Mitochondrial Collapse Responsible for Chagasic and Post Is-Chemic Heart Failure Is Reversed by Cell Therapy Under Different Transcriptomic Topology

1. Introduction

Mitochondria are critical organelles that regulate cell metabolism and survival, especially in the heart, where mitochondria comprise 30 - 40% of cardiomyocyte volume and play a primarily role for energy production, calcium homeostasis and regulation of cellular apoptosis. Mitochondrial dysfunction leads to impaired cardiomyocyte function, a hallmark of heart failure [1]. Since mitochondria are essential for the cellular energy-demanding cardiac contractile function, a decline in mitochondrial biogenesis and function [2] was associated with the development of ventricular dysfunction caused by myocardial infarction [3], type 2 diabetes [4], Chagas disease [5], or anticancer drug-related cardiotoxicity [6] among several cardiovascular afflictions.

Decades after initial infection with the parasitic euglenoid Trypanosoma cruzi [7], transmitted by the so-called “kissing bug” [8], ~30% of individuals can develop chronic Chagasic cardiomyopathy (CCC), a congestive heart failure and dilated cardiomyopathy [9,10,11]. It is estimated that CCC affects about 7 million of people worldwide, most of them in Latin America [12], and became endemic recently even in the United States [13]. Although the pathogenesis of CCC remains a matter of debate [14,15], the involvement of cardiac mitochondria was first demonstrated by Garg et al. in 2003 [16] and confirmed by Báez et al. in 2011 [17]. They found that a year post-infection parasite persistence and inflammation were associated with structural and functional alterations in mouse cardiac mitochondria in a parasite strain-dependent manner [18]. Common comorbidities include dyslipidemia, hypertension [19] and might lead to cryptogenic stroke [20]. As expected for any infectious disease, CCC triggers the immune response [21] that changes during the development of the disease [22].

Transcriptomic studies on cardiomyocytes and heart left ventricle of CCC mice revealed remodeling of immune response [23], extracellular matrix, cell adhesion [24], intercellular communication via gap junction channels [25,26], mitochondrial oxidative phosphorylation [27], JAK/STAT signaling and cell cycle [28] functional pathways.

The progenitor cells are well recognized for their therapeutic potential in cardiac regeneration [29]. We found that injecting bone marrow-derived stem cells restores most of the heart function and recovers most of the normal heart gene expression profile [30,31].

Myocardial infarction [32], described as the cardiomyocytes’ death caused by insufficient oxygen supply, whose definition and management are still under debate [33,34], is directly related to mitochondrial dysfunction [35] and affects several functional pathways [36,37].

Previous studies on mouse models have shown that myocardial infraction induced by interruption of blood supply [38] activates strong inflammatory response and leads to ventricular remodeling and ischemic heart failure (IHF) [39]. These effects were also reversed by injecting bone marrow-derived mononuclear stem cells into the cardiac scar tissue [40]. Whatever the cause, heart failure has severe consequences on all organs, including triggering various mental disorders [41].

This research investigated whether there are mitochondria-related genomic alterations in failing hearts due to CCC and IHF before and after stem cell treatment. We analyzed the KEGG-constructed oxidative phosphorylation functional pathway [42] as arranged in the mitochondrial module of the KEGG-constructed cardiovascular disease pathway of diabetic cardiomyopathy [43]. The study did not stop at the alteration of the gene expression profile but extended to encompass the entire mitochondrial transcriptomic topology, i.e. to include the control of transcripts’ abundances and the transcriptomic networks, as common in our Genomic Fabric Paradigm [31].

2. Materials and Methods

2.1. Experimental Data

We re-analyzed publicly accessible gene expression data generated in former Iacobas’ lab by profiling the transcriptomes of the left heart ventricle myocardium of mouse models of CCC [44] and IHF [45] before and after the treatment with bone-marrow derived mononuclear cells [46,47]. The treating stem cells were collected from femurs and tibiae of C57Bl/6 adult mice and purified by centrifugation in Ficoll gradient [30]. The Chagas disease was induced by infecting adult C57Bl/6j with trypomastigotes of Colombian Trypanosoma cruzi strain [48], while the permanent myocardial infarcts were obtained by ligating the descending branch of the left coronary artery [39]. The experimental protocols and raw and normalized expression data are fully described and publicly accessible in the Gene Expression Omnibus (GEO) of the (USA) National Center for Biotechnology Information (NCBI) [44,45,46,47]. For each disease, we examined two conditions: IN (infected/infarcted – not treated) and IT (infected/infarcted – treated) and compared with CN (healthy/control – not treated), all in four biological replicates.

