CD8+ T cell-mediated endotheliopathy is a targetable mechanism of neuro-inflammation in Susac syndrome

Abstract

Neuroinflammation is often associated with blood-brain-barrier dysfunction, which contributes to neurological tissue damage. Here, we reveal the pathophysiology of Susac syndrome (SuS), an enigmatic neuroinflammatory disease with central nervous system (CNS) endotheliopathy. By investigating immune cells from the blood, cerebrospinal fluid, and CNS of SuS patients, we demonstrate oligoclonal expansion of terminally differentiated activated cytotoxic CD8⁺ T cells (CTLs). Neuropathological data derived from both SuS patients and a newly-developed transgenic mouse model recapitulating the disease indicate that CTLs adhere to CNS microvessels in distinct areas and polarize granzyme B, which most likely results in the observed endothelial cell injury and microhemorrhages. Blocking T-cell adhesion by anti-α4 integrin-intervention ameliorates the disease in the preclinical model. Similarly, disease severity decreases in four SuS patients treated with natalizumab along with other therapy. Our study identifies CD8⁺ T-cell-mediated endotheliopathy as a key disease mechanism in SuS and highlights therapeutic opportunities.

Fig. 1. Accumulation of activated CD8⁺ T cells in SuS.

a Brain lesion from a patient with SuS stained for T cell markers CD3, CD8, and CD4, B cell marker CD20, and the plasma cell marker CD138. Bars: 50 µm. Most T cells belong to the CD8⁺ T cell subset; light brown cells are most likely microglia cells; B cells are rare (arrowhead). One B cell is enlarged in the insert. CD138⁺ plasma cells are absent from this and other lesions. The insert of the CD138 staining shows plasma cells from a positive control (Rasmussen encephalitis). b Graphs representing proportions of CD19⁺ B cells (top left), CD138⁺ plasma cells (top right) among lymphocytes and CD8⁺ T cells (middle left), CD4⁺ T cells (middle right), HLA-DR⁺CD8⁺ T cells (bottom left), and HLA-DR⁺CD4⁺ T cells (bottom right) among CD3⁺ T cells in the peripheral blood (PB, closed symbols; SoD = 76; SuS = 32; MS = 227) and cerebrospinal fluid (CSF, open symbols; SoD = 76; SuS = 14; MS = 227) of somatoform disorders (SoD, blue circles), SuS (cayenne squares), and MS patients (red triangles up). c Quantification of naive, central memory (CM), effector memory (EM), and effector memory expressing CD45RA (EMRA) CD8⁺ (left) and CD4⁺ (right) T cell subsets in the peripheral blood of healhy donors (HD; n = 20, closed blue circles), SuS (n = 20, closed cayenne squares), and MS patients (n = 10, closed red triangles up). Statistical analysis was performed using Kruskal–Wallis test with Dunn’s post-test. Error bars indicate the mean ± s.d.; p values: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 2. Perturbations in the CD8⁺ T_EMRA repertoire of SuS patients.

