SARS-CoV-2 infects adipose tissue in a fat depot- and viral lineage-dependent manner

Abstract

Visceral adiposity is a risk factor for severe COVID-19, and a link between adipose tissue infection and disease progression has been proposed. Here we demonstrate that SARS-CoV-2 infects human adipose tissue and undergoes productive infection in fat cells. However, susceptibility to infection and the cellular response depends on the anatomical origin of the cells and the viral lineage. Visceral fat cells express more ACE2 and are more susceptible to SARS-CoV-2 infection than their subcutaneous counterparts. SARS-CoV-2 infection leads to inhibition of lipolysis in subcutaneous fat cells, while in visceral fat cells, it results in higher expression of pro-inflammatory cytokines. Viral load and cellular response are attenuated when visceral fat cells are infected with the SARS-CoV-2 gamma variant. A similar degree of cell death occurs 4-days after SARS-CoV-2 infection, regardless of the cell origin or viral lineage. Hence, SARS-CoV-2 infects human fat cells, replicating and altering cell function and viability in a depot- and viral lineage-dependent fashion.

Fig. 1. SARS-CoV-2 infects human adipose tissue.

Thoracic subcutaneous adipose tissue samples of individuals who died of COVID-19. a Representative immunofluorescence images of 6–8 frames of two individuals. DAPI, blue. SARS-CoV-2 spike protein, red. Perilipin 2, green. The dashed box marks an area of an overlay with a concentration of spike labeling. The bottom panels are the overlay with the bright field. Scale bar = 50 µm. b–d Copy number of the SARS-CoV-2 genome normalized by 10⁶ copies of the endogenous RNase P gene as evaluated by RT-qPCR in the adipose tissue biopsies (n = 23) (b) and stratified by sex (n = 23) (c), and body mass index (BMI) (n = 16) (d). Violin plots show median, first and third quartiles of the means related to genomic copies, where each dot represents a biological sample. Two-tailed Student’s t-test and one-way ANOVA was applied in (c) and (d), respectively. No significant differences were found. e–g Correlation between copy number of the SARS-CoV-2 genome (normalized by 10⁶ copies of the endogenous RNase P gene) and age (n = 18) (e), body weight (n = 17) (f) and BMI (n = 16) (g). Only individuals that exhibited detectable levels of SARS-CoV-2 RNA in adipose tissue and had information about age, body weight, or BMI were used in the analysis (represented by dots in the correlation plot). Pearson correlation analysis was applied and no significant correlation was found. Source data are provided as a Source Data file and donor characteristics are shown in Supplementary Table 1 as averages. These experiments were repeated once with similar results.

Fig. 2. The degree of adipose tissue cell infection depends on the anatomical origin of the cells and the SARS-CoV-2 lineage.

a Human adipose-derived stromal-vascular cells (hADSCs) were obtained from abdominal subcutaneous (SAT) and omental visceral (VAT) adipose tissues, differentiated in adipocytes (Sub AD and Vis AD, respectively), and infected (INFEC) or not (MOCK) with the SARS-CoV-2 ancient strain [CoV-2 (B)] or the P.1 variant [CoV-2 (P.1)]. Created with Biorender.com. b Cells were infected with MOI = 0.1 for 1 h and harvested 3 days after infection to allow better contrast and resolution of the replicating viruses. Representative immunofluorescence of six to ten images acquired from two to six independent pools of cells from two to three donors. SARS-CoV-2 spike protein, magenta. Double-stranded RNA (dsRNA), green. LipidTOX, red. Scale bar = 50 µm (zoom = 10 µm). c, d Cells were harvested 24 h after infection with MOI = 1 for 1 h to assess their susceptibility to viral infection under conditions where virus availability is not limiting. c Viral load determined by RT-qPCR. eFFU, equivalent to focus forming units. Three-way ANOVA was performed to determine effects of infection (P < 0.0001), viral lineage (P > 0.05), anatomical origin (P < 0.0001), and interactions between anatomical origin × infection (P < 0.0001), anatomical origin × viral lineage (P < 0.01), infection × viral lineage (P < 0.0001), and infection, anatomical origin, and viral lineage (P < 0.0001). Significant differences between groups were determined by Sidak’s multiple comparison test and depicted as: a, P < 0.0001 vs. MOCK; b, P < 0.0001 vs. Sub AD; c, P < 0.0001 vs. CoV-2(B); d, P < 0.0001 vs. Vis AD. d Plate forming units (PFU) in the supernatant. Two-way ANOVA was performed to determine the effects of anatomical origin (P < 0.0001), viral lineage (P > 0.05), and the interaction between anatomical origin and viral lineage (P > 0.05). Significant differences between groups were determined by Tukey’s multiple comparison test and depicted as: b, P < 0.01 vs. Sub AD. Data were mean ± SEM of two to three independent pools of cells from two to three donors. Vero cells were used as reference. Different donors are distinguished by circles, squares, and triangles, while independent pools of cells from the same donor (or Vero) are marked by the same symbol. Source data are provided as a Source Data file. These experiments were repeated once with similar results.

Fig. 3. Visceral adipose tissue cells express more ACE2 than subcutaneous adipose tissue cells.

