Essential role of flotillin-1 palmitoylation in the intracellular localization and signaling function of IGF-1 receptor

Flotillin-1 (also known as reggie-2) and flotillin-2 (also known as reggie-1) are membrane-associated oligomeric proteins that are ubiquitously expressed (Lang et al., 1998; Neumann-Giesen et al., 2004; Solis et al., 2007; Stuermer et al., 2001). The flotillins form homo- and hetero-oligomers (Neumann-Giesen et al., 2004; Solis et al., 2007), which assemble into clusters defining specialized flat membrane microdomains at the plasma membrane independently of caveolins and caveolae (Fernow et al., 2007; Frick et al., 2007; Lang et al., 1998; Stuermer et al., 2001). Flotillins have been shown to be involved in various cellular processes including cell proliferation, T-cell activation and phagocytosis (Dermine et al., 2001; Gómez et al., 2010; Santamaría et al., 2005; Stuermer et al., 2004), and have been implicated in signaling activation of IgE and TrkA, and epidermal growth factor, insulin and G-protein-coupled receptors (Amaddii et al., 2012; Baumann et al., 2000; Kato et al., 2006; Limpert et al., 2007; Wehmeyer et al., 2014).

Insulin-like growth factor-1 receptor (IGF-1R) tyrosine kinase activation leads to downstream signaling activation of the mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K)–Akt pathways, which promote cell proliferation and differentiation and prevents apoptosis (Adams et al., 2000). IGF-1R localized in the noncaveolar microdomain has been reported to regulate IGF-1R signaling (Hong et al., 2004). However, it is not clear how the noncaveolar microdomain regulates IGF-1R signaling and which constituents and/or protein components of the noncaveolar microdomain are involved. Although flotillin-2 has been reported to regulate IGF-1-induced neutrite outgrowth in N2a neuroblastoma cells (Langhorst et al., 2008; Munderloh et al., 2009), whether flotillin-1 is required for IGF-1R signaling activation has not been investigated to date.

Flotillin-1 is palmitoylated at Cys34, and flotillin-2 is myristoylated at Gly2 and palmitoylated at Cys4, Cys19 and Cys20 (Liu et al., 2005; Morrow et al., 2002; Neumann-Giesen et al., 2004). Flotillin-1 palmitoylation is known to be required for its plasma membrane targeting (Liu et al., 2005; Morrow et al., 2002) and for protein kinase C (PKC)-induced endocytosis of dopamine transporter (Cremona et al., 2011). Recently, it has been shown that flotillin-1 and/or flotillin-2 can mediate receptor endocytosis, recycling and lysosomal targeting (Cremona et al., 2011; Pust et al., 2013; Solis et al., 2013). However, the regulatory role of flotillin palmitoylation in the intracellular trafficking and activation of IGF-1R is unknown.

Here, we report that palmitoylation of flotillin-1 in the endoplasmic reticulum (ER) is indispensable for the ER exit and the plasma membrane localization of flotillin-1 and IGF-1R. In addition, IGF-1 induced depalmitoylation and repalmitoylation of flotillin-1, thereby preserving its palmitoylation status in the plasma membrane and prolonging IGF-1R activation to promote cancer cell proliferation.

To explore the regulatory role of flotillin-1 in IGF-1R signaling, independently of caveolin-1, caveolin-2 and caveolae, HEK293T cells that expressed no caveolins (Kwon et al., 2011; Torres et al., 2006; Vial and Evans, 2005) were stably transfected with short-hairpin RNA (shRNA) against flotillin-1 and tested for IGF-1-induced signaling activation (Fig. 1). The depletion of flotillin-1 led to a reduction in flotillin-2 expression with no effect on its mRNA level (Fig. 1A). Treatment with MG132, a proteasome inhibitor, prevented flotillin-2 degradation (Fig. 1Ba) and caused accumulation of ubiquitylated flotillin-2 in flotillin-1-deficient cells expressing HA–ubiquitin (Fig. 1Bb), indicating that flotillin-1 downregulation results in ubiquitylation and subsequent proteasomal degradation of flotillin-2. A time course analysis of IGF-1-induced signaling activation in flotillin-1-deficient cells showed that the activation of IGF-1R, extracellular signal-regulated kinase 1/2 (ERK1/2) and Akt observed in control cells was abrogated with the IGF-1R level unaffected (Fig. 1C). Stable expression of IGF-1R was maintained in control and flotillin-1-deficient cells during a 6-h incubation with IGF-1 in the presence of cycloheximide (CHX), an inhibitor of protein biosynthesis (Fig. 1D), showing that flotillin-1 depletion has no effect on IGF-1R degradation. IGF-1R constitutively associated with flotillin-1 and flotillin-2 and the flotillin-associated IGF-1R was tyrosine-phosphorylated in response to IGF-1 (Fig. 1E). This activation was not detected and flotillin-2 association with IGF-1R was attenuated in flotillin-1-deficient cells (Fig. 1E). ERK1/2-mediated transcriptional activation of Elk-1 and cell proliferation in response to IGF-1 were inhibited in flotillin-1-deficient cells (Fig. 1F,G). These results show that expression of flotillin-1 and/or flotillin-2 is required for IGF-1-induced IGF-1R signaling activation in HEK293T cells deficient in caveolins.

