Polarized targeting of peripheral membrane proteins in neurons.

Differential targeting of neuronal proteins to axons and dendrites is essential for directional information flow within the brain, however, little is known about this protein-sorting process. Here, we investigate polarized targeting of lipid-anchored peripheral membrane proteins, postsynaptic density-95 (PSD-95) and growth-associated protein-43 (GAP-43). Whereas the N-terminal palmitoylated motif of PSD-95 is necessary but not sufficient for sorting to dendrites, the palmitoylation motif of GAP-43 is sufficient for axonal targeting and can redirect a PSD-95 chimera to axons. Systematic mutagenesis of the GAP-43 and PSD-95 palmitoylation motifs indicates that the spacing of the palmitoylated cysteines and the presence of nearby basic amino acids determine polarized targeting by these two motifs. Similarly, the axonal protein paralemmin contains a C-terminal palmitoylated domain, which resembles that of GAP-43 and also mediates axonal targeting. These axonally targeted palmitoylation motifs also mediate targeting to detergent-insoluble glycolipid-enriched complexes in heterologous cells, suggesting a possible role for specialized lipid domains in axonal sorting of peripheral membrane proteins.

Proper neuronal function requires selective protein targeting to specialized cellular and plasma membrane domains including the nerve terminal, node of Ranvier, axon hillock, and postsynaptic density. An early step in this targeting decision tree involves a polarized sorting of proteins to either dendritic (postsynaptic) or axonal (presynaptic) domains. However, the mechanisms by which neurons target specific proteins to dendrites versus axons are poorly understood. Better characterized is protein sorting to apical versus basolateral plasma membranes in polarized epithelial cells, which share certain features with axonal versus dendritic targeting in neurons (1,2). That is, short cytosolic C-terminal protein-sorting motifs are one route for both dendritic and basolateral targeting (2), whereas specialized lipid rafts can mediate both axonal and apical sorting of certain transmembrane and glycosylphos-phatidylinositol-anchored membrane proteins (3).
The concept of specialized lipid rafts mediating polarized protein targeting emerged from observations that apical and basolateral cell membranes have different lipid compositions. Apical membranes are enriched in sphingolipids that aggregate with cholesterol to form packed raft-like domains within the fluid membrane bilayer. These rafts are insoluble in nonionic detergents and, hence, are termed detergent-insoluble glycolipid-enriched complexes (DIGs). 1 These complexes form in the trans-Golgi network and incorporate certain transmembrane, GPI-anchored, and dually acylated proteins, which are then targeted to the apical plasma membrane (4,5). The inhibition of DIG formation by sphingolipid or cholesterol depletion disrupts this apical/axonal sorting pathway (3,6,7). However, the polarized targeting of cytosolic proteins via DIGs has not been explored.
Postsynaptic density-95 (PSD-95) is a peripheral membrane protein that localizes exclusively to the PSD in hippocampal neurons and is believed to mediate the targeting and assembly of other synaptic proteins, including neurotransmitter receptors and signaling enzymes (8 -11). The N terminus of PSD-95 is posttranslationally modified with palmitate (12), a 16-carbon-saturated fatty acid linked via thioester bonds to specific cysteine residues (13)(14)(15). Dual palmitoylation of PSD-95 is necessary for appropriate postsynaptic localization (16,17). However, not all dually acylated proteins are found at postsynaptic membranes; GAP-43 is a dually palmitoylated protein that occurs predominantly at axonal membranes (18). Both PSD-95 and GAP-43 accumulate in the secretory pathway in a palmitoylation-dependent manner (17,19), but it is unclear how they sort to separate vesicles destined for dendritic versus axonal membranes.
