Anatomical and molecular mapping of the left and right ventricular His–Purkinjeconduction networks
Andrew Atkinson a,1, Shin Inada a,1, Jue Li a, James O. Tellez a, Joseph Yanni a, Rakan Sleiman b,Eman Abd Allah a, Robert H. Anderson a, Henggui Zhang b, Mark R. Boyett a,⁎,2, Halina Dobrzynski a,⁎⁎,2
a Cardiovascular Medicine, University of Manchester, UKb Biological Physics, University of Manchester, UK
⁎ Correspondence to: M.R. Boyett, Cardiovascular MeHuman Sciences, University of Manchester, Core TechnoManchester M13 9NT, UK.⁎⁎ Correspondence to: H. Dobrzynski, Cardiovascular MHuman Sciences, University of Manchester, Core TechnoManchester M13 9NT, UK. Tel.: +44 161 275 1182.
Functioning of the cardiac conduction system depends critically on its structure and its complement ofion channels. Therefore, the aim of this study was to document both the structure and ion channel expressionof the left and right ventricular His–Purkinje networks, as we have previously done for the sinoatrial andatrioventricular nodes. A three-dimensional (3D) anatomical computer model of the His–Purkinje network ofthe rabbit heart was constructed after staining the network by immunoenzyme labelling of a marker protein,middle neurofilament. The bundle of His is a ribbon-like structure and the architecture of the His–Purkinjenetwork differs between the left and right ventricles. The 3Dmodel is able to explain the breakthrough pointsof the action potential on the ventricular epicardium during sinus rhythm. Using quantitative PCR, theexpression levels of the major ion channels were measured in the free running left and right Purkinje fibres ofthe rabbit heart. Expression of ion channels differs from that of the working myocardium and can explain thespecialised electrical activity of the Purkinje fibres as suggested by computer simulations; the expressionprofile of the left Purkinje fibres is more specialised than that of the right Purkinje fibres. The structure and ionchannel expression of the Purkinje fibres are highly specialised and tailored to the functioning of the system.The His–Purkinje network in the left ventricle is more developed than that in the right ventricle and this mayexplain its greater clinical importance.
dicine, Faculty of Medical andlogy Facility, 46 Grafton Street,
edicine, Faculty of Medical andlogy Facility, 46 Grafton Street,
The Purkinje fibres (PFs)were discovered by Jan Purkinje in 1837 [1]and their anatomy was elegantly studied and illustrated by SunaoTawara at the beginning of the 20th century [2]. The PFs form the finalportion of the cardiac conduction system — they provide a rapidconduction pathway through the ventricles ensuring a coordinatedcontraction of the ventricles [3]. They are insulated from the ventricularmyocardium by a connective tissue sheath, which is lost before thePFs form terminal connections with the ventricular myocardium viaspecialised junctions in the endocardium [4–7].
The PFs are specialised: most importantly they are fast conducting,in part as a result of a high upstroke velocity during phase 0 of the
action potential [8]. They have other distinct action potentialcharacteristics: a prominent early rapid repolarisation (phase 1), anegative plateau potential (phase 2), an increased action potentialduration, and spontaneous diastolic depolarisation (phase 4) [8].Normally the PFs do not exhibit pacemaker activity, because ofoverdrive suppression by sinus rhythm, but in heart block they act asan escape pacemaker [9]. They also play a role in the generationand maintenance of arrhythmias — they support reentry [10], sustainventricular fibrillation [7], are susceptible to arrhythmogenic earlyand delayed after-depolarisations [3,11,12], are linked to torsade depointes associated with long QT syndrome [13,14] and play a role inarrhythmias after electric shock defibrillation [15].
To understand the physiological and pathophysiological functioningof the PFs, the aim of the current study was to map the anatomy ofthe His–Purkinje conduction networks in the rabbit heart and theexpression of the major cardiac ion channels responsible for theelectrical activity of the PFs. In the human, right bundle branch block isrelatively common, but may be asymptomatic, whereas left bundlebranch block is less common, but more serious, and for this reasonboth left and right His–Purkinje networks were investigated, by usingsimilar molecular mapping techniques as we used previously formapping the sinoatrial and atrioventricular nodes in the rabbitheart [16,17].
690 A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
2. Materials and methods
Experiments were conducted on 1 to 2 kg male New Zealand whiterabbits,whichwere sacrificedbyanoverdoseof pentobarbital accordingto the United KingdomAnimals (Scientific Procedures) Act, 1986. Usingwhole mount immunoenzyme-histochemistry, a marker of the cardiacconduction system, middle neurofilament (NF-M) was labelled to stainHis–Purkinje tissue. Similar results were obtained from three heartsand data from two of the hearts are shown here. A 3D computer modelof the His–Purkinje networks in one of the hearts was constructed by
Fig. 1. His–Purkinje network in the left ventricle of the rabbit. A, macroscopic image of immuventricle. Labels indicate the location of high magnification images in Fig. 2. B, outline of thehas been segmented into different parts — shown in different colours. The dashed lines mabetween the interventricular septum (centre) and free wall (left and right of centre).
image analysis and Matlab software and used in numerical simulationsto calculate action potential conduction. From a further eight rabbits,samples of free running PFs were carefully micro-dissected under adissecting microscope using fine forceps without contamination fromthe left and the right ventricular muscle. Total RNA was extracted fromthe samples and reverse transcribed and the abundance of mRNA forthe major cardiac ion channels responsible for the electrical activityof the PFs was measured using quantitative PCR (qPCR). The abun-dances of mRNAs were normalised to the abundance of a housekeeper,28S. For further details of the methods used, see the Data Supplement.
