Lysosomes as well as a prelysosomal compartment rich in the mannose 6-phosphate receptor are clustered close to the Golgi apparatus in the perinuclear region of the microtubule organizing center in interphase human skin fibroblasts. The spatial organization of these organelles depends on an intact microtubule network. Depolymerization of the microtubules by treatment of cells with nocodazole leads to random scattering of Golgi elements, the prelysosomal compartment, and lysosomes throughout the cytoplasm. To test whether microtubules and the spatial organization of these organelles are important for efficient transport of lysosomal enzymes, the effect of microtubule depolymerization on the maturation of newly synthesized cathepsin D was studied. An up to fivefold inhibition of proteolytic maturation of cathepsin D was observed in drug-treated cells. This effect was due to a decreased rate of transport of cathepsin D from the Golgi apparatus to lysosomes. Depolymerization of microtubules did not inhibit transport of cathepsin D from the endoplasmic reticulum to the trans-Golgi network. Furthermore, synthesis of the phosphomannosyl marker present on cathepsin D was not affected by nocodazole. These results suggest that efficient transport of cathepsin D from the Golgi apparatus to a prelysosomal compartment and lysosomes is facilitated by microtubules and the spatial organization of these organelles.
The microtubule network plays a central role in the organization of the cytoplasm and in membrane traffic in animal cells (for reviews, see Schliwa, 1984; Vale, 1987). Movement of Golgi elements, endosomes and lysosomes along microtubules has been characterized in various systems (Herman and Albertini, 1984; Matteoni and Kreis, 1987; DeBrabander et al. 1988; Ho et al. 1989). In addition, the localization of most of the cytoplasmic organelles, including the endoplasmic reticulum, the Golgi apparatus and lysosomes, depends on microtubules (Kupfer et al. 1982; Terasaki et al. 1986; Matteoni and Kreis, 1987; Swanson et al. 1987; Heuser, 1989). The Golgi apparatus, a prelysosomal compartment rich in the cation-independent mannose 6-phosphate (Man-6-P) receptor and lysosomes are arranged in the perinuclear region of the microtubule organizing center (for reviews, see Thyberg and Moskalewski, 1985; Hopkins, 1986; Singer and Kupfer, 1986; Kreis, 1990; Griffiths et al. 1988, 1990). It has been speculated that this spatial apposition of the Golgi complex and endocytic organelles could facilitate intercompartmental transport (Pastan and Willingham, 1985; Kreis, 1990).
Vesicular traffic between the Golgi apparatus and the endocytic compartment occurs in both directions. Newly synthesized lysosomal enzymes are sorted via interaction with Man-6-P receptors in the Golgi apparatus and transported to a prelysosomal compartment before final delivery to lysosomes (for reviews, see von Figura and Hasilik, 1986; Kornfeld, 1987; Storrie, 1988; Komfeld and Mellman, 1989). On the other hand, recycling membrane proteins like the Man-6-P receptors or the transferrin receptor are transported from endosomes to the Golgi apparatus (Snider and Rogers, 1985; Duncan and Komfeld, 1988; Goda and Pfeffer, 1988). In addition LEP100, a lysosomal membrane glycoprotein, shuttles between lysosomes and endosomes (Lippincott-Schwartz and Fambrough, 1987).
The lysosomal hydrolase cathepsin D is synthesized in human skin fibroblasts as a precursor of Mr 53 000 carrying two N-linked oligosaccharides (Hasilik and Neufeld, 1980a; Hasilik and von Figura, 1981). During transport through the Golgi apparatus most of this precursor acquires Man-6-P residues (Hasilik and Neufeld, 1980b), which serve as the sorting signal for delivery to the lysosomes (Komfeld and Mellman, 1989). This precursor is then processed, most likely in a prelysosomal compartment, to an intermediate form of MT 47 000 (Gieselmann et al. 1983; Braulke et al. 1988). The precise characterization of this prelysosomal compartment and its location have not yet been established. It has been suggested, however, that it is identical to the late endosomal compartment that is rich in the cation-independent Man-6-P receptor (Griffiths et al. 1988, 1990; see also Geuze et al. 1988). Further processing of cathepsin D to the mature form (Mr 31000 and Mr 14000) occurs in lysosomes (Gieselmann et al. 1983,1985). We have used these different forms of cathepsin D as hallmarks of the lysosomal pathway to test the hypothesis that microtubules and the microtubule-dependent organization of the organelles involved in trafficking of lysosomal enzymes are essential for efficient intercompartmental transport.
