Histone serotonylation regulates ependymoma tumorigenesis

Abstract

Bidirectional communication between tumors and neurons has emerged as a key facet of the tumor microenvironment that drives malignancy^1,2. Another hallmark feature of cancer is epigenomic dysregulation, where alterations in gene expression influences cell states and interactions with the tumor microenvironment³. Ependymoma (EPN) is a pediatric brain tumor that relies on epigenomic remodeling to engender malignancy^4,5; how these epigenetic mechanisms intersect with extrinsic neuronal signaling during EPN tumor progression is unknown. Here we show that activity of serotonergic neurons regulates EPN tumorigenesis, while serotonin itself also serves as an activating modification on histones. We found that inhibition of histone serotonylation blocks EPN tumorigenesis and regulates expression of a core set of developmental transcription factors (TFs). High-throughput, in vivo screening of these TFs revealed that ETV5 promotes EPN tumorigenesis and functions by enhancing repressive chromatin states. Neuropeptide Y (NPY) is amongst the genes repressed by ETV5 and its overexpression suppresses EPN tumor progression and tumor-associated network hyperactivity via synaptic remodeling. Collectively, these studies identify histone serotonylation as a key driver of EPN tumorigenesis, while further revealing how neuronal signaling, neuro-epigenomics, and developmental programs are intertwined to drive malignancy in brain cancer.

Ependymomas (EPN) are aggressive pediatric brain tumors that are chemotherapy-resistant and lack targeted therapies, leading to poor survival and neurocognitive outcomes⁶. EPN can be divided into at least nine molecular subtypes associated with distinct genetic and epigenetic alterations. Approximately 70% of supratentorial EPNs (ST-EPN) are characterized by a gene fusion between ZFTA and RELA (ZFTA-RELA fusion, herein ZR^FUS) ^7,8. Among ST-EPN, those bearing ZR^FUS are particularly aggressive, reflected by decreased progression free survival compared to other forms of ST-EPN driven by YAP1 fusions ⁷. Pediatric brain tumors exhibit an exquisite dependency on epigenomic mechanisms, such as those regulated by core transcriptional regulatory proteins^4,5,9,10. Recent studies indicate that ZR^FUS functions as an aberrant transcription factor that recruits chromatin modifiers to activate oncogenic transcriptional programs linked to neural development^11-14. Despite detection of these global alterations, the precise mechanisms driving aberrant epigenomic states in ZR^FUS EPN are obscure, and likely multi-factorial encompassing both cell intrinsic and extrinsic factors. Separate investigations demonstrated that the neuromodulator serotonin can be integrated into histones and serve as an epigenetic mark regulating gene expression in both neurons and astrocytes, ultimately controlling circuit-level activities in the brain^15,16. Together, these observations raise the possibility that neurons can remodel the tumor epigenome through release of neuroactive compounds that are deposited on histones. Furthermore, recent studies highlighted bidirectional signaling between brain tumors and neurons, identifying paracrine factors, direct synaptic integration, and circuit-specific infiltration as key drivers of tumorigenesis^17-25. Nevertheless, how neuronal activity influences tumor growth in EPN is incompletely understood and whether neuromodulators serve as epigenetic regulators of brain tumor progression is unknown.

Serotonergic neurons suppress EPN growth

To understand how the ZR^FUS fusion promotes EPN tumorigenesis we compared patient transcriptomic profiles of ZR^FUS driven tumors with those lacking the fusion¹⁰. This comparison revealed selective enrichment of gene ontology (GO) sets involved in neuronal function, synaptic organization, and neurotransmission in EPN tumors containing ZR^FUS (Fig.1a; Supplementary Table 1). These pathways are enriched at the level of RNA-sequencing (RNA-Seq), and supported by the accumulation of active H3K27ac chromatin marks, which we previously demonstrated to be enriched in ion channels and neurotransmitter pathways¹⁰.