2.2. Primary Independent Characteristics of Individual Genes

A transcript was considered as properly quantified if the probing spot had a foreground fluorescence signal more than twice the background and no corrupted or saturated pixels in any microarray. Background subtracted foreground fluorescence signals were averaged for spots probing redundantly the same gene. According to our standard procedure [31], each properly quantified gene is characterized in each condition by three independent characteristics: AVE, REV and COR.

AVE is the average expression level across biological replicas normalized to that of the median gene expression in each sample. This type of normalization minimizes the errors caused by the differential impact of technical noise in the preparation and hybridization of the fluorescent samples.

REV is defined as the mid-interval of the chi-square estimate of the coefficient of variation of the expression levels on all spots probing redundantly the same gene across biological replicas of one condition. REV is used to calculate the REC (relative expression control) and RCS (relative control strength) as:

$$\forall \mathit{c}=\mathit{C}\mathit{N},\mathit{I}\mathit{N},\mathit{I}\mathit{T},{\mathit{R}\mathit{E}\mathit{C}}_{\mathit{i}}^{\left(\mathit{c}\right)}\equiv {\mathit{l}\mathit{o}\mathit{g}}_2\left({\displaystyle \frac{{〈\mathit{R}\mathit{E}\mathit{V}〉}^{\left(\mathit{C}\right)}}{{\mathit{R}\mathit{E}\mathit{V}}_{\mathit{i}}^{\left(\mathit{c}\right)}}}\right),{\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{c}\right)}\equiv {\displaystyle \frac{{〈\mathit{R}\mathit{E}\mathit{V}〉}^{\left(\mathit{C}\right)}}{{\mathit{R}\mathit{E}\mathit{V}}_{\mathit{i}}^{\left(\mathit{c}\right)}}}$$

(1)

where REC is the Relative Expression Control and 〈REV〉^(c) is the median REV of all genes quantified in condition: c = CN, IN, IT. A positive/negative REC value indicates that the expression of that gene is more/less controlled (i.e. more/less stably expressed across biological replicas) than the median in the condition. The control level of a gene indicates how much critical is that gene for the cell survival and phenotypic expression, with the top positive value pointing to the most important mitochondrial gene for that condition.

COR is the Pearson pair-wise correlation coefficient of the (log2) expression levels of two genes. There are three (p < 0.05) statistically significant cases when each of the two genes “i” and “j” in condition “c” are probed by single microarray spots:

1) $1\ge {\mathit{C}\mathit{O}\mathit{R}}_{\mathit{i}\mathit{j}}^{\left(\mathit{c}\right)}\ge 0.951$ synergistically expressed, i.e. their expression levels oscillate in phase across biological replicates.

2) $-1\ge {\mathit{C}\mathit{O}\mathit{R}}_{\mathit{i}\mathit{j}}^{\left(\mathit{c}\right)}\le -0.951$ antagonistically expressed, i.e. their expression levels oscillate in antiphase across biological replicates.

3) $\left|{\mathit{C}\mathit{O}\mathit{R}}_{\mathit{i}\mathit{j}}^{\left(\mathit{c}\right)}\right|<0.05$ independently expressed genes, i.e. expression of one gene has no influence on the expression of the other gene.

COR analysis was used to identify the (p < 0.05) significant transcriptomic networks based on the “Principle of transcriptomic stoichiometry” [49], requiring coordinated expressions of genes whose encoded products are linked in a functional pathway. The overall expression correlation of the mitochondrial genes responsible for adjacent complexes (Cx) of the respiratory chain was measured by their coordination (COORD) percentage:

$$\forall \mathit{c}=\mathit{C}\mathit{N},\mathit{I}\mathit{N},\mathit{I}\mathit{T}\&\forall \mathit{k}=1÷4,\mathit{C}\mathit{O}\mathit{O}\mathit{R}\mathit{D}\left({\mathit{C}\mathit{x}}^{\mathit{k}}\to {\mathit{C}\mathit{x}}^{\mathit{k}+1}\right)\equiv \mathit{S}\mathit{Y}\mathit{N}\left({\mathit{C}\mathit{x}}^{\mathit{k}}\to {\mathit{C}\mathit{x}}^{\mathit{k}+1}\right)+\mathit{A}\mathit{N}\mathit{T}\left({\mathit{C}\mathit{x}}^{\mathit{k}}\to {\mathit{C}\mathit{x}}^{\mathit{k}+1}\right)-\mathit{I}\mathit{N}\mathit{D}\left({\mathit{C}\mathit{x}}^{\mathit{k}}\to {\mathit{C}\mathit{x}}^{\mathit{k}+1}\right)$$

(2)

where: SYN, ANT, IND are the percentages of the p < 0.05 significant synergistically, antagonistically and independently expressed gene pairs out of all pairs that can be formed between the two complexes.

