Identification and characterization of self-association domains on small ankyrin 1 isoforms

Abstract

In striated muscles, the large scaffolding protein obscurin and a small SR-integral membrane protein sAnk1.5 control the retention of longitudinal SR across the sarcomere. How a complex of these proteins facilitates localization of longitudinal SR has yet to be resolved, but we hypothesize that obscurin interacts with a complex of sAnk1.5 proteins. To begin to address this hypothesis, we demonstrate that sAnk1.5 interacts with itself and identify two domains mediating self-association. Specifically, we show by co-precipitation and FLIM-FRET analysis that sAnk1.5 and another small AnkR isoform (sAnk1.6) interact with themselves and each other. We demonstrate that obscurin interacts with a complex of sAnk1.5 proteins and that this complex formation is enhanced by obscurin-binding. Using FLIM-FRET analysis, we show that obscurin interacts with sAnk1.5 alone and with sAnk1.6 in the presence of sAnk1.5. We find that sAnk1.5 self-association is disrupted by mutagenesis of residues Arg64-Arg69, residues previously associated with obscurin-binding. Molecular modeling of two interacting sAnk1.5 monomers facilitated the identification of Gly31-Val36 as an additional site of interaction, which was subsequently corroborated by co-precipitation and FLIM-FRET analysis. In closing, these results support a model in which sAnk1.5 forms large oligomers that interact with obscurin to facilitate the retention of longitudinal SR throughout skeletal and cardiac myocytes.

1. Introduction

The sarcoplasmic reticulum (SR) is an uninterrupted tubular network surrounding myofibrils that regulates the release and uptake of intracellular calcium necessary for synchronous excitation-contraction coupling in striated muscles. Based on its location in a myocyte, the SR is classified as junctional or longitudinal (also known as network or free SR). The junctional SR is defined as regions that are closely apposed to transverse-tubules and regulate the initiation of calcium-induced calcium release. In contrast, the longitudinal SR is the mesh-like network that runs parallel to myofibrils across the sarcomere connecting junctional SR. Given its proximity to the troponin-tropomyosin complex, the longitudinal SR facilitates uniform contractions in addition to playing an important role in taking up calcium during diastole.

Two proteins necessary for the proper subcellular localization of longitudinal SR in skeletal myocytes are obscurin and sAnk1.5. Obscurin is a large scaffolding protein initially implicated in the formation and alignment of myosin thick filaments at the M-line [1–11]. Interestingly, the most discernable effect of obscurin knockout is a significant decrease in longitudinal SR at the A-band of skeletal myocytes [12]. sAnk1.5 is a small, muscle-specific isoform that is the product of alternative splicing of the ANK1 gene [13–15]. It is tethered to the SR membrane via a transmembrane domain that is unique to this 19 kD isoform and the other small AnkR isoforms (sAnk1.6, 1.7, and 1.9) [13,14,16–18]. Similar to obscurin, sAnk1.5 knockdown or knockout causes the preferential loss of longitudinal SR compared to junctional SR in skeletal myocytes [19,20].

Many studies have demonstrated an interaction between obscurin and sAnk1.5 [16,18,21–25]. Specifically, the 800 kD isoform of obscurin, which localizes to the M-line of the sarcomere, has a unique C-terminal domain that contains two ankyrin-binding sites [26,27]. Likewise, two sites in sAnk1.5 that facilitate obscurin-binding have been reported [22–25]. Given their interaction and that genetic disruption of either protein results in a similar phenotype, it can be inferred that an interaction between sAnk1.5 and obscurin is necessary for the retention of longitudinal SR across the sarcomere. Yet it is unknown how an interaction between obscurin and a small integral membrane protein such as sAnk1.5 would maintain the subcellular localization of the longitudinal SR. We hypothesize that obscurin interacts with a complex of sAnk1.5 proteins, which serve as a protein scaffold within the SR membrane to retain longitudinal SR across the sarcomere. While two studies have reported sAnk1.5 complex formation [17,28], the domains mediating this formation have yet to be elucidated.

This study is the first to provide a detailed analysis of sAnk1.5 self-association. Specifically, we demonstrate that sAnk1.5 interacts with itself and with sAnk1.6 using co-precipitation experiments. Moreover, we demonstrate that obscurin interacts with a complex of sAnk1.5 proteins and that sAnk1.5 complex formation is enhanced by obscurin-binding. Using fluorescence-lifetime imaging microscopy fluorescence resonance energy transfer (FLIM-FRET), we demonstrate in cells that obscurin interacts with sAnk1.5 and with sAnk1.6 only when sAnk1.5 is co-expressed. Interestingly, we find that sAnk1.5 residues Arg64-Arg69, which have been previous associated with promoting obscurin-binding, in fact predominantly mediate sAnk1.5 self-association. We also characterize a second interaction site, Gly31-Val36, which was initially revealed by molecular modeling and subsequently corroborated by mutagenesis and co-precipitation experiments. We use FLIM-FRET to confirm sAnk1.5 and 1.6 interaction in cells. Finally, we also validate that residues Gly31-Val36 and Arg64-Arg69 mediate small AnkR self-association in cells using FLIM-FRET.

2. Methods

2.1. DNA plasmids and antibodies

Human sAnk1.5 (AF005213), sAnk1.6 (NM_020479), and sAnk1.9 (BC007930.2) were subcloned in frame with various C-terminal epitopes (HA, GFP, CFP, YFP, and GST). Human obscurin (NM_052843) was subcloned in frame with a C-terminal HA or GFP epitope (HA: amino acids 6148–6460, GFP: amino acids 6210–6380). Sarcolipin was isolated by reverse-transcriptase PCR from rat ventricular mRNA and subcloned in frame with a C-terminal fusion of CFP and YFP (separated by 6 serine and glycine residues). Human STIM1 (NM_001277962) was subcloned in frame with a C-terminal GST or YFP. Antibodies used in this study include HA (26183, Thermo Scientific), GFP (B-2, sc9996, Santa Cruz), and GST (B-14, sc138, Santa Cruz).

2.2. Co-immunoprecipitation and immunoblot analysis

HEK293TN cells (System Biosciences) were transfected with DNA plasmids using JetPRIME transfection reagent (Polyplus Transfection). After 24 h, transfected cells were subjected to three freeze-thaw cycles in lysis buffer (5 mM HEPES pH 8.0, 0.5 mM EDTA, 0.1 mM MgCl₂, 1 mM Dithiothreitol, 150 mM NaCl, 1% Triton X-100, 1 mM PMSF, 1× protease inhibitor cocktail). Protease inhibitor cocktail (P2714, Sigma-Aldrich) contains 2 mM AEBSF, 0.3 μM Aprotinin, 116 μM Bestatin, 14 μM E-64, 1 μM Leupeptin, and 1 mM EDTA). After incubating lysates for 20 mins at 4 °C, the supernatant was collected following centrifugation at 13,000 × g for 10 mins at 4 °C. 5% lysate was retained for input and the remaining lysate was incubated with 30 μl GST beads (GE Healthcare Life Sciences) or 15 μl GFP-Trap nanobeads (ChromoTek) in binding buffer (50 mM Tris pH 7.4, 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 0.1% Triton X-100) for 1 h at 4 °C. Immunocomplexes were pelleted and washed three times in wash buffer (binding buffer with 1% Triton X-100 and 500 mM to 1 M NaCl). Immunocomplexes were then separated by SDS-PAGE, and visualized by chemiluminescence (ProSignal Femto, Genesee Scientific) using antibodies to GFP (1:1000), GST (1:500), and HA (1:500). Quantification of signal intensity was performed using ImageJ software.

