β-Hydroxybutyrate Exposure is Associated with Increased TH⁺ Neuron Abundance and Dopaminergic-Related Transcriptional Features in Human Midbrain Organoids

Abstract

Neuronal differentiation is a highly regulated process in which morphogens establish cellular identities through coordinated transcriptional programs. However, the contribution of physiologically derived metabolites to human neural development remains poorly understood. β-hydroxybutyrate (bhb) has been recognized as an inflammatory, epigenetic, energetic, and neuroprotective regulator. We differentiated midbrain floor plate neural precursor cells (mfNPC) from human embryonic stem cells to generate midbrain organoids (MBOs), which were exposed to bhb. Although dopaminergic differentiation occurred in control MBOs, treatment with 1 mM bhb for 16 days significantly increased both the number of Tyrosine hydroxylase–positive (TH⁺) neurons, and the corresponding transcripts assessed by bulk RNA sequencing. Such transcriptomic profiling revealed modest but consistent bhb-associated changes in trophic, neuronal and dopaminergic transcripts. To further characterize these changes, we performed gene set enrichment analysis using curated cell phenotypes-associated transcriptional programs. These analyses indicated that bhb treatment is associated with shifts in the enrichment scores of dopaminergic programs in day 16 MBOs. At the epigenomic level, we used CUT&RUN assays to map H3K27 acetylation, and the selected candidates H3K27ac regions after bhb treatment were linked to their closest neighboring genes, suggesting positive enrichment of neural and dopaminergic gene ontology categories at both time points. Unsupervised analysis of H3K27ac-enriched regions revealed three major clusters detected across differentiation, although present in both control and bhb-treated MBOs. Motif enrichment analysis of clustering regions identified distinct predicted transcription factor binding motifs in mfNPC and MBOs. Finally, we integrated this motif analysis, enhancer annotation, and our transcriptomic data on a web platform, Enhancer Network Explorer, to explore relationships between H3K27ac profiles, transcriptional changes and bhb exposure during human MBOs differentiation. Together, our results indicate that bhb treatment is associated with increased TH-positive cell abundance and with population-level changes related to neurodevelopmental transcriptional programs. These findings suggest that bhb might support early human midbrain differentiation and provide a starting point for future studies addressing the mechanisms linking metabolic cues, chromatin-associated regulation and dopaminergic differentiation.

Graphical Abstract

Human embryonic stem cells were differentiated into midbrain floor plate neural progenitor cells and subsequently into midbrain organoids (MBOs) under control conditions or in the presence of β-hydroxybutyrate (bhb). Bhb treatment enhanced dopaminergic neuronal differentiation, as evidenced by increased TH⁺ cells. The administration of bhb modulates transcription factor–enhancer interactions, leading to changes of H3K27ac-marked regulatory regions and activation of enhancer networks. This reorganization promotes the expression of key dopaminergic genes, including TH, CORIN, FOXA2, ASCL2 and KCNJ6. Together, these findings reveal that bhb acts as a metabolic regulator of epigenetic and transcriptional programs that drive dopaminergic lineage specification.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s10571-026-01771-1.

Introduction

Ketone bodies (KB) are small molecules derived from the beta-oxidation of fatty acids; their production occurs predominantly in the mitochondria of hepatocytes. Notably, these molecules can serve as an alternative energy source for extrahepatic tissues when glucose levels are low, since they can be distributed through circulation (Berg et al. 2019; Steiner 2020). The principal KB are acetone, acetoacetate, and beta-hydroxybutyrate (bhb), with bhb comprising up to 70% of the KB pool in the body (Berg et al. 2019). Bhb exhibits diverse functions: it can be antioxidant (Haces et al. 2008; Rojas-Morales et al. 2020), anti-inflammatory (Youm et al. 2015; Achanta and Rae 2017), and neuroprotective (Li et al. 2024; Shang et al. 2024). Its protective role in models of neurodegenerative diseases, such as Parkinson’s disease (Imamura et al. 2006; Jiang et al. 2022; Yu et al. 2023), Alzheimer’s disease (Jin et al. 2023), and others (Lehto et al. 2022; Wang et al. 2023; Li et al. 2024; Qiao et al. 2024; Holstein et al. 2025) is well documented.

In contrast, the potential role of bhb during early human neurodevelopment remains less explored, although the levels of this KB are elevated during embryo development, as well as in the neonatal period (Chan et al. 2006; Nagai et al. 2010; Steiner 2020; Parenti et al. 2024). Importantly, the human placenta expresses monocarboxylate transporters, MCT, which facilitate bhb transport from the mother to the fetus (Iwanaga and Kishimoto 2015; Thomas et al. 2023; Molloy and Barry 2024). The possibility that the developing human fetal brain can use bhb as an energy substrate, is supported by the fact that it is consumed up to 1.47 times more efficiently (Adam et al. 1975).

During embryonic development, dopaminergic neurons arise from ventral midbrain progenitors in a spatially- and temporally-coordinated process governed by morphogens, and sequential transcriptional programs, where early patterning cues establish progenitor identity and progressively give rise to lineage-restricted neuronal populations. Importantly, subtle variations in the timing, intensity, and combination of signaling pathways can significantly alter dopaminergic lineage specification, highlighting the sensitivity of these developmental programs to environmental and intrinsic signals (Arenas et al. 2015). The specification of ventral midbrain dopaminergic neurons during development is tightly regulated by morphogen gradients, including SHH, WNT, and FGF8 signaling (Joyner et al. 2000; Puelles et al. 2004; Placzek and Briscoe 2005; Bonilla et al. 2008), that triggers the action of key transcription factors (TF), such as FOXA2 and LMX1A, that define floor plate progenitors and dopaminergic lineage identity (Andersson et al. 2006; Ferri et al. 2007; Jo et al. 2016; Fiorenzano et al. 2025). This developmental process follows a well-defined transcriptional program, in which early patterning factors give rise to lineage-specific regulators such as NURR1 (NR4A2) and PITX3, which are essential for dopaminergic neuron survival, maturation and functional identity (Zetterström et al. 1997; Ásgrímsdóttir and Arenas 2020).

Human brain organoids provide an innovative in vitro platform that recapitulates key cellular and molecular events of early embryonic brain development, including region-specific patterning, progenitor proliferation, and neuronal maturation, allowing the study of processes that are otherwise inaccessible in vivo (Lancaster et al. 2013; Jo et al. 2016; Fernandes et al. 2021). Importantly, these three-dimensional systems follow intrinsic developmental trajectories, and enable the emergence of spatially organized neural populations, thereby providing a better environment, compared to traditional monolayer cultures (Lancaster et al. 2013; Smits et al. 2019; Kelley and Pașca 2022; Fiorenzano et al. 2025). As such, organoids represent an intermediate model system connecting simplified in vitro differentiation, and some aspects of the complexity of the human brain. In this regard, neuronal differentiation is a highly dynamic and complex process, requiring precise molecular coordination to progress from a neural precursor to a non-dividing terminally differentiated neuron, with epigenetic and transcriptional changes related to functions as excitability and neurotransmitter identity. The variety of neuronal subtypes in the human brain, results from the different combination of signaling molecules present in the developing microenvironments (Bhaduri et al. 2020; Gordon et al. 2021; Kelley and Pașca 2022).

