Two major isoforms of protein 4.1R, a 135 kDa isoform (4.1R135) and an 80 kDa isoform (4.1R80), are expressed at distinct stages of terminal erythroid differentiation. The 4.1R135 isoform is exclusively expressed in early erythroblasts and is not present in mature erythrocytes, whereas the 4.1R80 isoform is expressed at late stages of erythroid differentiation and is the principal component of mature erythrocytes. These two isoforms differ in that the 4.1R135 isoform includes an additional 209 amino acids designated as the HP (head-piece) at the N-terminus of 4.1R80. In the present study, we performed detailed characterization of the interactions of the two 4.1R isoforms with various membrane-binding partners and identified several isoform-specific differences. Although both 4.1R135 and 4.1R80 bound to cytoplasmic domains of GPC (glycophorin C) and band 3, there is an order of magnitude difference in the binding affinities. Furthermore, although both isoforms bound CaM (calmodulin), the binding of 4.1R80 was Ca2+-independent, whereas the binding of 4.1R135 was strongly Ca2+-dependent. The HP of 4.1R135 mediates this Ca2+-dependent binding. Ca2+-saturated CaM completely inhibited the binding of 4.1R135 to GPC, whereas it strongly reduced the affinity of its binding to band 3. Interestingly, in spite of the absence of spectrin-binding activity, the 4.1R135 isoform was able to assemble on to the membrane of early erythroblasts suggesting that its ability to bind to membrane proteins is sufficient for its membrane localization. These findings enable us to offer potential new insights into the differential contribution of 4.1R isoforms to membrane assembly during terminal erythroid differentiation.

INTRODUCTION

Human protein 4.1R, an 80 kDa polypeptide, is a key structural component of the erythrocyte membrane skeleton and plays an essential role in maintaining cell shape and bestowing membrane mechanical stability. In mature erythrocytes, 4.1R interacts with spectrin and actin to form a junctional complex in the skeletal network and it also links the network to the membrane through interactions with GPC (glycophorin C) and band 3 [1,2]. It has been documented that a large number of 4.1R isoforms are generated through alternative splicing in a variety of cells and tissues ranging in size from 135 kDa to 60 kDa [35]. The multiple isoforms exhibit development-specific expression patterns in a variety of non-erythroid cells as well as in erythroblasts at different stages of development [5,6]. There is also marked heterogeneity in the sub-cellular localization of various 4.1R isoforms in cells. During erythropoiesis two major isoforms of protein 4.1R, a 135 kDa isoform (4.1R135) and an 80 kDa isoform (4.1R80), are expressed at distinct stages of terminal erythroid differentiation [3,4]. The 4.1R135 isoform is exclusively expressed in early erythroblasts and is not present in mature erythrocytes, whereas the 4.1R80 isoform is expressed at late stages of erythroid differentiation and is the principal component of mature erythrocytes. The erythroid 4.1R135 isoform includes an additional 209 amino acids designated as the HP (head-piece) at the N-terminus of 4.1R80, and it also lacks a stretch of 21 amino acids that are critical for the interaction of 4.1R with spectrin [6]. An extensive series of biochemical and biophysical studies have provided detailed insights into the nature of the interaction of erythroid 4.1R80 with its various binding partners and the functional sequeli of these interactions in mature erythrocytes [7]. In contrast, little or no such information is presently available on the 4.1R135 isoform expressed in erythroblasts.

Limited chymotryptic digestion of erythroid 4.1R80 generates four structural domains: the N-terminal 30 kDa membrane-binding domain {also termed the FERM (4.1/ezrin/radixin/moesin) domain [8]}, the 16 kDa regulatory domain, the 10 kDa SAB (spectrin- and actin-binding domain) and the 22/24 kDa CTD (C-terminal domain). In mature erythrocytes, the 30 kDa membrane-binding domain mediates the binding of 4.1R80 to the plasma membrane via GPC, band 3 and p55, a type II PDZ protein. The crystal structure of the 30 kDa domain of 4.1R has been determined and is found to be composed of N-, α- and C-lobes arranged in a clover-like shape [9]. The binding sites of band 3, GPC and p55 have been mapped to the N-, α- and C-lobes respectively, and the binding domain for CaM has been localized to the central region of the structure [7]. The binding sites for spectrin and actin in the 10 kDa SAB domain have also been delineated [10]. Phosphorylation of Ser312 in the 16 kDa regulatory domain by protein kinase C has been shown to decrease the binding affinity of 4.1R80 to GPC and to the SAB domain [11].

The binding of 4.1R80 to membrane proteins was previously shown to be regulated by Ca2+/CaM (Ca2+-saturated calmodulin) [1215]. The CaM-binding site is situated on exon 9 (Ca2+-sensitive site) and exon 11 (Ca2+-insensitive site) encoded regions in the 30 kDa domain of 4.1R80. Kelly et al. [16] noted Ca2+-dependent CaM binding to the HP of 4.1R135 in Xenopus sp. Subsequently, Leclerc and Vetter [17] reported a Ca2+-dependent CaM-binding sequence, S76RGLSRLFSSFLKRPKS, in the HP of human 4.1R135 [17]. The effects of CaM binding to the HP on the 30 kDa domain binding to membrane proteins has not been defined. Based on currently available information [7,13,17] there are three CaM-binding sites in 4.1R135: a Ca2+-dependent binding site on the HP and two binding sites in the 30 kDa domain (Figure 1). The question thus arises as to whether CaM binds to all three sites and which site(s) would be essential for the Ca2+-dependent regulation of 4.1R135 binding to membrane proteins through CaM binding.

