The patch clamp technique is the gold standard for functional assessment of single-cell electrophysiological properties since its introduction by Neher and Sakmann (Neher and Sakmann, 1976). It is utilized in various models, with neonatal rat ventricular cardiomyocytes (NRVCM) being frequently used for cardiac objectives due to their relatively robust nature, enabling prolonged culture duration, non-viral gene transfer, stressor- and drug-related assays (Louch et al., 2011; Chlopčíková et al., 2001). The ventricular action potential (AP), and its changes in various pathologies, are of great interest due to the high morbidity and mortality arising from ventricular arrhythmias, which constitute (in association with heart failure or myocardial infarction) a leading cause of death in the western population (Zeppenfeld et al., 2022). Therefore, the present work focuses on the heterogeneity of ventricular AP in NRVCM as measured by whole-cell patch clamp analysis, considering different methodological factors.

For patch clamp recording of APs in NRVCM, measurement of voltage over time data in whole-cell current clamp configuration is most often applied. Usually, ruptured patches (by application of negative pressure on the outside of the patch) are used. After resting membrane potential (RMP) measurement, some authors (Kilborn and Fedida, 1990; Lebeche et al., 2006; Hattori et al., 2012; Engels et al., 2015) report a current injection to reach a predefined holding membrane potential (HMP) of mostly −80 mV (−70 mV and −90 mV are also used). APs are typically induced by command current injections either with continuous repetition of single pulses at low frequency or in short trains (Kuryshev et al., 1999; Wickenden et al., 1997), often at higher frequencies.

Distinct parameters of the AP waveform are derived from the measured membrane potential. The maximal velocity of the depolarization (maximal upstroke velocity, dV/dt), characterizes the early phase of the AP, mostly defined by the density of I _Na (Attwell et al., 1979). The maximal AP amplitude (APA) is dependent on the relation of I _Na to the main early repolarizing currents I _Ca,L, I _to,f, I _to,s and, to smaller degree to I _Kr, I _K2P, I _f and I _NCX (Schmitt et al., 2014; Armoundas et al., 2003; Schram et al., 2002; Fenske et al., 2011). The subsequent early repolarization, plateau phase and late repolarization are most often described by the duration of the AP up to a defined percentage of repolarization. AP durations at 10%, 20%, 25%, 30%, 40%, 50%, 70%, 75%, 80%, and 90% of repolarization (APD₁₀, APD₂₀, APD₂₅, APD₃₀, APD₄₀, APD₅₀, APD₇₀, APD₇₅, APD₈₀ and APD₉₀) are described [e.g., (Zhang et al., 2019; Song et al., 2012; Korhonen et al., 2009; Fantini et al., 1990; Zasadny et al., 2020; Kang et al., 1995; Guo et al., 2007; Guo et al., 1996; Wickenden et al., 1997)], with APD₃₀, APD₅₀ and APD₉₀ being slightly more widely used. APD of low percentages of repolarization (APD₁₀ to APD₃₀) are used to describe the early repolarization. APD₅₀ is used to describe the plateau phase and the later APDs are used to describe the repolarization. Further parameters have been developed to describe and differentiate different shapes of APs. This includes solely visual characterization of the waveform as triangular or plateau-shape (Wu et al., 2013; Voitychuk et al., 2011; Bahrudin et al., 2011), but also quantification methods, e.g., the relation between maximal downstream velocity at plateau and repolarization phase (Ma et al., 2011; Syren, 2021a). There is, however, no common standard in reporting these AP properties. Most studies also report the RMP which is usually relatively stable and reflects mostly the density and reversal potential of potassium currents at rest (Bernstein, 1868; Bernstein, 1902).

