INTRODUCTION

Eucaryotic cells are subdivided into membrane-bounded compartments. These functional organelles contain sets of proteins and other molecules specific to themselves. The intracellular vesicular transport system delivers specific proteins to their destination. A knowledge of this mechanism is essential for understanding how these compartments are created and maintained within eucaryotic cells.

The oligodendrocyte provides both an opportunity and a challenge for studying the machinery of intracellular vesicular transport. Oligodendrocytes are glial cells in the central nervous system which synthesize unique functional component ``myelin.'' Myelin is composed of multilamellar stacks of plasma membrane surrounding individual axons and plays a significant role in supporting fast nerve conduction. To create and maintain this myelin, oligodendrocyte must deliver large amounts of proteins and lipids to this component via vesicular transport (1).

Generally, integral membrane proteins are co-translationally inserted into the rough endoplasmic reticulum membrane and then transported to the plasma membrane via the Golgi apparatus (2). Myelin proteolipid protein (PLP)¹ is the major integral membrane protein of central nervous system myelin. PLP mRNA is associated with polysomes on the rough endoplasmic reticulum (3) and an immunoreactive product has been detected in membranous structures, such as the Golgi apparatus, of oligodendrocytes in vivo (4, 5, 6). As expected for a protein being processed through the vesicular transport pathway, a significant lag exists between translation of PLP on the rough endoplasmic reticulum and its insertion into the myelin membrane (3, 7). Mutations within the PLP gene causes severe dysmyelination (8), at least in part caused by an impaired protein transport system. In one of the PLP mutants, the jimpy mouse, for example, the mutated PLP protein accumulates in the rough endoplasmic reticulum and very little PLP is found in myelin (6). In oligodendrocytes of the transgenic mouse overexpressing the wild type PLP gene there is a swelling in the Golgi apparatus and PLP is rarely found in myelin (9, 10). Therefore, it is very important to study the regulation of the PLP transport system in order to understand the efficient vesicle transport system of oligodendrocyte.

PLP is a highly conserved protein. In mammals, the amino acid sequences of PLP of bovine, rat, mouse, and human are 99% identical, suggesting that PLP has indispensable functions (11). DM-20 is a less abundant proteolipid of mammalian central nervous system myelin, the mRNA of which is produced by alternative splicing of the PLP-mRNA precursor (12, 13, 14). It is important to ascertain the function conferred upon DM-20 by the addition of this PLP-specific domain.

InsP₆ is found at concentrations from 10 to 100 µM in many kinds of cells (15, 16). Although, the function of InsP₆ has not yet been clarified, several recent findings have suggested a physiological role for InsP₆. Several proteins involved in intracellular vesicular transport have been identified as InsP₆-binding proteins. A clathrin assembly protein, AP-2 (17, 18, 19, 20), may be an essential protein in the endocytotic recycling pathway of all cells (21). Binding of InsP₆ inhibits the clathrin assembly mediated by AP-2 (22) and AP-3, a synapse-specific clathrin assembly protein (23, 24). Coatomer, a cytosolic protein complex containing subunits of non-clathrin-coated Golgi intercisternal transport vesicles, also binds InsP₆ (25). These findings indicate that InsP₆ is closely related to vesicular transport.

In this study, we have purified PLP using a non-organic solvent and showed that PLP is an InsP₆-binding protein, while DM-20 is not. Apparently, DM-20 acquired InsP₆ binding activity by gaining a PLP-specific domain and thereafter became the major central nervous system myelin component PLP with InsP₆ binding activity. Thus, this binding property of PLP may play a crucial role in targeting vesicles containing PLP to central nervous system myelin.

