|Home | About | Journals | Submit | Contact Us | Français|
The Arf exchange factor Grp1 selectively binds phosphatidylinositol 3,4,5-triphosphate (PtdIns(3,4,5)P3), which is required for recruitment to the plasma membrane in stimulated cells. The mechanisms for phosphoinositide recognition by the PH domain, catalysis of nucleotide exchange by the Sec7 domain, and autoinhibition by elements proximal to the PH domain are well characterized. The N-terminal heptad repeats in Grp1 have also been shown to mediate homodimerization in vitro as well as heteromeric interactions with heptad repeats in the FERM domain-containing protein Grsp1 both in vitro and in cells (1). Here, we have characterized the oligomeric state of Grsp1 and Grp1 family proteins (Grp1, ARNO, and Cytohesin-1) as well as the oligomeric state, stoichiometry, and specificity of Grsp1 complexes with Grp1, ARNO and Cytohesin-1. At low micromolar concentrations, Grp1 and ARNO are homodimeric whereas Cytohesin-1 and Grsp1 are monomeric. When mixed with Grsp1, Grp1 homodimers and Cytohesin-1 monomers spontaneously re-equilibrate to form heterodimers whereas approximately 50% of ARNO remains homodimeric under the same conditions. Fluorescence resonance energy transfer experiments suggest that the Grsp1 heterodimers with Grp1 and Cytohesin-1 adopt a largely anti-parallel orientation. Finally, formation of Grsp1-Grp1 heterodimers does not substantially influence Grp1 binding to the head groups of PtdIns(3,4,5)P3 or PtdIns(4,5)P2 nor does it influence partitioning with liposomes containing PtdIns(3,4,5)P3, PtdIns(4,5)P2 and/or phosphatidyl serine.
Stimulation of cells with agonists such as insulin and EGF results in activation of phosphatidylinositol 3 kinase (PI-3 kinase) (2-4), leading to transient accumulation of the lipid second messenger phosphatidylinositol (PtdIns) 3,4,5-trisphosphate (PtdIns(3,4,5)P3). Production of PtdIns(3,4,5)P3 controls diverse cellular processes through plasma membrane recruitment of proteins and protein complexes, including Grp1. Grp1 (also referred to as Cytohesin-3) belongs to the homologous Grp1 family of functionally related Arf guanine nucleotide exchange factors (GEFs) that includes ARNO (Cytohesin-2) and Cytohesin-1. Grp1, ARNO and Cytohesin-1 have a modular architecture consisting of N-terminal heptad repeats, a Sec7 domain with exchange activity for Arf1 and Arf6, a pleckstrin homology (PH) domain, and a C-terminal polybasic sequence (5). The Grp1 PH domain selectively binds PtdIns(3,4,5)P3 with high affinity and is essential for plasma membrane targeting (6, 7). Localization of Grp1 family proteins to the plasma membrane and subsequent activation of Arfs has been implicated in a variety of cellular processes including adhesion, endocytic trafficking, cell motility, T-cell anergy, helper T-cell activation, and insulin signaling (8).
EST and Affymetrix gene chip transcriptomes indicate that ARNO and Cytohesin-1 are ubiquitously expressed whereas Grp1 is broadly expressed, though at relatively low levels in certain tissues such as liver, thymus and peripheral blood lymphocytes (9, 10). Grsp1 was originally isolated from a mouse brain cDNA expression library probed with Grp1 and shown by Western blotting to be expressed at significant levels in brain and lung, where Grp1 is also highly expressed (1). A subsequent analysis by RT-PCR suggests that Grsp1 is expressed at high levels in other tissues as well, including kidney, spleen, heart and bone marrow (11). Grsp1-Grp1 complexes were readily detected by co-precipitation in lysates from co-transfected COS-1 cells, but not in mixed lysates from separately transfected cells, and complexes of the endogenous proteins have been co-precipitated from mouse lung homogenates (1). Grsp1 contains several putative protein-protein interaction domains including an N-terminal FERM domain, which is followed by two heptad repeat regions with a high propensity to form coiled-coils (12). Deletion mapping indicated that the interaction between Grp1 and Grsp1 is mediated by the N-terminal heptad repeats in Grp1 and the first of the two heptad repeat regions in Grsp1 (1). FERM domains have been shown to mediate high affinity protein-protein interactions with the cytoplasmic domains of integral membrane proteins including CD44 and ICAM-2 (13). The multiple protein-protein interaction motifs present in Grsp1 suggest that it may function as a molecular scaffold. Indeed, the Grsp1-Grp1 complex co-localizes with cortical actin rich regions in response to stimulation of CHO-T cells with insulin or EGF (1). Taken together, this data suggests Grp1 may function not only to activate Arf proteins at the cell membrane, but also to recruit additional functionality to the cell membrane in response to extracellular signals.
