Skip to main content
PMC home page
Scientific Reports logoLink to Scientific Reports
. 2021 Jul 22;11:14935. doi: 10.1038/s41598-021-94324-z

Modest changes in Spi1 dosage reveal the potential for altered microglial function as seen in Alzheimer’s disease

Ruth E Jones 1,2, Robert Andrews 1,4, Peter Holmans 3, Matthew Hill 2, Philip R Taylor 1,2,4,✉
PMCID: PMC8298495  PMID: 34294785

Abstract

Genetic association studies have identified multiple variants at the SPI1 locus that modify risk and age of onset for Alzheimer’s Disease (AD). Reports linking risk variants to gene expression suggest that variants denoting higher SPI1 expression are likely to have an earlier AD onset, and several other AD risk genes contain PU.1 binding sites in the promoter region. Overall, this suggests the level of SPI1 may alter microglial phenotype potentially impacting AD. This study determined how the microglial transcriptome was altered following modest changes to Spi1 expression in primary mouse microglia. RNA-sequencing was performed on microglia with reduced or increased Spi1/PU.1 expression to provide an unbiased approach to determine transcriptomic changes affected by Spi1. In summary, a reduction in microglial Spi1 resulted in the dysregulation of transcripts encoding proteins involved in DNA replication pathways while an increased Spi1 results in an upregulation of genes associated with immune response pathways. Additionally, a subset of 194 Spi1 dose-sensitive genes was identified and pathway analysis suggests that several innate immune and interferon response pathways are impacted by the concentration of Spi1. Together these results suggest Spi1 levels can alter the microglial transcriptome and suggests interferon pathways may be altered in individuals with AD related Spi1 risk SNPs.

Subject terms: Inflammation, Monocytes and macrophages, Neuroimmunology, Alzheimer's disease

Introduction

Alzheimer’s Disease (AD) is the most prevalent form of Dementia, affecting millions of people worldwide1. Studies investigating AD genetics and pathology have suggested immune gene networks may contribute to an increased risk of developing AD2,3. SPI1 encodes PU.1, a central transcription factor in microglial development and activation, and has a genome-wide significant genic association with AD in the IGAP GWAS (rs3740688 Odds Ratio 0.92 Meta p = 5.4 × 10−13), comprising 35,274 Alzheimer’s disease cases and 59,163 controls4. In addition, a 56-protein interaction network consisting of strongly enriched rare coding variants (p = 1.08 × 10–7, and common variants with Late-Onset AD gene-wide significance (p = 2.98 × 10–7) identified SPI1 as a central hub gene5. Several studies have suggested SPI1/PU.1 levels impact on the microglial transcriptome, therefore affecting the phenotype of these cells. Several single-nucleotide polymorphisms (SNPs) associated with an increased risk of AD are thought to lie within the Spi1 gene locus3,6,7. The SNP variant rs1057233a is thought to result in a higher level of Spi1 expression and earlier age of AD onset7. Moreover, Spi1 is thought to influence the expression of other AD risk genes5,7,8.

The PU.1 transcription factor is essential for the survival and function of macrophages9–13 and is well conserved between humans and mice (> 85% protein similarity, BLAST protein alignment (RefSeq ID’s NP_003111.2 and NP_035485, respectively). In hematopoietic development low levels of PU.1 drives B-lymphocyte development whereas cells expressing high levels of PU.1 are committed to the myeloid lineage14–16. In development PU.1 levels are regulated to commit precursors to a macrophage or B-cell lineage17–23. In these early experiments PU.1 appeared to have dose-dependent transcriptional thresholds in foetal liver macrophages24. PU.1 also regulates expression of several key macrophage receptors such as CSF1R, CD11b and CD4525–27. Moreover PU.1 interacts with other lineage-determining factors such as C/EPBα/β to alter the chromatin landscape resulting in a specialised macrophage epigenome28–30.

In both primary human microglia and the BV-2 mouse microglia cell line, reductions in PU.1 have resulted in changes to gene expression and a reduced phagocytic capacity7,31,32. Additionally, increased PU.1 expression in BV-2 cells resulted in increased zymosan phagocytosis, and amplified both ROS, NO and cytokine production after LPS stimulation32. Though the impact of altered Spi1 on the microglial transcriptome would ideally be studied in freshly isolated cells, such as those from a transgenic mouse model, at the time of this study no appropriate Spi1 over-expression transgenic mouse lines were available.

CSF1R inhibitors are a potential AD therapeutic and prevent AD-associated microgliosis by blocking the CSF1R/PU.1 survival signalling pathway33–36 but the impact on peripheral macrophages has not been reported. Knowing how subtle changes in SPI1/PU.1 levels contribute to the microglial tissue resident subset transcriptome could allow more specific pathways could be targeted.

RNA-sequencing was used to assess changes to the microglial transcriptome following modest changes to Spi1 expression in microglia from primary mixed-glia cultures. Modest changes to PU.1 protein were desirable as they reflect the expression changes caused by the risk allele, and therefore the biology underlying AD. A moderate reduction in microglial Spi1 resulted in altered expression of cell cycle related genes while small Spi1 increases upregulated immune response genes. A subset of genes identified as Spi1 dose-sensitive highlighted a potential dose-dependent interferon driven immune response regulated by Spi1.

Results

RNA-sequencing data shows impact of Spi1 dose on microglia transcriptome

The effect of Spi1 shRNA knock down (KD) and Spi1 pSIEW over expression (OE) was assessed in flow cytometric sorted microglia after 11 days of lentivirus infection. Following Spi1 knock-down (Spi1 shRNA compared to control shRNA) 1615 genes were up- and 1462 down-regulated, using an adjusted p value of < 0.05 cut-off (Benjamini–Hochberg corrected for multiple testing). A proportion of down-regulated genes (230 genes) surpassed the -2 log2 fold-change cut-off (Fig. 1A). In the Spi1 over-expression dataset (Spi1 pSIEW compared to pSIEW) 284 genes had an increased expression, 62 were down-regulated (p < 0.05). In this comparison 71 up-regulated and 4 down-regulated exceeded a log2 fold-change value of 2. Principal component analysis (PCA, Supplementary Fig. 1) confirmed that samples within the same group had greater similarity, while the Spi1 pSIEW and Spi1 shRNA samples clustered separately.

Figure 1.

Figure 1

Open in a new tab

Results of RNA-Seq experiment in primary mixed-glial cultures (A) Volcano plots summarising the distribution of genes in the Spi1 knock-down (purple) and Spi1 over-expression (orange). In the Spi1 knock-down dataset a 1462 genes were down-regulated (grey background) and 1615 up-regulated with a p value of ≤ 0.05 (solid lines), as indicated by the numbers on the graph. In the Spi1 over-expression dataset the majority of the genes, 284, were up-regulated and 62 genes had down-regulated expression. (B) Genes that were significantly changed in both the Spi1 knock down (KD, purple) and the Spi1 over expression (OE, orange) using a p ≤ 0.05 threshold. A Plot of adjusted p values from all genes in both datasets. Genes that were below the p ≤ 0.05 threshold in the Spi1 knock-down dataset are highlighted in purple, those that were below the p ≤ 0.05 cut-off in the Spi1 over-expression are within the orange bar. In the bottom left corner the red line surrounds the 196 genes that were significantly changed in both datasets. (C) Venn diagrams summarising the gene expression that were significantly changed (p ≤ 0.05) in either the Spi1 knock-down dataset (purple), the Spi1 over-expression dataset (orange) or changed in both datasets (red). Of interest were the genes with expression that appears to be sensitive to the Spi1 dose in microglia, namely the 194 genes that appear to be expressed relative to the dose of Spi1. (D) The log10 number of Spi1 mRNA fragments per Kilobase of transcript per Million mapped reads (FPKM) shows that Spi1 was lower in Spi1 shRNA infected microglia than in control shRNA infected microglia (One-tailed Unpaired T-test on the log10 transformed data, p = 0.0166, n = 4 for control shRNA and n = 3 for Spi1 shRNA). PU.1 protein expression was normalised to the non-infected (NI) samples in each experiment (as described in methods). Microglia infected with the Spi1 shRNA (solid purple) have reduced PU.1 expression compared to cells infected with a control shRNA (lined purple) (One-tailed Paired T-test, p = 0.0013, n = 3 per group). (E) Spi1 mRNA expression (FPKM) was increased in the samples infected with the Spi1 over-expression construct (filled orange) compared to empty vector control samples (outline orange), (One-tailed Paired T-test on the Log10 transformed data, p = 0.0013, n = 4 per group).The Spi1 over-expression virus (Spi1 pSIEW) increased PU.1 protein expression in microglia by roughly half compared to cells infected with an empty vector control (lined orange). One-tailed Paired T-test, p = 0.0131, n = 4 per group) (D–E) Each dot represents the value from one biological replicate, the means are indicated by the horizontal lines and the error bars display the standard deviation about the mean. All experiments were performed using mixed glia cultures from 8-week-old female C57BL/6J mice. Figures (A–C) were made in GraphPad PRISM 6 (version 3.07) and (D–E) in GraphPad PRISM 8 (version 8.4.3; both GraphPad Software, Inc.).

