In the present study, we demonstrate that endogenous APP is enriched in neuronal lysosomes as well as lysosomes of Cos7 and SN56 cells that were transfected to overexpress fluorescently tagged APP constructs. In addition to its expected transit from the cell surface to the early endosomal compartment, surface-labeled wild-type APP transits to the lysosomal compartment within minutes and appears to redistribute from the cell surface to LAMP1 vesicles that are found in close proximity to the plasma membrane. Wild-type APP is transported to the lysosome more rapidly than to either early or late endosomes. These observations suggest that APP is trafficked to the lysosome via a mechanism that is distinct from that utilized by the non-specific uptake of dextran. Unexpectedly, two APP mutations that cause increased Aβ peptide accumulation and early onset AD exhibit profound differences in their intracellular trafficking to lysosomes when compared to wild-type APP. Taken together, these experiments suggest distinct trafficking pathways for APP transit from the cell surface to the lysosome.
Although APP has been studied extensively, this is the first demonstration of rapid protein transport directly from the plasma membrane to the lysosome. That APP is highly enriched in lysosomes has been previously observed [28, 36–38]. APP has also been demonstrated to move to lysosomes within 15 minutes  to 2 hours  but these studies did not quantitate either internalization or colocalization. The easy detection of APP in lysosomes suggests that it has a functional role in this compartment, as the half life of lysosomal substrate proteins is 8 minutes and lysosomal substrate proteins are expected to be detectable only at very low levels . One possible parallel might be found in the prion protein (PrP) which normally traffics through early endosomes, but when pathologically misfolded appears to transit to the lysosome without being observed in early endosomal compartments [40, 41].
Our data demonstrating enrichment of APP in lysosomes has several limitations. The quantitation method used here is based on colocalization of the brightest pixels. It therefore is likely only useful for comparing colocalization amongst discreet compartments and does not represent an attempt to account for all of the APP in a cell. For example large amounts of APP in other compartments (such as the ER or the plasma membrane) would not be accounted for if they had a lower levels of fluorescence intensity. In addition, compartment marker proteins are often present in other compartments; for example, colocalization with LAMP1 might also include autophagosomes, which have also been identified as a likely site of APP processing [42, 43]. This study also specifically ignores possible cleavage events in biosynthetic compartments as β- or γ-cleavage of the construct before transport to the cell surface would remove the HA tag and render these proteins invisible to our cell surface labeling technique.
The appearance of APP in the lysosome before it appears in the early endosomes suggests that wt-βAPP can transit directly to the lysosome from the cell surface, bypassing the early and late endosomal compartments. In contrast both London and Swedish βAPP appear to be excluded from this pathway and transit to the lysosome with kinetics suggesting that they transit first through early endosomes and then through late endosomes. This might account for the fact that we observe less steady-state localization of both London and Swedish APP in lysosomes. Furthermore, the decreased abundance of APP-London in the late endosome, coupled with its accumulation in early endosomes, is consistent with the relative retention of APP-London in the early endosome. Interestingly, Swedish APP is delivered to the late endosome at a faster rate than either wild-type or London mutant APP and yet at steady state there is no apparent difference in the amount of London or wild-type APP in the Rab9 positive late endosome. This suggests that effective proteolysis of APP-Swedish, which is expected to be cleaved by BACE with much higher efficiency than APP-wt , may occur as the protein transits from the late endosome to the lysosome. Although this work does not rule out the possibility that some APP might be cleaved by secretases in biosynthetic compartments, it does demonstrate that uncleaved APP (including APP-Swedish) is clearly present on the cell surface. This data also suggests the presence of a novel and highly specific sorting system for APP at the cell surface and a novel pathway from the cell surface directly to the lysosome.
A model to account for the phenomena observed here is that APP might transit to the lysosome by 2 distinct pathways, the classical expected pathway through the early and then late endosomes which has been previously described , and another directly from the cell surface to the lysosome either through an intermediate transport vesicle, or by direct fusion and recovery of lysosomes with the plasma membrane. In this model, APP sorting would occur at the cell surface, and APP bearing Swedish and London mutations would be preferentially sorted through the endosomal pathway. There is now good evidence that different cell surface receptors can be directed into distinct clathrin-coated pits based upon binding domains in their intracellular C-terminus [44, 45]. This cell surface sorting system has been demonstrated to direct proteins to at least 2 distinct types of endosomes, with some cargo transported to slowly moving, slowly maturing rab5-labeled endosomes, and some cargo sorted to a rapidly maturing Rab7 labeled compartment . Because Rab7 can label both endosomes and lysosomes, this study might also be documenting direct lysosomal transport.
There is no simple explanation for how mutations in the extracellular or transmembrane domain of APP might alter its trafficking. Although proteins involved in APP trafficking typically bind the cytoplasmic C-terminus of APP (reviewed in ), there are poorly defined signals in the luminal domain of APP [23, 29, 48] and deletions of the extracellular juxtamembrane segment of APP have been shown to disrupt APP sorting , suggesting that a trafficking mechanism exists which recognizes changes in the extracellular domain of APP. One possible trafficking regulator might be the γ-secretase complex itself (either through presenilin or nicastrin), which can interact with APP in the transmembrane and extracellular domain  and only cleaves APP after a β-cleavage has occurred [50, 51]. This idea is supported by fact that mutations in PS1 (D385N) that block γ-secretase activity or pharmacological inhibition of γ-secretase activity also block APP internalization .
The pathways by which APP transits the endosomal/lysosomal system may have implications for Aβ production. For example, it is possible that the type of proteolysis that APP undergoes may be related to the mechanism by which the protein is being mobilized to the lysosome. Thus, endocytosis through the early and late endosome may lead to more efficient APP cleavage by secretases to produce Aβ while rapid transport to lysosomes might lead to degradation by lysosomal proteases. Because the aggregation of Aβ is also favored by low pH as well as by the intrinsic composition of the lysosomal membrane [53, 54], the route of delivery to the lysosome might also affect amyloid formation.