Developmental changes in the subcellular distribution of the 43K (v₁) polypeptides in Torpedo marmorata electrocyte: support for a role in acetylcholine receptor stabilization

Along with the nicotinic acetylcholine receptor (AChR) (rev. Bloch and Pumplin, 1988; Changeux et al. 1990), a number of cytoskeletal proteins are concentrated at the neuromuscular and Torpedo electrocyte postsynaptic membranes (Sobel et al. 1977; Hall et al. 1981; Bloch and Hall, 1983; Burden, 1985; Woodruff et al. 1987; Carr et al. 1989a; Cartaud et al. 1989; Jasmin et al. 1990). The 43×10³M_r protein (43K) is the only protein of the neuromuscular junction (NMJ) (Froehner et al. 1981) that is present as a major protein component associated with the AChR at the postsynaptic membrane of the electrocyte (Sobel et al. 1977). While the AChR is a transmembrane protein (Wennogle and Changeux, 1980; rev. Popot and Changeux, 1984), 43K protein is a peripheral protein (Neubig et al. 1979) specifically colocalized with the AChR at the cytoplasmic face of the cell membrane (Nghiêm et al. 1983; Sealok et al. 1984). Alkaline extraction of the 43K protein leads to an increase in AChR mobility within the membrane plane (Rousselet et al. 1979; Barrantes et al. 1980; Lo et al. 1980; Cartaud et al. 1981; Rousselet et al. 1982; Block and Froehner, 1987), and in its sensitivity to both thermal (Saitoh et al. 1979) and enzymatic degradation (Klymkowsky et al. 1980). In myotube cultures, clustering of AChR and its association with 43K protein are simultaneously detectable (Burden, 1985; Peng and Froehner, 1985; Bloch and Froehner, 1987). Therefore, the 43K protein has been postulated to be involved in the formation, stabilization and maintenance of AChR clusters. In vivo, AChR aggregates show a remarkable stability persisting up to several weeks after denervation (Bourgeois et al. 1978; Loring and Salpeter, 1980). The 43K cDNA and proteins present in the electrocyte share a high degree of homology with their mammalian (rodent) and Xenopus NMJ counterparts (Frail et al. 1988; Froehner, 1989). Thus, the Torpedo AChR-rich electrocyte represents a good model system of the NMJ for deciphering the mechanisms responsible for the formation and the maintenance of the postsynaptic machinery.

In the immature embryonic electrocyte, before any detectable innervation (45 mm-stage), AChR can cluster on the ventral face of the cell (Witzemann et al. 1983), independently of 43K protein which is mainly found dispersed in the cytoplasm (Kordeli et al. 1989); the 43K protein is codistributed with the AChR on the postsynaptic face of the cell only at a later stage of development (80 mm embryo) (Kordeli et al. 1989). These results, confirmed by La Rochelle et al. (1990), are consistent with the reported aggregation of AChR expressed in fibroblasts stably transfected with AChR subunit cDNAs and cultured in the presence of clustering agents derived from neural tissue or of basal lamina from muscle or Torpedo electric organ (Claudio et al. 1989, 1991). The 43K protein accordingly might not be the direct inducer of the AChR clustering (see however, Froehner et al. 1990; La Rochelle et al. 1990; Phillips et al. 1991). Its putative role in stabilizing and maintaining AChR clusters still remains. As an attempt to gain insight into these problems, we have investigated the compartmentalization of 43K protein during maturation of the electrocyte: two forms of 43K protein have consistently been detected. We have analyzed the two 43K protein forms and compared their distributions, as well as that of the AChR, in subcellular fractions of the electrocyte in relation to maturation of the postsynaptic membrane.

Torpedo marmorata embryos were collected from the Institut de Biologie Marine, (Arcachon, France) during the Torpedo gestation season. The developmental stages of the embryos were characterized by their body length and electric organs were immediately dissected out and frozen in liquid nitrogen. The cytosol and membrane-enriched fractions were prepared according to the following scheme. Electric organs were homogenized in cold 150mM Tris-HCl pH7.5, 3mM EDTA, ImM EGTA, 100μM phenyl-methylsulfonylfluoride containing 5 units of aprotinin, 5 μ g of pepstatin A, 10 of antipain and 10 μ g of leupeptin per ml. The homogenates were centrifuged 30min, 30000g to give a SI supernatant and a Pl pellet fraction. The cytosol fraction was obtained by further centrifugation of the SI supernatant two times (30 min, 140000g). The Pl pellet was washed three times (30 min, 37 000g) and used as the membrane-rich fraction. The whole procedure was performed at 4 °C.