For quantification of obscurin-binding to sAnk1.5 and 1.6, we normalized the amount of obscurin-precipitated to obscurin-input and then to sAnkR-precipitated, which was then divided by the average of three independent replicates to yield a percentage of obscurin-bound to sAnk1.5 or 1.6. For quantification of obscurin- and sAnk1.5-binding to sAnk1.5-GST, we normalized the amount precipitated to input and then to sAnk1.5-GST precipitated. We established sAnk1.5-GST precipitation of sAnk1.5-HA and wild-type obscurin-HA as 100% binding and normalized all other calculations to this standard.

For HA-peptide elution and resolution on native gels, we used the same lysis buffer minus Dithiothreitol and Triton X-100 was substituted with Triton X-114. Cell lysates were incubated with 2 μg HA antibody and precipitated with TrueBlot^® anti-mouse Ig IP agarose beads. Precipitated complexes were washed in a wash buffer containing 500 mM NaCl three times and then incubated at 37 °C for 10 min with 1 mM HA-peptide (APExBIO) to elute the sAnk1.5-HA oligomer. Eluted oligomers were resolved on native and SDS-denaturing gels.

2.3. Modeling and molecular dynamics simulation

An extended conformation of a sAnk1.5 partial peptide (residues Ile28-Gln155) was subjected to a 100 ns MD simulation to generate a rough initial model of a monomer. We then generated 3D printouts of the monomer to help visualize how they may interact. We then performed a 225 ns MD run of two monomers that were aligned in parallel orientation to investigate possible protein-protein interaction sites. Peptides were placed in a cubic box of side ~81 Å, containing 44,184 TIP3P waters and 44 NaCl ions for an ionic strength of 150 mM. The system was energy minimized for 5000 conjugate gradient steps and then equilibrated for 50,000 steps with a position restraint applied to backbone atoms. In the production phase, long-range electrostatic interactions were calculated by the Particle Mesh Ewald (PME) method [29] with SHAKE [30] restraints applied on bonds involving hydrogen atoms. Simulations were performed with a 2 fs timestep under a constant number of particles, temperature (300K) and pressure (1 bar) condition. The Nose-Hoover Langevin piston method was used to maintain constant pressure and Langevin thermostat to maintain constant temperature. Short-range non-bonded forces and long-range electrostatic forces were computed every step. All simulations were performed with the NAMD2.11 program [31] using the CHARMM27 force field [32].

2.4. FLIM-FRET analysis and confocal microscopy

FLIM experiments were performed with a lifetime fluorescence imaging attachment (Lambert Instruments, Leutingewolde, Netherlands) on an inverted microscope (Nikon). HEK293TN cells were plated on coverslips and transfected with CFP(donor)/YFP(acceptor) tagged constructs at a 1:1 ratio. After 24 h, transfected cells were washed in 1× PBS, fixed in 2% paraformaldehyde, quenched in 50 mM NH₄Cl for 10 min, and washed in 1× PBS again. Coverslips were mounted on slides in 10% Mowiol mounting media (Sigma-Aldrich) and dried at room temperature for 24 h. Transfected cells were excited by using a sinusoidally modulated 3-watt 448-nm light-emitting diode at 40 MHz under epi-illumination. Fluorescein at the concentration of 1 μM was used as a lifetime reference standard (4 ns). Cells were imaged with a 40× oil-immersion objective (numerical aperture 1.3) using a CFP filter set. The phase and modulation lifetimes were determined from 12 phase settings by using the manufacturer’s software (LIFA). The analysis yields the CFP lifetime of free CFP in donor only (τ1) and the CFP lifetime in donor-acceptor complexes (τ2). FLIM data were averaged on a per cell basis and three experiments (at least 65 cells in total) were performed for each condition. FRET efficiency is calculated as (1 - τ2/ τ1) and represented as a percentage in Table 1. CFP and YFP images corresponding to the LIFA image were acquired using a Nikon A1R confocal microscope with a 40× oil-immersion objective (numerical aperture 1.3).

Table 1.

FRET Efficiencies. FRET efficiency is calculated as (1 - τ2/ τ1) where τ1 is the CFP lifetime of obscurin-CFP alone (for Fig. 4) or sAnk1.5-CFP alone (for Figs. 9&10) and τ2 is the CFP lifetime in donor-acceptor complexes.

Donor	Acceptor	FRET efficiency (%)	n
obscurin-CFP		0.01 ± 0.12	99
obscurin-CFP	sAnk1.5-YFP	12.0 ± 0.35	88
obscurin-CFP	sAnk1.6-YFP	0.3 ± 0.08	125
sAnk1.5-CFP		−0.2 ± 0.14	67
sarcolipin-CFP-YFP		17.2 ± 0.4	50
sAnk1.5-CFP	STIM1-YFP	3.0 ± 0.3	84
sAnk1.5-CFP	sAnk1.5-YFP	15.3 ± 0.38	97
sAnk1.5-CFP	sAnk1.5ΔABD1-YFP	6.8 ± 0.42	78
sAnk1.5-CFP	sAnk1.5ΔABD2-YFP	6.5 ± 0.69	109
sAnk1.6-CFP	sAnk1.6-YFP	21.1 ± 0.51	60
sAnk1.6-CFP	sAnk1.6ΔABD1-YFP	2.5 ± 0.22	107
sAnk1.6-CFP	sAnk1.6ΔABD2-YFP	1.2 ± 0.39	78
sAnk1.5-CFP	sAnk1.6-YFP	12.1 ± 0.42	76
sAnk1.5-CFP	sAnk1.6ΔABD1-YFP	2.0 ± 0.16	108
sAnk1.5-CFP	sAnk1.6ΔABD2-YFP	−1.4 ± 0.37	69

2.5. Statistics

For Western blot analysis, signal intensity was quantified using ImageJ (NIH) and the results are represented as mean ± SEM with statistical significance determined by paired Student’s t-test (2-way). For FLIM-FRET analysis, the results are represented as mean ± SEM with statistical significance determined by unpaired Student’s t-test (2-way). P values of < 0.05, < 0.01 and < 0.001 are represented by *, **, and *** respectively.

3. Results

3.1. sAnk1.5 and 1.6 interact with themselves and each other

To investigate whether sAnk1.5 interacts with itself, we performed co-precipitation experiments to assess sAnk1.5-HA precipitation by sAnk1.5-GFP. Specifically, we used GFP-nanobeads to precipitate GFP associated protein complexes from protein lysates isolated from HEK293TN cells transiently expressing sAnk1.5-HA with GFP or sAnk1.5-GFP. Our results demonstrate that while sAnk1.5-HA was expressed at similar levels in both conditions, only sAnk1.5-GFP precipitated sAnk1.5-HA (Fig. 1A). In the sAnk1.5-GFP co-transfection condition, we believe the 28 kD product is GFP that has been cleaved from sAnk1.5-GFP. Considering GFP is expressed in the cytosol and sAnk1.5 is expressed in the ER membrane, we repeated these experiments using STIM-1 as a negative control because it is expressed in the ER membrane. In addition, we used an alternative precipitation strategy using glutathione S-transferase (GST) as the epitope tag for STIM-1 and sAnk1.5. In this experiment, sAnk1.5-HA does not co-precipitate with the negative control STIM1-GST, but does co-precipitate with sAnk1.5-GST even in the presence of increasingly stringent wash buffers with salt concentrations from 500 mM to 1 M (Fig. 1B). A longer exposure of the GST immunoblot in Fig. 1B is provided in Fig. S1 to demonstrate sAnk1.5-GST expression in the input condition. We also demonstrate that sAnk1.5-HA does not precipitate with GST protein alone (Fig. S2).