Multiple mechanisms drive neuronal differentiation, including non-coding RNA activity, TF binding, enhancer activation, histone post-translational modifications, and chromatin remodeling. In this context, studies have explored the roles of morphogens and small molecules during neuronal differentiation, as well as their impact in chromatin changes during dopaminergic neuron differentiation of human embryonic stem cells (Meléndez-Ramírez et al. 2021). Cellular metabolites, under physiological conditions, can directly induce chromatin modifications (Li et al. 2018). In addition to its metabolic roles, bhb can serve as a substrate that incorporates to histones, and can modulate the activity of chromatin-modifying enzymes (Cai et al. 2011; Shimazu et al. 2013; Li et al. 2018). Specifically, it has been reported bhb, increases the acetylation level patterns of the histone H3 lysine 27 (H3K27ac) in neurons, enhancing the expression of neurotrophic factors, such as brain-derived neurotrophic factor (BDNF), that confer neuroprotection (Sleiman et al. 2016; Marosi et al. 2016; Hu et al. 2018). Currently, it is unknown if bhb can induce the expression of trophic factors during neuronal differentiation of human cells.

The effect of bhb on early neural differentiation-associated programs has not been reported. A preprint reports that 0.02 mM bhb is capable of inducing proliferation of primary neural stem cells from the rat cerebral cortex, and immortalized mouse cerebellar neural stem cells, through ERK phosphorylation (Wang et al. 2024), suggesting a potential role of bhb during early phases of neural lineage progression. These observations raise the possibility that bhb may influence transcriptional and epigenetic programs, that operate during dopaminergic lineage specification and maturation in a developmental process. Given the tightly regulated, stage-dependent nature of dopaminergic differentiation, metabolic cues such as bhb might act as modulators, influencing maturation dynamics and regulatory network activity, rather than acting as a primary determinant of cell fate.

Here, we investigate whether bhb exposure is associated with population-level changes in transcriptional programs and with H3K27ac present in identified regions, during early human midbrain organoid differentiation. By integrating transcriptomic and epigenomic analyses, we aim to characterize how this relevant metabolite relates to neurodevelopmental, trophic, and dopaminergic maturation-associated programs in human midbrain organoids, although the underlying mechanisms remain to be functionally established.

Materials and Methods

Further methodological details, including experimental procedures, sequencing workflows, bioinformatic analyses, data processing strategies, and parameter settings, are provided in the Supplementary Methods.

Human Embryonic Stem Cell Line Culture

Human embryonic stem cells (hESCs; H9, WA09) were maintained under feeder-free conditions on Matrigel-coated plates in mTeSR1 medium at 37 °C in a humidified incubator with 5% CO₂. Cells were routinely tested for mycoplasma prior to experimental use.

Culture and Differentiation of mfNPCs from Human ESC

Midbrain floor plate neural progenitor cells (mfNPCs) were derived from hESCs following previously described protocols for ventral midbrain patterning and expansion.(Smits et al. 2019) Briefly, embryoid bodies were generated from dissociated hESCs and subjected to neuroectodermal induction using defined signaling conditions including CHIR99021, SAG, LDN193189, and SB431542, followed by transition to N2B27-based differentiation medium. At day 8, embryoid bodies were dissociated into smaller aggregates and plated onto laminin-coated plates for mfNPC expansion. Cells were passaged weekly, and stabilized mfNPC cultures (passages 7–9) were used for downstream two-dimensional and three-dimensional differentiation experiments.

Two-dimensional differentiation experiments were performed as exploratory assays, whereas three-dimensional mfNPC-derived organoids were used for all transcriptomic and epigenomic analyses.

Organoid Generation

Midbrain organoids (MBOs) were generated from pre-patterned midbrain floor plate neural progenitor cells. In this study, the term “midbrain organoids” or “MBO” is used to refer to these mfNPC-derived three-dimensional neural aggregates, in line with previously published protocols (Nickels et al. 2020) and for consistency throughout the manuscript. First, dissociated mfNPCs were seeded into AggreWell-800 plates to generate uniform aggregates of approximately 6,000 cells. After 24 h, aggregates were transferred to non-adherent three-dimensional culture conditions and patterned through staged withdrawal of SB431542 and LDN193189, followed by progressive reduction of CHIR99021. From day 8 onward, organoids were maintained in maturation medium supplemented with Brain-Derived Neurotrophic Factor (BDNF), Glial cell line-Derived Neurotrophic Factor (GDNF), Activin A, Transforming Growth Factor-beta 3 (TGF-β3), dibutyryl-cyclic AMP (db-cAMP), and γ-secretase inhibitor (DAPT) until day 16, with medium changes performed every other day.

Bhb Treatment

For bhb treatment, DL-β-hydroxybutyric acid was added at a final concentration of 1 mM from day 0 to day 16 of organoid differentiation. A fresh stock solution was prepared at each medium change using the corresponding stage-specific N2B27 medium as vehicle. Control organoids were cultured in parallel under identical differentiation conditions without bhb supplementation.

Immunofluorescence, Whole-Mount Staining, and Image Analysis

mfNPCs were analyzed by immunofluorescence after fixation, permeabilization, and incubation with antibodies against FOXA2, SOX2, NESTIN, β-III-tubulin, and Tyrosine hydroxylase (TH). Whole-mount immunostaining was performed in MBOs at days 8 and 16 after fixation, tissue clearing, permeabilization, and prolonged antibody incubation under agitation. Imaging was carried out using a Zeiss Lightsheet Z.1 fluorescence microscope (RRID: SCR_020919), and three-dimensional reconstruction was performed using Imaris (RRID: SCR_007370). Dopaminergic neuron quantification was based on TH⁺ cell counts obtained from systematically sampled optical sections throughout each organoid. Total TH⁺ neuron number and density were estimated using a stereological section-sampling approach with geometric correction to minimize overcounting across adjacent sections. Data are presented as median and interquartile range, and statistical significance was assessed using the Wilcoxon rank-sum test.

RNA Isolation, Library Construction, and RNA Sequencing

Total RNA was extracted from mfNPCs and MBOs at days 8 and 16 using the RNeasy Plus Micro Kit. For organoid samples, 20 MBOs were pooled per biological replicate, whereas 5 × 10⁶ mfNPCs were collected per replicate. RNA quality was assessed prior to poly(A)+ mRNA enrichment, library preparation, and paired-end 150-bp sequencing on an Illumina NovaSeq platform. Previously generated hESC RNA-seq data were incorporated from GEO accession GSE153005 (Meléndez-Ramírez et al. 2021).

Sequencing reads were processed using the snakePipes mRNA-seq workflow, including read quality assessment, adapter trimming, alignment to the human reference genome (GRCh38), and gene-level quantification. Differential gene expression analysis was performed in R using DESeq2. One biological replicate from the MBO control day 8 group was excluded from downstream RNA-seq analyses based on outlier behavior during quality control and exploratory analysis. For selected graphical representations, expression values were additionally normalized as FPKM.

Projection of Cell-Type Transcriptional Programs by ssGSEA

To estimate the relative enrichment scores of cell-type-associated gene expression programs across differentiation stages and treatment conditions, single-sample Gene Set Enrichment Analysis (ssGSEA) was performed using the GSVA package (Hänzelmann et al. 2013). Variance-stabilized expression values derived from DESeq2 were used as input after restricting the matrix to protein-coding genes, collapsing duplicated gene symbols, and harmonizing gene names. Literature-curated gene signatures representing pluripotent, progenitor, glial, and neuronal midbrain-related states were formatted as gene sets and evaluated using ssGSEA; these defining gene sets are listed in Supplementary Table S2. Enrichment scores were visualized with heatmaps and box plots, and differences between MBO groups were assessed using linear modeling with limma and Benjamini–Hochberg correction.

