Tubulin hyperacetylation is adaptive in cardiacproteotoxicity by promoting autophagyPatrick M. McLendona, Bradley S. Fergusonb, Hanna Osinskaa, Md. Shenuarin Bhuiyana, Jeanne Jamesa,Timothy A. McKinseyb, and Jeffrey Robbinsa,1

aThe Heart Institute, Department of Pediatrics, Division of Molecular Cardiovascular Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH45229; and bDepartment of Medicine, Division of Cardiology, Anschutz Medical Campus, University of Colorado Denver, Aurora, CO 80045

Edited by J. G. Seidman, Harvard Medical School, Boston, MA, and approved October 28, 2014 (received for review August 12, 2014)

Proteinopathy causes cardiac disease, remodeling, and heartfailure but the pathological mechanisms remain obscure. MutatedαB-crystallin (CryABR120G), when expressed only in cardiomyocytesin transgenic (TG) mice, causes desmin-related cardiomyopathy,a protein conformational disorder. The disease is characterizedby the accumulation of toxic misfolded protein species that pres-ent as perinuclear aggregates known as aggresomes. Previously,we have used the CryABR120G model to determine the underlyingprocesses that result in these pathologic accumulations and toexplore potential therapeutic windows that might be used to de-crease proteotoxicity. We noted that total ventricular protein ishypoacetylated while hyperacetylation of α-tubulin, a substrate ofhistone deacetylase 6 (HDAC6) occurs. HDAC6 has critical roles inprotein trafficking and autophagy, but its function in the heart isobscure. Here, we test the hypothesis that tubulin acetylation is anadaptive process in cardiomyocytes. By modulating HDAC6 levelsand/or activity genetically and pharmacologically, we determinedthe effects of tubulin acetylation on aggregate formation inCryABR120G cardiomyocytes. Increasing HDAC6 accelerated aggre-gate formation, whereas siRNA-mediated knockdown or pharma-cological inhibition ameliorated the process. HDAC inhibition invivo induced tubulin hyperacetylation in CryABR120G TG hearts,which prevented aggregate formation and significantly improvedcardiac function. HDAC6 inhibition also increased autophagic fluxin cardiomyocytes, and increased autophagy in the diseased heartcorrelated with increased tubulin acetylation, suggesting thatautophagy induction might underlie the observed cardioprotec-tion. Taken together, our data suggest a mechanistic link betweentubulin hyperacetylation and autophagy induction and points toHDAC6 as a viable therapeutic target in cardiovascular disease.

autophagy | proteotoxicity | heart | alphaB-crystallin | HDAC6

Proteotoxicity is an important yet understudied mechanism incardiac pathobiology (1), as maintaining tight control of

protein homeostasis is critical for proper cell function. This isespecially important in the unique context of the heart, as itis under constant mechanical and oxidative stress, and car-diomyocytes appear to be largely postmitotic soon after birth andare unable to readily regenerate (2, 3). Cellular stress events,including normal physiologic stimuli, can lead to altered car-diomyocyte function if the pathways of protein quality control(PQC) are compromised. Pathological cardiac stress, includingpressure overload-induced hypertrophy and ischemia–reperfu-sion (I/R) injury, can alter protein degradation pathways (4–6).Our laboratory has developed a model of cardiac proteotoxicitybased upon transgenically mediated cardiomyocyte expression ofa mutated αB-crystallin (CryABR120G), which causes desmin-related cardiomyopathy in humans (7–9).CryAB is a molecular chaperone for desmin, an intermediate

filament protein expressed in myocytes. Desmin is a crucialcardiomyocyte protein with vital signaling and structural roles,including maintaining the interconnectivity between the sarco-mere, sarcolemma, and nuclear lamina, as well as holding mi-tochondria adjacent to sarcomeres to ensure that the energetic

requirements of the contractile cycle are met (10–12). TheCryABR120G mutation leads to a loss in chaperone function, suchthat CryAB can no longer facilitate proper folding of desmin(13). As a result, CryABR120G leads to accumulation of CryAB-and desmin-containing aggregates in the sarcoplasm, which leadsto disruption of myofibrillar structure and eventually to car-diomyocyte death and heart failure (9, 10, 14–16). We andothers have described deficits in protein degradation resultingfrom the CryABR120G mutation, including reductions in deg-radation through the ubiquitin–proteasome system (UPS) andautophagy (17–19).There is an urgent need for new therapeutic interventions for

treating cardiac proteotoxic disorders, as heart failure arisingfrom diverse etiologies can lead to misfolded protein accumu-lations (16). As the UPS cannot eliminate large aggregatedproteins from the cytoplasm due to the size constraints of theproteasome, we have recently focused on using the autophagy–lysosome pathway to eliminate cellular inclusion bodies, dam-aged organelles, and aggresomes. Autophagy relies on a dynamicmicrotubule trafficking scaffold, as both aggresome formationand autophagy are dependent upon dynein-mediated retrogradetransport of the cargo along microtubules to the perinuclearregion and lysosomes (20–22). However, how microtubule mo-dification is linked to the modulation of autophagy mechanisti-cally remains obscure.The histone deacetylases (HDACs) have emerged as a therapeutic

target in diverse diseases, from cancer and neurodegeneration to

Significance

Proteotoxicity, or the accumulation of misfolded protein, cancause heart failure and effective therapeutics are needed toreduce protein accumulation in the myocardium. This study showsthat inhibiting tubulin deacetylation by histone deacetylase6 (HDAC6) is protective in a mouse model of proteinopathy-induced heart failure. Inhibiting tubulin deacetylation using theFDA-approved drug suberoylanilide hydroxamic acid (SAHA)reduced protein aggregates in cardiomyocytes and led tosubstantial improvement in cardiac function. Mechanistically,we show that inhibiting HDAC6 increases autophagy in car-diomyocytes, and that inducing autophagy with voluntaryexercise also induces tubulin acetylation. This study shows thattubulin acetylation is important for autophagy stimulation inthe heart and, importantly, sheds new light on the mechanismof autophagy induction with HDAC inhibitors.

Author contributions: P.M.M. designed research; P.M.M., B.S.F., H.O., M.S.B., and J.J.performed research; P.M.M., B.S.F., and T.A.M. contributed new reagents/analytic tools;P.M.M., H.O., J.J., and J.R. analyzed data; J.R. helped select data; and P.M.M. and J.R.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: jeff.robbins@cchmc.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1415589111/-/DCSupplemental.

