Erythroid protein 4.1 (4.1R) stabilizes the spectrin–actin network and anchors it to the plasma membrane. To contribute to the characterization of non-erythroid protein 4.1R, we used sedimentation, pull-down and co-immunoprecipitation assays to investigate the ability of protein 4.1R to establish inter-/intra-molecular associations. We demonstrated that the small 4.1R isoforms of 60 kDa (4.1R60), but not the larger isoforms of 80 and 135 kDa (4.1R80 and 4.1R135), were self-associated, and that a domain contained in all 4.1R isoforms, the core region, was responsible for 4.1R self-association. Results from denaturing–renaturing experiments, in which an initially non-self-associated 4.1R80 isoform became self-associated, suggested that an initially hidden core region was subsequently exposed. This hypothesis was supported by results from pull-down assays, which showed that the core region interacted with the N-terminal end of the FERM (4.1, ezrin, radixin, moesin) domain that is present in 4.1R80 and 4.1R135 isoforms but absent from 4.1R60 isoforms. Consistently, 4.1R80 isoforms bound neither to each other nor to 4.1R60 isoforms. We propose that 4.1R60 isoforms are constitutively self-associated, whereas 4.1R80 and 4.1R135 self-association is prevented by intramolecular interactions.
Protein 4.1R was first identified in human erythrocytes as an 80 kDa multifunctional protein of the membrane skeleton. Erythroid protein 4.1R is essential for maintaining erythrocyte shape and mechanical properties of the membrane, such as deformability and stability. In this cell type, protein 4.1R stabilizes the spectrin–actin network and mediates the attachment of the underlying cytoskeleton to the overlaying lipid bilayer through interactions with lipids (phosphatidylserine and phosphatidylinositol 4,5-bisphosphate) [1,2] as well as with integral membrane proteins . Deficiency of 4.1R in erythrocytes leads to the assembly of an unstable cytoskeleton structure that manifests itself as hereditary elliptocytosis, a disease characterized by the loss of normal discoid morphology and the presence of oval or elliptical erythrocytes with unstable membranes .
Subsequent studies showed that multiple immunoreactive 4.1R proteins, varying in size from 30 to 210 kDa, are present in nucleated cells [4,5]. Isoforms of 4.1R are expressed in many tissues and detected at different subcellular locations [6–9]. The roles and partners of 4.1R in non-erythroid cells are beginning to be elucidated [10–21].
The multiple isoforms of 4.1R are mainly expressed as a result of extensive alternative splicing of the 4.1R-encoding pre-mRNA [22,23]. This event is cell- and tissue-specific and dependent on the growth and differentiation stages of the cell [24–30]. The prototypical erythroid protein 4.1R80 is produced when 17 nt 5′-upstream from exon 2 are spliced out, and translation is initiated at the downstream start site present in exon 4 (ATG2). Larger isoforms, termed 4.1R135, containing up to 209 amino acids at the N-terminus of the erythroid 4.1R80, are synthesized when the 17-nt sequence containing the upstream ATG (ATG1) translation initiation codon is included. These isoforms are predominantly expressed in non-erythroid cells. A third type of isoforms, termed 4.1R60, can be produced in erythroid and non-erythroid cells when both the 17-nt sequence (containing the ATG1) and exon 4 (containing the ATG2) are spliced out and translation is initiated from a third translation-initiation codon (ATG3) present in exon 8 [31,32]. Thus three different types of 4.1R isoforms varying in their N-terminal extensions can be generated depending on whether ATG1, ATG2 or ATG3 is used as the translation-initiation codon.
Erythroid 4.1R80 protein contains four structural domains : a 30 kDa N-terminal membrane-binding domain that has recently been named the FERM domain (4.1, ezrin, radixin, moesin) [34,35], a 16 kDa domain, a 10 kDa SAB (spectrin/actin-binding domain) and the 22–24-kDa CTD (C-terminal domain). 4.1R135 isoforms contain an extra N-terminal extension domain [HP (Head-Piece)] comprised of up to 209 amino acids, whereas 4.1R60 isoforms lack the HP domain and the N-terminal end of the FERM domain.
