Dopaminergic dysfunction in neurodevelopmental disorders: recent advances and synergistic technologies to aid basic research

Abstract

Neurodevelopmental disorders (NDDs) represent a diverse group of syndromes characterized by abnormal development of the central nervous system and whose symptomatology includes cognitive, emotional, sensory, and motor impairments. The identification of causative genetic defects has allowed for creation of transgenic NDD mouse models that have revealed pathophysiological mechanisms of disease phenotypes in a neural circuit- and cell type-specific manner. Mouse models of several syndromes, including Rett syndrome, Fragile X syndrome, Angelman syndrome, Neurofibromatosis type 1, etc., exhibit abnormalities in the structure and function of dopaminergic circuitry, which regulates motivation, motor behavior, sociability, attention, and executive function. Recent advances in technologies for functional circuit mapping, including tissue clearing, viral vector-based tracing methods, and optical readouts of neural activity, have refined our knowledge of dopaminergic circuits in unperturbed states, yet these tools have not been widely applied to NDD research. Here, we will review recent findings exploring dopaminergic function in NDD models and discuss the promise of new tools to probe NDD pathophysiology in these circuits.

Graphical abstract

Introduction

In the last decade, the widespread adoption of technologies for functional circuit mapping in animal models has greatly enhanced our ability to understand the input-output relationships between populations of neurons and determine their function in vivo. These include techniques for the visualization, reconstruction, and analysis of intact circuits across micro- and macroscales. Examples include serial section electron microscopy [1,2], the Brainbow toolkit [3,4] and intersectional labelling strategies [5,6], improved neuroinformatic tools for neurite tracing [7], tissue clearing [8,9], light sheet microscopy [10,11], and serial two-photon tomography [12,13]. Additionally, optogenetic [14] and chemogenetic [15] actuators, genetically encoded indicators of neuronal activity [16,17], and advanced in vivo imaging modalities [18–23] have allowed for the functional deconstruction of genetically defined circuits in order to probe their contributions to complex behaviors. The development of viral vectors that can deliver transgenes in a pathway- and cell type-specific manner [24–28] or broadly transduce neurons across the CNS [29] have greatly facilitated efforts to anatomically and functionally characterize complex neurobiological systems in both basal and disease states.

New tools for ‘connectomic’ or circuit-centered research that can survey large scale functional connectivity patterns are particularly well suited to the study of neurodevelopmental disorders (NDDs), such as autism spectrum disorder (ASD), where diverse genetic and environmental insults during neurodevelopment can vastly perturb circuit architecture and physiology across brain areas [30,31]. While the neural substrates of ASD symptomatology are multifaceted, mesencephalic dopamine systems, consisting of A8 retrorubral, A9 nigrostriatal, and A10 mesocorticolimbic projections [32], represent circuits of interest given their potential contribution to several common ASD symptoms, including perseverant interests, stereotyped movements, impaired attention and executive function, and difficulty with social interactions [33]. Several recent studies implicate these circuits in behavioral phenotypes observed in rodent NDD disease models, including Angelman syndrome (AS), Rett syndrome (RS), fragile X syndrome (FXS), neurofibromatosis type 1 (NF1), etc. (Table 1), yet widespread adoption of new tools for functional circuit mapping has yet to occur in these models. In this review, we will highlight common patterns of cellular and circuit level phenotypic variation across NDD mouse models and discuss the promise of recent neurotechnological advances such as whole brain tissue clearing and gene delivery by systemic viral vectors to further elucidate NDD pathophysiology in dopaminergic circuits.

Table 1.

Diverse behavioral, synaptic, and cellular phenotypes are observed in nigrostriatal and mesocorticolimbic pathways in mouse models of neurodevelopmental syndromes and ASD candidate genes.

