This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Bouvard, D. Articles by Fässler, R. Search for Related Content PubMed PubMed Citation Articles by Bouvard, D. Articles by Fässler, R. Related Collections Cell biology/structural biology Cell signalling/signal transduction Developmental biology Genetically altered mice

(Circulation Research. 2001;89:211.)
© 2001 American Heart Association, Inc.

Review

Functional Consequences of Integrin Gene Mutations in Mice

Daniel Bouvard^*, Cord Brakebusch^*, Erika Gustafsson, Attila Aszódi, Therese Bengtsson, Alejandro Berna, Reinhard Fässler

* From the Department of Experimental Pathology, Lund University, Lund, Sweden.

Correspondence to Daniel Bouvard or Cord Brakebusch, Lund University Hospital, Department of Experimental Pathology, 22185 Lund, Sweden. E-mail Daniel.Bouvard{at}pat.lu.se or Cord.Brakebusch@pat.lu.se

	Abstract

Top Abstract Introduction Early Mouse Development References

 
Abstract— Integrins are cell-surface receptors responsible for cell attachment to extracellular matrices and to other cells. The application of mouse genetics has significantly increased our understanding of integrin function in vivo. In this review, we summarize the phenotypes of mice carrying mutant integrin genes and compare them with phenotypes of mice lacking the integrin ligands.

Key Words: integrin • knockout • extracellular matrix

	Introduction

Top Abstract Introduction Early Mouse Development References

 
Cell-cell and cell-matrix interactions play essential roles in development, tissue function, and repair. During morphogenesis and tissue homeostasis, specific signals emerge from the environment, cross the cell membrane, and control many aspects of cell behavior, including migration, proliferation, survival, and differentiation. Integrins are cell adhesion molecules that can transmit such signals. They are heterodimeric transmembrane glycoproteins, composed of noncovalently-associated and ß subunits. To date, 24 integrin receptors assembled from various combinations of 18 and 8 ß chains have been described (Figure 1). The extracellular domain of integrins binds the cognate ligand, whereas the cytoplasmic domain is linked to the actin cytoskeleton. Integrins can bind extracellular matrix (ECM) proteins, such as collagens, fibronectins, and laminins, and cellular receptors, such as vascular cell adhesion molecule-1 (VCAM-1) and the intercellular cell adhesion molecule (ICAM) family.¹ Certain integrins share the same extracellular ligand, but on the other hand, several ligands are recognized by the same integrin. Ligand binding is followed by recruitment of several signaling proteins to the integrin and initiation of a signaling cascade that modulates cell behavior and gene transcription.^2,3

View larger version (21K): [in this window] [in a new window]   Figure 1. Vertebrate integrin family. ß Integrin subunits associate with subunits in a noncovalent fashion. So far, 24 heterodimers have been described that can bind extracellular matrix proteins, other cell surface receptors, and bacterial and viral proteins.

During the last two decades, most of the information about integrin function has been derived from in vitro cell culture systems. Gene targeting technology recently made it possible to generate mice that lack specific integrins in a constitutive or cell type–specific manner. Analyses of these mice demonstrate how integrin-mediated adhesion and signal transduction affect development and maintenance of tissues and provide additional insight into integrin function in various diseases. In this review, we summarize the phenotypes of mice carrying mutant integrin genes and compare them with the relevant genetic models bearing mutations in some integrin ligands (Table).

View this table: [in this window] [in a new window]   Table 1. Ablation of Integrin Genes in Mice and Their Resulting Phenotypes In Vivo

	Early Mouse Development

Top Abstract Introduction Early Mouse Development References

 
Ablation of integrin genes leads to various phenotypes during mouse development, ranging from apparently normal mice to early lethality (Figure 2). For example, disruption of the ubiquitously expressed ß₁ integrin gene leads to the loss of at least 12 different integrin receptors and results in peri-implantation lethality characterized by an inner cell mass (ICM) failure.^4,5 Fertilization of ß₁-null oocytes and the entire preimplantation development is normal. This is most likely because of the presence of maternal ß₁ integrin mRNA and protein.⁶ A possible explanation of the ICM failure could be the loss of ß₁-mediated survival signals. At peri-implantation stage, the ICM cells facing the blastocyst cavity differentiate into primitive endodermal cells and lay down a basal membrane (BM). This leads to polarization of adjacent ICM cells and their differentiation into ectodermal cells. The inability of the ectodermal cells to interact with BM components through integrins could lead to their loss by apoptosis and arrest of additional development.⁷ Another possibility could be that the lack of ß₁ integrins leads to abnormal BM assembly. ß₁ integrin is crucial for normal expression and correct assembly of BM components into a supramolecular structure.^8–12 The complete absence of a BM is observed in mice carrying a null mutation in the laminin ₁ chain gene.¹³ Laminins are heterotrimeric glycoproteins composed of , ß, and subunits. Laminin ₁-null mice fail to develop a BM between primitive endoderm and ICM, and, like the ß₁-null mice, they die at the peri-implantation stage.¹³ A recent study demonstrated the role of the BM in regulating apoptosis and cavity formation using embryoid bodies derived from laminin ₁-null embryonic stem (ES) cells.¹⁴ In laminin ₁-null embryoid bodies, the ectoderm never polarizes and the inner cells never apoptose to form cavities. An identical phenotype is observed in ß₁-null embryoid bodies (R. Fässler, unpublished data, 1995).

View larger version (22K): [in this window] [in a new window]   Figure 2. Integrin function during mouse development. Diagram showing the embryonic and perinatal lethal phenotype of single and double integrin–null mutant mice. The sizes of the embryos from fertilization (E0) to birth (P0) are not drawn to scale. Fertilization occurs in all integrin mutants, possibly because of the presence of maternal mRNAs. The earliest lethal phenotype has been observed in ß1-deficient mice. Studies of double 5/v-null and 3/6-null mice revealed a redundant and/or compensatory function for 5/v during early mouse development and for 3/6 in several organs and tissues, ie, brain, lung, and AER.

