Cardiac Troponin I Directly Binds and Inhibits Mitochondrial ATP Synthase with a Noncanonical Role in the Post-Ischemic Heart

Abstract

Cardiac troponin I (cTnI) is a key regulator of cardiomyocyte contraction. However, its role in mitochondria is unknown. Here we show that cTnI localized to mitochondria in the heart, inhibited mitochondrial functions when stably expressed in non-cardiac cells, and increased the opening of the mitochondrial permeability transition pore under oxidative stress. Direct, specific, and saturable binding of cTnI to F1FO-ATP synthase was demonstrated in vitro using immune-captured ATP synthase and in cells using proximity ligation assay. cTnI binding doubled ATPase activity, whereas skeletal troponin I and several human pathogenic cTnI variants associated with familial hypertrophic cardiomyopathy did not. A rationally designed peptide, P888, inhibited cTnI binding to ATP synthase, inhibited cTnI-induced increase in ATPase activity in vitro, and reduced cardiac injury following transient ischemia in vivo. We suggest that cTnI-bound ATP synthase results in lower ATP levels, and releasing this interaction during cardiac ischemia-reperfusion may increase the reservoir of functional mitochondria to reduce cardiac injury.

The human heart produces and consumes an average of 6 kilograms of adenosine triphosphate (ATP) per day, relying primarily on mitochondrial metabolism for its energy demands¹; in the healthy heart, 60–70% of ATP is utilized for myocyte contraction^2–4. Actin and tropomyosin form the thin filament and are regulated by the troponin complex comprised of troponin T (tropomyosin-binding), troponin C (calcium-binding), and troponin I (inhibitory) subunits. In the relaxed state, troponin T binds tropomyosin, sterically blocking active sites on actin and preventing interaction with myosin. Upon calcium binding to troponin C, its affinity for troponin I increases, and conformational change in troponin and tropomyosin allows actin to form cross-bridges with myosin, enabling contraction⁵. While cytoskeletal elements such as actin and microtubule filaments interact with mitochondria to regulate mitochondrial fusion-fission^6–8, trafficking^9,10, membrane attachment¹¹, and cellular metabolism^12–15, whether contractile proteins have a direct role in regulating oxidative phosphorylation is not known.

Myocardial infarction, the leading cause of cardiovascular death¹⁶, causes impaired mitochondrial function, bioenergetics and redox balance in cardiomyocytes, which ultimately results in cell death, tissue remodeling and the development of heart failure^17–19. Under basal conditions, the majority of cardiac troponin I (cTnI) is myofibril-bound, but ~2–4% remains unbound to the contractile element²⁰. During cardiac ischemia-reperfusion (IR) injury, cTnI phosphorylation induces structural alterations in the protein and increases its proteolysis, affecting muscle contractility^5,21,22 concomitantly with reduced mitochondrial function²³. Ischemia also induces the release of troponin subunits from the contractile elements, and the increase in circulating cTnI levels is used as a biomarker for myocardial injury in patients^20,24.

In addition to its established structural role, non-cytoskeletal cTnI can localize to the nucleus²⁵ where it regulates transcription^26,27 and induces inflammatory signaling^28,29 in heart failure. Furthermore, human cTnI mutations are associated with aberrations in mitochondrial structure and function in the heart through unclear mechanisms^30–32, suggesting a potential direct connection between cTnI and mitochondrial functions. Here we identified how cTnI directly regulates mitochondrial functions by directly binding ATP synthase and a means to inhibit this non-canonical role of cTnI to reduce cardiac injury following ischemic insult.

Results

cTnI localizes to mitochondria and regulates their function.

To determine if cTnI directly regulates mitochondrial functions, we first examined whether cTnI localizes to mitochondria. Using mouse hearts, we found that some endogenous cTnI associates with both subsarcolemmal (SSM) and interfibrillar (IFM) mitochondria-enriched fractions under basal conditions, whereas another contractile protein, b-actin, did not (Figure 1a). To determine if cTnI enters the mitochondria, mitochondria-enriched fractions were subjected to proteinase K digestion. Intact mitochondria exposed to proteinase K showed complete degradation of the outer mitochondrial membrane protein Tom20, but only partial degradation cTnI and no loss of the matrix protein Hsp60 unless co-incubated with Triton-X to disrupt the integrity of the inner mitochondrial membrane (Figure 1b). Similarly, mitoplasts (generated by osmotic shock, OS), where the outer mitochondrial membrane is lost, contained a similar amount of cTnI to proteinase K-treated mitochondria, which was only degraded with Triton-X and proteinase K co-treatment. These experiments confirm the intramitochondrial localization of cTnI. To visualize the extent of mitochondrial cTnI localization in the intact mouse heart, electron microscopy was performed with immunogold labeling against anti-cTnI rabbit polyclonal IgG or rabbit IgG control (Figure 1c). Anti-cTnI labeling was greater in number than IgG control, and as expected, was primarily sarcomeric (87% of foci). Interestingly, Ten percent of anti-cTnI foci were associated with mitochondria and 3% localized elsewhere. To further visualize cTnI localization in a cell-based system, we transfected H9c2 cardiac myoblasts with a dual-labeled cTnI construct possessing fluorescent tags (GFP on the N-terminus and mCherry on the C-terminus) (Figure 1d). C-terminal mCherry fluorescence greatly overlapped with mitochondria labeled with MitoTracker, especially in the cell periphery. Notably, the localization of N-terminal-labeled cTnI was cytosolic and punctated and not co-localized with the mitochondria. This may indicate that a portion of the N-terminal extension of cTnI is cleaved along with the tag and doesn’t enter mitochondria.

Figure 1.

cTnI localizes to mitochondria. (a) Total, cytosolic, subsarcolemmal mitochondria (SSM)- and interfibrillar mitochondria (IFM)-enriched protein fractions were isolated from mouse hearts, and western blot was used to determine the presence of cTnI. VDAC1 was used as mitochondrial protein control and b-actin was used as a cytosolic protein control; n=3. (b) Proteinase K (PK) digestion of mitochondria and mitoplasts, generated with osmotic shock (OS), in the presence or absence of Triton-X (T-X). Experiment repeat three independent times. (c) Top: Immunogold-labeled electron micrograph of mouse heart with antibody against cTnI. Bottom: Quantification of immunogold foci localization to sarcomeric (Sarc), mitochondria (Mito) and other sites (left histogram) and the specificity of the immunogold staining in the mitochondria presented as a percentage of mitochondrial/non-sarcomeric localization; ~2X higher localization of anti-cTnI antibodies relative to control IgG (right histogram); n=9. Left: Two-way ANOVA with Sidak correction for multiple comparisons. Right: unpaired two-tailed t-test. (d) H9c2 rat cardiac myoblasts were transfected with a single cTnI construct possessing both N-terminal GFP and C-terminal mCherry tags. Images were taken for mCherry (red) on the C-terminus (top left), GFP on the N-terminus (bottom left), and mitochondria stained with Mitotracker (pseudo-color blue, middle left). Co-localization of C-terminal cTnI with mitochondria is observed (top right), but no co-localization of N-terminal cTnI is seen (bottom right), with a magnified view (middle right). Experiment repeated three independent times. All data are presented as mean +/− SD.

