A novel method for measuring protein expression at the cell surface

The membranes of eukaryotic cells are organized to form organelles with specific functions. Many of these organelles communicate by means of vesicular transport, resulting in a heterogeneous distribution of many membrane-associated proteins. To study processes such as cell surface expression and endocytosis of proteins it is essential to discriminate quantitatively between intracellular and extracellular localization. Several methods have been described in the literature that allow either selective labelling or removal of sur-face-disposed constituents. Protein labelling at the cell surface has for example been achieved by means of ¹²⁵I iodination (Hubbard and Cohn, 1972; Bretscher, 1989), cell surface biotinylation (Sargiacomo et al., 1989) and trini-trophenylation (Kaplan et al., 1979). These methods are usually not suitable for determining the extracellular versus intracellular distribution of proteins. However, some labels on surface-exposed proteins can be selectively removed or, alternatively, certain surface-exposed proteins may be removed or modified; intact cells may be treated with pro-teases such as trypsin or proteinase K (for examples, see Matlin and Simons, 1984, and Stoorvogel et al., 1989, respectively), neuraminidase (for example, see Krangel, 1987) or galactosyl transferase (Thilo, 1983) to modify sen-sitive externally disposed constituents. Unfortunately, not all surface-exposed proteins are sensitive to proteases, and methods that rely on modification of glycosyl moieties can be applied only for glycoproteins. Alternatively, specific cell surface proteins may be isolated by incubating intact cells in the presence of specific antibodies followed by cell lysis and immunoprecipitation (for example, see Schwartz and Rup, 1983). Cell-surface immunoprecipitations can be applied only when a specific antibody is available. Finally, labels that are linked by a disulphide group can be removed by membrane-impermeant reducing agents (Bretcher, 1989; Sargiacomo et al., 1989). Disulphide-linked labels can only be used for quantitative endocytosis studies, but not for quantitative determinations of biosynthetic arrival of proteins at the plasma membrane.

All the methods described above can be used only to measure endocytic and exocytic transport of subsets of membrane-associated proteins. To our knowledge no general method has been described that allows removal or modification of all surface-exposed proteins. Our objective was to develop a uniform method that would allow us to dis-criminate between an intracellular and an extracellular localization of any protein, to study both endocytosis and biosynthetic arrival of proteins at the plasma membrane.

In previous studies we accumulated horseradish peroxi-dase, which was conjugated to transferrin in endosomes, and employed its enzymic activity to polymerize 3,3′-diaminobenzidine (DAB) in endosomes with the conse-quent cross-linking of endosomal content proteins to the DAB polymer (Stoorvogel et al., 1989, 1991). Cross-linked proteins were separated from non-cross-linked proteins by SDS-PAGE. In this study we utilized the same principle selectively to cross-link cell surface-exposed proteins after coupling of horseradish peroxidase (HRP) to glycosyl moi-eties at the plasma membrane using concanavalin A (ConA). Using DAB cytochemistry, a coat of DAB poly-mer was formed around the cells, which encapsulated cell surface-exposed proteins. DAB-encapsulated proteins could easily be separated from freely soluble proteins by SDS-PAGE, thus allowing quantitative measurements of the intracellular/extracellular ratio of proteins.

All cell lines were cultured in a hydrated, 5% CO₂ atmosphere to 80% confluence on 60 mm tissue culture dishes. The human hepatoma cell line HepG2, clone A16 (Schwartz and Rup, 1983), was cultured as described earlier (Stoorvogel et al., 1988). Human intestinal epithelial-derived Caco-2 cells (Pinto et al., 1983) were grown in DME (4.5 g/l glucose), supplemented with 20% heat-inactivated FCS, non-essential amino acids, 100 i.u./ml penicillin, and 100 μg/ml streptomycin. Madin-Darby canine kidney cells (MDCK strain II) were cultured as described before (van Meer et al., 1987). Rabbit antiserum raised against the human ASGPR was a gift from Dr A. L. Schwartz (Washington University, St Louis, MO) and reacted specifically with the H1 subunit (Schwartz and Rup, 1983). Rabbit anti-human MPR serum was kindly provided by Dr K. von Figura (Georg-August University, Göttingen, Ger-many). Goat anti-serum to rabbit kidney β-Na⁺,K⁺-ATPase (Peters et al., 1984) was provided by Dr J. J. H. H. M. de Pont (Univer-sity of Nijmegen, Nijmegen, The Netherlands).

