Functional expression of the hyperpolarization-activated, non-selective cation current I_f in immortalized HL-1 cardiomyocytes

Abstract

HL-1 cells are adult mouse atrial myocytes induced to proliferate indefinitely by SV40 large T antigen. These cells beat spontaneously when confluent and express several adult cardiac cell markers including the outward delayed rectifier K⁺ channel. Here, we examined the presence of a hyperpolarization-activated I_f current in HL-1 cells using the whole-cell patch-clamp technique on isolated cells enzymatically dissociated from the culture at confluence. Cell membrane capacitance (C_m) ranged from 5 to 53 pF. I_f was detected in about 30 % of the cells and its occurrence was independent of the stage of the culture. I_f maximal slope conductance was 89.7 ± 0.4 pS pF⁻¹ (n = 10). I_f current in HL-1 cells showed typical characteristics of native cardiac I_f current: activation threshold between −50 and −60 mV, half-maximal activation potential of −83.1 ± 0.7 mV (n = 50), reversal potential at −20.8 ± 1.5 mV (n = 10), time-dependent activation by hyperpolarization and blockade by 4 mm Cs⁺. In half of the cells tested, activation of adenylyl cyclase by the forskolin analogue L858051 (20 μm) induced both a ≈6 mV positive shift of the half-activation potential and a ≈37 % increase in the fully activated I_f current. RT-PCR analysis of the hyperpolarization-activated, cyclic nucleotide-gated channels (HCN) expressed in HL-1 cells demonstrated major contributions of HCN1 and HCN2 channel isoforms to I_f current. Cytosolic Ca²⁺ oscillations in spontaneously beating HL-1 cells were measured in Fluo-3 AM-loaded cells using a fast-scanning confocal microscope. The oscillation frequency ranged from 1.3 to 5 Hz and the spontaneous activity was stopped in the presence of 4 mm Cs⁺. Action potentials from HL-1 cells had a triangular shape, with an overshoot at +15 mV and a maximal diastolic potential of −69 mV, i.e. more negative than the threshold potential for I_f activation. In conclusion, HL-1 cells display a hyperpolarization-activated I_f current which might contribute to the spontaneous contractile activity of these cells.

A variety of tissue cultures are used by cellular cardiologists to study cardiac function at the cellular level. Most of them use cardiomyocytes isolated from immature hearts, typically from chick embryonic or rat neonatal hearts, which retain their capacity to divide and maintain a cardiac phenotype in culture. However, in spite of a number of efforts (see e.g. Jacobson & Piper, 1986; Bugaisky & Zak, 1989), long-term cultures of cardiomyocytes isolated from adult hearts lead to a progressive change in phenotype, as the cells move through dedifferentiated and redifferentiated states. Therefore, it has been a challenge for a number of investigators to obtain a stable adult cardiac cell line retaining a differentiated cardiac phenotype while proliferating in culture.

Recently, one such effort gave rise to HL-1 cells (Claycomb et al. 1998). These cells were derived from the atria of a transgenic mouse expressing the SV40 large T antigen under the control of the atrial natriuretic factor (ANF) promoter. The oncoprotein is responsible for the unique property of these cells to proliferate indefinitely in vitro (Lanson et al. 2000), in contrast to other cardiac cell lines (Claycomb et al. 1998). Under suitable culture conditions, HL-1 cells show spontaneous contractile activity and express many of the cardiac-specific genes typical of a differentiated adult cardiac phenotype (Claycomb et al. 1998). These include α-type (adult isoform) myosin heavy chain, α-type-cardiac actin, ANF and connexin 43, the most abundant protein of cardiac gap junctions. Moreover, immunofluorescence methods have shown the existence of perinuclear granules containing ANF, cardiac-specific muscle desmin filaments and sarcomeric myosin. Compared to adult cardiomyocytes, and like mitotically active embryonic myoblasts, HL-1 cells show a less organized ultrastructure, their cytoplasm containing a lot of developing myofibrils and glycogen rich areas (Claycomb et al. 1998).

With regard to signalling pathways, HL-1 cells express α1-adrenergic receptors that are coupled, as in rat cardiac myocytes (McWhinney et al. 2000b), to phosphatidyl inositol hydrolysis and the mitogen-activated phosphorylated kinase signalling pathway (McWhinney et al. 2000a). Moreover, Neilan et al. (2000) showed the presence of functional δ-opioid receptors coupled to phosphatidyl inositol hydrolysis, in analogy with several studies performed in rat cardiomyocytes. However, currently there is no available information about the cAMP pathway in HL-1 cells.

Although confluent HL-1 cells typically show a spontaneous contractile activity, which suggests either a functional excitation-contraction coupling or the presence of spontaneous release of Ca²⁺ from sarcoplasmic reticulum stores (Viatchenko-Karpinski et al. 1999), little is known about the electrophysiological properties of these cells. Only the presence of the inward delayed rectifier K⁺ current (I_Kr) has been shown (Claycomb et al. 1998). This current is typical of cardiac tissue and is sensitive to dofetilide, an antiarrhythmic drug used clinically in atrial fibrillation and other arrhythmias. However, several reports (Zhou & Lipsius, 1992; Brochu et al. 1992; Robinson et al. 1997; Cerbai et al. 1999; Abi-Gerges et al. 2000; Yasui et al. 2001) have demonstrated that spontaneously beating cardiac myocytes possess a hyperpolarization-activated I_f current (DiFrancesco et al. 1986). Moreover, at least in humans (Porciatti et al. 1997), I_f current has been shown to be present in atrial myocytes, from which HL-1 cells are derived. Therefore, our aim in this study was to search for the presence of I_f current in HL-1 cells and to characterize some of its basic properties.

A preliminary report of this work has appeared elsewhere (Sartiani et al. 2001).

