Membrane skeletal protein 4.1R is the prototypical member of a family of four highly paralogous proteins that include 4.1G, 4.1N and 4.1B. Two isoforms of 4.1R (4.1R135 and 4.1R80), as well as 4.1G, are expressed in erythroblasts during terminal differentiation, but only 4.1R80 is present in mature erythrocytes. Although the function of 4.1R isoforms in erythroid cells has been well characterized, there is little or no information on the function of 4.1G in these cells. In the present study, we performed detailed characterization of the interaction of 4.1G with various erythroid membrane proteins and the regulation of these interactions by calcium-saturated calmodulin. Like both isoforms of 4.1R, 4.1G bound to band 3, glycophorin C, CD44, p55 and calmodulin. While both 4.1G and 4.1R135 interact with similar affinity with CD44 and p55, there are significant differences in the affinity of their interaction with band 3 and glycophorin C. This difference in affinity is related to the non-conserved N-terminal headpiece region of the two proteins that is upstream of the 30 kDa membrane-binding domain that harbours the binding sites for the various membrane proteins. The headpiece region of 4.1G also contains a high-affinity calcium-dependent calmodulin-binding site that plays a key role in modulating its interaction with various membrane proteins. We suggest that expression of the two paralogues of protein 4.1 with different affinities for band 3 and glycophorin C is likely to play a role in assembly of these two membrane proteins during terminal erythroid differentiation.

INTRODUCTION

Protein 4.1R [1] is the prototypical member of a family of four highly paralogous proteins that include 4.1G (generally expressed) [2], 4.1N (neuronal type) [3] and 4.1B (brain type) [4]. There are two major isoforms of 4.1R: the 135 kDa isoform (4.1R135) and the 80 kDa isoform (4.1R80) which lacks the HP (headpiece) region [5,6]. We previously reported that both 4.1R135 and 4.1R80 isoforms interact with cytoplasmic domains of GPC (glycophorin C) [7], band 3 [8] and CD44 [9], and with p55 [10,11], and that the 30 kDa domain of the proteins is responsible for these interactions. Interestingly, we noted that the interaction of 4.1R135 with GPC was an order of magnitude weaker than that of 4.1R80, whereas its interaction with band 3 was an order of magnitude stronger [12], indicating possible regulation of the interactions by the HP region. 4.1G and both isoforms of 4.1R are expressed in erythroblasts during terminal differentiation [13], but only 4.1R80 is present in mature erythrocytes. Although the function of 4.1R isoforms in erythroid cells has been well characterized [12,14], there is little or no information on the function of 4.1G in these cells. The primary structure of 4.1G is very similar to that of 4.1R135 (Figure 1). The primary amino acid sequence of the 30 kDa FERM (4.1, ezrin, radixin, moesin) domain of 4.1G is 71% identical with that of 4.1R. 4.1G is therefore predicted to bind to many of the 4.1R-binding partners described above. In contrast with the high conservation of the 30 kDa domains, the amino acid sequence identity of the HP region of 4.1G and 4.1R135 is quite low (35%). We hypothesize that the HP region may confer some distinct binding properties on the 30 kDa domain of 4.1G.

Structure of 4.1G, 4.1R and various 4.1G constructs used in the present study

Figure 1
Structure of 4.1G, 4.1R and various 4.1G constructs used in the present study

4.1R135 consists of a 209-amino-acid HP region, a 298-amino-acid 30 kDa domain (FERM domain), a 10 kDa spectrin–actin-binding domain (SAB) and a 22/24 kDa C-terminal domain (CTD). The HP is also referred to as unique region 1 (U1) whereas the 16 kDa domain between the 30 kDa domain and the SAB, and the domain between the SAB and the CTD are referred to as unique region 2 (U2) and 3 (U3) respectively. The apparent molecular masses of the SAB and CTD are 10 kDa and 22/24 kDa based on the migration of fragments obtained after α-chymotryptic digestion of 4.1R80 [1]. The GenBank® accession number for human 4.1R135 is P11171. The primary structure of 4.1G is registered under GenBank® accession number AAC16923. The amino acid sequence identity between 4.1R135 and 4.1G HPs (RHP and GHP respectively) and between the 30 kDa domains of 4.1R135 and 4.1G (G30) is displayed (35% and 71% respectively). Various domains of 4.1R135 and 4.1G used in binding assays are depicted including a chimaeric protein corresponding to the 4.1R135 HP and 4.1G 30 kDa domain, RHP–G30. The amino acid sequences at the boundary of each domain of this chimaeric protein are displayed, with the amino acid numbering for each domain referring to the protein of origin.

Figure 1
Structure of 4.1G, 4.1R and various 4.1G constructs used in the present study

4.1R135 consists of a 209-amino-acid HP region, a 298-amino-acid 30 kDa domain (FERM domain), a 10 kDa spectrin–actin-binding domain (SAB) and a 22/24 kDa C-terminal domain (CTD). The HP is also referred to as unique region 1 (U1) whereas the 16 kDa domain between the 30 kDa domain and the SAB, and the domain between the SAB and the CTD are referred to as unique region 2 (U2) and 3 (U3) respectively. The apparent molecular masses of the SAB and CTD are 10 kDa and 22/24 kDa based on the migration of fragments obtained after α-chymotryptic digestion of 4.1R80 [1]. The GenBank® accession number for human 4.1R135 is P11171. The primary structure of 4.1G is registered under GenBank® accession number AAC16923. The amino acid sequence identity between 4.1R135 and 4.1G HPs (RHP and GHP respectively) and between the 30 kDa domains of 4.1R135 and 4.1G (G30) is displayed (35% and 71% respectively). Various domains of 4.1R135 and 4.1G used in binding assays are depicted including a chimaeric protein corresponding to the 4.1R135 HP and 4.1G 30 kDa domain, RHP–G30. The amino acid sequences at the boundary of each domain of this chimaeric protein are displayed, with the amino acid numbering for each domain referring to the protein of origin.

CaM (calmodulin) binds to both isoforms of 4.1R. Although its binding to 4.1R80 was Ca2+-independent [15], its binding to 4.1R135 was strongly Ca2+-dependent [12]. Ca2+-saturated CaM completely inhibited the binding of 4.1R135 to GPC and markedly reduced the affinity of its binding to band 3 [12]. The 4.1R135 has three CaM-binding sites: a Ca2+-dependent site in the HP region and two sites in the 30 kDa domain [12,14,15]. The two CaM-binding sites in the 30 kDa domain are conserved in 4.1G and 4.1R. The sequence motif S76RGLSRLFSSFLKRPKS in the HP of 4.1R135 is responsible for Ca2+-dependent CaM binding [16]. There is a similar sequence, S71RGISRFIPPWLKKQKS, in the HP of 4.1G. The questions thus arise as to whether CaM binds to all three site(s) and which site(s) would be essential for the Ca2+-dependent regulation of 4.1G binding to membrane proteins through CaM binding.

In the present study, we performed detailed characterization of the interaction of various erythroid membrane proteins with 4.1G and the regulation of these interactions by Ca2+ and CaM.

MATERIALS AND METHODS

Reagents

The pET31b(+) bacterial expression vector was purchased from Novagen. pGEX-4T2 and pGEX-6P2 bacterial expression vectors, glutathione–Sepharose 4B, CaM–Sepharose 4B, heparin–Sepharose CL-6B, Q-Sepharose, Sephacryl S-200 and Sephacryl S-300 were purchased from GE Healthcare. Restriction enzymes were purchased from New England BioLabs. SFEM (serum-free expansion medium), FBS (fetal bovine serum), SCF (stem cell factor), IL (interleukin)-3 and EPO (erythropoietin) were purchased from Stemcell Technologies. All other reagents were purchased from Wako Pure Chemical Industries and Sigma–Aldrich unless noted otherwise. IAsys® cuvettes coated with aminosilane were supplied by Affinity Sensors. CaM was purified as described previously [9].

Antibodies

Polyclonal antibodies against 4.1G, 4.1R exon 2 and 4.1R exon 13 were raised in rabbit using His-tagged recombinant 4.1G HP, synthetic 4.1R exon 2 and exon 13 peptides as antigens. The antibodies were affinity-purified on Sulfolink Coupling Gel (Thermo Fisher Scientific). The HRP (horseradish peroxidase)-conjugated anti-rabbit immunoglobulins used for immunoblot analysis were purchased from Dako. The specificity of each antibody was confirmed by Western blotting on recombinant full-length protein 4.1 isoforms, on tissues taken from the cognate knockout mice as negative controls [17].

