Individual Phosphoinositide 3-Kinase C2α Domain Activities Independently Regulate Clathrin Function*

Phosphoinositide 3-kinase C2α (PI3K-C2α) is a member of the class II PI-3 kinases, defined by the presence of a second C2 domain at their C termini. The cellular functions of the class II enzymes are incompletely understood, though they have been implicated in receptor activation pathways initiated by epidermal growth factor, insulin, and chemokines. PI3K-C2α was recently found to be localized to clathrin-coated membranes in the trans-Golgi network and at endocytic sites on the plasma membrane. Further, a specific binding site was identified for clathrin on the N terminus of PI3K-C2α, whose occupancy resulted in lipid kinase activation. Expression of PI3K-C2α in cells dramatically affected clathrin distribution and function in cells, leading to accumulation of intracellular clathrin-coated structures, which are visualized here at the ultrastructural level, and inhibition of clathrin-mediated transport from both the plasma membrane and the trans-Golgi network. In this study we have demonstrated that the isolated clathrin binding domain of PI3K-C2α can drive clathrin lattice assembly and that both it and the lipid kinase activity of the protein can independently modulate clathrin distribution and function when expressed in cells. Together, these results suggest that PI3K-C2α employs both protein-protein interaction and localized production of 3-phosphoinositides to affect clathrin dynamics at sites of membrane budding and targeting.

It is well established that both cargo and informational movement across the plasma membrane, as well as certain aspects of intercellular signaling, involve vesicular transport initiated by endocytosis from clathrin-coated pits. The machinery of endocytosis has been extensively studied, and its complexity would seem to reflect the challenge of providing a highly efficient yet controllable interface between the environment and the cell (1). In addition to the structural protein clathrin, numerous other receptor binding adaptors are present, as are proteins involved in membrane curvature, detachment, coat dissociation, and recycling; these include AP-2 and other adaptors, epsin, dynamin, synaptojanin, and others (reviewed in Refs. [2][3][4][5]. How the functions of these factors are controlled in space and time during the lifetime of the coated pit and the initial phase of the endocytic process remains a major challenge to our understanding. It has recently become apparent that membrane lipids, and specifically phosphoinositides, play major roles in the coordination of these processes. There is considerable evidence for an important regulatory role of phosphoinositide lipids in membrane traffic between multiple intracellular compartments. In the endocytic pathway, phosphoinositides were initially demonstrated to bind with high affinity to the plasma membrane adaptor AP-2 and to modulate its properties (6,7). Subsequent studies have confirmed these observations and demonstrated the interaction of phosphoinositides with other proteins involved in clathrinmediated trafficking (reviewed in Ref. 8). The actin cytoskeleton has also been strongly implicated in the function of the endocytic machinery (9 -11), and phosphoinositide regulators (12,13) tightly control its assembly and disassembly. Thus there is considerable in vitro evidence for roles of both PtdIns(4,5)P 2 and 3-phosphorylated inositides in membrane trafficking, and the identity of many of the targets to which they bind has now been established. To dissect the spatiotemporal aspects of these interactions in the modulation of endocytic and cytoskeletal function, it is critical to identify and study the enzymes responsible for inositide metabolism at specific sites of action.
