Accepted Article
Role of dynamin and clathrin in cellular trafficking of flotillins
Melanie Meister, Alexandra Zuk#, Ritva Tikkanen
Institute of Biochemistry, Medical Faculty, University of Giessen, Friedrichstrasse 24, 35392 Giessen, Germany
# Present address: University of Cologne, Institute of Biochemistry II, Joseph-StelzmannStraße 52, 50931 Cologne, Germany
Corresponding author, e-mail: [email protected], Tel.: +49641-9947420, Fax: +49-641-9947429
Running title: Dynamin & clathrin in flotillin trafficking
Article type
: Original Article
The authors declare no conflict of interest
Key Words: endocytosis, dynamin, recycling, clathrin, epidermal growth factor
Abbreviations: PM, plasma membrane; CME, clathrin mediated endocytosis; CCP, clathrin coated pit; CCV, clathrin coated vesicle; CIE, clathrin independent endocytosis; ARF6, ADPribosylation factor 6; EGF, epidermal growth factor; EGFR, EGF receptor; PHB, prohibitin homology; GPI, glycophosphatidyl-inositol; dyn2, dynamin-2; DMSO, dimethyl sulfoxide; tfn, transferrin; LatA, Latrunculin A; EtOH, ethanol; MAPK, mitogen activated protein kinase; WT, wildtype; CHC, clathrin heavy chain; KD, knockdown; BACE1, ß-secretase 1; MPR300, cation-independent mannose 6-phosphate receptor; EE, early endosome; LE, late endosome; GAP, GTPase activating protein; ACAP1, Arf-GAP with coiled-coil ANK repeat and PH domain containing protein 1; GGA3, Golgi localized gamma ear containing ADP ribosylation factor binding protein 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; DABCO, 1,4-diazadicyclo[2,2,2]octane; PBS, phosphate buffered saline; DAPI, 4’,6diamidino-2-phenylindole
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/febs.12834 This article is protected by copyright. All rights reserved.
Accepted Article
Abstract Flotillin-1 and flotillin-2 are highly conserved, membrane microdomain associated proteins that have been shown to be involved in signal transduction, membrane trafficking and cell adhesion. Upon growth factor stimulation, flotillins are tyrosine phosphorylated and become endocytosed from the plasma membrane into endosomes from which they are recycled back to the plasma membrane. Although a role for flotillin-1 in the endocytosis of certain cargo proteins has been suggested, it is not known how the growth factor induced endocytosis of flotillins is regulated and which endocytosis pathway is used. However, this is likely to be different from the pathway used by flotillin dependent cargo. In this study, we have addressed the mechanistic details of flotillin trafficking during growth factor signaling. We here show that dynamin-2 activity is required for the uptake of flotillins from the plasma membrane upon epidermal growth factor stimulation, and inhibition of dynamin-2 GTPase activity impairs flotillin endocytosis. Surprisingly, recycling of flotillins from endosomes to the plasma membrane appears to require both dynamin-2 and clathrin. Upon overexpression of dynamin-2 mutants or depletion of clathrin heavy chain, flotillins are permanently trapped in endosomes. These data show that clathrin and dynamin are required for the endosomal sorting of flotillins, and the present study provides a mechanistic dissection of the thus far poorly characterized endosomal trafficking of flotillins.
Introduction Endocytosis serves e.g. to incorporate nutrients and receptor bound cargo into the cell and also plays a role in the regulation of signal transduction. On the other hand, the loss of membrane by endocytosis needs to be compensated by pathways of recycling that return lipids and proteins to the plasma membrane (PM). Different classes of endocytosis have been described so far (for a review, see [1, 2]). The best characterized and understood mechanism is clathrin mediated endocytosis (CME), during which the cargo is incorporated in clathrin coated pits (CCPs) that pinch off from the PM towards the cytosol and thereby form clathrin coated vesicles (CCVs). In contrast to CME, clathrin independent endocytosis (CIE) reflects several distinct pathways for cargo to enter the cells [3-5]. In many cases, small GTPases such as cdc42 or ADP-ribosylation factor 6 (ARF6) have been suggested to regulate these pathways (reviewed in [6]), but the molecular details of CIE pathways are not as well characterized as those for CME. Several cargo molecules and regulators of CIE have been characterized, and an important role for membrane rafts in some forms of CIE has become apparent. Membrane rafts are cholesterol and sphingolipid enriched microdomains that sub-compartmentalize membranes and thereby serve as signaling or endocytosis platforms [7, 8]. However, some cargo molecules, such as the epidermal growth factor receptor (EGFR), have been shown to utilize both CME and CIE pathways in a ligand dosage dependent manner [9, 10].
The three members of the dynamin GTPase family have been shown to regulate both CME and CIE (reviewed in [11, 12]). While dynamin-2 (dyn2) is ubiquitously expressed, dynamin-1 expression is restricted to the brain in organisms [13-15], although its expression in some cultured cells has also been shown [16]. Dynamin-3 in turn has been detected in the brain,
This article is protected by copyright. All rights reserved.
Accepted Article
lung and testis [17-19]. The best characterized function of dynamins lies within vesicle fission during endocytosis. However, dyn2 has been suggested to regulate Golgi export of cargo proteins such as p75 [16], suggesting that the different dynamin proteins are not fully redundant. Although CME is clearly dynamin dependent, also some of the CIE pathways have been designated as dynamin dependent, whereas others appear to be dynamin independent [6]. However, the role of dynamin in endocytic trafficking appears not to be limited to formation of endocytic carriers only, since several publications have shown that dynamin may also play a role in sorting within endosomes, i.e. in endocytic recycling of cargo molecules [20-24].
Recently, a role during CIE has been suggested for the members of the flotillin family, flotillin-1 and -2 ([25-27] and see below). The expression of flotillin-1 is highly dependent on flotillin-2 in most cell types and also in the respective knockout mice that have recently been generated [28-30]. Flotillins participate in several cellular processes, such as signal transduction, regulation of the cytoskeleton and endocytosis (reviewed in [31]). In their Nterminus, flotillins contain a prohibitin homology (PHB) domain which mediates the association with the inner leaflet of the PM and contains motifs for acylation, i.e. myristoylation and palmitoylation which together with oligomerization mediate a constitutive membrane association of flotillins [32, 33]. In line with this, flotillins form higher order oligomers, a property mediated by their C-terminal parts containing potential coiled-coils [26, 33-35].
The subcellular localization of flotillins is rather dynamic. Under growth conditions, flotillins predominantly localize to the PM and endosomal/vesicular structures [36-38]. At the PM, flotillin microdomains were described to form flat platforms or invaginations that apparently bud into the cell [26, 27]. The cellular localization of flotillins can be modulated e.g. by growth factor stimulation. Upon stimulation with epidermal growth factor (EGF), flotillins are tyrosine phosphorylated by members of the Src kinase family and translocate from the PM to endosomes [35, 39, 40]. However, it is not known which kinds of carriers are responsible for the growth factor induced flotillin endocytosis or if dynamin is required for it.
During the last years, several cargo molecules have been suggested to be internalized by flotillin mediated endocytosis [25-27, 41-48], although this has recently been disputed for some of the cargo molecules [48-50]. Furthermore, none of these cargo proteins appears to specifically use the suggested flotillin dependent CIE route. In some cases the role of flotillins in cargo trafficking is not the endocytosis per se, but a step preceding the formation of endocytic carriers [48-50]. Thus, it has still not been conclusively demonstrated that flotillins are proper endocytic proteins that mechanistically participate in cargo trafficking. Furthermore, in some context such as growth factor signaling, flotillins are more likely to act as cargo proteins due to their role in signal transduction [49, 51]. Studies on the role of flotillins in endocytic processes have also produced different conclusions on the dynamin dependency. Whereas e.g. the CD59 endocytosis was concluded to be dynamin independent [26], uptake of other cargo molecules such as proteoglycans [45] or even
This article is protected by copyright. All rights reserved.
Accepted Article
polarized uptake of glycophosphatidyl-inositol (GPI) -anchored proteins [25] have been suggested to depend on both flotillin and dynamin, suggesting cargo or cell type specific differences. Thus, the mechanisms of flotillin dependent endocytosis of cargo molecules are still not conclusively characterized. Even less is know about the endocytosis of flotillins during EGF signaling, which results in a rapid uptake of flotillins from the PM to endosomes [39]. However, these two pathways, flotillin dependent cargo trafficking vs. flotillin endocytosis as cargo appear to be fundamentally different, and the former one was not addressed in this study. Instead, we here sought to characterize the trafficking of flotillins themselves during EGF signaling. We here provide data that dyn2 is important not only for the growth factor induced endocytosis of flotillins from the PM to endosomes, but it may also be required for endosomal sorting of flotillins and possibly for their recycling back to the PM. We here also unravel a novel and unexpected role for clathrin in the cellular trafficking of flotillins.