Structures of 4.1R isoforms used in the present study

Figure 1
Structures of 4.1R isoforms used in the present study

4.1R135 consists of the HP (209 amino acid residues), FERM domain (30 kDa domain containing 297 amino acid residues), SAB and CTD. U1 and U2 represent unique region 1 and 2 respectively. The apparent molecular masses of these domains are 30 kDa, 16 kDa, 10 kDa and 22/24 kDa, as determined by α-chymotryptic digestion [7]. Arrowheads indicate CaM-binding sites on protein 4.1R. 4.1R135 has two initiation codons (AUG1 and AUG2) located in exons 2 and 4. R30, RHP and RHP–R30 are recombinant proteins that were expressed as GST-fusion proteins in Esherichia coli.

Figure 1
Structures of 4.1R isoforms used in the present study

4.1R135 consists of the HP (209 amino acid residues), FERM domain (30 kDa domain containing 297 amino acid residues), SAB and CTD. U1 and U2 represent unique region 1 and 2 respectively. The apparent molecular masses of these domains are 30 kDa, 16 kDa, 10 kDa and 22/24 kDa, as determined by α-chymotryptic digestion [7]. Arrowheads indicate CaM-binding sites on protein 4.1R. 4.1R135 has two initiation codons (AUG1 and AUG2) located in exons 2 and 4. R30, RHP and RHP–R30 are recombinant proteins that were expressed as GST-fusion proteins in Esherichia coli.

In contrast with our detailed understanding of the structure of the erythroid 4.1R80 isoform, little is known with regards to the 4.1R135 isoform. In the present study, we performed a comprehensive analysis of the membrane-binding characteristics of the 4.1R135 isoform expressed in early erythroblasts. We have documented that the presence of the HP domain adjacent to the membrane-binding domain in 4.1R135 modulates its functional characteristics.

MATERIALS AND METHODS

Materials

pET31b and pGEX-4T2 bacterial expression vectors were purchased from Novagen and GE Healthcare Sciences respectively. Glutathione–Sepharose CL-6B and CaM–Sepharose were from GE Healthcare Sciences. All other reagents were purchased from Wako Pure Chemical Industries unless otherwise noted. IAsys® cuvettes coated with aminosilane were obtained from Affinity Sensors.

Antibodies

A mouse monoclonal antibody against GPC (#M820) was from Dako and a mouse monoclonal antibody against band 3 (extracellular region) was from Sigma–Aldrich. Alexa Fluor® 546-conjugated goat IgG to mouse immunoglobulins was from Invitrogen. FITC-conjugated rabbit IgG to goat immunoglobulins was from Dako. A rabbit polyclonal antibody to the HP region of 4.1R for specific detection of the 4.1R135 isoform and a rabbit polyclonal antibody to p55 were prepared in our laboratory. Antibody specificity was assessed using an immunoblot assay using purified proteins.

Preparation of recombinant proteins

Recombinant proteins of full-length 4.1R135 and 4.1R80 were expressed as non-tagged proteins employing a modified pET31b vector. cDNA for the various proteins corresponding to different functional domains were cloned into a pGEX-4T2 vector and DNA sequencing was carried out to validate the fidelity of the cloned sequence. Recombinant protein expression was induced by the addition of 1 mM IPTG (isopropyl β-D-thiogalactoside) to the bacterial culture. The recombinant proteins were purified as follows: first, proteins were fractionated using 35% saturated ammonium sulfate, the protein bound to Q-Sepharose was equilibrated with 50 mM Tris/HCl (pH 7.5), containing 0.1 M NaCl, 1 mM EDTA and 1 mM 2-mercaptoethanol, and finally the bound protein was eluted with a gradient of NaCl (0.6 M). The eluted protein was applied to CaM–Sepharose equilibrated with 50 mM Tris/HCl (pH 7.5), containing 0.1 M NaCl and 0.5 mM CaCl2. The bound protein was eluted with 5 mM EGTA [15]. The recombinant proteins corresponding to various domains of 4.1R were expressed as GST (glutathione transferase)-fusion proteins. These included RHP (HP of 4.1R135), R30 (recombinant protein of the 30 kDa domain of 4.1R; Met210 to Phe507), the 16 kDa domain, the 10 kDa SAB domain, the 22 kDa CTD and RHP–R30 (a hybrid of RHP and R30) (Figure 1). At the final step, the target proteins were further purified by Q-Sepharose column chromatography to remove any bacterial contamination. Preparations of recombinant band 3 cyt (the N-terminal cytoplasmic domain of band 3), GPCcyt (the cytoplasmic domain of GPC) and p55 have been described previously [13,14]. Except for the 16 kDa domain–GST fusion protein which contains a thrombin-sensitive site, GST was cleaved from all other recombinant proteins by thrombin. The purity of the expressed proteins was assessed using SDS/PAGE analysis. The protein concentration was determined as previously reported [12].

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

IOVs depleted of peripheral protein (native IOV) and trypsinized IOVs were prepared as described by Danilov et al. [18] with minor modifications. SDS/PAGE of trypsinized IOVs indicated that band 3cyt was completely digested in this IOV preparation; however, in these IOVs GPCcyt remained intact [19]. The pelleting assay was performed as described previously [15]. Briefly, various concentrations of 4.1R proteins were incubated for 20 min at room temperature (25 °C) with IOVs in TBS [50 mM Tris/HCl (pH 7.5) containing 0.15 M NaCl]. The 4.1R proteins and IOV mixture were laid on to 150 μl of a 3.3% sucrose cushion equilibrated in TBS containing 1 mM EDTA and 1 mM dithiothreitol. The sample was centrifuged at 60000 rev./min for 10 min at 4 °C in a TL-100 ultracentrifuge using a TLA-100 rotor (Beckman Coulter). The supernatant and precipitate were collected, treated with sample buffer and subjected to SDS/PAGE. Proteins bound to IOVs were visualized on CBB (Coomassie Brilliant Blue G250, Gelcode Blue®; Pierce)-stained gels and analysed with a Digital Science EDAS290 System® (Kodak). An apparent dissociation constant at equilibrium estimated from the pelleting assay is represented as K′.

