Supersulphides provide airway protection in viral and chronic lung diseases

Abstract

Supersulphides are inorganic and organic sulphides with sulphur catenation with diverse physiological functions. Their synthesis is mainly mediated by mitochondrial cysteinyl-tRNA synthetase (CARS2) that functions as a principal cysteine persulphide synthase (CPERS). Here, we identify protective functions of supersulphides in viral airway infections (influenza and COVID-19), in aged lungs and in chronic lung diseases, including chronic obstructive pulmonary disease (COPD), idiopathic pulmonary fibrosis (IPF). We develop a method for breath supersulphur-omics and demonstrate that levels of exhaled supersulphides increase in people with COVID-19 infection and in a hamster model of SARS-CoV-2 infection. Lung damage and subsequent lethality that result from oxidative stress and inflammation in mouse models of COPD, IPF, and ageing were mitigated by endogenous supersulphides production by CARS2/CPERS or exogenous administration of the supersulphide donor glutathione trisulphide. We revealed a protective role of supersulphides in airways with various viral or chronic insults and demonstrated the potential of targeting supersulphides in lung disease.

Fig. 1. Supersulphide-mediated anti-influenza host defense in mice.

a Kaplan–Meier survival curves for WT (n = 11) and Cars2^+/− (n = 12) mice infected with influenza A virus (Flu). b Mean body weight changes (95% confidence interval, CI) of Flu-infected WT (n = 8) and Cars2^+/− (n = 11) mice. None-infected groups (a, b): WT and Cars2^+/− (n = 3). c Quantification of lung injury scores for hematoxylin and eosin (HE) staining images of lungs: Flu (−) WT, n = 7, Flu (−) Cars2^+/−, Flu (+) WT, n = 9; Flu (+) Cars2^+/−, n = 8 (3 days post-infection). Supplementary Fig. 2a shows representative images. d, Numbers of viral RNA in lungs at days 3 and 7 after infection. e Viral infectivity of Flu attenuated with 10, 100, and 1000 μM Na₂S₄ (37 °C for 30 min). f, g Total cells (f) and macrophages (g) in BALF of Flu-infected mice. h IL-6 concentration in BALF from Flu-infected mice (n = 4). i 8-OHdG levels in Flu-infected WT and Cars2^+/− mice (n = 3). j GSSG/GSH ratio in lungs in Flu-infected WT and Cars2^+/− mice (n = 4–6), determined by supersulphide metabolome shown in Supplementary Fig. 3a. k Kaplan–Meier survival curves for Flu-infected WT mice treated with PBS as control (n = 12) or 32 μg GSSSG (n = 14). l Mean body weight changes (95% CI) in Flu-infected WT mice given PBS (n = 4) or 32 μg GSSSG (n = 4). m Lung injury scores of WT mice (day 8 after infection) treated with PBS (n = 5) or 32 μg GSSSG (n = 5). Supplementary Fig. 2f shows representative images. n, o Numbers of total cells (n), and macrophages (o) in BALF from Flu-infected WT mice given PBS (n = 4) or 32 μg GSSSG (n = 4). p Concentration of IL-6 in BALF from Flu-infected WT mice given PBS (n = 4) or 32 μg GSSSG (n = 4). Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Source data are provided as a Source Data file.

Fig. 2. Anti-SARS-CoV-2 effects of supersulphides in VeroE6/TMPRSS2 cells with or without CARS2 KD.

a Plaque assay for SARS-CoV-2 with control (VeroE6/TMPRSS2 cells) and their CARS2 KD cells. Left, middle, and right panels show representative plaque formation, plaque number, and plaque area, respectively. P = 0.0031 and 0.012. b–d Plaque-reduction assay with GSSSG (b), GSSG (c), and GSH (d) for SARS-CoV-2 using CARS2 KD VeroE6/TMPRSS2 cells. The plaque-reduction efficacy was assessed in terms of the effect of various compounds on the plaque formation by measuring total areas of plaques (mm²) observed in each well. The related data for the plaque-reduction assay with control VeroE6/TMPRSS2 cells are shown in Supplementary Fig. 4f–h. All P < 0.0001. e, f Suppression of SARS-CoV-2 infectivity by supersulphides. SARS-CoV-2 was incubated with Na₂S₂, Na₂S₄, GSSSG, GSSG, or GSH at indicated concentrations (e), and with 1 mM Na₂S, Na₂S₂, or Na₂S₄ (f), at 37 °C for 30 min. The viral infectivity was determined by the plaque-forming assay. P values are described in (e). P < 0.0001 in both in f. Data are means ± s.d. (n = 3). ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Source data are provided as a Source Data file.

