Physiological and ultrastructural features of humaninduced pluripotent and embryonic stem cell-derivedskeletal myocytes in vitroGunnar Skoglunda, Jeanne Lainéa, Radbod Darabib,1, Emmanuel Fourniera, Rita Perlingeirob, and Nacira Tabtia,c,2

aDépartement de Physiologie, Université Pierre et Marie Curie, Site Pitié-Salpêtrière, 75013 Paris, France; bLillehei Heart Institute, Department of Medicine,University of Minnesota, Minneapolis, MN 55455; and cUniversité Paris Est Créteil, 94010 Créteil, France

Edited* by William A. Catterall, University of Washington School of Medicine, Seattle, WA, and approved April 29, 2014 (received for review December3, 2013)

Progress has recently been made toward the production of humanskeletal muscle cells from induced pluripotent stem (iPS) cells.However, the functional and ultrastructural characterization, whichis crucial for disease modeling and drug discovery, remains to bedocumented. We show, for the first time to our knowledge, thatthe electrophysiological properties of human iPS-derived skeletalmyocytes are strictly similar to those of their embryonic stem (ES)cell counterparts, and both are typical of aneural mammalianskeletal muscle. In both cell types, intracellular calcium signalingthat links membrane depolarization to contraction occurs in theabsence of extracellular Ca2+, a unique feature of skeletal mus-cle. Detailed analysis of the Ca2+ signal revealed diverse kineticsof the rising phase, and hence various rates in the release of Ca2+

from the sarcoplasmic reticulum. This was mirrored by ultrastruc-tural evidence of Ca2+ release units, which varied in location,shape, and size. Thus, the excitation–contraction coupling ma-chinery of both iPS- and ES-derived skeletal myocytes was func-tional and specific, but did not reach full maturity in culture. Thisis in contrast with the myofibrillar network, which displayed thesame organization as in adult skeletal muscle. Overall, the pres-ent study validates the human iPS-based skeletal myocyte modelin comparison with the embryonic system, and provides the func-tional and ultrastructural basis for its application to human skel-etal muscle diseases.

human iPS-myocyte | human ES-myocyte | electrophysiology |EC coupling | ultrastructure

The generation of induced pluripotent stem (iPS) cells bygenetic reprogramming of adult human somatic cells has

opened great opportunities for basic research and regenerativemedicine. Modeling human diseases with iPS cell technologyoffers a direct, noninvasive, and renewable experimental systemfor reproducing and studying pathological conditions. Within5–6 y after the first report on human iPS cells (1), an amazingnumber of diseases affecting various systems (neurological, met-abolic, cardiovascular, hematopoietic) have been modeled byreprogramming patient somatic cells (especially skin fibroblasts)into iPS cells followed by specific differentiation into cell typesaffected by the disease (2). Nevertheless, the efficiency of gen-erating a robust population of cell progenitors from humaniPS cells with a high differentiation potential in culture varieswidely between different tissues. In particular, producing skel-etal muscle cells from iPS cells turned out to be challenging, asreflected by the very limited number of reports and laboratorieswith successful results. This is in contrast to cardiac muscle,which has widely benefited from iPS cell technology, as varioushereditary heart diseases have been modeled and explored (3, 4).We have previously shown that inducible expression of pairedbox (PAX) 3 or 7 transcription factors in both murine embryonicstem (ES) and iPS cells promotes the production of myogenicprogenitors (5, 6). The differentiation potential of these pro-genitors has been demonstrated in vitro by the occurrence of

multinucleated myocytes expressing myogenin and myosin heavychain (MHC) and in vivo by the presence of functional donor-de-rived myofibers (5, 6). Recently, this approach has been appliedsuccessfully to human ES/iPS cells (7). Nonetheless, other im-portant aspects, such as those related to the physiological andultrastructural differentiation, need to be explored to assessthe applicability of the present model to pathophysiologicalresearch. The need for such characterization also applies toES-derived skeletal muscle cells, which we can refer to as thenatural counterpart to the iPS model.The present study elucidates the electrophysiological properties

and the excitation–contraction (EC) coupling of both human iPS-and ES-derived skeletal myocytes (ES- and iPS-myocytes) anddepicts the underlying ultrastructural organization. It also deter-mines the degree of maturity that stem-cell–derived myogenic pro-genitors can attain following several weeks of differentiation in vitro.

ResultsA preliminary survey led us to the conclusion that iPS- and ES-derived myogenic progenitors should be allowed at least 2 wkin the differentiation medium to reach a consistent degree offunctional and ultrastructural differentiation. Therefore, we fo-cused on cells that have differentiated for 2–8 wk, as the aimof the present study was not to describe the sequence of develop-mental events, but to assess the ability of myogenic progenitors

Significance

Patients with inherited skeletal muscle diseases are offeredlittle else than symptomatic relief, even when responsiblegenes have been identified. Access to human skeletal musclethrough induced pluripotent stem (iPS) cells has opened newprospects for basic research and drug discovery. Yet the exactnature of iPS-derived skeletal muscle cells needs to be exploredto assess their relevance to pathophysiological studies. Weshow, for the first time to our knowledge, that human iPS-derived skeletal myocytes acquire functional and ultrastruc-tural landmarks of skeletal muscle and reach, in vitro, the samedegree of maturity as their embryonic counterparts. This de-tailed characterization, besides validating the model, paves theway for exploring the pathophysiology and pharmacology ofskeletal muscle diseases using patient-derived iPS cells.