Neurodevelopmental Syndromes
Syndrome	Mouse Model	Major Findings	Citation
15q11–13 Duplication Syndrome	Ube3a-2×	Triplication of Ube3a synergizes with seizures to reduce expression of the glutamatergic synapse organizer Cbln1, impairs glutamatergic transmission in VTA neurons, and leads to loss of sociability.	[99]
16p11.2 Deletion Syndrome	16p11.2+/−	Mice carrying a homologous chromosomal deletion to 16p11.2 (7F3) exhibit abnormal synaptic signaling and increased numbers of dopamine D2 receptor (D2R)-expressing medium spiny neurons (MSNs) in the striatum, fewer D1 receptor (D1R)-expressing neurons in the cortex, locomotor hyperactivity, and deficits in motor control.	[157]
Angelman Syndrome	Ube3am−/p+	Ube3am−/p+ mice display enhanced electrically evoked dopamine release in the NAc and reward seeking but decreased sensitivity to drugs that increase dopamine overflow.	[100]
	TH-Cre∷Ube3am−/p+, TH-Cre∷Ube3aFLOX/p+	Loss of maternal Ube3a in tyrosine hydroxylase (TH)-expressing neurons enhanced optical self-stimulation via increased GABA release from dopaminergic terminals in the NAc.	[101]
Fragile X Syndrome	Fmr1−/y	Fmr1−/y mice are more sensitive to the rewarding effects but less sensitive to the motor effects of cocaine compared to wildtype; the number of TH-expressing neurons in reduced in the SNc but not VTA of these mice.	[158]
	Fmr1−/y	Locomotor sensitization, conditioned place preference, and synaptic changes in the NAc following repeated cocaine is reduced in Fmr1−/y mice.	[159]
Neurofibromatosis Type 1	Nf1+/−:Nf1FLOX∷GFAP-Cre:Nf1FLOX/FLOX	Mice with one non-functional Nf1 allele in all somatic cells and complete Nf1 knockout in glial fibrillary acid protein (GFAP)-expressing cells display reduced striatal dopamine and TH expression in vivo and reduced dopaminergic neurite outgrowth in vitro.	[47]
	Nf1+/−, TH-Cre∷Nf1FLOX/FLOX, GFAP-Cre∷GFAPFLOX/FLOX	Knockout of Nf1 in TH or GFAP-expressing cells is associated with reduced dopamine content in the hippocampus and deficits in spatial working memory.	[46]
Rett Syndrome	Mecp2+/−, Mecp2−/y	SNc neurons exhibit decreased somal size, dendrite count, and striatal dopamine release in Mecp2+/− mice and symptomatic Mecp2−/y males.	[52]
	Mecp2+/−	Mecp2+/− display aberrant motor coordination and motor skill learning secondary to reduced striatal dopamine content, down-regulation of tyrosine hydroxylase expression, and dopamine D2 receptor (D2R) up-regulation.	[102]
	Mecp2S421A	Loss of MeCP2 phosphorylation at position 421 results in accelerated amphetamine sensitization and changes in MSN excitability in the NAc.	[160]
	Dlx5/6-Cre∷DMecp2FLOX/y	Conditional knockout of Mecp2 in the striatum phenocopies Mecp2+/− mice in dopamine deregulation and motor dysfunction.	[161]
Non-Syndromic ASD Genes
Gene (Product)	Mouse Model	Major Findings	Citation
Cntnap4 (CNTNAP4)	Cntnap4−/−	Loss of Cntnap2 results in enhanced dopamine release in the NAc and dorsal striatum through a presynaptic mechanism and results in excessive grooming.	[53]
Nlgn1 (Neuroligin-1)	Nlgn1−/−	Nlgn1−/− display reduced GluN2A-containing NMDA receptor currents and glutmatergic inputs in dopamine D1 receptor (D1R) and D2R-expressing striatal medium spiny neurons (MSNs), respectively.	[162]
Nlgn2 (Neuroligin-2)	Nlgn2 miR knockdown	Striatal knockdown of Nlgn2 results in downregulation of dopaminergic synapses and upregulation of GABAergic synapses.	[163]
Nlgn3 (Neuroligin-3)	Nlgn3−/−, Nlgn3R451C	Both Nlgn3 knockout mice and mice modeling the R451C polymorphism demonstrate enhanced repetitive motor routines by impairing inhibitory transmission onto D1R-expressing MSNs in the NAc.	[164]
Shank3 (SHANK3) (Note: loss of SHANK3 is also seen in Phelan McDermid Syndrome)	Shank3 shRNA knockdown	Shank3 knockdown via short hairpin RNA (shRNA) in the VTA impairs excitatory synapse maturation, reduces dopaminergic neuron excitability via increased inhibitory tone, reduces social preference, and can be rescued an mGluR1 agonist or optogenetic stimulation of dopaminergic neurons	[98]
	Shank3b−/−	Loss of Shank3b alters the development of excitatory inputs to medium spiny neurons of the dorsomedial striatum, which can be rescued by chemogenetic inhibition of corticostriatal inputs.	[165]
	Shank3bfx/fx	Loss of Shank3b in a conditional knock-in model results in abnormal motor, social and exploratory behaviors; repetitive grooming; and synaptic changes in the striatum. These deficits are rescued with germline re-activation of Shank3 expression.	[166]

