Formation of the annexin II₂p11₂ complex upon differentiation of F9 teratocarcinoma cells

Annexin II and p11 are members of two distinct protein families of Ca²⁺-binding proteins, the annexin family and S100 protein family, respectively (for reviews see Hilt and Kligman, 1991; Moss et al., 1991). The two proteins are linked to each other in a tight heterotetrameric complex (annexin II₂p11₂) that is present in many tissues and cell lines of higher vertebrates (for review see Gerke, 1989). Within the cell the annexin II/p11 complex is concentrated at the cytoplasmatic face of the plasma membrane (Green-berg and Edelman 1983; Gerke and Weber, 1984; Osborn et al., 1988; Semich et al., 1989). This association most likely reflects the ability of annexin II to interact in a Ca²⁺-dependent manner with negatively charged phospholipids and components of the submembraneous cytoskeleton (for review see Gerke, 1989). Binding sites for these ligands reside in the 33 kDa protein core of annexin II, which can be generated by limited proteolysis and comprises four so-called annexin repeats (Glenney, 1986; Johnsson et al., 1986a). The p11 binding site in annexin II is contained within the N-terminal 14 amino acid residues which form an amphiphatic α-helix and are part of the protease sensi-tive N-terminal tail (residues 1-29; Johnsson et al., 1988a; Becker et al., 1990). In addition to the p11 binding site the N-terminal tail of annexin II harbors phosphorylation sites for protein kinase C and the tyrosine kinase encoded by the src oncogene (Glenney and Tack, 1985; Johnsson et al., 1986b; Gould et al., 1986). Whereas annexin II has all Ca²⁺ and lipid-binding properties typical of members of the annexin family, the other subunit of the complex, p11, has lost the Ca²⁺-binding activity displayed by other S100 pro-teins. This loss is explained by crucial amino acid deletions and substitutions in the two EF hand motives, which are thought to function as Ca²⁺-binding sites in the other mem-bers of the S100 protein family (Gerke and Weber, 1985; Glenney, 1986).

Formation of the annexin II₂p11₂ complex seems a pre-requisite for the association of the two subunits with the submembranous cytoskeleton. After microinjection into rat mammary cells carboxymethylated p11, which does not bind annexin II, shows a general cytoplasmic distribution and is not present in the submembraneous cytoskeleton (Osborn et al., 1988; Johnsson and Weber, 1990). Simi-larly, mutant annexin II molecules unable to bind p11 are not incorporated into the submembraneous network (Thiel et al., 1992). The percentage of annexin II which resides in the complex with p11 varies from cell type to cell type. When isolated from the mucosa of porcine small intestine more than 90% of annexin II is associated with p11, whereas fibroblasts contain approximately 50% of annexin II in a monomeric form, which appears to reside in the cyto-plasm (Gerke and Weber, 1984; Erikson et al., 1984; Zokas and Glenney, 1987). Complex formation also modulates biochemical properties displayed by annexin II. The com-plex but not monomeric annexin II has the capability to aggregate chromaffin granules at micromolar Ca²⁺ concen-trations (Drust and Creutz, 1988). Moreover, compared to monomeric annexin II the complex has an increased poten-tial to restore Ca²⁺-evoked exocytosis in permeabilized chromaffin cells (Sarafian et al., 1991).

Many established cell lines express annexin II as well as p11. In some cases a significant change in the level of expression is observed upon differentiation. U937 promyeloblastic cells show an increase in annexin II syn-thesis upon differentiation into a macrophage-like pheno-type (Isacke et al., 1987). Similarly, an induction of annexin II expression is observed in PC 12 pheochromocytoma cells which were induced to differentiate into neuron-resembling cells by the addition of nerve growth factor (NGF) (Schlaepfer and Haigler, 1990). p11 transcription shows a comparable increase in NGF-treated PC 12 cells (Masi-akowski and Shooter, 1988). Several other processes such as transformation of hepatocytes to hepatoma cells (Frohlich et al., 1990) or serum-induced growth stimulation of fibroblasts (Keutzer and Hirschhorn, 1990) are associ-ated with an increase in the levels of annexin II transcript and protein. However, potential alterations in the p11 expression have not been studied in these cases.