Methods

Culture and isolation of HL-1 cells

HL-1 cells at passage 71 were obtained from Dr W. C. Claycomb (Louisiana State University Health Science Center, New Orleans, LA, USA), who first established and characterized the cell line (Claycomb et al. 1998). Cells were grown in T25, gelatin- fibronectin coated flasks, as previously described (Claycomb et al. 1998). Cells were maintained in ‘Claycomb Medium’ (JRH Biosciences, Lenexa, KS, USA), supplemented with 10 % fetal bovine serum (Life Technologies, Cergy Pontoise, France), 4 mm l-glutamine (Life Technologies), 10 μm noradrenaline (norepinephrine; Sigma Aldrich, L'Isle d'Abeau Chesnes, France) and penicillin-streptomycin (Life Technologies). The cultures were grown at 37 °C, in an atmosphere of 5 % CO₂ and 95 % air at a relative humidity of approximately 95 %. The medium was changed every 24 h. Single HL-1 cells were isolated from cultures at different passages (from 74 to 93), in a state of confluence or 24 h or 48 h after splitting. To do this, the cells were detached using a 5 min enzymatic dissociation with trypsin-EDTA (Life Technologies). Digestion was stopped by adding medium and the sedimented cells were either re-plated at a dilution of 1:3 (new passage) or used for daily measurements.

Intracellular Ca²⁺ measurements

After dissociation from the culture flask, cells were plated on gelatin/fibronectin-coated coverslips. After reaching confluence they were loaded with Fluo-3 AM (Molecular Probes, Eugene, OR, USA) by incubation in 3 μm Fluo-3 AM in normal Tyrode solution at 37 °C for at least 30 min. They were put on the stage of an inverted confocal microscope (LSM510, Zeiss France, Le Pecq, France) and maintained at 37 °C. Images were acquired with a Plan-Apochromat × 63 1.4 NA oil-immersion objective (Zeiss). Excitation was at 488 nm and fluorescence detection between 530 and 600 nm. Solutions were applied by gravity perfusion (1-2 ml min⁻¹) and the excess liquid was aspirated with a vacuum pump (Medcalf Bros., Potters Bar, UK); in these conditions, the time for exchanging the solutions near the cells (selected from the centre of the dish to avoid turbulence) was approximately 17 s. A 64 pixel × 64 pixel zone was scanned repeatedly at intervals of 50 ms. Finally, the signal was stored on a computer and analysis was performed with the LSM program and Origin (OriginLab Corp., Northampton, MA, USA). The spontaneous beating frequency was determined by manual counting of the number of oscillations in 10 s intervals.

Electrophysiological recordings

Electrophysiological recordings were performed using the whole-cell configuration of the patch-clamp technique (Hamill et al. 1981; Cerbai et al. 1994, 1996) in isolated cells enzymatically dissociated from the culture at confluence or 24 or 48 h after splitting. The experimental set-up consisted of an inverted microscope (Nikon, Champigny sur Marne, France), a home-made perfusion system, which allowed changes of the extracellular solution by gravity, a patch-clamp amplifier (Visual-Patch 500, Biologic, Claix, France) and a PC-compatible computer. Data acquisition was performed by means of the Visual-Patch 500 V 1.31B software (Biologic) and data analysis by means of Biotools (Biologic) and Origin software.

After dissociation, the cells were re-suspended in normal Tyrode solution and kept at room temperature for 4–5 h prior to experiments. During recordings, the cells were superfused with normal Tyrode solution (to measure action potentials, APs) or modified Tyrode solution (to measure I_f), see below. Experiments were performed at room temperature. Patch-clamp pipettes, prepared from borosilicate glass capillary tubes (GC150-15, Clark Electromedical Instruments, Reading, UK) by means of a two-stage vertical puller (Narishige, Kyoto, Japan), had a resistance of 2–5 MΩ when filled with the internal solution. APs were measured in current-clamp conditions, elicited at a rate of 0.2 Hz by 2 ms square current pulses and sampled at 2 kHz. I_f current was evoked by 1.5 s hyperpolarizing steps to potentials ranging from −60 to −130 mV from a holding potential of −40 mV. A mono-exponential fit of the current traces evoked at different potentials allowed derivation of the time constant τ (in s) of current activation which was analysed as a function of the membrane potential. The initial delay (maximally 22 ms) of the rising phase of the current was excluded from the fitting. Since the protocol used to activate I_f current was not long enough to reach steady-state at all potentials, I_f amplitude was measured from the fit as the difference between the extrapolated value at steady state and that at the beginning of the test pulse, after the initial delay. These values were subsequently used to generate activation curves.

Current amplitudes were normalized with respect to cell membrane capacitance (C_m) in order to obtain current densities. C_m was automatically evaluated by the recording software (Visual-Patch, Biologic). C_m averaged 19.2 ± 1.7 pF (n = 34), 18.3 ± 2.8 pF (n = 6) and 17.1 ± 3.1 pF (n = 10) in HL-1 cells at confluence, 24 h and 48 h after splitting, respectively (data not shown). These values were not significantly different and a similar dispersion of values, ranging from 5 to 53 pF, was present in the different passages of the cultures.

The reversal potential of I_f was evaluated by tail current analysis. Briefly, tail currents were recorded by 1.5 s ‘tail’ steps to membrane potentials ranging from −70 to 0 mV in 5 or 10 mV step intervals, preceded by a 1 s conditioning potential step to −120 mV (holding potential −40 mV) at the end of which I_f was fully activated. To compensate for the presence of interfering currents and correct for the time-independent component (mainly leakage), the data were fitted to an exponential decay starting 30 to 50 ms after the beginning of the tail step. Current amplitudes were measured as the difference between the extrapolated value at steady state and that at the beginning of the tail step. Tail amplitudes were then plotted as a function of the test membrane potential to obtain the instantaneous current-voltage (I-V) relationship, which was then fitted by a linear function which intersected the x-axis at the reversal potential of I_f. Specific conductance was determined following the equation:

where g_f (pS pF⁻¹) is the conductance calculated at the membrane potential V_m (mV), I (pA pF⁻¹) the current density and V_rev (mV), the reversal potential, derived from the analysis of the tail currents of the fully activated I_f current. Activation data were fitted by a Boltzmann function expressed by the equation:

where V_m (mV) is the membrane test potential, V_h (mV) the fitted potential for half-maximal activation and k (mV) the slope factor of the activation curve.

