Eps15 Homology Domain-containing Protein 3 Regulates Cardiac T-type Ca2+ Channel Targeting and Function in the Atria*

Background: Endosome-based protein trafficking in the heart is poorly understood. Results: Functional targeting and expression of the T-type Ca2+ channel (TTCC) in the atria requires the endosomal protein, EHD3. Conclusion: Impaired endosomal transport leads to cardiac rhythm and conduction disorders. Significance: Understanding endosome-based protein trafficking in the heart may provide new therapeutic targets for cardiovascular disease. Proper trafficking of membrane-bound ion channels and transporters is requisite for normal cardiac function. Endosome-based protein trafficking of membrane-bound ion channels and transporters in the heart is poorly understood, particularly in vivo. In fact, for select cardiac cell types such as atrial myocytes, virtually nothing is known regarding endosomal transport. We previously linked the C-terminal Eps15 homology domain-containing protein 3 (EHD3) with endosome-based protein trafficking in ventricular cardiomyocytes. Here we sought to define the roles and membrane protein targets for EHD3 in atria. We identify the voltage-gated T-type Ca2+ channels (CaV3.1, CaV3.2) as substrates for EHD3-dependent trafficking in atria. Mice selectively lacking EHD3 in heart display reduced expression and targeting of both Cav3.1 and CaV3.2 in the atria. Furthermore, functional experiments identify a significant loss of T-type-mediated Ca2+ current in EHD3-deficient atrial myocytes. Moreover, EHD3 associates with both CaV3.1 and CaV3.2 in co-immunoprecipitation experiments. T-type Ca2+ channel function is critical for proper electrical conduction through the atria. Consistent with these roles, EHD3-deficient mice demonstrate heart rate variability, sinus pause, and atrioventricular conduction block. In summary, our findings identify CaV3.1 and CaV3.2 as substrates for EHD3-dependent protein trafficking in heart, provide in vivo data on endosome-based trafficking pathways in atria, and implicate EHD3 as a key player in the regulation of atrial myocyte excitability and cardiac conduction.

vesicular transport systems, endosomes facilitate neuronal plasticity, internalization of membrane proteins targeted for recycling or degradation, nutrient endocytosis, anterograde trafficking of newly synthesized proteins from the Golgi apparatus to the plasma membrane, as well as retrograde protein trafficking from the membrane (1)(2)(3)(4)(5). Because of the development of new in vivo models to study endosome-based protein trafficking, we are now beginning to understand the molecular underpinnings of endosome-based protein trafficking in the context of the intact physiological environment. Furthermore, we have now identified that defects in endosome-based processes may result in disease (4,6,7).
Cardiac membrane excitability depends on the proper organization and expression of specific ion channels, pumps, exchangers, and transporters within the plasma membrane. The localization of these membrane-bound proteins requires transport through endosomal protein trafficking pathways. C-terminal Eps15 homology domain-containing (EHD) 2 proteins are endocytic regulatory proteins. Four EHD gene products (EHD1-4) are differentially expressed across all tissue types (8). Years of research in surrogate cells systems and model organisms has established the general role of EHD proteins as regulators of endosomal trafficking. EHDs are known to mediate protein trafficking from the Golgi to the plasma membrane along with internalization, recycling, and degradation pathways (9 -13). EHDs primarily mediate clathrin-independent pathways. However, EHD1 mediates the endosome-based recycling of select membrane proteins initially internalized by clathrindependent pathways (14 -16). Disruption of their function results in impaired intracellular protein trafficking and devas-tating consequences for the cell. These proteins are highly conserved across species, emphasizing their fundamental role in cell biology. Indeed, the human and mouse isoforms of EHD1 share 99.6% sequence homology (17).
