Shear stress triggers insertion of voltage-gatedpotassium channels from intracellular compartmentsin atrial myocytesHannah E. Boycotta,b,c,d, Camille S. M. Barbiera,b,c,d, Catherine A. Eichela,b,c,d, Kevin D. Costaa,e,Raphael P. Martinsa,b,c,d, Florent Louaulta,b,c,d, Gilles Dilaniana,b,c,d, Alain Coulombea,b,c,d,Stéphane N. Hatema,b,c,d,f,1, and Elise Balsea,b,c,d,1,2

aLIA/Transatlantic Cardiovascular Research Center, Université Pierre et Marie Curie, 75005 Paris, France and Mount Sinai School of Medicine, New York,NY 10029-6574; bICAN, Institute of Cardiometabolism and Nutrition, 75013 Paris, France; cUniversité Pierre et Marie Curie-Paris 6, 75005 Paris, France;dInstitut National de la Santé et de la Recherche Médicale, Unité Mixte de Recherche Scientifique 956, 75013 Paris, France; eCardiovascular Research Center,The Icahn School of Medicine at Mount Sinai, New York, NY 10029-6574; and fDepartment of Cardiology, Assistance Publique-Hôpitaux de Paris, HôpitalPitié-Salpêtrière, 75013 Paris, France

Edited by Richard W. Aldrich, University of Texas at Austin, Austin, TX, and approved August 28, 2013 (received for review May 25, 2013)

Atrial myocytes are continuously exposed to mechanical forcesincluding shear stress. However, in atrial myocytes, the effects ofshear stress are poorly understood, particularly with respect to itseffect on ion channel function. Here, we report that shear stressactivated a large outward current from rat atrial myocytes, witha parallel decrease in action potential duration. The main ion chan-nel underlying the increase in current was found to be Kv1.5, therecruitment of which could be directly observed by total internalreflection fluorescence microscopy, in response to shear stress. Theeffect was primarily attributable to recruitment of intracellularpools of Kv1.5 to the sarcolemma, as the response was preventedby the SNARE protein inhibitor N-ethylmaleimide and the calciumchelator BAPTA. The process required integrin signaling throughfocal adhesion kinase and relied on an intact microtubule system.Furthermore, in a rat model of chronic hemodynamic overload,myocytes showed an increase in basal current despite a decreasein Kv1.5 protein expression, with a reduced response to shear stress.Additionally, integrin beta1d expression and focal adhesion kinaseactivation were increased in this model. This data suggests that,under conditions of chronically increased mechanical stress, theintegrin signaling pathway is overactivated, leading to increasedfunctional Kv1.5 at the membrane and reducing the capacity of cellsto further respond to mechanical challenge. Thus, pools of Kv1.5may comprise an inducible reservoir that can facilitate the repolari-zation of the atrium under conditions of excessive mechanical stress.

trafficking | cardiomyocytes | potassium current

The electrical properties of the myocardium are governed bythe interplay of ion channels, whose expression and regula-

tion determines the precise electrical responses of the tissue. Theactivity of ion channels can be regulated in a variety of ways: forexample, interaction with accessory subunits (1), phosphoryla-tion (2), oxidation state (3), and gene expression (4). Recently,increased attention has been focused on trafficking as a means toregulate ion channel function, notably by modulating the numberof active channels present at the plasma membrane (5, 6). Thisregulation is a complex process derived from a balance betweentrafficking of newly synthesized channel, endocytosis, and recy-cling/degradation. Trafficking of ion channels is known to bea dynamically regulated process that depends on Rab-GTPasesas well as dynamin motors (7, 8). Indeed, certain antiarrhyth-mogenic drugs have been shown to exert their activity bymodifying the number of channels at the plasma membrane (9).In this context, we previously showed that ion channels arerecruited from a submembranous pool in response to cholesteroldepletion (10). In addition, several ion channels are regulated bymechanical forces, which directly affect the gating of the channel

(11) or indirectly activate intracellular signaling pathways to alterchannel properties (12).The myocardium is subjected to a variety of forces with each

contraction and therefore must adapt to the various associatedmechanical stresses. The response to stretch has been well-studied and includes gene regulation (13), activation of stretch-activated ion channels (14–17), and the release of atrial natri-uretic peptide (ANP) (18–21). In addition, it has been shownthat direct stretch of β1 integrins activates ICl,swell as well asa cation current (22). Less well-studied are the responses ofcardiomyocytes to shear stress. Shear forces in the myocardiumarise from blood flow and the relative movement of sheets ofmyocytes, causing cell deformation as the myocardial layers slideagainst each other with each heart beat (23, 24). Although theeffect of shear stress upon cardiomyocytes has not been exten-sively explored, it has been shown that increased shear stressstimulates intracellular calcium transients (25, 26), induces anincrease in the beating rate of neonatal ventricular myocytes(27), and triggers propagating action potentials (APs) in mono-layers of ventricular myocytes (28). Thus far, the response toshear stress remains relatively unknown, particularly with regardto ion channel regulation. Ion channel activity determinesboth the shape of the AP and the firing frequency of excitablecells. Therefore, the response of cardiomyocytes to shear stress

Significance

The heart is continuously subjected to mechanical forces. Theatria in particular are susceptible to changes in the mechanicalenvironment due to their unique position as “pressure sen-sors.” Here, we show that increased shear stress induces therecruitment of potassium channels from intracellular storagepools to the plasma membrane, via signaling pathways thatlink the extracellular matrix to the cytoskeleton. This processis altered in myocytes experiencing chronically increased me-chanical stress. The incorporation of channels into the mem-brane causes changes in the electrical activity of the myocyteand may be an important way for cells to adapt to increasedmechanical forces.

Author contributions: H.E.B., A.C., S.N.H., and E.B. designed research; H.E.B., C.S.M.B., C.A.E.,K.D.C., R.P.M., F.L., G.D., A.C., and E.B. performed research; H.E.B., C.S.M.B., C.A.E., K.D.C.,R.P.M., F.L., A.C., S.N.H., and E.B. analyzed data; and H.E.B., S.N.H., and E.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1S.N.H. and E.B. contributed equally to this work.2To whom correspondence should be addressed. E-mail: elise.balse@upmc.fr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1309896110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1309896110 PNAS | Published online September 24, 2013 | E3955–E3964

PHYS

IOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

is important for normal cardiac excitability and could be centralin pathological conditions in which the working conditions ofthe myocardium are altered.In this study, we investigate the response of native adult rat

cardiomyocytes to shear stress, reproduced in vitro by laminarflow. Using a combination of whole-cell patch-clamp and single-channel recordings, high spatial resolution 3-dimensional andtotal internal reflection fluorescence (TIRF) microscopy, weshow that shear stress induces an increase in outward current andshortens AP duration within the range of a few minutes. Thisphenomenon is saturable and reversible, and is caused by Kv1.5exocytosis from the recycling endosome. We identify the mech-anotransduction pathway of this recruitment, which involvesintegrin/focal adhesion kinase (FAK) signaling. Finally, the re-sponse to shear stress is altered in chronically hemodynamicallyoverloaded and dilated atria.

ResultsShear Stress Causes an Increase in Outward Current from AtrialMyocytes. The effect of increased shear stress on atrial myo-cytes was investigated using the whole-cell patch-clamp tech-nique, at a membrane potential of +60 mV. As shown by Fig. 1A(and Fig. S2), increasing shear stress from 0.5 dyn·cm−2 to 4dyn·cm−2 elicited an increase in outward current from 4.5 ± 0.3pA/pF to 52.7 ± 2.3 pA/pF (n = 43, P < 0.001) without change inmembrane capacitance. Shear stress of 4 dyn·cm−2 did not in-duce current increase in ventricular myocytes (n = 10) (Fig. 1A).The effect was observed with a delay of 4–8 min, with a meanonset of 281 ± 23 s (n = 43). The response was slowly reversiblewith a t1/2 recovery of 781 ± 54 s (n = 9). It is noteworthy that∼60% of atrial cells tested responded to shear stress, and this

response was isolation-dependent. Shear stresses between 0.5and 10 dyn·cm−2 were tested, and the response was found to havea threshold of 2.8 dyn·cm−2. There was no change in the mag-nitude or kinetics of the response from 2.8 dyn·cm−2 to 10dyn·cm−2, indicating that if the system was activated sufficiently(2.8 dyn·cm−2) there was an “all-or-nothing” effect. To provide asuprathreshold stimulus, most studies were performed at a shearstress of ∼4 dyn·cm−2.

Estimation of Shear Stress in the Rat Atrial Myocardium. To evaluatethe relevance of the shear stress value used in vitro, interlaminarshear stress in the rat atria was estimated using the Couette flowmodel (Fig. S3). In this simplified model, shear stress is calcu-lated on the surface of two parallel plates, one of which is slidingrelative to the other. Shear stress (τ) is dependent on the vis-cosity of the fluid (μ), the distance between the plates (d), andthe velocity of the sliding plate (Uo). Due to the extremelylimited amount of data relating to the mechanics and anatomy ofthe atria, certain assumptions were made to estimate the shearstress (the choices of the parameters used to measure shearstress are detailed in Fig. S3). The plate velocity (Uo) can beestimated from the shear strain rate and the dimensions of myo-cardial laminae. Both myocardial laminae (19.9 ± 3.7 μm, n = 9)and gaps (6.5 ± 2.3 μm, n = 9) were measured from represen-tative phase-contrast images of the rat left atria using ImageJ (Fig.S3A). These measures were used to calculate the shear stress ofthe system at a resting heart rate in adult rat of ∼5 Hz [300 beatsper minute (bpm)], and the calculated atrial shear stress wasfound to be 0.43 dyn·cm−2 (Materials and Methods). This roughapproximation of the atrial shear stress is within an order ofmagnitude of the experimental threshold value of 2.8 dyn·cm−2.