E5178–E5186 | PNAS | Published online November 17, 2014 www.pnas.org/cgi/doi/10.1073/pnas.1415589111

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

heart disease (23–27). HDACs are enzymes that remove acetylgroups from lysine residues on histones and cytoplasmic proteins,but can also affect gene expression, protein function, and cellsignaling (28). In the context of cardiac stress, HDAC activity isincreased (25, 29) and HDAC inhibitors can reduce hypertrophyinduced by pressure overload (29, 30). Several HDACs, includingHDAC6, have been implicated in promoting autophagy and, assuch, HDAC inhibitors may also impact on these processes.Depending upon the metabolic context, this may be beneficial indecreasing cardiac hypertrophy, which can occur in response topressure overload or ischemic–reperfusion (I/R) injury (30).HDAC6, a unique class IIb HDAC, plays a significant role in

the cellular PQC response (20, 31–33). Unlike other HDACs,HDAC6 is predominantly cytoplasmic and affects cytoskeletaldynamics through deacetylation of α-tubulin and cortactin (34).Acetylated tubulin can affect the stability of microtubules, andrecent data suggest that acetylated tubulin may augment as-sembly of autophagic cargo along microtubules, thereby primingthe cell for increased autophagic degradation (21). In addition,HDAC6 contains ubiquitin- and dynein-binding domains, con-tributing to aggresome formation and autophagy of ubiquitinatedsubstrates by linking them to dynein motors for transport to theperinuclear region and subsequent aggresomal sequestration orautophagic degradation (32, 35, 36). However, the precise role ofHDAC6 and tubulin acetylation in autophagy and pathologicalaggregation is unresolved. Increased HDAC6 activity has beencorrelated with cardiovascular disease (37, 38) but its cardiacfunction is largely undefined (39).Considering the intense interest of the HDAC inhibitors’

therapeutic potential in other diseases, we wished to explore theirefficacy in the pathogenic processes underlying the cardiac pro-teinopathies. Here, we demonstrate aberrant HDAC activity inCryABR120G cardiac proteinopathy and explore the functionalconsequences. We observed global protein hypoacetylation in thesehearts but hyperacetylation of α-tubulin, a substrate of HDAC6.Using the FDA-approved drug suberoylanilide hydroxamic acid(SAHA), we show that inhibiting tubulin deacetylation reducedaggregate formation and attenuated the cardiac functional def-icits that develop as a result of CryABR120G expression in vivo.This protection is due in part to increased autophagy and ourdata suggest that tubulin acetylation may be an adaptive pro-cess and may play an important role in the cardioprotectivemechanisms of HDAC inhibition.

ResultsAltered Acetylation Patterns and HDAC Activity in CryABR120G Hearts.A global profile of protein hypoacetylation has been documentedin a number of diseases, an observation that is consistent withthe hypothesis that HDAC hyperactivity can be detrimental inpathological contexts. For example, diverse pathologies such asHuntington’s disease and cardiac I/R injury show proteotoxicphenotypes and display protein hypoacetylation (25, 40, 41),suggesting HDAC hyperactivity. We analyzed the acetylationprofile of our proteotoxic disease model, in which CryABR120G

is expressed in cardiomyocytes of TG mice, resulting in the de-velopment of heart disease between 3 and 7 mo (9, 14). Asexpected, an overall pattern of hypoacetylation was present inthe diseased hearts (Fig. 1A). In contrast, α-tubulin, a preferredsubstrate of HDAC6, was hyperacetylated (Fig. 1B). α-Tubulin isdeacetylated by HDAC6 but, paradoxically, we noted increasedHDAC6 protein levels (Fig. 2 A and B). Class I HDAC activitywas increased (Fig. S1); this has been observed in other modelsof hypertrophic cardiac disease (26, 39), whereas class IIaHDAC activity was unaffected (Fig. S1). Notably, the HDACactivity of class IIb, to which HDAC6 belongs, was significantlyincreased (Fig. 2C). These data show that, whereas overallHDAC6 levels and activity are not compromised, deacetylaseactivity directed to α-tubulin is clearly decreased. We subsequently

defined the consequences of tubulin acetylation on aggregate for-mation and cardiac function.

Inhibiting HDAC6 Reduces Aggregate Formation in Cardiomyocytes.The function of HDAC6 is obscure in the heart (39) but datafrom other cell types suggest that it plays a critical function inaggresome formation and autophagy (31). As HDAC6 activitydirected toward α-tubulin appears to be perturbed in theCryABR120G hearts, we wished to determine the functionalconsequences on aggresome accumulation, which is a hallmarkof proteotoxic disease (11, 16). We assessed the effects ofHDAC6 modulation on CryABR120G-mediated aggregate for-mation in rat neonatal cardiomyocytes (RNCs) by performinggain- and loss-of-function experiments. Adenoviral expression ofHDAC6 reduced acetylated tubulin levels (Fig. 3A), confirmingHDAC6’s role in α-tubulin deacetylation in cardiac myocytes.Subsequently, adenoviruses containing HDAC6 and CryABR120G

were coinfected into RNCs and the cells were analyzed after 2 d inculture. At this time, CryABR120G aggregates are still small and havenot fully matured and coalesced into perinuclear aggresomes (Fig.3B, Top). HDAC6 overexpression resulted in larger, mature peri-nuclear aggregates, suggesting enhanced aggresome formation (Fig.3B, Bottom). We directly confirmed this by collecting soluble and

Fig. 1. Protein acetylation is perturbed in hearts of CryABR120G TG mice. (A)Global profile of protein hypoacetylation is observed in CryABR120G hearts at6 mo. Total acetylation is reduced by ∼60%, **P < 0.01. (B) Hyperacetylationof α-tubulin is observed in CryABR120G hearts at 6 mo. *P < 0.05, n = 4per group.