EXPERIMENTAL PROCEDURES

[³H]InsP₄ (17 Ci/mmol), [³H]InsP₃ (17 Ci/mmol), [³H]InsP₆ (12 Ci/mmol), and PP-InsP₅ were obtained from DuPont NEN. Ins-1,4,5-P₃, Ins-1,3,4,5-P₄, and CHAPS were from Dojindo Laboratories. Ins-1,3,4,5,6-P₅ was from Boehringer Mannheim, while InsP₆ was from Sigma. The hybridoma of anti-PLP monoclonal antibodies (AA3 and AH7-2a) were kindly supplied from Dr. Marjorie B. Lees (26). The hybridomas were cultured in HYBRIDOMA-SFM (Life Technologies, Inc.) medium.

Protein concentrations were determined using the Bio-Rad protein assay (the Bradford protein assay) with bovine serum albumin as standard. All the purification steps and protein handling were performed at 4 °C or on ice. The pH of all buffers was adjusted at room temperature and was not corrected for cooling to 4 °C.

InsP₄ and InsP₆ binding were measured by the slightly modified polyethylene glycol precipitation method as described previously (27). Assay mixture contained 2 mg/ml -globulin, 20 mM HEPES-NaOH at pH 7.2 for the [³H]InsP₄-binding assay and 0.15 M KCl, 0.2 mg/ml -globulin, 50 mM HEPES-KOH at pH 7.2 for the [³H]InsP₆ binding assay.

Preparation of the P2/P3 membrane fraction from young adult male ddY mouse cerebella (18 g) and the solubilization with 1% Triton X-100 were carried out according to the method described previously (27). The supernatant (1% Triton X-100 extract) (~200 ml) was applied to a column of DE-52 (Whatman) ( 2.6 × 11 cm) equilibrated with 1% Triton X-100, 10% glycerol, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), 10 µM leupeptin, 10 µM pepstatin A, 1 mM 2-mercaptoethanol, and 50 mM Tris-HCl, pH 8.0 (Buffer 1). The column was washed with 70 ml of Buffer 1. DE-52 flow-through fractions were stored at 70 °C. Stored DE-52 flow-through fractions from two experiments were combined (~400 ml) and incubated with 5 ml of packed heparin-agarose (Sigma) for 1 h at 4 °C on a rotator. The heparin-agarose was poured into a column ( 1.0 cm), and nonadherent proteins were collected and discarded. The column was washed with 40 ml of Buffer 2 (10 mM CHAPS, 0.1 mM PMSF, 10 µM leupeptin, 10 µM pepstatin A, 1 mM 2-mercaptoethanol, and 50 mM EPPS-NaOH, pH 8.0) containing 0.05 M NaCl and 40 ml of Buffer 2 containing 0.25 M NaCl, and the InsP₄ binding activities were eluted with a linear gradient of 0.25-1.0 M NaCl, Buffer 2 (50 ml total) at 0.5 ml/min. Peak fractions of the InsP₄ binding activity were pooled and diluted with 15 volumes of the solution containing 0.1 mM PMSF, 10 µM leupeptin, 10 µM pepstatin A, 1 mM 2-mercaptoethanol, 50 mM HEPES-NaOH, pH 7.5 (Buffer 3), and incubated with 8 ml of packed cation-exchange gel CM-52 (Whatman) for 1 h at 4 °C on a rotator. They were poured into a column, and nonadherent proteins were discarded. The column was washed with 40 ml of Buffer 3 containing 10 mM CHAPS, 0.05 M NaCl and the binding activities were eluted with a linear gradient of 0.05-0.5 M NaCl, 10 mM CHAPS, Buffer 3 (50 ml total) at 0.5 ml/min. Peak fractions of the InsP₄ binding activity were pooled, concentrated, and applied to a column of Sephacryl S-300 (Pharmacia) ( 1.0 × 57 cm) equilibrated with 10 mM CHAPS, 0.5 M NaCl, Buffer 3. Peak fractions of InsP₄ binding activity were pooled, concentrated, and stored at 70 °C.

SDS-PAGE for protein profile analysis or immunoblotting were carried out by the method of Laemmli (28). The proteins were visualized with Coomassie Brilliant Blue R-250 or by silver staining with Silver Stain II kit (Wako).

