This report presents the characterization of a fluorescently tagged peptide probe, the SBD, for tracing the trafficking pathways of sphingolipid-containing membrane microdomains in living cells. We present evidence that SBD binds to artificial membranes with raft-like composition, with improved binding efficiency at low pH and in the presence of gangliosides. Quantitative time-course colocalizations based on the method of Costes 
are used to describe the endocytic trafficking pathway of fluorescently tagged SBD, and compare its intracellular trajectory with those of other presumptive microdomain-specific and endolysosomal markers.
Consistent with the model of Fantini et al, wherein aromatic and basic residue(s) in the SBD were proposed to interact with galactose-terminal glycolipids and sphingomyelin, we found that SBD-TMR interacted optimally with the combination of POPC/SM/Chol, and the strongest enhancement of binding was seen when ganglioside was added to this mixture. The ability to bind POPC/SM/Chol may be partially attributable to the tendency of this mixture to phase separate into liquid ordered domains (confirmed by atomic force microscopy [GT & RK, unpublished results]; 
), since a saturated glycerophospholipid with the same choline headgroup (SPPC), which would also be expected to phase separate, was able to substitute for sphingomyelin to some extent. However, SBD binding to the SPPC mixture was not as high as that which was seen with sphingomyelin. Moreover, addition of glycosphingolipid to the SPPC mixture abolished binding, whereas it strengthened the binding of the sphingomyelin mixture. Therefore, although an interaction with other PC-containing liquid ordered domains is possible, it appears likely that under conditions at the plasma membrane of a cell, where glycolipids are present, a PC-headgroup sphingolipid (i.e. sphingomyelin) is strongly preferred.
SBD showed a preference for multiply sialylated gangliosides at neutral pH, similarly to amyloid peptide Aβ1−40 
, but the specificity for a particular ganglioside is less strict than is the case with CtxB. Interestingly, the strength of SBD's interaction with ganglioside-containing membranes is greatly enhanced at low pH. Because of this, one can speculate that as SBD is endocytosed and the pH drops, its interaction with the membrane may be strengthened by masking of electrostatic repulsion between SBD (which has a pKa of ~4.25) and sialic acid. Given the overall negative charge of SBD, however, its predilection for GT1b, GQ1b, and GD1a (which have multiple sialic acids) over GM1 (which has one) at pH 7 is difficult to explain. Possibly, the positively charged Arg and His residues near the aromatic Tyr residue proposed by Fantini to be important in binding via π-bonding interactions 
, actually attract SBD to more heavily sialylated gangliosides. This would be supported by the work of Ariga et al 
, and by our experiments at neutral pH. On the other hand, SBD's preference for less sialylated glycolipids (e.g. GM1) in acidic endosomes suggests the possibility that Fantini's proposed π-bonding with the exposed galactose, or a specific structural interaction with terminal galNAC-gal-sialic acid, may begin to dominate when the charges are neutralized.
Confirming the spectrofluorimetric liposome binding assays above, SPR analysis showed improved binding of SBD-TMR to the same liposomes immobilized on a Dextran-coated gold chip, as the concentration of GD1a was increased. However, we note that binding of SBD reacted relatively weakly and gradually to low concentrations of ganglioside, and required up to 20% to give the maximal response, as compared to CtxB, which bound very strongly already at just 5% GM1. These results together with the lack of intracellular colocalization with CtxB (below) suggest that recognition of different glycolipids (in this case GD1a vs. GM1), perhaps in combination with membrane proteins, may mediate uptake into distinct intracellular trafficking routes, as suggested by Lencer and colleagues 
As would be expected for a raft-binding probe 
, SBD showed no detectable colocalization with Transferrin, which is internalized by clathrin-mediated endocytosis 
. Unexpectedly, however, SBD showed only low to moderate overlap upon uptake and trafficking in neuroblastomas with the sphingolipid-interacting marker CtxB. In some mammalian cell types, CtxB is thought to follow a cdc42-dependent uptake pathway 
but it can also be endocytosed by a mixture of non-clathrin and clathrin-mediated mechanisms in neurons 
. SBD showed higher colocalization with another raft-localized marker, Flotillin2-GFP which, like SBD, traffics to a large extent through the endolysosomal pathway 
The non-colocalization with CtxB was all the more surprising, because of reports that Aβ (from which SBD is derived) binds to the same glycosphingolipid as CtxB, namely GM1 
. Aβ, however, can also bind to other glycosphingolipids, dependent on conformation 
. From these results we conclude that SBD is internalized by a non-clathrin pathway that is distinct from other cholesterol-dependent pathways, reflecting the probable heterogeneity of lipid microdomains as has been noted in many other studies 
After uptake, SBD overlapped extensively with the endolysosomal tracer Dextran, which at low concentrations is endocytosed via a cdc42-dependent uptake pathway 
. Early, SBD trafficked through a rab5→FYVE→rab7 trajectory, finally reaching lysosomes after 3–4h (summarized in ). A significant amount (tM ~60%) of SBD also trafficked from late endosomes (1.5–2 h) to a rab11-GFP-positive recycling compartment, which has been reported to be particularly rich in sphingomyelin and cholesterol in mammalian cells 
. These findings are interesting with regard to the discovery that mutations in Alzheimer's disease-linked SORL1 result in shunting of App away from recycling endosomes toward the endolysosomal pathway 
. More experiments will be required to determine whether and how App trafficking and SBD trafficking are related.
