RyR-NCX-SERCA local crosstalk ensures pacemaker cell function at rest and during the fight-or-flight reflex

Abstract

Rationale

A recent study published in Circulation Research by Gao et al. employed SA node-targeted, incomplete Ncx1-knockout (KO) in mice to explore the role of the Na⁺/Ca²⁺ exchanger (NCX) in cardiac pacemaker. The authors concluded that NCX is required for increasing sinus rates, but not for maintaining resting heart rate. This conclusion was based, in part, on numerical model simulations performed by Gao et al. that reproduced their experimental results of unchanged action potentials in the KO SA node cells. The authors, however, did not simulate the NCX current (I_NCX), i.e. the subject of the study.

Objective

We extended numerical examinations to simulate I_NCX in their incomplete KO SA node cells that is crucial to interpret the study results.

Methods and Results

I_NCX and Ca²⁺ dynamics were simulated using different contemporary numerical models of SA node cells. We found that minimum diastolic Ca²⁺ levels and I_NCX amplitudes generated by remaining NCX molecules (only 20% of control) remained almost unchanged. Simulations using a new local Ca²⁺ control model indicate that these powerful compensatory mechanisms involve complex local cross-talk of Ca²⁺ cycling proteins and NCX. Specifically, lower NCX expression facilitates Ca²⁺-induced Ca²⁺ release and larger local Ca²⁺ releases that stabilize diastolic I_NCX. Further reduction of NCX expression results in arrhythmia and halt of automaticity.

Conclusions

Remaining NCX molecules in the incomplete KO model likely produce almost the same diastolic I_NCX as in wild-type cells. I_NCX contribution is crucially important for both basal automaticity of SA node cells and during the fight-or-flight reflex.

1. Introduction and background

The Na⁺/Ca²⁺ exchanger (NCX) in cardiac cells has two fundamental roles. Firstly, it maintains the cell Ca²⁺ balance by matching Ca²⁺ efflux to Ca²⁺ influx through the L-type Ca²⁺current (I_CaL). During diastole, NCX extrudes one Ca²⁺ from the cell in exchange for three extracellular Na⁺ (forward mode), generating an inward current, I_NCX. The second fundamental function of NCX specific to cardiac pacemaker cells is to contribute to the diastolic depolarization^1–4 (for review see⁵). However, the importance of NCX function remains unresolved due to a lack of specific NCX blockers and the challenges in measuring I_NCX under different conditions.

2. Original paper by Gao et al.: rationale and conclusions

In a recent study published in Circulation Research, Gao et al.⁶ employed a global myocardial and SAN-targeted conditional Ncx1 knockout in mice to explore the role of NCX in basal and reserve operation of the cardiac pacemaker. Surprisingly, SA node cells (SANC) isolated from the KO mice exhibited no changes in action potential (AP) firing rates, cytosolic Ca²⁺ transient shapes, or ion current densities (other than I_NCX). However, the cells were insensitive to β-adrenergic receptor (β-AR) stimulation. Based on these and other results, including in vivo measurements of heart rate and numerical model simulations, the authors concluded that genetic inhibition of I_NCX disables fight-or-flight sinoatrial node activity without affecting the resting heart rate (title) and therefore physiological Ncx1 expression is required for increasing sinus rates, but not for maintaining resting heart rate (abstract).

3. Incomplete NCX KO cells in Gao et al

An important characteristic of the NCX KO model developed by Gao et al. is that NCX was not completely eliminated. This raises an important issue relevant for their data interpretation and numerical modelling: which genotypes of SANC were actually measured by Gao et al. In the cre-loxP paradigm, each SANC from KO mice had a disruption of 0, 1 or 2 of the NCX alleles. Since the responses of their cells differed from those in wild type SANC, these cells had disruptions in 1 or both alleles and therefore a bimodal distribution of cell behavior would be expected, an issue not addressed by Gao et al. Their measurements of I_NCX, in fact, performed only in a limited number of cells (n=15, their Fig. 1E), showed a very large spread of values from 0 to 2 pA/pF. On average the I_NCX density and, therefore, the density of functional molecules was 20% of that in SANC from wild type mice. Finally, because these measurements were performed only five days after tamoxifen treatment, some NCX molecules will still remain even in the cells with 2 disrupted alleles, due to kinetics of the NCX protein turnover. Specifically, the half-life of 33 hours for the NCX protein reported by Slodzinski and Blaustein⁷ would indicate that after 5 days the fraction of the remaining NCX molecules is expected to be at least 100%*1/(2^(120/33)) = 8%. Thus, the NCX KO model developed by Gao et al. likely achieved, in fact, a selective, substantial (albeit cell-to-cell variable) NCX inhibition, i.e. incomplete KO (iKO). Due to the incompleteness of NCX knock-out, this model demonstrates unique behaviors and provides new insights into cardiac pacemaker function that cannot be achieved with a true KO model. SANC isolated from mice of other types of NCX knockout have demonstrated different behaviours, such as dys-rhythmic beating⁸ or absence of automaticity⁹.

