Big tau aggregation disrupts microtubule tyrosination and causes myocardial diastolic dysfunction: from discovery to therapy

Abstract

Background

Amyloid plaques and neurofibrillary tangles, the molecular lesions that characterize Alzheimer’s disease (AD) and other forms of dementia, are emerging as determinants of proteinopathies ‘beyond the brain’. This study aims to establish tau’s putative pathophysiological mechanistic roles and potential future therapeutic targeting of tau in heart failure (HF).

Methods and results

A mouse model of tauopathy and human myocardial and brain tissue from patients with HF, AD, and controls was employed in this study. Tau protein expression was examined together with its distribution, and in vitro tau-related pathophysiological mechanisms were identified using a variety of biochemical, imaging, and functional approaches. A novel tau-targeting immunotherapy was tested to explore tau-targeted therapeutic potential in HF. Tau is expressed in normal and diseased human hearts, in contradistinction to the current oft-cited observation that tau is expressed specifically in the brain. Notably, the main cardiac isoform is high-molecular-weight (HMW) tau (also known as big tau), and hyperphosphorylated tau segregates in aggregates in HF and AD hearts. As previously described for amyloid-beta, the tauopathy phenotype in human myocardium is of diastolic dysfunction. Perturbation in the tubulin code, specifically a loss of tyrosinated microtubules, emerged as a potential mechanism of myocardial tauopathy. Monoclonal anti-tau antibody therapy improved myocardial function and clearance of toxic aggregates in mice, supporting tau as a potential target for novel HF immunotherapy.

Conclusion

The study presents new mechanistic evidence and potential treatment for the brain–heart tauopathy axis in myocardial and brain degenerative diseases and ageing.

We present mechanistic evidence that tau tangles are involved in cardio and neurodegenerative diseases and that tau can be targeted by immunotherapy.

Structured Graphical Abstract

Structured Graphical Abstract.

Tangle-like aggregates of hyperphosphorylated tau are present in the heart of patients with dilated cardiomyopathy or Alzheimer’s disease. A mouse model of tauopathy recapitulates the pathological findings of the human disease. Disruption of microtubule de/tyrosination causes myocardial diastolic dysfunction observed in the mouse model, and it can be targeted with immunotherapy.

Abbreviations: iDCM, idiopathic dilated cardiomyopathy; PAO, pre-amyloid oligomers; TOMA, tau oligomer monoclonal antibody.

See the editorial comment for this article ‘Big tau aggregation and the broken heart’, by O.O. Nyarko and C.C. Sucharov, https://doi.org/10.1093/eurheartj/ehad204.

Translational Perspective.

The study has diagnostic and therapeutic translational implications. It presents the first evidence that human hearts express the high-molecular-weight isoform of the protein tau as well as tau aggregates in heart failure and/or Alzheimer’s hearts. This finding represents a novel molecular determinant of diastolic dysfunction and acts as an early sign of Alzheimer’s disease. In contrast to heart failure with reduced ejection fraction, there is limited guiding evidence for the treatment of heart failure with preserved ejection fraction, and the study provides initial evidence that myocardial tauopathy can be effectively targeted by immunotherapy, the first in the heart failure treatment toolkit.

Introduction

Alzheimer’s disease (AD) is a disorder of protein folding defined by the presence of deposits of amyloid-beta (Aβ) in senile plaques, and hyperphosphorylated microtubule-associated protein tau (MAPT), composing neurofibrillary tangles (NFTs). First described in AD, aggregates of hyperphosphorylated tau are linked to a wide spectrum of neurodegenerative disorders underlying the pathogenic importance of tau in a range of illnesses of adulthood. Notably, all tauopathies described so far are brain disorders.

Recently, AD and heart failure (HF) have been pathophysiologically linked beyond perfusion defects or abnormal vascular reactivity,^1,2 via the pathogenic mechanism of protein misfolding resulting in the hallmark of deposits of Aβ fibrillar and non-fibrillar soluble pre-amyloid oligomers.^3,4 Tau has instead been viewed as a brain-specific protein despite MAPT mRNA expression in other organs, proteomics⁵ evidence of tau expression in rodent myocardium, and the critical role of tau in promoting microtubule formation⁶ in virtually every cell in the body. However, neither the expression nor the possible pathological roles and mechanisms of tau in human HF have been previously studied.