After electrophoresis, the proteins were transferred to a nitrocellulose membrane (Hybond-C; Amersham) using semidry system (ATTO). The membrane was soaked in 5% skim milk in PBS containing 0.1% Tween 20 (T-PBS) for 1 h at room temperature to block nonspecific binding and then incubated in the culture supernatant of hybridoma producing monoclonal antibody for 60 min at 37 °C. After washing with T-PBS, the membrane was incubated with biotinylated goat anti-rat IgG (Fc region specific) (Jackson Immunoresearch Laboratories) (1:200) in 5% skim milk, T-PBS. Immunoreactive bands were visualized using Vectastain ABC kit (Vector Laboratories) according to the manufacturer's protocol. Finally, the membrane was treated with 0.08% diaminobenzidine, 0.009% hydrogen peroxide, 0.04% NiCl₂/imidazol (5 µg/ml), PBS.

The purified IP4BP2b (approximately 14.4 µg) was separated by SDS-PAGE on 12% gel. After separation, the proteins were transferred to a polyvinylidene difluoride membrane (0.2 µm) (Bio-Rad) with CAPS transfer buffer (CAPS-NaOH, pH 11.0, 10% methanol). These proteins were visualized with Coomassie Brilliant Blue R-250 and protein bands were cut from the blots. The membrane pieces were applied to a gas-phase protein sequencer (Applied Biosystem). The sequences were compared to those in the SWISS-PROT data base.

The AA3-IgG solution which was prepared by general ammonium sulfate-precipitation method, or normal rat IgG solution (Inter-Cell Technologies Inc.) were diluted with PBS to give a final protein concentration of 2 mg/ml. To each 500 µl of the above IgG solution (1 mg of protein), an equal volume of binding solution (1 M acetate buffer, pH 4.6, containing 3 M NaCl) was added, and incubated with 200 µl of packed Protein G-Sepharose (Sigma) equilibrated with the binding solution overnight at 4 °C on a rotator. Each of these solutions was poured into a column ( 5 mm), and washed with 2 ml of the washing solution containing 0.15 M NaCl, 20 mM HEPES-NaOH, pH 7.5. The washed gel particles were transferred into sample tubes and incubated with 20 µl of the purified IP4BP2b (PLP) (13.6 µg) for 1 h at 4 °C on a rotator. After incubation, these were re-poured into columns ( 5 mm) and washed four times with 100 µl of washing solution. Nonadherent proteins were collected at approximately 100 µl/fraction. A control experiment was also performed by the same method without IgG. Each of these four fractions was used for [³H]InsP₄-binding assay or SDS-PAGE analysis.

Myelin was purified from the medulla oblongata and the spinal cord of adult male ddY mouse by the procedure of Lucas et al. (29).

Stored DE-52 flow-through fraction (20 ml) was diluted with 19 volumes of the solution containing 1% Triton X-100, 0.1 mM PMSF, 10 µM leupeptin, 10 µM pepstatin A, 1 mM 2-mercaptoethanol, 1 mM EDTA, 1 mM EGTA, 50 mM acetate buffer, pH 5.0 (Buffer 4), and applied to a column of CM-52 ( 2.6 × 15 cm) equilibrated with Buffer 4. The column was washed with 50 ml of 0.05 M NaCl/Buffer 4. CM-52 flow-through fraction containing DM-20 was stored at 70 °C. PLP was eluted with a linear gradient of 0.05-0.5 M NaCl/Buffer 4 (100 ml total) and a further 100 ml of 1.0 M NaCl/Buffer 4 at 1.0 ml/min. Peak fractions of PLP detected by immunoblot analysis were pooled and stored at 70 °C. This mixture (8 ml) was diluted with an equal volume of the solution containing 1% Triton X-100, 0.1 mM PMSF, 10 µM leupeptin, 10 µM pepstatin A, 1 mM 2-mercaptoethanol, 1 mM EDTA, 1 mM EGTA, 50 mM HEPES-KOH, pH 7.2 (Buffer 5), and incubated with 0.5 ml of packed heparin-agarose for 1 h at 4 °C on a rotator. The heparin-agarose was poured into a column ( 0.5 cm), washed with 10 ml of the solution containing 10 mM CHAPS, 0.1 mM PMSF, 10 µM leupeptin, 10 µM pepstatin A, 1 mM 2-mercaptoethanol, 1 mM EDTA, 1 mM EGTA, 50 mM HEPES-KOH, pH 7.2 (Buffer 6). Nonadherent proteins were collected and discarded. PLP was eluted with 8 times of 0.2 ml of 1.0 M NaCl/Buffer 6, and peak fractions of PLP detected by immunoblot analysis were pooled and stocked as ``PLP-containing fraction'' at 70 °C.