-100-80-60-40 -20 20 40 60-10

10

20

30

40

50

604 dyn.cm-2 0.5 dyn.cm-2

+ DIDS4 dyn.cm-2+ 4-AP

Time (s)0 200 400 600 800 1000

AP

dura

tion

at 9

0%

repo

laris

atio

n (m

s)

0

20

40

60

804 dyn.cm-2

VENTRICLE

ATRIA

Cur

rent

Den

sity

(pA

/pF)

Cur

rent

Den

sity

(pA

/pF)

Time (s)

Time (s)

B Current Density (pA/pF)

mV

-80mV

Before shear

4 dyn.cm-2

200 ms1 nA

050

20

40

60

300 350 400 450 600 800 1000 1200 1400

0 200 400 600 8000

20

40

60

80

Time (ms)40 60 80 100 120 140 160

0mV

4 dyn.cm-2

BEFORE SHEAR 20mV

I(DR

UG

)/I(m

ax.s

hear

)

0 .0

0 .2

0 .4

0 .6

0 .8

1 .0

1 .2

*

ns

*

***

***

***

***

ns1.2

1.0

0.8

0.6

0.4

0.2

0.0

A

C

E

D

F

Fig. 1. Shear stress causes an increase in outwardcurrent from atrial myocytes. (A) Time series ofcurrent density showing the increase in sustainedoutward current recorded from atrial myocytes(pulse from −80 mV to +60 mV) induced by 4dyn·cm−2 shear stress (n = 37), and recovery under0.5 dyn·cm−2 (n = 9). Also shown are time seriesrecorded from ventricular myocytes subjected to thesame degree of shear stress (n = 9). (B) Current-voltage relationships obtained by depolarizing themembrane from −100 mV to +60 mV in 10 mVincrements before (n = 10) and after shear stress (n =8). Inset shows representative current traces fromthe same myocyte before and after shear stress(currents shown from −100 mV to +60 mV, 20 mVincrements). (C) Pharmacological properties of theshear stress-induced current illustrated as ratio(IDrug/Imax·shear): K+ channel blocker 4-AP (100μM, n = 26; ***P < 0.001; 1 mM, n = 14; ***P <0.001), the Kv1.5 inhibitor AVE0118 (10 μM, n = 8;***P < 0.001), the general potassium channelblocker TEA (20 mM, n = 8; n.s.), the chloride chan-nel blockers tamoxifen (20 μM, n = 8; *P < 0.05) andDIDS (100 μM, n = 6; *P < 0.05), 4-AP (1 mM) plusDIDS (100 μM) (n = 8; ***P < 0.001), and the stretch-activated ion channel blocker gadolinium (30 μM,n = 6; n.s.). (D) Time series of outward current den-sity from a representative myocyte in response toshear stress followed by the effect of 100 μM DIDSand 1 mM 4-AP. (E) Representative action potentialsrecorded from myocytes before and after shearstress. (F) Time series showing the reduction in ac-tion potential duration at 90% of repolarization inresponse to shear stress.

E3956 | www.pnas.org/cgi/doi/10.1073/pnas.1309896110 Boycott et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

Shear Stress Primarily Induces an Increase in Outward PotassiumCurrent. We characterized the nature of the current stimulatedby shear stress. The reversal potential of the shear stress-inducedcurrent approximately followed the calculated equilibrium po-tential for K+: at EK = −86 mV, EREV = −77.3 ± 1.2 mV (n =8); at EK = −6 mV, EREV = −9.6 ± 2.1 mV (n = 5), indicatingthat shear stress primarily activates a K+ conductance in atrialmyocytes. The voltage-gated K+ (Kv) channel blocker 4-amino-pyridine (4-AP) reversibly inhibited the shear stress-induced cur-rent by 59.7 ± 3.2% (n = 26; P < 0.001) and 69.2 ± 3.4% (n = 14;P < 0.001) at 100 μM and 1 mM, respectively (bar graph, Fig. 1C).Because at 100 μM 4-AP principally inhibits Kv1.5 channels (29),this data suggests that shear stress primarily modulates IKur. An-other recognized blocker of IKur, AVE0118 (30) (10 μM), alsoinhibited the shear stress-induced current by 49.3 ± 5.3% (n = 9;P < 0.001). The current increase was not inhibited by 20 mMtetraethylammonium (TEA) (n = 8, n.s., Fig.1C), further sup-porting the involvement of Kv1.5 channel in the shear-stress effect(31). The transient outward (Ito,fast and Ito,slow) was never in-creased upon shear stress, eliminating the involvement of Kv1.4/7and Kv4.2/3, the molecular basis of Ito.As K+ channel blockers did not fully inhibit the shear stress-

induced current, we assessed the involvement of other conduc-tances, such as chloride and stretch-activated ion channels(Fig.1C). The ICl,swell inhibitor tamoxifen (20 μM) reduced thecurrent by 23.2 ± 3.1% (n = 8; P < 0.05). However, tamoxifencan inhibit other currents in myocytes, including K+ currents(32). We used another inhibitor, 4,4′diisothio-cyanatostilbene-2,2′-disulfonic-acid (DIDS), at a concentration reported to spe-cifically block ICl,swell (100 μM) (33). DIDS inhibited the currentby 22.0 ± 3.1% (n = 6; P < 0.05). When 1 mM 4-AP was appliedfollowing 100 μM DIDS (bar graph, Fig. 1C and example time-course, Fig. 1D), the current was blocked to preshear stress levels(4.5 ± 0.4 pA/pF before shear vs. 11.9 ± 2.9 pA/pF after DIDSplus 4-AP, n = 8; n.s.). The current was insensitive to 30 μMgadolinium (n = 8; n.s.) excluding the involvement of stretch-activated ion channels (Fig. 1C). These results confirm that themajority of the shear stress-induced current is carried by K+,with ∼20% attributable to activation of chloride channels anda further 10% of the increased current remaining unidentified.The current density was increased across all voltages tested,

with an initial linear current-voltage relationship followed bya slight outward rectification (Fig. 1B). To investigate the bio-physical properties of the K+ current stimulated by shear stress,we isolated the 4-AP–sensitive component: total outward currentupon shear stress minus current recorded under 4-AP. The ac-tivation–Vm relationship of the 4-AP–sensitive current was shiftedto the left upon shear stress (unsheared, V0.5 = 12.2 ± 3.7 mV,n = 9 vs. sheared, V0.5 = −25.9 ± 2.4 mV, n = 5; P < 0.001), andthe slope factor (k) was decreased (unsheared, 18.3 ± 1.2 mV, n=9 vs. sheared, 13.9 ± 1.0 mV, n= 5, P < 0.001). By mathematicallyremoving the leftward shift of the activation–Vm relationship, wefound that only ∼20% of the increase of the 4-AP–dependent com-ponent was due to the activation change (Materials and Methods).As K+ currents have precise roles in shaping the AP, we in-

vestigated the effect of the shear stress-induced current on theAP duration measured at 90% repolarization (APD90) in freshlyisolated atrial myocytes. Indeed, as shown in Fig. 1E, shear stressshortened APD90 from 51.4 ± 4.5 ms to 22.1 ± 0.9 ms (n = 4,P < 0.01), with a time course that mirrored the increase in cur-rent (Fig. 1F). No change in the resting membrane potential wasobserved. These data are supportive of the primary ion channelinvolved in the response to shear stress being Kv1.5, this channelbeing responsible for the repolarization phases of the cardiac AP.

Shear Stress Increases the Surface Density of Potassium Channels.We previously showed that Kv1.5 is recruited from a subme-mbranous pool in response to cholesterol depletion (10). To

examine the hypothesis that the increase in current caused byshear stress results from K+ channel recruitment, atrial myocyteswere transduced with adenovirus encoding EGFP-Kv1.5. Thedistribution of EGFP-Kv1.5 was similar to Kv1.5 in nativemyocytes (freshly isolated or cultured myocytes) (Fig. S5). TheEGFP-Kv1.5 fluorescence measured in real time at the sarco-lemma by TIRF microscopy showed a progressive increase uponshear stress (Fig. 2A and Movie S1). The relative evanescent fieldfluorescence at the start of the recording was 2.6 ± 2.3% (n =10) and increased to a plateau of 75.6 ± 6.9% (n = 10) at ∼480 safter the initiation of shear stress (P < 0.001) (Fig. 2B). In pre-shear stress conditions, EGFP-Kv1.5 channels showed a punctatedistribution at the sarcolemma (vesicle mean size, 0.11 ± 0.01μm2, n = 6) with dynamic movement in the x/y axis (Movie S2).Shear stress of 4 dyn·cm−2 induced the development of an in-crease in brightness (t1/2 = 119.7 ± 10.2 s, n = 10) correspondingto accretion of EGFP-Kv1.5–containing vesicles into clusters inthe sarcolemma (cluster mean size, 11.44 ± 1.38 μm2, n = 6).The insertion of functional K+ channels was further inves-

tigated using single-channel recordings. After 4- to 8-min expo-sure to shear stress, an increase in channel activity was observedin freshly isolated atrial myocytes (Fig. 2C). Upward openings ofK+ channels were observed under control conditions (meancurrent from channel openings, 1.3 ± 0.7 pA, n = 4) with anelementary conductance of 12.4 ± 2.5 pS (n = 4). After ∼6 minof shear stress, a marked increase in channel activity occurred,reaching a maximum value of 9.2 ± 3.5 pA (n = 6). Exampletraces from the start of the recording, and from time pointsduring the exponential phase of the increase in channels (310 sand 350 s) are shown in Fig. 2C. Amplitude histograms havebeen determined from the corresponding current traces (Fig.2D) and are indicative of the increase in the number of activechannels present in the membrane patch upon shear stress.Taken together, these experiments indicate that shear stressstimulates the recruitment and the accretion of functional Kv1.5channels at the sarcolemma.

