Mapping the Rap1-binding interface in KRIT1

To analyze the functional importance of Rap1 binding to KRIT1, we mapped the KRIT1 region that binds to Rap1. Affinity chromatography showed that the KRIT1 FERM domain binds to Rap1A (Figure 1C), in agreement with a previous yeast two-hybrid analysis (Serebriiskii et al., 1997). As with the interaction of full-length KRIT1, the binding of the FERM domain (F123) to Rap1A was GTP dependent (Figure 1B). Domain analyses (http://smart.embl-heidelberg.de/, http://scansite.mit.edu/motifscan_seq.phtml) suggested a potential Ras association domain in the predicted F1 subregion of the KRIT1 FERM domain. The importance of this region was confirmed by the failure of Rap1A to bind to the F23 fragment of KRIT1, which lacks the F1 subregion (Figure 1C). We also expressed the isolated F1 subregion and found that F1 alone was sufficient for Rap1A binding (Figure 1D). Thus the FERM domain of KRIT1 binds to Rap1 in a GTP-dependent manner, and the F1 subregion mediates this interaction.

Having mapped the Rap1 binding site to the F1 subdomain, we built a homology model of the KRIT1 FERM domain using the moesin FERM domain structure (PDB 1EF1) as a template. In this model (Figure 2A), the predicted F1 subdomain (orange) contained a five-stranded β sheet interrupted by two α helices and structurally resembled the c-Raf RBD (Nassar et al., 1995). The model was subjected to conjugate energy minimization in a water-bound environment in the presence of Rap1 (Figure 2B). The model had good structure-packing quality (Z score of 0.83), indicating that the predicted structure is compatible with the sequence of the KRIT1 FERM domain. The major predicted binding interface between Rap1 (blue) and KRIT1 (orange) involved the β2 strand from Rap1 switch 1 domain, in agreement with our previous results (Figure 1B). In KRIT1 F1 subdomain, the β2 strand and the preceding loop are in close contact with Rap1A, and Arg452 of KRIT1 is proximal to Rap1A Glu-37 and Asp-38 residues. We had implicated these acidic Rap1A residues in the capacity of Rap1A to bind to KRIT1 (Figure 1B), lending additional credence to the model. KRIT1 Arg-452 corresponded to c-Raf Arg-89, a residue critical for Rap1 binding (Nassar et al., 1995).

Open in a separate window

FIGURE 2:

KRIT1 FERM domain homologous modeling reveals R452 residue mediating KRIT1–Rap1 interaction. (A) Homology model of KRIT1 FERM domain. Residues 410–736 from KRIT1 were modeled using the moesin crystal structure as the template. The modeled structure is formed by three subdomains that correspond to the F1 (orange), F2, and F3 (both in green) subregions of the FERM fold. F1 subregion structurally resembles the c-Raf RBD in a complex with Rap1. An arginine residue (R-452), whose corresponding residue in c-Raf (R-89) mediates the c-Raf–Rap1 binding, is also conserved in the KRIT1 F1 subregion. (B) A closer look at the modeled binding interface between KRIT1 F1 subregion (orange) and Rap1 (blue). The R452 residue in KRIT1 and switch 1 domain residues in Rap1 are shown in black. (C) The R452 residue is crucial for KRIT1–Rap1 binding. GST-F123 WT pulls down recombinant Rap1V12 from HEK293 cell lysate (second lane). The GST-F123(R452E) mutant form, which changes a basic residue to an acidic residue, does not bind to Rap1V12 (left lane). Bottom, the equal loading of both GST fusion proteins as judged by SDS–PAGE and Coomassie Blue staining. Blots are representative of five experiments.

