Cell, Vol. 62, 3-6,
July 13, 1990, Copyright
0 1990 by Cell Press
Leukocyte Adhesion to Endothelium in Inflammation Laurelee Osborn Biogen, Inc. 14 Cambridge Center Cambridge, Massachusetts
02142
The endothelium is a dynamic tissue, controlling, among other processes, traffic of molecules and cells between the blood and sites of immunologic challenge. A crucial, early step in mounting an effective inflammatory or immune response is the promotion of leukocyte adhesion to the vascular endothelium; since circulating leukocytes can crawl but not swim, they must attach to a surface before they can migrate chemotactically. Both endothelial cells and leukocytes undergo rapid change in the milieu of inflammatory mediators established by immune or nonimmune activation of the body’s defense mechanisms. Endothelial cells lining the postcapillary venules and microcirculation elaborate leukocyte-specific adhesion molecules both constitutively and in response to a wide range of inflammatory mediators. Circulating leukocytes routinely display receptors for these inducible endothelial cell adhesion molecules (CAMS). Equally important, when stimulated they undergo further changes in their adhesive properties for both resting and activated endothelium. Within the last two years a number of endothelial CAMS and their leukocyte ligands have been cloned and characterized, providing an increasingly detailed picture of how leukocytes are recruited from the bloodstream and guided to their targets. This review will describe these molecules and the current understanding of their structure and function. Endothellal CAMS Are Selective for Leukocyte Class Leukocytes are often divided into three major specialized types. Neutrophils, lymphocytes, and monocytes migrate into inflamed tissue sites with class-specific kinetics, as shown in skin-window and histological studies. Within minutes of stimulation, neutrophils arrive. They are the most abundant leukocytes in blood and are also known as polymorphonuclear cells, PMNs, “polys,” or granulocytes (of which they are a subclass). Neutrophils are relatively nonspecific, easily activated phagocytic cells, as is appropriate to their role as the first line of defense. Antigenspecific lymphocytes (T and B cells) and monocytes arrive within hours and may remain for days, targeting and destroying foreign cells and molecules. Sites of recalcitrant inflammation or infection such as granulomas, foreign bodies, or tubercular lesions can cause a chronic immigration and accumulation of differentiated monocytes called macrophages. These are phagocytic cells, enlarged and activated in response to inflammatory cytokines such as interferon y (IFNy), interleukin 1 (ILl), and tumor necrosis factor a (TNFa). Such mediators are released by a number of cell types, including lymphocytes, endothelial cells, and macrophages themselves. Activation stimulates all these cells to secrete more inflammatory mediators,
Minireview
setting up a positive feedback cycle that continues until the irritant has been eliminated. The endothelial response to inflammatory mediators can account for this hierarchy of lines of defense. As detailed below, each endothelial CAM preferentially (though not exclusively) recognizes a subset of leukocytes and is expressed according to an individualized time course in response to a subset of inflammatory mediators. It is convenient to consider migration of each class of leukocyte separately, although in reality there is overlap in the time course of migration and molecular pathways used by each class. Structures of Endothelial Leukocyte Receptors and Their Ligands As outlined in Table 1, currently known endothelial-leukocyte receptors belong to three structural groups: the immunoglobulin (lg) gene superfamily, the integrin family, and a newly discovered family of proteins called selectins or LECCAMs (for their homology to lectins, epidermal growth factor, and complement regulatory proteins). The lg gene superfamily (reviewed in Hunkapillar and Hood, 1989) encompasses a large group of molecules involved in recognition and adhesion, and is characterized by the presence of one or more lg homology units. Each unit is usually encoded by a discrete exon and consists of a primary sequence of 70-110 residues with a disulfide bridge spanning 50-70 residues; several other conserved residues are involved in establishing a tertiary structure referred to as an antibody fold. This structure provides a molecular scaffold on which countless variations of recognition sequences can be displayed, as demonstrated by the enormous diversity of the Igs themselves. On endothelial cells, ICAMl, ICAM2, and VCAMl (or INCAM-110) recognize their leukocytic ligands and enable the adhesion and migration of these cells out of the bloodstream, as described below. Ligands of the ICAMs and VCAMl are members of the widely distributed integrin family of adhesion molecules (reviewed by Hynes, 1987; Hemler, 1988), which mediate cell-extracellular matrix and cell-cell binding. The integrins are noncovalent heterodimers composed of an a and a 8 polypeptide, each of which has a large extracellular domain, a transmembrane segment, and a short cytoplasmic tail of 30-50 amino acids. Currently, more than ten a chains have been characterized, ranging in size between 130 and 210 kd. At least seven 8 subunits have been characterized to date. Three integrin subfamilies are commonly recognized, based on the three first-characterized 8 chains: the 8, (VLA), p2 (LeuCAM), and B3 (cytoadhesin) integrins. In contrast to the multifunctional lg and integrin families, the currently known members of the LECCAM or selectin family (ELAMl on endothelium, the lymphocyte homing receptor [mLHR] on murine leukocytes and its human homolog Leu8 or LAMl, and PADGEM of GMP140 on platelets and endothelium) all function in leukocyte-endothelial adhesion. Their structure is an interesting me-
Cell 4
Table
1. Endothelial
Cell-Leukocyte
Adhesion
Molecules Ligand Structure
Endothelial CAM
Structure
M, Size
Cell Type
Known
ICAMl
lg
100 kd
LFAl ; Mac1 7
92 integrin
ICAMP
ICI
mRNA
activated ECs; other ceils ECs
LFAl ; Macl?
f3* integrin
VCAMl ELAMl GMP140
kl LECCAM LECCAM
100 kd 115 kd 140 kd
activated activated activated activated
vLA4 ? ?