McLendon et al. PNAS | Published online November 17, 2014 | E5179

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

insoluble cellular fractions and measuring CryAB in the insolublefraction; this is a well-documented method for determining in-soluble aggregate content (16, 42, 43). The insoluble fraction de-rived from cells overexpressing HDAC6 showed elevated CryABlevels compared with controls (Fig. 3C). Subsequently, we partiallysilenced HDAC6 expression using siRNA transfection, which in-duced tubulin hyperacetylation (Fig. 3D). Immunocytochemicalanalyses showed smaller CryABR120G-positive aggregates scatteredthroughout the cytoplasm after HDAC6 knockdown, in contrast tothe predominantly perinuclear location of mature aggresomes incells treated with negative control siRNA (Negsi). At 5 d post-infection, we observed a 60% reduction in total aggregate/car-diomyocyte area (Fig. 3 E and F). To further confirm these resultswere due to HDAC6 inhibition, CryABR120G-infected RNCs weretreated with the specific HDAC6 inhibitor, tubastatin A (Fig. 3 Gand H), which also resulted in a similar reduction. These dataclearly show that HDAC6 promotes aggregate formation incardiomyocytes, and inhibiting HDAC6-mediated tubulindeacetylation inhibits aggregate accumulation.HDAC inhibitor therapies have been and are being explored

due to their apparent effectiveness in certain proteotoxic diseasemodels (25, 40, 44). We subsequently treated the cultures withSAHA, an FDA-approved hydroxamic acid-based pan-HDACinhibitor that inhibits class IIb HDAC activity. Consistent withthe above results, SAHA treatment induced tubulin hyper-acetylation in cardiomyocytes (Fig. 4A). We treated CryABR120G-infected cells with SAHA at 3 d postinfection and noted smalleraggregates scattered throughout the cytoplasm when fixed 2 dlater (Fig. 4B). The lack of visible, perinuclear aggresomes wasreflected by a corresponding decrease of aggregate content ona per cardiomyocyte basis (Fig. 4C). The data further supportthe hypothesis that tubulin hyperacetylation via HDAC6 inhi-bition effectively reduces aggregate formation in RNCs. Impor-tantly, the data also confirm that inducing tubulin acetylationcan prevent further aggregation, suggesting the potential fortherapeutic application.

SAHA Treatment Reduces Aggregate Levels and Improves CardiacFunction in Vivo. To assess if this process was therapeutically rel-evant in vivo, we induced tubulin hyperacetylation by treating micewith SAHA. After subjecting the mice to 2 wk of SAHA in thedrinking water, acetylated tubulin levels were significantly in-creased (Fig. 5 A and B). The mice showed no obvious adverseeffects and the hearts appeared normal and sarcomere periodicitywas maintained (Fig. S2). Subsequently nontransgenic (NTG)and CryABR120G TG mice were treated with SAHA or vehiclefor 7 wk, starting at 3.5 mo postbirth. The time point was chosenbecause sarcoplasmic aggregates are present without an accom-panying decline in cardiac function, allowing us to define drugefficacy under preexisting conditions of aggregate formation, butin the absence of overt heart failure. We could then determinewhether SAHA treatment reduced aggregates and could prevent

cardiac functional decline. SAHA treatment decreased cyto-plasmic aggregate content by ∼40% (Fig. 5 C and D), suggestingthat increased tubulin acetylation effectively decreased overallaggregate accumulation. To assess the effect of myocardial ag-gregate reduction on cardiac function, we examined the heartsusing echocardiography. There was substantial improvement incardiac function in the SAHA-treated mice at 5.5 mo of age, asfractional shortening and ejection fraction improved significantlycompared with the vehicle-treated controls (Fig. 5 E and F).Interestingly, and in contrast to recent reports (23, 24), leftventricular mass was unchanged (Fig. 5G).

SAHA-Mediated Cardioprotection Correlates with Increased Autophagy.We previously noted decreased autophagy in CryABR120G car-diomyocytes and showed that increasing basal autophagy, eitherthrough transgenic Atg7 overexpression or voluntary exercise, re-duced aggregate formation and improved cardiac function inCryABR120G mice (17, 45). We wished to determine if decreasingaggregates through inhibiting tubulin deacetylation was also car-dioprotective and if the mechanism involved modulation ofautophagy. SAHA can stimulate autophagy in the mammalianheart (24) and, consistent with those data, we detected increasedautophagic vacuoles in the SAHA-treated CryABR120G hearts(Fig. 6A). To determine if SAHA was affecting autophagic flux inthis disease setting, we used bafilomycin A1 (BafA1) treatment tomeasure autophagic flux in CryABR120G-infected cells after treat-ing the cultures with SAHA. Autophagic flux substantially increasedafter SAHA treatment (Fig. 6B). To support these data, we mea-sured p62 levels after SAHA treatment, as p62 is an importantmarker of autophagy that is degraded during the autophagic cycle(46), such that decreased p62 can be interpreted as increasedautophagy. Indeed, we noted decreased p62 levels after SAHAtreatment (Fig. S3), which further supports increased autophagy. Toconfirm that this effect was indeed mediated by HDAC6 inhibition,we repeated the experiment using the HDAC6-specific inhibitortubastatin A and noted a similar increase in autophagic flux, asmeasured by LC3-II levels post-BafA1 treatment (Fig. 6C). Thesedata strongly suggest that the observed aggregate reduction reflectsincreased aggresomal clearance via enhanced autophagic flux.As induction of tubulin acetylation via SAHA treatment in-

creased cardiac autophagy, we asked whether the converse wasalso true: Does inducing autophagy increase myocardial tubulinacetylation levels? Increasing autophagy is beneficial in thecontext of proteotoxic stress in the heart (17) and increasingautophagy via voluntary exercise (47) increased lifespan inCryABR120G TG mice (45, 48). We subjected CryABR120G tovoluntary exercise, which induces autophagy (45, 47). ExercisedCryABR120G mice showed increased levels of tubulin acetylationcompared with unexercised mice (Fig. 7 A and B), with a pres-ervation of HDAC6 activity (Fig. S4).

DiscussionProteotoxicity induced by misfolded protein accumulation isa causative factor in disease and appears to be particularly im-portant in organ systems where function is dependent upon celltypes with limited regenerative capacities. Proteotoxicity can bedue to different etiologies and a prominent cytoplasmic featurein a number of proteotoxic diseases is the formation of largeinclusion bodies visible by light microscopy. These proteina-ceous aggregates have been characterized in a number of dis-eases, including neurodegenerative disorders (49, 50) anddiabetes (51, 52). More recently, we have focused on proteinmisfolding in cardiovascular disease and uncovered a causativerole for proteotoxicity in desmin-related myopathies (8, 9, 16,53). However, proteotoxicity also appears to be associated withmore general cardiac disease including load-induced heart fail-ure, ischemic cardiomyopathy, and hypertrophic cardiomyopathy

Fig. 2. Hyperacetylation of α-tubulin occurs despite induction of HDAC6. (Aand B) HDAC6 protein levels increase concomitantly with tubulin acetylation.Acetylated tubulin and HDAC6 are both increased. **P < 0.01, n = 4–6 pergroup. (C) Class IIb HDAC activity is significantly increased. ***P < 0.01, n = 8–10per group.