Despite their putative roles, remarkably little is known about what specific membrane protein substrates require EHD-based endosomal systems for their proper trafficking and function, particularly in complex vertebrate cells including cardiomyocytes. Recent data indicates that EHD proteins are critical mediators of protein trafficking in the ventricle (4,6,7). For example, we identified EHD3 as a key regulator of anterograde trafficking of the Na ϩ /Ca 2ϩ exchanger (NCX), and the voltagegated L-type Ca 2ϩ channel (LTCC) in the ventricle (4,7). Each of these proteins plays an important role in maintaining cardiac membrane excitability and facilitating excitation-contraction coupling in the heart (18 -20). However, the role of EHD3 beyond the ventricle is unknown and unstudied.
Here, using an unbiased approach, we identify the cardiac voltage-gated T-type Ca 2ϩ channels (TTCC), Ca V 3.1 and Ca V 3.2, as novel substrates for EHD3-dependent trafficking. Ca V 3.1 and Ca V 3.2 are critical mediators of action potential initiation and conduction from the sinoatrial node through the atria and into the atrioventricular node (AVN) in heart (21,22). Using a cardiac-selective model of EHD3 deficiency, we demonstrate defects in Ca V 3.1 and Ca V 3.2 expression, localization, and function in EHD3 Ϫ/Ϫ atrial myocytes. Furthermore, EHD3 co-immunoprecipitates with Ca V 3.1 and Ca V 3.2. In line with known roles for Ca V 3.1 and Ca V 3.2 in cardiac conduction system, loss of EHD3 in heart results severe cardiac rhythm and conduction defects. These new data are the first to implicate EHD3-dependent endosomal trafficking as a key mediator of cell excitability and electrical conduction in the atria.

Generation of Cardiac-selective Mouse Model of EHD3
Deficiency-To test the in vivo cardiac intrinsic roles of EHD3, we utilized a conditional null mutant allele where the 5Ј UTR and exon 1 of the mouse EHD3 gene (Ehd3) was flanked by LoxP sites (Ehd3 f/f ) and therefore deleted in the presence of Cre recombinase (23) (Fig. 1A). We selectively eliminated EHD3 in cardiomyocytes by crossing these animals with mice expressing ␣MHC-Cre (24); homozygous conditional knock-out mice are referred to as EHD3 cKO throughout the report. Mice were born at expected Mendelian ratios, were healthy and fertile with body weights comparable with WT littermates, and displayed lack of EHD3 by PCR and immunoblot. Mice were genotyped by PCR. Three primers in a single duplex PCR amplified the WT allele (377 bp) and mutant allele (488 bp). Primers were as follows: 5Ј CAA CAA GAG TGT CAG GAA ACC TGA ACT A-3Ј; 5Ј-CTG GGA AAC TGC AGA ACA TCA GGG AAC A-3Ј; 5Ј-ATG AGG GAC TCA AGG GGC AAG TCC TGG A-3Ј (Fig. 1B). Mice utilized in this study were between 8 and 12 weeks of age.
ECG Experiments-ECG recordings of ambulatory mice were obtained using implanted radiotelemeters (DSI, St. Paul, MN) (25). Recordings were obtained from mice that were conscious, post-exercise, post-isoprenaline injection, and post-exercise plus isoprenaline injection. For baseline heart rate anal-ysis, continuous ECG data were collected from WT and EHD3 cKO mice for 1 h on seven separate days. For stress tests, mice were run on a treadmill for a maximum of 45 min or until exhaustion, then immediately injected with isoprenaline (0.5 mg/kg), or injected with isoprenaline in the absence of exercise. Non-sustained and sustained arrhythmias were identified using standard guidelines (26).