Results We have earlier shown that EGF stimulation results in flotillin uptake from the PM into endosomes [39]. On the other hand, flotillin-1 has been described to function in a dynamin independent, raft associated endocytosis pathway that mediates the uptake of ligands such as CD59 and GPI anchored proteins [27]. However, this pathway is constitutive [26, 27], whereas the EGF induced endocytosis takes place as a result of active signaling [35, 39, 40]. The dependency of the EGF induced flotillin uptake on dynamin function has not been studied, nor is the nature of the respective carriers known. To clarify the role of dynamin in flotillin endocytosis during EGFR signaling, we used two different modes to impair dyn2 function, namely chemical inhibition and overexpression of dyn2 mutants. To chemically inhibit dynamin, we used two different chemical inhibitors of dynamin function. Dynasore, a reversible, cell permeable inhibitor of dynamin GTPase activity, prevents vesicle fission at the PM [52-54], whereas MiTMAB interferes with the membrane association of dyn2 [55]. Mutation of lysine 44 (K44A) in dyn2 affects GTP binding and GTPase activity and results in blocked endocytosis [56-58], although in some cases, dyn2 K44A expression has also been shown to have an effect on endosomal trafficking of certain cargo molecules [20, 22]. R399A exchange impairs the membrane association of dyn2 and is considered as a milder dominant negative form of dynamin [59], whereas the T65A substitution produces a protein capable of GTP binding but deficient in GTP hydrolysis [60].
Starved HeLa cells were treated with Dynasore or dimethyl sulfoxide (DMSO) as a control, stimulated with EGF and stained for flotillins (Fig. 1A and Fig. 2A). Flotillin endocytosis upon growth factor stimulation can be detected either by live imaging [39] or in fixed cells [35, 39] as appearance of flotillin positive, perinuclear vesicular structures, whereas in starved cells, flotillins almost exclusively reside in the PM. The endocytosis is scored as percentage of cells showing a clear perinuclear accumulation of flotillins (see e.g. Fig. 1A, third row from top). Our quantitative analysis showed that Dynasore efficiently blocked the EGF induced uptake of flotillins from the PM (Fig. 1B). Uptake of fluorescently labeled transferrin, which was used as a control, was also prevented (Fig. 2B, middle). We also used another dynamin inhibitor, MiTMAB which prevents the membrane binding of dynamins by interfering with the
This article is protected by copyright. All rights reserved.
Accepted Article
lipid binding domain, without affecting the GTPase activity as such [55]. This inhibitor also prevented the endocytosis of both transferrin (Fig. 2B, right) and flotillins (Fig. 2C).
Recent findings have suggested that Dynasore inhibits fluid phase endocytosis and membrane ruffling also in cells that lack all three dynamin isoforms, implicating that it may have some unspecific effects on certain forms of endocytosis [61]. Since endocytosis by means of fluid phase and uptake by ruffling both require actin [62] and Dynasore has been suggested to directly affect the actin cytoskeleton [63], it was important to exclude that the inhibition of flotillin endocytosis by Dynasore was due to such adverse effects influencing actin rather than dynamin. Thus, we used Latrunculin A (LatA), an agent that inhibits actin polymerization, and analyzed the EGF dependent endocytosis of flotillins. LatA was applied for 1 h prior to EGF stimulation on starved cells. Despite disruption of the actin cytoskeleton, flotillin-2 was endocytosed in LatA treated cells (Fig. 3A, left row), although the staining appeared slightly more dispersed as compared to the control cells (10 min EGF + EtOH). Depolymerization of the actin cytoskeleton was controlled using phalloidin staining (Fig. 3A, right row). Quantification of the data showed that the translocation of flotillin-2 in LatA treated cells was about equally efficient as in EtOH treated cells (Fig. 3B). Very similar results were obtained when another actin depolymerizing drug, cytochalasin D, was used (data not shown). Thus, these data show that the inhibitory effects of Dynasore and MiTMAB on flotillin endocytosis are not due to the adverse effects of these dynamin inhibitors on actin.
Although EGF stimulation induces endocytosis of flotillins, it is not known if EGFR kinase activity or MAP kinase activation are directly required for it. Serum starved cells were incubated with the EGFR kinase inhibitor PD 153035 or the non-inhibiting control substance AG9 for 5 min and then stimulated with EGF in the presence of these substances for 15 min (Fig. 4A, left). Immunofluorescent staining for flotillin-2 and quantification of the data (Fig. 4B) demonstrated that flotillin endocytosis was equally efficient despite EGFR kinase inhibition. Inhibition of EGFR was controlled by means of Western Blot and quantified (Fig. 4C). Similarly, inhibition of the MAP kinase kinase MEK, the activity of which is essential for the activation of the MAP kinases ERK1 and ERK2 downstream of the EGFR, did not prevent the EGF induced endocytosis of flotillins (Fig. 4A, right & Fig. 4D). Inhibition of MEK activity was controlled using phospho-specific antibodies against its phosphorylated substrates ERK1/2 and quantified (Fig. 4E). Thus, these data show that neither EGFR nor MAP kinase activity are required for flotillin endocytosis.
To gain more insight into the kinetics of flotillin endocytosis, we used costaining of flotillins with different endosomal markers after EGF stimulation (Fig. 5). Flotillin-2 was found to rapidly localize to a compartment positive for the tetraspanin CD63/LAMP3 which is a marker of multivesicular bodies that represent an intermediate to late endosomal form. This colocalization was evident after 5 min EGF and became more intense after 10 and 30 min (Fig. 5A), as shown by analysis of the Pearson correlation coefficient (Fig. 5B). However, only a very low degree of colocalization of flotillin-2 with LAMP3 was detected upon Dynasore treatment (Fig. 5A-B). Furthermore, the colocalization of flotillin-2 with the early
This article is protected by copyright. All rights reserved.
Accepted Article
endosomal marker Rab5 was very limited at all time points in both Dynasore treated and DMSO treated cells (Fig. 5C-D). Similarly, the colocalization of flotillin-2 with transferrin receptor (TfnR) was very low in EGF stimulated cells without Dynasore (Fig. 6A). However, some degree of colocalization was seen at the PM upon Dynasore treatment, consistent with the inhibition of endocytosis of both flotillin-2 and TfnR by Dynasore (Fig. 6A).
Our earlier findings have suggested that in serum-grown cells, flotillins are continuously recycled between plasma membrane and endosomes and accumulate at the plasma membrane during starvation [39]. However, the possible role of dynamin in the retrograde trafficking of flotillins back to the plasma membrane has not been studied. Thus, we hypothesized that if Dynasore is applied during starvation, we might detect an effect on the retrograde transport of flotillins and see them accumulating in endosomes. We therefore incubated cells in serum-free medium overnight in the presence of Dynasore. Flotillin-1 and flotillin-2 were localized at the plasma membrane in starved cells treated with DMSO overnight (Fig. 6B, left). Similarly, in Dynasore treated, starved cells, flotillins were present at the plasma membrane (Fig. 6B, middle). The endosomal localization of transferrin receptor was also partially perturbed by overnight Dynasore treatment (Fig. 6B, right).
To inhibit dynamin function, we used an alternative strategy to chemical inhibition. Mutation of the Lys44 residue to Ala (K44A) in dyn2 results in a mutant protein that is not capable of high affinity binding of GTP and thus shows a lower activity than the wildtype (WT) protein [56-58], whereas T65A mutant has a very low GTPase activity although it is capable of GTP binding [60]. Mutation R399A in turn impairs the membrane association of dyn2 [59]. As a result of these mutations, dynamin dependent processes such as endocytosis are inhibited. We expressed GFP-tagged WT, K44A, T65A or R399A dyn2 in HeLa cells (Fig. 7 and 8). In GFP transfected, starved cells, flotillins were to a large degree localized in the PM. However, expression of K44A resulted in accumulation of flotillins in the perinuclear region already in starved cells (Fig. 7A). Quantification of the data showed that in starved or EGF stimulated, K44A expressing cells, over 90% of the cells show a perinuclear flotillin staining (Fig. 7B). A less profound but significant accumulation was also seen in the cells expressing the WT dyn2 (Fig. 7B). After EGF stimulation, a very strong condensation of flotillins in the same regions became apparent in K44A (Fig. 7A), R399A and T65A (Fig. 8) expressing cells. However, the signals for flotillin and dyn2 did not overlap, but flotillins appeared to localize in the vicinity of the dots containing K44A dyn2, and the compartment containing flotillins in these cells frequently appeared more condensed as compared to the GFP transfected control cells. In starved, K44A dyn2 expressing cells, flotillin-2 colocalized with LAMP3 (Fig. 9), implicating that it accumulates in late endosomes. Thus, K44A, T65A and R399A dyn2 expression clearly altered the cellular localization of flotillins in a similar fashion, resulting in late endosomal accumulation.
Given that the chemical inhibition of dyn2 activity and the overexpression of the mutant forms produced different results, we set out to identify the cause of this discrepancy. Since the inhibition of dynamin GTPase activity seemed not to be the cause of the accumulation of
This article is protected by copyright. All rights reserved.
