THE ANATOMICAL RECORD PART A 280A:1046 –1052 (2004) Recreating the Biological Pacemaker MICHAEL R. ROSEN,1–3* RICHARD B. ROBINSON,1 PETER BRINK,3 1,3 AND IRA S. COHEN 1 Center for Molecular Therapeutics, Department of Pharmacology, Columbia University, New York, New York 2 Department of Pediatrics, Columbia University, New York, New York 3 Institute of Molecular Cardiology, Departments of Physiology and Biophysics, State University of New York, Stony Brook, New York ABSTRACT In recent years, several groups have reported a variety of strategies for developing biological pacemakers whose ultimate function would be to supplement/replace electronic pacemakers. Strategies have included gene therapy using naked plasmids or viral vectors and cell therapy for which both adult human mesenchymal stem cells (hMSCs) and human embryonic stem cells have been employed. This article reviews the various approaches and summarizes our own research in which the pacemaker gene, HCN2, is administered via viral vector or in an hMSC platform to produce pacemaker function in the intact canine heart. © 2004 Wiley-Liss, Inc. Key words: pacemaker current; HCN channels; heart block; cardiac arrhythmias; sinoatrial node In a variety of settings of sinoatrial and/or atrioventricular nodal disease, the ideal solution would be to heal or to recreate the biological pacemaker of the heart, the sinoatrial node, itself. At present, electronic pacemakers represent the state of the art. However, limited battery life, the need for permanent catheter implantation into the heart, and lack of response to autonomic neurohumors all render them less than optimal therapy (Zivin and Bardy, 2001). Several gene therapy approaches have been explored as alternatives to electronic pacemakers, including ␤2-adrenergic receptor overexpression (Edelberg et al., 1998, 2001), suppression of inward rectiﬁer current when a dominant negative construct is expressed together with the wild-type gene Kir2.1 (Miake et al., 2002), and implantation of vectors carrying the pacemaker gene, HCN2, into the atrium (Qu et al., 2003) or the bundle branch system (Plotnikov et al., 2004). The ␤2-adrenergic receptor experiments used a naked plasmid as the vector for gene therapy; the others, an adenoviral vector. A problem inherent in many gene therapy experiments (Miake et al., 2002; Qu et al., 2003; Plotnikov et al., 2004) is the use of viruses to deliver the necessary genes. Use of replication-deﬁcient adenoviruses having little infectious potential raises the likelihood of only transient improvement in pacemaker function as well as potential inﬂammatory responses. Retroviruses and other viral vectors carry the risk of carcinogenicity and infectivity. One way to avoid the use of viral vectors is to rely on cell therapy using human embryonic stem cells to create pacemakers. Such approaches are not only in their infancy but carry the problems of identifying appropriate cell lineages, the © 2004 WILEY-LISS, INC. possibility of stem cell differentiation into lines other than pacemaker cells, and potential for neoplasia (Gepstein, 2002). Another cell therapy approach for making biological pacemakers would be to use adult mesenchymal stem cells as platforms for delivering pacemaker genes to the heart (Potapova et al., 2004). In this setting, the cells would be thought of as a biologically inert vector that could deliver molecular/genetic information to adjoining myocardium. We review here our use of both viral gene therapy and adult human mesenchymal stem cell (hMSC) approaches to test the hypothesis that biological pacemakers may become feasible alternatives to electrical pacemakers. These experiments followed our initial demonstration of proof of concept that an adenoviral vector containing HCN2 delivered to the canine left atrium could generate a pacemaker current 100- to 1,000-fold greater than the Grant sponsor: USPHS-NHLBI; Grant number: HL-28958, HL20558, GM-55263, HL-67101; Grant sponsor: Guidant Corporation. *Correspondence to: Michael R. Rosen, Center for Molecular