|Home | About | Journals | Submit | Contact Us | Français|
The FYVE domain associates with phosphatidylinositol 3-phosphate [PtdIns(3)P] in membranes of early endosomes and penetrates bilayers. Here, we detail principles of membrane anchoring and show that the FYVE domain insertion into PtdIns(3)P-enriched membranes and membrane-mimetics is substantially increased in acidic conditions. The EEA1 FYVE domain binds to POPC/POPE/PtdIns(3)P vesicles with a Kd of 49 nM at pH 6.0, however associates ~24 fold weaker at pH 8.0. The decrease in the affinity is primarily due to much faster dissociation of the protein from the bilayers in basic media. Lowering the pH enhances the interaction of the Hrs, RUFY1, Vps27p and WDFY1 FYVE domains with PtdIns(3)P-containing membranes in vitro and in vivo, indicating that pH-dependency is a general function of the FYVE finger family. The PtdIns(3)P binding and membrane insertion of the FYVE domain is modulated by the two adjacent His residues of the R(R/K)HHCRXCG signature motif. Mutation of either His residue abolishes the pH-sensitivity. Both protonation of the His residues and nonspecific electrostatic contacts stabilize the FYVE domain in the lipid-bound form, promoting its penetration and increasing the membrane residence time.
Phosphoinositide (PI) 3-kinases regulate membrane trafficking, protein sorting and signaling by generating phosphatidylinositol (PtdIns) derivatives phosphorylated at the third position of the inositol ring.1,2 Of the four known products of PI 3-kinases, PtdIns 3-phosphate [PtdIns(3)P] is the most abundant and is constitutively produced in the cytosolic leaflet of membranes of early endosomes. In contrast to the multiply phosphorylated PtdIns’s, PtdIns(3)P does not undergo hydrolysis into the secondary messengers, but rather functions as a signaling molecule itself and is specifically recognized by FYVE domains.3–5 The FYVE finger is a small ~70 residue module discovered in 1997 and named after the four proteins, Fab1, YOTB, Vac1, and EEA1 (reviewed in Refs. 6 and 7). The FYVE domain fold consists of two double-stranded antiparallel β-sheets and a carboxy-terminal α-helix and is stabilized by two zinc binding clusters. The three highly conserved regions, the N-terminal WxxD sequence, the basic ‘signature’ motif [R(R/K)HHCRXCG], and the C-terminal RVC region form the PtdIns(3)P binding site and distinguish the FYVE subfamily from other structurally related RING fingers. The majority of FYVE finger-containing proteins are involved in endo- and exocytosis, intracellular signaling and cytoskeletal tethering and their targeting to PtdIns(3)P-enriched membranes is mediated by the FYVE domains.
Human EEA1 (early endosome antigen 1), Hrs (hepatocyte growth factor-regulated tyrosine kinase substrate), RUFY1 (RUN and FYVE domains-containing 1), WDFY1 (WD repeat and FYVE domain containing 1), and yeast Vps27p are key downstream effectors of PI 3-kinases. EEA1 functions as a docking/tethering protein and is required for homo- and heterotypic fusion of endosomes.8,9 Owing to their robust association with early endosomes, EEA1 and WDFY1/FENS1 (Ref. 10) are commonly used as markers of these organelles. Hrs11,12 is involved in signal transduction and, as its yeast homologue Vps27p,13 in endocytic protein trafficking. RUFY1 is a human homologue of Rabip4, an effector of Rab4 GTPase that regulates recycling of endocytosed cargo.14,15 Recognition of PtdIns(3)P by the FYVE domains results in efficient translocation of these normally cytoplasmic proteins to endosomal, lysosomal, or vacuolar membranes. We have recently found that the interaction of the EEA1 FYVE domain with PtdIns(3)P is modulated by a histidine switch and is increased by lowering pH.16 Additional membrane anchoring is provided by the insertion of a hydrophobic loop of the FYVE domain into bilayers and by nonspecific electrostatic interactions with negatively charged membrane surfaces.17–19 However, it remains unclear if pH-sensitivity is a common function of the FYVE domain family and whether the pH fluctuations and nonspecific interactions have an effect on the docking, insertion and membrane affinity of this module. Here, we detail principles of the membrane targeting by the FYVE domain, show that the extent of the membrane penetration is increased in acidic media, and demonstrate that the pH-dependency is a general mechanism of FYVE domain regulation.
