Arrhythmogenic mechanisms of interleukin-6 combination with hydroxychloroquine and azithromycin in inflammatory diseases

Inflammatory diseases including COVID-19 are associated with a cytokine storm characterized by high interleukin-6 (IL-6) titers. In particular, while recent studies examined COVID-19 associated arrhythmic risks from cardiac injury and/or from pharmacotherapy such as the combination of azithromycin (AZM) and hydroxychloroquine (HCQ), the role of IL-6 per se in increasing the arrhythmic risk remains poorly understood. The objective is to elucidate the electrophysiological basis of inflammation-associated arrhythmic risk in the presence of AZM and HCQ. IL-6, AZM and HCQ were concomitantly administered to guinea pigs in-vivo and in-vitro. Electrocardiograms, action potentials and ion-currents were analyzed. IL-6 alone or the combination AZM + HCQ induced mild to moderate reduction in heart rate, PR-interval and corrected QT (QTc) in-vivo and in-vitro. Notably, IL-6 alone was more potent than the combination of the two drugs in reducing heart rate, increasing PR-interval and QTc. In addition, the in-vivo or in-vitro combination of IL-6 + AZM + HCQ caused severe bradycardia, conduction abnormalities, QTc prolongation and asystole. These electrocardiographic abnormalities were attenuated in-vivo by tocilizumab (TCZ), a monoclonal antibody against IL-6 receptor, and are due in part to the prolongation of action potential duration and selective inhibition of Na+, Ca2+ and K+ currents. Inflammation confers greater risk for arrhythmia than the drug combination therapy. As such, in the setting of elevated IL-6 during inflammation caution must be taken when co-administering drugs known to predispose to fatal arrhythmias and TCZ could be an important player as a novel anti-arrhythmic agent. Thus, identifying inflammation as a critical culprit is essential for proper management.


Results
In-vivo impact of the combination of IL-6, hydroxychloroquine (HCQ), and azithromycin (AZM) on guinea pig ECG. Because IL-6, AZM and HCQ are known to individually affect multiple ion channels 28,31,32 and because the net electrophysiological effect on the ECG is unknown, we set to first test these compounds in combination in-vivo and then evaluate their effects on the ECG. At baseline conditions, ECGs were first recorded from all the six, initially untreated, guinea pigs which served as their own control (Fig. 1a). Subsequently, guinea pigs were sequentially and cumulatively injected intravenously first with IL-6 (184 μg/kg) for 40 min to mimic the clinical settings where inflammation precedes drug therapy, followed by intravenous injection with AZM (1-time clinically relevant dose (CRD), 38.2 mg/kg) and intraperitoneal injection with HCQ (0.5-times CRD, 22.9 mg/kg) for 30 min followed by the cumulative injection of two higher dose of HCQ (1-time CRD, 45.8 mg/kg and 2-times CRD, 91.6 mg/kg) at 30 min intervals. ECGs were taken after each injection. IL-6 alone significantly reduced heart rate (Fig. 1b,f), prolonged PR interval (Fig. 1b,g) and QTc (Fig. 1b,i). IL-6 combined with AZM and HCQ (0.5-time CRD) or HCQ (1-time CRD) or HCQ (2-time CRD) caused further significant and dose-dependent bradycardia, PR, QRS and QTc prolongations, and complete atrioventricular dissociation (Fig. 1c-i and Table 1A). Interestingly, the administration of only the combination of AZM and HCQ (0.5-time CRD) without IL-6 in another set of six guinea pigs, resulted in no effect on heart rate and QRS, lesser PR prolongation (ΔPR = 6 ms vs. 20 ms with IL-6) and non-significant QTc prolongation (ΔQTc = 12 ms vs. 31 ms with IL-6) indicating that IL-6 significantly amplifies the abnormal ECG phenotype (Fig. 1j-m and Table 1A, B). To assess whether AZM and HCQ combination may have per se affected IL-6R expression, we measured the corresponding mRNA and proteins levels. IL-6R mRNA levels in hearts of guinea pigs injected with AZM (1) (38.2 mg/kg) and HCQ (2) (91.6 mg/kg) showed no significant change in transcript levels of IL-6R (1.0-fold and 0.96-fold change when mRNA was normalized to control hearts, experiments were performed in triplicates, p = 0.4, Fig. 2a). Likewise, protein levels of IL-6R were measured in western blots and densitometric analysis showed no significant change between the control and guinea pigs treated with AZM and HCQ (the mean intensity for the control guinea pigs was 0. 