Culturing patient-derived malignant hematopoietic stem cells in engineered and fully humanized 3D niches
Edited by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved August 20, 2021 (received for review August 4, 2021)
Significance
Recapitulating human blood cancers in vitro remains challenging due to the limitations of current models to culture patient-derived malignant hematopoietic stem and progenitor cells in entirely human bone marrow microenvironments. We demonstrate that progenitor cells from patients with acute myeloid leukemia and myeloproliferative neoplasms can be cultured for at least 3 wk in fully human cell-based three-dimensional osteoblastic niches engineered in perfusion bioreactors, while exhibiting key features found in native bone marrow. Furthermore, the system can be customized to include a human vascular component within the engineered stromal microenvironment, which enables investigation of human leukemogenesis under designed settings. This platform can be used to test the effectiveness of chemotherapy compounds, toward application in patient-personalized medicine.
Abstract
Human malignant hematopoietic stem and progenitor cells (HSPCs) reside in bone marrow (BM) niches, which remain challenging to explore due to limited in vivo accessibility and constraints with humanized animal models. Several in vitro systems have been established to culture patient-derived HSPCs in specific microenvironments, but they do not fully recapitulate the complex features of native bone marrow. Our group previously reported that human osteoblastic BM niches (O-N), engineered by culturing mesenchymal stromal cells within three-dimensional (3D) porous scaffolds under perfusion flow in a bioreactor system, are capable of maintaining, expanding, and functionally regulating healthy human cord blood-derived HSPCs. Here, we first demonstrate that this 3D O-N can sustain malignant CD34+ cells from acute myeloid leukemia (AML) and myeloproliferative neoplasm patients for up to 3 wk. Human malignant cells distributed in the bioreactor system mimicking the spatial distribution found in native BM tissue, where most HSPCs remain linked to the niches and mature cells are released to the circulation. Using human adipose tissue-derived stromal vascular fraction cells, we then generated a stromal-vascular niche and demonstrated that O-N and stromal-vascular niche differentially regulate leukemic UCSD-AML1 cell expansion, immunophenotype, and response to chemotherapy. The developed system offers a unique platform to investigate human leukemogenesis and response to drugs in customized environments, mimicking defined features of native hematopoietic niches and compatible with the establishment of personalized settings.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Adult bone marrow (BM) stem cell niches are natural blood factories in which BM mesenchymal stromal cells (MSC-BM) and their extracellular matrices interact with hematopoietic stem and progenitor cells (HSPCs) to ensure their maintenance, self-renewal, and differentiation (1, 2). The composition of these BM niches is critical to determine HSPC function. In the murine system, quiescent HSPCs were originally found close to bone surfaces in osteoblastic niches (3–5), but subsequent studies showed that the vast majority of HSPCs locate in niches close to the sinusoidal vasculature (6). These so-called perivascular niches host active HSPCs that circulate between the BM and the bloodstream (7, 8). BM microenvironments also contribute to the development of hematopoietic malignancies (9, 10). Acute myeloid leukemia (AML) and myeloproliferative neoplasms (MPN) are clonal hematopoietic malignancies characterized by overproduction of neoplastic hematopoietic cells (11, 12), which evolve and hijack BM microenvironments to create supportive niches for leukemic stem cells (13–19). In contrast to their murine counterparts, human adult BM niches (20) have remained largely unexplored due to the limited accessibility.
Multiple humanized animal models have been generated in the last years to model hematological disorders (21), including recently developed patient-derived xenograft models with human BM niche-forming cells (22). Alternatively, human MSC-BM–coated scaffolds have been implanted in mice to generate humanized bone organs capable of hosting healthy murine and human HSPCs (23–26), but also patient-derived leukemic cells (27–30). Nevertheless, none of these in vivo approaches allows studying human malignant HSPCs over prolonged time under controlled settings. In vitro models mimicking the complexity of human BM would represent a solution to investigate the mechanisms of leukemogenesis, while bypassing the ethical concerns associated with the use of experimental animals (31). Some two-dimensional (2D) coculture systems were reported to maintain self-renewing AML progenitor cells for several weeks (from 3 up to 24 wk); these are based on cocultures with either murine (MS-5) or human (Saos-2, HS-5) cell lines (32–34), or human undifferentiated/adipogenic MSC-BM (35, 36). However, they do not recapitulate the complex, specialized signals and cell interactions of native human BM microenvironments. Recently, a three-dimensional (3D) hydrogel triculture system including primary AML cells, MSC-BM, and human umbilical vein endothelial cells (HUVEC) was proposed to model cell–cell interactions in the AML vascular niche (37). Despite introducing the vascular component, this system is based on undifferentiated MSC-BM and thus lacks important features from mature BM niches (e.g., osteoblastic cells and matrix). Although few 3D systems have been developed to model multiple myeloma (38, 39), they are based on classic static cultures and thus, they cannot recapitulate features such as BM interstitial flow and hematopoietic cell circulation. Therefore, 3D multicellular niches mimicking the complexity of human BM niches are not available to culture patient-derived malignant HSPCs for at least 1 wk.
We have previously developed a bioreactor-based 3D culture system for MSC-BM that can be used to model in vitro human BM stem cell niches (40). The system enables perfusion in alternate directions of MSC-BM suspensions directly through the pores of 3D hydroxyapatite scaffolds, mimicking the mineral component of trabecular bone. The resulting constructs, recapitulating features of a native human osteoblastic BM niche (O-N), can be further perfused with umbilical cord blood (CB) HSPCs, which establish direct interactions with the O-N and thereby maintain functional properties (41). The culture model offers the unprecedented opportunity to investigate in an in vitro system not only the growth/differentiation of HSPCs within a solid 3D niche, but also the exit of defined subpopulations of HSPCs or their differentiated progenies from the niche. Indeed, hematopoietic cells distributed differentially in the two compartments: while HSPCs remained within the O-N, committed blood cells were released into the liquid phase (supernatant) (41). This system could thus be used to investigate the effects of extrinsic factors (noncell autonomous) on HSPCs, as well as for disease modeling (42).