The presence of a phosphoinositide specific PH domain in Grp1 and a putative protein or lipid binding FERM domain in Grsp1 is consistent with the possibility that both protein-lipid and protein-protein interactions may contribute to localization and/or assembly into higher order complexes. In mouse lung tissue extracts, a Grsp1 antibody co-precipitated only a small amount of the total Grp1 whereas a Grp1 antibody co-precipitated Grsp1 almost quantitatively (1). This suggests that a substantial portion of Grp1 would be available to interact with other proteins. Indeed, Grp1 has been shown to interact with several other proteins including CASP (Cytohesin-1 Associated Scaffold Protein) and GRASP (Grp1 Associated Scaffold Protein (14, 15). The epitope for interaction with these proteins has been mapped to the heptad repeats of Grp1. Among Grp1, ARNO and Cytohesin-1, this region shows the highest sequence diversity (50-63% identity; Figure 1A). The higher sequence variability within the heptad repeats raises the possibility that the individual members of the Grp1 family may be capable of assembling into functionally distinct complexes.
To gain further insight into the intrinsic properties of Grsp1 complexes, we have characterized the stability, specificity, and oligomeric state of Grsp1 complexes with Grp1, Cytohesin-1 and ARNO. The results demonstrate that Grsp1 readily forms heterodimers with both Grp1 and Cytohesin-1. Whereas Grp1 heterodimers form at the expense of less stable Grp1 homodimers, ARNO homodimers are at least as stable as Grsp1-ARNO heterodimers. FRET experiments suggest that Grsp1-Grp1 and Grsp1-Cytohesin-1 heterodimers preferentially assemble in an antiparallel orientation. Formation of the Grsp1-Grp1 complex, however, does not enhance partitioning with membranes containing PtdIns(3,4,5)P3 or PtdIns(4,5)P2.
Constructs of mouse Grp1 (‘2G’ diglycine splice variant, GenBank: AF001871), human ARNO (‘3G’ triglycine splice variant, GenBank: X99753), human Cytohesin-1 (‘3G’ triglycine splice variant, GenBank: BC050452) and mouse Grsp1 (GenBank: AF327857) were amplified using Vent DNA polymerase (New England Biolabs). For Grp1 and ARNO, the human and mouse proteins are identical within the heptad repeats. Human and mouse Cytohesin-1 differ by only 2 substitutions, both of which are conservative in nature and are not located in the a or d positions of the heptad repeat. The first heptad repeat region of human and mouse Grsp1 are 94% identical with 1 substitution at the first d position and 2 substitutions at other (non a or d) positions. Amplified constructs were inserted into a modified pET15b vector (Novagen) containing an N-terminal 6xHis tag (MGHHHHHHGS) using BamHI and SalI restriction sites. The GST-Grsp1350-400 construct used for co-precipitation assays was inserted into the BamHI and SalI restriction sites of pGEX-6P1 (Amersham Pharmacia Biotech). For FRET experiments, a cysteine residue was added to either the N- or C-terminus of each protein. Constructs were verified by sequencing the coding region from both 5' and 3' directions.
Constructs were expressed in BL21(DE3)RIL cells (Stratagene) grown in 2X YT-amp (16 g bacto tryptone, 10 g bacto yeast extract, 5 g NaCl and 100 mg ampicillin per liter). Cultures were grown to OD600 of 0.4 and induced with isopropyl-1-thio-β-D-galactopyranoside for 16 hours at 20 °C. For the purification of 6xHis proteins, cells were suspended in lysis buffer (50 mM Tris-HCl, pH 8.0, 0.1% β-mercaptoethanol, 0.1 mM phenylmethylsulfonylfluoride, 1 mg/ml lysozyme, 2 mM MgCl2, 10 g/ml DNase I) and disrupted by sonication. Triton X-100 was then added to 0.5% and the cell lysate centrifuged at 35,000 g for 45 minutes. The supernatant was loaded on a Ni-NTA column (Qiagen) equilibrated with buffer (50 mM Tris-HCl, pH 8.0, 0.1% β-mercaptoethanol), washed with 20 column volumes of buffer containing 15 mM Imidazole, and eluted with a gradient from 10-150 mM Imidazole. Subsequent ion exchange chromatography on Source-S or Source-Q (GE Health Care) followed by gel filtration chromatography over Superdex-200 (GE Health Care) yielded protein preparations that were >99% pure as judged by SDS-PAGE. For GST-Grsp1350-400, the clarified supernatant was mixed with glutathione-Sepharose (GE Health Care) equilibrated with buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% β-mercaptoethanol) and nutated at 4 °C for 30 minutes. The resin was centrifuged at 500 g and washed with three times with 20 ml of buffer. The GST-fusion protein was eluted with buffer containing 10 mM glutathione. The GST-Grsp1350-400 fusion protein was >99% pure as judged by SDS-PAGE.
