SARS-CoV-2 infection perturbs the gastrointestinal tract and induces modest microbial translocation across the intestinal barrier

Abstract

SARS-CoV-2 infects via the respiratory tract, but COVID-19 includes an array of non-respiratory symptoms, among them gastrointestinal (GI) manifestations such as vomiting and diarrhea. Here we investigated the GI pathology of SARS-CoV-2 infections in rhesus macaques and humans. Macaques experienced mild infection with USA-WA1/2020 and shed viral RNA in the respiratory tract and stool, including subgenomic RNA indicative of replication in the GI tract. Intestinal immune cell populations were disturbed, with significantly fewer proliferating (Ki67+) jejunal B cells in SARS-CoV-2-infected macaques than uninfected ones. Modest translocation of bacteria/bacterial antigen was observed across the colonic epithelium, with a corresponding significant increase in plasma soluble CD14 (sCD14) that may be induced by LPS. Human plasma demonstrated significant decreases in interleukin (IL)-6 and sCD14 upon recovery from COVID-19, suggesting resolution of inflammation and response to translocated bacteria. sCD14 significantly positively correlated with zonulin, an indicator of gut barrier integrity, and IL-6. These results demonstrate that GI perturbations such as microbial translocation can occur in even mild SARS-CoV-2 infections and may contribute to the COVID-19 inflammatory state.IMPORTANCEThis study investigates gastrointestinal (GI) barrier disruption in SARS-CoV-2 infections and how it may contribute to disease. We observed bacteria or bacterial products crossing from the colon interior (the lumen) to the lamina propria during SARS-CoV-2 infection in macaques. Bacteria/bacterial products are tolerated in the lumen but may induce immune responses if they translocate to the lamina propria. We also observed a significant increase in soluble CD14, which is associated with an immune response to bacterial products. In addition, we observed that humans recovering from COVID-19 experienced a significant decrease in soluble CD14, as well as the inflammatory marker interleukin (IL)-6. IL-6 and sCD14 correlated significantly across macaque and human samples. These findings suggest that SARS-CoV-2 infection results in GI barrier disruption that permits microbial translocation and a corresponding immune response. These findings could aid in developing interventions to improve COVID-19 patient outcomes.

Keywords: COVID-19; gut; immune response; microbiome; non-human primates; pathology.

Fig 1

SARS-CoV-2 infection of adult rhesus macaques results in viral RNA (vRNA) shedding and replication in the upper respiratory tract and gastrointestinal tract. (A) Total vRNA load in nasal swab fluid. (B) Total vRNA load in throat swab fluid. (C) Total vRNA load in stool. (D) Subgenomic vRNA (sgRNA) load in nasal swab fluid. (E) sgRNA load in throat swab fluid. (F) sgRNA load in stool. Means in black lines with open circles. Undetected samples are graphed as 0.5 while value 0 was used for averaging.

Fig 2

Mild SARS-CoV-2 infection of rhesus macaques does not induce considerable bacterial dysbiosis in the gastrointestinal tract. (A) Relative abundance of 13 bacterial phyla in stool microbial communities identified by 16S sequencing. Representative samples are shown, with additional in Fig. S3. (B) Alpha diversity (Shannon index) for stool bacterial community of each animal at baseline, day 2, and days 10 and 11 post-infection; no significant differences (Wilcoxon matched-pairs tests). (C) Principal coordinate analysis of unweighted UniFrac distances of stool bacterial communities at baseline, day 2, and days 10 and 11. Samples are not significantly separated by time point in Adonis testing. (D) Seven significantly differentially abundant taxa were identified between baseline and day 2, with positive log₂ fold change indicating increased abundance in day 2 samples.