Accepted Article
flotillins in endosomes and their impaired recycling, we considered the possibility that dyn2 mutants might sequester some factors that are required for flotillin recycling. Beyond its normal function in endocytosis, clathrin has been suggested to play a role as a coat protein in endocytic recycling [64, 65]. In the case of e.g. transferrin receptor, the recycling requires not only clathrin but also dynamin [23, 24]. Thus, we tested if clathrin depletion would display an effect on the recycling of flotillins. Clathrin heavy chain (CHC) was depleted in HeLa cells using two independent siRNA sequences, resulting in an efficient knockdown of CHC (Fig. 10A, C-E). The cells were starved and labeled for flotillin-2. Surprisingly, flotillin-2 exhibited an intracellular localization also in starved cells depleted of CHC, instead of the PM localization seen in control cells (Fig. 10A). Quantification of flotillin translocation showed that over 90% of the starved CHC knockdown cells exhibited flotillin translocation, whereas far less than 20% of the control siRNA cells did so (Fig. 10D). To test this further, we used a flotillin-2-GFP fusion protein in which Tyr163 has been mutated to Phe (F2-Y163F-GFP). We have previously shown that this mutant is exclusively localized at the PM and does not translocate to endosomes upon EGF stimulation [35, 39]. Intriguingly, an endosomal localization of this mutant was detected in starved CHC-KD cells (Fig. 10B). Furthermore, a prominent colocalization of flotillin-2 with LAMP3 was observed in starved CHC-KD cells (Fig. 11A), whereas no colocalization could be observed with the early endosomal marker Rab5 (Fig. 11B). This might implicate that impairment of clathrin function prevents the translocation of flotillins from endosomes to the PM, similarly to the expression of K44A, T65A or R399A dyn2.
To provide further insight into this surprising role of clathrin in the retrograde trafficking of flotillins, we overexpressed the C-terminal part of CHC, the so-called “Hub domain” which has previously been shown to inhibit clathrin mediated endocytosis [66-68]. In serum-starved cells expressing the Hub domain, flotillin-2 and flotillin-1 were partly localized to the PM, but a clear accumulation in the perinuclear region was also seen (Fig. 12A), and flotillin-2 exhibited a clear colocalization with LAMP3 (Fig. 12B), demonstrating that CHC Hub expression results in a similarly altered localization of flotillins as CHC depletion and mutant dyn2 expression. To test if the effect of K44A overexpression might also affect the localization of CHC, we stained WT and K44A dyn2 transfected cells for CHC. As shown in Figure 13, high level overexpression of WT or K44A dyn2 resulted in their accumulation in cytoplasmic patches which also contained CHC. Thus, K44A dyn2 expression not only causes an impairment of flotillin localization but also results in sequestration of CHC in cytoplasmic aggregates, which may also impair endosomal sorting steps that involve clathrin.
Discussion Our previous data have shown that during growth factor signaling, flotillins are endocytosed but they do not affect the early endocytic trafficking of EGFR [49]. Thus, due to their function as regulators of MAP kinase signaling (reviewed in [51]), flotillins are likely to be internalized as part of a signaling complex, i.e. as cargo. However, the immediate internalization pathways of flotillins vs. EGFR appear to be different, since no colocalization between flotillins and EGFR can be observed after short times, whereas a partial colocalization is
This article is protected by copyright. All rights reserved.
Accepted Article
observed after 15 min EGF stimulation [39], implicating that the originally separate endocytosis routes merge in late endosomes.
Flotillins have been suggested to define a non-clathrin, non-caveolar endocytosis pathway [27]. The GPI-anchored protein CD59 and the sphingolipid GM1 were shown to be partially endocytosed through this pathway. Furthermore, based on the additive effect of flotillin-1 depletion and K44A dyn2 expression on the uptake of cholera toxin, the ligand of GM1, it was concluded that the flotillin dependent endocytosis of these cargo molecules was dynamin independent. However, the direct influence of dynamin on the growth factor induced endocytosis of flotillins has not been studied so far, although these two pathways are likely to be separate. The idea that in some specific context, such as growth factor receptor activation, flotillins may not be part of the endocytic machinery, but rather represent cargo molecules has been neglected by most studies. Using two different chemical inhibitors of the GTPase dynamin, we here show that dyn2 is indeed involved in the EGF induced endocytosis of flotillins, since inhibition of dyn2 GTPase activity (Dynasore) or impairment of its membrane localization (MiTMAB) in starved cells prevented the uptake of flotillins from the PM upon EGF stimulation. Importantly, we here demonstrated that these effects are not due to the described adverse effects of dynamin inhibitors on actin cytoskeleton and endocytosis pathways that require actin [61, 62]. Although most of the endocytosis pathways described are actin dependent, some CIE pathways appear not to depend on actin [6, 69]. Since flotillins are endocytosed by means of a CIE pathway whose exact identity is not known, it seems in the light of our data that this pathway is also actin independent.
In serum grown HeLa cells, flotillins localize to both PM and LAMP3 positive endosomes, whereas upon growth factor deprivation, flotillins translocate to the PM [35, 39]. This implies that under growth conditions, flotillins continuously recycle between endosomes and PM. Recycling of flotillin containing vesicles has previously been shown [70], and very recent findings have even suggested that flotillin-2 might be involved in the recycling of E-cadherin, transferrin receptor [71] and the β-secretase BACE1 [72]. Our data here suggest that recycling of flotillins themselves may involve dynamin and clathrin.
Expression of K44A, T65A or R399A dyn2 and chemical dynamin inhibition appeared at first glance to produce somewhat contradictory effects on flotillin trafficking. Whereas Dynasore and MiTMAB clearly displayed an effect on flotillin uptake, K44A, T65A and R399A dyn2 appeared to trap flotillins in endosomal structures, which we never observed upon chemical inhibition of dynamin. Therefore, mutant dyn2 expression most likely specifically blocks flotillin recycling and translocation to the PM and thus prevents the normal accumulation of flotillins at the PM during starvation. Although the main function of dynamins may well be in endocytosis, some previous publications have already suggested that dynamin plays a role in endosomal recycling. For example, dynamin has previously been shown to play a role in the endocytic recycling of the Tfn receptor [23]. Likewise, it has been shown that expression of K44A dyn2 results in trapping of other recycled cargo molecules such as the cationindependent mannose 6-phosphate receptor (MPR300) [22] or ricin toxin [20] within
This article is protected by copyright. All rights reserved.
Accepted Article
endosomes. The findings of Llorente et al. [20] are especially intriguing in the light of a recent publication demonstrating that the transport of ricin and Shiga toxin from endosomes to Golgi was significantly reduced upon flotillin knockdown [73]. Thus, our present data suggest that the effect of K44A dyn2 on ricin toxin, as observed by Llorente et al. [20], may actually be indirect and caused by trapping of flotillins in endosomes, which in turn impairs ricin trafficking.
Interestingly, we found that clathrin heavy chain was accumulated in clusters in K44A dyn2 positive aggregates, suggesting that impairment of clathrin function might cause the aberrant localization of flotillins. In line with the hypothesis that K44A dyn2 overexpression results in clathrin sequestration and prevents the endosomal recycling of flotillins, knockdown of CHC and expression of the dominant negative CHC Hub fragment [66, 74] also resulted in endosomal accumulation of flotillins. Both CHC depletion and Hub expression have been suggested to result in altered localization and perinuclear accumulation of early endosomes (EE) and late endosomes (LE) [74, 75]. Thus, the collapse of EE into the perinuclear region might theoretically prevent the transfer of cargo, including flotillins, from EE to LE. On the other hand, overexpression of CHC Hub was shown to prevent the uptake of Tfn into cells, whereas receptor recycling appeared not to be impaired, despite the effects on endosome localization [74], speaking against this possibility. Furthermore, in CHC depleted cells, flotillins clearly colocalized with the multivesicular body marker LAMP3. Taken together, although flotillins are known as lipid microdomain associated proteins, their recycling appears to be dependent on clathrin. Furthermore, the endocytosis incompatible flotillin-2 Y163F mutant [35, 39] which is normally exclusively localized at the PM but not in endosomes was also found to accumulate in endosomes upon CHC knockdown, suggesting that also the biosynthetic transport of flotillin-2 may proceed through endosomes in a clathrin dependent fashion. This would be consistent with the findings showing that a considerable fraction of biosynthetic cargo passes the recycling endosomes before appearing at the PM [76].
Clathrin has been shown to play a role not only in endocytosis but also in endosomal sorting and recycling of proteins from endosomes to the PM [23, 64, 77-79]. It has been demonstrated that clathrin coats are present in endosomes in which they even appear to exhibit two different functions [80]. One class of endosomal clathrin coats mediates the sorting of ubiquitinated cargo towards degradation [77-79], whereas the other one is involved in recycling of proteins towards the PM [23]. In fact, clathrin and dynamin have even been suggested to collaborate in the endosomal recycling of some cargo proteins [23, 24], which may also take place during flotillin sorting and recycling. It has been shown that a specific coat containing clathrin and Arf-GAP with coiled-coil ANK repeat and PH domain containing protein 1 (ACAP1), a GTPase activating protein (GAP) for ARF6, is involved in the stimulus induced recycling of integrins and the glucose transporter Glut4 [81-83]. Clathrin also cooperates in endosome to PM recycling with the Golgi localizing gamma-ear containing ADP ribosylation factor binding protein 3 (GGA3) during the sorting of BACE1 and hepatocyte growth factor receptor (also known as MET) [84-87]. Our recent data point to a role for flotillins in the endosomal sorting of BACE1, as flotillin depletion impairs BACE1 recycling to PM [72].
