Altered high-density lipoprotein composition and functions during severe COVID-19

Abstract

Coronavirus disease 2019 (COVID-19) pandemic is affecting millions of patients worldwide. The consequences of initial exposure to SARS-CoV-2 go beyond pulmonary damage, with a particular impact on lipid metabolism. Decreased levels in HDL-C were reported in COVID-19 patients. Since HDL particles display antioxidant, anti-inflammatory and potential anti-infectious properties, we aimed at characterizing HDL proteome and functionality during COVID-19 relative to healthy subjects. HDLs were isolated from plasma of 8 severe COVID-19 patients sampled at admission to intensive care unit (Day 1, D1) at D3 and D7, and from 16 sex- and age-matched healthy subjects. Proteomic analysis was performed by LC-MS/MS. The relative amounts of proteins identified in HDLs were compared between COVID-19 and controls. apolipoprotein A-I and paraoxonase 1 were confirmed by Western-blot analysis to be less abundant in COVID-19 versus controls, whereas serum amyloid A and alpha-1 antitrypsin were higher. HDLs from patients were less protective in endothelial cells stiumalted by TNFα (permeability, VE-cadherin disorganization and apoptosis). In these conditions, HDL inhibition of apoptosis was blunted in COVID-19 relative to controls. In conclusion, we show major changes in HDL proteome and decreased functionality in severe COVID-19 patients.

Figure 1

Relative abundance of proteins identified in HDLs isolated from of the plasma of control subjects (n = 15) and COVID-19 patients (n = 8) at admission in ICU (D1). The relative abundance of proteins in HDLs from controls and D1 COVID-19 patients was expressed as the Log2 Abundance ratio (D1 COVID/Control), as described in the “Methods” section.

Figure 2

Percentage of protein abundance in HDLs from controls and COVID-19 patients. HDLs were isolated from the plasma of 15 controls and 8 patients with COVID-19 at admission (D1). Bar graphs represent the percentage of intensity ± SD of the two groups. Unpaired t-test was used. *p < 0.0332, **p < 0.0021, ***p < 0.0002 and ****p < 0.0001 as compared to the control group.

Figure 3

Western blot analysis for detection of SAA and PON-1 in HDL isolated from COVID-19 patients and controls. HDLs (3 μg protein /lane) were immunoblotted for SAA (serum amyloid A) (A) and PON-1 (paraoxonase 1) (B). (A) HDLs isolated from patients and controls are presented individually (C1+ = ICU-COVID-19 positive patient #1, C2+ for patient #2, etc.) at day 1, 3 and 7 after admission (D1, 3 and 7). Controls 1 and 1′ correspond to HDLs isolated from healthy subjects sex- and age-matched to patient #1. (B) Pools of LDLs and HDLs were blotted for PON-1 (left panel) after isolation from non-infected healthy subjects ("Controls"), COVID-19 positive patients in intensive care unit ("C+ICU") (n = 8) and SARS-CoV-2 seropositive cured caregivers ("sero+", sampled > 15 days after symptoms, n = 7). A pool of control HDLs containing PON-1 was incubated with elastase (El, 20 μg/mL), plasmin (Pn, 500 μg/mL) or both proteases (El + Pn) for 2 h at 37 °C. Enzymes alone (without HDLs) treated in the same experimental conditions were loaded as control. Uncropped gels are presented in Supplementary information file.

Figure 4

Real-time monitoring of HUVEC barrier dysfunction in response to TNFα+/−HDLs. Human umbilical vein endothelial cells (25,000 cells/well) were seeded on gelatin-coated xCELLigence 16 well E-plates. HDL were isolated from plasma of 16 controls and 8 D1 COVID ICU patients by ultracentrifugation and pooled into two groups: “Control HDL" and “COVID+ HDL”. The cell index was recorded continuously until confluence, when a plateau was reached [indicated cell growth in (A)]. Cells were then serum-deprived (- Fetal Calf Serum, FCS) for 6 h and then stimulated with 30 ng/mL TNFα+/−HDLs from COVID-19 patients or controls at 0.2 mg/mL. Cell indexes were normalized after stimulation [as indicated by the vertical line, in (A), 30 min. after the stimulation]. A magnification showing TNFα, TNFα + COVID + HDL and TNFα + control HDL is presented in (B). Representative results from 2 independent experiments. HDLs isolated from each patient and matched controls (2 controls/patients) were analyzed separately (Supplementary Fig. 4A).

Figure 5

Effect of HDLs from controls and COVID-19 (C+) subjects on HUVEC apoptosis cells after stimulation with 30 ng/mL TNFα for 24 h. HDL were isolated from plasma of 16 controls and 8 D1 COVID-19 subjects by ultracentrifugation and pooled into two groups respectively “Control HDL" and “COVID+ HDL” Upper panel: immunostaining for VE-Cadherin (red) and nuclear staining with DAPI (blue). Empty white arrowheads show VE-Cadherin re-organization. Plain white arrowheads show examples of apoptotic nuclei (condensed and/or fragmented).

Figure 6

TUNEL and DAPI staining- Quantification of apoptotic nuclei. In TNFα-treated HUVECs (30 ng/mL 24 h), dual staining shows a good correlation between morphological apoptotic features determined by nuclear condensation and fragmentation (DAPI in blue, A, C) and DNA fragmentation (green, B, C). Bar graph (D) represents the means ± SD of apoptotic nuclei count in 5 fields (between 700 and 900 nuclei counted) of a representative experiment. *p = 0.0332 (Control vs TNFα). **p = 0.002 (Control vs TNFα + COVID+ HDL). #p = 0.03 (TNFα + control HDL vs TNFα + COVID+ HDL).