Article Figures & Data
Figures
- Figure 1. Cx3cr1 loss alters the transcriptome of microglia from 2-mo-old mice.
Microglia were isolated from 2-mo-old Cx3cr1+/+ (WT, n = 8), Cx3cr1+/eGFP (Het, n = 5), or Cx3cr1eGFP/eGFP (KO, n = 8) mice. The RNA isolated from microglia of each mouse was used to generate an individual RNA-seq data set. (A) The PCA indicates that the samples separate by genotype in the first principal component and gender in the second. Each dot represents microglia from an individual mouse. (B) Most of the genes differentially expressed between Het and WT microglia are also differentially expressed by KO and WT microglia. (C) Unsupervised cluster analysis of the top 200 DEGs in microglia from 2-mo-old WT (n = 8), Het (n = 5), or KO (n = 8) mice. Most DEGs are expressed at lower levels in Het and KO samples compared to WT samples. Selected genes and the pathways they are associated with (manual annotation) are represented in different colors.
- Figure S1. qRT-PCR analysis of the expression of selected DEGs identified by RNA-seq.
Microglia were isolated from 2-mo-old Cx3cr1+/+ (WT, n = 4), Cx3cr1+/eGFP (Het, n = 4), or Cx3cr1eGFP/eGFP (KO, n = 4) mice. The RNA isolated from microglia of each mouse and used for qRT-PCR analysis. (A) Selected genes that show changes in expression after loss of Cx3cr1 by RNA-seq and confirmed by qRT-PCR. (B) Examples of genes that are not modulated in microglia by Cx3cr1 genotype. Egr1 was identified as a DEG, but Trem2 was not differentially expressed.
- Figure S2. Cx3cr1 loss does not alter the expression of microglia signature genes.
Microglia were isolated from 2-mo-old Cx3cr1+/+ (WT, n = 8), Cx3cr1+/eGFP (Het, n = 5), or Cx3cr1eGFP/eGFP (KO, n = 8) mice. The RNA isolated from microglia of each mouse was used to generate an individual RNA-seq data set. Unsupervised cluster analysis of microglial signature genes as defined by Butovsky et al (2014). Selected genes and the pathways they are associated with (manual annotation) are represented in different colors.
- Figure 2. Active promoter landscape of Cx3cr1-KO and –WT samples shows minor differences in Rbp2 binding.
Microglia were isolated from 2-mo-old Cx3cr1+/+ (WT) or Cx3cr1eGFP/eGFP (KO) mice, pooling brains from three gender-matched mice to obtain samples for chromatin immunoprecipitation followed by sequencing. (A) ChIP-seq for Rbp2, a mark of actively transcribed genes, indicates few significant differences between the genotypes. Only genes with at least four sequencing reads in at least one of the genotypes are shown, with genes reaching statistical significance highlighted in blue. (B) There is minimal correlation between mRNA expression (fold change determined by RNA-seq) and active transcription (Rbp2 binding determined by ChIP-seq). Genes with fold change > |1.5| are highlighted in coral. (C) ChIP-seq for H3K27Ac, a mark of open and active chromatin, indicates few significant differences between the genotypes. Only genes with at least four sequencing reads in at least one of the genotypes are shown, with genes reaching statistical significance highlighted in blue and coral. (D) The UCSC browser snapshots of Rbp2 binding for selected genes. Exons are represented by boxes, and direction of transcription is indicated by arrows in introns.
- Figure 3. Loss of Cx3cr1 does not affect microglial transcriptional response to LPS.
2-mo-old Cx3cr1+/+ (WT, n = 7), Cx3cr1+/eGFP (Het, n = 6), or Cx3cr1eGFP/eGFP (KO, n = 6) mice were injected with 1 mg/kg LPS i.p. (or saline control, four WT and three KO animals only). Twenty-four hours later, peritoneal cells were collected, and microglia were sorted from isolated brains. Both cell populations were used for RNA-seq. (A) PCA of all samples. Each dot within a cell type represents an individual animal, but there are matched microglia and peritoneal cells from the same animal. Samples, coded based on the cell type (open or solid fill), genotype (color), and treatment (symbol), separate by cell type and treatment. (B) The PCA of microglia only shows separation by treatment, but not genotype in LPS-injected mice. Samples are coded by gender (open or solid fill), genotype (color), and treatment (symbol). (C) Expression of selected genes that are modulated by Cx3cr1 genotype after LPS injection.
