Antibodies to basement membrane heparan sulfate proteoglycans bind to the laminae rarae of the glomerular basement membrane (GBM) and induce subepithelial GBM thickening.

Antibodies specific for the core protein of basement membrane HSPG (Mr = 130,000) were administered to rats by intravenous injection, and the pathologic consequences on the kidney were determined at 3 min to 2 mo postinjection. Controls were given antibodies against gp330 (the pathogenic antigen of Heymann nephritis) or normal rabbit IgG. The injected anti-HSPG(GBM) IgG disappeared rapidly (by 1 d) from the circulation. The urinary excretion of albumin increased in a dose-dependent manner during the first 4 d, was increased 10-fold at 1-2 mo, but remained moderate (mean = 12 mg/24 h). By immunofluorescence the anti-HSPG(GBM) was seen to bind rapidly (by 3 min) to all glomerular capillaries, and by immunoperoxidase staining the anti-HSPG was seen to bind exclusively to the laminae rarae of the GBM where it remained during the entire 2-mo observation period. C3 was detected in glomeruli immediately after the injection (3 min), where it bound exclusively to the lamina rara interna; the amount of C3 bound increased up to 2 h but decreased rapidly thereafter, and was not detectable after 4 d. Mononuclear and PMN leukocytes accumulated in glomerular capillaries, adhered to the capillary wall, and extended pseudopodia through the endothelial fenestrae to contact in the LRI of the GBM where the immune deposits and C3 were located. At 1 wk postinjection, staining for C3 reappeared in the glomeruli of some of the rats, and by this time most of the rats, including controls injected with normal rabbit IgG, had circulating anti-rabbit IgG (by ELISA) and linear deposits of rat IgG along the GBM (by immunofluorescence). Beginning at 9 d, there was progressive subepithelial thickening of the GBM which in some places was two to three times its normal width. This thickening was due to the laying down of a new layer of basement membrane-like material on the epithelial side of the GBM, which gradually displaced the old basement membrane layers toward the endothelium. The results show that the core proteins of this population of basement membrane HSPG (Mr = 130,000), which are ubiquitous components of basement membranes, are exposed to the circulation and can bind anti-HSPG(GBM) IgG in the laminae rarae of the GBM. Binding of these antibodies to the GBM leads to changes (C3 deposition, leukocyte adherence, moderate proteinuria, GBM thickening) considered typical of the acute phase of anti-GBM glomerulonephritis. Antibody binding interferes with the normal turnover of the GBM, presumably by affecting the biosynthesis and/or degradation of basement membrane components.

Budding roles for myosin II on the Golgi.

Myosin II--conventional myosin--has been typecast in muscle-man roles. While members of the Schwarzenegger clan from skeletal muscle have grabbed the limelight, myosin II motors in nonmuscle cells labour away in many varied and subtle roles. Recent findings show that nonmuscle myosin II, along with other myosins and cytoskeletal proteins, assembles on Golgi membranes. Nonmuscle myosin II associates transiently with membranes of the trans-Golgi network during the budding of a subpopulation of transport vesicles. The exact role of myosin II in vesicular trafficking is not yet understood, but its participation heralds a novel role for actin-based motors in vesicle budding.

Distinctive populations of basement membrane and cell membrane heparan sulfate proteoglycans are produced by cultured cell lines.

We have investigated the nature and distribution of different populations of heparan sulfate proteoglycans (HSPGs) in several cell lines in culture. Clone 9 hepatocytes and NRK and CHO cells were biosynthetically labeled with 35SO4, and proteoglycans were isolated by DEAE-Sephacel chromatography. Heterogeneous populations of HSPGs and chondroitin/dermatan proteoglycans (CSPGs) were found in the media and cell layer extracts of all cultures. HSPGs were further purified from the media and cell layers and separated from CSPGs by ion exchange chromatography after chondroitinase ABC digestion. In all cell types, HSPGs were found both in the cell layers (20-70% of the total) as well as the medium. When the purified HSPG fractions were further separated by octyl-Sepharose chromatography, very little HSPG in the incubation media bound to the octyl-Sepharose, whereas 40-55% of that in the cell layers bound and could be eluted with 1% Triton X-100. This hydrophobic population most likely consists of membrane-intercalated HSPGs. Basement membrane-type HSPGs were identified by immunoprecipitation as a component (30-80%) of the unbound (nonhydrophobic) HSPG fraction. By immunofluorescence, basement membrane-type HSPGs were distributed in a reticular network in Clone 9 and NRK cell monolayers; by immunoelectron microscopy, these HSPGs were localized to irregular clumps of extracellular matrix located beneath and between cells. The cells did not produce a morphologically recognizable basement membrane layer under these culture conditions. When membrane-associated HSPGs were localized by immunoelectron microscopy, they were found in a continuous layer along the cell membrane of all cell types. The results demonstrate that two antigenically distinct populations of HSPG--an extracellular matrix and a membrane-intercalated population--are found at the surface of several different cultured cells lines; these populations can be distinguished from one another by differences in their distribution in the monolayers by immunocytochemistry and can be separated by hydrophobic chromatography; and basement membrane-type HSPGs are secreted and deposited in the extracellular matrix by cultured cells even though they do not produce a bona fide basement membrane-like layer.