Stocked CM-52 flow-through fraction containing DM-20 (100 ml) was incubated with 1.0 ml of packed heparin-agarose for 1 h at 4 °C on a rotator. The heparin-agarose was poured into a column ( 0.5 cm) washed with 5 ml of Buffer 6. Nonadherent proteins were collected and discarded. DM-20 was eluted with 10 times of 0.2 ml of 1.0 M NaCl/Buffer 6, and peak fractions of DM-20 detected by immunoblot analysis were pooled and stocked as ``DM-20-containing fraction'' at 70 °C.

RESULTS

During the sequential purification process of the receptor protein for InsP₃ (IP3R) from mouse cerebella (27), we noticed that some fractions contained [³H]InsP₄ binding activity, which indicated the existence of InsP₄-binding proteins. We have already purified and identified two InsP₄- binding proteins (IP4BPs). One is IP4BP1/synaptotagmin II (30) and the other is IP4BP2a/aldolase A.² We purified and identified another IP4BP (IP4BP2b). The sequential purification procedure is depicted as a flow chart (Fig. 1).

IP4BP1/synaptotagmin II and the other concomitant proteins (IP3R and phosphatases) were separated from IP4BP2 (a and b) by the first anion-exchange chromatography on DE-52. Since the volume of the DE-52 flow-through fraction was large, InsP₄ binding activity was concentrated by heparin-agarose chromatography. At the heparin-agarose chromatography step, the detergent in the sample and purification buffers was changed from Triton X-100 to CHAPS because Triton X-100 inhibited the InsP₄ binding activity of IP4BP2b more than CHAPS and because the concomitant proteins had been effectively separated from IP4BP2b.

IP4BP2a/aldolase A was separated from IP4BP2b by washing the heparin-agarose with 0.25 M NaCl (Fig. 2A). The IP4BP2b was eluted with a 0.25-1.0 M NaCl linear gradient. The InsP₄ binding activity of this fraction seemed to be expressed by only IP4BP2b. After dilution to lower the NaCl concentration and pH, the InsP₄ binding activity was concentrated and enriched by cation-exchange chromatography on CM-52 (Fig. 2B). The final step was Sephacryl S-300 gel filtration (Fig. 2C). Since the solubilizing detergent had been changed and the InsP₄ binding assay modified (described below), the purification process cannot be summarized in a figure. Approximately 0.5 mg of IP4BP2b was obtained from 40 g of mouse cerebella.

The protein profile of each purification step was characterized by SDS-PAGE (Fig. 3A). Particular attention was given to not boiling the sample mixtures in SDS solution, but rather allowing them stand at room temperature. Since the IP4BP2b appeared to be extremely hydrophobic, IP4BP2b protein aggregated and did not enter the separation gel after boiling (data not shown).

Many contaminating proteins which appeared after heparin-agarose chromatography were efficiently eliminated by CM-52 chromatography (Fig. 3A, lanes 2 and 3). After the final step on Sephacryl S-300, a single protein band with a molecular weight of 26,000 was detected (Fig. 3A, lane 4). While the molecular weight of IP4BP2b was 26,000 on SDS-PAGE, the apparent molecular weight was estimated to be 440,000-669,000 by gel filtration chromatography (Sephacryl S-300) (Fig. 3B). Consequently, IP4BP2b was expected to be a homomultimer or to aggregate. The pattern of InsP₄ binding activity and that of the intensities of Coomassie Brilliant Blue R-250 staining of this 26,000 protein differed slightly (Fig. 2C and Fig. 3B). The results below describe the binding activity resides in this 26,000 protein.

