Background
Mobilisation of hematopoietic stem and progenitor cells (HSPC) into peripheral blood is used to enable collection of enriched hematopoietic stem cells (HSC) for transplantation. This procedure is a successful approach used to treat a broad range of immune and hematopoietic malignancies and deficiencies. In up to 40% of patients referred for autologous transplant, insufficient numbers of HSPC are mobilised due to underlying morbidity or prior treatment impacts on the HSPC pool [
1]. This deficiency can preclude HSC transplant in these poor mobilisers leaving no other effective treatment options [
2]. Development of approaches to achieve HSC expansion prior to mobilisation would address this treatment gap [
3].
HSC resides in specific locations called niches in the bone marrow (BM). BM resident macrophages are an integral component of HSC niches. Macrophage depletion in vivo is sufficient to drive mobilisation of HSC to blood [
4‐
6] and prevent successful re-establishment of the HSC niche after total body irradiation [
7]. Granulocyte colony-stimulating factor (G-CSF) was one of the first growth factors used to mobilise HSC [
8] and remains the main compound to elicit HSPC mobilisation in donors and patients [
9]. G-CSF triggers a complex array of mechanisms affecting HSC niche cellular and structural components [
10], including direct effects on BM macrophages to elicit stem cell mobilisation [
4,
11].
Colony-stimulating factor 1 (CSF1) (aka macrophage colony-stimulating factor, M-CSF) is required for the differentiation, survival and proliferation of tissue resident macrophages [
12,
13]. Although clinical trial of CSF1 as an adjunct therapy for patients receiving HSC transplantation showed some benefit [
14], it has not progressed into mainstream clinical care. Recombinant CSF1 is rapidly cleared by the kidneys and the clinical use of CSF1 required repeat high dose/continuous infusion (reviewed in [
12]), making clinical use and pre-clinical studies impractical and/or cost prohibitive. To address CSF1 therapeutic limitations, Gow et al. engineered a pig CSF1 molecule conjugated to the Fc (CH-3) region of pig immunoglobulin IgG1a that greatly increased its circulating half-life [
15]. Pig CSF1 is equally active on mouse and human macrophages [
16], and short-term daily CSF1-Fc injection was well tolerated and effectively increased murine blood monocyte and tissue resident macrophage populations, including those in BM, mice [
15], rats [
17] and pigs [
18]. AF647-labelled CSF1-Fc localised specifically to monocyte macrophages in tissues when injected into mice [
19] confirming receptor specificity. However, the improved drug qualities of CSF1-Fc are associated with supraphysiologic pharmacokinetic properties [
12,
15,
20]. This, as well as the resultant expansion of resident macrophages induced by treatment, could have complex consequences on hematopoiesis in BM and spleen that have not been investigated.
We hypothesised that CSF1-Fc treatment has direct and indirect impacts on the HSPC compartment that might be harnessed to improve HSC transplantation outcomes. Herein, we studied the temporal profile of the effects of an acute CSF1-Fc treatment regimen on resident macrophages, the HSPC pool and developing as well as mature leucocytes and erythroid cells. While the treatment initially reduced HSPC within the BM and triggered extramedullary hematopoiesis, the recovery phase following treatment was associated with an increase in total HSC that could be mobilised using subsequent G-CSF administration. We suggest that CSF1-Fc has potential utility in conditioning for more effective stem cell mobilisation.
Discussion
The current study characterised hematopoietic impacts of treatment with a modified CSF1-Fc molecule that has improved drug qualities [
15]. The expansion of monocytes and macrophages in BM and spleen following CSF1-Fc treatment was even more marked at day 7 (3 days after the last treatment) than at day 5 as studied previously [
15]. This is consistent with the prolonged half-life of CSF1-Fc. Importantly, treatment impacts were rapidly reversible, reminiscent of transient effects of either 14-day continuous [
55] or daily [
56] infusion of unmodified CSF1 in human clinical trials. At the time of peak response to CSF1-Fc, the prolonged excessive production of mature myeloid cells in BM and spleen occurred at the expense of normal hematopoiesis including disruption of BM HSC niche homeostasis. These impacts were again transient and largely resolved within a week of the peak treatment effect. We unexpectedly reveal that CSF1-Fc therapy caused a delayed increase in the total BM and spleen HSC pool, and we showed that this could be manipulated to achieve enhanced HSC mobilisation for transplantation.
