Integrin functions play a key role in the differentiation of thymocytes in vivo.

T cells express a variety of surface proteins as they develop to maturity in the thymus. In addition to the TCR-CD3 complex and the two major coreceptors, CD4 and CD8, other surface proteins expressed include receptors for cytokines, growth factors, counterreceptors, and extracellular matrix molecules. To determine the role of integrin adhesion receptors in T cell development, we have expressed a trans-dominant inhibitor of integrin function in the thymus. This inhibitor leads to a block of adhesion to fibronectin due to reduced activation of integrin receptors. This reduced adhesion leads to a partial block in differentiation from CD4-CD8- cells to CD4+CD8+ cells, after the CD25+ stage, suggesting that integrins are important during Lck-mediated differentiation. Furthermore, the overall production of CD4+ cells is reduced compared with that of CD8+ cells without changes in negative selection, suggesting that integrins may be involved in the determination of the fate of the cell as well. These results demonstrate that integrin receptor function is required for proper thymocyte development in vivo.

Focal adhesion motility revealed in stationary fibroblasts.

Focal adhesions (FAs) are clustered integrins and associated proteins that mediate cell adhesion and signaling. A green fluorescent protein-beta1 integrin chimera was used to label FAs in living cells. In stationary cells, FAs were highly motile, moving linearly for several plaque lengths toward the cell center. FA motility was independent of cell density and resulted from contraction of associated actin fibers. In migrating cells, FAs were stationary and only moved in the tail. FA motility in stationary cells suggests that cell movement may be regulated by a clutch-like mechanism by which the affinity of integrins to substrate may be altered in response to migratory cues.

Integrins interact with focal adhesions through multiple distinct pathways.

Integrin signaling involves oligomerization and a transmembrane conformational change induced by receptor occupancy. Previous work has shown that subsets of focal adhesion-associated proteins are recruited to integrins as a result of clustering, ligand binding, or both. However, it is unclear whether these discrete subsets reflect the differential binding of cytoplasmic proteins to the integrin or whether a single protein or set of proteins binds the integrin and is differentially activated by receptor occupancy or clustering. To address this question, we made mutations of the beta1 integrin cytoplasmic domain in the context of a single subunit chimera and studied their activation of various known integrin-mediated signaling pathways. We show here that the indirect association of the integrin with actin is distinct from its interactions with both preformed focal adhesions and FAK. Therefore, multiple independent signaling pathways exist from the integrin to the focal adhesion, which may reflect the association of independent factors with the integrin beta1 cytoplasmic domain.

Differential regulation of beta1 integrins by chemoattractants regulates neutrophil migration through fibrin.

Chemoattractants differ in their capacity to stimulate neutrophils to adhere to and to migrate through matrices containing fibrin. Formyl methionyl leucyl phenylalanine (fMLP) stimulates neutrophils to adhere closely to, but not to migrate into, fibrin gels. Leukotriene B4 (LTB4) stimulates neutrophils to adhere loosely to and to migrate through fibrin gels. We report that alpha5beta1 integrins regulate the different migratory behaviors on fibrin gels of neutrophils in response to these chemoattractants. fMLP, but not LTB4, activated neutrophil beta1 integrins, as measured by binding of mAb 15/7 to an activation epitope on the beta1 integrins. Antibodies or peptides that block alpha5beta1 integrins prevented fMLP-stimulated neutrophils from forming zones of close apposition on fibrin and reversed fMLP's inhibitory effect on neutrophil chemotaxis through fibrin. In contrast, neither peptides nor antibodies that block beta1 integrins affected the capacity of LTB4-stimulated neutrophils to form zones of loose apposition or to migrate through fibrin gels. These results suggest that chemoattractants generate at least two different messages that direct neutrophils, and perhaps other leukocytes, to accumulate at specific anatomic sites: a general message that induces neutrophils to crawl and a specific message that prepares neutrophils to stop when they contact appropriate matrix proteins for activated beta1 integrins.

The membrane proximal region of the integrin beta cytoplasmic domain can mediate oligomerization.

