Hematopoietic Stem Cell Mobilization: Current Collection Approaches, Stem Cell Heterogeneity, and a Proposed New Method for Stem Cell Transplant Conditioning

Abstract

Hematopoietic stem cells naturally traffic out of their bone marrow niches into the peripheral blood. This natural trafficking process can be enhanced with numerous pharmacologic agents – a process termed “mobilization” – and the mobilized stem cells can be collected for transplantation. We review the current state of mobilization with an update on recent clinical trials and new biologic mechanisms regulating stem cell trafficking. We propose that hematopoietic mobilization can be used to answer questions regarding hematopoietic stem cell heterogeneity, can be used for non-toxic conditioning of patients receiving stem cell transplants, and can enhance gene editing and gene therapy strategies to cure genetic diseases.

Hematopoietic stem cell transplantation

Hematopoietic stem cell (HSC) transplantation is the only curative option for many malignant and genetic diseases. HSCs are essential for the lifelong production of circulating blood cells and after transplantation are able to rebuild the blood system of cancer patients undergoing chemotherapy treatment that destroys normal HSCs along with the malignancy; cure genetic diseases that result in defective hematopoiesis; and restore normal immunity in autoimmune diseases after partial or full myeloablation. The transplantation process can be broken down into three components: collection of HSCs from the patient (autologous transplant) or from a healthy donor (allogeneic transplant), followed by conditioning that clears space in the bone marrow, and then (re)infusion of the HSCs into the patient to replace the hematopoietic pool in the bone marrow.

Currently, there are upwards of 50,000 HSC transplants per year ^[1,2]. Despite steady growth in transplant, there are still limitations in the procedure that reduce the number of patients transplanted who could conceivably benefit. These limitations include a restricted donor pool available for allogeneic transplants, conditioning-related toxicities, graft rejection, and graft-versus-host disease (GVHD). Numerous efforts by Be The Match, DKMS and others have worked to get more available donors on the registry. However, even of those who end up on the registry, around 50% ultimately decline to donate when they are discovered to be a potential match for a patient in need ^[3]. An inconvenient donation process and lack of time have been reported as barriers to retain donors in the registry, along with worries about time spent away from work, fear of the pain and needles, and worries that the process might damage their own health ^[4,5].

To increase available donors and matches for allogeneic transplant, and to facilitate the growing field of bone marrow transplant into non-malignant diseases, including those that can be treated with emerging gene therapy and gene editing strategies, a more efficient and less toxic stem cell collection method is needed. In addition, reduced or non-toxic methods of conditioning prior to transplant are necessary for expansion of HSC transplant into non-malignant diseases. In this spotlight paper, we summarize the standard-of-care of HSC collection agents and their limitations, update on several current clinical trials, and examine new pre-clinical strategies to improve HSC collection. We end this spotlight paper with a look towards future uses of HSC trafficking to answer unknown questions regarding HSC heterogeneity, facilitate gene therapy, and propose a strategy for non-toxic conditioning.

Current state of mobilization

Initially, HSCs were collected directly from the bone marrow, requiring anesthesia and aspiration from the iliac crest and other sites. The predominate method of HSC collection is now done by a process termed “mobilization”. Donors or patients receive a 5–7 day regimen of granulocyte-colony stimulating factor (G-CSF) that causes stem cells to traffic out of the bone marrow niche into the peripheral blood. Granulocyte-colony stimulating factor is still the current standard-of-care in HSC mobilization for transplantation and has been for over 20 years. Originally, G-CSF was approved by the FDA in 1991 to treat neutropenia after chemotherapy but shortly after several studies reported an increase in peripheral blood hematopoietic progenitor cells ^[6,7]. G-CSF mobilization is mediated by several mechanisms that founded the basis for discovering alternative mobilization agents ^[8]. G-CSF-mediated remodelling of osteoblasts leads to decreased production of CXCL12 – the ligand for CXCR4 – a chemokine receptor present on HSCs that mediated stem cell retention ^[9–11]. Accumulation of miRNA-126 microvesicles in the BM compartment after G-CSF administration are incorporated in endothelial and stromal cells leading to the reduction of VCAM1 – the ligand for VLA4 – an integrin present on the surface of HSCs ^[12]. An increase of the protease CD26 in endothelial cells by G-CSF reduces cell junction amongst endothelial cells and consequently increased vascular permeability and HSC egress ^[13,14]. Downregulation of c-kit ^[15] or upregulation of many other proteases ^[16] have also been described during G-CSF mobilization highlighting the pleiotropic nature of G-CSF treatment ^[17].

A 5-to-7-day regiment of G-CSF (5–10 ug/Kg/day) is able to achieve a minimum target dose of 2 million CD34⁺ cells/kg patient body weight. In comparison with bone marrow-derived grafts, donors do not have to undergo anaesthesia or bone marrow harvesting. However, the use of G-CSF has several drawbacks; healthy donors require at best 5-days of clinic visits (or must self-inject the G-CSF daily), and common side effects from G-CSF include bone pain, myalgia, nausea, fatigue and headache ^[4,18]. In rare cases, G-CSF toxicity can evolve to splenomegaly and splenic rupture ^[19,20]. In addition, for sickle-cell patients in which gene therapy or editing of HSCs could be a curative option, G-CSF toxicity is exacerbated, exemplified by the onset of vaso-occlusive crisis ^[21], preventing the use of G-CSF to acquire the high number of HSCs needed for successful gene therapy. Inconsistent mobilization response in autologous transplant for multiple myeloma (MM) and non-hodgkin’s lymphoma (NHL) have also been reported, with 22.5% not reaching the minimum target dose after two apheresis sessions ^[22], and 30% MM and 71% NHL patients failing to reach at least the optimum mobilization dose of 5 million CD34⁺ cell/Kg ^[23]. Therefore, while G-CSF has paved the way for HSC mobilization and collection for transplant, the shortcomings justify further investigation for alternative mobilization agents.