Immunoblots and Immunostaining-Immunoblots of atrial lysates were performed as described (29,30). Briefly, atria were harvested from the hearts of adult, age-matched littermates, and immediately placed into ice-cold homogenization buffer (in mM: 50 Tris-HCl, 10 NaCl, 320 sucrose, 5 EDTA, 2.5 EGTA; supplemented with 1:1000 protease inhibitor mixture and 1:1000 PMSF (Sigma)). Following quantification, tissue lysates were analyzed on Mini-PROTEAN tetra cell (Bio-Rad) on a 4 -15% precast TGX gel (Bio-Rad). Gels were transferred to a nitrocellulose membrane using the Mini-PROTEAN tetra cell (Bio-Rad). Membranes were blocked for 1 h at room temperature using a 3% BSA solution or 5% milk solution and incubated with primary antibody overnight at 4°C. Densitometry analysis was done using Image Lab software (Bio-Rad). For all experiments, protein values were normalized against an internal loading control (GAPDH).
Electrophysiology-Membrane currents were assessed by use of an Axopatch-200B amplifier and a CV-203BU head stage (Axon Instruments). Experimental control, data acquisition, and data analysis were accomplished with the use of software package PClamp 10 with the Digidata 1440A acquisition system (Axon Instruments). Patch pipettes were pulled from thinwalled glass capillary tubes (Sutter Instruments). The electrode resistance ranged from 2 to 4 M⍀. L-and T-type calcium currents were recorded at room temperature (20 -22°C) with pipette resistances Ͻ3.0 M⍀. The external solution contained the following (in mmol/liter): TEA-Cl (137); CsCl (5.4); MgCl 2 (1); CaCl (1); 4-AP (4); HEPES (10); glucose (11) adjusted to pH 7.4 with CsOH. The internal solution contained the following (in mmol/liter): CsCl (120); TEA-Cl: (20) MgCl-6H 2 O; Mg-ATP (5.2); HEPES (10); EGTA (10) adjusted to pH 7.2 with CsOH. For L-type calcium current measurement, myocytes were held at Ϫ50 mV followed by 300-ms depolarizing steps to voltages ranging from Ϫ40 to ϩ50 mV in 5-mV increments. For total calcium current, myocytes were clamped at Ϫ90 mV and subjected to the same step protocol as above. T-type current was determined by subtracting the former from the latter.
Microarray-Total RNA was isolated from the hearts of 3-4month-old WT and EHD3 Ϫ/Ϫ mice using the Qiagen RNeasy Mini Kit per the manufacturer's instructions (Qiagen Inc., Valencia, CA). RNA was purified and cRNA was generated as previously described (32). Relative expression levels of cRNA were assessed using Affymetrix GeneChip Mouse Gene 2.0 ST Array (Affymetrix, Santa Clara, CA). Signal intensities were analyzed by Affymetrix Expression Console software. Background correction and quantile normalization was performed to adjust technical bias, and gene expression levels were summarized by robust multiarray method (33). A filtering method based on percentage of arrays above noise cutoff was applied to FIGURE 1. Cacna1h transcript levels are reduced in EHD3 ؊/؊ heart. A, representative heat map of microarray data generated by analyzing relative transcript levels (ϳ25,000 mRNA targets) in WT versus EHD3 Ϫ/Ϫ hearts (n ϭ 3 each). B, select genes of interest with known roles in endosomal transport or excitation contraction coupling that were significantly different in expression between WT and EHD3 Ϫ/Ϫ hearts. C, quantitative PCR results of Cacna1h levels between WT and EHD3 Ϫ/Ϫ hearts. Both actin and GAPDH are included as controls (right).
filter out low expression genes. Linear model was employed to detect differentially expressed genes. To improve the estimates of variability and statistical tests for differential expression, a variance smoothing method with fully moderated t-statistic was employed for this study (34). The significance level was adjusted by controlling the mean number of false positives (35).