Preparation of CaM

CaM purification was performed as descibed previously [13].

Computer analysis

The theoretical molecular mass and isoelectric point (pI) of the proteins was computed based on the amino acid sequences, using the software package DNASIS® (Hitachi).

Binding assay by RMD (resonant mirror detection)

Protein–protein and protein–peptide interactions were examined by RMD of the IAsys® system (Affinity Sensors) [20,21]. Immobilized protein or peptide on the cuvette is referred to in the present study as the ‘ligand’ and protein in solution added to the cuvette is referred to as the ‘analyte’. The immobilization of proteins on the aminosilane cuvette has been described previously [12]. The binding assay was conducted in all cases in PBS at 25 °C with constant stirring. To quantify the effects of Ca2+ and CaM on the binding of 4.1R135 to different ligands, GPCcyt, band 3cyt and p55 were immobilized on the cuvette surface. 4.1R135 (1 nM–approx. 2 μM) was pre-incubated with 5 μM CaM in buffer [50 mM Tris/HCl (pH 7.5), 0.1 M NaCl with 0.1 mM EGTA or 1.1 mM CaCl2 and 1.0 mM EGTA] at 25 °C for 30 min. The binding assays used the same buffers [12,15].

Kinetic analysis and analysis of the ratio of analyte binding to ligand were carried out as previously described [12,15]. The dissociation constant in kinetic analysis [termed KD] was calculated as KD=kd/ka, where ka is the association rate constant and kd is the dissociation rate constant. The reported ka, kd and KD values are represented as means±S.D. from three separate measurements.

Binding ratio determination using the QCM (quartz crystal micro-balance) method

The QCM method Affinix® (Initium) was used to determine the stochiometry of 4.1R135 binding to CaM by directly measuring the changes in mass of immobilized CaM on the sensor chip following the binding of 4.1R135 [22]. CaM was immobilized on the sensor chip according to the Affinix® instructions. The maximal binding value (Bmax) of 4.1R135 to immobilized CaM on the sensor chip was determined. The theoretical molecular masses of 4.1R135 (93305.77 Da) and CaM (16705.54 Da) were used for estimating the binding ratio.

Cells and ex vivo generation of erythroid progenitor cells

GM-CSF (granulocyte/macrophage colony-stimulating factor)-mobilized human peripheral blood CD34+ cells were purified from healthy volunteers, who had signed informed consent forms approved by the Akita University School of Medicine Committee for the Protection of Human Subjects, as previously described [23]. The purified cells were stored in liquid nitrogen until use. The thawed CD34+ cells were suspended in IMDM (Iscove's modified Dulbecco's medium; Invitrogen) containing 30% (v/v) FCS (fetal calf serum) and 100 units/ml DNase, and then were washed twice with IMDM containing 20% (v/v) FCS. The cells were seeded in 50 ml polystyrene flasks (Corning Costar) at (3–5)×104 cells/ml in IMDM containing 10% (v/v) heat-inactivated pooled human AB serum, 1% BSA, 10 mg/ml insulin, 0.5 mg/ml vitamin B12, 15 mg/ml folic acid, 50 nM 2-mercaptoethanol, 50 units/ml penicillin and 50 units/ml streptomycin in the presence of 10 ng/ml interleukin-3, 10 ng/ml SCF (stem cell factor) and 2 units/ml EPO (erythropoietin) at 37 °C in a 5% CO2 incubator, as described previously [23]. After 7 days of culture, the generated cells were collected and washed twice with IMDM containing 0.3% BSA to obtain early erythroid progenitor cells.

Immunofluorescence study

Cell samples were rinsed twice in PBS, fixed for 10 min at room temperature in PBS containing 3.5% formaldehyde, permeabilized further with 0.2% Triton X-100 for 5 min at room temperature and finally rinsed twice in PBS. After incubation in PBS with 10% (v/v) goat serum and 10 mg/ml BSA to block non-specific protein binding, the fixed cells were incubated with antibodies for 1 h at room temperature in PBS containing 10 mg/ml of BSA, washed and then incubated with FITC- and rhodamine-conjugated secondary antibodies. The cells were washed again and the coverslips were mounted on slides using Vectashield containing DAPI® (4′,6-diamidino-2-phenylindole) for nuclear staining. Controls with equivalent amounts of non-immune IgG or primary antibody omitted were also included in each experiment. Immunofluorescent cells were imaged with a confocal microscope (LSM510-Ver3.2, Carl Zeiss) having a 63×, 1.4 NA oil-immersion objective lens.

RESULTS

4.1R135 binding to IOVs

SDS/PAGE analysis of non-tagged full-length recombinant 4.1R135 used for the binding assay showed a major band at 135 kDa corresponding to full-length protein and a minor band at ∼100 kDa, presumably a proteolytic-cleavage product (Figure 2A, lane 1). The membrane protein composition of native IOVs (Figure 2A, lane 2) and of trypsinized IOVs (Figure 2A, lane 4) are also shown. Note that trypsin treatment of IOVs resulted in the generation of band 3 fragments of 60 kDa that lack the cytoplasmic domain containing the 4.1R-binding site. Previous studies have shown that although 4.1R binds to both band 3 and GPC in native IOVs, it binds only to GPC in trypsinized IOVs [19]. The pelleting assay showed binding of 4.1R135 to native IOVs (Figure 2A, lane 3), as well as trypsinized IOVs (Figure 2A, lane 5).

