Tissue
|
MG⸸ (37)
|
NuMP* (1,041)
|
All⸹ (24,616)
|
MtMP⸹
(13)
|
Mt
tRNA¥
|
Mt
rRNAꝕ
|
Over-exp
|
Under-exp
|
Total DE
|
Over-exp
|
Under-exp
|
Total DE
|
Over-exp
|
Under-exp
|
Total DE
|
|
|
|
Adrenal cortex
|
0
|
0
|
0
|
66
|
157
|
223
|
2018
|
2532
|
4550
|
-
|
-
|
-
|
Adrenal medulla
|
2
|
0
|
2
|
124
|
89
|
213
|
1620
|
2260
|
3880
|
-
|
-
|
2
|
Brain caudal lobe
|
2
|
0
|
2
|
190
|
151
|
341
|
3354
|
4150
|
7504
|
-
|
-
|
2
|
Brain cerebellum
|
1
|
0
|
1
|
120
|
187
|
307
|
3282
|
3879
|
7161
|
-
|
-
|
1
|
Brain stem
|
0
|
0
|
0
|
95
|
161
|
256
|
1911
|
3294
|
5205
|
-
|
-
|
-
|
Heart
|
17
|
0
|
17
|
576
|
53
|
629
|
2185
|
2833
|
5018
|
13
|
2
|
2
|
Kidney cortex
|
5
|
0
|
5
|
315
|
30
|
345
|
1465
|
2851
|
4316
|
ATP8, CYB, COX2
|
-
|
2
|
Kidney medulla
|
0
|
0
|
0
|
93
|
58
|
151
|
1618
|
2444
|
4062
|
-
|
-
|
-
|
Latissimus d muscle
|
14
|
0
|
14
|
481
|
74
|
555
|
2668
|
2970
|
5638
|
13
|
1
|
-
|
Leg muscle
|
14
|
0
|
14
|
456
|
80
|
536
|
2773
|
3083
|
5856
|
13
|
1
|
-
|
Leukocytes
|
0
|
9
|
9
|
99
|
368
|
467
|
5270
|
3948
|
9218
|
ATP8, ND3, COX3, COX2, ATP6, CYB, COX1, ND1, ND2
|
-
|
-
|
Liver
|
0
|
0
|
0
|
360
|
82
|
442
|
2914
|
3124
|
6038
|
-
|
-
|
-
|
Loin muscle
|
14
|
0
|
14
|
514
|
114
|
628
|
3853
|
3707
|
7560
|
13
|
1
|
-
|
Lung
|
0
|
5
|
5
|
31
|
430
|
461
|
1793
|
2565
|
4358
|
ND3, ATP6, COX3, CYB, ND1
|
-
|
-
|
Lymph node
|
0
|
7
|
7
|
38
|
306
|
344
|
2649
|
2562
|
5211
|
ND1, ND3, ATP6, ND4, ND4L, CYB, ND2
|
-
|
-
|
Mammary
|
0
|
0
|
0
|
86
|
250
|
336
|
1805
|
1864
|
3669
|
-
|
-
|
-
|
Mediastinal LN
|
0
|
0
|
0
|
46
|
77
|
123
|
993
|
2432
|
3425
|
-
|
-
|
-
|
Omental fat
|
0
|
2
|
2
|
78
|
163
|
241
|
1107
|
1901
|
3008
|
-
|
|
2
|
Ovary
|
0
|
0
|
0
|
58
|
70
|
128
|
985
|
2018
|
3003
|
-
|
-
|
-
|
Pituitary gland
|
0
|
5
|
5
|
73
|
251
|
324
|
2522
|
3033
|
5555
|
ND1, ND3, CYB, ATP6, ATP8
|
-
|
-
|
Placenta
|
0
|
10
|
10
|
110
|
232
|
342
|
2842
|
3267
|
6109
|
ND3, ND1, ATP6, COX3, ND2, CYB, ND4L, ND4, COX2, ND5
|
-
|
-
|
Rib muscle
|
7
|
0
|
7
|
500
|
70
|
570
|
2325
|
2933
|
5258
|
ND1, ND3, COX2, ATP6, CYB, ND4L, ND2
|
-
|
-
|
Skin black
|
0
|
0
|
0
|
79
|
176
|
255
|
1814
|
2574
|
4388
|
-
|
-
|
-
|
Skin white
|
0
|
0
|
0
|
79
|
205
|
284
|
1780
|
2554
|
4334
|
-
|
-
|
-
|
Spleen
|
2
|
0
|
2
|
47
|
260
|
307
|
2003
|
2449
|
4452
|
-
|
-
|
2
|
Subcutaneous fat
|
0
|
0
|
0
|
105
|
95
|
200
|
1299
|
2238
|
3537
|
-
|
-
|
-
|
Thymus
|
0
|
10
|
10
|
63
|
363
|
426
|
3509
|
3208
|
6717
|
ND3, ATP6, ND1, CYB, ND4, COX3, COX1, ND4L, ATP8, COX2
|
-
|
-
|
Thyroid
|
0
|
2
|
2
|
33
|
329
|
362
|
2335
|
2335
|
4670
|
ND1, ND3
|
-
|
-
|
Tongue
|
14
|
0
|
14
|
413
|
30
|
443
|
1369
|
2365
|
3734
|
13
|
1
|
-
|