Fig. 1.

sAnk1.5 interacts with itself. (A) sAnk1.5-GFP immunoprecipitates sAnk1.5-HA. HA and GFP immunoblots demonstrate expression and precipitation of sAnk1.5-HA by sAnk1.5-GFP but not GFP alone. (B) sAnk1.5-GST precipitates sAnk1.5-HA in stringent wash conditions with increasing salt concentrations from 500 mM to 1 M. HA and GST immunoblots demonstrate expression and precipitation of sAnk1.5-HA by sAnk1.5-GST but not the negative control STIM1-GST.

Striated muscles express alternative sAnk1.5 isoforms including the isoform sAnk1.6 [13–16,18]. Fig. 2A presents the amino acid alignment of sAnk1.5 to 1.6, highlighting the location of the transmembrane domain and two previously-reported obscurin-binding domains (OBD). Interestingly, sAnk1.6 lacks the majority of residues that constitute the second obscurin-binding domain. To determine whether sAnk1.6 also interacts with itself, we assessed sAnk1.6-HA co-precipitation with sAnk1.6-GST or STIM1-GST. Our results demonstrate that sAnk1.6-HA co-precipitates with sAnk1.6-GST but not with STIM1-GST (Fig. 2B). These findings suggest that self-association domains are present in both sAnk1.5 and 1.6. When evaluating sAnk1.6 interaction with sAnk1.5, we find that sAnk1.5-HA co-precipitates with sAnk1.6-GST, but not with STIM1-GST (Fig. 2C). Taken together, these findings demonstrate that sAnk1.5 and 1.6 interact with themselves and each other.

Fig. 2.

sAnk1.6 interacts with itself and with sAnk1.5. (A) Amino acid alignment of sAnk1.5 and 1.6. Transmembrane domain (TMD) and obscurin-binding domains (OBD) are highlighted. HA and GST immunoblots demonstrate expression and precipitation of sAnk1.6-HA by sAnk1.6-GST but not STIM1-GST (B) and of sAnk1.5-HA by sAnk1.6-GST but not STIM1-GST (C).

Striated muscles also express sAnk1.9, another isoform that has a different C-terminal domain than sAnk1.5 after Glu123 (Fig. S3A). Like sAnk1.5 and 1.6, we find that sAnk1.9 interacts with itself and sAnk1.5 (Fig. S3B&C). We also assessed the self-association of AnkG107, a muscle-specific ankyrin-G isoform with similarly expressed obscurin-binding domains in its C-terminus [33]. Unlike the sAnk1 isoforms, AnkG107-GST did not precipitate AnkG107-HA, suggesting that AnkG107 does not interact with itself (Fig. S4).

3.2. Obscurin binds sAnk1.5 but not sAnk1.6

Considering the majority of the second obscurin-binding domain is not present in sAnk1.6, we measured obscurin-binding to sAnk1.5 and 1.6. As obscurin is an 800 kD scaffolding protein, we used an HA-tagged construct of the obscurin C-terminal domain encoding residues 6148 to 6460, which contain two ankyrin-binding domains. Fig. 3A shows the position of the ankyrin-binding domains relative to other domains in obscurin such as the immunoglobulin and fibronectin type 3 domains. Our findings demonstrate that sAnk1.5-GST precipitates obscurin, but sAnk1.6-GST precipitates obscurin minimally (Fig. 3B). We quantified the results normalizing obscurin-binding to input and sAnkR-precipitated from three independent experiments (see Methods for a more detailed description). While sAnk1.5 demonstrates maximal obscurin-binding (101.1%), sAnk1.6 demonstrates minimal obscurin-binding (4.1%) (Fig. 3C). These findings are consistent with previous reports that sAnk1.6 does not interact with obscurin [16,18]. We also measured obscurin-binding to sAnk1.9, which like sAnk1.5 has the two previously-reported obscurin-binding domains. Like sAnk1.5-GST, sAnk1.9-GST precipitates obscurin-HA (Fig. S3D).

Fig. 3.

sAnk1.5 interacts with sAnk1.5 and obscurin. (A) Schematic of the relative positions of obscurin domains including immunoglobulin (IG), fibronectin type 3 (FN3), calmodulin-binding motif (IQ), Src homology 3 (SH3), guanine nucleotide exchange factor for Rho/Rac/Cdc42-like GTPases (RhoGEF), pleckstrin homology (PH), and ankyrin-binding domains (ABD). (B) HA and GST immunoblots demonstrate expression and precipitation of obscurin-HA by sAnk1.5-GST but not sAnk1.6-GST. (C) Quantification of obscurin-binding by sAnk1.5 and 1.6. (D) HA and GST immunoblots demonstrate expression and precipitation of obscurin-HA and sAnk1.5-HA by sAnk1.5-GST but not STIM1-GST. sAnk1.6-GST only precipitates sAnk1.5-HA but not obscurin-HA.

3.3. sAnkR1.5 interacts with itself and obscurin

To assess whether sAnk1.5 or 1.6 interacts with both sAnk1.5 and obscurin, we performed co-precipitation assays. Our findings demonstrate that sAnk1.5-GST precipitates obscurin-HA and sAnk1.5-HA (Fig. 3D). In contrast, sAnk1.6-GST only precipitates sAnk1.5-HA but not obscurin-HA. Also, neither sAnk1.5-HA nor obscurin-HA co-precipitates with the negative control STIM1-GST. These findings demonstrate that sAnk1.5 can form a complex with itself and obscurin. We also found that sAnk1.9 can form a complex with sAnk1.5-HA and obscurin (Fig. S3E).

3.4. In cells obscurin interacts with sAnkR1.5 alone and with sAnk1.6 in the presence of sAnk1.5

To investigate obscurin’s interaction with sAnk1.5 or 1.6 in cells, we performed fluorescent lifetime imaging microscopy fluorescent resonance energy transfer (FLIM-FRET). With this technique, the lifetime of the donor fluorophore is reduced by energy transfer between the donor (CFP) and acceptor (YFP) fluorophores when they are within 10 nm of each other. Using a Lambert Instruments FLIM Attachment (LIFA) microscope, fluorescent lifetime per cell was measured and pseudocolored images represent the CFP lifetime, which ranges from 2.3 ns (red) to 1.9 ns (blue) based on experimental results. We also acquired representative LIFA images and the average lifetime (or τ_av) was calculated from > 80 cells from three independent experiments.