In this study, we aimed at exploiting 3D biomimetic niches engineered in perfusion bioreactors for culturing patient-derived blood cancer cells in fully human microenvironments. First, we assessed the culture of patient-derived AML and MPN CD34+ cells in engineered O-N niches for up to 3 wk. We investigated the impact of O-N on cell expansion, immunophenotype, gene expression, and distribution between the bioreactor compartments. Then, we tested the customization potential of our approach by engineering a stromal-vascular niche (SV-N) using human adipose tissue-derived stromal vascular fraction (SVF) cells, leading to stromal tissues enriched in endothelial cells and pericytes (43). We compared O-N and SV-N for culturing leukemic UCSD-AML1 cells and studied the influence of the different microenvironments on leukemic cell expansion, immunophenotype, gene expression, and response to gold-standard chemotherapy (Ara-C; cytarabine), the latter in comparison to classic 2D cultures.
Results
Engineered O-N Allow for the Long-Term Maintenance of Patient-Derived AML Progenitor Cells in Perfusion Bioreactors.
To assess the capacity of O-N to support the culture of human primary malignant cells, CD34+ cells were isolated from the BM of a myelodysplastic syndrome-diagnosed patient, which transformed into AML, and seeded in O-N generated using MSC-BM from healthy individuals. As control, AML cells were also seeded in cell-free scaffolds. We assessed weekly the maintenance/expansion and immunophenotype of leukemic cells released to the fluidic phase of the bioreactor system (supernatant) over the 3-wk culture using flow cytometry and gene-expression analysis. After 3 wk, bioreactors were dismounted for end-point analyses and cells were harvested from the two bioreactor compartments (i.e., supernatant and tissue contained within the chamber, based on O-N or cell-free scaffolds) (Fig. 1A).
Fig. 1.
Weekly analysis of CD45+ cells released into the supernatant indicated that AML cells were essentially undetectable after 1-wk culture in the absence of engineered niches (cell-free scaffolds). In contrast, a threefold increase in AML cell numbers was detected in the supernatant of bioreactors engineered with O-N as early as week 1, and stabilized along the 3-wk total culture period (Fig. 1B). Interestingly, while CD34+ progenitor populations decreased after 1-wk culture, more mature CD34− populations drastically increased, suggesting that most cells released to the supernatant underwent differentiation (SI Appendix, Fig. S1 A–D).
At end-point analyses (week 3), the comparison between cells collected from the supernatants and those harvested from digested O-N tissues revealed a threefold expansion of AML cells in the niche compartment (Fig. 1C). Importantly, the phenotypic assessment showed that the segregation of the different leukemic populations in the bioreactor system largely depends on their differentiation status. In fact, leukemic committed progenitor cells were significantly enriched in the O-N tissue, while mature CD34−CD38− leukemic cells were mostly detected in the supernatant (Fig. 1D and SI Appendix, Fig. S1E). In line with this, primitive CD33+CD34+CD117+ progenitors represented 4% of cells in the O-N tissue, but they were barely detectable in the supernatant (Fig. 1E and SI Appendix, Fig. S1F). Conversely, the proportion of mature CD33+ leukemic blasts was higher in the supernatant (Fig. 1F and SI Appendix, Fig. S1G). Thus, the distribution of AML cells in the bioreactor system mimicked that of the native BM tissue, where most HSPCs remain in their niches and mature cells are released to the bloodstream.
We next investigated if the specific AML cell distribution results from a differential gene expression in key chemotactic factors. The expression of CXCL12, VCAM1, and IL6, which are known to attract and regulate leukemic progenitor cells, was exclusively detected in cells retrieved from the O-N tissue (Fig. 1 G and H and SI Appendix, Fig. S1 H and L). Instead, macrophage colony-stimulating factor (M-CSF)—one of the main chemotactic and differentiating factors for mature myeloid cells—was almost five times more abundant in the supernatant fraction than in the O-N tissue (Fig. 1I). Inflammatory molecules, such as interleukin (IL)-8, MCP-1, and IL-1β were similar at gene and protein levels in both compartments (SI Appendix, Fig. S1 I–K and M–O). Altogether, our results indicate that patient-derived CD34+ AML cells can be cultured in perfusion bioreactors engineered with O-N, where they differentiate and distribute according to niche-specialized signals, as observed in native BM environments.
Patient-Derived MPN HSPCs Expand in O-N While Preserving the JAK2V617F Mutation.
In order to assess the applicability of the model to other hematological malignancies, we isolated CD34+ cells by phlebotomy in MPN patients. Similarly to the AML setting, MPN cells were cultured in O-N or in control cell-free scaffolds for 1 wk to assess potential maintenance/expansion, with CD34+ cells from buffy coats (BC) of healthy donors (eight pooled donors) as controls (Fig. 2A). After O-N tissue digestion, malignant cell expansion, immunophenotype, and distribution were analyzed by flow cytometry, whereas cell malignancy and function were assessed, respectively, through allele burden and colony forming-unit (CFU-C) assays (Fig. 2A).
Fig. 2.
While control healthy BC cells expanded similarly in bioreactors engineered with O-N and with cell-free scaffolds, MPN cells expanded significantly more (fivefold) in bioreactors containing O-N (Fig. 2B). Similar to AML cells and in contrast to healthy BC cells, CD34+ progenitor populations were reduced after 1-wk culture, whereas mature CD34− populations were significantly expanded (Fig. 2 C–F and SI Appendix, Fig. S2 A–D). Despite these differences in blood cells, the mesenchymal fraction remained unaltered (SI Appendix, Fig. S2E). These findings, together with high capacity of MPN cells to form burst forming unit colonies (SI Appendix, Fig. S2F), point toward a differentiation of most MPN cells after 1-wk culture in O-N. To assess the fraction of malignant cells among MPN CD34+ cells, we measured the JAK2V617F allele burden before and after the culture in bioreactors with O-N. While 20% of the starting MPN CD34+ cells carried the mutation, we observed a substantial increase in malignant cell development, with 70% of total cells carrying the malignant-driving mutation after 1-wk culture in bioreactors O-N (Fig. 2G). This suggests the enrichment of JAK2V617F CD34+ cells in O-N tissues and confirms the dependency of malignant hematopoietic cells on niche-derived signals. Consistent with the compartmentalized cell distribution observed with AML cells (Fig. 1 C–F), MPN progenitor cells were also found preferentially in the O-N tissue, whereas the supernatant fraction contained more than 60% of mature CD34−CD38− cells (Fig. 2 H and I).
