Human Pluripotent Stem Cell-Derived Medium Spiny Neuron-like Cells Exhibit Gene Desensitization

Gene desensitization in response to a repeated stimulus is a complex phenotype important across homeostatic and disease processes, including addiction, learning, and memory. These complex phenotypes are being characterized and connected to important physiologically relevant functions in rodent systems but are difficult to capture in human models where even acute responses to important neurotransmitters are understudied. Here through transcriptomic analysis, we map the dynamic responses of human stem cell-derived medium spiny neuron-like cells (hMSN-like cells) to dopamine. Furthermore, we show that these human neurons can reflect and capture cellular desensitization to chronic versus acute administration of dopamine. These human cells are further able to capture complex receptor crosstalk in response to the pharmacological perturbations of distinct dopamine receptor subtypes. This study demonstrates the potential utility and remaining challenges of using human stem cell-derived neurons to capture and study the complex dynamic mechanisms of the brain.


Introduction
Complex cellular programs (e.g., transcriptional networks, epigenetics) respond to spatiotemporally diverse extracellular perturbations. These responses, including gene desensitization to repeated stimuli, act as homeostatic mechanisms to stress and are important for processes such as addiction [1,2], learning, and memory [3]. These phenotypes have been primarily studied in rodents but sparsely investigated in human models [4][5][6][7]. However, there are likely species-specific differences between humans and other animals and these differences may be relevant for the success of therapeutic strategies [8,9]. Therefore, it is important to assess complex cellular responses in ethical human models.
A salient example of gene desensitization is the complex response of medium spiny neurons, the primary cell type of the striatum [1,2], to the chronic administration of substances of abuse. In particular, cocaine and other drugs of abuse modulate the timing and levels of dopamine that presynaptic dopaminergic neurons [10,11] expose postsynaptic MSNs to, with postsynaptic transcriptomic and functional effects on MSNs [12][13][14][15]. More generally, MSNs serve as the main input and projection cells of the basal ganglia and thus play a role in the generation of behaviors [16]. Thus, in addition to human MSNs (hMSNs) serving as a potential model to study gene desensitization in human cells, they may also yield insights into the mechanisms underlying aspects of human behavior.
Protocols have previously been developed to differentiate human pluripotent stem cells (hPSCs) into neurons expressing stereotypical markers of striatal MSNs [17][18][19][20][21][22]. However, these studies generally aimed to model disorders such as Huntington's Disease and genetic forms of Parkinson's Disease and their relatively static phenotypes. In this work, we evaluate the dynamic transcriptomic responses of hMSN-like cells to a range of acute and (RT) followed by 3 × 5 min 1X PBS washes. Cultures were then permeabilized and blocked in 0.3% Triton X-100 in PBS (PBST) and 5% (w/v) normal donkey serum (Jackson Immunoresearch) in PBS for 30 min at RT. Cultures were incubated with primary antibodies (see Table S1) in PBST at RT for 2 h or at 4 • C in a humidity chamber overnight. Cultures were washed by 3 × 10 min PBST washes and incubated with secondary antibodies in PBST for 2 h at RT, and nuclei were stained with DAPI (Invitrogen, Waltham, MA, USA). Slides were mounted using ProLong Antifade Diamond (Thermo Fisher Scientific, Waltham, MA, USA). Secondary antibodies used were donkey Alexa Fluor 546 and 647 conjugates (Invitrogen, 1:250). Images were taken using a Nikon A1R confocal microscope (Nikon Instruments, Tokyo, Japan) using a 40X oil immersion objective. To 'fill' the cell to assess morphology, hMSN-like cells were transfected with a pmaxGFP plasmid (Lonza, ThermoFisher Scientific, Melbourne, Australia) using Lipofectamine 3000 Transfection reagent (Thermo Fisher) 72 h before fixation following the manufacturer's instructions.