In recent years our group has been interested in characterizing non-erythroid protein 4.1R. Systematic studies using 4.1R135, 4.1R80 and 4.1R60 isoform types showed that the small 4.1R60 isoforms present significant differences in their subcellular distribution and function [15,32,36]. To gain further insights into nonerythroid protein 4.1R, in the present study we have analysed the biochemical behaviour of different 4.1R isoform types using sedimentation, co-immunoprecipitation and pull-down assays. These experiments show that 4.1R60, but not 4.1R135 or 4.1R80 isoforms, establishes intermolecular interactions through a constitutive domain, termed the core region. Since this region is present in all 4.1R isoforms and only 4.1R60 isoforms were found to be self-associated, we reasoned that 4.1R135 and 4.1R80 isoforms should also have the ability to self-associate, but that their core region needs to be exposed for such an interaction to occur. Results from subsequent experiments are consistent with this idea. Our results indicate that all 4.1R proteins have the ability to self-associate but that large 4.1R135 and 4.1R80 isoforms prevent their self-association by intramolecular interaction between the N-terminus of the FERM domain and the self-association region.
Cell culture and transfection
COS-7 cells were grown as described in . Transfection experiments were performed by electroporation using the Electro Cell Manipulator 600 (BTX, San Diego, CA, U.S.A.). Cells were processed 48 h after transfection.
cDNA cloning and composite cDNA constructs
4.1R135Δ16; 4.1R80Δ16; 4.1R60Δ16 and 4.1R60Δ16,18 cDNAs were cloned from Molt-4 T-cells and tagged as described in [32,36]. 4.1R135Δ16–GFP (green fluorescent protein); 4.1R80Δ16–GFP; and 4.1R60Δ16,18–GFP were constructed as detailed in . GST (glutathione S-transferase); GST–4.1R135Δ16; GST–4.1R80Δ16; GST–4.1R60Δ16,18; GST–Cter (calmodulin, troponin C, essential and regulatory light chains of myosin); GST–core and GST–coreΔLeu proteins were prepared as described in . GST–4.1RE4–E8 was constructed by PCR using pSRα4.1R80Δ16 as a template . Appropriate sense and antisense primers containing the BglII and XhoI restriction sites at the 5′- and 3′-ends respectively were used for the amplification reactions. The amplified cDNAs were inserted into the BamHI and XhoI sites of pGEX-6P1 vector (Amersham Biosciences, Piscataway, NJ, U.S.A.) in-frame with the GST coding sequence. GST fusion proteins were cleaved by PreScission protease (Amersham Biosciences) following the manufacturer's instructions.
Anti-4.1R (10b) antibody is an affinity-purified polyclonal antibody generated as described previously  and recognizes a sequence encoded by exon 17. Anti-4.1R (764) is a polyclonal antibody raised against a synthetic peptide (FRYSGRTQAQTRC) whose sequence is encoded by exon 12. Anti-GST and anti-FLAG antibodies are rabbit polyclonal antibodies (Sigma, St. Louis, MO, U.S.A.). Anti-GFP is a rabbit polyclonal antibody (Molecular Probes, Eugene, OR, U.S.A.). Horseradish-peroxidase-labelled secondary antibodies were obtained from Southern Biotechnology Associates.
Western blot analysis and immunoprecipitation assays
Protein samples were separated by SDS/PAGE and transferred to Immobilon (Millipore) in Tris/borate buffer (pH 8.2). Membranes were processed and developed as described in . For immunoprecipitation assays, COS-7 cells were washed twice with PBS and scraped from the plate into PBS containing 5 mM EDTA. Cells were centrifuged, and the pellet was resuspended in lysis buffer [10 mM Tris/HCl (pH 7.6) 150 mM NaCl, 1% Nonidet P40, 5 mM EDTA, 0.5 mM PMSF and 1 μg/ml each of leupeptin, aprotinin and pepstatin], incubated on ice for 20 min and centrifuged in a Minifuge for 10 min at 4 °C. The supernatant was incubated with antibody-coupled Sepharose beads for 3 h at 4 °C and processed as previously described .
