MARK4 controls ischaemic heart failure through microtubule detyrosination

Summary Myocardial infarction (MI) is a major cause of premature adult death. Compromised cardiac function after MI leads to chronic heart failure with systemic health complications and high mortality rate1. Effective therapeutic strategies are highly needed to improve the recovery of cardiac function after MI. More specifically, there is a major unmet need for a new class of drugs that improve cardiomyocyte contractility, because currently available inotropic therapies have been associated with high morbidity and mortality in patients with systolic heart failure2,3, or have shown a very modest risk reduction4. Microtubule detyrosination is emerging as an important mechanism of regulation of cardiomyocyte contractility5. Here, we show that deficiency of Microtubule-Affinity Regulating Kinase 4 (MARK4) substantially limits the reduction of left ventricular ejection fraction (LVEF) after acute MI in mice, without affecting infarct size or cardiac remodeling. Mechanistically, we provide evidence that MARK4 regulates cardiomyocyte contractility through promoting microtubule-associated protein 4 (MAP4) phosphorylation, thereby facilitating the access of Vasohibin 2 (VASH2), a tubulin carboxypeptidase (TCP), to microtubules for α-tubulin detyrosination. Our results show how cardiomyocyte microtubule detyrosination is finely tuned by MARK4 to regulate cardiac inotropy, and identify MARK4 as a promising druggable therapeutic target for improving cardiac function after MI.

Université de Paris, Institut National de la Santé et de la Recherche Médicale, U970, PARCC, Paris, France # These authors contributed equally to this work.

Summary
Myocardial infarction (MI) is a major cause of premature adult death. Compromised cardiac function after MI leads to chronic heart failure with systemic health complications and high mortality rate 1 . Effective therapeutic strategies are highly needed to improve the recovery of cardiac function after MI. More specifically, there is a major unmet need for a new class of drugs that improve cardiomyocyte contractility, because currently available inotropic therapies have been associated with high morbidity and mortality in patients with systolic heart failure 2,3 , or have shown a very modest risk reduction 4 . Microtubule detyrosination is emerging as an important mechanism of regulation of cardiomyocyte contractility 5 . Here, we show that deficiency of Microtubule-Affinity Regulating Kinase 4 (MARK4) substantially limits the reduction of left ventricular ejection fraction (LVEF) after acute MI in mice, without affecting infarct size or cardiac remodeling. Mechanistically, we provide evidence that MARK4 regulates cardiomyocyte contractility through promoting microtubule-associated protein 4 (MAP4) phosphorylation, thereby facilitating the access of Vasohibin 2 (VASH2), a tubulin carboxypeptidase (TCP), to microtubules for α-tubulin detyrosination. Our results show how cardiomyocyte microtubule detyrosination is finely tuned by MARK4 to regulate cardiac inotropy, and identify MARK4 as a promising druggable therapeutic target for improving cardiac function after MI.
Myocardial infarction, the main cause of ischaemic heart disease (IHD) and chronic heart failure, is a serious ischaemic syndrome in which the blood supply to the heart is blocked, thus causing substantial myocardial cell death and loss of function in the remaining viable cells 6 . Microtubule (MT) detyrosination, which is associated with DESMIN at forcegenerating sarcomeres 5 , is upregulated in the failing hearts of patients with ischaemic cardiomyopathy 5,7 and hypertrophic cardiomyopathies 5,7,8 , and suppression of microtubule detyrosination improves contractility in failing cardiomyocytes 7 . VASH1 or VASH2, coupled with a small vasohibin-binding protein (SVBP), forms TCP that are capable of tubulin detyrosination 9,10 . Depletion of VASH1 speeds contraction and relaxation in failing human cardiomyocytes 11 . Structural and biophysical studies have suggested that VASH interacts with the C-terminal tail of α-tubulin [12][13][14] . However, the regulatory mechanisms of this system are still poorly understood.
Microtubule stability is regulated by microtubule-associated proteins (MAPs), including classical MAPs such as MAP2, MAP4, and Tau 15 . MAP4 is expressed in the cardiomyocytes and MAP4 level significantly increases in human hearts with cardiomyopathy 7 . MAP4 dephosphorylation on microtubule network has been described in a feline model of pressure overload cardiac hypertrophy 16 , but the relationship of MAP4 phosphorylation with microtubule detyrosination has not been examined. MARK4 is an evolutionarily conserved serine-threonine kinase 17,18 known to phosphorylate MAPs including Tau, MAP2 and MAP4, on KXGS motif within their microtubule-binding motif [19][20][21] . The phosphorylation of MAPs triggered by MARK induces conformational changes that alter MAPs association with microtubules, and thereby regulates microtubule dynamics [19][20][21] . MARK4 is expressed in the hearts 20 , however the role of MARK4 in the cardiomyocyte has not been studied. Here, we examined whether MARK4 regulates the function of the failing cardiomyocyte through modulation of microtubule detyrosination.