Figure 1. Stimulation of serotonergic neurons suppresses EPN tumorigenesis.

a. GO-term analysis of up-regulated genes (log₂(fold change(FC)) >=1, P < 0.05) in ZR^FUS patients using GO Project datasets from Mouse Genome Informatics (MGI).

b. Immunofluorescence staining of SLC6A4 in PFA and ZR^FUS EPN patient samples (scale bar=25 μm).

c. Schematic of DREADD-hM3Dq activation of ipsilateral excitatory neurons in ZR^FUS EPN mice.

d. Low magnification image of EPN tumors and representative BrdU staining of EPN tumors after DREADD-based activation of ipsilateral excitatory neurons via CNO (scale bar=50 μm).

e. Quantification of BrdU staining in saline versus CNO treated EPN tumors (Saline: n=3, CNO: n=3, AAV_Saline: n=4, AAV_CNO: n=4, mean±SEM, unpaired Student’s two-sided t test, *P_{SalinevsAAV_CNO} = 0.0289, *P_{CNOvsAAV_CNO} = 0.0212, *P_{AAV_SalinevsAAV_CNO} = 0.0132).

f. Schematic of DREADD-hM3Dq activation of dRN neurons in ZR^FUS EPN mice.

g. Low magnification image of EPN tumors and representative BrdU staining of EPN tumors with DREADD-based activation of dRN neurons via CNO (scale bar=50 μm).

h. Quantification of BrdU staining in saline versus CNO treated EPN tumors (Saline: n=3, CNO: n=3, AAV_saline: n=4, AAV_CNO: n=3, mean±SEM, unpaired Student’s t two-sided test, **P_{SalinevsAAV_CNO} = 0.0024, ****P_{CNOvsAAV_CNO} < 0.0001, ****P_{AAV_SalinevsAAV_CNO} = 6.86E-05).

i. Schematic of DREADD-hM4Di inhibition of dRN neurons in ZR^FUS EPN mice.

j. Low magnification image of EPN tumors and representative BrdU staining of EPN tumors with DREADD-based inhibition of dRN neurons via CNO (scale bar=50 μm).

k. Quantification of BrdU staining in saline versus CNO treated EPN tumors (n=3 per group, mean±SEM, unpaired Student’s two-sided t test, **P_{SalinevsAAV_CNO} = 0.01, **P_{CNOvsAAV_CNO} = 0.0048, *P_{AAV_SalinevsAAV_CNO} = 0.0133).

Panel c, f, i were created using Biorender.com

These findings implicate aberrant neuronal activity in the brain microenvironment in EPN tumorigenesis, a phenotype that has not been functionally evaluated in this disease. To test this hypothesis, we leveraged our in utero electroporation (IUE) model of ZR^FUS EPN to conduct DREADD-based activation of cortical neurons in the hemisphere ipsilateral to the tumor^11,19,26. Accordingly, we generated ZR^FUS tumors and then injected adeno-associated virus (AAV) 2/9 syn1-hM3Dq-mCherry at P5, stimulating neuronal activity by treating mice with either saline or clozapine N-oxide (CNO) two times per day for 10 days, starting at P20 (Extended Data Figure.1a). To confirm activation of neurons, we used c-Fos staining in neurons as a proxy for increased neuronal activity and confirmed increases in neurons at the peritumoral margin in CNO-treated animals (Extended Data Fig.1h-i). Next, using incorporation of Bromodeoxyuridine (BrdU) as an index of cellular proliferation we found an increase in BrdU staining in tumors treated with CNO (Extended Data Figure 1b-c,g), which is coupled with a gross expansion in tumor size (Extended Data Figure 1b-c). These findings indicate that activation of cortical neurons may promote EPN tumor proliferation. To further dissect neuronal subtypes that contribute to activity dependent tumorigenesis, we used similar experimental paradigms, instead selectively targeting excitatory- (AAV-CaMKIIa promoter) or inhibitory- (AAV-Dlx5/6) neurons (Fig.1c; Extended Data Figure 1d). Activation of excitatory neurons promoted tumor proliferation and gross expansion (Fig.1d-e), while activation of inhibitory neurons had no effect on tumor growth (Extended Data Figure 1e-f). Together, these data indicate that activation of cortical excitatory neurons promotes ZR^FUS EPN tumorigenesis.