MATERIALS AND METHODS
Cell culture and drug treatment of cells
Human skin fibroblasts were grown in α-MEM supplemented with 10% fetal calf serum, 2 mM glutamine and 1% penicillin and streptomycin under standard culture conditions. Unless other-wise indicated, microtubules were depolymerized by incubating cells for 1h at 37 °C in culture medium containing 10 μM nocodazole (Sigma Chemie GmbH, Deisenhofen, FRG). In some experiments we used lower concentrations of nocodazole, or 10 μM colchicine (Calbiochem GmbH, Frankfurt, FRG) for 1 h at 37 °C in culture medium.
Cells grown on glass coverslips were fixed: (1) in 3% formaldehyde and 0.02% glutaraldehyde in PBS for 15 min at room temperature and permeabilized with 0.1% Triton in PBS for 4min; or (2) for 4min in methanol at −20°C after pre-extraction with detergent to remove soluble protein as described (Kreis, 1987). The Golgi apparatus was labeled with rabbit antibodies against galactosyltransferase (Roth and Berger, 1982). A murine monoclonal antibody against a lysosomal glycoprotein, hLAMP-2 (Mane et al. 1989), was used to label the lysosomes. The cation-independent Man-6-P receptor was labeled using a rabbit serum against the bovine receptor (Griffiths et al. 1988). Microtubules were labeled with a murine monoclonal antibody (1A2) against a- tubulin (Kreis, 1987). Incubation with fluorescently labeled secondary antibodies and fluorescence microscopy was performed as described (Kreis, 1986).
Metabolic labeling of cells
Subconfluent cultures of human skin fibroblasts were incubated for 15 min in methionine-free culture medium and then labeled with 500μCiml−1 of L-[35S]methionine (Amersham Buehler, Braunschweig, FRG). After 10–20 min cells were washed three times in PBS and chased for various periods of time in medium containing 1 mM methionine. A 2 ml sample of chase medium was added to a 60 mm dish for immunoprecipitation of cathepsin D secreted into the medium. Cells were either scraped into PBS (for subcellular fractionation) or solubilized with 0.5 ml of 1% Triton in 50 mM NaCl, 50 mM Tris-HCl, 1mM EDTA, 1mgml−1 bovine serum albumin (BSA), pH 7.4, per 60 mm dish.
Immunoprecipitation of cathepsin D
Cell lysates, media, subcellular fractions and fractions from the Man-6-P receptor affinity column were adjusted to 1% Triton X-100,150 mM NaCl, 50 mM Tris-HCl, 1mMEDTA, 1 mg ml−1 BSA, pH 7.4. Samples were centrifuged at 15000# for 10 min at 4°C and 0.1vol. of protein A-Sepharose (Pharmacia LKB GmbH, Freiburg, FRG) was added to the supernatants. These suspensions were incubated on a shaker for 1 h at 4 °C and the Sepharose with unspecifically bound material was removed by centrifugation as described above, and then 0.01 vol. of a rabbit antiserum against cathepsin D (Ludwig et al. unpublished data) was added to the supernatants and these mixtures were incubated overnight at 4°C. Immunocomplexes were sedimented with 0.1vol. of protein A-Sepharose after incubation for 1 h at 4 °C by centrifugation at 5000g for 2 min at 4°C. Protein A-Sepharose pellets were washed seven times with 0.5% Triton X-100, 0.4 M NaCl, 50 mM Tris-HCl, pH 7.4, and once with 50 mM Tris-HCl, pH 7.4. Final pellets were boiled in 60 μl gel sample buffer and immunoprecipitated material was separated by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS-PAGE) in 9% polyacrylamide gels according to Laemmli (1970). Gels were incubated in 1M sodium salicylate for 20min (Chamberlain, 1979), dried and fluorographs were taken on preflashed KODAK X-OMAT AR films (Eastman Kodak, Rochester, NY) (3–21 days at −70°C). Radioactivity associated with the different forms of cathepsin D was quantified by densitométrie scanning of the fluorographs (Quick scan R&D, Desaga, Heidelberg, FRG).