noenzyme-labelling (dark brown signal) of NF-M on the endocardial surface of the leftHis–Purkinje network (structures displaying positive immunolabelling). The networkrk the cut edge of the ventricular free wall, the papillary muscles, and the oval border
691A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
3. Results
3.1. Anatomy
Unstained, the PFs can be difficult to identify. NF-M is a positivemarker for the cardiac conduction system in the rabbit [16] and, in thewhole heart, NF-M was labelled using the immunoenzyme technique.After NF-M labelling, the PF tissue was clearly visible, because it wasstained a light brown colour (Figs. 1–3). Fig. 1A shows the His–Purkinje
Fig. 2. High magnification images of NF-M labelling of the His–Purkinje network in the rabbiimages in A–C, F and H are from the heart shown in Figs. 1 and 3. The images in D, E and G aaorta. B, midseptal LBB. C, division of the LBB prior to forming free running PFs. D, LBB and frthe left ventricle. F, RBB running around the base of the papillary muscle on the right ventricRBB runs endocardially on the septal surface crossing the ventricular chamber via the mod
network on the endocardial surface of the opened left ventricle and inFig. 1B the His–Purkinje network has been outlined and segmented intodifferent parts: the bundle branch, the free running PFs and the terminalPFs (shown in different colours); the free running PFswere divided intomultiple layers in order to ensure correct connectivity between fibres(see Data Supplement for details). The left bundle branch (LBB)emerged on the crest of the muscular ventricular septum in the aorticroot as a broad structure ~2 mmacross— it is shown in red in Fig. 1 andahighmagnification viewof it is shown in Fig. 2A. The LBBwasmadeupof
t ventricles. The locations of the high power images are indicated in Figs. 1A and 3A. There from a second heart. A, origin of LBB in the membranous septum near the root of theee running PFs. Connection to anterior papillary muscle is visible. E, free running PFs inular septum with free running fibres branching from it. G, RBB and moderator band. Theerator band. H, peripheral PF network on the RV free wall. PM, papillary muscle.
Fig. 3.His–Purkinje network in the right ventricle of the rabbit. A, macroscopic image of immunolabelling of NF-M on the endocardial surface of the right ventricle. Labels indicate thelocation of the high magnification images in Fig. 2. B, outline of the His–Purkinje network. The network has been segmented into different parts — shown in different colours. Thedashed lines mark the cut edge of the ventricular free wall, the papillary muscles, the border between the interventricular septum (left) and the free wall (right), and the tricuspidvalve.
692 A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
many individual strands, which ran in parallel over the surface of theseptum and descended towards the apex (Figs. 1 and 2B, C). Along itslength, the branch divided into three fascicles, which detached fromthe septal endocardium and traversed the ventricular chamber as freerunning PFs to reach the endocardial surface of the left ventricular free
wall — the free running PFs are shown in green in Fig. 1B. The freerunning PFs projected predominantly towards the septal and parietalpapillary muscles, around which the PFs divided and coalesced toform a complex arrangement (Figs. 2D, E). Finally, the free runningPFs contacted and travelled along the ventricular endocardium once
Fig. 4. 3D model of the His–Purkinje network. A, 3D model of the His–Purkinje networkin the left and right ventricles superimposed on an idealised geometry of the left andright ventricles. The colours show the activation sequence of the His–Purkinje network(see scale bar) and the arrows show the predicted ventricular epicardial breakthroughpoints in the two ventricles. The separation of the free running PFs in the chamber intodifferent levels was used as amodelling tool to remove erroneous connections in the 3Dmodel at points where separate PFs run past each other without making connectionwith each other. The distance between the levels is dependent on how the heart ismanipulated and is likely to have little relation to distances in the intact chamber.B, ventricular epicardial activation sequence (shown in greyscale — see scalebar) ofthe rabbit heart. The arrows show the ventricular epicardial breakthrough points inthe two ventricles. LV, left ventricle; RV, right ventricle.From Azarov et al. [19].
693A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
more — the terminal PFs (the site of action potential transmission tothe ventricular muscle) are shown in blue in Fig. 1B.
Fig. 3 shows the His–Purkinje network in the right ventricle. Theright bundle branch (RBB; shown in red in Fig. 3B) emerged beneaththe membranous part of the septum and descended as a fine thread ofaround ~200 μm, running down the septal surface towards the apex(Fig. 2F). Having reached the base of the anterior papillary muscle, itpassed along the moderator band to reach the parietal ventricularwall (Figs. 2G, H). Free running PFs (shown in green in Fig. 3B)branched from the RBB along its length connecting with terminal PFsin the parietal ventricular wall shown in blue in Fig. 3B. The terminalPFs divided and coalesced to form a network more complex than thatin the left ventricle (Fig. 3B) — the network covered the majority ofthe free wall and extended towards the tricuspid valve (Fig. 2H).
In the left ventricle, the bundle branch (red) accounted for 2.8% ofthe His–Purkinje network, the free running PFs (green) accounted for29.1% and the terminal PFs (blue) accounted for 67.3% (Fig. 1B). Inthe right ventricle, the bundle branch (red) accounted for 8.6% ofthe His–Purkinje network, the free running PFs (green) accountedfor 12.5% and the terminal PFs (blue) accounted for 78.4% (Fig. 3B).The percentages were calculated from number of pixels attributedto each category.
Interestingly, the terminal PFs (but not the bundle branches or thefree running PFs) showed striated labelling of NF-M (e.g. Fig. 2C).Tissue sections failed to reveal transmural PFs in the rabbit heart (datanot shown). Therefore, most of the His–Purkinje system in the rabbitis likely to exist on the endocardial surface and is, therefore, visibleusing the whole-mount technique.
We made a reconstruction of the ventricular conduction systemusing three rabbit hearts and there were no obvious differences.
3.2. A computer model of the His–Purkinje network in the rabbit
The outlining and segmentation of the His–Purkinje network(Figs. 1B and 3B) resulted in a pseudo two-dimensional (2D) model ofthe network in the two ventricles as shown in the Data Supplement(Figs. S1 and S2). It is pseudo-2D, because there are five layers ofHis–Purkinje tissue to ensure that the connectivity of the Purkinjefibres is correct (if two PFs cross over one another, they may or maynot be connected). The model of the His–Purkinje network in the leftand right ventricles is available as a mathematical array (with a totalof 8.8 million grid nodes) and can be obtained from the correspond-ing authors on request.
An idealised model of the ventricles (based on the dimensionsof the rabbit heart used to generate the pseudo-2D model) wasgenerated to visualise the His–Purkinje network (see Data Supple-ment for details). The pseudo-2D model of the His–Purkinje networkwas then transformed onto the endocardium of the left and rightventricles using a finite element method (see Data Supplement fordetails) and the result is shown in Fig. 4A and as a movie in the DataSupplement. A biophysically-detailed rabbit PF action potential model[18], as well as the cellular automaton model was used to predictthe activation sequence of the His–Purkinje network. The results withthe two approaches were similar and the results obtained usingthe cellular automaton model (and the pseudo-2D model) are shownin the Data Supplement (Fig. S2). Fig. 4A also shows the activationsequence in 3D. Fig. 4B is from Azarov et al. [19] and shows theactivation times (measured experimentally) on the ventricularepicardial surface of the rabbit heart and the black arrows show theventricular epicardial breakthrough points on the ventricular freewalls. The white arrows in Fig. 4A show the predicted breakthroughpoints in the left and right ventricles (the sites at which the actionpotential arrives at terminal PFs in the ventricular free wall earliest).In the experiment, in the left ventricle, the action potential breaksthrough towards the base of the heart near the junction with theright ventricle and the predicted breakthrough point is the same
(Fig. 4). In the experiment, in the right ventricle, the action potentialbreaks through at the apex and the predicted breakthrough point issimilar (Fig. 4).