On the basis of the homology model described, we created a KRIT1(R452E) mutant and tested its interaction with Rap1. Affinity chromatography (Figure 2C) revealed that the FERM domain of KRIT1(R452E) exhibited a markedly reduced affinity for Rap1A. We also quantified the binding affinity by isothermal titration calorimetry (ITC) (Figure 3, A and B). The KRIT1 FERM domain bound to Rap1 with K_d = 1.8 ± 0.25 μM; however, the KRIT1(R452E) FERM bound with a ~37-fold-lower affinity (K_d = 67 ± 18.7 μM; Table 1). It is significant that the R452E mutation did not affect the folding of the KRIT1 FERM domain, since it exhibited a sharp melting point (57.0 ± 0.1°C vs. 55.7 ± 0.1°C for wild-type FERM domain) in differential scanning calorimetry (Figure 3C). Thus the KRIT1(R452E) mutant remains well folded but binds to Rap1 with dramatically reduced affinity.

Open in a separate window

FIGURE 3:

Calorimetric characterization of KRIT1 FERM-domain binding to Rap1B bound to GMP-PNP, a GTP analogue. (A) Calorimetric titration of 400 μM Rap1B, out of the syringe, into 45 μM wild-type KRIT1 FERM domain in the sample cell. (B) Titration of 1.2 mM Rap1B into 45 μM KRIT1(R452E) FERM domain mutant protein. (C) KRIT1(R452E) FERM mutant does not disrupt protein folding. Differential scanning calorimetry results of wild-type and R452E FERM proteins both exhibit similar narrowly defined melting points, indicating that they are well folded.

The KRIT1–Rap1 interaction stabilizes endothelial cell–cell junctions

Rap1 acts to stabilize cell–cell junctions through several potential effectors (Bos, 2005), and we proposed KRIT1 as one such effector (Glading et al., 2007). Having created a folded KRIT1 point mutant with markedly impaired Rap1 binding, we used this mutant to test the role of KRIT1–Rap1 interaction in the stabilization of endothelial cell–cell junctions. Small interfering RNA (siRNA)–mediated depletion of KRIT1 led to disruption of cell junctions as measured by a twofold increase in monolayer permeability; cotransfection of KRIT1 reversed the increased permeability. In sharp contrast, cotransfection of KRIT1(R452E) failed to reverse the siRNA-induced permeability increase (Figure 4A), even though it was expressed at similar abundance to wild-type KRIT1 (Figure 4B).

Open in a separate window

FIGURE 4:

KRIT1–Rap1 binding regulates HuVEC permeability. (A) Knockdown of KRIT1 expression by KRIT1 siRNA causes an approximately twofold increase in permeability (left). Reconstitution of recombinant WT KRIT1 reverses increased permeability (center). Recombinant KRIT1(R452E) mutant does not reverse increased permeability (right). Data shown are mean increased percentage over control siRNA in permeability ± SEM; n = 3. *p < 0,05 compared with control siRNA. (B) A representative blot of KRIT1 depletion and reexpression of recombinant proteins. More than 80% knockdown of KRIT1 is observed by immunoprecipitation and blotting. Reconstituted wild-type and RE (arginine 452 to glutamic acid) mutant KRIT1 proteins are expressed at equivalent levels. (C, D) Knockdown of KRIT1 expression causes an approximately threefold increase in pMLC/total MCL ratio (left). Reconstitution of recombinant WT KRIT1 reverses MLC phosphorylation increase (center). Recombinant KRIT1(R452E) mutant does not reverse increased pMLC level (right). Data shown are mean increased percentage over control siRNA; n = 2. (D) Immunoblots of pMLC (top) and total MLC (bottom) in HuVECs.

KRIT1 suppresses the activity of RhoA and its effector, Rho kinase (ROCK), thereby stabilizing endothelial cell–cell junctions (Stockton et al., 2010). We therefore examined the capacity of KRIT1(R452E) to inhibit phosphorylation of myosin light chain (MLC), a major substrate of ROCK. Depletion of KRIT1 led to increased phosphorylated MLC (pMLC), an index of ROCK activity (Figure 4C). This increased pMLC was reversed by reconstitution of KRIT1 but not KRIT1(R452E). Reconstituted wild-type and mutant KRIT1 proteins were equally abundant in all transfections (unpublished data). Thus Rap1 binding to KRIT1 stabilizes endothelial cell–cell junctions by inhibiting the RhoA/ROCK signaling pathway.

