Cholera Toxin Is Found in Detergent-insoluble Rafts/Domains at the Cell Surface of Hippocampal Neurons but Is Internalized via a Raft-independent Mechanism*

A number of studies have demonstrated that cholera toxin (CT) is found in detergent-insoluble, cholesterol-enriched domains (rafts) in various cells, including neurons. We now demonstrate that even though CT is associated with these domains at the cell surface of cultured hippocampal neurons, it is internalized via a raft-independent mechanism, at both early and late stages of neuronal development. CT transport to the Golgi apparatus, and its subsequent degradation, is inhibited by hypertonic medium (sucrose), and by chlorpromazine; the former blocks clathrin recruitment, and the latter causes aberrant endosomal accumulation of clathrin. Moreover, both internalization of the transferrin receptor (Tf-R), which occurs via a clathrin-dependent mechanism, and CT internalization, are inhibited to a similar extent by sucrose. In contrast, the cholesterol-binding agents filipin and methyl-β-cyclodextrin have no effect on the rate of CT or Tf-R internalization. Finally, once internalized, CT becomes more detergent-soluble, and chlorpromazine treatment renders internalized CT completely detergent-soluble. We propose two models to explain how, despite being detergent-insoluble at the cell surface, CT is nevertheless internalized via a raft-independent mechanism in hippocampal neurons.

Cholera toxin (CT) 1 consists of a pentameric B subunit that binds with high affinity to ganglioside GM1 and an A subunit comprising two peptides, A1 and A2, linked by a disulfide bond. The A1 subunit is responsible for activation of adenylate cyclase via the stimulatory G protein, G s . Electron microscopy analysis in A431 cells, and in cultured liver cells, demonstrated that CT does not bind uniformly over the plasma membrane but is rather concentrated in membrane invaginations (1) identified as caveolae (2). Caveolae contain the coat protein, caveolin (3), and are enriched in glycosphingolipids (GSLs) and cholesterol (4). Biochemical analysis has shown that the GSLs and cholesterol found in caveolae are insoluble in nonionic detergents at low temperature (4). However, not all cells contain caveolin or morphologically distinct caveolae. Smooth muscle cells, fibroblasts, adipocytes, endothelial cells, and epithelial cells express caveolin/caveolae, but lymphocytes and neurons do not (3). Even in cells lacking caveolae, a significant fraction of cellular cholesterol and GSLs are found in detergent-insoluble complexes (5)(6)(7), sometime known as rafts (8); and these complexes are indistinguishable, using the criteria of detergent insolubility, from those associated with caveolae (9).
Due to its association with caveolae and/or detergent-insoluble domains, it is normally assumed that CT is internalized by the pinching off of caveolae from the plasma membrane (10), followed by transport to the Golgi apparatus and endoplasmic reticulum (11,12). Recent studies in A431 cells, which express high levels of caveolin, in CaCo-2 cells, which express low levels of caveolin, and in Jurkat cells, which express no caveolin, demonstrated that the cholesterol-binding agent filipin disrupted CT internalization and the subsequent generation of cAMP in all three cell types (13), presumably due to extraction of cholesterol from caveolae or detergent-insoluble domains.
In contrast to the cell types mentioned above, less is known about how CT is internalized in neurons; neurons contain high levels of GSLs, and particularly of GM1 (14). In cultured hippocampal neurons (15), high levels of GM1 can be detected after about 2 days in culture (16), and consequently CT binding increases as neurons mature (17). CT is internalized by an energy-and temperature-dependent mechanism to the Golgi apparatus (18), and at least in CT containing the A subunit (rather than holo-CT which only contains the B subunit, but still binds GM1), to the endoplasmic reticulum (17). Transport to the Golgi apparatus and cAMP elevation are inhibited by cationic amphiphilic drugs (i.e. chlorpromazine, imipramine, and sphingosine) (18), which block receptor recycling by disrupting the assembly-disassembly of clathrin from coated pits and endosomes (19). These data are consistent with the rapid appearance of CT in the same endosomal compartment as that labeled by ␣ 2 -macroglobulin, a ligand that enters cells via clathrin-coated pits (20), and data showing that CT is not excluded from clathrin-coated pits, even though it is enriched in caveolae in cells that contain these structures (1,2).