Sucrose-density-gradient centrifugation assays
COS-7 cells were harvested, lysed and sedimented as described in . Briefly, cells were lysed in a buffer containing 50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% Nonidet P40, 0.5 mM PMSF and 1 μg/ml each of leupeptin, aprotinin and pepstatin, and kept at 4 °C for 15 min. The lysates were centrifuged at 100000 g for 25 min at 4 °C in a Beckman TL-100 table-top ultracentrifuge using a TLA-100.1 fixed-angle rotor. The supernatants were loaded on to 4.2 ml linear 5–20% (w/v) sucrose gradients prepared in lysis buffer without Nonidet P40. Gradients were centrifuged in a TST-60.4 rotor (Kontron Instruments) at 26000 rev./min for 18 h at 4 °C and collected as 300 μl fractions. Sedimentation standards BSA (4.4 S), catalase (11.3 S) and thyroglobulin (19 S) were run in parallel. Equal volumes of each fraction were analysed by SDS/PAGE followed by Western blotting . For protein denaturation and renaturation (Figure 5), GST–4.1R80Δ16 protein was dialysed serially against buffer A (50 mM Tris/HCl, pH 7.5, 150 mM NaCl and 1 mM EDTA) containing 2, 4 or 8 M urea. The protein was slowly renatured by removing the urea by serially dialysing against buffer A containing 4, 2 and 0 M urea. After completely removing the urea, the GST–4.1R80Δ16 protein was loaded on to the sucrose gradients, centrifuged, fractionated and processed as described above. For recombinant proteins 10 μg of fusion protein was loaded on to the sucrose gradient.
In vitro protein expression
In vitro protein expression was achieved by coupled in vitro transcription and translation reactions using the TNT T7 reticulocyte lysate system (Promega, Madison, WI, U.S.A.) as previously described . Synthesized proteins were radiolabelled by including [35S]methionine (35S-Met; Amersham Biosciences) in the reaction.
GST and the recombinant proteins GST–4.1R80Δ16, GST–4.1R60Δ16,18, GST–Cter, GST–core, GST–coreΔLeu and GST–4.1RE4–E8 were prepared as indicated in . COS-7 cells were lysed and processed as described in . Briefly, COS-7 cell lysates or radiolabelled proteins were incubated for 1 h at 4 °C with the glutathione–Sepharose-4B column loaded with the corresponding GST proteins. After extensive washes the beads were resuspended in Laemmli buffer and boiled for 5 min. The bound proteins were separated by SDS/PAGE, transferred to Immobilon membranes according to standard procedures and visualized by immunoblotting as described above.
Gel filtration chromatography was carried out using a Superdex 200 HiLoad 16/60 column (Amersham Biosciences) controlled by an AKTA FPLC system (Amersham Biosciences). The column was equilibrated with two column volumes of a buffer containing 50 mM Tris (pH 7.5), 150 mM NaCl and 1 mM EDTA and run at 4 °C. Globular proteins of known molecular mass were used to calibrate the column before applying protein GST–4.1R60Δ16,18. The marker proteins used were thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), BSA (66 kDa), ovoalbumin (43 kDa) and chymotrypsinogen A (25 kDa). Proteins were eluted at a flow rate of 0.4 ml/min. The eluate was monitored by absorbance at 280 nm and the collected fractions (1.6 ml/fraction) were analysed by SDS/PAGE. The distribution of protein GST–4.1R60Δ16,18 was detected with the 10b anti-4.1R antibody.
Sedimentation velocity analysis
Analytical ultracentrifugation experiments were carried out at 45000 rev./min and 11.5 °C for 3 h in an XL-A analytical ultracentrifuge (Beckman Coulter) equipped with UV–visible absorbance optics, an An50Ti rotor and 12 mm double-sector centrepieces. The protein (loading concentration of 0.5 mg/ml) was equilibrated in 50 mM Tris (pH 7.5), 150 mM NaCl and 1 mM EDTA. Data were collected at 280 nm. Differential sedimentation coefficient distributions, c(s), were calculated by least-squares boundary modelling of sedimentation velocity data using the program SEDFIT .
