Solution structure of the human signaling protein RACK1
© Gonçalves et al; licensee BioMed Central Ltd. 2010
Received: 17 February 2010
Accepted: 8 June 2010
Published: 8 June 2010
Skip to main content
© Gonçalves et al; licensee BioMed Central Ltd. 2010
Received: 17 February 2010
Accepted: 8 June 2010
Published: 8 June 2010
The adaptor protein RACK1 (receptor of activated kinase 1) was originally identified as an anchoring protein for protein kinase C. RACK1 is a 36 kDa protein, and is composed of seven WD repeats which mediate its protein-protein interactions. RACK1 is ubiquitously expressed and has been implicated in diverse cellular processes involving: protein translation regulation, neuropathological processes, cellular stress, and tissue development.
In this study we performed a biophysical analysis of human RACK1 with the aim of obtaining low resolution structural information. Small angle X-ray scattering (SAXS) experiments demonstrated that human RACK1 is globular and monomeric in solution and its low resolution structure is strikingly similar to that of an homology model previously calculated by us and to the crystallographic structure of RACK1 isoform A from Arabidopsis thaliana. Both sedimentation velocity and sedimentation equilibrium analytical ultracentrifugation techniques showed that RACK1 is predominantly a monomer of around 37 kDa in solution, but also presents small amounts of oligomeric species. Moreover, hydrodynamic data suggested that RACK1 has a slightly asymmetric shape. The interaction of RACK1 and Ki-1/57 was tested by sedimentation equilibrium. The results suggested that the association between RACK1 and Ki-1/57(122-413) follows a stoichiometry of 1:1. The binding constant (KB) observed for RACK1-Ki-1/57(122-413) interaction was of around (1.5 ± 0.2) × 106 M-1 and resulted in a dissociation constant (KD) of (0.7 ± 0.1) × 10-6 M. Moreover, the fluorescence data also suggests that the interaction may occur in a cooperative fashion.
Our SAXS and analytical ultracentrifugation experiments indicated that RACK1 is predominantly a monomer in solution. RACK1 and Ki-1/57(122-413) interact strongly under the tested conditions.
The adaptor 36 kDa protein RACK1 (receptor of activated kinase 1) was originally identified as an anchoring protein for protein kinase C and contains seven WD repeats that mediate its protein-protein interactions [1, 2]. It is also found to be up-regulated in human carcinomas and during tissue regeneration after ischemic renal injury [3, 4]. Furthermore, RACK1 has been functionally implicated in the development of cardiac hypertrophy, in the increase of focal adhesion and regulation of cell adhesion . Also, recent data suggest that RACK1 may affect gene expression through translation regulation and assembly activation of ribosomes [8–10]. Overall, these data suggest that RACK1 is a multi-purpose protein whose regulatory roles may impact various functional contexts.
It has been reported that RACK1 interacts with several molecules including β PKC β; Src; β integrins; PDE4D5; STAT1 and the regulatory protein Ki-1/57, [16, 17] a cytoplasmic and nuclear protein initially identified in malignant cells from Hodgkin's lymphoma [18, 19]. Recent studies showing its phosphorylation, arginine methylation and interaction with other regulatory proteins, suggest that the functional role of Ki-1/57 in human cells seems to be related to splicing regulation and other events in RNA metabolism [16, 20, 17, 21].
RACK1 is a member of the family of β-propeller proteins and the first member of this protein family whose structure has been determined was the β-subunit of the mammalian heterotrimeric G protein [8, 22]. Significant sequence similarity with the G protein β-subunit has led to the prediction that RACK1 may also adopt a seven-bladed β-sheet propeller [22, 23]. This suggests the availability of multiple protein interaction surfaces. We previously modeled the three-dimensional structure of human RACK1 and the crystal structure of a RACK1 ortholog from Arabidopsis thaliana was recently elucidated with a resolution of 2.4 Å (PDB entry 3md0) .