The datasets with altered Spi1 expression were then compared to identify which subset of genes were likely affected by Spi1/PU.1 in a dose-dependent manner. This identified 196 genes which were significantly diminished in the Spi1 knock-down and upregulated in the Spi1 over-expression dataset (Adjusted p value of ≤ 0.05; Fig. 1B). When the direction of the gene fold-changes was compared 162 of these genes were lower in the dataset with lower Spi1 and higher in the Spi1 over-expression dataset and 32 genes where expression was increased following Spi1 knock-down and reduced in the Spi1 over-expression dataset (Fig. 1C). Therefore, 194 genes were classed as Spi1 dose-sensitive (Supplementary Table 1).

RNA-sequencing confirmed Spi1 fragments per Kilobase of transcript per Million mapped reads (FPKM) values were significantly lower in cultures infected with Spi1 shRNA as opposed to the control shRNA (Fig. 1D, One-Tailed Unpaired T-test on Log10 data p = 0.0166). Spi1 mRNA expression was increased in the Spi1 pSIEW samples compared to the control virus (Fig. 1E, One-Tailed Paired T-test on Log10 data p = 0.0013).

PU.1 protein was assessed in independent microglia samples (Fig. 1D–E) and was reduced by ~ 70–80% in Spi1 shRNA samples compared to control shRNA infected microglia (Fig. 1D One-tailed Paired T-test, p = 0.0013 and Supplementary Fig. 3B One-tailed Paired T-test, p = 0.001). Spi1 over-expression in microglia increased PU.1 protein expression by ~ 70% compared to microglia infected with pSIEW control virus (One-tailed Paired T-test, p = 0.0131). The values clearly cluster within each biological group and the means of Spi1 shRNA and Spi1 pSIEW microglia compared to the appropriate control viruses are disparate indicating PU.1 was altered by these Spi1 modulating viruses.

Gene ontology analysis in Spi1 knock-down and Spi1 over-expression datasets

Differentially expressed genes (p < 0.05 significance threshold) in the Spi1 KD and Spi1 OE datasets were separately assessed using DAVID (Database for Annotation, Visualization and Integrated Discovery, version 6.8). Figure 2 displays the Gene Ontology (GO) terms from the 20 most significantly enriched pathways in each dataset (corrected p values p < 0.05, FDR). In primary microglia with a lower Spi1 expression the most significantly changed pathways were related to cell cycle and DNA repair (Fig. 2A). However, microglia with higher Spi1 (Fig. 2B) had an over representation of genes related to antigen presentation pathways, immune system processes and response to interferon.

Figure 2.

Figure 2

Open in a new tab

The 20 most siginificantly changed biological pathways in the Spi1 knock-down (A) and Spi1 over-expression (B) datasets assessed using DAVID. In these graphs the Benjamini–Hochberg adjusted p value(− Log10) is displayed on the x-axis, the Gene Ontology (GO) term listed on the y-axis, the percentage of the gene list in each cluster is denoted by the size of the bubble and the colour denotes the fold-change, where a darker colour indicates a stronger enrichment. The vertical black line indicates the p value cut-off of 0.05. Bubble plots were made using the ‘tidyverse’37 and ‘ggpubr’38 packages in R39,40.

Spi1 dose-sensitive subset highlight interferon response pathways

The absolute log2 fold-changes of 162 genes reduced by Spi1 knock-down and increased by Spi1 over-expression significantly correlated to each other (Fig. 3A) and the 32 genes that were increased by Spi1 knock-down and reduced by Spi1 over-expression (Fig. 3B) were also found to correlate (Two-tailed Spearman Rank Test approximate p < 0.0001 and r > 0.8 for both). This further supports the hypothesis that the expression of these genes is controlled in a Spi1 dose-sensitive manner. Collective analysis of the 194 Spi1 dose-sensitive gene list using GO terms through DAVID indicated an enrichment of immune response genes (Fig. 3C), particularly pathways linked to interferon and viral defence responses. These results suggest that a high microglial Spi1 expression may result in more responsive microglia that produce more interferon, though further work would be required to confirm this experimentally.

Figure 3.

Figure 3

Open in a new tab

Absolute Log2 fold-changes in Spi1 dose-sensitive genes. (A) Absolute log2 fold-change values were used the 162 genes where p ≤ 0.05 with negative fold-change values in the Spi1 knock-down dataset (purple) and positive fold-change values in the Spi1 over-expression dataset (orange) were assessed separately from the 32 genes where p ≤ 0.05 and Spi1 dose had the opposite effect on expression (B). Two-tailed Spearman Rank test identified significant correlations between the log2 fold-changes in both graphs (approximate p < 0.0001 and r > 0.8). (C) Top 10 most significantly altered “Biological Processes” according to the Benjamini–Hochberg adjusted p value. The bubble size indicates the percentage of the gene list aligned to this pathway. The colour indicates the fold enrichment, which is the proportion of genes present in this list compared to the background gene expression. The vertical black line indicates the p value threshold of 0.05. Figures (A, B) were made in GraphPad PRISM 6 (version 3.07) and (C) using the ‘tidyverse’37 and ‘ggpubr’38 packages in R39,40.

Several gene clusters implicate immune system responses

The data were assessed using hierarchical clustering of the FPKM expression values (Fig. 4). Six discrete clusters were identified, and gene lists were assessed using DAVID as before. Genes with an increased expression in microglia over-expressing Spi1 (Cluster 1) are enriched for pathways involving MHCII processing and presentation of antigens. The genes in Cluster 2 appear to be the most Spi1 sensitive, where expression was reduced in the Spi1 shRNA and increased in the Spi1 over-expression samples compared to relative controls. Most pathways associated with the genes in this cluster are related to the immune response including the response to viruses. Cluster 3 contained genes that were down-regulated in the Spi1 shRNA samples compared to the controls and were related to interferon signalling and immune response pathways. Cluster 5 contains genes that had a higher expression in the Spi1 shRNA samples compared to the control shRNA samples. The GO TERMS associated with the genes in this cluster include those related to cell cycle and DNA replication pathways. Overall, this reinforces previous reports that reduced Spi1 seems to impact cell cycle which was not unexpected given the involvement of PU.1 in survival signalling11,41. A lowering of Spi1 appeared to reduce activation of immune response related pathways while increasing Spi1 expression seemed to have the opposite effect.

Figure 4.