Isoforms of protein 4.1R show different sedimentation behaviours
To gain insights into non-erythroid protein 4.1R, we analysed the biochemical behaviour of different 4.1R isoform types using sucrose-density-gradient centrifugation assays. We assayed 4.1R isoforms translated from the most upstream start codon, ATG1 (4.1R135 isoforms; ∼135 kDa), and from the downstream start sites, ATG2 (4.1R80 isoforms; ∼80 kDa) and ATG3 (4.1R60 isoforms; ∼60 kDa) (Figure 1A) fused to GST. We chose representative isoforms that correspond to the most abundant of each type of 4.1R isoforms in non-erythroid human T-cells and whose cDNAs we had previously isolated [32,36]. Fusion proteins comprised of GST and full-length 4.1R135Δ16, 4.1R80Δ16 or 4.1R60Δ16,18 isoforms were purified, loaded on to linear 5–20% sucrose gradients, centrifuged and fractionated as described in the Experimental section. Protein distribution along the gradient was analysed by immunoblotting using the 10b anti-4.1R antibody (Figure 1B) or an anti-GST antibody (results not shown). Proteins GST, GST–4.1R135Δ16 and GST–4.1R80Δ16 fractionated at the top of the gradient, whereas the small protein GST–4.1R60Δ16,18 did not remain at the top of the gradient but fractionated with a sedimentation coefficient higher than that of large 4.1R proteins. These results demonstrate the different sedimentation behaviour of the small protein GST–4.1R60Δ16,18, indicating that this protein is self-associated.
A fusion protein containing the small 4.1R60Δ16,18 isoform migrates with a high sedimentation coefficient in sedimentation assays
Protein 4.1R60Δ16,18 binds both to itself and to other 4.1R60 isoform types, but not to a 4.1R80 isoform type
We next investigated whether 4.1R60Δ16,18 protein was able to establish intermolecular interactions with other 4.1R60 proteins. Glutathione–Sepharose beads coupled with GST or GST–4.1R60Δ16,18 were incubated with 35S-Met-labelled in vitro-translated 4.1R60Δ16 (Figure 2A). As revealed by autoradiographs, protein 4.1R60Δ16 bound to GST–4.1R60Δ16,18 (Figure 2A, lane 3), but not to GST control beads (Figure 2A, lane 2). This result confirmed that 4.1R60 isoform types form intermolecular interactions. In addition, the fusion protein GST–4.1R60Δ16 behaved like GST–4.1R60Δ16,18 in sucrose-density-gradient centrifugation assays (results not shown).
Protein 4.1R60Δ16,18 binds to itself and to another 4.1R60 isoform, but not to a 4.1R80 isoform
To determine whether 4.1R60 and 4.1R80 isoform types establish intermolecular interactions, we performed pull-down assays using GST–4.1R60Δ16,18 bound to Sepharose beads on COS-7 lysates expressing isoform 4.1R80Δ16–GFP. As shown in Figure 2(B), these two proteins did not interact (lane 3). This Figure also shows that proteins 4.1R80Δ16–GFP and GST–4.1R80Δ16 are not associated (Figure 2B, lane 4), whereas the same type of experiments confirmed the interactions between proteins 4.1R60Δ16,18–GFP and GST–4.1R60Δ16,18 (Figure 2B, lane 10). Taken together, these results imply that 4.1R60 isoforms are able to establish intermolecular interactions, while 4.1R80 isoforms are not. They also indicate that 4.1R60 and 4.1R80 isoforms do not interact. These results are in close agreement with those from the sedimentation assays.
Protein 4.1R60Δ16,18 exhibits a relatively disperse distribution in the sucrose gradients, suggesting that it forms oligomers of more than two subunits. When analysed by size-exclusion chromatography, the purified GST–4.1R60Δ16,18 protein was eluted in several peak fractions covering a wide range of molecular sizes (Figure 2C). We observed a major peak at fractions 41–44 (which may correspond to the monomer and the dimer), a minor peak at fraction 37 (tetramer) and a third peak at fractions 29–32 (octamer). Further examination of GST–4.1R60Δ16,18 oligomerization by sedimentation velocity analysis (Figure 2D) showed the presence of several species with sedimentation coefficients, corrected at 20 °C in water, of 6.97, 12.1, 19.7 and 26.4 S, consistent with the oligomeric forms detected by size-exclusion chromatography.
A conserved region is involved in 4.1R self-association
To identify the region involved in 4.1R60 self-association, we divided the molecule into two halves, previously designated by us as the ‘core’ and ‘Cter’ regions, and each fragment was fused to GST (Figure 3A) [14,32]. Pull-down assays were performed using the recombinant protein GST–core bound to Sepharose beads and incubated with 35S-Met-labelled, in vitro-translated core or Cter protein regions (Figure 3B). As revealed by the autoradiographs, 35S-Met–core bound to GST–core but not to GST control beads (Figure 3B, autoradiograph, 35S-core), while 35S-Met–Cter bound neither to GST–core nor to GST (Figure 3B, autoradiograph, 35S-Cter). These results indicate that the core region is involved in 4.1R self-association.
