Meta-Analysis of Mechano-Sensitive Ion Channels in Human Hearts: Chamber- and Disease-Preferential mRNA Expression

The cardiac cell mechanical environment changes on a beat-by-beat basis as well as in the course of various cardiac diseases. Cells sense and respond to mechanical cues via specialized mechano-sensors initiating adaptive signaling cascades. With the aim of revealing new candidates underlying mechano-transduction relevant to cardiac diseases, we investigated mechano-sensitive ion channels (MSC) in human hearts for their chamber- and disease-preferential mRNA expression. Based on a meta-analysis of RNA sequencing studies, we compared the mRNA expression levels of MSC in human atrial and ventricular tissue samples from transplant donor hearts (no cardiac disease), and from patients in sinus rhythm (underlying diseases: heart failure, coronary artery disease, heart valve disease) or with atrial fibrillation. Our results suggest that a number of MSC genes are expressed chamber preferentially, e.g., CHRNE in the atria (compared to the ventricles), TRPV4 in the right atrium (compared to the left atrium), CACNA1B and KCNMB1 in the left atrium (compared to the right atrium), as well as KCNK2 and KCNJ2 in ventricles (compared to the atria). Furthermore, 15 MSC genes are differentially expressed in cardiac disease, out of which SCN9A (lower expressed in heart failure compared to donor tissue) and KCNQ5 (lower expressed in atrial fibrillation compared to sinus rhythm) show a more than twofold difference, indicative of possible functional relevance. Thus, we provide an overview of cardiac MSC mRNA expression in the four cardiac chambers from patients with different cardiac diseases. We suggest that the observed differences in MSC mRNA expression may identify candidates involved in altered mechano-transduction in the respective diseases.


Introduction
Cardiac cells are exposed to continually changing mechanical forces leading to stretching, compression, torsion, bending, shearing or a combination of these. Cardiac cells can sense and transduce mechanical cues and modify signaling pathways accordingly. Molecules involved in the process of mechano-sensation and -transduction are integrins, various transmembrane receptors such as G-protein coupled receptors and receptor tyrosine kinases and mechano-sensitive ion channels (MSC) [1].

Results
With the aim of identifying novel MSC genes underlying pathological conditions, we conducted a meta-analysis of five previously published bulk-tissue RNA sequencing datasets. Thus, we included data from non-diseased (no structural heart disease) cardiac tissue samples and a large number of patients with dilated cardiomyopathy (DCM), ischemic cardiomyopathy (ICM), coronary artery disease (CAD), heart valve disease (HVD) or atrial fibrillation (AF). Samples originated from the left ventricle (LV), right ventricle (RV), left atrium (LA) or right atrium (RA). LV samples with DCM and ICM constitute the HF group and samples from RA with CAD and HVD without AF compose the SR group.

Principal Component Analysis
When we conducted pairwise comparisons, whole genome differential gene expression between atria and ventricles explained 53% of the variance between individual nondiseased samples, whereas sex-associated differences in gene expression explained 10% of the variance (Figure 1a). In diseased LV samples, principal component 1 roughly separated samples originating from HF patients from those originating from donor hearts (explaining 22% variance; Figure 1b). In RA samples, samples originating from AF patients were interlaced with those originating from SR ( Figure 1c).

Mechano-Sensitive Ion Channels and Their Cardiac mRNA Expression
The mRNA expression of non-diseased and diseased human tissue from ventricles and atria was analyzed with respect to the MSC and MSC candidate genes listed in Table 1. Cardiac MSC were mainly selected based on [17,18]. Additional candidate MSC were extracted from the gene ontology resource [19,20] (GO:0008381; organism: Homo sapiens) where they have been assigned 'mechano-sensitive' based on phylogenetic analysis [3] and sequence similarity [4]. In total, considering MSC and candidate MSC, 73 genes were considered in this study. including coronary artery and heart valve disease) in right atrial tissue. ♂: male patient/donor; ♀: female patient/donor; each individual point represents data from one tissue sample; the whole genome was considered.