Figure 4

Open in a new tab

Hierarchical clustering analyses with the 20 most siginificantly changed biological pathways each cluster assessed using DAVID. (A) Hierarchical clustering of rows, where each row represents the scaled log10 FPKM values for each gene, with dendrogram higlighting cluster boundaries. The Pearson correlation was used to calculate the Z-Scores and UPGMA agglomeration methods were used. These 6 clusters were produced using the cutree function at a height of 1.59. (B) The gene lists were assessed using DAVID and a bubble plot created to highlight the top 5 most significatly enriched pathways (p value threshold of < 0.05) in each gene cluster. Cluster 1 contains genes that had increased expression following a higher Spi1 expression and are mainly linked to immune responses and MHCII antigen presentation. Cluster 2 appears to contain more of the Spi1 dose-sensitive genes and are enriched for viral/immune response signalling pathways. Genes in Cluster 3 had a lower expression in the Spi1 shRNA dataset and are related to immune response and interferon signalling pathways. Cluster 4 contained no significantly enriched pathways using the p ≤ 0.05 threshold. Cluster 5 contains multiple genes related to cell cycle and DNA replication pathways, whose expression was increased in the Spi1 shRNA samples. The bubble size denotes the fold enrichment and the colour Benjamini–Hochberg adjusted p values. Figures were made with the ‘tidyverse’37, ‘gplots’42, ‘plotly’43, ‘dendextend’44 and ‘colourspace’45 packages in RStudio39,40, code adapted from46,47 complete Markdown is available in Supplementary Information.

PU.1 dose-related transcriptomic differences were not limited by direct binding

Changes in gene expression after manipulation of PU.1 could arise from changes in direct DNA binding or via secondary effects. To identify possible direct targets of PU.1 that were differentially expressed, the proportion of protein coding genes that contained a PU.1 binding site sequence, especially within the promoter region, were assessed utilising a published ChIP-Sequencing (ChIP-Seq) dataset available online30. Figure 5 shows ~ 70% of genes differentially expressed in the Spi1 RNA-Seq datasets contained a PU.1 binding site and therefore expression was potentially directly modified by Spi1. Moreover, 30–40% of genes in Spi1 KD and Spi1 OE datasets had PU.1 binding sites within the promoter region suggesting their expression was directly altered by Spi1/PU.1. These proportions remained consistent even in the Spi1 dose-sensitive subset. Only differentially expressed genes in the Spi1 KD were enriched for PU.1 binding sites in the promoter region (Fisher’s exact test, p value of overlap = 1.6 × 10−16). Given the proportion of overlapping genes was similar across all three test sets (Spi1 KD, Spi1 OE and Spi1 dose-sensitive) it seems Spi1 was not limited to directly regulating genes, but likely indirectly mediates expression of multiple other genes via secondary downstream processes.

Figure 5.

Figure 5

Open in a new tab

Comparison of Spi1 KD, Spi1 OE and 194 Spi1 Dose-sensitive genes to Spi1/PU.1 Chip-Seq dataset. Spi1 binding sites were determined from PU.1 Chip-Seq published data available online30 as described in methods. Fisher’s exact tests were used to determine if there was a significant overlap between the datasets (A) includes the results from all protein coding genes with a Spi1 binding region (defined by Chip-seq in grey) that are also expressed in KD (purple), OE (orange) RNA-seq datasets or the Spi1 dose-sensitive genes (red). (B) The proportion of genes expressed in the RNA-seq datasets that contained a Spi1 binding site within the promoter region. The KD dataset contained a significant number of genes with a Spi1 binding sequence in the promoter (Fisher’s exact test, overlapping p value = 1.6 × 10−16). No significant enrichment was noted in either the Spi1 OE or dose sensitive genes. Figures were made in PowerPoint based on results from ‘GeneOverlap’48 package in RStudio39,40.

Comparison to human AD risk genes

Protein network analysis suggest Spi1/PU.1 is one of several “hub” genes within a network of AD risk genes5, which was further supported by Cis-eQTL analyses in monocytes and macrophages7. The RNA-Sequencing profiles generated in this paper were compared to the International Genomics of Alzheimer’s Project (IGAP) dataset, to investigate if there was an enrichment of AD genetic risk within the differential expressed sets of genes. Supplementary Table 2 denotes the gene sets used in this MAGMA analysis. In the Spi1 over-expression dataset a 21 gene set (Fig. 6), corresponding to a Benjamini–Hochberg adjusted p value of ≤ 1 × 10−6 for differential expression, were enriched for AD genetic risk (adjusted p value = 0.035). The individual IGAP p values can be viewed alongside the corresponding Spi1 over-expression data in Supplementary Table 3. This gene list (Fig. 6B) contained Ifit3, Oas1b and Oas2 which GO analysis linked to the interferon response pathway and Rnf144b and Treml4 which are related to antigen presentation pathways, implicating the immune system response in AD pathology.

Figure 6.

Figure 6

Open in a new tab

Summary of Spi1 over-expression genes that were significantly enriched in the IGAP dataset. (A) Volcano plot of the Spi1 over-expression gene expression changes relative to the empty vector control, highlighting the p ≤ 1 × 10−6 threshold that was shown to be significantly enriched by MAGMA analysis in the IGAP dataset (MAGMA’s empirical multiple testing corrected p value 0.035). The adjusted p value for this gene set was 0.0033 and an enrichment effect size of 0.51. (B) Bubble plot of the genes from the Spi1 over expression dataset that were enriched for AD genetic risk via MAGMA comparison to the human IGAP database, against the Benjamini–Hochberg adjusted p values from the Spi1 over-expression RNA-sequencing data with the fold-change shown in colour. The vertical black line indicates the p value threshold of 0.05. Figure (A) was made in GraphPad PRISM 6 (version 3.07) and (B) using the ‘tidyverse’37 and ‘ggpubr’38 packages in in RStudio39,40.

Discussion

Manipulation of Spi1 expression in primary mixed glia cultures were used to identify how microglial gene expression was altered following modest changes to Spi1/PU.1 (Fig. 1). Analysis of the RNA-sequencing datasets produced from these cultures have provided insight into how Spi1 dose affects the microglia transcriptome. Briefly Spi1 knock-down resulted in changes to gene expression of components of cell cycle checkpoint pathways while over-expression of Spi1 altered enriched pathways related to MHCII, interferon response and viral response (Figs. 3, 4). Comparing these RNA-Seq datasets to ChiP-Seq data suggests that approximately 70% of the differentially expressed genes were likely directly altered by Spi1 binding whereas the rest appeared to be altered via other mechanisms (Fig. 5).

Strengths of this experimental approach include assessing the bidirectional impact of Spi1/PU.1 changes (Fig. 1) on the microglial transcriptome in an unbiased manner in primary cell cultures. We supplemented our cultures with TGF-β which partially compensates for the loss of complex environmental cues microglia would normally receive. However, it does not fully replicate the in vivo context and there were no suitable transgenic mice available to study the impact of Spi1 dose on microglia in situ. The reduction in Spi1 FPKM values and PU.1 protein levels can be clearly seen in both microglia (Fig. 1D and Supplementary Fig. 3B) and in RAW264.7 cells (Supplementary Fig. 2). Figure 1E clearly shows that the Spi1 pSIEW plasmid results in increased Spi1 mRNA and PU.1 expression in microglia.

PU.1 has been previously linked to proliferation in macrophages. Bone-marrow derived Macrophages over-expressing PU.1 increased GM-CSF and M-CSF dependent proliferation and cell number, the opposite was seen in cells transfected with an anti-sense PU.19. However, there are some conflicting reports in the literature as to the role of PU.1 in microglia proliferation. In this model absolute cell numbers did not differ between Spi1 shRNA and control shRNA infected microglia, neither were there significant fold-change differences between normalised cell cycle data (Supplementary Fig. 3). In human microglia cultures an siRNA mediated reduction of PU.1 resulted in a lower cell number, disintegration and rounding of some cells but at the final timepoint viability appeared unaffected. This suggests some microglia could survive with reduced PU.141. Another similar study observed no reduction in microglia number, though PU.1 loss was measured at a culture level so the reduction could not be quantified on a single-cell level31. In summary, we did not observe significant changes in the proportion of cells in each phase of the cell cycle following a reduction in PU.1 (Supplementary Fig. 3), despite the differential expression of cell cycle associated genes within the Spi1 KD RNA-Seq dataset (Fig. 2).

