Out of Line or Altered States? Neural Progenitors as a Target in a Polygenic Neurodevelopmental Disorder

Abstract The genesis of a mature complement of neurons is thought to require, at least in part, precursor cell lineages in which neural progenitors have distinct identities recognized by exclusive expression of one or a few molecular markers. Nevertheless, limited progenitor types distinguished by specific markers and lineal progression through such subclasses cannot easily yield the magnitude of neuronal diversity in most regions of the nervous system. The late Verne Caviness, to whom this edition of Developmental Neuroscience is dedicated, recognized this mismatch. In his pioneering work on the histogenesis of the cerebral cortex, he acknowledged the additional flexibility required to generate multiple classes of cortical projection and interneurons. This flexibility may be accomplished by establishing cell states in which levels rather than binary expression or repression of individual genes vary across each progenitor’s shared transcriptome. Such states may reflect local, stochastic signaling via soluble factors or coincidence of cell surface ligand/receptor pairs in subsets of neighboring progenitors. This probabilistic, rather than determined, signaling could modify transcription levels via multiple pathways within an apparently uniform population of progenitors. Progenitor states, therefore, rather than lineal relationships between types may underlie the generation of neuronal diversity in most regions of the nervous system. Moreover, mechanisms that influence variation required for flexible progenitor states may be targets for pathological changes in a broad range of neurodevelopmental disorders, especially those with polygenic origins.


Introduction
Neuronal stem and progenitor cells have been suggested as prime targets for pathogenesis in a broad range of clinically diagnosed neurodevelopmental disorders (NDDs) including intellectual disability (ID), autistic spectrum disorder (ASD), attention deficit/ hyperactivity disorder (ADHD), and schizophrenia (Scz).This suggestion was presaged by the pioneering work of Verne Caviness and his colleagues, beginning in the 1970s.Verne's early work defined neural progenitor dynamics in the developing cerebral cortex, cerebellum, and hippocampus.It also established the vulnerability of these progenitors to single gene mutations that compromise neurogenesis in "neurological mutant" mice (reviewed by [1][2][3][4]).The subsequent arguments for NDD-mediated neural progenitor pathology were based primarily upon changes in numbers of neuronal progeny in the apparent absence of extensive cell death, especially in the neocortex of genetic NDD animal models including those based upon single genes mutated in humans with a broad range of NDD diagnoses [5][6][7].In humans with clinically or genetically defined NDDs, disrupted neurogenesis in the cortex and other sites has been inferred from postmortem tissue analyses or gray matter thickness measurements in structural imaging studies [8][9][10].Parallel analyses of single gene mutations as well as polygenic disruptions, especially copy number variants (CNVs) associated with NDD risk in genomically accurate animal models, has identified disrupted neural progenitor proliferation and neurogenic capacity as key contributors to pathogenesis underlying circuit dysfunction and behavioral deficits [11][12][13].In these instances, neural progenitor proliferative characteristics and subsequent neurogenesis are altered; however, the mechanisms underlying many of these changes remain uncertain.
Inspired by Verne's work on neurological mutant mice and the use of genetic approaches to better understand neural progenitor dynamics and NDD pathology it inspired, we have analyzed the consequences of polygenic, dosage-based CNV disruption for multiple classes of neural precursors in mouse models of 22q11.2deletion syndrome (22q11DS).22q11DS is one of the most frequent CNV syndromes, and it is associated with substantially elevated risk for multiple clinically defined NDDs including ID, ASD, ADHD, and Scz.Based upon our observations in the genomically accurate LgDel 22q11DS mouse model (heterozygous deletion on mouse chromosome 16 of 28/32 contiguous orthologues of human 22q11.2deleted genes [3,[14][15][16]), as well as consequences of mutations of single genes within the region, we suggest that progenitors within broadly defined molecular classes acquire quantitative "states" that regulate their neurogenic capacity instead of progressing through precisely defined lineages.Such state-dependent mechanisms provide flexibility to generate appropriate numbers of neurons with diverse but malleable differentiation programs for optimal circuit organization and function.These probabilistic, rather than deterministic, progenitor states may be more susceptible to pathology due to CNVs or unique combinations of single gene variants thought to underlie polygenic origins of a large proportion of NDDs [8,17].
Neural Progenitors as a Target for NDD Pathogenesis Long before current classification of clinically diagnosed behavioral diseasesincluding ID, ADHD, ASD, and Sczas NDDs, disrupted development was a suspect for pathogenesis in these disorders.Some NDD disruptions, particularly those resulting in ID, were associated with perinatal hypoxia/ischemia or other birth-related complications [18,19] that result in NDD-related behavioral deficits.Nevertheless, evidence for altered cortical neurogenesis during initial brain development in ASD, Scz, and many other clinically diagnosed NDDs remains indirectrigorous analysis of neurogenesis in human fetal material, where risk for these diseases in the absence of diagnostic behavioral deficits can (at best) be inferred is difficult if not impossible.Altered neurogenesis has been suggested as a potential explanation for differences in cortical size, functional connectivity, neuronal density, and transcriptome divergence associated with ASD as well as ADHD [8,20,21] or Scz [22][23][24].Many other mechanistic explanations, however, could account for these changes in the mature brains of individuals with ASD, ADHD, or Scz.Ventricular heterotopias and cortical dysplasia in individuals with ASD may reflect disrupted cortical neurogenesis ( [25,26], reviewed by [27]).This interpretation is supported by recent cellular analyses of the roles of multiple ASD risk genes in neurogenesis and neuronal migration [28,29].Observations using induced pluripotent stem cells (IPSCs) derived from individuals with Scz suggest that IPSC-derived precursors have divergent capacities for proliferation and self-organization into cortex-like aggregates (reviewed by [30]).Despite these observations, establishing a causal relationship between altered cortical progenitor proliferation and key behavioral pathologies in NDDs like Scz and ASD remains challenging.Nevertheless, cortical neurogenesis remains a compelling, if speculative, candidate for pathogenesis in common clinical NDDS like ASD and Scz because of its central role in producing projection neurons (PNs) from which association corticocortical circuitspresumed prime targets of NDD pathology [31,32] are built.
