Human erythroid models of sickle cell disease Authors: Deborah E. Daniels, Maria A. Lizarralde‐Iragorri, Fatima O. Oyawoye, David C. Rees, Jan Frayne, Sara El Hoss Published: 2026 ABSTRACT AbstractSickle cell disease (SCD) is a group of severe inherited red blood cell disorders causing a major global healthcare burden. Despite recent advances in gene therapy and stem cell transplantation, treatment options remain very limited, highlighting the urgent need for the development of new therapeutic approaches. Human erythroid model systems of SCD represent emerging tools for the discovery of disease‐modifying strategies through drug screening and mechanistic studies. This review describes the range of models available, including patient‐derived red blood cells, hematopoietic stem cells, induced pluripotent stem cells, and immortalized cell lines, and summarizes their relative advantages and disadvantages, as well as potential integration with more complex integrated co‐culture and organoid models. Overall, these cellular systems represent an expanding toolkit available to SCD researchers with the potential to reduce reliance on the use of animal models. INTRODUCTION Sickle cell disease (SCD) involves a group of conditions characterized by the presence of a gain of function mutation in the beta globin gene (HBB:c.20A>T, p.Glu7Val) which produces sickle hemoglobin (HbS). 1 , 2 The pathology in SCD results from the tendency of HbS to polymerize when deoxygenated, which damages red cells. This results in vaso‐occlusion and hemolysis, with a complex cascade of interrelated downstream events including anemia, tissue infarction, inflammation, hypoxemia, oxidative stress, hypercoagulability, vascular‐endothelial dysfunction, infection and organ damage. 3 SCD is characterized by episodes of acute illness with progressive organ damage, resulting in reduced life expectancy. In low‐income areas, the majority of people with SCD die in childhood. 4 Treatment options in SCD are very limited with hydroxyurea being the only effective disease‐modifying drug. Significant advances are being made in stem cell treatments; however, these remain restrictively expensive and unavailable to the majority of patients with SCD. 5 In SCD research, there has been a heavy reliance on the use of mouse models, in which human hemoglobins are expressed in mice. 6 Whilst these have been useful in understanding pathological processes, they have been less successful at identifying effective drugs in SCD. 5 Part of this problem relates to differences between mouse and human physiology, particularly regarding erythropoiesis and red cells. There has therefore been increasing interest in developing models of SCD using human red cells to complement mouse studies. The aim of this review is to describe human erythroid cell models currently available (summarized in Figure 1), with an assessment of their advantages and disadvantages, as well as models under development. A better understanding of these models should facilitate the development of new drugs as well as additional, improved models of SCD. Overview of the currently available human erythroid models of sickle cell disease (SCD). Directly patient‐derived models include red blood cells (RBCs) and hematopoietic stem and progenitor cells (HSPCs). Induced pluripotent stem cell models (iPSCs) are derived indirectly from patient somatic cells. Immortalized erythroid cell line models (imEry) are generated via CRISPR‐Cas9 genome editing of existing immortalized lines (BEL‐A and HUDEP‐2) to introduce SCD genotypes. Solid purple line represents self‐renewal; dashed red line depicts erythroid differentiation; and solid gray line represents induction of cell sickling for relevant models. OrthoE, orthochromatic erythroblast; ProE, proerythroblast; Retic, reticulocyte. CURRENT HUMAN SCD MODELS Cellular models using patient‐derived material Patient‐derived red blood cells (RBCs) remain the most direct ex vivo system for interrogating the biophysical consequences of HbS polymerization. To date, fully mature, enucleated RBCs cannot be robustly generated in vitro, and therefore freshly isolated erythrocytes from individuals with SCD provide an irreplaceable platform for assays of sickling, cation transport, rheology, membrane fragility, adhesion to endothelial cells, and drug screening for effects that manifest specifically in mature RBCs. Classic studies established the link between HbS polymerization, membrane damage, and abnormal adhesion using patient RBCs, 7 , 8 and this approach remains widely used in modern therapeutic screening and mechanistic studies. 2 , 9 , 10 However, these models are intrinsically limited by donor‐to‐donor variability, the need for repeated sample collection, and the inability to model erythroid differentiation or genetically manipulate cells to interrogate causal mechanisms. As a result, while patient RBCs accurately reflect end‐stage cellular pathology, they provide little insight into how these abnormalities arise during erythropoiesis. Patient‐derived cellular systems represent the most physiologically relevant platforms currently available to model SCD erythropoiesis, as these models capture the intrinsic heterogeneity of SCD, including the influence of many modifier genes. In addition, having been derived directly from SCD patients, these cells have the potential to retain the hallmarks of chronic inflammation, treatment exposure, and bone‐marrow stress, all of which profoundly shape erythroid differentiation and red cell function. Among these systems, hematopoietic stem and progenitor cells (HSPCs) isolated from SCD patients provide a powerful entry point to interrogate the developmental and molecular origins of the disease. 