REC and COR values were used to hierarchize the genes according to their Gene Commanding Height (CGH) score [50]:

$$\forall \mathit{c}=\mathit{C}\mathit{N},\mathit{I}\mathit{N},\mathit{I}\mathit{T},{\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{c}\right)}\equiv \mathit{e}\mathit{x}\mathit{p}\left({\mathit{R}\mathit{E}\mathit{C}}_{\mathit{i}}^{\left(\mathit{c}\right)}+〈{2{\mathit{C}\mathit{O}\mathit{R}}_{\mathit{i}\mathit{J}}^{\left(\mathit{c}\right)}}^2〉\right)$$

(3)

2.3. Transcriptomic Changes of Individual Genes and Functional Pathways

Our standard protocol [51] considers the expression of a gene as statistically significantly regulated (here in the left heart ventricle of the diseased mouse w/o cell treatment (d = IN, IT) with respect to the control counterpart CN) if the absolute expression fold-change $\left|{\mathit{x}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}\right|$ exceeds the appropriate cut-off and the p-value of the heteroscedastic t-test of the means equality is less than 0.05. The absolute fold-change cut off (CUT) is computed for each gene by considering both the expression variabilities in the compared conditions and the technical noises of the probing microarray spots in the two sets of biological replicas.

$$\forall \mathit{d}=\mathit{I}\mathit{N},\mathit{I}\mathit{T},{\mathit{x}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}=\left\{\begin{array}{ccc}{\displaystyle \frac{{\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{d}\right)}}{{\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}}& \mathit{i}\mathit{f}& {\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{d}\right)}\ge {\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\\ -{\displaystyle \frac{{\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}{{\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{d}\right)}}}& \mathit{i}\mathit{f}& {\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{d}\right)}<{\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\end{array}\right.$$

(4)

$$\left|{\mathit{x}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}\right|>{\mathit{C}\mathit{U}\mathit{T}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}\equiv 1+{\displaystyle \frac{1}{100}}\sqrt{2\left({{\mathit{R}\mathit{E}\mathit{V}}_{\mathit{i}}^{\left(\mathit{d}\right)}}^2+{{\mathit{R}\mathit{E}\mathit{V}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}^2\right)}\&{\mathit{p}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}<0.05$$

(5)

Contributions of individual genes to the transcriptomic alteration in condition “d” (CCC, IHF) with respect to their healthy counterparts were measured by their WIR (Weighted Individual (gene) Regulation) score:

$$\forall \mathit{d}=\mathit{C}\mathit{C}\mathit{C},\mathit{I}\mathit{H}\mathit{F},{\mathit{W}\mathit{I}\mathit{R}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}\equiv {\mathit{A}\mathit{V}\mathit{E}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\left(\left|{\mathit{x}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}\right|-1\right)\left(1-{\mathit{p}}_{\mathit{i}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}\right)$$

(6)

The overall alteration of a functional pathway Φ in a particular condition (CCC or IHF) was measured by the median WIR of the pathway genes. A positive median WIR indicates that total number of (significantly and not significantly) upregulated genes in the pathway exceeded the number of the downregulated and vice-versa.

The alteration of expression control in CCC and IHF mice w/o cell treatment with respect to the condition CN was measured by the Fold-Change (FC) defined as:

$$\forall \mathit{d}=\mathit{I}\mathit{N},\mathit{I}\mathit{T},{\mathit{F}\mathit{C}}_{\mathit{I}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}=\left\{\begin{array}{ccc}{\displaystyle \frac{{\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{d}\right)}}{{\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}}& \mathit{i}\mathit{f}& {\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{d}\right)}\ge {\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\\ -{\displaystyle \frac{{\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}{{\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{d}\right)}}}& \mathit{i}\mathit{f}& {\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{d}\right)}<{\mathit{R}\mathit{C}\mathit{S}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\end{array}\right.$$

(7)