The wild type FYVE domains of EEA1, Hrs, RUFY1, Vps27p and WDFY1, and H1371A, H1372A and H1371A/H1372A mutants of EEA1 FYVE domain were expressed in E. Coli BL21 cells. Two liters of LB or 15NH4Cl-supplemented minimal media containing 100 μg/mL ampicillin were inoculated with an overnight culture of BL21 cells harboring each construct and grown at 37°C until absorbance at 600 nm reached 0.6. After induction with 0.2 mM of isopropyl-1-thio-β-D-galactopyrano-side and incubation at 25°C for 16 h, cells were harvested for 10 min at 6000g, and the resulting pellet was resuspended in 10 mL of 20 mM Tris pH 8.0 buffer, containing 160 mM NaCl, 1 mM dithiothreitol (DTT) and 1.0% Triton X-100. The solution was then sonicated for 8 min using a 30 s sonication on ice followed by 30 s incubation on ice. The homogenate was centrifuged at 50,000g, and the supernatant was filtered into a 50 mL tube. The unlabeled and uniformly 15N-labeled GST-fusion proteins were purified using GST binding resin (EMD Biosciences, San Diego, CA) or a glutathione sepharose 4B column (Amersham). In case of EEA1 and RUFY1, the GST tag was cleaved with Thrombin (Sigma). The proteins were concentrated in millipore concentrators (Millipore) and purified by FPLC. The purity of the proteins was verified by SDS-PAGE and 1H NMR. Protein concentrations were determined using the BCA method (Pierce).
Insertion of the wild type and mutant EEA1 FYVE domains into a phospholipid monolayer was investigated by measuring the change in surface pressure (π) of invariable surface area during gradual addition of the proteins. The experiments were performed using a 1 mL circular Teflon trough and wire probe connected to a Kibron Micro-Trough X (Kibron, Helsinki) as described previously.20 A monolayer containing various combinations of phospholipids was spread onto the subphase composed of either 10 mM KH2PO4/160 mM KCl (pH 6.0), 10 mM HEPES/160 mM KCl (pH 7.4), or 10 mM HEPES/160 mM KCl (pH 8.0) until the desired initial surface pressure (π0) was reached. After stabilization of the signal (~5 min), 15 μg of protein was injected into the subphase through a hole in the wall of the trough. The surface pressure change (Δπ) was monitored for 45 min. The (Δπ) value reached a maximum after 30 min in all experiments.
The EGFP-EEA1-FYVE domain yeast expression plasmid was generated as described.3 The plasmid was transformed into a wild type yeast strain (SEY6210: MATα, his3-Δ200, ura3-52, trp1-Δ901, lys2-801, suc2-Δ9, leu2-3112). For microscopy, yeast strains were grown in selective media overnight and the following morning a fresh culture was inoculated and grown at 26°C until an OD600 of 0.5. The cells were visualized by fluorescence microscopy as described,21 using a Nikon Eclipse E800 microscope fitted with a cooled, high-resolution charge-coupled device camera (Hammamatsu Photogenics). Images were acquired using Phase 3 Imaging software (Phase 3 Imaging Systems).
All surface plasmon resonance (SPR) measurements were performed at 25°C as described previously.20 Equilibrium SPR measurements were carried out at the flow rate of 2 μL/min. After injecting 85 μL of the FYVE domain, the protein-lipid association and dissociation were monitored for 1020 and 500 s, respectively. The sensor-grams were obtained using five or more different concentrations of the protein (within a 10-fold range of Kd) for each condition tested. The sensorgrams were corrected for refractive index change by subtracting the control surface response. A total of three datasets were generated to obtain a standard deviation. The maximal response (Req, saturation value) was determined from the plots and plotted versus protein concentrations (C). The Kd values, determined by a nonlinear least squares analysis of the binding isotherm using the equation Req = Rmax/(1 + Kd/C),18 are summarized in Table I. Kinetic SPR measurements were performed at a flow rate of 30 μL/min. Ninety microliters of the wild type or mutant FYVE domain was injected, and the protein-lipid association and dissociation were monitored for 3 min and over 6 min, respectively.