46  To test whether IL-6 worsening of the electrocardiographic abnormalities of AZM and HCQ can be attenuated pharmacologically, TCZ (10 mg/kg), the IL-6R inhibitor, was administered intravenously in-vivo to six guinea pigs for 10 min first, followed by IL-6, AMZ and HCQ as outlined above. TCZ alone had no significant effects on heart rate ( Fig. 3b,g), PR interval (Fig. 3b,h), QRS (Fig. 3b,i) and QTc (Fig. 3b,j) compared to basal conditions (Fig. 3a,g-j). However, TCZ prevented the IL-6 (184 μg/kg) from reducing heart rate, prolonging PR interval and QTc (Fig. 3c, g-j and Table 1C). Similarly, TCZ also prevented the combination of IL-6, AZM (1-time CRD) and HCQ (0.5 or 1-time CRD) from reducing the heart rate and prolonging PR interval and QRS, attenuated QTc prolongation but not significantly when using one-way repeated measures analysis of variance ( Fig. 3 and Table 1C). However, TCZ prevented the atrioventricular dissociation induced by 2-times HCQ (Figs. 1e, 3f).
In-vitro impact of the combination of IL-6, AZM and HCQ on Langendorff perfused guinea pig hearts. Next, we aimed at assessing the direct impact of the combination of IL-6, AZM and HCQ on the heart by recording surface electrograms from isolated Langendorff perfused 5 guinea pig hearts (Fig. 4a). We first perfused the hearts with IL-6 (200 μg/L) alone for 40 min 28 , then cumulatively added AZM (1-time clinically relevant concentration (CRC), 41.5 mg/L) followed immediately by HCQ (0.5-time CRC, 24.9 mg/L) then by HCQ (1-time CRC, 49.8 mg/L) and HCQ (2-time CRC, 99.6 mg/L) at 8-10 min intervals. IL-6 resulted in significant PR (Fig. 4b,g) and QTc prolongations (Fig. 4b,i). The addition of AZM and HCQ (0.5-, 1-and 2-times CRC) to IL-6, resulted in a marked and significant concentration-dependent bradycardia, PR, QRS and QTc prolongations ( Fig. 4c-i), followed by a complete atrioventricular dissociation and asystole in 5/5 hearts ( Fig. 4e and Table 2A). Similar to the in-vivo studies above, the combination of only AZM and HCQ (0.5-time CRD) In-vitro impact of the combination of IL-6, AZM and HCQ on action potentials of guinea pig left ventricular myocytes. Next, we investigated the underlying mechanisms of the observed electrocardiographic abnormities seen in-vivo and in-vitro with the combination of IL-6, AZM and HCQ. The duration of AP is a surrogate for the QT interval and is impacted by I Na , I CaL and I Kr currents. As such the net impact of IL-6, AZM and HCQ on individual ion channels can be assessed accurately. The perfusion of IL-6 (200 μg/L) alone for 40 min resulted in a statistically significant prolongation of the APD 90 without significant change in the AP amplitude (n = 9). The cumulative addition of AZM (1-times CRC, 41.5 mg/L), HCQ (0.1-time CRC, 5.0 mg/L) and HCQ (0.5-time CRC, 24.9 mg/L and 1-time CRC, 49.8 mg/L) resulted in a concentration-dependent prolongation of the APD 90 and a decrease in AP amplitude (Fig. 5a,d) . The combination of IL-6 + AZM + HCQ when HCQ concentration reached 1-time CRC (49.8 mg/L) resulted in the failure of AP to completely repolarize ( Fig. 5a and Table 3A). Note that HCQ at 2-times CRC (99.6 mg/L) in combination with IL-6 + AZM completely abolished AP and hence no related data are provided. Noteworthy again the perfusion of only the combination AZM and HCQ (0.1-time CRD) without IL-6, in another set of 7 myocytes, resulted in lesser APD 90 prolongation (ΔAPD 90 = 204 ms vs. 312 ms with IL-6) confirming that IL-6 amplifies the prolongation of APD 90 and the reduction in AP amplitude (Fig. 6a,b and Table 3A,B).