Further analysis of the neurotransmitter profiles enriched in ZR^FUS EPN highlighted the expression of monoamine, dopamine, and catecholamine transporters (Extended Data Fig.2a-e; Supplementary Table 1). To validate these findings, we stained human and mouse ZR^FUS EPN tumors and identified elevated expression of the serotonin transporter Slc6a4, but not dopamine transporter Slc6a3 (Fig.1b; Extended Data Fig.2f-g). These observations suggest that levels of serotonin in the brain microenvironment may influence EPN tumorigenesis. Therefore, to manipulate the levels of serotonin in the brain, we used DREADD-based approaches to activate serotonergic neurons in the dorsal raphe nucleus (dRN), which are the predominant source of serotoninergic input to the cortex and are in a brain region remote to the primary tumor^27,28. Similar to the above studies, we used IUE to generate tumors, injected AAV-2/9 syn1-hM3Dq-mCherry in the dRN and stimulated neuronal activity by treating mice with CNO (or saline control) two times per day for 10 days, followed by BrdU injection prior to collection (Fig.1f). Due to anatomical constraints of the perinatal mouse brain our dRN injections occurred at P28, followed by CNO-based stimulation from P35-P46 (Fig.1f). Using c-Fos expression as a proxy for neuronal activity, we confirmed CNO-induced increases in c-Fos expression in serotonergic neurons residing in the dRN (Extended Data Fig.1j-I). Analysis of BrdU staining revealed a drastic decrease in cell proliferation in CNO treated tumors (Fig.1g-h), accompanied by a decrease in gross tumor volume (Fig.1g). Next, we performed the complimentary experiment, where we injected the dRN with AAV-2/9 syn1-hM4Di-mCherry which enabled us to inhibit neuronal activity in the dRN (Fig.1i). Analysis of ZR^FUS EPN tumors revealed an increase in proliferation, coupled with gross expansion of the tumor after inhibition of dRN neurons (Fig.1j-k). To understand how these manipulations of the remote dRN influence local neurons in the cortex, we stained the peritumoral neurons with synaptic markers, and did not observe any gross changes in the synaptic constituency of the peritumoral margin after manipulation of dRN neurons (Extended Data Figure 3a-h). Together, these unexpected findings demonstrate that activity of dRN, serotonergic neurons suppresses EPN tumor progression. Collectively, these observations reveal circuit-specific effects on EPN tumorigenesis, where broad stimulation of cortical neurons promotes growth, while stimulation of serotonergic neurons in the dRN suppresses growth.

Histone serotonylation governs EPN

The forgoing observations implicate serotonergic signaling in EPN tumorigenesis and highlight the tumor suppressive effects driven by serotonergic neurons. Therefore, we next asked how EPN cells respond to and process serotonin released from the brain microenvironment. Prior studies have shown that serotonin can be added directly to histones and serves as an epigenomic regulator of gene expression in neurons and astrocytes^15,16. Given that epigenomic dysregulation plays a central role in EPN tumorigenesis, we next examined whether histone serotonylation occurs in our ZR^FUS EPN model by staining these tumors with antibodies specific for the serotonylated histone mark on histone H3 (H3K4me3Q5ser, herein referred to as H3-5HT). Immunostaining showed abundant expression of H3-5HT in the nucleus of both human ZR^FUS EPN tumors and posterior fossa A (PFA, non-ZR^FUS) EPN tumors, indicating that this histone mark is present in both forms of EPN. (Fig.2a). Analysis of mouse ZR^FUS EPN tumors also demonstrated H3-5HT expression (Fig.2b), while immunoblotting revealed a significant increase in H3-5HT tumors compared to normal, non-malignant cortical tissue (Fig.2c). Moreover, manipulation of dRN neurons led to alterations in H3-5HT in ZR^FUS EPN tumors (Extended Data Figure 3i-I). These observations led us to investigate whether histone serotonylation contributes to EPN tumorigenesis. To test the role of histone serotonylation, we overexpressed a dominant negative form of H3.3 (H3.3-Q5A) that blocks H3Q5Ser in our ZR^FUS EPN model¹⁵ (Fig.2d). Strikingly, overexpression of H3.3-Q5A suppressed tumor formation, with only 4/17 mice forming tumors, compared to 9/10 of wild type H3.3 control expressing tumors (Fig.2e). In the few H3.3-Q5A expression EPN tumors that were generated, we observed a reduction in H3-5HT levels, coupled with decreased proliferation (Extended Data Figure 4a-d). These data demonstrate that serotonin has potentially disparate effects on EPN tumorigenesis, in contrast to the microenvironment where it suppresses growth (Fig.1f-k), while blockade of histone serotonylation in neoplastic cells significantly abrogates EPN tumorigenesis.

