Role of CXCR4 in the progression and therapy of acute leukaemia

Abstract CXCR4 is expressed on leukaemia cells and haematopoietic stem cells (HSCs), and its ligand stromal‐derived factor 1 (SDF‐1) is produced abundantly by stromal cells in the bone marrow (BM). The SDF‐1/CXCR4 axis plays important roles in homing to and retention in the protective BM microenvironment of malignant leukaemia cells and normal HSCs. CXCR4 expression is regulated by multiple mechanisms and the level of CXCR4 expression on leukaemia cells has prognostic indications in patients with acute leukaemia. CXCR4 antagonists can mobilize leukaemia cells from BM to circulation, which render them effectively eradicated by chemotherapeutic agents, small molecular inhibitors or hypomethylating agents. Therefore, such combinational therapies have been tested in clinical trials. However, new evidence emerged that drug‐resistant leukaemia cells were not affected by CXCR4 antagonists, and the migration of certain leukaemia cells to the leukaemia niche was independent of SDF‐1/CXCR4 axis. In this review, we summarize the role of CXCR4 in progression and treatment of acute leukaemia, with a focus on the potential of CXCR4 as a therapeutic target for acute leukaemia. We also discuss the potential value of using CXCR4 antagonists as chemosensitizer for conditioning regimens and immunosensitizer for graft‐vs‐leukaemia effects of allogeneic haematopoietic stem cell transplantation.

liver to the bone marrow (BM) along with the SDF-1 gradient participating in the transformation of haematopoietic sites in different stages of individual development. 11 After birth, SDF-1 secreted by stromal cells recruits HSCs into the BM niche to regulate quiescence or proliferation. 12 CXCR4 contributes to lung alveolar regeneration after pneumonectomy. 13 SDF-1 expression is upregulated after tissue injury, promoting the migration of CXCR4 + adult stem cells to injury lesions to protect or repair infarcted cardiac and ischaemic cerebral tissues. 14,15 The treatment of ischaemic diseases by mobilized tissue-committed stem cells was reviewed by Kwon et al. 16 CXCR4 is a co-receptor for the entry of HIV type 1 (HIV-1) into CD4 + T cells, which was prevented by SDF-1. 4,17 Therefore, increasing efforts have been made to develop new CXCR4 antagonists to control HIV infection (reviewed by Zhang et al). 18 CXCR4 also plays important roles in the development, invasion, angiogenesis, epithelial-mesenchymal transition and maintenance of stemness of tumour cells, [19][20][21][22][23] and targeting CXCR4 is a potential therapeutic strategy for treating malignant tumours. [24][25][26] Acute leukaemia (AL) includes acute myeloid leukaemia (AML) and acute lymphoblastic leukaemia (ALL). AML is the most common AL in adult patients, while ALL is the first and second most frequent AL in children and adults, respectively. 27,28 Except for acute promyelocytic leukaemia (APL), chemotherapy remains the backbone of treatment for other AL subtypes. In recent years, with the use of tyrosine kinase inhibitors and chimeric antigen receptor T cells in AL, patients' survival has increased to some extent. However, there is still a considerable scope for improving patients' outcomes. 27,[29][30][31] Haematopoietic stem cell transplantation (HSCT) is the regular treatment for patients with AL, including autologous HSCT (auto-HSCT) and allogeneic HSCT (allo-HSCT). The graft-vs-leukaemia (GVL) effects of allo-HSCT mediated by allogeneic T cells can effectively eradicate residual leukaemia cells. However, relapse remains a major obstacle for successful treatment. Therefore, more effective treatment methods are needed to eliminate residual leukaemia cells after allo-HSCT.