Shear Stress Stimulates Channel Exocytosis from the RecyclingEndosome and Requires an Intact Microtubule Network. Becausevesicles are known to use the microtubule system to transportcargo (34–36), we used colchicine to destroy the microtubulenetwork and examined the effect of shear stress. EGFP-Kv1.5–infected atrial myocytes showed association with the microtubulenetwork that was disrupted upon treatment with 10 μM colchicine(Fig. 3A). Colchicine treatment (2 h) attenuated the response toshear stress (17.9 ± 3.0 pA/pF vs. shear stress in control condition,52.6 ± 2.3 pA/pF, n = 8, P < 0.001), but the response was notcompletely prevented (colchicine vs. control, 4.3 ± 0.3 pA/pF, n =8, P < 0.05) (Fig. 3 B and C). In contrast, destruction of the actincytoskeleton with cytochalasin-D (5 μM, 24 h) did not affect theresponse of atrial myocytes to shear stress (4 dyn·cm−2 control,26.2 ± 8.1 pA/pF vs. 4 dyn·cm−2 with cytochalasin-D, 25.7 ± 7.9pA/pF, n.s.) (Fig. S4), suggesting that the actin cytoskeleton is notneeded for the recruitment of vesicles in response to shear stress.Exocytosis is reliant upon fusion of vesicles to the plasma

membrane, a process mediated by SNARE proteins. Intracellulardialysis of the SNARE protein inhibitor N-ethylmaleimide(NEM) (250 μM) followed by 4 dyn·cm−2 shear stress preventedthe increase in current (n = 8) (Fig. 3 D and E). As intracellularcalcium may be required for exocytosis, we tested the calcium de-pendency of the shear stress-induced current in freshly isolatedatrial myocytes. Intracellular dialysis of the calcium-specific chelator1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA)(10 mM) prevented the increase in current (n = 8) (Fig. 3 Dand E).To identify the compartment responsible for delivery of Kv1.5

channels, cultured atrial myocytes were transfected with a domi-nant negative (DN) form ofRab11 or with empty vector. Rab11DN

Boycott et al. PNAS | Published online September 24, 2013 | E3957

PHYS

IOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

overexpression significantly reduced the effect of shear stress(39.5 ± 9.0 pA/pF in empty vector, n= 8, vs. 12.4 ± 7.0 pA/pF inRab11DN, n = 6; P < 0.001) (Fig. 3F), indicating that the in-creased current is due to recruitment of channels from theRab11-associated recycling endosome.

Shear Stress Stimulates the Integrin/FAK Signaling Pathway. Integ-rins are central in mechanotransduction processes, conveyingmechanical forces into the cell. To investigate whether the in-tegrin system is involved in the response to shear stress in atrialmyocytes (22), we used the “disintegrin” echistatin toxin. Pre-incubation with echistatin (100 nM) for 30 min prevented theresponse to shear stress in freshly isolated atrial myocytes(n = 10) (Fig. 4 A and B).Focal adhesion kinase (FAK) is an early downstream effector

of integrins and has been shown to be activated in stretchedcardiomyocytes (22). The involvement of FAK in the responseto shear stress was studied by measuring phosphorylation ofFAK tyrosine 397. Cultured atrial myocytes were subjected to4 dyn·cm−2 shear stress in a laminar flow chamber, and the amountof FAKY397 was examined by deconvolution microscopy (Fig.4C). When quantified, FAKY397 fluorescence increased from0.24 ± 0.04 arbitrary units (AU) in unsheared myocytes to 0.67 ±0.06 AU in shear stressed myocytes (n = 3, n = 10 fields; P <0.001) (Fig. 4C). The subcellular distribution of FAKY397 wasalso modified upon shear stress: whereas FAKY397 staining wasrestricted to cell-to-cell contacts in unsheared cardiomyocytes,FAK397 was also detected in the entire cell bodies of shearedmyocytes (Fig. 4C, enlargements). FAK inhibitor 14 (1,2,4,5-benzenetetramine tetrahydrochloride) was used to prevent theactivation of FAK. When dialyzed intracellularly with FAK in-hibitor 14, freshly isolated atrial myocytes no longer respondedto shear stress (Fig. 4 D and E).Altogether, these results indicate that integrin is the mecha-

nosensor that conveys the shear stress signal via the activation ofFAK to trigger the release of Kv1.5 channels to the sarcolemma.

Up-Regulation of the Integrin/FAK Signaling Pathway Is Associatedwith a Reduced Response to Shear Stress in HemodynamicallyOverloaded Atria. To investigate the importance of the shearstress-induced current in situ, we used a rat model of heart

failure (HF) and atrial hemodynamic overload, dilation, and fi-brosis (37–39). As similar models have been shown to inducealterations in the integrin/FAK system (the mechanosensor) (40–42), we hypothesized that the response of myocytes to shearstress would be altered. As previously described (38), in thismodel, left-ventricular dysfunction (ejection fraction in HF rats,55.9 ± 3.6%, n = 11 vs. sham, 79.4 ± 0.6%, n = 11; P < 0.001)was associated with a marked atrial dilation as assessed byechocardiography (surface of left atria in HF rats, 0.37 ± 0.02cm2, n = 11 vs. sham, 0.19 ± 0.01 cm2, n = 11; P < 0.001).The expression of integrinß1 was increased twofold in HF rat

atria as quantified by Western blot (P < 0.001) and illustrated byintegrinß1D staining in atrial slices (Fig. 5A and Fig. S6). Westernblot quantification of FAK expression showed a 20% increase inHF rat atria (P < 0.001). Moreover, the endogenous activation ofFAK was also increased in myocytes from dilated atria as shown byimmunocytochemistry performed on freshly isolated cells (Fig. 5B).The membrane capacitance of atrial myocytes isolated from

HF rats was increased (74.2 ± 4.5 pF, n = 13) compared withsham 61.2 ± 3.8 pF (n = 11, P < 0.05), indicative of myocytehypertrophy. In preshear conditions, IKur was increased in hy-pertrophied myocytes (5.1 ± 2.3 pA/pF, n = 13) compared withsham (3.7 ± 1.4 pA/pF, n = 11, P < 0.05) (Fig. 5C, Upper). Incontrast, the amount of Kv1.5 protein was decreased by ∼40%(P < 0.05) in dilated atria (Fig. 5C, Lower). We cannot excludethat increased extracellular matrix protein accumulation and fi-broblast proliferation had resulted in an apparent decrease inKv1.5 from myocytes. However, the mRNA expression of theKCN5A (Kv1.5) gene was also reduced although this did notreach significance, which indicates that KCN5A was not stimu-lated during this hypertrophic process (Fig. S6). These observa-tions supported an up-regulation of integrin/FAK signaling inthe hemodynamically overloaded and dilated atria, which couldcontribute to the increase in density of functional channels at thesarcolemma, despite no increase of Kv1.5 protein.Due to the changes in the mechanosensor, we investigated the

response to shear stress in hypertrophied myocytes. The responseto shear stress was modified in hypertrophied myocytes. Firstly,the number of hypertrophied myocytes responding to shear stresswas decreased compared with sham myocytes [53% (13/28) v.s.78% (11/14), respectively] (Fig. 5D). Secondly, the kinetics of the

A B

C DN

umbe

r of e

vent

s

Current amplitude

0

200

100

400

300

500

4 pA

0C L1 L2

4 pA

0.5 dyn.cm-2

300

200

100

400

500

0

100

200

300

C L1 L2 L3

300

200

100

0

4 dyn.cm-2 310 s

0

50

100

150

C L1 L2 L3 L4

150

100

50

0

4 dyn.cm-2 350 s

0.5 dyn / cm2

2

4 dyn / cm2 310 s

2

C

4 dyn.cm-2 310s

4 dyn / cm2 350 s4 dyn.cm-2 350s

C

0.5 dyn.cm-2

C

100 ms

5 pA

500 sec300 sec200 sec

0 sec 100 sec 150 sec

***

Max

.fluo

resc

ence

(%Δ

EFF

)

0

100

80

60

40

20

0 100 200 300 400 500

0

20

40

60

80

100

%Δ

EFF

Time (s)

4 dyn.cm-2

0.5 dyn.cm-2

Fig. 2. Shear stress stimulates the recruitment offunctional Kv1.5 channels. (A) Representative TIRFimages recorded from a single myocyte expressingEGFP-Kv1.5 upon shear stress. The dashed linesoutline the cell border. (Scale bar: 10 μm.) (B, Left)Mean time course of the increase in fluorescence inresponse to shear stress vs. preshear (t1/2, 120 ± 5s,n = 10). (Right) Bar graph showing the percent in-crease in average fluorescence before and aftershear stress (n = 10; ***P < 0.001). (C) Samplechannel recordings from cell-attached membranepatches of adult rat atrial myocytes during 750 mspulses from −80 to +80 mV before application ofshear stress, after 310 s 4 dyn·cm−2 shear stress andafter 350 s 4 dyn·cm−2 shear stress. (D) Amplitudehistograms determined from the correspondingcurrent traces indicated. C, closed state; L1–L4, lev-els corresponding to the successive measurablecurrent levels that are the indication of the numberof active channels present in the membrane patch.Gaussian fitted curves corresponding to each cur-rent level are also shown.