Digestion of immunoprecipitated material with endoglycosidase H
Immunoprecipitated material bound to protein A-Sepharose was boiled with 30μl of 1% SDS, 200 mM sodium citrate, pH 5.5. Samples were cooled after 4 min and 120 μl of 200 mM sodium citrate, pH5.5, containing 2.5μgml−1 pepstatin was added. Samples were split into two aliquots, one of which received endoglycosidase H (50 munits ml−1 final concentration) (Seikagaku, Tokyo, Japan). Samples with and without endoglycosidase H were incubated for 18 h at 37 °C and then analyzed by SDS-PAGE.
Subcellular fractionation on Percoll gradients
Labeled fibroblasts (from 15 cm Petri dishes) were scraped into 5 ml of PBS using a rubber policeman. Cells were sedimented at 1000 #for 5 min at 4°C and resuspended in 1 ml of 250 mM sucrose, 2 mM EDTA, 3 mM imidazole, pH 7.2. Cells were homogenized by five strokes in a tight-fitting Dounce homogenizer (Wheaton Scientific, Millville, USA). A postnuclear supernatant was prepared by centrifugation of the homogenate at 800 g for 10 min at 4°C, loaded over 8 ml of 27% Percoll (Pharmacia LKB GmbH. Freiburg, FRG) in 250 mM sucrose, 3mM imidazole, 1mgmF BSA, pH 7.2, on a 0.5 ml cushion of 2 M sucrose and sedimented at 20000gmox for 2 h at 4 °C in a Ti50 rotor (Beckman Instruments, Inc., Fullerton, CA). Fractions of 0.6 ml were collected from the top and assayed for density (Pertoft et al. 1978), radioactivity, β-hexosaminidase activity (Pool et al. 1983) and galactosyltransferase activity (Strous and Berger, 1982).
Mannose 6-phosphate receptor affinity column
Cation-independent Man-6-P receptor purified from bovine liver was coupled to Affigel 10 (Bio-Rad Laboratories GmbH, Muenchen, FRG) as described elsewhere (Varki and Kornfeld, 1982; Hoflack et al. 1987). Cell lysate (0.5 ml) prepared as described above was supplemented with 5 mM β-glycerophosphate and applied to a 3 ml Man-6-P receptor affinity column equilibrated in CB (50 mM imidazole, 150 mM NaCl, 0.05% Triton X-100, 0.1 mg BSA ml−1, 5mM β-glycerophosphate, pH 7.0). The column was washed with CB after 2 h of incubation at 4 °C and fractions of 2 ml were collected. Fractions containing radioactivity were pooled. Lysosomal enzymes bound to the receptor were eluted with 5 mM Man-6-P in CB and fractions containing radioactivity were pooled.
Role of microtubules in the spatial organization of endocytic organelles and the Golgi apparatus
The Golgi apparatus, the prelysosomal compartment rich in the cation-independent Man-6-P receptor, and lysosomes were labeled in human skin fibroblasts with antibodies against galactosyltransferase (Roth and Berger, 1982), the cation-independent Man-6-P receptor (Griffiths et al. 1988), and hLAMP2 (Mane et al. 1989), respectively. It has been proposed that the compartment containing the bulk of the Man-6-P receptor is a prelysosomal compartment where lysosomal enzymes are released from the receptor (Griffiths et al. 1988). Cathepsin D colocalized with hLAMP2 by immunofluorescence (not shown). All three compartments were clustered in the perinuclear region of the microtubule organizing center (Fig. 1). The three compartments were distinct, but the Golgi apparatus and the prelysosomal compartment rich in the Man-6-P receptor were closely apposed (see also Griffiths et al. 1988, 1990).
Depolymerization of microtubules by treatment of the fibroblasts for 1h at 37 °C with 10 μM nocodazole resulted in random scattering of the Golgi apparatus, the prelysosomal compartment rich in the Man-6-P receptor, and lysosomes throughout the cytoplasm (Fig. 1). Thus, the close apposition of these organelles in the perinuclear region requires an intact microtubule network.
