Beyond phenotypic markers: rethinking dopaminergic identity in iPSC-derived neurons

Abstract

Human pluripotent stem cell-derived dopaminergic neurons have become central platforms for Parkinson’s disease (PD) modeling and regenerative medicine. Current differentiation systems can efficiently generate TH⁺/FOXA2⁺/LMX1A⁺ ventral midbrain-like populations. Although canonical markers remain important components of dopaminergic validation, the field increasingly complements them with single-cell profiling, developmental reference mapping, functional assessment, and transplantation-based evaluation. Nevertheless, marker-positive phenotypic resemblance alone may inadequately capture the full multidimensional identity of substantia nigra pars compacta neurons. Recent single-cell, developmental, and organoid studies indicate that human dopaminergic neurons exist across highly specialized multidimensional states shaped by developmental trajectory, epigenetic regulation, metabolic adaptation, biological aging, and ecosystem-level interactions. On the basis of these observations, we hypothesize that some current in vitro systems may generate neurons with incomplete or context-dependent identity because of accelerated differentiation, strong exogenous patterning, resetting of age-associated features, incomplete metabolic maturation, and the absence of supporting cells and tissue-derived signals. We distinguish these experimentally observed limitations from their conceptual interpretation and propose a testable framework for evaluating dopaminergic identity across developmental, molecular, metabolic, functional, aging-related, and environmental dimensions. This framework identifies improved maturation, representation of biological age, metabolic resilience, and multicellular context as priorities for PD modeling and regenerative neuroscience.

Introduction

Human pluripotent stem cell (hPSC)-derived dopaminergic neurons have rapidly become central tools for Parkinson’s disease (PD) modeling and regenerative medicine. Over the past decade, increasingly optimized differentiation protocols have enabled efficient generation of TH⁺/FOXA2⁺/LMX1A⁺ ventral midbrain-like populations, supporting disease modeling, organoid development, and clinical-grade transplantation strategies [1–4]. Despite these advances, unresolved limitations include heterogeneous lineage specification, incomplete maturation, variability in culture and graft composition, and uncertainty regarding how in vitro characteristics predict long-term graft function [1–5].

Canonical markers such as TH, FOXA2, LMX1A, EN1, PITX3, DAT, and GIRK2 remain widely used and biologically informative components of dopaminergic validation [1–4, 6, 7]. However, the field increasingly complements marker-based assessment with single-cell transcriptomics, developmental reference mapping, trajectory analysis, electrophysiological characterization, lineage tracing, and transplantation-based functional evaluation [3, 4, 6–13]. These advances demonstrate active efforts to resolve regional identity, cellular heterogeneity, maturation, subtype specification, and therapeutic competence. Because each approach captures selected dimensions of cellular identity, establishing equivalence to an endogenous substantia nigra pars compacta (SNpc) state requires convergent evidence rather than any individual marker panel or profiling modality used in isolation. Human midbrain dopaminergic neurons comprise molecularly and functionally diverse populations that emerge through developmental progression [5, 9, 10, 14]. Aging further shapes their vulnerability [14, 15], while multicellular and circuit context can influence neuronal maturation and functional organization in experimental human neural systems [16–18]. Consequently, phenotypic resemblance may not necessarily reflect genuine biological fidelity.

Experimental studies have directly documented mixed regional identities, variable differentiation trajectories, incomplete maturation, and culture-dependent cellular states in hPSC-derived dopaminergic systems [2, 5, 9]. Whether these observations reflect developmental immaturity, expected biological heterogeneity, or stable states without direct in vivo counterparts remains unresolved. These limitations may be particularly important in PD, where selective neuronal vulnerability emerges from the long-term interaction among aging, mitochondrial stress, inflammatory signaling, circuit-level activity, and multicellular ecosystem dynamics.

In this review, we propose that neuronal identity should be viewed not as a static marker profile, but as a dynamic multidimensional biological state. We use “phenotypic illusion,” “developmental compression,” and “spurious cell states” as conceptual and testable hypotheses rather than established biological classifications. These concepts are used to examine whether marker-positive cells can diverge from corresponding in vivo states across developmental, epigenetic, metabolic, functional, and ecosystem-level dimensions.

For clarity, we use “authenticity” to describe the overall degree to which an engineered neuron corresponds to its intended in vivo counterpart, recognizing that authenticity is multidimensional rather than binary. “Biological fidelity” refers to the measurable concordance between an engineered cell and an appropriately stage- and region-matched reference across molecular, epigenetic, metabolic, electrophysiological, functional, and ecosystem-level dimensions. “Developmental compression” denotes the acceleration or abbreviation of the temporal sequence of human neurodevelopment during in vitro differentiation; it does not, by itself, imply an abnormal outcome. Finally, “cell-state engineering” refers to the deliberate generation and stabilization of a desired multidimensional cellular state through control of developmental trajectory, biological age, metabolism, stress adaptation, and environmental context, rather than marker expression alone.

The phenotypic illusion: when marker positivity masks biological incompleteness

The development of human pluripotent stem cell (hPSC)-derived dopaminergic neurons has advanced substantially over the past decade, with increasingly refined protocols capable of generating TH⁺/FOXA2⁺/LMX1A⁺ ventral midbrain-like populations for disease modeling and regenerative applications in Parkinson’s disease (PD) [1–4, 7, 12]. These advances have enabled the emergence of clinical-grade donor products, including cryopreserved midbrain dopamine progenitors designed for transplantation into PD patients [1]. Nevertheless, despite impressive improvements in differentiation efficiency and graft survival, persistent limitations include variable differentiation and graft composition, incomplete functional maturation, and uncertainty regarding how donor-cell composition predicts in vivo performance [2–4, 7, 12]. These observations suggest that phenotypic resemblance alone may not equate to authentic biological identity.

Marker panels remain an essential first-line component of quality control and are increasingly integrated with single-cell profiling, developmental benchmarking, electrophysiological assessment, and transplantation outcomes. Nevertheless, many differentiation studies continue to use immunostaining or transcript-level detection of canonical ventral midbrain markers including TH, FOXA2, LMX1A, EN1, OTX2, PITX3 and NR4A2 as principal evidence of specification [1–4, 7, 12]. Accordingly, the principal limitation is not the use of these markers per se, but their interpretation as sufficient evidence of complete regional, developmental, and functional identity. However, developmental biology indicates that many of these markers are not uniquely restricted to SNpc dopaminergic neurons. For example, LMX1A⁺/FOXA2⁺ progenitors can include ventral mesencephalic and adjacent diencephalic floor-plate populations rather than exclusively SNpc-fated lineages [4, 9, 12]. Similarly, TH expression alone is insufficient to define mesencephalic dopaminergic identity because multiple floor plate-derived neuronal populations can express TH while lacking genuine nigrostriatal specification [4, 9]. EN1 provides additional information about caudal midbrain identity and graft-relevant specification, but its expression remains dependent on developmental stage and regional context [4, 7, 12]. Thus, conventional marker combinations may provide only partial lineage approximation rather than definitive evidence of authentic SNpc identity.

Single-cell transcriptomics has revealed substantial cellular heterogeneity within hPSC-derived dopaminergic cultures and transplanted grafts [4, 6, 12, 13]. Directed differentiation can generate populations resembling multiple adjacent developmental territories, including ventral midbrain, isthmic, and ventral hindbrain identities [9, 12]. Midbrain organoids likewise generate transcriptionally diverse dopaminergic and non-dopaminergic populations and exhibit variability in cellular composition, maturation, and developmental trajectories [4, 19]. This molecular diversity reflects both biological complexity in dopaminergic lineage specification and variability introduced by the differentiation and culture environment. Midbrain dopaminergic neurons are highly heterogeneous populations composed of transcriptionally and functionally distinct subtypes distributed across SNpc and ventral tegmental area (VTA) regions [5, 10, 11]. Recent single-nucleus transcriptomic profiling of the adult human substantia nigra pars compacta identified at least ten molecularly distinct dopaminergic neuron populations, including ventrally localized AGTR1⁺ subtypes that exhibit selective vulnerability and enrichment for PD-associated genetic risk [5]. Consequently, achieving generalized TH positivity may obscure profound subtype-level incompleteness relevant to disease modeling and transplantation biology.