AChR-rich membrane fractions from adult tissues were prepared according to Sobel et al. (1980) and Saitoh et al. (1980b) in presence of the above antiproteases. The AChR concentrations were determined with the a-bungarotoxin filter method according to Heidmann et al. (1983). The protein concentration was determined according to Lowry et al. (1951). The 43K-enriched extract (Sil) was prepared from adult Torpedo AChR-rich membrane vesicles at 4 °C according to Neubig et al. (1979).

The cytosol and membrane-rich fractions from E80mm embryos were used for immunoprecipitation of the corresponding 43K polypeptides. Prior to the immunoprecipitation step, the membrane-enriched fractions were subjected to a Bolton Hunter ¹²⁵I iodination (1973) and the cytosol fraction to a ¹⁴C reductive methylation with ¹⁴C-formaldehyde (C.E.A, France) (Jentoft and Dearborn, 1979). Antibodies directed against the SDS denatured and purified 43K protein from a Torpedo marmorata extract were developed in hyperimmunized rabbits. The specificity of the immune serum was tested by micro-immunoblot (Nghiêm, 1988) over the Sil extract and/or the AChR-rich membrane fraction. Selected immune serum was further characterized by a 2D immunoblot on the AChR-rich membrane fraction (Nghiêm et al. 1983). The antibodies from such an immune serum were first adsorbed on an immobilized protein A Sepharose 2B (Pharmacia) suspension. The resulting suspension was then incubated with SDS-treated cytosol and membrane-enriched fractions of Torpedo mormorata E80mm embryo in a SDS, Triton X100, Tris buffer saline pH7.4. After centrifugation, the pellets were washed with the same buffer. Adsorbed proteins were extracted with a SDS-PAGE running buffer for 2 min, 100°C, then subjected to SDS 10% polyacrylamide gel electrophoresis (Laemmli, 1970). Coomassie Blue was used to stain the resulting protein bands and fluorography on a ßmax film (Amersham) to detect the labeled polypeptides immunoprecipitated from the cytosol and membrane-rich fractions respectively.

Double fluorescence experiments on cryostat sections of adult T. marmorata electric tissue were carried out essentially as described in Kordeli et al. (1986).

Peptides corresponding to the N and C termini were chosen from the sequences of 43K polypeptides of Torpedo californica electrocyte based on the 2 cDNA sequences reported by Frail et al. (1987). The C7- and C7_i-terminal peptide sequences are present only on the protein encoded by the long mRNA transcript while the C1-terminal sequence is present as the next inner peptide on the same protein or as the C-terminal peptide of the 43K protein encoded by the short mRNA transcript. The N-terminal sequence is the same for both proteins. The N, C7, C7i peptides were synthesized by Neogene (Strasbourg, France), the 340–354 peptide by Appligene (Illkirch, France) and the Cl peptide by the Laboratoire de Microchimie, Hôpital Gustave Roussy (Villejuif, France). Antibodies were developed in rabbits. 200 μ g of peptide as is or coupled to chick ovalbumin or Keyhole Limpet Haemocyanin (Sigma) were injected into rabbits with complete Freund adjuvant. Boosting injections were given with incomplete Freund adjuvant 4 or 5 weeks later. One week after each boosting injection, animals were bled and their sera tested for the presence of anti-43K antibodies in an ELISA assay with the 43K-enriched S11 extract and sheep anti-rabbit IgG coupled to horseradish peroxidase (Biosys, France).