We reasoned that if cTnI can localize to mitochondria, this should also occur when cTnI is expressed in non-cardiac cells. We therefore engineered human embryonic kidney cells (HEK293T) to stably express full length cTnI (HEK-cTnI; Figure 2a, top). cTnI localized to the mitochondria-enriched fraction in HEK-cTnI cells (Figure 2a, bottom). Relative to control HEK cells, HEK-cTnI cells showed no significant change in mitochondrial mass by VDAC1 protein levels (Figure 2a, Extended Data Fig 1b, 1c), by MitoTracker Green fluorescence using flow cytometry (Figure 2b, Extended Data Fig 1g), and by citrate synthase activity (Extended Data Fig 1a), and yet exhibited a 30% decrease in ATP level (Figure 2c). Using flow cytometry, we found that mitochondrial membrane potential was lower in HEK-cTnI cells compared to HEK controls (Figure 2d), as evidenced by a decrease in TMRM fluorescence. Interestingly, FCCP-uncoupled TMRM fluorescence was similar between HEK and HEK-cTnI groups, as was oligomycin-treated TMRM fluorescence, also suggestive of unchanged mitochondrial mass in HEK-cTnI cells. HEK-cTnI cells also exhibited a decrease in mitochondrial oxygen consumption (Figure 2e), including basal (state II) rates (Figure 2f), proton leak (Figure 2g) and ATP-linked mitochondrial respiration (Figure 2h). Confirming that the change in mitochondrial function was specific to stable cTnI expression, HEK cells stably expressing GFP showed no difference in membrane potential or oxygen consumption relative to untransduced HEK controls, suggesting that the changes in mitochondrial membrane potential and oxygen consumption are specific to cTnI expression in these non-cardiac cells (Extended Data Fig 1d-f). Finally, when exposed to an oxidant stressor, H₂O₂, HEK-cTnI cells had a greater decrease in TMRM fluorescence, which was inhibited by the mitochondrial permeability transition pore (mPTP) inhibitor cyclosporine A (CsA), suggestive of a possible increase in the rate of opening of the mPTP (Figure 2i-k). These data suggest that cTnI expression in non-cardiac cells inhibits mitochondrial function and increases their vulnerability to oxidative stress without a change in mitochondrial mass.

Figure 2.

Expression of cTnI in HEK293T cells inhibits mitochondrial function. (a) HEK293T cells were engineered to stably express cTnI and compared to unmodified HEK293T cells (top). Mitochondria-enriched fraction shows association of cTnI with the mitochondrial fraction (bottom). Experiment repeated four independent times. (b) HEK cells expressing cTnI (HEK-cTnI) cells have unchanged mitochondrial mass as determined by MitoTracker Green flow cytometry (n=6), two-tailed t-test. (c) HEK-cTnI have decreased basal ATP levels as seen in a luciferase assay (n=3), two-tailed t-test. (d) HEK-cTnI cells have decreased mitochondrial membrane potential by TMRM fluorescence using flow cytometry (n=7), two-way ANOVA. HEK-cTnI cells have decreased mitochondrial oxygen consumption. (e) Representative Seahorse experiment with serial injection of oligomycin (Oligo), 2,4-dinitrophenol (DNP), antimycin and rotenone (Rot+AA) normalized per microgram of protein. Quantifications for basal respiration (f), proton leak (g), and ATP-linked respiration (h) normalized to microgram of protein; n=3, two-tailed t-test. HEK-cTnI cells have increased rate of mPTP opening as measured by rate of loss of TMRM signal after addition of 500μM H₂O₂. Cyclosporine A (CsA; 1μM), an inhibitor of mPTP opening, was used as control; n=3–4. (i) Representative image showing TMRM fluorescence 10 minutes and 30 minutes after addition of H₂O₂. Scale bar = 20μm. (j) Representative experiment showing a higher slope of loss of TMRM signal in HEK-cTnI cells relative to HEK controls; n=4. (k) Quantification of the relative change in TMRM fluorescence as measured as the slope between 10-minute and 30 minute time points. All data are presented as mean +/− SD.

cTnI directly binds ATP synthase.

A mass spectrometry analysis of protein interactions in the mouse heart previously suggested a potential interaction between cTnI and ATP synthase³³. We therefore examined the possibility that cTnI binds ATP synthase and directly affects its function. ATP synthase, a complex of 18 subunits, is a key component of the oxidative phosphorylation system and is proposed to be a critical component of the mPTP^34–37. To determine whether cTnI interacts with ATP synthase, we first used an in silico approach that we have successfully applied before to identify potential partners of cTnI; we often find that inducible protein-protein interactions are characterized by a short sequence of similarity between the two interacting proteins^38,39. While these sites of homology are unlikely to directly interact, they can predict the site of inducible interaction in one of the protein partners. A search for a short homology stretch between cTnI and potential binding partners identified ATP synthase as a potential partner; a ten amino acid sequence A₄₃SRKLQLKTL₅₂ in cTnI is 70% identical and 80% homologous to the N-terminal sequence in the d subunit of ATP synthase, A₂GRKLALKTI₁₁ (Figure 3a). This sequence in cTnI and ATP synthase subunit d is largely conserved among all mammals (Figure 3b).

Figure 3.

cTnI binds mitochondrial F₁F_O-ATP synthase. (a) Sequence homology predicted a possible inducible interaction between cTnI and ATP synthase based on the presence of a 10 amino acid stretch in subunit d, showing 80% similarity. (b) cTnI aa43–52 and ATP synthase subunit d aa2–11 are highly conserved in mammals. Portion of protein sequence shown. Red shading indicates frequency-based difference at each amino acid residue. Green box shows sequence of similarity between cTnI and ATP synthase subunit d. Analysis by constraint-based Multiple Alignment Tool (COBALT) plotted on NCBI Multiple Sequence Alignment Viewer. (c) A proximity ligation assay in H9c2 myoblasts reveals an interaction between cTnI and ATP synthase subunit d as strong punctuate signals (red foci) were observed and overlapped with mitochondria labeled with Mitotracker Deep Red (pseudocolored green). Scale bar = 20μm. Experiment repeated three independent times. (d) Best model for the molecular docking of cTnI in ATP synthase complex is shown. cTnI is shown as red surface and the proteins in ATP synthase complex interacting with cTnI (within 4.5Å distance) are shown as colored surfaces (ATP synthase subunit d in blue, ATP synthase subunit α in cyan, ATP synthase peripheral stalk-embrane subunit b in pink, ATP synthase-coupling factor 6 in green, ATP synthase subunit f in purple and ATP synthase protein 8 in gray). All other proteins in ATP synthase complex are shown as light blue ribbons. Box on the left is detailed scope showing the relative positions of short homology stretch in cTnI and ATP synthase subunit d in the model, corresponding to peptide P888 sequence (yellow); though these regions appear to be in close proximity in the model, these regions interact very weakly and don’t contribute significantly to the overall interaction of cTnI and ATP synthase.

To determine if there is direct binding between cTnI and ATP synthase, we next used a proximity ligation assay (PLA) in H9c2 cardiac myoblasts to visualize this protein-protein interaction at a single molecule resolution. Indeed, PLA demonstrated specific interactions between native cTnI and ATP synthase subunit d in mitochondria as evidenced by punctate foci overlapping a mitochondrial stain (Figure 3c). Since ATP synthase subunit d is located in the matrix sector of the peripheral stalk, the PLA provides further support of cTnI localizing within mitochondria as the cTnI-ATP synthase subunit d protein-protein interaction occurs in the mitochondrial matrix. In addition, a molecular docking simulation of cTnI on ATP synthase, using Molecular Operating Environment (MOE) software, further suggests that cTnI may bind in the peripheral stalk region of ATP synthase (Figure 3d and Extended Data video 1).

cTnI increases ATP hydrolysis and decreases ATP synthesis.

To determine if cTnI binding to ATP synthase affects ATP synthase activity, we used an immune-captured ATP synthase-based assay. We first confirmed that recombinant cTnI bound to ATP synthase and that this binding is saturable (Figure 4a, black). We also found that in rat heart mitochondria, some endogenous cTnI associated with immune-captured ATP synthase complex (Extended Data Fig 2a), and that ATP synthase subunit d co-immunoprecipitated with cTnI (Extended Data Fig 2b). Using blue-native gel electrophoresis, we found that cTnI and ATP synthase subunit d co-localize with the ATP synthase complex in isolated mouse heart mitochondria (Extended Data Fig 2c). A 10-amino acid peptide corresponding to the cTnI homology site in ATP synthase subunit d (Figure 3a), which we termed P888, inhibited cTnI binding to the immune-captured ATP synthase by ~78% (Figure 4a, purple). We then showed that cTnI binding doubled the ATPase activity (Figure 4b), and P888 blocked the cTnI-induced increase in ATPase activity (Figure 4b-c; IC₅₀=0.55 μM), supporting the conclusion that cTnI binding of ATP synthase increases the hydrolytic activity of this complex.

