Microglia replacement by peripheral delivery of CSF1R inhibitor-resistant hematopoietic cells

Abstract

Microglia replacement holds great promise for the study and treatment of neurological diseases. We previously achieved high-efficiency replacement by engineering an inhibitor-resistant human CSF1R variant, G795A. Here and in Chadarevian et al, we introduce a transgenic mouse line carrying the homologous murine inhibitor resistance mutation, G793A. G793A confers resistance to CSF1R inhibitors (CSF1Ris) without impacting cell or organismal function, allowing brain-wide microglia replacement after intravascular delivery of myeloid cells or progenitors. G793A macrophages have normal CSF1R signaling and effector responses in cell-based assays. CSF1Ri treatment paired with peripheral vascular delivery of G793A myeloid cells, including ex vivo expanded hematopoietic stem cells, leads to microglia replacement in neonates and adults, with low peripheral chimerism. With vascular adoptive transfer of GFP⁺ G793A donor cells, engraftment is inconsistent and variable. Rag1-deficient hosts or GFP^- donor cells, however, led to robust replacement of virtually all microglia, suggesting that G793A-based microglia replacement substantially spares host adaptive immune function. Although equally efficient, microglia replacement using G793A donors is less toxic to neuronal and oligodendrocyte progenitors than traditional hematopoietic stem cell transplantation. We present an approach for robust, reduced toxicity microglia replacement by intravenous adoptive transfer using a new research tool, the G793A mouse.

Keywords: CSF1R; microglia; microglia replacement; myeloid cell therapies; neuroimmunology; stem cell transplantation.

Figure 1:. G793A macrophages are resistant to CSF1R inhibitors.

Survival curves of WT (black) and G793A (blue) BM macrophages treated for seven days with PLX3397 (A), PLX5622 (B) or BLZ945 (C). IC50 values calculated from mean survival values ± SEM from n=6 biological replicates across two independent experiments. Plots depict n=3 replicates from one experiment, with normalized values and nonlinear fit line plotted. (D) Immunoblots showing pCSF1R Y723, total hCSF1R, pERK, and Actin expression in WT and G793A BMDMs at a range of PLX3397 concentrations, three minutes following spike-in of 360 ng/mL mCSF1, annotated with mass of protein ladder (kD, right). Representative of three independent experiments. (E,F) Representative flow cytometry plots from adult WT (E) and G793A (F) mice given PLX3397 diet (600 mg/kg) for two weeks. Oval indicates brain macrophage gate. Gating strategy: Cells/Singlet/Live/CD11b+;CD45+. (G) Microglia frequencies from flow cytometry analysis of WT (black striped), WT + PLX3397 (black), G793A (blue striped), and G793A + PLX3397 (blue) CNS suspensions. n=6-12/group across four experiments, gating in Fig S1H. (H) Representative histograms of TMEM119 expression in macrophage gate. (I) Average TMEM119 MFI in macrophage gate. n=5-6/group, across two experiments. (J) Quantification of microglia density by IBA1 immunostaining within the cortex. N=2-3/group, across two experiments. (K,L) Rendering of representative sagittal sections following 14D PLX treatment (600 mg/kg chow) from WT (K) and G793A (L) adults. IBA1+ cells are represented by red dots overlaid on DAPI (blue). Scale bar = 1,000 μm. (M) Representative immunostaining for IBA1 (red) with DAPI (blue) from WT, WT + PLX3397, G793A and G793A + PLX3397 animals. Scale bar = 50 μm. For all statistics unless otherwise noted: Compared by one-way ANOVA with multiple comparisons. Post-hoc multiple comparisons were performed using the Sidak correction. Error bars, ± SEM; ns = not significant or p >= 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 2:. G793A donor BM and exHSPCs engraft the CSF1Ri-treated neonatal brain after vascular delivery.