For this study we have chosen the murine teratocarci-noma cell line F9, which remains largely undifferentiated under standard culture conditions. Upon induction of differentiation F9 cells form cellular layers resembling early embryonic tissues (see Alonso et al., 1991, for a recent review). The relatively simultaneous differentiation of the cells allows a detailed study of proteins whose expression pattern changes during the differentiation process. Among the early genes induced are putative transcription factors such as the homeobox-containing protein Era 1 (La Rosa and Gudas, 1988). The level of histon H1° also increases rather early in the differentiation process, whereas the induction of many other genes such as laminin and keratins 8 and 18 occurs later with the first rise detectable after 24 hours (Alonso et al., 1988; Wang et al., 1985; Oshima, 1981). Interestingly, undifferentiated F9 cells show, in con-trast to the other cell lines studied so far, a strong mispro-portion in the amounts of annexin II and p11 transcript. Northern blots reveal the presence of p11 mRNA although an annexin II transcript seems absent (Saris et al., 1987).

Here we describe the regulation of annexin II and p11 expression during the F9 differentiation processes. In addi-tion we show that the intracellular distribution of p11 in undifferentiated cells is altered upon coexpression of annexin II, achieved by transfection of the respective cDNA. Since this appears to be a consequence of annexin II/p11 complex formation F9 cells represent a valuable system to investigate the unknown physiological function of the complex.

F9 cells obtained from Dr. Peter Gruss (Department of Molecu-lar Cell Biology, Max Planck Institute for Biophysical Chemistry, Göttingen) were grown at 37°C in 5% CO² in Dulbecco’s modi-fied Eagle’s medium, supplemented with 10% fetal calf serum and 200 units/ml of streptomycin and penicillin, respectively (all Gibco BRL, Eggenstein, FRG). The supporting tissue culture dishes were coated with 0.1% gelatine. Under these conditions F9 cells remained undifferentiated. Differentiation of F9 cells into parietal endoderm cells was induced by incubation for 4 days in presence of retinoic acid and dibuturyl cAMP (Sigma Chemicals, Neuulm, FRG) to final concentrations of 5 × 10⁻⁷ M and 10 ⁻³ M, respectively. This will be referred to as rc treatment throughout the rest of the text (Strickland and Mahdavi, 1978). Cells resem-bling primitive endoderm developed within approximately 4 days after addition of retinoic acid alone to a final concentration of 5×10⁻⁷ M. This will be referred to as ra treatment. Incubation of aggregates of 20-30 F9 cells in bacterial Petri dishes in medium containing 10⁻⁷ M retinoic acid for 7 days led to formation of embryoid bodies (eb), which are composed of an undifferentiated inner cell mass surrounded by a layer of cells with a phenotype resembling visceral endoderm.

Cells from two 10 cm tissue culture plates were lysed in 5 ml ice-cold extraction buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM DTT, 0.2 mM PMSF, 0.5% Triton X-100) using a glass/teflon homogenizer. Subsequently CaCl₂ and E64 (Peptide Institute, Osaka, Japan) were added to 3 mM and 2.5 μM, respec-tively. After centrifugation for 20 min at 25,000 g the pellet was washed twice in extraction buffer plus 3 mM CaCl₂ and 2.5 μM E64. The final pellet was resuspended in 2 ml cold extraction buffer containing 5 mM EGTA and incubated for 15 min on ice. After centrifugation (20 min, 25,000 g) protein present in 200 μl of the supernatant (EGTA extract) was precipitated with CHCl³/methanol (Wessel and Flügge, 1984) and subsequently sol-ubilized by boiling in 20 μl SDS-PAGE sample buffer. Equal amounts of protein were separated in SDS-12.5% polyacrylamide gels (Laemmli, 1970), transferred to nitrocellulose membranes (Towbin et al., 1979) and subjected to immunoblotting with an annexin II-specific antibody (see below).

