Promoting donor microglial replacement through augmented conditioning or radiation sensitivity

Abstract

Donor myeloid cells engraft the brain after hematopoietic cell transplantation (HCT), partially replacing native microglia. The conditions which promote or maintain engraftment are not well known. Previously, we demonstrated that mice receiving double HCT using radiation-based conditioning had higher brain engraftment but also mortality of 70%. We now show that mice that undergo double HCT with busulfan preconditioning have nearly 70%–80% brain engraftment and lower mortality of 8%. Using the radiation-sensitive NSG mouse model, we found up to 15% donor myeloid (CX3CR1⁺) brain engraftment after low-dose radiation (450 cGy) vs. 0.1% in C57BL/6 mice treated similarly. NBSGW mice engraft the marrow without the need for conditioning, and we found that in these mice the addition of radiation boosted brain engraftment greater than 10-fold, from 0.3% (no conditioning) to 4.8% after HCT. A further increase in brain engraftment (median 22.5%) was observed when animals were pretreated with PLX3397, a microglia inhibitor, which indicated that myeloid niche clearance promotes brain engraftment. In conclusion, preconditioning and niche clearance play an important role in brain engraftment. New ways of microglia replacement are needed if we are to avoid the use of preconditioning.

Graphical abstract

Lund and colleagues determined that myeloid donor engraftment of the brain is enhanced with increased intensity of preconditioning before hematopoietic cell transplantation. Alternatively, utilizing mice with genetic radiation sensitivity also permitted a high level of donor myeloid brain engraftment due to enhanced microglial niche clearance.

Introduction

Central nervous system (CNS) engraftment after hematopoietic cell transplantation (HCT) occurs as a result of migration to the brain of donor cells that become microglia-like in nature and replace a portion of the native microglia. This phenomenon has likely occurred since the initial HCT was performed.

Donor-derived cells in the CNS after HCT are known to be myeloid in origin. They are often referred to as peripheral derived myeloid cells, peripheral macrophage cells, or microglia-like cells. Pre-HCT conditioning regimens that penetrate the blood-brain barrier (BBB) are critical for donor cells to reach the CNS after HCT, and they are required to be myeloablative in nature. We have shown that a reduction in radiation from 900 cGy to 450 cGy (50%) led to a nearly complete loss (>90%) of CNS engraftment in a murine model.¹ Furthermore, Capotondo et al. showed that pre-HCT busulfan, a BBB-penetrating agent, led to robust donor myeloid CNS engraftment, while with the use of treosulfan, which does not penetrate the BBB, pre-HCT fails to promote CNS engraftment.² Using BBB-penetrating therapy (busulfan or radiation) to promote myeloid engraftment of the brain involves the clearance of native microglia, which is likely necessary to “make space” for incoming donor cells.³ In addition, data support that pre-HCT conditioning sets up an inflammatory signal cascade that recruits myeloid cells via CCR2, MCP-1, and other CCL family mediators.^4,5

We have previously reported the loss of myeloid donor cell chimerism in the brain with the use of nonmyeloablative therapy as opposed to myeloablative therapy in a murine model, although we were able to achieve 100% myeloid engraftment in the brain through back-to-back HCTs when lethal radiation was used both times as preconditioning.¹ Unfortunately, this scenario led to an overwhelming 70% mortality, given that lethal radiation was applied twice only 30 days apart. We now explore alternative variations in HCT conditioning to achieve myeloid donor brain engraftment, including the testing of animals that are sensitive to radiation.

Results

Our previous study used a double-transplant (DT) approach with 900 cGy radiation for both HCTs, which were separated by 30 days. While the data showed 80%–90% myeloid CNS chimerism, the mortality was 70%.¹ In an attempt to reduce mortality, we next combined busulfan and radiation as preconditioning in a DT approach (Figure 1A); busulfan was given for HCT #1, and 30 days later 900 cGy was used as preconditioning for HCT #2. As previously published, we utilized brain CX3CR1⁺ cells to measure myeloid donor cell chimerism in the brain (Figure 1B). We found that survival was excellent at 100%, and peripheral as well as marrow engraftment was also near 100% in this DT approach (Figure 1C). Brain engraftment showed myeloid donor chimerism at 47% and slightly less than our prior DT with radiation (Figure 1D).

Figure 1.

