The Role of MAPRE2 and Microtubules in Maintaining Normal Ventricular Conduction

Abstract

Background:

Brugada syndrome (BrS) is associated with loss-of-function SCN5A variants, yet these account for only ~20% of cases. A recent genome-wide association study identified a novel locus within MAPRE2, which encodes microtubule end-binding protein 2, implicating microtubule involvement in BrS.

Methods:

A mapre2 knockout zebrafish model was generated using CRISPR/Cas9 and validated by Western blot. Larval hearts at 5 days post-fertilization were isolated for voltage-mapping and immunocytochemistry. Adult fish hearts were used for ECG, patch clamping, and immunocytochemistry. Morpholinos were injected into embryos at one-cell stage for knock-down experiments. A transgenic zebrafish line with cdh2 tandem fluorescent-timer was used to study adherens junctions. Microtubule plus-end tracking and patch clamping were performed in human iPSC-derived cardiomyocytes (CM) with MAPRE2 knock-down and knock-out, respectively.

Results:

Voltage-mapping of mapre2 knockout hearts showed a decrease in ventricular action potential upstroke velocity (V_max) and conduction velocity, suggesting loss of cardiac Na_V function. ECG showed QRS prolongation in adult knockout fish and patch clamping showed decreased sodium current density in knockout ventricular myocytes and arrhythmias in knockout iPSC-CM. Confocal imaging showed disorganized adherens junctions and mislocalization of mature N-cadherin with mapre2 loss-of-function, associated with a decrease of detyrosinated tubulin. MAPRE2 knock-down in iPSC-CM led to an increase in microtubule growth velocity and distance, indicating changes in microtubule dynamics. Finally, knock-down of ttl encoding tubulin tyrosine ligase in mapre2 knockout larvae rescued tubulin detyrosination and ventricular V_max.

Conclusions:

Genetic ablation of mapre2 led to a decrease in Na_V function, a hallmark of BrS, associated with disruption of adherens junctions, decrease of detyrosinated tubulin as a marker of microtubule stability, and changes in microtubule dynamics. Restoration of the detyrosinated tubulin fraction with ttl knock-down led to rescue of Na_V-related functional parameters in mapre2 knockout hearts. Taken together, our study implicates microtubule dynamics in the modulation of ventricular conduction.

Summary:

Brugada syndrome is a devastating inherited cardiac arrhythmia syndrome associated with loss-of-function variants in the SCN5A gene encoding the main cardiac sodium channel Na_V1.5, yet these account for only ~20% of cases. A recent GWAS identified a novel locus within the MAPRE2 gene, which encodes microtubule end-binding protein 2, implicating microtubule involvement in BrS. Using multiple animal and cellular models, we show for the first time the mechanistic link between MAPRE2 loss-of-function, changes in microtubule dynamics, and decrease in cardiac Na_V function, a hallmark of BrS. Furthermore, we demonstrate rescue of the Na_V dysfunction by manipulating the tyrosination status of α-tubulin, which may be a novel therapeutic avenue for sodium channelopathies.

Graphical Abstract

INTRODUCTION

Inherited cardiac arrhythmias are a significant cause of sudden cardiac death in young individuals (<35 years of age) and remain a major clinical challenge due to high genetic and allelic heterogeneity, reduced penetrance and the lack of effective predictors or preventive treatments.¹ Genetic variants in SCN5A, the gene encoding the main cardiac sodium channel (Na_V1.5), have been directly implicated as a major cause of several forms of inherited arrhythmia,^2,3 yet mechanistic understanding remains incomplete. One prominent example of sodium channelopathy is Brugada syndrome (BrS), a condition affecting 5 per 10,000 people in Western countries and 12 per 10,000 in Southeast Asia, with 80–90% being males.^4,5 Among >20 reported genes, only SCN5A has been unequivocally implicated in BrS to date,⁶ though loss-of-function variants in the gene are found in only ~20% of probands.¹ Together with the low penetrance and variable expressivity observed for rare SCN5A variants, the high percentage of negative genetic testing (~80%) suggests that BrS inheritance is complex with multiple contributory genetic factors and/or environmental modifiers.^1,4

