1-s2.0-S0163725805000252-main

Embed Size (px)

Citation preview

  • 8/20/2019 1-s2.0-S0163725805000252-main

    1/21

    Associate editor: G.E. Billman

    Physiology and pharmacology of the cardiac pacemaker (bfunny Q ) current 

    Mirko Baruscotti, Annalisa Bucchi, Dario DiFrancescoT

     Laboratory of Molecular Physiology and Neurobiology, Department of Biomolecular Sciences and Biotechnology, University of Milano,

    via Celoria 26, 20133 Milan, Italy

    Abstract

    First described over a quarter of a century ago, the cardiac pacemaker   bfunny Q  ( I f ) current has been extensively characterized since, and its

    role in cardiac pacemaking has been thoroughly demonstrated. A similar current, termed I h, was later described in different types of neurons,where it has a variety of functions and contributes to the control of cell excitability and plasticity.  I f  is an inward current activated by both

    voltage hyperpolarization and intracellular cAMP. In the heart, as well as generating spontaneous activity, f-channels mediate autonomic-

    dependent modulation of heart rate:   h-adrenergic stimulation accelerates, and vagal stimulation slows, cardiac rate by increasing and

    decreasing, respectively, the intracellular cAMP concentration and, consequently, the f-channel degree of activation. Four isoforms of 

    hyperpolarization-activated, cyclic nucleotide-gated (HCN) channels have been cloned more recently and shown to be the molecular 

    correlates of native f-channels in the heart and h-channels in the brain. Individual HCN isoforms have kinetic and modulatory properties

    which differ quantitatively. A comparison of their biophysical properties with those of native pacemaker channels provides insight into the

    molecular basis of the pacemaker current properties and, together with immunolabelling and other detection techniques, gives information on

    the pattern of HCN isoform distribution in different tissues. Because of their relevance to cardiac pacemaker activity, f-channels are a natural

    target of drugs aimed at the pharmacological control of heart rate. Several agents developed for their ability to selectively reduce heart rate act 

     by a specific inhibition of f-channel function; these substances have a potential for the treatment of diseases such as angina and heart failure.

    In the near future, devices based on the delivery of f-channels in situ, or of a cellular source of f-channels (biological pacemakers), will likely

     be developed for use in therapies for diseases of heart rhythm with the aim of replacing electronic pacemakers.D  2005 Elsevier Inc. All rights reserved.

     Keywords: I f  current; Funny current; f-Channels; HCN; Bradycardic agents

     Abbreviations:   CNBD, cyclic nucleotide-binding domain; CNG, cyclic nucleotide-gated; HCN, hyperpolarization-activated, cyclic nucleotide-gated; SAN,

    sinoatrial node.

    Contents

    1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

    2. Sinoatrial  I f   current and the mechanism of cardiac pacemaking . . . . . . . . . . . . . . . . . . 61

    2.1. Early experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

    2.2. Biophysical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612.2.1. Voltage dependence of   I f  kinetics . . . . . . . . . . . . . . . . . . . . . . . . . 61

    2.2.2. Kinetics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

    2.2.3. Ionic nature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

    2.3. Contribution to automaticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

    2.4. Autonomic modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

    2.5. Single f-channel recording . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

    3. Molecular determinants of the  I f  current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

    3.1. Hyperpolarization-activated, cyclic nucleotide-gated clones. . . . . . . . . . . . . . . . . 65

    0163-7258/$ - see front matter  D  2005 Elsevier Inc. All rights reserved.

    doi:10.1016/j.pharmthera.2005.01.005

    T  Corresponding author. Tel.: +39 02 50314931; fax: +39 02 50314932.

     E-mail address:  [email protected] (D. DiFrancesco).

    Pharmacology & Therapeutics 107 (2005) 59–79

    www.elsevier.com/locate/pharmthera

  • 8/20/2019 1-s2.0-S0163725805000252-main

    2/21

    3.2. Structure–function relationships for hyperpolarization-activated,

    cyclic nucleotide-gated channels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

    3.2.1. Structures involved in voltage-dependent gating. . . . . . . . . . . . . . . . . . . 68

    3.2.2. Structures involved in cAMP-dependent gating . . . . . . . . . . . . . . . . . . . 69

    3.3. Hyperpolarization-activated, cyclic nucleotide-gated isoforms in sinoatrial node tissue . . . 69

    4. f-Channel blockers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

    4.1. Alinidine (ST567) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714.2. Zatebradine (UL-FS 49) and cilobradine (DK-AH269) . . . . . . . . . . . . . . . . . . . . 71

    4.3. ZD-7288 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

    4.4. Ivabradine (S16257) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

    5. Future directions: the biological pacemaker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

    6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

    Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

    References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

    1. Introduction

    In mammals, cardiac pacemaking originates in a highly

    specialized structure located in the wall of the right atrium,the sinoatrial node (SAN). SAN myocytes generate sponta-

    neous action potentials; these propagate, through specialized

    conduction systems, first to the atria and then to the

    ventricles, and thus drive cardiac rhythmic contractions.

    The control of cardiac rate is a process of major 

     physiological relevance, and it is not surprising that the

    identification of the electrical events underlying sinoatrial

     pacemaker activity has attracted the constant interest of 

    cardiac physiologists.

    During diastole, corresponding to mechanical relaxation,

    myocytes of the working myocardium (atrial and ventricular 

    cells) lack electrical activity and normally rest at a hyper-

     polarized voltage level. Spontaneously beating SAN myo-

    cytes, on the contrary, are characterized by the presence of a

    bslow diastolic Q  depolarization phase of the action potential.

    At the termination of one SAN action potential, the

    membrane voltage does not stabilize to a negative level

     but slowly creeps up with an approximately constant slope,

    until it reaches the threshold for a new SAN action potential

    (DiFrancesco, 1993).

    The slow diastolic   b pacemaker  Q  depolarization is respon-

    sible for the generation of repetitive activity and has

    therefore been the focus of intense studies aimed to the

    understanding of the cellular mechanisms that generate it.

    Although the cellular processes ultimately contributing,directly or indirectly, to the pacemaker depolarization are

    many, and their involvement is still a subject of inves-

    tigation and debate   (DiFrancesco & Robinson, 2002;

    Vinogradova et al., 2002), there is now general agreement 

    that a major role in the generation and control of this phase

    is played by the so-called   b pacemaker  Q   (funny,   I f ) current.

    Since its discovery in the SAN  (Brown et al., 1979), the

    funny current has been thoroughly investigated in cardiac

     pacemaker, and detailed knowledge of several of its

     properties has been obtained. Importantly, a hyperpolariza-

    tion-activated current similar to I f  ( I h  current) has also been

    described in a large number of different types of neurons,

    where it contributes to a wide range of physiological

    functions such as rhythmic firing, regulation of neuronal

    excitability, sensory transduction, synaptic plasticity, and

    more (Pape, 1996; Robinson & Siegelbaum, 2003). Despitethe early description of the funny current in the SAN, the

    molecular components of f-channels were particularly

    elusive, and their cloning was only achieved by chance in

    the late 1990s. The first clone to be obtained (BCNG1), was

    originally thought to be a novel K + channel related to Eag

    and cyclic nucleotide-gated (CNG) channels  (Santoro et al.,

    1997) and only later found to have properties expected from

    f/h-channel subunits (Santoro et al., 1998). Further cloning

    revealed the existence of four isoforms forming a new

    family of hyperpolarization-activated, cyclic nucleotide-

    gated channels (HCN1 to 4) and belonging to the super-

    family of voltage-dependent K + (Kv) and cyclic nucleotide-

    gated (CNG) channels (see reviews by Ludwig et al., 1999a;

    Santoro & Tibbs, 1999; Kaupp & Seifert, 2001; Accili et al.,

    2002; Robinson & Siegelbaum, 2003). A thorough inves-

    tigation has shown that the 4 HCN isoforms have different 

     properties and variable patterns of expression, and that 

    different isoforms can colocalize to form heteromultimers

    with specific kinetic and modulatory properties.

    Given its specific functional relevance in cardiac pace-

    making and rate regulation, the funny current has long been

    considered a primary target for the development of drug-

     based therapeutic strategies aiming to a selective control of 

    heart rate. Drugs targeting this current would selectively

    alter the diastole, avoiding undesirable and potentially proarrhythmic effects on other phases of the action

     potential. The selection of drugs specifically interacting

    with funny channels has advanced rapidly in the last few

    years and is now moving from basic research towards

    clinical application. A successful search for these drugs may

    have a significant impact on specific cardiac therapies. The

    use as pharmacological tools of drugs interacting with ion

    channels or with their function is indeed widespread, since

    several diseases such as epilepsy, long QT syndrome, cystic

    fibrosis, myopathies, and others have been found to depend

    on abnormalities of genes coding for channel proteins or 

    auxiliary subunit proteins.

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7960

  • 8/20/2019 1-s2.0-S0163725805000252-main

    3/21

    In this review we will focus on the properties of cardiac

    f- channels, most of which are analyzed in SAN pacemaker 

    cells, and address their physiological role, molecular 

    composition, localization, and interaction with drugs.

    Potential applications to the therapeutic use of the functional

    role of funny channels in the generation of spontaneous

    activity and heart-rate regulation will also be addressed.

