Cell‐type‐specific gene expression and regulation in the cerebral cortex and kidney of atypical Setbp1 S858R Schinzel Giedion Syndrome mice

Abstract Schinzel Giedion Syndrome (SGS) is an ultra‐rare autosomal dominant Mendelian disease presenting with abnormalities spanning multiple organ systems. The most notable phenotypes involve severe developmental delay, progressive brain atrophy, and drug‐resistant seizures. SGS is caused by spontaneous variants in SETBP1, which encodes for the epigenetic hub SETBP1 transcription factor (TF). SETBP1 variants causing classical SGS cluster at the degron, disrupting SETBP1 protein degradation and resulting in toxic accumulation, while those located outside cause milder atypical SGS. Due to the multisystem phenotype, we evaluated gene expression and regulatory programs altered in atypical SGS by snRNA‐seq of the cerebral cortex and kidney of Setbp1 S858R heterozygous mice (corresponds to the human likely pathogenic SETBP1 S867R variant) compared to matched wild‐type mice by constructing cell‐type‐specific regulatory networks. Setbp1 was differentially expressed in excitatory neurons, but known SETBP1 targets were differentially expressed and regulated in many cell types. Our findings suggest molecular drivers underlying neurodevelopmental phenotypes in classical SGS also drive atypical SGS, persist after birth, and are present in the kidney. Our results indicate SETBP1's role as an epigenetic hub leads to cell‐type‐specific differences in TF activity, gene targeting, and regulatory rewiring. This research provides a framework for investigating cell‐type‐specific variant impact on gene expression and regulation.

3][4] SETBP1 encodes for SET Binding Protein 1, a transcription factor (TF) which acts as an epigenetic hub and has many understudied functions in SGS. 5 SETBP1 variants associated with classical SGS cluster at mutational hotspots in the degron and lead to disruptions in protein degradation resulting in a toxic accumulation of SETBP1 protein, therefore making SGS a gain-of-function (GoF) disorder.[8] Prior human cell and mouse model studies uncovered multiple potential mechanisms altering neurodevelopment in SGS (reviewed in 9 ), documenting the role of SETBP1 accumulation in inhibiting p53 function, causing aberrant proliferation, deregulating oncogenes and suppressors, and promoting unresolved DNA damage and apoptosis resistance in neural progenitor cells (NPCs). 10SETBP1 variants were also implicated in decreased genome panacetylation. 4Furthermore, variants resulting in SETBP1-mediated PP2A inhibition have been found somatically in several myeloid malignancies as well as in germline variants related to altered neurodevelopment. 9SGS's syndromic presentation and SETBP1's TF role provide the opportunity to study the impact of a variant on cell-type-specific expression and regulation in a context-specific framework.However, no studies to date have determined if these molecular alterations are present in other affected cell types or tissues, like the kidney, where hydronephrosis and recurrent urinary tract infections are secondary phenotypes of SGS.
Multiple biological processes have been hypothesized to explain tissue-specific manifestations of disease phenotypes. 11To understand the multisystem phenotype of SGS, we evaluated gene expression and regulatory programs altered in atypical SGS through single-nuclei RNA-sequencing (snRNA-seq) of the cerebral cortex and kidney of 6-week-old male Setbp1 S858R heterozygous mice compared to age-and sex-matched wild-type (WT) mice and construction of cell-type-specific regulatory networks (Figure 1).The Setbp1 S858R variant corresponds to the human likely pathogenic SETBP1 S867R atypical SGS variant, located within the SKI homologous domain (amino acids 706-917), a region named for its partial homology to the SKI oncoprotein. 12In addition to determining cell-type-specific impacts of the Setbp1 S858R variant in two SGS-affected tissues, our research provides a framework for investigating changes introduced by a variant to mechanisms of cell-type-specific expression and regulation in this understudied rare disease.

| MATERIAL S AND ME THODS
A more detailed version of all data processing and analysis steps is available in the Appendix S2.

