RNA-sequencing reveals transcriptional signature of pathological remodeling in the diaphragm of rats after myocardial infarction

Abstract

The diaphragm is the main inspiratory muscle, and the chronic phase post-myocardial infarction (MI) is characterized by diaphragm morphological, contractile, and metabolic abnormalities. However, the mechanisms of diaphragm weakness are not fully understood. In the current study, we aimed to identify the transcriptome changes associated with diaphragm abnormalities in the chronic stage MI. We ligated the left coronary artery to cause MI in rats and performed RNA-sequencing (RNA-Seq) in diaphragm samples 16 weeks post-surgery. The sham group underwent thoracotomy and pericardiotomy but no artery ligation. We identified 112 differentially expressed genes (DEGs) out of a total of 9664 genes. Myocardial infarction upregulated and downregulated 42 and 70 genes, respectively. Analysis of DEGs in the framework of skeletal muscle-specific biological networks suggest remodeling in the neuromuscular junction, extracellular matrix, sarcomere, cytoskeleton, and changes in metabolism and iron homeostasis. Overall, the data are consistent with pathological remodeling of the diaphragm and reveal potential biological targets to prevent diaphragm weakness in the chronic stage MI.

1. Introduction

In the United States of America, there is an incidence of 600,000 to 800,000 myocardial infarctions (MI) every year (Benjamin et al., 2019). The progression post-MI is accompanied by systemic inflammation, neural adaptations, and hormonal compensations that affect various organs and cell types. After an MI, patients experience shortness of breath and exercise intolerance. These symptoms have a multifactorial nature, which includes decreased maximal inspiratory pressure and endurance (Neves et al., 2012, 2014). Loss of maximal inspiratory pressure that occurs post-MI, once heart failure ensues, exacerbates the disease pathophysiology and morbidity (Kelley and Ferreira, 2017).

The diaphragm is the main inspiratory muscle, and studies in rodents have shown that the chronic phase post-MI (>6 weeks in rodents) is characterized by diaphragm fiber atrophy, contractile dysfunction (loss of force per cross-sectional area), increased fatigability, and abnormal vascular control (van Hees et al., 2007; Empinado et al., 2014; Adams et al., 2019; Laitano et al., 2016; Stassijns et al., 1998). The pathophysiology of diaphragm abnormalities post-MI or heart failure is complex because inflammatory and neurohumoral responses are accompanied by increased work of breathing (Cross et al., 2011; Mancini et al., 1996), and the diaphragm fails to show typical training-like responses (Mangner et al., 2015). The mechanisms of diaphragm abnormalities post-MI include activation of sphingomyelinase and sphingolipid signaling (Empinado et al., 2014; Coblentz et al., 2019), increases in reactive oxygen species and protein oxidation (Ahn et al., 2015; Laitano et al., 2016; Supinski and Callahan, 2005), and activation of the ubiquitin–proteasome system (van Hees et al., 2008a; Adams et al., 2019). Moreover, a recent study showed DNA damage in the diaphragm in the chronic stage post-MI (Jaenisch et al., 2018). However, a comprehensive understanding of transcriptome changes associated with functional and structural diaphragm abnormalities induced by MI is still missing.

In the current study, we used RNA-sequencing (RNA-Seq) to determine the MI-related transcriptional changes and understand the genetic reprogramming of the diaphragm in the chronic phase post-MI. Our goal was to gain insights into mechanisms of dysfunction and identify potential biological targets to prevent or attenuate diaphragm abnormalities post-MI.

2. Materials and methods

2.1. Sample collection

2.1.1. Animals

We studied samples from 8 male adult Sprague-Dawley rats (Rattus norvegicus, Charles River Laboratories) that were a subset of Sham and severe heart failure animals from our recent study (Kelley et al., 2020). We originally performed surgeries in n = 10 (Sham) and n = 11 (MI), and mortality post-MI was 54%. We only had access to n = 4 samples from MI and selected a matched number of samples from the sham group for this study. The animals were maintained by University of Florida animal care services personnel, individually caged, and exposed to standard dark-light cycles with free access to food and water throughout the study. The Institutional Animal Care and Use Committee of the University of Florida approved all procedures performed in our study.

2.1.2. Surgical preparation and coronary artery ligation

We caused MI via ligation of the left anterior descending coronary artery (Pfeffer et al., 1979; Laitano et al., 2016). We shaved the left side of the thorax and cleaned the surgical area with 4% chlorhexidine and sterile saline. Thereafter, we performed orotracheal intubation and connected the animal to a mechanical ventilator (Model 683; Harvard Apparatus). While the animal was in the surgical plane of anesthesia (2% isoflurane), we exposed the heart via a left-sided thoracotomy by blunt dissection of the intercostal muscle (fourth or fifth intercostal space) while maintaining the ribs intact, removed the pericardium with attention to avoid phrenic nerve damage, and ligated the left anterior descending coronary artery near the left atrium using a 6–0 PGA suture (Demesorb; Demetech). After the ligation, we hyperinflated the lungs, approximated the ribs using a 6–0 PGA suture and closed the skin incision with 3–0 suture (Demelon; Demetech). Once extubated, the animals were transferred to a heated pad for recovery. Sham surgeries were similar to the MI procedure and included thoracotomy and pericardiectomy, except that we skipped the ligation of the coronary artery. The MI and Sham procedures were completed aseptically with minimal penetration of the thoracic cavity and manipulation of heart and lungs to avoid damage to major vessels, esophagus, and the phrenic nerve. During surgery, post-extubation, and tissue harvesting (described below) all animals showed signs of diaphragm contractions suggesting that the phrenic nerve remained intact during surgery. The animals received topical bupivacaine injection immediately after the skin was closed and subcutaneous buprenorphine for 3 days post-surgery. We collected tissue 16 weeks post-surgeries, as studies in rats show diaphragm abnormalities at this time (Kelley et al., 2020; van Hees et al., 2007, 2008b; Empinado et al., 2014; Laitano et al., 2016).

2.1.3. Tissue harvesting and infarct size

We collected the tissues (diaphragm and heart) around the same time of day (0800 to 1000 hr), with the animals in the surgical plane of anesthesia. Portions of the costal diaphragm were snap-frozen in liquid nitrogen, stored in vapor-phase liquid nitrogen, and later processed for RNA isolation. We further dissected the right (RV) and left ventricles (LV) for measurements of weight and infarct area. To determine the infarct area, we cut the interventricular septum from the base to the apex of the LV, pinned the infarct and non-infarcted areas of the LV + septum, and acquired a digital photograph. The transmural infarct area was determined by planimetry (Finsen et al., 2005; Ferreira et al., 2006). Based on previous studies in rodents (Finsen et al., 2005; van Hees et al., 2007, 2008b; Laitano et al., 2016) and the more pronounced diaphragm weakness in patients with severe heart failure with reduced ejection fraction (HFrEF) (Filusch et al., 2011), we studied only MI rats with infarct area > 35% of LV + septum. Thus, MI animals in this study constitute a group with cardiac morphological and functional signs of severe heart failure that show diaphragm contractile dysfunction and atrophy (Kelley et al., 2020). The animals did not receive any treatment for heart failure (e.g., β-blockers, angiotensin-converting enzyme inhibitors, or angiotensin-1 receptor blockers).