Deletion of the integrin ₅ gene leads to an embryonic lethality around embryonic day 9.5 (E9.5) and E10 with variable degrees of embryonic and extra-embryonic defects that partly depend on the genetic background.^15,16 ₅ß₁ plays a role in neural crest survival and maintenance of mesoderm derivatives after the gastrulation stage but is dispensable for initial commitment of mesodermic cells.¹⁷ Knockout of the ₅ß₁ ligand fibronectin (FN) results in a more severe phenotype, suggesting that other FN-binding integrins compensate in the ₅-null mouse. Fibronectin-null embryos implant and initiate gastrulation normally. Shortly thereafter, they show defects in neural tube closure, in the cardiovascular system, and in somite formation and die at E8.5.¹⁸

₄-Null embryos fail to fuse the allantois with the chorion during placentation, show defects in the heart and vasculature, and display abnormalities in the cranial and facial structure.¹⁹ ₄-Null embryos die at two different stages during development. The first wave is around E9.5 to E11.5 and is caused by the failure of allantois-chorion fusion, and the second is between E11.5 and E14 and is caused by massive heart hemorrhage. ₄ß₁ Integrin can bind fibronectin and VCAM-1. The mutant phenotype is mainly attributable to a defect in VCAM-1 binding, because mice lacking VCAM-1 also fail to fuse the allantois with the chorion and have cardiac defect similar to ₄-null mice.^20,21

_v Integrins can associate with ß₁, ß₃, ß₅, ß₆, and ß₈ subunits. They bind to both fibronectin and vitronectin, except for _vß₆ integrin, which binds to fibronectin, vitronectin, and tenascin-C, and _vß₈, which binds only to fibronectin.¹ Several reports have shown that _vß₃ and _vß₅ integrins are crucial for tumor angiogenesis.²² Therefore, it came by surprise that _v-null mice, which lack both of these receptors, have no defects in vasculogenesis and angiogenesis. Approximately 80% of the _v-null mice die between E10 and E12 because of a placenta defect characterized by a poorly developed labyrinthine zone and reduced interdigitation of fetal and maternal vessels. The remaining 20% die at birth from massive hemorrhages and cleft palates.²³ Because vitronectin, tenascin-C, and osteopontin-null mice have no obvious phenotype during the early mouse development,²⁴ the _v-null phenotype is most likely attributable to a lack of fibronectin binding.

A recent study investigated ₅^-/-_v^-/- double knockouts, which lack several major fibronectin receptors. The mutants die between E7.5 and E8 and display a severe gastrulation defect, with a lack of anterior mesoderm.¹⁶ It is noteworthy that this phenotype is even more severe than that observed in the fibronectin-null mutants, suggesting the presence of additional ligands for these two receptors at this developmental stage.

Hematopoiesis
Hematopoietic stem cells (HSCs) and endothelial cells are derived from a common precursor cell called hemangioblast. HSCs are generated outside the embryo proper, in the yolk sac (YS), and within the embryo in the para-aortic splanchnopleura/aorta-gonad mesonephros region. At around E8.5, when circulation starts in mouse, the HSCs are present in the fetal blood, and at E10 they start to colonize the fetal liver. Later, hematopoietic progenitors seed the thymus, spleen, other lymphoid organs, and bone marrow. In the adult mouse, HSCs reside in the bone marrow, but differentiated leukocytes circulate through the body and extravasate into infected tissues. Because of the early embryonic lethality of integrin ß₁-null mice, chimeric mice have been used to analyze ß₁ function in the hematopoietic system. These mice revealed that ß₁ integrin is not required for the generation of progenitor cells in the para-aortic splanchnopleura/aorta-gonad mesonephros region and the YS but is essential for them to colonize the fetal liver.^25,26 Differentiation of ß₁-null precursor cells in the presence of cytokines in vitro and in reaggregated fetal organ cultures of liver and thymus is normal, indicating that ß₁ integrins are crucial for the migration of HSCs into, but not for their development within, hematopoietic organs. Also, the homing of adult HSCs to the bone marrow is critically dependent on ß₁ integrin expression. Accumulation of transferred ß₁-null precursor cells in the blood in vivo and reduced adhesion of other cells to endothelioma cells in vitro suggests that ß₁ integrin is essential for the adhesion of HSC to the vessel wall.²⁶ Whether it plays an additional role at later processes (transmigration though the endothelium or extravascular migration) is not clear. Mouse HSCs express ₄ß₁, ₅ß₁, and ₆ß₁ integrin. Because individual knockouts of these subunits do not interfere with HSC homing,^27–29 either a combination of these receptors or another integrin of ß₁ family mediates HSC migration to the bone marrow.

Deletion of the ß₇ integrin subunit, which can associate with ₄ and _E, results in defective migration of T cells to Peyer’s patches and a reduced number of lamina propria and intraepithelial lymphocytes.³⁰ Migration of lymphocytes to the Peyer’s patches is mediated by ₄ß₇ binding to the endothelial MadCAM-1 receptors on the high endothelial venules in the gut. This was confirmed with mice lacking MadCAM-1 because of ablation of the transcription factor NKX2.3 gene, which controls MadCAM-1 expression. These mice have very small Peyer’s patches only visible after microscopic inspection.³¹ Targeted disruption of the _E gene leads to a reduction of T cells in lamina propria and gut epithelium but not in Peyer’s patches.³² Mice lacking both ß₇ integrin and L-selectin suffer from a nearly complete impairment of lymphocyte migration to mesenterial lymph nodes.³³

Because of the early embryonic lethality of ₄-null mice, the function of ₄ integrin in the hematopoietic system was studied in ₄-null chimeric mice. The ₄ subunit can associate with ß₁ and ß₇ integrin and is important for the differentiation of erythroid, myeloid, and B cell progenitors during embryogenesis.³⁴ Because in vitro differentiation in colony assays is normal, the defects observed in vivo are quite likely attributable to impaired interaction of the ₄-null progenitors with the microenvironment in the hematopoietic organs. Furthermore, no ₄-null erythrocytes, almost no B lymphoid cells, and only a few myeloid cells are present in adult chimeric mice, indicating an important role of ₄ integrins in the postnatal hematopoietic development.³⁴ The underlying mechanism proposed for this severe phenotype is an important role for the ₄ integrin for transmigration of progenitors through the stroma. In addition, proliferation of ₄-null precursor cells might be decreased, because ₄-deficient preB cells that did transmigrate in vitro showed a poor mitotic activity.³⁴ Similar to ß₇-deficient mice, the contribution of ₄-null T cells to Peyer’s patches is impaired.²⁷ ₄-Null chimeras that are older than 1 month after birth lack ₄-deficient thymocytes in the thymus, although T cell precursors are present in the bone marrow.²⁷ The few peripheral ₄-null T cells present show a reduced migration into the inflamed peritoneum.²⁹ Because mice with a disrupted VCAM-1 gene have normal hematopoiesis, the ₄-null phenotype is most likely attributable to impaired interactions of ₄ß₁ with fibronectin.³⁵ Mice with a conditional inactivation of the VCAM-1 gene show reduced number of mature B cells in the BM attributable to a defective migration of lymphocyte to the BM.^36,37 In addition, T cell–dependent immune response is impaired.³⁶

Integrins play a central role in leukocyte adhesion and migration. Leukocytes express several integrins in a low-affinity state, ensuring that they do not make inappropriate interactions with ligands normally present in blood or on endothelial cells. To leave the blood, leukocytes attach loosely to activated endothelium via selectins and begin to roll. Chemokines released by the endothelium or the surrounding tissue shift the leukocyte integrin into a high-affinity state that allows binding to endothelial cell adhesion molecules such as ICAMs and VCAM-1, resulting in firm adhesion of the blood cells to the vessel wall. Finally, the leukocytes cross the endothelial cell layer and underlying BM and adhere and migrate within the extravascular tissue.