PU.1 has been shown to bind the promoter of the Csf1-receptor (Csf1r) increasing expression of Csf1-receptor49, which is highly linked to microglia survival and proliferation. In BV-2 cells a reduction in PU.1 also results in a reduced Csf1r expression7 though no cell number/viability data was provided. It was recently shown that BV-2 cells with diminished PU.1 expression were more vulnerable to caspase-dependent cell death, while PU.1 over-expressing cells appear to have delayed onset of death. Baseline cell viability did not appear to be impacted by PU.1 modulation in these cells32. In the Spi1 RNA-seq datasets generated in this paper Csf1r expression was not significantly altered (Benjamini–Hochberg adjusted p values 0.665 in KD and 0.999 in OE), though the impact on Csf1r protein expression was not investigated here. Given that Csf1r inhibitors have been shown to prevent AD-related microgliosis and partially ameliorate disease pathology in vivo33–36 it would be interesting to see if Csf1R protein was affected in Spi1 shRNA infected microglia and if the reduction in PU.1 had a similar impact to Csf1r inhibitors on disease progress.

Over-expression of Spi1 resulted in differentially expressed genes enriched for MHCII related pathways, interferon response pathways and response to virus pathways (Fig. 2B). Though there may be minor concerns that the significance of viral immune response pathways could be an unintended effect of using lentiviruses to manipulate Spi1 expression this is unlikely as comparisons were made relative to an almost identical control virus that lacked the Spi1 coding sequence. Additionally, this study identified a subset of 194 genes that appear to be Spi1 dose sensitive, linked to GO terms such as “Cellular response to interferon” and “Defence response to virus” (Fig. 3). Therefore, an increased level of Spi1 appears to cause transcription of genes associated with an inflammatory phenotype that might be relevant to AD genetic risk mechanisms (Fig. 6).

Previous work has identified that Spi1 is able to influence multiple gene networks in microglia5,7,31 and has been proposed as central to a network of AD risk genes that are conserved between species50. Basal cytokine expression was higher in BV-2 cells over-expressing PU.1, and was further potentiated by LPS stimulation compared to control samples32. Media taken from LPS treated PU.1 over-expressing BV-2 cells promoted a reactive astrocyte phenotype which was not observed in astrocytes treated with media from LPS treated BV-2 cells with reduced PU.1 expression32. Together these results suggest that PU.1 dose likely modulates the cytokine response in stimulated BV-2 microglia cells. In co-transfected NIH-3T3 fibroblasts PU.1 was required to bind either IFN Regulator Factor 4 (IRF-4) or IFN Consensus Binding Protein (ICSBP) for maximal induction of an IL-1β reporter assay51. Therefore, it is likely that increased Spi1/PU.1 expression levels in microglia may result in a more inflammatory microglial profile, potentially in conjunction with additional transcription factors. GO analysis of the Spi1 knock-down and over-expression datasets suggest that the level of Spi1 may impact microglial phenotype. Work in BV-2 cells suggests that reduced PU.1 results in a less reactive phenotype whereas PU.1 over expression primes cells for a more reactive response32. Further profiling is needed to ensure the pathway analysis matches the phenotype presented in this model or ideally in vivo.

Several gene-expression studies in AD model mice have highlighted Spi1 as a target of interest. Salih et al. identified a network of microglial genes expressed by amyloid-associated microglia, collated from five transgenic mouse lines, including Spi1, and 74 other genes that were identified as Spi1 dose-sensitive in this study52. These genes included Oas1b, Oas2, Ifit3 which were all highly differentially expressed in the Spi1 over-expression dataset (Fig. 6) and are linked to interferon signalling53. Sierskma et al. identified 18 genes that both overlapped with AD GWAS datasets and were maintained between young and old APP transgenic mice, which were predominantly expressed by microglia and appear to be regulated by Spi150.

There are several reports in the literature that imply human and mouse microglia have phenotypic differences. For example, work in primary human microglia suggests that the responses to IFN-γ, TGF-β1 and M-CSF may be species specific54. As this work was carried out in primary mouse cultures there could be concerns that the impact of reduced or increased Spi1/PU.1 might not be present in human biology.

However, studies investigating PU.1 in human microglia suggest there are shared Spi1/PU.1 dose dependent pathways in microglia. In addition to SPI1 binding sites being located near other AD risk genes or regulatory elements5,7,8, SPI1 risk alleles associated with a higher gene expression have been shown to lower the age of onset in AD7. Transcriptional profiles of microglia isolated from frozen human AD and control cortical tissue suggest an overall species difference in the disease associated signature, though an increased SPI1 expression was measured in AD samples compared to controls55 validating previous observations by others31.

In primary human microglia cultures siRNA mediated reductions in PU.1 resulted in a decrease in Amyloid-β 1-42 peptide phagocytosis41, in a similar manner to observations in BV-2 cells7,32. Analysis of mixed microglia/pericyte cultures where PU.1 was reduced via siRNA showed reduced expression of DAP12 and HLA-DR/DP/DQ31. Moreover, similar gene expression changes to immune response and lipid metabolism genes were observed in BV-2 cells and human iPSC microglia following PU.1 siRNA knock-down32. Together these studies show that while the impact of SPI1 dose on microglia phenotype needs to be assessed in both species, some functional changes appear common to both human and mouse microglia.

In summary, disease relevant reductions in Spi1 highlighted dysregulation of genes involved in cell cycle checkpoint pathways. Modest increases in microglial Spi1 expression results in dysregulation of genes linked to immune response and interferon signalling, suggestive of a more pro-inflammatory microglial phenotype. This study highlights how relatively modest changes to Spi1/PU.1 expression alone can have a large impact on the microglial transcriptome of primary mixed-glial cultures, providing candidate pathways for future studies investigating Spi1 dependent processes and AD relevant biology.

Methods

Primary mixed glia cultures

Brains from 8-week-old C57BL/6 J mice (Charles River) were collected and transported in Hank’s Balanced Salt Solution (HBSS without Mg2+ or Ca2+; Gibco) on ice. These culture experiments were repeated 10 times (n = 10), 4 were used to generate RNA-Sequencing replicates and 6 for subsequent validation experiments. For most cultures three brains were sufficient per experiment (one 6 well-plate), though for some experiments more cells were required so cultures were scaled up accordingly. Brains were digested using the Neural Tissue Digest Kit-P in C-tubes as per manufactures directions (Miltenyi Biotec). Two brains were digested in each C-tube using program 37_ABDK on the GentleMACS OctoDissociator (Miltenyi Biotec) to produce a single cell suspension. Cell suspensions were passed through a 70 μM strainer (Falcon) into a 50 mL tube and centrifuged at 300 × g for 7 min. The supernatant was aspirated and the pellet resuspended in 10 mL Dulbecco’s Minimum Essential Media (DMEM containing 4.5 g/L D-Glucose, GlutaMAX and supplemented with 15% (v/v) heat-inactivated Foetal Bovine Serum (FBS) and 100 units/mL (v/v) Penicillin and 100 μg/mL Streptomycin (v/v); all Gibco) before centrifugation at 300 × g for 7 min. The supernatant was aspirated, and the pellet resuspended in 2 mL per brain of DMEM media containing 10 ng/mL recombinant murine M-CSF (Peprotech). Cell suspensions were then pooled as required and placed in a 6-well plate (2 mL per well) or 100 mm × 20 mm (diameter × height) tissue culture plates, 12 mL per plate, and moved to a humid incubator with 95% air and 5% CO2 overnight to allow the microglia to adhere.