Figure 1.

Kurata et al. model¹¹ predicts that diastolic I_NCX[from maximum diastolic potential (MDP) to −45 mV] remains almost unchanged with different k_NCX (mimicking various levels of NCX expression). In original model k_NCX=125 pA/pF (100%, left panels). Mechanisms of I_NCXstabilization of diastolic I_NCX(AP firing rate)include a compensatory increase of [Ca²⁺] in the network SR and submembrane space. APs do not occur in the simulations when k_NCX was set to zero.

4. We argue

Since Gao et al. achieved substantial inhibition of NCX expression (verified by PCR, Western blot, and voltage clamp recordings), they assumed that this would also substantially inhibit I_NCX under physiological conditions. We argue, however, that the remaining NCX molecules within SANC from iKO mice, nevertheless, are sufficient to conduct nearly the same current during diastolic depolarization (DD) as occurs in wild type cells. Thus NCX still provides the same crucial contribution to cell automaticity in iKO cells.

Another interesting aspect of the iKO model is linked to cell Ca²⁺ regulation. Prior studies showed that NCX knockout in ventricular myocytes activates powerful compensatory mechanisms to ensure cell Ca²⁺ balance in the KO cells, which include a reduction in I_CaL and AP shortening¹⁰. In iKO SANC of Gao et al., however, both I_CaL amplitudes and AP shapes remain unchanged. The classical mechanism of Ca²⁺-induced inactivation of I_CaL is also expected to remain unchanged because Ca²⁺ transient shapes were found unchanged in these cells. Thus, if in the iKO cells the Ca²⁺ influx remains unchanged, but the major Ca²⁺ efflux mechanism via NCX is indeed at only 20% of that in wild type, then it is unclear what mechanism generates the bulk of Ca²⁺ efflux in these cells.

In this Research Commentary we provide numerical model simulations, which, we believe, resolve, at least in part, these paradoxical results of Gao et al. and suggest an alternative interpretation of their experimental data with respect to the relevance of I_NCX to the resting heart rate.

5. SANC is a complex dynamical system, I_NCX is activated by local Ca²⁺ releases (LCRs)

Pacemaker cells from numerous species, including rabbits and mice, generate spontaneous rhythmic LCRs beneath the cell surface membrane during DD (review⁵). Gao et al. measured I_NCX only under voltage clamp as a function of voltage at a fixed intracellular [Ca²⁺] (Fig. 1C–E in Gao et al.). The diastolic I_NCX in cells that fire spontaneous APs under physiological conditions, however, is determined not only by membrane potential, but also by the LCRs, which, in turn, are determined by a complex, local crosstalk that includes Ca²⁺-dependent functions of NCX, sarcoplasmic reticulum (SR) release channels (RyR), and pump (SERCA) molecules. The feedback mechanism between LCRs and I_NCX within this complex dynamical system can rapidly change I_NCX as voltage and Ca²⁺ are allowed to vary during AP firing.