CUT&RUN and Epigenomic Analysis

CUT&RUN experiments were performed using the CUTANA ChIC/CUT&RUN Kit following the manufacturer’s instructions. For each biological replicate, 5 × 10⁵ cells were used. Cells were immobilized on concanavalin A-coated beads and resuspended in antibody buffer. H3K27ac profiling was performed using 0.5 µg of anti-H3K27ac antibody per reaction. Rabbit IgG control antibody was used as negative control. Samples were incubated overnight at 4 °C, washed, and subsequently treated with pAG-MNase. Reactions were stopped by addition of stop solution containing E. coli spike-in DNA as an internal normalization control. DNA was then purified and used for library preparation with the CUTANA CUT&RUN Library Prep Kit prior to sequencing. Due to technical and material limitations the number of biological replicates was constrained, which prevented the inclusion of triplicates for all conditions. Results derived from conditions with n = 2 biological replicates are interpreted cautiously. CUT&RUN experiments were performed with the following number of independent biological replicates: hESC (n = 2), mfNPC (n = 3), MBO ctrl day 8 (n = 2), MBO bhb day 8 (n = 2), MBO ctrl day 16 (n = 3), and MBO bhb day 16 (n = 2).

CUT&RUN sequencing data were processed using the snakePipes DNA-mapping and ChIP-seq workflows under CUT&RUN-optimized settings. Reads were aligned to a hybrid reference genome containing the human GRCh38 assembly and the Escherichia coli spike-in genome, and host- and spike-in-aligned reads were separated for normalization and quality control.

Peak calling was performed independently for each biological replicate using MACS2 with matched IgG controls. Condition-specific consensus peak sets were generated by retaining reproducible peaks detected in at least two biological replicates, and a global non-redundant consensus peak set was subsequently defined across all conditions. Read counts per consensus region were quantified with BEDTools and analyzed in R using DESeq2 package. Because some CUT&RUN comparisons were based on two biological replicates, exploratory differential enrichment analyses were interpreted cautiously and used nominal p-values together with effect size thresholds to identify hypothesis-generating regions. CUT&RUN signal was visualized using deepTools as normalized H3K27ac signal relative to IgG background or as spike-in-scaled log₂(H3K27ac/IgG) enrichment over peak-centered regions. Full experimental, alignment, peak calling, normalization, and visualization procedures are described in the Supplementary Methods.

Enhancer Network Explorer Platform

To comprehensively analyze and test the interactions between TF, enhancers and target genes, we designed a platform that integrates the information generated in this work. The code can be found in https://doi.org/10.5281/zenodo.17645300. This platform can be accessed at: https://132.247.213.138:15443/remote/login?lang=sp. The updated url is: https://computo.ifc.unam.mx/enhancer-network-explorer.

Statistical Analysis

Statistical analyses were performed using R 4.4.0. Normality of the data was assessed using the Shapiro–Wilk test. Data are presented as median and interquartile range. Differences between groups were evaluated using the Wilcoxon rank-sum test.

For multiple testing, p-values were adjusted using the Benjamini–Hochberg method to control the false discovery rate (FDR) and are reported as adjusted p-values (padj). Unless otherwise specified, p-values reported in figures and text correspond to unadjusted values; adjusted p-values are explicitly indicated where applicable. Statistical significance was defined as p < 0.05 or adjusted p-values (FDR) < 0.05, as appropriate.

Data Availability

RNAseq data and CUT and RUN regions with H3K27ac were deposited at the NCBI Gene Expression Omnibus (GEO) with reference numbers GSE313206 and GSE313207, respectively.

Results

Bhb Exposure is Associated with Pro-dopaminergic Features and Changes in H3K27ac Signal in Monolayer Cultures

To determine whether bhb supplementation is associated with changes in early dopaminergic differentiation-related features, we tested it on monolayer differentiation of human ESCs, using a published protocol for midbrain dopaminergic differentiation (Smits et al. 2019). Given the limited data on fetal brain concentrations of bhb, a concentration of 1 mM was selected as a proxy to physiologically relevant upper-range levels. Bhb treatment was initiated on day 0 and maintained through 14 days of the monolayer differentiation protocol, assessing TUJ1, FOXA2, KI67, and acetylation of lysine 27 of Histone H3 (H3K27ac) at day 6, and MAP2, TH and H3K27ac at day 14. Quantification of positive cells indicated significant increases for FOXA2 (day 6) and TH (day 14). Since bhb can influence chromatin-associated histone modifications, we also assessed H3K27ac patterns at days 6 and 14, observing significant decreases after incubation with bhb (Fig. S1). Additionally, Sholl analysis performed in monolayer differentiation cultures identified differences between conditions, including increases in Sholl area under the curve, maximum radius, and total intersections in bhb-treated cells at day 14 (Fig. S1H). Notably, these analyses were conducted independently of neuronal quantification, which showed an increased number of TH⁺ neurons under bhb treatment.

Midbrain Organoids Recapitulate Early Dopaminergic Differentiation Features

To further evaluate the generation of early midbrain and dopaminergic differentiation features, we generated midbrain organoids (MBOs) from hESC–derived mfNPCs following a previously described protocol with defined modifications (Fig. 1A) (Nickels et al. 2020). Brightfield imaging confirmed the progressive formation of uniform aggregates and their subsequent growth into three-dimensional structures (Fig. 1B). The identity of the starting mfNPC population was validated by immunofluorescence, showing expression of canonical ventral midbrain progenitor markers, including FOXA2, SOX2, and NESTIN (Fig. 1C). Consistent with this, three-dimensional cultures retained co-expression of FOXA2 and NESTIN, supporting maintenance of ventral midbrain progenitor features during early stages of organoid formation (Fig. 1D). We next assessed the temporal emergence of neuronal and dopaminergic features within the organoids. Immunostaining for the pan-neuronal marker TUJ1 and the dopaminergic marker TH revealed progressive neuronal differentiation. At day 8 (MBO d8), TUJ1⁺ cells were organized in rosette-like structures, and a small population of TH⁺ cells was already detectable (Fig. 1E). By day 16 (MBO d16), organoids exhibited increased neuronal complexity, with a higher proportion of TH⁺ cells co-expressing TUJ1 (Fig. 1F), consistent with the progression of early dopaminergic differentiation.

Fig. 1.

Midbrain organoids recapitulate early dopaminergic differentiation features and bhb treatment is associated with increased TH⁺ cell density at later stages. A Schematic overview of the differentiation protocol used to generate midbrain organoids (MBOs) from human embryonic stem cells (hESCs), including key developmental stages and timing of signaling modulators and bhb treatment. B Representative bright-field images showing morphological progression from hESCs to mesencephalic floor plate neural progenitor cells (mfNPCs), aggregation in AggreWell plates, and formation of MBOs. C Immunofluorescence of mfNPCs showing expression of FOXA2, SOX2, and NESTIN, with Hoechst nuclear staining. D Immunofluorescence of three-dimensional mfNPC aggregates showing co-expression of FOXA2 and NESTIN at two magnifications. E–F Immunofluorescence of MBOs at day 8 (E) and day 16 (F) stained for TUJ1 (neuronal marker) and Tyrosine hydroxylase (TH). Insets show higher-magnification views of TH⁺ regions. G, I Representative whole-organoid TH staining at day 8 (G) and day 16 (I) in control (ctrl) and bhb-treated conditions. H, J Quantification of TH⁺ cells density in MBOs at day 8 (H) and day 16 (J) using a stereological sampling approach. TH⁺ cells were counted in optical sections spaced 50 μm apart, and total neuron number was estimated and normalized to organoid volume. Data are presented as box plots (median and interquartile range; whiskers indicate minimum–maximum values) from n = 3 independent organoids per condition. Statistical significance was assessed using the Wilcoxon rank-sum test (*p-value < 0.05)

Bhb Treatment is Associated with Increased TH⁺ Cell Numbers at Day 16

To determine whether bhb exposure is associated with changes in TH⁺ cell abundance during organoid differentiation, we quantified TH⁺ neurons in organoids treated with 1 mM bhb, compared to controls. The number of TH-positive neurons was normalized by the MBO volume, by counting optical sections separated by 50 μm. At day 8, no significant difference in TH⁺ neuron density was observed between conditions (Fig. 1G–H). In contrast, at day 16, bhb-treated MBOs showed a significant increase (p-value < 0.05) in TH⁺ neuron density, compared to controls (Fig. 1I–J).