Treatment of cells with either 10 μM. nocodazole or 10 μM colchicine during 1 h at 37 °C led to the complete depolymerization of microtubules (Fig. 2C,E) and dispersal of the Golgi apparatus (Fig. 2D,F). Treatment of cells with 0.3 μM nocodazole for up to 4h did not completely depolymerize microtubules (see also Kreis, 1987) but led to some dispersal of Golgi elements (Fig. 2G,H). Scattering of the prelysosomal compartment rich in the Man-6-P receptor, and lysosomes, was comparable to dispersal of the Golgi apparatus. On the other hand, 0.1 μM nocodazole induced minor changes in the microtubule network (compare Fig. 2A and I) and the Golgi apparatus remained apparently normal (compare Fig. 2B and J). Treatment of cells with 10 μM nocodazole for 1 h at 37 °C had no significant effect on the morphology of the endoplasmic reticulum visualized by immunofluorescence labeling with antibodies against protein disulfide isomerase (data not shown).
Effect of depolymerization of microtubules on the maturation of cathepsin D
Three major forms of cathepsin D can be distinguished during maturation of this lysosomal hydrolase. A precursor form of Mr53000, an intermediate form of Afr47 000, and the mature heterodimeric form consisting of a Mr 31000 heavy and a Mr 14 000 light chain are predominantly present in the endoplasmic reticulum and the Golgi apparatus, a prelysosomal compartment, and the lysosomes, respectively (Gieselmann et al. 1983; Braulke et al. 1988). A small fraction of the 53000Mr precursor form (<10%) is secreted into the medium of cultured cells (Hasilik and Neufeld, 1980a). In this study ‘cathepsin D’ refers to all three forms of the protein.
Maturation of cathepsin D was followed by metabolic labeling of human skin fibroblasts with [35S]methionine. Cells were pulsed with 500μCiml−1 for 15 min at 37 °C and chased for various periods of time. Labeled cathepsin D was immunoprecipitated both from the cell lysates and the corresponding cell culture medium with specific antibodies and analyzed by polyacrylamide gel electrophoresis. Processing of cathepsin D from precursor to intermediate and mature forms was followed in control and nocodazole-treated cells (Fig. 3), as well as in cells treated with colchicine or low concentrations of nocodazole (Fig. 4). The rate of processing of cathepsin D to the intermediate and mature forms was reduced up to fivefold in cells without microtubules (Fig. 3). A total of 54% of labeled cathepsin D was present in the precursor form after 2.5 h of chase in nocodazole-treated cells, whereas only 9% was found in the control cells (Fig. 3B). The intermediate form reached its highest concentration after 90-120 min of chase in control cells, but only after 3h in the nocodazole-treated cells (Fig. 3B). After 2.5 h of chase 71% of total cathepsin D has been processed to the mature form (only the Afr31 000 form was analyzed here) in control cells, compared to 13% in nocodazole-treated cells (Fig. 3B). Complete processing to the mature form occurred also in nocodazole-treated cells after long periods of chase (9 h; Fig. 3).
We observed no differences in the inhibition of processing of cathepsin D after 2 and 3h of chase when microtubules were completely depolymerized with either nocodazole or colchicine (Fig. 4). Treatment of cells with low doses of nocodazole (0.1 μM, see also Fig. 2), however, had no effect on the processing of cathepsin D. These results suggest that reduced rates of maturation of cathepsin D are probably due to depolymerization of microtubules, rather than non-specific effects of nocodazole.
About 5% of the precursor was secreted in both control and nocodazole-treated cells after 9 h of chase. Secretion of cathepsin D, however, was significantly delayed in the nocodazole-treated cells (Fig. 3). Similar effects of microtubule depolymerization on the kinetics of transport of vesicular stomatitis virus glycoprotein from the trans-Golgi network to the cell surface have been measured (W. C. Ho and T. E. Kreis, unpublished results).
Analysis of transport of cathepsin D to lysosomes by cell fractionation
To test whether the inhibitory effect of nocodazole on processing of cathepsin D was due to inhibition of transport, fibroblasts labeled with [36S]methionine were homogenized and fractionated by centrifugation through a Percoll gradient. Control or nocodazole-treated cells were pulsed with [35S]methionine for 15 min and subsequently chased for 30 min, 1 h and 6 h, respectively, before homogenization. Postnuclear supernatants were loaded on top of a 27% Percoll gradient and centrifuged as described in Materials and methods. Fractions were collected and analyzed for radioactivity, β-hexosaminidase and galac-tosyltransférase activity (Fig. 5). The light fractions (2-6) containing the bulk of the Golgi apparatus and the prelysosomal compartment rich in the Man-6-P receptor (Griffiths et al. 1990), the dense fractions (12-16) enriched in lysosomes, and the intermediate fractions (7–11), were pooled as indicated (Fig. 5A). Cathepsin D was immunoprecipitated from these three pools (Fig. 5B).