Figure 4.

P888 modulates cTnI-ATP synthase binding and ATPase activity in vitro. (a) cTnI demonstrates a dose-dependent binding to ATP synthase by ELISA; binding is inhibited in the presence of P888 (2μM); n=3. (b) ATPase activity increases in the presence of cTnI, and P888 co-incubation prevents this effect; n=3 independent experiments conducted in 3–4 replicates each, two-way ANOVA with Sidak correction. (c) Dose-dependent inhibition by P888 of cTnI’s (2μg) effect on ATPase activity, with IC₅₀ 0.55μM: n=6 independent experiments. (d) Schematic of cTnI (top; white) with mutations at R21C, R141Q, S166F, and ΔK177 and the phospho-mimetic substitution S42/44D, IR-induced C-terminal truncated fragment of cTnI (middle; light blue), and N-terminal truncation in ssTnI (bottom; dark blue). (e) C-terminal truncation of cTnI (cTnI_1–193) binds ATP synthase similar to wild-type cTnI, as does cTnI-ΔK177, whereas ssTnI, and hypertrophic cardiomyopathy variants (cTnI-R21C, cTnI-R141Q, and cTnI-S166F) do not; n=6. cTnI-S42/44D (mimetic of phosphorylated cTnI) exhibits greater maximal binding to ATP synthase than unphosphorylated cTnI. Two-way ANOVA with Sidak correction. #: p<0.001 vs cTnI 1–210. (f) C-terminal truncated cTnI (cTnI 1–193) increases ATPase activity similar to full recombinant cTnI, whereas ssTnI does not. Hypertrophic cardiomyopathy variants in cTnI differentially impact ATPase activity; n=3 independent experiments. Two-way ANOVA with Sidak correction. (g) cTnI binding to ATP synthase correlates with ATPase activity (R²=0.9); n=6. (h) Phosphomimetic cTnI-S42/44D exhibits greater binding affinity and maximal binding to ATP synthase when compared to unphosphorylated cTnI; n=4. (i) Schematic of experimental procedure: Mitochondria were isolated from mouse heart and brain and co-incubated with cTnI (4μg mitochondria:1μg cTnI), and oxygen consumption and ATP synthesis rates were measured. (j) Co-incubation of mitochondria isolated from mouse heart or brain decreases the rate of maximal ATP synthesis by luciferase assay. n=3 with 3–4 replicates each, two-tailed t-test. (k) Co-incubation of isolated mitochondria with cTnI decreases ADP-linked (state III) oxygen consumption rate in brain and heart; n=3, two-tailed t-test. Panels (d) and (i) were created with BioRender.com. All data are presented as mean +/− SD.

N-terminal truncation of cTnI occurs at low levels in the normal heart⁴⁰, increases in adaptation to hemodynamic stress⁴¹, and is associated with improved myocardial relaxation⁴². C-terminal truncation of cTnI occurs during IR²² and myocardial stunning⁴³ and is associated with impaired contractile function⁴⁴. We determined whether protein fragments of cTnI (Figure 4d) differentially bind ATP synthase (Figure 4e) and affect ATPase activity (Figure 4f). Indeed, we found that C-terminal truncated cTnI (cTnI_1–193; Figure 4d, light blue) bound ATP synthase and increased ATPase activity in vitro similar to intact protein. Conversely, slow skeletal troponin I (ssTnI; Figure 4d, dark blue), which lacks the N-terminal 30-amino acid extension of cTnI, did not bind ATP synthase and had no effect on ATPase activity. This suggests that the N-terminal region of cTnI is necessary in the modulation of F₁F_O-ATPase activity, whereas the C-terminus is not.

Pathogenic variants in cTnI lead to cardiomyopathy with heterogeneous phenotypes (hypertrophic, restrictive and dilated) and disease course (progression rate, risk of sudden cardiac death)⁴⁵. We determined if select pathogenic mutations in cTnI⁴⁶ would affect its binding to ATP synthase and ATPase activity. Indeed, cTnI-R21C, cTnI-R141Q, and cTnI-S166F did not bind to and had no effect on ATPase activity, whereas a single codon deletion (cTnI-ΔK177) bound ATP synthase and increased ATPase activity similar to wild-type cTnI protein (Figure 4e, 4f); binding of a cTnI variant to ATP synthase tightly correlated with ATPase activity with R² of 0.90 (Figure 4g). This suggests that some pathogenic variants in cTnI do not affect mitochondrial ATPase activity because they cannot bind the ATP synthase complex.

Phosphorylation of cTnI during IR is associated with mitochondrial dysfunction. To determine if cTnI phosphorylation affects its ability to bind ATP synthase and increase ATPase activity, we generated recombinant cTnI with phosphomimetic amino acid substitutions (serine to aspartate) at Ser42 and Ser44, both of which are known targets of δPKC during IR. Phosphomimetic cTnI-S42/44D exhibited greater maximal binding and higher binding affinity (Figure 4h) to ATP synthase, and an exaggerated effect on ATPase activity relative to wildtype cTnI. This suggests that cTnI phosphorylation may increase its effect on ATP synthase.

In respiring mitochondria, F₁F_O-ATP synthase catalyzes the conversion of ADP to ATP, while in hypoxic conditions where mitochondrial membrane potential is impaired, it reverses to hydrolyze ATP⁴⁷. Given the profound effect of cTnI on ATP hydrolysis, we next measured its effect on oxygen consumption and ATP synthesis in isolated mitochondria co-incubated with purified cTnI (Figure 4i). Consistent with the HEK cell data (Figure 2c, h), cTnI treatment decreased the rates of ATP synthesis (Figure 4j) and ADP-dependent respiration (Figure 4k) in mitochondria from both mouse brains and mouse hearts. This suggests that cTnI has distinct and opposing effects on ATP synthesis and hydrolysis by F₁F_O-ATP synthase complex.

Inhibitor of cTnI-ATP synthase binding prevents IR injury.

We reasoned that if cTnI regulates mitochondrial function in vivo by directly binding to ATP synthase, this effect will be inhibited by P888, the peptide inhibitor of this interaction. We therefore assessed the effect of P888 treatment on the outcome of myocardial IR injury in rats subjected to transient (30 minutes) ligation of the left anterior descending (LAD) coronary artery. At reperfusion, rats were injected intraperitoneally with TAT peptide (a control peptide that enables biological membrane crossing⁴⁸) or TAT-P888 conjugate (single injection, 3mg/kg; Figure 5a). Infarct size as measured by TTC staining was increased in IR, and this was prevented by TAT-P888 treatment (Figure 5b). Serum creatine kinase (CK) measured at one time point, 24 hours after reperfusion, increased with IR, and was attenuated with P888 treatment (Figure 5c). Three days after IR, lactate dehydrogenase (LDH) levels were increased, and this was prevented with P888 treatment (Figure 5d). Echocardiography in the presence and absence of isoproterenol demonstrated a decrease in systolic function (Figures 5e-f) as measured by an increase in LV end-systolic dimension and a decrease in fractional shortening in the vehicle-treated IR group compared to sham surgery controls; this decrease was almost completely prevented in rats treated with TAT-P888 at reperfusion only. There was also a trend towards an increase in LV end-diastolic dimension (LV EDD). Mitochondrial respiration measured 24 hours after reperfusion was impaired following IR, but was not improved in the P888-treated group. These data suggest that cTnI’s interaction with ATP synthase at reperfusion contributes to cardiac IR injury.

Figure 5.