2.2. RNA isolation, library preparation and sequencing

2.2.1. RNA isolation

Frozen rat diaphragm muscle tissues were homogenized in a bead mill homogenizer (Bullet Blender, Next Advance, USA) using RNAse-free stainless steel beads (Next Advance, USA). Total RNA was isolated using TRI Reagent (Sigma, USA), according to the manufacturer’s specifications. Total RNA was solubilized in nuclease free-water and treated to eliminate genomic DNA carryover with DNA-free kit (Life Technologies, USA) according to the manufacturer’s specifications. The RNA concentration was determined with a Nanodrop 2000 (Thermo Fisher Scientific, USA). Acceptable absorbance ratio of 260 nm/280 nm and 260 nm/230 nm were > 1.8 and ≥ 2, respectively. RNA integrity (RIN) was determined with the RNA 6000 Nano Chip on the 2100 Bioanalyzer automated electrophoresis system (Agilent Technologies Inc., Santa Clara, CA, USA). Samples with RIN ≥ 8 were used for sequencing (Supplementary Fig. S1A).

2.2.2. RNA-Seq library preparation

RNA-Seq libraries (4 RNA-Seq libraries per each study group) were generated using the SMARTer^® Ultra-Low input RNA Kit (Clontech, USA) and KAPA Hyper Library Preparation Kit Illumina platforms (Kapa Biosystems, USA) following the manufacturers recommended protocols. Briefly, PCR tubes containing 200 ng total RNA in 3.5 μl of reaction buffer were thawed on ice. First-strand cDNA synthesis was initiated by adding 1 μl of 3 SMART CDS Primer II A (12 μM) at 72°C for 3 min, and then 5.5 μl master mix containing SMARTScribe^™ Reverse Transcriptase (100 U) was added and the reaction was incubated at 42°C for 90 min, followed by inactivation at 70°C for 10 min. First-strand cDNA was isolated using SPRI AMPure Beads and double-stranded cDNA was generated by long-distance PCR. Amplified cDNA (acDNA) was purified using SPRI AMPure beads and the final product was assessed for its quantity and size distribution using Agilent High Sensitivity D5000 ScreenTape Kit (Agilent Technologies, USA) (Supplementary Fig. S1B). The double-stranded cDNA fragments were sheared by Covaris fragmentation, then end-repaired with a combination of T4 DNA polymerase, Klenow fragment polymerase, and polynucleotide kinase to ensure blunted ends, and then adenylated with a single A-base at the 3′ -end of the fragment by the Klenow 3′−5′ exoenzyme accordingly. The tails were ligated with Illumina adapters by T4 DNA ligase. The adapter-ligated samples were amplified in 13 cycles of PCR using primers that only anneal to adapter-ligated fragments. The final RNA-Seq libraries were assessed for its size distribution using Bioanalyzer DNA 1000 Kit (Agilent Technologies. USA) (Supplementary Fig. S1C) and for its concentration using Qubit 4 fluorimeter (Thermo Fisher Scientific, USA).The adapter sequences of the pooled RNA-Seq libraries were annealed to the primers on the Illumina flow cell during bridge PCR in the Illumina NextSeq 500 sequencer (NextSeq 500 sequencer, Illumina; Next-Generation Sequencing Core, ICBR, University of Florida), which generated the clusters necessary to view fluorescence during the 76 cycles sequencing-by-synthesis process.

2.2.3. RNA-Seq analysis

Read quality assessment, sample demultiplexing, and adapter trimming were performed using Illumina BaseSpace software applications. Demultiplexed and quality filtered RNA-Seq reads were aligned to rat reference assembly (Rattus norvegicus UCSC rn5 (RefSeq gene annotation)), using TopHat App (TopHat Alignment | Version: 1.0.0; BaseSpace, Illumina). The normalized expression level of the transcriptome of each individual sample and differential gene expression analysis were generated using Cufflinks App (BaseSpace^® Cufflinks Assembly & DE v2.1.0 App; BaseSpace, Illumina). A False Discovery Rate (FDR)- adjusted P < 0.05 was used to determine significant differences. The FDR correction guarantees that the proportion of false positives is <5% of the total number of positive tests (Benjamini and Hochberg, 1995; Benjamini and Yekutieli, 2001). Volcano plot of statistically significant differentially expressed genes (DEGs) at P < 0.05, functional annotation of differentially expressed genes for Sham and MI groups, significantly impacted pathways, and gene ontology (GO) analyses were obtained using iPathwayGuide App (https://www.advaitabio.com/ipathwayguide, Advaita Bioinformatics, USA) in the context of the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Release 84.0+/10–26, Oct 17) (Kanehisa, 2002), and the Gene Ontology Consortium database. Pathway analysis was performed on DEGs, and FDR-corrected P < 0.05 was used to determine significant differences. The Elim pruning method, which iteratively removes the genes mapped to a significant GO term from more general (higher level) GO terms, was used in GO analysis to overcome the limitation of errors introduced by considering genes multiple times. Elim-corrected P < 0.05 was used to determine significant differences. Hierarchical clustering heat map and exon-level expression analysis were performed using DNASTAR App (DNASTAR, USA). The data have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=) and are accessible through GEO Series accession number GSE141859.

2.2.4. Analysis of differential gene expression

We also performed an analysis of gene expression signatures of biological networks specifically identified in skeletal muscle (Wang et al., 2012; Smith et al., 2013). These networks include neuromuscular junction, excitation–contraction coupling, extracellular matrix, sarcomere contraction, cytoskeletal elements, energy metabolism, inflammation, muscle atrophy and hypertrophy (skeletal muscle size), and myofibrillar protein isoforms (muscle fiber types). We also analyzed the expression of genes involved in oxidative stress response, which is an important component of diaphragm abnormalities post-MI (Laitano et al., 2016; Ahn et al., 2015), and metal ion homeostasis. The use of skeletal muscle-specific biological networks along with GO and pathway analysis facilitates the identification of genes and biological networks critical to skeletal muscle function.

3. Results

3.1. Animal characteristics

Animals weighed 255 ± 21 g (Sham) and 244 ± 21 g (MI) at the beginning of the study and 652 ± 37 g (Sham) and 619 ± 31 g (MI) at the time of terminal experiments. Rats in the MI group had an infarct area of 41 ± 0.5% and right ventricle (RV) hypertrophy, evident by increased RV weight-to-tibial length (in mg/mm; Sham 7.2 ± 0.91 and MI 11.0 ± 3.11). These are consistent with a large MI and severe heart failure (Pfeffer et al., 1979; Nahrendorf et al., 2000).