To ascertain the mechanism of CT internalization in neurons, and the relationship of detergent insolubility to the mode of internalization, we now examine the effect of various drugs that inhibit clathrin-mediated endocytosis or bind cholesterol and hence disrupt detergent-insoluble domains. In addition, we compare CT internalization with that of the transferrin receptor (Tf-R), which is internalized exclusively via a clathrinmediated mechanism (21). We confirm that CT is found in detergent-insoluble membrane domains at the cell surface of 8 -9-day-old hippocampal neurons (22,23), as is the GPI-anchored protein Thy-1 (24), but CT is nevertheless internalized via a raft-independent mechanism. EXPERIMENTAL PROCEDURES Materials-CT was purchased from Calbiochem. Horseradish peroxidase-conjugated CT was from List Biological Laboratories, Inc., Campbell, CA. Bodipy succinimidyl ester was from Molecular Probes Inc., Eugene, OR. Na 125 I (carrier-free) was from Amersham Pharmcia Biotech. IODO-BEADS were from Pierce. Chlorpromazine (CPZ), filipin, methyl-␤-cyclodextrin (M␤CD), and Tf were from Sigma.
Incubation of Hippocampal Neurons with Bodipy-CT-Bodipy-CT was prepared by conjugation of CT with Bodipy succinimidyl ester according to the manufacturer's instructions. Hippocampal neurons cultured at low density (6,000 -12,000 cells/13-mm glass coverslip) (15,18) were incubated with 10 nM Bodipy-CT for 30 min at 13-16°C in Hanks' balanced salt solution (HBSS) containing 10 mM HEPES, pH 7.4, and 0.1% (w/v) bovine serum albumin (BSA). After washing, neurons were incubated in N2.1 medium (serum-free medium (minimal essential medium) that included the N2 supplements, ovalbumin and pyruvate (15)) at 37°C for 1 (for 2-day-old neurons) or 2 h (for 8 -9-dayold neurons). Neurons were fixed with 4% formaldehyde in phosphatebuffered saline containing 4% sucrose for 5 min at 37°C. Fluorescence microscopy was performed using Plan Neofluar 40ϫ/1.3 numerical aperture and Plan Apochromat 63ϫ/1.4 numerical aperture oil objectives of a Zeiss Axiovert 35 microscope equipped with a filter for Bodipy fluorescence. Cells were photographed using a Contax 167MT camera and Kodak Tmax p3200 film. The number of cells in which the Golgi apparatus was labeled was quantified as described (17).
To examine the effect of drugs, neurons were preincubated with sucrose, CPZ, filipin, or M␤CD in N2.1 medium without glia for 30 min at 37°C. Neurons were incubated with Bodipy-CT as described above in the continuous presence of the drugs.
CT Degradation-CT was iodinated as described (25) to a specific activity of 57 Ci/mmol. Neurons (25,000 cells/13-mm glass coverslip) were incubated with 5 nM 125 I-CT for 30 min at 13-16°C in HBSS containing 10 mM HEPES, pH 7.4, and 0.1% (w/v) BSA. After washing with HBSS, neurons were incubated in N2.1 medium without glia at 37°C for 8 h and then cooled on ice. Coverslips were removed from the culture dishes, and the radioactivity of coverslips (cell-associated CT) and trichloroacetic acid-soluble material in the culture medium (degraded CT) was determined. The amount of degraded CT was expressed as a percentage of total cell-associated radioactivity. To examine the effect of drugs on CT degradation, neurons were preincubated in N2.1 medium without glia for 30 min at 37°C. The CT degradation assay was performed as above in the presence of drugs.
