Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

Contents lists available at ScienceDirect

Journal of Molecular and Cellular Cardiology

j ourna l homepage: www.e lsev ie r.com/ locate /y jmcc

Original article

Distribution of the pacemaker HCN4 channel mRNA and protein in the rabbitsinoatrial node

Chiara Brioschi a,1, Stefano Micheloni a,1, James O. Tellez b, Giuliano Pisoni c, Renato Longhi d,Paolo Moroni c, Rudi Billeter e, Andrea Barbuti a,f, Halina Dobrzynski b, Mark R. Boyett b,Dario DiFrancesco a,f, Mirko Baruscotti a,f,⁎a University of Milano, Department of Biomolecular Sciences and Biotechnology, Laboratory of Molecular Physiology and Neurobiology, via Celoria 26, 20133 Milano, Italyb Cardiovascular Research Group, Faculty of Medical and Human Sciences, University of Manchester, 46 Grafton Street, M13 9NT, UKc University of Milano, Dept. of Animal Pathology, Hygiene and Veterinary Public Health, via Celoria 20, 20133 Milano, Italyd National Research Center of Milano, Institute of Chemistry of Molecular Recognition, via M. Bianco 9, 20131 Milano, Italye University of Nottingham, School of Biomedical Sciences, Queen's Medical Centre, Nottingham NG7 2UH, UKf Centro Interuniversitario di Medicina Molecolare e Biofisica Applicata (CIMMBA), University of Milano, Italy

⁎ Corresponding author. Department of BiomoleculaLaboratory of Molecular Physiology and NeurobiologyMedicina Molecolare e Biofisica Applicata (CIMMBA), U26, 20133 Milano, Italy. Tel.: +39 02 5031 4931; fax: +

E-mail address: mirko.baruscotti@unimi.it (M. Barus1 These authors have contributed equally to the work

0022-2828/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.yjmcc.2009.04.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 19 November 2008Received in revised form 20 March 2009Accepted 10 April 2009Available online 24 April 2009

Keywords:SANPacemaker channelsHCN clonesIfCx43NF-M

Several studies of the pacemaker mechanisms in mammalian cells, most of which were carried out in cellsisolated from the rabbit sinoatrial node (SAN), have highlighted the role of the If current. While thedistribution of Hyperpolarization-activated Cyclic Nucleotide-gated (HCN) channels, the molecular correlatesof f-channels, is known at the mRNA level, the identification of f-channel proteins in this tissue is stillundetermined. Here we investigate HCN protein expression in the rabbit pacemaker region. We found thatHCN4 is the main isoform, and set therefore to analyze its distribution within the SAN and surrounding areaswith the aim of correlating protein expression and pacemaking function. The analysis was carried out intissue slices and single cells of the intercaval area, which includes the crista terminalis (CT), the SAN, and theseptum interatrialis (SI). Immunolabeling, in situ hybridization, qRT-PCR analysis, and electrophysiologicalrecordings identified the SAN as a region characterized by high HCN4 signal and current levels, while theexpression in the CT and in the SI was either negligible or absent. Detailed analysis of the central SAN areashowed that cells are predominantly distributed in islets interconnected by cell prolongations, and single-cellHCN4 labeling suggested sites of channel clustering. Our data indicate that in the rabbit SAN, HCN4 proteinsare major constituents of native f-channels, and their distribution matches closely the SAN as definedmorphologically and electrophysiologically. Until recently, the SAN was identified as the region where Cx43and atrial natriuretic peptide are not expressed; we propose here that expression of HCN4 is an appropriatetool to map and identify the cardiac SAN pacemaker region.

© 2009 Elsevier Inc. All rights reserved.

1. Introduction

In mammals, the pacemaker region of the heart (SAN) is located inthe posterior wall of the right atrium; a coarse anatomical definitionidentifies this structure as the intercaval region adjacent to the atrialmuscle of the CT extending from the superior to near the inferior venacava (SVC, IVC). The histological investigation reveals a heterogeneousarchitecture with myocytes sparsely distributed in a dense connectivematrix [1,2]. The identification of the SAN as the leading pacemakerzone of the rabbit heart was originally accomplished by electro-

r Sciences and Biotechnology,, Centro Interuniversitario diniversity of Milano, via Celoria39 02 5031 4932.cotti)..

ll rights reserved.

physiological mapping studies and later confirmed in rabbit and otherspecies. Since the repertoire of transcribed mRNAs and expressedproteins changes in different types of myocytes, in more recent yearscriteria based on molecular properties have also been extensivelyemployed to identify SAN cells. For example SAN cells, unlike adjacentatrial working myocytes, are rich in connexin 45 (Cx45) and (in therabbit) neurofilament M (NF-M), but lack connexin 43 (Cx43) andatrial natriuretic peptide (ANP). The specific pattern of mRNA/proteinexpression can be used therefore as a cellular “footprint” of the SANarea [3].

