Biochemical Journal

Research article

Structural stabilization of protein 4.1R FERM domain upon binding to apo-calmodulin: novel insights into the biological significance of the calcium-independent binding of calmodulin to protein 4.1R

Wataru Nunomura , Daisuke Sasakura , Kohei Shiba , Shigeyoshi Nakamura , Shun-ichi Kidokoro , Yuichi Takakuwa

This article has a correction. Please see:

Abstract

In erythrocytes, 4.1R80 (80 kDa isoform of protein 4.1R) binds to the cytoplasmic tail of the transmembrane proteins band 3 and GPC (glycophorin C), and to the membrane-associated protein p55 through the N- (N-terminal), α- (α-helix-rich) and C- (C-terminal) lobes of R30 [N-terminal 30 kDa FERM (4.1/ezrin/radixin/moesin) domain of protein 4.1R] respectively. We have shown previously that R30 binds to CaM (calmodulin) in a Ca2+-independent manner, the equilibrium dissociation constant (Kd) for R30–CaM binding being very similar (in the submicromolar range) in the presence or absence of Ca2+. In the present study, we investigated the consequences of CaM binding on R30's structural stability using resonant mirror detection and FTIR (Fourier-transform IR) spectroscopy. After a 30 min incubation above 40°C, R30 could no longer bind to band 3 or to GPC. In contrast, R30 binding to p55, which could be detected at a temperature as low as 34°C, was maintained up to 44°C in the presence of apo-CaM. Dynamic light scattering measurements indicated that R30, either alone or complexed with apo-CaM, did not aggregate up to 40°C. FTIR spectroscopy revealed that the dramatic variations in the structure of the β-sheet structure of R30 observed at various temperatures were minimized in the presence of apo-CaM. On the basis of Kd values calculated at various temperatures, ΔCp and ΔG° for R30 binding to apo-CaM were determined as −10 kJ·K−1·mol−1 and ~−38 kJ·mol−1 at 37°C (310.15 K) respectively. These data support the notion that apo-CaM stabilizes R30 through interaction with its β-strand-rich C-lobe and provide a novel function for CaM, i.e. structural stabilization of 4.1R80.

  • apo-calmodulin
  • 4.1/ezrin/radixin/moesin domain (FERM domain)
  • protein 4.1R
  • β-sheet structure
  • structural stability

INTRODUCTION

Protein 4.1R is a key membrane skeletal protein in human erythrocytes where it is expressed as an 80 kDa isoform (4.1R80). 4.1R80 comprises four major chymotryptic domains: an N-terminal 30 kDa domain also known as a FERM (4.1/ezrin/radixin/moesin) domain, a 16 kDa domain, a 10 kDa domain and a C-terminal 24 kDa domain [1,2]. The N-terminal domain, which consists of 279 amino acid residues, is the focus of the present study. We refer to it as R30 in the present paper. R30 binds to various transmembrane proteins including band 3 [3], GPC (glycophorin C) [4], CD44 [5] and to the erythrocyte membrane-associated protein p55 [6,7]. The 10 kDa domain of 4.1R80 binds to spectrin and actin filaments [1,2]. Through these multiple interactions, 4.1R80 is a key component for the maintenance of the mechanical stability of human erythrocytes.