- Figure S3. Loss of Cx3cr1 does not affect microglial transcriptional response to LPS.
2-mo-old Cx3cr1+/+ (WT, n = 7), Cx3cr1+/eGFP (Het, n = 6), or Cx3cr1eGFP/eGFP (KO, n = 6) mice were injected with 1 mg/kg LPS IP (or saline control, four WT and three KO animals only). Twenty-four hours later, microglia were sorted from isolated brains and subjected to RNA-seq analysis. Unsupervised cluster analysis of microglial of top 200 DEGs (saline versus LPS comparison) shows a strong effect of LPS, but no clear separation of the samples by genotype after LPS treatment.
- Figure 4. Effect of aging and Cx3cr1 deletion on microglial transcriptome.
Microglia were isolated from 2-mo-, 1-yr-, or 2-yr-old Cx3cr1+/+ (WT), Cx3cr1+/eGFP (Het), or Cx3cr1eGFP/eGFP (KO) mice. (A) Fewer microglia were recovered from 2 yr mice, but the recovery was independent of genotype. (B, C) RNA-seq was performed on all samples together. Each symbol represents microglia from an independent animal. (B) The PCA for all samples suggests separation by age as the primary factor driving variability, with genotype second and gender third (B). (C) For microglia from 2-yr-old mice (C), the samples separate by genotype and by gender within each genotype. (D) Overlap of DEGs between 2-yr and 2-mo microglia in WT and in KO mice. There is a large amount of shared DEGs between the genotypes.
- Figure 5. Young Cx3cr1-deficient mice display a premature aging phenotype.
Microglia were isolated from 2-mo-, 1-yr-, or 2-yr-old Cx3cr1+/+ (WT), Cx3cr1+/eGFP (Het), or Cx3cr1eGFP/eGFP (KO) mice and used for RNA-seq analysis, followed by quantification and unsupervised hierarchical clustering for the top 100 DEGs (KO versus WT, by P-value) at each time point, resulting in a combined list of 254 genes. In 2-mo-old mice, the samples separate by genotype, whereas in older mice, they first separate by age and then genotype. Overall, 2-mo-old Cx3cr1-Het and KO microglia resemble aged mice of any genotype. Some genes increase with both age and Cx3cr1 deletion and show the highest expression in aged Cx3cr1-KO microglia.
- Figure S4. Unsupervised hierarchical clustering by age shows that similar pathways are modulated by aging as by Cx3cr1 loss.
Microglia were isolated from 2-mo-, 1-yr-, or 2-yr-old Cx3cr1+/+ (WT), Cx3cr1+/eGFP (Het), or Cx3cr1eGFP/eGFP (KO) mice and used for RNA-seq analysis, followed by quantification and unsupervised hierarchical clustering for the top 200 DEGs (by P-value) between WT 2-yr- versus 2-mo-old mice. Aging alters the expression of genes in pathways modulated by Cx3cr1 deletion as well. Although separation by genotype is visible in microglia form 2-mo-old mice, it is lost in aged mice.
- Figure 6. Increased inflammatory gene expression in microglia from 2-yr female mice.
Quantification and unsupervised hierarchical clustering for microglia from 2-yr-old Cx3cr1+/+ (WT, n = 9), Cx3cr1+/eGFP (Het, n = 8), or Cx3cr1eGFP/eGFP (KO, n = 11) mice highlights a strong effect of gender on gene expression in old mice. The samples primarily separate by gender and not Cx3cr1 genotype. Selected genes and associated pathways (manual annotation) are represented in different colors.
- Figure 7. Cx3cr1 deletion alters microglial morphology in young and aged mice.