E5180 | www.pnas.org/cgi/doi/10.1073/pnas.1415589111 McLendon et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

(6, 54, 55). Elimination of the toxic misfolded protein is a legit-imate therapeutic target for the proteotoxic phenotypes but,despite substantial efforts across multiple fields of study, no ef-fective therapeutic intervention to reverse pathological proteinaggregation has been found. This highlights the need to betterunderstand the pathogenic mechanisms and here we investigatethe potential role of aberrant protein acetylation. This appears tobe a common feature in proteinopathies (25, 29, 41, 56) and,importantly, drugs to inhibit overactive deacetylation are avail-able and in clinical use.HDACs are implicated in a broad spectrum of disease, in-

cluding proteotoxicity. Protein acetylation is an important post-

translational modification and controls many crucial cellularprocesses, including transcription, cell motility, and metabolism(28, 39, 57). Protein acetylation can also influence cytoplasmicprotein activity (58) and assembly of autophagic vacuoles (21).The acetylase and deacetylase activities must remain balanced topreserve proper cell function. Indeed, aberrant acetylation pro-files are observed in disease, which has led to using HDAC in-hibition as a therapeutic approach in neurodegeneration, inwhich hypoacetylation of cellular proteins is observed (40, 59).Several reports have described beneficial effects of broad-spec-trum HDAC inhibition in the progression of neurodegenera-tive phenotypes (44, 60). The therapeutic potential of HDAC

Fig. 3. Tubulin acetylation status influences aggregate formation in CryABR120G-expressing cardiomyocytes. (A) HDAC6 overexpression leads to reducedtubulin acetylation in RNCs. n = 4 per group. (B) HDAC6 overexpression increased aggregate formation in CryABR120G-infected RNCs. (Scale bar, 50 μm.) n =50–100 cells per group. (C) HDAC6 overexpression increased levels of insoluble CryAB-positive cellular protein, confirming increased aggregate formation.*P < 0.05, n = 5 per group. (D) siRNA knockdown increased tubulin acetylation. (Scale bar, 25 μm.) (E) Aggregate formation was inhibited in CryABR120G-infected RNCs when HDAC6 was reduced, compared with cells treated with negative control (scrambled) siRNA (Negsi). n = 50–100 cells per group. (F) Re-duced HDAC6 reduced aggregate/myocyte ratios compared with CryABR120G RNCs transfected with scrambled siRNA. ***P < 0.001. (G and H) Similar effectswere seen using a specific HDAC6 inhibitor, tubastatin A. ***P < 0.001. (Scale bar, 25 μm.) n = 50–100 cells per group.

McLendon et al. PNAS | Published online November 17, 2014 | E5181

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

inhibition has also been explored for different cancers and theHDAC inhibitor, SAHA, is FDA-approved to treat cutaneousT-cell lymphoma. Importantly, it was recently shown that SAHAinduces autophagy in cardiomyocytes (24), but the underlyingmechanism was not determined. The effects of modulatingHDAC activity likely are a function of the specific disease, celltype, and dosing protocols used, and should be completely un-derstood to expand the therapeutic reach of HDAC inhibition.In this paper, we describe both global and specific changes in

protein acetylation in CryABR120G hearts (9, 10, 16, 43). In con-trast to the general hypoacetylated global protein profile, weobserved hyperacetylation of α-tubulin, the primary componentof microtubules. We noted that hyperacetylated α-tubulin ispresent even with increased HDAC6 expression and class IIbHDAC activity. We therefore began a systematic analysis of theeffects of increased and decreased HDAC6 levels by manipu-lating the enzyme’s overall activity in RNCs. The data show thatHDAC6 can control tubulin acetylation: HDAC6 knockdown orinhibition led to a reduction in perinuclear aggregate formation,whereas increased HDAC6 levels accelerated the aggregationprocess. To assess whether tubulin hyperacetylation was anadaptive or maladaptive compensatory response, we further in-duced tubulin acetylation by treating CryABR120G TG mice withSAHA. Our goal was to directly correlate tubulin acetylationwith reduced aggregate formation and determine whether ag-gregate reduction achieved by inhibiting HDAC6 was protectivein an animal model of proteotoxicity. In agreement with the datafrom the RNCs, we noted reduced aggregate formation inCryABR120G TG mice, which led to a significant improvement

in cardiac function. SAHA treatment increased autophagic fluxin CryABR120G-expressing cardiomyocytes and led to increasedautophagic vacuoles in the myocardium.We then tested the general hypothesis of linking tubulin hyper-

acetylation to autophagic modulation by subjecting CryABR120G

TG mice to voluntary exercise. We recently showed that voluntaryexercise can elicit an increase in autophagic gene expression,which led to improved cardiac function and lifespan inCryABR120G mice compared with control mice (45). We wereinterested in whether increased tubulin acetylation would ac-company increased autophagy in this context, which we haveshown to be protective. Indeed, increased tubulin hyper-acetylation was observed in the exercised cohort (Fig. 7) despitepreserved HDAC6 activity (Fig. S4), which is consistent with thehypothesis that tubulin acetylation is a cardiomyocyte adaptiveresponse that can lead to increased autophagic clearance of toxicprotein accumulation. As SAHA inhibits class IIb HDACs,inhibiting HDAC6-mediated tubulin deacetylation may be re-sponsible for the observed increase in autophagic flux, providingimportant mechanistic data on how HDAC inhibition caninduce autophagy.The multifaceted role of HDAC6 in autophagic stimulation

is currently unclear and somewhat controversial. There are nu-merous examples describing the positive role of HDAC6 onautophagy stimulation (20, 32, 33), but more recent data suggestthat α-tubulin hyperacetylation is essential for providing anassembly platform for the autophagosome (21). HDAC6 is notrequired for inducing autophagy, but rather plays a role in theautophagosome–lysosome fusion step (61). As HDAC6 also hasubiquitin-binding domains and can modulate the unfolded pro-tein response, its role in selective autophagy may be more criticalthan in other autophagic modes such as those induced duringstarvation.The preservation of HDAC6 levels and activity levels despite