KRIT1 requires Rap1 binding to support cardiovascular development

The foregoing experiments established the importance of Rap1 binding for KRIT1 function in stabilizing endothelial cell–cell junctions in vitro. In addition to its effects regulating vascular permeability, KRIT1 is important in vertebrate cardiovascular development. Zebrafish embryos lacking krit1 (also known as santa) exhibit a dilated heart phenotype (Mably et al., 2006), and zebrafish krit1 Arg-449 corresponds to human KRIT1 Arg-452. Injection of an antisense morpholino oligonucleotide targeting splice donor site at exon 14 of krit1 (san MO) (Mably et al., 2006) resulted in cardiac dilation in greater than 95% of san MO–injected fish at 48 h postfertilization (hpf). Authentic dilation was verified by identifying the endocardium in the fluorescence images of flk1:EGFP zebrafish (Figure 5A). Coinjection of wild-type krit1 cRNA together with san MO resulted in less than 5% of fish with dilated hearts (Figure 5B). In sharp contrast, 60% of the san MO + krit1(R449E) cRNA–injected fish exhibited dilated hearts. Both wild-type and R449E zebrafish krit1 cRNAs contained N-terminal hemagglutinin (HA) tag, and similar protein expression levels were confirmed by immunofluorescence microscopy (Figure 5C). In addition, both wild-type and mutant proteins were well expressed in HEK293 cells (Supplemental Figure S3). Thus the direct interaction between Rap1 and KRIT1 is required for KRIT1 function in cardiovascular development.

Open in a separate window

FIGURE 5:

KRIT1–Rap1 interaction is required for normal zebrafish cardiovascular development in vivo. (A) Flk1:EGFP transgenic krit1-morphant (san MO) embryo showed enlarged heart phenotype. Coinjecting the cRNA encoding HA-tagged wild-type (WT) zebrafish krit1 protein reduced the heart size of injected fish, whereas fish coinjected with cRNA encoding HA-tagged Krit1(R449E) (RE, homologous to human R452E mutant) still exhibited a dilated heart phenotype. Dilated heart phenotype was scored by cardiac dilation in living fish in combination with a slower heart rate. Authentic dilation was verified by identifying the endocardium in the fluorescence images, which is indicated by the dotted lines. All microscopic images were taken at 48 hpf. Scale bars, 500 μm. (B) Bar graphs showing effects of Krit1 RE mutant on zebrafish cardiovascular development. Data are expressed as number of embryos with dilated heart phenotype divided by total number of embryos used per experiment times 100%, mean ± SD. Asterisk indicates p < 0.05 compared to san MO + HA-WT cRNA group. Data are from three independent experiments. Total number of animals used: 85 in san MO + HA-WT group, 120 in san MO + HA-RE group. (C) Lateral views of fixed animals at 48 hpf. Right, immunofluorescence images revealing that both HA-WT and HA-RE proteins were expressed at similar levels and throughout the entire fish. Scale bar, 500 μm.

Antibodies, cDNAs, siRNA, and reagents

Monoclonal anti-HA and monoclonal anti–FLAG M2 antibodies (both from Sigma-Aldrich) were used for immunoblotting at 1:4000. Monoclonal anti-KRIT1 (15B2) and polyclonal anti-KRIT1 (6832) antibodies were previously described (Glading et al., 2007). Mouse anti–KRIT1 15B2 antibody was used for immunoprecipitation. Rabbit anti–KRIT1 6832 antibody was used for immunoblotting at 1:1000. Polyclonal anti-pMLC antibody (Cell Signaling, Danvers, MA) was used for immunoblotting at 1:1000. Polyclonal anti-GFP antibody (Clontech, Mountain View, CA) was used for immunoprecipitation. Monoclonal anti-GFP antibody (Clontech) was used for immunoblotting at 1:5000. Monoclonal anti–β-tubulin antibody (BD Biosciences, San Jose, CA) was used for immunoblotting at 1:2000. Monoclonal anti–β-actin antibody (Sigma-Aldrich) was used for immunoblotting at 1:4000.

