c-Kit M541L variant is related to ineffective hemopoiesis predisposing to clonal evolution in 3D in vitro biomimetic co-culture model of bone marrow niche

Abstract

Hematopoietic stem cell (HSC) maintenance in vitro is challenging because stem cell survival relies on cell-to-cell contacts and paracrine signals from bone marrow (BM) microenvironment. Indeed, HSCs easily differentiate in conventional culture systems, and in vitro study of stem cell biology, leukemogenesis, and evolutionary trajectories is limited. 3D-culture systems can mimic tissue architecture and microenvironment thus preserving HSC phenotype. In this study, we developed a calcium alginate hydrogel-based 3D co-culture system of BM mononuclear cells (BMMCs) and BM-derived mesenchymal stem cells (BM-MSCs) to study hemopoiesis in health and disease, such as biological roles of c-Kit M541L somatic mutation of unknown significance. BMMCs and peripheral blood stem cells were obtained from an acute myeloid leukemia patient who experienced graft failure and his haploidentical donor, and from a healthy donor. Cells embedded in alginate scaffolds were cultured for up to 21 days, and flow cytometry immunophenotyping was performed at baseline and every seven days. Our results showed suitability of our 3D culture system in preserving HSC vitality and phenotype throughout the culture period, and also in maintaining composition and vitality of total BMMCs. Moreover, 3D in vitro culture results suggested that M541L c-Kit somatic mutation could be a loss-of-function alteration by reducing HSC maintenance ability thus quickly promoting differentiation, as documented by in vivo graft failure and in vitro absence of long-term culture stability. In conclusions, our 3D BM-like biomimetic culture system allowed long-term stemness maintenance, making it a valid and effective tool for in vitro study of physiological and pathological hemopoiesis.

Highlights

• •HSC maintenance in vitro is challenging, and alginate-based hydrogel can be easy and suitable biomimetic scaffolds for mimicking BM niche.
• •c-Kit M541L somatic mutation could be a loss-of-function alteration by reducing stemness potential.

3D cultures; Bone marrow niche; Clonal evolution, Hematopoiesis.

1. Introduction

Acute Myeloid Leukemia (AML) is a heterogeneous group of clonal hematologic malignancies characterized by differentiation block and increased proliferation of myeloid neoplastic cells harboring various cytogenetic and molecular abnormalities [1]. Despite somatic mutations can theoretically and stochastically occur in every gene, a set of recurrent driver mutations in a small number of them seems to be linked to leukemogenesis, such as mutations in DNA methyltransferase 3 alpha (DNMT3A) or Tet methylcytosine dioxygenase 2 (TET2), with various clinical significance [2]. Within tumor bulk, a small population of leukemic stem cells (LSCs) resides and reconstitutes in close contact to the bone marrow (BM) niche thus propagating the disease [3]. LSCs can self-renewal and be quiescent as their normal hematopoietic stem cell (HSC) counterpart; however, LSCs produce more differentiated leukemic blasts with increased survival and senescence leading to accumulation in the BM and displacement of normal hemopoiesis [1, 3]. HSC and LSC homeostasis is fine-tuned not only by mesenchymal and endothelial cells, but also by the “immune niche” composed by T and B cells, plasma cells, dendritic cells, neutrophils, and macrophages [3, 4, 5]. Impairment in immune niche composition can lead to both an autologous activation of the immune system against HSCs and hematopoietic stem progenitor cells (HSPCs), such as in acquired aplastic anemia, or to immune tolerance [1, 6, 7].

Clonal evolution, defined as the selection of a particular cancer cell population from tumor heterogeneity [8, 9], is a Darwinian-like model to explain the process by which tumor cells survive throughout continuous external selection pressures, such as immune surveillance or cancer treatments [10]. Currently, the most accepted model is the branched evolution in which a common ancestor posits multiple co-existing clones diverging and evolving in parallel with the acquisition of additional somatic mutations [11, 12]. In vitro evolutionary trajectory model development for studying associations between somatic mutation occurrence and leukemogenesis is challenging because in vitro HSC maintenance and expansion is difficult, as these cells easily differentiate in conventional culture systems [13, 14, 15, 16]. Co-cultures with mesenchymal stem cells (MSCs), a component of the BM niche characterized by multi-differentiation potential and immunomodulatory activity, can support HSC maintenance in bi- (2D) and tri-dimensional (3D) conditions through secretion of soluble factors and by inter-cellular interactions [17, 18]. Therefore, culture systems resembling BM architecture can improve HSC survival and proliferation and reproduce in vitro the complex biological in vivo cross-talks between stem cells, stroma, and microenvironment to better investigate the impacts of somatic mutations in AML-related genes [19, 20].

HSC maintenance in vitro is challenging because stem cell survival relies on cell-to-cell contacts and paracrine signals from BM microenvironment. 3D-culture systems can mimic tissue architecture and microenvironment thus preserving HSC phenotype [21]. For example, sodium and calcium alginate hydrogels have already been employed to mimic hematopoietic niche in vitro and to expand CD34⁺ CD38⁻ stem and CD34⁺ CD38⁺ progenitor cells in several 3D bioengineered scaffolds also fabricated by bioplotting methods [21, 22]. Calcium alginate is a stable and biocompatible biomaterial that can be used as a 3D matrix for human stem cell culture, because polysaccharide chains can be easily polymerized to form a solid matrix in the presence of calcium by ionic gelation resulting in pre-mixed cell suspension immobilization [23, 24]. Another important advantage of alginate-based scaffold is that crosslinking reaction in presence of divalent ions, like calcium chloride, to form polysaccharide chains can be easily reverted by calcium chelation resulting in alginate hydrogel dissolution and full cell recovery [23, 24, 25]. Moreover, hydrogel dissolution does not require enzymatic reactions, differently from collagen or fibrin hydrogels, not affecting cell viability, gene expression, and surface marker integrity [26].