Therapeutics, Departments of Pharmacology and Pediatrics, Columbia University, 630 West 168 Street, PH 7 West-321, New York, NY 10032. Fax: 212-305-8351. E-mail: email@example.com Received 11 March 2004; Accepted 8 June 2004 DOI 10.1002/ar.a.20073 Published online 15 September 2004 in Wiley InterScience (www.interscience.wiley.com). RECREATING THE BIOLOGICAL PACEMAKER native atrial If and could initiate atrial beats (Qu et al., 2003). In the gene therapy experiment to be summarized here, we implanted an adenoviral construct into the left bundle branch (LBB) system and determined whether the rhythms that developed during transient atrioventricular block were stable and organized and would result from IHCN2 induction of automaticity in the region injected. In the stem cell experiment, we tested the hypothesis that the expressed pacemaker current need not be generated within a myocyte, but that it would be sufﬁcient if it were generated within any cell electrotonically coupled to a myocyte. For this purpose, we investigated whether hMSCs could be transfected by electroporation with HCN2, could express functional HCN2 channels, and could drive the canine ventricle, mimicking the HCN2 overexpression by adenoviral constructs (Qu et al., 2003), but without the need for in situ viral delivery. MATERIALS AND METHODS All studies were performed using protocols approved by the Columbia University Institutional Animal Care and Use Committee. Viral/Genetic Preparation An adenoviral construct of murine HCN2 (mHCN2; Genbank AJ225122) driven by the CMV promoter was prepared (Qu et al., 2003), puriﬁed through plaque assay, ampliﬁed to a large stock, and harvested and titrated after CsCl banding. The identical procedure was used to construct an adenoviral vector of enhanced GFP (AdGFP), whose sequence was taken from its original vector pIRES2-EGFP (Clontech, Palo Alto, CA) at BamHI and NotI sites and subcloned into the shuttle vector pDC516. Final titers were AdHCN2, 3.4 ⫻ 1011 ffu/mL; AdGFP, 1.4 ⫻ 1012 ffu/mL; 2–3 ⫻ 1010 ffu of each virus was injected per experiment. Dissociation of Myocytes and Studies of HCN Current Purkinje myocytes were dissociated by modifying a previously published procedure (Boyden et al., 1989). Isolated cells were transferred to a stage-mounted chamber of an inverted epiﬂuorescence microscope to identify GFP-expressing cells. To measure pacemaker currents, cells were superfused with 35°C Tyrode solution containing (in mmol/L): NaCl, 140; NaOH, 2.3; MgCl2, 1; KCl, 5.4; CaCl2, 1.0; MnCl2, 2; BaCl2, 4; HEPES, 5; glucose, 10; pH 7.4. Pipette solution contained (in mmol/L) aspartic acid, 130; KOH, 146; NaCl, 10; CaCl2, 2; EGTA-KOH, 5; Mg-ATP, 2; HEPES-KOH, 10; pH 7.2. To record pacemaker current, cells were held at ⫺55 mV and stepped to ⫺55 to ⫺125 mV for 6 sec, followed by an 8-sec step to ⫺115 mV to measure tail current. Human Mesenchymal Stem Cell Maintenance and Transfection hMSCs (Poietics hMSC, mesenchymal stem cells, human bone marrow) were purchased from Clonetics/BioWhittaker (Walkersville, MD) and cultured in MSC growing medium (Poietics MSCGM; BioWhittaker) at 37°C in a humidiﬁed atmosphere of 5% CO2. Cells were used from passages 2– 4. A full-length murine HCN2 cDNA was subcloned into a pIRES2-EGFP vector (BD Biosciences Clon- 1047 tech, Palo Alto, CA). Cells were transfected by electroporation using the Amaxa Biosystems Nucleofector (Amaxa, Cologne, Germany) technology (Hamm et al., 2002). Transfection efﬁciency was 30 – 45%. Study of membrane currents in hMSCs was via the techniques described above. Action Potential Recordings in Coculture hMSCs were plated onto ﬁbronectin-coated coverslips using a cloning cylinder to restrict initial plating to a 4 mm diameter circular area. The cells expressed EGFP alone or EGFP ⫹ mHCN2. Four hours later, the cloning cylinder was removed and neonatal rat ventricular myocytes, prepared as described previously (Protas and Robinson, 1999), were plated over the entire coverslip. Four to ﬁve days later, the coverslips were placed in a superfusion chamber maintained at 