11 , 12 CD34+ cells isolated from peripheral blood, often readily available from exchange transfusion products without additional burden to patients, can be differentiated ex vivo into the erythroid lineage, enabling the study of erythroid differentiation, oxidative stress, inflammatory priming, and globin gene regulation in a patient‐specific genetic context. 12 Importantly, SCD HSPC‐derived erythroblasts recapitulate multiple hallmarks of pathological erythropoiesis, including delayed and ineffective terminal differentiation, increased apoptosis, heightened oxidative and proteotoxic stress, dysregulated globin synthesis, and aberrant inflammatory priming. 12 These phenotypes mirror the stressed erythropoietic state observed in the bone marrow of SCD patients and provide a mechanistic framework for understanding how HbS polymerization and inflammatory signaling converge to impair red cell production. Moreover, because these cultures retain patient‐specific genetic backgrounds, they capture clinically relevant inter‐individual variability in HbF levels, stress responses, and susceptibility to ineffective erythropoiesis. Primary HSPCs have been central to the successful translation of gene editing from a conceptual strategy to a clinically approved therapy for SCD. 13 Patient‐derived CD34+ HSPCs faithfully recapitulate the chromatin landscape, DNA repair pathways, and lineage commitment programs of the cells that ultimately reconstitute the patient's hematopoietic system. 14 , 15 As a result, they provided the only reliable platform in which the efficiency, precision, and durability of genome editing approaches, such as CRISPR‐Cas9–mediated disruption of BCL11A enhancers, 15 could be rigorously optimized and validated. Ex vivo editing of autologous SCD HSPCs made it possible to directly measure on‐target editing, off‐target risk, clonal fitness, and erythroid differentiation capacity before transplantation, thereby facilitating safe clinical translation. 16 These systems also revealed how edited stem cells maintain long‐term repopulating ability while giving rise to erythroblasts with stable fetal hemoglobin induction or corrected globin expression, providing the mechanistic foundation for durable therapeutic benefit in patients. In this way, primary HSPC platforms were not merely experimental tools, but the indispensable bridge that enabled gene therapy to move from proof‐of‐concept to a transformative, one‐time treatment now administered to individuals with SCD. As highlighted above, a major strength of patient‐derived HSPC systems is their suitability for genetic and pharmacological manipulation, particularly regarding the promotion of fetal hemoglobin synthesis. Lentiviral knockdown, CRISPR‐based gene editing, and small‐molecule modulation can be applied at defined stages of erythroid differentiation, enabling direct causal testing of candidate regulators of HbS polymerization, oxidative stress pathways, inflammasome activation, and stress erythropoiesis. 17 This makes it possible to move beyond correlative observations in patient samples and to establish mechanistic links between molecular pathways and downstream erythroid phenotypes in a fully human, disease‐relevant context. Chronic inflammation and cellular stress in SCD extend far beyond mature RBCs and profoundly reshape the HSPC compartment, altering lineage commitment, proliferative capacity, and responsiveness to therapeutic manipulation. 18 By combining single‐cell transcriptomics of SCD HSPCs with clinical data from a lentiviral gene‐therapy trial, it was shown that the most primitive stem and progenitor compartments are pre‐programmed toward heightened inflammatory and interferon signaling and skewed myeloid lineage commitment, before any ex vivo processing. These intrinsic alterations were associated with impaired engraftment and reduced contribution of genetically corrected cells following transplantation, providing a mechanistic link between SCD‐associated inflammation at the stem cell level and variability in therapeutic efficacy. 18 By demonstrating that the SCD bone‐marrow environment imprints long‐lasting molecular programs onto HSPCs, this study established hematopoietic stem cell dysfunction as an important component of SCD pathophysiology and highlights why primary HSPC models are indispensable for understanding both disease biology and response to advanced therapies. Nonetheless, HSPC‐based systems also have significant constraints, including limited proliferative capacity, restricted access to bone‐marrow–derived cells, and substantial donor‐to‐donor variability, with different genetic backgrounds and thus genetic modifiers of disease severity, including but not solely due to modifiers of γ‐globin levels. 