All data are expressed as means ± s.e.m. Statistical analysis was performed by means of Student's t test. A P value of less than 0.05 was considered significant.

Solutions and chemicals

Normal Tyrode solution contained (mm): NaCl 140; KCl 5.4; CaCl₂ 1.8; MgCl₂ 1.2; d-glucose 10; Hepes 5 (pH adjusted to 7.35 with NaOH). Modified Tyrode solution for I_f current was obtained from normal Tyrode solution supplemented with (mm): NiCl 2; BaCl₂ 2; 4-aminopyridine 0.5 to eliminate Ca²⁺ current (T- and L-type), inward rectifier K⁺ current, I_K,I and transient outward K⁺ current, I_TO, respectively. KCl was increased to 25 mm to amplify I_f.

Pipette solution contained (mm): potassium aspartate 120; TEACl 10; Na₂GTP 0.4; Na₂ATP 5; MgCl₂ 2; EGTA (acid form) 11; CaCl₂ 5 (pCa 6.9); Hepes 10 (pH adjusted to 7.2 with KOH).

L858051 was from Calbiochem (Meudon, France) and 8-bromo-cAMP from Sigma Aldrich.

Reverse transcription-polymerase chain reaction amplification (RT-PCR)

Total RNA was isolated from HL-1 cells culture and mouse brain by the combined acid phenol-guanidinium thiocyanate- chloroform method, as used in the Extract-All protocol (Eurobio, Les Ulis, France). Prior to reverse transcription (RT), RNA was treated with DNAse 1 (Life Technologies). cDNA was synthesized using Moloney Murine Leukaemia Virus reverse transcriptase (Life Technologies) following the manufacturers’ instructions and using poly dT oligonucleotide (12-18 bp) as primer. At the end of the RT, the cDNA was aliquoted in water and kept at −80 °C. Amplifications were performed on fresh aliquots.

PCR for hyperpolarization-activated, cyclic nucleotide-gated channel isoforms (HCN) was performed using primer constructions chosen on the basis of the mouse mRNA sequences coding for HCN1 (Accession Number (AC): AJ225123; Ludwig et al. 1998), HCN2 (AC: AJ225122; Ludwig et al. 1998), HCN3 (AC: AJ225124; Ludwig et al. 1998) and HCN4 (AC: AF064875, partial coding sequences; Santoro et al. 1998) isoforms (Table 1). Multiple alignment of the four sequences was generated with CLUSTAL W (Thompson et al. 1997). Selectivity of the primers was assured by the presence of mismatches at the 3′ end with their non-target sequences. Based on the human gene organization (Ludwig et al. 1999), primers were chosen on each side of the putative exons. The software Oligo (Rychlik & Rhoads, 1989) provided the thermodynamic characteristics relative to the compatibility of the primer pairs, as well as an indication of the annealing temperature. PCR was conducted on 5 μl aliquots of the RT products, each corresponding to 100 ng of initial total RNA, using 2.5 U Eurobiotaq DNA polymerase (Thermus aquaticus YT1, Eurobio), 10 pmol of each primer (Life Technologies) and 1.5 mm MgCl₂ (Eurobio) in a final volume of 50 μl. The reaction was conducted in a thermal cycler (Gene Amp PCR System 2400, Perkin Elmer, Courtabeuf, France) using the following protocol: 3 min at 94 °C, followed by 40 cycles consisting of 30 s at 94 °C, 30 s at 52 °C (HCN1 and HCN2, 59 °C (HCN3) or 54 °C (HCN4), 90 s at 72 °C and a final elongation at 72 °C for 5 min. After amplification, 10 μl of PCR product were analysed on a 1.5 % agarose gel stained with ethidium bromide.

Table 1.

Detection of the hyperpolarization-activated, cyclic-nucleotide-gated channels in HL-1 cells; primer sequences for the amplification and predicted digestion patterns

Sequence and accession number	Upstream primer	Downstream primer	Size of product (bp)	Restriction enzyme	Sizes of fragment (bp)
mHCN1 AJ225123	5′-ACCTGCTACGCAATGTTTG-3′	5′-TCAGCTTCATTTCTTTACTGGA-3′	448	BglII	176,272
mHCN2 AJ225122	5′-CGCATCTGTAACCTGATCA-3′	5′-GGCTGGAAGACCTCAAATT-3′	605	KpnI	429,176
mHCN3 AJ225124	5′-ATCGTGGTGGAGGAAGGTG-3′	5′-CTGCAGCATAGGGACCAGA-3′	383	BglII	121,262
mHCN4 AF064874	5′-GGTCAACAAATTCTCCCTAA-3′	5′-CAATGCGCACAGCCCTA-3′	455	TfiI	151,304
mβ-actin X03672	5′-CACCTTCTACAATGAGCTGCGTGTGGC-3′	5′-TTGCTGATCCACATCTGCTGGAAGGTGGA-3′	796	—	—

The primers and the sizes of the expected products and digestion fragments are based on the mouse sequences.

β-Actin PCR was carried out on the RT products from both HL-1 cells and mouse brain in order to test the efficiency of the reverse transcription reaction (RT+ products) and the deoxyribonuclease treatment quality (RT- product). The primers (Life Technologies) had sequences chosen on the basis of the mouse cytoskeletal β-actin (AC: X03672; Table 1). The PCR protocol was as described above except for the annealing temperature, which was 54 °C, and the number of cycles, which was 35.