Identification of the 4.1R region involved in self-association
The involvement of the core region in 4.1R self-association was further analysed by sedimentation assays (Figure 3C). While GST and GST–Cter localized in the upper fractions of the gradient, GST–core distributed along the gradient with a peak in fractions 7–9, thus resembling the distribution of the full-length protein GST–4.1R60Δ16,18. These results confirm that the core region is responsible for 4.1R self-association.
As the core region contained 22 amino acids comprised of heptad repeats of leucine residues resembling a putative leucine zipper motif , we wondered whether they were involved in this intermolecular interaction. The behaviour of a fusion protein lacking the 22 amino acids and designated GST–coreΔLeu was analysed by sedimentation assays (Figure 3C, lower panel). The results show that while the core region was involved in 4.1R60 self-association, the heptad repeats of leucine residues were not responsible for such interactions. Although the sedimentation assays revealed that GST alone localized in the upper fractions of the gradient (Figure 3C), we wanted to demonstrate that the fusion of GST to 4.1R isoforms does not artificially contribute to 4.1R self-association. The biochemical behaviour of 4.1R80Δ16, 4.1R60Δ16,18 and the core region, after removing GST (see the Experimental section), was analysed by sucrose-density-gradient centrifugation assays (Figure 4). Protein 4.1R80Δ16 fractionated at the top of the gradient, whereas the small 4.1R60Δ16,18 isoform and the core region fractionated with a high sedimentation coefficient. These results demonstrate that GST influences neither the self-association of the 4.1R60 isoform and the core region nor the differential sedimentation behaviour of 4.1R isoform types.
Removal of the GST does not influence the self-association of protein 4.1R60
The 4.1R80 isoform 4.1R80Δ16 was able to self-associate after denaturing–renaturing treatments
It was intriguing that all 4.1R isoforms contained the core region responsible for 4.1R60 self-association, but that not all of them were able to self-associate. One possible explanation for this is that the core region was exposed in 4.1R60, but not in 4.1R80 and 4.1R135 isoforms. We reasoned that 4.1R80 and 4.1R135 isoforms would have the ability to self-associate if their core region were exposed. Thus we surmised that submitting the protein to denaturing conditions first and allowing it to renature slowly would favour the exposure of the core region, thus facilitating self-association before protein folding. To determine whether this indeed occurred, protein GST–4.1R80Δ16 was dialysed with a series of the appropriate buffer (see the Experimental section) containing consecutively greater concentrations of urea. The protein was then allowed to renature slowly by dialysing with a series of the same buffer containing consecutively lower concentrations of urea. The protein was loaded on to a 5–20% sucrose-density gradient, centrifuged and fractionated as described in the Experimental section, and the results were analysed by immunoblotting using the 10b anti-4.1 antibody (Figure 5). It was observed that protein GST–4.1R80Δ16 distributed along the entire gradient with two peaks: the first is in fraction 3 and the other is in fraction 10, which may correspond to the monomeric and the self-associated species respectively. These results indicate that protein 4.1R80Δ16 is also able to self-associate, but that the core region of the molecule needs to be exposed for this interaction to occur. As it could be argued that the higher sedimenting species corresponded to aggregated protein that had not properly refolded, we performed sedimentation velocity analysis with protein GST–4.1R80Δ16 after denaturation–renaturation treatments. We detected species with sedimentation coefficients, corrected at 20 °C in water, of 6.5, 10 and 15.9 S (results not shown), which may correspond to the monomeric, dimeric and tetrameric forms. These results indicate that protein GST–4.1R80Δ16 has the ability to oligomerize.
