1, and alpha 4beta 7/mucosal addressin cell adhesion molecule-1 (MAdCAM-1) (29) an CD44/hyaluronan (64, 76). The physiological role of tenascin and CD44/hyaluronan in leukocyte emigration from blood vessels is yet unclear.

The molecular dynamics of the transition from L-selectin- to beta 2-integrin-dependent neutrophil adhesion was analyzed under defined hydrodynamic shear (330). Neutrophils were allowed to aggregate (a process which depends on L-selectin and beta 2-integrin) in a cone-plate viscosimeter. From this study, the binding kinetics of selectin and integrin appear to be optimized to function at discrete shear rate and stress, providing an intrinsic mechanism for the transition from neutrophil tethering to stable adhesion (330). A mathematical model for cellular aggregation under these conditions was formulated (243).

Although the flowing bloodstream drags on leukocytes that try to bind to the blood vessel wall, surprisingly, it was found that shear above a critical threshold was necessary to promote and maintain rolling interactions through L-selectin (97). Although this was first thought to be a special requirement for L-selectin and not for E- or P-selectin, it was recently reported that all three selectins share this requirement for a threshold level of fluid shear (192). This could even be demonstrated in vivo in L-selectin-deficient mice, under conditions where rolling was exclusively mediated by P-selectin (192). It was suggested that at low shear forces "the fluid shear may generate a moment which induces additional bond formation as the cell experiences a torque into the wall of the flow chamber during the lifetime of existing bonds. . . . Fluid shear may stabilize leukocyte rolling by deforming the cell slightly after the first bond cluster forms, thereby increasing the time and cell/substratum contact area to favor further bond formation" (192).

C. How the Cell Surface Distribution of a Selectin Affects Its Function

Soon after the selectins were found to be responsible for the initiation of leukocyte endothelial contact formation, leading to the rolling of leukocytes (193, 202, 351), L-selectin was found to be located on tips of microvilli, as was first examined by immunogold electron microscopy on frozen thin sections of neutrophils (266) and then by immunogold scanning electron microscopy (91, 129). On the basis of the analysis of sectioned cells, 78% of neutrophil, 72% of monocyte, and 71% of lymphocyte L-selectin was observed on microvilli (48).

The presentation of adhesion receptors on microvilli has been shown to facilitate the establishment of primary interactions between leukocytes and the vascular lining under physiological shear forces (352). This report examined the distribution of L-selectin and CD44 on transfected lymphoid cells. Although L-selectin was concentrated on microvilli, CD44 was restricted to the planar cell surface. With the use of chimeric molecules, it was demonstrated that the transmembrane and intracellular domains of CD44 targeted the extracellular part of L-selectin to the planar body. Analogously, the extracellular part of CD44 was directed to microvilli when fused to the transmembrane and intracellular domain of L-selectin. These experiments establish a mechanism for the specific targeting or anchoring of surface proteins to the two cell-surface domains on leukocytes. In addition, this study suggests that the expression of L-selectin on microvilli strongly improves its ability to initiate contacts of the transfected cells to ligand bearing substrates under flow. L-selectin-CD44 chimeric molecules that were excluded from microvilli initiated leukocyte rolling under flow only very inefficiently. In agreement with these findings, other adhesion molecules that have been demonstrated to mediate cell contact formation under flow are also found to be enriched on microvillous processes. This was shown for the P-selectin ligand PSGL-1 (234) and for the integrin alpha 4beta 7 (29). In contrast, beta 2-integrins, which are essential for leukocyte adhesion to endothelium but which are not able to initiate contacts under flow conditions, are excluded from microvillous processes.

Deletion of the COOH-terminal 11 amino acids of the 17-amino acid cytoplasmic tail of L-selectin eliminated binding of transfected cells to HEV in frozen sections of lymph nodes and also abolished rolling of these cells in vivo in exteriorized rat mesenteric venules (162). Interestingly, carbohydrate recognition was not affected, arguing for a function of the COOH-terminal amino acids in the correct anchoring to the cytoskeleton and for the importance of cytoskeleton interactions for the rolling process. In line with these results, treatment of the cells with cytochalasin B, which disrupts actin microfilaments, had the same effects as observed for the mutant (162). Pavalko et al. (262) showed that L-selectin binds directly to alpha -actinin. Lack of the COOH-terminal 11 amino acids of L-selectin disrupted the binding of L-selectin to alpha -actinin. However, this mutant L-selectin still localized normally to the microvillar projections on the cell surface (262). Thus L-selectin does not only have to be positioned on the tips of microvilli to be able to support leukocyte rolling; it also has to be anchored to the cytoskeleton. It is still unknown which molecular interactions lead to the presentation of L-selectin on the microvilli tips.

