University of Groningen Regulation of polarity development in hepatocytes van der Wouden, Johanna IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2004 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Wouden, J. M. V. D. (2004). Regulation of polarity development in hepatocytes: a novel role for oncostatin M Groningen: s.n. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date: 18-06-2017 CHAPTER 1 Membrane dynamics and the regulation of epithelial cell polarity Johanna M. van der Wouden, Olaf Maier, Sven C.D. van IJzendoorn, and Dick Hoekstra Department of Membrane Cell Biology, Faculty of Medical Sciences, University of Groningen, Groningen, the Netherlands International Reviews of Cytology, 226:127-164 (2003) Chapter 1 Abstract Plasma membranes of epithelial cells consist of two domains, an apical and a basolateral domain, the surfaces of which differ in composition. The separation of these domains by a tight junction and the fact that specific transport pathways exist for intracellular communication between these domains and distinct intracellular compartments relevant to cell polarity development, has triggered extensive research on issues that focus on how polarity is generated and maintained. Apart from proper assembly of tight junctions, their potential functioning as landmark for the transport machinery, cell-cell adhesion is obviously instrumental in barrier formation. In recent years, distinct endocytic compartments, defined as subapical compartment or common endosome were shown to play a prominent role in regulating membrane trafficking to and from the polarized membrane domains. Sorting devices remain to be determined but likely include distinct rab proteins, and evidence is accumulating to indicate that signaling events may direct intracellular membrane transport, intimately involved in the biogenesis and maintenance of polarized membrane domains and hence the development of cell polarity. 8 Chapter 1 I. Introduction Epithelial cells form a barrier between distinct extracellular environments, e.g. the blood and underlying tissues on the one hand and a lumen on the other. For biogenesis and maintenance of such a barrier, epithelial cells must regulate their cell shape and internal processes. A major feature of the epithelial cell is that it exhibits a plasma membrane polarity, i.e. the plasma membrane has two domains, an apical and basolateral domain, which are separated by a tight junction. The basolateral domain is faces the blood and underlying tissue and the microvilli-covered apical domain is facing the lumen (e.g. the lumen of the intestine, bile canaliculi and kidney tubules; figure 1). Figure 1: Schematic representation of a polarized Hepatocyte. The apcical bile canalicular membrane (BC) is separated by tight junctions (TJ) from the basolateral membrane which is facing the blood and underlying tissues. Because the molecular composition of either domain differs, consistent with differences in functional properties and likely, mechanical stability, intracellular transport from and towards the different membrane domains must be highly regulated to secure that difference. Indeed, for that purpose several transport pathways and sorting stations are operational in epithelial cells, which are discussed in this chapter. Other important parts of polarity biogenesis and maintenance are the extracellular 9 Chapter 1 environment and cell attachments, which are closely related to the cytoskeleton. The cytoskeleton needs to be properly organized to provide epithelial cell shape and to constitute tracks for vesicle movement. Presumably, this complex composition of epithelial morphology is highly regulated by signalling cascades to provide a proper barrier under each condition. Disruption of the barrier function, due to malfunctioning or disregulation of signalling cascades results often in carcinogenesis. On the other hand, stimulation of fetal hepatocytes or metanephros with interleukins such as Oncostatin M or Leukemia Inhibitory Factor, respectively, result in epithelialization. In this chapter, membrane polarity will be discussed with a focus on the role of the cytoskeleton and tight junctions, signalling and intracellular traffic in the development and regulation of epithelial cell polarity. II. Structural organization of epithelial cell polarity Because epithelial cells constitute a boundary between tissues and extracellular interstitium, their plasma membranes face simultaneously different environments, it is obvious that distinct functions require a concentration at their site of action. To meet such a requirement, appropriate barrier and transport functions of epithelia depend on the socalled polarized distribution of proteins and lipids in the plasma membrane. To maintain such a polarized distribution, tight junctions separate these two membrane domains. Overall cell architecture and its organization represent fundamental aspects in the maintenance and development of plasma membrane polarity. However, plasma membrane polarity is often described in terms of specific intracellular transport processes, which deliver and retrieve compounds so that the specificity of either domain is generated and/or maintained. 10 Chapter 1 A. Cell-cell contacts For epithelial cells to constitute a barrier, cells need to form cell-cell and cell-substratum contacts. Important players in the formation and maintenance of cell polarity are the cadherins, the adhesion receptors for cell-cell adhesion, which assemble the adherens junction, present on lateral membrane domains. They comprise a family of calcium-dependent cell adhesion molecules. Cadherins not only function in cell-cell contact but are also involved in the programming of cell differentiation and polarization. For example, E-cadherin plays an important role in maintaining the normal phenotype of epithelial cells, while N-cadherin is needed to prevent epithelial to mesenchym conversion, as frequently occurs in tumors with an inappropriate expression of N-cadherin (Kim et al., 2000). The extracellular domain of E-cadherin is responsible for calcium-dependent homophilic interactions with adjacent cells and the cytoplasmic domain associates with proteins that belong to the catenin family, i.e., α- catenin, β-catenin and γ-catenin (plakoglobin) are all known to associate directly with the E-cadherin (Nathke et al., 1994). In fact, catenins anchor E-cadherin to the cytoskeleton thereby allowing the cells to engage in strong cell-cell interactions. Abnormal expression of cadherins, such as in human carcinomas, correlates with a dedifferentiated epithelial phenotype, invasion and metastasis. More specifically, MDCK cells in which a mutated E-cadherin is overexpressed display a reduced expression of endogenous cadherin, which results in a decreased cell adhesion and they are more easily dissociated from the cyst (Troxell et al., 2001). Thus, a loss of cadherin does not induce transformation, but it promotes the development of an invasive, non-polarized phenotype, often observed in carcinogenesis. Further details concerning the association of E-cadherin with the cytoskeleton will be discussed below. 11 Chapter 1 B. Tight junctional complex The basolateral and apical plasma membrane domains are separated by a tight junction (TJ), which is in fact a multicomponent protein complex. TJs not only create a primary barrier to prevent paracellular transport of solutes, but they also restrict the lateral diffusion of apical and basolateral proteins and lipids in the outer leaflet of the plasma membrane. Inner leaflet lipids appear to be capable of randomizing over the entire inner leaflet, suggesting that an effective barrier exists for the outer leaflet only. Occludin, an integral transmembrane protein, was the first protein identified in the TJ, where it functions as the actual seal of the junction. To localize to and function at the TJ, the protein requires to be phosphorylated (Wong, 1997). Detailed insight on the assembly of occludin into the TJ complex has been obtained from a wound healingmodel, which relies on the establishment of cell polarity in mouse epithelia. In this system, Ando-Akatsuka et al. (1999) showed that occludin first associates with E-cadherin and ZO-1, another TJ protein, at cell-cell contacts. Subsequently, occludin and ZO-1 separate from Ecadherin; the latter forms the adherens junction whereas ZO-1 and occludin localize at the TJ, occludin being in its tyrosine phosphorylated state, as a result of phosphorylation by nonreceptor tyrosine kinase c-Yes (Chen et al., 2002). ZO-1 is a member of the membrane-associated guanylate kinase-like homologues (MAGUK) family, which includes ZO-2 and ZO-3, both of which also localize to the TJ. MAGUKs contain several PDZ domains, one SH3 domain and a guanylate kinase like domain. Not only occludin binds to ZOs, but the actin cytoskeleton (see below) and proteins belonging to the claudin-family also bind (via their C-terminus) to ZO-1/ZO-2/ZO-3 (Itoh, et al., 1999). Partners for the SH3 domain of ZO-1 are ZONAB-A and -B as Y-box transcription factors, which are located at the tight junction (Balda and Matter, 2000). ZONAB per se is also present in the nucleus where it regulates gene expression of ErbB2 and p27kip1, which is involved in cell cycle regulation 12 and most likely in intestinal Chapter 1 differentiation (Tian, et al., 1999a,b). ErbB-2 is involved in alveolar branching and morphogenesis of the mammary gland (Niemann et al., 1998). How these functions correlate with those expressed by these proteins in the tight junction assembly is as yet undetermined. Given their specific localization at the boundary between the apical and basolateral domains, the TJ could represent a landmark for the transport machinery, acting in targeting vesicles to the correct plasma membrane domain, thereby facilitating formation and maintenance of the polarized epithelial phenotype (see below). In this context it is interesting to note that TJs indeed form a microdomain in the plasma membrane, as hyperphosphorylated occludin and ZO-1 are both enriched in the Triton X-100-insoluble fractions, isolated from T84 epithelial cells (Nusrat et al., 2000), suggesting an association with a detergent resistant microdomain (DRM). Furthermore, the small GTPase rab 13, belonging to the family of proteins thought to regulate vesicle docking and fusion, is recruited from a cytosolic pool to the junctional complexes after cell-cell contacts have been established. Overexpression of dominant active rab13 (Q67L) causes a delay in the recruitment of claudin-1 to the junctional complex, which presumably explains the delay in the formation of a sealed monolayer of polarized MDCK cells, and an enhanced leakage of a small nonionic tracer from the apical domain (Marzesco et al., 2002). These data suggest that rab13 functions as a regulator in TJ assembly, and hence as a potential parameter in governing development and maintenance of polarity. A role of rab13 in vesicular transport has thus far not been described, but rab13, in conjunction with rab8, is closely related to yeast small GTPase sec4, which is involved in controlling the assembly of the exocyst (Guo, et al., 1999), required for the polarized delivery of cargo proteins to the plasma membrane in budding yeast. 