Author contributions: G.S., J.L., and N.T. designed research; G.S., J.L., R.D., and N.T. per-formed research; G.S., J.L., R.D., E.F., R.P., and N.T. analyzed data; and G.S., J.L., E.F.,R.P., and N.T. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1Present address: Center for Stem Cell and Regenerative Medicine, University of Texas,Houston, TX 77030.

2To whom correspondence should be addressed. E-mail: nacira.tabti@upmc.fr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1322258111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1322258111 PNAS | June 3, 2014 | vol. 111 | no. 22 | 8275–8280

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Oct

ober

10,

202

0

to acquire skeletal muscle-specific features in vitro. Fig. S1 showsthe occurrence of MHC, a protein typical of skeletal muscle (7),at 7 and 14 d postdifferentiation in ES- and iPS-myocytes,however sarcomeric organization takes place at later stages,becoming manifest at 30 d, as illustrated by α-actinin immu-nostaining of the Z-lines (Fig. 1).

Electrophysiology. The resting membrane potential (RMP) recor-ded from proliferating cells was relatively low in both ES- andiPS-derived myogenic progenitors, with mean values (±SEM)of –39.41 ± 4.31 mV (n = 12) and –36.52 ± 1.25 mV (n = 12),respectively. Differentiated cells (myocytes) displayed higherRMP than proliferating progenitors; similar RMP mean valueswere found in ES- (–72.25 ± 4.57 mV) and iPS-myocytes(–72.12 ± 2.55 mV). To assess the ability of the cells to gen-erate action potentials (APs), a series of 15 ms depolarizingcurrent pulses of increasing intensity was delivered in 50 pAincrements. Before stimulation, the cell membrane was hyper-polarized to a standard level of –100 ± 2 mV by injecting a steadyinward current, which eliminated variation in the initial RMPbetween cells and enabled assessment of membrane excitabilityunder similar conditions of maximum Na+ channel availability. Inboth ES- and iPS-derived myogenic progenitors, electrical stim-ulations of increasing intensity failed to trigger APs and gave riseto passive voltage changes (Fig. 2 A1 and A2). Proliferatingprogenitors are therefore unexcitable as confirmed by the lackof detectable voltage-dependent Na+ currents (Fig. 2 C1 andC2). In contrast, differentiated muscle cells from both stem celltypes were electrically excitable and displayed fast rising APsthat clearly overshot the zero level (Fig. 2 B1 and B2). APsgenerated by ES- and iPS-myocytes had a similar magnitudeand kinetics. In both ES- (n = 10) and iPS-myocytes (n = 10),the AP was initiated from a threshold potential of –42.5 ± 1.4and –42.0 ± 1.7 and reached 45.4 ± 1.5 and 44.8 ± 1.6 mV, witha maximum rising slope of 103.4 ± 8.8 and 105.9 ± 10.4 V/s,respectively. The fast upstroke was followed by a slower decaytoward the resting level at a maximum rate of –32 ±0.2.5 and–35.9 ± 4.3 V/s; the AP duration relative to the threshold was2.9 ± 0.2 and 2.6 ± 0.2 ms, respectively.Voltage-gated ion channels underlying membrane excitability

were investigated by recording whole-cell currents under thesame near-physiological conditions (and in most cases from thesame cells) as those used to record APs. In proliferating myo-genic progenitors from both stem cell types, ion currents werehardly detectable following membrane depolarization up to 60mV (Fig. 2 C1 and C2). In a few cases, early, transient inwardcurrents, presumably carried by Na+ ions, could be observedbut did not exceed a few tens of pA. Outward currents werehardly detectable within this voltage range, however stronger de-polarization up to 100 mV led to noticeable outward currentsthat activated at ∼50 mV and increased steeply with furtherdepolarization. As can be seen from Fig. 2 C1 and C2, thesecurrents led to noisy recordings characterized by fluctuationsthat became larger as the voltage was stepped toward positivevalues. They were carried by K+, as they were suppressed by mMconcentrations of TEA and had a reversal potential of ∼–90 mV

as measured from the tail currents, a value close to the Nernstequilibrium potential for K+ ions (∼−97 mV under the presentconditions). In contrast to their progenitors, both ES- and iPS-myocytes produced large inward currents in response to a seriesof depolarizing voltage steps from −70 to 60 mV (VH, −100 mV)(Fig. 2 D1 and D2). Total currents, which represent the sum ofcurrents of different ion species, consisted of an early inwardNa+ current followed by a sustained outward delayed rectifier K+

current, as confirmed by their respective blockade with TTX andTEA (Fig. 2 F and G). Accurate study on the voltage dependence

Fig. 1. α-actinin immonostaining of ES/iPS-derived skeletal myocytes. ES- (A)and iPS- (B) myocytes labeled with anti–α-actinin antibody at 30 d post-differentiation. (Scale bar, 20 μm.)