3.2. RNA-Seq quality control

After quality control and data filtering, on average, 33,972,308 raw reads per sample were obtained (Table 1). For each sample, 88.49% to 91.70% reads were uniquely mapped to the rat reference genome, indicating sufficient coverage. Median CV coverage uniformity, the median coefficient of variation of coverage of the 1000 most highly expressed transcripts (ideal value = 0), as reported by the CollectRnaSeqMetrics utility from Picard tool, was 0.45 ± 0.01. Distribution of mapped reads over DNA regions showed 48% of reads mapped to the coding region, 20% of reads mapped to untranslated region (UTR), 12% of reads mapped to intron region, and 20% reads mapped to intergenic region (Supplementary Fig. S2). Transcript coverage distribution, showing a visual representation of coverage uniformity across transcripts, is displayed in Supplementary Fig. S3. The resulting Sham and MI transcriptomes, created from generated transcriptomes of each individual sample in each group and merged transcriptomes within the same group, were used for differential gene expression analysis (Supplementary Table S1). Volcano plot of statistically significant DEGs identified from the RNA-Seq libraries of Sham and MI diaphragm muscle is shown in Fig. 1A. Hierarchically clustered heat map of DEGs between MI and Sham is shown in Fig. 1B. A total of 112 DEGs were identified out of a total of 9664 genes with measured expression in MI and Sham groups (Table 2). Among DEGs, 42 were upregulated and 70 were downregulated in the MI group.

Table 1.

Statistics of the sequence and alignment quality for the diaphragm Sham and MI rat.

Sample ID	Reads	Number of Reads	% Total Aligned	% Abundant	% Unaligned	Median CV Coverage Uniformity	% Stranded
Sham-1	1×76	32,657,230	90.93%	25.44%	9.07%	0.46	49.76%
Sham-2	1×76	37,438,152	89.38%	21.48%	10.62%	0.44	49.93%
Sham-3	1×76	58,551,865	91.70%	25.02%	8.30%	0.46	49.86%
Sham-4	1×76	18,828,056	88.49%	22.89%	11.51%	0.44	49.84%
MI-1	1×76	36,206,005	91.28%	25.67%	8.72%	0.46	49.88%
MI-2	1×76	38,047,491	91.70%	26.09%	8.30%	0.46	49.95%
MI-3	1×76	28,926,169	91.58%	25.86%	8.42%	0.45	49.86%
MI-4	1×76	21,123,498	91.57%	27.51%	8.43%	0.48	49.71%

Fig. 1.

Visualization of differentially expressed genes detected in MI rat diaphragm tissues. A) Volcano plot: All significantly differentially expressed genes are represented in terms of their measured expression change (X-axis represents log2 expression change) and the significance of the change (Y-axis). The significance is represented in terms of the negative log (base 10) of the p-value so that more significant genes are plotted higher on the y-axis. The dotted lines represent the thresholds of expression change significance used to select the DE genes; B) Hierarchically clustered heat map of log2 fold change of diaphragm genes significantly altered by myocardial infarction (MI). Rows are differentially expressed genes following RNA sequencing. Fold change was calculated as log2 (MI) - log2 (Sham). Color intensity shows the relative value for each gene.

Table 2.

List of 112 DEGs in diaphragm of rats after myocardial infarction.