Computational analysis of structures of the HP and 30 kDa domain

The disorder probability for the structure of each domain was calculated using the PrDOS software package (http://prdos.hgc.jp) [18]. Their secondary structure was predicted by the PAPIA software package (http://mbs.cbrc.jp/papia/). The three-dimensional structure of the 4.1G 30 kDa domain was predicted by multiple alignment using the Modeller™ software (http://salilab.org/modeller/) based on the structure of the 30 kDa domains of 4.1R (PDB code 1GG3, chainA) and 4.1B (PDB code 2HE7). The 50 potential models obtained were ranked based on evaluation function [19]. The secondary structure and corresponding amino acid sequence of FERM domains were drawn using software package DSSP (http://swift.cmbi.ru.nl/gv/dssp/) [20].

Preparation of recombinant human 4.1G and 4.1R proteins

Recombinant proteins corresponding to various combinations of the HP region of 4.1G (GHP), the 4.1G 30 kDa domain (G30) and the 4.1R 30 kDa domain (R30) were expressed as GST (glutathione transferase)-fusion proteins in Escherichia coli BL21(DE3). They included recombinant proteins corresponding to the HP and 30 kDa domains of 4.1G (GHP–G30) and to the chimaeric protein RHP–G30 (Figure 1). Preparation of recombinant 4.1R135 has been described previously [12]. A 40 ml bacterial suspension in lysis buffer [50 mM Tris/HCl (pH 8.0) containing 200 mM NaCl, 1 mM EDTA, 1 mM 2-ME (2-mercaptoethanol), 1 mM benzamidine, 2 mM PMSF, 2 mM NaF and 0.5% Triton X-100] was sonicated at 10 kHz (50% power out) for 5 min on ice using an ultrasonic homogenizer (Smurt NR-50M, Microtec Nition). After sonication, bacterial lysates were loaded on to a glutathione affinity column to purify GST-fusion proteins. Recombinant proteins were eluted from the column after cleavage of the GST tag with thrombin as described previously [9]. After desalting, proteins were further purified on heparin–Sepharose equilibrated with 50 mM Tris/HCl (pH 7.5) containing 200 mM NaCl, 1 mM EDTA and 1 mM 2-ME and Sephacryl S-300 for GHP–G30 and Sephacryl S-200 for G30 to remove contaminants and breakdown products. Sephacryl S-300 and Sephacryl S-200 were equilibrated with 50 mM Tris/HCl (pH 7.5) containing 500 mM NaCl, 1 mM EDTA, 1 mM 2-ME, 2 mM NaF and 1% glycerol (buffer A). The retention time is recorded by Akta Prime™ Plus (GE Healthcare). The purity of recombinant proteins was assessed by SDS/PAGE and Western blot analysis. Preparations of p55 and the cytoplasmic domains of band 3 (band 3cyt), GPC (GPCcyt) and CD44 (CD44cyt), have been described previously [9,11,15]. Protein concentrations were determined as described previously [9].

Cloning of the HP, 30 kDa domain and chimaeric constructs

Human RHP, R30, GHP and G30 were cloned using 5′-NsiI-XhoI-3′ sites into the pET31b(+) vector or 5′-EcoRI-XhoI-3′ sites into the pGEX-4T2 vector. Full-length human 4.1G was cloned using 5′-EcoRI-SalI-3′ sites into the pGEX-6P2 vector (the internal SalI site in the human 4.1G coding sequence being mutated prior to cloning without altering the amino acid sequence of the protein). A chimaeric protein corresponding to RHP and G30 (RHP–G30) was generated by the ‘sewing method’. Briefly, the RHP was amplified with a forward primer (primer A: 5′-TCCAGGAATTCCCATGACAACAGAGAAGAGTTTAGTGACTGAGGC-3′) creating an EcoRI site (underlined) upstream of the ATG initiation site of 4.1R (bold) and a reverse primer (primer B’: 5′-TACTAAGAGGGTCACTTTACACTGGACGTTCCTGTGTTTTCTGATTGGTTTTTGGGAAG-3′) containing the end of RHP and the beginning of the G30 (italics). In a similar fashion, G30 was amplified with a forward primer (primer B: 5′-TCCCAAAAACCAATCAGAAAACACAGGAACGTCCAGTGT-AAAGTGACCCTCTTAGAT-3′) complementary to the reverse primer B’ described above and a reverse primer (primer C: 5′-GTTCCGCTCGAGTCAAAATTTGGACCCCAAGGTCAGG-AACTTG-3′) corresponding to the end of the G30 coding sequence followed by a stop codon (bold) and a XhoI site (underlined). The two PCR products were then mixed in equal amounts and subjected to a second amplification using primers A and C. The ~1.5 kb chimaeric PCR product was digested with EcoRI-XhoI and cloned into the pGEX-4T2 vector. All constructs were checked by DNA sequencing (Elim BioPharmaceuticals) prior to expression in bacteria or mammalian cell transfection. The GenBank® accession numbers for nucleotide sequences of human 4.1R and 4.1G are BC039079 and AF027299 respectively.

Preparation of IOVs (inside-out vesicles) and binding assay

IOVs depleted of all peripheral proteins were prepared using the method of Danilov et al. [21] with minor modifications [22]. The pelleting assay has been described previously [12]. Briefly, various concentrations of G30 and GHP–G30 were incubated for 14 h at 5 °C with IOVs in TBS [Tris-buffered saline: 50 mM Tris/HCl (pH 7.5), containing 150 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 2 mM NaF and protease inhibitors (Complete®, Roche Diagnostics)]. The mixture of G30 and IOVs were laid on 150 μl of 3.3% sucrose cushion equilibrated in TBS. The sample was centrifuged at 60000 rev./min for 20 min at 4 °C in a TL-100 ultracentrifuge using a TLA-100 rotor (Beckman Coulter). The supernatant and pellet were collected, treated with sample buffer and subjected to SDS/PAGE, using 12% and 8.4% gels for G30 and GHP–G30 respectively. Proteins bound to IOVs were visualized on CBB (Coomassie Brilliant Blue) G250 (Gelcode Blue®, Pierce)-stained gels. Bound GHP–G30 to IOVs was analysed by immunoblot. The PI (photometric intensity) of bound proteins was analysed with a ChemiDoc XRS Plus® (Bio-Rad Laboratories). The PI values of corresponding bands were plotted against the concentration of added proteins. An apparent dissociation constant at equilibrium estimated from the pelleting assay is represented as K′. The results are presented as means±S.D. calculated from three different measurements.

Purification of non-tagged full-length 4.1G

Bacteria expressing full-length recombinant human 4.1G were sonicated in 50 mM Tris/HCl (pH 8.0), 200 mM NaCl, 1 mM EDTA, 2 mM EGTA, 1 mM 2-ME, 1 mM benzamidine, 1 mM PMSF and 1 mM di-isopropylfluorophosphate on ice. The bacterial lysate was desalted stepwise by saturation with 35–50% ammonium sulfate. The 35–50% fraction containing 4.1G was dialysed against 50 mM Tris/HCl (pH 8.0), 1 mM EDTA, 1 mM 2-ME (buffer B) then loaded on to a Q-Sepharose column (1 cm × 29 cm, flow rate 0.7 ml/min) pre-equilibrated with buffer B. Protein elution, monitored by absorbance at 280 nm (A280), was achieved with a linear gradient of NaCl (0–500 mM) in buffer B. The fractions enriched in 4.1G (eluted with 320 mM NaCl) were further purified on a CaM–Sepharose column to remove the ~50 kDa contaminant protein.

Binding analysis using CaM–Sepharose 4B

Full-length 4.1G and GHP were dialysed against 50 mM Tris/HCl (pH 7.5), 100 mM NaCl and 1 mM CaCl2 (buffer C). Protein samples were loaded on to a 1.6 cm × 8 cm CaM–Sepharose 4B column pre-equilibrated with buffer C and run at a flow rate of 0.2 ml/min. The column was washed with an excess of buffer C until the A280 returned to the baseline level. Bound proteins were eluted with buffer C supplemented with 500 mM NaCl or 50 mM Tris/HCl (pH 7.5), 100 mM NaCl and 5 mM EGTA.