Fig. 3. Anti-COVID-19 effects of supersulphides in ACE2-Tg mice and hamsters.

a Schematic drawing of the method for generation of ACE2-Tg::CPERS-deficient (AINK::ACE2-Tg) mice. Cars2^AINK/+ mice were produced via CRISPR-Cas9 system and crossed with ACE2-Tg mice to generate AINK::ACE2-Tg mice. Supplementary Fig. 7 shows genomic modification at Cars2 locus by the CRISPR-Cas9 system. b–d SARS-CoV-2 infection in mice. b, Mean body weight changes (95%, CI) of SARS-CoV-2-infected ACE2-Tg (n = 12) and AINK::ACE2-Tg mice (n = 10). All mice were infected intratracheally (i.t.) with 100 pfu (per 50 μl) of SARS-CoV-2; body weight was monitored until day 8 post-infection. P = 0.0045. c Kaplan–Meier survival curves for ACE2-Tg (n = 12) and AINK::ACE2-Tg mice (n = 10) infected with SARS-CoV-2; survival was monitored until day 14 post-infection. P = 0.035. d Viral yield in the infected lungs (homogenates) at 4 days post-infection. The viral titers were determined by plaque-forming assay and are expressed as a plaque-forming unit (pfu)/lung. P = 0.048. e–g SARS-CoV-2 infection in hamsters. Hamsters were infected i.t. with 6 × 10⁶ pfu (120 μl) of SARS-CoV-2; with simultaneous i.t. administration of 100 or 300 μg GSSSG. The hamsters were subsequently administered intraperitoneally (i.p.) 500 or 1000 μg GSSSG daily from day 1 to day 4. e Mean body weight changes (95% CI) in SARS-CoV-2-infected hamster that were given PBS (n = 6) or GSSSG (n = 6), as being monitored until day 4 post-infection. P = 0.035 (3 days) and 0.020 (4 days). f Semi-quantitative measurement of the area of consolidation (pathological lesion) of SARS-CoV-2-infected hamster lungs with or without GSSSG treatment (n = 6). P = 0.0004 and 0.0015. g The amounts (pfu) of virus yielded in the lungs of hamsters treated or untreated with GSSSG after infection, as assessed by the plaque-forming assay at 4 days post-infection. P < 0.0001 in both. Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Source data are provided as a Source Data file.

Fig. 4. Anti-SARS-CoV-2 activity of supersulphides: impairment of viral thiol proteases and S protein integrity.

a Protease activity (SARS-CoV-2 PL^pro and 3CL^pro, 1 μM each) was measured after treatment of proteases with 300 μM each GSSG, GSSSG, or ebselen. b Docking models of PL^pro triad (left panel) and 3CL^pro dyad (right panel) bound with GSSSG obtained by using SwissDock Yellow, blue, red, and green indicate sulphur, nitrogen, oxygen, and carbon atoms, respectively. c Inhibitory effects of GSSSG on PL^pro, 3CL^pro, papain, and cathepsin B (1 μM each). *P < 0.05, ***P < 0.001 vs. 0 μM GSSSG. P values are described in (c). d Inhibitory effects of Na₂S, Na₂S₂, Na₂S₃, and Na₂S₄ (3 μM each) on PL^pro and 3CL^pro (1 μM each). ***P < 0.001 vs. GSSSG (–). P values are described in (d). e, f Proteome analysis of GSSSG treated PL^pro (e) and 3CL^pro (f). LC-Q-TOF-MS chromatograms obtained from proteome analysis of PL^pro and 3CL^pro: peptide fragments containing cysteine residues C111 and C145 are shown. Cys modification identified in the peptide fragments are shown in the panel headlines. Perthioglutathionylation (GSSH adducts) of cysteine residues in PL^pro and 3CL^pro were detected by monitoring at m/z 920.1017 and 1224.1257, respectively: corresponding MS/MS spectra are illustrated in Supplementary Fig. 9. The GSSSG reaction with thiol proteases is schematized in the right panel. g, h Proteome of SARS-CoV-2 S protein, which includes each cysteine residue (C391, g and C525, h) identified via the LC-Q-TOF-MS. S proteins that were reacted with Na₂S₄ (0–300 µM), were subjected to the MS analysis (see Supplementary Fig. 10 for the spectra). The intensity of each peptide with Cys-modified was normalized to the 906-FNGIGVTQNVLYENQK-921 of S protein. Each signal intensity is shown as a relative value compared to the total intensity of corresponding fragments derived from DTT-treated S protein. The reaction of Na₂S₄ with S protein is schematized in the right panel. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 vs. Peptide-SH/0 μM Na₂S₄; ^##P < 0.01, ^###P < 0.001 vs. Peptide-SSH/0 μM Na₂S₄; ^¶¶¶P < 0.001 vs. Peptide-SSSH/0 μM Na₂S₄. P values are described in (g, h). Data are means ± s.d. (n = 3). Source data are provided as a Source Data file.