| CXCR4 participates in homing and residence of AML cells in BM
CXCR4 was critical for murine BM engraftment by human severe combined immunodeficient repopulating stem cells. 32 Human cells pretreated with CXCR4 antibodies impeded engraftment and in vitro CXCR4-dependent migration to SDF-1 of CD34 + CD38 -/low cells associated with in vivo engraftment and stem cell function. 32 CXCR4 expression influences the engraftment of autologous stem cells in patients undergoing auto-HSCT. Significantly faster haematologic recovery was found in patients who received transplanted CD34 + cells that showed high spontaneous and SDF-1-induced migration. 33 Therefore, SDF-1/CXCR4 plays a critical role in homing to and retention in the BM of normal HSCs.
Like normal HSCs, CXCR4 is also closely associated with the migration of AML cells. 34 Higher SDF-1-induced migration was observed in AML for CD34 + BM-derived cells than in paired CD34 + peripheral blood (PB)-derived cells, and a lower percentage of circulating leukaemia blasts in patients with a relatively high level of SDF-1 induced migration indicated the role of CXCR4 in the anchoring of leukaemia cells in the BM. 34 In 2004, Monaco et al evaluated the engraftment of AML cells into NOD/SCID mice. 35 Six of the 11 patient samples were engrafted successfully. Poor prognosis was observed to be inversely correlated with engraftment, and the median overall survival (OS) was 26.1 weeks for patients with cell engraftment and 95.9 weeks for those without. No correlation between CXCR4 expression and engraftment was found, and anti-CXCR4 antibody failed to block the engraftment of AML cells. 35 Concurrently, CXCR4-dependent engraftment of AML cells into NOD/SCID mice has been reported. 36 Although AML cells from some patients did not express cell surface CXCR4, intracellular CXCR4 expression was detected in all samples. Pretreatment of human AML cells with neutralizing CXCR4 antibodies blocked their homing to the BM and spleen of NOD/SCID/β2M null mice and treating mice previously engrafted with AML cells with antibodies against CXCR4 resulted in a dramatic decrease in leukaemia cell levels in a dose-and time-dependent manner. 36 Subsequently, a debate on whether engraftment of AML cells into mouse BM was dependent on SDF-1/CXCR4 between these two groups was published. 37 The opposite observations may be associated with different mice used and if newly expressed CXCR4 was inhibited. 37 Recently, a murine MLL-AF9-driven AML model was used to evaluate the engraftment of leukaemia cells into mouse BM. 38 The deletion of cxcr4 in AML cells eradicated leukaemia cells in vivo, but their homing to the BM was not impaired. Furthermore, SDF-1 is dispensable for the development of leukaemia in mice. Thus, CXCR4 signalling may play an essential role in AML stem cells, preventing differentiation independent of SDF-1. 38 Using high-resolution 2-photon and confocal intravital microscopy of mouse calvarium BM, chemoresistant MLL-AF9 AML cells were found to become less motile and unaffected by AMD3100. 39 Therefore, there may be other factors that regulate the homing and retention of AML cells within the BM. However, whether such phenomena possess leukaemiatype specificity remains unclear.

| CXCR4 expression and its regulation in AML
AML cells exposed to low oxygen partial pressure showed upregulated expression of CXCR4, and the underlying mechanisms involved alteration of lipid rafts. 40 NPM1 is one of the most common mutated genes in AML, and increased CXCR4 expression was observed when NIH3T3 cells were transfected with plasmids encoding NPM1 mutation A with enhanced migration and invasion abilities. 41 AML blasts with mutated NPM1 displayed significantly higher CXCR4 expression than those without. 42 However, no significant correlation between NPM1 mutation and CXCR4 or phosphorylated CXCR4 (pCXCR4) expression was observed in the BM specimens of untreated AML patients. 43 CEBPA mutations consist of unilateral and bilateral mutations, whereas only bilateral mutations indicate a favourable prognosis. N-terminal CEBPA mutations may impair CXCR4 expression, as only CEBPA p42 can recognize the CXCR4 promoter by chromatin immunoprecipitation assays. 44 FLT3-ITD mutation is an indicator of poor prognosis for patients with AML and associates with upregulated CXCR4 expression in a series of studies. [45][46][47] The downstream pathways may involve STAT5 and Pim-1. 45 Epigenetic regulation of CXCR4 expression by miR-146a has been reported in patients with different subtypes of AML. 48 Chemotherapy-induced upregulation of CXCR4 expression was observed in both AML cell lines and clinical samples, which may represent a mechanism of treatment-induced resistance in AML. 49 Accordingly, the expression of CXCR4 in AML is regulated by multiple mechanisms, indicating a complicated role of CXCR4.