Here, we analyzed the structural properties of human RACK1 in solution through a set of biophysical approaches. We showed that RACK1 is a predominantly monomeric protein in solution accordingly to data obtained from analytical ultracentrifugation (AUC) and small angle X ray scattering (SAXS) experiments. We restored the human RACK1 envelope from SAXS curves through ab initio approaches and compared it with previously obtained related structures. These results suggest a strong similarity between the structures of human RACK1 and RACK1 isoform A from Arabidopsis thaliana, in accordance with the high conservation between both amino acids sequences. The functional activity of RACK1 in solution was also accessed through its binding to Ki-1/57 protein. Fluorescence spectroscopy experiments after titration of Ki-1/57(122-413) in solution with RACK1 and sedimentation equilibrium experiments showed that this association is strong and might involve cooperative events.
The crystallographic structure of the RACK1 isoform A from Arabdopsis thaliana was recently solved and deposited in the Protein Data Bank under the code 3dm0 . Furthermore, the homology model of the human RACK1 was previously published,  allowing to calculate the scattering intensity curve from the atomic coordinates of these two structures as described in the "material and methods" section. We compared these calculated curves with the experimental data obtained for the human RACK1 in solution. The results are shown in Figure 2B. Both calculated curves compare quite well with the experimental data, which indicates that the human RACK1 may have a very similar structure compared to both the homology model and the RACK1 isoform A from Arabdopsis thaliana, although the calculated intensity for the homology model (red curve) seems to be nearer to the experimental data for human RACK1.
One of the great advantages of the SAXS technique is that it offers the possibility of restoring the three dimensional molecular envelope from the experimental data. This modeling can be improved when portions of the structure are already known, solving the possible ambiguities resulting from the calculation. This seems to be the case of the human RACK1, since both a homology model of human RACK1 as well as the crystal structure of RACK1 from Arabidopsis thaliana, are available. Since we studied the protein fused to a 6xHis tag at the N-terminal and we already observed that the homology modeled human RACK1 structure provided a calculated SAXS curve in agreement with the experimental SAXS data, we restored the human RACK1 molecular envelope using the method of addition of missing loops. Fixing the known part of the human homology model, we built dummy residues (DR) chains in the position attached to the proper amino acids in the N-terminal region of the structure. The number of DRs was known from the amino acid sequence. Both programs CHADD and GLOOPY provided very similar SAXS data based models for the human RACK1.
The resulting model is presented in Figure 2C (yellow spheres) superimposed to the ab initio homology model (ribbons representation). For the correct alignment of the two models, we used the program SUPCOMB [30, 31]. To show the region of the ab initio model where the (6xHis tag) was attached to the protein, the dummy elements are shown as blue spheres. The intensity curve calculated from the model (derived from SAXS data) is displayed in the Figure 2B (black curve), showing, as expected, an even better fit to the experimental data. These results give a strong indication of the similarity of the structure of human RACK1 and RACK1 isoform A from Arabdopsis thaliana and confirms the structure predicted by the homology model of human of RACK1.
RACK1 hydrodynamic properties.
RACK1 Hydrodynamic property
Predicted for a sphere
HydroPro of homology model
36.4 ± 0.8&
36.4 ± 0.8&
s 0 20,w (S)
3.00 ± 0.01
3.16 ± 0.02
7.7 ± 0.2
8.0 ± 0.1
63 ± 1
20.0 ± 0.1
20.8 ± 0.8$
20.6 ± 0.4#
Table 1 summarizes the RACK1 hydrodynamic properties calculated for a sphere of 37 kDa (RACK1's MM) and the experimental data determined by dynamic light scattering and sedimentation velocity. We also used the RACK1 homology model previously calculated  to obtain its hydrodynamic properties using HydroPro software (see Material and Methods) . Both D20,w and s020,w determined experimentally were in accordance to the hydrodynamic properties calculated for the homology model, suggesting that it is a reasonable model for RACK1 in solution. All hydrodynamic data, together with the frictional ratio calculated by either SedFit software (~1.4) or estimated from the ratio of ssphere to s020,w (~1.3), suggest that RACK1 is a slightly asymmetric or distorted particle.
Thus, our findings that RACK1 is predominantly a monomer in solution seem at first to contrast with some data found in the literature, that relate that human RACK1 can engage in dimerization interaction in in vitro pull-down assays and in vivo, using cross-linker experiments . A more recent study however, demonstrates that such a RACK1 dimerization in vivo may depend crucially on its previous phosphorylation, while dephosphorylation of RACK1 causes the disintegration of the homodimeric complex  Altogether, these findings are in good agreement with our results, since our bacterially expressed RACK1 is not phosphorylated, whereas RACK1 in human cells is likely to be phosphorylated and regulated by phosphorylation. Therefore, the dephosphorylated RACK1 should present low KB while the phophorylated RACK1 may be able to dimerize with higher KB. However, we cannot exclude that the relative low KBM-D and KD-T described above may be an artifact induced by the experimental conditions tested.