In summary, our results demonstrate that malignant CD34+ cells derived from patients with hematopoietic malignancies such as AML or MPN require O-N signals to engraft and expand in our 3D biomimetic cultures within perfusion bioreactors. Moreover, malignant primary cells not only engraft in O-N, but they also recapitulate some features found in native niches, such as differentiation, migration, and expansion of the mutant cell fraction.
Adipose Tissue-Derived Cells Generate an Alternative SV-N for Malignant HSPCs.
Native BM contains multiple microenvironments that differentially regulate healthy and malignant hematopoiesis. Therefore, in order to assess the culture of the same leukemic cells in different engineered niches and to exemplify the customization potential of the bioreactor model, we adapted an experimental set-up previously validated as angiogenic niche (43) to mimic the BM stromal-vascular environment. In contrast to the O-N, the SV-N was generated by culturing human adipose tissue-derived SVF cells (known to include stromal progenitors and endothelial lineage cells) for 3 wk within collagen-based scaffolds, which do not provide specific osteoblastic differentiation signals, and in the presence of FGF-2, to support the growth of both stromal and endothelial cells. The SV-N was then compared to the O-N by histological, gene expression and cytofluorimetric analyses (SI Appendix, Fig. S3A).
Immunofluorescence characterization of whole-mount engineered tissues documented the formation of self-organized 3D vascular structures composed by CD31+ endothelial cells and NG2+ perivascular cells in the SV-N (Fig. 3 A–E and SI Appendix, Fig. S3 B and F–M), while those structures were totally absent in the O-N (Fig. 3F). While VEGF proteins could be detected in stromal cells from both engineered niches, gene-expression analysis reported a significant CD31 and VWF gene expression exclusively detectable in the SV-N (SI Appendix, Fig. S3 C–E). Consistently, cytofluorimetry analyses enabled to quantify a significant enrichment in endothelial (CD34+CD31+) and perivascular cells (CD34−CD146+) in the SV-N compared to the O-N (Fig. 3 G and H), while both engineered niches contained a substantial nonvascular CD90+ mesenchymal cell fraction (Fig. 3I). In order to better characterize those nonvascular cells, we performed immunostainings for osteocalcin (OCN) and collagen I (COL1A1). As expected, we detected a dense osteogenic matrix in O-N, but those markers were drastically reduced in the SV-N (Fig. 3 J–O). This was also confirmed by the reduced alkaline phosphatase (ALPL) and osteocalcin (OCN) gene expression in the SV-N (Fig. 3 P and Q). These results indicate the successful engineering of two distinct microenvironments (O-N and SV-N) displaying specific structural, cellular, and molecular features.
Fig. 3.
Engineered O-N and SV-N Differentially Influence Leukemic UCSD-AML1 Cells in 3D Culture.
In order to test whether our differently engineered niches could modulate the culture of malignant HSPCs, we used the EVI-1–overexpressing leukemic cell line UCSD-AML1 (44). This cell line exhibits an immunophenotype similar to human HSPCs (CD45+CD34+CD38−CD45RA+CD90+/−) (SI Appendix, Fig. S4A). The use of a cell line enables availability of large numbers of leukemic cells, required to then compare their regulation by two different engineered niches (O-N and SV-N), without exposing the experiments to the large variability of patient-derived malignant cells.
Before assessing UCSD-AML1 cells in bioreactors with engineered niches, we evaluated cell maintenance and expansion for 3 wk in cell-free hydroxyapatite scaffolds inside the perfusion bioreactors. Cells released from the cell-free scaffold into the supernatant were weekly collected. After the 3-wk culture, the scaffolds were enzymatically digested to harvest cells attached to them. UCSD-AML1 cells expanded along the 3-wk culture period (SI Appendix, Fig. S4 B–E), and almost all of them maintained their stem cell-like immunophenotype (CD34+CD38−) (SI Appendix, Fig. S4 F–H). Interestingly, this cell expansion correlated with increased expression of leukemic cell marker genes, such as EVI-1 and CD52 (SI Appendix, Fig. S4 I and J) (45), and proteins like ANGPT-1 and ANGPT-2 (SI Appendix, Fig. S4 M and N) (46). In contrast, CALRCL and ITGA6 mRNA expression (SI Appendix, Fig. S4 K and L), and MCP-1 protein content (SI Appendix, Fig. S4P) were not affected, and M-CSF content decreased over culture time (SI Appendix, Fig. S4O). These results demonstrate that, in the absence of an engineered niche, UCSD-AML1 leukemic cells can be maintained and expanded in 3D cell-free scaffolds under perfusion flow for at least 3 wk, with induced changes in gene and protein expression.
We then investigated the impact of both engineered niches (O-N and SV-N) on the culture of UCSD-AML1 cells in perfusion-based bioreactors. For this purpose, we first analyzed the leukemic cells released to the supernatant of bioreactors engineered with the O-N and SV-N over a 3-wk culture. UCSD-AML1 cells expanded in bioreactors with both engineered niches (approximately 40-fold) (Fig. 4A and SI Appendix, Fig. S5 A and B), although to a lower extent than in bioreactors with cell-free scaffolds (approximately 100-fold) (SI Appendix, Fig. S4 B–D). Interestingly, while almost all UCSD-AML1 cells cultured in cell-free scaffolds were CD34+CD38− (SI Appendix, Fig. S4F), a small fraction of UCSD-AML1 cells (6 to 8%) lost the CD34 expression and expanded when cultured in bioreactors with engineered niches (Fig. 4D and SI Appendix, Fig. S5C). At the gene-expression level, UCSD-AML1 cells cultured in engineered niches did not up-regulate the oncogenic genes EVI-1 and CD52 (Fig. 4 E and F) in contrast to cells cultured in cell-free scaffolds (SI Appendix, Fig. S4 I and J). During UCSD-AML1 cell culture in bioreactors with engineered niches, we also observed increased expression of CXCR4 (SI Appendix, Fig. S5F). This is the receptor for CXCL12, a key chemokine promoting HSPC maintenance in BM niches (47).
