Defining new mechanistic roles for αII spectrin in cardiac function

Spectrins are cytoskeletal proteins essential for membrane biogenesis and regulation and serve critical roles in protein targeting and cellular signaling. αII spectrin (SPTAN1) is one of two α spectrin genes and αII spectrin dysfunction is linked to alterations in axon initial segment formation, cortical lamination, and neuronal excitability. Furthermore, human αII spectrin loss-of-function variants cause neurological disease. As global αII spectrin knockout mice are embryonic lethal, the in vivo roles of αII spectrin in adult heart are unknown and untested. Here, based on pronounced alterations in αII spectrin regulation in human heart failure we tested the in vivo roles of αII spectrin in the vertebrate heart. We created a mouse model of cardiomyocyte-selective αII spectrin-deficiency (cKO) and used this model to define the roles of αII spectrin in cardiac function. αII spectrin cKO mice displayed significant structural, cellular, and electrical phenotypes that resulted in accelerated structural remodeling, fibrosis, arrhythmia, and mortality in response to stress. At the molecular level, we demonstrate that αII spectrin plays a nodal role for global cardiac spectrin regulation, as αII spectrin cKO hearts exhibited remodeling of αI spectrin and altered β-spectrin expression and localization. At the cellular level, αII spectrin deficiency resulted in altered expression, targeting, and regulation of cardiac ion channels NaV1.5 and KV4.3. In summary, our findings define critical and unexpected roles for the multifunctional αII spectrin protein in the heart. Furthermore, our work provides a new in vivo animal model to study the roles of αII spectrin in the cardiomyocyte.

Cardiovascular disease is the number one cause of mortality in the United States, accounting for ϳ31% of all deaths (1). Although significant initiatives in both prevention and therapeutics have reduced cardiovascular mortality, the field lacks effective solutions for a host of cardiovascular phenotypes. Furthermore, the literature is just beginning to address the striking complexity of organ, cell, and molecular pathways linked with human cardiovascular disorders. Thus, there remains a clear need to understand the fundamental cellular and molecular pathways associated with normal cardiac function as well as dysfunction of critical pathways in cardiovascular disease.
Cardiac function requires finely tuned integration of myocyte mechanical and electrical pathways. Membrane-associated ion channels, transporters, receptors, signaling proteins, and cell adhesion molecules are important for cardiac function. The organization of these proteins within the vertebrate myocyte is essential for normal excitation-contraction coupling. In fact, decades of research have illustrated the role of the submembrane cytoskeleton and cytoskeletalassociated proteins in excitable cell biology (2)(3)(4)(5)(6)(7). Moreover, in the heart alone, defects in myocyte cytoskeletal proteins have been linked with a host of structural and electrical phenotypes in human cardiovascular disease as well as in animal disease models (4).
Spectrins are cytoskeletal proteins, initially identified through their roles in erythrocyte structure and flexibility (8 -10). In diverse tissues, spectrins act as an essential link between the actin-based cytoskeleton and membrane, serving to maintain cellular structure and polarity while providing flexibility and strength. In excitable cells and nonexcitable cells, spectrins form complexes with ankyrins and membrane-associated proteins, playing essential roles in the localization and regulation of critical ion channels and transporters (11)(12)(13)(14). In humans, there are five ␤ but only two ␣ spectrin genes (15). SPTAN1 encodes nonerythrocytic ␣II spectrin, which forms functional heterotetramers with ␤ spectrins (16). ␣II spectrin has been extensively studied in neurons due in part to its link to human neurologic disease. Variants in SPTAN1 are linked to early infantile epileptic encephalopathy, type 5 (EIEE5) 2 or West Syndrome, characterized by refractory seizures, intellectual arrest/regression, agenesis of the corpus callosum, and hypomyelination, carrying a poor prognosis (11,17,18). Beyond the brain, recent work has defined roles for ␤-spectrins in cardiac structure, excitability, and signaling and linked dysfunction in ␤ spectrin pathways with both acquired and congenital forms of human cardiovascular disease (19,20). In contrast, the number of studies on ␣II spectrin in heart are limited (21)(22)(23)(24)(25)(26), and the in vivo roles of cardiac ␣ spectrins are essentially unstudied.
Here, we report dysregulation of ␣II spectrin in human heart failure. Moreover, using a newly engineered model of cardiomyocyte-specific deletion of ␣II spectrin, we illustrate the role of ␣II spectrin in normal cardiac physiology. ␣II spectrin cKO mice display both structural and electrical phenotypes at baseline that are intensified by physiological and pathological stress. Furthermore, we illustrate striking post-transcriptional roles for ␣II spectrin for global spectrin family protein regulation and unanticipated roles for ␣II spectrin in ion channel expression, targeting, and regulation. Together, our findings define critical in vivo roles for ␣II spectrin in the stabilization of the cardiomyocyte spectrin network, ion channel regulation, cardiac electrophysiology, structure, function, and stress response.

␣II spectrin is dysregulated in human heart failure
To evaluate potential dysregulation of ␣II spectrin in human cardiovascular disease, we performed immunoblot experiments on failing and nonfailing human left ventricle. Spectrins are cleaved by both calpain and caspase, forming specific degradation products (27). Specifically, ␣II spectrin is proteolytically cleaved by both caspase and calcium-activated calpain, forming ϳ120and ϳ150-kDa degradation products, respectively (27,28). Notably, the ϳ150-kDa (calpain-mediated) cleavage product of ␣II spectrin was significantly increased in both ischemic and nonischemic heart failure (HF) samples compared with samples from nonfailing hearts (Fig. 1, A-D). Furthermore, levels of the ϳ120-kDa (caspase-mediated) cleavage were increased in human HF samples compared with samples from nonfailing hearts (p Ͻ 0.05, n ϭ 3 nonfailing, 6 HF). Levels of full-length ␣II spectrin were similar between samples likely due to the observed trend toward a significant transcriptional up-regulation of ␣II spectrin in disease (Fig. S1). In summary, cardiac ␣II spectrin degradation products are significantly increased in human heart failure.