RMD (resonant mirror detection) binding assays

Protein–protein interactions and protein–peptide interactions were measured using the IAsys® based on the RMD method (Affinity Sensors) [23]. Below, the protein or peptide immobilized on the cuvette is referred to as the ‘ligand’, whereas the protein added in solution to the cuvette is referred to as the ‘analyte’. Protein and peptide immobilization on aminosilane cuvettes has been described previously [9]. All binding assays were carried out at 25 °C with constant stirring. In experiments designed to quantify the effects of Ca2+ and CaM on 4.1G and chimaeric protein binding profiles, full-length recombinant 4.1G (50 nM–1 μM) was pre-incubated with 5 μM CaM in 50 mM Tris/HCl (pH 7.5) and 100 mM NaCl (buffer D), and either 0.1 mM EGTA or 1.1 mM CaCl2 and 1.0 mM EGTA at 25 °C for 30 min prior to binding assays with immobilized GPCcyt, band 3cyt, CD44cyt and p55 [9,11,15].

The kinetic analysis and the stoichiometry of analyte binding to ligand were calculated according to previously published equations [9,24]. In brief, a dissociation constant [termed ‘K(D)’] was calculated as

 
formula

where ka is the association rate constant and kd is the dissociation rate constant. The K(D) was obtained from the means of three to five measurements for ka and kd. The K(D) was confirmed by Scatchard plotting using the maximum binding value (Bmax) and the added molar concentrations of analyte. The Bmax was calculated from the binding profile using the software package Fastfit® version 2.1. The method for estimation of binding ratios has been described previously [15]. The Bmax of GHP–G30 represented as arc seconds was obtained from the Scatchard plot as described previously [15]. The amount of immobilized CaM on the aminosilane cuvette was determined as the difference of arc seconds between bis(sulfosuccinimidyl)suberate and CaM under equilibrium conditions. The stoichiometry of GHP–G30 binding to CaM was calculated according to the equation below described in the Method Guide of the IAsys® system. Stoichiometry of GHP–G30: CaM=(Bmax of GHP–G30/58892): (amount of immobilized CaM on aminosilane cuvette/16705), where 58892 and 16705 are apparent molecular masses (in Da) of GHP–G30 and CaM respectively. The cuvettes were re-used after cleaning with 20 mM HCl. Original binding curves could be replicated after HCl washes implying that the washing procedure did not denature the bound ligands.

In vitro culture of erythroblasts

CD34+ HSC (haematopoietic stem cell) precursor cells were purified from cord blood (permission and approval to use cord blood was from the New York Blood Center) by Percoll separation followed by a CD34 MicroBead® kit (Miltenyi Biotec). Cells were cultured using a two-phase culture system with modifications. In the first phase (days 0–6), cells (105 cells/ml) were cultured for 3 days in SFEM supplemented with 10% FBS in the presence of SCF (50 ng/ml), IL-3 (10 ng/ml), EPO (1 unit/ml), α-thioglycerol (60 μM), penicillin (100 units/ml) and streptomycin (100 μg/ml). On day 4, cells were diluted to a density of 105 cells/ml with fresh medium and the culture was continued for a further 3 days. In the second phase (days 7–13), cells were cultured at 105 cells/ml in SFEM medium supplemented with 30% FBS in the presence of EPO, α-thioglycerol and penicillin/streptomycin. Cellular morphology was assessed by cytospin on a daily basis followed by May–Grünwald Giemsa staining and light microscopy. The vast majority of the cells were proerythroblasts on day 7 and orthochromatic erythroblasts on day 13.

Calculation of molecular mass and isoelectric point

Theoretical molecular masses and isoelectric points were calculated based on peptide amino acid sequences using the software package DNASIS® (Hitachi).

RESULTS

4.1G interaction with 4.1R-binding partners

4.1G interacted in vitro with band 3cyt, GPCcyt, p55 and CD44cyt with K(D) values of the order of ~200 nM (Table 1). Importantly, the binding affinities of 4.1G for band 3cyt and GPCcyt were different from those of 4.1R135. 4.1G interacted with band 3cyt with a much lower affinity [high K(D) value] than did 4.1R135, whereas it interacted with much higher affinity with GPCcyt. These differences were mainly derived from differences in the association rate constant ka. In contrast, both 4.1G and 4.1R135 interacted with CD44cyt and p55 with similar affinities.

Table 1
4.1G binding to membrane proteins

K(D) values for the interactions of 4.1G and 4.1R with band 3cyt, GPCcyt, p55 and CD44cyt are shown. Each analyte (at 50 nM–2 μM) was applied to the ligands listed and immobilized on aminosilane cuvettes as described in the Materials and methods section. From the binding curves obtained by the RMD method, K(D) values were determined using the software package Fastfit®. ka, kd and K(D) were calculated from three independent experiments; values are means±S.D.

Analyte Ligand ka (M−1·s−1kd (s−1K(D) (nM) 
4.1G Band 3cyt (8.0±0.1) × 104 (1.4±0.3) × 10−2 185±23 
4.1R Band 3cyt (3.1±0.2) × 105 (7.1±0.2) × 10−3 23±2 
4.1G GPCcyt (5.6±0.1) × 104 (8.1±0.2) × 10−3 144±5 
4.1R GPCcyt (8.0±0.2) × 103 (1.1±0.1) × 10−2 1327±103 
4.1G p55 (4.7±0.1) × 104 (8.5±0.1) × 10−3 181±10 
4.1R p55 (1.9±0.1) × 104 (4.7±0.1) × 10−3 248±17 
4.1G CD44cyt (3.8±0.1) × 104 (6.6±0.2) × 10−3 178±19 
4.1R CD44cyt (4.2±0.2) × 104 (1.5±0.1) × 10−2 329±64 
Analyte Ligand ka (M−1·s−1kd (s−1K(D) (nM) 
4.1G Band 3cyt (8.0±0.1) × 104 (1.4±0.3) × 10−2 185±23 
4.1R Band 3cyt (3.1±0.2) × 105 (7.1±0.2) × 10−3 23±2 
4.1G GPCcyt (5.6±0.1) × 104 (8.1±0.2) × 10−3 144±5 
4.1R GPCcyt (8.0±0.2) × 103 (1.1±0.1) × 10−2 1327±103 
4.1G p55 (4.7±0.1) × 104 (8.5±0.1) × 10−3 181±10 
4.1R p55 (1.9±0.1) × 104 (4.7±0.1) × 10−3 248±17 
4.1G CD44cyt (3.8±0.1) × 104 (6.6±0.2) × 10−3 178±19 
4.1R CD44cyt (4.2±0.2) × 104 (1.5±0.1) × 10−2 329±64 

Binding affinities of G30 and GHP–G30 to these membrane proteins were very similar to those of full-length 4.1G, suggesting that 4.1G interacted with its binding partners primarily through the 30 kDa domain (G30) and GHP did not affect these binding interactions. This is in marked contrast with interactions of the 30 kDa domain of 4.1R (R30) which were significantly affected by RHP [14], indicating that while RHP influences R30 binding to its partners, GHP has little influence on G30 binding to its partners. Interestingly, a recombinant chimaeric protein consisting of RHP and G30 (RHP–G30) showed similar binding affinities to G30 and GHP–G30 (results not shown) implying significant differences in the structure and function of RHP and GHP. It should be noted that neither GHP nor RHP bound to any of these membrane proteins.

To validate 4.1G binding to intact membranes, we assayed binding of G30 or GHP–G30 to IOVs prepared from erythrocyte membranes. Scatchard plot analysis showed that these proteins bound to IOVs in a saturable manner (Figure 2). The apparent K′ values for G30 and GHP–G30 binding to IOVs were 169±67 nM and 207±49 nM respectively. These values were similar to those obtained using RMD (Table 1). These findings demonstrate that 4.1G can bind to transmembrane proteins of the erythrocyte membrane through its 30 kDa domain.

Binding profiles of the 4.1G 30 kDa domain (G30) and GHP–G30 polypeptide binding to human erythrocyte IOVs

Figure 2
Binding profiles of the 4.1G 30 kDa domain (G30) and GHP–G30 polypeptide binding to human erythrocyte IOVs

G30 and GHP–G30 at different concentrations were incubated with IOVs and the IOVs were subsequently pelleted by centrifugation. The PI of IOVs bound G30 (A) and GHP–G30 (B) in the pellet was measured by densitometry. The PI values are plotted as a function of protein concentration and the Scatchard plots are shown in the inserts.