In control cells, cathepsin D immunoprecipitated from the pool of the light fractions was 96%, 70% and 4% of total cathepsin D after periods of chase of 30 min, 1 h and 6h, respectively (Fig. 5B). Cathepsin D immunoprecipitated from the pooled heavy fractions was 2%, 27% and 95%, after 30 min, 1 h and 6 h, respectively (Fig. 5B). Only precursor and intermediate forms of cathepsin D were found in the pooled light fractions, whereas the pooled heavy fractions contained only the intermediate and mature forms of cathepsin D. Thus, we find that the proteolytic cleavage steps during maturation of cathepsin D and its transport to the fraction containing the lysosomes are in good agreement with previous findings reported by Gieselmann et al. (1983). Virtually no difference in the amounts of cathepsin D immunoprecipitated from the light and heavy fractions could be detected in control and nocodazole-treated cells after 30 min and 6h of chase (Fig. 5B). This result suggests that sorting and targeting of cathepsin D is unaffected by the nocodazole treatment of the cells. At the intermediate chase period (1 h), however, only 5% of the total cathepsin D could be immunoprecipitated from the pooled heavy fractions in the nocodazole-treated cells (Fig. 5B). Therefore, transport of cathepsin D to lysosomes was significantly delayed in the absence of microtubules, when Golgi elements, prelysosomal compartment and lysosomes were scattered throughout the cytoplasm. In the first 60 min of chase this delay was equivalent to an inhibition of about fivefold.
Depolymerization of microtubules does not inhibit transport of cathepsin D to the Golgi apparatus
A fraction of the two oligosaccharide side-chains of the cathepsin D precursor (Mr 53 000) is converted in the Golgi apparatus to endoglycosidase H-resistant forms (Hasilik and von Figura, 1981). Digestion of metabolically labeled cathepsin D precursor with endoglycosidase H was used to measure its transport to the Golgi apparatus (Fig. 6). Control cells and cells treated with nocodazole were labeled for 10 min with [:35S]methionine and chased for up to 60 min. Cathepsin D was immunoprecipitated and half of the precipitated material was digested with endoglycosidase H as described in Materials and methods. About 20% of the precursor acquired (partial or complete) endoglycosidase H resistance after 60 min of chase (Fig. 6). No significant difference in the rate and extent of acquisition of endoglycosidase H resistance could be detected between control and nocodazole-treated cells (Fig. 6). This suggests that transport of cathepsin D from the endoplasmic reticulum to the Golgi apparatus is not affected by nocodazole.
Synthesis of the phosphomannosyl marker is not affected by treatment of cells with nocodazole
Binding of lysosomal enzymes to Man-6-P receptors allows the segregation in the frans-Golgi network of these proteins routed to lysosomes from those destined to be secreted (for reviews, see von Figura and Hasilik, 1986; Komfeld, 1987; Kornfeld and Mellman, 1989). Cathepsin D acquires the Man-6-P recognition signal in a two-step process, which occurs in a pre-Golgi compartment and the cis-cistemae of the Golgi apparatus (for a review, see Komfeld and Mellman, 1989). To confirm that phosphorylation of mannose residues was not affected by the nocodazole treatment of cells, binding of cathepsin D to the cation-independent Man-6-P receptor was analyzed (Fig. 7). Fibroblasts with intact or depolymerized microtubules were pulsed for 10 min with [35S]methionine and chased for 45 min to allow transport of the cathepsin D precursor to the Golgi apparatus. Cell lysates were prepared and fractionated on a Man-6-P receptor affinity column as described in Materials and methods. No significant difference in binding of cathepsin D precursor to the Man-6-P receptor affinity column was measured with lysates from control or nocodazole-treated cells (58% versus 57%; Fig. 7). Since binding of cathepsin D to the Man-6-P receptor was also not modified by the nocodazole treatment, we concluded that inhibition of transport to lysosomes was not due to alterations in the state of phosphorylation of cathepsin D by the drug. Therefore, nocodazole does not affect the sorting of lysosomal enzymes by the Man-6-P receptors. This is also consistent with the pulse-chase experiment showing that secretion of cathepsin D was not increased in nocodazole-treated cells (Fig. 3).