CT Endocytosis-Neurons (50,000 cells/13-mm glass coverslip) were incubated with 5 nM 125 I-CT for 30 min at 13-16°C in HBSS containing 10 mM HEPES, pH 7.4, and 0.1% (w/v) BSA. After washing with HBSS, neurons were incubated in N2.1 medium without glia at 37°C for the indicated period and then placed on ice for 5 min to inhibit endocytosis. To remove cell-surface 125 I-CT, neurons on coverslips were incubated on ice for 5 min with acid buffer (0.5 M NaCl and 0.2 M acetic acid, pH 2.5), followed by two rapid washes with ice-cold HBSS. The radioactivity of coverslips (internal CT) and of the acid wash was measured. After acid washing, Ͼ95% of cell-surface CT was removed.
Tf Endocytosis-Iron loading, iodination, and measurement of Tf endocytosis was performed according to McGraw and Subtil (26). The effect of drugs on CT and Tf endocytosis was measured in the presence of drugs after preincubation of neurons with the drugs for 30 min at 37°C.
Detergent Insolubility and Flotation on Sucrose Gradients-Neurons at high density (180,000 cells/24-mm coverslip) (16) were labeled with 125 I-CT or 125 I-Tf as above and removed by scraping with a rubber policeman into ice-cold TNE buffer (25 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, pH 7.5). The cells were lysed for 20 min at 4°C with Triton X-100 (final concentration; 1%, v/v). Lysates were centrifuged for 10 min at 16,000 ϫ g at 4°C, and radioactivity in the supernatant (detergent-soluble fraction) and in the pellet (detergent-insoluble fraction) was counted.
Sucrose gradient analysis of lysates was performed as described with slight modifications (27). The cell lysate was brought to 40% sucrose using 80% sucrose in TNE buffer without Triton X-100, and placed in an ultracentrifuge tube. A discontinuous sucrose gradient was layered over the lysate. Gradients were centrifuged for 15-17 h at 200,000 ϫ g at 4°C. Fractions (400 l each) and the pellet were harvested, and radioactivity was counted. For analysis of endogenous GM1 distribution, lipids were extracted (16) from lyophilized fractions and separated by thin layer chromatography (TLC) using chloroform, methanol, 0.2% CaCl 2 (60:35:8, v/v/v) as the developing solvent. GM1 ganglioside was detected by TLC immunoblotting using horseradish peroxidase-conju-gated CT and the ECL detection system (Amersham Pharmacia Biotech). Total lipids were separated by TLC using methyl acetate, 1-propanol, chloroform, methanol, 0.25% KCl (25:25:25:10:9, v/v) as developing solvent and visualized with cupric sulfate.

CT and Ganglioside GM1 Are Detergent-insoluble in Hip-
pocampal Neurons-When the neuronal cell surface of 8-dayold neurons was labeled with 125 I-CT and extracted with 1% Triton X-100 at low temperature, 67% of the CT was detergentinsoluble ( Fig. 1A); in contrast, all of the Tf-R was detergentsoluble (not shown). After depletion of cholesterol using M␤CD (28), the detergent solubility of CT increased in a dose-dependent manner (Fig. 1A), demonstrating that cell-surface CT is associated with detergent-insoluble, cholesterol-enriched domains in hippocampal neurons.
We further analyzed detergent extracts by flotation on sucrose gradients. In young neurons (i.e. 2-day-old), 27% of cell-

FIG. 1. Detergent insolubility of CT and Tf-R.