A requirement for flotillin-1 palmitoylation in IGF-1R signaling activation was tested by using a palmitoylation-deficient flotillin-1-C34A mutant. First, we tested that re-expression of wild-type (WT) flotillin-1 but not the flotillin-1-C34A mutant ameliorated the retarded IGF-1R tyrosine kinase activation in flotillin-1-deficient cells (Fig. 2Aa, lanes 4,6,8). In addition, we found that the attenuated downstream activation of ERK1/2 and Akt was rescued by WT flotillin-1 but not the flotillin-1-C34A mutant (Fig. 2Ab, lanes 4,6,8). The constitutive association of flotillin-2 with IGF-1R and stable expression of flotillin-2 depended on flotillin-1 palmitoylation (Fig. 2Aa,b, lanes 3–8). Likewise, the abrogated IGF-1-induced Elk-1 transactivation and cell proliferation in flotillin-1-deficient cells were restored by WT flotillin-1 but not the flotillin-1-C34A mutant (Fig. 2Ba,b, lanes 5–8). These results show that flotillin-1 palmitoylation at Cys34 is required for flotillin-2 stability and the activation of IGF-1R signaling for cell proliferation.

When the effect of flotillin-1 palmitoylation on the intracellular localization of IGF-1R was examined by confocal microscopy in HEK293T cells expressing IGF-1R and flotillins, we found that flotillin-1 colocalized with IGF-1R in the plasma membrane (Fig. 3Aa). However, IGF-1R failed to localize in the plasma membrane in flotillin-1-deficient cells (Fig. 3Ab). To test whether this failure is due to a defect in its ER exit, colocalization of IGF-1R with CFP–ER, an ER marker, was analyzed. IGF-1R colocalized with CFP–ER in flotillin-1-deficient cells (Fig. 3Ac). Re-expression of WT flotillin-1 led to translocation of IGF-1R along with WT flotillin-1 to the plasma membrane (Fig. 3Ad) but the re-expressed flotillin-1-C34A mutant colocalized with IGF-1R and CFP–ER, and had no effect (Fig. 3Ae,f), indicating that the flotillin-1 palmitoylation defect leads to accumulation of IGF-1R and flotillin-1 in the ER and inhibits their translocation to the plasma membrane. The requirement of flotillin-1 palmitoylation for the plasma membrane localization of IGF-1R and flotillin-1 was further confirmed by analyzing purified plasma membrane fractions from WT-flotillin-1- and flotillin-1-C34A-expressing HEK293T cells (Fig. 3B). We found that ectopic WT flotillin-1 co-fractionated with the endogenous IGF-1R, flotillin-1 and flotillin-2 in the plasma membrane fraction, but that the IGF-1R plasma membrane localization was diminished by 40.1% in cells expressing the flotillin-1-C34A mutant despite there being no change in the total IGF-1R protein level in whole-cell lysates (Fig. 3B). Consistent with this, the flotillin-1-C34A mutant was unable to translocate to the plasma membrane, but substantial plasma membrane localization of IGF-1R and flotillin-2 was detected, reflecting that endogenous palmitoylated flotillin-1 mediated their plasma membrane localization. The results show that flotillin-1 palmitoylation is required for the ER exit of IGF-1R and localization of flotillin-1 and the IGF-1R to the plasma membrane.

Given that flotillin-1 expression and its palmitoylation defect affected flotillin-2 stability (Fig. 1; Fig. 2A), we tested whether the flotillin-2 stability in the plasma membrane was important for the plasma membrane targeting of IGF-1R mediated by flotillin-1 palmitoylation. Flotillin-1-deficient cells expressing WT flotillin-1 or the flotillin-1-C34A were treated with MG132 or dynasore, an inhibitor of dynamin-dependent endocytosis, to block proteasomal degradation of flotillin-2 by ubiquitin-mediated endocytosis and were then subjected to plasma membrane fractionation. As shown in Fig. 2A, the level of flotillin-2 was significantly reduced in the whole-cell lysates and the plasma membrane fraction of flotillin-1-deficient cells expressing the flotillin-1-C34A as compared to the cells expressing WT flotillin-1 (Fig. 4A, lanes 1,4 and 7,10). This reduction was prevented by MG132 or dynasore treatment (Fig. 4A, lanes 4–6 and 10–12) and flotillin-1-C34A and IGF-1R were not detected in the plasma membrane (Fig. 4A, lanes 7–9 versus 10–12). These results show that flotillin-1 palmitoylation mediates plasma membrane trafficking of IGF-1R independently from the flotillin-2 stability in the plasma membrane.