Fig. 4.
Interestingly, leukemic cells displayed different profiles if cultured in bioreactors engineered with the O-N or SV-N. UCSD-AML1 cells cultured in bioreactors with the SV-N for 1 wk had lower expression of EVI-1, CD52, and ANGPT-1 (Fig. 4 E–G), increased levels of ANGPT-2 (the antagonist of ANGPT-1) and proinflammatory cytokines, such as IL-6, IL-8, and MCP-1 (Fig. 4H and SI Appendix, Fig. S5 G–I). These changes preceded a reduction in CD34+CD38−CD90− UCSD-AML1 cells after the first week of culture (Fig. 4C and SI Appendix, Fig. S5B) and disappeared at the end of the culture period (3 wk). This might reflect the adaptation of leukemic cells to different microenvironments and changes in the niche cell composition over culture time. Thus, our results indicate that engineered niches can differentially control/impact the proliferation of UCSD-AML1 leukemic cells and modulate their gene expression and phenotype.
At the end of the 3-wk culture, bioreactors were dismounted and tissues were fixed for histological analyses or enzymatically digested to analyze their cellular composition by flow cytometry. Whole-mount immunofluorescence analyses of control engineered niches confirmed the presence of CD34+ UCSD-AML1 leukemic cells in the osteogenic matrix of the O-N (Fig. 5 A–D), and around the vasculature (composed by CD31+CD34+ endothelial cells and NG2+ pericytes) of the the SV-N (Fig. 5 E–H). In agreement with supernatant data (Fig. 4), UCSD-AML1 cell expansion was reduced in engineered niche tissues in comparison to cell-free scaffolds (Fig. 5I). However, we also observed here that a small fraction of UCSD-AML1 cells lost CD34 expression in engineered niches. Importantly, these CD34−CD38− UCSD-AML1 cells, which exhibit a phenotype compatible with differentiated cells, expanded eight times more than in cell-free scaffolds (Fig. 5 J–L and SI Appendix, Fig. S6 A–C). UCSD-AML1 cells also exhibited lower capacity to form CFU-C when cultured in engineered niches (SI Appendix, Fig. S6D). The CD34+CD38−CD90− population was specifically reduced in the SV-N (Fig. 5K and SI Appendix, Fig. S6B), and the increase in CD34−CD38− cells was more pronounced in the SV-N than in the O-N (Fig. 5L).
Fig. 5.
In order to determine the relative contribution of the scaffolding material to the differences observed between both engineered niches, we also assessed the culture of UCSD-AML1 leukemic cells in the SV-N engineered on hydroxyapatite scaffolds (Engipore; SV-N EP) and compared it to the culture in the SV-N engineered on collagen scaffolds (ZimmerPatch; SV-N ZP). UCSD-AML1 cells cultured in bioreactors with the SV-N EP and SV-N ZP had similar expansion and phenotypic patterns (SI Appendix, Fig. S7), suggesting that the contribution of the scaffolding material is rather minimal. We next investigated whether defined signals derived from the O-N and the SV-N could explain the differential UCSD-AML1 cell expansion and immunophenotype. Gene and protein expression analyses of engineered niches showed striking differences. HSPC chemotactic genes, such as CXCL12 and VCAM1, and matrix degradation genes, like CTSK and MMP13, were significantly reduced in the SV-N (Fig. 5 M–P). In contrast, we generally did not detect significant differences in terms of proinflammatory cytokines at gene or protein level (SI Appendix, Fig. S6 E–L). Reduced expression of genes related to hematopoietic cell maintenance and niche remodeling in the SV-N might explain reduced leukemic cell expansion and CD34 expression in this engineered niche. In summary, we showed that our system can be customized for the generation of biomimetic O-N and SV-N, supporting and controlling leukemic cell expansion and differentiation in a distinct fashion.
Engineered Niches Protect Leukemic Cells from Chemotherapy Treatment.
In order to validate our system as a possible drug-testing platform, we assessed the effectiveness of gold-standard chemotherapy (Ara-C; cytarabine) against UCDS-AML1 leukemic cells cultured in 2D classic systems (monoculture or coculture with MSC-BM) or in bioreactors with 3D engineered niches (O-N and SV-N) by measuring cell viability (SI Appendix, Fig. S8A). As expected, Ara-C treatment induced massive cell death in UCSD-AML1 leukemic cells in both 2D monoculture and coculture conditions. Still, we detected a minor but significant reduction in the mortality of UCSD-AML1 cells cocultured with MSC-BM, suggesting a minor protection derived from the monolayer of stromal cells (Fig. 6A and SI Appendix, Fig. S8 B–G). Then, we analyzed the impact of Ara-C on leukemic cells cultured in bioreactors with engineered niches. Mirroring the effect observed in 2D culture, the vast majority of UCSD-AML1 cells isolated from the supernatant fraction of the bioreactors were either apoptotic or dead (Fig. 6B). In sharp contrast to leukemic cells isolated from the supernatant or 2D cultures, those cells isolated from the engineered niche tissues exhibited higher viability (Fig. 6 A–C), suggesting that the O-N and SV-N can protect leukemic cells from chemotherapy. Interestingly, the survival of leukemic cells in the O-N was considerably higher than in the SV-N (Fig. 6C), while stromal cells were not affected by the chemotherapy (SI Appendix, Fig. S8H). These results point out the importance of suitable 3D niche models to test the effectiveness of gold-standard chemotherapy on leukemic cells. It also supports the relevance of the O-N and SV-N as possible niche configurations to investigate specific regulatory factors mediating the leukemic cell response to candidate therapeutic compounds.