Figure 2
Binding profiles of the 4.1G 30 kDa domain (G30) and GHP–G30 polypeptide binding to human erythrocyte IOVs

G30 and GHP–G30 at different concentrations were incubated with IOVs and the IOVs were subsequently pelleted by centrifugation. The PI of IOVs bound G30 (A) and GHP–G30 (B) in the pellet was measured by densitometry. The PI values are plotted as a function of protein concentration and the Scatchard plots are shown in the inserts.

Computational analysis of 4.1G

Based on the crystal structures of the 30 kDa domain of 4.1R (PDB code 1GG3) and of 4.1B (PDB code 2HE7), the structure of 4.1G was modelled by Modeller™ software based on the multiple alignment obtained by ClustalW (http://align.genome.jp/) [25] (Figure 3). The secondary structures in the FERM domains are conserved among 4.1G, 4.1R and 4.1B (Supplementary Figure S1 at http://www.BiochemJ.org/bj/432/bj4320407add.htm). The structure derived was very similar to that of 4.1R, except for a deviation in the N-terminal lobe due to the LFQESPEQ sequence in 4.1G. The results of this computational analysis validated the structural basis for 4.1G binding to previously defined 4.1R binding partners through the 30 kDa domain, and that the 30 kDa domain of 4.1G could interact with CaM in a Ca2+-independent manner as for the 30 kDa domain of 4.1R.

In silico prediction of the three-dimensional structure of the 4.1G 30 kDa domain

Figure 3
In silico prediction of the three-dimensional structure of the 4.1G 30 kDa domain

Predicted ribbon model (upper panel) and surface charge distribution (bottom panel) for 4.1G (left-hand column) based on the defined structure of the 4.1R 30 kDa domain (upper and lower panels in the right-hand column; PDB code 1GG3). N, α and C represent N-lobe, α-lobe and C-lobe respectively [26].

Figure 3
In silico prediction of the three-dimensional structure of the 4.1G 30 kDa domain

Predicted ribbon model (upper panel) and surface charge distribution (bottom panel) for 4.1G (left-hand column) based on the defined structure of the 4.1R 30 kDa domain (upper and lower panels in the right-hand column; PDB code 1GG3). N, α and C represent N-lobe, α-lobe and C-lobe respectively [26].

Calculation of the disorder probability of each domain was carried out based on PrDOS software analysis, a value greater than 0.5 reflecting a disordered structure, the probability of false prediction reaching only 5%. This analysis predicted a highly disordered structure for the HP region (amino acids 1–209) and, as expected, a highly ordered structure for the 30 kDa domain (amino acids 210–507). Of particular note, whereas the overall HP region adopted a disordered structure, a polypeptide (amino acids 70–80), corresponding to a previously identified Ca2+-dependent CaM-binding site [16], exhibited an ordered structure (Supplementary Figure S2 at http://www.BiochemJ.org/bj/432/bj4320407add.htm). SDS/PAGE analysis and SEC (size-exclusion chromatography) (for reviews see [2628]) lent some indirect support for the structural prediction that the HP domain is an unfolded protein. The theoretical molecular mass of GHP is 23 kDa, but the apparent molecular mass was estimated as 36 kDa by SDS/PAGE, as shown in Figure 4(C). Although the theoretical molecular masses of GHP–G30 and G30 are 57 kDa and 33 kDa respectively, they migrated as polypeptides of 75 kDa and 35 kDa respectively on SDS/PAGE (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/432/bj4320407add.htm). Although we cannot rule out that the acidic nature of GHP could affect its binding to SDS, other acidic proteins with similar isoelectric points (5.3–5.6) such as BSA or GFAP (glial fibrillary acidic protein) show no increase in apparent molecular mass [29]. Furthermore, SEC analysis, employing Sephacryl S-300, showed that GHP–G30 was eluted between IgG (150 kDa) and albumin (68 kDa) (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/432/bj4320407add.htm), further suggesting that GHP adopts an abnormal shape. Taken together, these findings suggest, but do not prove, that the GHP is an unfolded domain, whereas G30 is folded. PrDOS-based analysis of full-length 4.1G and 4.1R135 also predicted that the 30 kDa domain was the only region in these full-length proteins than can adopt an ordered structure (results not shown).

4.1G and GHP binding to CaM–Sepharose 4B

Figure 4
4.1G and GHP binding to CaM–Sepharose 4B

(A) Monitoring of non-tagged full-length 4.1G purification on SDS/PAGE (10% gel) stained with CBB (see the Materials and methods section). Lane 1, bacterial lysate; lane 2, 35% saturated ammonium sulfate precipitate; lane 3, fraction after dialysis; lane 4, 50% saturated ammonium sulfate precipitate; lane 5, 50% saturated ammonium sulfate supernatant; lane 6, active fraction eluted from the Q-Sepharose column; lane 7, flow-through fraction from the CaM–Sepharose 4B column in the presence of Ca2+, lane 8, 5 mM EGTA elution fraction from CaM–Sepharose 4B. The arrowhead indicates the position of purified full-length 4.1G. (B) Elution profile of GST and of GHP from a CaM–Sepharose 4B column in the presence of 1 mM CaCl2 and 5 mM EGTA respectively. Protein elution is monitored by absorbance at 280 nm. (C) GHP binding to a CaM–Sepharose 4B column was analysed by SDS/PAGE (12% gel) stained with CBB. Lane 1, GST–GHP fusion protein purified on a glutathione–Sepharose column; lane 2, affinity-purified thrombin-treated GST–GHP fusion protein; lanes 3, ‘1 mM CaCl2’ flow-through fractions shown in (B) enriched in GST (open arrowhead); lanes 4, 5 mM EGTA elution fractions shown in (B) and enriched in GHP (closed arrowhead). Note that GST as a control did not bind to CaM–Sepharose 4B. For (A and C), the molecular mass in kDa is indicated on the left-hand side.

Figure 4
4.1G and GHP binding to CaM–Sepharose 4B

(A) Monitoring of non-tagged full-length 4.1G purification on SDS/PAGE (10% gel) stained with CBB (see the Materials and methods section). Lane 1, bacterial lysate; lane 2, 35% saturated ammonium sulfate precipitate; lane 3, fraction after dialysis; lane 4, 50% saturated ammonium sulfate precipitate; lane 5, 50% saturated ammonium sulfate supernatant; lane 6, active fraction eluted from the Q-Sepharose column; lane 7, flow-through fraction from the CaM–Sepharose 4B column in the presence of Ca2+, lane 8, 5 mM EGTA elution fraction from CaM–Sepharose 4B. The arrowhead indicates the position of purified full-length 4.1G. (B) Elution profile of GST and of GHP from a CaM–Sepharose 4B column in the presence of 1 mM CaCl2 and 5 mM EGTA respectively. Protein elution is monitored by absorbance at 280 nm. (C) GHP binding to a CaM–Sepharose 4B column was analysed by SDS/PAGE (12% gel) stained with CBB. Lane 1, GST–GHP fusion protein purified on a glutathione–Sepharose column; lane 2, affinity-purified thrombin-treated GST–GHP fusion protein; lanes 3, ‘1 mM CaCl2’ flow-through fractions shown in (B) enriched in GST (open arrowhead); lanes 4, 5 mM EGTA elution fractions shown in (B) and enriched in GHP (closed arrowhead). Note that GST as a control did not bind to CaM–Sepharose 4B. For (A and C), the molecular mass in kDa is indicated on the left-hand side.

4.1G binding to CaM in a Ca2+-dependent manner

We next investigated whether CaM binds to 4.1G in the absence and presence of Ca2+. Both full-length 4.1G and GHP bound to a CaM–Sepharose 4B column in the presence of Ca2+ and could be eluted with 5 mM EGTA (Figure 4). The K(D) for CaM binding to 4.1G and GHP increased dramatically following chelation of Ca2+ with EGTA (Table 2). These findings established that CaM bound to the HP region of 4.1G in a Ca2+-dependent manner. These data recapitulated previous observations made for CaM binding to RHP and 4.1R135 [12].