We have recently shown that a specific phosphoinositide 3-kinase, phosphoinositide 3-kinase C2␣ (PI3K-C2␣) 3 is a component of the clathrin-mediated transport machinery at the plasma membrane and in the trans-Golgi network (TGN) (14). PI3K-C2␣ is a member of the class II PI 3-kinases and is distinguished from other PI 3-kinases by its additional C-terminal C2 domain (15,16). Three distinct mammalian members of the class II PI 3-kinase family have been characterized (17), PI3K-C2␣, PI3K-C2␤, and PI3K-C2␥. Although all share the distinguishing C-terminal C2 domain as well as the characteristic 3-kinase motifs, the enzymes differ predominantly in their N-terminal regions. The cellular functions of class II enzymes are not well understood, though there is evidence that PI3K-C2␤, and to a lesser extent PI3K-C2␣, can be stimulated by activation of several receptor systems including epidermal growth factor (18), insulin (19), chemokines (20), and integrins (21); the former is also implicated in cell adhesion and movement (22). By analogy with the Class I PI 3-kinase family, it has been suggested that activation of class II enzymes by signaling receptors may mediate specific aspects of downstream receptor activity. In vivo evidence for this possibility has recently been reported in studies of the single Class II PI 3-kinase in Drosophila melanogaster, PI3K-68D, which demonstrated its involvement in patterning processes involving the epidermal growth factor and Notch receptor pathways (23). Interestingly, the N-terminal region of PI3K-68D more closely resembles that of mammalian PI3K-C2␤, consistent with the receptor-mediated signaling role for this family member implied by the studies in mammalian cells noted above.
In addition to localizing PI3K-C2␣ to endocytic and TGN clathrincoated membranes (14), we found that PI3K-C2␣ is enzymatically activated by clathrin through binding to a discrete clathrin binding domain * This work was supported by National Institutes of Health Grant GM-49217. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  (CBD) present in the unique N-terminal region of the protein (24). Exogenous expression of PI3K-C2␣ in cells dramatically affected both clathrin distribution and function. Although plasma membrane clathrin-coated pits could still be observed, PI3K-C2␣ expression induced a striking accumulation of intracellular clathrin staining consistent with the presence of numerous cytoplasmic clathrin-coated structures. Such structures, distinct from those in the TGN, are virtually undetectable in control cells (24,25). Inhibition of clathrin-mediated transport from both the plasma membrane and TGN was also observed under these conditions. It should be noted that the proliferation of uniform, punctate intracellular clathrin signal under these conditions was quite distinct from the aggregates of clathrin in granules that have been observed upon expression of dominant-negative forms of other clathrin-binding proteins such as auxilin or AP180 (26). Given the ubiquitous roles played by many different phosphoinositides in membrane trafficking, clathrin binding and activation of PI3K-C2␣ suggests a possible mechanism for the localized and timed formation of 3-phosphoinositides at sites of membrane budding during endocytosis or Golgi transport. This notion is supported by the observation that both PI3K-C2␣ overexpression (24) and the putatively reciprocal knockout of synaptojanin phosphatase activity (27) result in the accumulation of unusual intracellular clathrin-coated structures, together suggesting defects in the coat recycling pathway. Elucidation of PI3K-C2␣ function under these conditions can be expected to provide insight into the mechanisms by which phosphoinositides regulate these aspects of membrane trafficking. To provide a basis for understanding PI3K-C2␣ function in cells, we report here an analysis of the roles of the CBD and of the lipid kinase activity in modulating PI3K-C2␣ effects on clathrin behavior in cells. Interestingly, we found that both the enzymatic and the clathrin binding activities of PI3K-C2␣ play distinct and apparently independent roles in regulating clathrin distribution and function.
