Molecular characterization of Rab11 interactions with members of the family of Rab11-interacting proteins.

The Rab11 subfamily of GTPases plays an important role in vesicle trafficking from endosomes to the plasma membrane. At least six Rab11 effectors (family of Rab11-interacting proteins (FIPs)) have been shown to interact with Rab11 and are hypothesized to regulate various membrane trafficking pathways such as transferrin recycling, cytokinesis, and epidermal growth factor trafficking. In this study, we characterized interactions of FIPs with the Rab11 GTPase using isothermal titration calorimetric studies and mutational analysis. Our data suggest that FIPs cannot differentiate between GTP-bound Rab11a and Rab11b in vitro (50-100 nm affinity) and in vivo. We also show that, although FIPs interact with the GDP-bound form of Rab11 in vitro, the binding affinity (>1000 nm) is not sufficient for FIP and GDP-bound Rab11 interactions to occur in vivo. Mutational analysis revealed that both the conserved hydrophobic patch and Tyr628 are important for the GTP-dependent binding of Rab11 to FIPs. The entropy and enthalpy analyses suggest that binding to Rab11a/b may induce conformational changes in FIPs.

, and phagocytosis (9). Whereas Rab11a and Rab11b are ubiquitously expressed, Rab25 is present exclusively in epithelial cells (10,11), hence its proposed role in regulating membrane trafficking pathways that are specific to epithelial cells. The differential roles of the Rab11a and Rab11b isoforms remain unclear. Although several Rab11-binding proteins have so far been isolated (12)(13)(14)(15), the differences in their binding to Rab11a and Rab11b isoforms have not been investigated.
Rab proteins cycle between inactive GDP-and active GTPbound forms. In the GTP-bound form, they interact with effector proteins, and each Rab⅐effector complex is proposed to regulate a unique trafficking step/event such as vesicle docking, budding, transport, or fusion (3,16). Recently, several Rab11 effector proteins, Rip11, FIP2, 1 Rab-coupling protein (RCP), FIP3/eferin, and FIP4, were identified using biochemical and yeast two-hybrid methods (12)(13)(14)(15). In addition to Rab11 binding, FIP3 and FIP4 also interact with ADP-ribosylation factor GTPases, thereby coupling these small GTPase families, allowing for potential cross-talk between two signaling pathways (17). All FIPs have a conserved C-terminal motif that is known as the Rab11/25-binding domain (RBD) (18). Based on the presence of additional structural domains, FIPs are classified into three groups: class I FIPs (Rip11, FIP2, and RCP) containing a C2 domain, class II FIPs (FIP3/eferin and FIP4) containing EF-hand motifs, and the class III FIP (FIP1) with no homology to known protein domains (reviewed in Ref. 19).
Despite the identification of the RBD that mediates the interaction of FIPs with Rab11, we are only beginning to understand the mechanisms of Rab11 and FIP interactions. It remains unclear whether different FIPs exhibit preferences for interaction with Rab11a or Rab11b isoforms. We still do not know whether these multiple effectors are capable of binding to Rab11a or Rab11b with affinities that are relevant to their physiological concentrations in cells. Interestingly, some recent work suggests that FIP2 and RCP bind to Rab11a in a GTPindependent manner (14,20). In addition, although several effector proteins have been identified for Rab11, little information is available on the affinities and thermodynamic properties of Rab11⅐effector complex formation. Thus, to understand the properties of Rab11 and FIP binding, we have performed isothermal titration calorimetric (ITC) analysis on the interaction of class I FIPs with GDP-and Gpp(NH)p-bound forms of Rab11a and Rab11b. ITC analysis has been used previously to study several protein-protein and protein-ligand interactions, as it directly measures enthalpy and the binding constant involved (22,23). Our data indicate that all class I FIPs are capable of binding to the Gpp(NH)p-bound form of Rab11 with similar affinity. A weak interaction was also seen with the GDP-bound form of Rab11, although it was not sufficient to mediate GDP-bound Rab11 and FIP interactions in vivo. FIPs bound equally well to both isoforms of Rab11 in vitro and in vivo, suggesting that Rab11a and Rab11b mediate redundant and overlapping functions. The free energy (⌬G) for Rab11⅐FIP complex formation is contributed by enthalpic factors and exhibits a strong enthalpy-entropy compensation effect. Mutational analysis of the conserved amino acids in the RBD of Rip11 indicates that the hydrophobic patch is crucial for Rab11⅐FIP complex formation. In addition to these conserved hydrophobic residues, Rip11 Tyr 628 also plays an important role in FIP association with Rab11. Finally, our data suggest that FIPs form a high affinity homodimer, which interacts with two Rab11 molecules to form a heterotetramer.