We also used sedimentation equilibrium methods to assess how functional RACK1 protein was in our experimental conditions. Previously, we described the interaction of RACK1 with the regulatory protein Ki-1/57 and analyzed their interaction through spectroscopic methods [17, 24]. We then used here a truncated form of Ki-1/57 encompassing the amino acids 122-413 to obtain further thermodynamic details of this association. We analyzed the absorbance data by the SedPhat program as multi-speed equilibrium data and the results are presented in Figure 4B by using the Equation 3 presented above for a hetero-association system with a stoichiometry between RACK1 and Ki-1/57(122-413) of 1:1 (Equation 4). The lower panel (Figure 4B) shows the residuals fitting randomly dispersed around zero, which points to a good data fitting. The KB observed for RACK1-Ki-1/57(122-413) interaction was of around (1.5 ± 0.2) × 106 M-1 that resulted in a KD of 0.7 ± 0.1 μM. This data suggest that the equilibrium interaction between these two proteins is strong in the tested conditions. The KB for RACK1-Ki-1/57(122-413) interaction is 7.5 times higher than RACK1 KBM-D suggesting that, in the protein concentration assayed, around 91% of the RACK1 is in the hetero-dimeric complex and that the concentration of RACK1 as a homo-dimer must be lower than 0.01 μM and could be neglected from this analysis. This highly productive association may explain why around 54% of clones restored from our yeast two-hybrid screenings using Ki-1/57 as bait represented the RACK1 protein .
At concentrations higher than 0.3 μM the Ki-1/57 titration induced RACK1 fluorescence emission quenching which follows a sigmoidal curve (Figure 5, inset), that represents another effect due to protein association. Figure 5B represents the difference of fluorescence intensity at 334 nm (F0 - F = θ) as function of Ki-1/57 concentration range from 0.3 μM up to 1.4 μM. Adjusting this titration curve with a modified Hill equation (Equation 5) we could estimate the KD for the interaction between Ki-1/57 and RACK1 as about 0.60 μM and the number of cooperative sites n as 8 ± 1. This points to an apparent cooperativity of the interaction, since the stoichiometry determined by sedimentation equilibrium was 1:1. It is also interesting to note that this KD corresponds to the KD of RACK1\Ki-1/57(122-413) observed in sedimentation equilibrium experiments. These data emphasize that the interaction between human RACK1 and Ki-1/57 was strong in the experimental conditions tested.
The human RACK1 protein contains thirteen tryptophan residues, which yield strong signal fluorescence emission intensities but also increase the complexity of fluorescence emission spectra. Despite the high number of Trp residues in the RACK1 structure, the homology model developed for RACK1 suggests the presence of, at least, two kinds of Tryptophans on RACK1: 1) the exposed Trps on the RACK1 surface and 2) the buried or partially buried Trps . Thus, these different classes of Trps may suffer distinct effects upon Ki-1/57(122-413) binding: while some of them may be more protected, others may be rather quenched. However, we cannot exclude the participation of the 3 Trps on the Ki-1/57(122-413) recombinant protein. Ki-1/57(122-413) has been described as an intrinsically unfolded protein  and therefore it could bind to RACK1 by several contact points, possibly inducing conformational changes on either protein. The sigmoidal titration curve observed may be explained by possible cooperative effects in the binding of the two proteins. However, more detailed emission fluorescence or other spectroscopic experiments are required to address this newly raised issue.
Human RACK1 has significant amino acid sequence and structurally similarity to RACK1 isoform A from Arabidopsis thaliana. Analytical ultracentrifugation experiments showed that human RACK1 has similar hydrodynamic properties to that predicted to homology model previously developed. Moreover, the SAXS data demonstrated that the human RACK1 in solution may have the same or very similar molecular envelope as that of the RACK1 isoform A from Arabidopsis thaliana. Both techniques also showed that the homology model previously proposed by shows a good agreement with the experimental data presented here .