A 2 ml solution of 30 μM Grp113-248 was mixed with an equal volume of 30 μM Grsp12-400 in buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10% glycerol, 0.1% β-mercaptoethanol) and incubated at 4 °C for 16 hours. The sample was then filtered through a 0.2 μm acrodisc and loaded on Superdex-200 column (GE Health Care). Equal amounts of the free proteins were loaded on a Superdex-200 column under identical conditions to allow direct comparison of the elution profiles. Fractions were analyzed on 15% SDS-PAGE with coomassie blue staining. The Superdex-200 column was calibrated using a gel filtration calibration kit (GE Health Care) with ribonuclease A (13.7 kDa), Chymotrypsinogen A (25 kDa), ovalbumin (43 kDa), albumin (67 kDa), aldolase (158 kDa), catalase (232 kDa), ferritin (440 kDa) and thyroglobulin (654 kDa) as molecular weight standards.
GST-Grsp1350-400 at a concentration of 21 μM was incubated with an equivalent molar quantity of either 6xHis Grp113-399, 6xHis ARNO2-400 or 6xHis Cytohesin-12-398 in buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% β-mercaptoethanol) at 4 °C for 2 hours on a nutator. Glutathione-sepharose beads were added to the protein mixture and nutated for 1 hour at 4 °C. After centrifugation at 500 g, the supernatant was collected, the pellet washed 3 times with 100 μl buffer, and the beads nutated for 1 hour at 4 °C in buffer containing 10 mM glutathione. The supernatants and bead elutions were analyzed by 15% SDS-PAGE with coomassie blue staining. Controls with GST substituted for GST-Grsp1350-400 were also carried out to determine the amount of non-specific binding. The gel was scanned and the integrated intensity of each band on the gel determined using the NIH Image program (http://rsb.info.nih.gov/nih-image).
Full length Grp1, Cytohesin-1 and ARNO in pEGFP-c1 and the HA-tagged brain isofrom of Grsp1 in pCMV5 were generated as described (1, 16). COS-1 cells on 15 cm dishes were co-transfected using Fugene-6 (Roche) according to the manufacturer's instructions. The cells were lysed 48 hours after transfection in ice-cold lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 25 mM sodium fluoride, 1 mM sodium orthovanadate, 1% Nonidet P-40, and 1 mM dithioerythritol) supplemented with complete EDTA-free Protease Inhibitor Cocktail Tablets (Roche). One milligram of protein lysate was incubated with 50 μl of HA antibody Agarose Immobilized conjugate (Bethyl Laboratories, S190-107) overnight at 4°C on a nutator. Immunoprecipitates captured by HA-agarose conjugate were washed in lysis buffer four times and resuspended in 2x SDS-PAGE gel loading buffer. A rabbit anti-GFP (Abcam, ab6556) and HRP conjugated goat anti-rabbit IgG (Promega, W4011) were used for immunoblotting with chemiluminescence based detection.
6xHis tagged constructs of individual proteins and 1:1 complexes were dialyzed overnight against buffer (10 mM Tris-HCl, pH 8.0, 200 mM NaCl) and centrifuged to equilibrium at 20 °C in an Optima XL1 analytical ultracentrifuge (Beckman instruments). Sedimentation analysis was carried out at various protein concentrations. The absorbance at 280 nm was measured as a function of radial distance (r) from the axis of rotation using the dialysis buffer as a blank. The abscissa was transformed as σm (r02 – r2) / 2, where r0 was taken as the data point furthest from the axis of rotation and σm was calculated with SEDINTERP (17) using the monomer molecular mass for each construct. Data were fit to the function
where C0 and Ci are constants and ni represents the order of the ith oligomeric species. In the case of the Grsp1 complexes, σm was calculated using the molecular mass for the heterodimeric complex.