| In vivo mouse samples and nuclei isolation
We obtained decapsulated kidneys and cerebral cortex hemispheres from three 6-week-old male C57BL/6J-Setbp1 em2Lutzy /J mice heterozygous for Setbp1 S858R (JAX Stock #033235) and three wild-type (WT) age-and sex-matched C57BL/6 J mice (JAX Stock #000664) (Table S1). 13 adapted and conducted single nuclei isolation for the kidney as previously described. 14We resuspended the final nuclei pellet in Nuclei Suspension Buffer with a final yield of 1.3-3.0 million nuclei per kidney.We adapted our single nuclei isolation for cerebral cortex samples for mice from the previously described 15 and sorted nuclei using fluorescence-activated cell sorting (FACS) on an Atlas BD ARIA II with a yield of 1.5-3.0 million nuclei per cerebral cortex hemisphere.

| snRNA-seq processing
We generated single-nuclei RNA-sequencing (snRNA-Seq) libraries using the 10X Chromium Single Cell 3′ v3 (10X#: PN-1000121) protocol by loading 20,000 nuclei per sample with a targeted cell recovery rate of 10,000 nuclei per sample.All cerebral cortex samples were processed together on the same day.Kidney samples were processed together on a separate day.GEMs were generated individually.The resulting cDNA was multiplexed and the libraries were indexed and sequenced together on a single S4 flow cell with a NovaSeq 6000 (Illumina).

| SETBP1 protein quantification
We used the Mouse SETBP1 ELISA kit (MyBioSource Cat No. MBS9335445) to quantify SETBP1 protein abundance in kidney and cerebral cortex samples for both conditions (n = 3 replicates per condition per tissue).We performed incubation and manual washing as described in the MyBioSource Kit protocol and measured absorbance using a Tecan Infinite M Plex plate reader at 450 nm for 10 flashes.The proportion of total protein abundance in սg/mL was measured in all tissues (n = 3 per condition per tissue) using the BCA with the Pierce protocol. 19Significance was calculated using a paired t-test between groups (p-value <0.05) (Figure S2).

| Detection of differentially expressed genes
We applied the Seurat FindAllMarkers function with an average log 2 FC threshold of 0.1.Results for all differentially expressed genes (p-adjusted value <0.05) for all S858R cerebral cortex and kidney cell types (cell type markers; Table S2) are in Tables S3 20 We performed functional enrichment analysis (FEA) on the differentially expressed genes filtered for the SETBP1 target gene set (gene set; Appendix S2, Table S5) and applied a p-adjusted threshold of 0.05 and a Bonferroni procedure for multiple hypothesis correction.
F I G U R E 1 Schematic overview of our approach (A) We generated single-nuclei RNA-seq (snRNA-seq) from male C57BL/6J (WT) and Setbp1 S858R cerebral cortex and kidney tissues.(B) We processed and aggregated our data across samples to create cell-type-specific count matrices.(C) Next, we assessed cell-type-specific expression for Setbp1 and genes SETBP1 is known to target.(D) With decoupleR, we measured cell-type-specific TF activity.(E) We acquired protein-protein interaction data (PPI; green) from STRING and Transcription Factor-motif data (TF-motif; yellow) from CIS-BP enriched for SETBP1 targets.We then built cell-type-specific TF-gene regulatory networks with the data we generated by using the message-passing algorithm PANDA to identify regulatory relationships between TFs (circles), genes (rectangles), and proteins (triangles) for each S858R and WT cell type in both tissues and investigated differential communities using differential community detection with ALPACA, differential gene targeting, cooperativity analysis, and functional enrichment analysis.

| TF activity
We inferred transcription factor (TF) activity using decoupleR 25 (v.2.6.0) by combining our scRNA-seq data with prior knowledge from CollecTRI, a comprehensive network of TFs and their direction of regulation on transcriptional targets (accessed May 2023). 26 calculated activity scores for all 732 TFs by using a Multivariate Linear Model, only including TFs with a minimum of five targets.We summarized average TF activity by each cell type.Then we scaled and centered data to prioritize TFs with the largest change in activity between conditions using absolute percentage change of the average activity.More positive percent changes (above Q3) represented a larger magnitude of TF activity in the S858R, and the more negative percent changes (below Q1) represented a larger magnitude of TF activity in WT.