Fig. 6.
Discussion
In this study we demonstrate that cellular niches can be engineered in a 3D scaffold- and perfusion flow-based bioreactor system and exploited for prolonged ex vivo maintenance, expansion, as well as phenotypic and functional regulation of patient-derived malignant HSPCs. The fully humanized model was applied to investigate human leukemogenesis in the presence of customized niche components (i.e., osteoblastic vs. stromal-vascular elements) and to test responsiveness to chemotherapy (Fig. 7).
Fig. 7.
Classic 2D cultures based on pioneering studies performed by Dexter et al. (48) have contributed to the understanding of specific molecular pathways in hematopoiesis. Yet, they were not sufficient to model more complex processes, such as BM disruption by neoplastic cells or address niche-dependent factors that protect them from treatment. This is due to the limitations of 2D systems to recapitulate native human BM microenvironments and to maintain leukemic cells in long-term cultures without adding hyperphysiological doses of certain cytokines (49). Some 3D culture systems mimicking features of human BM niches have been assessed for the culture of malignant hematopoietic cells. A static biomimetic 3D osteoblastic niche was shown to better maintain BM mononuclear cells from chronic myeloid leukemia patients than traditional 2D culture systems with a feeder-cell layer. Nevertheless, the mutant cell fraction was progressively reduced during the culture (50). Highly porous scaffolds made from polymeric materials coated with human extracellular matrix (ECM)/MSC-BM proteins were shown to support the growth of AML subtype-specific cell lines in the absence of exogenous growth factors (51), and increase their resistance to drug-induced apoptosis (52–54). Similarly, umbilical cord-derived ECM was shown to support leukemic cell lines with stem cell features (55). Recently, the TF-1 erythroleukemic cell line cultured in a 3D microfluidic platform was shown to exhibit an increased proliferation rate and drug resistance compared to a 2D microfluidic device (56). However, none of these systems has fully demonstrated the ex vivo long-term maintenance of patient-derived leukemic cells in 3D fully human cell-based microenvironments.
In this study, we propose a 3D biomimetic microenvironment (O-N) engineered entirely from human primary material (human MSC-BM) that provides key niche signals and influences the localization of patient-derived malignant HSPCs in the system during prolonged culture time. In this regard, human primary HSPCs were not just maintained but, mimicking native BM niches, they exhibited a specific spatial distribution in which most HSPCs located within the niche compartment and committed cells were released to the supernatant (mimicking the circulation). Another bioreactor-based 3D culture model also allowed the parallel analysis of cells retained inside the niche and those released to the supernatant and was exploited to assess chronic lymphocytic leukemia cell mobilization in response to pharmacological agents (57); however, this system used a BM stromal cell line (HS-5) instead of primary MSC-BM and only short time culture (3 d).
Supporting this compartmentalization, key HSPC chemotactic factors, such as CXCL12 and VCAM1, were detected in our engineered niches. The counterpart of murine CXCL12-abundant reticular (CAR) cells was recently identified in human adult BM to correspond to adipo-osteogenic progenitors with hematopoiesis-supporting abilities (58). Accordingly, the O-N exhibited double CXCL12 expression compared to the SV-N. Moreover, VCAM1 expression was also higher expressed in the O-N, suggesting that vascular cells are not the main source of VCAM1 in our engineered niches. In this regard, a nonendothelial VCAM1+ stromal population was reported in human BM (59). Finally, matrix remodeling genes, such as CTSK and MMP13, were exclusively detected in the O-N, which might indicate that osteoblasts are the cells expressing these genes in engineered niches.
In order to demonstrate that our system allows for customization, we have engineered an SV-N that adds to our previously validated O-N. Human adipose tissue-derived SVF cells were shown to generate an angiogenic niche with vascular structures after 5-d culture in 3D culture within perfusion-based bioreactors (43, 60). Here we demonstrate that these vascular cells are still present after a 3-wk culture, although they are less abundant. In particular, we could detect around 8% of CD34−CD146+ perivascular cells in the SV-N, while this population represented ∼20% in the angiogenic niche generated after the 5-d culture of SVF cells (43). CD34+CD31+ endothelial cells were also slightly reduced but, in contrast to the previous study, most of them were integrated in large capillary structures composed by five or more cells. Beyond the relative abundance of vascular cells, our results demonstrate that the SV-N can differentially control the expansion and immunophenotype of the leukemic cell line UCSD-AML1 in comparison to the avascular O-N. Therefore, our system allows the culture of malignant human hematopoietic cells in two functional engineered niches that mimic features from the O-N and SV-N. Since these microenvironments are known to exert different effects on malignant HSPCs, our system might be combined with omics-based approaches to compare and predict critical interactions between malignant HSPCs and specific niches without interferences from other BM components or systemic effects. This is challenging to achieve in vivo without genetic manipulations or depletions of specific cell populations, since both HSPCs and stromal cells are exposed to multiple and complex signals at the same time.
While cells engrafted in humanized mouse models exhibit a similar mutant allele burden than that detected in HSPCs isolated from the primary sample (61–63), we observed a 3.5-fold increase of JAK2V617F allele burden in HSPCs after 1-wk culture in the O-N, suggesting that these engineered niches selectively enrich the mutant cell fraction from patient-derived biopsies. This might reflect the superior capacity of human mutant cells to expand and hijack the niche in absence of murine factors.
Since leukemic cells cause several abnormalities in the murine vascular network that decisively contribute to disease progression (18), it is essential to confirm and investigate this cross-talk in a model with human vasculature. Different 3D microfluidic approaches have been developed to recreate perivascular niches that, in some cases, were exploited as preclinical models to assess vasculature-malignant hematopoietic cell interactions (64–66). Despite offering great advantages, microfluidic devices are often limited by their own dimensions and their capacity to reproduce large and complex biological structures like vascular networks with appropriate cell densities (67). At larger scale, AML cells, MSC-BM, and HUVEC were cocultured in a 3D hydrogel system aiming to recapitulate the AML vascular niche (37). However, this more-advanced system still lacks a crucial element in vascular niches, namely perivascular cells. In this context, the SV-N engineered in perfusion bioreactors circumvents these size-related limitations and contains entirely human vessels with endothelial and perivascular cells. Moreover, the model bypasses the challenge to decouple the contribution of the murine vasculature and circulating cytokines in humanized in vivo models.