In this study, we developed a novel ex vivo 3D co-culture system for mimicking BM niche using total BM mononuclear cells (BMMCs) depleted of MSCs co-cultured with BM-derived MSCs. This system was then used for establishing a long-term co-culture model in health and disease and for investigating biological impact on hemopoiesis of clones harboring somatic mutations of unknown significance, such as M541L c-Kit variant, in a BM-like microenvironment.

2. Methods

2.1. Patients

In this study, 3D co-culture systems were assembled using calcium alginate as scaffold, BM-MSCs from healthy subject, and MSC-deprived BMMCs or peripheral blood stem cells (PBSCs) obtained from an AML patient, its haploidentical transplant BM donor, or a healthy donor. Standard 2D cultures were used as a control over a culture period of 21 days, and cell viability and hemopoiesis was monitored by live&dead assay and by flow cytometry, respectively. In details, BM-MSCs were from two healthy subjects (two males aged 30 and 48 years), BMMCs were from an AML patient (male, 51 years old) and a healthy BM donor (male, 22 years old), and PBSCs were from a haploidentical donor (female, 46 years old). BM or post-apheresis peripheral blood (PB) was collected for establishment of the 3D BM niche after informed consent obtained in accordance with Declaration of Helsinki [27] and protocols approved by local Ethic Committee “Campania Sud”, Brusciano, Naples, Italy (prot./SCCE n. 24988). Two subjects were healthy HSC donors: in one subject (named haploidentical donor), blood was collected after CD34⁺ mobilization protocol and apheresis as per international guidelines; in the other donor (named healthy), blood was collected after BM blood collection. Cells were stored in liquid nitrogen until HSC transplantation (HSCT) procedure. After infusion, residual blood was used for research. The third subject was diagnosed with AML based on 2016 World Health Organization (WHO) criteria [28] and received chemotherapy as per international protocols [29] at the Hematology and Transplant Center, University Hospital “San Giovanni di Dio e Ruggi d’Aragona”, Salerno, Italy. BM blood of this patient was collected at the time of graft failure after haploidentical HSCT. BMMCs were isolated by Ficoll-Paque density gradient centrifugation (Cytiva™), and cells were stored in liquid nitrogen until use. After thawing, cells were allowed to restore for 24h in MyeloCult™ H5100 (STEMCELL Technologies) supplemented with 100nM dexamethasone (Sigma-Aldrich). After recovery, cells in suspension were washed with phosphate buffer saline (PBS; Gibco, Thermo Fisher Scientific, Waltham, MA, US), and used for further experiments, while attached cells were removed.

2.2. Flow cytometry

Immunophenotyping was conducted on whole fresh PB or BM blood. Briefly, 50 μL of sample were stained with antibodies according to manufacturers' instructions and as previously described [30]. For co-cultured cells, first, alginate scaffolds were disrupted in the presence of 50mM EDTA, buffered in 1000mM HEPES, calcium was withdrawn. Cells were released, fixed in 4% PFA for 20 min, and then stained with CD90, CD105, CD3, CD117, CD11b, CD14, CD34, CD33, CD16, CD45, HLA-DR, CD73. Immunophenotyping was performed at day 0, 7, 14 and 21. Manufacturer's characteristics of antibodies are displayed in Supplementary Table 1.

Samples were acquired on a DxFlex cytometer (Beckman Coulter), equipped with violet (405 nm), blue (488 nm), and red (638 nm) lasers. Instrument daily quality control, external quality controls, and compensation were performed as previously [30]. Samples were run using the same PMT voltages, and at least 50,000 events were recorded, and post-acquisition analysis was performed with Kaluza software (v2.1; Beckman Coulter). Leukemic cell identification and myelogram were performed as previously described [26]. For 3D culture immunophenotyping, cells were first identified based on side scattered area (SSC-A) and CD45 expression, and then myeloid cells were discriminated from mesenchymal stem cells (MSCs) based on HLA-DR and CD73 expression. On CD73⁻ cells, CD33 and CD34 expression was investigated, and CD33⁺ and/or CD34⁺ precursors were further studied for CD117 and CD11b expression [31]. On CD33⁻CD34⁻ cells, CD3 expression for T lymphocytes and CD14 levels for monocytes were further investigated. On CD73⁻ population, a radar plot was made showing CD117, HLA-DR, CD33, CD11b, and CD16 expression simultaneously; while on CD73⁺ cells, a radar plot was designed using CD73, CD34, CD117, HLA-DR, CD33, and CD11b expression.