Cells grown on 6 cm culture dishes were surface ¹²⁵I-labelled as described previously (Stoorvogel et al., 1989). Alternatively, cells were metabolically labelled; cells were washed with, and incu-bated for 10 minutes in, Eagle’s minimum essential medium (MEM) lacking methionine, followed by a 10 minute pulse in the presence of 3.7 MBq/ml [³⁵S]methionine (Tran-S-label, 185 Mbq/ml, 40 TBq/mmol, ICN, CA). The cells were chased as indi-cated in MEM containing 100 μM unlabelled methionine at 37°C in a 5% CO₂ atmosphere. After radiolabelling the cells were washed 3× with PBS containing 1 mM EDTA at 0°C. Next, pro-tease-sensitive cell surface proteins were removed and the cells were detached from the culture dish during a 60 minute incuba-tion at 0°C on a rocker platform in 2 ml PBS, 1 mM EDTA, 0.5 mg/ml proteinase K (Boehringer Mannheim, Germany). The pro-tease activity was neutralized by adding 1 ml Ca²⁺- and Mg²⁺-free PBS supplemented with 1 mM EDTA and 1 mM phenyl-methylsulphonyl fluoride (PMSF). The detached cells were col-lected and washed twice with NT-buffer (130 mM NaCl, 1 mM CaCl₂, 1 mM MnCl₂, 0.1 mM PMSF, 20 mM Tris/NaOH, pH 7.8) by centrifugation at 300 g for 5 minutes at 0°C. Virtually, no cells were leaky, as determined by trypan blue exclusion. Next, the cells were resuspended in 1 ml NT-buffer and ConA (type V; Sigma, St Louis, MO) was added to a final concentration of 100 μg/ml from a 100× stock solution in NT. The cells were incubated at 4°C on a rocker platform for 30 minutes, after which 4 ml NT-buffer containing 100 μg/ml HRP (type VI; Sigma, St Louis, MO) was added. The incubation was continued for another 30 minutes. Care was taken to agitate the cells constantly, thus avoiding sed-imentation. The cell suspension was layered carefully on top of 5 ml 10% Percoll (Pharmacia LKB Biotechnology AB, Uppsala, Sweden) in NT-buffer and centrifuged for 5 minutes at 300 g, during which the cells migrated through the Percoll solution and were pelleted at the bottom of the tube. Non-cell-bound ConA and HRP were carefully removed together with the Percoll by aspiration, after which the cells were taken up in 2 ml NT containing 100 μg/ml DAB and resuspended carefully using a Pasteur pipet. Care was taken to avoid HRP-containing NT-buffer adhering to the top of the tube to mix with the sample. Samples of 1 ml were supplemented with H₂O₂ (final concentration 0.02%) and vortexed at the lowest setting, while H₂O₂ was omitted from control samples. After an incubation of 10 minutes at 0°C the cells were pelleted by centrifugation for 5 minutes at 300 g. The supernatant was removed and the cells were lysed in 0.5 ml Ca²⁺- and Mg²⁺-free PBS containing 1 mM EDTA, 0.1 mM PMSF, 1% Triton X-100, 0.1% SDS and 0.2% NaN₃. HRP is inactivated in the presence of NaN₃ and H₂O₂ (Ortiz de Montellano et al., 1988). Control cells, which were not treated with proteinase K, were kept on cul-ture dishes at 0°C during the entire procedure and were finally lysed in 1 ml lysis buffer. After 30 minutes in lysis buffer at 0°C the samples were vortexed and the nuclei removed by centrifugation (5 minutes at 10,000 g in an Eppendorf centrifuge at 4°C). Lysates were stored at −20°C.

