Novel insights into residual hematopoiesis from stem cell populations in pediatric B-acute lymphoblastic leukemia

Hematopoiesis is a complex process that ensures production of highly specialized blood cells, emerging from bone marrow (BM) every day.¹ This continuous blood cell production is finely regulated by a molecular network of intrinsic and microenvironmental signals that control a rare pool of multipotent hematopoietic stem cells (HSCs), niched in the BM, which stands at the top of the hematopoietic hierarchy organization. During malignant hematological disorders, including acute leukemias (AL), oncogenic event(s) change(s) the key regulator of pathological hematopoietic cells (more or less differentiated) and give(s) them a capacity for unlimited self-renewal, by reversing normal proliferation control, blocking cell differentiation, and promoting resistance to apoptosis.² Contrary to adult AL, 80% of pediatric AL are developed from lymphoid lineage (ALL). The uncontrolled proliferation of these malignant progenitors primarily affects B cells (80%) and supports tumor growth at the expense of normal blood cell production with an often-massive BM invasion with more than 90% of lymphoblasts.³ As a result, malignant niche outcompetes native hematopoietic stem and progenitor cells (HSPCs) niche by the mobilization of CD34⁺ cells, hence disrupting normal HSC and progenitor functions.^4,5,6 Today, the status of residual naive HSPC populations in the BM at initial diagnosis or relapse in pediatric B-ALL is still poorly investigated. In this work we show that a residual hematopoiesis is still present in many patients, even when BM is fully invaded, and presents different properties depending on the B-ALL status (initial diagnosis or relapse), with residual hematopoiesis from relapsed patients being in a better shape than residual hematopoiesis from patients at initial diagnosis.

We first sorted a large panel of medullary mononuclear cells (MMC) from pediatric B-ALL patients (n = 30), including 25 initial diagnosis and 5 relapses (Table 1). BM samples from children and adolescents without malignant hemopathies (n = 4) were used as control (Ct). CD34⁺CD38⁻ HSPCs (or CD34⁺ when no CD38⁻ fraction could be detected—indicated by + in Table 1) were sorted from MMC isolated on Ficoll-Hypaque gradient (Pancoll, PAN Biotech GmbH). As B-ALL BM are massively invaded by leukemic blasts, we needed to use a very stringent protocol of cell sorting to isolate residual HSPCs free of any blast contamination: as shown in Fig. 1a, after isolation of CD45⁺ cells, CD19⁻ and CD10⁻ cell populations were selected to eliminate leukemic blasts. Residual CD34⁺CD38⁻ HSPCs were sorted from this CD45⁺CD19⁻CD10⁻ population.

Table 1 Clinical and biological characteristics of B-ALL and control patients.

Fig. 1: Residual HSPCs sorted from relapse B-ALL BM give rise to a better hematopoietic production than residual HSPCs from B-ALL BM at initial diagnosis.

a Gating strategy for CD34⁺CD38⁻ HSPC sorting: 1. CD45^lo/neg blast exclusion by gating on CD45⁺ population, and selecting CD19⁻ and CD10⁻ cells; 2. CD34⁺CD38⁻ HSPCs sorting. b Total number of HSPCs sorted from patients BM MNCs. c Hematopoietic production of sorted residual HSPCs from primary patients BM or e after amplification in the NSG PDX model, after co-culture on MS5 stromal support during 14 days in the presence of 10 ng/ml of SCF, IL-7, IL-2, IL-3, and TPO.⁷ Hematopoietic production was assessed by the ratio of the number of total live CD45⁺ cells present in the supernatant of the culture to the input number of sorted HSPCs. d Percentage of live CD45+ cells produced by HSPCs from initial diagnosis after 14 days of cocultures, compared to CD45+ cells produced by HSPCs from relapse and controls. The asterisks indicate statistical significance based on nonparametric Kruskal–Wallis and two-sided Wilcoxon–Mann–Whitney tests calculated with GraphPad Prism 9 (GraphPad Software Inc.) *p: 0.01; **p: 0.002; ***p < 0.001. Initial diagnosis: B-ALL patient detected at initial diagnosis; Relapse: B-ALL patient diagnosed at relapse. Ct control patients, Pt patient, B-ALL B-acute lymphoblastic leukemia.

Next, we co-cultured these residual HSPCs during 14 days with murine MS-5 cells in myelo-lymphoid conditions as described by Hussein et al.⁷ Cells present in the supernatant were analyzed by flow cytometry to evaluate hematopoietic production and lympho-myeloid differentiation. Despite the fact that the number of sorted HSPCs was lower in relapse patients than patients at initial diagnosis (mean: 47 cells vs 230 cells, p: 0.01) (Fig. 1b), their hematopoietic production assessed by the ratio of the total number of live CD45⁺ cells obtained after 14 days of culture to the input number of sorted HSPCs was much better (mean 6.7 vs 0.6 for patients at initial diagnosis), and comparable to the hematopoietic production obtained with HSPCs sorted from control pediatric BM (mean 5.9) (p: 0.0001) (Fig. 1c). Interestingly, only 31.8% of CD45 cells were still alive after 14 days of cocultures with HSPCs from initial diagnosis, compared to 72.7 and 77.8% for HSPCs from relapsed and controls, respectively (p: 0.002) (Fig. 1d). No difference was observed between patients with good or poor prognosis factors (data not shown). Altogether, these results indicate that residual HSPCs present in the BM of relapsed patients are in better conditions than residual HSPCs in the BM of patients at initial diagnosis. Analysis of myeloid and lymphoid content did not show any significant difference between B-ALL (initial diagnosis or relapse) and Ct.