NRVCMs have, beside their already mentioned advantages, several limitations when it comes to comparability with human (Varró et al., 2021) and adult rat physiologic properties. First, even slight differences in pregnancy duration and age of the neonatal rats at euthanization affect the differentiation process. APD decreases with increased differentiation of NRVCM, likely influenced by reduced I _K1, I _to and increased I _f at an early differentiation stage (Kilborn and Fedida, 1990; Cerbai et al., 1999). Commonly, NRVCM utilized for patch clamp originate from animals in between 2.5 ± 0.37 days of postnatal age (Supplementary Table S1). Increased culture duration also shortens the AP and increases maximal upstroke velocity (dV/dt), likely by increased on channel expression and density, increased amplitude of I _to and decreased I _Ca in long-term culture (Guo et al., 1996; Meiry et al., 2001; Fast and Kléber, 1993; Kamiya et al., 1999; Wickenden et al., 1997). Culture durations for patch clamp analysis vary mostly in between 1 and 5 days, while APs from NRVCM after as long as 15 days of culture have been described (Guo et al., 1996). In culture, both spontaneously beating and quiescent NRVCM appear, without differences in APs (Zhang et al., 2010). If mentioned in literature, there is no clear preference in patch clamp assays for either cell type, with beating (Fantini et al., 1990; Gaughan et al., 1998; Michels et al., 2008; Wu et al., 2013; Sun et al., 2015; Zhao et al., 2016), quiescent (Snopko et al., 2007; Guo et al., 2007; Engels et al., 2015; Zhang et al., 2019; Oshiyama et al., 2022) or both (Guo et al., 1996; Rastan et al., 2005; McSpadden et al., 2012; Silva Dos Santos et al., 2023) cell types included in analysis. Culture medium and coating of surfaces affect NRVCMs (Viero et al., 2008; Parameswaran et al., 2013; Louch et al., 2011; Chlopčíková et al., 2001) after prolongated culture duration. In particular, high fetal bovine serum (FBS) concentrations are known to result in dedifferentiation (Viero et al., 2008). Furthermore, paracrine factors of cardiac fibroblasts (and likely the amount of fibroblasts in NRVCM culture) cause reduction of dV/dt and prolong AP due to reduced I _Na, I _K1 and I _to (Pedrotty et al., 2009).

In summary, biological variance, chosen methodological approaches, data analysis and choice of output parameters vary strongly and limit comparability in between available data. Furthermore, data on the impact of variation of the mentioned parameters are sparse. In the present study, whole-cell patch clamp measurements of NRVCM under various conditions were used to elucidate the respective influences on the resulting AP and contextualize these findings with existing literature.

Firstly, an overview regarding AP heterogeneity in NRVCM is presented. Influences of technical aspects of patch-clamp methodology are presented in the following order: starting with setup temperature, electrode resistance, holding (respective resting) membrane potential, induction current frequency, duration and strength, and lastly duration of the experiment. In total, 464 publications were screened and data from 54 publications were included. For an in-depth overview of the chosen parameters and included data, compare Supplementary Table S1. Details of the original data analyzed in this work are provided in Supplementary Figure S1; Supplementary Table S2.

High variability of the AP in NRVCM poses a relevant problem for data interpretation in individual studies themselves and to an even greater degree for the comparability between different publications. The data illustrated in Figure 1 with reported mean APD₉₀ differing over more than one order of magnitude is combined with several reports of several clusters of different AP shape and consecutively different APD (Bahrudin et al., 2011; Wu et al., 2013). Possibly, these subgroups could arise from different cellular origin, e.g., epi- and endocardial localization (Voitychuk et al., 2011; Bahrudin et al., 2011), or could be caused by (de-)differentiation, as indicated by the influence of prolonged culture duration on different AP shapes (Wu et al., 2013). This data, indicating distinct subgroups, is consistent with several normality tests suggesting a non-Gaussian distribution of APD and APA in our data (Supplementary Section S2), further complicating statistical approaches and comparability. While the issue of significant variability in APD has been described earlier (Liang et al., 2015), and primarily biological causes like animal age, differentiation state of the NRVCM, culture duration (data from this work provided in Supplementary Section S6) and choice of culture media have already been discussed before (Viero et al., 2008; Parameswaran et al., 2013; Louch et al., 2011; Chlopčíková et al., 2001), the influence of patch clamp related parameters, to our knowledge, has not been discussed to the following extent before.