Real-time PCR-Total RNA was isolated from the hearts of 3-4-month-old WT and EHD3 Ϫ/Ϫ mice using the Qiagen RNeasy Mini Kit per the manufacturer instructions (Qiagen Inc., Valencia, CA). RNA was subsequently treated with TURBO DNase (Invitrogen). First strand cDNA was synthesized from 200 ng of RNA as a template using the SuperScript III First Strand Synthesis System (Invitrogen). Real-time PCR was conducted on fast optical 96-well reaction plates using the StepOne Plus real-time PCR system (Invitrogen) in a 10-l final volume, using 0.5 l of TaqMan primer (FAM-labeled), 2 l of PCR-grade H 2 O, and 5.0 l of TaqMan Master Mix per well. 2.5 l of a 1:5 cDNA dilution was added to each well. The oligonucleotide primer for Cacna1h (Mm00445382_m1) was from Applied Biosystems. Gene expression was calculated relative to GAPDH (Mm99999915_g1) and ␤-actin (Mm00607939_s1). Data are reported as fold-change (2 (Ϫ⌬⌬CT) ). C T values are normalized to control samples.
Immunofluorescence-Isolated atrial myocytes from WT and EHD3 cKO hearts were fixed and permeabilized in 100% etha-nol at Ϫ20°C. Cells were blocked in 2% gelatin from cold water fish skin (Sigma, G7041) and 0.075 Triton X-100 in PBS. Cells were stained with primary antibodies overnight at 4°C in blocking solution. Cells were washed three times in ice-cold blocking solution. After washes were complete, secondary antibodies were applied in blocking solution for 2 h at room temperature. Secondary antibodies included Alexa-conjugated donkey antimouse 488 and 568, and donkey anti-rabbit 488 and 568. Cells were imaged on an LSM 780 confocal microscope (Carl Zeiss). Myocytes were images using identical confocal settings between genotypes. At least 20 myocytes were examined for each staining protocol.
Co-immunoprecipitation-Protein A-conjugated agarose beads (AffiGel; Bio-Rad) were incubated with either control mouse Ig or anti-EHD3 Ig in co-immunoprecipitation binding buffer (PBS with 0.1% Triton X-100 and protease inhibitor mixture (Sigma)) for 12 h at 4°C. Beads were centrifuged and washed three times in ice-cold PBS. WT atrial tissue lysates were added to the washed beads, along with protease inhibitor mixture and co-immunoprecipitation binding buffer, and incubated for 12 h at 4°C. The reactions were washed three times in ice-cold co-immunoprecipitation buffer. The samples were eluted and the proteins were separated by SDS-PAGE prior to immunoblots with antibodies specific for Ca V 3.1, Ca V 3.2, or

Unbiased Screen Reveals Reduced Cacna1h in EHD3 Null
Heart-EHD-based protein pathways are poorly defined in heart. As an initial and unbiased first step to define EHD3-dependent pathways in the mammalian heart, we identified differentially expressed mRNAs between WT and EHD3 Ϫ/Ϫ mice (mice globally lacking EHD3). Approximately 25,000 unique transcripts were evaluated. Surprisingly, fewer than 300 transcripts showed significant differences in mRNA expression between WT and EHD3 Ϫ/Ϫ hearts (Fig. 1). In fact, only a small subset of transcripts met the strict criteria of having both a 1.5-fold change in expression and a p value of Ͻ0.05 as assessed by analysis of variance. These data suggest that the EHD3-de-pendent pathways have a relatively selective impact on the regulation of the cardiac transcriptional program.
From this transcript profile, we focused on gene candidates with direct roles in either cardiac excitation-contraction coupling or endosomal protein trafficking, including Slc8a1 (Na/Ca exchanger, NCX), Sln (sarcolipin), or Strn (striatin) (19, 36 -38). These genes (33 in all) are summarized in Table 1 and Fig. 1B. A representative heat map reflecting this data are show in Fig. 1A. Notably, the gene encoding one isoform of the cardiac TTCC. Specifically, Cacna1h mRNA was reduced by 67% in the EHD3 Ϫ/Ϫ hearts (Fig. 1B, p Ͻ 0.01). Although the depolarization phase of the atrial action potential is primarily mediated by the voltage-gated sodium channel, Na V 1.5, the T-type calcium current may also facilitate this process (21,39). Furthermore, T-type calcium current is dysregulated in cardiac disease and linked with the maladaptive hypertrophic response associated with heart failure (40 -42). Based on the severity of the expression change and role in vertebrate physiology, we focused on our studies on the putative link between EHD3 and TTCC in heart.