BM microenvironments are known to influence the response of leukemic cells to chemotherapy, conditioning its effectiveness (68). Indeed, both osteoblastic and vascular niches have been shown in murine models to provide protection of leukemic cells from chemotherapy (69–71). In fully humanized in vitro settings, our results reveal that both engineered 3D niches tested could offer a superior protection against chemotherapy in comparison to 2D monocultures o cocultures systems. The fact that leukemic cells were more susceptible to chemotherapy when cultured in the SV-N than in the O-N might be explained by the possibly differential accessibility of the niches to the drug, or by specific cellular/molecular components of the niches. In this regard, VCAM1/VLA4 signaling was reported to activate survival and proliferative pathways in leukemic cells, leading to higher chemotherapy resistance (72). The significantly reduced presence of VCAM1 in the SV-N might partially explain the decreased survival of leukemic cells cultured in bioreactors with this engineered niche.
In summary, here we propose a bioengineering approach generating patient-derived (human malignant HSPCs and healthy stromal cells), biomimetic, and customizable (O-N and SV-N) 3D microenvironments that can be exploited to study leukemogenesis or as a drug-testing platform (Fig. 7). In this regard, our primary cells-based system might be a suitable model for patient personalized medicine, since it allows to study leukemic cells interactions not only with healthy allogenic MSCs, but also with patient-matched stroma.
Materials and Methods
Human BM aspirates were collected from healthy donors and AML patients at the University Hospital Basel to extract human MSC-BM for engineering of O-N and isolate CD34+ AML cells. Blood cells were collected from BCs from healthy donors at the Blutspendezentrum Basel and from phlebotomies of MPN patients at the University Hospital Basel. SVF cells were isolated from human adipose tissue samples collected from patients who underwent liposuction or abdominoplasty operations. Informed consent was obtained preoperatively and the local ethics committee (Ethikkommission Nordwest und Zentralschweiz, ref. 78/07) approved the protocol. All experiments were compliant with European Union recommendations. Detailed methods for human MSC-BM and SVF cells extraction and culture, engineering of the O-N and SV-N, UCSD-AML1 cell line culture, HSPC extraction and purification, malignant hematopoietic cell seeding, cell harvesting, flow cytometry, immunostaining, Luminex assay, qPCR, allele burden assay, and CFU-C assay are available in SI Appendix.
Data Availability
All study data are included in the main text and SI Appendix.
Acknowledgments
We thank all members of the Tissue Engineering group for advice and support; Department of Biomedicine facilities for assistance and support; and C. Korn for data discussion and critical reading. This work was supported by the Swiss National Foundation Div 3 Grant 31003A-179259 (to I.M.) and National Centre of Competence in Research (NCCR) Molecular Systems Engineering Grant 51NF40-141825 (to I.M.), as well as by the Freenovation program (Novartis Foundation to P.E.B. and M.H.).
Supporting Information
Appendix (PDF)
- Download
- 1.29 MB
References
1
S. Méndez-Ferrer et al., Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature 466, 829–834 (2010).
2
S. J. Morrison, D. T. Scadden, The bone marrow niche for haematopoietic stem cells. Nature 505, 327–334 (2014).
3
L. M. Calvi et al., Osteoblastic cells regulate the haematopoietic stem cell niche. Nature 425, 841–846 (2003).
4
J. Zhang et al., Identification of the haematopoietic stem cell niche and control of the niche size. Nature 425, 836–841 (2003).
5
F. Arai et al., Tie2/angiopoietin-1 signaling regulates hematopoietic stem cell quiescence in the bone marrow niche. Cell 118, 149–161 (2004).
6
M. J. Kiel, S. J. Morrison, Uncertainty in the niches that maintain haematopoietic stem cells. Nat. Rev. Immunol. 8, 290–301 (2008).
7
S. Méndez-Ferrer, D. Lucas, M. Battista, P. S. Frenette, Haematopoietic stem cell release is regulated by circadian oscillations. Nature 452, 442–447 (2008).
8
A. García-García et al., Dual cholinergic signals regulate daily migration of hematopoietic stem cells and leukocytes. Blood 133, 224–236 (2019).
9
D. S. Krause, D. T. Scadden, A hostel for the hostile: The bone marrow niche in hematologic neoplasms. Haematologica 100, 1376–1387 (2015).
10
C. Korn, S. Méndez-Ferrer, Myeloid malignancies and the microenvironment. Blood 129, 811–822 (2017).
11
T. C. Somervaille, M. L. Cleary, Identification and characterization of leukemia stem cells in murine MLL-AF9 acute myeloid leukemia. Cancer Cell 10, 257–268 (2006).
12
C. H. Jamieson et al., Granulocyte-macrophage progenitors as candidate leukemic stem cells in blast-crisis CML. N. Engl. J. Med. 351, 657–667 (2004).
13
K. J. Hope, L. Jin, J. E. Dick, Acute myeloid leukemia originates from a hierarchy of leukemic stem cell classes that differ in self-renewal capacity. Nat. Immunol. 5, 738–743 (2004).
14
F. Ishikawa et al., Chemotherapy-resistant human AML stem cells home to and engraft within the bone-marrow endosteal region. Nat. Biotechnol. 25, 1315–1321 (2007).
15
A. Colmone et al., Leukemic cells create bone marrow niches that disrupt the behavior of normal hematopoietic progenitor cells. Science 322, 1861–1865 (2008).
16
D. S. Krause et al., Differential regulation of myeloid leukemias by the bone marrow microenvironment. Nat. Med. 19, 1513–1517 (2013).
17
K. Schepers et al., Myeloproliferative neoplasia remodels the endosteal bone marrow niche into a self-reinforcing leukemic niche. Cell Stem Cell 13, 285–299 (2013).