In order to analyze the expression of the p11 protein we par-tially purified the protein by affinity chromatography. Therefore, extracts from two 10 cm plates of undifferentiated or differenti-ated F9 cells were prepared by lysis in 5 ml of 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 2 mM EGTA, 1 mM DTT, 0.2% Triton X-100, 0.2 mM PMSF and subsequent centrifugation at 100,000 g for 30 min. p11, which was present in the super-natant, was enriched by affinity chromatography, which employed a peptide resembling the amino acids 1-14 (including the N-ter-minal acetyl group) of annexin II coupled to AH-Sepharose (Phar-macia, Freiburg). This peptide and the matrix bind p11 with high affinity (Johnsson et al., 1988a; Kube et al., 1992). A 15 μl sample of a matrix slurry (approximately 20 μg of peptide, i.e. a high molar excess to possibly competing annexin II molecules) was incubated for 30 min at room temperature with 1 ml of the 100,000 g supernatants (after adjusting their protein contents) and subse-quently washed twice with TBS (20 mM Tris-HCl, pH 7.4, 150 mM NaCl) and TBS containing 0.2% Triton X-100. The protein was released from the matrix by boiling in 50 μl SDS-PAGE sample buffer (Laemmli, 1970), subjected to electrophoresis in tricine-SDS-16.5% polyacrylamide gels (Schägger and von Jagow, 1987) and transferred to nitrocellulose membranes following stan-dard procedures (Towbin et al., 1979).

Immunostaining of the nitrocellulose filters was performed using either a polyclonal antibody against annexin II (Gerke and Weber, 1984) or a monoclonal p11 antibody (clone LC 129, kindly provided by Dr. J.R. Glenney) at concentrations of 50 μg/ml or 10 μg/ml, respectively. After thorough washing with TBS plus 0.1% Triton X-100, the filters were incubated for 1 h with a 1:1000 dilution of peroxidase-coupled second antibodies directed against mouse or rabbit immunoglobulin, respectively (Dako Immuno-histochemicals, Klostrup, Denmark), and subsequently washed with TBS plus 0.1% Triton X-100. Bands decorated by the anti-bodies were detected with the ECL chemoluminescence system (Amersham Buchler, Braunschweig, FRG) following the manu-facturer’s protocol.

For immunofluorescence the cells were grown on coverslips coated with 0.1% gelatine. After washing with PBS the cells were fixed in 3.7% formaldehyde in PBS (10 min at room temperature) and subsequently permeabilized with methanol for 6 min at −10°C. After rinsing in PBS the coverslips were incubated with the pri-mary antibody for 1 h at 37°C in a humidified atmosphere. Murine (F9 cell-) annexin II was stained with undiluted hybridoma cul-ture supernatant containing the mouse monoclonal antibody HH7 (Thiel et al., 1992) or with an affinity-purified polyclonal rabbit antibody (Gerke and Weber, 1984) used at a concentration of 30 μg/ml. Annexin II introduced into F9 cells by transient transfec-tion was stained with undiluted hybridoma culture supernatant containing the mouse monoclonal antibody H28 (Osborn et al., 1988). H28 has a limited cross-species reactivity and does not react with the endogenous mouse annexin II (Johnsson et al., 1988b). Due to a reconstruction of the H28 epitope, the exoge-nous annexin II, which is derived from a human cDNA clone con-taining a glutamic acid for alanine replacement at position 65, is recognized by this antibody (Thiel et al., 1991, 1992). The mouse monoclonal antibody H42 (Osborn et al., 1988) was used to dec-orate p11. Keratin was visualized by staining with the mouse mon-oclonal antibody lu 5 (von Overbeck et al., 1985). Both H42 and Lu5 were used as undiluted culture supernatants. Polyclonal rabbit antibodies at a concentration of 20 μg/ml were used to stain fodrin (Glenney et al., 1982) and actin (a kind gift from Dr. M. Osborn). After incubation with the first antibodies (1 h, 37°C) the cover-slips were washed 3 times in PBS and decorated with FITC-cou-pled goat anti-mouse antibodies (20 μg/ml, Cappel Laboratories, Cochranville, PA, USA) and/or with rhodamine-coupled goat anti-rabbit antibodies (Dianova, Hamburg, FRG) for 1 h at 37°C. After a final PBS wash (3 times at room temperature) the coverslips were embedded in Mowiol 4-88 (Hoechst, Frankfurt, FRG) and analyzed in a Zeiss Axiophot photomicroscope. Photography employed Kodak Tri X Pan 400 film.

