Hematopoietic stem cell transplantation (HSCT) has remained an unrivaled and promising treatment for a range of hematopoietic disorders and blood cancers over the past decades. During HSCT, circulating hematopoietic stem cells (HSCs) must migrate back to the bone marrow (BM) niche (homing), lodge within this niche, survive, and continually generate new blood and immune cells (engraftment). Inefficient homing or premature differentiation of transplanted HSCs can result in delayed hematopoietic recovery or graft failure [1,2]. Therefore, precise prediction of homing and engraftment efficiency, along with the long-term repopulation capacity of transplanted HSCs, is critical for ensuring successful HSCT outcomes.

Recent advances in gene editing technologies, such as CRISPR-based editors, have ushered in a new era for HSCT by expanding its therapeutic potential. HSCT is particularly advantageous for gene therapy. Unlike other gene therapy approaches, HSCs are ideally suited for ex vivo gene therapy because stem cells can be extracted from the body, edited during the culture period, and then conveniently reinfused [3,4]. However, recent studies indicate that genotoxic stress, including DNA damage responses that occur during the gene editing process, can impair cell viability. This may lead to insufficient engraftment and oligoclonal hematopoiesis, increasing the risk of infections and hematologic malignancies [5,6]. Therefore, more research and technology development are needed to make HSCT safer and more effective.

Among the multiple hurdles facing HSCT therapy, a major challenge is ensuring that transplanted HSCs efficiently home to the patient’s BM and engraft to reconstitute long-term hematopoiesis. Currently, there is a lack of effective model platforms capable of predicting long-term clinical outcomes, but this gap may be addressed by the development of BM organoids (BMOs)—3-dimensional (3D) constructs that closely mimic the BM microenvironment [7]. Traditional animal models, such as mouse xenografts or non-human primate (NHP) autologous models, have been invaluable in translational medicine but often fail to fully replicate the human-specific milieu and entail considerable expenses [8]. Human BMOs comprise multiple cell populations and structural elements organized to reproduce the native architecture and function of BM. With these characteristics, BMOs may enable researchers to examine the dynamic processes by which gene-edited cells—despite internal damage—interact with the BM environment and contribute to hematopoietic reconstitution, thus providing more accurate predictions of engraftment outcomes [7,9,10].

In this review, we investigate the key factors influencing HSC homing and engraftment, and discuss how gene editing may affect these processes. We then review recent advancements in BMO systems and their application as platforms for evaluating HSC-niche interactions. Our aim is to highlight how integrating gene-edited HSCs with state-of-the-art organoid models could overcome current limitations of HSCT, laying the groundwork for improved engraftment and personalized transplantation therapies.

Ethics statement: This study is a systematic review of previously published literature and was therefore exempt from institutional review board approval.

HSCs are classified into short-term HSCs (ST-HSCs) and long-term HSCs (LT-HSCs). ST-HSCs are proliferative and readily differentiate into blood lineage cells, while LT-HSCs remain in a quiescent state, rarely entering the cell cycle. Although LT-HSCs cycle and differentiate less frequently, this quiescence allows them to remain dormant, supporting lifelong participation in hematopoiesis [11–13]. In HSC transplantation, the efficient engraftment and preservation of LT-HSCs are essential for sustained hematopoietic reconstitution, making the enrichment and maintenance of their stemness a critical priority [14].

After infusion into the peripheral circulation, HSCs relocate to the BM niche through transendothelial migration—a process known as homing. Once in the marrow, HSCs lodge within a specialized microenvironment that supports their survival and the production of mature blood cells [15]. The BM niche can be subdivided into the perivascular niche, which promotes immediate HSC proliferation and differentiation, and the endosteal niche, which preserves dormant LT-HSCs. Each contains distinct cellular populations. The perivascular niche, located near sinusoidal blood vessels, comprises endothelial cells, LepR-expressing CXCL12-abundant reticular cells, Nestin⁺ BM mesenchymal stromal cells (BMSCs), dendritic cells, and adipocytes. These cells secrete factors such as CXCL12 (also known as SDF-1), stem cell factor, and vascular cell adhesion molecule-1 (VCAM-1), which facilitate HSC transmigration. The endosteal niche, lining the trabecular bone, contains osteoblasts, osteoclasts, Schwann cells, and associated stromal cells, and produces factors including angiopoietin-1, thrombopoietin, and osteopontin, which promote HSC quiescence [16]. These niche-organizing cellular components and their secreted factors play distinct and coordinated roles in orchestrating the balance required after transplantation.