Elucidating abnormal patterns of dopaminergic connectivity in NDD models

Dopaminergic projection neurons are a heterogenous population whose function, activity, neurotransmitter content, and pattern of connectivity varies with brain region and connection target [34–36]. For example, efferents arising from the midbrain ventral tegmental area (VTA) project throughout the extended amygdala [including the nucleus accumbens (NAc)], hippocampus, and prefrontal cortex (the mesocorticolimbic pathway) and have been widely studied for their role in cognition, reinforcement, and motivation [37,38], while dopaminergic populations in the substantia nigra pars compacta (SNc) project primarily to the dorsal striatum (nigrostriatal pathway) and are critical for the selection and execution of motor programs and habitual behavior [39,40]. Other populations outside the mesencephalon include those in the dorsal raphe nucleus (DRN)/ventral periaqueductal gray area (vPAG) that affect social behavior, nociception, and arousal [41–43] and tuberoinfundibular projections from the hypothalamic arcuate nucleus to the median eminence that regulate prolactin release [44].

Mesencephalic dopaminergic neurons in mice arise from a pool of progenitors in the midbrain floor plate under the control of numerous signaling molecules, including sonic hedgehog, WNT1, engrailed 1 and 2, fibroblast growth factor-8, etc., undergo radial migration to their final positions in either the VTA or SNc by embryonic day 13.5, and exhibit extensive axonal outgrowth along the anteroposterior and dorsoventral axes with synaptogenesis in downstream targets continuing into postnatal development [45]. Consequently loss of NDD-associated genes, such as EN2, MECP2, CNTNAP2, and NF1, produce hypo- or hyperdopaminergic behavioral phenotypes, such as abnormal motor, cognitive, or social behavior, in mouse models by perturbing neuronal maturation, migration, or neurite outgrowth [46–53]. Efforts to understand these phenotypes would benefit from a comprehensive ultrastructural understanding of how specific NDD-associated genetic changes alter dopaminergic circuit architecture and function and inform new therapeutic strategies, such as whole brain gene therapy or genome editing, to help ameliorate NDD symptomatology.

Several recent viral vector-based labeling methods are likely to greatly enhance our understanding of the input-output relationship between dopaminergic efferent and afferent connections in NDD models (Table 2A). This toolkit includes a new adenoassociated viral (AAV) vector for retrograde labeling (AAV2-retro) [26] (in addition to existing retrograde labeling vectors [24,25,27]), intersectional strategies to target individual neuronal projections and their inputs (INTERSECT [5], TRIO [6]), mGRASP for fluorescent labeling of connections between synaptic partners [54,55], and a single cell projection mapping via RNA barcoding (MAP-seq [56]). The recently developed brain-penetrant AAV PHP.eB can efficiently deliver viral transgenes to the CNS after peripheral administration (Figure 1A–B) [29,57], including Brainbow reagents [58] for multicolor labeling via stochastic expression of fluorescent proteins (XFPs) [57] and genetically encoded calcium indicators (GECIs; [59,60]). This tool should also prove useful for non-invasive delivery of optogenetic [14] or chemogenetic [15] tools, AAV-optimized CRISPR-Cas9 effectors for genome editing (e.g. [61–63]), and therapeutic transgenes across large brain volumes. Additionally, PHP.eB can deliver the cargo of interest co-administered with a titratable inducer vector for controlled sparseness while maintaining high viral transgene copy number [57], which is beneficial for effective neurite tracing with methods such as mGRASP [55] or Brainbow [58] (Figure 1C–D). This method is also likely to benefit sensors that need sparse expression to reduce background fluorescence [64–66].