The fold-change alteration (negative for downregulation) of GCH in a treated (IT) or untreated (IN) Chagas and post-ischemic heart failure is computed as:

$$\forall \mathit{d}=\mathit{I}\mathit{N},\mathit{I}\mathit{T},\mathit{G}\mathit{C}\mathit{H}-{\mathit{F}\mathit{C}}_{\mathit{I}}^{\left({\displaystyle \frac{\mathit{d}}{\mathit{C}\mathit{N}}}\right)}=\left\{\begin{array}{ccc}{\displaystyle \frac{{\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{d}\right)}}{{\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}}& \mathit{i}\mathit{f}& {\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{d}\right)}\ge {\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\\ -{\displaystyle \frac{{\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}}{{\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{d}\right)}}}& \mathit{i}\mathit{f}& {\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{d}\right)}<{\mathit{G}\mathit{C}\mathit{H}}_{\mathit{i}}^{\left(\mathit{C}\mathit{N}\right)}\end{array}\right.$$

(8)

3. Results

3.1. Both Chagasic Disease and Post-Ischemic Heart Failure Are Characterized by Substantial Downregulation of Mitochondrial Genes

Figure 1 presents the mitochondrial genes that were significantly regulated (according to the composite criterion of the absolute fold-change and p-value (5)) in Chagasic disease and post-ischemic heart failure. Thus, in the untreated CCC, four of the five complexes contain only downregulated genes, with Cx 4 making the exception having one upregulated (Cox6a1) although two (Cox5a, Cox6b1) were downregulated. The percentage of down-regulated genes is even larger in the post-ischemic heart despite the upregulation of Ndufa9.

The large numbers of the downregulated genes in the five complexes indicate a significant reduction of the oxidative phosphorylation, and by consequence less production of ATP. Remarkably, in both cardiomyopathies, Ca²⁺ storage in the mitochondrion was diminished by downregulation of one of the involved genes (Vdac1 in CCC and Slc25a5 in IHF mice). The downregulation of these transporters led consistently to Ca²⁺ accumulation in the cardiomyocyte that affected the contractility function of this heart muscle cell.

3.2. Cell Treatment Restores the Normal Expression of Most Mitochondrial Genes Altered in Both Chagasic Disease and Post-Ischemic Heart Failure

As illustrated in Figure 2, we found that the expressions of most mitochondrial genes that have been altered in both Chagasic and ischemic mice were markedly restored following the cell treatment.

Moreover, in the case of CCC mice, the initially not affected Ndufb11 was upregulated by treatment, while the originally downregulated Ndufs4 was even switched to upregulated. Of note is that normal Ca²⁺ storage in the mitochondrion was restored in the treated CCC mice but remained diminished in the treated IHF now because of the downregulation of Vdac2 despite recovering the normal expression of initially downregulated Slc25a5.

3.3. The Largest Mitochondrial Gene Contributors to the Transcriptomic Alterations in Both Chagasic and Post-Ischemic Mice

Figure 1 and Figure 2 considered implicitly significantly regulated genes as equal contributors to the transcriptomic alteration presenting the overall change as percentages of up- and downregulated genes that is an oversimplified description of the biological reality. A better representation of the contribution would be by the pair (expression ratio, p-value) of each affected gene. However, we think that a more comprehensive characterization is by the weighted individual (gene) regulation (WIR), computed with formula (6), that incorporates the total expression change AVE^(CN)(|x|-1) and the statistical confidence (1 – p-val) of this change.

Table 1 presents the largest five contributors (as absolute value of WIR) in Chagasic (CCC) and ischemic (IHF) mice, together with their normal average expression levels (AVE), expression ratios (X) and p-values of their regulation with respect to the normal expression. As expected, all top contributors had negative expression ratios.

The effectiveness of the cell treatment on restoring the expression profiles of the mitochondrial genes is evident by the change of the median WIR from -18.74 in infected not treated to -4.50 after treatment in CCC mice and from -16.83 to -1.81 in IHF mice. These results indicate both substantial reduction of the mitochondrial function caused by the parasitic infection and partial recovery following the cell treatment. While in CCC mice, the non-mitochondrial gene Pln was the largest contributor (WIR = - 222), in the ischemic mice one mitochondrial gene, Cox5a, tops the list with the impressive WIR = - 2603(!).

Interestingly, the median WIR value for the entire transcriptome was positive in all conditions, meaning that overall, the contributions of the up-regulated genes (like those involved in the inflammatory response as previously reported [23,39]) exceeded those of the down-regulated. As expected, the median WIRs of the entire transcriptome were larger for both untreated CCC (0.67 vs. 0.13) and IHF (1.63 vs. 0.70) mice, pointing to an overall reduction of the alteration of the expression profile.