Site-directed Mutagenesis and Expression and Purification of Proteins-Site-specific mutations were introduced into the pGEX-Rip11-F1 or pEGFP-Rip11-F1 construct using a Stratagene PCR-based mutagenesis kit. The GST-Rab11, GST-FIP2-F1, GST-RCP-F1, and GST-Rip11-F1 constructs were transformed into Escherichia coli BL21 Codon Plus (Stratagene). GST fusion proteins were expressed and purified from E. coli as described previously (24). The Gpp(NH)p-or GDP-bound forms of Rab11a and Rab11b proteins were prepared by processing through a series of nucleotide (Gpp(NH)p or GDP) exchange reactions in the presence of EDTA as described previously for Rab5 GTPase (25). Full-length FIPs were cloned into the baculovirus expression plasmid, and the GST fusion proteins were purified from insect cells using glutathione affinity column chromatography. Protein concentrations were determined by the Bradford assay.
In Vitro Binding Assays-Glutathione beads (50 l) were coated with 5 g of GST fusion protein and incubated with varying amounts of soluble protein in a final volume of 0.5 ml of reaction buffer (50 mM HEPES (pH 7.4), 150 mM NaCl, 5 mM MgCl 2 , 0.1% Triton X-100, 0.1% bovine serum albumin, and 1 mM phenylmethylsulfonyl fluoride) in the presence of either GTP␥S or GDP␤S (0.5 mM) as indicated. Samples were incubated at 4°C for 1 h on a nutator with constant rotation. The samples were pelleted at 2000 ϫ g for 3 min and washed three times with 1 ml of reaction buffer. Bound proteins were eluted with 1% SDS, analyzed by SDS-PAGE, and either stained with Coomassie Blue or immunoblotted.
Gel Filtration Analysis-The oligomeric status of Rab11, Rip11-F1, and Rip11-F1 mutants and the Rab11⅐Rip11-F1 complex was determined by analyzing the proteins on a Superdex S200 column that was connected to a fast protein liquid chromatograph equipped with a UV280 monitor and a mini DAWN detector (Wyatt Technology Corp.). The native molecular mass of the protein was calculated from light scattering analysis using the software provided (Wyatt Technology Corp.).
ITC-ITC experiments were performed using a VP-ITC calorimeter (Microcal LLC, Northampton, MA) as recommended by the manufacturer. Rab11 (8 M) was loaded in the sample cell (in phosphatebuffered saline containing 5 mM MgCl 2 and 0.5 mM Gpp(NH)p or GDP; 1.426-ml volume) and titrated with Rip11-F1, FIP2-F1, GST-RCP-F1, and Rip11-F1 proteins in the same buffer (5-l injections up to total of 40 -50 injections). The titrations were performed while samples were stirred at 300 rpm at 25°C or at the indicated temperature. An interval of 4 min between each injection was allowed for the base line to stabilize. The blank ITC titration was performed against buffer by injecting the corresponding FIP that was used in Rab11 titration. The blank subtraction was done for all data used for analysis. The data were fitted via the one-set-of-sites model to calculate the binding constant (K) using Origin software (Microcal, LLC). All proteins used in the ITC studies were either in the thrombin-cleaved form (Rab11, Rip11-F1, and FIP2-F1) or in the GST fusion form (GST-RCP-F1). The presence of GST did not interfere with the binding studies.