9, integrin 7 ?
1.4 kb
ECs = endothelial cells. See text for other abbreviations. References: ICAMl, Simmons et al., 1999; ICAMP, Staunton Johnston et al., 1989; see also text.
ECs ECs ECs platelets
et al., 1999; VCAMl,
lange of adhesive motifs (Bevilacqua et al., 1989). Their amino-terminal domains resemble the carbohydrate binding domain of animal lectins, suggesting that they bind to sugars. However, they also have an EGF-homologous domain and a number of cysteine-rich tandem motifs homologous to the short consensus repeats typical of complement regulatory proteins, which could bind and perhaps signal other cells, as recently shown for the EGF-homologous domain of mLHR (Siegelman et al., 1990). The ligands of ELAMl, mLHR, LAMl, and GMP140 have not yet been characterized. PhYgocyfo Recru/fmenf Neutrophils represent 70% of a human’s circulating leukocytes. They respond to a wide range of activators and chemotactic factors, migrate quickly out of the bloodstream and into the tissue, and degranulate readily, releasing a host of defensive antimicrobial agents. Defects in neutrophil adhesion and migration cause recurrent, lifethreatening infections, as seen in patients suffering from leukocyte adhesion deficiency (LAD); in these individuals, neutrophils lack the 32 integrins and are therefore incapable of migration into inflammatory sites under most circumstances (reviewed in Anderson and Springer, 1987). Conversely, the normal neutrophil’s ease of activation and degranulation can cause destruction of viable host cells in the vascular bed and other tissues. This phenomenon is clearly seen in a number of serious, common illnesses involving ischemiafollowed by reperfusion, such as infarction of myocardial or other tissue and hypovolemic shock following trauma (Harlan, 1987). Currently there appear to be two basic pathways by which neutrophils are arrested in the circulation near sites of tissue damage or infection, both of which are activated relatively early in the inflammatory process. One of these pathways is immediate and does not require protein synthesis. It appears to be partly due to a transient increase in adhesiveness of the neutrophil for resting endothelial cells, as well as to a transient increase in endothelial adhesiveness for neutrophils. The other pathway is activated within l-2 hr of the inflammatory stimulus, requires protein synthesis, and depends on endothelial cell elaboration of surface molecules that bind to both resting and activated neutrophils.
Ligands
Osborn
et al., 1999; ElAMl,
Ligand
Cell Type
lymphocytes; PMNs; monocytes lymphocytes; PMNs; monocytes lymphocytes; monocytes PMNs; monocytes PMNs; monocytes PMNs; monocytes
Bevilacqua
et al., 1989; GMP140,
Within seconds of exposure to a wide range of substances including leukotrienes, platelet-activating factor, complement fragment C5a, bacterial lipopolysaccharide, and TNFc, neutrophils become markedly more adhesive for endothelium. This adhesion appears to be mediated primarily by a transient effect on the 32 integrins LFAl and Macl, as it can be partially blocked by monoclonal antibodies (MAbs) to either of these molecules, and almost completely blocked by MAbs to both (Lo et al., 1989). LFAl binds to both ICAMl, which is inducible on endothelium (although not quickly enough to account for the immediate response, see below), and ICAMP, which is constitutively present. It seems likely that Mac1 also binds to both these endothelial receptors, although this is controversial. Activation of one or more endothelial molecules may also contribute to transient neutrophil-endothelial adhesion in vivo. Although direct evidence has been lacking until very recently, participation of the endothelium in the rapid adhesive response is an attractive model, as it would provide a defined target for the adhesion and emigration of neutrophils to the site where they are immediately needed. The best candidate for such an endothelial CAM is GMP140 or PADGEM, a LECCAM that has been shown to bind neutrophils when translocated from platelet secretory granules to their plasma membranes (Larsen et al., 1989). GMP140 is translocated from analogous endothelial granules to the endothelial cell surface within 5 min after treatment by thrombin in culture (Hattori et al., 1989). Recently it has been shown that a MAb to GMP140 inhibits binding of neutrophils to endothelial cells activated with phorbol ester or histamine for 30 min, providing more direct evidence that endothelial as well as platelet GMP140 is functional in binding neutrophils at early time points after stimulation. Fixed neutrophils or HL80 cells will bind to GMP140-coated plastic, indicating that activation of neutrophils is unnecessary for binding to GMP140, and implying that resting neutrophils in the circulation could be recruited (and then quickly activated) within a few minutes after release of fast-acting mediators such as histamine (Geng et al., 1990). Interestingly, GMP140 is structurally similar to ELAMl, a neutrophil binding protein that appears on endothelium about 2 hr after exposure to IL1 or TNFa (see below).