    2. Sinoatrial   I f  current and the

    mechanism of cardiac pacemaking

    2.1. Early experiments

    Early experiments investigating the nature of the

    diastolic depolarization were conducted in Purkinje fibres

    and led to the hypothesis that this process was due to the

    decay of the delayed K + conductance activated during the

     preceding action potential (Weidmann, 1951). Resultsapparently compatible with this assumption were obtained

     by measurements of conductance changes in voltage clamp

    conditions (Vassalle, 1966). The idea received further 

    confirmation with the finding in Purkinje fibres of the so-

    called   b pacemaker  Q    ( I K2) current, described as a pure K +

    current activated upon depolarization in the diastolic range

    of voltages ( Noble & Tsien, 1968; Peper & Trautwein,

    1969); the assumption of a pure K + ionic nature for   I K2depended essentially on evidence for a current reversal near 

    the expected K + equilibrium potential. Finally, the relevance

    of  I K2 to pacemaking was strengthened by evidence that this

    component was modified by catecholamines and was

    responsible for the acceleration of rate caused by sympa-

    thetic stimulation (Hauswirth et al., 1968).

    The experimental evidence ruling in favour of the   bK +

    current-decay Q   hypothesis was therefore apparently incon-

    trovertible, and for over a decade, the   I K2   current was

    considered as one of the most typical K + cardiac compo-

    nents (Hille & Schwarz, 1978). Yet, as we now know, the

     I K2   interpretation and, consequently, the K +-current decay

    hypothesis were deeply incorrect. In the late 1970s and early

    1980s, a new set of experimental data appeared which led to

    the demonstration that the Purkinje fibre pacemaker current 

    was not at all an outward current activated on depolariza-

    tion, it was no less than just the opposite, an inward current activated on hyperpolarization (DiFrancesco, 1981a,b). The

    main reason for the incorrect interpretation was the presence

    of a   bfake Q    reversal potential close to the expected K +

    equilibrium potential during voltage-clamp hyperpolariza-

    tion, due to the superimposition of an inward activating

    current ( I f ) and a large inward decaying component caused

     by the depletion of K + ions from the extracellular clefts

    (DiFrancesco & Ojeda, 1980).

    A crucial finding that contributed essentially to this

    reintrepretation was the discovery of the   bfunny Q  current in

    the mammalian SAN. In 1979, the first detailed report of 

    this current appeared, describing its elementary properties

    and involvement in the generation and catecholamine-

    induced acceleration of SAN spontaneous activity (Brown

    et al., 1979). Among the unusual features which justified the

    name   bfunny Q    were a mixed Na+ and K +  permeability,

    activation on hyperpolarization, and very slow kinetics

    (Brown & DiFrancesco, 1980; Yanagihara & Irisawa, 1980).

    Although uncommon, these features were quite appropriatefor    I f    to function as an efficient    b pacemaker  Q    current 

    involved in the generation of diastolic depolarization and

    in rate control by   h-adrenergic stimulation. The Purkinje

    fibre’s pacemaker current   I K2   was then re-interpreted and

    shown to be identical to  I f  in the SAN based on several bits

    of evidence, including the abolishment of  I K2  reversal near 

    the K + equilibrium potential by the simple perfusion with

    Ba2+, a K + current blocker (DiFrancesco, 1981a). This latter 

    result was particularly impressive in that it unmasked the

    real inward nature of the Purkinje fibre’s pacemaker current 

    and allowed, for the first time, to visualize the conversion of 

     I K2  into an inward, hyperpolarization-activated current. Theinward ionic nature of   I K2  was confirmed by ion-substitu-

    tion experiments (DiFrancesco, 1981b), revealing a mixed

     Na+ and K +  permeability similar to that of the nodal  I f .

    The novel description of  I f  and the reinterpretation of the

    Purkinje fibre’s I K2 confirmed the identity of the pacemaker 

    mechanisms in the 2 cardiac tissues and paved the way to an

    integrated view of cardiac pacemaking in the different 

     pacing regions of mammalian heart. According to this view,

    the diastolic depolarization simply reflects the slow activa-

    tion of f-channels taking place when, at the termination of 

    an action potential, the membrane voltage enters the range

    of   I f  activation (DiFrancesco, 1985, 1993).

    2.2. Biophysical properties

    The typical features of the   I f   current include hyper-

     polarization-induced activation, slow kinetics of activation

    and deactivation, Na+ and K + ionic nature, and modulation

     by cAMP. Following its original description in the SAN,

    several studies identified   I f   also in atrial and ventricular 

    myocytes (Carmeliet, 1984; Zhou & Lipsius, 1992, 1993;

    Yu et al., 1993, 1995; Cerbai et al., 1994; Porciatti et al.,

    1997; Hoppe & Beuckelmann, 1998; Zorn-Pauly et al.,

    2004). We will now discuss these properties in the cardiac

    SAN and extend the analysis to other cardiac tissues whereappropriate.

    2.2.1. Voltage dependence of I   f    kinetics

    The voltage range where diastolic depolarization occurs

    in pacing myocytes is determined primarily by the voltage

    range of   I f  activation and by its kinetics. This explains, for 

    example, why diastolic depolarization in SAN myocytes

    occurs at more depolarized voltages than in Purkinje fibres,

    where the   I f  activation range is normally tens of millivolts

    more negative than in the SAN (see Table 1). Even within

    the SAN area, the  I f  activation range varies from cell to cell,

    shifting, for example, to more negative levels when moving

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   61

  • 8/20/2019 1-s2.0-S0163725805000252-main

    4/21

  • 8/20/2019 1-s2.0-S0163725805000252-main

    5/21

    at physiological voltages, but its function is lost with

    adulthood (Robinson et al., 1997). On the other hand, in an

    animal model of cardiac hypertrophy (spontaneous hyper-

    tensive rats), adult   I f    expression increases substantially

    relative to control animals (Cerbai et al., 1996).

    As well as by cAMP, which mediates autonomic-

    dependent modulation (see below), I f  kinetics and activationrange are modified by other mechanisms, such as auxiliary

    subunits (Qu et al., 2004), phosphorylation-dependent 

     processes (Chang et al., 1991; Accili et al., 1997), and

    interacting structural proteins that can affect the channel

    sub-membrane localization (Barbuti et al., 2004; Gravante et 

    al., 2004). These mechanisms may act synergistically to

    fine-tune the current activation range and kinetics and thus

    set the amount of current that can be recruited at appropriate

    times during cell activity. It is likely that additional, still

    unidentified processes play a role in the control of   I f activation range. Evidence for such processes comes, for 

    example, from the presence of current run-down duringwhole-cell recording in SAN myocytes, involving a leftward

    shift of the current activation curve (DiFrancesco et al.,

    1986), and the abrupt negative shift of the activation curve

    when membrane macro-patches are excised from the

    membrane during transition from cell-attached to inside– 

    out configuration (DiFrancesco & Mangoni, 1994). These

     phenomena suggest that cytoplasmic and/or cytoskeleton

    integrity is a necessary requirement for a correct channel

    functioning and that dilution by patch pipette or bath

    solution of the channel intracellular micro-environment 

    affects the channel properties. This view is strengthened by more recent evidence indicating the existence of a

    bcontext dependence Q    of pacemaker channel properties,

     based on the observation that identical pacemaker channel

    isoforms (see Section 3) have quantitatively different 

     properties when expressed in different expression systems

    (Qu et al., 2002).

    2.2.2. Kinetics

    Several types of mechanisms have been used in the

    literature to describe   I f    activation and deactivation pro-

    cesses, including single- (DiFrancesco & Noble, 1985;

    McCormick & Pape, 1990) and double-exponential Hodg-kin–Huxley kinetics ( Noble et al., 1989; van Ginneken &

    Giles, 1991; Demir et al., 1994), and more complex, non-

    Hodgkin–Huxley kinetics (DiFrancesco, 1984), often

    reflecting the different kind of accuracy required for specific

     Notes to Table 1:

    V 1/2   is the half-activation voltage and  t 1/2   is the half-activation time upon stepping to a given potential.

     Notes: (1) Isoprenaline 1–10   AM; (2) acetilcholine 1–10   AM; (3) saturating effect from Hill plot; (4) cAMP 10–100   AM; (5) room T; (6)  h1 stimulation

    (noradrenaline 1   AM),  h2 stimulation (isoprenaline 1   AM); (7) forskolin 10   AM; (8) carbachol 100   AM.