Myelin was resuspended in the media (with 24 h of digest) by gentle plate agitation and the media containing the myelin debris was discarded. The cell monolayer was carefully washed with DMEM media. The wash media was then replaced with fresh DMEM media with 10 ng/mL M-CSF and this was again replaced two days later. On day 5 of culture the media was replaced with DMEM media containing 10 ng/mL M-CSF and 50 ng/mL recombinant mouse TGF-β1 (eBioscience or Biolegend). TGF-β supplementation was used to promote a more homeostatic phenotype in vitro56,57. Recombinant murine TGF-β1 was purchased either from eBioscience or Biolegend, both of which were produced in a similar way. The Biolegend recombinant TGF-β1 was diluted to a 100 μg/mL stock in an equal volume of sterile filtered 2% BSA, 0.2 M glycine in DPBS to match the eBioscience product.

From this point forward all culture media contained M-CSF and TGF-β1 was replaced every two days. On day 10, these cultures were infected with Spi1 shRNA, control shRNA, Spi1 pSIEW or pSIEW lentivirus particles as appropriate. To achieve this the culture media was replaced with 3 mL fresh media and 100–300 μL of lentivirus was added to each plate. The volumes of Spi1 targeting and control viruses were kept the same within each experiment. After 6 h each plate received an additional 3 mL of culture media. The media on these cultures continued to be changed every 2 days, supplemented with M-CSF and TGF-β1 as before.

All animal experiments were conducted in accordance with UK Home Office Guidelines and Animal [Scientific Procedures] Act 1986 which included full review and approval by the local ethical review board (Animal Welfare and Ethical Review Body, AWERB, of the Biological Standards Committee) and the granting of a UK Home Office Project Licence. The study was conducted in compliance with the ARRIVE guidelines.

Sample processing for RNA-sequencing

After 21 days the media was removed, set-aside, and the mixed glial cells harvested by incubating with ~ 10 mL Accumax (Sigma) at 37 °C for 10–20 min. The monolayer was gently washed with DMEM media and any remaining attached cells were carefully removed using a plastic scraper (Greiner). The cell suspension was added back to the media and centrifuged at 300 × g for 7 min. The supernatant was removed via pipetting and the cell pellet was resuspended in 0.5% BSA in DPBS with a 1:1000 dilution of LIVE/DEAD near-IR staining solution for 30 min on ice per manufacturers direction (Molecular Probes).

Samples were centrifuged at 300 × g for 7 min and the supernatant aspirated via pipette. Each sample was resuspended in 500 µL block solution (4 μg/mL Rat anti-mouse FcγRII/III (clone 2.4G2) and 5% (v/v) filtered rabbit serum in 0.5% BSA (w/v), 5 mM EDTA in DPBS) and kept on ice for 10 min. CD11b (Clone M1/70 PerCP-Cy5.5 conjugate, BD Biosciences final dilution 2 μg/mL) and CD45 (Clone 30-F11 eFlour 450 conjugate, eBioscience final dilution 2 μg/mL) antibodies were diluted to a 4 μg/mL concentration in 0.5% BSA (w/v), 5 mM EDTA in DPBS. 500 μL of the antibody staining solution was added to each sample, to give a final antibody concentration of 2 μg/mL, and the samples were incubated on ice for a further 15 min. The tubes were then centrifuged at 300 × g for 7 min and the supernatant removed.

The cells were resuspended in 1 mL of 0.5% BSA (w/v), 5 mM EDTA in DPBS and kept cool before undergoing fluorescent activated cell sorting on a FACS Aria III (BD Biosciences). Dead cells were excluded from the sort and microglia were selected using CD11b/CD45 double staining and GFP as a marker of infection. The sorted cells were then pelleted via centrifugation, the supernatant aspirated and the pellet lysed for RNA using the Mini RNeasy kit (Qiagen) per manufacturer’s directions.

Lentivirus preparation

The shRNA vector construct and lentivirus production method has been described in58 and sucrose-gradient purification in59. The insert sequences for the control shRNA and Spi1 shRNA can be seen in Table 1.

Table 1.

shRNA sequences including termination sequence (bold).

Spi1 shRNA sequence (5′ → 3′)
GATGTGCTTCCCTTATCAAACCTCGAGGTTTGATAAGGGAAGCACATCTTTTT
Control shRNA sequence (5′ → 3′)
GTCTCGCTTGGGCGAGAGTAAGTAGTGAAGCCACAGATGTACTTACTCTCGCCCAAGCGAGACTTTTT
Open in a new tab

The over-expression plasmid used a spleen focus forming virus (SFFV) promoter to drive expression of murine Spi1 (Ensemble reference CCDS 16,425.1). An Internal Ribosome Entry Site (IRES) was used to initiate translation of a downstream eGFP reporter. Lentiviruses were purified by overlaying the virus containing media over a 20% sucrose solution before ultracentrifugation at 26,000 rpm for 90 min at 4 °C in a SW28Ti swinging-bucket ultracentrifuge rotor assembly in an Optima XPN-80 Ultracentrifuge (Beckman Optima Ultracentrifuge) and the viral pellet was resuspended in AimV media.

RNA-sequencing data analysis

RNA was isolated from cell sorted GFP + microglia using the RNeasy Mini kit (Qiagen) per manufacturer’s instructions and eluted in nuclease free water. RNA integrity and concentration were assessed using the Agilent 2100 Bioanalyzer (Aligent). Complementary (cDNA) libraries were generated using the Truseq stranded total RNA with Ribo-Zero GOLD kit (Illumina). Paired end sequencing was then performed using the Illumina HiSeq 2500 sequencing platform to a read depth of between 30 and 40 million pairs.

The FASTQ files were processed with Trimmomatic60 to remove paired-end reads and quality was confirmed in FastQC using default parameters61. Following this sequence reads were mapped to the mm10 genome (GRCm38) using the STAR pipeline62 and featureCounts was used to assign counts to transcripts63 with the GRCm38.84 Ensembl gene build GTF. The Ensembl FTP site was used to download the reference genome and GTF64. Differential gene expression was performed using the DESeq2 package65. Genes that were not significant after the differential expression analysis were discarded; significance was defined as an adjusted p value (Padj) of < 0.05 (Benjamini–Hochberg correction for multiple testing).

Assessing PU.1 promoter binding sites

A microglial PU.1 ChIP-sequencing (ChIP-seq) dataset (GSM1533906) isolated from C57BL/6 mice of a similar age (8–9 weeks)30 was accessed via the Cistrome database66. This dataset was then run through HOMER28 to generate an annotated peak file with genomic features. The promoter region was defined as 1000 base pairs upstream or 100 base pairs downstream of the transcription start site.

Duplicates were removed from the annotated ChIP-seq dataset and compared to gene discoveries from knock-down, over-expression and dose-sensitive Spi1 RNA-sequencing datasets. The background genome size was the number of genes detected in each RNA-Sequencing dataset defined as either all control and/or all experimental samples had a raw read count of ≥ 5. In the Spi1 knock-down dataset 17,962 genes were detected and 16,998 genes in the Spi1 over-expression data. For the 194 Spi1 dose-sensitive genes a merged list of genes expressed in either the Spi1 knock-down or over-expression RNA-sequencing datasets was generated, and duplicates removed, to give a background gene number of 16,753.

Finally, the ChIP-seq datasets were compared against their respective RNA-Seq dataset (Padj ≤ 0.05) in R using the ‘tidyverse’ and ‘GeneOverlap’ packages. Here, a Fisher’s exact test was utilised to confirm if there was any significant enrichment of PU.1 ChIP-seq binding sites in genes expressed in Spi1 knock-down, over-expression or dose-sensitive datasets.

Gene ontology analysis

Enriched gene Ontology (GO) terms were identified using the Database for Annotation, Visualization and Integrated Discovery (DAVID, version 6.8) and compared against a background of genes expressed in at least one Spi1 RNA-seq dataset (16,998 genes). Gene expression in the Spi1 RNA-seq datasets was defined as having a raw read count ≥ 5 in all control and/or all experimental samples. The most significantly altered terms from the GOTERM_BP_DIRECT list, henceforth called Biological Process GO Terms, were downloaded for subsequent analysis. Bubble plots were generated in R using the ‘ggplot2’ package, and an adjusted p value cut-off of 0.05 as indicated on each plot. Data utilised in these plots included the Benjamini–Hochberg adjusted p value, the percentage of the genes from the RNA-seq datasets that aligned to this pathway and the fold enrichment, which was defined as the proportion of genes present in this list compared to the Spi1 RNA-Seq gene expression background.