Human KRIT1 siRNA, cDNAs encoding HA- and GFP-tagged full-length KRIT1, and KRIT1 FERM domain were previously described (Glading et al., 2007). Human KRIT1 R452E mutant was generated by site-directed mutagenesis using the QuikChange SD mutagenesis kit (Agilent, Santa Clara, CA). The mutant was then subcloned into pGEX2TK (GE Healthcare, Piscataway, NJ), pcDNA3.1 (Invitrogen), and pEGFP-C1 (Clontech) vectors for expressing GST-, HA-, and GFP-tagged proteins, respectively. Zebrafish krit1 cDNA was a gift from B. Weinstein (National Institutes of Health, Bethesda, MD) and subcloned into pcDNA3.1 vector (Invitrogen) with an N-terminal HA tag. Zebrafish krit1 (R449E) mutant was generated using the QuikChange SD mutagenesis kit. pRK5-CCM2 encoding FLAG-tagged human CCM2 has been previously described and was the generous gift of Doug Marchuk (Duke University, Durham, NC) (Zawistowski et al., 2005). The recombinant human HEG1 cytoplasmic tail model protein affinity matrix was previously described (Lim et al., 2007; Kleaveland et al., 2009). Mammalian expression constructs for HA-Rap1A-G12V (Rap1V12) and HA-Rap1GAP (RapGAP) have been described (Han et al., 2006). HA-Rap1A-G12V E37A, D38A, S39A, Y40A, and R41A mutants were generated using the QuikChange SD mutagenesis kit. pGEX-H-Ras construct was a gift from D. Broek (University of Southern California, Los Angeles, CA).

Homology modeling of KRIT1 FERM domain–Rap1 interaction

We used a DeepView Swiss-PdbViewer (Swiss Institute of Bioinformatics, Lausanne, Switzerland) (Guex and Peitsch, 1997) to align the KRIT1 FERM and moesin FERM sequences and to build an initial model of KRIT1 FERM domain. We then sent the alignment to I-TASSER server for protein structure and function prediction (Zhang, 2008, 2009; Roy et al., 2010) and built a revised model of the KRIT1 FERM domain. We replaced the c-Raf RBD from the Rap1/c-Raf RBD complex crystal structure (PDB 1C1Y) with our KRIT1 FERM model and adjusted the spatial restraints with Rap1 by using Coot (Emsley et al., 2010). This Rap1/KRIT1 FERM complex model was then completely immersed in a water sphere, and the whole system was subjected to energy minimization by molecular dynamics simulation using NAMD program (Phillips et al., 2005). The simulation continued for 5 ns until the total energy reached the minimum and the RMSD was stabilized. The quality of the model was assessed by the program QMEAN (Benkert et al., 2009). The final model was rendered for publication in PyMOL (Schrödinger, New York, NY).

In vitro protein interaction assay

Bacterial expression plasmids encoding glutathione S-transferase (GST)–KRIT1 FERM, partial KRIT1 FERM domains (F23, F1), R452E mutant, GST–RalGDS RBD, or GST vector were expressed in BL21 (DE3) (Novagen, Madison, WI), and recombinant proteins were purified on glutathione-Sepharose beads according to manufacturer's instructions (GE Healthcare). HEG1 and αIIb intracellular tail model proteins were expressed and purified as previously described (Pfaff et al., 1998). To perform pulldown assays using cell lysates, HEK293 cells were transfected with indicated cDNAs. At 24 h after transfection, cells were lysed in cold lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP-40, and 5 mM MgCl₂) plus protease and phosphatase inhibitor cocktails (Roche). A total of 10 μg of immobilized bead-bound proteins was added to 350 μg of clarified cell lysates. Reactions were kept at 4°C overnight. After washing the beads with lysis buffer, samples were fractionated on 4–20% SDS–PAGE gel (Invitrogen). Bound proteins were analyzed by immunoblotting or Coomassie Blue staining.