2.3. 2D static culture

For 2D culture, PBSCs were seeded at a final density of 3.75 × 10⁶ cells/mL in MyeloCult™ H5100 (STEMCELL Technologies) supplemented with 100 nM dexamethasone (Sigma-Aldrich, St. Louis, MO, US) filtered using a 0.22 μm filter (Millipore, Burlington, MA, US). Cells were incubated for 24h at 37 °C and 5% CO₂ to recover before starting experiments. After 24h incubation, culture medium was changed, cells were cultured up to 21 days, and medium was changed twice per week. Samples were harvested after 2h post the first culture medium change (time point [T] 0), and then every seven days (T7, T14, and T21). As a control, right after thawing, a total of 2 × 10⁵ PBSCs were directly fixed and stained as described above for flow cytometry immunophenotyping.

2.4. 3D culture with calcium alginate matrix and cell recovery

Sodium alginate was dissolved in a solution of 0.16 M Hepes and 1 M NaCl (all from Sigma-Aldrich, Milan, Italy) at room temperature (RT) under gentle stirring overnight, to obtain a 4% (w/v) polymer solution. Alginate was then combined to an equal volume of cell suspension containing 1 × 10⁶ cells, and this cell-alginate mix solution was extruded in a cross-linking solution of 0.1 M CaCl₂ (Sigma-Aldrich) through an 18 Gauge needle to produce alginate beads with a mean diameter of 3 mm. Subsequently, beads were washed in 1X PBS (Corning, NY, US) for 10 min and cultured in MyeloCult™ H5100 (STEMCELL Technologies) supplemented with 100nM dexamethasone (Sigma-Aldrich), and cultured at 37 °C in an atmosphere of 5% CO₂ and 95% relative humidity up to 21 days (Figure 1). At each time point, beads were washed with 1X PBS (Corning). and hydrogel was disassembled with a 50 mM EDTA solution (cat.no. 15575020; Thermo Fisher Scientific) and 10 mM Hepes (cat.no. 15630080; Thermo Fisher Scientific) by pipetting for 10 min at RT. Fisher Scientific) and 10 mM Hepes (cat.no. 15630080; Thermo Fisher Scientific) for 10 min at RT. Cells were pelleted by centrifugation at 400g for 5 min for further characterization.

Figure 1.

Schematic representation of 3D co-culture system assembling. First, bone marrow mononuclear cells (BMMCs) or peripheral blood mononuclear cells (PBSC)s were collected from donors and isolated by Ficoll-Paque density gradient centrifugation, and then cells were seeded for 24h in MyeloCult™ H5100 supplemented with 100nM dexamethasone. After recovery, attached cells were removed, while cells in suspension were washed with phosphate buffer saline, and mixed in equal volume with a sodium alginate solution of 0.16 M Hepes and 1M NaCl and bone marrow-derived mesenchymal stem cells (BM-MSCs) obtained from a healthy donor. The cell-alginate mix solution was then extruded in a cross-linking solution of 0.1 M CaCl₂ through an 18 Gauge needle to produce alginate beads with a mean diameter of 3 mm. Parts of the figure were drawn by using pictures from Servier Medical Art. Servier Medical Art by Servier is licensed under a Creative Commons Attribution 3.0 Unported License (https://creativecommons.org/licenses/by/3.0/).

2.5. Live&Dead assay

Cell viability was detected by fluorescence live&dead assay at day 0 (right after bead preparation), 7, 14, and 21. Live cells were stained with calcein AM solution (cat.no. C1359; Sigma-Aldrich), while dead cells with cell membrane-impermeable Ethidium homodimer I solution (cat.no. E1903; Sigma-Aldrich). Cells were incubated for 30 min at RT and then washed in 1X PBS (Corning). Images were acquired at 4× magnification by a fluorescence microscope (Eclipse Ti Nikon Corporation, Tokyo, Japan). Images were acquired using the same settings (e.g., light intensity, exposure time, and gains). Signal intensity quantification was performed using ImageJ software (rel.1.52p National Institutes of Health, Bethesda, MD, USA), as previously described [32].

2.6. Statistical analysis

Data were analyzed using Prism (v.9.3.1; GraphPad software, La Jolla, CA, US). For flow cytometry data, populations were reported as percentage of positive cells, and for CD117 expression, median fluorescence intensity (MFI) values were calculated on total CD73⁻ cells. For live&dead data, images were converted from RGB to 8-bit gray scale format, tagged areas were expressed as pixel intensity mean values, and results were expressed as percentage of live and dead cells. The number of replicates for each analysis was two. Multiple comparisons for percentage of positive cells, MFI values, or live/dead ratios between each sample and timepoint were performed using one-way analysis of variance (ANOVA) fitting a main effects only model and uncorrected Fisher's LSD. Statistically significant differences were defined as: ∗ = P < 0.05, ∗∗ = P < 0.01, ∗∗∗ = P < 0 .001, and ∗∗∗∗ = P < 0 .0001.

3. Results

3.1. 3D in vitro culture system mimicking BM niche

To reproduce in vitro BM niche, a 3D culture system was assembled using calcium alginate as a synthetic extracellular matrix for BMMCs and BM-derived MSCs. First, cell viability was determined by Live&Dead assay in the 3D culture to assess the applicability of our system for long-term in vitro culture of HSCs. For healthy donor and AML patient, cells remained alive for the whole length of culture (up to 21 days) (Figures 2A and B, and Figure 3A). Conversely, the number of live cells from the haploidentical donor was low with a high rate of mortality, already at baseline and after 14 days of culture (P = 0.0436, healthy vs haploidentical donor) (Figure 3B).

Figure 2.