19 Moreover, experiments using gene‐edited cells are typically performed with heterogeneous cell populations containing unedited and variably edited cells, as it is not possible to isolate and grow pure populations of cells with the required edit, complicating data interpretation and contributing to inter‐experiment variability. Finally, differentiation protocols only generate cells up to the reticulocyte stage, and the lack of standardization across laboratories further limits cross‐study comparability. Despite these challenges, patient‐derived HSPCs offer a valuable bridge between molecular perturbation and disease‐relevant erythroid biology, making them indispensable for mechanistic and translational SCD research. Induced pluripotent stem cells Induced pluripotent stem cells (iPSCs) are generated from somatic patient tissue via overexpression of transcription factors that stimulate de‐differentiation to pluripotency. 20 Upon induction of haematopoietic differentiation using various culture systems, iPSCs can provide a near‐infinite source of HSPCs with the largely intact genetic background of the individual they were sourced from, although the epigenome is influenced by the tissue of origin 21 and the reprogramming process itself. 22 While several studies have successfully generated iPSCs from SCD individuals, full phenotypic modeling of the disease is prevented by several factors, most notably the predominantly embryonic/fetal globin expression pattern of iPSC‐derived erythroid cells precluding βS polymerization and induction of sickling under hypoxia. 23 , 24 Further challenges arise from the complex, multi‐stage culture systems typically required to perform hematopoietic and erythroid differentiation from iPSCs which, together with variability between protocols and iPSC line performance, hinders reproducibility. 25 , 26 In addition, incomplete erythroid maturation remains a barrier to comprehensive disease modeling from iPSC cultures, with limited enucleation restricting the production of reticulocytes for downstream analyses, a feature that impedes their use in SCD research. 27 While Kobari et al. reported some enucleation in iPSC‐derived βS/βS erythroid cells, high HbF levels precluded cell sickling under hypoxia or laminin adherence under capillary flow. 23 A further study reported 6%–10% enucleation in iPSC erythroid cultures; however, these reticulocytes were not subjected to sickling or flow adhesion assays. 24 Due to their inability to recapitulate the major pathological features of SCD, use of SCD iPSCs has instead focused on investigating gene correction strategies. This includes approaches utilizing engineered zinc finger nucleases 28 , 29 or CRISPR‐Cas9 gene editing. 24 , 30 More recently, some studies have demonstrated success in inducing limited HbS production from SCD iPSC‐derived erythroid cells. Haro‐Mora et al. reported generation of erythroid cells with 2% βS‐globin of total globin by RP‐HPLC, also demonstrating CRISPR‐Cas9–mediated gene correction of these cells to WT β‐globin. 30 Similarly, Conway et al. reported detectable HbS levels by HPLC, with some hypoxia‐induced HbS polymerization visible by electron microscopy. 31 While this represents promise for the future, the level of HbS production in iPSC‐derived SCD erythroid cells is currently not sufficient to reflect that seen in patients, preventing phenotypic disease manifestation and, in turn, negating their application in drug screening assays and mechanistic studies. In addition, reports of karyotypic abnormalities should be taken into account when considering iPSC‐based models for gene editing studies. 32 However, ongoing improvements to iPSC culture systems show promise in boosting erythroblast enucleation rates 33 , 34 and adult globin production, 35 , 36 indicating potential for a more prominent role for iPSCs in future SCD modeling. Immortalized cell lines Immortalized erythroid progenitor cell lines, most notably the cord blood–derived cell line HUDEP‐2 37 and the adult bone marrow–derived line BEL‐A 38 have emerged as valuable tools for the provision of a sustainable supply of erythroid cells ex vivo. These lines enable researchers to overcome the donor‐to‐donor variability and limited scalability associated with primary cultures, while offering an adult globin expression profile and improved erythroid differentiation compared to iPSC‐derived erythroid cells, together with a shorter and simpler culture regimen. Another advantage of immortalized erythroid lines lies in their amenability to precise genome manipulation, making them well suited to the investigation of mechanistic questions and evaluation of gene correction or gene modulation strategies. Moreover, the modified lines can be banked, providing a sustainable supply for downstream applications. In recent studies, both HUDEP‐2 and BEL‐A have been used as founder lines for the generation of SCD model systems via CRISPR‐Cas9 genome editing to introduce the homozygous E6V sickle cell anemia (SCA) mutation. Demirci et al. successfully generated a sickle HUDEP‐2 (sHUDEP‐2) cell line, which showed robust HbS production and >40% morphological sickling in the late‐stage erythroid cells produced from differentiation cultures (majority orthochromatic erythroblasts). 