Table 2.

Selected recent advances in neurotechnologies for structural analysis of circuit architecture in rodent models.

A. Viral Vectors and Vector-Based Labeling Methods
Tool	Summary	Citation
rAAV2-retro	A recombinant AAV2 variant for retrograde targeting of projection neurons that can be used for both functional and tracing studies. Efficient retrograde transduction was observed in many cortical and subcortical regions.	[26]
PHP.eB and PHP.S	AAV9-based vectors for efficient CNS (PHP.eB) or PNS (PHP.S) transduction after peripheral (intravenous or retro-orbital) virus administration. Can be used with a titratable inducer vector for controlled sparseness of multicolor labels that preserves color diversity; inducers involve use of tetracycline-controlled transactivator (tTA) or Cre-dependent Flp-based constructs.	[29,57]
TRIO	Tracing the Relationship Between Input and Output; A combinatorial two-vector system that maps the input-output relationship of a population of neurons. In this method, canine adenovirus-2 (CAV-2) [25] is used to deliver a Flp recombinase transgene to axons in a specific projection terminal field for retrograde transduction of the cell bodies; Flp-dependent RVdG [24] component vectors are later delivered to the cell bodies for monosynaptic retrograde tracing of inputs. Cre-dependent Flp can be used for cell type-specific targeting using a Cre driver line (cTRIO). Cannot be used for functional studies due to lethality of RVdG.	[6]
INTERSECT	INTronic Recombinase Sites Enabling Combinatory Targeting; A two-component system that allows for functional projection targeting using Flp- and Cre-dependent viral vectors via axonal targeting in a downstream region using replication incompetent herpes simplex virus (HSV) [27] carrying a Cre-dependent Flp recombinase transgene. Either Cre- and Flp-ON and -OFF strategies can be used.	[5]
MAP-Seq	Multiplexed Analysis of Projections by Sequencing; allows for parallel mapping of single neuron axonal arbors via recovery of RNA barcodes in from terminal fields after delivery of AAV viral barcode libraries to the cell body. Does not distinguish fibers of passage, so downstream regions must be chosen carefully for RNA recovery and sequencing.	[56]
mGRASP	Mammalian GFP Reconstitution Across Synaptic Partners; A method for fluorescently labeling synaptic connections that employs AAV-mediated delivery of synapse-targeted split GFP fragments in genetically defined pre- and post-synaptic neuronal partners. Cre-ON and Cre-OFF strategies can be used for studying microcircuits.	[54,55]
B. Large Volume Imaging Modalities
Light Sheet Microscopy (LSM)	Originally developed over 100 years ago, LSM illuminates the sample with a thin sheet of light and detects the emitted fluorescent signal with an orthogonally arranged detection objective. Variants include CLARITY optimized LSM (COLM) for use in cleared tissue [76], SPED (Spherical Aberration-assisted Extended Depth of Field) LSM that improves scan speed via extended depth of field [68], and an adaptive LSM that integrates multiple fields of view with 10 degrees of freedom that are autonomously adjusted in real time for improved spatial resolution and image quality [11].	[10,11,68]
High-Speed Volumetric STP Tomography	High-speed Volumetric Serial Two-Photon Tomography; A high speed imaging platform based on Serial Two-Photon Tomography (STP) [13] that creates 3D reconstructions of neuronal axonal arbors via the integration of fast volumetric 2-photon microscopy and a vibrating microtome to image bright, sparsely labelled neurons in cleared samples embedded in gelatin. Includes computational tools for the registration and visualization of large (up to 100 TB) data sets, although labeling must be sufficiently sparse to prevent neurite reconstruction errors when axons from different neurons are closely positioned.	[12]
