MSC transplantation ameliorates depression in lupus by suppressing Th1 cell-shaped synaptic stripping

Abstract

Systemic lupus erythematosus (SLE), an autoimmune disease, can cause psychiatric disorders, particularly depression, via immune activation. Human umbilical cord mesenchymal stromal cell (hUCMSC) transplantation (MSCT) has been shown to ameliorate immune dysfunction in SLE by inducing immune tolerance. However, whether MSCT can relieve the depressive symptoms in SLE remains incompletely understood. Here, we demonstrate that MSCT relieved early-onset depression-like behavior in both genetically lupus-prone (MRL/lpr) and pristane-induced lupus mice by rescuing impaired hippocampal synaptic connectivity. Transplanted hUCMSCs targeted Th1 cell-derived IFN-γ to inhibit neuronal JAK/STAT1 signaling and downstream CCL8 expression, reducing phagocytic microglia apposition to alleviate synaptic engulfment and neurological dysfunction in young (8-week-old) lupus mice. Systemic delivery of exogenous IFN-γ blunted MSCT-mediated alleviation of synaptic loss and depressive behavior in lupus mice, suggesting that the IFN-γ/CCL8 axis may be an effective therapeutic target and that MSCT is a potential therapy for lupus-related depression. In summary, transplanted hUCMSCs can target systemic immunity to ameliorate psychiatric disorders by rescuing synaptic loss, highlighting the active role of neurons as intermediaries between systemic immunity and microglia in this process.

Keywords: Autoimmune diseases; Autoimmunity; Depression; Immunotherapy; Inflammation; Neuroscience.

Figure 1. MSCT alleviates depression in MRL/lpr mice.

(A) Timeline of the experimental procedure for MRL/lpr mice. (B–D) Evaluation of depression-like behavior in MRL/lpr mice and wild-type (MRL/mpj) controls (n = 11 mice/group). The SPT (B), TST (C), and FST (D) were performed at 8 and 18 weeks. (E) Experimental protocol for the treatment of MRL/lpr mice by MSCT. (F–H) Effects of MSCT on depression-like behavior in MRL/lpr mice (n = 11 mice/group). Five-week-old MRL/lpr mice were intravenously injected with hUCMSCs (5 × 10⁵ cells in 500 μL of PBS) or PBS (as a control). The SPT (F), TST (G), and FST (H) were performed 3 weeks after MSC/PBS injection. The data are presented as mean ± SEM. *P < 0.05; **P < 0.01, determined using 1-way ANOVA followed by Tukey’s post hoc test (B–D), or 2-tailed unpaired t test (F–H). SPT, sucrose preference test; TST, tail suspension test; FST, forced swim test. See also Supplemental Figures 1–4.

Figure 2. MSCT inhibits microglia-mediated synaptic stripping to rescue dendritic loss in the brains of lupus mice.

(A and B) Images and quantification of Golgi-stained dendritic spines in dentate gyrus granule neurons from MRL/mpj mice and MSCT/PBS-treated MRL/lpr mice (n = 5 mice/group). Scale bar: 5 μm. (C–E) Immunostaining of presynaptic (synaptophysin, red) and postsynaptic (PSD-95, green) proteins in hippocampal sections from MRL/mpj mice and MSCT/PBS-treated MRL/lpr mice (C) and quantification of the levels of these proteins (D and E) (n = 5 mice/group, with an average of 3–4 slices per mouse). Scale bar: 10 μm. (F) Heatmap showing the relative expression of significantly altered genes in the hippocampi of MRL/mpj and MRL/lpr mice treated with or without MSCT according to RNA-seq (see Methods). The color scale represents the genewise z score calculated from the normalized gene expression levels. Each column represents an individual group (n = 3 mice/group). (G and H) Orthogonal view of a microglial cell (IBA-1, red) showing PSD-95 inclusions (green, upper in G) and colocalization with CD68 (red, upper in H). These findings suggest that microglia eliminate synapses in the hippocampi of MRL/lpr mice. Bottom: 3D reconstruction of the cell in G and lysosome in H (red) and PSD-95 inclusions (green) with Imaris software. Scale bars: 5 μm. (I) Quantification of reconstructed PSD-95⁺ spheres and PSD-95⁺CD68⁺ spheres inside microglia in the indicated mice (n = 20 cells from 4–5 mice). The data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 by 1-way ANOVA followed by Tukey’s post hoc test. See also Supplemental Figures 5 and 6.