Cell Culture, Transient Transfection, and Immunofluorescence-HeLa cells were cultured as described previously (18). The cells were transfected with wild-type and mutant GFP-Rip11-F1 constructs using LipofectAMINE 2000 (Invitrogen). At 18 -24 h post-transfection, the cells were plated on collagen-coated glass coverslips, grown overnight, and fixed with 4% paraformaldehyde, followed by quenching with 0.1 M glycine. The cells were permeabilized with phosphate-buffered saline containing 0.4% saponin, 2% fetal bovine serum, and 1% bovine serum albumin for 30 min, followed by incubation with rabbit anti-Rip11 or anti-Rab11 polyclonal antibody (Zymed Laboratories Inc.). After extensive washing, cells were incubated with Alexa 594-conjugated secondary antibodies (Molecular Probes, Inc.) for 30 min, washed, and mounted with Vectashield (Vector Laboratories). Cells were imaged with a Zeiss Axiovert 200M deconvolution microscope. RNA Interference Analysis-Rab11a and Rab11b isoforms were knocked down using small interfering RNAs (siRNAs) that were designed using human Rab11a and Rab11b sequences (Rab11a, 5Ј-aatgtcagacagacgcgaaaa-3Ј; and Rab11b, 5Ј-aagcacctgacctatgagaac-3Ј). Rab11a siRNA, Rab11b siRNA, or a mixture of Rab11a and Rab11b siRNAs was cotransfected into HeLa cells using LipofectAMINE 2000. Transfected cells were incubated for 74 h and analyzed for Rab11a and Rab11b expression by Western blotting or used for subcellular fractionation. The remaining cells were plated on collagen-coated glass coverslips, fixed, and analyzed by immunofluorescence microscopy using a Zeiss Axiovert 200M inverted deconvolution microscope. Anti-Rab11 primary antibodies were labeled with Alexa 488 and anti-Rip11 primary antibodies with Alexa 594 using a Zenon labeling kit (Molecular Probes, Inc.) and were used for staining HeLa cells.
FACS-based Tf uptake and recycling assays were done as described previously (36). Briefly, for Tf uptake assays, the cells were incubated at 4°C for 30 min with 20 g/ml Alexa 647-conjugated Tf, followed by incubation at 37°C for various time intervals in the continuous presence of Alexa 647-conjugated Tf. To measure plasma membrane-associated transferrin receptor (TfR), cells were washed and analyzed after incubation at 4°C. In all cases, the cells were washed and terminated by pelleting and resuspending the cells in 3% paraformaldehyde. Cellassociated Alexa 647-conjugated Tf was determined by FACS analysis. For recycling assays, cells were incubated at 4°C for 30 min, followed by internalization for 20 min at 37°C in the continuous presence of Alexa 647-conjugated Tf. Cells were then washed and incubated in complete medium supplemented with 50 g/ml unlabeled Tf for various times prior to fixation. The experiment was terminated as described above and analyzed on a BD Calibur flow cytometer (BD Biosciences) equipped with 488-and 647-nm lasers, gating for transfected cells (10,000 GFP-positive cells), and the amount of Tf internalized was determined.

Rab11 Effector Proteins (FIPs) Form Homodimers-It has been previously suggested that FIPs can form homodimers and
heterodimers, yet it remains unclear whether dimerization is dependent on Rab11 binding (26). To test this, we produced recombinant full-length GST-FIP fusion proteins in insect cells and truncated FIPs in E. coli and tested their ability to form homodimers or heterodimers with other FIP family member proteins in the presence or absence of Rab11a. As shown in Fig.  1, soluble full-length Rip11 bound to GST-Rip11. The binding was specific to Rip11 (Fig. 1A), as there was no binding observed with GST alone or with other FIP family member proteins, FIP2 (Fig. 1B) or FIP3 (Fig. 1A). Similarly, FIP2 and FIP3 interacted with only FIP2 and FIP3, respectively (Fig. 1,  A and B), indicating that FIPs form only homodimers in vitro. Rab11a had no influence on FIP homodimerization, as FIPs formed homodimers in both its presence and absence (Fig. 1, A  and B). Deletion of the N-terminal region (amino acids 1-490) of Rip11 had no effect on the Rip11-Rip11 interaction, whereas deletion of the C-terminal 28 amino acids completely abolished this interaction, indicating that the C-terminal region including the RBD (amino acids 630 -652) is very important for homodimer formation (Fig. 1A). These results were further supported by cross-linking experiments showing that the Rip11-F1 protein formed a dimer in solution (data not shown). To confirm that Rip11 forms a homodimer, we performed gel filtration followed by light scattering analysis on Rip11-F1 (amino acids 490 -652) alone or complexed with Rab11 (Fig. 2). Consistent with cross-linking results, Rip11-F1 eluted as a dimer (36 kDa), whereas Rab11 eluted as monomer (24.3 kDa) and the Rab11⅐Rip11 complex as a heterotetramer of 85 kDa (Rip11-F1 dimer ϩ two Rab11 molecules).
