Retinoic acid as a stage-specific modulator of hematopoietic lineage fate from human pluripotent stem cells Authors: Seo-Hyun An, Ji-Yeon Kim, Ying-Ying Mao, Xing Zhen, Gyu-Seo Bae, Kyungjun Uh, Young-Hyun Kim, Chan Young Park, Jong-Hee Lee Published: 2026 ABSTRACT Retinoic acid (RA), a bioactive metabolite of vitamin A, plays roles in early embryogenesis and hematopoietic development. However, its precise function in directing the hematopoietic lineage outcomes of human pluripotent stem cells (hPSCs) remains unclear. Here, we uncovered a distinct, stage-specific role for RA as a lineage-specifying modulator during late-stage hematopoietic differentiation, rather than as a promoter of hematopoietic progenitor generation. Using a stepwise hPSC differentiation system, we demonstrated that RA exerted minimal or inhibitory effects when applied during early mesoderm or hemogenic endothelial stages. In contrast, RA treatment during days 13–15 significantly enhanced progenitor maturation, proliferation, and functional output. Notably, RA acted cooperatively with external cytokines to modulate lineage fate. In the presence of erythropoietin (EPO), RA strongly promoted erythroid differentiation by activating EPOR signaling and upregulating erythroid transcriptional programs, including GATA1, KLF1, and globin gene expression. Conversely, under GM-CSF/M-CSF stimulation, RA biased progenitor differentiation toward macrophages, consistent with its role as an amplifier of the prevailing cytokine-directed lineage fate rather than an independent suppressor of erythropoiesis. These effects were highly dose- and context-dependent, with low-dose RA optimally enhancing lineage bias without cytotoxicity. Importantly, RA modulated the transcriptional and proliferative dynamics of committed progenitors. Taken together, our findings reveal a previously unrecognized role of RA as a versatile and tunable modulator of hematopoietic lineage fate that offers a novel strategy for in vitro blood cell engineering. This study advances approaches for lineage-specific blood production relevant to disease modeling, drug screening, and regenerative medicine.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03012-w. Introduction In vitro generation of hematopoietic cells from human pluripotent stem cells (hPSCs) has emerged as a promising approach for disease modeling, hematopoietic research, and regenerative therapies [1]. Unlike donor-derived sources, hPSCs offer a renewable and genetically customizable cell source, making them ideal for exploring hematopoietic ontogeny and producing specific blood cell types on demand [2]. While recent landmark work has successfully demonstrated long-term multilineage engraftment from hPSC-derived hematopoietic cells [3], the stage-specific signaling mechanisms required to efficiently direct hPSC-derived progenitors toward desired downstream lineages, such as erythroid and myeloid cells, and to promote their robust maturation remain to be fully resolved. Addressing these regulatory mechanisms will be critical for improving the yield, purity, and functional properties of hPSC-derived blood products for disease modeling and translational applications. Hematopoietic specification is a complex, multistep process that unfolds in a spatially and temporally coordinated manner during embryogenesis [4, 5]. It begins with mesodermal patterning and proceeds through the formation of hemogenic endothelium, followed by the endothelial-to-hematopoietic transition (EHT), which gives rise to multipotent progenitors [6]. These progenitors then undergo lineage commitment under the influence of intrinsic transcriptional programs and extrinsic signals from the surrounding microenvironment [4, 7, 8]. Key developmental pathways, such as Wnt, BMP4, Activin A, and Notch, have been shown to guide early mesoderm and hematoendothelial commitment in vitro [9], while hematopoietic cytokines, including VEGF, SCF, TPO, Flt3L, IL-3, and IL-6, support the emergence and expansion of progenitor populations [10]. These signals converge on transcriptional networks centered on factors such as RUNX1, GATA2, and TAL1, which are required for hematopoietic emergence from hemogenic endothelium [11–13]. Despite these insights, the precise regulatory mechanisms that govern late-stage lineage commitment remain poorly understood. Retinoic acid (RA), a metabolite of vitamin A, is a well-characterized morphogen with a broad influence on embryonic patterning, germ layer formation, and organogenesis [14, 15]. It functions through ligand-activating nuclear receptors, RA receptors (RARs) and rexinoid X receptors (RXRs), which act as transcriptional regulators [16, 17]. In early development, RA signaling has been implicated in both promoting and restricting mesodermal differentiation depending on the context, dosage, and timing of exposure [18, 19]. In hematopoiesis, RA plays multifaceted roles including contribution to the regulation of HSC quiescence in the adult bone marrow niche, influencing HOX gene expression during definitive hematopoietic development, and modulating lineage outcomes [20–22]. Notably, RA has also been shown to cooperate with hematopoietic cytokines such as erythropoietin (EPO) or granulocyte macrophage-colony stimulating factor (GM-CSF) to influence erythroid and myeloid differentiation, respectively [23–26]. Although the roles of RA in early mesoderm induction, neuroectodermal specification, and primitive hematopoiesis have been studied extensively, far less is known about how RA influences the fate decisions of committed hematopoietic progenitors during the terminal stages of differentiating from hPSCs. This distinction is important because hPSC-derived progenitors lack the full self-renewal capacity of bona fide HSCs and are highly sensitive to extrinsic cues during lineage restriction [6, 27]. In addition, the ability of RA to bias lineage outcomes in normal, developmentally patterned progenitors derived from hPSCs has not been characterized in detail. A major bottleneck in advancing hPSC-based hematopoiesis is the lack of fine-tuned methods for directing lineage specification without compromising cell viability or function. This is particularly relevant for translational applications such as transfusion-ready red blood cells, myeloid cells for immunotherapy, or disease-specific cell types for modeling genetic hematological disorders. As a small molecule with pleiotropic and context-dependent effects, RA represents an attractive candidate for such control, provided its application can be precisely staged and dosed. Nevertheless, the temporal dynamics of RA signaling and its interactions with cytokine environments during hPSC-derived hematopoietic differentiation are poorly understood. In this study, we investigated the role of RA as a stage-specific modulator of hematopoietic lineage commitment in hPSC-derived differentiation systems. Our aim was to determine whether RA functions as a permissive enhancer of hematopoiesis or as an context-dependent modulator capable of biasing lineage outcomes in response to external cytokine cues. By applying RA during discrete time windows and in combination with erythroid- or myeloid-promoting factors, we assessed its ability to influence progenitor maturation, proliferation, and fate decision. This approach seeks to clarify the functional relevance of RA not only as a maturation agent but also as a tunable modulator of blood cell identity. Our findings address critical knowledge gaps in late-stage hematopoietic programming from hPSCs and offer new strategies for engineering lineage-specific blood cells in vitro. Materials and methods Culture and passaging of hPSCs In this study, we utilized the human induced pluripotent stem cell (hPSC) line KRIBB1005-A. This cell line was previously generated from human CD34⁺ umbilical cord blood cells using integration-free Sendai virus-based vectors (CytoTune™-iPS 2.0 Sendai Reprogramming Kit, Life Technologies) and has been thoroughly characterized [28]. The hPSCs were cultured in six-well plates pre-coated with Matrigel (Corning; diluted 1:30 in KO-DMEM) using TeSR™-E8™ medium (Stemcell Technologies). Cells were maintained at 37 °C, in 5% CO₂ and passaged at ~ 90% confluence. For passaging, cultures were treated with ReLeSR for 1 min at room temperature, then incubated at 37 °C for 3 min. Detached clumps were collected in KO-DMEM, centrifuged, and reseeded in TeSR™-E8™ medium with 5 μM ROCK inhibitor Y-27632 (MedChemExpress) for the first 24 h. Medium was changed daily to maintain the cells. hPSCs differentiation into hematopoietic lineage When hPSCs reaches 90% confluence, they were dissociated into single-cell suspensions using Accutase (Sigma) and counted. On day −1, cells were resuspended at a density of 2.5 × 104 cells per