To elucidate mechanisms for axonal versus dendritic sorting of peripheral membrane proteins, we analyzed the polarized targeting of PSD-95 and GAP-43 in hippocampal neurons. We find that the palmitoylation motif of PSD-95 is necessary but not sufficient for dendritic targeting, whereas the palmitoylation motif of GAP-43 is sufficient for axonal targeting. Systematic mutagenesis of these two palmitoylation motifs reveals that axonal targeting by the GAP-43 motif requires two adjacent cysteines as well as nearby basic residues, features that are conserved in other palmitoylated axonal proteins. Palmitoylation motifs that mediate axonal targeting also localize to DIGs in heterologous cells, indicating that lipid rafts probably mediate axonal targeting of certain cytosolic proteins.  1 The abbreviations used are: DIG, detergent-insoluble glycolipidenriched complexes; PSD, postsynaptic density; GAP, growthassociated protein; GFP, green fluorescent protein; HBS, 150 mM NaCl, 20 mM Hepes, pH 7.4; MAP, microtubule-associated protein; A/D ratio, ratio of axonal versus dendritic expression; ANOVA, analysis of variance; TEE, 50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA; PAGE, polyacrylamide gel electrophoresis; SAP, synapse-associated protein; par, paralemmin.

EXPERIMENTAL PROCEDURES
cDNA Cloning and Mutagenesis-GW1 PSD-95, PSD-95(C3,5S), and PSD-95(1-26) fused to GFP were described previously (16,17). The mutations of the palmitoylation motif of PSD-95, the addition of the GAP-43 N-terminal palmitoylation motif to PSD-95, and the mutations within the GAP-43 palmitoylation motifs of the 43-PSD-95 chimera were constructed with oligos encoding the appropriate wild-type or mutated motif and restriction sites that were annealed and subcloned into GW1 PSD-95 GFP at a HindIII site upstream of the starter methionine and a silent KpnI site at amino acid 13 of PSD-95. The addition of the C-terminal prenyl-palmitoylation motif of paralemmin was added to the extreme C terminus of PSD-95(C3,5S) GFP with primers encoding the appropriate wild-type or mutated motif and restrictions sites, which were used to amplify the C-terminal GFP. Dr. David Sretavan (University of California, San Francisco) kindly provided wild-type GAP-43. Paralemmin was obtained by reverse transcriptase-polymerase chain reaction from mouse brain RNA and subcloned into pEGFP (CLONTECH) at the BglII and HindIII sites.
Primary Neuronal Culture and Transfection-Neuronal cultures were prepared from the hippocampi of E18/E19 rats. Hippocampi were dissociated by enzyme digestion with papain followed by brief mechanical trituration. Cells were plated on poly-D-lysine (Sigma)-treated glass coverslips (12 mm in diameter) and maintained in neurobasal media (Life Technologies, Inc.) supplemented with B27, penicillin, streptomycin, and L-glutamine as described in Brewer et al. (39). Hippocampal cultures were transfected by lipid-mediated gene transfer just before plating as described previously (20). 2 g of DNA and 10 l of 1,2dioleoyl-sn-glycero-3-trimethylammonium-propane (Roche Molecular Biochemicals) were mixed in 25 l of HBS and added to the cells (1 million cells/0.25 ml) with immediate and gentle mixing. Cells were incubated for 1 h at 37°C and then plated at a density of 600/mm 2 on glass coverslips (Fisher) in 24-well plates (Falcon). To visualize transfected cells, coverslips were removed from the wells and mounted live onto slides (Frost Plus slides, Fisher) with Fluoromount-G (Southern Biotechnology Associates, Inc.). Transfection efficiency was never Ͼ0.01%, and on average, 15-30 transfected cells were obtained for each independent transfection.
Immunofluorescence-Coverslips were removed from culture wells and fixed in 4°C paraformaldehyde for 15-20 min. The cells were washed with Tris-buffered saline containing 0.1% Triton X-100 Trisbuffered saline and blocked in Triton X-100 Tris-buffered saline with 3% normal goat serum for 1 h at room temperature. Primary antibodies against MAP-2 (monoclonal) (Pharmingen) or Thy-1 (MRC OX-7, Serotec) to stain dendrites or axons, respectively, were added to blocking solution for 1 h at room temperature followed by donkey anti-mouse or donkey anti-rabbit antibodies conjugated to Cy3 or 7-amino-4-methylcoumarin-3-acetic acid fluorophores (diluted 1:200 in blocking solution) for 1 h. at room temperature. Coverslips were then mounted on slides (Frost Plus) with Fluoromount-G, and images were taken under fluorescence microscopy with a ϫ 60 objective affixed to a Zeiss-inverted microscope.