C. Tissue Clearing Methods
CLARITY	Hydrogel-based clearing method that utilizes 4% SDS for lipid removal after sample has been embedded in an acrylamide-bisacrylamide gel and cross-linked with formaldehyde. Clearing can be accelerated with electrophoresis at the expense of tissue integrity. Compatible with immunolabeling and endogenous fluorescence. The EDC-CLARITY variant is compatible with HCR (hybridization chain reaction) probes for bulk RNA labeling.	[75,76,86]
PACT PACT-deCAL	Passive CLARITY Technique; A passive CLARITY-based clearing method for rapid clearing of thick sections that employs 8% SDS as the detergent. Compatible with immunolabeling, endogenous fluorescence, smFISH (single molecule), and smHCR probes for single and bulk RNA labeling. Produces reversible expansion of tissue and can be used with RIMS (Reflective Index Matching Solution), a non-viscous mounting medium that decreases the refractive index of the sample for better optical access. PACT-deCAL uses EDTA/EGTA to decalcify samples for bone clearing.	[71,78,85,167]
PARS	Perfusion Assisted Agent Release In Situ; An active CLARITY-based clearing method that involves intracranial and/or transcardial perfusion of reagents for whole body clearing.	[71,78]
SWITCH	System-Wide Control of Interaction Time and Kinetics of Chemicals; A fixation and clearing method that exploits the pH dependence of glutaraldehyde-tissue gel formation for uniform fixation prior to delipidation with SDS. This method provides added tissue integrity for multiplexed immunolabeling. Not compatible with smFISH or smHCR probes.	[84]
uDISCO	A whole-body clearing method based on 3DISCO; it utilizes dehydration with tert-butanol followed by delipidation with diphenyl ether for fast sample clearing. Maintenance of endogenous fluorescence is improved relative to 3DISCO and other solvent based methods, whereby fluorescence deteriorates within several days after clearing. Shrinks tissues by approximately 40% for faster LSM imaging.	[79]
Sca/eS	An improved version of Sca/eA2 [168] that achieves tissue transparency via partial delipidation and hyperhydration via urea, sorbitol, glycerol, and Triton X-100. Preserves endogenous fluorescence and limits expansion better than most other methods, although large0 sample clearing can take several weeks. A simplified protocol Sca/eSQ can be used in thick (<500 micron) sections.	[72]
CUBIC	Clear, Unobstructed, Brain/Body Imaging Cocktails and Computational Analysis; A clearing method based on Sca/eA2 that uses urea, aminoalcohols, TRITON X-100, and high sucrose concentrations. Maintains endogenous fluorescence, can be perfused for whole body clearing, produces reversible tissue expansion, and exhibits superior decolorization (i.e. loss of the heme chromophore) relative to other techniques.	[73,74]
Tissue Expansion Methods for High Resolution Microscopy
ExM	Expansion Microscopy; A tissue expansion technology whereby the fixed and permeabilized sample is embedded in a superabsorbent hydrogel containing sodium acrylate and acrylamide, cross-linked with N-N′-methylenebisacrylamide, and digested with a protease to produce a 4.5-fold sample expansion. Newer variants display improved protein retention (proExM) and are compatible with immunolabeling, smFISH, and smHCR (ExFISH).	[80–82]
ePACT	Expansion PACT; Variant of the PACT tissue clearing method that utilizes a superabsorbent hydrogel and enzymatic digestion to increase sample size up to 5-fold for high resolution imaging with preserved endogenous fluorescence.	[78]
MAP	Magnified Analysis of Proteome; a hydrogel-based clearing method that expands tissue without the use of enzymatic digestion via treatment with high acrylamide concentrations (up to 20%) prior to SDS treatment. Compatible with immunolabeling but not RNA detection.	[83]