well in TeSR™-E8™ medium supplemented with 5 μM Y-27632 and 0.5% penicillin–streptomycin (P/S), seeded into ultra-low attachment 96-well plates (Corning), and centrifuged at 1500 rpm for 5 min to promote aggregation into embryoid bodies (EBs). Days 0–5 (Mesoderm/early hematoendothelial specification) (StepI) On day 0, E8 medium was removed and cells were differentiated towards mesoderm in APEL II medium supplemented with 1% P/S, BMP4 (40 ng/mL, R&D Systems), VEGF (50 ng/mL, R&D Systems), SCF (40 ng/mL, R&D Systems), and bFGF (5 ng/mL, R&D Systems). Days 5–10(Hematoendothelial transition) (StepII) Cells were differentiated utilizing our previously described defined media conditions [29]. To facilitate endogenous niche signaling during the hematoendothelial transition, EBs were transferred from ultra-low attachment 96-well plates and attached onto vitronectin-coated 12-well plates on day 5 in the absence of stromal feeder cells. This attachment step enables EBs to interface with the extracellular matrix and supports endogenous activation of DLL4/Notch signaling during the Step II window (days 5–10), as confirmed by RT-qPCR analysis of DLL4 expression (Supplementary Fig. 1D). Simultaneously, the medium was replaced with hematoendothelial differentiation medium: APEL II medium supplemented with 1% P/S, BMP4 (40 ng/mL), VEGF (50 ng/mL), SCF (40 ng/mL), bFGF (5 ng/mL), TPO (30 ng/mL, R&D Systems), and Flt3L (50 ng/mL, R&D Systems). Days 10–15(Hematopoietic progenitor maturation) (StepIII) On day 10, hematoendothelial differentiation medium was replaced with hematopoietic lineage specification medium: APEL II supplemented with 1% P/S, VEGF (50 ng/mL), SCF (40 ng/mL), bFGF (5 ng/mL), TPO (30 ng/mL), Flt3L (50 ng/mL), and IL-3 (30 ng/mL, R&D Systems). On day 15, cells were analyzed by flow cytometry for hematopoietic lineage markers. Differentiation of hPSC-derived hematopoietic progenitors into erythroid cells On Day 12, CD34+ hematopoietic progenitors were isolated using MACS. Importantly, MACS was applied specifically to the floating (non-adherent) blood cell fraction of the differentiation culture, which is highly enriched for hematopoietic cells and largely depleted of adherent hemogenic endothelial and stromal cells. Post-sort flow cytometric analysis confirmed that greater than 90% of the magnetically isolated CD34⁺ cells co-expressed CD45 (CD34⁺CD45⁺ double-positive; Supplementary Fig. 6 A), validating that the isolated population represents authentic hematopoietic progenitors. The isolated cells were cultured in APEL II medium supplemented with 1% P/S, BMP4 (40 ng/mL), VEGF (50 ng/mL), SCF (40 ng/mL), bFGF (5 ng/mL), TPO (30 ng/mL), Flt3L (50 ng/mL), IL-3 (30 ng/mL), and EPO. The cells were maintained for one day. From day 13 to day 15, cultures were treated with either All-trans Retinoic Acid (ATRA; Sigma-Aldrich, Cat# R2625) (treatment group) or DMSO (control group). On day 25, cells were harvested for flow cytometry, RT-PCR, and Western blot analysis. Differentiation of hPSC-derived hematopoietic progenitors into macrophages Using a protocol similar to erythroid differentiation, CD34+ hematopoietic progenitors were cultured in APEL II medium supplemented with 1% P/S, BMP4 (40 ng/mL), VEGF (50 ng/mL), SCF (40 ng/mL), bFGF (5 ng/mL), TPO (30 ng/mL), Flt3L (50 ng/mL), IL-3 (30 ng/mL), IL-6 (30 ng/mL), GM-CSF (100 ng/mL), and M-CSF (100 ng/mL). From day 13 to day 15, cells were treated with RA (treatment group) or DMSO (control group). On day 30, cells were harvested and analyzed. Methylcellulose-based hematopoietic colony forming units (CFU) assay On day 15, both non-CD34+ isolated single cells (5 × 104) and CD34+ isolated single cells (1 × 104) were collected, counted, and seeded in MethoCult™ H4436 (STEMCELL Technologies) in non-adherent 12-well plates. Cells were incubated at 37 °C for 15 days. After incubation, colony-forming units (CFUs) were imaged and quantified, including CFU-E and BFU-E erythroid colony, CFU-G granulocyte colony, CFU-M monocyte colony, CFU-GM granulocyte-monocyte colony, and CFU-GEMM multipotent hematopoietic progenitor colonies. Giemsa staining Differentiated cells were washed with PBS, centrifuged onto glass slides using a Cytospin (Thermo Fisher Scientific) at 1000 rpm for 5 min, and fixed in methanol for 5 min. Following the manufacturer’s instructions, giemsa staining solution (Abcam), diluted 2:1 with PBS, was applied for 5 min. Excess stain was removed with PBS, slides were rinsed with distilled water, air-dried, and imaged under a microscope. Flow cytometry Cells were resuspended in DPBS (Welgene) containing 3% FBS and incubated with appropriate fluorophore-conjugated antibodies for 30 min at 4 °C in the dark. After incubation, cells were washed with DPBS supplemented with 3% FBS and analyzed on a BD FACSAria III flow cytometer (BD Biosciences). A hierarchical gating strategy was applied consistently across all experiments. Briefly, cells were first gated on the FSC/SSC scatter plot to identify the primary cell population and exclude debris. A subsequent live cell gate was applied within the FSC/SSC-gated population to exclude dead cells and residual debris based on scatter properties. Marker-positive populations were then identified within the live cell gate using isotype-matched control antibodies to set background thresholds. The complete gating scheme, including representative FSC/SSC, live cell, and isotype control plots for NK cells (CD56⁺CD45⁺), erythroid cells (CD71⁺CD235a⁺), and macrophages (CD11b⁺CD14⁺), is shown in Supplementary Fig. 1C. Data were processed and the percentage of cells within each defined gate was quantified using FlowJo software (TreeStar, Inc.). Detailed information on all antibodies used is provided in Supplementary Table. Quantitative real-time PCR Total RNA was extracted from the RNeasy Plus Mini Kit (Qiagen) according to the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized from purified RNA using the ReverTra Ace® qPCR RT Master Mix (Toyobo). Quantitative PCR was performed with TB Green™ Premix (TaKaRa) and gene-specific primers. Cycle threshold (Ct) values were determined, and relative gene expression levels were calculated. Primer sequences are listed in Supplementary Table. Immunofluorescence Cells were fixed and permeabilized using Fixation/Permeabilization Solution (BD Biosciences) for 20 min, then washed with 1 × Wash/Perm Buffer (BD Biosciences) and resuspended in 100 µL of the same buffer. Samples were cytospun onto glass slides at 500 rpm for 5 min and air-dried. Slides were incubated overnight at 4 °C with primary antibodies (1:100 dilution), followed by Alexa Fluor 488 conjugated goat anti-mouse secondary antibodies (Thermo Fisher Scientific; 1:500 dilution) for 1 h in the dark. Nuclei were counterstained with DAPI for 20 min at room temperature, and images were acquired using a fluorescence microscope. A full list of primary and secondary antibodies is provided in Supplementary Table. EdU incorporation assay Cell proliferation was evaluated using the EdU Cell Proliferation Kit (Invitrogen) according to the manufacturer’s protocol. Cells were treated with RA or vehicle from day 13 to day 15. On day 15, after removal of RA or vehicle, the total differentiating cultures were incubated with 10 µM EdU for 4 h to allow incorporation into newly synthesized DNA. Following treatment, cells were fixed with 4% paraformaldehyde for 15 min at room temperature and permeabilized with 0.5% Triton® X-100 in PBS for 20 min. After two washes with 3% BSA in PBS, the Click-iT® reaction mixture was applied for 30 min in the dark to fluorescently label EdU via a copper-catalyzed azide-alkyne cycloaddition. Nuclei were counterstained with DAPI for 20 min at room temperature, and images were acquired with fluorescence microscope. Western blotting Cells were lysed on ice for 30 min in western blot lysis buffer supplemented with protease and phosphatase inhibitor cocktails (Sigma). Protein concentrations were measured using the Pierce™ BCA Protein Assay Kit. Equal amounts of protein, typically 15–20 μg, were resolved by 10% SDS-PAGE and transferred onto nitrocellulose membranes. Membranes were blocked with Every Blot Blocking Buffer (Bio-Rad) for 30 min at room temperature, followed by overnight incubation at 4 °C with primary antibodies. After three washes with 1 × TBST, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Protein bands were detected using enhanced chemiluminescence reagent and visualized with a chemiluminescence imaging system. GAPDH was used as a loading control. Antibody information is listed in Supplementary Table S2. Statistical analysis All data are presented as mean ± standard deviation (SD). The differences between two groups were evaluated using the unpaired two-tailed Student’s t-test. For