Creation and validation of in vivo model of cardiomyocyte ␣II spectrin deficiency
Based on the critical role of spectrins in excitable cells and dysregulation in heart failure ( Fig. 1), we investigated the role of ␣II spectrin for in vivo cardiac function. To date, due to the embryonic lethality of global ␣II spectrin knockout mice (29), the role of ␣II spectrin in the adult heart is unknown and unstudied. Thus, we generated a cardiomyocyte-specific ␣II spectrin mouse knockout model (␣II spectrin cKO) using the Cre-lox system in which exon 8 of Sptan1 is flanked by LoxP sites (30) and the expression of Cre recombinase is driven under the ␣-myosin heavy chain promoter ( Fig. 2A) (31).
␣II spectrin cKO mice are viable, showed no gross visible phenotypes, and reproduced normally. As expected, cardiomyocyte preparations from ␣II spectrin cKO mouse myocytes displayed near-complete loss of ␣II spectrin protein expression (Fig. 2, B-D). This loss was selective for cardiomyocytes as ␣II spectrin protein levels were similar between control and ␣II spectrin cKO cerebellum (Fig. 2, E-G) and skeletal muscle. In the control mouse heart, ␣II spectrin is highly expressed and localizes to the Z-disc, lateral membrane, and intercalated disc (Fig. 2, H and I). Notably, this intracellular distribution does not completely overlap with ␣I spectrin (localized to the Z-disc and lateral membrane, absent from the intercalated disc (25)) suggesting potential unique roles for ␣II spectrin versus ␣I spectrin. Consistent with immunoblot data, ␣II spectrin cKO mice displayed complete loss of ␣II spectrin in the cardiomyocyte by immunofluorescence (Fig. 2J). Thus, the newly generated knockout mouse line provides a viable model to study in vivo ␣II spectrin deficiency in cardiomyocytes.

␣II spectrin cKO mice display cardiac dysfunction
To test the in vivo role of ␣II spectrin in vertebrate heart, we performed detailed cardiac structural and functional phenotyp- representative, and C and D, quantification of immunoblots of human control (nonfailing) and ischemic (IHF) or nonischemic (NIHF) failing left ventricular tissue for ␣II spectrin, normalized to GAPDH. C, levels of full-length ␣II spectrin (black arrowhead), are unchanged, whereas for D, levels of the 150-kDa degradation product (red arrowhead) are increased in ischemic and nonischemic failing human LV (n ϭ 6, 6, 6 for NF, IHF, NIHF LV, respectively, p value Ͻ0.0001 for the 150-kDa degradation product). Values are represented as mean Ϯ S.E.

␣II spectrin regulates cardiac function
ing of ␣II spectrin cKO mice and control littermates. In adult mice (20 -24 weeks of age), we observed a significant decrease in stroke volume, ejection fraction, and fractional shortening, indicative of early stages of heart failure and/or cardiac remodeling (Fig. 3, A-F). Furthermore, ␣II spectrin cKO mice displayed cardiomyocyte hypertrophy, as assessed by histologic analysis of cross-sectional cardiomyocyte size (Fig. 3L, Fig. S2). However, this did not translate into gross increases in heart weight, even when normalized to body weight or to tibia length (Fig. S3). Histologically, we observed increased cardiac fibrosis in ␣II spectrin cKO mice, as assessed by automated quantification of Masson's Trichrome-stained heart samples (Fig. 3, I-K). Fibrosis was not accompanied by an increase in apoptosis, as assayed by TUNEL staining (Fig. S4). Notably, this dysfunction develops with age, as ␣II spectrin hearts from younger mice (12-16 weeks of age) displayed no significant changes in cardiac structure or function and no fibrosis, necrosis, or hypertrophic phenotypes (Fig. S5). Thus, ␣II spectrin deficiency promotes an age-dependent decline in cardiac ejection fraction and increased cardiac fibrosis. A, strategy for cardiomyocyte-specific knockout of ␣II spectrin in mice. B-G, representative and quantification of immunoblots for ␣II spectrin normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in isolated adult ventricular myocytes and cerebellum from ␣II spectrin cKO and control mice. Results demonstrate (B-D) reduction of full-length (Ͼ250 kDa) and degraded (ϳ150 kDa) ␣II spectrin from isolated adult ventricular myocytes from ␣II spectrin cKO mice compared with control mice (n ϭ 8.5 for control and ␣II spectrin cKO mice, respectively, p Ͻ 0.0001). E-G, no differences in ␣II spectrin levels were observed in cerebellum. Values are represented as mean Ϯ S.E. H-J, staining of ␣II spectrin (red) and desmin (green) in adult cardiomyocytes isolated from (H) control and (J) ␣II spectrin cKO mice. I, magnified image of control myocytes demonstrates localization of ␣II spectrin at lateral membrane (white arrows), Z-lines (white arrowheads), and intercalated disc (yellow arrowheads). Scale bars ϭ 20 m.

␣II spectrin regulates cardiac function ␣II spectrin cKO mice display in vivo electrical dysfunction
We next tested the impact of ␣II spectrin deficiency on cardiac excitability. We observed electrical dysfunction in ␣II spectrin cKO mice, regardless of age. In young 12-16 -weekold mice, prior to structural dysfunction, we observed ECG phenotypes in ␣II spectrin cKO mice compared with control mice including minor, but significant increases in P wave and QRS duration by conscious recordings (Table S1; no change in QRS by DSI telemetry), indicating slowed conduction through the atria and ventricles. Notably, this slowed conduction is not associated with reduced levels or altered localization of connexin 43 (Fig. S6).
Treatment of conscious ␣II spectrin cKO, but not control animals with epinephrine revealed significant differences in heart rate as well as PR, QT, and rate corrected QT (QTc) intervals when compared with control littermates (Table S1). Furthermore, epinephrine-treated ␣II spectrin cKO mice (but not control littermates) displayed arrhythmia phenotypes including sustained runs of premature ventricular contractions (PVCs) (Fig. 4, A-C). In summary, 12-16 -week-old ␣II spectrin cKO mice display defects in electrical activity that precede cardiac structural and functional phenotypes. Furthermore, electrical phenotypes are increased by catecholamine treatment.