For these experiments, we transiently transfected HEK293TN cells with different CFP and YFP constructs and measured their expression by immunoblot analysis with a GFP antibody that recognizes both the CFP and YFP fluorophores (Fig. 4E). The positive control is the small ER membrane protein sarcolipin, which oligomerizes into pentamers [34], conjugated to a fusion protein of CFP and YFP fluorophores separated by six amino acids. We calculated the average CFP lifetime of donor alone (i.e. obscurin-CFP) is 2.29 ± 0.003 ns (n = 99) while the average lifetime of the positive control sarcolipin-CFP-YFP is 1.91 ± 0.004 ns (n = 102) (Fig. 4A&D). Consistent with our biochemical results, FLIM-FRET demonstrates a molecular interaction between obscurin-CFP and sAnk1.5-YFP alone. Specifically, the energy transfer between CFP and YFP fluorophores attenuates the average CFP lifetime of obscurin-CFP from 2.29 ± 0.003 ns to 2.02 ± 0.008 ns (n = 88) in the presence of sAnk1.5-YFP (Fig. 4B&D). No molecular interaction is detected between obscurin-CFP and sAnk1.6-YFP alone in cells as the CFP lifetime of obscurin-CFP is 2.29 ± 0.002 ns (n = 125) in the presence of sAnk1.6-YFP alone (Fig. 4C&D).

Fig. 4.

FLIM-FRET analysis demonstrating obscurin interaction with sAnk1.5 alone or with sAnk1.6 in the presence of sAnk1.5 in cells. (A) Images from FLIM-FRET analysis (pseudocolored images showing average CFP lifetime: τ_av) of CFP donor alone (obscurin-CFP) and the positive control (sarcolipin-CFP-YFP). (B) Images of FLIM-FRET analysis of obscurin-CFP (donor) with sAnk1.5-YFP (acceptor) in the presence of sAnk1.5-HA or sAnk1.6-HA. (C) Images of FLIM-FRET analysis of obscurin-CFP (donor) with sAnk1.6-YFP (acceptor) in the presence of sAnk1.5-HA or sAnk1.6-HA. (D) The graph shows mean fluorescence lifetime of CFP ± SEM (n ≥ 80, from three independent experiments). Statistical analysis was performed by unpaired Student’s t-test (*** p < .001). (E) HA and GFP immunoblots demonstrate expression of obscurin-CFP, sAnk1.5-YFP, sAnk1.6-YFP, sAnk1.5-HA and sAnk1.6-HA.

To examine whether sAnk1.5 forms a complex with sAnk1.6 and obscurin in cells, we performed FLIM-FRET analysis on cells expressing obscurin-CFP, sAnk1.6-YFP, and sAnk1.5-HA or sAnk1.6-HA. Interestingly, the co-expression of sAnk1.5-HA facilitates a molecular interaction between obscurin-CFP and sAnk1.6-YFP as the CFP lifetime of obscurin-CFP decreases to 2.02 ± 0.008 ns (n = 108) in the presence of sAnk1.6-YFP (Fig. 4C&D). In contrast, the CFP lifetime of obscurin-CFP is 2.29 ± 0.003 ns (n = 119) in the presence of sAnk1.6-YFP and sAnk1.6-HA (Fig. 4C&D), demonstrating that obscurin does not interact with a complex of sAnk1.6 proteins. Furthermore, immunoblot analysis demonstrates that co-expression of sAnk1.5-HA or sAnk1.6-HA does not noticeably change the expression of obscurin-CFP or sAnk1.5-YFP (Fig. 4E); therefore, the enhanced interaction is not mediated by increased fluorophore expression. These findings suggest that when sAnk1.5-HA interacts with obscurin-CFP and sAnk1.6-YFP, the fluorophores are in close enough proximity to manifest FRET.

To assess whether sAnk1.5 and obscurin interactions could be enhanced or inhibited by sAnk1.5 or sAnk1.6 respectively, we performed similar FLIM-FRET experiments on cells expressing obscurin-CFP, sAnk1.5-YFP, and sAnk1.5-HA or sAnk1.6-HA. It is noteworthy that the CFP lifetime of obscurin-CFP is slightly decreased to 1.98 ± 0.007 ns (n = 103) in the presence of sAnk1.5-YFP and sAnk1.5-HA (Fig. 4B&D), suggesting that the additional sAnk1.5-HA may enhance obscurin-CFP and sAnk1.5-YFP interactions. In contrast, co-expression of sAnk1.6-HA did not alter the CFP lifetime of obscurin-CFP (2.03 ± 0.01 ns, n = 86) in the presence of sAnk1.5-YFP (Fig. 4B&D). Unfortunately, sAnk1.6-HA expresses much less than sAnk1.5-HA (Fig. 4E) so an effect of sAnk1.6-HA co-expression on obscurin-CFP lifetime may not be detectable due to its low expression.

3.5. Obscurin enhances sAnkR1.5 self-association

Two ankyrin-binding domains have been characterized in the C-terminal domain of obscurin consisting of residues 6232–6240 and 6325–6335 [16,18,23–25] (Fig. 5A). To assess whether obscurin-binding alters sAnk1.5 self-association, we generated conservative mutations in the ankyrin-binding sites, converting acidic residues to uncharged glutamine residues. We performed co-precipitation experiments evaluating sAnk1.5 precipitation of itself and various obscurin constructs. The experiments were repeated in triplicate and immunoblot signal intensity was measured with ImageJ to quantify the results. Considering co-transfection of wild-type or mutant obscurin constructs increases expression of sAnk1.5 constructs by 2–3 fold, we normalized the signal intensity measurements to construct expression and the amount of GST precipitated. We then established 100% binding as the measurements of sAnk1.5-GST precipitation of sAnk1.5-HA and wild-type obscurin.

Fig. 5.

sAnk1.5 proteins independently self-associate, but obscurin-binding enhances complex formation. (A) Alignment of obscurin amino acid sequences of wild-type and ankyrin-binding domain mutants (ΔABD1, ΔABD2, ΔABD1&2). (B) HA and GST immunoblots demonstrate expression and precipitation of wild-type (wt) and mutant (ΔABD1, ΔABD2) obscurin-HA and sAnk1.5-HA by sAnk1.5-GST. (C) Quantification of sAnk1.5-GST precipitation of sAnk1.5-HA and obscurin-HA.

Consistent with our previous findings, sAnk1.5-GST precipitates sAnk1.5-HA alone and in combination with wild-type obscurin-HA (Fig. 5B). Interestingly, co-transfection of wild-type obscurin increases sAnk1.5-GST precipitation of sAnk1.5-HA by 5-fold (Fig. 5C). This effect is negated when obscurin’s first ankyrin-binding domain is mutated. Also, this mutation reduces sAnk1.5-GST precipitation of obscurin by ~85%. In contrast, mutating obscurin’s second ankyrin-binding domain did not significantly reduce sAnk1.5-GST precipitation of sAnk1.5-HA, although it did reduce obscurin precipitation by ~40%. When both ankyrin-binding domains in obscurin are mutated, sAnk1.5-GST no longer precipitates obscurin, but it still precipitates sAnk1.5-HA at levels similar to the transfection condition without obscurin. These findings suggest that obscurin enhances complex formation between sAnk1.5 proteins and obscurin, an effect that is regulated predominantly by obscurin’s first ankyrin-binding domain.