( ) total number of genes in a category; Over-exp Over-expression, under-exp under-expression; MG⸸ Genes from mitochondrial genome including tRNA and rRNAs, NuMP* Mitochondrial protein genes encoded by the nuclear genome, MtMP⸹ Mitochondrial protein genes encoded by mitochondrial genome, Mt tRNA¥ Mitochondrial transfer RNA, Mt rRNAꝕ Mitochondrial ribosomal RNA, All⸹ all genes from nuclear and mitochondrial genomes
|
Table 2. Number of differentially expressed (DE) genes in tissues by gene categories averaged for two cows in the Main Cows dataset
Table 3
KEGG functional annotation of overall differentially expressed genes of selected tissues with the largest number of genes averaged across two cows in the Main Cows dataset
Tissues | Enrichment | No. of genes (Overlap NuMP*) | Adj. p | Other pathways |
Heart | Oxidative phosphorylation | 101 (93) | 3.3e− 53 | Parkinson’s disease, Alzheimer's disease, Huntington’s disease, NAFLD, carbon metabolism, Metabolic pathways, cardiac muscle contraction, TCA cycle, |
Leg muscle | Oxidative phosphorylation | 83 (78) | 1.0e− 32 | Parkinson's disease, Alzheimer's disease, NAFLD, Huntington's disease, Carbon metabolism, metabolic pathways, Proteasome, cardiac muscle contraction Biosynthesis of antibiotics |
Loin muscle | Oxidative phosphorylation | 89(86) | 5.8e− 39 | Parkinson's disease, Alzheimer's disease, NAFLD, Huntington's disease, Huntington's disease, carbon metabolism, Proteosome, Cardiac muscle contraction |
Rib muscle | Oxidative phosphorylation | 86 (81) | 1.8e− 35 | Parkinson's disease, Alzheimer's disease, NAFLD, Huntington's disease, Metabolic pathways, carbon metabolism, Proteasome |
Tongue | Oxidative phosphorylation | 98(91) | 3.0e− 50 | Parkinson's disease, Alzheimer's disease, NAFLD, Huntington's disease, Metabolic pathways, carbon metabolism, cardiac muscle contraction, |
Kidney cortex | Metabolic pathways | 312(118) | 3.9e− 28 | Biosynthesis of antibiotics, Carbon metabolism, Valine, leucine and isoleucine degradation, Glycine, serine and threonine metabolism, tryptophane metabolism, Fatty acid metabolism |
Kidney medulla | Focal adhesion | 67(0) | 3.0e− 10 | Tight junction, calcium signalling pathway, Gastric acid secretion. ECM-receptor interaction, cGMP-PKG signalling pathway, Gastric acid secretion, Dilated cardiomyopathy |
Liver | Metabolic pathways | 387(123) | 4.1e− 49 | Biosynthesis of antibiotics, Peroxisome, valine, leucine and isoleucine degradation, complement and coagulation cascades, fatty acid degradation, tryptophan metabolism, carbon metabolism |
Brain caudal lobe | Axon guidance | 57(0) | 8.0e− 18 | Glutamatergic Synapse, Domaminergic synapse, MAPK signalling pathway, Adrenergic signalling in cardiomyocytes, Retrograde endocannabinoid signalling, cAMP signalling pathway, Synaptic vesicle cycle, GABAergic synapse, Morphine addiction, Glutamatergic synapse, Circadian entrainment, Dopaminergic synapse |