When indicated, cells were lysed in a lysis buffer containing 1% Triton X-114 as detergent, replacing Triton X-100 and SDS. Phase separation and isolation of hydrophobic proteins (com-plexes) was performed principally according to the method of Bordier (1981). In short, nuclei and large DAB polymers were removed from the lysate by centrifugation for 5 minutes at 10,000 g at 4°C in an Eppendorf centrifuge. The supernatant was incu-bated for 5 minutes at 37°C and the detergent phase was pelleted for 1 minute at 10,000 g at room temperature. The extraction was repeated 3 times by mixing the detergent phase with 450 μl Ca²⁺- and Mg²⁺-free PBS, 1 mM EDTA, 0.1 mM PMSF at 0°C, fol-lowed by the procedure described above. Since small amounts of cells could have been lost during the procedure before the DAB treatment, quantitative analysis is performed best by comparing the amount of non-cross-linked protein from a sample that was incubated in the presence of DAB and H₂O₂ with total protein in a sample from the same cell suspension that was incubated in the presence of DAB only.

ASGPR, β-Na⁺,K⁺-ATPase and MPR serum were separately quantitatively immunoprecipitated from 50 μl samples of SDS-containing lysates, and analysed by SDS-PAGE (10% PAA) under reducing conditions, according to the method described previously (Stoorvogel et al., 1989). Gels containing [³⁵S]methionine-labelled proteins were soaked in Amplify (The Radiochemical Center, Amersham, United Kingdom) and fluorographed; ¹²⁵I-labelled proteins were detected on Kodak X-OMAT S film using an inten-sifier screen. Silver staining was performed using the Bio-Rad silver stain kit (Bio-Rad, Richmond, VA).

For the detection of ConA-binding proteins, transfer of proteins after SDS-PAGE to Immobilon PVDF membranes (Millipore, Bedford, MA, USA) was performed in 50 mM Tris, 380 mM glycine, 0.1% SDS, 20% methanol, using a Trans-blot cell (Bio-Rad Laboratories Inc. Brussels, Belgium) during 16 hours at 50 V. Low-range prestained SDS-PAGE standards were from Bio-Rad (Bio-Rad Laboratories Inc. Brussels, Belgium). The mem-branes were blocked in 150 mM NaCl, 20 mM Tris-HCl, pH 7.2, 1 mM CaCl₂, 1 mM MnCl₂, 0.1% Tween-80 for 2 hours at room temperature. Subsequently the membranes where incubated for 1 hour in blocking buffer containing 1 μg/ml ¹²⁵I-labelled ConA. When indicated, 100 mM α-methyl D-mannoside was present during the incubation to compete for specific binding of ¹²⁵I-ConA. Non-bound ¹²⁵I-ConA was removed by extensive washing in blocking buffer, after which the membranes were autoradi-ographed. To label ConA with ¹²⁵I, 200 μg ConA in 200 μl PBS was incubated for 15 minutes with 1 mCi ¹²⁵I (essentially carrier free; Amersham Crp., Arlington Heights, IL, USA) and five iodobeads (Pierce Chemical Co., Rockford, IL, USA). Free ¹²⁵I was removed by chromatography on a Sephadex G-25 column using 150 mM NaCl, 20 mM Tris-HCl, pH 7.2, 1 mM CaCl₂, 1 mM MnCl ₂ as an elution buffer. The specific activity was 6×10⁶ cpm/μg ¹²⁵I-ConA.

