Regulation of Cell–Cell Adhesion by Rac and Rho Small G Proteins in MDCK Cells

The Rho small G protein family, consisting of the Rho, Rac, and Cdc42 subfamilies, regulates various cell functions, such as cell shape change, cell motility, and cytokinesis, through reorganization of the actin cytoskeleton. We show here that the Rac and Rho subfamilies furthermore regulate cell–cell adhesion. We prepared MDCK cell lines stably expressing each of dominant active mutants of RhoA (sMDCK-RhoDA), Rac1 (sMDCK-RacDA), and Cdc42 (sMDCK-Cdc42DA) and dominant negative mutants of Rac1 (sMDCK-RacDN) and Cdc42 (sMDCK-Cdc42DN) and analyzed cell adhesion in these cell lines. The actin filaments at the cell–cell adhesion sites markedly increased in sMDCK-RacDA cells, whereas they apparently decreased in sMDCK-RacDN cells, compared with those in wild-type MDCK cells. Both E-cadherin and β-catenin, adherens junctional proteins, at the cell–cell adhesion sites also increased in sMDCK-RacDA cells, whereas both of them decreased in sMDCK-RacDN cells. The detergent solubility assay indicated that the amount of detergent-insoluble E-cadherin increased in sMDCK-RacDA cells, whereas it slightly decreased in sMDCK-RacDN cells, compared with that in wild-type MDCK cells. In sMDCK-RhoDA, -Cdc42DA, and -Cdc42DN cells, neither of these proteins at the cell–cell adhesion sites was apparently affected. ZO-1, a tight junctional protein, was not apparently affected in any of the transformant cell lines. Electron microscopic analysis revealed that sMDCK-RacDA cells tightly made contact with each other throughout the lateral membranes, whereas wild-type MDCK and sMDCK-RacDN cells tightly and linearly made contact at the apical area of the lateral membranes. These results suggest that the Rac subfamily regulates the formation of the cadherin-based cell– cell adhesion. Microinjection of C3 into wild-type MDCK cells inhibited the formation of both the cadherin-based cell–cell adhesion and the tight junction, but microinjection of C3 into sMDCK-RacDA cells showed little effect on the localization of the actin filaments and E-cadherin at the cell–cell adhesion sites. These results suggest that the Rho subfamily is necessary for the formation of both the cadherin-based cell– cell adhesion and the tight junction, but not essential for the Rac subfamily-regulated, cadherin-based cell– cell adhesion.     

Polarized Organization of the Cytoskeleton: Regulation by Cell Polarity Proteins

Polarity is one of the fundamental properties displayed by living organisms. In metazoans, cell polarity governs developmental processes and plays an essential role during maintenance of forms of tissues as well as their functions. The mechanisms of establishment and maintenance of cell polarity have been investigated extensively in the last two decades. This has resulted in identification of “core cell polarity modules” that control anterior–posterior, front–rear and apical–basal polarity across various cell types. Here, we review how these polarity modules interact closely with the cytoskeleton during establishment and maintenance of cytoskeletal polarity. We further suggest that reciprocal interactions between cell polarity modules and the cytoskeleton consolidate the initial weaker polarity, arising from an external cue, into a committed polarized system.     

Tyrosine Phosphorylation and Src Family Kinases Control Keratinocyte Cell–Cell Adhesion

In their progression from the basal to upper differentiated layers of the epidermis, keratinocytes undergo significant structural changes, including establishment of close intercellular contacts. An important but so far unexplored question is how these early structural events are related to the biochemical pathways that trigger differentiation. We show here that β-catenin, γ-catenin/plakoglobin, and p120-Cas are all significantly tyrosine phosphorylated in primary mouse keratinocytes induced to differentiate by calcium, with a time course similar to that of cell junction formation. Together with these changes, there is an increased association of α-catenin and p120-Cas with E-cadherin, which is prevented by tyrosine kinase inhibition. Treatment of E-cadherin complexes with tyrosine-specific phosphatase reveals that the strength of α-catenin association is directly dependent on tyrosine phosphorylation. In parallel with the biochemical effects, tyrosine kinase inhibition suppresses formation of cell adhesive structures, and causes a significant reduction in adhesive strength of differentiating keratinocytes. The Fyn tyrosine kinase colocalizes with E-cadherin at the cell membrane in calcium-treated keratinocytes. Consistent with an involvement of this kinase, fyn-deficient keratinocytes have strongly decreased tyrosine phosphorylation levels of β- and γ-catenins and p120-Cas, and structural and functional abnormalities in cell adhesion similar to those caused by tyrosine kinase inhibitors. Whereas skin of fyn−/− mice appears normal, skin of mice with a disruption in both the fyn and src genes shows intrinsically reduced tyrosine phosphorylation of β-catenin, strongly decreased p120-Cas levels, and important structural changes consistent with impaired keratinocyte cell adhesion. Thus, unlike what has been proposed for oncogene-transformed or mitogenically stimulated cells, in differentiating keratinocytes tyrosine phosphorylation plays a positive role in control of cell adhesion, and this regulatory function appears to be important both in vitro and in vivo.     