Table 2
Ca2+-dependence of binding of 4.1G, its fragments and RHP–G30 to CaM

K(D) values for the interactions of 4.1G, recombinant proteins corresponding to the HP and the 30 kDa membrane-binding domain of 4.1G (GHP–G30), HP of 4.1G (GHP), the 30 kDa membrane binding domain of 4.1G (G30) and chimaeric protein of the HP of 4.1R and the 30 kDa FERM domain of 4.1G (RHP–G30) with CaM-immobilized aminosilane cuvettes in the absence (EGTA/CaM) or presence (Ca2+/CaM) of Ca2+ are shown. Each analyte (at 50 nM–2 μM) was incubated with CaM immobilized on aminosilane cuvettes as described in the Materials and methods section. From the binding curves obtained by the RMD method, K(D) values were determined using the software package Fastfit®. ka, kd and K(D) were calculated from three independent experiments; values are means±S.D.

Analyte Ligand ka (M−1·s−1kd (s−1K(D) (nM) 
4.1G EGTA/CaM (3.7±0.2) × 103 (8.3±0.1) × 10−3 2245±41 
 Ca2+/CaM (9.4±0.2) × 104 (5.1±0.2) × 10−3 54±3 
GHP-G30 EGTA/CaM (3.0±0.3) × 103 (8.8±0.2) × 10−3 2967±313 
 Ca2+/CaM (1.8±0.1) × 105 (8.0±0.4) × 10−3 44±3 
GHP EGTA/CaM (3.7±0.2) × 103 (8.0±0.1) × 10−2 21644±628 
 Ca2+/CaM (1.8±0.1) × 105 (9.0±0.1) × 10−3 62±8 
G30 EGTA/CaM (8.8±0.3) × 103 (1.1±0.1) × 10−3 130±8 
 Ca2+/CaM (7.8±0.1) × 104 (1.6±0.1) × 10−2 190±16 
RHP-G30 EGTA/CaM (4.1±0.1) × 104 (1.8±0.1) × 10−2 441±41 
 Ca2+/CaM (1.0±0.1) × 105 (2.3±0.1) × 10−2 213±22 
Analyte Ligand ka (M−1·s−1kd (s−1K(D) (nM) 
4.1G EGTA/CaM (3.7±0.2) × 103 (8.3±0.1) × 10−3 2245±41 
 Ca2+/CaM (9.4±0.2) × 104 (5.1±0.2) × 10−3 54±3 
GHP-G30 EGTA/CaM (3.0±0.3) × 103 (8.8±0.2) × 10−3 2967±313 
 Ca2+/CaM (1.8±0.1) × 105 (8.0±0.4) × 10−3 44±3 
GHP EGTA/CaM (3.7±0.2) × 103 (8.0±0.1) × 10−2 21644±628 
 Ca2+/CaM (1.8±0.1) × 105 (9.0±0.1) × 10−3 62±8 
G30 EGTA/CaM (8.8±0.3) × 103 (1.1±0.1) × 10−3 130±8 
 Ca2+/CaM (7.8±0.1) × 104 (1.6±0.1) × 10−2 190±16 
RHP-G30 EGTA/CaM (4.1±0.1) × 104 (1.8±0.1) × 10−2 441±41 
 Ca2+/CaM (1.0±0.1) × 105 (2.3±0.1) × 10−2 213±22 

In addition to the CaM-binding site in the HP region of 4.1G, two CaM-binding sites have been mapped in the 30 kDa domain, one of them being Ca2+-independent. Indeed, G30 alone bound to CaM in a Ca2+-independent manner, as did R30 alone [14,15] (Table 2). We therefore probed a recombinant 4.1G expressing both the HP region and 30 kDa domain (GHP–G30). Recombinant GHP–G30 protein behaved like full-length 4.1G, i.e. it bound to a CaM–Sepharose 4B column in the presence of Ca2+ and was eluted in the presence of EGTA (results not shown). The K(D) for CaM binding to GHP–G30 increased dramatically in the presence of EGTA, as observed for full-length 4.1G (Table 2). We also determined the stoichiometry of GHP–G30 binding to CaM using immobilized CaM in a IAsys® system assay (Figure 5). The binding ratio was calculated as (2.2/58892):(0.7/16800)=1.1:1, where 2.2 arc seconds is the Bmax of GHP–G30, 0.7 arc seconds is the immobilized CaM concentration and 58892 Da and 16705 Da are the molecular masses of GHP–G30 and CaM respectively. The fact that the stoichiometry of binding of GHP–G30 and CaM is ~ 1:1, implies that only one of the three potential CaM-binding sites in GHP–G30 is functional in 4.1G. It is to be noted that the chimaeric protein of RHP–G30 (where GHP is replaced with the RHP of GHP–G30) completely lost its ability to regulate G30 binding to CaM by Ca2+ (Table 2). This observation strongly supported the importance of GHP on the Ca2+-dependence of CaM binding to 4.1G.

Analysis of GHP–G30 binding to CaM by RMD

Figure 5
Analysis of GHP–G30 binding to CaM by RMD

CaM was immobilized on an aminosilane cuvette. The binding assay was carried out in the presence of 1 mM Ca2+. The Bmax (arc seconds) at each GHP–G30 concentration was calculated using the software package Fastfit® and Bmax is plotted against the ratio Bmax/concentration of added GHP–G30 (Bmax/GHP–G30). The maximum binding (Bmax) at the crossing point on the x-axis was 2.4. The half-maximum binding occurred at 87 nM GHP–G30, an affinity that is similar to the K(D) value calculated from kinetic binding analysis (44 nM, Table 2). The open circle represents the Bmax in the absence of Ca2+.

Figure 5
Analysis of GHP–G30 binding to CaM by RMD

CaM was immobilized on an aminosilane cuvette. The binding assay was carried out in the presence of 1 mM Ca2+. The Bmax (arc seconds) at each GHP–G30 concentration was calculated using the software package Fastfit® and Bmax is plotted against the ratio Bmax/concentration of added GHP–G30 (Bmax/GHP–G30). The maximum binding (Bmax) at the crossing point on the x-axis was 2.4. The half-maximum binding occurred at 87 nM GHP–G30, an affinity that is similar to the K(D) value calculated from kinetic binding analysis (44 nM, Table 2). The open circle represents the Bmax in the absence of Ca2+.

Ca2+/CaM regulation of 4.1G binding to membrane proteins

In the light of these observations and our previous studies documenting Ca2+/CaM-dependent regulation of 4.1R binding to membrane proteins through its 30 kDa domain [12], we examined the Ca2+/CaM-dependent regulation of 4.1G binding to its various binding partners. As shown in Table 3, Ca2+/CaM binding to the HP of 4.1G resulted in complete inhibition of 4.1G binding to band 3cyt, CD44cyt and p55, and a significant increase in the K(D) for 4.1G binding to GPCcyt.

Table 3
Ca2+/CaM-mediated inhibition of 4.1G interactions with membrane proteins

K(D) values for the interactions of the complex of 4.1G and CaM (4.1G/CaM) with the cytoplasmic domain of band 3 (band 3cyt), with that of GPC (GPCcyt), with that of CD44 (CD44cyt) or with p55 in the absence (EGTA) or presence (Ca2+) of Ca2+ are shown. Full-length protein 4.1G (at 50 nM to 1 μM) was pre-incubated with CaM (5 μM) and either 0.1 mM EGTA (EGTA) or 1.1 mM CaCl2 and 1.0 mM EGTA (Ca2+) for 30 min at 25 °C in buffer D. The complex of 4.1G and CaM was applied to band 3cyt, GPCcyt, p55 or CD44cyt immobilized on aminosilane cuvettes. ka, kd and K(D) values were calculated from three independent experiments; values are means±S.D. No binding, no signal appeared when 2 μM analyte was applied to the ligand-immobilized cuvette.