Direct binding of E- or P-selectin cytoplasmic tails to alpha -actinin was not observed, although similar conditions were tried as had been successfully used for L-selectin (163). Furthermore, deletion of the cytoplasmic tails of E- and P-selectin neither affected the cell surface expression of these selectins nor their adhesion function as was tested in nonstatic/rotation assays with transfected COS cells (163). However, a function for the interaction of E-selectin with the cytoskeleton could well be important for events downstream of the leukocyte docking process. Yoshida et al. (380) showed that leukocyte binding to activated HUVEC could increase the fraction of E-selectin that was detergent insoluble, i.e., could not be extracted and presumably was associated with the cytoskeleton. This was not seen with a cytoplasmic deletion mutant of E-selectin. In addition, cross-linking of E-selectin with antibodies allowed to coprecipitate cytoskeleton proteins such as alpha -catinin, vinculin, paxillin, filamin, and even focal adhesion kinase (FAK) in E-selectin immunoprecipitations. These proteins did not copurify if the cross-linking step was omitted (380). Cytoskeletal linkage of E-selectin might be important for cell-cell signaling or for mechanical stabilization of leukocyte-endothelial interactions immediately after the first interaction of leukocytes with E-selectin.

Examining the effect of cell shape on neutrophil tethering and rolling on endothelial selectins revealed that microvilli are essential to allow neutrophil's initial binding to a support under flow conditions, but are not important for the subsequent rolling movement (96). Disruption of microfilaments by cytochalasin B caused an ~50% reduction of the numbers of microvilli, whereas hypotonic swelling reduced the number of these protrusions by 80%. Both treatments almost completely wiped out tethering, but when tethering was allowed at subphysiological levels of shear stress of <= 0.35 dyn/cm2, subsequent increase of shear stress removed control cells at much lower shear from the support than cytochalasin B-treated or hypotonically swollen cells.

    IV. SELECTIN LIGANDS: CARBOHYDRATE MOIETIES THAT ARE PRESENTED ON A SELECTED NUMBER OF CARRIER MOLECULES
Top
Previous
Next
References

Unlike most other cell adhesion molecules that bind to their ligands on the basis of protein-protein interactions, the ligands of the selectins are composed of a scaffold protein, or perhaps a lipid carrier molecule, which is modified by certain carbohydrates. Thus the carrier molecule is not sufficient to define a selectin ligand; it needs to be expressed in the right cellular background that provides the necessary repertoire of glyosylation enzymes which confer selectin-binding activity to the carrier molecule. Lectin recognition systems have been described as functional triads: receptors, ligands, and carriers. The receptors are the lectins, the ligands are the oligosaccharides, and the carriers are molecules on which these oligosaccharides are optimally assembled and presented for binding to the lectin (68). Because the physiological binding partners of the selectins are most likely glycoproteins and most publications in the field refer to such binding partners as ligands, the term ligand will be used for the glycoproteins that bind to the selectins throughout this review.

Some of the uncertainty about which glycoproteins are the physiological selectin ligands is based on the fact that oligosaccharides that can bind with some affinity to a selectin can indeed transfer selectin-binding activity to many different carrier proteins (347). Even BSA, chemically modified with the selectin binding oligosaccharide sLex, can bind to a selectin (25) (27). Furthermore, depending on the technique with which binding is detected, ligands with physiologically irrelevant, low affinities could be mistaken for physiological ligands (347).

Physiological ligands are most likely distinct glycoproteins that actively take part in the formation of the ligand molecule. Evidence is evolving that oligosaccharides are not the only modifications that are necessary for selectin binding on a certain carrier, as was shown for the tyrosine sulfation of the P-selectin ligand PSGL-1 (269, 284, 370). Furthermore, some carrier molecules seem to be preferential targets for the generation of certain carbohydrate modifications that enable this molecule to bind to a selectin, as was shown for the E-selectin ligand-1 (ESL-1) and for PSGL-1 (36, 369, 385). Thus a limited number of discrete glycoproteins, modified with certain oligosaccharides and in some cases with other posttranslational modifications, define physiological ligands of the selectins.