We then investigated whether plasma membrane targeting of flotillin-2 depends on flotillin-1 palmitoylation (Fig. 4B). Confocal microscopy analysis showed the colocalization of endogenous flotillin-1 with flotillin-2 in the plasma membrane (Fig. 4Ba) observed in control cells was not detected in flotillin-1-deficient cells (Fig. 4Bb). Upon re-expression of WT flotillin-1 but not flotillin-1-C34A, flotillin-1-deficient cells exhibited the plasma membrane localization of flotillin-2 and it colocalized with WT flotillin-1 (Fig. 4Bc versus Bd). Consistent with this, dynasore treatment, which blocks endocytosis, led to the plasma membrane accumulation of flotillin-2 in flotillin-1-deficient cells expressing flotillin-1-C34A, and flotillin-1-C34A and IGF-1R failed to localize to the plasma membrane (Fig. 4Be,f). Blockage of transferrin receptor endocytosis by the dynasore treatment is shown in Fig. 4Bg. These results show that anterograde trafficking of flotillin-2 to the plasma membrane is independent of flotillin-1 palmitoylation, but that the plasma membrane targeting of palmitoylated flotillin-1 from the ER is required for the stable maintenance of flotillin-2 in the plasma membrane.

To explore dynamics of flotillin-1 palmitoylation in the plasma membrane, the effect of treatment with 2-bromopalmitate (2-BP), a non-metabolizable analog of palmitate, was analyzed in CHX-treated HEK293T cells stimulated with IGF-1. The irreversible inhibition of flotillin-1 palmitoylation by 2-BP started after 10 min exposure and was complete by 180 min of exposure (Fig. 5A), showing that IGF-1-induced depalmitoylation of the preexisting palmitoylated flotillin-1 and that repalmitoylation of the depalmitoylated flotillin-1 occurred in the plasma membrane within this time frame. Degradation of flotillin-2 started 30 min after the beginning of the treatment with 2-BP and IGF-1 (Fig. 5A). Inhibition of the tyrosine kinase activation of IGF-1R started after 10 min of exposure and was complete after 180 min exposure (Fig. 5B), which coincided with the 2-BP-mediated inhibition of flotillin-1 palmitoylation. We then tested whether IGF-1 induces palmitoylation of newly synthesized flotillin-1 in cells treated with CHX and 2-BP, to inhibit de novo synthesis and palmitoylation of flotillin-1, and chased in serum-free medium or IGF-1-containing medium for various periods of time (Fig. 5C). Palmitoylation of preexisting endogenous flotillin-1 was completely blocked after the pretreatment with CHX and 2-BP (at 0 time chase) and palmitoylation of newly synthesized flotillin-1 started as early as 10 min after the chase regardless of the presence of IGF-1, indicating that palmitoylation of newly synthesized flotillin-1 is independent of IGF-1. Taken together, the results suggest that the preservation of flotillin-1 palmitoylation in the plasma membrane controlled by the IGF-1-stimulated rapid reversible palmitoylation turnover is required for prolonged activation of IGF-1R.

When we checked the effect of flotillin-1 palmitoylation turnover on flotillin-1 phosphorylation status in response to IGF-1, we found that tyrosine-phosphorylated flotillin-1 in basal conditions (serum starvation; 0 time) became dephosphorylated after 30 min incubation with IGF-1 (Fig. 5Da). However, inhibition of flotillin-1 palmitoylation turnover by 2-BP treatment had no effect on IGF-1-induced dephosphorylation of flotillin-1. Phosphorylation of flotillin-1 at Tyr160 by the Fyn kinase is required for its endocytosis from the plasma membrane (Riento et al., 2009). Hence, we tested the requirement for Tyr160 phosphorylation of flotillin-1 during IGF-1R signaling activation by using the phosphomimetic flotillin-1-Y160E and phosphorylation-deficient flotillin-1-Y160F mutants. We found that ectopic expression of both of flotillin-1-Y160E and flotillin-1-Y160F rescued the retarded activation of IGF-1R tyrosine kinase, ERK1/2 and Akt in flotillin-1-deficient cells, as did WT flotillin-1, in response to IGF-1 (Fig. 5Db). These results show that it is not the change in flotillin-1 tyrosine phosphorylation status but the turnover of flotillin-1 palmitoylation in the plasma membrane that is responsible for the regulation of IGF-1R activation in response to IGF-1.