The electrical oscillatory property of SAN cells reflects the lack of astable resting potential and is associated with the slow diastolicdepolarization of the action potential. Among the various ion channelsand transporters contributing to spontaneous activity, the “funny” (f-)channels play a key role in the generation of the early fraction of thediastolic depolarization and in the autonomic modulation of cardiacrhythm, achieved through the control of the diastolic depolarization

Fig. 2. Immunolabeling of rabbit SAN cell proteins. (A) Bright field images of single cellsisolated from the SAN are shown either alone (bottom) or after merging with video-confocal sections of the same cells immunolabeledwith antibodies against HCN4, HCN1,and HCN2 proteins, as indicated (top). All cells investigated resulted positive to HCN4staining (left) and negative to HCN2 staining (right); labeling of HCN1 yielded nodetectable signal in most experiments (centre right), although in some cases a faintsignal could be observed (centre left). Calibration bar: 30 μm. (B) Typical staining of twocells isolated from the SAN and immunolabeled with antibodies against the HCN4protein. The images represent a single video-confocal section scanning, and the arrowsindicate membrane patches with high signal level (hot spots). Nuclei labeled with DAPI(blue). Calibration bar: 10 μm. (C) Bright field images of HEK293 cells transfected withHCN4 (left), HCN1 (centre), and HCN2 (right) are shown in bottom panels; thecorresponding video-confocal images are shown at the top. For all antibodies a strongmembrane-associated fluorescence could be detected. When primary antibodies wereomitted no signal was detected (data not shown). Calibration bar: 10 μm.

222 C. Brioschi et al. / Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

rate [4]. Importantly, it has recently been confirmed that thepacemaker current is present in the human SAN [5].

Crystallographic and electrophysiological investigation of themolecular structure of HCN channels indicates that they assemble astetramers [6]. Extensive work has addressed the issue of the isoformcomposition of native f-channels, and both homomeric and hetero-meric assemblies of the four isoforms known have been demonstratedin vitro and/or in vivo [7–10], but conclusive evidence on theidentification of HCN isoforms contributing to f-channels are stillmissing.

mRNA signals and the corresponding protein expression are wellcorrelated in the mouse SAN [11,12]; however, while electrophysio-logical, morphological, structural, and histochemical data on SANtissue are largely based on experiments in the rabbit heart, no directmeasurements of HCN protein in the rabbit SAN are yet available.

Several data indicate that HCN4 is the major HCN subunitcomposing f-channels, including evidence for the high expression ofthe mRNA of HCN4 relative to other HCN isoforms in nodal tissue[11,13–16] and for the contribution of HCN4 to kinetic and ionicproperties of native f-channels in the SAN [8], as well as recentevidence correlating mutations of HCN4 protein in humans to sinusbradycardia or more complex cardiac pathologies involving distur-bances of rhythm [17].

We present here a set of experiments aimed to identify thedistribution of the HCN4 protein in the rabbit SAN region. Our resultsindicate that the HCN4 isoform is abundantly expressed in nodal cells,but is virtually absent in surrounding atrial myocytes. Thus, HCN4 canbe considered as a specific molecular marker of the nodal pacemakercells.

2. Materials and methods

Themethods used to dissect the SAN region from rabbit hearts andisolate single myocytes were as previously published [18]. Theseprocedures conformed with the guidelines for the care and use oflaboratory animals as established by State (D.L. 116/1992) andEuropean directives (86/609/CEE). Antibodies used for HCN detectioninwestern blot analysis and single cell or tissue section immunolabel-ing were either purified in our laboratories (HCN4) or commerciallyavailable (HCN1, HCN2, Alomone, Israel). In situ hybridizationdigoxigenin-labeled cRNA probes were used. qPCR was performedon cDNA obtained from reverse transcription of mRNA. If currentrecordings were carried out by patch-clamp experiments (whole-cellconfiguration) at room temperature. Further details are available inthe Supplementary material and methods.

3. Results

Western blot experiments were carried out to evaluate thepresence of the HCN1, HCN2, and HCN4 protein isoforms in the rabbit

Fig. 1. Western blot analysis of rabbit SAN proteins. Isoform specific anti-HCN4, anti-HCN 1, and anti-HCN 2 antibodies were employed on amembrane extract of SAN tissues(left lanes). A clear signal at the expected molecular size (∼160 kDa) was observed onlywith anti-HCN4 antibodies. Similar experiment performed with protein extracts fromHCN4-, HCN1-, and HCN2-transfected HEK293 cells (right lanes) confirmed thespecificity of the antibodies employed.

SAN tissue. Analysis performed on membrane extracts with isoformspecific antibodies led to the identification of a positive signal for theHCN4 protein, while no signal was ever detected for either HCN1 andHCN2 (Fig. 1, left lanes). Antibodies against HCN4 identified a clearband at about 160 kDa, as expected for the rabbit HCN4 protein [19],and a second fainter band at a slightly lower molecular weight whichcould be due to a de-glycosylated form of the protein. The presence ofa band at a lower than expected weight is a frequent finding in HCNwestern blot experiments [10]. Control experiments carried out inHEK293 HCN4-, HCN1-, or HCN2-transfected cells confirmed thespecificity of the antibodies employed (Fig. 1, right lanes).

We further employed immunofluorescence analysis of single nodalmyocytes to both evaluate the subcellular localization of HCN4proteins, and to verify whether this technique could reveal thepresence of HCN1 and HCN2. In Fig. 2A bright field images of singlespindle-shaped cells isolated from the SAN either alone (bottom) ormerged with the corresponding video-confocal images (superimposi-tion of several sections) obtained by immunolabeling of specific HCN

Fig. 3. Endocardial view of a typical rabbit SAN. A 9mmwide, 1 mm interdistance grid isdrawn onto the intercaval region (levels 1 through 10). Each line indicates the positionfromwhere 3 adjacent sections of 10 μm thickness were collected for investigation. SVC,IVC: Superior and Inferior Vena Cava, CT Crista Terminalis, SI Septum Interatrialis.