6xHis tagged proteins corresponding to the heptad repeat region of Grp1, ARNO, Cytohesin-1 and Grsp1 were purified as described above. A single cysteine was added at either the amino (MGHHHHHHGSC-peptide) or carboxyl (MGHHHHHHGS-peptide-C) terminal ends to facilitate selective labeling of either termini. Prior to labeling, peptide samples were exhaustively dialyzed against 50 mM sodium borate, pH 7.5, 100 mM NaCl for 24-36 hours to remove 2-mercaptoethanol present in the storage buffer. The single cysteine in each peptide was labeled with Alexa Fluor-546 (donor) or Alexa Fluor-647 (acceptor) by dilution of stock protein solution to 100-150 μM in buffer (50 mM sodium borate, pH 7.5 or 7.0, 100 mM NaCl). Potential disulfide bonds were reduced by addition of a 10-fold molar excess of TCEP (tris-(2-carboxyethyl)phosphine, Pierce Scientific) to the samples. The stock Alexa Fluor dye solution in dimethylsulfoxide was diluted to 10 mM and added to the reduced peptide solution to give a 1.1 fold molar excess of the dye. The reaction was carried out at 25 °C for 2 hours or at 4 °C overnight. During labeling steps, samples were protected from light. Labeling was terminated by addition of β-mercaptoethanol to 0.1%. Free dye was separated from labeled peptide using a D-salt gel filtration column (Pierce Scientific) followed by exhaustive dialysis against 50 mM sodium borate, pH 7.5 or 7.0. The labeling efficiency was determined by comparison of the absorbance at the λmax of the dye and the amount of peptide present determined by amino acid analysis. Labeling efficiencies of 85% for Grsp1-N and 50% for Grp1 and Cytohesin-1 peptides were obtained.
FRET measurements were carried out using a Tecan Safire microplate spectrometer and Corning half area 96 well microplates. Alexa Fluor labeled peptides were mixed in 1:1 stoichiometric amounts (2.5 μM each) in buffer (10 mM Tris, pH 8.0, 100 mM NaCl). The optimal excitation wavelength was determined to minimize direct excitation of the acceptor Alexa-647 while maximizing the donor emission. Bandwidths of 5.0 nm were used for both the excitation and emission monochromators to minimize bleed through of donor fluorescence into the acceptor channel. The emission spectra of the donor, in the presence and absence of acceptor, was used to calculate the observed FRET efficiency (FEobs) as
where Ida is the donor emission intensity of the sample containing both donor and acceptor and Id is the emission intensity of the donor alone. Assuming the labeled and unlabeled peptides bind with equal affinity, the observed FRET efficiency was corrected for the labeling efficiency by
where LEd and LEa are the labeling efficiency for the donor and acceptor, respectively. Predicted distances were determined using the EEA1 coiled-coil structure as a model (PDB ID code 1joc). The uncertainty in the predicted distances was estimated by measuring the distance from the cysteine linker to the center of the fluorophore in an extended conformation.
Small unilamellar vesicles (SUVs) were prepared by mixing lipid stocks in chloroform in the appropriate molar ratios followed by drying to form a lipid film. In order to facilitate efficient sedimentation of liposomes, 1,2-distearoyl(dibromo)-sn-glycero-3-phosphocholine was used. Lipids resuspended in assay buffer (50 mM Tris, pH 8.0, 200 mM KCl, 1 mM MgCl2) were frozen in liquid nitrogen and thawed for 10 cycles followed by bath sonication for 30 minutes and extrusion through polycarbonate membranes with a 100 nm pore size. Partitioning assays were carried out with 2 μM protein or protein complex and varying amounts of total lipid. Reactions were incubated at 25 °C for 1 hour followed by centrifugation at 100,000 g for 1 hour at 25 °C. Vesicle pellets were suspended in SDS sample buffer in a volume equal to the supernatant. Samples were analyzed on SDS-PAGE and the amount of bound and free proteins determined using the program GelEval. Each band was enclosed with a rectangle and the intensity of all the pixels in the rectangle integrated after background subtraction. The background was determined by averaging the pixel intensity above and below the band of interest. The percent pelleted was calculated as the integrated intensity of the pellet divided by the sum of the integrated intensity in the supernatant and pellet.