This section mainly focuses on those glycoprotein ligands, which have been identified as so-called high-affinity ligands, based on their specific and selective isolation from cellular detergent extracts with selectin affinity probes (Fig. 3). Several excellent reviews were published recently about these ligands (220, 280, 347, 361). Here we focus on the most recent results.


View larger version (20K):
[in this window]
[in a new window]
 
FIG. 3.   Selectin ligands that have been identified by affinity isolation with respective selectin as affinity probe. Except E-selectin ligand-1 (ESL-1), depicted ligands are sialomucins or contain at least a sialomucin domain. L-selectin ligand mucosal addressin cell adhesion molecule-1 (MAdCAM-1) was originally found as a ligand for integrin alpha 4beta 7. Sequencing revealed a sialomucin domain. A subpopulation of MAdCAM-1 molecules in high endothelial venule of mesenteric venules can indeed be expressed as an L-selectin binding glycoform, carrying posttranslational modifications that define peripheral node addressins. L-selectin is a major carbohydrate-presenting ligand for E-selectin on human neutrophils; however, L-selectin of human lymphocytes or mouse neutrophils is unable to bind E-selectin. P-selectin glycoprotein ligand-1 (PSGL-1) is the only selectin ligand so far that has been demonstrated to mediate leukocyte rolling on endothelium (251) and leukocyte recruitment into inflamed tissue in vivo (35, 37). Ig, immunoglobulin; GlyCAM-1, glycosylation-dependent cell adhesion molecule.

A. Glycolipids as Binding Partners for the Selectins

Numerous reports have documented that glycolipids bind specifically to the selectins. Fucosylated monosialogangliosides mediating binding to E-selectin were isolated from human myeloid cells (239, 323). Sialyl Lewisx-carrying glycolipids and sialyl Lewisa-carrying neoglycolipids were shown to support rolling of E-selectin-transfected cells and of L-selectin-expressing leukocytes (7). An unusual class of sulfated glycosphingolipids, sulfoglucuronyl-containing neolactosyl-ceramides (SGNL lipids), that are recognized by the mouse MAb HNK-1 bind to L- and P-selectin, but not to E-selectin (242). Sulfatides bind to P-selectin (14) as well as to the other two selectins (242), although they only support weak tethering of L-selectin-expressing cells and do not support rolling (7). It is still uncertain to what extent glycolipids are relevant for selectin-mediated adhesive events. Because several selectin ligands are mucins that are rigid and highly extended molecules, likely to project from the leukocyte surface, and L-selectin is exposed on the tips of microvilli, it has been emphasized that projection above the cell surface may facilitate cellular interactions under flow (193, 266). Furthermore, several studies have demonstrated that protease treatment blocks the binding of myeloid cells to P-selectin (185, 315). Treatment of myeloid cells with tunicamycin blocked cell binding to E- and P-selectin (185). This argues against glycolipids as ligands involved in selectin-mediated leukocyte capturing. However, it is possible that glycolipids might be involved in the rolling process after initial tethering has occurred, strengthening selectin-mediated cell contacts during rolling.

B. High-Affinity Glycoprotein Ligands of the Selectins

1. P-selectin ligands

In contrast to the other selectin ligands, PSGL-1 was identified by expression cloning, using a P-selectin-Ig fusion protein as panning reagent and an expression library from the human monocytic cell line HL-60 transfected into COS cells (
283). The same protein had been identified as a 250-kDa disulfide-linked dimeric protein 1 year earlier by affinity isolation using purified P-selectin as affinity probe (235). This protein was demonstrated to be identical to PSGL-1 (233). A 230/130-kDa pair of glycoproteins, which had been identified on mouse neutrophils by affinity isolation with P-selectin-Ig and E-selectin-Ig chimeras (196), turned out to be the mouse homolog of PSGL-1 (35) that had been cloned with the help of a human PSGL-1 cDNA probe (376). The PSGL-1 polypeptide chain is broadly expressed on cells of myeloid, lymphoid, and dendritic lineage and on some nonhematopoetic cells, such as the epithelium of the fallopian tube and sporadically on microvascular endothelium in some pathologic tissues (189). P-selectin glycolipid-1 was also reported to be expressed on the zona pellucida of porcine oocytes, and P-selectin was found on the acrosomal membrane of porcine sperm cells (108).