3D in vitro culture cell viability in healthy. Live & Dead assay of 3D in vitro co-cultures using bone marrow mononuclear cells (BMMCs) from a healthy donor (A) or peripheral blood stem cells (PBSCs) after mobilization from the haploidentical donor (B) and bone marrow-derived mesenchymal stem cells (BM-MSCs) from a healthy subject in an alginate-based hydrogel scaffold was performed at day 0, 7, 14, and 21. Viable cells are in green, dead cells in red. Scale bar: 500 μm (A) and 200 μm (B).

Figure 3.

3D in vitro culture cell viability in disease. (A) Live & Dead assay of 3D in vitro co-cultures using bone marrow mononuclear cells (BMMCs) obtained from an acute myeloid leukemia (AML) patient after graft failure and bone marrow-derived mesenchymal stem cells (BM-MSCs) from a healthy subject in an alginate-based hydrogel scaffold was performed at day 0, 7, 14, and 21. Viable cells are in green, dead cells in red. Scale bar: 500 μm. (B) Signal intensity was quantified using ImageJ software, and converted in percentage of live and dead cells for each subject and time point. N. of replicates = 2.

BMMCs cultured in 2D static conditions failed to reach 21 days. At T0, cells were mostly constituted by CD34⁺ HSCs and more mature progenitors (CD117⁺CD11b⁺), while at T14, cells were mostly represented by terminally differentiated cells and only a small proportion of CD34⁺ cells (Figure 4A). In 3D culture conditions, BMMCs alone showed a normal distribution between immature and mature cells at T0. At T14, the small proportion of CD34⁺ cells was described in 3D culture; however, terminally differentiated cells were lower than that observed in 2D culture conditions (Figure 4B). At T21, the CD34⁺ cell population showed a more differentiated profile with the same distribution of myeloid progenitors as described at T14. No significant variations were documented in the CD73⁺ population in both conditions (Supplementary Figure 1).

Figure 4.

Hematopoiesis evolution in 2D and 3D culture systems. Peripheral blood stem cells (PBSCs) were cultured in standard monolayer cultures (A) and in 3D alginate-based hydrogel system (B). Cells were harvested ad day 0 (T0), 14 (T14), and 21 (T21), and flow cytometry immunophenotyping was performed. On CD73⁻ cells, radar plots display cell distribution based on CD117, HLA-DR, CD33, CD11b, and CD16 expression to identify stem to differentiated cell evolution over culture time. Monocytes, lymphocytes, myeloid progenitors (CD117⁺CD11b⁺, CD117⁺CD11b^+/−, and CD117⁻CD11b⁻ cells), CD34⁺ hematopoietic stem cells, and CD34-CD33- (not myeloid) cells are shown in different colors to monitor surface marker expression variations throughout the culture period.

3.2. 3D in vitro culture system as a model of BM niche in health and disease

Next, based on good cell viability and conserved cell composition up to 21 days of culture, we sought to implement this system by adding BM-MSCs and to test this in vitro model in health and disease. Therefore, a case of AML patient who underwent haploidentical HSCT and experienced a graft failure was chosen as a source of pathological BMMCs, while the haploidentical PBSC donor and another BMMC donor were chosen as a source of healthy HSCs.

3.3. Case 1: AML patient

A 51-year-old male with anemia and neutropenia received a diagnosis of AML FLT3-ITD mutated, NPM1 wild type and normal karyotype in May 2021 in another Institution and received daunorubicin and cytarabine following a 3+7 scheme plus midostaurin as first-line induction treatment achieving a complete remission (CR) in June 2021. Consolidation therapy with high dose cytarabine and midostaurin was initiated. In August 2021, the patient arrived at our Institution for hematopoietic stem cell transplantation (HSCT) evaluation; however, increased liver enzymes and total bilirubin were observed, and a diagnosis of acute hepatitis C virus (HCV) infection was made (HCV copy number, 10,400 copies/mL); therefore, antiviral therapy was started and complete negativization was obtained in September 2021. In October 2021, the patient was re-evaluated showing a CR by light microscopy, and flow cytometry documented the absence of CD45⁺CD34⁻CD117⁻CD64⁺⁺CD33⁺CD16⁻CD56⁻minimal residual disease (MRD). Hematopoietic CD34⁺ stem cells (0.5% of total nucleated cells) were represented by myeloid CD33⁺CD117⁺DR⁺ (89%) and lymphoid CD19⁺DR⁺ (10%) precursors, while CD117⁺CD34⁻ progenitor cells (2.5% of total nucleated cells) were composed by CD33⁺DR^dim myeloid (1.4%) and CD36⁺CD33⁺DR^dim erythroid (1.1%) progenitors. CD19⁺CD34⁻CD45⁺⁻ hematogones were also present (1.5%). Neutrophils were mostly represented by CD16^dimCD11b⁺ intermediated (53%) and CD16⁺CD11b⁺ mature (43%) forms. Next-generation sequencing analysis found a pathogenetic missense Arg882His (2645 G > A; exon 23) somatic mutation in DNMT3A (31.8%), two missense mutations in SETBP1 of uncertain significance (Val231Leu, 691 G>C, 51.1%; and Val1101Ile, 3301 G>A, 47.9%; both on exon 4), and one missense somatic mutation in TET2, exon 11 (Ile1762Val, 5284 A>G, 48.6%). Because the patient was still in CR, he underwent haploidentical HSCT after myeloablative conditioning regimen with busulfan (3.2 mg/kg/day) and fludarabine (40 mg/m²/day on day -6 and -3) and selecting the HLA-haploidentical 52-year-old sister as donor (serology positive for cytomegalovirus (CMV) for donor and recipient, as well as blood group 0 + for both subjects). Graft versus Host Disease (GvHD) prophylaxis included: cyclosporine and mycophenolate mofetil, started on day 0 and +1, respectively; and cyclophosphamide infused only on days +3 and +4. No post-transplant complications or signs of acute GvHD were observed; however, at +22 days post HSCT, a partial donor chimerism was observed (62% donor, 38% recipient residual DNA), and on day +39, no residual donor DNA was observed in the peripheral blood of the recipient thus a diagnosis of graft failure was made. Therefore, supportive therapies were started with red blood cell and platelet transfusion, and growth factor stimulation (granulocyte-colony stimulating factor [G-CSF], erythropoietin, and eltrombopag switched to romiplostim for lack of response). Interestingly, the patient completely recovered using only growth factors, and he is still in complete remission (CR) at the time of writing.