Integrin-ligand binding generates many intracellular signals, including signals to initiate focal contact formation and to regulate cellular decisions concerning growth and differentiation. Oligomerization of the beta subunit cytoplasmic domain appears to be required for many of these events. In order to study these processes, we have generated a novel chimeric protein, consisting of the chicken integrin beta 1 cytoplasmic domain connected to the central rod domain of a neuronal intermediate filament, alpha-internexin. This chimeric protein, when expressed transiently in 293T cells, oligomerizes in a beta cytoplasmic domain-dependent manner. This oligomerization requires the membrane proximal amino acids LLMII of the beta 1 cytoplasmic domain, as demonstrated by deletion analysis. Therefore, the integrin beta cytoplasmic domain in this system contains an oligomerization function, which may provide some insight as to the function of intact integrins in vivo.

Role of the I-domain in collagen binding specificity and activation of the integrins alpha1beta1 and alpha2beta1.

Adhesion to collagens by most cell types is mediated by the integrins alpha1beta1 and alpha2beta1. Both integrin alpha subunits belong to a group which is characterized by the presence of an I domain in the N-terminal half of the molecule, and this domain has been implicated in the ligand recognition. Since purified alpha1beta1 and alpha2beta1 differ in their binding to collagens I and IV and recognize different sites within the major cell binding domain of collagen IV, we investigated the potential role of the alpha1 and alpha2 I domains in specific collagen adhesion. We find that introducing the alpha2 I domain into alpha1 results in surface expression of a functional collagen receptor. The adhesion mediated by this chimeric receptor (alpha1-2-1beta1) is similar to the adhesion profile conferred by alpha2beta1, not alpha1beta1. The presence of alpha2 or alpha1-2-1 results in preferential binding to collagen I, whereas alpha1 expressing cells bind better to collagen IV. In addition, alpha1 containing cells bind to low amounts of a tryptic fragment of collagen IV, whereas alpha2 or alpha1-2-1 bearing cells adhere only to high concentrations of this substrate. We also find that collagen adhesion of NIH-3T3 mediated by alpha2beta1 or alpha1-2-1beta1, but not by alpha1, requires the presence of Mn2+ ions. This ion requirement was not found in CHO cells, implicating the I domain in cell type-specific activation of integrins.

Integrin receptor signaling: the propagation of an alpha-helix model.

Upon ligand binding to integrin receptors, a transmembrane conformation change occurs, which is required for the engagement of the actin cytoskeleton. Integrin receptor latency clearly involves the proximal portions of the alpha and beta cytoplasmic domains. Several experiments suggest that these two regions, which are highly conserved among integrins, may be associated, and this association is the structural basis for latency. We propose that ligand binding leads to a disruption of this association, which allows for the folding of the proximal beta cytoplasmic domain. Thus, in this model, the alpha chain association keeps the beta unfolded, and ligand binding leads to the propagation of an alpha helix from the transmembrane domain through the proximal beta cytoplasmic domain, leading to signal transduction.

The adaptor protein Shc couples a class of integrins to the control of cell cycle progression.

We provide evidence that a class of integrins combines with the adaptor Shc and thereby with Grb2. Coimmunoprecipitation and mutagenesis experiments indicate that the recruitment of Shc is specified by the extracellular or transmembrane domain of integrin alpha subunit and suggest that this process is mediated by caveolin. Mutagenesis and dominant-negative inhibition studies reveal that Shc is necessary and sufficient for activation of the MAP kinase pathway in response to integrin ligation. Mitogens and Shc-activating integrins cooperate to promote transcription from the Fos serum response element and transit through G1. In contrast, adhesion mediated by integrins not linked to Shc results in cell cycle arrest and apoptosis even in presence of mitogens. These findings indicate that the association of specific integrins with Shc regulates cell survival and cell cycle progression.

The membrane-cytoplasm interface of integrin alpha subunits is critical for receptor latency.

Localization of integrin receptors to focal contact sites occurs upon ligand binding. This activity is latent, since unoccupied integrin receptors do not localize to focal contacts. Deletion analysis has revealed that the alpha cytoplasmic domains is required for the maintenance of integrin receptor latency. Our current hypothesis for the mechanism of integrin post-ligand binding events is that there is a change in relationship of alpha and beta cytoplasmic domains, which overcomes receptor latency. One possible mechanism for such a change would involve the amino acid residues at the membrane-cytoplasm interface. To test this hypothesis, we have produced point mutations in the human integrin alpha 1 subunit. These mutations had no effect on the adhesion via alpha 1 beta 1 to its ligand, collagen IV. However, receptor latency is lost in one of these mutants, leading to constitutive focal contact localization. This effect did not occur in receptors with an exchange of intracellular domains, suggesting that the mechanism of loss of latency involves a relative motion of the integrin chains. These results suggest a model in which post-ligand binding events in integrin receptors are associated with changes in the position of the alpha and beta cytoplasmic domains.