Recent single-cell studies have extended quality-control analysis beyond mDA marker expression to identify diencephalic and adjacent basal-plate lineages. In Xu et al., PITX2 was expressed by two mesencephalic basal-plate progenitor clusters and by two off-target glutamatergic populations detected after transplantation, one of which was characterized by BARHL1 expression; the BARHL1-positive population was found in unsorted but not CLSTN2- or PTPRO-sorted grafts [12]. In Kim et al., PITX2 marked a rostral dopaminergic population and was also expressed in a small group of cells displaying hybrid mDA and subthalamic transcriptional features [7]. Accordingly, PITX2 and BARHL1 should be interpreted within broader regional and lineage-specific transcriptional signatures rather than as stand-alone indicators of cellular identity.

Functional maturation may remain incomplete despite successful acquisition of dopaminergic markers. Although hPSC-derived neurons can synthesize dopamine and display spontaneous electrical activity, their electrophysiological and molecular properties may continue to mature during prolonged culture or following transplantation [4, 7, 13]. Adult SNpc dopaminergic neurons are characterized by autonomous pacemaking, extensive axonal arborization, calcium-dependent activity, and high bioenergetic demands [20, 21]. These properties are only partially reconstructed by current in vitro systems. Single-cell studies additionally reveal developmental and cellular heterogeneity across differentiated cultures and transplanted grafts [4, 6, 12]. Midbrain organoids likewise remain limited by incomplete maturation, necrotic regions, and organoid-to-organoid variability despite continuing advances in bioengineering [4, 19].

Importantly, transplantation success does not by itself establish equivalence to endogenous SNpc neurons. Clinical-grade donor products such as MSK-DA01 can survive transplantation, improve motor behavior, and generate TH-positive neurons in vivo [1]. However, behavioral efficacy alone does not define the developmental, transcriptional, subtype, or functional state of the graft-derived neurons [12].

Recent transplantation studies provide direct evidence that hPSC-derived dopaminergic progenitors can undergo substantial maturation, subtype specification, and functional integration in vivo. Using TARGET-seq one year after homotopic intranigral transplantation, Fiorenzano et al. identified nine molecularly distinct human dopaminergic neuron subtypes and linked their transcriptional identities to projections targeting the prefrontal cortex or dorsolateral striatum [13]. Storm et al. combined single-cell multiomics with molecular barcode-based lineage tracing to demonstrate that the three major graft-derived populations dopaminergic neurons, astrocytes, and vascular leptomeningeal-like cells originated from a common tripotent progenitor present at the time of transplantation, showing that graft heterogeneity can emerge through post-transplantation lineage diversification [6]. More recently, Kim et al. showed that FGF18 treatment combined with WNT inhibition increased the yield and fidelity of A9-like dopaminergic neurons, enhanced EN1, PITX3, ALDH1A1, and DAT expression, and reduced diencephalic off-target populations in vitro and after transplantation [7]. Boost+ neurons exhibited improved dopaminergic and electrophysiological properties in vitro, while graft-derived neurons displayed spontaneous activity and received functional synaptic inputs from the host brain; both Boost and Boost+ grafts restored amphetamine-induced rotation and adhesive-removal performance, whereas only Boost+ grafts significantly improved ladder-rung walking [7]. Collectively, these studies demonstrate that in vivo maturation and therapeutic competence should be evaluated through graft composition, subtype identity, target-specific innervation, synaptic integration, and behavioral recovery rather than dopaminergic marker expression alone.

Importantly, graft heterogeneity should not automatically be equated with undesirable off-target contamination. The shared lineage identified by Storm et al. indicates that some non-neuronal graft components may arise through post-transplantation lineage diversification rather than differentiation failure [6]. However, whether these non-neuronal populations support or impair dopaminergic maturation and functional integration remains unresolved and is likely to depend on their identity, abundance, state, and spatial organization. Quality-control strategies should therefore distinguish developmentally related graft heterogeneity from incorrectly specified, proliferative, or otherwise unsafe populations rather than assuming that maximal neuronal purity is always preferable.

Collectively, these observations indicate that marker positivity provides an incomplete measure of dopaminergic cell state, but they do not establish that marker-positive cells are biologically artificial. We therefore propose, as a hypothesis, that a subset of in vitro-derived populations may occupy “spurious cell states”: transcriptionally coherent states that acquire selected dopaminergic features but cannot yet be matched to a corresponding state along available human developmental or adult reference trajectories. Such apparent divergence may alternatively reflect developmental immaturity, physiological heterogeneity, incomplete reference atlases, or culture-induced adaptation. Distinguishing among these possibilities will require longitudinal trajectory analysis, multimodal benchmarking, and functional validation rather than endpoint marker assessment alone.

Building on ongoing advances in single-cell, developmental, functional, and transplantation-based validation, we propose a multidimensional cell-state perspective in which SNpc identity emerges from coordinated developmental, molecular, epigenetic, metabolic, stress-adaptive, functional, and ecosystem-level integration (Fig. 1).

Fig. 1.

Phenotypic convergence may not ensure biological fidelity. Current dopaminergic differentiation platforms use ventral midbrain patterning and canonical markers to assess specification. However, marker-positive iPSC-derived dopaminergic neurons may remain incomplete across developmental, epigenetic, metabolic, functional, and ecosystem-level dimensions. The figure presents a conceptual framework in which phenotypic convergence may not establish equivalence to a stage- and region-matched human SNpc reference state. The proposed incomplete state is hypothetical and requires longitudinal, multidimensional validation

Developmental trajectory divergence and the emergence of spurious states

The generation of dopaminergic neurons from hPSCs represents an attempt to reconstruct a temporally and spatially coordinated developmental program. During embryogenesis, ventral midbrain specification unfolds through prolonged and tightly coordinated morphogenic signaling involving SHH, WNT, FGF8, OTX2, GBX2, and multiple downstream transcriptional regulatory networks that progressively establish anterior-posterior, dorsal-ventral, and mediolateral identity [2, 9, 22]. Importantly, these developmental processes occur across extended temporal windows in humans and are accompanied by continuous transitions in progenitor competence, proliferation dynamics, and lineage diversification [9]. Single-cell studies have identified multiple radial-glial and intermediate progenitor populations during ventral midbrain development, while developmental and adult reference datasets demonstrate extensive dopaminergic subtype diversity [5, 9, 10]. Thus, SNpc identity is unlikely to arise from a single binary specification event and is better understood as the product of a progressive developmental trajectory and subsequent state stabilization.

This temporal architecture presents a challenge for in vitro differentiation systems, which reconstruct selected stages of human midbrain development over several weeks through staged exposure to morphogens and small-molecule patterning signals [2, 4, 12, 18]. Contemporary protocols commonly combine dual-SMAD inhibition, SHH- or SAG-mediated ventralization, WNT activation, and timed FGF8 exposure to generate FOXA2⁺/LMX1A⁺ ventral midbrain floor-plate progenitors [4, 12, 18, 23, 24]. Although these strategies efficiently promote dopaminergic specification, they simplify the spatial gradients, evolving tissue architecture, and extended temporal progression of endogenous human midbrain development.

The contrast between prolonged human embryonic midbrain development and temporally compressed in vitro raises the possibility that accelerated protocols reconstruct only selected aspects of endogenous developmental progression (Fig. 2).