Plerixafor (AMD3100)

One early proposed G-CSF mechanism of action indicated that disruption of the CXCL12/CXCR4 axis is a key step for HSC egress from the bone marrow ^[9,24]. Desensitization of the CXCR4 receptor with a CXCL12 analog was shown to cause HSC mobilization ^[25]. Later, follow-up studies of a CXCR4 antagonist, AMD3100, originally developed as a treatment for HIV, showed that inhibition of CXCR4 leads to HSC mobilization ^[26]. FDA approval followed in 2008 for Plerixafor, with an original indication of use in combination with G-CSF ^[27,28]. Addition of AMD3100 to a regimen of G-CSF increased the number of CD34⁺ cells by 2–3 fold in MM and NHL patients when compared to G-CSF alone. However, up to 25% of NHL patients still failed to achieve the minimum CD34 dose after 4 days of apheresis using this combination regimen. An economic report estimated that each additional day of apheresis entails an additional $6,600 in medical costs ^[29] and the use of AMD3100 with G-CSF adds significantly to the total costs of collection ^[30,31]. While AMD3100 can be used as a standalone mobilization agent in healthy donors, single day collection is still not achieved in a substantial portion of patients ^[32]. Robustness of mobilization with AMD3100 alone is also highlighted in sickle-cell patients where 75% of patients require multiple collections of up to 32 days to obtain sufficient cells for gene therapy ^[33]. While repetitive dosing and collection may be possible, not only is this approach less ideal for the donor or patient, but a recent study showed that after initial exposure to AMD3100 in nonhuman primates, redosing with AMD3100 6-weeks later lead to a poor mobilization of CD34⁺ cells, indicating HSC de-sensitization to CXCR4 antagonism ^[34]. While AMD3100 has improved clinical collection of HSCs for transplant, improvements are needed for healthy, unrelated donor collection and for collection of large numbers of quality HSCs for emerging gene therapies.

Clinical investigation of new approaches

Meloxicam

Non-steroidal anti-inflammatory drugs (NSAIDs) like aspirin and ibuprofen are inexpensive with well-known tolerability and safety. NSAIDs are commonly used to treat fever, pain, and inflammation by inhibiting the cyclooxygenase enzymes 1 and 2 (COX1 and COX2) that synthesize prostaglandin E₂ (PGE₂) ^[35]. In the bone marrow, PGE₂ is mainly produced by osteoblasts, monocytes and macrophages ^[36–38] and its levels are important to maintain CXCR4 expression on murine and human HSCs ^[39]. In fact, ex vivo treatment of HSCs with 16,16 dimethyl PGE₂ (dmPGE₂) prior to transplantation into a myeloablated host enhances engraftment ^[39,40], due to upregulation of CXCR4 that increases homing to the marrow, and upregulation of the anti-apoptotic protein Survivin increasing survival of the homed HSCs in the myeloablated bone marrow environment ^[39], and possibly facilitating increased self-renewal ^[39,41]. Based on these prior findings, it was hypothesized that the opposite effect may also occur – blocking PGE₂ synthesis with NSAIDs would decrease CXCR4 expression on HSCs and result in increased mobilization. Indeed, NSAID treatment in mice, non-human primates and healthy human volunteers results in HSC egress from the bone marrow ^[42]. Furthermore, NSAID addition to a standard regimen of G-CSF leads to a 2.6-fold enhancement in long-term repopulating HSCs and a 4-day faster recovery of neutrophils and platelets when the mobilized graft is transplanted into lethally irradiated mice ^[42]. Surprisingly, despite the original hypothesis, the mechanism of action was CXCR4-independent and rather revealed a differential mobilization mechanism for stem and progenitor cells. Hematopoietic progenitor cells were expanded within the bone marrow, by removing the PGE₂ inhibitory effect on myelopoiesis ^[37,43,44]. In contrast, HSC mobilization was not due to direct effects on the HSCs but instead was caused by niche inhibition resulting in downregulation of osteopontin, CXCL12, and other retention factors. To our knowledge, this is the first agent shown to have differing mechanisms of actions for HSC mobilization versus hematopoietic progenitor mobilization.

Meloxicam is an FDA-approved NSAID and was selected as the prime candidate for HSC mobilization. Meloxicam is a dual COX1 and COX2 inhibitor, but has increased selectivity to COX2, resulting in better tolerability ^[45,46]. Meloxicam also only requires once-a-day oral dosing, decreasing the burden on donors and patients. Based on the pre-clinical data ^[42], a Phase 2 clinical trial in Switzerland compared the use of Meloxicam in combination with G-CSF and non-myelosuppressive chemotherapy with vinorelbine or gemcitabine to mobilize CD34⁺ cells in MM patients in first remission ^[47]. Meloxicam (n = 66) was well tolerated and showed similar engraftment and survival rates than the no-meloxicam group (n = 84). Consistent with prior results, the Meloxicam group significantly mobilized a higher number of CD34⁺ cells (1.5-fold; p = 0.007) within the first harvest, resulting in fewer patients requiring an extra apheresis day (9% in meloxicam group versus 19%) or AMD3100 administration (11% in meloxicam group versus 21%), reducing the overall cost and complexity for the HSC harvesting ^[47].

