|Home | About | Journals | Submit | Contact Us | Français|
Phosphotyrosine signaling in anchored epithelial cells constitutes a spacially ordained signaling program that largely functions to promote integrin-linked focal adhesion complexes, serving to secure cell anchorage to matrix and as a bidirectional signaling hub that coordinates the physical state of the cell and its environment with cellular functions including proliferation and survival. Cells release their adhesions during processes such as mitosis, migration or tumorigenesis, but the fate of signaling through tyrosine phosphorylation in unanchored cells remains poorly understood. In an examination of epithelial cells in the unanchored state, we find abundant phosphotyrosine signaling, largely recommitted to an anti-adhesive function mediated through the Src family phosphorylation of their transmembrane substrate Trask/CDCP1/gp140. Src-Trask phosphorylation inhibits integrin clustering and focal adhesion assembly and signaling, defining an active phosphotyrosine signaling program underlying the unanchored state. Src-Trask signaling and Src-focal adhesion signaling inactivate each other, constituting two opposing modes of phosphotyrosine signaling that define a switch underline cell anchorage state. Src kinases are prominent drivers of both signaling modes, identifying their position at the helm of adhesion signaling capable of specifying anchorage state through substrate selection. These experimental studies along with concurring phylogenetic evidence suggest that phosphorylation on tyrosine is a signaling function fundamentally linked with the regulation of integrins.
Phosphorylation on tyrosine is one of the fundamental signaling mechanisms developed in metazoans to coordinate cell compliance to a tissue architecture. In adherent cells, much of phosphotyrosine signaling is occurring at focal adhesions where the cellular actin cytoskeleton is engaging with the underlying extracellular matrix (ECM).1–3 The focused nature of this basic signaling mechanism is confirmed in immunostaining experiments using anti-phosphotyrosine antibodies that show predominant staining at focal adhesion sites.3–14 Phosphotyrosine signaling at focal adhesions is largely driven by focal adhesion kinase (FAK) and the Src family kinases (SFKs) and function to promote the assembly of macromolecular complexes that anchor the actin cytoskeleton to sites of adhesion and also serve as signaling hubs that coordinates the physical state of the cell and its environment with many other cellular functions including proliferation and survival.15–21
Preserving adhesion to matrix is particularly important in epithelial tissues, but epithelial cells have the authority to negotiate the requirement for adhesion in order to execute critical functions such as cell migration during development and wound healing or cell rounding during mitosis. Since tyrosine phosphorylation appears to be largely operating at the sites of cell adhesion, what happens to this fundamentally important signaling mechanism in epithelial cells that are not anchored to matrix? It's possible that unanchored cells have a reduced usage of this signaling mechanism or alternatively that phosphotyrosine signaling is recommitted to a biologic function more appropriate to unanchored cells. This question has been of interest to us since independence from anchorage is a hallmark of tumor cells and identifying signaling mechanisms operating during anchorage deprivation may enhance our understanding of anchorage-independent growth. In the analysis of phosphotyrosine signaling in MCF10A immortalized epithelial cells we find abundant tyrosine phosphorylation in the unanchored state. On phosphotyrosine immunoblot analysis it is apparent that distinctly different proteins are phosphorylated in the unanchored state compared to the anchored state (Fig. 1A). This is evident in the different size migration of the predominant phosphotyrosinylated proteins in the two states. Much of the phosphotyrosine signal of the anchored state is contained in proteins within the 110–130 kD size range. Within this size range are many of the well known proteins of the focal adhesion complexes including FAK, p130Cas, α-actinin and vinculin.1,2,17,22,23 In contrast, much of the phosphotyrosine signal of the unanchored state is contained within proteins of 80 and 140 kD sizes. The phosphorylation of these proteins is highly dependent on Src kinases as treatment with any one of several structurally distinct Src-selective tyrosine kinase inhibitors results in the rapid dephosphorylation of these two bands (Fig. 1A and reviewed in ref. 24).