3.3. qPCR
The results of the qPCR experiments are shown in Figs. 5–7. Themean expression of mRNAs for markers, ion channels and Ca2+-handling proteins (as compared to that of a housekeeper gene, 28S) inthe PFs (of the left and right ventricles) is shown; for comparison,expression is also shown in ventricular muscle from the left ventricle.The qPCR data are summarised in Table 1 and Fig. S3 in the DataSupplement — they show the mRNA expression in the PFs as apercentage of that in the ventricular muscle.
3.4. Marker genes
Atrial natriuretic peptide (ANP) is well known to be present inthe atria. It is also present in PFs [20]. ANP has been reported to bedetectable in some ventricular myocytes, but absent in most [21];immunolabelling of ANP confirmed that ANP protein was absentin rabbit ventricular muscle (data not shown). T-box 3 (Tbx3) is acardiac conduction system transcription factor (e.g. [22]), and NF-Mis known to be exclusively expressed in the cardiac conductionsystem in the rabbit [16]. Fig. 5 shows that there was significantlyhigher mRNA expression for ANP and NF-M in the PFs comparedto the ventricular muscle. However, there was no significantdifference in Tbx3 mRNA expression, although surprisingly therewas a trend for lower expression in the PFs than in the ventricularmuscle (Fig. 5).
Fig. 5. Relative abundance of mRNA for markers, connexins, Na+ channel α and β subunits, Ca2+ channels, and HCN channels. Means+SEM shown (ventricular muscle, n=8; PFs,n=7). a, significantly different (Pb0.05) from ventricular muscle; b, significantly different (Pb0.05) from left ventricular PFs; c, significantly different (Pb0.05) from rightventricular PFs (one-way ANOVA).
694 A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
3.5. Connexins
Gap junctions between myocytes are formed by the connexinfamily. The main connexins in the heart are Cx40, Cx43 and Cx45[23] and they form gap junctions with different conductances:
Cx40NCx43NCx45 [24]. Fig. 5 shows that the most abundant isoformwas middle conductance Cx43, and Cx43 mRNA was similarlyexpressed in all of the samples (Fig. 5). Expression of mRNA forthe high conductance Cx40 was significantly higher in the leftventricular PFs than in the ventricular muscle (Fig. 5). The expression
Fig. 7. Relative abundance of mRNA for Ca2+-handling proteins. Same format as Fig. 5.
696 A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
of mRNA for the low conductance Cx45 was significantly lower in thePFs than in the ventricular muscle (Fig. 5). The connexin expressionpattern in the PFs is consistent with the high conduction velocityof the PFs.
Table 1Mean expression of genes in left and right ventricular PFs versus ventricular muscle.Data are expressed as a percentage.
Rapid depolarisation during the upstroke (phase 0) of the actionpotential is due to the Na+ current, INa [25]. Nav1.5 is the major Na+
channel responsible for INa, but Nav1.1 also makes a contribution [26].Consistent with this, the expression of Nav1.5 mRNA was N10–100fold greater than the expression of Nav1.1 mRNA (Fig. 5). Nav1.5 washighly expressed in all of the tissues, but expression in the leftventricular PFs tended to be higher than expression in the ventricularmuscle and it was significantly higher than in the right ventricularPFs (Fig. 5). Navβ1 is a major β subunit for the Na+ channel and theexpression pattern of Navβ1 mRNA was the same as that of Nav1.5mRNA; however, the expression of Navβ1 mRNA expression in theleft ventricular PFs was higher as compared to both the ventricularmuscle and the right ventricular PFs (Fig. 5). Nav1.1 mRNA expressionwas significantly higher in the PFs than in the ventricular muscle.In summary, the expression of Na+ channels in the left ventricularPFs, at least, is consistent with the high upstroke velocity and highconduction velocity of the PFs.
The L-type Ca2+ current, ICa,L, is primarily formed by Cav1.2.However, in nodal tissues, Cav1.2 is generally more poorly expressedthan in the working myocardium and an alternative isoform, Cav1.3,is expressed [27,28]. The expression pattern in PFs was similarto that in nodal tissues: Cav1.2 mRNA expression was significantlylower in the PFs than in the ventricular muscle and Cav1.3 mRNAexpression was significantly higher in the left ventricular PFs than inthe ventricular muscle (Fig. 5).
Rabbit PFs, like PFs from other species, show pacemaker activityand they possess the funny current, If, which is known to beresponsible for pacemaking [29,30]. HCN channels are known to beresponsible for If [25,31]. mRNA levels for both HCN1 and HCN4isoforms were significantly higher in the PFs than in the ventricularmuscle (Fig. 5). HCN1 was the more abundant isoform (Fig. 5).
3.7. K+ channels
Kv1.4, Kv4.2 and Kv4.3 channels are responsible for the transientoutward current, Ito, which is responsible for the early phase ofrepolarisation (phase 1) during the actionpotential [25]. The expression
Fig. 8. Relating mRNA expression to function. A, current densities in PFs — comparisonof the best estimates from voltage clamp experiments with the predictions from ionchannel mRNA expression levels. Black bars, current density in rabbit left ventricularPurkinje cells (as a fraction of current density in rabbit left ventricular endocardialmyocytes; based on Aslanidi et al. [18]). The peak INa (Na+ current) density wasmeasured during a voltage clamp pulse to −25 mV from a holding potential of−75 mV. The peak ICa,L (L-type Ca2+ current) density was measured during a pulseto +10 mV from a holding potential −40 mV. The IK,r (rapid delayed rectifier K+
current) tail current density was measured at the holding potential of −50 mV after a300 ms pulse to +20 mV. The IK,s (slow delayed rectifier K+ current) tail currentdensity was measured at the holding potential of−50 mV after a 1 s pulse to +20 mV.The IK,1 (background inward rectifier K+ current) density was measured as the positivepeak value at −70 mV. Grey bars, expression level of the ion channel or ion channelsresponsible for each of the ionic currents in rabbit left ventricular PFs (as a fractionof the expression level in rabbit left ventricular muscle). See Table S2 in the DataSupplement for further details. B, computed action potentials. Black trace, actionpotential computed for a rabbit ventricular endocardial cell using the model of Aslanidiet al. [18]. Grey trace, action potential computed using a modified endocardial cellmodel by incorporating changes of some ionic current conductances based on the ratioof corresponding ion channel mRNAs between the PFs and ventricular muscle as shownin Table S2. The computed action potential resembles that of a PF cell. Black dashedtrace, action potential computed from the rabbit PF cell model of Aslanidi et al. [18].