    References for table:

    1. Accili et al. (1996).   Pflugers Arch   431, 757

    2. Accili et al. (1997).  Am J Physiol   272, H1549

    3. Accili et al. (1997).  J Physiol   500, 6434. Bois et al. (1997).  J Physiol   501, 565

    5. Cerbai et al. (1996).  Circulation   94, 1674

    6. Cerbai et al. (1999).   Cardiovasc Res   42, 121

    7. Cerbai et al. (2001).  J Mol Cell Cardiol   33, 441

    8. Denyer et al. (1990).  J Physiol   428, 405

    9. DiFrancesco (1981).  J Physiol   314, 377

    10. DiFrancesco et al. (1986).  J Physiol   377, 61

    11. DiFrancesco et al. (1988).  J Physiol   405, 493

    12. DiFrancesco et al. (1989).   Science   243, 669

    13. DiFrancesco et al. (1991).   Pflugers Arch   417, 611

    14. DiFrancesco et al. (1991).  Nature   351, 145

    15. DiFrancesco et al. (1994).  J Physiol   474, 473

    16. Fares et al. (1998).  J Physiol   506, 73

    17. Frace et al. (1992).  J Physiol   453, 307

    18. Hagiwara et al. (1989).  J Physiol   409, 121

    19. Hoppe et al. (1998).   Cardiovasc Res   38, 788

    20. Hoppe et al. (1998).  Circulation   97, 55

    21. Mangoni et al. (2001).   Cardiovasc Res   52, 51

    22. Maruoka et al. (1994).  J Physiol   477, 423

    23. Porciatti et al. (1997).  Br J Pharmacol   122, 963

    24. Ranjan et al. (1998).  Biophys J   74, 1850

    25. Robinson et al. (1997).   Pflugers Arch   433, 533

    26. Shibata et al. (1999).  Br J Pharmacol   128, 1284

    27. Yasui et al. (2001).  Circ Res   88, 536

    28. Yu et al. (1993).  Circ Res   72, 232

    29. Yu et al. (1995).  J Physiol   485, 469

    30. Zaza et al.  J Physiol  491, 347

    31. Zhou et al. (1992).  J Physiol   453, 503

    32. Zhou et al. (1993).   Pflugers Arch   423, 442

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   63

  • 8/20/2019 1-s2.0-S0163725805000252-main

    6/21

    analysis. Detailed investigation reveals that several kinetic

    features of  I f  cannot be accommodated by simple Hodgkin– 

    Huxley type of gating and require complex multistate

    kinetic modelling based on the existence of distinct 

    bdelaying Q    and proper    bgating Q    processes   (DiFrancesco,

    1984). A similar approach has been used to describe the

    kinetics of HCN channels. As is further discussed below(see Section 3.2), an   ballosteric Q   dual voltage and cAMP-

    dependent kinetic model, which views HCN channels as

    tetramers carrying voltage sensors that can be gated

    individually by voltage and undergoes concerted open/ 

    closed allosteric transitions, can reproduce most of their 

    kinetic properties   (DiFrancesco, 1999; Altomare et al.,

    2001).

    2.2.3. Ionic nature

    Early evidence on the ion permeation properties of 

     pacemaker channels derives from experiments in Purkinje

    fibres and in isolated rabbit SAN myocytes   (DiFrancesco,1981b; DiFrancesco et al., 1986). Values of the reversal

     potential measured in these experiments were in the range of 

    10 to   20 mV, pointing to a mixed ionic permeability.Ionic substitution experiments indeed identified Na+ and K +

    ions as the physiological carriers of the current, with a Na+/ 

    K +  permeability ratio of about 0.27   (DiFrancesco, 1981 b;

    Frace et al., 1992). The conductance of f-channels was also

    shown to increase with external K + concentration (DiFran-

    cesco, 1981b) in a way similar to that of other K +-permeable

    channels   (Hille, 2001). This effect is physiologically

    relevant since a higher external K + concentration tends to

    decrease, by depolarization, the fraction of f-channels

    activated at the termination of an action potential and thus

    decrease substantially the slope of diastolic depolarization

    and heart rate; an increase in the   I f  conductance may thus

    compensate, at least partially, for the bradycardic effect of 

    hyperkalemia.

    2.3. Contribution to automaticity

    Several direct and indirect experimental observations

    substantiate the key role of  I f   in cardiac pacemaking. These

    include the actual ionic and kinetic properties of   I f mentioned above, which are particularly appropriate for 

    the generation of the slow diastolic depolarization in the pacemaker range of voltages, and the correlation between I f expression and the presence of spontaneous activity in

    cardiac cells. This correlation is evident by simply compar-

    ing different cell types in an adult mammalian heart, where

     pacemaking cells such as SAN or atrio-ventricular cells do,

    while atrial and ventricular myocytes, quiescent if not 

    stimulated, do not normally express   I f    at physiological

    voltages. Further, the expression of   I f    correlates with

    spontaneous activity in the developing newborn chick 

    (Satoh & Sperelakis, 1993) or mammalian  (Robinson   et 

    al., 1997; Yasui et al., 2001) ventricular cells, where  I f  and

    spontaneous activity are simultaneously present at early

    stages and disappear together at later stages of development,

    and in zebrafish heart, where a bradycardic mutant was

    found to express, among all current systems investigated,

    only a reduced   I f  in cardiac myocytes  (Baker et al., 1997;

    Warren et al., 2001).

    A more recent investigation based on the overexpression

    of HCN2 channels, one of the isoforms composing native f-channels (see Section 3), in embryonic rat ventricular 

    myocytes has shown a large increase of the rate of 

    spontaneous activity correlated to the expression of pace-

    maker channels  (Qu et al., 2001). Further, the overexpres-

    sion of a nonfunctional HCN2 isoform induced dominant-

    negative suppression of native pacemaker channel activity

    and strongly depressed pacing in newborn ventricular 

    myocytes (Er et al., 2003).

    The use of molecules specifically inhibiting   I f    also

     provides evidence for a strict correlation between I f  current 

    availability and the presence of spontaneous activity in

     pacemaker cells. For example, drugs which block f-channelswith a high degree of selectivity, such as UL-FS49

    (zatebradine), ZD7288, and S16257 (ivabradine), act as

     pure   bheart rate-reducing Q  agents by decreasing the slope of 

    diastolic depolarization and, as such, have a potential for 

    therapeutic use in cardiac diseases whose prognosis can be

    alleviated by moderate bradycardia (DiFrancesco & Camm,

    2004). Heart rate-reducing drugs slow heart rate in a

    concentration-dependent way, reflecting the proportionality

     between   I f   inhibition and slowing effect. This aspect is

    further developed below (see Section 4).

    Finally, a direct demonstration of the  I f  role in pacemaker 

    generation and control is illustrated by the action of 

    autonomic neurotransmitters. As is outlined in the Section

    2.4, opposite chronotropic effects exerted by sympathetic

    and parasympathetic stimuli, particularly at low levels of 

    autonomic activity, are mediated by cAMP-dependent 

    modulation of   I f , which underlies rate changes by the

    modification of the slope of diastolic depolarization.

    2.4. Autonomic modulation

    The mammalian SAN region is richly innervated by the

    autonomic nervous system, which exerts a direct control of 

     pacemaker activity. Sympathetic stimulation accelerates,

    and parasympathetic stimulation slows, heart rate actingthrough h-adrenergic and muscarinic receptors, respectively.

    The original description of  I f  already indicated the involve-

    ment of this current system not only in the generation, but 

    also in the positive chronotropic effect of adrenaline (Brown

    et al., 1979). Subsequent work showed that this action is due

    to a shift of the  I f  activation curve to more positive voltages

    (DiFrancesco et al., 1986; DiFrancesco & Mangoni, 1994).

    These results were in agreement with previous indications

    on I K2 in Purkinje fibres: despite the incorrect interpretation

    of   I K2   as a pure K + current, investigation of the action of 

    adrenaline had provided evidence for a rightward shifting

    action of adrenaline on the   I K2  activation curve (Hauswirt h

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7964

  • 8/20/2019 1-s2.0-S0163725805000252-main

    7/21

    et al., 1968), an effect mediated by increased intracellular 

    cAMP (Tsien et al., 1972).

    How does therefore adrenaline accelerate the heart? The

    adrenaline-induced shift of the   I f  activation curve to more

     positive voltages, caused by increased cAMP levels

    subsequent to   hAR-stimulation of adenylate cyclase,

    increases the degree of steady-state current activated at any potential (within the activation range), thus increasing

    the current availability during diastolic depolarization and,

    as a consequence, the rate of development of diastolic

    depolarization itself (DiFrancesco, 1993).

    The realization of the three basic steps of the mechanism

     by which sympathetic activity   increases I f  and accelerates

    cardiac rate, i.e. (1)   hAR stimulation and coupling to the

    stimulatory G-protein Gs, (2) activation of adenylate cyclase

    and cAMP increase, and (3) activation of f-channels and rate

    acceleration, naturally posed the question whether this

    mechanism could operate in the opposite direction, i.e. to

    decrease I f , thereby slowing the rate. Indeed, later work showed that, as well as being activated by adrenaline,   I f   is

    also inhibited by acetylcholine (DiFrancesco & Tromba,

    1987). The stimulation of muscarinic receptors was shown

    to shift the   I f   activation curve to more negative voltages,

    according to a process exactly opposite to that of   hAR 

    stimulation, and involving Gi-protein-dependent inhibition

    of cAMP synthesis (DiFrancesco & Tromba, 1988a,b). ACh

    induced a negative shift of the   I f   activation curve without 

    changing the fully activated   I / V   relation, indicating a

    modification of gating, but not of the channel conductance

    (DiFrancesco & Tromba, 1988a).

    The results above suggested a major role for   I f   in the

    negative chronotropic action of vagal activity. However,

    early work had identified an ACh-activated K + current 

    ( I K,ACh) as the main process underlying slowing (Sakmann

    et al., 1983). What was the relative importance of the 2

     processes? This issue was resolved with the demonstration

    that ACh acts to inhibit   I f    at much (20-fold) lower 

    concentrations than those required to activate   I K,ACh, and

    that concentrations of ACh inhibiting   I f   were effective to

    slow rate (DiFrancesco et al., 1989). The comparison

     between ACh action on   I f    and   I K,ACh   was the first 

    demonstration that cardiac rate slowing by low ACh doses

    (i.e. moderate vagal activity) is due to muscarinic-induced  I f 

    inhibition, caused by decreased cAMP levels and a negativeshift of the I f  activation curve, and the consequent reduction

    of the rate of diastolic depolarization.