CaM (calmodulin), a regulator of cellular signalling, binds to and activates more than 100 known target proteins [8,9]. In human erythrocytes, saturation of CaM with Ca2+ (Ca2+–CaM) destabilizes the mechanical stability of membranes [2]. Although CaM binds to R30 in a Ca2+-independent manner, Ca2+–CaM regulates R30 binding to membrane proteins [2,5,7,10,11] and to the spectrin–actin complex [12,13]. Ca2+–CaM binding to 4.1R80 results in a destabilization of membrane stability. Although the stoichiometry of R30 binding to Ca2+–CaM has been shown to be 1:1 [11], two CaM-binding sites have been identified in R30. The A264KKLWKVCVEHHTFFRL peptide, located in the exon 11-encoded region of R30 (pep11), mediates Ca2+-independent CaM binding. The A181KKLSMYGVDLHKAKDL peptide, located in the exon 9-encoded region of R30 (pep9), is responsible for Ca2+-sensitive CaM binding, with Ser185 being critical for Ca2+-dependency [11]. We have shown previously that the binding affinity of R30 for band 3 and GPC decreases when Ca2+–CaM binds simultaneously to pep11 and to Ser185 [5,7,11], with Ca2+–CaM losing its regulatory effect when Ser185 is mutated to tryptophan or proline [11,14]. In most cases, CaM binding to target proteins strongly depends on Ca2+ saturation of CaM [8,9]. In that respect, the characteristics of CaM binding to R30 are unique. These unique properties raised the question as to why R30 binds to apo-CaM with the same Kd as Ca2+–CaM.

X-ray crystal structure reveals that R30 adopts the shape of a three-lobe clover [15], as depicted in Figure 1 (PDB code 1GG3). The cytoplasmic domains of band 3 and of GPC, and the HOOK domain of p55 bind to the N- (N-terminal), α- (α-helix-rich) and C- (C-terminal) lobes respectively [15]. The three-dimensional structure shows that each domain possesses a distinct secondary structure. The C-lobe contains seven β-strands that form three sets of β-sheet structures (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/440/bj4400367add.htm). The CaM-binding pep11 sequence, which adopts an α-helix structure, is located in the C-lobe and the Ca2+-sensitive Ser185 is located on a loop structure between the α- and C-lobes [15]. The dynamic binding of R30 to multiple proteins suggests that the native structure of R30 may be structurally unstable, its free energy being high compared with that of the CaM-bound state. [Ca2+]i (intracellular Ca2+ concentration) is maintained at ~10 nM. In contrast, the equilibrium dissociation constant (Kd) of CaM binding to Ca2+ is in the submicromolar range [13]. Since saturation of one molecule of CaM requires four molecules of Ca2+, one can predict that nearly all CaM molecules in a cell are in a Ca2+-free state, i.e. in an ‘apo-’ state.

Figure 1 Three-dimensional structure of R30

In the three-dimensional structure of R30 (PDB code 1GG3), the Ca2+-insensitive CaM-binding sequence (pep11), and the Ca2+-sensitive site (Ser185 appearing as sphere) are labelled. The side chains in pep11 are presented as a stick model. The cytoplasmic domains of band 3 and GPC, and p55 bind to the N-, α- and C-lobe respectively (reviewed in [2]).

Our goal is to explain our previous observation that the kinetic parameters for binding of apo-CaM to R30 are the same as those for binding of Ca2+–CaM to R30 [11] and to determine whether apo-CaM binding confers on R30 its structural stability on the basis of RMD (resonant mirror detection) and FTIR (Fourier-transform IR) spectroscopy analyses. The results of the present study clearly indicate that apo-CaM stabilizes the β-strand-rich C-lobe of R30 by binding to the pep11 sequence and unveil a novel function for apo-CaM in stabilizing the structure of proteins with which it interacts, such as R30.

EXPERIMENTAL

Materials

pGEX-4T2 bacterial expression vector, glutathione–Sepharose CL-6B, heparin–Sepharose, phenyl-Sepharose 4B, Sephacryl S-200 and Akta Prime Plus® were purchased from GE Healthcare. All other reagents were purchased from Wako Pure Chemicals and Sigma, unless noted otherwise. IAsys® cuvettes coated with aminosilane were obtained from Affinity Sensors.