Figure 3. MSCT inhibits microglial apposition and synaptic stripping by targeting neuronal CCL8 signaling.

(A and B) Mouse hippocampal tissues were harvested, and RNA-seq was subsequently performed. The heatmap shows all genes whose expression was altered in MRL/lpr mice compared with MRL/mpj mice and whose expression was further reversed by MSCT (A). The Venn diagram (B) shows overlapping genes that were significantly upregulated in MRL/lpr mice versus controls and whose expression was reversed by MSCT (downregulated in MSCT-treated mice versus MRL/lpr mice). (C) Heatmap showing the significantly altered genes enriched in “chemokine signaling pathway.” Each column represents an individual group (n = 3 mice/group). The color scale represents the genewise z score calculated from the normalized gene expression levels. Genes are ordered by hierarchical clustering (A and C). (D) Validation of selected genes in a unique set of mice by qPCR (n = 4 mice/group). (E) Validation of CCL8 levels in the serum and hippocampal homogenates of each group by ELISA (n = 6 mice/group). (F) Representative images of FISH for Ccl8 (green) and the neuronal marker Eno2 (red) in hippocampal sections from the indicated mice. Scale bar: 20 μm. (G) Quantitative results for Ccl8⁺ neurons in the hippocampal sections (n = 7 mice/group). (H–J) Evaluation of depression-like behavior in pristane-treated Ccl8^fl/fl and Syn1^Cre;Ccl8^fl/fl mice (n = 9 mice/group). Ten weeks after pristane injection, the mice were subjected to the SPT (H), TST (I), and FST (J). The data are presented as mean ± SEM. *P < 0.05; ***P < 0.001, determined using 1-way ANOVA followed by Tukey’s post hoc test (D and E), or 2-tailed unpaired t test (H–J). NS, not significant; DEGs, differentially expressed genes; SPT, sucrose preference test; TST, tail suspension test; FST, forced swim test. See also Supplemental Figures 7 and 8.

Figure 4. IFN-γ/JAK/STAT1 signaling contributes to MSCT-mediated improvements in neuronal function in recipient MRL/lpr mice.

(A) The expression of IFN response genes was significantly altered by MSCT in the hippocampi of MRL/lpr mice (n = 3 mice/group). (B and C) qPCR analysis of Ifng and Ifitm3 expression in the spleens and hippocampi of the indicated mice (n = 5 mice/group). (D–F) Representative flow cytometric analysis of splenic CD4⁺IFN-γ⁺ (Th1) cells in the mice. The percentages (E) and numbers (F) of Th1 cells are shown in the right 2 panels (n = 5 mice/group). (G and H) Representative images of brain sections coimmunostained for p-STAT1 together with a neuronal marker (NeuN⁺) and quantification of positive cells in the CA1 region of the indicated groups (n = 5 mice/group). Scale bar: 20 μm. (I–K) IFN-γ (10 ng/mL) was used to treat hippocampal neurons. (I) p-JAK1, p-JAK2, and p-STAT1 levels were measured by Western blotting 30 minutes after treatment. (J) The expression of CCL8 was measured by ELISA 24 hours after treatment. (K) Ccl8 expression was assessed by RT-PCR 6 hours after treatment with IFN-γ and fludarabine. (L–O) Pharmacological inhibition of JAK1/2 or STAT1 in MRL/lpr mice. Ruxolitinib (Rux, p.o.) or fludarabine (Flu, i.p.) was given to MRL/lpr mice from 5 to 8 weeks of age. Immunostaining (L) and quantification (M) of p-STAT1⁺NeuN⁺ cells in the hippocampal sections were performed 3 weeks after pharmacological inhibitor treatment (n = 5 mice/group). Scale bar: 20 μm. The TST (N) and FST (O) were performed at the end of the experiment (n = 7–8 mice/group). The data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, determined using 1-way ANOVA followed by Tukey’s post hoc test (B, C, E, F, H, K, M and N) or Dunnett’s multiple comparisons test (O), or 2-tailed unpaired t test (J). Ctrl, control; TST, tail suspension test; FST, forced swim test. See also Supplemental Figure 9.