Fig. 2.

Temporal compression during in vitro dopaminergic differentiation. During human embryonic midbrain development, dopaminergic identity emerges through coordinated morphogen signaling, lineage diversification, multicellular interactions, and progressive maturation. In vitro differentiation uses staged SHH, WNT, and FGF8 signaling to generate dopaminergic progenitors over a compressed timescale. Although these approaches efficiently produce marker-positive neurons, temporal compression and simplified culture environments may alter lineage progression, subtype resolution, maturation, and functional stability. The figure presents the hypothesis that some resulting populations may occupy biologically incomplete states characterized by limited maturation and batch variability despite apparent marker positivity

Importantly, emerging evidence suggests that accelerated differentiation may fundamentally distort lineage trajectory fidelity rather than merely shortening developmental duration. Single-cell analysis by Xu et al. demonstrated that mDA-directed differentiation concurrently generated populations resembling ventral midbrain, isthmic, and ventral hindbrain territories despite successful dopaminergic induction [12]. Ventral midbrain organoids likewise generate diverse progenitor, neuronal, and non-neuronal populations and exhibit inter-organoid variability in cellular composition and maturation [4]. In addition, ventral midbrain–striatal–cortical assembloids permit experimental reconstruction of dopaminergic projections and interregional activity, while remaining engineered approximations of human circuit organization [18]. Together, these studies establish regional heterogeneity, variable maturation, and context dependence in current models; they do not, however, demonstrate that accelerated differentiation intrinsically distorts developmental trajectories. Whether temporal compression produces stable divergence from an endogenous human developmental state therefore remains a testable hypothesis.

Accordingly, we hypothesize that temporal compression and simplified culture environments may alter the duration, sequence, or stabilization of intermediate states, potentially allowing marker-positive cells to arise through trajectories that only partially align with human development. This hypothesis is motivated by observations of regional heterogeneity, variable developmental trajectories, and culture-dependent maturation in hPSC-derived dopaminergic systems [2, 4, 9, 10, 18]. Testing it will require longitudinal comparison with stage-matched human reference trajectories rather than endpoint marker assessment alone.

Here, we operationally define a putative “spurious cell state” as a persistent, internally coherent in vitro state that expresses selected dopaminergic features but cannot be aligned with an appropriately stage-matched human developmental or adult reference state across multiple biological dimensions. This designation should require convergent evidence of: (i) persistent divergence from reference developmental trajectories after extended maturation; (ii) discordance among regional identity, transcriptomic or epigenomic state, and dopaminergic function; and (iii) failure to converge toward a recognized in vivo state after restoration of relevant environmental cues or transplantation. These criteria constitute a proposed experimental framework rather than an established classification system.

This proposed state differs from developmental immaturity, in which cells align with an earlier but recognizable stage in an appropriate developmental reference and retain the capacity for further maturation. It also differs from physiological cellular heterogeneity, in which distinct populations correspond to reproducible developmental stages, regional identities, functional states, or dopaminergic subtypes observed in vivo [5, 9, 10]. Mixed marker expression or transcriptional divergence at a single time point is therefore insufficient to classify a population as spurious because it may reflect transient immaturity, physiological heterogeneity, technical variation, or limitations of the available reference atlases. Accordingly, “spurious cell state” remains a provisional designation that requires longitudinal and multimodal validation.

Several experimental studies provide observations relevant to, but not proof of, this hypothesis. Time-resolved single-cell analysis by Xu et al. identified 19 cellular clusters across 25,776 cells and showed that mDA-directed differentiation concurrently generated ventral midbrain-, isthmic, and ventral hindbrain-like populations. Prospective enrichment of CLSTN2- or PTPRO-positive human progenitors increased mDA neuronal yield and improved the consistency of graft composition [12]. Although these markers were identified in human hPSC-derived progenitors and evaluated using human reporter lines, developmental localization of Clstn2 was additionally examined in the mouse embryonic midbrain, and graft composition, circuit integration, and therapeutic efficacy were assessed after xenotransplantation into parkinsonian mice. Species-specific differences in development and the host environment therefore limit direct extrapolation to human development or clinical performance. CLSTN2 and PTPRO should consequently be regarded as promising enrichment markers that require validation in clinically compatible human donor-cell preparations. Complementing these approaches, Schörling et al. benchmarked eight published cell-surface markers across seven regionalized human neural progenitor populations and identified APCDD1 as the marker with the highest specificity for caudal ventral midbrain progenitors. In deliberately mixed caudal ventral midbrain, rostral ventral midbrain, and ventral hindbrain populations, APCDD1-high cells generated grafts containing a higher proportion of TH⁺ neurons co-expressing FOXA2 and LMX1A and produced full recovery in the amphetamine-induced rotation test, whereas APCDD1-low cells did not improve behavior [25].

Additional studies have demonstrated prospective enrichment using human PSC-derived dopaminergic preparations. Samata et al. identified LRTM1 as a cell-surface marker for enriching human ES- and iPSC-derived mDA progenitors; sorted LRTM1-positive cells generated mDA neurons, improved motor behavior after transplantation into parkinsonian rats, and survived following transplantation into an MPTP-treated monkey [26]. Using human PSC knock-in reporter lines, de Luzy et al. showed that isolation of LMX1A-eGFP-positive ventral midbrain progenitors enriched vmDA neurons, excluded proliferative and serotonergic populations, and produced more predictable graft composition and earlier functional recovery in parkinsonian rats. In contrast, PITX3-eGFP-positive precursors survived poorly, demonstrating that enrichment outcomes depend on both marker identity and developmental stage [27]. Lehnen et al. further showed that immunomagnetic enrichment of IAP/CD47-positive human PSC-derived progenitors enriched ventral midbrain floor-plate progenitors and generated grafts with more consistent volume and dopaminergic neuron density in a rat model of PD [28]. Collectively, these studies support cell sorting as a means of improving donor-cell composition and reproducibility, while also emphasizing that marker selection, sorting stage, and validation context materially influence the resulting graft.

Beyond prospective cell enrichment, organoid and lineage-restriction studies provide complementary evidence that developmental context influences the composition and reproducibility of hPSC-derived dopaminergic cultures. Fiorenzano et al. profiled 91,034 cells across five stages of ventral midbrain organoid development and observed sequential progenitor, neuronal, vascular leptomeningeal-like, astrocytic, and oligodendroglial populations, together with inter-organoid variability that was reduced by laminin-functionalized scaffolding [4]. Maimaitili et al. further showed that engineered restriction of competing lineage programs increased the generation of caudal midbrain floor-plate progenitors from approximately 25% in conventional hESCs to 69% in lineage-restricted undifferentiated stem cells [2]. These findings demonstrate that experimentally constraining alternative lineage potential can materially alter differentiation outcome. These findings also highlight a broader source of variation that precedes dopaminergic differentiation. Large-scale characterization of 711 iPSC lines derived from 301 donors showed that inter-individual differences accounted for a substantial proportion of variation across multiple iPSC phenotypes, including differentiation capacity and cellular morphology, while recurrent copy-number alterations provided an additional source of variation [29]. More specifically, population-scale single-cell profiling of 215 iPSC lines differentiated toward a midbrain neural fate demonstrated highly reproducible line-to-line differences in neuronal differentiation efficiency and identified molecular features in the pluripotent state that predicted subsequent differentiation outcome [30]. Together, these population-scale observations demonstrate that molecular differences present before differentiation can materially influence the efficiency and cellular composition of the resulting neural cultures. The lineage-restriction study described above provides complementary experimental evidence that constraining alternative lineage potential can redirect dopaminergic differentiation.