35°C and action potentials recorded from near the center of the coverslip using a perforated patch electrode (Protas and Robinson, 1999) and normal physiologic solution. Recordings were conducted with an Axopatch 200 ampliﬁer and PClamp 8 software (Axon Instruments). The perforated patch technique was used with amphotericin B (400 g/mL; Sigma) in the pipette solution. Intact Animal Studies With Viral Injection or hMSCs Adult mongrel dogs were anesthetized using Na-thiopental induction (17 mg/kg IV) followed by inhalational isoﬂurane (1.5–2.5%). For viral injection, a custom-modiﬁed bipolar 8 Fr steerable catheter (Guidant, St. Paul, MN) was used for subendocardial delivery of the adenoviral constructs or saline. A 29 G needle, which could be advanced and retracted by 3 mm, was incorporated into the core lumen of the catheter. The catheter ﬁlled with sterile saline was introduced into the LV under ﬂuoroscopic control via an 8 Fr arterial introducer sheath, a stable bundle branch potential electrogram recording was obtained, the needle tip was veriﬁed in the LV wall, and an adenoviral construct (AdGFP ⫹ AdHCN2 or AdGFP alone) or normal saline solution (0.6 ml) was injected into the LBB (Plotnikov et al., 2004). Animals recovered for 4 –7 days, after which a terminal study was performed in which they were anesthetized as above and both cervical vagal trunks isolated as described previously (Plotnikov et al., 2004). Graded right and left vagal stimulation was performed via bipolar platinum iridium electrodes (Rosenshtraukh et al., 1994) to suppress sinus rhythm such that escape pacemaker function might occur. The chest was then opened and LBB excised from the left ventricular sites that had been injected. Comparable procedures were performed for AdGFP- or saline-injected or control dogs. These were used for microelectrode study or study of ion channels. For microelectrode studies, preparations were placed in a 4 ml chamber perfused with Tyrode’s solution (37°C; pH 7.3–7.4) and impaled with 3 mol/l KCl-ﬁlled glass capillary microelectrodes. Calibration of the system and recording of transmembrane action potential were performed as described previously (Anyukhovsky et al., 1996). For hMSC studies, anesthesia was as above, a thoracotomy was performed, and we injected 106 hMSCs containing mHCN2 ⫹ GFP or GFP alone subepicardially in 0.6 ml of solution into the left ventricular anterior wall, approx- 1048 ROSEN ET AL. imately 2 mm deep to the epicardium via a 21 gauge needle. Animals recovered for 4 –10 days and were anesthetized as above. Both cervical vagal trunks were isolated, the chest opened, and ECGs monitored. Graded right and left vagal stimulation was performed to suppress sinus rhythm such that escape pacemaker function might occur. Tissues were then removed for histological study. Statistical Analysis Measurements of cycle lengths of cardiac rhythms and of isolated LBB were made from at least ﬁve beats. Fisher’s exact test, Student’s t-test, or one-way ANOVA were used as appropriate. Data are expressed as mean ⫾ SEM. P ⬍ 0.05 was considered signiﬁcant. RESULTS Gene Therapy Protocols We studied seven animals injected with mHCN2 ⫹ AdGFP, three with AdGFP alone, and three with saline. Seven of the animals were monitored round the clock following surgery. We noted that for 2 days following the injection procedure, there was increased ventricular irritability expressed as multiple premature ventricular depolarizations in the saline-injected, AdGFP-injected, and AdGFP ⫹ AdHCN2-injected animals. This ectopy was attributed to local tissue injury, as was exempliﬁed by hematoma formation at the injection site in all three groups. Hence, no studies of pacemaker function were done until day 4 –7. Figure 1 demonstrates representative experiments showing records from one dog injected with AdGFP alone (Fig. 1A) and one with AdGFP ⫹ AdHCN2 (Fig. 1B). Prior to vagal stimulation, both dogs had comparable sinus rates. Following vagal stimulation, the rate of the AdGFP ⫹ AdHCN2-injected heart was over two times faster than