Cryostat sections (10 μm) of embryoid bodies were cut using a Reichert Jung model 2700 Frigocut cryostat after embedding in Tissue Tec (Miles, Elkhart, IN, USA) and freezing in liquid nitro-gen. After transferring the sections to slides they were air dried, fixed in 3.7% formaldehyde in PBS, permeabilized with ice-cold methanol and stained as described above.

Total cellular RNA from F9 cells was prepared according to Chomzynski and Sacchi (1987). 20 μg samples of RNA were sep-arated in a 1.2% agarose/formaldehyde gel and transferred to a Qiabrane noncharged membrane (Diagen, Düsseldorf, FRG) as described (Sambrook et al., 1989). The RNA was UV-crosslinked with 0.3 J/cm² at 312 nm. Subsequently the blots were prehy-bridized for 12 h at 40°C in 50% formamide, 50 mM Tris-HCl pH 7.5, 5 × SSC, 1 × Denhardt’s, 1% SDS, and 100 μg/ml denatured salmon sperm DNA (Boehringer, Mannheim, FRG). Hybridization was carried out for 16 h at 40°C simultaneously with 1 ng/ml of radiolabeled murine annexin II cDNA (V. Gerke, unpublished) and 1 ng/ml of labeled human p11 cDNA (Kube et al., 1991). After removal of the annexin II and p11 probes by incubation at 90°C in 0.01 × SSC for 10 min, the same blot was hybridized with a radiolabeled chicken β-actin probe (Cleveland et al., 1980). The probes were radiolabeled with [α-³²P]dCTP (Amersham-Buchler, Braunschweig, FRG) to a specific activity of 5 × 10⁸ cts/min per μg using a random priming kit (Pharmacia, Freiburg, FRG). Blots were washed at a final stringency of 2 × SSC, 0.1% SDS at 60°C. Autoradiography used Kodak X Omat films with an intensifying screen at −70°C.

The human annexin II cDNA was cloned into the eucaryotic expression vector pCMV 5 as described previously (Thiel et al., 1992). A CaPO⁴ precipitate containing 10 μg of purified plasmid DNA was formed by standard procedures (Graham and van der Eb, 1973) and added to a 5 cm Petri dish containing F9 cells grown to 30% confluency on gelatine-coated coverslips. After incubation for 12 h at 37°C the precipitate was removed from the cells by washing in full medium prewarmed to 37°C. Subsequently the cells were incubated for an additional 12 h and the distribution of the exogenous annexin II was monitored with the monoclonal antibody H28 (see above). In double-labeling experiments, endogenous p11 and exogenous annexin II were decorated simultaneously by immunofluorescence using the mouse monoclonal antibody H42 directed against p11 (Osborn et al., 1988) and a rabbit polyclonal antibody directed against annexin II (Gerke and Weber, 1984).

F9 cells can be induced to differentiate into cells resem-bling those of a primitive, visceral or parietal endoderm by addition of retinoic acid alone or in combination with dibu-turyl cAMP. Addition of retinoic acid together with dibu-turyl cAMP results in a F9 phenotype reminiscent of a pari-etal endoderm (Strickland et al., 1980). Retinoic acid alone induces the development of primitive endoderm-like cells even though the population appears somewhat inhomoge-neous (Strickland and Mahdavi, 1978). Incubation of aggre-gates of F9 cells at low concentrations of retinoic acid induces the development of embryoid bodies, which con-sist of a relatively undifferentiated inner cell mass sur-rounded by a layer of cells resembling a visceral endoderm (Hogan et al., 1981). To determine the expression and sub-cellular distribution of annexin II and p11 in the various differentiation states we employed immunological as well as Northern blot analyses.