Minireview 5
The second pathway of neutrophii recruitment mediates migration into tissues within a few hours after the original insult. Two adhesion molecules, ICAMl and ELAMl, that bind neutrophils are strongly induced on endothelial cells within 2 hr after exposure to inflammatory cytokines. ICAMl is an lg superfamily member that is present on many cell types constitutively or inducibly, is induced more than 30-fold on cultured human umbilical vein endothelial cells (HUVECs) by IFNy, ILl, or TNFa, and can bind LFAl (CDllaKD18) (and possibly Macl, CDllW CD18, as mentioned above), present on all leukocytes. ELAMl belongs to the LECCAM family, is induced more than lOO-fold on HUVECs by IL1 or TNFa but not IFNy, and has been found only on induced endothelial cells. Its ligand has not been identified. What roles do ICAMl and ELAMl play in neutrophil recruitment? ICAMl is probably important in both adhesion and migration, assuming that it is the major ligand for CDllKD18. Antibodies to CD18 can completely inhibit migration of neutrophils into certain tissues in vivo, and as mentioned above, LAD patients, who have a genetic defect in the 6z integrin CD18, show an almost total lack of neutrophil migration to sites of inflammation, resulting in a seriously compromised ability to fight infection. Thus a CD1 KDlSdependent mechanism is necessary (although perhaps not sufficient) for neutrophil recruitment in response to infection. The relative contribution of the two known endothelial ligands for CDllICD18, the inducible ICAMl and the constitutive ICAMP, to this process remains to be determined. What is the function of ELAMl? In vitro, antibody to CD18 blocks neutrophil binding to 4 hr cytokine-stimulated HUVECs by 50%-60%, and anti-ELAMl antibodies block most of the remainder of this binding. It is not yet known what role ELAMl plays in vivo. It may act to recruit neutrophils in tissues or circumstances in which the ICAMl-CDll/CD18 interaction is relatively unimportant. Alternatively, it may function as a preliminary step to CDll/ CDl&dependent adhesion and migration; one such scenario has been suggested by Lawrence et al. (1990) who measured CDl&dependent and -independent adhesion to Ill-treated HUVECs at different shear stresses, to mimic venous flow conditions. They found that CD18-independent (presumably ELAMl-mediated) adhesion withstands significantly greater shear stresses than does CDlB-dependent adhesion, supporting a model in which ELAMl helps “catch” the circulating phagocyte and then passes it to ICAMl or ICAM2, resulting in CDlbdependent migration across the endothelium. Published information on monocyte migration lags behind that on the neutrophil, perhaps due to the monocyte’s relative scarcity and the difficulty of isolating this cell type from peripheral blood. Monocytes share certain adhesion molecules with both neutrophils (e.g., Macl, the GMP140/PADGEM ligand) and lymphocytes (VLA4, LFAl), implying that mechanisms of migration will overlap in these leukocyte classes. The rapidly progressing generation of antibodies and cDNA clones to both endothelial and leukocyte adhesion molecules should facilitate study of monocyte-endothelial adhesion and migration.
Lymphocyte Recruitment Both T and B lymphocytes leave the bloodstream for two destinations, lymphoid tissue or sites of inflammation. To enter lymphoid tissue, they bind and migrate via adhesion molecules termed homing receptors through high endothelial venules (HEVs), areas of activated endothelium characterized by columnar rather than flat cells. The molecules on HEVs that bind lymphocytes are known as addressins, because there are antigenically distinct forms present in various types of lymphoid tissue such as gut or Peyer’s patch, peripheral lymph nodes, and lung-associated lymphoid tissue. When lymphocytes enter the lymphoid tissue, they are exposed to an array of antigen-presenting cells and are thus stimulated to become activated and multiply if their particular antigen is currently present in the body. Although lymphoid tissue can be activated, the addressins are presumed to be constitutively present on HEVs. In contrast, to enter sites of inflammation, circulating lymphocytes bind and migrate through temporarily activated endothelium; similar in appearance to HEVs, the activated endothelium is likely induced by local release of inflammatory mediators in the affected tissue. At sites of inflammation, lymphocytes in concert with cells of the monocytelmacrophage lineage accomplish their task of recognizing and destroying malignant cells or pathogens displaying foreign antigens, Comparison of these two systems has been complicated because almost all homing receptor research has been done in murine tissue sections, while the response of endothelium to cytokines has been extensively studied in cultured human endothelial cells. It appears that some overlap between the two systems exists; for example, a murine homolog of VLA4 is involved in homing to Peyer’s patches (Holzmann and Weissman, 1989), while in humans VLM serves as a ligand for VCAM, a molecule induced by inflammatory cytokines on the nonlymphoid endothelium (see below). The ongoing molecular cloning of lymphoid and endothelial adhesion molecules from both mouse and human, accompanied in some cases by their functional expression, will facilitate direct comparison of results from these lines of work. Lymphocyte homing has been recently reviewed by Stoolman (1989) and Jutila et al. (1990). In cultured human endothelial cells, two types of molecules are known that affect lymphocyte adhesion. ICAMl and/or ICAM is thought to be responsible for the basal binding of lymphocytes to unstimulated HUVECs, indicated by the ability of MAbs against CD18 to inhibit this binding almost completely. Although ICAMl is strongly induced on cytokine-stimulated endothelial cells, evidence obtained both in vivo and in vitro suggests that one or more other CAMS are required for lymphocyte recruitment. LAD patients, who lack CD18 and therefore the known lymphocyte ligand of ICAMl and ICAMP, demonstrate essentially normal migration of lymphocytes, in contrast to the seriously defective migration of neutrophils seen in these patients. In vitro, binding of T lymphocytes from normal donors to cytokine-treated HUVECs can be inhibited only partially by an anti-CD18 MAb, indicating