Figure 2. Histone serotonylation is required for EPN tumorigenesis.

a. Immunofluorescence staining of H3-5HT in PFA and ZR^FUS EPN patient samples (scale bar=50 μm).

b. Immunofluorescence staining of 5HT and H3-5HT in mouse EPN tumors (scale bar=25 μm).

c. Immunoblots of histone serotonylation marks in mouse non-tumor cortex and EPN tumors (n=3 per group).

d. Schematic of mouse EPN tumors expressing wild-type (H3.3) and dominant negative form (H3.3-Q5A) of histone variants.

e. Representative low magnification image of H3.3 wild-type (P40) and H3.3-Q5A (P160) tumor (scale bar=50 μm), Kaplan–Meier survival curve of EPN H3.3 wild-type (n=10, median=80.5 days) and H3.3-Q5A (n=17, median=undefined, log-rank test, ***P = 0.0008), and table of tumor bearing mice vs all mice.

f. ChIP-seq heatmap profiles demonstrating co-occupancy between ZR^FUS-HA, H3-5HT, and H3K27ac in mouse EPN tumors.

g. ZR^FUS-HA, H3-5HT and H3K27ac ChIP-seq peaks at Ccnd1 locus.

h. Significant TF motif (P < 0.05, cumulative binomial distribution test in HOMER software suite) of the genes annotated with ZR^FUS-HA and H3-5HT peaks in mouse tumors.

i. Venn diagram depicting core TFs annotated with H3-5HT peaks in mouse tumors and core TFs identified in ZR^FUS patients from previous study.

j. Representative core TF Etv5 locus with ZR^FUS-HA, H3-5HT and H3K27ac ChIP-seq peaks.

Panel k was created using Biorender.com

The robust effect of H3.3-Q5A on EPN tumorigenesis, led us to next examine whether attenuating serotonin transport also impacted tumorigenesis and H3-5HT status. Focusing on Slc6a4, as this serotonin transporter exhibits expression in EPN tumors (Fig.1b), we used a CRISPR-Cas9 strategy to knock it out in our ZR^FUS EPN model. These studies revealed a modest extension in overall survival, coupled with a decrease in cell proliferation and reduced levels of H3-5HT in tumors (Extended Data Figure 5a-e). Next, we acutely treated mice bearing ZR^FUS EPN tumors with the selective serotonin reuptake inhibitor (SSRI) sertraline, also finding a decrease in H3-5HT levels in tumors (Extended Data Figure 5h-j). These studies suggest that H3-5HT in EPN is mediated by transport from external sources, which is corroborated by our observations that key enzymes involved in serotonin synthesis are not expressed in EPN (Extended Data Figure 5f-g). Together, these studies indicate that manipulating serotonin transport into EPN tumors directly impacts the levels of H3-5HT.

Because H3-5HT serves as an active histone mark that drives gene expression, we next used chromatin immunoprecipitation sequencing (ChIP-Seq) to evaluate its distribution throughout the genome in ZR^FUS EPN tumors. We found that H3-5HT enrichment was closely aligned with H3K27 acetylation (Fig.2f-g), an active transcriptional mark, and bound at sites co-localized with ZR^FUS protein (Fig.2f-g). To identify intersectional regulatory networks between H3-5HT and ZR^FUS we focused on genes with adjacent peaks and performed transcription factor (TF) motif analysis which identified a host of developmental TFs as regulating their expression (Fig.2h;Supplementary Table 2). Comparing TFs associated with H3-5HT peaks and core developmental TFs previously identified in ZR^FUS EPN patients, prioritized 12 TFs conserved between datasets (Fig.2i-j; Extended Data Fig.6a; Supplementary Table 3) ¹⁰. Collectively, these results indicate that H3-5HT plays an essential role in EPN tumorigenesis and is tightly associated with gene regulatory programs established by the ZR^FUS protein.