The 43K protein was quantified by inhibition experiments applied to the ELISA techniques. Wells of a 96-well microtiter plate were coated with a 43K-rich alkaline extract (S11) of the AChR-rich membrane from adult Torpedo marmorata electrocyte. After saturation of the remaining protein-binding sites with bovine serum albumin (BSA), wells were reacted with serial dilutions of an anti-C7i (and anti-C7) antiserum of which the specificity has been demonstrated. Horseradish-peroxidase-labeled anti-rabbit immunoglobulins were used as secondary antibodies and orthophenylene diamine as substrate. Readings of the plates at 490 nm–630 nm permit quantification of the immunoreaction and design of the primary serum dilutions for inhibition assay. Conditions of assay were chosen to make the assay linear with the serum dilutions. The antigen is maintained constant. After preincubation with increasing amounts of SDS-pretreated preparations to test, the anti-C7i antiserum is probed with the 43K-rich S11 extract. The 100% control is obtained with a preincubation of the same dilution of the anti-C7i antiserum with buffer. The 0 % control (0D₀) is performed with a prior incubation with an excess of the S11 extract. The inhibition percentage I=(0D₁₀₀−0D_i) × 100)/0D₁₀₀−0D₀ is plotted versus corresponding volumes of preparations. Within the linear range of 40 to 60% inhibition, those giving the same percentage of inhibition are considered to possess the same amount of 43K protein [Affinities of the two forms of 43K protein for the C-terminal antibody were assumed to be similar on the basis of parallel intensities of the immunoblots and of the 43K protein bands in Ponceau red (protein staining)]. Relative concentrations in 43K proteins are then deduced. Absolute values of 43K concentrations are obtained when data are normalized to the 43K content of the 43K-rich S11 extract. The concentration of 43K present in the S11 is itself determined by densitometric scanning of the 43K band after separation of the extract on a SDS gel, a range of BSA concentrations being used as standards. A certain uncertainty is inherent in this method of protein estimation and the derived absolute values for the content in 43K protein must be taken with care. However, this restriction does not stand when ratios of concentrations of 43K proteins in two preparations are considered, as the values have been obtained through a standardization with the same S11 extract.

We have previously shown that both cytosol-enriched and particulate fractions (centrifugation at 30000g 30 min) of Torpedo marmorata electrocytes at three developmental (45 mm embryo (E45mm), 80 mm embryo (E80mm) and adult) stages are stained with monospecific anti-43K protein antibodies (Fig. 4 in Kordeli et al. 1989). The faint staining at the 43 K polypeptide region in the cytosolic fraction of adult electrocyte, compared with the strong staining observed in the membrane-enriched fraction raises the issue of possible contamination of the cytosolic fraction by AChR- and 43K-rich membrane vesicles during experimental manipulation. This led us to probe for the presence of AChR (^L25I-α-bungarotoxin probe) and 43K protein (micro-immunoblot with anti-43K antibodies) molecules after a higher speed centrifugation (140 000 g 30 min). While AChR are detected only in the membrane fractions, 43K polypeptides are found in both cytosol and membrane fractions of all three stages (data not shown), confirming the coexistence of soluble and membrane-bound 43K proteins in the electrocyte.

To characterize further the two 43K polypeptide pools, immunoprecipitation was performed using a mono-specific anti-43K protein antibody and radioactive-labelled electrocyte extracts. Fig. 1 shows that the immunoprecipitated proteins stained with Coomassie Blue at 43×10³ molecular mass are radioactive and thus correspond to the 43K proteins immunoprecipitated from the extracts. The two immunoprecipitated 43K polypeptide forms present nearly the same migration profile in SDS gels, the cytolosic one migrating slightly more slowly.

cDNAs encoding two 43K proteins differing by the presence of an additional 23 amino acid sequence at the C terminus were isolated by Frail et al. (1987). We, therefore, were interested in analyzing the potential correlation of the two 43K protein isoforms to one or the other mRNA transcripts. Rabbit antibodies were developed against the short and the long C-terminal peptide sequences of the 43K proteins. Specific antibodies capable of recognizing the 43K proteins were selected in an ELISA assay according to the following criteria: no inhibition with irrelevant peptides and 100% inhibition with the relevant one. The selected antibodies were further tested for their specificity in immunoblot. They were used to probe for the presence of related determinants in the cytosol and membrane-associated 43K polypeptides, respectively. Results shown in Fig. 2 indicate that the two 43K polypeptides can be recognized by different antisera directed against the N terminus, an inner peptide sequence, the short C terminus as well as the longer C-terminal peptide sequences of the 43K proteins. The two forms of 43K polypeptides thus did not result from a translation of different mRNA transcripts. Staining of the 43K bands was, however, lower in the case of the cytosol-associated 43K proteins, especially with anti-N-terminal peptide antibodies. Upon preabsorption with relevant peptides, the immunoreactivity observed at the 43K polypeptide band drastically decreased or disappeared, demonstrating the specificity of the 43K polypeptide staining.

Immunocytolocalization of the 43K proteins performed on cryostat sections of the adult electrocyte (Fig. 3) shows that antibodies directed against the last 11 amino acid (C-terminal sequence) of the long 43K polypeptide decorated mainly the postsynaptic membrane but also, although faintly, the cytoplasm of the electrocyte. Both stainings were specific since they disappeared after preabsorption of the anti-43K C-terminal antibodies with the relevant C-terminal peptide.