Intracellular calcium and calcineurin regulate neutrophil motility on vitronectin through a receptor identified by antibodies to integrins alphav and beta3.

Buffering of intracellular calcium ([Ca2+]i) or inhibition of the calcium/calmodulin-dependent phosphatase, calcineurin, results in neutrophils being unable to detach from vitronectin with a consequent loss of motility. Treatment of [Ca2+]i-buffered or calcineurin-inhibited neutrophils with monoclonal antibodies (MoAbs) to beta3 or alphav beta3 integrins allowed neutrophils to detach and restored motility. Quantitative immunofluorescence and flow cytometry showed that MoAbs specific for beta3, alphav, or alphav beta3 integrins bind to neutrophils. Immunolocalization studies using antibodies to the highly conserved cytoplasmic domains of alphav and beta3 also identified the receptor on neutrophils. Whereas antibodies to alphav, alphav beta3, and beta3 recognized the receptor in intact cells, only the beta3 MoAb immunoprecipitated the receptor from a neutrophil cell lysate. The alpha subunit co-immunoprecipitated by the beta3 antibody reacted with an antibody to alphav by Western blot. Peptide maps of V8 protease digests showed a strong similarity in alpha and beta chains precipitated by antibodies to beta3 from neutrophils and endothelial cells. These results indicate that [Ca2+]i and calcineurin regulate neutrophil motility on vitronectin through an alphav beta3-like receptor. Although we cannot rule out the possibility that neutrophils have an isoform of alphav, such an isoform would have to be similar enough to react with alphav- and alphav beta3-specific MoAbs in intact cells.

Cell adhesion to extracellular matrix regulates the life cycle of integrins.

The expression of alpha 5 beta 1 integrin on the surface of fibroblasts requires adhesion to substratum. We have examined the basis for this adhesion-dependent surface expression by comparing the life cycle of integrins in parallel cultures of adherent and nonadherent cells. Results of biosynthetic labeling experiments in NRK fibroblasts showed that the synthesis and biosynthetic processing of the beta 1 integrin subunit proceed in the absence of cell attachment; however, when examining the behavior of preexisting cell surface integrins, we observed that the alpha beta 1 integrins are internalized and degraded when adhesion to substratum is blocked. A kinetic analysis of integrin internalization in cycloheximide-treated NRK cells showed that each of the fibroblast integrins we examined (in both the beta 1 and beta 3 families) are lost from the cell surface after detachment from substratum. Thus, the default integrin life cycle in fibroblasts involves continuous synthesis, processing, transport to the cell surface, and internalization/degradation. Interestingly, studies with NIH-3T3 cells expressing alpha 1 beta 1 integrin showed that the loss of cell-surface alpha 5 beta 1 integrin is blocked by adhesion of cells to dishes coated with type IV collagen (a ligand for alpha 1 beta 1 integrin) as well as fibronectin. Similarly, adhesion of these cells to dishes coated with type IV collagen stabilizes the surface expression of alpha 5 beta 1 as well as alpha 1 beta 1 integrin. We propose that the adhesion of fibroblasts to extracellular matrix protein alters the integrin life cycle and permits retention of these proteins at the cell surface where they can play important roles in transmitting adhesion-dependent signals.

Assembly and function of integrin receptors is dependent on opposing alpha and beta cytoplasmic domains.

The membrane proximal regions of integrin alpha and beta subunits are highly conserved in evolution. In particular, all integrin alpha subunits share the KXGFFKR sequence at the beginning of their cytoplasmic domains. Previous work has shown that this domain is important in integrin receptor assembly. Using chimeric integrin alpha and beta subunits, we show that the native cytoplasmic domains of both subunits must be present for efficient assembly. Most strikingly, chimeric alpha 1 and beta 1 subunits with reciprocally swapped intracellular domains dimerize selectively into collagen IV receptors expressed at high levels on the surface. However, these receptors, which bind ligand efficiently, are deficient in a variety of post-ligand binding events, including cytoskeletal association and induction of tyrosine phosphorylation. Furthermore, deletion of the distal alpha cytoplasmic domain in the swapped heterodimers leads to ligand-independent focal contact localization, which also occurs in wild-type subunits when the distal cytoplasmic domain is deleted. These results show that proper integrin assembly requires opposed alpha and beta cytoplasmic domains, and this opposition prevents ligand-independent focal contact localization. Our working hypothesis is that these two domains may associate during receptor assembly and provide the mechanism for integrin receptor latency.

