Selection of COVID datasets
In order to examine the NEM and mtDNA expression signature in SARS-CoV-2 infection, we utilized data sets that were uploaded to GEO (GSE147507 and GSE110551) and the BIG Data Center (CRA002390). These RNASeq data sets were derived from A549, A549 (ACE2), Calu-3, and NHBE cells as well as from SARS-CoV-2 patients’ lung autopsies and bronchoalveolar lavage fluid (BALF). A549 cells were infected with seasonal influenza A virus (IAV), human orthopneumovirus (respiratory syncytial virus; RSV), human parainfluenza virus 3 (HPIV3), and SARS-CoV-2. ACE2-expressing A549 cells, Calu-3 cells, and NHBE cells were infected with SARS-CoV-2. NHBE cells were also infected with IAV. The original authors who curated the in vitro data infected SARS-CoV-2 in A549 cells at low and high multiplicities of infection (MOI). They found that the rate of SARS-CoV-2 replication after low MOI was comparable to the replication rate after high MOI in ACE2-expressing A549 cells.4 The original authors also observed that low MOI SARS-CoV-2 infection stimulated a relative muted proinflammatory response, which was ablated in high MOI SARS-CoV-2 infection in ACE2-expression A549 cells. We specifically contrasted infection conditions by using low MOI SARS-CoV-2 in order to (1) stay consistent with previously published results and limit confounding effects from stoichiometry disruption of high SARS-CoV-2 components. The number of differentially expressed genes (DEGs) and NEM DEGs per biological source is listed in Table 1. Significant DEGs were filtered by an adjusted p value of 0.2.
Source
|
Total Samples
|
Total DEGS
|
Total Nuclear-Encoded Mitochondrial (NEM) DEGs
|
NEM DEGs as Percentage of Total DEGs
|
NEM DEGs as Percentage of NEM GO Annotations
|
NHBE
|
9
|
2840
|
313
|
12.7%
|
17.0%
|
A549 (ACE2)
|
6
|
3265
|
293
|
9.0%
|
16.1%
|
Calu-3
|
6
|
4219
|
455
|
10.8%
|
24.7%
|
BALF
|
5
|
5353
|
411
|
7.7%
|
22.3%
|
Lung
|
10
|
475
|
28
|
5.9%
|
1.6%
|
SARS-CoV-2 Differentially Regulates mtDNA-Encoded Genes
We hypothesized that SARS-CoV-2 infection would upregulate mtDNA-encoded gene expression due to the effects of SARS-CoV on patient PBMCs.12 In our analyses, however, we observed minimal regulation of mtDNA-encoded genes after SARS-CoV-2 infection and – against our original hypothesis – downregulation of mtDNA-genes in BALF (Figure 1). SARS-CoV-2 infection only upregulated mt-CytB and downregulated mt-TN in primary NHBE cells, whereas IAV and IAVdNS1 (i.e., IAV with a null interferon antagonist NS1 mutant) downregulated every mtDNA-encoded protein gene and several tRNAs (Figure 1A). Such strong regulatory effects of IAV on mtDNA-encoded genes was not consistent in cancerous cell lines. In A549 cells, while IAV downregulated mt-ND6 and mt-ATP6, we did not observe the same global downregulation as we did in primary NHBE cells, and we found 16S rRNA was upregulated only in these cancerous cells (Figure 1B). HPV did not regulate expression of any mtDNA-encoded protein gene (Figure 1B). RSV induced dramatic upregulation of every mtDNA-encoded protein along mt-rRNA and a few tRNAs (Figure 1B). SARS-CoV-2 did not upregulate any mtDNA-encoded proteins but did upregulate 16S rRNA in Calu-3 and ACE2-expressing A549 cells (Figure 1B). In BALF, SARS-CoV-2 surprisingly downregulated nearly every mtDNA-encoded gene along with several mt-tRNAs (Figure 1C). Overall, following SARS-CoV-2 infection, we observed downregulation of mtDNA-encoded genes in BALF, which is opposite to the minimal upregulation we observed in primary and cell lines. The complete list of significant mtDNA-encoded genes and fold changes are included in Supplementary Table: mtDNA Differentially Expressed Genes.
SARS-CoV-2 Does Not Downregulate MAVS Expression
We hypothesized that MAVS expression would not be significantly downregulated after SARS-CoV-2 infection. Indeed, there were no significant effects of SARS-CoV-2 on MAVS expression in ACE2-expressing A549 cells, Calu-3 cells, NHBE cells, BALF, and lung (Figure 2A). In contrast, IAV, RSV, and HPIV all induced a statistically significant downregulation of MAVS. IAV-infected A549 cells induced the most dramatic downregulation of MAVS (Log2FC = -0.98; Padj = 1.32E-08), followed by IAVdNS1 in NHBE cells (Log2FC = -0.93; Padj = 5.35E-04), IAV in NHBE cells (Log2FC = -0.52; Padj = 1.11E-01), RSV in A549 cells (Log2FC = -0.33; Padj = 7.00E-02) and HPIV in A549 cells (Log2FC = -0.20; Padj = 2.02E-01).
NEMS Sufficiently Classifies SARS-CoV-2
Given that MAVS and mtDNA-encoded gene expression differs among SARS-CoV-2, HPIV, RSV, and IAV, we hypothesized that the global NEM signature would sufficiently classify SARS-CoV-2. Therefore, we conducted a principal component analysis (PCA) exclusively on an NEM-extracted gene set. As expected, the first two principal components sufficiently reduced NEM expression variance in a manner that classified SARS-CoV-2 in primary cells, cell lines, and clinical samples (Figure 3). The amount of variance that the first two NEM-specific two principal components explain total 81%, 60%, and 56% for primary cells, cell lines, and clinical samples, respectively.
