SARS-CoV-2 innate effector associations and viral load in early nasopharyngeal infection

Abstract

COVID-19 causes severe disease with poor outcomes. We tested the hypothesis that early SARS-CoV-2 viral infection disrupts innate immune responses. These changes may be important for understanding subsequent clinical outcomes. We obtained residual nasopharyngeal swab samples from individuals who requested COVID-19 testing for symptoms at drive-through COVID-19 clinical testing sites operated by the University of Utah. We applied multiplex immunoassays, real-time polymerase chain reaction assays and quantitative proteomics to 20 virus-positive and 20 virus-negative samples. ACE-2 transcripts increased with infection (OR =17.4, 95% CI [CI] =4.78-63.8) and increasing viral N1 protein transcript load (OR =1.16, CI =1.10-1.23). Transcripts for two interferons (IFN) were elevated, IFN-λ1 (OR =71, CI =7.07-713) and IFN-λ2 (OR =40.2, CI =3.86-419), and closely associated with viral N1 transcripts (OR =1.35, CI =1.23-1.49 and OR =1.33 CI =1.20-1.47, respectively). Only transcripts for IP-10 were increased among systemic inflammatory cytokines that we examined (OR =131, CI =1.01-2620). We found widespread discrepancies between transcription and translation. IFN proteins were unchanged or decreased in infected samples (IFN-γ OR =0.90 CI =0.33-0.79, IFN-λ2,3 OR =0.60 CI =0.48-0.74) suggesting viral-induced shut-off of host antiviral protein responses. However, proteins for IP-10 (OR =3.74 CI =2.07-6.77) and several interferon-stimulated genes (ISG) increased with viral load (BST-1 OR =25.1, CI =3.33-188; IFIT1 OR =19.5, CI =4.25-89.2; IFIT3 OR =245, CI =15-4020; MX-1 OR =3.33, CI =1.44-7.70). Older age was associated with substantial modifications of some effects. Ambulatory symptomatic patients had an innate immune response with SARS-CoV-2 infection characterized by elevated IFN, proinflammatory cytokine and ISG transcripts, but there is evidence of a viral-induced host shut-off of antiviral responses. Our findings may characterize the disrupted immune landscape common in patients with early disease.

Keywords: SARS-CoV-2; biomarkers; host shut-off; innate immunity; interferon.

FIGURE 1

Relationships with Age. In A: male patients were older and in B: patients without infection were older, but in C: this relationship is seen to be limited to male patients (see also Table 3). Each box‐plot includes boxes that show median, upper, and lower quartile values and whiskers or single points that show upper and lower extremes (McGill et al., 1978);p values were calculated using linear regression (Chambers, 1998).

FIGURE 2

Association of ACE2 but not TMPRSS2 Expression with SARS‐CoV‐2 Infection. A: ACE‐2 mRNA is increased approximately three‐fold in patients with SARS‐CoV‐2 infection over ACE‐2 mRNA expression in patients without infection, and B: the increase in expression is associated with viral load (OR =1.16, CI =1.1–1.23, p < 0.001). However, the expression of TMPRSS‐2 is C: neither increased nor decreased with infection and D: is not associated with viral load. Adjustments for age and sex were not significant for either molecule. In each panel, A‐D, there are six infected and six noninfected status patients. Each box‐plot includes boxes that show median, upper, and lower quartile values and whiskers or single points that show upper and lower extremes (McGill et al., 1978), and p values were calculated using linear regression (Chambers, 1998).

FIGURE 3

Transcripts for Genes Associated with Inflammation. Transcription of A: IL‐8, B: IP‐10, C: TNF‐α mRNA are all increased approximately sixfold, whereas transcription of D: TRAF‐1 is reduced. When assessed for relationship with viral load, E: decreasing TRAF‐1 mRNA is associated with increasing viral N1 protein mRNA, whereas increasing mRNA for F: IL‐8, G: TNF‐α (less strongly) and H: IP‐10 are associated with increasing viral N1 protein mRNA. In contrast, I: decreasing IP‐10 mRNA is associated with increasing age. Similar relationships are seen in viral E1 protein transcripts (not shown). In each panel, A‐C, there are 20 infected and 20 noninfected status patients. In panel D, there are six infected and six noninfected. Each box‐plot includes boxes that show median, upper, and lower quartile values and whiskers or single points that show upper and lower extremes (McGill et al., 1978), and p values were calculated using linear regression (Chambers, 1998).

FIGURE 4

Transcripts for Genes Associated with Interferons. Transcription of A: IFN‐λ1 is increased nearly sixfold although B: IFN‐λ1 protein is unchanged with infection. Transcription of C: IFN‐λ2 is increased about 3‐fold but D: IFN‐λ2 protein is decreased even if the combined measurement of IFN‐λ2 and IFN‐λ3 proteins are attributed solely to IFN‐λ2. The mRNA fold changes for E: IFN‐λ1 and F: IFN‐λ2 are directly associated with increasing viral N1 protein mRNA. For G: IFN‐λ1 protein there is no association with viral N1 protein mRNA, but for H: combined IFN‐λ2 and IFN‐λ3 protein measurement, there is an inverse relationship with viral N1 protein mRNA. For E‐H, substitution of viral E1 mRNA produced similar relationships and figures (not shown). In each panel, A‐D, there are 20 infected and 20 noninfected status patients. Each box‐plot includes boxes that show median, upper, and lower quartile values and whiskers or single points that show upper and lower extremes (McGill et al., 1978), and p values were calculated using linear regression (Chambers, 1998).

FIGURE 5

Transcripts for Pre‐selected ISGs. Infection with SARS‐CoV‐2 is associated with increased A: MX‐1, B: IFIT‐1, C: IFIT‐3, and D: Tetherin (BST‐2) mRNAs. There were significant associations between increasing E: IFIT‐3 mRNA and F: IFIT‐3 protein fold changes and increasing viral N1 protein mRNA. Protein fold‐change for IFIT‐3 was measured using DDA mass spectrometry. Similar relationships were seen using viral E1 protein mRNA as in E and F. In each panel, A‐F, there are six infected and six noninfected status patients. Each box‐plot includes boxes that show median, upper, and lower quartile values and whiskers or single points that show upper and lower extremes (McGill et al., 1978), and p values were calculated using linear regression (Chambers, 1998).