This article is protected by copyright. All rights reserved.
Accepted Article
Our present data are of special interest since there are findings suggesting that also clathrin mediated endocytosis and flotillin function may be connected [43, 48, 50]. Flotillins have been suggested to be necessary for the clathrin mediated endocytosis of the Alzheimer amyloid precursor protein [48] and the dopamine transporter DAT [43]. However, flotillins appear not to be directly involved in the endocytosis of these proteins, but rather mediate their clustering in specific domains at the PM, which then cooperate with the CME machinery. Similarly, flotillin-1 appears to mediate the clustering of EGFR upon EGF stimulation [49]. It has been postulated that flotillins might serve as platforms that preassemble clathrin dependent cargo which is then handed over to CCPs for CME [88]. However, this possibility has not been studied in detail and is thus mechanistically not well understood. Our present data suggest that clathrin and flotillin functions may indeed be connected.
We have here provided important novel insight into cellular trafficking of flotillins which are associated with lipid microdomains by means of fatty acid modifications. Dynamin is apparently important for the growth factor induced endocytosis of flotillins. However, the most important findings of the present study revealed an unexpected role for clathrin in the endocytic recycling and possibly also biosynthetic trafficking of flotillins. These findings, together with those suggesting a collaboration of flotillin microdomains and CME components at the PM, imply that the division of trafficking pathways into clathrin dependent and raft dependent ones may not be justified, and that these pathways may intermingle and their components interact in an unexpected manner. It will thus be of special interest to characterize the exact role of flotillins in these trafficking processes, keeping in mind the still remaining question if flotillins are mechanistically involved in endocytosis/trafficking or merely cargo proteins.
Materials and Methods Antibodies and constructs Mouse monoclonal antibodies against clathrin heavy chain and flotillin-1 were purchased from Transduction Labs, and the rabbit polyclonal antibodies against flotillin-1 and -2 were from Sigma-Aldrich (Taufkirchen, Germany). Mouse monoclonal antibodies against LAMP3/CD63 and pERK1/2 as well as a polyclonal antibody against total ERK 1/2 (H-72) and Rab5 were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). An antibody directed towards the transferrin receptor was purchased from Life Technologies (Darmstadt, Germany) and an antibody towards glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from Abcam. Mouse monoclonal anti-myc and anti-phospho tyrosine (pY100) and rabbit polyclonal anti-EGFR antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). Secondary antibodies for immunofluorescence coupled to Cy3 and Cy5 were purchased from Jackson Immunoresearch (Newmarket, UK) and secondary antibodies coupled to Alexa488 were from Life Technologies. HRP-coupled goat anti-mouse and antirabbit secondary antibodies for detection of Western blots were purchased from Dako (Glostrup, Denmark).
This article is protected by copyright. All rights reserved.
Accepted Article
Rat dynamin-2-GFP fusion constructs (WT and K44A) were kindly provided by M. McNiven (Mayo College of Medicine, Rochester, MN, USA). The R399A and T65A dyn2 mutant were generated by site directed mutagenesis of the WT construct. The flotillin-2-Y163F-EGFP mutant and its transfection have been described earlier [39]. The dominant-negative clathrin heavy chain Hub region [74] was cloned in pCMV-Tag5A vector which contains a C-terminal myc-tag (Stratagene/Agilent Technologies, Waldbronn, Germany).
Cell culture and siRNA knockdown Human cervix adenocarcinoma cells (HeLa) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Life Technologies) with high glucose, supplemented with 10% fetal calf serum (Life Technologies), 100 units/ml penicillin and 100 µg/ml streptomycin (Life Technologies) at 37°C and 8% CO2. Transient depletion of clathrin heavy chain by siRNAmediated knockdown was performed as described earlier for flotillins [35]. The siRNA oligonucleotide duplexes as well as a negative control with medium GC content that does not target any human sequences were obtained from Life Technologies (StealthTM siRNA system). The experiments were performed 72 hours post-transfection.
Immunofluorescence HeLa cells grown on coverslips were treated with growth factors and inhibitors as indicated and then fixed in methanol for 10 min at -20°C. After blocking unspecific protein binding sites with 1% bovine serum albumin (BSA) in phosphate buffered saline (PBS), the cells were labeled with primary antibodies and Alexa-488 (Life Technologies), Cy5 or Cy3 conjugated (Jackson Immunoresearch) secondary antibodies. The incubation with secondary antibodies was combined with the nuclei counterstaining with 4',6-diamidino-2-phenylindole (DAPI). The specimen were mounted in Gelmount supplemented with 1,4-diazabicyclo[2.2.2]octane (DABCO, Sigma-Aldrich) and examined using a confocal laser scanning microscope (Zeiss LSM 710 combined with Axio Observer, Carl Zeiss, Jena, Germany).
Treatment with growth factors and inhibitors For growth factor treatments, HeLa cells were serum-starved for at least 16 hours and then stimulated with 100 ng/ml epidermal growth factor (EGF, Sigma-Aldrich) for the indicated time points.
For EGFR tyrosine kinase inhibitor treatments, the cells were pretreated with the inhibitor PD153035 (20 nM, Calbiochem), a non-inhibiting control AG9 (20 nM, Calbiochem) for 5 min before EGF was applied. To inhibit MEK, the cells were pretreated with the specific MEK inhibitor U0126 (10 µM, Cell Signaling) or DMSO for 120 min and then stimulated with 100 ng/ml EGF. In general, EGF stimulation was performed in the presence of inhibitors and their non-inhibiting analogs as negative control.
This article is protected by copyright. All rights reserved.
Accepted Article
To inhibit the GTPase activity of dynamin, the cells were pretreated with 80 µM Dynasore (Sigma-Aldrich) or DMSO as a negative control for 2 hours and then stimulated with EGF. Another dynamin inhibitor MiTMAB (Abcam, Cambridge, UK) interferes with the PH-domain of dynamins and inhibits dynamin mediated fission [55]. Cells were pretreated with 30 µM MiTMAB for 30 min at 37°C and then stimulated with 100 ng/ml EGF in the presence of the inhibitor.
For the uptake of transferrin (Tfn), the cells were chilled on ice, incubated with AlexaFluor546 conjugated Tfn (1:500) (Molecular Probes/Life Technologies) for 10 min on ice, washed with PBS and then transferred to pre-warmed serum free medium containing negative controls or dynamin inhibitors. Cells were allowed to internalize Tfn for 10 min. Surface bound Tfn was stripped in stripping buffer (300 mM NaCl, 200 mM acetic acid, pH 2.9) before fixation to facilitate the detection of intracellular Tfn. Cells were fixed in 4% paraformaldehyde for 10 min (80 mM PIPES pH 6.8, 2 mM MgCl2, 5 mM EGTA pH 8.0, 4% PFA) at room temperature before mounting in Gelmount (Biomeda, Foster City, CA, USA) supplemented with 50 mg/ml DABCO.
To check the influence of the actin cytoskeleton on flotillin translocation the actin cytoskeleton was depolymerized with 1 µM LatA for 1 hour at 37°C prior to stimulation with 100 ng/ml EGF for 10 min. Thereafter the cells were either fixed in methanol for a subsequent immunofluorescent staining of flotillins or fixed in 4% PFA, permeabilized with 50 µg/ml digitionin, labeled with Alexa488-coupled phalloidin (Molecular Probes/Life Technologies) and counterstained with DAPI. Cells were embedded in Gelmount supplemented with DABCO.
Cell lysis and Western blot Cells were washed in PBS, scraped in lysis buffer (50 mM Tris HCl pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% NP-40) supplemented with protease inhibitor cocktail (Sigma-Aldrich) and 1 mM sodium fluoride and cleared by centrifugation. Total protein concentration was estimated from the cleared lysates with the Bio-Rad protein assay reagent (Bio-Rad, Munich, Germany). Equal protein amounts were analyzed by SDS-PAGE and Western blot.
Statistical analysis Unless otherwise stated, all experiments were performed at least three times independent from each other. Densitometric analysis of Western blots was performed with Quantitiy One Software (Bio-Rad). For quantification of flotillin translocation after immunofluorescent staining, at least 50 cells per condition were counted and the percentage of cells showing translocation from the PM to intracellular membranes was assessed and used for data analysis. For the analysis of colocalization, the Pearson’s correlation coefficient (Rr) was calculated from a defined region of interest (ROI) for at least 15 cells per condition using the JACoP plug-in for ImageJ (NIH, Bethesda, MD, USA). Statistical comparisons were
This article is protected by copyright. All rights reserved.