a Complexity score (left) and CDR3 length distribution (right) of peripheral CD8⁺ and CD4⁺ T cell repertoires of HD (closed blue circles, n = 11), SuS (closed cayenne squares, n = 7), and MS patients (closed red triangles up, n = 12) assessed by TCR Vβ-CDR3 spectratyping. b Representative Manhattan plots showing TCR repertoires of CD8⁺ T cells from a HD (top left), a SuS patient (top right, patient #10), a MS patient (bottom left), and SuS CD8⁺ T_EMRA cells (bottom right, patient #10). The x-axis depicts all the Vβ genes, the z-axis the Jβ genes, and the column heights represent the percentage reads for each V/J gene combination. c Clonality in the CD8⁺ T cell repertoires of HD (closed blue circles, n = 12), SuS (closed cayenne squares, n = 14), and MS (closed red triangles up, n = 15) patients (left). Clonality in the total CD8⁺ T cell and CD8⁺ T_EMRA repertoire of SuS (n = 6) patients (right). d Clonality of SuS patients with clinically active disease (closed cayenne squares, n = 6) or in clinical remission (open cayenne squares, n = 8). e Quantification of the ten most prevalent clones in the CD8⁺ T cell repertoires of HD (closed blue circles, n = 12), SuS (closed cayenne squares, n = 14), and MS (closed red triangles up, n = 15) patients (left), as well as in the total CD8⁺ T cell and CD8⁺ T_EMRA repertoire of SuS (closed cayenne squares, n = 6) patients (right). f Graph representing the proportion of viral, public, SuS-specific public, and SuS-specific private clones in the ten most prevalent clones in the total CD8⁺ T cell (left) and T_EMRA (right) repertoire of SuS patients. Statistical analysis was performed using Kruskal–Wallis test with Dunn’s post-test (a, c, e left graph) or unpaired Student’s t test (c right graph; d, e right graph), respectively. Error bars indicate the mean ± s.d.; p values: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 3. CTLs are more prevalent in the periphery of SuS patients.

a Representative examples (top) and quantification (bottom) of granzyme B (GrB; left) and perforin (right) expressing CD8⁺ T cells circulating in the blood of HD (closed blue circles, n = 18), SuS patients (closed cayenne squares, n = 13), and MS patients (closed red triangles up, n = 9). b Representative examples (top) and quantification (bottom) of CD107a-expressing CD8⁺ T cells from HD (closed blue circles, n = 15), SuS patients (closed cayenne squares, n = 14), and MS patients (closed red triangles up, n = 9) under the indicated conditions of re-directed lysis assay with αCD3-linked P815 cells. c Middle: Experimental set-up of allogenic suppression assays to test the suppressive capacity of regulatory T cells. Left: Graphs representing the degree of suppression of proliferation of HD CD8⁺ T cells upon co-culture with titrated numbers of T_reg from HD (closed blue circles) and SuS (closed cayenne squares) (n = 3). Right: Graphs representing the degree of suppression of HD (closed blue circles, n = 7) and SuS patient (closed cayenne squares, n = 7) CD8⁺ T cell proliferation upon co-culture with titrated numbers of independent HD T_reg. Statistical analysis was performed using Kruskal–Wallis test with Dunn’s post-test (a, b) or two-way ANOVA with Bonferroni post-test (c), respectively. Error bars indicate the mean ± s.d.; p values: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 4. CTLs accumulate in damaged microvessels of SuS patients CNS biopsies of SuS patients (n = 7).

a Accumulation of CD8⁺ T cells (brown) in a brain microvessel. b GrB⁺ cells in a brain microvessel; arrowheads indicate GrB⁺ cells attached to the vessel wall. c MHC class-I expression on ECs (arrowhead). Several leukocytes adhere to the endothelium. d Yellow arrowheads point to CD8⁺ T cells (blue) attached to CD31⁺ ECs (red). The enlargement of the upper CD8⁺ T cell shows its GrB⁺ (green) granules polarized toward the ECs. Owing to close proximity of the green GrB⁺ granules to the blue CD8⁺ cell membrane, these granules have a cyan blue color. The lower CD8⁺ T cell is migrating through the damaged (white arrowhead) vessel wall. e The same triple staining for GrB (green), CD8 (blue), and CD31 (red) shows a cytotoxic T cell attached to an intact vessel wall (yellow arrowhead). The upper left corner shows a microhemorrhage (green arrow) with parenchymal erythrocytes. f An apoptotic CD31⁺ EC with a condensed nucleus (arrowhead) within a microhemorrhage. g Staining for CD34 shows an EC with an apoptotic condensed (arrowhead) nucleus within a microhemorrhage encircled by the blue dots. h TUNEL-positive nuclei (arrowheads) of CD31⁺ ECs. i Turnbull blue (TBB) shows iron deposition around a blood vessel. j–l Small ischemic lesion combining loss of GFAP⁺ astrocytes, MBP⁺ myelin, and axonal damage shown by APP⁺ spheroids. m Proportions of CD8⁺ T cells in contact with ECs in brain specimen of SuS (cayenne squares, n = 7) and MS patients (red triangles up, n = 6). n Quantification of CD34⁺ endothelial apoptosis in brain specimen derived from non-inflammatory controls (blue circles, n = 6), SuS (cayenne squares, n = 7), and MS patients (red triangles up, n = 6). Scale bars: 20 µm (a, b, f–i), 25 µm (c–e), 200 µm (j, l). Statistical analysis was performed using Mann–Whitney test (m) or one-way ANOVA with Bonferroni post-test (n), respectively. Error bars indicate the mean ± s.d.; p values: **p < 0.01; ***p < 0.001. Source data are provided as a Source Data file.

