SARS-CoV-2-induced dysregulation in ADAR editing patterns persists post viral clearance in individuals with mild COVID-19

Figures

• 

  Figure 1. Expression dynamics of ADAR genes, in transcripts per million (TPM), across three distinct stages of SARS-CoV-2 infection: pre-, mid-, and post-infection. (A) ADAR1 (ADAR) expression increased significantly mid-infection compared to pre-infection (paired t-test, p value = 1.7 x 10-6). However, this increase in expression is not retained in the post-infection stage, and no significant difference in ADAR expression is seen between pre- and post-infection samples (paired t-test p value = 0.96). (B) ADAR2 (ADARb1) expression increased mid-infection (median TPM 8.309; paired t-test, p value = 0.07) and decreased minimally post-infection (median TPM 4.958; paired t-test, p value = 0.22) compared to pre-infection (median TPM 6.991), although the observed differences did not achieve statistical significance in both comparisons. (C) ADAR3 is minimally expressed across all stages of infection and showed no significant change mid- (paired t-test, p value = 0.36) or post-infection (paired t-test, p value = 0.5) compared to pre-infection. Effect size (Cohen’s d) was large (-0.824) only for ADAR pre- versus mid-infection comparison; same comparison for ADARB1 had a small effect size, with d =…

• 

  Figure 2. Changes in expression of ADAR isoforms, ADARp110 and ADARp150 as DESeq2 normalized counts, across mid-, and post-infection, compared to pre-infection. (A) ADARp110 expression is significantly increased mid infection compared to pre-infection (paired t-test, p value = 0.006), although this increase in expression is not retained post-infection, where no significant change is observed in the expression of ADARp110 when compared to pre-infection (paired t-test, p value = 0.98). (B) Transcript level analysis of ADARp150 showed that expression increased mid- infection compared to pre-infection, although statistical significance was not achieved (paired t-test, p value = 0.31). ADARp150 expression showed no change post-infection compared to pre-infection (paired t-test p value = 0.95). Effect sizes (Cohen’s d) ranged from small (-0.43 for pre- versus mid-infection ADARp110 comparison) to negligible (d values of 0.00433, -0.154, and -0.0092 for ADARp110 pre- versus post-infection, and ADARp150 pre- versus mid-, and pre- versus post-infection comparisons, respectively).

• 

  Figure 3. Changes in the number of ADAR editing sites including within individual changes in total number of putative ADAR edits, number of unique ADAR edits, distribution of unique ADAR edits across genic regions, and pathway enrichment analysis of genes incorporating unique edits. (A) Total number of putative ADAR edits varied within each individual mid- and post- infection relative to respective pre-infection values. (B) The upset plot shows the number of consistently and uniquely edited sites across pre-, mid-, and post-infection stages of infection. The plot also shows the number of sites shared between different infection stages. (C) Unique editing sites by genic regions and (D) Pathways overrepresented among genes incorporating unique editing sites pre-infection, mid-infection, and post-infection.

• 

  Figure 4. Alterations overall ADAR editing levels mid- and post- infection compared to pre-infection. Panel A and B show alterations in overall ADAR editing levels at cohort and individual specific levels respectively. Each bar in panel B represents an individual and values for mid-infection (red bars) and post-infection (blue bars) are represented relative to pre-infection overall ADAR editing values. (A) Overall ADAR editing levels showed statistically significant decrease mid-infection compared to pre-infection stage (paired t-test, p-value 0.011) and although overall ADAR editing levels remained decreased post -infection compared to pre-infection, the difference was not statistically significant (paired t-test, p value = 0.14). (B) Decreased levels of overall ADAR editing mid-infection compared to pre-infection was generally consistent across each patient. However, individual specific differences in overall editing levels are observed post-infection, where some individuals retained the dysregulated levels of overall ADAR editing while in others overall ADAR editing levels returned to around pre-infection levels post-infection.

• 

  Figure 5. Differences in overall and site-specific ADAR editing within the two groups of individuals identified using K-means algorithm (A) Group 1, which include individuals identified by K-means clustering to retain dysregulated editing, show statistically significant differences in overall ADAR editing levels post- compared to pre-infection (paired t-test, p value=0.005). (B) No significant differences in overall ADAR editing levels were identified in group 2, which include individuals identified by K-means for whom editing returns to pre-infection levels (paired t-test, p value=0.74). Scatter plot in panel C and D show differences in site-specific editing levels, each data point represents an editing site, and the reference line indicates equal editing level pre- and post-infection (C) Analysis of site-specific editing levels for high confidence ADAR editing sites within group 1 individuals revealed marked differences, with most sites showing reduced editing levels post-infection. (D) In contrast, site specific editing level remains largely unchanged for high confidence editing sites within group 2 individuals.