At protein concentrations in the low micromolar range, Grp1 and ARNO form stable homodimers whereas Cytohesin-1 is predominately monomeric (16, 18, 19). Homodimerization requires the heptad repeats, which also mediate heteromeric interactions with other proteins. In the case of Grp1, heteromeric complexes with Grsp1 have been detected in vitro and in cells (1). The region of Grsp1 that binds to Grp1 has been mapped to the first of two heptad repeat regions C-terminal to the FERM domain and has a high propensity to form a coiled-coil structure (12) as do the heptad repeats in Grp1, ARNO and Cytohesin-1. However, the stability of the heteromeric complexes compared with Grp1 homodimers is poorly characterized as is the specificity for Grp1 family proteins.
As one approach to determine the specificity of Grsp1 for the Grp1 family, a 6xHis Grsp1 construct spanning the N-terminal FERM domain and first heptad repeat region (Grsp12-400) was incubated with an equivalent molar quantity of a 6xHis Grp1, ARNO or Cytohesin-1 construct consisting of the heptad repeats and Sec7 domain (hr-Sec7). After incubation for 16 hours at 4 °C, the protein mixtures were analyzed by gel filtration chromatography (Figure 1; see also Figure 4 for calculated molecular weights). The hr-Sec7 construct was used for these experiments to allow homodimers to be clearly distinguished from heterodimers. In the absence of other proteins, Grsp12-400 and Cytohesin-1 elutes as a monomeric species whereas Grp1 and ARNO have elution volumes larger than expected for a globular homodimer but less than expect for a globular homotrimer, consistent with an elongated homodimeric species. After incubation, Grp1 and Grsp12-400 co-elute as a single peak with an elution volume less than that of either protein alone and in the expected range for a heterodimer. A similar elution profile is observed after incubation of Grsp12-400 and Cytohesin-1. The small tails in the elution profiles likely represent residual free species and could be due to incomplete formation of heteromeric complexes. In the case of the ARNO construct, on the other hand, two overlapping peaks are observed. The first peak contains both Grsp12-400 and ARNO and has an elution volume corresponding to that of the apparent heterodimeric species observed for the Grp1 and Cytohesin-1 complexes. The second peak contains predominately Grsp12-400 and has an estimated elution volume similar to that of the free monomer. Using Gaussian functions to model the individual peaks in the elution profiles and calculated extinction coefficients, it was estimated that approximately 50% of the Grsp12-400 is bound to ARNO whereas greater than 90% is bound to Grp1 and Cytohesin-1.
As an alternative method to analyze the specificity of Grsp1 for Grp1 family proteins, a Grsp1 construct containing the first heptad repeat region (Grsp1350-400) fused to GST was used to co-precipitate Grp1, ARNO, or Cytohesin-1 constructs containing the heptad repeats and Sec7 domain. Stoichiometric amounts of either Grp1, ARNO or Cytohesin-1 were mixed with GST-Grsp1350-400 and incubated overnight at 4 °C. At a concentration of 21 μM, the majority of Grp1 and at least 50% of Cytohesin-1 co-precipitate with GST- Grsp1350-400 whereas the majority of ARNO remains in the supernatant (Figure 2A). The extent of co-precipitation in all three cases is not significantly altered by incubation for 2 or 24 hours at 25 °C or 2 hours at 37 °C, suggesting the binding reactions have reached equilibrium within the 2 hour incubation period (Figure 2B). Thus, the relative amounts of Grsp1 complex formed is unlikely to be due to a kinetic effect limited by the dissociation rate of the homodimers.
As a third approach, the full length HA-tagged brain isoform of Grsp1 (HA-Grsp1) was used to co-immunoprecipitate EGFP fusions of Grp1, Cytohesin-1 or ARNO following co-transfection of COS-1 cells. As shown in Figure 2C, the EGFP fusions of all three proteins were detected in immunoprecipitates with HA-Grsp1. However, the amount of EGFP-ARNO in the immunoprecipates was consistently lower than that of EGFP-Grp1 and EGFP-Cytohesin-1, even though EGFP-ARNO appeared to express at a higher level. Taken together, the gel filtration, co-precipitation and co-immunoprecipitation experiments indicate that: i) Grsp1-Grp1 heterodimers are considerably more stable than Grp1 homodimers; ii) Cytohesin-1 can form stable heterodimers with Grsp1; and iii) ARNO homodimers are at least as stable if not moderately more stable than Grsp1-ARNO heterodimers.