Immunostaining Quantification
To quantify FOSB+ and DARPP32+ cell percentages, immunostained slides were imaged using a 40X oil immersion objective for DAPI, FOSB, MAP2, and DARPP32 simultaneously using the 405, 488, 561, and 640 nm lasers. ND2 stacks were imported into FIJI [23], Z-projected for max intensity, converted to grayscale, split into individual channels and background subtracted for each channel individually using a 50-pixel rolling ball radius. Images were thresholded using the "Moments" algorithm with the "Auto" setting and measured for area fraction. Overlap percentages were obtained by combining thresholded FOSB or DARPP32 channels with either DAPI or MAP2 using the Image Calculator function with the "AND" operation, measuring area fraction, and dividing the overlap area fraction by the DAPI or MAP2 area fraction. For FOSB overlap with DARPP32/MAP2, "AND" combined DARPP32 and MAP2 channels were "AND" combined with the thresholded FOSB channel. Error bars were calculated as standard deviation. For each sample, 2-3 independent replicates (n = 2-3) in 8-well chamber slides were dosed, fixed, and immunostained.

RNA Extraction and QRT-PCR
For RNA extraction, total RNA was extracted as previously described [24]. In brief, hMSN-like cultures were washed 1 time in PBS. Total RNA was isolated using Direct-zol RNA MicroPrep Kit (Zymo Research) according to the manufacturer's protocol. RNA samples were collected in 2 mL RNAse-free tubes and chilled on ice throughout the procedure. Total RNA was then used for either QRT-PCR or RNA-seq. cDNA synthesis was performed using 1000 ng of total RNA and the iScript Reverse Transcription Kit (BIO-RAD) according to the manufacturer's protocol. QRT-PCR reactions were performed using SsoAdvanced™ Universal SYBR ® Green Supermix (BIO-RAD) on a BIORAD 384-well machine (CXF384) with customdesigned primers using Primer3 [25] (see Table S2). Analysis of target gene expression along with the reference genes HPRT1 and GUSB was performed using Excel. Data are presented as expression level (2 −∆∆Ct ) relative to HPRT1 and GUSB and normalized to control dose. For each QRT-PCR sample, 3 independent replicates (n = 3) in individual 24-well plate wells were collected. Two technical replicates per sample were also included with each QRT-PCR experiment. All QRT-PCR analyses were performed on RNA extracted from DIV 45 hMSNcells.

RNA-Sequencing and Analysis
Sequencing and preparation of Illumina libraries from total RNA was performed and then sequenced using Illumina Hi-seq 2 × 150 bp (Azenta, Research Triangle Park, NC, USA) and NovaSeq S1 2 × 100 bp (UNC High Throughput Sequencing Facility) for 20-30 million reads per sample. Raw sequencing data are publicly available on the Gene Expression Omnibus (GEO Accession Number: GSE195492). Raw FASTQ formatted sequence reads were imported into CLC Genomics Workbench (v. 21.0.5 Qiagen, https://digitalinsights.qiagen.com/, accessed on 12 January 2022). Adaptor sequences and bases with low quality were trimmed and reads were mapped to the reference genome (GRCh38.102) using the RNA-seq analysis tool with the default parameters recommended for RNA-seq analysis. Principal component analysis and differential expression analysis were performed using 'PCA for RNA-seq' and 'Create Heat Map for RNA-seq' toolsets. Heatmaps were generated using Euclidean distance with complete linkage. All genes were displayed. Ingenuity Pathway Analysis (v. 22.0, QI-AGEN Inc., https://digitalinsights.qiagen.com/IPA, accessed on 12 January 2022) was used to predict upstream regulators [26]. Functional enrichment analysis was performed using gProfiler to extract KEGG pathways, biological process and molecular function gene ontologies, and transcription factor binding motifs associated with an inputted list of genes [27]. Other plots and analyses were generated using RStudio [28]. Overlaps between gene sets were performed using BioVenn [29] and the Venn Diagram tool from the Bioinformatics & Evolutionary Genomics Department at Ghent University [30]. Log 2 (fold change) heatmaps were generated using pheatmap (v. 1.0.12). Boxplots and Lineplots were plotted using ggplot2 (v. 3.3.5) [31]. Significance was defined as p < 0.05. Statistical analyses for RNA-seq were performed in CLC Genomics Workbench (v. 21.0.5 Qiagen, https://digitalinsights.qiagen.com/, accessed on 12 January 2022).