Recycling of E-cadherin: a potential mechanism for regulating cadherin dynamics.

E-Cadherin plays critical roles in many aspects of cell adhesion, epithelial development, and the establishment and maintenance of epithelial polarity. The fate of E-cadherin once it is delivered to the basolateral cell surface, and the mechanisms which govern its participation in adherens junctions, are not well understood. Using surface biotinylation and recycling assays, we observed that some of the cell surface E-cadherin is actively internalized and is then recycled back to the plasma membrane. The pool of E-cadherin undergoing endocytosis and recycling was markedly increased in cells without stable cell-cell contacts, i.e., in preconfluent cells and after cell contacts were disrupted by depletion of extracellular Ca2+, suggesting that endocytic trafficking of E-cadherin is regulated by cell-cell contact. The reformation of cell junctions after replacement of Ca2+ was then found to be inhibited when recycling of endocytosed E-cadherin was disrupted by bafilomycin treatment. The endocytosis and recycling of E-cadherin and of the transferrin receptor were similarly inhibited by potassium depletion and by bafilomycin treatment, and both proteins were accumulated in intracellular compartments by an 18 degrees C temperature block, suggesting that endocytosis may occur via a clathrin-mediated pathway. We conclude that a pool of surface E-cadherin is constantly trafficked through an endocytic, recycling pathway and that this may provide a mechanism for regulating the availability of E-cadherin for junction formation in development, tissue remodeling, and tumorigenesis.

Heparan sulfate proteoglycans are concentrated on the sinusoidal plasmalemmal domain and in intracellular organelles of hepatocytes.

The distribution of cell surface heparan sulfate proteoglycans (HSPGs) was determined in rat liver by immunocytochemistry. A polyclonal antibody was raised against HSPGs purified from rat liver microsomes which specifically immunoprecipitated liver membrane HSPGs. It was shown to recognize both the heparin-releasable and membrane-intercalated form of membrane HSPGs and to recognize determinants on the core protein of these HSPGs. By immunocytochemistry membrane HSPGs were localized to hepatocytes. The distribution of HSPGs at the cell surface of the hepatocyte was restricted to the sinusoidal domain of the plasmalemma; there was little or no staining of the lateral or bile canalicular domains. Intracellularly, HSPGs were occasionally detected in cisternae of the rough endoplasmic reticulum and were regularly found in Golgi cisternae--usually distributed across the entire Golgi stack. HSPGs were also localized in some endosomes, lysosomes, and cytoplasmic vesicles of hepatocytes. We conclude that the HSPGs recognized by this antibody have a restricted distribution in rat liver: they are largely confined to the sinusoidal plasmalemmal domain and to biosynthetic and endocytic compartments of hepatocytes.

A heterotrimeric G protein, G alpha i-3, on Golgi membranes regulates the secretion of a heparan sulfate proteoglycan in LLC-PK1 epithelial cells.