Independent efforts using phosphotyrosine immunoaffinity purification and mass spectroscopy techniques from mitotically detached or trypsin detached epithelial cells led to the purification and identification of Trask/gp140.24,25 This protein or its RNA have also been identified by other investigators searching for gene expression alterations in cancer and named CDCP1/SIMA135.26,27 Trask is a 140 kD type I transmembrane glycoprotein that undergoes proteolytic cleavage within its extracellular domain (ECD) by serine proteases including MTSP1, plasmin and trypsin, generating a smaller 80–85 kD product.24,25,28 Trask is not just a minor component of the phosphotyrosine proteome of unanchored epithelial cells; rather it accounts for the entirety of the abundant 140 and 80 kD signals seen on phosphotyrosine immunoblots. To determine this we depleted Trask from detached MCF10A cell lysates by two methodologies. In the first, we depleted Trask in the cells by expression of an anti-Trask shRNA that eliminates Trask expression. In the second, we depleted the Trask protein from cell lysates, prior to western blotting, by an immunodepletion method using anti-Trask antibodies at concentrations required to remove all Trask protein from the whole cell lysate. Both of these techniques to remove the Trask protein led to removal of the 80 and 140 kD protein bands seen on phosphotyrosine immunoblots (Fig. 1B–E). This shows that Trask alone accounts for the abundant 80 and 140 kD bands seen on phosphotyrosine immunoblots in detached MCF10A cells. This is not unique to MCF10A cells and is a general property of epithelial cells. Trask is phosphorylated in most epithelial cell types when unanchored25,29,30 and an analysis of phosphotyrosine signaling in a panel of adherent or detached epithelial cell lines shows that many, if not all, show the predominant phosphorylation of 80/140 kD proteins during cell detachment (Fig. 2). The relative abundance of 80 kD or 140 kD phosphotyrosine bands varies among different cell types. This parallels the preferential expression of the 80 kD or 140 kD forms of Trask protein expressed since Trask is expressed as a blend of its cleaved and uncleaved 80 kD and 140 kD forms in different cell types.29,31 The phosphorylation of Trask bears no relationship to its cleavage and both the 80 kD and 140 kD cleaved or uncleaved forms undergo tyrosine phosphorylation upon loss of anchorage. The 80 kD form is a proteolytic product of the 140 kD form and can be generated by serine proteases including cellular proteases such as MT-SP1 or proteases used in cell culture such as trypsin.24,25,28 As such, the use of trypsin to induce cell detachment results in the concomitant cleavage and phosphorylation of Trask. But the use of EDTA more clearly reveals the detachment-induced phosphorylation of Trask without the confounding attribute of cleavage.32 Taken together, these experiments show that the phosphorylation of Trask accounts for the majority of phosphotyrosine signaling activity in unanchored epithelial cells. Such signaling activity is not easily seen in fibroblasts since Trask expression is very low or undetectable in mesenchymal cell types.32 The phosphorylation of Trask is tightly linked temporally to the unanchored state such that it is phosphorylated instantly upon loss of anchorage, it's phosphorylation is maintained for as long as cells remains unanchored, and it is dephosphorylated upon re-establishment of anchorage.30,32
Immunostaining of MCF10A cells in the adherent and suspended states is also consistent with the biochemical data shown above that phosphotyrosine signaling adopts a new character in the unanchored state. On immunofluorescent microsopy it is apparent that phosphotyrosine signaling largely localizes to focal adhesion sites as does pY-FAK, one of the tyrosine phosphorylated proteins within focal adhesions (Fig. 3). However, although focal adhesions are disrupted and FAK is dephosphorylated upon loss of anchorage, phosphotyrosine signaling persists in the unanchored state and now localizes to the cell membrane in a punctate pattern similar to phosphorylated Trask (Fig. 3). This is also consistent with the identity of Trask as the dominant phosphotyrosine signaling protein of the unanchored state.
