Ira Mellman 
Ludwig Institute for Cancer Research, New York, NY
Yale University School of Medicine, New Haven, CT

Endocytosis and Antibody-Based Cancer Therapy

All cells exhibit one or more forms of endocytosis, in either a fashion which can be either constitutive or regulated. Each form of endocytosis, ranging from receptor-mediated uptake via small clathrin-coated vesicles to macropinocytosis to phagocytosis, is mechanistically distinct. Nevertheless, they are all related by a common result: the internalization of macromolecules bound to the plasma membrane or soluble in the extracellular fluid. There is thus an unavoidably close relationship between any therapeutic approach based on the use of anti-cancer antibodies and the fate of those antibodies after they bind to their target cells. It is highly likely that the endocytosis and possibly the intracellular compartmentalization of therapeutic antibodies influences efficacy. As a result, it is important to consider the various scenarios that can follow antibody binding to cancer cells. Recent work has revealed an unexpected complexity which is manifested not only at the internalization step, but also in endocytic organelles, particularly in polarized cells such as epithelia and neurons. The purpose of this presentation is to provide a brief summary of our current understanding of endocytosis and the endocytic system.

Classically, endocytosis is most often associated with the internalization of plasma membrane receptors via clathrin-coated pits. This is often presumed as the primary mode by which “internalizing” antibodies are taken up. The formation of clathrin-coated pits involves the sequential binding of the heterotetrameric adapter complex AP-2 to the membrane followed by the assembly of hexameric clathrin subunits to form the characteristic lattice of coated pits. Plasma membrane proteins may selectively or non-selectively internalize, with selectivity being mediated by the interaction of tyrosine-based cytoplasmic domain signals that interact with the AP-2 µ2 subunit. Although still true, this relatively simple scenario is far from the complete story. Recent work has now strongly suggested that multiple adapter complexes exist that work either in concert with, or separately from, AP-2. Moreover, these alternative potential adapters – including
b-arrestin, AP180/CALM, Disabled-2, Huntingtin-interacting protein-1 (HIP1), GGA’s — interact with internalization signals that are distinct from the motifs that are decoded by µ2 (e.g., Y-X-X-f). These other signals, even more degenerate and heterodisperse than the µ2 signal, make it very difficult to know from sequence inspection whether a particular surface antigen is efficiently internalized. Even “obvious” signals may not be functional since they may be sterically masked or nullified by cytoskeletal interactions that inhibit movement to coated pits.

Surface receptor endocytosis can also be induced by ligand binding. Particularly in the case of receptor tyrosine kinases, induced autophosphorylation, or phosphorylation of downstream effectors such as Eps15, epsin, amphiphysin, or even clathrin can selectively enhance recruitment of a surface antigen to coated pits.

Endocytosis can also occur in the absence of clathrin. Small uncoated vesicles are now known to form and, indeed, can substitute entirely for clathrin-mediated uptake when the clathrin pathway is inhibited for example by the expression of dominant negative alleles of dynamin, a GTPase thought to act in the scission of clathrin-coated buds from the plasma membrane. Non-coated uptake may occur via caveolae or perhaps by an actin-dependent mechanism reminiscent of the normal clathrin-independent pathway typical of S.cerevisiae.

The two other major forms of endocytosis, macropinocytosis and phagocytosis, are generally, but not always, induced by ligand binding. Antibodies that bind to the appropriate signalling receptors, for example, can induce generalized formation of membrane folds which can form macropinosomes. Induced macropinocytosis reflect the activation of Rho family GTPases such as Cdc42 and assembly of branched actin filaments via N-WASP and the Arp2/3 complex.

Regardless of the mode of uptake, internalized receptors or antibodies are subject to multiple potential fates after delivery to endosomes, the collective term for all non-lysosomal organelles. The most common fate is recycling back to the plasma membrane, a pathway emanating from “early endosomes” via either of two routes. First, there is the rapid (1-2 min) route involving direct transfer; proteins involved in this pathway do not even accumulate an appreciable intracellular pool and bound antibodies may not even seem to be internalized. There is also a slower indirect route involving perinuclear “recycling endosomes,” the slower kinetics resulting in an accumulation of an internal pool of receptor (or bound antibody). The factors determining if one or the other paths are used remain poorly understood. In polarized epithelial cells or neurons, recycling is further complicated by the fact that internalized proteins have the possibility of returning to their surface of origin or to a second plasma membrane domain. Here, our recent work has revealed a second array of cytoplasmic domain signals that determine the surface to which recycling is directed. In epithelia, a specific AP-2-like adapter complex exists (AP-1B) to mediate specific targeting to the basolateral plasma membrane region.

Rapid recycling, either polarized or non-polarized, may subvert the efficacy of a given antibody if that antibody requires intracellular accumulation for action. Unless the antibody were to dissociate in acidic endosomes, it would be most likely lost to the extracellular fluid upon return to the plasma membrane.

Not all endocytosis is followed by recycling, however. In some cases, internalized plasma membrane proteins are diverted in early endosomes from the recycling pathway to the lysosomal pathway. This process, classically referred to as down regulation, may be critical to a number of antibody-based therapies due to the increased intracellular accumulation of therapeutic antibody. When antibodies are directed against signalling receptors, antibody-induced down regulation may also result in an overall desensitization to growth factor or other hormones. While it has long been appreciated that antibodies can induce down regulation by receptor cross-linking, only in the past 2-3 years has the biochemical basis for this effect begun to emerge. Receptors “targeted” for lysosomal delivery are selected at the level of early endosomes, usually by virtue of a specific ubiquitination event. In some cases, ubiquitination itself can be mediated by a receptor-specific ubiquitin ligase, such as in the case of the proto-oncogene c-Cbl and EGF receptor. This modification, together with a signal transduction pathway mediated by PI3 kinase, leads to the formation of internal inclusions within endosomes onto which modified receptors are selectively sequestered. Appearance in the internal membranes of these late endosomes or “multivesicular bodies” invariably leads to receptor delivery to lysosomes and ultimately to receptor degradation.

Finally, a newly appreciated pathway would appear to permit the escape of macromolecules from endosomes to the cytosol. This event, now well characterized in the case of dendritic cells during the cross presentation of antigen, would thus permit therapeutic antibodies, or bound substituents, access to intracellular components not normally accessible to extracellular macromolecules. Thus far, little is known mechanistically regarding how endosomal egress works, although it seems to be favored as a consequence of macropinocytosis. Conceivably, antibodies could be selected that favor this pathway.