The Grapefruit Flavonoid Naringenin Inhibits Multiple Cardiac Ion Channels

Drinking fresh grapefruit juice is associated with a signicant prolongation of the QT segment on the electrocardiogram (ECG) in healthy volunteers. Among the prominent avonoids contained in citrus fruits, the avanone naringenin is known to be a blocker of the human ether-a-go-go related gene (hERG) potassium channel. We hypothesized that naringenin could interfere with other major ion channels shaping the cardiac ventricular action potential (AP). To this end, we examined the effects of naringenin on the seven currents comprising the Comprehensive in vitro Pro-Arrhythmia (CiPA) panel for early arrhythmogenic risk assessment in drug discovery and development. We used automated patch-clamp of human ion channels heterologously expressed in mammalian cell lines to evaluate half-maximal inhibitory concentrations (IC 50 ). Naringenin blocked all CiPA currents tested with IC 50 values in the 30 µM – 100 µM concentration-range. The rank-order of channel sensitivity was the following: hERG > K ir 2.1 > Na V 1.5 late > Na V 1.5 peak > K V 7.1 > K V 4.3 > Ca V 1.2. This multichannel inhibitory prole of naringenin suggests exercising caution when large amounts of grapefruit juice or other citrus juices enriched in this avanone are drunk in conjunction with QT prolonging drugs or by carriers of congenital long QT syndromes.


Introduction
Taking certain medication or carrying inheritable mutations in genes encoding cardiac ion channels are not the sole circumstances that can lead to prolongation of the QT segment on the ECG. As recently emphasized in the concept of "arrhythmogenic foods", components of common diet can also affect this important cardio-safety metric (Tisdale, 2019;Woosley, 2020). Grapefruit-based beverages are probably the most emblematic example in this regard. Drinking grapefruit juice was rst shown to enhance the QT length brought about by terfenadine through drug-metabolism interference due to the well-known cytochrome P450 inhibition produced by grapefruit juice (Benton et al., 1996). However, it was later shown that the avonoid naringenin, which is abundant in citrus fruits and notably in grapefruit, is actually by itself a blocker of the human ether-a-go-go related gene (hERG) channel (Zitron et al., 2005). Outward potassium currents owing through hERG channels are primarily responsible for the repolarization of the cardiac ventricular action potential (AP), and drug-induced inhibition of hERG function (or tra cking to the cardiomyocyte membrane) have historically been known to be associated with QT prolongation (Rampe & Brown, 2013).
The QT prolonging effect of grapefruit intake was recently con rmed in a "thorough QT" (TQT) study.
Clinical TQT studies are rigorously controlled and adequately powered studies designed to anticipate the proarrhythmic risk of new chemical entities (NCEs) during their development. Notably, the sensitivity of TQT studies is monitored by the contemporary administration of a positive control (Darpo, 2010).
Regarding grapefruit juice speci cally, Chorin et al. (Chorin et al., 2019) recently established in a well powered randomized TQT study in healthy volunteers, that the length of the QT segment on the ECG corrected for heart rate (i.e. QTc) increased signi cantly as soon as three hours after drinking 1 liter of grapefruit juice. Although the effect was relatively modest (QTc increase reached a maximum of 14 ms after ingestion of another 0.5 L juice), it was notably greater in a subgroup of patients suffering congenital long QT syndrome. Among the latter, the QTc increase peaked at nearly 22 ms, an effect that was comparable to the QTc prolongation brought about by the positive control moxi oxacin.
Several other currents besides hERG-mediated I Kr signi cantly impact the AP duration and are increasingly considered as important cardio-safety targets for NCEs. In this perspective, the Comprehensive in vitro Pro-Arrhythmia (CiPA) initiative has been de ned and is increasingly being adopted early in the discovery process to re ne the prediction of arrhythmogenic risk of NCEs heading towards the clinic (Fermini et al., 2016). The CiPA paradigm aims at better evaluating the degree of ECG surveillance that will be needed later to safely accompany the development of NCEs. Interestingly enough, the CiPA paradigm predicts that drug effects at multiple cardiac ion channels may compensate each other, a contention that has received support from the retrospect pro ling of a series of marketed drugs that were revealed safer than their sole "hERGo-centric" pharmacology would have suggested contained: CsF, 150 ; EGTA/CsOH, 1/5 ; NaCl, 10 ; MgCl 2 , 1 ; CaCl 2 , 1 ; HEPES, 10, and the pH was adjusted to 7.2 with CsOH. For the recording of the late component of Na V 1.5, intracellular CsF was reduced to 130 mM and the extracellular buffer was supplemented with 10 nM of the sea anemone toxin ATX-II to slow channel inactivation (Wu et al., 2019). All channels were exposed to 6 concentrations of Naringenin applied cumulatively in an ascending order up to 300 µM. At the end of each recording, a reference inhibitor was added at a maximally active concentration to isolate leak currents if any. E-4031 (10 µM) was used for the hERG currents, lidocaine (3 mM) for the Na V 1.5 currents, CdCl 2 (0.2 mM) for Ca V 1.2, SKF-96365 (30 µM) for K V 4.3, HMR-1556 (30 µM) for K V 7.1 and BaCl 2 (3 mM) for the K ir 2.1 channels.
Racemic naringenin (CAS Number 67604-48-2) was obtained from Sigma-Aldrich (catalog # N5893). Concentrated stock solutions prepared in pure DMSO were diluted in extracellular buffer containing 1% Pluronic F-68 so as to obtain the following six nal concentrations which were applied to the cells in ascending order : 1.