    Data collected by experimentation in the SAN had

    therefore provided evidence for a key role of   I f   in both

    the adrenergic and cholinergic modulation of cardiac rate

    mediated by the increase and decrease, respectively, of 

    intracellular cAMP (DiFrancesco et al., 1986; DiFrancesco

    & Tromba, 1987, 1988a,b). This evidence was in accord-

    ance with the established notion that, in cardiac cells, the

    stimulation of   h-adrenergic receptors activates, and of 

    muscarinic receptors inhibits, adenylate cyclase and cAMP

     production; it also identified a role for cAMP, which, by

     promoting a depolarizing shift of the  I f  activation curve, acts

    as a second messenger in f-channel modulation. The mode

    of action of cAMP was investigated by the use of inside–out 

    macro-patch analysis in SAN cells and led to the surprising

    finding that f-channels are activated by cAMP through the

    direct  binding of cAMP molecules to channels, and not by

     phosphorylation, as it occurs with other channels such as,for example, the L-type Ca2+ channels (DiFrancesco &

    Tortora, 1991). This was the first evidence that funny

    channels, as well as being voltage-gated, shared with

    another class of channels, the cyclic nucleotide-gated

    (CNG) channels of sensory neurons, the property of being

    directly gated by cyclic nucleotides.

    2.5. Single f-channel recording 

    There are few reports of single-channel recording for the

     pacemaker current, performed in isolated SAN myocytes

    (DiFrancesco et al., 1986; DiFrancesco & Mangoni, 1994).This limited amount of data likely reflects the intrinsic

    difficulty of measuring tiny, single-channel events (order of 

    0.05–0.1 pA) given the small single-channel conductance of 

    f-channels (about 1 pS,   DiFrancesco, 1986). Despite the

     paucity of data, single f-channel recording has allowed us to

    understand features of the molecular mechanisms under-

    lying f-channel dual gating by voltage hyperpolarization

    and cAMP, and the neurotransmitter-induced f-channel

    modulation.

    For example, single-channel data have shown that the

    hAR-induced current activation is due to an increased open

    channel probability rather than to an increased single-

    channel conductance, in agreement with the action of  hAR 

    stimulation in whole-cell conditions, which leads to a shift 

    of the current activation curve to more positive voltages

    without modification of the fully activated   I / V   relation

    (DiFrancesco et al., 1986). cAMP was shown to activate f-

    channels by binding directly to the intracellular aspect of the

    channel (DiFrancesco & Tortora, 1991), which, in accord-

    ance to the whole-cell data, resulted in an increase of single-

    channel open probability, with no changes of channel

    conductance (DiFrancesco & Mangoni, 1994).

    3. Molecular determinants of the   I f  current

    3.1. Hyperpolarization-

    activated, cyclic nucleotide-gated clones

    A major progress in the understanding of the molecular 

     basis of the properties of pacemaker channels was achieved

    with the cloning of the Hyperpolarization-activated, Cyclic

     Nucleotide-gated (HCN) family of channels in the late

    1990s (Zagotta et al., 2003). The search for the pacemaker 

    channel clone had been a major task in several laboratories

    for decades, but no direct approach yielded useful results

    until, as it sometimes happens, the crucial step towards

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   65

  • 8/20/2019 1-s2.0-S0163725805000252-main

    8/21

    cloning was made by chance while looking for proteins

    interacting with the SH3 domain of the neural form of Src

    (Santoro et al., 1997). The sequence thus identified was

    homologous to eag K + and cyclic nucleotide-gated (CNG)

    channels and, although not immediately, was later used to

    identify a complete ORF and recognized as a pacemaker 

    channel subunit based on its biophysical properties  (Santor oet al., 1998). Other members of the same family were soon

    cloned in mammalian and non-mammalian tissues (Gaug  et 

    al., 1998; Ludwig et al., 1998, 1999b; Ishii et al., 1999;

    Seifert et al., 1999; Vaccari et al., 1999). In mammals, four 

    isoforms have been cloned (HCN1-4); human genes coding

    for the 4 isoforms are all located in different chromosomes

    (HCN1: 5p12; HCN2: 19p13.3, Vaccari et al., 1999; HCN3:

    1q22; HCN4: 15q24-25, Seifert et al., 1999). In Fig. 1, the

    alignment of the four human HCN isoforms is shown.

    Based on their sequence, HCN channels are classified as

    members of the superfamily of voltage-gated K + (Kv) and

    CNG channels; as such, they appear to have a tetramericcomposition   (Ulens & Siegelbaum, 2003; Zagotta et al.,

    2003), and are characterized by the presence of 6 trans-

    membrane domains (S1–6) with a voltage sensor located in

    the positively charged S4 domain, the GYG pore sequence

    typical of K +-permeable channels, and a cyclic nucleotide-

     binding domain (CNBD) homologous to that of CNG

    channels located in the C-terminus.

    Sequence alignment revealed that the HCN   bcore Q 

    region, comprising the transmembrane and CNB domains,

    is highly conserved among the different isoforms (z80%

    identity), while sequences diverge at the N- and C-termini,

    suggesting that the terminal regions are responsible for some

    of the differences in the biophysical properties among

    isoforms   (Viscomi et al., 2001). Indeed, the properties of 

    different isoforms differ quantitatively. For example, the

    activation/deactivation kinetics of HCN2 are faster than

    those of HCN4 and slower than those of HCN1; typical

    values of activation time constant at  95 mV are 0.11, 1.13,and 2.52 s for HCN1, HCN2, and HCN4, respectively

    (Altomare et al., 2001). HCN2 has a more negative

    activation threshold than either HCN1 or HCN4 does;

    typical values of the half-maximal activation voltage are

    73,   81, and   92 mV for HCN1, HCN4, and HCN2,respectively, when channels are expressed in HEK293 cells

    (Accili et al., 2002), but several conditions may alter significantly these values (see below). Finally, as with

    native f-channels, cAMP activates HCN channels by

    shifting the activation curve to more positive voltages, but 

    maximal shifts vary among isoforms, with HCN1 being

    much less responsive (range 4.3–5.8 mV) than either HCN2

    (range 16.9–19.2 mV) or HCN4 (range 11.1–23 mV; Ishii et 

    al., 1999; Ludwig et al., 1999b; Seifert et al., 1999; Moroni

    et al., 2000; Viscomi et al., 2001; Wainger et al., 2001;

    Wang et al., 2001; Altomare et al., 2003; Zagotta et al.,

    2003). These differences appear to be determined by

    differential inhibitory interactions of the C-termini with

    the core transmembrane domains in the various isoforms,

    more than by a variable cAMP binding affinity to the CNBD

    (Wang et al., 2001). The properties of the HCN3 isoform

    have been only partially investigated so far; HCN3 kinetics

    are intermediate between those of HCN2 and HCN4

    (Moosmang et al., 2001).

    The different kinetic and modulatory properties of native

    currents in various regions of the heart and brain  (DiFran-cesco, 1993; Pape, 1996) may thus simply reflect a different 

    tissue distribution of HCN isoforms. However, simple

    electrophysiological analysis is not sufficient to reveal the

    isoform composition of native channels, since several

    conditions can contribute to modify the channel properties.

    For example, native channels can be formed by heteromul-

    timers of different isoforms, with properties intermediate

     between those of individual components   (Chen et al.,

    2001b; Ishii et al., 2001; Ulens & Tytgat, 2001; Xue et 

    al., 2002; Altomare et al., 2003); also, HCN activity can be

    modified by interaction with auxiliary subunits such as

    MiRP1  (Qu et al., 2004) or with scaffold proteins such asfilamin-A (for HCN1, Gravante et al., 2004), or by specific

    subcellular compartmentation such as the caveolar compart-

    mentation of HCN4 in SAN cells  (Barbuti et al., 2004);

    finally, the expression of a given isoform may yield

    quantitatively different biophysical properties according to

    whether the isoform is expressed in heterologous or in

    homologous expression systems, suggesting that a   bcontext-

    dependent  Q   modulation occurs  (Qu et al., 2002). A more

    detailed assessment of tissue distribution of specific HCN

    isoforms therefore requires message and protein detection

    assays such as Northern blot, RNase protection assay, and

    immunolabelling. In heart HCN1, HCN2, and HCN4 are all

    expressed, with HCN4 being the major component in the

     pacemaker region, although low expressions of HCN1 and

    HCN2 have also been reported (Santoro et al., 1998; Shi et 

    al., 1999; Moroni et al., 2001).

    3.2. Structure–function relationships for 

    hyperpolarization-activated, cyclic nucleotide-gated channels

    Among the peculiar properties of   bfunny Q  channels are

    the dual dependence on voltage and cAMP, and the

    hyperpolarization-induced activation   (DiFrancesco, 1999).

    These properties are typical of HCN channels and are not 

    shared by the other members of the superfamily of Kv andCNG channels, which are gated by voltage only and cAMP

    only, respectively; also, Kv channels open on depolariza-

    tion, which raised the intriguing question of how similar 

    voltage sensor (S4) sequences could give rise to opposite

    voltage dependences of gating.

    The effect of cAMP is to shift the  I f  activation curve (i.e.

    the single-channel open-probability curve) to more positive

    voltages and to accelerate activation and slow deactivation

    kinetics (DiFrancesco et al., 1986; DiFrancesco & Tortor a,

    1991; DiFrancesco & Mangoni, 1994). How does cAMP

    cause these changes? The presence of a hyperpolarizing

    shift of voltage dependence suggests that the gating

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7966

  • 8/20/2019 1-s2.0-S0163725805000252-main

    9/21

    mechanism operated by voltage hyperpolarization and the

    one operated by cAMP are the same, and that HCN channels

     behave in the presence of a given cAMP concentration

    simply as if they were experiencing a stronger voltage drop.