The NH₂-terminal sequence of purified IP4BP2b after Sephacryl S-300 gel filtration chromatography was determined with a gas-phase protein sequencer. The NH₂-terminal sequence of IP4BP2b was checked against the SWISS-PROT data base. All 10 amino acid residues identified out of 14 NH₂-terminal amino acid residues of IP4BP2b were identical with mouse myelin proteolipid protein (PLP) (Table I). With the sequencing method used, cysteine (C) and arginine (R) are undetectable.

Table I. Comparison of NH2-terminal amino acid sequences from IP4BP2b and mouse PLP/DM-20 Amino acid sequencing was done as described under ``Experimental Procedures.'' ? refers to an amino acid that was not positively identified. Mouse 1        10  PLP/DM20 GLLECCARC LVGAPFA         IP4BP2b GLLE??A?? LVGAP

Purified IP4BP2b after Sephacryl S-300 gel filtration chromatography and purified myelin sample as a positive control were applied to SDS-PAGE, and the proteins were electrotransferred to nitrocellulose membranes. The membranes were either stained with Amido Black (Fig. 4, lanes 1 and 4) or analyzed immunochemically using monoclonal antibodies against PLP (epitope: amino acid residues number 209-217 (AH7-2a) or 264-276 (AA3)). Fig. 4 (lanes 2 and 3) shows that monoclonal antibodies against PLP recognized the 26,000 molecules. However, these monoclonal antibodies have also been shown to recognize DM-20 (26), an alternative splicing variant of PLP. Immunoblot analysis of the purified myelin containing both PLP and DM-20 revealed that the 26,000 band comigrated with the band corresponding to PLP (Fig. 4). These observations suggested that IP4BP2b (26,000 band) is PLP.

To further confirm that IP4BP2b is PLP, we determined whether the InsP₄ binding activity of IP4BP2b could be immunoabsorbed by anti-PLP antibody. The purified sample obtained by Sephacryl S-300 gel filtration chromatography was incubated with Protein G-Sepharose resin coupled with AA3-IgG or normal rat IgG, or non-coupled. The resins were poured into columns and washed with washing solution (100 µl × 4). The nonadherent fractions were collected, and assayed for their [³H]InsP₄ binding activity and analyzed on SDS-PAGE with silver staining (Fig. 5, A and B). The InsP₄ binding activity and 26,000 protein bands were immunoabsorbed by AA3-Protein G-Sepharose (Fig. 5, A and B; +AA3, Fraction No. 2), but not by non-coupled Protein G-Sepharose or normal rat IgG-Protein G-Sepharose (Fig. 5, A and B, IgG and +Rat IgG; Fraction No. 2). These results together with the NH₂-terminal sequences indicate that the purified IP4BP2b is in fact mouse PLP.

Analysis of the InsP₄ binding described thus far was performed under the hypotonic conditions. To investigate its physiological significance, we measured InsP₃, InsP₄, and InsP₆ binding activities of purified IP4BP2b (PLP) in an isotonic buffer containing 0.15 M KCl, 20 mM HEPES-KOH, at pH 7.2. Binding activity was detectable only against InsP₆ (data not shown).

These findings suggested that InsP₆ is the true ligand for PLP. Therefore, the K_d and B_max under isotonic conditions were determined for [³H]InsP₆ binding. Since the purified PLP (Sephacryl S-300 fraction) was unstable and occasionally showed two types of binding sites (high and low affinity) (data not shown), we determined the K_d and B_max of heparin-agarose fraction, which showed only one type (high affinity) of binding site and in which most, if not all, InsP₆ binding activity is attributable to PLP. Scatchard analysis of InsP₆ binding to the heparin-agarose fraction showed that the K_d was 52 nM, the B_max 6.5 pmol/µg of protein (Fig. 6). This value of K_d was nearly the same as that of high affinity binding site of the purified PLP.