Fig. 5. Breath omics analyses in human and hamster.

a Sulphur metabolome of EBC of healthy subjects and patients with COVID-19. EBC were quickly collected by freezing expired air at −20 °C; levels of sulphur metabolites including HSH (H₂S, S²⁻), HSSH (H₂S₂, S₂²⁻), HSSSH (H₂S₃, S₃²⁻), HSO₃⁻, and HS₂O₃⁻ were quantified by using LC–MS/MS analysis after β-(4-hydroxyphenyl)ethyl iodoacetamide (HPE-IAM) labeling. Red arrows show higher levels of sulphur metabolites for a patient with exacerbated COVID-19. Each dot represents data from healthy subjects and COVID patients (n = 22 per group). P < 0.0001 (HSH), =0.0006 (HSSH), <0.0001 (HSSSH), = 0.0009 (HSO₃⁻), and <0.0001 (HS₂O₃⁻). b Non-invasive hamster EBC collection system. We developed here a hamster model especially for the breath analysis to experimentally identify biomarkers of COVID-19. c Mean body weight changes (95% CI) in SARS-CoV-2-infected WT hamster (n = 6). All hamster were intratracheally infected with 6 × 10⁶ pfu (per 100 μl) of SARS-CoV-2; survival was monitored until day 8 post-infection. d Copy numbers of SARS-CoV-2 N protein-encoded RNA analyzed by quantitative RT-PCR with EBC recovered from hamsters for 8 days after infection. e Viral proteome analysis with EBC of infected hamster until day 8 post-infection as performed vis LC-ESI-MS/MS. f Sulphur metabolome conducted with the EBC obtained from hamsters for 8 days after SARS-CoV-2 infection. P values are described in (f). Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Source data are provided as a Source Data file.

Fig. 6. Attenuated CARS2 expression in COPD and enhancement of COPD pathology in elastase-induced COPD model in Cars2^+/− mice.