Urokinase-Type Plasminogen Activator Receptor Is Internalized by Different Mechanisms in Polarized and Nonpolarized Madin–Darby Canine Kidney Epithelial Cells

Accumulated data indicate that endocytosis of the glycosylphosphatidyl-inositol-anchored protein urokinase plasminogen activator receptor (uPAR) depends on binding of the ligand uPA:plasminogen activator inhibitor-1 (PAI-1) and subsequent interaction with internalization receptors of the low-density lipoprotein receptor family, which are internalized through clathrin-coated pits. This interaction is inhibited by receptor-associated protein (RAP). We show that uPAR with bound uPA:PAI-1 is capable of entering cells in a clathrin-independent process. First, HeLa^K44A cells expressing mutant dynamin efficiently internalized uPA:PAI-1 under conditions in which transferrin endocytosis was blocked. Second, in polarized Madin–Darby canine kidney (MDCK) cells, which expressed human uPAR apically, the low basal rate of uPAR ligand endocytosis, which could not be inhibited by RAP, was increased by forskolin or phorbol ester (phorbol 12-myristate 13-acetate), which selectively up-regulate clathrin-independent endocytosis from the apical domain of epithelial cells. Third, in subconfluent nonpolarized MDCK cells, endocytosis of uPA:PAI-1 was only decreased marginally by RAP. At the ultrastructural level uPAR was largely excluded from clathrin-coated pits in these cells and localized in invaginated caveolae only in the presence of cross-linking antibodies. Interestingly, a larger fraction of uPAR in nonpolarized relative to polarized MDCK cells was insoluble in Triton X-100 at 0°C, and by surface labeling with biotin we also show that internalized uPAR was mainly detergent insoluble, suggesting a correlation between association with detergent-resistant membrane microdomains and higher degree of clathrin-independent endocytosis. Furthermore, by cryoimmunogold labeling we show that 5–10% of internalized uPAR in nonpolarized, but not polarized, MDCK cells is targeted to lysosomes by a mechanism that is regulated by ligand occupancy.     

In Vitro Assay of Bacterial Adhesion onto Mammalian Epithelial Cells

To cause infections, bacteria must colonize their host. Bacterial pathogens express various molecules or structures able to promote attachment to host cells¹. These adhesins rely on interactions with host cell surface receptors or soluble proteins acting as a bridge between bacteria and host. Adhesion is a critical first step prior to invasion and/or secretion of toxins, thus it is a key event to be studied in bacterial pathogenesis. Furthermore, adhered bacteria often induce exquisitely fine-tuned cellular responses, the studies of which have given birth to the field of ''cellular microbiology''². Robust assays for bacterial adhesion on host cells and their invasion therefore play key roles in bacterial pathogenesis studies and have long been used in many pioneer laboratories^3,4. These assays are now practiced by most laboratories working on bacterial pathogenesis.Here, we describe a standard adherence assay illustrating the contribution of a specific adhesin. We use the Escherichia coli strain 2787⁵, a human pathogenic strain expressing the autotransporter Adhesin Involved in Diffuse Adherence (AIDA). As a control, we use a mutant strain lacking the aidA gene, 2787ΔaidA (F. Berthiaume and M. Mourez, unpublished), and a commercial laboratory strain of E. coli, C600 (New England Biolabs). The bacteria are left to adhere to the cells from the commonly used HEp-2 human epithelial cell line. This assay has been less extensively described before⁶.Download video file.^{(34M, mov)}     