    Indeed, the shifting action of cAMP can be explained by

    assuming that channels behave according to an allosteric

    model of channel kinetics (DiFrancesco, 1999; Altomare et 

    al., 2001). In this model, HCN channels are described as

    Fig. 1. Amino acid sequence alignment of the 4 human HCN isoforms. Transmembrane domains S1 to S6, pore helix P and the CNBD are indicated.

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   67

  • 8/20/2019 1-s2.0-S0163725805000252-main

    10/21

    tetramers whose four subunits undergo simultaneous,

    concerted open/closed transitions according to an allosteric

    reaction, and where each of the four subunits carries a

    voltage sensor which is independently gated by voltage. The

    channel open probability   po   can then be described by a

    modified Boltzmann equation:

     po ¼  1=   1 þ exp   V    V 1=2   s

    =v 

      ð1Þ

    where  V   is voltage,  V 1/2  is the half-activation voltage,   v   is

    the inverse slope-factor, and   s   is a cAMP concentration-

    dependent term which represents the effects of cAMP on the

    channel open/closed equilibrium   (DiFrancesco, 1999).   s   is

    clearly a shift of the voltage dependence of  po, showing that 

    the allosteric model assumptions are able to correctly

    interpret the action of cAMP experimentally observed.

    The shift depends on the cAMP concentration according

    to the equation:

     s   cAMP½ ð Þ ¼ v  ln 1 þ  cAMP½ = K o

    ð Þ=   1 þ   cAMP½ = K c

    ð Þð Þ

    ð2Þ

    where   K o   and   K c   are the dissociation constants of cAMP

     binding to the channel in the open and closed states,

    respectively. As illustrated in  Fig.  2, this relation accounts

    quantitatively for the observed sigmoidal dependence of the

    shift on cAMP concentration (DiFrancesco & Tortora, 1991;

    DiFrancesco, 1999). It is interesting to note that according

    to this interpretation, the action of cAMP derives simply by

    the assumption that cAMP molecules have a higher binding

    affinity to open than to closed channel states ( K ob K c).

    Several structure–function studies have addressed these

    aspects by investigating the HCN channel domains

    involved in hyperpolarization-dependent and cAMP-

    dependent gating.

    3.2.1. Structures involved in voltage-dependent gating 

    HCN channels share several structural similarities with

    voltage-gated K + (Kv) channels, among which a positively

    charged S4 domain, the putative voltage sensor which, in

    HCN channels, includes 10 basic residues whose mutation

    strongly affects the channel voltage dependence  (Chen et 

    al., 2000; Vaca et al., 2000); however, while Kv channels

    open on depolarization, HCN channels open on hyper-

     polarization. This difference could be due to an   binverted Q 

    movement of the voltage sensor of HCN vs. Kv channels in

    response to the same voltage change, or to an   binverted Q 

    coupling between S4 movement and channel gating. Thelatter possibility appears more likely since cysteine acces-

    sibility experiments have shown that, like in Kv channels,

    hyperpolarization induces an inward movement of the S4

    segment in HCN channels   (Mannikko et al., 2002).

    Although details of the mechanism coupling S4 to gating

    are still not fully understood, there are indications for the

    involvement of the S4–S5 linker. These studies are based on

    the observation that a point mutation in the HERG K +

    channel S4–S5 linker (D540K) is able to induce hyper-

     polarization-dependent activation (Sanguinetti & Xu, 1999).

    A thorough investigation of the amino acid residues of the

    S4–S5 linker in HCN2 channels by an alanine-scanning

    mutagenesis supports a critical role of this linker in

    hyperpolarization-induced opening   (Chen et al., 2001a).

    The mutation of most of these residues was shown to

    modify the channel voltage dependence and kinetics;

    specifically, the mutation of E324, Y331, and R339 induced

    a disruption of channel closure. The location in the S4–S5

    linker makes it unlikely that these residues are a structural

     part of the gating machinery and suggests that they act as

    coupling elements between the S4 voltage sensor and the

    channel gate. The region at the boundary between S4 and

    the S4–S5 linker contains a histidine residue and is

    important also in the pHi   sensitivity of HCN channels

    (Zong et al., 2001).Differences in the kinetics of HCN isoforms can also

     be exploited to gain insight into the gating mechanisms.

    This approach takes advantage of the fact that, as

    mentioned above, different isoforms can form heteromul-

    timeric channels with properties intermediate between

    those of homomeric channels and that chimeric channels

    can be constructed by swapping various domains between

    components.

    The replacement of the HCN4 C-terminus, but not of the

     N-terminus, by corresponding C- or N-termini from HCN1

    led to a substantial acceleration of channel kinetics and loss

    of cAMP sensitivity, demonstrating a role of the C-terminus

    Fig. 2. Activation of f-channels by cAMP as explained according to an

    allosteric model of channel gating. (A) The f-channel open probability

    curve ( P o) measured by the voltage-ramp protocol in an inside-out macro-

     patch in control conditions and during perfusion with 10   AM cAMP. The

    action of cAMP is to shift the  P o curve to more positive voltages (by about 

    14 mV in this case), thereby increasing the current availability for the

    generation of diastolic depolarization, hence steepening its rate. (B) Dose– 

    response relationship of the cAMP-induced shifts (squares, mean F SEM)

    from 47 macro-patches. The full line is a best fit to data performed with Eq.

    (2), yielding the values  K o = 0.0578  AM and K c = 0.5416  AM. (C) Cartoon

    illustrating the basic scheme of channel modulation by cAMP. Panels  A  and

    B  are modified from DiFrancesco (1999, with permission).

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7968

  • 8/20/2019 1-s2.0-S0163725805000252-main

    11/21

    in voltage- and cAMP-dependent gating (Viscomi et al.,

    2001). These results were in agreement with previous data

    on native f-channels of SAN myocytes demonstrating that 

    the perfusion of pronase onto the intracellular side of 

    inside–out patches, acting by the cleavage of internal

     portions of channels, causes a dramatic shift of the channel

    open probability curve to more positive voltages (+56 mV)and eliminates the cAMP dependence (Barbuti et al., 1999).

    The above experiments introduced the concept of a basal

    inhibitory action exerted by a proteolysis-sensitive internal

    domain (possibly in the C-terminus) on channel gating,

    which could be removed by either hyperpolarization or 

    cAMP binding, a concept that was later confirmed by a

    study using HCN1 and HCN2 CNBD-deletion mutants

    (Wainger et al., 2001) or chimeras (Wang et al., 2001;  see

     below).

    HCN channel kinetics can be modified by mutations in

    several channel regions, including S1, the S1–S2 linker, S6,

    and the S3–S4 linker, in the latter probably by a surface-charge action (Ishii et al., 2001; Henrikson et al., 2003).

    Residues in the S4–S5 linker appear to be specifically

    relevant to the coupling of S4-mediated voltage sensing and

    channel activation in HCN2 (Chen et al., 2001a), while the

    C-linker (i.e. the region linking S6 to the CNBD) is

    important in mediating an inhibitory action of the CNBD

    on channel activation through an interaction with core

    transmembrane regions (Wang et al., 2001).

    Although the outermost pore rim affects the gating of 

    HCN channels (Xue & Li, 2002), the voltage-dependent 

    gate appears to be located at the intracellular side of the

    channel (Shin et al., 2001). This conclusion is based on

    evidence that the HCN channel blocker ZD7288, applied

    from the intracellular side, can enter and leave the channel

     pore only when it is open. A similar reasoning applies to the

     blocking action of other HCN channel blockers such as

    ivabradine (Bucchi et al., 2002).

    The present data therefore suggest that gating is a

    complex mechanism involving interactions among several

    regions, which include the CNBD, the C-linker, the S4–S5

    linker, and the core transmembrane regions.

    3.2.2. Structures involved in cAMP-dependent gating 

    cAMP exerts its modulatory effects by directly binding to

    HCN channels (DiFrancesco & Tortora, 1991), and its binding is more likely to occur when channels are in the

    open state. The binding of cAMP to the channel thus

    stabilizes the open configuration by partial removal of a

    tonic inhibition, thereby shifting the channel distribution

    equilibrium towards the open state (DiFrancesco, 1999).

    The region involved in ligand binding and functional

    transfer of cAMP-mediated channel gating is located in the

    C-terminus (Viscomi et al., 2001; Wang et al., 2001). The C-

    terminus can be subdivided into different regions, including

    a central cyclic nucleotide-binding domain (CNBD), and a

    C-linker (about 80 aa long) that connects the C-terminal part 

    of S6 to the CNBD.

    Homology with the CNBD structure of both the bacterial

    catabolite-activating protein and the regulatory subunit of 

    cAMP-dependent protein kinase, and the more recent data

    from X-ray crystallography of the C-terminus of HCN2,

    have contributed to a more detailed identification of 

    functionally relevant sub-domains of the C-terminus and

    specifically of the CNBD (Weber & Steitz, 1987; Su et al.,1995; Wainger et al., 2001; Zagotta et al., 2003).

    The CNBD structure comprises a   h-roll sub-domain,

    acting as a constitutive inhibitor of the channel, and an   a-

    helix located at the C-terminal end (termed C-helix), which

    contributes to the interaction with the purine ring of cAMP.

    cAMP binds with a greater affinity to the open, rather than

    the closed state of the channel, and the mutation of residues

    in the C-helix of HCN2 and CNG channels uncouple ligand

     binding from gating modulation (Varnum et al., 1995;

    Matulef et al., 1999; Wainger et al., 2001; Zagotta et al.,

    2003).