Two additional phase 2 clinical trials assessing the addition of meloxicam to a G-CSF mobilization regimen in MM patients have recently been completed in the US. In the Massachusetts General Hospital study (NCT02003625), there was a modest increase in the number of CD34⁺ collected in the Meloxicam group (n = 15) compared to placebo (n = 12) (30.7 vs 26 cells/ul) however only 1 patient from the meloxicam group (6.7%) vs 5 patients in the placebo control group (33.4%) required a third or fourth apheresis session to collect enough CD34⁺ cells. There was no difference in the time to engraftment of neutrophils or platelets between groups, but patients in the Meloxicam group needed less platelet (75% vs 100%) and red blood cell (25% vs 42.9%) transfusions after transplantation. In the study conducted at Indiana University School of Medicine (NCT02078102) (In this issue of Stem Cell Reviews and Reports) no placebo control was employed, but instead all MM (n = 24) patients received meloxicam in combination with G-CSF. As in the prior studies, Meloxicam was safe and well tolerated and all patients achieved successful stem cell acquisition and engraftment. However, there were no differences in the total number of CD34+ cells collected at the first apheresis in contrast to the Swiss study, nor were there differences in neutrophil and platelet recovery following transplant. In the pre-clinical mouse studies, NSAID mobilized grafts were found to have more multi-centric, CFU-GEMM colonies ^[42], perhaps indicative of more active progenitors. Intriguingly, transcriptomic analysis of the CD34+ cells mobilized with a Meloxicam + G-CSF regimen in the Indiana study had reduced expression of genes associated with oxidative phosphorylation compared to G-CSF mobilized (In this issue of Stem Cell Reviews and Reports). Self-renewing HSCs rely on glycolysis, rather than mitochondrial oxidative phosphorylation ^[48] suggesting that the quality of the CD34+ cells obtained with the addition of Meloxicam may be superior to those with G-CSF alone, despite no appreciable difference in the total number.

Taken together, the current clinical data for Meloxicam in HSC mobilization demonstrates that Meloxicam is safe, well tolerated, and does not impair collection of CD34+ cells. There may be some increases in total number and/or the quality of the CD34+ cells acquired, but the current clinical reports are mixed as to the effectiveness of increasing total CD34+ yield. Further cohorts, particularly using a randomized placebo-controlled design, should be performed to ultimately confirm the potential utility of NSAIDs in mobilization. Based on the pre-clinical data, and the current reports from Phase 2 studies, increases in total yield are likely to be modest, particularly in comparison to AMD3100.

As MM patients likely have a considerably altered inflammatory status within their bone marrow niche, specifically in relation to PGE₂ levels, it is possible that increased dosing of NSAIDs is needed to result in similar levels of mobilization enhancement as was seen in the pre-clinical models ^[42]. Considering that meloxicam is well tolerated, using a higher dose of meloxicam and/or trying different staggering regimens of meloxicam in combination with G-CSF may therefore prove more beneficial in this setting. As meloxicam is relatively inexpensive and safe, particularly compared to AMD3100, and data so far suggests that there is at least no adverse effect on mobilization, further studies exploring its use in combination with other mobilization agents, like G-CSF, are warranted.

Rapid mobilization with GROβ

Reducing the fear, pain, and ultimate lifestyle disruption of stem cell collection is likely to help improve the number of willing, available donors. While mobilized peripheral blood donation reduces the fear and risk of anesthesia and bone marrow harvest, in rare cases, serious G-CSF toxicity in the form of splenomegaly, splenic rupture, and pro-coagulant effects have been reported ^[19,20,49]. Importantly, female donors are at higher risk for adverse reactions during the procedure and are twice as likely to require extended hospitalization ^[4]. In addition to this toxicity, the multi-day regimen of G-CSF needed to achieve the required CD34⁺ cell dose for HSC transplantation is inconvenient and time consuming, one of the main discouraging factors for unrelated donors to join the bone marrow registry and to decline donation when called upon ^[5,50,51]. Donors are worried about the number of days away from work, and especially about the uncertainty of recovery following donation.

The addition of Meloxicam to a regimen of G-CSF does not adequately address the shortcomings of the multi-day regimen and the associated side effects. As previously mentioned, while AMD3100 mobilization could conceivably be accomplished in a day, the amount of HSCs acquired is still too low for many donors and patients. With the goal of developing a new mobilization regimen to overcome the shortcomings of the current standard of care, the human chemokine GROβ has been extensively characterized. While animal models showed that GROβ could act as a HSC mobilizer ^[52–54] a first-in-human study demonstrated that while GROβ was well tolerated and increased peripheral CD34+ cells, the magnitude of hematopoietic mobilization was not expected to be sufficient for clinical harvesting for transplantation ^[55]. This human study then led to the exploration of a combination with the CXCR4 antagonist AMD3100.

An extensive series of experiments have demonstrated that the combination therapy of GROβ + AMD3100 robustly mobilizes hematopoietic stem and progenitor cells, and results in a hematopoietic graft superior to the current standard of care ^[55]. These studies evaluated and compared the engraftment properties of the hematopoietic grafts rapidly mobilized after a single treatment with GROβ + AMD3100 compared to a standard multi-day regimen of G-CSF. Lethally irradiated mice transplanted with peripheral blood mononuclear cells (PBMC) from mice treated with the combination regimen had a 4-day faster recovery of neutrophils and a 6-day faster recovery of platelets compared to mice transplanted with a G-CSF graft. In competitive repopulation assays, GROβ + AMD3100 grafts were significantly more competitive than G-CSF. Surprisingly, in studies assessing transplantation of the exact same number of phenotypically-defined HSC from the peripheral blood of mice mobilized with G-CSF or those mobilized with GROβ + AMD3100, HSCs from GROβ + AMD3100 mobilized blood were twice as competitive as those from G-CSF-treated mice. Intriguingly, in transplants that resulted in 10% or less chimerism, the G-CSF sourced cells demonstrated low engraftment in 5 out of 17 mice, while no mice were below that threshold in the GROβ + AMD3100 grafts. These results demonstrate that the rapid combination regimen of GROβ + AMD3100 mobilizes a distinct, highly engraftable hematopoietic stem cell (heHSC) population that has superior competitiveness compared to G-CSF-mobilized HSCs. Not only does this regimen appear to have a superior graft and higher functioning stem cells, a feature likely to help improve emerging ex vivo gene therapies, but the rapid nature of the regimen and the elimination of G-CSF is likely to significantly remove the inconvenience and fear attributes currently associated with HSC donation.