increased tubulin hyperacetylation in CryABR120G TG mice is in-triguing. As HDAC6 positively influences autophagy, it is logicalthat inhibiting HDAC6 might be detrimental to maintainingautophagic flux. This does not appear to be the case and suggeststhat HDAC6 inhibitors may function by inhibiting deacetylationof tubulin by HDAC6, allowing HDAC6 to be recruited else-where and/or to the autophagic machinery. Clearly, HDAC6 isstill functional after inhibition and overall HDAC6 protein levelsmay be a poor indicator of autophagic function (45, 61). We donot fully understand how the CryABR120G mutation is affectingHDAC6 activity and/or regulation. Given the myriad roles ofHDAC6 in maintaining PQC, further study is needed to fullyelucidate its role in protein degradation in the heart.Our interpretation of the data is tempered somewhat by the

fact that SAHA is a pan-HDAC inhibitor that has activity towardother class I/II HDACs. However, the present study focuses onthe effects of acetylated tubulin on aggregate formation. Theobserved CryAB aggregate:myocyte area ratios were similar afterSAHA treatment or decreasing/inhibiting HDAC6, suggestingthat the observed effects on aggregate formation were primarilydue to increased acetylated tubulin via HDAC6 inhibition. In-deed, specific inhibition of HDAC6 with tubastatin A inducedtubulin acetylation and increased autophagic flux. We certainlycannot rule out the possibility that other HDACs may be im-pacting the disease phenotype, but there are clear differences inthe effects of HDAC inhibition in our model versus other cardiacstress models. For example, class I HDACs are generally thoughtto be prohypertrophic (39) and can be inhibited by SAHA; in-hibition of these HDACs would be expected to be cardio-protective. However, we did not observe substantial reductions inhypertrophy, which differs from results observed with HDACinhibition in a model of cardiac pressure overload stress (23, 39),suggesting that the protective effects of SAHA in this context arederived from HDAC6 inhibition.

Fig. 4. SAHA treatment of RNCs promotes tubulin acetylation and reducesaggregates. (A) SAHA induced α-tubulin acetylation, confirming the in-hibitory activity toward HDAC6. n = 4 per group. (B and C) SAHA treatmentreduced aggregate formation in CryABR120G-infected RNCs. ***P < 0.001.(Scale bar, 50 μm.)

E5182 | www.pnas.org/cgi/doi/10.1073/pnas.1415589111 McLendon et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

Our data support the hypothesis that hyperacetylated tubulin isan adaptive response in the heart that can stimulate autophagy inthe face of proteotoxic stress. Inhibiting HDAC6-mediated tubulindeacetylation increases autophagy in cardiomyocytes, and this mayhelp explain the increases in autophagy observed when usingHDAC inhibitors in cardiac disease. The cardioprotection ob-served by inducing tubulin acetylation underscores the complexrole of HDAC6 in selective cardiac autophagy and its potentialtherapeutic value.

Materials and MethodsReagents and Immunocytochemistry. Antibodies for immunoblotting were asfollows: rabbit anti-HDAC6 antibodies were from Novus and Assay Biotech.Mouse antiacetyl tubulin and anti–α-sarcomeric actin were from Sigma. Rabbitanti-LC3B and rabbit antiacetyl lysine were from Cell Signaling. Mouse anti-GAPDH was from Millipore. Antibodies for immunofluorescence staining areas follows: anti-CryAB (Enzo), and anti-troponin I (Millipore). BafA1 was pur-chased from LC Laboratories, and tubastatin A was from Sigma.

Primary RNCs. All cell culture reagents were from Life Technologies, unlessnoted otherwise. Primary cardiomyocytes were prepared as follows: Ven-

tricular tissue of 2- to 3-d-old Sprague-Dawley rat pups was digested withcollagenase at 37 °C overnight, followed by further digestion in trypsin.After a preplating step to remove cardiac fibroblasts, isolated cardio-myocytes were plated at 1.5 × 106 cells/plate in 10 cm2 dishes, 2 × 105 cells/well in six-well plates, or 1 × 105 cells/chamber in two-well glass chamberslides (Thermo) in α-MEM containing 10% (vol/vol) FBS and 1% Ab/Am (LifeTechnologies) and incubated in a humid 5% CO2 atmosphere at 37 °C.Experiments were initiated no less than 24 h after plating and were done inDMEM containing 2% (vol/vol) FBS and 1% Ab/Am.

Adenoviral and siRNA Treatment of RNCs. Replication-deficient adenoviruseswere generated using the AdEasy system (Clontech). CryABR120G-FLAG andHDAC6-FLAG were prepared by insertion of mouse cDNA into the multiplecloning site of pShuttle-CMV. For adenoviral infection, plating media wasremoved and the cells were incubated at 37 °C for 2 h in serum-free DMEMcontaining the appropriate amount of virus. Virus-containing media was re-moved and replaced with DMEM containing 2% (vol/vol) FBS and 1% Ab/Am.

Animals. CryABR120G TG mice have been described previously (9) and are inthe FVB/N background. NTG control animals were in the same background.A minimum of four animals was used for each experiment, unless notedotherwise. Animals were handled in accordance with the principles and

Fig. 5. Modulating tubulin acetylation in vivo in CryABR120G TG mice. (A and B) SAHA treatment (p.o.) increased tubulin acetylation after 2 wk of treatment inmice. n = 4 per group. (C) SAHA treatment in CryABR120G TG mice resulted in substantial reduction in aggregate formation. (Top) cross-sectional areas; (Bottom)longitudinal views. Area of aggregates from n = 300–380 myocytes was measured, n = 2 hearts per group. (D) Aggregate area per cell was reduced by ∼40%. ***P <0.001. (Scale bar, 50 μm.) (E and F) SAHA-mediated aggregate reduction led to significantly improved cardiac fractional shortening (FS) and ejection fraction (EF)compared with vehicle-treated controls. *P < 0.05 (G) Cardiac hypertrophy, as assessed by left ventricular mass, was unchanged. n = 5–7 per group; for NTG, n = 3.

McLendon et al. PNAS | Published online November 17, 2014 | E5183

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

procedures of the Guide for the Care and Use of Laboratory Animals (62).The Institutional Animal Care and Use Committee at Cincinnati Children’sHospital approved all experimental procedures.