comparisons involving more than two groups, one-way ANOVA followed by Tukey's multiple comparisons post-hoc test was applied. In all cases, p < 0.05 was considered statistically significant. Unless otherwise specified, experiments were conducted in triplicate. Statistical analyses and graph generation were performed using GraphPad Prism software 8.0. Results RA regulates stage-specific hematopoietic differentiation and lineage bias Before assessing the stage-specific effects of RA, we first validated the developmental progression of our feeder-free hematopoietic differentiation system, which utilizes our previously described defined media conditions combined with the physical attachment of EBs to facilitate endogenous niche signaling [29]. Using phase-contrast microscopy to observe morphological changes (Supplementary Fig. 1 A) alongside time-course flow-cytometric analysis (Supplementary Fig. 1B), we defined mesodermal cells as KDR⁺CD31⁻, hemogenic endothelial cells as CD144⁺CD43⁻CD73⁻, vascular/HE-associated cells as KDR⁺CD31⁺, endothelial cells as KDR⁻CD31⁺, hematopoietic progenitors as CD34⁺CD45⁺, and mature hematopoietic cells as CD34⁻CD45⁺. These phenotypic assignments were based on established studies of hPSC-derived hematopoiesis, in which hematopoietic potential is enriched within the CD73-negative endothelial fraction, whereas CD73-positive endothelial cells correspond to non-hemogenic vascular endothelium [29, 30]. Based on this developmental progression, we defined Step I (days 0–5) as mesoderm/early hematoendothelial specification, Step II (days 5–10) as hematoendothelial transition, and Step III (days 10–15) as hematopoietic progenitor maturation. KDR⁺CD31⁻ mesoderm emerged during days 3–5, followed by the appearance of a CD144⁺CD43⁻CD73⁻ hemogenic endothelial population around days 4–5. Transient KDR⁺CD31⁺ and KDR⁻CD31⁺ populations were observed during days 6–8, preceding the accumulation of CD34⁺CD45⁺ hematopoietic progenitors and CD34⁻CD45⁺ mature hematopoietic cells from days 13 to 15 (Fig. 1B). Notably, a key feature of this system is the physical attachment of EBs onto vitronectin-coated plates at Day 5 in the absence of stromal feeder cells. This matrix-mediated niche interaction promoted hemogenic endothelial specification, as evidenced by SOX17 expression detected by immunocytochemistry during the specification window (Days 7–10) (Fig. 1C). In parallel, endogenous DLL4 expression was induced under these conditions, with Western blot analysis showing peak DLL4 protein levels at Day 7 (Fig. 1D) and RT-qPCR analysis confirming increased DLL4 transcript expression during the hemogenic endothelial specification window (Supplementary Fig. 1D). Together, these findings indicate that EB attachment to vitronectin-coated plates provides a biologically relevant hemogenic endothelial specification cue without exogenous DLL4 supplementation. Consistent with hemogenic endothelial specification during this window, RT-qPCR analysis confirmed robust expression of SOX17 (Supplementary Fig. 1D), a transcription factor required for arterial and hemogenic endothelial identity, which was further validated by immunofluorescence staining at days 7, 8, and 10 (Fig. 1C). In addition, downstream differentiation yielded erythroid (CD235a⁺), myeloid, and CD56⁺ natural killer (NK) cells (Supplementary Fig. 1 C), supporting the interpretation that this system generated definitive hematopoietic progenitors and is not restricted to a primitive erythromyeloid output. To directly visualize active EHT within this system, we performed immunofluorescence staining of Day 10 and 15 cultures using antibodies against CD31 and CD45, counterstained with Hoechst. This analysis revealed CD45⁺ hematopoietic cells in direct physical association with CD31⁺ vessel-like endothelial structures (Supplementary Fig. 1E), providing morphological evidence that hematopoietic cells actively emerge from endothelial structures in our feeder-free culture system, consistent with bona fide EHT. To further validate the definitive hematopoietic identity of progenitors generated in this system at the molecular level, we analyzed expression of mesodermal and hematoendothelial markers, including HOXA9 and HOXA10, by RT-qPCR. Both HOXA genes showed progressive upregulation during differentiation, peaking at Day 10–15 (Supplementary Fig. 1D). Activation of medial HOXA genes is the field-standard molecular criterion distinguishing definitive from primitive hematopoietic progenitors, as primitive progenitors derived from the yolk sac lack HOXA cluster expression [3, 31]. Taken together, the sequential phenotypic progression, endogenous DLL4/SOX17 expression, multilineage output, active EHT morphology, and HOXA9/10 upregulation are consistent with progenitors showing features of a definitive-type hematopoietic identity.Fig. 1RA accelerates hematopoietic progenitor maturation and alters colony distribution in a stepwise hPSC differentiation system. A Schematic overview of the three-step protocol for differentiating hPSCs into hematopoietic lineages. B Sequential flow cytometric analysis of hematopoietic cell development from days 3 to 15. C Immunofluorescence staining of SOX17 (green) and nuclei (Hoechst, blue) during hemogenic endothelial specification at days 7, 8, and 10. Scale bar, 100 μm. D Western blot analysis of DLL4 protein expression during the Step II differentiation window (Days 0, 7, 8, 10). GAPDH was used as a loading control. The numbers above the blot represent relative protein levels calculated as the ratio of DLL4 to GAPDH, with Day 7 set to 1.0. E Schematic of RA treatment windows during Step 3. F Representative flow cytometry plots showing progenitors (CD34+CD45+) and mature hematopoietic cells (CD34-CD45+), with percentages, in control versus RA-treated groups (n=18). G Fold change in total hematopoietic cell numbers relative to control during differentiation (n=18). H Total colony-forming units (CFUs) per 50 K input cells and CFU frequency per 1 K CD34⁺CD45⁺ progenitors in control and RA-treated groups. The number of CD34⁺CD45⁺ progenitors used for normalization was calculated from FACS analysis shown in (F), by multiplying total live cell counts by the percentage of CD34⁺CD45⁺ double-positive cells. (n=6). I CFUs obtained from 50 K cells showing colony subtype distribution (BFU-E, CFU-E, CFU-G, CFU-M, CFU-GM, CFU-GEMM) in control and RA-treated groups, represented as both pie charts and stacked bar graphs showing the percentage of total CFUs (Colony Distribution, % of Total CFUs) for each subtype (n=6). J Representative CFU colony morphologies. Scale bars, 50 μm and 100 μm. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p<0.05; **p<0.01; ***p<0.001 RA accelerates hematopoietic progenitor maturation and alters colony distribution in a stepwise hPSC differentiation system. A Schematic overview of the three-step protocol for differentiating hPSCs into hematopoietic lineages. B Sequential flow cytometric analysis of hematopoietic cell development from days 3 to 15. C Immunofluorescence staining of SOX17 (green) and nuclei (Hoechst, blue) during hemogenic endothelial specification at days 7, 8, and 10. Scale bar, 100 μm. D Western blot analysis of DLL4 protein expression during the Step II differentiation window (Days 0, 7, 8, 10). GAPDH was used as a loading control. The numbers above the blot represent relative protein levels calculated as the ratio of DLL4 to GAPDH, with Day 7 set to 1.0. E Schematic of RA treatment windows during Step 3. F Representative flow cytometry plots showing progenitors (CD34+CD45+) and mature hematopoietic cells (CD34-CD45+), with percentages, in control versus RA-treated groups (n=18). G Fold change in total hematopoietic cell numbers relative to control during differentiation (n=18). H Total colony-forming units (CFUs) per 50 K input cells and CFU frequency per 1 K CD34⁺CD45⁺ progenitors in control and RA-treated groups. The number of CD34⁺CD45⁺ progenitors used for normalization was calculated from FACS analysis shown in (F), by multiplying total live cell counts by the percentage of CD34⁺CD45⁺ double-positive cells. (n=6). I CFUs obtained from 50 K cells showing colony subtype distribution (BFU-E, CFU-E, CFU-G, CFU-M, CFU-GM, CFU-GEMM) in control and RA-treated groups, represented as both pie charts and stacked bar graphs showing the percentage of total CFUs (Colony Distribution, % of Total CFUs) for each subtype (n=6). J Representative CFU colony morphologies. Scale bars, 50 μm and 100 μm. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p<0.05; **p<0.01; ***p<0.001 Having established the baseline developmental trajectory, we systematically evaluated the effects of continuous RA treatment during each step (Fig. 1A). RA treatment during Step I completely abolished hematopoietic development and led to significantly reduced total cell numbers, accompanied by extensive cell death (Supplementary Fig. 2A–E). Consistently, colony-forming unit (CFU) assays confirmed the absence of hematopoietic colonies (Supplementary Fig. 2 F). Molecular analysis revealed that RA severely downregulated key mesodermal markers (Brachyury, MIXL1) while upregulating ectodermal genes (PAX6, SOX2), implying a lineage shift away from the mesodermal lineage (Supplementary Fig. 2H). Endodermal markers (GATA4, FOXA2) were also suppressed. The concentration used in all single-dose Step I experiments was 1 µM ATRA, consistent with the pharmacological concentration widely used in prior hPSC hematopoietic differentiation studies [32, 33]. To determine whether these effects were concentration-dependent, we performed a dose–response analysis using 0.01, 0.1, and 1 µM ATRA during Step I (Supplementary Fig. 3A-C). Flow cytometric analysis at multiple developmental stages revealed a clear dose-dependent pattern. First, increasing RA concentrations progressively reduced the KDR⁺CD31⁻ mesodermal population (Supplementary Fig. 3 A), indicating that RA exposure during Step I suppresses mesodermal differentiation itself, consistent with the concomitant reduction in Brachyury and MIXL1 and upregulation of ectodermal markers observed under the same conditions (Supplementary Fig. 2H). Second, the frequency of CD144⁺CD43⁻CD73⁻ hemogenic endothelial cells decreased in a dose-dependent manner (Supplementary Fig. 3B), confirming that RA impairs hemogenic endothelial specification across all concentrations tested. Third, the final hematopoietic output (CD34⁺CD45⁺ progenitors) was reduced dose-dependently (Supplementary Fig. 3 C). Together, these data confirm that the inhibitory effect of continuous Step I RA treatment on hematopoietic development is concentration-dependent, with suppression evident from 0.01 µM. When RA was applied during Step II, corresponding to HET stage, a small population of hematopoietic cells still emerged, but their frequency and CFU output were significantly reduced compared to controls (Supplementary Fig. 2A–G). Unlike Step I, the reduction in hematopoietic output at this stage was not due to compromised viability (Supplementary Fig. 2 C), but was rather, the hematoendothelial population was significantly suppressed. The residual surviving fraction retained its endothelial transcriptional identity, evidenced by the expression of HE-associated genes (EPHB4, EFNB2) (Supplementary Fig. 2I). This pattern is mechanistically consistent with RA reinforcing endothelial transcriptional identity and thereby blocking EHT progression, rather than eliminating cells outright. EPHB4 and EFNB2 are arterial and venous endothelial markers, respectively, whose expression reflects a pre-EHT endothelial state rather than hematopoietic potential. Their persistence under Step II RA treatment therefore suggests that RA maintains the surviving fraction in a hemogenic endothelial state and prevents its conversion to hematopoietic fate, consistent with prior reports that RA promotes arterial endothelial specification and can antagonize the transcriptional program required for hematopoietic emergence [34]. To determine whether the Step II effects were also concentration-dependent, we performed a dose–response analysis using 0.01, 0.1, and 1 µM ATRA during Step II (Supplementary Fig. 3D). This analysis revealed a qualitatively distinct, biphasic pattern from that observed at Step I. At low doses (0.01 and 0.1 µM), CD34⁺CD45⁺ progenitor frequency was modestly reduced while the CD34⁻CD45⁺ mature hematopoietic population paradoxically increased, suggesting that sub-optimal RA concentrations during Step II may accelerate EHT and promote progenitor differentiation toward mature hematopoietic cells rather than blocking this transition. In contrast, 1 µM ATRA during Step II nearly abolished both populations, recapitulating the high-dose inhibitory phenotype. This biphasic dose–response at Step II, where low-dose RA promotes and high-dose RA inhibits hematopoietic output, is consistent with the concentration-dependent role of RA in hematopoietic development described by Luff et al. [35], and further supports the interpretation that RA functions as a tunable, dose- and stage-dependent modulator of hematopoietic fate in this feeder-free system. Notably, RA treatment during Step III (days 10–15) altered the composition of hematopoietic output without affecting total hematopoietic cell numbers. Flow cytometric analysis revealed a decrease in progenitor cells (CD34+CD45+) and a corresponding increase in mature hematopoietic cells (CD34−CD45+), indicating that RA promoted terminal maturation over progenitor maintenance (Fig. 1F, Supplementary Fig. 4B). Although overall cell numbers slightly decreased, possibly due to modest cell death (Fig. 1F, G, Supplementary Fig. 4A), RA-treated cultures exhibited a significant increase in CFU formation, especially when normalized to input progenitor numbers (Fig. 1H; here defined as the FACS-quantified CD34⁺CD45⁺ double-positive cell count derived from Fig. 1F, to ensure normalization was based exclusively on bona fide hematopoietic progenitors). Importantly, RA induced a notable shift in lineage composition. Erythroid colonies (CFU-E, BFU-E) and multipotent CFU-GEMM were significantly enriched, while myeloid colony types, including CFU-M and CFU-GM, were markedly reduced (Fig. 1I). Wright-Giemsa staining confirmed the presence of various hematopoietic cell types in these colonies (Fig. 1J). Taken together, these results establish a dual role for RA during hPSC-derived hematopoietic differentiation. RA inhibited early hematopoietic specification when applied during the mesoderm or HE stages, but enhanced progenitor maturation and skewed lineage fate when applied during the late hematopoietic phase. These findings identify RA as a context-dependent, stage-specific modulator of hematopoietic differentiation that promotes erythroid lineage biases at the expense of myeloid potential in hPSC-derived blood progenitors. Temporal and dose-dependent effects of RA on hematopoietic lineage specification The transition from HE to hematopoietic progenitors (days 10–15) encompasses two distinct developmental windows: the EHT and subsequent progenitor maturation. To examine its stage-specific effects, we applied RA during either the early (days 10–12) or late (days 13–15) phase of development (Fig. 2A). RA treatment during days 13–15 significantly enhanced hematopoietic maturation, as evidenced by an increased proportion of CD34−CD45+ mature hematopoietic cells and a concurrent reduction in CD34+CD45+ progenitors (Fig. 2B, Supplementary Fig. 5 A). Total hematopoietic cell numbers were also elevated in this group, whereas continuous RA treatment throughout Step 3 impaired overall cell expansion (Figs. 1G, 2C). In contrast, RA exposure during days 10–12 failed to promote progenitor maturation, and led to decreased total cell numbers, implying that RA may negatively affect the EHT phase.Fig. 2RA modulates hematopoietic differentiation in a time- and dose-dependent manner. A Schematic of RA treatment windows during Step 3: days 10–12 and days 13–15. B Flow cytometry analysis of progenitors (CD34+CD45+) and mature hematopoietic cells (CD34−CD45+) after RA exposure (n = 3). C Fold change in total hematopoietic cell numbers following RA treatment at days 10–12 or days 13–15 (n = 3). D Total CFUs per 50 K input cells and CFU frequency per 1 K CD34+CD45+ progenitors after RA treatment at days 10–12 or days 13–15 (n = 3). E CFU subtype counts (BFU-E, CFU-E, CFU-G, CFU-M, CFU-GM, CFU-GEMM) per 50 K cells following RA treatment at days 10–12 or days 13–15 (n = 3). F Experimental design for dose–response RA treatment (0.01, 0.1, 1, or 2 µM) applied to isolated CD34+ progenitors during days 13–15 (n = 6). G Flow cytometry showing RA dose-dependent changes in CD34+CD45+ progenitors and CD34−CD45+ mature hematopoietic cells (n = 6). H Fold change in total hematopoietic cell numbers across RA doses (n = 6). I CFUs per 10 K total cells and CFU frequency per 1 K CD34+CD45.