that injected with AdGFP alone. Overall, baseline sinus cycle length did not differ between control and AdHCN2-injected groups (484 ⫾ 46 and 506 ⫾ 11 msec, respectively; P ⬎ 0.05). During vagal stimulation, right ventricular escape rhythms occurred in four of six controls and three of seven HCN2 dogs (P ⬎ 0.05), while left ventricular escape rhythms occurred in four of six controls and six of seven HCN2-injected dogs (P ⬎ 0.05). Whereas idioventricular escape of right or left ventricular origin occurred in all animals, the rate generated was signiﬁcantly more rapid in the hearts of animals that had received a left bundle branch injection of AdHCN2 than was seen for left ventricular rhythms of control animals and for right ventricular rhythms of control or AdHCN2-injected animals (P ⬍ 0.05) (Plotnikov et al., 2004). Pacemaker function was studied in 16 experiments on bundle branches from animals injected with AdHCN2 ⫹ AdGFP, AdGFP alone, or saline as well as from the same site in ﬁve control dogs that had not been injected at all. Rate was most rapid (P ⬍ 0.05) in the AdHCN2-injected regions (95 ⫾ 10 vs. 42 ⫾ 3 bpm at control sites; P ⬍ 0.05) (Plotnikov et al., 2004). Figure 2 demonstrates the magnitude of IHCN2 in a representative Purkinje myocyte from an animal injected with AdHCN2 ⫹ AdGFP and from a control myocyte. Note the greater magnitude of current in the setting of HCN2 overexpression. Stem Cell Therapy Protocols Whereas nontransfected hMSCs demonstrated no signiﬁcant time-dependent currents during hyperpolarizations, HCN2-transfected hMSCs expressed a large time-dependent inward current activating upon hyperpolarization (Fig. 3). Moreover, the reversal potential (⫺37.5 ⫾ 1.0 mV at [K⫹]o ⫽ 5.4 mM) and response to cesium were consistent with the pacemaker current, If. Given these properties, the current expressed in hMSCs would be expected to activate in the physiologic range of diastolic potentials as long as adequate gap junctional coupling was present between hMSCs and adjacent myocytes. That such coupling (involving connexins43 and 40 in hMSCs and connexin43 in myocytes) is in fact a reality among hMSCs and between hMSCs and myocytes was demonstrated by us recently (Potapova et al., 2004; Valiunas et al., 2004). Because a potential advantage of biological over electronic pacemakers would be their hormonal regulation, we examined the effects of ␤-adrenergic and muscarinic agonists on If recorded in the hMSCs. Isoproterenol induced a signiﬁcant leftward shift in the activation curve for If, consistent with the literature on ␤-adrenergic stimulation in myocytes. Whereas acetylcholine had no direct effect on If, it did antagonize the effects of isoproterenol. This is consistent with the accentuated antagonist action of acetylcholine on If in Purkinje ﬁbers (Chang et al., 1990). Given the expression of the pacemaker gene in hMSCs, we hypothesized that HCN2-transfected hMSCs could inﬂuence excitability of coupled heart cells. Maximum diastolic potential was ⫺74 ⫾ 1 mV (n ⫽ 5) in neonatal rat ventricular myocytes cocultured with EGFP-expressing hMSCs and ⫺67 ⫾ 2 mV (n ⫽ 6) in myocytes cocultured with hMSCs expressing HCN2 (P ⬍ 0.05). Spontaneous rate was 93 ⫾ 16 bpm in the former group (n ⫽ 5) and 161 ⫾ 4 bpm in the latter (n ⫽ 6; P ⬍ 0.05). The reduced maximum diastolic potential is consistent with the observed threshold potential of the expressed current in the HCN2-transfected hMSCs and indicates an effect of this depolarizing current on the electrically coupled myocytes (Potapova et al., 2004). We next injected HCN2-expressing hMSCs into canine heart in situ to test whether pacemaker function was demonstrable. During sinus arrest, ﬁve of six animals receiving hMSCs expressing EGFP ⫹ HCN2 developed rhythms originating from and pace-mapped to the left ventricle at a site whose origin approximated that of the hMSC injection (Potapova et al., 2004). Moreover, the idioventricular rates of these animals was 61 ⫾ 5 vs. 45 ⫾ 1 bpm in animals receiving hMSCs expressing EGFP alone (P ⬍ 0.05). Study of the