Heart function of Mark4 -/mice post-MI
To evaluate the effect of MARK4 in the setting of IHD, we used a murine model of permanent left anterior descending (LAD) coronary artery ligation to induce a large MI 22,23 (Extended Data Fig.1a). We detected Mark4 mRNA (Fig.1a) and MARK4 protein ( Fig.1b) expression in the heart tissues, peaking between day 3 and day 5 post-MI ( Fig.1a-1c). MARK4 was almost exclusively detected in the cytoskeleton-enriched insoluble fraction of the whole heart extracts (Fig.1b), and was localized in the cardiomyocytes ( Fig.1c; Extended Data Fig.2a). MARK4 deficient mice (Mark4 -/-) displayed a remarkable preservation of LVEF, which was 63.6% (± 5.8 %) higher compared with their wild-type littermate controls on the first week post LAD surgery (Fig.1d), without any alteration of cardiac remodeling (Supplementary Table1). Interestingly, infarct scar size was similar between the two groups of mice (Fig.1e), indicating that the substantial difference in cardiac function between wild-type and Mark4 -/mice was not attributable to differences of size in viable cardiac tissues.

MARK4 regulates cardiac contractility
We found that the protective effect of MARK4 deficiency on the preservation of cardiac function was already apparent at 24 hours post-MI (Extended Data Fig.1b; Fig.2a), despite similar extent of myocardial injury, shown by comparable serum cardiac troponin I (cTnI) level (Fig.2b), and comparable infarct size analyzed by triphenyltetrazolium chloride (TTC) staining (Fig.2c), in Mark4 -/and wild-type mice. MARK4 has previously been shown to regulate NLRP3 activation in macrophages 24,25 , which could affect the outcome of post-ischaemic injury given the role of NLRP3 inflammasome in this setting 26,27 . However, MARK4 deficiency did not significantly alter local and systemic inflammatory responses to myocardial injury at day 3 post-MI (Supplementary Table 2; Extended Data Fig.2b) when the preservation of LVEF was already evident in Mark4 -/mice (Extended Data Fig.2c). Moreover, bone marrow transfer of Mark4 -/haematopoietic cells into wild-type mice (Extended Data Fig.1c; validation in Extended Data Fig.3a-3b) did not improve cardiac function after MI in comparison with the transfer of wild-type bone marrow cells (Fig.2d), indicating that the protective effect of MARK4 deficiency post-MI cannot be explained by the role of MARK4 in haematopoietic cells. In contrast, using an inducible conditional deletion of Mark4 in cardiomyocytes (Mark4cKO) (Extended Data Fig.1d; validation in Extended Data Fig.3c), we found a substantial preservation of LVEF in Mark4cKO mice post-MI, which was 56.8% (± 6.2%) higher when compared with their littermate control mice at day one post-MI (Fig.2e). The protective effect seen in Mark4cKO started as early as the first day after MI and lasted until the end of the observation at four weeks post-MI (Fig.2e). Very impressively, Mark4cKO mice had only 4.3% (±3.8%) LVEF reduction at day one post-MI, as compared with 37.9% (±5.5 %) LVEF reduction in the control mice (Fig.2e), without any difference in infarct size (Extended Data Fig.3e). The data further show an impact of the remaining/viable MARK4-deficient cardiomyocytes on the contractile function. Collectively, our data demonstrate an intrinsic role of cardiomyocyte-expressed MARK4 in controlling cardiac function post-MI.
To examine the effect of MARK4 on cardiomyocyte function, we subjected freshly isolated primary cardiomyocytes 28 from wild-type and Mark4 -/mice to a single cell contractility assay using an electrical stimulator (Fig.2f-2j). We found that sarcomere peak shortening of isolated cardiomyocytes strongly correlated with the in vivo LVEF (Fig.2g), indicating that isolated cardiomyocyte contraction measured ex vivo reflects LVEF assessed in vivo (Fig.1d,   Fig.2a, and Fig.2e). At baseline (BL), wild-type and MARK4-deficient cardiomyocytes had similar levels of resting sarcomere length (Extended Data Fig.4a-4b), sarcomere peak shortening and contraction/relaxation velocities (Fig.2h-2j), an observation consistent with the absence of LVEF difference between wild-type and Mark4 -/mice prior to MI (Fig.1d) Fig.4f) when compared with wild-type cells. Upstream changes of calcium influx in excitation-contraction coupling could contribute to the contractile alterations, however, we did not observe any significant difference of Ca 2+ transients between the electrically stimulated Mark4 -/and wild-type cardiomyocytes at baseline or at day 3 post-MI (Extended Data Fig.4g-4m). These data strongly demonstrate that MARK4 deficiency substantially improves both contractile and relaxation functions of cardiomyocytes after MI.