3.4. The Most and the Least Controlled Mitochondrial Genes in CCC and IHF Mice

The experimental design with four biological replicates allowed us to estimate the Relative Expression Control (REC) of individual genes in each condition and how much both treated and untreated cardiomyopathies affected them. Table 2 presents the REC values (computed according to equation 1) of the most (REC > 0) and the least (REC < 0) controlled mitochondrial genes in all profiled conditions (healthy, treated and untreated CCC and IHF mice). We listed also the most and the least controlled genes in the entire transcriptome and the fold-change (negative for downregulation) of diseased states with respect to the healthy mice.

3.5. Both CCC and IHF Alter the Transcriptomic Networks of the Mitochondrial Genes by Partially Decoupling the Oxidative Phosphorylation Complexes

Figure 3 presents the (p < 0.05) significant transcriptomic networks coupling the five complexes of the oxidative phosphorylation in the heart left ventricles of healthy and untreated CCC and IHF mice. One observes that the adjacent complexes are coupled in all conditions mostly by synergistic expression correlations and that the coordination degree (computed with formula (2) is significantly lower in the untreated cardiomyopathies, indicating a partial decoupling of the complexes. The coordinated percentages were computed for the 80 possible gene-pairs between Cx1 and Cx2, 27 between Cx2 and Cx3, 84 between Cx3 and Cx4, and 140 between Cx4 and Cx5.

The reduction was much more considerable for the untreated IHF mice, going to even negative values (COORD^(Cx1-Cx2) = -2.50%, COORD^(Cx3-Cx4) = -1.19%, which means that more significantly independently that synergistically + antagonistically expressed gene pairs were identified. For instance, the number of synergistically + antagonistically expressed gene pairs between complexes Cx1 and Cx2 decreased from 18 (= 18 + 0) in CN to 15 (= 15 + 0) in CCC and to 6 (4 + 2) in IHF, while the number of independently expressed pairs goes from 5 in CN to 4 in CCC and 8 in IHF. In general, there are fewer antagonistically than independently expressed gene pairs, explaining the synchrony of the respiratory chain.

3.6. Stem Cell Treatment Benefits for the Transcriptomic Coupling of the Oxidative Phosphorylation Complexes

Figure 4 presents the (p < 0.05) significant gene expression correlations in treated CCC and IHF mice and the coordination degrees in all five conditions. As summarized in Figure 4c, both cardiomyopathies reduced the expression coordination degrees among all pairs of coupled complexes. However, the stem cell treatment restored partially the coordination degree in the infarcted mice while even substantially increasing it in the CCC mice.

One important observation is that although the cell treatment restored the normal expression levels for most mitochondrial genes, the transcriptomic coupling of the complexes exhibits different than normal patterns. Importantly, the patterns of treated CCC and IHF are also largely distinct to one another.

3.7. Both CCC and IHF Alter the Hierarchy of Mitochondrial Genes

Table 3 presents the most prominent five mitochondrial genes in treated and untreated Chagasic and ischemic mice according to their Gene Commanding Height (GCH) scores computed with formula (3). It contains also the fold-change (GCH-FC, negative for downregulation) of the GCHs in disease states with respect to the healthy condition, computed with formula (8). For comparison, the table presents also the GCH scores of the top five mitochondrial genes in healthy mice and their GCHs in the untreated CCC and IHF mice.

4. Discussion

Mitochondrial respiratory alterations and excessive generation of reactive oxygen species (ROS) are regarded as the main causes of heart failure [52,53], given that mitochondria provide about 90% of the energy used by heart in normal functioning [54,55]. Using the Genomic Fabric Paradigm [51], which provides the most theoretically possible comprehensive transcriptomic characterization, our present study has brought new insights into the current understanding of mitochondrial genes involvement underlying Chagas disease [56] and post-ischemic heart failure [57]. Although several previous studies specifically analyzed alteration of the expression of genes/proteins that regulate mitochondrial oxidative phosphorylation [16,17], none of them studied the alteration and recovery of the gene expression control and transcriptomic networking.

Our transcriptomic characterization of untreated CCC and IHF hearts indicated a unified failure of mitochondrial energetics that were substantially reversed by bone-marrow-derived cell therapy. At the level of Complex 1 [58], the downregulation of 11/32 Nduf genes in untreated CCC and 22/32 in untreated IHF mice likely predict a severe block in NADH oxidation, a surge in reactive-oxygen species, and consequent energy depletion. Complex 1 (NADH ubiquinone oxidoreductase [59]) is responsible in NADH oxidation and proton pumping along the electron transport chain [60]. For instance, the downregulation of Ndufa7 found by us (x = -1.92 in the untreated CCC and x = - 1.72 in untreated IHF) was reported also in cardiac hypertrophy [61]. Although Ndufa10, known for its role in spontaneous hypertension [62,63] was equally downregulated (x = -1.60) in both untreated cardiomyopathies, its downregulation was statistically significant only in IHF mice (p^(IHF) = 0.009, p^(CCC) = 0.142). Interestingly, while Ndufa9 was (even not statistically significant) downregulated in untreated CCC, it was upregulated (x = 4.18) in untreated IHF as happens in high-intensity training [62].