| Cell-type-specific network construction and analysis
To model cell-type-specific regulatory relationships between TF and genes within both conditions, we constructed cell-type-specific networks using Passing Attributes between Networks for Data Assimilation (PANDA) (v.1.2.1). 27,28Network edge weights represent the strength in interactions between nodes (coexpressed genes, proteins, or pairs of TFs and genes) where more positive weights correspond to higher confidence interactions in a given cell-typespecific network.All analysis was performed in a high-performance computing environment at UAB. Gene targeting enabled us to quantify all in-degree edge weights of each gene.A higher targeting score indicated a gene had a higher sum of all in-degree edge weights.By calculating a differential targeting score for a gene between two networks, we could compare if a gene is targeted more or more 'important' in one network versus the other.We calculated differential gene targeting between conditions by taking the difference in targeting scores for each gene, with positive scores indicating enrichment in S858R mice.Genes with differential targeting scores above the third quartile were selected within each cell type and further investigated using GSEA after filtering for the SETBP1 gene set (p-adjusted <0.05).
We conducted a comparative analysis of regulatory network community structures in all cell types for both conditions by employing the bipartite community detection algorithm, COmplex Network Description Of Regulators (CONDOR) (v.1.2.1), 29 on positive edge weights from the PANDA networks.We used ALtered Partitions Across Community Architectures (ALPACA) (v.1.2.1), 30 a message passing and modularity optimization method, to detect structural changes in closely interacting groups of nodes between conditions referred to as differential communities and quantified the similarities between SETBP1 communities across cell types within each tissue using the Jaccard similarity index (JI).
To understand regulatory network rewiring due to the Setbp1 S858R variant, we filtered regulatory (TF-gene) and cooperativity (protein-protein) networks to only include edges associated with the SETBP1 gene set.We then annotated the regulatory and cooperativity network edges, categorizing them based on the presence or absence of regulation and cooperation between genes and proteins, and investigated specific proteins involved in mechanisms related to altered cell cycle, chromatin remodeling, DNA damage, and phosphorylation in SGS. 9

| SETBP1 targets are expressed in a cell-type-specific manner in atypical SGS Setbp1 S858R mouse cerebral cortex and kidney tissues
As prior studies using patient-derived cell lines and cerebral organoids indicated that variants in SETBP1 lead to an accumulation of SETBP1 protein in SGS, 10 we confirmed increased SETBP1 protein abundance in the S858R cerebral cortex and kidney tissues compared to WT (p-value <0.05, paired t-test) (Figure S2).After profiling single-nuclei RNA from the cerebral cortex (n = 51,038 total nuclei) and kidney (n = 78,751 total nuclei), we confirmed the presence of the variant in all S858R samples (Figure S1) and assigned cell types (Table S2, and Figures S3, S4).We identified differentially expressed genes (p-adjusted <0.05 Wilcoxon rank sum) between S858R and WT and found widespread gene expression changes impacting all cell types in our study (Tables S3, S4).Setbp1 was ubiquitously expressed across all cell types in both conditions, in line with prior studies (Figure 2A, B). 31 However, Setbp1 was only significantly differentially expressed (upregulated) in S858R excitatory neurons compared to WT (p-adjusted <0.05, Wilcoxon rank sum) (Figure 2C).Functional enrichment analysis (FEA) on all statistically significant differentially expressed genes in atypical SGS mice revealed cell-type-specificity in the cerebral cortex.These genes were overrepresented in pathways related to previously noted SGS signatures and clinical phenotypes such as cell death, apoptosis, language impairment, and neurodegeneration (Figures S5-S10).
There was no enrichment for differentially expressed genes in kidney cell types associated with polycystic kidney disease to support a cystic phenotype in SGS hydronephrosis 32 or in genes related to failed repair, a common kidney disease mechanism 33 (Figures S11, S12, Appendix S2).DNA damage accumulation has been previously confirmed in NPCs for SGS; however, to our knowledge, it has not been studied within the context of SGS kidneys.Interestingly, we found kidney cell-type-specific decreases in DNA repair, WNT-Beta Catenin signalling, and p53 pathway expression signatures in atypical SGS mouse kidneys (Figure S13B).Due to SETBP1's role as a TF, we investigated mouse orthologs of known SETBP1 targets for cell-type-specific differential expression between S858R and WT (Figure 2C, D).We found that many target genes with functions in SGS-associated pathways were differentially expressed (p-adjusted <0.05, Wilcoxon rank sum) (Table 1).Furthermore, Piazza et al. reported MECOM as a main target of SETBP1 epigenetic activity in blood malignancies. 5While we do not see differential expression between WT and S858R in cerebral cortex, Mecom was significantly differentially expressed in Loop of Henle (LOH), Distal Convoluted Tubule (DCT), fibroblasts, and proximal tubule cells of S858R kidneys compared to WT (Table S3).cells (OPCs), and excitatory neurons exhibited a decreased enrichment score for the p53 pathway compared to WT.We also found S858R cortical pericytes had reduced DNA repair, p53, and apoptosis, while microglia, OPCs, and excitatory neurons had decreased G2M enrichment compared to WT.Additionally, S858R excitatory neurons were enriched for apoptosis compared to WT (Figure S13A).
Due to the presence of these SGS-associated signatures, we profiled the expression of the proliferative NPC marker Sox2 and found exclusive expression in S858R astrocytes, suggesting they may have a disease-specific proliferative NPC-like population (Figure 2E, Appendix S2). 10,47While previous research has indicated inflammatory conditions or disease states such as Alzheimer's disease 48 drive mature astrocytes to reacquire NPC-like properties, 49 we did not find statistically significant differences in gene expression of the reactive astrocyte marker Gfap in astrocytes between conditions (Figure S14, Appendix S2).