Synthesis and purification of recombinant proteins

Recombinant R30 was expressed as a GST (glutathione transferase)-fusion protein in BL21 bacteria. Following sonication, the bacterial lysate was loaded on to a glutathione-affinity column for purification, and the recombinant GST-fusion protein was eluted from the column after cleavage of the GST tag with thrombin, as described previously [5,7,11]. After desalting, the protein was purified further on a heparin–Sepharose column to remove contaminants and breakdown products. Finally, R30 was loaded on to a Sephacryl S-200 size-exclusion chromatography column equilibrated with 50 mM Tris/HCl (pH7.5) containing 0.5 M NaCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, 1 mM benzamidine, 0.1% glycerol and 2 mM NaF (Figures 2A and 2B). Preparation of the cytoplasmic domains of band 3 and GPC, and p55 was conducted as described previously [5,7,11]. Protein purity was assessed by SDS/PAGE (12.5% gel). Proteins were stained with Gelcode Blue® (Pierce). The R30 concentration was determined by measuring the absorbance at 280 nm, with the E1% corresponding to 14 at the molar concentration for tyrosine (ϵ=1340), tryptophan (ϵ=5550) and cycteine (ϵ=200) [16].

Figure 2 Purification of R30 and CaM

(A) Elution profile of R30 loaded on to a Sephacryl S-200™ size-exclusion chromatography column. Arrows indicate elution position of marker proteins: a, Blue Dextran (2000 kDa); b, BSA (68 kDa); c, ovalbumin (43 kDa); d, chymotrypsinogen (25 kDa). (B) SDS/PAGE assessment of R30 purity on a 12.5% gel. (C) MS analysis of apo-CaM. (D) SDS/PAGE assessment of apo-CaM purity on a 15% gel. Proteins were stained with Gelcode Blue®. Molecular masses are indicated in kDa.

Purification of CaM

CaM was purified from bovine brain by phenyl-Sepharose affinity chromatography with slight modifications, as described previously [11]. The purity of CaM was assessed by TOF (time-of-flight)-MS (Figure 2C) and SDS/PAGE (15% gel shown in Figure 2D). For SDS/PAGE analysis, 5 μg of CaM in 50 mM Tris/HCl (pH 7.5) containing 0.15 M NaCl and 1 mM EDTA was loaded on to the gel. The CaM concentration was calculated based on the absorbance at 280 nm and an E1% of 1.6 for CaM.

RMD binding assays

Kinetic analysis

Interactions of R30 with apo-CaM were examined using the IAsys® RMD system following the manufacturer's instructions (Affinity Sensors) [17]. The protein immobilized on the cuvette is referred to as the ‘ligand’, whereas the protein added to the cuvette in solution is referred to as the ‘analyte’. CaM was immobilized on aminosilane cuvettes as described previously [11]. Binding assays were conducted at temperatures ranging from 9 to 39°C with constant stirring. R30 was dissolved in 50 mM Tris/HCl (pH 7.5), 0.1 M NaCl, 1 mM EDTA and 1 mM 2-mercaptoethanol (buffer A1) and with 4 mM (final concentration) CaCl2 (buffer A2) and used at concentrations ranging from 50 nM to 1 μM. Kinetic analysis of analyte binding to ligand was conducted using equations reported previously [5,11]. Dissociation constants at equilibrium (termed Kd) were calculated using eqn (1): Embedded Image (1) where ka is the association rate constant, and kd is the dissociation rate constant. Kd was obtained from the means of three to five measurements for ka and kd. Kd was confirmed by Scatchard plotting using maximum binding (Bmax) and molar concentrations of analyte [11,18]. The Bmax was calculated from binding characteristics using the software package FASTfit®, version 2.1.

R30 was pre-incubated in buffer A1 at various temperatures (5–50°C) for 30 min before binding assays with immobilized cytoplasmic domains of band 3 and GPC, or p55. Binding assays using IAsys® were carried out at 25°C. The cuvettes were reused after cleaning with 20 mM HCl. Original binding curves could be replicated after HCl washing, indicating that the washing did not denature the bound ligands. R30 (0.4 μM) in buffer A1 or buffer A2 was incubated for 30 min at temperatures ranging from 5°C to 50°C with or without CaM (4.4 μM) before binding assays with immobilized p55. The maximum response expressed as Beq (represented by ‘arc second’) was estimated from the binding profile using the software package FASTfit®, version 2.1. The Beq for R30 binding to each binding partner at 5°C was 100% and the binding ratio at each temperature was calculated. Temperatures resulting in 50% binding corresponded to the Beqhalf.