The earliest genetic and cell biological models for pathological neurogenesis in NDDs were based upon observations, many made by Verne and his collaborators (particularly Pasko Rakic), in multiple strains of "neurological mutant" mice [2].Monogenic mutations in these mice were identified initially by mode of inheritance and distinct behavioral phenotypes (reeling, lurching, staggering, weaving) without knowledge of a specific gene or protein product target.Many of these mutations were shown to selectively disrupt differentiation of neurons and their arrangement into orderly cortices in the cerebellum, hippocampus, and cerebrum [2,[33][34][35].The likely relevance of these early studies in mutant mice to understanding cortical malformations in humans, as well as the impact of altered genetic architecture on connections and function of neural circuits, were acknowledged by Verne and Pasko throughout their now classic 1978 review "Mechanisms of Cortical Development: A View from Mutations in Mice" [2].Nevertheless, they recognized at the time (45 years ago!) that few single gene mutations had been associated with altered cortical development in humans and thus wisely refrained from additional speculation of how such disruptions result in what are now recognized as NDDs.
Seventeen years later, in 1995 [1], when Verne and Pasko revisited a central "neurological mutant" reeler, described at length in their 1978 review, they were more confident of the relevance of the genetic approach for understanding human cortical malformations and their consequences.Summarizing the significance of the identification of the reeler gene in mice and humans in this context: "Molecular and cell biological explorations of the role of the reeler gene in normal cortical histogenesis now take their place in a widening strategy of analysis of normal and abnormal histogenesis.Mutations associated with consistent patterns of cortical malformation in man will figure preeminently in this strategy" [1].Thus, observations in reeler and other neurological mutant mice anticipated, and in some ways inspired, the confluence of human genetics, human in vivo imaging, and clinical characterization of a broad range of primarily monogenic behavioral disorders as well as parallel cellular and molecular neuroscience in animal models.Together, these observations eventually confirmed that NDDs can indeed be as disorders of neurogenesis, neuronal migration, and/or subsequent cortical circuit development (reviewed by [17,32,36]).
In the intervening years, remarkable progress has been made in establishing the key role of cortical histogenesis in NDD pathology based in large measure upon identification of multiple human monogenic disorders that result in microcephaly (reviewed by [37,38]) as well as subsequent defects in migration of postmitotic cortical neuroblasts (reviewed by [39,40]).Some of these mutations target genes that control brain size by modulating cell proliferation and division [41].In addition, many human mutations, including autosomal recessive mutations in the human RELN gene [42], compromise the next step in cortical histogenesis, migration of postmitotic cortical neurons (reviewed by [43][44][45]).RELN in humans (and mice [46]) encodes an extracellular matrix protein that acts as a ligand to regulate neuronal migration and placement.RELN receptors and signaling mediators include cell surface proteins VLDLR, ApoER2, and the β1 integrin as well as additional interacting proteins like Dab1.RELN binding and subsequent activation of these proteins influences cell adhesion and motility [47][48][49].Distinctions between microcephaly and migration disorders suggest a boundary between the two processes: there may be "neurogenesis genes" that control cell cycle and cytokinesis and "migration genes" that control adhesion, signaling, cytoskeletal integrity, and motility for postmitotic neurons.Verne and his colleagues, however, recognized that neurogenesis versus migration most likely did not define an absolute boundary of molecular functions in the developing cortex [50,51].Based upon a series of studies of cortical progenitor proliferative dynamics and their modulation by cell cycle regulators and downstream transcription factors [52,53] as well as cell-cell interactions mediated by Notch signaling [54,55], Verne et al. [1] suggested that the control of cortical neurogenesis via pathways not directly engaged in cell cycle regulation, cytokinesis, or even cell adhesion and migration can influence cortical neuron laminar identity, regional specificity, and circuit differentiation.
It remains uncertain whether gene networks and local signaling engaged by these additional contributors can establish discretely defined progenitor cell types identified by singular molecular markers that are specified to generate distinct neuron classes in the cerebral cortex or indeed anywhere in the CNS or PNS (Fig. 1a).It is also possible that dynamic, statistically variable cell states, established by transient quantitative modulation of expression independent of singular molecular markers (reviewed by [56][57][58][59]; Fig. 1b), bias postmitotic neuroblasts toward potential fates.Such variable, dynamic, progenitor cell states would establish a far greater range of transcriptional activation and thus provide more flexible outcomes than those possible within a rigidly prescribed lineage defined by binary expression of single genes.These progenitor states may reflect differing probabilities of local signaling via soluble factors as well as cell surface ligands and receptors.They may also rely upon epigenetic regulation of transcription including DNA methylation and histone modifications that modify chromatin conformation and levels or probability of gene expression.In each instance, progenitor states would change quantitatively without altering binary expression distinctions that identify currently established progenitor types in prescribed lineages.The contrasting possibilities of progenitor types or states are particularly relevant for and another from non-Wnt1 neural tube progenitors (blue), migrate and coalesce with trigeminal cranial placode-derived progenitors (red).These populations, with distinct derivations, give rise, respectively, to mechanosensory (red) and nociceptive neurons (green and blue) axons extending into the distal and proximal branches of the trigeminal nerve (c after [60,61]; d based upon data from [16,62]).
analyzing how cortical histogenesisa term used frequently by Verneis targeted by genetic or environmental mechanisms that result in clinically/behaviorally defined NDDs like ASD, ADHD, and Scz.Recent analyses of disrupted neurogenesis associated with heterozygous deletion of the genes associated with 22q11DS, a polygenic CNV NDD that elevates risk for behavioral deficits parallel to those in ID, ADHD, ASD, and Scz, address fundamental questions raised by Verne and colleagues regarding dynamic regulation of neurogenesis and its impact on establishing optimal versus pathological neuron identities.Analyses in 22q11DS mouse models suggest that broad progenitor classes may be targeted by diminished dosage of 22q11 genes; however, these results also suggest that probabilities of divergent or disrupted proliferation and neurogenesis for individual progenitors within broad classes defined by singular molecular markers vary substantially.We argue that CNS and PNS neural precursor dysregulation in 22q11DS mouse models reflects mechanisms that destabilize cell states rather than disrupt specific progenitor types.Accordingly, variable quantitative, probabilistic cell states underlie divergent neurogenesis and circuit differentiation in NDDs, especially those with polygenic origins.
Studying Neurogenesis: Locations, Signals, Lineages, and States The fundamental challenge to understanding neurogenesis in any part of the nervous system is relating remarkable neuronal (and glial) diversitydefined cytologically, molecularly, and anatomically as well as by connections and physiological functionto a far more limited number of known progenitor types defined by expression of single molecular markers.It seems likely that the two extremesrigidly determined lineages and postmitotic tabule rasa (see Fig. 1a, b)are integrated so that progenitors within broad classes retain flexibility to produce postmitotic neuroblasts with biased fates that can subsequently acquire additional molecular distinctions to achieve maximal diversity.At least two cell biological mechanisms could mediate less rigidly determined changes of cell states within molecular markerdefined populations of CNS or PNS neural progenitors: (1) soluble extracellular signals available widely to responsive target precursors and external cell surface ligand/receptor partners whose signaling relies upon cell-cell contact between immediately adjacent precursors (Fig. 1b).These mechanisms, because of their flexibility and the large number of molecular mediators they engage, may also be targets of multigenic disruptions that underlie NDDs.