The specificity of the InsP₆-binding site was characterized by adding several inositol polyphosphates to the fraction containing purified PLP (Sephacryl S-300 fraction). We used the purified PLP in this experiment to rule out the presence of other InsP_X-binding proteins (Table II). While InsP₆ suppressed [³H]InsP₆ binding, Ins-1,4,5-P₃, Ins-1,3,4,5-P₄, and Ins-1,3,4,5,6-P₅ displayed much lower affinity. PP-InsP₅ displaced [³H]InsP₆ binding with higher potency than InsP₆. It appeared that PP-InsP₅ had a higher affinity for PLP than InsP₆.

Table II. Inhibition of specific [3H]InsP6-binding by various inositol phosphates Binding assay mixtures contained 0.17 µg of the purified IP4BP2b/PLP, 2.4 nM [3H]InsP6, 50 or 200 nM inositol phosphates, 20 µg of -globulin in 50 mM HEPES-KOH at pH 7.2 containing 0.15 M KCl (100 µl) (isotonic condition). Nonspecific binding was determined by removing the sample. Samples were incubated for 10 min at 0 °C and binding activity was measured by the polyethyleneglycol precipitation method described under ``experimental procedures.'' Each value represents the mean from two independent duplicated experiments. Displacing agent Total specific binding of [3H]InsP6 (% of control) 50 nM 200 nM % Ins(1,4,5)P3 100 90 Ins(1,3,4,5)P4 100 100 Ins(3,4,5,6)P4 100 88 Ins(1,3,4,5,6)P5 80 85 InsP6 75 55 PP-InsP5 55 12

All of the results obtained thus far clearly indicate that PLP has InsP₆ binding activity. However, it is not unknown whether DM-20 has this activity. During purification of IP4BP2b/PLP, DM-20 separated from PLP at the heparin-agarose step. DM-20 was recovered from the heparin-agarose flow-through fraction, although we could not use this fraction to study the InsP₆ binding activity of DM-20 because it also contained PLP as revealed by immunoblot analysis (data not shown). To separate DM-20 from PLP, we devised several modifications of the purification method.

First, we changed the pH of the DE-52 flow-through fraction from 8.0 into 5.0, by dilution with acetate buffer to achieve pH 5.0. The DE-52 flow-through fraction used was the same as that of the IP4BP2b/PLP purification procedure. DM-20 separated from PLP and was recovered from the flow-through fraction after CM-52 chromatography at pH 5.0. The adsorbed fraction did not contain DM-20. Because both PLP and DM-20 are extremely hydrophobic and the InsP₆ binding activity of PLP was apparently stable in the solution containing 1% Triton X-100, we performed the CM-52 chromatography with this solution. PLP separated from the other concomitant proteins (including IP4BP2a/aldolase A) by elution with a 0.05-0.5 M NaCl gradient. Both the CM-52 flow-through fraction containing DM-20 and the CM-52 adsorbed fraction containing PLP were concentrated by heparin-agarose chromatography and the buffer detergent was changed from Triton X-100 to CHAPS because the InsP₆ binding activity was inhibited more by Triton X-100 than by CHAPS. After these steps, we obtained fractions containing either PLP or DM-20. PLP or DM-20 was the major protein in PLP- or DM-20-containing fractions, respectively (Fig. 7A). Importantly, as shown by immunoblot analysis (Fig. 7B), PLP-containing fraction did not contain a detectable amount of DM-20 and either DM-20-containing fraction did not contain detectable amounts of PLP.