Brain cerebellum | Glutamatergic synapse | 49(2) | 3.1e− 14 | GABAergic synapse, Retrograde endocannabinoid signalling, Morphine addiction, Circadian entrainment, Dopaminergic synapse, cAMP signalling pathway, Adrenergic signalling in cardiomyocytes, axon guidance, |
Brain stem | GABAergic synapse | 36(2) | 3.1e− 9 | Glutamergic synapse, Morphine addiction, Retrograde endocannabinoid signalling, Dopaminergic synapse, Circadian entrainment |
Leukocytes | Chemokine signalling pathways | 74 | 4.7e− 15 | Focal adhesion, leukocyte transendothelial migration, Rap1 signalling pathway, natural killer cell mediated cytotoxicity, regulation of actin cytoskeleton, B cell receptor signalling pathway |
*we show the number of genes in the top enriched pathway that overlap mitochondrial proteins |
More than 99% of the MP genes originate from the nuclear genome (NuMP). The proportion of DE NuMP genes across the tissues varied from 12 to 60% with higher proportions (> 50%) in heart and skeletal muscles. The proportion of under or over-expressed DE NuMP genes within tissues varied considerably. A relatively greater proportion of NuMP genes were over-expressed in heart, kidney cortex, skeletal muscles, and tongue, and under-expressed in blood leukocytes, lymph nodes, placenta, lungs, mammary, and thymus (Fig. 2). The expression of NuMP genes was similar between animals in the Main Cows as indicated by the clustering together of same tissues, with the exception of five tissues (Fig. 2; adipose, ovary, kidney cortex, and leukocytes).
In contrast to NuMP, differential expression of MtMP genes were observed in less than 50% of tissues (14 out of 29 tissues). The proportion of DE MtMP genes within tissues ranged widely from 0 (no genes) to 100% (all 13 MtMP genes). Specifically, MtMP genes were 100% DE in heart, leg muscle, latissimus dorsi muscle, loin muscle, and tongue, and ranged between 50–75% in other tissues (leukocytes, placenta, thymus, rib muscle, and lymph node). Unlike NuMP genes, all DE MtMP genes were expressed in a single direction (i.e. either all over-expressed or all under-expressed) meaning every DE MtMP gene was over-expressed in heart, tongue, muscles and kidney cortex, and under-expressed in blood leukocytes, placenta, lymph node, pituitary, thymus, and thyroid (Fig. 3). Further, there were similarities between the expression of DE MtMP and NuMP genes within a tissue. For instance, every tissue showing over-expression of DE MtMP genes invariably showed predominant over-expression of NuMP genes and similarly for under-expression. In addition to MtMP genes, some of the non-protein coding genes from the mitochondrial genome were also DE in several tissues (16 s rRNA, 12 s rRNA, tRNA-Pro and tRNA-Ser).