Of all the markers tested, the highest degree of colocalization was found between SBD and the glycosphingolipid analog BODIPY-lac-Cer. Interestingly, this was only true when the two labels were present at the membrane simultaneously; sequential addition resulted in nearly no colocalization. The finding that SBD and lac-Cer trafficked so closely together was surprising, since lac-Cer's trafficking itinerary is different from that of SBD in both mammalian 
and Drosophila cells (RH, EL, SH, SS & RK, submitted). Their co-trafficking may result from SBD's affinity for glycosphingolipids, or may reflect a phenomenon we have documented in a separate study, termed “hijacking”, whereby one label influences the trafficking behavior of the other. The hijacking mechanism is apparently different from the simple stimulation of uptake reported by Pagano and colleagues 
, since we find that even over extended chasing times, the second label added sequentially never reaches the same compartment as the first, which would be expected if only endocytic stimulation were responsible (RH, EL, SH, SS & RK, submitted).
The interaction of SBD with glycosphingolipids is consistent with our finding that its trafficking to lysosomes is strongly affected by cholesterol levels, since glycolipids are also known to traffic aberrantly in cholesterol storage disease fibroblasts 
. Paradoxically, however, we find that SBD trafficking to the lysosome is ultimately diminished by excess cholesterol, in contrast to glycosphingolipids, which accumulate to excess in endolysosomes under similar conditions. In summary, it appears that SBD trafficking is strongly influenced by cellular cholesterol levels, especially cholesterol overload, and this may reflect an underlying event involving the trafficking of its glycosphingolipid partner(s). Moreover, changes in SBD targeting may provide insights into the effects of cholesterol perturbation on the trafficking behavior of Aβ, since their binding characteristics appear to be similar.
In this study, we describe the intracellular trafficking of a novel fluorescent sphingolipid-interacting marker, the SBD, and present evidence that it can interact with raft lipids sphingomyelin, cholesterol, and glycosphingolipids, but without a highly specific interaction to a particular ganglioside, like CtxB. Thus, SBD can potentially be used to trace the intracellular pathways of sphingolipid-containing domains, including at least some gangliosides. There is little doubt about the utility of being able to follow the trafficking routes of sphingolipids and microdomains in living cells, yet very few non-transgenic probes are currently available that track these lipids.
The SBD probe is a short, non-toxic peptide, and can therefore be produced easily as a fluorophore-linked conjugate and applied exogenously to cells. Together with the observation that SBD's trafficking route reacts strongly to cholesterol perturbation, these properties make it well suited for applications in diagnostic and drug screening assays. Additionally, the fact that SBD behaves similarly in Drosophila and mammalian neurons validates its use as a tool for studying lipid trafficking disease models in the fly. Further characterization of the behavior of the SBD in the context of genetic and lipid perturbations will allow the use of this probe in models of neurodegenerative and other diseases that are linked with the metabolism and trafficking of sphingolipids. Information gained from such studies would contribute to our understanding of the involvement of sphingolipid trafficking in neurodegenerative disease pathologies.