region of injection via H&E staining demonstrated basophilic cells that were immunohistochemically positive for vimentin and for the CD44 antigen. Interdigitation between hMSCs and myocardium was demonstrable (Potapova et al., 2004). DISCUSSION Electronic pacemakers having high reliability and low morbidity are a primary therapy for complete heart block or sinus node dysfunction (Zivin and Bardy, 2001). Important negative aspects of electronic pacemakers are their lack of the biological responsiveness required for the autonomic and physiologic demands on the heart, the need for maintaining catheter electrodes in the heart, and the requirement for battery replacement. These limitations RECREATING THE BIOLOGICAL PACEMAKER Fig. 1. Representative experiments using AdGFP injection (A) and AdGFP ⫹ AdHCN2 injection (B). Note that at the outset (left side of each panel), sinus rhythm of comparable rate occurs in both animals. Vagal stimulation was followed by a slow idioventricular rhythm in the AdGFPinjected dog (A; the interval between left and right traces lasted 22 sec, near the outset of which vagal stimulation was initiated). In the AdHCN2- 1049 injected dog (B), vagal stimulation (see arrow) was followed by a far more rapid idioventricular rhythm. The interval between left and right traces was of 5-sec duration. Insets are magniﬁcations of the lead II and RV electrogram impulses indicated in the basic traces. Reprinted with permission from Plotnikov et al. (2004). 1050 ROSEN ET AL. have provided the impetus for developing biological pacemakers. Among the approaches attempted to provide such pacemaker function have been upregulation of ␤2-adrenergic receptors, downregulation of the background K⫹ current IK1, and overexpression of HCN2 channels, the molecular correlate of the endogenous cardiac pacemaker current If (Edelberg et al., 1998, 2001; Miake et al., 2002; Qu et al., 2003; Plotnikov et al., 2004). All three approaches have initially employed either plasmid injection into the heart (␤2-adrenergic receptors) or viral vectors (IK1 downregulation and HCN2 overexpression). One potential ﬂaw in ␤2-adrenergic receptor overexpression is that endogenous cardiac pacemaker mechanisms (which may be diseased) are left intact and the ␤-receptor is used as a nonspeciﬁc stimulator of heart rate: action will be on not only pacemaker current but on other catecholamine-sensitive currents. With regard to IK1, downregulation of this current subtracts an important determinant of repolarization, such that cells lacking this current manifest prolonged repolarization (Miake et al., 2002). This may set the stage for excess dispersion of repolarization. Moreover, it is not yet certain what inward current drives the pacemaker in this setting; existing data suggest it is a hybrid current rather than the wild-type If pacemaker. We have elected to work with the HCN channel family, as this family encodes the ␣-subunit of the wild-type If that provides primary pacemaker initiation in the heart (DiFrancesco, 1981, 1982). Clearly, HCN2 overexpression locally in left atrium (Qu et al., 2003) or in the proximal bundle branch system (Plotnikov et al., 2004) induces both If-like currents and in situ pacemaker function in transfected myocytes. The voltage dependence of the If conductance results in current ﬂow during phase 4 but not during the action potential plateau limiting the potential arrhythmogenicity resulting from altered action potential waveforms. Yet our use of an adenoviral construct to deliver the HCN2 gene to the heart (Qu et al., 2003; Plotnikov et al., 2004) is not an optimal approach because adenoviruses are episomal and the nucleic acids they deliver do not integrate into the genome. Other viral systems that generate more sustained transgene expression are accompanied by other important drawbacks that hinder their application in vivo. The use of stem cell therapy as an alternative to viral gene transfer is attractive ﬁrst of all because there is no need to depend on a virus. Both embryonic stem cells and mesenchymal