As a first step to identify the potential link between EHD3 and TTCC, we performed quantitative PCR measurements of mRNA transcripts levels in WT and EHD3 Ϫ/Ϫ heart, probing specifically for Cacna1h to validate our microarray analyses (Fig. 1C). Using both actin and GAPDH probes as controls, Cacna1h mRNA expression was decreased by Ͼ50% in EHD3 Ϫ/Ϫ hearts compared with WT (p Ͻ 0.05; Fig. 1C).
EHD3 cKO Atria Display Reduced Ca V 3.1 and Ca V 3.2 Protein Expression-Cardiac activity may be influenced by noncardiac extrinsic factors (i.e. sympathetic and parasympathetic inputs). To avoid confounding effects on the heart due to loss of EHD3 in non-cardiac tissues we generated mice selectively lacking EHD3 in cardiac myocytes (see "Experimental Procedures," EHD3 cKO). Consistent with the design of the targeting construct (Fig. 2, A and B), myocytes from EHD3 cKO hearts displayed nearly complete elimination of EHD3 protein expression (Fig. 1C). As this animal line represents a more selective model to study myocyte-specific effects of EHD3 on cardiac function, the remaining experiments in this study utilized the EHD3 cKO mouse.
Although not enriched in ventricle, the vertebrate atria expresses two TTCC gene products Ca V 3.1 (Cacna1g) and Ca V 3.2 (Cacna1h) (21). The dominant isoform in the adult atria is Ca V 3.1, whereas Ca V 3.2 is linked with embryonic and postnatal cardiac development (43). Therefore, we analyzed the protein expression of both Ca V 3.1 and Ca V 3.2 in WT and EHD3 cKO atria. Notably, expression of Ca V 3.1 and Ca V 3.2 is reduced by ϳ40 and 60% in atria isolated from EHD3 cKO mice, respectively (Fig. 3, A and B, p Ͻ 0.01). As expected, Ca V 3.1 and Ca V 3.2 levels were minimal in ventricle compared with mouse atria. Notably, expression of Cacna1g (Cav3.1) was not significantly reduced in microarray experiments. This suggests that the loss of this major TTCC subtype from EHD3 cKO atria is likely due to a post-translational mechanism.

EHD3 Is Required for Ca V 3.1 and Ca V 3.2 Targeting in Atrial
Myocytes-Based on our immunoblot data and known roles for EHDs, we tested if the membrane targeting of Ca V 3.1 and Ca V 3.2 was disrupted in EHD3 cKO atrial myocytes (Fig. 4). In WT myocytes, both Ca V 3.1 and Ca V 3.2 were expressed both on the plasma membrane and over the atrial Z-line (mouse atrial myocytes have a less well developed T-tubule network, but maintain well organized Z-line structure) (44). In line with previously published data, ankyrin-B expression and localization were disrupted in the EHD3-deficient atrial myocytes (4). In contrast, whereas markers of the Z-line are clearly present in EHD3 cKO atrial myocytes, we observed loss of both Ca V 3.1 and Ca V 3.2 localization. In fact, both proteins show diffuse punctate staining in EHD3 cKO myocytes, consistent with reduced Ca V 3.1 and Ca V 3.2 expression observed by immunoblot (Fig. 3). In summary, whereas these immunofluorescence data do not explicitly assign a role for EHD3 in the insertion of these ion channels into the surface membrane, the results clearly a role for EHD3 in the targeting of Ca V 3.1 and Ca V 3.2 in atrial myocytes.