The functions of this phosphotyrosine signaling mechanism in unanchored epithelial cells were interrogated in a number gain-of-function and loss-of-function experiments published recently in reference 30. In a number of monolayer and cell-bead adhesion assays it was shown that phosphorylated Trask inhibits cell adhesion. This is specifically mediated through its phosphorylation, since phosphorylation-defective mutants of Trask fail to reproduce this anti-adhesive phenotype. The inhibition of adhesion is not due to effects on integrin affinity state as shown by two different assays of integrin conformation. Rather Trask, when phosphorylated, is found within integrin immune complexes and functions to inhibit integrin clustering and prevent or disrupt the formation of focal adhesion complexes.
The anti-adhesive function of Trask is specifically and entirely mediated through the tyrosine phosphorylation of its ICD and the role of the ECD remains unclear at this time. The functions of the Trask ECD were interrogated through the generation of a number of Trask mutant constructs.31 A Trask mutant lacking its ECD shows an inhibition of cell adhesion similar to wildtype Trask. However, the deletion of the Trask ICD or the mutation of phosphorylated tyrosine residues within the ICD completely abolishes its anti-adhesive function. The ECD also does not appear to function to regulate phosphorylation of the ICD. A cleavage-resistant mutant or a cleavage-product mimick of Trask are equally competent at phosphorylation when overexpressed.31 Furthermore, cells exhibit varying blends of uncleaved and cleaved Trask. Whichever form is expressed undergoes phosphorylation when anchorage is lost.29,31 One group has suggested a link between Trask/CDCP1 cleavage and its phosphorylation, but this was shown in experiments using protease-induced Trask cleavage.28 We believe the use of proteases can confound the study of Trask phosphorylation, since even the disruption of a fraction of cellular focal adhesion complexes, before the onset of visible signs of the loss of cell adhesion, induces the phosphorylation of Trask. EDTA-induced cell detachment and Trask phosphorylation is not accompanied by the proteolytic cleavage of Trask.30–32 The preponderance of evidence suggests that the phosphorylation of Trask is induced by cell detachment and not by the proteolytic cleavage of its ECD.
The identification of Trask as an active phosphotyrosine signal of the unanchored state reveals that signaling through tyrosine phosphorylation is largely linked with the regulation of cell adhesion, both in the anchored and in the unanchored state in epithelial cells. In the anchored state, phosphotyrosine signaling is concentrated at focal adhesions, and functions to secure them. In the unanchored state, phosphotyrosine signaling functions to prevent the formation of focal adhesions or disrupt them. In fact it appears that the two functionally opposing modes of phosphotyrosine signaling are mutually exclusive with eachother and constitute a phosphotyrosine switch that underlies the state of anchorage. The switch function is revealed by the finding that the tyrosine phosphorylation of Trask leads to the rapid dephosphorylation of focal adhesion proteins and similarly the assembly and phosphorylation of focal adhesion proteins leads to the rapid dephosphorylation of Trask.30 Experimental manipulations also show that the dephosphorylation of Trask is required for the phosphorylation of focal adhesion proteins and the dephosphorylation of focal adhesion proteins is required for the phosphorylation of Trask. These two phosphotyrosine signaling mechanisms do not appear to temporally co-exist and constitute a switch that defines the state of anchorage. In fact, Trask knockdown cells have an abnormal persistence of focal adhesion protein phosphorylation even when detached,30 consistent with a functional reciprocity in the two phosphotyrosine signaling modes.