A heterotrimeric G alpha i subunit, alpha i-3, is localized on Golgi membranes in LLC-PK1 and NRK epithelial cells where it colocalizes with mannosidase II by immunofluorescence. The alpha i-3 was found to be localized on the cytoplasmic face of Golgi cisternae and it was distributed across the whole Golgi stack. The alpha i-3 subunit is found on isolated rat liver Golgi membranes by Western blotting and G alpha i-3 on the Golgi apparatus is ADP ribosylated by pertussis toxin. LLC-PK1 cells were stably transfected with G alpha i-3 on an MT-1, inducible promoter in order to overexpress alpha i-3 on Golgi membranes. The intracellular processing and constitutive secretion of the basement membrane heparan sulfate proteoglycan (HSPG) was measured in LLC-PK1 cells. Overexpression of alpha i-3 on Golgi membranes in transfected cells retarded the secretion of HSPG and accumulated precursors in the medial-trans-Golgi. This effect was reversed by treatment of cells with pertussis toxin which results in ADP-ribosylation and functional uncoupling of G alpha i-3 on Golgi membranes. These results provide evidence for a novel role for the pertussis toxin sensitive G alpha i-3 protein in Golgi trafficking of a constitutively secreted protein in epithelial cells.

Identification of a 200-kD, brefeldin-sensitive protein on Golgi membranes.

A mAb AD7, raised against canine liver Golgi membranes, recognizes a novel, 200-kD protein (p200) which is found in a wide variety of cultured cell lines. Immunofluorescence staining of cultured cells with the AD7 antibody produced intense staining of p200 in the juxtanuclear Golgi complex and more diffuse staining of p200 in the cytoplasm. The p200 protein in the Golgi complex was colocalized with other Golgi proteins, including mannosidase II and beta-COP, a coatomer protein. Localization of p200 by immunoperoxidase staining at the electron microscopic level revealed concentrations of p200 at the dilated rims of Golgi cisternae. Biochemical studies showed that p200 is a peripheral membrane protein which partitions to the aqueous phase of Triton X-114 solutions and is phosphorylated. The p200 protein is located on the cytoplasmic face of membranes, since it was accessible to trypsin digestion in microsomal preparations, and is recovered in approximately equal amounts in membrane pellets and in the cytosol of homogenized cells. Immunofluorescence staining of normal rat kidney cells exposed to the toxin brefeldin A (BFA), showed that there was very rapid redistribution of p200, which was dissociated from Golgi membranes in the presence of this drug. The effect of BFA was reversible, since upon removal of the toxin, AD7 rapidly reassociated with the Golgi complex. In the BFA-resistant cell line PtK1, BFA failed to cause redistribution of p200 from Golgi membranes. Taken together, these results indicate that the p200 Golgi membrane-associated protein has many properties in common with the coatomer protein, beta-COP.

Entry of cholera toxin into polarized human intestinal epithelial cells. Identification of an early brefeldin A sensitive event required for A1-peptide generation.

The effect of brefeldin-A (BFA), a reversible inhibitor of vesicular transport, on cholera toxin (CT)-induced Cl- secretion (Isc) was examined in the polarized human intestinal cell line, T84. Pretreatment of T84 monolayers with 5 microM BFA reversibly inhibited Isc in response to apical or basolateral addition of 120 nM CT (2.4 +/- 0.5 vs. 68 +/- 3 microA/cm2, n = 5). In contrast, BFA did not inhibit Isc responses to the cAMP agonist VIP (63 +/- 7 microA/cm2). BFA had no effect on cell surface binding and endocytosis of a functional fluorescent CT analog or on the dose dependency of CT induced 32P-NAD ribosylation of Gs alpha in vitro. In contrast, BFA completely inhibited (> 95%) the ability of T84 cells to reduce CT to the enzymatically active A1-peptide. BFA had to be added within the first 10 min of CT exposure to inhibit CT-elicited Isc. The early BFA-sensitive step occurred before a temperature-sensitive step essential for apical CT action. These studies show that sequential steps are required for a biological response to apical CT: (a) binding to cell surfaces and rapid endocytosis; (b) early, BFA-sensitive vesicular transport essential for reduction of the A1-peptide; and (c) subsequent temperature-sensitive translocation of a signal (the A1-peptide or possibly ADP-ribose-Gs alpha) to the basolateral domain.

Fluid-phase markers in the basolateral endocytic pathway accumulate in response to the actin assembly-promoting drug Jasplakinolide.