ß₂ integrin is specifically expressed on leukocytes and associates with _L (lymphocyte function–associated antigen-1 [LFA-1]), _M (Mac-1), _X, and _D subunits. Important counter-receptors of ß₂ integrins are members of the ICAM family.¹ However, Mac-1, _Xß₂, and _Dß₂ also bind to other proteins: Mac-1 to inactivated complement factor (iC3b), fibrinogen, factor X, _Xß₂ to iC3b and type I collagen,³⁸ and _Dß₂ to VCAM-1.³⁹ In addition, Mac-1 and _Xß₂ can interact specifically with various polysaccharides, such as ß-glucan, mannose, N-acetyl-D-glucosamine, and glucose, which mediate attachment to bacteria.^40–42

Humans with a mutation in the ß₂ gene develop leukocyte adhesion deficiency type I. These patients have a massive leukocytosis, recurrent bacterial infections, impaired wound healing, and severe gingivitis. Almost no neutrophils extravasate into infected tissues. Mice lacking ß₂ integrin display a very similar phenotype.^43,44 The degree of neutrophil extravasation, however, is variable and depends on the model of inflammation.^43–45 Some neutrophil extravasation processes like the croton oil-induced migration into skin are completely abolished, whereas others, like the sterile peritonitis, still occur. These migrations are perhaps mediated by ß₁ integrins. In chronic inflammations, ₄ß₁ integrin seems to play an important role.⁴⁶ Another candidate is ₉ß₁, which is highly expressed on human neutrophils and involved in transendothelial migration in vitro.⁴⁷ ß₂ integrin also plays an important role in T cell proliferation⁴⁴ and contributes to the accumulation of dendritic cells in the lung.⁴⁸

LFA-1 and, to a lesser degree, Mac-1 contribute to neutrophil adhesion to endothelium. In a tumor necrosis factor (TNF)-induced air pouch inflammation model, neutrophil migration was equally reduced in ß₂-deficient and LFA-1–deficient mice, whereas mice lacking Mac-1 showed a marked increase of neutrophil extravasation.⁴⁹ In the absence of LFA-1, immune responses against systemic viral infections are normal, but alloantigen triggered T cell proliferation and cytotoxicity are severely impaired, leading to defective host-versus-graft reaction and abrogated tumor rejection.^50–52 LFA-1 is also involved in the homing of lymphocytes to peripheral lymph nodes and, to a lesser degree, to mesenteric lymph nodes and Peyer’s patches.⁵³ Mac-1–null neutrophils have an impaired phagocytosis and degranulation, leading to a delayed neutrophil apoptosis.^54,55 Furthermore, mast cells were reduced in some but not all locations of Mac-1–deficient mice.⁵⁶ In an acute glomerulonephritis model, Mac-1–null neutrophils extravasated normally but did not stay in the tissue, apparently because of an absent cytoskeletal rearrangement mediated by Mac-1.⁵⁷ Unexpectedly, mice lacking Mac-1 or the Mac-1 ligand ICAM-1 are obese, suggesting a role of Mac-1 in the regulation of fat metabolism.⁵⁸ No knockouts have been reported so far for _X and _D integrin.

Hemostasis
Platelets express several integrins on their surface, including ₂ß₁, ₅ß₁, ₆ß₁, _vß₃, and _IIbß₃, which play different roles during blood coagulation. _IIbß₃ is crucially important for the aggregation and adhesive spreading of platelets during hemostasis.⁵⁹ On resting platelets, _IIbß₃ is present in a low-affinity form. Inside-out signaling triggered by various agonists, including thrombin, ADP, and collagen induces a shift to a high affinity conformation. Activated _IIbß₃ then binds to fibrinogen, which crosslinks platelets, leading to thrombus formation and finally sealing of wounds. In addition, outside-in signaling induced by ligand binding to _IIbß₃ results in calcium mobilization, tyrosine phosphorylation of several proteins, including ß₃ integrin itself, and additional cytoskeletal rearrangement, which is especially important for stable clot formation.⁵⁹ This was confirmed in mice with a selective loss of outside-in signaling caused by replacement of the two cytoplasmic tyrosines of the ß₃ subunit by phenylalanine residues. These mice are impaired in the late phase of platelet aggregation and in clot retraction in vitro and show an increased rebleeding tendency in vivo.⁶⁰ Patients having no or only a mutated _IIbß₃ integrin on platelets suffer from Glanzmann thrombasthenia characterized by impaired platelet aggregation and prolonged bleeding time. The same phenotype was replicated in mice lacking the ß₃ or the _IIb integrin gene.^61,62

When platelets get in contact with injured or diseased blood vessel walls, they tether to the subendothelial matrix through interactions with von Willebrand factor. Integrin ₂ß₁ was thought to be essential for the subsequent platelet adhesion to collagens, facilitating interactions with the activating platelet collagen receptor glycoprotein VI (GPVI). This hypothesis was based on observations from patients with reduced expression of ₂ß₁ integrin on platelets who suffered from excessive posttraumatic bleeding and menorrhagia.^63,64 In addition, their platelets display profound defects in collagen-induced adhesion/aggregation.⁶⁵ However, it is presently unclear whether these patients also carried other defects contributing to the pathological condition. A few ₂ß₁ integrin function–blocking antibodies have been shown to interfere with collagen-induced platelet aggregation.^66–68 Surprisingly, however, mice lacking ß₁ integrin on megakaryocytes show no significant changes in platelet counts or bleeding time.⁶⁹ Aggregation of ß₁-null platelets on collagen is delayed but not reduced. On the other hand, platelets lacking GPVI cannot activate integrins and consequently fail to adhere to and aggregate on collagen.⁶⁹ Therefore, GPVI plays the central role in platelet-collagen interactions by activating different adhesive receptors, including ₂ß₁ integrin. ₂ß₁ Strengthens adhesion but is not essential for it.

Vasculogenesis and Angiogenesis
Studies of mice carrying integrin-null mutations support the concept that endothelial integrins are involved in the formation and maintenance of the vascular network. Endothelial cells express at least 11 different integrins, including ₁ß₁, ₂ß₁, ₃ß₁, ₄ß₁, ₅ß₁, ₆ß₁, _vß₁, _vß₃, _vß₅, _vß₈, and ₆ß₄.