P888 treatment prevents IR-induced injury in vivo in rats. (a) Time course of in vivo myocardial IR in the presence of vehicle (TAT) or TAT-P888 treatment. (b) TAT-P888 attenuates IR-induced infarct size as measured by TTC staining, n=5 animals per group. (c) Serum creatine kinase (CK) at 1 day after IR, n=15 animals per group, two-way ANOVA with Sidak correction. (d) TAT-P888 attenuates serum lactate dehydrogenase (LDH) increase at 3 days after IR compared to vehicle, n=15 animals per group, two-way ANOVA with Sidak correction. Basal (e) and isoproterenol (f) echocardiography 3 days after IR demonstrated a decrease in systolic function, as measured by fractional shortening and an increase in LV end-systolic dimension, in vehicle-treated animals. The decrease in fractional shortening and increase in LV end-systolic dimension were prevented by treatment at time of reperfusion with TAT-P888. Non-significant trends in LV end-diastolic dimension are also shown, n=5 animals per group, two-way ANOVA with Sidak correction. (g) Mitochondrial oxygen consumption at 1 day after IR measured by Seahorse, n=15 animals per group, two-way ANOVA with Sidak correction. Panel (a) created with BioRender.com. All data are presented as mean +/− SD.

Discussion

Cardiac troponin I regulates cardiomyocyte contraction in conjunction with troponin C and troponin T via the calcium-mediated interaction between actin and myosin. Here we identified a mitochondrial role of cTnI: the inhibition of mitochondrial respiration by direct binding of cTnI to the ATP synthase complex. This conclusion is based on the following evidence: First, cTnI is associated with mitochondria in the heart, in cardiac myoblasts, and in non-cardiac HEK cells expressing cTnI. Second, a proximity ligation assay indicates that cTnI interacts with the ATP synthase complex. Third, expression of cTnI in non-cardiac HEK cells inhibited mitochondrial function as evidenced by impaired oxidative phosphorylation, a 30% decrease in ATP level, a decrease in mitochondrial membrane potential and increased sensitivity to oxidative stress induced by H₂O₂ treatment. Fourth, cTnI bound to immune-captured ATP synthase in a saturable manner and P888, a peptide corresponding to the sequence homology between ATP synthase subunit d and cTnI, blocked this binding at IC50=0.55 μM. Fifth, the functional consequences of the cTnI-ATP synthase interaction was demonstrated in vitro; recombinant cTnI increased the ATPase activity of immune-captured F₁F_O ATPase, which was inhibited by P888, and treatment of isolated mitochondria with cTnI decreased mitochondrial respiration and ATP synthesis. Sixth, phosphorylation of cTnI at Ser42/44 led to higher ATPase binding affinity and increased ATPase activity, and the decreased or loss of effect of several pathogenic mutants of cTnI and the skeletal isoform of troponin I on ATPase activity correlated tightly with the decrease in their ability to bind to the ATP complex. Finally, myocardial IR-induced injury after transient coronary artery occlusion in rats was greatly reduced with TAT-P888 treatment at time of reperfusion. The acute protection resulted in inhibition of the IR-induced decrease in systolic function as measured three days after infarction. Together, our work reveals a cTnI interaction with ATP synthase and the negative functional consequences of this interaction in IR-induced injury (Figure 6).

Figure 6.

cTnI localizes to mitochondria, where it binds ATP synthase (peripheral stalk shown in red), decreases mitochondrial respiration and membrane potential, increases opening of the mitochondrial permeability transition pore (mPTP) in the inner mitochondrial membrane after reactive oxygen species (ROS) exposure, and increases IR injury. The peptide P888 prevents binding of cTnI to ATP synthase and decreases IR injury. Figure 6 was created with BioRender.com.

The F₁F_O ATP synthase is a protein complex with dual roles. It is a key component of oxidative phosphorylation in the production of ATP in respiring mitochondria. In addition, ATP synthase may serve as a key component of the mPTP, opening in response to cellular stress leading to cell death^34–37. A protein-protein interaction between cTnI and ATP synthase subunit a (ATP5f1a) was previously observed in a crosslinking mouse heart proteomics dataset³³ and an atlas of protein-protein interactions across mouse tissues identified an interaction between cTnI and ATP synthase subunits⁴⁹. However, the consequence of this interaction on cardiac physiology was not assessed. Here we found that HEK cells stably expressing cTnI have decreased oxidative phosphorylation under basal conditions and increased loss of membrane potential following H₂O₂ treatment, suggestive of mPTP opening. We also found that cTnI increased mitochondrial F₁F_O ATPase activity in vitro and decreased ATP synthesis in isolated mitochondria. In healthy respiring mitochondria, high membrane potential favors ATP synthesis by F₁F_O ATP synthase; however, when respiration is compromised, membrane potential decreases and ATP synthase can reverse to hydrolyze ATP^50–52. It is possible that cTnI may function to inhibit ATP synthase and mitochondrial functions in respiring mitochondria (under basal conditions) and increases ATP hydrolysis during conditions of hypoxia. It is also possible that cTnI may affect the stability of the ATP synthase complex under stress conditions, or that cTnI may bind other mitochondrial proteins. Regardless, mitochondrial cTnI clearly exacerbates mitochondrial dysfunction and increases cell injury during excessive stress, such as during acute myocardial infarction.

Mutations within the TNNI3 gene (which encodes cTnI) result in substantial phenotypic variability, with most mutations associated with hypertrophic cardiomyopathy, whereas others cause dilated or restrictive phenotypes⁴⁵. The factors leading to this phenotypic heterogeneity are not well-understood, and the avenues of targeted therapy for familial cardiomyopathies with TNNI3 mutations are limited⁵³. TNNI3 mutations are associated with mitochondrial abnormalities in mouse models through unknown mechanisms. Importantly, mitochondrial dysfunction has also been seen in hypertrophic cardiomyopathy attributed to mutations in other sarcomeric genes such as MYH7 and MYBPC3⁵⁴. Here, we show that some genetic variants of TNNI3 with pathological phenotype have heterogeneous binding to F₁F_O-ATP synthase which correlates with effects on ATPase activity. Note that the variants that we have studied represent only few of those identified; benign and pathogenic variants have been reported in humans in at least half of the amino acids in this 210 amino acid protein. While not yet known, it is possible that single amino acid substitutions (or deletions) in cTnI affect protein structure to change its affinity to ATP synthase. However, it remains to be seen whether the differences in binding to ATP synthase and ATPase activity play any role in the TNNI3 cardiomyopathy phenotype. The downstream effects of TNNI3 mutation on mitochondrial function beyond ATPase activity and the role of this effect in disease pathogenesis are areas of ongoing investigation in our lab.

Phosphorylation of cTnI by δPKC contributes to contractile dysfunction during pathological hypertrophy and heart failure⁵⁵. We previously showed that inhibiting phosphorylation of cTnI by δPKC during IR in rats was associated with an attenuation of IR-induced mitochondrial dysfunction²³. Here, we find that phosphomimetic recombinant cTnI at two δPKC sites (cTnI-S42/44) exhibited higher affinity of binding to F₁F_O-ATP synthase and had a more pronounced effect than unphosphorylated cTnI on ATPase activity. This may shed some light on non-contractile mechanisms by which cTnI phosphorylation can lead to contractile dysfunction.

N-terminal truncation of cTnI occurs at low levels in normal hearts and increases in response to stress (e.g., IR injury, beta-adrenergic activation, and aging). Overexpression of this fragment (lacking cTnI aa 1–30) improves myocardial relaxation^40–42. Interestingly, differential expression of N-terminal cTnI fragments has been shown to occur in stunned myocardium of patients after coronary artery bypass graft⁵⁶. Similarly, in transgenic mice, expression of the slow skeletal isoform of troponin I (ssTnI, which lacks the cardiac-specific N-terminal region) in the heart improved metabolic efficiency in heart failure⁵⁷ and IR injury^58,59 and provided protection from endotoxemia⁶⁰. Here, we show that unlike cTnI, ssTnI does not increase ATP hydrolysis. It is thus possible that the beneficial effect of N-terminal truncation could be at least in part due to abrogation of cTnI’s effect on mitochondrial function, supporting a damaging effect of cTnI’s N-terminal extension on mitochondria.