Fig 3

Considerable dysbiosis in GI bacterial communities of hospitalized individuals with moderate/severe COVID-19 relative to recovered individuals. (A) Relative abundance of 15 bacterial phyla identified with 16S sequencing in stool or rectal swabs. Abundance plots are grouped by disease state of the individual: recovered, moderate/severe (non-fatal), and moderate/severe (fatal). (B) Alpha diversity for each bacterial community. Data from individuals with fatal infection are shown in squares with black outlines. Diversity was significantly lower in samples from moderate/severe infection than in recovery samples (Welch’s test). Within moderate/severe infection samples, diversity was not significantly different between fatal and non-fatal cases (Mann-Whitney). (C) Principal coordinate analysis of unweighted UniFrac distances of GI bacterial communities. Samples are significantly separated by disease state (Adonis testing). (D) 279 significantly differentially abundant taxa between bacterial communities of individuals with moderate/severe COVID-19 vs recovered individuals. A positive log₂ fold change indicates a greater abundance in samples from individuals with moderate/severe COVID-19. (E) SARS-CoV-2 N gene copies in stool samples from moderate/severe COVID-19. Data from individuals with fatal infection are shown in squares with black outlines. Undetected samples are graphed as 0.5.

Fig 4

Microbial translocation across colon and expansion of relative germinal center (GC) area in MLNs of SARS-CoV-2-infected animals. (A) Immunohistochemistry (IHC) images for representative uninfected (left, macaque Rh22), SARS-CoV-2-infected (center, Rh6), and SIV-infected (right, Rh13) colon tissue. DAPI stain in blue and polyclonal E. coli antibody staining in white. Microbial products translocated across the gut epithelial layer are indicated with yellow arrows and magnified in the SIV+ sample inset. (B) Median summary scores of microbial translocation in uninfected, SARS-CoV-2-infected, and SIV-infected colon tissue stained by IHC and represented in A. Group medians in horizontal black bars. No significant difference between uninfected and SARS-CoV-2-infected scores (Mann-Whitney test). (C) Representative hematoxylin and eosin (H&E) staining of MLNs from uninfected (left, Rh19) and SARS-CoV-2-infected (right, Rh2) macaques with a GC structure indicated in each image in black arrow. (D) Mean GC area as a percent of the total MLN area determined from H&E slides shown in C. Group medians in horizontal black bars. No significant difference between uninfected and SARS-CoV-2-infected scores (Mann-Whitney test).

Fig 5

Lymphocyte frequency and proliferative capacity perturbations in macaque intestinal tissues with SARS-CoV-2 infection. Lymphocytes isolated from macaque colon (A–D) and jejunum (E–H) as identified with flow cytometry. B-cell frequency (A, E) and T-cell frequency (B, F) of the parent population given for necropsy tissues of uninfected (green), SARS-CoV-2-infected (blue), and acutely SIV-infected (red) macaques. (C, G) Frequency of Ki67-positive B cells in three infection groups. (D, H) Frequency of Ki67-positive T cells in three infection groups. Mann-Whitney tests were conducted between all group pairings for each cell and tissue type, with P values < 0.1 reported.

Fig 6

Increases in plasma IL-6 and soluble CD14 (sCD14) in macaques after SARS-CoV-2 infection and decreases in humans recovered from COVID-19. (A) Plasma IL-6 in rhesus macaques at baseline and days 10 and 11 of infection. No significant difference (paired t-test). (B) Plasma IL-6 in paired human samples from within approximately 2 weeks of symptom onset and 2 months post-symptom onset. Significant decrease with recovery (Wilcoxon test). (C) Plasma IL-6 in human samples of individuals with mild or moderate/severe COVID-19. Significantly higher levels were observed in moderate/severe cases (Mann-Whitney test). (D) Scatterplot of contemporaneous plasma sCD14 and IL-6 values in human and rhesus samples. A significant correlation was observed (Spearman). (E) Plasma sCD14 in rhesus macaques at baseline and days 10 and 11 of infection. Significant difference by paired t-test. (F) Plasma sCD14 in paired human samples. Significant difference with recovery (paired t-test). (G) Plasma sCD14 in human samples of individuals with mild or moderate/severe COVID-19. No significant difference between groups (unpaired t-test). (H) Scatterplot of contemporaneous plasma sCD14 and zonulin values in human and rhesus samples. A significant correlation was observed (Spearman).