In-vitro effects of the combination of IL-6, AZM and HCQ on I CaL , I Na and I Kr on guinea pig left ventricular myocytes. Because the observed ECG and AP effects of the combination of IL-6, AZM and HCQ caused conduction delays, QTc and APD 90 prolongations, we focused on characterizing the corresponding currents such as I Na , I CaL and I Kr , which can affect conduction, APD and also QT interval respectively. Preincubation of IL-6 (200 μg/L) alone for 40 min significantly reduced densities of I CaL (Figs. 4e, 5b and Table 3A) and I Kr (Fig. 5h,l and Table 3A) without significant change in I Na density (Fig. 5c/5f. and Table 3A). Myocytes pre-  In-vivo impact of the combination of IL-6, HCQ, AZM, and TCZ on guinea pig ECG. All values are expressed as mean ± SE, n = 6 guinea pigs for (A), n = 6 guinea pigs for (B) and n = 6 guinea pigs for (C).

HR (bpm) PR (ms) QRS (ms) QTc (ms)
A. Basal 150 ± 12.5 81 ± 1.7 12 ± 0.3 284 ± 9.  (2) n/a n/a n/a n/a www.nature.com/scientificreports/ www.nature.com/scientificreports/ incubated with IL-6 and subsequently superfused with AZM (1-times CRC, 41.5 mg/L) and HCQ (0.1 or 0.5 or 1-times CRC, 5.0 or 24.9 or 49.8 mg/L) for 5-8 min periods for each concentration, significantly decreased I CaL , I Na and I Kr densities in a concentration-dependent manner (Fig. 5b,c,e-l and Table 3A). Finally, the perfusion of only the combination AZM and HCQ (0.1-time CRD) without IL-6, in another two sets of 6 and 5 cardiac myocytes, resulted in lesser reduction in I CaL density (ΔI CaL = 1.1 pA/pF vs. 1.5 pA/pF with IL-6) and similarly lesser reduction in I Kr densities (ΔI Kr = 0.09 pA/pF vs. 0.2 pA/pF with IL-6). Because IL-6 did not have any significant effect on I Na , no conclusion for this current was drawn (Fig. 6d). Collectively, IL-6 continued to amplifies the inhibitory effect on both I CaL and I Kr (Fig. 6c,e and Table 3A,B).

Materials and methods
Animal and ethical approval. Equal number of female and male Dunkin-Hartley guinea pigs (300-350 g) aged 4-5 weeks from the experimental animal center of Nanjing University of Chinese Medicine (Nanjing, China) were used in this study. All experiments involving animals were approved by the Animal-Care and Use-Committee of Nanjing University of Chinese Medicine (#202005A045), whose policies adhere to the USA National Institutes of Health Guide for the Care and Use of Laboratory Animals. The study was carried out in compliance with the ARRIVE guidelines. TCZ at 10 mg/kg was used for in-vivo ECG studies as previously described 35 .

Electrocardiographic (ECG) recordings and drug administration. For in-vivo recordings, ECG
(lead II) was used in anesthetized guinea pigs (20% urethane 5 mL/kg, intraperitoneally) by placing the leads at the left forelimb as the positive electrode, the right forelimb as the negative electrode and the hindlimb as the reference electrode. The heart rate (HR), PR, QRS, QT and QTc intervals were analyzed using Bazett formula (QTc = QT/RR 1/3 , RR in s). Lyophilized azithromycin lactobionate was dissolved in saline and administrated intravenously. Hydroxychloroquine sulfate was dissolved in saline and injected intraperitoneally. Recombinant human IL-6 and tocilizumab were dissolved in saline and injected intravenously. At the completion of experiments, the deeply anaesthetized animals were sacrificed by cervical dislocation. Electrophysiological studies. For in-vitro electrogram recordings, isolated Langendorff perfused guinea pig hearts were used. The heart from anaesthetized guinea pigs was excised, cannulated through the aorta above the coronary ostia and perfused at a constant pressure (70 mmHg) with a solution (37 °C) containing (in mM): NaCl 117, CaCl 2 1.8, KCl 5.7, NaHCO 3 4.4, MgCl 2 1.7, HEPES 20, NaH 2 PO 4 1.5, glucose 11, gassed with 95% O 2 plus 5% CO 2 , pH7.4 adjusted with NaOH. Drugs were added in the perfusion solution via the retrograde perfusion of the coronary artery. Electrograms were obtained using a positive electrode placed at the apex, the negative electrode at the right atrium and the reference electrode at the root of the aorta. The analysis of electrogram parameters was similar to those in in-vivo experiments.
