1 Clathrin-independent pathways do not contribute significantly to endocytic 2 flux 3 4 Vassilis Bitsikas1, Ivan R. Corrêa Jr.2, Benjamin J. Nichols1* 5 1. Medical Research Council Laboratory of Molecular Biology, CB2 0QH, Cambridge, 6 UK. 7 2. New England Biolabs, Inc., 240 County Road, Ipswich, MA 01938, USA. 8 *[email protected] 9 10 Abstract 11 Several different endocytic pathways have been proposed to function in mammalian 12 cells. Clathrin-coated pits are well defined, but the identity, mechanism and function of 13 alternative pathways have been controversial. Here we apply universal chemical labelling 14 of plasma membrane proteins to define all primary endocytic vesicles, and labelling of 15 specific proteins with a reducible SNAP-tag substrate. These approaches provide high 16 temporal resolution and stringent discrimination between surface-connected and 17 intracellular membranes. We find that at least 95% of the earliest detectable endocytic 18 vesicles arise from clathrin-coated pits. GPI-anchored proteins, candidate cargoes for 19 alternate pathways, are also found to enter the cell predominantly via coated pits. 20 Experiments employing a mutated clathrin adaptor reveal distinct mechanisms for sorting 21 into coated pits, and thereby explain differential effects on the uptake of transferrin and 22 GPI-anchored proteins. These data call for a revision of models for the activity and 23 diversity of endocytic pathways in mammalian cells. 24 25 26 27 28 29 30 31 32 33 34 35 1 36 Introduction 37 Endocytosis has central roles in many cell biological processes [1,2]. Since the mid- 38 nineties evidence has accumulated to suggest that mammalian cells utilise additional 39 endocytic mechanisms beyond clathrin-coated pits [3-5]. Whilst the molecular detail of 40 how clathrin-coated pits work is understood in ever-increasing detail [6], similarly complete 41 mechanistic descriptions of how endocytosis may take place outside of clathrin-coated pits 42 are lacking. 43 One central difficulty in defining clathrin-independent endocytic pathways has been the 44 paucity of rigorously validated endocytic markers and pathway-specific cargoes. 45 Dissection of clathrin-mediated endocytosis benefited greatly from signature cargoes such 46 as the transferrin receptor, which are efficiently concentrated in the forming endocytic 47 vesicle, and from the fact that the forming vesicle is marked specifically in space and time 48 by transient assemblies of clathrin, adaptors and associated proteins [7-10]. By contrast, 49 clathrin independent endocytosis has largely been defined by morphological criteria, and 50 by the persistent uptake of cargoes that may utilise multiple pathways following 51 perturbation of the clathrin machinery [11-13]. 52 Endocytic structures that can be defined morphologically include macropinosomes, which 53 can readily be resolved by light microscopy [14], and caveolae which are distinctive in 54 electron micrographs [15]. The extent to which caveolae are involved in endocytosis is, 55 however, controversial [16]. Morphology also forms a large part of the definition of more 56 recently characterised endocytic membranes termed CLICs, for clathrin-independent 57 carriers [17,18]. 58 GPI-anchored proteins have been extensively studied as potential cargoes for clathrin- 59 independent endocytosis [4,19]. The apparent presence of these proteins in endosomes 60 devoid of transferrin, and uptake in the presence of inhibitors of clathrin-coated pits, 61 provides evidence for GPI-enriched endosomal compartments (GEECs) that are fed from 62 the cell surface independently from coated pits [20-22]. It is not clear, however, that GPI- 63 anchored proteins are ever highly concentrated in nascent endocytic vesicles in a manner 64 analogous to transferrin receptor, so they may enter the cell via multiple mechanisms 65 [19,22-24]. The extent to which other types of cargo associated with clathrin-independent 66 endocytosis, including glycosphingolipid-binding bacterial toxins as well as various viruses, 67 are efficiently sorted during uptake is similarly unclear [11,12,19,25-27]. 2 68 In the absence of demonstrably specific cargoes, much of the literature on clathrin- 69 independent endocytosis relies on the use of overexpressed mutant proteins to perturb 70 clathrin function, and observation of differential effects on the uptake of transferrin and 71 potential clathrin-independent cargoes. Dominant negative mutants used in this manner 72 include the C-terminal clathrin-binding domain of AP180/CALM [28], and the K44A mutant 73 of dynamin, which renders this GTPase involved in the scission of clathrin coated pits 74 enzymatically inactive [29,30]. Additionally, differential effects on endocytosis can be 75 observed using overexpression of inactive forms of small GTPases such as ARF6, ARF1 76 or cdc42 [20,21,31-33]). Blocking one type of endocytosis may up-regulate alternative 77 mechanisms, over-expression of mutant proteins may induce non-physiological cellular 78 responses, and small GTPases may, via different sets of effectors, directly or indirectly 79 control the activity of multiple endocytic pathways. Plainly, these types of experiment need 80 to be interpreted carefully. 