Several limitations of our study should be highlighted: (1) Although cTnI is present in murine cardiac fraction enriched in mitochondria, by itself, it can represent contamination of this fraction by a very abundant cardiac protein. However, the resistance of mitochondria-associated cTnI in the mitochondrial fraction to proteinase K, the mitochondrial association of cTnI by immunogold electron microscopy, the localization of fluorescent-tagged cTnI at the mitochondria and the demonstration through proximity ligation assay (PLA) of the association of endogenous cTnI with ATP synthase, strongly indicate the localization of cTnI within the mitochondrial matrix. Because cTnI’s molecular weight after PK proteolysis is slightly lower, it is formally possible that it resides, at least in part, within the intermembrane space. Immunogold EM foci show association with mitochondria rather than sub-organellar localization. Fluorescent dual-labeling of cTnI could impact its localization. While each method may have limitations, taken together, they all strongly suggest the mitochondrial matrix localization of cTnI. (2) The characteristics of intramitochondrial cTnI are not yet known. Fluorescent-tagged cTnI localization suggests that the N-terminus (GFP tagged) is cleaved prior to mitochondrial localization, a common feature of mitochondrial imported proteins⁶¹. The cTnI fragment identified in mitochondria by immunoblot after proteinase K treatment is close in size to the full-length protein, and the antibody epitope used for the PLA assay is raised against amino acids 23–29, suggesting that only a small truncation of the N-terminal of cTnI occurs. (3) While quite unlikely, it is also possible that cTnI could affect the mitochondrial import of ATP synthase subunits to modulate ATPase complex assembly or function. However, our data support mitochondrial localization of cTnI and a direct effect of cTnI on ATP synthase activity. (4) Our study does not define the exact site where cTnI binds to the ATP synthase complex. We propose a docking model of cTnI within ATP synthase that relies on the best available structure of cTnI and ATP synthase, but this may not be complete and further validation of this model is required. (5) Although we demonstrated that P888 prevents cTnI binding to ATP synthase and decreases ATPase activity in vitro and in cells, we cannot rule out other mechanisms by which this peptide may have in vivo effects. Evaluation of mitochondrial function at one timepoint after IR demonstrated no significant effect of P888 on mitochondrial function. Additionally, demonstrating P888’s localization in mitochondria in vivo with traditional biochemical methods is limited by peptide kinetics and binding affinity, and would require radiolabeled peptide studies⁶². (6) While previous studies suggest that a single injection of TAT-conjugated short peptides does not trigger an immune reaction^63,64, the potential immunogenic effect of P888 was not assessed.

Together, our study indicates that cTnI binds to mitochondrial ATP synthase, which inhibits ATP synthesis under basal conditions and possibly augments ATP hydrolysis during cardiac ischemia. We suggest that releasing this inactivating effect of mitochondrial cTnI and preventing ATP hydrolysis may increase the reservoir of functional mitochondria to reduce cardiac injury. If confirmed in humans, the protein-protein interaction elucidated in our study between ATP synthase and cTnI can be utilized as a potential drug target to reduce ischemic injury and sequalae of myocardial infarction and may provide mechanistic insight into the pathologies associated with mutations in cTnI.

Methods

Animals

The Administrative Panel on Laboratory Animal Care at Stanford University (protocol #10363), the Animal Care and Use Committee at NHLBI (protocol # H-0308R2), and the Ethic Committee on Animal Use of the Institute of Biomedical Sciences at University of Sao Paulo (protocol #60/2017) approved all animal protocols. Mice were kept on under a 12 h/12 h light/dark cycle, room temperature (~22 °C), and ~50% humidity. Male and female wild-type C57Bl/6 mice (10–12 weeks old), and male and female Wistar rats (10–12 weeks old) were used throughout the study as described below.

Blue-Native Polyacrylamide Gel Electrophoresis

Mitochondria isolated from mouse heart (50 μg) were pelleted at 10,000xg for 10 minutes at 4ºC. Supernatant was aspirated and pellet was solubilized in 5μL of Buffer A (50mM NaCl, 50mM imidazole, 1mM EDTA, pH = 7). Sample was homogenized and digitonin was added (0, 2, 4, or 5 mg digitonin/mg protein) and gently mixed. The sample was incubated on ice for 5 minutes and centrifuged at 20,000xg for 20 minutes at 4ºC. The supernatant was retained and run on Invitrogen NativePAGE 4–16% Bis-Tris gel (ThermoFisher, catalog #BN1002BOX) with sample buffer (1.5M HCl, 12.5% glycerol, 0.25M bis-tris pH = 7, 0.25M NaCl, 0.001% Ponceau stain). Wells were washed three times with dark cathode buffer (15mM bis-tris, 0.02% Coomassie G250 dye, 50mM tricine, pH = 7) before loading samples. Gel was run at 150V for 1 hour on ice with dark cathode buffer and anode buffer (50mM bis-tris, pH = 7). Then, the dark cathode buffer was removed and light cathode buffer added (15mM bis-tris, 0.002% Coomassie G250 dye, 50mM tricine, pH = 7). The gel was run for 2 hours at 250V on ice. NativeMARK Unstained Protein Standard (ThermoFisher, catalog #LC0725) was used. After running, ladder was cut off the gel and stained with EZBlue Gel Staining Reagent (Sigma-Aldrich, catalog #G1040) for 1 hour at room temperature. Protein was transferred to PVDF membrane using the TurboBlot Transfer System (BioRad, catalog #1704150) at 25V for 1 hour at room temperature. Then, membrane was blocked in 5% milk in TBST for 30 minutes, and incubated with rabbit monoclonal antibody against cTnI (abcam, ab209809, 1:1000 in 5% milk in TBST) overnight at 4ºC. Membrane was washed 3 times with TBST and incubated with anti-rabbit HRP antibody (Invitrogen, catalog number A27036, Lot #274089, 1:10,000 dilution in 5% milk in TBST) for 3 hours at room temperature, washed 3 times with TBST, and incubated with SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher, Catalog #34096, Lot #YJ376605) for 5 minutes in the dark. Then, the membrane was developed on the c600 Azure Biosystems Imager on the highest sensitivity setting. Membrane was then stripped with Restore Western Blot Stripping Buffer (ThermoFisher, catalog #21059) at room temperature for 30 minutes and re-blocked with 5% milk in TBST for 30 minutes. The membrane was incubated overnight with rabbit antibody against ATP5H (MyBioSource, catalog #MB59127683, 1:1000 in 5% milk in TBST) at 4ºC overnight. Membrane was washed 3 times with TBST, incubated with SuperSignal West Femto Maximum Sensitivity Substrate for 5 minutes in the dark and developed on the c600 Azure Biosystems Imager (Azure cSeries Capture Software 2017).