13 Chapter 1 C. Role of the extracellular matrix and the cytoskeleton Cell adhesion to the extracellular matrix (ECM) is important for the determination of the axis of polarity, i.e., in defining the orientation of the apical and basolateral membrane domains relative to the biological environment and the direction of transport of solutes. As previously mentioned, an important indicator for the axis of polarity is the TJ and cell-cell contacts, but it is the cytoskeleton which provides the cell's epithelial-like shape. We will first discuss some aspects of the involvement of the ECM in the development of an apical plasma membrane domain. In early studies it could already be demonstrated that surface adhesion of MDCK cells suffices to target a 184 kDa apical resident protein to the non-contacting, apical domain of the cells. By contrast, at this early phase of polarity development, a 63 kDa basolateral protein marker randomly distributes over the entire surface of the cell (Vega-Salas et al., 1987). In this study, the medium contained a relatively low calcium concentration, which prevents the formation of cell-cell contacts and TJs. Accordingly, the presence of TJs and the occurrence of cell-cell adhesion is apparently not needed to induce recruitment of apical resident proteins at distinct membrane domains, seemingly reflecting an initiation of polarity. Evidently, full development of polarity requires cell-cell contacts and the formation of TJs. Following cell attachment, MDCK cells initially organize their microtubules radiating from an intracellular region where also the centrosomes and Golgi complex are located. In the fully polarized state, the microtubules are organized as an apical web, and are longitudinally arranged in microtubule bundles in the apical-basal axis with the minus end spread over the apical region and the plus ends directed towards the basolateral region (Bacallao et al., 1989). Such an organization of the microtubule network has been thought to be specific for polarized epithelial cells, as it would strongly facilitate vesicle transport to the correct plasma membrane domain. However, in nonpolarized fibroblasts, transport of apical and basolateral proteins in 14 Chapter 1 different vesicles along microtubules occur in a similar manner. Thus, it is more likely that microtubules in addition to regulating overall vesicular transport in general, are also involved in functions as maintaining cell shape. Apart from a microtubule network, epithelial cells also contain a highly ordered actin network. In epithelial cells, a dense actin network is found just underneath the apical surface. The actin network is linked to actin filaments, which constitute the core of the microvilli, covering the apical surface. Fath and Burgess (1993, 1994) proposed that cytoplasmic dynein moves Golgi-derived vesicles along microtubules to the cell cortex, where the actin-based motor protein myosin-I then mediates the local delivery to the apical membrane across the dense actin network. Not only apical delivery depends on actin. When actin filaments in MDCK cells are depolymerized, following treatment with latrunculin B, both the apical construct FcLR(5-22) and the basolaterally localized transferrin receptor (TfR) become redistributed over both membrane domains, without perturbation of the tight junctional complex (Sheff, el al., 2002). This suggests that the integrity of actin filaments is also crucial in regulating in a direct or indirect manner protein-sorting events, for example by orchestrating protein retention in specific domains. In this context, the involvement of additional regulators, connecting the actin cytoskeleton with cell polarity, is of relevance, such as the Rho family GTPases, including RhoA, Rac1 and cdc42. Mutations in these small GTPases cause a loss or misorientation of polarity and sorting of proteins. Expression of dominant-active cdc42 in MDCK cells slowed endocytic and biosynthetic transport, whereas the dominant-negative mutant slowed apical endocytosis and basolateral-to-apical transcytosis, but stimulated biosynthetic traffic (Rojas et al., 2001). Both mutants cause altered TJ morphology and function, as measured by a decrease in transepithelial resistance. A downstream effector of Rho is Rho kinase (ROCK). Inhibition of ROCK by a specific inhibitor, Y-27632, resulted in a reorganization of F-actin structures and enhances paracellular permeability, but it did not alter the distribution and detergent solubility 15 Chapter 1 of TJ-proteins (Walsh et al., 2001). Therefore it is suggested that ROCK regulates intact TJ via its effects on the F-actin cytoskeleton. Effects on cell-cell adhesion by the Rho family of GTPases has been shown in keratinocytes, where Rac1 regulates adherens junctions via endocytosis of E-cadherin (Akhtar, 2001) possibly regulated by Protein kinase C activity (Le et al., 2002) and effects on F-actin. In conclusion, the data suggest that regulation of actin organization by Rho GTPases is involved in the proper functioning of cell adhesion and cell-cell contacts, and thereby plays an important role in the development and maintenance of epithelial cell polarity. III. Regulation of cell polarity A. Interleukins and epithelial cell differentiation Several cytokines and hormones are known to be involved in epithelial differentiation, in particular in liver development and regeneration. An important group of cytokines in hepatic differentiation and regeneration are the interleukin (IL)-6 type of cytokines. This family consists of IL-6, IL-11, CNTF, CT-1, LIF and OSM, and these cytokines all share the common signal transducer gp130 in combination with cytokine-specific subunits. This shared use of gp130 presumably explains their almost identical biological response. Fetal murine hepatocytes differentiate upon stimulation with OSM in combination with glucocorticoid, and as a result several mature hepatic markers become expressed, such as albumin secretion (Kamiya et al, 1999). Furthermore, livers deficient in gp130 display defects in liver maturation. Not only does liver development depend on an IL-6 type cytokine action, but mesenchymal-to-epithelial conversion of rat metanephros is also induced by an IL-6 type of cytokine, LIF, and following OSM treatment the same epithelialization was observed (Barasch, et al., 1999). The classical signal transduction via gp130 involves the JAK/STAT pathway, but also Ras, PI3K and MAPK can also be activated by OSM via a gp130-dependent mechanism (Heinrich, et 16 Chapter 1 al.1998, and Taga and Kishimoto, 1997). In the murine fetal hepatocyte system, overexpression of dominant negative STAT3 resulted in suppression of the differentiated phenotypes obtained upon OSM stimulation (Ito et al, 2000). This indicates that the STAT3 pathway is an essential component of OSM induced hepatic differentiation. On the molecular level, further insight into downstream events of OSM-mediated signalling is gradually emerging. OSM-induced formation of E-cadherinbased adherens junctions (AJ), which are fundamental for proper organization of epithelial cells, including hepatocytes, is dependent on direct K-Ras activation (Matsui et al., 2002). This indicates that OSM can regulate cell adhesion in fetal hepatocytes, thereby functioning in the organization of the cellular architecture. However the tight junctional organization, when monitored by ZO-1 expression and localization, is not affected by OSM. The IL-6 type of cytokines are not only involved in epithelialization of fetal hepatocytes. Recently, evidence was obtained that these cytokines can also specifically induce a plasma membrane polarization in primary cell systems. This issue was investigated in cultured hepatocytes, which develop a bile canalicular (BC) structure, like HepG2 hepatoma cells, a system that has been shown to represent a valuable tool to study development of plasma membrane polarity (van IJzendoorn et al., 2000; Maier et al., 2001; van der Wouden et al., 2002). Thus, van der Wouden et al. showed that OSM can stimulate plasma membrane development via a gp130-dependent mechanism. Furthermore, van IJzendoorn and Hoekstra (2000) provided evidence that a distinct apically-directed vesicular transport pathway is specifically activated by an intrinsic protein kinase A-mediated mechanism during polarity development of HepG2 cells (see below). An inhibition of this pathway by protein kinase A (PKA) inhibitors abolishes polarity development. Similarly, stimulation of HepG2 cells with OSM or IL-6 results in a more polarized phenotype of HepG2 cells (see fig. 2) and, interestingly, this hyperpolarization is a consequence of stimulation of an apically directed transport pathway, 17 Chapter 1 B. Cell cycle regulation and cell polarity During development, coordinately regulated cell division events are crucial for proper tissue morphogenesis. Other key features of morphogenesis are cellular polarization and membrane dynamics. These two intimately entwined features are well illustrated in polarized epithelia in which, as noted above, the plasma membrane is segregated into different domains with distinct protein and lipid compositions, and intracellular