Mechano-Sensitive Ion Channels and Their Cardiac mRNA Expression
The mRNA expression of non-diseased and diseased human tissue from ventricles and atria was analyzed with respect to the MSC and MSC candidate genes listed in Table  1. Cardiac MSC were mainly selected based on [17,18]. Additional candidate MSC were extracted from the gene ontology resource [19,20] (GO:0008381; organism: Homo sapiens) where they have been assigned 'mechano-sensitive' based on phylogenetic analysis [3] and sequence similarity [4]. In total, considering MSC and candidate MSC, 73 genes were considered in this study. including coronary artery and heart valve disease) in right atrial tissue. ♂: male patient/donor; ♀: female patient/donor; each individual point represents data from one tissue sample; the whole genome was considered. Table 1. Human mechano-sensitive ion channels (MSC) and candidates in alphabetical order. Gene and protein names are shown together with references demonstrating their direct or indirect involvement in mechano-sensing. When mechano-sensitivity has been attributed to specific subunits only these were selected-if not, all subunits were considered. MMCs, mechanically modulated ion channels; VACs, volume-activated ion channels; SACs, stretch-activated ion channels; SAC K , K + -selective SACs; SAC NS , cation non-selective SACs; Ca 2+ , calcium; Na + , sodium; K + , potassium; H + , hydrogen; Cl − , chloride. We characterized cardiac mRNA expression of MSC in non-diseased samples of all heart chambers ( Figure 2). The heatmap represents mRNA expression in normalized counts for each tissue sample and for each MSC gene. The dendrograms on both sides of the heatmap represent the similarity between the levels of mRNA expression. We observed a clustering of the samples by tissue provenance (i.e., depending on chamber): in no case were ventricular and atrial samples clustered together. All MSC genes are expressed in the heart, except for ASIC5 and KCNK4, which clustered together because both are not systematically expressed. This cluster showed highest similarity to ASIC2, GRIN1, TRPC5 and KCNA1, which are among the least-expressed MSC genes in the human heart. Furthermore, we observed two clusters of MSC genes with atria-preferential mRNA expression-on the one hand, KCNA5 and CHRNE and, on the other hand, KCNMB2, KCNMB1, KCNQ3, KCNMA1 and SCN9A. In contrast, GJA3, TMC6 and KCNJ2 may cluster for their ventricle-preferential mRNA expression. Finally, CLCN3, CACNA1C and SCN5A clustered together for their strong mRNA expression throughout heart chambers. Taken together, the heatmap illustrates that MSC have a wide spectrum of mRNA expression in the heart. no case were ventricular and atrial samples clustered together. All MSC genes are ex-pressed in the heart, except for ASIC5 and KCNK4, which clustered together because both are not systematically expressed. This cluster showed highest similarity to ASIC2, GRIN1, TRPC5 and KCNA1, which are among the least-expressed MSC genes in the human heart. Furthermore, we observed two clusters of MSC genes with atria-preferential mRNA expression-on the one hand, KCNA5 and CHRNE and, on the other hand, KCNMB2, KCNMB1, KCNQ3, KCNMA1 and SCN9A. In contrast, GJA3, TMC6 and KCNJ2 may cluster for their ventricle-preferential mRNA expression. Finally, CLCN3, CACNA1C and SCN5A clustered together for their strong mRNA expression throughout heart chambers. Taken together, the heatmap illustrates that MSC have a wide spectrum of mRNA expression in the heart.

Chamber-Preferential MSC mRNA Expression
Chamber-preferential MSC mRNA expression was analyzed using non-diseased human cardiac tissue samples. Depending on grouping of samples (i.e., atria vs. ventricles, only LA vs. LV or only RA vs. LA), we found that a total of 21 MSC displayed chamberpreferential gene expression. Of these 21, 17 were higher expressed in the atria than in the ventricles (atria-preferential). 4 of these 21 were higher expressed in the ventricles than in the atria (ventricle-preferential). An overview for all comparisons is provided in Table A1. MSC genes are considered differentially expressed when the adjusted p-value is lower than 0.05 (Benjamini-Hochberg procedure). Figures 3 and 4 show differentially expressed genes having a |log2(fold difference)| ≥ 1, to increase chances to select differential expressions that are most likely to be biologically relevant.
Comparing both-sides atria to both-sides ventricles (atria vs. ventricles; Figure 3a), we found chamber-preferential gene expression for 20 MSC. A total of 16 of these 20 were