The collapse of 2/3 of the quantified Complex 2 genes, critical for the Krebs cycle [64,65,66,67], in both untreated heart diseases investigated by us, resulted from the significant down-regulation of Sdha (x^(CCC) = -2.49, x^(IHF) = -2.29) and Sdhb (x^(CCC) = -2.07, x^(IHF) = -1.59).

Electron flow stops at Complex 3: the hinge protein Uqcrch shows one of the strongest negative WIR scores (WIR = -44.38), consistent with knockout data in which its loss alone causes contractile failure [68], thereby supporting and broadening those functional observations to two distinct heart-failure models. Complex 3 (CoQ: cytochrome c-oxidoreductase [69]) deficiency is responsible for fewer mitochondrial disorders [70] although carbonylation of genes like Uqcrc2 (found as downregulated in both untreated CCC and IHF) is a direct response to the stress of Trypanosoma cruzi infection [71].

Complex 4 is compromised as suppression of the ATP/ADP sensor Cox4i1 in CCC, which matches marked reduction in state-3 respiration and cytochrome-c-oxidase deficits documented clinically [72]. The downregulation in untreated IHF mice of Cox7b (x = -2.3, WIR = -80) likely undermines Complex 4 assembly, highlighting a novel candidate for future functional probing. Cox5a, the largest contributor listed in Table 1 to the downregulation of Cx4 in IHF (WIR = -2603) is considered a biomarker for the blood stasis syndrome [67], cyanotic heart disease [73] and acute myocardial infarction [74].

Adaptive changes in energy preservation were evident at Complex 5, where the ATP synthase ε-subunit Atp5f1e was down-regulated (x = -2.22) in the ischemic hearts and then normalized after therapy. This not only agrees with, but now quantifies, the prior enzymology-based findings [75,76]. Our results about the downregulation of Cx 5 contradict the report of its upregulation found by another group [77] through meta-analysis of 13 patients with heart failure subjected to heart transplantation compared to 10 healthy counterparts. However, Cpt2 (Carnitine palmitoyltransferase 2), not affected by either CCC and IHF, is down-regulated in treated IHF, that, according to some research [78], might result in rapamycin-resistant, acetylation-independent hypertrophy.

By comparing Figure 1 and Figure 2, one observes that the cell treatment triggered marked recovery of the expression levels of almost all mitochondrial genes in both CCC and IHF mice. The recovery happens in all five oxidative phosphorylation complexes. While the commonly decreased expression of Ndufs1 in infarcted heart [79] is restored to the normal level by the cell treatment in our model, it remains (even less but still) downregulated in the treated CCC mice (from x^(IN) = -3.02 to x^(IT) = -2.20). However, the downregulation of Ndufs4 in T. cruzi infected mice was reversed to upregulation by the cell treatment, confirming the cardioprotection role of its upregulation [80].

A very important part of the study refers to the control of the transcripts’ abundances by the cellular homeostatic mechanisms and its modification during the disease development and following treatment. Among the most controlled genes (Table 2), there were 5 subunits of NADH:ubiquinone oxidoreductase (i.e. mitochondrial complex I), including Ndufa10, Ndufb10, Ndufb5, Ndufb11, and Ndufaf4. These genes can affect mitochondria function profoundly. Especially Ndufa10, the mitochondrial gene with the largest REC (relative expression control) in healthy mice whose RCS (relative control strength) was markedly reduced in both treated (by 32.80x) and untreated (by 22.70x) CCC, meaning a high expression flexibility to adapt to changeable environmental conditions that might eventually be used to alleviate diabetic cardiomyopathy [81]. Ndufa10 has been also implicated with diabetic cardiomyopathy [63] and ischemic injury [72]. Note also the substantial reduction of the expression control in both treated and untreated CCC mice for other stably expressed MITO genes in the healthy mice: Cox7b and Ndufb10. To our best knowledge, this is the first report revealing the significant alteration of the expression control of these mitochondrial gene in Chagasic heart. Ndufb10 observed destabilization agrees with its known role in holoenzyme assembly failure [82].