Phase 1 clinical studies (NCT04154670 and NCT03932864) using GROβ in combination with plerixafor showed reliable, predictable, and rapid single-day mobilization and collection of CD34⁺ HSCs. The administration of GROβ was safe, as a monotherapy or in combination with plerixafor, in 79 volunteers and the majority of low level adverse events – mostly back pain – were transient (< 20 minutes). The number of functional stem cells (CD34⁺ CD90⁺) mobilized by GROβ + plerixafor was 35% of the total CD34+ pool and when tested in vivo in NSG mice they were 20 times more engraftable than G-CSF mobilized HSCs ^[56]. These promising results corroborating the mouse studies led to initiation of phase 2 clinical trials in an autologous setting for MM (NCT04552743), and in allogeneic setting for leukemias (NCT04762875). Preliminary results from the MM trial presented at the 2021 ASCO conference demonstrated 100% (15/15) of patients mobilized sufficient CD34+ cells within 2 apheresis sessions, no severe adverse reactions with only transient Grade 1 adverse events, and a greater enrichment for long-term engrafting CD34+CD90+ cells ^[57]. Further clinical exploration is needed to validate these early findings, but a stem cell collection method using GROβ + plerixafor would be considerably better for the donor and may potentially result in higher quality grafts for ex vivo gene therapy.

New areas of biology mediating mobilization

A prior review of hematopoietic mobilization ^[58] highlighted the various “branches” of biology that have been demonstrated to cause HSC mobilization. Since that time, these branches have not only grown, but new branches of biology have emerged that can be potentially utilized to improve mobilization. We highlight a few of the promising pre-clinical findings in this section but encourage the reader to look to some of the other references for a more comprehensive set.

Opening the gates for HSC trafficking

The hematopoietic stem cell niche within the bone marrow is highly vascularized, and studies in mice suggest that the most potent HSCs reside adjacent to various blood vessels ^[59–61]. For HSC trafficking into, or out of the bone marrow, these vascular “gates” may need to be opened to increase mobilization and homing efficiency. Earlier studies with G-CSF showed an increase in dipeptidylpeptidase 4/CD26 (DPP4/CD26) during HSC egress from the bone marrow ^[13,14,62]. DPP4 is an enzyme that cleaves and activates neuropeptide Y (NPY), a neurotransmitter released by endothelial cells and nerve fibers in the bone marrow ^[63]. When truncated NPY – its active form – binds its receptors NPYR2 and NPYR5 on endothelial cells it leads to a reduction of VE-cadherin and CD31 expression near endothelial cell junctions, giving rise to enhanced vascular permeability and consequently HSC mobilization ^[64]. Treatment with truncated NPY synergizes with AMD3100 to mobilize HSCs in a rapid manner, and NPY2 and 5 receptor antagonists are also able to mobilize HSCs ^[64] and are therefore attractive therapeutic options that warrant further investigation.

Another modulator of the bone marrow endothelial gateway is Viagra (sildenafil citrate), a phosphodiesterase type 5 (PDE5) inhibitor causing vasodilation that is commonly used to treat vascular disorders, preeclampsia, and erectile dysfunction ^[65,66]. Administration of Viagra on its own is not sufficient to mobilize HSCs, but in combination with AMD3100 is able to mobilize within 2 hours more HSCs than AMD3100 alone and similar numbers to G-CSF treated mice ^[67]. Transplantation assays demonstrated better engraftment than AMD3100 alone, however the G-CSF graft had a higher repopulation ability ^[67]. Like NSAIDs, Viagra has a long history of safety and tolerability data and is relatively accessible. Clinical exploration in combinatorial mobilization strategies therefore should be attainable and may help in scenarios where large numbers of CD34+ cells are needed – like ex vivo gene therapy applications. The use of Viagra to acquire stem cells for gene therapy, however, is likely disease dependent. For instance, patients with sickle cell have adverse events associated with G-CSF, and AMD3100 is currently the only available agent, but is incapable of acquiring sufficient numbers of HSCs on its own ^[21,33]. A previous trial using Viagra in sickle cell patients to treat pulmonary hypertension was stopped due to an increase in pain compared to the placebo group ^[68,69] suggesting that its use in this particular setting may be limited.

How about a spicy taco?

The peripheral nervous system ^{[70,71,80–84,72–79]}, and perhaps even the brain ^[85], are regulators of the bone marrow microenvironment and hematopoiesis. The sympathetic nervous system, in the form of norepinephrine and exercise, regulates physiological HSC trafficking ^[77] as well as G-CSF-induced osteoblast suppression and bone CXCL12 downregulation during HSC egress ^[78]. Pharmacological administration of the norepinephrine uptake inhibitor desipramine enhances G-CSF-mediated mobilization in mice, but only modestly in MM patients undergoing autologous stem cell transplantation ^[79,80]. In addition, cholinergic signals from the hypothalamus stimulate adrenal glucocorticoid hormone production which in turn acts through its receptor NR3C1 directly on hematopoietic cells to enforce HSC migration in G-CSF treated mice ^[81].