E3958 | www.pnas.org/cgi/doi/10.1073/pnas.1309896110 Boycott et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

response were substantially slower in hypertrophied myocytes:maximum slope of 0.70 ± 0.09 pA/(pF·s), n = 13 vs. 1.18 ± 0.13pA/(pF·s) in control, n= 11 (P <=0.05) (Fig. 5 E and F). Finally,the percent current increase induced by shear stress was reducedin hypertrophied myocytes: whereas sham myocytes showed ∼15-fold increase in current density (54.7 ± 7.8 pA/pF; n = 11), hy-pertrophied myocytes only showed approximately 9-fold increase(45.2 ± 4.5 pA/pF; n = 13) (Fig. 5G). These results indicate thatthe shear stress-induced current is altered in dilated and fibroticatria, which is also characterized by an up-regulation of theintegrin/FAK signaling pathway.

DiscussionIn this study, we describe a previously undescribed mechanism bywhich voltage-gated ion channels are regulated in atrial myocytes(for summary, see Fig. 6). Shear stress stimulates a large outward

current within 4–8 min, the majority of which is carried by K+

channels despite a small but significant contribution from chlo-ride channels. Using direct measurement and visualization ofchannels, we show that shear stress triggers Kv1.5 channel de-livery to the sarcolemma. Integrins are the sensors of shear stressand convey the mechanical stimulation intracellularly to theRab11-associated slow-recycling endosome, via activation ofFAK. Consequently, shear stress represents a previously unde-scribed means to modulate the density of repolarizing currentand therefore to tune the electrical activity of atrial myocytes.Constitutive exocytosis shares several common features with

regulated (fast) exocytosis. Cargo proteins are transported intovesicles, tethered to microtubules and the actin cytoskeleton viasmall Rab GTP-ases, and inserted into the membrane after fu-sion of donor and acceptor membranes, under the control ofSNARE proteins (43, 44). Three important parameters emerge

CONTROL

CONTROL

NEMBAPTA

Empty vector Rab 11 DN

***

ns

4 dyn.cm-2

Cur

rent

Den

sity

(pA

/pF)

Cur

rent

Den

sity

(pA

/pF)

Cur

rent

Den

sity

(pA

/pF)

Time (s)

***ns

ns

*** ***

Cur

rent

Den

sity

(pA

/pF)

*

-tubulin

GFP

Overlay

CONTROL COLCHICINE

21 COLCHICINE

4 dyn.cm-2

Cur

rent

Den

sity

(pA

/pF)

Time (s)

2

1

0

20

40

60

0

20

40

60

0 100 200 300 400 5000

20

40

60

0

20

40

60

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 00

2 0

4 0

6 0

8 0

1 0 0

**

100

60

40

20

0

80

1000 200 300 400 500

A B C

D E F

Fig. 3. Shear stress triggers Kv1.5 channel exocytosis. (A) Immunocytochemistry showing the localization of EGFP-Kv1.5 (green) and association withmicrotubules (red) in the absence or presence of 10 μM colchicine. (Scale bar: 10 μm.) Nuclei are stained with DAPI (blue). (B) Time series of current densityrecorded from an atrial myocyte treated or not with 10 μM colchicine and subjected to shear stress. Insets are enlargements of region of interest (ROI) in Fig.3A showing that 2 h colchicine treatment is sufficient to destroy the microtubule network in atrial myocytes stained with anti-tubulin antibody. (C) Bar graphshowing the average current recorded before (white bar) and after shear stress in control medium (black bar) (vs. before shear, n = 8; ***P < 0.001) and after10 μM colchicine treatment (red bar) (n = 8; *** P < 0.001). Note that under colchicine treatment the shear stress-induced current is reduced but not abolished(n = 8; *P < 0.05). (D) Time series of current density from a representative myocyte subjected to shear stress in conjunction with the intracellular dialysis of theSNARE protein inhibitor 250 μM NEM (red), or 10 mM BAPTA (blue). (E) Bar graph showing average current densities in response to shear stress in controlintracellular solution (n = 8; ***P < 0.001) or in the presence of 250 μM NEM (vs. before shear, n = 8; n.s.) or 10 mM BAPTA (vs. before shear, n = 8; n.s.). (F) Bargraph showing the average current density from myocytes transfected with empty vector (n = 8) or Rab11DN (n = 6) and subjected to shear stress. The effectof shear stress was significant only with the empty vector (***P < 0.001).

Boycott et al. PNAS | Published online September 24, 2013 | E3959

PHYS

IOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

from our experiments: the time dependency, the saturability, andthe reversibility of the shear-stress effect. Despite the increasein fluorescence observed with TIRF microscopy preceding themean onset of the current increase, initiation of both processesoccurred with a delay between the onset of shear stress and theonset of the response, suggesting the involvement of an in-tracellular cascade. However, once initiated, both EGFP-Kv1.5fluorescence and current increase followed a Boltzmann-likedistribution and reached a plateau at ∼400 s. These two obser-vations suggest that (i) the reservoirs of channels recruited byshear stress are not located just beneath the sarcolemma and (ii)a single compartment is likely responsible for the delivery ofKv1.5 channels. Our results support Kv1.5 exocytosis, as theshear stress-induced current increase was abolished by theSNARE inhibitor NEM and by intracellular calcium bufferingof submembrane domains with BAPTA. Moreover, we dem-onstrated dependency on the Rab11-associated slow recyclingendosome and an intact microtubule network for the effect tooccur. This finding is reminiscent of our previous observationthat cholesterol depletion stimulated trafficking of Kv1.5 fromthe recycling endosome to the sarcolemma in ∼7 min (10, 45). Inaddition, the delivery to the sarcolemma of the Glut4 transportertriggered by insulin stimulation is dependent on Rab11 withinthe same time scale (46). The Rab11-associated endosome isusually considered a slow route for recycling of integral proteins.For instance, the turnover of Kv1.5 from the recycling endosometo the plasma membrane takes ∼72 h (47). Together with ourprevious observations (10), these results point to the slow recy-cling endosome being a preferential storage compartment forstimulated exocytosis of Kv1.5 channels. Finally, the current re-quired over 20 min to return to baseline levels. This observationsuggests that compensatory endocytosis mechanisms occur. Atthe present time, the endocytosis pathways for Kv1.5 channelshave not been elucidated, nor are the kinetics of constitutiveendocytosis in atrial myocytes known.It is noteworthy that changes in the activation properties of the

4-AP–sentitive component account for ∼20% of the shear-acti-vated current. One can speculate that accretion of Kv1.5 channelsupon shear stress into distinct lipid and protein environments canmodulate their biophysical properties. Such changes in the voltagedependency and activation slope of Kv1.5-mediated current have

been reported previously, for example, when associated with Kvβ-subunits (1) or following cholesterol/caveolae modifications (48).The importance of integrins in regulating the activity of ion

channels has been demonstrated in several cell types, includingvascular smooth muscle (49, 50), endothelium, (51) and neurons(52). In the myocardium, integrin activation is necessary formechano-electrical transduction, as evidenced by the fact that theincreased beating rate caused by shear stress in ventricular myo-cytes can be prevented by integrin inhibition (27). Integrins are alsoinvolved in calcium homeostasis and regulate the L-type calciumchannel (53). In addition, and Baumgarten showed that directstretch of integrins activates a mixed (chloride/cationic) current viaactivation of FAK in ventricular myocytes (22). Here we show thatthe shear stress-induced current is preventable with echistatin andFAK inhibitor 14. As such, the integrin/FAK transduction pathwayprobably constitutes a major mechanism by which myocytes mod-ulate their electrical activity in response to mechanical stimuli.The range of shear stress used in this study is consistent with

that shown to have physiological effects on cardiomyocytes inother reports. The 1 dyn·cm−2 shear stress affects the APD ofneonatal rat ventricular myocytes (NRVM) (54) whereas 0.5–5dyn·cm−2 activates ERK in a 3D cardiomyocyte array (55). Inaddition, shear stresses of 5–10 dyn·cm−2 induce differentiationof mesenchymal stem cells into cardiomyocytes capable of ANPsecretion (56, 57). There is currently no technique to directlymeasure the shear stress experienced by cardiomyocytes in theworking myocardium. Using the Couette flow model, the shearstress was estimated to be ∼0.5 dyn·cm−2 in rat atria (Fig. S3).This value is below the threshold of 2.8 dyn·cm−2 and is consis-tent with the increase in current seen in this study being a re-sponse to elevated shear stress. Changes in shear stress level couldoccur physiologically as a result of increased heart rate or in-creased contractile reserve, or pathophysiologically as a result ofincreased laminar thickness (hypertrophy) or decreased gapdistance (fibrosis). In hemodynamically overloaded and dilatedatria, we observed increased expression of integrinβ1D, as well asFAK activation, suggesting an overactivation of the integrin/FAK signaling pathway as previously reported by others (40–42).In addition, in this model, the capacity of myocytes to respond toshear stress was reduced. Taken together, these data support theidea that the shear stress-induced current increase is affected

Unsheared 4 dyn.cm-2

ECHISTATIN

***ns

Cur

rent

Den

sity

(pA

/pF)

0

20

40

60

CONTROL

4 dyn.cm-2

Cur

rent

Den

sity

(pA

/pF)