To define the regulatory role of flotillin-1 palmitoylation, independently of caveolin-1, caveolin-2 and caveolae, on IGF-1-induced cancer cell proliferation, HeLa cells expressing flotillins and caveolins were chosen to test IGF-1R signaling activation. As in flotillin-1-deficient HEK293T cells (Fig. 1), knockdown of flotillin-1 caused retardation in IGF-1-induced IGF-1R, ERK1/2, and Akt activation and reduced flotillin-2 expression but did not affect the expression of caveolins in HeLa cells (Fig. 6A). To test the effect of the flotillin-1 palmitoylation defect on IGF-1R signaling activation, we measured the degree of association between IGF-1-activated IGF-1R tyrosine kinase and flotillins, and found that this was increased in HeLa cells expressing WT flotillin-1 as compared to those expressing vector control, as was downstream activation of ERK1/2 and Akt (Fig. 6Ba,b, lanes 2 versus 4). Neither the association of IGF-1-activated of IGF-1R and flotillins nor the activation of ERK1/2 and Akt in response to IGF-1 was detected in HeLa cells expressing the flotillin-1-C34A mutant (Fig. 6Ba,b, lanes 4,6). Ectopic WT flotillin-1, but not the flotillin-1-C34A mutant, promoted IGF-1-induced HeLa cell proliferation (Fig. 6C, lanes 4–6). Caveolins did not interact with IGF-1R (Fig. 6Ba). We then tested the effect of IGF-1-induced flotillin-1 palmitoylation turnover on IGF-1R tyrosine kinase activation in HeLa cells. Like the results from HEK293T cells (Fig. 5A,B), the inhibition of flotillin-1 palmitoylation turnover by 2-BP treatment abrogated the activation of IGF-1R but had no effect on the stability of flotillin-2 in HeLa cells (Fig. 6Da,b), indicating that the flotillin-1 palmitoylation turnover, but not the change in flotillin-2 stability, is the regulatory factor responsible for activation of IGF-1R. In contrast to the results from HEK293T cells (Fig. 5Da), IGF-1-induced dephosphorylation of flotillin-1 was not observed, and the flotillin-1 palmitoylation turnover had no effect on the flotillin-1 phosphorylation status in HeLa cells (Fig. 6Db). Taken together, these results show that flotillin-1 palmitoylation and intracellular trafficking is important for supplying the plasma membrane with the mature IGF-1R and that this process accounts for an increased abundance of the IGF-1R at the plasma membrane and a boost in IGF-1R signaling. We also showed that IGF-1-induced flotillin-1 palmitoylation turnover in the plasma membrane facilitated prolonged activation of IGF-1R, independently of caveolin-1, caveolin-2 and caveolae, to stimulate cell proliferation in more than one cell type.

Finally, we analyzed the oligomeric status of flotillin-1 and flotillin-2 in the plasma membrane, the chemical stability of flotillin-enriched plasma membrane microdomains and IGF-1R activation in the microdomain in response to IGF-1. We did this in purified plasma membrane from HEK293T cells expressing no caveolins, and found that flotillin-1 and flotillin-2 formed large hetero-oligomeric complexes (supplementary material Fig. S1A, high molecular mass fractions 9–12). Methyl-β-cyclodextrin (MβCD) treatment, to deplete cholesterol from the plasma membrane, did not affect the hetero-oligomeric status (supplementary material Fig. S1A). We found that tyrosine kinase activation of IGF-1R took place in detergent-resistant membrane domains that were enriched in flotillin-1 and flotillin-2 (fractions 4–6) in response to IGF-1 (supplementary material Fig. S1B). Four-step sucrose density gradient fractionation of the purified plasma membrane showed that both flotillin-1 and flotillin-2 localized in the heavy low-density insoluble membrane (HLDM) (fractions 9–12), as opposed to the light low-density insoluble membrane (LLDM), which is enriched in the ganglioside GM1 (supplementary material Fig. S1Ca). Cholesterol was enriched in the HLDM and could be depleted by MβCD treatment (supplementary material Fig. S1Cb). The HLDM localization of flotillin-1 and flotillin-2 was unaffected by the cholesterol depletion. It has been reported that flotillin microdomains are formed by oligomerization of the flotillins rather than by clustering mediated by binding to the cortical actin cytoskeleton (Langhorst et al., 2007). In agreement, we found that disrupting the actin cytoskeleton by treatment with latrunculin B, an actin polymerization inhibitor, had no effect on the HLDM localization of flotillin-1 and flotillin-2 (supplementary material Fig. S1Ca). IGF-1R localized in flotillin-1- and flotillin-2-enriched HLDM where IGF-1R tyrosine kinase activation occurred upon IGF-1 stimulation (supplementary material Fig. S1D). Thus, the data show that IGF-1R activation is initiated in the plasma membrane microdomain where hetero-oligomeric flotillin-1 and flotillin-2 was enriched and that this membrane domain is refractory to cholesterol depletion and actin cytoskeleton disruption, which is not the case for cholesterol-sensitive membrane microdomains, including caveolae and GM1-enriched plasma membrane microdomain.

Flotillin-1 palmitoylation was known to be important for its plasma membrane targeting (Liu et al., 2005; Morrow et al., 2002). The present data are in agreement with these reports and advance the previous findings to show that loss of flotillin-1 expression and a defect in flotillin-1 palmitoylation in the ER inhibit the ER exit and plasma membrane localization of IGF-1R, with the result that IGF-1R is not activated in response to IGF-1. More importantly our data show the regulatory function of flotillin-1 palmitoylation turnover in the plasma membrane – a dynamic IGF-1-induced process of depalmitoylation and repalmitoylation of the plasma membrane-targeted palmitoylated flotillin-1 – was required for IGF-1R activation. The preservation of flotillin-1 palmitoylation turnover also facilitated prolonged activation of IGF-1R in response to IGF-1.

Given that the C34A mutant of flotillin-1 is retained in the ER, our data indicate that palmitoylation of newly synthesized flotillin-1 in the ER is catalyzed by ER-resident palmitoyl acyltransferases. Having shown that plasma-membrane-targeted palmitoylated flotillin-1 undergoes depalmitoylation and repalmitoylation in response to IGF-1, our data suggest that the turnover of flotillin-1 palmitoylation is processed by IGF-1-activated palmitoyl thioesterases and palmitoyl acyltransferases in the plasma membrane. Further studies are needed to identify the responsible enzymes and address the mechanisms controlling palmitoylation and depalmitoylation of flotillin-1.