Statistical Analysis
Statistics were performed for QRT-PCR using the one-way ANOVA with Tukey's post hoc test to test for significance. All analyses were performed using Rstudio. Significance was defined as p < 0.05. Statistical analyses for RNA-seq were performed in CLC Genomics Workbench (v. 21.0.5 Qiagen, https://digitalinsights.qiagen.com/, accessed on 13 January 2022). Significance was selected as log 2 (fold change) ≥ |1| in expression between comparison groups with a threshold false discovery rate (FDR), adjusted p-value < 0.05, and max group mean ≥ 1 for acute and chronic dopamine datasets. Significance for all other RNA-seq datasets was selected as (FDR) adjusted p-value < 0.05. All sequencing data passed default quality filters for the CLC Genomic Workbench version 21.0.5 RNA-Seq pipeline analysis. Significance for Ingenuity Pathway (v. 22.0, QIAGEN Inc., https://digitalinsights.qiagen.com/IPA, accessed on 13 January 2022) and gProfiler [27] analyses were defined as p-value < 0.05 and g:SCS threshold < 0.05, respectively.

hMSN-like Cells Exhibit Dose and Time-Dependent Responses to Dopamine
MSNs are expected to respond to dopamine, a key neurotransmitter in the striatum, although it remains unclear what the dose dependency and kinetics of the responses are in hMSN-like models [20]. Therefore, we first characterized the dose-dependent response of hMSN-like cells to dopamine. It is still unclear what concentrations would best mimic the effective local concentrations of dopamine in the synaptic cleft. Estimates of neurotransmitter concentrations vary widely from 1 µM to 1 mM and are heavily dependent on whether measurements were made directly in the cleft or in the extracellular space. Prior work in ex vivo rodent models suggested 100 nM-100 µM and 30-60 min time points would be reasonable to elicit physiologically relevant responses [35][36][37][38][39][40][41]. For example, FOSB and FOS are immediate early genes (IEG) previously observed to be induced in rodents by a minimum of 100 µM dopamine [40,42]. We, therefore, dosed both H1-and H9-derived DIV45 hMSN-like cells with a range of dopamine concentrations from 1 µM to 10 mM and tracked FOSB and FOS expression by QRT-PCR after 0 to 120 min ( Figure 2A). Gene expression induction was observed at all concentrations and reached maximal expression at a concentration of 1 mM dopamine. We also tracked IEG induction at 0, 30, 60, and 120 min post-addition of 1 mM dopamine and observed that maximal induction was observed between 60 to 120 min ( Figure 2B). The FOSB and FOS measurements indicated that hMSN-like cells respond acutely to dopamine. Additionally, immunostaining revealed FOSB expression in 11.21% of DIV45 H9 DAPI+ and 4.7% of DIV 45 H1 DAPI+ cells after acute dopamine dosage ( Figure S2). Having observed greater responses in H9-derived cells, we decided to perform all remaining experiments using H9 hMSN-like cells. We next assessed the transcriptome-wide responses and performed an RNA-seq 1 h after dosing the DIV45 H9 hMSN-like cells using either 1 µM or 1 mM dopamine ( Figure 2C,D). The overlap in the DEGs was low between the two concentrations ( Figure 2C, Dataset S1); yet, reassuringly, among the DEGs that were common included many genes canonically relevant to substance-use disorders and neuronal stimulation such as EGR1, FOSB, FOS, JUNB, and ARC. The KEGG pathway analysis suggested some general differences between 1 µM and 1 mM concentrations where the 1 mM dose led to DEGs associated more with the signaling pathways (i.e., TGFβ, Wnt, & Hippo Signaling) ( Figure 2D). As expected, the larger 1 mM concentration of dopamine led to a greater response compared to the 1 µM dopamine dose [43][44][45]. Furthermore, the presence of some specific DEGs suggests that the 1 mM concentration may be more informative for studies related to substances of abuse. For example, GADD45B (implicated in memory of cocaine reward) [46], TENT5B (the top DEG observed in dopamine-dosed rodent MSNs) [40], and MMP1 (extracellular matrix and synaptic plasticity) [47] were elevated only in the 1 mM dopamine conditions. However, the expression of individual DEGs should not be overinterpreted as they were not individually confirmed by QRT-PCR. Instead, here we focus on collections or sets of related DEGs through ontologies or key phenotypes that reflect neuronal responses.