Accordingly, reproducible dopaminergic modeling requires rigorous characterization and reporting of the starting iPSC material, including donor and cell-line provenance, derivation method, cell-line authentication, genomic integrity, functional evidence of pluripotency where appropriate, passage history, culture conditions, and relevant genetic modifications [31]. Experimental designs should additionally account for clone-to-clone, line-to-line, batch-to-batch, and operator-dependent variability through replication across independent differentiations and, where appropriate, multiple donor lines, subclones, and isogenic controls [31]. Such practices can reduce technical variability and identify unstable or poorly differentiating lines, but they should not eliminate biologically meaningful donor-specific effects. Instead, biological variation should be distinguished explicitly from variability introduced by reprogramming, cell expansion, differentiation, and experimental handling. Collectively, these studies document cellular and regional heterogeneity and demonstrate that differentiation outcomes are sensitive to the starting cell state, lineage restriction, and culture context. They do not, however, demonstrate that any resulting population constitutes a persistent artificial cellular state without a recognizable in vivo counterpart; that interpretation remains a testable hypothesis.

A major challenge in testing the spurious-state hypothesis is that developmental benchmarking itself remains incomplete. No single human reference currently spans embryonic specification, fetal maturation, postnatal development, and adulthood while providing matched transcriptomic, epigenomic, spatial, and functional information. Early ventral midbrain studies profiled human cells primarily between 6 and 11 post-conceptional weeks [9], whereas a more recent first-trimester atlas expanded single-cell and spatial coverage to brain samples from 5 to 14 post-conceptional weeks [32]. These resources substantially improve developmental resolution but remain concentrated in early development. Adult human substantia nigra profiling has additionally revealed extensive dopaminergic subtype diversity and differential vulnerability that cannot be reconstructed fully from fetal references alone [5]. In a recent preprint, Bocchi et al. integrated a single-cell and spatial atlas of the developing human diencephalon-midbrain-hindbrain axis with 19 hPSC-derived mDA datasets and harmonized adult mDA references. Their analysis reported recurrent off-target populations, a greater prevalence of hybrid lineage states in vitro than in fetal references, and partial independence between dopaminergic identity and maturation across protocols [33]. These findings remain provisional pending peer review. Consequently, apparent divergence from any individual atlas may reflect biological mismatch, incomplete reference coverage, or both.

This incompleteness creates an important benchmarking limitation: similarity to an available reference does not establish complete developmental fidelity, whereas divergence from that reference does not by itself demonstrate an artificial state. Evaluation should therefore integrate multiple stage- and region-matched references with orthogonal molecular and functional measurements.

Collectively, these findings demonstrate that differentiation outcomes are influenced by developmental timing, regional specification, starting-cell state, and culture context. They do not yet establish that abbreviated differentiation necessarily produces an inauthentic or therapeutically inferior neuronal state. We hypothesize that developmental compression becomes biologically consequential when it produces persistent divergence from appropriately stage-matched human reference trajectories or impairs state stability and function. Testing this hypothesis will require longitudinal single-cell and epigenomic profiling, stage-matched fetal and adult benchmarking, lineage tracing, and functional assessment during extended in vitro maturation and, where relevant, before and after transplantation.

The aging reset paradox: rejuvenation as a barrier to Parkinson’s disease modeling

One of the central paradoxes in PD modeling is that the very process used to generate patient-specific neurons may simultaneously erase the biological features most relevant to disease vulnerability. Although iPSC technology revolutionized human disease modeling by enabling the derivation of patient-matched dopaminergic neurons, reprogramming through pluripotency resets many donor-age-associated molecular and cellular features, producing rejuvenated derivatives that may inadequately reproduce aspects of late-onset pathogenesis [34–38]. This presents a major conceptual challenge because aging is the strongest risk factor for PD and contributes to the cumulative cellular dysfunction underlying disease emergence [15].

Reprogramming through pluripotency induces extensive cellular rejuvenation, including resetting of epigenetic age and age-associated transcriptional features and remodeling of telomere and mitochondrial states [34–38]. Huh et al. demonstrated that microRNA-mediated direct neuronal conversion preserved donor-age-associated DNA methylation, oxidative-stress responses, telomere length, and DNA-damage features [35]. Mertens et al. likewise demonstrated that directly converted neurons retained age-associated transcriptomic signatures and nucleocytoplasmic transport defects that were not retained following passage through pluripotency [38]. In a subsequent study of sporadic Alzheimer’s disease, patient-derived iNs exhibited age-dependent transcriptional and epigenetic abnormalities that were largely absent from rejuvenated iPSC-derived neurons [34].

This rejuvenation has important implications for PD modeling because aging is the strongest established risk factor for PD. Disease-associated vulnerability develops within an aging cellular environment characterized by progressive mitochondrial dysfunction, impaired proteostasis, oxidative stress, genomic instability, inflammatory dysregulation, and epigenetic alteration [15, 39, 40]. SNpc dopaminergic neurons may be particularly sensitive to these processes because autonomous pacemaking, calcium handling, extensive axonal arborization, and dopamine metabolism impose substantial bioenergetic and oxidative demands [5, 15, 39, 41, 42]. Human single-nucleus profiling has additionally identified molecularly distinct SNpc dopaminergic populations with differential vulnerability in PD and Lewy body dementia [5]. Thus, PD vulnerability likely reflects interactions among subtype identity, genetic susceptibility, aging, and cumulative cellular stress.

Reprogramming-associated rejuvenation therefore creates an application-specific limitation for modeling late-onset PD rather than an unavoidable contradiction. It may attenuate phenotypes that depend on cumulative aging or prolonged cellular stress. Experimental strategies including progerin expression, mitochondrial or oxidative challenges, proteostatic perturbation, and accelerated α-synuclein pathology have consequently been used to reveal or amplify disease-associated phenotypes in iPSC-derived neurons [36, 37, 39]. For example, progerin-induced aging exposed dendritic, mitochondrial, and neuromelanin-associated abnormalities in PD iPSC-derived dopaminergic neurons [37]. Such approaches are valuable experimental stressors, but the resulting phenotypes should not automatically be interpreted as reproducing the natural trajectory or timescale of sporadic PD.

Importantly, the issue is not simply chronological age, but the preservation of age-associated cellular state. Aging involves interacting epigenetic, transcriptional, mitochondrial, inflammatory, and proteostatic alterations [40]. Using directly converted neurons from patients with sporadic Alzheimer’s disease, Mertens et al. identified age-dependent loss of mature neuronal identity, activation of immature and stress-associated programs, and epigenetic erosion; these disease-associated signatures were largely absent from rejuvenated iPSC-derived neurons [34]. Herdy et al. subsequently identified a postmitotic senescence-like state in aged and Alzheimer’s disease-derived iNs, characterized by inflammatory signaling, metabolic dysfunction, and DNA-damage-associated programs, and found related signatures in human postmortem brain tissue [43]. In a dopaminergic context, Riessland et al. demonstrated that loss of SATB1 activated a p21-dependent senescence program in postmitotic human stem cell-derived and mouse dopaminergic neurons, accompanied by senescence-associated phenotypes and inflammatory signaling [44]. These findings suggest that aging-associated neuronal pathology can involve destabilization of mature neuronal identity and activation of senescence-like programs. Such phenomena are relevant to PD because the selective vulnerability of SNpc dopaminergic neurons emerges over prolonged timescales through interactions among cellular aging, subtype-specific physiology, genetic susceptibility, and environmental stress [5, 15, 39, 41, 44]. Consequently, rejuvenated iPSC-derived neurons may reproduce genetic predisposition while underrepresenting age-dependent cellular states that contribute to neurodegeneration. This limitation may be particularly relevant to sporadic PD, for which aging and non-genetic influences cannot be modeled adequately through disease-associated genotype alone.

