EH domain proteins regulate cardiac membrane protein targeting

Abstract

Rationale

Cardiac membrane excitability is tightly regulated by an integrated network of membrane-associated ion channels, transporters, receptors, and signaling molecules. Membrane protein dynamics in health and disease are maintained by a complex ensemble of intracellular targeting, scaffolding, recycling, and degradation pathways. Surprisingly, despite decades of research linking dysfunction in membrane protein trafficking with human cardiovascular disease, essentially nothing is known regarding the molecular identity or function of these intracellular targeting pathways in excitable cardiomyocytes.

Objective

We sought to discover novel pathways for membrane protein targeting in primary cardiomyocytes.

Methods and Results

We report the initial characterization of a large family of membrane trafficking proteins in human heart. We employed a tissue-wide screen for novel ankyrin-associated trafficking proteins and identified four members of a unique Eps15 homology (EH) domain-containing protein family (EHD1, EHD2, EHD3, EHD4) that serve critical roles in endosome-based membrane protein targeting in other cell types. We show that EHD1-4 directly associate with ankyrin, provide the first information on the expression and localization of these molecules in primary cardiomyocytes, and demonstrate that EHD1-4 are co-expressed with ankyrin-B in the myocyte perinuclear region. Notably, the expression of multiple EHD proteins is increased in animal models lacking ankyrin-B, and EHD3-deficient cardiomyocytes display aberrant ankyrin-B localization and selective loss of Na/Ca exchanger expression and function. Finally, we report significant modulation of EHD expression following myocardial infarction, suggesting that these proteins may play a key role in regulating membrane excitability in normal and diseased heart.

Conclusions

Our findings identify and characterize a new class of cardiac trafficking proteins, define the first group of proteins associated with the ankyrin-based targeting network, and identify potential new targets to modulate membrane excitability in disease. Notably, these data provide the first link between EHD proteins and a human disease model.

Introduction

Membrane trafficking, endocytosis, and endocytic recycling of integral membrane proteins are critical cellular functions for normal membrane excitability and cardiomyocyte physiology. Mechanistically, these cellular pathways are controlled by highly organized protein networks that tightly regulate anterograde and retrograde trafficking of membrane proteins, their stability at the cell surface, and their degradation following internalization. While the last few years have witnessed the identification and preliminary characterization of these protein networks in cultured cells, essentially nothing is known regarding these pathways in primary excitable cells, including cardiomyocytes. In fact, the field of molecular cardiology lacks fundamental understanding even of the identity of these critical trafficking proteins in primary cardiac tissue.

Ankyrins are a family of proteins required for membrane targeting and stability of membrane ion channels, transporters, cell adhesion molecules, and signaling proteins.⁴ The importance of ankyrins for normal physiology is illustrated by human diseases linked with ankyrin dysfunction and mice lacking ankyrin gene products. In heart, ankyrins have key roles in the membrane trafficking and regulation of sodium and calcium channels, Na/Ca exchanger, Na/K ATPase, and IP3 receptor.¹⁶ Ankyrin-B dysfunction is associated with a human arrhythmia syndrome affecting multiple excitable cell types resulting in sinus node disease, atrial fibrillation, conduction defects, catecholaminergic polymorphic ventricular arrhythmia, and sudden death.^{13, 18–20} While ankyrins are recognized as critical components for anterograde targeting of ion channels and transporters, little is known regarding the underlying molecular machinery and associated proteins required for ankyrin-based trafficking.

Here, we report the initial characterization of a family of membrane trafficking proteins in human heart. We employed a tissue-wide screen for novel ankyrin-associated trafficking proteins and identified four members of a family of EH (Eps15 homology) domain-containing (EHD) gene products (EHD1-4) that provide critical roles for endosome-based membrane protein targeting and recycling in other cell types. Specifically, EHD proteins regulate endosomal anterograde and retrograde trafficking, as well as membrane protein recycling and lipid homeostasis.^{2, 5, 29} We report that EHD1-4 directly associate with ankyrin and provide the first information on the expression and localization of these molecules in cardiomyocytes. EHD1-4 are expressed with ankyrin-B in the myocyte perinuclear region and EHD expression is increased in ankyrin-B-deficient hearts. Myocytes deficient in EHD3 display striking loss of Na/Ca exchanger membrane trafficking and function, thus establishing the first functional role for cardiac EHD proteins. Finally, we report modulation of EHD expression in a large animal model of arrhythmias, suggesting that these proteins may play a key role in regulating membrane excitability in normal and diseased heart. Together, our findings identify a new class of cardiac membrane trafficking proteins, define the first group of proteins associated with the cardiac ankyrin-based targeting network, and identify potential new targets to modulate membrane excitability in human disease.