| Relationship between CXCR4 expression and prognosis of AML
The unfavourable prognostic indication of CXCR4 expression in AML has been well documented in many studies. 42,46,47,[50][51][52][53][54] AML patients with <20% CXCR4 + /CD34 + cells had significantly superior OS and relapse-free survival (RFS) than those with ≥20%. 46 In a prospective study, patients with AML were divided into groups with low, intermediate or high levels of CXCR4 expression, as determined by CXCR4 mean fluorescence intensity ratio thresholds of <5, 5-10 and ≥10, respectively, which resulted in significantly different outcomes. 50 AML patients with normal karyotype showed higher percentages of CXCR4 + cases than those without, and high CXCR4 expression predicted poor prognoses in multivariate analysis. 52 A combination of CXCR4 and VLA-4 expression can divide AML patients into different groups with various prognoses. 53 In paediatric patients with AML, high CXCR4 expression indicated an unfavourable prognosis only in the low-risk group. 54 Taken together, CXCR4 expression levels show prognostic indications in AML and may be a potential marker for restratifying the prognosis of patients with AML.

| CXCR4 small molecular antagonist AMD3100/ AMD3465
The first generation of CXCR4 antagonist AMD3100 inhibited the migration of AML blasts induced by SDF-1 and their proliferation in vitro and reversed the enhanced engraftment of AML blasts into NOD/SCID mice mediated by SDF-1. 55 Tavor et al found that AMD3100 could significantly inhibit proliferation and induce apoptosis in multiple AML cell lines 56 and upregulate the expression of CD15 and CD11b. 56 AMD3465 is the second generation of CXCR4 antagonist that can inhibit the migration of AML cells induced by SDF-1 and multiple intracellular signalling pathways responsible for cell survival. 57 AMD3465 partially reversed the protective effects of stromal cells on leukaemia cells in vitro. AMD3465 alone or combined with granulocyte colony-stimulating factor (G-CSF) mobilize leukaemia cells from the BM and render them killed by chemotherapeutic drugs or sorafenib in leukaemic mice, leading to reduced leukaemia burden and prolonged survival. 57 In a similar study of a murine APL model, AMD3100 also reversed the drug resistance of AML cells mediated by stromal cells in vitro and reduced leukaemia burden and prolonged survival of leukaemic mice when used with chemotherapy. 58 Cocultivation of FLT3-ITD mutated AML blasts or haematopoietic progenitor cells (HPCs) on BM stromal cells resulted in a strong proliferation advantage compared with FLT3-wide-type AML blasts, and addition of AMD3100 to the co-culture significantly reduced the proliferation of FLT3-ITD mutated cells, but did not affect FLT3wide-type cells. 59 AMD3100 promoted the death of leukaemia cells with high CXCR4 expression and reduced NOG leukaemia-initiating cells but had no efficacy when AML cells did not express CXCR4. 60 This suggests that CXCR4 expression levels may be a potential marker for identifying candidates who can benefit from CXCR4 antagonists. A triple combinational therapy using AMD3100 and anti-PD-L1 plus chemotherapy was investigated in a mouse AML model. Noticeable benefits of triple combinational therapy could be achieved to eradicate leukaemia blasts that transformed into prolonged survival of mice. The frequencies of regulatory T cells (Tregs) and myeloid-derived suppressor cells in the PB of mice treated with triple combinational therapy consistently decreased. 61 Collectively, conventional chemotherapeutic drugs, kinase inhibitors or immune checkpoint inhibitors are potential strategies to be combined with CXCR4 antagonists to enhance the eradication of AML.