Midbrain Differentiation Transcriptional Trajectories are Preserved After bhb Treatment

To study transcriptomic differences between bhb-treated MBOs and their corresponding controls, we performed RNA-seq across three critical time points of midbrain organoid differentiation: mfNPCs, MBO d8 (Ctrl and bhb) and MBO d16 (Ctrl and bhb). We also included in the analysis previously generated data for pluripotent hESC (Meléndez-Ramírez et al. 2021). To identify differentially expressed genes (DEGs), we performed pairwise comparisons across the six sample groups (hESC, mfNPC, MBO d8 ± bhb, and MBO d16 ± bhb), applying a threshold of |log₂FC| > 2, adjusted p-value < 0.05. Following quality control and exploratory data analysis, one biological replicate from the MBO control day 8 group was identified as an outlier and excluded from DEG analysis. This analysis identified 15,613 DEGs across groups, reflecting the extensive transcriptional remodeling associated with the transition from pluripotency to neural differentiation states in organoids (Fig. 2A). Unsupervised clustering of these genes revealed six major expression patterns corresponding to distinct developmental programs. Early clusters were highly expressed in hESCs and mfNPCs and were progressively downregulated, being enriched for processes related to DNA replication, stem cell proliferation, and early developmental programs. In contrast, late clusters showed increased expression at day 16 and were associated with neuronal maturation, including axonogenesis, synaptic signaling, and neurotransmitter-related processes. These late-stage clusters also included terms and genes related to neuronal maturation and dopaminergic-associated processes. Particularly, these late clusters were linked to dopaminergic function. Intermediate clusters captured transitions associated with neural specification, regional identity, and differentiation processes. Importantly, across all identified clusters, bhb-treated MBO closely mirrored their corresponding control conditions.

Fig. 2.

Global transcriptional programs during midbrain organoid differentiation are largely preserved under bhb treatment. A Heatmap representing the transcriptomic landscape of hESCs, mfNPCs, and midbrain organoids (MBOs) at day 8 and day 16, under control (ctrl) and bhb conditions. Differential expression analysis (|log₂FC| > 2, adjusted p-value < 0.05) identified 15,613 differentially expressed genes (DEGs) across all groups. Rows correspond to DEGs and columns to individual biological replicates. Gene expression values were transformed to z-scores for visualization and clustered using hierarchical clustering, yielding six major expression clusters. Gene Ontology (GO) enrichment analysis was performed for each cluster; bar plots on the right indicate significantly enriched biological processes, with bar length representing –log₁₀(q-value) and numbers indicating the number of genes associated with each term. B Distribution of DEGs by transcript biotype (lncRNA and protein-coding genes) in pairwise comparisons between hESCs and differentiated states (mfNPCs, MBO ctrl d8, MBO bhb d8, MBO ctrl d16, and MBO bhb d16). C Distribution of DEGs by transcript biotype in comparisons between mfNPCs and MBOs (day 8 and day 16, ctrl and bhb conditions). Bars represent the number of significantly up- and down-regulated genes (|log₂FC| > 2, adjusted p-value < 0.05). RNA-seq analysis was performed using n = 3 biological replicates per condition, except for MBO ctrl day 8 (n = 2 after exclusion of one outlier sample identified during quality control)

To further explore transcriptional composition, we examined the distribution of differentially expressed transcripts by biotype. Comparisons between hESCs and differentiated states revealed a marked reduction in lncRNA expression as differentiation progressed, whereas protein-coding genes exhibited more balanced bidirectional changes (Fig. 2B). This trend remained consistent across comparisons between mfNPCs and MBOs, as well as between control and bhb-treated samples (Fig. 2C). Together, these findings indicate that while extensive transcriptional remodeling occurs during midbrain organoid differentiation, bhb treatment does not induce large-scale shifts in global transcriptional programs. This supports a model in which bhb-associated effects are modest and context-dependent, rather than driven by broad transcriptional reprogramming, consistent with Principal Component Analysis (PCA) using the 200 most variable genes (Fig. S2A).

Bhb Treatment is Associated with Late-Stage Changes in Neurodevelopmental and Dopaminergic Phenotype-Associated Programs

To further characterize transcriptional dynamics during differentiation and evaluate whether bhb treatment influences cell-type–associated transcriptional programs, we performed ssGSEA, using curated gene signatures that define neural phenotypes (Nichols and Smith 2012; Takashima et al. 2014; La Manno et al. 2016; Fiorenzano et al. 2021; Budinger et al. 2026). Across the differentiation trajectories, ssGSEA revealed a progressive transition from pluripotency-associated programs in hESCs toward neural progenitor and dopaminergic-related transcriptional programs in MBOs. Pluripotency signatures were enriched in hESCs and decreased upon transition to mfNPCs and organoids, whereas progenitor-associated and ventral midbrain programs increased at intermediate stages and remained elevated in MBOs. At later stages of differentiation (day 16), gene sets associated with dopaminergic neuroblast, and mature dopaminergic neurons showed increased enrichment scores, consistent with the emergence of dopaminergic-associated transcriptional features within MBO. Notably, alternative neuronal programs, including serotonergic-associated signatures, were also examined and did not display consistent enrichment differences between conditions, supporting the interpretation that the observed transcriptional changes occur within the expected midbrain developmental context (Fig. 3A).

Fig. 3.

Modulation of cell-type–associated transcriptional programs by bhb during midbrain organoid differentiation. A Heatmap showing ssGSEA enrichment scores for curated gene sets representing neural developmental states across hESCs, mfNPCs, and MBOs at day 8 and day 16, in control (ctrl) and bhb-treated conditions. Each column represents a biological replicate and each row a gene set. Color scale indicates relative enrichment score. The analysis reveals a progressive transition from pluripotency-associated program to neuronal progenitor and dopaminergic-related transcriptional programs during differentiation, with similar global patterns between conditions. B Boxplots showing ssGSEA enrichment scores for selected gene sets across differentiation stages and conditions. The dopaminergic neuron program shows increased enrichment at day 16, with a significant increase in bhb-treated organoids compared to controls. In contrast, A10-like dopaminergic signatures are relatively enriched in control samples, whereas A9-like signatures remain comparable between conditions. Boxplots display median, interquartile range, and individual data points (n = biological replicates per group as indicated in Methods). Statistical significance was assessed using linear modeling (limma), with Benjamini–Hochberg FDR correction; adjusted p-values < 0.05 were considered significant. C Expression of selected dopaminergic-related genes derived from RNA-seq data (FPKM values) at day 16, including FOXA2, KCNJ6, TH, and CALB2. Genes associated with dopaminergic identity (FOXA2, KCNJ6, TH) show significantly increase expression in bhb-treated organoids, whereas CALB2, associated with A10-like dopaminergic populations, is reduced. Importantly, ssGSEA enrichment scores reflect the relative activity of predefined transcriptional programs and do not directly measure cell-type abundance. Therefore, observed differences are interpreted as changes in transcriptional program activity at the population level rather than direct evidence of altered cell-type proportions

Despite this overall similarity, quantitative comparison of enrichment scores revealed modest but consistent treatment-associated differences at day 16 (Fig. 3B). In particular, the global dopaminergic neuron program showed a significant increase in bhb-treated organoids compared to controls (adjusted p-value < 0.05). In contrast, A10-like dopaminergic signatures were relatively enriched in control organoids and reduced in bhb-treated samples, whereas A9-like signatures did not significantly differ between conditions. To further examine these observations at the gene level, we analyzed the expression of selected dopaminergic markers. Consistent with the enrichment analysis, genes such as FOXA2, KCNJ6 (GIRK2) and TH showed significantly increased normalized expression in bhb-treated MBOs at day 16, while CALB2, associated with A10-like dopaminergic populations, showed reduced expression (Fig. 3C). These findings support treatment-associated differences in selected dopaminergic maturation-related transcriptional features at day 16, with the subtype-associated comparison primarily reflecting reduced A10-like scores.