The liposome binding assays were performed as described in.16 Briefly, solutions of phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS) (Avanti), C16-PtdIns(3)P (Echelon Biosciences) dissolved in CHCl3:MeOH:H2O (65:25:4) were mixed and dried down under vacuum. The lipids were resuspended in 50 mM MOPS, pH 7.0, 150 mM KCl, and 1 mM DTT, and incubated at 65°C for 1 h. The liposomes were then frozen in liquid nitrogen and thawed at 37°C for three cycles. The liposome solution was passed through an Avanti extruder to make 1.0 μm liposomes. Liposomes were collected by centrifugation at 25,000g for 20 min and resuspended to a final concentration of 4 mM total lipids in 100 μL of binding buffers, containing 150 mM NaCl, 1 mM DTT, and 50 mM either MES pH 5.5, BisTris pH 6.5, HEPES pH 7.5, Tris pH 8.0, or Bicine pH 8.5. Liposomes were incubated with GST-fusion Hrs, Vps27p and WDFY1 FYVE domains, and untagged RUFY1 FYVE domain (200–250 μg/mL final protein concentration) for 30 min at room temperature and then collected again by centrifugation. The liposome pellets were resuspended in 100 μL of buffer and analyzed using SDS-PAGE and Coomassie brilliant blue staining.
The N-terminal-tagged EGFP-fusion FYVE domains of RUFY1, Vps27p, and WDFY1 were generated using corresponding pGEX-FYVE domains as templates and Bam-HII and EcoRI restriction sites. The HeLa cells were grown in Dulbecco modified essential medium (DMEM) (Mediatech) supplied with 10% fetal bovine serum (FBS) (Mediatech). The cells were transfected with mammalian plasmids encoding EGFP-RUFY1, Vps27p or WDFY1 FYVE fusion proteins using Effectene reagent (Qiagen) and 24 h later were replated onto 25-mm glass coverslips. Microscopic imaging of cells was performed 72 h after transfection. Before imaging, cells were washed three times with phosphate buffered saline (PBS) followed by addition of buffered solutions (pH = 6.0–8.5) containing 10 mg/mL nigericin (Sigma) and 10 mg/mL monensin (Sigma). The buffered solutions consisted of 120 mM KCl, 20 mM NaCl, 0.5 mM CaCl2, 0.5 mM MgSO4, and Hydrion Chemvelope pH buffer powder (Micro Essential Laboratory) or 25 mM of either Mops (Sigma), citrate (Acros), or Tris (Fisher Scientific). After pH equilibration for 15 min, the living cells were visualized at 33°C using Olympus IX81 inverted microscope equipped with FITC filter set (Chroma).
NMR experiments were performed at 298 K on a Varian INOVA 500 MHz spectrometer as described.16 The 1H,15N HSQC spectra of 0.2 mM H1371A and H1372A EEA1 FYVE domains were recorded as C4-PtdIns(3)P (up to 1.0 mM), and imidazole (up to 50 mM) were added stepwise.
The pKa values of His1371 and His1372 of the EEA1 FYVE domain were calculated by monitoring changes in the chemical shifts of δ2H and ε1H of both His residues upon varying pH. The 1D 1H NMR spectra of 0.5 mM FYVE domain were recorded in the presence of 120 mM deuterated d38-dodecylphosphocholine (DPC) (Cambridge Isotopes) containing 10% w/w PS in 2H2O as the pH of the samples was adjusted stepwise from 4.7 to 9.4. Proton chemical shifts were calibrated and referenced to the internal standard, 2,2-dimethyl-2-silapentane-5-sulfonate (Sigma). The pKa values of the His residues were obtained by a nonlinear least squares fitting of the chemical shifts observed in 1D spectra as a function of pH using a modified Henderson-Hasselbalch equation: δobs = [δacid + δbase10(pH − pKa)]/[1 + 10(pH − pKa)], where the δobs value is chemical shift at a given pH, and δacid and δbase are chemical shifts at the low and high extremes of pH, respectively.
