A Novel Microglia-Specific Transcriptional Signature Correlates With Behavioral Deficits in Neuropsychiatric Lupus

Abstract

Neuropsychiatric symptoms of systemic lupus erythematosus (NP-SLE) affect over one-half of SLE patients, yet underlying mechanisms remain largely unknown. We demonstrate that SLE-prone mice (CReCOM) develop NP-SLE, including behavioral deficits prior to systemic autoimmunity, reduced brain volumes, decreased vascular integrity, and brain-infiltrating leukocytes. NP-SLE microglia exhibit numerical expansion, increased synaptic uptake, and a more metabolically active phenotype. Microglia from multiple SLE-prone models express a "NP-SLE signature" unrelated to type I interferon. Rather, the signature is associated with lipid metabolism, scavenger receptor activity and downregulation of inflammatory and chemotaxis processes, suggesting a more regulatory, anti-inflammatory profile. NP-SLE microglia also express genes associated with disease-associated microglia (DAM), a subset of microglia thought to be instrumental in neurodegenerative diseases. Further, expression of "NP-SLE" and "DAM" signatures correlate with the severity of behavioral deficits in young SLE-prone mice prior to overt systemic disease. Our data are the first to demonstrate the predictive value of our newly identified microglia-specific "NP-SLE" and "DAM" signatures as a surrogate for NP-SLE clinical outcomes and suggests that microglia-intrinsic defects precede contributions from systemic SLE for neuropsychiatric manifestations.

Keywords: DAM; NP-SLE; SLE; behavior; interferon; lupus; microglia.

Figure 1

Behavioral deficits occur in young CReCOM mice. 3-4-month-old female MRL^lpr/lpr (n = 7), WT (n = 9), and CReCOM (n = 8) mice underwent behavioral testing. Data are combined from 3 independent experiments. (A) Morris water maze: % of mice that reached the hidden platform, latency to platform, and distance traveled. (B) Fear conditioning: % freezing following reintroduction of “shock” environment (CONTEXT) or tone associated with “shock” (CUE). (C) Prepulse inhibition: amplitude of acoustic startle response at 70, 80, 90, and 100 dB and % prepulse inhibition at 4, 12, and 20 KHz frequencies. (D) Rotarod: time spent on rod prior to drop and speed at drop during acceleration phase. (*p < 0.05; **p < 0.005).

Figure 2

CReCOM mice exhibit reduced brain volumes, decreased connectivity, and diminished vascular integrity. Four-month-old female MRL^lpr/lpr (n = 4) and 12-month-old female WT (n = 4) and CReCOM (n = 4) mice underwent MRI. (A) Brain volumes extracted from 3D MRI. (B) Body weight. (C) Fractional anisotropy (FA) maps extracted from diffusion MRI images. A set threshold value (FA ~ 400) generated semi-automated ROI. Volumes of voxels/region with FA values>~400 were determined and thresholded volumetric FA (TVFA) values are shown. (D) TVFA renderings overlaid on wired mouse head renderings. (E) Images of k^trans level (measure of vascular integrity) extracted from dynamic contrast-enhanced MRI (*p < 0.05).

Figure 3

Evidence of diffuse leukocyte infiltration in CReCOM brains. Brains of 11-12-month-old female WT (n = 4) and CReCOM (n = 4) mice were split into two portions and analyzed by flow cytometry. Data are representative of three independent experiments. (A) Whole brain gating strategy. (B) Regional breakdown for subsequent analysis. (C) Analysis of cortical region infiltration of T-cells (CD4⁺CD45^hi and CD8⁺CD45^hi), B-cells (CD19⁺CD45^hi), neutrophils (CD11b⁺Ly6G⁺CD45^hi), NK cells (CD11b⁺NK1.1⁺CD45^hi), macrophages (CD11b⁺CD64⁺CD45^hi), monocytes (CD11b⁺CD64⁻CD45^hi), as well as microglia (CD11b⁺CD64⁺CD45^lo). (D) Analysis of aforementioned populations in cerebellar region (*p < 0.05).