TA B L E 1
Highlighted SGS-signature associated differentially expressed SETBP1 targets in S858R cerebral cortex and kidney cells.

Ddi2
Function in DNA damage response by promoting protein turnover. 43wnregulated: • Macrophages, PT, and PCTS2

Naaladl2
Knockdown of Naaladl2 has shown decreased cell viability accompanied by increased apoptosis in previously published cancer studies. 44regulated: • Kidney fibroblasts Downregulated: • Macrophages, DLH, PT, PST, and LOH F I G U R E 3 SETBP1 TF activity is altered in pericytes and microglia and multiple kidney cell types of S858R mice: Heatmap of decoupleR TF activity scores of SETBP1 and its target genes that displayed the largest percent change in activity in the cerebral cortex (A) and kidney (B).Teal and brown correspond to TFs acting as activators or repressors, respectively.
Accumulation of SETBP1 is known to result in the stabilization of SET and inhibition of PP2A in SGS. 10 We found ubiquitous expression of Set in all cell types and tissues (Figure S15, Appendix S2); however, Ppp2ca (the mouse ortholog for PP2A) was expressed in WT but not S858R astrocytes and S858R but not WT microglia (Figure S16, Appendix S2).There was no difference in other cell types between the S858R and WT cells of both tissues.PP2A's tumor suppressor function is known to be disrupted in SGS, and a lack of apoptosis in astrocytes has previously been noted as well. 9,10As reported above, our gene set signature enrichment analysis revealed a decrease in the G2M checkpoint signature in S858R astrocytes compared to WT, suggesting there may be more astrocytes in the G0/G1 cell cycle phase (Figure S13A).This is further supported by a subsequent downregulation of cell death and apoptotic processes in the pathways associated with downregulated SETBP1 target genes in S858R astrocytes (Figure S13A).

| SETBP1 and known targets' TF activity is altered across cerebral cortex and kidney cell types
Due to cell-type-specific changes in the gene expression of known SETBP1 targets, we inferred the TF activity of Setbp1 and its TF targets in the cerebral cortex and kidney using a multivariate linear model. 25We prioritized transcriptional regulators displaying the highest absolute percent change between WT and S585R and found that SETBP1 and its targets display cell-type-and disease-specific alterations in TF activity in the cerebral cortex and kidney (Figure 3).More specifically, in the cerebral cortex (Figure 3A), SETBP1 is upregulating its targets in WT pericytes and microglia.However, SETBP1 functions as a transcriptional repressor in S858R astrocytes.Interestingly, The TFs regulated by SETBP1 display cell-type-and disease-specific TF activity changes.We quantified the percent change of all TFs between WT and S858R and then subsetted for those that SETBP1 transcriptionally regulates.We see 14 of these transcriptional regulators in the cerebral cortex (Figure 3A).Interestingly, 13 of the 14 TFs were not differentially expressed.Mef2c was differentially expressed in excitatory and inhibitory neurons and microglia.We see the largest change in the TF activity of Mef2c in microglia and no change in TF activity in inhibitory neurons.In the kidney, we see altered TF activity in SETBP1 and 18 of its targets.These TF activity changes occur in every kidney cell type (Figure 3B).Although subtle, these results indicate that the S858R variant alters the gene regulatory landscape not only through differential gene expression of direct targets but also by altering the activity of additional transcriptional regulators.