Recently, Bezzina and colleagues conducted a large genome-wide association study (GWAS) using 2820 BrS patients which not only confirmed prior findings⁷ but also uncovered 9 additional loci.⁸ One of these new loci is situated within intron 2 of MAPRE2, which encodes microtubule end-binding protein 2 (EB2). In mammals, EB2 is member of a family of three proteins (EB1, EB2, and EB3), all of which bind to the growing plus-ends of microtubules via their N-terminal domain and interact with other partners via their C-terminus. The three EBs share a high degree of homology with the notable exception of a 43 amino acid segment at the N-terminus of EB2, not present in the other two family members and of unknown function.⁹ Previously, EB1 has been shown to play a pivotal role in the subcellular localization of Na⁺ and K⁺ channels at the initial axonal segment and Ranvier nodes in neurons.^10–12 EB1 has also been shown to regulate Na⁺ channel density in cardiomyocytes¹³ and the targeted delivery of the gap junction protein connexin-43 (Cx43) to intercellular junctions.^14,15 By contrast, nothing is known about the role of EB2 in cardiomyocytes, although it appears to have distinct functions from EB1 and EB3 in vitro.¹⁶ Interestingly, human mutations in MAPRE2 have been implicated in congenital symmetric circumferential skin creases type 2, although the genetic and molecular mechanisms are not well defined, and the cardiac phenotype has not yet been characterized.^17,18

To understand the role of EB2 in cardiac arrhythmia, we employed the zebrafish model which exhibits cardiac electrophysiology similar to that of humans.¹⁹ Previously, we knocked out mapre2 acutely in F0 zebrafish using a multi-gRNA CRISPR/Cas9 approach²⁰ and observed a ventricular phenotype consistent with SCN5A loss-of-function associated with BrS.⁸ Because that particular model was limited by somatic mosaicism, we have now generated a germline mapre2 knockout (KO) in order to perform detailed mechanistic studies. Here we demonstrate that mapre2 loss-of-function leads to ventricular conduction slowing and Na_V dysfunction, associated with disruption of adherens junctions. Mechanistically, we show that mapre2 loss-of-function leads to a decrease in tubulin detyrosination and changes in microtubule dynamics. Based on this we knocked down ttl encoding tubulin tyrosine ligase and were able to rescue the effects of mapre2 loss-of-function on detyrosinated tubulin, adherens junction, and Na_V-related functional parameters. Together, our study directly implicates microtubule dynamics in the modulation of normal cardiomyocyte electrophysiology and ventricular conduction, opening the door for novel therapeutic approaches.

METHODS

All materials and methods are described in the Supplemental Materials and available from the corresponding author upon reasonable request.

RESULTS

Generation of mapre2 loss-of-function models

Like 70% of human genes which have at least one zebrafish homolog,²¹ MAPRE2 is homologous to the zebrafish mapre2 with 87.4% similarity at the protein level (EMBOSS matcher²²). Based on this we designed a gRNA to target exon 2 of zebrafish mapre2 and generated a germline KO mutant using CRISPR/Cas9 (Figure S1A-B). Although there is no significant change in the mRNA level of the edited mapre2 (Figure S1C), the protein product EB2 was not detected on Western blot by antibody targeting its C-terminus (Figure S1D). Quantitative real-time PCR also showed no significant changes in the mRNA levels of other mapre family members except splice variant 1 of mapre1a, which is a minor homolog of MAPRE1 (Figure S1C). Finally, to confirm mapre2 loss-of-function in an orthologous model, we injected wild-type (WT) zebrafish embryos with morpholinos (MOs) targeting the same splice acceptor site in exon 2 of mapre2 and recapitulated the loss of EB2 expression by Western blot (Figure S1E).

mapre2 loss-of-function leads to slowed ventricular conduction and decreased Na_V current

To understand the impact of mapre2 loss-of-function on cardiac electrophysiology, we performed high-resolution optical voltage mapping on hearts isolated from both germline KO and MO-injected lines at 5 days post-fertilization (dpf). Compared to WT siblings, homozygous null larvae had a nonsignificant decrease in ventricular conduction velocity (WT: 16.3±1.1, HET: 15.8±1.3; HOMO: 13.4±0.8 mm/s; P=0.32 WT vs. HOMO; Figure 1A), a significant decrease in maximum action potential upstroke velocity (V_max; WT: 86.0±2.7, HET: 77.4±2.2, HOMO: 74.0±3.3 1/s; P<0.05 WT vs. HOMO; Figure 1B), and a nonsignificant increase in action potential duration (APD; WT: 242±8, HET: 247±9, HOMO: 264±16 ms; P=0.39 WT vs. HOMO; Figure 1C). Notably for ventricular V_max, the heterozygous (HET) KO larvae had an intermediate phenotype between WT and HOMO KO larvae (P<0.05 WT vs. HET). In the mapre2 MO-injected model, the phenotypes all trended in the same direction as the germline KO but were more severe, which is common for morphants due to the genetic compensation induced by deleterious mutations.²³ Compared to larvae injected with control (CTL) morpholinos, mapre2 MO-injected larvae exhibited a significant decrease in ventricular conduction velocity (18.2±1.6 vs. 9.3±1.0 mm/s; P<0.0001; Figure 1D) and V_max (81.7±1.7 vs. 62.4±2.3 1/s; P<0.0001; Figure 1E), and a significant increase in APD (222±6 vs. 256±9 ms; P<0.01; Figure 1F).