Double transplants enhance donor myeloid brain engraftment

(A) Paradigm for busulfan-irradiation HCT. Note that busulfan was given as 15 mg/kg intraperitoneally per day for 5 days to achieve the given total dose in both experiments. BM, bone marrow cells. (B) Paradigm for brain myeloid enrichment. (C) Peripheral blood mononuclear cell (PBMC) subset and marrow LSK engraftment at 90 days following second HCT. (D) Donor myeloid engraftment in the brain 90 days following second HCT. (E) Paradigm for busulfan-busulfan HCT. (F) PBMC subset and marrow LSK engraftment at 90 days following second HCT. (G) Donor myeloid engraftment in the brain 90 days following second HCT. (H) Paradigm for busulfan-radiation HCT ± head shield (HS) with different donors. (I) Donor myeloid engraftment in the brain 60 days following second HCT. (J) PBMC, whole bone marrow (WBM), and marrow LSK engraftment at 60 days following second HCT. Note that two animals experienced graft failure after the second HCT and were in the HS group as noted. They were also the lowest in CD45.1 brain engraftment in the HS group. (I) shows a p value from a Mann-Whitney test of GFP engraftment and from a Welch’s t test of CD45.1 engraftment. (G) and (J) show medians. All other panels show means and SD.

We next tested an “all-busulfan”-based DT strategy whereby 75 mg/kg of busulfan was given followed by HCT; a second HCT was performed 30 days after the first, again with busulfan pretreatment (Figure 1E). Using this approach, we were able to achieve high levels of peripheral and marrow donor chimerism (Figure 1F). Myeloid donor chimerism in the brain was greater than that in the prior experiment, at close to 80% in most animals (median 82%, Figure 1G), while mortality was 8.3%. To investigate the histology of the engrafted cells, we costained for CD68, generally a marker upregulated on activated microglia cells, and GFP; we found that many of the GFP⁺ engrafted cells clearly expressed CD68, although some host residual cells remained (Figure S1). Border-associated macrophages (BAMs) have also been shown to be replaced by cells from the circulation after CSF1R inhibitor exposure.^6,7 In our busulfan-based DT, we did not find a significant amount of BAM replacement after performing immunostaining for CD206 and GFP (Figure S2), which may be explained by differences between our model system and the published studies.

To understand whether the first HCT brain engrafted myeloid cells persisted during the second HCT, we performed HCT using GFP⁺ donor cells following busulfan conditioning, followed by a second HCT using radiation but with (or without) a head shield to protect the brain from radiation (Figure 1H). With the head shield, the HCT GFP⁺ donor cells remained engrafted and were 100% depleted without the head shield. The second HCT donor (CD45.1) was highly engrafted without the head shield and much less so in mice with the shield, emphasizing the significant contribution of radiation exposure to peripheral myeloid brain engraftment.

Recent work by Sailor et al. showed that resident microglia niche clearance is required for robust CNS engraftment.³ Of note, microglia are myeloid lineage cells with a relative radioresistance.⁸ The NSG mouse (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) is a highly immune-deficient and radiosensitive (due to a defect in DNA repair) model allowing for superior bone marrow engraftment of human hematopoietic stem cells following preconditioning.⁹ We found that NSG mice as recipients (Figure 2A) had lower levels of peripheral and marrow engraftment without preconditioning, with donor HSC content being <10% (Figure 2B). CX3CR1⁺ myeloid brain engraftment was nominal without conditioning (<0.1%, Figure 2C). When animals were preconditioned with 450 cGy (their lethal dose), near 100% peripheral and marrow engraftment was achieved, and brain engraftment increased 41-fold to 3.3% donor chimerism (Figures 2B and 2C), which was significantly greater than in C57BL/6 mice treated with the same dose of radiation (p = 0.0029).

Figure 2.

Radiation sensitivity enhances donor myeloid brain engraftment following HCT

(A) Paradigm for HCT in NSG mice. (B) Peripheral blood mononuclear cell (PBMC), bone marrow mononuclear cell (BMMC), marrow LSK, and marrow LSK-SLAM engraftment at 30 days following HCT. (C) Donor myeloid brain engraftment in the brain 30 days following HCT. (D) Paradigm for NBSGW in NSG mice. (E) PBMC, BMMC, marrow LSK, and marrow LSK-SLAM engraftment at 30 days following HCT. (F) Donor myeloid brain engraftment in the brain 30 days following HCT. (G) Donor myeloid brain engraftment as a function of residual host brain CX3CR1⁺ cell count. Shaded area represents 95% confidence interval. p value derived from least-squares linear regression. (H) Paradigm for HCT in NBSGW mice preceded by PLX3397 treatment. (I) PBMC, BMMC, marrow LSK, and marrow LSK-SLAM engraftment at 30 days following HCT. (J) Donor myeloid brain engraftment in the brain 30 days following HCT. Shown are means and SD with p values derived from a Student’s t test.