A, detergent insolubility of cell-surface CT. Eight-day-old neurons were incubated with or without M␤CD and then incubated with 5 nM 125 I-CT at 13-16°C for 30 min. Neurons were removed from the coverslips by scraping and treated with Triton X-100 (1% v/v, 4°C, 20 min). Detergent extracts were centrifuged at 16,000 ϫ g for 10 min, and radioactivity in the supernatant (detergent-soluble fraction) and pellet (detergent-insoluble fraction) was counted. Data are means Ϯ S.D., n ϭ 2. B, sucrose gradient analysis of cell-surface CT. Eight-day-old (closed squares) and 2-day-old neurons (open squares) were labeled with 125 I-CT (5 nM, 13-16°C, 30 min) and then lysed with detergent. After centrifugation of the detergent extracts on sucrose gradients, fractions (400 l) were collected from the top (fraction 1) of the gradients. Fraction 13 is the pellet. C, distribution of Tf-R (closed circles) and endogenous GM1 (open circles) in 8-day-old neurons. Tf receptors were labeled with 125 I-Tf for 30 min at 4°C, and detergent extracts were analyzed on a sucrose gradient as in B. Endogenous GM1 ganglioside was detected by TLC immunoblotting of lipid extracts from each fraction, using horseradish peroxidase-CT; the inset shows a typical immunoblot.
surface CT was found in low density fractions (fractions 2-4, Fig. 1B), but in older neurons (i.e. 8 -9 day-old), 50% of CT was found in the low density fractions (Fig. 1B), as was Thy-1 (not shown) (23,24). Similarly, ϳ65% of endogenous GM1 was obtained in low density fractions (Fig. 1C), which also contained neutral lipids (presumably cholesterol), demonstrating that binding of CT to GM1 does not alter the distribution of endogenous GM1 on sucrose gradients. The Tf-R was found exclusively in detergent-soluble fractions (Fig. 1C), which also contained most of the glycerolipids (not shown).
Cholesterol-binding Agents Have No Effect on CT Transport to the Golgi Apparatus or on CT Degradation-To determine the relationship between detergent insolubility and the mechanism of internalization, we performed a series of experiments to quantify CT transport to the Golgi apparatus, CT degradation, and the rate of CT internalization.
CT was transported to the Golgi apparatus in control neurons (Fig. 2, A and F), with ϳ90% of cells labeled after 1 h in 2-day-old neurons ( Fig. 3) and ϳ60% of cells labeled after 2 h in 8 -9-day-old neurons (Fig. 3); the differences in the rate of Golgi apparatus labeling are similar to those reported previously (17).
Hypertonic medium (i.e. sucrose), which renders clathrin unavailable for assembly into normal coated pits (29), signifi-cantly inhibited CT transport to the Golgi apparatus in both 2- (Fig. 2B) and 8 -9-day-old neurons ( Fig. 2G), with the highest sucrose concentration (0.3 M) completely inhibiting CT transport to the Golgi apparatus (Fig. 3). Likewise, CPZ inhibited CT transport to the Golgi apparatus in 2- (Fig. 2C) and 8 -9-day-old neurons ( Fig. 2H and Ref. 18), with ϳ80 -90% inhibition of Golgi apparatus labeling (Fig. 3); CPZ disrupts receptor recycling by stimulating the recruitment of the AP-2 adaptor protein and of clathrin to an uncoated, late endosomal compartment but does not affect the initial uncoating of clathrin-coated vesicles or their transport to early endosomes (19). The inhibitory effects of sucrose (data not shown) and CPZ (18) could be restored after incubation of neurons in fresh medium that did not contain either sucrose or CPZ.
Although M␤CD reduced the detergent insolubility of CT (Fig. 1A), it did not affect the pattern of Golgi labeling (Fig. 2, D and I). However, M␤CD-treated cells generally displayed distorted cell bodies and had thinner neurites and, at high concentrations, detached from the coverslips, rendering it impossible to quantify accurately the percentage of cells in which the Golgi apparatus was labeled (Fig. 3); this was particularly apparent in 2-day-old neurons. It should be emphasized that Golgi apparatus labeling could be clearly detected in all of the cells that remained attached to the coverslips (Fig. 2, D and I).
The sterol-binding agent, filipin, had no effect on Golgi apparatus labeling (Fig. 2, E and J) even though this agent has been shown to disrupt caveolae structure and function (30,31). Filipin-treated cells also detached from coverslips, rendering accurate quantification difficult (Fig. 3).