Chamber-Preferential MSC mRNA Expression
Chamber-preferential MSC mRNA expression was analyzed using non-diseased human cardiac tissue samples. Depending on grouping of samples (i.e., atria vs. ventricles, only LA vs. LV or only RA vs. LA), we found that a total of 21 MSC displayed chamberpreferential gene expression. Of these 21, 17 were higher expressed in the atria than in the ventricles (atria-preferential). 4 of these 21 were higher expressed in the ventricles than in the atria (ventricle-preferential). An overview for all comparisons is provided in Table A1. MSC genes are considered differentially expressed when the adjusted p-value is lower than 0.05 (Benjamini-Hochberg procedure). Figures 3 and 4 show differentially expressed genes having a |log 2 (fold difference)| ≥ 1, to increase chances to select differential expressions that are most likely to be biologically relevant.
Comparing both-sides atria to both-sides ventricles (atria vs. ventricles; Figure 3a), we found chamber-preferential gene expression for 20 MSC. A total of 16 of these 20 were atria-preferential: CHRNE, KCNA5, ASIC4, KCNQ5, PIEZO2, TRPM3, KCNQ3, SCNN1A, KCNMB2, KCNA1, CACNA1B, ASIC1, TMC5, KCNMB1, FAM155A and TRPV4. 4r of these 20 were ventricle-preferential: KCNK2, KCNJ2, KCNJ8 and TMC6. At the whole genome level, KCNA5 ranged among the top 10 differentially expressed genes, CHRNE among the top 20 and KCNJ2 among the top 110 ( Figure 4a). erential expression with one order of magnitude difference. For the other genes with atriapreferential expression, the order of magnitude difference was < 1 for at least one comparison. None of the MSC genes with ventricle-preferential mRNA expression reached two orders of magnitude difference. KCNJ2 ventricular/LV-preferential expression was the most pronounced with one order of magnitude difference for both comparisons. For the other genes with ventricular-preferential expression, the order of magnitude difference was <1 for at least one comparison (Table 2).     Comparing LA to LV (Figure 3b), we found chamber-preferential gene expression for 18 MSC; 14 of these 18 were atria-preferential. Except for KCNA1, FAM155A and TRPV4, all of the atria-preferential MSC genes observed in Figure 3a showed LA-preferential mRNA expression. Therefore, atria-preferential mRNA expression of KCNA1, FAM155A and TRPV4 could be due, at least in part, to high expression in the RA. One gene, KCNMA1, was expressed higher in LA compared to LV, but comparing both-sides atria to both-sides ventricles, KCNMA1 mRNA expression was not significantly different. A total of 4 of these 18 chamber-preferential MSC genes observed in Figure 3b were ventricle-preferential. All of the ventricle-preferential MSC genes observed in Figure 3a showed LV-preferential mRNA expression. At the whole genome level, KCNA5 was the top differentially expressed gene, KCNJ2 ranged among the top 60 differentially expressed genes and CHRNE among the top 80 ( Figure 4b).
Comparing both atrial sides (RA vs. LA; Figure 3c), we found chamber-preferential gene expression for three MSC. CACNA1B and KCNMB1 were expressed LA preferentially. TRPV4 was expressed RA preferentially. At the whole genome level, KCNMB1 ranged among the top 140 differentially expressed genes (Figure 4c).
CHRNE atria/LA-preferential expression was the most pronounced with two orders of magnitude difference, followed by KCNA5, ASIC4, KCNQ5 and TRPM3 atria/LApreferential expression with one order of magnitude difference. For the other genes with atria-preferential expression, the order of magnitude difference was < 1 for at least one comparison. None of the MSC genes with ventricle-preferential mRNA expression reached two orders of magnitude difference. KCNJ2 ventricular/LV-preferential expression was the most pronounced with one order of magnitude difference for both comparisons. For the other genes with ventricular-preferential expression, the order of magnitude difference was <1 for at least one comparison ( Table 2). Table 2. Summary of chamber-preferential MSC expression in non-diseased human cardiac tissue samples. Arrows indicate the general direction and the orders of magnitude of the difference in expression. Arrows indicate higher or lower expression, no significant difference is indicated by (−): significant differences < 1 order of magnitude ( or ), significant differences of 1 order of magnitude or more (↑ or ↓); alphabetical order. LA, left atrium; LV, left ventricle; RA, right atrium; LA, left atrium.