In order to determine the isoform identity of the amplified products, the PCR products were digested by specific restriction enzymes (Table 1). Enzymatic digestion was performed as recommended by the manufacturer (New England Biolabs, Beverly, MA, USA).

Results

Rhythmic oscillations of cytosolic Ca²⁺ in HL-1 cells

During cardiac contraction, a transient increase in cytosolic Ca²⁺ (${\text{Ca}}_{\text{i}}^{\mathrm{2+}}$) occurs due to the sequential activity of voltage-dependent Ca²⁺ channels, intracellular Ca²⁺ stores and Ca²⁺ extrusion mechanisms (Fabiato, 1985; duBell & Houser, 1987; Barcenas-Ruiz & Wier, 1987; Nabauer et al. 1989; Bers, 1997). In spontaneously beating cells, the rapid increase in ${\text{Ca}}_{\text{i}}^{\mathrm{2+}}$ coincides with the occurrence of an action potential (Huser et al. 2000), allowing ${\text{Ca}}_{\text{i}}^{\mathrm{2+}}$ transients to be used as measures of the beating frequency. Figure 1 (lower panel, trace A) shows the rhythmic variation in fluorescence intensity of Fluo-3 AM-loaded cells measured in a spontaneously beating region of an HL-1 cell culture (region A, upper panel). As a control, the variations of fluorescence intensity (lower panel, traces B and C) relative to two non-beating zones (regions B and C, upper panel) are reported. The spontaneous beating frequency of different contracting spots of cells was calculated as reported in Methods and ranged between 1.3 and 5 Hz (n = 8).

Figure 1. Spontaneous activity measured by fluorescence intensity oscillations in Fluo-3 AM-loaded HL-1 cells.

HL-1 cells at confluence were incubated in 3 μm Fluo-3 AM in normal Tyrode solution at 37 °C for at least 30 min. Cells were excited at 488 nm wavelength and fluorescence emission was measured between 530 and 600 nm through an inverted confocal laser microscope. A 64 pixel × 64 pixel zone was scanned repeatedly at 50 ms intervals. Upper panel, mean of fluorescence images over several cycles. Lower panel, fluorescence measurement in one beating (A, ▪) and two non-beating regions (B, ♦, and C, □) of the above field. a.u., arbitrary units.

As shown in Fig. 2, the rhythmic oscillation of ${\text{Ca}}_{\text{i}}^{\mathrm{2+}}$ of a spontaneously beating spot of cells was strongly affected by application of 4 mm extracellular Cs⁺, which completely blocked the spontaneous ${\text{Ca}}_{\text{i}}^{\mathrm{2+}}$ rhythm (as seen in three other experiments). After washout of Cs⁺, spontaneous activity readily recovered (Fig. 2). Since Cs⁺ is a well known blocker of the hyperpolarization-activated cationic (I_f) channel, which has been shown to be associated with cardiac pacemaker centres (DiFrancesco, 1993), the blocking action of Cs⁺ on spontaneous beating activity of HL-1 cells suggested a possible contribution of I_f in the activity of these cells.

Figure 2. Blockade of spontaneous beating of HL-1 cells by Cs⁺.

Fluorescence emission was measured from a group of HL-1 cells that were spontaneously beating. Upper panel, the cell preparation was first superfused with normal Tyrode solution and spontaneous oscillations developed with a frequency of 2.7 Hz. When 4 mm Cs⁺ was added to the solution, spontaneous activity stopped (detail in lower panel, A). After washout of Cs⁺, spontaneous activity restored progressively (detail in lower panel, B) and reached a stable frequency of 3.8 Hz.

Action potential characteristics of HL-1 cells

As it is generally accepted that I_f contributes to the diastolic depolarization phase of the AP in spontaneously beating cells (DiFrancesco, 1993), we recorded APs from HL-1 cells under whole-cell current-clamp conditions. Figure 3A and B shows two representative APs recorded from two different isolated HL-1 cells. On average, the APs had a triangular shape, with an overshoot at +15.3 ± 1.9 mV and a maximal diastolic potential of −68.8 ± 1.6 mV (n = 19). Action potential duration was 35.3 ± 4.7 ms when measured at −20 mV and 69.2 ± 8.9 ms at −50 mV (n = 19). The two APs in Fig. 3, however, differ in that the one in Fig. 3A exhibits a clearly spontaneous diastolic depolarization phase, while in Fig. 3B, the diastolic potential was almost flat. Voltage-clamp experiments were performed on these two cells to check for the presence of a hyperpolarization-activated inward current. When the cells were voltage-clamped at −40 mV and hyperpolarizing steps to membrane potentials ranging from −60 to −130 mV were applied, a hyperpolarization-activated inward current was observed in the cell with a spontaneous diastolic depolarization phase (Fig. 3C), but not in the other cell (Fig. 3D). Out of 19 cells in which APs were successfully recorded, only three cells developed a spontaneous diastolic depolarization. Unfortunately, only the cell illustrated in Fig. 3A and C supported the additional voltage-clamp protocol and elicited a hyperpolarization-activated inward current. Of the 16 other cells, 11 were successfully voltage-clamped but none of them developed a hyperpolarization-activated inward current.

Figure 3. Action potential characteristics of HL-1 cells.

A and B, two representative action potentials (APs) recorded under whole-cell current-clamp condition in two different HL-1 cells superfused with Tyrode solution. Note that the AP in A exhibits a clear spontaneous diastolic depolarization phase, while the diastolic potential is almost flat in the AP shown in B. C and D, voltage-clamp of the same cells as in A and B, respectively. Currents elicited in modified Tyrode solution during 1.5 s hyperpolarizing steps to potentials ranging from −60 to −130 mV (holding potential −40 mV) are superimposed for each cell. This protocol evoked a family of hyperpolarization-activated inward currents in C, but not in D.