Class I FIPs Form Strong GTP-dependent Complexes with Rab11a and Rab11b-The binding affinity and thermodynamics of the interactions between Rab11 and FIP complexes were characterized using ITC analysis. Fig. 3 (A and B, upper panels) shows a typical calorimetric titration of Rip11-F1 with Rab11a at 25°C in the presence of GDP and Gpp(NH)p, respectively. The size of the injection peaks decreased gradually as the injections progressed due to saturation of the binding sites. Fig. 3 (lower panels) shows the binding isotherm, in which the total heat per injection (kilocalories/mol of Rip11-F1 injected) has been plotted against the molar ratios of Rip11-F1 and Rab11a. Curve fitting the data using the identical-site model (Origin software) resulted in the following parameters: the binding constant (K B ), stoichiometry (n), enthalpy (⌬H), and entropy (⌬S). Both GDP-and Gpp(NH)p-dependent binding exhibited exothermic heat responses. The binding stoichiometry for the Rip11-F1 and Rab11a interaction was 0.5, indicating that one Rip11-F1 dimer bound to two Rab11a molecules. This agreed with the gel filtration analysis (Fig. 2). The affinity of Rip11-F1 for Gpp(NH)p-bound Rab11a (54 nM) was 17 times higher than that for GDP-bound Rab11a (950 nM). Similar results were obtained for the interaction of Rip11-F1 with Rab11b (Fig. 3, C and D). There was no exothermic heat response produced by the buffer control or with other Rab GTPases, indicating that the Rab11 and Rip11 interaction was specific and that ITC-based assays can be used to measure affinities between Rab proteins and their effector proteins (Fig.  3, E and F; and data not shown).
The ITC data suggest that Rip11 interacts with Rab11a and Rab11b in vitro with similar affinity. To examine these interactions in vivo, we used siRNAs designed against Rab11a and Rab11b isoforms. As shown in Fig. 4, knockdown of Rab11a or Rab11b alone had no effect on Rip11 subcellular distribution, indicating that Rab11b can compensate for Rab11a and vice versa. In contrast, knockdown of both Rab11 isoforms (Fig. 4, C and F) resulted in redistribution of Rip11 from endosomes to the cytosol. To confirm this, we fractionated cells into the cytosol and membranes. As shown in Fig. 5, only double Rab11a and Rab11b knockdown resulted in translocation of Rip11 from membranes to the cytosol.
To gain additional insights into the interaction of Rab11 with FIP family effectors, we extended the ITC-based study to other Rab11 effector proteins, FIP2 and RCP, belonging to the Class I group of FIPs. The interactions of FIP2 and GST-RCP with Rab11a and Rab11b were studied under conditions similar to those for the Rab11-Rip11 interaction as described above. It is evident from Table I that a strong Gpp(NH)p-dependent interaction existed between FIP2 and Rab11a/b (ϳ40 nM), whereas the interaction with GDP-bound Rab11 was weak (Ͼ1000 nM), similar to the conditions for Rab11⅐Rip11 complex formation. Similar results were also obtained for the interaction of GST-RCP with Rab11, but with slightly decreased affinity (ϳ2-3fold) compared with the Rab11-Rip11 interaction (Table I).