223C. Brioschi et al. / Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

isoforms (top) are shown. While cells were always positive for HCN4protein expression (top, left), labeling of HCN2 was never detected(top right); also, even if a faint labeling of HCN1 was seen in a fewcases, in the majority of cells investigated we could not detect anyHCN1 signal (centre, right). In Fig. 2B, images of single video-confocalsections collected in two separate cells labeled with anti HCN4antibodies are shown; it is evident that membrane labeling is notevenly distributed, but is rather organized in “hot spot” regions wherethe signal shows high intensity (arrows) and regions where the signalis fainter. Control experiments carried out in HEK293 cells transfectedwith either HCN4, HCN1 or HCN2 confirmed the functionality of allanti-HCN antibodies used (Fig. 2C).

The experiments presented so far suggest that native SAN f-channels are mainly made of HCN4 proteins. To further confirm thisconclusion, we performed quantitative RT-PCR experiments to assessthe relative abundance of mRNA messengers for the three isoformsinvestigated (HCN1, HCN2, and HCN4). The amounts of transcriptswere measured relative to the housekeeping gene 28S, which isknown to be uniformly distributed in SAN cells [16]. Results indicatean HCN4/HCN1 ratio of 7.4±0.5 (n=4) and an HCN4/HCN2 ratio of46.4±7.7 (n=4), each experimentwas repeated three times (see also

Fig. 4. Histological identification and molecular mapping of sections of SAN tissue. Tissue serepresents a different treatment. (A) The Masson-trichrome labeling was employed to idenhybridization) and (C) (immunofluorescence) show enlargements of panel (A) (squares). Tbetween mRNA and protein signals is apparent. Calibration bars: 1 mm.

Supplementary Fig. 1 and Supplementary materials and methods formore details). Thus, the results relative to mRNA abundance are inagreement with the data in Figs. 1 and 2 indicating fairly low levels ofHCN1 and essential lack of HCN2 expression in the SAN.

The SAN is not a homogeneous structure since both the cytologicalorganization and ion channel expression profile vary substantiallywithin the node itself [20,21]. We thus set to determine thedistribution of the HCN4 protein in the whole node since this is theonly isoform clearly detectable at the protein level, and to do so weanalyzed tissue sections obtained from intact SAN preparations. Thesections were cut perpendicularly to the CT up to the SI. Starting fromthe SVC and proceeding caudally towards the IVC, we cut 3 adjacenttissue slices every 1 mm (Fig. 3) for a total length of 10 mm.

This procedure allowed the collection of 10 sets of 3 sections each,which extended throughout the entire node (levels L1–L10, Fig. 3).The slices were used for Masson's trichrome, HCN4 in situ hybridiza-tion, and HCN4 immunofluorescence labeling (Figs. 4A–C, respec-tively). The four sections shown in Figs. 4A–C represent L2, L4, L6, andL8 as indicated in Fig. 3. TheMasson's trichrome staining (Fig. 4A) wasinitially used to identify the histological composition of the nodaltissue. The presence of compact muscle tissue (violet) and of limitedamount of connective tissue (blue) characterizes the region of the CT.This situation is partially reversed in the central region of the node,where few muscular fibers are intertwined with abundant connectiveelements. To better resolve the distribution of HCN4 signal, panels Band C of Fig. 4 show enlargements of selected portions of panels A(boxes). Immunolabeling of HCN4 proteins (Fig. 4C) is strong in thecentral node, identified as the region starting off the right side of theCT, but thin strips of signal lining the upper and lower borders of theCT are also clearly stained. This pattern of staining thus confirms thatnodal tissue also partially wraps around the CT. On the contrary, on theright-hand side of the slices, i.e. towards the SI, the HCN4 signaldecreases until it eventually fades away, indicating poor HCN4expression (see also Fig. 5C). Similar observations can be made onthe corresponding in situ hybridization images (Fig. 4B) confirmingthe tight correlation between expression of mRNA and of correspond-ing protein.

In Fig. 5 a comparative histological (A), in situ hybridization (B)and immunofluorescence (C) analysis of the tissue slices collected at

ctions collected from levels 2, 4, 6, and 8 as in Fig. 3 are aligned vertically. Each columntify muscular tissue (violet), connective tissue (blue), and nuclei (black). (B) (in situhe central area of the node is extensively stained; in addition, a clear correspondence

Fig. 5. Staining of selected regions of the intercaval area. Top, regions of nodal tissueinvestigated. The CT, SAN and SI regions delimited by the boxes are enlarged in thebottom (A–C) panels. A, Masson-trichrome staining of the three selected regions of theintercaval area (boxes). B, in situ hybridization experiments showing dense expressionof HCN4 mRNA in the centre of the node. Note that upper and lower layers of the nodalstrip are mostly made of connective tissue; epithelial surfaces were lost duringprocessing of tissue slices. C, immunofluorescence staining confirms that the proteinsignal arising from primary pacemaker cells (SAN) is higher than that from other moreperipheral regions. Calibration bar: 20 μm.