39 The authors went on to demonstrate dose‐dependent anti‐sickling activity of lentiviral βT87Q‐globin addition, exhibiting the utility of this technology for evaluating SCD therapeutic strategies. Similarly, the sHUDEP‐2 line has been used to demonstrate efficient one‐time correction of the SCA mutation in the endogenous βS‐globin gene via an all‐in‐one non‐integrating lentiviral system encoding Cas9, guide RNA and donor DNA, achieving up to 42% correction at protein level. 40 However, the use of sHUDEP‐2 to study ineffective erythropoiesis in SCD revealed contrasting results to those observed in primary SCA erythroid cell cultures—while primary cell‐derived SCA erythroblasts exhibited elevated ROS and markedly reduced GATA1 protein during differentiation, this was not observed in the sHUDEP‐2 cell line cultures. 12 This suggests that secondary effects from a pathological bone marrow microenvironment in SCD patients may be the main driver of ineffective erythropoiesis in SCD, as opposed to manifesting from an erythroid‐intrinsic property. 12 In 2025, Daniels et al. reported generation of a SCA BEL‐A line with validated HbS production and confirmed sickling phenotype in both nucleated orthochromatic erythroblasts and reticulocytes. 41 The line also demonstrated decreased sickling activity in response to increased γ‐globin production, 41 in line with the known anti‐sickling effect of HbF. 42 While a degree of ineffective erythropoiesis was observed in the SCA BEL‐A line as demonstrated by flow cytometry analysis, this was reduced compared to that reported in SCA primary cell cultures. 12 Cell line approaches to modeling SCD offer several advantages, including the provision of a genetically tractable and easily modifiable platform to study the involvement of individual proteins in sickling modulation and downstream events. In addition, short differentiation protocols allow for the screening of small molecules that require cell exposure during erythroid development to observe a potentially anti‐sickling effect, for example via cell proteome alterations (rather than being applied to mature RBCs derived from SCD individuals). Future integration with hypoxia live cell imaging and analysis pipelines would enable high‐throughput screening of such drug candidates, including monitoring of phenotypic changes throughout differentiation. As with primary cell cultures, one limitation of cell line–derived erythroid cells is their inability to mature from the reticulocyte to the erythrocyte stage, at least under current protocols. While cell line studies have clearly demonstrated the capacity of the reticulocytes (and even nucleated orthochromatic erythroblasts) to sickle, 39 , 41 differences in the biophysical and cytoskeletal properties compared to mature RBCs lead to altered sickling kinetics, increased rigidity and decreased propensity to sickling reversal. 43 As a result, where possible it will be important to test any findings from cultured reticulocytes in patient‐derived RBCs to assess transferability. Another limitation of cell line model systems is the suboptimal enucleation efficiency compared to primary cultures. 44 However, at least in the case of SCA BEL‐A, sufficient reticulocytes can be obtained via filtration to perform cell sickling assays under hypoxia from relatively small culture volumes (~20 mL) (Figure 2). As reactivation of gamma globin expression upon gene editing is a possibility in both cell lines, 39 , 41 appropriate screening of subclones or cells in long‐term culture is also an important consideration for researchers due to the phenotypic modifying effects of increased HbF. In addition, while the BEL‐A cell line has a verified increased HbF response to hydroxyurea 44 , 45 reports for HUDEP‐2 are variable, with some groups reporting a positive HU response 46 , 47 and others reporting no response, 44 , 48 , 49 findings which should be taken into account when using such lines for the screening of γ‐globin inducing compounds. Finally, it is important to consider the previously described karyotypic abnormalities in HUDEP‐2 and BEL‐A, 44 , 50 , 51 as well as any potential clonal variation, especially for downstream gene editing applications (although notably no changes in the globin locus containing chromosomes have been reported). Despite these factors, cell line model systems of SCD present an attractive option for studying the underlying disease pathology and testing novel therapeutic approaches in a sustainable and scalable cellular system. A comparative summary of the currently available human erythroid models of SCD discussed is provided in Table 1. Hypoxia‐induced sickling of cultured reticulocytes. (A) Schematic of workflow for performing sickling assays from cultured erythroid cells. (B) Example images from filtered immortalized erythroid cell (imEry) cultures. WT and sickle cell anemia (SCA) BEL‐A 41 cells were differentiated for 8 days in culture, before being filtered through a 5 µm pore to remove nucleated orthochromatic erythroblasts. Filtered reticulocytes were resuspended in Hemox solution (25 mM HEPES, 135 mM NaCl, 5 mM KCl, 0.9 mM CaCl2, 0.5 mM MgCl2, 9 µm FeCl2.4H2O, and 5 mM glucose; pH 7.4) and exposed to 0% O2 (95% N2, 5% CO2) for 4 h at 37°C. Images were acquired using a Cell Discoverer 7 environment‐controlled microscopy system. Red dots indicate sickled reticulocytes. HSPC, hematopoietic stem and progenitor cell; OrthoE, orthochromatic erythroblast; Retic, reticulocyte. Table 1Summary of