Recent studies have shown that the FYVE domain inserts an exposed hydrophobic loop into membranes and membrane-mimetics.6,17–19,22,23 To test whether changes in pH influence the extent of the membrane insertion, we examined the lipid monolayer penetrating ability of the EEA1 FYVE domain in acidic, neutral, and basic conditions. The monolayers containing POPC/POPE (80:20) or POPC/POPE/PtdIns(3)P (78:20:2) were spread over an invariable area with initial surface pressure π, and after adding the FYVE domain into the subphase, the change in surface pressure (Δπ) was analyzed. Extrapolating Δπ versus π, one can obtain the critical surface pressure πc, which defines an upper limit of the monolayer pressure that a protein can insert into.24 Given that π of cell membranes and large unilamellar vesicles is in the range of 30–35 dyne/cm,25,26 for the FYVE domain to be able to penetrate these bilayers, the πc value should be above 30 dyne/cm.
Figure 1(a) shows that the EEA1 FYVE domain has no intrinsic POPC/POPE penetrating ability, and πc (~20 dyne/cm) remains essentially unchanged at pH 6.0, 7.4, and 8.0. Thus, in the absence of PtdIns(3)P, the EEA1 FYVE domain does not penetrate POPC/POPE membranes nor does the pH influence the insertion. However, when PtdIns(3)P was incorporated in the monolayer, the πc value increased to 31 dyne/cm at pH 7.4, indicating a moderate insertion. A similar augmentation in penetrating power due to PtdIns(3)P was observed for the FYVE domains of Hrs, Vps27p and FENS-1, where πc raised to ~30–34 dyne/cm.18,19 We next examined the ability of the EEA1 FYVE domain to penetrate PtdIns(3)P-containing monolayers in variable pH buffers. We found that πc increases from 31 dyne/cm at pH 7.4 to 35 dyne/cm at pH 6.0 but decreases to 27 dyne/cm at pH 8.0. Thus, insertion of the EEA1 FYVE domain into PtdIns(3)P-enriched membranes is substantially enhanced in acidic environment.
The robust increase of membrane penetration in vitro upon lowering pH paralleled changes in localization of the enhanced green fluorescent protein (EGFP)-fusion EEA1 FYVE domain in vivo [Fig. 1(b)]. The EGFP-EEA1 FYVE domain was expressed in yeast SEY6210 cells, and after brief incubation of the cells in buffers of pH 6.0, 7.0, 8.0, and 9.0, the intracellular distribution of the protein was examined by fluorescence microscopy. Although the EGFP-FYVE domain was strongly anchored to PtdIns(3) P-containing vacuolar and endosomal membranes in acidic conditions (pH < 7.0), the protein was dispersed in the cytosol at pH 8.0 or higher. Furthermore, in vivo association with PtdIns(3)P-enriched membranes, particularly in the physiological pH range of 6.0–8.0, had the trend fully consistent with the trend seen for in vitro binding of the EEA1 FYVE domain to PtdIns(3)P-containing membrane-mimetics [Fig. 1(a)].
To fully evaluate the effect of pH, we investigated association of the EEA1 FYVE domain with PtdIns(3)P-containing vesicles by SPR. The equilibrium binding constants were measured at near physiological pH levels of 6.0, 7.4, and 8.0 using POPC/POPE/PtdIns(3)P (78:20:2) vesicles and various protein concentrations [Fig. 2(a) and Materials and Methods]. We found that the EEA1 FYVE domain binds POPC/POPE/PtdIns(3)P vesicles with a Kd of 49 nM at pH 6.0 (Table I). However, its association with vesicles became considerably weaker as the pH of the buffer was raised from 6.0 to 8.0. The binding affinity of the EEA1 FYVE domain dropped by 10- and 24-fold upon alkalinization of the media to pH 7.4 and 8.0, respectively (Table I). These data demonstrate that acidity mediates recruitment of the EEA1 FYVE domain to PtdIns(3)P-containing bilayers, amplifying the binding affinity by 10 fold per one pH unit in the pH range of 6.0 to 8.0. The dependence of membrane anchoring on pH is summarized in Table I, which reveals a significant increase of the EEA1 FYVE domain activity in acidic conditions and a decrease in basic conditions.