In this study, we provide new evidence that the high-molecular-weight (HMW) tau isoform,⁷ pathogenic hyperphosphorylated tau, and toxic tau oligomers are detectable in murine and human myocardium and participate in the pathogenesis of diastolic dysfunction. First, we linked tau pathology to HF in a humanized tau mouse model (hTau) of AD. Second, we mechanistically connected cardiac tauopathy to tau oligomer deposits and to altered microtubule detyrosinated/tyrosination equilibrium, the imbalance of which has been previously associated with an increase myocardial stiffness, diastolic dysfunction, and HF.⁸ Finally, we tested the efficacy of a novel immunotherapy promoting the increased clearance of toxic tau oligomers, and improving myocardial diastolic performance. The last two decades have seen a surge in the number of clinical trials for AD, identifying oligomer disaggregation as a target to reduce brain pathology and improve clinical symptoms.^9,10 Our results indicate that tau oligomer immunotherapy offers a potential new approach to treat Alzheimer’s cardiomyopathy by targeting multiple mechanisms of the disease.

Methods

Details of the methods are included in the Supplementary material online.

Human samples

Failing explanted hearts were obtained from patients with a primary diagnosis of idiopathic dilated cardiomyopathy (iDCM). Heart and brain tissue from controls and AD patients were obtained from organ donors.

Mouse models

Three-, seven-, and twelve-month-old hTau and C57Bl/6J wild-type (WT) mice and sections of a mouse model of transverse aortic constriction were used in the study.

Functional studies

Myocardial function was studied in vivo in WT and hTau mice by transthoracic echocardiography and in vitro in isolated cardiomyocytes. Cognitive function was measured by Open Field, Novelty Y-maze, and Morris Water Maze tests conducted at the VA Small Animal Behavioural Core Facility.

Immunoblotting and enzyme-linked immunosorbent assay (ELISA)

Immunoblotting and ELISA were performed using standard methods.

Immunofluorescence staining

Immunofluorescence staining and analysis of human and mouse brains, hearts, and cardiomyocytes were carried out independently in the laboratories of Drs del Monte, Kayed, Norris, and Prosser.

Tau oligomer monoclonal antibody (TOMA) treatment

Twelve-month-old hTau mice were treated with a single i.p. injection of either IgG or TOMA as previously described.¹¹ The same mice were followed by echocardiography 10 and 30 days after injection. Mice were studied 10 days after injection to compare with the results published on the brain and at 1 month to recapitulate common therapeutic regimens with monoclonal antibodies (mAbs) for human diseases.

Statistical analysis

Details of the statistical analysis are described in the Supplementary material online for the different experiments. The adjustment for multiple comparisons was applied for all pairwise comparisons within a specific outcome.

Results

HMW tau is the heart’s isoform

The clinical data from iDCM, AD, and control patients are presented in the Supplementary material online, Table S1. The molecular analysis of both human and mouse heart tissue demonstrated that the most represented tau isoform in the heart is the HMW (100–110 kDa) from, whereas in the brain tau is mostly expressed as a ∼55 kDa form (Figure 1). While the 100–110 kDa band may represent the 55 kDa dimer, the absence of the 55 kDa band in the heart and the proteinase K treatment (see Supplementary material online, Figure S1) suggest that the band represents the HMW tau. Total tau expression, using tau 2E9, was not significantly different in the iDCM (soluble: control 0.81 ± 0.28, iDCM 0.80 ± 0.22, NS; insoluble: control 0.93 ± 0.23, iDCM 0.98 ± 0.26, NS) and the AD (soluble: control 1.02 ± 0.15, AD 1.09 ± 0.40, NS; insoluble: control 1.24 ± 0.05, AD 1.26 ± 0.08, NS) hearts by SDS–PAGE (Figure 1A–F). ELISA, using tau5, confirmed that total tau was not significantly different in iDCM hearts (control 501.69 ± 46.08, iDCM 470.83 ± 63.47, NS) (Figure 1G), whereas it showed higher expression in both AD heart (control 551.02 ± 62.35, AD 734.02 ± 166.35, P = 0.008) (Figure 1G) and brain (control 429.78 ± 89.51, AD 748.78 ± 37.71) (Figure 1I). Immunofluorescence showed that tau co-localizes with tau oligomers stained with the structural antibody T22 in diseased tissue (Figure2).

Figure 1.