+ progenitors across RA doses (n = 6). J CFU subtype counts per 10 K cells across RA doses (n = 6). Data are presented as mean ± SD. P values were calculated using one-way ANOVA followed by Tukey’s multiple comparisons test. *p < 0.05; **p < 0.01; ***p < 0.001 RA modulates hematopoietic differentiation in a time- and dose-dependent manner. A Schematic of RA treatment windows during Step 3: days 10–12 and days 13–15. B Flow cytometry analysis of progenitors (CD34+CD45+) and mature hematopoietic cells (CD34−CD45+) after RA exposure (n = 3). C Fold change in total hematopoietic cell numbers following RA treatment at days 10–12 or days 13–15 (n = 3). D Total CFUs per 50 K input cells and CFU frequency per 1 K CD34+CD45+ progenitors after RA treatment at days 10–12 or days 13–15 (n = 3). E CFU subtype counts (BFU-E, CFU-E, CFU-G, CFU-M, CFU-GM, CFU-GEMM) per 50 K cells following RA treatment at days 10–12 or days 13–15 (n = 3). F Experimental design for dose–response RA treatment (0.01, 0.1, 1, or 2 µM) applied to isolated CD34+ progenitors during days 13–15 (n = 6). G Flow cytometry showing RA dose-dependent changes in CD34+CD45+ progenitors and CD34−CD45+ mature hematopoietic cells (n = 6). H Fold change in total hematopoietic cell numbers across RA doses (n = 6). I CFUs per 10 K total cells and CFU frequency per 1 K CD34+CD45.+ progenitors across RA doses (n = 6). J CFU subtype counts per 10 K cells across RA doses (n = 6). Data are presented as mean ± SD. P values were calculated using one-way ANOVA followed by Tukey’s multiple comparisons test. *p < 0.05; **p < 0.01; ***p < 0.001 To assess the functional output of hematopoietic differentiation, we performed CFU assays. RA treatment during days 13–15 resulted in a significant increase in both total colony numbers and CFUs per 1 × 104 progenitors, while treatment during days 10–12 led to substantial reductions in both metrics (Fig. 2D, E). Colony subtype analysis revealed that late RA exposure (days 13–15) specifically increased erythroid colonies (BFU-E, CFU-E) while reducing myeloid colonies (CFU-M), indicating a shift in lineage potential (Fig. 2E, Supplementary Fig. 5B, C). These findings imply that RA primarily regulates lineage specification and promotes maturation during the later stage of hematopoietic development, rather than during the EHT. We next investigated the dose-dependent effects of RA during the late window by treating cells with increasing concentrations of RA (0.01, 0.1, 1, and 2 μM) from days 13 to 15 (Fig. 2F). All concentrations of RA promoted maturation, as indicated by a dose-dependent increase in CD34−CD45+ mature hematopoietic cells (Fig. 2G). Among these, 0.1 μM RA yielded the most pronounced enhancement of total cell numbers and CFU output (Fig. 2H, I). Although the proportion of CD34+CD45+ progenitors remained relatively stable across conditions, the number of mature hematopoietic cells increased significantly (Supplementary Fig. 5D), supporting the conclusion that RA promotes progenitor maturation rather than expansion of the progenitor compartment itself. Colony subtype analysis further demonstrated that all RA concentrations tested biased differentiation toward the erythroid lineage. RA significantly decreased the proportion of CFU-M colonies while increasing both BFU-E and CFU-E colonies in a dose-dependent manner (Fig. 2I, J). Notably, erythroid output appeared to plateau beyond 0.1 μM RA, implying that this concentration represents an optimal dose for inducing erythroid lineage bias, with no additional benefit at higher levels. Taken together, these results show that RA functions as a temporal and dose-specific regulator of hematopoietic differentiation. RA promoted progenitor maturation and selectively enhanced erythroid lineage commitment when applied during the late stages of hematopoietic development. In contrast, RA exposure during the EHT impaired hematopoietic output. Among the concentrations tested, 0.1 μM RA emerged as the most effective, providing a strong foundation for future strategies to direct lineage output from hPSC-derived hematopoietic progenitors. RA promotes progenitor maturation and functional output via proliferation and lineage specification To investigate the intrinsic effects of RA on hematopoietic progenitor cell function, we isolated CD34+ cells by magnetic-activated cell sorting from the floating blood cell fraction. Post-sort flow cytometric analysis confirmed that greater than 90% of the isolated CD34⁺ cells co-expressed CD45 (Supplementary Fig. 6 A), validating their identity as CD34⁺CD45⁺ hematopoietic progenitors. Isolated cells were then treated them with 0.1 μM RA for 2 days during the late maturation phase (days 13–15), based on our previous observation that RA exerts strong effects during this window (Fig. 3A). Flow cytometric analysis showed that RA treatment significantly increased the proportion of mature hematopoietic cells (CD34−CD45+) while reducing the proportion of progenitor cells (CD34+CD45+) (Fig. 3B), indicating that RA promotes terminal differentiation of hematopoietic progenitors. In addition, RA treatment significantly increased total hematopoietic cell numbers (Fig. 3C). To assess proliferative capacity following the late RA treatment window, an EdU incorporation assay was performed on the total differentiating culture at day 15 after RA removal. The results showed that RA treatment markedly enhanced overall EdU incorporation (Fig. 3D). Crucially, although the absolute number of CD34⁺CD45⁺ progenitor cells remained unchanged, the absolute number of mature hematopoietic cells (CD34⁻CD45⁺) increased significantly (Supplementary Fig. 6B). Because the progenitor pool size remained static while total and mature cell numbers expanded, these findings are most consistent with increased EdU incorporation reflecting proliferation associated with downstream hematopoietic maturation, rather than expansion of the progenitor compartment itself. Functional differentiation potential was assessed by CFU assay. RA-treated cells exhibited significant increases in both total CFU numbers and CFUs per 1 × 104 input cells, demonstrating improved clonogenic capacity (Fig. 3E, Supplementary Fig. 6 C). In MethoCult medium containing EPO, RA treatment also promoted erythroid colony formation, indicating enhanced responsiveness to EPO signaling (Fig. 3F). This erythroid bias was accompanied by upregulation of key components of the EPO signaling pathway, including EPOR, STAT5A, STAT5B, and BCL_XL. Furthermore, increased expression of RARB, RARG, and CYP26A1 confirmed activation of the RA signaling axis (Fig. 3G). To determine whether these transcriptional changes were accompanied by functional changes at the protein level, we performed Western blot analysis of p-STAT5 and total STAT5 in control and RA-treated progenitors (Fig. 3H). RA treatment markedly increased total STAT5 protein levels (~ 10–17-fold), consistent with the transcriptional upregulation observed in Fig. 3G, and absolute p-STAT5 levels were correspondingly elevated (~ 7.5–10-fold). The p-STAT5/total STAT5 ratio was maintained between conditions, suggesting that RA expands the STAT5 protein pool rather than directly enhancing phosphorylation efficiency. As a result, the net amount of active p-STAT5 is substantially amplified, consistent with the enhanced erythroid differentiation observed in this system. These findings show that RA enhanced the maturation, proliferation, and lineage output of hematopoietic progenitors, primarily by promoting erythroid differentiation through upregulation of EPO pathway components. This underscores the role of RA as a powerful regulator of progenitor fate and functional hematopoietic output.Fig. 3RA modulates hematopoietic differentiation, proliferation, and colony-forming capacity. A Experimental timeline for 0.1 µM RA treatment applied to isolated CD34⁺ hematopoietic progenitors during days 13–15. B Flow cytometry analysis of progenitors (CD34⁺CD45⁺) and mature hematopoietic cells (CD34⁻CD45⁺) under control versus RA-treated conditions (n = 15). C Total hematopoietic cell numbers following RA treatment (n = 15). D EdU incorporation assay performed on the total culture at day 15 after removal of RA or vehicle, showing increased proliferation in RA-treated cells compared with controls (n = 3). E Total CFUs per 50 K input cells and CFU frequency per 1 K CD34+CD45.+ progenitors after RA treatment, calculated using flow cytometry-derived proportions (n = 9). F CFU subtype counts (BFU-E, CFU-E, CFU-G, CFU-M, CFU-GM, CFU-GEMM) per 50 K cells in control versus RA-treated groups (n = 9). G Relative expression of EPO-pathway genes (EPOR, BCL-XL, STAT3, STAT5A, STAT5B) and RA receptors (RARB, RARG) and the canonical RA target gene CYP26A1 in progenitors following RA exposure (n = 6). H Western blot analysis of p-STAT5, total STAT5, and GAPDH in control and RA-treated progenitors at day 15. Numbers above each band indicate relative protein levels normalized to GAPDH and relative to control. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p < 0.05; **p < 0.01; ***p < 0.001 RA modulates hematopoietic differentiation, proliferation, and colony-forming capacity. A Experimental timeline for 0.1 µM RA treatment applied to isolated CD34⁺ hematopoietic progenitors during days 13–15. B Flow cytometry analysis of progenitors (CD34⁺CD45⁺) and mature hematopoietic cells (CD34⁻CD45⁺) under control versus RA-treated conditions (n = 15). C Total hematopoietic cell numbers following RA treatment (n = 15). D EdU incorporation assay performed on the total culture at day 15 after removal of RA or vehicle, showing increased proliferation in RA-treated cells compared with controls (n = 3). E Total CFUs per 50 K input cells and CFU frequency per 1 K CD34+CD45.