Fig. 5. EC-HA⁺ mice as an in vivo model for CTL-mediated endotheliopathy.

a Investigation of T cell CNS infiltration in the brain, retina, and inner ear of EC-HA⁻ or EC-HA⁺ recipient mice on day 7 after adoptive transfer. Left: Representative histological sections documenting T cell infiltration (CD3⁺, brown). Right: Quantification of tissue-infiltrating T cells in EC-HA⁻ (closed blue circles) or EC-HA⁺ (closed cayenne diamonds) recipient mice (n = 4–8 per group); data for retina and inner ear originate from 2 independent experiments, involving transfer of cytotoxic CD8⁺ T cells alone or with control IgG. b Quantification of the cytolytic activity of CNS-infiltrating T cells analyzed as CD107a expression on day 7 after adoptive transfer in CD45.2⁺ EC-HA⁻ (closed blue circles) or EC-HA⁺ recipient mice (closed cayenne diamonds) (n = 7–8 per group, 2 independent experiments). c Weight loss (left; 11–12 mice per group) and rotarod motor performance (right; 7–8 mice per group) following adoptive transfer of HA-specific CTLs in EC-HA⁻ (closed blue circles) or EC-HA⁺ (closed cayenne diamonds) mice. Data are from three (body weight) or two (motor performance) independent experiments. Statistical analysis was using the unpaired Student’s t test (a, b) or two-way ANOVA with Bonferroni post-test (c), respectively. Error bars indicate the mean ± s.d.; p values: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 6. CTLs accumulate in damaged microvessels of EC-HA⁺ mice.

Representative brain images of EC-HA⁺ mice 4 days after adoptive transfer of cytotoxic CD8⁺ T cells (n = 6). a Accumulation of leukocytes in CNS microvessels. b Expression of MHC class-I molecules (green) on CD31⁺ (red) ECs; nuclei marked with TO-PRO-3 (blue). c Presence of GrB⁺ (red) CD8⁺ (blue) T cells in vessel lumen (L) of a battered CD31⁺ (green) blood vessel. Par parenchyma d GrB⁺ granules (red, arrowheads) CD8⁺ cells (blue) in close contact with the lumen of a small CD31⁺ blood vessel (green). e Representative image exhibiting CD31⁺ (brown) endothelium loss (arrowheads) in CNS microvessels of EC-HA⁺ mice following CD8⁺ T cell transfer. f Apoptosis and g DNA fragmentation (arrowheads) of CD31⁺ ECs (brown) detected with TUNEL staining (black). h Microhemorrhages in close proximity to destroyed vessels. Scale bars 10 µm. i Percentage of endothelial apoptosis assessed condensed nuclei in CD34⁺ endothelial cells (74–206 cells per mouse) of EC-HA⁻ (closed blue circles) or EC-HA⁺ mice (closed cayenne diamonds) (n = 6 per group). Scale bars 100 µm. Statistical analysis was performed by unpaired Student’s t test. Error bars indicate the mean ± s.d. p value: **p < 0.01. Source data are provided as a Source Data file. j Deposition of IgG in the hippocampus and turnbull blue staining (TBB) of the cerebellum of EC-HA⁻ and EC-HA⁺ mice after CTL transfer. Scale bars 500 µm. Data from two independent experiments, 5–6 mice per genotype are shown. k Representative small ischemic lesion in and above the corpus callosum at day 7 post-transfer in an EC-HA⁺ mouse. APP labeling (arrowhead) reveals axonal damage. Focal loss of astrocytes is visualized by the absence of GFAP (arrowhead). Detection of carbonic anhydrase II (CAII) reveals the focal loss of oligodendrocytes (arrowhead). The CD31 staining shows loss of ECs (arrowhead). Nuclear counterstaining was performed with hematoxylin (blue) (n = 6 per group).

