CSF1R inhibition with PLX5622 affects multiple immune cell compartments and induces tissue-specific metabolic effects in lean mice

Aims/hypothesis Colony stimulating factor 1 (CSF1) promotes the proliferation, differentiation and survival of macrophages, which have been implicated in both beneficial and detrimental effects on glucose metabolism. However, the physiological role of CSF1 signalling in glucose homeostasis and the potential therapeutic implications of modulating this pathway are not known. We aimed to study the composition of tissue macrophages (and other immune cells) following CSF1 receptor (CSF1R) inhibition and elucidate the metabolic consequences of CSF1R inhibition. Methods We assessed immune cell populations in various organs by flow cytometry, and tissue-specific metabolic effects by hyperinsulinaemic–euglycaemic clamps and insulin secretion assays in mice fed a chow diet containing PLX5622 (a CSF1R inhibitor) or a control diet. Results CSF1R inhibition depleted macrophages in multiple tissues while simultaneously increasing eosinophils and group 2 innate lymphoid cells. These immunological changes were consistent across different organs and were sex independent and reversible after cessation of the PLX5622. CSF1R inhibition improved hepatic insulin sensitivity but concomitantly impaired insulin secretion. In healthy islets, we found a high frequency of IL-1β+ islet macrophages. Their depletion by CSF1R inhibition led to downregulation of macrophage-related pathways and mediators of cytokine activity, including Nlrp3, suggesting IL-1β as a candidate insulin secretagogue. Partial restoration of physiological insulin secretion was achieved by injecting recombinant IL-1β prior to glucose stimulation in mice lacking macrophages. Conclusions/interpretation Macrophages and macrophage-derived factors, such as IL-1β, play an important role in physiological insulin secretion. A better understanding of the tissue-specific effects of CSF1R inhibition on immune cells and glucose homeostasis is crucial for the development of targeted immune-modulatory treatments in metabolic disease. Data availability The RNA-Seq dataset is available in the Gene Expression Omnibus (GEO) under the accession number GSE189434 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE189434). Graphical Abstract Supplementary Information The online version of this article (10.1007/s00125-023-06007-1) contains peer-reviewed but unedited supplementary material.


Hyperinsulinemic-euglycemic clamp studies
Mice were anesthetized with isoflurane and eye ointment was applied to both eyes (Vitamin A, Bausch & Lomb Swiss AG, Switzerland).A catheter (MRE 025, Braintree Scientific, Braintree, MA, USA) was inserted into the right jugular vein and exteriorized at the neck.Five to seven days after the surgery, clamp was performed in freely moving mice after 5 h fasting.
Glucose disposal rate was calculated by dividing the rate of 3-[ 3 H] glucose infusion by the plasma 3-[ 3 H] glucose specific activity.Endogenous glucose production during the clamp was calculated by subtracting the glucose infusion rate from the glucose disposal rate.In order to assess tissue specific glucose uptake, a bolus (3.7*10 5 bq) of 2-[1-14 C] deoxyglucose (Perkin Elmer) was administered via catheter at the end of the steady state period.Blood was sampled 2, 15, 25 and 35 min after bolus delivery.Area under the curve of disappearing plasma 2-[1-14 C] deoxyglucose was used together with tissue-concentration of phosphorylated 2-[1- 14 C] deoxyglucose to calculate glucose uptake.Accumulation of 2-[1-14 C] deoxyglucose was determined in an aqueous tissue extract after homogenization.Phosphorylated 2-[1- 14 C] deoxyglucose was separated by Poly-Prep columns (#731-6212, BioRad, USA).
Cells of the peritoneum were isolated by peritoneal lavage with 10mL FACS buffer.
Erythrocytes were lysed as described above.
To isolate colon immune cells, the colon was washed in HBBS, cut into small pieces, washed twice 10 min in HBSS containing 2 mM EDTA (Sigma) shaking at 37 °C, rinsed twice with HBSS and digested with 1mg/mL collagenase VIII (Sigma-Aldrich) and 12.5µg/mL DNase I (Roche), 30min shaking at 37 °C, followed by homogenization (gentleMACS Dissociator: mouse intestine program; Milteny Biotec, Germany).Leukocytes were enriched by Percoll (as described for liver).
Spleens were mashed with a syringe plunger on a 70µm filter in red cell lysis buffer, washed and subsequently stained.
Thymus was mashed using a syringe plunger and a 70µm filter in FACS buffer, washed and subsequently stained.
Blood was collected in FACS buffer and stained 30min and lysed using BD FACS lysis solution (BD, USA), according to the manufacturer's protocol.
Cell analysis was performed on a FACS LSRII Fortessa (BD).Acquired data were analyzed using FlowJo software (Version 10.6 or higher, TreeStar Inc. Ashland, OR, USA).
Gating strategies are summarized in ESM Figures 1 and 2.
For surface staining, single cell suspensions were stained 30min at 4°C.For intracellular staining, the Foxp3 staining kit (Thermo Fischer Scientific) was used according to the manufacturer's protocol.For cytokine staining, cells were cultured 4h in the presence of cellular protein transport inhibitors Brefeldin A (5µg/mL, Sigma) and Monensin (1µg/mL, Sigma) in 2 or 20mM glucose, followed by washing and staining (30 min, 4°C in the dark).
Intracellular cytokine staining was performed using IC Fixation Buffer (Thermo Fischer Scientific) according to the manufacturers protocol.Antibody concentrations were verified by titrating experiments.