Cells were labelled with HRP as described above. After centrifu-gation through the Percoll solution, the cells were suspended in NT containing 2 mg/ml DAB. The pH of this DAB solution was readjusted to 7.6 using 1 M NaOH, after which the solution was filtered through a 0.22 μm Millex-GS filter (Millipore SA, Mol-sheim, France). H₂O₂ was added to a final concentration of 0.02%, and the cells were incubated for 10 minutes at 0°C. Next, the cells were washed twice with PBS by centrifugation for 3 minutes at 300 g, and finally resuspended in 2% glutaraldehyde in PBS. After fixation for 30 minutes at room temperature the cells were washed in 100% human serum, and embedded overnight by overlaying the cell pellet in serum with 2% glutaraldehyde. The cells were postfixed and stained in 2% OsO₄, dehydrated in ethanol and embedded in Epon 812 according to standard procedures. Thin sections were studied by transmission electron microscopy.

In previous studies (Stoorvogel et al., 1989, 1991; Rijnboutt et al., 1992) we used the peroxidase activity of endocytosed HRP or HRP conjugates to polymerize DAB in endosomes and or lysosomes with the consequent cross-linking of the protein contents to the DAB polymer. In the current study we used DAB cytochemistry to remove proteins quantitatively and selectively that are expressed at the cell surface from a cell lysate. Adherent cells may form plasma mem-brane areas that are sealed from the medium by tight junctions. To gain access to, and achieve formation of DAB polymer at the entire plasma membrane, cells were first brought in suspension by proteinase K treatment at 4°C in the presence of EDTA. Next, plasma membranes were labelled with ConA (specific for α-D-mannose/α-D-glucose; Nicolson, 1974), followed by incubation in the presence of HRP. In this way the glycoprotein HRP was linked to α-D-mannose and α-D-glucose moieties of glycoproteins at the plasma membrane using the multivalent lectin ConA as a linker (Geoghegan and Ackerman, 1977; Roth, 1983).

To visualize DAB polymer at the plasma membrane, HepG2 cells were fixed in glutaraldehyde and processed for electron microscopy (Fig. 1). DAB polymer was found as an electron-dense coat on the outer surface of the entire plasma membrane, including microvilli (Fig. 1, inset). Adherent cells that were not treated with proteinase K/EDTA lacked DAB polymer on large areas of plasma membrane facing the tissue culture dish, which were sealed off from the medium for ConA/HRP (not shown). Thus, proteinase K/EDTA treatment was an absolute requirement for complete accessibility to the plasma membrane of these adherent cells. Even after the proteinase K/EDTA treatment the cells seemed tightly linked to each other (Fig. 1). How-ever, this was mainly due to the presence of ConA, a pro-tein known to induce the formation of cell aggregates, since no intimate binding between cells was monitored when the cells were processed for electron microscopy immediately after the proteinase K/EDTA treatment (not shown). This finding led us to develop the present protocol (see Materi-als and Methods) in which the suspended cells were incu-bated in the presence of ConA and HRP subsequently, with-out a centrifugation step in between. No DAB polymer was detected in endoplasmic reticulum, Golgi, endosomes or lysosomes, indicating no significant DAB-polymerizing activity in the secretory and endocytic pathways of these cells (not shown).