Analyte Ligand Condition ka (M−1·s−1kd (s−1K(D) (nM) 
4.1G/CaM Band 3cyt EGTA (5.1±0.2) × 105 (6.4±0.2) × 10−2 127±6 
  Ca2+ No binding No binding No binding 
 GPCcyt EGTA (8.4±0.2) × 104 (1.3±0.2) × 10−2 155±22 
  Ca2+ (1.2±0.1) × 104 (1.5±0.1) × 10−2 1251±21 
 p55 EGTA (1.1±0.2) × 105 (2.3±0.2) × 10−2 215±43 
  Ca2+ No binding No binding No binding 
 CD44cyt EGTA (7.1±0.2) × 104 (1.4±0.2) × 10−2 197±29 
  Ca2+ No binding No binding No binding 
Analyte Ligand Condition ka (M−1·s−1kd (s−1K(D) (nM) 
4.1G/CaM Band 3cyt EGTA (5.1±0.2) × 105 (6.4±0.2) × 10−2 127±6 
  Ca2+ No binding No binding No binding 
 GPCcyt EGTA (8.4±0.2) × 104 (1.3±0.2) × 10−2 155±22 
  Ca2+ (1.2±0.1) × 104 (1.5±0.1) × 10−2 1251±21 
 p55 EGTA (1.1±0.2) × 105 (2.3±0.2) × 10−2 215±43 
  Ca2+ No binding No binding No binding 
 CD44cyt EGTA (7.1±0.2) × 104 (1.4±0.2) × 10−2 197±29 
  Ca2+ No binding No binding No binding 

On the basis of the fact that CaM bound to the 30 kDa domain on 4.1R80 in the absence of Ca2+ and decreased K(D) for its binding partners in a Ca2+-dependent manner [14,15], we also examined the effect of CaM binding to G30 on its binding to membrane proteins using RHP–G30. Binding affinities of RHP–G30 for band 3cyt, GPCcyt, p55 and CD44cyt were measured in the presence or absence of Ca2+ and CaM (Table 4). The K(D) obtained for each binding partner in the absence of CaM was similar to that obtained with full-length 4.1G (Table 1). In contrast, binding assays performed with RHP–G30 pre-incubated with Ca2+/CaM showed a major decrease in binding affinity [7–10-fold decrease in K(D)] of RHP–G30 for band 3cyt, GPCcyt and p55, and a less pronounced decrease [2.5-fold decrease in K(D)] for CD44cyt. These results indicated that although CaM can bind to G30 independently of Ca2+, G30 interactions with membrane proteins were regulated by CaM in a Ca2+-dependent manner. The RHP–G30 indeed co-pelleted with IOVs (see Supplementary Figure S5 at http://www.BiochemJ.org/bj/432/bj4320407add.htm).

Table 4
Ca2+-dependent CaM regulation of RHP–G30 binding to membrane proteins

K(D) values for the interactions of the complex of RHP–G30 and CaM complex (RHP–G30/CaM) with the cytoplasmic domain of band 3 (band 3cyt), with that of GPC (GPCcyt), with that of CD44 (CD44cyt) or with p55 in the absence (EGTA) or presence (Ca2+) of Ca2+ are shown. RHP–G30 as an analyte (at 50 nM–2.5 μM) was incubated with CaM (5 μM) in either 0.1 mM EGTA (EGTA) or 1.1 mM CaCl2 and 1.0 mM EGTA (Ca2+) for 30 min at 25 °C. The RHP–G30/CaM complex was applied to band 3cyt, GPCcyt, p55 or CD44cyt immobilized on aminosilane cuvettes. From the binding curves obtained by the RMD method, K(D) values were determined using the software package Fastfit®. ka, kd and K(D) values were calculated from three independent experiments; values are means±S.D.

Analyte Ligand Condition ka (M−1·s−1kd (s−1K(D) (nM) 
RHP–G30/CaM Band 3cyt EGTA (6.4±0.4) × 104 (1.2±0.2) × 10−2 183±32 
  Ca2+ (2.0±0.2) × 104 (4.3±0.4) × 10−2 2188±78 
 GPCcyt EGTA (6.5±0.3) × 104 (1.5±0.3) × 10−2 237±23 
  Ca2+ (1.4±0.1) × 104 (1.9±0.4) × 10−2 1478±182 
 p55 EGTA (3.0±0.2) × 105 (3.2±0.4) × 10−2 113±12 
  Ca2+ (6.8±0.4) × 103 (1.2±0.1) × 10−2 1763±43 
 CD44cyt EGTA (5.1±0.2) × 104 (1.6±0.4) × 10−2 314±41 
  Ca2+ (2.5±0.2) × 104 (1.9±0.1) × 10−2 737±50 
Analyte Ligand Condition ka (M−1·s−1kd (s−1K(D) (nM) 
RHP–G30/CaM Band 3cyt EGTA (6.4±0.4) × 104 (1.2±0.2) × 10−2 183±32 
  Ca2+ (2.0±0.2) × 104 (4.3±0.4) × 10−2 2188±78 
 GPCcyt EGTA (6.5±0.3) × 104 (1.5±0.3) × 10−2 237±23 
  Ca2+ (1.4±0.1) × 104 (1.9±0.4) × 10−2 1478±182 
 p55 EGTA (3.0±0.2) × 105 (3.2±0.4) × 10−2 113±12 
  Ca2+ (6.8±0.4) × 103 (1.2±0.1) × 10−2 1763±43 
 CD44cyt EGTA (5.1±0.2) × 104 (1.6±0.4) × 10−2 314±41 
  Ca2+ (2.5±0.2) × 104 (1.9±0.1) × 10−2 737±50 

The Ca2+-concentration-dependence of the CaM-modulated interaction of GHP–G30 with band 3cyt and GPCcyt was measured (Figure 6). At Ca2+ concentrations greater than 0.01 μM (pCa=8), the extent of GHP–G30 binding to band 3cyt and GPCcyt started to decline, and maximal inhibition of the binding was noted at Ca2+ concentrations of 100 μM and higher (pCa=4). A half-maximal effect was seen at a Ca2+ concentration of ~0.1 μM (pCa=7). In the absence of CaM, the binding affinities were not altered (results not shown).

Ca2+-concentration-dependence of 4.1G (GHP–G30) binding to the cytoplasmic domain of band 3 and GPC

Figure 6
Ca2+-concentration-dependence of 4.1G (GHP–G30) binding to the cytoplasmic domain of band 3 and GPC

GHP–G30 binding to the cytoplasmic domain of band 3 (●) and GPC (○) was measured at various concentrations of Ca2+, in the presence of 5 μM CaM. Ca2+ concentrations were maintained by a Ca2+/EGTA buffer system. The maximal extent of binding under different experimental conditions was quantified as described in the Materials and methods section. Maximal binding in the presence of EGTA was used to normalize the extent of binding under different experimental conditions. pCa represents the ionized Ca2+ concentration. The extent of GHP–G30 binding to the cytoplasmic domain of band 3 and GPC is plotted as a function of Ca2+ concentration. In the absence of CaM, there was no change in the binding of GHP–G30 to the cytoplasmic domain of the transmembrane proteins as a function of Ca2+ concentration (results not shown).

Figure 6
Ca2+-concentration-dependence of 4.1G (GHP–G30) binding to the cytoplasmic domain of band 3 and GPC

GHP–G30 binding to the cytoplasmic domain of band 3 (●) and GPC (○) was measured at various concentrations of Ca2+, in the presence of 5 μM CaM. Ca2+ concentrations were maintained by a Ca2+/EGTA buffer system. The maximal extent of binding under different experimental conditions was quantified as described in the Materials and methods section. Maximal binding in the presence of EGTA was used to normalize the extent of binding under different experimental conditions. pCa represents the ionized Ca2+ concentration. The extent of GHP–G30 binding to the cytoplasmic domain of band 3 and GPC is plotted as a function of Ca2+ concentration. In the absence of CaM, there was no change in the binding of GHP–G30 to the cytoplasmic domain of the transmembrane proteins as a function of Ca2+ concentration (results not shown).

Expression of 4.1G and 4.1R135 in human erythroblasts

To establish the biological significance of our biochemical findings, we studied the expression pattern of 4.1G and 4.1R135 during terminal differentiation of human erythroblasts (Figure 7). Western blot analysis using antibodies specific for 4.1R and 4.1G showed that both 4.1G and 4.1R135 were highly expressed in proerythroblasts (day 7 of culture), but their expression level was markedly decreased in orthochromatic erythroblasts (day 13 of culture). In contrast, the expression level of 4.1R80 that was low in proerythroblasts was markedly elevated in orthochromatic erythroblasts.

Expression of 4.1G, 4.1R135 and 4.1R80 during human erythroblast differentiation

Figure 7
Expression of 4.1G, 4.1R135 and 4.1R80 during human erythroblast differentiation

(A) Immunoblot analysis of protein extracts from human erythroblasts at early (day 7) and late (day 13) stages of differentiation (see the Materials and methods section) using antibodies specific for 4.1G and 4.1R. The molecular mass in kDa is indicated on the left-hand side. (B) Cellular morphology was assessed by cytospin on a daily basis followed by May–Grünwald Giemsa staining and light microscopy (magnification is ×100).