Beta-cell mass
For heat-induced antigen retrieval, 5µm thick sections were boiled 30min at 93°C in epitope retrieval solution (Biosystems, Switzerland) and stained overnight (4°C) with an insulin antibody (Agilent, USA), washed twice in PBS (5min), stained 2h at room temperature for CD45 (Rat Anti-CD45; 30-F11, BD Bioscience), washed twice with PBS (5min) before applying the secondary antibodies (2h, room temperature; Alexa647 goat anti-guinea pig IgG and Alexa555 goat anti-rat IgG; Thermo Fisher Scientific), then washed twice (PBS), before mounting with a fluorescence mounting media (Dako).Pictures were acquired using a Nikon microscope (4x magnification).For alpha-to beta-cell ratio, slides were stained as described above using primary antibodies against insulin and glucagon (Thermo Fischer Scientitic) and secondary antibodies Alexa488 goat anti-guinea pig IgG and Alexa647 goat anti-rabbit IgG, respectively (Thermo Fischer Scientific).For details on antibodies ESM Table 1.
Images were analyzed using Fiji software.Analysis was performed in a semi-automated way; ilastik software (Version 1.3.2,https://www.ilastik.org/)was trained twice, once to recognize pancreas area excluding background and lymph nodes, and the second time to recognize islets.The masks ilastik generated were used to quantify the areas with Fiji.Beta-cell mass was defined as the ratio of insulin-positive area to total pancreas area multiplied by the weight of the pancreas.Two sections per animal 100µm apart were quantified.