Distinct Actions and Cooperative Roles of ROCK and mDia in Rho Small G Protein-induced Reorganization of the Actin Cytoskeleton in Madin–Darby Canine Kidney Cells

Rho, a member of the Rho small G protein family, regulates the formation of stress fibers and focal adhesions in various types of cultured cells. We investigated here the actions of ROCK and mDia, both of which have been identified to be putative downstream target molecules of Rho, in Madin-Darby canine kidney cells. The dominant active mutant of RhoA induced the formation of parallel stress fibers and focal adhesions, whereas the dominant active mutant of ROCK induced the formation of stellate stress fibers and focal adhesions, and the dominant active mutant of mDia induced the weak formation of parallel stress fibers without affecting the formation of focal adhesions. In the presence of C3 ADP-ribosyltransferase for Rho, the dominant active mutant of ROCK induced the formation of stellate stress fibers and focal adhesions, whereas the dominant active mutant of mDia induced only the diffuse localization of actin filaments. These results indicate that ROCK and mDia show distinct actions in reorganization of the actin cytoskeleton. The dominant negative mutant of either ROCK or mDia inhibited the formation of stress fibers and focal adhesions, indicating that both ROCK and mDia are necessary for the formation of stress fibers and focal adhesions. Moreover, inactivation and reactivation of both ROCK and mDia were necessary for the 12-O-tetradecanoylphorbol-13-acetate-induced disassembly and reassembly, respectively, of stress fibers and focal adhesions. The morphologies of stress fibers and focal adhesions in the cells expressing both the dominant active mutants of ROCK and mDia were not identical to those induced by the dominant active mutant of Rho. These results indicate that at least ROCK and mDia cooperatively act as downstream target molecules of Rho in the Rho-induced reorganization of the actin cytoskeleton.     

Ezrin Is an Effector of Hepatocyte Growth Factor–mediated Migration and Morphogenesis in Epithelial Cells

The dissociation, migration, and remodeling of epithelial monolayers induced by hepatocyte growth factor (HGF) entail modifications in cell adhesion and in the actin cytoskeleton through unknown mechanisms. Here we report that ezrin, a membrane–cytoskeleton linker, is crucial to HGF-mediated morphogenesis in a polarized kidney-derived epithelial cell line, LLC-PK1. Ezrin is a substrate for the tyrosine kinase HGF receptor both in vitro and in vivo. HGF stimulation causes enrichment of ezrin recovered in the detergent-insoluble cytoskeleton fraction. Overproduction of wild-type ezrin, by stable transfection in LLC-PK1 cells, enhances cell migration and tubulogenesis induced by HGF stimulation. Overproduction of a truncated variant of ezrin causes mislocalization of endogenous ezrin from microvilli into lateral surfaces. This is concomitant with altered cell shape, characterized by loss of microvilli and cell flattening. Moreover, the truncated variant of ezrin impairs the morphogenic and motogenic response to HGF, thus suggesting a dominant-negative mechanism of action. Site-directed mutagenesis of ezrin codons Y145 and Y353 to phenylalanine does not affect the localization of ezrin at microvilli, but perturbs the motogenic and morphogenic responses to HGF. These results provide evidence that ezrin displays activities that can control cell shape and signaling.     

“Harpoon” Model for Cell–Cell Adhesion and Recognition of Target Cells by the Natural Killer Cells

The mechanism of recognition by natural killer (NK) cells is still unknown. A dynamic model is formulated describing recognition or NK-sensitive target cells (TCs) by NK cells of NK-like cells. This model does not assume the presence of the specific NK-receptor(s) on the membrane of NK cells and corresponding specific ligands on the NK-sensitive TCs. We suggest: (1) the expression of various kinds of “non-NK receptors” and corresponding ligands (counter-receptors) on the plasma membrane of the same NK cell and, possibly, of TCs (e.g. LFA-1 and ICAM-1-ICAM3, CD2 and LFA-3; receptors for TNF and corresponding ligand etc.); (2) the presence of multiple disorders in the organization of “extracellular matrix-surface membrane-submembrane cytoskeleton” assembly of the NK-sensitive TCs; (3) non-specific primary linking of NK cell with TCs, which induces a transfer of vesicles or membrane fragments from the NK surface to the target cell surface (and perhaps vice versa). These processes may also permit the transfer of many types of receptor and counter-receptor molecules from the surface of one conjugated cell to another by vesicles or membrane fragments. After transferral through the intercellular cleft, the free receptors and counter-receptors will be localized on both cell surfaces at the contact region between conjugated cells. By this model the NK cell can “harpoon” the TC and enhance the binding forces between cells up to the critical level and then switch on killing mechanisms for the TC. By means of this “harpoon” model of cell recognition, it seems possible to explain the nature of the wide polymorphism of TCs which are sensitive to the effect of NK and NK-like cells. A mathematical model of the NK cell cytotoxic reaction is described. The model describes many nonlinear peculiarities of the cytotoxic process and predicts some new phenomena. We suggest new approaches of manipulation of cell membranes which can transform NK-resistant target cells in NK sensitive cells and vice versa.     