Soluble signals, including regulators of fundamental developmental mechanisms like induction and patterning, including Shh, Fgfs, BMPs, Wnts, and retinoic acid [63][64][65][66][67][68][69], cytokines [70,71], morphogenetic signals including EGF [72] and VEGF [73], and even neurotransmitters [74,75], are known to regulate neural progenitor proliferation and modes of division, both in vitro and in vivo.In the embryonic periphery, such signals released from nascent vasculature or by secretion from adjacent cells and tissues into the extracellular space [76][77][78] likely reach neural progenitors that generate peripheral sensory and autonomic ganglia and the enteric plexus.In addition, signals from amniotic fluid may access neural progenitors in neurogenic placodes that generate olfactory receptor neurons as well as those that contribute to the nascent sensory ganglia of the developing head [79].Signals in amniotic fluid could also influence the enteric nervous system via fetal fluid ingestion [80].In the central nervous system, these signals can be provided via the cerebrospinal fluid [80,81] as well as local secretion by subsets of progenitors or differentiating neurons: e.g., Shh by Purkinje cells for granule cell precursors [82,83], retinoic acid by subsets of adjacent spinal cord neurons [84,85], multiple signals from differentiating glia or immune cells [86], circulating signals from developing brain vasculature [87,88].
Neural progenitors also respond to contact-mediated signaling facilitated by direct interactions with neighboring cells.Such direct receptor-ligand interactions require membrane-to-membrane contact between progenitors that either signal or respond to their immediate neighbors.Thus, the position of each individual progenitor, its complement of cell surface receptors and ligands, and those of immediately adjacent cellsother progenitors, newly generated neurons and glia, vascular cells, or peripheral target tissuesis critical for establishing its proliferative and neurogenic capacity.Signals from immediate neighbors likely provide an additional, essential regulation of proliferation, modes of division, and initial differentiation capacity.These signals include ligands and receptors of the Notch/Delta pathway [54,55], the Eph/Ephrin pathway [89,90], and the cadherin/catenin pathway [91].In each instance, cell positions and the direction of interaction based upon differential signal transmission or transduction capacity of receptor/ligand pairs may result in individual responses that distinguish immediately adjacent, otherwise apparently equivalent, Neural Progenitor States and Neurodevelopmental Disorders neural progenitors.For Notch/Delta signaling, the consequences of direct local interaction include maintenance of progenitor state versus progression toward differentiation.Progenitor state dynamics via Notch signaling-mediated reflect transcriptional responses that influence proliferation as well as mode and plane of cell division: symmetric/asymmetric, self-renewing/ neurogenic, and in the developing cortex, perpendicular or parallel to the ventricular surface [55,[92][93][94].The influence of other direct cell-cell contact signaling pathways is less well characterized, but in each case, ligand/receptor interactions that depend upon direct contact between adjacent cells elicit intracellular signaling changes that influence progenitor proliferation and differentiation.
Soluble and cell-contact-dependent signals most likely diversify progenitor proliferation and the fates of their progeny beyond that inferred from classifications using a single or limited combination of molecular markers [60,61,95].Nevertheless, the identification of molecular marker-defined precursors (Fig. 1c, d) throughout the developing and mature CNS and PNS [96][97][98][99][100][101][102][103] has led to inferences of lineages considered parallel to those defined for hematopoietic differentiation.Indeed, over the last three decades, the term "neuropoiesis" has been coined to emphasize general as well as specific similarities between hematopoietic lineages and apparent progression through molecularly labeled neural progenitor subclasses to distinct postmitotic neuronal progeny [104][105][106][107].These comparisons have identified broadly shared characteristics of all stem cellse.g., transcriptional regulation of proliferative states, the cell cycle, or cellular mechanisms for cytokinesisthat can be applied to neurogenic progenitors.There are, however, significant limitations.Hematopoiesis and its parallel, neuropoiesis, in simplest forms generate a limited number of cell types from a parallel set of molecularly distinct progenitors [108] and thus cannot easily accommodate cellular diversity in the CNS or PNS.Strictly applied, neuropoiesis anticipates linear progression through progenitor types that generate committed postmitotic neuroblasts that differentiate into a singular identity.This sort of linear progression, however, especially in the mammalian CNS and PNS, may not facilitate adjustments of identity and function necessary for flexible neural circuit differentiation including neuron/target interactions and activity-dependent plasticity that occur long after neurons become postmitotic.
A new synthesis is needed to integrate the contribution of linear progression through neural progenitor classes defined by one to a few molecular markers and the transcriptional flexibility needed as progenitors generate highly diverse and adaptable neurons that comprise complex and dynamic circuits of the mammalian brain.Verne recognized that this synthesis was necessary to fully explain typical neurogenesis in the cortex as well as its likely atypical execution in multiple NDDs.The substantial progress in identifying rare monogenic human diseases that compromise neural progenitors broadly and cerebral cortex precursors specifically or selectively [17,44,[109][110][111][112] has provided essential insight into singular vulnerabilities of neural progenitors to genetic disruption.There is also, however, a need to understand how more subtle changes in neurogenesis arise in the context of the apparently probabilistic, polygenic disruptions [113,114] associated with clinically defined NDDs including (but not limited to) ADHD [115,116], ASD [8,117], and Scz [118,119].Our analyses of CNS and PNS neurogenesis (Fig. 1c, d) in the "model" CNV syndrome, 22q11DS [4,14,75] using genomically valid mouse models [75] provide insight into how polygenic disruption of neural progenitors might contribute to circuit pathology and behavioral dysfunction [3,15].We argue that these polygenic disruptions may target progenitor "states" that reflect highly variable quantitative regulation of gene expression, cell-cell interactions, and modes of cell division.