␣II spectrin is required for normal myocyte excitability
Based on in vivo ␣II spectrin cKO mouse phenotypes, we investigated electrical properties at the individual myocyte level. Compared with control myocytes, we observed statistically significant changes in action potential duration (APD) at 50, 75, and 90% repolarization (APD 50 , APD 75 , and APD 90 , respectively) or in maximum upstroke velocity (dV/dT) in ␣II spectrin cKO myocytes (Fig. S7, A-C; 0.5 Hz). Furthermore, we observed a statistically significant increase in action potential amplitude (APA) in ␣II spectrin cKO myocytes compared with control cells (Fig. S7D). These changes occurred in the absence of changes in resting membrane voltage (Fig. S8). Notably, we observed a failure of ␣II spectrin cKO myocytes to initiate an action potential at higher stimulation frequencies (Fig. S7, E-G). For example, unlike control myocytes that initiated normal action potentials at frequencies up to 10 Hz, ␣II spectrin cKO myocytes began failing to excite at just 2 Hz, indicating a ␣II spectrin cKO mice. D, wall thickness is unchanged, whereas C, stroke volume; E, fractional shortening; and F, ejection fraction are decreased in ␣II spectrin cKO mice when compared with littermate controls (n ϭ 10 and 6 for control and ␣II spectrin cKO mice, respectively). G-L, ␣II spectrin cKO mice exhibit cellular hypertrophy and increased fibrosis compared with littermate controls. G and H, representative images of cardiac structure in control (G) and ␣II spectrin (H) cKO mice. Scale bar ϭ 2 mm. I and J, representative Masson's trichrome-stained cardiac tissue from control (I) and ␣II spectrin (J) cKO mice demonstrating increased fibrosis (K). Scale bars ϭ 100 m. L, ␣II spectrin cKO mice have increased average cross-sectional myocyte size (n ϭ 5 and 4 for control and ␣II spectrin cKO mice, respectively). All values are represented as mean Ϯ S.E. * indicates a statistical differences from controls, p Ͻ 0.05.
␣II spectrin regulates cardiac function key role of ␣II spectrin in myocyte excitability (Fig. S7G). In summary, ␣II spectrin cKO myocytes display multiple electrical phenotypes.

Defining ␣II spectrin pathways in heart: ␣II spectrin is required for cardiac spectrin family regulation
Our findings support significant in vivo and cellular defects in the absence of cardiac ␣II spectrin. We therefore directly tested the impact of ␣II spectrin deficiency on spectrin family regulation in heart. Consistent with our hypothesis, ␤ spectrin proteins (obligate ␣ spectrin-binding partners) were significantly reduced in ␣II spectrin cKO cardiomyocytes. We observed a 77.9% reduction in ␤I spectrin protein levels with ␤II spectrin protein levels reduced 93.2% (Fig. 5, A-D). Remaining ␤II spectrin was normally localized (Fig. 5H). Normal mRNA expression of ␤I (Sptb) and ␤II (Sptbn1) spectrin ( Fig. 5G) support that altered protein levels are due to post-transcriptional regulation of ␤ spectrins. ␣I spectrin, a protein unique to mammals, is structurally similar to ␣II spectrin, having likely arisen from a gene duplication of ␣II spectrin (32). Notably, in cardiomyocytes lacking ␣II spectrin, ␣I spectrin is strongly up-regulated at both transcript ( Fig. 5G) and protein levels (Fig. 5, E and F).
In control cardiomyocytes, ␣I spectrin localizes to the Z-disc and lateral membrane, and is absent from the intercalated disc (Fig. 5, I and J). However, in ␣II spectrin cKO cardiomyocytes, ␣I spectrin is present at the intercalated disc (Fig. 5, I and J), supporting a putative compensatory role of ␣I spectrin in the ␣II spectrin cKO model. In summary, our data support a key role of ␣II spectrin expression for the normal regulation of both ␤I spectrin and ␤II spectrin expression in cardiomyocytes. Moreover, our findings support that ␣I spectrin levels are significantly increased in ␣II spectrin cKO heart as a potential compensatory mechanism to preserve cardiac function. However, our data suggest that this compensatory pathway is insufficient to restore normal function to ␣II spectrin cKO mice, potentially due to the inability of ␣I spectrin to associate with cardiac ␣II spectrin-binding partners.

␣II spectrin is required for cardiac Na V 1.5 expression and function
Spectrin dysfunction is linked with altered voltage-gated Na v channel regulation in the heart and brain (2, 3, 6, 33-35). We therefore tested the impact of ␣II spectrin-deficiency on cardiac Na v 1.5 regulation. Na v 1.5 expression was significantly reduced in ␣II spectrin cKO hearts compared with hearts from control littermates (Fig. 6, A and B; reduced 59.5% when normalized for total protein expression). This decrease was likely due to post-translational dysregulation, as mRNA levels of Scn5a (encoding Na V 1.5) were unchanged between control and ␣II spectrin hearts (Fig. 6C). In the control heart, Na v 1.5, whereas present at multiple membrane domains, is preferentially expressed at the intercalated disc membrane (Fig. 6D). In ␣II spectrin cKO hearts, we observed modestly decreased intercalated disc staining relative to lateral membrane expression of Na v 1.5 (Fig. 6, D and E). However, neither immunoblot nor immunofluorescent microscopy are quantitative measures of functional Na V 1.5. To determine the function of the remaining Na V 1.5, analysis by single-cell electrophysiology was conducted, which revealed nearly a 2-fold decrease in peak I Na in ␣II spectrin cKO myocytes compared with littermate controls (Fig. 7, A and B). However, we observed no difference in voltagedependent inactivation or time-dependent recovery from inactivation between myocytes from control and ␣II spectrin cKO hearts ( Fig. 7, C and D). In summary, ␣II spectrin is required for normal Na v 1.5 expression, localization, and function in mouse heart.