Gene ID	Gene name (abbreviation)	log2 (FC)	Gene Description
29,275	Actc1	0.65	Actin, alpha, cardiac muscle 1
246,253	Adipoq	0.57	adiponectin, C1Q and collagen domain containing
24,180	Agtr1a	−0.56	angiotensin II receptor, type 1a
25,748	Alas2	1.07	5′-aminolevulinate synthase 2
25,239	Apod	1.45	apolipoprotein D
29,657	Arntl	−0.72	aryl hydrocarbon receptor nuclear translocator-like
306,344	Arrdc2	−0.97	arrestin domain containing 2
25,389	Atf3	−2.72	activating transcription factor 3; expression is associated with neuronal injury
24,214	Atp1b2	−0.58	ATPase Na+/K + transporting subunit beta 2; plays a role in Na + and K + transport
24,888	Bcl2l1	−0.53	Bcl2-like 1; inhibits neuronal apoptosis
24,225	Bdnf	0.66	brain-derived neurotrophic factor
29,619	Btg2	−0.61	BTG family, member 2, an anti-proliferative protein
365,603	Bves	−0.63	blood vessel epicardial substance
307,100	Calml3	2.09	calmodulin-like 3
361,431	Cbfa2t3	−0.6	CBFA2/RUNX1 translocation partner 3
307,199	Cd226	1.76	CD226 molecule
83,681	Cish	0.93	cytokine inducible SH2-containing protein
83,620	Cit	−1.2	citron rho-interacting serine/threonine kinase
84,588	Cldn11	−1.14	claudin 11; component of the inter- Sertoli cell tight junctions in the testis
500,336	Clec1b	3.09	C-type lectin domain family 1, member B
83,842	Crot	−0.59	carnitine O-octanoyltransferase
498,335	Cxcl13	−4.12	C-X-C motif chemokine ligand 13;
266,761	Cyp4v3	−0.58	cytochrome P450, family 4, subfamily v, polypeptide 3
29,167	Dctn1	−0.5	dynactin subunit 1; component of dynein microtubule activated ATPase
140,942	Ddit4	−1.05	DNA-damage-inducible transcript 4
297,989	Ddx58	1.12	DEXD/H-box helicase 58
305,448	Dok7	−0.82	docking protein 7
94,204	Ece1	−0.67	endothelin converting enzyme 1
25,043	Eln	−0.72	elastin
85,496	Enpp1	0.62	ectonucleotide pyrophosphatase/phosphodiesterase 1
84,050	Enpp2	−0.68	ectonucleotide pyrophosphatase/phosphodiesterase 2
65,030	Ephx2	−0.51	epoxide hydrolase 2
54,319	Ezr	−0.61	ezrin; connects the microvillar cytoskeleton to the plasma membrane
29,636	F2rl2	1.5	coagulation factor II (thrombin) receptor-like 2;
310,540	Fam198b	0.56	protein FAM198B
83,517	Fcn1	−0.93	ficolin A; mouse homolog is a plasma protein that binds elastin and GlcNAc
24,367	Fgg	1.04	fibrinogen gamma chain
360,457	Figf	−1.95	c-fos induced growth factor
361,810	Fkbp5	−0.99	FK506 binding protein 5
114,110	Flt4	−1.27	fms-related tyrosine kinase 4
246,245	Fmo2	−0.58	flavin containing monooxygenase 2;
362,106	Fubp3	−1.06	far upstream element binding protein 3
361,288	Fundc2	0.72	FUN14 domain containing 2
363,091	Gcom1	−0.9	myocardial zonula adherens protein
25,236	Gpc3	−0.7	glypican 3; heparan sulfate cell surface proteogycan
494,500	Gsta3	0.6	glutathione S-transferase alpha 3
305,882	Haus4	−0.96	HAUS augmin-like complex, subunit 4
25,632	Hba1	1.09	hemoglobin, alpha 1;
360,504	Hba2	1.09	hemoglobin, alpha 2
24,440	Hbb	1.04	hemoglobin subunit beta; beta-globin chain of hemoglobin
361,619	Hbb-b1	1.07	hemoglobin, beta adult major chain
303,163	Igtp	−0.76	interferon gamma induced GTPase
171,060	Il13ra2	−2.53	interleukin 13 receptor subunit alpha 2
24,499	Il6r	−0.74	interleukin 6 receptor; alpha subunit of the interleukin 6 receptor complex
315,962	Ky	0.57	kyphoscoliosis peptidase
689,064	LOC689064	0.89	beta-globin
29,469	Lbp	0.96	lipopolysaccharide binding protein
315,714	Loxl1	−1	lysyl oxidase-like 1
81,682	Lum	0.8	lumican; member of leucine-rich proteoglycan family
293,186	Lyve1	−0.75	lymphatic vessel endothelial hyaluronan receptor 1
246,760	Mafk	−0.55	v-maf avian musculoaponeurotic fibrosarcoma oncogene homolog K
362,911	Mal2	−1.35	mal, T-cell differentiation protein 2
171,341	Mgst1	0.59	microsomal glutathione S-transferase 1
365,352	Mical2	−0.69	microtubule associated monooxygenase, calponin and LIM domain containing 2
24,564	Mpz	1.23	myelin protein zero
60,333	Msln	−0.48	mesothelin; may play a role in cell adhesion and cell shape dynamics
83,708	Mybph	−0.85	myosin binding protein H; skeletal muscle myosin binding protein
29,556	Myh6	−1.44	myosin, heavy chain 6, cardiac muscle, alpha
688,228	Myl4	0.64	myosin, light chain 4
117,558	Mylk2	−0.59	myosin light chain kinase 2
306,616	Myom2	−0.49	myomesin 2; a major M− line structural protein
314,648	Ncln	−0.52	nicalin
114,519	Nfil3	−0.6	nuclear factor, interleukin 3 regulated
81,526	Nov	−1.34	nephroblastoma overexpressed
494,198	Oas1k	−0.9	2 ’ − 5 ’ oligoadenylate synthetase 1 K
29,511	Padi2	−0.63	peptidyl arginine deiminase 2
362,282	Pck1	0.96	phosphoenolpyruvate carboxykinase 1
360,918	Pf4	1.78	platelet factor 4; member of low-molecular weight chemokines superfamily
117,276	Pfkfb3	0.86	6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3
287,644	Phospho1	−0.55	phosphoethanolamine/phosphocholine phosphatase 1
29,692	Pla2g2a	0.68	phospholipase A2 group IIA
301,265	Pla2g7	0.65	phospholipase A2 group VII
361,352	Pqlc1	−0.49	PQ loop repeat containing 1
25,521	Prkar1b	−1.57	protein kinase cAMP-dependent type 1 regulatory subunit beta
289,138	RGD1304982	−0.88	TP53I3; tumor protein p53 inducible protein 3
60,383	Rbck1	−0.62	RANBP2-type and C3HC4-type zinc finger containing 1; PKC binding protein
362,662	Rbp7	0.86	retinol binding protein 7
24,718	Reln	−1.09	reelin
691,411	Sbk2	−0.75	SH3 domain binding kinase family, member 2
246,074	Scd1	− 0.64	stearoyl-CoA desaturase
25,719	Scg5	−1.08	secretogranin V
24,795	Serpina3n	1.45	serine (or cysteine) peptidase inhibitor
310,552	Sfrp2	−1.24	secreted frizzled-related protein 2
302,363	Sh3bgrl	0.51	SH3 domain binding glutamate-rich protein like
304,785	Slc45a3	−0.88	solute carrier family 45, member 3
289,378	Tatdn3	0.67	TatD DNase domain containing 3
364,134	Tecrl	0.52	trans-2,3-enoyl-CoA reductase-like
24,825	Tf	−0.63	transferrin; plays a role in iron transport and homeostasis
363,013	Tmem123	0.59	transmembrane protein 123
338,474	Tmem132a	−0.89	transmembrane protein 132A; ER/Golgi protein
314,472	Tmem179	0.69	transmembrane protein 179
362,401	Tmem43	−0.62	transmembrane protein 43
64,565	Tmprss11d	0.7	transmembrane protease, serine 11d
302,965	Tnfrsf12a	−0.99	tumor necrosis factor receptor superfamily, member 12a
64,104	Tnmd	1.07	tenomodulin; transmembrane protein; may be an angiogenesis inhibitor
315,159	Tob2	−0.7	transducer of ERBB2, 2
291,966	Tppp3	− 0.5	tubulin polymerization-promoting protein family member 3
29,268	Tpsb2	−0.64	tryptase beta 2; serine protease; plays a role during the inflammatory process
364,398	Trim13	− 0.87	tripartite motif-containing 13
140,939	Trim63	−0.44	tripartite motif containing 63; ubiquitin ligase
312,225	Wdr91	−0.75	WD repeat domain 91
353,227	Zbtb16	−1.01	zinc finger and BTB domain containing 16

Rows show genes listed in alphabetical order. Statistical cut-off threshold 0.05 (5%) FDR. Positive log2 fold change (log2FC) means higher expression in MI vs. Sham.

3.3. Gene ontology and pathway analysis

We performed GO analysis of the cellular components (Table 3 and Supplementary Table S2), biological processes (Table 3 and Supplementary Table S3), and molecular functions (Table 3 and Supplementary Table S4). The results showed that MI elicited diaphragm gene expression changes in particular associated with extracellular space, muscle myosin complex, M–band, Z disc (cellular components), actin-myosin filament sliding, regulation of interleukin-6 production, regulation of chemokine production, G-protein coupled receptor signaling pathway, striated muscle contraction (biological processes), iron-binding, protein kinase A catalytic subunit binding, chemokine activity, myosin light chain kinase activity, and glutathione transferase activity (molecular functions).

Table 3.

Top identified cellular components, biological processes and molecular functions.