MARK4 alters microtubule detyrosination
Detyrosinated MTs represent tunable, compression-resistant elements that impair cardiac function in the human failing hearts 5,29 . We confirmed that detyrosinated α-tubulin level was significantly higher in cardiomyocytes isolated from ischaemic hearts compared with cardiomyocytes isolated from sham animals, in contrast with the remaining cell pool (immune cells, fibroblast, endothelial cells) which did not display such a change in α-tubulin detyrosination (Extended Data Fig.2d-2e). Previous data indicate that MARK4 affects posttranslational microtubule detyrosination and polyglutamylation in ciliated cells 30 . Therefore, we hypothesized that MARK4 deficiency may affect microtubule detyrosination in cardiomyocytes after MI. We found a significantly lower level of detyrosinated microtubules in whole heart tissue extracts (Fig.3a-3b), and in isolated cardiomyocytes (together with reduced polyglutamylated microtubules) (Fig.3e-3g; Extended Data Fig.2f-2g) of Mark4 -/mice compared with their littermate wild-type controls after MI. In the absence of MARK4, we observed reduced ratio of α-tubulin in the soluble fraction versus its level in the insoluble fraction (Fig.3c), indicating a reduced percentage of free tubulin level without MARK4. More interestingly, we found that the

Europe PMC Funders Author Manuscripts
level of tubulin detyrosination inversely correlated with LVEF ( Fig.3d), suggesting a major role of MARK4-dependent modulation of microtubule detyrosination in controlling cardiac function after MI.
To further address the hypothesis that MARK4 deficiency improves cardiomyocyte contractility through its impact on microtubule detyrosination, we employed a genetic approach to overexpress tubulin tyrosine ligase (TTL) using an adenovirus system (Extended Data Fig.5a-5c) to reverse the effect of TCP 31 (Fig.3h-3k). TTL overexpression robustly improved peak shortening ( level of detyrosinated microtubules in Mark4 -/cardiomyocytes. We further confirmed these data by using a pharmacological approach with parthenolide (PTL) to inhibit microtubule detyrosination 5,7 (Extended Data Fig.5j-5s). Taken together, our data show that MARK4 regulates cardiac inotropic function through its impact on microtubule detyrosination in cardiomyocytes.