Chronic Administration of Dopamine Leads to Desensitization of Genes Implicated in Cocaine and Dopamine Responses
A key physiological response to substances of abuse is desensitization at the cellular level. The chronic dosage or self-administration of substances of abuse in rodents leads to

Chronic Administration of Dopamine Leads to Desensitization of Genes Implicated in Cocaine and Dopamine Responses
A key physiological response to substances of abuse is desensitization at the cellular level. The chronic dosage or self-administration of substances of abuse in rodents leads to a unique molecular phenotype known as gene desensitization [1,14,48], in which differentially expressed genes (DEG) reduce in magnitude of expression when compared to a single acute exposure. To assess if this important response also occurs in hMSN-like cells, we acutely (DIV45) or chronically (DIV50) dosed hMSN-like cells with dopamine, extracted RNA at 1 h or 24 h after the last dose, and performed RNA-seq (Figures 3 and S3). As both the 1 µM and 1 mM concentrations elicited gene expression responses but to distinct degrees ( Figure 2A) and it remains unclear what concentration is truly physiological, we tested both. Importantly, at both concentrations, chronic dosing led to a dramatic reduction in the number of DEGs compared to single acute doses ( Figure 3B). As expected, the number of DEGs decreased when sampling at 24 h versus 1 h after the last dopamine exposure, although a substantial number of DEGs remained (see Dataset S1-S4). To identify whether the same set of genes was responding to both acute and chronic dosing, we compared DEGs across conditions and observed that 50-60% of the chronically dosed hMSN-like cell DEGs were also differentially expressed after acute dosing ( Figure 3C). This was true for both concentrations. However, although the overall number of DEGs decreased, this did not mean that individual genes were necessarily desensitizing. Therefore, we narrowed our analysis to identify the desensitized genes. To do this, we ratioed the gene expression changes of acute 1 h hMSN-like cells to chronic 1 h hMSN-like cells and did so for 1 µM and 1 mM dopamine conditions separately. Desensitized genes were defined as those with a ratio > |1.1| as previously defined [1]. In general, the 1 mM dopamine conditions led to a larger number of desensitized genes compared to the 1 µM conditions ( Figure 3D; see Dataset S5). However, both concentrations shared overlap in important desensitized genes ( Figure 3D). Specifically, genes that were desensitized in both concentrations of dopamine included many immediate early genes (e.g., FOSB, FOS, JUNB) ( Figure 3D). Additional overlap was identified through IPA analysis including genes that were associated with the upstream regulators CREB1 and SRF, transcription factors that are necessary for ∆FOSB induction including in response to cocaine; and NFκB1/RELA, which are involved in regulating the reward properties of drugs of abuse ( Figure 3E) [49,50]. Interestingly, we also noted an increase of FOSB+ cells after chronic dopamine dosage (DIV50) compared to acute dosage (DIV45) for both H1 and H9 hMSN-like cells ( Figure S2). This may be indicative of an increase in the ∆FOSB concentration after chronic dosage. We also performed a regulatory motif analysis of desensitized DEGs present for both concentrations and found that the motifs for SRF and CREB1 were enriched, which was similar to the IPA analysis results. In addition, the motifs for E2F3 binding, a transcription factor that regulates cocaine-induced locomotor and place-conditioning behavior, were also enriched ( Figure 3E) [51]. To identify differences in the nature of the desensitized genes between the 1 mM and 1 µM concentrations, we also analyzed genes that were highly desensitized to a 1 mM dopamine dose which we defined as beyond the maximum level of desensitization observed after a 1 µM dose (Acute 1 h to Chronic 1 h ratio > |2|). Genes dosed with 1 mM dopamine with an Acute 1 h to Chronic 1 h ratio |>| 2 were enriched for biological process GO related to ERK1/2 cascades ( Figure 3F). Thus, differences in desensitization between the 1 mM and 1 µM concentrations were in large part acting through ERK signaling. Cells 2022, 11, x FOR PEER REVIEW 9 of 17

Time Course of Chronic Dopamine Administration Reveals Peak in DEGs at Day 3 and Desensitization at Day 5