The oligomeric state of the individual proteins and heteromeric complexes was further analyzed by sedimentation equilibrium experiments in the concentration range from 9-18 μM. In an earlier study, we noted that the catalytic activity of Grp1, ARNO and Cytohesin-1 constructs as well as a Grsp1-Grp1 complex, does not correlate with a qualitative assessment of the oligomeric state consistent with sedimentation equilibrium experiments (16). Here, we present the data for the sedimentation equilibrium experiments along with quantitative analyses of oligomeric state and comparisons with the calculated models for monomeric, dimeric and heterodimeric species. We also extend the experiments and analyses to include Grsp1 alone and in combination with ARNO and Cytohesin-1. At concentrations similar to those used in the gel filtration and co-precipitation experiments, constructs of Grp1 and ARNO that include the heptad repeats are uniformly dimeric whereas the analogous Cytohesin-1 constructs have an equilibrium distribution consistent with a mixture of monomers and dimers (Figures 3, ,44 and S1). Grsp12-400, which includes the FERM domain and first heptad repeat region, is uniformly monomeric as is a shorter construct corresponding to the FERM domain alone. The Grsp12-400 complexes with the Grp113-248 and Cytohesin-12-244 spanning the heptad repeats and Sec7 domain centrifuge with a nearly uniform size distribution close to the predicted model for an ideal heterodimeric species and clearly distinguishable from Grsp12-400 and Cytohesin-12-244 monomers as well as Grp113-248 dimers (Figures 3 and and4).4). Conversely, the solution containing Grsp12-400 and the ARNO2-252 construct has a size distribution in the range expected for ARNO2-252 homodimers. Note that Grp1, ARNO and Cytohesin-1 constructs lacking the linker, PH domain and C-terminal helix were used for the sedimentation equilibrium experiments with Grsp12-400 so that heterodimeric complexes could be distinguished from monomers and homodimers.
To determine the orientation of the Grsp1-Grp1 and Grsp1-Cytohesin-1 coiled coils, FRET experiments were employed using Alexa Fluor labeled peptides corresponding to the heptad repeat regions. For these experiments, the donor fluorophore (Alexa 546) was covalently attached to a single cysteine residue at the N- or C-terminus of Grp1 and Cytohesin-1 whereas the acceptor fluorophore (Alexa 647) was attached to a single cysteine residue at the N-terminus of Grsp1. As expected for an antiparallel orientation, the donor quenching as well as sensitized emission were considerably larger when the donor was attached to the C-terminus of the Grp1 or Cyothesin-1 heptad repeats (Figure 5A and 5B). To further analyze the FRET data, predicted donor-acceptor distances were estimated using the coiled coil region in the crystal structure of the EEA1 C-terminus (20) as an approximate model (Figure 5C). Based on the estimated distances, the observed FRET efficiency was compared to the theoretical 1/R6 distance dependence calculated with a Forster radius of R0 = 74 Å for the Alexa-546/647 donor/acceptor pair (Figure 5D). Taking into account the experimental error in the observed FRET efficiency (vertical bars) and the uncertainty in the estimated distances (horizontal bars), the data are most consistent with either an antiparallel orientation or a mixed population with a predominately antiparallel orientation.
Recruitment of Grp1 to the plasma membrane in response to insulin stimulation requires binding of PtdIns(3,4,5)P3 to the PH domain (21). In CHO-T cells, Grp1 and Grsp1 co-localize to membrane ruffles in response to insulin stimulation (1); however, it is unclear whether Grsp1 can partition with membranes in the absence of Grp1 or whether formation of the complex with Grsp1 affects membrane partitioning of Grp1. In addition, FERM domains in proteins such as Ezrin and Radixin have been shown to bind PtdIns(4,5)P2 (22-24). Homology modeling suggests that the Grsp1 FERM domain conserves some of the basic residues implicated in the interaction with the PtdIns(4,5)P2 head group in the crystal structure of the Radixin FERM domain in complex with Ins(1,4,5)P3 (23).