Methods

An expanded Methods section is available in the Online Data Supplement at http://circres.ahajournals.org.

Statistics

Values are expressed as mean ± SEM. P values were assessed with a Student’s t test or ANOVA, as appropriate. The Bonferroni test (electrophysiological measurements) or least squares difference was used for post-hoc testing. The null hypothesis was rejected for P <.05. To analyze immunoblot densitometric measurements from multiple experiments on EHD expression across species and heart chambers, immunoblot bands were normalized to actin and then to a control (human or left ventricle), resulting in a mean value of one with no standard error for the control group.

Protein modeling

Modeling of EHD2 was performed using Cn3D4.1 software based on the EHD2 structure of Daumke.⁵

In vitro binding

Protein products were in vitro translated using the TNT Coupled Reticulocyte Lysate System and labeled with [³⁵S]- methionine. Products were incubated with GST fusion proteins of ankyrin-B and ankyrin-G domains (MBD, SBD, CTD) in 500 µl binding buffer (50 mM Tris (pH 7.4), 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 0.1% Triton X-100 and Protease inhibitor overnight at 4 °C. GST-protein complexes were washed five times (50 mM Tris (pH 7.4), 1 mM EDTA, 1 mM EGTA, 500 mM NaCl, 0.1% Triton X-100), and analyzed by SDS-PAGE. To visualize equal loading of GST proteins, gels were stained with Coomassie Blue. [³⁵S]-labelled products that bound GST-ankyrin fusion proteins were detected by phosphorimaging. All binding experiments were replicated >3 times.

Results

Ankyrin-B associates with key membrane trafficking protein

Loss of ankyrin-B results in the development of a proarrhythmic substrate due to the loss of membrane expression of a select group of critical cardiac ion channels and transporters.^18–20 As a first step in defining the key molecular components of the cardiac ankyrin-B-based membrane trafficking pathway, we performed a screen for novel ankyrin-binding proteins. We screened a human heart yeast two-hybrid library with the central region of ankyrin-B as bait for candidate interacting proteins with specific roles in protein trafficking, vesicle fusion, or organelle biogenesis. A screen of >two million independent clones identified one such candidate (Figure 1A). Sequencing of a ~500 base pair clone revealed a nucleotide match with NM_014600.1 that encodes residues 192–349 of the human Eps15 homology domain-containing 3 (EHD3) protein (Figure 1A). Previously unidentified in human cardiac tissue, this protein is a member of the four-member human EHD protein family (EHD1-4) with recently defined roles in membrane protein trafficking in other cell types.^{22, 26} Notably, the ankyrin-binding partial sequence bears strong similarity to the corresponding sequences in EHD1 (86% identical), EHD2 (77% identical) and EHD4 (78% identical) (Figure 1A). Like EHD3, these other EHD products have not been previously studied in heart.

Figure 1. Ankyrin-B associates with EHD3.

A) Yeast two-hybrid screen of human heart library with ankyrin-B identified clone-HHL278 corresponding to the central region of human EHD3 (NM 014600.1). This clone was similar to the corresponding nucleotide sequence of EHD1 (86% identical), EHD2 (78% identical), and EHD4 (77% identical). B-C) EHD3 directly associates with membrane-binding domain of ankyrin-B, but not ankyrin-G. B) Ankyrin-B and ankyrin-G are comprised of three domains including a membrane-binding domain (MBD), spectrin-binding domain (SBD), and C-terminal domain (CTD, death domain and regulatory domain). C) Radiolabeled EHD3 directly associates with the ankyrin-B MBD, but not ankyrin-G. Equal protein quantities were evaluated for each domain. All proteins were evaluated on the same gel by phosphorimaging, but ankyrin-G lanes were moved to the right of panel for clarity. D) Domain organization of EHD proteins. E) Amino acid identity of EHD-containing protein domains compared with EHD3. F–I) Radiolabeled EHD1-4 directly associate with GST-ankyrin-B but not GST. Binding data were replicated in four independent experiments.

