Compensation between FOXP transcription factors maintains proper striatal function

Abstract

Spiny projection neurons (SPNs) of the striatum are critical in integrating neurochemical information to coordinate motor and reward-based behavior. Mutations in the regulatory transcription factors expressed in SPNs can result in neurodevelopmental disorders (NDDs). Paralogous transcription factors Foxp1 and Foxp2, which are both expressed in the dopamine receptor 1 (D1) expressing SPNs, are known to have variants implicated in NDDs. Utilizing mice with a D1-SPN specific loss of Foxp1, Foxp2, or both and a combination of behavior, electrophysiology, and cell-type specific genomic analysis, loss of both genes results in impaired motor and social behavior as well as increased firing of the D1-SPNs. Differential gene expression analysis implicates genes involved in autism risk, electrophysiological properties, and neuronal development and function. Viral mediated re-expression of Foxp1 into the double knockouts was sufficient to restore electrophysiological and behavioral deficits. These data indicate complementary roles between Foxp1 and Foxp2 in the D1-SPNs.

Figure 1:

Schematic showing the genotypes used in this study. D1-cre mediated loss of Foxp1 (Foxp1^D1; purple), Foxp2 (Foxp2^D1; cyan), or both (Foxp1/2^D1; gold), as well as cre-negative controls (grey).

Figure 2:

Quality control of nuclei sequenced from adult snRNA-Seq. (A) Pie charts showing the total number of nuclei sequenced from each genotype as well as the number of D1-SPNs profiled for each condition. (B) Violin plots showing the number of genes and UMI for each cell-type annotated within the dataset. (C) UMAP showing all of the annotated clusters from the dataset. A total of 11 major cell-types were identified. SPNs and progenitors were subset for further analysis. (D) Bubble plot showing overlap with marker genes used to annotate clusters. (E) Stacked bar plot showing the proportion of each genotype in all the cell-types as well as within the neuron-only subset (F) that was later generated. (G) Violin plots showing the number of genes and UMI for the remaining neuronal sub-types in the subset dataset.

Figure 3:

Loss of both Foxp1 and Foxp2 results in amplified loss of transcriptional regulation in D1-SPNs in juvenile mice. (A) UMAP plot generated from the neuron-only subset with colors indicating the different annotated cell-types. D1-SPNs were used for further DEG analysis. Genes were determined to be differentially expressed from controls if they had an adjusted p-value < 0.05 and absolute logFC >|0.25|. (B) Semi-scaled Venn diagram showing number of unique and overlapping DEGs in each knockout condition in the D1-SPNs. (C) Bar plots showing the number of up- and downregulated genes in each knockout condition. (D) Bubble chart showing enrichment of DEGs from each knockout condition. The - log₁₀(p-value) for each enrichment is also indicated. ASD, SFARI ASD risk genes; ASD 1–3, SFARI ASD risk genes with scores of 1–3; FMRP, Fragile X Syndrome; ID, Intellectual disability; SYN, synaptic genes. Gene Ontology (GO) analysis of (E) Foxp1^D1, (F) Foxp2^D1, and (G) Foxp1/2^D1 DEGs reveals enrichment for terms associated with electrophysiological properties and synaptic properties.

Figure 4:

Quality control of juvenile snATAC-Seq. (A) Pie charts showing the total number of nuclei sequenced from each genotype as well as the number of D1-SPNs profiled for each condition. (B) Stacked bar plots showing the proportion of each genotype within the cell-types in the dataset. Plots showing the number of peaks (D-C), percent of reads in peaks (C), and number of reads in peaks (E) in each knockout condition. (F) Trackfile for Kcnip1 with the differentially accessible region highlighted.

