Increased Persistent Sodium Current Due to Decreased PI3K Signaling Contributes to QT Prolongation in the Diabetic Heart

Insulin-deficient C57BL/6-Ins2^Akita/J (Ins2^Akita) and insulin-resistant B6.BKS(D)-Lepr^db/J (db/db) mice in the C57BL/6J background were purchased from Jackson Laboratory (Bar Harbor, ME). For both groups, control animals were wild-type C57BL/6J mice. Animals were analyzed between 2 and 3 months of age. Hyperglycemia (>500 mg/dL) was confirmed in Ins2^Akita and db/db mice by tail vein blood glucose measurements using a OneTouch UltraMini glucometer (LifeScan, Milpitas, CA). All animal-related experimental protocols were approved by the Stony Brook University Institutional Animal Care and Use Committee and conform to National Institutes of Health (NIH) standards.

Mouse ventricular myocytes were isolated from mice fed ad libitum as previously described (8). To make myocyte cultures, hearts harvested from adult mice were first perfused with Ca²⁺-free myocyte buffer (137 mmol/L NaCl, 5.4 mmol/L KCl, 2 mmol/L MgSO₄, 0.33 mmol/L NaH₂PO₄, 10 mmol/L HEPES, 10 mmol/L taurine, 10 mmol/L glucose, and penicillin-streptomycin, pH 7.4) followed by digestion with Liberase TH (Roche) and 0.1 mmol/L Ca²⁺ in myocyte buffer. Digestion was stopped by perfusion with myocyte buffer containing 0.2 mmol/L Ca²⁺ and 1% BSA. The tissue was dissociated by mechanical force, and isolated cells were filtered through 210-µm nylon mesh to remove debris. Settled myocytes were then washed with myocyte buffer containing 1% BSA with increasing concentrations of Ca²⁺ up to 1.0 mmol/L. Viable myocytes were resuspended in culture medium (medium 199 supplemented with 2 mmol/L carnitine, 5 mmol/L creatine, 5 mmol/L taurine, 0.2% BSA, 100 units/mL penicillin, 100 µg/mL streptomycin, 1.75 μmol/L bovine insulin, 5.5 μg/mL transferrin, 6.7 ng/mL selenium, and 25 μmol/L blebbistatin with physiological levels of Ca²⁺) with 5% FBS and cultured on laminin-coated dishes at 37°C with a 5% CO₂ atmosphere for 2 h and then changed to serum-free culture medium for another 4 h prior to adding adenoviruses. The next day, cell culture medium was changed to contain either 1 nmol/L bovine insulin (wild-type and db/db) or no insulin (Ins2^Akita). Electrophysiological or biochemical analyses were performed 24 h later.

Site-directed mutagenesis was used to generate hemagglutinin (HA) epitope–tagged p110α H1047R from wild-type mouse p110α cDNA. Recombinant adenoviruses carrying p110α H1047R were constructed using a method similar to the one described by He et al. (13). In brief, HA-tagged p110α H1047R was cloned into the pAdTrack-CMV shuttle vector using the XbaI and KpnI restriction sites. The resulting plasmid was linearized and then electroporated into BJ-5183-Ad1 cells (Stratagene). Bacterial colonies were screened for appropriate recombination. The resulting plasmid was linearized with PacI and transfected into HEK293 cells for viral production. Viruses were collected and purified by centrifugation in cesium chloride. Green fluorescent protein (GFP)-expressing control adenoviruses were produced using a similar method (14).

Recordings were made at room temperature. Only relaxed quiescent cells displaying clear cross striations were used. Standard whole-cell patch-clamp techniques were performed with an Axopatch-1D amplifier with a CV-4 1/100 headstage (Axon Instruments). A personal computer equipped with 12-bit AD/DA converters (model 1360; Cambridge Electronic Design) was used for data acquisition, generation of pulse protocols, and data analysis. Currents were sampled at 10 kHz and filtered with a four-pole Bessel filter at 2 kHz. Current amplitude was normalized to cell capacitance to obtain current density in picoamp/picofarad (pA/pF). In some experiments, 1 μmol/L phospholipids (all di-C8; Echelon Biosciences) were added to the pipette solution. Mexiletine hydrochloride (Sigma-Aldrich) was added to cells for 2 h at room temperature prior to or acutely perfused during patch clamping as indicated. Wild-type myocytes were incubated with Akt inhibitor (Akti) VIII (EMD Millipore) for 2 h prior to patch clamping.