Human hearts express HMW tau: SDS–PAGE of tau expression in myocardial tissue. The age of patients is indicated at the bottom of the blots. Soluble (A) and insoluble (B) fractions of iDCM and age/sex/ethnicity-matched controls; (C) quantification of tau expression. Soluble (D) and insoluble fraction (E) of AD and age/sex/ethnicity-matched controls. (F) quantification of tau expression; (G–I) total tau quantification by ELISA using tau5 and phosphor-tau employing p-Ser^396/404 antibodies in the heart of patients with iDCM and AD and in the brain of patients with AD and controls. Data are presented as means ± SEM. P-values have been calculated using the non-parametric two-tailed Mann–Whitney test for the SDS–PAGE. Statistical analysis for the ELISA was performed using one-way ANOVA with Tukey post-hoc analysis. When values were normalized by age, the P = value for total tau was 0.870 for Con vs. iDCM and 0.003 for Con vs. AD. Abbreviations: H = heart; B = brain; iDCM = idiopathic dilated cardiomyopathy; AD = Alzheimer’s disease.

Figure 2.

Oligomerized tau accumulates in myocardial tissue in iDCM and AD. (A) Immunofluorescence images of total tau (tau5) and tau oligomers (T22) and merged images in control (A–C) and iDCM (D–I) myocardial tissue; (B) immunofluorescence images of total tau (tau5) and tau oligomers (T22) and merged images in control (A–D), iDCM (E–H), and AD (I–L) myocardial tissue. Abbreviations: iDCM = idiopathic dilated cardiomyopathy; AD = Alzheimer’s disease.

Tau phosphorylation and pathology in human hearts

SDS–PAGE showed that tau is hyperphosphorylated on Ser³⁹⁶ (Figure 3A–C) in both the soluble (control 0.81 ± 3.8, iDCM 2.85 ± 2.8, P = 0.047) and insoluble fractions (control 1.62 ± 0.23, iDCM 1.98 ± 0.38, P = 0.028) and on Ser^396/404 (see Supplementary material online, Figure S2) in iDCM (control 0.47 ± 0.20, iDCM 0.96 ± 0.55, NS) and AD (control 0.45 ± 0.06, AD 0.59 ± 0.06, P = 0.031) hearts. p-Ser³⁹⁶ was not significantly increased in the heart of AD patients, possibly due to the variability of the samples, whereas the brain signal could not be quantified due to saturation of the signal (Figure 3D–F). Increased tau p-Ser^396/404 was shown by ELISA both in iDCM heart (control 100.13 ± 22.93, iDCM 185.32 ± 39.46, P < 0.001) and in AD heart (control 79.70 ± 14.87, AD 240.14 ± 8.52, P < 0.001) and brain (control 42.08 ± 4.07, AD 237.43 ± 1.47) (Figure 1H and I).

Figure 3.

Phospho-tau Ser³⁹⁶ and p-Ser²⁶² in human iDCM heart and AD heart and brain. SDS–PAGE of phospho-tau in myocardial tissue. The age of patients is indicated at the bottom of the blots. Tau is hyperphosphorylated on Ser³⁹⁶ (A) and segregated in the aggregates (B) in human iDCM hearts, while no changes appear in AD hearts (D–F). p-Ser²⁶² in iDCM (G, H) and AD hearts (I, L). AD brain expression of tau p-Ser³⁹⁶ could not be quantified due to signal saturation. Data are presented as means ± SEM. P-values have been calculated using the non-parametric two-tailed Mann–Whitney test for the SDS–PAGE. Abbreviations: H = heart; B = brain; iDCM = idiopathic dilated cardiomyopathy; AD = Alzheimer’s disease.

Ser²⁶² phosphorylation instead tends to be decreased, although non-significantly, in iDCM (control 0.87 ± 0.36, iDCM 0.65 ± 0.32, NS) and AD hearts (control 0.97 ± 0.07, iDCM 0.89 ± 0.26, NS) (Figure 3G and H) and it was undetectable in the human brains (Figure 3I and L).

By immunofluorescence p-Ser³⁹⁶ co-localizes in the diseased myocardium with the toxic tau oligomers stained with structural antibodies (T22) (Figure 4A and B) and is present inside the cardiomyocytes in iDCM and AD (Figure 4C; Supplementary material online, Figure S3) while tau oligomers are undetectable in the control tissue and cardiomyocytes (Figure 4).

Figure 4.