+ progenitors after RA treatment, calculated using flow cytometry-derived proportions (n = 9). F CFU subtype counts (BFU-E, CFU-E, CFU-G, CFU-M, CFU-GM, CFU-GEMM) per 50 K cells in control versus RA-treated groups (n = 9). G Relative expression of EPO-pathway genes (EPOR, BCL-XL, STAT3, STAT5A, STAT5B) and RA receptors (RARB, RARG) and the canonical RA target gene CYP26A1 in progenitors following RA exposure (n = 6). H Western blot analysis of p-STAT5, total STAT5, and GAPDH in control and RA-treated progenitors at day 15. Numbers above each band indicate relative protein levels normalized to GAPDH and relative to control. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p < 0.05; **p < 0.01; ***p < 0.001 RA enhances erythroid commitment via EPO signaling and represses myeloid differentiation Building on our observation that RA upregulates erythropoietic signaling pathways (Fig. 3), we next examined whether RA can modulate hematopoietic lineage decisions in the presence of EPO added on day 10 with application of 0.1 μM RA during the late maturation window (days 13–15) and cultures were maintained until day 25 (Fig. 4A). RA treatment significantly increased the total numbers of red blood cells (Supplementary Fig. 7A) and the proportion of mature erythroid cells (CD71−CD235a+) (Fig. 4B), indicating enhanced erythroid differentiation. To assess lineage bias further, we analyzed the proportions of erythroid (CD235a+) and myeloid (CD33+) populations by flow cytometry. This experiment was designed to determine whether RA not only promotes erythroid lineage specification but also limits residual differentiation into the myeloid lineage. Indeed, RA treatment significantly promoted erythroid commitment while concurrently reducing the frequency of CD33+ myeloid cells (Fig. 4C). Within the myeloid population, RA markedly suppressed macrophage generation, as reflected by reductions in CD11b+CD14+ and CD14+ subsets (Supplementary Fig. 7B), confirming that RA represses macrophage lineage specification under EPO-driven conditions. To validate terminal erythroid maturation, we performed Wright-Giemsa staining to assess morphological progression. RA-treated cultures exhibited a higher proportion of late-stage erythroid cells, particularly polychromatic and orthochromatic erythrocytes (Fig. 4D, Supplementary Fig. 7C). In parallel, β-globin immunofluorescence staining revealed that RA enhanced β-globin expression and increased the proportion of enucleated cells, both hallmarks of terminal erythroid differentiation (Fig. 4E). At the transcriptional level, RA treatment significantly upregulated erythroid-specific transcription factors including GATA1, KLF1, NFE2, and BCL11A, along with globin genes (α-globin, γ-globin, β-globin) (Fig. 4F). To contextualize the level of β-globin induction in our system, we generated hematopoietic progenitors using the differentiation protocol of Ng et al. [3], which produces CD34⁺ progenitors with long-term multilineage engraftment capacity, and subsequently differentiated them into erythroid cells using our own RBC differentiation protocol under identical EPO-containing conditions, serving as a definitive erythropoiesis reference (Ref). β-globin expression in untreated control cultures (Ctrl) was comparable to the Ng et al. -derived reference [3], with no significant difference between Ref and Ctrl (Fig. 4F), indicating that our system generates erythroid cells with β-globin expression levels consistent with those of a validated definitive hematopoietic benchmark. RA treatment significantly increased β-globin expression beyond that of both the reference and untreated control (Fig. 4F), demonstrating that RA enhances β-globin induction above the level achievable with established definitive protocol-derived progenitors. These results imply that RA acts in a cooperative manner with EPO to activate a transcriptional program that supports erythroid specification and maturation. Taken together, these findings demonstrate that RA acts cooperatively with EPO signaling to enhance erythropoiesis while concurrently repressing myeloid and macrophage lineage output. These observations underscore the potential of RA as a powerful modulator of cytokine-guided lineage commitment in hPSC-derived hematopoietic progenitors.Fig. 4RA accelerates EPO-induced erythroid differentiation from hematopoietic progenitors. A Schematic of RA treatment timing during erythroid differentiation. B Flow cytometry analysis showing proportions of erythroid precursors (CD71⁺CD235a⁺) and mature erythroid cells (CD71⁻CD235a⁺) in control versus RA-treated groups (n = 9). C Lineage bias analysis comparing erythroid (CD235a⁺) and myeloid (CD33⁺) populations (n = 3). D Wright-Giemsa staining illustrating erythroid maturation stages: proerythroblasts, basophilic, polychromatic, and orthochromatic erythroblasts, and reticulocytes (n = 3). E Immunofluorescence analysis of β-globin (green) and nuclei (DAPI, blue), with quantification of relative fluorescence intensity and enucleation efficiency (n = 16). F qPCR of erythroid differentiation genes-transcription factors (GATA1, KLF1, BCL11A, NFE2) and globins (α-, β-, γ-globin)- normalized to β-actin (n = 3). In the β-globin panel, 'Ref' denotes a definitive erythropoiesis reference generated using the Ng et al. (2024) protocol. ns, not significant. Scale bar, 100 μm. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p < 0.05; **p < 0.01; ***p < 0.001 RA accelerates EPO-induced erythroid differentiation from hematopoietic progenitors. A Schematic of RA treatment timing during erythroid differentiation. B Flow cytometry analysis showing proportions of erythroid precursors (CD71⁺CD235a⁺) and mature erythroid cells (CD71⁻CD235a⁺) in control versus RA-treated groups (n = 9). C Lineage bias analysis comparing erythroid (CD235a⁺) and myeloid (CD33⁺) populations (n = 3). D Wright-Giemsa staining illustrating erythroid maturation stages: proerythroblasts, basophilic, polychromatic, and orthochromatic erythroblasts, and reticulocytes (n = 3). E Immunofluorescence analysis of β-globin (green) and nuclei (DAPI, blue), with quantification of relative fluorescence intensity and enucleation efficiency (n = 16). F qPCR of erythroid differentiation genes-transcription factors (GATA1, KLF1, BCL11A, NFE2) and globins (α-, β-, γ-globin)- normalized to β-actin (n = 3). In the β-globin panel, 'Ref' denotes a definitive erythropoiesis reference generated using the Ng et al. (2024) protocol. ns, not significant. Scale bar, 100 μm. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p < 0.05; **p < 0.01; ***p < 0.001 RA acts cooperatively with M/GM-CSF to amplify myeloid lineage commitment and macrophage differentiation Given our earlier finding that RA promotes erythroid lineage specification under EPO stimulation, we next examined whether this effect is maintained under alternative cytokine environments that direct differentiation toward myeloid fates. Specifically, we tested whether RA enhances macrophage differentiation under M/GM-CSF conditions, in line with its proposed role as an amplifier of the prevailing cytokine-directed lineage fate. This experiment assessed whether the lineage-modulating effects of RA are context-dependent or permissive in the context of cytokine-driven specification. For this purpose, we supplemented hematopoietic cultures with M/GM-CSF and treated cells with 0.1 μM RA during the late differentiation window (days 13–15) (Fig. 5A). RA treatment significantly increased total cell numbers in the presence of myeloid-promoting cytokines (Fig. 5B). Flow cytometric analysis revealed that RA markedly enhanced macrophage differentiation, as shown by increased frequencies of CD11b+CD14+ cells (Fig. 5C). Correspondingly, the proportion of erythroid cells was reduced within this M/GM-CSF cytokine context, as evidenced by decreases in CD71+ populations (Fig. 5D), indicating a shift in lineage output from the erythroid to myeloid fate under these conditions. To examine the effects of RA on macrophage development further, we analyzed specific stages within the myeloid lineage, including monoblasts, promonocytes, monocytes, and mature macrophages. RA treatment significantly reduced early-stage precursors (monoblasts, promonocytes, monocytes) and increased the proportion of mature macrophages (Fig. 5E, Supplementary Fig. 8 A), implying that RA not only promotes lineage specification but also accelerates myeloid maturation. In addition, Wright-Giemsa-stained