FIP3 is a Rab11 binding protein essential for the completion of abscission [8, 9, 20]. Previous work from our group and others has shown that FIP3 exhibits considerable changes in subcellular distribution throughout the cell cycle. Here, we have investigated the potential role of phosphorylation as a mechanism to control FIP3 distribution. We have shown that FIP3 is a cell cycle-regulated phosphoprotein (Figure 7), and that levels of phosphorylation of FIP3 decrease as cells move from anaphase into telophase (Figure 7). This dephosphorylation of FIP3 is accompanied by a shift of FIP3 from a largely cytosolic fraction to a membrane fraction (Figure 7), consistent with the observed distribution of GFP-FIP3 reported by us and others [8, 21]. Using MS approaches we have identified four putative sites of FIP3 phosphorylation: S102, S280, S347 and S450 (Figure 1), extending the list of potential phospho-acceptor sites within FIP3 - see Figure 2C. Of these, we show that phosphorylation of S102 is cell cycle regulated.
S102 lies within a consensus site for proline-directed kinases, including Cdk1 [17, 22, 23]. Since Cdk1 activity is tightly regulated during the cell cycle [3, 17], we tested the hypothesis that Cdk1-cyclin B could phosphorylate this site in vitro. Figure 4 shows that recombinant Cdk1-cyclin B phosphorylated recombinant FIP3 in vitro solely on S102. Consistent with Cdk1-cyclin B kinase phosphorylating FIP3 in vivo, levels of pS102-FIP3 fall as cells progress from anaphase into telophase (Figure 3C, D). In an attempt to determine the functional significance of this phospho-acceptor site, we generated S102A and S102D mutants and expressed these in HeLa cells. Expression of GFP-FIP3-S102D but not GFP-FIP3-S102A resulted in a modest but reproducible increase in the number of binucleate cells, consistent with a defect in cytokinesis (Figure 5). However, this effect was significantly less than that observed upon expression of GFP-FIP3-I737E, a Rab11-binding defective mutant . The molecular basis of the role of S-102 in cytokinesis remains to be established (see below).
We examined the functional consequences of S > A or S > D mutation at all of the sites identified here (S-102, S-280, S-347 and S-450). With the exception of S102D, over-expression of these mutants in HeLa cells had no effect on the numbers of binucleate cells or the distribution of FIP3 in the cell cycle (not shown). Of note, we also expressed all combinations of double, triple and quadruple Aspartate mutants. None of these exhibited significant increases in the numbers of binucleate cells (compared to S102D alone). Similarly, the interphase role of FIP3 in the spatial control of the recycling endosomal compartment and in transferrin recycling did not appear to be impaired by over-expression of any of the phospho-mutants (Figure 6). One exception to this concerns GFP-FIP3-S450A. We consistently observed that transfection of HeLa cells with GFP-FIP3-S450A led to significant cell death, with little, if any, immuno-reactive GFP-FIP3-S450A observed (not shown). By contrast, GFP-FIP3-S450D did not modulate cell growth, and its localisation was identical to that of GFP-FIP3. Interestingly, when double, triple or quadruple phospho-mutants were generated, any mutant containing S450A failed to express (data not shown). Such observations suggest that this residue may be involved in FIP3 stability, processing or folding, but further studies will be required to determine the basis of this.
More than 20 phospho-acceptor sites on FIP3 (and its fly homologue Nuclear Fallout) have recently been identified . Phosphorylation of Nuf by IKK-ε has been shown to regulate the trafficking of Rab11-positive recycling endosomal-derived vesicles into and out of the fly bristle tip by coordinating its interaction with dynein [14, 24]. Mutation of S-488, 538, 647 and 648 to A (HA-FIP3-4[A]) was also shown to modulate the ability of IKK-ε to control recycling endosome directionality . We hypothesised that de/phosphorylation of FIP3 at these sites may control the interaction of FIP3-positive vesicles with the microtubule network and thus modulate the function and/or distribution of FIP3 in the cell cycle. However, over-expression of HA-FIP3 (wild-type) or HA-FIP3-[4A] did not affect cytokinesis as revealed by quantification of the levels of binucleate cells (Additional file 1: Figure S1). Similarly, the membrane/cytosol ratio of HA-FIP3 (wild-type) and HA-FIP3-[4A] in metaphase were similar (Additional file 1: Figure S1), consistent with these mutations not exerting a significant effect on the interaction of FIP3 with membranes. Similarly, the distribution of this mutant during the cell cycle was indistinguishable from the HA-tagged wild-type FIP3 (Figure 5).