Gene
Atria vs. Ventricles LA vs. LV RA vs. LA

Disease-Preferential MSC mRNA Expression
To address disease-preferential MSC expression, we compared mRNA levels between diseased and control (non-diseased or SR) human cardiac tissue samples. MSC genes are considered differentially expressed when the adjusted p-value is lower than 0.05 (Benjamini-Hochberg procedure). In Figures 5 and 6, differentially expressed genes satisfying addi-tionally |log 2 (fold difference)| < 1 are included, and highlighted by a lighter shade of red/blue than those with a more than twofold difference in expression levels. We found that 15 MSC have a differential expression depending on the disease considered. Among those, expression of CHRNE, KCNJ4 and TRPC6 was higher and expression of SCN9A, CFTR, ASIC3, LRRC8A, KCNJ11 and TMEM63A was lower in HF samples compared to donor hearts (Figure 5a). SCN9A is the only MSC gene in HF with a |log 2 (fold difference)| ≥ 1. For TRPC6, SCN9A, LRRC8A and ASIC3, HF-related differential expression seemed to be driven largely by DCM (Figure 5b). In AF, expression of PKD1, KCNJ4 and KCNQ4 was higher and expression of KCNQ5, TMC5 and KCNJ5 was lower compared to samples from patients in SR (Figure 5c). KCNQ5 is the only MSC gene in AF with a |log 2 (fold difference)| ≥ 1. Except for KCNJ4, this differential mRNA expression seemed to be dominated by CAD (Figure 5d). One gene, TMEM120A, was expressed higher in AF than in CAD, but showed no significant difference in expression when comparing AF to SR (Figure 5c,d).
fying additionally |log2(fold difference)| < 1 are included, and highlighted by a lighter shade of red/blue than those with a more than twofold difference in expression levels. We found that 15 MSC have a differential expression depending on the disease considered. Among those, expression of CHRNE, KCNJ4 and TRPC6 was higher and expression of SCN9A, CFTR, ASIC3, LRRC8A, KCNJ11 and TMEM63A was lower in HF samples compared to donor hearts (Figure 5a). SCN9A is the only MSC gene in HF with a |log2(fold difference)| ≥ 1. For TRPC6, SCN9A, LRRC8A and ASIC3, HF-related differential expression seemed to be driven largely by DCM (Figure 5b). In AF, expression of PKD1, KCNJ4 and KCNQ4 was higher and expression of KCNQ5, TMC5 and KCNJ5 was lower compared to samples from patients in SR (Figure 5c). KCNQ5 is the only MSC gene in AF with a |log2(fold difference)| ≥ 1. Except for KCNJ4, this differential mRNA expression seemed to be dominated by CAD (Figure 5d). One gene, TMEM120A, was expressed higher in AF than in CAD, but showed no significant difference in expression when comparing AF to SR (Figure 5c,d).
At the whole genome level, no MSC ranged among the top 10 differentially expressed genes. SCN9A was within the top 200 differentially expressed genes when comparing HF vs. donor (Figure 6a). When comparing DCM vs. donor, none of the MSC genes reached a differential expression with a |log2(fold difference)| ≥ 1 (Figure 6b). KCNQ5 ranged among the top 140 and the top 170 when comparing AF vs. SR and AF vs. CAD, respectively (Figure 6c,d). None of these MSC genes with disease-preferential expression exceeded one order of magnitude difference (Table 3).  At the whole genome level, no MSC ranged among the top 10 differentially expressed genes. SCN9A was within the top 200 differentially expressed genes when comparing HF vs. donor (Figure 6a). When comparing DCM vs. donor, none of the MSC genes reached a differential expression with a |log 2 (fold difference)| ≥ 1 (Figure 6b). KCNQ5 ranged among the top 140 and the top 170 when comparing AF vs. SR and AF vs. CAD, respectively (Figure 6c,d). None of these MSC genes with disease-preferential expression exceeded one order of magnitude difference ( Table 3).
To conclude, we found that MSC have a wide range of expression in the heart, that sex was correlated with gene expression, and 21 MSC were expressed chamber-preferentially (among them, 11 MSC displayed ≥ one order of magnitude difference). From diseased tissues, 15 MSC were differentially expressed (among them only SCN9A [lower in HF vs. donor] and KCNQ5 [lower in AF vs. SR or CAD] showed a |log 2 (fold difference)| ≥ 1, and none displayed ≥ 1 order of magnitude difference).