Expression of SLC2A9 Isoforms in the Kidney and Their Localization in Polarized Epithelial Cells

Background

Many genome-wide association studies pointed out that SLC2A9 gene, which encodes a voltage-driven urate transporter, SLC2A9/GLUT9 (a.k.a. URATv1), as one of the most influential genes for serum urate levels. SLC2A9 is reported to encode two splice variants: SLC2A9-S (512 amino acids) and SLC2A9-L (540 amino acids), only difference being at their N-termini. We investigated isoform-specific localization of SLC2A9 in the human kidney and role of N-terminal amino acids in differential sorting in vitro.

Methodology/Principal Findings

Isoform specific antibodies against SLC2A9 were developed and human kidney sections were stained. SLC2A9-S was expressed in the apical side of the collecting duct while SLC2A9-L was expressed in the basolateral side of the proximal tubule. GFP fused SLC2A9s were expressed in MDCK cells and intracellular localization was observed. SLC2A9-S was expressed at both apical and basolateral membranes, whereas SLC2A9-L was expressed only at the basolateral membrane. Although SLC2A9-L has a putative di-leucine motif at 33th and 34th leucine, deletion of the motif or replacement of leucine did not affect its subcellular localization. When up to 16 amino acids were removed from the N-terminal of SLC2A9-S or when up to 25 amino acids were removed from the N-terminal of SLC2A9-L, there was no change in their sorting. Deletion of 20 amino acids from SLC2A9-S was not expressed in the cell. More than 30 amino acids deletion from SLC2A9-L resulted in expression at both apical and basolateral membranes as well as in the lysosome. When amino acids from 25th and 30th were changed to alanine in SLC2A9-L, expression pattern was the same as wild-type.

Conclusions/Significance

SLC2A9-L was expressed in the basolateral membrane of kidney proximal tubules in humans and this isoform is likely to responsible for urate reabsorption. N-terminal amino acids unique to each isoform played an important role in protein stability and trafficking.     

Selective Perturbation of Apical Membrane Traffic by Expression of Influenza M2, an Acid-activated Ion Channel, in Polarized Madin–Darby Canine Kidney Cells

The function of acidification along the endocytic pathway is not well understood, in part because the perturbants used to modify compartmental pH have global effects and in some cases alter cytoplasmic pH. We have used a new approach to study the effect of pH perturbation on postendocytic traffic in polarized Madin–Darby canine kidney (MDCK) cells. Influenza M2 is a small membrane protein that functions as an acid-activated ion channel and can elevate the pH of the trans-Golgi network and endosomes. We used recombinant adenoviruses to express the M2 protein of influenza virus in polarized MDCK cells stably transfected with the polymeric immunoglobulin (Ig) receptor. Using indirect immunofluorescence and immunoelectron microscopy, M2 was found to be concentrated at the apical plasma membrane and in subapical vesicles; intracellular M2 colocalized partly with internalized IgA in apical recycling endosomes as well as with the trans-Golgi network marker TGN-38. Expression of M2 slowed the rate of IgA transcytosis across polarized MDCK monolayers. The delay in transport occurred after IgA reached the apical recycling endosome, consistent with the localization of intracellular M2. Apical recycling of IgA was also slowed in the presence of M2, whereas basolateral recycling of transferrin and degradation of IgA were unaffected. By contrast, ammonium chloride affected both apical IgA and basolateral transferrin release. Together, our data suggest that M2 expression selectively perturbs acidification in compartments involved in apical delivery without disrupting other postendocytic transport steps.     