Figure 5. Blockage of IFN-γ signaling rescues depression phenotypes in MRL/lpr mice.

(A) Experimental protocol for model establishment and analysis. (B) qPCR analysis of Ifitm3 expression in the hippocampi of the indicated mice (n = 4 mice/group). (C and D) Representative images of brain sections coimmunostained for p-STAT1 together with a neuronal marker (NeuN⁺) and quantification of positive cells in the CA1 region of the indicated groups (n = 4 mice/group). Scale bar: 20 μm. (E and F) Quantification of IBA-1⁺CD68⁺ phagocytes (E) and CD68 staining intensity (F) in the hippocampi of the indicated groups (n = 5 mice/group). (G and H) Representative 3D reconstruction and quantification of engulfed PSD-95⁺ puncta (green) in CD68⁺ lysosomes (red) in the CA1 region of the indicated groups (n = 15–17 cells/group from 5 mice). Scale bar: 2 μm. (I and J) Immunostaining and quantification of presynaptic (SYP, red) and postsynaptic (PSD-95, green) proteins in hippocampal sections from the indicated mice (n = 5 mice/group, with an average of 3–4 slices per mouse). Scale bar: 10 μm. (K–M) The depression-like behavior of the indicated mice was assessed. The SPT (K), TST (L), and FST (M) were performed 3 weeks after antibody treatment (n = 8 mice/group). (N) Quantification of Golgi-stained dendritic spines of dentate gyrus granule neurons in MSCT (or MSCT plus IFN-γ–treated) MRL/lpr mice (n = 5 mice/group). (O and P) Depression-like behavior of the indicated mice was assessed via the TST (O) and the FST (P) (n = 8–10 mice/group). The data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, determined using 1-way ANOVA followed by Tukey’s post hoc test (B, D, H and J), or 2-tailed unpaired t test (E, F and K–P). IOD, integrated optical density; SPT, sucrose preference test; TST, tail suspension test; FST, forced swim test. See also Supplemental Figure 10.

Figure 6. Patients with SLE showed activation of the IFN-γ/JAK/STAT pathway and elevated CCL8 levels in CSF.

(A) ELISA revealed that patients with SLE (n = 15) had higher serum levels of IFN-γ than normal controls did (n = 11). (B) Flow cytometric analysis revealed that the number of CD4⁺IFN-γ⁺ cells in the peripheral blood was greater in patients with SLE (n = 15) than in normal controls (n = 11). (C and D) qPCR analysis showing the levels of IFNG and IFITM3 in sorted CD4⁺ T cells from patients with SLE (n = 15) and normal controls (n = 11). (E) Western blot analysis revealed that p-JAK1, p-JAK2, and p-STAT1 levels were increased in CD4⁺ T cells from patients with SLE. (F) Protein levels of IFN-γ in the cultured supernatants of CD4⁺ T cells cocultured with/without MSCs under Th1-polarizing conditions for 5 days (n = 5). (G and H) IFN-γ and CCL8 levels in CSF samples from controls (n = 6), non-NPSLE patients (n = 9), and NPSLE patients (n = 14). (I) Analysis of the correlation between CCL8 levels in the CSF and SLEDAI scores in patients with SLE (n = 23). The data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, determined using 2-tailed unpaired t test (A-D and F), or 1-way ANOVA followed by Sidak’s post hoc test (G and H). NS, not significant. The correlation coefficient r and P values were calculated via Spearman’s r test in I. SLEDAI, System Lupus Erythematosus Disease Activity Index. See also Supplemental Figure 11.

Figure 7. Graphical abstract.

Schematic diagram of the mechanism by which MSCs inhibit neuron-coordinated synaptic stripping to alleviate depression in lupus mice by targeting IFN-γ/JAK/STAT1/CCL8 signaling.