a Expression of CARS2 in primary bronchial epithelial cells from patients with COPD (n = 10), healthy never-smokers (HNS, n = 6), and healthy ex-smokers (HES, n = 6) as determined by western blotting (upper panel) and quantitative results (lower panel). b Correlations between the amounts of CARS2 and FEV₁% predicted (left panel) or D_LCO/V_A predicted (right panel) of study subjects. Black, blue, and red circles indicate HNS, HES, and COPD, respectively. D_LCO diffusing capacity for carbon monoxide, VA alveolar volume. c Representative HE-stained images showing the extent of airspace enlargement. Scale bars, 100 μm. d Quantification of the mean linear intercept in c. PBS-treated WT mice (n = 10); PBS-treated Cars2^+/− mice (n = 11); elastase-treated WT mice (n = 8); elastase-treated Cars2^+/− mice (n = 9). e Representative CT images of a lung of an elastase-induced emphysema model mouse. Blue indicates the low-attenuation area. f Representative 3D micro-CT images of lungs of PBS-treated or elastase-induced WT and Cars2^+/− mice. Yellow indicates the low-attenuation area; the total lung area appears as a transparent shape. g The emphysematous area in (f) was quantified by calculating low-attenuation volume to total lung volume (LAV/TLV). PBS-treated WT (n = 8); PBS-treated Cars2^+/− mice (n = 4); elastase-induced WT (n = 7); elastase-induced Cars2^+/− mice (n = 8). Each dot represents data from an individual mouse (n = 4–8 per group). h Airflow obstruction (FEV_0.1/FVC) of lungs of elastase-induced WT and Cars2^+/− mice measured by using the flexiVent system on day 21. i–k Pulmonary function measured on day 21. i Representative flow-volume curves. j Pressure–volume curves. k Static compliance. PBS-treated WT mice (n = 7); PBS-treated Cars2^+/− mice (n = 6); elastase-treated WT mice (n = 7); elastase-treated Cars2^+/− mice (n = 7). l Concentrations of TNF-α, monocyte chemoattractant protein-1 (MCP-1), and IL−6 in BALF obtained from elastase-induced WT and Cars2^+/− mice on day 5. Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Source data are provided as a Source Data file.

Fig. 7. Sulphur metabolome and COPD pathology of elastase-induced COPD model in Cars2 mutant mice.

a Endogenous levels of GSSH and GSSSG in BALF obtained during 21 days from elastase-induced WT and Cars2^+/− mice, identified via LC–MS/MS analysis with HPE-IAM labeling. All P < 0.0001, but ^###P = 0.0002 (5 days), ^###P = 0.0004 (14 days) in CysSSH. All P < 0.0001 in GSSH. Data are means ± s.d. ^****P < 0.0001 vs. elastase-treated WT mice/day 0; ^###P < 0.001, ^####P < 0.0001 vs. elastase-treated Cars2^AINK/+ mice/day 0. b Representative 3D micro-CT images of lungs of elastase-treated WT and Cars2^AINK/+ mice. Yellow indicates the low-attenuation area; the total lung area appears as a transparent shape. c LAV/TLV for images in (b) was quantified by calculating the ratio of LAV to TLV. Elastase-treated WT mice (n = 5); elastase-treated Cars2^AINK/+ mice (n = 5). P = 0.036. Data are means ± s.d. ^*P < 0.05. d Numbers of total cells in BALF from Flu-infected WT and Cars2^AINK/+ mice. P = 0.034. Data are means ± s.d. ^*P < 0.05. e FEV_0.1/FVC for WT and Cars2^AINK/+ mice. P = 0.036. Data are means ± s.d. ^*P < 0.05. f Representative flow-volume curves for WT and Cars2^AINK/+ mice, as measured by the flexiVent system on day 21. Source data are provided as a Source Data file.

Fig. 8. Enhanced COPD pathology of CSExt-induced COPD model and lung premature ageing in Cars2^+/− mice.

a Quantification of the mean linear intercept for lungs of CSExt-treated COPD model mice (n = 4 per group). b, c Pulmonary function was measured by using the flexiVent system on day 28 after CSExt administration. b Airflow obstruction as represented by FEV_0.1/FVC. c Static compliance. PBS-treated WT mice (n = 4); PBS-treated Cars2^+/− mice (n = 4); CSExt-treated WT mice (n = 4); CSExt-treated Cars2^+/− mice (n = 4). d Representative western blot analysis of p53 and p21 in lungs of WT and Cars2^+/− mice (upper panel) and quantitative results of p53 (bottom left panel) and p21 (bottom right panel) (n = 3 for each group). e Representative immunohistochemical images showing localization of p21 (upper panels) and quantitative p21-positive cells (lower panels) in lungs of PBS- and CSExt-treated mice. Scale bars, 500 μm (low magnification) and 100 μm (high magnification). f p53 and p21 protein levels in whole lungs from aged WT and Cars2^+/− mice (88 weeks old) were measured by using Western blotting (left panel); quantitative results of p53 (middle panel) and p21 (right panel) in whole lungs (n = 5). g p21 immunostaining in lungs from aged WT and Cars2^+/− mice (88 weeks old) shown as representative images (left panel) and quantitative results (right panel). Scale bars, 500 μm (low magnification) and 100 μm (high magnification). h Representative images of airspace enlargement as visualized by HE staining (left panel) and quantification of the mean linear intercept (right panel). Scale bars, 500 μm. i Representative 3D micro-CT images of lungs from aged WT and Cars2^+/− mice (88 weeks old) (left panel). Yellow indicates the low-attenuation area; the total lung area appears as a transparent shape (right panel). LAV/TLV was quantified by calculating the ratio of LAV to TLV. Aged WT mice (n = 4); aged Cars2^+/− mice (n = 4). Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001. P values are described in this figure. Source data are provided as a Source Data file.