To specifically label cell surface proteins, cells were ¹²⁵I-labelled on ice using lactoperoxidase-mediated iodination. After cell lysis in buffer containing Triton X-100 and SDS, ¹²⁵I-labelled proteins were analysed by SDS-PAGE and flu-orography (Fig. 2A, lane 1). Proteinase K treatment of intact HepG2 cells at 0°C removed many but not all ¹²⁵I-labelled proteins (Fig. 2A, lane 2). When the proteinase K treatment was followed by DAB cytochemistry all ¹²⁵I-label was removed (Fig. 2A, lane 3), indicating efficient removal of all cell surface proteins by this method. Coomassie Blue staining of the same lanes (Fig. 2B, lanes 1-3) revealed no detectable loss of intracellular proteins as a consequence of proteinase K treatment of intact cells or DAB cytochem-istry, indicating that the procedure was specific for cell sur-face-exposed proteins only. To determine the efficiency of the procedure for amphiphylic (integral membrane) pro-teins, a parallel experiment was performed in which the cells were lysed in a lysis buffer containing Triton X-114 instead of Triton X-100 and SDS. Triton X-114 extraction was performed to separate integral membrane proteins from non-hydrophobic proteins (Bordier, 1981). Most surface ¹²⁵I-labelled proteins did not fractionate into the detergent phase (Fig. 2A, lane 4 and 7), indicating either a weak asso-ciation with membrane proteins or an extracellular matrix origin. Many, but not all, ¹²⁵I-labelled hydrophobic and hydrophillic proteins and extracellular proteins were affected by the proteinase K treatment (Fig. 2A, lanes 5 and 8). Using DAB cytochemistry both classes of ¹²⁵I-labelled proteinase-K-resistant proteins were completely removed (Fig. 2A, lanes 6 and 9). No apparent loss of most the intra-cellular hydrophobic (Fig. 2C, lanes 1-3; silver stained) or hydrophillic proteins (Fig. 2B, lanes 4-9; Coomassie Blue stained) occurred. However, we noted that some specific integral membrane proteins localized to intracellular mem-branes were not quantitatively recovered from DAB-coated cells that were lysed in Triton X-114 or Triton X-100 only (not shown). We therefore favour the presence of both 1% Triton X-100 as well as 0.1% SDS in the lysis buffer (Fig. 2, lanes 1-3). In these conditions intracellular integral mem-brane proteins that were not embedded in DAB polymer were quantitatively solubilized (see below), whereas chro-matin was not released from the nuclei and could be removed by centrifugation.

To test whether surface proteins of other cell lines were also efficiently cross-linked by the procedure we applied the method to Caco-2 and MDCK cells (Fig. 3). Similar results to those for HepG2 cells were obtained, suggesting that the method is applicable to a wide variety of cell lines. Next, we tested the requirement for ConA, HRP, DAB and H₂O₂ for cross-linking cell surface proteins (Fig. 4). Incubation of ¹²⁵I-labelled, proteinase K treated cells in the presence of ConA resulted in the loss of some cell-associ-ated proteins (lanes 2, 3 and 6). When the cells were pelleted after incubation with ConA, some dissociated proteins could be recovered non-quantitatively from the medium (Fig. 4, lane 8). The nature of the association of these pro-teins with the cells remains unresolved, but may involve either specific or non-specific binding that is sensitive to competition with ConA for α-D-mannose/α-D-glucose binding. Incubation with both HRP and ConA did not cause removal of additional ¹²⁵I-labelled proteins (Fig. 4, lane 2). Both DAB and H₂O₂ were required for cross-linking of pro-tease-resistant ¹²⁵I-labelled proteins (Fig. 4, lanes 3 and 4). No cross-linking occurred when cells that were not treated with ConA and HRP were incubated in the presence of DAB/H₂O₂ (Fig. 4, lane 5). Con-A was an absolute require-ment for efficient HRP binding to the plasma membrane, since cross-linking of ¹²⁵I-labelled proteins was not observed when ConA was omitted (Fig. 4, lane 7). We con-clude that concanavalin A, HRP, DAB and H₂O₂ were all essential components for cross-linking of cell surface pro-teins. Optimal cross-linking efficiencies were achieved at 100 μg/ml DAB, 30-300 μg/ml ConA, and 30-300 μg/ml HRP (not shown).