At graft failure, BM aspirate was performed. No leukemic cells were observed by light microscopy, and flow cytometry showed the absence of CD45⁺CD34⁻CD117⁻CD64⁺⁺CD33⁺CD16⁻CD56⁻ MRD (Supplementary Figure 2). Hematopoietic CD34⁺ stem cells (0.4% of total nucleated cells) were represented by myeloid CD33⁺CD117⁺DR⁺ (90%) and lymphoid CD19⁺DR⁺ (8%) precursors, while CD117⁺CD34⁻ progenitor cells (4.8% of total nucleated cells) were composed by CD33⁺DR^dim myeloid (3.6%) and CD36⁺CD33⁺DR^dim erythroid (1.2%) progenitors. CD19⁺CD34⁻CD45⁺⁻ hematogones were also present (0.02%). Neutrophils were mostly represented by CD16^dimCD11b⁺ intermediated and CD16⁺CD11b⁺ mature forms. Next-generation sequencing performed at this time showed increased variant allele frequency (VAF) for the pathogenetic somatic mutation in DNMT3A (44.3%), while the two missense mutations in SETBP1 and the one in TET2 remained stable (51.6%, 48%, and 51.9%, respectively). BM blood obtained at the time of graft failure was employed for in vitro 3D co-cultures. The patient is still in CR at the time of writing (Supplementary Figure 3, MRD evaluation by flow cytometry after 3 months post-HSCT).

3.4. Case 2: haploidentical HSCT donor

A 52-year-old HLA-haploidentical sister was selected as donor (CMV positive and blood group 0+), and PBSCs were obtained after G-CSF administration and stored in liquid nitrogen until HSCT. After infusion, leftovers were used for in vitro 3D dynamic co-cultures. Next-generation sequencing and flow cytometry immunophenotyping were also performed in the donor using PB. Surprisingly, the donor had two of the three missense mutations of uncertain significance found in the recipient: the Val231Leu SETBP1 (691 G>C on exon 4, 47.5%), and the one on TET2, exon 11 (Ile1762Val, 5284 A>G, 48.5%). Moreover, the donor had a missense somatic mutation of uncertain significance on c-Kit (Met541Leu, 1621 A>C) with a 99.1% VAF. Flow cytometry immunophenotyping did not show any abnormalities and circulating CD34⁺CD117⁺DR⁺CD33⁺ cells were 0.02% of total nucleated cells (Supplementary Figure 4).

3.5. 3D in vitro skewing of hemopoiesis might reflect in vivo events

To investigate physiological and pathological hemopoiesis in our 3D system, BMMCs or PBSCs were co-cultured with BM-MSCs in an alginate scaffold up to 21 days. Cells were harvested at 0, 7, 14, and 21 days, and myeloid progenitor cell distribution variations were monitored by flow cytometry immunophenotyping (Figure 5). At the time of seeding, post-graft failure AML patient showed a hemopoiesis mostly composed by differentiated cells, and no variations were described throughout the culture period. Interestingly, the two donors showed completely different behaviors (Figure 5). At T0, the haploidentical HSC donor displayed an even distribution between immature and mature progenitors, and terminally differentiated cells. At T14, the small CD34⁺ population described in 2D static culture conditions and in 3D culture without BM-MSCs was also present. At T21, an equal distribution of progenitors and differentiated cells was described in the presence of mesenchymal cells in contrast with that observed in 3D culture conditions without BM-MSCs at the same time point. Conversely, at T0, cells obtained from the healthy donor showed a comparable distribution of that observed in AML patient at baseline, while at T7, the hematopoietic profile was more similar to that documented in the haploidentical HSC donor at the time of seeding. At T14 and T21, cell distribution was like that shown at baseline. No significant modifications were described for CD73⁺ population distribution (Supplementary Figure 5). In details, CD117⁻CD11b⁻ and CD117⁺CD11b^+/− progenitors were significantly reduced at day 14 compared to baseline for all conditions, while no variations were described for CD34⁺ stem cells, mature CD117⁺CD11b⁺ myeloid progenitors, and terminally differentiated cells (Figure 6A). Interestingly, CD117 expression was significantly lower in the haploidentical HSC donor throughout the culture period compared to both AML patient and healthy donor, while no significant changes were described between patient and healthy subjects (Figure 6B). Moreover, CD34⁻CD33⁻ cells were the predominant population in AML patient and in both HSC donors; however, this subset represented almost half of total cells in the haploidentical HSC donor, and immature progenitors tended to reduce over culture time (Figure 6C).