P-selectin glycolipid-1 requires carbohydrate modifications such as sialic acid and fucose (283) as well as branched carbohydrate side chains generated by the core-2 beta -1,6-N-acetyl-glucosaminyltransferase (core-2 enzyme) for its binding activity (180, 204, 233, 369). Detailed analysis of the O-linked carbohydrate side chains revealed an interesting trifucosylated O-glycan, beta -1,6-linked to core-2 structures on PSGL-1 of HL-60 cells, which was not found on a control sialomucin of this cell line (369) (Fig. 4). In addition to carrying the correct oligosaccharides, PSGL-1 needs to be sulfated at one of the three tyrosine residues at its NH2 terminus for binding to P-selectin (269, 284, 370) and probably also for binding to L-selectin (303). P-selectin glycolipid-1, as almost all other glycoprotein ligands of the selectins, is a sialomucin. The clusters of O-linked carbohydrate side chains make it a rigid and extended molecule (203). The protein is processed in the endoplasmic reticulum by a paired basic amino acid converting enzyme (PACE) that cleaves off the pro-peptide (345). The following 19 amino acids are important for the binding of PSGL-1 to P-selectin, since they carry the tyrosine residues that can be sulfated, harboring the epitopes of several adhesion blocking antibodies against human and mouse PSGL-1 (35, 37, 234, 300, 376). Furthermore, cleaving the 10 NH2-terminal amino acids of the mature form of human PSGL-1 with a cobra venom metalloproteinase, mocarhagin, abolishes binding to P-selectin (77).


View larger version (12K):
[in this window]
[in a new window]
 
FIG. 4.   Analysis of O-glycans of human PSGL-1, isolated from HL-60 cells, revealed a minority of alpha -1,3-fucosylated structures that occurred as 2 core-2-based species, depicted here (369).

P-selectin glycolipid-1 is the best-characterized selectin-ligand to date, which fulfills all criteria for a physiologically relevant ligand. It is the major ligand for P-selectin on human neutrophils, as the MAb PL-1 to PSGL-1 completely blocks rolling of these cells on P-selectin (234), and it is the major ligand for P-selectin on stimulated T cells (345). A soluble, recombinant form of human PSGL-1, produced in FucTIII- and core-2 enzyme-transfected CHO cells, can block ischemia/reperfusion injury in the rat (328). The MAb PL-1 can also block rolling of human neutrophils in venules of exteriorized rat mesentery (251). Polyclonal antibodies against mouse PSGL-1 were found to block the entry of Th1 cells into inflamed skin of the mouse (37), and a MAb against mouse PSGL-1 was able to inhibit the migration of neutrophils into chemically inflamed peritoneum (35). Thus, in vitro as well as in vivo, PSGL-1 has been demonstrated to be necessary for neutrophil and lymphocyte binding to P-selectin, despite the fact that PSGL-1 only presents <1% of the sLex on the cell surface (204, 249, 369). The demonstrated functional importance of PSGL-1 establishes that a single glycoprotein can be responsible for the generation of high-affinity and biologically relevant cellular interactions with a selectin, despite the lectin character of the selectins and their ability to bind to many sLex-bearing glycoproteins when presented at high density in certain in vitro assays. Because PSGL-1 could be affinity isolated as a major ligand by P-selectin-based affinity probes (196, 235), this ligand is a good example for the validity of this approach to identify physiologically relevant ligands.

E-selectin can also bind to PSGL-1, as has been demonstrated in numerous reports (15, 114, 196, 204, 233, 259). This binding requires the same carbohydrate modifications as the binding to P-selectin; however, sulfation of tyrosine residues is not necessary for the binding to E-selectin (114, 204). With the use of recombinant fragments of human PSGL-1, coated on microspheres, it was shown that P- and E-selectin both bind to the first 19 amino acids of PSGL-1 and that E-selectin can additionally bind to a site located between amino acids 19 and 148 (114). Whether PSGL-1 is indeed relevant for the binding of cells to human E-selectin has been questioned. First, the binding affinity of human PSGL-1 to human E-selectin was reported to be 50-fold lower than to P-selectin (233). Second, the binding of human T-cell clones to E-selectin was reported to be independent of sialomucins (10). Third, the inhibitory effect of the anti-PSGL-1 MAb PL-1 on the rolling of human neutrophils on E-selectin (259) was not due to the blocking of direct interactions of PSGL-1 with E-selectins, but rather due to the blocking of leukocyte-leukocyte interactions (258). This interaction between leukocytes, the so-called secondary tethering, is mediated in part by PSGL-1 (357). Thus evidence for a direct involvement of PSGL-1 in cellular interactions with E-selectin is still lacking.