While a temperature of 37°C is usually applied for mammalian cell culture, except for specific purposes (Brown, 1990), the effect of temperature in the short time span of electrophysiological measurement is less clear. While measurement at 37°C reflects in vivo conditions best, it comes with the disadvantage of decreased cellular viability (and possible artificial APs of compromised cells), as described before (Chung and Campbell, 2013) and consistent with our (qualitative) observations. The significant increase of dV/dt at 37°C in our study is consistent with the finding of steep conduction increase of Na_V1.5 (mediating I _Na) at higher temperature (Milburn et al., 1995). We did neither observe significant changes of APA and APD in comparison of 28°C and 37°C, nor did observe a clear qualitative effect in literature data. This is in contrast to several works directly focusing on hypothermia, which are reporting a negative correlation of temperature with mainly QT in clinical observations and, less pronounced, of APD in vitro, attributing these changes mainly to decreased I _Ca, I _K and I _K1 (Kiyosue et al., 1993; Shah et al., 2006; Manzini et al., 1986; Dietrichs et al., 2019). Two possible reasons might explain these discrepancy: Firstly, temperature dependency of APD might be attributed to local cellular differences, as in particular an increase by over 300% in the dispersion of repolarization between different cardiac origins has been described in hypothermia (Piktel et al., 2011) and differences in hypothermia-associated APD changes between Purkinje fibres and ventricular myocardium have been described (Watanabe, 1975). Secondly, most studies addressing hypothermia-associated APD changes compare room temperature (20°C–23°C) or even lower temperatures with 37°C. Possible, the temperature dependency of the AP is not linear and more robust to changes closer to physiologic values. Supporting this hypothesis, a relative temperature insensitivity of calcium sparks above 25°C was described in cardiomyocytes (Fu et al., 2005) and guinea pig and ground squirrel cardiac APD₉₀ showed a distinct change between 10°C and 24°C, which was not apparent in the guinea pig and reduced for ground squirrel from 24°C to 36°C (Herve et al., 1992). Under these observations, regular measurement at 37°C was recommended (Chung and Campbell, 2013). Our data support this approach, if the early phase of the AP (and in particular I _Na) are within the scope of the research project. For analysis of cardiac repolarization, the decision is less clear and dependent on the assay design, where the faster cellular degradation and resulting artifacts under 37°C should be taken into consideration.

Lower electrode (and therefore access) resistance contributes to a lower series resistance, leading to artificial measurements in particular in voltage clamp (in particular for small currents (Armstrong and Gilly, 1992), while the effect size is debated and might be of low relevance for large currents (Gray and Santin, 2023)). For current clamp, RMP measurement is altered, likely due to increased and varying tip potentials, at last at vast differences in comparison to sharp microelectrodes (Li et al., 2004). Reported electrode resistance did not affect APA, APD or dV/dt significantly. Additionally, series resistance and its changes over the experiments’ duration can influence the RMP (Armstrong and Gilly, 1992) and possibly AP shape and duration. Input resistance reflects membrane resistance of the cell and might help in the categorization of NRVCM AP due to its sensibility to changes in potassium conductance (Cameron et al., 2000). Furthermore, its strong voltage dependency (Zaniboni et al., 2005) might influence cell capacitance and AP specifics observed in variable RMPs. The exact influence on AP heterogeneity in NRVCM of both series and input resistance are unknown and require further research.

Reported data on the RMP of NRVCM under experimental conditions also shows a wide variability with mean values between −50.5 (Snopko et al., 2007) and −86.4 mV (Wickenden et al., 1997). Whether current is injected during measurement to achieve a specific HMP or even to stabilize the initially measured RMP, is rarely mentioned, with −80 mV as most often described HMP (Metzele et al., 2011; Engels et al., 2015). In our experience, RMP under experimental conditions is seldomly stable and drops significantly over increased assay duration, making an increase of current injection necessary for stable HMP (data not shown). Furthermore, depolarized cells with initially low RMPs correlate with structurally compromised cells (Boyden et al., 1984) and should be discarded prior analysis (we suggest exclusion at RMP > −50 mV). In our study, we included such cells to better reflect the resulting data. The 5 cells included cells exhibited trendwise reduced APA of 97.6 ± 1.9 mV, APD₉₀ of 46.6 ± 2.9 ms and APD₅₀ of 25.5 ± 1.4 ms. APD is increased at a more positive RMP/HMP, while dV/dt and APA are negatively related. The observed behavior of dV/dt is likely highly influenced by the reduced steady-state channel availability of Na_V1.5 at a more positive HMP, conducting I _Na, whose peak current is elevated at a more negative HMP (Ruben et al., 1992; Walton and Fozzard, 1979). These kinetics of Na_V1.5 might also explain the observed decrease of APA with a more positive HMP, with a reduced peak current of I _Na leading to a smaller overshoot and reduced dV/dt (Kodama et al., 1981; Coraboeuf et al., 1979). Prolonged APD in more positive HMP results most likely from altered kinetics of a multitude of ion channels active in repolarization. Block of delayed rectifier currents (I _K) and the transient outward current (I _to) with 4-AP respectively TEA reduced HMP-dependent AP prolongation in neurons (Gong et al., 2008), implicating HMP-dependent kinetics of the respective ion channel K_V subunits. Considering these observations, we suggest the use of a stable membrane potential either by stabilizing the measured RMP in single NRVCM for comparisons of APs in the same cell over time or, if comparisons in between different cells are of interest, a predefined HMP for all cells.