HDAC Activity Assays. HDAC activity assays were performed as described (37).A minimum of six mice per group was used for activity assays. Briefly, tissuewas lysed in PBS (pH 7.4, 0.5% Triton X-100, 300 mM NaCl and protease/phosphatase inhibitor mixture; Thermo) using a Bullet Blender homogenizer(Next Advance). Tissue extracts were diluted into PBS buffer (100 μL totalvolumes) in a 96-well plate (60 μg protein per well). Substrates were added(5 μL of 1 mM DMSO stock solutions) and plates returned to the 37 °C in-cubator for 2–3 h. Developer/stop solution was added (50 μL per well of PBSwith 1.5% (vol/vol) Triton X-100, 3 μM trichostatin A, and 0.75 mg/mL

trypsin), with another 20 min incubation at 37 °C. AMC (7-amino-4-methyl-coumarin) fluorescence was measured using a BioTek Synergy 2 plate reader,with excitation and emission filters of 360 nm and 460 nm (each with band-width of 40 nm), along with a 400-nm dichroic top mirror. Background signalsfrom buffer blanks were subtracted, and data were normalized as neededusing appropriate controls.

SAHA Administration. SAHA was administered in the drinking water as de-scribed (44). Briefly, SAHA (Vorinostat; LC Laboratories) was mixed with 5molar equivalents of 2-hydroxypropyl-β-cyclodextrin (AK Scientific) andMilliQ water, heated at 55 °C until dissolved, and cooled on ice to roomtemperature. The final concentration of SAHA was 0.33 mg/mL in water,corresponding to ∼50 mg/kg/d dosing. Drinking water was replaced withfreshly prepared solution once weekly.

Immunocytochemistry. RNCs were rinsed with PBS and fixed with 4% (vol/vol)paraformaldehyde (Electron Microscopy Sciences) in PBS for 15 min at roomtemperature. Cells were permeabilized with 0.25% Triton X-100 in PBS for15 min, and incubated in 0.1 mol/L glycine, pH 3.5 for 30 min. Cells wereincubated in blocking buffer (1% BSA, 0.1% cold water fish gelatin, 0.1%Tween 20 in PBS) for 60 min. Primary and secondary antibodies were dilutedin blocking buffer. Cells were counterstained with DAPI and mounted withVectashield Hard Set (Vector Laboratories). Quantification of aggregate:myocyte area used MetaMorph software.

Micewere anesthetizedwith isoflurane and hearts were fixed by perfusionwith 10% (vol/vol) normal-buffered formalin or 4% (vol/vol) para-formaldehyde. Butterflied hearts were fixed overnight at room temperature,transferred into 70% EtOH, and embedded in paraffin or subjected to a su-crose gradient under refrigeration and frozen in O.C.T. compound (Tissue-Tek). Tissue slices were sectioned (5 μm) and paraffin sections were depar-affinized in xylene and alcohol. Slices were rehydrated in dH2O and PBS, andantigen retrieval was carried out by boiling in citrate buffer (0.1 mol/L so-dium citrate, pH 6.0). Sections were blocked and probed with primary andsecondary antibodies diluted in blocking buffer. Sections were counter-stained with DAPI and mounted with Vectashield Hard Set. Aggregate areaper cell was calculated by manually outlining 25–50 myocytes per field fromat least 10 fields per heart (n = 2). Images were obtained on a Nikon A1confocal microscope using a 40× H20 objective (NA = 1.15). Aggregate areawas calculated using National Institutes of Health ImageJ software.

Fig. 6. SAHA induces autophagy in cardiomyocytes and increased autophagy underlies the protective effects. (A) Increased autophagic vacuoles (whitearrows) are increased in SAHA-treated hearts and are proximal to aggregates (*). (Scale bar, 2 μm.) (B) SAHA treatment (10 μM) in CryABR120G-infectedcardiomyocytes increased autophagic flux. n = 4 per group. (C) Tubastatin A treatment (10 μM) confirms autophagy induction is mediated by HDAC6. ***P <0.001, n = 4 per group.

Fig. 7. Voluntary exercise increases acetylated tubulin levels. (A and B) In-creased acetylated tubulin in CryABR120G hearts after voluntary exercise.

E5184 | www.pnas.org/cgi/doi/10.1073/pnas.1415589111 McLendon et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

Electron Microscopy. Hearts of isoflurane-anesthetized mice were initiallyfixed by perfusion with 1% paraformaldehyde/2% (vol/vol) glutaraldehyde incardioplegic solution (50 mmol/L KCl, 5% dextrose in PBS) and next in 1%paraformaldehyde/2% (vol/vol) glutaraldehyde in 0.1 mol/L cacodylatebuffer, pH 7.2. The heart was removed and immersed into the latter fixative(ice cold) and then left and right ventricular free walls and septa were iso-lated. Each region was divided into small fragments and fixed further in thesame fixative, at 4 °C, then postfixed in 1% OSO4 (in water) before de-hydration in acetone and embedding in epoxy resin. Ultrathin sections werecounterstained with uranium and lead salts. Images were acquired on aHitachi 7600 electron microscope equipped with an AMT digital camera.

Immunoblotting. RNCs were lysedwith CelLyticM (Sigma) containing proteaseinhibitor mixture (Complete Mini; Roche) and scraped before brief sonicationand centrifugation to pellet the insoluble fraction. For cardiac tissue, micewere euthanized via CO2 inhalation. Ventricular tissue (fresh or flash frozenin liquid N2) was placed in ice-cold CelLytic M and minced on ice. Tissue washomogenized with Zr-Si beads and protein concentration was measured viathe Bradford method. Proteins were separated by SDS/PAGE and transferredonto PVDF before blocking in 5% (wt/vol) milk in TBS-T. Primary antibodieswere diluted in milk buffer and incubated with the membranes overnight at4 °C. Membranes were probed with appropriate secondary AP-conjugatedantibodies (1:5,000 in milk buffer) and incubated 1 h at room temperature.Membranes were visualized with ECF reagent (GE Healthcare) and scannedwith a STORM scanner. Data were analyzed by densitometry and normalizedto a housekeeping gene (GAPDH or actin).

Voluntary Exercise. One-month-old mice were housed in regular cages or incages equipped with voluntary running wheels with one mouse per cage.Hearts from exercised and unexercised (sedentary) control mice were col-lected after 4 mo of running and homogenized for immunoblotting. Six maleCryABR120G TG mice were used in each group.