stem cells have been tested here. Several laboratories are exploring the use of embryonic stem cells that can be differentiated along a cardiac lineage and might provide a cell-based control of cardiac rhythm. Among the advantages of these cells is that they make functional gap junctions and generate spontaneous rhythms (Gepstein, 2002). There are problems as well, however, including the immunogenicity of the cells, the potential for neoplasia, spatial nonuniformity of the implants, and the proper engineering of pure cardiac lineages (Gepstein, 2002). We found hMSCs to be an attractive platform for delivering genes to the heart for several reasons: they can be obtained in relatively large numbers through standard clinical interventions, are easily expanded in culture, are capable of long-term transgene expression, and their administration can be autologous or via banked stores (as they are immunoprivileged). While they might in theory be differentiated in vitro into cardiac-like cells capable of spontaneous activity, we have taken a genetic engineering approach that does not depend on differentiation along a speciﬁc lineage. Moreover, the ex vivo transfection method we used allows evaluation of DNA integration and engineering of the cell carriers with fail-safe death mechanisms. Both our adenoviral and our genetically engineered hMSCs expressed an If-like current and were capable of functioning as biological pacemakers in intact canine heart. Control adenoviruses or control hMSCs expressing only EGFP did not exert these effects. Hence, the electrical effects of hMSCs transfected with the mHCN2 gene were similar to the effects of overexpression of the same gene in the myocytes in vitro and in vivo. These ﬁndings suggest that hMSCs may serve as an alternative approach for the delivery of pacemaker genes for cardiac implantation. Yet there are several important caveats with regard to our results. Some relate to the technique of construct injection, particularly in experiments in which adenoviral constructs were injected into the canine left bundle branch (Plotnikov et al., 2004). This intervention induces hematoma formation that is for a short period arrhythmogenic. Importantly, hematoma formation occurred regardless of whether we injected AdHCN2 ⫹ AdGFP, AdGFP alone, or saline. In other words, it was the result of trauma from the needle per se. It is likely that this injury contributed to the tachycardias that occurred in the ﬁrst 24 – 48 hr after cardiac catheterization and that were universal across the three groups of animals. Moreover, the fact that both in vivo and in vitro accelerated pacemaker function was seen only in the animals that received HCN2, despite the fact that all groups underwent injury, suggests that the expression of pacemaker function was not the result of injury. A major question relates to the duration of efﬁcacy of these pacemakers. In the present study, we were only concerned with demonstrating the feasibility of applying gene therapy via an adenoviral vector or using hMSCs as a gene delivery system. Since our studies in vivo lasted less than 2 weeks, transient transfections were sufﬁcient to demonstrate proof of concept. Before this approach can be considered clinically relevant, far longer periods of study will be required. In this regard, our transfected hMSCs maintain their green ﬂuorescence for at least 3 months, indicating that we have selected for stable clones expressing HCN2. Hence, it is likely that persistence of expression will not pose signiﬁcant difﬁculties for more prolonged studies. Nonetheless, it remains to be determined if the differentiation state of the hMSCs is altered in situ in the long term, or whether such differentiation would affect HCN2 expression or biophysical properties. In addition, we used a murine gene, which is quite close but not completely identical in sequence to the human gene. It would be most advantageous to use human genes and to explore various mutations to optimize activation and recovery characteristics, as well as neurohumoral response. Such