Oligomerized tau accumulates in tissue and isolated cardiomyocytes in iDCM and AD. (A) Immunofluorescence images of tau oligomers in control, iDCM, and AD myocardial tissue; (B) co-localization and Manders coefficient of co-localization of tau p-Ser³⁹⁶ and tau oligomers in control, iDCM and AD myocardial tissue. The arrow indicates the site of measurement of the co-localization. (C) Tau p-Ser³⁹⁶ and tau oligomers localize inside cardiomyocytes as shown by the staining of isolated cardiomyocytes. Green, phospho-tau; red, tau oligomers; blue DAPI, nuclei. Abbreviations: iDCM = idiopathic dilated cardiomyopathy; AD = Alzheimer’s disease.

Functional readout of cardiac tauopathy in mice

The behavioural test in hTau mice showed that spatial learning and memory, tested by Morris Water Maze (see Supplementary material online, Figure S4A and B) and Novelty Y-maze (see Supplementary material online, Figure S4C), were unaffected at 3 and 7 months of age, but became significantly impaired in the 12-month-old mice, while the open field test showed no significant abnormalities for ages and genotypes (see Supplementary material online, Figure S4D). Tau pathology detected with TOMAs showed age-dependent accumulation in hTau mouse brain (see Supplementary material online, Figure S4E).

Different from the brain, progressive myocardial dysfunction was shown to be present in 3-month-old hTau mice by echocardiography (Figure 5A–I; see Supplementary material online, Figure S5 and Table S2) paralleled by progressive accumulation of tau oligomers (Figure 5J and K; Supplementary material online, Figure S6). Systolic function was lower, but non-significant, at all ages in hTau mice, with the worse values at 12 months (Figure 5A and B). Diastolic function (Figure 5C–I) appeared progressively impaired with age in hTau mice when parameters were compiled as defined by the degrees of diastolic dysfunction (see Supplementary material online, Figure S7A–C). The E/A ratio (Figure 5E) was not significantly different; however, it showed an averaged reduced value at 7 months and an increased value at 12 months of age mirrored by changes in mitral valve deceleration time (Figure 5F). This pattern indicates a progressive worsening of diastolic function from grade II to grade III (shown in the insert between Figure 5E and F). E/e′ was increased at 3 months of age while the value progressively increased in the WT to approach that in hTau mice (Figure 5G); the left atrium (LA) was enlarged (Figure 5H) and the mean LA pressure was consistently higher (Figure 5I; see Supplementary material online, Table S2) in 3- and 7-month-old hTau mice. Right ventricular (RV) pressure was severely increased in some mice where it caused paradoxical septal movement (see Supplementary material online, Figure S7D) impeding the accurate haemodynamic measurements of the left ventricular end-diastolic pressure (LVEDP).

Figure 5.

Functional effect and mechanisms of tauopathy in the heart in hTau and WT mice. (A) Representative M-Mode echocardiographic images and (B) quantification of the contractile parameter: fractional shortening (FS). Representative echocardiographic images of (C) pulsed-wave Doppler of the mitral valve (MV) and (D) flow tissue Doppler at the level of the mitral valve annulus. Quantification of the diastolic function parameters: (E) MV E/A ratio; (F) MV deceleration time. An insert with the graphical representation of the progression of the MV flow from normal to grade I to II–III diastolic dysfunction is shown between the data panels (E) and (F); (G) E/e′ ratio; (H) left atrium (LA) dimensions; (I) LA calculated pressure. Data are presented with dotted lines since the data were not collected longitudinally from the same mice. Immunostaining of tau oligomers using TOMA antibodies (green): (J) myocardial tissue of all three age groups and (K) quantitative analysis of the signal. Data were analysed in R using t-test with Welch’s correction for unequal variances. Data are presented as untransformed means ± SEM.

In vivo echocardiographic findings were confirmed in cardiomyocytes isolated from 6-month-old hTau and WT mice. Relaxation was impaired as shown by reduced Ca²⁺ peak amplitude and the significant delayed rise and decay velocities (Figure 6A–D; see Supplementary material online, Figure S8 and Table S3), while cell and sarcomere shortening were lower and cell shortening single exponential tau (the exponential decay time constant) higher, although non-significant (Figure 6E; see Supplementary material online, Table S3). Accumulation of tau oligomers was evident in the cardiomyocytes although it showed an inhomogeneous pattern (Figure 6F and G; see Supplementary material online, Figure S9) that may explain the variability of cardiomyocyte functional values.