Characteristics of a hyperpolarization-activated inward current in HL-1 cells

In the experiments below, the cells were voltage-clamped immediately after patch-break, in order to make a more systematic assessment of the presence of I_f and to determine its characteristics. In ≈30 % of cells, hyperpolarization steps (1.5 s long) ranging from −60 mV to −130 mV (10 mV intervals) elicited a time-dependent, non-inactivating inward current, with an amplitude that increased as the potential became more negative. Representative current traces recorded in a single HL-1 cell are shown in Fig. 4A. Specific conductance was calculated as reported in Methods and normalized with respect to the maximal conductance value for each cell. Mean normalized conductance from 50 cells is plotted in Fig. 4B as a function of membrane potential. The activation curve was obtained by fitting the experimental data points to a Boltzmann function (see Methods). This yielded an activation threshold for the current between −50 and −60 mV, a half-maximal activation potential (V_h) of −83.1 ± 0.7 mV and a slope factor (k) of 9.0 ± 0.6 mV. Time constants (τ) of activating currents were determined by mono-exponential fitting of individual current traces. Mean τ values from 50 cells are plotted in Fig. 4C as a function of membrane potential. As the voltage became more negative, the activation of the current became progressively faster, the time constant going from 7.2 ± 0.7 s at −60 mV to 0.20 ± 0.03 s at −130 mV.

Figure 4. Voltage dependence of a hyperpolarization-activated I_f in HL-1 cells.

Isolated HL-1 cells were hyperpolarized for 1.5 s from a −40 mV holding potential to various potentials ranging from −60 to −130 mV (in 10 mV intervals), as shown by the voltage-protocol in the lower panel of A. Tail currents were obtained by a subsequent depolarization to +40 mV, as indicated. A, top, representative activating and tail current traces obtained with this protocol. Mean I_f fractional activations (B) and I_f activation time constants (C) were obtained at each membrane potential from the fit of the activating current traces, as described in Methods. Each symbol represents the mean ± s.e.m. of 50 similar experiments carried out as in A.

The measurement of tail current amplitudes was used to evaluate the reversal potential (V_rev) of the current. Tail currents traces displayed in Fig. 5A were elicited in a 39 pF cell by depolarizing steps ranging from −60 to 0 mV in 5 mV increments, following a hyperpolarizing pulse to −120 mV, which was sufficient to maximally activate the current. Tail current amplitudes were measured as described in Methods and normalized with respect to C_m. The fully activated I-V relationship shown in Fig. 5B was obtained by plotting the mean values of normalized tail currents from 10 cells as a function of membrane potential. Best fit through the data points gave a linear relationship that intersected with the voltage axis (V_rev) at −20.8 ± 1.5 mV and had a slope maximal conductance of 89.7 ± 0.4 pS pF⁻¹.

Figure 5. Current-voltage relationship of fully activated I_f in HL-1 cells.

Isolated HL-1 cells were hyperpolarized during 1.5 s to −120 mV from a −40 mV holding potential to fully activate I_f. Tail currents were obtained by subsequent depolarizations to potentials ranging from −70 to 0 mV (in 5 or 10 mV steps) as indicated in the voltage- protocol in the lower part of A. A, upper part, representative current traces obtained with this protocol. B, mean peak tail current amplitudes were normalized to cell membrane capacitance to give current densities shown as a function of membrane potential. Each symbol shows the mean ± s.e.m. of 10 experiments similar to those in A.

Sensitivity of the hyperpolarization-activated inward current to extracellular Cs⁺

Blockade by extracellular Cs⁺ is a hallmark of I_f, since millimolar concentrations of Cs⁺ were shown to inhibit I_f in sino-atrial node cells (DiFrancesco et al. 1986), latent atrial pacemaker cells (Zhou & Lipsius, 1992), atrial cells (Porciatti et al. 1997) and ventricular cells (Cerbai et al. 1994), as well as Purkinje fibers (DiFrancesco, 1982) and cells (Callewaert et al. 1984). The block is voltage dependent as the current is reduced only at potentials negative to its reversal potential, V_rev.

Figure 6 shows that the hyperpolarization-activated inward current recorded in HL-1 cells was also sensitive to Cs⁺. Tail currents were elicited as described in the Methods in the absence or presence of 4 mm Cs⁺ added to the extracellular solution. As shown by the individual current traces in Fig. 6A, application of Cs⁺ resulted in a complete block of the current elicited during the conditioning pulse at −120 mV. Moreover, the I-V relationship in 4 mm Cs⁺ (Fig. 6B) showed a clear reduction of the inward currents at potentials negative to V_rev, while the outward currents were unaffected. Similar voltage-dependent block of the f-channel by Cs⁺ has been reported in several other studies (DiFrancesco, 1982; DiFrancesco et al. 1986; Zhou & Lipsius, 1992). As shown in the activation curves of Fig. 6C, 4 mm Cs⁺ blocked the hyperpolarization-activated inward current in HL-1 cells at all membrane potentials ranging from −60 to −130 mV and its effect was mostly reversible upon washout. At −120 mV, 4 mm Cs⁺ blocked 92.0 ± 7.9 % (n = 4) of the current.

Figure 6. Blockade of I_f by Cs⁺.

A, the same protocol as in Fig. 5 was applied to an HL-1 cell in a modified Tyrode solution (control) and after application of 4 mm Cs⁺. Representative current traces are shown. B, the I-V relationship of the corresponding tail current density under control conditions (▪) and in the presence of 4 mm Cs⁺ (○). C, specific I_f conductance (g_f) was calculated as described in Methods and is plotted as a function of membrane potential in control conditions (▪), in the presence of 4 mm Cs⁺ (○) and after washout of Cs⁺ (□).