Comparison of 4.1R135 and 4.1R80 binding to IOVs of human erythrocytes

Figure 2
Comparison of 4.1R135 and 4.1R80 binding to IOVs of human erythrocytes

(A) Input 4.1R135 (lane 1), native IOV (lane 2), native IOV mixed with 4.1R135 (lane 3), trypsinized IOV (lane 4) and trypsinized IOV mixed with 4.1R135 (lane 5). Proteins were stained with CBB. The bands corresponding to 4.1R135, native band 3 and trypsin-cleaved band 3 are indicated by arrowheads and asterisks respectively. Bound proteins were analysed by SDS/PAGE stained with CBB. The band density was determined using a densitometer. (B) The binding profile and (C) Scatchard plot of 4.1R135 (● and ▲) and 4.1R80 (○ and Δ) toward native IOV (● and ○) and trypsinized IOV (▲ and Δ).

Figure 2
Comparison of 4.1R135 and 4.1R80 binding to IOVs of human erythrocytes

(A) Input 4.1R135 (lane 1), native IOV (lane 2), native IOV mixed with 4.1R135 (lane 3), trypsinized IOV (lane 4) and trypsinized IOV mixed with 4.1R135 (lane 5). Proteins were stained with CBB. The bands corresponding to 4.1R135, native band 3 and trypsin-cleaved band 3 are indicated by arrowheads and asterisks respectively. Bound proteins were analysed by SDS/PAGE stained with CBB. The band density was determined using a densitometer. (B) The binding profile and (C) Scatchard plot of 4.1R135 (● and ▲) and 4.1R80 (○ and Δ) toward native IOV (● and ○) and trypsinized IOV (▲ and Δ).

Binding of 4.1R135 to both band 3 and GPC on native IOVs increased with an increasing concentration of 4.1R135, reaching a maximum at 600 nM (Figure 2B). Scatchard analysis indicated an apparent dissociation constant at equilibrium, K′, of 76 nM (Figure 2C). In contrast, K′ for 4.1R80 binding to band 3 and GPC on native IOVs was higher at 340 nM. Surprisingly, 4.1R135 binding to trypsinized IOVs, in which binding is exclusively to GPC, was markedly weaker with a K′ of ∼2000 nM, whereas the K′ for 4.1R80 binding to trypsinized IOVs was 230 nM, similar to that of native IOVs. These findings suggest that the presence of the HP in 4.1R135 differentially regulates the binding of the two isoforms to band 3 and GPC.

We also studied the binding of the recombinant RHP–R30 to IOVs and found that its binding affinity to both native and trypsinized IOVs was the same as those of full-length 4.1R135 (results not shown). This finding suggests that all of the binding characteristics of 4.1R135 to band 3 and GPC are accounted for by the HP and the 30 kDa domain of native protein.

Kinetic analysis of 4.1R135 binding to membrane proteins using the RMD method

In order to obtain independent confirmation of the binding affinities of 4.1R135 to band 3cyt and GPCcyt we used the IAsys® system based on the RMD method (Table 1). In agreement with our binding data using IOVs, there was a profound difference in the binding of 4.1R135 to band 3cyt and GPCcyt with much stronger binding to band 3cyt. The KD values for binding to these two proteins differed by almost two orders of magnitude (15±1 nM for band 3 compared with 1317±104 nM for GPC). In marked contrast, the KD values for binding of 4.1R80 to both band 3cyt and GPCcyt were very similar, in the sub-micromolar range (Table 1), suggesting that an important role for the HP is in regulating binding affinities. In contrast with the marked differences in the binding affinities of 4.1R135 and 4.1R80 to band 3cyt and GPCcyt, the two isoforms bound to p55 with very similar affinities in the sub-micromolar range (Table 1).

Table 1
4.1R isoforms binding to membrane proteins

KD values for the interactions of 4.1R135 or 4.1R80 to GPC, p55 and band 3 are shown. Each analyte (1 nM–2 μM) was incubated with the identified ligand immobilized on the aminosilane cuvette as described in the Materials and methods section. From the binding curves obtained by the RMD method, the values were determined using the software package FAST-Fit®. ka, kd and KD are means±S.D. calculated from three independent experiments.

Analyte Ligand ka (M−1·s−1kd (s−1KD (nM) 
4.1R135 Band 3cyt (4.0±0.20)×105 (6.1±0.06)×10−3 15±1 
4.1R135 GPCcyt (9.1±0.19)×103 (1.2±0.12)×10−2 1317±104 
4.1R135 p55 (1.4±0.08)×105 (1.7±0.08)×10−2 122±10 
4.1R80 Band 3cyt (7.0±0.46)×104 (1.4±0.01)×10−2 194±6 
4.1R80 GPCcyt (2.9±0.16)×105 (2.7±0.32)×10−2 95±12 
4.1R80 p55 (3.8±0.08)×105 (3.6±0.22)×10−2 94±7 
Analyte Ligand ka (M−1·s−1kd (s−1KD (nM) 
4.1R135 Band 3cyt (4.0±0.20)×105 (6.1±0.06)×10−3 15±1 
4.1R135 GPCcyt (9.1±0.19)×103 (1.2±0.12)×10−2 1317±104 
4.1R135 p55 (1.4±0.08)×105 (1.7±0.08)×10−2 122±10 
4.1R80 Band 3cyt (7.0±0.46)×104 (1.4±0.01)×10−2 194±6 
4.1R80 GPCcyt (2.9±0.16)×105 (2.7±0.32)×10−2 95±12 
4.1R80 p55 (3.8±0.08)×105 (3.6±0.22)×10−2 94±7 

To clarify the effects of RHP on the R30 interaction with its binding partners, we measured its ability to bind band 3cyt and GPCcyt with and without RHP (Table 2). RHP by itself did not bind to either band 3cyt or GPCcyt. As we previously reported, R30 by itself interacted with similar sub-micromolar affinities to both band 3cyt and GPCcyt (Table 2) [13,14]. Importantly, the addition of RHP to R30 (RHP–R30), resulted in a profound differential change in its ability to bind band 3cyt and GPCcyt. The binding affinity of RHP–R30 to band 3cyt was 35-fold stronger than for GPCcyt, a finding that is very similar to that noted above for the native full-length protein in the IOV-binding assays. These findings have identified an important role for RHP in modulating the interaction of R30 with its two membrane-binding partners.