| Cell-type-specific changes in gene targeting recapitulate SGS molecular features in many atypical SGS Setbp1 S858R mouse cell types
We constructed TF-gene regulatory networks for each cell type by condition using the message-passing algorithm PANDA 27 to calculate the 'responsibility' of a TF for regulating a gene and the 'availability' of that gene for regulation by that TF.We then averaged the responsibility and availability networks to obtain a consensus gene regulatory network per cell type.From these networks, we calculated the normalized differential targeting score for all genes by cell type in the cerebral cortex (Figure 4A) and kidney (Figure 4B).This revealed higher total differential targeting scores in specific cell types of S858R compared to WT for both tissues, suggesting increased gene regulation, consistent with gain of function, in the S858R (Table S7).We performed FEA on both tissues' differentially targeted genes from the SETBP1 gene set.In the cerebral cortex, microglia and excitatory and inhibitory neurons had differential expression of pathways indicating altered H3K9 acetylation and regulation in S858R compared to WT (Figure 4C).
These findings are in line with recent studies that show SET overexpression alters histone acetylation and accumulation of SET can impair chromatin accessibility at distal regions of the genome, specifically, chromatin loop contacts in SGS NPCs. 4 Our FEA also identified sloping forehead as a phenotype term associated with the excitatory neuron S858R gene signature, concordant with clinical indications that individuals with SGS have pronounced craniofacial features (i.e.excessive posterior sloping of the forehead). 50In the kidney, we also identified altered H3K9 signatures in S858R Collecting Duct Intercalated Cells (CDIC) type A and B and pericytes.Furthermore, S858R LOH, DLH, Proximal Convoluted Tubule Segment 1 (PCTS1), and DCT cells displayed differential pathway enrichment for cellular response to lectin, previously implicated in renal injury (Figure 4D). 51,52Overall these results demonstrate that SETBP1 with the S858R variant differentially targets genes by cell type and that molecular programs associated with typical SGS are reflected in the cerebral cortex and kidney of this model of atypical SGS.

| Differential communities between S858R and WT containing SETBP1 indicate shared SETBP1 regulatory patterns between sets of cell types
We identified communities in each cell-type-specific network based on differential modularity (a metric measuring the extent to which edges in communities of S858R networks differ from those of reference WT networks) for both the cerebral cortex and kidney using the ALPACA framework. 30When we analyzed differential communities containing SETBP1 as a TF, we found that excitatory neurons and astrocytes had the highest Jaccard Similarity Index (JI) (0.42, Figure 5A) and were also more similar to OPCs (0.44 and 0.38, respectively) than to other cerebral cortex cell types.
Oligodendrocytes and pericytes, not previously suggested to have a role contributing to SGS, had a JI of 0.59, concordant with having similar pathway signatures in S858R compared to WT as described above (Figure S17).These two cell types also had high JI with fibroblasts (0.41 for both).In kidney, we found SETBP1-specific differential communities were most similar between DCT, DLH, PCTS2, Proximal Tubule (PT), endothelial, and LOH cell types (JI = 0.30 to 0.53), between fibroblasts, macrophages, pericytes, podocytes, and smcs (JI = 0.38-0.64),and between B cells, Proximal Straight Tubule (PST), CDIC type A and B and CDPC (JI = 0.34-0.51)(Figure 5B).
Overall, we found that there are sets of cell types with more similar SETBP1 differential communities, further underscoring the multifunctional role of SETBP1 in atypical SGS and that additional cell types are impacted by, may contribute to, and work together in disease aetiology and pathogenesis.