Cardiac Ischemia Reperfusion – in vivo LAD Coronary Artery Ligation Rat Model

Myocardial infarction was induced by ligation of the left anterior descending (LAD) coronary artery for 30 minutes⁶⁵. 10–12-week-old male and female Wistar rats were anesthetized with 3% isoflurane and intubated with a rodent ventilator at 80 breaths/minute. Body temperature was maintained at 37°C using a heating blanket. Left thoracotomy between the fourth and fifth ribs was performed, and the LAD coronary artery was ligated close to its origin from the aortic root. Normoxia control animals (sham) were exposed to the same procedure with no ligation. The free ends of the ligature were used to form a noose around a syringe plunger placed flat on the myocardium. Coronary occlusion was achieved by tightening the noose around the plunger for 30 minutes. Occlusion was determined by immediate blanching of the infarcted area and reperfusion was achieved by release of the ligature just after injecting an intraperitoneal injection (IP) 3mg/kg of the respective peptides. Serum creatine kinase was measured at 1 day after MI by commercially available kit. Hearts were homogenized in isotonic mitochondrial buffer (300 mM sucrose, 10 mM Hepes, 2 mM EGTA, pH 7.2, 4°C) ⁶⁶. The suspension was homogenized in a 40 mL tissue grinder and centrifuged at 950×g for 5 min. The resulting supernatant was centrifuged at 9500×g for 10 min. Mitochondrial pellet was washed, resuspended in isolation buffer and centrifuged (9500×g for 10 min). The mitochondrial pellet was washed and resuspended in isolation buffer. Mitochondrial oxygen consumption was measured from isolated mitochondria (0.125 mg/mL) in experimental buffer containing 125 mM sucrose, 65 mM KCl, 10 mM Hepes, 2 mM inorganic phosphate, 2 mM MgCl₂, 100 μM EGTA, and 0.01% BSA, pH 7.2 in the presence of succinate, malate, and glutamate substrates (2 mM of each) with continuous stirring at 37 °C using a computer-interfaced Clark-type electrode (OROBOROS Oxygraph−2k). ADP (1 mM—Amresco 0160) was added to induce State 3 respiratory rates. Addition of oligomycin (1 μg/mL—Sigma 4876) was used to determine State 4 rates. 0.1 mM FCCP [Carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone—Enzo BML-CM120] was added to evaluate O₂ consumption in uncoupled mitochondria. Lactate dehydrogenase level was measured at 3 days after MI following manufacturer’s instructions (Eton Bioscience, San Diego, CA, USA). Fractional shortening, left ventricular end-systolic dimension, and left ventricular end-diastolic dimension were determined under basal conditions and after isoproterenol stimulation (10μg/kg, IP) at 3 days after MI by M-mode echocardiography (Acuson Sequoia model 512 echocardiographer equipped with a 14-MHz linear transducer). Treatments were performed with TAT (molecular weight: 1559.83 g/mol; CAS Number: 191936–91-1) and TAT-P888 (TATAGRKLALKTI; N-terminus acetylated C-terminus amidated; molecular weight: 2854 g/mol) at 1μM.

Cell Culture and Treatments

H9c2 cells (ATCC #CRL-1446) were used at passages 7–10 at a confluency of 70%. H9c2 cells were transfected with a cTnI construct with GFP on N-terminus and mCherry on C-terminus. H9c2 cells were grown in DMEM with L-glutamine, 4.5g/L glucose and sodium pyruvate (Corning) and 10% FBS as per ATCC recommendation. A HEK293T cell line (ATCC #CRL-3216) stably expressing human cTnI was generated through retroviral transduction. Untreated HEK cells and HEK cells stably expressing GFP via viral transduction were used as control. For mitochondrial function studies, cells were incubated in media containing 1 mM pyruvate, 2 mM L-glutamine, and 2 mM D-glucose.

Cellular ATP Assay

Cellular ATP levels were determined using the ATP detection kit (Cayman Chemical, #700410) according to the manufacturer’s protocol. ATP levels detected were normalized to total protein levels. Protein concentration was measured using bicinchoninic acid (BCA) assay kit (ThermoFisher Scientific, USA).

Citrate Synthase Activity Assay

Citrate synthase activity was determined per manufacturer protocol (Cayman Chemical, #701040). Cell lysates from HEK, HEK-GFP and HEK-cTnI cells were incubated at 1:200 dilution with citrate synthase enzyme, oxaloacetate and acetyl-CoA, and production of SH-CoA was measured as absorbance of Citrate Synthase Reagent at 412nm in 96-well format by plate reader using SoftMax Pro (6.5.1). Each experiment was conducted in duplicates which were averaged, and repeated in three independent experiments. Relative Citrate synthase activity was measured as slope of the linear portion of the curve for absorbance versus time normalized to HEK control condition.

Co-Immunoprecipitation Assay

Mitochondria were isolated as detailed below from rat hearts. 100μg of mitochondria were incubated overnight at 4^oC with cTnI antibody (Abcam (ab47003); lot #: GR3248433–1, 1:500). This lysate was incubated with Pierce^™ Protein A/G Magnetic Beads (Thermo Fisher Scientific 88803; lot #: TL277527) at 4^oC for 4 hours to bind cTnI protein-protein interaction partners. IP protocol was followed according to manufacturer’s instructions and the western blot for input and IP was run as detailed below in Western Blot. Secondary antibodies that don’t bind to heavy- or light-chain IgG were used: Rat monoclonal antibody to mouse IgG (HRP; Abcam ab131368; lot # GR3371049–5, 1:1000) and VeriBlot for IP Detection Reagent (HRP; Abcam ab131366; lot # GR3369925–2, 1:1000).

cTnI Binding to ATP Synthase by ELISA

Microplates with plate-bound antibody to ATP synthase (Abcam ab109714) were blocked with 2% BSA and co-incubated with 4μg of solubilized rat heart mitochondria (Abcam ab110347) in the presence of recombinant human cTnI, ssTnI, cTnI-R21C, cTnI-R141Q, cTnI-S166F, cTnI-delK177, cTnI-S42/44D (2μg) in the absence or presence of P888 peptide (at increasing concentrations or at 2μM). After washes with assay solution, plates were incubated with HRP anti-cTnI antibody (Abcam ab24460 1:5000) or HRP anti-ssTnI antibody (LSBio LS-C712349 1:5000) for one hour, followed by washes, and read after addition of TMB substrate solution (Thermo Fisher N301) followed by 2N sulfuric acid. Absorbance was then quantified by spectrophotometer plate reader at 450nm using SoftMax Pro (6.5.1).

Electron Microscopy with Immunogold Labeling

Three-month-old, male C57BL/6J mouse was anesthetized with inhaled isoflurane. Cardiac tissue was perfused through the apex of the left ventricle with 3 ml of fixative solution (4% formaldehyde, 0.1M phosphate buffer, pH 7.4, Electron Microscopy Sciences) using a 27G needle for 3 minutes. After initial fixation, the left ventricle was removed, placed in fixative solution, and cut into 1 mm³ cubes. Tissue pieces were placed in fixative solution for two hours at room temperature. After fixation, tissue was washed with 0.1M phosphate buffer three times for five minutes each and dehydrated with 10-minute incubations with the ethanol (30%, 50%, 70%, and 90% in water). Tissue was then infiltrated with LR White (LRW) resin (Electron Microscopy Sciences) starting with a 30-minute incubation in two parts 90% ethanol and one part LRW at room temperature, followed by an overnight incubation in equal parts 90% ethanol and LRW at room temperature. The following morning, tissue was incubated in 100% LRW for 4 hours at room temperature. Individual pieces were embedded in gelatin capsules using fresh LRW resin, and blocks were cured in a 50°C oven over the course of two days.

For immunogold labeling, 65 nm-thick sections were collected on copper grids using an ultramicrotome (Leica Microsystems, EM UC7). Grids were incubated in blocking buffer (5% goat serum and 1% bovine serum albumin in PBS) for 30 minutes at room temperature followed by 1-hour incubation at room temperature with primary antibody - anti-cardiac troponin I, rabbit polyclonal IgG (Abcam, ab47003) or rabbit IgG, polyclonal isotype control (Abcam, ab171870) at 1:50 in blocking buffer. Grids were washed five times for three minutes each in blocking buffer followed by 1-hour incubation at room temperature in secondary antibody - goat-anti-rabbit IgG 10 nm gold (Electron Microscopy Sciences, 25109) at 1:20 in blocking buffer. Grids were then washed in PBS five times for three minutes each, fixed with 1% glutaraldehyde (Electron Microscopy Sciences) in PBS for 30 minutes at room temperature, and washed in double distilled water five times for three minutes each. After sections dried, micrographs were acquired with a JEOL JEM 1200 EXII transmission electron microscope operating at 80 kV and equipped with an AMT XR-60 digital camera. Blinded analysis of immunogold foci sites (“sarcomere”, “mitochondria”, “other”) was conducted by two readers and averaged for 8–10 selected images from each condition of similar magnification and quality.