Action potential recordings from single ventricular myocyte. Guinea pig ventricular myocytes were obtained by enzymatic dissociation. Whole-cell current clamp configuration was used to record action potential (AP) using an amplifier (Axopatch 200B, Axon Instruments) as previously described 32  For I Kr recording, the external solution contained (mM): NaCl 145, KCl 4.5, MgCl 2 1, CaCl 2 1.8, Glucose 10, HEPES 10, pH 7.4 adjusted with NaOH. The internal solution contained (mM): KCl 140, MgCl 2 1, EGTA 11, HEPES 10, CaCl 2 1, MgATP 5, K 2 ATP 5, pH 7.30 with KOH. Currents were recorded in the whole-cell configuration of the patch-clamp technique using an Axopatch-200B amplifier (Axon Instruments, Inc, CA, USA). I Kr was recorded from a holding potential (HP) of − 50 mV using a short 200 ms depolarizing pulses from − 40 to + 70 mV in a 10-mV increment before returning to − 40 mV for the tail current recording. I CaL was blocked by the addition of 5 μM nifedipine in the bath solution and the slow delayed rectifier K current (I Ks ) was blocked with 100 μM chromanol. The I Kr current at + 70 mV was used for current density analysis.
Data and statistical analyses. ECGs were analyzed using LabChart Pro software (RM6240BD, Biosignal analysis system, China) and AP and ion currents were analyzed using Sigmaplot 12.5 (Sigmaplot, Northampton, MA, USA). One-way repeated measures analysis of variance was used to compare the multiple interventions in the same preparation. Student's paired t-test was used to compare before (baseline) and after the interventions in the same preparation. Unpaired t-test was used to compare control and the interventions between independent preparations. Data are presented as means ± SE. A value of p < 0.05 was considered significant.

Discussion
Here, we report a novel mechanistic explanation of the potentially fatal arrhythmogenic effect (severe bradycardia, conduction disturbances, QTc prolongation and cardiac arrest) of IL-6 in the settings of inflammation/ cytokine storm in the presence of drugs known to predispose to cardiac arrhythmias. Specifically, IL-6 amplifies the arrhythmogenic impact of AZM and HCQ, and TCZ attenuated these arrhythmic events in-vivo. The observed electrocardiographic abnormalities are explained by the underlying APD prolongation and the inhibition of key ion currents (I Na , I CaL and I Kr ) at the single ventricular myocyte. Our data provide evidence that in the setting of inflammation characterized by high IL-6 titers, the use of medications such as AZM + HCQ increases the arrhythmogenic risk with the potential for cardiac arrest without any significant changes in IL6R gene expression by AZM + HCQ. This observation may be compounded in COVID-19 patients with the increased risk for cardiac arrhythmias secondary to acquired conditions such as the use of antiviral and other concomitant QTprolonging drugs, fever, electrolyte imbalance, co-morbidities, and/or inherited arrhythmogenic syndromes 7-9,38 . www.nature.com/scientificreports/ The most important determinant of risk for malignant arrhythmias in patients with acquired QT-prolongation, is the use of one or more QTc prolonging drugs 29 in the setting of severe manifestations of COVID-19 9 . Our data show that the combination of IL-6 + AZM + HCQ, resulted in dose-dependent bradycardia, conduction abnormalities, QTc prolongation and cardiac arrest (asystole) but without any episodes of TdP which is consistent, so far, with the clinical data from COVID-19 patients where TdP events are rare 3,39 .