To provide additional information of gene expression patterns of key biological processes within dopaminergic-related pathways, relevant to midbrain and dopaminergic neuron development, a heatmap was built across differentiation stages. This analysis revealed expression patterns consistent with progression from progenitor states to dopaminergic differentiation-associated profiles, including early expression of developmental regulators and increased expression of genes associated with neuronal function at later stages (Fig. S2B). This directed analysis also provided an initial overview of expression trends in genes related to neurodevelopmental and neurotrophic pathways that may be influenced by bhb treatment. However, we did not find differences between control and bhb-treated MBOs in genes such as BDNF, CDNF nor PITX3; only SOS1 and PHOX2B showed decreased expression after bhb in day 16 organoids (data not shown). These results are presented to complement the global transcriptomic and ssGSEA analyses. Together, these results indicate that dopaminergic transcriptional programs emerge progressively during MBO differentiation and show modest treatment-associated differences at later stages. Importantly, these changes are consistent with modulation of transcriptional program scores at the population level, rather than evidence of large-scale transcriptional reprogramming or direct cell-type proportion changes.

Bhb Exposure is Associated with Stage-Dependent Transcriptional Changes in Midbrain Organoids

To formally analyze and further characterize transcriptional differences associated with bhb treatment, we performed direct pairwise comparisons between control and bhb-treated MBOs, at day 8 and at day 16, separately. Differential expression analysis revealed stage-dependent sets of genes associated with bhb exposure at day 8 or at day 16, as shown by volcano plots (Fig. 4A, B). At day 8, a small number of genes reached significance, indicating that early transcriptional differences between conditions are modest (Fig. 4A). In contrast, at day 16, a larger subset of DEGs was observed (Fig. 4B), suggesting that bhb-associated transcriptional effects become more evident at later stages. To visualize these changes, DEGs were represented in heatmaps for each timepoint (Fig. 4C), including both protein-coding genes and lncRNAs. While overall expression patterns remained similar between conditions, a structured subset of genes displayed coordinated differences between control and bhb-treated MBOs. These differences were more evident at day 16, supporting the emergence of a stage-dependent bhb-associated transcriptional profile at later stages of differentiation. Notably, these patterns were observed across both protein-coding and non-coding transcripts, suggesting that bhb-associated effects involve coordinated regulation across multiple layers of gene expression.

Fig. 4.

Bhb is associated with stage-dependent transcriptional changes during dopaminergic differentiation of midbrain organoids (MBOs). A, B Volcano plots showing differential gene expression between control and bhb-treated organoids at day 8 (A) and day 16 (B). The x-axis represents log₂ fold change (bhb vs. control), and the y-axis shows –log10 adjusted p-value (padj). Vertical dashed lines indicate |log₂FC| > 1, and the horizontal dashed line indicates padj < 0.05. Genes meeting both thresholds are highlighted in red; genes significant only by padj are shown in blue, and those exceeding fold-change threshold only are shown in green. Non-significant genes are shown in grey. C Heatmaps of bhb-dependent differentially expressed genes (DEGs) at day 8 (left) and day 16 (right). Rows correspond to DEGs and columns represent experimental conditions. Expression values are z-score scaled and hierarchically clustered. Gene biotype is indicated by the annotation bar (purple: protein-coding; green: lncRNA). D Boxplots showing expression levels (FPKM) of selected DEGs in MBO ctrl d8 and MBO bhb d8. E Boxplots showing expression levels (FPKM) of selected DEGs in MBO ctrl d16 and MBO bhb d16. Boxplots display median, interquartile range, and individual data points (n = 3 biological replicates per group, unless otherwise indicated). Statistical significance was assessed using the Wilcoxon rank-sum test (p-value < 0.05)

Among the treatment-associated DEGs, several genes are linked to neurodevelopmental, trophic signaling, and neuronal-related features. At day 8 (Fig. 4D), bhb-treated organoids showed significant increased expression of GDNF and GBX2, whereas control organoids showed higher expression of INHA. At day 16 (Fig. 4E), bhb-treated organoids showed significantly higher expression of GBX2, VSX2, LHX3, PCP4, LRRK1, FOXD3-AS1, SYT2, and CHD5, whereas control MBOs showed higher expression of INHA and LIF. Only INHA and GBX2 showed significant changes at both timepoints. Together, these results identify a stage-dependent set of transcriptional differences associated with bhb exposure at the population level.

Candidate H3K27ac Regions Present After bhb Treatment are Linked to Neurodevelopmental and Dopaminergic Programs

To explore chromatin-associated correlates of bhb-related transcriptional changes, we assessed the genomic distribution of the transcriptional activation-associated H3K27ac post-translational modification, using the Cleavage Under Targets and Release Using Nuclease (CUT&RUN) assay. We first assessed the global distribution of H3K27ac signal by generating metaplots and heatmaps centered on the midpoint of sample-specific H3K27ac peaks (± 2 kb), using spike-in normalized signal. All conditions showed a clear enrichment of H3K27ac at peak centers, consistent with its expected localization at active regulatory regions (Fig. 5A). In contrast, IgG controls displayed low and uniform signal without central enrichment, supporting the specificity of the H3K27ac profiles (Fig. S3A). Overall signal intensity was higher in less differentiated states (hESC and mfNPC) compared to MBOs at later stages, indicating progressive changes in chromatin activity during differentiation rather than a global increase associated with bhb treatment. Genomic annotation of H3K27ac peaks revealed that the most of enriched regions were located in intronic and intergenic regions across all conditions, with a smaller fraction mapping to promoter–TSS regions (Fig. S3B). While minor differences in peak distribution were observed between conditions, including a modest increase in promoter-associated regions in bhb-treated organoids at day 8, the overall genomic distribution remained comparable. Consistently, clustering based on Jaccard similarity grouped samples primarily by developmental stage, rather than treatment with bhb (Fig. S3C).

Fig. 5.

Bhb is associated with locus-specific changes in H3K27ac enrichment linked to neuronal gene programs. A Metaplot and heatmap showing H3K27ac signal distribution centered on the midpoint of sample-specific H3K27ac peaks (± 2 kb), obtained by CUT&RUN using spike-in normalized signal. Profiles are shown for hESC, mfNPC, and MBOs at day 8 and day 16, under control and bhb-treated conditions. B Overlap of candidate H3K27ac-enriched regions across experimental groups. Venn diagrams show shared and unique peaks between hESC and mfNPC, and among MBO conditions at day 8 and day 16 with or without bhb treatment. C, D Volcano plots showing differential candidate-H3K27ac enrichment between bhb-treated and control MBOs at day 8 (C) and day 16 (D), based on DESeq2 analysis of read counts across consensus peaks. Each point represents a genomic candidate-enriched region for H3K27ac. Regions are color-coded according to significance: red (|log₂FC| ≥ 2 and p-value < 0.05), blue (significant by p-value only), green (significant by log₂FC only), and gray (not significant); regions highlighted using nominal p-value and fold-change thresholds are presented as candidate regions for exploratory analyses. E Bar plot showing the log₂ fold change (bhb vs. ctrl) of genes associated with candidate H3K27ac-enriched regions, based on GeneHancer annotation, at day 8 and day 16. Statistical significance is indicated according to DESeq2 analysis. Gene labels are color-coded by functional category (neuronal, metabolic, or other), and highlighted genes correspond to those associated with day 16-specific acetylated regions. Data were derived from independent biological replicates (n = 2–3 per condition, depending on sample type), as detailed in Methods