approaches are currently being explored. In comparing results between the two approaches, it is important to emphasize the conceptual and practical differences that exist in the design of gene therapy and stem cell therapy. While both have one endpoint in common— the delivery of a biological pacemaker— gene therapy uses speciﬁc HCN isoforms to engineer a cardiac myocyte into a pacemaker cell, whereas hMSC therapy uses stem cells RECREATING THE BIOLOGICAL PACEMAKER 1051 Fig. 2. Patch clamp recording of native If (control dog) and overexpressed IHCN2 in Purkinje cells. Note the different scales in both panels and the markedly greater current magnitude in the presence of overexpressed HCN2. Reprinted with permission from Plotnikov et al. (2004). as a platform to carry speciﬁc HCN isoforms to a heart whose myocytes retain their original function. Gene therapy makes use of preexisting homeotypic cell-cell coupling among myocytes to facilitate propagation of the pacemaker impulses from those myocytes in which pacemaker current is overexpressed to those that retain their original function. In contrast, stem cells depend on heterotypic coupling of cells with somewhat dissimilar populations of connexins to deliver pacemaker current alone from a stem cell to a myocyte whose function is left unchanged. Importantly, and unlike sinus node cells, HCN2-transfected hMSCs are not excitable, because they lack the other currents necessary to generate an action potential. However, when transfected, these cells generate a depolarizing current, which spreads to coupled myocytes, driving myocytes to threshold. In effect, the myocyte acts like a trip wire whose hyperpolarization turns on pacemaker current in the stem cell and whose depolarization turns off the current. We believe that as long as the hMSCs contain the pacemaker gene and couple to cardiac myocytes via gap junctions, they will function as a cardiac pacemaker in an analogous manner to the normal primary pacemaker the sinoatrial node. We must emphasize that a biological pacemaker needs an optimal size (in terms of cell mass) and an optimal cell-to-cell coupling for long-term normal function. In a sense, the outcome of the initial experiments was Fig. 3. Functional expression of If in hMSCs transfected with mHCN2 gene. If was expressed in hMSCs transfected with the mHCN2 gene (B) but not in nontransfected stem cells (A). C: The ﬁt by the Boltzmann equation to the normalized tail currents of If gives a midpoint of ⫺91.8 ⫾ 0.9 mV and a slope of 8.8 ⫾ 0.5 mV (n ⫽ 9). If was fully activated around ⫺140 mV with an activation threshold of ⫺60 mV. The inset shows representative tail currents used to construct If activation curves. The voltage protocol was to hold at ⫺30 mV and hyperpolarize for 1.5 sec to voltages between ⫺40 and ⫺160 mV in 10 mV increments followed by a 1.5-sec voltage step to 20 mV to record the tail currents. Reprinted with permission from Potapova et al. (2004). fortunate in that the construct used and the number of cells administered coupled and functioned as well as they did. We currently are performing experiments to identify what might be the optimal cell numbers and coupling ratios needed to optimize function. Both gene therapy and stem cell approaches currently have as one of their shortcomings the passage of seconds between the last normal heart beat and the onset of pacemaker function. Ideally, such an interval should be 1–2 sec in duration, which may require the engineering of mutant genes. Given the differences in cell-cell coupling in gene therapy and stem cell therapy, it is highly likely that the engineering of the mutant genes necessary to function as the ultimate biological pacemaker gene will differ importantly between the two approaches. It remains to be seen, with both biological pacemaker approaches, whether 1052 ROSEN ET AL. the duration of effect and the functional characteristics of the biological pacemakers will be competitive with electronic units currently in use and in their planning stages. Finally, the delivery of hMSCs