Accepted Article
performed with Student’s t-test (unpaired, two-tailed) or analysis of variance (one- or twoway ANOVA), where appropriate, using Graph Pad Prism 5 Software (GraphPad Software, Inc., La Jolla, CA, USA). Data are shown as mean ± SD. Values of p
Role of dynamin and clathrin in cellular trafficking of flotillins
Melanie Meister, Alexandra Zuk#, Ritva Tikkanen
Institute of Biochemistry, Medical Faculty, University of Giessen, Friedrichstrasse 24, 35392 Giessen, Germany
# Present address: University of Cologne, Institute of Biochemistry II, Joseph-StelzmannStraße 52, 50931 Cologne, Germany
Corresponding author, e-mail: [email protected], Tel.: +49641-9947420, Fax: +49-641-9947429
Running title: Dynamin & clathrin in flotillin trafficking
Article type
: Original Article
The authors declare no conflict of interest
Key Words: endocytosis, dynamin, recycling, clathrin, epidermal growth factor
Abbreviations: PM, plasma membrane; CME, clathrin mediated endocytosis; CCP, clathrin coated pit; CCV, clathrin coated vesicle; CIE, clathrin independent endocytosis; ARF6, ADPribosylation factor 6; EGF, epidermal growth factor; EGFR, EGF receptor; PHB, prohibitin homology; GPI, glycophosphatidyl-inositol; dyn2, dynamin-2; DMSO, dimethyl sulfoxide; tfn, transferrin; LatA, Latrunculin A; EtOH, ethanol; MAPK, mitogen activated protein kinase; WT, wildtype; CHC, clathrin heavy chain; KD, knockdown; BACE1, ß-secretase 1; MPR300, cation-independent mannose 6-phosphate receptor; EE, early endosome; LE, late endosome; GAP, GTPase activating protein; ACAP1, Arf-GAP with coiled-coil ANK repeat and PH domain containing protein 1; GGA3, Golgi localized gamma ear containing ADP ribosylation factor binding protein 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; DABCO, 1,4-diazadicyclo[2,2,2]octane; PBS, phosphate buffered saline; DAPI, 4’,6diamidino-2-phenylindole
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/febs.12834 This article is protected by copyright. All rights reserved.
Accepted Article
Abstract Flotillin-1 and flotillin-2 are highly conserved, membrane microdomain associated proteins that have been shown to be involved in signal transduction, membrane trafficking and cell adhesion. Upon growth factor stimulation, flotillins are tyrosine phosphorylated and become endocytosed from the plasma membrane into endosomes from which they are recycled back to the plasma membrane. Although a role for flotillin-1 in the endocytosis of certain cargo proteins has been suggested, it is not known how the growth factor induced endocytosis of flotillins is regulated and which endocytosis pathway is used. However, this is likely to be different from the pathway used by flotillin dependent cargo. In this study, we have addressed the mechanistic details of flotillin trafficking during growth factor signaling. We here show that dynamin-2 activity is required for the uptake of flotillins from the plasma membrane upon epidermal growth factor stimulation, and inhibition of dynamin-2 GTPase activity impairs flotillin endocytosis. Surprisingly, recycling of flotillins from endosomes to the plasma membrane appears to require both dynamin-2 and clathrin. Upon overexpression of dynamin-2 mutants or depletion of clathrin heavy chain, flotillins are permanently trapped in endosomes. These data show that clathrin and dynamin are required for the endosomal sorting of flotillins, and the present study provides a mechanistic dissection of the thus far poorly characterized endosomal trafficking of flotillins.
Introduction Endocytosis serves e.g. to incorporate nutrients and receptor bound cargo into the cell and also plays a role in the regulation of signal transduction. On the other hand, the loss of membrane by endocytosis needs to be compensated by pathways of recycling that return lipids and proteins to the plasma membrane (PM). Different classes of endocytosis have been described so far (for a review, see [1, 2]). The best characterized and understood mechanism is clathrin mediated endocytosis (CME), during which the cargo is incorporated in clathrin coated pits (CCPs) that pinch off from the PM towards the cytosol and thereby form clathrin coated vesicles (CCVs). In contrast to CME, clathrin independent endocytosis (CIE) reflects several distinct pathways for cargo to enter the cells [3-5]. In many cases, small GTPases such as cdc42 or ADP-ribosylation factor 6 (ARF6) have been suggested to regulate these pathways (reviewed in [6]), but the molecular details of CIE pathways are not as well characterized as those for CME. Several cargo molecules and regulators of CIE have been characterized, and an important role for membrane rafts in some forms of CIE has become apparent. Membrane rafts are cholesterol and sphingolipid enriched microdomains that sub-compartmentalize membranes and thereby serve as signaling or endocytosis platforms [7, 8]. However, some cargo molecules, such as the epidermal growth factor receptor (EGFR), have been shown to utilize both CME and CIE pathways in a ligand dosage dependent manner [9, 10].
The three members of the dynamin GTPase family have been shown to regulate both CME and CIE (reviewed in [11, 12]). While dynamin-2 (dyn2) is ubiquitously expressed, dynamin-1 expression is restricted to the brain in organisms [13-15], although its expression in some cultured cells has also been shown [16]. Dynamin-3 in turn has been detected in the brain,
This article is protected by copyright. All rights reserved.
Accepted Article
lung and testis [17-19]. The best characterized function of dynamins lies within vesicle fission during endocytosis. However, dyn2 has been suggested to regulate Golgi export of cargo proteins such as p75 [16], suggesting that the different dynamin proteins are not fully redundant. Although CME is clearly dynamin dependent, also some of the CIE pathways have been designated as dynamin dependent, whereas others appear to be dynamin independent [6]. However, the role of dynamin in endocytic trafficking appears not to be limited to formation of endocytic carriers only, since several publications have shown that dynamin may also play a role in sorting within endosomes, i.e. in endocytic recycling of cargo molecules [20-24].
Recently, a role during CIE has been suggested for the members of the flotillin family, flotillin-1 and -2 ([25-27] and see below). The expression of flotillin-1 is highly dependent on flotillin-2 in most cell types and also in the respective knockout mice that have recently been generated [28-30]. Flotillins participate in several cellular processes, such as signal transduction, regulation of the cytoskeleton and endocytosis (reviewed in [31]). In their Nterminus, flotillins contain a prohibitin homology (PHB) domain which mediates the association with the inner leaflet of the PM and contains motifs for acylation, i.e. myristoylation and palmitoylation which together with oligomerization mediate a constitutive membrane association of flotillins [32, 33]. In line with this, flotillins form higher order oligomers, a property mediated by their C-terminal parts containing potential coiled-coils [26, 33-35].
The subcellular localization of flotillins is rather dynamic. Under growth conditions, flotillins predominantly localize to the PM and endosomal/vesicular structures [36-38]. At the PM, flotillin microdomains were described to form flat platforms or invaginations that apparently bud into the cell [26, 27]. The cellular localization of flotillins can be modulated e.g. by growth factor stimulation. Upon stimulation with epidermal growth factor (EGF), flotillins are tyrosine phosphorylated by members of the Src kinase family and translocate from the PM to endosomes [35, 39, 40]. However, it is not known which kinds of carriers are responsible for the growth factor induced flotillin endocytosis or if dynamin is required for it.
During the last years, several cargo molecules have been suggested to be internalized by flotillin mediated endocytosis [25-27, 41-48], although this has recently been disputed for some of the cargo molecules [48-50]. Furthermore, none of these cargo proteins appears to specifically use the suggested flotillin dependent CIE route. In some cases the role of flotillins in cargo trafficking is not the endocytosis per se, but a step preceding the formation of endocytic carriers [48-50]. Thus, it has still not been conclusively demonstrated that flotillins are proper endocytic proteins that mechanistically participate in cargo trafficking. Furthermore, in some context such as growth factor signaling, flotillins are more likely to act as cargo proteins due to their role in signal transduction [49, 51]. Studies on the role of flotillins in endocytic processes have also produced different conclusions on the dynamin dependency. Whereas e.g. the CD59 endocytosis was concluded to be dynamin independent [26], uptake of other cargo molecules such as proteoglycans [45] or even
This article is protected by copyright. All rights reserved.
Accepted Article
polarized uptake of glycophosphatidyl-inositol (GPI) -anchored proteins [25] have been suggested to depend on both flotillin and dynamin, suggesting cargo or cell type specific differences. Thus, the mechanisms of flotillin dependent endocytosis of cargo molecules are still not conclusively characterized. Even less is know about the endocytosis of flotillins during EGF signaling, which results in a rapid uptake of flotillins from the PM to endosomes [39]. However, these two pathways, flotillin dependent cargo trafficking vs. flotillin endocytosis as cargo appear to be fundamentally different, and the former one was not addressed in this study. Instead, we here sought to characterize the trafficking of flotillins themselves during EGF signaling. We here provide data that dyn2 is important not only for the growth factor induced endocytosis of flotillins from the PM to endosomes, but it may also be required for endosomal sorting of flotillins and possibly for their recycling back to the PM. We here also unravel a novel and unexpected role for clathrin in the cellular trafficking of flotillins.