22q11 Genes and Neural Precursor Diversity
To assess potential influence of 22q11 gene dosage on neural progenitors, we first asked whether one, some, or all of the murine orthologues of the 32 genes in the minimal critical deleted region of human chromosome 22 whose heterozygous deletion results in the full spectrum of 22q11DS phenotypes [120] are expressed in neural precursors in the developing mouse CNS or PNS.A variety of observations using quantitative PCR, in situ hybridization, and immunolocalization indicate that a fairly substantial number of the 22q11 deleted genes (at least 22 of the full set of murine orthologues) are expressed throughout the developing nervous system, including in neural progenitors in the cerebral cortex, spinal cord, and neural crest [3,4,[121][122][123]. Indeed, subsequent analyses from several laboratories [124] as well as assessment of images from Genepaint (Fig. 2), a publicly available in situ localization database [125], indicate that multiple 22q11 genes are expressed both in the ventricular and subventricular zone (VZ/SVZ) of the developing cortex as well as in cranial sensory ganglia including the trigeminal ganglion (CNgV) which is derived from both neural crest-derived and cranial placode-derived precursors [126].Thus, multiple 22q11 genes are expressed in neural progenitors and may influence their capacity to proliferate as well as generate new neurons.
The diverse functional identities of the 22q11 genes (see Fig. 2) suggest an additional level of complexity of precursor regulation.Of the 32 minimal critical deleted region genes, plus 20 additional genes in the larger 3 MB region more commonly deleted in 22q11DS [120,127], none are direct cell cycle regulators, although some like Hira [128], Cdc45l [129], Ranbp1 [15], and Trmt2a [130] have been functionally implicated in control of progenitor proliferation.Most protein-coding 22q11 deleted genes fall into categories not specifically related to progenitor regulation: these include six mitochondrially localized genes [131], a limited number of transcription factors including Tbx1, a suggested 22q11DS "candidate" gene that may regulate cardiac as well as neural differentiation [132,133], modulators of chromatin integrity and DNA repair, and several cell-cell or intracellular signaling mediators and adhesion molecules [124].Thus, many 22q11 genes could influence progenitor progression; however, their dosage-dependent influence in the context of polygenic 22q11 deletion may be subtle, probabilistic and rely upon mechanisms that transcend uniform effects on defined progenitor types.To begin to determine whether and how genetic disruptions alter neural progenitors in the context of NDDsone of Verne's major goals and that of his many trainees and collaboratorswe analyzed the consequences of multigenic heterozygous 22q11 gene deletion for cortical neurogenesis as well as genesis of CNgV sensory neurons (see Fig. 1).

22q11 Deletion Disrupts Central and Peripheral Neurogenesis
The first question we asked was whether heterozygous deletion of murine orthologues of human genes associated with essential diagnostic 22q11DS phenotypes (Fig. 3) (thus, the genes in the 1.5 MB minimal critical deleted region; [14]) compromises CNS or PNS neural progenitor proliferation.We also asked whether such disruptions result in altered frequency, proportions, or identities of with cortical circuitry and function, infants and toddlers with 22q11DS have perinatal dysphagiadifficulties with suckling, feeding, and swallowingfollowed by oromotor difficulties including aberrant food ingestion, swallowing, and speech throughout the lifespan.Bottom left: location of heterozygous deletion of contiguous 22q11 orthologues that define the minimal critical deleted region associated with 22q11DS on mouse chromosome 16.Bottom right: although it is not possible to identify behavioral states in mice that fully parallel those in ADHD, ASD, Scz, or anxiety disorder, LgDel mice with a deletion on mmChr16 orthologous to that on hChr 22 have a spectrum of behavioral deficits and changes in cortical as well as peripheral neural circuits that can be related to aspects of human clinical disorders seen in individuals with 22q11DS throughout the lifespan (figure modified from [75]).neurons generated from these progenitor populations.Mice with heterozygous deletions genomically equivalent to those in 22q11DS have behavioral deficits (see Fig. 3) that can be related to those seen in individuals with 22q11DS as well as in clinically defined NDDs including ID, Scz, and ASD (as reviewed in [75]).Thus, using 22q11 mouse models, it may be possible to assess critically whether polygenic disruption alters CNS or PNS neural progenitor proliferation or the capacity of these precursors to generate neurons that contribute to specific circuits whose dysfunction underlies NDD pathology.
We focused first on cortical progenitors in the VZ/SVZ of the developing cortical mantle (see Fig. 1c) in the LgDel mouse 22q11DS model as well as related single gene mutations.This focus seemed critical because of the robust association of 22q11DS with vulnerability for clinically defined disorders of cortical circuit development including Scz and ASD that are thought to compromise PN progeny of VZ/SVZ progenitors [134,135].We found that the frequency of proliferating basal progenitors in the SVZ was selectively diminished in LgDel fetuses at midgestation [3] (Fig. 4a, top).Additional experiments suggested that the proliferative capacity of VZ (apical) precursors was not appreciably disrupted.
Our observations suggest that basal progenitors are generated by apical progenitors at similar frequencies in the LgDel, and the proliferative capacity of a subset is altered subsequently.The selective decrease in frequency of actively proliferating basal progenitors indicated that the frequency of their primary progeny: layer 2/3 PNs might also be selectively diminished compared to their primarily apical precursor-derived layer 5/6 PN counterparts.This was indeed the case (Fig. 4a, bottom).The overall frequency of NeuN (RbFox3) labeled neurons was diminished in layers 2/3 but not layers 4, 5, or 6.Moreover, there was a diminished frequency of presumed layer 2/3 PNs, labeled by Cux1, but not of layer 4, 5, or 6 PNs identified by Tbr1, a marker for these cells.
To better formulate hypotheses regarding the significance of selectively disrupted cortical neurogenesis in divergent neural circuit development underlying NDD pathology [32,136], we asked whether partial depletion of layer 2/3 PNs, apparently reflecting randomly diminished proliferation of a subset of basal progenitors, might contribute to functional deficits later in life that reflect altered association cortico-cortical connections, presumably made by layer 2/3 PNs.One such deficit, the performance of a visual reversal learning task, relies critically upon association cortico-cortical connections between the medial frontal and lateral entorhinal cortex [137,138].In a cohort of individual LgDel and WT control littermate mice, we assessed the correlation between performance of this layer 2/3 PN connectiondependent task and the frequency of Layer 2/3 PNs versus that of parvalbumin-labeled interneurons (also altered in LgDel mice [3,139]) in medial frontal association cortex in each individual mouse [140].Diminished medial frontal association cortex layer 2/3 PN frequency in individual mice was highly correlated with the magnitude of reversal learning deficit [140].In contrast, the frequency of parvalbumin-labeled interneurons does not correlate with visual reversal performance in individual mice.These results suggest that selective disruption of basal progenitor proliferation, followed by decreased frequency and perhaps divergent specification of layer 2/3 PNs, may contribute to altered association cortical circuit and related behavioral deficits due to 22q11 deletion.