Figure 5.

Hematopoiesis evolution in 3D culture system in health and disease. Bone marrow mononucleated cells (BMMCs) depleted of mesenchymal stem cells (MSCs) from an acute myeloid leukemia (AML) patient after graft failure and from a healthy donor or peripheral blood stem cells (PBSCs) from an haploidentical donor were co-cultured with bone marrow-derived MSCs (BM-MSCs) in 3D alginate-based hydrogel system. Cells were harvested ad day 0 (T0), 14 (T14), and 21 (T21), and flow cytometry immunophenotyping was performed. On CD73⁻ cells, radar plots display cell distribution based on CD117, HLA-DR, CD33, CD11b, and CD16 expression to identify stem to differentiated cell evolution over culture time for AML patient (A), haploidentical (B), and healthy donor (C). Monocytes, lymphocytes, myeloid progenitors (CD117⁺CD11b⁺, CD117⁺CD11b^+/−, and CD117⁻CD11b⁻ cells), CD34⁺ hematopoietic stem cells, and CD34⁻CD33⁻ (not myeloid) cells are shown in different colors to monitor surface marker expression variations throughout the culture period.

Figure 6.

Hematopoietic stem and progenitor cell distribution and variation in 2D and 3D cultures. (A) On CD73⁻ cells, CD34⁺ hematopoietic stem cell, progenitor (CD117⁻CD11b⁻, CD117⁺CD11b⁻, and CD117⁺CD11b⁺ cells), and differentiated cell (CD33⁻CD34⁻) frequencies were calculated for each time point (day 0, 7, 14, and 21), and compared between cells cultured in 2D standard monolayers, 3D in vitro cultures using only peripheral blood stem cells (PBSCs), or hematopoietic cells co-seeded with bone-marrow-derived mesenchymal stem cells (BM-MSCs) from an acute myeloid leukemia (AML) patient after graft failure, his haploidentical donor, and a healthy subject. (B) On CD73⁻ cells, median fluorescence intensity (MFI) values of CD117 (c-Kit) were calculated for each sample and time point to investigate expression variations along the culture period and between healthy subjects and AML patient. (C) Frequency population perturbations were also plotted for each subject alone to display patient-specific modifications. ∗P < 0.05; ∗∗, P < 0.01. N. of replicates = 2.

5. Discussion

In this study, we developed a 3D co-culture system with BMMCs and BM-MSCs seeded in an alginate scaffold hydrogel selected as an in vitro model of hematopoiesis in health and disease used to explore clinical significance of somatic mutations occurring in AML-related genes. Our results showed suitability of our novel 3D in vitro model to mimic the hematopoietic niche by mixing alginate hydrogel, human BM-MSCs from a healthy subject, and human total BM cells (comprised of HSCs, progenitors, and mature cells) deprived of MSCs. Moreover, our 3D bioengineered scaffold was cultured for up to 21 days, longer than standard 2D culture periods. Calcium alginate 3D scaffold successfully mimicked marrow stroma by preserving HSC compartment and differentiation abilities. We also showed the applicability of our system to successfully recover cultured BMMCs for further characterizations, such as immunophenotyping.

In vitro hematopoietic models are still a chimera because HSC maintenance and differentiation are complex processes that require not only an appropriate culture medium composition, but also fine-tuned cross-talks between stem and stromal cells through cell-to-cell contacts and cytokine release [19, 33]. 3D cultures are frequently used in tissue engineering as already described in regeneration models [26, 34], and 3D hydrogel co-cultures are of increasing interest because of the ability to combine multiple cell populations and to mimic physiological conditions, for reproducing stem cell niche environments, such as satellite cell niche in muscle cell regeneration [18, 26, 35].

3D hydrogel co-cultures are frequently used in tissue engineering for reproducing stem cell niche environments, such as satellite cell niche in muscle cell regeneration [18, 26, 35]. Hydrogels are also a suitable biomaterial for in vitro studies, as they can be easily degraded, and cells can be harvested and used for further experiments [19, 20, 21, 22, 23, 24, 25, 26, 34, 36, 37, 38]. Moreover, the three-dimensionality of hydrogels creates a hypoxic gradient, naturally present in the marrow, thus better mimicking the BM niche [18]. In 3D cultures, MSCs express HIF1α and VEGF, that favor the establishment of a hypoxic gradient contributing to HSC quiescence, metabolism, and survival [39]. For example, umbilical cord-derived MSCs cultured in an alginate-gelatin bio-printed hydrogel scaffold show increased long-term viability and stemness maintenance compared to 2D monolayer cultures [40, 41]. 3D co-cultures induce proliferation of CD34⁺ stem cells and increased expression of homing markers, such as C-X-C chemokine receptor type 4 (CXCR-4), the alpha-chemokine receptor for stromal-derived-factor-1 involved in HSC homing to the bone marrow and quiescence [21, 42]. Indeed, CD34⁺ cells cultured in a 3D hydrogel system express higher levels of stemness markers and anchoring molecules, such as CD133, CXCR4 and N-Cadherin, with increased engraftment potential compared to those cells cultured in 2D standard conditions [18]. Moreover, in other models, alginate hydrogels have been mixed with umbilical cord-derived HSCs or mouse long-term stem cells, or human mobilized CD34⁺ HSCs have been seeded in polydimethylsiloxane [19, 21, 22, 36]. In our study, we developed a 3D co-culture system using BMMCs depleted of MSCs or PBSCs and BM-MSCs immersed in an alginate-based hydrogel as a scaffold to reproduce in vitro architecture and composition of the BM niche. Total BMMCs or PBSCs instead of single cell populations were employed to prove the feasibility of our system in mimicking the complex environment in which hematopoiesis happens, and in maintaining HSC compartment and differentiation over a long-term culture period (up to 21 days). Our 3D in vitro model assured high cell viability throughout the co-culture for AML-derived BMMCs and healthy PBSCs. Of note, for the haploidentical donor of the AML patient who experienced graft failure, the number of live cells was low with a high rate of mortality, already after 14 days of culture. Our in vitro observations reflected clinical outcomes documented in vivo, as the AML patient after graft failure recovered his hematopoiesis without additional cell therapies and he is still in complete hematological remission; conversely, his donor's cells failed the BM engraftment, while the other healthy donor's cells allowed complete BM reconstitution after HSCT.