currently available human erythroid models of sickle cell disease (SCD).Model approachComplexity of culture systemEase of genetic manipulationVerified response to HU?Key advantagesKey limitationsPatient‐derived RBCsN/AN/AN/AEase of availability; mature RBCsNo genetic manipulation or HbF induction possiblePatient‐derived HSPCsModerateLow‐moderateYesModeling of erythroid differentiation from CD34+ stage; cells have been exposed to SCD BM; high enucleation rate compared to cell linesReduced ease of genetic manipulation compared to cell lines and iPSCs; donor‐to‐donor variability; no final maturation to RBCsiPSCsHighHighN/ASustainable supply of cells; ease of genetic manipulation; modeling of erythroid differentiation; can generate from SCD patient somatic cellsComplex and lengthy culture systems; lack of adult globin production; limited enucleation; no final maturation to RBCs; potential for karyotypic abnormalitiesImmortalized linesLow‐moderateHighYes (variable in the case of HUDEP‐2)Sustainable supply of cells; ease of genetic manipulation; consistent genetic background; modeling of erythroid differentiation from ProE stageReduced enucleation compared to primary cells; no final maturation to RBCs; karyotypic abnormalitiesAbbreviations: BM, bone marrow; HSPC, hematopoietic stem and progenitor cell; iPSC, induced pluripotent stem cell; ProE, proerythroblast; RBC, red blood cell; SCD, sickle cell disease.John Wiley & Sons, Ltd. Summary of currently available human erythroid models of sickle cell disease (SCD). Abbreviations: BM, bone marrow; HSPC, hematopoietic stem and progenitor cell; iPSC, induced pluripotent stem cell; ProE, proerythroblast; RBC, red blood cell; SCD, sickle cell disease. HUMAN SCD MODELS UNDER DEVELOPMENT The growing recognition that SCD is driven not only by intrinsic red cell defects but also by pathological interactions between erythroid cells, the vascular endothelium, immune cells, and the bone‐marrow microenvironment has fueled the development of increasingly complex in vitro model systems. 52 , 53 Co‐culture platforms and organoid‐based approaches aim to move beyond isolated erythroid cultures by reconstructing key cellular and biophysical components of the SCD niche, thereby enabling the study of multicellular interactions that shape hemolysis, vaso‐occlusion, inflammation, and ineffective erythropoiesis. Integrated microfluidics co‐culture systems Microfluidic systems incorporating endothelial cells, immune cells, and flowing RBCs have provided powerful tools to model SCD‐specific microvascular pathology. 54 These platforms enable the controlled recreation of microvascular shear stress, hypoxia‐reoxygenation cycles, and inflammatory activation, allowing real‐time visualization of HbS polymerization, red cell sickling, adhesion to the endothelium, and occlusive events under physiologically relevant flow conditions.55, 56, 57 Integration of endothelial and immune compartments further permits investigation of how leukocytes, platelets, and inflammatory cytokines amplify vaso‐occlusion and vascular injury, a central feature of SCD pathophysiology. 58 In SCD, despite these strengths, microvasculature‐on‐a‐chip systems remain largely restricted to modeling events occurring in the circulation. They typically rely on mature RBCs or reticulocytes introduced into the device and therefore do not capture the upstream processes of erythroid differentiation, stress erythropoiesis, or ineffective maturation that originate in the bone marrow. However, microfluidic technologies still hold substantial promise for the future of SCD research beyond their current use in modeling vascular occlusion. By integrating controlled fluid flow with defined oxygen gradients, together with the range of cell types present in the bone marrow, including patient‐derived HSPCs, next‐generation microfluidic platforms could be engineered to mimic key features of the bone‐marrow niche in which sickle erythropoiesis occurs. Such systems would make it possible to study how hypoxia, inflammatory cytokines, and cell–cell interactions within erythroblastic islands shape erythroid differentiation, stress responses, and HbS polymerization in real time. In the context of SCD, where chronic inflammation and hypoxic stress profoundly alter hematopoietic stem cell function and erythroid maturation, marrow‐on‐a‐chip models 59 could provide an unprecedented window into the earliest cellular events driving ineffective erythropoiesis and hemolysis. If successfully developed, these platforms would bridge the gap between erythroid cultures and the in vivo aspect of the disease, offering a powerful and scalable framework for mechanistic studies and therapeutic screening in a physiologically relevant microenvironment. Figure 3 illustrates a proposed next‐generation microfluidic platform designed to integrate bone marrow erythropoiesis, inflammatory activation, and downstream vaso‐occlusive events within a single controllable system. Conceptual framework for an integrated bone marrow‐to‐vaso‐occlusion microfluidic platform in sickle cell disease (SCD). The proposed system combines a 3D stromal bone marrow hydrogel containing mesenchymal stromal cells (MSCs) and patient‐derived CD34+ hematopoietic stem and progenitor cells (HSPCs) with localized macrophage‐rich erythroblastic