The ability of the Grsp1 FERM domain to bind Ins(1,4,5)P3 or Ins(1,3,4,5)P4 was assessed by isothermal titration microcalorimetry (ITC) at a protein concentration of 40 μM. Under the conditions of these experiments, no detectable binding was observed (Figure S2). Although it is hypothetically possible that binding occurs without a significant change in enthalpy, and would therefore be difficult to detect by ITC, such entropically driven binding events are more typical of protein-protein interfaces that bury a substantial quantity of non-polar surface area rather than the well characterized high affinity binding modes for polyphosphoinositides, which primarily involve ionic/polar interactions between basic residues and phosphate groups. To determine whether oligomeric state or heterodimerization with Grsp1 directly influence head group binding, ITC was used to measure the affinity of Ins(1,3,4,5)P4 for Grp1 homodimers or the Grp1 complex with Grsp12-400 (Figure S2 and Table S1). In both cases, the dissociation constant (Kd) is similar to that of monomeric Grp1 constructs lacking the heptad repeat region (Table S1) and is also similar to Kd values (27.3 nM and 32.2 nM) reported previously for the isolated PH domain (6, 25).
The ITC experiments suggest that Grsp12-400 has weak if any affinity for the head groups of PtdIns(4,5)P2 and PtdIns(3,4,5)P3; however, it is possible that a membrane environment may be required. For example, FYVE domains have weak affinity for the head group of PtdIns(3)P (Kd ~30 μM) and even lower affinity for soluble, short chain PtdIns(3)P analogs yet partition strongly with liposome membranes in a PtdIns(3)P dependent manner. To address this possibility, the partitioning of Grsp12-400, Grp113-399, and the Grp113-399-Grsp12-400 complex with phospholipid vesicles was evaluated using a sedimentation assay. Whereas Grp113-399 partitions efficiently (~90% at 1.25 mM total phospholipid) with liposomes containing 20% PtdSer, 10% PtdIns(4,5)P2, and 3% PtdIns(3,4,5)P3, the majority of Grsp12-400 (~80-85% at 1.25 mM total phospholipid) remains in the soluble fraction in both the presence and absence of PtdIns(4,5)P2 (Figures 6 and S3). The Grsp12-400-Grp113-399 complex also partitions efficiently (~80-90% at 1.25 mM total phospholipid), although the fraction in the pellet appears somewhat reduced compared with Grp1 alone. Likewise, Grsp1 does not enhance partitioning of Grp1 with liposomes containing PtdIns(4,5)P2 but not PtdIns(3,4,5)P3.
The highly homologous proteins Grp1, ARNO and Cytohesin-1 share 80-85% identity overall and >95% identity within the Sec7 and PH domains. Consistent with the high degree of sequence similarity, the proteins are effectively indistinguishable with respect to the exchange activity of the Sec7 domain and phosphoinositide recognition by the PH domain. Nevertheless, there is evidence of functional divergence. For example, a splice variant that differs only by the insertion of a single glycine residue in the β1/β2 loop of the PH domain has markedly reduced affinity for PtdIns(3,4,5)P3 (26). In brain, Grp1 is predominately expressed as the high affinity diglycine variant, whereas ARNO and Cytohesin-1 are predominately expressed as the low affinity triglycine variant. In cells treated with the phorbal ester PMA, Cytohesin-1 and ARNO but not Grp1 are phosphorylated at PKC sites in polybasic region (27, 28). In addition to variations in alternative splicing and post-translation modification, sequence diversity within the heptad repeats (50-63% identity) may also contribute to functional differences. A classic example of this would be the b/ZIP family of transcription factors which utilize hetero and homo-oligomerization of a coiled-coil region to precisely modulate the transcriptional activity of target genes (29-31).
In this study, stable complexes of Grsp1 with Grp1 or Cytohesin-1 were isolated by gel filtration chromatography following incubation of stoichiometric amounts of the purified proteins. Conversely, only a fraction of ARNO (50% or less) was associated with Grsp1 and this fraction could not be increased by prolonged incubation. Although Grsp1 and Cytohesin-1 are monomeric under the conditions of these experiments, Grp1 and ARNO are uniformly homodimeric in the absence of Grsp1. Thus, whereas Grp1 homodimers readily re-equilibrate with Grsp1 to form more thermodynamically stable heterodimers, this is not the case for Grsp1-ARNO heterodimers, which are approximately as stable as ARNO homodimers. The substantially greater thermodynamic stability of Grsp1-Grp1 heterodimers compared with Grp1 homodimers provides a plausible thermodynamic explanation for the previously reported observation that Grsp1-Grp1 complexes could be detected by co-immune precipitation even though Grsp1-Grp1 homo-oligomers were not (1). The observed differences in the interaction of Grsp1 with ARNO might in principle reflect greater stability of the ARNO homodimer, lower stability of the Grsp1-ARNO heterodimer, or a combination of both. In any case, it is noteworthy that the sequence divergence in the heptad repeat region is greater between Grp1 and ARNO (52% identity) than between Grp1 and Cytohesin-1 (63% identity). Most of the amino acid substitutions occur in the C-terminal half of the heptad repeat region and involve three of the hydrophobic residues in the predicted a and d positions, which are important for forming the helix interface between coiled-coils (29, 32). Finally, given that our studies employed a truncation construct of Grsp1 previously shown to be sufficient for Grp1 binding (but not tested with respect to ARNO or Cytohesin-1), it is possible that other regions of Grsp1 could contribute to the stability of complexes with ARNO or Cytohesin-1. However, the results of the co-immunoprecipitation experiments with the full length proteins argue against this possbility.