No HSPC could be sorted for 3 out of the 25 initial B-ALL patients (Patient #13, #18, and #22—Table 1). Taking advantage of the PDX NSG model,⁸ we were able to amplify BM cells from Patient #13 and Patient # 22 after intravenous injection of initial BM cells into irradiated (1.25 Gy) adult or intra-hepatic injection into newborn NSG mice, respectively. We also amplified cells from Patient #3 and from control Patient #33. CD19⁺ blast cells sorted from the BM or the spleen of NSG mice had the same phenotypic characteristics as initial CD19+ blast cells from patients (not shown). In addition, CD34⁺CD38⁻ cells could be sorted, which indicates that even if in a bad shape, initial medullary residual HSPCs have also been able to amplify in the NSG context. These HSPCs were tested in co-culture, like initial medullary HSPCs. As shown in Fig. 1d, a low ratio of the total number of live CD45⁺ cells obtained to the input number of sorted HSPCs residual hematopoiesis was also found for Patients #13 and #22, as expected for B-ALL at initial diagnosis. Ratio observed for PDX-HSPCs sorted from Patient #3 and control Patient #33 were similar to the ratio obtained from initial medullary HSPCs sorted from these patients.

Residual HSPCs from relapse B-ALL patients (isolated or combined medullary relapses) demonstrated properties closer to Ct HSPCs, contrasting to residual HSPCs from B-ALL at initial diagnosis. Interestingly, principal component analysis (PCA) of RNA sequencing of CD19⁺ blast cells from these patients (n = 32) and CD19⁺ cells from control pediatric BM also showed that out of the six relapse B-ALL studied, four were found in the Ct group cluster (patients #27, #28, #29 and #30—Fig. 2a). Among these four patients, two had a combined relapse (ophthalmic and BM), with less than 10% blasts in their BM (Patients #27 and #28), while two had an isolated BM relapse with 60 and 80% blasts (Patients #30 and #29, respectively). Analysis of the contrast between RNA content from these four relapsed B-ALL and from Ct demonstrated only 159 differentially expressed genes (DEG) (adjusted p value <0.05), with 82 up- and 77 down-regulated (fold-change >1.5 or <−1.5). Gene ontology (GO) enrichment analysis, performed using DAVID (database used for annotation, visualization, and integrated discovery),⁹ indicated that the main biological process (BP) down-regulated in CD19⁺ blasts from relapsed B-ALL is involved in regulation of immune response while up-regulated BP encompass positive regulation of cell cycle, and apoptotic signaling in response to DNA damage (Fig. 2b). Contrast between RNA content from B-ALL patients at initial diagnosis and control demonstrated much more DEG (4481), with 2473 up- and 1724 down-regulated, with main down-regulated BP implicated in apoptotic process and immune response, and up-regulated BP involved in epigenetic regulation of gene expression and in cell adhesion.

Fig. 2: RNA content from relapse B-ALL CD19+ blasts cells is closer to RNA content of control CD19+ cells than that of B-ALL CD19+ at initial diagnosis.

a Principal component analysis (PCA) of the top 500 most variable genes from RNA content of sorted relapse B-ALL CD19⁺ blasts cells (blue triangles, and indicated with R and a number on the label), B-ALL CD19⁺ blasts at initial diagnosis (blue triangles) and control CD19⁺ cells (red circles). b Enriched GO biological process (BP) terms for up- (left panel) and down- (right panel) regulated genes in the CD19⁺ blasts from the four relapse B-ALL patients whose RNA content clusterize with that of control CD19⁺ cells, contrasted with control CD19⁺ cells RNA. GO enrichment analysis was performed using DAVID (database used for annotation, visualization, and integrated discovery).⁹ GO BP as –Log10 (p value). c Enriched GO biological process (BP) terms for up- (left panel) and down- (right panel) regulated genes in the CD19⁺ blasts from B-ALL patients at initial diagnosis, contrasted with of control CD19⁺ cells RNA. B-ALL ID patient detected at initial diagnosis, B-ALL R patient diagnosed at relapse, Ct control patients, B-ALL B-acute lymphoblastic leukemia, GO BP Gene ontology biological pathway.

In summary, our study demonstrates for the first time that a residual hematopoiesis is still present in many patients at diagnosis of B-ALL. However, our results indicate that the few sorted residual HSPCs present a poor ability to produce hematopoietic cells, with a high mortality rate (Fig. 1c, d), which can be explained by their low intrinsic functional activity. Indeed, culture of these HSPCs in vitro in a stable microenvironment able to support normal hematopoiesis, in the presence of cytokines revealed a low hematopoietic differentiation burden. In contrast, residual hematopoiesis of relapsed patients is in a better shape, closer to the hematopoiesis of control pediatric BM, even in the context of a major BM invasion (≥60%). These results were quite surprising because relapses are sign of aggressiveness of the disease. Actually, transcriptomic analysis of CD19⁺ B-lymphoblast cells of four out of six relapse patients also indicated that their RNA content is close to that of control B-lymphoid CD19⁺ cells.

B-lymphoblastic invasion of pediatric BM results in a negative impact on the functionality of resident HSPCs and microenvironment.¹⁰ As a result, at diagnosis normal residual hematopoiesis is very weak. By killing B-lymphoblasts, chemotherapy allows BM microenvironment to resume its supportive role and the few residual HSPCs to function again during remission. Our original results indicate for the first time that when the relapse occurs residual HSPCs might have gained a better ability to self-renew and to differentiate in an adverse environment, and can behave normally, at least for a while.