Figure 6.

(A–E) Sarcomeres and Ca²⁺ transient parameters indicating diastolic defect in isolated cardiomyocytes from five male/one female WT mice and five male hTau mice. Data were analysed using a linear mixed effects model (LMM) with post-hoc comparison with Holm–Bonferroni correction within outcomes. (F) Increased deposits of tau oligomers stained in green with monoclonal antibody (TOMA) in isolated cardiomyocytes and (G) quantification of the signal. (H) Representative single-plane images of cardiomyocytes stained with a pan-microtubule antibody (cyan) and the nuclear dye Hoechst (grey). The thresholded images showing the microtubule skeleton are displayed below. (I) Quantification of microtubule density (n = 29, 38 non-carrier; n = 3, 4 hTau). (J) Representative single-plane images of cardiomyocytes stained with an antibody specific for detyrosinated (deTyr) microtubules (magenta) and the nuclear dye Hoechst (grey). The thresholded images showing the deTyr microtubule skeleton are displayed below. (K) Quantification of deTyr microtubule density (n = 31, 40 non-carrier n = 3, 4 hTau). (L) Representative single-plane images of the same cardiomyocytes from (C) stained with an antibody specific for tyrosinated (Tyr) microtubules (orange) and the nuclear dye Hoechst (grey). The thresholded images showing the Tyr microtubule skeleton are displayed below. (M) Quantification of Tyr microtubule density (n = 31, 40 non-carrier; n = 3, 4 hTau). Each data point represents the average from a single cell and the box-plot depicts the mean ± SEM. Statistical significance was determined via one-way ANOVA.

Tau oligomer accumulation (see Supplementary material online, Figure S10) was also observed in an established mouse model of cardiac hypertrophy with diastolic disfunction by transverse aortic constriction (TAC).

Dual mechanisms of tau oligomers for cardiac disease

By immunofluorescence, cardiomyocytes isolated from hTau mice showed a preferential reduction in tyrosinated microtubules and a modest, variable reduction in detyrosinated microtubules and total microtubules (Figure 6H–M).

TOMA therapy for cardiac tauopathy

Male and female hTau mice of the same age were injected (once) i.p. with TOMA or IgG control (Figure 7A) as previously published in neurological studies.¹¹ Echocardiography (Figure 7B–G; Supplementary material online, Table S4 and S5) showed no significant changes at the 10-day timepoint, whereas 30 days after treatment diastolic function was improved as shown by the E/e′ ratio, LA dimension, and mean LA pressure. Although the measurements of filling pressure were not significant, the pseudo-normalization in the most severe dysfunction might explain the result. The degree of dysfunction based on the algorithm that includes the E/A ratio, E-wave deceleration time, and LA dimension showed a grade IV diastolic dysfunction in >1/3 of IgG-treated mice compared with TOMA-treated mice where most mice (>1/3) showed grade I/II. The functional improvement was accompanied by a reduction in the accumulation of tau oligomers (Figure 7H, I), and a more ordered distribution of Ca²⁺ handling proteins as exemplified by staining of the ryanodine receptors (RyRs) (Figure 7J), providing the first evidence of anti-proteotoxic mAb therapy for HF.

Figure 7.

Structural and functional myocardial outcomes of TOMA immunotherapy for hTau mice. (A) Injection and readout measurements protocol. (B–D) Representative echocardiographic images of IgG- (six male/six female) or TOMA- (six male/six female) treated 12-month-old hTau mice 1 month after injection. (B) M-mode; (C) pulsed-wave Doppler of the mitral valve flow; (D) tissue Doppler at the level of the mitral valve annulus. (E–G) Diastolic function parameters change by echocardiography: (E) E/e′ ratio; (F) left atrium (LA) dimension; (G) mitral valve deceleration time. Changes from the value at 10 days and 1 month are linked by a straight line since the same mice were followed over time. Data were analysed in R using a linear mixed effects model (LMM) approach. Within- and between-group differences are indicated in each graph. (H) Immunohistochemistry of the IgG- and TOMA-treated hTau mice to visualize total tau (tau5) (blue), tau oligomers (T22) (red); WGA (wheat germ agglutinin; green) defines the cell membrane to localize the intra-/extracellular accumulation of oligomers; (I) quantification of the oligomers; (J) polymerized tubulin (red), and RyRs (green). Data are presented as means ± SEM.