Cell 6
the presence of at least one more inducible molecule for mediating binding (Dustin and Springer, 1988). Such a molecule, VCAMl, has recently been cloned; it is induced strongly on cytokine-stimulated endothelial cells and like the ICAMs is a member of the lg superfamily (Osborn et al., 1989; Rice et al., 1990). It binds lymphocytic and monocytic cell lines via the 3, integrin VLM (which is present constitutivelyon most leukocytes), and binds to a different site on VLA4 than that which binds fibronectin (Elites et al., 1990). VCAMl is present on the cell surface by 3 hr after cytokine stimulation and remains for at least 72 hr in the continued presence of cytokine, consistent with the prolonged time course of lymphocyte and monocyte migration seen in inflammatory lesions in vivo. MAbs to VCAMl and to VLM can be expected to help determine physiological and pathological functions of this newly defined receptor-ligand pair. It is likely to play a role in inflammatory disorders with an immune component, such as rheumatoid arthritis, asthma, inflammatory bowel diseases, sepsis, and dermatoses. POt8nt/8/ hf/8MI8tOfy
for
ht8fVSlttiOfl
ifl th8
hOC88S
It is clear that the recruitment of leukocytes from the bloodstream into areas of injury or infection is essential for host defense, but also that many acute and chronic allergic, autoimmune, or inflammatory diseases are exacerbated by the presence of excess leukocytes. The potential of adhesion-blocking reagents to reduce or prevent leukocyte-mediated tissue damage has been demonstrated in a number of animal model systems. Many of the published reports have utilized MAb 80.3, an antibody to CD18 that prevents binding of leukocytes via LFAl or Mac1 to endothelial cells. This antibody cross-reacts with rabbit and primate CD18 and has a remarkable ability to prevent various forms of neutrophil-mediated ischemia-reperfusion damage in these animals, when administered at the time of reperfusion (reviewed by Carlos and Harlan, 1990). Therapeutic blocking of the endothelial rather than the leukocytic adhesion partner was demonstrated in a model of bronchial asthma in which airway eosinophilia and hyperresponsiveness were attenuated by daily intravenous treatment of primates with a MAb to ICAMl (Wegner et al., 1990). These results are intriguing, not only as proof that leukocyte migration is an important factor in these disease states, but as evidence that new therapies could be based on blocking specific adhesion pathways. Currently available antiinflammatory drugs have limited efficacy in interrupting the self-perpetuating cycle of tissue damage seen in chronic diseases of this type. Furthermore, many of these drugs have severe side effects. Therefore it would be useful to have other forms of therapy that could replace or alternate with currently used drugs. Although the MAbs that are now being used experimentally to block leukocyte migration have theoretical drawbacks as drugs to treat chronic diseases, the expected side effects, such as vascular damage due to immune complex deposition on the endothelium, may not inevitably occur in practice (see, e.g., Wegner et al., 1990). Alternatively, it may be possible to develop small peptides or other molecules that will specifically block adhesive inter-
actions of ligand-receptor pairs. Thus, study of leukocyte adhesion and migration through the endothelium may make accessible a new point of intervention in both acute and chronic inflammatory disorders.
Anderson, 175-194.
D. C., and
Springer,
T. A. (1987).
Bevilacqua, M. P, Stengelin, S., Gimbrone, (1969). Science 243, 1160-1165. Carlo%
T. M., and Harlan,
J. M. (1990).
DUStin, fvt. L., and Springer, Elites, M. J., Csborn, ler, M. E., and Lobb,
Annu.
Med.
38,
M. A., Jr., and Seed,
Immunol.
T A. (1988).
Rev.
6.
Rev., in press.
J. Cell Biol. 707, 321-331.
L., Takada, Y., Grouse, C., Luhowskyj, R. R. (lQ90). Cell 60, 577-584.
S., Hem-
Geng, J.-G., Bevilacqua, M. P.. Moore, K. L., McIntyre, T. M., Prescott, S. M., Kim, J. M., Bliss, G. A., Zimmerman, G. A., and McEver, R. P (1990). Nature 343, 757-760. Harlan,
J. M. (1987). Acta Med. Stand.
775, 123-129.
Hattori, R., Hamilton, K. K., Fugate, R. D., McEver, P. J. (1989). J. Biol. Chem. 264, 7766-7771. Hemler,
M. E. (1988).
Holzmann, Hunkapillar, Hynes,
Immunol.
B., and Weissman, T., and Hood,
R. P, and Sims,
Today 9, 109-113. I. L. (1989).
L. (1989).
EMBO
Adv. Immunol.
J. 8, 1735-1741. 44, l-63.
R. 0. (1987). Cell 46, 549-554.
Johnston, 10331044.
G. I., Cook,
R. G.,
and
McEver,
R. P. (1989).
Cell
56,
Jutila, M. A., Lewinsohn, D. M., Berg, E. L., and Butcher, E. C. (1990). In Leukocyte Adhesion Molecules, T. A. Springer, D. C. Anderson, A. S. Rosenthal, and R. Rothlein, eds. (New York: Springer-Verlag), pp. 227-235. Larsen, E., Celi, A., Gilbert, G. E., Furie, B. C., Erban, J. K., Bonfanti, R., Wagner, D. D., and Furie, B. (1989). Cell 59, 305-312. Lawrence, M. B., Smith, C. W., Eskin, Blood 75, 227-237.
S. G., and Mclntire,
L. V. (1990).
Lo, S. K., Van Seventer, G. A., Levin, J. Immunol. 743, 3325-3329.
S. M., and Wright,
S. D. (1989).
Osborn, L., Hession, C., Tizard, Rosso, G., and Lobb, R. (1989). Rice, G. E., Munro, 777. 1369-1374.
R., Vassallo, C., Luhowskyj, Cell 59, 1203-1211.
J. M., and Bevilacqua,
Siegelman, M. H., Cheng, (1990): Cell 61, 611-622.
I. C., Weissman,
Simmons,
M. W., and
D., Makgoba,
Seed,
M. P (1990).
S., Chi-
J. Exp. Med.
I. L., and Wakeland, B. (1988).
E. K.
Nature
331,
T. A. (1989). Nature
339,
624-627. Staunton, 61-64.
D. E., Dustin,
M. L., and Springer,
Stoolman,
L. M. (1989).
Cell 56, 907-910.
Wegner, C. D., Gundel, R. H., Reilly, P, Haynes, Rothlein, R. (1990). Science 247, 458-459.