Flotillins were traditionally defined as caveolae-associated membrane proteins (Bickel et al., 1997). However, flotillins have now been shown to generate membrane microdomains that are morphologically similar to caveolae but distinct from caveolae containing caveolin-1 (Fernow et al., 2007; Frick et al., 2007; Lang et al., 1998; Stuermer et al., 2001). Our findings show that upon IGF-1 stimulation, IGF-1R tyrosine kinase activation occurs in the hetero-oligomeric flotillin-1- and flotillin-2-enriched noncaveolar microdomain in the plasma membrane. The present data show that both flotillin-1 and flotillin-2 interact with IGF-1R, and that flotillin-1 depletion leads to flotillin-2 endocytosis and degradation through a ubiquitin–proteasome pathway, causing inhibition of IGF-1R signaling activation in HEK293T cells. Flotillin-2 depletion is known to induce proteasomal degradation of flotillin-1 (Solis et al., 2007), and IGF-1-mediated cell proliferation is also reduced in flotillin-2-depleted N2a cells (Munderloh et al., 2009). Collectively the reports and our results indicate that a reciprocal dependency between flotillin-1 and flotillin-2 is required for the formation of hetero-oligomeric flotillin-1- and flotillin-2-enriched microdomains and the recruitment of IGF-1R to this noncaveolar microdomain in the plasma membrane to initiate IGF-1R activation in response to IGF-1. However, our results show that flotillin-2 in the plasma membrane of flotillin-1-deficient cells after the prevention of its proteasomal degradation neither co-fractionated nor colocalized with IGF-1R because IGF-1R is captured in the ER owing to the lack of flotillin-1-palmitoylation-dependent plasma membrane targeting. Thus, our results indicate that palmitoylated flotillin-1 is a major regulatory component of the noncaveolar microdomain in the plasma membrane, which is needed to facilitate the plasma membrane microdomain targeting of IGF-1R for initiation of IGF-1-induced signal transduction.

Flotillin-1 palmitoylation has been shown to be involved in the PKC-triggered endocytosis and membrane microdomain localization of dopamine transporter (Cremona et al., 2011). Phosphorylation of flotillin-1 at Tyr160 by Fyn kinase is important for its endocytosis from plasma membrane (Riento et al., 2009). Tikkanen and coworkers, however, have shown that hetero-oligomerization of flotillin-1 and flotillin-2, but not tyrosine phosphorylation of flotillin-2 mediated by Src kinase, is required for EGF-induced endocytosis of flotillin-1 and flotillin-2 (Babuke et al., 2009). Although our data show that a defect in flotillin-1 Cys34 palmitoylation, but not in Tyr160 phosphorylation, abrogates IGF-1-induced IGF-1R signaling activation in the hetero-oligomeric flotillin-1- and flotillin-2-enriched plasma membrane microdomain, whether IGF-1-induced palmitoylation turnover and/or dephosphorylation of flotillin-1 are responsible for endocytosis of flotillin-1 and/or IGF-1R remains to be elucidated. The present study shows that tyrosine phosphorylated flotillin-1 is dephosphorylated after IGF-1 stimulation in HEK293T cells, but not in HeLa cells, and that this dephosphorylation coincides with IGF-1-induced turnover of flotillin-1 palmitoylation. However, in both HEK293T and HeLa cells, IGF-1R activation was controlled by the flotillin-1 palmitoylation turnover but not the change in its phosphorylation status. Depletion of flotillin-1 caused a reduction of flotillin-2 expression in HEK293T and HeLa cells, but different flotillin-2 stability between HEK293T and HeLa cells was observed upon IGF-1-induced flotillin-1 palmitoylation turnover. Taken together, our data show that IGF-1-induced flotillin-1 palmitoylation turnover in the hetero-oligomeric flotillin-1- and flotillin-2-enriched plasma membrane microdomain regulates tyrosine kinase activation of IGF-1R independently of caveolins and caveolae, regardless of cell type. However, the results suggest that the endocytosis-mediated flotillin-2 degradation that is induced by the flotillin-1 palmitoylation turnover does depend on cell type.

A genetic mutation that prevented flotillin-1 palmitoylation, as well as pharmacological blockage of flotillin-1 palmitoylation, inactivated IGF-1-induced IGF-1R signaling and hence prevented HeLa cell proliferation. Our findings thus identify flotillin-1 palmitoylation as a potential mechanism to control tumorigenesis. For now, the regulatory function of flotillin-1 palmitoylation and its turnover in various cancers remains to be explored. However, our finding that the intracellular localization and activation of IGF-1R is regulated by flotillin-1 palmitoylation provides a new approach to explore unknown mechanisms of IGF-1R-mediated signaling propagation in cancers.