During transplantation, circulating HSCs are ideally attracted by chemoattractants that facilitate transmigration through the BM microvasculature and proper lodging within the marrow [17]. After arrival, ST-HSCs and progenitor cells occupy the perivascular region, where they can rapidly proliferate and differentiate. In contrast, LT-HSCs lodge in the BM endosteal region, where they remain quiescent and dormant, ensuring continued hematopoiesis throughout the patient’s life [12,18]. If transplanted HSCs prematurely differentiate and lose their stemness, their repopulation capacity is exhausted, leading to graft failure and an inability to sustain long-term hematopoiesis. Therefore, it is crucial to secure a sufficient quantity and quality of engrafted cells, especially LT-HSCs, and maintain a niche environment conducive to their quiescence and viability after homing for successful HSCT outcomes.

Current gene editing techniques—including zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and the CRISPR/Cas9 system—enable precise correction of genetic mutations for treating various Mendelian diseases [34]. More recently, next-generation editors such as base editors (BEs) and prime editors (PEs) have been developed, allowingsingle-nucleotide changes or small insertions without generating double-strand DNA breaks [35].

ZFNs use engineered zinc-finger protein domains to recognize specific DNA triplets, fused to a FokI nuclease domain that introduces DNA cuts with high specificity [36]. However, their design is complex and costly for each new target. TALENs, although easier to design than ZFNs, still require labor-intensive protein engineering. The CRISPR/Cas9 system, by contrast, uses a guide RNA (gRNA) to direct Cas9 nuclease to complementary DNA target sites, where it induces DNA strand breaks. This approach is comparatively simple and cost-effective, as each new target only requires changing the gRNA sequence. For these reasons, CRISPR/Cas9 is now widely used as the “standard” gene editing platform [34,37]. One notable concern with CRISPR/Cas9 technology is the risk of off-target cleavage at sites with partial homology to the gRNA, potentially resulting in genotoxic damage [38].

Beyond nucleases, BEs enable direct conversion of one DNA base to another without creating double-strand breaks. Base editing allows precise correction of point mutations in HSCs with lower rates of insertions or deletions (indels) and potentially fewer DNA damage responses [39]. However, BEs are limited to certain types of single-base changes and have a narrow “editing window”—only a few nucleotides near the target site can be efficiently edited [40]. PEs use a modified Cas9 fused to a reverse transcriptase and a prime editing guide RNA (pegRNA) to insert new genetic information at the target site. This approach allows a broader range of genetic modifications and offers specialized options for precise editing with potentially fewer adverse effects [41], although PEs are currently less efficient in HSCs compared to standard CRISPR/Cas9 tools.

The delivery method for editing tools is also crucial. Editing components (nucleases and guide RNAs) can be delivered as DNA plasmids, RNA, or protein complexes. Common approaches include transfection of CRISPR ribonucleoprotein (RNP) complexes or transduction with viral vectors such as adeno-associated virus serotype 6 (AAV6). Among various methods, electroporation of RNPs has become a preferred approach for HSCs, as it enables transient presence of the editor, reducing off-target activity over time and avoiding insertional mutagenesis from integrating vectors [42,43]. However, electroporation can reduce cell viability. AAV6 vectors are used to deliver DNA donor templates for homology-directed repair (HDR) and offer high gene transfer efficiency in HSCs [44], but they introduce foreign DNA that may provoke immune responses [43]. Non-integrating lentiviral vectors, Sendai viral vectors, and novel lipid nanoparticles are also being investigated.