    The deletion of the C-helix of HCN1 and HCN2abolished the cAMP response, but did not alter normal

    gating (Wainger et al., 2001). The deletion of the whole

    CNBD shifted the activation curves of HCN1 and HCN2 by

    different amounts, such that the half-activation voltages of 

    the truncated channels were nearly coincident. These data

     provide a quantitative confirmation of the idea that cAMP

    removes a basal inhibitory mechanism operated by the C-

    terminus on HCN channel activation, originally proposed

    for native f-channels of SAN myocytes (Barbuti et al.,

    1999), and shows that the reduced cAMP sensitivity of 

    HCN1 can be simply explained by a reduced basal

    inhibition of the CNBD on gating for HCN1 channels

    (Wainger et al., 2001).

    Additional information revealed by crystallographic

    analysis provided further ground to the evidence for a

    tetrameric nature of HCN channels and revealed that the C-

    linkers are responsible for subunit–subunit interactions

    (Zagotta et al., 2003).

    3.3. Hyperpolarization-activated, cyclic

    nucleotide-gated isoforms in sinoatrial node tissue

    The identification of the molecular components of native

     pacemaker channels in different tissues, and specifically in

    the cardiac pacemaker region, is important not only for amore detailed understanding of channel gating and inter-

    action with other components, but also for pharmacological

    and genetic therapy approaches.

    As a first step in the identification of the molecular 

    composition of native f- channels, their properties can be

    compared with those of individual HCN isoforms. However,

    this comparison does not always enable the identification of 

    the channel composition. For example, the kinetics and

    cAMP modulation of HCN2 are not too dissimilar from

    those of native f-channels of SAN myocytes, but the

     position of the activation curve is far too negative relative

    to that of the   I f   current (see Section 3.1); similarly, f-

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   69

  • 8/20/2019 1-s2.0-S0163725805000252-main

    12/21

    channels are unlikely to be homotetramers of either HCN1

    or HCN4 subunits only, since HCN1 has too fast kinetics

    and weak cAMP sensitivity, while the HCN4 kinetics are

    too slow   (Altomare et al., 2003). Complementary ap-

     proaches, based on RNase protection assay, Northern

     blotting, and in situ hybridization techniques, indicate that,

    in the SAN region, HCN4 is the prevalent isoform (about 80% of the total HCN signal), while HCN1 and HCN2 are

    expressed weakly; also, the expression of HCN1 and HCN2

    appears to be species dependent (Shi et al., 1999; Ishii et al.,

    2001; Moosmang et al., 2001; Moroni et al., 2001; Stieber et 

    al., 2003).

    As mentioned above, several mechanisms may contribute

    to modify the functional properties of HCN channels,

    including the heteromerization of different isoforms, as

    well as interaction with auxiliary proteins and with the

    intracellular environment. Studies using heterologous coex-

     pression of different isoforms (HCN1 and HCN2 or HCN1

    and HCN4), individually or in concatenated form, or dominant-negative channel constructs, together with studies

    using the yeast 2-hybrid system, clearly revealed the ability

    of HCN isoforms to form heterotetrameric complexes, with

     properties distinct from those of individual isoforms (Chen

    et al., 2001b; Ulens & Tytgat, 2001; Proenza et al., 2002;

    Xue et al., 2002; Altomare et al., 2003; Er et al., 2003;

    Much et al., 2003). Heteromeric channel constructs exist in

    vivo  (Much et al., 2003), demonstrating the physiological

    relevance of heteromerization. It is interesting to note that 

    HCN2–HCN3 heteromers cannot be formed, suggesting that 

    specific channel domains may be required for the stabiliza-

    tion of subunit interactions   (Much et al., 2003). The

    tetramerization domains must be different from the N-

    terminal domains shown to underlie channel assembly and

    trafficking to the plasma membrane, since the latter are

    conserved among HCN isoforms (Proenza et al., 2002; Tran

    et al., 2002).

    The hypothesis that f-channels of rabbit SAN myocytes

    are composed of HCN4–HCN1 heteromers was investigated

     by constructing tandem HCN4–HCN1 constructs, with the

    HCN4 C-terminus covalently linked to the HCN1 N-

    terminus   (Altomare et al., 2003). The expression of this

    construct generated currents with kinetics similar to those of 

     I f , but did not reproduce the same voltage dependence of 

    activation and cAMP sensitivity. A normal cAMP sensi-tivity could be obtained, on the other hand, with the mirror 

    tandem construct HCN1–HCN4, with the HCN4 N-termi-

    nus now linked to the HCN1 C-terminus, suggesting that a

    restricted mobility of the C-terminus reduces the response to

    cAMP. The position of the activation curves of HCN

    constructs was, however, always more negative than that of 

     I f . These data, together with the observation of a 4:1 ratio of 

    HCN4 vs. HCN1 mRNA (Shi et al., 1999), indicate that f-

    channels may indeed be heteromultimers of HCN4 and

    HCN1, with a prevalence of HCN4 (Altomare et al., 2003),

     but that a   bcontext  Q -dependent mechanism also likely

    modulates the channel properties (Qu et al., 2002).

    Most voltage-dependent ionic channels are composed by

    a main pore-forming   a-subunit, able to generate functional

    channels, and by additional accessory   h-subunits.   h-

    subunits do not have channel-like properties per se but 

    interact functionally with   a-subunits and modulate the

     properties of the channel. The minK-related protein 1

    (MiRP1), a single transmembrane-spanning protein highlyexpressed in the SAN, acts as a modulatory  h-subunit of the

    K + channels responsible for the currents   I Kr ,   I Ks, and   I to(Abbott et al., 1999; Tinel et al., 2000; Zhang et al., 2001).

    The effects of MiRP1 on HCN currents heterologously

    expressed have been investigated, but the results in the

    literature are not uniform.   Yu et al. (2001)   coexpressed

    MiRP1 with HCN1 or HCN2 in   Xenopus   oocytes and

    observed an enhanced protein expression and acceleration

    of the activation kinetics.   Altomare et al. (2003), on the

    other hand, did not have evidence for an effect of MiRP1 on

    either HCN1, HCN4, or a tandem HCN4–HCN1 construct 

    when coexpressed in HEK293 cells. When MiRP1 modu-lation of HCN4 currents was analyzed by the expression in

    CHO cells and in  Xenopus  oocytes, the results included an

    increase of maximal current, a slowing of activation

    kinetics, and a negative shift of the voltage dependence of 

    activation (Decher et al., 2003). The reason for the observed

    differences is still unclear and awaits further investigation.

    More recently,   Qu et al. (2004)   investigated the effect 

    of MiRP1 on HCN2 channels homologously expressed in

    neonatal rat ventricle myocytes, thus using a physiological

    cell substrate. In these experiments MiRP1 strongly

    enhanced (4-fold) the maximal conductance and acceler-

    ated the rates of activation and deactivation; also, structural

    interaction was demonstrated by co-immunoprecipitation.

    A possible interpretation of the results with homologous

    vs. heterologous channel expression is therefore that the

    interaction between HCN and   h-subunits depends on the

    intracellular environment according to still unknown

    mechanisms.

    4. f-Channel blockers

    Molecules interfering specifically with ion channels are

    important tools in the characterization of their structural and

    functional properties. Ion channel blockers are moleculeswhich inhibit ionic flow through channel pores; when their 

    action is specific for a given channel, their use allows the

     pharmacological dissection of the contribution of this

    channel to cellular electrical activity. Ion channels dysfunc-

    tions are the basis of several types of diseases known as

    channelopathies, and the development of drugs specifically

    targeting ion channels has naturally become a rapidly

    growing concern of pharmacological research. It is not a

    case that a large number of drugs on the market today are

    targeted to modify ion channel functions.

    The relevance of   I f   to cardiac pacemaking makes it an

    obvious target for the search of drugs able to interfere

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7970

  • 8/20/2019 1-s2.0-S0163725805000252-main

    13/21

    selectively with f-channels and thus control the pacemaker 

    function. If the concept of   I f   activation as the primary

    mechanism responsible for the generation and autonomic

    modulation of spontaneous activity is true, then a specific

    alteration of   I f   is expected to affect, via a modification of 

    the slope of diastolic depolarization, the cardiac chrono-

    tropic state, without modifying other cardiovascular  parameters.

    In the last few years, substances able to act as specific

     blockers of the pacemaker current have been developed.

    These molecules, originally known as   b pure bradycardic Q 

    agents and termed today as   bheart rate-lowering Q  agents, are

     potentially important therapeutic agents able to induce rate

    slowing without the inotropic side effects typical of drugs

    currently used to slow heart rate, such as Ca2+ antagonists or 

    h-blockers (Yusuf & Camm, 2003; DiFrancesco & Camm,

    2004). Because of the early incorrect interpretation of the

    origin of the pacemaker depolarization and of the pacemaker 

    current (see Section 2.1 above), heart rate-lowering agentswere sometimes thought to act as Ca2+-channel blockers

    (Doerr & Trautwein, 1990) and were shown to be specific  I f  blockers only based on more careful experimentation (Van

    Bogaert & Goethals, 1987; DiFrancesco, 1994).