3.6. Identification of two domains that mediate small AnkR self-association

Considering sAnk1.6 had minimal binding to obscurin, we examined the arginine residues in the putative obscurin-binding site (Arg64-Phe71) for ankyrin-binding via electrostatic interactions. Specifically, we mutated the basic arginine residues to uncharged glutamine residues (Fig. 6A). Mutating these residues in sAnk1.5-HA markedly decreases its precipitation by sAnk1.5-GST compared to wild-type sAnk1.5-HA (Fig. 6B). Interestingly, when the mutation (ΔABD2) is introduced into sAnk1.5-GST, sAnk1.5-HA precipitation is notably reduced, but there is no effect on the co-precipitation of obscurin (Fig. 6C). Likewise, mutating these residues in sAnk1.6-GST dramatically reduces co-precipitation of sAnk1.6-HA or sAnk1.5-HA (Fig. 7B&C). Mutating these residues in sAnk1.9 similarly decreases its self-association (Fig. S3B). These findings suggest that Arg64-Arg69 has minimal obscurin-binding activity and in fact predominantly mediates small AnkR self-association.

Fig. 6.

Identification of ankyrin-binding site in sAnk1.5. (A) Amino acid alignment of wild-type and mutant (ΔABD2) sAnk1.5. (B) HA and GST immunoblots demonstrate expression and precipitation of wild-type but not ΔABD2 sAnk1.5-HA by sAnk1.5-GST. (C) HA and GST immunoblots demonstrate expression and precipitation of obscurin-HA and sAnk1.5-HA by wild-type sAnk1.5-GST. In contrast, mutant sAnk1.5-GST (ΔABD2) precipitates far less sAnk1.5-HA but maintains its interaction with obscurin-HA. (D) YFP, HA, and GST immunoblots demonstrate expression and precipitation of sAnk1.5-HA and wild-type or mutant sAnk1.5-YFP by obscurin-GST. The decrease in mutant sAnk1.5-YFP precipitated corresponds with an increase in sAnk1.5-HA precipitated by obscurin-GST. (E) Quantification of increased sAnk1.5-HA precipitated and decreased mutant sAnk1.5-YFP precipitated by obscurin compared to obscurin precipitation of wild-type sAnk1.5-YFP and -HA (represented as % change in binding).

Fig. 7.

Identification of ankyrin-binding site in sAnk1.6. (A) Amino acid alignment of wild-type and mutant (ΔABD2) sAnk1.6. (B) HA and GST immunoblots demonstrate expression and precipitation of obscurin-HA and sAnk1.6-HA by wild-type sAnk1.6-GST. In contrast, mutant sAnk1.6-GST (ΔABD2) does not precipitate sAnk1.6-HA or obscurin-HA. (C) HA and GST immunoblots demonstrate expression and precipitation of obscurin-HA and sAnk1.5-HA by wild-type sAnk1.6-GST. In contrast, mutant sAnk1.6-GST (ΔABD2) precipitates far less sAnk1.5-HA and does not interact with obscurin-HA.

Based on our hypothesis that obscurin interacts with a complex of sAnk1.5 proteins, we predicted that obscurin would precipitate more wild-type (self-associating) sAnk1.5 proteins than mutant ones. To test this prediction, we measured obscurin precipitation of wild-type sAnk1.5-HA and wild-type or mutant (ΔABD2) sAnk1.5-YFP (Fig. 6D). Using different epitope tags allowed the sAnk1.5 proteins to be discriminated by size on protein gels. Our results demonstrate that obscurin precipitates similar amounts of wild-type sAnk1.5-HA and -YFP. In contrast, obscurin precipitates less mutant sAnk1.5-YFP and more wild-type sAnk1.5-HA. Specifically, quantifying the results from three independent experiments reveals that compared to obscurin precipitation of the wild-type sAnk1.5 proteins, obscurin precipitates 46.7 ± 6.6% less mutant sAnk1.5-YFP and 36.7 ± 11.9% more wild-type sAnk1.5-HA (Fig. 6E). These results are consistent with the hypothesis that obscurin interacts with a complex of self-associating sAnk1.5 proteins.

Initially, we attempted to identify a complementary binding site to Arg64-Arg69 focusing on the four basic arginine residues. Specifically, we performed co-precipitation experiments with sAnk1.5 mutants that removed the negative charges associated with acidic residues Glu43-Glu50 or Glu87-Glu93. Neither mutant disrupted sAnk1.5 self-association (data not shown). We then decided to create a model of sAnk1.5 dimers to identify other potential sites of interaction considering the structure of sAnk1.5 has not been determined. First, we used molecular dynamics (MD) simulation and energy-minimization to generate a rough model of a sAnk1.5 monomer consisting of residues Ile28-Gln155. This stretch of 128 residues lacks the transmembrane domain and has no predicted domains. We then generated 3D printouts of the monomers to help visualize how they may interact. With this insight, we then combined two monomers in a parallel arrangement and conducted a second MD simulation with the purpose of identifying additional sites of self-association, but not to determine the structure of sAnk1.5 dimers. Other groups have used MD simulation in a similar manner to model the folding of peptides and protein-protein interactions [35,36]. Fig. 8A displays a cartoon representation of two interacting sAnk1.5 monomers (one in green, the other in silver) with a putative interaction site highlighted in red. As we had demonstrated that Arg64-Arg69 mediates ankyrin-binding, we focused on opposing residues in close proximity, specifically Gly31-Val36 (Fig. 8B). As there are no acidic residues in this region to mediate an electrostatic interaction, we generated a mutant sAnk1.5 with decreased hydrophobicity in this region (Fig. 8C). When evaluating the role of Gly31-Val36 in sAnk1.5 self-association, we find a marked reduction in sAnk1.5-GST co-precipitation of mutant compared to wild-type sAnk1.5-HA (Fig. 8D). In contrast, mutating these residues has no effect on obscurin co-precipitation of sAnk1.5 (Fig. 8E). We also investigated the hydrophobic residues flanking ABD1 by generating another mutant that substituted residues Leu33, Leu37, Lys38, and Ile40 with alanines. This mutant construct displayed similar levels of sAnk1.5 self-association as that of wild-type sAnk1.5 (data not shown). Based on our findings, we propose that sAnk1.5 self-association is mediated in part by two ankyrin-binding domains consisting of Gly31-Val36 (ABD1) and Arg64-Arg69 (ABD2).

Fig. 8.