MARK4 directs VASH2 access to MTs
Detyrosination of α-tubulin preferentially occurs on polymerized microtubules 32 . Apart from binding to VASH, C-terminal tubulin tails of the polymerized microtubules are also important for MAP binding 33,34 . MAP4 bound to the C-terminal tubulin tail along the protofilament stabilizes the longitudinal contacts of the microtubule, and this interaction can affect other microtubule binding partners such as the motor protein Kinesin-1 34 . MARK4, as a kinase, is expected to phosphorylate MAP4 on its KXGS motif (including S941 and S1073 in human MAP4, or S914 and S1046 in murine MAP4) within its microtubule-binding repeats 19,20 (Extended Data Fig.6a), and alters MAP4 binding status on the protofilament (Extended Data Fig.6b). We therefore hypothesized that MARK4, through modifying MAP4 phosphorylation, may affect VASH accessibility to C-terminal α-tubulin tail and therefore influence microtubule detyrosination. To address this, we firstly used an in vitro microtubule co-sedimentation assay. Both MAP4 (Extended Data Fig.6c-6d) and VASH2/SVBP (Extended Data Fig.6e-6f) were able to incrementally bind to polymerized microtubules when incremental amounts were separately applied in the assays, consistent with the results of the past studies 12,34 . Interestingly, we found that VASH2/SVBP bound to polymerized microtubules gradually decreased in the presence of incremental amounts of previously bound MAP4 (with four microtubule-binding repeats, 4R-MAP4) ( Fig.4a-4b).
Therefore, these results support the hypothesis that the level of MAP4 occupancy on the polymerized microtubules influences the level of VASH2 access to the microtubule protofilaments.
To confirm this hypothesis in in vivo, we performed biochemical subcellular fractionation on primary cardiomyocytes isolated from non-ischaemic and ischaemic hearts of wild-type and Mark4 -/mice using a commercial kit, which we have validated (Extended Data Fig.7a-7b).
We firstly confirmed that MAP4 was expressed in the cardiomyocytes and its level was higher post-MI (Extended Data Fig.7c), a result consistent with data showing that MAP4 levels significantly increase in human hearts with cardiomyopathies 7 . MAP4 was detected in its S914 phosphorylated (within KXGS motif) form (pMAP4 S914 ) in the pellet extraction buffer (PEB), and also in its S1046 form (pMAP4 S1046 ) in the cytosolic extraction buffer (CEB) (Extended Data Fig.7c-7e). Knocking down MAP4 by small hairpin RNA (shRNA) in the isolated cardiomyocytes post-MI led to increased VASH2 levels in the PEB fraction, confirmed by both western blot and immunocytochemistry (Extended Data Fig.7f-7i), which was in line with the results of in vitro microtubule co-sedimentation assay ( Fig.4a-4b).
VASH2 was detected as a specific band (validated by specific knock-down using shRNA, Extended Data Fig.8a) of around 50 kDa in the PEB fraction (Extended Data Fig.8a-8b), higher than its theoretical molecular weight of 40 kDa, presumably due to the formation of a stable complex with SVBP, because adding a denaturing agent (urea) reduced its size to around 40 kDa (Extended Data Fig.8b). Upon MI, pMAP4 S914 was detected as a 110 kDa form in the PEB fraction whereas pMAP4 S1046 was detected as a 220 kDa form in the CEB fraction ( Fig.4c and Extended Data Fig.7c). MAP4 was detected as giant puncta in the cytosol of cardiomyocytes isolated post-MI, and these puncta were barely present at baseline (Extended Data Fig.8c-8d). pMAP4 S1046 (in the CEB fraction) formed oligomerized structures (at 440 kDa or higher) as revealed on the native gel (Extended Data Fig.8e-8f), and these pMAP4 S1046 oligomers could be further reduced to the 220 kDa form in presence of urea as revealed on the denaturing gel (Extended Data Fig.8g). The data suggest that MAP4 phosphorylation at S1046 is associated with its presence as oligomers/giant puncta in the cytosol in situ. Our results are consistent with a structural model, in which S914 is within the weak microtubule binding repeat of MAP4, whereas S1046 is within the strong anchor point of MAP4 binding repeat to the microtubules 34 (Extended Data Fig.6b), so that S1046 phosphorylation can lead to detachment of MAP4 from polymerized microtubules and accumulation in the cytosol. Accordingly, a higher pMAP4 S1046 level was strongly and positively correlated with increased VASH2 levels in the PEB fraction (there also was a weaker correlation between pMAP4 S914 levels and VASH2 levels in the PEB fraction) (Extended Data Fig.7c, 7e) in wild-type cardiomyocytes, indicating an association between phosphorylated MAP4 (at S941 and S1046) levels and VASH2 levels on the polymerized microtubules. Strikingly, levels of pMAP4 S914 and pMAP4 S1046 were substantially reduced in Mark4 -/cardiomyocytes after MI ( Fig.4c-4d), confirming S914 and S1046 of MAP4 as MARK4 kinase substrate sites. Reduced levels of pMAP4 S1046 in the CEB fraction and pMAP4 S914 in the PEB fraction correlated well with a reduced level of VASH2 in the PEB fraction (r 2 =0.6165, P=0.0025; r 2 =0.4529, P=0.0165, respectively) ( Fig.4c-4e), with a stronger association between pMAP4 S1046 and VASH2. In addition, we found that VASH2 levels were positively correlated with DESMIN levels in the PEB fraction (Extended Data Fig.8h-8j), supporting previous data that detyrosinated microtubules are positively correlated with DESMIN levels in cardiomyocytes 5 . In summary, our results suggest that MARK4 kinase, through phosphorylation of MAP4 at S914 and S1046, changes MAP4 status to allow VASH2 access to the polymerized microtubule for its TCP activity.
To further confirm the causal effect of MARK4 on VASH2 localization, we overexpressed MARK4 in primary cardiomyocytes, which caused the appearance of pMAP4 S1046 (Extended Data Fig.9a-9c) and giant MAP4 puncta in the cytosol (Extended Data Fig.9d-9e), and led to increased VASH2 levels in the PEB fraction (Extended Data Fig.9a-9c). By using stimulated emission depletion (STED) super-resolution microscopy 35 , we found a strong co-localization of VASH2 on the polymerized microtubules in primary cardiomyocytes isolated from wild-type hearts post-MI when compared with the samples isolated from wild-type at baseline (Extended Data Fig.10a-10b). Total VASH2 levels were comparable between Mark4 -/cardiomyocytes and Mark4 +/+ cells post-MI (Extended Data Fig.10c-10d). However, there was a significant reduction of VASH2 association with polymerized microtubules in Mark4 -/compared to wild-type cardiomyocytes ( Fig.4f-4g).
In conclusion, our results demonstrate that MARK4 regulates microtubule detyrosination by phosphorylating MAP4 and controlling VASH2 accessibility to the microtubules (Extended Data Fig.10e).