In 2009, the first case report of using AMD3100 in a relapsed patient with AML who underwent sibling donor allo-HSCT was reported. 62 A significant decrease in leukaemia cell levels was observed after the patient was treated with AMD3100 plus chemotherapy, and a second allo-HSCT was performed thereafter.
Complete remission (CR) was achieved one month after transplantation. Five months after allo-HSCT, the patient died of severe graft-vs-host disease (GVHD), but maintained continuous CR. 62 Three years later, the first clinical trial of combination therapy with AMD3100 and chemotherapy in 52 patients with relapsed or refractory AML was reported. 63 AMD3100 was increased to a maximum of 240 μg/kg/d without any dose-limiting toxicities. An overall CR and CR with incomplete blood count recovery (CRi) rate of 46% were achieved in 46 patients treated with AMD3100 plus chemotherapy. Furthermore, no evidence of symptomatic hyperleukocytosis or delayed haematopoietic cell recovery was found. 63 The efficacy and safety of chemotherapy combined with AMD3100 and G-CSF in the treatment of relapsed or refractory AML were evaluated by the same group. 64 No dose-limiting toxicities were observed when AMD3100 was increased to a maximum of 750 μg/kg/d. However, this clinical trial was terminated early due to unsatisfactory responses after interim analysis. 64 The POE 10-03 trial was released in 2017 by the paediatric oncology experimental therapeutics investigators' consortium. 65 Nineteen patients were enrolled, including 13 with AML, 5 with ALL and 1 with myelodysplastic syndromes (MDS). AMD3100 was administered for 5 days at four dose levels (6,9,12 and 15 mg/m 2 /dose daily) followed by high-dose cytarabine (every 12 hours) and etoposide (daily) 4 hours later. No dose-limiting toxicities were found, and febrile neutropenia and hypokalaemia were the most common grade 3 or higher non-haematologic toxicities attributable to AMD3100. Mobilization of leukaemia blasts into the PB was observed in 14 of 16 evaluable patients. All three patients achieved CR/CRi with AML. 65 In phase I/II study of AMD3100 in combination with fludarabine, idarubicin, cytarabine and G-CSF (FLAG-Ida) for the treatment of patients with early-relapsed or refractory AML, the CR/CRi rate was 50% among primary refractory and 47% among early-relapsed patients, and three patients died during induction. 66 Thus, AMD3100 plus FLAG-Ida resulted in a relatively high CR/CRi rate in adult patients with primary refractory or early-relapsed AML with acceptable toxicity. AMD3100 combined with the hypomethylating agent decitabine was used to treat newly diagnosed elderly patients with AML in phase I clinical trial (n = 69), with an overall response of 43%, and the most common side effects were myelosuppression and infection. 67

| New peptide or antibody antagonists of CXCR4
New antagonists of CXCR4 in preclinical and clinical studies are summarized in Table 1. These antagonists not only inhibit SDF-1 or stromal cell-induced chemotaxis of leukaemia cells, but also impair the proliferation or induce death of leukaemia cells directly. Thus, when used alone or in combinational therapies, CXCR4 antagonists were found to significantly inhibit the growth of leukaemia cells and prolong the survival of leukaemic mice. It is worth noting that LY2510924 and PF-06747143 have entered phase I clinical trials.
Although some of these antagonists were suggested to be more potent than AMD3100, further preclinical and clinical studies are needed to confirm it.