Figure 7
Expression of 4.1G, 4.1R135 and 4.1R80 during human erythroblast differentiation

(A) Immunoblot analysis of protein extracts from human erythroblasts at early (day 7) and late (day 13) stages of differentiation (see the Materials and methods section) using antibodies specific for 4.1G and 4.1R. The molecular mass in kDa is indicated on the left-hand side. (B) Cellular morphology was assessed by cytospin on a daily basis followed by May–Grünwald Giemsa staining and light microscopy (magnification is ×100).

DISCUSSION

In the present study, we demonstrated that 4.1G binds through its 30 kDa domain to various previously defined binding partners for 4.1R, including transmembrane proteins such as band 3, GPC, CD44 and p55. These interactions were not affected by the HP region, but Ca2+-dependent CaM binding to the HP region had a profound effect on the interaction of 4.1G with its binding partners. The documented binding profiles of 4.1G are markedly different from that previously reported for 4.1R135 [12]. Since the primary structure of the 30 kDa domain of 4.1G and 4.1R is highly conserved (71% sequence identity), the differences in binding profiles arise primarily from the non-conserved HP region. We provide here extensive support for this hypothesis through the use of several recombinant proteins, GHP, G30 and GHP–G30, as well as the chimaeric protein RHP–G30.

Computational calculation indicates that the three-dimensional structure of the 30 kDa domain of 4.1G is very similar to that of 4.1R, a clover-like structure adopting a globular conformation [30]. The high similarity of three-dimensional structure is supported by in vitro binding assays. Strikingly, the binding sites for each major binding partner of this domain are located on a different lobe of the clover-like structure [26]: band 3 on the N-lobe [31], GPC on the α-lobe [11] and p55 [11] and CD44 on the C-lobe [32]. We previously showed an important role for the HP region in regulating 4.1R135 30 kDa domain binding to membrane proteins [12]; the HP region itself improved accessibility of the N-lobe to band 3, impaired accessibility of the α-lobe to GPC and did not have much influence on the C-lobe [12]. In contrast, the HP of 4.1G does not seem to affect the 30 kDa domain interactions with membrane proteins, the binding profile of 4.1G and GHP–G30 being similar to that of G30. This indicates that in contrast with the HP region of 4.1R135, the HP region of 4.1G has little effect on the binding sites in the 30 kDa domain of 4.1G. The calculated isoelectric points of GHP and GHP–G30 are 5.16 and 6.65, whereas those of RHP and RHP-R30 are 4.46 and 5.30 respectively. Thus differences in electrostatic interactions may account for the documented differences in the affinity of interaction of 4.1G and 4.1R135 with band 3cyt and GPCcyt. Considering that the HP region of both 4.1G and 4.1R135 appears to adopt an intrinsically disordered structure, the spatial organization of the HP region and the 30 kDa domain must be different in 4.1G and 4.1R135. The present study highlights that GHP–G30 and RHP–G30 represent unique structural and functional modules, despite the high amino acid sequence identity of their 30 kDa domains and the conservation of CaM-binding sites. Delineation of the structure of the HP region alone and in combination with the 30 kDa domain will be critical for validating these hypotheses.

The present results indicate that CaM bound to the HP region of 4.1G in a Ca2+-dependent manner, but not to the 30 kDa domain as has been previously documented for 4.1R80 [15]. The HP region of 4.1G contains the sequence S71RGISRFIPPWLKKQKS which is 76% identical (13 out of 17 residues) with the CaM-binding site in the HP region of 4.1R (S76RGLSRLFSSFLKRPKS) [12,16]. Although the Ca2+-independent CaM-binding sequence previously identified in the 30 kDa domain of 4.1R80 is also conserved in 4.1G [15], our findings imply that CaM binds to the HP region, but not to the 30 kDa domain of 4.1G. It should be emphasized that although the HP by itself does not affect the binding to the 30 kDa domain of 4.1G to various membrane proteins, Ca2+/CaM binding to the HP markedly inhibits the ability of 4.1G to interact with its various binding partners. These findings have enabled us to document similarities and differences in the structural and functional properties of 4.1G and 4.1R135.

Our finding that the half-maximal binding of GHP–G30 to band 3cyt and GPCcyt occurred at Ca2+ concentrations in the submicromolar range suggests potential biological relevance of our biochemical findings. Ca2+/CaM modulations of 4.1G binding to membrane proteins may be modulated upon signal transduction during erythroid development, since at early stages of erythropoiesis it has been documented that, following binding of EPO to its receptor, intracellular Ca2+ levels increased to 259±49nM from a basal level of 55±5nM [33,34]. This level of increase in intracellular Ca2+ levels is sufficient to modulate the interaction of 4.1G with its binding partners in erythroid cells.

Our findings further suggest that 4.1G offers a unique opportunity to explore divergence of protein structure and function during evolution and development. In erythroblasts, we showed that 4.1G and 4.1R135 are both expressed during terminal erythroid differentiation consistent with an earlier report [13], and both of these proteins can interact with the transmembrane proteins band 3, GPC and CD44. Different binding affinities and Ca2+/CaM modulation to band 3 and GPC suggest that these 4.1 proteins may play specific roles in membrane biogenesis during terminal erythroid differentiation.

Abbreviations

     
  • CaM

    calmodulin

  •  
  • CBB

    Coomassie Brilliant Blue

  •  
  • EPO

    erythropoietin

  •  
  • FBS

    fetal bovine serum

  •  
  • FERM

    4.1, ezrin, radixin, moesin

  •  
  • GPC

    glycophorin C

  •  
  • GST

    glutathione transferase

  •  
  • HP

    headpiece

  •  
  • IL

    interleukin

  •  
  • IOV

    inside-out vesicle

  •  
  • 2-ME

    2-mercaptoethanol

  •  
  • PI

    photometric intensity

  •  
  • RMD

    resonant mirror detection

  •  
  • SCF

    stem cell factor

  •  
  • SEC

    size-exclusion chromatography

  •  
  • SFEM

    serum-free expansion medium

  •  
  • TBS

    Tris-buffered saline

AUTHOR CONTRIBUTION

Wataru Nunomura conceived the study, designed the experiments, performed biochemical experiments, analysed and interpreted the data, and wrote the manuscript. Kengo Kinoshita performed in silico structural analyses of protein 4.1G. Marilyn Parra established the bacterial system for expression of recombinant proteins. Philippe Gascard designed the chimaeric protein expression system and contributed to pre-submission editing of the manuscript. Xiuli An performed the experiments using cultured erythroblasts. Narla Mohandas and Yuichi Takakuwa contributed to the interpretation of the data and edited the manuscript 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.)]; and by the National Institutes of Health [grant numbers DK 26263, HL31579, DK 32094 (to N.M.)].