SARS-CoV-2 Specific NEM-Enriched Pathways
Since we showed that NEMs are sufficient to classify SARS-CoV-2, HPIV, RSV, and IAV, we attempted to unravel the biological processes that these NEMs modify. All NEMs were extracted from the complete list of statistically significant DEGs. Hierarchical clustering of NEMs showed distinct signatures by viral infection (Figure 4A). We then then conducted gene enrichment analyses by inputting this set of significant NEMs against a universe background of all total significant DEGs. Four separate gene enrichment analyses were conducted (i.e., primary cells, cell lines, BALF, and lung).
The GO enriched terms of small molecule metabolism, phosphorus metabolism, oxidation-reduction, and cellular amide metabolism were all shared between SARS-CoV-2 and IAV in NHBE primary cells (Figure 4B; filtered by >20% NEM within gene set). SARS-CoV-2 particularly induced greater enrichment for mitochondrion organization and catabolism compared to IAV and IAVdNS1 (Figure 4B). Furthermore, the top 10 most significant enriched for SARS-CoV-2 mapped back to mitochondrial translation, mitochondrial organization, and cellular respiration (Figure 4C and 4D). SARS-CoV-2 not only induced global downregulation of the metabolic pathways shown in Figure 4C (Log2FC), but the degree of downregulation was greatest in SARS-CoV-2, as illustrated in the right panel of Figure 4C with hierarchical clustering scores colored (SARS-CoV-2 Score). Several mitochondrial ribosome protein genes (e.g., MRPL55, MRPL47, MRPL42, etc.) and Complex 1 related genes (e.g., NDUFB11, NDUFB2, NDUFC1, etc.) were expressed much less after SARS-CoV-2 compared to IAV and IAVdNS1 (Figure 4C and 4D).
In cell lines, as we noted in primary cells, enrichment for oxidation-reduction metabolism was shared among all viral infections (i.e., SARS-CoV-2, HPIV, RSV, and IAV). Hierarchical clustering showed distinct NEM signatures by viral infection (Figure 5A). For SARS-CoV-2 in ACE2-expressing A549 cells, we observed more NEMs involved in catabolism and small molecule metabolism, but we did not observe this same degree of enrichment in Calu-3 cells (Figure 5B). In addition, Calu-3 cells contained more NEMs enriched for mitochondrion organization than in ACE2-expressing A549 cells. We still observed reduced expression of mitochondrial Complex 1 related genes (e.g., NDUFS2, NDUFB7, NDUFS6, etc.) after SARS-CoV-2 compared to IAV, HPIV, and RSV IAVdNS1 (Figure 5D). Notably, unlike the downregulation of mitochondrial ribosome protein genes in primary cells, we did not observe similar enrichment for mitochondrial translation after SARS-CoV-2 infection. Instead, a greater number of differentially expressed genes related to carboxylic acid metabolism (FASN, ACAT1, ACAT2, etc.) were observed in cell lines after SARS-CoV-2 infection.
The top 10 most significant enriched processes after SARS-CoV-2 in ACE2-expressing A549 cells mapped back to cellular respiration, oxidation-reduction, small molecule metabolism, among many other interrelated metabolic pathways (Figure 5C). As we observed in primary NHBE cells, we found similar downregulation of genes mapping back to cellular respiration and mitochondrion organization after SARS-CoV-2 infection. Additionally, while we observed downregulation of catabolic genes after SARS-CoV-2 infection, the degree of downregulation relative to other viruses was not as great (i.e., some NEMs were expressed at greater levels after SARS-CoV-2 compared to other viruses), which is in conflict to that observed in primary cells (i.e., Figure 5C SARS-CoV-2 Score shows positive scores for genes that are downregulated by log 2 fold change). We also observed greater expression of PINK1 after SARS-CoV-2 infection in ACE2-expressing cells, suggesting differential mitochondrial dynamics specific to cell lines. Upregulation of PINK1 could explain why we a greater catabolic gene set percentage in SARS-CoV-2 compared to other viruses. Overall, we also observed attenuation of cellular respiration, oxidation-reduction, and related interconnected pathways in cell lines after SARS-CoV-2 infection relative to IAV, HPIV, and RSV.
BALF and lung clinical samples also contained downregulation of genes involved in cellular respiration and Complex 1 assembly after SARS-CoV-2 infection (BALF: NDUFAF6, NDUFB9, NDUFV2, etc.; Lung: NDUFB1, NDUFB7, NDUFAL2). However, the effect of SARS-CoV-2 on the NEM transcriptome was not as dramatic in lung compared to BALF. In BALF, the most significant NEM-enriched included orhanophosphate metabolism, mitochondrial gene expression, cellular respiration, oxidation-reduction, etc. (Figure 6A-B). The majority of these genes downregulated after SARS-CoV-2 infection, which was similar to the signature in primary cells and cell lines. In lung, despite identifying just 28 significant NEMs, mitochondrion organization, phosphorus metabolism, and overall energy metabolism were enriched (Figure 6C-D).
Generally, across cellular and clinical samples, several metabolic pathways were enriched after viral infection, but the SARS-CoV-2-specific signature included downregulation of NEMs involved in cellular respiration and Complex 1 assembly (NDUF family of proteins). Mitochondrial ribosome gene expression was particularly downregulated after SARS-CoV-2 infection in primary cells (greater downregulation compared to IAV) and clinical samples, although we did not observe this similar mitochondrial translation signature in cell lines. There were clear tissue and cell-specific differences across all analyses related to oxidation-reduction, small molecule metabolism, and carboxylic metabolism, suggesting SARS-CoV-2 may affect metabolism differently per cell type.