The high RCS of Tmem186 in the entire normal transcriptome was significantly reduced (RCS-FC = -39.13) in untreated IHF that persisted (RCS-FC = -22.50) after treatment, but the role of this gene is much less understood. Tmem186 is a component of the mitochondrial Complex 1 intermediate assembly [58]. In contrast, IHF increased by 36.56x the control strength of the antigen Cd164 but the cell treatment restores the normal transcription control.

There are several other interesting data in Table 2. For instance, while the RCS of Vdac2, whose downregulation is one of the main responsible factors for heart failure [83], increased by 2.89x in the not-treated CCC mice (IN) with respect to the healthy animals (CN), it decreased in treated mice (IT) by 3.23x. The control of Cox6b1, known to relieving hypoxia/reoxygenation injury [83] increased by 8.31x in the untreated IHF but decreased by 2.48 in treated IHF. The situation is opposite for Ndufaf4 whose expression control decreased in untreated CCC but increased in treated CCC. We found substantial changes in ischemic mice, with some genes exhibiting more expression control (e.g. Atp5f1e, Cox6b1). The control of Atp13a2 decreased in untreated IHF to considerably increase (by 14.82x) in treated IHF.

All these differences indicate that the homeostatic control mechanisms to keep the expression levels within certain intervals are neither uniform among genes nor similarly altered by each of the two cardiomyopathies or restored following the same therapeutic approach. A very important observation comparing data from Table 2 with Figure 1 and Figure 2 is that, even though the cell treatment restored the normal expression of several genes, their expression control was not. For instance, in the ischemic mice, the expression of the significantly down-regulated Cox6b1 (x^(IN) = -2.87) was restored (x^(IT) = 1.1) by the treatment but its RCS was decreased by 2.48x, while the recovered down-regulation of Atp5f1e (x^(IN) = -2.22) was accompanied by an increase of the RCS by 2.77x). This observation indicates that whereas the treatment recovered the expression profile of most mitochondrial genes, it did not restore the initial control of the transcripts’ abundances.

Recent work in acute experimental models of Chagas disease showed that an imbalance between ROS and NO creates a strongly proarrhythmic substrate independent of major structural remodeling and directly links mitochondrial redox output to electrical instability [2,84]. The same study demonstrates that NO inhibition reverses early after-depolarizations and action potential alternans, underscoring redox–calcium crosstalk as a therapeutic lever. Our findings of Complexes 1-5 suppression and Vdac1 dysregulation align with this mechanism and suggest that transcriptomic loss of the ETC (electron transfer chain) control may be an upstream driver of the ROS/NO dis-equilibrium [85]. It was proved that excessive ROS generation is associated with low oxygen levels as in acute hypoxia [86] or in the ischemic heart.

Additional analysis highlighted that mitochondrial injury is shaped by both inflammatory environment and genotype [87]. This supports our observation that nuclear encoded ETC subunits (Ndufb11, Cox7b) are vulnerable nodes that could synergize with inflammatory signaling to depress ATP output. It also strengthens the rationale to stratify future cohorts by mitochondrial genotype and inflammatory profile when validating control metrics [81,88]. There was evidence for adipose tissue as a chronic reservoir for Trypanosoma cruzi and a source of adipokines-associated mitochondrial stress, implying that systemic metabolic inputs can burden cardiac bioenergetics [85]. This concept aligns with our Cpt2 signal finding that reduced fatty acid oxidation capacity in the heart may be compounded by parasite driven adipose dysfunction and argues for integrating peripheral metabolic tissues into future mechanistic and therapeutic designs [89].

Figure 3 presents for the first time to our knowledge the transcriptomic networks that couple the respiratory chain complexes. It shows also how infection with Trypanosoma cruzi and ischemia alter the transcriptomic coupling, indicating profound remodeling of the oxidative phosphorylation functional pathway. The most dramatic alteration occurred between Cx1 and Cx2 in the IHF mice, where the coordination degree decreased from +16.25% to -2.50%, meaning a significant decoupling of the paired genes between the two complexes. The marked coupling decrease between the adjacent complexes is an additional explanation of the mitochondrial failure in the two investigated cardiomyopathies.