Several proteins have been reported to associate with the heptad repeat regions of Grp1 family proteins, including Grsp1, CASP and GRASP (1, 14, 15). In general, the stoichiometry and quaternary structure of these complexes has not been characterized. We have found that Grsp1 forms heterodimers with Grp1 and Cytohesin-1. The results of FRET experiments are consistent with an antiparallel coiled coil arrangement or a mixture of species with a predominately antiparallel arrangement. This arrangement would in principle place the FERM domain of Grsp1 in proximity to the Sec7 and PH domains of Grp1; however, formation of the Grsp1 complex does not appear to directly affect the functional properties of either domain.
FERM domains in some proteins have been shown to function as membrane targeting modules. In Radixin, the head group of PtdIns(4,5)P2 binds at the interface between two of the three subdomains which comprise the FERM domain (33, 34). The head group binding site is located on an extended flat surface with a positive electrostatic potential that may contribute to membrane partitioning through non-specific interactions with negatively charged phospholipids. Despite conservation of roughly 50% of basic residues that contact the PtdIns(4,5)P2 head group in the Radixin FERM domain, the Grsp1 FERM domain shows no indication of binding to either Ins(1,4,5)P3 or Ins(1,3,4,5)P4 by ITC. Consistent with the absence of detectable head group binding, the partitioning of Grsp1 with liposomes is not enhanced by inclusion of 10% PtdIns(4,5)P2 in the presence of absence of 3% PtdIns(3,4,5)P3. Although a small fraction of Grsp1 was observed in the membrane fraction in both the presence and absence of PtdIns(4,5)P2, indicative of a weak intrinsic ability to partition with acidic liposomes, Grsp1 did not appear to enhance the partitioning of 2G Grp1. However, it remains possible that Grsp1 could enhance partitioning with the 3G variants of Grp1 family proteins, which have considerably lower affinity for PtdIns(3,4,5)P3 (21), or perhaps contribute to alternative targeting mechanisms described for Grp1 family proteins, including Arf6 and Arl4 dependent plasma membrane recruitment (35, 36).
Finally, the FERM domains of Ezrin, Radixin and Moesin have been shown to bind polybasic regions in the cytoplasmic tails of integral membrane proteins (13). In contrast, we find no evidence for interaction of the Grsp1 FERM domain with the polybasic region of Grp1 (J.D., unpublished observations). Likewise, formation of the complex with Grsp1 does not affect the ability of the polybasic region to inhibit the exchange activity of the Sec7 domain (16). In view of these observations, it seems likely that the Grsp1 FERM domain will have functions that are yet to be identified, possibly including interactions with polybasic regions in proteins other than Grp1.
Supporting Information Available
Figure S1 - Sedimentation equilibrium experiments for Grp1, ARNO and Cytohesin-1 constructs lacking the heptad repeat region.
Figure S2 - ITC analysis of Ins(1,3,4,5)P4 and Ins(1,4,5)P3 binding to Grp1, Grsp1 and the Grsp1-Grp1 complex.
Figure S3 - Sedimentation of Grp1, Grsp1 and Grsp1-Grp1 complexes with liposomes.
Table S1 - Dissociation constants for inositol polyphosphate binding to Grp1, Grsp1 and the Grsp1-Grp1 complex.
We thank Kim Crowley for assistance with analytical ultracentrifugation, Michael Czech for use of cell culture equipment, and Xiaochu Chen, Jes Klarlund, Tse-Chun Kuo, Xiarong Shi, Qionglin Zhou for advice on co-immunoprecipitation experiments.
This work was supported by NIH Grant DK60564.
This material is available free of charge via the Internet at http://pubs.acs.org.