Fig. 6. HCN4, Cx43 and NF-M immunolabeling of the rabbit intercaval area. (A) HCN4(green) and Cx43 (red) immunolabeling of an intercaval tissue slice; enlargementscorresponding to the regions identified by arrows in (A) are shown in panels (B) and(C). HCN4 signal clearly labels the whole SAN region, while Cx43 is only detectableoutside the node or in its peripheral regions (B). (D) HCN4 (green) and NF-M (red)immunolabeling of an intercaval tissue slice; enlargements corresponding to theregions identified by arrows in (D) are shown in panels (E) and (F). For both proteinsthe signal distributions show a selective labeling of the peripheral and central SANareas. Calibration bars: (A, D) 1 mm; (B, C, E, F) 20 μm.

224 C. Brioschi et al. / Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

level L5 is shown. This analysis aimed to identify any differences inpatterns of expression of HCN4 in the CT, the SAN and the SI.

Visual inspection clearly indicates that while in the nodal area astrong HCN4-specific signal was detected, the same was much lowerin the region of the CT, and almost virtually absent from the SI region(Figs. 5B, C). It also appears that cells of the node positive for HCN4 areconcentrated in the middle portions of the slice and are bordered byconnective tissue at both the epicardial and endocardial sides. Theconnective tissue is however not only confined to these surface layers,but is also present within the HCN4-positive area, where blue stainingis apparent (Fig. 5A), in agreement with the notion that nodal cells areembedded in a dense matrix of collagen fibers [2,20].

Typically, the rabbit SAN area is identified by the lack of expressionof Cx43 and by the expression of NF-M [1,20]; we therefore carried outdouble labeling experiments (Cx43/HCN4, and NF-M/HCN4) to verifythe spatial distribution of HCN4 relative to that of known markers ofthe rabbit SAN.

Typical tissue slices labeled with HCN4 and Cx43, and with HCN4and NF-M antibodies are shown in panels A and D, respectively;details at highermagnification corresponding to the regions identifiedby the arrows are also shown (Figs. 6B, C and E, F). Signal distributionof the central part of the node shows a marked prevalence of HCN4

(green) within the SAN area, whereas Cx43 (red) is abundantlyexpressed in the working myocytes of the CT (panels A); imagesshown in panels B and C confirm that, although Cx43 is absent fromthe central SAN area (C), it partially infiltrates the nodal area at theboundaries with the CT (B). Double labeling of HCN4 (green) and NF-M (red) shows that both proteins are highly expressed within the SANarea (Fig. 6D); enlargements of the SAN regions confirm that bothHCN4 and NF-M (Fig. 6E) signals are also present at the periphery ofthe node.

A quantitative description of the regional distribution andfunctional properties of HCN4 channels was achieved by fluorescenceand electrophysiological measurements. Single cells were isolatedfrom the CT (n=22), SAN (n=34) and SI (n=24) regions and themean (±SEM) fluorescence densities were plotted in Fig. 7A.Electrophysiological experiments were also performed to evaluatethe If current density, i.e. current amplitude normalized to cellcapacitance, in single cells isolated from the same regions (CT, n=18;SAN, n=35; SI, n=18, Fig. 7B). Statistical analysis (one-way ANOVAand post-hoc Bonferroni test) confirmed that the three groups differsignificantly for both parameters (pb0.05). Normalizing mean signals

Fig. 7. Quantification of HCN4 expression in cells isolated from CT, SAN, and SI. (A, B)Plots of mean immunofluorescence and If current densities measured in single cellsisolated from CT, SAN, and SI. The current was normalized against cell capacitance.Current densities were measured at the end of a 4 s hyperpolarizing step to −125 mV.In order to measure the HCN4 fluorescence density, a single-section of visual scan wasacquired for each cell, and the signal intensity was calculated over an areacorresponding to the cell membrane. This area had a length variable from cell to celland a fixed depth of 0.77 μm. (C, D) Samples of the fluorescence images and currenttraces corresponding to plots in (A, B). Calibration bars: 10 μm.

Fig. 8. Cellular organization of the central region of the SAN. (A, B) High resolutionimages of sections obtained in the central area of the node. Cells are organized in groups(islets) interconnected by thin cytoplasmic bridges (arrows). (C) Whole-tissue labelingwith anti-HCN4 antibody confirms the cellular organization in islets and bridges.Calibration bars: 10 μm.

225C. Brioschi et al. / Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

to the values obtained in SAN cells yields fluorescence density ratios of0.39 and 0.22, and current density ratios of 0.38 and 0.10 for CT and SIcells, respectively.

Data in Figs. 7A and B confirm that HCN4 proteins are denselyexpressed in the nodal area and that their expression decreasesprogressively in the bordering regions. The observation that the twoparameters used to estimate the expression of HCN4 proteins per cell(fluorescence and current density) have similar ratios in each group(SI and CT) highlights the correlation between functional andmolecular aspects of HCN4 channel expression. Sample images ofHCN4 signal and plots of the pacemaker current from the threeregions investigated (CT, SAN, SI) are shown in panels C and D.

While fluorescence and functional data therefore converge toindicate that expression of HCN4 channels is highest in the centralnode, they do not provide information as to how single nodal cells aredistributed within this area. Images shown in Figs. 5C and 6D abovesuggest that in the central region, nodal myocytes are organized insmall agglomerates (islets), rather than in uniform cell layers. Thisaspect is further investigated in Figs. 8A and Bwhere details of a tissueslice obtained from the central nodal region (position L5 in Fig. 3) areshown at a higher magnification.