A recent study shows that spicy food may be able to cause HSC mobilization via activation of TRPV1⁺ nociceptive nerves (pain-sensing neurons) in the bone marrow ^[82]. Mechanistically, G-CSF and spicy food containing capsaicin are able to stimulate the secretion of calcitonin gene-related peptide (CGRP) from nociceptive nerves that binds to the heterodimer formed by receptor activity modifying protein 1 (RAMP1) and the calcitonin receptor-like receptor (CALCRL) present on the cell-surface of HSCs. Interestingly, CGRP binding to RAMP1/CALRL does not modify the expression of known receptors involved in HSC trafficking such CXCR4 or VLA4. Instead, gene expression analysis of Ramp1 KO HSCs indicates a role for Gα_s/adenylyl cyclase/cAMP pathway, previously known to regulate HSC mobilization ^[83]. Indeed, stimulation of adenylyl cyclase with forskolin rescues the mobilization defect of Ramp1 KO mice ^[82]. In addition to G-CSF, CGRP also enhances AMD3100 alone or G-CSF + AMD3100-mediated mobilization. To achieve the mobilization effect driven by spicy food in mice, an adult person would need to ingest the equivalent capsaicin of ~10 jalapeño peppers a day for four days ^[82]. However, novel TRPV1 agonists are currently under evaluation for the treatment of migraines, chronic pain, dysphagia, etc. ^[84,86,87] and may represent an additional tool to facilitate stem cell collection.

Medicinal marijuana – can we get more stem cells and donors?

If a plate of spicy tacos doesn’t attract new HSC donors, perhaps medicinal marijuana can help improve the collection and experience. The main active ingredient in marijuana is tetrahydrocannabinol (THC), a chemical in a broader family of cannabinoids. These molecules act on cannabinoid receptors, CB1 and CB2, that are expressed on mouse and human HSCs ^[88,89]. While plant and medicinal molecules are potent agonists of these receptors, normal signaling is facilitated by endocannabinoids. These endogenous agonists belong to the eicosanoid family, like PGE₂, and are catabolized from arachidonic acid to endocannabinoid (2-arachidonoylglycerol or 2-AG) via the fatty acid amide hydrolase (FAAH) or monoacylglycerol lipase (MAGL) ^[90]. It has been shown that in vitro treatment with the dual CB1 and CB2 agonist CP55940 ^[91] leads to downregulation of the stem cell retention factors CXCR4 and VLA4 in HSPCs ^[88], and that in vivo treatment of mice with specific or dual CB agonists rapidly mobilizes hematopoietic progenitor cells within 2 hours of administration, and was dependent specifically on CB2 agonism ^[88,92,93]. Furthermore, CP55940 synergizes with G-CSF to induce progenitor mobilization ^[88]. Intriguingly, in addition to the action on traditional CB1 and CB2 receptors, cannabinoids have also been shown to act on TRPV1 receptors ^[94,95] suggesting further avenues of crosstalk to explore.

With the increasing availability of both recreational and medicinal products capable of activating cannabinoid receptors, a comprehensive study of their effects on HSC trafficking should be done. Prior experiments in our laboratory with a variety of mobilization agents in mice has shown that the mobilization response attenuates after repeated exposures (unpublished data). If potential donors are exposed to mobilizing agents (spicy food, cannabinoids, NSAIDs, etc.) in their normal day to day lives, this repeated mobilization might result in lower-than-expected yields during the donation process. There is considerable variability in donor response to G-CSF, with some otherwise healthy donors resulting in poor yields of CD34+ cells. We hypothesize that one potential cause of poor mobilization are compensatory mechanisms in some individuals as a result of repeated mobilizations due to exposure to various agents. Detailed donor history studies, particularly regarding recreational cannabinoid use, chronic NSAID exposure, and diet, may allow for future predictors. It is also likely that many donors abruptly change their normal lifestyle leading up to HSC donation. For example, normal NSAID use may be curtailed, recreational marijuana use paused, etc. If HSC trafficking “homeostasis” was established in those donors with these agents normally in the system, abrupt withdrawal may actually put those donors in a trough for HSC trafficking, resulting in poor yields. If this is the case, encouraging those donors to not change their normal habits may allow for better yields. This hypothesis could potentially be experimentally demonstrated using animal models or controlled human studies.

The Future of Mobilization – Gene Therapy and Non-Malignant Diseases

The vast majority of HSC mobilizations and subsequent transplants are for malignant blood diseases like myelomas, lymphomas, leukemias, and others. With the currently existing FDA approved agents, and several new agents on the horizon, HSC mobilization for malignant disease is very successful. Future research efforts in collection for malignant disease likely will be around patient and donor convenience – less time, less side effects – and overall graft composition to achieve the goldilocks balance of graft versus leukemia (GVL) versus graft versus host disease (GVHD). We believe the future bulk of mobilization research in the coming decade should be to improve non-malignant transplantation, specifically with gene therapy and non-toxic conditioning. In this final section of the spotlight, we highlight three new areas of exploration for HSC mobilization.