Discussion
Detyrosinated microtubules impede contractile function of cardiomyocytes from failing human hearts 7 , and targeting the regulatory mechanism controlling microtubule detyrosination could represent a new inotropic strategy for improving cardiac function. We show a major role of MARK4 in the alteration of cardiomyocyte contractility through modulation of microtubule detyrosination in the ischaemic heart. It will be interesting to examine whether this protective effect of MARK4 inactivation on cardiac function after MI is sustained in the very long term (several months after MI) without inducing any harmful side effects, and whether MARK4 inhibition can improve contractile function in the setting of non-ischaemic heart failure. Furthermore, the marked improvement in relaxation kinetics in the absence of MARK4 raises the possibility of a potential beneficial effect of MARK4 inhibition in the setting of heart failure with preserved ejection fraction, an increasingly common cardiac syndrome associated with high morbidity and mortality. The molecular and structural mechanisms of MARK4 coupled with MAP4 and VASH2/SVBP in modifying microtubule detyrosination will need to be probed in other settings such as mitosis where regulation of detyrosinated microtubules has significant pathophysiological relevance 9,36 , and the differential role of other TCPs (e.g. VASH1) will need to be further studied in the future.

LAD coronary artery ligation model
Permanent left anterior descending coronary artery ligation was performed on experimental animals as described 22,23 previously with minor modification. Mice, at 8-10 weeks of age, were anesthetized using ketamine at 100mg/kg (body weight) and xylazine at 10mg/kg (body weight) via i.p., and then intubated and ventilated with air (supplemented with oxygen) using a small-animal respirator. A thoracotomy was performed in the fourth left intercostals space. The left ventricle was visualized and the pericardial sac was ruptured to expose the LAD. The LAD was permanently ligated using a 7-0 Prolene suture. The suture was passed approximately 2mm below the tip of the left auricle. Significant colour changes at the ischaemic area and ECG changes were monitored as an indication of successful coronary artery occlusion. The thoracotomy was closed with 6-0 Prolene sutures. Sham-operated animals underwent the same procedure without coronary artery ligation. The endotracheal tube was removed once spontaneous respiration resumed, and the mice were placed on a warm recovery cage maintained at 37 °C until they were completely awake. At the indicated time points in the experimental timeline, the mice were sacrificed by CO 2 asphyxiation, and the tissues were subsequently harvested for analysis.

Bone marrow transplants
Eight to ten-week old C57BL/6 mice were maintained overnight with Baytril (Bayer AG) before irradiation with two doses of 5.5 Gy (separated by 4 hours) followed by reconstitution with 1×10 7 sex-matched donor bone marrow cells. Animals were randomly assigned to receive the Mark4 -/or Mark4 +/+ bone marrow. Mice were then maintained on Baytril for a 4-week recovery period before performing LAD ligation.

Echocardiography
Transthoracic echocardiography was performed on all mice using Vevo 3100 with a MX400 linear array transducer (VisualSonics), 30 MHz. Mice were anesthetized with 2-3% isoflurane and kept warm on a heated platform (37 °C). The chest hairs were removed using depilatory cream and a layer of acoustic coupling gel was applied to the thorax. After alignment in the transverse B-mode with the papillary muscles, cardiac function was measured on M-mode images. Echocardiography data were collected by using VisualSonics Vevo 3100, and analyzed by using Vevo LAB3.1.1.

Histological analysis
Whole hearts were excised at different time point after LAD ligation, rinsed in PBS and fixed with 4% PFA overnight at 4°C. Fixed tissues were thoroughly washed in PBS, and then sinked in 30% sucrose. Tissues were embedded and sectioned by a cryostat into 10μm thick slices, which started at the apex and ended at the suture ligation site. Masson's trichrome staining was performed to determine scar size. Scar size (in %) was calculated as total infarct circumference divided by total left ventricle circumference. Some hearts were excised at 24 hours post-MI and quickly sliced into four 1.0 mm thick sections perpendicular to the long axis of the heart. The sections were then incubated with 1% triphenyltetrazolium chloride (TTC, Sigma) for 15 minutes at 37°C and then digitally photographed. For infarct size at 24 hours post-MI, TTC-stained area, and TTC-negative staining area (infarcted myocardium) were measured using ImageJ (v2.0). Myocardial infarct size was expressed as a percentage of the total left ventricle area. Images were obtained by using Leica DM6000 B Microscope, collected by using LAS AF software (2.4.0 build 6254), and analyzed by using ImageJ (v2.0) analyze tools.