Despite these technical advances, gene editing in HSCs triggers cellular stress pathways that can impair engraftment. Introducing a double-strand break in HSCs activates the DNA damage response [45]. Edited HSCs show rapid activation of the p53 pathway, and excessive p53 activity can induce apoptosis or cell cycle arrest, depleting the pool of transplantable stem cells [2,46]. Although p53 is essential for maintaining genomic integrity, its activation may also eliminate many correctly edited HSCs, leading to increased apoptosis, delayed proliferation, and ultimately failure of hematopoietic repopulation. Thus, transient and target-specific p53 suppression during editing has been proposed to improve the yield of edited HSCs without permitting survival of cells with oncogenic mutations [2,47].

The p38 mitogen-activated protein kinase–ROS (MAPK-ROS) pathway is also reported to be activated during gene editing, driving HSCs out of quiescence and promoting differentiation or even senescence in LT-HSCs [48,49]. Depletion of LT-HSCs from such stress may lead to suboptimal or short-lived engraftment after transplantation, resulting in poor clinical outcomes. To address this, strategies such as the use of antioxidants or p38 inhibitors during ex vivo editing culture have been reported to preserve more primitive HSCs and enhance their repopulating capacity after editing [48,50].

Thus, researchers must optimize editing protocols to maximize on-target corrections while minimizing side effects. Looking ahead, gene-edited HSCs will likely be evaluated in BMO models or humanized mice to ensure they can home and engraft comparably to unedited HSCs. The next section explores how organoid technologies are being developed for these purposes.

A model that accurately recapitulates the human BM environment and the interactions between its cellular components is essential for understanding hematological physiology and related disorders. However, the complexity of the BM microenvironment poses significant challenges. Over recent decades, extensive research has focused on developing alternative BM models.

BM models can generally be categorized as 2-dimensional (2D) culture systems, animal models, and 3D culture platforms. Each model has distinct advantages and limitations, making careful selection critical for appropriate experimental validation and eventual clinical translation. Among animal models, murine models are most widely used due to extensive prior study and well-established protocols. Thus, most current knowledge about HSC homing and engraftment has been derived from mouse models. Murine models of hematological disorders include several subtypes: spontaneous models, xenograft models (cell-line or patient-derived), humanized mouse models, germline transgenic models, and conditional transgenic models. Humanized mouse models, for example, are developed by transplanting human HSCs into immunodeficient mice such as the NSG (NOD/SCID/IL2rγ-null) strain, resulting in a functional immune system that supports tumor–immune interactions and serves as a powerful platform for hematological malignancy research [68,69].

NHPs are also employed in preclinical HSCT studies, as their hematopoietic processes and BM environment more closely resemble those of humans. Compared to mice, NHPs have lifespans exceeding 30 years, permitting long-term tracking of transplanted HSC clones. They also exhibit lineage-specific differentiation patterns similar to human physiology, and their larger size enables clinically relevant autologous transplantation procedures with subsequent monitoring of multilineage reconstitution [70,71].

Nevertheless, animal models are limited by cross-species differences. For instance, mice have a much faster HSC turnover rate and distinct niche composition compared to humans [72]. Thus, human HSCs may not exhibit native behavior within mouse BM due to species-specific differences in homing receptor interactions and cytokine signaling. Moreover, the small size of murine bones limits their ability to fully model the 3D structure and cellular complexity of the human BM environment, such as the proportion of adipose tissue and the presence of larger bone trabeculae [73,74]. The use of NHPs is also constrained by high costs and ethical considerations. Therefore, there is a need for alternative preclinical models that more accurately replicate the function and structure of the human BM niche.

The primary objective of BMO systems is to recapitulate the complex microenvironment and cellular interactions of the BM niche in vitro. This includes not only HSCs and progenitors, but also multiple stromal cell types (such as mesenchymal stem cells, osteoblasts, adipocytes, and endothelial cells) as well as extracellular matrix (ECM) proteins, all organized within a trabecular bone structure [75]. To achieve this, diverse approaches have been developed to construct BMOs.