Discussion
In this project we aimed to describe MSC mRNA expression in cardiac physiology and pathophysiology. We sought to characterise MSC mRNA expression in human cardiac health and disease with the aim of obtaining new insights on the involvement of MSC in cardiac pathophysiology. Our main results are (i) expression of acetylcholine receptor nicotinic subunit ε (CHRNE) shows strong preferential expression in the atria (compared to ventricles), (ii) TRPV4 (TRPV4) is expressed RA preferentially (compared to LA), (iii) Ca v 2.2 (CACNA1B) and BK Ca β-1 subunits (KCNMB1) are expressed LA preferentially (compared to RA), (iv) TREK-1 (KCNK2) and K ir 2.1 (KCNJ2) are higher expressed in ventricles compared to atria, (v) Na v 1.7 (SCN9A) is lower expressed in HF (compared to donor) and (vi) K v 7.5 (KCNQ5) is lower expressed in AF (compared to SR or CAD).
Principal component analysis showed that differential gene expression correlates with heart chamber, sex and health status (Figure 1). Our observation that samples originating from AF did not differ clearly from those originating from SR may be explained by the fact that all patients with AF also had CAD or HVD as an indication for cardiac surgery (just as SR patients), but tissue samples were only grouped by atrial rhythm. In addition, all diseased RA samples were obtained from male patients. Thus, no sex-associated differences could be explored. We did not account for age and sex in the various sample groups because we performed a retrospective study. That means that we could not match sample sizes to the different parameters (but prioritized a higher sample size). For the same reason, we could not conduct all theoretically possible pairwise comparisons. For example, for the non-diseased samples we could not compare RA vs. RV and LV vs. RV because these samples groups were not balanced within the previously published reference datasets that we had access to.
The heatmap visualizes the normalized counts per MSC gene across all samples (Figure 2). On the one hand, it allows for an assessment of the amount of mRNA expression across heart chambers. We found that a large number of MSC are expressed in the heart, and most of them are widely present throughout the four chambers. Only ASIC5 and KCNK4 are not systematically expressed in the human heart (they are predominantly expressed in the intestinal tract and in the brain, respectively). On the other hand, the heat map enabled a first evaluation of chamber-preferential mRNA expression as a complementary approach to the analysis presented in Figure 3. Consequently, for KCNA5, CHRNE, KCNMB2 and KCNQ3, atria-preferential and for TMC6 and KCNJ2 ventricle-preferential mRNA expression was identified by both approaches. With this, we confirm the wellknown atria-preferential expression of KCNA5 [70,71]). That said, we should not forget that minimally expressed MSC may still play a major role in cardiac physiology [72]. They simply may be expressed in a small subpopulation of cells, e.g., nerve cells, a subgroup of immune cells. In addition, a few channels with a large conductance, e.g., BK, may have major roles although lowly expressed.
Differentially expressed genes in non-diseased human cardiac tissue derived from different heart chambers are summarized in Table 2 (cf. Figure 3). Overall, 21 MSC genes of the 73 considered were differentially expressed in cardiac chambers. Comparing LA to LV, we confirmed higher KCNA5 expression and lower KCNJ2 expression but we could not confirm higher SCN5A, SCN9A and KCNJ5 expression in LA when compared to LV [73,74]. Gaborit et al. also included the comparison RA vs. RV into their analysis, which showed that KCNA5, KCNJ4 and KCNJ5 are predominantly expressed in RA, and KCNJ2 and KCNJ8 are predominantly expressed in RV [74]. When comparing RA to LV by microarray analysis, Barth et al. found CHRNE, KCNA5, PKD2 and TRPC1 to be higher expressed in RA, and KCNJ2, KCNJ4, KCNJ8, KCNQ4, SCN5A and TRPM3 to be predominantly expressed in LV [75].
Differentially expressed genes in cardiac tissue derived from patients with different health conditions are summarized in Table 3 (cf. Figure 5). Overall, 15 MSC genes of the 73 considered were dysregulated in the context of the cardiac diseases included in this study. We confirmed lower CFTR [76,77] and KCNJ11 [78] expression in HF as well as higher TRPC6 [11] expression in DCM. While we only found lower RA mRNA expression of KCNJ5 in AF compared to SR and CAD, an earlier study showed increased KCNJ5 expression in AF compared to controls [79]. Furthermore, we could not confirm lower ventricular expression of KCNJ8 in HF compared to non-diseased samples [80]. When comparing ICM and DCM, Liu et al. showed that KCNJ5 and KCNMB1 are higher expressed in ICM [12].
Interestingly, the largest differences in MSC mRNA expression were observed when comparing chambers (donor hearts), not when considering the health status of the tissue. Indeed, only SCN9A and KCNQ5 showed a |log 2 (fold difference)| ≥ 1 in disease, compared to 11 genes that showed one-to-two orders of magnitude difference when comparing chambers. Mechanical activity and constraints are very different and probably most conserved in donor hearts between atria and ventricles. Therefore, it is not surprising to have several MSC genes with a marked chamber-preferential mRNA expression. When comparing chambers, CHRNE ranges among the top 20 most differentially expressed genes at the whole genome level. However, very little is known about its role in the heart, making it a very exciting target to investigate further.
When considering MSC mRNA expression in the context of diseases, among the more highly expressed genes this meta-analysis identifies PKD1. Although it is differentially expressed in the context of AF (p-value < 0.05), the biological relevance of this difference |log2(fold difference)| < 1 remains to be considered. This channel is ubiquitous in mammalian cells, involved in various mechano-transduction pathways and is known to contribute to regulate Piezo1 channels [48,81], a channel also ubiquitously expressed and related to AF [82]. Although not much is known about the role of TRPP1 in the heart, it was recently reported to assemble with K v channels to change cardiomyocytes repolarization and contractility [83]. Taken together, this observation suggests that TRPP1 might be an interesting mechano-sensor to investigate further in the context of the heart and particularly for AF.
The top 10 differentially expressed genes associated with a heart chamber or cardiac diseases confirm earlier findings which help to validate our study (Figures 4 and 6). Although mRNA expression of the matrix metallopeptidase 3, MMP3, was strongly lower expressed in AF when comparing to SR, its low average expression (19.3 counts) suggests that it may not be biologically relevant. Of note, in the context of AF, an upregulation of MMP3 was reported previously [84][85][86] which is contradictory to our result.
Especially for ion channels, including MSC, it may not be a change in expression but a change in activity that contributes to cardiac pathophysiology. In addition, it is difficult to correlate gene expression with the number of functional proteins in the plasma membrane. For example, in atrial myocytes, K v 1.5 has been shown to respond to shear stress by increased insertion into the plasma membrane, i.e., recruitment from intracellular compartments [87]. Consequently, the validation of our findings at the protein level will be the subject of future research. Having said that, the necessity of "genomic approaches to identify and investigate genes associated with AF and HF susceptibility" was recently demonstrated by Patel et al. [88].
Whole tissue RNA sequencing does not allow for interpretation about the cell type responsible for differential gene expression and for quantification of heterogeneous responses in individual cell populations. This implies that large differences in expression may exist within one cell type but may not be detected in this study because the whole tissue is considered. This could be particularly significant if the gene of interest is expressed in multiple cell types but differed in only one. Most functional data were acquired in a specific cell type (often cardiomyocytes). The obtained results will be advanced by single-cell or single-nucleus RNA sequencing. Moreover, quantitative polymerase chain reaction and functional experiments like patch-clamp can be used to confirm expression and characterise candidate MSC genes. Furthermore, the genetic background was not consistent within the dataset but could not be accounted for in the analysis, in part, because the information could not be retrieved from all original studies.
This meta-analysis gives an overview of MSC mRNA expression in cardiac health and disease, and highlights MSC chamber-and disease-preferential mRNA expression, suggesting novel potential molecular targets involved in cardiac mechano-transduction.