Figure 5:

Loss of both Foxp1 and Foxp2 dysregulates chromatin state in D1-SPNs. (A) UMAP showing annotated cell-types in nuclei collected for snATAC-Seq analysis. D1-SPNs were used for further DAR analysis. A region was deemed to be differentially accessible if it had an adjusted p-value < 0.05 and a logFC > |0.1375|. (B) Semi-scaled Venn diagram showing the number of unique and overlapping DARs within each condition. (C) Venn diagrams showing the overlap between differentially expressed genes and DARs in Foxp1/2^D1 D1-SPNs. (D) Bar plot showing the number of more open, more closed, or both more open and more closed regions in each knockout condition. Motifs enriched in the DARs of each knockout were identified. (E) Semi-scaled Venn diagram showing the number of unique and overlapping motifs in each genotype. (F) Bar plot showing the number of motifs enriched within more open, more closed, or both more open and more closed chromatin regions in each knockout condition. FOX and MEF2 motifs highlighted to indicate where they were enriched. (G) Trackfile for Pde1c with the differentially accessible region highlighted.

Figure 6:

Loss of Foxp1 results in K_Leak mediated hyperexcitability with amplification by further loss of Foxp2. (A) Number of action potentials recorded under current clamp conditions. (B) Input resistance recorded from the same cells in the same conditions. (C) Contribution of K_Leak channels was determined by finding the difference in current density plots. Values recorded in presence of cesium (see Figure S3D) were subtracted from those recorded in its absence (see Figure S3C) with no significant differences observed. (D) Schematic showing pAAV-hSYN-Foxp1-T2A-eGFP construct which was injected into mice at postnatal day 1. Dual presence of td-Tomato and GFP was used to identify which neurons had taken up FOXP1 construct. (E) Current clamp recordings done in mice that were injected with the FOXP1 or control construct to record number of action potentials and (F) input resistance. Repeated measures Two-Way ANOVA with Holm-Sidak’s post-hoc test; only significant differences between knockouts and controls are shown. *p<0.05 (A & B) n=56 (control), 32 (Foxp1^D1), 40 (Foxp2^D1), and 41 (Foxp1/2^D1). (C) n=61, 43, 38, and 41. (E & F) n=44 (control with control virus), 34 (Foxp1/2^D1 with control virus), 32 (Foxp1/2^D1 with FOXP1 construct), and 19 (control with FOXP1 construct).

Figure 7:

Foxp1 and Foxp2 mediate hyperexcitability in D1-SPNs. (A) Table with significant differences when comparing D1-SPN excitability between all genotypes (related to Figure 4A). (B) D1-SPNs from Foxp2^D1 have greater Foxp1 expression than those from control mice as assessed by RT-qPCR. (C) Current voltage (IV) plots were generated to identify K_IR step density in absence and presence (D) of cesium to block K_IR currents; both Foxp1^D1 and Foxp1/2^D1 D1-SPNs show differences compared to controls in both conditions. (E) Input resistance is increased in Foxp1^D1 and Foxp1/2^D1 D1-SPNs even in presence of cesium and is decreased in Foxp2^D1. (F) Expression of Kcnk2 is increased specifically in D1-SPNs from Foxp2^D1 mice. (B, F) One-way ANOVA with Tukey’s post-hoc analysis, **** p < 0.0001, n=3 for all conditions. (C-E) Repeated measures Two-Way ANOVA with Holm-Sidak’s post-hoc test, *p<0.05, n=61, 43, 38, and 41 respectively.

Figure 8:

Quality control of nuclei sequenced from adult snRNA-Seq. (A) Pie charts showing the total number of nuclei sequenced from each genotype as well as the number of D1-SPNs profiled for each condition. (B) Violin plots showing the number of genes and UMI for each cell-type annotated within the dataset. (C) UMAP showing all of the annotated clusters from the dataset. A total of 11 major cell-types were identified. SPNs and progenitors were subsetted for further analysis. (D) Bubble plot showing overlap with marker genes used to annotate clusters. (E) Stacked bar plot showing the proportion of each genotype in all of the cell-types as well as within the neuron-only subset (F) that was later generated. (G) Violin plots showing the number of genes and UMI for the three SPN sub-types in the subset dataset. (H) Bar plot showing the number of DEGs in cortical cell-types in each knockout condition. EX23_IT: Excitatory cells in Layer 2/3 with intratelencephalic projections, EX45_IT: Excitatory cells in Layer 4/5 with intratelencephalic projections, EX5_ET: Excitatory cells in Layer 5 with extratelencephalic projections, EX5_IT: Excitatory cells in Layer 5 with intratelencephalic projections, EX6_CT: Excitatory cells in Layer 6 with cortico-thalamic projections, EX6_IT: Excitatory cells in Layer 6 with intratelencephalic projections, EX6B: Excitatory cells in Layer 6B, IN_PVALB: Pvalb+ inhibitory cells, IN_SCNG: Scng+ inhibitory cells, and IN_SST: Sst+ inhibitory cells.