organelles and structures such as the centrosome are positioned in a polarized manner. In addition to the well-defined function of TJs in maintaining a fence between apical and basolateral membrane domains, recent evidence suggests that TJs are also involved in basic cellular processes such as the regulation of cell growth and differentiation (Balda and Matter, 1998; 2000), as well as polarized processes such as the orientation of the division apparatus (Lecuit and Wieschaus, 2002). In cells that are stimulated to proliferate, e.g. during the process of wound healing, the localization of the prominent TJ protein zonula occludens (ZO-1) is shifted to the nucleus (Gottardi et al., 1996). Furthermore, ZO-1, by virtue of its interaction with ZONAB, directly participates in the control of the expression of the ErbB-2 gene (Balda and Matter, 2000), which is implicated in cell differentiation and proliferation. ZO-1 displays homology to Discs large, a Drosophila tumor suppressor (Bilder et al., 2000). In Drosophila epithelial cells, Discs large, together with two other proteins, lethal giant larvae and scribble, act to properly localize apical proteins and adherens junctions and as such contribute to epithelial architecture. Studies with mutants of these proteins suggest that in epithelial-derived cancers, the breakdown of the apical-basolateral architecture may have been the primary defect, resulting in a loss of differentiation, an increased proliferation rate (Bilder et al., 2000), and possibly a disorganized tissue architecture. The establishment and maintenance of epithelial cell polarity may thus be functionally related to cell cycle control. For instance, circulating growth factors, their receptors 19 Chapter 1 typically displaying a polarized distribution in epithelial cells, may elicit different signaling cascades that lead to changes in cell cycle control when no longer restricted to a specific surface or membrane domain. Many signaling events are organized in specialized PM microdomains, called caveolae (Fielding, 2001). The main structural component of caveolae, caveolin-1, behaves as a tumor suppressor. Thus, downregulation of caveolin-1 is sufficient to trigger cell transformation (Galbiati et al., 1998) and its (over)expression negatively regulates cell cycle progression (Galbiati et al., 2001). Accordingly, caveolin-1-deficient mice display increased cell numbers and a disorganized cell architecture (Drab et al., 2001). The molecular mechanism by which caveolin-1 controls cell growth and architecture is not clear, but may involve a p53/p21(WAF1/CIP1)-dependent pathway (Galbiati et al., 2001) and/or caveolin-1-mediated regulation of protein kinase A signaling (Razani et al., 1999). In addition to its involvement in signaling, caveolin-1 also plays a role in the delivery of certain newly synthesized proteins, many of them being sequestered in detergent-insoluble sphingolipid- and cholesterol-enriched PM microdomains (‘rafts’; Simons and Ikonen, 1997), to the apical PM (Scheiffele et al., 1998). As noted in section II B, the TJ area appears to be a targeting patch for polarized traffic pathways (Louvard et al., 1980; Zahraoui et al., 2000). Moreover, the TJ area displays raft-like characteristics (Nusrat et al., 2000) and harbors many signaling complexes (Zahraoui et al., 2000) and integrins. Proteins that are known to direct polarized trafficking, for example, the exocyst members sec6/sec8 (Grindstaff et al., 1998), cdc42 (Kroschewski et al., 1999; Cohen et al., 2001; Zhang et al., 2001; Rojas et al., 2001) and the mammalian homologue of Drosophila lethal giant larvae (Müsch et al., 2002), localize at the side of cell-cell contact in close proximity to the TJ area. The intracellular sorting and trafficking of proteins and lipids are fundamental to the establishment and maintenance of cell polarity. Indeed, these processes ensure the selective supply of molecules to the different PM domains and control the (intra)cellular localization of signaling molecules activated at the cell 20 Chapter 1 surface, and thereby the ultimate biological response. Well-known sorting sites are the Golgi apparatus and PM. Recently, several elements of the endosomal system have been identified that control the polarized sorting and targeting of both proteins and lipids, and include the early sorting endosome and the subapical compartment (SAC) (reviewed in van IJzendoorn and Hoekstra, 1999, van IJzendoorn et al., 2000). The latter compartment is rich in caveolin-1 and other raft components such as cholesterol and sphingolipids (Gagescu et al., 2000; Hoekstra and van IJzendoorn, 2000). Early studies revealed that membrane traffic ceases during mitosis, which suggests a functional relationship between the regulation of cell cycle progression and intracellular transport. The cessation of intracellular transport during the mitotic phase was reported not to perturb the existing epithelial architecture and polarity (Reinsch and Karsenti, 1994). However, polarity is transiently rearranged without loss of cell-cell contact during growth factor-induced tubulogenesis (Pollack et al., 1998). It is not clear whether epithelial polarity can be directly influenced by changes in cell cycle progression or by an altered expression pattern of cell cycle-regulating molecules. However, terminal differentiation of epithelial cells and the establishment of polarity are often associated with withdrawal from the cell cycle following cell cycle arrest in the G1 phase. Recent data suggest that proteins that are primarily implicated in cell cycle control may play additional functions in epithelial cell morphogenesis and polarity. For instance, the cyclin-dependent kinase inhibitor p27kip1, which is involved in mediating G1 arrest, was found to be required for mouse mammary gland morphogenesis and function (Muraoka et al., 2001), and was also implicated in the differentiation of kidney and intestinal epithelium (Combs et al., 1998; Deschenes et al., 2001; Tian and Quaroni, 1999b). Furthermore, the overexpression or activation of p53, a tumor suppressor that regulates cell cycle progression, was reported to inhibit cdc42-dependent cell effects that promote cell polarity, including actin cytoskeletal dynamics (Gadea et al., 2002) and, possibly, polarized membrane transport (Kroschewski et al., 21 Chapter 1 1999; Cohen et al., 2001; Zhang et al., 2001). Recent data indicate that proteins involved in the regulation of the cell cycle may modulate changes in the organization and function of the endomembranous system. Indeed, several cyclins, e.g. cyclin A (S-phase), cyclin B (G2/mitosis) and cyclin E (G1 phase), cdc2 and the cyclin-dependent kinase (cdk) 2 have been found in isolated endosomal fractions from liver (Vergés et al., 1997). Cyclin B has also been detected in the Golgi apparatus (Jackman et al., 1995) and both cdk2 (Gaulin et al., 2000) and the cdk2-inhibitor p27kip1 (Yaroslavskiy et al., 2001) have been found in PM fractions. Moreover, the amount and activity of cell cycle regulatory proteins in endosome and PM fractions is dynamically and specifically regulated, e.g. during liver regeneration, a process that involves a highly increased proliferation rate of hepatocytes (Vergés et al. 1997), following insulin stimulation (Gaulin et al., 2000), or upon cellular activation (Yaroslavskiy et al., 2001). The endosomal targets of cdk2 kinase activity are not known but are likely governed by the cyclin type that is complexed with cdk2. Cdk2 complexes integrate tyrosine phosphorylation and dephosphorylation signals, and in this way may control the number of rounds of membrane fusion at discrete domains. This leads to changes in the intracellular location of internalized receptors and, ultimately, in their biological response (Fiset and Faure, 2001). A target of the mitotic cyclin B-complexed cdk2 is the monomeric small GTPase rab4, which in epithelial cells regulates endosomal membrane trafficking and transport to the apical PM domain (Mohrmann et al., 2002). Phosphorylation of rab4 in mitotic cells by cyclin B/cdk2 leads to its relocalization from endosomal membranes to the cytosol (van der Sluijs et al., 1992; Ayad et al., 1997). Phosphorylation of rab4 by cyclin B/cdk2 inhibits both the recruitment of rab4 effector proteins to endosomes and the docking of rab4-containing transport vesicles. Thus in this way endocytic membrane fusion and transport are affected during the G2 and mitotic phase. The presence of active pools of cdk2, complexed with the G1-specific cyclin E, at the plasma membrane and in endosomes of polarized hepatocytes (Gaulin et al., 2000) strongly suggests that a functional relation between 22 Chapter 1 Figure 2: Oncostatin M treatment causes an increase in the number and membrane surface area of bile canaliculi in HepG2 cells. Hep2 cells were cultured for 72 hrs in the absence (A) or presence (B) of rhOSM (10 ng/ml). The cells were fixed in ethanol and stained with TRITC-labeled Phalloidin to visualize actin, which is abundantly present underneath the apical, bile canalicular surface (van der Wouden et al., 2002). Arrows indicate BCs and the bar is 5 µm. marked by sphingomyelin, from the subapical compartment SAC to the bile canalicular membrane. In epithelial development, particularly with respect to the development of plasma membrane domain polarity, IL-6 type cytokines have thus far been studied most extensively. Nevertheless, other interleukins could also interfere in correct maintenance and regeneration of polarity after tissue damage. For example, IL-15 is upregulated during inflammatory bowel disease as well as during colitis (Liu et al., 2000), cases in which the intestinal barrier function can be disrupted. Under these conditions, IL15 likely contributes to the reassembly of the tight junctional complex, separating the basolateral domain from the apical domain, by upregulation and recruitment of claudins, occludin, ZO-1 and ZO-2 into the complex (Nishiyama et al., 2001). Although knowledge in this area is still scanty, we anticipate that other cytokines will be identified that may, in addition, control polarity during development, inflammation or cancer, the latter also often being accompanied by a loss of cell polarity. 