Miscellaneous Procedures-For immunofluorescence studies and transferrin uptake experiments, COS1 cells transiently transfected with vector, full-length wild type, or mutant PI3K-C2␣ constructs were processed as described previously (24). Quantitation of Alexa-594 transferrin uptake (20 g/ml for 10 min) and of clathrin-coated structure number was performed by analyzing pairs of PI3K-C2␣-expressing and non-expressing cells in the same field, following background correction. Ten separate fields from multiple experiments were evaluated for each construct. For transferrin uptake, average pixel intensity in the perinuclear recycling compartment was quantitated, while the number of anticlathrin antibody (R5 antibody, Ref. 25)-labeled spots within a 30.25m 2 (50 ϫ 50 pixel) box toward the periphery of the cell were counted for determination of clathrin coat structures. Image processing was performed using MetaMorph software (Molecular Devices, Inc.). For immunoprecipitation studies, cells were lysed 48 h after transfection in 100 mM MES-Na (pH 6.8), 0.1% Triton X-100, 1 mM EGTA, 0.5 mM MgCl 2 , 0.02 NaN 3 , supplemented with protease inhibitors. Clarified lysates (12,000 rpm, 10 min) were centrifuged at 80,000 rpm for 15 min in a TLA100 rotor. Pellets were resuspended in lysis buffer and challenged with anti-Glu antibody. The immunoprecipitates were blotted with anti-Glu or anti-clathrin antibody TD1 (ATCC, Manassas, VA). Clathrin coat assembly was performed as described (30). Briefly, 100 g of clathrin alone or mixed with equimolar amounts of either AP-2 or purified PI3K-C2␣-(2-134) was dialyzed overnight against 0.1 M MES-Na, pH 6.5, 1 mM EGTA, 0.5 mM MgCl 2 , 0.02% NaN 3 . Assembled clathrin structures were sedimented by ultracentrifugation (80,000 rpm, 10 min, TLA100 rotor), and equal proportions of pellets were analyzed by SDS-PAGE. Assembled clathrin structures induced by PI3K-C2␣-(2-134) were further visualized by negative staining electron microscopy as described previously (30).
rin-coated structures and inhibition of clathrin function in endocytosis ( Fig. 2A) and TGN sorting (24). To investigate the role of the kinase activity in these processes, we generated a lipid kinase-defective mutant. In this mutant, residue Asp-1250 in the catalytic loop region (corresponding to Asp-166 in the protein kinase A catalytic subunit), conserved between all classes of PI 3-kinases as well as other lipid and protein kinases, was changed to alanine. In addition, two residues (Lys-1138 and Asp-1157) involved in ATP binding, and corresponding to protein kinase A catalytic subunit residues Lys-72 and Asp-89, were replaced with alanine. The resulting epitope-tagged kinase-dead mutant (KD-PI3K-C2␣) and wild type (WT) proteins were immunoprecipitated from lysates of expressing cells and tested for kinase activity. At similar levels of overall expression, no catalytic activity toward phosphatidylinositol was detected in immunoprecipitates containing the KD protein in contrast to the wild type protein (Fig. 1B), confirming that kinase activity had indeed been ablated by the mutations.
When epitope-tagged KD-PI3K-C2␣ protein was transiently expressed in COS1 cells and subjected to immunofluorescence, it exhibited a punctate pattern throughout the cytoplasm with some concentration in the perinuclear region characteristic of the TGN, a pattern indistinguishable from that of the WT protein ( Fig. 2A and Ref. 24). As with the wild type protein, localization in the nucleus (31) was only very occasionally observed. These results suggest that, as expected, the protein was correctly folded. Further, the results indicate that the kinase activity of the protein is not required for its correct targeting.
Next we wished to evaluate the effect of transiently expressing KD-PI3K-C2␣ on the proliferation of novel intracellular clathrin-coated structures and inhibition of clathrin-mediated endocytosis that characterize expression of the wild type protein. Surprisingly, expression of the mutant protein also resulted in proliferation of intracellular clathrin-coated structures, virtually indistinguishable from the effect of wild type enzyme ( Fig. 2A). Similarly, cells expressing KD-PI3K-C2␣ were also defective in endocytosis of transferrin (Fig. 2B). Quantitation of multiple fields confirmed these effects (Fig. 3). The effects on clathrin distribution and on transferrin internalization were observed at similar levels of wild type and mutant enzyme overexpression. These data indicate that determinants other than the lipid kinase activity of PI3K-C2␣ are sufficient to elicit the effects on clathrin distribution and function.