Both CSF1 therapeutic potential and elucidation of CSF1-mediated in vivo biology have been hampered by the challenging practicalities of exogenous delivery of native CSF1. Continuous infusion of 150 µg/kg/day of native CSF1 in clinical trial resulted in only a transient increase in blood monocytes, peaking around day 7–8. Similarly, ongoing growth factor infusion [
55] and repeat high dose regimens did not result in cumulative effects in animal models [
57,
58]. Macrophages themselves clear CSF1 from the circulation by CSF1R-mediated endocytosis [
59]. Consequently, the combination of treatment induced expansion of tissue macrophages, added to efficient renal clearance of native CSF1, likely culminate in rapid depletion of available growth factor from circulation despite ongoing treatment. Our and previous observations [
15,
20] demonstrate that avoidance of renal clearance through use of modified CSF1-Fc is sufficient to achieve additive and sustained growth factor effects that parallel drug dose and predicted bioavailability. Further preclinical application of CSF1-Fc has potential to accelerate discovery regarding the usefulness of targeting the CSF1-CSF1R axis to modulate macrophages in clinical applications including organ regeneration [
60], chemotherapy consolidation [
55] and HSC transplantation [
14,
61]. Delivery of a hematopoietic growth factor to patients with underlying cancer, particularly hematological malignancies, would need to proceed with caution. However, at least 10 clinical trials have been conducted using high dose CSF1 therapy in cancer patients, including melanoma, refractory solid tumours, lymphoma and leukaemia. No adverse impacts relating to accelerated cancer progression were reported (reviewed in [
12]). Importantly, clinical trials using high dose CSF1 in leukaemias, including acute myeloid leukaemia, did not impact relapse rate [
14,
56].
We provide compelling evidence that HSC does not express CSF1R and consequently the HSC treatment impacts are not due to direct action of CSF1-Fc. Furthermore, a next-generation knock-in
Csf1r reporter model dependent on target translation confirmed CSF1R protein expression is restricted to the monocyte/macrophage lineage with expression initiating in Lin
−Kit
+Sca1
+CD48
+ multipotent progenitors [
62]. Mossadegh-Keller et al
. have previously suggested that HSC are directly responsive to CSF1 using a PU.1 reporter model [
47]. However, it is now appreciated that the ‘HSC’ gating strategy used in this earlier study does not achieve precision segregation of HSPC subsets, with the reported data likely reflecting CSF1 action on CSF1R
+ HPC within their gated population. Our observations that
Csf1r RNA and CSF1R protein are undetectable in HSC aligns with a recently published single cell RNA sequencing study [
48]. They are also consistent with
Spi1 knockout models which showed hematopoietic cells failed to express
Csf1r in the absence of PU.1 [
63,
64], and early studies indicating minimal proliferative effects on CFU-M in vitro [
65]. It should also be noted that at the time HSC pool expansion was noted herein (i.e. 11 days after last CSF1-Fc injection), CSF1-Fc would have cleared from circulation.
In spleen, resident macrophages can contribute to retention of HSC during extramedullary myelopoiesis [
66]. However, the observed delayed-CSF1-Fc-associated increase in splenic HSC was not accompanied by local extramedullary hematopoiesis, suggesting the lodged HSC were in a resting state. Acute CSF1-Fc treatment expanded mature F4/80
+ resident macrophages within spleen at the expense of lymphocyte, granulocyte and even monocyte frequency/retention. The expansion was skewed toward maturation of a subset of F4/80
+CD169
+ red pulp macrophages that under physiologic conditions are present at very low frequency in spleen. This unusual F4/80
+CD169
+ red pulp macrophage phenotype persisted even after splenic morphology was reinstated, and macrophage frequency had returned to normal. F4/80
+CD169
+ macrophages support HSC in BM [
5,
7]. We speculate that the phenotype shift in splenic macrophages reflects functional adaptation toward creating pseudo-BM niches for temporary maintenance of long-term repopulating HSC. This need for a temporary increase in HSC niche capacity BM may be in response to increased demand on the HSC pool, including the initial myelopoiesis and secondary lymphopoiesis demands that were triggered by CSF1-Fc treatment. Consequently, direct effects of CSF1-Fc on macrophages, combined with secondary effects on hematopoiesis homeostasis, converged to achieve the total HSC pool increase. Further studies are required to explicitly link the changed splenic macrophage phenotype with improved HSC-supportive function. Additional studies are also needed to confirm that CSF1-Fc treatment has not compromised HSC long-term repopulation integrity. Our data could be interpreted to conclude that CSF1-Fc has not compromised HSC quality. Disrupting BM HSC niche homeostasis does not automatically equate to compromise HSC quality, and our data show that the expression of c-KIT, which can be downregulated in response to stress [
52], is maintained in both the HSC that rapidly relocated to liver and the expanded HSC pool within spleen. Additionally, the repopulation capacity was not compromised in competitive transplant, nor was there any impact on lineage potential of mobilised HSC after pre-treatment with CSF1-Fc. A possible interpretation of the rapid relocation of phenotypic long-term repopulating HSC to liver during the acute response to CSF1-Fc is that it represents a proactive protective measure due to CSF1-Fc-triggered alterations in BM microenvironment homeostasis. Future studies should include investigation of these possible outcomes.