697A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
levels of the mRNAs for the three channels were comparable (b10 folddifferences; Fig. 6). The expression of Kv4.2 mRNA was similar in alltissues (Fig. 6). Whereas Kv1.4 mRNA was significantly more highlyexpressed in ventricular muscle than in the PFs, Kv4.3 mRNA wassignificantly more highly expressed in the left ventricular PFs than inthe ventricular muscle (and also in the right ventricular PFs; Fig. 6).This suggests a switch from a slowly recovering isoform (Kv1.4) to afast recovering one (Kv4.3) in the left ventricular PFs. KChIP2 andDPP6 (dipeptidyl aminopeptidase-like protein 6) are β subunits forKv4 channels [32,33] and, interestingly, whereas KChIP2 mRNA wassignificantly more highly expressed in the ventricular muscle thanin the PFs, DPP6 mRNA was significantly more highly expressed in thePFs than in the ventricular muscle (Fig. 6).
The plateau phase (phase 2) and the final phase of repolarisation(phase 3) of the action potential are shaped by the ultra rapid (IK,ur),rapid (IK,r) and slow (IK,s) delayed rectifier K+ currents formed byKv1.5, ERG and KvLQT1, respectively [25]. The expression of mRNA forall three K+ channels tended to be lower in the PFs than in theventricular muscle (expression of Kv1.5 and KvLQT1 mRNAs wassignificantly lower in both left and right ventricular PFs, whereasexpression of ERG mRNA was only significantly lower in the rightventricular PFs; Fig. 6). minK is a β subunit for KvLQT1 and contributesto IK,s [25]. Unlike KvLQT1 mRNA, expression of minK mRNA wassignificantly higher than in the left ventricular PFs than in theventricular muscle (and also the right ventricular PFs; Fig. 6).
The Kir2 family of K+ channels (Kir2.1–2.4) generates thebackground inward rectifier K+ current, IK,1, responsible for theresting potential [25]. All members showed a tendency for lowermRNA expression in the PFs versus the ventricular muscle (Fig. 6). TheKir2.1 isoform (at the mRNA level) was the most abundant isoform(by 10 fold or more) in most tissues and expression of Kir2.1 mRNAwas significantly higher in the ventricular muscle than in the PFs.Rabbit PFs are controlled by the parasympathetic neurotransmitter,ACh [29,30]. The ACh-activated K+ current (IK,ACh) is in part generatedby Kir3.1 [25] and expression of Kir3.1 mRNA was significantlyhigher (N20 fold) in the PFs than in the ventricular muscle (Fig. 6).The ATP-sensitive K+ current (IK,ATP) is generated in part by Kir6.2and SUR2 [14] and both were distributed in the same way as the Kir2channels: there was significantly lower expression of the two mRNAsin the PFs than in the ventricular muscle (Fig. 6).
TWIK1 is a weak inwardly rectifying twin-pore K+ channel andlike the Kir2 channels has been suggested to be involved in thegeneration of the resting potential, and there was a significantlyhigher expression of TWIK1 mRNA in the PFs than in the ventricularmuscle (Fig. 6).
3.8. Ca2+-handling proteins
Four Ca2+-handling proteins were investigated: the main and asubsidiary sarcoplasmic reticulum Ca2+ release channel (RYR2 andRYR3, respectively), the sarcoplasmic reticulum Ca2+ pump (SERCA2a)and the Na+–Ca2+ exchanger (NCX1). PFs are known to be poorlycontractile [34] and, therefore, it is perhaps not surprising that theexpression of mRNAs for all four Ca2+-handling proteins wassignificantly less in the PFs than in the ventricular muscle (Fig. 7).
3.9. Relating mRNA expression to function
If it is assumed that, generally, there is a reasonable correlationbetween ion channel mRNA expression and ion channel density (orwhole cell conductance for an ionic current), the expression level ofion channel mRNAs can be used to predict ionic current densities —
we have previously used this approach to predict feasible actionpotentials for the human sinoatrial and atrioventricular nodes (e.g.[35]). The expression of mRNAs for Nav1.1, Nav1.5, Cav1.2, Cav1.3, ERG,KvLQT1, Kir2.1, Kir2.2, Kir2.3 and Kir2.4 in left ventricular PFs (as a
fraction of that in ventricular muscle from the left ventricle) was usedto predict the ionic conductances for INa, ICa,L, IK,r, IK,s and IK,1 in leftventricular PFs (as a fraction of that in ventricular muscle from the leftventricle); see Table S2 in the Data Supplement for details. Theconductance for Ito could not be predicted in the same way, because itis unclear which subunit (Kv1.4, Kv4.2, Kv4.3, KChIP2 for example) isthe dominant isoform determining the maximal channel conductanceof Ito. Whereas the conductance for INa in the left ventricular PFs ispredicted to be higher than in the ventricular muscle, the otherconductances are predicted to be lower (Fig. 8A). These predictionswere tested by using our recently developed biophysically-detailedaction potential models for rabbit PFs and ventricular muscle,which were based on voltage clamp data from rabbit Purkinje and
698 A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
ventricular myocytes [18]. Fig. 8A also shows the ionic conductancesfor the various ionic currents in left ventricular PFs (as a fraction ofthat in ventricular muscle from the left ventricle) from the models.There is generally good agreement between the conductancespredicted from ion channel mRNA expression and conductancesbased on experimental data. A further test was performed to seewhether or not one can turn a ventricular cell model to a PF cell-likecell model based on the measured fractions of ion channel mRNAexpression, by assuming the same fraction for each corresponding ionchannel conductance. Fig. 8B shows left ventricular PF and ventricularendocardial action potentials simulated by the Aslanidi et al. models[18]. Fig. 8B also shows the action potential simulated by theleft ventricular endocardial model when the ionic conductances forINa, ICa,L, IK,r, IK,s and IK were modified based on ion channel mRNAexpression in the left ventricular PFs (in addition, one ionic current,the late Na+ current, for which there is no knownmolecular correlate,was added). The resulting action potential is qualitatively similar tothe action potential predicted by the left ventricular PF model basedexclusively on experimental data.