We observed desensitization in hMSN-like cells after 5 days of chronic dopamine dosing, similar to the length of administration in rodent models [1,14]. We next asked what the kinetics of gene expression changes and (de)sensitization were over time. We dosed hMSN-like cells with 1 mM dopamine once a day for five days and performed RNA-seq 1 h (DIV50) after each dose ( Figures 4A and S4, Dataset S6). We calculated the total number of DEGs over time and observed a peak on the third day of chronic dosing, which then dropped. We did observe a common set of genes that were differentially expressed at all time points ( Figure 4B) and enriched for biological process GOs related to responses to stimuli and KEGG pathways such as MAPK signaling ( Figure 4C). MAPK/ERK responses were observed prior to day 5, in agreement with our prior functional enrichment analysis of desensitized genes. However, despite this commonality, intriguingly, the nature of the response each day dynamically shifted. Genes unique to day 3 were related to the extracellular matrix organization and cell adhesion or morphogenesis. This then shifted, where day 4 was enriched for the nicotine addiction KEGG pathway, a pathway associated with nicotine-induced alterations in glutamatergic and GABAergic receptors (e.g., GRIN1 and GABRR2) on GABAergic MSNs [52]. By day 5, the expression profile had shifted yet again to cholesterol and fatty acid metabolism. It is notable that the number of DEGs peaked at day 3, with many genes not activating or repressing until multiple doses of dopamine were administered.

hMSN-like Cells Capture Some Features of Dopamine Receptor Cross-Interactions
hMSN-like cells respond acutely to dopamine and exhibit desensitization to repeated exposures. However, one of the enduring challenges for MSN biology in rodents and humans is that MSNs exist as complex subtypes with multiple distinct receptors that may be expressed in many combinations [53][54][55]. Furthermore, these receptors can have complex downstream cross interactions. For example, D2-like receptor expression is normally associated with ADORA2A expression in the same MSNs in vivo [55]. Furthermore, D2-like receptors inhibit the activity of adenylyl cyclase downstream of ADORA2A [53][54][55]. We directly tested whether this crosstalk interaction between D2-like receptors and ADORA2A was maintained in human cells. We acutely and separately dosed DIV45 hMSN-like cells with the D2-like receptor agonist quinpirole and ADORA2A agonist CGS21680 (Figures 5 and S5, Dataset S7). DEGs were the most abundant after dosage with CGS21680 ( Figure 5A). CGS21680-induced DEGs were enriched for molecular function GO and reactome pathways related to G protein-coupled receptor signaling ( Figure 5B), as expected for ADORA2A binding [53]. More interesting than the effects of individual agonists was when quinpirole and CGS21680 were added together. DEGs induced by quinpirole alone largely overlapped with those induced by CGS21680 alone ( Figure 5C). However, preincubation with quinpirole an hour before the CGS21680 dosage negated the significant differential expression induced by CGS21680. This matches much of what is understood about D2-like receptors and ADORA2A. This result is also in alignment with previously studied interactions between the adenosine and dopamine systems [56]. This cross-interaction suggests that hMSN-like cells may be capable of capturing complex MSN signaling.
3 were related to the extracellular matrix organization and cell adhesion or morphogenesis. This then shifted, where day 4 was enriched for the nicotine addiction KEGG pathway, a pathway associated with nicotine-induced alterations in glutamatergic and GABAergic receptors (e.g., GRIN1 and GABRR2) on GABAergic MSNs [52]. By day 5, the expression profile had shifted yet again to cholesterol and fatty acid metabolism. It is notable that the number of DEGs peaked at day 3, with many genes not activating or repressing until multiple doses of dopamine were administered.  quinpirole alone largely overlapped with those induced by CGS21680 alone ( Figure 5C). However, preincubation with quinpirole an hour before the CGS21680 dosage negated the significant differential expression induced by CGS21680. This matches much of what is understood about D2-like receptors and ADORA2A. This result is also in alignment with previously studied interactions between the adenosine and dopamine systems [56]. This cross-interaction suggests that hMSN-like cells may be capable of capturing complex MSN signaling.

