Triggered activity results from afterdepolarizations, which are secondary depolarization events arising prematurely before generation of the next AP (Cranefield, 1977; January et al., 1991). They can occur before or after full repolarization, termed early afterdepolarizations (EADs) or delayed afterdepolarizations (DADs), respectively. EADs typically develop when action potential durations (APDs) are prolonged, due to a greater magnitude of inward depolarizing currents (I_Na, I_Ca, I_NCX) compared to the outward repolarizing currents (I_Kr, I_Ks, I_K1) (Tse, 2015). This would enable reactivation of the L-type calcium channel, thereby increasing I_{Ca, L} (January and Riddle, 1989). By contrast, DADs occur when there is intracellular Ca²⁺ overload, which induces Ca²⁺-induced Ca²⁺ release from the ER. This activates the sodium-calcium exchange current, I_NCX, the nonselective cationic current, I_NS and the calcium-activated chloride current, I_{Cl, Ca} (Guinamard et al., 2004). The net result of EAD or DAD is membrane depolarization and initiation of another AP. If the afterdepolarizations are not sufficiently large to induce an AP, they can exacerbate the regional differences in repolarization, which may lead to alternans, unidirectional conduction block and reentry (Tse et al., in press).

Electrophysiological remodeling is observed in heart failure, with mostly downregulation of ion channels, leading to APD prolongation and/or altered calcium release, predisposing to EADs, DADs, and reentry (Marfella et al., 2013; Tse et al., 2016a). Moreover, the activities of ion channels may condition heart failure and arrhythmic state, regulating epigenetic mechanisms related to cardiac adaptive responses (Sardu et al., 2014c, 2015). The renin-angiotensin system (RAS) becomes activated in diabetes, hypertension and heart failure. This pathway is responsible for enzymatic cleavage of angiotensinogen, which is converted to angiotensin I and angiotensin II (Ang-II) by renin and angiotensin converting enzyme (ACE), respectively. Ang-II can activate the AT₁ receptor, which increases NADPH oxidase activity to promote ROS production, resulting in downregulation of the calcium-independent transient outward K⁺ channel (I_to) (Lebeche et al., 2006). ROS can increase the activity of MAPK-activated protein kinase-1 (p90^rsk), which phosphorylates and decreases the activity of I_{to, f}, I_{K, slow} and I_SS channels (Lu et al., 2008). ROS can prolongs the late sodium current, I_{Na, L} (Ma et al., 2005; Song Y. et al., 2006).

Moreover, prolonged activation of UPR during ER stress in heart failure leads to increased oxidative stress, downregulation of I_to and increased Ca²⁺ release from the SR. RyR2, the cardiac-specific isoform, is affected by post-translational modification from intracellular signaling cascades. Thus, they can be oxidized by ROS or reactive carbonyl species (RCS) (Eager et al., 1997; Xu et al., 1998; Bidasee et al., 2003), or phosphorylation by protein kinases such as protein kinase A or Ca²⁺/calmodulin-dependent protein kinase II (CAMKII), leading to abnormal gating (Witcher et al., 1991; Hain et al., 1995; Wehrens et al., 2004). This in turn leads to reduced amplitude but increased frequency of Ca²⁺ sparks and increased RyR2 sensitivity to Ca²⁺ activation (Shao et al., 2007, 2009). RyR2 decoupling from LTCCs can lead to desynchronized Ca²⁺ release (Song L.S. et al., 2006). ROS derived from the mitochondria could inhibit SERCA and stimulate RyR2, thereby activating NCX, Ca²⁺-sensitive cationic channels and Ca²⁺-induced Ca²⁺-release (Li et al., 2015).

Aside from regulation of energy production, mitochondria can also take up and release Ca²⁺ through several mechanisms, and may therefore serve as a buffer system. It may be that under physiological conditions, mitochondria do not have a major role in buffering Ca²⁺ (O'Rourke and Blatter, 2009; Williams et al., 2013). However, in heart failure, there is significant Ca²⁺ overload in both the cytoplasm and the mitochondria of cardiomyocytes (Santulli et al., 2015b). Experiments in a mouse model of heart failure generated by myocardial infarction, leaky RyR2 was shown to be responsible for mitochondrial Ca²⁺ overload (Santulli et al., 2015b). This would lead to mitochondrial dysfunction and increased ROS production. A recent study found that in mice with RyR2 mutations leading to diastolic Ca²⁺ leak, increased RyR2 oxidation, mitochondrial dysfunction and increased ROS production are observed, and these findings were associated with an increase in atrial fibrillation (AF) susceptibility (Xie et al., 2015a). Other studies also support the critical role of increased oxidative stress in the development and maintenance of AF. Thus, ROS from the atria correlated with the duration and substrate of AF in goats and humans (Reilly et al., 2011). Increased mitochondrial oxidative stress leading to RyR2 leak has also been observed in the ventricles of a peroxisome proliferator-activated receptor-γ (PPARg) cardiac overexpression mouse model (Joseph et al., 2016). The net electrophysiological effect was AP prolongation and development of ectopic activity.