Stable Cell Lines Reveal Ultrastructural Changes upon PI3K-C2␣ Expression-To more fully characterize the effects of WT-and KD-PI3K-C2␣ expression upon clathrin morphology, we established stable inducible cell lines expressing each of these proteins in Tet-regulatable MEF cells. As with the transiently expressing cells (Fig. 2A),  light microscopy revealed that both the WT and the KD-PI3K-C2␣ proteins are localized to punctate spots throughout the cytoplasm and are concentrated in the perinuclear region; further they show an increase in intracellular clathrin and a modest inhibition of transferrin uptake (data not shown). Ultrastructural analysis of the stable cell lines revealed important changes in coated membrane profiles. In parental MEF cells, characteristic bristle-coated pits were observed on the plasma membrane and occasionally as discrete profiles nearby (Fig. 4, upper left). In the WT-PI3K-C2␣-expressing cells, no significant change was observed in the frequency of plasma membrane-coated pits (data not shown). However, a striking increase in the number of intracellular coated profiles was observed (Fig. 4, right panels). Most frequently these had the appearance of coated buds on the ends of tubules. Quantitation of the number of these coated buds suggests that their appearance is ϳ10-fold more frequent in the WT-PI3K-C2␣-expressing cells (2.4 coated structures/m 2 ; 9 fields analyzed) than in the parental line (0.23 coated structures/m 2 ; 8 fields analyzed). Evaluation of the KD-PI3K-C2␣ cell line (Fig. 4, lower left panel) also revealed a pronounced increase in intracellular coated membrane profiles (1.8 coated structures/m 2 ; 11 fields analyzed).
These results confirm the immunofluorescence observations of a dramatic proliferation of intracellular clathrin-coated structures reported here (Fig. 2) and previously (24). Further, they indicate that the lipid kinase activity of the exogenous PI3K-C2␣ molecule is not required to elicit the effect. In most cases the ultrastructural images suggest the presence of coats on small, circular membrane profiles of ϳ70 -90-nm diameter. Whether these structures are in fact discrete coated vesicles, coated buds on the end of tubular segments, or other coated membrane structures will require more elaborate ultrastructural studies such as serial section analysis. Observation of fluorescently tagged protein molecules in live cells may provide an alternative perspective, and preliminary results in our laboratory are consistent with the presence of clathrin-coated buds exhibiting highly dynamic but restricted local mobility. 4 Functional Domains in PI3K-C2␣-The finding that the lipid kinase activity of PI3K-C2␣ was not required to elicit either the marked changes in clathrin distribution or inhibition of transferrin uptake prompted us to further analyze which domain(s) of the protein contributed to these phenomena. Accordingly, several constructs comprising different domains of PI3K-C2␣ were prepared and transiently expressed in cells. First, we expressed the N-terminal half of PI3K-C2␣ (amino acids 2-873), truncated before the catalytic regions (Fig. 1A,  NT). This N-terminal fragment expressed well and exhibited the punctate pattern of cytoplasmic fluorescence with perinuclear concentration (Fig. 5A) characteristic of both the endogenous and WT protein (14,24). Furthermore, it was able to potently induce formation of intracellular clathrin-coated structures and, like the WT protein, it colocalized with these structures (Fig. 5A, insets). Finally, again like the full-length WT protein, it was able to inhibit endocytosis of transferrin at expression levels comparable with those of full-length WT or KD proteins (Fig. 5D). These results (quantitated in Fig. 3) extend the finding that neither functional kinase activity nor the structural presence of the C-terminal half of the protein is required to elicit the proliferation of intracellular clathrin structures and concomitant inhibition of endocytosis.
As the N-terminal half of PI3K-C2␣ contains the clathrin binding domain that we have identified previously at the extreme N terminus of the protein (24), we then focused our attention on this region. First, expression of the N-terminal half of the protein lacking the CBD (residues 143-873) neither colocalized with clathrin nor had any demonstrable effect on its distribution (Fig. 5B), suggesting that the activity of this half of the protein can indeed be ascribed to the CBD. In confirmation, we found that expression of this fragment alone (residues 2-143) was indeed able to induce the formation of intracellular clathrin-coated structures, though only at higher levels of expression (Fig. 5C). This may reflect less effective folding or stability of this short fragment upon expression or that there are additional functional determinants located toward the C terminus of the protein. In any case, the results demonstrate that the ability of the catalytically inactive PI3K-C2␣ to elicit 4 Y. Zhao, I. Gaidarov, and J. Keen, manuscript in preparation.   DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 changes in clathrin distribution is attributable at least in part to the CBD of the enzyme.