To examine the effect of pH on the on- and off-rates, we analyzed kinetics of the EEA1 FYVE domain’s interaction with POPC/POPE/PtdIns(3)P vesicles at pH 6.0, 7.4, and 8.0. As shown in Figure 2(b), the dissociation rate is substantially slower in acidic (pH 6.0) conditions than in neutral (pH 7.4) or basic (pH 8.0) conditions. The association rate, however, remains largely unchanged at all pHs. Thus, enhancement of the EEA1 FYVE domain binding to PtdIns(3)P-containing membranes in low pH buffers is primarily due to a much slower dissociation rate of the protein from bilayers.
To determine whether the pH-sensitive membrane anchoring is a function of the FYVE family, FYVE domains of Hrs, RUFY1, Vps27p and WDFY1 were investigated by liposome binding assays and fluorescence microscopy. The GST-fusion Hrs, Vps27p and WDFY1 FYVE modules and untagged RUFY1 FYVE domain were incubated with small unilamellar vesicles (SUVs) composed of PC, PE, PS and PtdIns(3)P at pH 5.5, 6.5, 7.5, 8.0 or 8.5. The distribution of the proteins between supernatant and SUVs was examined by centrifugation [Fig. 3(a)]. Densitometry analysis of the gel bands revealed that 88% of the Hrs1 FYVE domain, 85% of the RUFY1 FYVE domain, 78% of the WDFY1 FYVE domain, and 58% of the Vps27p FYVE domain were retained in the pelleted liposome fraction at a low pH of 5.5. However, FYVE domains of all proteins were increasingly redistributed to the supernatant as the pH of the buffer was progressively increased. Particularly, binding of the Hrs1, RUFY1, and Vps27p FYVE domains to PtdIns(3)P-enriched vesicles was reduced to ~20% as a result of alkalinization of the buffer from pH 6.0 to 8.5. Although the WDFY1 FYVE domain clearly displayed the pH dependence toward PtdIns(3)P, almost a half of this protein (47%) was still bound to the SUVs at pH 8.5. This is consistent with the fact that the WDFY1 FYVE domain has a 10-residue extension in the membrane insertion loop and associates with membranes particularly tightly.10
A similar pH-dependency was observed for the in vivo localization of the EGFP-fusion RUFY1, Vps27p, and WDFY1 FYVE domains to endosomal membranes in human HeLa cells [Fig. 3(b)]. Following acidification of the cytosol, most of the FYVE domains were found to co-localize with PtdIns(3)P-enriched early endosomes. However, when the cytosolic pH was gradually increased, the EGFP-FYVE domains became more soluble, and at pH > 7.5 were entirely cytoplasmic. Thus, FYVE domains of RUFY1, Vps27p, and WDFY1 translocate to PtdIns(3)P-containing membranes at low cytosolic pH, demonstrating that the pH-sensitivity represents a general characteristic of the FYVE module.
Of the all residues in the R(R/K)HHCRXCG “signature” motif, the two histidine residues (His1371 and His1372 of EEA1) are absolutely conserved among all FYVE domains (see Fig. 3(c, d)). They form hydrogen bonds to the 3-phosphate and the 4- and 5-hydroxyl groups of PtdIns(3)P (Ref. 27) and are necessary for the lipid recognition.16 To determine if the membrane insertion of the EEA1 FYVE domain requires both histidine residues, H1371 and H1372 were substituted with Ala, and the mutant proteins were probed by monolayer penetration experiments [Fig. 4(a, b)]. In acidic conditions (pH 6.0), in which the wild type EEA1 FYVE domain inserts furthest, penetration of the H1371A and H1372A mutants into a POPC/POPE/PtdIns(3)P monolayer was significantly compromised. The πc value was decreased by 3 dyne/cm for H1371A and by 2.5 dyne/cm for H1372A. The double H1371A/H1372A mutation caused a further drop in penetrating power, with the surface pressure decreasing to 30 dyne/cm. Thus, both His residues are necessary for the substantial penetration of the EEA1 FYVE domain into PtdIns(3)P-containing membranes. Furthermore, deprotonation of the His residues abolished insertion. The πc value was decreased to 30 dyne/cm at pH 7.4 and to 27 dyne/cm at pH 8.0 for the single mutants. At pH 8.0, both His residues became fully deprotonated based on their pKa values determined by chemical shift changes in imidazole’s δ2H and ε1H upon varying pH [Fig. 4(c)]. Consequently, in a basic environment, neither mutants nor wild type protein were able to insert into the monolayer, suggesting that the two additional positive charges in the PtdIns(3)P binding site are essential for the robust interaction and membrane penetration.