Figure 1. mapre2 loss-of-function leads to decreased ventricular conduction and Na_V function.

A-C. Voltage mapping of hearts isolated from WT vs. heterozygous (HET) and homozygous (HOMO) mapre2 KO zebrafish. In HOMO hearts, there was a nonsignificant decrease in ventricular conduction velocity (CV; A), a significant decrease in ventricular V_max (maximum action potential upstroke velocity, dV/dt; P=0.0121 vs. WT; B), and a nonsignificant increase in ventricular action potential duration (APD; C). There was also a significant decrease in ventricular V_max in HET hearts compared to WT (P=0.0363). Multiple comparisons were done using Dunnett’s test if one-way ANOVA was significant (P=0.0147 for V_max in B). D-F. Voltage mapping of hearts isolated from control (CTL) vs. mapre2 morpholino (MO) injected larvae. In MO hearts, there was a significant decrease in ventricular CV (unpaired t-test: P<0.0001; D) and V_max (unpaired t-test: P<0.0001; E), and a significant increase in ventricular APD (Mann-Whitney test: P=0.0021; F). All hearts (represented by dots) were isolated from 5 dpf zebrafish larvae. The dotted squares in A and D reflect the main ventricular area in the hearts from which the parameters were measured. APD was measured at 80% repolarization while the hearts were paced at 100 bpm.

To further support that these observations reflect a loss in cardiac Na_V function, we repeated the same measurements in WT larvae injected with translational blocking MOs targeting scn12aa and scn12ab (homologs of human SCN5A), and compared these with control MO. Similar to mapre2 loss-of-function models, larvae injected with scn12aa and scn12ab MOs showed a significant and more profound decrease in ventricular conduction velocity (17.1±3.0 vs. 6.5±1.0 mm/s, P<0.001; Figure S2A) and V_max (73.6±3.6 vs. 49.7±3.7 1/s; P<0.01; Figure S2B), but with unchanged APD (243±4 vs. 235±23 ms; P=N.S.; Figure S2C). Together, these findings confirm that mapre2 loss-of-function phenocopies Na_V loss-of-function in larval hearts.

To better define the role of mapre2 in cardiac electrophysiology, we raised the mapre2 KO line to adulthood and performed surface ECGs in anesthetized fish. In a standard heterozygote incross, the different genotypes were observed in the expected Mendelian ratios with no increased mortality or gross abnormalities observed in the heterozygous and homozygous KO siblings. As shown in Figure 2A, the zebrafish ECG is similar to human ECG albeit with inverted T waves and allows for accurate measurement of key ECG parameters. Compared to WT siblings, we found a significant increase in QRS duration in both the heterozygous and homozygous KO fish (WT: 26.8±0.7, HET: 29.3±0.8; HOMO: 32.1±0.7 ms; P<0.05 WT vs. HET, P<0.0001 WT vs. HOMO; Figure 2F). Notably, the heterozygotes had an intermediate QRS phenotype, similar to that observed with V_max from isolated larval hearts (Figure 1B), suggesting a gene-dose dependent effect. These findings demonstrate that mapre2 loss-of-function leads to conduction slowing in the adult ventricle.

Figure 2. mapre2 loss-of-function leads to conduction slowing in adult fish.

Two-lead surface ECG was performed in anesthetized fish from the mapre2 KO line. A. Representative averaged ECG tracings demonstrating similarity to human ECG with the notable exception of the inverted T wave. B-G. In homozygous mapre2 KO, there is a nonsignificant increase in P wave duration and a significant increase in QRS duration, suggesting ventricular conduction slowing, with heterozygotes showing an intermediate phenotype (one-way ANOVA P<0.0001 followed by Tukey’s multiple comparisons test: P=0.0435 WT vs. HET; P=0.0314 HET vs. HOMO; P<0.0001 WT vs. HOMO; F). Each dot represents one fish.

To directly measure the Na⁺ current (I_Na), we performed patch clamp experiments on enzymatically isolated ventricular myocytes and found a reduced I_Na density in the mapre2 KO cells (Figure 3A-B). For example, at −20 mV, the I_Na density was −64.5±9.6 and −35.9±4.8 in WT and KO myocytes, respectively, indicating a 44% reduction in current density. Neither the speed of current inactivation (Figure 3C) nor the voltage dependency of (in)activation (Figure 3D and F) were significantly different between WT and KO myocytes, indicating that the gating properties were not affected by mapre2 loss-of-function. Taken together, these experiments demonstrate that mapre2 loss-of-function leads to slowed ventricular conduction and decreased Na_V current.

Figure 3. mapre2 loss-of-function leads to decreased Na⁺ current.

