Innate immune activation and mitochondrial ROS induce acute and persistent cardiac conduction system dysfunction after COVID-19

Figures

• 

  Figure 1. SARS-CoV-2 infection causes weight loss, hypothermia, and tachypnea in the hamster model. (A) Body weight decreased after infection, reached a minimum at 6–7 dpi, and then recovered back to normal. (B) SARS-CoV-2 infection induced a transient decrease in body temperature peaking at 3 dpi and recovering by 7 dpi. (C) Representative traces of subpleural pressure recorded in mock-infected (left) and SARS-CoV-2 (right) –infected hamsters at 0 (upper), 3 (middle), and 7 (lower) dpi. Vertical bar: 10 mmHg; horizontal bar: 1 second. (D) Breathing rate increased following SARS-CoV-2 infection, peaked at 7 dpi, and recovered by 14 dpi. †P < 0.01; ‡P < 0.0001.

• 

  Figure 2. Effects of SARS-CoV-2 infection on cardiac electrophysiology in the hamster model. (A–E) Representative ECG recording showing normal baseline rhythm (A) and arrhythmias induced by SARS-CoV-2 infection, including bradycardia (B), sinus pauses (C; arrows), second-degree AV block (D), and third-degree AV block (E). Arrowheads indicate P waves. (F–I) Analysis of ECGs from 0 to 28 dpi typically showed a triphasic pattern of SARS-CoV-2 effects on cardiac rhythm: an acute peak within 7 dpi; recovery to, or below, baseline; and a long-term effect developing between 7 dpi and 28 dpi. Mean RR interval (normalized to 0 dpi) (F), incidence of long sinus pauses (RR > meanRR+100 ms; normalized to 0 dpi) (G), and RMSSD (H) peaked at 1–3 dpi; returned to levels close to or lower than baseline; and then gradually increased to levels significantly higher than baseline at 21 and 28 dpi. The rate of AV block events (I) peaked early and did not return to baseline level, remaining significantly higher than baseline between 7 and 28 dpi. Red symbols denoting P values *P < 0.05; †P < 0.01; ‡P < 0.0001 on the figures compare mock- vs. SARS-CoV-2–infected groups. P values comparing SARS-CoV-2 data from…

• 

  Figure 3. Macrophage remodeling in the CCS. (A) Representative images showing the conduction tissue marker contactin 2 (red) and the distribution of Iba1+ macrophages (green) in the region. (B) Representative images showing Iba1+ and CD163+ macrophages in the conduction region in mock- or SARS-CoV-2–infected hearts at 4 and 28 dpi. (C and D) Density of Iba1+ cells and CD163+ cells in CCS region of the mock- or SARS-CoV-2–infected hearts at 4 and 28 dpi. (E and F) Gene expression of cytokines and IFN-stimulated genes in lung (E) and heart (F) at 4 dpi, evaluated with qPCR. Unpaired 2-tailed t test was used to test for significance in normally distributed lung data in E. Mann-Whitney U test was used as the nonparametric significance test in heart data in F. (G and H) SARS-CoV-2 nucleocapsid protein was detected in lung, but not in the heart, by Western blot (G) or by immunofluorescence (H) at 4 dpi. (I) Viral spike protein mRNA was detected in lung, but not in the heart, by qPCR at 4 dpi. Unpaired 2-tailed t test.

• 

  Figure 4. The dsRNA mimetic PIC induces cardiac arrhythmias and innate immune responses in the absence of viral infection in guinea pigs. (A–D) Representative ECG traces showing regular cardiac rhythm in a vehicle-injected control (A) and bradycardia (B), sinus pauses (C; marked by arrows), and AV block (D) after cardiac injection of polyinosinic:polycytidylic acid (PIC) in guinea pigs (arrowheads in D indicate abnormal P waves; horizontal bar equals 2 seconds). (E–I) Summary data showing increase in RR interval (E), sinus pauses longer than mean RR+2SD (F), corrected QT (H), and QT interval (I) but no increase in PR interval (G) 4 days after myocardial PIC injection in guinea pigs. Welch’s 2-tailed t test was done for B–F. (J) Increased expression of innate immune response genes induced by PIC, expressed as fold-change from vehicle-injected controls, was observed. Mann-Whitney test was done for IL-1β, CCL2, and TNF-α. Welch’s 2-tailed t test was performed for TGF-β, Casp3, and OAS1. (K) Ca2+ transient analysis of adult cardiomyocytes isolated from guinea pig hearts injected with either PIC+AdV-GFP or AdV-GFP alone (AdV-GFP was used to identify myocytes near the injection site; unpaired…