Rab11-FIP Interactions Are Thermodynamically Similar and Enthalpically Driven-The enthalpy changes associated with the formation of the Rab11⅐FIP complex were measured directly using ITC. The net ⌬G and entropy (⌬S) were calculated using the following equations: ⌬G ϭ ϪRT ln K and ⌬G ϭ ⌬H Ϫ T⌬S, respectively. Negative enthalpy (Ϫ12.8 kcal/mol for GDP and Ϫ21.4 kcal/mol for Gpp(NH)p) and entropy (Ϫ4.6 kcal/mol for GDP and Ϫ11.6 kcal/mol for Gpp(NH)p) values were obtained for GDP-as well as Gpp(NH)p-dependent Rab11⅐Rip11 complex formation; thus, this interaction must be driven by enthalpic rather than entropic factors, as the net favorable ⌬G is contributed by enthalpic factors (Table I). The enthalpy for Gpp(NH)p-dependent Rab11⅐Rip11-F1 complex formation was about two times higher, and the ⌬G was about Ϫ1.5 to Ϫ2.1 kcal/mol larger than that with GDP-dependent Rab11⅐Rip11 complex formation at each temperature used in the study (Fig. 6A), indicating that Rab11 binds to Rip11 in the presence of Gpp(NH)p with higher affinity. The enthalpy change upon binding showed a strong temperature dependence, which was essentially linear (r ϭ 0.99) (Fig. 6A). GDPand Gpp(NH)p-dependent Rab11⅐Rip11 complex formation both displayed a strong enthalpy-entropy compensation, resulting in a relatively constant ⌬G at all temperatures tested (Fig.  6B). The change in heat capacity (⌬C p ) values were calculated by performing linear regression analysis on a ⌬H versus T plot, yielding Ϫ0.84 and Ϫ0.97 kcal/mol/K, respectively, for both GDP-and Gpp(NH)p-dependent Rab11⅐Rip11 complex formation. The large negative ⌬C p values may be indicative of a hydrophobic component to the binding and may mean that a conformational change accompanies the association of Rip11 with Rab11.
The Conserved Hydrophobic Patch and Tyr 628 Are Important for Rab11⅐Rip11 Complex Formation-The thermodynamic analysis of Rip11 and Rab11 interactions suggested that hydrophobic amino acid residues may play an important role in Rab11⅐Rip11 complex formation. The amino acid sequence alignment of FIPs in the RBD indicates that the conserved hydrophobic residues (marked with asterisks) form a hydrophobic patch (Fig. 7A) (18). To test the importance of these hydrophobic residues, we introduced a conservative substitution of the central core isoleucine residue (Ile 629 ) with either valine or a non-conservative charged glutamic acid. All of the mutant proteins were analyzed using ITC to test their ability to bind to Rab11. Consistent with the involvement of the "hydrophobic patch" in Rab11 binding, the non-conservative replacement (I629E) abolished the Rip11 interaction completely (Ͼ500-fold decreased affinity), whereas the conservative substitution (I629V) had little effect on Rab11⅐Rip11 complex formation (Fig. 7, A and B; and Table II).
In addition to the I629E mutation, we also mutated other highly conserved residues such as Tyr 628 , Asp 630 , and Glu 638 . Interestingly, the Y628A mutant weakly interacted with the Gpp(NH)p-bound form of Rab11 (1900 nM) (Fig. 7B), whereas substitution with Phe partially rescued this effect, but still exhibited weak affinity (530 nM) compared with the wild-type Rip11-F1 protein (Fig. 7C). The Ala substitution at highly conserved negatively charged residues (Asp 630 and Glu 638 ) had no effect on Gpp(NH)p-dependent Rab11⅐Rip11 complex formation (Table II). None of the Rip11 mutants used in this study

Rab11-FIP Interactions
had any effect on dimerization (as assessed by gel filtration analysis) (data not shown). Inactive Rip11 Mutants Do Not Affect Transferrin Uptake or Recycling-The above-described ITC binding studies showed that the I629E mutation in the hydrophobic patch and an Ala mutation at the conserved Tyr 628 , a residue neighboring the hydrophobic patch, significantly reduced Rip11 interaction with the Gpp(NH)p-bound form of Rab11 in vitro. To test whether these mutations have a similar effect on Rip11 binding in vivo, we transfected wild-type and mutant GFP-Rip11-F1 constructs into HeLa cells and analyzed their colocalization with endogenous Rab11. As shown in Fig. 8 (A-C), wild-type Rip11-F1 was localized to Rab11-positive compartments, which were tubulated and aggregated due to Rip11-F1 overexpression, consistent with previously published data (18). In contrast, the Rip11-F1 mutants were predominantly cytosolic and had no effect on the cellular distribution of Rab11 (Fig. 8, D-I), indicating that these mutants failed to interact with Rab11 in vivo, consistent with their interaction in vitro. Interestingly, co-overexpression of Rab11 caused cytosolic Y628A (Fig. 8, J-L), but not I629E (data not shown), to be recruited to a Rab11 compartment. Thus, the low affinity binding of Y628A to Rab11 can be compensated by the presence of excess Rab11.