islands and granulopoiesis‐supportive regions to recapitulate key features of the human marrow niche. Controlled hypoxia‐reoxygenation and inflammatory stimulation would enable investigation of stress erythropoiesis, HbS polymerization during erythroid differentiation, oxidative stress responses, neutrophil generation, and cell–cell interactions within a physiologically relevant microenvironment. The marrow compartment connects to an endothelial sinusoid that regulates hematopoietic cell egress and provides a controlled vascular interface prior to entry into a downstream endothelialized vaso‐occlusion (VOC) module. This VOC network incorporates bifurcations and stenotic segments to model shear‐dependent adhesion, aggregate formation, and inflammatory vaso‐occlusion under controlled flow conditions. Together, this envisioned platform would provide a unified framework to investigate how hypoxia, inflammation, and bone marrow–derived cellular alterations contribute to adhesion‐mediated microvascular obstruction in SCD. Organoid models Three‐dimensional (3D) organoid and tissue‐engineered systems (reviewed by Silva‐Pedrosa et al. 60 ) are an emerging frontier for disease modeling and could represent a promising approach for modeling SCD in a more integrated and physiologically relevant manner. By combining hematopoietic cells with stromal, endothelial, and immune components within biomimetic scaffolds or hydrogels, these approaches could recreate aspects of the bone‐marrow niche or secondary lymphoid organs such as the spleen, where sickle red cells are filtered and destroyed. In principle, such systems would allow simultaneous modeling of erythroid differentiation, inflammatory priming, and cell–cell interactions that contribute to SCD pathophysiology. Indeed, advances in the development of bone marrow niche organoid models in particular 61 , 62 show promise in these cutting‐edge platforms for potential adaption into SCD modeling. Nevertheless, organoid‐based SCD models remain in early developmental stages. Key challenges include controlling cellular composition, achieving reproducible erythroid maturation and enucleation, maintaining long‐term culture stability, and scaling these systems for mechanistic studies or drug testing. While their potential to bridge erythropoiesis, vascular pathology, and immune dysregulation is substantial, significant technical and biological hurdles must be overcome before 3D organoid platforms can be routinely used to model SCD pathophysiology or therapeutic responses. METHODOLOGY FOR ASSESSING CELL SICKLING EX VIVO Development of methods to both successfully induce and quantify the degree of cell sickling in erythrocytes and erythroid cells, either those produced ex vivo or obtained from SCD individuals, is critical for enabling a phenotypic readout for these cellular model systems, including in response to pharmaceutical and gene editing interventions. This section will summarize current approaches to achieving hypoxia‐induced sickling ex vivo, as well as detection and quantification methods for the identification of normal versus sickled cells (Table 2). Table 2Summary of sickle cell assessment methodology.MethodPrinciple/readoutLive cell compatibleThroughputApplicationsLight microscopy (live imaging)Direct visualization of cell morphology during oxygenation/deoxygenation cyclesYesModerate‐highKinetic profiling of sicklingLight microscopy (fixed cells)End‐point morphological assessment of chemically fixed cellsNoModerate‐highEvaluation of cell sickling at a fixed timepointScanning electron microscopy (SEM) and x‐ray tomographyHigh‐resolution surface morphology imagingNoLowDetailed structural assessment of sicklingImaging flow cytometryHigh‐speed single‐cell imaging morphology‐based classificationNoHighPopulation‐level sickling quantification at a fixed timepointEktacytometryLaser diffraction–based measurement of RBC deformability under shear stressYesModerateAssessment of oxygen‐dependent deformability and sickling‐related rheological changesJohn Wiley & Sons, Ltd. Summary of sickle cell assessment methodology. Induction of cell sickling As HbS fiber polymerization is hypoxia‐dependent, inducing low oxygen conditions in the immediate cellular environment is key to triggering the nucleation mechanism for formation and elongation of these hemoglobin bundles, leading to RBC distortion into the characteristic sickle shape. Arguably the most cost‐effective and least technically involved approach to decreasing the local oxygen concentration is the use of sodium metabisulphite, an oxygen scavenging compound that converts oxygen gas to sulfate ions. As a result, this approach is commonly employed in point of care screening tests to induce cell deoxygenation and dehydration. 63 , 64 Other chemical approaches to generating a hypoxic environment include the use of protocatechuic acid and protocatechuate 3,4‐dioxygenase enzyme, which has been shown to allow reproducible sample deoxygenation over seconds to hours. 65 However, these oxygen scavenging–based approaches to hypoxia induction are hampered by a lack of fine control and adaptability, for example, to allow for exposure to multiple deoxygenation and reoxygenation cycles. To address this, many research groups utilize composition‐controlled gas chambers, often coupled with microscopy or other techniques to allow assessment of sickling in real time.10, 11, 66, 67, 68 Polymerization can also be induced by photodissociation of the hemoglobin‐carbon monoxide complex, resulting in rapid fiber formation. 