(A, B) Representative images of brain sections from the cortex (A) or cerebellum (B) of young (2 mo) or old (18–24 mo) Cx3cr1+/+ or Cx3cr1eGFP/eGFP male or female mice immunoreacted with anti-Iba1 (purple) and anti-Pu.1 (brown). Note the overall different microglial morphology in the cerebellum in both young and old mice compared with the cortex. Many microglial clusters were detectable in the cerebellum of old animals of either genotype. Scale bar: large image, 200 μm; inset, 50 μm. (C, D) Quantification of microglial cell numbers and morphology in the cortex (C) and the cerebellum (D). Microglial cell numbers were quantified as the number of Pu.1-positive nuclei per unit area. Microglial morphology was assessed as the total Iba1 immunoreactive area per Pu.1+ nucleus, dendritic process area per Pu.1+ nucleus, and dendritic process perimeter per Pu.1+ nucleus. Only processes attached to nuclei were included in the dendritic process calculations. Statistics: two-way ANOVA (for genotype and age) and Tukey’s post hoc test. Only select significant intergroup comparisons are shown. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Ctx cell number: genotype: F(1,32) = 4.892, P = 0.0342; age: F(1,32) = 0.0001061, P = 0.9918; interaction: F(1,32) = 0.61, P = 0.4405. Ctx Iba1 area: genotype: F(1,32) = 8.461, P = 0.0065; age: F(1,32) = 3.225, P = 0.0820; interaction: F(1,32) = 2.182, P = 0.1494. Ctx process area: genotype: F(1,32) = 9.908, P = 0.0035; age: F(1,32) = 5.665, P = 0.0234; interaction: F(1,32) = 3.165, P = 0.0847. Ctx process perimeter: genotype: F(1,32) = 11.31, P = 0.0020; age: F(1,32) = 2.868, P = 0.1000; interaction: F(1,32) = 2.777, P = 0.1054. Cbm cell number: genotype: F(1,32) = 4.982, P = 0.0327; age: F(1,32) = 31.08, P < 0.0001; interaction: F(1,32) = 1.119, P = 0.2817. Cbm Iba1 area: genotype: F(1,32) = 2.602, P = 0.1165; age: F(1,32) = 6.502, P = 0.0158; interaction: F(1,32) = 0.6807, P = 0.4154. Cbm process area: genotype: F(1,32) = 0.5592, P = 0.4601; age: F(1,32) = 113, P < 0.0001; interaction: F(1,32) = 2.322, P = 0.1374. Cbm process perimeter: genotype: F(1,32) = 0.5271, P = 0.4731; age: F(1,32) = 118.5, P < 0.001; interaction: F(1,32) = 2.745, P = 0.1073.
- Figure S5. Cx3cr1 deletion alters microglial morphology in the HC and striatum of young and aged mice.
(A, B) Representative images of the brain section from the HC (A) or striatum (B) of young (2 mo) or old (18–24 mo) Cx3cr1+/+ or Cx3cr1eGFP/eGFP male or female mice immunoreacted with anti-Iba1 (purple) and anti-Pu.1 (brown). Scale bar: large image, 200 μm; inset, 50 μm. (C, D) Quantification of microglial cell numbers and morphology in the HC (C) and the striatum (D). Microglial cell numbers were quantified as the number of Pu.1-positive nuclei per unit area. Microglial morphology was assessed as the total Iba1 immunoreactive area per Pu.1+ nucleus, dendritic process area per Pu.1+ nucleus, and dendritic process perimeter per Pu.1+ nucleus. Only processes attached to nuclei were included in the dendritic process calculations. Statistics: two-way ANOVA (for genotype and age) and Tukey’s post hoc test. Only select significant intergroup comparisons are shown. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. HC cell number: genotype: F(1,32) = 2.159, P = 0.1515; age: F(1,32) = 8.952, P = 0.0053; interaction: F(1,32) = 0.1583, P = 0.6934. HC Iba1 area: genotype: F(1,32) = 11.1, P = 0.0022; age: F(1,32) = 5.603, P = 0.0241; interaction: F(1,32) = 1.685, P = 0.2036. HC process area: genotype: F(1,32) = 2.69, P = 0.1108 Age: F(1,32) = 20.21, P < 0.0001; interaction: F(1,32) = 1.566, P = 0.2199. HC process perimeter: genotype: F(1,32) = 3.972, P = 0.0548; age: F(1,32) = 16.07, P = 0.003; interaction: F(1,32) = 1.647, P = 0.2086. Str cell number: genotype: F(1,32) = 0.005704, P = 0.9403; age: F(1,32) = 14.24, P = 0.0007; interaction: F(1,32) = 0.9689, P = 0.3323. Str Iba1 area: genotype: F(1,32) = 5.419, P = 0.0264; age: F(1,32) = 1.819, P = 0.1870; interaction: F(1,32) = 8.379, P = 0.0068. Str process area: genotype: F(1,32) = 5.746, P = 0.0225; age: F(1,32) = 34.86, P < 0.0001; interaction: F(1,32) = 10.53, P = 0.0027. Str process perimeter: genotype: F(1,32) = 8.114, P = 0.0076; age: F(1,32) = 27.4, P < 0.0001; interaction: F(1,32) = 10.9, P = 0.0024.