Thermodynamic analysis

Change in standard Gibbs free energy (ΔG°) as a result of binding was determined using eqn (2): Embedded Image (2) where R is the gas constant, 8.314 J·K−1·mol−1, T is the absolute temperature, Kd is the average value for the dissociation constant of two to five measurement from IAsys®, as indicated above, ΔH is the change in enthalpy, and ΔS° is the change in standard entropy. Change in heat capacity (ΔCp) was determined by fitting to eqn (3), which reflects the correlation between temperature and ΔG [19]. Curve fitting was performed using the SALS software package [20]. Embedded Image (3)

ΔH is shown as a function of temperature in eqn (4), where T0 is 300.15 K: Embedded Image (4)

In order to determine the thermodynamic parameters of the transition state of R30 upon binding to CaM, the Eyring equation (5) was used [21]: Embedded Image (5) where R is the gas constant, 8.314 J·K−1·mol−1, ka is the association rate constant from eqn (1), h is the Planck constant, 6.63×10−34 J·s, and kB is the Boltzmann constant, 1.38×10−23 J·K−1. ΔH and −TΔS were calculated as described above, using plots of ΔG with temperature.

FTIR spectroscopy of the ATR (attenuated total reflection) spectrum

IR spectra of solutions of proteins R30, CaM or a 1:1 (molar ratio) mixture of R30 and CaM dissolved in 50 mM Tris/HCl (pH 7.5), 0.15 M NaCl, 1 mM EDTA and 1 mM 2-mercaptoethanol (buffer B), were recorded with a Tensor27 spectrometer (Bruker Optik). Protein samples were prepared in a BioATR celli II (Harrick Scientific Products), connected to a thermostat (DC30-K20, Thermo Scientific Haake Products).

The BioATR sample cell was used to analyse protein samples in solution. For each spectrum, a 64 scan interferogram was obtained at a single beam mode at 4 cm−1 resolution. Reference spectra for buffer B alone in the cell were recorded under similar conditions. Recorded and evaluated IR spectra were analysed with the Opus 6.5 software (Bruker Optik). The temperature interval was 2°C and the temperature range was 20–54°C. Second-derivative amide I spectra were determined using nine smoothing points according to the Savitzky–Golay algorithms [22].

DLS (dynamic light scattering) analysis

The apparent molecular mass of CaM was calculated from the molecular diameter determined by DLS [23] using a Zetasizer NanoZS (Malvern Instruments). R30, CaM and the complex (1:1 in molar ratio) were dissolved in buffer B. Cell temperature during measurement was strictly controlled by the system. Z-Average diameter was determined by Cumulant analysis (ISO13321). All samples were filtered through a 0.22 μm pore-size membrane following dialysis and degassed before analysis.

Visualization of R30 and apo-CaM and plotting of B-factor of R30

Three-dimensional structures of R30 (PDB code 1GG3) could be visualized as a ribbon structure and a surface model respectively using the MolFeat Ver. 4.6 (FiatLux) and PyMOL software packages (http://www.pymol.org). B-factors of R30 were obtained from the RSCB PDB (http://www.rcsb.org/pdb/). Each ribbon was assigned a particular colour in accordance to the temperature factor ramped from cold-blue to hot-red for B-factors ≤20 to ≥80 Å2 (1 Å=0.1 nm) respectively.

RESULTS

Structurally unstable sites in R30

The site-specific structural stability of R30 was assessed by binding analysis at various temperatures using the IAsys® system. Following R30 incubation at each temperature for 30 min, the binding activity was computed as Beqhalf (the 50% binding ratio). The Beqhalf for GPC and band 3 occurred at ~40°C (Figure 3A). Although Beqhalf of R30 binding to p55 occured at 34°C, R30 was still able to bind to p55 at 44°C in the presence of apo-CaM (Figure 3B). This suggested that the p55-binding site, located in the C-lobe of R30, might be structurally unstable and that it was stabilized in the presence of apo-CaM. We hypothesized that this stabilization would result primarily from the interaction of apo-CaM with the β-sheet in the C-lobe of R30 (Figure 1 and Supplementary Figure S1).