Panning for gold

There is an ever-increasing list of clinical studies demonstrating the potential of stem cell transplant in non-malignant diseases like multiple sclerosis, diabetes, sickle cell anemia, and others. Along with the emergence of advanced gene editing and gene therapy options, stem cell transplant is poised for tremendous growth. For these therapies to be successful, HSCs need to be highly competitive and capable of robust engraftment following transplantation. Specifically for gene therapy, there are several limitations that reduce the efficacy of treatment. Firstly, the ex vivo culture conditions required for lentiviral transduction, or CRISPR editing, reduce the engraftment potential of HSCs ^[96,97]. As a result, a large number of starting CD34+ cells are needed, which is not possible to obtain from many patient groups (eg. young children). Or, the engraftment efficiency, and hence total chimerism of corrected cells, is reduced as a result of the reduced potential of the HSCs. Limitations in HSC number can be overcome in some scenarios, where there is a survival or competitive advantage to the edited gene. For example, this was seen in an ADA-SCID gene therapy trial, where discontinuation of enzyme replacement resulted in a competitive advantage for edited HSCs versus mutants ^[98].

A second related limitation, recently described in clinical translation, is the availability of viral vectors ^[99]. Because of inefficiencies in the transduction process, coupled with the reduced competitiveness in transplantation, and a lack of ability to accurately define the “engraftable cell”, large numbers of CD34+ cells are transduced, hoping to get relatively few of them to finally engraft. As a result, costs are increased, shortages of virus result, and the overall procedure is expensive and limited to just a fraction of patients. Increasing HSC competitiveness and ability to engraft will improve the efficacy of gene therapy and stem cell transplantation broadly. In addition, reliably identifying the “engraftable cell” could allow for a more focused starting cell material, reducing the amount of vector and culture media needed for HSC gene therapy/editing approaches. As described above, a new mobilization method using GROβ + AMD3100 that selectively mobilizes a highly engraftable hematopoietic stem cell (heHSC) ^[55] has been developed. This heHSC was both functionally and transcriptomically distinct from those acquired with G-CSF or to randomly collected bone marrow-resident HSCs, suggesting that it is the ideal starting population for gene therapy.

A fundamental hypothesis to this suggestion is that not all stem cells are the same and that the heHSC is unique. This hypothesis itself is not new as HSC heterogeneity has been proposed, and demonstrated, by many others. Identifying stem cell heterogeneity has previously been performed in several ways. For example, single cell transplantation assays revealed that individual HSCs have differing lineage outputs ^[100–104], with some HSCs giving “balanced” contribution to myeloid and lymphoid lineages, and other HSCs giving “biased” myeloid or lymphoid output, or others specifically going towards megakaryocyte lineages. However, these studies did not identify HSCs that were more or less competitive than others and did not have a trackable system to study the HSC before it was transplanted.

Several clonal tracking systems have also been used to demonstrate HSC heterogeneity including lentiviral barcoding ^[105,106] sleeping beauty transposase ^[107], Cre-loxP in vivo barcoding “polylox” ^[108], fluorescent protein recombination creating a “rainbow” of colors distinct to individual HSC clones ^[109], etc. These various approaches have resulted in several different variations of the “hematopoietic tree” and have challenged or supported the “clonal succession” or “clonal maintenance” models of hematopoiesis. These studies have been important advancements in the field and have increased our understanding of stem cell hierarchy and heterogeneity. However, like the single cell assays before, it is very difficult if not impossible to study the actual HSC in these scenarios – rather the majority of the research has studied the output of theoretical HSC clones. As such, many of the molecular determinants of the various HSC clones are still unknown. Similarly, while some clones were more predominate in blood production than others, identifying why one HSC clone produces more blood, and perhaps is more competitive, than another is still unknown. Finally, in some of these models it is impossible to tell what steady state hematopoiesis (i.e., the donor) looked like, versus the transplanted HSCs, as is the case with lentiviral barcoding. In those models that can determine both, the results have been mixed. For example, using the Sleeping Beauty Transposase system ^[107], the authors used their mouse model in a transplantation assay in which they determined the clonal repertoire in a mouse 64 weeks after genetic tagging, when presumably transient progenitor function should be exhausted and blood should be produced from HSCs, and transplanted the bone marrow into lethally irradiated recipients. Surprisingly, only about 5–8% of the clones that were contributing to blood production in the donor mouse were found in the recipient mice. Furthermore, when genetic tags were evaluated in HSCs, less than 5% of the HSC tags were found in mature cells. This suggests that the “engraftable HSC” is perhaps different than the steady state HSC. In another model using fluorescently labeled clones dubbed the Hue mouse, HSCs appeared to be stochastic as transplantation into multiple mice from the same donor resulted in similar clonal distribution ^[109]. However, this study was limited by relatively few clones labeled. In addition, we would argue this study and the others cited still do not answer the question of relative HSC competitiveness, which we believe is essential to addressing the critical barriers in clinical transplantation.

The age of HSCs, and assays that address age, have specifically addressed the question of competitiveness. There is an accumulation of phenotypically defined HSCs after aging ^[110–112]. Despite the increase in phenotypic number, hematopoietic output shifts towards a myeloid bias, and the repopulating ability of old HSCs is only about one-quarter as efficient as young HSCs. While most of the published work has used mouse models, a retrospective analysis by the National Marrow Donor Program, assessing donor characteristics on recipient outcome, found that age was the only donor trait significantly associated with overall and disease-free survival. Recipients who received a graft from a donor older than 45 years had significant decreases in five-year survival, demonstrating clear deficits in HSC function as a result of aging. Intriguingly, fetal liver HSCs have a robust engraftment capability compared to adult bone marrow HSCs ^[113]. In addition, virtually all of the fetal liver HSCs are actively cycling, while almost all adult bone marrow HSCs are quiescent ^[114]. Since the vast majority of transplants, and experiments, are done using adult sources of HSCs, the current paradigm is that the “true long-term hematopoietic” stem cell is quiescent. Indeed, about 90–95% of phenotypically defined murine HSCs are found in G0 ^[114–117], and human adult HSCs are estimated to only divide once every 40 weeks ^[118].