We measured the InsP₆ binding activity of PLP-containing and DM-20-containing fractions at equal volumes. Since CHAPS also inhibited the InsP₆ binding activity, although to a lesser extent than Triton X-100, we had to make the same dilution of the samples. The concentration of DM-20 was similar or slightly higher than that of PLP as shown semiquantitatively by immunoblot analysis (Fig. 7B). However, the InsP₆ binding activity of the DM-20-containing fraction was not detectable and only the PLP-containing fraction showed InsP₆ binding activity (Fig. 7C).

DISCUSSION

Recently, inositol polyphosphates (InsP_X), such as InsP₄, InsP₅, InsP₆, and PP-InsP₅ have been shown to accumulate intracellularly in response to several stimuli (15, 16). We identified three InsP₄-binding proteins (IP4BP); one, an IP4BP from the DE-52 adsorbed fraction, was called IP4BP1 and the others, from the DE-52 flow-through fraction, were called IP4BP2a and IP4BP2b (Fig. 1). We previously showed that IP4BP1 is synaptotagmin II (30). We also purified IP4BP2a and identified it as aldolase A, which is one of the three isoforms of fructose 1,6-bisphosphate aldolase.² A similar observation had already been reported by Koppitz et al. (31). In this study, we succeeded in purifying IP4BP2b and identifying it as PLP.

Evidence for identifying IP4BP2b as PLP includes: (i) identity of their NH₂-terminal amino acid sequences (Table I); (ii) nearly the same molecular weights of IP4BP2b, as determined by SDS-PAGE (26K), and of PLP (Fig. 3); (iii) immunoreactivity of 26,000 molecules against anti-PLP monoclonal antibodies (AA3 and AH7-2a) (Fig. 4), and immunoabsorption of InsP₄ binding activity of IP4BP2b by AA3 antibody (Fig. 5). From these results, we concluded that the purified IP4BP2b is mouse PLP. In addition, the further analysis of InsP_X binding activity of PLP under the isotonic condition demonstrated that PLP is not really IP4BP but InsP₆-binding protein.

PLP is a major integral membrane protein of central nervous system myelin. The amino acid and nucleotide sequences of cow, rat, mouse, and human PLP are closely homologous (32, 33, 34, 35, 36, 37, 38, 39, 40), and mutations within the PLP gene cause severe dysmyelination (8). Thus, PLP seems to play a significant role in central nervous system myelination, presumably by promoting the apposition of extracellular surfaces of the myelin lamellae. However, many PLP mutations also result in profound abnormalities in premyelinating oligodendrocytes. These include (a) a decrease in the number of mature oligodendrocytes, (b) premature cell death of oligodendrocytes, (c) abnormal oligodendrocyte inclusions and organelle distentions, and (d) increased oligodendrocyte proliferation (41). Therefore, it is important to understand the premyelinating functions of PLP. We have shown that expression of the PLP gene results in secretion of a factor influencing oligodendrocyte development (42). The present results indicate that another function of PLP is the binding of InsP₆.

Recently, some InsP_X-binding proteins have been reported by several groups. Partial amino acid sequencing of one protein has revealed that it is clathrin assembly protein 2 (AP-2), which is possibly an essential protein in the endocytotic or recycling pathway of all cells (17, 18, 19, 20). The K_d of InsP_X-binding protein toward InsP₆ reported by Theibert et al. (17) is 12 nM and that reported by Chadwick et al. (18) is 120 nM. Dependence of the InsP_X binding on the salt concentration in the assay system seems to account for this difference. Another clathrin assembly protein, AP-3, has also been reported to have InsP_X binding activity (23, 24). AP-3 is expressed in neurons and localized to synapses. It has been suggested that AP-3 is involved in synaptic vesicle biogenesis and recycling (43). The K_d value for InsP₆ reported by Norris et al. (23) is 1.2 µM, that by Ye et al. (24) 239 nM. In addition, coatomer, which is a cytosolic protein complex containing subunits of non-clathrin-coated Golgi intercisternal transport vesicles, was shown to have the InsP_X binding activity (K_d for InsP₆ = 0.2 nM, K_d for InsP₄ = 0.1 nM) by Fleischer et al. (25).