Pruning Type: None				Pruning Type: Elim		Pruning Type: Weight
GO Term: Cellular components	p-value	p-value (FDR)	p-value (Bonferroni)	GO Term	p-value	GO Term	p-value
hemoglobin complex	2.6E-10	9.412E-08	9.412E-08	hemoglobin complex	2.6E-10	hemoglobin complex	2.6E-10
blood microparticle	0.0000003	0.0000579	0.0001158	blood microparticle	0.0000003	blood microparticle	0.0000003
haptoglobin-hemoglobin complex	0.0000073	0.0008809	0.003	haptoglobin-hemoglobin complex	0.0000073	haptoglobin-hemoglobin complex	0.0000073
extracellular region part	0.000016	0.001	0.006	muscle myosin complex	0.000098	muscle myosin complex	0.000098
extracellular region	0.000023	0.002	0.008	proteinaceous extracellular matrix	0.00034	I band	0.00024
GO Term: Biological Processes	p-value	p-value (FDR)	p-value (Bonferroni)	GO Term	p-value	GO Term	p-value
oxygen transport	5.50E-09	1.83E-05	1.83E-05	oxygen transport	5.50E-09	oxygen transport	5.50E-09
gas transport	6.40E-08	1.07E-04	2.13E-04	cellular response to cAMP	3.10E-06	cellular response to cAMP	3.10E-06
multicellular organismal process	1.30E-07	1.33E-04	4.33E-04	cellular response to dexamethasone stimulus	2.30E-05	cellular response to dexamethasone stimulus	2.30E-05
system development	1.60E-07	1.33E-04	5.32E-04	response to electrical stimulus	3.00E-04	actin-myosin filament sliding	1.50E-04
immune system process	1.50E-06	9.43E-04	0.005	positive regulation of interleukin-8 production	3.60E-04	response to electrical stimulus	3.00E-04
GO Term: Molecular Functions	p-value	p-value (FDR)	p-value (Bonferroni)	GO Term	p-value	GO Term	p-value
oxygen carrier activity	4.90E-09	2.27E-06	2.27E-06	oxygen carrier activity	4.90E-09	oxygen carrier activity	4.90E-09
oxygen binding	1.00E-07	2.32E-05	4.64E-05	oxygen binding	1.00E-07	oxygen binding	1.00E-07
molecular carrier activity	2.80E-07	4.33E-05	1.30E-04	haptoglobin binding	1.70E-06	haptoglobin binding	1.70E-06
haptoglobin binding	1.70E-06	1.97E-04	7.89E-04	iron ion binding	1.70E-05	iron ion binding	1.70E-05
iron ion binding	1.70E-05	0.002	0.008	heme binding	1.10E-04	structural molecule activity conferring elasticity	5.90E-05

Top scoring cellular components, biological processes and molecular functions for Elim and Weight pruning type.

Pathway analysis revealed that MI elicited changes in diaphragm pathways related to cytokine-cytokine receptor interaction, PI3K-Akt signaling pathway, and cardiac muscle contraction (p-value Fisher’s method) (Table 4.). However, reliance on p-value based on Bonferroni and FDR correction showed only significant alteration in the cytokine-cytokine receptor interaction pathway.

Table 4.

Top pathways and their associated p-values. P-values are presented by using Fisher’s method, and then corrected for multiple comparisons using false discovery rate (FDR) and Bonferroni corrections.

Pathway Name	Pathway ID	p-value	p-value (FDR)	p-value (Bonferroni)
African trypanosomiasis	5143	3.88426E-06	0.00058652	0.000586524
Malaria	5144	3.89219E-05	0.00293860	0.005877213
Cytokine-cytokine receptor interaction	4060	0.000153046	0.00770329	0.023109898
Adrenergic signaling in cardiomyocytes	4261	0.002860454	0.08140411	0.43192861
PI3K-Akt signaling pathway	4151	0.002980936	0.08140411	0.450121393
Cardiac muscle contraction	4260	0.003234601	0.08140411	0.488424716
Ras signaling pathway	4014	0.004489364	0.08219767	0.677894016
Gastric acid secretion	4971	0.004643523	0.08219767	0.701172005
Ether lipid metabolism	565	0.004899199	0.08219767	0.739779028

The p-value corresponding to the pathway was computed using only overrepresentation analysis. Bonferroni reduces the false discovery rate by imposing a stringent threshold on each comparison adjusted for the total number of comparisons.

3.4. Biological networks

Next, we systematically categorized and analyzed DEGs in the novel framework of “biological networks,” specifically identified in skeletal muscle (Wang et al., 2012; Smith et al., 2013). Although we mostly considered and will discuss below genes which function has been studied in skeletal muscle, we also included certain genes which function has been established in other cell types, but with a role in skeletal muscle that is still unclear. The changes in gene expression within skeletal muscle-specific biological networks are displayed in Fig. 2. Our findings within each biological network are summarized below.

Fig. 2.

Differentially expressed genes in the MI rat diaphragm categorized according to skeletal muscle-specific networks. Networks are labeled in each panel. X-axis represents the log2 fold change of gene expression. Red indicates upregulation while blue indicates the downregulation of MI-induced gene expression relative to control Sham. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.4.1. Neuromuscular junction and phrenic nerve (Fig. 2A)

Most genes associated with neuromuscular junction and phrenic nerve were downregulated. There was downregulation of Dctn1 (dynactin subunit 1) and Dok7 (docking protein 7), which are respectively involved in synaptic stabilization and activation of MuSK (muscle-specific tyrosine kinase) with subsequent acetylcholine receptor clustering (Bloch-Gallego, 2015). Regarding the phrenic nerve, there were changes in two genes encoding structural components of myelin assembly: Cldn11 (downregulated) and Mpz (upregulated). Bdnf (brain-derived neurotrophic factor), which is expressed by neuronal and skeletal muscle cells, was upregulated.

3.4.2. Membrane potential and Ca²⁺ handling (Fig. 2B)

There was downregulation of Atp1b2 (Na⁺/K⁺ ATPase β subunit 2), whereas genes encoding Ca²⁺ handling proteins Calml3 (calmodulin-like protein 3) and Tecrl (Trans-2,3-Enoyl-CoA Reductase Like) were upregulated.

3.4.3. Extracellular matrix (Fig. 2C)

There was a downregulation of the genes Loxl1 (Lysyl oxidase-like 1 enzyme) and Eln (elastin). There was upregulation of Lum (Lumican), encoding a collagen-associated proteoglycan in the basal lamina, and Fam198B (protein FAM198B), linked to fibrotic-related processes (Lu et al., 2012).

3.4.4. Sarcomere and cytoskeleton (Fig. 2D)

The majority of DEGs related to skeletal muscle sarcomere and cytoskeleton were downregulated. Specifically, the list of downregulated genes included Mylk2 (myosin light chain kinase 2), Mybph (myosin binding protein H), Myom2 (M–line structural protein myomesin 2), and Gcom1/Myzap (myocardial zonula adherens protein), localized in the Z-line. An isoform of essential myosin light chain, Myl4 (myosin light chain 4), and Ky (kyphoscoliosis peptidase) were upregulated. Two cardiac isoforms of sarcomeric proteins emerged in our data and changed in opposite directions: Actc1 (sarcomeric α-actin) was upregulated, whereas Myh6 (MHC-α, myosin heavy chain 6) was downregulated. Along with differential gene expression, Myh6 also showed differential exon-level expression, suggesting potential changes in the resultant alternatively spliced transcripts in MI compared with Sham. Specifically, exon 3 of Myh6 was upregulated in the MI group (Fig. 3). There were no differences in exon-level expression of any skeletal muscle isoforms of myosin heavy chain identified in our data (Myh1, Myh2, Myh4, and Myh7) (data not shown).

Fig. 3.

Myocardial infarction effects on Myh6 exon usage in the diaphragm. A: Plots show the fitted expression values of each exon in gene Myh6, for each of the two conditions, MI and Sham. The dotted lines represent exon 3 that showed significant upregulation in MI. B: Myh6 exon 3 normalized expression level in RPM (reads assigned per million mapped reads) units in the diaphragm.