These data link phosphotyrosine signaling with cell adhesion even more tightly than had previously been envisioned. It appears that phosphotyrosine signaling is a fundamental mechanism by which cell adhesion is regulated. This also makes sense in an evolutionary context. In a recent phylogenetic analysis of the integrin adhesome, Zaidel-Bar found that while many gene families important in adhesion appear to predate the emergence of integrins, tyrosine kinases developed simultaneously with the integrin family.33 The evolutionary co-emergence of these gene families provides yet another level of evidence of their functional relationship. Further consistent with the functional link between the integrins and phosphotyrosine signaling is evident in studies of unicellular organisms that exhibit colonial behavior reminiscent of multicellular organisms. In particular, the genome of the choano-flagellate Monosiga brevicollis has been studied in some depth. This organism, which may be a unicellular precursor to metazoans, has both a family of tyrosine kinases and a family of integrin receptor genes, not typically seen in unicellular organisms.34 The phylogenetic evidence is consistent with the biochemical and experimental evidence in suggesting that the phosphorylation of tyrosine is a signaling mechanism intimately related to the functions of integrins. Orthologues of the Trask gene itself, however, are not found in very simple organisms and its possible that other protein substrates of tyrosine kinases mediate the negative regulation of integrins in these organisms.
What introduces significant complexity in understanding how phosphotyrosine signaling regulates cell adhesion is the unexpected vacillating role played by SFKs. In the simplest schema to explain how phosphotyrosine signaling could operate in two modes, one would envision that certain tyrosine kinases would function to promote cell adhesion while another group of tyrosine kinases would function to oppose it. But it appears that SFKs play both roles in a decisive way. Trask appears to be uniquely phosphorylated by SFKs as evidenced by studies with Src-selective inhibitors as well as co-transfection studies in SYF cells24,29 and therefore SFKs inhibit cell adhesion through the phosphorylation of Trask. However, SFKs are also well known central players in the formation of focal adhesion complexes, in particular in complex with activated FAK and phosphorylate many substrates at focal adhesions, stabilizing them and linking them to the actin cytoskeleton.8,35–37 This apparent dichotomous role of SFKs suggests that they may embody functions more complex than workhorse functions. The traditional view of kinases is that they are regulated at the level of activity and by turning on and off, they function to transmit signals along a pathway. The fact that SFKs can phosphorylate focal adhesion targets and can phosphorylate Trask in exclusion of one another with opposing biological affects suggests that SFKs function as branch points in cell signaling with the capacity for temporal discrimination in selecting substrates.
The dichotomous role of SFKs in the regulation of both pro-adhesive and anti-adhesive signaling mechanisms also may help reconcile some of the conceptually conflicting evidence regarding the functions of SFKs. The role of SFKs in regulating focal adhesion formation and signaling downstream of integrin activation is well established. Activated Src localizes to focal adhesions, an event that is required for cell spreading on fibronectin.38 Fibroblasts deficient in SFKs have focal contacts but reduced tyrosine phosphorylation at focal adhesions and defective cell adhesion to matrix.36,38,39 But the kinetics of focal adhesion assembly are more rapid in these cells, introducing complexity in the model.36 Furthermore, the pro-adhesive role of SFKs suggested by these loss-of-function experiments are not reciprocated by gain-of-function experiments. The constitutively activated v-src oncogene product interacts with focal contacts, phosphorylating target proteins within them.40,41 However, the activities of the v-src product are destructive to focal adhesions and in fact v-src transformed cells appear to have significantly reduced focal adhesions.42 To reconcile these datasets, SFKs have been proposed to function in the turnover of focal adhesions, a dynamic model that is tolerant to potentially opposing roles for SFKs.43 But the SFK effectors that could mediate the proposed anti-adhesive component of the turnover model have not been well defined. The identification of Trask as an anti-adhesive effector of SFKs now fills a conceptual gap that enables a better mechanistic model of how SFKs can promote the turnover of focal adhesions.
A caveat to this is that much of the cited studies of focal adhesions were performed in fibroblasts and Trask is abundantly expressed in epithelial cells but appears to have very low expression in mesenchymal cells.32 But there is some, albeit low expression in fibroblasts and although we can't detect the expression of Trask in 3T3 cells, we can detect pY-Trask in v-src transformed 3T3 cells by immunoprecipitation and phosphotyrosine immunoblotting (Fig. 4). This may be due to the sensitivity limits of our antibodies and their weaker affinity for the mouse Trask protein and the likely much higher sensitivity and cross-reactivity of anti-phosphotyrosine antibodies with mouse Trask. The higher expression of Trask in epithelial cells may reflect the more complex nature of adhesion in epithelial cells, including the existence of both cell-matrix and cell-cell adhesions and junctions, the link with cell polarity, survival and many other functions.