Figure 1. Broadly transducing AAVs permit brainwide transgene expression and facilitate neurite tracing.

A. Workflow for multicolor labelling with PHP.eB and neurite tracing. Viral particles carrying red, green, or blue XFP transgenes with or without a titratable inducer are systemically introduced via retro-orbital injection (1). Following transduction (4–8 weeks), brains are fixed and cleared (2). Tissue samples can then be imaged with light sheet or confocal microscopy (3) prior to neurite tracing (4). B. Brainwide transduction of neurons (green) or astrocytes (red) using cell type-specific promoters (hSyn1 and GFAP, respectively) and gene regulatory elements following retro-orbital injection of PHP.eB (1×10¹² viral genomes/mouse). C–D. Sparsely labelled striatal neurons were successfully traced after transduction by PHP.eB multicolor XFPs.

The utility of viral vector-based mapping tools has been improved by microscopic techniques for rapid imaging of large samples, such as light sheet microscopy [11,67,68] or high-speed volumetric serial two-photon (STP) tomography [12] (Table 2B), and tissue clearing protocols that render biological samples optically transparent for analysis of intact circuits in whole brains or thick slices (Figure 2) [8,9]. Several tissue clearing strategies have been recently described or refined (Table 2C); these include immersion clearing with high refractive index (RI) solutions (SeeDB [69], FRUIT [70], RIMS [71]), clearing via hyperhydration (Sca/eS [72], CUBIC [73,74]), hydrogel embedding followed by detergent delipidation (CLARITY [75,76], PARS [77,78], PACT [77,78]), and solvent-based clearing methods (uDISCO [79]). Clearing methods that build upon water-absorbent CLARITY hydrogels to create hyperabsorbant hydrogels have also been implemented to facilitate high resolution imaging of small structures, such as individual dendrites or neurites (ExM [80–82], ePACT [78], and MAP [83]). Hydrogel-based methods preserve endogenous fluorescence while maintaining compatibility with tools for proteomic analysis [76,78,82–84], RNA profiling (smFISH or smHCR probes [71,81,85,86]), and time-stamped fluorescent readouts of neuronal activity (e.g. ArcTRAP [87,88]).

Figure 2. Visualization of intact circuits using hydrogel-based clearing methods.

A. A mouse brain before and after clearing with PARS and long-term storage in RIMS (adapted from [71]). The sample demonstrates moderate tissue expansion due to acrylamide embedding. B. Confocal image of dopaminergic neurons in the SNc (red) and cholinergic afferents from the pedunculopontine tegmental nucleus (green) visualized in a 1mm-thick PACT-cleared section. C–D. Whole brain imaging of fluorescently-labelled cells in Thy1-eYFP mice using light sheet microscopy.

Several recent studies have successfully integrated these technologies to probe the structure of dopaminergic and related circuits in healthy mice. For example, retrograde labeling, tissue clearing, and LSM have been used to parse SNc subcircuit connectivity and function [89], identify an anatomically distinct projection to the posterior striatum [90] that preferentially encodes novel cue information rather than reward prediction errors [91], and refine our knowledge of cholinergic inputs to the SNc and VTA [92]. An input-output analysis of VTA connections using TRIO uncovered a novel projection from the anterior cingulate cortex to the lateral NAc that produces behavioral reinforcement using an optogenetic intracranial self-stimulation paradigm [93].

Bridging the gap between synaptic function and neural circuit dynamics in NDD models

One common feature amongst NDDs is that the causative genes (e.g. FMR1 in FXS, UBE3A in AS, MECP2 in RS, NF1 in NF1, EN1 and EN2, SHANK genes, etc.) affect synapse formation, maintenance, and plasticity in rodent models [94,95]. As such, there have been considerable efforts to characterize synapse function in dopaminergic circuits in these mice. For example, reduction in SHANK-3, an excitatory synapse scaffolding protein whose loss of function is associated with Phelan McDermid Syndrome (also called 22q13 deletion syndrome; see [96] for a review) and some non-syndromic ASD cases [97], via delivery of a short hairpin RNA (shRNA) into the VTA impairs maturation of excitatory synapses and reduces dopaminergic neuron excitability and social preference via increased inhibitory tone [98]. Mice modeling 15q11–13 Duplication Syndrome (where Ube3A protein levels are increased three-fold) exhibit a loss of sociability due to downregulation of the glutamatergic synapse organizer CBLN1 in the VTA [99]. Altered neurotransmitter content or release from VTA or SNc neurons in downstream targets has been reported in mouse models of AS [100,101], NF1 [46,47], and RS [52,102] and in Cntnap2 knockout mice [53]. While there is a large body of research delineating the role of synaptic or microcircuit deficits in NDD models [103–106], less is known about how those changes alter population dynamics or neuron ensemble activity to produce behavioral phenotypes; improvements in optical tools to monitor neural activity across multiple spatial scales [59,107–109] should help bridge this divide.