69 Alternatively, cells can be fixed either under hypoxia using a glove box set‐up 70 or a deoxygenated buffer system 71 and even under normoxia to reveal irreversibly sickle cells, for later analysis. 41 Sickle cell detection and quantification methods Light microscopy is a popular approach for assessing sickle cell morphology, either dynamically within a live imaging system or by assessing fixed cells from end‐point assays. Live imaging systems allow continuous imaging over several hours in a multi‐well plate format followed by extraction of sickling profiles from each image sequence to produce frame‐by‐frame counts of normal and sickled cells. 10 , 68 With this approach, multiple cycles of oxygenation and deoxygenation can be applied to the cells to mimic the circulatory environment. Alternatively, quantification of sickling can be performed from images of fixed cells, as mentioned previously.2, 72, 73, 74 In either case, automated image analysis and quantification are possible by leveraging a supervised classification approach to “train” software to identify sickle cells. 2 , 10 , 68 Furthermore, higher resolution visualization of cell sickling can be achieved by scanning electron microscopy 75 , 76 or x‐ray tomography, 71 enabling detailed structural characterization of the morphological effects of anti‐sickling compounds, albeit at much lower throughput compared to other methods. In recent years, imaging flow cytometry has emerged as an alternative approach for the detection of sickled cells.70, 77, 78, 79 Imaging flow cytometry offers advantages over microscopy techniques in terms of high throughput capacity, enabling detection of thousands of cells per second compared to hundreds of cells per slide in an imaging well. However, as current approaches are restricted to data acquisition under normoxia, cell fixation is required, therefore preventing imaging of live cells. In addition, the nature of the technique means that tracking of individual cells over time and direct assessment of sickling kinetics is not possible. Beyond morphological analysis, alterations in cell deformability provide another potential read‐out for detecting and quantifying sickle cells. Ektacytometry is a commonly used laser diffraction–based biophysical technique that measures RBC deformability, 80 allowing differentiation between normal and sickle RBCs due to the increased membrane rigidity induced by polymerization of deoxy‐HbS. 81 By introduction of a controlled oxygen tension gradient during deformability measurements, cells can be exposed to a physiologically relevant oxygenation/deoxygenation cycle, 82 , 83 similarly to that seen in live imaging approaches. This oxygen gradient ektacytometry method has been used as a platform to evaluate anti‐sickling agents, 82 assess differences in cell deformability between different SCD genotypes 82 , 84 and evaluate different treatment regimens. 84 However, a more recent study found that oxygen gradient ektacytometry was not predictive of vaso‐occlusive episodes in children with SCD, 85 therefore the clinical predictive value of this methodology remains uncertain. When considering drug screening approaches for SCD, high‐throughput compatibility of these cell‐based assays is essential for enabling automated imaging and analysis for large numbers of drug candidates. While drug repurposing library screens have been demonstrated using fixed cell imaging 2 and live cell imaging of patient RBCs in 384‐well plates, 10 future assays could benefit from integration with HSPC and immortalized cell line cultures to enable screening for compounds where their efficacy requires exposure to the differentiating erythroblasts, as would be the case physiologically. In addition, combining high‐throughput approaches with microvasculature‐on‐a‐chip models or other more complex multi‐cell systems would enable drug screening in more physiologically relevant systems, with the hope of increasing clinical translation potential. FUTURE PERSPECTIVES This review summarizes the currently available human erythroid models of SCD and highlights their individual advantages and disadvantages, from RBCs through HSPC, iPSC, and cell line approaches. These relatively simple cell‐based model systems offer valuable tools for human‐specific disease modeling and show promise for future integration into more complex modeling platforms for more systemic aspects of SCD. While patient‐derived RBCs provide the only source of fully mature erythroid cells at present and, as such, provide unique advantages, erythroid models that enable recapitulation of erythroid maturation ex vivo (including from HSPCs and erythroid cell lines) allow the study of SCD erythroid differentiation and, in turn, the potential for more comprehensive disease modeling. Indeed, future integration with inflammatory agents and macrophage co‐culture systems might generate more representative model systems of SCD and enable unraveling of physiologically relevant factors affecting the sickle bone marrow microenvironment. For example, improving understanding of the