EHD3 Co-immunoprecipitates with Cav3.1 and Cav3.2-In non-cardiac tissue, EHD proteins are linked with a host of membrane ion channels, transporters, and receptors (4,(45)(46)(47)(48). However, loss of Ca V 3.1 and Ca V 3.2 expression and targeting in EHD3 cKO atria may represent an indirect response to myocyte remodeling in the face of altered cell trafficking pathways. Therefore, as a first step to establish the relationship between EHD3 and Ca V 3.1/Ca V 3.2 we performed coimmunoprecipitation experiments from detergent-soluble lysates from mouse atria, immobilizing EHD3 Ig on beads. Notably, even in high salt conditions, we observed co-immunoprecipitation of both Ca V 3.1 and Ca V 3.2 with EHD3 (Fig. 5, A  and B). In contrast, Na V 1.5 did not co-immunoprecipitate with EHD3 supporting that Na V 1.5 trafficking is likely independent of EHD3 (Fig. 5E). These data support that EHD3 is localized within the same in vivo protein complexes with Ca V 3.1 and Ca V 3.2.
EHD3 Is Required for Atrial T-type Ca 2ϩ Current-To test the functional relevance of impaired EHD3-dependent targeting of Ca V 3.1 and Ca V 3.2, we assessed I Ca , T in atrial myocytes isolated from WT and EHD3 cKO hearts (Fig. 6). I Ca,T was significantly decreased across all physiologically relevant voltages (Fig. 6, A and B), and peak I Ca,T was reduced by ϳ50% (Fig.  6C) in EHD3 cKO atrial myocytes compared with WT (p Ͻ 0.05). Notably, we also observed loss of L-type calcium channel (I Ca,L ; LTCC) in EHD3 cKO atrial myocytes, consistent with our prior report linking EHD3 with Ca v 1.2 targeting in ventricular myocytes (4) (Fig. 6D). Notably, loss of EHD3 was selective for I Ca versus other atrial myocyte currents, as we observed no difference in I Na properties in EHD3 cKO atrial myocytes compared with WT (Fig. 6, E and F). The data are consistent with results from co-immunoprecipitation experiments showing inability of EHD3 to associate with Nav1.5 (Fig. 5E). Atrial myocyte size was comparable between WT and EHD3 cKO myocytes as measured by cell capacitance (56.9 Ϯ 21.3 versus 57.2 Ϯ 17.8 pF, respectively, p Ͼ 0.99). Together, our data demonstrate that EHD3 is requisite for normal atrial I Ca,T and I Ca,L .

EHD3 cKO Mice Display Defects in Atrial Function-Both
TTCC and LTCC play critical roles in atrial myocyte automaticity, atrial conduction, and atrioventricular conduction (49). Therefore, we hypothesized that disruption of both TTCC and LTCC function in EHD3 cKO mice would result in defects in atrial physiology. We implanted WT and EHD3 cKO mice with radiotelemeters and assessed cardiac ECGs in conscious mice at both baseline and following acute stress (injection of ␤-adrenergic agonist, isoprenaline). Consistent with our hypothesis, compared with age-matched WT littermates, EHD3 cKO mice displayed severe rhythm and conduction abnormalities (Fig. 7). For example, at baseline, EHD3 cKO mice displayed significant heart rate variability compared with WT littermates (Fig. 7, A,  B, and F). In resting EHD3 cKO mice, we regularly observed heart rate fluctuations from 400 to 650 bpm within the span of 10 s. This variation was regularly sustained for more than 5 min with interspersed durations of normal sinus rhythm. This phenotype was not observed in WT mice (100 versus 0%, EHD3 cKO versus WT; p Ͻ 0.01). In response to sympathetic stimulation (0.2 mg/kg of Iso), EHD3 cKO mice displayed a high frequency of atrioventricular block, with multiple episodes of block observed in all mice. Unlike EHD3 cKO mice, atrioventricular block was only observed in one WT mouse after Iso treatment, and this occurred in only one instance in this mouse (100 versus 17%, p Ͻ 0.05, Fig. 7F). EHD3 cKO conduction phenotypes are clearly shown in Fig. 7, C and D, by the presence of non-propagating P waves. These traces indicate that whereas the heart displayed normal sinus rhythm, the cardiac action potential did not properly conduct into the ventricle (lack of QRS complex). Last, EHD3 cKO mice showed a high incidence of sinus pause following adrenergic stimulation (Fig. 7E). This phenotype was only regularly observed in EHD3 cKO mice (but observed only once in one WT mouse after Iso). In summary, at baseline and following sympathetic stimulation, EHD3 cKO mice display aberrant atrial phenotypes.