Radiolabeled cDNAs of murine annexin II and human p11 were used to probe a Northern blot containing total RNA from undifferentiated F9 cells (Fig. 1, un, two independent preparations), and embryoid bodies (eb) as well as primi-tive endoderm (ra), and parietal endoderm (rc, two inde-pendent preparations) resembling cells. Fig. 1 reveals that undifferentiated F9 cells show only a weak hybridization signal with the annexin II-specific probe, thus confirming a previous Northern blot analysis (Saris et al., 1987). Treat-ment with retinoic acid (ra) or retinoic acid plus cAMP (rc) leads to a more than 20-fold increase in the annexin II tran-script when compared to undifferentiated cells. The rise in the level of annexin II mRNA is somewhat lower in embry-oid bodies (eb) which formed after incubation of aggregates of F9 cells in the presence of low concentrations of retinoic acid. Most likely, this reflects the fact that only the outer layer of cells of embryoid bodies resembles a visceral endo-derm whereas the inner cell mass remains undifferentiated. The signals resulting from the hybridization with a p11-specific probe have the same intensity in each state of differentiation. Thus, in contrast to the annexin II mRNA, the level of p11 mRNA is unaffected by the differentiation processes studied.

The induction of annexin II expression also reflects itself on the protein level as revealed by Western blot analysis of EGTA extracts of the four differentiation states (Fig. 2A). While the extract from undifferentiated cells shows very little annexin II immunoreactivity (visible only after overexposure of the chemoluminescence reaction, not shown), an annexin II band is readily visible in the extracts from differentiated F9 cells. In line with the Northern blot results a lower level of annexin II is detected in the EGTA extract from embryoid bodies as compared to the EGTA extracts from ra- and rc-treated cells. Thus the increase in annexin II mRNA is accompanied by a rise in the annexin II protein, indicating that the transcript is effectively trans-lated. To elucidate whether the amount of p11 protein pre-sent in various differentiation states also correlates with the respective mRNA levels, the protein was partially purified from untreated and ra-treated cells. Therefore extracts from each cell type were incubated with a peptide comprising the N-acetylated 14 N-terminal amino acids of annexin II coupled to AH-Sepharose beads. This matrix binds p11 with high affinity (Kube et al., 1992). p11 of the cellular extracts was retained by the affinity matrix and subsequently released by boiling in SDS-PAGE sample buffer (Laemmli, 1970) and analysed by immunoblotting with a mouse mon-oclonal antibody (kindly provided by Dr. J.R. Glenney). Fig. 2B shows that a considerable amount of p11 protein is present in both undifferentiated and ra-treated F9 cells. Thus undifferentiated F9 cells contain a substantial amount of p11, whose translation and stability is independent of annexin II expression.

Important information as to the possible function of the annexin II/p11 complex in the differentiation of F9 terato-carcinoma cells can be obtained by determining the time point of the onset of annexin II synthesis after addition of the inducing agents. The Northern blot in Fig. 3 shows the signals obtained with annexin II- and p11-specific hybridization probes on total RNA from F9 cells harvested at different time points after the addition of the inducing agents retinoic acid and dibuturyl cAMP. During the first 12 h of the rc treatment the amount of annexin II transcript remains low. A substantial increase in annexin II is appar-ent after 24 h. The level of annexin II mRNA rises further during the differentiation process and reaches a plateau after 96 h. Thus induction of annexin II expression occurs relatively late in the differentiation process and seems coupled to the acquisition of the fully differentiated phenotype and not to early events in F9 cell differentiation. p11 expres-sion, on the other hand, remains basically unaffected during the entire differentiation process.

The subcellular distribution of annexin II and p11 in differentiated F9 cells was analysed by immunofluorescence microscopy. A typical staining of the submembraneous region and the enrichment at sites of cell-to-cell contact described for other cell lines of epithelial origin (Greenberg and Edelman, 1983; Osborn et al., 1988; Zokas and Glen-ney, 1987) is clearly visible in ra-(Fig. 4A) and rc-(Fig. 4B) treated cells. Not all cells are stained equally with the annexin II-specific antibody, most likely due to some vari-ations in the degree of differentiation. The p11 distribution of differentiated cells is highly similar to that of annexin II (not shown) whereas a diffuse cytoplasmatic p11 staining is visible in undifferentiated cells which show very little annexin II expression (see Fig. 5).