Results And Discussion
We observed that naringenin inhibits all seven cardiac CiPA currents tested in a concentration-dependent manner. This nding con rms our hypothesis that this prominent avonoid of grapefruit and other citrus fruits can indeed alter the function of multiple other ion channels generating the ventricular AP besides hERG. Block potency values fall within a half-log window ranging approximately between 30 µM and 100 µM. Fig. 1 illustrates typical current traces collected before and during exposure to naringenin. Most concentration-response curves presented similar Hill slope factors around unity, suggesting that naringenin interact with single target sites on these channels (see table for IC 50 and n H values).
The K ir 2.1-mediated outward component of the inward recti er I K1 , as well as the hERG-mediated, rapidly activating component of the delayed recti er I K were the most sensitive currents to naringenin. The K V 7.1mediated, slowly activating I Ks component of I K was less potently inhibited by the avonoid. All three currents contribute to the terminal repolarization phase of the cardiac action potential and constitute the "repolarization reserve" (Roden, 2008). The potency of naringenin on hERG we determined here by automated patch-clamp in CHO cells con rms early ndings by two-electrode voltage-clamp in Xenopus oocytes (Zitron et al., 2005). To our knowledge, the inhibitory effects of naringenin on K ir 2.1 and on the ve other cardiac currents comprising the CiPA panel has not been documented before. Our data indicate that, among the inward cationic currents, the Ca V 1.2-mediated I Ca,L current appeared the least sensitive to naringenin, followed by the I Na,peak and I Na,late currents transiting through Na V 1.5 channels. Overall, the inhibitory potency of naringenin on these three depolarizing currents is roughly 2 -3 fold weaker than its effect on hERG outward currents which are prominently associated with QT prolongation. Although it is uncertain whether the IC 50 differences are signi cant, our data suggest that the inhibitory pro le of naringenin at these multiple channels incompletely compensate for each other, leaving an overall effect of prolonging the QT by a small, but signi cant extent.
Patients developing life threatening Torsade de Pointe (TdP) in response to drugs often present with additional risk factors such as hypokalemia, bradycardia or inheritable mutations in genes encoding cardiac ion channels. Regarding the latter, genotype-phenotype correlation studies have estimated that the prevalence of congenital long-QT could reach 1 : 2500 (Schwartz et al., 2009). Therefore, many clinically silent carriers of inheritable risk factors could develop proarrhythmic events when ingesting QTprolonging substances. De facto, up to 1 : 3 patients presenting with drug-induced long-QT syndromes actually harbor mutations in ion channel genes related to congenital long-QT (Itoh et al., 2016). Moreover, it is noteworthy that loss-of-function mutations in KCNQ1 or KCNE1, the gene encoding the pore-forming alpha-subunit K V 7.1 and its gating-modulating ancillary subunit minK, respectively, and in KCNH2 (i.e. hERG), are associated with the vast majority of autosomal dominant or autosomal recessive Romano-Ward or Jervell and Lange-Nielsen long-QT syndromes (Bokil et al., 2010). We found that both channels are targeted by naringenin with half-maximal inhibitory concentrations ranging between 30 µM and 100 µM.
In summary, although the plasma concentrations of naringenin after drinking 1 -2 L grapefruit juice were not monitored in the available TQT study (Chorin et al., 2019), the signi cant QTc prolongation previously observed in the clinic combined with our present ion channel pro ling data suggest that known carriers of congenital long-QT syndromes should avoid dietary sources enriched in this avonoid. Furthermore, taking high volumes of fresh grapefruit juice with medications known to prolong the QT interval should be discouraged. Figure 1 Effects of Narigenin on seven CiPA currents. Traces in each panel are representative population patch currents elicited by the voltage-protocol shown in inset next to them. Green traces were recorded in control buffer and superposed to red traces collected in the presence of the indicated concentration of naringenin. Panel A = Na V 1.5 peak ; B =Ca V 1.2 ; C = Na V 1.5 late ; D = hERG ; E = K ir 2.1 outward ; F = K V 4.3 ; G = K V 7.1.