These sample images show details of the multicellular organiza-tion of nodal cells within the SAN. The staining pattern reveals a nonhomogeneous distribution of the HCN4 signal; in particular, multipleregions of high density can be seen. According to this pattern, SANmyocytes within the nodal structure are organized as islets composed

by several highly intertwined cells and these regions of intense signalappear to be linked by thin “bridge-like” structures (arrows), possiblyinvolved in electrical coupling between different islets.

Additional evidence supporting a high density spot-type of tissueorganization was obtained by immunofluorescence analysis of wholeSAN tissues. In these experiments intact SA nodes were immunola-beled for the presence of HCN4 (green), and nuclei were identified byDAPI staining (blue). In Fig. 8C a portion of the whole-SAN is shownand the HCN4 labeling confirms the presence of thin positively stainedprolongations connecting regions where the signal was more intense(islets).

4. Discussion

Funny (f-) channels are expressed in several excitable tissueswhere they provide the basis for a variety of cellular functionsincluding automaticity, modulation of excitability, signal integrationand plasticity [22,23]. Most of these functions are associated with theability of funny channels to induce and control a depolarizing processin response to an appropriate stimulus, such as a membrane potentialhyperpolarization or an increased level of intracellular cAMP. Severalregulatory mechanisms, including compartmentation, accessory sub-units, macrocomplex proteins, protein kinases, phorbol esters andphosphoinositides, have been shown to modulate HCN properties andthus provide additional means to control their function [24–30].

Evidence indicates that native pacemaker channels are tetramers[6] and that functional channels can arise from both heteromeric andhomomeric assemblies of all subunits [7–10,31]. Kinetics and cAMPsensitivity vary among isoforms; for example, activation/deactivationrates are faster for HCN1, slower for HCN4, and intermediate for HCN2and HCN3, while cAMP-dependent modulation of HCN2 and HCN4 ismore efficient than modulation of HCN1 and HCN3 [32]. Heteromericchannels made of different HCN isoforms have properties intermedi-ate between those of individual homomeric channels [7–9,31,33],

226 C. Brioschi et al. / Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

which further improves adaptability of HCN channels to comply withcell type-specific functional requirements.

The leading cardiac pacemaker site is defined as the area locatedroughly midway between the two venae cavae and adjacent to the CT.Our data show that this region coincideswith the onewhere the HCN4protein is most densely expressed, as illustrated in Figs. 4–7. Thisfinding nicely adds to the current knowledge according to which thenode is defined as the area lacking expression of ANP and Cx43 and, inrabbit, showing expression of NF-M [1] as it is clearly shown in Fig. 6.In fact, it allows the association of the main functional role of the SAN,i.e. pacemaking, with the specific expression of HCN4 channels in thenode itself. The observation that Cx43 partially infiltrates also theregions at the borders of the SAN (Fig. 6B) strengthens the hypothesisthat a high conductance connexin is required to facilitate the exit fromthe node of the excitation wave [20,34].

Our evidence that HCN4 expression is highly concentrated in therabbit SAN agrees with the established role of f-channels in cardiacpacemaking initiation and rate modulation, and with the notion thatspontaneous activity is typical of pacemaker tissue and is notfunctional (except under special pathological conditions) in theworking muscle surrounding the SAN. This result is important inthat it establishes a link between existing electrophysiological data onthe cellular pacemaker mechanisms, mostly collected from the rabbitSAN, and the main channel isoform responsible for these functionalproperties.

In order to understand towhat extent the datawe have collected inthe rabbit can also be applied to other species, it is important toestablish whether high level expression of HCN4 is a property specificof the rabbit SAN or it is a general feature of nodal tissues acrossmammals. To date molecular identification of cardiac pacemakerchannels, either at the message or protein level, has been carried outin several species, including rabbits, mice, rats, dogs, and humans [22],and studies have shown that themajor component of the native SAN Ifis generally the HCN4 isoform [11–15,35]. Interestingly, this is trueregardless of the main basal heart rate that can vary substantiallyamong different species. This evidence supports the notion that HCN4channels provide a depolarizing contribution for a wide range of rates.Also relevant is the observation that the HCN4 signal can beconsidered as a marker for the early identification of the SAN regionduring mice embryonic development in mice [36].

Our results show that the contribution of isoforms other thanHCN4 to the assembly of pacemaker channels in the SAN rabbit issubstantially poorer. In particular while the HCN2 signal was neverdetected (either with western blotting or with single cell immuno-labeling), the HCN1 signal was occasionally observed but only insingle cell labeling experiments (Figs. 1 and 2). These data are furthersupported by the quantitative comparison of the relative abundance ofHCN1, HCN2, and HCN4 mRNAs by qRT-PCR. Quantitative results onHCN4 vs HCN1 abundance indicate a 7.4-fold prevalence of HCN4, infull agreement with previous data obtained both by qRT-PCR [16] andby RNase protection assay [14], which indicate a 4- to 10-foldprevalence of HCN4. The quantitative relevance of HCN2 mRNA waspreviously assessed only by RNase protection assay; our dataindicating a relative HCN4/HCN2 abundance of 46.4-fold is thereforethe first quantitative result obtained by qRT-PCR experiments.