Time (s)0 200 400 600 800

0

20

40

60

FAKi14

CONTROL

4 dyn.cm-2

Cur

rent

Den

sity

(pA

/pF)

Time (s)0 200 400 600

0

20

40

60

Cur

rent

Den

sity

(pA

/pF)

***ns

800 0

20

40

60

***

Inte

nsity

(AU

)

0.8

0.6

0.4

0.2

0

FAK397 -ac nin DAPI

A B C

D E

Fig. 4. Integrin/FAK signaling is the mechanotrans-ducer for shear stress effect. (A) Time series ofcurrent density recorded from a myocyte pretreatedfor 30 min with the disintegrin echistatin (100 nM)and subjected to shear stress. (B) Bar graph showingaverage current before (white bar) and after shearstress in control medium (black bar) (vs. before shear,n = 10; ***P < 0.001). The 100 nM echistatin treat-ment (red bar) prevented the effect of shear stress(vs. no shear, n = 10; n.s.). (C, Upper) Shear stressincreases FAK activation in cultured atrial myocytesstained with anti-phospho FAKY397 antibody (red),α-actinin (green), and DAPI (blue). (Scale bar: 10 μm.)Arrows indicate FAKY397 at cell-to-cell contacts inunsheared and sheared myocytes. Note thatFAKY397 is redistributed in shear-stressed myocytesas show by enlargement of ROI. (Scale bar: 5 μm.)(Lower) Bar graph quantifying FAK activation (vs.before shear, n = 3; ***P < 0.001). (D) Time series ofcurrent density recorded from a myocyte subjected toshear stress in the presence of intracellular FAK in-hibitor FAKi14 (50 μM). (E) Bar graph showing theaverage current amplitude of the response before(white bar) and after shear stress in control (blackbar) (vs. before shear, n = 11; ***P < 0.001) or in thepresence of the focal adhesion kinase inhibitorFAKi14 (red bar) (vs. before shear, n = 11; n.s.).

E3960 | www.pnas.org/cgi/doi/10.1073/pnas.1309896110 Boycott et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

when the working conditions of the atria are altered. This ex-cessive recruitment of potassium channels following shear stressmight overcompensate for the decrease in channel content in thediseased atrial myocardium. These results constitute further ev-idence that the shear stress-induced current is functional in situ. Itis noteworthy that, in the working myocardium, various mechan-ical stresses are likely to activate the integrin-signaling pathwayincluding direct stretching of the integrins anchored to the ex-tracellular matrix; these mechanical stresses might also trigger therecruitment of Kv1.5 channels to the plasma membrane.

Physiological Relevance. As the shear stress response was atria-specific, we speculate that the increased current (and subsequent

decreased APD) is related to the function of the atria. The atriahave an important reservoir function that is essential for both thefilling of ventricles and ANP secretion. Indeed, ANP secretion isaffected by changes in the extracellular space and fluid volume, bothof which may impact the shear stress of the atria (58). ANP releaseis calcium- (59, 60) and stretch-dependent (19, 20). Additionally,there is evidence that shear stress elicits calcium sparks in atrialmyocytes (26) and that acceleration of membrane repolarization(such as the shortening of the AP seen in this study) inhibits stretch-induced secretion of ANP by shortening APD (61). We hypothesizethat the shortened AP upon shear stress may be part of this systemof ANP regulation in conditions of atrial mechanical overload andmay constitute a mechanism by which the atria can repolarize.

1

1

1 1

22

2 2

FAK397 - DAPI

SHAM HF

Fres

hly

isol

ated

atr

ial m

yocy

tes

40

130SHAM HF

GAPDH

FAK

SHAM HF

integrin 1D - DAPI

GAPDH

Integrin 1

40

SHAM HF

130

B

6040

SHAM HF

GAPDH

Kv1.5

Cur

rent

Den

sity

(pA

/pF)

HFSHAM

*6

4

2

0

HFFo

ld in

crea

se u

pon

shea

r

**

SHAM

20

15

10

5

0

% c

ell r

espo

ndin

g to

she

ar

HFSHAM

11/14

13/28

0

20

40

60

80

100

HF

Time (s)

**

HF

Max

slo

pe (p

A/(p

F.s )

)

SHAM0.0

0.5

1.0

1.5

4 dyn.cm-2

SHAM

Cur

rent

Den

sity

(pA

/pF)

0 200 400 6000

20

40

60

A C

D E F G

Rata

tria

lon

sse

cti

Fig. 5. The response to shear stress is altered in a rat model of atrial hemodynamic overload. (A, Upper) Atrial sections from sham and HF rats stained withanti-integrinß1D (green), α-actinin (red), and DAPI (blue) showing the increased integrinß1D staining in HF tissue. (Lower) Representative examples ofintegrinß1 protein expression in the left atria showing ∼100% increase in HF rats after normalization to GAPDH (sham, n = 4; HF, n = 9; **P < 0.01). (B, Upper)Freshly isolated atrial myocytes from sham and HF rats stained with anti phospho-FAKY397 (green), α-actinin (red), and DAPI (blue), showing endogenousactivation of FAK. (Scale bar: 10 μm.) Enlargement of ROI showing FAK397 staining (1) within the cell and (2) at the intercalated disk. (Lower) Representativeexamples of FAK protein expression in the left atria showing ∼20% increase in HF rats after normalization to GAPDH (sham, n = 6; HF, n = 10; **P < 0.01). (C,Upper) Bar graph showing basal current density from control and hypertrophied myocytes showing the increase in current density before shear stress in HFrats (*P < 0.05). (Lower) Representative examples of Kv1.5 protein expression in the left atria showing ∼40% decrease in HF rats after normalization toGAPDH (sham, n = 4; HF, n = 9; *P < 0.05). (D) Quantification of the percentage of atrial myocytes isolated from sham or HF rats responding to shear stress(sham, 78%; n = 14 vs. HF, 53%; n = 28). (E) Example time series from myocytes from sham and HF rats exposed to shear stress showing the reduced kinetics ofthe response in hypertrophied myocytes. (F) Corresponding plots of the kinetics measured as the maximum slope of the response to shear stress (sham, 1.18 ±0.13 pA/(pF·s), n = 11; HF, 0.70 ± 0.09 pA/(pF·s), n = 13; **P < 0.01). (G) Plot of the response to shear stress from myocytes from sham (n = 10) and HF (n = 12)rats expressed in fold increase compared with preshear stress current densities. Note the reduced ability to respond to shear stress in hypertrophied myocytes(sham, ∼15-fold increase, n = 11 vs. HF, approximately 9-fold increase, n = 23; **P < 0.01).

Boycott et al. PNAS | Published online September 24, 2013 | E3961

PHYS

IOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

Conclusion. The lifetime of ion channels at the membrane hasemerged as a major determinant of current properties and cellexcitability. Most of the knowledge on the processes regulat-ing ion channel density, particularly K+ channels, concernstheir internalization (8, 62–64) whereas little is known aboutforward trafficking. Our discovery of the existence of reser-voirs of Kv1.5 channels that are recruitable in response to me-chanical stimulation of ECM/integrin complexes constitutes animportant breakthrough in the understanding of Kv1.5 channelregulation.Our results suggest that the density of functional Kv1.5

channels at the sarcolemma results from an equilibrium betweenpools of channels inserted in the membrane and pools of chan-nels recruitable from intracellular vesicles. In addition to con-stitutive exocytosis, these results indicate that Kv1.5 channels canalso undergo regulated exocytosis, which appears to depend onthe integrity of the extracellular matrix.Importantly, Kv1.5 expression is not confined to the heart but

is expressed in various cell types, including neurons, pulmonaryarteries, skeletal muscle, and endocrine cells (4). In addition, thisKv channel is involved in several pathological processes, in-cluding atrial fibrillation (65, 66), pulmonary hypertension (67),multiple sclerosis (68), and the proliferation of cancer cells (69).As such, the regulation of this channel through manipulation ofits availability is likely to be relevant in these diseases, particu-larly in conditions in which integrin activity or the mechanicalenvironment is altered.

Materials and MethodsAnimals. Adult male Wistar rats (Janvier) were treated in accordance withour institutional guidelines (Ministère de l’Agriculture, France, authorization75–1090), and treatment conformed to the Directive 2010/63/EU of theEuropean Parliament.

Cell Isolation, Cell Culture, and Gene Transfer. Adult cardiomyocytes isolatedby Langendorff perfusion were used for patch-clamp experiments. Briefly,hearts were cannulated and retrogradely perfused at 5 mL/min through theaorta, first with a solution containing (mM): NaCl 100, KCl 4, MgCl2 1.7,glucose 10, NaHCO3 5.5, KH2PO4 1, Hepes 22, 2,3-butanedione monoxime(BDM) 15, taurine 10, pH 7.4 (NaOH) at 37 °C for 5 min, then with enzymaticsolution containing (mM): NaCl 100, KCl 4, MgCl2 1.7, glucose 10, NaHCO3

5.5, KH2PO4 1, Hepes 22, BDM 15, and 200 UI/mL collagenase A (RocheDiagnostics) plus 0.5% BSA, pH 7.4 (NaOH) at 37 °C for ∼17 min. Solutionswere bubbled with 95% O2/5% CO2 throughout. Atria and ventricles wereseparated and placed into Kraft-Brühe (KB) buffer containing (mM): gluta-mic acid potassium salt monohydrate 70, KCl 25, taurine 20, KH2PO4 20,MgCl2 3, EGTA 0.5, glucose 10, Hepes 10, pH 7.4 (KOH). The tissue was thencut into small pieces and gently agitated to dissociate single myocytes.Myocytes were plated onto laminin-coated Petri dishes and maintained inKB buffer until use. Atrial myocytes for use in overexpression experiments,or for experiments requiring more than 24 h treatment with drugs, wereisolated as previously described (45). Cultured myocytes were transfectedwith the dominant negative (DN) construct of Rab11 (S25N) cloned intoa pEGFP-C1 (kind gift from D. Fedida, Department of Anaesthesiology,Pharmacology and Therapeutics, University of British Columbia, Vancouver)using a liposome-based approach and low CO2 condition (37 °C, 1%) (70) ortransduced with the Kv1.5 construct cloned in pEGFP-N3 and subcloned inadenovirus (10).