Protein 4.1R80Δ16 self-association is achieved after denaturing–renaturing treatments
The N-terminal region of the FERM domain interacts with the core region
The 4.1R FERM domain has been shown to have a cloverleaf-like crystal structure . One major difference between the 4.1R80 and 4.1R60 isoform types is that the former contains a complete FERM domain, while the latter lacks the N-terminal end of this domain, the N-lobe and most of the α-lobe (Figure 6A and 6B). It is reasonable to hypothesize that the N-terminal region of the FERM domain present in 4.1R80 proteins masks the core region in this type of isoforms. To investigate this, we performed pull-down assays using a GST fusion protein containing the 158 N-terminal amino acids of the FERM domain that are absent from 4.1R60 isoforms (GST–4.1RE4–E8). The recombinant protein GST–4.1RE4–E8 bound to Sepharose beads (Figure 6C, lane 1, Coomassie) was incubated with 35S-Met-labelled in vitro-translated 4.1R80Δ16 protein, 4.1R60Δ16 protein, the core region or the Cter region (Figure 6C, inputs). As revealed by the autoradiographs, GST–4.1RE4–E8 bound the small 4.1R60Δ16 isoform and the core region (Figure 6C, lanes 7 and 8), but not the large 4.1R80Δ16 isoform or the Cter region (Figure 6C, lanes 6 and 9). These results support the hypothesis that the core region is hidden in the large 4.1R isoforms by the intramolecular interaction between the N-terminus of the FERM domain and the core region.
The N-terminal sequence of the FERM domain interacts with the core region
4.1R60 isoforms, but not those of 4.1R80, are isolated within the same complex in vivo
The results reported above indicate that small 4.1R60 isoforms self-associate in vitro. We next investigated whether multiple 4.1R60Δ16,18 molecules could be isolated within the same complex in vivo. For this purpose we co-expressed proteins 4.1R60Δ16,18 tagged with two different tags, one at the N-terminus, FLAG–4.1R60Δ16,18, and one at the C-terminus, 4.1R60Δ16,18–GFP, and performed co-immunoprecipitation experiments with the appropriate antibodies (Figure 7A). Immunoprecipitation with the anti-GFP antibody, followed by immunoblotting with the anti-FLAG antibody, revealed the co-immunoprecipitation of the two proteins from lysates of co-transfected COS-7 cells (Figure 7A, lane 2). When the anti-FLAG antibody was used for the immunoprecipitation assay and the anti-GFP antibody for the immunoblotting, both proteins also co-immunoprecipitated (Figure 7A, lane 3). In contrast with these results, when the 4.1R80 isoform 4.1R80Δ16 was co-expressed with the two different tags, FLAG–4.1R80Δ16 and 4.1R80Δ16–GFP, and assayed in co-immunoprecipitation experiments, they did not co-immunoprecipitate (Figure 7A, lane 5). These results are in agreement with those obtained from the pull-down assays (Figure 2).
The small 4.1R60Δ16,18 isoform is self-associated in vivo
Isoforms of protein 4.1R exogenously expressed in COS-7 cells show different sedimentation behaviours
We next performed sucrose-density-gradient centrifugation assays using extracts from COS-7 cells transfected with 4.1R135Δ16–GFP, 4.1R80Δ16–GFP or 4.1R60Δ16,18–GFP cDNAs to compare the sedimentation behaviour of the expressed proteins with that of the recombinant proteins analysed in Figure 1. The lysates were loaded on to linear 5–20% sucrose-density gradients, centrifuged and fractionated as described in the Experimental section. The distribution of the proteins along the gradient was determined by immunoblotting using an anti-GFP antibody (Figure 7B). We observed that proteins 4.1R135Δ16–GFP and 4.1R80Δ16–GFP remained at the top of the gradient, whereas the small protein 4.1R60Δ16,18–GFP did not, but instead fractionated with a sedimentation coefficient higher than that of the large 4.1R proteins. Similar distribution patterns were observed for the 4.1R135 isoforms 4.1R135Δ16,19 and 4.1R135Δ16,18,19 and the 4.1R80 isoforms 4.1R80Δ16,18 and 4.1R80Δ16,19 (results not shown). The similar sedimentation behaviour observed for isoform 4.1R60Δ16,18 isolated from COS-7 cells and for the recombinant protein GST–4.1R60Δ16,18 supports the notion that the small 4.1R60Δ16,18 isoform also self-associates in vivo.
Protein self-association is a very common phenomenon that can confer several structural and functional advantages on proteins, including improved stability, control over the accessibility and specificity of active sites, and increased complexity. Recent structural and biophysical studies show that protein self-association is a key factor in the regulation of proteins such as enzymes, ion channels, receptors and transcription factors . The present study shows that protein 4.1R can also be included within the growing list of proteins that are known to be capable of self-association. The sedimentation data indicate that isoform 4.1R60Δ16,18, previously isolated from human T-cells and whose sequence is contained in 4.1R135 and 4.1R80 isoforms , oligomerizes. Pull-down assays confirmed that 4.1R60Δ16,18 interacts with itself, but not with a larger 4.1R80 isoform. Interestingly, a region conserved in all 4.1R isoforms, previously designated by us as the core region , is involved in self-association.