| Cell-type-specific DNA damage and cell cycle associated network rewiring in S858R cerebral cortex and kidney cells
While little is known about the role of specific patient variants, sev- cycle control, increased DNA damage, and modified PP2A complex activity. 9To further evaluate the molecular impact of the S858R variant, we investigated the magnitude and direction of the regulatory network and changes in cooperativity network edge weights where SETBP1 is acting as a TF.Similar to the regulatory network, a more positive edge weight between two proteins indicates a greater likelihood they cooperate to regulate gene expression.By contrast, a more negative edge weight indicates greater confidence that the two proteins do not cooperate.We used these regulatory or cooperativity edge weight magnitude and direction changes to infer potential regulatory rewiring due to the S858R variant.While altered cell cycling, DNA damage, chromatin remodeling and phosphorylation have been previously implicated by disrupted SETBP1 interactions in the brain, 9 we do not capture SETBP1 interacting with proteins associated with chromatin remodeling and phosphorylation but do identify interactions with proteins related to DNA damage and cell cycle interactions in our networks (Figure 6A).

We filtered edges to include only those involving SETBP1 as a TF
where the target is also a protein in the cooperativity network (these interactions were previously measured by ChIP-seq 5 ) and then filtered for APEX1 (cooperating protein from the proposed DNA damage mechanism) and TRP53 (cooperating protein from the proposed cell cycle mechanism), therefore identifying 4 categories depending on the changes in edge weight and direction in the regulatory and cooperativity networks (Figure 6B).
For both the cerebral cortex and kidney, the proposed altered DNA damage and cell cycle mechanisms were cell-type-specific (Figure 6 C-F, Figure S18).
We found evidence of several rewiring events.For example, SETBP1 regulation of Hmga1 is present in S858R but not WT inhibitory neurons, and HMGA1 subsequently cooperates with APEX1, a protein involved in DNA damage.Microglia also had regulatory changes indicative of altered DNA damage mediated through a SETBP1-E4f1/ E4F1-APEX1 interaction.However, a lack of cooperation between E4F1 and APEX1 is introduced in the S858R condition (Figure 6C).does not infer direct interaction or binding, we confirmed interactions for all proteins above using scores from STRING (Table S8).

| DISCUSS ION
We profiled the impact of Setbp1 S858R on cell-type-specific expression and regulation within mouse cerebral cortex and kidney.damage, and apoptosis resistance as a result of SETBP1 accumulation in the brain, 4,5,9,10 here we investigated S858R's impact on multiple cell types in both the cerebral cortex and kidney showing these properties extend to additional cell types.Setbp1 is known to be ubiquitously expressed across cells and tissues of the body, and we found this to be the case in our study.Setbp1 was only differentially expressed between Setbp1 S858R and WT excitatory neurons, but many known targets of SETBP1 were differentially expressed and regulated in multiple cerebral cortex and kidney cell types.Here, the opposing decreased G2M enrichment we identified in atypical SGS microglia, inhibitory neurons, OPCs, and astrocytes contrasts in vitro SGS models exhibiting overproliferative SGS NPCs, with a higher fraction of the cells in the S and G2M phases.