Identification of a second ankyrin-binding site in sAnk1.5. (A) Cartoon representation of sAnk1.5 dimer formation with one labeled in green and the other labeled in silver. Predicted sites of interaction are highlighted in red. (B) Magnification of interaction site with gold residues labeling the predicted ankyrin-binding domain (ABD1) and the blue residues labeling ankyrin-binding domain 2 (ABD2). (C) Alignment of sAnk1.5 amino acid sequences of wild-type and ankyrin-binding domain 1 mutant (ΔABD1). (D) HA and GST immunoblots demonstrate expression and precipitation of wild-type sAnk1.5-HA by sAnk1.5-GST. In contrast, sAnk1.5-GST precipitates less mutant sAnk1.5-HA (ΔABD1). (E) HA and GFP immunoblots demonstrate expression and precipitation of wild-type and mutant sAnk1.5-HA by obscurin-GFP. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To determine whether mutating ABD1 or ABD2 would alter sAnk1.5 protein complex formation, we assessed wild-type and mutant sAnk1.5-HA (ΔABD1 or 2) oligomers resolved on native and SDS-denaturing gels. HEK293 cell lysates expressing wild-type or mutant sAnk1.5-HA protein were immunoprecipitated with an HA antibody and TrueBlot^® anti-mouse Ig IP agarose beads. Precipitated complexes were washed in a wash buffer containing 500 mM NaCl and 1% Triton X-114 three times and then incubated at 37 °C for 10 min with 1 mM HA-peptide (APExBIO) to elute the sAnk1.5-HA oligomers. Eluted oligomers were resolved on native and SDS-denaturing gels. On the native gel, there were no appreciable differences in eluted protein complexes between wild-type and mutant sAnk1.5-HA suggesting that mutating ABD1 or ABD2 does not disrupt sAnk1.5-HA oligomer formation (Fig. S5). The SDS-denaturing gel demonstrates that sAnk1.5-HA proteins were expressed at similar levels.

3.7. FLIM-FRET analysis corroborates sAnkR self-association and sites of interaction

To examine sAnk1.5 self-association in cells, we performed FLIM-FRET analysis. We measured the fluorescent lifetime per cell and acquired pseudocolored images representing CFP lifetimes, which range from 2.3 ns (red) to 1.7 ns (blue). We also acquired confocal images of representative LIFA images displaying the CFP, YFP, and merged channels. The average lifetime (or τ_av) was calculated from > 50 cells from three independent experiments.

For our experiments, we transiently transfected HEK293 cells with different CFP and YFP constructs in a 1:1 ratio and confirmed similar expression levels using immunoblot analysis with a GFP antibody that recognizes both fluorophores (Fig. S6). The positive control is the small ER membrane protein sarcolipin conjugated to a fusion protein of CFP and YFP fluorophores separated by six amino acids. The negative control is YFP-tagged STIM1. The average CFP lifetime of sAnk1.5-CFP alone is 2.26 ± 0.03 ns (n = 67) (Fig. 9A&C). In contrast, the average CFP lifetime of the positive control sarcolipin-CFP-YFP is 1.87 ± 0.06 ns (n = 50), while the average lifetime of the negative control (i.e. sAnk1.5-CFP and STIM1-YFP) is 2.19 ± 0.06 ns (n = 84). Consistent with our biochemical results, FLIM-FRET demonstrates a molecular interaction between sAnk1.5-CFP and sAnk1.5-YFP. Specifically, the energy transfer between CFP and YFP fluorophores attenuates the average CFP lifetime from 2.26 ± 0.03 ns (sAnk1.5-CFP alone) to 1.91 ± 0.08 ns (n = 97) (Fig. 9B&C). Moreover, mutating either ankyrin-binding domain in sAnk1.5-YFP results in increased average CFP lifetimes (i.e. ΔABD1: 2.10 ± 0.08 ns, n = 78; ΔABD2: 2.11 ± 0.10 ns, n = 109) due to a decrease in energy transfer between fluorophores, which is indicative of decreased sAnk1.5 interactions.

Fig. 9.

FLIM-FRET analysis demonstrating sAnk1.5 self-association in cells. (A) Images from confocal microscopy (CFP, YFP, merge) and FLIM-FRET analysis (pseudocolored images showing average CFP lifetime: τ_av) of CFP donor alone (sAnk1.5-CFP), positive control (sarcolipin-CFP-YFP), and negative control (sAnk1.5-CFP and STIM1-YFP). (B) Images from confocal microscopy and FLIM-FRET analysis of wild-type sAnk1.5-CFP (donor) with acceptors: wild-type sAnk1.5-YFP or mutant sAnk1.5-YFP (either ΔABD1 or ΔABD2). Scale bar represents 10 μm. (C) The graph shows mean fluorescence lifetime of CFP ± SEM (n ≥ 65, from three independent experiments). Statistical analysis was performed by unpaired Student’s t-test (*** p < .001).

We also performed FLIM-FRET experiments to examine sAnk1.6 interaction with itself or with sAnk1.5. The average CFP lifetime of cells expressing sAnk1.6-CFP and sAnk1.6-YFP is 1.78 ± 0.09 ns (n = 60) (Fig. 10A&C), demonstrating an energy transfer due to sAnk1.6 interaction with itself. Moreover, there is a significant decrease in energy transfer when either ankyrin-binding domain is mutated. Specifically, the average CFP lifetimes of sAnk1.6-CFP co-transfected with sAnk1.6ΔABD1-YFP is 2.20 ± 0.05 (n = 107) or sAnk1.6ΔABD2-YFP is 2.23 ± 0.08 ns (n = 78). Regarding sAnk1.5 interaction with sAnk1.6, we demonstrate that the CFP lifetime of cells expressing sAnk1.5-CFP and sAnk1.6-YFP is 1.98 ± 0.08 ns (n = 76) (Fig. 10B&C). Also, mutating either ankyrin-binding domain in sAnk1.6-YFP significantly attenuates the energy transfer between sAnk1.5-CFP and sAnk1.6-YFP. In fact, the average CFP lifetimes of sAnk1.5-CFP cells co-transfected with sAnk1.6ΔABD1-YFP is 2.21 ± 0.04 ns (n = 108) or sAnk1.6ΔABD2-YFP is 2.29 ± 0.06 ns (n = 69). FRET efficiencies for donor/acceptor pairs in Figs. 4, 9 and 10 have been calculated and are presented in Table 1. As the FRET efficiencies of the sAnkR fluorophores are similar to the positive control in which the CFP and YFP fluorophores are physically coupled, this data strongly supports that homodimers and heterodimers of sAnk1.5 and 1.6 are highly efficient FRET partners. Taken together, the FLIM-FRET results strongly corroborate our biochemical data that [1] sAnkR isoforms interact with themselves and each other and [2] sites mediating self-association are Gly31-Val36 (ABD1) and Arg64-Arg69 (ABD2).

Fig. 10.

FLIM-FRET analysis demonstrating sAnk1.6 interaction with itself or with sAnk1.5 in cells. (A) Images from confocal microscopy (CFP, YFP, merge) and FLIM-FRET analysis (pseudocolored images showing average CFP lifetime: τ_av) of wild-type sAnk1.6-CFP (donor) alone or with acceptors: wild-type sAnk1.6-YFP or mutant sAnk1.6-YFP (either ΔABD1 or ΔABD2). (B) Images from confocal microscopy and FLIM-FRET analysis of wild-type sAnk1.5-CFP (donor) with acceptors: wild-type sAnk1.6-YFP or mutant sAnk1.6-YFP (either ΔABD1 or ΔABD2). Scale bar represents 10 μm. (C) The graph shows mean fluorescence lifetime of CFP ± SEM (n ≥ 70, from three independent experiments). Statistical analysis was performed by unpaired Student’s t-test (*** p < .001).

4. Discussion

The distribution of the sarcoplasmic reticulum as junctional and longitudinal structures ensures the rapid and efficient release and uptake of intracellular calcium necessary for uniform and consistent excitation-contraction coupling in striated muscle. Previous studies have demonstrated that the maintenance of the longitudinal SR requires the large scaffolding protein obscurin and the small SR-integral adaptor protein sAnk1.5 [12,19,20]. How a complex of these proteins regulates the retention of longitudinal SR across the sarcomere has yet to be fully elucidated.