The fact that the phosphorylation of Trask defines a state of anchorage loss makes pY-Trask a marker of the unanchored state in epithelial cells. This is consistent and confirmed across the various in vitro and in vivo experimental models and tissue sections. The phosphorylation of Trask is widely seen in epithelial cells in vitro when detached (Fig. 1).25,32 When MCF10A cells are grown in 3D, Trask is phosphorylated in the early days when cells are still in an unanchored state. But upon the secretion and establishment of a basement membrane and anchorage to it, Trask is dephosphorylated (Fig. 5). In human tissue sections, Trask phosphorylation is not seen in the normal epithelium, which is entirely anchored. But pY-Trask can be found if one searches for physiologically detaching cells. This is best seen during mitotic detachment or in cells undergoing physiologic shedding.32 The mechanical induction of anchorage loss by experimental skin wounding in the mouse epidermis also leads to the phosphorylation of Trask.30 A conceptual enigma develops when interrogating tumor tissues. The phosphorylation of Trask is seen in many epithelial tumors, including preinvasive, invasive and metastatic tumors.29 The phosphorylation is usually seen in a regional and patchy distribution (Fig. 6). If pY-Trask is a marker of the unanchored state, this seemingly indicates that these regions of tumor have lost anchorage. This is not readily apparent in their histological or architectural appearance and the implications of pY-Trask in these tumor regions are currently not understood. Traditional concepts lead us to assume that tumors are anchored to a surrounding matrix, even if it may be disorganized and without architecture. But the phosphorylation of Trask in tumors suggests that there is at least an abnormal or deficient state of anchorage in these regions. What an abnormal or deficient state of anchorage could mean is up to speculation. It could mean much reduced number of adhesion sites or defective assembly of adhesion complexes or defective signaling through the integrin adhesion complex. This may be due to abnormalities in the composition of the surrounding matrix. Specifically, this may be due to the absence of a continuous basal lamina, which typically underlies the basal surface of epithelial cells in the normal epithelium but is highly abnormal or missing in epithelial tumors.44–46 Much more work is needed to understand the relevance of phosphorylated and unphosphorylated Trask in tumors in vivo. The study of anchorage in vivo is inherently difficult since we currently lack a firm understanding of cell anchorage in tissues. Work of the past decades has provided considerable insight into the mechanisms that mediate cell adhesion including the conformational regulation of integrin affinity state and its relationship to the regulation of matrix protein assembly and polymerization, and the formation of macromolecular complexes at the intracellular tail of clustered integrins linking the adhesion structures to the actin cytoskeleton.37,47–49 However, these insights come almost entirely from the study of cultured cells in monolayer and such adhesion structures are not readily apparent in 3D models or in vivo and the nature of cell adhesion in vivo continues to be the subject of debate and speculation.50–54
Signaling through the phosphorylation of tyrosine appears to have two opposing functional modes intimately linked with the regulation of integrin receptors and the function of cell adhesion. The two modes appear to function in reciprocity to one another revealing a phosphotyrosine switch function that underlies the state of anchorage. The positive and negative regulation of cell adhesion through SFKs reveals a complexity in the functions of this tyrosine kinase family that awaits further mechanistic insight.
This work was funded by the National Institutes of Health CA113952 (M.M.M.). D.S. was funded by a Susan G. Komen for the Cure Postdoctoral Fellowship. CHW was funded by a California Breast Cancer Research Program Postdoctoral Fellowship. We acknowledge the use of core facilities of the UCSF Helen Diller Family Comprehensive Cancer Center including the immunohistochemistry core and The Laboratory for Cell analysis.