Identification of putative ␣II spectrin-dependent cardiac pathways
Reduced Na v 1.5 (I Na ) in ␣II spectrin cKO hearts (Figs. 6 and 7) supports the observed decrease in ␣II spectrin cKO myocyte excitability (Fig. S7, E-G), but is striking in the context of a normal dv/dt max . As assessed by an unbiased partial leastsquares regression analysis using the Hund-Rudy action potential model (36 -39), Na v 1.5-dependent dysregulation alone is unlikely to produce observed experimental data from control and ␣II spectrin cKO action potential measurements (e.g. APD, APA, and dv/dt max ). Our computational analyses strongly predicted secondary current alterations in ␣II spectrin cKO myocytes, most specifically associated with several repolarizing potassium currents.
To support the computational model, we performed transcriptional analysis of mRNA expression of ion channel subunits in 12-16 -week-old ␣II spectrin cKO versus control littermate hearts. In support of the modeling predictions, we observed significant changes in ion channel subunits involved ␣II spectrin regulates cardiac function in cardiac repolarization, particularly potassium channel ␣and ␤-subunits (Fig. S9). Analysis of mRNA expression of Kcnd2 (encoding K V 4.2) and Kcnd3 (encoding K V 4.3), major contributors to I K,peak, I TO,fast , revealed normal expression of Kcnd3, but a 0.57-fold decrease in Kcnd2 expression in ␣II spectrin cKO hearts (Fig. S9). Additionally, we observed a downregulation of K V 1.7 (Kcna7), a potassium channel subunit that modulates both I Kur and I TO in ␣II spectrin cKO hearts. Although we did not observe a difference in the resting membrane potential of isolated ␣II spectrin cKO cardiomyocytes (Fig. S8), there was a significant down-regulation of K ir 2.1 (Kcnj2) in ␣II spectrin cKO hearts compared with controls. Furthermore, we observed increased expression of the twopore potassium channel TREK1 (Kcnk2), but not TASK1 (Kcnk3) (Fig. S9) in ␣II spectrin cKO hearts. Similarly, GIRK1 (Kcnj3) expression was decreased, whereas GIRK4 (Kcnj5) expression was unchanged in ␣II spectrin cKO hearts. Kcne1, linked with long QT syndrome type 5 (LQT5) showed increased expression, whereas Kcnq1, which is linked with long QT syndrome type 1, is unchanged in ␣II spectrin cKO hearts. Finally, Kcnip2 (Kchip2), Kcnb1 (contributing to I K,SS ), Kcna5 (contributing to I Kur ), Kcnj8 (K ir 6.1), Kcnj11 (K ir 6.2), Kcnj12 (K ir 2.2), and Kcnj14 (K ir 2.4) were all similarly expressed between control and ␣II spectrin cKO hearts (Fig. S9). Thus, whereas we observed dysregulated expression of many potassium channels, many remained unchanged, demonstrating a selective disruption of potassium channel subunit expression in ␣II spectrin cKO hearts.
Beyond potassium channel subunits, we observed alterations in other ion channel subunits. Although the mRNA expression Figure 5. Loss of cardiac ␣II spectrin results in broad spectrin dysregulation. A, C, and E, representative, and B, D, and F, quantification, of immunoblots of cardiomyocytes for ␤I, ␤II, and ␣I spectrin (n ϭ 7 and 5 for control and ␣II spectrin cKO, respectively, p value Ͻ0.0001, Ͻ0.0001, ϭ 0.043 for ␤I, ␤II, and ␣I spectrin, respectively). G, mRNA expression of Sptb, Sptbn1, and Spta1 in adult cardiac tissue (n ϭ 8 and 8, p ϭ 0.57). H-J, immunofluorescence staining of ␤II (H) and ␣I spectrin and plakoglobin (I and J) in ventricular tissue from control and ␣II spectrin cKO mice demonstrating increased intercalated disc localization of ␣I spectrin (yellow arrowhead) in ␣II spectrin cKO tissue compared with controls. Scale bar ϭ 20 m. J, magnified images of intercalated discs. Scale bar ϭ 5 m.
␣II spectrin regulates cardiac function of the ␣-subunit of Na v 1.5, Scn5a, was unchanged, the expression of the ␤1-subunit, Scn1b, was up-regulated ϳ2-fold in ␣II spectrin cKO hearts (Fig. S9). Furthermore, we noted an increase in the expression of two ␣ subunits of the L-type calcium channel (Cacna1s and Cacna1c) accompanied by an upregulation of the ␤1 (Cacnb1) and ␤3 (Cacnb3) subunits in ␣II spectrin cKO hearts (Fig. S9). In summary, consistent with computational modeling, transcriptional analysis of ␣II spectrin cKO hearts illustrates alterations in multiple unexpected cardiac ion channels, with a notable association with cardiac K ϩ channels.

␣II spectrin cKO myocytes display altered repolarizing potassium currents
Based on modeling and transcript analyses, we hypothesized that select potassium currents may be altered in ␣II spectrin cKO hearts. In line with our data showing no difference in resting membrane voltage (Fig. S8), we observed no change in I K1 between genotypes (Fig. 8, A-C). We did, however, observe significant decreases in I TO at both peak (I TO,peak ) and steadystate potassium currents (I TO,SS ) in ␣II spectrin cKO mice compared with littermate controls (Fig. 8, D-F). In further support of a role for ␣II spectrin in potassium channel regulation, we observed a significant increase in protein expression of K V 4.3 (Fig. 9, A and B). Of note, the increased expression of K V 4.3 in ␣II spectrin cKO mice was accompanied by more diffuse localization of K V 4.3 when compared with control mice (Fig. 9, C and D), suggesting a potential role for ␣II spectrin in K V 4.3 trafficking and/or localization. In summary, these data support an unexpected role for ␣II spectrin in regulation of cardiomyocyte potassium currents. Future experiments will be critical to define the mechanistic relationship for this regulation (direct versus compensatory due to I Na loss).

␣II spectrin associates with Na V 1.5 and K V 4.3
To investigate the relationship of ␣II spectrin with ion channel subunits underlying I Na and I TO , we examined potential association between ␣II spectrin and Na V 1.5 and K V 4.3. We observed association of ␣II spectrin with both Na V 1.5 and K V 4.3 in co-immunoprecipitation experiments from detergent-soluble lysates of adult mouse heart (Fig. S10, A and B). In contrast, we observed no association of the structurally similar ␣I spectrin with Na V 1.5 or K V 4.3 in parallel co-immunoprecipitation experiments (Fig. S10, C and D). Thus, in addition to supporting association of ␣II spectrin with multiple ion channel subunits, these new data suggest selectivity of ␣ spectrin polypeptides for myocyte membrane targets. Furthermore, these data suggest that ␣I spectrin may not completely compensate for loss of ␣II spectrin in the ␣II spectrin cKO mouse model. Future experiments will be important to define the ␣II spectrin regulates cardiac function mechanisms (direct versus indirect) and structural requirements for ␣II spectrin association with Na v 1.5 and K v 4.3.