Direct conversion preserves donor-age features more effectively than pluripotency-based reprogramming, but directly converted neurons lack an expandable pluripotent or progenitor intermediate and may exhibit variable conversion efficiency and cellular heterogeneity [34–36, 38]. Because direct conversion also bypasses intermediate developmental stages, model selection requires balancing age fidelity against experimental scalability, cellular consistency, and the capacity to investigate developmental lineage specification.

Collectively, these findings suggest that biological youthfulness is an important limitation when iPSC-derived models are used to investigate age-dependent PD mechanisms. The challenge is not merely generating dopaminergic neurons, but determining which aging-associated features are necessary for the disease mechanism under investigation. Future progress may therefore require strategies that combine developmental fidelity with question-specific induction or reconstruction of aging-associated cellular programs. By contrast, cellular rejuvenation may be advantageous for regenerative applications in which graft survival, stability, and functional integration are priorities.

Metabolic vulnerability and stress adaptation in iPSC-derived dopaminergic neurons

SNpc dopaminergic neurons have distinctive bioenergetic demands arising from autonomous pacemaking, calcium-dependent activity, extensive axonal arborization, and dopamine metabolism [41, 42]. These features couple neuronal activity to mitochondrial oxidative phosphorylation and expose the cells to sustained oxidative and proteostatic challenges. Whether hPSC-derived dopaminergic cultures reproduce the magnitude and chronicity of these demands in adult human SNpc neurons remains uncertain; conversely, nutrient delivery, oxygenation, and culture composition may themselves introduce stress responses that complicate interpretation.

Extensive axonal arborization increases the energetic requirements for axonal transport, membrane-potential maintenance, vesicle cycling, and neurotransmission [41]. Autonomous pacemaking and calcium entry also couple electrical activity to mitochondrial oxidative phosphorylation in SNpc dopaminergic neurons, helping to meet the energetic requirements of sustained spiking while increasing mitochondrial oxidant stress [42]. Dopamine oxidation can generate reactive quinones and other oxidized species that impair mitochondrial and lysosomal function and promote α-synuclein accumulation [45]. Together, these processes impose a substantial bioenergetic and oxidative burden, although their relative contributions to human PD vulnerability remain context dependent.

In a conditional mouse model, González-Rodríguez et al. showed that selective disruption of mitochondrial complex I in dopaminergic neurons induced an early Warburg-like metabolic shift that supported neuronal survival, followed by progressive loss of the dopaminergic phenotype, nigrostriatal dysfunction, and parkinsonism [46]. This temporal sequence demonstrates that dopaminergic neurons can initially adapt to respiratory-chain impairment, whereas prolonged complex I dysfunction ultimately compromises neuronal identity and function. However, the extent to which this experimentally induced mouse phenotype reproduces human sporadic PD or the behavior of hPSC-derived dopaminergic neurons requires direct validation.

Human iPSC-derived dopaminergic models reproduce selected mitochondrial and oxidative phenotypes in genotype- and protocol-dependent ways. Ryan et al. found that A53T SNCA-mutant dopaminergic neurons exhibited basal and toxin-enhanced nitrosative and oxidative stress, S-nitrosylation of MEF2C, impairment of the MEF2C–PGC-1α transcriptional pathway, mitochondrial dysfunction, and increased apoptotic vulnerability relative to isogenic mutation-corrected controls [47]. Burbulla et al. identified a time-dependent cascade linking mitochondrial oxidant stress, oxidized dopamine, lysosomal dysfunction, and α-synuclein accumulation in dopaminergic neurons derived from familial and idiopathic PD lines [45]. Zambon et al. further associated α-synuclein pathology in A53T SNCA and SNCA-triplication dopaminergic neurons with altered mitochondrial morphology and bioenergetic dysfunction [48]. These studies establish that human iPSC models can capture defined components of PD-associated metabolic pathology; they do not demonstrate a single convergent metabolic state across all PD genotypes or differentiation systems.

α-synuclein proteinopathy illustrates the difficulty of distinguishing disease-relevant pathology from experimentally induced phenotypes. Current iPSC-derived dopaminergic models generate or accelerate aggregation through SNCA mutation or multiplication, exogenous fibril seeding, fibrils generated de novo or amplified in the presence of PD or multiple-system-atrophy brain homogenates, and optogenetically or transgenically inducible systems [49–53]. These complementary approaches demonstrate that α-synuclein abundance, seed conformation, and mode of induction can influence aggregation, inclusion composition, mitochondrial or calcium dysfunction, and neurotoxicity. However, models based on genetic dosage, exogenous seeding, or engineered induction may not reproduce the abundance, molecular composition, anatomical distribution, or prolonged development of α-synuclein pathology in sporadic PD.

Disease relevance should therefore be evaluated longitudinally using complementary measurements of soluble and oligomeric α-synuclein, pS129 accumulation, detergent insolubility, fibrillar structure, intracellular localization, seeding activity, and neuronal toxicity rather than any single aggregation marker. Where feasible, these measurements should be combined with isogenic controls, independent differentiations, prolonged maturation, aging-associated or multicellular contexts, and comparison with human PD tissue [49–52]. These constitute proposed validation criteria rather than an established definition of a physiologically normal or pathological α-synuclein state. Importantly, α-synuclein-positive inclusions alone are insufficient to establish that a culture reproduces human PD progression; stronger disease relevance requires demonstrating a temporal relationship between molecularly characterized α-synuclein species and impaired proteostasis, synaptic or dopaminergic dysfunction, mitochondrial injury, or selective neuronal vulnerability.

A related unresolved question is whether metabolic and stress phenotypes observed in vitro arise primarily from cell-intrinsic disease mechanisms, from the culture environment, or from interactions between the two. In vivo, dopaminergic neurons interact with astrocytes, oligodendrocytes, microglia, vascular cells, extracellular matrix components, and regionally organized neuronal circuits. These cellular and extracellular components can influence neurotransmitter recycling, antioxidant defense, lipid metabolism, axonal maintenance, inflammatory signaling, nutrient delivery, and metabolic-substrate availability [54–56]. Because simplified neuronal cultures reproduce only selected components of this environment, comparisons across monoculture, defined multicellular co-culture, organoid or assembloid systems, and transplantation will be necessary to determine which metabolic phenotypes are cell intrinsic and which are context dependent [16, 18].

One important limitation involves metabolic and lysosomal incompleteness. Lipid metabolism and lysosomal glucocerebrosidase activity are closely connected to α-synuclein homeostasis. Mazzulli et al. demonstrated a bidirectional pathogenic relationship in which reduced glucocerebrosidase activity promoted α-synuclein accumulation, while increased α-synuclein impaired lysosomal glucocerebrosidase trafficking and activity [57]. Lipid and ceramide alterations associated with GBA1 dysfunction may therefore influence lysosomal function, proteostasis, and dopaminergic vulnerability. However, whether conventional hPSC-derived cultures reproduce the lipid composition and lysosomal environment of aging human SNpc tissue remains insufficiently established.

Ferroptosis represents another candidate mechanism connecting iron dyshomeostasis, lipid peroxidation, and oxidative injury in PD. Experimental inhibition of ferroptosis protected dopaminergic cells in toxin-based cellular and animal models, supporting the potential involvement of ferroptotic cell death in dopaminergic degeneration [58]. Nevertheless, the extent to which ferroptosis drives the selective degeneration of human SNpc neurons, rather than accompanying other forms of cellular injury, remains unresolved. hPSC-derived models could help address this question by combining measurements of labile iron, lipid peroxidation, glutathione and GPX4 activity, mitochondrial function, and cell-death rescue by mechanistically distinct ferroptosis inhibitors. Prolonged maturation or aging-associated paradigms may be necessary to model cumulative iron and oxidative burdens.