Recording of action potentials in mouse myocytes was initiated in current clamp mode by applying pulses 120 pA in amplitude and 10 ms in duration with a cycle length of 1 s as previously described (8). 4-Aminopyridine (4-AP, 4 mmol/L; Sigma-Aldrich) was added to the external solution to block most of the transient outward current (I_to).

The tetrodotoxin (TTX)-sensitive Na⁺ current was elicited by 750-ms depolarizing voltage steps ranging from −80 to 0 mV at 10-mV increments from a holding potential of −80 mV. The pipette solution contained 111 mmol/L CsCl, 20 mmol/L tetraethylammonium chloride, 14 mmol/L EGTA, 5 mmol/L MgATP, 10 mmol/L HEPES, and 10 mmol/L glucose. The pH of the pipette solution was adjusted to 7.4 with 39 mmol/L CsOH. The external solution contained 90 mmol/L CsCl, 1.2 mmol/L MgCl₂, 1 mmol/L CaCl₂, 10 mmol/L tetraethylammonium chloride, 5 mmol/L HEPES, 11 mmol/L glucose, and 50 mmol/L NaCl. The pH of the external solution was adjusted to 7.4 with CsOH. I_NaP was measured as the main inward current between 700 and 750 ms at the end of depolarization. TTX-sensitive currents were measured by subtracting a trace obtained in the presence of 10 μmol/L TTX from a trace obtained in its absence. I_NaP records were filtered at 20 Hz. For the single trace of TTX-sensitive current shown in the figures, the current was activated at a test voltage of −20 mV from a holding potential of −80 mV.

Isolated mouse hearts were mounted on the IH-SR isolated heart perfusion system (Harvard Apparatus) and perfused with Krebs-Henseleit solution (118 mmol/L NaCl, 4.7 mmol/L KCl, 2.52 mmol/L CaCl₂, 1.64 mmol/L MgSO₄, 24.88 mmol/L NaHCO₃, 1.18 mmol/L KH₂PO₄, 5.55 mmol/L glucose, and 2 mmol/L sodium pyruvate aerated with 5% CO₂ and 95% O₂) at 37°C for 30 min to reach a stable baseline prior to data collection. For electrocardiographic recording, one electrode was placed at the base of the heart next to the left atrium, and a second electrode was placed at the heart apex held by pressure. Recordings were collected under control conditions, and then human insulin (Novolin R; Novo Nordisk), PI-103 (Cayman Chemical), or mexiletine was added to the perfusate reservoir and circulated through the system for 30 min prior to collecting another set of recordings. QT intervals were measured automatically by the LabChart 7.1.2 (ADInstruments) software system from >30 consecutive heart beats, and QTc was calculated using the correction described by Mitchell et al. (15) or by Bazett’s formula.

Mice were fasted overnight. Harvested hearts were perfused with Krebs-Henseleit solution as described above, and then either normal saline or human insulin (1 unit/L) was added to the perfusate reservoir and circulated through the hearts for 10 min. The left ventricles were collected and stored in liquid nitrogen. For PI3K assays, pieces of tissue were homogenized in RIPA buffer containing 50 mmol/L HEPES, pH 7.5, 10 mmol/L sodium pyrophosphate, 50 mmol/L NaCl, 50 mmol/L NaF, 5 mmol/L EDTA, 1 mmol/L sodium orthovanadate, 0.25% sodium deoxycholate, 1% NP-40, 0.2 mmol/L phenylmethylsulfonyl fluoride, and 2 µL/mL protease inhibitor cocktail (Sigma-Aldrich). After centrifugation, aliquots of supernatant containing equal amounts of protein (Bradford assay; Bio-Rad) were mixed with antiphosphotyrosine antibody 4G10 coupled to agarose (Millipore) for 3 h at 4°C. After washing the beads several times with RIPA buffer and then with PI3K assay buffer (20 mmol/L HEPES, pH 7.4, 100 mmol/L NaCl, and 0.5 mmol/L EGTA), they were resuspended in 40 µL PI3K assay buffer. Assays were started by adding l-α-phosphatidylinositol (Sigma-Aldrich) and a reaction mix containing [γ-³²P]ATP (PerkinElmer Life Sciences), and the reactions were carried out and processed as previously described (16). After thin-layer chromatography, radioactive spots containing phosphatidylinositol phosphate were visualized by autoradiography, cut out of the plate, and quantified by scintillation counting.