A. Voltage clamp protocol used to measure I_Na (top panel) and typical I_Na traces (bottom) of a freshly isolated WT and KO myocyte. As detailed in the Supplement, we collected ventricles from 4–5 adult fish which were pooled for each cell isolation. The ‘n’ in this figure indicates the number of cells measured from three isolations. B. Average current voltage (I-V) relationships (left panel) and dot plots of I_Na density at −20 mV (right panel) in WT and KO myocytes showing a decrease of I_Na density in KO myocytes. I-V relationships were compared with the two-way repeated measures ANOVA (P=0.036) followed by pairwise comparison using the Student–Newman–Keuls test. The I_Na densities at −20 mV were compared using the Student’s t-test. The numbers indicate the P values lower than 0.05. C. The time course of current inactivation at −20 mV was fitted by a double-exponential equation: I/I_max=A_f×exp(−t/τ_f)+A_s×exp(−t/τ_s), where A_f and A_s are the fractions of the fast and slow inactivation components, and τf and τs are the time constants of the fast and slow inactivating components, respectively. Data did not differ significantly (Student’s t-test). Neither the time constants nor the relative amplitudes were different between WT and KO myocytes. D. Voltage-dependency of activation (left panels) and V_1/2 and k of the Boltzmann fits of every cell measured (right panels). Solid lines are Boltzmann fits to the average data. Data did not differ significantly (Student’s t-test). E. Voltage clamp protocol used to measure voltage dependency of I_Na inactivation (top panel) and typical I_Na inactivating traces (bottom panels). F. Voltage-dependency of inactivation and V_1/2 and k of the Boltzmann fits of all cells (right panels). Solid lines are Boltzmann fits to the average data. Data did not differ significantly (Student’s t-test).

Previously, we have shown that MAPRE2 KO in human iPSC-derived cardiomyocytes led to decreased V_max and I_Na,⁸ similar to what we report now in the germline mapre2 KO fish. To further evaluate the iPSC model, we performed current clamp to observe spontaneous action potentials in individual iPSC-derived cardiomyocytes. Whereas all 21 WT cells exhibited normal trains of action potentials, 17 of 40 KO cells had either early afterdepolarization (EAD; 6 cells), delayed afterdepolarization (DAD; 11 cells), and/or irregular rhythm (6 cells; P<0.001; Figure S3). These data suggest that, in addition to decreasing I_Na, MAPRE2 loss-of-function may be arrhythmogenic.

mapre2 loss-of-function does not result in gross structural remodeling or changes in ion channel expression

To determine if the observed electrophysiologic abnormalities were associated with structural remodeling, we measured the ventricular dimensions of hearts isolated from the mapre2 fish models. Figure S4A-C shows that there was no significant difference in the ventricular length, width and area between the WT and HOMO larvae in the mapre2 germline KO model. Similarly, there was no significant difference in these parameters in the mapre2 MO knockdown model (Figure S4D-F). In addition, we performed high-speed in vivo video-microscopy²⁴ in the mapre2 KO model and found no significant difference in ventricular contractility parameters between the WT and homozygous embryos (Figure S5). Finally, there was also no significant change in the ventricular dimensions between the WT and homozygous mapre2 KO adult fish (Figure S6).

To evaluate the extent of cellular transcriptional remodeling we measured ion channel gene expression, using real-time PCR experiments in the germline KO model. These experiments showed no significant difference in the mRNA level of all major ion channel genes (Figure S7), including notably scn12aa and scn12ab which are the direct homologs of SCN5A. This was further corroborated in adult ventricles where there was no significant difference in the expression of the main cardiac Na_V channel encoded by scn12ab on Western blot (Figure S8).

mapre2 loss-of-function leads to adherens junction disruption

Based on these findings and a previous study implicating microtubules in the regulation of adherens junctions,²⁵ we hypothesized that mapre2 loss-of-function disrupts cell-cell adhesion via dysregulation of microtubule networks, which would in turn disrupt Na_V localization and function. To test this, we first examined the integrity of adherens junctions by immunostaining for N-cadherin (Ncad). As shown in Figure 4A, we found a general disorganization of ventricular Ncad in mapre2 KO hearts compared to WT siblings.

Figure 4. mapre2 loss-of-function leads to disruption of adherens junctions.

A. Representative immunostaining of hearts from WT versus homozygous mapre2 KO larvae shows a general disorganization of ventricular N-cadherin (Ncad). Representative of 2 WT vs. 2 KO hearts. B. Immunocytochemistry of hearts from WT versus homozygous mapre2 KO larvae on the transgenic background with cdh2 tandem fluorescent timer (tFT). Immunostaining of zonula occludens-1 (ZO-1) was used to mark cell borders. Signal from GFP which takes 5 min to fold marks nascent Ncad whereas signal from RFP which takes 100 min to fold marks stable Ncad. Quantification of GFP and RFP signals using ZO-1 signal as a mask shows no significant change in nascent Ncad localization (C) but decreased stable Ncad localization at cell borders (unpaired t-tests: P=0.0020 in D and P=0.0278 in E), suggesting disruption of adherens junctions. Each dot represents one heart. Representative images were chosen based on closeness to group mean and image quality.