6. Common pool model simulations by Gao et al. and Severi et al

These complex interactions can be explored by numerical model simulations. Gao et al.⁶ employed the 2002 Kurata et al. model¹¹ (Kurata model). In this model I_NCX = − k_NCX × F(V_m,Ca_sub), where k_NCX is a scaling factor that reflects the density of functional NCX molecules (per pF). The function F increases as submembrane space [Ca²⁺] (Ca_sub) increases, but decreases as the membrane depolarizes (V_m). F has an extremely complicated formulation; as a reference, it decreases roughly from 1 to −1 when membrane depolarizes from −100 mV to +100 mV at Ca_sub=1 μmol/l. Although Gao et al. did not simulate I_NCX, their numerical simulations show that a 5-fold reduction in k_NCX would not markedly affect the spontaneous AP firing rate. Recent studies¹², using a modification of Maltsev-Lakatta model¹³ with similar I_NCX formulations, have demonstrated that not only is spontaneous AP firing rate preserved, but also that the magnitude of diastolic I_NCX remains almost unchanged (Fig. 10C in¹²) when k_NCX is reduced to values as low as 25% of its normal value.

7. Our simulations of Kurata model: Mechanism of I_NCX stabilization and failure to keep Ca²⁺ low

Here we show that in Kurata model (utilized by Gao et al.) the average diastolic I_NCX in the iKO cells also remains almost unchanged (0.35 vs. 0.33 pA/pF, Fig. 1B,H). In simulations with I_NCX=0, AP firing stops (Fig. 1N), representing the case with a true NCX knockout. The results of these simulations point to the existence of powerful stabilization mechanisms that preserve the diastolic I_NCX in the context of a reduced NCX expression. Given that I_NCX = −k_NCX × F(V_m,Ca_sub), the preservation of I_NCX (as k_NCX decreases) requires a substantial increase in F. Since AP shape (i.e. V_m) remains almost unchanged, this is accomplished via an increase in Ca_sub.

Our additional simulations of Kurata model established the following mechanism of the required Ca_sub increase. NCX and SERCA compete for the Ca²⁺ released via RyR. When the density of NCX molecules decreases (mimicking iKO cells), the balance shifts in favor of SERCA. Hence, more Ca²⁺ becomes available for pumping into the SR by SERCA, which increases RyR Ca²⁺ release, leading to an increase in I_NCX. In our simulations of Kurata model, this stabilization mechanism is manifested by an increase in SR Ca²⁺ uptake flux (by ~50%, Fig. 1C,J), a higher SR Ca²⁺ load (Fig. 1D,K), and a persistent Ca²⁺ release (Fig. 1E,L). While these three processes increase diastolic Ca_sub in the iKO cells (Fig. 1F,M), this increase is associated with an unphysiologically high nadir (> 790 nmol/l, Fig. 1M). The failure to predict the required normal diastolic level of Ca²⁺ (Indo-1 ratio, Online Fig. V in Gao et al.) points to fundamental limitations of this model.

8. First limitation of Kurata model: insufficient SR Ca²⁺ pumping

One fundamental limitation of the Kurata model (developed in 2002) is its insufficient SR Ca²⁺ pumping (adopted from Luo-Rudy ventricular myocyte formulations¹⁴). In 2006 Vinogradova et al.¹⁵ discovered that phospholamban in SANC is highly phosphorylated in the basal state by PKA, suggesting a relatively higher rate of SR Ca²⁺ pumping. Thus, more recent pacemaker cell models (of Maltsev-Lakatta type)^{12,13,16–18} feature a higher SR Ca²⁺ pumping rate, reflecting these important experimental findings. Specifically, in Kurata model the maximum pumping rate P_up was 6 mM/s, while in our model it is 12 mM/s. The insufficient Ca²⁺ pumping accounts, in part, for the unrealistic diastolic Ca_sub predicted by Kurata model (Fig. 1M). For example, simulations using a recent update of the Maltsev-Lakatta model ¹² exhibit a much smaller, albeit still unphysiologically high, diastolic Ca_sub of 326 nmol/l (not shown) when k_NCX is reduced to 20% of its normal value.

9. Second limitation of Kurata model: lack of local Ca²⁺ control. Our new model

Another fundamental limitation of Kurata model and also of all models of Maltsev-Lakatta type^{12,13,16–18} is that they belong to so-called “common pool” models that simulate only total (“global”) Ca²⁺ signals, but contain no terms that explore local Ca²⁺ control and RyR recruitment in SANC (i.e. LCRs). To overcome this limitation, we employed a recent model of rabbit SANC¹⁹ that describes local Ca²⁺ dynamics (including diffusion and buffering) within the submembrane space, cytosol, and SR. In this model the SR has the ability to locally pump and release Ca²⁺, and RyRs are grouped in clusters (Ca²⁺ release units). Each Ca²⁺ release unit can generate a Ca²⁺ spark and several nearby sparks can create a multi-spark release i.e. an LCR. The LCRs are generated via locally propagating Ca²⁺-induced Ca²⁺ release (CICR) of individual sparks. Here, we supplemented this LCR model with the entire ensemble of ion currents described by an earlier Maltsev-Lakatta model¹³ in order to simulate pacemaker potentials.