- Figure S6. Effect of gender on the morphology of microglia in young and old mice.
Sections from young (2 mo) or old (18–24 mo) Cx3cr1+/+ or Cx3cr1eGFP/eGFP male or female mice were immunoreacted with anti-Iba1 (purple) and anti-Pu.1 (brown). (A, B, C, D) Quantification of microglial cell numbers and morphology in the cortex (A), cerebellum (B), HC (C) and the striatum (D). Microglial cell numbers were quantified as the number of Pu.1-positive nuclei per unit area. Microglial morphology was assessed as dendritic process area per Pu.1+ nucleus. Only processes attached to nuclei were included in the dendritic process calculations. Statistics: Two-way ANOVA (for genotype and age) and Tukey’s post-hoc test. Only select significant intergroup comparisons are shown. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Ctx Cell number: Genotype: F(3,28) = 2.136, P = 0.1181; Age: F(1,28) = 0.138, P = 0.2581; Interaction: F(3,28) = 1.419, P = 0.2581. Ctx Process area: Genotype: F(3,28) = 5.209, P = 0.0055; Age: F(1,28) = 5.439, P = 0.0271; Interaction: F(3,28) = 9.132, P = 0.0002. Cbm Cell number: Genotype: F(3,28) = 2.752, P = 0.0613; Age: F(1,28) = 36.38, P < 0.0001; Interaction: F(3,28) = 0.8772, P = 0.4647. Cbm Process area: Genotype: F(3,28) = 1.401, P = 0.2631; Age: F(1,28) = 140.5, P < 0.0001; Interaction: F(3,28) = 6.773, P = 0.0014. HC Cell number: Genotype: F(3,28) = 1.309, P = 0.2909; Age: F(1,28) = 6.797, P = 0.0145; Interaction: F(3,28) = 0.9511, P = 0.4294. HC Process area: Genotype: F(3,28) = 1.124, P = 0.3561; Age: F(1,28) = 18.97, P = 0.0002; Interaction: F(3,28) = 3.947, P = 0.0.0182. Str Cell number: Genotype: F(3,28) = 0.3224, P = 0.8091; Age: F(1,28) = 11.39, P = 0.0022; Interaction: F(3,28) = 0.4544, P = 0.7163. Str Process area: Genotype: F(3,28) = 2.551, P = 0.0757; Age: F(1,28) = 36.56, P < 0.0001; Interaction: F(328) = 8.065, P = 0.0005.
Tables
Pathway 2 mo 1 yr 2 yr No. of up No. of down No. of up No. of down No. of up No. of down 1 Immune response 14 34 8 9 12 11 2 Transcription factor binding 27 5 4 2 4 19 3 Regulation of immune response 11 27 9 4 12 9 4 Regulation of response to external stimulus 11 25 5 8 6 14 5 Positive regulation of the immune system process 16 28 10 4 12 11 6 External side of the plasma membrane 3 20 5 6 5 12 7 Leukocyte migration 5 16 1 5 3 6 8 Vacuole 15 14 9 6 15 8 9 Regulation of cell activation 9 24 7 6 10 9 10 Lysosome 14 14 8 6 13 6 11 Cell migration 18 26 7 5 9 16 12 Response to wounding 14 27 7 6 12 11 13 Endosome 15 24 5 9 12 15 14 Carbohydrate binding 4 23 2 1 3 7 15 Leukocyte chemotaxis 2 12 0 4 2 5 16 Regulation of lymphocyte activation 8 20 7 5 8 7 17 Regulation of defense response 9 22 2 5 4 10 18 Regulation of locomotion 14 21 7 6 7 14 19 G-protein–coupled peptide receptor activity 2 11 2 4 2 5 20 Regulation of cell migration 13 19 7 6 7 14 Top 20 pathways affected by Cx3cr1 genotype (KO versus WT) are shown, with the number of genes up-regulated or down-regulated at each time point. Pathways showed an overall increase (tan highlight) or decrease (blue highlight) in expression. Only genes with FDR < 0.05, fold change > 30%, and average expression > 4 TPM were included in the enrichment analysis.