ETV5 regulates EPN tumor progression

The significant role of H3-5HT in EPN tumorigenesis and its regulation of developmental TFs, led us to further investigate their contribution to tumorigenesis. To examine the functional impact of these core developmental TFs upon EPN tumorigenesis, we performed a barcoded overexpression screen by generating a PiggyBac-based, barcoded library containing 38 of these developmental TFs (Fig.3a; Supplementary Table 4) ^10,23. Following introduction of the TF library into our ZR^FUS EPN model, we harvested tumor-bearing mice at P70 and performed barcode sequencing, using barcode enrichment as a proxy for relative abundance within the tumor. Our screen identified LIM homeobox 2 (LHX2), LIM homeobox 4 (LHX4), ETS variant transcription factor 5 (ETV5), and KLF transcription factor 12 (KLF12) as enriched in EPN tumors. To validate the role of these nominated candidates in EPN tumorigenesis, we individually overexpressed them in our ZR^FUS EPN model and used overall survival as a surrogate for tumorigenesis. We found that ETV5 overexpression decreased overall survival and enhanced proliferation, as measured by BrdU incorporation (Fig.3b; Extended Data Fig.7a,d-e, g-h), while LHX2, LHX4, or KLF12 overexpression did not impact overall survival (Extended Data Fig.7a-e; Extended Data Figure 8). Next, we used a CRISPR-Cas9 strategy to knockout ETV5 from ZR^FUS EPN tumors and found a complementary increase in overall survival and decrease in proliferation (Fig.3b; Extended Data Fig.7d-h). These complementary gain-of-function (ETV5-GOF) and loss-of-function (ETV5-LOF) experiments implicate ETV5 as an important transcriptional regulator of EPN tumorigenesis that we prioritized for further downstream analysis. Moreover, the ETV5 locus exhibits local H3-5HT and ZR^FUS binding sites (Fig.2j) and its expression is reduced in H3.3-Q5A expressing tumors (Extended Data Figure 4e-f), reinforcing its regulation by histone serotonylation and ZR^FUS.

Figure 3. ETV5 regulates EPN progression and repressive chromatin states.

a. Schematic of in vivo screening and barcode enrichment from mouse EPN tumors (n=5, unpaired two-sided Student’s t test).

b. Kaplan–Meier survival curve of EPN control (n=51, median=70 days), ETV5-GOF (n=37, median=59 days, log-rank test, *P = 0.0167), and Etv5-LOF (n=40, median=92 days, log-rank test, *P = 0.0191).

c. Comparison of H3K27ac and H3K27me3 ChIP-seq heatmap profiles between control versus ETV5-GOF tumors. TSS: transcription start site, TES: transcription end site.

d. Schematic of ETV5 IP-MS in mouse non-tumor cortex and EPN tumors (n=3 per group).

e. Volcano plot depicting ETV5 interactome in mouse EPN tumors (log₂FC >=1, P < 0.05, two-sided Wald test, fold change compared to control samples).

f. Venn diagram depicting ETV5 binding partners in mouse non-tumor cortex and EPN tumors.

g. Left panel: Immunoblots of ZR^FUS-HA, Etv5, Hdac1, and Cbx3 from mouse non-tumor cortex and EPN tumors. Right panel: Immunoprecipitation of Cbx3 and immunoblot of Etv5, Hdac1, and Cbx3 in mouse non-tumor cortex and EPN tumors. Arrowhead labels the protein of interest.

h. Volcano plot of RNA-seq analysis from ETV5-GOF tumors versus control (n=3 per group, two-sided Wald test, log₂FC >=1 or =< −1, P < 0.05).

i. Venn diagram depicting downregulated DEGs acquiring H3K27me3 peaks in ETV5-GOF tumors.

j. GO-terms analysis of the overlapping genes from i. performed using datasets from Enrichr (two-sided Fisher’s exact test).

k. RT-qPCR fold-enrichment of NPY transcript (ddCT) in human normal brain and supratentorial EPN tissues (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, ****P = 2.98E-06).

l. Immunofluorescence staining of NPY in control versus ETV5-GOF tumors (scale bar=50 μm).

m. Quantification of NPY in control versus ETV5-GOF tumors (n=4 per group, mean±SEM, unpaired two-sided Student’s t test. **P = 0.0031).