The hydrodynamic properties of the human RACK1 were established by both sedimentation velocity and dynamic light scattering. The samples presented an apparent aggregation in the hydrodynamic experiment and they showed polydisperse behavior. Although the equilibrium sedimentation analytical ultracentrifugation results indicated a tendency of the protein for di- and tetramerization, the data showed that human RACK1 is predominantly a monomer in solution (over 90%). SAXS data also confirmed the human RACK1 monomeric conformation in solution, although we previously adjusted the glycerol content in the buffer and centrifuged the sample to assure its monodispersity. This condition was checked by DLS before the SAXS experiments (data not shown).
The interaction of RACK1 with the Ki-1/57(122-413) is strong and follows the stoichiometry of 1:1 as observed by sedimentation equilibrium experiments. RACK1 is a Trp-rich protein and its fluorescence emission spectrum presents a maximum of around 334 nm suggesting that some of these Trp are exposed to the solvent. The interaction of RACK1 with the Ki-1/57(122-413) monitored by fluorescence emission spectroscopy agreed with the interaction strength observed in the sedimentation equilibrium experiments. Furthermore, the titration curve resembles a sigmoidal curve suggesting a cooperative effect in the interaction of RACK1 with the Ki-1/57(122-413). Considering that Ki-1/57(122-413) is an intrinsically unfolded protein, its interaction with RACK1 may involve several binding sites and seems to induced conformational changes upon binding.
Cloning of the cDNA coding for RACK1 full-length into the pET-28a plasmid (Novagen/EMD Biosciences, San Diego, CA) was performed as described previously . The recombinant vector pET-28a-RACK1 expresses a construct containing 24 additional amino acids at the N-terminal of RACK1. These extra residues encopass the 6 × His-tag and other poly-linker encoded amino acids. cDNA encoding human Ki-1/57 protein fragments was amplified by PCR and then directionally cloned into the bacterial expression pPROEx (Invitrogen, Carlsbad, USA), as previously described .
BL21(Δ SlyD) strain (Invitrogen) was transformed with recombinant pET28a- RACK1 vector and grown in LB medium with 30 ug/ml of kanamycin. When reaching the logarithmic growth phase the recombinant bacteria were induced for protein production with 1.7 mM of isopropyl-b-D-thiogalactoside (IPTG) at 30 °C for 4 h. After harvesting (6000 × g, 10 min), the bacterial cells were resuspended and incubated for 40 min at 4 °C in lysis buffer: 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 20% glycerol, 1 mg/ml lysozyme, 1 mM phenylmethylsulfonyl fluoride, and 0.05 mg/ml DNase. Additional pellet disruption was performed by 10 cycles of sonication in an ice bath, followed by centrifugation at 18 000 × g, 4 °C for 30 min. The obtained supernatant was loaded onto a HiTrap Chelating HP column (Ge Healthcare) and eluted by a gradient of 0-400 mM imidazole. The obtained Ni2+-affinity purified fractions were pooled and dialyzed against the buffer: 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 20% glycerol.
To obtain protein preparations free of degradation and up to 95% of purity, Ki-1/57(122-413) was purified in two steps as follows as described previously . The concentration of the recombinant protein was spectroscopically determined using the calculated extinction coefficient for the denatured proteins .
The SAXS experiments were performed at the D02A-SAXS2 beamline of the Laboratório Nacional de Luz Síncrotron (LNLS, Campinas, Brazil). Before the analysis, the samples of human RACK1 were centrifuged at 356,000 × g for 30 min at 18 °C to remove any possible aggregates. The composition of the buffer solution was adjusted to avoid aggregation, [20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 20% glycerol]. Dynamic light scattering (DLS) was used to test the monodispersity of the solution. Measurements were performed with a monochromatic X-ray beam with a wavelength of λ = 1.488 Å and the X-ray scattering patterns were recorded using a two-dimensional position-sensitive MARCCD detector. The sample-to-detector distances were set at 506.8 mm and 1722.5 mm, corresponding to the scattering vector range of 0.01 < q < 0.50Å-1 where q is the magnitude of the q-vector defined by q = (4π/λ)sinθ (where 2θ is the scattering angle). Three successive frames of 300 s each were recorded for each sample. The scattering data from the buffer were recorded before and after the measurement of each sample. The experiments were performed with two different sample concentrations: 1.95 and 2.54 mg/mL. The sample and buffer intensity curves were individually corrected for detector response and scaled by the incident beam integrated intensity and the sample absorption. Subsequently, the buffer scattering was subtracted from the corresponding sample scattering. The resulting curves were inspected for radiation-induced damage, but no such effect was observed. A 5 mg/ml BSA (66 kDa) solution in the same sample buffer was used as MM standard sample to estimate the MM of RACK1. This value was inferred from the ratio of the extrapolated values of the intensity at the origin, I(0), from both sample and BSA solutions scattering [38, 39].