We also found that synaptotagmin II is an InsP_X-binding protein (30). Synaptotagmin is an integral membrane protein of synaptic vesicles considered to play a significant role in the docking and fusion of synaptic vesicles at presynaptic release sites (44, 45).

From the viewpoint of InsP_X affinity, the IP4BP2b/PLP purified in this study resembles IP4BP1/synaptotagmin II and several proteins involved in vesicular transport (AP-2, AP-3, and coatomer), whereas IP4BP2a/aldolase A does not fit into any groups because of its comparatively lower affinity and different specificities for InsP_X (31).²

PP-InsP₅ is a newly discovered pyrophosphorylated derivative (46, 47). The inhibition of InsP₆ binding activities of AP-3 and coatomer by PP-InsP₅ was stronger than by InsP₆ (24, 25). PP-InsP₅ also inhibited InsP₆ binding to PLP more strongly than InsP₆. This suggests that PLP is one of the member of a family of InsP_X-binding proteins, including AP-2, AP-3, and coatomer.

All of these reports indicate that InsP_X-binding protein is involved in vesicular transport, suggesting the involvement of PLP in vesicular transport. Abnormal transport of PLP found in the PLP mutants further supports this hypothesis (see Introduction).

The current model of PLP topology in the plasma membrane was proposed by Popot et al. (48) and Weimbs et al. (49). In this model, PLP has four -helical transmembrane regions, two pairs of disulfide bonds (Cys¹⁸³-Cys²²⁷, Cys²⁰⁰-Cys²²⁹), and six cysteine residues (in positions 5, 6, 9, 108, 138, and 140) acylated with long-chain fatty acids. The NH₂ terminus and the COOH terminus of PLP are on the cytoplasmic side. Therefore, highly basic residues (amino acid residues number 115-150) specific for PLP are on the cytoplasmic side between the second and third loop, which is spliced out in the DM-20 molecule. Thus far, the structure of DM-20 is thought to be the same as PLP with a small cytoplasmic domain between the second and third loop. Since we detected InsP₆ binding activity in PLP but not in DM-20 (Fig. 7), it is suggested that the PLP-specific domain contains the InsP₆-binding site. Therefore, the conclusion can be drawn that the ancestral DM-20 molecule acquired the PLP-specific domain as an InsP₆-binding site (or at least changed conformation of other part of DM-20 to create InsP₆-binding site).

The InsP_X-binding site of synaptotagmin II was determined to be in the C2B region (amino acid residues, 315-346) (50, 51). Although the C2A domain also contains a lysine-rich sequence, InsP₄ bound only to the C2B domain. This observation indicated that the C2A and C2B domains of synaptotagmin II have different conformations and functions. The C2B domain of synaptotagmin is highly conserved from C. elegans to humans (52, 53, 54, 55), suggesting that the InsP_X binding capacity has also been maintained during evolution. On the other hand, Voglmaier et al. (20) reported that the InsP₆-binding site of AP-2 lies on an -subunit of AP-2 as shown by a specific photoaffinity label. The details of the InsP₆-binding site of AP-2 have not yet been clarified. In addition, Ye and Lafer (56) showed that the InsP₆-binding site of AP-3 exists in the 33,000 amino terminus of AP-3. This NH₂-terminal region is known to be a basic and clathrin-binding domain (57). However, since there are few similarities among these InsP₆-binding proteins (including PLP) in their primary structure, the consensus sequence of the InsP₆-binding site is still unknown. Furthermore, the InsP₆-binding sites of proteins other than synaptotagmin II (PLP, AP-2, AP-3, and coatomer) have not been precisely determined. In all likelihood, the InsP₆-binding sites are in the basic region and are governed by as yet unknown mechanisms.

In conclusion, we identified PLP as an InsP₆-binding protein and showed that DM-20 does not have InsP₆ binding activity. InsP₆ might regulate the vesicular transport of PLP having PLP-specific domain as InsP₆-binding site. We are currently investigating how InsP₆ affects the transport of PLP to the myelin membrane.