The most severe vascular defect is present in mice lacking the ₅ subunit. They have impaired development of the extraembryonic (YS) and embryonic (heart and aorta) vasculature.¹⁵ Blood islands form, and hematopoiesis occurs in the YS. Subsequently, the mesoderm and endoderm layers are separated, and the primitive blood cells lie in sacs between the two germ layers and leak into the exocoelomic space. This splitting of the germ layers could be explained by defective cell-matrix interaction or mesoderm defects. All ₅-null embryos form heart and blood vessels that contain a few primitive blood cells. The vessels, however, are distended and leak. A very similar but more severe vascular phenotype is present in mice lacking fibronectin,^18,70 the major ligand of ₅ß₁. The fibronectin mutants display variable cardiovascular defects depending on the genetic background.⁷⁰ On the C57BL/6 background, the defects are less severe, whereas on the 129/Sv background, no hearts have properly formed vessels. In C57BL/6 mutants that form hearts, the myocardial and endocardial cells are disorganized, resulting in a bulbous heart tube. The aorta contains endothelial cells that lack contact with the surrounding mesenchyme, whereas the yolk sac is completely devoid of vessels. On a 129/Sv background, endothelial cells differentiate but fail to assemble into vessels. These findings clearly suggest that ₅ß₁ is not required for initial vasculogenesis but is necessary for the proper formation and maintenance of blood vessels. Fibronectin, on the other hand, seems necessary for vasculogenesis, suggesting that in the absence of ₅, other fibronectin receptors can compensate. Similar to the ₅- and fibronectin-null mice, mice lacking vascular endothelial growth factor (VEGF)-R1 (flt-1) have normal hematopoietic progenitors but show impaired blood vessel formation.⁷¹ Mice lacking the tie-2 receptor⁷² or its ligand angiopoetin-1⁷³ show defects in the interactions of endothelial cells with the surrounding matrix and mesenchyme. These abnormalities are reminiscent of those observed in ₅- and fibronectin-deficient mice, suggesting that these signaling systems cooperate in blood vessel assembly.

Null mutations of other subunits associated with ß₁ show milder or no obvious vascular defects. Mice lacking integrin ₄ die during embryonic development because of defects in heart and placenta.¹⁹ In addition, these mutants lack coronary vessels.¹⁹ Because epicardial coronary vessels connect with those in the myocardium, blood from the myocardial vessels leaks, resulting in cardiac hemorrhage in the ₄-deficient mice. It is not clear, however, if this is a direct consequence of the loss of ₄ or an indirect secondary defect attributable to a loss of the epicardium. A vascular abnormality is also present in ₃-null mice. Their glomerular capillaries show reduced branching and a distended lumen.⁷⁴ Glomerular endothelial cells express ₃ß₁, and, therefore, the defect could result from an abnormal formation of capillary structure of the glomerulus. It is also possible that the altered capillaries are caused by the failure of podocytes to provide adequate scaffolding, within which the newly formed capillaries would form. Mice lacking the ₁ subunit show normal development of the blood vessel system⁷⁵ but have a reduced tumor angiogenesis.⁷⁶ This reduction in tumor capillary numbers is caused by overexpression of metalloproteinase 7 (MMP-7) and MMP-9 in the mutants leading to elevation of angiostatin that has been shown to inhibit endothelial cell proliferation in tumors but not during embryogenesis. Isolated ₁-null endothelial cells from adult lung, however, show a reduced proliferation on both ₁-dependent and -independent substrata, indicating that this response is not restricted to tumor endothelial cells.⁷⁶ The formation of a normal embryonic vasculature suggests that there are phenotypic differences between embryonic versus adult and tumor endothelial cells. These data demonstrate that ₁ integrins regulate MMP expression and that a tightly controlled MMP expression is crucial for angiogenesis. Although MMP synthesis is essential for new blood vessel formation by degrading the ECM and thus facilitating remodelling and invasion, an increased MMP expression can also lead to the release of factors that inhibits cell growth, such as angiostatin.

Endothelial cells express both ₆ß₁ and ₆ß₄ integrins. Mice lacking ₆ or ß₄ integrin show no obvious vascular defects.^77,78 Other members of the integrin family or -dystroglycan may compensate for the loss of ₆ function. Mice lacking -dystroglycan die before the first vessels form,⁷⁹ and, therefore, only tissue-specific gene inactivation of the -dystroglycan gene will reveal whether it has a function during vessel formation. Targeted null mutation of the ß₁ integrin gene leads to embryonic lethality before the vascular system begins to form. With the use of ß₁-null ES cells, however, it is possible to demonstrate an important role for the ß₁ subfamily during blood vessel formation. Chimeric mice derived from ß₁-null ES cells lack ß₁-null endothelial cells in liver,⁴ indicating that endothelial cells either fail to form or fail to participate in blood vessel formation. In vitro differentiation of ß₁-null ES cells into embryoid bodies show that endothelial cells differentiate but initiation of vasculogenesis is significantly delayed.⁸⁰ In addition, the responsiveness to VEGF is impaired,⁸⁰ suggesting that VEGF action is upstream of ß₁ integrin function. Finally, in ß₁-null teratomas, the ß₁ mutant ES cells fail to participate in tumor vessel formation, which is in sharp contrast to wild-type ES cells.⁸⁰

Endothelial cells express at least 4 members of the _v-integrin subfamily, _vß₁, _vß₃, _vß₅, and _vß₈. Numerous studies using blocking antibodies suggested that this family of integrins is of crucial importance for angiogenesis, as shown in chorioallantoic membrane assays,⁸¹ tumor models,⁸² and neovascularization studies of the mouse retina.^83,84 The generation and analyses of _v-null mice, however, revealed that 20% of the mutants are born alive with no defects in vasculogenesis and angiogenesis.²³ Basic endothelial processes of proliferation, migration, tube formation, branching, and basement membrane assembly occur in the mutants. The intracerebral and intestinal vasculature, however, becomes distended and eventually ruptures, resulting in severe hemorrhage. Interestingly, ß₈-null mutant mice share many similarities to those of the _v mutants, with early defects in placental development and later deficits in formation and stabilization of the vascularization of the central nervous system (Z. Jiangwen and L. Reichardt, personal communication, 2001). Mice lacking ß₃ or ß₅, two of the _v-associated ß subunits, are viable and show no obvious defects in vascular development.^61,85 The ß₃-null mutants were additionally investigated for defects in postnatal retinal angiogenesis, but no abnormalities were observed.⁶¹ In addition, null mutants of many _v ligands, including vitronectin,⁸⁶ tenascin-C,^87,88 osteopontin,⁸⁹ fibrinogen,⁹⁰ perlecan,⁹¹ and von Willebrand factor,⁹² show no defects in vessel formation.