Discussion

Evidence that misfolded protein aggregates accumulate in organs other than the brain and play key pathogenic roles in human disease is steadily emerging.^12–15 Our group previously described misfolded, toxic protein aggregate deposits in HF and AD hearts in human^3,16 and animal models.^17,18 In our prior work, Aβ fibrillar and non-fibrillar pre-amyloid oligomer and amyloid deposit accumulation³ were associated with progressive age-dependent HF with preserved ejection fraction (HFpEF). In this study, we present new evidence that hearts in patients with HF and AD also harbour misfolded, toxic tau aggregates, extending the mechanistic role for toxic protein aggregates in human disease as we previously described for Aβ. Furthermore, our findings mechanistically link hyperphosphorylated tau pathology and loss of tubulin tyrosination with diastolic function in HFpEF and afford novel future potential therapeutic targets (Structured Graphical Abstract). These novel finding supported a shared ‘heart and brain’ human AD and HFpEF pathogenesis.

Tau is present in the myocardium as the HMW isoform

While Aβ has been linked to HF pathogenesis, the role of tau remained unexplored prior to this study. We speculated that tau may play a role in HF pathogenesis given its ubiquitous association with tubulin and the importance of the tau–tubulin interaction in cell microtubule structure and function. Tubulin is a major component of the cytoskeleton present in virtually all eukaryotic cells including cardiomyocytes. As in other cells, the normal tubulin-rich microtubule cytoskeleton confers structural and functional integrity upon cardiomyocytes. The vital role played by tau in promoting tubulin stability and normal microtubule structure and function suggests that a ‘tauopathy’—by either altered tau structure, isoform expression, or post-translational modifications—could induce cardiomyocyte dysfunction via cytoskeletal destabilization or other mechanisms including the deposition of misfolded, toxic tau-enriched aggregates. In this study, we therefore sought to explore ‘tauopathy’ in HF and AD.

Interestingly, the predominant isoform in the myocardium is the HMW tau (100–110 kDa)⁷ (Figure 1). Tau is expressed as multiple isoforms ranging from 45 to 110 kDa due to alternative splicing of the MAPT gene and the resultant presence of a different number of repeats of the microtubule-binding domain in addition to the expression/lack of an 8 kb transcript of exons 4a and 6. The presence of the latter gives rise to the HMW tau. It was previously shown that the extra 250 amino acids added to the N-terminus of the protein increases microtubule spacing, lowering resistance and reducing the energy of transport,¹⁹ and that HMW tau separates the phosphatase activity domain from the microtubule–motor interface, releasing the obstacle to organelle transport. In the myocardium, these functions would result in a more efficient Ca²⁺ handling protein recruitment under stress and enhanced contractile efficiency.

As in the brain, tau oligomer pathology in the heart is linked to hyperphosphorylation

Since tau is highly soluble, its aggregation and polymerization should be unlikely. However, even with low efficiency, tau microtubule-binding repeat domains can facilitate the formation of paired-helical filaments (PHFs)²⁰ containing aggregates of hyperphosphorylated tau. Physiological tau phosphorylation acts as a regulatory mechanism in inhibiting tau’s ability to bind microtubules. When pathological, tau hyperphosphorylation alters protein stability, promoting aggregation and NFT formation.^21,22 As in the AD brain, we found that tau is hyperphosphorylated on Ser³⁹⁶ (which is a marker of mid-stage tau tangles in the brain)²² (Figures 1H and 3A–C) and Ser^396/404 (see Supplementary material online, Figure S2) (although non-significantly) in diseased hearts. Tau was also significantly hyperphosphorylated in the aggregates as shown by the expression in the insoluble fraction (Figure 3B and C) and by the co-localization with pre-amyloid oligomers in diseased cardiomyocytes (Figure 4). Taken together, these findings may indicate that hyperphosphorylated HMW tau is prone to aggregation. This finding is important as it calls for caution with the idea of expanding the size of tau as a therapeutic approach to prevent toxic misfolding, based on the assumption that HMW tau has a lower propensity to polymerize in fibrils²³ proposed to explain the lower impact of tauopathy in the peripheral nervous system. However, in view of the difference in isoform expression, different patterns of phosphorylation in the heart and the brain may confer differential risks of phosphorylated HMW tau aggregation that require further investigation.