Key cardiac regulatory pathways are altered in ␣II spectrin hearts
As heart failure and arrhythmia are complex pathways involving the interplay between structural, membrane, signaling, and transcriptional pathways, we investigated ␣II spectrindependent transcriptional pathways in heart in 12-16 -weekold mice, prior to cardiac electrical or structural remodeling using RNAseq (Fig. S11). Using GO Pathway analysis, we identified significantly altered regulation of many pathways, including several that may be relevant to the observed phenotypes of ␣II spectrin cKO mice, such as extracellular matrix organization, cell adhesion, voltage-gated channel activity, and collagen metabolic process. We focused on the dysregulation in two major transcript classes. First, several cardiac intracellular proteases including calpains and caspases, known to regulate neuronal and cardiac spectrins (24,27,40,41), were significantly altered (Fig. S12A). Second, we observed significant alterations in transcripts associated with extracellular matrix (ECM) organization and remodeling (Fig. S12B). For example, we observed significant transcriptional up-regulation of TIMP1 (over 12fold) that is essential for membrane metalloprotease regulation. Additionally, increased RNA expression of many collagens (Fig.  S12C) is consistent with increased fibrosis and with increased ECM turnover. Thus, consistent with a critical role of ␣II spectrin for cardiac remodeling, ␣II spectrin cKO mice display significant alterations in key pathways that promote cardiac protein turnover and ECM remodeling, thus promoting heart failure phenotypes. Furthermore, established markers of car-diac hypertrophy were significantly altered, foreshadowing the increased cardiomyocyte size observed in older (20 -24 -weekold) mice (Fig. S13). We hypothesize that these changes serve to augment and/or accelerate phenotypes in ␣II spectrin cKO mice.
␣II spectrin mice display mortality and severe cardiac phenotypes in response to stress ␣II spectrin levels are dysregulated in human HF. Moreover, ␣II spectrin cKO mice display significant structural and electrical phenotypes at baseline and in response to catecholaminergic stress and aging. Furthermore, in addition to observing alterations in critical global spectrin pathways in ␣II spectrin mice, we also observed alterations in pathways that would favor structural and electrical remodeling. We used a well-validated model for afterload-induced heart failure to test the hypothesis that ␣II spectrin cKO mice would display accelerated disease phenotypes in response to stress. When using our standard protocol that produces heart failure in 6 -8 weeks with no early mortality (5), ϳ75% of ␣II spectrin cKO mice died within 2 weeks of transverse aortic constriction (TAC).
Based on the high mortality of ␣II spectrin cKO mice following standard TAC, a less severe model of pressure overload (using a 25-gauge constriction) was performed. Consistent with a critical role for ␣II spectrin in normal cardiac function, ␣II spectrin cKO mice displayed an accelerated heart failure phenotype when compared with control mice following TAC (Fig.  10). Notably, ␣II spectrin cKO TAC mice exhibit accelerated and decreased contractility (decreased ejection fraction, Fig.  10A) and cardiac dilation (increased left ventricular internal diameter (LVID) (Fig. 10B)), without compensatory hypertro- Figure 7. Cardiac ␣II spectrin is essential for I Na . A, representative recordings of whole cell I Na from cardiomyocytes isolated from 12-16 -week-old control and ␣II spectrin mice. B, current-voltage relationship of I Na in ␣II spectrin cKO and control ventricular myocytes. C, voltage-dependent inactivation, and D, time-dependent recovery, of I Na in ␣II spectrin cKO and control adult ventricular myocytes. No significant differences were observed in V1 ⁄2 as determined by Boltzmann fits of the steady-state voltage-dependent inactivation or time-dependent recovery (n ϭ 3 and 3; n ϭ 11 and 11 for control and ␣II spectrin cKO, respectively). Values are represented as mean Ϯ S.E. * indicates a statistical differences from controls.
␣II spectrin regulates cardiac function phy (decreased left ventricular posterior wall (LVPW) thickness (Fig. 10C)) compared with control TAC littermates. Furthermore, compared with control TAC littermates, ␣II spectrin cKO TAC mice displayed more pronounced ECG changes, including QT and QTc prolongation, and T-wave depression (Fig. 10, F and G, Table S2). Additionally, histologic analysis with automated quantification of fibrosis in hearts of ␣II spectrin cKO and control mice indicated increased fibrosis in ␣II spectrin cKO hearts following TAC (Fig. S14, A-C). TUNEL staining reveled a large, but statistically insignificant (p ϭ 0.0689) increase in apoptosis in ␣II spectrin cKO mice (Figs. S14, G-I, and S15), suggesting that the observed fibrosis may be replacement fibrosis. Finally, increased vacuolization of cardiac tissue was observed in ␣II spectrin cKO mice but not control mice (Fig. S14, B, C, E, and F). Notably, these changes occur in the absence of generalized disruption of the cardiomyocyte sarcomere or intercalated disc structure (Fig. S16). In summary, ␣II spectrin is required for normal physiologic cardiac remodeling, as loss of ␣II spectrin results in accelerated heart failure phenotypes in response to standard experimental models of heart failure and hypertrophy.

Discussion
Spectrins are broadly expressed, forming a submembrane network with actin essential for membrane organization, flexibility, and stability (8 -10). Furthermore, recent work has demonstrated roles for spectrins in the trafficking, regulation, and stabilization of membrane-associated receptors and channels in diverse tissue types (12,13,20). Here, we define new roles of ␣II spectrin in normal cardiac function, in vivo. Based on obser- Figure 8. ␣II spectrin cKO myocytes display altered I TO . A-C, representative recordings of whole cell I K1 from cardiomyocytes isolated from control (A) and ␣II spectrin (B) cKO mice at 12-16 weeks of age. C, current-voltage relationship, of I K1,peak and I K1,SS in ␣II spectrin cKO and control ventricular myocytes (n ϭ 4 and 2; n ϭ 16 and 6 for control and ␣II spectrin cKO, respectively). D-F, representative recordings of whole cell I TO from control (D) and ␣II spectrin (E) cKO cardiomyocytes. F, current-voltage relationship of I TO,peak and I TO,SS in ␣II spectrin cKO and control ventricular myocytes from 12-to 16 -week-old mice (n ϭ 3 and 3; n ϭ 10 and 13 for control and ␣II spectrin cKO, respectively). Values are represented as mean Ϯ S.E. * and # indicate a statistically significant difference between genotypes at a given voltage in I TO,peak and I TO,SS , respectively.