ACK N O WLE D G E M ENTS
The authors thank the Lasseigne Lab members Amanda Clark, Anisha Haldar, Avery Williams, Emma Jones, Nathaniel DeVoss, and Victoria Flanary for their feedback throughout this study.
Because prior work in SGS NPCs has indicated SETBP1 accumulation leads to a neurodegenerative phenotype, inducing p53 inhibition and genotoxic stress accompanied by altered cell cycle and DNA damage,10 we hypothesized S858R neurons and astrocytes (cell types derived from NPCs45 ) and macrophages like microglia (previously implicated in inflammatory processes in other neurodegenerative disorders46 ) may also exhibit these molecular phenotypes.We determined the S858R oligodendrocytes, oligodendrocyte precursor F I G U R E 2 Differential expression of SETBP1 targets reveals cell-type-specific signatures of apoptosis resistance, oncogene disruption, and altered cell cycling in NPC-derived cells: Expression of Setbp1 in S858R and WT (A) cerebral cortex and (B) kidney cells.Differentially expressed known SETBP1 target genes (average log 2 FC threshold of 0.1 and p-adjusted value <0.05) in S858R and WT (C) cerebral cortex and (D) kidney cells.Teal and brown denote up and downregulated genes, respectively.(E) Split violin plot of the expression of Sox2 (a marker of proliferative NPCs) in S858R and WT cerebral cortex cell types.
itself was not differentially expressed in these cell types.In the kidneys (Figure 3B), SETBP1 serves as a repressor in WT LOH, smooth muscle cells (smcs), B cells, Collecting Duct Principal Cells (CDPC), and DCT and an activator in pericytes.On the contrary, in the S585R variant, SETBP1 had a transcriptional activator role in Distal Loop of Henle (DLH) and Proximal Convoluted Tubule Segment 2 (PCTS2) cells.Similar to gene expression, the TF activity changes for SETBP1 due to the S585R variant in the cerebral cortex and kidney are subtle.However, we see additional changes in the TF activity of transcriptional regulators targeted by SETBP1.
eral mechanisms are hypothesized to contribute to altered neurodevelopment in SGS, including chromatin remodeling, disrupted cell F I G U R E 4 Increased differential gene targeting in S858R cerebral cortex and kidney cells indicates increased gene regulation and altered cell-type-specific mechanisms associated with SGS: Dot plot of the normalized total differential gene targeting by (A) cerebral cortex and (B) kidney cell types (teal for increased gene targeting in S858R compared to WT and brown for decreased gene targeting in S858R compared to WT; dot size corresponds to gene number).Functional Enrichment Analysis (FEA) of differentially targeted SETBP1 gene set in (C) cerebral cortex and (D) kidney cell types.
In addition to rewiring involving DNA damage mechanisms mediated through Parp1/PARP1, E4f1/E4F1 and Mef2c/MEF2C, cerebral cortex pericytes also exhibited rewiring for altered cell cycling through Mxf1/ MXF1 and Pbrm1/PBRM1 with Trp53/TRP53 (the mouse ortholog for human TP53) (Figure 6C,E).While the kidney had fewer SETBP1 targets involved in altered mechanisms of cooperation and regulation, more cell types were disrupted.Macrophages, CDIC type B cells, pericytes, B cells, LOH, and DCT all had rewiring involving APEX1 and TRP53 in S858R cells (Figure 6D,F).However, fibroblasts only had regulatory changes involving DNA damage rewiring, represented by a loss of Mef2c APEX1 cooperation in S858R compared to WT. CDPC and DLH exhibited SETBP1-Mzf1/MZF1-TRP53 mediated cell cycle rewiring through gained cooperation between MZF1 and TRP53 in the S858R mice compared to WT (Figure 6D,F).Since cooperation While previous research primarily focused on the role of NPCs in driving SGS neurodegeneration through aberrant proliferation, deregulated oncogenes and tumor suppressors, unresolved DNA F I G U R E 5 Differential community analysis of SETBP1 communities reveals groups of cell types with more similar SETBP1 differential communities: Jaccard Similarity Index (JI) between all SETBP1 differential communities in (A) cerebral cortex and (B) kidney.F I G U R E 6 S858R pericytes and microglia and multiple kidney cell types exhibit altered cooperation and regulation in proposed cell cycle and DNA damage mechanisms in SGS: Visualization of changes in regulation and cooperation of cell types in S858R cerebral cortex and kidney.(A) Schematic of proposed increased DNA damage and increased cell cycle mechanisms in the context of SGS.(B) Classification of altered regulation and cooperation through a change in network scores between TF SETBP1 (yellow) and its target genes from the gene set (light grey) associated with the corresponding target protein and proposed mechanisms of altered neurodevelopment in SGS mediated through Apex1 or Trp53 (dark grey).Heatmap showing the sign change in regulation and cooperation between conditions in the cerebral cortex (C) and kidney (D).Dot plot showing the magnitude of regulatory (direction and size of triangle) and cooperativity (teal to brown) network scores in the cerebral cortex (E) and kidney (F).