Based on studies using brefeldin A (BFA), it was shown that CT must be transported to the Golgi apparatus prior to degradation (32). Thus, in hippocampal neurons, BFA inhibited CT degradation by ϳ60% (Fig. 4). Sucrose, CPZ, and M␤CD also inhibited CT degradation in a dose-dependent manner (not shown), with maximal inhibition of degradation of ϳ50 -70% (Fig. 4); M␤CD inhibited CT degradation to a lower extent in older neurons. The inhibition of CT degradation by M␤CD can be explained by a reduction in the rate of clathrin-mediated endocytosis (33,34). In contrast, filipin did not inhibit degradation to any significant extent (Fig. 4). These data on CT transport to the Golgi apparatus, and its degradation, support the idea that CT internalization does not occur via detergentinsoluble rafts/domains in hippocampal neurons.
The Rate of CT and Tf Internalization Are Modified by Sucrose and CPZ but Not by M␤CD or Filipin-We next compared the effects of the drugs on CT and Tf internalization by kinetic analysis. The rate of CT internalization was calculated using Equation 1, for 8-day-old neurons (Fig. 5B). Due to the relatively slow binding of CT to the plasma membrane, we could not directly measure an internalization rate constant for CT; however, the initial rate of internalization is an approximation of the internalization rate constant, which was faster in 2-day-old (0.0238/min; Fig. 5A) than in 9-day-old neurons (0.0051/min; Fig. 5B). We had previously shown that the rate of CT transport to the Golgi apparatus (17), and its rate of degradation (see above), are slower in older neurons, but the current kinetic analysis directly demonstrates that the rate of CT internalization is a rate-limiting step in regulating CT transport. In both 2-and 8 -9-day-old neurons, CT endocytosis reached steady state after 60 min.
The kinetics of internalization of Tf differed from that of CT, as Tf binding to the cell surface is very fast (26). Thus, the amount of cell-surface Tf is constant at all time points, and the internalization rate of Tf is determined as a ratio of internal Tf to cell-surface Tf (0.0129/min in 9-day-old neurons) (Fig. 5C). No analysis of Tf binding or internalization could be performed in 2-day-old neurons since they do not express high enough levels of the Tf-R (36,37).
Even though sucrose and CPZ inhibited CT transport to the Golgi apparatus (Figs. 2 and 3) and its degradation (Fig. 4), they had opposite effects on the rate of CT internalization. Sucrose completely inhibited CT internalization in 2- (Fig. 6A) and 8 -9-day-old (Fig. 6B) neurons, whereas CPZ increased the rate of CT internalization (Fig. 6, A and B). The effect of sucrose on internalization is consistent with it affecting clathrin availability for coated pit assembly (29), and the effect of CPZ is consistent with it affecting a more downstream step in the endosomal pathway (19), possibly receptor recycling. In contrast, neither M␤CD nor filipin had any effect on CT internalization. Since we were unable to determine directly a kinetic rate constant for M␤CD-and filipin-treated neurons, as many M␤CD-and filipin-treated cells detached from the coverslips, we analyzed their effects at steady state (i.e. after 60 min of incubation). Neither drug affected internalization in 2- (Fig. 6, C and D) or 8 -9-day-old (Fig. 6, E and F) neurons at the concentrations used.
To compare directly CT and Tf internalization, we examined the effects of the drugs on steady state levels of CT internalization and on the rate of Tf internalization. Sucrose inhibited CT and Tf internalization, whereas CPZ increased the amount of internalized CT and increased the rate of Tf internalization (Fig. 7); the effect of CPZ is probably due to inhibition of receptor recycling. Neither M␤CD nor filipin had any effect on the amount of internalized CT or on the rate of Tf internalization (Fig. 7). Based on these data, and that presented above, we suggest that CT is internalized by a similar mechanism to that of Tf-R internalization, namely via clathrin-dependent endocy-tosis, although we cannot exclude other clathrin-or raft-independent mechanisms.