To assess candidate locus-specific differences, we next identified differentially enriched H3K27ac regions showing nominal differential enrichment, between control and bhb-treated organoids (Fig. 5B). Using BH adjusted p < 0.05, one region reached FDR significance at day 8, whereas no regions reached FDR significance at day 16. Therefore, subsequent analyses based on nominal p-value and fold-change thresholds were considered exploratory. At day 8, nominally defined candidate regions showed enrichment (Fig. 5C), whereas candidate regions were detected at day 16 (Fig. 5D), indicating that bhb-associated changes are modest and restricted to a subset of loci. Functional annotation of regions with increased H3K27ac signal in bhb-treated organoids suggested exploratory associations with GO terms associated with neurodevelopmental processes, including neurogenesis, neuronal differentiation, and central nervous system development at day 8 (Fig. S4A). Further, dopaminergic-related processes such as dopamine response and dopaminergic neuron differentiation were also represented among these exploratory annotations (Fig. S4B). At day 16, candidate regions in bhb-treated MBOs were associated with processes related to neuronal maturation, including neurite outgrowth, synapse organization, and neuronal migration (Fig. S4C), as well as response to dopamine (Fig. S4D). We then specifically examined candidate acetylated regions after bhb treatment and analyzed the expression of their connected-genes, using information from the GeneHancer database (v4.14). This analysis revealed that loci showing increased acetylation in bhb-treated MBOs were associated with gene sets related predominantly to neuronal differentiation and, to a lesser extent, metabolic processes (Fig. 5E). Together, these findings indicate that bhb treatment is associated with exploratory locus-specific H3K27ac differences that are compatible with, but do not establish, the transcriptional patterns observed, rather than global remodeling of histone acetylation.

Integrated Motif and Expression Profiling Explores Links Between bhb-Associated H3K27ac Profiles and Dopaminergic Maturation-Related Programs

To further integrate the transcriptional and epigenetic changes associated with bhb treatment, we analyzed H3K27ac enrichment patterns across all samples to identify shared regulatory dynamics during differentiation. Unsupervised clustering of H3K27ac signal revealed three major groups of regulatory regions (Fig. 6A): Cluster 1 (13,203 regions), enriched at early stages (hESC and mfNPC); Cluster 2 (14,183 regions), active at early and intermediate stages, but reduced at day 16; and Cluster 3 (14,588 regions), which progressively gained H3K27ac enrichment at day 16, in both control and bhb-treated MBOs. Given that promoter-distal H3K27ac regions are commonly associated with active cis-regulatory elements (CREs), we assessed their overlap with annotated enhancers. A high proportion of regions across all clusters overlapped with GeneHancer elements, with Cluster 3 showing the strongest association with CREs linked to dopaminergic lineage genes (TH and KCNJ6) and TF (PITX3, FOXA2, EN1/2, and LMX1A/B) (Fig. 6B). This cluster was preferentially associated with late differentiation stages, consistent with the emergence of dopaminergic maturation-associated transcriptional features. Inspection of representative loci revealed increased H3K27ac signal across differentiation, particularly at regulatory regions dopaminergic linked to lineage-associated such as LMX1B (Fig. 6C–D), supporting progressive establishment of lineage-associated regulatory elements during differentiation.

Fig. 6.

Integrated motif and expression profiling explores candidate regulatory relationships during dopaminergic-associated differentiation. A Hierarchical clustering of candidate H3K27ac-enriched regions across all experimental conditions, identifying three major clusters (Cluster 1: 13,203 regions; Cluster 2: 14,183 regions; Cluster 3: 14,588 regions). B Overlap between H3K27ac-enriched regions and GeneHancer-annotated cis-regulatory elements (CREs) associated with selected dopaminergic lineage genes. Each dot represents a CRE–gene association, with dot size reflecting the GeneHancer interaction score. C Representative IGV tracks showing H3K27ac signal at Cluster 3 regions linked to LMX1B. D RNA-seq expression levels (FPKM) of LMX1B across experimental conditions. E Left: heatmap showing normalized expression (z-score) of transcription factors (TF) whose binding motifs are associated within each H3K27ac cluster. Right: relative frequency of motif enrichment across clusters (C1–C3). F–I Representative IGV tracks showing H3K27ac signal at GeneHancer-annotated regulatory regions associated with key dopaminergic genes: TH (F), FOXA2 (G), KCNJ6 (H), and CORIN (I). Signal is shown across hESCs, mfNPCs, and MBOs at day 8 and day 16 under control or bhb treatment; scale for each track is indicated in brackets. GeneHancer elements and linked target genes are shown below each locus. J Integrative TF–enhancer–target gene network based on GeneHancer-linked regulatory regions. Node color represents Log₂FC (bhb vs. ctrl) in MBOs at day 16. Enhancers are additionally classified by H3K27ac cluster (C1–C3). Edges represent predicted regulatory interactions (TF→enhancer and enhancer→target gene), with edge thickness proportional to GeneHancer interaction score. This network illustrates potential regulatory relationships associated with dopaminergic differentiation in d16 MBOs bhb-treated

To identify potential transcriptional regulators, we performed motif enrichment analysis across H3K27ac clusters. Cluster-specific enrichment of TF binding motifs was observed, and integration with RNA-seq data revealed associations between motif availability and TF expression (Fig. 6E). Early-stage clusters were enriched for pluripotency-associated factors such as POU5F1, whereas late-stage Cluster 3 was enriched for motifs associated with neurodevelopmental and dopaminergic TF, including PITX3, FOXP1, and SOX21, consistent with their roles in neuronal differentiation and maturation. Several TF exhibited increased expression at day 16 alongside enrichment of their corresponding motifs in Cluster 3 regions, suggesting candidate links between regulatory chromatin profiles/states and transcription factor expression. Notably, some factors showed modest increases in bhb-treated organoids, consistent with the subtle transcriptional differences observed at this stage. To further examine the relationship between H3K27ac enrichment and gene expression, we analyzed selected GeneHancer-linked regulatory regions associated with key dopaminergic genes (TH, FOXA2, KCNJ6, and CORIN). The loci for TH, KCNJ6 and CORIN displayed increased acetylation at day 8 after bhb addition; at day 16, both control and bhb-treated MBOs had similar H3K27ac signal (Fig. 6F–I), suggesting a transient or stage-dependent association between bhb exposure and selected regulatory regions.

Finally, we integrated TF motifs, enhancer regions, and target gene expression into a regulatory interaction network (Fig. 6J). To facilitate systematic exploration of these relationships, we developed the Enhancer Network Explorer, an interactive platform that integrates multi-omic data generated in this study across multiple experimental contrasts. While the present analysis focuses on day 16 MBOs, where differences are more evident, the platform enables broader interrogation of TF–enhancer–target gene interactions across developmental stages and conditions. The resulting network highlights potential TF–enhancer–gene relationships associated with dopaminergic maturation-related programs and reveals candidate relationships between regulatory chromatin state and transcriptional profiles. While these interactions are predictive and reflect population-level associations, bhb-associated differences in regulatory chromatin state are compatible with transcriptional changes observed at the population level, providing a framework to explore candidate regulatory relationships associated with neuronal differentiation in the context of bhb exposure.

Discussion

Midbrain dopaminergic differentiation involves morphological and metabolic changes accompanied by transcriptional remodeling and dynamic chromatin-associated regulation. Our findings support a model in which bhb modulates regulatory programs linked to neuronal maturation within a conserved dopaminergic differentiation trajectory. Utilizing an established human midbrain organoid model, our integrated transcriptomic and CUT&RUN analyses revealed that bhb exposure, beyond its canonical metabolic role, is linked to population-level transcriptional differences and changes in candidate region-specific H3K27ac profiles at neuronal and dopaminergic gene loci.