␣II spectrin regulates cardiac function
vations of striking alterations in ␣II spectrin regulation in human heart failure and the lack of knowledge about the role of cardiac ␣II spectrin, we tested the impact of ␣II spectrin deficiency on the vertebrate heart. We establish that ␣II spectrin is required for normal cardiac structure and function, as adult ␣II spectrin cKO mice display reduced contractility. Histologically, ␣II spectrin cKO mice display simultaneous increases in cellular hypertrophy and fibrosis. Furthermore, unlike control mice, ␣II spectrin cKO mice displayed striking mortality in response to standard TAC protocols and significant remodeling and tissue damage in response to a less severe pressure-overload protocol. Underlying this dysfunction, ␣II spectrin cKO mice demonstrate a trend toward an increase in apoptosis and increased fibrosis following TAC. In addition to structural phenotypes, ␣II spectrin cKO mice display electrical phenotypes at baseline that are exacerbated in response to catecholamines. At the molecular level, we illustrate a central role of ␣II spectrin for the dynamic regulation of global ␣ and ␤ spectrin expression and localization in vivo. Furthermore, we illustrate a critical role of ␣II spectrin in the expression and targeting of Na V 1.5 and K V 4.3, resulting in reductions of I Na and I TO in the heart. Finally, we define the impact of ␣II spectrin expression on upstream spectrin regulatory pathways. These findings provide new insight of the central nodal role of ␣II spectrin in the formation and regulation of key structural, electrical, and signaling pathways in heart.

␣II spectrin regulates global cardiac spectrin pathways
␣II spectrin is abundantly expressed in the heart where it forms heterotetramers with ␤I, ␤II, and ␤IV spectrin. Impor-tantly, the role of cardiac ␣II spectrin, specifically, is relatively unexplored. Given the essential role of ␣II spectrin in excitable cells (21-26, 30, 42), in vivo roles of ␣II spectrin were anticipated. However, the severity of in vivo and in vitro ␣II spectrin cKO phenotypes were unexpected, particularly the impact of ␣II spectrin loss on both ␣ and ␤ spectrin expression. For most functions, the spectrin heterotetramer is the functional unit, with the presence of both ␣ and ␤ spectrin being required for spectrin stability (29). Thus, work on the in vivo role of ␤ spectrins provides insight into the function of ␣ spectrins. Cardiomyocyte-specific ␤II spectrin-deficient (␤II spectrin cKO) mice experience a variety of electrophysiological abnormalities, including increased heart rate variability, atrioventricular block, prolonged QT intervals, and widened QRS complexes at baseline, along with pronounced ventricular arrhythmias and death following catecholaminergic stress (19). Similar to ␣II spectrin cKO mice, ␤II spectrin cKO mice displayed accelerated heart failure phenotypes following transverse aortic constriction. The near absence of ␤II spectrin in the ␣II spectrin model (that did not display such striking electrical phenotypes) supports that remodeling of other cardiac pathways is sufficient to mitigate more severe phenotypes.

␣II spectrin relationship with ␣I spectrin
Cardiac ␣II spectrin localizes to the Z-disc, lateral membrane, and intercalated disc. Alternately, ␣I spectrin is localized only to the Z-disc and lateral membrane, leaving ␣II spectrin as the only ␣ spectrin at the intercalated disc. Although we hypothesized that this would lead to deficits in intercalated disc organization and function in ␣II spectrin cKO mice, we were ␣II spectrin regulates cardiac function surprised to observe translocation of ␣I spectrin to the intercalated disc in the absence of cardiac ␣II spectrin (Fig. 5). This change in localization, along with a somewhat mild phenotype at baseline suggests functional redundancy among cardiac ␣ spectrins, and a partial compensation by ␣I spectrin for ␣II spectrin loss. This is, perhaps not surprising, as only mammals have two ␣ spectrin genes, ␣I having arisen from a gene duplication in terrestrial vertebrates (43). However, it is apparent from our characterization of ␣II spectrin cKO mice that this compensation is not complete. Although young, unchallenged mice are able to maintain normal cardiac function and conduction, age, physiologic (catecholamines) and pathologic (TAC) stress resulted in dysfunction. Although compensatory dysregulation of the spectrin cytoskeleton and opposing reductions of currents in the heart are able to preserve cardiac function at baseline, even somewhat subtle disruptions (age, mild pressure overload) resulted in a failure to maintain cardiac function, resulting in reduced contractility and arrhythmia. This is likely due to the relative inability of ␣I spectrin to associate with ␣II spectrin targets.
When cardiac ␤II spectrin is knocked out, there is an up-regulation of other ␤ spectrins, whereas ␣ spectrins are decreased.
The opposite is observed here, with loss of ␣II spectrin causing an increase in ␣I spectrin and a decrease in ␤ spectrins. Based on these observations and on the nature of ␣-␤ spectrin tetramer formation, we hypothesize the dysregulation occurs due to increased degradation of unbound spectrin monomers, which are unable to form stable ␣-␤ tetramers due to perturbed ratios of ␣ and ␤ spectrins.

