Blood-labyrinth barrier damage mediated by granzymes from cytotoxic lymphocytes results in hearing loss in systemic lupus erythematosus

Abstract

Hearing loss (HL) in patients with systemic lupus erythematosus (SLE) has been widely reported, though the underlying mechanisms remain unclear. In this study, we demonstrate significant HL in an imiquimod-induced mouse model of SLE, accompanied by blood-labyrinth barrier (BLB) damage in the stria vascularis (SV). We found that cytotoxic CD8⁺ T lymphocytes and natural killer cells mediate BLB disruption, with granzyme b (Gzmb) acting as the primary factor inducing cell death of the cochlear capillary endothelial cells. Remarkably, inhibiting Gzmb by Serpinb9 significantly alleviates the elevations of hearing thresholds in imiquimod-induced SLE model mice, suggesting a therapeutic target for autoimmune-related HL. Our findings provide insights into the pathogenesis of SLE-related HL and propose Gzmb inhibition as a potential therapeutic strategy.

Keywords: blood–labyrinth barrier; granzymes; hearing loss; immune-mediated cytotoxicity; systemic lupus erythematosus.

Fig. 1.

Characteristics of SLE and hearing loss (HL) in the IMQ-induced SLE mouse model. (A) Spleen index (spleen weight/body weight) measured at different time points (weeks 2, 4, 6, 8, 10). n = 6 mice each. (B) Urine protein levels measured by ELISA. n = 6 mice each. (C) Serum anti-dsDNA antibody levels detected by ELISA. n = 6 mice each. (D) Serum IgG levels detected by ELISA. n = 6 mice each. (E) Spleen morphology at weeks 2, 4, 6, 8, 10 of model mice and weeks 10 of control mice. (F) H&E staining images of the kidney from IMQ-induced SLE and control mice. (G–J) ABR thresholds in IMQ-induced SLE and control mice. n = 8 mice each. (K–N) ABR threshold shifts in IMQ-induced SLE and control mice. (O–R) DPOAE thresholds in IMQ-induced SLE and control mice. n = 8 mice each. Data are presented as mean ± SD. One-way ANOVA was used in (A–D). Two-way ANOVA was used in (G–R). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 2.

Single-cell transcriptome atlas of the cochlea from IMQ-induced SLE mice. (A) UMAP plot showing the distribution of various cochlear cell types. (B) Dot plot showing representative marker genes across different cell types. (C) Histogram displaying the proportion of distinct cell types in SLE and control mice. (D and E) Heatmaps of different interaction counts and strengths between cells in SLE and control mice (SLE vs control). The vertical axis is the sources. The horizontal axis is the targets. Heat map colors from blue to red mean the interaction counts or strengths from decrease to increase. (F and G) Circle plots depicting VE-cadherin and JAM signaling networks in SLE and control mice. (H–L) Violin plots of junction molecule expression levels in CECs, including VE-cadherin, occludin, ZO-2, JAM-2, and JAM-3.

Fig. 3.

Increased permeability and compromised integrity of the BLB in IMQ-induced SLE mice. (A and B) Images and quantification of Evans blue (961 Da) accumulation in the cochlear lateral wall of IMQ-induced SLE mice, absent in control mice. n = 6 of each group. (C and D) Capillary leakage of BLB in IMQ-induced SLE mice, but not in control mice, visualized by FITC-dextran (40 kDa) and FITC-IgG (150 kDa). Leakage is indicated by arrows pointing to the tracer fluorescence (green). (E and F) Quantification of tracer leakage from capillaries, showing significant leakage in IMQ-induced SLE mice. n = 6 of each group in both the FITC-dextran experiments and the FITC-IgG experiments. (G–M) Western blot images and quantitative analysis showing reduced levels of junction proteins VE-cadherin, occludin, ZO-2, JAM-2, and JAM-3 in SV tissues of IMQ-induced SLE mice compared to controls. n = 3 of each group. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by unpaired Student’s t test.

Fig. 4.

Key role of cytotoxic immune cells in BLB damage identified by integrated analyses of scRNA-seq and bulk RNA-seq. (A) KEGG enrichment analysis of differentially expressed genes (DEGs) of CECs from scRNA-seq. (B) GSVA of CECs based on KEGG pathways. (C) GSVA of CECs based on GO pathways of scRNA-seq data. (D) Accumulation of CTLs and NK cells in the stria vascularis (SV). CTLs and NK cells were marked by CD8a and NKp46, respectively. Arrowheads indicate the accumulated immune cells.

Fig. 5.

Granzymes from cytotoxic lymphocytes mediate apoptosis of CECs in the BLB. (A) Heatmap of the top DEGs in CTLs/NK cells. (B) Hub gene analysis in CTL/NK cells via protein–protein interaction (PPI) networks. (C) CellChat interaction analysis between CTLs/NK cells (source of ligands) and CECs (target of receptors). (D and E) Violin plots showing the expression levels of Gzma, Gzmb, Bax, and Bcl-2. (F–M) Relative expression levels of Gzma, Gzmb, Bax, Bcl-2, pro-Caspase-3, and cleaved Caspase-3 in SV tissues analyzed by western blot. n = 3 of each group. (N and O) Proportions of dead CECs (labeled by CD31 and TUNEL) detected and calculated using flow cytometry. n = 3 of each animal group. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by unpaired Student’s t test.

Fig. 6.

Rescue of HL by inhibiting granzymes. (A) Experimental schematic illustrating the application of Gzma or Gzmb inhibitors via tail intravenous injections. (B–D) ABR thresholds (B), ABR threshold shift (C), and DPOAE thresholds (D) of various experimental groups (SLE, control, Serpinb9, and Serpinb6b) at week 4. n = 4 of each animal group. (E, and G–K) Relative expression levels of Bax, Bcl-2, pro-Caspase-3, and cleaved Caspase-3 in the SV tissues of Serpinb9-treated mice analyzed by western blot. n = 3 of each group. (F) Expression levels of Bax, Bcl-2, pro-Caspase-3, and cleaved Caspase-3 in the SV tissues of Serpinb6b-treated mice analyzed by western blot. n = 3 of each group. (L and M) Proportions of dead CECs (labeled by CD31 and TUNEL) detected and calculated using flow cytometry. n = 3 of each animal group. Data are expressed as mean ± SD. Two-way ANOVA was used in (B–D). One-way ANOVA was used in (G–K and M). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.