Figure 3 Effect of apo-CaM on the structural stability of R30

(A) R30 was pre-incubated at each indicated temperature for 30 min, loaded on to a band 3 cytoplasmic domain, GPC cytoplasmic domain or p55-immobilized aminosilane cuvette and subjected to IAsys® binding analysis. The binding (%) at each temperature was compared with the maximum (100%) binding detected at equilibrium at 5°C. Beqhalf shows 50% binding. Small and large circles are raw data and mean values respectively. (B) R30 was pre-incubated with (●) or without (○) apo-CaM at each indicated temperature for 30 min, loaded on to a p55-immobilized aminosilane cuvette and subjected to IAsys® binding analysis. Binding (%) and Beqhalf in the presence or absence of apo-CaM were calculated as in (A). Small and large circles are presented as raw data and mean values respectively. (C) R30 binding to p55 in the absence (white bars) or presence (black bars) of Ca2+–CaM. Results are means±S.E.M. (n=3).

It has been reported previously that the structural stability of a peptide derived from MLCK (myosin light-chain kinase) complex is significantly higher in an apo-CaM-bound state than in a Ca2+–CaM-bound state [2325]. We therefore investigated the effects of Ca2+–CaM on the binding properties of R30 at various temperatures. Surprisingly, binding of R30 to p55 was very similar at 34, 44 and 50°C in the presence or absence of Ca2+–CaM. As shown in Figure 3(C), R30 binding to p55 was already observed at 34°C and was comparable in the presence or absence of Ca2+–CaM. Although binding of R30 to p55 was still observed at 44°C, R30 could no longer bind to p55 at 50°C either in the presence or absence of Ca2+–CaM. These results indicated that the structural stabilization of R30 mediated by CaM was not altered by Ca2+. Thus R30 appears to adopt a unique behaviour with respect to Ca2+-dependency of regulatory properties mediated by CaM.

Temperature-induced secondary-structural change in R30

The second derivative (d2A/dx2) in the ATR analysis of R30 in the presence or absence of apo-CaM is shown in Figure 4(A) (original data are shown in Supplementary Figure S2 at http://www.BiochemJ.org/bj/440/bj4400367add.htm). Amide I bands ranged from 1720 to 1600 cm−1. The downward band centred at 1628 cm−1 indicated that the β-sheet structure had undergone structural changes with temperature. Of particular note, this band was not as pronounced in the presence of apo-CaM (Figure 4B), indicating that the β-strand structure of R30 was more structurally stable in the presence of apo-CaM. In contrast, the α-helix structure of R30, detected at 1652 cm−1, did not change up to 55°C, in either the presence or absence of apo-CaM (Figure 4B). This suggested a specific effect of apo-CaM on the stabilization of the β-sheet structure of R30. We could not detect any significant change in the secondary structure of apo-CaM in the range of temperatures tested (Figure 4A).

Figure 4 Changes in secondary structure of R30 in the presence and absence of apo-CaM as a function of temperature

FTIR measurements enabled the visualization of the second derivative of change in the Amide I region (1720–1600 cm−1). Second derivative of change in the Amide I region of R30 (top panel) and of R30 bound to apo-CaM (middle panel) as a function of temperature is shown (A). The derivative of apo-CaM alone is shown as control (bottom panel). Changes in absorbance for specific regions of R30 (1628 cm−1 for α-helix and 1652 cm−1 for β-sheet structure) in the presence or absence of apo-CaM as a function of temperature are shown (B). Small and large circles are presented as raw data and mean values respectively. ○, R30; ●, R30–apo-CaM complex. The same d2A/dx2 scale (y-axis) is used for both 1628 cm−1 and 1652 cm−1.