Quantitative Measurement of Polarized Protein Expression-Quantification of polarized protein expression in dendrites versus axons was performed on 15-50 neurons from 2-3 independent transfections. Images of neurons were acquired with a charge-coupled device camera and quantitated using Metamorph imaging software (Universal Imaging). The exposure time of the camera was adjusted to limit photobleaching, so that the maximum pixel intensity was approximately one-half to three-fourths saturating for cells with low to moderate expression levels as determined by total pixel counts. Because high protein expression can saturate targeting mechanisms, cells expressing to pixel saturation were not included in the analysis.
The degree of polarized expression was determined by calculating the average pixel intensity in the axon versus that in the dendrites. The axon was identified as the thin process extending farthest from the cell body and not immunoreactive for MAP-2. The average pixel intensity was calculated by drawing a line through three dendrites and three sections of the axon at defined distances from the cell body. These averages were then converted into a ratio of axonal versus dendritic expression (A/D ratio) and compared with that for diffusely expressed GFP. The A/D ratio represents the percentage of protein in the axon as compared with that in the dendrites, and because the volume of the axon is considerably smaller than that of the dendrites, the ratio is always Ͻ1. Data were analyzed by one-way ANOVA with Bonferroni corrections for multiple comparisons with Prism software (GraphPAD, San Diego, CA).
Cell Transfection, Metabolic Labeling, and Immunoprecipitation-COS-7 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, penicillin, and streptomycin. Cells were transfected using LipofectAMINE reagent according to the manufacturer protocol (Life Technologies, Inc.). For studies of palmitoylation, transfected COS-7 cells were labeled in media containing 1 mCi/ml [ 3 H]palmitic acid (50 Ci/mmol) (DuPont NEN). Cells were washed with ice-cold phosphate-buffered saline and resuspended in 0.4 ml of lysis buffer containing TEE, 150 mM NaCl, and 0.2% SDS. After extracting for 20 min at 4°C, Triton X-100 was added to 1% to neutralize the SDS, and insoluble material was removed by centrifugation at 10,000 ϫ g for 10 min. For immunoprecipitation experiments, the samples were then incubated with GFP antibodies (1:150 dilution, CLONTECH) for 1 h at 4°C. After the addition of 20 l of protein A-Sepharose beads (Amersham Pharmacia Biotech), samples were incubated for 1 h at 4°C. Immunoprecipitates were washed three times with buffer containing TEE, 150 mM NaCl, and 1% Triton X-100, boiled in SDS-PAGE sample buffer with 1 mM dithiothreitol for 2 min, and analyzed by SDS-PAGE. For fluorography, protein samples were separated by SDS-PAGE and stained with Coomassie Blue. Gels were treated with Amplify (Amersham Pharmacia Biotech) for 30 min, dried under vacuum, and exposed to Hyperfilm-MP (Amersham Pharmacia Biotech) at Ϫ80°C for 3-5 days.

COS-7 Cell Transfection and Isolation of Detergent-insoluble Glycolipid-enriched
Complexes-Detergent-insoluble fractions were prepared as described previously with modifications (21). COS-7 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, penicillin, and streptomycin. Cells were transfected using Lipo-fectAMINE reagent according to the manufacturer protocol (Life Technologies, Inc.). Cells (50 ϫ 10 6 ) were washed with ice-cold phosphatebuffered saline and resuspended in 1 ml of lysis buffer containing 25 mM Tris-HCl, pH 7.6, 5 mM EDTA, 150 mM NaCl, 0.5% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml each of aprotinin and leupeptin. After extracting for 10 min at 4°C, the homogenate was adjusted to 40% sucrose and placed at the bottom of an ultracentrifuge tube. A linear sucrose gradient (5-30%) was overlaid on top, and samples were centrifuged for 18 h at 200,000 ϫ g in a SwTi 50 rotor (46,245 rpm) at 4°C. Sequential fractions across the gradient were collected, separated by SDS-PAGE, and immunoblotted using antibodies to caveolin-1 (Transduction Laboratories, Lexington, KY) and GFP (1:150 dilution, CLONTECH).