mechanisms of inflammatory priming in SCD may inform new approaches for addressing the high engraftment rejection rates seen in SCD patients, 18 a key barrier to overcome for consistent outcomes in gene‐therapy approaches. The observation that orthochromatic erythroblasts are capable of hypoxia‐induced sickling 11 , 12 , 41 , 86 raises the question of the potential influence of these sickled erythroblasts within the erythroblastic islands and wider bone marrow niche. Ex vivo reconstitution of the central macrophage and developing erythroblasts provides a possible approach for investigating these interactions and their potential implication on the dysregulated bone marrow microenvironment in SCD, 87 allowing interrogation of the macrophage‐erythroid cross‐talk in SCD. Beyond the SS genotype, modeling of other genotypes within the SCD family will be key to interrogating similarities and differences in the molecular mechanism of disease and therapeutic response. In the case of the less studied HbSC disease, it is still not clearly established why HbS polymerizes in red cells containing only about 50% HbS. Additionally, clinical features of people with HbSC are largely unexplained, including an increased rate of splenomegaly and retinopathy. 88 Combining HbSC patient‐derived cells or cell lines with microvasculature modeling may yield mechanistic explanations to these clinical observations and, in turn, inform combative therapeutic strategies. SCD model lines derived from BEL‐A and HUDEP‐2 have the advantage of a uniform genetic background, enabling reproducibility and effects specific to the SS mutation to be analyzed. This consistency also enables drug evaluation without confounding variation from patient‐specific genetic differences. In addition, these models provide a controlled platform for introducing and studying candidate disease‐modifying variants. However, as SCD disease severity is modulated by different genetic backgrounds, it would be valuable to also directly create lines from patient HSPCs, for example representing the main βS haplotypes: Senegal, Bantu (or Central African Republic), Benin and Arab‐Indian, which vary in severity partly due to genetic modifiers of HbF levels, as well as HbSC, HbSD, HbS/beta thalassemia variants and HbS/α‐thalassemia trait variants. Such a catalog would enable broader validation of therapeutic efficacy across genotypes, helping determine whether candidate treatments are broadly effective or best suited to specific patient subgroups. Currently available cell sickling assay approaches present a range of complementary methods that can be used to assess the therapeutic impact of gene editing or pharmaceutical interventions across SCD model systems. The balance of advantages and limitations of each approach should be considered when determining which is most appropriate for a particular study, for example weighing up the capability to live image under controlled hypoxia for assessment of sickling kinetics against the more detailed structural assessment possible when using fixed samples. Beyond whole cell sickling assay methods, biophysical studies that directly assess HbS polymerization (reviewed by Eaton 89 ) also play an important role, especially when determining the mechanistic basis of compounds with a direct anti‐sickling effect. In any case, there is a need for further development of standardized and reproducible methods based on accessible platforms that can be readily adopted by the SCD research community. CONCLUSION Human erythroid model systems of SCD provide valuable platforms for investigation of molecular mechanisms of the disease and for screening of new therapeutic strategies. Each approach offers different advantages, including fully mature erythroid cell dynamics (patient RBCs), the ability to recapitulate erythropoiesis from progenitors originating from a SCD environment (patient HSPCs), and ease of genetic manipulation with sustainable supply (erythroid cell lines). Integration of these models with other cell types and/or 3D scaffolds brings the potential of added complexity and, therefore, accuracy of recapitulating the human disease phenotype ex vivo. Ultimately, such models provide an important step on the journey toward the replacement, reduction, and refinement of the use of animal models in SCD research. AUTHOR CONTRIBUTIONS Deborah E. Daniels: Conceptualization; writing—original draft; writing—review and editing. Maria A. Lizarralde‐Iragorri: Writing—original draft; writing—review and editing. Fatima O. Oyawoye: Generation of sickling assay data. David C. Rees: Conceptualization; writing—original draft; writing—review and editing. Jan Frayne: Conceptualization; writing—review and editing. Sara El Hoss: Conceptualization; writing—original draft; writing—review and editing. CONFLICT OF INTEREST STATEMENT The authors declare no conflicts of interest. ETHICS STATEMENT This article is a review of previously published literature and does not involve human participants or animals. Therefore, ethics committee approval was not required. FUNDING F.O.O. is supported by a Faculty of Health and Life Sciences International University of Bristol PhD studentship. S.E.H. and M.A.L.‐I. are supported by the GR‐Ex, France 2030 program through the Idex Université Paris Cité (ANR‐018‐0001_GR‐Ex).