Quiescence is important. Loss of quiescence in HSCs leads to HSC exhaustion ^[119–121]. This central tenet of maintaining HSC quiescence was part of the original hypothesis of the hematopoietic stem cell “niche” put forth by Schofield ^[122]. However, we propose that contrary to the predominate focus on the quiescent HSC paradigm, the ideal HSC for transplantation is actually much closer to the fetal liver HSC. An HSC capable of entering the bone marrow space, rapidly repopulating, and self-renewing, allowing for robust engraftment and life-long blood production.

We suggest that a line of research for mobilization in the future should be to use mobilization as a means to “pan for gold”. Use the natural trafficking mechanisms as a biologic sieve to acquire HSCs with differing function. The mobilization studies we previously performed ^[55], were done in 10–12 week old adult mice. However, just using differing mobilization as a biologic sieve, we have found heHSCs that are 1.) more competitive than traditionally acquired HSCs, 2.) give long-term, multi-lineage engraftment, and 3.) in preliminary RNA sequencing studies, have a transcriptome similar to fetal liver HSCs. We believe similar approaches using mobilization regimens like GROβ + AMD3100 allows investigators to study HSC competitiveness in a manner not previously possible. First, this approach allows for study of differences in HSC function in the exact same cohort of mice – the sources of HSCs with differing function come from mice that are the same strain, same age, same sex, and harvested at the same time on the same flow cytometry run. Yet, predictably and reliably, the HSCs harvested using GROβ + AMD3100 mobilization outcompete those harvested using G-CSF or from random bone marrow HSCs as well. This approach is much closer to clinical reality, and we believe will allow others to identify unique mechanisms that govern HSC competitiveness from adult HSCs. Importantly, because the harvesting method reliably predicts functional outcome, this mobilization approach can prospectively isolate HSCs of differing function, which is not possible with the previously outlined techniques. Multiple different combinations of mobilizing agents may allow for predictable collection of differing subsets of HSCs, allowing for further sub-categorization of HSCs based on functional output. Coupled with ongoing efforts with single cell transcriptomics and proteomics, we predict that the characteristics of an “engraftable” HSC will be further defined – giving clear targets to meet for clinical HSC gene therapy products.

Playing Musical Chairs

There is another reason to improve the competitiveness of an HSC and its relative ability to engraft. For non-malignant diseases, a major barrier to implementing HSC transplantation is the associated toxicities with modern conditioning. For malignant disease, chemotherapeutic approaches and/or irradiation make sense to eliminate the underlying cause of disease. Some form of immune ablation also most likely is needed for transplant treatments for autoimmune disease. However, for diseases such as sickle cell anemia, thalassemia, SCID, fanconi anemia, Wiskott-Aldrich syndrome, metachromatic leukodystrophy, and many others, cell ablation using genotoxic approaches adds considerably to the morbidity/mortality of transplant. These toxicities preclude transplant as a front-line therapy. To circumvent this problem, transplants using various forms of “reduced intensity” conditioning, or foregoing conditioning altogether have been tried. However, in these cases, there is reduced engraftment and multi-lineage reconstitution, reducing the efficacy of the treatment due to low levels of chimerism.

Using toxic chemotherapy/radiation is clearly prohibitive for widespread use of HSC transplant for non-malignant disease. Mobilizing cells out of the niche prior to transplant may serve as an alternative to the current toxic approaches. Non-toxic conditioning using mobilization agents has been tried in the past with not much success. G-CSF on its own leads to little levels of hematopoietic engraftment but can marginally increase chimerism when coupled with low dose irradiation ^[123,124]. AMD3100 has been tried before as a standalone agent ^[125]. However, these studies only achieved ~9% chimerism, and that required 3 consecutive treatments and transplants over 3 weeks. We believe these previous attempts may not have succeeded because of several factors. G-CSF, while it mobilizes a significant number of stem cells out of the marrow and creates vascular permeability, it also causes a remodeling of the bone marrow niche with an attenuation of niche retention factors ^[17], causes a proliferation of progenitor cells in the marrow, and kinetically requires many days of treatment, making cell infusion at an “optimal” time-point impractical. AMD3100 does not mobilize many HSCs on its own, presumably leading to only small increases in niche vacancy, and does not result in increases in vascular permeability. As discussed above, we believe the “opening of the gates” may be a facilitator of hematopoietic egress into the marrow during transplantation.

We propose that if HSCs can rapidly exit the bone marrow, leaving empty niches behind, and the vascular gateway for entry was open, that this scenario could facilitate HSC engraftment without toxicity. We have dubbed this approach the “musical chairs” strategy for bone marrow conditioning. In the childhood game, there is one less chair than there are children playing. Each must compete for one of the empty chairs, or else they will be eliminated from the game. In this case, available bone marrow niche spaces may be like the chairs in the game. When HSCs are transplanted into non-conditioned mice, there are not enough niche spaces for the HSCs to “sit” into, and they are unable to compete. Rapid mobilization strategies, such as one using GRO-B and AMD3100, may be able to create a game of musical chairs, facilitating transplant.

Step 1 requires “starting the music” – endogenous HSCs need to mobilize out of their seats. Rapid mobilization, which does not alter the niche (like G-CSF does) is a key requirement for this step. In addition to GRO-B and AMD3100, it has been reported that a single injection of a small molecule inhibitor targeting α9β1/α4β1 integrins [BOP (N-(benzenesulfonyl)-L-prolyl-L-O-(1-pyrrolidinylcarbonyl)tyrosine))] rapidly mobilizes long-term, multi-lineage reconstituting HSC ^[126]. The single injection and rapid kinetics of this agent, coupled with the unique molecular target compared to GROβ + AMD3100 make BOP an attractive molecule to explore in this conditioning context. In addition, the combination of GROβ with α4β1(VLA-4) inhibitors, such as firategrast, leads to robust levels of stem and progenitor cell mobilization ^[127]. While this combination does not appear to mobilize the heHSC like GROβ + AMD3100 in preliminary studies, the addition of a VLA-4 inhibitor may allow for further bone marrow clearance that allows for increased engraftment efficiency.