To investigate the role of filamentous actin in the endocytic pathway, we used the cell-permeant drug Jasplakinolide (JAS) to polymerize actin in intact polarized Madin-Darby canine kidney (MDCK) cells. The uptake and accumulation of the fluid-phase markers fluorescein isothiocyanate (FITC)-dextran and horseradish peroxidase (HRP) were followed in JAS-treated or untreated cells with confocal fluorescence microscopy, biochemical assays, and electron microscopy. Pretreatment with JAS increased the uptake and accumulation of fluid-phase markers in MDCK cells. JAS increased endocytosis in a polarized manner, with a marked effect on fluid-phase uptake from the basolateral surface but not from the apical surface of polarized MDCK cells. The early uptake of FITC-dextran and HRP was increased more than twofold in JAS-treated cells. At later times, FITC-dextran and HRP accumulated in clustered endosomes in the basal and middle regions of JAS-treated cells. The large accumulated endosomes were similar to late endosomes but they were not colabeled for other late endosome markers, such as rab7 or mannose-6-phosphate receptor. JAS altered transport in the endocytic pathway at a later stage than the microtubule-dependent step affected by nocodazole. JAS also had a notable effect on cell morphology, inducing membrane bunching at the apical pole of MDCK cells. Although other studies have implicated actin in endocytosis at the apical cell surface, our results provide novel evidence that filamentous actin is also involved in the endocytosis of fluid-phase markers from the basolateral membrane of polarized cells.

Vesicle budding on Golgi membranes: regulation by G proteins and myosin motors.

One of the main functions of the Golgi complex is to generate transport vesicles for the post-Golgi trafficking of proteins in secretory pathways. Many different populations of vesicles are distinguished by unique sets of structural and regulatory proteins which participate in vesicle budding and fusion. Monomeric and heterotrimeric G proteins regulate vesicle budding and secretory traffic into and out of the Golgi complex. An inventory of G protein alpha subunits associated with Golgi membranes highlights their diverse involvement and potential for coupling Golgi trafficking, through various signal transduction pathways, to cell growth or other more specialized cell functions. Cytoskeletal proteins are now also known to associate specifically with the Golgi complex and Golgi-derived vesicles. Amongst these, conventional and unconventional myosins are recruited to vesicle membranes. Several roles in vesicle budding and vesicle trafficking can be proposed for these actin-based motors.

Localization and post-Golgi trafficking of tumor necrosis factor-alpha in macrophages.

Tumor necrosis factor-alpha (TNF-alpha) is a proinflammatory cytokine secreted by activated macrophages. In this study, we examined the intracellular distribution and trafficking of TNF-alpha. Immunofluorescence and immunogold localization demonstrated that in lipopolysaccharide (LPS)-stimulated RAW264 macrophages, the greatest concentration of TNF-alpha is found in the perinuclear Golgi complex. Staining of the Golgi complex appeared 20 min after activation of cells and persisted for 2-12 h, and TNF-alpha appeared on the cell surface only transiently during this time. The rate of disappearance of Golgi staining correlated with the release of the cleaved, mature TNF-alpha into the medium. Pulse chase labeling and subcellular fractionation studies indicated that both 26-kDa and 17-kDa forms of TNF-alpha may be present at the level of the Golgi complex. Post-Golgi trafficking of TNF-alpha was modulated by agents that disrupt the cytoskeleton. Interferon-gamma (IFN-gamma), which primes macrophages for TNF-alpha-dependent cellular cytotoxicity, potentiated the effect of LPS by sustaining enhanced intracellular pools of TNF-alpha and also promoted redistribution of TNF-alpha into post-Golgi vesicular compartments. We propose that the primary pool of biologically active TNF-alpha in activated macrophages is held in the Golgi complex and that the cytokine is recruited directly from this intracellular pool for release in response to tumor cells or pathogens.

Recycling endosome-dependent and -independent mechanisms for IL-10 secretion in LPS-activated macrophages.

IL-10 is a key anti-inflammatory cytokine secreted by activated macrophages as a feedback control mechanism to prevent excessive inflammatory responses. Here, we define multiple intracellular trafficking pathways involved in the secretion of newly synthesized IL-10 from macrophages following TLR4 activation with LPS, and show how this relates to the previously defined trafficking pathways for IL-6 and TNF in macrophages simultaneously producing these proinflammatory cytokines. IL-10 exits the Golgi in multiple tubular carriers, including those dependent on p230GRIP. Some of the IL-10 is then delivered to recycling endosomes, where cytokine sorting may occur prior to its release. Another portion of the IL-10 is delivered to the cell surface in distinct vesicles colabeled for apoE. Thus, we show at least two post-Golgi pathways via which IL-10 is trafficked, ensuring its secretion from activated macrophages under different physiological conditions.