The expression of annexin II in different cell types of the embryoid bodies was revealed by staining cryosections of frozen embryoid bodies with the mouse monoclonal anti-body HH7 (Thiel et al., 1992). Fig. 4C and D shows that the annexin II immunoreactivity is restricted to the outer layer of differentiated cells, which form a visceral endo-derm. A hallmark for this cell layer is the expression of keratin (Paulin et al., 1982), which is revealed by staining with the mouse monoclonal antibody lu 5 (Overbeck et al., 1985). The inner cell mass of undifferentiated cells, which is annexin II negative, is also devoid of keratin staining (Fig. 4E). This observation explains that relatively weak annexin II signals were obtained in the Northern and West-ern blot analyses of embryoid bodies when compared to the ra- and rc-treated F9 cells (Figs 1 and 2A). While the inner cells do not contain annexin II it seems that the annexin II expression in the outer layer of differentiated cells reaches a high level, which is similar to that in ra- and rc-treated cells. However, the RNA and protein samples from total embryoid bodies represent a mixture obtained from the different cell types and thus contain intermediate levels of annexin II. Immunofluorescence staining of cryosections revealed that p11 seems expressed in all cell types of the embryoid bodies (not shown). However, this analysis is complicated by the fact that uncomplexed p11 is easily extracted upon sectioning and fixation.

Since undifferentiated F9 cells contain considerable amounts of p11 protein but no or very little annexin II, it was of interest to investigate whether the intracellular dis-tribution of the presumably free p11 is altered upon an ele-vated expression of annexin II. To increase the level of annexin II we employed transient transfection experiments and introduced an annexin II expression construct into undifferentiated cells. This approach also enabled us to elu-cidate the subcellular distribution of annexin II artificially introduced into cells devoid of endogenous annexin II. Effi-cient expression of the exogenous annexin II gene driven by the cytomegalovirus promoter was revealed by immunostaining with the mouse monoclonal antibody H28 (not shown), which reacts with the exogenous but not with the endogenous murine annexin II (Osborn et al., 1988; Johnsson et al., 1988b; Thiel et al., 1992). The p11 distri-bution in the transfected cells was monitored in double-labeling experiments using a polyclonal rabbit antibody specific for annexin II (Gerke and Weber, 1984; Fig. 5A) and the mouse monoclonal antibody H42 directed against p11 (Osborn et al., 1988; Fig. 5B). The intracellular local-ization of p11 is clearly altered in transfected cells, which are easily identified by the decoration with the annexin II antibody. While p11 is diffusely distributed throughout the cytoplasm in untransfected cells, the cells expressing exoge-nous annexin II show a staining of the submembraneous region with both p11- and annexin II-specific antibodies (Fig. 5). This relocalization of p11 most likely reflects its association with the exogenous annexin II, which in turn acquires a submembraneous localization with the typical concentration at the sites of cell to cell contact. The p11 staining is stronger in cells which had received the annexin II expression construct than in the untransfected controls. Most likely, this is a consequence of the annexin II₂p11₂ complex formation, which renders the endogenous p11 less susceptible to extraction during the fixation procedure. No gross morphological alterations are observed in the trans-fected cells, indicating that the annexin II/p11 complex alone is not sufficient to induce the significant changes in cell morphology which occur during F9 differentiation. Fur-thermore, the intracellular localization of typical elements of the submembraneous cytoskeleton, actin and fodrin, appears unaltered in the transfected cells (not shown).

Comparative Northern and Western blot studies have shown that most cell lines and tissues contain roughly equal amounts of both subunits of the annexin II₂p11₂ complex (see Introduction for references). A notable exception is the teratocarcinoma cell line F9, which contains significant amounts of p11 but almost no annexin II both on the RNA (Saris et al., 1987; this study) and on the protein level (this study). In order to examine whether this misproportion is associated with the highly undifferentiated state of these cells, we studied the expression of annexin II and p11 during the differentiation of F9 cells.

Undifferentiated F9 cells contain very little annexin II transcript and protein. This could either be due to a very weak expression in all cells, which is too low to be detected by immunofluorescence, or could result from a small pop-ulation of spontaneously differentiated F9 cells showing strong annexin II expression. The level of annexin II expres-sion indeed rises very strongly upon differentiation. A more than 20-fold increase in annexin II transcript and protein can be detected in differentiated F9 cells resembling a prim-itive or parietal endoderm. Similarly, the visceral endo-derm-like cells surrounding embryoid bodies show a high level of annexin II expression, which is probably similar to that reached in ra- or rc-treated cells. The expression of p11, on the other hand, is unaffected during the F9 differ-entiation processes both at the mRNA level and the protein level. Thus in contrast to all other cell lines and tissues studied so far the synthesis of annexin II and p11 is uncou-pled in the F9 system.