FIP3 is a cell cycle regulated phosphoprotein
The identification of a range of phospho-acceptor sites within FIP3 begs the question of the functional role of these sites. Given the significant redistribution of GFP-tagged FIP3 from cytosol to membranes (Figure 1) during the cell cycle and our observation that S-102 is dephosphorylated during progression from metaphase to telophase (Figure 3), we further explored whether phosphorylation of FIP3 is modulated during the cell cycle. To test this, we used a phospho-protein affinity column to selectively bind phosphorylated proteins from HeLa extracts, isolated from prophase/metaphase arrested cells or from cells in telophase (Figure 7A). The phospho-affinity column retained FIP3 from prophase/metaphase arrested cells, but the extent of FIP3 binding was considerably reduced in telophase extracts. Such data suggest that FIP3 is phosphorylated in early stages of the cell cycle (metaphase) and dephosphorylated in later stages (telophase). As this corresponds to a movement of FIP3 from largely cytosolic (metaphase) to membrane associated (telophase), we tested this hypothesis by examining the membrane/cytosol distribution of FIP3 in metaphase homogenates, treated with or without phosphatase, and then subsequently separated into membrane and cytosol fractions. As revealed in Figure 7B, dephosphorylation of FIP3 consistently resulted in more FIP3 distributing to the membrane fraction. Such data support the hypothesis that FIP3 is a phosphoprotein. As a control for these experiments, we examined the distribution of another Rab11 effector, Rab Coupling Protein (RCP)/FIP1. The distribution of RCP/FIP1 between membrane and cytosolic fractions was not altered during the cell cycle. Similarly, the levels of phosphorylated RCP/FIP1 did not differ between metaphase and telophase. Such data suggest that the cell cycle-induced changes in localization and phosphorylation are specific to FIP3.
These data suggest that dephosphorylation of FIP3 accompanies its redistribution from a cytosolic fraction to a membrane fraction. Several lines of argument suggest that dephosphorylation of FIP3 alone is unlikely to be sufficient to mediate this. Firstly, we and others have shown that FIP3 is recruited to membranes via interaction with Rab11 [8, 9, 25]. Indeed, inspection of the data in Figure 6 reveals that a FIP3 mutant (I737E) unable to bind Rab11 is cytosolic . Hence, the interaction with Rab11 is likely to be the most important factor controlling FIP3 membrane distribution. Secondly, recombinant FIP3 does not bind either to phospholipid vesicles or to intracellular membranes isolated from HeLa cells (Figure 7C). Finally, neither S > D nor S > A mutants exhibited altered distribution of GFP-FIP3 throughout the cell cycle (Additional file 2: Figure 2 and data not shown). Collectively, these data suggest that although FIP3 is a cell cycle-regulated phosphoprotein, the consequences of phosphorylation are not primarily to control the localisation of FIP3. This is also consistent with the location of the phospho-sites within FIP3: all four sites studied here are located towards the N-terminal domain of FIP3, and are not within the coiled-coil domain which encompasses the sites of Rab11, Arf6 and Cyk-4 interaction (Figure 2C) [8, 11]. It should be noted that FIP3 has multiple other potential phospho-acceptor sites, and over 20 such sites have been reported to be phosphorylated in vitro . Some of these sites are located within the coiled-coil domain, and may modulate the interaction of FIP3 with other proteins or possibly control FIP3 homo-dimerisation. Mutation of the sites reported here did not in our hands modulate the distribution of dynein (not shown), nor affect the re-distribution of recycling endosomes characteristic of FIP3 over-expression (Figure 6). Similarly, traffic of cargo into and out of the clustered endosomes was not modulated by either S > A or S > D mutation of these sites singly (Figure 6) or collectively.