In order to investigate further the effect of apo-CaM on R30 stability, we compared the aggregation state of R30 at 35°C, in the presence or absence of apo-CaM, by DLS. DLS enables the estimation of polydispersity or the percentage of width (nm) to diameter (nm) of a protein, an indicator of the heterogeneity of its tertiary structure [23]. Restriction of molecular fluctuation of a protein in response to the binding of specific ligands will result in a decrease in polydispersity. The molecular size of R30 in the presence or absence of apo-CaM was ~5–6 and ~8–9 nm respectively. Polydispersity of R30 at 35°C, in the presence or absence of apo-CaM, was 38.6% and 23.9% respectively, indicating stabilization of R30 in the presence of apo-CaM. The increase in β-sheet structure was not due to aggregation (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/440/bj4400367add.htm). Indeed, DLS measurements showed no aggregation at 39°C for R30 and 40°C for R30 in the presence of apo-CaM. The sensitivity of the apparatus used in the present study was not sufficient to detect smaller fluctuations in temperature (<1°C) for R30 with and without apo-CaM.

Thermodynamic parameters of CaM binding to R30

The IAsys® system can be used at different temperatures, thus enabling thermodynamic analysis of interactions. The Kd for R30 binding to apo-CaM changed considerably with temperature. The minimum Kd was detected at 25°C. The temperaturedependence of Kd was due primarily to that of ka, with kd showing little change with temperature (Figure 5A). In order to characterize the mechanism for the temperature-dependence of R30 binding to apo-CaM, the thermodynamic parameters of R30 binding to apo-CaM were inferred from these results.

Figure 5 Changes in kinetic parameters as a function of temperature and thermodynamic parameters

(A) Changes in Kd, ka and kd for R30 binding to apo-CaM as a function of temperature. Small and large circles indicate raw and mean values respectively. Kinetic parameters were determined with the IAsys® system software. (B) Theoretical changes in thermodynamic parameters for R30 binding to apo-CaM as a function of temperature.

View this table:
Table 1 Thermodynamic parameters of R30 binding to apo-CaM

A change in the association rate constant with temperature is an indication of the activation energy in the transition state of R30 upon binding to apo-CaM. This energy was determined from Eyring plots [21]. In the transition state, the change in activation Gibbs free energy (ΔG) for R30 bound to apo-CaM was 49 kJ·mol−1 at 37°C (310.15 K). The activation energy of binding was derived from the entropic effect, as −TΔS, 153.1 kJ·mol−1, whereas activation enthalpy, ΔH, was negative, −104.1 kJ·mol−1 at 37°C (310.15 K). The negative activation heat capacity, ΔCp (−8.0 kJ·K−1·mol−1) reflected a decrease in hydrophobic hydration and/or conformational change (partial folding) of R30 upon CaM binding.

The ΔG upon CaM binding at equilibrium was derived from eqn 2. ΔCp was obtained by non-linear least-squares fitting of eqn 3 to temperature as a factor of ΔG. Characteristics of R30 binding to apo-CaM at 300.15 K (27°C) were as follows: ΔCp, −10 kJ·K−1·mol−1; ΔG°, −39.1 kJ·mol−1; ΔH, −21.3 kJ·mol−1; and −TΔS°, −17.8 kJ·mol−1 (Figure 5B and Table 1). The magnitude of the negative heat capacity change ΔCp, −10 kJ·K−1·mol−1, observed in the present study is typical of that observed during protein folding/unfolding [26]. This observation once again strongly supports the notion that the structure of R30 is dramatically changed by apo-CaM binding. Changes in structure could involve a decrease in the hydrophobic hydration and/or a conformational change of R30 upon apo-CaM binding.

Importantly, there were no significant differences for thermodynamic parameters between R30 binding to apo-CaM or Ca2+–CaM, and thus in both transient and stable states (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/440/bj4400367add.htm). For example, ΔCp values were very similar when R30 was bound to either Ca2+–CaM or apo-CaM (~−10 kJ·K−1·mol−1).