EHD3 directly interacts with ankyrin-B

We confirmed the ankyrin-B/EHD3 interaction in vitro using radiolabeled full-length EHD3 and bacterially-expressed and affinity-purified ankyrin-B fused to GST. Full-length EHD3 cDNA was cloned from a human heart cDNA library, and in vitro translated in the presence of [³⁵S]-methionine. Consistent with the two-hybrid screen, we observed interaction of radiolabeled EHD3 with GST-ankyrin-B (Figure 1B–C). Ankyrin-B is comprised of three structural domains including a membrane-binding domain that primarily associates with ion channels and transporters, a spectrin-binding domain that associates with cytoskeletal and signaling proteins, and a C-terminal domain that modulates ankyrin function in vivo (Figure 1B) We observed specific interaction of EHD3 with the ankyrin-B membrane-binding domain (Figure 1C). In contrast, GST-ankyrin-B spectrin-binding domain, GST-ankyrin-B C-terminal domain, and GST alone lacked EHD3 binding (Figure 1C). Moreover, despite ~75% amino acid identity with the ankyrin-B membrane-binding domain, the ankyrin-G (product of ANK3 gene) membrane-binding domain lacked EHD3 binding activity (Figure 1C). We also observed no interaction of EHD3 with the ankyrin-G spectrin-binding domain or C-terminal domain (Figure 1C). Moreover, EHD3 did not associate with a third ankyrin-family gene product, ankyrin-R (not shown). Therefore, EHD3 directly associates with ankyrin-B in vitro. This interaction is specific to ankyrin-B versus other ankyrin-gene products and requires the ankyrin-B membrane-binding domain.

EH domain-containing proteins associate with ankyrin-B

While yeast and in vitro binding studies (Figure 1A–C) demonstrated that ankyrin- B can associate with EHD3, human EHD3 protein paralogs display high amino acid identity across all domains (Figure 1D–E). While EHD3 is most similar to EHD1 (87% identity), EHD2 and EHD4 sequences also closely resemble EHD3 (71% and 75% amino acid identity, respectively). We therefore tested whether ankyrin-B associates with multiple EHD proteins. Consistent with the high homology of EHD proteins, particularly within the EHD central region, we observed association of ankyrin-B with all EHD proteins (Figure 1F–I). Specifically, purified GST-ankyrin-B membrane-binding domain interacts with radiolabeled full-length EHD1, EHD2, EHD3, and EHD4 (Figure 1F–I). In contrast, radiolabeled EHD proteins (EHD1-4) displayed no binding activity for GST alone. Similar to findings with EHD3 (Figure 1C), we observed no interaction of ankyrin-G with EHD1, 2, or 4.

EHD3 interaction with ankyrin requires coiled-coil domain

EHD proteins display a conserved primary structure consisting of a short N-terminal region (NT), a dynamin-like domain (DLD) that includes the nucleotide-binding “P-loop” known to bind ATP (together NTD and DLD often referred to as ‘G-domain’), a ‘coiled-coil’ region, and a C-terminal EH domain that binds to NPF motifs in target proteins (Figure 2A–C). The EH domain consists of two EF hands of which the second is capable of binding to Ca²⁺; this is thought to be important for the folding of the EH domain. In addition to associating with Ca²⁺, lipid, and membrane trafficking proteins, EHD proteins form homo- and hetero-oligomers via their coiled-coil region. Nucleotide binding at the P-loop also regulates membrane-binding and oligomerization (Figure 2A–C). Therefore, we investigated the structural requirements on EHD3 for ankyrin-binding activity. We generated a library of mutants corresponding to each EHD3 domain/region (Figure 2D). Additionally, we created mutants corresponding to all combinations of EHD3 domains (Figure 2D). Each EHD3 mutant was radiolabeled and incubated with purified GST-ankyrin-B or GST. We observed consistent interaction with ankyrin-B for all EHD3 mutants that contained the coiled-coil region (Figure 2D–E). In contrast, the EHD3 N-terminal region, dynamin-like domain, and EH domain lacked ankyrin-binding activity. We observed no binding of any EHD3 mutant with GST (Figure 2E). Thus, we have defined the structural requirements on both ankyrin-B and EHD3 for interaction. Specifically, the ankyrin-B membrane-binding domain is critical for EHD3-binding activity, whereas the coiled-coil region mediates the EHD3 interaction with ankyrin-B. It is noteworthy that the association of the EHD coiled-coil region with ankyrin-B is the first assigned non-EHD protein binding partner for this domain.