For Akt phosphorylation, Western blotting was performed on the heart lysates as previously described (17) using antibodies to phospho-T473 Akt (Cell Signaling) and total Akt (Santa Cruz Biotechnology). PDK1 and PTEN antibodies were purchased from Cell Signaling; Nav1.5 antibodies were from Alomone Laboratories or EMD Millipore. The bands visualized on film were quantified by NIH ImageJ software. Bands visualized by ProteinSimple FluorChem E (Santa Clara, CA) were quantified by software provided by the company.

Freshly isolated ventricular myocytes from hyperglycemic (>500 mg/dL) insulin-deficient Ins2^Akita and insulin-resistant db/db mice were studied in patch-clamp experiments to determine the APD at 90% repolarization (APD₉₀). As in our previous study (8), we performed our action potential recording in the presence of 4-AP to eliminate most of the large transient outward current I_to to produce a longer action potential that allowed us to more easily determine the effects on APD, which might be more relevant to larger mammals. Representative action potentials recorded from Ins2^Akita, db/db, and wild-type myocytes are shown in Fig. 1A–C. APD₉₀ in both types of diabetic myocytes was markedly longer than in wild-type cells (Fig. 1D). When phosphatidylinositol 3,4,5-trisphosphate (PIP3), the second messenger produced by PI3K, was added to the patch pipette to dialyze the cell interior, APD₉₀ of Ins2^Akita and db/db myocytes decreased to near wild-type levels (Fig. 1A and D). In contrast, addition of the control phospholipids phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2] or phosphatidylinositol 3,5-bisphosphate [PI(3,5)P2] had no effect on APD₉₀ (Fig. 1D). To test if the long APD of diabetic myocytes is associated with increased I_NaP, we incubated the cells for 2 h with the I_NaP blocker mexiletine. This drug treatment shortened APD₉₀ in both Ins2^Akita and db/db myocytes to wild-type levels (Fig. 1B and D). Acute perfusion of Ins2^Akita myocytes with mexiletine had a similar shortening effect on the APD (Supplementary Fig. 1). When action potentials were recorded in the absence of 4-AP, PIP3 and mexiletine still reversed APD prolongation in the diabetic myocytes, but the effect was partial (Supplementary Fig. 2).

The results with mexiletine suggested that APD lengthening in diabetic cardiac myocytes might be due to an increase in I_NaP. This was validated in patch-clamp experiments showing an approximately twofold increase in I_NaP current density in both Ins2^Akita and db/db myocytes as compared with wild-type cells (Fig. 2A). This enhancement in current was present across the entire range of voltages tested (Fig. 2B), as was seen in our previous study (8). Infusion of both diabetic cell types with PIP3 caused a reduction in I_NaP to wild-type levels (Fig. 2). We saw no consistent change in expression of the Nav1.5 sodium channel that conducts I_NaP in the diabetic mouse hearts (Supplementary Fig. 3A). Therefore, the increase in I_NaP associated with diabetes might be due to alterations in trafficking, gating properties, and/or single channel conductance.

To determine if the increases in APD and I_NaP in the diabetic hearts translate into prolongation of the QT interval, we analyzed recordings of electrical activity of Langendorff-perfused hearts. QTc in Ins2^Akita hearts was increased more than twofold as compared with nondiabetic wild-type hearts (Fig. 3A and Supplementary Fig. 4). Furthermore, mexiletine treatment reversed the QTc prolongation, emphasizing the importance of increased I_NaP in lengthening the QT interval in this mouse model (Fig. 3A and Supplementary Fig. 4). Insulin treatment corrected the abnormal QTc of Ins2^Akita hearts, and treatment with PI-103, a PI3K inhibitor, blocked the insulin effect (Fig. 3A and Supplementary Fig. 4).

Like the Ins2^Akita hearts, QTc in db/db hearts was lengthened more than twofold in comparison with control hearts (Fig. 3B and Supplementary Fig. 4). Treatment with mexiletine also shortened the QT interval in this diabetic mouse model, but the effect was only partial (Fig. 3B and Supplementary Fig. 4).