Skin
The skin is composed of an epidermal and a dermal layer, separated by a basement membrane. The epidermis is made of keratinocytes at different stages of differentiation. The proliferating basal keratinocytes adhere to the basement membrane via specialized adhesion junctions called hemidesmosomes, containing ₆ß₄ integrin, and via other ECM receptors, including ₃ß₁ integrin and -dystroglycan. Both ₆ß₄ and ₃ß₁ integrins bind laminin 5 (₃ß₃₁) of the BM. During differentiation, the keratinocytes lose their integrin expression, detach from the basement membrane, move to the suprabasal layers, and terminally differentiate to form the stratum corneum. Mutations in the ₆ or the ß₄ gene result in a reduction of hemidesmosomes, causing the epidermolysis bullosa in humans, a severe blistering disease leading in most cases to early postnatal lethality.⁹³ Deletion of the ₆ or the ß₄ gene in mice leads to a complete absence of hemidesmosomes, a severe detachment of the epidermis and death shortly after birth.^77,78,94 Interestingly, the basal lamina is not affected by the loss of ₆ß₄ integrin.

₃-Null mice die shortly after birth because of defects in kidney and lung organogenesis.⁷⁴ These mice also display mild blisters caused by disruption of the basement membrane organization.⁹⁵ Keratinocyte-restricted disruption of the ß₁ integrin gene during early development causes a similar but more severe phenotype with death shortly after birth, suggesting that ₂ß₁ or ₅ß₁, which both are expressed in skin, play an additional role in basement membrane formation and skin homeostasis.⁹⁶ Skin-specific ablation of the ß₁ integrin gene at around birth demonstrated a role of ß₁ integrin in the processing of the BM components and in the growth and maintenance of hair follicles. ß₁-deficient basal keratinocytes have an aberrant morphology and a reduced proliferation rate, but are still able to terminally differentiate.¹⁰ Interestingly, in the absence of both ₃ and ₆ or ß₄ integrin, basal keratinocytes can initially attach to the basal lamina and proliferate, indicating additional adhesion mechanisms.⁹⁷ A candidate receptor for maintaining proliferation could be -dystroglycan.

The critical importance of laminin-5 as a ligand for ₆ß₄ integrin is demonstrated by patients with mutations in the laminin subchains ₃, ß₃, or ₂, which develop lethal junctional epidermolysis bullosa.⁹³ Similarly, mice lacking a functional laminin ₃ gene have severe blistering and abnormal hemidesmosomes but an ultrastructurally normal basal lamina.⁹⁸ Adhesion experiments with laminin-5–deficient basement membrane suggested that ₃ß₁ also interacts with other basement membrane components, in contrast to ₆ß₄ integrin.

The integrin _vß₆, which interacts with fibronectin, vitronectin, and tenascin-C, is only expressed on epithelial cells during local injury or inflammation. Mice deficient for ß₆ integrin show normal embryonic development and wound healing.⁹⁹ Surprisingly, however, mutant mice exhibit juvenile baldness attributable to macrophage infiltration of the dermis of the affected areas. Furthermore, ß₆-null mice demonstrate an increased airway responsiveness, a hallmark of asthma, and an infiltration of activated B and T cells into the lung. This phenotype was explained by the finding that _vß₆ can bind to a latent form of the transforming growth factor-ß₁ (TGF-ß₁), inducing a conformational change in the ligand, which allows its interaction with TGF-ß receptors. Therefore, loss of _vß₆ reduces the amount of active TGF-ß₁ in epithelia, which at least partially mimics a local TGF-ß₁–null phenotype.¹⁰⁰ Expression of _vß₆ in injury and inflammation, therefore, leads to increased local activation of TGF-ß₁, which, among other effects, modulates inflammatory cell function.

Skeletal Muscle
The ß₁ subunit is expressed throughout skeletal muscle development. Initially, muscle precursor cells and fetal muscle express the cytoplasmic ß_1A splice variant. After birth, the ß_1A splice variant becomes substituted by the ß_1D variant.¹⁰¹ During early muscle development, ß_1A is expressed with ₁, ₃, ₄, ₅, ₆, ₇, ₉, and _v. With the exception of the ₇ integrin, the other subunits are downregulated around birth. In vitro studies using Drosophila cells revealed an important role for integrins in generating the sarcomeric cytoarchitecture. Normal Drosophila embryo cells fuse and develop sarcomeres in which integrins and actins concentrate at Z-bands. In contrast, integrin-deficient fly cells fail to form sarcomeres.¹⁰² In mice, antibody perturbation experiments suggested an important role of ß₁ integrin for migration of muscle precursor cells from the dermomyotome to the limbs, fusion of myoblasts into primary and secondary myotubes, and assembly of the contractile proteins/apparatus.^103–105 Interaction between ₄ß₁ on the primary myotubes and VCAM-1 on the secondary myoblasts was suggested to be essential for the formation of the secondary myotubes.¹⁰⁵ Such a function for ₄ integrin, however, was neither confirmed in ₄-null chimeric mice nor in ₄-null embryoid bodies.³⁴ In addition, no muscle phenotype was found in the few VCAM-1 null mice that survive the embryonic defects.^20,21

₅ß₁ Integrin is expressed in adhesion plaque–like structures along the myotube during development¹⁰⁶ but is downregulated in adult muscle. ₅-Null mice die during early development before muscle is formed. ₅-Null chimeras with a high contribution of ₅-deficient cells to skeletal muscles develop typical alterations resembling muscular dystrophy, including giant muscle fibers, vacuoles, and centrally located nuclei.²⁸ The cause of the muscle degeneration is reduced adhesion and survival of ₅-null myoblasts.²⁸

The most abundant integrin in skeletal muscle is ₇ß₁, which is expressed during all stages of muscle development^107,108 and interacts with laminin 2 (₂ß₁₁) and laminin 4 (₂ß₂₁) in skeletal muscle.^109–111 ₇ Exists in two extracellular (X1 and X2) and three cytoplasmic (A, B, and C) splice variants, which display a tissue-specific and developmentally regulated expression pattern.^112–115 Skeletal muscles express _7A, _7B, _7C, and _7X2, whereas _7B, _7X1, and _7X2 are found in the heart. After birth, _7A and _7B concentrate at neuromuscular and myotendinous junctions, whereas _7C is the predominant extrajunctional splice variant.^107,113 Mice with a targeted deletion of the ₇ integrin develop a progressive muscular dystrophy after birth. The major defect is severe disruption of the myotendinous junctions.¹¹⁶ Mutations in the ₇ gene have also been identified in patients suffering from congenital myopathies characterized by delayed motor milestones.¹¹⁷ The genetic findings from mouse and human suggest no role for ₇ß₁ in myogenesis but a maintenance function on the mature muscle by firmly linking the muscle fibers to the extracellular matrix. Muscle expresses laminin 4 that binds the ₇ß₁ integrin. Naturally occurring mutations in the laminin ₂ gene are found in human and mouse (dy/dy mouse strain) and lead to a congenital muscular dystrophy¹¹⁸ resembling the phenotype produced by loss of ₇ integrin expression.