Another cell system capable of differentiation in vitro, for which the expression of annexin II and p11 has been studied, is the pheochromocytoma cell line PC12. Upon application of nerve growth factor these cells differentiate and acquire a neuronal phenotype, as indicated by the devel-opment of neurite like extensions. This process is accom-panied by a strong increase in the amount of annexin II protein (Schlaepfer and Haigler, 1990; Gerke, unpublished observation) and p11 mRNA (Masiakowski and Shooter, 1988). It seems that the expression of annexin II and p11 is simultaneously induced upon differentiation of PC12 cells leading to an efficient annexin II₂p11₂ complex for-mation. This co-regulation could occur at the transcriptional level as similar sequence elements, e.g. the βDRE motif (Stuve and Meyers, 1990), are found in the annexin II and the p11 promoter region (Spano et al., 1990; Harder et al., 1992). However, other sequence elements present in the murine p11 promotor must be responsible for the activation of transcription in the undifferentiated F9 cells, since these cells show no significant annexin II expression. Alter-natively, specific elements in the annexin II promotor could act as negative regulators of the transcription of this gene in undifferentiated cells.

The time course of annexin II induction during F9 differentiation resembles that of several other proteins studied so far, e.g. keratin 8 and 18 (Oshima, 1981) or laminin (Wang et al., 1985). In each case the first increase in expres-sion is detected 24 h after addition of the inducing agents. This relatively late onset of expression indicates that these proteins function in later stages of differentiation and not in the early events which establish the commitment of F9 cells towards the differentiation process.

In differentiated F9 cells, p11 and annexin II are found underneath the plasma membrane showing a particular enrichment at the sites of cell to cell contact. This distrib-ution is very similar if not identical to that of fodrin and cortical actin and thus places the annexin II/p11 complex in the submembraneous cytoskeleton. A similar concentration in the cell cortex is also observed for exogenous annexin II introduced via transfection into undifferentiated F9 cells. Interestingly, p11 follows the exogenous annexin II to the plasma membrane whereas the staining of p11 in untransfected cells is diffuse and cytoplasmatic. This translocation indicates that the association of p11 with the cytoplasmic face of the plasma membrane is dependent on the formation of the annexin II₂p11₂ complex. Similar con-clusions have been drawn from microinjection experiments using rat mammary carcinoma cells. A microinjected p11 derivative incapable of annexin II binding remains diffusely distributed in the cytoplasm whereas injected wild-type p11 is incorporated into the submembranous region, presumably through its interaction with annexin II (Osborn et al., 1988). Annexin II also requires the formation of an annexin II₂p11₂ complex for its stable integration into the submembraneous cytoskeleton. Mutant annexin II molecules which do not bind p11 are not incorporated into the Triton X-100-insol-uble cortical cytoskeleton of fibroblasts (Thiel et al., 1992). Our transfection experiments show that the actin- and fodrin-based cell cortex of undifferentiated F9 cells fulfills all requirements for the incorporation of the annexin II/p11 complex. Nevertheless, the morphology of the undifferen-tiated F9 cells expressing exogenous annexin II appears unaltered. Similarly, the cell cortex itself is not disturbed or otherwise affected by the annexin II₂p11₂ incorporation as judged by immunofluorescence with actin and fodrin antibodies. Thus the annexin II₂p11₂ complex alone is not sufficient to induce significant alterations in the submem-braneous cytoskeleton and/or the morphology of the cell. To analyze the possible biological activities of annexin II₂p11₂ it will be of interest to establish an undifferentiated F9 cell line stably expressing annexin II and compare its properties with those of wild-type F9 cells in more detail.

We thank Klaus Weber for support and Frauke Melchior and Klaus Weber for stimulating discussions. This work was supported in part by a grant from the Deutsche Forschungsgemeinschaft (Ge 514/2-1).