We showed in a previous study that PI3K signaling to the protein kinase Akt is downregulated in the heart of Ins2^Akita mice as compared with nondiabetic wild-type mice, but activation of PI3K/Akt by exogenous insulin is intact (18). Because db/db mice develop hyperglycemia despite hyperinsulinemia at a young age, we next investigated if basal and insulin-activated PI3K/Akt signaling are altered in the db/db hearts. Spontaneously beating hearts from wild-type and db/db mice were mounted on the Langendorff apparatus and perfused with insulin or vehicle for 10 min. Heart lysates were then assayed for PI3K activity or analyzed for Akt phosphorylation by Western blotting. Basal PI3K activity tended to be lower in the db/db samples than in wild-type samples (Fig. 4A), whereas basal Akt phosphorylation was clearly reduced in db/db versus wild-type hearts (Fig. 4B). Insulin stimulated a robust increase in PI3K activity (Fig. 4A) and Akt phosphorylation (Fig. 4B) in the wild-type hearts. In contrast, insulin activation of PI3K and Akt was reduced by 35–40% in the db/db hearts as compared with wild-type (Fig. 4). Decreased Akt activation in the two diabetic mouse models does not appear to be due to changes in cardiac expression levels of PTEN or PDK1 (Supplementary Fig. 3B). Thus, db/db mice exhibit cardiac insulin resistance with respect to PI3K signaling, which we believe leads to QT interval prolongation.

We previously showed that genetic ablation of the PI3K p110α catalytic subunit in cardiac myocytes caused APD prolongation due at least in part to an increase in I_NaP, demonstrating the importance of this PI3K isoform in regulating the mouse cardiac action potential (8). Therefore, we assessed whether expression of the constitutively active p110α H1047R mutant (CAp110α) in db/db and Ins2^Akita myocytes shortens the APD. Primary cultures of ventricular myocytes isolated from db/db and Ins2^Akita mice were infected with adenoviruses carrying either CAp110α or GFP as a control. As expected, expression of CAp110α increased Akt phosphorylation (Supplementary Fig. 5). Patch-clamp measurements performed in the presence (Fig. 5A and B) or absence (Supplementary Fig. 6) of 4-AP showed that APD₉₀ in db/db and Ins2^Akita myocytes expressing CAp110α was dramatically shortened as compared with the same batch of myocytes infected with adenoviral GFP. Expression of adenoviral CAp110α in cultured wild-type myocytes did not cause a change in APD₉₀ as compared with the GFP control (Fig. 5A and B). It was interesting that APD₉₀ in GFP-infected wild-type myocytes was still shorter than in GFP-infected diabetic cells after 2 days in culture (Fig. 5B).

Expression of adenoviral CAp110α in cultured db/db and Ins2^Akita myocytes also caused a marked decrease in I_NaP (Fig. 5C–F). As with the effect of PIP3 on freshly isolated myocytes shown in Fig. 2, the effect of CAp110α on I_NaP was evident at all voltages studied. Furthermore, the current density of I_NaP in cultured db/db and Ins2^Akita myocytes infected with GFP was similar to the level observed in freshly isolated diabetic myocytes (compare Fig. 2B and Fig. 5E and F), whereas the current density in diabetic myocytes expressing CAp110α was similar to the values for wild-type and diabetic myocytes infused with PIP3 (compare Fig. 2B and Fig. 5E and F). These results indicate that the cultured myocytes retain some of the properties of freshly isolated myocytes and suggest that PI3K is responsible for the regulation of I_NaP.

In a converse experiment, we tested whether inhibition of Akt activity in nondiabetic myocytes increases I_NaP. Myocytes freshly isolated from wild-type mice were incubated with an Akti prior to patch clamping. We found that this treatment indeed caused an increase in I_NaP, but not to the levels seen in myocytes from diabetic mice (Supplementary Fig. 7).