18
D. Passaro et al., Increased vascular permeability in the bone marrow microenvironment contributes to disease progression and drug response in acute myeloid leukemia. Cancer Cell 32, 324–341.e6 (2017).
19
S. Méndez-Ferrer et al., Bone marrow niches in haematological malignancies. Nat. Rev. Cancer 20, 285–298 (2020).
20
B. Sacchetti et al., Self-renewing osteoprogenitors in bone marrow sinusoids can organize a hematopoietic microenvironment. Cell 131, 324–336 (2007).
21
S. Goyama, M. Wunderlich, J. C. Mulloy, Xenograft models for normal and malignant stem cells. Blood 125, 2630–2640 (2015).
22
A. Abarrategi et al., Modeling the human bone marrow niche in mice: From host bone marrow engraftment to bioengineering approaches. J. Exp. Med. 215, 729–743 (2018).
23
J. Lee et al., Implantable microenvironments to attract hematopoietic stem/cancer cells. Proc. Natl. Acad. Sci. U.S.A. 109, 19638–19643 (2012).
24
B. M. Holzapfel et al., Tissue engineered humanized bone supports human hematopoiesis in vivo. Biomaterials 61, 103–114 (2015).
25
M. S. Ventura Ferreira et al., An engineered multicomponent bone marrow niche for the recapitulation of hematopoiesis at ectopic transplantation sites. J. Hematol. Oncol. 9, 4 (2016).
26
K. Fritsch et al., Engineered humanized bone organs maintain human hematopoiesis in vivo. Exp. Hematol. 61, 45–51.e5 (2018).
27
A. Antonelli et al., Establishing human leukemia xenograft mouse models by implanting human bone marrow-like scaffold-based niches. Blood 128, 2949–2959 (2016).
28
P. Sontakke et al., Modeling BCR-ABL and MLL-AF9 leukemia in a human bone marrow-like scaffold-based xenograft model. Leukemia 30, 2064–2073 (2016).
29
M. Carretta et al., Genetically engineered mesenchymal stromal cells produce IL-3 and TPO to further improve human scaffold-based xenograft models. Exp. Hematol. 51, 36–46 (2017).
30
A. Abarrategi et al., Versatile humanized niche model enables study of normal and malignant human hematopoiesis. J. Clin. Invest. 127, 543–548 (2017).
31
P. E. Bourgine, I. Martin, T. Schroeder, Engineering human bone marrow proxies. Cell Stem Cell 22, 298–301 (2018).
32
D. van Gosliga et al., Establishing long-term cultures with self-renewing acute myeloid leukemia stem/progenitor cells. Exp. Hematol. 35, 1538–1549 (2007).
33
E. Griessinger et al., A niche-like culture system allowing the maintenance of primary human acute myeloid leukemia-initiating cells: A new tool to decipher their chemoresistance and self-renewal mechanisms. Stem Cells Transl. Med. 3, 520–529 (2014).
34
B. A. Cardoso, H. Belo, J. T. Barata, A. M. Almeida, The bone marrow-mediated protection of myeloproliferative neoplastic cells to vorinostat and ruxolitinib relies on the activation of JNK and PI3K signalling pathways. PLoS One 10, e0143897 (2015).
35
S. Ito et al., Long term maintenance of myeloid leukemic stem cells cultured with unrelated human mesenchymal stromal cells. Stem Cell Res. (Amst.) 14, 95–104 (2015).
36
M. Azadniv et al., Bone marrow mesenchymal stromal cells from acute myelogenous leukemia patients demonstrate adipogenic differentiation propensity with implications for leukemia cell support. Leukemia 34, 391–403 (2020).
37
L. J. Bray et al., A three-dimensional ex vivo tri-culture model mimics cell-cell interactions between acute myeloid leukemia and the vascular niche. Haematologica 102, 1215–1226 (2017).
38
P. de la Puente et al., 3D tissue-engineered bone marrow as a novel model to study pathophysiology and drug resistance in multiple myeloma. Biomaterials 73, 70–84 (2015).
39
H. Fairfield et al., Development of a 3D bone marrow adipose tissue model. Bone 118, 77–88 (2019).
40
N. Di Maggio et al., Toward modeling the bone marrow niche using scaffold-based 3D culture systems. Biomaterials 32, 321–329 (2011).
41
P. E. Bourgine et al., In vitro biomimetic engineering of a human hematopoietic niche with functional properties. Proc. Natl. Acad. Sci. U.S.A. 115, E5688–E5695 (2018).
42
P. Sommerkamp, F. E. Mercier, A. C. Wilkinson, D. Bonnet, P. E. Bourgine, Engineering human hematopoietic environments through ossicle and bioreactor technologies exploitation. Exp. Hematol. 94, 20–25 (2021).
43
G. Cerino et al., Engineering of an angiogenic niche by perfusion culture of adipose-derived stromal vascular fraction cells. Sci. Rep. 7, 14252 (2017).
44
J. Oval, O. W. Jones, M. Montoya, R. Taetle, Characterization of a factor-dependent acute leukemia cell line with translocation (3;3)(q21;q26). Blood 76, 1369–1374 (1990).
45
Y. Saito et al., CD52 as a molecular target for immunotherapy to treat acute myeloid leukemia with high EVI1 expression. Leukemia 25, 921–931 (2011).
46
H. R. Saha et al., Suppression of GPR56 expression by pyrrole-imidazole polyamide represents a novel therapeutic drug for AML with high EVI1 expression. Sci. Rep. 8, 13741 (2018).
47
T. Sugiyama, H. Kohara, M. Noda, T. Nagasawa, Maintenance of the hematopoietic stem cell pool by CXCL12-CXCR4 chemokine signaling in bone marrow stromal cell niches. Immunity 25, 977–988 (2006).
48
T. M. Dexter, E. G. Wright, F. Krizsa, L. G. Lajtha, Regulation of haemopoietic stem cell proliferation in long term bone marrow cultures. Biomedicine (Paris) 27, 344–349 (1977).