Banfi et al. 2021, previously showed that SGS inhibitory neurons inherited DNA damage and were prone to degeneration that could be alleviated by inhibiting PARP-1.10Our rewiring analyses indicated that Setbp1 S858R pericytes and microglia, in addition to inhibitory neurons, had SETBP1 targets that exhibited altered cooperation with APEX1, potentially identifying additional cell types that may benefit from PARP1 inhibition in SGS.10 Cortical pericytes were the only cell type involving TRP53-associated cell cycling mediated through MZF1 and PBRM1.They exhibited the most rewiring across all mechanisms in the cerebral cortex, further emphasizing a potential role for pericytes in atypical SGS.While more cell types involve DNA damage mechanisms mediated through altered cooperation with APEX1 in the kidney, less is known about how this may contribute to a hydronephrosis phenotype.Further functional studies investigating the role of cell-type-specific DNA damage and altered cell cycling would be useful for understanding both the SGS mechanism and treatment opportunities.There are several limitations to the present study.For example, here we focused on tissues from 6-week-old male mice, so it remains to be seen how cell-type-specific expression and regulation change across developmental time points and in female mice.Furthermore, our TF activity was inferred based on TFs included in the CollecTRI prior.PANDA relied on information from STRING, CIS-BP, and prior SETBP1 ChIP-Seq experiments.As more comprehensive data sets become available, our analytical approaches can be expanded and refined.One major benefit to the TF-gene regulatory networks is that they provide a comprehensive understanding of the impact of a point mutation not just at the gene coexpression or protein-protein cooperation levels but facilitate in silico profiling of Setbp1 S858R mice across all three facets of the central dogma of biology.However, further experiments are needed to determine if the absence of alterations in SGS-associated phosphorylation and chromatin remodeling mechanisms in regulatory and cooperativity rewiring we found were a direct result of the S858R variant or if this was due to a lack of prior knowledge in the regulatory networks.Additional experiments are also needed to confirm the role of each cell type in SGS etiology and progression.Finally, our study was performed in mice, and while the protein domain is conserved between mouse and human, findings in mouse models do not always translate to patients.In summary, we report that classical SGS molecular signatures are present across cerebral cortex and kidney cell types in the Setbp1 S858R mouse model of atypical SGS, underscoring the multifaceted role of SETBP1 and the impact of many cell types in atypical SGS.Future work is needed to further understand how SETBP1's role changes across cell types, developmental time points, and in the context of other patient variants.Further, experiments and analyses investigating disrupted cell-cell communication between cell types (e.g.astrocytes and excitatory neurons) may reveal novel insights into the role additional cell types play in contributing to SGS neurodegeneration and clinical manifestations outside of the brain.AUTH O R CO NTR I B UTI O N S Jordan H Whitlock: Conceptualization (supporting); data curation (lead); formal analysis (lead); investigation (equal); methodology (equal); software (lead); visualization (lead); writing -original draft (lead); writing -review and editing (equal).Tabea M Soelter: Investigation (equal); software (supporting); validation (equal); writing -review and editing (supporting).Timothy C Howton: Investigation (equal); methodology (equal); project administration (supporting); validation (equal); writing -review and editing (equal).Elizabeth J Wilk: Software (supporting); writing -review and editing (supporting).Vishal H Oza: Conceptualization (supporting); formal analysis (supporting); methodology (equal); project administration (supporting); software (supporting); validation (equal); visualization (supporting); writing -review and editing (supporting).Brittany Lasseigne: Conceptualization (lead); funding acquisition (lead); methodology (equal); project administration (lead); resources (lead); supervision (lead); writing -review and editing (equal).

Furthermore, they
thank Dr. Anna Thalacker-Mercer's lab at UAB for providing equipment and reagents for the ELISA and BCA experiments and Dalton Patterson for his advice for these experiments.In addition, we thank Shanrun Liu at the UAB CFCC (supported by the Center for AIDS Research grant AI027767 and The O'Neal Comprehensive Cancer Center grant CA013148) for his expertise and preparation of snRNA-seq libraries for this study as well as Dr. Michael Crowley and the team at the UAB Heflin Center for Genomic Sciences at the UAB Sequencing Core