Figure 2. EHD3 coiled-coil domain is required for ankyrin-binding.

A) EHD proteins are comprised of an N-terminal region (NT), dynamin-like domain (DLD), a coiled-coil region (CC), and an EH domain (EHD). B–C) Depiction of ribbon and space filling model of an EHD3 homolog⁵ (EHD2) to illustrate potentially exposed sites on the polypeptide. Yellow denotes AMP-PNP molecule in structure. D) Library of protein mutants generated for mapping the ankyrin-binding region on EHD3. E) Purified GST-ankyrin-B membrane-binding domain (MB) associates with radiolabeled EHD3 mutants containing the central helical domain. We observed background binding for all other mutants. Input (In) and control binding to GST alone (Cn) are also shown. All binding data were successfully replicated in four independent experiments. Note that the translated DLD/CC/EHD product displays a second, slightly smaller (migrates ~3 kD lower) degradation product that also associates with ankyrin-B.

EHD family proteins are expressed in vertebrate heart

Four genes encode EHD proteins in vertebrates. However, to date there is no detailed information on the cellular expression, localization, or function of any of these gene products in human heart. We first performed RT-PCR from normal human left ventricle to establish the presence of EHD1-4 message in human heart. All EHD mRNA products were observed in left ventricle and sequencing revealed matches with full-length EHD1-4 cDNAs previously identified from non-cardiac sources (Figure 3A). Consistent with mRNA findings, we observed EHD1-4 proteins by immunoblot using isoform-specific antibodies on human left ventricle (Figure 3B). Parallel lysates prepared from mouse, rat, and canine left ventricle demonstrated expression of EHD1-4 (Figure 3B–F). Finally, immunoblots of normal human heart chambers revealed EHD1-4 expression in all cardiac chambers (Figure 3G–K). Together, these findings demonstrate that the all members of the EHD protein family (EHD1-4) are expressed in heart across a range of vertebrates.

Figure 3. EHD proteins are expressed in human heart.

A) RT-PCR of mRNA prepared from human left ventricle demonstrates expression of full-length EHD1-4. Identical results were obtained from three independent experiments. B–F) Immunoblots and densitometric measurements of EHD1, EHD2, EHD3, and EHD4 from total protein lysates of mouse, rat, canine, and human left ventricle. Band densities were normalized to actin and expressed relative to human heart for comparison across experiments. Data in C–F represent n=4 normal independent samples. Asterisk denotes p<0.05 compared to human heart, whereas the pound symbol represents p<0.05 compared with dog. G–K) EHD proteins are expressed throughout the heart. Immunoblots and summary data from analyses of total protein lysates of normal human heart chambers for EHD1-4. Actin was utilized as a loading control and expression is normalized to left ventricle. Data in H–K represent n=3 normal independent samples (p=N.S. for all groups).

Finally, we tested the ability of each EHD-containing protein to interact with ankyrin-B in cardiac tissue lysates. In agreement with in vitro binding assays, ankyrin-B associates with all EHD-containing polypeptides in heart. Specifically, co-immunoprecipitation studies using affinity-purified ankyrin-B Ig and detergent-soluble lysates from adult mouse heart demonstrate interaction of ankyrin-B with EHD1-4 (Online Figure I). This interaction was specific to ankyrin-B, as no EHD interaction was observed with control Ig (Online Figure I), or with ankyrin-G or ankyrin-R (not shown). Collectively, our data provide the first evidence for the cellular expression of the EHD protein family across vertebrates.

EHD proteins are localized to cardiomyocyte peri-nuclear region

To date, the cellular distribution of any EHD-containing protein in any cardiac cell type is unknown. We defined the subcellular distribution of EHD1-4 in heart using primary cardiomyocytes and EHD isoform-specific antibodies.⁸ While not fully differentiated, neonatal myocytes were chosen for these studies, as these cells are thin (~2 microns in X–Z dimension compared with ~20 microns in adult myocytes), which allows optimal imaging of the organelles involved in the membrane targeting pathway. In neonatal cardiomyocytes, ankyrin-B is localized to cytoskeleton-enriched domains underlying the myocyte membrane (Figure 4A–D, left panel). In addition, ankyrin-B is localized to the perinuclear region of cardiomyocytes (Figure 4A–D). EHD1-4-positive staining was observed in clusters near the perinuclear region (Figure 4). This is similar to the localization of EHD proteins in HeLa cells where they reside in endosomes and regulate trafficking through this endocytic compartment.^{26, 33} Together, these findings establish the subcellular distribution of EHD proteins in isolated primary ventricular cardiomyocytes.