| Other strategies that target CXCR4
Ibrutinib, an inhibitor of Bruton's tyrosine kinase (BTK), is used to treat Waldenström's macroglobulinaemia, mantle cell lymphoma and lymphoblastic leukaemia, which also inhibits SDF-1 induced AKT and MAPK activation, leading to the inhibition of the migration and proliferation of leukaemia cells. 80

| CXCR4 is potential target for immunotherapy
The frequencies of Tregs in PB and BM of AML patients were higher than those in healthy controls. Increased CXCR4 expression robustly promoted the migration of Tregs towards BM, which played critical roles in immunosuppression of conventional T cells through proliferation inhibition, apoptosis promotion and suppression of IFNγ production. 84 Using a murine MLL-AF9 AML model, blocking CXCR4 was found to reduce Treg accumulation in the leukaemia haematopoietic microenvironment and promote anti-leukaemic effects of CD8 + T cells, and delay leukaemia progression. 85

| CXCR4 and differentiation syndrome
Differentiation syndrome is a common complication of APL.
Differentiated APL cells expressed high levels of CXCR4, and SDF-1 secreted by lung cells could help these cells migrate to lung tissues, which was reduced by pretreatment with an anti-CXCR4 antibody.
Therefore, targeting CXCR4 may provide the basis for potential prophylaxis or treatment of differentiation syndrome. 86

| CXCR4 in the pathogenesis of ALL
The precursor B-cell line Nalm-6 selectively localized within the BM stroma, which was partially controlled by the SDF-1/CXCR4 axis in vitro. 87  This difference may be due to the distribution of intracellular and cell surface CXCR4 between normal and malignant human HPCs. 95 Moreover, VLA-4 and Rho proteins are critical for B-ALL cell homing to BM, but not for normal CD34 + cells. 95 In a study using 27 clinical samples, leukaemia cells from all the patients showed SDF-1dependent proliferation, but some did not undergo chemotaxis in response to SDF-1 due to the absence of phosphorylation of p38 MAPK. 96 However, loss of the chemotactic response of ALL cells to SDF-1 did not impede their engraftment in NOD/SCID mice. 96

| CXCR4 expression and its regulation in ALL
Most studies indicated that ALL cell lines and primary ALL blasts expressed high CXCR4 levels, 92,94,99,100 while heterogeneity of CXCR4 expression in ALL was also reported in 100 paediatric patients with relapsed BCP-ALL. 101 Inactivation of Rac1 significantly prolonged the chemotactic response of ALL cells to SDF-1, and this effect was associated with an alteration of CXCR4 internalization. 102 CXCR4 expression was reduced in calcineurin-deficient T-ALL cells due to downregulation of cortactin expression, impinges CXCR4 trafficking. 93 Inhibitors of histone deacetylases extensively downregulated CXCR4 expression at both mRNA and protein levels in leukaemia cell lines and lymphoblasts from patients. 103 Accordingly, CXCR4 expression in ALL may show heterogeneity.

| Role of CXCR4 in extramedullary invasion of ALL cells
Levels of CXCR4 expression determined by flow cytometry in lymphoblasts were associated with extramedullary organ infiltration (EOI) in childhood ALL (n = 73). EOI was defined as ultrasonographically measured enlargement of the liver or spleen. The fluorescence intensity of CXCR4 in leukaemia cells was significantly higher in patients with EOI than those without. 104

| Relationship between CXCR4 expression and prognosis of patients with ALL
Expression of CXCR4 and pCXCR4 in 54 adults with newly diagnosed B-ALL, including 19 patients with Philadelphia chromosome (Ph), was analysed. CXCR4 expression levels were not related to clinical or laboratory findings or survival. However, pCXCR4 levels are associated with high leukocyte counts, serum bilirubin levels and patients' outcomes. 105 The prognostic significance of CXCR4 and VLA-4 expression was evaluated in 29 adult and 25 paediatric patients with ALL, 106 and only in adult patients, high CXCR4 expression was associated with shorter disease-free survival (DFS) and OS and low VLA-4 expression associated with shorter DFS. 106