Evaluation of Sub Organelle Localization of cTnI within the Mitochondria

Mitochondria isolated from mouse hearts (2μg/μl) were incubated with proteinase K (5μg/ml) in presence or absence of 0.1% Triton-X for 5 minutes at room temperature, followed by addition of 5mM PMSF to stop the reaction. Mitochondria were then spun at 9000xg for 10 minutes and resuspended in mannitol-sucrose buffer. Mitoplasts were generated from isolated mitochondria by incubation in hypotonic buffer (60mM sucrose, 1.25mM HEPES, 250μM EDTA) and were subjected to similar proteinase K digestion with and without Triton-X. All fractions were analyzed by western blot.

F₁F_O-ATPase Activity Assay

ATPase activity was measured using microplate assay (Abcam ab109714) according to manufacturer protocol. Plate-bound antibody to ATP synthase was co-incubated with 4μg of solubilized rat heart mitochondria (Abcam ab110347) in the presence or absence of recombinant human cTnI (2μg) and P888 peptide (2μM), and absorbance was measured in a spectrophotometer plate reader over a period of 120 minutes. Oligomycin treatment was used as negative control.

Microscopy

H9c2 rat cardiac myoblasts were stained with Hoechst (1:10,000 in staining media) and MitoTracker Deep Red (1.6:10,000 in staining media) for 30 minutes in the dark. Fluorescence images were acquired using an All-in-One Fluorescence Microscope BZ-X700 (Keyence). Pseudo-color was used to better visualize co-localization.

Mitochondrial ATP Synthesis Rate Assay

ATP synthesis rate was determined using the luciferin/luciferase-based ATP bioluminescence Assay Kit CLS II (Roche) with modifications. Isolated mitochondria were suspended in buffer with 5mM pyruvate/malate. Measurements were started immediately by adding luciferin/luciferase and ADP (0.5mM final) in luminescence plate reader. The initial slope of increase in ADP-supported luciferase chemiluminescence was used to determine the rate of ATP production after subtraction of background. Oligomycin was used as control to determine the rate of non-mitochondrial ATP production.

Mitochondrial Isolation

Male and female wild-type C57Bl/6 mice (10–12 weeks old) were anesthetized by isoflurane inhalation with confirmation by toe pinch reflex, followed by cervical dislocation, and excision of heart and brain. Mitochondria were isolated from heart and brain tissue using mannitol-sucrose (MS) buffer with protease inhibitor, and mitochondrial subpopulations were isolated^67,68. Tissue was homogenized using a Potter-Elvehjem glass/teflon homogenizer and spun at 800xg for 10 minutes. Total lysate was saved at this point. The remaining supernatant was spun at 10,000xg for 20 minutes to separate out the subsarcolemmal mitochondrial pellet from the cytosolic fraction. The tissue debris pellet from the first spin was re-homogenized and re-spun at 800xg for 10 minutes, and supernatant was spun at 10,000g for 20 minutes to separate the interfibrillar mitochondrial pellet from the cytosolic fraction. Mitochondrial pellets were washed with MS buffer and spun at 10,000xg for 5 minutes. When used for western blot, the final pellet was resuspended in MS buffer with 1% Triton-100X.

Mitochondrial Mass

Mitochondrial mass was assessed in cells by staining for the mitochondrial outer membrane protein VDAC-1 or using MitoTracker Green FM. For VDAC-1 staining, cells were detached, fixed and permeabilized using eBioscience Intracellular Fixation Permeabilization kit (Invitrogen) and stained with VDAC-1 antibody (CL647–55259,1:500, Proteintech) or isotype control for 20 min and used for flow cytometry analysis. For Mitotracker staining, HEK cells were incubated with MitoTracker Green (100 hM) for 20 minutes in medium without serum at 37oC, washed with PBS, detached and used for flow cytometry analysis. Flow cytometry was performed using a Cytek DxP10 flow cytometer (Cytek Biosciences, Inc) or a Novocyte Quanteon flowcytometer (Agilent). Gating strategy used for analysis excluded doublets and 7-AAD stained dead cells. Mitochondrial mass was also measured by western blot for mitochondrial proteins as described below.

Mitochondrial Membrane Potential

Mitochondrial membrane potential was assessed using tetramethylrhodamine, methyl ester, perchlorate (TMRM). HEK cells were incubated with TMRM (25 ηM) for 20 minutes in medium without serum at 37^oC. Cells were washed with PBS, detached and analyzed by flow cytometry using a Cytek DxP10 flow cytometer (Cytek Biosciences, Inc) or Novocyte Quanteon flow cytometer (Agilent). Cells were pre-treated with FCCP (5μM) or oligomycin (2.5μM) for 30 minutes prior to TMRM incubation to determine depolarized and hyperpolarized fluorescence. Gating strategy used for analysis excluded doublets and 7-AAD positive dead cells. 20,000 events were acquired for each sample.

Mitochondrial Permeability Transition Pore (mPTP) Opening Assay

mPTP opening was measured after TMRM and Hoechst staining⁶⁹. Live cell microscopy was completed using BZ-X700 (Keyence) in a 37°C 5% CO₂ chamber with image acquisition every 1–2 minutes for 30–40 minutes using BZ-X Viewer (1.03.00.05). Relative fluorescence intensity was used to measure mitochondrial membrane potential, and after 500μM H₂O₂ was added to cells, the slope of decrease of TMRM signal calculated from all obtained images over a 30-minute period was used to measure the relative rate of mPTP opening. Fluorescence intensity quantification was done in FIJI ImageJ.

Molecular Docking of cTnI in ATP Synthase Complex

PDB ID: 4Y99 was loaded into Molecular Operating Environment (MOE) 2019.0102 software and 4Y99.C chain (cTnI) was kept and other chains were deleted. 4Y99.C was prepared using QuickPrep functionality in MOE at default settings. QuickPrep optimizes the H-bond network and performs energy minimization on the system. PDB ID: 6J5I⁷⁰ (ATP synthase complex) was also loaded in MOE 2019.0102 and all the ligands and metals were removed and the system was prepared using QuickPrep functionality as above. 6J5I.d chain (ATP5H) and other chains within 4.5 Å distance from 6J5I.d were kept whereas all other chains were deleted. 6J5I.d and all other chains of 6J5I within 4.5 Å distance from 6J5I.d were defined as receptor and 6J5I.d was defined as the dock site. 4Y99.C was docked in the receptor using Protein-Protein dock functionality in MOE 2019.0102 at default settings. The lowest energy docking pose of cTnI in ATP synthase complex is reported in this study. Molecular graphics performed with UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases⁷¹.

Peptide Synthesis

P888 peptide was synthesized on solid support using a fully automated microwave peptide synthesizer (Liberty, CEM Corporation) using homology sequence⁷². For in vivo studies, P888 was conjugated to TAT_47–57 carrier peptide, a short positively charged peptide that is used as a carrier for the delivery of the peptide into the cell, using a Gly-Gly spacer. The peptide was synthesized by SPPS (solid phase peptide synthesis) methodology with a fluorenylmethoxycarbonyl (Fmoc)/tert-butyl (tBu) protocol. The final cleavage and side chain deprotection was done manually without microwave energy. Peptide was analyzed by analytical reverse-phase high-pressure liquid chromatography (RP-HPLC) (Shimadzu, MD, USA) and matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (MS) and purified by preparative RP-HPLC (Shimadzu, MD, USA).

Protein Sequence Conservation Analysis

Mammalian protein sequences were obtained from NCBI Orthologs tool for cTnI (TNNI3) and ATP synthase subunit d (ATP5H). Alignment was conducted using NCBI Constraint-Based Multiple Alignment Tool (COBALT) (https://www.ncbi.nlm.nih.gov/tools/cobalt/re_cobalt.cgi) and visualized using the NCBI Multiple Sequence Alignment Viewer to show Frequency-Based Differences in sequences.