Redox and endoplasmic-reticulum stress phenotypes also require careful contextual interpretation. In cortical organoids, Bhaduri et al. identified ectopic glycolytic and endoplasmic-reticulum stress programs associated with impaired neuronal subtype specification; transplantation reduced these stress signatures [59]. Seiler et al. subsequently showed that microfluidic culture reduced glycolytic stress relative to conventional suspension culture in cerebral organoids [60]. Although these studies were not conducted in dopaminergic systems, they demonstrate that culture configuration, nutrient delivery, and tissue environment can alter neuronal stress programs. Elevated ROS, mitochondrial fragmentation, unfolded-protein responses, and stress-associated transcriptional changes in hPSC-derived dopaminergic neurons should therefore be evaluated across culture conditions and against isogenic controls before being interpreted as disease-specific phenotypes.

Evidence from multicellular models provides an empirical basis for examining how circuit organization and culture context influence dopaminergic state. Spatially arranged ventral midbrain–striatum–cortex assembloids formed long-range dopaminergic projections into striatal and cortical compartments and enabled analysis of target innervation, optogenetically evoked dopamine release, and functional trans-synaptic connectivity [18]. These capabilities cannot be assessed fully in isolated neuronal cultures; however, the findings demonstrate the experimental value of reconstructing circuit context rather than establishing that monocultured dopaminergic neurons are intrinsically abnormal.

Accordingly, the possibility that simplified culture environments contribute to stress-associated dopaminergic states should be treated as a testable hypothesis rather than an established conclusion. This hypothesis could be examined by comparing isogenic dopaminergic neurons across monoculture, glial co-culture, assembloid, and transplantation conditions using longitudinal transcriptomic stress signatures, metabolic flux, mitochondrial function, electrophysiology, dopamine handling, and alignment to stage- and region-matched reference states. Persistence of a phenotype across these environments would strengthen evidence for a cell-intrinsic or disease-related origin, whereas partial or complete normalization following restoration of environmental support would indicate a substantial culture-context contribution. Thus, the proposed framework builds upon established evidence that cellular context influences neural maturation while defining experiments required to determine its specific relevance to human SNpc dopaminergic neurons.

Ecosystem dependency and emerging multicellular reconstruction strategies

Dopaminergic neuronal phenotype and function reflect interactions between cell-intrinsic programs and the surrounding developmental and physiological environment. Although iPSC-derived dopaminergic neurons can acquire canonical markers such as TH, FOXA2, LMX1A, GIRK2, PITX3, and DAT, these markers do not capture multicellular interactions that may influence maturation, metabolism, inflammatory responses, synaptic activity, and stress resilience. Astrocytes, oligodendrocytes, microglia, vascular cells, extracellular matrix components, and regionally organized neuronal circuits each reconstruct distinct dimensions of this context [18, 54, 55, 61]. Consequently, isolated neuronal cultures reproduce selected cell-intrinsic features but cannot assess every context-dependent dimension of dopaminergic state.

Astrocytes can modify dopaminergic neuronal survival and α-synuclein proteostasis through non-cell-autonomous mechanisms. In an iPSC-based co-culture model, di Domenico et al. showed that LRRK2 G2019S astrocytes induced neuritic degeneration, reduced survival, and astrocyte-derived α-synuclein accumulation in control ventral-midbrain dopaminergic neurons. Conversely, control or gene-corrected isogenic astrocytes partially reduced these phenotypes in LRRK2 G2019S neurons [61]. These findings demonstrate that astrocytic genotype and proteostatic state can influence dopaminergic pathology. Distinguishing cell-intrinsic from astrocyte-dependent effects therefore requires comparison of the same neurons in monoculture and defined co-culture, preferably using isogenic astrocyte controls.

Oligodendrocytes represent another important but frequently underrepresented component of the dopaminergic ecosystem. Beyond their established role in myelin formation and maintenance, oligodendrocytes contribute to long-term axonal integrity through metabolic support. Fünfschilling et al. showed that mature oligodendrocytes maintained CNS myelin and axonal integrity after genetic disruption of mitochondrial respiration, demonstrating their capacity to rely substantially on glycolytic metabolism [62]. Lee et al. demonstrated that monocarboxylate transporter 1 (MCT1) is enriched in oligodendroglia and that genetic reduction or pharmacological inhibition of MCT1 produces axonal injury and neuronal loss; lactate supplementation reduced neuronal toxicity under the tested in vitro conditions [55]. These studies establish myelin-independent oligodendroglial metabolic support in the broader CNS, but they did not examine human SNpc dopaminergic neurons specifically. The contribution of oligodendrocytes to the maintenance and selective vulnerability of human SNpc dopaminergic neurons therefore remains to be established directly.

Beyond their role in myelination, oligodendrocytes may support the dopaminergic ecosystem through metabolic and trophic interactions. Human substantia nigra single-nucleus profiling identified an association between PD genetic risk and oligodendrocyte-specific gene-expression patterns, supporting further investigation of their involvement in PD biology but not establishing a causal role [63]. Dopaminergic cultures lacking oligodendrocyte-lineage cells may therefore incompletely reproduce axonal metabolic support and long-term neuronal maintenance. However, the direct contributions of oligodendrocytes to the maturation and selective vulnerability of human SNpc dopaminergic neurons remain to be established experimentally.

Microglial interactions introduce an additional layer of ecosystem complexity. Microglia regulate neuronal development, synaptic remodeling, immune surveillance, and tissue homeostasis, while maladaptive microglial activation and chronic neuroinflammation contribute to PD pathogenesis [16, 17, 64, 65]. Their identity and function are strongly shaped by the surrounding tissue environment. Schafer et al. showed that microglia within xenotransplanted human neuroimmune organoids acquired more mature, homeostatic, and physiologically relevant states than those maintained in vitro [17]. Similarly, Park et al. demonstrated that iPSC-derived microglia incorporated into brain organoids promoted neuronal progenitor differentiation, axonogenesis, synaptogenesis, and electrophysiological maturation through cholesterol transfer to neural progenitor cells [16]. These studies demonstrate that microglial state and neuronal maturation can be modified by multicellular and in vivo environmental context.

The vascular compartment also plays a crucial role in neuronal homeostasis. In vivo, the blood-brain barrier (BBB) is a specialized neurovascular interface formed through interactions among endothelial cells, pericytes, astrocytes, immune cells, and neurons [66]. Barrier regulation, nutrient and oxygen delivery, metabolic exchange, and neurovascular signaling influence the environment in which neurons function. These interactions may be particularly relevant to SNpc dopaminergic neurons because of their high energetic demands and vulnerability to mitochondrial dysfunction. Human BBB assembloids provide an experimental platform for reconstructing selected components of this neurovascular interface. Dao et al. generated human BBB assembloids by fusing PSC-derived brain and blood-vessel organoids, producing spatially organized interactions among neural, endothelial, pericyte-like, and astrocytic populations together with measurable barrier properties [67]. Similarly, Pediaditakis et al. developed a microfluidic human brain-chip containing dopaminergic neurons, astrocytes, microglia, pericytes, and endothelial cells and demonstrated that α-synuclein pathology was accompanied by neuronal dysfunction, inflammatory responses, and impairment of BBB integrity [68]. Although neither platform reproduces the complete circulation or neurovascular architecture of the human brain, these studies demonstrate that vascular and neuroimmune context can materially influence the expression and interpretation of PD-related neuronal phenotypes.

Circuit-level integration remains incompletely represented in conventional iPSC-derived dopaminergic cultures. Reumann et al. developed spatially organized ventral midbrain–striatum–cortex assembloids that formed long-range dopaminergic projections into striatal and cortical compartments and enabled analysis of target innervation, optogenetically evoked dopamine release, and functional trans-synaptic connectivity [18]. These capabilities cannot be evaluated fully in isolated neuronal cultures, although their absence does not establish that monocultured dopaminergic neurons are intrinsically abnormal.