The initial analyses of the SAXS data were performed following the standard procedures. The radius of gyration and the forward scattering intensity I(0) were obtained from the Guinier approximation [26, 40, 41] where we assumed that the intensity can be represented as I(q) = I(0)exp(-(qR g )2/3) at very small values of q (qR g < 1). Thereupon, the pair distance distribution function p(r) was calculated by indirect Fourier transform of the scattering intensity through a regularization method implemented in the program GNOM . This p(r) function also provided the I(0) and R g values. These parameters proved to be in agreement with those obtained from the Guinier approximation. This function also allows determination of the maximum dimension of the protein in solution since p(r > Dmax) = 0.
We compared the experimental scattering curve from the human RACK1 protein with its calculated scattering curve from the homology model from the human RACK1 and with the calculated curve from the crystallographic structure of the RACK1 isoform A from Arabdopsis thaliana. The calculation of the scattering intensity from the atomic coordinates of the latter structures was performed using the program CRYSOL . Since the sample we studied was in a solution with 20% glycerol, we calculated the electron density of the solvent for this condition (0.3497 electrons/Å3). The contribution of the other components of the buffer solution to the the electron density of the water (0.3344 electrons/Å3) was considered negligible, due to their low concentration.
The ab initio models of the human RACK1 in solution were calculated from the SAXS data using the homology model previously obtained. We used a modeling method that is an extension of the original Dummy-Residues (DR) approach . It can be used when parts of the protein are known and the location of the interface of the known and missing parts is also known . Here, as we studied the protein fused to a 6xHis tag at the N-terminal, the homology model of the human RACK1 is used as the known part. The 6xHis tag is represented as a chain of DRs attached to the proper residues (known from the sequence) in this structure. We performed random modification in this chain of DRs using the simulated annealing method for global minimization of the scoring function to provide the best fit to the experimental scattering curve. This method is implemented in two programs:  CHADD and GLOOPY. Both programs are similar but the second one takes into account the primary structure (amino acid sequence) of the protein.
where R is the gas constant, T is the absolute temperature.
The HydroPro software  was applied to estimate the standard diffusion coefficient (D 20,w ), standard sedimentation coefficient (s20,w) and hydrodynamic maximum distance (D max,h ) from the RACK1 model previously developed  in order to compare to the experimentally hydrodynamic data.
where C is the protein concentration at radial position r, C mononer,r0 is the protein concentration at radial position r 0 (initial radial position), ω is the centrifugal angular velocity, KB is the binding constant and n is the stoichiometry. Equation 3 can be developed for equilibrium systems with more species.
The software Sednterp was used to estimate the molar absorbance of the folded protein at 280 nm for RACK1 (80.940 mol/L-1.cm-1) and the C-terminal of Ki-1/57(264-413) (32.890 mol/L-1.cm-1). These parameters were used in SedPhat software for hetero-association fitting.
where F0 is the fluorescence intensity at 0.3 μM of Ki-1/57, F is the fluorescence intensity at a given Ki-1/57 concentration (in μM), θ is the difference of fluorescence intensity, θ0 is θ at 0.3 μM of Ki-1/57; θmax is θ at Ki-1/57 saturating concentration, and n is the number of cooperative sites. The software OriginPro® (Version 8) was used to this fitting routine.
Financially supported by: Fundação de Amparo à Pesquisa do Estado São Paulo, the Conselho Nacional de Pesquisa e Desenvolvimento and the LNLS. We thank Maria Eugenia R. Camargo for technical assistance, Rodrigo Martinez for the technical support at the SAXS2 beamline, and Renata Rocha de Oliveira for assistance with Fluorescence data acquisitions.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.