To date, there is still an unmet need for understanding of the electrophysiological impact of cytokine storminduced IL-6 overactivation in COVID-19 patients. The cytokine storm or cytokine release syndrome refers to an excessive and overwhelming release of proinflammatory mediators by an overly activated immune system 40 . Although, the immunologic mechanism of cytokine storm caused by COVID-19 is not fully understood, high IL-6 titers have been reported in COVID-19 patients [18][19][20]41 and IL-6 is known to prolong APD as a result of inhibition of I Kr 28 , and prolong QTc interval during acute infections regardless of concomitant antimicrobial therapy 42 . APD prolongation observed with IL-6 + AZM + HCQ led to QTc prolongation thereby increasing the arrhythmic risk for COVID-19 patients. The IL-6 effect on I Kr is mediated through binding to IL-6R which interacts with membrane-bound gp130 and its downstream Janus-kinase signaling pathways 28,43 . In support of the role of proinflammatory IL-6 in COVID-19 patients, TCZ which is a recombinant humanized monoclonal anti-IL-6R antibody approved by FDA for the treatment of severe or life-threatening chimeric antigen-receptor-T-cell-induced cytokine release syndrome 44 , is being tested in COVID-19 patients [21][22][23][24] . Our data demonstrating that TCZ prevented the electrocardiographic abnormalities induced by IL-6 + AZM + HCQ suggest that targeting key molecules within the inflammatory cytokine storm such as IL-6, could be a novel strategy for limiting the direct electrophysiological and arrhythmogenic effects of IL-6 in COVID-19 patients 26 and beyond. Accordingly, two clinical studies demonstrated the effectiveness of TCZ in rapidly reversing QTc prolongation in rheumatoid arthritis patients, by controlling systemic inflammation 45,46 .
HCQ has an established safety profile at appropriate doses and is used to treat malaria, and autoimmune diseases such as rheumatoid arthritis and lupus erythematosus 47,48 . Because HCQ demonstrated antiviral activity 49 and ability to regulate the immune system 50 , it was assumed that it may be useful in the treatment of COVID-19. Based on initial favorable outcomes from a small number of non-randomized clinical studies 17,51,52 , FDA had issued an emergency use authorization to HCQ for hospitalized COVID-19 patients on March 28, 2020, to just revoke this authorization on June 15th 2020 based a randomized double blind HCQ clinical study which showed no additional benefit for COVID-19 patients compared to placebo control 12,53 . The macrolid antibiotic drug AZM is employed in the clinical practice worldwide to treat different types of common bacterial infections, more frequently pneumonia and other respiratory infections. More recently, AZM has been used in combination with HCQ for the treatment of COVID-19 but there was no consensus on the clinical outcomes with some studies demonstrating that the combination of AZM + HCQ resulted in a favorable clinical outcomes 17,52 and others reporting arrhythmia safety concerns 4,7,15 . The risk of QTc prolongation and TdP has been reported for both HCQ and AZM when used alone 54 but QTc prolongation seems to be the major concern when these drugs are used in combination, specifically in COVID-19 patients 39,55 . In fact, several recent studies reported that this combination therapy was associated with significant QTc lengthening (on average ~ 25-35 ms increase when compared to baseline) and high rates of marked QTc prolongation > 500 ms (~ 10-35% of treated subjects 39,56,57 ). Conversely, in a previous study involving 116 healthy controls, co-administration of AZM(500-1,500 mg/ day) + chloroquine(1,000 mg/day) induced only mild QTc lengthening(5-9 ms) 58 , thereby supporting a key role for inflammatory activation, specifically IL-6 in boosting QT-prolonging potential of these drugs. In agreement with these findings are our in-vivo experiments demonstrating that while the combination of IL-6 + AZM + HCQ resulted in a marked QTc prolongation, AZM + HCQ when administered alone without IL-6, only mild and not significant QTc changes were observed.