RNA-Sequencing
A first batch of 9 male mice from different litters of age between 4 and 5 weeks were fed a PLX5622 or a control diet for 5.5.months.A second batch of 4 male mice of age 5 weeks were fed a PLX5622 or control diet for 8.5.months.All mice were euthanized on the same day and RNA isolation was performed on lysed pancreatic islets using a NucleoSpin RNA kit (Macherey Nagel, Germany).RNA samples were quantified and quality-controlled using a High Sensitivity RNA ScreenTape on an Agilent TapeStation instrument.RNA-seq library preparation (Illumina Truseq stranded kit, USA) was performed at the Genomics Facility Basel of the ETH Zurich, Basel, starting from 80ng of total RNA, and controlled on an Agilent Fragment Analyzer instrument.Sequencing was performed on the Illumina NextSeq 500 platform to produce paired-end 38nt reads.
Data analysis was performed by the Bioinformatics Core Facility, Department of Biomedicine, University of Basel.Read quality was assessed with the FastQC tool (version 0.11.5).Reads were mapped to the mouse genome mm10 with STAR (version 2.7.0c) 26 with default parameters, except filtering out multimapping reads with more than 10 alignment locations (outFilterMultimapNmax=10) and filtering reads without evidence in the spliced junction table (outFilterType="BySJout"). All subsequent analyses were performed using the R software (version 4.0.0) and Bioconductor 3.11 packages.The featureCounts function from the Bioconductor Rsubread package (version 2.2.3) 27 was used to count the number of reads (5' ends) overlapping with the exons of each gene (Ensembl release 96) assuming an exon union model 28,29 .
The Bioconductor package edgeR (version 3.30.3) 30was used for differential gene expression analysis.Between samples normalization was performed using the TMM method 31 .
Genes with CPM values above 1 in at least 2 samples were retained for the differential expression analysis.To test for gene expression differences between mice fed PLX5622 or a control diet, a model accounting for the treatment and treatment duration effects was fitted to the read counts using a quasi-likelihood testing framework (edgeR functions glmQLFit and glmQLFTest) 32 .P-values were adjusted by controlling the false discovery rate (FDR; Benjamini-Hochberg method) and genes with a FDR lower than 5% were considered significant.Gene set enrichment analysis was performed with the function camera 33 from the edgeR package (using the default parameter value of 0.01 for the correlations of genes within gene sets) using gene sets from the c5 collection (Gene Ontology categories) and c8 (cell type signatures) of the MSigDB Molecular Signature Database (version 7.2) 34 .We filtered out sets containing less than 10 genes, and gene sets with a FDR lower than 5% were considered significant.A list of mediators of cytokine activity was compiled by downloading the mouse genes annotated to both the GO terms "transcription factor" (GO:0003700) and "regulation of cytokine production" (GO:0001817) using the QuickGO website (download date: May 23 rd 2022) and crossed to the differential expression results (ESM Table 4).

ESM Table 3: Differential gene expression of islets upon CSF1R-inhibition by PLX5622
(separate excel file).Full results table for the differential expression analysis comparing islets isolated from mice on PLX5622 or control diet (first tab).

ESM Table 4: Gene set enrichment analysis based on the differential expression analysis, performed on the MSigDB collection c5 (separate excel file).
Full results from the gene set enrichment analysis based on the differential expression analysis, performed on the MSigDB collection c5 (second tab).

ESM Table 5: Gene set enrichment analysis based on the differential expression analysis, performed on the MSigDB collection c8 (separate excel file).
Full results from the gene set enrichment analysis based on the differential expression analysis, performed on the MSigDB collection c8 (third tab).

ESM Table 6: Gene set enrichment analysis based on the differential expression analysis, performed on the hallmark pathways (separate excel file).
Full results from the gene set enrichment analysis based on the differential expression analysis, performed on the hallmark pathways (fourth tab).