Table 2
The 30 kDa domain binding to membrane proteins in the absence and the presence of the HP

KD values for the interactions of RHP, R30 and RHP–R30 proteins to band 3cyt and GPCcyt are shown. Each analyte (50 nM–2 μM) was incubated with the identified ligand immobilized on the aminosilane cuvette. From the binding curves obtained by the RMD method, the values were determined using the software package FAST-Fit®. ka, kd and KD are means±S.D. calculated from three independent experiments. No binding, no signal appeared when 2 μM of RHP was applied to the ‘ligand’-immobilized cuvette.

Analyte Ligand ka (M−1·s−1kd (s−1KD (nM) 
RHP Band 3cyt No binding No binding No binding 
R30 Band 3cyt (6.0±0.23)×104 (1.2±0.11)×10−2 198±18 
RHP–R30 Band 3cyt (2.0±0.11)×105 (8.1±0.23)×10−3 40±3 
RHP GPCcyt No binding No binding No binding 
R30 GPCcyt (7.0±0.56)×104 (1.2±0.10)×10−2 173±21 
RHP–R30 GPCcyt (7.5±0.20)×103 (1.1±0.10)×10−2 1481±145 
Analyte Ligand ka (M−1·s−1kd (s−1KD (nM) 
RHP Band 3cyt No binding No binding No binding 
R30 Band 3cyt (6.0±0.23)×104 (1.2±0.11)×10−2 198±18 
RHP–R30 Band 3cyt (2.0±0.11)×105 (8.1±0.23)×10−3 40±3 
RHP GPCcyt No binding No binding No binding 
R30 GPCcyt (7.0±0.56)×104 (1.2±0.10)×10−2 173±21 
RHP–R30 GPCcyt (7.5±0.20)×103 (1.1±0.10)×10−2 1481±145 

4.1R135 binding to CaM

We previously documented that 4.1R80 binds to CaM with a KD in the range of 100 nM both in the presence and absence of Ca2+, suggesting Ca2+-independent binding (Table 3) [13]. In the present study, we examined the nature of the interaction between 4.1R135 and CaM. Kinetic analysis of the 4.1R135 interaction with CaM using the RMD method identified a very strong interaction with a KD of 51±5 nM in the presence of Ca2+. In the absence of Ca2+ the affinity of the interaction decreased by over 100-fold. Thus, in contrast with 4.1R80, the interaction of 4.1R135 with CaM was strongly Ca2+-dependent. As with binding to membrane proteins, this differential change in binding of 4.1R135 to CaM was recapitulated by RHP–R30 (Table 3). Furthermore, the HP region by itself exhibited Ca2+-dependent binding with CaM, suggesting that this region harbours the CaM-binding site (Table 3).

Table 3
4.1R isoforms and RHP binding to CaM

KD values for the interactions of 4.1R135, 4.1R80 and RHP–R30 to a CaM-immobilized aminosilane cuvette in the presence (Ca2+) and absence (EGTA) of Ca2+ are shown. Each analyte (1 nM–2 μM) was incubated with the identified ligand immobilized on the aminosilane cuvette. From the binding curves obtained by the RMD method, the values were determined using the software package FAST-Fit®. ka, kd and KD are means±S.D. calculated from three independent experiments. No binding, no signal appeared when 2 μM of RHP was applied to the CaM-immobilized cuvette.

Analyte Ligand Condition ka (M−1·s−1kd (s−1KD (nM) 
4.1R135 CaM EGTA (2.6±0.23)×103 (1.5±0.15)×10−2 6117±534 
  Ca2+ (2.1±0.16)×105 (1.1±0.11)×10−2 51±5 
4.1R80 CaM EGTA (7.8±0.34)×104 (1.6±0.14)×10−2 207±23 
  Ca2+ (2.3±0.22)×105 (2.3±0.21)×10−2 100±15 
RHP–R30 CaM EGTA (1.4±0.22)×103 (1.6±0.13)×10−2 11659±2890 
  Ca2+ (2.0±0.13)×105 (1.5±0.11)×10−2 78±10 
RHP CaM EGTA No binding No binding No binding 
  Ca2+ (2.6±0.17)×105 (1.8±0.12)×10−2 68±7 
Analyte Ligand Condition ka (M−1·s−1kd (s−1KD (nM) 
4.1R135 CaM EGTA (2.6±0.23)×103 (1.5±0.15)×10−2 6117±534 
  Ca2+ (2.1±0.16)×105 (1.1±0.11)×10−2 51±5 
4.1R80 CaM EGTA (7.8±0.34)×104 (1.6±0.14)×10−2 207±23 
  Ca2+ (2.3±0.22)×105 (2.3±0.21)×10−2 100±15 
RHP–R30 CaM EGTA (1.4±0.22)×103 (1.6±0.13)×10−2 11659±2890 
  Ca2+ (2.0±0.13)×105 (1.5±0.11)×10−2 78±10 
RHP CaM EGTA No binding No binding No binding 
  Ca2+ (2.6±0.17)×105 (1.8±0.12)×10−2 68±7 

To determine the stoichiometry of 4.1R135 binding to CaM in the presence of Ca2+, we assessed the change in molecular mass following the interaction of the two proteins using the QCM method. To 1.57×10−14 mol of CaM immobilized on to the sensor chip, 1.75×10−14 mol of 4.1R135 bound in the presence of Ca2+ (Figure 3). The binding ratio of 4.1R135/CaM was calculated to be 1:1.1. The apparent dissociation constant of the 4.1R135–CaM interaction was determined to be 54 nM in the presence of Ca2+, consistent with a KD of 51 nM obtained by IAsys®. 4.1R135 did not bind to CaM-immobilized sensor chips in the absence of Ca2+. These findings independently validate our findings with the RMD method and establish a 1:1 stoichiometry of CaM binding to 4.1R135 in a highly Ca2+-dependent manner.