Panel a, d were created using Biorender.com

The role of ETV5 in EPN tumorigenesis has not been investigated, therefore we next examined the mechanisms by which it regulates gene expression and chromatin states. Because epigenetic dysregulation plays a central role in EPN progression, we assessed active (H3K27ac) and repressive (H3K27me3) chromatin marks via ChIP-Seq in ETV5-GOF tumors. These studies revealed a reduction in H3K27ac peaks (52,658 peaks in control and 27,197 peaks in ETV5-GOF), coupled with an increase in H3K27me3 peaks (227 peaks in control and 1343 peaks in ETV5-GOF) in ETV5-GOF EPN tumors (Fig.3c; Extended Data 7j; Supplementary Tables 5-8), while immunoblotting demonstrated no differences in the overall levels of these marks (Extended Data Fig.7i). To understand how ETV5 regulates these epigenomic states, we performed immunoprecipitation followed by mass spectrometry (IP-MS) on ZR^FUS EPN and normal brain, identifying 297 proteins that specifically interact with ETV5 in EPN tumors (Fig.3d-f; Extended Data Fig.7k; Supplementary Tables 9,10). A subset of these proteins have established roles in regulating polycomb repressive complex (PRC) and using co-immunoprecipitation (co-IP) we validated interactions between ETV5 and chromobox 3 (CBX3) and histone deacetylase 1 (HDAC1) specifically in ZR^FUS EPN (Fig.3g). Collectively, these data indicate that ETV5 promotes a repressive H3K27 tri-methylation signature in ZR^FUS EPN tumors, through recruitment of PRC-associated proteins.

ETV5 suppresses NPY gene expression

Having established that ETV5 can remodel the epigenome in ZR^FUS EPN, we next sought to identify downstream effector genes that mediate its role in tumorigenesis. We performed RNA-Seq on ETV5-GOF tumors (and controls) and found that a vast majority of the differentially expressed genes (DEGs) in the ETV5-GOF tumors were downregulated (85.7%, 258 total) (Fig.3h; Supplementary Table 11), corroborating our observations that ETV5 increases epigenomic repressive states. Next, we integrated these RNA-Seq data with the H3K27me3 ChIP-Seq data and found a subset of genes that are both transcriptionally repressed and acquire H3K27me3 modification when ETV5 is overexpressed (Fig.3i; Supplementary Table 12). GO analysis of this subset revealed an association with chemical synaptic transmission and neuropeptide signaling pathways (Fig.3j; Supplementary Tables 12,13). Furthermore, cross comparison of these gene lists with ChIP-Seq data from ETV5-GOF tumors, further resolved this list of genes (Extended Data Figure 6b; Supplementary Tables 11,14). Focusing on the neuropeptide gene sets, we validated the downregulation of Neuropeptide Y (NPY) in human ZR^FUS EPN compared to normal brain and used immunostaining to demonstrate that it is downregulated in mouse ETV5-GOF tumors (Fig.3k-m; Extended Data Fig.7l). This downregulation of NPY in ETV5-GOF tumors is complemented by increased H3K27me3 repressive marks at the NPY locus (Extended Data Fig.7m) implicating the PRC complex as an important regulator of NPY expression.

NPY suppresses brain hyperactivity

NPY is a potent modulator of neuronal activity in the brain and also has separate functions in a host of other tissues outside the central nervous system (CNS) ^29-34. NPY has also been implicated in a host of malignancies, however its role in EPN remains undefined^35,36. Because ETV5 promotes tumor progression and NPY is downregulated in ETV5-GOF tumors, we hypothesized that its overexpression would suppress EPN tumor growth. Indeed, NPY-GOF in our ZR^FUS EPN model extended overall survival and suppressed cell proliferation, as assayed by BrdU incorporation (Fig.4a-c; Extended Data Figure 8; Extended Data Fig.9a). To understand how NPY suppresses tumorigenesis we performed RNA-Seq on NPY-GOF tumors (and control), and similar to ETV5, we found that a vast majority of the DEGs (89%, 995 total) were downregulated (Fig.4d). GO analysis identified synaptic development and synapse signaling as the top gene sets downregulated in NPY overexpressing tumors (Extended Data Fig 9e; Supplemental Table 15,16). These results suggest that NPY release from EPN tumors influences the synaptic constituency of the microenvironment. Consistent with this notion, immunostaining of the peritumoral margin revealed nominal expression of NPY receptor 2 (NPY2R) in mouse EPN tumors and high expression in neurons surrounding the tumor (Extended Data Fig.9b-d).