Available multicellular platforms reconstruct complementary dimensions of the dopaminergic ecosystem. Microglia-containing organoids enable investigation of neuroimmune interactions, including cholesterol transfer from microglia to neural progenitor cells that promotes axonogenesis, synaptogenesis, and electrophysiological maturation [16]. Xenotransplantation of neuroimmune organoids additionally supports vascularization and the acquisition of more mature, homeostatic microglial states [17]. BBB assembloids and microfluidic brain-chip systems reconstruct selected neurovascular interactions, barrier organization and permeability, and disease-associated BBB dysfunction, including α-synuclein-associated barrier impairment [67, 68]. Oligodendrocyte-containing cultures provide complementary platforms for examining myelination, axonal maintenance, and oligodendrocyte–neuron metabolic coupling [54, 55]. Transplantation provides an integrated host environment in which dopaminergic progenitors can undergo long-term maturation, subtype specification, target-related innervation, and synaptic integration, while also revealing lineage diversification, graft heterogeneity, and off-target populations [6, 7, 13]. Because no individual platform reproduces the complete human SNpc ecosystem, model selection should be guided by the biological question: circuit assembloids for connectivity, microglia-containing organoids for neuroimmune interactions, oligodendrocyte-containing systems for axonal and metabolic support, BBB and brain-chip systems for neurovascular biology, and transplantation for integrated maturation and therapeutic function.

Despite substantial advances, current organoid and co-culture systems reconstruct only selected components of the human CNS ecosystem. Common limitations include incomplete vascularization and oxygen delivery, immature or variable cellular states, absent systemic immune interactions, limited long-range connectivity, and incomplete regional patterning [16–18, 67, 69]. Microglia-containing organoids likewise do not fully reproduce human developmental timing, immune diversity, or mature neurovascular organization, although xenotransplantation can improve vascularization and microglial maturation [17]. These limitations do not negate the value of these platforms but require their findings to be interpreted relative to the specific biological features reconstructed by each model.

Collectively, these advances extend the field’s transition toward a systems-level understanding in which dopaminergic identity is shaped by interactions between intrinsic cellular programs and the surrounding developmental and physiological environment. Assembloids, multicellular organoids, microfluidic systems, and transplantation models enable complementary investigation of circuit, neuroimmune, neurovascular, and metabolic biology. No individual platform reproduces the complete human SNpc ecosystem, but their question-guided application provides a framework for determining how environmental context influences dopaminergic maturation, function, and vulnerability.

Toward authentic cell-state engineering

The next generation of PD modeling and regenerative strategies will likely depend not on increasingly efficient production of TH-positive cells, but on the ability to generate bona fide midbrain dopaminergic neurons with biologically coherent and functionally stable cellular states. Current differentiation pipelines often emphasize endpoint marker acquisition, yet accumulating evidence suggests that neuronal identity is a multidimensional systems-level property shaped by developmental trajectory, epigenetic memory, metabolic adaptation, circuit integration, immune interactions, and long-term stress resilience [4, 5, 16–18, 70–72]. Consequently, future progress may require a conceptual transition from phenotype engineering toward authentic cell-state engineering.

Accordingly, future validation frameworks should combine transcriptomic, epigenomic, electrophysiological, metabolic, spatial, and functional assessments rather than relying on any individual modality [69]. Single-cell multi-omics can resolve relationships among transcriptional programs, chromatin accessibility, developmental trajectory, and cellular context, while RNA-velocity and pseudotemporal approaches can identify transient states and infer developmental progression [11]. Successful cell-state engineering may therefore require evaluation of both endpoint phenotypes and the developmental trajectories through which those phenotypes arise.

The organoid, assembloid, transplantation, and direct-conversion evidence reviewed above indicates that developmental architecture, circuit activity, multicellular context, and biological age can influence dopaminergic cell-state maturation and stability [4, 16–18, 71]. These observations support incorporating question-specific stress testing into future quality-control frameworks rather than relying on baseline phenotyping alone. Engineered neurons should be evaluated under defined oxidative, mitochondrial, inflammatory, dopamine-associated oxidative, ferroptotic, or sustained metabolic challenges, with longitudinal assessment of molecular state, electrophysiology, dopamine handling, and survival [46, 47, 73, 74]. Where appropriate, responses should be compared across monoculture, multicellular culture, assembloid, and transplantation environments to distinguish cell-intrinsic vulnerability from culture-dependent adaptation.

Computational approaches may further improve evaluation of cellular fidelity. Single-cell reference datasets increasingly enable construction of cell-state landscapes integrating transcriptomic, epigenomic, developmental, and spatial information across healthy and diseased tissues [5, 70, 72]. CONCORD provides a framework for integrating large-scale single-cell datasets while preserving biologically informative cell-state structure, lineage relationships, and rare populations [72]. Rather than relying on canonical markers alone, engineered neurons should be compared with appropriately stage- and region-matched human reference datasets, while recognizing that existing SNpc reference atlases remain incomplete. Such approaches may eventually support probabilistic evaluation of differentiation fidelity, stress vulnerability, long-term state stability, and disease relevance.

Application-specific criteria for disease modeling and cell replacement

The requirements for PD disease modeling and cell-replacement therapy overlap but are not identical. Disease models should reproduce the mechanisms relevant to the specific research question, which may include donor genotype, biological aging, α-synuclein and proteostatic abnormalities, mitochondrial dysfunction, stress susceptibility, and selective degeneration of vulnerable dopaminergic subtypes [8, 15, 46, 49, 50, 73]. By contrast, therapeutic donor cells need not reproduce every aging-associated or disease-vulnerable feature of endogenous SNpc neurons because the objective is to restore dopaminergic function rather than recapitulate disease progression. Features that increase stress susceptibility or compromise survival would be undesirable in a therapeutic graft. Accordingly, cell-replacement products should instead be evaluated for appropriate regional and lineage identity, controlled cellular composition, safety, capacity for in vivo maturation, target-appropriate host innervation, striatal dopamine restoration, and sustained functional benefit [1, 6, 7, 12, 13]. Biological fidelity should therefore be evaluated relative to the intended application rather than through a single universal standard.

This distinction also helps explain why stage-defined ventral midbrain dopaminergic progenitors, rather than terminally differentiated neurons, are generally used for transplantation. Clinical-development studies have demonstrated large-scale manufacture, cryopreservation, transplantation, and subsequent in vivo maturation of dopaminergic progenitors, while isolation of more differentiated PITX3-positive precursors resulted in poor graft survival [1, 6, 27]. Nevertheless, no single in vitro biomarker or assay has been clinically validated to predict graft composition, maturation, safety, and therapeutic performance. Donor-product characterization therefore combines regional and dopaminergic-lineage identity, cellular composition, viability, residual pluripotency and proliferative-cell assessment, sterility and genomic-quality testing, and preclinical evaluation of tumorigenicity and functional efficacy [1, 7, 12, 26–28]. CLSTN2, PTPRO, LRTM1, LMX1A-based selection, and IAP/CD47 have improved progenitor enrichment or graft consistency in preclinical studies, but their performance depends on differentiation stage, donor-cell preparation, and transplantation context [12, 26–28]. Establishing clinically compatible release criteria that prospectively correlate with graft safety, composition, integration, and functional benefit remains an important translational priority.

To operationalize these application-specific priorities within the broader framework, we outline testable predictions and corresponding experimental strategies for each conceptual dimension (Table 1). These approaches are not intended as mandatory components of every study, but as question-specific modules through which the framework can be supported, refined, or challenged.

Table 1.