• 

  Figure 5. PIC increases expression of antiviral interferon signaling proteins and alters excitation-contraction coupling in human ventricular and sinoatrial nodal myocytes. (A) Activation of interferon signaling pathway proteins in hiPSC-CM with PIC treatment (200 µg/mL for 72 hours). Left panel: Western blot; right panels: quantification of proteins after normalizing to total protein. (n = 3; unpaired 2-tailed t test.) (B) Field potential durations in hiPSC ventricular myocytes were significantly increased by 24 hours after PIC treatment (n = 11 for control and n = 12 for PIC treated; 2-way repeated measures ANOVA with Holm-Šídák’s multiple-comparison test for panels B and D). Multielectrode array measurements were taken at the indicated times before and after treatment. (C) Left: Ca2+ transient peak amplitudes were significantly reduced 5 days after PIC treatment in monolayers of hiPSC-CM (data points represent 20–30 replicates from 3 independent experiments; a nested analysis with Mann-Whitney 2-tailed test was used). Right: Ca2+ transient recordings in PIC-treated (red trace) and control (black trace) monolayers. (D) In hiPSC-CM atrial/sinoatrial nodal cell monolayers, RR intervals…

• 

  Figure 6. JAK/STAT inhibition, but not mitochondrial antioxidant treatment, suppresses interferon responses in hiPSC-CM. (A) Schematic of molecular mechanisms of COVID-19 and dsRNA induced IFN response. Ruxolitinib suppresses IFN-stimulated response and mitoTEMPO suppresses mitochondrial damage and oxidative stress. (Created with BioRender.com.) (B and C) Inhibition of JAK/STAT signaling with ruxolitinib (Rux) (1 μM) significantly suppressed PIC-mediated induction of the interferon response signaling proteins STAT1, p-STAT1, IRF9, and MX1 in hiPSC-CM monolayers. MitoTEMPO (MT) (1 μM) does not significantly suppress the IFN response. Bar graphs show relative band intensity normalized to total protein loading (1-way ANOVA with Tukey’s multiple-comparison test). Lysates were from 3–5 different hiPSC-CM samples. (C) Inhibition of JAK/STAT signaling in hiPSC-derived engineered heart tissues (EHTs) with ruxolitinib, but not mitoTEMPO (MT), decreased the interferon response of signaling proteins STAT1, pSTAT1, IRF9, and MX1. Lysates loaded for the Western blot in C were from 3 different EHTs. (D) Cytokine secretion into the media induced by PIC, measured using a human Proteome Profiler array…

• 

  Figure 7. Connexin dysregulation in vitro and in vivo after innate immune activation. (A) In hiPSC-CM, Cx43 and Cx45 gene expression was suppressed by PIC (72 hours) and prevented by concomitant ruxolitinib treatment, but not by mitoTEMPO (n = 3, 1-way ANOVA with Holm-Šídák’s multiple comparisons test). (B) Optical mapping of hiPSC-CM monolayers showed that 4–5 days of PIC treatment significantly decreased conduction velocity (n = 7; unpaired nested t test). (C) Images: Aberrant Cx43 localization in the ventricular myocardium 4 days after SARS-CoV-2 infection in the hamster COVID-19 model. Cx43 was mislocalized away from the intercalated discs (ICD) and toward the nuclear/perinuclear domains. Right-hand panel: Percentage of Cx43 at the ICD. DAPI, nuclear stain; WGA, wheat germ agglutinin membrane marker; Cx43, connexin 43. (D) Images: In PIC-injected guinea pig myocardium, Cx43 decreased at the ICD and increased at lateral membranes. Right-hand panel: Percentage of lateralized Cx43 in PIC-injected guinea pig hearts.

• 

  Figure 8. Effects of mitochondrial ROS scavenging or JAK/STAT inhibition on pulmonary and cardiac sequelae of COVID-19. (A and B) COVID-19–induced changes in body weight (A) or temperature (B) were not significantly altered by mitoTEMPO (mTEMPO) or ruxolitinib (Ruxo) treatments. (C) SARS-CoV-2–induced tachypnea was significantly attenuated by both mTEMPO and Ruxo treatments. †P < 0.005, ‡P < 0.0001. (D) mTEMPO, but not Ruxo, attenuated the increase in RR interval at 1 and 3 dpi. After returning to baseline level at 7 dpi, both treatments showed no impact on the redeveloped bradycardia. *P < 0.05, ‡P < 0.0001. (E and F) The early spike in sinus pauses (E) and RMSSD (F) at 1 dpi were not significantly decreased by either treatment. Both Ruxo and mTEMPO attenuated the late effects of SARS-CoV-2 infection on sinus pauses at 21 or 28 dpi, respectively (*P < 0.05). Only mTEMPO significantly suppressed RMSSD at 28 dpi (*P < 0.05). (G) Both treatments significantly attenuated the incidence of AV block 1–7 dpi and abrogated the sustained increase in AV block at 14–28 dpi. (*P < 0.05, †P < 0.01; compared with SARS-CoV-2 alone, blue symbols are Ruxo treatment, green symbols are mTEMPO treatment; see…