Fine-Tuning of the Actin Cytoskeleton and Cell Adhesion During Drosophila Development by the Unconventional Guanine Nucleotide Exchange Factors Myoblast City and Sponge

The evolutionarily conserved Dock proteins function as unconventional guanine nucleotide exchange factors (GEFs). Upon binding to engulfment and cell motility (ELMO) proteins, Dock–ELMO complexes activate the Rho family of small GTPases to mediate a diverse array of biological processes, including cell motility, apoptotic cell clearance, and axon guidance. Overlapping expression patterns and functional redundancy among the 11 vertebrate Dock family members, which are subdivided into four families (Dock A, B, C, and D), complicate genetic analysis. In both vertebrate and invertebrate systems, the actin dynamics regulator, Rac, is the target GTPase of the Dock-A subfamily. However, it remains unclear whether Rac or Rap1 are the in vivo downstream GTPases of the Dock-B subfamily. Drosophila melanogaster is an excellent genetic model organism for understanding Dock protein function as its genome encodes one ortholog per subfamily: Myoblast city (Mbc; Dock A) and Sponge (Spg; Dock B). Here we show that the roles of Spg and Mbc are not redundant in the Drosophila somatic muscle or the dorsal vessel. Moreover, we confirm the in vivo role of Mbc upstream of Rac and provide evidence that Spg functions in concert with Rap1, possibly to regulate aspects of cell adhesion. Together these data show that Mbc and Spg can have different downstream GTPase targets. Our findings predict that the ability to regulate downstream GTPases is dependent on cellular context and allows for the fine-tuning of actin cytoskeletal or cell adhesion events in biological processes that undergo cell morphogenesis.     

Neuronal Pathfinding and Recognition: Roles of Cell Adhesion Molecules

Neurons contact their neighbors through a diverse array of cell adhesion and other surface molecules. These molecules can exhibit highly regulated patterns of expression, underscoring their multiple roles in establishing specific interactions between neurons and their environment. Recent studies are beginning to ask how these membrane-bound neural recognition molecules interact with each other and intracellular signaling pathways within an individual neuronal growth cone, and direct the formation of neural connections during development.     

Direct Ca2+-dependent Heterophilic Interaction between Desmosomal Cadherins, Desmoglein and Desmocollin, Contributes to Cell–Cell Adhesion

Human fibrosarcoma cells, HT-1080, feature extensive adherens junctions, lack mature desmosomes, and express a single known desmosomal protein, Desmoglein 2 (Dsg2). Transfection of these cells with bovine Desmocollin 1a (Dsc1a) caused dramatic changes in the subcellular distribution of endogenous Dsg2. Both cadherins clustered in the areas of the adherens junctions, whereas only a minor portion of Dsg2 was seen in these areas in the parental cells. Deletion mapping showed that intact extracellular cadherin-like repeats of Dsc1a (Arg¹-Thr¹⁷⁰) are required for the translocation of Dsg2. Deletion of the intracellular C-domain that mediates the interaction of Dsc1a with plakoglobin, or the CSI region that is involved in the binding to desmoplakin, had no effect. Coimmunoprecipitation experiments of cell lysates stably expressing Dsc1a with anti-Dsc or -Dsg antibodies demonstrate that the desmosomal cadherins, Dsg2 and Dsc1a, are involved in a direct Ca²⁺-dependent interaction. This conclusion was further supported by the results of solid phase binding experiments. These showed that the Dsc1a fragment containing cadherin-like repeats 1 and 2 binds directly to the extracellular portion of Dsg in a Ca²⁺-dependent manner. The contribution of the Dsg/ Dsc interaction to cell–cell adhesion was tested by coculturing HT-1080 cells expressing Dsc1a with HT-1080 cells lacking Dsc but expressing myc-tagged plakoglobin (MPg). In the latter cells, MPg and the endogenous Dsg form stable complexes. The observed specific coimmunoprecipitation of MPg by anti-Dsc antibodies in coculture indicates that an intercellular interaction between Dsc1 and Dsg is involved in cell–cell adhesion.     