4. Discussion
We have constructed a detailed 3D model of the anatomy of theHis–Purkinje networks in the rabbit — the model shows thecomplexity of the system, shows the His–Purkinje networks to beasymmetrical in the left and right ventricles, and predicts theventricular epicardial break-through points in the rabbit heartmeasured experimentally. We have also shown that the PFs displaya distinct expression profile of ion channels at the mRNA level, whichin part explains their distinct action potential profile and theirspecialisation for rapid conduction.
4.1. Anatomy
The work of Tawara greatly enhanced our knowledge of thecardiac conduction system and in particular the ventricular compo-nent consisting of the His bundle and the PF networks [2]. We havepreviously shown that NF-M is a positive marker of the cardiacconduction system in the rabbit [16,17] and we have now shown thatit is possible to use whole mount immunohistochemical labelling ofNF-M to label the His–Purkinje networks, the ventricular componentsof the cardiac conduction system (Figs. 1–3). For the first time, thisstudy has shown that the bundle branches have a ribbon cable-likestructure with a number of individual strands running in parallel(Fig. 2B). The His–Purkinje system initially projects predominantlytowards the papillary muscles and undergoes numerous divisionsto form networks around the papillary muscle as well as theendocardium of the free wall. The peripheral networks cover largeportions of the ventricular free walls. The labelling of the His–Purkinjesystem was used to construct a 3D model of the His–Purkinje systemand this predicted correctly the points of ventricular epicardialbreakthrough (Fig. 4). The overall network can be divided into thebundle branches, the free running strands and the terminalramifications attached to the ventricular endocardium. Why shouldthe strands, or false tendons, be free running? One advantage is thattransmission of the action potential to the ventricular muscle occursat appropriate sites in order to obtain the correct activation sequenceof the ventricular muscle. However, another advantage is that it mayallow faster conduction of the action potential to the transmissionsites. This hypothesis was tested out by simulating the propagation ofthe action potential along a single strand of the specialised myocytescoupled together with a diffusion coefficient (proportional to gapjunction coupling conductance) of 0.42 mm2/ms, chosen to producea longitudinal conduction velocity similar to that expected of PFmyocytes (~1.4 m/s). This strand was then coupled on either side toa 2D layer of ventricular endocardial muscle as illustrated in the
Data Supplement (Fig. S4). Action potentials were simulated usingbiophysically-detailed models of the action potential of rabbit PFsand endocardial ventricular muscle [18]. The ventricular muscleacted as a ‘load’ on the PFs and it slowed the conduction velocity ofthe action potential along the strand of these specialised myocytes by43% (from ~1.4 to 0.8 m/s; Fig. S4; see Data Supplement for furtherdetails of the simulation).
Although we have developed an anatomical model of the His–Purkinje system (Fig. 4), it does not identify Purkinje-ventricularjunctions and these remain to be defined (Fig. 4 shows activationtimes for the PFs only). Only the terminal PFs are expected to havejunctions with the ventricular muscle and, as shown in Fig. 2C, theterminal PFs, but not the bundle branches or the free running PFs,showed striated labelling of NF-M — could these represent Purkinje-ventricular junctions?
4.2. ANP, Tbx3 and connexins
ANP mRNA was more abundant in the PFs than in ventricularmuscle. It has been suggested that ANP could play an autocrine/paracrine role in PFs and may be linked to the regulation of actionpotential conduction [36,37]. The Tbx3 transcription factor tendedto be more poorly expressed in the PFs than in the ventricularmuscle (Fig. 5). Tbx3 is known to be highly expressed in thesinoatrial node of many species (e.g. [35]) and is important in theregulation of pacemaker gene expression and phenotype [38]. Thereduced levels could be due to the fact that the PFs are the terminalportion of the cardiac conduction system and there may be agradient in expression of Tbx3 down the cardiac conduction system.It has been suggested that the PFs differentiate from ventricularmyocytes via recruitment and differentiation under the control oftranscription factors such as HF1-b and Nkx2.5 [39,40], rather thanpre-existing conduction cells under the control of Tbx3 as in the caseof the sinoatrial node.
Connexin expression in the PFs is consistentwith the specialisationof the PFs for rapid conduction: the high conductance Cx40 was morehighly expressed in the left ventricular PFs than in the ventricularmuscle, the medium conductance Cx43 was highly expressed in alltissues and the low conductance Cx45 was more poorly expressed inthe PFs than in the ventricular muscle (Fig. 5). In Cx40 knockout mice,the conduction velocity of the bundle branches is significantly slowerand right bundle branch block is observed [41]. A high expressionof Cx40 mRNA has been reported in human [42] and mouse [43] PFs.
4.3. Channels carrying inward ionic currents
Nav1.5 and Navβ1 mRNA expression was higher in the leftventricular PFs than in the ventricular muscle (Fig. 5) and this canhelp explain the higher upstroke velocity and conduction velocity ofPFs as compared to ventricular muscle. Nav1.1 mRNA expression washigher in the PFs than in the ventricular muscle (Fig. 5). However,in the PFs, the abundance of Nav1.5 cDNA was N10× more abundantthan Nav1.1 cDNA (Fig. 5) and, therefore, Nav1.5 is likely to be thedominant isoform determining INa. The PFs showed decreased Cav1.2and increased Cav1.3 mRNA expression compared to the ventricularmuscle (Fig. 5). The same isoform switch (between the cardiacconduction system and working myocardium) has been observed atrabbit and human sinoatrial and atrioventricular nodes [27,28,35]and in human PFs [42]. Cav1.3 channels are more appropriate forpacemaking than Cav1.2 channels and Cav1.2 is associated withintracellular Ca2+-handling [42] and, therefore, it is not unexpectedthat there is an isoform switch in the cardiac conduction system,with pacemaker activity and limited contractility. Overall, Cav1channel expression, responsible for ICa,L, was lower in the PFs than inthe ventricular muscle (Fig. 8A) and this is consistent with a lowerdensity of ICa,L in PFs as compared to ventricular muscle (Fig. 8A).