Tissue immunohistochemistry
Whole hearts were excised, quickly washed in PBS, and flash frozen. Tissues were then embedded and cryo-sectioned. Slices were fixed in pre-chilled methanol for 10 minutes at -20°C. After washing with PBST (0.1% tween-20 in 1 × PBS), slices were incubated with 3% H 2 O 2 (in PBS) for 10 minutes, and then with blocking buffer (5% BSA in PBST) for 1 hour at room temperature. The primary antibody against MARK4 (Abcam, ab124267, used at 1:200), or rabbit IgG isotype control (Novus Biologicals, NB810-56910, used at 1:1000) was used for overnight at 4°C. Extensive washing steps were performed to remove nonspecific binding antibody. Slices were incubated with the biotinylated secondary antibody (Abcam, ab6720, used at 1:800) for 1 hour at room temperature. Reagents A and B from Avidin-Biotin Complex kit (VECTOR, PK-4000) were diluted and added to the slides. The slides were stained with ImmPACT DAB peroxidase substrate (VECTOR, SK-4105), and counterstained with hematoxylin. Images were obtained by using Leica DM6000 B Microscope, collected by using LAS AF software (2.4.0 build 6254), and analyzed by using ImageJ (v2.0) analyze tools.

Microtubule co-sedimentation assay
Lyophilized porcine brain tubulin (T240) was purchased from Cytoskeleton, Inc (Denver, USA). Recombinant proteins of 4R-MAP4 and VASH2/SVBP were previously described 12,34 . The desiccated tubulin was reconstituted in the microtubule polymerization buffer to 10 mg/mL. To generate polymerized microtubules, tubulin was diluted to 2 mg/mL in the polymerization buffer (80mM K-PIPES, pH 6.8, 1mM MgCl2, 1mM EGTA and 1mM DTT), supplemented with 5% glycerol and 1mM GTP at 37°C for 30 minutes, and then stabilized by incubating with 2.5 μM taxol at 37°C for 15 minutes. The taxol stabilized MTs were centrifuged over cushion buffer (polymerization buffer with 40% glycerol) at 131,700g at 37°C for 15 minutes to remove the free tubulin. The pellet was suspended in the polymerization buffer with 1 μM taxol. Taxol influenced the association of 4R-MAP4 with the MTs in our assay. 4R-MAP4 association was facilitated when taxol was completely excluded from the buffer. The MTs without taxol were susceptible to depolymerisation if stored at room temperature. In these conditions, the polymerized microtubules were maintained at 37°C throughout the experiment. For the co-sedimentation assay, the MTs were mixed with various concentrations of 4R-MAP4 (1-6 μM) and VASH2/SVBP (1-4 μM) in the polymerization buffer. In the competition experiments, the MTs were incubated with specified 4R-MAP4 concentrations (1-4 μM) for 10 minutes, followed by addition of constant amount of VASH2/SVBP (3 μM) with further incubation of 10 minutes. Subsequently, the reaction mixture was centrifuged in TLA120.2 rotor at 55,000 rpm for 15 minutes. The pellet fraction containing the MTs and bound proteins was resuspended in the loading buffer. The samples were loaded on 10 % SDS-PAGE gel and stained with Colloidal Coomassie blue dye (ThermoFisher). The experiments were repeated at least 3 times. The band intensities were analyzed using ImageJ (v2.0).

Murine cardiomyocyte isolation
Cardiomyocytes preparation was accomplished as previously described 28 . In brief, mice were anesthetized, and the chest was opened to expose the heart. Descending aorta was cut, and the heart was immediately flushed by injection of 7 mL EDTA buffer into the right ventricle. Ascending aorta was clamped, and the heart was transferred to a 60 mm dish containing fresh EDTA buffer. Digestion was achieved by sequential injection of 10 mL EDTA buffer (NaCl, 130mM; KCl, 5mM; NaH 2 PO 4 , 0.5mM; HEPES, 10mM; Glucose, 10mM; BDM, 10mM; Taurine, 10mM; EDTA, 5mM; pH to 7.8), 3

Cardiomyocyte contractility assay
Sarcomere shortening and relaxation were measured in freshly isolated left ventricular cardiomyocytes of murine hearts using the integrated IonOptix contractility/photometry system. Cardiomyocytes were maintained in normal Tyrode's solution (NaCl, 140mM; MgCl 2 , 0.5 mM; NaH 2 PO 4 , 0.33mM; HEPES, 5mM; glucose, 5.5mM; CaCl 2 , 1mM; KCl, 5mM; NaOH, pH to 7.4) at room temperature, electrically stimulated at 2 Hz using a field stimulator, and changes in sarcomere length were recorded. Basal and peak sarcomere length, maximum departure/return velocities and time to peak were measured. All measurements were performed at room temperature. For PTL experiments, cardiomyocytes were treated with 10 μM PTL (Sigma P0667) or vehicle at room temperature in normal Tyrode's solution for 1 hour before contractility measurements, and the vehicle dimethyl sulfoxide (DMSO) diluted in the same way was applied as control. All measurements were performed at room temperature within 4 hours. Data were collected and analyzed by using IonWizard 7.4.