Echocardiography. Echocardiogramswere performedon isoflurane-anesthetizedmice using a VisualSonics Vevo 2100 imaging system. Two-dimensionalM-mode echocardiography was used to image the left ventricle to determineleft ventricular (LV) fractional shortening (% FS), a primary measure of LVfunction. From these data, LV end-systolic and end-diastolic diameter, LV mass,and ejection fraction were calculated.

Statistics. Data are expressed as mean ± SEM. All statistical tests were donewith KyPlot 2.0 software. Statistical analyses between two groups werecarried out with Student t test and Tukey’s post hoc test. A value of P < 0.05was considered to be statistically significant.

ACKNOWLEDGMENTS. This work was supported by National Institutesof Health (NIH) Grants P01HL69779, P01HL059408, R01HL05924, andR011062927 and a Trans-Atlantic Network of Excellence Grant from LeFondation Leducq (to J.R.); NIH Grants F32 HL112558 and T32 HL007382 (toP.M.M.); and an American Heart Association postdoctoral fellowship (to M.S.B.).T.A.M. was supported by NIH Grants HL116848 and AG043822 and B.S.F. wassupported by a postdoctoral fellowship from the American Heart Association(12POST10680000).

1. Willis MS, Patterson C (2013) Proteotoxicity and cardiac dysfunction—Alzheimer’sdisease of the heart? N Engl J Med 368(5):455–464.

2. Senyo SE, et al. (2013) Mammalian heart renewal by pre-existing cardiomyocytes.Nature 493(7432):433–436.

3. Porrello ER, et al. (2011) Transient regenerative potential of the neonatal mouseheart. Science 331(6020):1078–1080.

4. Hill JA, Olson EN (2008) Cardiac plasticity. N Engl J Med 358(13):1370–1380.5. Zhu H, et al. (2007) Cardiac autophagy is a maladaptive response to hemodynamic

stress. J Clin Invest 117(7):1782–1793.6. Tannous P, et al. (2008) Intracellular protein aggregation is a proximal trigger of

cardiomyocyte autophagy. Circulation 117(24):3070–3078.7. Vicart P, et al. (1998) A missense mutation in the alphaB-crystallin chaperone gene

causes a desmin-related myopathy. Nat Genet 20(1):92–95.8. Wang X, et al. (2003) AlphaB-crystallin modulates protein aggregation of abnormal

desmin. Circ Res 93(10):998–1005.9. Wang X, et al. (2001) Expression of R120G-alphaB-crystallin causes aberrant desmin

and alphaB-crystallin aggregation and cardiomyopathy in mice. Circ Res 89(1):84–91.10. McLendon PM, Robbins J (2011) Desmin-related cardiomyopathy: An unfolding story.

Am J Physiol Heart Circ Physiol 301(4):H1220–H1228.11. Goldfarb LG, Dalakas MC (2009) Tragedy in a heartbeat: Malfunctioning desmin

causes skeletal and cardiac muscle disease. J Clin Invest 119(7):1806–1813.12. Capetanaki Y (2002) Desmin cytoskeleton: A potential regulator of muscle mito-

chondrial behavior and function. Trends Cardiovasc Med 12(8):339–348.13. Bova MP, et al. (1999) Mutation R120G in alphaB-crystallin, which is linked to a des-

min-related myopathy, results in an irregular structure and defective chaperone-likefunction. Proc Natl Acad Sci USA 96(11):6137–6142.

14. Maloyan A, et al. (2005) Mitochondrial dysfunction and apoptosis underlie thepathogenic process in α-B-crystallin desmin-related cardiomyopathy. Circulation112(22):3451–3461.

15. Maloyan A, Sayegh J, Osinska H, Chua BHL, Robbins J (2010) Manipulation of deathpathways in desmin-related cardiomyopathy. Circ Res 106(9):1524–1532.

16. Sanbe A, et al. (2004) Desmin-related cardiomyopathy in transgenic mice: A cardiacamyloidosis. Proc Natl Acad Sci USA 101(27):10132–10136.

17. Pattison JS, Osinska H, Robbins J (2011) Atg7 induces basal autophagy and rescuesautophagic deficiency in CryABR120G cardiomyocytes. Circ Res 109(2):151–160.

18. Chen Q, et al. (2005) Intrasarcoplasmic amyloidosis impairs proteolytic function ofproteasomes in cardiomyocytes by compromising substrate uptake. Circ Res 97(10):1018–1026.

19. Wang X, Osinska H, Gerdes AM, Robbins J (2002) Desmin filaments and cardiac dis-ease: Establishing causality. J Card Fail 8(6, Suppl):S287–S292.

20. Iwata A, Riley BE, Johnston JA, Kopito RR (2005) HDAC6 and microtubules are re-quired for autophagic degradation of aggregated huntingtin. J Biol Chem 280(48):40282–40292.

21. Geeraert C, et al. (2010) Starvation-induced hyperacetylation of tubulin is requiredfor the stimulation of autophagy by nutrient deprivation. J Biol Chem 285(31):24184–24194.

22. Mackeh R, Perdiz D, Lorin S, Codogno P, Poüs C (2013) Autophagy and microtubules:New story, old players. J Cell Sci 126(Pt 5):1071–1080.

23. Kong Y, et al. (2006) Suppression of class I and II histone deacetylases blunts pressure-overload cardiac hypertrophy. Circulation 113(22):2579–2588.

24. Xie M, et al. (2014) Histone deacetylase inhibition blunts ischemia/reperfusion injuryby inducing cardiomyocyte autophagy. Circulation 129(10):1139–1151.

25. Granger A, et al. (2008) Histone deacetylase inhibition reduces myocardial ischemia-reperfusion injury in mice. FASEB J 22(10):3549–3560.

26. Cavasin MA, et al. (2012) Selective class I histone deacetylase inhibition suppresseshypoxia-induced cardiopulmonary remodeling through an antiproliferative mecha-nism. Circ Res 110(5):739–748.

27. McConkey DJ, White M, YanW (2012) HDAC inhibitor modulation of proteotoxicity asa therapeutic approach in cancer. Adv Cancer Res 116:131–163.

28. Haberland M, Montgomery RL, Olson EN (2009) The many roles of histone deacety-lases in development and physiology: Implications for disease and therapy. Nat RevGenet 10(1):32–42.

29. Zhao TC, Cheng G, Zhang LX, Tseng YT, Padbury JF (2007) Inhibition of histone de-acetylases triggers pharmacologic preconditioning effects against myocardial ische-mic injury. Cardiovasc Res 76(3):473–481.