Taken together, these data thus suggest that: 1) HCN2 proteins canbe assumed to be functionally absent from the rabbit SAN, and 2)HCN1 proteins are expressed in the rabbit SAN, but their level ofexpression is low, suggesting a limited HCN1 contribution tofunctional f-channel properties.

Analysis of the HCN protein expression in the murine SAN tissueconfirmed the robust expression of HCN4, but also revealed the lack ofdetectable signal for HCN1 and HCN2 [12]. Investigation of HCNmRNAs also confirmed the prevalence of HCN4 [11,15] but yielded lessclear results for the relative abundance of HCN1 and HCN2.Quantitative PCR experiments revealed. a ∼3-fold prevalence of

HCN1 (Fig. 7 from [11]), while in situ hybridization analysis reportedmoderate and low levels for HCN2 and HCN1, respectively [15].Important functional data on the relevance of HCN2 also come fromthe analysis of HCN2 knockout mice. In single SAN myocytes of thesemice the If current is reduced by about 30%, suggesting a contributionof HCN2 proteins to the native f-channels [37].

Data collected from cells isolated from the contractile part ofmouse embryoid bodies indicate that, despite the presence of mRNAcoding for all HCN isoforms, only HCN1 and HCN4 proteins aredetectable. Furthermore, the fraction of cells expressing mostly HCN4increases at later stages of cell differentiation [38]. Data on HCNmRNAs abundance in the human SAN similarly indicate a highexpression of HCN4 mRNA, along with faint signals for HCN1 andHCN2 [35,39]. Interestingly the relative abundances of HCN mRNAsmeasured in the human SAN (HCN4/HCN1=6.3; HCN4/HCN2=33.8, data from Chandler et al. [39] Figs. 2J–L) correspondwell to our results in the rabbit SAN (HCN4/HCN1=7.4 and HCN4/HCN2=46.4). Our data agree with the view that HCN1 subunits maycontribute, albeit to an extent limited by their low expression rate, tonative f-channel properties, as suggested by previous work [8,40]. Alimit in the exact estimation of the contribution of individual isoformsresults from a “context” dependence of f-channel properties [41],likely due to local interactions of HCN channel α-subunits withancillary and/or modulatory elements such as local cAMP concentra-tion, MiRP1, caveolin-3, PIP2, and protein kinases [17].

As well as providing information on the expression of HCN4withinthe SAN region and its surroundings, immunolabeling experimentsallow the identification of membrane-associated spots of highintensity signal (Figs. 2, 7, and 8) indicating the presence of scatteredclusters of high-density HCN4 expression. This confirms previousobservations that regions of high current densities (“hot spots”) canbe occasionally identified by cell-attached patch-clamp recordings[18]. Clustering of HCN channels is likely due to a specific localizationof the HCN4 protein to lipid rafts due to the interaction of the HCN4proteinwith caveolin-3 (see Supplementary Fig. 2 for a representativeimage showing the colocalization of these two proteins in a single SANcell). A recent quantitative analysis of the colocalization of HCN4 andcaveolin-3 has indicated a degree of colocalization of about ∼77% [30].This structural arrangement can also provide compartmentation ofionic channels and membrane receptors involved in autonomiccontrol of cardiac rate [30]. Of relevance is also the observation thatthe confinement of the channel in membrane microdomains mod-ulates the biophysical and functional properties of the channel itself;for example disruption of caveolar structures determines a positiveshift of the If activation curve [29]. It is thus possible that some of thedifferences observed among pacemaker currents recorded in variousspecies partly derive from differences in the properties of caveolae inSAN cells.

Several pieces of evidence indicate that the cell composition of theSAN is quite heterogeneous; for example about 50% of the rabbit nodeis represented by connective tissue [42] and similar values (45–75%)have also been found in humans [21]. Although our experiments didnot address the detailed structural organization of SAN cardiomyo-cytes within the rabbit node, some interesting considerations arisefrom the images shown in Figs. 8A and B. HCN4-positive cells appearto be organized in isolated small multi-cellular aggregates (islets)bridged by thin cytoplasmic extensions to form a three-dimensionalmesh-like structure. Branch-like cell extensions have previously beenobserved by other groups in isolated SAN cells, but their function wasunknown [17]. Our findings confirm the presence of cell extensionsalso in the context of a multicellular organization, and further showthat these structures appear to be involved in connecting otherwiseseparated islets of SAN cells within the nodal area. Whether thisbranch-like organization is relevant for the correct spreading of theimpulse within the node must be further investigated. The structurethat results from our data based on labeling of the HCN4 pacemaker

227C. Brioschi et al. / Journal of Molecular and Cellular Cardiology 47 (2009) 221–227

protein nicely agrees with histological analysis [3], also reporting theexistence in the rabbit node of clusters of nodal myocytes separated byconnective tissue. It is also worth noting that we did not identify inany part of the node a “compact” region characterized by an almostexclusive presence of nodal myocytes as recently described in themouse SAN [12].