Studies Considered for the RNA Sequencing Meta-Analysis
We accessed deposited transcriptomic datasets via the National Center for Biotechnology Information sequence read archive database [89]. Studies had to fulfill the following criteria for inclusion into this meta-analysis: organism: homo sapiens; tissue type: cardiac; source: RNA; strategy: RNA sequencing; platform: Illumina; library layout: paired end; file type: FASTQ. The sequencing files of human cardiac tissue samples were then grouped, based on the patients' health status, into non-diseased (no structural heart disease), DCM, ICM, CAD, HVD or AF. Non-diseased atrial and ventricular myocardial samples originated from donor hearts not suitable for transplantation due to size mismatch or logistic reasons. Diseased myocardial samples were obtained from patients undergoing open heart surgery. All patients had given informed consent within the respective studies. Only tissue from patients with sustained (non-paroxysmal) AF was included in the AF group. Of note, all patients with AF also had CAD or HVD as the reason for cardiac surgery but tissue samples were only grouped into AF. Furthermore, the samples had to have clearly identified origins (tissue provenance) from the LV, RV, LA or RA. This meta-analysis included paired (different tissue provenances from the same individual) and non-paired tissue samples. In order to relate our results to published work, especially results of functional experiments, LV samples with DCM and ICM constitute the HF group and samples from RA with CAD and HVD without AF compose the SR group. Considering these inclusion criteria, the analysis of RNA sequencing studies with a total of 108 samples from 62 individuals is summarized in Table 4. Details on sample acquisition, including ethical approvals, and technical procedure can be found in the corresponding original publications. Patient characteristics are summarized in Table 5.