Among all the numerous tau phosphosites, p-Ser²⁶² has been reported to be involved in tau aggregation.²⁰ In contrast to other studies reporting hyperphosphorylated tau p-Ser²⁶² in AD brains, in our study tau p-Ser²⁶² was less expressed in the diseased hearts, although non-significantly (Figure 3G and H), and was undetectable in the brains (Figure 3I). Ser²⁶² is positioned in the first repeat, a region that forms the backbone of PHFs, and has been reported to reduce the affinity of tau for microtubules. This would indicate that hyperphosphorylation on this site should favour tau aggregation. p-Ser²⁶² was also reported to be required for the pathogenic interaction between Aβ₄₂ and tau in Drosophila²⁴ and rats,²⁵ potentially driving aggregation. However, p-Ser²⁶² was also described to inhibit tau’s assembly in NFTs, and only weak phosphorylation on this site was previously reported in human brains, with minor progression with increased Braak stage.²⁶ Finally, absence of p-Ser²⁶² was shown to drive tau aggregation in vitro,²⁷ making it unlikely that there is a role for this phosphorylation in driving tau aggregation. Taken together, the various results of these studies highlight the need for further investigations on the significance of this PTM for the onset and progression of tau-related brain and heart diseases. Beyond this, Ser²⁶² is phosphorylated by protein kinase A, which has been previously suggested as a therapeutic target in HF.²⁸ Modulation of protein kinase activity may thus presage an additional future novel therapeutic route for cardiac and brain tau-associated proteotoxicity.

Diastolic dysfunction is the functional phenotype of myocardial tauopathy

Overall, the human data provide indirect evidence that tau may contribute to myocardial dysfunction. To establish a more direct link, although with the limitations of mouse models, we tested the age-dependent molecular–functional link in male and female hTau mice at three ages. While the functional deficit in the brain was delayed compared with accumulation of pre-amyloid oligomers (see Supplementary material online, Figure S4), progressive myocardial dysfunction (Figure 5A–I; see Supplementary material online, Figures S5 and S7, and Table S2) paralleled by progressive pre-amyloid oligomer deposition (Figure 5J, K). As in AD patients,³ hTau mice showed diastolic disfunction also at the single-cell level (see Supplementary material online, Figure S6A–E and S8). Cardiac function in WT mice progressively worsened, approaching the age-matched myocardial dysfunction in hTau. Of note, while AD hearts were not in end-stage HF showing the HFpEF phenotype,³ human iDCM hearts were explanted from patients with end-stage HF at the time of cardiac transplantation. Although HFpEF and heart failure with reduced ejection fraction (HfrEF) have strikingly different phenotypes, the two may co-exist.²⁹ Systolic function may gradually decline in HFpEF and evolve, in some patients, to HFrEF.³⁰ Additionally, left ventricular ejection fraction has been found to be preserved in 38% and reduced in 34% of patients with dementia.³¹ Hence, proteinopathy, as a known defect in ageing and in age-related diseases, may at least in part underlie novel operative mechanisms in HFpEF.

Tau-induced loss of tyrosinated microtubules can disrupt cardiomyocyte homeostasis

Together with tau and other structural microtubule-associated proteins (MAPs), tubulin isoforms, and post-translational modifications, executing the tubulin code, regulate microtubule composition and functional specialization.^32,33 Among the post-translational modifications, tyrosination/detyrosination regulates microtubule stability, dynamics, interactome, and cargo transport.³⁴ Notably, changes in the detyrosination/retyrosination cycle have been recognized as important regulators of cardiac remodelling in response to stress, contributing to diastolic dysfunction in human and animal models of HF.^8,33–36 Here, we found a reduced dynamic microtubule population and tyrosinated microtubules in hTau mouse cardiomyocytes (Figure 6H–M). Due to the long life of cardiomyocytes and neurons, protein must be degraded, cleared, and replaced, requiring an efficient removal of aged and degraded protein and delivery of mRNA and new proteins. This equilibrium relies on the microtubular network for cargo trafficking, including of Ca²⁺ handling proteins, contributing to the uncoupling known to occur in HF.³⁷ Finally, since autophagy-mediated protein clearance is driven by dynein retrograde transport, loss of tyrosination may impair the clearance of protein aggregates, sustaining the accumulation of aggregates in both organs.

