Intrathecal activation of CD8+ memory T cells in IgG4-related disease of the brain parenchyma

Abstract

IgG4-related disease (IgG4-RD) is a fibroinflammatory disorder signified by aberrant infiltration of IgG4-restricted plasma cells into a variety of organs. Clinical presentation is heterogeneous, and pathophysiological mechanisms of IgG4-RD remain elusive. There are very few cases of IgG4-RD with isolated central nervous system manifestation. By leveraging single-cell sequencing of the cerebrospinal fluid (CSF) of a patient with an inflammatory intracranial pseudotumor, we provide novel insights into the immunopathophysiology of IgG4-RD. Our data illustrate an IgG4-RD-associated polyclonal T-cell response in the CSF and an oligoclonal T-cell response in the parenchymal lesions, the latter being the result of a multifaceted cell-cell interaction between immune cell subsets and pathogenic B cells. We demonstrate that CD8⁺ T effector memory cells might drive and sustain autoimmunity via macrophage migration inhibitory factor (MIF)-CD74 signaling to immature B cells and CC-chemokine ligand 5 (CCL5)-mediated recruitment of cytotoxic CD4⁺ T cells. These findings highlight the central role of T cells in sustaining IgG4-RD and open novel avenues for targeted therapies.

Keywords: CSF single-cell sequencing; IgG4-related disease; cytotoxic T helper cell; inflammatory pseudotumor; pathogenic B-cell.

Figure EV1. Disease course of recurrent CNS‐restricted inflammatory pseudotumor

1. A

   Top: Serial axial MR imaging (T1‐weighted, contrast‐enhanced) representative of the patient’s disease course with indicated study dates. White arrowheads indicate disease manifestations as contrast enhancements. Bottom: Hematoxylin and eosin (H&E) stainings of representative resection tissues at indicated study dates.

2. B

   Sagittal MR imaging representative of the patient’s disease course with indicated study dates. White arrows indicate optic nerve lesion pre and post steroid treatment.

3. C, D

   Longitudinal development of inflammation markers and immunoglobulin levels in serum and cerebrospinal fluid (CSF). Gray box indicates duration of methylprednisolone treatment C. left, serum C‐reactive protein (CRP) [mg/dl] and leukocyte count [10E9/l]; right: serum levels of IgG, IgA, and IgM [g/l]. D. left, CSF total protein [mg/l] and total cell count [n/µl]; right, CSF levels of IgG, IgA, and IgM [mg/l].

Source data are available online for this figure.

Figure EV2. Histopathological features of IgG4‐RD in recurrent CNS‐restricted inflammatory pseudotumor

1. A, B

   Exemplary H&E staining (A) and IgG4 immunohistochemistry staining (B) of 2015 temporal lobe parenchyma, highlighting obliterative phlebitis and lymphocytic infiltration.

2. C, D

   Exemplary H&E staining (C) and Elastica van Gieson (EvG) staining (D) of 2019 optic nerve tissue, highlighting storiform fibrosis.

Source data are available online for this figure.

Figure EV3. Histopathological consensus criteria of IgG4‐RD

1. A, B

   Immunohistochemistry DAB staining of IgG4 (A) and total IgG (B) on resection tissues (study dates indicated). As suggested by the consensus statement on the pathology of IgG4‐RD (Deshpande et al, 2012), three 40x fields with the highest number of IgG4⁺ and IgG⁺ cells each were selected, counted, and averaged within these fields. Cell counts as indicated.

2. C

   Quantification of IgG4⁺ and IgG⁺ cells from panels A‐B. IgG4/IgG ratio shown on the right. Individual values, mean ± SEM.

Source data are available online for this figure.

Figure EV4. Automated IgG4 quantification and immunoglobulin ELISA in recurrent CNS‐restricted inflammatory pseudotumor

1. A–C

   Immunohistochemistry DAB staining of IgG4 and total IgG on resection tissues (study dates indicated). Bottom, automatic quantification results as per default DAB parameters of QuPath 0.2.3 software.

2. D

   Immunoglobulin ELISA of IPT CSF, control CSF, and MS patient‐derived CSF and respective sera. IgG4 and total IgG concentrations were measured [mg/dl]. IgG4/IgG ratio for CSF and serum shown on the right. Individual values, mean ± SD shown. n = 4 experimental repeats with technical replicates. Significance was assessed by two‐way ANOVA analysis with Tukey’s post hoc testing (*P < 0.05, **P < 0.01, ***P < 0.001).

Figure EV5. Intraparenchymal IgG4‐RD and peripheral T helper‐like cells in IgG4‐RD cerebrospinal fluid

1. A

   Immunofluorescence staining of 2015 resection tissue of GFAP (Alexa Fluor Plus 488, red) and IgG4 (Alexa Fluor Plus 546, green). DAPI (blue) was used for nuclear staining. 10× (left) and 20× (right) field shown.

2. B

   Stacked bar chart depicting mean relative expression levels of T helper cell‐associated cytokines in cell subsets as identified by Seurat v4 reference mapping in IPT CSF as shown in Fig 1A.

3. C

   Dot plot of mean relative PDCD1 (x‐axis) and CXCR5 (y‐axis) transcript expression in CD4 T‐cell subset in inflammatory pseudotumor CSF. Dotted lines indicate PDCD1⁺CXCR5⁻ peripheral T helper (Tph)‐like cells. TEM, T effector memory; TCM, T central memory.

4. D

   Heatmap depicting average relative expression of relevant transcripts in CD4 T‐cell subset in inflammatory pseudotumor. PDCD1⁺ = mean relative PDCD1 expression > 0.5; CXCR5⁺ = mean relative CXCR5 expression > 0.5.