18 Chapter 1 cell cycle control and membrane dynamics may not be restricted to mitosis. Several factors that stimulate cell polarization also cause cells to arrest in the G1 phase of the cell cycle, and modulate membrane traffic events. As discussed above, the IL-6 cytokine family member OSM is required for fetal liver maturation and, in mature hepatocytes, stimulates polarity development (van der Wouden et al., 2002). OSM also induces G1 arrest via p27kip1-mediated inhibition of cdk2 activity (Klausen et al., 2000). In polarized hepatocytes, OSM-elicited signaling targets PKA-dependent membrane trafficking events at the SAC (van der Wouden et al., 2002) that have previously been demonstrated crucial for polarity development (van IJzendoorn and Hoekstra, 1999; 2000). Direct stimulation of PKA activity with cAMP analogs enhances the development of polarity in hepatic cells (Zegers and Hoekstra, 1998; van IJzendoorn and Hoekstra, 2000). The cAMP-mediated PKA activation pathway also increases the expression of the cdk2-inhibitor p27Kip1 by interfering with its degradation, in this manner controlling cell cycle progression (van Oirschot et al., 2001). Functional molecular links that integrate these events in a mechanism(s) that regulates the process of polarity development are as yet unknown, but are likely to be revealed in the near future. IV. Membrane transport in epithelial cells A. Polarized sorting in epithelial cells As described above a major feature of polarized epithelial cells is the formation of distinct PM domains, the apical and basolateral membrane, which are characterized by distinct protein and lipid compositions. This implies that sites exist in the cell where membrane components are sorted from each other and from where they are directed specifically to the different membrane domains. Although the polarized distribution of PM proteins and lipids can be generated during their biosynthesis, it is obvious that due to continuous internalization from and recycling to the 23 Chapter 1 PM, efficient sorting in the endosomal compartment(s) is required to maintain the polarized phenotype of epithelial cells. Accordingly, two major compartments have been implicated in polarized sorting: the transGolgi network (TGN) in the biosynthetic transport pathway and an endosomal compartment that has been referred to as an apical recycling endosome (ARE) or common endosome (CE) in epithelial cells such as MDCK, and as the subapical compartment (SAC) in liver cells (Traub, 1997; van IJzendoorn et al., 1999; 2000). The extent of sorting in the TGN is dependent on the cell type. Although in MDCK cells most apical and basolateral proteins are directly sorted from the TGN to the different domains (Ikonen and Simons, 1998), hepatocytes transport most (Bastaki et al., 2002), but not all (Kipp and Arias, 2000), apical proteins first to the basolateral membrane, where they are internalized and transported through the endosomal membrane system to the apical membrane, a process called transcytosis. This transcytotic pathway is also used for the transport of molecules across the epithelial cell layer, as, for example in case of transport of dimeric IgA after its binding to the polymeric Ig-receptor (pIgR) at the basolateral membrane of epithelial cells (Apodaca et al. 1991). The differences in the sorting capacities of biosynthetic and endocytotic pathways in MDCK cells and hepatocytes are exemplified by their distinct sorting of sphingolipids. Sphingolipids, a class of lipids which includes glycosphingolipids and sphingomyelin (SM), are enriched to different extent in the apical membrane of polarized epithelial cells (Holthuis et al., 2001), with glycosphingolipids showing a more pronounced polarity than SM. This implies that glycosphingolipids and SM are (partly) sorted from each other during their transport to the PM. Indeed, when fully polarized MDCK cells were incubated with fluorescently labeled (C6NBD)-ceramide, the newly synthesized C6NBD-SM and –glucosylceramide (GlcCer) are transported predominantly to the basolateral and apical membrane, respectively (van Meer et al., 1987). In contrast, no sorting of these sphingolipid analogues was observed in the transcytotic pathway of these cells (van Genderen and van Meer, 1995). Careful examination of 24 Chapter 1 transcytotic lipid transport in polarized HepG2 hepatoma cells revealed, however, that sorting in the transcytotic pathway of polarized cells can occur. This could be demonstrated after loading the endosomal compartment of these cells, accomplished by bulk-flow transport of sphingolipid analogues from the basolateral to the apical membrane. After their accumulation in the SAC, and continuing the incubation at a transport permissive temperature, C6NBD-SM was efficiently transported back to the basolateral membrane whereas C6NBD-GlcCer remained in the apical pole of the cells, where it was recycling between SAC and the apical (bile canalicular, BC) membrane (van IJzendoorn and Hoekstra, 1998; see fig. 2). HepG2 cells can transport newly synthesized sphingolipids directly from the TGN to the BC (Zegers and Hoekstra, 1997), although direct evidence for sorting of sphingolipids in the TGN was not provided at that time. However, more recently indirect evidence was obtained that newly synthesized fluorescent sphingolipid analogues are indeed sorted in the TGN of HepG2 cells. As in MDCK cells, SM is transported predominantly to the basolateral membrane while GlcCer is, in relative amounts, the major lipid going to the BC (Maier and Hoekstra, 2003). Most studies on polarized protein sorting in the endosomal system have been performed in MDCK cells, often using TfR and pIgR as markers for recycling and transcytosis, respectively. After internalization at the apical and basolateral membrane the endocytosed material encounters first apical and basolateral early endosomes, respectively (Mostov et al., 2000). Here the membrane-bound proteins are sorted from soluble components; most of the soluble molecules are transported further into the lysosomal degradation pathway, but a major part of the membrane-bound components is recycled directly back to the original plasma membrane, a mechanism which may help to restrict the intermixing of basolateral and apical molecules. Nevertheless, in non-polarized cells approximately 50 % of the internalized TfR is transported further into the cell to the pericentriolar recycling compartment (PRC) from where it is efficiently recycled back to the plasma membrane (Hao and Maxfield, 2000). 25 Chapter 1 In polarized cells, proteins internalized from the apical and the basolateral membrane both have access to a tubular endosomal compartment, which is localized around the centriole in the apical pole of the cell, and has been defined as common endosome (CE; Knight et al., 1995). From the CE, which is presumably homologous to the SAC in hepatocytes, apical and basolateral proteins are recycled back to their original membrane. Recently it was shown that in MDCK cells TfR does also colocalize with the transcytosing pIgR in the CE; TfR is recycled back from this compartment to the basolateral membrane and pIgR is transported via the apical recycling endosome (ARE) to the apical membrane (Brown et al., 2000). Because the ARE is a specialized compartment involved in the apical delivery of molecules in polarized cells it was postulated to represent the site of polarized sorting in the endocytotic pathway. However, these data clearly indicate that it is the CE that is responsible for polarized sorting, whereas the ARE may have additional functions or it may act as a “back-up” sorting station, where proteins that should be transported to the basolateral membrane and which have escaped the sorting process in the CE are ultimately removed from the apical pathway. That CE and ARE are indeed distinct compartments or that they may represent different subregions in a pleiotropic compartment, perhaps dictated by tubular extensions; this has been verified by their distinct pH values, which are 5.8 for the CE and 6.5 for the ARE, indicating that intermixing of their lumenal contents is restricted (E. Wang et al., 2000). Promising markers to further distinguish the different endosomal (sub)compartments are small GTPases of the rab family. Interestingly rab11, which colocalizes to the TfR-positive recycling compartment in non-polarized cells is localized to the TfR-negative apical recycling endosomes in MDCK cells, but not to the CE (Brown et al. 2000), indicating that regulatory proteins can fulfill different tasks depending on the requirements for the trafficking pathways, and indicating in this case that ARE and recycling endosomes are not identical compartments. Although there is evidence that rab11, as well as rab17 and rab25, may 26 Chapter 1 function in apical sorting at the level of the common endosome and associated subcompartments (Zacchi, et al, 1998; X. Wang et al., 2000; Prekeris et al., 2000), their exact role is not known. Recently a direct role for another rab protein, rab3b, has been demonstrated for the polarized trafficking of pIgR in MDCK cells (van IJzendoorn et al., 2002). Rab3b is primarily localized at an apical endosomal compartment, probably the ARE, and binds in its GTP-form to the ligand-free pIgR. In MDCK cells, the unoccupied pIgR is constitutively internalized at the basolateral membrane and the binding of rab3b-GTP mediates its transport back to the basolateral membrane. Association of dimeric IgA to pIgR at the basolateral membrane causes GTP-hydrolysis and dissociation of rab3b, which in turn enables the dIgA-pIgR-complex to reach the apical surface. Hence, in this case rab proteins provide a mechanism that carefully regulates optimal utilization of a receptor in polarized transcytotic transport, regulating apical-basolateral receptor recycling. Furthermore, binding of dIgA to pIgR induces a phosphorylation cascade, which involves the tyrosine kinase p62yes and tyrosine phosphorylation of PLCγ1, and which as such is important for the efficient transcytosis of the dIgA-pIgR complex, (Luton et al., 1999). This cascade depends on an increase of the intracellular Ca2+-concentration [via inositol 1,4,5triphosphate (IP3) signaling] and accordingly is activated by PKC stimulation (Cardone et al., 1994). In addition, pIgR can be a substrate for serine kinases and it was shown that phosphorylation of newly synthesized pIgR at Ser-726 facilitates its transport from the TGN to the basolateral PM (Orzech et al., 1999) as well as its internalization at the basolateral membrane (Okamoto et al., 1994), whereas basolateral to apical transcytosis is independent of pIgR phosphorylation. In contrast to its stimulatory effect on pIgR-transcytosis, PKC activation dramatically reduces polarity of HepG2 cells (Zegers and Hoekstra, 1997). Indeed PKC inhibits the transport of sphingolipids from both the TGN and the SAC to the apical membrane in these cells. In contrast, stimulation of PKA stimulates the transport of sphingolipids to the apical 27 Chapter 1 membrane of HepG2 cells and also increases the size and number of BC structures in these cells (Zegers, and Hoekstra, 1997). In particular, the size increase after a short-term incubation with cAMP analogues indicates that PKA causes a redistribution of membrane components from an endosomal compartment to the BCs. Indeed, it was shown that SM, which in fully matured cells is predominantly transported to the basolateral membrane after its accumulation in the endosomal SAC, is redirected from this compartment to the apical membrane upon PKA activation (van IJzendoorn and Hoekstra, 1999). Figure 3: Schematic representation of sphingolipid sorting in the SAC of HepG2 cells. Under normal growth conditions, glucosylceramide (GC) and sphingomyelin (SM) are laterally segregated in the SAC membrane ( ). GC is recycled back to the membrane of the bile canaliculus (BC), while SM is transported to the basolateral plasma membrane. Activation of PKA, naturally occurring in young developing cells or artificially-induced by exogenous addition of cAMP analogs, causes a redirection of SM trafficking from the SAC. The lipid now enters a nocodazole-sensitive subregion of SAC (SM*) while on its way to the apical (BC) membrane. Moreover, this pathway is distinct from the GCrecycling pathway. Interestingly, the apical transport of SM after PKA activation follows the same route as the transcytosing pIgR, which is clearly distinct from the apical recycling pathway of GlcCer (van IJzendoorn and Hoekstra, 1999), 28 Chapter 1 indicating that at least two different apical transport pathways originate from the SAC (fig. 3) In several studies it has been reported that PKA activation can induce the redistribution of proteins from an internal endosomal compartment to the apical membrane. Well-known examples are the water channels aquaporin 1 and 2 that are redistributed to the apical membrane of kidney cells after PKA activation induced by binding of vasopressin to the basolateral membrane (Han and Patil, 2000, Katsura et al., 1997). The redistribution of aquaporin 2 in renal principal cells depends on the presence of an A kinase-anchoring protein (AKAP), either at the ARE or the apical membrane, where it is thought to sequester the RII subunit of PKA, thereby providing the link between vasopressin signaling and trafficking of aquaporin 2 (Klussmann and Rosenthal, 2001, Jo et al., 2001). In hepatocytes, PKA activation induced the redistribution of several ABC transporters, including the bile salt exit pump (BSEP), from endosomal pools to the BC membrane (Kipp et al., 2001). Although a similar redistribution of BSEP was found after incubating hepatocytes with taurocholate, the effects were clearly distinct, indicating that the increase in bile salts is not the physiological stimulus for PKA-induced apical transport of ABC transporters. The extensive membrane redistribution in the epithelium of the urinary bladder upon stretching is another example for PKA-mediated apical transport in the endosomal system (Truschel et al., 2002). Most likely in these cells it is the ARE rather than the CE or the AEE that acts as a reservoir for proteins destined for the apical membrane. From here, they can be transported instantly to the apical membrane upon a physiological stimulus. Although there is no indication for the presence of such a ‘storage’ compartment in the basolateral endocytic pathway, it may exist in some nonpolarized cells. For example, the features that are associated with the redistribution of GLUT4 transporters in muscle cells and adipocytes from a specialized recycling endosomal compartment upon binding of insulin (Lampson et al., 2001), are reminiscent of events following signaling in the apical pathway. 29 Chapter 1 Recently it was shown that PKA activation not only redirects SM in fully polarized cells, but that intrinsic activation of PKA activity is required in young HepG2 cells to develop apical membrane domains (van IJzendoorn and Hoekstra, 2000). Thus, a polarized phenotype does not develop when the cells are cultivated in the presence of the PKA inhibitor H89. At such conditions, apical-directed transport of SM is inhibited, emphasizing once again the close correlation between the rerouting of a SM-marked membrane flow, exiting from SAC, and development of polarity. Whether the rerouting of SM as such is relevant to this development remains to be seen, for example as a necessary compound for raft-assembly (see Hoekstra and van IJzendoorn, 2000), a sorting principle thought to be relevant to apical membrane targeting. In this context it is interesting to note the importance of SM for development of polarity in neurons. Here, SM is required for axonal outgrowth (Ledesma et al., 1999), although it is not clear whether PKA activation is involved. B. Polarized sorting during vesicular transport Most transport processes between membranes within eukaryotic cells are mediated by membrane-bound vehicles that are released from the donor compartment and fuse, after their transport, specifically with the target membrane (Rothman, 1996). To establish and maintain the specific protein and lipid composition of the cellular organelles it is necessary to tightly regulate all aspects of this vesicular transport event. Originally it was thought that transport in general is mediated by small spherical vesicles, although in cells overexpressing GFP-tagged cargo proteins also large irregularly shaped containers have been identified that mediate the transport from the TGN to the apical and basolateral membrane (Hirschberg et al., 1998; Keller, et al., 2001). Most data available deal with the formation of spherical vesicles and it is unclear whether the formation of large transport containers requires different mechanisms and regulation steps, although some general features are likely to be similar. For example: a necessary step during vesicle formation is the 30 Chapter 1 specific sequestration of proteins to be transported into a domain that will bud into a vesicle, while resident proteins of the donor organelle excluded. In this section we will concentrate on recent findings that may shed light on the possible mechanisms that govern the sorting of proteins and lipids into the budding vesicle. Because most studies so far concentrated on the sorting within the biosynthetic pathway, the focus will be on vesicle formation at the TGN, but we will also present some recent mechanistic data on polarized sorting in endosomes. 1. Sorting during vesicle formation The minimal requirements for vesicle budding and fission have been analyzed in reconstituted systems using large liposomes. These studies revealed that interaction of cytoplasmic coat proteins with lipids and/or cytoplasmic tails of transmembrane proteins of the donor membrane is sufficient for the formation of vesicles (see e.g. Matsuoka et al., 1998, Bremser et al., 1999, Zhu et al., 1999). Consequently, the presence of cargo proteins in the donor compartment is not required for vesicle formation per se. In contrast, the specific sorting of transport proteins into the forming vesicle at cellular membranes requires the interplay of the cytoplasmic budding machinery with components of the donor membrane as well as the cargo molecules (Schmid, 1997; Kuehn et al., 1998, Lanoix, et al., 2001). So far three distinct coat protein (COP) complexes have been characterized. The COPII coat mediates exit of newly synthesized proteins from the ER. COPI-coated vesicles are required for the retrograde (and possibly anterograde) transport within the Golgi and from the Golgi to the ER, and several classes of clathrin-coated vesicles have been described that originate from the TGN, the plasma membrane and endocytotic organelles. These are distinguished by distinct adaptor protein complexes (up to now AP1 to 4 have been characterized), which mediate the binding of clathrin to the forming bud (for a review see Hirst and Robinson, 1998). These AP-complexes consist of two large subunits, one medium 31 Chapter 1 chain (µ1-4) and one small chain (σ1-4). Although one of the large subunits (β-adaptin 1-4) is highly homologous and mediates the binding to clathrin, the other is more specific and is used for the characterization of the different AP complexes. These are α-adaptin (for AP2), γ-adaptin (AP1), δ-adaptin (AP3) and ε-adaptin (AP4). Several of these AP complexes have been localized to the TGN where they are involved in the exit of distinct classes of proteins, indicating that they are important for sorting in the TGN. The “classical” AP1 complex binds to mannose 6-phosphate (M6P) receptors (MPR) and mediates the transport of soluble lysosomal enzymes to the endosomal system (Le Borgne and Hoflack, 1997). The small GTPase ARF1 mediates the binding of the AP1 complex to the TGN (West et al., 1997). Association of GTPase to the membrane is regulated by exchange of GDP for GTP, which causes the exposure of an N-terminal myristoyl fatty acid anchor. ARF1 also mediates the binding of COPI complexes to the cis-Golgi (Palmer et al., 1993), indicating that additional factors are required to recruit coat protein complexes to distinct membranes. Indeed, although the interaction of MPRs with M6Pcontaining proteins in the Golgi lumen and AP1 components in the cytoplasm might be sufficient for sorting, recently an additional class of proteins, the Golgi-localizing, gamma-adaptin ear homology domain, ARF-binding