The Detergent Insolubility of CT Is Altered after Internalization and after CPZ Treatment-In the data presented above, we have shown that although cell-surface CT is found in detergent-insoluble domains, disruption of these domains by cholesterol-binding agents does not inhibit CT endocytosis. To determine whether internalized CT is also detergent-insoluble, neurons were solubilized with Triton X-100 before or after acid washing (which removes cell-surface CT). In contrast to cellsurface CT, which was essentially completely detergent-insoluble, ϳ40% of internalized CT was detergent-soluble (Fig. 8).
Remarkably, after CPZ treatment, all internal CT became detergent-soluble (Fig. 8). This suggests that internalized CT may be partially excluded from detergent-insoluble domains and is completely excluded from detergent-insoluble domains after its accumulation in the aberrant clathrin-coated endosomes that accumulate after CPZ treatment.

DISCUSSION
In neurons, CT could potentially be internalized via either a clathrin-dependent mechanism, together with receptor-bound ligands such as Tf, or via a mechanism that depends on the integrity of detergent-insoluble domains/rafts. Our previous studies (18) could not distinguish between these two possibilities but did demonstrate that CT passes through the same endosomal compartment as receptor-bound ligands internalized via clathrin-mediated endocytosis (18). In the current study, kinetic analysis of the initial rates of internalization and direct comparison with Tf-R internalization strongly suggest that CT is internalized by a similar mechanism to that of Tf-R. Since Tf-R is internalized exclusively via clathrin-mediated endocytosis (21), our data suggest that CT is also internalized via a clathrin-mediated mechanism, although we cannot exclude the possibility that additional nonclathrin-mediated pathways are involved.
There are at least two models that could explain the apparent paradox between the association of CT with detergentinsoluble domains at the cell surface and its internalization via a clathrin-dependent mechanism in neurons (Fig. 9). In the first model, CT escapes the domains prior to internalization ( Fig. 9). In the second model, the whole domain, including CT, is internalized (Fig. 9). In both scenarios, hypertonic medium (i.e. high sucrose concentrations) would inhibit internalization, and CPZ (and other cationic amphiphilic drugs) would block subsequent transport to the Golgi apparatus.
The first model implies that CT association with membrane domains at the neuronal cell surface is transient. Indeed, even in cells that contain morphologically distinct caveolae, CT is enriched ϳ4-fold in caveolae but is also associated with areas of the plasma membrane that lack caveolae (2). This model also implies that the rate of diffusion out of the domains, and the rate of diffusion into clathrin-coated pits, might be a means to regulate the rate of endocytosis. This is consistent with observations in primary neurons that the prion protein, a GPIanchored protein associated with detergent-insoluble domains, is more detergent-soluble than another GPI-anchored protein, Thy-1, and is also internalized much more rapidly than Thy-1 (24). In hippocampal neurons, CT is internalized more rapidly in 2-versus 9-day-old neurons, which also correlates with increased detergent insolubility in older neurons; the reason for increased detergent insolubility is not known but may be related to changes in glycosphingolipid, sphingomyelin, or cholesterol composition as neurons mature (16,23,38). However, other membrane components, which are not necessarily associated with detergent-insoluble domains, are also internalized more slowly in older neurons (25,36,39). Interestingly, in hippocampal neurons, only 13% of cell-associated CT is internalized at steady state in 8-day-old neurons, but ϳ70% of CT is internalized after 30 min of incubation in CaCo-2 cells (13). These different rates of internalization might be due to tighter association of CT with domains in neurons than in CaCo-2 cells. In addition, internalization from caveolae and also clathrin-coated pits in CaCo-2 cells may increase the rate of CT internalization compared with neurons in which internaliza-tion does not occur via caveolae.