Calcium measurement
Measurement of intracellular calcium was performed in freshly isolated left ventricular cardiomyocytes using integrated IonOptix contractility/photometry system. Cardiomyocytes were loaded with 1 μM Fura-2-AM for 20 minutes (protected from light), and then washed to allow de-esterification for 20 minutes. Cells were then rinsed with a normal Tyrode's Solution. Cells were stimulated at 2 Hz using a field stimulator with dual excitation (at 360 and 380 nm), and emission light was collected at 510 nm. Changes in calcium transients were recorded using IonOptix software. All the cells analyzed were beating. All measurements were performed at room temperature within 4 hours. Data were collected and analyzed by using IonWizard 7.4

Immunofluorescence and image acquirement
Cardiomyocytes were fixed with pre-chilled methanol for 10 minutes, then washed twice using PBST (0.1% tween-20 in 1×PBS) with 5 minutes intervals. Cells were blocked for 1 hour at room temperature with blocking buffer (5% BSA in PBST) and incubated with primary antibodies overnight at 4°C.

STED imaging and image analysis
Cardiomyocytes on coverslips were fixed with 100% methanol for 15 minutes at room temperature and then washed three times with PBS (5 minutes intervals). Cells were blocked with buffer (5% BSA and 0.2% TX-100 in PBS) for 30 minutes, then incubated with primary antibodies (diluted in blocking buffer) overnight at 4°C. The primary antibodies were VASH2 (Abcam, ab224723, used at 1:200), MAP4 (Abcam, ab245578, used at 1:200) and α-tubulin (CST, 3873S, used at 1:200). The cells were washed three times using wash buffer (0.05% TX-100 in PBS) at room temperature, then incubated with the secondary antibody for 1 hour at room temperature. The secondary antibodies were: Atto 594 goat anti-Rabbit IgG (Sigma, 77671, used at 1:500), and Atto 647N goat anti-mouse IgG (Sigma, 50185, used at 1:500). Cells were then washed three times in wash buffer. Cells were fixed (3% Paraformaldehyde and 0.1% glutaraldehyde diluted in PBS) followed by three washes in PBS. The coverslips were then mounted on the slide.
STED imaging was carried out on a custom multicolour system with three pulsed excitation lines, one fixed depletion line, fast 16 kHz beam scanning and gated detection centered around an Olympus IX83 microscope base. This system uses identical hardware, and a closely matched optical arrangement, to the system previously published by Bottanelli and co-workers 35 . In brief, two-colour STED imaging was performed sequentially. Images were acquired with a 100X oil immersion objective lens (Olympus, UPLSAPO 100XO/PSF). MAP4 oligomerized puncta (with diameter longer than 400 nm) were measured and calculated using ImageJ (v2.0). The number of puncta was normalized against the cell area on each image.
For the acquired images, a dynamic thresholding algorithm was used for the image analysis. Images were converted into HSV colour images (C) with information of Hue (h), Saturation (s), and Value (v). C (I, j) was assumed as a non-background image pixel, N was the total number of non-background image pixels. The average of all the non-background image pixels was calculated as: k = (Σ i = 1 ℎ Σ j = 1 w C(i, j))/N. The collected pellets were dissolved in 2xSDS buffer (4% SDS; 20% glycerol; Tris-HCl, 0.25M; pH to 6.5). The dissolved lysates were then centrifuged at 14,000g for 5 minutes at 4°C. The supernatants were collected as F2 (extraction from the stable pellet fraction), and the residual pellets were kept.
The primary antibodies used for fractionation assays were: Detyrosinated α-tubulin (Abcam, ab48389, used at 1:1000), α-tubulin (CST, 3873S, used at 1:1000), TTL (Proteintech, 13618-1-AP, used at 1:1000), VASH1 (Abcam, ab199732, used at 1:1000), VASH2 (Abcam, ab224723, used at 1:1000), MAP4 (phospho S1073) (Abnova, PAB15916, used at 1:1000), MAP4 (Abcam, ab245578, used at 1:1000), MAP4 (phospho S941) (Abcam, ab56087, used at 1:1000), GAPDH (CST, 5174S,used at 1:1000), DESMIN (R&D, AF3844, used at 1:1000). Membranes were revealed with ECL. Quantification of western blot band was performed using ImageJ (v2.0). The band density was normalized in two steps: 1). The density of the targeted band was first normalized against the density of the loading molecular weight marker band (Norm 1). 2). The value of Norm 1 was internally normalized against the average value of Norm1 of the control group (Norm2). The finalized value (Norm 2) was used to compare the fold changes against the value of control groups across different gels. DESMIN was used as marker for the pellet fraction, and GAPDH was used as a marker for the cytosolic fraction. Coomassie blue stained gels loaded with the same amounts of proteins as used in western blotting experiments, or Ponceau S stained membranes after the transferring step were used to confirm the equal loading. All the immunoblots, gels and membranes associated with the data presented in the Figures and Extended Data Figures are provided (Supplementary Figure 1).