Prior to this study, it was known that sAnk1.5 interacts with obscurin and that higher-order sAnk1.5 complexes are present in skeletal muscle SR [17]. In addition, two ankyrin-binding sites had been characterized in the C-terminal domain of obscurin [16,18,23–25] and two obscurin-binding sites had been characterized in sAnk1.5 [22–25]. This study is the first to demonstrate self-association of heterologously expressed sAnk1.5, 1.6, and 1.9. Specifically, we demonstrate that sAnk1.5, 1.6, and 1.9 interact with themselves and each other in co-precipitation experiments. While the results demonstrate sAnkR self-association, they do not preclude the possibility that sAnk1.5 forms higher order complexes that merely resolve to the two different populations (GST- or HA-tagged) on an SDS-denaturing gel. In fact, sAnk1.5-HA protein complexes eluted with an HA peptide resolve on a native gel as a large protein complex spanning ~90–200 kD (Fig. S5).

This study is also the first to use FLIM-FRET to visualize small AnkR interaction in cells. Compared to co-precipitation experiments, this technique is more sensitive detecting protein-protein interactions at the level of a single cell. While the CFP lifetime of sAnk1.5-CFP alone is 2.26 ± 0.03 ns, it is significantly reduced to 1.87 ± 0.06 ns with the positive control, sarcolipin conjugated to a fusion protein of CFP and YFP fluorophores. For the negative control (sAnk1.5-CFP and STIM1-YFP), the CFP lifetime is 2.19 ± 0.06 ns, corroborating the biochemical results that these proteins do not interact. In contrast, the CFP lifetimes are significantly reduced in transfection conditions demonstrating interaction of sAnk1.5-CFP with sAnk1.5-YFP (1.91 ± 0.08 ns) or with sAnk1.6-YFP (1.98 ± 0.06 ns) and sAnk1.6-CFP with sAnk1.6-YFP (1.78 ± 0.09 ns). Moreover, the FRET efficiencies of these various combinations of sAnkR fluorophores are similar to the positive control (Table 1) suggesting that sAnk1.5 and 1.6 homodimers and heterodimers are highly efficient FRET pairs. Taken together, these results further validate our biochemical results demonstrating small AnkR self-association.

This study is the first to identify two segments that mediate small AnkR self-association, Arg64-Arg69 and Gly31-Val36. Previous studies have demonstrated that residues Arg64-Phe71 promote sAnk1.5 binding to obscurin [22–25]. Specifically, alanine mutagenesis (or charge reversal, Arg to Glu) of individual arginine residues 64, 67, 68, or 69 significantly reduced sAnk1.5 binding to obscurin as assessed by Far Western blot analysis and surface plasmon resonance [22,25]. For the Far Western experiments, binding was assessed by incubating fusion proteins of maltose-binding protein (MBP) conjugated to wild-type or mutant sAnk1.5 with affinity-purified GST-obscurin constructs separated by SDS-PAGE and transferred to a nitrocellulose membrane. While these previous studies did not assess whether MBP-sAnk1.5 could self-associate, it is possible to re-interpret their results with this new perspective. Specifically, if MBP-sAnk1.5 self-associates, then the GST-obscurin construct would bind more wild-type sAnk1.5 and mutations that disrupt self-association (i.e. R64A) would manifest as decreased sAnk1.5 bound to obscurin. It is our understanding that the surface plasmon resonance experiments were performed using the same fusion proteins thus it’s possible that the difference in binding kinetics may reflect obscurin binding to an oligomer of sAnk1.5 versus binding to individual sAnk1.5 proteins.

We gained insight into the function of Arg64-Arg69 while performing co-precipitation experiments with the alternative isoform sAnk1.6, which lacks the majority of the obscurin-binding domain Lys100-Asp111. First, sAnk1.6 interaction with itself revealed that self-association domains are present in both isoforms. Also, sAnk1.6 displayed minimal obscurin-binding activity (4.1%) compared to sAnk1.5 (101.1%). These findings were consistent with previous studies that demonstrated obscurin preferentially interacts with sAnk1.5 but not with sAnk1.6 [16,18]. The studies that investigated Arg64-Phe71 for obscurin-binding were performed using sAnk1.5 and it was never demonstrated that this domain alone was sufficient for obscurin-binding [22–25]. Using sAnk1.6 in co-precipitation experiments, we demonstrate that Arg64-Phe71 manifest minimal obscurin-binding, thus we reassessed these residues in the context of mediating sAnkR self-association. First, we generated conservative mutations (arginine to glutamine) to remove the positive charges of arginines 64, 67, 68 and 69. In co-precipitation experiments, mutating these residues dramatically reduces sAnk1.5 and sAnk1.9 self-association, but does not perturb obscurin co-precipitation. In contrast, this mutation in sAnk1.6 abolishes self-association and obscurin co-precipitation. We maintain that obscurin’s interactions with sAnk1.5 and sAnk1.9 are principally mediated by residues Lys100-Asp111 while residues Arg64-Arg69 predominantly mediate sAnkR self-association.

The FLIM-FRET results corroborate the conclusion that residues Arg64-Arg69 mediate sAnkR self-association. Specifically, mutating these residues increased the CFP lifetimes of sAnk1.5-CFP (1.91 ± 0.08 to 2.11 ± 0.1 ns) and sAnk1.6-CFP (1.78 ± 0.09 to 2.23 ± 0.08 ns), which is consistent with a decrease in sAnkR self-association. While there are alternative interpretations of these results, we contend that [1] sAnkR structural changes are minimized by substituting arginine with glutamine and [2] wild-type and mutant sAnkR-YFP proteins displayed similar expression as shown by immunoblot and confocal microscopy.

Our results demonstrate that obscurin-binding enhances sAnk1.5-GST precipitation of sAnk1.5-HA by 5-fold. There are two possible mechanisms underlying this effect. First, sAnk1.5-GST may precipitate more sAnk1.5-HA because of the two independent ankyrin-binding domains in wild-type obscurin. Previous studies have demonstrated each ankyrin-binding domain in obscurin is sufficient to bind sAnk1.5 [23–25]. Consistent with these studies, our data demonstrates that sAnk1.5-GST precipitates obscurin even when the first or second ankyrin-binding domain is mutated. Interestingly, this effect is significantly abrogated by mutating the first ankyrin-binding domain, but not the second ankyrin-binding domain. These findings suggest that the first domain plays an important role in complex formation between obscurin and sAnk1.5 proteins. The second mechanism is that obscurin co-expression significantly increases sAnk1.5 expression. It has been previously demonstrated that sAnk1.5 expression is regulated by ubiquitination via the adaptor protein potassium channel tetramerization domain containing 6 (KCTD6) and the E3 ligase cullin-3 [28]. Moreover, it was suggested that obscurin-binding may increase sAnk1.5 protein stability by disrupting the ubiquitination of lysine residues associated with sAnk1.5 degradation by the proteasome [28]. Therefore, obscurin-binding would increase the protein stability of sAnk1.5-HA and sAnk1.5-GST resulting in enhanced sAnk1.5-HA precipitation.