expressing mHCN2 to the canine heart is not only a demonstration of feasibility of preparing hMSC-based biological pacemakers, but is the ﬁrst concrete example of a general principle; that is, hMSCs can be used to deliver a variety of genes to inﬂuence the function of tissues capable of forming gap junctional connections. Hence, the payload delivered by hMSCs need not be restricted to membrane channels; any gene product or small molecule that can permeate gap junctions (MW ⬍ 1,000, minor diameter ⬍ 1.2 nm) can be incorporated into the hMSCs and delivered to a gap junctional-coupled tissue as its therapeutic target. ACKNOWLEDGMENTS The authors thank Nimee Bhat for assistance in performing the studies and Ms. Eileen Franey for her careful attention to the preparation of the manuscript. Work on gene therapy alone was supported by Guidant Corporation. LITERATURE CITED Anyukhovsky EP, Sosunov EA, Rosen MR. 1996. Regional differences in electrophysiological properties of epicardium, midmyocardium, and endocardium: in vitro and in vivo correlations. Circulation 94:1981–1988 Boyden PA, Albala A, Dresdner KP Jr. 1989. Electrophysiology and ultrastructure of canine subendocardial Purkinje cells isolated from control and 24-hour infarcted hearts. Circ Res 65:955–970. Chang F, Gao J, Tromba C, Cohen I, DiFrancesco D. 1990. Acetylcholine reverses effects of ␤-agonists on pacemaker current in canine cardiac Purkinje ﬁbers but has no direct action: a difference between primary and secondary pacemakers. Circ Res 66:633– 636. DiFrancesco D. 1981. A study of the ionic nature of the pacemaker current in calf Purkinje ﬁbres. J Physiol (Lond) 314:377–393. DiFrancesco D. 1982. Block and activation of the pacemaker channel in calf Purkinje ﬁbres: effects of potassium, caesium and rubidium. J Physiol (Lond) 329:485–507. Edelberg JM, Aird WC, Rosenberg RD. 1998. Enhancement of murine cardiac chronotropy by the molecular transfer of the human ␤2adrenergic receptor cDNA. J Clin Invest 101:337–343. Edelberg JM, Huang DT, Josephson ME, Rosenberg RD. 2001. Molecular enhancement of porcine cardiac chronotropy. Heart 86:559 – 562. Gepstein L. 2002. Derivation and potential applications of human embryonic stem cells. Circ Res 91:866 – 876. Hamm A, Krott N, Breibach I, Blindt R, Bosserhoff AK. 2002. Efﬁcient transfection method for primary cells. Tissue Eng 8:235–245. Miake J, Marban E, Nuss HB. 2002. Gene therapy: biological pacemaker created by gene transfer. Nature 419:132–133. Plotnikov AN, Sosunov EA, Qu J, Shlapakova I, Anyukhovsky EP, Liu L, Janse MJ, Brink PR, Cohen IS, Robinson RB, Danilo P Jr, Rosen MR. 2004. A biological pacemaker implanted in the canine left bundle branch provides ventricular escape rhythms having physiologically acceptable rates. Circulation 109:506 –512. Potapova I, Plotnikov A, Lu Z, Danilo P Jr, Valiunas V, Qu J, Doronin S, Zuckerman J, Shlapakova IN, Gao J, Pan Z, Herron AJ, Robinson RB, Brink PR, Rosen MR, Cohen IS. 2004. Human mesenchymal stem cells as a gene delivery system to create cardiac pacemakers. Circ Res 94:952–959. Protas L, Robinson RB. 1999. Neuropeptide Y contributes to innervation-dependent increase in ICa,L via ventricular Y2 receptors. Am J Physiol 277:H940 –H946. Qu J, Plotnikov AN, Danilo P Jr, Slapakova I, Cohen IS, Robinson RB, Rosen MR. 2003. Expression and function of a biological pacemaker in canine heart. Circulation 107:1106 –1109. Rosenshtraukh L, Danilo P Jr, Anyukhovsky EP, Steinberg SF, Rybin V, Brittain-Valenti K, Molina-Viamonte V, Rosen MR. 1994. Mechanisms for vagal modulation of ventricular repolarization and of coronary occlusion-induced lethal arrhythmias in cats. Circ Res 75:722–732. Valiunas V, Doronin S, Valiuniene L, Potapova I, Zuckerman J, Walcott B, Robinson RB, Rosen MR, Brink PR, Cohen IS. 2004. Human mesenchymal stem cells make cardiac connexins and form functional gap junctions. J Physiol 555(Pt 3):617– 626. Zivin A, Bardy GH. 2001. Cardiac pacemakers. In: Spooner PM, Rosen MR, editors. Foundations of cardiac arrhythmias. New York: Marcel Dekker. p 571–598.