Microtubule-depolymerizing agents have been instrumental in analyzing the involvement of microtubules in the processes of intracellular transport (see, for example, Freed and Lebowitz, 1970; Herman and Albertini, 1984; Phaire-Washington et al. 1980; DeBrabander et al. 1988; Ho et al. 1989; Kreis, 1990). They not only disrupt the tracks along which movement of vesicles proceeds, but they also have profound effects on the spatial organization of cytoplasmic organelles; for example, both elements of the Golgi apparatus and lysosomes are scattered randomly throughout the cytoplasm in the absence of an intact microtubule network (Kupfer et al. 1982; Matteoni and Kreis, 1987; Swanson et al. 1987; Heuser, 1989). We have used the maturation of the lysosomal hydrolase, cathepsin D, as a model in which to study whether microtubules and the microtubule-dependent spatial organization of the Golgi apparatus and lysosomes are important for efficient transport between these two organelles.
The rate of transport of cathepsin D from the Golgi apparatus to lysosomes was reduced up to fivefold in the absence of microtubules as monitored by the proteolytic processing of cathepsin D and delivery of cathepsin D from the Golgi apparatus (light membrane fraction) to lysosomes (heavy membrane fraction) followed by cell fractionation. Comparison of the kinetics of processing of cathepsin D in control and nocodazole-treated cells suggests that: (1) transport of cathepsin D from the trans-Golgi network to the compartment where processing to the intermediate form occurs was inhibited significantly in nocodazole-treated cells; and (2) transport from this prelysosomal compartment to lysosomes was inhibited to a smaller extent. Neither transport from the endoplasmic reticulum to the Golgi apparatus, nor sorting of cathepsin D was affected by nocodazole (see Figs 5 and 6). Furthermore, it has been shown previously that formation of coated vesicles at the plasma membrane remains unaffected by the treatment of cells with nocodazole (Gruenberg et al. 1989).
Inhibition of transport of cathepsin D by depolymerization of microtubules could be due to the disappearance of the microtubule tracks along which movement of vesicles might occur, or to the disruption of the spatial arrangement of the organelles involved. It is assumed, however, that microtubules are not involved in short-range vesicular transport like that from the endoplasmic reticulum to the Golgi apparatus, in between Golgi cisternae, or from the plasma membrane to early endosomes (see, for example, Kelly, 1990). We have shown here that, normally, the Golgi apparatus, the prelysosomal compartment that is rich in the Man-6-P receptor, and the clustered lysosomes are closely apposed. Depolymerization of microtubules results in random dispersal of these organelles throughout the cytoplasm. Thus, the inhibitory effect of nocodazole and colchicine on transport of cathepsin D to lysosomes may to a large extent be due to the dramatic changes in the spatial organization of these organelles. It is conceivable that the nearest neighbour distances between the organelles are increased upon scattering, since these same organelles are randomly dispersed in a much greater volume than initially. In contrast, the endoplasmic reticulum remains extended throughout the cytoplasm and its morphology is not significantly changed by treatment with nocodazole for short periods of time. Therefore, distances between the endoplasmic reticulum and elements of the Golgi apparatus should, in principle, not alter very much, which is consistent with our observation that nocodazole does not affect transport from the endoplasmic reticulum to the Golgi apparatus. Although we cannot exclude the possibility that vesicular transport from the Golgi apparatus to lysosomes occurs along microtubules, we consider it more likely that it is the microtubule-dependent apposition of Golgi apparatus, prelysosomal compartment that is rich in the Man-6-P receptor, and lysosomes, in the perinuclear region that ensures efficient transport of newly synthesized cathepsin D to lysosomes.
Evidence has been presented recently that microtubule-independent factors may be involved in the positioning of the Golgi apparatus (Turner and Tartakoff, 1989; Ho and Kreis, unpublished observations), and perhaps also the lysosomes (Matteoni and Kreis, 1987), in the region of the microtubule organizing center. Further work will be required to characterize the molecules that are involved in the spatial organization of these organelles.
We thank Drs Thomas August, Eric Berger, Steve Fuller and Stuart Komfeld for their generous gift of antibodies, Rodrigo Bravo for providing the human skin fibroblasts, and Gareth Griffiths for critically reading the manuscript.