Neural circuit and behavioral disruption in NDDs including 22q11DS are not confined to the cerebral cortex and its functional capacities or even the CNS [122,141,142].We found that neonatal LgDel mice have signs of altered suckling, feeding, and swallowing (S/F/S; see Fig. 3) that parallel perinatal dysphagia in infants with 22q11DS [122,126,143].Based upon the essential role of cranial sensory ganglia, particularly CNgV, in S/F/S oropharyngeal function, we asked whether neural progenitors that yield multiple classes of CNgV sensory neurons are compromised in developing LgDel mice.We analyzed cranial placode-and neural crest-derived neural progenitors (see Fig. 1d; [62]) at midgestation, right after CNgV has coalesced as a recognizable ganglion (see Fig. 2).There is a decline in the frequency of neural crest progenitors [16] and a proportional increase in placode-derived progenitors in the LgDel CNgV (Fig. 4b).This disruption is accompanied by a change in neighbor relations of placodal as well as neural crest progenitors.In the LgDel CNgV, placode cells are more likely to have a placode neighbor, while neural crest cells are less likely to have neural crest neighbors.These changes suggest that optimal local cell-cell signaling as well as neurogenic capacities of subsets of CNgV progenitors may be altered by heterozygous 22q11 deletion.
To explore this possibility, we asked whether the timing and frequency of neurogenesis changes in the LgDel CNgV.In typically developing CNgV, placodal progenitors divide more frequently early to generate mechanosensory neurons, while neural crest progenitors divide more at a slightly later time to generate nociceptive neurons [144,145].We found an increase in the frequency of postmitotic neurons in the LgDel CNgV at midgestation, accompanied by an apparent decrease in the proportion of presumed nociceptive neurons in the early postnatal ganglion (Fig. 4b, top) [16].The early increase could reflect a greater proportion of placode-derived neurons or, counterintuitively, increased frequency of neuronal progeny of neural crest precursors generated at earlier times.This latter possibility may be the case since LgDel CNgV neural crestderived progenitors yield more frequent asymmetric neurogenic divisions at the apparent expense of selfrenewing progenitor divisions (Fig. 4b, middle).This change could explain several differences in the LgDel CNgV.First, the proportion of proliferating placodederived progenitors would appear higher because the pool of self-renewing neural crest progenitors is diminished.Second, placode-derived progenitors would likely be more aggregated since the addition and presumably intercalation of additional neural crest progenitors is diminished.Third, altered proportions of mechanosensory versus nociceptive neurons would reflect diminished nociceptive neurogenesis at later stages of CNgV development due to a depleted progenitor pool.Thus, 22q11 deletion could alter proportions of subsets of CNgV sensory neurons by altering neighbor relationships and modes of division.These changes may reflect differences in progenitor state rather than identity: affected subsets of neural crest progenitors retain expression of established molecular markers.Due to probabilistically altered cell-cell interactions, however, their proliferative properties diverge from those of others in the same lineage.
The potential consequences of these changes among subsets of LgDel CNgV progenitors may include altered axon growth and guidance for the peripheral and central branches of CNgV neurons during initial ganglion differentiation [122,146] as well as changes in proportions, sizes, and distribution of CNgV neurons with molecular characteristics of mechanoreceptive as well as nociceptive sensory neurons (Fig. 4b, bottom) [16].These disruptions of CNgV differentiation suggest that altered interactions between subsets of placode and neural crest progenitors due to 22q11 deletion may impact oropharyngeal sensory function.We have shown that newborn LgDel mice have S/F/S-related difficulties [122,147] and that feeding behaviors continue to be altered in adult LgDel mice [143].We have not yet established definitive evidence of a causal connection between early CNgV neural progenitor disruption, premature neurogenesis, quantitative changes in CNgV mechanosensory and nociceptive neurons, and oropharyngeal dysfunction in S/F/S.Nevertheless, changes in nociceptive versus mechanoreceptive sensory neuron identities could disrupt peripheral sensory feedback that controls milk intake early in life and successful ingestion of solid food later in the lifespan.

Up-and Downstream of Diminished Dosage: 22q11 Genes as Regulators of Progenitor States
The two instances of altered neural progenitor proliferation in the LgDel 22q11DS mouse model indicate that multigenic, dosage-dependent alterations alter neurogenic capacity of somebut not allprogenitors within molecular marker-defined subclasses.An essential question is how these disruptions result from dosage changes of 22q11 genes expressed in CNS and PNS progenitors (see Fig. 2; [4,131]).We and others have approached this question by analyzing the consequences of loss-or gain-of-function of individual 22q11 genes as well as characterizing molecular changes that are likely downstream of diminished multigenic 22q11 gene dosage.Our results suggest that a combination of changes in mechanisms that regulate cell proliferation as well as more global changes in transcriptional regulation modify neural progenitors in the context of heterozygous 22q11 gene deletion.
CNV disorders are often analyzed as "contiguous gene syndromes" in which individual genes within a larger deleted or duplicated chromosomal region are associated with distinct phenotypes.Our functional genomic, informatic, and literature-based assessments of genes within the 1.5 MB 22q11DS minimal critical deleted as well as the more common 3.0 MB typically deleted region suggest that none of the 22q11 genes regulate primarily cell cycle or cytokinesis [3,4,124].Nevertheless, several of the 22q11 deleted genes may influence neural progenitor progression, based upon expression localization (see Fig. 2) and function in pathways that can modulate cell proliferation and division in addition to other signal transduction cascades.These include the chromatin elongation/duplication regulator CDC45 and the nuclear/cytoplasmic transport regulator RANBP1.Based upon its function in chromosomal duplication during cell division as well as localization in subsets of dividing cells, CDC45 is a potential candidate for disrupted neural progenitor disruption due to 22q11 deletion.Nevertheless, potential functions of CDC45 in neural progenitors have not been explored in animal models due to early postimplantation lethality of homozygous Cdc45 null mutations, as well as a lack of apparent phenotypes in Cdc45 heterozygous mutant mice [148].