Regulation of vesicular transport by GTP-binding proteins.

Intracellular protein trafficking occurs in a series of transport vesicles. Vesicle trafficking is regulated both by heterotrimeric and monomeric GTP-binding proteins (G proteins). Recent studies have explored effector systems used by heterotrimeric G proteins and by monomeric ADP-ribosylation factor G proteins for regulation of vesicle budding. New members of the Rab monomeric G protein family have been identified in polarized cells and new evidence confirms the function of Rab proteins in vesicle targeting. From these data we can begin to reconstruct the signal transduction pathways that regulate intracellular transport.

Immunological studies of brain-specific antigens.

Central nervous system antigens have been investigated, using immunofluorescence, to demonstrate the presence of antisera raised against regional homogenates of neonatal and adult rat brain. Heterologous antisera raised in rabbits against regions of rat brain contained several antibodies demonstrated by immunofluorescence. Brain-specific activity directed against neuronal and glial elements was difficult to characterise. Regional and age-dependent specificity of the antisera was not demonstrated. Homologous antisera could not be raised against neonatal rat brain regions, but low titre antibodies to neurones were detected in rats over 21 days old. These animals did not develop experimental allergic encephalomyelitis (EAE). In other rats which developed EAE following a different immunisation schedule similar antibodies were seen. It is concluded that heterologous antisera raised against unpurified antigens have too wide a spectrum of antibody specificities for use as anatomic markers. Homologous antisera are difficult to obtain in sufficiently high titre. There is production of antibodies towards neurones in one form of EAE.

Distinct coated vesicles labeled for p200 bud from trans-Golgi network membranes.

Golgi-associated cytoplasmic proteins, such as the coatomer protein complex, are required for vesicle budding and trafficking. We have previously described a cytoplasmic phosphoprotein, p200, which binds dynamically and specifically to Golgi membranes. The p200 protein is dissociated from Golgi membranes in the presence of brefeldin A and it is induced to bind to Golgi membranes by activation of guanine nucleotide binding proteins (G proteins) with guanosine 5'-[gamma-thio]triphosphate or aluminum fluoride. To establish the role of p200 in vesicle budding, we localized membrane-bound p200 in intact cells and on isolated Golgi membranes. We show that p200 is preferentially associated with vesicles in the trans-Golgi network (TGN). Activation of G proteins induced budding and accumulation of small, coated vesicles from Golgi membranes and p200 was localized on the cytoplasmic surface of some of these vesicles. Using immunogold labeling we further demonstrate that p200 and beta-COP are localized on different populations of Golgi-derived vesicles. These data establish that p200 is involved in the budding and coating of a class of Goli vesicles that are likely to be derived from the TGN. The data also show that there are distinct populations of non-clathrin-coated vesicles budded from Golgi membranes, and vesicles labeled for either beta-COP or p200 may represent transport vesicles for separate steps of protein transport.

Protein trafficking and polarity in kidney epithelium: from cell biology to physiology.

The transepithelial movement of fluids, electrolytes, and larger molecules is achieved by the activity of a host of specialized transporting proteins, including enzymes, receptors, and channels, that are located on either the apical, basal, or lateral plasma membrane domains of epithelial cells. In the kidney as well as in all other organs, this remarkable polarity of epithelial cells depends on the selective insertion of newly synthesized and recycling proteins and lipids into distinct plasma membrane domains and on the maintenance and modulation of these specialized domains once they are established during epithelial development. This review addresses the mechanisms by which epithelial cells control the movement of membrane components within the cell to ensure that they are delivered to the correct target membrane. Among the topics discussed are targeting signals within membrane proteins, the role of the cytoskeleton and the tight junctional barrier in cell polarity, and the requirement for accessory proteins in the targeting process, including GTP-binding proteins, and proteins that are involved in vesicle docking and fusion events. The final part of the review is devoted uniquely to the polarized targeting of functionally defined proteins in various kidney cell types. In concluding, examples of how a breakdown in these trafficking pathways may be related to some disease states are presented.

Distribution and role of heterotrimeric G proteins in the secretory pathway of polarized epithelial cells.