Despite this lack of evidence for the physiological relevance of PSGL-1 as a ligand for E-selectin, it was shown recently that PSGL-1 on activated T cells is the major if not even the only glycoprotein carrier for a carbohydrate epitope, called cutaneous lymphocyte antigen (CLA) (36, 103), which is thought to be relevant for lymphocyte binding to E-selectin but not to P-selectin. This carbohydrate epitope, defined by the MAb HECA452, is further described in section IVA. Borges et al. (36) used a CD8+ mouse T-cell clone that expressed CLA transiently after antigen-specific activation. Cells could only bind to E-selectin when they expressed CLA, whereas cell binding to P-selectin was independent of CLA expression. P-selectin glycolipid-1 was found to be the only ligand that could be affinity isolated from these cells with E-selectin-Fc, and this was only observed at an activation stage when PSGL-1 was modified with the CLA carbohydrate epitope (36) (see also sect. IVB). Binding of PSGL-1 to P-selectin occurred in the presence as well as in the absence of the CLA epitope, arguing for the independent regulation of the binding activity of PSGL-1 for E- and P-selectin on activated T cells.

The P-selectin-binding glycoform of PSGL-1 on CD4+ T cells seems to be induced by activation stimuli which also lead to the differentiation of these cells to Th1 cells. In a mouse DTH model, it was recently shown that Th1 cells, generated by in vitro differentiation of primary isolated mouse CD4+ T cells, can migrate into inflamed skin, whereas Th2 cells cannot. Th1 cell migration into the skin was almost completely blocked by antibodies against E- and P-selectin (17). In agreement with this, only Th1 cells, but not Th2 cells, bound readily to the P-selectin-Ig chimera. Biochemical analysis revealed that Th1 and Th2 cells carried similar amounts of PSGL-1 molecules on their cell surface, but only the PSGL-1 on Th1 cells was modified in a way that allowed binding to P-selectin (37).

In addition to P-selectin and perhaps E-selectin, L-selectin seems to be an important receptor for PSGL-1. In adhesion assays under flow, neutrophils (20) and also lymphocytes (155) were found to roll on leukocytes that had already established contact to the selectin coated support. This secondary tethering (8) was found to be at least in part dependent on L-selectin on the adherent leukocytes (8, 20) and on PSGL-1 as an L-selectin ligand on the flowing leukocytes (303, 342, 357), although one of the reports did not confirm PSGL-1 as an L-selectin ligand in this process (8).

Platelets were reported to be involved in the interactions of lymphocytes with HEV of mouse peripheral lymph nodes, thereby mediating lymphocyte homing. P-selectin on platelets was suggested to mediate this process. It was shown that activated platelets expressing P-selectin on their surface could "rescue" lymphocyte homing in mice treated with the anti-L-selectin MAb MEL14 (80). Circulating activated platelets could reconstitute lymphocyte homing and immunity in L-selectin-deficient mice (79). Platelets bound via P-selectin to PSGL-1 on mononuclear lymphocytes and to peripheral node addressin on HEV, suggesting that the peripheral node vascular addressins can function as P-selectin ligands.