We have previously reported that Rip11-F1 has a strong dominant-negative effect on Tf uptake and recycling since the overexpression of truncated Rip11 sequesters Rab11 and inhibit the formation of native Rab11⅐FIP complexes (18,36). Here, we utilized a FACS-based assay to analyze the effects of overexpression of GFP-Rip11-F1 mutants on Tf uptake and recycling (see "Experimental Procedures"). Consistent with previous report (18,36), Rip11-F1 had an inhibitory effect on Alexa 647-conjugated Tf uptake (Fig. 9A). This inhibition is most likely due to the inhibition of TfR recycling from endosomes to the plasma membrane (Fig. 9C), thus resulting in fewer TfRs at the cell surface. Indeed, overexpression of GFP-Rip11-F1 decreased the amount of TfR at the plasma membrane (Fig. 9B). In contrast, the Rip11-F1 mutants Y628A and I629E had little effect on Alexa 647-conjugated Tf recycling (Fig. 9, A-C). These results indicate that the Rip11 mutants that bind with low affinity to Rab11 in vitro also do not interact with Rab11 in vivo and are unable to compete with endogenous FIPs for Rab11.   19). Furthermore, different FIPs appear to play distinct roles in regulating membrane traffic (12-15, 17, 27). The presence of several Rab11 effectors and two Rab11 isoforms complicates our ability to understand the roles of Rab11 and FIPs. One possible mechanism is that various FIPs differentially bind to Rab11a or Rab11b, thus regulating distinct membrane transport steps or pathways. Alternatively, FIPs may bind to both Rab11 isoforms with similar affinity, thus competing with each other for this interaction. Our ITC analysis indicates that the affinities of both Rab11 isoforms for different FIPs are nearly identical. This was further supported by RNA interference analysis, where the absence of one isoform could compensate for the other, as the knockdown of either of the Rab11 isoforms had no effect on Rip11 localization, whereas the knockdown of both isoforms redistributed Rip11 into the cytosolic fraction. Thus, FIPs appear to play a role of scaffolding proteins that may regulate Rab11 function by colocalizing them with various regulatory proteins. Consistent with this idea, different FIPs are reported to interact with distinct sets of proteins that are known to regulate various endocytic pathways. For instance, FIP2 interacts with myosin Vb, whereas FIP3 and FIP4 bind to ADP-ribosylation factor GTPases, thereby generating putative Rab11⅐FIP2⅐myosin Vb or Rab11⅐FIP3⅐ADP-ribosylation factor complexes to regulate different transport steps and/or pathways (17,27).
It has been previously reported that FIPs may form homodimers or even heterodimers (26). This raises an interesting possibility of cross-talk between different FIPs via heterodimerization. Our data did show that FIPs form homodimers in vitro. The homodimerization appears to be independent of Rab11 binding. Each of the dimer subunits can independently interact with Rab11, forming a final heterotetramer that contains two FIPs and two Rab11 molecules. Surprisingly, we saw very little evidence of heterodimerization between different FIPs. However, we cannot fully discount the possibility that FIPs heterodimerize in vivo. The functional meaning of FIP dimerization remains unclear. The dimerization of Rab11⅐FIP complexes on the opposing membranes may play a role in transport vesicle docking at its target membrane. Indeed, it has been suggested that dimerization of EEA1⅐Rab5 complexes mediates organelle docking during homotypical early endosome fusion (28 -30).