699A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
Both HCN1 and HCN4 mRNAs, responsible for If, were more highlyexpressed in the PFs than in the ventricular muscle (Fig. 5). Thepresence of the channels is consistent with the pacemaker activity ofPFs in the rabbit [29]. An If-like current has been reported in rabbit PFs[30]. HCN channels and pacemaker activity in PFs have been observedin other species [42,44,45]. The pacemaker activity in PFs is maskedunder normal conditions as a result of overdrive suppression bythe primary pacemaker, the sinoatrial node [46], but in heart block thePFs have the potential to provide a ventricular escape rhythm. Thisproperty also makes them a candidate for the creation of ventriculararrhythmias.
4.4. K+ channels
In the PFs, there was a lower expression of Kv1.4 and KChIP2mRNAs and increased expression of Kv4.3 mRNA (Fig. 6). Similarchanges in Kv1.4, KChIP2 and Kv4 mRNAs have been observed at thesinoatrial and atrioventricular nodes in the rabbit [27,28] and Gaboritet al. [42], reported that KChIP2 mRNA is more poorly expressedand Kv4.3 mRNA is more highly expressed in PFs than in ventricularmuscle in the human. Ito in PFs has been shown to be functionallydifferent from Ito in the ventricular muscle [47,48].
The delayed rectifier K+ currents, IK,ur, IK,r and IK,s, are largelyresponsible for repolarisation during the plateau phase. Expressionof the K+ channels responsible (Kv1.5, ERG and KvLQT1) wassignificantly lower in the PFs than in the ventricular muscle (Fig. 6).This is consistent with a lower density of IK,r and IK,s (at least) in PFs ascompared to ventricular muscle (Fig. 8A). PFs have a longer actionpotential than ventricular muscle [14] (Fig. 8B) and the poorexpression of Kv1.5, ERG and KvLQT1 must be in part responsible forthis. When IK,r and IK,s are selectively blocked, there is an increase ofaction potential duration in rabbit PFs [49], confirming that bothcurrents play an important role in determining action potentialduration. All the K+ channels that contribute to IK,1 and are involvedin the generation of the resting potential (Kir2.1–2.4) tend to bepoorly expressed in the PFs (Fig. 6). This is consistent with a lowerdensity of IK,1 in PFs as compared to ventricular muscle (Fig. 8A). Poorexpression of Kir2.1 (at least) and a low density of IK,1 are commonto all tissues of the cardiac conduction system, because it permitspacemaking [28,35,50].
Kir3.1, in part responsible for IK,ACh, was more highly expressed inthe PFs than in the ventricular muscle (Fig. 6). Consistent with this,ventricularmuscle is generally poorly sensitive to the parasympatheticneurotransmitter, ACh, whereas PFs are sensitive [29,30].
In the rabbit, the lower expression of Kir6.2 and SUR2, responsiblefor IK,ATP, in PFs as compared to ventricularmuscle (Fig. 6) is consistentwith what is seen in the human [42]; similar expression of the otherKir channels is also observed in human PFs [42].
4.5. Intracellular Ca2+-handling
All Ca2+-handling proteins investigated were more poorlyexpressed at the mRNA level in the PFs than in the ventricular muscle(Fig. 7) and this is consistent with their poor contractility. A similarexpression pattern of RYR2, SERCA2 and NCX1 mRNAs has beenobserved in human PFs [42].
4.6. Left ventricular PFs versus right ventricular PFs
The initial part of the left bundle branch is extensive compared tothe right bundle branch (Figs. 1 and 3). This may be consistent withthe higher prevalence of right bundle branch block in humans. Theconsequences of left bundle branch block are more serious than rightbundle branch block and this may be a reflection of the greater sizeof the left bundle branch. The asymmetry in the size of the right andleft bundle branches has been noted before in the mouse and human
[3]. Perhaps the left ventricular His–Purkinje network may be moredeveloped, because the left ventricle generates higher pressures andis, therefore, more susceptible to improper mechanical activation. Inaddition, the right ventricle is a later evolutionary feature than theleft ventricle and this may be another reason why its His–Purkinjenetwork is less well developed.
There were significant differences in mRNA expression for anumber of ion channels between the left and right ventricular PFs thatsuggest that the left ventricular PFs aremore specialised than the rightventricular PFs (Figs. 5 and 6). These findings suggest that conductionmay be faster (Cx40, Nav1.5, Navβ1) and pacemaking more vigorous(Cav1.3, possibly HCN1, HCN4) in the left ventricular PFs than inthe right ventricular PFs and this may also be consistent with theprevalence and clinical importance of left and right bundle branchblock in the human. In our study, there were also differences in theexpression of ERG, minK, SUR2A and NCX1 mRNAs between theleft and right ventricular PFs (Figs. 6 and 7). ERG and minK mRNAexpression was significantly higher in the left ventricular PFs (than inthe right ventricular PFs)— KvLQT1mRNAwas distributed in a similarmanner, although the difference between the left and right ventric-ular PFs did not reach significance (Fig. 6). The higher expression ofERG, possibly KvLQT1 and minK mRNAs could explain the shorterduration of the left ventricular PF action potential reported in twoprevious studies [8,51]. However, the higher expression of Nav1.5and Navβ1 mRNAs in the left ventricular PFs are expected to mitigatethe difference in action potential duration.
4.7. Comparison of rabbit, human and dog PF gene expression
Previously Gaborit et al. [42] and Han et al. [45] studied geneexpression in the PFs of human and dog hearts and Table S3 (in theData Supplement) compares the expression of gap junction channels,ion channels and Ca2+-handling proteins in the PFs of the rabbit (datafrom this study), human and dog. Comparison of the three speciesshowsmany similarities (although there are also species-differences).For example, in the PFs of all species (if measured), Cx40, Cav1.3,HCN1, HCN4, Kv4.3, Kir3.1 and TWIK1 were more highly expressedand Cav1.2, KChIP2, KvLQT1, Kir2.1–Kir2.3, Kir6.2, RYR2, NCX1 andSERCA2Aweremore poorly expressed (as compared to the ventricularmuscle; see the Data Supplement for more details).
4.8. Study limitations
The study has a number of limitations: (i) in terms of ion channelexpression, we only analysed free running PFs, although ion channelexpression may vary from the bundle branches to the terminal PFs(action potential duration declines from the bundle branches to theterminal PFs [51]). (ii) We were unable to measure the expressionlevels of some important ion channels such as Cav3 channels, KCNE3and Kir3.4, because there is no rabbit sequence available for thesetargets. We did attempt to measure Cav3.1 mRNA expression, but itwas undetectable and perhaps Cav3.2 or Cav3.3 is responsible for ICa,Tin the rabbit. (iii) In this study, we did not measure the expression ofion channels at the protein level (partly because commerciallyavailable ion channel antibodies are often raised in rabbit and are,therefore, not suitable for this species). However, our data on thecardiac conduction system have shown that generally proteinexpression reflects mRNA expression (e.g. [35]). These limitationswill have to be addressed in future studies. Another potentiallimitation is contamination of one tissue by another. However, byworking with free running PFs, contamination should not be possible.Furthermore, NF-M (specific marker of cardiac conduction system inrabbit) was not detected in the ventricular tissue, whereas Kir2.1(marker of ventricular tissue) was highly abundant in the ventriculartissue but not in the PFs (Figs. 5 and 6).