proteins (GGAs), has been described that has been implicated in providing specificity to this pathway by binding to both ARF1 and MRPs (Takatsu et al., 2001; Puertollano et al., 2001; Doray et al., 2002). Although it remains to be shown whether GGAs have a role in polarized transport, they can bind to various ARFs, including ARF6 (Takatsu et al., 2002), which is found at endosomes, indicating that they are involved in various transport pathways originating not only from the TGN but also from endosomal compartments. Although it was postulated that vesicle formation requires the attachment of a cytosplasmic coat complex to the site of vesicle formation, it was only recently that a protein coat was identified that is involved in the transport from the TGN to the plasma membrane. A variant of the AP1 complex 32 Chapter 1 containing a distinct medium chain, APµ1b, was described that mediates the transport of newly synthesized proteins to the basolateral membrane of LLC-Pk1 cells (Fölsch et al., 1999), presumably by binding tyrosinebased basolateral sorting signals, as was shown for the LDL receptor (Fölsch et al., 2001). Nevertheless there are several indications that APµ1b expression is not sufficient for basolateral versus apical sorting in all epithelial cells. First, several basolateral proteins are correctly transported and targeted, independently of APµ1b in LLC-Pk1 cells, whereas fibroblast, which do not express APµ1b, are able to sort at least some apical and basolateral cargo proteins into distinct vesicles (Yoshimori et al., 1996). Second, some polarized cells, e.g. hepatocytes, do not express APµ1b, indicating that in these cells a different sorting mechanism must exist. AP4, which has been implicated in basolateral transport from the TGN, may be a candidate for mediating such an alternative pathway (Simmen et al., 2002). But it is striking that in hepatocytes most apical proteins reported thus far are first transported to the basolateral plasma membrane before they are transcytosed to the canalicular membrane. This implies that polarized sorting in hepatocytes takes place mainly at the level of plasma membrane and/or endosomes and that other mechanisms exist in these cells to separate apical and basolateral proteins and lipids. Whether those mechanisms involve AP/clathrin-coated vesicles is not known, but recently AP1/clathrin-coated vesicles budding from recycling endosomes were described in MDCK cells, that are enriched in TfR (Futter et al., 1998). These vesicles may therefore separate basolateral cargo from the transcytosing pIgR, which was enriched in cup-shaped vesicles (Gibson et al., 1998), although in another study AP1 was reported to colocalize with pIgR at the level of the ARE (Wang et al., 2001). In addition endosomederived ARF6-containing vesicles were isolated from nonpolarized HEK293 cells, which were covered with an unknown coat complex and which contain also TfR, indicating that they may be involved in basolateral transport in polarized cells (Peters et al., 2001). On the other hand, because ARF6 was localized to the apical membrane in MDCK cells 33 Chapter 1 (Altschuler et al., 1999), it may have a direct function in apical transport in polarized cells. However, current evidence generally implies that the major function of protein coat complexes is the active sorting of cargo molecules away from apical cargo and resident membranes into vesicles destined for the basolateral membrane. In contrast, so far no coat protein complex has been described that is involved in the formation of transport vesicles containing apical cargo. This may reflect differences in the localization of targeting sequences in the cargo molecules. So far the only signals that have been localized in the cytoplasmic domain of cargo proteins are those mediating basolateral targeting, and coat components have been shown to interact specifically with these signals. In contrast, apical sorting signals have been shown to be localized almost exclusively in the membrane domain or the lumenal domain of cargo proteins (Keller and Simons, 1997). Consequently coat proteins cannot directly interact with sorting determinants of apical proteins and would therefore require a transmembrane receptor that recognizes both the lumenal cargo and the cytoplasmic coat. This is actually the case for AP1/clathrin-mediated transport of M6P-containing proteins from the TGN to endosomes (see above), but so far no similar receptor for apical targeting has been identified. The evidence for the polarized targeting of pIgR from AREs indicates that in this case only basolateral transport is mediated by AP1/clathrin-coated vesicles, because brefeldin A, which interferes with the binding of ARF1 to the membrane, disturbs the basolateral recycling of the TfR and results in its mistargeting to the apical membrane (Wang et al., 2001). On the other hand special AP1/clathrin-coated vesicles containing a distinct β-subunit, which were implicated in the transport of H+/K+-ATPase to the apical membrane in gastric acid secretory cells (Okamoto and Jeng, 1998), may be of a coated vesicle fraction that mediates transport from the ARE to the apical membrane. However, if coated vesicles would mediate only basolateral targeting, the specific binding of adaptor proteins to the targeting signals could cause the clustering of basolateral proteins into the forming vesicles, thereby 34 Chapter 1 resulting in their separation from apical proteins. This mechanism would also explain the observed hierarchy of sorting signals, i.e., basolateral targeting signals being dominant over apical ones (Ikonen and Simons, 1998). Yet, there is good evidence that apical proteins and lipids form clusters in the donor membrane, thus being actively sorted from basolaterally directed cargo. In particular the formation of cholesterol- (glyco)sphingolipid-enriched membrane domains in the TGN has been implicated in this clustering of apical cargo (Verkade and Simons, 1997). Sphingomyelin and most glycosphingolipids are synthesized in the lumenal membrane leaflet of the Golgi. The biophysical properties of sphingolipids, bearing a long-chain saturated fatty acid and alkyl chain of high melting temperature, favor, their packaging, together with cholesterol, into small membrane domains at the level of the TGN (Brown and London, 2000). These membrane domains, often called ‘rafts’, can be identified by their resistance to extraction with nonionic detergents and the low buoyant density of these extracts in sucrose gradients. Glycosylphosphatidylinositol (GPI)-anchored proteins as well as several transmembrane proteins, which are usually expressed in the apical membrane of epithelial cells, are associated with these rafts, whereas basolateral proteins are excluded from these domains (Verkade and Simons, 1997). Raft association of newly synthesized apical proteins occurs in the TGN and disruption of rafts, e.g. by depleting the cholesterol pool of the cells, perturbs raft association and retards apical transport. In addition, it may cause randomization of protein transport, indicating that rafts are important for apical targeting (Keller and Simons, 1998). Interestingly, although cholesterol depletion disrupts raft integrity in FRT-cells, only inhibition of sphingolipid synthesis causes the randomization of polarized transport (Lipardi et al., 2000), indicating that sphingolipids are especially important for the polarized sorting. Although rafts may be important for the separation of apical and basolateral cargo, it is clear that they are not sufficient for apical targeting. Both N- and O-glycans have been identified as apical targeting 35 Chapter 1 signals (Scheiffele et al., 1995; Alfalah et al., 1999), indicating that lectins, e.g., VIP36 (Fiedler et al., 1994), may have a function in delivery to the apical membrane. Other proteins that have been implicated in targeting of vesicles to the apical membrane are the lipid-binding protein annexin XIIIb (Lafont et al., 1998) and the transmembrane protein MAL (Cheong et al., 1999), which were both shown to be raft-associated. Indeed, MAL can interact with GPI-anchored proteins (Millan and Alonso, 1998) and this interaction may be important for apical sorting of these proteins. It is still unknown whether MAL and other proteins directly target rafts to the apical membrane or whether they constitute an intermediate link to other targeting devices, such as SNAREs (see below). However, so far no evidence has been provided that would support an interaction of MAL with SNARE proteins. The observations mentioned above indicate that interactions between the apically directed molecules in the lumen of the vesicle are important for their clustering into rafts and their sorting into the assembling vesicle. Consistent with this notion, it has been shown that rafts can also mediate transport of secretory proteins to the plasma membrane (Martin-Belmonte et al., 2001). Intermolecular interactions, important for polarized sorting, also occur in the lumen of the SAC, as has been shown for the sorting of fluorescently labeled SM and GlcCer. As described above, in fully polarized HepG2 cells, C6NBD-SM and GlcCer are transported from the SAC to the basolateral and apical membrane, respectively. Because these probes differ only in their head group (choline versus glucose) it is evident that head group-specific sorting is operational in this case. Moreover, C6NBDgalactosylceramide, an epimer of C6NBD-GlcCer, is transported from the SAC to the basolateral membrane, further emphasizing head groupdependent sorting but also reflecting a very specific interaction of C6NBDGlcCer with the apical transport pathway (van IJzendoorn and Hoekstra, 1998). Although direct extrapolation of results obtained with fluorescent lipid analogues requires a direct comparison with their natural counter parts, which is often hampered due to technical limitations (see Maier et al., 36 Chapter 1 2002), these results could be taken as a first indication that membrane domain formation also plays a role in polarized sorting in endosomal compartments. That rafts do exist in endosomal compartments was confirmed in a recent study that demonstrated that the CE of MDCK cells is enriched in raft