ESM
Gating strategies of macrophages, dendritic cells and eosinophils in different organs.Gating strategy in (a) colon (b) adipose tissue, (c) peritoneum, (d) brain (e) lung, (f) liver, (g) spleen, (h) blood and (i) islets.All gating strategies included a leukocyte gate, followed by doublet exclusion and exclusion of dead cells (only shown in panel a).Further gating included CD45 and lineage (CD19, Nk1.1, CD3 and Ly6G).ESM Fig. 2: Gating strategy for Innate lymphoid cells and adaptive immunity.Leukocytes were gated for live cells, and doublets excluded by FSC W and H. CD45 + cells were analyzed for T and B-cells and ILCs.T-cells were defined as CD3 + and either CD8 + and CD4 + , the latter ones were further analyzed for Tregs defined as Foxp3 + and CD25 + , Th2 identified by the expression of GATA3 and TH17 identified by RORt.ILCs were defied as CD3 -/lineage -cells, different subpopulations were further defined as RORt + (ILC3), and GATA3 + (ILC2).ESM Fig. 3: Absolute numbers of tissue resident macrophages upon PLX5622 treatment.Set-up see Fig. 1. a-c, Total macrophages (left panel) and subpopulations (right panel) of (a) colon, (b) adipose tissue and (c) lung.Data are shown as meanSEM.Data are representative of 3 (a, b) independent experiments or one experiment (c) with each data point representing an individual mouse.*p<0.05,**p<0.01,***p<0.001,unpaired Mann-Whitney U test with twotailed distribution.ESM Fig. 4: CSF1R expression in macrophages across different organs.(a) Dataset obtained from Lavin, 2014.(b) Data obtained from Brykcznska, 2020.(c) Correlation between macrophage depletion (%) and CSF1R expression in relation to CSF1R expression in microglia.ESM Fig. 5: Changes in immune cells upon CSF1R-inhibition by PLX5622.Mice were fed a PLX5622-containing (purple circles) or control diet (green triangles) starting from 4-5 weeks of age for up to 5.5 months (a, b) or starting from 8-9 weeks of age for 3 weeks (c-g).a, Absolute numbers of eosinophils.b, Absolute numbers of dendritic cells (DCs) und their subpopulations in the colon and lung.c-f, Frequencies of B and T-cells, CD4, CD8 T-cells and their subsets, NK cells, neutrophils and ILC in colon (c), lung (d), adipose tissue (e) and liver (f).g, Frequencies of B and T cells and neutrophils in blood.Data are shown as meanSEM.Pooled data of three independent experiments (c-g) or representative of three or one experiment (a, b) with each data point representing an individual mouse.*p<0.05,**p<0.01,***p<0.001,unpaired Mann-Whitney U test with two-tailed distribution.ESM Fig. 6: Depletion of macrophages does not affect T cell development.Mice were fed a PLX5622-containing (purple circles) or control diet (green triangles) for 3 weeks.Total thymocytes (left panel) and double negative thymocytes (right panel).Data are shown as meanSEM.One experiment with 3 mice each, each data point representing an individual mouse.treatment.Mice were treated for three weeks with the CSF1R antibody AFS98 (blue circles) or a rat IgG2a isotype control (green triangles) and immune cells of different organs were assessed by flow a-c, Total macrophages (left panel) and subpopulations (right panel) of (a) colon, (b) adipose tissue and (c) lung.Data are shown as meanSEM.One experiments, with 5 mice each, each data point representing an individual mouse.**p<0.01,unpaired Mann-Whitney U test with two-tailed distribution.ESM Fig. 8: Changes in immune cells upon CSF1R antibody treatment.Mice were treated for three weeks with the CSF1R antibody AFS98 (blue circles) or a rat IgG2a isotype control (green triangles).a-d, Frequencies of B and T-cells, CD4, CD8 T-cells and their subsets, NK cells, neutrophils and ILCs in colon (a), lung (b), adipose tissue (c) and liver (d).e, Frequencies of B and T cells and neutrophils in blood.Data are shown as meanSEM.One experiments, with 5 mice each, each data point representing an individual mouse.*p<0.05,**p<0.01,unpaired Mann-Whitney U test with two-tailed distribution.ESM Fig. 10: Macrophage depletion reduces insulin secretion in wild type islets.Islet macrophages were depleted either ex-vivo by clodronate treatment for 24 hours or by AFS98 antibody (i.p. 1mg/mouse) once weekly over 3 weeks.a,b Macrophages were depleted by clodronate (turquoise circles) or PBS liposomes (control; black triangles) treatment over 24 hours.(a) Frequency of islet macrophages after 24 hours of culture with clodronate or PBS liposomes.(b) Ex-vivo insulin release in a glucose stimulated insulin secretion assay (GSIS) after ex-vivo treatment of islets with clodronate or PBS liposome.c, d Mice were treated over 3 weeks once weekly with the CSF1R anti-body AFS98 (blue circles) or its corresponding control anti-body (green triangles) at a dose of 1mg/mouse (i.p.).(c) Frequency of islet macrophages after treatment.(d) Insulin release ex-vivo in a glucose stimulated insulin secretion assay (GSIS) over one hour of islets from treated mice.Data are shown as meanSEM, Pooled data from two independent experiments (b), three independent experiment (c) or one experiment (c, d), with each point representing one mouse (d) or technical replicates (b) unpaired Mann-Whitney U test with two-tailed distribution.ESM Fig. 11: Heatmap showing the centered and scaled normalized expression levels of genes differentially expressed in islets of Langerhans upon CSF1R-inhibition.FDR<5%