Another mouse model often utilized to engraft human cells is the NBSGW mouse (NOD.Cg-KitW-41J Tyr + Prkdcscid Il2rgtm1Wjl/ThomJ), which does not require preconditioning. This mouse, which has an NSG background, also harbors a Kit (stem cell factor receptor) mutation that allows HSC (with normal Kit) to outcompete the resident cells after HCT, which can be performed without pre-HCT radiation or chemotherapy. We showed that without radiation, peripheral blood and marrow can engraft with donor cells at levels of 30%–50%, although there was minimal brain engraftment (Figures 2D and 2E). Only after the addition of radiation (450 cGy) did we achieve a 15.5-fold increase in levels of brain chimerism to 4.34% (Figure 2F). Additionally, we found an inverse relationship between numbers of residual host CX3CR1⁺ cells and percentage of brain donor myeloid chimerism when evaluating both NSG and NBSGW experiments (p = 0.0036, Figure 2G), suggesting that reduction in host CX3CR1⁺ cells plays a role in myeloid brain engraftment and lending support to the idea that “space” in the brain needs to be made for efficient engraftment to occur.

To further deplete host microglia specifically, we utilized PXL339,7 a known CSF1R inhibitor that can deplete microglia cells efficiently (<4 weeks) when given in mouse chow.¹⁰ When we combined PLX3397 with conditioning in NBSGW mice, we were able to augment donor CX3CR1⁺ brain chimerism to 22.6% (median), which was significantly greater than with radiation alone (p = 0.0066, Figure 2J). Interestingly, the use of PLX3397 by itself did not result in elevated levels of CX3CR1⁺ brain (or marrow) chimerism, suggesting that conditioning is important in our system.

Discussion

CNS engraftment is important as HCT is being used or being considered for various neurological disorders. Some of these are “microgliopathies,” such as adult-onset leukoencephalopathy with axonal spheroids or adrenoleukodystrophy (ALD),^11,12,13 for which microglia replacement via HCT offers effective treatment. Other “storage disorders” such as mucopolysaccharidosis type 1H (MPS-1H) or metachromatic leukodystrophy rely on HCT to be the definitive BBB-penetrating therapy to allow enzyme delivery (via donor myeloid engraftment) to the CNS. We and others have shown a clear drop in CNS opening pressure and significant decrease in cerebrospinal fluid (CSF) glycosaminoglycans after HCT in MPS-1H patients, which are believed to be surrogate markers of CNS engraftment.^14,15 CNS engraftment is probably also important in the control of autoimmune diseases such as multiple sclerosis and neuromyelitis optica,¹⁶ but whether it acts via the same mechanism as with the pediatric storage diseases is unknown.

Previously, we demonstrated 6% donor myeloid engraftment of the murine brain at 100 days following HCT. This was consistent with the findings of Loeb et al. who reported 8.1% donor microglia brain engraftment in humans after HCT utilizing autopsy specimens and sex-mismatched donors.¹⁷ They further demonstrated that the greatest level of donor myeloid brain engraftment (mean 16.3%) occurred in patients who had received multiple HCTs (n = 4), which agreed with our data in the murine model involving multiple HCTs, although mortality was 70% in our prior model.¹ We went on to further demonstrate that a reduction in preconditioning radiation of 50% led to a >90% reduction in brain donor myeloid cell engraftment. We now have utilized murine models with enhanced radiosensitivity as HCT recipients to improve myeloid donor engraftment, summarized in Table S2.

The development of humanized mouse models has greatly accelerated the field of hematopoietic stem cell biology. SCID animals were discovered to be lymphopenic and to have the ability to engraft human cells in 1983.¹⁸ It wasn’t until 1995 that the gene responsible for the lack of double-stranded DNA repair, Prkdc, was determined to be the root cause of the SCID mouse’s immune defects.¹⁸ Given the ubiquitous expression of Prkdc, most cells are radiosensitive in animals harboring this mutation, but especially cells with a hemopoietic origin given their propensity for cell division. While we show superior myeloid brain engraftment in radiosensitive murine models compared to wild type, our data do not exclude the possibility that an immune process directed against GFP-expressing cells (our most common donor) may slow brain engraftment in wild-type animals given that NSG mice lack functional B cells, T cells, and natural killer cells,¹⁹ although we have tested brain engraftment in RAG-deficient mice that have defects in B and T cells and have found them to be no different from wild-type mice (data not shown).