References

References
1
Diakowski
W.
Grzybek
M.
Sikorski
A. F.
Protein 4.1, a component of the erythrocyte membrane skeleton and its related homologue proteins forming the protein 4.1/FERM superfamily
Folia Histochem. Cytobiol.
2006
, vol. 
44
 (pg. 
231
-
248
)
2
Parra
M.
Gascard
P.
Walensky
L. D.
Snyder
S. H.
Mohandas
N.
Conboy
J. G.
Cloning and characterization of 4.1G (EPB41L2), a new member of the skeletal protein 4.1 (EPB41) gene family
Genomics
1998
, vol. 
49
 (pg. 
298
-
306
)
3
Walensky
L. D.
Blackshaw
S.
Liao
D.
Watkins
C. C.
Weier
H. U. G.
Parra
M.
Huganir
R. L.
Conboy
J. G.
Mohandas
N.
Snyder
S. H.
A novel neuron-enriched homolog of the erythrocyte membrane cytoskeletal protein 4.1
J. Neurosci.
1999
, vol. 
19
 (pg. 
6457
-
6467
)
4
Parra
M.
Gascard
P.
Walensky
L. D.
Gimm
J. A.
Blackshaw
S.
Chan
N.
Takakuwa
Y.
Berger
T.
Lee
G.
Chasis
J. A.
, et al. 
Molecular and functional characterization of protein 4.1B, a novel member of the protein 4.1 family with high level, focal expression in brain
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
3247
-
3255
)
5
Conboy
J. G.
Chan
J.
Mohandas
N.
Kan
Y. W.
Multiple protein 4.1 isoforms produced by alternative splicing in human erythroid cells
Proc. Natl. Acad. Sci. U. S. A.
1988
, vol. 
85
 (pg. 
9062
-
9065
)
6
Gascard
P.
Lee
G.
Coulombel
L.
Auffray
I.
Lum
M.
Parra
M.
Conboy
J. G.
Mohandas
N.
Chasis
J. A.
Characterization of multiple isoforms of protein 4.1R expressed during erythroid terminal differentiation
Blood
1998
, vol. 
92
 (pg. 
4404
-
4414
)
7
Anderson
R. A.
Lovrien
R. E.
Glycophorin is linked by band 4.1 protein to the human erythrocyte membrane skeleton
Nature
1984
, vol. 
307
 (pg. 
655
-
658
)
8
Pasternack
G. R.
Anderson
R. A.
Leto
T. L.
Marchesi
V. T.
Interactions between protein 4.1 and band 3. An alternative binding site for an element of the membrane skeleton
J. Biol. Chem.
1985
, vol. 
260
 (pg. 
3676
-
3683
)
9
Nunomura
W.
Takakuwa
Y.
Tokimitsu
R.
Kawashima
M.
Krauss
S.
Mohandas
N.
Regulation of CD44-protein 4.1 interaction by Ca2+ and calmodulin. Implications for modulation of CD44-ankyrin interaction
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
30322
-
30328
)
10
Marfatia
S. M.
Morais-Cabral
J. H.
Kim
A. C.
Byron
O.
Chishti
A. H.
The PDZ domain of human erythrocyte p55 mediates its binding to the cytoplasmic carboxyl terminus of glycophorin C. Analysis of the binding interface by in vitro mutagenesis
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
24191
-
2417
)
11
Nunomura
W.
Takakuwa
Y.
Parra
M.
Conboy
J.
Mohandas
N.
Regulation of protein 4.1R, p55, and glycophorin C ternary complex in human erythrocyte membrane
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
24540
-
24546
)
12
Nunomura
W.
Parra
M.
Hebiguchi
M.
Swada
K.
Mohandas
N.
Takakuwa
Y.
Marked difference in membrane protein binding properties of the two isoforms of protein 4.1R expressed at early and late stages of erythroid differentiation
Biochem. J.
2009
, vol. 
417
 (pg. 
141
-
148
)
13
Delhommeau
F.
Vasseur-Godbillon
C.
Leclerc
P.
Schischmanoff
P. O.
Croisille
L.
Rince
P.
Morinière
M.
Benz
E. J.
Jr
Tchernia
G.
Tamagnini
T.
, et al. 
A splicing alteration of 4.1R pre-mRNA generates 2 protein isoforms with distinct assembly to spindle poles in mitotic cells
Blood
2002
, vol. 
100
 (pg. 
2629
-
2636
)
14
Nunomura
W.
Takakuwa
Y.
Regulation of protein 4.1R interactions with membrane proteins by Ca2+ and calmodulin
Front. Biosci.
2006
, vol. 
11
 (pg. 
1522
-
1539
)
15
Nunomura
W.
Takakuwa
Y.
Parra
M.
Conboy
J.
Mohandas
N.
Ca2+-dependent and Ca2+-independent calmodulin binding sites in erythrocyte protein 4.1. Implications for regulation of protein 4.1 interactions with transmembrane proteins
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
6360
-
6367
)
16
Leclerc
E.
Vetter
S.
Characterization of a calcium-dependent calmodulinbinding domain in the 135-kD human protein 4.1 isoform
Eur. J. Biochem.
1998
, vol. 
258
 (pg. 
567
-
571
)
17
Kang
Q.
Wang
T.
Zhang
H.
Mohandas
N.
An
X.
A Golgi-associated protein 41B variant is required for assimilation of proteins in the membrane
J. Cell Sci.
2009
, vol. 
122
 (pg. 
1091
-
1099
)
18
Ishida
T.
Kinoshita
K.
PrDOS: prediction of disordered protein regions from amino acid sequence
Nucleic Acids Res.
2007
, vol. 
35
 (pg. 
W460
-
W464
)
19
Marti-Renom
M. A.
Stuart
A.
Fiser
A.
Sánchez
R.
Melo
F.
Sali
A.
Comparative protein structure modeling of genes and genomes
Annu. Rev. Biophys. Biomol. Struct.
2000
, vol. 
29
 (pg. 
291
-
325
)
20
Kabsch
W.
Sander
C.
Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features
Biopolymers
1983
, vol. 
22
 (pg. 
2577
-
2637
)
21
Danilov
Y. N.
Fennell
R.
Ling
E.
Cohen
C. M.
Selective modulation of band 4.1 binding to erythrocyte membranes by protein kinase C
J. Biol. Chem.
1990
, vol. 
265
 (pg. 
2556
-
2562
)
22
An
X.
Takakuwa
Y.
Nunomura
W.
Manno
S.
Mohandas
N.
Modulation of band 3-ankyrin interaction by protein 4.1. Functional implications in regulation of erythrocyte membrane mechanical properties
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
33187
-
33191
)
23
Cush
R.
Cronin
J. M.
Stewart
W. J.
Maule
C. H.
Molloy
J.
Goddard
N. J.
The resonant mirror: a novel optical biosensor for direct sensing of biomolecular interactions. Part I: Principle of operation and associated instrumentation
Biosens. Bioelectron.
1993
, vol. 
8
 (pg. 
347
-
353
)
24
Nunomura
W.
Takakuwa
Y.
Cherr
G. N.
Murata
K.
Characterization of protein 4.1R in erythrocytes of zebrafish (Danio rerio): unique binding properties with transmembrane proteins and calmodulin
Comp. Biochem. Physiol. Part B
2007
, vol. 
148
 (pg. 
124
-
138
)
25
Higgins
D. G.
Thompson
J. D.
Gibson
T. J.
Using CLUSTAL for multiple sequence alignments
Methods Enzymol.
1996
, vol. 
266
 (pg. 
383
-
402
)
26
Tompa
P.
Intrinsically unstructured proteins
Trends Biochem. Sci.
2002
, vol. 
27
 (pg. 
527
-
533
)
27
Radivojac
P. R.
Iakoucheva
L. M.
Oldfield
C. J.
Obradovic
Z.
Uversky
V.
Dunker
A. K.
Intrinsic disorder and functional proteomics
Biophys. J.
2007
, vol. 
92
 (pg. 
1439
-
1465
)
28
Uversky
V. N.
Dunker
A. K.
Understanding protein non-folding
Biochim. Biophys. Acta
2010
, vol. 
1804
 (pg. 
1231
-
1264
)
29
Porchet
R.
Probst
A.
Bouras
C.
Dráberová
E.
Dráber
R.
Riederer
B. M.
Analysis of glial acidic fibrillary protein in the human entorhinal cortex during aging and in Alzheimer's disease
Proteomics
2003
, vol. 
3
 (pg. 
1476
-
1485
)
30
Han
B. G.
Nunomura
W.
Takakuwa
Y.
Mohandas
N.
Jap
J. K.
Protein 41R core domain structure and insights into regulation of cytoskeletal organization
Nat. Struct. Biol.
2000
, vol. 
7
 (pg. 
871
-
875
)
31
Jöns
T.
Drenckhahn
D.
Identification of the binding interface involved in linkage of cytoskeletal protein 4.1 to the erythrocyte anion exchanger
EMBO J.
1992
, vol. 
11
 (pg. 
2863
-
2867
)
32
Hattori
M.
Nunomura
W.
Ito
E.
Ohta
H.
Takakuwa
Y.
Regulation of ankyrin interaction with CD44 by protein 4.1 in HeLa cells
Membrane
2006
, vol. 
31
 (pg. 
107
-
114
)
33
Yelamarty
R. V.
Miller
B. A.
Scaduto
R. C.
Jr
Yu
F. T. S.
Tillotson
D. L.
Cheung
J. Y.
Three-dimensional intracellular gradients in single human burst-forming units-erythroid-derived erythroblasts induced by erythropoietin
J. Clin. Invest.
1990
, vol. 
85
 (pg. 
17799
-
1809
)
34
Miller
B. A.
Scaduto
R. C.
Jr
Tillotson
D. L.
Botti
J.
Cheung
J. Y.
Erythropoietin stimulates a rise in intracellular free calcium concentartion in single early human erythroid precursors
J. Clin. Invest.
1988
, vol. 
825
 (pg. 
309
-
315
)

Supplementary data