DISCUSSION

In the present study, we have found strong evidence for the structure of R30 being stabilized upon apo-CaM binding. Our conclusion is based on the results of a set of sophisticated biophysical and biochemical analyses. Highly significant findings regarding apo-CaM binding to R30 include: (i) stabilization of the β-sheet-rich C-lobe of R30 by apo-CaM as shown by site-specific binding assays and FTIR analysis; and (ii) folding of R30 upon apo-CaM binding as documented by the thermodynamic parameters derived from the kinetic analysis.

The thermodynamic parameters support the notion that apo-CaM may induce conformational changes in R30. The structural analysis of various complexes consisting of apo-CaM and target peptides has been reported [2729]. The N- and C-lobes of apo-CaM have been shown to wrap the target peptide as in the case of the Ca2+–CaM complex [30]. We hypothesize that, in the case of R30, apo-CaM may interact with the surface of R30 and pull out the pep11 helix usually masked in the non-complexed structure as shown in Figure 6. A flexible loop sequence, L256PS, upstream of pep11 [15] could acts as a hinge during this process. Thermodynamic parameters, particularly the negative value for ΔCp (−10 kJ·K−1·mol−1) have indicated that R30 undergoes conformational changes subsequent to binding to apo-CaM. On the basis of the location of specific residues (shown in bold in the following sequence), the sequence of pep11, AKKLWKVCVEHHTFFRL, has been classified as a 1-5-10 motif [31]. A 1-5-10 motif has been described to act as a Ca2+-dependent CaM-binding site in CaMKI (Ca2+/CaM-dependent protein kinase I) and in Hsp90 (heat-shock protein 90) [31]. In that respect, R30 pep11 is an exception as it constitutes a Ca2+-independent CaM-binding site. Although pep11 is a Ca2+-independent CaM-binding sequence, it does not contain the IQ motif [29]. Further investigation, such as X-ray diffraction analysis of crystals of the R30–apo-CaM complex, should enable us to elucidate the unique structural properties of this complex.

Figure 6 Proposed model for the structure of the R30–apo-CaM complex

The unstable C-lobe through which pep11 passes diagonally and the stable N- and α-lobes are shown. X-ray crystallization of R30 reveals that pep11 is part of a compact α-helix in the C-lobe. Apo-CaM pulls this α-helix out to form a more structurally stable complex than R30.

DLS is a sensitive method to detect changes in the hydrodynamic diameter of proteins with exposed amino acid hydrophobic side chains upon heat denaturation [23]. Using this technique, we could not detect any difference in the distribution of hydrodynamic diameters for R30 in the presence or absence of apo-CaM when varying the temperature (see Supplementary Figure S3). The aggregation of R30 may involve a portion of the hydrophobic surface that does not mediate the binding of R30 to apo-CaM. Alternatively, the apparent distribution of the hydrophobic surface of R30 may not change upon interaction with apo-CaM. If so, and since FTIR measurements and Iasys® binding assays indicate a potential change in the β-sheet structure of R30 with temperature, this would indicate that the region of the β-sheet structure affected by temperature is not situated at the surface of R30. The region of the β-sheet structure of R30 that is stabilized by apo-CaM still remains to be identified. Mapping of this region will require additional studies of the structure of the R30–apo-CaM complex.

Examination of chain mobility may lead to a better understanding of the mechanism for stabilization. B-factors, or ‘temperature factors’, are related to the amplitude of the motion of atoms, with greater B-factors indicating more extensive atom disorder [32,33]. Smith et al. [34] have studied the mobility of the ezrin FERM domain and we have characterized that of R30 using the same method. As is evident from Supplementary Figures S5(A) and S5(B) (http://www.BiochemJ.org/bj/440/bj4400367add.htm), B-factors along the chain of R30 differ in a complex manner. Local peaks of mobility are associated primarily with loops between secondary-structural elements and termini (see Supplementary Figure S1). The N- and α-lobes in R30 constitute a rather tightly knitted unit, deviating only slightly from the mean B-factor. In contrast, the C-lobe shows less cohesiveness and possesses greater average mobility (see Supplementary Figure S5C).