The fact that all isoforms of protein 4.1R contained the core region and only the small isoform 4.1R60Δ16,18 was found to be self-associated suggested to us that the core region must be hidden in 4.1R80 and 4.1R135 isoform types, but not in 4.1R60 isoforms. This idea was supported by the results of denaturing–renaturing experiments in which an initially non-self-associated 4.1R80 isoform became self-associated (Figure 5). Thus large isoforms of protein 4.1R have the ability to self-associate, but the core region of the molecule needs to become exposed for this interaction to occur.
The core region comprises the N-terminal end of 4.1R60 isoforms; however, up to 367 and 158 amino acids are added N-terminally to this region in the 4.1R135 and 4.1R80 isoforms respectively. It is reasonable to propose that these extra N-terminal extensions might be responsible for hiding the core region. This hypothesis was supported by the results from the pull-down assays showing that the N-terminal region of the FERM domain, which was absent from 4.1R60 isoforms but present in the extensions of 4.1R135 and 4.1R80 isoforms, specifically bound to the core region (Figure 6). It is of note that hiding of the core region was specific to the 4.1R amino acid sequence, as adding an extra N-terminal extension comprised of the GST sequence to that of the small protein 4.1R60Δ16,18 did not prevent self-association of the fusion protein (Figure 1). Taken together, our results imply that 4.1R60 isoforms, lacking the N-terminal region of the FERM domain, have exposed the core region and are therefore self-associated. By contrast, self-association of proteins 4.1R containing a complete FERM domain appears to be modulated by intramolecular interactions.
It is noteworthy that 4.1R60 isoforms are the smallest 4.1R isoforms and that the presence or absence of the N-terminal extension of 4.1R135 and 4.1R80 must confer special characteristics to the different 4.1R isoforms. The results presented here are consistent with those of our previous studies of the nuclear/cytoplasmic distribution of 4.1R proteins, wherein we observed that 4.1R60 isoforms were predominantly distributed to the nucleus due to the nuclear targeting effect of the core region . By contrast, 4.1R135 isoforms also contain the core region and were predominantly localized at non-nuclear sites, implying that the core region is hidden by their extra N-terminal domain [36,41].
A mechanism preventing protein self-association through intramolecular interactions has been reported for many proteins, such as the ERM (ezrin, radixin, moesin) family of proteins belonging to the band 4.1 superfamily. Indeed, monomeric ezrin possesses a conformationally hidden C-terminal domain that, when exposed, can bind to an N-terminal domain (the FERM domain) of a second molecule . It is noteworthy that while two different domains, the N- and C-termini, are involved in ERM intermolecular interactions, only one region, the core region, is involved in 4.1R intermolecular interactions. The WASP (Wiskott–Aldrich syndrome protein) family and the formins mDia1 (a mammalian homologue of the Drosophila diaphanous protein) and FHOD1 (forming homology 2 domain containing protein) also form intramolecular interactions in order to hide regions involved in protein interactions [46,47]. These designs virtually ensure that self-association is physiologically important and imply that self-interacting protein complexes have capabilities that monomers do not. Self-association of spectrin, one of the major structural erythrocyte proteins, proved to be essential for normal erythroid shape and mechanical stability . Mutations in spectrin that impair its ability to self-associate lead to clinically significant forms of HE (hereditary elliptocytosis) and HPP (hereditary pyropoikilocytosis) . Future investigation will shed light on the functional consequences of protein 4.1R self-association.
We thank Dr M. A. Alonso (Centro de Biología Molecular ‘Severo Ochoa’, Madrid, Spain) for invaluable discussions. We are very grateful to Dr C. Alfonso (Centro de Investigaciones Biológicas, Madrid, Spain) for advice and assistance on sedimentation velocity assays. We also thank A. Gosálbez for her excellent technical support. This work was supported by grant number BFU2005-01825 from the Ministerio de Educación y Ciencia, Spain. C. M. P.-F. and E. L. were post-doctoral and pre-doctoral fellows of the Ministerio de Educación y Ciencia, Spain.