DISCUSSION
The in vivo roles of endosome-based pathways for the targeting/transport of membrane proteins in heart are not well stud-  ied. Here, we define a key role for EHD3 in atrial function, as EHD deficiency in mice results in defects in cardiac automaticity and conduction. At the cellular level, we identify that EHD3 is critical for the selective expression, targeting, and function of both T-type and L-type calcium channels. Observed phenotypes in EHD3 cKO mice are consistent with atrial phenotypes found in animals deficient in either LTCC or TTCC (50 -52). In summary, these findings identify the TTCC as a novel target for EHD3-dependent endosomal trafficking and demonstrate that loss of a single endosome-based protein is sufficient to disrupt normal cardiac function.
At baseline, the atrial action potential is primarily initiated by the I Na , with both the LTCC and TTCC facilitating membrane depolarization (53). Furthermore, LTCC and TTCC play unique roles related to Ca 2ϩ homeostasis and contractility (54). Intriguingly, data from the Molkentin lab suggest that Ca V 3.1mediated Ca 2ϩ entry may have anti-hypertrophic effects (55). During heart failure, TTCC expression and function is increased and plays a more prominent role in the atria (40 -42, 56). For these reasons, targeting of the TTCC as a potential therapeutic against heart failure has recently gained interest, but its effectiveness remains unclear (41,57,58). EHD3 expression is significantly increased during heart failure (6). Our current study implicates EHD3 in the regulation of Ca V 3.1 and Ca V 3.2 in the atria. It is intriguing, then, to speculate that the increased expression and function of the TTCC during heart failure relies on EHD3-dependent mechanisms. Furthermore, the TTCC is well established to facilitate both sinoatrial and atrioventricular pacemaker activity (59,60). We previously demonstrated EHD3 to be expressed within the sinoatrial node (4). Thus, the observation of rhythm and conduction disorders in EHD3 cKO mice was consistent with this prior work. It is important to note that whereas whole animal atrial phenotypes are consistent with loss of TTCC and LTCC function (50,52), we cannot rule out secondary molecular defects in EHD3 cKO hearts that may contribute to observed phenotypes. Further-FIGURE 7. EHD3 deficiency results in severe conduction and rhythm abnormalities. A and B, representative ECG data from conscious WT (A) and EHD3 cKO (B) mice, note the high heart rate variability in the EHD3 cKO mouse. C, representative ECG data from WT mouse post-Iso injection. D, representative ECG data from EHD3 cKO mouse post-Iso injection demonstrating repeated incidences of atrioventricular conduction block (indicated by arrowheads). E, representative ECG data of an EHD3 cKO mouse post-Iso injection demonstrating repeated incidences of sinoatrial node pause. Black arrows indicate R to R interval present during normal sinus rhythm. F, statistical analysis of observed incidence of heart rate variability, atrioventricular, and sinoatrial node pause block shown as relative percentage (p Ͻ 0.01). more, beyond membrane protein function and electrical activity, EHDs have been linked with the biogenesis/organization of key membrane domains in skeletal muscle (61). As atrial dysfunction is linked to both dysfunction in electrical function as well as defects in atrial substrate, future experiments will be important to define any potential new roles of EHD3 in atrial function, both at baseline and in disease. Finally, our microarray experiments were utilized as an unbiased method to predict new pathways regulated by EHD3 in heart. These data noted a decrease in the levels of Cacna1h that encodes the minor component of T-type calcium current in the atria, Ca v 3.2. In contrast, we did not observe a significant difference in Cacna1g that encodes Ca v 3.1, the dominant component of atrial T-type current in the adult mouse. Our subsequent findings in EHD3 cKO adult atria revealed reduced Ca v 3.1 protein expression, aberrant Ca v 3.1 immunostaining, and reduced T-type current (despite no difference in Cacna1g mRNA levels). We therefore surmise that observed Ca v 3.1 defects in EHD3 cKO atria are most likely related to a post-translational trafficking defect. The roles of EHD3 in the transcriptional regulation of Cacna1h are not currently known but an important area of future investigation.