To test whether intracellular ConA-binding proteins were not effected by the procedure, non-labelled HepG2 cells were treated with proteinase K/ConA/HRP and DAB cyto-chemistry was performed. Proteins from the lysates were separated by SDS-PAGE and transferred to Immobilon paper using a blotting apparatus. ConA-binding proteins were detected by incubating the blot with ¹²⁵I-ConA fol-lowed by autoradiography (Fig. 5, lanes 1 and 2). No ¹²⁵I-ConA-binding proteins were captured by the DAB com-plex, indicating that the major ConA-binding proteins are localized within the cell and are not effected by the proce-dure. To ensure that the ¹²⁵I-ConA binding was specific, a parallel blot of the same samples was incubated in the pres-ence of both ¹²⁵I-ConA and excess of α-methyl D-manno-side, a competitor for ConA binding. No signal was obtained in the presence of α-methyl D-mannoside (Fig. 5, lanes 3 and 4).

To test whether the method could be used to measure endo-cytic uptake of specific integral membrane proteins, HepG2 cells were ¹²⁵I-labelled at 0°C followed by an incubation at 37°C prior to proteinase K treatment and DAB cyto-chemistry. After cell lysis, cation-independent mannose 6-phosphate receptor/insulin-like growth factor II (MPR), asialoglycoprotein receptor (ASGPR), and the β-chain of Na⁺,K⁺-ATPase were immunoprecipitated and analysed by SDS-PAGE (Fig. 6). When the cells were not incubated at 37°C prior to the protease treatment, no endocytosis could occur and consequently all ¹²⁵I-labelled proteins were accessible to the protease. Both MPR (Fig. 6, lanes 1 and 2) and ASGPR (Fig. 6, lanes 6 and 7) were digested whereas β-Na⁺,K⁺-ATPase (Fig. 6, lanes 11 and 12) was resistant to proteinase K. However, all surface-exposed β-Na⁺,K⁺-ATPase was removed when DAB cytochemistry was performed, showing the value of the method for mea-suring cell surface exposure of an individual membrane protein. When the cells were incubated for 30 minutes at 37°C, significant pools of surface ¹²⁵I-labelled MPR (Fig. 6 lane 4) and ASGPR (Fig. 6, lane 9) became resistant to pro-teinase K, indicating endocytosis (see also Stoorvogel et al., 1989). These endosomal receptor pools were not affected when the protease treatment was followed by DAB cyto-chemistry (Fig. 6, lanes 5 and 10), indicating that endoso-mal proteins were not touched by the procedure. Only a small amount of ¹²⁵I-β-Na⁺,K⁺-ATPase became resistant to DAB-mediated cross-linking after incubating the cells at 37°C (Fig. 6, lanes 14 and 15), confirming that this protein is not endocytosed efficiently (Strous et al., 1988).

To determine if the method could be used to measure pro-tein arrival at the plasma membrane, we used β-Na⁺,K⁺-ATPase as a model protein (Fig. 7). HepG2 cells were pulse-labelled for 10 minutes with [³⁵S]methionine, and chased at 37°C for either 0, 1, 2 or 4 hours in the presence of excess non-labelled methionine. After proteinase K treat-ment and DAB cytochemistry ‘high mannose’ glycosylated precursors of β-Na⁺,K⁺-ATPase, present in the endoplas-mic reticulum, were not affected by the procedure, indi-cating that no endogenous peroxidase activity, which may be present in the ER, was able to cross-link proteins in the ER. After 1 hour of chase most of the newly synthesized ³⁵S-β-Na⁺,K⁺-ATPase was converted to a complex glyco-sylated form, part of which could be cross-linked to the DAB polymer, indicating arrival at the plasma membrane. After 4 hours of chase, almost all ³⁵S-β-Na⁺,K⁺-ATPase was converted to the ‘complex glycosylated’ form and had become accessible for cross-linking, demonstrating the applicability of the method.