    Moderate reduction of heart rate is therapeutically

     beneficial in a variety of cardiovascular conditions, includ-

    ing chronic angina, ischaemic heart disease, and heart 

    failure. A lower heart rate decreases oxygen demand and

    improves myocardial perfusion during a prolonged diastole,

    thus improving myocardial oxygen balance. Also, the

    concept of heart-rate reduction as a valid therapeutic

    intervention in a range of cardiovascular diseases is

    validated by several studies showing a link between

    elevated heart rate and mortality in patients with coronary

    heart disease and in the general population (DiFrancesco &

    Camm, 2004).

    Specific heart rate-lowering agents include ST567

    (alinidine), UL-FS49 (zatebradine), ZD-7288, and S16257

    (ivabradine; Van Bogaert & Goethals, 1987; BoSmith et al.,

    1993; DiFrancesco, 1994; Gasparini & DiFrancesco, 1997;

    Bucchi et al., 2002; Yusuf & Camm, 2003) and their 

     properties are discussed below.

    4.1. Alinidine (ST567)

    The term   bspecific bradycardic agent  Q    was originally

     proposed as a way to define drugs acting directly on

     pacemaker function and able to decrease heart rate at doses

    having minimal or no side effects (Kobinger, 1985).

    Alinidine ( N -allyl-clonidine), the first member of the family

    of specific bradycardic agents, is an imidazoline compound

    derived from the antihypertensive drug clonidine. Although

    structurally similar to clonidine, it has different pharmaco-

    logical effects, including analgesic (Stockhaus, 1977)   and

     bradycardic properties (Kobinger et al., 1979; Traunecker &

    Walland, 1980; Millar & Vaughan-Williams, 1981; Lillie &

    Kobinger, 1983).

    In a study in rabbit SAN, 0.3  Ag/ml alinidine induced the

    slowing of spontaneous activity accompanied by a prolon-

    gation of the action potential duration by about 10% (Satoh

    & Hashimoto, 1986). No effects were observed on the

    maximum diastolic potential (MDP), maximal dV /dt , and

    action potential amplitude. Further experiments in sheep

    Purkinje fibres showed that this drug is able to slow heart rate by reducing the steepness of diastolic depolarization,

    with limited side effects on action potential duration and

    inotropic state (Snyders & Van Bogaert, 1987). In these

    experiments, alinidine (28  AM) was reported to inhibit  I f  by

    shifting its activation curve to more negative voltages (by

    7.8 mV) and reducing its fully activated conductance (by

    27.0%). No use or frequency dependence was observed,

    suggesting that the drug binds equally well to open and

    closed channel states.

    Although alinidine was the first substance shown to act 

    as a bradycardic agent with little inotropic side effects, it 

    was not pursued as a pharmacological tool because itsselectivity for phase 4 changes was limited and partial

     prolongation of action potential duration was observed, as

    mentioned above, at moderate drug concentrations (0.3  Ag/ 

    ml). This effect implied that alinidine, as well as inhibiting

    f-channels, was acting as an inhibitor of other types of 

    channels, namely K + channels involved in repolarization, at 

    only slightly higher doses.

    4.2. Zatebradine (UL-FS 49) and cilobradine (DK-AH269)

    The search for more specific bradycardic agents led to a

    second group of molecules derived from the Ca2+ channel

    inhibitor verapamil, among which falipamil (AQ-A39),

    zatebradine (UL-FS49), and cilobradine (DK-AH26) have

     been investigated with some detail. Like alinidine, these

    drugs slow heart rate by the inhibition of the   I f  current. A

    slowing of spontaneous activity was observed in Purkinje

    fibres, intact SAN tissue, and single pacemaker cells (Van

    Bogaert & Goethals, 1987, 1992; Van Bogaert et al., 1990;

    Goethals et al., 1993; DiFrancesco, 1994; Thollon et al.,

    1994; Bois et al., 1996; Van Bogaert & Pittoors, 2003).

    In contrast to alinidine and derivatives, zatebradine and

    cilobradine do not modify the position of the   I f  activation

    curve but reduce the maximal conductance in a use- and

    frequency-dependent fashion (Goethals et al., 1993; DiFran-cesco, 1994). Preclinical studies were promising, and for 

    this reason, the specific action of zatebradine was analyzed

    extensively. Zatebradine acts from the cytoplasmic side of 

    the membrane (Van Bogaert et al., 1990; DiFrancesco,

    1994).

    In a study in SAN myocytes (DiFrancesco, 1994), the

    development of UL-FS49-induced   I f  block was slow, with

    time constants of the order of several tens of seconds; also,

     block only developed when the channel was open.

    Apparently in contrast with the requirement of negative

    voltages for channel opening, hyperpolarization relieved the

     block. These data indicated that UL-FS49 behaves as an

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   71

  • 8/20/2019 1-s2.0-S0163725805000252-main

    14/21

    bopen channel blocker  Q , and that block occurs when the

     protonated form of the drug binds at a site located ~39% of 

    the way across the membrane electrical field from the

    intracellular side of the membrane. Thus, use and frequency

    dependence of block arises from the need of drug molecules

    to move through part of the voltage drop across the channel

     pore in order to reach its binding site (DiFrancesco, 1994).Zatebradine is a more specific f-channel blocker than

    alinidine is. According to one report in SAN cells, 1   AM

    zatebradine caused a strong (65–95%) use-dependent block 

    of  I f  and only small reductions of sinoatrial Ca2+ and delayed

    K + currents (Goethals et al., 1993). Other reports confirmed

    these findings (BoSmith et al., 1993; Thollon et al., 1994;

    Bois et al., 1996; Valenzuela et al., 1996). Zatebradine

    reduced heart rate both at rest and during exercise, but these

    results were not accompanied by a significant prevention of 

    angina (Frishman et al., 1995). In addition, some undesired

    effects on vision, including the persistence of images in the

    visual field and flashes  (Frishman et al., 1995; Usui et al.,1995; Glasser et al., 1997), further hindered the use of 

    zatebradine as a cardiac selective drug. Experimentation in

     photoreceptors showed that this side effect of zatebradine

    was still due to the block of h-channels, the neuronal type of 

    f-channels, which are expressed in several different types of 

    neurons   (Pape, 1996; Robinson & Siegelbaum, 2003),

    including photoreceptors and retinal bipolar and ganglion

    cells. Incidentally, these data allowed an interpretation of the

    function of hyperpolarization-activated channels in the

    retina, since their inhibition by zatebradine led to a reduced

    ability to respond to high-frequency sinusoidal stimuli

    (explaining the persistence of images), thus indicating the

    involvement of  I h   in the response to rapid changes of light 

    intensity (Gargini et al., 1999).

    A newer congener of zatebradine, cilobradine (DK-

    AH269), was later found to induce a more effective and

    faster block of  I f  in SAN myocytes; for example, the block 

    due to 1   AM cilobradine was over 2-fold higher (82% vs.

    36%) and developed faster than that due the same concen-

    tration of zatebradine (Van Bogaert & Pittoors, 2003).

    4.3. ZD-7288

    Early data on the action of ZD-7288, a compound

    originally named ICI D7288, were obtained in isolated beating atria and intact SAN tissue of the guinea pig

    (Marshall et al., 1993) and in anaesthetized dog  (Rouse &

    Johnson, 1994). The results showed that in intact right atria,

    ZD-7288 (0.1–100  AM) reduced spontaneous rate (maximal

    slowing was 50%) but did not modify the contractile force

    of electrically stimulated left atria   (Marshall et al., 1993);

    similar effects were observed in intact SAN tissue (slowing

    of 61% at 0.3   AM concentration,   BoSmith et al., 1993),

    although a small prolongation of the action potential

    duration was also present (10%).

    The slowing of heart rate suggested the possibility that 

    ZD-7288 has an inhibitory action on f-channels, and

    experiments on single cells isolated from the guinea pig

    SAN confirmed this hypothesis (BoSmith et al., 1993). The

     I f   inhibitory effect of ZD-7288 (0.3   AM) consisted of a

    negative shift of the current activation curve (16.2 mV) and

    a decrease of the maximal conductance (by 52%; BoSmit h

    et al., 1993). Furthermore, ZD-7288, like alinidine but 

    unlike zatebradine, blocked   I f   in a use- and frequency-independent way; the simplest interpretation of this property

    is that ZD-7288 interacts with its channel binding site

    independently from the channel gating configuration  (Gas-

     parini & DiFrancesco, 1997). The development of ZD-

    7288-induced   I f   block in SAN myocytes and in other cell

    types was slow (BoSmith et al., 1993; Harris & Constant i,

    1995; Gasparini & DiFrancesco, 1997), suggesting an

    intracellular block like that of zatebradine.

    The selectivity of ZD-7288 for f-channels was inves-

    tigated by testing its action on Ca2+ and K + currents.

    Experiments in isolated SAN cells from the guinea pig

    (BoSmith et al., 1993) showed that both Ca2+

    and K +

    currents were only slightly affected by the drug. Concen-

    trations equal or less than 1   AM elicited a Ca2+ current 

    reduction of only 18%, while  I f  was strongly reduced (82%

    reduction); 0.1   AM ZD-7288 decreased the delayed K +

    current by about 1%.

    When the effects of ZD-7288 were evaluated in the

    central nervous system, substantial block of the hyper-

     polarization-activated current ( I h) was observed in guinea

     pig substantia nigra neurons, rat hippocampal CA1 cells, cat 

    ventrobasal thalamocortical neurons, and newt photorecep-

    tors   (Harris & Constanti, 1995; Gasparini & DiFrancesco,

    1997; Williams et al., 1997; Satoh & Yamada, 2002). The

    high drug sensitivity of the neuronal pacemaker current is a

    serious limitation to its use as a specific heart rate-limiting

    agent. It is, however, worth noting that the block of 

    hippocampal pacemaker current appears to require higher 

    drug doses than those blocking I f  in the SAN, which may be

    due to the different HCN isoform composition of native

    channels in these 2 tissues   (Gasparini & DiFrancesco,

    1997). The different heteromeric composition of f-channels

    in different tissues is the basis for a molecular approach

    aimed to develop drugs with tissue-specific properties,

    which explains the interest of knowing the precise isoform

    distribution in target tissues.