Cell culture reagents were purchased from Gibco. Antibodies and reagents used were as follows: anti-flotillin-1 (BD 610820), anti-caveolin-1 (BD 610406), anti-caveolin-2 (BD 610685), anti-pan-Akt (BD 610860), anti-phosphotyrosine (PY20) (BD 610000) and anti-E-cadherin (BD 610182) antibodies from BD Transduction Laboratories; anti-flotillin-2 (number 3436), anti-IGF-1R (number 3027), anti phosphorylated (p)-ERK 1/2 (number 9101), anti-ERK1/2 (number 9102) and anti-pan-p-Akt (number 9271) antibodies from Cell Signaling Technology; anti-GM130 (ab31561) antibody from Abcam; anti-actin (I-19) (sc-1616), anti-calnexin (sc-11397), anti-α-tubulin (sc-5286) and horseradish peroxidase (HRP)-conjugated anti-goat-IgG (sc-2020) antibodies from Santa Cruz Biotechnology; transferrin receptor (13-6800) antibody from Invitrogen; anti-FLAG (F1804), anti-HA (H3663), HRP-conjugated anti-mouse-IgG (A4416) and anti-rabbit-IgG (A6154) antibodies and IGF-1 (I1271), CHX (C7698), 2-BP (21604), dynasore (D7693), and MβCD (C4555) from Sigma-Aldrich; MG132 (474790) and latrunculin B (428020) from Calbiochem; and HRP-conjugated cholera toxin subunit B (C-34780) from Molecular Probes.

Human embryonic kidney (HEK) 293T cells, HEK293T cells stably transfected with flotillin-1 shRNA (see below) and HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM, Gibco/BRL) containing 5 mM D-glucose supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich) and 1% penicillin-streptomycin (Gibco/BRL) in a 5% CO₂ incubator at 37°C.

To generate HEK293T cells stably transfected with flotillin-1 shRNA, HEK293T cells were transfected with a flotillin-1-specific shRNA lentiviral plasmid set (MISSION^® shRNA Bacterial Glycerol Stock, Sigma-Aldrich) and the MISSION^® Non-Target shRNA Control Vector (pLKO.1-puro, Sigma-Aldrich) for 48 h and subjected to incubation with 1 µg/ml puromycin (Clontech) to select for puromycin-resistant clones for 1 week. Independent colonies were then picked using a cloning cylinder (Sigma), subcultured and tested for flotillin-1 expression by reverse transcription (RT)-PCR and immunoblot analysis. Stable cell lines that expressed a flotillin-1 shRNA were then selected.

A full-length Homo sapiens flotillin-1 cDNA (GeneBank accession number NM_005803.2) was subcloned into the pDonor207 and moved into pDS-X-FLAG vector using Gateway^TM Technology (Invitrogen), and into pEGFP-N1 vector (Clontech Laboratores). Flotillin-1-C34A–FLAG and flotillin-1-Y160E–GFP and flotillin-1-Y160F–GFP were generated by using the WT flotillin-1–FLAG and flotillin-1–GFP as templates with a EZchange site-directed mutagenesis kit (Enzynomics, Daejeon, Korea). All expression vectors were verified by sequencing. The CFP–ER plasmid, encoding enhanced CFP fused to an ER-targeting sequence (the targeting sequence of calreticulin), was obtained from Clontech Laboratories. The HA–ubiquitin plasmid was kindly provided by Jiyun Yoo (Gyeongsang National University, Jinju, Korea).

Total RNA was extracted with TRIzol reagent (SolGent Co., Ltd.) according to the manufacturer's instructions and cDNA was generated using the Accupower RT PreMix (Bioneer). The cDNA was used as the template for subsequent PCR amplification. PCR primers were as follows: flotillin-1, 5′-ATGTTTTTCACTTGTGGCCC-3′ and 5′-TCAGGCTGTTCTCAAAGGCT-3′; flotillin-2, 5′-ATGGGCAATTGCCACACGGTA-3′ and 5′-TCACACCTGCACACCAGTGGC-3′; GAPDH, 5′-CTCAGTGTAGCCCAGGATGCC-3′ and 5′-ACCACCATGGAGAAGGCTGG-3′. PCR was performed using KOD FX Neo polymerase (Toyobo). The PCR fragments were separated by running on 1% agarose gels.

For the ubiquitylation assay, denaturing immunoprecipitation was performed as described previously (Hayer et al., 2010). HA–ubiquitin-expressing shRNA vector control and HEK293T stably expressing flotillin-1 shRNA were lysed with denaturing lysis buffer (150 mM KCl, 50 mM Tris-HCl pH 7.4, 5 mM MgCl₂, 1% Triton X-100, 5% glycerol, 2 mM β-mercaptoethanol, 0.2 mM PMSF, 1 μg/μl aprotinin and 1 μg/μl leupeptin). 1% SDS was added to the cell lysates, which were then boiled for 5 min, and diluted to yield 0.1% SDS final concentration using the denaturing lysis buffer. The denatured lysates were immunoprecipitated with anti-flotillin-2 antibody and subjected to immunoblot analysis.