Within groups of tissues with either over-expression of MP genes (heart, skeletal muscles, liver and kidney cortex) or under-expression (leukocytes, thymus, placenta and lymph node), we examined all overlapping genes and their functional enrichment. In heart and skeletal muscles, there were 1,088 over-expressed genes in common including 320 NuMP and seven MtMP genes. Altogether across these 1,088 genes, there was significant enrichment for OXIPHOS, metabolic pathways and neurodegenerative disease pathways as in these individual tissues. Similarly, liver and kidney cortex had 1,249 over-expressed genes in common including 223 NuMP genes (0 MtMP genes) and these were significantly enriched for metabolic pathways and peroxisome, valine, leucine and isoleucine degradation. In contrast, the DE genes in common for tissues in the under-expression group was low (63) with only 20 NuMP genes (0 MtMP genes). Across all 63 genes, there was enrichment for adrenergic signalling in cardiomyocytes, dilated cardiomyopathy, cardiac muscle contraction and hypertrophic cardiomyopathy. Altogether, these results indicated a significant role of the over-expressed MP genes contributing to the enriched pathways in the over-expression tissue group, while this pattern was not observed in the under-expression group.
Main Cows: Foetuses
The analysis for functional enrichment of overall DE genes in six foetal tissues showed significant enrichment of OXIPHOS and metabolic pathways only in heart and lungs but not in leg muscle (Additional file 5). The NuMP genes were over-expressed in the heart and under-expressed in the remaining tissues, including leg muscles (Additional file 6). Similarly, the MtMP genes were prominently over-expressed in heart, under-expressed in the lungs and not significant in the remaining tissues (Additional file 7). Between the two foetuses, liver of the male fetus showed a relatively higher expression of NuMP genes compared to the female.
Co-expression network analysis of mitochondrial protein genes
The gene co-expression network constructed based on the affinity matrix from genes correlated in expression > |0.95| had altogether 3643 genes clustered into four major network clusters I-IV (Fig. 4). The NuMP genes were concentrated in two main clusters (I and IV) indicating co-expression among NuMP genes and the remaining NuMP genes were sparsely scattered across all other clusters. Similarly, MtMP genes were all grouped in cluster I. Clusters I and IV containing subgroups of highly co-expressed NuMP and MtMP and NuMP genes are referred to as NuMP-MtMP and NuMP clusters respectively. Within the NuMP-MtMP cluster, the MP genes from the respective genomes were highly co-expressed. The NuMP-MtMP cluster was significantly enriched for OXIPHOS, metabolic pathways and mitochondrial diseases’ pathways. Similarly, the NuMP cluster (cluster IV) was over-represented for signalling pathways, contraction, metabolic pathways and myopathies related to the heart (Table 4). The gene functions of the non-mitochondrial protein genes (Non-MP) in the NuMP-MtMP cluster were associated with heart and muscle functioning, signalling, and contraction (Additional file 8,9).
Table 4
Summary of gene, composition, and functional enrichment of KEGG pathways of genes in co-expression clusters (FDR < 1e-05) in the Main Cows dataset
Cluster | No. of MtMP genes | No. of NuMP Genes | No. of Non-MP genes | Total No. of genes | Enrichment of pathways |
I (NuMP-MtMP Cluster) | 13 | 216 | 584 | 813 | Parkinson’s disease, Oxidative phosphorylation, Alzheimer’s diseases, Huntington diseases, Non-alcohol fatty liver diseases, metabolic pathways, Citrate cycle, carbon metabolism, Cardiac muscle contraction, Proteosome |
II | 0 | 10 | 871 | 881 | Retrograde endocannabinoid signaling, GABAergic synapse, Nicotine addiction, Morphine addictions, Glutamatergic synapse, Dopaminergic synapse, Synaptic vesicle cycle, Neuroactive ligand-receptor interaction |
III | 0 | 12 | 923 | 935 | Cell adhesion molecules (CAMs), Staphylococcus aureus infection, intestinal immune network for IgA production, Leishmaniasis, Antigen processing and presentation, viral myocarditis, Allograft rejection, primary immunodeficiency, Hematopioetic cell lineage, Natural killer cell-mediated cytotoxicity |
IV (NuMP cluster) | 0 | 79 | 466 | 545 | Chemical carcinogenesis, Complement and coagulation cascade, Drug metabolism – cytochrome p450 metabolism, steroid hormone biosynthesis, retinol metabolism, Metabolic pathways, Complement and coagulation cascades, bile secretion, primary bile acid biosynthesis, tryptophan metabolism, carbon metabolism, fatty acid metabolism |
We tested if the co-expression of NuMP genes in the NuMP-MtMP cluster was due to random chance using a Chi-square goodness of fit test. The frequency of NuMP genes in the cluster was significantly higher than random (ꭓ2= 307.6, p < 0.01), supporting that the cluster was enriched with co-expressed MPs.