In Vitro Generation of Shear Stress. Shear stress was replicated in vitro usinga perfusion system placed 40 cm above the experimental chamber and at-tached to a flow regulator. The magnitude of the shear stress at the outletwas calculated using the equation:

τ ¼ 4μQπr3

,

where τ is the shear stress in dyn·cm−2, μ is the viscosity of the fluid(1.002·10−2 dyn·s/cm2 for water at 20 °C), Q is the volumetric flow rate(cm3·s−1), and r is the internal radius of the perfusion tip (cm). The shearstress generated by this system could be varied by changing the speed of theperfusion such that shear stresses between 0.5 and 10 dyn·cm−2 (corre-sponding to 0.05 N·m−2 to 0.1 N·m−2) could be generated. The tip of theperfusion (internal diameter of 0.58 mm) was placed ∼100 μm from the cell;therefore the calculated shear stress represents a maximum value experi-enced by the cell. To shear stress fields of cells, a VC-LFR-18-SS laminar flowchamber was used (C&L Instruments). The chamber was attached to the flowregulator and the shear stress was calculated using the equation:

τ ¼ 6μQbh2 ,

where τ, μ, and Q have the same meaning as above, b is the chamber width(cm), and h is the chamber height (cm). For the VC-LFR-18-SS chamber,a perfusion rate of 2 mL·min−1 generates a shear stress of 4 dyn·cm−2. SeeFig. S1 for schematic representations of the devices used to generate shearstress using the above methods.

Electrophysiology. Whole-cell patch-clamp recordings were obtained frommyocytes in voltage- or current-clamp mode using a patch-clamp amplifier(Axopatch 200B;Molecular Devices). Patch pipettes had resistances between 1and 2 MΩ for whole-cell and 10 and 15 MΩ for single-channel recordings.Currents were low-pass filtered at 10 kHz (−3 db) and digitized with NI PCI-6251 (National Instruments). Data were acquired and analyzed with Elphy2.0 software (G. Sadoc, CNRS, Gif-sur-Yvette, France). For AP recording, pi-pette solution contained (mM): K-aspartate 115, NaCl 5, EGTA 5, Hepes 10,Mg-ATP 3, MgCl2 2, Tris phosphocreatine-5 5, pH 7.2 (KOH). Perfusate con-tained (mM): NaCl 140, KCl 4, CaCl2 2, MgCl2 2, NaH2PO4 1, pyruvate 2.5,Hepes 10, glucose 20, pH 7.4 (NaOH). APs were evoked by 10 ms currentpulses at 0.2 Hz. For whole-cell current recording, pipette solution contained(mM): K-aspartate 115, KCl 10, KH2PO4 2, MgCl2 3, Hepes 10, glucose 10,CaCl2 0.1, and Mg-ATP 5, pH 7.2 (KOH). Standard perfusate contained (mM):NaCl 140, KCl 4, MgCl2 2, NaHPO4 1, glucose 20, Hepes 10, CaCl2 2 (EK = −86mV), pH 7.4 (NaOH). For time series analysis, currents were evoked by de-polarization from −80 mV to +60 mV for 750 ms at 0.2 Hz. IKur amplitudeswere measured at the end of the test pulse relative to the holding current.For the current-voltage protocol, voltage steps ranging from −100 mV to+60 mV were applied from a holding potential of −80 mV in 10 mV incre-ments. Leak current was numerically compensated. To modify the equilib-rium potential for K+, a high [K+] extracellular solution was used (mM): KCl

Sarcolemma

Kv1.5

Exocytosis

AP shortening

PFAK

Shear-stress

Microtubules Rab11

SNAREs

Integrin

Recycling endosome

ECM

Ca2+ Ca2+

Ca2+

Fig. 6. Schematic representation of Kv1.5 recruitment to the sarcolemma inresponse to shear stress. Increased shear stress is sensed by the extracellularmatrix (ECM) and integrins that convey the signal through phosphorylationof focal adhesion kinase (PFAK). This intracellular signaling pathway leads tothe stimulation of the Rab11-associated slow recycling endosome (RE)walking along microtubules. The fusion of the donor compartment (RE) withthe acceptor membrane (sarcolemma) causes the release of Kv1.5 and itsaccretion at the sarcolemma. As a consequence, an increase in currentdensity is observed together with a shortening of the action potential (AP).During hemodynamic overload, a condition in which both the ECM and themyocardium are remodeled, this mechanotransduction pathway is likelychronically stimulated, leading to an increase in the number of Kv1.5channels at the sarcolemma.

E3962 | www.pnas.org/cgi/doi/10.1073/pnas.1309896110 Boycott et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

90, MgCl2 2, NaCl 40, NaH2PO4 1, glucose 20, Hepes 10, (EK = −6 mV), pH 7.4(KOH). For single-channel recordings, pipette solution contained (mM):NMDG-aspartate 130, K-aspartate 5.4, MgCl2 1, Hepes 10, CaCl2 1, NaH2PO4

0.33, glucose 10, pH 7.4 (KOH). Perfusate contained (mM): K-aspartate 135,MgCl2 1, Hepes 10, CaCl2 1, NaH2PO4 0.33, glucose 10, pH 7.4 (NaOH).

The chord conductance (G) was calculated as follows: G = Gmax/(1 + exp[−(Vm – V0.5)/k]), where Gmax is the maximum chord conductance, G thechord conductance calculated at membrane potential Vm, V0.5 the potentialat which the conductance is half-maximally activated, and k the slope factordescribing the steepness of the activation curve. The Nernst potential for K+

ions in our experiments was EK = −86 mV. The activation-Vm curves andsteady-state availability-Vm curves were fitted with a single Boltzmannfunction: y = 1/(1+exp[−(Vm − V0.5)/k)]) where V0.5 is the half-activation po-tential, Vm the test voltage, and k the slope factor. Mathematical subtractionof the shift of the activation was performed using the following equation:IKur(Vm, t →∞) = Gmax(Vm)·m(Vm, t →∞)·(Vm-EK)/Cm, where Vm is the mem-brane potential, IKur(Vm,t →∞) is the resulting steady state current ampli-tude, Gmax maximum chord conductance of the 4-AP–sensitive component,m(Vm, t →∞) is the steady state activation parameter obtained in unshearedmyocytes, EK is the Nernst equilibrium potential for K+ ion, and Cm is themean cell capacitance. Single-channel recordings were performed in the cell-attached configuration, the patch membrane was clamped from −80 mV to+80 mV for 750 ms at 0.2 Hz.

Couette Flow Model. The shear stress (τ) was estimated in the rat atria usingthe following equation:

τ ¼ μUod,

where μ is the viscosity of the fluid, d the distance between the plates, andUo the velocity of the sliding plate. The plate velocity can be estimated fromthe shear strain rate and the dimensions of myocardial laminae. The followingmeasures were used to calculate the shear stress of the system: laminarthickness = 20 μm; gap height d = 6.5 μm; height of laminar unit a = 1/2 (20μm) + 6.5 μm + 1/2 (20 μm) = 26.5 μm; shear strain Esn = 1/2 tan (θ) = 1/2 b/a∼0.15 in canine LV (24); laminar shear motion b = 2 a Esn ∼2(26.5 μm)·(0.15) =7.95 μm; time t = 0.1 s for resting heart rate in adult rate of ∼5 Hz (300 bpm)with contraction taking half the cycle; viscosity of interstitial fluid μ = 3.5 cP =0.035 dyn·s·cm−2. Finally, the shear stress can be calculated as follows:

τ ¼�0:035 dyn · s · cm−2�:

�7:95 μm0:1 s

�

6:5 μm¼ 0:43 dyn · cm−2:

Immunofluorescence and 3D Microscopy. Immunofluorescence (IF) was per-formed on cardiomyocytes or on atrial cryosections (8-μm sections) fixed with4% paraformaldehyde (PFA) for 10 min at room temperature. Preparationswere incubated for 1 h at room temperature with permeabilizing/blockingbuffer (PBS containing 0.1% Triton X-100, 1% BSA, 10% goat serum, and10% chicken serum) and then incubated with primary antibodies diluted inblocking buffer (PBS containing 1% BSA, 3% goat serum, and 3% chickenserum) overnight at 4 °C. Detection was performed with a 1 h incubationwith secondary antibodies AlexaFluor488 or AlexaFluor594 (1:500; MolecularProbes), and DAPI (1:500; Sigma) to detect nuclei. Images were acquired witha cooled CoolSnap camera (Roper-Scientific) on an Olympus epifluorescentmicroscope (60×, UPlanSApo, 0.17). Images were processed and analyzedusing Metamorph software (Molecular Devices) supplemented with the 3D-deconvolution module. For each sample, a series of consecutive planes (stackof images) were acquired (sectioning step, 0.2 μm) and deconvoluted usingacquired point spread function. Fluorescence was quantified using ImageJsoftware (freeware; National Institutes of Health).