Thus, from the voltage dependence, the activation curve and the sensitivity to extracellular Cs⁺, we conclude that the hyperpolarization-activated inward current measured in HL-1 cells is carried by f-channels. This current will be denominated I_f from now on.

Presence of I_f throughout the cultures in HL-1 cells

As shown in Fig. 3, I_f was not found to be present in every cell. Therefore, we tested whether the state of the cultures might affect the occurrence of the current. At confluence, 31 % of cells (34 out of 109 tested cells) exhibited an I_f; when examined 24 h and 48 h after splitting, I_f was observed in 32 % of cells (6 out of 19 tested cells) and in 23 % of cells (10 out of 44 tested cells), respectively. Thus, the degree of confluence was not a determinant factor for the occurrence of I_f in HL-1 cells.

Figure 7A shows the scatter of I_f densities (pA pF⁻¹) elicited at −120 mV in a large number of individual experiments and their mean values in each of the three culture conditions. The relative time constants of I_f activation are reported in a similar manner in Fig. 7B. No significant difference in current density and time constants of current activation was found in the different culture conditions. Moreover, the activation curves obtained by fitting the experimental data did not show a significant difference either. Indeed, V_h values were −82.6 ± 0.6 mV (n = 34), −80.9 ± 1.1 mV (n = 6) and −83.9 ± 0.9 mV (n = 10) for cells at confluence, 24 h and 48 h after splitting, respectively. In the same three conditions, the slope factors k were 9.3 ± 0.7 mV, 8.9 ± 1.0 mV and 9.5 ± 1.0 mV, respectively.

Figure 7. Influence of the culture conditions on I_f density and activation time constant.

I_f densities (A) and activation time constants (B) were measured at −120 mV membrane potential in HL-1 cells that were at confluence, or 24 h or 48 h after splitting. Values from individual experiments (○) and means ± s.e.m. (▪, with the number of experiments indicated near each symbol) are shown for each culture condition.

In conclusion, I_f is present in about 30 % of HL-1 cells and its characteristics are rather constant over the course of the culture. Thus, all data obtained at different degrees of confluence were pooled for subsequent analysis.

Effect of adenylyl cyclase stimulation on I_f in HL-1 cells

A classic hallmark of the cardiac I_f is its stimulation by intracellular cAMP (for reviews, see e.g. DiFrancesco, 1993; Kaupp & Seifert, 2001; Accili et al. 2002). Thus, the direct binding of cAMP to the intracellular cyclic nucleotide-binding domain of f-channels leads to a positive shift in the activation curve of I_f (DiFrancesco & Tortora, 1991; Ludwig et al. 1998; Santoro et al. 2000; Wainger et al. 2001; Moroni et al. 2001; Viscomi et al. 2001). Moreover, a cAMP-dependent phosphorylation via cAMP-dependent protein kinase (PKA) has been reported to modulate I_f (Chang et al. 1991; Abi-Gerges et al. 2000) producing an increase of the fully-activated current (Accili et al. 1997; Proenza & Accili, 2001).

To examine whether I_f was also regulated by cAMP in HL-1 cells, we tested the effect of L858051, a water soluble and stable analogue of forskolin. Forskolin is known to directly stimulate adenylyl cyclase activity and increase intracellular cAMP. Like forskolin, L858051 has been shown to increase the L-type Ca²⁺ current by a cAMP-dependent mechanism (Hartzell & Budnitz, 1992). Figure 8 shows the mean fractional activation of I_f in HL-1 cells plotted as a function of membrane potential, both in the absence and presence of 20 μm L858051. A fit of the data points to the Boltzmann function revealed a shift of the activation curve towards less negative potentials in the presence of L858051. Indeed, V_h varied from −96.0 ± 0.9 mV to −90.6 ± 1.4 mV (n = 4, P < 0.05). In the presence of 20 μm L858051, the maximally activated conductance was increased by 37 ± 4 % (n = 4, P < 0.001) compared to the control solution in the same cells. Thus, activation of the cAMP pathway in HL-1 cells led to both a positive shift in the I_f activation curve and an increase in the fully activated I_f.

Figure 8. Effect of the forskolin analogue L858051 on I_f in HL-1 cells.

Mean activation curves of I_f were obtained in modified Tyrode solution in the absence (▪) and presence (□) of 20 μm L858051. Each symbol shows the mean ± s.e.m. of results from four cells. The curves were obtained by measuring the specific I_f conductance g_f at each membrane potential, as described in Methods. The conductance values were then normalized to the maximal conductance obtained in control conditions, demonstrating a clear increase in g_f during application of L858051.

Molecular characterization of HCN isoforms present in HL-1 cells

So far, our results indicated that a time-dependent inward current was present in HL-1 cells with the typical characteristics of an I_f, as found in cardiac pacemaker cells (Porciatti et al. 1997; DiFrancesco, 1993). In the following experiments, our aim was to pin down the molecular identity of the f-channel(s) expressed in HL-1 cells.

Previously, four different isoforms of hyperpolarization-activated, cyclic nucleotide-gated (HCN) channels have been described to possibly underlie the I_f (Kaupp & Seifert, 2001). Since several reports show the presence of the four isoforms (HCN1, HCN2, HCN3 and HCN4) in mouse brain (Ludwig et al. 1998, 1999; Santoro et al. 1998, 2000), we used this tissue as a positive control to test the efficiency of the primer pairs, each being designed to selectively amplify one isoform (see Methods). Products obtained by PCR amplification had the size expected from the published sequences (Ludwig et al. 1998; Santoro et al. 2000; see Table 1). Figure 9A shows an agarose gel of the PCR amplification products obtained from mouse brain using the RT-PCR amplification protocol described in Methods. Four fragments of the expected size are clearly detectable, thus verifying the efficiency of the four primer pairs.

Figure 9. PCR amplification of hyperpolarization-activated channel cDNAs.