Together, all of the above mechanisms can prolong APD and/or lead to abnormal Ca²⁺ release, inducing EADs and DADs and therefore triggered activity.

The conduction of cardiac APs through the working myocardium has traditionally been described by the core conductor model. This is based on the cable theory, whose equations can be derived by applying Kirchoff's circuit law. Conduction velocity (CV) is determined by both passive and active properties of the cell membrane. The passive components are axial resistance of the myoplasm (r_i), resistance of the extracellular space (r_o), and membrane capacitance (C_m). Diameter of the cardiomyocyte is important because it alters both r_i and C_m. Active membrane properties refer to the voltage-gated conductances, mainly the sodium current (I_Na) mediating the AP upstroke. This model was subsequently modified to account for the presence of gap junctions (Saffitz et al., 1994; Rohr, 2004). Each gap junction consists of two connexons, which are hexameric proteins of the connexin (Cx) subunit, mediating electrical communication by electrotonic spread. However, it forms a high resistance pathway that can decrease CV and produce discontinuous propagation (Spach et al., 1981; Spach, 2003). Its expression is not equally distributed across the myocardium, higher numbers are found at the ends of myocytes compared to the lateral aspects of the cell membrane (Kumar and Gilula, 1996). Consequently, CV is higher in the longitudinal direction than in the transverse direction, which is termed anisotropic conduction (Sano et al., 1959; Clerc, 1976). The role of ephaptic coupling has been neglected until recently, involving a direct electrical field mechanism for mediating conduction (Rhett and Gourdie, 2012; Lin and Keener, 2013; Rhett et al., 2013; Veeraraghavan et al., 2014a,b,c; George et al., 2015; Veeraraghavan et al., 2015). Finally, myocardial fibrosis cam reduce the coupling between cardiomyocytes, increasing r_i, and enhance the coupling between fibroblasts and cardiomyocytes (Camelliti et al., 2004; Chilton et al., 2007), increasing c_m, both of which would reduce CV and increase the dispersion of CV (Tse and Yeo, 2015).

Circus-type reentry occurs when an AP travels around an obstacle and re-excites its site of origin, and requires both conduction and repolarization abnormalities (Tse, 2015; Tse et al., 2016d,h). Firstly, conduction velocity (CV) must be sufficiently slowed to prevent the AP depolarization wavefront from colliding with its repolarization wave back, where the local myocardium is still in its effective refractory period (ERP) and cannot be excited. If this occurs, conduction will be blocked and the AP will extinguish, resulting in a break of the reentrant circuit. Secondly, unidirectional conduction block must be present. Otherwise, wavefront APs traveling in opposite directions can collide and extinguish, thereby terminating the arrhythmia. Thirdly, an obstacle can be a non-conducting or refractory region, arising from structural defects such as interstitial fibrosis, or dynamically from ectopic activity. When the excitation wavelength of the AP wave (λ, CV × ERP) is smaller than the path length, reentry can occur. The ERP usually coincides with the APD. Thus, reduction in CV, APD, or ERP can promote reentry by reducing λ, whereas increased λ reduces the likelihood of reentry (Tse et al., 2012; Choy et al., 2016; Tse, 2016b; Tse et al., 2016c,e,f,g,h,i,k,l).

All of the above factors are influenced by ROS, altering ion channel function and promoting extracellular matrix (ECM) remodeling, which together would produce conduction and repolarization abnormalities, thereby leading to reentry (Tse and Yeo, 2015). For example, the SCN5A gene encoding for the cardiac Na⁺ channels is regulated by the transcription factor NF-κB (Shang and Dudley, 2005). Elevated angiotensin II levels and the increased oxidative stress increase NF-κB binding to the SCN5A promoter region, thereby decreasing its transcriptional activity (Shang et al., 2008). Ang-II can also downregulate Cx43 and Cx40, reducing intercellular coupling (Kasi et al., 2007). During ER stress, PERK activation downregulates the Na⁺ channels, although this is a non-ion channel specific effect since I_to expression is also reduced (Gao et al., 2013). Furthermore, abnormal Ca²⁺ release can have direct and indirect effects. Ca²⁺ can bind directly to the evolutionarily conserved EF hand motif at the carboxyl end of the Na⁺ channel, thereby producing a depolarizing shift in its voltage-dependent inactivation, which prolongs the late sodium current, I_{Na, L} (Wingo et al., 2004; Song Y. et al., 2006). Moreover, the Na⁺ channel also possesses an IQ domain for Ca²⁺/ CaM binding and additionally several serine/threonine residues amenable to phosphorylation, for example by CaMKII (Ashpole et al., 2012), leading to channel inactivation (Tan et al., 2002).