Palmitoylation Is Required for Axon Exclusion of PSD-95-
PSD-95 and SAP-97 are highly homologous membrane-associated guanylate kinase proteins, however, PSD-95 is restricted to postsynaptic sites in forebrain neurons, whereas SAP-97 occurs both pre-and postsynaptically (11). To test whether this polarized sorting can be reproduced in cell transfections, we expressed GFP fusions of PSD-95 or SAP-97 in hippocampal neurons ( Table 1). As shown in Fig. 1, we find that PSD-95 is restricted to postsynaptic clusters in the dendrites, whereas SAP-97 occurs both in dendrites and axon. As PSD-95 is palmitoylated and SAP-97 is not, we asked whether differential lipid modifications of these proteins account for their differential sorting to axons versus dendrites. Indeed, we found that mutating the palmitoylated cysteines of PSD-95 to serine (PSD-95(C3,5S)) disrupts axon exclusion and causes the mutant to enter the axon. Conversely, appending the palmitoylated N terminus of PSD-95 to SAP-97 yields a chimera that is excluded from the axon (Fig. 1).
Axons were unambiguously identified both by their morphology and by the absence of MAP-2 immunoreactivity (Fig. 1). To quantitate the extent of axonal targeting, the average fluorescent pixel intensity of GFP fusions in the axon versus the dendrites were determined (A/D ratio) (see under "Experimental Procedures") and compared with the A/D ratio of GFP alone. This analysis revealed that wild-type PSD-95 protein and 95-SAP-97 are excluded from the axon, whereas SAP-97 and PSD-95(C3,5S) are present in the axon to a similar extent as is GFP alone (Fig. 2C).
The Palmitoylation Motif of GAP-43 Targets PSD-95 to the Axon-Whereas palmitoylation of PSD-95 is required for axon exclusion, not all palmitoylated neuronal proteins are re-stricted to dendrites. For example, dually palmitoylated GAP-43 predominantly localizes to axonal membranes (18). Given these different localizations, we asked whether the palmitoylation motifs play opposing roles in protein trafficking. To accomplish this task, we first replaced the dually palmitoylated N-terminal 13 amino acids of PSD-95 with those of GAP-43. The GAP-43 palmitoylation motif on PSD-95 (43-PSD-95) maintains palmitoylation and partially maintains postsynaptic targeting (17). However, the 43-PSD-95 chimera also dramatically targets to the axon, a localization not observed with wild-type PSD-95 (Fig. 2).
Differential Targeting by the Palmitoylation Motifs of PSD-95 and GAP-43-Given the relocation of PSD-95 by the GAP-43 palmitoylation motif, we next assessed differential targeting mediated by these motifs. When the PSD-95 palmitoylation motif alone is fused to GFP, the A/D ratio of PSD-95(1-26) is not statistically different from GFP (Fig. 2). In contrast, the GAP-43 palmitoylation motif, fused to GFP (GAP-43 (1-14)) though not exclusively axonal, is enriched in the axon compared with GFP alone (Fig. 2). These data suggest that the palmitoylation motif of PSD-95 is necessary but not sufficient for axon exclusion, whereas the palmitoylation motif of GAP-43 is sufficient for axonal targeting.
We considered the possibility that axon exclusion of PSD-95 may simply result from its clustering at postsynaptic sites. However, we find that the 43-PSD-95 chimera is both more postsynaptically clustered and more axonally localized than PSD-95(C3,5S), indicating that postsynaptic clustering and axonal exclusion are separable processes.
To help to understand the constituents of the GAP-43 palmitoylation motif necessary for axonal targeting, we compared the sequences of the PSD-95 and GAP-43 palmitoylation motifs. The palmitoylated cysteines of the GAP-43 motif are adjacent, whereas those in PSD-95 are separated by a leucine. Strikingly, a PSD-95 mutant containing adjacent cysteines (PSD-95(CC)) is not exclusively dendritic; the protein also localizes to the axon though not to the degree of 43-PSD-95 (Fig.  3). On the other hand, the addition of a single amino acid between the contiguous cysteines of the GAP-43 palmitoylation motif (43-PSD-95(CXC)) does not affect axonal targeting of 43-PSD-95, and the A/D ratio is unchanged (Fig. 3). These mutated constructs are all efficiently palmitoylated (Fig. 5), so changes in protein targeting are not attributed to alterations in protein palmitoylation. These results suggest that the spacing of the cysteines is important but is not the only feature of these motifs that determines targeting.