Source data are available online for this figure.

Figure 1. Comparative cerebrospinal fluid single‐cell profiling of an inflammatory pseudotumor

1. A

   Uniform Manifold Approximation and Projection (UMAP) of sequenced single cells from inflammatory pseudotumor (IPT) CSF (n = 1 sample, n = 4,324 single cells, left), control CSF (n = 6 samples, n = 15,467 single cells, middle), and multiple sclerosis (MS) patient‐derived CSF (n = 6, n = 18,412 cells, right). Indicated immune cell subsets as identified by Seurat v4 reference (ref) mapping.

2. B

   Stacked bar chart of relative B‐cell abundances as identified by Seurat v4 reference mapping in inflammatory pseudotumor (IPT) CSF, control CSF, and MS patient‐derived CSF.

3. C

   Circle plots representing the relative abundance of B‐cell subsets as identified by Seurat v4 reference mapping in inflammatory pseudotumor CSF, control CSF, and MS patient‐derived CSF.

4. D

   Pseudotime analysis of intrathecal B‐cell subsets in inflammatory pseudotumor CSF with the naïve B‐cell cluster, identified by canonical markers, as root node. Percentage of cycling naïve B cells and IgG4 B cells depicted in pie charts.

5. E

   Retrospective immunohistochemistry DAB staining of IgG4 and IgG on archival temporal lobe resection tissue. As suggested by the consensus statement on the pathology of IgG4‐RD (Deshpande et al, 2012), three 40x fields with the highest number of IgG4⁺ and IgG⁺ cells were selected, counted, and averaged within these fields. Cell counts as indicated.

6. F

   C‐X‐C motif chemokine 13 (CXCL13) concentrations measured by ELISA from inflammatory pseudotumor (IPT) CSF, control CSF, and MS patient‐derived CSF. Individual values, mean ± SEM; n = 4 experiment repeats with technical replicates.

7. G

   Stacked bar chart depicting mean relative expression levels of T helper cell‐associated cytokines in T‐cell subsets as identified by Seurat v4 reference mapping in inflammatory pseudotumor CSF as in (A). TEM, T effector memory; TCM, T central memory; Treg, regulatory T cell.

8. H

   Violin plot depicting relative expression levels of IL4R (left) and IL10RA (right) in B‐cell subsets as identified by Seurat v4 reference mapping in inflammatory pseudotumor CSF.

Data information: (B, C) Cell subsets as indicated by the legend on the right.

Figure 2. CNS manifestation of IgG4‐RD with distinct cytotoxic T‐cell–B‐cell interactions

1. A

   Relative abundances of cell types identified by Seurat v4 reference mapping as in Fig 1A in IPT CSF compared with control or MS patient‐derived CSF. Boxplot depicting 25^th–75^th percentiles with median shown as central band and whiskers extending from minimum to maximum values. FC, fractional difference.

2. B

   TCR repertoire of IPT CSF as analyzed by single‐cell VDJ sequencing. Top, all sequenced and paired TCR clonotypes shown, clonotypes ordered counterclockwise according to abundance. Bottom, clonotype frequency stratified by T‐cell subset.

3. C

   Feature plot of UMAP of single cells from pseudotumor CSF shown in Fig 1A, depicting cell‐wise representations of indicated transcripts. Relative expression shown.

4. D

   Heatmap of cell–cell interaction analysis depicting top predicted interactions based on receptor‐ligand co‐expression and reference‐based cell subsets. Interaction z‐score shown.

5. E

   Dot plot representation of top 30 receptor–ligand interactions based on molecule co‐expression and reference‐based cell subsets. Mean relative expression of both interaction partners (dot color) and interaction P‐value (dot size) shown. P‐values are derived from one‐sided permutation tests and refer to the enrichment of the interacting ligand–receptor pair in each of the interacting pairs of cell types.

6. F

   Relative expression of CD40, CD74, and CCL5 on indicated combined meta‐clusters classified by reference‐based cell identification as shown in Fig 1A.

7. G

   Circos plot representation of highlighted cell–cell interactions in IPT CSF between indicated cell subsets. Colored receptor–ligand interaction pairs from (E‐F).

Figure 3. Clonal evolution of the TCR repertoire in IgG4‐RD revealed by longitudinal VDJ‐ sequencing of disease manifestations

1. A

   TCR repertoire clonality in inflammatory pseudotumor lesions (2015–2019) and 2019 CSF as analyzed by VDJ sequencing. Clonotype abundance represented by angular diameter.

2. B

   TCR repertoire distribution assessed as clonal space homeostasis, i.e., proportion of the repertoire occupied by the clones of a given size.

3. C

   Rarefaction analysis indicating estimated diversity of clonotype richness in inflammatory pseudotumor lesions (2015–2019). Sample size indicated on the x‐axis.

4. D

   TCR sequence overlap between inflammatory pseudotumor lesions (2015–2019) measured by Morisita’s overlap index.

5. E

   Shared (absolute number) and unique (proportion) clonotypes between inflammatory pseudotumor lesions (2015–2019).

6. F

   Motif probability matrix depicting occurrences of amino acids in each position of all computed CDR3 beta chains.

7. G

   Longitudinal tracking of most abundant clonotypes in inflammatory pseudotumor lesions (2015–2019). CDR3 amino acid sequence and v‐chain usage for each clonotype annotated in the legend. Private clones are limited to each time point. Shared clones are found at all time points.

8. H

   Automated CDR3 beta chain high‐resolution modeling of dominant T‐cell clone CASSQEYSPYEQY.