3.4.5. Muscle fiber size (Fig. 2E)

Genes involved in the regulation of muscle size or signaling through atrophy/hypertrophy pathways were down-regulated. These included members of the family of E3 ubiquitin ligases Rbck1, Trim13, and Trim63 (Tatematsu et al., 2008; Tomar et al., 2012; Attaix et al., 2005), Ddit4 (DNA-damage-inducible transcript 4), a negative regulator of mTOR, and glucocorticoid receptor (GR) target Fkbp5 (FK506 binding protein 5) (Shimoide et al., 2016).

3.4.6. Bioenergetics and metabolism (Fig. 2F)

Myocardial infarction shifted the expression of genes involved in glycolysis and fatty acid β-oxidation. There was an upregulation of genes involved in glucose metabolism, namely Enpp1, Pfkfb3 and Pck1. Genes involved in β-oxidation, including Crot, Fmo2 and Scd1, were downregulated. Fkbp5, which also regulates metabolism during stress (Balsevich et al., 2017), was downregulated.

3.4.7. Transcription factors (Fig. 2G)

Myocardial infarction caused a general down-regulation of transcription factor genes. These included Arntl/Bmal1 (brain and muscle Arnt-like 1) (Andrews et al., 2010), Atf3, a stress-response gene (Seijffers et al., 2014; Fernández-Verdejo et al., 2017), Nfil3, regulator of inflammation and activating transcription factors (Cowell, 2002; Keniry et al., 2014), and Zbtb16, regulator of cell cycle, metabolism, cardiac hypertrophy, and fibrosis (Šeda et al., 2017).

3.4.8. Cytokine and chemokine-mediated signaling (Fig. 2H)

Our study revealed that MI elicited downregulation of Il6r (interleukin 6 receptor) and Tnfrsf12a /FN14 (tumor necrosis factor receptor superfamily, member 12a: TWEAK receptor). In addition, Cxcl13 (C-X-C motif chemokine ligand 13) was downregulated and Pf4 (platelet factor 4) was upregulated.

3.4.9. Neurohumoral signaling (Fig. 1I)

Myocardial infarction decreased the expression of genes involved in neurohumoral signaling such as Agtr1a (angiotensin II receptor, type 1a) and Prkar1b (protein kinase cAMP-dependent type 1 regulatory subunit beta).

3.4.10. Metal ion homeostasis (Fig. 1J)

We found upregulation of Alas2 (5′-aminolevulinate synthase 2) and downregulation of Tf (transferrin) genes, which are involved in heme biosynthesis and iron homeostasis, respectively (Messaoudi et al., 2017; McNabney and Essig, 1992; Moreno-Navarrete et al., 2016).

4. Discussion

We completed the current study to define the diaphragm transcriptional profile in the chronic stage after a large MI, which is accompanied by diaphragm pathological remodeling and abnormal function. Our RNA-Seq analysis revealed 112 DEGs, with approximately two-thirds being downregulated. The skeletal muscle-specific biological networks that showed predominantly downregulated genes were neuromuscular junction, extracellular matrix, excitation–contraction coupling, and sarcomere and cytoskeleton. Surprisingly, the level of transcripts related to inflammatory/immune signaling and muscle atrophy were also downregulated.

4.1. Neuromuscular junction and phrenic nerve

The changes in voluntary inspiratory function reported in patients after an MI can arise from abnormalities in phrenic nerve, diaphragm neuromuscular junction, and muscle fibers. Our data revealed changes in the expression of genes involved in the stabilization of the neuromuscular synapse (downregulation of Dctn1 and Dok7, and upregulation of Ky), but there were no changes in agrin (Agrn) and acetylcholine receptor (Ach receptor; CHRN) subunits that are associated with neurodegeneration and denervation (Elliott et al., 2016; Wu et al., 2014; Adams et al., 1995). We also found that MI caused upregulation of Bdnf in the diaphragm that contrasts the decrease seen in denervation and aging that are accompanied by diaphragm neuromuscular dysfunction (Mantilla et al., 2013). Overall, these results suggest potential neuromuscular remodeling in the diaphragm, without changes in the mRNA level indicative of overt denervation or neurodegeneration in the chronic stage post-MI.

4.2. Membrane potential and Ca⁺² handling

Intact muscle contraction involves a process where membrane depolarization leads to calcium release from the sarcoplasmic reticulum and activation of sarcomeric proteins responsible for muscle force generation and shortening. The notable change in genes related to excitation–contraction coupling was a downregulation of Atp1b2 (Na⁺/K⁺ ATPase β subunit 2), which is crucial for normal function of the Na⁺/K⁺ ATPase and maintenance of steady-state membrane potential (Clausen, 2003; Geering, 2008; Pirkmajer and Chibalin, 2016) and may contribute to impairments in contractile function, especially during repetitive contractions (McKenna et al., 2008).

Myocardial infarction impairs diaphragm calcium flux resulting in elevated levels of intracellular calcium (Dominguez and Howell, 2003), which is evidence of abnormal Ca⁺² handling. We found upregulation of two genes that are reportedly involved in calcium handling: Calml3 (Bennett et al., 2013) and Tecrl (Devalla et al., 2016). Calml3 competes with calmodulin and affects free calcium dynamics in non-muscle cells (Bennett et al., 2008). Although Calml3 has been considered to be exclusively expressed in epithelial cells (Yaswen et al., 1992), limb muscles have Calml3 FPKM ranging from 0.1 to 4.1. In healthy rat (sham) diaphragm, FPKM was 0.397 and elevated 4 fold in animals with large MI. Tecrl mutation causes cardiac arrhythmia and decreased protein expression of RYR2 and CASQ2, calcium-handling genes in cardiomyocytes (Devalla et al., 2016).

4.3. Extracellular matrix

The extracellular matrix is involved in signal transduction, structural support for muscle regeneration, and is an important determinant of muscle stiffness (Grzelkowska-Kowalczyk, 2016; Gillies and Lieber, 2011; Anglister and McMahan, 1984). Diaphragm from animals in the MI group showed downregulation of extracellular matrix-related and collagen genes Loxl1 (lysyl oxidase-like 1) and Eln (elastin). Loxl1 is involved in collagen cross-linking that decreases collagen proteolysis and increases stiffness (Smith et al., 2016; Zenkel et al., 2014). Eln is mainly responsible for tissue elasticity and structural support of connective tissue (Gilbert et al., 2016). Therefore, the downregulation of Eln may be a consequence of decreased Loxl1 mRNA (Liu et al., 2004; Schlötzer-Schrehardt et al., 2012). Myocardial infarction also caused upregulation of Fam198b and Lum. Fam198b is upregulated in cardiomyocytes post-MI and has been postulated to promote fibrotic or stress-related processes (Lu et al., 2012). Lum is unpregulated in skeletal muscle by overload (Mendias et al. 2017), encodes a collagen-associated proteoglycan in the basal lamina that promotes fibrocyte differentiation (Pilling et al., 2015), regulates collagen assembly into fibrils (Dupuis et al., 2015), and heightens the activation of Toll-like receptor 4 signaling by pathogen-associated molecular patterns (Wu et al., 2007). Overall, our data suggest that the chronic stage after a large MI leads to diaphragm extracellular matrix remodeling.