Additionally, the GAP-43 motif contains two basic amino acids that is one residue away from the palmitoylated cysteines, whereas the PSD-95 motif does not have basic residues near the cysteines. Remarkably, adding a single basic amino acid to PSD-95 two amino acids away from the cysteines (PSD-95(T8R)) redistributes the mutant to the axon with almost half the efficiency of 43-PSD-95 (Fig. 3). Conversely, mutating the basic amino acids of the GAP-43 palmitoylation motif to isoleucines (43-PSD-95(R6I,R7I)) does not alter protein palmitoylation but reduces axonal targeting of 43-PSD-95 by approximately half (Fig. 3). These results suggest that the basic amino acids are critically involved but do not entirely explain the targeting differences between these motifs.
Finally, we combined alternations in spacing between the cysteines and mutations in the basic amino acids. A PSD-95 palmitoylation motif with juxtaposed cysteines and the T8R mutation (PSD-95(CC-T8R) targets PSD-95 to the axon similar to 43-PSD-95 (Fig. 3). In addition, a GAP-43 palmitoylation motif both with cysteines separated and with mutations in the basics (43-PSD-95(CXC-R6I,R7I)) no longer targets 43-PSD-95 to the axon, but rather the protein localizes solely to the postsynaptic membrane similar to wild-type PSD-95 (Fig. 3). These results suggest that these two features account entirely for differential targeting by these domains.
The Palmitoylation Motif of Paralemmin Resembles That of GAP-43 and Mediates Axonal Targeting-We next asked whether the adjacent cysteine/basic amino acid motif might play a general role in the axonal sorting of palmitoylated proteins. Paralemmin is a neuronal prenyl-palmitoyl-anchored protein that is found at axonal membranes (22). The C-terminal palmitoylation motif of paralemmin contains adjacent prenylated/palmitoylated cysteines and nearby basic amino acids and thereby resembles the lipidated domain of GAP-43 (Fig. 4). GFP-tagged paralemmin is significantly targeted to the axon, similar to GAP-43. Furthermore, the isolated prenyl-palmitoylation motif of paralemmin fused to GFP (GFP-par) is sufficient for axonal targeting (Fig. 4), and the addition of this motif to the C terminus of palmitoylation-deficient PSD-95 targets the chimera (PSD-95(C3,5S-par)) to the axon (Fig. 4).
We next determined whether axonal targeting by the paralemmin motif requires similar features as the GAP-43 motif.  Adjusting the spacing of the cysteines interferes with prenylation and subsequent palmitoylation (data not shown), so we were unable to assess the importance of cysteine residue spacing in protein trafficking. However, mutating the palmitoylated cysteines to serines (PSD-95(C3,5S-par-palmitoyl)) re-tains prenylation but blocks palmitoylation (Fig. 5) and disrupts axon targeting (Fig. 4). In addition, mutations in the basic amino acids maintain lipidation but reduce axonal targeting, consistent with a requirement for nearby basic amino acids in axonal targeting (Fig. 4).

Targeting of Peripheral Membrane Proteins in Neurons 44987
Palmitoylation Motifs That Mediate Axonal Targeting Are Incorporated into DIGs-Dually acylated proteins can be incorporated into DIGs, and these complexes have been implicated in targeting to axonal membranes (4,7,23). We, therefore, asked whether the palmitoylation motifs of GAP-43 and PSD-95 might differentially associate with DIGs. Resident proteins of DIGs float in sucrose gradients and are found in light membrane fractions together with ␣-caveolin (24). As previously published, only a small amount of PSD-95 associates with DIGs, and this is independent of palmitoylation ( Fig. 6) (25,26). In contrast, the palmitoylation motif of GAP-43 efficiently targets a GFP reporter to DIGs (Fig. 6) (40) as does the axonally targeted prenyl-palmitoylation motif of paralemmin (Fig. 6). Furthermore, when the palmitoylation motif of PSD-95 is replaced with that of GAP-43 or paralemmin, these chimeras also associate with DIGs, whereas the 43-PSD-95(R6I,R7I) shows an intermediate degree of DIG partitioning (Fig. 6, and data not shown). Thus, there is a strong correlation between the ability of palmitoylation motifs to target proteins to DIGs and to axonal membranes. DISCUSSION This analysis of polarized sorting of peripheral membrane proteins demonstrates that palmitoylation motifs can mediate either dendritic/postsynaptic or axonal targeting. Previous work shows that dual palmitoylation is necessary for targeting PSD-95 to postsynaptic membranes (16,17). We now find that palmitoylation is also necessary to exclude PSD-95 from axons, although the isolated palmitoylation motif is not sufficient for axonal exclusion. In contrast, the dually palmitoylated motifs of the axonal proteins GAP-43 and paralemmin are sufficient to mediate protein targeting to axonal membranes and, therefore, are the first identified axonal targeting motifs for peripheral membrane proteins.