␣II spectrin regulates cardiac I TO
The relatively normal action potential duration in ␣II spectrin cKO mice supports the impressive pathways likely evolved in cardiac tissue to maintain excitability. The association of ␣II spectrin with Na v 1.5 was not unexpected (3,5). However, our new findings illustrate altered K V 4.3 and I TO in ␣II spectrin cKO myocytes. Although co-immunoprecipitation experiments demonstrate association of ␣II spectrin with K V 4.3 (and Na V 1.5), the relationship of these proteins may be indirect and instead related to secondary cytoskeletal interactions or functional interactions between channel subunits. Prior work from Remme and colleagues (44) has elegantly demonstrated regulation of I Na by K V 4.3, beyond electrophysiologic interference. Furthermore, work by Deschênes et al. (45) has demonstrated Figure 10. ␣II spectrin cKO mice exhibit accelerated heart failure phenotypes following TAC. A-D, following TAC, ␣II spectrin cKO mice have decreased cardiac function, as evaluated by echocardiography. A, ejection fraction decreased precipitously in ␣II spectrin cKO mice, whereas, B, LVID at systole, increases compared with control mice. C, LVPW at systole failed to increase in ␣II spectrin cKO mice. D and E, representative M-mode traces from control (D) and ␣II spectrin (E) cKO mice at 12 weeks following TAC (n ϭ 5 and 6 for control and ␣II spectrin cKO mice, respectively). F and G, representative ECG traces from control (F) and ␣II spectrin (G) cKO mice demonstrating T-wave depression (n ϭ 5 and 6 for control and ␣II spectrin cKO mice, respectively). H and I, representative images of cardiac structure in control (H) and ␣II spectrin (I) cKO mice. Scale bar indicates 2 mm (n ϭ 3 and 4 for control and ␣II spectrin cKO mice, respectively). All values are represented as mean Ϯ S.E. * indicates a statistical differences from controls, p Ͻ 0.05.
␣II spectrin regulates cardiac function physical association of subunits responsible for I Na and I TO in neonatal rat ventricular myocytes. Given the dependence of both currents on ␣II spectrin for stability at the membrane, it is also possible that there is a co-trafficking mechanism contributing to their concurrent reductions, as has been observed between Na V 1.5 and K ir 2.1 (46). Whether I TO (K V 4.3) remodeling is a direct impact of ␣II spectrin loss, or a secondary compensatory factor to preserve action potential dynamics will be an important future area for research (43).
Although our study notes extensive transcriptional dysregulation in hearts lacking ␣II spectrin, similar to the dysregulation observed following the loss of binding partner, ankyrin G (5), the mechanism of this dysregulation is not known. Although ␣II spectrin does localize (albeit at low levels) to the nucleus and is known to play a role in DNA repair (47,48), a role for ␣II spectrin in direct transcriptional regulation in heart has not been described to our knowledge. It is likely that the transcriptional changes observed are a cellular response to the described disruptions of the cytoskeleton network, however, the mechanism of this regulation remains unexplored.
There are important aspects of the ␣II spectrin cKO mouse phenotype that are not yet fully elucidated. First, the normal dV/dt in the setting of reduced I Na is surprising, and is incompletely explained by the simultaneous reduction in I TO . Furthermore, fibrosis and hypertrophy were observed to occur simultaneously, preventing an investigation of any potential causal relationship. These limitations elucidate important areas for future study.
It is important to note that although experiments were done using isolated cardiomyocytes when possible (electrophysiology, immunoblotting), experiments conducted on whole heart tissue (RNAseq, histology) are confounded by the presence of noncardiomyocytes in heart tissue. Although efforts were made to obtain the most cardiomyocyte-rich sample possible, the contribution of blood, fibroblasts, adipocytes, and immune cells cannot be eliminated.
Finally, heart failure results in dysregulated calcium handling, which can contribute to increased calcium-activated calpain activity. We have shown increased calpain-mediated degradation of ␣II spectrin is associated with heart failure. Consistent with our new findings, work from Jain et al. (24) have previously demonstrated ␣II spectrin breakdown products in the serum of neonates with congenital heart disease. However, future studies investigating the localization of ␣II spectrin and other spectrins in human heart failure, in addition to the potential effect of ␣II and ␤II spectrin degradation in heart failure are essential to further our understanding of the role of spectrin dysregulation in both congenital and acquired heart failure.

Human heart tissue
Ischemic and nonischemic failing left ventricular tissue samples from explanted hearts of patients undergoing heart transplantation were obtained through The Cooperative Human Tissue Network: Midwestern Division at The Ohio State University. Nonfailing hearts were obtained through the Lifeline of Ohio Project. The Ohio State University Institutional Review Board approved the use of human subject tissue. This investigation conforms to the principles outlined in the Declaration of Helsinki.

Animal studies
Cardiomyocyte-specific ␣II spectrin knockout (␣II spectrin cKO) mice were produced using the Cre-flox system. Mice with exon 8 of the Sptan1 gene flanked by LoxP sites (Sptan1 f/f mice) (30) were backcrossed onto a C57BL/6J background for greater than five generations. Mice with Cre recombinase expression driven by the ␣-myosin heavy chain promoter were acquired from Jackson Laboratories (B6N.FVB(B6)-Tg(Myh6-cre)2182Mds/J, stock number 018972). Sptan1 f/WT breeders were established by maintaining one cre-positive parent per breeding pair. The genotype was confirmed with PCR (primers: Cre forward, ATGA-CAGACAGATCCCTCCTATCTCC; Cre reverse, CTCATCAC-TCGTTGCATCATCGAC; Cre internal control forward, CAA-ATGTTGCTTGTCTGGTG; Cre internal control reverse, GTCAGTCGAGTGCACAGTTT; Sptan1flox forward, AAC-AGTCACACCCTCTGAGTGCCA; Sptan1flox reverse, ATTC-AGTGGAAAGCTGAGAAGCCAG). Male and female mice between 16 and 24 weeks of age were used for experiments, unless otherwise noted. Littermate Sptan1 WT/WT , Cre ϩ , or Sptan1 f/f Cre Ϫ mice were used as controls for Sptan1 f/f , Cre ϩ , and ␣II spectrin cKO mice. Young mice were between 12 and 16 weeks of age. Adult mice were 20 -24 weeks of age. The Ohio State University IACUC approved all animal studies. The investigation conformed to the Guide for the Care and Use of Laboratory Animals published by the NIH.

Echocardiography
Echocardiographic analysis was performed on mice lightly anesthetized with isoflurane (1.75% in 1 liter/min oxygen). Mice were immobilized on a heated imaging stage during image acquisition. HR was monitored throughout imaging and recordings that obtained heart rates Յ400 bpm were excluded. Long and short axis analyses were conducted using the GE LOGIQ E, whereas Doppler analysis was conducted on a VEVO 2100. Analysis was conducted following acquisition using at least three nonadjacent contractions. Researchers blinded to genotype performed the image collection and analysis.

Electrocardiogram
Surface electrocardiogram analysis was conducted on mice anesthetized with isoflurane (2% in 1 liter/min oxygen). Mice were immobilized on a heated imaging stage during acquisition. Lead II ECGs were collected using PowerLab equipment (ADInstruments). Conscious ECGs were collected using: 1) the ECGenie system with eMouse analysis software (Mouse-Specific Inc.) or 2) implanted radiotelemetry using an ETA-F10 miniature telemeter (DSI) and Ponemah acquisition software. All conscious ECGs were analyzed using LabChart software. Researchers blinded to genotype performed the collection and analysis.