Fig. 2. Membrane excitability and voltage-gated ion currents in humanES/iPS-derived myogenic progenitors and myocytes. (A1 and A2) Represen-tative tracings of passive membrane depolarization evoked by a series ofstep currents (100 pA, 30 ms) in ES- (A1) and iPS-derived myogenic progen-itors (A2). (B1 and B2) Superimposed tracings showing initiation of the AP inES- (B1) and iPS-derived skeletal myocytes (B2). Prior hyperpolarization, −100mV; step stimulus, 100 or 50 pA (last step); 15 ms (B1) or 20 ms (B2); dashedlines delimit AP overshoot. (C and D) Representative whole-cell currentsfrom myogenic progenitors (C1 and C2) and from skeletal myocytes (D1 andD2) of ES or iPS origin as indicated. Currents were elicited by 20 mV voltagesteps from −80 to 100 mV (C1 and C2) or to 70 mV (D1 and D2). VH, −100 mV.[Scale bar, 1 nA, 5 ms (C1 and C2); 20 pA/pF, 5 ms (D1 and D2).] (E1–E3)Current/voltage relationships (E1 and E2) and maximum Na+ and K+ currentdensities (E3) computed at VH −100 or −50 mV as indicated. (F1–F3) Effect ofTTX on Na+ currents in ES- (F1) and iPS-myocytes (F2), and related dose–response curves (F3). (G1 and G2) Effect of TEA on K+ currents in ES- (G1) andiPS-myocytes (G2). (H1–H3) DHP-sensitive L-type Ca2+ currents recorded us-ing the prepulse protocol shown under the tracings. Representative tracingselicited by a test pulse to 30 mV in ES- (H1) and iPS-myocytes (H2) before (C,control) and after the addition of nifedipine. The small residual current in H2is plausibly carried by the T-type Ca2+ current. (Scale bar, 2 pA/pF, 50 ms.)Symbols and graphs represent mean values ± SEM; capacity currents havebeen blanked.

8276 | www.pnas.org/cgi/doi/10.1073/pnas.1322258111 Skoglund et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

10,

202

0

of Na+ and K+ currents usually requires their separation usingspecific solutions and ion channel blockers. However, individualcurrent/voltage curves could still be drawn from our recordingsand yielded useful information on the ion currents underlyingmembrane excitability (Fig. 2 E1–E3). In both cell types, Na+

currents were activated from −50 mV and reached maximumintensity at ∼−30 mV. Delayed rectifier K+ currents displayeda more positive activation threshold of ∼−30 mV and a slightrectification from 40 mV (Fig. 2E1). Changing the holdingpotential from −100 to −50 mV caused full inactivation of theNa+ component, leaving only outward K+ currents (Fig. 2E2),with similar voltage dependence to those recorded at a VH of−100 mV. Their intensity, however, decreased by ∼40%, sug-gesting thereby a degree of inactivation for the voltage-gatedK+ channels expressed in these cells (Fig. 2E3). Analysis ofNa+ and K+ current density (Fig. 2E3) did not reveal any sig-nificant difference between the two cell types. Indeed, maximumcurrent density was 78.4 ± 6.8 and 80.8 ± 9.8 pA/pF for Na+

and 29.77 ± 2.39 and 27.46 ± 2.59 for K+ currents (at 60 mV),in ES- (n = 13) and iPS-myocytes (n = 13), respectively (VH,−100 mV). Recordings obtained at a VH of −50 mV in ES- (n =6) and iPS-myocytes (n = 8) showed negligible maximal Na+

current density (−0.51 ± 0.09 and −1.23 ± 0.28, respectively)and a clear decrease in K+ current densities (17.22 ± 2.54 and16.70 ± 2.99 pA/pF, respectively) compared with those obtainedat a VH of −100 mV. Again, there was no difference between ES-and iPS-myocytes.Pharmacological dissection of the total current was achieved

using TTX to block Na+ channels and TEA to block K+ chan-nels. It is well established that vertebrate skeletal muscles expressboth TTX-sensitive and TTX-insensitive Na+ channels duringtheir development and retain essentially TTX-sensitive channelsafter full maturation and innervation in vivo (8). Similarly, myo-cytes from primary cultures express both types of Na+ channels(9). We used different doses (from 0.1 to 10 μM) of TTX todistinguish between these two Na+ channel types based on K0.5values previously reported for developing myocytes in culture(9). In both ES- and iPS-myocytes, low doses of TTX (0.1 μM)suppressed ∼25% of the Na+ current, demonstrating thereby thepresence of TTX-sensitive Na+ channels (Fig. 2 F1–F3). Higherconcentrations of TTX further blocked the Na+ current ina dose-dependent manner, with a break in the slope of the dose–response curve suggesting the presence of two types of Na+

channels with a different TTX affinity (Fig. 2F3). In both celltypes, 10 μM of TTX eliminated ∼80% of the current. Theoutward current was substantially altered by the addition of TEAin both ES- and iPS-myocytes. On average, up to 75% of the K+

current was blocked by 10 mM TEA (Fig. 2 G1 and G2), and theremaining outward current disappeared after the addition of100 μM 3,4-Diaminopyridine (DAP). Taken together, these datasuggest the presence of TEA- and DAP-sensitive K+ channels,which is in line with the existence of different voltage-dependentK+ channels in mammalian skeletal muscle (10).L-type Ca2+ currents were recorded using a 1 s depolarizing

prepulse (11) to inactivate Na+ and T-type Ca2+ currents (Fig. 2H1–H3). In both ES- and iPS-myocytes, currents evoked duringvarious test potentials (Fig. 2H3) had the characteristics of skeletalmuscle Cav1.1 channels (12)—that is, slow activation kinetics(reached a peak value in >100 ms), high activation threshold(∼−10 mV), and complete inhibition by 50 μM of the dihy-dropyridine receptor (DHPR) blocker nifedipine (Fig. 2 H1 andH2). The current peaked at ∼30 mV, and maximum densitiesobtained from ES- (1.93 ± 0.25 pA/pF, n = 7) and iPS-myocytes(1.86 ± 0.33 pA/pF, n = 9) were comparable and of the sameorder of magnitude as those reported for human myotubes frommuscle biopsies (13).