Amplification from mouse brain cDNA (A) and from HL-1 cell cDNA (B). The sizes (bp) of markers are indicated on the left. The primers and restriction enzymes used are listed in Table 1. Note that all four HCN isoforms were detected equally in mouse brain whereas HCN1 and HCN2 were predominant in HL-1 cells.

RT-PCR of RNA from HL-1 cells was performed with the same primers and the same thermal cycle conditions as on that from the brain and produced a pattern of four bands Comparison of the signals obtained from HL-1 cells and mouse brain clearly shows that HCN1 and HCN2 isoforms are detectable as strong bands both in HL-1 cells and in the brain. Conversely, while the PCR conditions used allowed us to obtain rather strong HCN3 and HCN4 bands from brain tissue, the same conditions in HL-1 cells yielded weak bands, which were just above the detectable level. Therefore, we expect HCN1 and HCN2 isoforms to be expressed in a larger amount than HCN3 and HCN4 isoforms and to contribute predominantly to the functional I_f in HL-1 cells.

To check the identity of each amplified fragment, the products were digested using specific restriction endonucleases (Table 1). Figure 10 shows the results obtained using PCR products from brain and HL-1 cells. Enzymatic digestion of HCN1 and HCN2 products from both brain and HL-1 cells were as expected from the sequences (Table 1). Due to the weak signal obtained in HL-1 cells for HCN3 and HCN4 isoforms, the identity of the amplified products was assessed only in the brain products. There, the amplified fragments were abundant enough to produce detectable signals after the digestion. In both cases, the size of the fragments was in accordance with a specific amplification of the right isoform sequences.

Figure 10. Characterization of the products from PCR amplification of the HCN cDNAs.

Restriction enzyme digests of PCR products amplified from mouse brain cDNA (A) and from HL-1 cell cDNA (B). The sizes (bp) of markers are indicated on the left. The primers and restriction enzymes used are listed in Table 1.

Discussion

In this study, we have described the presence of a hyperpolarization-activated inward current in immortalized HL-1 cardiomyocytes. The current has many of the electrophysiological characteristics of the I_f present in pacemaker cardiac cells (DiFrancesco, 1993): (1) it is typically activated upon a hyperpolarization to membrane potentials < −50 mV; (2) its activation is time dependent and its amplitude increases as the potential becomes more negative; (3) it is blocked by 4 mm Cs⁺ and is insensitive to Ba²⁺; and (4) it is activated by cAMP.

I_f was present in only about 30 % of HL-1 cells. Several studies have reported a cellular heterogeneity with respect to the presence of I_f in myocytes isolated from native cardiac tissues, such as sino-atrial node (Denyer & Brown, 1990; Belardinelli et al. 1988) and atrial tissue (Zhou & Lipsius, 1992). However, the morphology of HL-1 cells was rather homogeneous, characterized by a flat, triangular spindle-like shape. After enzymatic dissociation, all HL-1 cells rounded up so that it was impossible by looking at the cell to predict which cell would express I_f and which would not. We examined the possibility that f-channel density might increase together with an increase in cell size, as has been reported in dedifferentiated adult rat ventricular cells in primary culture (Fares et al. 1998) and in rabbit sino-atrial node cells (Honjo et al. 1996). However, we found no significant difference in cell membrane capacitance between cells which expressed an I_f (18.7 ± 1.3 pF, n = 50) and those which did not (16.7 ± 0.8 pF, n = 145), which is consistent with an independent regulation of f-channel expression and cell dimension (Cerbai et al. 1996, 1999). As shown in Fig. 7, the passage number or the degree of confluence of the cell culture was also not a determinant factor for the presence of I_f. A possible hint for the limited presence of I_f in HL-1 cells might come from the spontaneous activity of the cells. Indeed, prior to enzymatic dissociation, when the cells were attached to the protein matrix of the flask, there were regions where the cells showed a spontaneous contractile activity and other regions where the cells were quiet (Fig. 1). As shown in Fig. 2, these beating regions stopped their spontaneous activity in the presence of Cs^+, which strongly suggests the presence of I_f in these cells as well as its contribution to spontaneous activity.

The I-V relationship of the fully activated I_f in HL1 cells was linear with a reversal potential of −21 mV, as determined in 25 mm K⁺ external solution. This value is consistent with a selectivity of the channel for Na⁺ and K⁺ (DiFrancesco, 1981; DiFrancesco et al. 1986; Ho et al. 1994). The permeability ratio (P_Na/K) calculated using the Goldman-Hodgkin-Katz-related equation (Frace et al. 1992) is 0.37 ± 0.02, a value in the range of those reported for I_f channels in cardiac myocytes (Frace et al. 1992; Irisawa et al. 1993; Cerbai et al. 1994, 1997; Hoppe et al. 1998).

Analysis of the voltage dependence of I_f activation in HL-1 cells allowed us to determine a potential for half-maximal activation (V_h) of −3.1 ± 0.7 mV and a slope factor of the activation curve (k) of 9.0 ± 0.6 mV, values which are similar to those reported for I_f in native human atrial myocytes (Porciatti et al. 1997; Hoppe et al. 1998) and cat atrial pacemaker cells (Zhou & Lipsius 1992). Moreover, a recent study by Ulens & Tytgat (2001a) showed that the functional expression of concatenated heteromers of HCN1 and HCN2 isoforms in Xenopus oocytes causes them to display a current with similar activation parameters to those reported in our study. Finally, the hyperpolarization-activated I_h current in CA1 hippocampal neurons, where coexpression of HCN1 and HCN2 isoforms has been detected at the single-cell level, shows activation parameters that closely resemble those found in HL-1 cells (Franz et al. 2000).