Our findings also suggest that a complex of self-associating sAnk1.5 proteins interacts with obscurin. We hypothesized that the population of sAnk1.5 proteins in this complex could be altered by mutating the Arg64-Arg69 self-association domain. To test this hypothesis, we measured obscurin precipitation of two populations of sAnk1.5 (-HA and -YFP) and whether mutating Arg64-Arg69 altered the population precipitated by obscurin. When both populations of sAnk1.5 are wild-type, obscurin precipitates similar levels of sAnk1.5-HA and -YFP. In contrast, mutating Arg64-Arg69 results in obscurin precipitating 46.7 ± 6.6% less mutant sAnk1.5-YFP and a corresponding increase of 36.7 ± 11.9% in wild-type sAnk1.5-HA. Considering obscurin-binding is not altered by mutating Arg64-Arg69 in sAnk1.5, we interpret these results to suggest that obscurin interacts with a large complex of sAnk1.5 proteins that interact with each other in part through this self-association domain. Furthermore, mutating this domain decreases the number of mutant proteins in the complex while allowing for more wild-type proteins to integrate into the sAnk1.5 complex.

This hypothesis is further supported by our FLIM-FRET results demonstrating that obscurin interacts with sAnk1.6 in the presence of sAnk1.5 in cells. While obscurin-CFP and sAnk1.6-YFP alone do not result in FRET, the addition of sAnk1.5-HA significantly reduces obscurin-CFP lifetime due to the energy transfer to sAnk1.6-YFP. Our data supports a model in which sAnk1.5 would form a complex with obscurin-CFP and sAnk1.6-YFP, effectively bringing the fluorophores in close proximity resulting in FRET.

While a previous study created a model of sAnk1.5 binding regions to obscurin (consisting of residues 29–122) based on homology to ankyrin repeats 4–6 of human Notch 1 [22], we decided to pursue an alternative strategy to model sAnk1.5 dimers to identify potential sites of interaction, but not to determine the structure of sAnk1.5 dimers. Using MD simulation of a sAnk1.5 monomer (consisting of residues 28–155) followed by MD simulation of two monomers aligned in parallel, we identified Gly31-Val36 as a proximal site to Arg64-Arg69. Using biochemical and FLIM-FRET analyses, we subsequently demonstrated that sAnk1.5 self-association is attenuated by mutating these residues. Specifically, substituting Gly31-Val36 with less hydrophobic residues reduced sAnk1.5 self-association in co-precipitation assays, but had no effect on its interaction with obscurin. This mutation also significantly reduced energy transfer between sAnk1.5 with itself and with sAnk1.6 as demonstrated by FLIM-FRET.

While the modeling suggests that Gly31-Val36 are in close proximity to Arg64-Arg69, we have not provided evidence demonstrating that these regions are in fact complementary binding sites. First, we anticipate sAnk1.5 will have additional sites of interaction that mediate self-association, thus confounding the interpretation of experiments intended to investigate whether ABD1 and ABD2 are complementary binding sites. Second, we modeled sAnk1.5 monomers lacking transmembrane domains, which would constrain how sAnk1.5 monomers associate.

Another study has predicted that [1] the alignment of hydrophobic residues in and surrounding ABD1 (Leu33, Leu37, Ile40, Leu44 and Leu48) on two adjacent monomers facilitated sAnk1.5 dimer formation and [2] disulfide bond formation between cysteine residues 22 and 34 facilitated oligomer formation [28]. While that study did not investigate dimer nor oligomer formation, the authors concluded that oligomer formation protected sAnk1.5 from ubiquitination and protein degradation based on the finding that mutagenesis of cysteine residues 22 and 34 increased sAnk1.5 protein degradation [28]. In addition to examining ABD1 (Gly31-Val36), we also investigated residues Leu33, Leu37, Lys38, and Ile40 (via alanine substitution) for sAnk1.5 binding. These mutations did not disrupt sAnk1.5 self-association (data not shown). In our study, disrupting possible disulfide bond formation with Cys34 and reducing the hydrophobicity of Leu33, Phe35, and Val36 significantly reduce sAnkR self-association. Taken together, our data suggests that both hydrophobic and electrostatic forces contribute to sAnk1.5 self-association. It should be noted that MD simulation of the two sAnk1.5 monomers predicted hydrogen bonds between Lys47 in one monomer with Gln63 in the other monomer.

While this study has been the first to identify two sites that mediate sAnkR self-association, we anticipate that there will be additional sites of interaction mediating sAnk1.5 self-association. Moreover, sAnk1.5 most likely forms heteromeric complexes with other SR resident proteins such as the sarco/endoplasmic reticulum calcium ATPase (SERCA1) and sarcolipin, a regulator of SERCA. Both the transmembrane and cytoplasmic domains of sAnk1.5 have been shown to interact with SERCA1 resulting in reduced SERCA1 calcium affinity [37]. Likewise, sAnk1.5 has been shown to interact with sarcolipin, which appears to increase sAnk1.5 association with SERCA1 [38]. These interactions provide an explanation for why the mutant sAnk1.5-HA proteins display a similar protein complex on a native gel as wild-type sAnk1.5 protein. Specifically, mutant sAnk1.5 proteins may still associate with sarcolipin or SERCA1 to form a large protein complex. Measuring for SERCA1 and sarcolipin expression in sAnk1.5 protein complexes is a future line of inquiry.

In closing, we propose a model in which the M-line isoform of obscurin interacts with a complex of sAnk1.5 proteins integrated into the sarcoplasmic reticulum to facilitate the elongation and retention of longitudinal SR throughout cardiac and skeletal myocytes. Consistent with this model, genetic disruption of either obscurin or sAnk1.5 results in the preferential loss of longitudinal SR in skeletal muscle [12,20]. In addition, sAnk1.5 knockout mice also display decreased L-type calcium current and altered contractile performance suggesting that loss of sAnk1.5 results in diminished excitation-contraction coupling [20]. Another sAnk1.5-interacting protein that is important for SR structure in muscle fibers is tropomodulin-3, which connects the SR to cytoplasmic γ-actin and localizes to the M-line and regions flanking the Z-line [39]. In skeletal myocytes, tropomodulin-3 compensates for the genetic disruption of tropomodulin-1 resulting in its redistribution to the pointed ends of actin thin filaments, swollen SR vesicles near the Z-lines, and reduced acetylcholine-induced calcium release [39].

While this study provides an important foundation for subsequent studies to uncover the basis of sAnk1.5 oligomer formation and incorporation of other SR proteins (SERCA1 and sarcolipin) into heteromeric protein complexes, many questions remain unanswered. For example, is the formation of sAnk1.5 oligomers regulated? While it has been shown that sAnk1.5 is modified by acetylation, neddylation and ubiquitination, these modifications were evaluated in the context of sAnk1.5 protein degradation [12,28]. Also, what is the function of other small AnkR isoforms (1.6, 1.7, and 1.9) in a large complex of sAnkR proteins. While all isoforms express the self-association domains, only sAnk1.5 and 1.9 have the obscurin-binding domain Lys100-Asp111. We anticipate that future experiments addressing these questions will provide insight into sAnk1.5 regulation of SR membrane protein complexes and the formation of longitudinal SR in skeletal and cardiac myocytes.