N., Letts,
L. G., and
July 13, 1990, Copyright
0 1990 by Cell Press
Leukocyte Adhesion to Endothelium in Inflammation Laurelee Osborn Biogen, Inc. 14 Cambridge Center Cambridge, Massachusetts
02142
The endothelium is a dynamic tissue, controlling, among other processes, traffic of molecules and cells between the blood and sites of immunologic challenge. A crucial, early step in mounting an effective inflammatory or immune response is the promotion of leukocyte adhesion to the vascular endothelium; since circulating leukocytes can crawl but not swim, they must attach to a surface before they can migrate chemotactically. Both endothelial cells and leukocytes undergo rapid change in the milieu of inflammatory mediators established by immune or nonimmune activation of the body’s defense mechanisms. Endothelial cells lining the postcapillary venules and microcirculation elaborate leukocyte-specific adhesion molecules both constitutively and in response to a wide range of inflammatory mediators. Circulating leukocytes routinely display receptors for these inducible endothelial cell adhesion molecules (CAMS). Equally important, when stimulated they undergo further changes in their adhesive properties for both resting and activated endothelium. Within the last two years a number of endothelial CAMS and their leukocyte ligands have been cloned and characterized, providing an increasingly detailed picture of how leukocytes are recruited from the bloodstream and guided to their targets. This review will describe these molecules and the current understanding of their structure and function. Endothellal CAMS Are Selective for Leukocyte Class Leukocytes are often divided into three major specialized types. Neutrophils, lymphocytes, and monocytes migrate into inflamed tissue sites with class-specific kinetics, as shown in skin-window and histological studies. Within minutes of stimulation, neutrophils arrive. They are the most abundant leukocytes in blood and are also known as polymorphonuclear cells, PMNs, “polys,” or granulocytes (of which they are a subclass). Neutrophils are relatively nonspecific, easily activated phagocytic cells, as is appropriate to their role as the first line of defense. Antigenspecific lymphocytes (T and B cells) and monocytes arrive within hours and may remain for days, targeting and destroying foreign cells and molecules. Sites of recalcitrant inflammation or infection such as granulomas, foreign bodies, or tubercular lesions can cause a chronic immigration and accumulation of differentiated monocytes called macrophages. These are phagocytic cells, enlarged and activated in response to inflammatory cytokines such as interferon y (IFNy), interleukin 1 (ILl), and tumor necrosis factor a (TNFa). Such mediators are released by a number of cell types, including lymphocytes, endothelial cells, and macrophages themselves. Activation stimulates all these cells to secrete more inflammatory mediators,
Minireview
setting up a positive feedback cycle that continues until the irritant has been eliminated. The endothelial response to inflammatory mediators can account for this hierarchy of lines of defense. As detailed below, each endothelial CAM preferentially (though not exclusively) recognizes a subset of leukocytes and is expressed according to an individualized time course in response to a subset of inflammatory mediators. It is convenient to consider migration of each class of leukocyte separately, although in reality there is overlap in the time course of migration and molecular pathways used by each class. Structures of Endothelial Leukocyte Receptors and Their Ligands As outlined in Table 1, currently known endothelial-leukocyte receptors belong to three structural groups: the immunoglobulin (lg) gene superfamily, the integrin family, and a newly discovered family of proteins called selectins or LECCAMs (for their homology to lectins, epidermal growth factor, and complement regulatory proteins). The lg gene superfamily (reviewed in Hunkapillar and Hood, 1989) encompasses a large group of molecules involved in recognition and adhesion, and is characterized by the presence of one or more lg homology units. Each unit is usually encoded by a discrete exon and consists of a primary sequence of 70-110 residues with a disulfide bridge spanning 50-70 residues; several other conserved residues are involved in establishing a tertiary structure referred to as an antibody fold. This structure provides a molecular scaffold on which countless variations of recognition sequences can be displayed, as demonstrated by the enormous diversity of the Igs themselves. On endothelial cells, ICAMl, ICAM2, and VCAMl (or INCAM-110) recognize their leukocytic ligands and enable the adhesion and migration of these cells out of the bloodstream, as described below. Ligands of the ICAMs and VCAMl are members of the widely distributed integrin family of adhesion molecules (reviewed by Hynes, 1987; Hemler, 1988), which mediate cell-extracellular matrix and cell-cell binding. The integrins are noncovalent heterodimers composed of an a and a 8 polypeptide, each of which has a large extracellular domain, a transmembrane segment, and a short cytoplasmic tail of 30-50 amino acids. Currently, more than ten a chains have been characterized, ranging in size between 130 and 210 kd. At least seven 8 subunits have been characterized to date. Three integrin subfamilies are commonly recognized, based on the three first-characterized 8 chains: the 8, (VLA), p2 (LeuCAM), and B3 (cytoadhesin) integrins. In contrast to the multifunctional lg and integrin families, the currently known members of the LECCAM or selectin family (ELAMl on endothelium, the lymphocyte homing receptor [mLHR] on murine leukocytes and its human homolog Leu8 or LAMl, and PADGEM of GMP140 on platelets and endothelium) all function in leukocyte-endothelial adhesion. Their structure is an interesting me-
Cell 4
Table
1. Endothelial
Cell-Leukocyte
Adhesion
Molecules Ligand Structure
Endothelial CAM
Structure
M, Size
Cell Type
Known
ICAMl
lg
100 kd
LFAl ; Mac1 7
92 integrin
ICAMP
ICI
mRNA
activated ECs; other ceils ECs
LFAl ; Macl?
f3* integrin
VCAMl ELAMl GMP140
kl LECCAM LECCAM
100 kd 115 kd 140 kd
activated activated activated activated
vLA4 ? ?