Intracellular Calcium Signaling. Calcium signaling that mediatesEC coupling was evoked by depolarizing the cells with high

extracellular K+ ([K+]e) and in the absence of extracellular Ca2+

to discriminate between cardiac and skeletal muscle. [K+]e wasincreased from 5 to 50 mM, which should shift the membranepotential by ∼60 mV from its resting level and activate DHPRCa2+ channels. ES- and iPS-derived muscle progenitors didnot generate any intracellular Ca2+ signal upon membranedepolarization, but once differentiated, both cell types displayeda transient change in [Ca]i characterized by a sharp increasefollowed by a rapid, then a slow decay toward the baseline. Fig. 3A1 and A2 shows typical changes in the fluorescence ratioobtained from individual cells and expressed as absolute R340/380values. This demonstrates that the intracellular calcium signalinduced by membrane depolarization does not require extra-cellular Ca2+. Importantly, the Ca2+ transient was accompaniedby cell contraction, which confirms its physiological role in ECcoupling. Note that cell contraction, which transiently reducedthe fluorescent area under assessment, did not have any signifi-cant effect on the peak R340/380. Baseline and peak fluorescence-induced high K+ depolarization were comparable between ES-and iPS-myocytes (Fig. 3 A1 and A2). This yielded a similarincrease in fluorescence between ES- and iPS-myocytes, withmean values of a 1.44 ± 0.01- and 1.48 ± 0.03-fold increasefrom baseline, respectively. The upstroke of the calcium signalwas best fitted to a sigmoid function of the Boltzmann type,and described by the slope factor (dt). There was no statisti-cally significant difference (P > 0.5) in the mean values (±SEM)of the slope factor between ES- (2.58 ± 0.19, n = 80) and iPS-myocytes (2.65 ± 0.16, n = 95). The dt values were widely spread,ranging from 0.25 to 8.25 s in ES- and from 0.75 to 7.75 s in iPS-myocytes. Histograms of upstroke dt values revealed two peaksfor both cell types: a common primary peak at 1.8 s shared bythe majority of cells and a secondary peak at 6.0 s and 4.8 s forES- and iPS-myocytes, respectively (Fig. 3 B1 and B2). Note that

Fig. 3. Intracellular Ca2+ signaling in human ES/iPS-derived skeletal myo-cytes. (A1 and A2) Dynamics of intracellular Ca2+ following K+-induced de-polarization in Ca2+-free extracellular medium expressed as change withtime in fura-2 fluorescence ratio [R(340/380)]. (B1 and B2) Kinetics of theupstroke phase showing the distribution of the slope factor (dt) in ES- (B1)and iPS-myocytes (B2). (C1 and C2) Kinetics of the decay phase computed fromaveraged data at seven different time points in ES- (C1) and iPS-myocytes(C2); dashed lines depict the two-exponential fit of the decay. (D1–D4) Dis-tribution of τfast and τslow for ES- (D1 and D2) and iPS-derived myocytes(D3 and D4).

Skoglund et al. PNAS | June 3, 2014 | vol. 111 | no. 22 | 8277

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Oct

ober

10,

202

0

dt values were not correlated with the differentiation time inculture (beyond the minimum period of 2 wk applied herein).The decay phase of the Ca2+ signal was analyzed by normalizingthe fluorescence R340/380 values to the peak and averaging thedata at seven subsequent time points following the peak (Fig. 3C1 and C2). The averaged decay curves could be nicely fitted toa biexponential function characterized by a fast (τf) and a slow(τs) time constant. The fast decay component was identical inES- and iPS-myocytes, with τf mean values of 7.41 (n = 80) and7.47 s (n = 95), respectively. In contrast, the slow decay com-ponent was faster in ES- than in iPS-myocytes with a clear dif-ference in τs (60.4 s and 97.0 s, respectively). We also analyzedthe distribution of individual τf and τs values for more accuracy.As shown in Fig. 3 D1 and D3, the values for τf ranged from 1.0to 15.0 s in ES- and from 1.0 to 19.0 s in iPS-myocytes. Thevalues of τs displayed greater variability than τf between in-dividual cells within the same experiment (Fig. 3 D2 and D4) anddid not fit to a normal distribution. Corresponding histogramsclearly depict two populations among ES-myocytes: the first withrelatively fast τ values around 25 s, which we will refer to as theslow time constant per se (τs), and the second with values over100 s, which we named ultraslow (τus). This kind of bimodaldistribution was not manifest in iPS-myocytes, which mostlydisplayed ultraslow values over 100 s. Indeed, the fraction of cellswith τus was significantly higher in iPS- (56.8%) than in ES-myocytes(27.5%). Altogether, our results demonstrate that both ES- andiPS-myocytes respond to K+ depolarization by a similar rise in[Ca2+]i and show differences in their ability to bring back cyto-solic Ca2+ to baseline levels.