To further investigate this observation, we employed a transgenic zebrafish line which has Ncad tagged with a tandem fluorescent timer (tFT; TgBAC(cdh2:cdh2-sfGFP-TagRFP)).²⁶ This tFT includes a superfolder green fluorescent protein that takes five minutes to fold (thereby marking nascent Ncad) and a tag red fluorescent protein which takes 100 minutes to fold (thereby marking stable Ncad). Using embryos from this transgenic line, we knocked down mapre2 using MO and performed immunofluorescence studies on hearts isolated at 5 dpf (Figure S9). By immunostaining for zonula occludens-1 (ZO-1), which marks tight junctions or cell borders, we were able to quantify the amount of nascent and stable Ncad signals that overlapped with ZO-1 signals. As shown in Figure S9, mapre2 knockdown did not lead to changes in nascent Ncad localization but significantly decreased stable Ncad localization at cell borders.

To confirm these findings, we crossed our mapre2 KO line onto this TgBAC transgenic background and repeated the experiment using sibling controls. As shown in Figure 4B-E, the germline homozygous mapre2 KO showed the same phenotype as the MO knock-down line where the ratio of stable to nascent Ncad is significantly decreased at cell borders (Figure 4E). Together, these data show that mapre2 loss-of-function leads to disruption of adherens junctions, which likely contributes to the decreased ventricular conduction and Na_V function given the known tight co-regulation between cardiac Na_V and Ncad in the connexome.^27,28

mapre2 loss-of-function leads to altered microtubule dynamics

To directly investigate the effect of mapre2 loss-of-function on microtubules, we first performed immunostaining on hearts isolated from WT versus homozygous mapre2 KO larvae for α-tubulin (total) and detyrosinated (Glu)-tubulin, which is a marker of stabilized microtubules. Using α-tubulin signal as mask, we found an approximately 25% decrease in the fraction of detyrosinated-tubulin in the KO ventricles (1.00±0.06 vs. 0.77±0.04 a.u., P<0.01; Figure 5A-B). To confirm this finding, we performed a similar experiment in ventricular myocytes isolated from adult fish. As shown in Figure 5C-D, we observed the same decrease in ventricular detyrosinated tubulin (Glu-tubulin) relative to total α-tubulin.

Figure 5. mapre2 loss-of-function leads to decreased detyrosinated tubulin.

A. Representative immunostaining of hearts from WT (n=8) and homozygous KO (n=16) larvae showing a decrease in ventricular detyrosinated tubulin (Glu-tubulin) relative to total α-tubulin. B. Quantification of ventricular Glu-tubulin signal using α-tubulin signal as a mask (unpaired t-test: P=0.0072). C. Representative immunostaining of ventricular myocytes isolated from adult WT (21 cells from 2 fish) and homozygous KO fish (20 cells from 3 fish) also showing a decrease in ventricular detyrosinated tubulin (Glu-tubulin) relative to total α-tubulin. D. Quantification of ventricular Glu-tubulin signal using α-tubulin signal as a mask (unpaired t-test: P=0.0232). Representative images were chosen based on closeness to group mean and image quality.

Given EB2’s role in regulating microtubule dynamics in vitro,²⁹ we knocked down MAPRE2 in human iPSC-derived cardiomyocytes and performed microtubule plus-end tracking experiments to directly measure microtubule dynamics. We first validated two siRNAs (#1 and #2) by Western blot in HeLa cells (Figure S10A) and by immunofluorescence cell staining in iPSC-derived cardiomyocytes (Figure S10B). We transfected human iPSC-derived cardiomyocytes with these two siRNAs, along with EB3-GFP to enable the visualization of microtubule plus ends, then performed live-cell imaging to define the behavior of individual microtubules (Figure 6A and Supplemental Videos). From these videos we obtained kymographs for individual microtubules and calculated their growth velocity and distance (Figure 6B). Compared to control siRNA, MAPRE2 knockdown with siRNAs #1 and #2 resulted in 1.11-fold and 1.26-fold increases in microtubule growth velocity (Control: 8.05±0.18 μm/min; siRNA #1: 8.91±0.19 μm/min; siRNA #2: 10.17±0.23 μm/min; P=0.002 Control vs. siRNA #1; P<0.0001 Control vs. siRNA #2; Figure 6C-D), and in 1.22-fold and 1.34-fold increases in microtubule growth distance (Control: 5.76±0.19 μm; siRNA #1: 7.03±0.20 μm; siRNA #2: 7.69±0.28 μm; P<0.0001 Control vs. siRNA #1; P<0.0001 Control vs. siRNA #2; Figure 6E-F), respectively. Taken altogether, these data provide direct evidence that mapre2 loss-of-function leads to altered microtubule dynamics and modifications, disrupted adherens junctions, and decreased Na_V function, without gross cardiac morphological changes.