10. Normal Ca²⁺ in our local control model

In contrast to 2002 Kurata model, our local Ca²⁺ control model predicts normal diastolic [Ca²⁺] levels (Fig. 2B) over a wide range of k_NCX. For example, at k_NCX=37.5 pA/pF (20% of k_NCX in Maltsev-Lakatta model) the model predicts a realistic nadir of 166 nmol/l for spatially averaged Ca_sub. (Of note, diastolic [Ca²⁺] in SANC ranges from 141 to 190 nmol/l²⁰). This realistic prediction happens not only because of accelerated SR Ca²⁺ pumping (mimicking phospholamban phosphorylation mentioned above), but also because the released Ca²⁺ is more efficiently controlled by SERCA. Specifically, pumping rate is determined by a sigmoid curve on cytosolic [Ca²⁺]. Since each LCR results in very high local Ca²⁺ concentration, it fully activates Ca²⁺ pumping of the SR in its vicinity. On the other hand, in common pool models (i.e. those lacking local Ca²⁺ control), a diastolic Ca²⁺ release of a similar magnitude is spatially averaged over the entire pool, and therefore the Ca²⁺ concentrations do not reach the high levels of those in the local model. Therefore the SERCA operates in the region close to the foot of the sigmoid curve associated to low levels of Ca²⁺ characterized by a relatively slow pumping rate.

Figure 2.

Reduction in NCX expression (parameter k_NCX) increases diastolic Ca²⁺ spark rate and recruitment of Ca²⁺ release units (CRU) to fire via CICR. Simulations depicted in the figure employed our recent model of LCRs¹⁹ supplemented with the entire ensemble of ion currents described by Maltsev-Lakatta model¹³. A,B: Simulated AP firing rates and the nadir of spatially averaged [Ca²⁺] in submembrane space (Ca_sub) are plotted as function of k_NCX values, mimicking various levels of NCX expression. Red line plot in panel A shows the effect of β-AR stimulation (labeled “ISO”). C–E: Simulations of time series for membrane potential (blue), I_NCX (green) and Ca_sub (red), with respective k_NCX values indicated at the top of each panel. Images in panel C illustrate the instantaneous local submembrane [Ca²⁺] at −40 mV by red shades, from black (200 nM) to red (1000 nM). Ca²⁺ release units, i.e. clusters of RyRs, are shown by dots in blue and white (when firing). Further details of 2D Ca²⁺ dynamics that include these images are illustrated in supplemental Movies. Simulations of resting state AP firing were performed with I_CaL conductance g_CaL=0.3712 nS/pF, I_Kr conductance g_Kr=0.08113973 nS/pF, voltage for I_f half-activation V_1/2h=−64 mV, I_spark=1.25 pA, restitution period T_r=250 ms, maximum SR Ca²⁺ pumping rate P_up=12 mM/s. Parameters for simulations of β-AR stimulation were chosen as previously suggested¹⁶: g_CaL=0.6496 nS/pF, g_Kr=0.1217096 nS/pF, V_1/2h=−56.2 mV, T_r=200 ms, and P_up=24 mM/s.