The major finding in this study is that an increase in the inward sodium current I_NaP plays an important role in causing long QT syndrome in murine models of diabetes. Previous studies suggested that the cardiac repolarization defect in diabetes is due mainly to a decrease in outward potassium currents. Using streptozotocin-induced diabetic rats as a model of type 1 diabetes, it was concluded that APD prolongation is due to reductions in the 4-AP–sensitive transient outward current I_to and a 4-AP–insensitive steady-state current referred to as I_K (19–21). A decrease in the inward L-type calcium current I_CaL was also seen in streptozotocin-induced diabetic rats (21) and Ins2^Akita mice (18), but this would tend to shorten the APD. APD prolongation and suppression of outward potassium currents and I_CaL were also seen in cardiac myocytes from type 2 diabetic db/db mice (22–25). The affected potassium currents in db/db mice were not extensively characterized, but were probably equivalent to I_to and I_K in the rat studies described above. In comparison with rodents, the magnitude of I_to is small in humans and other large mammals. Instead of I_to, the outward delayed rectifier potassium currents I_Ks and I_Kr play a prominent role in regulating cardiac repolarization in large mammals, and some reports demonstrated that these currents can also be altered in diabetes. For example, alloxan-induced diabetic rabbits (a model of type 1 diabetes) exhibited QTc prolongation and reductions in current density of I_Kr, I_Ks, I_to, and I_CaL, with no change in peak sodium current (26,27). Computer modeling of the rabbit ventricular action potential suggested that the decrease in I_Kr was the major ionic mechanism for diabetic QT prolongation in this model (26). By contrast, no differences I_Kr, I_CaL, or the inward rectifier potassium current I_K1 were seen in cardiac tissue from alloxan-induced diabetic dogs as compared with controls, whereas the current densities of I_Ks and I_to were significantly lower (28).

To our knowledge, this is the first report in which I_NaP was examined in diabetic hearts. We found that I_NaP current density was higher in myocytes from Ins2^Akita and db/db mice as compared with wild-type myocytes. Furthermore, treatment of diabetic myocytes with a sodium channel blocker at concentrations somewhat selective for I_NaP reduced APD₉₀. In the presence of the I_to blocker 4-AP, alterations in I_to or other 4-AP–sensitive currents did not contribute to the differences in APD that were observed in control versus diabetic cells. Therefore, it is perhaps not surprising that inhibition of I_NaP with mexiletine was sufficient to normalize APD in the diabetic cells under these conditions. The effect of mexiletine was reduced in the absence of 4-AP, indicating that a decrease in I_to and other 4-AP–sensitive currents plays a role in determining APD₉₀ in the diabetic myocytes. On the other hand, perfusion of Ins2^Akita hearts with mexiletine resulted in complete normalization of QTc even in the absence of 4-AP. The action potential studies and the ECG studies occur at different stimulation/heart rates and so the effects on APD and ECG are not simply comparable. However, taking the APD and ECG results together suggests that the increase in I_NaP is, at a minimum, a significant contributor to QT interval lengthening in the Ins2^Akita mouse. Although mexiletine also significantly decreased QTc in db/db hearts, the effect was partial. It is possible that QT prolongation in db/db hearts is due to an increase in I_NaP as well as a decrease in outward potassium currents such as I_to and I_K (22).

Patients with type 3 congenital long QT syndrome (LQT3) have gain-of-function mutations in the Nav1.5 sodium channel protein (encoded by SCN5A) that cause an increase in I_NaP (2). Transgenic mice that express SCN5A mutants exhibit many of the phenotypes of LQT3 patients, including elevated I_NaP, QT interval prolongation, and cardiac arrhythmias (11,29,30). Mexiletine treatment shortened APD in mouse myocytes expressing an SCN5A mutant (29), and it also shortened QTc in LQT3 patients (31). Although mouse models of altered I_NaP function have been informative with regard to human LQT3, it is not yet known whether I_NaP is increased in diabetic patients. Further studies using cardiac myocytes from diabetic humans or large animal models need to be performed to answer this question and to determine the feasibility of using a sodium channel blocker as a therapy to reduce QT prolongation in diabetes.

The central role of insulin signaling in maintaining normal cardiac electrophysiology was demonstrated in cardiac myocyte–specific insulin receptor knockout mice. Cardiac myocytes from these nondiabetic animals exhibited significant APD prolongation and reduced outward potassium currents (24). Our results provide the following evidence suggesting that low insulin/PI3K signaling is the cause of the cardiac repolarization defect in the diabetic mice studied here. First, intracellular delivery of PIP3, the second messenger produced by PI3K, shortened APD and reduced I_NaP in freshly isolated diabetic myocytes. Second, the same effects were produced in cultured diabetic myocytes upon expression of CAp110α. Interestingly, CAp110α did not cause a change in APD₉₀ in wild-type cells. Myocytes from transgenic mice with cardiac-specific expression of a different form of constitutively active p110α were also reported to exhibit no significant differences in action potential waveform or ECG parameters as compared with controls (32). Third, perfusion of Ins2^Akita hearts with insulin caused QTc to shorten, an effect that was blocked by a PI3K inhibitor. We showed earlier that Akt activity is low in Ins2^Akita hearts as compared with nondiabetic hearts, and insulin injection strongly activated cardiac Akt in this diabetic mouse model (18). The ability of insulin to rapidly correct QTc suggests that expression of ion channel proteins that regulate the cardiac action potential (especially Nav1.5) is not greatly altered in Ins2^Akita hearts, in contrast to what has been observed in some other animal models of diabetes (23,26,28).