Figure 4

Microglia and infiltrating macrophages in CReCOM brains show distinct transcriptional profiles. FACS-purified macrophages of 11-12-month-old female WT (n = 3) and CReCOM (n = 3) mice and microglia of 11-12-month-old female WT (n = 4) and CReCOM (n = 4) mice were analyzed by RNA-seq. (A) Pairwise Pearson correlation of gene expression between macrophage and microglia samples from CReCOM and WT mice. (B) Volcano plot depicting the most differentially expressed genes (167) (DESeq2, false discovery rate (FDR) <10%, fold change>1.5) between WT and CReCOM macrophages. Downregulated genes (105) in CReCOM shown in green; upregulated genes (62) in CreCOM shown in purple. (C) GO processes associated with differentially expressed genes in CReCOM macrophages. (D) Heat map depicting relative expression [log₂[fold change compared to mean of WT replicates]] of the most differentially expressed genes (120) (DESeq2, FDR <10%, fold change>1.5) between WT and CReCOM microglia. Downregulated genes (46) in CReCOM shown in green; upregulated genes (74) in CReCOM shown in purple. Macrophage expression data for microglia-specific differentially expressed genes are depicted. (E) GO processes associated with differentially expressed genes in CReCOM microglia.

Figure 5

Evidence of a common “NP-SLE signature” in microglia. FACS-purified microglia of 11-12-month-old female WT (n = 4), and CReCOM (n = 4) mice were analyzed by RNA-seq and compared to data from 8-10-month-old female B6 and B6.Sle1Sle3 mice (17). (A) A significant overlap exists between upregulated genes (DESeq2, p < 0.05, fold change>1.5) in CReCOM (256) and B6.Sle1Sle3 (214) microglia (p < 2.54 × 10⁻⁴, hypergeometric distribution). (B) Expression values (counts per million=CPM) for shared genes. (C) GO processes associated with shared genes.

Figure 6

Expression of DAM-associated genes in NP-SLE microglia. FACS-purified microglia of 11-12-month-old female WT (n = 4), and CReCOM (n = 4) mice were analyzed by RNA-seq and compared to data from 8-10-month-old female B6 and B6.Sle1Sle3 mice (17). Differential genes (DESeq2, p < 0.05, fold change>1.5) in CReCOM (581) and B6.Sle1Sle3 (472) microglia compared to their respective controls were determined. Differential genes (fold change in average UMI count>1.5) between Microglia3 (DAM) and Microglia1 populations (48) were determined (29). (A) Enrichment of significantly upregulated (UP) and downregulated (DOWN) genes in microglia or macrophages from NP-SLE models in the over-(OVER) or under-(UNDER)-expressed genes in DAMs. Enrichment=ratio of the observed number of genes that overlap DAM-associated genes divided by the expected number. Dotted line at enrichment value of “1” denotes observed number equals expected number of overlapping genes due to chance. P-values reflect hypergeometric distribution. (B) Expression values (CPM) for shared genes in microglia from NP-SLE-like disease models and DAM, termed “DAM signature.” (C) GO processes associated with 15-gene “DAM signature.” (D) Overlapping genes between upregulated genes in “NP-SLE” and “DAM” signatures (p < 2.70 × 10⁻⁷, hypergeometric distribution).

Figure 7

Expression of “NP-SLE” and “DAM” signatures correlates with behavioral deficits in young CReCOM mice. FACS-purified microglia of 5-6-month-old and 11-12-month-old female WT (n = 4) and CReCOM mice (n = 4) mice were analyzed by RNA-seq. Correlation coefficients (R) were determined between behavior (y-axis) and gene expression (x-axis) scores. Lines suggest pattern of correlation. Scatter plots depicting correlation between (A) “NP-SLE signature” or (B) downregulated and upregulated “DAM signature” gene expression scores and behavior scores.