Among additional candidate genes, we focused upon Ranbp1 as a potential regulator of basal progenitor function based upon its high level of expression in the SVZ and its modulation of signaling that can influence cell cycle progression and cell division [149].We developed a mouse model with a null mutation of the Ranbp1 gene [15].Full Ranbp1 loss-of-function is lethal around birth and has a variety of developmental consequences including acting as an apparent microcephaly gene [146].Gross microcephaly is seen in late gestation Ranbp1 −/− fetuses.The reduced size of the cerebral cortex reflects disruptions of both early apical (Fig. 5, middle) and later basal progenitor populations (Fig. 5, bottom).The consequences for the basal progenitor population appear most substantial: layer 2/3 PN numbers are dramatically depleted (Fig. 5, bottom) so that cortical thickness as well Neural Progenitor States and Neurodevelopmental Disorders Dev Neurosci 2024;46:1-21 DOI: 10.1159/000530898 Fig. 5. Homozygous null mutation of the 22q11 deleted gene Ranbp1 results in microcephaly due to diminished apical and basal progenitor proliferation.Top panels: at E18.5, Ranbp1 −/− mice, who die at birth, are visibly microcephalic, based upon head size (not shown) and brain size, especially that of the cerebral hemispheres.These changes in brain size are visible both from a dorsal (left) and latral (right) view.Middle panels: at E10.5, when the cortical rudiment is still a proliferative neurepithelium with a small number of postmitotic neurons, active proliferation of neuroepithelial/apical progenitors, labeled with Ki67, is diminished in Ranbp1 −/− fetuses, as is the frequency of PH3+ presumed mitotic progenitors.Lower panels: by E14, when basal progenitor genesis from apical progenitors has begun, there are fewer Tbr2+ basal progenitors acutely labeled by BrdU in Ranbp1 −/− fetuses.By E17.5, when an appreciable number of layer 2/3 PNs, labeled here by Cux1, have accumulated in the WT cortex, there are significantly fewer Cux1+ layer 2/3 PNs in the Ranbp1 −/− cortex; however, the frequency of Ctip2+ layer 5/6 PNs is not significantly reduced (all panels adapted from [15]).as cortical size is diminished.The Ranbp1 −/− cortical phenotype suggests that heterozygous deletion of Ranpb1, acting as a contiguous gene, may account for cortical progenitor phenotypes seen with broader heterozygous 22q11 deletion.Basal progenitor proliferation, however, does not change significantly in heterozygous Ranbp1+/− mutants [146].Thus, diminished basal progenitor proliferation in LgDel likely reflects dosage-dependent interactions between Ranbp1 and additional 22q11 genes with diverse expression and functions in cortical Fig. 6.Transcriptional divergence of E10.5 LgDel CNgV neural progenitors and newly generated neurons.Top panels: we dissected CNgV from WT and LgDel E10.5 fetuses, an age when substantial numbers of placodal (Six1+, red) and neural crest (Wnt1 Cre-recombined + or −, green and blue, respectively) can be seen in the ganglion (far right).These dissected CNgV from multiple embryos and multiple litters were pooled to collect 5 biological replicates for RNAseq transcriptome analysis.Middle panels: comparison of the 5 biological replicate CNgV transcriptomes identified 134 genes that are differentially expressed in LgDel versus WT, based upon a significance threshold of FDR q < 0.1 (left).Of these 134 genes, 38 can be informatically identified as potential targets for transcriptional regulation by Six1, a diagnostic transcription factor expressed in placodederived CNgV progenitors, and 130 can be identified as potential targets for transcriptional regulation by Sox10, a diagnostic transcription factor marker for a subpopulation of neural crest-derived CNgV progenitors (right).Lower panels: there is a significant transcriptome-wide increase in the coefficient of variation for genes expressed in the E10.5 LgDel versus WT CNgV.Increased variation is seen for 22q11 genes whose expression is detected in the E10.5 CNgV (red dots) as well as the 134 genes differentially expressed between LgDel and WT (orange dots).Increased variation can be seen across functional categories (right) that include fundamental cellular metabolic and protein/membrane trafficking processes (amino acid synthesis and Golgi) as well as those specific for neural progenitors and differentiating neurons (proneural genes, axon extension genes).Asterisks indicate significance levels as shown in the top left corner (all panels adapted from [123]).
Neural Progenitor States and Neurodevelopmental Disorders progenitors (see Fig. 2).Disruption of these interactions by 22q11 deletion is likely to contribute to aberrant differentiation of layer 2/3 PNs as well as related behavioral deficits [75,150].
The impact of diminished dosage of multiple 22q11 genes on transcriptional regulation that ultimately defines neural progenitor proliferative and neurogenic capacities is not yet clear.To begin to address this question, we analyzed the transcriptomes of LgDel and WT CNgV at E10.5 (Fig. 6, top), when there is peak proliferation of both placode-and neural crest-derived progenitors in the nascent ganglia (see Fig. 5 and [62]).We assessed differential gene expression in WT versus LgDel CNgV (Fig. 6, middle right).We identified differentially expressed genes among the approximately 17,200 significantly expressed transcripts shared by WT and LgDel CNgV by comparing mean expression levels of each across 5 WT and LgDel replicate transcriptome samples [123].There is a relatively modest number -134 (0.7% of total shared transcripts)of significantly differentially expressed genes, including a subset of 22q11 genes whose expression diminishes by the expected 50%.The range of fold changesincrease or decreaseis substantial: from nearly 200 to less than 2-fold.The absolute expression levels of most differentially expressed transcripts, however, are modest.Moreover, there is no single functional signature among the group.We further assessed the association of the differentially expressed genes with progenitor classes whose identity is defined by distinct transcription factors (Fig. 6, middle right).Based upon informatic genomic sequence analysis, most differentially expressed loci have binding sites for Six1 (placode), as well as Sox10 and FoxD3 (neural crest subpopulations) [62,151,152].In contrast, no differentially expressed genes were informatically associated with Brn3a, a transcription factor seen in differentiating CNgV neurons.It is possible that levels of differential CNgV gene expression reflect distinct transcriptional regulation of the same genes in placodal versus neural crest progenitors.
In addition to diversity of transcriptional regulation of common targets, expression levels in the LgDel versus WT CNgV appear to vary more across the entire transcriptome, despite presumed reduction of variability realized by pooling samples across embryos and litters [123].To quantify this impression, we determined the coefficient of variation for each of the 17,000+ shared transcripts and found that approximately 15,000 (88%: including several 22q11 and other differentially expressed genes) had a greater coefficient of variation in LgDel than WT (Fig. 6, bottom left).Moreover, expression levels of transcripts from gene ontology categories to which the differentially expressed genes belong are significantly more variable in LgDel CNgV (Fig. 6).These data suggest that progenitor identities reflect transcript levels, rather than singular expression or repression, of a broad spectrum of genes.When diminished dosage of 22q11 genes disrupts these quantitatively defined transcriptional states, the proliferative characteristics and neurogenic capacity of individual progenitors within a quantitatively diverse population may be destabilized.These potentially stochastic changes in variable expression independent of progenitor types identified by unique markers could result in changes in number of neurons generated as well as the capacity of postmitotic progeny to differentiate optimally.