Discussion
New experimental models are needed to better understand the complex humanspecific responses to dynamically variable perturbations. In this study, we highlighted the ability of hMSN-like cells to capture many simple gene expression responses as well as the more complex dynamic and signaling properties of the cells. These included gene desensitization in response to chronic dopamine exposure and receptor crosstalk in response to agonists of dopamine receptor subtypes. However, further development will be needed to improve these models, in particular, to improve the homogeneity and purity of hMSNlike cultures and to connect observed responses to physiological contexts. Identifying the similarities and differences to rodent models may also identify areas for improvement in hMSN-like models.
We showed that hMSN-like cells exhibited gene desensitization to a wide range of concentrations of dopamine. We also found that desensitized immediate early genes in hMSN-like cells, such as FOS, FOSB, JUNB, and DUSP14, matched those found in prior rat studies [1,57]. We also found that highly desensitized genes were associated with signaling through ERK1/2, a phenomenon also observed in the mice striatum in response to drugs of abuse [58]. Additionally, we described how the number of DEGs peaked at 3 days of chronic dopamine exposure and were functionally enriched at that point for processes thought to underlie neuroadaptations in substance use disorders. Specifically, alterations in the extracellular matrix organization may relate to the synaptic modifications observed after chronic cocaine exposure in mice [59]. These processes shifted to metabolic-related ontologies and pathways, along with a reduction in the total number of DEGs, after day 3. Finally, we showed that the D2-like receptor agonist quinpirole and the ADORA2A agonist CGS21680 induced DEGs. In addition, DEGs induced by CGS21680 were inhibited by quinpirole preincubation. Thus, D2-like receptors function as expected in hMSN-like cultures, and complex cross-interactions between receptors were observed. Overall, we observed overlap in many aspects of acute and chronic transcriptional responses and dynamics between hMSN-like cells and rodent MSNs. Understanding the transcriptional dynamics in hMSN-like cells may improve our understanding of the role specific genes play in the development of addictive behaviors in humans.
Despite many similarities, many differences also exist between the responses of hMSNlike cultures and rodent primary MSNs. Importantly, heterogeneity is observed in hMSNlike cells. For example, we observed the expression of genes characteristic of medial ganglionic eminence interneurons when MSNs should express only lateral ganglionic eminence genes [17]. This is partly due to the fact that differentiated culture systems in general do not generate pure populations of cells [60]. Furthermore, they may not be able to differentiate stem cells into fully mature, subtype-specific neurons, with neurons in development co-expressing mixtures of markers not observed in adult neurons [61]. It is important to also note that this heterogeneity may contribute to the differential responses observed in hMSN-like cells compared to rodent MSNs. For example, there were many more downregulated DEGs due to acute dopamine exposure observed in this study compared to what has previously been reported in rodents [40]. Furthermore, chronic dopamine dosage led to fewer DEGs in hMSN-like cells compared to more DEGs after chronic cocaine dosage in rats [1]. These differences could arise due to species-specific differences, heterogeneity, as well as experimental differences. For example, although we time-matched dosages and RNA extractions in most experiments, when comparing acute to chronic conditions, they cannot both be simultaneously matched. One additional experimental consideration is how dopamine dosing is performed in this study, where whole-cell incubation of dopamine does not necessarily replicate in vivo exposures. In vivo, dopamine exposure occurs at synapses on the length scale of tens of nanometers and time scale of microseconds [36]. However, given the rapid autooxidation rate of dopamine in vitro, higher concentrations of dopamine are potentially required to bind dopamine receptors despite precautions to maintain stability in acidic aqueous solutions [62]. A deviation in hMSN-like culture and rodent striatal MSN responses in culture may also be due to different distributions of MSN subtypes and precursors. For example, FOS expression in predominantly D2-like hMSN-like cells may be partly due to the natural lack of dopamine in basal hMSN-like culture media [17], as dopamine depletion in the globus pallidus has been shown to lead to FOS expression in response to quinpirole [63]. Additionally, the expression of dopamine receptor heteromers, which are more highly expressed in neonatal striatal neurons than in adult MSNs, may lead to unexpected transcriptional dynamics not seen in vivo [64]. hMSN-like cells may also not compare well with adult rodent studies because hMSN-like cells tend to lack functional maturity when compared to adult rodent neurons [61,65]. The lack of noticeable spines, characteristic of in vivo MSNs, also highlights the relative immaturity of hMSN-like cells [66]. Longer experiments may provide additional time for maturation but face practical experimental challenges. Functional maturation may also be due to the lack of other relevant cell types that MSNs reside with in vivo such as astrocytes, glutamatergic neurons, and dopaminergic neurons.