9, integrin 7 ?
1.4 kb
ECs = endothelial cells. See text for other abbreviations. References: ICAMl, Simmons et al., 1999; ICAMP, Staunton Johnston et al., 1989; see also text.
ECs ECs ECs platelets
et al., 1999; VCAMl,
lange of adhesive motifs (Bevilacqua et al., 1989). Their amino-terminal domains resemble the carbohydrate binding domain of animal lectins, suggesting that they bind to sugars. However, they also have an EGF-homologous domain and a number of cysteine-rich tandem motifs homologous to the short consensus repeats typical of complement regulatory proteins, which could bind and perhaps signal other cells, as recently shown for the EGF-homologous domain of mLHR (Siegelman et al., 1990). The ligands of ELAMl, mLHR, LAMl, and GMP140 have not yet been characterized. PhYgocyfo Recru/fmenf Neutrophils represent 70% of a human’s circulating leukocytes. They respond to a wide range of activators and chemotactic factors, migrate quickly out of the bloodstream and into the tissue, and degranulate readily, releasing a host of defensive antimicrobial agents. Defects in neutrophil adhesion and migration cause recurrent, lifethreatening infections, as seen in patients suffering from leukocyte adhesion deficiency (LAD); in these individuals, neutrophils lack the 32 integrins and are therefore incapable of migration into inflammatory sites under most circumstances (reviewed in Anderson and Springer, 1987). Conversely, the normal neutrophil’s ease of activation and degranulation can cause destruction of viable host cells in the vascular bed and other tissues. This phenomenon is clearly seen in a number of serious, common illnesses involving ischemiafollowed by reperfusion, such as infarction of myocardial or other tissue and hypovolemic shock following trauma (Harlan, 1987). Currently there appear to be two basic pathways by which neutrophils are arrested in the circulation near sites of tissue damage or infection, both of which are activated relatively early in the inflammatory process. One of these pathways is immediate and does not require protein synthesis. It appears to be partly due to a transient increase in adhesiveness of the neutrophil for resting endothelial cells, as well as to a transient increase in endothelial adhesiveness for neutrophils. The other pathway is activated within l-2 hr of the inflammatory stimulus, requires protein synthesis, and depends on endothelial cell elaboration of surface molecules that bind to both resting and activated neutrophils.
Ligands
Osborn
et al., 1999; ElAMl,
Ligand
Cell Type
lymphocytes; PMNs; monocytes lymphocytes; PMNs; monocytes lymphocytes; monocytes PMNs; monocytes PMNs; monocytes PMNs; monocytes
Bevilacqua
et al., 1989; GMP140,
Within seconds of exposure to a wide range of substances including leukotrienes, platelet-activating factor, complement fragment C5a, bacterial lipopolysaccharide, and TNFc, neutrophils become markedly more adhesive for endothelium. This adhesion appears to be mediated primarily by a transient effect on the 32 integrins LFAl and Macl, as it can be partially blocked by monoclonal antibodies (MAbs) to either of these molecules, and almost completely blocked by MAbs to both (Lo et al., 1989). LFAl binds to both ICAMl, which is inducible on endothelium (although not quickly enough to account for the immediate response, see below), and ICAMP, which is constitutively present. It seems likely that Mac1 also binds to both these endothelial receptors, although this is controversial. Activation of one or more endothelial molecules may also contribute to transient neutrophil-endothelial adhesion in vivo. Although direct evidence has been lacking until very recently, participation of the endothelium in the rapid adhesive response is an attractive model, as it would provide a defined target for the adhesion and emigration of neutrophils to the site where they are immediately needed. The best candidate for such an endothelial CAM is GMP140 or PADGEM, a LECCAM that has been shown to bind neutrophils when translocated from platelet secretory granules to their plasma membranes (Larsen et al., 1989). GMP140 is translocated from analogous endothelial granules to the endothelial cell surface within 5 min after treatment by thrombin in culture (Hattori et al., 1989). Recently it has been shown that a MAb to GMP140 inhibits binding of neutrophils to endothelial cells activated with phorbol ester or histamine for 30 min, providing more direct evidence that endothelial as well as platelet GMP140 is functional in binding neutrophils at early time points after stimulation. Fixed neutrophils or HL80 cells will bind to GMP140-coated plastic, indicating that activation of neutrophils is unnecessary for binding to GMP140, and implying that resting neutrophils in the circulation could be recruited (and then quickly activated) within a few minutes after release of fast-acting mediators such as histamine (Geng et al., 1990). Interestingly, GMP140 is structurally similar to ELAMl, a neutrophil binding protein that appears on endothelium about 2 hr after exposure to IL1 or TNFa (see below).