Lining Up for Altered States: Balancing Progenitor Identities with Neurogenic Flexibility Among the complexities of cortical neurogenesis that Verne recognized in the 1970s and explored in depth over the subsequent four decades, he identified two fundamental characteristics of neural progenitors that remain central for understanding brain development and its disruption by NDDs.First, broad molecular identities of neural progenitors (i.e., those resolved by cell class-specific markers or distinct cell cycle times) vary, indicating lineage constraints.Second, additional variables must contribute to the diversity of neuron types these progenitors generate.The vulnerability of cortical neurogenesis, and indeed neurogenesis more broadly, to genetic mutation and environmental disruptions that underlie NDDs may reflect, at least in part, the second class of variables, i.e., those that establish progenitor states rather than progenitor classes.These mechanisms may regulate the amount of variability among lineally related progenitors belonging to apparently singular classes, leading to as yet unrecognized neurogenic diversity or flexibility.NDD-associated mutations may modify or destabilize signaling, cell cycle phases, or modes of division for subsets of otherwise similar progenitors and thus alter numbers of neuronal progeny or their capacity for optimal differentiation.
In the decades following the 1978 landmark review by Caviness and Rakic, great insight into the first issue of broad progenitor classes, their proliferative characteristics, and lineage constraints has been generated by identifying a number of selective markers for neural progenitors in distinct brain regions [95,[153][154][155][156], analyzing flexibility as well as limitations of precursor capacities and progeny fates using real-time imaging in vitro [72,157,158] and by studying single gene mutations that compromise CNS or PNS progenitors [2,[159][160][161][162][163][164].Nevertheless, it remains difficult to map this still fairly limited number of molecularly distinct neural progenitor types onto the diversity of postmitotic neuronal progeny in any CNS or PNS region.The extensive literature on Notch-related signaling [55] as well as signaling via Wnt-β catenin [165] and other pathways [64,65,166] in neural progenitors suggests that these signals contribute to variation that modifies the states of neural progenitors without changing their membership in broadly defined progenitor classes.Small variations in local soluble or secreted signals may also add to this state-dependent variability (Fig. 7).For example, the cilium of one cortical precursor might access sufficient concentration of Sonic hedgehog [65,166] to elicit signal transduction and a subsequent transcriptional response, while that of its near neighbors may not.Such local divergence could account for significant additional variation among otherwise molecularly similar progenitors and their progeny; its disruption could have substantial consequences for neurogenesis and neural circuit differentiation.
Recent analysis of progenitor subpopulations based upon single-cell transcriptomes [167][168][169][170][171] from the developing cortex suggests substantial quantitative heterogeneity [168,169,[172][173][174].In comparisons of sorted cortical progenitor types, single molecular marker-defined populations have dynamic, quantitatively diverse transcriptional signatures [170].Regulation via miRNAs [175,176] and other RNAmediated processes [177] may contribute to progenitor diversity.Dynamic changes in chromatin conformation could further diversify neural progenitor states [178][179][180][181] and thus increase flexibility of neuronal differentiation.Some chromatin modifications may be transient, depending upon local interactions, cell cycle phases, and the abundance and activity in individual progenitors of proteins that regulate DNA methylation [182,183] or histone modifications [171,180].Indeed, variation in the sequence or quantity of chromatin modifications among different progenitors with the same broad molecular identity may provide flexibility for increasing the possible range of progenitor states that bias postmitotic neuroblasts toward a broader spectrum of differentiated neuron types.The durability of DNA methylation or histone modification and how these chromatin marks are transferred to the progeny of individual progenitors is likely key for relating the proliferative history to the differentiation capacity of any neuron.The sequential accumulation or removal of direct DNA methylation or specific histone modifications may "keep time" for individual neural progenitors and provide a history of signaling, division, and transcriptional states as a foundation for a distinct program of differentiation for postmitotic progeny.For example, divergent DNA methylation in a subset of apical or basal cortical progenitors could bias differentiation of apparently similar progeny to PN subclasses that share a general laminar destination and similar time of final division but have distinct axonal targets, synaptic inputs, and functional signatures.This epigenetic adaptability, however, may come with a price: dynamic and flexible epigenetic progenitor proliferation could easily be disrupted by multiple genetic or environmental insults associated with NDD pathology.Increased or decreased transcriptional variability among neural progenitors due to NDD-related pathogenic mechanisms that target chromatin modification could substantially alter neurogenesis and subsequent neuronal differentiation (Fig. 7, right).

Conclusion
The two examples of 22q11 deletion-mediated disruption of neural progenitor proliferation and genesis of related neuron classes in 22q11DS mouse models provide a starting point to consider how molecularly identified precursor types that get "in line" can be further diversified to yield highly flexible progenitors, each in a distinct "state" based upon quantitative differences in signaling and epigenetic modifications (Fig. 7).During cortical neurogenesis, 22q11DS compromises basal progenitors as a broad class; however, it is not clear whether changes are targeted to a molecularly distinct basal progenitor subset.These changes may be random, due to 22q11 gene dosage-dependent modulation of signaling, modes of division, epigenetic modification, and transcriptional signature of individual progenitors.The robust, apparently mosaic, expression of multiple 22q11 genes in cortical and CNgV progenitors may further vary due to stochastic allelic regulation [184].Thus, even in neighboring cells, expression levels of heterozygously deleted 22q11 genes may rise below or above the presumed 50%, creating a matrix of progenitors with effectively null to wild-type levels or beyond.These variable 22q11 gene dosages among single cells may initiate additional pathogenesis among otherwise apparently equivalent progenitors.Such 22q11 gene-dependent "altered states" may diminish proliferation for some progenitors, alter modes of division for others, and modify the differentiation capacity of subsets of layer 2/3 PNs or CNgV sensory neurons.Downstream of 22q11 gene dosage, local signaling including that leading to epigenetic modifications may vary beyond adaptive limits, changing a spectrum of optimal progenitor states to a mosaic of suboptimal states that compromise circuit development, organization, and behavior in 22q11DS [75,150,[185][186][187].

Fig. 1 .