Early experiments by Priller et al. demonstrated that GFP-marked hematopoietic cells could engraft the CNS after radiation-based conditioning.²⁰ Historically, some type of conditioning (chemical or radiation) seems to be necessary to engraft the CNS given that parabiosis experiments demonstrated little to no donor cell transfer to the CNS from one mouse to another.²¹ Radiation has been historically used to study donor brain engraftment given its history in human (and mammalian) HCT, ease of administration, and predictability. Unfortunately, radiation can cause long-term damage to the host including the brain and other organs.²² Clearance of the microglia niche using busulfan (a strong alkylator) is also an effective preconditioning agent for HCT and also achieves brain engraftment, although which method is most effective is of some debate.^5,22 A recent report by Sailor et al. showed that busulfan conditioning caused host microglia to undergo cellular senescence followed by depletion and subsequent replacement by donor cells (in this study termed macrophages).³

We were able to show partial donor myeloid brain engraftment after ablative HCT in wild-type animals. One of the key determinants of CNS engraftment was radiation sensitivity, with the elevated ability of NSG mice to engraft donor cells at greater amounts even with reduced conditioning compared to wild-type mice. The argument that DNA sensitivity to radiation in these animals is the significant factor is supported by the data from NBSGW mice, which are capable of robust HSC engraftment without conditioning; however, their CNS engraftment is very low unless pre-HCT conditioning is given, suggesting that an important element is the radiation.

We believe that the findings of the current study are important, as newer microglia-targeted replacement therapies are being developed to target a wider range of neurological diseases such as Alzheimer’s and Parkinson’s.^23,24 Understanding the critical determinants of CNS engraftment is key to optimizing therapies targeting the CNS.^24,25 Alternative technologies and methodologies, allowing for microglial replacement while avoiding the use of toxic agents, are advantageous due to their lack of genotoxicity and overall systemic toxicity from chemotherapy and/or radiation.

Materials and methods

Bone marrow transplantation

B6.SJL-Ptprca Pepcb/BoyJ (“CD45.1” mice) were obtained from Jackson Labs (strain #0021014). NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (“NSG,” strain #005557) and NOD.Cg-Kit^W41J Tyr ⁺ Prkdc^scid Il2rg^tm1Wjl/ThomJ (“NBSWG,” strain #026622) were initially purchased from Jackson Labs following by in-house breeding. “GFP” mice were bred in-house and used as donor as previously described.¹ All protocols followed were approved by the Institutional Animal Care and Use Committee, University of Minnesota (protocol #2504-42891A). Mice were irradiated (450 cGy or 900 cGy) using an X-RAD 320 (Precision X-Ray, North Branford, CT). In some experiments, the head was shielded using a Quad Fixture (Precision X-Ray, Madison, CT). Twenty-four hours post irradiation, mice received up to 5 × 10⁶ sterile donor whole bone marrow cells administered in a total volume of 200 μL of phosphate-buffered saline (PBS) (pH 7.4) via tail vein injection. Alternative conditioning consisted of 15 mg/kg busulfan (Busulfex) given intraperitoneally for 5 days prior to HCT (last dose 24 h prior to HCT). All mice were monitored daily for 12 days after conditioning for toxicity.