700 A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
Disclosures
None.
Acknowledgments
This study was supported by the British Heart Foundation.
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.1016/j.yjmcc.2011.05.020.
References
[1] Silverman ME, Grove D, Upshaw Jr CB. Why does the heart beat? The discoveryof the electrical system of the heart. Circulation 2006;113:2775–81.
[2] Tawara S. The conduction system of the mammalian heart. London: ImperialCollege Press; 2000.
[3] Miquerol L, Meysen S, MangoniM, Bois P, van Rijen HV, Abran P, et al. Architecturaland functional asymmetry of the His–Purkinje system of the murine heart.Cardiovasc Res 2004;63:77–86.
[4] Tranum-Jensen J, Wilde AA, Vermeulen JT, Janse MJ. Morphology of electrophys-iologically identified junctions between Purkinje fibers and ventricular muscle inrabbit and pig hearts. Circ Res 1991;69:429–37.
[5] Oosthoek PW, Virágh S, Lamers WH, Moorman AF. Immunohistochemicaldelineation of the conduction system. II: the atrioventricular node and Purkinjefibers. Circ Res 1993;73:482–91.
[6] Ansari A, Ho SY, Anderson RH. Distribution of the Purkinje fibres in the sheepheart. Anat Rec 1999;254:92–7.
[7] Dosdall DJ, Tabereaux PB, Kim JJ, Walcott GP, Rogers JM, Killingsworth CR, et al.Chemical ablation of the Purkinje system causes early termination and activationrate slowing of long-duration ventricular fibrillation in dogs. Am J Physiol 2008;295:H883–9.
[8] Anumonwo JM, Tallini YN, Vetter FJ, Jalife J. Action potential characteristics andarrhythmogenic properties of the cardiac conduction system of the murine heart.Circ Res 2001;89:329–35.
[9] Callewaert G, Carmeliet E, Vereecke J. Single cardiac Purkinje cells: generalelectrophysiology and voltage-clamp analysis of the pace-maker current. J Physiol1984;349:643–61.
[10] Wit AL, Cranefield PF. Reentrant excitation as a cause of cardiac arrhythmias. Am JPhysiol 1978;235:H1–17.
[11] Cordeiro JM, Bridge JH, Spitzer KW. Early and delayed afterdepolarizations inrabbit heart Purkinje cells viewed by confocal microscopy. Cell Calcium 2001;29:289–97.
[12] Dumaine R, Cordeiro JM. Comparison of K+ currents in cardiac Purkinje cellsisolated from rabbit and dog. J Mol Cell Cardiol 2007;42:378–89.
[13] Asano Y, Davidenko JM, Baxter WT, Gray RA, Jalife J. Optical mapping of drug-induced polymorphic arrhythmias and torsade de pointes in the isolated rabbitheart. J Am Coll Cardiol 1997;29:831–42.
[14] Dun W, Boyden PA. The Purkinje cell; 2008 style. J Mol Cell Cardiol 2008;45:617–24.
[15] Li HG, Jones DL, Yee R, Klein GJ. Defibrillation shocks produce different effects onPurkinje fibers and ventricular muscle: implications for successful defibrillation,refibrillation and postshock arrhythmia. J Am Coll Cardiol 1993;22:607–14.
[16] Dobrzynski H, Li J, Tellez J, Greener ID, Nikolski VP, Wright SE, et al. Computerthree-dimensional reconstruction of the sinoatrial node. Circulation 2005;111:846–54.
[17] Li J, Greener ID, Inada S, Nikolski VP, Yamamoto M, Hancox JC, et al. Computerthree-dimensional reconstruction of the atrioventricular node. Circ Res 2008;102:975–85.
[18] Aslanidi OV, Sleiman RN, Boyett MR, Hancox JC, Zhang H. Ionic mechanisms forelectrical heterogeneity between rabbit Purkinje and ventricular cells and tissue.Biophys J 2010;98:2410–31.
[19] Azarov JE, Shmakov DN, Vityazev VA, Roshchevskaya IM, Roshchevsky MP.Activation and repolarization patterns in the ventricular epicardium under sinusrhythm in frog and rabbit hearts. Comp Biochem Physiol A Mol Integr Physiol2007;146:310–6.
[20] Toshimori H, Toshimori K, Oura C, Matsuo H, Matsukura S. Immunohistochemicalidentification of Purkinje fibers and transitional cells in a terminal portion of theimpulse-conducting system of porcine heart. Cell Tissue Res 1988;253:47–53.
[21] Howarth FC, Adem A, Adeghate EA, Al Ali NA, Al Bastaki AM, Sorour FR, et al.Distribution of atrial natriuretic peptide and its effects on contraction andintracellular calcium in ventricular myocytes from streptozotocin-induced diabeticrat. Peptides 2005;26:691–700.
[22] Hatcher CJ, Basson CT. Specification of the cardiac conduction system bytranscription factors. Circ Res 2009;105:620–30.
[23] Severs NJ, Rothery S, Dupont E, Coppen SR, Yeh HI, Ko YS, et al. Immunocyto-chemical analysis of connexin expression in the healthy and diseased cardiovas-cular system. Microsc Res Tech 2001;52:301–22.
[24] Desplantez T, Dupont E, Severs NJ, Weingart R. Gap junction channels and cardiacimpulse propagation. J Membr Biol 2007;18:13–28.
[25] Schram G, Pourrier M, Melnyk P, Nattel S. Differential distribution of cardiac ionchannel expression as a basis for regional specialization in electrical function. CircRes 2002;90:939–50.
[26] Lei M, Jones SA, Liu J, Lancaster MK, Fung SS, Dobrzynski H, et al. Requirementof neuronal- and cardiac-type sodium channels for murine sinoatrial nodepacemaking. J Physiol 2004;559:835–48.
[27] Tellez JO, Dobrzynski H, Greener ID, Graham GM, Laing E, Honjo H, et al.Differential expression of ion channel transcripts in atrial muscle and sinoatrialnode in rabbit. Circ Res 2006;99:1384–93.