RNA Sequencing Data Analysis
RNA sequencing data analysis was carried out using the Galaxy platform [90], following guidelines from the tutorial "Reference-based RNA-Seq data analysis" [91,92]. In short, data were downloaded and extracted in FASTQ format from the National Center for Biotechnology Information sequence read archive [89]. Thomas et al. [15] deposited 20 sequencing runs/sample to achieve higher coverage. All runs for one sample, i.e., FASTQ files separated into forward and reverse datasets, were merged using Concatenate datasets tail-to-head [93]. Quality control checks on raw sequence data were performed using FastQC [94] and MultiQC [95]. Then, Cutadapt [96] was used to remove adapter sequences. The splice-aware aligner STAR [97] was used to map the RNA sequencing reads onto the human reference genome (hg19). Mapping results were visualized by the integrative genome viewer, IGV [98]. Thereafter, the strandness of the RNA sequencing data (reads mapping to the forward or reverse DNA strand) was estimated using Infer Experiment from RSeQC [99]. Gene expression was measured by featureCounts [100].
From the published reference datasets, we selected and combined the samples into three different groups of pairwise comparisons where the condition samples were balanced in each respective group: AF vs. SR (from RA tissue samples), HF vs. non-diseased (from LV tissue samples) and atria vs. ventricles (from non-diseased tissue samples). For the comparison of AF with SR, we selected 25 samples from the reference datasets of Thomas et al. [15] and Darkow et al. [16]. To compare HF with non-diseased, we selected 28 samples from the reference datasets of Darkow et al. [16], Schiano et al. [13] and Liu et al. [12]. To compare atria with ventricles, we selected 53 samples from the reference datasets of Johnson et al. [14] and Darkow et al. [16].
For each pairwise comparison, we used the DESeq2 package [101] in R [102] to test for differential expression of genes, and employed the gglot2 package [103] to visualize the overall effect of experimental covariates with the plots of principal component analysis. To perform the differential gene expression analysis in DESeq2, we utilized the formula "design~batch + condition" to model the gene expression data according to the experimental design. This formula represents the relationship between the gene expression data and the experimental conditions, while also accounting for potential batch effects. The formula incorporates "design" which specifies how samples are categorized. The "~" symbol signifies the relationship between variables in the R formula syntax. Including "batch" corrects technical variations derived from the different references, while "condition" represents the health status or the tissue provenance. By integrating both "batch" and "condition," the formula "design~batch + condition" allows for accurate identification of differentially expressed genes, mitigating technical artifacts. The resulting "normalized" counts were transformed by the function vst() of DESeq2 and further underwent batch removal with the function removeBatchEffect() of the limma package [104], which served as the input to generate the principal component analysis plots. For checking differential expression, we first ran the function DESeq(), and then applied the function lfcShrink() with the shrinkage estimator ashr [105]. The analysis parameters and the versions of the tools and packages used until this step can be found here: https://dyusuf.github.io/analysis_ collection/Peyronnet-lab/Darkow2023_data_analysis.html.
Differential gene expression levels with adjusted p-values < 0.05 (Benjamini-Hochberg procedure) were regarded as significantly different (statistical significance); the condition |log 2 (fold difference)| ≥ 1 was regarded as indicative functional relevance (biological significance); data representation was performed with OriginPro 2020. DESeq2 output files were re-uploaded to the Galaxy platform and used to generate a heatmap and volcano plots (Wickham, 2009) [103].    Table A1. Summary of chamber-preferential MSC gene expression in non-diseased human cardiac tissue samples. Genes significantly (adjusted p-value < 0.05 and |log2(fold difference)| ≥ 1) higher expressed in the heart chamber indicated in the column than in the heart chamber indicated in the row; grey cells indicate that the corresponding comparisons could not be performed with our dataset; genes are sorted in alphabetical order. LV, left ventricle; RV, right ventricle; LA, left atrium; RA, right atrium; prefer., preferential.