Others have functionalized porous hydroxyapatite scaffolds resembling bone structure and functions with MSCs and specific extracellular matrix components, and repopulated with human purified cord blood derived CD34⁺ cells under perfusion [43]. In other studies, BM-like structure has been produced using polydimethylsiloxane or polystyrene, cryopreserved pre-isolated human CD34⁺ HSCs, and optimized culture medium [19]. However, despite polydimethylsiloxane or polystyrene can preserve the HSC compartment, these polymers can not be simultaneously assembled with live cells, and cell collection is difficult along the culture [19, 36]. Conversely, in our 3D system, alginate hydrogels can be quickly assembled with cells immobilized within, and easily depolymerized to further collect cells along the culture for further characterization. Moreover, alginate 3D culture systems allowed an optimal cell perfusion within the scaffold and an appropriate hypoxic gradient, as documented by high cell viability up to 21 days [44]. Most of published studies using CD34⁺ cells in 2D cell suspension or in 3D culture systems investigate in vitro hematopoiesis up to 14 days, because of the propensity of HSCs to terminally differentiate after a certain culture period [17]. In our model, BMMCs or PBSCs were mixed with calcium alginate hydrogel and BM-MSCs and maintained in culture up to 21 days without specific cytokines or growth factors in the culture medium, in contrast with literature reporting the importance of appropriate cytokines or developmental regulators to maintain HSC pluripotency [19, 36, 45]. Alginate-based hydrogels are largely employed in regenerative medicine because of their biodegradability and good biocompatibility [46], as they are already used and commercialized for wound dressing applications and are under investigation for nanodelivery of drugs and for tissue engineering [46]. Moreover, in 3D bioprinted BM niche models assembled with sodium and calcium alginate scaffolds have already been described as a biocompatible material for HSC and progenitor cell expansion and differentiation [21, 22]. In our study, we confirmed that alginate-based microbeads as a biocompatible material suitable for long-term HSC culture by preserving self-renewal and differentiation potentials through cell-to-cell contacts and paracrine feedbacks. Moreover, alginate scaffolds can be easily disassembled and cells can be recovered after culture for further experiments.