Mostly, current pulses are used for AP induction, with only rare occurrences of intrinsic APs being analyzed (33 out of 44 or 75% sources providing information on this topic in contrast to 7 (16%) analyzing only spontaneous APs and 4 (9%) analyzing both entities of APs). We found no significant differences between induced APs from spontaneously beating and quiescent NRVCM (compare Supplementary Figure S5; Supplementary Table S5), as was described earlier (Zhang et al., 2010). The choice of induction protocol is rarely reported. A rectangular pulse is most often utilized, but varies highly in regard to frequency (from 0.2 Hz (Voitychuk et al., 2011) up to 10 Hz (Hattori et al., 2012)), induction pulse duration (from 1 ms (Rastan et al., 2005) up to 10 ms (McSpadden et al., 2012)) and induction current strength. Since the latter is relative, different approaches are used, from application of a relative percentage above the individual cells threshold (varying from 1.1-fold the threshold (McSpadden et al., 2012) to 2.0-fold the threshold (Rastan et al., 2005; Sun et al., 2015)) over fixed current strength steps (Metzele et al., 2011; Oshiyama et al., 2022; Liang et al., 2015; Kang et al., 1995) to (high suprathreshold) fixed current strength (Voitychuk et al., 2011; Yang et al., 2021). While duration of the induction pulse (inside the chosen range from 2 up to 10 ms) did not influence the AP in our study, current durations above 10ms change kinetics of ion channels similar to altered HMP (e.g., described for I _Na (Walton and Fozzard, 1979)), and should therefore be avoided. In contrast, the total amplitude of the induction pulse directly influences the maximal upstroke and consecutively APA for currents above 1.5-fold the threshold energy via overlap of induction pulse and early AP. Interestingly, we did not observe a significant difference in dV/dt, even at induction pulses which are active over the whole upstroke phase of the AP. Probably, the conduction of Na_V1.5 at the timepoint of dV/dt exceeds other currents to such an amount that their reaction to the induction pulse is irrelevant, as others have already pointed out the validity of dV/dt even under induced APs (Roberge and Drouhard, 1987). While the plateau phase and repolarization (including AP duration) seem to be relatively independent from induction until around 2.0-fold the threshold induction pulses, higher induction energies shorten the APD. First, the calculation method from APD, which is usually calculated from APA, contributes to these measurements, and artificially shortens APs at higher induction energy (and higher APA). Rarely described alternative APD calculations, such as fixed AP measurements at 0 mV and −60 mV (Liu et al., 2010), pose other problems, as described later. However, the APD prolongation at induction >2.0-fold the threshold cannot solely be explained by this confounding aspect. For several ion currents (in varying models, for following sources often own interpretations of I-V-relationships mentioned, since statistical tests for this question could not be found), voltage dependency with nonconstant channel kinetics beyond the regular maximal overshoot (at around +20 to +50 mV) has been described. For I _Kr, the tail current is qualitatively higher at +60 and +80 mV compared to +0 and +20 mV (Haworth et al., 2014; Wu et al., 2015; Lu et al., 2014; Zhou et al., 1998). Similar results have been described for I _to (Lu et al., 2014) and I _Ks (Kaboua et al., 2022; Wu et al., 2015), including the tail current (Gou et al., 2018). while I _Ca,L, including its later components, is qualitatively steeply declining at values higher than +30 mV (Lu et al., 2014). Taken together, increased tail currents of positive currents contributing to repolarization (I _to, I _Ks and I _Kr) and decreased negative currents stabilizing the AP plateau (I _Ca,L) at a very high overshoot induced by far suprathreshold induction pulses, shorten the APD. We suggest firstly the measurement of the induction energy threshold and following a slightly higher (e.g., 1.2-fold–1.5-fold the threshold) induction pulse energy. For projects addressing specifics of the early AP, e.g., I _Na, we suggest an even stricter choice of induction pulse energy very close to the threshold, with a clear separation of induction pulse from the depolarization phase. Furthermore, we suggest omission of any distorted AP waveforms from analysis, since they are likely to originate from compromised cells or methodological issues.