Further, we investigated the effect of TAD on the co-expression by comparing the number of 651 TAD mapped genes in the NuMP-MtMP cluster with the mean from 100 randomly generated samples of 651 genes from 3022 TAD mapping genes across the clusters. It showed involvement of NuMP-MtMP genes in a similar number TADs (472 ± 10 vs 484), but NuMP-MtMP were more likely to be present in groups of two or more within TADs. The total number of genes occurring in a two or more in a TAD was 282 and 116 (± 6) in NuMP-MtMP cluster and random samples respectively. This indicated that co-expression in the NuMP-MtMP cluster was enriched within TADs.
Validation of patterns of mitochondrial protein gene expression
The key findings on the expression of MP genes from the Main Cows dataset were validated using two independent datasets (Validation Cow and Validation Sheep). Both validation sets confirmed the general trends of MP gene expression and coexpression in tissues.
Firstly, both validation datasets confirmed the over-expression of MP genes in heart and skeletal muscles, and under-expression in blood leukocytes as in the adult tissues of the Main Cows dataset (Additional file 10–17). Further, expression of MP genes within tissues, as indicated by LFC values between Main Cows and Validation Cow (Fig. 5), were highly correlated (R2 0.67–0.96) except for thyroid (R2 0.013). Similarly, the correlation of LFC values between Main Cows and Validation Sheep was high (R2 0.6–0.87), except for mammary and lungs (R2 0.34, 0.37) (Additional file 18). We investigated the poor correlation of gene expression in thyroid between the Main Cows and the Validation Cow. At least 35 DE NuMP genes in common between the datasets were expressed in opposite directions. These genes were mainly enriched for metabolic pathways, pyruvate metabolism and synthesis of antibiotics. Interestingly, the expression of NuMP genes between Validation Cow and Validation Sheep were moderately correlated in the thyroid (R2 0.59) (Additional file 19).
Secondly, the ‘either all over-expression or all under-expression’ of DE MtMP genes within tissues was supported by findings from both validation datasets. Further, the expression of MtMP genes in the direction of the dominant DE NuMP genes also remained evident across datasets.
Thirdly, the co-expression of MtMP and NuMP genes in a cluster were reproduced in the Validation Cow (Additional file 20), and to some extent in Validation Sheep (Additional file 21). The co-expression of MP genes in the NuMP-MtMP cluster in the Validation Cow was more than expected by random chance (ꭓ2=207.847, p < 0.01) indicating that this cluster was enriched for co-expression of MP genes.
Finally, the overlap of genes in NuMP-MtMP clusters across the Main Cow and validation datasets was higher than would be expected if genes were randomly allocated to clusters. In particular, the occurrence of MtMP genes were almost coincidental (13/13) between cow datasets and 12/13 genes in common between cow and sheep datasets. Similarly, a considerable proportion of NuMP genes and also non-mitochondrial protein genes, were in common across datasets (Fig. 6).