Rotavirus Infection Reduces Sucrase-Isomaltase Expression in Human Intestinal Epithelial Cells by Perturbing Protein Targeting and Organization of Microvillar Cytoskeleton

Rotavirus infection is the most common cause of severe infantile gastroenteritis worldwide. These viruses infect mature enterocytes of the small intestine and cause structural and functional damage, including a reduction in disaccharidase activity. It was previously hypothesized that reduced disaccharidase activity resulted from the destruction of rotavirus-infected enterocytes at the villus tips. However, this pathophysiological model cannot explain situations in which low disaccharidase activity is observed when rotavirus-infected intestine exhibits few, if any, histopathologic changes. In a previous study, we demonstrated that the simian rotavirus strain RRV replicated in and was released from human enterocyte-like Caco-2 cells without cell destruction (N. Jourdan, M. Maurice, D. Delautier, A. M. Quero, A. L. Servin, and G. Trugnan, J. Virol. 71:8268–8278, 1997). In the present study, to reinvestigate disaccharidase expression during rotavirus infection, we studied sucrase-isomaltase (SI) in RRV-infected Caco-2 cells. We showed that SI activity and apical expression were specifically and selectively decreased by RRV infection without apparent cell destruction. Using pulse-chase experiments and cell surface biotinylation, we demonstrated that RRV infection did not affect SI biosynthesis, maturation, or stability but induced the blockade of SI transport to the brush border. Using confocal laser scanning microscopy, we showed that RRV infection induces important alterations of the cytoskeleton that correlate with decreased SI apical surface expression. These results lead us to propose an alternate model to explain the pathophysiology associated with rotavirus infection.     

PDGF, TGF-β, and Heterotypic Cell–Cell Interactions Mediate Endothelial Cell–induced Recruitment of 10T1/2 Cells and Their Differentiation to a Smooth Muscle Fate

We aimed to determine if and how endothelial cells (EC) recruit precursors of smooth muscle cells and pericytes and induce their differentiation during vessel formation. Multipotent embryonic 10T1/2 cells were used as presumptive mural cell precursors. In an under-agarose coculture, EC induced migration of 10T1/2 cells via platelet-derived growth factor BB. 10T1/2 cells in coculture with EC changed from polygonal to spindle-shaped, reminiscent of smooth muscle cells in culture. Immunohistochemical and Western blot analyses were used to examine the expression of smooth muscle (SM)-specific markers in 10T1/2 cells cultured in the absence and presence of EC. SM-myosin, SM22α, and calponin proteins were undetectable in 10T1/2 cells cultured alone; however, expression of all three SM-specific proteins was significantly induced in 10T1/2 cells cocultured with EC. Treatment of 10T1/2 cells with TGF-β induced phenotypic changes and changes in SM markers similar to those seen in the cocultures. Neutralization of TGF-β in the cocultures blocked expression of the SM markers and the shape change. To assess the ability of 10T1/2 cells to contribute to the developing vessel wall in vivo, prelabeled 10T1/2 cells were grown in a collagen matrix and implanted subcutaneously into mice. The fluorescently marked cells became incorporated into the medial layer of developing vessels where they expressed SM markers. These in vitro and in vivo observations shed light on the cell–cell interactions that occur during vessel development, as well as in pathologies in which developmental processes are recapitulated.     

Acute loss of Cell–Cell Communication Caused by G Protein–coupled Receptors: A Critical Role for c-Src

Gap junctions mediate cell–cell communication in almost all tissues, but little is known about their regulation by physiological stimuli. Using a novel single-electrode technique, together with dye coupling studies, we show that in cells expressing gap junction protein connexin43, cell–cell communication is rapidly disrupted by G protein–coupled receptor agonists, notably lysophosphatidic acid, thrombin, and neuropeptides. In the continuous presence of agonist, junctional communication fully recovers within 1–2 h of receptor stimulation. In contrast, a desensitization-defective G protein–coupled receptor mediates prolonged uncoupling, indicating that recovery of communication is controlled, at least in part, by receptor desensitization. Agonist-induced gap junction closure consistently follows inositol lipid breakdown and membrane depolarization and coincides with Rho-mediated cytoskeletal remodeling. However, we find that gap junction closure is independent of Ca²⁺, protein kinase C, mitogen-activated protein kinase, or membrane potential, and requires neither Rho nor Ras activation. Gap junction closure is prevented by tyrphostins, by dominant-negative c-Src, and in Src-deficient cells. Thus, G protein–coupled receptors use a Src tyrosine kinase pathway to transiently inhibit connexin43-based cell–cell communication.