We observed few proliferating cells in spleen at day 7 post-CSF1-Fc treatment, indicating that in situ proliferation of resident splenic macrophages was not a major mechanism of local expansion. Instead, the increase in splenic macrophage number is more likely due to influx of blood monocytes that subsequently differentiates into macrophages. These newly arrived monocytes undergo specific adaptation to the CSF1-Fc-altered splenic environment, which may be contributing to the shift in splenic macrophage phenotype. Gow et al. reported increased PCNA
+ cells in spleen at day 5 post-CSF1-Fc treatment, so it is possible that the in situ macrophage proliferative response to CSF1-Fc had resolved by day 7 [
15]. However, in this previous study the proliferating spleen cells were not confirmed to be macrophages and were predominantly located in the white pulp, which was not the dominant site of splenic macrophage expansion [
15].
The potent monocytosis elicited by CSF1-Fc probably triggers known negative feedback loops directing compensatory reductions in lymphopoiesis and/or erythropoiesis [
67]. We observed a significant impairment of BM B lymphopoiesis and temporary loss of lymphocytes in spleen. The observed reduction in BM B lymphopoiesis may be a secondary impact through altered osteoblast-lineage frequency or function, as per has been previously reported for suppressed B lymphopoiesis associated with mobilising regimens of G-CSF [
68,
69]. Osteoblast function and frequency can be influenced by both osteoclasts and osteal macrophages [
70]. CSF1-Fc treatment causes rapid expansion of bone-resorbing osteoclasts [
15], and osteal macrophages are CSF1-responsive [
26,
71], and paradoxically, systemic CSF1 treatment has an anabolic impact on bone [
72]. Osteoblast-lineage cells in turn support B cell maturation [
53]. Activation of this complex cellular feedback loop was not specifically examined in this study. The CSF1-Fc-induced increase in monocyte/macrophages could also result in supraphysiologic accumulation of growth factors and cytokines that are expressed by these cells, many of which have the capacity to influence hematopoiesis. For example, excessive monopoiesis could result in elevated secretion of interleukin-1 and/or interferons, which are known to trigger HSC proliferation [
73]. Gene expression profiling of liver at day 5 post-CSF1-Fc treatment exposed increases in both pro- (
Il1, Il6 and
Tnf) and anti-inflammatory (
Il10) cytokines based on gene expression profiling [
15]. Further investigation is required to understand the secondary indirect impacts of CSF1-Fc treatment.
A clinical challenge associated with autologous HSC transplantation is collection of sufficient HSC following mobilisation to achieve the required graft cell dose for successful transplant [
1,
2]. The increase of total available HSC pool induced by CSF1-Fc treatment presented herein could address this unmet need. Enhanced mobilisation of HSC into blood of mice treated with CSF1-Fc + G-CSF therapy was accompanied by increased CFU activity in blood. Importantly increased reconstitution of all blood lineages in recipient mice was demonstrated using grafts from combination therapy versus G-CSF alone. As BM and spleen still contained HSC reserves after this mobilisation regimen, it is possible that HSC egress into blood could be further enhanced by treatment with a regimen that also included a CXCR4 antagonist [
74,
75]. Clinical trials have explored the ability of granulocyte macrophage (GM)-CSF to enhance HSPC mobilisation in combination with G-CSF, including as a sequential regimen of GM-CSF prior to G-CSF. This combination therapy provided minimal or no advantage over G-CSF alone [
76‐
79]. Therefore, expansion of the HSC pool is possibly a unique consequence of signalling through the CSF1R in myeloid lineage cells.
A remaining question is whether the long-term potential of the HSC within the CSF1-Fc + G-CSF donor graft is reduced due to CSF1-Fc exposure. Long-term serial transplant assays would be required to address this potential limitation. As discussed earlier, the collective data presented herein suggest that CSF1-Fc treatment induced 'stress' on the HSPC and committed progenitor compartment is contained and reversible. Mossadegh-Keller et al
. similarly provided evidence that CSF1 treatment did not compromise long-term hematopoietic repopulating activity [
47].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.