Figure 4. EHD proteins are expressed in the myocyte perinuclear region.

Mouse neonatal cardiomyocytes were immunostained for ankyrin-B (red) and EHD1-4 (green). Cells were co-labeled with topro-3AM to stain nuclei (blue). Scale bar=ten microns. Experiments were repeated four times for each EHD protein from four different wild-type mice (>100 individual myocytes/experiment were analyzed for each EHD).

Loss of ankyrin-B affects EHD1-4 expression and localization in heart

To examine the functional relationship between ankyrin-B and EHD proteins, we evaluated the expression of EHD polypeptides in hearts of adult wild-type and ankyrin-B^+/− mice (ankyrin-B^−/− mice die immediately after birth). Interestingly, EHD1-4 levels were not reduced in ankyrin-B^+/− hearts. Instead, we observed significant increases in all EHD protein levels (Figure 5A–E, n=7/genotype; p<0.05). These findings strongly support a relationship between ankyrin-B and EHD proteins in heart. We further examined the relationship of ankyrin-B and EHD1-4 at the level of the single cell by examining the distribution of all EHD proteins in isolated primary cardiomyocytes lacking ankyrin-B. In agreement with immunoblot data, expression of all EHD–containing proteins (EHD1-4) was increased in the complete absence of ankyrin-B (Figure 5F–M). EHD1 expression was increased in the perinuclear region of ankyrin-B^−/− neonatal cardiomyocytes (Figure 5F–G). Increased peripheral cytoplasmic staining was observed for both EHD2 and EHD3 in ankyrin-B^−/− cardiomyocytes (Figure 5H–K). Finally, EHD4 showed a strong increase preferentially in the perinuclear area of ankyrin-B^−/− cells (Figure 5L–M).

Figure 5. EHD protein expression is increased by loss of ankyrin-B.

A) Representative immunoblots and B–E) densitometric measurements of EHD1-4 from lysates prepared from adult wild-type and ankyrin-B^+/− mouse left ventricle. Equal protein quantities were loaded for each genotype. Band densities were normalized to actin and expressed relative to WT (n=7 experiments/genotype, *p<0.05). F–M) Ankyrin-B^−/− cardiomyocytes display increased expression of EHD proteins. Localization of ankyrin-B and EHD proteins in wild-type (F, H, J, L) and ankyrin-B^−/− (G, I, K, M) primary cardiomyocytes. All cells were stained and imaged using similar protocols. Scale bar=10 microns. Cells are co-labeled with topro-3AM to stain nuclei. Experiments were repeated from four mice/genotype and >100 myocytes/experiment were analyzed.

EHD3 over-expression selectively increases myocyte I_NCX

Ankyrin-B is critical for targeting select cardiac ion channels and transporters. Myocytes lacking ankyrin-B expression display decreased membrane expression and function of Na/Ca exchanger 1 (NCX1), while the localization and function of the L-type Ca²⁺ channel, Ca_v1.2, is unaffected.^{3, 16, 17, 19} We therefore assessed the ability of EHD-overexpression to enhance ankyrin-B-associated protein membrane expression and function. As expected, whole-cell patch clamp recordings revealed a decrease in membrane Na/Ca exchanger current (I_NCX) in ankyrin-B^−/− cardiomyocytes, consistent with the notable perinuclear localization in ankyrin-B^−/− myocytes (Figure 6A–C). Specifically, we observed differences in whole cell I_NCX at nearly all voltages in ankyrin-B^{− /−} cardiomyocytes compared with wild-type myocytes (n=12 cells/genotype, p<0.05). Overexpression of human GFP-EHD3 increased I_NCX in wild-type cardiomyocytes at select voltages compared with untransfected myocytes (Figure 6D–I, n=12 cells/condition, p<0.05). Overexpression of GFP-EHD3 had no significant effect on I_NCX in ankyrin-B^−/− myocytes (n=12 cells/condition, N.S.). We observed no difference in Na/Ca exchanger (or ankyrin-B) localization in wild-type or ankyrin-B^−/− cardiomyocytes in the presence or absence of EHD3 overexpression. Finally, EHD3 overexpression had no affect on an ankyrin-B-independent myocyte current, I_Ca,L (Figure 6J–K, n=12 cells/genotype/treatment, N.S.). Together, these data support the requirement of ankyrin-B for anterograde delivery of select myocyte membrane proteins to the cell surface. Moreover, these data demonstrate that EHD3-overexpression enhances this cellular pathway in wild-type cardiomyocytes.