Ultrastructural Features. Proliferating myogenic progenitors ofboth ES and iPS origin had a round or polygonal shape with

a prominent nucleus and unspecific subcellular organization (Fig. S2).No muscle features such as myofilaments were detected at thisstage. Upon their differentiation, both ES- and iPS-derivedmyogenic cells gradually developed a muscle phenotype, witha typical transition from randomly organized myofilament net-work to clearly striated myofibril pattern (Fig. 4). The degree ofmyofibrillar organization varied between cells within the sameculture dish and between different dishes of the same age, butbecame more homogenous beyond 2 wk of differentiation. Fromthe third week of induced differentiation, a myofibril cross-stri-ation pattern was noticeable in both ES- and iPS-myocytes. TheA–I alternating pattern could be observed before parallelalignment of myosin and actin filaments was fully achieved.This is illustrated in Fig. 4 A and H, which shows longitudinalsections of ES- and iPS-myocytes, with somewhat wavy Z-linesand A-bands containing fascicles of thick filaments of variousorientations. After 1 mo of differentiation, we commonly en-countered ES- and iPS-myocytes with the typical sarcomericarrangement alternating A- (overlapping thick and thin fila-ments) and I-bands (thin filaments). Electron-dense Z-linesclearly delimited the sarcomeres, and M-lines were identifiedacross the A-bands (Fig. 4 E and I). Myofibril organization wasgenerally accompanied and sometimes preceded by the occur-rence of specific junctions between T-tubule–like structures (T)with their electrolucent lumens and sarcoplasmic reticulum(SR) sacs containing electron-dense granular material. These so-called internal Ca2+ release units (CRUs) (14) displayed variousconfigurations, orientations, and localizations. They could ap-pear as dyads (one T-tubule and one SR cisterna), triads (oneT-tubule between two SR cisternae), or “inverted” triads (two T-tubules separated by one SR cisterna). Multiple junctions made

Fig. 4. Ultrastructural features of human ES- (A–G) and iPS-myocytes (H–K). (A) Incomplete sarcomeric organization at 16 d postdifferentiation. (B–D)T-tubules–SR junctions (CRUs) with visible feet-like structure from the same fiber. (E) Highly organized myofibrillar pattern at 20 d postdifferentiation witha CRU at the A–I border (arrow). (F) Multiple SR–T-tubule junctions. (G) Coexistence of peripheral and internal CRUs. [Scale bar, 1 μm (A and E), 100 nm (G,applies to B–F).] (H) Incomplete sarcomeric organization at 15 d postdifferentiation. (H1–H3) SR–T-tubules junctions are enlarged on the right. (I) Fully maturesarcomeric banding pattern at 30 d postdifferentiation. (J) Another well-differentiated myocyte with several CRUs. (J1–J3) Selected CRUs enlarged un-derneath. (J1 and J2) Internal CRUs near the periphery of the cell with visible RyR feet. (J3) Mixed CRU with a flat SR saccule facing the plasmalemma on oneside and a T-tubule–like one on the other; both sides are lined with RyR feet. (K) A distinct example of multiple internal CRUs with noticeable RyR feet. PM,plasma membrane. [Scale bar, 1μm (H and I), 2 μm (J), 100 nm (H3, applies to H1 and H2), 100 nm (K, same scale in J1–J3).]

8278 | www.pnas.org/cgi/doi/10.1073/pnas.1322258111 Skoglund et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

10,

202

0

of several T–SR junctions in a row were also often observed (Fig.4 F and J2). At the interface with the T-tubules, SR membraneswere typically lined by rows of regularly spaced electron-densestructures corresponding most likely to the “feet” of ryanodinereceptors (RyRs), as clearly illustrated in Fig. 4 D, G, and K.Regardless of their configuration, these internal couplings weregenerally localized in a cytoplasmic area at some distance fromthe myofibrils, and only rarely occurred at the A–I junction.Besides these internal junctions, both cell types displayed the so-called peripheral couplings—that is, SR cisternae with RyR feetclosely apposed to the plasma membrane (Fig. 4 G and J3). Bothinternal and peripheral couplings could coexist within the samemyocyte. Interestingly, regardless of the type of CRUs, the gapbetween T-tubule and SR membranes had a constant thicknessof ∼15 nm. Triads with typical localization at the A–I junctionand transversal orientation were not found in the present elec-tron microscopy survey, suggesting that these highly organizedstructures characteristic of mature skeletal muscle fibers are ei-ther very rare or absent in the present model.

DiscussionAlthough human iPS-derived skeletal myocytes have been suc-cessfully produced in vitro, their functional characteristics anddegree of maturity remain unknown. We show, for the first timeto our knowledge, that these cells acquire the ion channel ma-chinery to generate APs and EC coupling typical of skeletalmuscle. At the ultrastructural level, they are endowed with CRUs,which remain diverse in shape and distribution in an otherwisehighly organized myofibrillar network. All these characteristics arecomparable to those found in ES-myocytes.Myogenic progenitors from both stem cell types have a low

RMP and lack detectable voltage-gated Na+ currents, the keymarkers of muscle excitability. In this respect, they are clearlydifferent from human satellite cells from adult muscles, whichexpress noticeable voltage-gated Na+ currents (15). This sug-gests that satellite cells undergo some degree of electrophysi-ological differentiation toward the muscle phenotype, whereasES- and iPS-derived myogenic precursors remain electricallyundifferentiated. Voltage-gated ion currents present in stem-cell–derived myogenic progenitors are predominantly K+ cur-rents resembling MaxiK (BK) currents reported in humanmesenchymal stem cells (16). MaxiK channels, which respondto both voltage and [Ca2+]i, are thought to regulate calcium-dependent signaling pathways and may act as effectors fortrophic factors in the surrounding cell milieu (17). Neverthe-less, definite identification of these presumed MaxiK channelswill require further investigation.In both cell types, the RMP became more negative with cell