11. Additional I_NCX stabilization via local Ca²⁺ control

Our new model simulations have also discovered the existence of another stabilization mechanism of I_NCX (Fig. 2C and Supplemental Movies) via the local control of CICR²¹. During DD, the Ca²⁺ released via a RyR in SANC can recruit its neighboring RyRs to release more Ca^{2+ 19}. This local CICR mechanism generates spontaneous Ca²⁺ wavelets (i.e. LCRs), i.e. larger than Ca²⁺ sparks, but smaller than global, full-cell-length Ca²⁺ waves^2,3. The extent of this local RyR recruitment depends upon the extent to which Ca²⁺ released into the subspace diffuses to and thus interacts with neighboring RyRs. As the NCX “steals” Ca²⁺ in the vicinity of each RyR, it restrains CICR since the amount of Ca²⁺ available for diffusion and recruitment of further RyRs decreases. This restraint wanes as NCX expression becomes reduced and the spread of Ca²⁺ release between RyRs via CICR is enhanced, resulting in a more effective activation of the remaining NCX molecules by LCRs (Fig. 2C). In other words, the diastolic I_NCX increase is determined roughly by the formula: I_NCX = n_sparks × i_{NCX_spark}, where n_sparks is the number of sparks and i_{NCX_spark} is the NCX current generated by one spark. As we decrease k_NCX in our local model from 150 to 37.5 pA/pF (Fig. 2A, blue plot), the number of Ca²⁺ sparks (within all these LCRs) increases. However at the same time, a smaller k_NCX results in a smaller i_{NCX_spark}. Therefore, the two processes compensate each other and the product (n_sparks × i_{NCX_spark}) remains almost the same (Fig. 3).

Figure 3.

Simulations of our local control model demonstrate that diastolic I_NCX is preserved (purple arrow) within a broad range of NCX expression (given by parameter kNCX in the model). The I_NCX traces are the same as in Figure 2C,D (green) but shown with substantial zoom to clearly see the I_NCX magnitude and dynamics. For clear comparison within the diastolic period, the traces were synchronized at their peak induced by I_CaL (red arrow) via CICR mechanism at the beginning of an AP. I_NCX substantially decreased only when k_NCX decreased dramatically (black dotted curve).

12. I_NCX stabilization failure at extremely low NCX expression: arrhythmia and pacemaker failure

In our simulations, when k_NCX is reduced further to or below 9.375 pA/pF (Fig. 2D), i_{NCX_spark} continues to decrease, but n_sparks cannot further increase because it becomes equal to the total number of Ca²⁺ release units in the cell (i.e. all RyRs fire during DD). Thus the aforementioned mechanism of I_NCX stabilization becomes saturated. In this case the diastolic I_NCX becomes dramatically reduced (Figure 3, black dotted curve) and Ca²⁺ release becomes uncoupled (partially or fully, Fig. 2D,E) from AP generation. Specifically, I_NCX becomes too small to depolarize the membrane sufficiently to reach the L-type Ca²⁺ channel activation threshold (approximately −45 mV) and thus I_CaL activation and AP generation fail. The dys-rhythmic firing happens when diastolic I_NCX brings membrane potential close to the I_CaL threshold, making further I_CaL activation and AP generation unstable (Fig. 2D). A close inspection of Ca²⁺ traces recorded in iKO cells by Gao et al. reveals that some of the cells do exhibit dys-rhythmic beating. In their Online figure V (panel B), inter-peak times are approximately as follows: 192, 269, 270, 184 ms, showing a substantial beat-to-beat variation (up to 47%).

13. About the response to β-AR stimulation

Using our local control model, we also performed simulations of β-AR stimulation in SANC at different NCX expression levels (red-line plot in Fig. 2A). We found that the AP firing rate increase varies from 0 to 44% of control (rate before β-AR stimulation), which encompasses physiological responses ranging from 20 to 30% in the rabbit (our model is for rabbit SANC). Interestingly, this relationship is bell-shaped, indicating that both high and low levels of NCX expression inhibit β-AR stimulation. In the case of a low NCX expression (k_NCX = 18.75 pA/pF) our numerical model closely reproduces Gao et al. experimental result of a complete absence of response to β-AR stimulation (shown by the vertical red arrow in Fig. 2A). On the other hand, excessive NCX expression in the model inhibits response by preventing the increased diastolic release that normally occurs via LCRs (observed experimentally ²²). In this case NCX removes Ca²⁺ very efficiently and prevents local CICR so that wavelet-like LCRs do not occur.