4.4. Sarcomere and cytoskeleton

The coordinated action of the sarcomere and structural integrity of the cytoskeleton and cell membrane are critical for proper contractile function and force transmission. The structural components Myom2, and Gcom1 were downregulated, and Ky was upregulated in the diaphragm post-MI. Myom2 (myomesin 2 or M− protein) is an elastic protein localized in the M–band and the main constituents of M–bridges that connect the thick filament and stabilizes the sarcomere during contraction and relaxation (Henderson et al., 2017; Tskhovrebova and Trinick, 2012). Gcom1, also known as Myzap (myocardial zonula adherens protein) localizes in intercalated discs of cardiomyocytes and is involved in mechano-sensing (Seeger et al., 2010). In skeletal muscles, Gcom1 is downregulated by denervation and genetically-induced atrophy (Adams et al., 2012). The downregulation of Myom2 and Gcom1 in the diaphragm can result in impaired mechano-transduction and susceptibility to disruption of sarcomere integrity. The upregulation of Ky, which also occurs in aged skeletal muscle (Lin et al., 2018), might increase the structural integrity of the Z-disk and mechanical support of the myofiber (Baker et al., 2010). We also observed downregulation of Mybph and Mylk2, and upregulation of Myl4, which are involved in regulation of myofibrillar protein function and muscle contractile properties (Henderson et al., 2017; Yamamoto, 1988, 1984; Takashima, 2009; Sweeney et al., 1993; Klug et al., 1982; Ryder et al., 2007; Stull et al., 2011; Schiaffino et al., 2015). In general, our findings are consistent with abnormal mRNA levels of genes involved in the modulation of force generation and transmission. However, there are no signs of diminished expression at the transcriptional level for the most abundant contractile proteins, actin and myosin.

4.5. Muscle fiber size and remodeling

The ubiquitin–proteasome and autophagy-lysosome systems are important regulators of muscle catabolism. Upregulation of ubiquitin ligases, such as MURF-1, promotes and knockout prevents muscle atrophy (Bodine et al., 2001; Attaix et al., 2005; Bilodeau et al., 2016). We found that an MI caused downregulation of members of the family of E3 ubiquitin ligases Trim63 (MuRF-1), Trim13, and Rbck1. MuRF-1 has a well-defined role in muscle atrophy (Bodine et al., 2001; Adams et al., 2019), and previous studies have shown that MI increased diaphragm MuRF-1 protein (Mangner et al., 2015) and mRNA (van Hees et al., 2008a), and pharmacological blockade of MuRF1 prevents diaphragm weakness post-MI (Adams et al., 2019). These observations contrast our findings of decreased MuRF-1 mRNA. Apart from using different techniques (RT-PCR vs. RNASeq), rat strains (Wistar vs Sprague-Dawley), or species (mice vs. rats), there are no obvious discrepancies to explain the findings for MuRF-1. Trim13 participates in the initiation of autophagy and response to endoplasmic reticulum stress (Tomar et al., 2012; Parry and Willis, 2016), but its effects on skeletal muscle are unknown. Rbck1 deficiency causes skeletal and cardiac myopathy (Nilsson et al., 2013; Vallentin and Mochly-Rosen, 2007). Additional modulators of myofiber size downregulated post-MI were Ddit4 (REDD1) and Fkbp5 that are involved in skeletal muscle anabolism (Goodman et al., 2011; Gordon et al., 2014; Shimoide et al., 2016). In general, the DEGs involved in the regulation of protein turnover and myofiber size suggest a profile that is consistent with stimulation of muscle hypertrophy and regeneration or inhibition of muscle atrophy, presumably as a result of overload. However, the diaphragm fiber cross-sectional area is either unchanged or decreased in animals (Stassijns et al., 1998; Laitano et al., 2016; Hwee et al., 2015; Lima et al., 2014; Kelley et al., 2020) and patients with heart failure (Lindsay et al., 1996). These observations suggest that in the chronic stage post-MI the diaphragm is in a state of ‘anabolic resistance’, defined as abnormal protein synthesis and muscle fiber growth in response to hypertrophic stimuli (Haran et al., 2012).

4.6. Bioenergetics and metabolism

During normal breathing, the diaphragm relies approximately equally on carbohydrates and lipids as energy sources (Rochester and Briscoe, 1979). The effects of an MI on diaphragm mRNA levels of genes involved in energetics and metabolism suggest an increase in glycolysis and fatty acid β-oxidation. Three genes involved in carbohydrate metabolism were upregulated in the diaphragm of MI animals: Enpp1, Pfkfb3, and Pck1. Enpp1 inhibits insulin receptor-mediated signaling (Du and Wei, 2014) and overexpression of Enpp1 causes insulin resistance (Pan et al., 2012), which occurs post-MI (Ohta et al., 2011). Pfkfb3 promotes glycolysis (Niwa et al., 2005), whereas Pck1 (phosphoenolpyruvate carboxykinase) plays a role in gluconeogenesis and glyceroneogenesis (Yang et al., 2009) and increases in the diaphragm with starvation (Snell and Duff, 1979). The genes involved in fatty acid β-oxidation modified post-MI were Scd1, Fmo2, and Crot. These genes were downregulated, which would have the collective effect of increasing fatty acid β-oxidation (Paton and Ntambi, 2009; Rogowski et al., 2013; Dobrzyn and Dobrzń, 2006; Flowers and Ntambi, 2008; van den Bosch et al., 1992; Poirier et al., 2006) and xenobiotics (Veeravalli et al., 2014). Collectively, the mRNA levels of metabolic genes are consistent with increased diaphragmatic activity. However, the elevated Pck1 may be a compensatory response to metabolic disturbance and the response of Enpp1 suggests a pathological effect that could lead to insulin resistance in the diaphragm.