Human adipose tissue-derived stromal-vascular cells from subcutaneous and visceral fat depots were differentiated into adipocytes (Sub AD and Vis AD, respectively). a Representative immunofluorescence of Vis AD and Sub AD cells infected with CoV-2(B) (MOI = 0.1 for 1 h and harvested 3 days after infection) evidencing ACE2 expression. DAPI, blue. ACE2, green. LipidTOX, red. Scale bar = 50 µm. Representative of eight to ten images acquired from two to five independent pools of cells from two to three donors. b Western blotting comparing the basal levels of ACE2 in uninfected Sub AD (n = 3 donors) and Vis AD (n = 2 donors) cells. Two-tailed Student’s t-test: b, P < 0.01. MW, molecular weight. c–e RT-qPCR showing basal relative expression of genes related to SARS-CoV-2 entry in uninfected Sub AD and Vis AD cells. One-way ANOVA was applied in order to detect the effect of anatomical origin on c ACE2 (P < 0.0001), d NRP1 (P < 0.0001), and e TMPRSS2 (P > 0.05) expression. Tukey’s multiple comparison test was applied to detect differences between the groups. Significant differences are depicted in letters: b, P < 0.01 vs. Sub AD; d, P < 0.01 vs. Vis AD; e, P < 0.0001 vs. Vero. Data represent the mean ± SEM of experiments performed using three independent pools of cells from two to three donors. Vero cells were used as a reference. Different donors are distinguished by circles, squares, and triangles, while independent pools of cells from the same donor (or Vero) are marked by the same symbol. Source data are provided as a Source Data file. The experiments were repeated once with similar results, with the exception of c–e, which were not repeated; although the calculated interassay variability was low, the controls worked as expected, and the biological and technical variability was accounted for in the analysis; moreover, the main findings were corroborated by a, b using other approaches.

Fig. 4. SARS-CoV-2 infection of fat cells results in different cellular responses depending on the depot of origin and the viral lineage.

Stromal-vascular cells from human subcutaneous and visceral fat depots were differentiated into adipocytes (Sub AD and Vis AD, respectively) and infected (INFEC) or not (MOCK) with the SARS-CoV-2 ancient strain [CoV-2(B)] or the P.1 variant [CoV-2(P.1)]. a Overlap between differentially expressed proteins as determined by proteomics. Arcs represent different infection conditions (outside) or differentially expressed proteins (inside). Dark and light orange represent proteins that are differentially expressed in multiple conditions (linked by purple lines) or in only one condition, respectively. b Enriched ontology clusters. Statistically enriched terms were hierarchically clustered and the top-ranked terms within each cluster were displayed. c HSL abundance. Significant differences were determined by ANOVA and depicted as: a, P < 0.01 vs. MOCK CoV-2(B). d Western blotting of phosphorylated HSL (p-HSL), total HSL, ATGL, and α-tubulin as the loading control. Quantification represents the mean ± SEM of three independent pools from one donor. Significant differences were determined by a two-tailed Student’s t-test: a, P < 0.01. e Glycerol released in the medium. Significant differences were determined by a two-tailed Student’s t-test: a, P = 0.0018. f SARS-CoV-2 genome in the supernatant of cells treated or not with 10 μM isoproterenol (ISO). Significant differences were determined by one-way ANOVA with Tukey’s multiple comparison test and depicted as: a, P < 0.0001 vs. MOCK and b, P < 0.05 vs. INFEC[CoV-2(B)]. g Protein–protein interaction analysis using proteins related to the Reactome term “Interferon Signaling”. e, f are mean ± SEM of two to six independent pools of cells from two to three donors. Proteomics were run using samples from two to three donors in technical triplicate. Different donors are distinguished by circles, squares, and triangles, while independent pools of cells from the same donor are marked by the same symbol. Source data are provided as a Source Data file and in Supplementary Data 1 and 2. The experiments were repeated once with similar results, with the exception of (e, f), which were not repeated; although the interassay variability was low, controls worked as expected, and biological and technical variability was accounted for in the analysis.

Fig. 5. Genes associated with inflammation are differentially expressed in SARS-CoV-2-infected adipose tissue cells.

a–l Stromal-vascular cells from human subcutaneous and visceral fat depots were differentiated into adipocytes (Sub AD and Vis AD, respectively) and infected (INFEC) or not (MOCK) with the SARS-CoV-2 ancient strain [CoV-2 (B)] (a–f) or the P.1 variant [CoV-2 (P.1)] (g–l). Gene expression as quantified by RT-qPCR 24 h post-infection. Two-way ANOVA with Tukey’s multiple comparison test was applied to identify differences between groups. These differences are depicted in letters: a, P < 0.05 vs. respective MOCK; b, P < 0.05 vs. respective Sub AD. Data represent the mean ± SEM of two to three independent pools of cells from two to three donors. Different donors are distinguished by circles, squares, and triangles, while independent pools of cells from the same donor are marked by the same symbol. Source data are provided as a Source Data file. These experiments were not repeated; although the calculated interassay variability for these genes was low, controls worked as expected, and biological and technical variability was accounted for in the analysis.

Fig. 6. SARS-CoV-2 infects adipose tissue in a fat depot- and viral lineage-dependent manner.

CoV-2(B), original B SARS-CoV-2 lineage. CoV-2(P.1), SARS-CoV-2 gamma variant P.1. IFN interferon, IFNA interferon alpha gene, IL1B interleukin 1 beta gene, CCL2 C-C motif chemokine ligand 2 gene. Created with Biorender.com.