differentiation, reaching for both cell types comparable levels tothose reported for human myotubes from muscle biopsies (18).Such change is probably related to the expression of K+ leakchannels known to set the RMP (19).APs elicited by electrical stimulation had similar characteristics

in ES- and iPS-myocytes and displayed a sharp rising phase fol-lowed by a slower decay, but no plateau phase, which distinguishthem from those generated by cardiomyocytes in culture (20). Inboth cell types, AP threshold and overshoot were comparable tothose reported for cultured human primary myotubes from musclebiopsies (18, 21), and for mammalian skeletal muscle, in general(22). AP kinetics vary with experimental conditions (temperature,protocols, etc.) and are therefore difficult to compare with pub-lished data. However, the total duration of AP recorded from ES-and iPS-myocytes was relatively long, mainly because of the decayrate. In contrast, the rate of the rise was in the same order as thatfound in primary cultured myocytes when bath temperature istaken into account (18). This suggests that voltage-gated Na+

channels, which are responsible for the depolarizing phase of theAP, reach a higher level of efficiency than K+ channels known fortheir critical role in membrane repolarization. This is corroborated

by the higher density of voltage-gated Na+ compared with K+

currents in both ES- and iPS-myocytes. Although this seems to bea common characteristic in mammalian skeletal muscles and inother excitable tissues (23), a relatively small K+ conductancemay favor spontaneous firing and hence contractions in thesedifferentiating cells. The coexistence of TTX-sensitive andTTX-resistant Na+ currents is in line with established knowl-edge on the expression of both Nav1.4 and Nav1.5 (adultskeletal muscle and cardiac Na+ channel isoforms, respectively)in noninnervated skeletal muscle (8, 9). Although the contri-bution of each channel isoform cannot be precisely inferredfrom the present study, our results indicate a predominantexpression of the cardiac isoform in both cell types. This bringsto attention the limitation of using noninnervated iPS-derivedskeletal myocytes as an experimental model for studying thepathophysiological mechanisms underlying human diseases as-sociated with ion channels (or other proteins) that requireneuromuscular interactions for full expression. A natural al-ternative would be to coculture human iPS-derived skeletalmuscle cells and motoneurons, the latter being currently avail-able for investigation (24).In different muscle types (cardiac, smooth, skeletal), excita-

tion–contraction coupling relies on homologous molecular ma-chinery—that is, the DHPR L-type Ca2+ channels in the plasmamembrane and the RyR Ca2+ release channels in the SR mem-brane (25, 26). Crosstalk between these two molecular players is,however, unique in skeletal muscle, where it depends primarily onmembrane depolarization and proceeds in the absence of extra-cellular Ca2+. Therefore, the occurrence of a biphasic change in[Ca2+]i upon K+-induced membrane depolarization in the absenceof extracellular Ca2+ provides evidence for a skeletal-muscle–specific EC coupling in ES- and iPS-myocytes, and is in line withthe presence of skeletal-muscle L-type Ca2+ currents. The rate ofrise in [Ca2+]i, which reflects the rate of Ca2+ release, showeda nonuniform distribution, suggesting various degrees of differ-entiation for EC coupling in both cell types, reminiscent of whathas been reported in developing skeletal muscle (27–29). None-theless, the fastest rates of release (lowest dt values), which shouldreflect more mature EC coupling, concerned the majority of cells.This functional heterogeneity is substantiated by the occurrence ofvarious structural arrangements of the CRUs.The biexponential decay of the intracellular Ca2+ signal sug-

gests that at least two distinct processes account for the bufferingof Ca2+ ions released from the SR to the cytosol. This is in linewith previous reports on skeletal muscle EC coupling using ei-ther K+-induced depolarization or electric stimulation (27–29).The initial, relatively fast decay phase likely corresponds to thebinding of cytosolic Ca2+ to fast Ca2+ buffers (troponin) (30).This initial process is relatively uniform in both ES- and iPS-myocytes, suggesting a well-developed fast buffering system inboth cell types. In contrast, the slow decay phase displayed a highervariability for both cell types and slower rates in iPS-myocytes. In-terestingly, it has been shown that the slow component of the in-tracellular Ca+2 decay becomes faster with vertebrate development(27–29). This decay involves slow Ca2+ buffering via Ca2+-bindingproteins (parvalbumin), and Ca2+ transfer from the cytosol to theSR lumen by the sarco/endoplasmic reticulum Ca2+-ATPase (30) orto the extracellular space by plasma membrane Ca2+ATPase andNa+/Ca2+ exchanger (31). Taken together, our data suggest that theslow Ca2+ buffering processes that follow muscle contraction maystill remain immature under these culture conditions in both celltypes and to a higher extent in iPS-myocytes.So far there have been very few reports on the ultrastructural

changes of human skeletal muscle embryogenesis in general (32,33), and no detailed analysis on myofibrillogenesis or the for-mation of calcium release units. We show herein that both ES-and iPS-derived myogenic cells develop similarly and reach,within a month of differentiation, a high degree of sarcomericorganization with noticeable A-, I-, Z-, and M-bands. T–SR