To date, CaM has been viewed as an important ‘regulator’ of Ca2+-mediated signal transduction in cells [35,36]. The present study represents the first report of a biological function for apo-CaM in stabilizing and/or promoting folding of one of its binding partners. We have shown previously that the FERM domain of 4.1R in zebrafish (Danio rerio) binds to CaM in a Ca2+-independent manner, but that Ca2+–CaM is unable to regulate 4.1R FERM domain binding to transmembrane proteins since the key residue for Ca2+-dependent regulation, serine, is replaced with proline in zebrafish [14]. Although the biological significance of CaM binding to the FERM domain of zebrafish 4.1R remains to be determined, it is important to note that the FERM domain of zebrafish 4.1R requires CaM binding for structural stabilization. Ezrin and moesin, two members of the FERM domain-containing protein family, also bind to apo-CaM [37]. The three-dimensional structures of ezrin and moesin FERM domains closely resemble that of 4.1R FERM domain [15,37]. Although ezrin and moesin FERM domains form a ternary complex with apo-CaM and L-selectin, the FERM domain interacts directly only with apo-CaM [37]. Thus the requirement for CaM-mediated stabilization of the FERM domain may be a general feature among members of the protein 4.1 family of proteins. As stated earlier, the FERM domain appears to adopt a unique behaviour with regard to Ca2+-dependency of regulatory properties mediated by CaM. The goal of our future studies is to decipher the mechanisms responsible for these unique properties.

The CaM concentration in erythrocytes is 2–8 μM, whereas that of ionic Ca2+ is below the submicromolar range [13]. Moreover, one molecule of CaM requires four molecules of Ca2+ for saturation. Therefore, under physiological conditions, CaM is mostly unsaturated (Ca2+-free) in erythrocytes. Given a mean volume of ~100 fl for erythrocytes and a mean concentration of CaM of 5 μM [13], one can estimate the copy number of CaM as 3×105 molecules per erythrocyte and that of 4.1R80 as 2×105 molecules [13]. This means that all 4.1R80 molecules can recruit CaM in erythrocytes. We have shown in the present study that the recruitment of apo-CaM may confer on 4.1R80 stabilization of its structure. One can hypothesize that the remaining 105 molecules of CaM may bind to (and potentially stabilize) other target proteins, such as Ca2+-ATPase or the membrane skeletal protein adducin [38,39].

A p55-binding site has been initially identified in the exon 10-encoded region of R30 C-lobe [7]. However, the L37EEDY sequence in the N-lobe, shown to mediate R30 interaction with band 3, has been proposed as an alternative p55-binding site [40,41]. The present study strongly supports the notion that the p55-binding site resides in the C-lobe of R30. Our findings should aid in selecting for appropriate constructs to be used in ongoing crystallization studies of the R30–apo-CaM complex, and may have general implications for subcellular targeting of other membrane-associated FERM domain-containing proteins.

AUTHOR CONTRIBUTION

Wataru Nunomura conceived the study, designed the experiments, performed biochemical experiments, analysed and interpreted the data, and wrote the paper. Daisuke Sasakura performed FTIR measurements. Kohei Shiba performed DLS measurements. Shigeyoshi Nakamura and Shun-ichi Kidokoro analysed the thermodynamic parameters and edited the paper before submission. Yuichi Takakuwa contributed to editing of the paper before submission.

FUNDING

This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education Culture, Sport, Science and Technology of Japan [grant number 15570123 (to W.N.)].

Acknowledgments

We thank Dr Philippe Gascard (Department of Pathology, University of California, San Francisco, San Francisco, CA, U.S.A.) for a critical reading and editing of the paper before submission.

Abbreviations: ATR, attenuated total reflection; CaM, calmodulin; DLS, dynamic light scattering; FTIR, Fourier-transform IR; GPC, glycophorin C; GST, glutathione transferase; 4.1R80, 80 kDa isoform of protein 4.1R; R30, N-terminal 30 kDa FERM domain of protein 4.1R; RMD, resonant mirror detection

References

View Abstract