Fig. 9. Alleviation of COPD pathology in lungs of elastase- and CSExt-induced COPD models by supersulphides.

a Representative 3D micro-CT images of lungs of elastase-treated Cars2^+/− mice administered with 1 mM GSSSG. Yellow indicates the low-attenuation area; the total lung area appears as a transparent shape. Each dot represents data from an individual mouse (n = 3 per group). b LAV/TLV was quantified by calculating the ratio of LAV to TLV. Elastase-treated Cars2^+/− mice treated (n = 8) or untreated (n = 6) with 1 mM GSSSG. P = 0.0024. c FEV_0.1/FVC for elastase-treated Cars2^+/− mice with or without 1 mM GSSSG administration (n = 4 per group), as measured by using the flexiVent system. P = 0.014. d Numbers of total cells in BALF from elastase-treated Cars2^+/− mice administered with 1 mM GSSSG. P = 0.0009. e Quantification of the mean linear intercept for CSExt-treated WT and Cars2^+/− mice with or without administration of 1 mM GSSSG (n = 4 per group). P = 0.0055. f p21 immunostaining with lungs of CSExt-induced emphysema in WT and Cars2^+/− mice with or without GSSSG treatment, shown as representative images (left panel) and quantitative results (right panel). P < 0.0001 (left), = 0.0002, and <0.0001 (right). Scale bars, 500 μm (low magnification) and 100 μm (high magnification). g p53 and p21 protein levels as analyzed by using Western blotting (top panel) and quantitative results of p53 (bottom left panel) and p21 (bottom right panel). All P < 0.0001, but ^*P = 0.036. h Lack of effect of GSSG on the LAV/TLV for micro-CT images of lungs of elastase-induced Cars2^+/− mice. P = 0.0003 and <0.0001. Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001, N.S., not significant. Source data are provided as a Source Data file.

Fig. 10. Enhanced pulmonary fibrosis in Cars2^+/− mice (bleomycin-induced IPF model).

a Body weight changes in bleomycin-treated WT (n = 5) and Cars2^+/− (n = 4) mice. b On day 21 after intranasal bleomycin administration, mice underwent micro-CT evaluation of the lungs. Representative 3D reconstruction of lungs from bleomycin-treated WT and Cars2^+/− mice. Tissue volume appears orange-red and airspace volume appears green. Conducting airways that were ignored in the analysis appear in blue. c Quantification of tissue and airspace volumes in lungs in (b). d Representative images (left panel) and semi-quantification (right panel) of HE staining in lungs from bleomycin-treated WT and Cars2^+/− mice on day 6. P = 0.044. Scale bars, 500 μm (low magnification, ×40) and 100 µm (high magnification, ×100). e Representative images (left panel) and semi-quantification (right) of Masson’s trichrome staining in lungs from bleomycin-treated WT and Cars2^+/− mice on day 21. P = 0.034. Scale bars, 500 μm (low magnification, ×40) and 100 μm (high magnification, ×100). f Concentration of keratinocyte chemoattractant (KC), IL-1β, IL-6, MCP-1, and TNF-α in BALF of WT and Cars2^+/− mice at 6 days after intranasal administration of bleomycin. Each dot represents data from an individual mouse (n = 4 per group). P = 0.0017 (KC), 0.041 (IL-1β), 0.024 (IL-6), 0.033 (MCP-1), 0.0004 (TNF-α). g Apoptotic cells in lung sections obtained on day 6 after intranasal administration of bleomycin were identified by means of the TUNEL method. Scale bars, 100 μm. Data are means ± s.d. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001. Source data are provided as a Source Data file.