Skoglund et al. PNAS | June 3, 2014 | vol. 111 | no. 22 | 8279

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Oct

ober

10,

202

0

junctions appeared during the first 2 wk of differentiation. Theywere abundant, peripheral, or internal with various shapes, sizes,locations, and seldom association to myofibrils. Transverse ori-entation at the A–I junction, which represents the final step inthe triad organization (34), was not encountered. This supportsthe idea that T–SR junctions occur distantly from the myofibrilsand do not reach maturity in muscle cell cultures, whereasmyofibrils do (35). Internal CRU polymorphism and nonassociationwith the myofibrils have been reported in rodent aneural primarymyocytes in vitro (28, 36) and in denervated muscle in vivo (37),speaking for the plausible influence of innervation on CRU or-ganization. Peripheral couplings, which are considered as rudi-mentary structures in skeletal muscle, were frequent in bothES- and iPS-myocytes at different stages of their myofibrillar dif-ferentiation. This is reminiscent of in vitro vertebrate skeletalmuscle development (28) but differs from the in vivo situationwhere peripheral couplings precede the formation of internalCRUs (14, 38). Although peripheral couplings represent a land-mark of cardiac muscle structure, signs of myocardiogenesis suchas the presence of adhering structures (“primitive” intercalateddisks) were not encountered in the present survey. Furthermore,in contrast to iPS-derived skeletal myocytes, cardiomyocytes dis-play very few internal T–SR junctions (39) and no M-bands evenafter 4 mo in culture (40).In conclusion, our results show that human iPS-derived myo-

cytes acquire, in culture, functional and ultrastructural land-marks of skeletal muscle. Their characteristics and degree of

maturity are comparable to those of their ES counterparts, val-idating thereby their use in pathophysiological studies. Cocul-turing iPS-derived skeletal myocytes with motoneurons shouldenhance their degree of maturity and optimize the model forskeletal muscle research and drug discovery.

Materials and MethodsA detailed description of the materials and methods is provided in SI Materialsand Methods.

Myogenic progenitors were generated from the human H9 ES cell line anda fully characterized iPS cell line, both stably transfected with an induciblelentiviral PAX7 gene construct (7) and differentiated in DMEM containing2% (vol/vol) horse serum (Gibco). APs and voltage-gated ion currents wererecorded with the whole-cell patch-clamp technique. EC coupling was trig-gered by high K+-induced depolarization, and the ensuing changes in [Cai]iwere measured using fura-2–based microfluorometry (41). MHC and α-acti-nin were labeled using indirect immunofluorescence. The ultrastructure ofES- and iPS-derived cells was analyzed on ultrathin sections (70 nm) usingtransmission electron microscopy. All results are expressed as means ± SEM.Unpaired, two-tailed Student t test was generally used for statistical analysis.In the case of nonparametric distribution determined by a Shapiro–Wilk Wtest, the Mann–Whitney U test was applied.

ACKNOWLEDGMENTS. We thank F. Moueza for her technical assistance,Laurence Lanfumey for kindly lending us the calcium imaging system, StefanElmquist for reading the manuscript, and the Association Française contre lesMyopathies for financial support (to N.T.). R.P. was supported by NationalInstitutes of Health Grants R01 AR055299 and RC1AR058118.

1. Takahashi K, et al. (2007) Induction of pluripotent stem cells from adult human fi-broblasts by defined factors. Cell 131(5):861–872.

2. Park IH, et al. (2008) Disease-specific induced pluripotent stem cells. Cell 134(5):877–886.

3. Itzhaki I, et al. (2011) Modelling the long QT syndrome with induced pluripotent stemcells. Nature 471(7337):225–229.

4. Moretti A, et al. (2010) Patient-specific induced pluripotent stem-cell models for long-QT syndrome. N Engl J Med 363(15):1397–1409.

5. Darabi R, et al. (2011) Functional myogenic engraftment from mouse iPS cells. StemCell Rev 7(4):948–957.

6. Darabi R, et al. (2011) Assessment of the myogenic stem cell compartment followingtransplantation of Pax3/Pax7-induced embryonic stem cell-derived progenitors. StemCells 29(5):777–790.

7. Darabi R, et al. (2012) Human ES- and iPS-derived myogenic progenitors restoreDYSTROPHIN and improve contractility upon transplantation in dystrophic mice. CellStem Cell 10(5):610–619.

8. Yang JS, Sladky JT, Kallen RG, Barchi RL (1991) TTX-sensitive and TTX-insensitive so-dium channel mRNA transcripts are independently regulated in adult skeletal muscleafter denervation. Neuron 7(3):421–427.

9. Gonoi T, Sherman SJ, Catterall WA (1985) Voltage clamp analysis of tetrodotoxin-sensitive and -insensitive sodium channels in rat muscle cells developing in vitro.J Neurosci 5(9):2559–2564.

10. Brinkmeier H, Zachar E, Rüdel R (1991) Voltage-dependent K+ channels in the sar-colemma of mouse skeletal muscle. Pflugers Arch 419(5):486–491.

11. Adams BA, Tanabe T, Mikami A, Numa S, Beam KG (1990) Intramembrane chargemovement restored in dysgenic skeletal muscle by injection of dihydropyridine re-ceptor cDNAs. Nature 346(6284):569–572.

12. Beam KG, Knudson CM (1988) Calcium currents in embryonic and neonatal mam-malian skeletal muscle. J Gen Physiol 91(6):781–798.

13. Morrill JA, Brown RH, Jr., Cannon SC (1998) Gating of the L-type Ca channel in humanskeletal myotubes: An activation defect caused by the hypokalemic periodic paralysismutation R528H. J Neurosci 18(24):10320–10334.

14. Franzini-Armstrong C, Jorgensen AO (1994) Structure and development of E-C cou-pling units in skeletal muscle. Annu Rev Physiol 56:509–534.