Monoclonal antibodies against tau oligomers: a novel therapeutic approach for myocardial proteinopathy and heart failure

mAb therapy is a rapidly expanding modality for the treatment of human diseases. However, several trials using immunotherapy to treat AD failed to improve clinical outcome despite the resolution of fibril accumulation, possibly due to the parallel increase of toxic oligomers.¹⁸ Anti-oligomer mAb has never been tested for HF. Here we show that targeting tau oligomers with TOMA improves diastolic function 1 month after therapy (Figure 7B–G; see Supplementary material online, Tables S4 and S5) and reduces tau oligomer aggregates in a tauopathy mouse model (Figure 7H and I). In view of the current knowledge on the mechanisms of toxicity of oligomers to both neurons and cardiomyocytes via Ca²⁺ dyshomeostasis and the novel discovery of tubulin tyrosination/detyrosination imbalance, TOMA treatment may improve cardiomyocyte function by targeting both the direct (Ca²⁺ fluxes) and indirect (protein trafficking) effect of tauopathy on Ca²⁺ handling, microtubule dynamics, and their interaction.^16,38–41 Further future studies will investigate this important mechanism.

Thus, TOMA treatment acting on the dual pathogenic mechanisms of tauopathy provides the first evidence for anti-proteotoxic immunotherapy for HF.

Study limitations

As detailed in the Supplementary data, we recognize the limitations of the size of our patient cohort and the use of mouse models that carry the limitation of not fully recapitulating the human disease.

Conclusions

HF affects 26 million people worldwide, with more than half suffering from HFpEF. AD affects >44 million people worldwide, with >30% of AD patients also suffering from HFpEF. Our study provides evidence for a new pathophysiological paradigm of ‘heart and brain tauopathy’ operative in both AD and HFpEF. Our findings provide insight into shared functional and structural heart and brain pathophysiological mechanisms and afford a novel therapeutic approach for AD, HFpEF, or their co-existence.

Author contributions

Marco Luciani (Data curation: Lead), Benjamin Prosser (Data curation: Supporting; Methodology: Supporting; Supervision: Supporting; Visualization: Supporting; Writing—review & editing: Equal), Thomas G Di Salvo (Writing—review & editing: Supporting), Russel A. Norris (Supervision: Supporting), José Francisco Abisambra (Supervision: Supporting; Writing—review & editing: Supporting), Onder Albayram (Data curation: Supporting; Writing—review & editing: Supporting), Thamonwan Diteepeng (Graphical Abstract: Supporting), Jordan Morningstar (Data curation: Supporting), Kelsey Moore (Data curation: Supporting), Rakez Kayed (Conceptualization: Equal; Data curation: Equal; Formal analysis: Equal; Methodology: Equal; Writing—review & editing: Equal), Cortney Gensemer (Data curation: Supporting), Justin Kiel (Data curation: Supporting), Helen Butler (Data curation: Supporting), Bethany Jacobs Wolf (Statistical Analysis: Supporting), Gianlorenzo Daniele (Data curation: Equal), Keita Uchida (Data curation: Supporting), Camilla Bacchin (Data curation: Supporting), Luca Troncone (Data curation: Equal), Mauro Montalbano (Data curation: Equal), Stefano Berto (Statistical Contribution: Supporting), and Federica del Monte, MD, PhD (Data curation: Lead; Formal analysis: Lead; Funding acquisition: Lead; Methodology: Lead; Resources: Lead; Supervision: Lead; Writing—original draft: Lead; Writing—review & editing: Lead)

The work was performed at Beth Israel Deaconess Medical Center—Massachusetts General Hospital and Harvard Medical School in Boston (MA), and the Medical University of South Carolina, Charleston (SC)

Supplementary data

Supplementary data are available at the European Heart Journal online.

Data availability

The data underlying this article are available in the article and in the online supplementary data.

Funding

The American Heart Association [17CSA33620007 and 20SRG35540029 to F.d.M., R.K., and J.F.A.]; National Institute of Health [1R01HL156116-01 to F.d.M. and AG054025 and AG055771 to R.K.]; Christie Heart and Brain Program and MUSC (discretional funds) to F.d.M.; Mitchell Center for Neurodegenerative Diseases to R.K.; Forschungskredit University of Zurich [FK-19-043 to M.L.]; Italian Society of Cardiology Fellowship supported by Merck Sharp & Dohme Italia to M.L.; Medical University of South Carolina Scientist Training Program [5T32 GM132055 to H.B.]; and Alzheimer’s Association Fellowship Program and MUSC (discretional funds) to OA.