Hierarchical clustering analysis

Hierarchical clustering was performed utilising ‘gplots’, ‘dendextend’ and ‘colorspace’ packages in R with code adapted from46,47, to produce a heatmap and dendrogram to better visualise the clusters. Complete markdown can be found in Supplementary Methods. FPKM values were log10 transformed, and z-scored (Pearson Correlation method) prior to clustering. Hierarchical clustering resulted in six distinct clusters which are highlighted in different colours in the dendrogram (Fig. 4).

Enrichment of association signal in IGAP GWAS data

As Spi1 has been implicated in the regulation of other AD risk gene expression, both RNA-Seq datasets were tested for enrichment of association with AD risk in the International Genomics of Alzheimer’s Project (IGAP) GWAS dataset67. Firstly, the BioMart feature in Ensembl was used to convert mouse genes differentially expressed in the Spi1 knock-down and over-expression datasets into human orthologs68.

Gene sets were then determined for each Spi1 RNA-Seq dataset using p value cut-offs between 0.05 and 1 × 10−10. Direction of differential expression was not considered when defining gene sets. These gene sets (Supplementary Table 2) were tested for enrichment in the IGAP dataset using Multi-marker Analysis of GenoMic Annotation (MAGMA) analysis69.

Flow cytometric analysis of PU.1

Mixed glial cultures were harvested on day 21. Culture media was carefully removed and retained. Cells were detached by incubating with Accumax for 10–20 min at 37 °C, added to the appropriate culture media and centrifuged at 300 × g for 7 min. Approximately 2 × 105 cells per sample were fixed in 4% formalin solution (in DPBS) on ice for 30 min. Formalin was removed by centrifugation (300 × g for 7 min) and cells were permeabilised on ice in 90% ice cold methanol (v/v in PBS). Samples were centrifuged at 300 × g for 7 min and the supernatant discarded. To ensure the methanol was completely removed an additional wash step using 500 μL of wash solution (0.5% (w/v) BSA, 5 mM EDTA and 2 mM NaN3 in DPBS) and centrifugation at 300 × g for 7 min. The supernatant was discarded, and the cell pellets were then re-suspended in 50 µL of block solution (5% (v/v) filtered Rabbit Serum, 4 μg/mL Rat Anti-mouse FcγRII/III 2.4G2 clone in wash solution) and incubated on ice for 20–30 min. Following the blocking step 50 μL of CD11b, CD45 and PU.1 antibodies (listed below) were added and cells were incubated for 30–60 min on ice in the dark. The following antibodies were used in Flow cytometric experiments; Anti-CD11b FITC (56C, 8 μg/mL, produced in house), Anti-CD11b BV421 (M1/70, 2 μg/mL, Biolegend), Anti-CD11b PerCP-Cy5.5 (M1/70, 2 μg/mL, BD Biosciences), Anti-CD45 eFlour450 or PE-Cyanine7 conjugates (30-F11, 2 μg/mL, eBioscience), Anti-PU.1 AF647 (7C2C34, 5 μg/mL, Biolegend) and Rat IgG2a,k AF647 Isotype Control (RTK2758, 5 μg/mL, Biolegend).

Cells were washed with wash solution 3 times and centrifuged at 300 × g for 7 min. After the final wash the cell pellets were re-suspended in 500 μL of wash solution and acquired on Attune NxT cytometer (Thermofisher). Data was analysed using FlowJo software (version 10; FlowJo LLC). Single colour and isotype controls were used as appropriate. Post data collection relative PU.1 protein levels were determined using the Median Fluorescent Intensity (MFI) values in the PU.1 antibody channel. First the isotype background signal from the same channel was subtracted. These values were then normalised between experiments by diving each value by the average MFI value (minus isotype) of the appropriate control sample.

Statistics and figures

Statistical analyses were performed using GraphPad PRISM 6 (version 3.07) and GraphPad PRISM 8 (version 8.4.3; both GraphPad Software, Inc.), unless otherwise stated. All statistical tests will be described as appropriate. p values of > 0.05 were taken as non-significant (ns). p values of ≤ 0.05 will be denoted with a single asterisk*, p values of ≤ 0.01 will be written as **, p values of ≤ 0.001 by *** and p value of ≤ 0.0001 as ****. Figures were made using both GraphPad PRISM 6, GraphPad PRISM 8 (version 3.07 & 8.4.3; GraphPad Software, Inc.) and R Studio (Version 1.2.5042 copyright 2009–2020 RStudio, Inc.39) with base R version 4.0.0 (2020–04-24, copyright 2020 The R Foundation for Statistical Computing40). The following R packages were used ‘tidyverse’37, ‘ggpubr’38, ‘gplots’42, ‘plotly’43, ‘dendextend’44, ‘colourspace’45 and ‘GeneOverlap’48.

Supplementary Information

Supplementary Information. (3.2MB, docx)

Acknowledgements

We would like to thank our animal facility staff for their care for the animals used in this study. We also thank the Cardiff University data clinic volunteers for their advice, and especially to thank Dr John Watkins for his guidance. We acknowledge Wales Gene Park and the Advanced Research Computing Team at Cardiff University, for performing Illumina sequencing and provision of high-performance computing infrastructure. This Project was funded by the MRC-Doctoral Training Programme Studentship. PRT is funded by a UK Dementia Research Institute Professorship and a Wellcome Trust Investigator Award (107964/Z/15/Z).

Author contributions

R.J. conducted the work; R.A., P.H. and M.H. assisted with data analysis and interpretation; R.J. and P.R.T. conceived and designed the study and wrote the manuscript; All authors commented on and approved the manuscript.

Competing interests

M.H. is now employed by Vertex Pharmaceuticals. The other authors declare that there are no other competing interests.

Footnotes

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-021-94324-z.