Ca²⁺ can activate protein kinase C (PKC) (Luo and Weinstein, 1993), which can phosphorylate both Na⁺ channels (Qu et al., 1996) and Cx43 (Moreno et al., 1994; Kwak et al., 1995) to reduce I_Na and I_gap, respectively. Ca²⁺ overload is also associated with dephosphorylation of gap junctions (Huang et al., 1999), leading to their uncoupling (Beardslee et al., 2000) and lateralization (Smith et al., 1991; Lampe et al., 2000). Interestingly, a recent report demonstrated that streptozotocin-induced diabetic mice have increased phosphorylation levels of Cx43 at the serine 262 residue (Palatinus and Gourdie, 2016). Previous experiments showed that Ser 262 phosphorylation of Cx43 is associated with a cardiac injury-resistant state (Srisakuldee et al., 2009). Poor myocardial healing could increase the likelihood of conduction abnormalities, which would increase arrhythmogenicity. Finally, homocysteine, which is raised in hypertension and diabetes, causes EEC dysfunction and apoptosis (Miller et al., 2000; Wei et al., 2010), suppression of superoxide dismutase and activation of matrix metalloproteinases (MMPs) in the plasma membrane, causing Cx43 disruption (Rosenberger et al., 2006). Increased Ca²⁺ leak through the RyR2 channels results in Ca²⁺ alternans, which in turn can drive APD alternans and reentry (Xie et al., 2013).

Mitochondrial ROS production is elevated in diabetes and hypertension (Kakkar et al., 1995; Kanazawa et al., 2002; Dikalov and Ungvari, 2013). Mitochondrial ROS can itself increase its own production in a positive feedback loop, termed ROS-induced ROS release (RIRR), a mechanism recently recognized as a key initiator of mitochondrial depolarization (Zorov et al., 2000). Thus, Ang-II can activate the AT₁ receptor, which stimulates Nox2, an enzyme that catalyzes electron transfer from NADPH to oxygen, generating oxygen free radicals and hydrogen peroxide. The oxygen free radical can enter the mitochondria and activate PKCε at the mitochondrial inner membrane (Jaburek et al., 2006). The downstream target is mitochondrial K_ATP channels (mitoK_ATP), resulting in reverse electron transfer from complex II to complex I, which generates more oxygen free radicals (Dikalov et al., 2014). The latter can be converted to hydrogen peroxide, activating c-src (Aikawa et al., 1997) and in turn further activating NADPH oxidases, production of free radicals and NOS uncoupling. Mitochondrial K_ATP channel activation in cardiomyocytes in diabetes led to impaired APD adaptation, which promoted the occurrence of VT (Xie et al., 2015b). Ang-II mediated mitochondrial ROS production can lead to c-src activation and reduced Cx43 expression (Sovari et al., 2013). Homocysteine can activate mitochondrial MMPs, which leads to opening the mitochondrial permeability transition pore (MPTP) (Moshal et al., 2008; Montaigne et al., 2013) and collapse of the mitochondrial inner membrane potential (ΔΨ_m) (Brown et al., 2010).

ROS can also activate the mitochondrial inner membrane anion channels (IMAC) to result in mitochondrial depolarization (Yang et al., 2010). RIRR-induced regional mitochondrial depolarization results in the formation of a metabolic sink (Zhou et al., 2014). Uncoupling of oxidative phosphorylation can reverse mitochondrial ATP synthase, thereby depleting intracellular ATP. This causes sarcolemmal K_ATP activation, increased K⁺ efflux and APD shortening (Garlid et al., 1997; Zhou et al., 2014). Moreover, ROS can promote myocardial fibrosis via a NOX4-mediated, ROS-ERK1/2-MAP Kinase-dependent mechanism (Aragno et al., 2006; Kuroda et al., 2010).

Together, the above factors can lead to reduction in CV, together with increased repolarization gradients, can promote unidirectional conduction block and reentry (Tse et al., in press).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer QY and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.