Fig. 1.Neural progenitor specification and its relationship with CNS and PNS neuronal diversity.a Neuronal diversity may depend upon progenitor identities (distinct colors in dividing cells) that determine differentiation capacity of postmitotic progeny (related colors, top).Extrinsic signals may establish or reinforce discrete progenitor identities (multicolored arrows) during lineage progression.b Neuronal diversity may be independent of lineage, relying primarily upon combinations of extrinsic signals (multicolor arrows) that act on postmitotic neuroblasts to establish distinct identities.c Current views of lineage progression in the cerebral cortex.Radial glia (apical progenitors [AP]) in the ventricular zone ([VZ], adjacent to the cerebral ventricles [V]) primarily generate projection neurons (PNs) in layers 5/6 early in cortical histogenesis.They also act as guides for postmitotic neuroblasts (NBs) as they migrate radially to their laminar positions.At later stages, radial glia generate basal progenitors (BPs) which diversify in the subventricular zone (SVZ) to include basal radial glia (blue cell with process).BPs generate primarily layer 2/3 PNs.d Trigeminal ganglion progenitor diversity and presumed lineages.Two distinct neural crest progenitor types: one from Wnt1-expressing neuroepithelial neural tube progenitors (green) and another from non-Wnt1 neural tube progenitors (blue), migrate and coalesce with trigeminal cranial placode-derived progenitors (red).These populations, with distinct derivations, give rise, respectively, to mechanosensory (red) and nociceptive neurons (green and blue) axons extending into the distal and proximal branches of the trigeminal nerve (c after[60,61]; d based upon data from[16,62]).

Fig. 2 .
Fig. 2. Multiple contiguous mouse orthologues on mouse chromosome 16 of genes that define the region of human chromosome 22 deleted in 22q11.2deletion syndrome are selectively expressed in the VZ/SVZ and trigeminal ganglion during neurogenesis.Top: a schematic of the region of mouse chromosome 16 where 28 colinear orthologues of the genes from the minimal critical deleted region associated with 22q11DS are located.Gray bars indicate the loci of protein-coding genes.Several 22q11 genes are expressed selectively in the developing cortical mantle (top, each panel) as well as the trigeminal ganglion (CNgV; bottom, each panel) during cortical histogenesis or gangliogenesis.Lines extend from protein-coding loci on the chromosomal map to identify images of cortical and CNgV expression patterns of the encoded transcript.Expression of a subset of 22q11 genes (Dgcr8, Ufd1, Ranbp1, Comt, Cldn5, Trmt2a, Txnrd2, Cldn5, Cdc45) is selectively enhanced in the VZ/SVZ (inset, top right, each panel).The color code (bottom) indicates the diverse functional classes to which each of these 22q11 genes localized to the VZ/SVZ and CNgV belong (top, after[75]; images from Genepaint; functional category key from[124]).

Fig. 3 .
Fig. 3. Heterozygous deletion at human chromosome 22q11 and parallel heterozygous deletion of contiguous murine orthologues on mouse chromosome 16 results in behavioral disruption and cellular pathology associated with neurodevelopmental disorders (NDDs).Top left: the location of the heterozygous deletion proximal to the centromere on the q arm of chromosome 22 that is associated with 22q11DS.Top right: the phenotypic spectrum of 22q11DS includes substantially elevated frequency of behavioral disorders that parallel clinically defined NDDs including ADHD, autism spectrum disorders (ASD), schizophrenia (Scz), and anxiety disorder.Functional imaging studies of individuals with 22q11DS indicate cortical circuit dysfunction, and structural imaging as well as a limited postmortem analyses have found likely foci of cellular brain pathology.In addition to clinically diagnosed NDDs associated

Fig. 4 .
Fig. 4. CNS and PNS neural progenitors are compromised by heterozygous deletion of murine orthologues of human 22q11 genes.a Selective disruption of basal progenitor proliferation during cortical histogenesis by heterozygous deletion of murine 22q11 orthologues.Top right: in E13.5 LgDel fetuses, as the initial population of basal progenitors is being generated, there is a diminished frequency of Tbr2+ basal progenitors that can also be labeled acutely with BrdU, indicating active proliferation.Bottom right: in neonatal LgDel mice (postnatal day P8), the frequency of layer 2/3 PNs, labeled selectively for Cux1 as well as the general neuronal marker NeuN (RbFox3), declines significantly; however, the frequency of NeuN+ or Ctip2+ neurons (a selective marker for layer 5/6 PNs) in layer 5/6 is not substantially changed.b Disrupted neurogenesis and progenitor proliferation during trigeminal ganglion (CNgV) differentiation due to heterozygous deletion of murine 22q11 orthologues.Top panels: there is an apparent increase in the frequency of newly generated neurons, labeled with NeuN, in the E10.5 LgDel CNgV.Middle panels: increased frequency of early generated CNgV neurons in LgDel apparently reflects increased symmetric neuronneuron as well as increased asymmetric precursor-neuron divisions of neural crest-associated (Sox2+) CNgV neural progenitors.These data were generated using a pair-cell assay in which CNgV cells are dissociated, plated in microwells at very low density, and allowed to divide for 21 h.Isolated pairs of cells are presumed to derive from individual progenitors, and their identities are then probed with specific progenitor markers (in this case Sox2) as well as neuronal markers (in this case, βIII-tubulin).Bottom panels: the apparent frequency of subclasses of TrpV1+ nociceptive sensory neurons, particularly those labeled for TrpV1 and via recombination driven by Wnt1 to mark a subpopulation of neural crest-derived progenitors at earlier stages, increases in P8 LgDel CNgV.This increase is paralleled by increased abundance of TrpV1 mRNA in LgDel versus WT, measured by qPCR in samples of dissected P8 CNgV (a adapted from[3]; b from[16]).

Fig. 7 .
Fig. 7. State-dependent regulation and diversification of neural progenitor classes whose broad identities are defined by molecular markers.These progenitors, of which two examples are shown here in detail, retain the unique expression of their diagnostic molecular marker, often reflecting derivation from another broad class of progenitors.Their transcriptional identities (indicated here by the colored bars on the histograms within the nuclei of the two cells) are at one level of analysisabsolute presence of multiple transcriptsequivalent.Their states, however, are divergent based upon variation of levels of expression of the set of transcripts shared by the two otherwise indistinguishable progenitors.These states, indicated here based upon levels of transcription of identified genes, are likely established by diffusible extrinsic signals as well as signaling that relies upon cell-cell contact.The targets of these signals for modifying progenitor state likely include activation or repression of signaling pathways that influence transcription as well as chromatin regulation via direct DNA methylation or changes in histone modifications.NDD pathogenic processes may target these statemodulation mechanisms without changing broad identities of molecular marker-defined progenitor classes.The consequences of these "altered states" would be recognized as increases or decreases in the transcription variability due to quantitative alteration of signaling pathways and their downstream targets including posttranslational modification of transcriptional regulators as well as chromatin-modifying enzymes.