Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) explosively spreads and has clinical manifestations ranging from asymptomatic infection to respiratory failure and even death1,2,3. SARS-CoV-2 is a novel virus belonging to the Beta coronavirus genus and exhibits high similarities to another two coronaviruses, severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV), which have caused large-scale outbreaks over the past two decades4,5. Previous studies found that the SARS-CoV-2 lifecycle commences with classic binding of the spike (S) protein to its cognate receptor on the surface of the host cell, human angiotensin-converting enzyme 2 (hACE-2). The cleavage of the S1/S2 site by the membrane protease serine 2 (TMPRSS2) and virus–cell membrane fusion mediated by lysosomal cathepsin L determine the efficiency of virus entry6,7,8,9. Single-cell sequencing data exhibit that the hACE-2 receptor is expressed in less than 1% cells in certain tissues at low expression levels, including heart, liver, brain, lung, and trachea, yet SARS-CoV-2 RNA may still be detected in these organs10,11,12,20. Current antibody therapies are divided into anti-viral and anti-inflammatory treatments. Among antibody options, convalescent plasma (CP) treatment is receiving significant attention, which may provide patients with immediate passive immunity21,22. However, CP therapy is suboptimal and fails to reverse respiratory failure and reduce mortality23,24. Another promising treatment option was monoclonal antibodies designed to mainly target the S protein of the virus membrane or the hACE-2 receptor of the host cell plasma membrane, thereby preventing viral binding with its receptor. To date, at least eight antibody candidates targeting the S protein have entered different stages of clinical studies20. The LY-CoV555 antibody from Lilly was the first neutralizing antibody to receive FDA emergency use authorization for the treatment of COVID-19. In a phase II trial, LY-CoV555 appeared to accelerate the natural decline in viral load at day 2 in outpatients diagnosed with mild or moderate COVID-19, but did not have the same effect in patients with severe COVID-19 or those with prolonged illness25,26. However, LY-CoV555 exhibited an unsatisfactory therapeutic effect on SARS-CoV-2 variant B.1.1.727. In particular, the “cocktail antibodies” BRII-196 plus BRII-198, developed against variants, were stopped early due to lack of utility28. Appearance of spontaneous mutations in SARS-CoV-2 is the main reason for the unsatisfactory effect of neutralizing antibodies targeting the virus. SARS-CoV-2 variants may dampen the efficacy and specificity of antibodies and further lead to new viral strains that may gradually develop resistance to existing antibodies29,30. Thus, the problem of SARS-CoV-2 esca** from antibodies needs much effort to be solved.

Results

The SARS-CoV-2 envelope protein induces extracellular vesicles containing virus particles

Our previous research has demonstrated that the SARS-CoV-2 structural envelope (2-E) protein forms a type of pH-sensitive cation channel, and that heterogeneous expression of 2-E channels causes host cell death44,45. Multiple characteristic visual fields for CoV-2-EVs were captured. First, there were numerous dense virus particles and mitochondria in a shedding CoV-2-EV. Second, a large number of SARS-CoV-2 virions were encapsulated in the shed CoV-2-EVs (Fig. 3a, yellow dotted circles). These virions displayed an average diameter of 75 ± 10 nm, consistent with previous reports43. To further verify the presence of SARS-CoV-2 virions in CoV-2-EVs, we performed immunohistochemistry analyses using SARS-CoV-2 nucleocapsid immunogold labeling. SARS-CoV-2-infected (MOI = 1) Vero E6 cells showed strong labeling for the nucleocapsid in the cytosol and in viral particles that accumulated intracellularly (Fig. 3b). In shed CoV-2-EVs, in addition to cellular contents, viral particles were visible and marked by gold particles (Fig. 3b, yellow arrows). Plaque-reduction assay and qRT-PCR also supported that these CoV-2-EVs contained a large number of infectious viruses, as high as 2.3 × 107 PFU/mL and 2 × 109 viral copies/mL (Fig. 3c–e). We noticed that the virial copies in the 2-E-EVs were higher than those in the virus-induced EVs, implying that redundant 2-E proteins may facilitate virus production, packaging, or EV secretion.

Fig. 3: CoV-2-EVs contain a large amount of virus particles.
figure 3

a TEM images of Vero E6 cells infected with SARS-CoV-2. Blue triangles, double-membrane vesicles (DMVs); red triangles, SMVs; yellow dotted circles, virus particles; red arrow, damaged mitochondria; blue dotted circle, lysosomes (MOI = 1). b Immunoelectron microscopy analyses of infected Vero E6 cells using SARS-CoV-2 nucleocapsid immunogold labeling (MOI = 1). Yellow arrows, virus particles. c Schematic representation of infection experiments. Vero E6 cells were infected with SARS-CoV-2. On the next day, the cells, supernatant, and CoV-2-EVs were collected. d qRT-PCR analysis of SARS-CoV-2 RNA in supernatant and isolated EVs from SARS-CoV-2-infected Vero E6 cells. e Plaque reduction assay of CoV-2-EVs. Data are the means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; unpaired Student’s t-test.

Full size image

SARS-CoV-2-induced vesicles help viruses escape from neutralizing antibodies and establish a productive infection

Some earlier studies reported that EVs, such as exosomes can protect viruses from the neutralizing antibodiesVirus loading and qRT-PCR analysis

The cells, supernatants, and CoV-2-EVs were collected as described above. Cell total RNA was extracted using Trizol (Invitrogen, USA). Cell supernatants and CoV-2-EV RNA were isolated with MiniBEST Viral RNA/DNA Extraction Kit (Takara, Japan) as described in the instruction, and cDNA was transcribed with PrimeScriptTM RT reagent Kit with gDNA Eraser (Takara, Japan). In detail, 50 μL supernatant was collected for RNA isolation with a MiniBEST Viral RNA/DNA Extraction Kit Ver.5.0 (Takara, AK41820A), and the total RNA was eluted with 30 μL RNase-free water. cDNA was transcribed from 3 μL total RNA in 20 μL reaction system with PrimeScriptTM RT reagent Kit with gDNA Eraser (Takara, AK71648A). Viral copies were quantified from 1 μL cDNA template viral cDNA by a standard curve method on ABI 7500 (Takara TB Green Premix Ex Taq II, AK81975A) with a pair of primers targeting S gene. The forward primer (5′-3′) is: CAATGGTTTAACAGGCACAGG; the reverse primer (5′-3′) is: CTCAAGTGTCTGTGGATCACG. The standard curve was set from six points in 20 µL reaction system (2.35 × 109 copies, 2.35 × 108 copies, 2.35 × 107 copies, 2.35 × 106 copies, 2.35 × 105 copies, 2.35 × 104 copies).

Statistical analysis

All measurements were derived from distinct samples. Statistics were performed in GraphPad Prism. Statistical significance was determined by one-way ANOVA and two-way ANOVA followed by pairwise Student’s t-test with Tukey’s or Sidak’s correction. Two-tailed unpaired Student’s t-test was performed if only two conditions were compared. Data in the text are presented as means ± SEM. Adjusted P values are reported in the figures.