Although FIPs interacted strongly with GTP-bound Rab11 (ϳ50 nM), weaker (Ͼ1000 nM; K d ) GDP-dependent interactions were also observed, which raised the possibility that FIPs may also interact with GDP-bound Rab11 in vivo. We have determined the concentrations of Rab11, Rip11, and RCP in HeLa cells by Western blot analysis using available antibodies for these proteins. These analyses revealed that there is ϳ270 nM total Rab11, 200 nM Rip11, and 430 nM RCP present in HeLa cells. Therefore, in the cell, GDP-dependent interactions will not be favored due to weaker affinity, and the realistic concentrations of available GDP-bound Rab11 at any given time will be much lower compared with the estimated total concentration as described above. Indeed, the Rip11 mutant Y629A, which bound to GTP-bound Rab11 with an affinity comparable to that of wild-type Rip11 binding to GDP-bound Rab11, showed a predominant cytosolic distribution and was not capable of inhibiting TfR recycling. This ϳ20-fold difference in FIP binding affinity appears to be sufficient to ensure the specific interaction between FIPs and GTP-bound Rab11 in vivo. Even though we are able to determine the total Rab11 and FIP concentrations in the cell, they do not fully reflect the local effective concentrations of Rab11 and FIPs on endocytic membranes. Thus, it is possible that relatively low affinity interaction between GDP-bound Rab11 FIG. 9. Inactive Rip11 mutants suppress the inhibitory Rip11-F1 effect on Tf uptake and recycling. HeLa cells were transfected with GFP alone, GFP-Rip11-F1, GFP-Rip11-F1(I629E), or GFP-Rip11-F1(Y628A) and analyzed for Tf uptake (A), plasma membrane-associated TfR (B), and Tf recycling (C) using Alexa 647-conjugated Tf as described under "Experimental Procedures." In A, plasma membrane-bound Tf signal (as determined by cell-associated Alexa 647-conjugated Tf signal at time 0) was subtracted from the total signal. In C, to compare the rates of recycling, data are expressed as a percentage of total endocytosed (time 0) Alexa 647-conjugated Tf. and FIPs could be compensated by high concentrations of these proteins on the endosomes.
The thermodynamic properties of the formation of the Rab11⅐FIP complex described in this study showed invariably favorable enthalpies and unfavorable entropies, indicating that the FIP and Rab11 interactions are energetically driven by exothermic enthalpy. This is in contrast to most protein-protein interactions, which possess favorable entropy because of the similarity of many protein interfaces to the interiors of proteins (31). The favorable enthalpy may arise from the significant number of polar interactions such as formation of salt bridges and hydrogen bonds. On the other hand, the negative ⌬C p values for both GDP-dependent (Ϫ0.84 kcal/mol/K) and Gpp(NH)p-dependent (Ϫ0.97 kcal/mol/K) Rab11⅐Rip11 complex formation suggest the presence of significant hydrophobic interactions. This is further supported by the presence of a conserved hydrophobic patch among all FIPs. The substitution of the central core residue within this patch with a charged glutamic residue severely affected the interaction of Rab11 with Rip11. The large negative ⌬C p may also be the result of reduction in conformational entropy that occurs due to a structural change upon binding. Similarly, a large negative ⌬C p was also observed for the interaction between TCF4 and ␤-catenin (32). ␤-Catenin ligands were shown to undergo significant structural changes upon ␤-catenin binding, and the interfaces are mixture of polar and non-polar contacts (32)(33)(34)(35). Interestingly, both GDP-and Gpp(NH)p-dependent Rab11-Rip11 interactions show similar negative ⌬C p values, and this is probably true for all FIPs, as FIP2 and RCP showed a similar trend (unfavorable entropy factor), indicating that FIP, but not Rab11, may undergo conformational changes upon binding. Our thermodynamic as well as mutational analyses suggest that the Rab11-FIP interface is probably contributed by both polar as well as non-polar interactions.
The RBD in FIPs is a highly conserved domain among different species (reviewed in Ref. 19). It is predicted to exist in an ␣-helical conformation with hydrophobic and charged residues clustered on the opposite sides of the helix (18). Consistent with the involvement of hydrophobic residues in mediating Rab11 binding, mutations within the hydrophobic cluster inhibit Rip11 interaction with Rab11. In addition, mutational analysis of the highly conserved Tyr 628 indicated that this residue may contribute to both hydrophobic and polar interactions. Surprisingly, mutations of the highly conserved Asp 630 and Glu 638 had no effect on Rab11 and Rip11 interactions or on homodimerization. Even though these residues do not directly participate in Rab11 binding, they still could be important for some other functions, such as in determination of the specificity of Rab11 binding or in the Rab11-independent interaction between FIPs and other binding proteins such as myosin Vb and/or the ADPribosylation factor GTPases (17,27).
The data described in this study are the first glimpse into understanding the molecular properties of Rab11-FIP interactions, but many questions remain to be answered. Perhaps the most intriguing one is how cells regulate the timing and location of Rab11 interactions with a specific FIP molecule. Since most of the cells express several FIPs, it is unlikely that they simply compete with each other for a limited number of Rab11 molecules. The presence of multiple phosphorylation and Ca 2ϩbinding motifs suggests that additional cellular factors probably regulate Rab11 and FIP interactions in vivo. The basic characterization of Rab11 and FIP interactions allows us to start addressing these questions.