Proximity Ligation Assay (PLA)

PLA was performed according to manufacturer protocols using Sigma Duolink^® In Situ Red Starter Kit Mouse/Rabbit (DUO92101)⁷³ with antibodies to cTnI (Adv Immunochemical mouse monoclonal anti-cTnI (4C2cc) 1:50) and ATP synthase subunit d (ProteinTech rabbit polyclonal anti-ATP5H (17589-I-AP) 1:50) in H9c2 cardiac myoblasts. Mitochondria were labeled with Mitotracker Deep Red (pseudocolored Green) and imaged at 60x or 100x.

Rat Heart Infarct Size

Hearts were sliced into 1-mm-thick transverse sections and incubated in 2,3,5-triphenyltetrazolium chloride (TTC) solution (1% in phosphate buffer, pH 7.4) at 37 °C for 15 minutes then fixed in 10% formalin. Tissue slices were photographed, and quantification of infarct size was measured objectively by setting the color threshold in FIJI ImageJ to differentiate infarcted from viable myocardium and calculated as percent infarcted relative to total tissue area.

Recombinant Protein Expression and Purification

cTnI was expressed using BL 21 (DE3) E. coli strain transformed with pET28 plasmid containing full-length human cTnI sequence or C-terminal-truncated cTnI_1–193 cloned downstream of T7 promoter in frame with N-terminal 6xHis and thrombin cleavage site. E. coli cells were grown in LB media supplemented with 50 μg/mL Kanamycin at 37 °C in a shaking incubator (200 rpm) until the optical density (OD₆₀₀) of the culture reached 0.6. Culture was induced with 0.5 mM IPTG and grown for 4 hours, cells harvested by centrifugation, and pellet resuspended in lysis buffer (50 mM Tris pH 8 and 150 mM NaCl) and sonicated. Whole cell lysate was then centrifuged at 20,000xg to separate soluble fraction from cell debris, and clarified lysate was loaded into Ni-NTA Agarose (Qiagen, USA) gravity column pre-equilibrated with lysis buffer. The column was washed with wash buffer (50 mM Tris pH 8, 150 mM NaCl and 40 mM Imidazole). Column bound HT-cTnI was eluted using elution buffer (50 mM Tris pH 8, 150 mM NaCl and 400 mM Imidazole). The eluent was buffer exchanged to storage buffer (50 mM Tris pH 8 and 150 mM NaCl) using Zeba spin desalting column 7K (ThermoFisher Scientific, USA) and flash frozen and stored at −80°C. Protein identity and purity was determined using SDS-PAGE and western blots. Protein concentration was measured using BCA assay (ThermoFisher Scientific, USA).

Seahorse Assay

For experiments with cells, cells were plated in a Seahorse XF24 Cell Culture Microplate (Agilent). Cells were washed twice with Agilent Seahorse XF Media (Agilent) with 1 mM pyruvate, 2 mM L-glutamine, and 2 mM D-glucose. Cells were incubated in 0% CO₂ chamber at 37 °C for 1 hour and placed into Seahorse XFe24 Analyzer (Agilent). For oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) experiments, cells were treated with 1 μM oligomycin, 100 μM 2,4-dinitrophenol (DNP), and 0.5 μM rotenone/antimycin. Three oxygen and pH measurements were taken after each compound was administered. All Seahorse experiments were repeated at least three times. At the end of the Seahorse run, cells were lysed and total protein content was determined using BCA assay (ThermoFischer Scientific, #23225). Raw seahorse values obtained were corrected for differences in protein concentration. For experiments with isolated mitochondria, mouse heart or brain mitochondria (10μg) were loaded on a Seahorse XF24 Microplate with mitochondrial assay solution (MAS: 70mM sucrose, 220mM mannitol, 5mM KH₂PO₄, 5mM MgCl₂, 2mM HEPES, 1mM EGTA, 0.2%BSA fatty acid-free, pH 7.4) plus pyruvate (5mM) and malate (5mM). Mitochondria were sequentially treated with substrate+ADP (2.5mM), oligomycin (2μM), FCCP (4μM), and Antimycin A (4μM).

Western Blot Analysis

BCA assay (Thermo Scientific, #23225) was used for protein quantification. Samples were boiled with Laemmli buffer containing 2-mercaptoethanol, loaded on SDS–PAGE, and transferred to PVDF membrane, 0.45 μm (Bio-Rad). Membranes were probed with the antibody and visualized by ECL (0.225 mM p-coumaric acid; Sigma), 1.25mM 3-aminophthalhydrazide (Luminol; Fluka) in 1 M Tris pH 8.5. Relative densitometry was measured in FIJI ImageJ. The antibodies used in this study are: Anti-VDAC1 Antibody [20B12AF2] (Abcam; ab14734; 1:1000; lot #: GR3296736–19); β-Actin (8H10D10) Mouse mAb Antibody (Cell Signaling Technologies; #3700; 1:1000; lot #: 17); Tom20 (Santa Cruz; sc-136211; 1:500; lot #: JO520); Hsp60 (Santa Cruz; sc-13115; 1:500; lot #: L3015); Cardiac Troponin I (Abcam; ab47003; 1:500; lot #: GR3248433–1); Cardiac Troponin C (Abcam; ab137130; 1:500; lot #: GR106421–9); HRP anti-cTnI mouse mAb (Abcam; ab24460; 1:500; lot #:GR34365881); Mouse IgG HRP linked whole Ab (Sigma; #NA931–1ML, 1:5000; lot #: 17041904), and Rabbit IgG HRP linked whole Ab (Sigma; #NA934–1ML, 1:5000; lot #: 17065614).

Statistical Analysis

Values are presented as mean ± S.D. relative to the average value for the control group unless otherwise stated. In experiments with two groups, group differences were assessed by Student’s unpaired two-tailed t-test. In experiments with more than two groups, ANOVA with Sidak multiple comparison test was used. Statistical analyses conducted with Graph Pad Prism.

Extended Data

Extended Data Figure 1.

HEK-cTnI cells have similar mitochondrial mass and decreased mitochondrial respiration. (a) Citrate synthase activity is unchanged in HEK-cTnI cells relative to HEK-GFP cells and HEK control cells; n=4. (b-c) HEK-cTnI cells have unchanged VDAC1 protein levels relative to HEK-GFP and HEK control cells by immunoblot (n=3) and by flow cytometry (n=3). (d) HEK-GFP cells have similar ATP-linked oxygen consumption to untransduced HEK control cells by Seahorse; n=10. (e) HEK-cTnI cells have similar protein levels of representative subunits of complexes I, II, IV and V of the electron transport chain; n=3. (f) Left: HEK-GFP cells have similar mitochondrial membrane potential to untransduced HEK control cells; n=8. Right: Representative figure of HEK, HEK-GFP and HEK-cTnI TMRM fluorescence by flow cytometry. (g) representative figure of HEK and HEK-cTnI MitoTracker Green fluorescence by flow cytometry.

Extended Data Figure 2.

cTnI associates with ATP synthase complex. (a) cTnI co-purifies with immune-captured ATP synthase complex by ELISA from rat heart mitochondria. Rat brain mitochondria used as “cTnI-free” negative control; n=3. (b) ATP synthase subunit d co-purifies with immunoprecipitated cTnI from rat heart mitochondria exposed to normoxia (Nor) and ischemia/reperfusion (IR). Experiment repeated three independent times. (c) Blue-Native polyacrylamide gel electrophoresis (BN-PAGE) (with 0, 2, 4, & 5 mg digitonin/mg protein from left to right) of mouse heart mitochondria shows co-localization of cTnI and ATP synthase subunit d in the ATP synthase complex. Experiment repeated three independent times.

Extended Data video 1.

Download video file ^{(67.1MB, mp4)}

Data Availability

All data supporting the findings of this study are included in Source Data.