Integrin beta 1 cytoplasmic domain dominant negative effects revealed by lysophosphatidic acid treatment.

Integrin receptors localize to focal contact sites and interact with the cytoskeleton via the beta 1 cytoplasmic domain. To study the role of this domain in adhesion, we have expressed in NIH 3T3 cells a cDNA consisting of the interleukin 2 receptor alpha subunit extracellular and transmembrane domains, connected to the integrin beta 1 cytoplasmic domain (IL2R-beta 1). Since the extracellular domain of the chimeric protein has no role in adhesion, this protein could uncouple adhesion from intracellular events. As expected, in a cell line expressing IL2R-beta 1, this chimera was directed to focal contact sites. Unexpectedly, the cells exhibited normal adhesion to fibronectin (FN). However, when a rapid reorganization of the cytoskeleton was induced using lysophosphatidic acid (LPA), IL2R-beta 1 cells detached from FN in contrast to wild-type cells. The detachment in response to LPA could be prevented with cytochalasin D, an inhibitor of actin polymerization. These results imply that a beta 1 cytoplasmic domain, which is uncoupled from adhesion, can compete with the cytoplasmic domain of native integrin beta 1 for cytoskeletal proteins. As a consequence, the IL2R-beta 1 protein acts as a dominant negative effector of adhesion by disrupting the integrin-cytoskeleton connection.

The role of the I domain in ligand binding of the human integrin alpha 1 beta 1.

We report here the analysis of potential ligand binding domains within the human integrin alpha 1 subunit, a known collagen/laminin receptor. This integrin is effectively blocked by the mouse monoclonal antibody 1B3.1. A truncated version of the alpha 1 subunit lacking the NH2-terminal half of the extracellular domain is not recognized by monoclonal antibody 1B3.1. Furthermore, we have isolated a cDNA containing the I domain from chicken alpha 1 bearing significant homology to the human and rat alpha 1 sequences. Replacing the human I domain with its chicken counterpart led to the surface expression of a functional heterodimer with endogenous mouse beta 1 on NIH 3T3 cells. However, 1B3.1 does not bind to the chicken/human chimera, demonstrating that the human alpha 1 I domain is required for epitope recognition. Mutation of Asp253 within the I domain to alanine resulted in surface expression of an alpha beta heterodimer recognized by 1B3.1 but with markedly reduced binding to collagen IV or laminin. Since a previously reported mutation of a homologous Asp in the Mac-1 I domain has similar consequences, these results suggest a central role for the I domain in ligand recognition for all integrin alpha subunits containing this domain.

Ligand-dependent and -independent integrin focal contact localization: the role of the alpha chain cytoplasmic domain.

Many integrin receptors localize to focal contact sites upon binding their ligand. However, unoccupied integrin receptors do not localize to focal contact sites. Because the integrin beta 1 cytoplasmic domain appears to have a focal contact localization signal, there must be a mechanism by which this domain is kept inactive in the unoccupied state and becomes exposed or activated in the occupied receptor. We considered that this mechanism involves the alpha subunit cytoplasmic domain. To test this hypothesis, we have established two NIH 3T3 cell lines that express either the human alpha 1 wild-type subunit (HA1 cells) or the cytoplasmic domain deleted alpha 1 subunit (CYT cells). Both cell lines express similar levels of the human alpha 1 subunit, and there is no significant effect of the deletion on the dimerization and surface expression of the receptor. Furthermore, the deletion had no effect on the binding or adhesion via alpha 1 beta 1 to its ligand collagen IV. However, when these two cell lines are plated on fibronectin (FN), which is a ligand for alpha 5 beta 1 but not for alpha 1 beta 1, there is a striking difference in the cellular localization of alpha 1 beta 1. The HA1 cells show only alpha 5 in focal contacts, without alpha 1, demonstrating that all of the integrin localization is ligand dependent. In contrast, when the CYT cells are plated on FN, the mutant alpha 1 appears in focal contacts along with the alpha 5/beta 1. Thus, there is both ligand-dependent (alpha 5/beta 1) and ligand-independent (alpha 1/beta 1) focal contact localization in these cells. The truncated alpha 1 also localized to focal contacts in a ligand-independent manner on vitronectin. We conclude that the mutant alpha 1 no longer requires ligand occupancy for focal contact localization. These data strongly suggest that the alpha cytoplasmic domain plays a role in the normal ligand-dependent integrin focal contact localization.