81 Ideally, different types of endocytosis would be defined by the presence of specific 82 molecular determinants analogous to clathrin or adaptor proteins in the case of clathrin 83 coated pits [6]. Candidates for such determinants include caveolin and cavin proteins 84 which make caveolae [15,34], flotillin proteins which define specific plasma membrane 85 microdomains potentially involved in endocytosis [35], and the protein GRAF1 that may be 86 important for the formation of CLIC / GEEC endosomes [36]. The precise mechanisms by 87 which these proteins are involved in endocytosis remain to be fully understood. Both 88 flotillins and caveolins plus cavins form protein assemblies that are stable over time, and 89 thus can not define endocytic events temporally in the way in which coordinated assembly 90 and disassembly of the clathrin machinery can [34,35,37-39]. 91 Thus, although several non-clathrin endocytic pathways have garnered varying degrees of 92 supportive evidence, none has been unambiguously established in molecular and 93 functional terms. Furthermore, the relative contributions of the multiple putative pathways 94 to overall endocytic flux has been unclear. To resolve some of these uncertainties, ideally 95 one would require a means to examine endocytosis in unperturbed cells in a global 96 manner. This would allow simultaneous evaluation of a large number of cargoes and their 97 relationship with clathrin and putative non-clathrin markers. In this study we apply a 98 combination of new and established methods that satisfy these requirements, and thereby 99 provide a systematic and quantitative analysis of total endocytic protein flux in cultured 100 mammalian cells. 3 101 102 4 103 Results 104 Experimental strategy and assay validation 105 We sought to establish endocytic assays and protein labelling strategies to satisfy four 106 main requirements: 1. Achieve very high topological specificity in discrimination between 107 endocytosed and extracellular protein, 2. Provide a means to analyse all, or nearly all, 108 surface proteins simultaneously, 3. Provide high signal to noise, and thereby high temporal 109 resolution for detection of primary endocytic vesicles, and 4. Allow a means to follow 110 uptake of specific cargoes. 111 Biotinylation of extracellular free amine groups with the small, monovalent label sulfo-NHS- 112 SS-biotin offered a way to satisfy the first three of these requirements [40]. The biotin 113 moiety can be removed from labelled proteins by reduction of the disulfide bond with 114 membrane-impermeant sodium 2-mercaptoethanesulfonate (MESNa) at 4°C, so this 115 approach provides a powerful way to detect internalisation of surface proteins [41,42]. We 116 assessed the efficiency of MESNa treatment in removing biotin from sulfo-NHS-SS-biotin- 117 labelled proteins, and compared this with two widely used alternatives, washing at low pH 118 to remove antibodies bound to the outside of the cell and cleavage of extracellular GPI- 119 anchors with PI-PLC [21]. MESNa treatment could remove over 99.9% of biotin from cells 120 labelled at 4°C (Figure 1A). Washes at pH3 to remove bound antibody, or PI-PLC to 121 cleave GPI-anchors, were around 2 orders of magnitude less efficient, removing up to 50% 122 and 90% of antibody bound to the GPI-anchored protein CD59 respectively (Figure 1B) 123 [43]. Reducible biotin and MESNa therefore offer a highly accurate way of assaying 124 protein internalisation. 125 To detect internalised protein in confocal images, we labelled HeLa cells with sulfo-NHS- 126 SS-biotin, allowed internalisation for defined periods of time, removed extracellular biotin 127 by reduction with MESNa, and used fluorescent streptavidin to detect intracellular biotin. 128 All subsequent experiments are also in HeLa cells unless otherwise stated. Control 129 experiments comparing incubation at 4°C with incubation for defined periods at 37°C 130 confirmed that there was no specific signal detected unless internalisation at 37°C was 131 allowed to take place (Figure 1C, Figure 1D). Endocytic vesicles could be observed after 132 as little as 20 seconds of internalisation. This is a significant improvement in time 133 resolution, and hence in our confidence that the detected vesicles have just budded from 134 the