49
H. J. Sutherland, A. Blair, R. W. Zapf, Characterization of a hierarchy in human acute myeloid leukemia progenitor cells. Blood 87, 4754–4761 (1996).
50
L. Hou et al., Long-term culture of leukemic bone marrow primary cells in biomimetic osteoblast niche. Int. J. Hematol. 90, 281–291 (2009).
51
T. M. Blanco, A. Mantalaris, A. Bismarck, N. Panoskaltsis, The development of a three-dimensional scaffold for ex vivo biomimicry of human acute myeloid leukaemia. Biomaterials 31, 2243–2251 (2010).
52
O. S. Aljitawi et al., A novel three-dimensional stromal-based model for in vitro chemotherapy sensitivity testing of leukemia cells. Leuk. Lymphoma 55, 378–391 (2014).
53
E. G. Velliou et al., Towards unravelling the kinetics of an acute myeloid leukaemia model system under oxidative and starvation stress: A comparison between two- and three-dimensional cultures. Bioprocess Biosyst. Eng. 38, 1589–1600 (2015).
54
M. Karimpoor et al., Development of artificial bone marrow fibre scaffolds to study resistance to anti-leukaemia agents. Br. J. Haematol. 182, 924–927 (2018).
55
D. Li et al., Decellularized Wharton jelly matrix: A biomimetic scaffold for ex vivo hematopoietic stem cell culture. Blood Adv. 3, 1011–1026 (2019).
56
M. Houshmand et al., Mimicking the acute myeloid leukemia niche for molecular study and drug screening. Tissue Eng. Part C Methods 23, 72–85 (2017).
57
F. Barbaglio et al., 3D co-culture model of chronic lymphocytic leukemia bone marrow microenvironment predicts patient-specific response to mobilizing agents. Haematologica 106, 2334–2344 (2020).
58
K. Aoki et al.; HANDAI Clinical Blood Club, Identification of CXCL12-abundant reticular cells in human adult bone marrow. Br. J. Haematol. 193, 659–668 (2021).
59
D. H. Schaumann, J. Tuischer, W. Ebell, R. A. Manz, R. Lauster, VCAM-1-positive stromal cells from human bone marrow producing cytokines for B lineage progenitors and for plasma cells: SDF-1, flt3L, and BAFF. Mol. Immunol. 44, 1606–1612 (2007).
60
A. Scherberich, R. Galli, C. Jaquiery, J. Farhadi, I. Martin, Three-dimensional perfusion culture of human adipose tissue-derived endothelial and osteoblastic progenitors generates osteogenic constructs with intrinsic vascularization capacity. Stem Cells 25, 1823–1829 (2007).
61
A. Reinisch et al., A humanized bone marrow ossicle xenotransplantation model enables improved engraftment of healthy and leukemic human hematopoietic cells. Nat. Med. 22, 812–821 (2016).
62
H. Celik et al., A humanized animal model predicts clonal evolution and therapeutic vulnerabilities in myeloproliferative neoplasms. Cancer Discov., (2021).
63
V. Lysenko et al., Enhanced engraftment of human myelofibrosis stem and progenitor cells in MISTRG mice. Blood Adv. 4, 2477–2488 (2020).
64
B. Carrion et al., Recreating the perivascular niche ex vivo using a microfluidic approach. Biotechnol. Bioeng. 107, 1020–1028 (2010).
65
S. S. Kotha et al., Engineering a multicellular vascular niche to model hematopoietic cell trafficking. Stem Cell Res. Ther. 9, 77 (2018).
66
D. B. Chou et al., On-chip recapitulation of clinical bone marrow toxicities and patient-specific pathophysiology. Nat. Biomed. Eng. 4, 394–406 (2020).
67
K. Haase, R. D. Kamm, Advances in on-chip vascularization. Regen. Med. 12, 285–302 (2017).
68
H. S. Zhou, B. Z. Carter, M. Andreeff, Bone marrow niche-mediated survival of leukemia stem cells in acute myeloid leukemia: Yin and yang. Cancer Biol. Med. 13, 248–259 (2016).
69
I. G. Winkler et al., Vascular niche E-selectin regulates hematopoietic stem cell dormancy, self renewal and chemoresistance. Nat. Med. 18, 1651–1657 (2012).
70
S. Azizidoost, V. Vijay, C. R. Cogle, E. Khodadi, N. Saki, The role and clinical implications of the endosteal niche and osteoblasts in regulating leukemia. Clin. Transl. Oncol. 19, 1059–1066 (2017).
71
M. G. Poulos et al., Activation of the vascular niche supports leukemic progression and resistance to chemotherapy. Exp. Hematol. 42, 976–986.e3 (2014).
72
R. Jacamo et al., Reciprocal leukemia-stroma VCAM-1/VLA-4-dependent activation of NF-κB mediates chemoresistance. Blood 123, 2691–2702 (2014).
Information & Authors
Information
Published in
Classifications
Copyright
© 2021. Published under the PNAS license.
Data Availability
All study data are included in the main text and SI Appendix.
Submission history
Accepted: August 20, 2021
Published online: September 27, 2021
Published in issue: October 5, 2021
Keywords
Acknowledgments
We thank all members of the Tissue Engineering group for advice and support; Department of Biomedicine facilities for assistance and support; and C. Korn for data discussion and critical reading. This work was supported by the Swiss National Foundation Div 3 Grant 31003A-179259 (to I.M.) and National Centre of Competence in Research (NCCR) Molecular Systems Engineering Grant 51NF40-141825 (to I.M.), as well as by the Freenovation program (Novartis Foundation to P.E.B. and M.H.).
Notes
This article is a PNAS Direct Submission.
Authors
Competing Interests
The authors declare no competing interest.
Metrics & Citations
Metrics
Altmetrics
Citations
Cite this article
Culturing patient-derived malignant hematopoietic stem cells in engineered and fully humanized 3D niches, Proc. Natl. Acad. Sci. U.S.A.
118 (40) e2114227118,
https://doi.org/10.1073/pnas.2114227118
(2021).
Copied!
Copying failed.
Export the article citation data by selecting a format from the list below and clicking Export.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.