Understanding how networks of interconnected neurons encode and translate relevant environmental stimuli into a motivated behavior requires a high throughput readout of neuron firing with single-cell resolution. Metal electrodes or electrode arrays are a robust tool to measure spiking with high temporal precision and can be coupled with optogenetic tools to manipulate activity or infer cell identity (i.e. opto-tagging, [110]). Opto-tagging has been used to monitor diverse populations across the CNS (e.g. cortical interneurons [111,112], AgRP neurons in the arcuate nucleus [113], dopaminergic neurons in the VTA [114], etc.), yet this technique is limited in the number of neurons it can sample and may not be suitable for all populations due to the challenges in efficiently and accurately opto-tagging (e.g. cortical pyramidal neurons), as well as genetically similar populations that are too sparse or dense to be reliably identified. In contrast, optogenetic stimulation of dopaminergic circuitry during blood oxygen level dependent contrast (BOLD) fMRI imaging can approximate mesocorticolimbic or nigrostriatal network activity in rodents [115–117], yet this technique lacks both cellular resolution and temporally precise neurophysiological readouts.

Alternatively, genetically encoded calcium indicators (GECIs; e.g. the GCaMP6 family of proteins [16]) provide cell type-specific fluorescent readouts of neuron activity during behavior that is stable over months of testing and is scalable [21]. Using two-photon mesoscopes with wide field of view objectives [108] or random access scanning strategies [107] to image through large cranial windows in head-fixed mice, researchers can record the calcium dynamics of hundreds to thousands of neurons at once. Bulk calcium signals can also be measured across superficial cortical areas using a wide field fluorescence macroscope featuring a 12 mm field of view, which has been used to assess global representations of motivated behavior in multiple cell types [59]. While these tools have been optimized for relatively superficial (< 1mm deep) structures, several tools should help extend the depth of non-invasive optical access, such as three-photon microscopy [118], the implementation of axially elongated Bessel foci [119], photoacoustic tomography [120], and guidestar-assisted wavefront engineering techniques to limit optical scattering [121], such as time reversal of ultrasound encoded light (TRUE) [122,123].

Several recent technologies have provided optical access to deep brain areas in behaving mice for activity measurements in bulk or with single-cell resolution. For bulk measurements, fiber photometry [124] and TEMPO [125] allow for quantification of calcium or voltage sensor dynamics, respectively, using implanted optical fibers in order to correlate activity of genetically defined populations with behavioral events. Calcium imaging via implanted gradient index microendoscopes (GRIN lenses) provides single cell resolution at depths >4mm below the skull surface [126]. While two-photon GRIN lens imaging is most commonly performed in head-fixed mice [127], several strategies such as 2-photon fiberscopes [128,129] and miniaturized head-mounted 2-photon microscopes [130,131] have been developed for freely moving behavior. Head-mounted miniaturized epifluorescence microscopes [132] are also available and have been more widely adopted for use in behaving animals. Single cell calcium dynamics have been imaged via GRIN lens in the VTA [133], SNc [134], and interconnected regions, including the dorsal striatum [134], lateral hypothalamus [134,135], medial preoptic area [136], medial prefrontal cortex [137,138], bed nucleus of the stria terminalis [133], hippocampus [139,140], etc. Additionally, chronic imaging windows have permitted monitoring of sparsely labelled SNc axons in the dorsal striatum, which revealed distinct temporal and spatial encoding of reward and motor signals [39]. While several groups employ cortical two-photon calcium imaging in NDD models, including RS [141] and FXS [142] mice, analysis of deeper structures has not been reported to date.