To address the issue of the putative role of 43K protein in the stabilization and/or immobilization of AChR clusters, we followed the evolution of 43K polypeptides with maturation of the electrocyte and examined their partition between the cytosol and membrane-rich fractions during development. Quantitation of AChR molecules was performed in parallel on the same samples.

Since both 43K polypeptide forms reacted strongly with anti-C-terminal peptide sequence antibodies, we used a specific anti-C terminus antibody to quantify the contents of 43K polypeptides in the two pools upon maturation of the electrocyte. Three stages were examined: the physiologically immature E45mm (preinnervated) and E80mm (postinnervated) stages (Fox and Richardson, 1979; Krenz et al. 1980) and the physiologically mature adult stage. Accurate quantification is difficult to obtain from immunoblot assays; we thus developed a sensitive assay based on specific inhibition of binding of the anti-43K C-terminus antibodies in ELISA to measure the amount of 43K protein, both in cytosol and membrane-rich fractions of the electrocyte. Assuming that the affinities of the 43K proteins for the antipeptide antibodies are the same for both pools and at all stages tested, it follows that equivalent amounts of 43K proteins are required to provide the same percentage of inhibition (see Material and methods).

Results are shown in Fig. 4 and Table 1. In the early stages of development, during transition from E45mm to E80mm stage, there is a dramatic increase in the 43K polypeptide content which affects equally both cytosol and membrane-associated pools (335-±18 and 354±39 fold respectively) (Fig. 4). The increase in membrane-associated AChR molecules (no AChR molecules were detectable in the cytosolic pool) is also dramatic (60- to 70-fold) (Table 1), although not as strong as that observed with the 43 K polypeptides. Upon further maturation of the electrocyte, the increase in 43K polypeptide decfines for both pools and is restricted rather to the membrane-associated pool (the ratio of adult over embryonic E80mm 43K polypeptide was 1.3±0.3 and 13.6±2.6 for the cytosol and membrane-associated pools, respectively) (Fig. 4). The amount of AChR increases only slightly (2- to 3fold during transition from E80mm to adult stage) (Table 1). In the membrane-associated pool, the ratio of AChR/43K decreases with maturation of the electrocyte: from 6- to 12-fold at the E45mm stage, it decreases to about 1.4- to 3.0-fold at the E80mm stage and to 0.4- to 1.0-fold in the adult electrocyte (Table 1).

Concerning the distribution of the 43K polypeptides (Fig. 5), similar total contents of 43K protein are found in both membrane and cytosol fractions of early electrocytes (1.3±0.7 membrane/cytosol at E45mm stage). At later developmental stages, more 43K protein is present in the membrane-enriched pools: the ratio (membrane over cytosol) increases slightly in the embryonic E80mm stage electrocyte (3.8±0.5). A quite different partition is observed between the two compartments in the adult electrocyte in which the 43K polypeptides are massively associated with the membrane compartment (24.7±9.4 more membrane-associated than cytosol-associated 43K polypeptide). Similar results expressed as ratios of the amounts of 43K polypeptide present per mass unit of protein in the membrane over that of the cytosol-rich fractions are observed (see Table 1).

In this report we confirm the coexistence in the electrocyte of two cytoplasmic and membrane 43K protein pools, the latter showing the expected coextensive distribution with AChR clusters.

The two 43K protein pools are very similar with respect to their antigenic determinants and migration in SDS gels. They both derive from the longer 43K mRNA transcript although translation from the shorter one cannot be excluded. However, they evolve differently with maturation of the electrocyte.

In the E45mm developing electrocyte, no clustering of 43K protein could be detected (Kordeli et al. 1989; La Rochelle et al. 1990). Nevertheless, the 43K proteins are present as two small but readily measurable membrane and cytoplasmic pools (in a 1:1 ratio). At this stage, AChR clustering is already detectable. The membrane ratio of 6 to 12 AChR per 43K protein (in E45mm electrocyte) indicates that AChR clustering plausibly precedes that of 43K protein. The present analysis provides a quantitative support to our previous results (Kordeli et al. 1989, see also La Rochelle et al. 1990) which reported failure to detect 43K protein clusters at the ventral face of E45mm electrocyte in which AChR clusters have been observed (Witzemann et al. 1983). These data conform to the hypothesis that, in electrocyte tissue, massive 43K protein accumulation is not required for induction of early AChR clustering.