The diastolic depolarization is a complex event and, as such is theresult of several contributing cellular processes, among whichhowever the activation during diastole of the funny current plays anestablished key role. According to our results, funny channelsrepresent specific molecular markers of automatic tissue. Indeed,while other contributing mechanisms such as INa, ICaL, Na/Caexchangers and others are ubiquitously present in almost all excitablecells, the If/HCN4 current is selectively and exclusively expressed (atfunctionally relevant levels) in those cells exhibiting automaticitysuch as SAN, Purkinje fibers, atrioventricular cells, and precursor ofcardiac cells [18,43,44]. We therefore believe that the tight link wehave found between the cardiac pacemaker region identified byhistological and electrophysiological criteria and the SAN regionexpressing high levels of HCN4, represents a further strong indicationthat f-channels are essential to basic pacemaker activity. In conclusion,our data indicate that, as well as being characterized by the presenceof Cx45 and NF-M and by the absence of both ANP and Cx43 [1], rabbitSAN myocytes typically express a high density of HCN4 channels,which can therefore be regarded as markers of pacemaker myocytes.

Acknowledgments

This work was supported by grants from Ministero dell' Istruzionedell'Università e della Ricerca (Cofin 2007WB35CW) to MB and byEuropean Union (Normacor) and MIUR-FIRB (RBLA035A4X_003)grants to DD. We wish to thank B. Terragni, Tea Sala and A. Atkinsonfor their valuable technical assistance.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.yjmcc.2009.04.009.

References

[1] Boyett MR, Dobrzynski H, Lancaster MK, Jones SA, Honjo H, Kodama I.Sophisticated architecture is required for the sinoatrial node to perform itsnormal pacemaker function. J Cardiovasc Electrophysiol 2003;14:104–6.

[2] Camelliti P, Green CR, LeGrice I, Kohl P. Fibroblast network in rabbit sinoatrialnode: structural and functional identification of homogeneous and heterogeneouscell coupling. Circ Res 2004;94:828–35.

[3] 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.

[4] Bucchi A, Baruscotti M, Robinson RB, DiFrancesco D. Modulation of rate byautonomic agonists in SAN cells involves changes in diastolic depolarization andthe pacemaker current. J Mol Cell Cardiol 2007;43:39–48.

[5] Verkerk AO, Wilders R, van Borren MM, Peters RJ, Broekhuis E, Lam K, et al.Pacemaker current (If) in the human sinoatrial node. Eur Heart J 2007;28:2472–8.

[6] Craven KB, Zagotta WN. CNG and HCN channels: two peas, one pod. Annu RevPhysiol 2006;68:375–401.

[7] Ulens C, Tytgat J. Functional heteromerization of HCN1 and HCN2 pacemakerchannels. J Biol Chem 2001;276:6069–72.

[8] Altomare C, Terragni B, Brioschi C, Milanesi R, Pagliuca C, Viscomi C, et al.Heteromeric HCN1–HCN4 channels: a comparison with native pacemakerchannels from the rabbit sinoatrial node. J Physiol 2003;549:347–59.

[9] Chen S, Wang J, Siegelbaum SA. Properties of hyperpolarization-activated pace-maker current defined by coassembly of HCN1 and HCN2 subunits and basalmodulation by cyclic nucleotide. J Gen Physiol 2001;117:491–504.

[10] Much B, Wahl-Schott C, Zong X, Schneider A, Baumann L, Moosmang S, et al. Roleof subunit heteromerization and N-linked glycosylation in the formation offunctional hyperpolarization-activated cyclic nucleotide-gated channels. J BiolChem 2003;278:43781–6.

[11] Marionneau C, Couette B, Liu J, Li H,MangoniME, Nargeot J, et al. Specific pattern ofionic channel gene expression associated with pacemaker activity in the mouseheart. J Physiol 2005;562:223–34.

[12] Liu J, Dobrzynski H, Yanni J, Boyett MR, Lei M. Organisation of the mouse sinoatrialnode: structure and expression of HCN channels. Cardiovasc Res 2007;73:729–38.

[13] Ishii TM, Takano M, Xie LH, Noma A, Ohmori H. Molecular characterization of thehyperpolarization-activated cation channel in rabbit heart sinoatrial node. J BiolChem 1999;274:12835–9.

[14] Shi W, Wymore R, Yu H, Wu J, Wymore RT, Pan Z. Distribution and prevalence ofhyperpolarization-activated cation channel (HCN) mRNA expression in cardiactissues. Circ Res 1999;85:e1–6.

[15] Moosmang S, Stieber J, Zong X, Biel M, Hofmann F, Ludwig A. Cellular expressionand functional characterization of four hyperpolarization-activated pacemakerchannels in cardiac and neuronal tissues. Eur J Biochem 2001;268(6):1646–52.

[16] 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.

[17] Mangoni ME, Nargeot J. Genesis and regulation of the heart automaticity. PhysiolRev 2008;88:919–82.

[18] DiFrancesco D, Ferroni A, Mazzanti M, Tromba C. Properties of the hyperpolarizing-activated current (If) in cells isolated from the rabbit sino-atrial node. J Physiol1986;377:61–88.

[19] 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.

[20] Boyett MR, Honjo H, Kodama I. The sinoatrial node, a heterogeneous pacemakerstructure. Cardiovasc Res 2000;47:658–87.

[21] Camelliti P, Borg TK, Kohl P. Structural and functional characterisation of cardiacfibroblasts. Cardiovasc Res 2005;65:40–51.

[22] Baruscotti M, Robinson RB. Electrophysiology and pacemaker function of thedeveloping sinoatrial node. Am J Physiol Heart Circ Physiol 2007;293:H2613–23.