Analysis of 4.1R135 binding to CaM using the QCM method

Figure 3
Analysis of 4.1R135 binding to CaM using the QCM method

(A) The binding profile and (B) the Scatchard plot of the CaM interaction with 4.1R135 in the presence (●) and absence (○) of Ca2+ are shown. From the binding profile, 1.75×10−14 mol of 4.1R135 bound to 1.57×10−14 mol of CaM immobilized on the sensor tips. The binding ratio of CaM/4.1R135 was approx. 1:1.1. The equivalent dissociation constant was 54 nM obtained by the Scatchard plot.

Figure 3
Analysis of 4.1R135 binding to CaM using the QCM method

(A) The binding profile and (B) the Scatchard plot of the CaM interaction with 4.1R135 in the presence (●) and absence (○) of Ca2+ are shown. From the binding profile, 1.75×10−14 mol of 4.1R135 bound to 1.57×10−14 mol of CaM immobilized on the sensor tips. The binding ratio of CaM/4.1R135 was approx. 1:1.1. The equivalent dissociation constant was 54 nM obtained by the Scatchard plot.

Regulation of interactions of 4.1R135 with membrane–protein interactions by Ca2+/CaM

In a series of studies, we examined the effect of Ca2+/CaM on the binding of 4.1R135 to its various binding partners (Table 4). The binding affinity of 4.1R135 to band 3cyt was decreased by almost two orders of magnitude by Ca2+/CaM. Moreover, Ca2+/CaM completely abolished the ability of 4.1R135 to bind either GPCcyt or p55. Either 5 μM CaM or 100 μM Ca2+ alone had no effect on binding affinities (results not shown).

Table 4
Ca2+/CaM down-regulation of the 4.1R135–membrane–protein interaction

KD values for the interaction between band 3cyt, GPCcyt or p55 and 4.1R135 in the presence and absence of Ca2+ are shown. 4.1R135 (1 nM–2 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 [50 mM Tris/HCl (pH 7.5) with 0.1 M NaCl]. The complex of CaM and 4.1R135 was incubated with band 3cyt, GPCcyt or p55 immobilized on the aminosilane cuvette. Binding assays were carried out as described in the Materials and methods section. ka, kd and KD are means±S.D. calculated from three independent experiments. No binding, no signal appeared when 2 μM of 4.1R135 was mixed with 5 μM of Ca2+/CaM applied to the ‘ligand’-immobilized cuvette.

Analyte Ligand Condition ka (M−1·s−1kd (s−1KD (nM) 
4.1R135+CaM Band 3cyt EGTA (5.1±0.20)×105 (6.4±0.19)×10−3 13±1 
  Ca2+ (1.1±0.13)×104 (1.2±0.10)×10−2 1126±154 
4.1R135+CaM GPCcyt EGTA (8.4±0.32)×103 (1.3±0.11)×10−2 1538±146 
  Ca2+ No binding No binding No binding 
4.1R135+CaM p55 EGTA (1.1±0.10)×105 (2.3±0.21)×10−2 211±27 
  Ca2+ No binding No binding No binding 
Analyte Ligand Condition ka (M−1·s−1kd (s−1KD (nM) 
4.1R135+CaM Band 3cyt EGTA (5.1±0.20)×105 (6.4±0.19)×10−3 13±1 
  Ca2+ (1.1±0.13)×104 (1.2±0.10)×10−2 1126±154 
4.1R135+CaM GPCcyt EGTA (8.4±0.32)×103 (1.3±0.11)×10−2 1538±146 
  Ca2+ No binding No binding No binding 
4.1R135+CaM p55 EGTA (1.1±0.10)×105 (2.3±0.21)×10−2 211±27 
  Ca2+ No binding No binding No binding 

Next we examined the Ca2+-dependence of CaM-regulation of the binding of 4.1R135 to band 3cyt (Figure 4). 4.1R135 binding to band 3cyt started to decline beginning at Ca2+ concentrations greater than 10 nM (pCa<8), with maximal inhibition at a Ca2+ concentration of 100 μM (pCa>4). Half-maximal binding was seen at a Ca2+ concentration of 3.2 μM (pCa=5.5).

Ca2+ concentration dependence of 4.1R135 binding to band 3cyt

Figure 4
Ca2+ concentration dependence of 4.1R135 binding to band 3cyt

4.1R135 binding to band 3cyt was measured at various concentrations of Ca2+ in the presence (●) and absence (○) of 5 μM CaM. The Ca2+ concentration was maintained using 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 in the presence of different concentrations of Ca2+. pCa represents −log[Ca2+]. The extent of 4.1R135 binding to band 3cyt is plotted as a function of pCa.

Figure 4
Ca2+ concentration dependence of 4.1R135 binding to band 3cyt

4.1R135 binding to band 3cyt was measured at various concentrations of Ca2+ in the presence (●) and absence (○) of 5 μM CaM. The Ca2+ concentration was maintained using 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 in the presence of different concentrations of Ca2+. pCa represents −log[Ca2+]. The extent of 4.1R135 binding to band 3cyt is plotted as a function of pCa.