These data demonstrate an important function of ₅ß_1A and ₇ß_1D integrins in the maintenance of muscles and show that they cannot compensate each other. However, in ß₁-null chimeric mice, ß₁-deficient myoblasts migrate, form myotubes, and develop a normal sarcomeric cytoarchitecture in ß₁-null chimeric mice.¹¹⁹ The formation of myotubes in ß₁-null embryoid bodies is delayed but otherwise normal.^119,120 The absence of any obvious muscular dystrophy in the ß₁-null chimeric mice could be explained by both a low contribution of ß₁-null muscle cells and the presence of wild-type cells, which can functionally compensate. Replacement of the muscle-specific splice variant ß_1D by ß_1A had no effect on the formation of skeletal muscles.¹²¹ No muscle phenotype has been reported from knockouts of ₁, ₃, ₄, ₆, ₉, and _v integrins, respectively.

Heart
Cardiomyocytes express several integrin subunits during development: ß_1A, ß_1D, ₁, ₃, ₅, ₆, ₇, and _v. Like in the skeletal muscle, after birth, adult cardiac muscle expressed the ß_1D variant. Expression of ₁, ₅, and ₆ is downregulated at birth. On the basis of in vitro analysis, integrin-mediated attachment to the ECM has been suggested to be important for controlling growth and differentiation of cardiomyocyte.¹²² Furthermore, integrins were proposed to function as mechanoreceptors that transform mechanical stimuli into biochemical signals that affect cellular function.¹²² Several genetic mouse models demonstrate an important function of ß₁ integrin in cardiac muscle in vivo. Mice expressing ß_1A instead of ß_1D in heart develop, without exposure to mechanical stress, mild cardiac defects characterized by an increase in atrial natriuretic peptide (ANP) synthesis. ANP is a vasorelaxant diminishing volume overload and hypertension. Increased levels of ANP have been correlated to, among other things, reduced ventricular function and induction of ventricular hypertrophy. The increased expression of ANP suggests that the ß_1A subunit, at least in the heart, is not fully compensating for ß_1D.¹²¹ Even more severe defects occur when both ß_1A and ß_1D are absent or functionally inactivated. The areas with ß₁-null cardiac muscle cells in the heart of ß₁-null chimeras become smaller with time and show signs of degeneration. Ultrastructural analysis revealed alterations in the sarcomeric architecture. In addition, transgenic mice expressing a dominant-negative form of ß₁ integrin, in which the extracellular and transmembrane domain of CD4 is fused to cytoplasmic domain of ß₁ integrin, show hypertrophic changes in the heart. Mice that express high levels of the transgene die around birth and display a replacement fibrosis.¹²³ Overexpression of a constitutively active form of ₅ integrin in the heart results in electrocardiographic abnormalities, cardiomyopathy, and death within one month after birth.¹²⁴

Skeleton
The vertebrate skeleton develops via two distinct ways: by replacing cartilaginous precursors (endochondral ossification) or by direct development within a mesenchymal tissue (intramembranous ossification). The major cell types of the skeleton, the chondrocytes, the bone depositing osteoblasts, and the bone resorbing osteoclasts, express various integrin receptors at their surfaces. Despite numerous in vitro studies, their function in vivo during skeletogenesis is not well understood. Mice deficient in ß₃ integrin produce dysfunctional osteoclasts, leading to osteosclerosis characterized by increased bone mass.¹²⁵ The mutant osteoclasts fail to form the proper ruffled membrane important for their resorptive function, leading to hypocalcemia in vivo and decreased resorption of whale dentin in vitro. Furthermore, ß₃-deficient osteoclasts do not spread in culture and lack the characteristic actin rings, suggesting that _vß₃-mediated signals are important for the organization of cytoskeleton in these cells.

The most severe skeletal abnormalities have been identified in ₃^-/-₆^-/- double-knockout mice.¹²⁶ They display exencephaly attributable to the failure of neural tube closure, kinked tail, and limb anomalies characterized by shortening, abnormal shape, lack of digit separation, and fusion of phalanges between digits 2 and 3. Detailed analyses revealed discontinuity of the surface ectoderm of the distal limb and defective formation of a specialized limb bud epithelial structure, called apical ectodermal ridge (AER). AER is located at the tip of the bud and is involved in the control of limb outgrowth and differentiation. Both ₃ and ₆ integrin chains are expressed in this region, and a lack of both integrins results in structural disorganization and reduced proliferation of the ridge cells. Because single ₃ or ₆ mutants do not show such defects, the observations suggest that the presence of both integrin chains is crucial for the organization and proper function of AER and reveal the synergistic actions of these integrins in limb morphogenesis. Similar abnormalities of the distal limbs are present in the laminin ₅ knockout mice.¹²⁷ Altogether, these observations indicate that ₃ and ₆ integrins are the major cell-surface receptors for laminins containing ₅ chains in the distal limb ectoderm.

Transgenic mice expressing a dominant-negative ß₁ integrin subunit under the control of the osteoblast-specific osteocalcin promoter show impaired membranous bone formation.¹²⁸ The phenotype is characterized by a decreased rate of cortical bone formation and reduced bone mass of cortical and flat bones in young animals, indicating the importance of ß₁ integrin in osteoblast function. The defect of flat bones was normalized in older males but not in females, suggesting sex-specific differences in adult animals.