Differential sorting of the PSD-95 and GAP-43 palmitoylation motifs depend on two features, the spacing of the cysteine residues and the presence of nearby basic amino acids. The PSD-95 and GAP-43 palmitoylation motifs also differ in their capacity to associate with DIGs; PSD-95 is only faintly incorporated into DIGs, whereas the isolated palmitoylation motif of  GAP-43 is sufficient for association with these complexes. These data suggest that the incorporation of peripheral membrane proteins into lipid rafts may mediate axonal trafficking.
Whether dendritic sorting of PSD-95 results from active dendritic targeting or axonal exclusion is unclear. Interestingly, when axonally targeted palmitoylation motifs from GAP-43 and paralemmin are added to PSD-95, the resulting chimeras are not as polarized as their isolated palmitoylation motifs, nor are they as well associated with DIGs. These results suggest that additional dendritic targeting or axonal exclusion signals reside within PSD-95. Consistent with this finding, the palmitoylation motif of PSD-95 is insufficient for dendritic targeting. Yet because unpalmitoylated PSD-95(C3,5S) is distributed similar to GFP, the expression of the dendritic targeting signal within the body of PSD-95 probably requires an association with membranes via palmitoylation. The identity of the additional region(s) of PSD-95 involved in dendritic targeting/ axonal exclusion remains to be uncovered.
In contrast to the absolute polarization of dendritically targeted PSD-95, axonally targeted constructs, such as GAP-43 and paralemmin, are also expressed in dendrites. These proteins are considered axonal because of their enhanced density in the axon compared with diffusely expressed GFP. Previous studies have also found that exogenous expression of axonal proteins often yields some protein in dendrites (30). This dendritic expression may occur for a number of reasons, including inadequate axonal retention or missorting due to saturation of targeting mechanisms. Alternatively, dendritic localization may be explained by the presence of dendritic targeting signals and/or the absence of dendritic exclusion signals, as axonal proteins can also occur in dendritic/postsynaptic localizations (22). Indeed, we found that PSD-95 constructs containing GAP-43 or paralemmin palmitoylation motifs concentrate in the axon but also occur at the PSD.
PSD-95 and related membrane-associated guanylate kinases display complex expression patterns that depend on the specific protein isoform and neuronal cell type. In hippocampal neurons, PSD-95 and PSD-93 are excluded from axons, whereas SAP-97 and SAP-102 occur both in axons and dendrites (17). As PSD-95 and PSD-93 are palmitoylated and SAP-97, and SAP-102 are not (16), this differential targeting in vivo may reflect the lipid-dependent mechanisms as described here.
Unlike its dendritic localization in hippocampal and other forebrain neurons, PSD-95 occurs prominently in axons of cerebellar basket cells (27). Interestingly, basket cells are unusual in that their axons are devoid of microtubules (28). Axons of hippocampal and most other neurons contain plus end distal microtubules, and dendrites contain both plus and minus enddirected microtubules (29). Therefore, dendritic localization of PSD-95 may reflect a selective association with minus enddirected microtubule motors that cooperate with palmitoylation to target PSD-95 into dendrites and to exclude PSD-95 from axons. Thus, the presence of PSD-95 in basket cell axons is potentially explained by the loss of microtubule-dependent axon exclusion.