Figure 9:

Differentially expressed genes in adults SPNs have similar biological functions as those observed in juveniles. (A) UMAP plot generated from the neuron-only subset with colors indicating the different annotated cell-types. D1-SPNs were used for further DEG analysis. Genes were determined to be differentially expressed from controls if they had an adjusted p-value < 0.05 and absolute logFC >|0.25|. (B) Semi-scaled Venn diagram showing number of unique and overlapping DEGs in each knockout condition in the D1-SPNs. (C) Bar plots showing the number of up- and downregulated genes in each knockout condition. (D) Bubble chart showing enrichment of DEGs from each knockout condition. The - log₁₀(p-value) for each enrichment is also indicated. ASD, SFARI ASD risk genes; ASD 1–3, SFARI ASD risk genes with scores of 1–3; FMRP, Fragile X Syndrome; ID, Intellectual disability; SYN, synaptic genes. Gene Ontology (GO) analysis of (E) Foxp1^D1, (F) Foxp2^D1, and (G) Foxp1/2^D1 DEGs reveals enrichment for terms associated with electrophysiological properties and synaptic properties.

Figure 10:

Loss of Foxp1 and Foxp2 results in impaired motor and social behavior. (A) Foxp1/2^D1 mice show motor learning deficit as assessed by latency to fall using the rotarod paradigm and in (B) motor activity measured as observed by amount of movement during 5-minute open field assay. (C) Nest building quality was assessed after single housing for 24 hours where Foxp1^D1 had an impairment. Impairment amplified in Foxp1/2^D1. AAV-mediated re-expression of Foxp1 restored measures to baseline in rotarod (D), motor activity (E), and nest building (F). (A-C) Behavior performed in control, Foxp1^D1, Foxp2^D1, and Foxp1/2^D1. (D-F) Performed in control or Foxp1/2^D1 mice with either control AAV9-hSYN1-eGFP or pAAV-hSYN-Foxp1-T2A-eGFP construct. Two- (A&D) or One- (B, C, E, & F) Way ANOVA with Tukey’s post-hoc analysis used to determine significance. *p<0.05, **p<0.01, ***p<0.001. n=25 (control), 13 (Foxp1^D1), 14 (Foxp2^D1), and 16 (Foxp1/2^D1) for (A&B), n=15, 13, 18, and 12 respectively for (C), and n=14, 11, 12, and 10 respectively for (D-F).

Figure 11:

Loss of Foxp1 and/or Foxp2 has no effect on limb strength or anxiety. No deficits were seen in either fore- (A) or hind-(B) limb grip strength. (C) No changes were seen in anxiety levels as assessed by time spent in periphery during the 5-minute open-field assay. (D) Violin plots showing the number of pup isolation ultrasonic vocalizations made by mice of all four genotypes at postnatal days 4, 7, and 10. No changes were observed when comparing number of calls made at each time point across genotypes. One- (A-C) or Two- (D) Way ANOVA with Tukey’s post-hoc analysis used to determine significance. n=17, 12, 13, and 14 respectively for (A-C), and n=310, 46, 44, and 47 for (D).

Figure 12:

Foxp1 and Foxp2 largely compensate to maintain repression of target genes. We propose one mechanism of compensation wherein both Foxp1 and Foxp2 bind to and regulate similar targets. Upon loss of one transcription factor, the remaining transcription factor takes over regulation of these shared targets and maintains chromatin in a closed state. Upon loss of both factors, we observe dysregulation of target genes and altered chromatin state.