Nonhuman Primate Model of Long Covid: Immune Insights of Glycometabolic Dysfunctions

ABSTRACT

Background: A major consequence of COVID-19 is long-term metabolic complications (metabolic PASC or Long COVID) following acute disease resolution leading to hyperglycemia, increased risk of diabetes or defects in glucose metabolism. However, the mechanisms underlying the links between COVID-19 and glycometabolic disruptions remain unclear. Method(s) 15 African green monkeys (AGM;Chlorocebus aethiops) were infected with SARS-CoV-2 (Wuhan stain) and divided into two groups unvaccinated (n=10) and vaccinated (BNT162b2 (Pfizer) 4-days post infection;n=5). Subgenomic SARS-CoV-2 mRNA (sgRNA) reflecting active replication was quantified in nasal and pharyngeal swabs, and blood chemistry analysis was performed longitudinally up to 18 weeks post-infection. We quantified liver glycogen at necropsy using Periodic acid-Schiff staining. Finally, we longitudinally analyzed 96 plasma proteins using a proximity extension assay (Olink). STRING was used to identify enriched protein networks. Comparisons between the two groups over time were performed using PERMANOVA. Result(s) All animals had detectable sgRNA ( >3.64x106) at day 3, and only two were undetectable at week 5. Post-infection BNT162b2 vaccination partially inhibited the SARS-CoV-2 mediated disruption of glucose levels (P=0.001, Fig. 1A). Liver glycogen levels following necropsy correlated positively with blood glucose levels at week 12 (r=0.74, P =0.003). Histopathological analysis revealed no marked evidence of long-term inflammation or fibrosis of pancreatic islets. Using the plasma proteomic data, we identified a signature of 15 SARS-CoV-2-modulated plasma proteins coinciding with early onset hyperglycemia during acute infection (P=0.001, Fig. 1B). These proteins are enriched for biological processes linked to chemotaxis (FDR=1.38E-06), and viral protein interaction with cytokines (FDR=1.01E-12) (Fig. 1C). Of these, CCL25 and glial cell derived neurotrophic factor (GDNF) remained persistently elevated post-acute infection and correlated with blood glucose levels (r=0.57, P=0.0003;and r=0.64, P<0.0001, respectively, Fig. 1D). Conclusion(s) Our AGM model validates phenotypes of metabolic PASC and offers an opportunity to mechanistically study the manifestations of PASC. Our preliminary data suggest that vaccine-preventable early insults by metabolicregulating immune factors may contribute to long-term dysregulated liver and systemic glucose homeostasis during PASC. These immune factors warrant further investigation for their mechanistic links to PASC. (Figure Presented).