Understanding the importance of some of these expression correlations for the normal heart function and what pathological consequences bring their alterations might open a very interesting and challenging field of research. For instance, Chagas disease turned the significant synergism of the downregulated Nfdufa6 and Sdha, whose simultaneous upregulation is an indication of acute myeloid leukemia [90], into statistically significant independence. We found that the pairs Uqcrb, Coxcb, and Atp5pf, Cox5b, reported as simultaneously upregulated in paradoxical low transvalvular flow and low gradient patients who develop advanced heart failure symptoms [91], are synergistically expressed in normal conditions. The synergism of the downregulated by ischemia Sdhb and Uqcrffs1 supports the efforts to improve the diagnosis and gene targeted therapies of Alzheimer’s disease [92].

As illustrated in Figure 4c, cell therapy recovered part of the decrease of the four coordination degrees in the IHF mice and even increased them over the original levels in the CCC mice. In the CCC mice the cell therapy not only restored the normal expressions of the down regulated Ndufv1 and Sdhb (downregulated also in diabetic patients caused by high fat diet [93]) but also made significant their synergistic expression. To prove this assertion, one can use an online available calculator [94] to determine the p-values of the statistical significance of the expression correlations in each the three conditions: COR^(CN) = 0.832 (p-value = 0.168), COR^(IN) = 0.397 (p-value = 0.603), COR^(IT) = 0.996 (p-value = 0.004). Interestingly, Ndufa13 and Sdhd, that are independently expressed in healthy mice housed in normal atmospheric conditions, maintained their independence in untreated IHF but became synergistically expressed in both treated CCC and IHF mice. The treatment turned the practically neutral correlations: COR^(CN) = 0.320 (p-val = 0.680), COR^(IN) = 0.222 (0.778) between the downregulated Cox17 and Atp5mc1 in ischemic mice into a significant antagonistic expression: COR^(IT) = -0.952 (p-val = 0.048). However, together with NDUFB1, COX17 and ATP5MC1, were reported as upregulated in Parkinson’s, Alzheimer’s and Huntington’s diseases [95].

Nevertheless, we have to recognize that the correlation analysis cannot determine what gene of the pair is the master, i.e. its expression level commands the expression of the other, nor about the energy coupling [96] and the electric charges transfer within or between the respiratory complexes [97,98,99].

The Prominence analysis ranked the mitochondrial genes according their importance for the functioning the cell powerhouse. From Table 3 one may learn that:

i) gene hierarchy is altered in cardiomyopathy as revealed by the differences between the top five genes in healthy mice compared to the sets of five in the other heart conditions.

ii) each type of cardiomyopathy induces distinct alteration of the gene hierarchy (see GCH differences between CCC and IHF for the most prominent genes in healthy mice.

iii) restoration of the normal expression level is not accompanied by the reinstate of the gene in the right hierarchy, indicated by the non-unit GCH-FC. For instance, Uqcrh, found by us as down-regulated in both untreated CCC and IHF mice, was reported as upregulated in patients with hypertrophic cardiomyopathy [100]. Cell treatment restored the normal expression in the CCC but not in the IHF mice. However, the restored expression level is not followed by the full restoration of its GCH after the treatment. Instead, its RCS became 6.67x stronger than in the normal condition (even stronger than in the untreated CCC, where it was 4.66x).

iv) cell treatment has different effects on the two types of heart afflictions. For instance, the somehow low GCH of Cox6b1 healthy mice that was considerably raised in both untreated cardiomyopathies (GCH-FC^(IN-CCC) = 4.81, GCH-FC^(IN-IHF) = 5.53), is raised even more in treated CCC (GCH-FC) = 7.34) but downgraded in treated IHF (GCH-FC = -2.10).

In previous publications, we have shown that the most prominent gene of a condition might be the most effective target for the gene therapy of that condition [50,101]. Therefore, this analysis may open new therapeutic avenues for both ischemic failure and Chagas disease.

5. Conclusions

Taken together, the findings of this study establish an exhaustive profile of mitochondrial collapse in Chagas disease and post ischemic heart failure. The investigation demonstrated that targeted cell therapy can partially restore the functional integrity of key ETC components. It further confirmed several established mechanisms, and introduce new signals. Their broad rescue with cell therapy underscores mitochondria as a convergent, tractable target for cardiac repair.

Interestingly, although the cell treatment recovers the normal expression of the most of the regulated genes, it does not restore the normal expression control or the inter-complexes transcriptomic coupling. Moreover, both treated cardiomyopathies present different expression control and correlation patterns to each-other. Since the treated animals displayed almost normal electrophysiological parameters and excitatory and contractility properties of the ventricular myocardium, our results suggest that a physiological state is compatible with several transcriptomic topologies. This conclusion points to the need of defining a kind of “transcriptomic entropy of a state”.