Figure 6. mapre2 loss-of-function leads to changes in microtubule dynamics.

A. Representative still images of microtubule plus-end tracking experiments in iPSC-derived cardiomyocytes with MAPRE2 knockdown (KD) using 2 different siRNA versus control siRNA. Scale bars = 10 μm. Please refer to the Supplement for representative videos (Video S1 is Control, Video S2 is siRNA#1, Video S3 is siRNA#2). B. Representative kymographs obtained from the live-cell imaging, which were used for measurement of microtubule parameters. Vertical scale bars = 10 seconds; horizontal scale bars = 5 μm. Compared to control, MAPRE2 knockdown with siRNA #1 and #2 resulted in 1.11-fold (P=0.0079) and 1.26-fold (P<0.0001) increases in microtubule growth velocity (C-D), and in 1.22-fold (P<0.0001) and 1.34-fold (P<0.0001) increases in microtubule growth distance (E-F), respectively. Dunn’s multiple comparisons tests were used following significant Kruskal-Wallis tests (P<0.0001 for both D and F). Data extracted from 368 Control microtubules, 314 siRNA #1 microtubules, and 194 siRNA #2 microtubules, in 5 sets of cells. Representative videos and images were chosen based on closeness to group mean and quality.

Rescue of mapre2 loss-of-function by knockdown of tubulin tyrosine ligase

Based on our finding that mapre2 loss-of-function leads to a decrease in the fraction of detyrosinated α-tubulin and changes in microtubule dynamics (Figure 6), we were interested in restoring the fraction of detyrosinated tubulin as a potential means of rescue. The ttl gene encodes tubulin tyrosine ligase which is responsible for re-tyrosination of detyrosinated tubulins.³⁰ Based on this we knocked down ttl using MO in the homozygous mapre2 KO larvae. Immunostaining of isolated larval hearts at 5 dpf showed that ttl knockdown restored the ventricular fraction of detyrosinated α-tubulin relative to total α-tubulin (1.00±0.08 vs. 1.46±0.10 a.u., P<0.01; Figure 7A-B). To determine if this restoration of detyrosinated α-tubulin is associated with rescue of adherens junction disruption, we performed the same immunocytochemistry experiment using homozygous mapre2 KO on the TgBAC background. As shown in Figure 7C-D, MO knockdown of ttl significantly increased the ratio of stable to nascent Ncad, suggesting a restoration of Ncad balance at adherens junctions.

Figure 7. Knockdown of ttl restores fraction of detyrosinated tubulin and stable to nascent Ncad in mapre2 KO hearts.

A. Representative immunostaining of hearts from homozygous mapre2 KO larvae injected with control (ctl) versus ttl morpholinos showing a restoration of ventricular detyrosinated tubulin (Glu-tubulin) signal relative to total α-tubulin signal. B. Quantification of ventricular Glu-tubulin signal using α-tubulin signal as a mask (unpaired t-test: P=0.0017). C. Immunocytochemistry of hearts from homozygous mapre2 KO larvae on the transgenic background with cdh2 tandem fluorescent timer (tFT). Immunostaining of zonula occludens-1 (ZO-1) was used to mark cell borders. D. Quantification of GFP and RFP signals using ZO-1 signal as mask shows a significant increase in stable to nascent Ncad localization at cell borders (unpaired t-test: P=0.0028), suggesting a restoration of Ncad balance at adherens junctions. Representative images were chosen based on closeness to group mean and image quality. Each dot represents one heart.

Finally, to determine if ttl knockdown is associated with functional rescue of Na_V-related parameters such as conduction velocity and V_max, we performed voltage mapping also in homozygous mapre2 KO larvae injected with control (ctl) versus ttl MO. As shown in Figure 8, ttl knockdown resulted in a significant increase of ventricular conduction velocity (ctl: 12.1±1.6, ttl: 14.5±2.2 mm/s; P<0.05; Figure 8A) and V_max toward WT levels (ctl: 74.3±3.1, ttl: 88.4±2.3 1/s; P<0.0001; Figure 8B). There was also a significant decrease in ventricular APD with ttl knockdown (ctl: 249±17, ttl: 222±14 ms; P<0.05; Figure 8C). These findings demonstrate that ttl knockdown is able to partially rescue mapre2 loss-of-function as evaluated by detyrosinated α-tubulin, Ncad balance, as well as Na_V-related functional parameters. Altogether, our data demonstrate a novel role for microtubular dynamics in the regulation of cardiomyocyte junctions and ventricular conduction as illustrated in the Graphical Abstract.