The BM niche is a complex microenvironment composed of hematopoietic and non-hematopoietic cells, including MSCs, osteoblasts, ECs, adipocytes, and ECM components [75]. Within this niche, the hematopoietic component plays a crucial role in regulating HSC self-renewal, differentiation, and quiescence. In addition to fully self-organized organoids, many researchers have developed defined 3D co-culture systems to model specific aspects of the BM niche. The endosteal niche, located near trabecular bone, preserves LT-HSCs for continuous hematopoietic replenishment, whereas the perivascular niche, near sinusoidal blood vessels, supports myeloid and lymphoid cell production [16]. To mimic the endosteal niche, a 3D co-culture system was established by cultivating hematopoietic progenitor cells (HPCs) with BM-derived MSCs (BM-MSCs) in a collagen-based scaffold. In this model, BM-MSCs provide osteoblast-like signals and an adhesive substrate within the collagen matrix, resembling the endosteal surface and supporting HPC self-renewal while preventing premature differentiation. Leisten et al. [93] demonstrated that HSC and MSC co-cultures maintain primitive CD34⁺CD38^– HSCs more effectively than standard liquid cultures. Numerous studies have shown that co-culture of HSCs and MSCs improves cell activation, significantly enhances engraftment success, and restores long-term multilineage potential [93–95]. However, increased cellular activity and proliferation in these co-cultures may also have adverse effects on long-term hematopoiesis following engraftment. For example, 3D co-culture of HSCs and BMSCs within a soft GelMA hydrogel (5%) has been shown to enhance HSC self-renewal, myelopoiesis, and lymphopoiesis [84]. The cellular source also matters: umbilical cord-derived MSCs promote rapid HSC proliferation and differentiation, making them less suitable for long-term HPC maintenance compared to BM-MSCs [93].

To address the perivascular niche, co-culture models incorporate endothelial cells. One approach involves seeding human BM-MSCs and human umbilical vein endothelial cells (HUVECs) onto a bio-derived bone scaffold to establish a microvascular network adjacent to stromal cells. Endothelial cells in this system induce lumen formation and release angiocrine factors that influence HSC behavior. For instance, HUVECs activate Wnt/β-catenin signaling in hBMSCs, leading to the upregulation of Notch ligands (Jagged-1, DLL-1, DLL-4) and subsequent activation of Notch signaling in HSCs [96]. This cascade promotes HSC self-renewal and expansion of primitive CD34⁺CD38^– cells [96,97].

BMOs significantly advance both basic and translational research. One major application is as alternative transplantation models. For instance, a BMO model incorporating MSCs and ECs has been used to study the migration and nesting of HSPCs and leukemic stem cells. HSPC migration into BMOs was observed, with homing regulated in part by the CXCL12–CXCR4 axis. BMOs have also been applied in leukemia research; when CD34⁺ leukemic blasts were seeded into 3-day-old BMOs and monitored for 96 hours, migration occurred more efficiently than proliferation [64]. To assess whether BMOs can replicate the tumor microenvironment of hematologic malignancies, patient-derived cells from multiple myeloma, acute myeloid leukemia, and chronic myeloid leukemia were transplanted into BMOs, and successful engraftment was observed. BMOs also enable drug screening on patient-specific cells. For example, treatment of a myelofibrosis BMO model with a TGF-β inhibitor and a BET bromodomain inhibitor effectively suppressed fibrosis, suggesting potential for drug discovery [9]. Genetic disease modeling is also feasible with BMOs. Frenz-Wiessner et al. [7] generated VPS45-deficient BMOs using CRISPR/Cas9-edited iPSC lines. These BMOs exhibited expansion of α-smooth muscle actin-positive myofibroblast-like stromal cells, mirroring the pathology of neutropenia and myelofibrosis. Similarly, a myelodysplastic syndrome (MDS) BMO model was established by Ren et al. [10]: CD34⁺ HSPCs isolated from MDS patients were transplanted into BMOs, and whole-exome sequencing showed that patient-derived HSPCs proliferated while retaining disease-associated mutations. Notably, CD34⁺ cells re-isolated from BMOs and subsequently transplanted into new BMOs retained partial hematopoietic function even after secondary transplantation. Collectively, these results highlight the utility of BMOs for studying HSC homing, disease modeling, gene editing, and drug screening—advancing research in hematopoiesis and leukemia.