14. Summary

In summary, given the new stabilization mechanism of I_NCX revealed by our simulations, the spontaneous AP rate is likely regulated differently from what Gao et al. suggest. In general, the rate is regulated by the duration of the DD, and more specifically by the DD rate that is determined by the sum of transmembrane ion currents. Gao et al. argues that I_NCX is “dispensable” (i.e. provides only a small contribution) because the AP rate and DD rate remain unchanged despite their “genetic inhibition of NCX current” (cited from their title). In other words, if I_NCX were a major pacemaker current, the AP firing rate must have decreased, but it remained unchanged. While their logic is correct, we argue that Gao et al. have not demonstrated “genetic inhibition of NCX current” under physiological conditions, i.e. during spontaneous beating of SANC. Our numerical simulations can be interpreted to indicate that I_NCX contribution to the DD is crucially important for basal automaticity of SANC and the remaining NCX molecules in Gao et al. iKO model still produce the same I_NCX during DD as in wild type cells (Fig. 3). The I_NCX preservation, however, comes at a cost: When NCX expression is low, the remaining NCX molecules become fully utilized under resting conditions (Fig. 2C, “Everywhere Wavelets”), leaving no reserve to support a further rate increase, e.g. in the presence of β-AR stimulation (Fig. 2A, shown by red arrow), which is similar to the findings of Gao et al.

15. Future studies

While we demonstrate the importance of local Ca²⁺ control mechanisms for NCX function in SANC, the details of the complex crosstalk between NCX and Ca²⁺ cycling molecules in cardiac pacemaker cells await further elucidation in new experimental and numerical modeling approaches. Specifically, with respect to the iKO cells reported by Gao et al, it would be interesting to test whether these indeed exhibit larger size LCRs and dys-rhythmic firing (Fig. 2D). Another interesting comparison would be with respect to ryanodine sensitive current (i.e. I_NCX) during DD³. If compensatory mechanisms actually exist, the diastolic I_NCX in the iKO cells must be much larger than 20% of that in wild type cells. Novel numerical modeling approaches ought to extend the full local control theory²¹ to pacemaker cells to explore further important details of local crosstalk NCX-RyR-SERCA-L-type Ca²⁺ channels in 3D. Adding terms to simulate the effect of protein phosphorylation will be also important for future models to be realistic.

16. Postscript: the case of I_NCX =0

While the case of a complete NCX KO^8,9 is indeed interesting (I_NCX=0, Fig. 1N–S and Fig. 2E), we have not elaborated on it further in our short Commentary, because (i) the original study by Gao et al. explored only the effects of partial NCX inhibition and (ii) cells with complete NCX knockout exhibit dys-rhythmic beating ⁸, complete absence of automaticity⁹, substantial down-regulation of I_CaL⁹, and persistent periodic Ca²⁺ release in the absence of automaticity²³, i.e. behaviors different from those reported by Gao et al. Thus, while our simulations do reproduce some of these behaviors (dys-rythmic beating and absence of automaticity accompanied by persistent periodic Ca²⁺ Release, Fig. 2D,E), this special case of I_NCX=0 requires a separate, dedicated consideration of all the unique properties of such cells, including additional Ca²⁺ efflux mechanisms such as sarcolemmal Ca²⁺ ATPase not included in our present model.

Non-standard abbreviations and non-standard acronyms

AP

Action potential

β-AR

β-adrenergic receptor

Ca_sub

Submembrane space [Ca²⁺]

CICR

Ca²⁺-induced Ca²⁺ release

CRU

Ca²⁺ release unit

DD

Diastolic depolarization

F(V_m,Ca_sub)

Na⁺/Ca²⁺ exchanger current dependence on [Ca²⁺] and membrane potential

I_CaL

L-type Ca²⁺ current

iKO

Incomplete knock-out

I_NCX

The whole-cell Na⁺/Ca²⁺ exchanger current

i_{NCX_spark}

The Na⁺/Ca²⁺ exchanger current generated by one Ca²⁺ spark

k_NCX

The whole cell Na⁺/Ca²⁺ exchanger current scaling factor

KO

Knock-out

LCR

Local Ca²⁺ release

NCX

The Na⁺/Ca²⁺ exchanger

n_sparks

Number of sparks within all local Ca²⁺ releases

MDP

Maximum diastolic potential

RyR

Sarcoplasmic reticulum Ca²⁺ release channel

SANC

Sionatrial node cell

SERCA

Sarco/Endoplasmic Reticulum Ca²⁺ -ATPase

SR

Sarcoplasmic reticulum

V_m

Membrane potential