After an early phase of unstable APs, for which an stabilization period of 5 min before measurement was proposed (Kang et al., 1995; Rastan et al., 2005), different approaches for pacing are applied. Steady pacing with frequencies between 0.2 Hz (Voitychuk et al., 2011) and 10 Hz (Hattori et al., 2012) were described. We observed faster cell degeneration and death during continuous pacing (qualitative observation) over prolonged time. This might be the reason for the application of repeated series of induced APs (Wickenden et al., 1997; Kuryshev et al., 1999). Single APs differ from continuous paced cells in the latter case, APs are frequency-dependent (Wickenden et al., 1997). In our data, APD stabilized after 3 repeated inductions at 1 Hz and stayed stable without significant changes for up to at least 15 repetitions. This effect, also termed “post-rest adaptation,” is well known in adult cardiomyocytes of several mammalian species (Becher and Ravens, 1982; Ishida et al., 1998; Bassani et al., 2004), but is less pronounced described in the adult rat (Bassani et al., 2004). Lastly, prolonged AP measurements are of interest, in particular concerning pharmacological effects on the AP. Considering the issue of high AP variability between different cells, this approach is likely superior in comparison to measurement of different drug-exposed NRVCMs. Unfortunately, prolongated measurements shorten APD significantly. Furthermore, prolongated measurements are more time consuming due to a high rate of cell death during measurements. In this study, 23 out of 48 NRVCM survived 20 min of measurement, and (whereas altered by drug exposure as additional stressor after 20 min, data not shown) only 6 out of initial 48 cells survived 50 min of measurement. We suggest an initial stabilization phase and afterwards either series of induced APs (in particular if a prolongated measurement period is planned) or steady pacing at a suitable frequency for measurement.

Data analysis of the voltage over time AP data is performed differently, with limitations to each approach. Due to variation between individual cells’ APs and the possibility of technical artifacts, mean or median values from several data traces in an AP series are often analyzed (Christ et al., 2001). While slightly different algorithms for smoothing, calculation data mean/median or maximal values are applied for the calculation of RMP, APA and dV/dt (Puddu et al., 2010; Árpádffy-Lovas and Nagy, 2023), we did neither find reported major issues with these values in literature nor did we observe such in our analysis. Calculation of APD, however, has several limitations. Most authors calculate APD from APA. Issues include firstly the heterogenous AP shapes in NRVCM (as discussed in the first paragraph), which influence in particular “early” APDs such as APD₁₀ - APD₄₀ due to the high potential differences in between maximal amplitude and plateau phase. Secondly, induction pulse energy influences APA and consecutively APD, as mentioned before. Furthermore, the starting point of measurement is sometimes calculated from maximal upstroke, e.g., in (Sallé et al., 2008) and our study, and alternatively from the begin of phase 0, e.g., (Avula et al., 2021; Burton and Cobbe, 2001). While the latter approach is including phase 0 and therefore the whole AP, it is vulnerable to artifacts arising from induction, with in particular “late” APD (APD₈₀, APD₉₀) being prone to inclusion of induction pulse variants in calculation (as observable in Figures 3C, 4B; Supplementary Figures S3A–C). As a further method, measurement of the AP width at fixed potential levels (e.g., 0 mV and −60 mV, roughly comparable to APD₃₀ and APD₈₀ (Liu et al., 2010)) is sometimes utilized. This approach is independent from induction pulse energy, but more influenced by different AP shapes and degeneration processes than the former variants. We recommend measurement from maximal upstroke until XX% of repolarization due to its robustness regarding AP shape, with compulsory awareness of the effect of induction with this calculation method and careful choice of suitable induction pulse energy. Concerning statistical approach and taking into account our findings of likely non-Gaussian APD- and APA-distribution (as discussed earlier), we suggest nonparametric tests, as already applied previously for NRVCM-APD (Oshiyama et al., 2022; Silva Dos Santos et al., 2023).

In summary, native cardiac APs of NRVCMs are highly variable in both shape and duration, hindering analysis and comparability. We suggest the use of a constant HMP, variable and closely adjusted to the individual cells’ threshold induction pulse energies and the use of repetitive AP-inductions. After an initial stabilization phase and repetitive induction series, a prolongated measurement period might be advisable, dependent on the objective. Setup temperature has to be considered if phase 0 and I _Na are regarded. For analysis, we suggest the measurement of APD from maximal upstroke and nonparametric statistic test for statistical analysis. Improved measurement consistency and consecutively reduced data variability might lead to more reliable and comparable results, improving electrophysiological cardiac research options.