Figure 8. Knockdown of ttl restores ventricular conduction in mapre2 KO hearts.

Voltage mapping of hearts isolated from homozygous mapre2 KO larvae injected with control (ctl) versus ttl morpholinos (MO). ttl knockdown resulted in a significant increase of ventricular conduction velocity (CV; Mann-Whitney test: P=0.0177; A) and V_max (maximum action potential upstroke velocity, dV/dt; unpaired t-test: P<0.0001; B), as well as a significant decrease in ventricular action potential duration (APD; unpaired t-test: P=0.0401; C). All hearts (represented by dots) were isolated from 5 dpf zebrafish larvae. The dotted squares in (A) reflect the main ventricular area from which the parameters were measured. APD was measured at 80% repolarization while the hearts were paced at 100 bpm.

DISCUSSION

In this study, we provide genetic and functional evidence that MAPRE2 and microtubules play an integral role in normal cardiac electrophysiology and when perturbed, lead to a phenotype consistent with a sodium channelopathy. Using both zebrafish and human iPSC-derived cardiomyocyte models, we demonstrate that mapre2 loss-of-function results in ventricular conduction slowing, decreased Na_V function, and arrhythmogenesis (Figures 1–3, S3), associated with disruption of adherens junctions (Figures 4 and S8). Mechanistically, mapre2 loss-of-function leads to a decrease in the fraction of detyrosinated α-tubulin and changes in microtubule dynamics (Figures 5 and 6). These molecular and functional changes are rescued by ttl knockdown which restores the detyrosinated α-tubulin fraction and Ncad ratio at cell borders (Figure 7) as well as Na_V-related functional parameters (Figure 8), suggesting a potential novel approach for treating conduction disorders.

Since it was first described around 1990,^31,32 our genetic understanding of Brugada syndrome has evolved from it being considered a monogenic disorder that is primarily determined by the inheritance of a single large-effect genetic factor, to being a polygenic disorder affected by the inheritance of multiple genetic risk factors that range in population frequency and effect size.¹ Genetic loci that have been associated with BrS provide strong support for loss of sodium current as a central mediator of BrS pathogenesis.¹ Rare large-effect loss-of-function variants in the coding region of SCN5A are found in ~20% patients.^33,34 Furthermore, common non-coding susceptibility variants identified by GWAS in and around SCN5A and at loci that harbor transcription factors that regulate SCN5A expression have implicated transcriptional effects of SCN5A expression.⁸ The GWAS association signal within MAPRE2 supports effects on sodium channel trafficking that involve microtubule function, yet this has not been explored.⁸ In silico evidence pointing to a potential causal role for MAPRE2 at this locus includes: 1) expression quantitative trait locus effects where the risk allele is significantly associated with a lower MAPRE2 mRNA expression in human LV compared to the non-risk allele (P=2.87×10⁻⁵; GTEx), and 2) chromatin interaction between MAPRE2’s promoter region and the risk haplotype (Hi-C from human left ventricle).³⁵

Based on these findings we previously acutely knocked out the highly homologous mapre2 gene in zebrafish using a multi-gRNA CRISPR/Cas9 approach,²⁰ and observed a significant decrease in both ventricular conduction velocity and V_max.⁸ However, that particular model was limited by somatic mosaicism which was not amenable to more detailed mechanistic studies. Here, we generated a definitive germline mapre2 KO as well as a complementary MO-knockdown model, both of which recapitulate the Na_V loss-of-function phenotype (Figure 1 and S2). In addition, we provide both ECG and patch clamp data from adult KO fish which directly demonstrate ventricular conduction slowing (Figure 2) and loss of Na_V current (Figure 3). Finally, we extend our previous work by showing that MAPRE2 KO leads to EAD, DAD, and irregular rhythms in spontaneously beating human iPSC-derived cardiomyocytes (Figure S3). Importantly, these findings are not explained by changes in Na_V expression (Figures S7 and S8), ventricular remodeling (Figures S4, S5 and S6), or intrinsic Na_V channel biophysics (Figure 3D and F). These data thus imply that the mapre2 loss-of-function acts through effects on Na_V trafficking and/or subcellular localization, resulting in an effective loss of Na_V function at the plasma membrane.

In epithelial cells, MAPRE2 knockdown leads to straighter and sometimes bundled microtubules, associated with changes in microtubule dynamics.²⁹ However, because the role of MAPRE2 in cardiomyocytes is not known, we knocked down MAPRE2 in human iPSC-derived cardiomyocytes and tracked the growing plus-ends of microtubules using live-cell imaging (Figure 6A-B, Supplemental Videos). This experiment demonstrated that MAPRE2 loss-of-function increases the velocity and distance of microtubule growth (Figure 6C-F). This increase in microtubule dynamics might lead to decreased microtubule stability and account for the reduction of microtubule detyrosination³⁰ (Figure 5). Together, these data suggest that MAPRE2 loss-of-function affects microtubule dynamics and post-translational modifications, which lead to an altered Na_V density at the membrane and a resultant ventricular conduction defect (Graphical Abstract).