The movement of newly synthesized proteins in the constitutive secretory pathway, from their site of synthesis in the endoplasmic reticulum to the cell surface or to intracellular destinations, requires an orderly sequence of transport steps between membrane-bound compartments. Until recently, the trafficking and secretion of proteins through this pathway was thought to occur as a relatively automatic, unregulated series of events. Recent studies show that protein trafficking in the constitutive secretory pathway requires GTP hydrolysis by families of GTP-binding proteins (G proteins), which at multiple steps potentially provide regulation and specificity for protein trafficking. Many monomeric G proteins are known to be localized and functional on membrane compartments in the constitutive secretory pathway. Now, members of the heterotrimeric G protein family have also been localized on intracellular membranes and compartments such as the Golgi complex. We have studied the localization and targeting of G alpha subunits to distinct membrane domains in polarized epithelial cells. The distribution of different G alpha subunits on very specific membrane domains in cultured epithelial cells and in epithelial cells of the kidney cortex, is highly suggestive of roles for these G proteins in intracellular trafficking pathways. One of these G protein subunits, G alpha i-3, was localized on Golgi membranes. Studies on LLC-PK1 cells overexpressing G alpha i-3 provided evidence for its functional role in regulating the transport of a constitutively secreted heparan sulfate proteoglycan through the Golgi complex. Inhibition or activation of heterotrimeric G proteins by pertussis toxin or by aluminium fluoride respectively, have provided further evidence for regulation of intracellular transport by pertussis toxin-sensitive G proteins.(ABSTRACT TRUNCATED AT 250 WORDS)

Heterogeneous localization of G protein alpha-subunits in rat kidney.

The Gs alpha and Gi alpha 1-3 subunits of GTP-binding proteins were localized in sections of rat kidney using antibodies against unique synthetic decapeptides from the different G alpha subunits. All of the G alpha subunits were found to have a polarized distribution on renal tubule epithelial cells, and staining was typically found on either basolateral or apical membranes in a given cell type. Gi alpha 1 was localized to the apical pole of both thick ascending limb cells and cells forming the papillary epithelium, Gi alpha 2 labeled the basolateral plasma membrane and the cytoplasm of collecting duct principal cells, and Gi alpha 3 was most abundant in the apical region of proximal tubule cells of the S1 segment, where it was concentrated in sub-brush-border invaginations. It was also found in the perinuclear Golgi complex in these cells. Gs alpha was heavily concentrated on the basolateral plasma membranes of thick ascending limb cells and both principal and intercalated cells of the collecting duct. Less intense subapical staining of G alpha s was also found in proximal tubule cells. The cells of the macula densa had a unique G protein distribution that was distinct from the surrounding cells of the thick ascending limb of Henle. Antibodies specific for the Gi alpha 1 and Gi alpha 3 subunits both stained intracellular vesicles clustered at the basal pole of the cell. A heterogeneous distribution of G alpha subunits was also found by Western blotting on isolated cortical membrane fractions.

Disruption of microtubules alters polarity of basement membrane proteoglycan secretion in epithelial cells.

Basement membrane proteins such as the heparan sulfate proteoglycan (HSPG) are secreted in a polarized fashion from the basolateral membrane of epithelial cells. We have used the microtubule-disrupting drug colchicine to study the role of the microtubule network in directing constitutive secretion to the basolateral membrane of LLC-PK1 renal epithelial cells. Microtubule depolymerization induced by colchicine resulted in fragmentation and redistribution of fluorescently labeled trans-Golgi membranes. Increased immunofluorescent staining of HSPG was associated with these dispersed Golgi cisternae. The biosynthetic processing of HSPG was not significantly altered by the loss of microtubules or by the dispersal of the Golgi elements. The most striking effect of microtubule disruption was the loss of polarity of HSPG secretion. Immunoprecipitation studies showed that HSPG was secreted from both apical and basolateral surfaces of LLC-PK1 cells treated with colchicine, and a similar result was found for the delivery of laminin, another basement membrane protein. In contrast, there was no change in the distribution of an integral basolateral membrane protein, Na(+)-K(+)-ATPase, following colchicine treatment. Our results provide the first demonstration that microtubules are involved in the constitutive trafficking of basolateral secretory proteins. These data also suggest that there may be an inherent difference in the targeting or delivery of membrane and secretory proteins to the basolateral cell surface.