Localization of peripheral membrane proteins in the erythroblast

As 4.1R135 has previously been shown to be expressed in early erythroid progenitors including pro-erythroblasts [6], we examined the cellular localization of 4.1R135 and its binding partners in cells at day 7 of CD34+ cultures where the predominant population of cells is pro-erythroblasts. As shown in Figure 5, band 3 and GPC are localized to the plasma membrane, and p55 is broadly distributed in these early erythroblasts. When we looked for localization of 4.1R135 in these cells, we noted that the protein was present both at the membrane and in the cytoplasm.

Immunochemical localization of membrane proteins in erythroblasts at day 7

Figure 5
Immunochemical localization of membrane proteins in erythroblasts at day 7

In each column, images from the immunostaining (top panels) and by phase-contrast microscopy (bottom panels) are shown. In the top panels, green and red staining indicates that the primary antibodies used were either rabbit or mouse respectively. The same antibodies were used for immunoblot analysis and immunocytochemistry. Magnification in all cases is ×1000.

Figure 5
Immunochemical localization of membrane proteins in erythroblasts at day 7

In each column, images from the immunostaining (top panels) and by phase-contrast microscopy (bottom panels) are shown. In the top panels, green and red staining indicates that the primary antibodies used were either rabbit or mouse respectively. The same antibodies were used for immunoblot analysis and immunocytochemistry. Magnification in all cases is ×1000.

DISCUSSION

In the present study, we have described the significant differences in the nature of the interaction of the two 4.1R isoforms, 4.1R135 and 4.1R80, with their various binding partners. The binding of 4.1R135 to band 3 was an order of magnitude stronger than that of 4.1R80, whereas its ability to bind GPC was an order of magnitude weaker. Furthermore, although the binding of 4.1R80 to CaM was Ca2+-independent, CaM binding to 4.1R135 was strongly Ca2+-dependent. All of these differences are entirely attributable to the additional 209 amino acids at the N-terminus of 4.1R135.

A previous study identified L246EEDY in the 30 kDa membrane-binding domain of 4.1R as the binding site for band 3cyt [24]. The sequences S95EEEG and L122DEEI in the HP of 4.1R135 are similar to L246EEDY, and as such may serve as potential additional binding sites for band 3, accounting for observed stronger binding of band 3 to 4.1R135; however, our finding that the HP by itself does not bind band 3 rules out this possibility. It is more likely that alterations in the conformation of the band 3-binding site L246EEDY in the 30 kDa domain by the HP domain accounts for increased affinity. In this regard it is interesting to note that the 209 amino acid HP domain is rich in positively charged amino acids (42 glutamic acid and nine aspartic acid residues) and has an isoelectric point of 4.46. Thus the presence of this domain in 4.1R135 with a large number of positively charged amino acids may enhance the electrostatically mediated interaction with band 3.

In contrast with the much stronger binding of 4.1R135 to band 3, its binding to GPC was significantly weaker than that of 4.1R80. The R86HK motif in GPCcyt has previously been shown to bind to a site within residues Tyr94–Arg166 in the 30 kDa binding domain of 4.1R80 [14,25]. In this stretch of 73 amino acids, 25 amino acids in the N-terminal region are predominantly hydrophobic, whereas 48 amino acids at the C-terminus are hydrophilic. It has been suggested that the E153LEE sequence in the hydrophilic region is likely to mediate its interaction with the R86HK sequence in GPC [25], although the precise GPC-binding site on 4.1R80 remains to be identified. The significantly weaker interaction of 4.1R135 with GPC implies that the HP region, in contrast with its positive effect on the band 3-binding site, must somehow mask the GPC-binding site and have a negative effect on binding.

Another significant difference noted during the present study between 4.1R135 and 4.1R80 is the Ca2+-dependence of the binding of these two isoforms to CaM. In contrast with Ca2+-independent binding of CaM to 4.1R80, CaM binding to 4.1R135 was strongly Ca2+-dependent. This difference was once again directly attributed to the HP region of 4.1R135. Importantly, in contrast with band 3 and GPC that do not directly bind to the HP region, this region by itself binds to CaM in a Ca2+-dependent manner. Leclerc and Vetter [17] have previously identified the Ca2+/CaM-binding site in the HP region of Xenopus. sp. 4.1 [16,17]. This CaM-binding sequence is highly conserved between human and Xenopus 4.1. Thus it must be inferred that the CaM-binding site in the HP is the dominant binding site for CaM in 4.1R135, and that this site prevents the binding of CaM to the Ca2+-independent binding site in 4.1R80. Furthermore, our finding that CaM, in a Ca2+-dependent manner, dramatically decreases the binding of 4.1R135 to band 3 and abolishes its binding to GPC and p55 has implications for the function of this protein in early erythroblasts. Although low levels of Ca2+ in early erythroblasts will lead to membrane association through a high-affinity interaction with band 3, increasing levels of Ca2+ during erythroid differentiation will lead to displacement of the protein from the membrane and possible degradation and loss of this isoform from erythroblasts [26,27]. Our findings that a fraction of 4.1R135 in early erythroblasts is associated with the membrane lends support to this hypothesis; however, it should be noted that quantitative measurement of Ca2+ levels in erythroblasts at different stages of maturation needs to be established to validate this hypothesis.

We gratefully acknowledge Dr Philippe Gascard (Genome Biology, Lawrence Berkeley National Laboratories, University of California, Berkeley, CA, U.S.A.) for valuable comments and Ms Hiromi Kataho (Internal Medicine III, School of Medicine, Akita University, Akita, Japan) for competent technical assistance. Thanks are also due to Dr Tetsuya Kawagoe (Initium, Japan) for assistance in conducting Affinix® (the QCM system).