| Small antagonists that targeting CXCR4
In precursor B ALL and stromal cell co-cultures, AMD3100 enhanced the cytotoxicity of vincristine and dexamethasone. 89 Treating murine ALL cells with Ph with low doses of dasatinib over an extended period allowed the emergence of drug-resistant cells with upregulated CXCR4 expression on their surfaces. A combination of dasatinib and a CXCR4 antagonist resulted in increased cell death, 107 indicating that this may be a promising strategy to kill ALL cells with Ph. However, it should be noted that CXCR4 antagonists may attenuate the cytotoxicity of cytarabine against ALL cells with MLL rearrangements. 94 Using human B-ALL patientderived xenograft (PDX) and murine leukaemia models, CXCR4 antagonists have reportedly mobilized ALL cells into PB. Compared with control mice, extended administration of a CXCR4 antagonist to leukaemic mice resulted in a reduction in leukaemia levels in PB and spleens and in the dissemination of ALL cells to extramedullary sites. This is the first study to present the concept of using CXCR4 antagonists to potentiate the effects of chemotherapy. 108 The mobilization responses of human ALL cells in the PDX model and haematopoietic stem/progenitor cells (HSPCs) in BALB/c mice were compared. ALL cells remained in the circulation for up to 6 hours after AMD3100 administration, when normal HPCs could not be detected. AMD3100 also increased the proportion of actively cycling ALL cells in PB. 109 The combination of CXCR4 antagonists with tyrosine kinase inhibitor, chemotherapeutic agents and FLT3 inhibitors can effectively eradicate ALL cells in multiple PDX models. [109][110][111][112] Chemotherapy-induced upregulation of CXCR4 expression led to drug resistance, which was reversed by AMD3100. 112 The POE 10-03 trial enrolled five patients with ALL, but no response was observed. 65 Consequently, whether patients with ALL can benefit from chemotherapy plus CXCR4 antagonists is unknown, and further studies are needed.

| Other treatment strategies based on CXCR4
CXCR4 may be a target for virotherapy in patients with T-ALL. A minimized derivative of HIV-1 was constructed to selectively remove leukaemia cells. 113 Thereafter, this group constructed doxycyclinedependent mini-HIV-1 variants that may improve the safety of virotherapy. 114 Ibrutinib may be a potential drug to treat B-ALL because it can inhibit the phosphorylation of CXCR4 induced by SDF-1 and the expression and activation of ERK and BCL-xL. 115 CXCR4targeted endoradiotherapy efficiently reduced leukaemia cells in the T-ALL PDX model. 116 In summary, CXCR4 plays very important roles in leukemogenesis and the biological characteristics of AL. CXCR4 is also a prognostic marker and can be used as a target for the treatment of AL.
The major progress of research on SDF-1/CXCR4 in AL is presented in Figure 1.

| Mobilization of autologous HSCs by AMD3100-contained regimens
In the preliminary phase I study for evaluating the pharmacokinetics and safety of AMD3100, all participants experienced a dose-related elevation of the leukocyte counts in PB, indicating that CXCR4 antagonists may have the ability to mobilize HSCs. 117  These results were also supported by other studies, 122-124 including a phase III prospective randomized double-blind placebo-controlled trial. 125 Additionally, an obvious increase in primitive PB progenitor cells (PBPCs) with high repopulation capacity was observed in subjects mobilized with G-CSF +AMD3100 than in those treated with G-CSF alone. 126 Compared with PBPCs mobilized by G-CSF alone, 81 genes were upregulated and 29 genes were downregulated in those treated with G-CSF +AMD3100. Increased expression was observed in the categories of apoptosis, cell cycle, replication/DNA repair, cell motility and oxygen transport, while decreased expression was found in the proapoptosis gene group and CXCR4 receptor gene itself. 127 Therefore, the addition of AMD3100 to G-CSF facilitates the collection of sufficient HSCs for transplantation. AMD3100 was shown to be utilized for mobilizing autologous HSCs in patients with multiple myeloma and lymphoma in 2008 by the FDA. Furthermore, G-CSF +AMD3100 is an alternative mobilization regimen for patients who fail to collect sufficient HSCs for auto-HSCT previously mobilized with G-CSF or G-CSF plus chemotherapy. 128,129 This will benefit more patients who have not acquired auto-HSCT when mobilized with conventional regimens.