PI3K-C2␣ Domain Activities Regulate Clathrin Function
The PI-3KC2␣ CBD Possesses Clathrin Assembly Activity-Given the ability of PI3K-C2␣ fragments that contain the CBD to induce the proliferation of intracellular clathrin-coated structures, we hypothesized that the CBD might itself possess clathrin assembly activity that could account for the formation of these structures through clathrin polymerization. Our previous work demonstrated that this region of the protein could indeed bind to intact clathrin triskelia in precleared cell lysates (24). The clathrin binding motif we noted in PI3K-C2␣ has been shown to interact specifically with the terminal domain of the clathrin heavy chain (29,32). As expected from these characteristics, we found that clathrin terminal domain (heavy chain amino acids 1-579) fused to GST was able to efficiently pull out the PI3K-C2␣ fragment of amino acids 2-873 from expressing COS1 cells (Fig. 6A).
Next we incubated purified clathrin triskelia with bacterially expressed polyhistidine-tagged PI3K-C2␣-(2-134) under conditions in which clathrin assembly requires a separate factor to polymerize into coat structures (30). Clathrin assembly was analyzed by ultracentrifugation followed by SDS-PAGE analysis. The N-terminal PI3K-C2␣ fragment supported efficient polymerization of clathrin into a sedimentable state, comparable in extent to AP-2-mediated assembly used as a con-trol (Fig. 6B). The macromolecular products of the assembly reaction were visualized by negative staining electron microscopy, revealing the formation of complete clathrin lattices in the presence of the PI3K-C2␣ N-terminal fragment (Fig. 6C). These data demonstrate that PI3K-C2␣ binds clathrin terminal domain and can induce clathrin assembly. It seems likely that this activity is responsible for the formation of assembled clathrin structures observed upon expression of PI3K-C2␣ constructs.
PI3K-C2␣ Kinase Activity and the CBD Operate Independently-Our results to this point indicate that the CBD is sufficient to induce the proliferation of intracellular clathrin structures and that the lipid kinase activity of PI3K-C2␣ is not required. To test whether the catalytic activity of PI3K-C2␣ can independently play a role in the modulation of clathrin distribution in cells it was necessary to remove the clathrin binding domain from PI3K-C2␣. Accordingly epitope-tagged deletion mutants of PI3K-C2␣ lacking N-terminal residues 1-142 in catalytically active (⌬CBD) or inactive (⌬CBD/KD) proteins (Fig. 1A) were constructed and expressed in COS1 cells.
Cells transiently expressing the indicated constructs were lysed, the epitope-tagged proteins were immunoprecipitated from resuspended high speed pellets, and the presence of associated clathrin was then determined by immunoblotting (Fig. 7A). As expected, the presence of the CBD conferred the ability to bind clathrin, as the latter was present in immunoprecipitates of both full-length WT-and KD-PI3K-C2␣. Moreover, immunoprecipitation of the CBD itself also revealed the presence of clathrin, providing evidence of a stable association. These  A, beads containing GST (10 g) or GST fused with clathrin terminal domain (heavy chain residues 1-579) were incubated with the cytosolic fraction of COS1 cells expressing epitopetagged PI3K-C2␣ residues 2-873. The bound PI3K-C2␣ fragment was detected by immunoblotting with anti-epitope antibody. B, purified bovine brain clathrin, alone or mixed with equimolar amounts of either AP-2 or purified PI3K-C2␣-(2-134), was incubated under conditions that support coat assembly (30). Assembled structures were sedimented, and equal proportions of pellets were separated by SDS-PAGE and stained with Coomassie Blue. C, negative staining electron microscopy reveals complete clathrin lattices formed in the presence of the CBD of PI3K-C2␣ (N-terminal fragment 2-134). data support our immunofluorescence (Figs. 3 and 5) and ultrastructural (Fig. 4) results showing that enzymatically inactive PI3K-C2␣ as well as the CBD alone are each able to mediate the formation of intracellular clathrin-coated vesicular structures and support the hypothesis that this activity is a direct consequence of the presence of the clathrin binding and assembly activity of the CBD (Fig. 6).