15. Hamann M, et al. (1994) Sodium and potassium currents in freshly isolated and inproliferating human muscle satellite cells. J Physiol 475(2):305–317.

16. Heubach JF, et al. (2004) Electrophysiological properties of human mesenchymal stemcells. J Physiol 554(Pt 3):659–672.

17. Alioua A, et al. (2002) Coupling of c-Src to large conductance voltage- and Ca2+-activated K+ channels as a new mechanism of agonist-induced vasoconstriction. ProcNatl Acad Sci USA 99(22):14560–14565.

18. Jurkat-Rott K, et al. (2000) Voltage-sensor sodium channel mutations cause hypoka-lemic periodic paralysis type 2 by enhanced inactivation and reduced current. ProcNatl Acad Sci USA 97(17):9549–9554.

19. Enyedi P, Czirják G (2010) Molecular background of leak K+ currents: Two-pore do-main potassium channels. Physiol Rev 90(2):559–605.

20. Sartiani L, et al. (2007) Developmental changes in cardiomyocytes differentiated fromhuman embryonic stem cells: A molecular and electrophysiological approach. StemCells 25(5):1136–1144.

21. Merickel M, Gray R, Chauvin P, Appel S (1981) Electrophysiology of human muscle inculture. Exp Neurol 72(2):281–293.

22. Wood SJ, Slater CR (1995) Action potential generation in rat slow- and fast-twitchmuscles. J Physiol 486(Pt 2):401–410.

23. Almers W, Roberts WM, Ruff RL (1984) Voltage clamp of rat and human skeletal muscle:Measurements with an improved loose-patch technique. J Physiol 347:751–768.

24. Dimos JT, et al. (2008) Induced pluripotent stem cells generated from patients withALS can be differentiated into motor neurons. Science 321(5893):1218–1221.

25. Amador FJ, Stathopulos PB, Enomoto M, Ikura M (2013) Ryanodine receptor calciumrelease channels: Lessons from structure-function studies. FEBS J 280(21):5456–5470.

26. Endo M (2009) Calcium-induced calcium release in skeletal muscle. Physiol Rev 89(4):1153–1176.

27. Capote J, Bolaños P, Schuhmeier RP, Melzer W, Caputo C (2005) Calcium transients indeveloping mouse skeletal muscle fibres. J Physiol 564(Pt 2):451–464.

28. Flucher BE, Andrews SB, Daniels MP (1994) Molecular organization of transversetubule/sarcoplasmic reticulum junctions during development of excitation-contrac-tion coupling in skeletal muscle. Mol Biol Cell 5(10):1105–1118.

29. Zhang XH, Wu J, Shen MX, Zhu PH (1999) Calcium release induced by high K+ andcaffeine in cultured skeletal muscle cells of embryonic chicken. Pflugers Arch 438(6):827–836.

30. Baylor SM, Hollingworth S (2012) Intracellular calcium movements during excitation-contraction coupling in mammalian slow-twitch and fast-twitch muscle fibers. J GenPhysiol 139(4):261–272.

31. Berchtold MW, Brinkmeier H, Müntener M (2000) Calcium ion in skeletal muscle: Itscrucial role for muscle function, plasticity, and disease. Physiol Rev 80(3):1215–1265.

32. Tomanek RJ, Colling-Saltin AS (1977) Cytological differentiation of human fetal skeletalmuscle. Am J Anat 149(2):227–245.

33. Tanaka O, Shinohara H, Oguni M, Yoshioka T (1995) Ultrastructure of developingmuscle in the upper limbs of the human embryo and fetus. Anat Rec 241(3):417–424.

34. Takekura H, Fujinami N, Nishizawa T, Ogasawara H, Kasuga N (2001) Eccentric exer-cise-induced morphological changes in the membrane systems involved in excitation-contraction coupling in rat skeletal muscle. J Physiol 533(Pt 2):571–583.

35. Flucher BE (1992) Structural analysis of muscle development: Transverse tubules,sarcoplasmic reticulum, and the triad. Dev Biol 154(2):245–260.

36. Tanaka H, Furuya T, Kameda N, Kobayashi T, Mizusawa H (2000) Triad proteins andintracellular Ca2+ transients during development of human skeletal muscle cells inaneural and innervated cultures. J Muscle Res Cell Motil 21(6):507–526.

37. Schiaffino S, Settembrini P (1970) Studies on the effect of denervation in developingmuscle. I. Differentiation of the sarcotubular system. Virchows Arch B Cell Pathol InclMol Pathol 4(4):345–356.

38. Kelly AM (1971) Sarcoplasmic reticulum and T tubules in differentiating rat skeletalmuscle. J Cell Biol 49(2):335–344.

39. Gherghiceanu M, et al. (2011) Cardiomyocytes derived from human embryonic and in-duced pluripotent stem cells: Comparative ultrastructure. J Cell Mol Med 15(11):2539–2551.

40. Lundy SD, Zhu WZ, Regnier M, Laflamme MA (2013) Structural and functional mat-uration of cardiomyocytes derived from human pluripotent stem cells. Stem Cells Dev22(14):1991–2002.

41. Grynkiewicz G, Poenie M, Tsien RY (1985) A new generation of Ca2+ indicators withgreatly improved fluorescence properties. J Biol Chem 260(6):3440–3450.

8280 | www.pnas.org/cgi/doi/10.1073/pnas.1322258111 Skoglund et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

10,

202

0