Results We have earlier shown that EGF stimulation results in flotillin uptake from the PM into endosomes [39]. On the other hand, flotillin-1 has been described to function in a dynamin independent, raft associated endocytosis pathway that mediates the uptake of ligands such as CD59 and GPI anchored proteins [27]. However, this pathway is constitutive [26, 27], whereas the EGF induced endocytosis takes place as a result of active signaling [35, 39, 40]. The dependency of the EGF induced flotillin uptake on dynamin function has not been studied, nor is the nature of the respective carriers known. To clarify the role of dynamin in flotillin endocytosis during EGFR signaling, we used two different modes to impair dyn2 function, namely chemical inhibition and overexpression of dyn2 mutants. To chemically inhibit dynamin, we used two different chemical inhibitors of dynamin function. Dynasore, a reversible, cell permeable inhibitor of dynamin GTPase activity, prevents vesicle fission at the PM [52-54], whereas MiTMAB interferes with the membrane association of dyn2 [55]. Mutation of lysine 44 (K44A) in dyn2 affects GTP binding and GTPase activity and results in blocked endocytosis [56-58], although in some cases, dyn2 K44A expression has also been shown to have an effect on endosomal trafficking of certain cargo molecules [20, 22]. R399A exchange impairs the membrane association of dyn2 and is considered as a milder dominant negative form of dynamin [59], whereas the T65A substitution produces a protein capable of GTP binding but deficient in GTP hydrolysis [60].
Starved HeLa cells were treated with Dynasore or dimethyl sulfoxide (DMSO) as a control, stimulated with EGF and stained for flotillins (Fig. 1A and Fig. 2A). Flotillin endocytosis upon growth factor stimulation can be detected either by live imaging [39] or in fixed cells [35, 39] as appearance of flotillin positive, perinuclear vesicular structures, whereas in starved cells, flotillins almost exclusively reside in the PM. The endocytosis is scored as percentage of cells showing a clear perinuclear accumulation of flotillins (see e.g. Fig. 1A, third row from top). Our quantitative analysis showed that Dynasore efficiently blocked the EGF induced uptake of flotillins from the PM (Fig. 1B). Uptake of fluorescently labeled transferrin, which was used as a control, was also prevented (Fig. 2B, middle). We also used another dynamin inhibitor, MiTMAB which prevents the membrane binding of dynamins by interfering with the
This article is protected by copyright. All rights reserved.
Accepted Article
lipid binding domain, without affecting the GTPase activity as such [55]. This inhibitor also prevented the endocytosis of both transferrin (Fig. 2B, right) and flotillins (Fig. 2C).
Recent findings have suggested that Dynasore inhibits fluid phase endocytosis and membrane ruffling also in cells that lack all three dynamin isoforms, implicating that it may have some unspecific effects on certain forms of endocytosis [61]. Since endocytosis by means of fluid phase and uptake by ruffling both require actin [62] and Dynasore has been suggested to directly affect the actin cytoskeleton [63], it was important to exclude that the inhibition of flotillin endocytosis by Dynasore was due to such adverse effects influencing actin rather than dynamin. Thus, we used Latrunculin A (LatA), an agent that inhibits actin polymerization, and analyzed the EGF dependent endocytosis of flotillins. LatA was applied for 1 h prior to EGF stimulation on starved cells. Despite disruption of the actin cytoskeleton, flotillin-2 was endocytosed in LatA treated cells (Fig. 3A, left row), although the staining appeared slightly more dispersed as compared to the control cells (10 min EGF + EtOH). Depolymerization of the actin cytoskeleton was controlled using phalloidin staining (Fig. 3A, right row). Quantification of the data showed that the translocation of flotillin-2 in LatA treated cells was about equally efficient as in EtOH treated cells (Fig. 3B). Very similar results were obtained when another actin depolymerizing drug, cytochalasin D, was used (data not shown). Thus, these data show that the inhibitory effects of Dynasore and MiTMAB on flotillin endocytosis are not due to the adverse effects of these dynamin inhibitors on actin.
Although EGF stimulation induces endocytosis of flotillins, it is not known if EGFR kinase activity or MAP kinase activation are directly required for it. Serum starved cells were incubated with the EGFR kinase inhibitor PD 153035 or the non-inhibiting control substance AG9 for 5 min and then stimulated with EGF in the presence of these substances for 15 min (Fig. 4A, left). Immunofluorescent staining for flotillin-2 and quantification of the data (Fig. 4B) demonstrated that flotillin endocytosis was equally efficient despite EGFR kinase inhibition. Inhibition of EGFR was controlled by means of Western Blot and quantified (Fig. 4C). Similarly, inhibition of the MAP kinase kinase MEK, the activity of which is essential for the activation of the MAP kinases ERK1 and ERK2 downstream of the EGFR, did not prevent the EGF induced endocytosis of flotillins (Fig. 4A, right & Fig. 4D). Inhibition of MEK activity was controlled using phospho-specific antibodies against its phosphorylated substrates ERK1/2 and quantified (Fig. 4E). Thus, these data show that neither EGFR nor MAP kinase activity are required for flotillin endocytosis.
To gain more insight into the kinetics of flotillin endocytosis, we used costaining of flotillins with different endosomal markers after EGF stimulation (Fig. 5). Flotillin-2 was found to rapidly localize to a compartment positive for the tetraspanin CD63/LAMP3 which is a marker of multivesicular bodies that represent an intermediate to late endosomal form. This colocalization was evident after 5 min EGF and became more intense after 10 and 30 min (Fig. 5A), as shown by analysis of the Pearson correlation coefficient (Fig. 5B). However, only a very low degree of colocalization of flotillin-2 with LAMP3 was detected upon Dynasore treatment (Fig. 5A-B). Furthermore, the colocalization of flotillin-2 with the early
This article is protected by copyright. All rights reserved.
Accepted Article
endosomal marker Rab5 was very limited at all time points in both Dynasore treated and DMSO treated cells (Fig. 5C-D). Similarly, the colocalization of flotillin-2 with transferrin receptor (TfnR) was very low in EGF stimulated cells without Dynasore (Fig. 6A). However, some degree of colocalization was seen at the PM upon Dynasore treatment, consistent with the inhibition of endocytosis of both flotillin-2 and TfnR by Dynasore (Fig. 6A).
Our earlier findings have suggested that in serum-grown cells, flotillins are continuously recycled between plasma membrane and endosomes and accumulate at the plasma membrane during starvation [39]. However, the possible role of dynamin in the retrograde trafficking of flotillins back to the plasma membrane has not been studied. Thus, we hypothesized that if Dynasore is applied during starvation, we might detect an effect on the retrograde transport of flotillins and see them accumulating in endosomes. We therefore incubated cells in serum-free medium overnight in the presence of Dynasore. Flotillin-1 and flotillin-2 were localized at the plasma membrane in starved cells treated with DMSO overnight (Fig. 6B, left). Similarly, in Dynasore treated, starved cells, flotillins were present at the plasma membrane (Fig. 6B, middle). The endosomal localization of transferrin receptor was also partially perturbed by overnight Dynasore treatment (Fig. 6B, right).
To inhibit dynamin function, we used an alternative strategy to chemical inhibition. Mutation of the Lys44 residue to Ala (K44A) in dyn2 results in a mutant protein that is not capable of high affinity binding of GTP and thus shows a lower activity than the wildtype (WT) protein [56-58], whereas T65A mutant has a very low GTPase activity although it is capable of GTP binding [60]. Mutation R399A in turn impairs the membrane association of dyn2 [59]. As a result of these mutations, dynamin dependent processes such as endocytosis are inhibited. We expressed GFP-tagged WT, K44A, T65A or R399A dyn2 in HeLa cells (Fig. 7 and 8). In GFP transfected, starved cells, flotillins were to a large degree localized in the PM. However, expression of K44A resulted in accumulation of flotillins in the perinuclear region already in starved cells (Fig. 7A). Quantification of the data showed that in starved or EGF stimulated, K44A expressing cells, over 90% of the cells show a perinuclear flotillin staining (Fig. 7B). A less profound but significant accumulation was also seen in the cells expressing the WT dyn2 (Fig. 7B). After EGF stimulation, a very strong condensation of flotillins in the same regions became apparent in K44A (Fig. 7A), R399A and T65A (Fig. 8) expressing cells. However, the signals for flotillin and dyn2 did not overlap, but flotillins appeared to localize in the vicinity of the dots containing K44A dyn2, and the compartment containing flotillins in these cells frequently appeared more condensed as compared to the GFP transfected control cells. In starved, K44A dyn2 expressing cells, flotillin-2 colocalized with LAMP3 (Fig. 9), implicating that it accumulates in late endosomes. Thus, K44A, T65A and R399A dyn2 expression clearly altered the cellular localization of flotillins in a similar fashion, resulting in late endosomal accumulation.
Given that the chemical inhibition of dyn2 activity and the overexpression of the mutant forms produced different results, we set out to identify the cause of this discrepancy. Since the inhibition of dynamin GTPase activity seemed not to be the cause of the accumulation of
This article is protected by copyright. All rights reserved.