A recent report showed that insulin/PI3K/Akt signaling in left ventricular biopsies was increased in diabetic subjects as compared with control subjects (33), suggesting that PI3K/Akt activity is enhanced in the type 2 diabetic heart due to hyperinsulinemia, rather than reduced as a consequence of insulin resistance. On the other hand, Akt phosphorylation in right atrial appendage biopsies tended to be lower in diabetic patients than in control subjects (34). Limitations of these two studies are the small number of subjects involved, cardiovascular disease in the control patients, and the different medications being taken by the control and diabetic groups that could affect the results. In addition, only basal (i.e., fasted) signaling was studied. We show here that insulin activation of PI3K/Akt is blunted in the db/db heart. Other investigators have also noted a tendency toward decreased insulin-induced PI3K/Akt activation in cardiac preparations from db/db mice (35,36). Attenuated insulin activation of cardiac PI3K/Akt was also reported in ob/ob type 2 diabetic mice (33,37) and a porcine model of diet-induced obesity and insulin resistance (38).

The molecular mechanisms involved in insulin/PI3K regulation of I_NaP are unknown. A number of protein kinases, including cAMP-dependent protein kinase, protein kinase C, and calmodulin-dependent kinase II, affect the trafficking and gating of Nav1.5 (39). Eleven “basal” phosphorylation sites have been identified in Nav1.5 purified from adult mouse ventricular tissue (40). It remains to be determined whether PI3K signaling alters the modification of these or other phosphorylation sites to regulate I_NaP. The ability of Akti treatment to increase I_NaP suggests that Akt might phosphorylate Nav1.5 to regulate the current. It is interesting that the increase in I_NaP observed in the presence of Akti was less than that observed in myocytes from diabetic mice (Fig. 2) or from mice lacking the p110α isoform of PI3K (8), suggesting that other kinases downstream of PI3K might also be involved in regulating the current. A recent study examined transgenic mice with cardiac-specific expression of a constitutively active form of serum- and glucocorticoid-regulated kinase-1 (CA-SGK1), a downstream effector of PI3K that is structurally related to Akt. Unlike our results and those using transgenic mice expressing constitutively active p110α (32), cardiac myocytes from CA-SGK1 mice exhibited APD prolongation and an increase in I_NaP (41). The apparent discrepancy between these results could be due to differential regulation of I_NaP and other currents that define the cardiac action potential by different protein kinases downstream of PI3K.

In conclusion, we have shown that decreased PI3K signaling in the diabetic mouse heart leads to an increase in I_NaP, which plays a major role in provoking QT prolongation. These findings are consistent with our study showing that cardiac-specific ablation of the PI3K catalytic subunit p110α resulted in elevated I_NaP and QT prolongation in mice, and that pharmacological inhibition of PI3K signaling in canine myocytes led to APD prolongation and alterations in multiple currents, with the increase in I_NaP and decrease in I_Kr contributing the most to APD prolongation (8). Based on these results, we predict that diabetic patients have multiple cardiac ion current abnormalities that predispose them to long QT syndrome and that decreased PI3K signaling due to insulin resistance is a mechanistic explanation for QT prolongation in this patient population.

This work was funded by NIH grants DK-62722 (R.Z.L.), HL-67101 (I.S.C.), and HL-94410 (I.S.C.) and a Veterans Affairs Merit Award (R.Z.L.).

No potential conflicts of interest relevant to this article were reported.

Z.L., Y.-P.J., C.-Y.C.W., L.M.B., and S.L. performed the experiments, analyzed data, and wrote the manuscript. E.S.C. generated the CAp110α adenovirus. M.R.R. contributed to the discussion, analyzed data, and wrote the manuscript. I.S.C. and R.Z.L. supervised the study, analyzed data, provided funding, and wrote the manuscript. R.Z.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.