smears revealed the presence of granulocytes, including eosinophils and neutrophils, under control conditions. RA treatment suppressed the generation of these granulocytic lineages, reinforcing its selective enhancement of the monocyte/macrophage axis (Supplementary Fig. 8B, C). In summary, RA acted cooperatively with M/GM-CSF to amplify myeloid lineage commitment, particularly macrophage differentiation. The proportional reduction in erythroid markers observed under these conditions reflects RA's role in reinforcing the myeloid-biased cytokine environment rather than an independent suppression of erythropoiesis, an interpretation consistent with the well-established principle that the external cytokine milieu is the primary determinant of hematopoietic lineage fate. Importantly, the observed shifts in cell proportions, upregulation of lineage-specific transcription factors, and accelerated progression through defined myeloid maturation stages collectively argue that RA acts as a genuine fate-amplifying signal within the cytokine context, rather than exerting effects solely through non-specific proliferative enhancement. These results highlight the context-dependent nature of RA activity and support its potential as a tunable modulator of cytokine-guided hematopoietic fate decisions in hPSC-derived blood cell production.Fig. 5RA accelerates macrophage differentiation induced by M-CSF and GM-CSF. A Schematic of RA treatment timing during differentiation of hPSC-derived hematopoietic progenitors into macrophages in the presence of M-CSF and GM-CSF. B Total hematopoietic cell numbers under control versus RA treatment (n = 3). C Flow cytometry analysis of CD11b⁺CD14⁺ myeloid cells with quantification of positive populations (n = 3). D Lineage distribution assessed by flow cytometry analysis of erythroid (CD71⁺) versus myeloid (CD33⁺) populations (n = 3). E Wright-Giemsa morphology of myeloid maturation stages-monoblasts, promonocytes, monocytes, and macrophages-with quantification (n = 8). Scale bar, 100 μm. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p < 0.05; **p < 0.01; ***p < 0.001 RA accelerates macrophage differentiation induced by M-CSF and GM-CSF. A Schematic of RA treatment timing during differentiation of hPSC-derived hematopoietic progenitors into macrophages in the presence of M-CSF and GM-CSF. B Total hematopoietic cell numbers under control versus RA treatment (n = 3). C Flow cytometry analysis of CD11b⁺CD14⁺ myeloid cells with quantification of positive populations (n = 3). D Lineage distribution assessed by flow cytometry analysis of erythroid (CD71⁺) versus myeloid (CD33⁺) populations (n = 3). E Wright-Giemsa morphology of myeloid maturation stages-monoblasts, promonocytes, monocytes, and macrophages-with quantification (n = 8). Scale bar, 100 μm. Data are presented as mean ± SD. P values were calculated using an unpaired two-tailed t-test. *p < 0.05; **p < 0.01; ***p < 0.001 Discussion Our study revealed a multifaceted and stage-specific role of RA in directing hematopoietic differentiation from hPSCs, providing new mechanistic insights and translational opportunities. Using a stepwise differentiation model, we showed that RA exerted inhibitory effects on early mesoderm and HE specification but promoted maturation, proliferation, and lineage bias during late hematopoietic development. The context and dose dependence of RA signaling provides a versatile regulatory node that can be tuned to enhance specific lineage outputs, with implications for regenerative medicine, disease modeling, and blood product manufacturing. Early exposure to RA during mesoderm induction (days 0–5) or HE specification (days 5–10) substantially suppressed hematopoietic progenitor formation, consistent with previous reports demonstrating that RA acts as an antagonist of Wnt and BMP4 signaling, both of which are essential for mesoderm patterning and hematoendothelial commitment [19, 36–38]). The inhibitory effects of RA were associated with downregulation of mesodermal transcription factors such as Brachyury and MIXL1, consistent with studies showing RA-driven fate diversion away from mesoderm toward ectodermal and endodermal fates [39, 40]. Moreover, high-dose RA was found to severely limit hematopoietic specification, as observed in human stem cell models [32]. This is further consistent with the findings of Szatmari et al., who demonstrated in murine embryonic stem cell models that activation of the retinoid signaling pathway inhibits hematopoiesis and uncouples RA signaling from HOX gene activation during hematopoietic development [41]. Importantly, however, the inhibitory effect we observe at Step I is not a selective redirection of mesoderm toward an alternative fate such as cardiac specification, but rather a broad suppression of mesodermal emergence itself, as reflected by the downregulation of Brachyury and MIXL1 and the concomitant upregulation of ectodermal markers (Supplementary Fig. 2H). This distinction is important: our data indicate that premature RA activation disrupts the mesodermal competence required for hematopoietic specification, consistent with the general inhibitory role described by Szatmari et al. [41], rather than invoking a fate switch toward any specific alternative mesoderm-derived lineage. Taken together, these findings indicated that hematopoietic specification requires a tightly regulated, low-RA signaling environment. Disruption of this balance during early developmental stages impairs mesodermal emergence and hematopoietic competence, an effect strongly reflected in our system. Strikingly, the role of RA reversed during the EHT and progenitor maturation phases (days 10–15). Our data demonstrate that RA treatment specifically between days 13–15 acted on hematopoietic progenitors rather than on the EHT itself, which occurs between days 10 and 12. We observed that RA during this late window promoted progenitor maturation, as evidenced by a reduction in the CD34+CD45+ fraction and an increase in mature CD34−CD45+ hematopoietic output. This observation is consistent with recent findings that RA contributes to the establishment of the hematopoietic program, in part by activating HOX genes, which are critical components of definitive hematopoietic development [9, 20, 31]. Furthermore, we observed that RA accelerated proliferation, as confirmed by EdU incorporation assays, implying that RA supports the expansion and terminal differentiation of committed hematopoietic progenitors. It is important to note that the cells characterized in this study are committed hPSC-derived progenitors, which lack the full self-renewal capacity of bona fide HSCs, and the in vivo engraftment potential of the RA-modulated cells has not been assessed. The proliferative effect of RA observed here is therefore associated with terminal lineage maturation rather than maintenance of a multipotent undifferentiated state, consistent with previous reports that while RA promoted progenitor proliferation and differentiation, the bone marrow niche maintained adult HSC quiescence by sustaining a low-retinoid environment [21]. Notably, our finding that RA-treated progenitors displayed enhanced functional capacity, as evidenced by improved CFU output, raises intriguing new questions. Specifically, RA may be acting not only as a driver of maturation but also as a functional regulator that enhances the intrinsic output potential of progenitor cells. This implies the existence of novel RA-mediated mechanisms that operate within committed progenitor pools, which requires further investigation. To place the temporal regulation of RA observed in our study into an in vivo context, the robust promotion of maturation and lineage bias during the late stage of differentiation may parallel the developmental transition in which newly formed hematopoietic progenitors enter the fetal liver microenvironment [4, 5]. During embryonic development, the fetal liver serves as the major site of hematopoietic expansion and erythropoiesis [5]. Previous studies have further shown that RA signaling contributes to the onset of fetal liver erythropoietin expression and early erythropoietic differentiation [24]. In this regard, our in vitro temporal model supports the idea that RA can potentiate lineage maturation once progenitors are formed, at least in part by enhancing responsiveness to lineage-promoting cytokine cues. In particular, our erythroid data are consistent with a model in which RA cooperates with EPO signaling to promote late-stage erythroid maturation, whereas the extent to which analogous RA-dependent mechanisms regulate myeloid/macrophage maturation in vivo remains to be determined. One of the most intriguing and clinically relevant findings of our study was the ability of RA to bias lineage