Accepted Article
flotillins in endosomes and their impaired recycling, we considered the possibility that dyn2 mutants might sequester some factors that are required for flotillin recycling. Beyond its normal function in endocytosis, clathrin has been suggested to play a role as a coat protein in endocytic recycling [64, 65]. In the case of e.g. transferrin receptor, the recycling requires not only clathrin but also dynamin [23, 24]. Thus, we tested if clathrin depletion would display an effect on the recycling of flotillins. Clathrin heavy chain (CHC) was depleted in HeLa cells using two independent siRNA sequences, resulting in an efficient knockdown of CHC (Fig. 10A, C-E). The cells were starved and labeled for flotillin-2. Surprisingly, flotillin-2 exhibited an intracellular localization also in starved cells depleted of CHC, instead of the PM localization seen in control cells (Fig. 10A). Quantification of flotillin translocation showed that over 90% of the starved CHC knockdown cells exhibited flotillin translocation, whereas far less than 20% of the control siRNA cells did so (Fig. 10D). To test this further, we used a flotillin-2-GFP fusion protein in which Tyr163 has been mutated to Phe (F2-Y163F-GFP). We have previously shown that this mutant is exclusively localized at the PM and does not translocate to endosomes upon EGF stimulation [35, 39]. Intriguingly, an endosomal localization of this mutant was detected in starved CHC-KD cells (Fig. 10B). Furthermore, a prominent colocalization of flotillin-2 with LAMP3 was observed in starved CHC-KD cells (Fig. 11A), whereas no colocalization could be observed with the early endosomal marker Rab5 (Fig. 11B). This might implicate that impairment of clathrin function prevents the translocation of flotillins from endosomes to the PM, similarly to the expression of K44A, T65A or R399A dyn2.
To provide further insight into this surprising role of clathrin in the retrograde trafficking of flotillins, we overexpressed the C-terminal part of CHC, the so-called “Hub domain” which has previously been shown to inhibit clathrin mediated endocytosis [66-68]. In serum-starved cells expressing the Hub domain, flotillin-2 and flotillin-1 were partly localized to the PM, but a clear accumulation in the perinuclear region was also seen (Fig. 12A), and flotillin-2 exhibited a clear colocalization with LAMP3 (Fig. 12B), demonstrating that CHC Hub expression results in a similarly altered localization of flotillins as CHC depletion and mutant dyn2 expression. To test if the effect of K44A overexpression might also affect the localization of CHC, we stained WT and K44A dyn2 transfected cells for CHC. As shown in Figure 13, high level overexpression of WT or K44A dyn2 resulted in their accumulation in cytoplasmic patches which also contained CHC. Thus, K44A dyn2 expression not only causes an impairment of flotillin localization but also results in sequestration of CHC in cytoplasmic aggregates, which may also impair endosomal sorting steps that involve clathrin.
Discussion Our previous data have shown that during growth factor signaling, flotillins are endocytosed but they do not affect the early endocytic trafficking of EGFR [49]. Thus, due to their function as regulators of MAP kinase signaling (reviewed in [51]), flotillins are likely to be internalized as part of a signaling complex, i.e. as cargo. However, the immediate internalization pathways of flotillins vs. EGFR appear to be different, since no colocalization between flotillins and EGFR can be observed after short times, whereas a partial colocalization is
This article is protected by copyright. All rights reserved.
Accepted Article
observed after 15 min EGF stimulation [39], implicating that the originally separate endocytosis routes merge in late endosomes.
Flotillins have been suggested to define a non-clathrin, non-caveolar endocytosis pathway [27]. The GPI-anchored protein CD59 and the sphingolipid GM1 were shown to be partially endocytosed through this pathway. Furthermore, based on the additive effect of flotillin-1 depletion and K44A dyn2 expression on the uptake of cholera toxin, the ligand of GM1, it was concluded that the flotillin dependent endocytosis of these cargo molecules was dynamin independent. However, the direct influence of dynamin on the growth factor induced endocytosis of flotillins has not been studied so far, although these two pathways are likely to be separate. The idea that in some specific context, such as growth factor receptor activation, flotillins may not be part of the endocytic machinery, but rather represent cargo molecules has been neglected by most studies. Using two different chemical inhibitors of the GTPase dynamin, we here show that dyn2 is indeed involved in the EGF induced endocytosis of flotillins, since inhibition of dyn2 GTPase activity (Dynasore) or impairment of its membrane localization (MiTMAB) in starved cells prevented the uptake of flotillins from the PM upon EGF stimulation. Importantly, we here demonstrated that these effects are not due to the described adverse effects of dynamin inhibitors on actin cytoskeleton and endocytosis pathways that require actin [61, 62]. Although most of the endocytosis pathways described are actin dependent, some CIE pathways appear not to depend on actin [6, 69]. Since flotillins are endocytosed by means of a CIE pathway whose exact identity is not known, it seems in the light of our data that this pathway is also actin independent.
In serum grown HeLa cells, flotillins localize to both PM and LAMP3 positive endosomes, whereas upon growth factor deprivation, flotillins translocate to the PM [35, 39]. This implies that under growth conditions, flotillins continuously recycle between endosomes and PM. Recycling of flotillin containing vesicles has previously been shown [70], and very recent findings have even suggested that flotillin-2 might be involved in the recycling of E-cadherin, transferrin receptor [71] and the β-secretase BACE1 [72]. Our data here suggest that recycling of flotillins themselves may involve dynamin and clathrin.
Expression of K44A, T65A or R399A dyn2 and chemical dynamin inhibition appeared at first glance to produce somewhat contradictory effects on flotillin trafficking. Whereas Dynasore and MiTMAB clearly displayed an effect on flotillin uptake, K44A, T65A and R399A dyn2 appeared to trap flotillins in endosomal structures, which we never observed upon chemical inhibition of dynamin. Therefore, mutant dyn2 expression most likely specifically blocks flotillin recycling and translocation to the PM and thus prevents the normal accumulation of flotillins at the PM during starvation. Although the main function of dynamins may well be in endocytosis, some previous publications have already suggested that dynamin plays a role in endosomal recycling. For example, dynamin has previously been shown to play a role in the endocytic recycling of the Tfn receptor [23]. Likewise, it has been shown that expression of K44A dyn2 results in trapping of other recycled cargo molecules such as the cationindependent mannose 6-phosphate receptor (MPR300) [22] or ricin toxin [20] within
This article is protected by copyright. All rights reserved.
Accepted Article
endosomes. The findings of Llorente et al. [20] are especially intriguing in the light of a recent publication demonstrating that the transport of ricin and Shiga toxin from endosomes to Golgi was significantly reduced upon flotillin knockdown [73]. Thus, our present data suggest that the effect of K44A dyn2 on ricin toxin, as observed by Llorente et al. [20], may actually be indirect and caused by trapping of flotillins in endosomes, which in turn impairs ricin trafficking.
Interestingly, we found that clathrin heavy chain was accumulated in clusters in K44A dyn2 positive aggregates, suggesting that impairment of clathrin function might cause the aberrant localization of flotillins. In line with the hypothesis that K44A dyn2 overexpression results in clathrin sequestration and prevents the endosomal recycling of flotillins, knockdown of CHC and expression of the dominant negative CHC Hub fragment [66, 74] also resulted in endosomal accumulation of flotillins. Both CHC depletion and Hub expression have been suggested to result in altered localization and perinuclear accumulation of early endosomes (EE) and late endosomes (LE) [74, 75]. Thus, the collapse of EE into the perinuclear region might theoretically prevent the transfer of cargo, including flotillins, from EE to LE. On the other hand, overexpression of CHC Hub was shown to prevent the uptake of Tfn into cells, whereas receptor recycling appeared not to be impaired, despite the effects on endosome localization [74], speaking against this possibility. Furthermore, in CHC depleted cells, flotillins clearly colocalized with the multivesicular body marker LAMP3. Taken together, although flotillins are known as lipid microdomain associated proteins, their recycling appears to be dependent on clathrin. Furthermore, the endocytosis incompatible flotillin-2 Y163F mutant [35, 39] which is normally exclusively localized at the PM but not in endosomes was also found to accumulate in endosomes upon CHC knockdown, suggesting that also the biosynthetic transport of flotillin-2 may proceed through endosomes in a clathrin dependent fashion. This would be consistent with the findings showing that a considerable fraction of biosynthetic cargo passes the recycling endosomes before appearing at the PM [76].
Clathrin has been shown to play a role not only in endocytosis but also in endosomal sorting and recycling of proteins from endosomes to the PM [23, 64, 77-79]. It has been demonstrated that clathrin coats are present in endosomes in which they even appear to exhibit two different functions [80]. One class of endosomal clathrin coats mediates the sorting of ubiquitinated cargo towards degradation [77-79], whereas the other one is involved in recycling of proteins towards the PM [23]. In fact, clathrin and dynamin have even been suggested to collaborate in the endosomal recycling of some cargo proteins [23, 24], which may also take place during flotillin sorting and recycling. It has been shown that a specific coat containing clathrin and Arf-GAP with coiled-coil ANK repeat and PH domain containing protein 1 (ACAP1), a GTPase activating protein (GAP) for ARF6, is involved in the stimulus induced recycling of integrins and the glucose transporter Glut4 [81-83]. Clathrin also cooperates in endosome to PM recycling with the Golgi localizing gamma-ear containing ADP ribosylation factor binding protein 3 (GGA3) during the sorting of BACE1 and hepatocyte growth factor receptor (also known as MET) [84-87]. Our recent data point to a role for flotillins in the endosomal sorting of BACE1, as flotillin depletion impairs BACE1 recycling to PM [72].
