Background
Long COVID (LC) is associated with dysregulated adaptive immune function and inflammation, which may originate from or contribute to the persistence of SARS-CoV-2 in post-acute COVID-19. COVID-19 can also induce adaptive NK cell responses during the acute phase, but our understanding of NK cell phenotypes in the post-acute phase and their relationship to LC is unknown.
Methods
Using PBMCs cryopreserved 4 months after initial SARS-CoV-2 infection, we studied individuals with LC (n=21), defined as ongoing COVID-attributed symptoms >3 months post-COVID, in comparison to matched comparators who had fully recovered from COVID-19 (n=14). All samples were collected in the pre-Omicron era, prior to confounders like vaccination or reinfection. Most participants (74%) had been outpatients with mild/moderate COVID-19. We performed NK cell phenotyping using high-dimensional spectral flow cytometry and examined associations between NK cell parameters and clinical factors.
Results
The frequencies of CD56+/CD3- NK cells were similar across groups. However, in comparison to those who fully recovered, the mean percentage of mature, cytotoxic CD56dim/CD16+ NK cells was lower in those with LC (69% vs. 57.9%; P=0.037). An even lower percentage was seen in those with severe LC (>5 symptoms; 50.6%; P=0.004), and in participants reporting neurocognitive or gastrointestinal symptoms (P=0.037, P=0.04, respectively). Overall, the percentage of CD56dim/CD16+ NK cells negatively correlated with the number of LC symptoms (Spearman r=-0.5, P=0.022). Compared to those who recovered, participants with severe LC also had lower percentages of CD56dim NK cells expressing KIR2D(L1/S1/S3/S5) (23.7% vs 16.1%; P=0.036). In multidimensional analyses, we did not find significant differences in NK cell UMAP clusters with adaptive NK cell features, as have been described in acute COVD-19, between those with and without LC. However, we did observe a significant increase in percentage of adaptive NK clusters in CMV IgG+ participants versus IgG- participants (P=0.024).
Conclusions
We observed a lower percentage of NK cells of the canonical mature, cytotoxic phenotype in people with LC compared to those who recovered. Reduced cytotoxic killing and impaired clearance of SARS-CoV-2 infected cells could be related to the development of LC. Alternatively, these data could suggest chronic tissue redistribution in response to persistent viral triggers in people with LC.
Table or Figure