While only rare cases of TdP have been reported in COVID-19 patients 3,56,59 , other arrhythmic episodes included cardiac arrest, significant bradyarrhythmias, and non-sustained ventricular tachycardia were also reported 3 . These clinical findings are consistent with our observation that the combination of IL-6 + AZM + HCQ, but not AZM + HCQ, results in marked bradycardia, QTc prolongation and cardiac arrest (asystole). Bradycardia and conduction disturbances are likely due to HCQ inhibition of the "funny" current, I f 31 and/or AZM inhibition of I CaL 32 . With the concentration of 200 µg/L of IL-6, we observed an inhibition of I CaL which may compound the effects of AZM in causing further bradycardia and atrioventricular conduction abnormalities. The observed QTc prolongation with the combination of IL-6 + AZ + HCQ both in-vivo and in-vitro is due to the inhibition of I Kr , which resulted in the prolongation of ventricular-myocyte AP and QT interval. Because the combination of IL-6 + AZM + HCQ also inhibited I CaL , this is expected to counterbalance the extent of AP prolongation and hence QTc. This multi-channel affinity with opposing effects may, in part, explain the absence or rarity of TdP arrhythmia. The combination of IL-6 + AZM + HCQ also inhibited I Na likely resulting, together with the concomitant I CaL inhibition, in conduction abnormalities like PR interval prolongation. It is also plausible that I Na inhibition in addition to I CaL inhibition may have both reduced susceptibility to TdP. In this regard, it has been previously proposed that multi-channel effects must be considered when evaluating the TdP risk 29 . Notably, beyond the combination therapy for COVID-19, AZM and HCQ are frequently employed also uncombined, still always during active inflammatory processes, i.e. acute bacterial infections or immune-inflammatory diseases, respectively. Based on the above mechanistic considerations, it is very likely that also in these pathologic conditions IL-6 could increase the arrhythmogenic potential of these drugs, even critically if, as commonly occurs, other pro-arrhythmic medications and/or non-pharmacologic risk factors are concomitantly present.
While guinea pigs exhibit the advantage of having an ECG and AP morphology similar to humans, the findings from this study may not be similar in human especially in the presence of a particular pathology like COVID-19 where often other comorbidities are concomitant. Consistent with our data is the recent work from Li et al. 60 , demonstrating that the clinically observed QT prolongation caused by treatment with HCQ could be recapitulated in human induced pluripotent stem cells derived cardiomyocytes (iPSC-CMs) by measuring field www.nature.com/scientificreports/ potential duration (FPDc) which is a surrogate for an electrocardiogram from cell monolayer. The authors showed that HCQ-induced FPDc prolongation was markedly enhanced by the combined treatment with AZM in iPSC-CMs. The authors have not examined the effects of IL-6 on HCQ + AZM. The fact that we could not induce any episodes of TdP in these healthy guinea pigs, does not mean that TdP cannot be triggered by the combination of an inflammatory state (IL-6) and the use of both AZM and HCQ in humans. Nevertheless, our animal data is consistent with several clinical reports demonstrating that TdP is not frequent in COVID-19 patients and that more in general in the setting of inflammation, IL-6 is a potential additional non-conventional risk factor for the development of cardiac arrhythmias alone or in combination with antimalarial or antibiotic drugs.

Conclusions
In the settings of severe infection, high levels of IL-6 combined with AZM and HCQ, can cause fatal arrhythmias and cardiac arrest via excessive conduction abnormalities, prolongation of the QTc and APD as a result of the multi-channels' inhibition of I Na , I CaL and I Kr . HCQ continues to be used in several countries for COVID-19 patients 14 , and for malaria and for autoimmune diseases such as lupus erythematosus and rheumatoid arthritis both conditions known to also experience high levels of IL-6 along with increased prevalence of QTc prolongation and sudden cardiac death [61][62][63] . In addition, there seems to be a renewed interest in the use of HCQ not only for COVID-19 patients but also as a prophylaxis for asymptomatic COVID-19 subjects as well as healthcare professionals 14,64 . Our findings are clinically relevant in assessing the arrhythmic risk when HCQ is combined with the commonly used AZM or other QTc prolonging factors during infection and/or inflammation. Intriguingly, the enhancer role demonstrated here for IL-6 on the arrhythmogenic potential associated with HCQ/ AZM treatment, could be extended to many other medications, particularly those listed at crediblemeds.org as QT-prolonging drugs. In fact, similarly to HCQ and AZM, it is well-recognized that most of these molecules can delay ventricular repolarization by blocking I Kr , in turn representing the key electrophysiological mechanism underlying IL-6-mediated QTc prolongation. In this way, COVID-19 could provide the opportunity to more generally decipher drug-induced arrhythmogenesis during inflammatory processes. Finally, as we move forward with COVID-19 pandemic, and in order to reduce mortality and for better preparedness, there continue to be an urgent need for the understanding of the electrophysiologic mechanism of drug combinations and the discovery of novel drug combination with a safe arrhythmic profile.