[23] Stieber J, Hofmann F, Ludwig A. Pacemaker channels and sinus node arrhythmia.Trends Cardiovasc Med 2004;14:23–8.

[24] Zong X, Eckert C, Yuan H,Wahl-Schott C, Abicht H, Fang L, et al. A novel mechanismof modulation of hyperpolarization-activated cyclic nucleotide-gated channels bySrc kinase. J Biol Chem 2005;280:34224–32.

[25] Arinsburg SS, Cohen IS, Yu HG. Constitutively active Src tyrosine kinase changesgating of HCN4 channels through direct binding to the channel proteins.J Cardiovasc Pharmacol 2006;47:578–86.

[26] Fogle KJ, Lyashchenko AK, Turbendian HK, Tibbs GR. HCN pacemaker channelactivation is controlled by acidic lipids downstream of diacylglycerol kinase andphospholipase A2. J Neurosci 2007;27:2802–14.

[27] Pian P, Bucchi A, Decostanzo A, Robinson RB, Siegelbaum SA. Modulation of cyclicnucleotide-regulated HCN channels by PIP(2) and receptors coupled to phospho-lipase C. Pflugers Arch 2007;455:125–45.

[28] Gravante B, Barbuti A, Milanesi R, Zappi I, Viscomi C, DiFrancesco D. Interaction ofthe pacemaker channel HCN1 with filamin A. J Biol Chem 2004;279:43847–53.

[29] Barbuti A, Gravante B, Riolfo M, Milanesi R, Terragni B, DiFrancesco D. Localizationof pacemaker channels in lipid rafts regulates channel kinetics. Circ Res 2004;94:1325–31.

[30] Barbuti A, Terragni B, Brioschi C, DiFrancesco D. Localization of f-channels tocaveolae mediates specific beta2-adrenergic receptor modulation of rate insinoatrial myocytes. J Mol Cell Cardiol 2007;42:71–8.

[31] Er F, Larbig R, Ludwig A, Biel M, Hofmann F, Beuckelmann DJ, et al. Dominant-negative suppression of HCN channels markedly reduces the native pacemakercurrent I(f) and undermines spontaneous beating of neonatal cardiomyocytes.Circulation 2003;107:485–9.

[32] Baruscotti M, Bucchi A, DiFrancesco D. Physiology and pharmacology of the cardiacpacemaker (“funny”) current. Pharmacol Ther 2005;107:59–79.

[33] Zha Q, Brewster AL, Richichi C, Bender RA, Baram TZ. Activity-dependentheteromerization of the hyperpolarization-activated, cyclic nucleotide gated(HCN) channels: role of N-linked glycosylation. J Neurochem 2008;105:68–77.

[34] Joyner RW, van Capelle FJ. Propagation through electrically coupled cells. How asmall SA node drives a large atrium. Biophys J 1986;50:1157–64.

[35] Thollon C, Bedut S, Villeneuve N, Coge F, Piffard L, Guillaumin JP, et al. Use-dependent inhibition of hHCN4 by ivabradine and relationship with reduction inpacemaker activity. Br J Pharmacol 2007;150:37–46.

[36] Mommersteeg MT, Hoogaars WM, Prall OW, Gier-de Vries C, Wiese C, Clout DE,et al. Molecular pathway for the localized formation of the sinoatrial node. Circ Res2007;100:354–62.

[37] Ludwig A, Budde T, Stieber J, Moosmang S, Wahl C, Holthoff K, et al. Absenceepilepsy and sinus dysrhythmia in mice lacking the pacemaker channel HCN2.EMBO J 2003;22:216–24.

[38] Barbuti A, Crespi A, Capilupo D, Mazzocchi N, Baruscotti M, DiFrancesco D.Molecular composition and functional properties of f-channels in murineembryonic stem cell-derived pacemaker cells. J Mol Cell Cardiol 2009;46:343–51.

[39] Chandler NJ, Greener ID, Tellez JO, Inada S, Musa H, Molenaar P, et al. Moleculararchitecture of the human sinus node: unsights into the function of the cardiacpacemaker. Circulation 2009;46:1562–75.

[40] Moroni A, Gorza L, Beltrame M, Gravante B, Vaccari T, Bianchi ME, et al.Hyperpolarization-activated cyclic nucleotide-gated channel 1 is a moleculardeterminant of the cardiac pacemaker current I(f). J Biol Chem 2001;276:29233–41.

[41] Qu J, Altomare C, Bucchi A, DiFrancesco D, Robinson RB. Functional comparison ofHCN isoforms expressed in ventricular and HEK 293 cells. Pflugers Arch 2002;444:597–601.

[42] Opthof T. The mammalian sinoatrial node. Cardiovasc Drugs Ther 1988;1:573–97.[43] Dobrzynski H, Nikolski VP, Sambelashvili AT, Greener ID, Yamamoto M, Boyett MR,

et al. Site of origin andmolecular substrate of atrioventricular junctional rhythm inthe rabbit heart. Circ Res 2003;93:1102–10.

[44] Hescheler J, Fleischmann BK, Lentini S, Maltsev VA, Rohwedel J, Wobus AM, et al.Embryonic stem cells: a model to study structural and functional properties incardiomyogenesis. Cardiovasc Res 1997;36:149–62.