References

  • 1.Prince, M. et al.Dementia UK: Second Edition - Overview 61 (Alzheimer's Society, 2014).
  • 2.Jones L, Holmans PA, Hamshere ML, Harold D, Moskvina V, Ivanov D, et al. Genetic evidence implicates the immune system and cholesterol metabolism in the aetiology of Alzheimer’s disease. PLoS ONE. 2010;5(11):13950. doi: 10.1371/journal.pone.0013950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Escott-Price V, Bellenguez C, Wang LS, Choi SH, Harold D, Jones L, et al. Gene-wide analysis detects two new susceptibility genes for Alzheimer’s disease. PLoS ONE. 2014;9(6):94661. doi: 10.1371/journal.pone.0094661. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Kunkle BW, Grenier-Boley B, Sims R, Bis JC, Damotte V, Naj AC, et al. Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019;51:414–430. doi: 10.1038/s41588-019-0358-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Sims R, van der Lee SJ, Naj AC, Bellenguez C, Badarinarayan N, Jakobsdottir J, et al. Rare coding variants in PLCG2, ABI3, and TREM2 implicate microglial-mediated innate immunity in Alzheimer’s disease. Nat. Genet. 2017;49(9):1373–1384. doi: 10.1038/ng.3916. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Gjoneska E, Pfenning AR, Mathys H, Quon G, Kundaje A, Tsai LH, et al. Conserved epigenomic signals in mice and humans reveal immune basis of Alzheimer’s disease. Nature. 2015;518:365–369. doi: 10.1038/nature14252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Huang K, Marcora E, Pimenova AA, di Narzo AF, Kapoor M, Jin SC, et al. A common haplotype lowers SPI1 (PU.1) expression in myeloid cells and delays age at onset for Alzheimer’s disease. Nat. Neurosci. 2017;20:1052–1061. doi: 10.1038/nn.4587. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Tansey KE, Cameron D, Hill MJ. Genetic risk for Alzheimer’s disease is concentrated in specific macrophage and microglial transcriptional networks. Genome Med. 2018;10(14):1–10. doi: 10.1186/s13073-018-0523-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Celada A, Borras FE, Soler C, Lloberas J, Klemsz M, van Beveren C, et al. The transcription Factor PU.1 is involved in macrophage proliferation. J. Exp. Med. 1996;184:61–69. doi: 10.1084/jem.184.1.61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Anderson KL, Smith KA, Conners K, McKercher SR, Maki RA, Torbett BE. Myeloid development is selectively disrupted in PU.1 null mice. Blood. 1998;91(10):3702–3710. doi: 10.1182/blood.V91.10.3702. [DOI] [PubMed] [Google Scholar]
  • 11.Beers DR, Henkel JS, Xiao Q, Zhao W, Wang J, Yen AA, et al. Wild-type microglia extend survival in PU.1 knockout mice with familial amyotrophic lateral sclerosis. PNAS. 2006;103(43):16021–16026. doi: 10.1073/pnas.0607423103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Scott EW, Simon MC, Anastasi J, Singh H. Requirement of transcription factor PU.1 in the development of multiple hematopoietic lineages. Science. 1994;265:1573–1577. doi: 10.1126/science.8079170. [DOI] [PubMed] [Google Scholar]
  • 13.McKercher SR, Torbett BE, Anderson KL, Henkel GW, Vestal DJ, Baribault H, et al. Targeted disruption of the PU.1 gene results in multiple hematopoietic abnormalities. EMBO J. 1996;15(20):5647–5658. doi: 10.1002/j.1460-2075.1996.tb00949.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.DeKoter RP, Singh H. Regulation of B lymphocyte and macrophage development by graded expression of PU.1. Science. 2000;288:1439–1441. doi: 10.1126/science.288.5470.1439. [DOI] [PubMed] [Google Scholar]
  • 15.Back J, Dierich A, Bronn C, Kastner P, Chan S. PU.1 determines the self-renewal capacity of erythroid progenitor cells. Blood. 2004;103(10):3615–3623. doi: 10.1182/blood-2003-11-4089. [DOI] [PubMed] [Google Scholar]
  • 16.Back J, Allman D, Chan S, Kastner P. Visualizing PU.1 activity during hematopoiesis. Exp. Hematol. 2005;33:395–402. doi: 10.1016/j.exphem.2004.12.010. [DOI] [PubMed] [Google Scholar]
  • 17.Hu Z, Gu X, Baraoidan K, Ibanez V, Sharma A, Kadkol SH, et al. RUNX1 regulates corepressor interactions of PU.1. Blood. 2011;117(24):6498–6508. doi: 10.1182/blood-2010-10-312512. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Jin H, Li L, Xu J, Zhen F, Zhu L, Liu PP, et al. Runx1 regulates embryonic myeloid fate choice in zebrafish through a negative feedback loop inhibiting Pu.1 expression. Blood. 2012;119(22):5239–5249. doi: 10.1182/blood-2011-12-398362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Zarnegar MA, Chen J, Rothenberg EV. Cell-type-specific activation and repression of PU.1 by a complex of discrete, functionally specialized cis-regulatory elements. Mol. Cell. Biol. 2010;30(20):4922–4939. doi: 10.1128/MCB.00354-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Li Y, Okuno Y, Zhang P, Radomska HS, Chen H, Iwasaki H, et al. Regulation of the PU. 1 gene by distal elements. Blood. 2001;98(10):2958–2965. doi: 10.1182/blood.V98.10.2958. [DOI] [PubMed] [Google Scholar]
  • 21.Okuno Y, Huang G, Rosenbauer F, Evans EK, Radomska HS, Iwasaki H, et al. Potential autoregulation of transcription factor PU.1 by an upstream regulatory element. Mol. Cell. Biol. 2005;25(7):2832–2845. doi: 10.1128/MCB.25.7.2832-2845.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Zarnegar MA, Rothenberg EV. Ikaros represses and activates PU.1 cell-type-specifically through the multifunctional Sfpi1 URE and a myeloid specific enhancer. Oncogene. 2012;25(31):4647–4554. doi: 10.1038/onc.2011.597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Leddin M, Perrod C, Hoogenkamp M, Ghani S, Assi S, Heinz S, et al. Two distinct auto-regulatory loops operate at the PU.1 locus in B cells and myeloid cells. Blood. 2011;117(10):2827–2838. doi: 10.1182/blood-2010-08-302976. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Kamath MB, Houston IB, Janovski AJ, Zhu X, Gowrisankar S, Jegga AG, et al. Dose-dependent repression of T-cell and natural killer cell genes by PU.1 enforces myeloid and B-cell identity. Leukemia. 2008;22:1214–1225. doi: 10.1038/leu.2008.67. [DOI] [PubMed] [Google Scholar]
  • 25.Lloberas J, Soler C, Celada A. The key role of PU.1/SPI-1 in B cells, myeloid cells and macrophages. Immunol. Today. 1999;20(4):184–189. doi: 10.1016/S0167-5699(99)01442-5. [DOI] [PubMed] [Google Scholar]
  • 26.Pahl HL, Scheibe RJ, Zhang DE, Chen HM, Galson DL, Maki RA, et al. The proto-oncogene PU.1 regulates expression of the myeloid-specific CD11b promoter. J. Biol. Chem. 1993;268(7):5014–5020. doi: 10.1016/S0021-9258(18)53496-7. [DOI] [PubMed] [Google Scholar]
  • 27.Anderson KL, Nelson SL, Perkin HB, Smith KA, Klemsz MJ, Torbett BE. PU.1 is a lineage-specific regulator of tyrosine phosphatase CD45. J. Biol. Chem. 2001;276(10):7637–7642. doi: 10.1074/jbc.M009133200. [DOI] [PubMed] [Google Scholar]
  • 28.Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell. 2010;38:576–589. doi: 10.1016/j.molcel.2010.05.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Ghisletti S, Barozzi I, Mietton F, Polletti S, De Santa F, Venturini E, et al. Identification and characterization of enhancers controlling the inflammatory gene expression program in macrophages. Immunity. 2010;32:317–328. doi: 10.1016/j.immuni.2010.02.008. [DOI] [PubMed] [Google Scholar]