A surprising, yet consistent observation of this work (as well as past studies) is the apparent selectivity of EHD-based endosome pathways. For example, the unbiased mRNA screen identified fewer than 300 of ϳ25,000 transcripts differentially expressed between WT and EHD3 Ϫ/Ϫ hearts ( Fig. 1; Table 1). Furthermore, although also found at the plasma membrane of atrial and ventricular myocytes, we observed no difference in Na V 1.5 in EHD3 cKO hearts and Na V 1.5 did not co-immunoprecipitate with EHD3 (Figs. 5 and 6) (4). These data suggest that EHD3-dependent protein trafficking in the heart is independent of "bulk" protein trafficking mechanisms, and more likely have evolved to mediate the transport of a specific subset of membrane bound proteins. The selectivity of EHD3-dependent transport mechanisms may be advantageous in developing new approaches to therapies that target the endosome. It is of particular interest that, thus far, EHD3-mediated protein trafficking in the heart is responsible for the proper trafficking, targeting, and expression of several proteins that are intimately involved in excitation-contraction coupling. Currently our knowledge of EHD3-dependent mechanism indicates that the NCX, LTCC, and TTCC all require EHD3 for their proper function. Notably, these proteins are differentially regulated during electrical remodeling. In particular, the increased function and expression NCX and TTCC in heart failure are considered to be substrates for arrhythmia (62). We hypothesize that by targeting EHD3-dependent trafficking (e.g. decreasing EHD3-mediated endosomal transport), a new approach to anti-arrhythmic therapy may be developed. This approach may offer a way to specifically tune ion channel trafficking to tune cardiac excitability in disease. Indeed, this potential has sparked the active development of EHD inhibitors (63,64). We have previously demonstrated that in vitro EHD3 overexpression is sufficient to tune cell excitability (7), suggesting that targeting endosomal transport may be a feasible approach. However, further characterization of the EHD-dependent trafficking system in the heart is needed.
Although it will be important to define additional membrane protein substrates of the EHD3 pathway, it will also be critical to define the molecular machinery of EHD-linked endosomes. There are four EHD gene products expressed in the heart. Given the number of proteins that are trafficked through endosomes, it is unlikely that these four proteins alone are sufficient to confer selectivity for cardiac membrane protein trafficking. We predict that the Rab family of GTPases may play important roles in EHD-dependent pathways in heart. In fact, our microarray data indicate that several Rab GTPases or Rab effector proteins are altered by the loss of EHD3 (Table 1). Rab GTPases and Rab effectors are tightly linked in all components of the endosomal trafficking system, and have long been known to coordinate vesicular trafficking (65,66). Furthermore, both Rab and Rab effectors are known binding partners of EHD proteins (46,(67)(68)(69). Currently, there are Ͼ60 Rab GTPases with an equal number of Rab effectors. The combinatorics of these proteins is believed to confer target specificity of endosomal transport. An important future goal will be to decode how Rab/ Rab effector interactions with EHDs may dictate substrate specificity. In fact, the use of in vivo models for EHD tissue-and cell-specific deletion may offer important insight into defining the endosome machinery in heart, as well as how defects in these pathways may cause defects in cardiac physiology.