Fig. 7. Effects of anti-VLA4 mAb in a preclinical and clinical setting.

a, b Preclinical data. a EC-HA⁺ mice received HA-specific cytotoxic CD8⁺ T cells and were treated every 4 days with the anti-mouse α4 integrin mAb (PS/2; open cayenne diamonds) or with an isotype control mAb (IgG2b; closed cayenne diamonds) from day 0 onwards. Clinical signs including weight loss (left) and rotarod performance (right) were assessed daily. The data are from 2 independent experiments involving 10–11/mice per group. b Absolute numbers of transferred (Thy1.2⁺CD45.1⁺) T cells infiltrating within the CNS of EC-HA⁺ mice (n = 4–8/group) 7 and 28 days after treatment with a control IgG2b (closed cayenne diamonds) or the anti-α4 integrin mAb (open cayenne diamonds). c–f Clinical data. c Quantification of unstimulated SuS CD8⁺ T cells (n = 7) adhering to HBMEC monolayer under flow conditions in the absence (closed cayenne squares) or presence (open cayenne squares) of the anti-human α4 integrin mAb natalizumab. Error bars indicate the mean ± SD. d Clinical episodes (closed black diamonds) of 4 SuS patients receiving treatments with 300 mg natalizumab monthly (cayenne line), oral prednisolone, tapered from 1 mg/kg body weight (yellow line), pulses of methylprednisolone, 1 g for 5 days (closed yellow diamonds), monthly IVIg 0.4 mg/kg body weight for 5 days (open white line), or other treatments, including cyclophosphamide, plasma exchange, mycophenolate mofetil, and azathioprine (blue line). *Patient 23 is still under treatment with natalizumab, while patients 25, 12, and 22 are not. e Bar graph representing the disease score of the four SuS patients before and during treatment with natalizumab. f Detection of CNS lesions (arrows) with sagittal FLAIR MRI sequence in patient 25 before (left) and 15 months after the beginning of natalizumab treatment (right). Statistical analysis was performed using unpaired (b), paired Student’s t test (c), or two-way ANOVA with Bonferroni post-test (a), respectively. Error bars indicate the mean ± s.d. p values: **p < 0.01; ***p < 0.001. Source data are provided as a Source Data file.

Fig. 8. Role of CD8⁺ T cells in the pathophysiology of SuS.

Working model of role of CD8⁺ T cells in the pathophysiology of SuS. A yet unknown antigenic stimulus drives activation, clonal expansion, and differentiation of CD8⁺ T cells into GrB- and perforin-expressing CD8⁺ T_EMRA cells. CD8⁺ T cells accumulate in microvessels of the brain, retina, and inner ear, where they adhere to the endothelium, recognize HLA:peptide complex(es), polarize their cytolytic vesicles toward the endothelial plasma cell membrane, and induce apoptosis of ECs, most likely in a perforin/GrB-dependent manner. Death of ECs and focal disruption of the blood–brain barrier result in microhemorrhages, whereas occlusion of small blood vessels leads to small ischemic lesions with loss of astrocytes, oligodendrocytes, neurons, and axons. Finally, ischemic lesions become gliotic by infiltration of surrounding astrocytes. Illustration^©2019-Heike Blum, Department of Neurology with Institute of Translational Neurology, University Hospital Münster.