Cells were lysed in immunoprecipitation buffer (1% Triton X-100, 150 mM NaCl, 10 mM Tris-HCl pH 7.4, 1 mM EDTA, 1 mM EGTA, 0.2 mM sodium ortho-vanadate, 0.2 mM PMSF, and 0.5% Nonidet P-40) containing 60 mM n-octylglucoside (Calbiochem). The cell lysates were then put on ice for 30 min and centrifuged at 13,500 g for 20 min at 4°C the supernatants were subjected to immunoprecipitation with anti-IGF-1R or the PY20 antibody. The lysates were rotated overnight at 4°C and then incubated with protein A or G plus agarose (Calbiochem) for 4 h at 4 °C. The immunocomplexes were collected by centrifugation at 13,500 g for 10 min at 4°C and washed three times in immunoprecipitation buffer. The immunoprecipitates were then resolved, separated by SDS-PAGE, and transferred onto polyvinylidene fluoride membrane (Millipore). Membranes were blocked for 2 h in 2% non-fat dried milk powder in Tris-buffered saline with 0.1% Tween-20 (TBS-T) at room temperature and incubated overnight at 4°C in primary antibody in 2% BSA in TBS-T, followed by three washes in TBS-T. The membranes were incubated for 1 h at room temperature in HRP-conjugated secondary antibody in 2% non-fat dried milk powder in TBS-T and washed three times in TBS-T. The membranes were developed using a Luminata^TM Crescendo western HRP substrate (Millipore).

An Elk translucent reporter vector (Elk-1-Luc), used to monitor the transcription factor binding activity of Elk-1, was purchased from Panomics. shRNA vector control and stable flotillin-1 shRNA HEK293T cells were transiently transfected with Elk-1-Luc or Elk-1-Luc and the FLAG-encoding vector, flotillin-1–FLAG or flotillin-1-C34A–FLAG for 48 h in each well of a 24-well plate. The Renilla reporter construct pRL-TK was used to normalize the transfection efficiency. The cells were treated with or without 10 nM IGF-1 for 24 h, washed twice with ice-cold PBS and lysed in 100 μl/well of passive lysis buffer (Promega). Luciferase activity was measured using a dual-luciferase reporter assay system (Promega).

The MTT [3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay was performed as described previously (Lee et al., 2011). Stable flotillin-1 shRNA HEK293T cells and HeLa cells were plated in 96-well plates and transfected with FLAG vector, flotillin-1–FLAG or flotillin-1-C34A–FLAG for 48 h. The cells and shRNA vector control cells were treated with or without 10 nM IGF-1 for 24 h and incubated with 0.5 mg/ml MTT solution (Sigma-Aldrich) at 37°C for 4 h. The cells were then incubated with 100 μl of DMSO (BioShop, CA) at 37°C for 30 min to solubilize the final product of MTT metabolism, the formazan precipitate. The optical density of each well was measured using a microplate reader set to 540 nm (Model 550, Bio-Rad Laboratories).

Immunofluorescent labeling of cells and confocal analysis were performed as described previously (Jeong et al., 2015). HEK293T and flotillin-1 shRNA stable HEK293T cells were transfected with flotillin-1–FLAG or flotillin-1-C34A–FLAG together with CFP–ER for 48 h and treated with or without 80 μM dynasore for 30 min. The cells and shRNA vector control cells were fixed with 3.7% paraformaldehyde in PBS for 20 min and permeabilized with 0.1% Triton X-100 for 20 min at room temperature. Permeabilized cells were rinsed with PBS and blocked with 1% BSA in PBS for 1 h at room temperature. Cells were rinsed with PBS and incubated with anti-flotillin-1 and/or anti-IGF-1R, anti-flotillin-1 and/or anti-flotillin-2, or anti-transferrin receptor (TfR) antibodies in 1% BSA in PBS for overnight at 4°C. After washing three times with PBS, the cells were incubated with TRITC-conjugated anti-mouse-IgG and/or FITC-conjugated anti-rabbit-IgG antibodies in 1% BSA in PBS for 2 h at room temperature. Fluorescence images were obtained using an Olympus Fluoview 1000 confocal microscope attached to an Olympus BX61 vertical microscope equipped with PlanApo 60×/1.40 NA oil immersion objective (Olympus). FITC signals were excited using an argon laser at 488 nm, TRITC signals using a He-Ne laser at 543 nm, and CFP signals using a diode laser at 405 nm. Images were acquired as a single stack of optical sections in the z axis, and dual-color immunofluorescence images were collected in simultaneous two-channel mode. FV10-ASW software (Olympus) was used to merge the channels for TRITC, FITC and CFP.

As previously described (Kwon et al., 2013), flotillin-1–FLAG or flotillin-1-C34A–FLAG-expressing HEK293T cells and stable flotillin-1 shRNA HEK293T cells were scraped into isotonic buffer (250 mM sucrose, 20 mM Tris-HCl pH 7.5, 1 mM EDTA, 15 mM KCl, 2 mM MgCl₂, 1 mM PMSF, 1 μg/μl aprotinin and 1 μg/μl leupeptin) and homogenized using ten strokes of a Dounce homogenizer. The homogenates were centrifuged at 10,000 g for 45 min at 4 °C and the pellet was resuspended in the isotonic buffer. The resuspended pellet was loaded onto a 1.12 M sucrose cushion and centrifuged at 100,000 g for 2 h at 4°C in a SW41Ti rotor (Beckman Instruments). The interface plasma membrane fraction was collected, diluted with ice-cold PBS and pelleted by centrifugation at 200,000 g for 1 h at 4°C in a SW41Ti rotor.