Abbreviations

     
  • band 3 cyt

    the N-terminal cytoplasmic domain of band 3

  •  
  • CaM

    calmodulin

  •  
  • CBB

    Coomassie Brilliant Blue

  •  
  • CTD

    C-terminal domain

  •  
  • FCS

    fetal calf serum

  •  
  • FERM

    4.1/ezrin/radixin/moesin

  •  
  • GPC

    glycophorin C

  •  
  • GPCcyt

    the cytoplasmic domain of GPC

  •  
  • GST

    glutathione transferase

  •  
  • HP

    head-piece

  •  
  • IMDM

    Iscove's modified Dulbecco's medium

  •  
  • IOV

    inside-out vesicle

  •  
  • R30

    recombinant protein of the 30 kDa domain of 4.1R

  •  
  • 4.1R80

    80 kDa human erythrocyte-type protein 4.1R

  •  
  • 4.1R135

    135 kDa human erythroblast-type protein 4.1R

  •  
  • RHP

    HP of 4.1R135

  •  
  • RHP–R30

    a hybrid of RHP and R30

  •  
  • RMD

    resonant mirror detection

  •  
  • QCM

    quartz crystal micro-balance

  •  
  • SAB

    spectrin- and actin-binding domain

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 numbers 12680702, 15570123] (to W. N.); the Center of Excellence Programme of the Ministry of Education, Science, Technology, Sports, and Culture of Japan [grant numbers 14370075, 17590978, 17925036] (to K.-I. S.); and the National Institutes of Health [grant numbers DK 26263, DK 32094] (to N. M.).

References

References
1
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
)
2
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
)
3
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
)
4
Tang
T. K.
Leto
T. L.
Correas
I.
Alonso
M. A.
Marchesi
V. T.
Benz
E. J.
Jr
Selective expression of an erythroid-specific isoform of protein 4.1
Proc. Natl. Acad. Sci. U.S.A.
1998
, vol. 
85
 (pg. 
3713
-
3717
)
5
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
)
6
Chasis
J. A.
Coulombel
L.
Conboy
J.
McGee
S.
Andrews
K.
Kan
Y. W.
Mohandas
N.
Differentiation-associated switches in protein 4.1 expression synthesis of multiple structural isoforms during normal human erythropoiesis
J. Clin. Invest.
1993
, vol. 
91
 (pg. 
329
-
338
)
7
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
)
8
Chishti
A. H.
The FERM domain: a unique module involved in the linkage of cytoplasmic proteins to the membrane
Trends Biochem. Sci.
1998
, vol. 
23
 (pg. 
281
-
282
)
9
Han
B. G.
Nunomura
W.
Takakuwa
Y.
Mohandas
N.
Jap
B. K.
Protein 4.1R core domain structure and insights into regulation of cytoskeletal organization
Nat. Struct. Biol.
2000
, vol. 
7
 (pg. 
871
-
875
)
10
Gimm
J. A.
An
X.
Nunomura
W.
Mohandas
N.
Functional characterization of spectrin-actin-binding domains in 4.1 family of proteins
Biochemistry
2002
, vol. 
41
 (pg. 
7275
-
7282
)
11
Manno
S.
Takakuwa
Y.
Mohandas
N.
Modulation of erythrocyte membrane mechanical function by protein 4.1 phosphorylation
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
7581
-
7587
)
12
Nunomura
W.
Takakuwa
Y.
Tokimitsu
R.
Krauss
S.
Kawashima
M.
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
)
13
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
)
14
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
)
15
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
)
16
Kelly
G. M.
Zelus
B. D.
Moon
R. T.
Identification of a calcium-dependent calmodulin-binding domain in Xenopus membrane skeleton protein 4.1
J. Biol. Chem.
1991
, vol. 
266
 (pg. 
12469
-
12472
)
17
Leclerc
E.
Vetter
S.
Characterization of a calcium-dependent calmodulin-binding domain in the 135-kD human protein 4.1 isoform
Eur. J. Biochem.
1998
, vol. 
258
 (pg. 
567
-
571
)
18
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
)
19
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
)
20
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
)
21
Watts
H. J.
Lowe
C. R.
Optical biosensor for monitoring microbial cells
Anal. Chem.
1994
, vol. 
66
 (pg. 
2465
-
2470
)
22
Okahata
Y.
Kawase
M.
Niikura
K.
Ohtake
F.
Fyrusawa
H.
Ebara
Y.
Kinetic measurements of DNA hybridization on an oligonucleotide-immobilized 27-MHz quartz crystal microbalance
Anal. Chem.
1998
, vol. 
70
 (pg. 
1288
-
1296
)
23
Oda
A.
Sawada
K.
Druker
B. J.
Ozaki
K.
Takano
H.
Koizumi
K.
Fukada
Y.
Handa
M.
Koike
T.
Ikeda
Y.
Erythropoietin induces tyrosine phosphorylation of Jak2, STAT5A and STAT5B in primary cultured human erythroid precursors
Blood
1998
, vol. 
92
 (pg. 
443
-
451
)
24
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
)
25
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
-
24197
)
26
Imagawa
S.
Smith
B. R.
Palmer-Crocker
R.
Bunn
H. F.
The effect of recombinant erythropoietin on intracellular free calcium in erythropoietin-responsive cells
Blood
1989
, vol. 
73
 (pg. 
1452
-
1457
)
27
Misiti
J.
Spivak
J. L.
Erythropoiesis in vitro
Role of calcium. J. Clin. Invest.
1979
, vol. 
64
 (pg. 
1573
-
1579
)