Immunoprecipitation

HEK293 cells were transfected with indicated cDNAs and incubated for 24 h. Cells were scraped on ice in lysis buffer plus protease and phosphatase inhibitor cocktails (Roche). A total of 2 μg of monoclonal anti-GFP antibody was added to 350 μg of cell lysates and incubated at 4°C overnight. Protein G-Sepharose (Invitrogen) was added to the reaction mixture and further incubated for 4 h at 4°C. After three washes with lysis buffer, beads were mixed with sample buffer and subjected to SDS–PAGE. Immunoprecipitated proteins were detected by immunoblotting.

Isothermal titration calorimetry and differential scanning calorimetry

Proteins were dialyzed into the ITC buffer (20 mM sodium phosphate, pH 6.5, 150 mM NaCl, 3 mM MgCl₂, and 0.1 mM guanylimidodiphosphate [GMP-PNP]). The thermodynamic parameters are determined using an isothermal titration calorimeter iTC200 (MicroCal, Piscataway, NJ) at 25°C in 20 mM sodium phosphate (pH 6.5), 150 mM NaCl, 3 mM MgCl₂, and 0.1 mM GMP-PNP. For binding, 0.4 or 1.2 mM Rap1B bound to GMP-PNP was titrated from the syringe into the sample cell containing 45 μM KRIT1 FERM domain wild-type or R452E mutant protein, respectively. Titration was carried out by injecting volumes of 2.5 μl into the sample cell, where the time between injections was 2 min. Further data evaluation was done using the MicroCal Origin program.

Differential scanning calorimetry experiments were performed at a scanning rate of 1 K/min under 3.0 atm of pressure using N-DSC II differential scanning calorimeter (Calorimetry Sciences Corp, Provo, UT). Differential scanning calorimetry samples contained phosphate-buffered saline (PBS) (pH 7.4) and 0.8 mg/ml KRIT1 F123 or KRIT1F123(R452E).

Protein expression and purification for ITC and differential scanning calorimetry

The cDNA encoding human KRIT1 residues 417–736 was synthesized by PCR using a human KRIT1 cDNA as template and cloned into the expression vector pLEICS-07 (Protex, Leicester, United Kingdom). Constructs were expressed in Escherichia coli BL21 Star (DE3) (Invitrogen) cultured in LB media. Recombinant His-tagged polypeptides were purified by nickel-affinity chromatography following standard procedures. The His tag was removed by cleavage with AcTEV protease (Invitrogen), and the protein was further purified by cation-exchange chromatography. The protein concentration was assessed using the A280 extinction coefficient of 45,090 M⁻¹ before dialysis into buffers. Human Rap1 isoform Rap1B (residues 1–167) was cloned into pTAC vector and expressed in the E. coli strain CK600K (Tucker et al., 1986). Cultures were grown at 37°C to A600 of 0.8, when they were induced with 200 μM IPTG and then cultured at 18°C overnight. Rap1B protein was purified by ion exchange, followed by Superdex-75 gel filtration. Nucleotide exchange was achieved as previously described (Gorzalczany et al., 2000), and the final sample was dialyzed into ITC buffer. The protein concentration was assessed using the CB protein assay (EMD, San Diego, CA).

In vitro endothelial permeability assay

Human umbilical vein endothelial cells were detached and transfected with indicated DNA and/or siRNA using an Amaxa (Lonza) nucleoporation device, then plated onto fibronectin-coated 6-mm Transwell filters and grown for 48 h. In some experiments, cells were plated into filters and grown to confluence, then transfected using FuGENE 6. Upper and lower chamber medium was replaced with phenol-free DMEM. Fluorescein isothiocyanate (FITC)–labeled 40-kDa dextran (2 μg/ml) was added to upper chambers for 2 h, and then FITC was quantified in lower chamber medium by fluorimetry. Raw fluorescence units were normalized as percentage of control treatment.