Experimental roadmap for testing the proposed cell-state framework

Framework component	Proposed experimental design	Principal readouts	Interpretation
Phenotypic illusion	Compare independently generated TH⁺/FOXA2⁺/LMX1A⁺ populations with similar marker profiles using stage- and region-matched single-cell reference mapping, electrophysiology, dopamine handling, and transplantation [4, 6, 7, 9, 12, 13].	Regional identity scores, transcriptomic and epigenomic concordance, dopamine release, pacemaking, projection specificity and graft function.	Divergence despite similar marker expression would support the hypothesis; consistent multidimensional concordance would weaken it.
Developmental compression	Differentiate isogenic hPSC lines using accelerated versus developmentally paced patterning, with matched longitudinal sampling and lineage tracing [4, 9, 11, 12].	Single-cell RNA/ATAC sequencing, RNA velocity, lineage composition, temporal competence markers and functional maturation.	Persistent divergence or impaired function after accelerated differentiation would support a consequential effect of compression; eventual equivalent convergence would argue that compression is primarily temporal.
Putative spurious cell states	Track candidate populations across prolonged maturation, repeated batches, environmental rescue, and transplantation.	State persistence, atlas alignment, chromatin accessibility, functional identity and convergence after environmental restoration.	A persistent multimodally discordant state that fails to converge would support the designation; mapping to a recognized developmental stage or subsequent convergence would classify it as immaturity or physiological heterogeneity.
Biological-age resetting	Compare isogenic donor cells differentiated through iPSC reprogramming and direct neuronal conversion, with and without experimentally induced aging [34, 35, 39, 43, 71].	Epigenetic clocks, DNA damage, nucleocytoplasmic transport, mitochondrial function, proteostasis and PD-related stress susceptibility.	Restoration of age-associated vulnerability by direct conversion or induced aging would support the aging-reset hypothesis.
Metabolic and stress-adaptive incompleteness	Apply metabolic maturation and defined oxidative, mitochondrial, dopamine-related, proteotoxic, and ferroptotic challenges [14, 46, 47, 73, 75, 76].	Seahorse analysis, isotope tracing, mitochondrial respiration, dopamine oxidation, ROS, lipid peroxidation, calcium handling and survival.	Improved alignment with adult SNpc reference states and physiologically appropriate stress responses would indicate increased biological fidelity.
Ecosystem dependency	Compare the same neuronal preparation in monoculture; astrocyte-, oligodendrocyte-, microglial-, or vascular-cell co-culture; organoids; assembloids; microfluidic systems; and transplantation [16–18, 55, 62–65, 67, 68, 74].	Multicellular signaling, neuronal maturation, synaptic activity, myelination, axonal integrity, oligodendrocyte–neuron metabolic coupling, inflammatory responses, barrier function, and circuit integration.	Context-dependent changes would quantify which ecosystem components are necessary, beneficial, neutral, or detrimental.
Cell-state engineering	Use factorial perturbation, CRISPR-based screening, or adaptive protocol optimization to manipulate developmental timing, metabolism, aging and environmental context [69, 70, 72].	Longitudinal multi-omic state scores, functional stability, stress resilience, reproducibility and transplantation outcomes.	Successful engineering should produce sustained multidimensional convergence rather than improvement in marker expression alone.
Application-specific validation priorities	Apply distinct validation workflows according to intended use. For disease modeling, compare patient-derived and isogenic-control neurons under aging- and disease-relevant challenges. For cell replacement, evaluate stage-defined ventral midbrain progenitors using donor-product characterization and preclinical transplantation [1, 5–7, 12, 13, 15, 26–28, 46, 49, 50, 73].	Disease modeling: genotype-dependent phenotypes, α-synuclein pathology, mitochondrial and proteostatic dysfunction, stress susceptibility, and selective neuronal degeneration. Cell replacement: regional identity, composition, viability, residual pluripotency or proliferation, graft safety, maturation, host innervation, dopamine restoration, and functional benefit.	Disease-model validity requires reproducible recapitulation of the mechanism under investigation, whereas therapeutic suitability requires a safe progenitor product capable of in vivo maturation and functional restoration. No single universal benchmark is sufficient for both applications.

Implementing every dimension of this framework within a single standardized culture model would be technically and resource intensive and is unlikely to be feasible for many research groups at present. Complex neural models commonly require prolonged culture, multiple cell lines and differentiation batches, sophisticated functional and omics assays, and, in transplantation studies, specialized surgical expertise and extended follow-up [69]. We therefore propose a modular, tiered implementation strategy in which core validation could comprise well-documented and reproducible differentiation, regional identity assessment, basic electrophysiology, and dopamine-related function, whereas advanced modules could incorporate developmental reference mapping, stress-response profiling, multicellular co-culture, organoids or assembloids, multi-omics, and transplantation according to the scientific question [69]. Cross-laboratory implementation will additionally require well-characterized and accessible reference cell lines, transparent protocols, benchmark datasets, standardized reporting criteria, replication across lines and users, and access to shared technical infrastructure [31, 69]. Because prolonged culture, high-dimensional profiling, advanced imaging, and animal transplantation impose substantial technical and resource demands, sustained support from funding agencies, multicenter collaborations, and shared research infrastructure will be important for establishing, comparing, and disseminating these models. Funding mechanisms should also recognize the costs of appropriate cell-line characterization, quality control, and reproducibility measures as necessary components of rigorous stem-cell research [31, 69].

Conclusion

The field of dopaminergic neuron engineering is approaching an important conceptual transition. The central challenge is no longer simply generating TH-positive cells, but reconstructing biologically authentic dopaminergic cell states that faithfully reproduce the developmental, metabolic, aging-associated, and ecosystem-dependent properties of vulnerable human substantia nigra neurons. Increasing evidence suggests that phenotypic resemblance and marker positivity alone may inadequately capture genuine biological identity, as in vitro-derived neurons can remain transcriptionally incomplete, metabolically distorted, developmentally compressed, or environmentally artificial relative to authentic in vivo states.

Current iPSC-based systems have transformed PD research by enabling patient-specific disease modeling and regenerative investigation. However, persistent limitations including developmental trajectory divergence, epigenetic rejuvenation, metabolic incompleteness, ecosystem deprivation, and unstable maturation states indicate that neuronal identity cannot be reduced to static marker acquisition alone. Emerging single-cell atlases, organoid systems, assembloids, and neuroimmune co-culture models increasingly demonstrate that authentic neuronal states arise through dynamic interactions among developmental timing, chromatin regulation, metabolism, circuit activity, immune signaling, vascular coupling, and multicellular ecosystem integration.

For PD modeling, this framework provides criteria for distinguishing disease-relevant phenotypes from effects introduced by reprogramming, developmental immaturity, or artificial culture conditions by evaluating aging-associated vulnerability, α-synuclein and proteostatic pathology, mitochondrial dysfunction, selective dopaminergic degeneration, and multicellular context. For regenerative neuroscience, it instead prioritizes stage-appropriate progenitor identity, controlled donor-product composition, safety, capacity for in vivo maturation, host-circuit integration, and sustained functional benefit. Thus, biological fidelity should be judged relative to the intended application: faithful mechanistic recapitulation for disease models and safe, reproducible functional restoration for cell-replacement therapies.

Collectively, these findings support a shift from phenotype engineering toward authentic cell-state engineering. Future translational success will likely depend not only on generating increasingly pure TH-positive cell populations, but on engineering bona fide midbrain dopaminergic neurons with biologically coherent states capable of maintaining authentic functionality across aging, stress, and long-term environmental challenge. Achieving this goal may represent one of the most important remaining frontiers in human dopaminergic neuroscience.

Authors’ contributions

All authors contributed to the manuscript.

Funding

No funding was received to support this study.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.