Western Blotting and Quantification. Left atria were transferred to lysis buffercontaining (mM) Tris-Hcl 50 (pH 7.5), NaCl 150, EDTA 2, 0.5% Triton X-100,and a protease mixture inhibitor (Sigma) and homogenized using a polytron.

The soluble fraction retrieved after 15 min centrifugation at 15,000 × g at 4 °Cwas used. Then 10 μg of protein was separated on 10% Bis-Tris gels (Invi-trogen) and transferred for 2 h at 80 V onto nitrocellulose membrane (Bio-Rad). Membranes were incubated with primary and secondary antibodiesand revealed using the Pierce ECL Plus kit. Anti-rabbit and anti-mouse IgG-linked HRP antibodies were used at 1:1,000 (Cell Signaling). All blots wereimaged using the Ettan DIGE imager (GE Healthcare), and ImageJ softwarewas used to quantify individual bands.

Antibodies. Primary antibodies used were mouse anti–α-tubulin (1:400; Sigma),rabbit anti-FAKY397 (1:50; Santa Cruz), mouse anti-sarcomeric α-actinin(1:2,000; Sigma), mouse anti-FAK (1:500; Millipore), rabbit anti-GFP (1:300;Torrey Pines Biolabs), and rabbit anti-GAPDH (1:2,000; Abcam). Integrinß1Dantibody was a kind gift from Robert Ross (Veterans AdministrationHealthcare, San Diego) (1:10,000 for Western Blot; 1:5,000 for IF).

Evanescent Field Microscopy and Image Analysis. Live imaging experimentswere performed at ∼27 °C with cultured myocytes seeded on glass-bottomedmicrodishes (170-μm thickness; Ibidi) bathed with standard extracellular so-lution. Kv1.5-EGFP channel dynamics were visualized with total internal re-flection fluorescence (TIRF) microscopy using the Olympus Celltirf system.Cells were placed on an Olympus IX81-ZDC2 inverted microscope, and TIRFillumination was achieved with the motorized integrated TIRF illuminatorcombiner (MITICO-2L) using a 60×/1.49 APON OTIRF objective. EGFP wasvisualized using a 50-mW solid-state 491-nm laser for excitation and dual-band optical filter (M2TIR488-561). All image acquisition, TIRF angle ad-justment, and some of the analysis were performed using the xcellencesoftware (Olympus). Time series of 100 images at 5-s intervals were recordedusing a digital CCD camera ORCA-ER (Hamamatsu). Cells were imaged for1 min before applying shear stress of 4 dyn·cm−2 to establish the baselinewhole-cell evanescent field fluorescence (EFF). To quantify changes in EFF,the cell perimeter was delineated, and the projection of minimum intensity(EFFmin) of the whole movie was subtracted from each image [EFF(t)] andnormalized to the background intensity measured in a region of interestoutside the cell (EFFbck) using the following equation:

ΔEFF ¼�EFFðtÞ − EFFmin

EFFmin − EFFbck

�·100:

Experimental Model of Atrial Dilatation. Myocardial infarction was induced in11maleWistar rats by ligating the left coronary artery as previously described(38, 71). Sham rats (n = 11) underwent the same surgical procedure withoutleft coronary artery ligation. The cardiopathy was characterized by trans-thoracic echocardiography in 3 mo after surgery. Rats were killed 4 mo aftersurgery, and single myocytes were obtained using the Langendorff method.Residual tissue remaining after isolation of single cells was collected, placed inlysis buffer, homogenized, and frozen at −80 °C for use in Western blotexperiments. Left atria were fixed with 4% PFA for 10 min at room tempera-ture and embedded in cryomatrix, and 8-μm cryosections were prepared for IF.

Statistics. Data were tested for normality using the D’Agostino and Pearsonnormality test. Statistical analysis was performed using Student t test orANOVA followed by post hoc Bonferroni test on raw data. Results are givenas means ± SEM and P values of less than 0.05 were considered significant(*P < 0.05, **P < 0.01, and ***P < 0.001).

ACKNOWLEDGMENTS. We thank Dr. Robert S. Ross for kindly providing theintegrinß1D antibody and Dr. Nathalie Mougenot and Mrs. Adeline Jacquetof Platform PECMV (Plateforme d’Expérimentation Coeur, Muscle, Vais-seaux), Pitié-Salpêtrière Hospital, France for their helpful assistance withthe animal model. This work was supported by Fondation Leducq “StructuralAlterations in the Myocardium and the Substrate for Cardiac Fibrillation”(to E.B., H.E.B., and S.N.H.) and European Union Grant EUTRAF-261057 (toE.B. and S.N.H.). This work was also supported by the French National Agencythrough the national program “Investissements d’avenir” with ReferenceANR-10-IAHU-05.

1. Uebele VN, et al. (1998) Distinct domains of the voltage-gated K+ channel Kv

beta 1.3 beta-subunit affect voltage-dependent gating. Am J Physiol 274(6 Pt 1):

C1485–C1495.2. Park KS, Yang JW, Seikel E, Trimmer JS (2008) Potassium channel phosphorylation in

excitable cells: Providing dynamic functional variability to a diverse family of ion

channels. Physiology (Bethesda) 23:49–57.3. Liu Y, Gutterman DD (2002) Oxidative stress and potassium channel function. Clin Exp

Pharmacol Physiol 29(4):305–311.

4. Levitan ES, Takimoto K (1998) Dynamic regulation of K+ channel gene expression in

differentiated cells. J Neurobiol 37(1):60–68.5. Schumacher SM, Martens JR (2010) Ion channel trafficking: A new therapeutic hori-

zon for atrial fibrillation. Heart Rhythm 7(9):1309–1315.6. Oancea E, Wolfe JT, Clapham DE (2006) Functional TRPM7 channels accumulate at the

plasma membrane in response to fluid flow. Circ Res 98(2):245–253.7. Choi WS, et al. (2005) Kv1.5 surface expression is modulated by retrograde trafficking

of newly endocytosed channels by the dynein motor. Circ Res 97(4):363–371.

Boycott et al. PNAS | Published online September 24, 2013 | E3963

PHYS

IOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

8. McEwen DP, et al. (2007) Rab-GTPase-dependent endocytic recycling of Kv1.5 in atrialmyocytes. J Biol Chem 282(40):29612–29620.

9. Schumacher SM, et al. (2009) Antiarrhythmic drug-induced internalization of theatrial-specific k+ channel kv1.5. Circ Res 104(12):1390–1398.

10. Balse E, et al. (2009) Cholesterol modulates the recruitment of Kv1.5 channels fromRab11-associated recycling endosome in native atrial myocytes. Proc Natl Acad SciUSA 106(34):14681–14686.

11. Zhao HC, et al. (2010) Membrane stretch and cytoplasmic Ca2+ independently mod-ulate stretch-activated BK channel activity. J Biomech 43(15):3015–3019.

12. Nam JH, et al. (2007) Mechanosensitive activation of K+ channel via phospholipaseC-induced depletion of phosphatidylinositol 4,5-bisphosphate in B lymphocytes.J Physiol 582(Pt 3):977–990.

13. Frank D, et al. (2008) Gene expression pattern in biomechanically stretched car-diomyocytes: Evidence for a stretch-specific gene program. Hypertension 51(2):309–318.

14. Nishimura S, et al. (2008) Responses of single-ventricular myocytes to dynamic axialstretching. Prog Biophys Mol Biol 97(2-3):282–297.

15. Zhang YH, et al. (2000) Stretch-activated and background non-selective cationchannels in rat atrial myocytes. J Physiol 523(Pt 3):607–619.

16. Liu X, et al. (2008) Stretch-activated potassium channels in hypotonically inducedblebs of atrial myocytes. J Membr Biol 226(1-3):17–25.

17. Dyachenko V, Husse B, Rueckschloss U, Isenberg G (2009) Mechanical deformationof ventricular myocytes modulates both TRPC6 and Kir2.3 channels. Cell Calcium45(1):38–54.

18. Lang RE, et al. (1985) Atrial natriuretic factor—a circulating hormone stimulatedby volume loading. Nature 314(6008):264–266.

19. Ruskoaho H, Thölken H, Lang RE (1986) Increase in atrial pressure releases atrialnatriuretic peptide from isolated perfused rat hearts. Pflugers Arch 407(2):170–174.

20. Zhang YH, Youm JB, Earm YE (2008) Stretch-activated non-selective cation channel: Acausal link between mechanical stretch and atrial natriuretic peptide secretion. ProgBiophys Mol Biol 98(1):1–9.

21. Ryu SY, Lee SH, Isenberg G, HoWK, Earm YE (2002) Monitoring of ANP secretion fromsingle atrial myocytes using densitometry. Pflugers Arch 444(4):568–577.

22. Browe DM, Baumgarten CM (2003) Stretch of beta 1 integrin activates an outwardlyrectifying chloride current via FAK and Src in rabbit ventricular myocytes. J GenPhysiol 122(6):689–702.

23. LeGrice IJ, Takayama Y, Covell JW (1995) Transverse shear along myocardial cleavageplanes provides a mechanism for normal systolic wall thickening. Circ Res 77(1):182–193.

24. Costa KD, Takayama Y, McCulloch AD, Covell JW (1999) Laminar fiber architectureand three-dimensional systolic mechanics in canine ventricular myocardium. AmJ Physiol 276(2 Pt 2):H595–H607.

25. Morad M, Javaheri A, Risius T, Belmonte S (2005) Multimodality of Ca2+ signalingin rat atrial myocytes. Ann N Y Acad Sci 1047:112–121.