components, although these were isolated by antibodies against the basolaterally directed TfR (Gagescu et al., 2000). This indicates that either transport of the TfR from the CE to the basolateral membrane is mediated by rafts or, alternatively, that these raft components were coisolated with the TfR, although they are actually localized in distinct parts of the CE. The first possibility seems unlikely because cholesterol depletion has no effect on the recycling of TfR in nonpolarized cells. In contrast, the recycling of GPI-anchored proteins, which is much slower than TfR recycling under normal conditions, is accelerated upon cholesterol depletion and indistinguishable from TfR recycling (Chatterjee et al., 2001). Surprisingly, the effects of cholesterol depletion on endocytotic trafficking appear to be different from those on the biosynthetic pathway. Although it generally slows down the transport from the TGN to the plasma membrane, recycling from the endosomal system is accelerated. This indicates that rafts in the CE cause the retention of GPI-linked proteins and such a mechanism may facilitate their polarized sorting to the apical membrane in polarized cells. Evidence that sphingolipid-cholesterol-enriched membrane domains in apical endosomes may be involved in the transcytosis of pIgR comes from studies in enterocytes (Hansen et al., 1999). In contrast, pIgR transport seems to be raft independent in MDCK and FRT cells, although transient interactions, e.g., in the CE or ARE, cannot be excluded (Sarnataro et al., 2000). Clearly, the formation of specified membrane domains may play an important role in the clustering and sorting of proteins and lipids towards the apical domain. Nevertheless many reports of raft-independent apical transport exist. In this respect it should be mentioned that proteins may display differences in resistance to detergent solubilization, a feature used for raft characterization. One 37 example is prominin, a Chapter 1 multimembrane spanning membrane protein in the apical membrane of epithelial cells, which is completely soluble in Triton X-100, but is resistant to extraction with lubrol WX (Roper et al., 2000). Similar features have been observed in the sorting of the mulitmembrane spanning proteins MDR1 and the copper transporter ATP7B in HepG2 cells. These proteins are readily extracted with Triton X-100, but are insoluble in Lubrol WX, implying that sorting and direct transport of apical resident proteins in liver cells could be mediated by Lubrol instead of Triton X-100 rafts. Remarkably, Triton-rafts are operating in the transport of apical resident proteins that travel via the indirect transcytotic pathway (for a further discussion, see Aït Slimane and Hoekstra, 2002; Aït Slimane et al., 2003). The data indicate that in a (polarized) cell type-dependent manner, various membrane domains with distinct biophysical properties appear to exist that mediate apical transport, and some of these may be formed independently of sphingolipids or cholesterol. Presumably, as a rule, the domains display a distinct degree of liquid ordering that could be provided by sphingolipids, cholesterol and saturated lipids, (co)dictated by the lipid and perhaps protein composition of the apposed leaflet. The absence of distinct lipids might be compensated for by others, provided that the overall liquid-ordered state is maintained. Also in MDCK cells, distinct vesicles for two apically transported proteins have been identified, the raft-associated sucrase-isomaltase and the raft- independent (Triton X-100-soluble) lactase-phlorizin hydrolase (Jacob and Naim, 2001), which were probably segregated after their exit from the TGN. Mechanisms and players, instrumental in governing these distinct sorting and vesicular transport pathways that lead to the same site of destination, remain to be elucidated. In this context, it has been shown that MAL is involved in the apical targeting of both raft-dependent and raft-independent proteins (Martin-Belmonte et al., 2000). In addition to their function in clustering and sorting of apical cargo, sphingolipid-enriched membrane domains may also facilitate the formation of the transport vesicles. The generation of a membrane 38 Chapter 1 curvature at the site of budding is one of the main features of vesicle formation. As described above, coat proteins can induce the formation of vesicles from artificial liposomes, proving that they are sufficient to provide the driving force for vesicle budding and fission. Membrane curvature can be induced by the lipid composition of the membrane itself, involving the asymmetric distribution of lipids between the lumenal and cytoplasmic leaflets of the membrane bilayer. Because of their relative rigid, i.e., tight packing, the formation of sphingolipid-cholesterol domains in the lumenal leaflet of the membrane would induce a curvature of the membrane, provided that packing properties in the cytoplasmic leaflet remain unaltered. As a consequence, the result would be the formation of a vesicular bud at the site of raft assembly. Association of proteins with raft lipids would therefore not only separate these proteins from the basolaterally directed clathrin-coated vesicles, but may also be sufficient to concentrate them into apically directed transport carriers (for further discussion see Huttner and Zimmerberg, 2001). For the further characterization of the molecular mechanisms that govern and regulate polarized sorting and vesicle formation, in vitro assays using permeabilized cells or isolated organelles are desirable. Indeed various “budding assays” have been developed to study vesicle formation from the TGN, and it was demonstrated that MDCK cells, after infection with vesicular stomatitis virus (VSV) and influenza virus, release the basolateral marker VSV glycoprotein and the apical marker influenza hemagglutinin in distinct transport vesicles (Wandinger-Ness et al., 1990). Also, newly synthesized sphingolipids are released in distinct vesicles from the TGN of MDCK and HT29 cells (Kobayashi et al., 1992; Babia et al., 1994). Although it should be possible to use these in vitro systems to further characterize the mechanisms of apical versus basolateral sorting during vesicle formation, so far they were mainly used to characterize the regulation of vesicle release and to distinguish the formation of distinct basolateral vesicles at the TGN. Thus distinct vesicles for secretory and transmembrane proteins as well as serum 39 Chapter 1 albumin and heparansulfate proteoglycan (HSPG) have been identified in hepatocytes (Saucan and Palade, 1994; Nickel et al., 1994). The impact of various protein kinases on vesicle formation from the TGN has been investigated using specific activators and inhibitors. In most systems basolateral transport proteins, especially VSV glycoprotein, but also HSPG, were used to monitor vesicle release. However, little is known about the release of apical transport vesicles. In these assays it was shown that both PKA and PKC are involved in mediating vesicle release from the TGN, although it is not clear whether the kinase activity of PKC is actually required (Muniz et al., 1997; Simon et al., 1996; Westermann et al., 1996). Other regulatory proteins that have been implicated in vesicle formation are tyrosine kinases and phosphatases as well as trimeric G proteins (Austin and Shields, 1996; Leyte et al., 1992). These data show that vesicle formation is a highly regulated process, although the mode of action of these proteins is currently unknown. Possible kinase substrates include cargo molecules (see above for pIgR), components of protein coats, or putative sorting receptors. The phosphorylation of MPR by a casein kinase II-like protein at the Golgi, which facilitates its interaction with the AP1 complex, is one example for such a mechanism (Mauxion et al., 1996). Phosphorylation of subunits of the AP complexes may be important for their binding specificity, as was shown for AP2 (Fingerhut et al., 2001), but so far there is no indication that such a mechanism is involved in polarized sorting. The analysis of polarized vesicle release from endosomes was hindered by difficulties to exclusively label the CE with specific marker molecules. Therefore only few in vitro assays with endosomal membranes have been established. Recently the possibility to accumulate fluorescent sphingolipid analogues in the SAC of HepG2 cells has been used to characterize the sorting capacity of the SAC membrane and indeed distinct vesicles enriched in either C6NBD-SM or -GlcCer were identified after permeabilization of the cells (Maier and Hoekstra, 2003). Other assays resulted in the discovery of ARF6-containing vesicles covered by a new coat (Peters et al., 2001; see above) and the characterization of 40 Chapter 1 H+/K+-containing vesicles derived from apical endosomes of gastric parietal cells (Okamoto et al., 2002). The isolation and characterization of distinct endosomal compartments from LDL-loaded hepatocytes (Enrich et al., 1996; Vergès et al., 1999) may provide new opportunities for the further analysis of polarized sorting in endosomal compartments. 