Kidney
So far, two integrins, ₃ß₁ and ₈ß₁, have been shown to be crucial for kidney development. In ₃^-/- mutant mice, the glomerular basement membrane of the kidney is disorganized and glomerular podocytes are unable to form mature foot processes, suggesting that ₃ integrin might be important for basal membrane organization. In addition, these mutant mice have a lung defect with reduced branching of the conducting airway⁷⁴ and show an alteration of the basal membrane organization of the submandibular gland.¹²⁹

Inactivation of the ₈ integrin leads to a failure in kidney development, with different levels of severity ranging from a reduction of kidney size to a complete loss of kidneys.¹³⁰ The kidney defect is characterized by abnormal growth of the ureteric bud and abnormal branching and formation of the ureteric epithelium. Fibronectin, tenascin-C, and vitronectin bind to ₈ß₁. However, they seem not to be involved in the phenotype, because they are not detectably expressed during this stage of kidney development. Osteopontin can also bind ₈ß₁ and has been proposed to mediate this interaction. Osteopontin-null mice, however, are normal and develop no apparent phenotype in the kidney.^131,132

Most integrin ₈-null mice die soon after birth, but a few survive until adulthood. These mice are deaf and have balance defects attributable to a structural inner-ear defect, indicating a role of ₈ß₁ integrin and its ligand fibronectin in hair-cell differentiation and function.¹³³

Brain
Evidence coming from fly and worm suggest a critical role for the integrin family during brain development and maintaining brain function.¹³⁴ This was also confirmed in mice. ₃-Null mice display a defect in neuron migration and a disorganized layering of the cerebral cortex, suggesting that this integrin is involved in radial neuronal cell migration.¹³⁵ A very similar phenotype is observed in reeler mice, which lack the extracellular matrix protein reelin. In these mutant mice, migration of Cajal-Retzius cells is impaired, leading to an abnormal lamination of the cerebral cortex. A recent study showed that ₃ß₁ integrin can bind reelin.¹³⁶ This interaction may provide a stop signal and arrest neuronal migration.

₆-Null mice have abnormalities in the laminar organization of the developing cerebral cortex and retina, ectopic neuroblastic outgrowth on the brain surface and in the eye, and an abnormal laminin deposition.¹³⁷ In the ₃^-/-₆^-/- double-knockout mice, the cortical disorganization appears earlier and is more severe, likely because of an additive effect of the two mutations. Ablation of the laminin ₅ gene leads also to a brain phenotype similar to the ₃^-/-₆^-/- double knockouts. In addition, the ₃^-/-₆^-/- mutants have several other abnormalities, including exencephaly, syndactyly, and kidney defects.^126,127,138

Axotomy leads to an increased expression of ₇ß₁ integrin on axons and growth cones during peripheral nerve regeneration. An impaired axonal regeneration has been reported in ₇-null mutants, suggesting that this integrin plays a role in this process.¹³⁹

Signaling
Integrin-mediated adhesion results in the activation of various signaling cascades inside the cell, including activation of tyrosine kinases like focal adhesion kinase (FAK) and Src, serine/threonine kinases like mitogen activated protein kinase (MAPK), and small G proteins.^3,140 Direct evidence for signaling defects in integrin-deficient mice is scarce. In ₇-null mice, a constitutive activation of c-Raf-1 kinase was observed in muscle, suggesting that ₇ß₁ integrin may be involved in the negative control of this pathway.¹⁴¹ However, this finding could also be caused indirectly by the muscle degeneration observed in these mice. Analysis of the inner ear development in ₈-null mice revealed a defect in stereocilia formation correlating with a loss of FAK at the cell membrane.¹³³

Cell cultures studies revealed that integrin-mediated signals are involved in cell proliferation, cell survival, prevention of apoptosis (also called anoikis), and cell differentiation.¹⁴² Presently, it is not clear to what degree the integrin signaling observed in vitro also has a role in vivo. The observations collected so far suggest an involvement of integrins in regulating cell proliferation and programmed cell death in vivo.

Several integrin-null mice show reduced proliferation of certain cell populations. Mice lacking ₃ and ₆ integrin have a reduction of cell proliferation in the AER,¹²⁶ and mice lacking ß₁ integrin in keratinocytes lose proliferating hair matrix cells and display a reduced proliferation of basal keratinocytes.¹⁰ Deletion of the ₁ gene results in decreased proliferation of dermal fibroblasts.¹⁴³ In vitro culture of these fibroblasts demonstrated a failure to recruit and activate the adaptor protein Shc, which triggers the MAPK activation cascade leading to cell proliferation. In a dominant-negative approach, the cytoplasmic and transmembrane domain of ß₁ integrin was fused to the extracellular domain of CD4, which interferes with endogenous ß₁ integrin function in vitro. Expression of such a chimeric protein in the mammary gland of transgenic mice results in decreased proliferation, increased apoptosis, and impaired differentiation of the mammary epithelial cells.¹⁴⁴ Additional biochemical analysis of these mice revealed no change in FAK activation in the mammary gland but a significant decrease in the phosphorylation of Shc, Grb2, Akt, c-Jun N-terminal kinase, and extracellular signal–regulated kinase.¹⁴⁵ Lack of the ß₄ integrin cytoplasmic domain has been shown to reduce proliferation in the skin and intestine,¹⁴⁶ which agrees with in vitro data showing that ß₄ integrin is able to bind and activate Shc.

Cell survival defects have been reported in some integrin-null mice. For instance, ₅ integrin was shown to be important for the survival of neural crest cells in vivo,¹⁷ and ₃^-/-₆^-/- mice have syndactyly resulting from an impaired cell death of interdigital cells.¹²⁶ On the other hand, loss of ß₁ integrin–mediated adhesion of basal keratinocytes does not result in cell death, in contrast to expectations from cell culture experiments.¹⁰

Concerning differentiation processes, so far gene ablation approaches revealed only a minor role for integrins. Absence of ß₁ integrin does not interfere with differentiation of neurons and neural crest–derived cells,⁴ fetal hematopoietic cells,²⁶ or keratinocytes.¹⁰ Double knockouts for ₃ and ₆ integrin show normal keratinocyte differentiation, despite the absence of the major integrin receptors ₃ß₁ and ₆ß₄.¹²⁶ In vitro myoblast differentiation and fusion occurs independently of ß₁ integrin but is delayed.¹¹⁹ A similar delay was found in blood vessel formation in ß₁-null embryoid bodies.⁸⁰ Such a delay could be attributable to the absence of ß₁ integrin or the experimental design that does not reflect the in vivo conditions, eg, growth factors that are absent in fetal calf serum but not in vivo or absence of complex matrices in vitro. On the other hand, teratomas using ß₁-null cells never give rise to an endothelial cell population, suggesting that ß₁ integrin could be involved during their differentiation or survival.⁸⁰

Future
Analysis of integrin function by gene-targeted mice often confirmed but sometimes also contradicted previous results obtained by antibody or peptide inhibition studies. For instance, blocking _vß₃ function by inhibitory antibodies suggested an essential role of _vß₃ integrin in angiogenesis. However, both _v and ß₃ knockout mice show blood vessel formation. Similarly, a few ₂ blocking antibodies prevent collagen-induced platelet aggregation, whereas deletion of ß₁ integrin on platelets revealed only a minor and clearly dispensable role for ₂ß₁ during this process. One obvious disadvantage of antibodies is that they do not completely block protein function, whereas a targeted gene disruption leads to a complete loss of the protein. In addition, antibodies might have unexpected side effects by steric hindrance of other ligand-receptor interactions or only partial inhibition of the integrin signaling. However, constitutive gene knockout also has disadvantages. The mutated mouse might react to the loss of a protein by upregulating the activity or expression of other molecules, thus compensating the defect. Because totipotent embryonic cells, at least theoretically, have a high capacity for compensation, this risk is real in constitutive knockouts, and compensation has already been demonstrated on several occasions. A way to reduce the chance of compensation is to induce the knockout in more differentiated cells, which are less able to change their expression pattern. This is possible using the cre/loxP system generating mice with tissue restricted or inducible gene deletions.^10,26,69,96 The generation of mice carrying floxed integrin genes to induce cell type–specific deletions, but also the introduction of subtle mutations into the extracellular and intracellular domain of integrin subunits and the interbreeding of integrin-null mice to create double and triple mutant animals, will additionally advance our understanding of how integrins function in vivo. In addition, the mutant mice will allow the establishment of cell lines that can be used to study in vivo observations at a molecular level.