26. Woo SH, Risius T, Morad M (2007) Modulation of local Ca2+ release sites by rapid fluidpuffing in rat atrial myocytes. Cell Calcium 41(4):397–403.

27. Lorenzen-Schmidt I, et al. (2006) Chronotropic response of cultured neonatal ratventricular myocytes to short-term fluid shear. Cell Biochem Biophys 46(2):113–122.

28. Kong CR, Bursac N, Tung L (2005) Mechanoelectrical excitation by fluid jets in mon-olayers of cultured cardiac myocytes. J Appl Physiol 98(6):2328–2336, discussion 2320.

29. Guo W, Kamiya K, Toyama J (1997) Immunocytochemistry of Kv1.5 potassium channelprotein in cultured neonatal rat ventricular myocytes. Heart Vessels 12(Suppl 12):165–167.

30. Wirth KJ, et al. (2003) Atrial effects of the novel K(+)-channel-blocker AVE0118in anesthetized pigs. Cardiovasc Res 60(2):298–306.

31. Grissmer S, et al. (1994) Pharmacological characterization of five cloned voltage-gated K+ channels, types Kv1.1, 1.2, 1.3, 1.5, and 3.1, stably expressed in mammaliancell lines. Mol Pharmacol 45(6):1227–1234.

32. He J, Kargacin ME, Kargacin GJ, Ward CA (2003) Tamoxifen inhibits Na+ and K+ currentsin rat ventricular myocytes. Am J Physiol Heart Circ Physiol 285(2):H661–H668.

33. Sorota S (1994) Pharmacologic properties of the swelling-induced chloride currentof dog atrial myocytes. J Cardiovasc Electrophysiol 5(12):1006–1016.

34. Rappoport JZ, Taha BW, Simon SM (2003) Movement of plasma-membrane-associatedclathrin spots along the microtubule cytoskeleton. Traffic 4(7):460–467.

35. Takahashi S, et al. (2012) Rab11 regulates exocytosis of recycling vesicles at the plasmamembrane. J Cell Sci 125(Pt 17):4049–4057.

36. Loewen ME, et al. (2009) Shared requirement for dynein function and intact micro-tubule cytoskeleton for normal surface expression of cardiac potassium channels. AmJ Physiol Heart Circ Physiol 296(1):H71–H83.

37. Milliez P, et al. (2005) Spironolactone reduces fibrosis of dilated atria during heartfailure in rats with myocardial infarction. Eur Heart J 26(20):2193–2199.

38. Rucker-Martin C, et al. (2006) Chronic hemodynamic overload of the atria is animportant factor for gap junction remodeling in human and rat hearts. CardiovascRes 72(1):69–79.

39. Leung DY, Boyd A, Ng AA, Chi C, Thomas L (2008) Echocardiographic evaluation ofleft atrial size and function: Current understanding, pathophysiologic correlates, andprognostic implications. Am Heart J 156(6):1056–1064.

40. Babbitt CJ, Shai SY, Harpf AE, Pham CG, Ross RS (2002) Modulation of integrins andintegrin signaling molecules in the pressure-loaded murine ventricle. Histochem CellBiol 118(6):431–439.

41. Brancaccio M, et al. (2006) Integrin signalling: The tug-of-war in heart hypertrophy.Cardiovasc Res 70(3):422–433.

42. Ross RS (2002) The extracellular connections: The role of integrins in myocardialremodeling. J Card Fail 8(6, Suppl):S326–S331.

43. Gundelfinger ED, Kessels MM, Qualmann B (2003) Temporal and spatial coordinationof exocytosis and endocytosis. Nat Rev Mol Cell Biol 4(2):127–139.

44. Balse E, et al. (2012) Dynamic of ion channel expression at the plasma membrane ofcardiomyocytes. Physiol Rev 92(3):1317–1358.

45. Abi-Char J, et al. (2007) Membrane cholesterol modulates Kv1.5 potassium channeldistribution and function in rat cardiomyocytes. J Physiol 582(Pt 3):1205–1217.

46. Tengholm A, Meyer T (2002) A PI3-kinase signaling code for insulin-triggeredinsertion of glucose transporters into the plasma membrane. Curr Biol 12(21):1871–1876.

47. Zadeh AD, et al. (2008) Internalized Kv1.5 traffics via Rab-dependent pathways.J Physiol 586(Pt 20):4793–4813.

48. Martens JR, Sakamoto N, Sullivan SA, Grobaski TD, Tamkun MM (2001) Isoform-specific localization of voltage-gated K+ channels to distinct lipid raft populations.Targeting of Kv1.5 to caveolae. J Biol Chem 276(11):8409–8414.

49. Yang Y, et al. (2010) Alpha5beta1 integrin engagement increases large conductance,Ca2+-activated K+ channel current and Ca2+ sensitivity through c-src-mediatedchannel phosphorylation. J Biol Chem 285(1):131–141.

50. Chao JT, Gui P, Zamponi GW, Davis GE, Davis MJ (2011) Spatial association of theCav1.2 calcium channel with α5β1-integrin. Am J Physiol Cell Physiol 300(3):C477–C489.

51. Kawasaki J, Davis GE, Davis MJ (2004) Regulation of Ca2+-dependent K+ current byalphavbeta3 integrin engagement in vascular endothelium. J Biol Chem 279(13):12959–12966.

52. Carlson SS, Valdez G, Sanes JR (2010) Presynaptic calcium channels and α3-integrinsare complexed with synaptic cleft laminins, cytoskeletal elements and active zonecomponents. J Neurochem 115(3):654–666.

53. Wang YG, Samarel AM, Lipsius SL (2000) Laminin acts via beta 1 integrin signalling toalter cholinergic regulation of L-type Ca(2+) current in cat atrial myocytes. J Physiol526(Pt 1):57–68.

54. Zhang Y, Sekar RB, McCulloch AD, Tung L (2008) Cell cultures as models of cardiacmechanoelectric feedback. Prog Biophys Mol Biol 97(2-3):367–382.

55. Dvir T, Levy O, Shachar M, Granot Y, Cohen S (2007) Activation of the ERK1/2 cascadevia pulsatile interstitial fluid flow promotes cardiac tissue assembly. Tissue Eng 13(9):2185–2193.

56. Huang Y, Jia X, Bai K, Gong X, Fan Y (2010) Effect of fluid shear stress on car-diomyogenic differentiation of rat bone marrow mesenchymal stem cells. Arch MedRes 41(7):497–505.

57. Huang Y, et al. (2012) Effect of cyclic strain on cardiomyogenic differentiation of ratbone marrow derived mesenchymal stem cells. PLoS ONE 7(4):e34960.

58. Cho KW, et al. (2002) Mechanical control of extracellular space in rabbit atria: Anintimate modulator of the translocation of extracellular fluid and released atrialnatriuretic peptide. Exp Physiol 87(2):185–194.

59. Suzuki E, et al. (1992) Cellular mechanisms for synthesis and secretion of atrialnatriuretic peptide and brain natriuretic peptide in cultured rat atrial cells. Circ Res71(5):1039–1048.

60. Ronkainen JJ, Vuolteenaho O, Tavi P (2007) Calcium-calmodulin kinase II is thecommon factor in calcium-dependent cardiac expression and secretion of A- and B-type natriuretic peptides. Endocrinology 148(6):2815–2820.

61. Saegusa N, et al. (2005) Kir6.2-deficient mice are susceptible to stimulated ANP se-cretion: K(ATP) channel acts as a negative feedback mechanism? Cardiovasc Res 67(1):60–68.

62. Guo J, et al. (2009) Extracellular K+ concentration controls cell surface density of IKrin rabbit hearts and of the HERG channel in human cell lines. J Clin Invest 119(9):2745–2757.

63. Svoboda LK, et al. (2012) Redox-sensitive sulfenic acid modification regulates surfaceexpression of the cardiovascular voltage-gated potassium channel Kv1.5. Circ Res111(7):842–853.

64. Pongs O (2009) Ins and outs of cardiac voltage-gated potassium channels. Curr OpinPharmacol 9(3):311–315.

65. Tamargo J, Caballero R, Gómez R, Delpón E (2009) I(Kur)/Kv1.5 channel blockers forthe treatment of atrial fibrillation. Expert Opin Investig Drugs 18(4):399–416.

66. Brendel J, Peukert S (2003) Blockers of the Kv1.5 channel for the treatment of atrialarrhythmias. Curr Med Chem Cardiovasc Hematol Agents 1(3):273–287.

67. Cogolludo A, et al. (2006) Serotonin inhibits voltage-gated K+ currents in pulmonaryartery smooth muscle cells: Role of 5-HT2A receptors, caveolin-1, and KV1.5 channelinternalization. Circ Res 98(7):931–938.

68. Mullen KM, et al. (2006) Potassium channels Kv1.3 and Kv1.5 are expressed on blood-derived dendritic cells in the central nervous system. Ann Neurol 60(1):118–127.

69. Bielanska J, et al. (2009) Voltage-dependent potassium channels Kv1.3 and Kv1.5in human cancer. Curr Cancer Drug Targets 9(8):904–914.

70. El-Haou S, et al. (2009) Kv4 potassium channels form a tripartite complex with theanchoring protein SAP97 and CaMKII in cardiac myocytes. Circ Res 104(6):758–769.

71. Michel JB, et al. (1988) Hormonal and cardiac effects of converting enzyme inhibitionin rat myocardial infarction. Circ Res 62(4):641–650.

E3964 | www.pnas.org/cgi/doi/10.1073/pnas.1309896110 Boycott et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0