30. Cao DJ, et al. (2011) Histone deacetylase (HDAC) inhibitors attenuate cardiac hyper-trophy by suppressing autophagy. Proc Natl Acad Sci USA 108(10):4123–4128.

31. d’Ydewalle C, Bogaert E, Van Den Bosch L (2012) HDAC6 at the intersection of neu-roprotection and neurodegeneration. Traffic 13(6):771–779.

32. Kawaguchi Y, et al. (2003) The deacetylase HDAC6 regulates aggresome formationand cell viability in response to misfolded protein stress. Cell 115(6):727–738.

33. Pandey UB, et al. (2007) HDAC6 rescues neurodegeneration and provides an essentiallink between autophagy and the UPS. Nature 447(7146):859–863.

34. Li Y, Shin D, Kwon SH (2013) Histone deacetylase 6 plays a role as a distinct regulatorof diverse cellular processes. FEBS J 280(3):775–793.

35. Kraft C, Peter M, Hofmann K (2010) Selective autophagy: Ubiquitin-mediated rec-ognition and beyond. Nat Cell Biol 12(9):836–841.

36. Olzmann JA, et al. (2007) Parkin-mediated K63-linked polyubiquitination targetsmisfolded DJ-1 to aggresomes via binding to HDAC6. J Cell Biol 178(6):1025–1038.

37. Lemon DD, et al. (2011) Cardiac HDAC6 catalytic activity is induced in response tochronic hypertension. J Mol Cell Cardiol 51(1):41–50.

38. Demos-Davies KM, et al. (2014) HDAC6 contributes to pathological responses of heartand skeletal muscle to chronic angiotensin-II signaling. Am J Physiol Heart Circ Physiol307(2):H252–H258.

39. Xie M, Hill JA (2013) HDAC-dependent ventricular remodeling. Trends CardiovascMed 23(6):229–235.

40. Thomas EA, et al. (2008) The HDAC inhibitor 4b ameliorates the disease phenotypeand transcriptional abnormalities in Huntington’s disease transgenic mice. Proc NatlAcad Sci USA 105(40):15564–15569.

41. Steffan JS, et al. (2001) Histone deacetylase inhibitors arrest polyglutamine-dependent neurodegeneration in Drosophila. Nature 413(6857):739–743.

42. Sanbe A, et al. (2009) Protective effect of geranylgeranylacetone via enhancement ofHSPB8 induction in desmin-related cardiomyopathy. PLoS ONE 4(4):e5351.

43. Sanbe A, et al. (2005) Reversal of amyloid-induced heart disease in desmin-relatedcardiomyopathy. Proc Natl Acad Sci USA 102(38):13592–13597.

44. Hockly E, et al. (2003) Suberoylanilide hydroxamic acid, a histone deacetylase in-hibitor, ameliorates motor deficits in a mouse model of Huntington’s disease. ProcNatl Acad Sci USA 100(4):2041–2046.

45. Bhuiyan MS, et al. (2013) Enhanced autophagy ameliorates cardiac proteinopathy.J Clin Invest 123(12):5284–5297.

46. Su H, Wang X (2011) p62 Stages an interplay between the ubiquitin-proteasomesystem and autophagy in the heart of defense against proteotoxic stress. TrendsCardiovasc Med 21(8):224–228.

47. He C, et al. (2012) Exercise-induced BCL2-regulated autophagy is required for muscleglucose homeostasis. Nature 481(7382):511–515.

McLendon et al. PNAS | Published online November 17, 2014 | E5185

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0

48. Maloyan A, Gulick J, Glabe CG, Kayed R, Robbins J (2007) Exercise reverses preamyloidoligomer and prolongs survival in alphaB-crystallin-based desmin-related cardiomy-opathy. Proc Natl Acad Sci USA 104(14):5995–6000.

49. Kopito RR (2000) Aggresomes, inclusion bodies and protein aggregation. Trends CellBiol 10(12):524–530.

50. Rubinsztein DC (2006) The roles of intracellular protein-degradation pathways inneurodegeneration. Nature 443(7113):780–786.

51. Aguzzi A, O’Connor T (2010) Protein aggregation diseases: Pathogenicity and ther-apeutic perspectives. Nat Rev Drug Discov 9(3):237–248.

52. Hull RL, Westermark GT, Westermark P, Kahn SE (2004) Islet amyloid: A critical entityin the pathogenesis of type 2 diabetes. J Clin Endocrinol Metab 89(8):3629–3643.

53. Wang X, et al. (2001) Mouse model of desmin-related cardiomyopathy. Circulation103(19):2402–2407.

54. Bulteau A-L, et al. (2001) Oxidative modification and inactivation of the proteasomeduring coronary occlusion/reperfusion. J Biol Chem 276(32):30057–30063.

55. Sarikas A, et al. (2005) Impairment of the ubiquitin-proteasome system by truncatedcardiac myosin binding protein C mutants. Cardiovasc Res 66(1):33–44.

56. Yi J, et al. (2013) Sodium valproate alleviates neurodegeneration in SCA3/MJD viasuppressing apoptosis and rescuing the hypoacetylation levels of histone H3 and H4.PLoS ONE 8(1):e54792.

57. Zhang X, et al. (2007) HDAC6 modulates cell motility by altering the acetylation levelof cortactin. Mol Cell 27(2):197–213.

58. Wang L, de Zoeten EF, Greene MI, Hancock WW (2009) Immunomodulatory effects ofdeacetylase inhibitors: Therapeutic targeting of FOXP3+ regulatory T cells. Nat RevDrug Discov 8(12):969–981.

59. Brochier C, et al. (2013) Specific acetylation of p53 by HDAC inhibition prevents DNAdamage-induced apoptosis in neurons. J Neurosci 33(20):8621–8632.

60. Winter L, et al. (2014) Chemical chaperone ameliorates pathological protein aggre-gation in plectin-deficient muscle. J Clin Invest 124(3):1144–1157.

61. Lee JY, et al. (2010) HDAC6 controls autophagosome maturation essential for ubiquitin-selective quality-control autophagy. EMBO J 29(5):969–980.

62. Committee for the Update of the Guide for the Care and Use of Laboratory Animals(2011) Guide for the Care and Use of Laboratory Animals (National Academies Press,Washington, DC).

E5186 | www.pnas.org/cgi/doi/10.1073/pnas.1415589111 McLendon et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 29

, 202

0