In the second model ( Fig. 9), the raft itself diffuses into the coated pit and is internalized together with raft components, including CT, cholesterol, and GSLs. There is ongoing debate about the size of membrane rafts, but recent measurements (40,41) suggest that they might be small enough to be internalized via clathrin-coated pits. As in the first model, CT association with domains is transient, since CT is less detergentinsoluble once internalized, but in contrast to the first model, CT exits the domains in early endosomes rather than at the plasma membrane. Rafts exist in recycling endosomes (42,43), and the sorting of membrane components out of the rafts depends on their preference for association with the domains (44) and can be regulated by cholesterol and GSL levels (43). The reason that CPZ alters the detergent solubility of CT may be a result of the fact that other endocytic compartments, such as late endosomes, do not contain rafts. We suggest that CPZ renders CT detergent-soluble due to recruitment of clathrin to an endosomal population (involved in transport to the Golgi apparatus) that does not contain rafts (see Fig. 9).
CT is not the only example of a molecule that is detergentinsoluble at the cell surface but is internalized via a clathrindependent mechanism. For instance, the epidermal growth factor receptor is detergent-insoluble at the plasma membrane but is internalized via a clathrin-dependent mechanism (reviewed in Ref. 3). Likewise, the GSL-binding toxin, Shiga toxin, is also found in detergent-insoluble domains (45) but is internalized via clathrin-coated pits (46), whereas verotoxin is internalized via both clathrin-dependent and -independent pathways (47). Similar to our findings, Shiga toxin internalization cannot be inhibited by cholesterol-binding agents. Shiga toxin and verotoxin both bind to the neutral GSL, Gb3 globoside, whereas CT binds to the acidic GSL, GM1. These two GSLs may be differentially distributed over the cell surface in different cell types, since GM1 is enriched in caveolae in cells that contain these structures (2), whereas there is no evidence that Shiga toxin is found in caveolae.
Do membrane domains play a role in regulating neuronal development, similar to their proposed roles in the development of polarity in epithelial cells? There is some evidence to support this possibility, based on analysis of the effect of depleting cholesterol and GSLs on axonal sorting (22), and on the effect of manipulating lipid levels on detergent insolubility of GPI-anchored proteins (23). However, it should be considered note that the acid wash had no effect on 125 I-CT solubility. The distribution of internalized 125 I-CT in CPZ-treated neurons is shown as closed circles; CPZ treatment had no effect on 125 I-CT solubility at the cell surface. The total number of counts/min loaded onto the gradient was 107,000 for cell-surface 125 I-CT, 13,000 for internal 125 I-CT, and 19,000 for internal 125 I-CT in CPZ-treated neurons.

FIG. 9. Mechanisms of CT internalization in neurons.
At the cell surface, CT is located in detergent-insoluble domains (rafts). In model I, CT escapes the domains prior to internalization via clathrin-coated pits. In model II, the microdomain is internalized via clathrin-coated pits together with CT and is sorted away from or excluded from other domain components only after internalization. Irrespective of which model is correct, CT is subsequently transported to the Golgi apparatus. CPZ blocks CT transport to the Golgi apparatus due to recruitment of coat proteins to endosomes. that rafts play no specific role in the targeting of GPI-anchored proteins to axons (48) but rather act to cluster proteins together into functionally important complexes (i.e. signaling complexes). In support of this are data showing that some raft components (i.e. GPI-anchored proteins (24) and CT (49)) are distributed uniformly over both the axonal and dendritic surfaces in polarized neurons, whereas they are nonuniformly distributed (i.e. polarized) in some epithelial cells (reviewed in Ref. 50); the lack of polarized distribution of raft components would not negate the possibility that they function to cluster proteins at the cell surface. An extension of the "clustering" hypothesis is the idea that a function of rafts is to regulate the rate of endocytosis of membrane components. This would not necessitate the polarized distribution of raft components but would rather depend solely on raft composition, which would in turn impinge upon the rate of diffusion out of the raft and hence the rate of endocytosis. If this hypothesis is correct, then the rate of internalization to early endosomes would be regulated by the distribution of CT between cell-surface rafts and areas of the plasma membrane that are not enriched in raft components (model 1 in Fig. 9), and it would be difficult to envisage how the association of CT with endosomal rafts (model 2) could regulate its rate of internalization.