  • 30.Gosselin D, Link VM, Romanoski CE, Fonseca GJ, Eichenfield DZ, Spann NJ, et al. Environment drives selection and function of enhancers controlling tissue-specific macrophage identities. Cell. 2014;159:1327–1340. doi: 10.1016/j.cell.2014.11.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Rustenhoven J, Smith AM, Smyth LC, Jansson D, Scotter EL, Swanson MEV, et al. PU.1 regulates Alzheimer’s disease-associated genes in primary human microglia. Mol. Neurodegener. 2018;13:44. doi: 10.1186/s13024-018-0277-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Pimenova AA, Herbinet M, Gupta I, Machlovi SI, Bowles KR, Marcora E, et al. Alzheimer’s-associated PU.1 expression levels regulate microglial inflammatory response. Neurobiol. Dis. 2021;148:105217. doi: 10.1016/j.nbd.2020.105217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Olmos-Alonso A, Schetters STTT, Sri S, Askew K, Mancuso R, Vargas-Caballero M, et al. Pharmacological targeting of CSF1R inhibits microglial proliferation and prevents the progression of Alzheimer’s-like pathology. Brain. 2016;139:891–907. doi: 10.1093/brain/awv379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Dagher NN, Najafi AR, Kayala KMN, Elmore MRP, White TE, Medeiros R, et al. Colony-stimulating factor 1 receptor inhibition prevents microglial plaque association and improves cognition in 3xTg-AD mice. J. Neuroinflam. 2015;12(139):1–14. doi: 10.1186/s12974-015-0366-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Sosna J, Philipp S, Albay RI, Reyes-Ruiz JM, Baglietto-Vargas D, LaFerla FM, et al. Early long-term administration of the CSF1R inhibitor PLX3397 ablates microglia and reduces accumulation of intraneuronal amyloid, neuritic plaque deposition and pre-fibrillar oligomers in 5XFAD mouse model of Alzheimer’s disease. Mol. Neurodegener. 2018;13:1–11. doi: 10.1186/s13024-018-0244-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Spangenberg EE, Lee RJ, Najafi AR, Rice RA, Elmore MRP, Blurton-Jones M, et al. Eliminating microglia in Alzheimer’s mice prevents neuronal loss without modulating amyloid-β pathology. Brain. 2016;139:1265–1281. doi: 10.1093/brain/aww016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, et al. Welcome to the Tidyverse. J. Open Source Softw. 2019;4(43):1686. doi: 10.21105/joss.01686. [DOI] [Google Scholar]
  • 38.Kassambara, A. ggpubr: ‘ggplot2’ Based Publication Ready Plots. https://rpkgs.datanovia.com/ggpubr/index.html
  • 39.RStudio Team. RStudio: Integrated Development for R. RStudio (2020).
  • 40.R Core Team. R: A Language and Environment for Statistical Computing [Internet]. R Foundation for Statistical Computing; 2020. Available from: https://www.R-project.org/.
  • 41.Smith AM, Gibbons HM, Oldfield RL, Bergin PM, Mee EW, Faull RLM, et al. The transcription factor PU.1 is critical for viability and function of human brain microglia. Glia. 2013;61:929–942. doi: 10.1002/glia.22486. [DOI] [PubMed] [Google Scholar]
  • 42.Warnes, G. R., Bolker, B., Bonebakker, L., Gentleman, R., Huber, W., Liaw, A. et al. gplots: Various R Programming Tools for Plotting Data [Internet]. 2020. Available from: https://cran.r-project.org/web/packages/gplots/index.html.
  • 43.Plotly Technologies Inc. Collaborative Data Science. Plotly Technologies Inc. (2015).
  • 44.Galili T. dendextend: An R package for visualizing, adjusting and comparing trees of hierarchical clustering. Bioinformatics. 2015;31(22):3718–3720. doi: 10.1093/bioinformatics/btv428. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Zeileis, A., Fisher, J. C., Hornik, K., Ihaka, R., McWhite, C. D., Murrell, P. et al. colorspace: A toolbox for manipulating and assessing colors and palettes. J. Stat. Softw.96, 5–6 (2020).
  • 46.Liu, Y. How to Draw Heatmap with Colorful Dendrogram [Internet]. Yang’s Research Blog. 2018 [cited 2020 Apr 23]. p. 1. Available from: https://liuyanguu.github.io/post/2018/07/16/how-to-draw-heatmap-with-colorful-dendrogram/.
  • 47.Brandon, Y. How to plot a Heatmap in Rstudio, the easy way [Internet]. 2019 [cited 2020 Apr 23]. Available from: https://www.youtube.com/watch?v=OWWHfXgRw3k.
  • 48.Shen, L. Sinai ISoMaM. GeneOverlap: Test and visualize gene overlaps. R package version 1.28.0. 10.18129/B9.bioc.GeneOverlap (2021).
  • 49.Zhang DE, Hetherington CJ, Chen HM, Tenen DG. The macrophage transcription factor PU.1 directs tissue-specific expression of the macrophage colony-stimulating factor receptor. Mol. Cell. Biol. 1994;14(1):373–381. doi: 10.1128/mcb.14.1.373. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Sierksma, A. et al. Novel Alzheimer risk genes determine the microglia response to amyloid-β but not to TAU pathology. EMBO Mol. Med.12(3), e10606 (2020). [DOI] [PMC free article] [PubMed]
  • 51.Marecki S, Riendeau CJ, Liang MD, Fenton MJ. PU.1 and multiple IFN regulatory factor proteins synergize to mediate transcriptional activation of the human IL-1β gene. J. Immunol. 2001;166:6892–6838. doi: 10.4049/jimmunol.166.11.6829. [DOI] [PubMed] [Google Scholar]
  • 52.Salih DA, Bayram S, Guelfi S, Reynolds RH, Shoai M, Ryten M, et al. Genetic variability in response to amyloid beta deposition influences Alzheimer’s disease risk. Brain Commun. 2019;1(1):1–13. doi: 10.1093/braincomms/fcz022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Schoggins JW. Interferon-stimulated genes: What do they all do? Annu. Rev. Virol. 2019;6(1):567–584. doi: 10.1146/annurev-virology-092818-015756. [DOI] [PubMed] [Google Scholar]
  • 54.Smith AM, Graham ES, Feng SX, Oldfield RL, Bergin PM, Mee EW, et al. Adult human glia, pericytes and meningeal fibroblasts respond similarly to IFNy but not to TGFβ1or M-CSF. PLoS ONE. 2013;8(12):e80463. doi: 10.1371/journal.pone.0080463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Srinivasan K, Friedman BA, Etxeberria A, Huntley MA, van der Brug MP, Foreman O, et al. Alzheimer’s patient microglia exhibit enhanced aging and unique transcriptional activation. Cell Rep. 2020;31(13):107843. doi: 10.1016/j.celrep.2020.107843. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Butovsky O, Jedrychowski MP, Moore CS, Cialic R, Lanser AJ, Gabriely G, et al. Identification of a unique TGF-β-dependent molecular and functional signature in microglia. Nat. Neurosci. 2014;17(1):131–143. doi: 10.1038/nn.3599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Bohlen CJ, Bennett FC, Tucker AF, Collins HY, Mulinyawe SB, Barres BA. Diverse requirements for microglial survival, specification, and function revealed by defined-medium cultures. Neuron. 2017;94:759–773. doi: 10.1016/j.neuron.2017.04.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Rosas M, Davies LC, Giles PJ, Liao C-T, Kharfan B, Stone TC, et al. The transcription factor Gata6 links tissue macrophage phenotype and proliferative renewal. Science. 2014;344:645–648. doi: 10.1126/science.1251414. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Ipseiz N, Czubala MA, Bart VMT, Davies LC, Jenkins RH, Brennan P, et al. Effective in vivo gene modification in mouse tissue-resident peritoneal macrophages by intraperitoneal delivery of lentiviral vectors. Mol. Ther. Methods Clin. Dev. 2020;16:21–31. doi: 10.1016/j.omtm.2019.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Bolger AM, Lohse M, Usadel B. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114–2120. doi: 10.1093/bioinformatics/btu170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Simon, A. FastQC A quality control tool for high throughput sequence data [Internet]. Babraham Bioinformatics. Available from: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/.
  • 62.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. doi: 10.1093/bioinformatics/bts635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Liao Y, Smyth GK, Shi W. featureCounts: An efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30(7):923–930. doi: 10.1093/bioinformatics/btt656. [DOI] [PubMed] [Google Scholar]
  • 64.Ensembl FTP Download [Internet]. Available from: ftp://ftp.ensembl.org/pub.
  • 65.Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. doi: 10.1186/s13059-014-0550-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Zheng R, Wan C, Mei S, Qin Q, Wu Q, Sun H, et al. Cistrome data browser: Expanded datasets and new tools for gene regulatory analysis. Nucleic Acids Res. 2019;47(D1):D729–D735. doi: 10.1093/nar/gky1094. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Lambert JC, Ibrahim-Verbaas CA, Harold D, Naj AC, Sims R, Bellenguez C, et al. Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat. Genet. 2013;45(12):1452–1458. doi: 10.1038/ng.2802. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Herrero, J. How to get all the orthologous genes between two species [Internet]. Ensembl Blog. 2009 [cited 2018 Nov 20]. Available from: http://www.ensembl.info/2009/01/21/how-to-get-all-the-orthologous-genes-between-two-species/.
  • 69.De Leeuw CA, Mooij JM, Heskes T, Posthuma D. MAGMA: Generalized gene-set analysis of GWAS data. PLoS Comput. Biol. 2015;11(4):1–19. doi: 10.1371/journal.pcbi.1004219. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information. (3.2MB, docx)

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES