Fig 2: L-SIGN is a receptor for SARS-CoV-2 virus.(A) Representative confocal microscopy images of SARS-CoV-2 nucleocapsid protein in LSECs from a COVID-19 liver autopsy sample. An uninfected normal human liver autopsy sample was used as a negative control. LSECs were L-SIGN+ (green). Arrows mark SARS-CoV-2 (red). DAPI (gray), nuclear staining. Scale bar: 10 µm. (B) Representative of 3D rendering of Z-stacked microscopy of liver sections from COVID-19 liver autopsy samples. Scale bars: 3 µm. Arrows mark SARS-CoV-2 protein particles (red) inside LSECs. Scale bar: 5 µm. (C) Representative 3D reconstructed confocal microscopy images of sections from a COVID-19 liver autopsy sample (left). Images on the right show orthogonal xy, xz, and yz projections of the dotted area to reveal the virus staining inside the LSEC. Scale bars: 10 µm. The results represent data from biopsy of autopsy samples of 4 independent patients with COVID-19 and 3 uninfected patients. (D) Representative confocal microscopy images of cultured L-SIGN-flag–expressing LSECs infected with authentic SARS-CoV-2 (MOI = 1). Scale bar: 5 µm. Blue (DAPI), nuclear staining; green, L-SIGN; red, nucleocapsid protein. (E) A 3D microscopy view of L-SIGN-flag–expressing LSECs infected with authentic SARS-CoV-2 (MOI = 1). Scale bar: 5 µm. Green, L-SIGN; red, nucleocapsid protein. (F) Percentage of nucleocapsid protein+ cells per total cells in 4–8 ×40 fields in mock- or L-SIGN-flag–transduced LSECs. (G) Median tissue culture infectious dose (TCID50) virus titer assay of supernatants 2 days after infection in mock-, L-SIGN-flag–, or ACE2-myc3–transduced LSECs. Infected Vero-E6 cells were positive control in terms of their permissiveness for high viral productivity. (H) Percentage of nucleocapsid protein+ cells per total L-SIGN-flag–transduced LSECs in the presence of mannan (100 and 500 µg/ml) in 4–8 ×40 fields in each group. *P < 0.01; **P < 0.0001. Bars indicate the mean; error bars represent mean ± SD. All experiments were at least repeated 3 times.

Fig 3: LSECs from liver autopsy samples from patients with COVID-19 express increased levels of L-SIGN, vWF, and FVIII.(A) Representative 3D rendering of Z-stacked confocal microscopy images of liver sections from normal human and COVID-19 liver autopsy samples. Scale bars: 20 µm. Green, L-SIGN; red, vWF; blue (DAPI), nuclear staining. (B) Representative single-color or merged-color confocal images of immune-stained human liver sections. Arrow indicates central vein. L-SIGN marks LSECs. vWF (green), L-SIGN (blue), FVIII (red), DAPI (white). Scale bar: 20 µm. (C) Quantification of relative mean fluorescence intensity of vWF or FVIII associated with L-SIGN+ LSECs. **P < 0.01. For all analyses, n = 4. Bars indicate the mean; error bars represent mean ± SEM. (D) Representative confocal microscopy images of Lyve1 (red) in a COVID-19–infected liver autopsy sample. An uninfected normal human liver autopsy sample was used as a positive control. Scale bar: 10 µm.

Fig 4: L-SIGN is expressed on LSECs in human biopsy or autopsy tissue samples.(A) Representative images of immunohistochemical staining of L-SIGN (ORIGENE, TA810067) in different normal human autopsy tissues. Arrows indicate liver sinusoids. Arrowheads indicate blood vessels. No L-SIGN was detected in blood vessels. Scale bar: 30 µm. (B) Representative confocal microscopy images of immunofluorescence staining of different normal human autopsy liver samples. Arrows indicate LSECs. Arrowheads indicate blood vessels. LSECs are vWF+Lyve1+L-SIGN+. Blood vessels are vWF++Lyve1–L-SIGN–. DAPI, nuclear staining. Scale bar: 20 µm. The results represent at least 3 experiments from biopsy or autopsy samples of 3 uninfected patients.

Fig 5: L-SIGN binds to a high-mannose–type N-glycan on the spike protein of SARS-CoV-2.(A) Protein domain structures of spike/Fc (left) (numbers represent amino acids). Spike/Fc forms dimer (right). RBD, receptor-binding domain; hFc, human IgG Fc domain. (B) Percentages of high-mannose–type versus complex/hybrid–type N-glycans in control Fc and spike/Fc proteins produced from COS-7 cells determined by MALDI-TOF MS. See also Supplemental Figure 1. (C) CBB staining of SDS-PAGE gels of COS-7 cell–derived spike/Fc under reducing conditions, with or without PNGaseF or Endo-H treatment (top). Immunoblotting of spike/Fc produced from COS-7 and HEK293T cells with or without PNGaseF or Endo-H treatment (bottom). (D) Silver staining of SDS-PAGE gel of COS-7 cell–derived purified SpikeN343Q/Fc protein or of WT spike/Fc treated with or without PNGaseF. Immunoblots and GNL lectin blots of WT and SpikeN343Q/Fc. The reduced size of SpikeN343Q/Fc protein relative to that of WT indicates the loss of an N-glycan on the RBD. (E) Representative flow cytometry plots of spikeN343Q/Fc binding to mock-, ACE2-myc3–, and L-SIGN-flag–transfected HEK293T cells (left) and quantification of the percentage of Fc protein binding (right). Dashed boxes indicate positive population. (F) Representative flow cytometry plot of spike/Fc binding to L-SIGN-flag–transfected HEK293T cells with the pretreatment of isotype IgG or anti–L-SIGN mAb (left) and quantification of the percentage of Fc protein binding (right). Isotype, isotype IgG pretreated; mAb, anti–L-SIGN mAb pretreated. (G) Representative flow cytometry plots of L-SIGN/Fc binding to spike protein–transfected HEK293T cells (left) and quantification of the percentage of Fc protein binding (right). Mannan (100 µg/ml), a mannose polymer, was used as a competitive inhibitor of L-SIGN/Fc. L-SIGNQPD/Fc is a carbohydrate recognition domain mutant of L-SIGN/Fc. EGTA (1 mM) was used as Ca2+ chelator. All experiments were repeated at least 3 times. For all analyses, n = 3. Bars indicate the mean; error bars represent mean ± SEM. Significance was calculated using a 1-way ANOVA for multiple groups: **P < 0.01.