(A) Schematic of intrahepatic delivery of G793A-GFP bone marrow into WT neonatal hosts. Hosts transplanted p0-4 with donor cells and treated with PLX3397 (25 mg/kg) daily for 21 days (solid red line), then off PLX for 2 weeks until harvest (dashed red line). (B) Rendering showing GFP+ donor cells (green dots) from a well engrafted animal after transplant as described in (A). Bottom right, average engraftment area +/− SEM. N=6, across two experiments. (C) Schematic of intrahepatic delivery of G793A-GFP exHSPCs into WT neonatal hosts. Hosts transplanted p0-4 with donor cells and treated with PLX3397 (25 mg/kg) daily for 21 days (solid red line), then off PLX for 2 weeks until harvest (dashed red line). (D) Rendering showing GFP+ donor cells (green dots) five weeks after intrahepatic delivery of G793A-GFP+ exHSPCs. Bottom right, average engraftment area +/− SEM. N=5 across two experiments. (E) Representative images of G793A-bone marrow (top) and -exHSPCs (bottom) derived brain macrophages five weeks post-transplant. IBA1 immunostaining (red), DAPI (blue) GFP-florescence (green). Scale bar = 50um. (F) Percent brain area occupied by GFP+ donors: wildtype (WT) or G793A (“IRR”) bone marrow (BM) or exHSPC transplants with (“+”) or without (“−”) PLX3397 regimen described in (A,C). WT BM, n=7; WT exHSPCs, n=5; IRR BM, n=6; IRR exHSPCs, n=5 across two experiments. One-way Anova, with post-hoc pairwise comparisons adjusted for multiple comparisons. (G) Donor cell density within engrafted regions of host cortex five weeks post- transplant. N=3/group, across two experiments. (H) Representative flow cytometry plots showing GFP+ and GFP− brain cells in CD45;CD11b gate from animals receiving G793A-BM (left), WT-exHSPCs (middle), or G793A-exHSPCs (right) donor cells as in (A,C). Green, GFP+ events, Gray, GFP− events. Arrow denotes endogenous repopulated microglia. Gating: Cells/Live/Single Cells/CD11b⁺/CD11b⁺;CD45⁺. (I) Histogram showing GFP fluorescence within brain CD11b⁺;CD45⁺ gate from n=3 animals receiving WT- (black) or G793A-exHSPCs (blue). (J) TMEM119 MFI of brain CD11b+/CD45+ cells as gated in (H). (K) Brain chimerism measured by % GFP+ cells within CD11b+;CD45+ gate of animals treated as in (A,C). (L) Peripheral myeloid chimerism as measured by % GFP+ among CD11B cells within spleen, bone marrow and blood samples via flow cytometry. n=5 per group across two experiments. All paired comparisons are ns by one-way ANOVA with multiple comparisons. Post-hoc multiple comparisons were performed using the Sidak correction. For all statistics unless otherwise noted: Error bars, ± SEM; ns = not significant or p >= 0.1, #p<0.1 but >=.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 3:. G793A bone marrow, monocytes, and exHSPCs extensively but variably engraft the CSF1Ri-treated adult brain after vascular delivery.

(A) Top, experimental schematic showing timing of PLX only (blue) and PLX/busulfan (red→ blue) treatments. Bottom, rendering of GFP+ donor cells (green dots) in a well engrafted PLX only brain, five weeks after retroorbital (RO) delivery of G793A-GFP+ exHSPCs. Average engraftment area=39.39% +/− 10.40 SEM. N=18, across three experiments. Scale bar = 1,000 μm. (B) Representative images from adult cortex transplanted with either G793A- or WT-exHSPCs. IBA1 (red), DAPI (blue), donor GFP (green). Scale bar = 100 μm. (C) Brain area engrafted following transplant, by tissue immunostaining (percent engrafted by GFP+ donor cells) in G793A+PLX and control transplants. Row labels indicate: PLX treatment (top row, given PLX3397 = “+”, untreated = ”−”), busulfan conditioning (second row, given busulfan = “+”, untreated = ”−”), donor cell CSF1R genotype (third row, “WT” or “IRR”), and donor cell type (bottom row, no donors (“−”), monocyte (“mono”), or exHSPCs. Red dots represent additional animals maintained on PLX3397 for 6-months. PLX only, n=5; WT exHPSCs - PLX, n=2; IRR exHPSCs - PLX, n=6; WT exHSPCs + PLX, n=6; IRR exHSPCs + PLX, n=18; IRR monocytes, n=3; IRR exHPSCs + Busulfan, n=6, across three experiments. (D) Brain chimerism by flow cytometry, measured as percent GFP⁺ in CD45⁺;CD11b⁺ gate. (E,F) Representative histograms showing GFP (E) and TMEM119 (F) expression within brain CD45⁺/CD11B⁺ gate. (G-H) Renderings showing animals receiving G793A-exHSPCs by RO injection and maintained on PLX3397 until harvest, with significant (G) or dominant (H) contribution from IBA1⁺/GFP⁻ macrophages. Scale bars = 1,000 μm. (I-J) Renderings of Ms4a7 and Gfp in situ from same animals shown in G-H, non-adjacent sections. Red dots represent Ms4a7+ and yellow dots represent Ms4a7+/Gfp+ cells. Scale bars = 1,000 μm. (K) Representative images showing Ms4a7 (red) and Gfp (green) in situ hybridization with IBA1 immunostaining (white) and DAPI (blue) from GFP+ and GFP− cells from same animals as in G (Ki) an H (kii), respectively. Scale bar 50 μm. (L) Brain engraftment by tissue immunostaining (percent engrafted by IBA1⁺ cells) in G793A+PLX and control transplants shows greatly increased apparent chimerism compared to panel C, which measured only GFP+ cells. PLX only, n=5; WT exHPSCs - PLX, n=2; IRR exHPSCs - PLX, n=6; WT exHSPCs + PLX, n=6; IRR exHSPCs + PLX, n=18; IRR monocytes, n=3; IRR exHPSCs + Busulfan, n=6, across three experiments. (M) Brain chimerism, measured as percent CD45⁺ in CD11b⁺ gate, similarly showing increased apparent chimerism compared to panel D. (N) Brain chimerism of hosts receiving G793A-GFP monocytes + PLX3397 by percent GFP⁺ (left bar) vs percent CD11b⁺;CD45⁺ (right bar, as also appears in panel M). n=3, from one experiment. Dotted lines connect corresponding values for a single animal. Post-hoc multiple comparisons were performed using the Sidak correction. For all statistics unless otherwise noted: Error bars, ± SEM; ns = not significant or p >= 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Unmarked paired comparisons are ns by one-way ANOVA with multiple comparisons.