A recent study (30) has determined the axonal exclusion signal for a metabotropic glutamate receptor. This work shows that dendritic/axonal targeting of these transmembrane proteins relies on signals within their cytoplasmic C termini. By contrast, our analysis of palmitoylation-dependent sorting of peripheral membrane proteins implicates a role for DIGs. These complexes are rich in glycosphingolipids and cholesterol, and their formation in the secretory pathway is thought to serve as a sorting platform to direct proteins to the apical membrane of epithelial cells (4). Polarized sorting of proteins to axonal membranes has been compared with this apical sorting (23). Indeed, an incorporation into sphingolipid-cholesterol rafts appears to mediate axonal targeting of the GPI-anchored protein, Thy-1 (3, 7).
We find that dual palmitoylation motifs that associate with these rafts can also target proteins to axonal membranes. In contrast, the dual palmitoylation motif of PSD-95 is not incorporated into DIGs, and rather than mediating axonal targeting, it plays a role in dendritic targeting/axonal exclusion. To FIG. 6. Palmitoylation motifs that target to axonal membranes also associate with DIGs. A, the isolated palmitoylation motif of PSD-95 fused to GFP (PSD-95(1-26)) is found in the heavy fractions (lanes 8 -10) of a sucrose gradient similar to cytosolic GFP and, thus, is not significantly associated with DIGs. In contrast, the palmitoylation motif of GAP-43 (GAP-43(1-14)) or paralemmin (GFP-par) is sufficient to incorporate GFP into lipid rafts, and these proteins are found in the lighter membrane fractions. B, PSD-95 is only modestly associated with lipid rafts similar to the non-DIG-associated transmembrane protein Tac. In contrast, replacing the palmitoylation of PSD-95 with that of GAP-43 (43-PSD-95) or paralemmin (PSD-95(C3,5S-par)) localizes a portion of these chimeric proteins to lighter membrane fractions containing ␣-caveolin, a DIG-resident protein.
uncover a more direct relationship between DIG association and axonal targeting, we attempted to disrupt DIG formation using lovastatin and methyl-␤-cyclodextrin to deplete cholesterol. However, this treatment did not affect axonal targeting of endogenous Thy-1, 2 suggesting that cholesterol depletion was not successful in our cultures.
Both GAP-43 and PSD-95 are found in the secretory pathway (17,19) where lipid rafts are first formed. And it is here that these two proteins may be segregated to separate secretory vesicles for transport to dendritic versus axonal membranes. The palmitoylation motif of GAP-43 can mediate lipid raft association for protein trafficking to the axonal membrane. Indeed, previous studies have found that GAP-43 is enriched in DIGs from brain homogenates (31) and is transported to axons on vesicles derived from the secretory pathway (32). On the other hand, signals within the body of PSD-95 working together with palmitoylation may mediate its association with dendritic targeting vesicles and its exclusion from rafts. Near its C terminus, PSD-95 possesses a tyrosine-based proteintrafficking motif that is sufficient to mediate protein endocytosis via clathrin-coated vesicles and that is required for postsynaptic targeting (33). As an exogenous axonally targeted lipid raft protein, influenza virus hemagglutinin is efficiently excluded from clathrin-coated endocytotic vesicles (34,35). Adding an internalization motif to hemagglutinin induces endocytosis (35,36) and potentially blocks association with DIGs (37,38). Thus, the inclusion of PSD-95 in clathrin-coated pits may aid in its exclusion from lipid rafts.
Alternatively, PSD-95 and GAP-43 may differentially associate with DIGs because different palmitoyl-transferase enzymes recognize their distinct palmitoylation motifs. Although such transferase enzymes have not yet been isolated, the sequence specificity for palmitoylation identified here suggests that this is indeed an enzymatic process. For instance, palmitoylation motifs similar to GAP-43 and paralemmin that contain adjacent cysteines with nearby basic amino acids may be recognized by a palmitoyl-transferase enzyme in the trans-Golgi network followed by the incorporation into lipid rafts for axonal trafficking. In contrast, the palmitoylation motif of PSD-95 may be recognized by a separate enzyme that is cytosolic or associated with membranes other than the trans-Golgi network, and thus PSD-95 is not incorporated into DIGs. Identification of the putative palmitoyl-transferase enzyme(s) that mediate palmitoylation of neuronal proteins will help to clarify these issues. Furthermore, elucidating how protein palmitoy-lation contributes to polarized trafficking may provide fundamental insights for understanding many different sorting decisions in the cell.