DAB cytochemistry has been well established for the local-ization of HRP in fixed tissue (Graham and Karnovsky, 1966). This same principle has been used by Courtoy et al. (1984) to distinguish non-fixed HRP-containing endosomes from cellular organelles of a similar size and equilibrium density. Later, Ajioka and Kaplan (1987) and we (Stoorvogel et al., 1988, 1989; Rijnboutt et al., 1992) found that DAB-containing endosomes do not only increase in equilibrium density, but that their protein contents became detergent-insoluble as well. We have used DAB cytochemistry to quan-tify the presence of certain receptors (Stoorvogel et al., 1989, 1991), lysosomal enzymes (Rijnboutt et al., 1992) in the endocytic pathway, and secretory proteins in the trans-Golgi network (Strous et al., 1993), by making use of fluid-phase-endocytosed HRP or transferrin receptor-mediated uptake of HRP-transferrin conjugates. In this study we have developed a new application of DAB cytochemistry involving cross-linking of cell surface proteins, by which the degree of cell surface expression of all proteins measured can be deter-mined quantitatively. We achieved selective labelling of the plasma membrane with peroxidase activity by incubating ConA-coated cells on ice, thus preventing endocytic uptake, in the presence of HRP. In endosomes, using endocytosed HRP-transferrin conjugates, optimal cross-linking was achieved at 100-300 μg/ml DAB (Stoorvogel et al., 1992). Similar concentrations of DAB were required for optimal cross-linking at the plasma membrane (not shown), indicating that HRP-mediated protein cross-linking at the plasma membrane and in endosomes is based on the same principle. Although the biochemical nature of the cross-linking reac-tion has not been established, it may involve both protein encapsulation by the DAB polymer and the formation of covalent bonds between the DAB polymer and the proteins (Stoorvogel et al., 1992). Depending on the reaction con-ditions, HRP catalyses peroxidation, oxidation and hydroxylation reactions (for a review see Whitaker, 1972). Under the standard in vitro assay conditions, in which a phenolic sub-strate (such as 3,3′-diaminobenzidine) is used, peroxidation is the main reaction, resulting in the formation of indamine and/or phenazine as reaction products (for a review see Deimann, 1984). However, HRP can also catalyse the hydroxylation of a variety of aromatic compounds, including tyrosine, phenylalanine and sialic acid. These residues may partly substitute for DAB as reductant, thus leading to cross-linking.

Many mammalian cell types such as leucocytes, glandular acinar cells, epithelial cells, endothelial cells and other cell types express endogenous peroxidase activities (reviewed by Deimann, 1984). These peroxidases are con-fined to the nuclear envelope, the endoplasmic reticulum and other organelles in the secretory pathway. Some have a microbicidal function whereas others are involved in the biosynthesis of prostaglandin and thyroxin. Besides perox-idases, other enzymes are also capable of polymerizing DAB. Cytochrome oxidase in mitochondria has been reported to react strongly in its native state (reviewed by Lewis and Knight, 1977). Peroxisomal catalase is not reac-tive in its native state but oxidizes DAB after exposure to aldehyde fixatives or in alkaline buffers (Fahimi, 1969). Most studies to which the current method seems applica-ble will probably involve measuring endocytosis or biosyn-thetic arrival of proteins at the plasma membrane. It is there-fore important to establish that endogenous peroxidases that may be present in the vacuolar system do not interfere with the essay. We have shown that such peroxidases play no disturbing role in HepG2 cells (Figs 6, 7). However, con-trol experiments have to be performed when other cell lines, especially those mentioned above, are being used.

In conclusion, we have developed a new technique with which protein expression at the cell surface can be measured quantitatively. This method should be useful in studying the dynamics of total plasma membrane protein or, alternatively, the dynamic behaviour of proteins that cannot be monitored by other methods.

We thank René Scriwanek for preparing the figures, and Drs G. van Meer and G. J. Strous for comments on the manuscript. We are grateful to Dr K. von Figura (Georg-August University, Göttingen), Dr A. L. Schwartz (Washington University, St. Louis), and Dr J. J. H. H. M. de Pont (University of Nijmegen, Nijmegen) for their gifts of antisera to MPR, ASGPR and β-Na⁺,K⁺-ATPase, respectively. These investigations were sponsored by a senior investigatorship from the Royal Netherlands Academy of Arts and Sciences (W. Stoorvogel).