Likewise, the time constant of activation of I_f in HL-1 cells (τ) varied as a function of voltage, from approximately 7.2 ± 0.7 s at −60 mV to 0.20 ± 0.03 s at −130 mV, in a similar manner to that reported for I_f in native human atrial myocytes (Hoppe et al. 1998) and cat atrial pacemaker cells (Zhou & Lipsius, 1992). At −120 mV, τ was 0.29 ± 0.04 s in HL-1 cells, which corresponds to an intermediate value between the slow and fast time constants reported for the current obtained at this potential upon expression of concatenated HCN1-HCN2 channels (Ulens & Tytgat, 2001a). For a closer comparison with the latter study, we performed a bi-exponential fit of I_f traces evoked at −120 mV in HL-1 cells. This analysis yielded fast and slow time constants of 0.11 ± 0.02 s (n = 10) and 0.58 ± 0.11 s (n = 10), respectively, which are quite similar to the values reported by Ulens & Tytgat (2001a) and those obtained for I_h current in CA1 hippocampal neurons (Franz et al. 2000). These strong similarities between the functional properties of I_f in HL-1 cells and the current induced by expression of a concatenated HCN1-HCN2 channel (Ulens & Tytgat, 2001a) or the I_h current in CA1 hippocampal neurons (Franz et al. 2000) strongly suggest that HCN1 and HCN2 channel isoforms also account for the I_f in HL-1 cells. Further support for this hypothesis came from our RT-PCR experiments, which showed that HCN1 and HCN2 channel mRNAs are expressed in larger amounts than HCN3 and HCN4 isoforms in HL-1 cells. Thus, HL-1 cells differ from sino-atrial myocytes, where the I_f is carried mainly by HCN1 and HCN4 channel isoforms (Ishii et al. 1999; Moroni et al. 2001).

Activation of the cAMP pathway by the forskolin analogue L858051 showed a small but significant shift of the I_f activation potential (V_h) towards more positive values. Although this effect was observed in a small portion of cells (four out of eight cells), several reasons make us confident that this phenomenon was real: (1) the effect of L858051 was mimicked by 8-bromo-cAMP (400 μm, 2 out of 10 cells), a membrane-permeant cAMP analogue which also activates PKA (data not shown); (2) there was no spontaneous shift of V_h in control conditions after an equivalent amount of time (11 ± 3 min) as that used when measuring the effect of L858051 (data not shown); (3) the shift in V_h was quantitatively similar to that reported by others for the heteromeric HCN1-HCN2 current (Ulens & Tytgat, 2001a), the homomeric HCN2 current (Ulens & Tytgat, 2001b) and the native I_h current in CA1 hippocampal neurons (Franz et al. 2000). Why then was the effect of cAMP on I_f so rare? One possible reason is that the culture medium contained 10 μm noradrenaline (norepinephrine), thus providing a chronic stimulation of cAMP synthesis likely to maximally shift V_h towards positive voltages (Graf et al. 2001). In support of this hypothesis is the finding that V_h was −96.0 ± 0.9 mV (n = 4) under basal conditions in the cells responding to L858051, while V_h was −83.2 ± 1.5 mV (n = 4) in the cells not responding to the forskolin analogue.

As a consequence of the relatively low occurrence of I_f in HL-1 cells and its low responsiveness to cAMP, it was rather difficult to study in detail the intracellular mechanisms involved in this regulation. However, our data demonstrated that, in addition to a positive shift of the I_f activation, L858051 also significantly increased the I_f maximal conductance in HL-1 cells. This effect resembles that of 8-bromo-cAMP on I_h maximal conductance in CA1 hippocampal neurons (Gasparini & DiFrancesco, 1999) and of calyculin A, a phosphatase inhibitor, on I_f maximal conductance in rabbit sino-atrial node myocytes (Accili et al. 1997). According to the latter study, the increase in I_f maximal conductance might result from phosphorylation via PKA of the f-channels (Accili et al. 1997), which clearly differs from the direct binding of cAMP to the HCN channels leading to the positive shift of the activation curve (see e.g. Moroni et al. 2001; Viscomi et al. 2001). Preliminary data by Proenza & Accili (2001) show that the homomeric mouse HCN2 channel displays an increase in maximal current due to PKA activation. This behaviour may be specific to the HCN2 channel isoform, since no effect of cAMP on maximal conductance was found when expressing the HCN1 (Moroni et al. 2001) or HCN4 isoforms (Ishii et al. 1999). This suggests that, in HL-1 cells, modulation of the maximal conductance of f-channels is dominated by the HCN2 isoform.

Although the precise relationship between I_f activation and HL-1 automaticity was beyond the scope of this study, action potential recordings demonstrated that: (1) the maximal diastolic potential of HL-1 cells was more negative than the threshold potential for I_f activation; and (2) I_f was always absent in cells lacking a spontaneous diastolic depolarization. Moreover, the spontaneous ${\text{Ca}}_{\text{i}}^{\mathrm{2+}}$ oscillations were blocked by Cs⁺ at the same concentration at which it fully blocked I_f. While this may indicate a possible participation of I_f in the spontaneous contractile activity of these cells, a more precise pharmacological dissection of the different potential pacemaker currents is needed to evaluate their respective contribution to the electrical activity of these cells. For instance, in proliferating immature or early developed mouse myocytes, which also express a functional I_f channel (Abi-Gerges et al. 2000), electrical activity can be generated by spontaneous Ca²⁺ release from the sarcoplasmic reticulum (Viatchenko-Karpinski et al. 1999). This indicates that the presence of I_f may not necessarily imply that this current is necessary or sufficient to generate spontaneous activity.

In conclusion, HL-1 cells displayed a hyperpolarization-activated I_f that shares several functional properties with native cardiac f-channels. Molecular characterization of the HCN channels responsible for I_f demonstrated a dominant contribution of HCN1 and HCN2 channel isoforms. HL-1 cells may thus provide a useful tool for testing pharmacological drugs acting on I_f as well as identifying key factors in the cellular environment acting on HCN channel expression.