Another ligand that was described for P-selectin is CD24, also called heat-stable antigen (HSA). Although it has not yet been possible to directly affinity isolate this molecule from myeloid cells with a P-selectin probe, its binding to P-selectin has been analyzed in several reports. Heat-stable antigen is a cell surface glycoprotein expressed by neutrophils, B cells, immature thymocytes, and red blood cells. It consists of a very small polypeptide chain of only 27 amino acids, which is phospholipid anchored and highly glycosylated resulting in molecular masses of up to 70 kDa in lymphoid cells. Glycosylation varies strongly among different types of leukocytes. Different isoforms of HSA were purified from different types of leukocytes, coated on microtiter plates, and binding of E- and P-selectin-Ig chimeras was analyzed. No binding was observed with E-selectin-Ig, whereas P-selectin-Ig chimera bound well to HSA from neutrophils, B cells, and a monocytic cell line, and binding was not seen to HSA from red blood cells (285). Antibodies against mouse HSA could partially block the binding of mouse neutrophils and monocytic cells to LPS-activated mouse endothelioma cells (bEND.3) and blocked the binding of HSA-coated latex beads to endothelioma cells or platelets (3). All these interactions were also sensitive to a P-selectin blocking antibody. The relevance of HSA/CD24 for the binding of neutrophils to P-selectin is not yet clarified, especially since PSGL-1 appears to be the dominant ligand. However, a breast and a small cell lung carcinoma cell line (both of human origin) that are negative for PSGL-1 were shown to bind to P-selectin. CD24 purified from these cells and coated onto latex beads bound to P-selectin-Ig chimeras and P-selectin-transfected cells in a Ca2+-dependent way (4). CD24 transfected human adenocarcinoma cells showed increased binding to P-selectin-expressing platelets. Adding soluble, purified CD24 to the assay and removing CD24 with phospholipase C from the cell surface was recently shown to block rolling of the PSGL-1 negative breast carcinoma cell line on P-selectin; the ability to roll was positively correlated with the expression level of CD24 (2). It is possible that CD24 is also involved in the rolling of the PSGL-1-negative human colon carcinoma cell line KM12-L4 on P-selectin (113).

2. L-selectin ligands

Four glycoprotein ligands have been identified for L-selectin: GlyCAM-1, CD34, MAdCAM-1, and Sgp200 (the latter is not yet cloned). All of them are expressed as L-selectin-binding glycoforms by HEV of lymph nodes. Both GlyCAM-1 and CD34 were identified by affinity isolation with an L-selectin-Ig chimera from 35SO4-labeled mouse lymph node tissue (
146). The two proteins were first named Sgp50 and Sgp90, according to their apparent molecular weights. Sufficient quantities of these ligands were purified, and microsequencing led to the identification of Sgp50 as a new soluble sialomucin (GlyCAM-1) (187) and Sgp90, a known antigen (CD34) (22).

Glycosylation-dependent cell adhesion molecule-1 is a 50-kDa secreted sialomucin that is specifically synthesized by HEV endothelial cells where its expression is affected by afferent lymphatic flow (224). It is also expressed as a nonbinding glycoform by mammary epithelial cells, where its expression is induced by lactation (85, 246). The gene for murine GlyCAM-1 was cloned and characterized (84, 87), and the GlyCAM-1 homologs were cloned in the rat (86) and in bovine (148, 149). Glycosylation-dependent cell adhesion molecule-1 can bind to all three selectins (225, 297).

The posttranslational modifications of GlyCAM-1 have been intensively studied. Glycosylation-dependent cell adhesion molecule-1 needs to be sulfated on oligosaccharide side chains to bind to L-selectin (145). The major sulfated mono- and disaccharides on GlyCAM-1 were identified as Gal-6-SO4 , GlcNAc-6-SO4 , (SO4-6)Galbeta 1right-arrow4GlcNAc, and Galbeta 1right-arrow4(SO4-6)GlcNAc (131). With the use of lectins and exoglycosidases, a major capping structure of GlyCAM-1 was identified that contained all three structural elements known to be critical for L-selectin binding: sialaic acid, fucose, and sulfate. This capping structure was determined as 6'-sulfated sialyl Lewisx: Siaalpha 2right-arrow3(SO4-6)Galbeta 1right-arrow4(Fucalpha 1right-arrow3)GlcNAc (134). By examining the complete structure of beta -eliminated oligosaccharide side chains of GlyCAM-1, two sulfated O-glycans were identified which represent 6'-sulfo-sLex (with SO4 linked to position 6 of Gal) and 6-sulfo-sLex (with SO4 linked to position 6 of GlcNAc) (133) (Fig. 5). Both these oligosaccharides were found in core-2 structures, i.e., beta 1right-arrow6 linked to GalNAc. Comparison of various sulfated Lex analogs revealed that 6-sulfo-sLex can block the binding of L-selectin-Ig to GlyCAM-1 better than sLex or 6'-sulfo-sLex (287). In another report, in contrast, CHO cells modified on their cell surface with 6'-sulfo-sLex were reported to bind to plastic-coated L-selectin-Ig chimera,