More information about the role of integrins in signaling in vivo will come from the analysis of mutant tissues with antibodies specific for activated forms of signaling molecules, from knockin mice carrying subtle mutations of the integrin molecules interfering specifically with certain in vitro signaling functions, from knockouts of integrin-associated proteins, and from the rescue of integrin-null mice with transgenes encoding activated forms of signal transduction molecules suspected to act downstream of the respective integrins.

	Acknowledgments

 
This work was supported by the Craaford Foundation (E.G.) and the Swedish National Research Foundation (R.F. and C.B.). D.B. is a Wenner-Gren fellow. We are grateful to Drs Z. Jiangwen and L. Reichardt for sharing information before publication. We thank Drs T. Sakai, Kamin Johnson, and Michael Dictor for critical reading the manuscript.

	Footnotes

 
*Both authors contributed equally to this work.

Received March 6, 2001; accepted June 20, 2001.

	References

Top Abstract Introduction Early Mouse Development References

 

1. Plow EF, Haas TA, Zhang L, Loftus J, Smith JW. Ligand binding to integrins. J Biol Chem. . 2000; 275: 21785–21788.[Free Full Text]
2. Hynes RO. Integrins: versatility, modulation, and signaling in cell adhesion. Cell. . 1992; 69: 11–25.[Medline] [Order article via Infotrieve]
3. Giancotti FG. Complexity and specificity of integrin signalling. Nat Cell Biol. . 2000; 2: E13–E14.[Medline] [Order article via Infotrieve]
4. Fässler R, Meyer M. Consequences of lack of ß₁ integrin gene expression in mice. Genes Dev. . 1995; 9: 1896–1908.[Abstract/Free Full Text]
5. Stephens LE, Sutherland AE, Klimanskaya IV, Andrieux A, Meneses J, Pedersen RA, Damsky CH. Deletion of ß₁ integrins in mice results in inner cell mass failure and peri-implantation lethality. Genes Dev. . 1995; 9: 1883–1895.[Abstract/Free Full Text]
6. Brakebusch C, Hirsch E, Potocnik A, Fässler R. Genetic analysis of ß₁ integrin function: confirmed, new and revised roles for a crucial family of cell adhesion molecules. J Cell Sci. . 1997; 110: 2895–2904.[Abstract]
7. Coucouvanis E, Martin GR. Signals for death and survival: a two-step mechanism for cavitation in the vertebrate embryo. Cell. . 1995; 83: 279–287.[Medline] [Order article via Infotrieve]
8. Aumailley M, Pesch M, Tunggal L, Gaill F, Fassler R. Altered synthesis of laminin 1 and absence of basement membrane component deposition in ß₁ integrin-deficient embryoid bodies. J Cell Sci. . 2000; 113 (pt 2):259–268.[Abstract]
9. Bagutti C, Wobus AM, Fassler R, Watt FM. Differentiation of embryonal stem cells into keratinocytes: comparison of wild-type and ß₁ integrin-deficient cells. Dev Biol. . 1996; 179: 184–196.[Medline] [Order article via Infotrieve]
10. Brakebusch C, Grose R, Quondamatteo F, Ramirez A, Jorcano JL, Pirro A, Svensson M, Herken R, Sasaki T, Timpl R, Werner S, Fassler R. Skin and hair follicle integrity is crucially dependent on ß₁ integrin expression on keratinocytes. EMBO J. . 2000; 19: 3990–4003.[Medline] [Order article via Infotrieve]
11. Henry MD, Satz JS, Brakebusch C, Costell M, Gustafsson E, Fassler R, Campbell KP. Distinct roles for dystroglycan, ß₁ integrin and perlecan in cell surface laminin organization. J Cell Sci. . 2001; 114: 1137–1144.[Abstract]
12. Sasaki T, Forsberg E, Bloch W, Addicks K, Fassler R, Timpl R. Deficiency of ß₁ integrins in teratoma interferes with basement membrane assembly and laminin-1 expression. Exp Cell Res. . 1998; 238: 70–81.[Medline] [Order article via Infotrieve]
13. Smyth N, Vatansever HS, Murray P, Meyer M, Frie C, Paulsson M, Edgar D. Absence of basement membranes after targeting the LAMC1 gene results in embryonic lethality due to failure of endoderm differentiation. J Cell Biol. . 1999; 144: 151–160.[Abstract/Free Full Text]
14. Murray P, Edgar D. Regulation of programmed cell death by basement membranes in embryonic development. J Cell Biol. . 2000; 150: 1215–1221.[Abstract/Free Full Text]
15. Yang JT, Rayburn H, Hynes RO. Embryonic mesodermal defects in 5 integrin-deficient mice. Development. . 1993; 119: 1093–1105.[Abstract]
16. Yang JT, Bader BL, Kreidberg JA, Ullman-Cullere M, Trevithick JE, Hynes RO. Overlapping and independent functions of fibronectin receptor integrins in early mesodermal development. Dev Biol. . 1999; 215: 264–277.[Medline] [Order article via Infotrieve]
17. Goh KL, Yang JT, Hynes RO. Mesodermal defects and cranial neural crest apoptosis in ₅ integrin-null embryos. Development. . 1997; 124: 4309–4319.[Abstract]
18. George EL, Georges-Labouesse EN, Patel-King RS, Rayburn H, Hynes RO. Defects in mesoderm, neural tube and vascular development in mouse embryos lacking fibronectin. Development. . 1993; 119: 1079–1091.[Abstract]
19. Yang JT, Rayburn H, Hynes RO. Cell adhesion events mediated by 4 integrins are essential in placental and cardiac development. Development. . 1995; 121: 549–560.[Abstract]
20. Gurtner GC, Davis V, Li H, McCoy MJ,