Transverse aortic constriction
Mice were anesthetized with 2% isoflurane and intubated for artificial ventilation at 120 -160 breaths per minute, tidal vol-␣II spectrin regulates cardiac function ume of 0.2-0.35 ml. Heating pads were used to keep body temperatures at 37°C throughout the procedure. The transverse aorta was accessed via a left lateral thoracotomy and 6-0 suture is used to ligate the aorta overlying a blunted 25-or 27-gauge needle. The needle was removed immediately following ligation leaving a discrete region of stenosis of the aorta. Successful constriction was confirmed by measuring the velocity of blood flow at the aortic root and after the construction site using echocardiography. The surgeon was blinded to genotype.

Cardiomyocyte isolation
Adult cardiomyocytes were isolated using aortic cannulation and retroperfusion of enzymes (for action potential and sodium current recordings, protease and collagenase type II (Worthington biochemical) or for potassium current recordings, Liberase TH (Roche Applied Science)) into the coronary circulation, as previously described (49 -51). Following isolation, cells were fixed with ethanol for staining, lysed for immunoblot, or processed for electrophysiology experiments.

Cardiac histologic analysis
Hearts were excised and flash frozen or fixed in 4% paraformaldehyde for 24 h. Masson's trichrome staining was performed on 5-m sections at the Comparative Pathology and Mouse Phenotyping Lab at the Ohio State University. TUNEL staining was performed using a fluorescein in situ cell death detection kit (Roche, 11684795910). Whole heart images were collected using a PathScan Enabler IV. Images of crosssectional myocytes from Masson's trichrome-stained and TUNEL-stained cardiac sections were collected on an EVOS microscope (Thermo Scientific). Cross-sectional areas of 50 myocytes per sample were measured using ImageJ software (52). Researchers blinded to genotype performed the image collection and analysis. Levels of fibrosis were quantified using an add-on to MATLAB (Mathworks). The add-on converts images from the RGB color space into CIELAB color space, then segments the images using the k means algorithm. Finally, the fibrosis (blue) segment is filtered through a color mask to remove noise, and the ratio of fibrosis to other tissues was calculated. TUNEL-stained images were analyzed using ImageJ "Color Threshold" and "Analyze Particles" functions. Only particles between 200 and 5000 pixels were counted as nuclei. Numbers of TUNEL-positive nuclei were normalized to total nuclei (by 4Ј,6-diamidino-2-phenylindole).

RNA sequencing
RNA-Seq of RNA isolated from male and female ␣II spectrin cKO and littermate control mice was conducted in collaboration with Ocean Ridge BioSciences (Deerfield Beach, FL). Mice were sacrificed using an isoflurane overdose, and the heart were immediately excised. The aorta was then cannulated and the heart was retroperfused with ice-cold Hanks' balanced salt solution to remove blood contamination. Ventricles were snapfrozen in liquid nitrogen. Total RNA was isolated from the tissue using the TRI Reagent (Molecular Research Center, part number TR118). Total RNA was quantified and assessed for quality on a 1% agarose, 2% formaldehyde gel. The RNA was then treated with RNase-free DNase I (Epicenter; part number D9905K) and re-purified using Agencourt RNAClean XP beads (Beckman Coulter; part number A63987). Final RNA samples were then quantified by spectrophotometry. cDNA libraries were prepared from 250 ng of DNA-free total RNA using the TruSeq Stranded mRNA Library Prep (96 samples) (Illumina Inc.; part number 20020595). The quality and size distribution of the amplified libraries were determined by CHIP-based capillary electrophoresis (Bioanalyzer 2100, Agilent Technologies). Libraries were Bioanalyzed (Bioanalyzer 2100, Agilent Technologies) and quantified using the KAPA Library Quantification Kit (Kapa Biosystems, Boston, MA). The libraries were loaded onto an Illumina HiSeq 4000 flowcell and bridge amplified to create sequence clusters and sequenced with 150 nucleotide paired-end reads plus dual index reads.
Quality-filtered and base-trimmed reads were used for alignment. Sequence alignment was performed using HISAT2 version 2.0.5. The read summarization program feature Counts2 version 1.5.1 was used for exon-and gene-level counting. Normalized RPKM values were calculated from the raw feature Counts read, and were then filtered to retain a list of genes with a minimum of ϳ50 mapped reads in 25% or more samples. The threshold of 50 mapped reads is considered the Reliable Quantification Threshold. GO Pathway Analysis was conducted to guide analysis and interpretation.

Co-immunoprecipitation
Co-immunoprecipitation experiments were conducted as previously described (53). WT mouse heart samples were homogenized in buffer (containing 0.025 M Tris-HCl, 0.15 M NaCl, 0.001 M EDTA, 1% (v/v) Nonidet P-40, 5% (v/v) glycerol, pH 7.4) using a Dounce homogenizer. Lysates were centrifuged for 30 min at 13,000 rpm at 4°C on using a benchtop centrifuge. One mg of supernatant was incubated and rotated with 2 g of Kv4.3 (Neuromab; clone K75/41), Na V 1.5 (Covance (3)), Mena (a generous gift from Dr. Benz), or control IgG at 4°C overnight. Following incubation, lysates with antibodies were rotated with washed Protein A/G Magnetic Beads (Pierce, number 88802) for 5 h at 4°C. After a 5-h incubation, the supernatant was removed from the beads using a magnetic stand, and the beads were washed 3 times with PBS. Bound protein was eluted with 2ϫ Laemmli sample buffer and ␤-mercaptoethanol and heated to 95°C for 10 min before immunoblotting with ␣I spectrin (BioLegend; 803101) or ␣II spectrin (BioLegend; 803201) antibodies. 60 g of lysate was used as an input loading control for each experiment.

Statistical analysis
All continuous variables are represented as mean Ϯ S.E. A multivariate one-way analysis of variance with Tukey's Honestly Significant Difference post hoc was used to identify differences among groups when data were normally distributed (passed Shapiro-Wilk normality test) in experiments with greater than two experimental groups. For experiments with just two experimental groups, an unpaired two-tailed Student's t test was performed, provided data were normally distributed based on a Shapiro-Wilk normality test. When data failed to pass a normality test, a Mann-Whitney test was performed. For experiments where the same mice were followed over time (TAC mice), a repeated measured two-way (genotype by time) analysis of variance was conducted. When examining differences in a categorical variable (failure to capture) between genotypes, a Fisher's exact test was used. Differences were considered significant at p Ͻ 0.05. Statistical analysis was performed using SPSS 25.0 (IBM SPSS Statistics) and GraphPad Prism (version 7.01 for Windows, GraphPad Software).