Step 2 of the musical chairs strategy requires “opening the gates” to facilitate easy exit and entry into the bone marrow. Our prior studies with GROβ+AMD3100 using live imaging of mice ^[55] clearly demonstrated an increase in vascular permeability within the bone marrow space, potentially opening the gates for transplanted HSC to enter. In addition, it is clear that neuropeptide Y (NPY) in regulates vascular permeability ^[64], and increases in vascular permeability have also been reported with mechanisms involving nitric oxide synthase ^[61,128], and norepinephrine/dopamine signaling ^[78,129]. A combinatorial balance of these agents may lead to increases in HSC engraftment with reduced toxicities compared to current chemotherapy regimens.

Rapid mobilization most likely causes a transient increase in niche vacancy. However, unlike traditional toxic regimens, this niche clearance is not the result of endogenous HSC death. Therefore, in any conditioning regimen in which endogenous HSCs remain in the recipient, the transplanted HSCs must outcompete for the available niche vacancies. This is like the children competing for the available seats when the music stops. We believe there are two logical ways to overcome this critical barrier. In Step 3, of our proposed process you can “blindfold” the endogenous HSCs to make them less competitive and less able to return to the bone marrow. Non-steroidal anti-inflammatory drugs (NSAIDs) cause a reduction in the ability of HSCs to engraft ^[42] by reducing the surface expression of CXCR4 – the blindfold in this scenario. Timely administration of NSAIDs and other agents, such as anti-VLA-4 molecules like BIO-1211 ^[130] may facilitate a decreased in the ability of the endogenous HSCs to return to the marrow, leaving those seats open for the newly transplanted HSCs. Similarly, Step 4 in this process requires that the transplanted HSCs are more competitive, so they are more likely to fill the empty “seats”. These HSCs can be acquired using GROβ+AMD3100, chemically altered with PGE₂ ^[39,131,132] or CD26-inhibition ^[14] or can be altered/engineered based on future studies exploring HSC competitiveness.

As the field of research in HSC mobilization and competitiveness advances, new agents targeting new areas of biology can be used in a “plug and play” manner with this musical chairs strategy to come up with the optimal regimen to facilitate engraftment. At first glance, the multi-agent strategy that may be developed by this proposed approach can seem overly complex and impractical to translate into the clinic. However, we do not believe the proposed strategy is any more complex than currently used protocols. Take for instance the Hopkins protocol used for haplo-transplant ^[133].

In vivo gene editing and gene therapy

The focus of increased HSC competitiveness and non-toxic conditioning is important for current ex vivo approaches where the transplanted HSCs must engraft in the bone marrow and potentially compete against remaining endogenous HSCs. Another approach for a variety of genetic diseases is in vivo gene editing and gene therapy. In vivo HSC gene therapy has many advantages for treating genetic diseases because it avoids HSC collection, ex vivo manipulation and multi-day culturing that creates GMP challenges, there is no need for toxic conditioning to remove HSCs, and the cost of vector manufacturing is lower ^[134]. As HSCs are normally in specialized niches within the bone marrow, they can be difficult to target with high efficiency with various in vivo vectors or nanoparticles. However, if HSCs are mobilized into the peripheral blood they are more available to be targeted by an intravenous infusion of an adenoviral vector, or other delivery vehicle to correct the gene defect prior to their re-homing back into the bone marrow ^[134]. This mobilization strategy to increase HSC transfection has been successful in pre-clinical studies using G-CSF + AMD3100 or GROβ + AMD3100 to correct sickle cell and thalassemia in mice ^[135–137] although G-CSF is not recommended due to its toxic profile in sickle cell disease ^[21,137]. Despite this success, one issue that arises is that when few HSC are transduced it can lead to the potential loss of the corrected clones over time. One approach explored to overcome this deficit is the addition of a chemotherapy resistance cassette included in the adenoviral vector, where subsequent chemotherapy treatment selectively ablates non-transduced HSCs ^[138]. However, in diseases like Fanconi anemia, where the corrected cells would have a competitive advantage, lower levels of HSC transduction may be more than sufficient to result in disease correction ^[139,140]. The field of in vivo genetic therapies is growing, and we believe that HSC mobilization research should advance in parallel as a potential method to improve efficacy.

Final remarks

One of us (JH) has often heard from colleagues, mentors, reviewers, and audience members at the microphone after a talk, that there is no longer a reason to study hematopoietic mobilization. We disagree. In fact, we believe this next decade may ultimately reflect a golden era of HSC mobilization research. The proposals in this spotlight paper are but a few of the possible directions the field can take. With the inevitable increase in transplantation and in vivo genetic therapies, equally fervent parallel research in HSC competitiveness and trafficking mechanisms will be needed to ultimately make these promising therapies a reality for patients.

Figure 1: A new approach to non-toxic HSC transplant.

In this game of musical chairs we will (1) start the music – using rapid mobilizers to cause HSCs to vacate the niche; (2) open the gates – explore modulators of vascular permeability; (3) blindfold endogenous HSCs – identify ways to decrease competition; and (4) transplant superior HSCs – discover new mechanisms of competitiveness so HSCs can out-compete for the open niche spaces (gold medals).