    4.4. Ivabradine (S16257)

    Ivabradine is, at present, the only member of the family

    of specific heart rate-reducing agents to have completed

    clinical development for stable angina. Its viability for 

    clinical use in ischaemic heart disease and cardiac failure

    has been carefully investigated by both in vitro and in vivo

    studies; its rate-reducing action is entirely attributable to a

    specific and selective inhibition of   I f    (DiFrancesco   &

    Camm, 2004).

    Early studies comparing the action of ivabradine and

    zatebradine in rabbit Purkinje fibres and in guinea pig

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7972

  • 8/20/2019 1-s2.0-S0163725805000252-main

    15/21

     papillary muscles showed that the 2 drugs are nearly

    equipotent in slowing rate, and that their effect is mediated

     by a reduced steepness of the slow diastolic depolarization

    (Thollon et al., 1994). The same analysis revealed that the

    action potential duration increases by about 29% with 3  AM

    zatebradine, and by only about 9% with the same dose of 

    ivabradine.Investigation in isolated SAN myocytes showed that the

    reduction of the slope of diastolic depolarization is due to

    the use-dependent intracellular blockade of f-channels (Bois

    et al., 1996). In this study, 3   AM ivabradine did not affect 

    either T-type or L-type Ca2+ currents, nor did it modify the

    delayed K + current, whereas  I f  was strongly reduced (about 

    60%). The overall effect of ivabradine is similar to that of 

    zatebradine and cilobradine since all drugs affect   I f   by

    inhibiting the maximal conductance without modifications

    of the voltage dependence of current activation (Bois et al.,

    1996).

    Additional studies of ivabradine on f-channels in SANcells have shed light on some of the molecular aspects of its

     blocking mechanism (Bucchi et al., 2002). These studies

    have shown that binding–unbinding reactions are restricted

    to the open state of the f-channel, implying that ivabradine is

    an open channel blocker like zatebradine. Also, similar to

    Fig. 3. Properties of the  I f  block by ivabradine. (A) I f  block induced by ivabradine during repetitive stimulation (100 mV/+5 mV) is partially removed by along hyperpolarizing step to 100 mV (compare traces a and c in inset). (B) When the same protocol is applied in the presence of Cs+, no block removaloccurs, indicating that current flow is required for block removal. (C) The voltage dependence of block shifts to more negative voltages when the external Na+

    concentration is reduced, and the shift is similar to that of the  I f   reversal potential, as measured by plotting fully activated  I / V  relations in the 2 solutions

    (bottom panel); also, there is a steep change of block efficiency across the reversal potentials in both conditions. This confirms that the block depends on

    current flow rather than on voltage per se. (D) The current dependence could be due to the interaction of ivabradine with permeating ions within the pore

    (ivabradine structure courtesy Dr. Peglion Servier). Panels  A   through  C, modified from Bucchi et al. (2002, with permission).

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–79   73

  • 8/20/2019 1-s2.0-S0163725805000252-main

    16/21

    other heart rate-reducing agents, ivabradine blocks f-

    channels more efficiently at depolarized voltages and blocks

    them from the intracellular side of the membrane. This

     property reflects the positively charged nature of the

     blocking molecule, which includes a quaternary ammonium

    ion, and its tendency to enter the channel from the internal

    side more easily at depolarized than at hyperpolarizedvoltages. Ivabradine, however, has a distinctive property in

    that its blocking action changes with voltage not because of 

    an intrinsic voltage dependence, but because of a depend-

    ence on the ion flow across the channel pore; ivabradine

     block of f-channels is therefore   bcurrent  Q  dependent  (Bucchi

    et al., 2002).

    Evidence for the current dependence of block is

    illustrated in   Fig.   3. In panels A and B, full block was

    induced by a repetitive activation/deactivation protocol in

    the presence of 3  AM ivabradine before the application of a

    long hyperpolarizing step in the presence (A) or in the

    absence of 5 mM Cs+

    (B), a known extracellular blocker of the  I f  current  (DiFrancesco, 1982). While in the absence of 

    Cs+ the long hyperpolarization clearly removes part of the

     block (compare in the inset the current records just before

    (A) and just after (C) the long step), in the presence of Cs+

    no block removal occurs. A simple interpretation of these

    data suggests that voltage hyperpolarization alone is not 

    enough to induce block removal, and that an inward ionic

    flow is required (Bucchi et al., 2002). Additional evidence

    for an   bion flow-dependent  Q  block is presented in panel C,

    where the voltage dependence of fractional block induced

     by 3   AM ivabradine is plotted for 2 different external Na+

    solutions. The 2 curves do not coincide, and in both cases, a

    steep change of block occurs across the reversal potential of 

    the current for the 2 Na+ concentrations, indicating that the

    electrochemical gradient, more than absolute voltage,

    determines the extent of the block. The   bcurrent  Q   depend-

    ence of a block is an atypical property of ivabradine that is

    not shared by other heart rate-limiting agents. A biophysical

    analysis of the current dependence suggests that f-channels

    are multi-ion, single-file pores and that ivabradine blocks

    ion flow by entering channels from the intracellular side and

    competing with permeating ions in their binding to specific

    ion-binding sites in the permeation pathway   (Fig.   3D;

    Bucchi et al., 2002).

    Preclinical tests have confirmed that the in vitrospecificity of ivabradine as an f-channel blocker correlates

    with a selective heart rate-reducing action with little or no

    cardiac side effects. In experimental animals, the heart rate

    was reduced without modification of the QT interval or 

    myocardial contractility; in addition, ivabradine decreased

    oxygen consumption during exercise and preserved coro-

    nary vasodilatation in conscious dogs (Thollon et al., 1994;

    Simon et al., 1995; Monnet et al., 2001; Colin et al., 2002,

    2004; Camm & Lau, 2003; Vilaine et al., 2003). Trials

    conducted with ivabradine on experimental animal models

    of cardiac ischemia and on patients with stable angina

    confirmed a cardioprotective therapeutic efficacy and

    showed an improved recovery of the regional contractility

    of the stunned myocardium; most importantly, drug

    concentrations affecting rate did not elicit undesired

    symptoms typical of other specific bradycardic agents,

    except for limited visual disturbances, which were anyway

    compatible with clinical tolerability and voluntary continu-

    ation of the therapy (Monnet et al., 2001; Borer et al., 2003;Vilaine et al., 2003).

    5. Future directions: the biological pacemaker

    The development of f-channel-specific drugs acting

    selectively on heart rate and having a potential for 

    therapeutic use represents an important clinical application

    of the concept of   b pacemaker  Q  current. The way these   bheart 

    rate-reducing Q    drugs act is to inhibit, by partial channel

     block, the   I f   function and thus slow pacemaker activity.

    Ideally, however, the concept of   b pacemaker  Q 

     current should be applicable not only to the aim of inhibiting pacemaking,

     but also to the opposite purpose, i.e. induce pacemaking.

    The question then becomes, is it possible to enhance

     pacemaking by means of an increased contribution of funny

    channels to activity?

    This idea forms the basis for the development of another 

     potentially important clinical application of the concept of 

    b pacemaker  Q  current, the so-called   b biological Q  pacemaker.

    In essence, this new application relies on the possibility to

    engineer a biological substrate that can be transferred in

    vivo to identified cardiac locations and enhance, or induce,

    automaticity. Ideally, such a device could replace the

    electronic pacemakers used today.

    The feasibility of this approach is supported by several

    b proofs of concept  Q , which have been reviewed elsewhere,

    along with the evaluation of advantages and drawbacks of 

     biological vs. electronic pacemakers (Rosen et al., 2004). A

    first demonstration that HCN channels can be used to

    enhance pacemaker activity was provided by evidence that 

    the overexpression of HCN2 in neonatal rat ventricular 

    myocytes by adenoviral infection substantially accelerated

    their spontaneous rate (Qu et al., 2001). The same approach

    used in canine hearts in vivo showed that the overexpression

    of HCN2 was able to induce spontaneous left-atrium rhythm

    after sinus arrest  (Qu et al., 2003).Cell-based methods have been adopted more recently.

    One method relies on the in vitro differentiation of stem

    cells towards a pacemaker-like phenotype; differentiated

    cells can eventually be implanted in vivo onto the heart, by

    grafting, to provide a new pacemaker source. Although in

    origin the new pacemaker site is exogenous, the implanted

    cells should, in principle, be able to integrate with the

    surrounding cardiac tissue. An obvious advantage of this

    approach is that the new pacemaker might be able to

    respond to autonomic modulation if the new cells express

    the whole chain of proteins mediating adrenergic and

    cholinergic stimulation.

     M. Baruscotti et al. / Pharmacology & Therapeutics 107 (2005) 59–7974

  • 8/20/2019 1-s2.0-S0163725805000252-main

    17/21

    Some of the crucial evidence showing the feasibility of 

    this approach has been provided.  Kehat et al. (2004)   and

    Xue et al. (2005) have shown that human Embryonic Stem

    Cells (hESCs) can differentiate following specific treat-

    ments into spontaneously