In the context of HSCT and gene therapy, BMOs provide an ex vivo platform to evaluate gene-edited HSCs and novel transplantation strategies prior to clinical use. For example, BMOs can be used to test whether gene-edited HSCs exhibit improved homing and engraftment in a 3D marrow-like environment compared to unedited cells. Organoids structured with endothelial networks and stromal components offer a feasible platform for assessing homing capacity and the balance between HSC proliferation and differentiation after transplantation. Additionally, BMOs can be utilized to screen co-administrative materials (e.g., cytokines, small molecules, antibodies) that may improve engraftment. They allow for detailed observation of niche interactions and HSC dynamics within a controlled, human BM–like microenvironment.

Organoids are also employed as a platform for toxicology and pharmacology studies. BMOs provide a predictive, human-specific model to assess drug toxicity on the BM (myelotoxicity), outperforming animal and 2D cell culture models. BMOs can recapitulate toxic responses relevant to patients, serving as valuable preclinical testing tools. A recent review by Matsui and Shinozawa highlights how human organoids are becoming increasingly instrumental for predictive toxicology and reducing the reliance on animal testing [98].

BMO systems present several significant advantages as platforms for HSC research and therapeutic development. First, BMOs offer species-specific modeling, providing closer relevance to human BM physiology than traditional models. For example, while immunodeficient NSG mice display low engraftment efficiency of MDS HSPCs, Ren et al. [10] demonstrated successful engraftment of these cells in BMOs. Furthermore, organoid systems address ethical concerns associated with animal research and enable personalized medicine by supporting patient-specific disease modeling and therapeutic testing. Patient-derived organoids allow modeling of individual pathologies, facilitating the development and screening of tailored therapies. Additionally, organoids enable real-time 3D imaging, allowing researchers to monitor cellular behaviors, niche interactions, and responses to interventions [65,99]. The scalability and adaptability of organoid systems also support high-throughput drug screening and susceptibility assessment [100].

Despite ongoing advances in organoid technology, several limitations remain. Most notably, organoids incompletely recapitulate native tissue complexity. Standard organoid protocols often rely on Matrigel, which poses challenges in nutrient and oxygen delivery, resulting in insufficient vascularization. They also generally lack mechanical stimulation from blood flow, robust cell–cell interactions, and full cellular diversity [60–62]. These limitations make it difficult to completely replace animal models. In addition, organoid cultures require specialized equipment, advanced technical expertise, and expensive reagents, which can create operational challenges compared to conventional 2D cell cultures or established animal models. Batch-to-batch variability can also affect long-term maintenance and reproducibility [63]. Finally, current BMOs exhibit limited long-term stability, restricting their use in chronic or longitudinal studies [9]. To address these issues, next-generation organoid engineering should employ advanced microfluidic systems, novel biomaterials, innovative co-culture strategies, and state-of-the-art gene editing technologies.

This review has highlighted recent advances in gene editing and BMO technologies as alternative HSCT model systems. Effective homing and engraftment of HSCs are fundamental to successful transplantation, yet animal models are constrained by species-specific differences. BMOs represent a promising human-specific alternative, reducing ethical concerns and enabling personalized, patient-specific disease modeling. As discussed, multiple factors influence HSC homing and engraftment, and before HSCT strategies can be clinically adopted, robust validation for safety and efficacy is essential. Although gene editing offers the potential for precise genetic modifications, it also introduces challenges such as off-target effects and cellular stress. BMOs can serve as preclinical testbeds for gene-edited HSCs, allowing researchers to observe edited cell behavior within a complex, human-like marrow microenvironment.

To fully realize the potential of these innovations, continued efforts must focus on enhancing the precision of gene editing and minimizing off-target responses. In parallel, advancing BMO models through increased cellular complexity, improved functional maturation, and greater biomechanical fidelity is critical. The combination of refined gene editing with advanced organoid systems will drive personalized medicine forward, ultimately enabling safer and more effective HSCT strategies. In conclusion, the convergence of translational research in HSCT and organoid technologies holds great promise for improving transplant outcomes, minimizing complications, and expanding therapeutic opportunities for patients who currently lack effective treatment options.