Purification of palmitoylated proteins by the fatty acyl biotin exchange was performed as described previously (Kwon et al., 2015; Tulloch et al., 2011). HEK293T and HeLa cells were lysed with buffer A (1% SDS in PBS containing 0.2 mM PMSF, 1 μg/μl aprotinin and 1 μg/μl leupeptin) and free cysteine residues were alkylated by overnight incubation at 4°C with 25 mg/ml N-ethylmaleimide. Excess soluble N-ethylmaleimide was removed by chloroform–methanol precipitation. Protein pellets were resolubilized in buffer A. The samples were treated with 200 mM hydroxylamine (pH 7.4) and then incubated with 1 mM of the pyridyldithiol-activated cysteine-reactive biotinylation reagent N-[6-(Biotinamido)hexyl]-3′-(2′-pyridyldithio)-propionamide for 1 h in the dark at room temperature. Excess biotinylation reagent was removed by chloroform–methanol precipitation. Protein pellets were resolubilized in buffer B (1% Triton X-100 and 0.2% SDS in PBS containing 0.2 mM PMSF, 1 μg/μl aprotinin and 1 μg/μl leupeptin). The samples were incubated with NeutrAvidin–agarose resin beads (Thermo Scientific) overnight at 4°C. The beads were collected by centrifugation at 13,500 g for 10 min at 4°C, followed by four washes in buffer B and biotinylated proteins were eluted in 5× SDS-PAGE sample buffer. The biotinylated proteins were then resolved, separated by SDS-PAGE, and subjected to immunoblot analysis.

Sucrose velocity gradient centrifugation was performed as described previously (Hayer et al., 2010; Kwon et al., 2013). The purified plasma membrane pellet of HEK293T cells was resuspended in 0.5% Triton X-100 lysis buffer (100 mM NaCl, 20 mM Tris-HCl, pH 7.4, 5 mM EDTA and protease inhibitors) for 20 min at 4°C and loaded onto 10–40% linear sucrose gradients containing 0.5% Triton X-100 lysis buffer. After centrifugation in a SW41Ti rotor (Beckman Instruments) at 237,020 g for 255 min at 4°C, fourteen 720-μl fractions were collected from the top to the bottom. An equal volume from each gradient fraction was separated by SDS-PAGE and subjected to immunoblot analysis.

Detergent-resistant membrane isolation was performed as described previously (Langlois et al., 2008). HEK293T cells were treated with or without 10 nM IGF-1 for 10 min. The cells were scraped with MBS buffer (25 mM MES pH 6.5 and 150 mM NaCl) containing 1% Triton X-100, and homogenized using ten strokes of a Dounce homogenizer. The homogenates were mixed with 90% sucrose and overlaid with 35% and 5% sucrose. The gradients were centrifuged at 200,000 g for 18 h at 4°C in a SW41Ti rotor. A total of 12 fractions of 1 ml were collected from the top to the bottom. An equal volume from each fraction was separated by SDS-PAGE and subjected to immunoblot analysis.

Four-step sucrose density gradient fractionation was performed as described previously (Kwon et al., 2013; Yao et al., 2009) to separate the HLDM from the LLDM plasma membrane fraction. The purified plasma membrane pellet from HEK293T cells was resuspended with 2 ml of 500 mM sodium carbonate, pH 11.0, homogenized using ten strokes of a Dounce homogenizer, and mixed with 2 ml of 90% sucrose in MBS buffer. The mixture was placed at the bottom of an ultracentrifuge tube and overlaid with 3 ml of 35% sucrose, 4 ml of 21% sucrose and 1 ml of 5% sucrose in MBS buffer containing 250 mM sodium carbonate, pH 11.0. The gradients were centrifuged at 280,000 g for 18 h at 4°C in a SW41Ti rotor. From the top of each gradient, 0.5 ml gradient fractions were collected to yield a total of 24 fractions. An equal volume from each fraction was separated by SDS-PAGE and subjected to immunoblot analysis.

An equal volume of the fractionated samples from the four-step sucrose density gradient fractionation as described above was dot-blotted onto Protran nitrocellulose membranes (Whatman). The membranes were blocked overnight at 4°C with 5% non-fat dried milk powder in TBS-T and incubated for 1 h at room temperature with HRP-conjugated cholera toxin subunit B, a ligand for the ganglioside GM1, in 2% non-fat dried milk powder in TBS-T. The membranes were washed three times in TBS-T and developed using a Luminata^TM Crescendo western HRP substrate (Millipore).

Cholesterol was measured by using an Amplex^® Red Cholesterol Assay Kit (Molecular Probes, A12216) according to the manufacturer's protocols.

Immunoblots were analyzed by scanning densitometry using a Kodak Gel Logic 100 imaging System (Eastman Kodak Co.). Data are expressed as mean±s.d., and P-values were calculated using a Student's t-test (*P<0.05, **P<0.01, ***P<0.001).

We thank Jae-Yong Park (School of Biosystem and Biomedical Science, College of Health Science, Korea University, Seoul, Korea), Eun Mi Hwang (Center for Functional Connectomics, Korea Institute of Science and Technology, Seoul, Korea), and Jiyun Yoo (Division of Life Science, Research Institute of Life Sciences, Gyeongsang National University, Jinju, Korea) for sharing reagents.