The critical role of His residues in pH-sensitivity was underscored by measuring the monolayer penetration of PtdIns(3)P-binding p40phox PX domain. Unlike FYVE domain, the p40phox PX module contains no His residues in the PtdIns(3)P binding site. As expected, its penetrating ability remained unchanged in all pH buffers [Fig. 4(d)].
To determine the contribution of individual His1371 and His1372 residues in the association of the EEA1 FYVE domain with bilayers, binding of the mutants to POPC/POPE/PtdIns(3)P vesicles was tested by SPR. We found that the H1371A mutant associates with vesicles ~13-fold weaker than the wild type protein at pH 6.0, 9-fold weaker at pH 7.4 and at least 6-fold weaker at pH 8.0 (Table I). Overall, the H1371A mutation caused at least a 10-fold drop in affinity when the pH of the buffer was changed from 6.0 to 8.0. Likewise, the H1372A mutant showed a steady decrease in binding upon alkalinization of the media. Together, these results reveal that protonation of both His residues is required for the membrane recruitment. The H1371A/H1372A double mutation abolished the pH sensitivity of the FYVE domain (Kd > 5 μM in all conditions tested), indicating that the two His residues play a major role in the pH dependent behavior of the FYVE domain. The tandem histidines occupy strikingly similar positions in the binding pockets of FYVE modules [Fig. 3(d)], suggesting that they may regulate general function of the FYVE proteins.
The kinetic studies of the interaction involving H1371A or H1372A EEA1 FYVE domain and PtdIns(3) P-containing vesicles demonstrate that the loss of binding activity by the mutants was primarily due to a much faster off-rate when compared with the off-rate of the wild type protein [Fig. 2(c)]. This underscores the importance of the His residues and their protonation in stabilization of the FYVE-PtdIns(3)P complex at membranes. Addition of imidazole to either His mutant did not restore their binding activity based on the lack of resonance perturbations in 1H,15N heteronuclear single quantum coherence (HSQC) spectra of the mutants [Fig. 4(e) and data not shown], demonstrating that the intact side chains of His residues are necessary for the proper coordination of PtdIns(3)P.
It has been shown that membrane association of the FYVE domain is facilitated by nonspecific electrostatic interactions, which involve basic residues of the protein and anionic phospholipids in the bilayer.18,28 A strong positive potential surrounding the insertion loop and PtdIns(3)P binding pocket may drive initial association with the membrane and facilitate PtdIns(3)P recognition.18,28 Protonated His1371 and His1372 residues may contribute to the electrostatic component because their pKa values were increased by 0.2–0.3 pH units in the presence of PS in DPC micelles [Fig. 4(c)]. However, their contribution to the association appears to be small because the on-rate of the EEA1 FYVE domain was not significantly affected by the changes in pH [Fig. 2(b)]. We concluded that Arg and/or Lys residues surrounding the insertion loop and PtdIns(3)P binding site are primarily involved in the electrostatic interactions, whereas protonation of His residues strengthens the coordination of PtdIns(3)P and promotes membrane penetration. This is consistent with the notion that PI-coordination and close range interactions mediate the dissociation of other peripheral proteins from membranes.7,29
In conclusion, the data presented here provide new insight into the mulivalent mechanism of docking and penetration into PtdIns(3)P-enriched membranes and membrane-mimetics by the FYVE domain. Our results reveal that the membrane insertion is enhanced in acidic media and that the pH-dependency represents a general mechanism for regulation of the FYVE domain family.
National Institutes of Health, Indiana University, the American Heart Association, the American Cancer Society.
The authors thank W. Cho, P. Hawkins, W. Hong, J. Hurley, Y. Qiu, and F. Quiocho for kindly providing original cDNAs of FYVE domains, and R. Eyeson, S. Lee, and Y. Xue for helping with experiments.