Future work could aim to improve the functional maturity and other phenotypes of hMSN-like cells that are prominently seen in primary rodent cultures. Prior work using primary rodent cultures has shown enhancements in functional characteristics such as increased dendritic spine density and spontaneous action potentials after cocultures with cortical neurons and astrocytes, respectively [67,68]. Similarly, cocultures between hMSN-like cells and other physiologically relevant human cell types may enhance the functional characteristics. Three-dimensional models of hMSN-like cells such as cerebral organoids could also potentially yield more mature neurons. In addition to improving neuronal maturation, new protocols that could either create or pan for purer populations of MSN subtypes would provide improved reductionist control to deconvolve the biology of MSNs. In addition, many protocols already exist to differentiate human stem cells to DARPP32+ striatal neurons [18,69] and it would be advantageous for the field to identify the differences in heterogeneity produced using each protocol using single-cell sequencing technologies. Future work could also include the purification of target cell types by tagging the DARPP32 gene with a fluorescent reporter and sorting via flow cytometry [70]. Finally, it would be interesting to investigate the post-translational and epigenetic changes in histone modifications that may underlie the mechanisms of desensitization, whether these changes are observed in hMSN-like cells in response to chronic dopamine administration as they are observed in rodents, as well as expand to other dosing regimens to investigate transcriptional or epigenetic changes in experiments mimicking withdrawal [71,72]. For example, we observed transcriptional desensitization of the ERK pathway genes; determining if post-translational phosphorylation of ERK1/2 is also desensitized would be of interest mechanistically. The work here highlights the promise as well as current limitations of stem cell-derived hMSN-like cells as a model to study complex responses to substances of abuse and other diverse perturbations in humans.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cells11091411/s1, Figure S1: QRT-PCR of target genes as compared to GUSB; Figure S2: FOSB immunostaining highlights differential responses to acute and chronic dopamine; Figure Table S1: Primary Antibodies; Table S2: RT-qPCR Primers; Dataset S1: Overlapping and unique differentially expressed genes between 1 µM and 1 mM acute dopamine conditions one hour after an acute dosage (Acute 1 h); Dataset S2: Overlapping and unique differentially expressed genes between 1 µM and 1 mM acute dopamine conditions twenty four hours after an acute dosage (Acute 24 h); Dataset S3: Overlapping and unique differentially expressed genes between 1 µM and 1 mM chronic dopamine conditions one hour after a chronic dosage (Chronic 1 h); Dataset S4: Overlapping and unique differentially expressed genes between 1 µM and 1 mM chronic dopamine conditions twenty four hours after a chronic dosage (Chronic 24 h); Dataset S5: Overlapping and unique desensitized differentially expressed genes between 1 µM and 1 mM dopamine conditions one hour after dosage (1 h); Dataset S6: Overlapping and unique differentially expressed genes between 2-day through 5-day chronic dopamine dosed hMSN-like cells one hour after dosage; Dataset S7: Overlapping and total differentially expressed genes between agonist dosed conditions 1 h after dosage as compared to water control.