Our results contrast with those from Xenopus oocyte expression studies (Froehner et al. 1990) and fibroblast transfection experiments (Phillips et al. 1991) where expression of 43K protein induces the formation of small AChR clusters in the cell surface membrane. These discrepancies may possibly reflect environmental or developmental differences. For example, as early as stage E45mm, the developing electrocyte is surrounded by a basal lamina (Richardson et al. 1987) and laminin (Kordeli et al. 1989). Components of the basal lamina from brain or motor neuron, such as ARIA (Falls et al. 1990) or agrin (McMahan, 1990) have been shown to play a role in AChR aggregate formation. Thus, AChR aggregation in the E45mm electrocyte may arise from the effects of extracellular elements absent from in vitro expression systems. In any event, our data emphasize that, in the developing electrocyte, an established model of the NMJ, massive presence of 43K protein is not essential for the early formation of AChR clusters. The function of 43K protein, then, is less straightforward than that suggested by reconstitution studies.

Rather, the 43K protein seems to play a more complex and later role in the postsynaptic membrane of the electrocyte. The two 43K protein pools observed prior to synaptogenesis undergo rapid synthesis during the electrogenic and early synaptogenic phases. Then, during the maturation phase (Fox and Richardson, 1978, 1979; Mellinger et al. 1978; Krenz et al. 1980), as fully functional synapses form, 43K protein partitioning drastically changes. Now, 43K protein preferentially accumulates at the membrane (the ratio of membrane to cytoplasmic 43K increases from 4:1 to 25:1 during transition from E80mm to adult electrocyte), and codistributes with AChR at the cytoplasmic face of the postsynaptic membrane, reaching the approximate equimolar ratio (La Rochelle and Froehner, 1986; this study) present in the adult electrocyte. This switch in the compartmentalization of the 43K protein in the adult electrocyte may result from co- or posttrans-lational modifications. The membrane 43K protein is myristoylated (Musil et al. 1988; Carr et al. 1989b), and phosphorylated (Saitoh and Changeux, 1980a; Hill et al. 1991). Concomitance of synaptic maturation events and the preferential partitioning of 43K protein in the postsynaptic membrane after AChR has clustered argue for a potential stabilizing role for 43K protein. In several species, stabilization of AChRs has been observed well after aggregation (see Burden, 1977; Schuetze et al. 1978; Betz et al. 1980 for chick; Berg and Hall, 1975; Steinbach et al. 1979; Michler and Sakmann, 1980; Rainess and Weiberg, 1981; Salpeter and Harris, 1983 for rat; Brehm, 1986 for Xenopus).

Antibody labeling (Nghiêm et al. 1983; Sealok et al. 1984), cross-linking experiments (Burden et al. 1983), electron microscopic observations (Bridgman et al. 1987) and high resolution image analysis of postsynaptic membranes (Toyoshima and Unwin, 1988; Mitra et al. 1989) suggest close association of 43K protein and AChR. In rat myotube cultures, 43K protein extraction unmasks cytoplasmic domains of AChR leading to an increase in the accessibility of anti-cytoplasmic AChR antibodies and in the susceptibility of the related epitopes to chymotrypsin (Bloch and Froehner, 1987). Taken together, these and our observations strengthen the hypothesis that 43K protein plays a primary role in the stabilization and maintenance of already clustered AChRs.

As a working hypothesis, the following scheme might be proposed: In the developing electrocyte, prior to innervation, the 43K protein does not play a crucial role in the initial clustering of AChR. Rather, other components such as dystrophin (Jasmin et al. 1990), laminin and the basal lamina, which have been detected at the future postsynaptic domain, may be good candidates. Some, although scarce, 43K protein is already present. Then, upon arrival of nerve endings, there are signals inducing a dramatic increase of the synthesis of AChR and 43K proteins. Upon further maturation of the cell, accumulation of the 43K protein at the subsynaptic membrane and its association (still putative) to cytoplasmic portions of the AChR molecules and cytoskeletal proteins (rev. in Steinbach and Bloch, 1986) would result in the stabilization and maintenance of the postsynaptic structure.

We are grateful to Drs Daniel Louvard and Rolf Kemler for helpful comments. We thank Dr R. Cazaux, Marine Station, Arcachon, France, for managing the collect of Torpedoes.

This work was supported by grants from the Centre National de la Recherche Scientifique, the Collège de France, the Fondation pour la Recherche Médicale, the Institut National de la Santé et de la Recherche Médicale, and the Direction des Recherches et Etudes Techniques (contract 87/211).