[28] Greener ID, Tellez JO, Dobrzynski H, YamamotoM, GrahamGM, Billeter R, et al. Ionchannel transcript expression at the rabbit atrioventricular conduction axis. CircArrhythm Electrophysiol 2009;2:305–15.
[29] Mubagwa K, Carmeliet E. Effects of acetylcholine on electrophysiologicalproperties of rabbit cardiac Purkinje fibers. Circ Res 1983;53:740–51.
[30] Carmeliet E, Mubagwa K. Changes by acetylcholine of membrane currents inrabbit cardiac Purkinje fibres. J Physiol 1986;371:201–17.
[31] Santoro B, Tibbs GR. The HCN gene family: molecular basis of the hyperpolarization-activated pacemaker channels. Ann N Y Acad Sci 1999;868:741–64.
[32] Pourrier M, Schram G, Nattel S. Properties, expression and potential roles ofcardiac K+ channel accessory subunits: MinK, MiRPs, KChIP, and KChAP. J MembrBiol 2003;194:141–52.
[33] Radicke S, Cotella D, Graf EM, Ravens U, Wettwer E. Expression and function ofdipeptidyl-aminopeptidase-like protein 6 as a putative beta-subunit of humancardiac transient outward current encoded by Kv4.3. J Physiol 2005;565:751–6.
[34] Boyett MR, Hart G, Levi AJ, Roberts A. Effects of repetitive activity on developedforce and intracellular sodium in isolated sheep and dog Purkinje fibres. J Physiol1987;388:295–322.
[35] Chandler NJ, Greener ID, Tellez JO, Inada S, Musa H, Molenaar P, et al. Moleculararchitecture of the human sinus node: insights into the function of the cardiacpacemaker. Circulation 2009;119:1562–75.
[36] Anand-Srivastava MB, Thibault G, Sola C, Fon E, Ballak M, Charbonneau C, et al.Atrial natriuretic factor in Purkinje fibers of rabbit heart. Hypertension 1989;13:789–98.
[37] Hansson M, Forsgren S. Immunoreactive atrial and brain natriuretic peptides areco-localized in Purkinje fibres but not in the innervation of the bovine heartconduction system. Histochem J 1995;27:222–30.
[38] Hoogaars WM, Engel A, Brons JF, Verkerk AO, de Lange FJ, Wong LY, et al. Tbx3controls the sinoatrial node gene program and imposes pacemaker function onthe atria. Genes Dev 2007;21:1098–112.
[39] Gourdie RG, Severs NJ, Green CR, Rothery S, Germroth P, Thompson RP. The spatialdistribution and relative abundance of gap-junctional connexin40 and connexin43correlate to functional properties of components of the cardiac atrioventricularconduction system. J Cell Sci 1993;105:985–91.
[41] van Rijen HV, van Veen TA, van KempenMJ,Wilms-Schopman FJ, PotseM, KruegerO, et al. Impaired conduction in the bundle branches of mouse hearts lacking thegap junction protein connexin40. Circulation 2001;103:1591–8.
[42] Gaborit N, Le Bouter S, Szuts V, Varro A, Escande D, Nattel S, et al. Regional andtissue specific transcript signatures of ion channel genes in the non-diseasedhuman heart. J Physiol 2007;582:675–93.
[43] RemmeCA, Verkerk AO, HoogaarsWM,AanhaanenWT, Scicluna BP, Annink C, et al.The cardiac sodium channel displays differential distribution in the conductionsystem and transmural heterogeneity in themurine ventricularmyocardium. BasicRes Cardiol 2009;104:511–22.
[44] ShiW,Wymore R, Yu H,Wu J,Wymore RT, Pan Z, et al. Distribution and prevalenceof hyperpolarization-activated cation channel (HCN) mRNA expression in cardiactissues. Circ Res 1999;85:1–6.
[45] HanW, BaoW,Wang Z, Nattel S. Comparison of ion-channel subunit expression incanine cardiac Purkinje fibers and ventricular muscle. Circ Res 2002;91:790–7.
[46] Boyett MR, Fedida D. Changes in the electrical activity of dog cardiac Purkinjefibres at high heart rates. J Physiol 1984;350:361–91.
[47] Han W, Zhang L, Schram G, Nattel S. Properties of potassium currents in Purkinjecells of failing humanhearts. Am J Physiol Heart Circ Physiol 2002;283:H2495–503.
[48] Han W, Wang Z, Nattel S. A comparison of transient outward currents in caninecardiac Purkinje cells and ventricular myocytes. Am J Physiol Heart Circ Physiol2000;279:H466–74.
[50] Dobrzynski H, Boyett MR, Anderson RH. New insights into pacemaker activity:promoting understanding of sick sinus syndrome. Circulation 2007;115:1921–32.
[51] Myerburg RJ, Stewart JW, Hoffman BF. Electrophysiological properties of thecanine peripheral A–V conducting system. Circ Res 1970;26:361–78.
Glossary
ANP: Atrial natriuretic peptideCx40, Cx43, Cx45: Connexins 40, 43 and 45Cav1.2, Cav1.3: Ca2+ channels responsible for the L-type Ca2+ current, ICa,L3D: Three-dimensional2D: Two-dimensionalDPP6: β subunit for Kv4 channelsERG: Ion channel responsible for rapid delayed rectifier K+ current, IK,rHCN1, HCN4: Ion channels responsible for hyperpolarization-activated (‘funny’)
current, If
701A. Atkinson et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 689–701
HF1-b: Transcription factorKChIP2: β subunit for Kv4 channelsKir2.1–Kir2.4: Ion channels responsible for background inward rectifier K+ current, IK,1Kir3.1: Ion channel responsible for the acetylcholine-activated K+ current, IK,AChKir6.2, SUR2: Ion channels responsible for the ATP-sensitive K+ current, IK,ATPKv1.4, Kv4.2, Kv4.3: Ion channels responsible for the transient outward K+ current, ItoKv1.5: Ion channel responsible for the ultra-rapid delayed rectifier K+ current, IK,urKvLQT: Ion channel responsible for the slow delayed rectifier K+ current, IK,sLBB: Left bundle branchLV: Left ventricleminK: Accessory protein for KvLQT1mRNA: Messenger ribonucleic acidNav1.1, Nav1.5, Navβ1:Na+ channel subunits responsible for the inward Na+ current, INaNCX1: Na+–Ca2+ exchanger