Expression of native and truncated forms of the human integrin alpha 1 subunit.

We report here the molecular cloning of cDNAs encoding for the human integrin alpha 1 subunit. The sequence is characteristic of an I domain containing integrin alpha subunit, with a high degree of homology to the rat integrin alpha 1 subunit, including complete identity of the transmembrane and cytoplasmic domains between the two species. The human cDNA directs the expression in mouse NIH 3T3 cells of authentic human alpha 1 protein as demonstrated by the reactivity of this subunit with two human-specific anti-alpha 1 monoclonal antibodies. This exogenous integrin specifically binds to type IV collagen in a Mg(2+)-dependent fashion. We have expressed in both transient systems and in stable cell lines truncated, soluble forms of the human alpha 1 subunit combined with truncated, soluble forms of beta 1 subunits. Although soluble beta 1 subunit was found in the media when the corresponding cDNA was used, the secretion of the soluble alpha 1 subunit was found to be dependent on dimerization with soluble beta 1. Co-transfection of truncated human alpha 1 cDNA with truncated forms of either the human or avian beta 1 cDNA led to efficient secretion of alpha 1.beta 1 heterodimers. These soluble heterodimers specifically bind to collagen IV in a manner similar to their full-length counterparts. Biosynthetic studies using stably expressing cell lines demonstrate that the soluble heterodimers and the native heterodimers are formed independently, strongly suggesting that the transmembrane or cytoplasmic domains of alpha and beta subunits are involved in the assembly of native heterodimers.

Cell attachment controls fibronectin and alpha 5 beta 1 integrin levels in fibroblasts. Implications for anchorage-dependent and -independent growth.

We have examined the effect of cell attachment on fibronectin and alpha 5 beta 1 integrin levels in three distinct anchorage-dependent fibroblast cell lines. Analysis of long-term biosynthetically labeled proteins from parallel cultures of adherent and nonadherent cells showed that the steady-state level of extracellular fibronectin is decreased upon loss of attachment. Pulse labeling studies and Northern blot analyses showed that the decrease occurs post-synthetically. A combined approach of surface radioiodination and biosynthetic labeling also demonstrated a selective post-synthetic decrease in cell-surface expression of the alpha 5 beta 1 integrin upon loss of cell attachment. Overall, we estimate that extracellular fibronectin and cell surface alpha 5 beta 1 integrin levels are reduced 5-7-fold in NIH-3T3, 15-20-fold in AKR-2B, and 50-fold in NRK fibroblasts. Finally, we find decreased total (serum- and cell-derived) fibronectin bound to the surface of nonadherent cells consistent with the reduced expression of alpha 5 beta 1 integrin. These results demonstrate a systematic down-regulation of fibronectin and its major receptor upon loss of attachment and suggest a potential mechanism involved in maintenance of the anchorage-dependent phenotype.

Mapping of the functional determinants of the integrin beta 1 cytoplasmic domain by site-directed mutagenesis.

We describe here the expression of deletion mutants of the cytoplasmic domain of the avian integrin beta 1 subunit. These mutants, which contain termination codons at positions 767, 776, 791, and 800, were transfected into mouse 3T3 cells to determine which sequences were essential for localization of integrins into focal contact sites. In all cases, high-level expression of the truncated avian integrins was obtained. Heterodimers were formed between the exogenous truncated avian beta 1 subunits and endogenous mouse alpha subunits, and these heterodimers were efficiently exported to the cell surface. The longest truncated beta 1 subunit tested, which is only four amino acids shorter than the wild type, does localize to focal contacts. In contrast, beta 1 subunits with moderately long truncations of the cytoplasmic domain failed to localize to focal contacts, including one which contains the consensus sequence for tyrosine phosphorylation. Surprisingly, a mutant subunit in which the bulk of the cytoplasmic domain was missing (but the segment nearest the membrane including the dibasic residues (RR) remained) did localize weakly to focal contacts. These results implicate the peptide segment nearest to the transmembrane region in focal contact localization. In addition, mutant subunits that included this segment together with a larger portion of the cytoplasmic domain did not localize as well as the shorter form, suggesting that these cytoplasmic domain segments are defective, presumably because of abnormal folding.