Importantly, these observations should be interpreted within a physiologically framework. Although circulating bhb levels in healthy pregnant women are generally low (typically below 0.1 mM in maternal serum) (Noshiro et al. 2022; Parenti et al. 2024), recent evidence indicates a maternal–fetal gradient, with higher concentrations detected in amniotic fluid (~ 0.5 mM), reflecting both placental transfer and local ketone body production (Shibata et al. 2025). Because the developing brain expresses monocarboxylate transporters and efficiently utilizes ketone bodies as oxidative substrates (Adam et al. 1975), local bhb concentrations within fetal neural tissue may be shaped by tissue-specific transport and differ from maternal serum. Thus, the 1 mM bhb used here, represents a high-end exposure intended to model an upper-range physiological scenario, while remaining well below the levels associated with pathological ketoacidosis (Qian et al. 2020). This provides a clear rationale for examining bhb effects in human MBOs and supports the potential relevance of the modest transcriptional and chromatin-associated differences observed, while recognizing that direct in vivo fetal brain concentrations remain difficult to define.

Transcriptomic Responses to bhb During Midbrain Organoid Differentiation

Our RNA-seq profiling revealed that bhb did not alter the global transcriptional progression of midbrain organoid differentiation, as both control and treated MBO groups followed the expected developmental transitions from hESCs to mfNPCs and, subsequently, into MBO stages enriched in neuronal and dopaminergic features (Fig. 2). However, unsupervised analyses identified a subset of differentially expressed protein-coding genes associated with neurodevelopmental, trophic, and dopaminergic-related processes (Fig. 4), indicating bhb-related effects within this differentiation framework.

To resolve these localized effects, we performed a directed examination of cell-type and dopaminergic-related programs using ssGSEA (Fig. 3). At day 8, bhb-treated MBOs showed significantly higher enrichment scores for neuronal progenitors, oligodendrocyte precursor cells, and radial glia 3 programs, whereas control MBOs showed higher A10 dopaminergic neuron-associated scores. At day 16, bhb-treated MBOs showed significantly higher enrichment of the global dopaminergic neuron-associated program and radial glia 3 signatures, while control MBOs showed higher radial glia 4 scores (Fig. 3B). Interestingly, bhb-treated MBOs showed higher normalized expression of TH, FOXA2 and KCNJ6 (GIRK2), a potassium channel expressed by A9 dopaminergic neurons, characterized by projections from the substantia nigra pars compacta to the striatum (Reyes et al. 2012; Mishra et al. 2025). Conversely, the A10-specific marker CALB2 (Chung et al. 2005; Arenas et al. 2015) was reduced in day 16 treated MBOs (Fig. 3C). More broadly, directed inspection of a wider developmental panel verified the expected, progressive expression of core midbrain and dopaminergic markers (OTX2, EN1/2, LMX1A/B, FOXA1, DRD1/2, and PITX3) across both groups at days 8 and 16 (Fig. S2B).

Rather than altering the overall dopaminergic developmental trajectory, or directly driving subtype specification, our findings suggest that bhb subtly fine-tunes cellular state transitions and developmental timing. This is supported by the enrichment of progenitor and radial glia programs alongside an increased day 16 dopaminergic signature, which suggests bhb modulates how responsively neural populations react to maturation cues. Furthermore, because A9 signatures remained comparable between conditions, and A10 signatures were higher in controls at day 8, our results points toward a role for bhb in shifting the balance of emerging cell states within the differentiating organoid, rather than rewriting their identities.

The treatment-associated DEGs suggests that bhb exposure is linked to stage-dependent changes in trophic and developmental signaling-related transcripts. At day 8, before exposure to the maturation medium, bhb-treated MBOs showed higher expression of GDNF and GBX2, whereas INHA was higher in control MBOs. This pattern is notable because GDNF is a trophic factor widely associated with midbrain dopaminergic neuron survival and maturation (Clarkson et al. 1997); GBX2 on the other hand, is involved in midbrain–hindbrain developmental patterning (Joyner et al. 2000; Li and Joyner 2001), and INHA belongs to the Activin/Inhibin branch of TGF-β superfamily signaling (Rodriguez-Martinez and Velasco 2012). By day 16, following exposure to maturation conditions containing trophic molecules (GDNF, Activin A, BDNF, TGFβ3, and dbcAMP), bhb-treated MBOs maintained higher GBX2 expression, whereas control MBOs exhibited higher levels of INHA and LIF. INHA encodes the Inhibin α subunit, which antagonizes Activin signaling (Rodriguez-Martinez and Velasco 2012). Because Activin A has been implicated in the survival of midbrain dopaminergic neurons (Stayte et al. 2015), the consistent reduction of INHA expression in bhb-treated MBOs at both days 8 and 16 could point to sustained, permissive Activin signaling in these cells. However, given that the specific role of INHA in human midbrain dopaminergic neurodevelopment is not yet defined, these results warrant cautious interpretation rather than direct extrapolation as a beneficial effect on dopaminergic maturation.

Regarding LIF, it is established that it activates STAT3 to induce astrocyte differentiation of cortical neural stem/progenitor cells (Chang et al. 2004; Rodriguez-Rivera et al. 2009); in our organoid system, decreased LIF expression after bhb treatment at day 16 would favor neuronal versus astrocyte differentiation. Although bhb has been reported to promote BDNF expression in hippocampal neurons under normoglycemic conditions (Hu et al. 2018), we did not observe evidence of differential BDNF expression in our midbrain organoid model. In agreement, CDNF, another neurotrophic factor did not change when comparing bhb and control MBOs. Thus, bhb exposure was not associated to positive regulation of all dopaminergic genes, but rather with a redistribution of transcriptional features related to regional patterning, trophic signaling, and neuronal-related expression. Future single-cell and functional studies will be required to determine whether bhb directly modulates expression of important components of these pathways or alters the relative representation of specific neural populations within MBOs.

Bhb-Associated H3K27ac Profiles in Midbrain Organoids

H3K27ac is a histone mark strongly associated with enhancer and promoter activation and transcriptional competence (Gao et al. 2020; Park et al. 2022). Our CUT&RUN analysis revealed that bhb treatment is associated with candidate region-specific differences in H3K27ac profiles in midbrain organoids, with increased acetylation at genomic regions linked to neuronal maturation, neurite outgrowth, and dopaminergic neurotransmission (Fig. 5). However, these data should be interpreted as exploratory; thus, the resulting GO analyses are best viewed as hypothesis-generating rather than definitive evidence of bhb-driven chromatin remodeling.

Interestingly, these candidate H3K27ac differences paralleled population-level transcriptional changes in GeneHancer-linked target genes, suggesting a potential relationship between regulatory chromatin state and gene expression during differentiation. Rather than driving broad activation or global remodeling of H3K27ac, bhb appears to fine-tune a selective subset of regulatory elements, consistent with the moderate but reproducible transcriptional changes observed. Nevertheless, whether these localized H3K27ac differences act as causal drivers of transcription or merely serve as parallel correlates of developmental state remains to be established.

This distinction is critical when comparing differentiation contexts. In monolayer cultures, bhb exposure was associated with reduced global H3K27ac signal; conversely, organoid CUT&RUN analysis identified region-specific variations rather than global alterations. This divergence suggests that bhb’s chromatin-associated effects are not uniform and might depend heavily on developmental stage, culture architecture, cellular composition, and genomic region. Future studies combining global H3K27ac quantification in MBOs, higher-powered chromatin profiling, and functional manipulation of candidate regulatory pathways will be required to determine if bhb directly modulates specific loci or if these profiles simply reflect broader shifts in the organoid’s developmental trajectory.