In contrast to the decrease in α-tubulin detyrosination observed in our model (Figure 5), several recent studies have found an increase of detyrosinated α-tubulin in mouse models of cardiomyopathy including the mdx model of Duchenne’s muscular dystrophy³⁶ and the MYBPC3_2373insG model of hypertrophic cardiomyopathy.³⁷ These differences suggest that the relationship between microtubule dynamics and membrane protein trafficking is complex. It is likely that the balance of detyrosinated to tyrosinated α-tubulin, and thus the dynamics of microtubule turnover, require to be tightly matched to the mechanical and functional requirements of the cardiomyocyte. Nevertheless, because we observed a decrease of detyrosinated α-tubulin, we knocked down the ttl gene encoding tubulin tyrosine ligase, the enzyme that tyrosinates detyrosinated α-tubulin.³⁰ This resulted in the restoration of α-tubulin detyrosination and stable to nascent Ncad ratio at cell borders (Figure 7) as well as Na_V-related functional parameters to WT levels (Figure 8).

Microtubule detyrosination has been causally linked to cell polarity and control of microtubule dynamics in migrating fibroblasts and epithelial cells.^38–41 In our mapre2 KO model, the primary defect is the ablation of EB2, a microtubule plus-end binding protein, which leads to changes in microtubule dynamics including faster and longer growth. This also disrupts adherens junctions implicating a broader role for microtubules in localizing membrane proteins which are known to interact with the sodium channel. Importantly, a previous study has shown that the intracellular pool of Ncad is tightly associated with the microtubule network, and the formation of adherens junctions require microtubules.²⁵ Furthermore, there is evidence that in cardiomyocytes there is additional organization of adherens junctions, and interacting proteins including sodium channels, as the heart undergoes maturation during development.⁴²

Taken together, we propose that MAPRE2 loss-of-function ultimately leads to a defect in formation or maintenance of adherens junction and associated features of cell polarity^29,43 which then result, directly or indirectly, in Na_V mis-localization and the emergence of a ventricular conduction defect.²⁷ Similar to an earlier in vitro study,³⁸ we were able to rescue the dysfunction of conduction velocity and V_max by knocking down ttl and restoring the detyrosinated α-tubulin fraction, in addition to restoring the balance of stable to nascent Ncad at adherens junction. These observations implicate microtubule dynamics in the regulation of sodium channel density at the cell membrane and in the regulation of mechanical junctions, contributing to our understanding of the range of clinical phenotypes observed in BrS. These data also suggest that modulating microtubule dynamics may offer novel inroads for the development of antiarrhythmic therapies.

Study limitations

Our models induced large expression changes in mapre2 whereas in humans any change in MAPRE2 expression associated with the SNP is likely much smaller with a modest overall effect size. Because of the lack of a suitable antibody for immunostaining Na_V in zebrafish, we were unable to directly visualize the proposed mislocalization of Na_V. Nevertheless, this was inferred based on staining of Ncad for adherens junctions and the tight association between the two.²⁷

NOVELTY AND SIGNIFICANCE.

What is known?

• Brugada syndrome (BrS) is associated with loss-of-function SCN5A variants, yet these account for only ~20% of cases.

• Recent genome-wide association study (GWAS) identified a novel locus within MAPRE2, which encodes microtubule end-binding protein 2, implicating microtubule involvement in BrS.

What new information does this article contribute?

• MAPRE2 is important for maintaining normal cardiac Na_V function and ventricular conduction.

• MAPRE2 loss-of-function leads to changes in microtubule dynamics, disruption of cell-cell junctions, and decrease in cardiac Na_V function, a hallmark of BrS.

• Manipulation of the tyrosination status of α-tubulin rescues the Na_V dysfunction caused by MAPRE2 loss-of-function.

NON-STANDARD ABBREVIATIONS AND ACRONYMS

APD

action potential duration

BrS

Brugada syndrome

CTL

control

CV

conduction velocity

Cx43

connexin-43

DAD

delayed afterdepolarization

dpf

days post-fertilization

EAD

early afterdepolarization

EB

microtubule end-binding protein

Glu-tubulin

detyrosinated tubulin

GWAS

genome-wide association study

HET

heterozygous

HOMO

homozygous

I _Na

sodium current

KO

knock-out

MO

morpholino

Na_V

voltage-gated sodium channel

Ncad

N-cadherin

tFT

tandem fluorescent timer

WT

wild-type

V_max

maximum action potential upstroke velocity

ZO-1

zonula occludens-1