commitment in a cytokine-dependent manner. Specifically, under EPO stimulation, RA robustly promoted erythroid differentiation, as shown by the increased CFU-E and BFU-E colony output, elevated expression of erythroid master regulators such as GATA, KLF1, and NFE2, which orchestrate hemoglobin synthesis, erythroid maturation, and enucleation [42–44], and subsequent activation of globin genes. Importantly, RA also upregulated key components of the EPO receptor (EPOR) signaling cascade, including EPOR itself, STAT5A/B, and BCL-XL, indicating a mechanistic cooperative interaction that amplified erythropoiesis. This cooperative effect was dose-dependent, with low-dose (0.1 μM) RA maximizing erythroid output without impairing progenitor viability. Conversely, under myeloid-inducing conditions with M-CSF/GM-CSF, RA acted cooperatively with external signals to amplify macrophage maturation. The proportional reduction in erythroid markers observed in this context reflects reinforcement of the myeloid-biased cytokine environment rather than an independent RA-driven suppression of erythropoiesis. We observed a pronounced expansion of CD11b+CD14+ macrophages alongside reduced erythroid markers (CD71+, CD235a+), underscoring the ability of RA to integrate extrinsic cytokine cues and bias lineage outputs. These observations are consistent with those of previous studies in which RA produced in tumor microenvironments (TMEs) polarized monocyte differentiation toward tumor-associated macrophages (TAMs) and away from dendritic cells (DCs), reshaping immune cell composition [45]. In addition, RA has been shown to play an essential role in the proper development and differentiation of colonic CD169+ macrophages [46], reinforcing its importance in macrophage biology. Our colony formation assays under varying cytokine environments, including EPO, M-CSF, and GM-CSF, revealed that short-term RA treatment significantly biased progenitor differentiation toward the erythroid lineage when combined with EPO, while reducing myeloid colony formation. However, in the presence of M-CSF/GM-CSF without EPO, RA supported macrophage differentiation, indicating that RA alone does not override the external cytokine milieu but rather enhances cellular responsiveness to dominant external signals. This implies that RA may act as a sensitizer or amplifier, tuning progenitor responsiveness to lineage-directing cytokines rather than imposing a fixed intrinsic fate decision. Importantly, in contexts in which EPO and myeloid cytokines are combined, EPO likely provides a stronger lineage-promoting signal, as RA-treated cells showed preferential commitment to the erythroid lineage. The integration of these findings with previous reports provides a more nuanced understanding of the dualistic and context-dependent role of RA. Previous studies reported that RA inhibition via DEAB enhanced early hematopoietic commitment and maintained primitive progenitor pools [20, 32], while RA activation promoted maturation and differentiation. This biphasic model reconciles conflicting observations in the literature, implying that the effects of RA depend not only on concentration but also on developmental timing and external cues [32, 41, 47]. Importantly, our data add temporal granularity to the concentration-dependent framework described by Luff et al. [35]. While Luff et al. demonstrated concentration-dependent roles for RA in promoting specific hemogenic endothelial subpopulations at intermediate developmental stages, our dose-titration analysis reveals that at the earlier mesoderm induction window (Step I, days 0–5), RA exerts a uniformly inhibitory effect across all concentrations tested (0.01–1 µM), suppressing the KDR⁺CD31⁻ mesodermal, CD144⁺CD43⁻CD73⁻ hemogenic endothelial, and CD34⁺CD45⁺ hematopoietic progenitor populations without a biphasic pattern. In contrast, the biphasic concentration–response we observed at Step II, where low-dose RA (0.01–0.1 µM) modestly reduces progenitor frequency while paradoxically increasing the mature hematopoietic fraction, and high-dose RA (1 µM) inhibits both, aligns directly with the HE-stage patterns reported by Luff et al. [35]. Together, these stage-resolved concentration–response profiles define a developmental “window of competence” for RA signaling in hPSC hematopoiesis: uniformly inhibitory during primitive streak and mesoderm formation (Step I), qualitatively biphasic during hemogenic endothelial transition (Step II), and dose-dependently promoting progenitor maturation during late progenitor specification (Step III). This temporal framework highlights the importance of specifying both developmental stage and concentration when characterizing RA activity, and provides a unifying context for reconciling the diverse roles of RA reported across hematopoietic differentiation systems. Moreover, recent studies have shown that RA interacts with microenvironmental factors, including the bone marrow niche, to regulate HSC dormancy and lineage allocation, thus underscoring its physiological relevance [9, 20, 21, 48]. Our study advances the field by having directly tested the lineage-modulating effects of RA in the context of hPSC-derived hematopoiesis, a setting with major translational relevance. While generating transfusion-ready erythrocytes or functional macrophages from hPSCs remains a challenge, primarily due to inefficient lineage specification and maturation of progenitors [1], our findings imply that precise modulation of RA is crucial for functional differentiation. RA treatment, when timed to the late differentiation phases and combined with lineage-specific cytokines, could improve the yield and purity of desired blood cell types. This has immediate implications for in vitro disease modeling and the development of regenerative blood therapies. Despite these promising findings and their translational potential, our study has several limitations that warrant consideration. First, our functional assessments were conducted exclusively in vitro. Although recent landmark advances have demonstrated long-term multilineage engraftment of hPSC-derived hematopoietic cells in immunodeficient models [3], further in vivo studies will be needed to evaluate the physiological function, safety, and persistence of the RA-modulated hematopoietic lineages generated in our system. Second, although our data demonstrate that RA cooperates with specific cytokine networks, the precise molecular mechanisms underlying this effect, including direct transcriptional targets and chromatin-level regulation by RA within heterogeneous progenitor populations, remain to be defined. Future studies using single-cell transcriptomics and chromatin accessibility profiling, together with direct mapping of RAR-associated regulatory sites, will be important to clarify the molecular architecture of this RA-driven lineage bias. Third, while the increases in mature cell numbers observed in this study are consistent in direction, they are modest in magnitude and are unlikely to be sufficient for clinical-scale blood product manufacturing as a standalone modification. The translational significance of these findings is therefore best understood in two complementary respects. The improvements in functional maturation, including increased enucleation rates, elevated β-globin expression, and enhanced lineage purity, represent gains in product quality that are as clinically important as absolute yield. Additionally, RA modulation, as a chemically defined and dose-tunable intervention, is well suited to serve as one component within a broader manufacturing strategy that integrates cytokine optimization, bioreactor scale-up, and gene editing approaches to achieve clinically relevant outputs. Whether the interaction between RA and lineage-specific cytokines represents true synergy, in which the combined effect exceeds the sum of individual contributions, remains to be formally established through factorial experimental designs with consistent normalization, and constitutes an important direction for future work. In conclusion, RA has emerged as a versatile and tunable regulator of hematopoietic differentiation that is capable of orchestrating both the maturation and lineage fates of hPSC-derived progenitors in a dose- and context-dependent manner. To elucidate its role fully, further studies are required to explore the downstream effectors of RA, including noncoding RNAs, epigenetic regulators, and metabolic pathways. In addition, integration of RA modulation with gene editing or synthetic biology approaches may enable scalable generation of lineage- biased blood products, thus advancing the clinical application of hPSC-based hematopoiesis. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 1. Supplementary Material 2. Supplementary Material 3.