Heart tissue digestion and flow cytometry
Hearts were collected and the left ventricle was isolated, minced with fine scissors, and

Real-time PCR
For gene expression analysis, RNA from heart tissues or separated cardiomoycytes was isolated using an RNAeasy mini kit (Qiagen). Reverse transcription was performed using a QuantiTect reverse transcription kit (Qiagen). qRT-PCR was performed with SYBR Green qPCR mix (Eurogentec) using the Roche LightCycler 480II. Primer sequences are as follows: Mark4 (For. The average of three housekeeping genes (Hprt, Rpl4, and Rpl13a) was used as reference for qPCR gene expression analysis.

Measurement of cTnI and inflammatory cytokines
Serum was collected within 24 hours post-MI or at day 3 post-MI. Measurements of cardiac injury biomarker (collected within 24h) and cytokines (collected at day 3 Post-MI) were performed by core biochemical assay laboratory of Cambridge University Hospitals.

Statistics and Reproducibility
All values in the text and figures are presented as mean ± s.e.m. of independent experiments with given n sizes. Statistical analysis was performed with Prism 7.05 (GraphPad) and Excel (Microsoft Excel 2102). Violin plots were created with Prism 9.1.0 (216) (GraphPad). Data were tested for normality using a Kolmogorov-Smirnov test. Group comparisions were analyzed using two-tailed analyses. Comparisons of 3 groups or more were analyzed using one-way (one variable) or two-way ANOVA (two variables) followed by the Bonferroni post-hoc correction for multiple comparisons when appropriate. P<0.05 was considered statistically significant. the same hearts used as control. Representative WBs (d). Ratio of dTyr-tubulin over total αtubulin quantified using western blot data from biologically independent samples (S group: n=4 mice; MI group: n=5 mice) (e). f-g, Western blots of cell lysates from the isolated cardiomyocytes of Mark4 -/or control mice at day 3 post-MI, to detect detyrosinated α-tubulin (dTyr-tub), polyglutamylated α-tubulin (Polyglu-tub), acetylated α-tubulin (Acetub), and a-tubulin (α-tub). Representative images (f). Ratio of dTyr-tub, or polyglu-tub, or ace-tub over total α-tubulin quantified using western blot data from biologically independent samples (n=3 mice per group) (g). The box bounds represent the 25 th and 75 th percentiles, the middle line shows the median, and the whiskers show the minimum and maximum (b).
Mean±s.e.m.; two-tailed unpaired t-test (c, e, g). P values are indicated on the graphs.  Mean±s.e.m.; two-tailed unpaired t-test (b, c, d, e). P values are indicated on the graphs. Fig. 4. The effect of MARK4 deficiency on sarcomere length, peak shortening, velocity, and calcium transients in cardiomyocytes before and after myocardial infarction.    n=38 CMs examined over 3 independent experiments). c, Real-time PCR on post-MI CMs, from the following groups: Mark4 +/+ MI (n= 5 mice), and Mark4 -/-MI (n=6 mice).

Extended Data
d, Quantification of VASH2 mean fluorescence intensity (MFI) within cell area (region of interest, ROI) using the STED images from the following groups: Mark4 +/+ MI (n=6 mice / n= 38 CMs examined over 3 independent experiments), and Mark4 -/-MI (n= 6 mice/ n= 47 CMs examined over 3 independent experiments). Mean ± s.e.m.; two-tailed unpaired t-test (b, c, d). P values are indicated on the graphs. e, A working model for MARK4-dependent regulation of microtubule detyrosination after MI: Upon ischaemic injury, increased MARK4 phosphorylates MAP4 at its KXGS motifs. Phosphorylated MAP4 either changes its conformation on the polymerized microtubules, or detaches itself from the polymerized microtubules to form oligomerized MAP4 structures in the cytosol. The phosphorylation of MAP4 by MARK4 allows for space access of VASH2 to the polymerized microtubules, thereby promoting α-tubulin detyrosination. As a consequence, the increased level of detyrosinated microtubules causes a reduction in contractile function of the cardiomyocyte.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.