Metabolic–Epigenetic Coupling and Regulatory Network Dynamics During Dopaminergic Differentiation

The effects of bhb on H3K27ac enrichment support the concept of metabolic–epigenetic coupling, where cellular metabolites influence chromatin state and transcriptional regulation through their role as cofactors for chromatin-modifying enzymes (Peng et al. 2016; Berger et al. 2016; Sivanand et al. 2018). In this context, the stage- and locus-associated H3K27ac profiles observed in this study suggest that bhb exposure could be linked to selective differences at regulatory elements, rather than inducing widespread chromatin changes. Although bhb has been described as an endogenous inhibitor of class I histone deacetylases (HDACs), low millimolar concentrations like those used in this study may not substantially affect HDAC1 deacetylase activity, since the reported IC50 for bhb is 5.3 mM (Shimazu et al. 2013). Specifically, in hippocampal neurons, bhb increased global histone acetylation, and promotes BDNF gene expression (Hu et al. 2018). Although the present data do not establish the molecular mechanism underlying the observed chromatin and transcriptional profiles, they are consistent with the possibility that bhb influences developmental gene regulation through pathways other than broad HDAC inhibition, a possibility that merits further investigation. Together, these findings support a model in which bhb exposure is associated with modest, region-specific chromatin-associated differences and stage-dependent transcriptional changes, rather than a uniform increase in H3K27ac or a demonstrated H3K27ac-mediated mechanism of dopaminergic differentiation. Alternative mechanisms, including metabolism-linked changes in acetyl-CoA availability, local chromatin-associated metabolic regulation, or indirect effects on developmental signaling pathways, remain plausible but unresolved. Recent evidence indicates that the acetyl-CoA–producing enzyme ACSS2 can sense bhb concentrations, and translocate to the nucleus, where it locally supplies acetyl-CoA for histone acetylation (Li et al. 2017; Wang et al. 2025). Such mechanisms provide a plausible link between metabolic state and localized chromatin profiles during neuronal differentiation.

Consistent with this framework, previous studies have shown that bhb-mediated histone acetylation facilitates recruitment of transcriptional regulators such as CREB and CBP/p300 to neuronal gene promoters (Hu et al. 2018). Building on this concept, our motif enrichment and integrative analyses identified candidate associations among bhb-linked H3K27ac profiles, predicted TF motifs, and gene expression patterns. These associations should be interpreted as predictive and hypothesis-generating, rather than as evidence of direct TF recruitment or enhancer activation by bhb.

The integration of TF motif enrichment with enhancer annotation and RNA-seq data revealed potential TF-enhancer-target gene relationships associated with dopaminergic-related differentiation (Fig. 6). While canonical dopaminergic regulators (e.g., LMX1A, PITX3) were represented within these networks, additional TF, including KLF family members, WT1, ZIC3, PAX4, and NHLH1, emerged as potential contributors to the regulatory landscape. Rather than indicating a shift in regional identity, the involvement of these factors may reflect an expanded regulatory repertoire associated with neuronal differentiation and maturation, or the heterogeneous cellular states present within developing MBOs.

In parallel, we identified the upregulation of long non-coding RNA transcripts such as LNCNEF, located near FOXA2, supporting the possibility that non-coding regulatory elements may contribute to enhancer-mediated control of dopaminergic gene expression. Together, these observations highlight a complex and interconnected regulatory architecture in which canonical and non-canonical factors may converge at the level of enhancer activity.

Taken together, these findings support a model in which bhb acts as a candidate metabolic–epigenetic modulator associated with modest, population-level differences in regulatory chromatin profiles and transcriptional programs during human midbrain organoid differentiation. Thus, bhb exposure is linked to selective candidate differences in H3K27ac at regulatory elements, and also to predicted TF–enhancer–target gene interactions. While these relationships are predictive and do not establish direct causality, they provide an integrative systems-level view of how metabolic cues could be associated with neuronal differentiation programs at the population level.

Concluding Remarks

In summary, our study provides an integrated view of how bhb exposure is associated with population-level transcriptional differences, and candidate region-specific H27K27ac profiles during human midbrain organoid differentiation. Across multiple analytical layers, incorporating bulk RNA-seq, ssGSEA-based pathway enrichment, CUT&RUN H3K27ac profiling, and integrative TF–enhancer–target gene modeling, we consistently observed moderate, stage-dependent differences following bhb treatment. These molecular shifts converge on neurodevelopmental patterning, trophic signaling, and dopaminergic maturation programs, suggesting that bhb fine-tunes the transcriptional response of differentiating organoids to stage-specific cues. Together, these findings contribute to a growing body of evidence linking cellular metabolism and chromatin-associated gene regulation in neural systems, providing a foundation for future functional dissection of metabolite-driven epigenetic changes. In addition, the development of the Enhancer Network Explorer platform offers an accessible framework for interrogating candidate TF–enhancer–gene interactions across multiple experimental conditions, extending the utility of this dataset beyond the scope of the present study.

Limitations and Perspectives

Although the present study identifies coordinated transcriptomic and epigenomic differences associated with bhb exposure, it does not establish direct causal links between bhb exposure, enhancer activation, and downstream transcriptional regulation. In addition, the use of bulk RNA-seq and CUT&RUN in heterogeneous organoid systems limits the resolution at which cell-type composition and cell-state transitions can be distinguished. Similarly, ssGSEA enrichment scores reflect the relative representation of predefined transcriptional programs and should not be interpreted as direct measurements of cell-type abundance. Therefore, the observed differences should be interpreted as population-level changes, which may reflect a combination of altered maturation dynamics, shifts in cell-type proportions, or differences in cellular states.

Although this study focuses on histone acetylation dynamics, alternative or parallel epigenetic mechanisms may also contribute to the observed effects. In particular, the potential role of histone β-hydroxybutyrylation (e.g., H3K9bhb) as an additional regulatory layer during dopaminergic differentiation remains to be explored. Additionally, we cannot exclude that changes in progenitor proliferation or cell survival may contribute to the increased abundance of dopaminergic markers observed in bhb-treated organoids, as these parameters were not directly assessed. Furthermore, the inferred TF–enhancer–target gene relationships remain predictive and require functional validation to establish their regulatory impact. Future experiments using different concentrations of bhb, broader 3D immunophenotyping, proliferation and apoptosis markers, single-cell transcriptomics, and targeted modifications of candidate pathways will be required to distinguish direct bhb-associated transcriptional effects from changes in organoid cellular composition.

Despite these limitations, the findings presented here highlight a previously underappreciated link between metabolic state and transcriptional/chromatin-associated regulation during human midbrain organoid differentiation. Future studies integrating single-cell approaches, functional assays, and targeted validation of regulatory elements will be essential to dissect the mechanistic basis of these observations. In addition, combining metabolic modulation strategies with cell selection approaches, such as enrichment of CORIN-positive populations, may further improve the generation of mature and functionally competent dopaminergic neurons for downstream applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

J.S-C designed and conceptualized the project, performed the main experiments and formal analysis, designed figures, and drafted the original manuscript. C.M-R co-executed the CUT&RUN assay and contributed to the writing of the manuscript. R.CD-D provided guidance in formal analysis, discussed data, and contributed to manuscript review and editing. A.L-S acquired lightsheet confocal microscopy at the ADMiRA unit. M.R-A contributed to experiments included in the supplementary materials. I. E-A contributed to stereological quantification. M.C-C performed Sholl analysis in 2D neuronal cultures. I.V. supervised the project, got funding, and wrote, reviewed and edited the manuscript. The manuscript was approved by all authors.

Funding

SECIHTI CF23-I-1668 and UNAM-PAPIIT IN229025. J.S-C (CVU: 774097; 2020-CLAVE: 000026-02NACF) received a graduate fellowship from SECIHTI. Samples were sequenced at the National Genomic Sequencing Laboratory Tec-BASE through the “Core Lab Genomics Tec-BASE Seed Fund for Research Projects” grant.

Data Availability

RNAseq data and CUT and RUN regions with H3K27ac were deposited at the NCBI Gene Expression Omnibus (GEO) with reference numbers GSE313206 and GSE313207, respectively.

Declarations

Conflict of interest

The authors declare no competing interests.