Blood and bone marrow isolation for flow cytometry

For complete blood counts, mice had peripheral blood drawn via facial vein bleeding with subsequent analysis on a Hemavet (Drew Scientific, Miami Lakes, FL). For marrow isolation, mice were euthanized by CO₂ according to approved animal protocols, and the euthanasia result was assessed by palpebral and pedal reflexes. Mice were then transcardially perfused with ice-cold PBS (pH 7.4). After perfusion, the hip, femur, and tibia were removed and placed in 5 mL of sterile ice-cold PBS+ (PBS + 2% fetal bovine serum [FBS]). The bones in PBS+ solution were crushed with a mortar and pestle until finely ground. The mixture was collected and filtered through a 40-μm filter into a 50-mL conical tube and placed on ice. The bones were washed and recrushed with enough PBS+ to cover the base of the mortar. To extract leftover bone marrow, bone fragments from the mortar were transferred to a conical tube that contained Accutase (#07922; STEMCELL Technologies, Cambridge, MA). Tubes were shaken for 5 min at 37°C. To stop the Accutase reaction, 5 mL of PBS was added to the fragment/Accutase mixture and shaken vigorously for 15 s. The supernatant was filtered through a 40-μm filter into the conical tube containing the previously filtered bone marrow cells, and another 5 mL of PBS was added and filtered into the same conical tube. Bone marrow mixtures were spun down at 600 × g for 5 min at 4°C. After the supernatant was discarded, 1× red blood cell lysis solution (RBC lysis,#420301; BioLegend, San Diego, CA) was added to the bone marrow pellet and incubated for 5 min on ice. An equal volume of PBS was added to stop the reaction of the lysis buffer and spun down at 600 × g for 5 min at 4°C. Bone marrow pellets were resuspended with PBS+ and filtered through a 40-μm filter to obtain a single-cell suspension of bone marrow. Cells were then incubated with antibodies of interest on ice for 1 h and washed once with fluorescence-activated cell sorting (FACS) buffer prior to flow cytometry (Table S1). In some experiments, cells were stained with CellROX Green (Thermo Scientific, #C10444) at a final concentration of 5 μM. Staining was done in darkness for 30 min at 37°C. Cells were washed with PBS and resuspended in 100 μL of FACS PBS+ for further surface marker staining. Stained cells were analyzed using a BD FACSymphony A3 Cell Analyzer, and results were obtained via analyses using FlowJo v.10.9.0. The antibodies used are listed in Table S1.

Microglia enrichment

The remaining hemisphere of the brain was placed into a 50-mL conical tube containing 10 mL of cold Hank’s balanced salt solution (HBSS, with no calcium, no magnesium, and no phenol red; Thermo Fisher, #14-185-052). Once all harvests were completed, the brain was removed from the HBSS and placed into a GentleMACS C tube containing 3 mL of room-temperature Accutase (STEMCELL Technologies, #07922). Each brain was run through three brain programs on the gentleMACS dissociator (Miltenyi Biotec, Gaithersburg, MD). The C tubes were then pulsed to 300 × g in a centrifuge and moved into an incubated shaker at 37°C for 40 min with gentle shaking. While the brains were incubating, 70-μm filters were placed atop 50-mL conical tubes and rinsed with 1 mL of HBSS. After incubation, 7 mL of HBSS was added to each C tube, and the homogenates were poured through the 70-μm filter. Using the plunger of a syringe, the homogenized brain was ground through the filter. The C tubes were rinsed with 5 mL of HBSS, and the remaining contents were filtered into the same tube. They were then spun down at 300 × g for 10 min at 22°C. After aspiration of the supernatant, the pellet was resuspended in 7 mL of HBSS+ (HBSS + 2% heat-inactivated FBS) and 3 mL of SIP (9 parts Percoll [Sigma #P1644-500ML, pH 8.5–9.5] and 1 part 10× PBS). One 15-mL conical tube containing 750 μL of 1× HBSS and 1,750 μL of SIP, to make a final concentration of 70%, was prepared per brain. The 10-mL cell suspension was gently layered on top of the 70% SIP and centrifuged for 30 min at 300 × g with slow start and no break. The P1000 tips were cut so that the myelin fat layer at the top of the tube could be easily removed and discarded with a micropipette. A new 15-mL conical tube containing 9 mL of HBSS+ was prepared for each brain. 3.5 mL of the interphase near the bottom third of the tube was placed into the tube containing 9 mL of HBSS+ and centrifuged for 7 min at 500 × g. The supernatant was aspirated, and the pellet was resuspended in the desired amount of FACS buffer. Cells were then incubated with antibodies of interest on ice for 1 h and washed once with FACS buffer prior to flow cytometry.

Data and code availability

A de-identified dataset that underlies the results reported in this article may be made available on reasonable request to academic investigators without commercial employment for the purpose of verifying and replicating results.

Acknowledgments

This work was funded, in part, by X out ALD and Knockout ALD.

Author contributions

T.C.L. analyzed the data and wrote the initial draft; A.O.G. and P.J.O. critically edited the manuscript; and W.D., E.S., E.E.N., M.M.C., and K.B. performed murine HCT experiments.

Declaration of interests

The authors declare no competing interests.