Considerations and future outlook

The identification of causative genetic defects in neurodevelopmental syndromes and subsequent creation of transgenic mouse models has greatly enhanced our understanding of the developmental perturbations that produce synaptic, cellular, and behavioral phenotypes in these mice. While several recent studies examining dopaminergic circuitry have uncovered pathophysiological mechanisms underlying aberrant social interactions, positive reinforcement, stereotyped behavior, etc., few studies have employed new technologies for functional circuit mapping in NDD models. This may be due to several factors; first, given that phenotype expression is dependent on genetic background in many mouse models, such as NF1 [143], it will be important to continue identifying and developing minimal gene regulatory elements (promoters, enhancers, miRNA binding sites) that can be accommodated within well-tolerated viral capsids for cell type-specific targeting without the need to cross mice to Cre or Flp driver lines. Several cell type-specific promoters have been developed to target different cell populations in the CNS, including catecholaminergic (tyrosine hydroxylase promoter), serotonergic (FEV), Purkinje (PCP2) [144], and forebrain GABAergic (mDlx5/6) neurons [145], although they vary in leakiness and promoter size, which can limit packageable transgene size due to AAV carrying capacity of 4.7 kb [146]. Second, many of these techniques require specialized equipment, reagents, or expertise that makes implementation challenging. Several helpful imaging, tissue clearing, and data analysis protocols have recently been published [54,76,78,133,147–149] that can help guide potential users.

To effectively integrate measures of neural activity with comprehensive dopaminergic connectomes in mouse models obtained with such tools for precise structural and functional analysis of intact circuits, several advances will be required. First, we will need better computational methods for automated detection, segmentation, and tracing of individual fluorescently labelled neurons in whole cleared brains. This task is currently labor intensive and works poorly for neurons with complex morphology, such as catecholaminergic neurons with large axonal arbors that traverse several mm of brain tissue. Recent successes in overcoming these challenges include the reconstruction of single projection neurons in the claustrum, which branch extensively throughout the entire cerebrum [150]. Second, we will need improved tools for converting neural activity states into fluorescent labels that can be superimposed upon neuronal reconstructions. Several technologies show promise, such as CaMPARI [151] and iTANGO [152], which provide light timestamped indicators of intracellular calcium or dopaminergic neurotransmission, respectively. Hybridization chain reaction (HCR) probes for single-cell, multiplexed RNA detection have been validated for hydrogel-based clearing methods [86,153] and could allow for medium throughput identification of projection- or activity-dependent changes in gene activity in mutant and wildtype mice. At this time, only PACT/PARS [71], EDC-CLARITY [86], and Ex-FISH [81] have been demonstrated to be compatible with RNA profiling, yet clearing methods are advancing rapidly and will likely be useful for a broader range of applications in the future.

When examining the role of functional circuit mapping technologies in elucidating dopaminergic connectivity in NDD models, one cannot neglect the ontogeny of these circuits. Several methods have been used to clear mouse embryos at various stages of development (reviewed by [154]), yet it is difficult to employ viral vector-based tracing and labeling techniques in the developing mouse. It is thus of great interest to identify AAVs that cross the blood-placenta barrier and selectively target the developing embryonic nervous system. AAV selection platforms, such as CREATE (Cre recombination–based AAV targeted evolution), which has been used to develop vectors that efficiently target the central (PHP.B/PHP.eB) or peripheral (PHP.S) nervous systems [29,57] when given systemically, could yield new vectors for in utero transgene delivery. Additionally, tools for large volume functional imaging of developing organisms, such as a two-beam light sheet microscope with adaptive optics and automated cell tracking [11], have been applied to early embryonic mice [155,156], yet new methods to maintain optical access within the amnion will be necessary to image and track post-implantation fetal cells.

Going forward, we anticipate that continued technological advances can yield progressively more precise and comprehensive functional and connectomic maps of dopaminergic circuitry across development. As these tools become more widely adopted by NDD researchers, we will likely gain newfound understanding of how functional and structural abnormalities synergize to produce behavioral and cognitive phenotypes in mouse models and reveal putative mechanisms of disease symptomatology in human populations. Ultimately these discoveries can inform the creation of behavioral and pharmacological therapies that target circuit- or cell-type specific mechanisms of disease in order to benefit the health of affected children and adults.

Highlights.

• Dopaminergic dysfunction is common in many neurodevelopmental disorders.

• New circuit mapping tools have refined our knowledge of these circuits at baseline.

• When used in animal models, these tools will likely elucidate NDD pathophysiology.