Minireview 5
The second pathway of neutrophii recruitment mediates migration into tissues within a few hours after the original insult. Two adhesion molecules, ICAMl and ELAMl, that bind neutrophils are strongly induced on endothelial cells within 2 hr after exposure to inflammatory cytokines. ICAMl is an lg superfamily member that is present on many cell types constitutively or inducibly, is induced more than 30-fold on cultured human umbilical vein endothelial cells (HUVECs) by IFNy, ILl, or TNFa, and can bind LFAl (CDllaKD18) (and possibly Macl, CDllW CD18, as mentioned above), present on all leukocytes. ELAMl belongs to the LECCAM family, is induced more than lOO-fold on HUVECs by IL1 or TNFa but not IFNy, and has been found only on induced endothelial cells. Its ligand has not been identified. What roles do ICAMl and ELAMl play in neutrophil recruitment? ICAMl is probably important in both adhesion and migration, assuming that it is the major ligand for CDllKD18. Antibodies to CD18 can completely inhibit migration of neutrophils into certain tissues in vivo, and as mentioned above, LAD patients, who have a genetic defect in the 6z integrin CD18, show an almost total lack of neutrophil migration to sites of inflammation, resulting in a seriously compromised ability to fight infection. Thus a CD1 KDlSdependent mechanism is necessary (although perhaps not sufficient) for neutrophil recruitment in response to infection. The relative contribution of the two known endothelial ligands for CDllICD18, the inducible ICAMl and the constitutive ICAMP, to this process remains to be determined. What is the function of ELAMl? In vitro, antibody to CD18 blocks neutrophil binding to 4 hr cytokine-stimulated HUVECs by 50%-60%, and anti-ELAMl antibodies block most of the remainder of this binding. It is not yet known what role ELAMl plays in vivo. It may act to recruit neutrophils in tissues or circumstances in which the ICAMl-CDll/CD18 interaction is relatively unimportant. Alternatively, it may function as a preliminary step to CDll/ CDl&dependent adhesion and migration; one such scenario has been suggested by Lawrence et al. (1990) who measured CDl&dependent and -independent adhesion to Ill-treated HUVECs at different shear stresses, to mimic venous flow conditions. They found that CD18-independent (presumably ELAMl-mediated) adhesion withstands significantly greater shear stresses than does CDlB-dependent adhesion, supporting a model in which ELAMl helps “catch” the circulating phagocyte and then passes it to ICAMl or ICAM2, resulting in CDlbdependent migration across the endothelium. Published information on monocyte migration lags behind that on the neutrophil, perhaps due to the monocyte’s relative scarcity and the difficulty of isolating this cell type from peripheral blood. Monocytes share certain adhesion molecules with both neutrophils (e.g., Macl, the GMP140/PADGEM ligand) and lymphocytes (VLA4, LFAl), implying that mechanisms of migration will overlap in these leukocyte classes. The rapidly progressing generation of antibodies and cDNA clones to both endothelial and leukocyte adhesion molecules should facilitate study of monocyte-endothelial adhesion and migration.
Lymphocyte Recruitment Both T and B lymphocytes leave the bloodstream for two destinations, lymphoid tissue or sites of inflammation. To enter lymphoid tissue, they bind and migrate via adhesion molecules termed homing receptors through high endothelial venules (HEVs), areas of activated endothelium characterized by columnar rather than flat cells. The molecules on HEVs that bind lymphocytes are known as addressins, because there are antigenically distinct forms present in various types of lymphoid tissue such as gut or Peyer’s patch, peripheral lymph nodes, and lung-associated lymphoid tissue. When lymphocytes enter the lymphoid tissue, they are exposed to an array of antigen-presenting cells and are thus stimulated to become activated and multiply if their particular antigen is currently present in the body. Although lymphoid tissue can be activated, the addressins are presumed to be constitutively present on HEVs. In contrast, to enter sites of inflammation, circulating lymphocytes bind and migrate through temporarily activated endothelium; similar in appearance to HEVs, the activated endothelium is likely induced by local release of inflammatory mediators in the affected tissue. At sites of inflammation, lymphocytes in concert with cells of the monocytelmacrophage lineage accomplish their task of recognizing and destroying malignant cells or pathogens displaying foreign antigens, Comparison of these two systems has been complicated because almost all homing receptor research has been done in murine tissue sections, while the response of endothelium to cytokines has been extensively studied in cultured human endothelial cells. It appears that some overlap between the two systems exists; for example, a murine homolog of VLA4 is involved in homing to Peyer’s patches (Holzmann and Weissman, 1989), while in humans VLM serves as a ligand for VCAM, a molecule induced by inflammatory cytokines on the nonlymphoid endothelium (see below). The ongoing molecular cloning of lymphoid and endothelial adhesion molecules from both mouse and human, accompanied in some cases by their functional expression, will facilitate direct comparison of results from these lines of work. Lymphocyte homing has been recently reviewed by Stoolman (1989) and Jutila et al. (1990). In cultured human endothelial cells, two types of molecules are known that affect lymphocyte adhesion. ICAMl and/or ICAM is thought to be responsible for the basal binding of lymphocytes to unstimulated HUVECs, indicated by the ability of MAbs against CD18 to inhibit this binding almost completely. Although ICAMl is strongly induced on cytokine-stimulated endothelial cells, evidence obtained both in vivo and in vitro suggests that one or more other CAMS are required for lymphocyte recruitment. LAD patients, who lack CD18 and therefore the known lymphocyte ligand of ICAMl and ICAMP, demonstrate essentially normal migration of lymphocytes, in contrast to the seriously defective migration of neutrophils seen in these patients. In vitro, binding of T lymphocytes from normal donors to cytokine-treated HUVECs can be inhibited only partially by an anti-CD18 MAb, indicating