4.7. Cytokine and chemokine-mediated signaling

Myocardial infarction and the resulting heart failure increase cytokine levels systemically and in limb muscles (Paulus, 2000; Adams et al., 2002). To our surprise, cytokine and chemokine-related genes were mostly downregulated. The most prominent downregulated genes were Il6r, Tnfrsf12a/Fn14, and Cxcl13. Il6r is an important component of IL-6 signaling (Muñoz-Cánoves et al., 2013; Zhou et al., 2016), whereas Fn14 acts as a receptor for tumor necrosis factor (TNF)-like weak inducer of apoptosis (TWEAK) (Kumar et al., 2012). Both Il6 and TWEAK activate atrophy signaling (Muñoz-Cánoves et al., 2013; Kumar et al., 2012). The decrease in mRNA levels for Il6 and TWEAK receptors may reflect a compensatory response to a sustained elevation in Il6 and TWEAK after MI. These findings are consistent with the decrease in atrophy-related genes (e.g., MuRF1) in the diaphragm. The chemokine Cxcl13, which showed the greatest fold change among all DEGs, is implicated in tumor growth (Tan et al., 2018). The role of Cxcl13 in skeletal muscle is unknown, but the gene is upregulated during myogenesis (Griffin et al., 2010). The chemokine Cxcl4/Pf4, one of the most abundant platelet chemokines that promote neutrophil accumulation and inhibits angiogenesis (Deppermann and Kubes, 2018), was upregulated in the diaphragm post-MI. The role of Cxcl4/Pf4 on skeletal muscle is undefined, but our data suggest that Cxcl4/Pf4 upregulation was not associated with the accumulation of neutrophils or macrophages in the diaphragm. Specifically, levels of mRNA for neutrophil and macrophage markers (Ly6G, Mpo, and Cd68) were unchanged in the rat diaphragm post-MI (Supplementary Table S1).

4.8. Neurohumoral signaling

The renin-angiotensin-aldosterone and sympathetic nervous systems are hyperactive and pathogenic in ischemic and non-ischemic heart failure. Angiotensin II is elevated and contributes to diminished inspiratory pressure in patients with ischemic and non-ischemic heart failure (Coirault et al., 2001) and diaphragm weakness in a rabbit model of non-ischemic heart failure (Coirault et al., 1999). Interestingly, we found that diaphragm Agtr1 (angiotensin II receptor type 1) mRNA was downregulated post-MI, which likely reflects a compensatory response to increased angiotensin II in the circulation. Hyperactivity of the sympathetic nervous system also contributes to skeletal muscle contractile dysfunction through enhanced beta-adrenergic receptor and protein kinase A signaling that disrupts calcium handling (Reiken et al., 2003). In contrast to this notion, we observed a decrease in Prkar1b (regulatory subunit of protein kinase A) in the diaphragm post-MI.

4.9. Transcription factors

Myocardial infarction caused downregulation of transcription factors Zbtb16, Atf3, Nfil3, and Arntl/Bmal1 in the diaphragm. Zbtb16 has been implicated in the metabolic syndrome (Šeda et al., 2017). In contrast, Zbtb16 is upregulated in skeletal muscle during acute cold exposure to promote thermogenesis, and overexpression promotes mitochondrial biogenesis in C2C12 myotubes (Plaisier et al., 2012). Atf3 (Activating Transcription Factor 3) is a stress-response gene induced by cytokines and chemokines (Rohini et al., 2018). In limb muscle, ATF3 is upregulated by exercise and ATF3 knockout impairs the muscle responses to exercise training (Fernández-Verdejo et al., 2017). In this sense, the downregulation of ATF3 is a paradoxical finding as the work of breathing is increased and cytokines are elevated in ischemic and non-ischemic heart failure (Paulus, 2000; Cross et al., 2011). Nfil3 is upregulated in the failing heart and counteracts angiotensin II signaling in cardiomyocytes in culture (Chen et al., 2018). Nfil3 is a transcriptional repressor involved in the anti-inflammatory response and a potential modulator of circadian controlled genes (Cowell, 2002). Indeed, Nfil3 exhibits circadian oscillation in skeletal muscle (McCarthy et al., 2007). In this context, MI also decreased diaphragm Bmal1 (brain and muscle Arnt-like 1), one of the core components of the molecular clock. Interestingly, muscle-specific knockout of Bmal1 impairs skeletal muscle contractile function and metabolism (Lefta et al., 2012; Andrews et al., 2010).

4.10. Metal ion homeostasis

Iron deficiency has been linked with inspiratory muscle weakness in patients with heart failure (Tkaczyszyn et al., 2018). Transferrin (Tf) is involved in extracellular iron-binding and transport and was downregulated in the diaphragm post-MI (Wang and Pantopoulos, 2011). Transferrin is upregulated with intracellular iron deficiency and downregulated with overload in adipocytes (Moreno-Navarrete et al., 2016), but not skeletal muscle (Rodriguez et al., 2007). Another iron-related DEG in the diaphragm post-MI is Alas2, a rate-limiting enzyme in heme synthesis. Alas2 was upregulated in the diaphragm post-MI in agreement with data from cardiomyocytes (Sawicki et al., 2015). Cardiac overexpression of Alas2 increases iron accumulation, promotes oxidative stress, and induces cardiomyocyte death (Sawicki et al., 2015). Thus, up-regulation of ALAS2 may be involved in increased heme accumulation and iron overload in the post-MI diaphragm.

4.11. Methodological considerations and limitations

We opted to use Sham-operated animals as ‘controls’ to isolate the effects of MI on the diaphragm. Some studies suggest that the Sham procedure has minimal effect on the systemic and cardiac physiology compared to non-surgical controls (Iyer et al., 2016). We are not aware of comparisons of diaphragm and respiratory variables between non-surgical controls and Sham. Thoracotomy causes a pneumothorax and alters the mechanics of breathing with potential impact on the diaphragm. However, the pneumothorax appears to resolve within 72 hrs post-surgery. We completed experiments 16 weeks post-surgery and the impact of surgery, if any, is likely minimal compared to non-surgical control at that time. A limitation of our study is the focus on the transcriptome. While we have a published analysis of whole-tissue proteomics and phenotype in these animals (Kelley et al., 2020), we did not attempt a multi-omics approach and the transcriptome changes we report here may translate to proteome changes at a later time point. Correlation of transcriptional changes with whole-tissue proteomics yields limited agreement (Van Pelt et al., 2020); a better approach to draw mechanistic conclusions would establish the relationship between transcriptomics and proteomics of subcellular fractions (e.g., nuclear, cytosolic, mitochondrial).

5. Conclusions

The RNA-Seq data from the diaphragm of rats in the chronic stage post-MI suggest remodeling of the neuromuscular junction, extracellular matrix, and sarcomere and cytoskeleton. The changes in the mRNA level were not indicative of overt denervation or neurodegeneration. The mRNA levels of genes involved in metabolism and iron homeostasis suggest greater reliance on fatty acid oxidation and iron/heme-group accumulation. Atrophy- and hypertrophy-related gene changes were consistent with increased anabolic signaling. Overall, these changes seem to be reflective of diaphragm overload in a pathological state that culminates in anabolic resistance. Genes linked to molecular circadian rhythms were downregulated and potentially contribute to the diaphragm abnormalities induced by MI. To our surprise, genes related to cytokine, chemokine, and neurohumoral signaling were mostly downregulated, presumably as a compensatory response to elevated circulating levels of cytokines, chemokines, and neurohumoral activation.

Abbreviations:

MI

myocardial infarction

RV

right ventricle

LV

left ventricle

HFrEF

heart failure with reduced ejection fraction

DEGs

differentially expressed genes

RNA-Seq

RNA-sequencing

acDNA

amplified cDNA

FDR

false discovery rate

GO

gene ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

GEO

Gene Expression Omnibus