| Healthy donor mobilization by AMD3100contained regimens
Mobilization responses of G-CSF alone and G-CSF +AMD3100 for allo-HSCT in an animal model were compared. 130 A significantly higher proportion of c-kit + Sca-1 + HSCs and plasmacytoid dendritic cells was found in grafts collected after combinational mobilization than in those mobilized by G-CSF alone. Recipient mice receiving allografts from G-CSF +AMD3100 mobilization showed higher mortality associated with increased acute GVHD clinical scores and higher pathology scores in the intestine than those that received G-CSF grafts, which may be related to the upregulation of CCR6 expression on both CD4 + and CD8 + T cells. 130 In 2011, physicians from Italy presented a case report in which G-CSF +AMD3100 was used for a normal donor of allo-HSCT who failed marrow harvest due to failure of intubation after anaesthesia. A single leukapheresis yielded sufficient HSCs for transplantation. During the 8 months' follow-up, immunosuppressants were withdrawn, and no significant appearance of GVHD was observed. 131 Patients achieved continuous CR and complete donor cell chimerism. 131 Mobilization with AMD3100 alone has also been explored. The durable engraftment of AMD3100-mobilized allogeneic PB mononuclear cells in a canine transplantation model was reported. 132 In rhesus macaques, AMD3100 could mobilize true long-term repopulating HSCs with more cells in the G1 phase of the cell cycle and those expressing CXCR4 and VLA-4 compared with CD34 + cells mobilized by G-CSF. 133 A comparison of cells mobilized by AMD3100 or G-CSF was performed in healthy volunteers. AMD3100-mobilized human mononuclear cells (MNCs) showed enhanced repopulating frequency compared with G-CSF-mobilized MNCs from paired donors, and purified CD34 + progenitors were at least as efficient as G-CSF-mobilized cells in NOD/SCID mice. 134 The immune characteristics of leukapheresis products mobilized by AMD3100 have been analysed. 135,136 In the rhesus macaque model, AMD3100 mobilization significantly increased both effector and Treg populations in PB and the resulting leukapheresis products compared with G-

CSF. CD8 + T cells (including effector memory T cells) were mobilized
to a greater extent than CD4 + cells compared with G-CSF alone. 135 Considering the high number of effector memory and Tregs in leukapheresis harvests, AMD3100 mobilization may induce less GVHD after allo-HSCT. T cells mobilized by AMD3100 had a similar phenotype, mixed lymphocyte reactivity, and Foxp3 gene expression levels in CD4 + T cells, and expression levels of 84 genes associated with Th1/Th2/Th3 pathways were not altered compared with nonmobilized T cells. However, G-CSF mobilization decreased CD62L expression on both CD4 + and CD8 + T cells and altered the expression of 16 cytokine-associated genes in CD3 + T cells. 136 In a murine GVHD model, recipient mice that received allografts mobilized by AMD3100 showed a higher incidence of skin GVHD than those receiving G-CSF mobilized allografts. 136 The impact of AMD3100 mobilization on engraftment of donor cells and GVHD was evaluated in two clinical trials. 137,138 In the first, and 19 days, respectively, in those who received MAC. 138 Therefore, mobilization with AMD3100 alone seems sufficient for allo-HSCT.