Cell 6
the presence of at least one more inducible molecule for mediating binding (Dustin and Springer, 1988). Such a molecule, VCAMl, has recently been cloned; it is induced strongly on cytokine-stimulated endothelial cells and like the ICAMs is a member of the lg superfamily (Osborn et al., 1989; Rice et al., 1990). It binds lymphocytic and monocytic cell lines via the 3, integrin VLM (which is present constitutivelyon most leukocytes), and binds to a different site on VLA4 than that which binds fibronectin (Elites et al., 1990). VCAMl is present on the cell surface by 3 hr after cytokine stimulation and remains for at least 72 hr in the continued presence of cytokine, consistent with the prolonged time course of lymphocyte and monocyte migration seen in inflammatory lesions in vivo. MAbs to VCAMl and to VLM can be expected to help determine physiological and pathological functions of this newly defined receptor-ligand pair. It is likely to play a role in inflammatory disorders with an immune component, such as rheumatoid arthritis, asthma, inflammatory bowel diseases, sepsis, and dermatoses. POt8nt/8/ hf/8MI8tOfy
for
ht8fVSlttiOfl
ifl th8
hOC88S
It is clear that the recruitment of leukocytes from the bloodstream into areas of injury or infection is essential for host defense, but also that many acute and chronic allergic, autoimmune, or inflammatory diseases are exacerbated by the presence of excess leukocytes. The potential of adhesion-blocking reagents to reduce or prevent leukocyte-mediated tissue damage has been demonstrated in a number of animal model systems. Many of the published reports have utilized MAb 80.3, an antibody to CD18 that prevents binding of leukocytes via LFAl or Mac1 to endothelial cells. This antibody cross-reacts with rabbit and primate CD18 and has a remarkable ability to prevent various forms of neutrophil-mediated ischemia-reperfusion damage in these animals, when administered at the time of reperfusion (reviewed by Carlos and Harlan, 1990). Therapeutic blocking of the endothelial rather than the leukocytic adhesion partner was demonstrated in a model of bronchial asthma in which airway eosinophilia and hyperresponsiveness were attenuated by daily intravenous treatment of primates with a MAb to ICAMl (Wegner et al., 1990). These results are intriguing, not only as proof that leukocyte migration is an important factor in these disease states, but as evidence that new therapies could be based on blocking specific adhesion pathways. Currently available antiinflammatory drugs have limited efficacy in interrupting the self-perpetuating cycle of tissue damage seen in chronic diseases of this type. Furthermore, many of these drugs have severe side effects. Therefore it would be useful to have other forms of therapy that could replace or alternate with currently used drugs. Although the MAbs that are now being used experimentally to block leukocyte migration have theoretical drawbacks as drugs to treat chronic diseases, the expected side effects, such as vascular damage due to immune complex deposition on the endothelium, may not inevitably occur in practice (see, e.g., Wegner et al., 1990). Alternatively, it may be possible to develop small peptides or other molecules that will specifically block adhesive inter-
actions of ligand-receptor pairs. Thus, study of leukocyte adhesion and migration through the endothelium may make accessible a new point of intervention in both acute and chronic inflammatory disorders.
Anderson, 175-194.
D. C., and
Springer,
T. A. (1987).
Bevilacqua, M. P, Stengelin, S., Gimbrone, (1969). Science 243, 1160-1165. Carlo%
T. M., and Harlan,
J. M. (1990).
DUStin, fvt. L., and Springer, Elites, M. J., Csborn, ler, M. E., and Lobb,
Annu.
Med.
38,
M. A., Jr., and Seed,
Immunol.
T A. (1988).
Rev.
6.
Rev., in press.
J. Cell Biol. 707, 321-331.
L., Takada, Y., Grouse, C., Luhowskyj, R. R. (lQ90). Cell 60, 577-584.
S., Hem-
Geng, J.-G., Bevilacqua, M. P.. Moore, K. L., McIntyre, T. M., Prescott, S. M., Kim, J. M., Bliss, G. A., Zimmerman, G. A., and McEver, R. P (1990). Nature 343, 757-760. Harlan,
J. M. (1987). Acta Med. Stand.
775, 123-129.
Hattori, R., Hamilton, K. K., Fugate, R. D., McEver, P. J. (1989). J. Biol. Chem. 264, 7766-7771. Hemler,
M. E. (1988).
Holzmann, Hunkapillar, Hynes,
Immunol.
B., and Weissman, T., and Hood,
R. P, and Sims,
Today 9, 109-113. I. L. (1989).
L. (1989).
EMBO
Adv. Immunol.
J. 8, 1735-1741. 44, l-63.
R. 0. (1987). Cell 46, 549-554.
Johnston, 10331044.
G. I., Cook,
R. G.,
and
McEver,
R. P. (1989).
Cell
56,
Jutila, M. A., Lewinsohn, D. M., Berg, E. L., and Butcher, E. C. (1990). In Leukocyte Adhesion Molecules, T. A. Springer, D. C. Anderson, A. S. Rosenthal, and R. Rothlein, eds. (New York: Springer-Verlag), pp. 227-235. Larsen, E., Celi, A., Gilbert, G. E., Furie, B. C., Erban, J. K., Bonfanti, R., Wagner, D. D., and Furie, B. (1989). Cell 59, 305-312. Lawrence, M. B., Smith, C. W., Eskin, Blood 75, 227-237.
S. G., and Mclntire,
L. V. (1990).
Lo, S. K., Van Seventer, G. A., Levin, J. Immunol. 743, 3325-3329.
S. M., and Wright,
S. D. (1989).
Osborn, L., Hession, C., Tizard, Rosso, G., and Lobb, R. (1989). Rice, G. E., Munro, 777. 1369-1374.
R., Vassallo, C., Luhowskyj, Cell 59, 1203-1211.
J. M., and Bevilacqua,
Siegelman, M. H., Cheng, (1990): Cell 61, 611-622.
I. C., Weissman,
Simmons,
M. W., and
D., Makgoba,
Seed,
M. P (1990).
S., Chi-
J. Exp. Med.
I. L., and Wakeland, B. (1988).
E. K.
Nature
331,
T. A. (1989). Nature
339,
624-627. Staunton, 61-64.
D. E., Dustin,
M. L., and Springer,
Stoolman,
L. M. (1989).
Cell 56, 907-910.
Wegner, C. D., Gundel, R. H., Reilly, P, Haynes, Rothlein, R. (1990). Science 247, 458-459.
N., Letts,
L. G., and