Figure 6. EHD3 overexpression alters I_NCX in cardiomyocytes.

A–B) Ankyrin-B (green) and Na/Ca exchanger (red) expression in wild-type and ankyrin-B^−/− neonatal myocytes. Nuclei are stained in blue. C, Na/Ca exchanger current (I_NCX) is significantly reduced in cardiomyocytes derived from ankyrin-B^−/− mice (p<0.05, n=12 myocytes/genotype). D–F) Control myocytes (D), and myocytes from wild-type (E) and ankyrin-B^−/− mice (F) expressing GFP-EHD3. Note untransfected cells lack signal on the green channel. G–I) EHD3-GFP overexpression increases INCX in wild-type (n=12 cells/treatment, *p<0.05 compared to WT untransfected), but not ankyrin-B^−/− myocytes (n=12 cells/treatment, p = N.S. compared to ankyrin-B^−/− untransfected). J–K) GFP-EHD3 overexpression does not affect myocyte I_Ca,L (n=9 myocytes/experiment, p = N.S. compared to untransfected). Scale bars=10 microns.

EHD3 is required for Na/Ca exchanger membrane expression and function

We further probed the role of EHD3 for myocyte function by reducing EHD3 expression using EHD3-specific siRNA. Notably, myocytes lacking EHD3 expression (see Online Figure II) displayed striking reduction in targeting and function of I_NCX (Figure 7A,C–D,F–G, n=11 cells/treatment for I_NCX experiments, p<0.05). Moreover, EHD3-deficient cardiomyocytes displayed defects in the localization (Figure 7F–G, note decreased expression and prominent large cytosolic puncta) and expression (see Online Figure II) of ankyrin-B. This loss was specific to the ankyrin-B-associated cellular trafficking pathway as we observed no difference in membrane expression and function of the voltage-gated calcium channel, Ca_v1.2 (Figure 7B, E, n=9 cells/treatment, N.S.). Moreover, we observed no differences in expression or function of ankyrin-B, Na/Ca exchanger 1, or Ca_v1.2 in the presence of a control siRNA (Figure 7A–E, n=9 cells/treatment for I_Ca,L, 11 cells/treatment for I_NCX, N.S.). Together, these data provide the first functional evidence of a role for EHD3 in the membrane trafficking of a critical cardiac transporter. Moreover, the notable selectivity of the targeting defect (Na/Ca exchanger versus Ca_v1.2), as well as the direct affect of EHD3 knock-down on ankyrin-B expression strongly support a role for EHD3 in the cardiac ankyrin-B membrane protein trafficking pathway.

Figure 7. EHD3 knockdown alters ankyrin-B distribution and NCX1 targeting and function.

A, C-D) I_NCX (pA/pF) in wild-type myocytes following EHD3 siRNA treatment (also untreated, control siRNA). Data represent 11 myocytes/treatment, *p<0.05 compared to control or untreated. B,E) EHD3 siRNA treatment does not affect I_Ca,L. Data represent 9 myocytes/treatment, p=N.S. for all groups F–G) Expression and distribution of EHD3, ankyrin-B, and Na/Ca exchanger in control and EHD3 siRNA-treated cardiomyocytes. Images are representative of >100 examined cells; no defects in myocyte proteins were observed in the presence of the control siRNA. Note abnormal distribution, decreased expression, and large ankyrin-B-positive puncta in EHD3 siRNA treated myocytes. Bars=10 microns.