| Blocking CXCR4 in conditioning of allo-HSCT
In phase I study (n = 12), AMD3100 was used (240 μg/kg) on days −4, −4 to −3, or −4 to −2 in conditioning for second allo-HSCT in paediatric patients with refractory or relapsed leukaemia. No dose-limiting toxicity was found, and grade 1 or 2 gastrointestinal side effects were the most common adverse events. 147 In phase I/II study (n = 45), both G-CSF (10 μg/kg, days −9 to −4) and AMD3100 (0-240 μg/kg, days −7 to −4) were added to the conditioning regimen for patients with haematologic malignancies undergoing allo-HSCT. Compared with historical controls, patients in this study showed increased myeloid chimerism and lower GVHD rates, but no difference in long-term outcomes. 148 In another phase I study, 12 patients were enrolled in four sequential cohorts. Patients in the first cohort received one dose of AMD3100 (240 μg/kg) before the first dose of chemotherapeutic agents, and subsequent cohorts received injections before 2, 3, and 4 days of conditioning chemotherapy. All patients were successfully engrafted. Six patients died due to infection (n = 3), relapse (n = 2), or chronic GVHD (n = 1), and the remaining patients maintained continuous CR. 149 Thus, the addition of CXCR4 antagonists in conditioning was well-tolerated and associated with increased myeloid chimerism.
However, enhanced eradication of residual leukaemia cells or survival benefits were not observed.

| Blocking CXCR4 after allo-HSCT
Because CXCR4 is widely expressed on multiple immune cells, its antagonists will cause tissue redistribution of these cells, which may be relevant to GVHD or GVL effects. AMD3100 can redistribute leukocytes from primary immune organs to peripheral tissues or blood. 150 The impact of AMD100 on haematopoietic and immune cell reconstitution was evaluated in phase I/II clinical trial (n = 30). AMD3100 was administrated every other day from day +2 to day +21 or until neutrophil recovery. Compared with historic controls, AMD3100 treatment promoted engraftment of neutrophils and platelets, but no significant difference was observed in GVHD occurrence, longterm outcomes, secretion of inflammatory factors, or immune cell reconstitution. 151 Using immune-compromised mice grafted with human B-ALL generated from human CD34 + cells with forced MLL-AF9 overexpression, we found that injection of AMD3100 after allogeneic lymphocyte infusion could enhance GVL effects, leading to more efficient eradication of leukaemia cells within the immune-privileged site BM. 152 Thus, CXCR4 antagonists combined with donor lymphocyte infusion may be a potential treatment option for relapsed patients post-allo-HSCT.
In summary, CXCR4 antagonists can be used alone or in combination with G-CSF to mobilize HSCs. Administration of CXCR4 antagonists in conditioning or post-allo-HSCT to enhance leukaemia cell eradication by high-dose chemotherapy or GVL effects was attempted. Further studies or optimized designs may be necessary to improve outcomes. The major progress in research on SDF-1/CXCR4 in HSCT is shown in Figure 2.

F I G U R E 2
The major progress of research on SDF-1/CXCR4 in HSCT. Shown is CXCR4 blockade with AMD3100 in mobilization of HSCs, conditioning of allo-HSCT, and recipients undergoing allo-HSCT during the last three decades inhibitor and CXCR4 antagonists will improve outcomes for these patients needs to be determined. CXCR4 antagonists may mobilize drug-resistant leukaemia cells or leukaemia-initiating cells, chemotherapy or molecular target drugs may not kill them effectively, and new combinational strategies still need to be explored. Immunotherapies may be more suitable in combination with CXCR4 antagonists because they are not affected by drugresistant leukaemia cells. Post-transplantation administration of AMD3100 had no significant impact on GVL effects in clinical trials, which may be associated with the time and frequency of drug usage. AMD3100 was administered when immunosuppressants were used, and the intensity was too low. The influence of using CXCR4 antagonists in conditioning regimens or posttransplantation on outcomes of patients undergoing allo-HSCT needs to be further verified. Furthermore, donor lymphocyte infusion followed by CXCR4 antagonists may be a promising therapeutic combination for treating relapsed patients after allo-HSCT or for pre-emptive treatment in patients with high-risk diseases or positive minimal residual disease.

ACK N OWLED G EM ENTS
This work was supported by grants from National Natural Science

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N
SL wrote the manuscript. SL, HZ and Y.Y-G. collected, analysed and summarized the data. SL, HZ and Y.Y-G. conceptualized this review.
Y.Y-G. and HZ revised the review. The final manuscript was read and approved by all authors.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.