Figure 4:. Host immunodeficiency or immunologically silent donor cells increase engraftment and rescue transplant variability.

(A) Rendering of representative sagittal section showing GFP⁺ donor cells (green) overlaid on DAPI (blue) after RO delivery of G793A-GFP+ exHSPCs into adult immunocompromised hosts (Rag1^−/−) treated with PLX3397 for 5 weeks, as depicted in Fig 3A. Average engraftment area=90.13 +/− 1.65% (by percent GFP+ in macrophage gate by flow cytometry). n=8, across two experiments. (B) Representative sagittal brain section showing IBA1 immunostaining (green) overlaid on DAPI (blue) after RO delivery of G793A-CD45.2+ exHSPCs into congenically mismatched hosts (CD45.1) treated with PLX3397 for 5 weeks following donor cell injection as depicted in Fig 3A. Average donor engraftment 86.75 +/− 4.64% (by percent CD45.2 positive in macrophage gate by flow cytometry). n= 4, from one experiment (C) Chimerism measured as percent GFP+ or CD45.2+ in myeloid gate by flow cytometry. Engraftment of GFP+ donor cells in immunocompetent hosts from figure 3D is shown for reference in gray. Conditions: With/without PLX3397 (“+”/”−”), with/without busulfan (“+”/”−”), with either no donor cells (“−”), hematopoietic stem and progenitor cells (“exHSPC”), or monocytes (“Mono”), into wildtype C57bl6J hosts (“WT”), Rag1 KO hosts (“Rag1^−/−”) or C57bl6J-CD45.1 hosts (“45.1”). Gating: Cells/Single Cells/Live/CD11b+/ %GFP or %CD45.2. Donor cells all CD45.2. WT, n=5; WT + G793A-exHSPCs + Busulfan, n=2; Rag1^−/− + G793A-exHSPCs + PLX, n=8; Rag1^−/− + WT-exHSPCs + PLX, n= 3; CD45.2 + G793A-exHSPCs + PLX, n=4; CD45.2 + WT-exHSPCs + PLX, n=3; CD45.2 + G793A-monos + PLX, n=4; CD45.2 + WT-monos + PLX, n=4, across four experiments. All paired comparisons analyzed by one-way ANOVA with multiple comparisons. Post-hoc multiple comparisons were performed using the Sidak correction. Error bars, ± SEM; ns = not significant or p >= 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 5:. IRR-CSFRi-based microglia replacement avoids neurotoxicity caused by busulfan conditioning.

(A) Schematic of adult transplants using PLX3397 only (top, D-5 to D35) or Busulfan (bottom, D-5 to D-1) + PLX3397 (D24-35). “PLX On” and “Busulfan” groups harvested at D35. An additional group, “PLX On/Off”, was subjected to a washout period from D35 until harvest on D49. Inset shows representative hippocampus with white dotted line depicting dentate gyrus (DG) region for immunostaining quantifications. (B) DCX counts from DG across unmanipulated controls (white bar), busulfan pretreatment (diagonal stripes), PLX On (horizontal stripes), and PLX On/Off (chevron). WT/unmanipulated, n=3; Busulfan, n=5; Plx On, n=6; PLX On/Off, n=3 from three experiments. (C) PDGFRα counts from DG across unmanipulated controls (white bar), busulfan pretreatment (diagonal stripes), PLX On (horizontal stripes), and PLX On/Off (dots). WT/unmanipulated, n=3; Busulfan, n=8; Plx On, n=7; PLX On/Off, n=4 from three experiments. (D) Representative images of DCX (red) immunostaining overlaid on DAPI (blue) within DG region across treatment conditions. (E) Representative images of PDGFRa (white) immunostaining overlaid on DAPI (blue) and GFP+ donor cells (green) within DG region across treatment conditions. All paired comparisons analyzed by one-way ANOVA with multiple comparisons. Post-hoc multiple comparisons were performed using the Sidak correction. Error bars, ± SEM; ns = not significant or p >= 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.