EHD proteins are differentially regulated in cardiovascular disease

Aberrant ankyrin-B regulation is observed in rare forms of congenital arrhythmia. ¹⁸ Moreover, common variants in the ankyrin-B gene (ANK2) modulate cardiac function in the general human population.³⁰ Finally, ankyrin-B expression is significantly reduced in a large animal model of arrhythmias following myocardial infarction (MI).¹⁰ In contrast, there are no data on the role of EHD1-4 in human arrhythmia or in animal models of human disease. As a first step in identifying a potential role of EHD proteins in disease, we examined the expression of EHD1-4 in a well-validated large-animal (canine) model of arrhythmias following MI.^{10, 15} At 48 hours post-coronary artery occlusion, we observed an increase in EHD3 expression in a specific heart region where arrhythmias originate (infarct border zone) compared to control tissue (remote from infarct; remote EHD3 levels are no different from remote regions of non-infarcted hearts at 48 hrs or 5 days after occlusion: 1.19±0.05 and 1.11±0.10, respectively, p=N.S. compared to control, n=3) (Figure 8A–B). By five days post-occlusion (when action potential and ion channel changes are most prominent²⁸), we observed significant increases in both EHD3 and EHD4 levels in the infarct border zone (Figure 8A–B). EHD1 and EHD2 levels were not significantly changed at 48 hours or five days post-occlusion (Figure 8A–B). Notably, ankyrin-B levels are reduced in this post-MI arrhythmia model (representative immunoblots shown in Figure 8).¹⁰

Figure 8. EHD3 and EHD4 expression levels are elevated following myocardial infarction.

EHD1-4 densitometric measurements (left) and representative immunoblots (right) from canine border zone and remote tissues 48 hours (A) or five days (B) post coronary artery occlusion (n=4 for remote and n=3 for BZ, *p<0.05 compared to remote).

Discussion

Our findings identify the EHD family of endosome-associated proteins as novel components of the ankyrin-B-based targeting pathway in heart. Specifically, ankyrin-B directly associates with EHD1-4 via a conserved sequence in the coiled-coil domain. EHD proteins are expressed in human heart, and are co-expressed with ankyrin-B in the perinuclear region of cardiomyocytes. Our findings define the first group of ankyrin-B-associated proteins that are not negatively affected by ankyrin-B loss. In fact, loss of ankyrin-B results in increased EHD1-4 expression. Myocytes lacking EHD3 expression phenocopy ankyrin-B^−/− myocytes, displaying both decreased expression and function of Na/Ca exchanger, but maintaining the expression and activity of Ca_v1.2. Finally, we demonstrate dramatic alteration of the EHD proteins following MI, suggesting that these proteins may play an important role in action potential and ion channel changes in heart disease. Our findings provide the initial functional characterization of a new class of membrane trafficking proteins in heart, and provide the first reported link between EHD proteins and a human disease model. We believe these findings open a new avenue to understand the molecular determinants of cardiac membrane excitability.

EHD proteins appear to regulate endosomal anterograde and retrograde trafficking, as well as membrane protein recycling and lipid homeostasis.^{2, 5, 12} Structural analysis has revealed a similarity of EHD2 with the dynamin superfamily. This similarity may translate into parallel functional roles of EHD-containing proteins with dynamin, including the ability to tubulate liposomes, dimerize, oligomerize around lipid tubules in ring-like structures, and to stimulate nucleotide hydrolysis in response to lipid-binding. ⁵ Among EHD proteins, EHD1 is the best characterized with roles in membrane recycling of the transferrin receptor, major histocompatibility class 1 receptor, and beta-1 integrin.¹² Additionally, EHD1 associates directly with phosphatidylinositol compounds^{11, 24} and modulates cholesterol homeostasis.²⁵ While molecular mechanisms underlying EHD1 function are still being identified, EHD1-binding partners including MICAL-L1³³, ARF6, and Rab effector proteins argue that EHD1 plays a central role in endosome/Golgi vesicular trafficking.^{2, 5, 21, 26} Similarly, EHD2 regulates glucose-stimulated GLUT4 membrane transport and endocytosis in adipocytes.⁹ Potentially relevant for cardiac muscle, EHD2 interactions with myoferlin are important for myoblast fusion.⁶ Additionally, EHD1 has been demonstrated to interact with amphiphysin-1²⁷, a member of a protein family implicated in biogenesis of transverse-tubules in skeletal muscle.¹⁴ Finally, both EHD3 and EHD4 have roles in endosome vesicular targeting and recycling.^{23, 34} Specifically, EHD3 directly associates with critical trafficking proteins including Rab11-FIP2.²⁶ Finally, two reports describe testicular phenotypes in mouse models of EHD1 and EHD4 deficiency. Deletion of EHD1 and EHD4^{7, 31} result in reduction in testes size; however, deletion of EHD1 leads to male infertility while deletion of EHD4 does not. In summary, while much is unknown about the roles of these recently identified proteins, the clear linkage of EHD proteins with protein trafficking, endocytosis, and endocytosis, suggest that these proteins may have critical roles in controlling trafficking of membrane proteins that maintain cardiac excitability.