To evaluate the contribution of the kinase activity to the effect of PI3K-C2␣ on clathrin, cells expressing truncated PI3K-C2␣ constructs lacking the CBD (⌬CBD) were then examined. Interestingly, clathrin was efficiently recovered in immunoprecipitates of cells expressing the ⌬CBD at levels comparable with the full-length protein (when corrected for expression level). However, it was greatly diminished or absent in immunoprecipitates from cells expressing the kinase-dead version of the protein (Fig. 7A), demonstrating that the CBD and catalytic activity of PI3K-C2␣ can independently contribute to stable clathrin binding.
We then tested ⌬CBD catalytically active and inactive mutants for the ability to modulate clathrin distribution in intact cells (Fig. 7B). Analysis showed that expression of the catalytically active ⌬CBD mutant leads to the proliferation of intracellular clathrin-coated vesicular structures. In contrast, the enzymatically inactive ⌬CBD/KD mutant was completely incapable of affecting clathrin distribution upon expression. These data directly confirm independent roles for the catalytic and clathrin binding activities of PI3K-C2␣ in modulation of clathrin function in cells. Further, they suggest that the effect on clathrin-mediated processes observed upon expression of PI3K-C2␣ in cells is likely the summary result of both of these activities.
The observations reported here indicate that the lipid kinase and clathrin binding activities of PI3K-C2␣ each contribute, and can even operate independently, to affect clathrin distribution and function. This bimodal mechanism of action, though revealed by utilizing non-physiological expression levels, is nonetheless useful in probing aspects of the involvement of PI3K-C2␣ in normal membrane trafficking processes. For example, the results provide a plausible mechanism for recruitment of PI3K-C2␣ to nascent sites of clathrin coat formation. This likely occurs as a consequence of both PI3K-C2␣ affinity for clathrin and of membrane binding by its PX and C2 domains (33). This dual mode of interaction, involving both protein-protein and protein-lipid interactions, is similar to targeting mechanisms that have been suggested to operate in other systems (13,34,35). On the other hand, the assembly activity of the CBD of PI3K-C2␣ that we report here may also contribute directly to clathrin lattice assembly upon PI3K-C2␣ recruitment.
An important implication of the formation of stable clathrin-PI3K-C2␣ complexes at sites of endocytosis is the localized activation of the lipid kinase activity of PI3K-C2␣ (24). This may have several consequences. Local production of charged 3-(poly)phosphoinositides may affect the membrane directly or, because many of the key proteins involved in endocytosis (e.g. AP-2, AP180/CALM, epsin, and dynamin) show relatively low inositide specificity and a preference for more highly phosphorylated derivatives (8,36), may serve to help recruit and stabilize these proteins. Clearly it will be important to temporally resolve the appearance of these proteins in the coated pit, and further work with live cell systems should help to address these issues. Finally, another important potential outcome of the activation of PI3K-C2␣ in the coated pit may be the production of local PtdIns-3-P pools in membrane regions destined to enter the endocytic pathway. Lipids in these pools may be the downstream binding targets of the many 3-phosphoinositide-binding FYVE and PX domain proteins that have been identified and may be a key source of the 3-phosphoinositides that characterize the early endosomal compartment. Further work on the in situ regulation of the individual PI3K-C2␣ activities identified here will be essential to further refine our understanding of these possibilities.