Immunofluorescence

Human umbilical vein endothelial cells were grown to ~80% confluence on fibronectin-coated glass coverslips, then transfected using FuGENE 6 reagent and the indicated KRIT1 cDNA. Transfected cells were grown for 24 h and then treated for 2 h at 37°C with 5 μM nocodazole or vehicle, then for 60 min at 37°C with 2 μM taxol or vehicle (Invitrogen). Cells were fixed with 3.7% formaldehyde for 60 min at 37°C, permeabilized for 7 min with 0.15% Triton X-100, and blocked for 1 h with 10% normal goat serum (Invitrogen) in Tris-buffered saline, pH 7.4, with 0.01% Triton X-100 (TBST). Cells were probed with mouse anti–β-catenin antibody (BD Biosciences), anti–VE-cadherin (Cell Signaling), or mouse antitubulin (Sigma-Aldrich) in TBST and with rabbit anti-GFP (Clontech) (all at 1:1000) overnight in a humidified chamber. Coverslips were washed with TBST and then incubated with indicated Alexa 568–labeled anti–mouse immunoglobulin (IgG) and anti–rabbit IgG Alexa 488 (Invitrogen) at 1:1000 overnight at 4°C. Coverslips were washed in TBST and PBS (pH 7.4) and then mounted using Prolong Gold (Invitrogen). Imaging was performed by a microscope (Leica LR23) with a charge-coupled device camera (model 2.2.1; Diagnostic Instruments, Sterling Heights, MI) or a confocal microscopy system (Leica TCS SP2 AOBS with DMRE microscope and HCX PL APO 63×/1.32 oil objective). Images were acquired for each experiment using the same gamma setting for all samples and processed using Photoshop CS II software (Adobe, San Jose, CA). Images shown are representative of results from at least three independent experiments.

Zebrafish studies

We maintained and bred antibody wild-type zebrafish and Tg(flk1:EGFP) mutant zebrafish. We injected antisense morpholino oligonucleotides (Gene-Tools, Eugene, OR) that target splice junctions of krit1 (Mably et al., 2006) or standard nontargeting control morpholino oligonucleotides (Gene-Tools) into the yolks of one-cell-stage embryos at a dose of 1 ng. To rescue the enlarged heart phenotype caused by the krit1 morpholino, we coinjected the one-cell-stage embryos with 1 ng of the krit1 morpholino and 400 pg of cRNA encoding HA-tagged krit1 or krit1 (R449E). We used the mMessage mMachine T7 ultra kit (Ambion, Austin, TX) to prepare cRNAs. We mounted the 48-hpf live embryos laterally in 1.2% agarose and acquired images using a fluorescence imaging system (Leica AF6000 with M205 FA stereomicroscope and Leica DFC310 FX digital color camera).

Paraformaldehyde-fixed (4%) embryos were kept in methanol and rehydrated in incubation buffer containing 1× PBS, 0.5% Triton X-100, 5% bovine serum albumin, and 2% goat serum. The primary antibody, rabbit anti-HA tag (Clontech), was used at 1:200 dilution. The secondary antibody, goat anti–rabbit IgG (H+L) Alexa Fluor 488 (Invitrogen), was used at 1:500 dilution. Stained embryos were mounted in 50% glycerol images acquired as described earlier. Images were acquired for each experiment using the same gamma setting for all examples and processed using Photoshop CS II software.

Microtubule sedimentation assay

U2OS cells were transfected with plasmids and treated with either vehicle or nocodazole as indicated. The microtubule fraction was prepared using a cytoskeleton isolation and assay kit (Cytoskeleton, Denver, CO). After centrifugation, the pellets were redissolved in cold water on ice for 1 h. To assess GFP-KRIT1 cosedimentation, polyclonal anti-GFP antibody (Clontech) was used for immunoprecipitation at 4°C overnight. Protein G-Sepharose was added to the reaction mixture and further incubated for 4 h at 4°C. After three washings with lysis buffer, beads were mixed with sample buffer and subjected to SDS–PAGE. Proteins were detected by immunoblotting with monoclonal anti-GFP antibody. The presence of tubulin was assessed by immunoblotting with monoclonal anti–β-tubulin antibody (BD Biosciences).

The authors are grateful to I. L. Barsukov (University of Liverpool, Liverpool, United Kingdom) for providing access to the ITC facility and help with the ITC experiments and B. T. Goult (Leicester University, Leicester, United Kingdom) for technical advice and assistance. J.J.L. was supported by an American Heart Association postdoctoral fellowship. These studies were supported by grants from the National Institutes of Health to M.H.G.