Loss of ankyrin-B in this study resulted in increased EHD family protein expression. Based on these findings, we hypothesize that that the EHD/ankyrin in vivo interaction occurs either upstream or concurrently with the interaction of ankyrin with membrane-associated proteins. Thus, in response to loss of ankyrin-B-associated proteins at the membrane, the myocyte may attempt to augment anterograde membrane trafficking by up-regulating the EHD/ankyrin pathway. EHD protein expression levels may also be regulated by the local signaling environment.^{1, 32} Thus, abnormal ion homeostasis created by the loss of ankyrin-B^{13, 19} may trigger a parallel ankyrin-independent EHD-containing protein pathway to accelerate anterograde membrane protein trafficking. Interestingly, EHD3 and EHD4 expression was increased in the canine model of myocardial infarction that also shows parallel loss of ankyrin-B (Figure 8).¹⁰ Clearly, additional studies in adult cardiomyocytes and in vivo will be necessary to carefully evaluate the functional relationship between EHD proteins and ankyrin-B.

Finally, based on the demonstrated roles of EHD proteins in other tissues^{8, 29}, we predict that loss of EHD-containing proteins in heart will affect on cardiac membrane protein regulation and excitability. In support of this hypothesis and in agreement with our findings in neonatal cardiomyocytes, we observed striking perinuclear staining for all EHD proteins (EHD1-4) in adult cardiomyocytes (Online Figure 3). Our findings that EHD proteins are upregulated in animal models of cardiovascular disease suggest that this pathway may be a nodal point for the regulation of electrical remodeling in the damaged myocyte, due to aberrant Na⁺ or Ca⁺² homeostasis or myocyte membrane damage/ischemia. However, based on previous findings in other tissues and high sequence identity, together with our results that ankyrin-B interacts with all EHD proteins, we predict that cardiac EHD-containing proteins may display functional redundancy in the constitutive absence of a single family member in the context of an animal. For example, while EHD4-deficient mice display defects in germ cell development and testis size, changes in the levels of EHD1-3 likely prevents EHD4^−/− mice from displaying more overt phenotypes.^{7, 31} Interestingly, a proportion of EHD1-deficient mice are neonatal lethal (Rainey et al, accepted) due to unknown causes. Future work will help unravel if this is due to defects in cardiac excitability. Thus, while siRNA experiments may be effective in assessing phenotypes in isolated cells acutely (Figure 7), elucidating the physiological relevance of EHD proteins in vivo may in fact require knock-out of multiple EHD protein isoforms.

Novelty and Significance

What is known?

• Cardiac excitability requires cellular pathways that tightly regulate integral membrane protein trafficking, endocytosis, and endocytic recycling.

• There is a fundamental lack of understanding of the cellular pathways and molecular mechanisms that control membrane protein trafficking and recycling in heart.

What New Information Does This Article Contribute?

• In the present study it was identified that the family of Eps15 homology (EH) domaincontaining proteins (EHD1-4) is critical for membrane protein trafficking in primary cardiomyocytes.

• EHD proteins interact with the previously identified ankyrin-dependent pathway for membrane protein targeting.

• EHD protein levels are altered in small animal models of ankyrin-B deficiency and in a large animal model of human cardiovascular disease.

Cardiac membrane excitability requires tightly regulated intracellular pathways that control anterograde and retrograde trafficking of membrane proteins, their stability at the cell surface, and their degradation. This study identifies four members of the Eps15 homology domain-containing protein family in cardiomyocytes that serve critical roles in endosome-based membrane protein targeting in other cell types. Cardiac EHD proteins were found to associate with the ankyrin-based pathway for membrane protein targeting in myocytes, and decreased EHD3 expression altered the membrane targeting and electrophysiological properties of select cardiac ion channels and transporters. Finally, altered EHD expression was identified in a large animal model of arrhythmias following myocardial infarction, suggesting that EHD proteins may serve as key nodal points for regulating membrane excitability in normal and diseased hearts. Collectively, these findings characterize a new class of cardiac membrane targeting proteins, identify key components of the ankyrin-based cellular pathway for membrane protein trafficking, and define potential new targets to regulate membrane excitability in disease.