This article is protected by copyright. All rights reserved.
Accepted Article
Our present data are of special interest since there are findings suggesting that also clathrin mediated endocytosis and flotillin function may be connected [43, 48, 50]. Flotillins have been suggested to be necessary for the clathrin mediated endocytosis of the Alzheimer amyloid precursor protein [48] and the dopamine transporter DAT [43]. However, flotillins appear not to be directly involved in the endocytosis of these proteins, but rather mediate their clustering in specific domains at the PM, which then cooperate with the CME machinery. Similarly, flotillin-1 appears to mediate the clustering of EGFR upon EGF stimulation [49]. It has been postulated that flotillins might serve as platforms that preassemble clathrin dependent cargo which is then handed over to CCPs for CME [88]. However, this possibility has not been studied in detail and is thus mechanistically not well understood. Our present data suggest that clathrin and flotillin functions may indeed be connected.
We have here provided important novel insight into cellular trafficking of flotillins which are associated with lipid microdomains by means of fatty acid modifications. Dynamin is apparently important for the growth factor induced endocytosis of flotillins. However, the most important findings of the present study revealed an unexpected role for clathrin in the endocytic recycling and possibly also biosynthetic trafficking of flotillins. These findings, together with those suggesting a collaboration of flotillin microdomains and CME components at the PM, imply that the division of trafficking pathways into clathrin dependent and raft dependent ones may not be justified, and that these pathways may intermingle and their components interact in an unexpected manner. It will thus be of special interest to characterize the exact role of flotillins in these trafficking processes, keeping in mind the still remaining question if flotillins are mechanistically involved in endocytosis/trafficking or merely cargo proteins.
Materials and Methods Antibodies and constructs Mouse monoclonal antibodies against clathrin heavy chain and flotillin-1 were purchased from Transduction Labs, and the rabbit polyclonal antibodies against flotillin-1 and -2 were from Sigma-Aldrich (Taufkirchen, Germany). Mouse monoclonal antibodies against LAMP3/CD63 and pERK1/2 as well as a polyclonal antibody against total ERK 1/2 (H-72) and Rab5 were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). An antibody directed towards the transferrin receptor was purchased from Life Technologies (Darmstadt, Germany) and an antibody towards glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from Abcam. Mouse monoclonal anti-myc and anti-phospho tyrosine (pY100) and rabbit polyclonal anti-EGFR antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). Secondary antibodies for immunofluorescence coupled to Cy3 and Cy5 were purchased from Jackson Immunoresearch (Newmarket, UK) and secondary antibodies coupled to Alexa488 were from Life Technologies. HRP-coupled goat anti-mouse and antirabbit secondary antibodies for detection of Western blots were purchased from Dako (Glostrup, Denmark).
This article is protected by copyright. All rights reserved.
Accepted Article
Rat dynamin-2-GFP fusion constructs (WT and K44A) were kindly provided by M. McNiven (Mayo College of Medicine, Rochester, MN, USA). The R399A and T65A dyn2 mutant were generated by site directed mutagenesis of the WT construct. The flotillin-2-Y163F-EGFP mutant and its transfection have been described earlier [39]. The dominant-negative clathrin heavy chain Hub region [74] was cloned in pCMV-Tag5A vector which contains a C-terminal myc-tag (Stratagene/Agilent Technologies, Waldbronn, Germany).
Cell culture and siRNA knockdown Human cervix adenocarcinoma cells (HeLa) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Life Technologies) with high glucose, supplemented with 10% fetal calf serum (Life Technologies), 100 units/ml penicillin and 100 µg/ml streptomycin (Life Technologies) at 37°C and 8% CO2. Transient depletion of clathrin heavy chain by siRNAmediated knockdown was performed as described earlier for flotillins [35]. The siRNA oligonucleotide duplexes as well as a negative control with medium GC content that does not target any human sequences were obtained from Life Technologies (StealthTM siRNA system). The experiments were performed 72 hours post-transfection.
Immunofluorescence HeLa cells grown on coverslips were treated with growth factors and inhibitors as indicated and then fixed in methanol for 10 min at -20°C. After blocking unspecific protein binding sites with 1% bovine serum albumin (BSA) in phosphate buffered saline (PBS), the cells were labeled with primary antibodies and Alexa-488 (Life Technologies), Cy5 or Cy3 conjugated (Jackson Immunoresearch) secondary antibodies. The incubation with secondary antibodies was combined with the nuclei counterstaining with 4',6-diamidino-2-phenylindole (DAPI). The specimen were mounted in Gelmount supplemented with 1,4-diazabicyclo[2.2.2]octane (DABCO, Sigma-Aldrich) and examined using a confocal laser scanning microscope (Zeiss LSM 710 combined with Axio Observer, Carl Zeiss, Jena, Germany).
Treatment with growth factors and inhibitors For growth factor treatments, HeLa cells were serum-starved for at least 16 hours and then stimulated with 100 ng/ml epidermal growth factor (EGF, Sigma-Aldrich) for the indicated time points.
For EGFR tyrosine kinase inhibitor treatments, the cells were pretreated with the inhibitor PD153035 (20 nM, Calbiochem), a non-inhibiting control AG9 (20 nM, Calbiochem) for 5 min before EGF was applied. To inhibit MEK, the cells were pretreated with the specific MEK inhibitor U0126 (10 µM, Cell Signaling) or DMSO for 120 min and then stimulated with 100 ng/ml EGF. In general, EGF stimulation was performed in the presence of inhibitors and their non-inhibiting analogs as negative control.
This article is protected by copyright. All rights reserved.
Accepted Article
To inhibit the GTPase activity of dynamin, the cells were pretreated with 80 µM Dynasore (Sigma-Aldrich) or DMSO as a negative control for 2 hours and then stimulated with EGF. Another dynamin inhibitor MiTMAB (Abcam, Cambridge, UK) interferes with the PH-domain of dynamins and inhibits dynamin mediated fission [55]. Cells were pretreated with 30 µM MiTMAB for 30 min at 37°C and then stimulated with 100 ng/ml EGF in the presence of the inhibitor.
For the uptake of transferrin (Tfn), the cells were chilled on ice, incubated with AlexaFluor546 conjugated Tfn (1:500) (Molecular Probes/Life Technologies) for 10 min on ice, washed with PBS and then transferred to pre-warmed serum free medium containing negative controls or dynamin inhibitors. Cells were allowed to internalize Tfn for 10 min. Surface bound Tfn was stripped in stripping buffer (300 mM NaCl, 200 mM acetic acid, pH 2.9) before fixation to facilitate the detection of intracellular Tfn. Cells were fixed in 4% paraformaldehyde for 10 min (80 mM PIPES pH 6.8, 2 mM MgCl2, 5 mM EGTA pH 8.0, 4% PFA) at room temperature before mounting in Gelmount (Biomeda, Foster City, CA, USA) supplemented with 50 mg/ml DABCO.
To check the influence of the actin cytoskeleton on flotillin translocation the actin cytoskeleton was depolymerized with 1 µM LatA for 1 hour at 37°C prior to stimulation with 100 ng/ml EGF for 10 min. Thereafter the cells were either fixed in methanol for a subsequent immunofluorescent staining of flotillins or fixed in 4% PFA, permeabilized with 50 µg/ml digitionin, labeled with Alexa488-coupled phalloidin (Molecular Probes/Life Technologies) and counterstained with DAPI. Cells were embedded in Gelmount supplemented with DABCO.
Cell lysis and Western blot Cells were washed in PBS, scraped in lysis buffer (50 mM Tris HCl pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% NP-40) supplemented with protease inhibitor cocktail (Sigma-Aldrich) and 1 mM sodium fluoride and cleared by centrifugation. Total protein concentration was estimated from the cleared lysates with the Bio-Rad protein assay reagent (Bio-Rad, Munich, Germany). Equal protein amounts were analyzed by SDS-PAGE and Western blot.
Statistical analysis Unless otherwise stated, all experiments were performed at least three times independent from each other. Densitometric analysis of Western blots was performed with Quantitiy One Software (Bio-Rad). For quantification of flotillin translocation after immunofluorescent staining, at least 50 cells per condition were counted and the percentage of cells showing translocation from the PM to intracellular membranes was assessed and used for data analysis. For the analysis of colocalization, the Pearson’s correlation coefficient (Rr) was calculated from a defined region of interest (ROI) for at least 15 cells per condition using the JACoP plug-in for ImageJ (NIH, Bethesda, MD, USA). Statistical comparisons were
This article is protected by copyright. All rights reserved.
Accepted Article
performed with Student’s t-test (unpaired, two-tailed) or analysis of variance (one- or twoway ANOVA), where appropriate, using Graph Pad Prism 5 Software (GraphPad Software, Inc., La Jolla, CA, USA). Data are shown as mean ± SD. Values of p