HSC maintenance in the BM niche relies on several interactions between stem cells, stromal cells -including MSCs-, extracellular matrix, or bone, and on the composition of BM microenvironment [44, 45, 46, 47]. The complexity of this system and these fine-tuned cross-talks between different cell types by direct contacts or by indirectly release of molecules makes challenging in vitro studies for better understanding of human BM niche biology and HSC maintenance and stem cell fate [44]. The use of 3D systems, largely optimized in regenerative medicine and tissue engineering, are promising in vitro models for mimicking BM niche architecture and functions [47, 48]. Moreover, advanced culture systems with high biological complexity are of increasing interest because they can reproduce physiological tissue organization both in health and diseases bypassing xenograft and mouse model utilization [43]. As a proof-of-concept, here we applied our 3D system to reproduce in vitro healthy and pathological hemopoiesis. We showed that frequencies of HSCs and myeloid progenitors were maintained throughout the culture period, except for the haploidentical HSC donor whose cells failed to engraft in the AML patient. Of note, patients' cells obtained after graft failure were mostly alive at the end of the culture, and their composition did not significantly change over time. This in vitro result confirmed clinical outcome of our AML patient who gradually recovered after graft failure under growth factor stimulation, and he is still in complete remission at the time of writing. We could only speculate on the reasons why AML patient's hemopoiesis recovered after graft failure and why his donor's cells failed to engraft in the BM. Interestingly, the two siblings shared two missense mutations of uncertain significance: Val231Leu on SETBP1 (exon 4, 691 G>C); and Ile1762Val on TET2 (exon 11, 5284 A>G). Conversely, the AML patient carried an Arg882His pathogenetic somatic mutation in DNMT3A (exon 23, 2645 G>A), while the donor had a missense somatic mutation of uncertain significance in c-Kit (Met541Leu, 1621 A>C) with a 99.1% VAF. DNMT3A is responsible for de novo addition of methyl groups to cytosine residues, frequently on a cytosine-guanine dinucleotide pair, an epigenetic DNA modification silencing gene expression [49, 50]. DNMT3A mutations are frequently found in several solid and hematological malignancies, including AML and myelodysplastic syndromes (MDS), and might act as a driver oncogenic mutation because they often coexist with secondary somatic alterations, such as ASXL1 or FLT3 mutations [51, 52, 53]. Mechanisms by which DNMT3A loss predisposes to leukemic transformation are still under investigation. In mouse models, DNMT3A loss predisposes to malignant transformation and increased self-renewal of hematopoietic stem cells (HSCs) rather than differentiation [53], while in vitro DNMT3A ablation can lead to abnormal RNA splicing, genomic instability, and DNA hypomethylation that might favor leukemogenesis [50]. However, DNMT3A mutations are also frequently described with ageing in healthy individuals, and these alterations increase the risk of AML development, all-cause mortality, and cardiovascular accidents [54]. The stem cell factor (SCF) receptor, c-Kit (also known as CD117), is a type III receptor tyrosine kinase involved in hematopoiesis, and is highly expressed on hematopoietic progenitors, mast cells, and other terminally differentiated cells, such as melanocytes [55]. c-Kit is quickly activated by SCF, and downstream signaling includes PI3-kinase, Src family kinases, mitogen-activated protein kinase, and phospholipases pathways [55]. Gain-of-function mutations are involved in neoplastic transformation, leading to hematological diseases, such as AML and mastocytosis, and solid tumors, including gastrointestinal stromal tumors [55, 56]. Conversely, loss-of-function mutations have been poorly described and found in benign conditions, such as piebaldism, an inherited disease with patches of white skin or hair [57]. The Met541Leu variant located on exon 10 of c-Kit (corresponding to the transmembrane domain), also found in our haploidentical donor, have been described in a case series of chronic myeloid leukemia Japanese patients, and in aggressive astrocytic gliomas [58, 59]. However, others have reported a similar frequency (8.1%) of this Met541Leu variant in normal population; therefore, clinical significance of this missense mutation is still unclear [60]. Here, we reported that hematopoietic progenitors from the haploidentical donor carrying the Met541Leu variant displayed a more differentiated profile compared to AML patient and healthy donor in vivo and in vitro throughout the 3D co-culture period of 21 days. Moreover, CD117 expression was significantly lower in the variant carrier compared to the other two subjects who did not show significant variations from baseline to the end of culture. Our findings suggest that the Met541Leu missense mutation on c-Kit gene might be a loss-of-function modification, likely causing an impaired membrane location of the receptor, as this mutation occurs in the transmembrane domain. Therefore, CD34⁺ HSC compartment might be not affected, as CD34⁺ were also documented in our 3D culture system at 14 days, while once committed, myeloid differentiation is altered. Indeed, as a proof-of-concept, PBSCs carrying Met541Leu mutation failed to engraft in the AML patient, while residual recipient's hemopoiesis normally repopulated the BM. Interestingly, the DNMT3A variant was also present after graft failure and recovery constituting almost half of total hemopoiesis, suggesting that DNMT3A carrying clones might have only a relative proliferative advantage in the setting of an ineffective hemopoiesis, such as in HSCs with loss-of-function of c-Kit. Moreover, DNMT3A altered functions might predispose to the acquisition of additional somatic mutations over time leading to leukemogenesis, as described in our AML patient [61].

Limitations of our model are: (i) lacking further characterization, such as proliferation capacity and apoptosis rate analysis by flow cytometry, or gene expression profiling by RT-qPCR, because of the paucity of available samples as they were obtained from an AML patient in an aplastic phase caused by graft failure or from residual PBSCs after stem cell infusion. (ii) Repopulation capacity after harvesting should be proven by additional assay, like colony-forming assay. (iii) Lacking comparison between different c-Kit mutation models because of the rarity of those molecular alterations.

In conclusions, studying hemopoiesis in vitro is still challenging because HSCs tend to quickly differentiate in culture, and appropriate expensive culture media are required. Moreover, cells carrying certain mutations, such as DNMT3A, PIGA, or clones harboring deletion of chromosome 7, are even more challenging to culture in vitro and no stable cell lines can be produced [50]. Therefore, the development of 3D culture systems is essential to reproduce in vitro physiological interactions and tissue structures to mimic the BM niche and cross-talks between normal and neoplastic cells and to better understand pathophysiology behind hematological malignancies [62]. Here, we provided a simple 3D co-culture system using BMMCs and BM-MSCs in an alginate-based hydrogel used as scaffold to replicate in vitro hematopoiesis in health and diseases and to elucidate biological mechanisms of somatic mutations of uncertain significance. Further perspectives of this study will be to develop a bio-plotting standardized protocol empowered by a 3D rapid prototyping system that can be also used as advanced diagnostic device in clinical setting for cell expansion and further characterization.

Declarations

Author contribution statement

Valentina Giudice, Giovanna Della Porta and Carmine Selleri: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper.

Paola Manzo, Pasqualina Scala, Marisa Gorrese, Angela Bertolini, Denise Morini, Antonio Pedicini, Barbara Izzo, Francesco Verdesca, Maddalena Langella and Bianca Serio: Performed the experiments; Analyzed and interpreted the data; contributed reagents, materials, analysis tools or data; Wrote the paper.

Francesca D’Alto and Rita Pepe: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

Data included in article/supp. material/referenced in article.

Declaration of interest’s statement

The authors declare no conflict of interest.

Additional information

Supplementary content related to this article has been published online at [URL].

Appendix A. Supplementary data

The following is the supplementary data related to this article:

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.