2. Specific targeting of vesicles After budding from the donor membrane the vesicles have to be transported to the distinct plasma membrane domains. Efficient transport requires an intact microtubular cytoskeleton, and disruption of microtubules affects both apical and basolateral transport, indicating that microtubules are important for overall targeting of the transport vesicles (Lafont et al., 1994). In addition, microtubule disruption not only disturbs several transport steps, but also interferes directly with the structure of the sorting organelles. At the same time, this feature may make it difficult to determine whether they have a direct role in polarized sorting and targeting to the plasma membrane. Nevertheless, recent evidence indicates that interaction of vesicles with certain kinesins, directed to the minus end of microtubules, can be affected by the lipid composition of the membrane so that they may associate specifically to apically directed vesicles. Thus the raft lipids cholesterol, sphingomyelin and GM1 enhance the binding of KIF1a to liposomes (Klopfenstein et al., 2002), and KIFc3 has been found in detergent-resistant domains associated with annexin XIIIb (Noda et al., 2001). The involvement of the actin cytoskeleton in polarized transport is even less clear, but interestingly myosin II has been localized to coated vesicles at the TGN (Müsch, et al., 1997; Ikonen et al., 1997). Association of myosin II to the TGN was disrupted by brefeldin A, indicating that it may bind to AP1 complexes (Ikonen et al., 1997) and thus be involved in basolateral transport. Indeed it was shown that myosin II is involved in the release of VSV glycoprotein from the TGN, but the protein has no effect on the exit of HA, indicating that attachment to actin filaments may 41 Chapter 1 be specifically involved in efficient formation of vesicles directed to the basolateral membrane (Müsch et al., 1997). The next step is the specific docking and subsequent fusion of the vesicle with the target membrane. SNARE proteins have been localized to distinct organelles and play a crucial role in the targeting process. Formation of trans-complexes between SNAREs present at the vesicle (v-SNARE) and target (t-SNARE) membrane are required for specific docking and possibly also for subsequent fusion (Pelham, 1997; Pecheur et al., 2000). The tSNAREs syntaxin 3 and syntaxin 4 are localized to the apical and basolateral membrane of polarized epithelial cells and therefore have been implicated in the polarized targeting of transport vesicles (Gaisano et al., 1996), whereas cellubrevin is a ubiquitous v-SNARE that may be involved in basolateral targeting of vesicles from TGN and/or endosomes. Indeed it was shown that cellubrevin is a component of the ARF6containing vesicles derived from endosomes (Peters et al., 2001). The function of SNAREs in vesicle targeting was originally identified by the sensitivity of the fusion of synaptic vesicles to botulinus and tetanus toxin and the insensitivity of apical transport to tetanus toxin was taken as indication that apical targeting is independent of the SNARE machinery (Ikonen et al., 1995). Recently, a tetanus toxin-insensitive SNARE (TI-SNARE) was identified that may be the v-SNARE that mediates transport to the apical membrane (Galli et al., 1998). To target vesicles to their destination v-SNAREs have to be incorporated into the vesicle during its formation. Recent evidence as to how this may be achieved comes from studies of ER-Golgi transport. Here, ARF1 mediates the binding of the COPI coat complex and the GTPase activity of ARF, regulated by an ARF-GAP (ARF GTPase activating protein), is required for efficient incorporation of cargo into the budding vesicle. In addition ARF-GAP interacts with v-SNAREs and this interaction is required for efficient association of COPI components with the membrane. Thus ARF-GAP links v-SNARE incorporation with incorporation of cargo proteins into the forming vesicle and vesicle budding (Rein et al., 2002). Whether this is a ubiquitous mechanism that is also operational at the 42 Chapter 1 level of the TGN and of endosomes remains to be seen. In addition, the association of SNAREs with rafts, as described for TI-SNARE and syntaxin 3, may provide a mechanism for their specific targeting to the apical membrane (Lafont et al., 1999). Although SNAREs are required for specific targeting, they are not sufficient. In budding yeast a protein complex, the exocyst, marks the location where new membrane material has to be internalized (Guo et al., 1999). In mammalian cells a protein complex homologous to the exocyst has been described, which is required for basolateral transport of LDL receptors, but not for apical transport. This complex is localized close to the tight junctions indicating that lateral cell-cell adhesion is important for the generation of cell surface polarity (Grindstaff et al., 1998). Additional regulators for the assembly of SNARE complexes at the site of the exocyst are members of the Sec1 protein family (Carr et al., 1999). In polarized epithelial cells the mammalian sec1 homologues munc18-2 and munc18c have been implicated in apical and basolateral transport by inhibiting syntaxin 3 and syntaxin 4-mediated fusion of vesicles with the respective plasma membrane domains (Riento et al., 2000; Tellam et al., 1997). In pancreatic acinar cells zymogen granules are stored in the cell and exocytosed after incubation with cholecystokinin (CCK). Incubation with supramaximal CCK concentrations abolishes apical exocytosis but has only a minor effect on overall secretion. It was shown recently that CCK incubation caused the dissociation of munc18c from the basolateral membrane thereby exposing fusion complexes. This in turn results in the basolateral exocytosis of zymogen granules which may result in mild acute pancreatitis (Gaisano et al., 2001). Interestingly the dissociation of munc18c is also mediated by PKC activation, although it has not yet been shown that munc18c is a PKC substrate. Such a mechanism may provide an explanation for the decrease in polarization of HepG2 cells caused by PKC activation (Zegers and Hoekstra 1997). 43 Chapter 1 V. Conclusions and perspectives In the last years the investigation of polarized transport has become one of the major subjects in cell biology. Especially the notion that certain lipids play a crucial role in membrane domain formation and that this domain formation is an important step for polarized sorting, opened a new field in membrane trafficking research. Evidently, the presence of such rafts will also have a major impact on other aspects of cell biology, as is shown by the growing evidence that similar membrane domains are also involved in many signal transduction cascades, including those that might regulate polarity development. Further impact in uncovering polarized membrane transport will come from the analysis of the various GFP-variants, displaying distinct fluorescence spectral properties. Thus different probes can be used to monitor simultaneously the flow of apical and basolateral proteins. Until recently, studies on polarized transport have been mainly restricted to the biosynthetic pathway, whereas the involvement of endosomes has been largely ignored. However, not surprisingly, cells exhibit a much greater complexity as originally anticipated and the discovery of the subapical compartment or common endosome in polarized cells will now bring the focus of research also on the role of the endocytotic pathway in development and maintenance cell polarity. A major challenge will be to characterize functions of these compartments on the molecular level. Among others, their isolation and purification would thus be highly desirable. 44 Chapter 1 Scope of this thesis In this thesis a novel role for the IL-6 type cytokine Oncostatin M (OSM) in the regulation of plasma membrane polarity development is described, as investigated in polarized HepG2 cells. HepG2 cells are human-derived hepatocellular carcinoma cells, and are well known for their ability to adopt polarity when cultured in vitro. In this system the effect of OSM on polarity development and sphingolipid trafficking was studied as propagated by virtue of signaling via its receptor, present on the basolateral membrane domain. This work, as described in chapter 2 revealed a correlation between OSM-mediated activation of PKA and polarity development, involving a mechanism that appears to rely on a reorganization of the intracellular distribution of PKA rather than on a net enhancement in PKA activity. Since sphingolipids play an important role in both cell polarization and proliferation, we analyzed the effects of polarity stimulating-signaling cascades on sphinganine turnover, using the mycotoxin fumonisin B1. Fumonisin B1 is known to inhibit dihydroceramide synthase, which causes an elevation of the intracellular levels of sphinganine, and is also associated with the development of hepatocellular carcinoma. In chapter 3, we investigated the effects of elevated sphinganine levels on polarized endosomal membrane traffic, and the data indicated a correlation between polarity development on the one hand, and the level of sphinganine and the expression of the cell cycle regulatory protein p27kip1 on the other. Moreover, dihydroceramide synthase turned out to be a target for the cAMP/PKA signaling cascade, known to be involved in promoting cell polarity development. In chapter 4 we therefore analyzed in detail the effects of p27kip1, a CDK inhibitor, on differentiation and subsequent polarization of HepG2 cells. Increased levels of p27kip1 are associated with a block in G1-S transition, thus keeping the cells in G1. Eventually they may leave the cell cycle and enter G0 to undergo terminal differentiation. The data demonstrate that both OSM and activation of PKA increased the intracellular levels of p27kip1, which was associated with a block in proliferation and an increase of 45 Chapter 1 polarization. Furthermore, we show that during early polarity development p27kip1 is reallocated from the nucleus to the cytosol where it possibly facilitates events that precede and/or promote the acquirement of cell polarity. The dynamics of the actin cytoskeleton is of importance for tissue morphogenesis and is carefully controlled by regulatory proteins, including the small GTPase Rho and its downstream effector ROCK. A loss of tissue architecture is often observed in carcinogenesis. In chapter 5 it is shown that suppression of ROCK activity results in a displacement of the adherens junction protein E-cadherin and the induction of lamellipodia, which culminates in the formation of elongated bile canalicular lumens, shared by multiple cells. OSM stimulation in ROCK suppressed cells counteracted both E-cadherin junction disassembly and complex apial lumen formation. These data indicate that OSM-controlled, E-cadherin-based cell-cell contacts and the formation of lamellipodia are important factors for apical lumen morphogenesis. Taken together, the data indicate that OSM is closely involved in controlling plasma membrane polarity development by regulating membrane traffic, cell cycle control and cell-cell adhesion, as summarized in chapter 6, in which chapter some suggestion are presented for future work. 46
© Copyright 2026 Paperzz