Figure 4. NPY suppresses EPN progression.

a. Kaplan–Meier survival curve of EPN control (n=11, median=70 days) and NPY-GOF (n=15, median=95 days, log-rank test, *P = 0.0174).

b. Representative BrdU staining of control versus NPY-GOF tumors (scale bar=50 μm).

c. Quantification of BrdU staining in control versus NPY-GOF tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, ***P = 0.0008).

d. Volcano plot of RNA-seq analysis from NPY-GOF tumors versus control (n=3 per group, two-sided Wald test, log₂FC >=1 or =< −1, P < 0.05).

e. Low magnification view of tumor margin and representative higher magnification images (derived from dashed box) of peri-tumoral inhibitory synaptic staining in control versus NPY-GOF tumors (scale bar=25 μm). Vgat: vesicular GABA transporter.

f. Quantification of inhibitory synaptic staining in control versus NPY-GOF tumors (n=5 per group, mean±SEM, two-sided Wilcoxon rank sum test, P = 0.0556).

g. Low magnification image of the tumor margin and representative higher magnification images (derived from dashed box) of peri-tumoral excitatory synaptic staining in control versus NPY-GOF tumors (scale bar=25 μm). Vglut1: Vesicular glutamate transporter 1; Psd95: postsynaptic density protein 95.

h. Quantification of excitatory synaptic staining in control versus NPY-GOF tumors (control: n=6, NPY-GOF: n=5, mean±SEM, two-sided Wilcoxon rank sum test, *P = 0.0173).

The forgoing observations suggest that NPY exerts its effect on EPN tumor growth via paracrine interactions with neurons in the microenvironment, prompting us to assess excitatory and inhibitory synapses at the peritumoral margin in NPY-GOF EPN tumors. These studies showed a decrease in excitatory synapses (Vglut1/PSD95) and an increase in inhibitory synapses (VGAT) at the peritumoral margins in NPY-GOF tumors (Fig.4e-h). The synaptic constituency at the peritumoral margin can influence brain hyperactivity and tumor progression. Our observations of synaptic remodeling towards decreased excitatory synapses, coupled with prior studies demonstrating that NPY suppresses brain hyperactivity³¹, led us to examine whether NPY overexpression in EPN suppresses progressive brain hyperactivity. We performed serial electroencephalography (EEG) starting at P30 and recording for 48 hours, every 10 days (Fig.5b). Mice bearing ZR^FUS EPN control tumors demonstrated increased brain network hyperactivity as early as P32, which led to 50% of mice exhibiting observable seizures by P42 (Fig.5i-k; Extended Data Fig 9f). In contrast, across the P30-P52 recording interval, mice bearing NPY-GOF EPN tumors did not exhibit seizures and had a significant delay in the onset of network hyperactivity (Fig.5i-k). To confirm that NPY overexpression leads to increased inhibitory synaptic activity, we generated NPY-GOF tumors, injected AAV 2/9 syn1-mCherry to label peritumoral neurons and performed whole cell recordings on labeled cortical neurons from tumor bearing brain slices (Fig.5a). These studies revealed an increase in the frequency of inhibitory post-synaptic currents (ISPC), coupled with a decrease in frequency of excitatory post-synaptic current (EPSC) (Fig.5c-h), indicating an overall increase in inhibitory activity in the brains of mice bearing NPY-GOF tumors. Together, these data demonstrate that NPY suppresses EPN tumor progression by remodeling the brain microenvironment towards synaptic inhibition, which in turn blunts brain hyperactivity, and impedes tumor progression.

Figure 5. NPY suppresses EPN-induced brain hyperactivity.

a. Schematic of electrophysiology recording in mCherry-labeled neurons around ZR^FUS EPN tumor.

b. Schematic of EEG recording in ZR^FUS EPN mice.

c. Traces of sEPSC recording in control and NPY-GOF mice

d. Amplitude of sEPSC recording in control and NPY-GOF mice (n=4 mice per group, mean±SEM, unpaired two-sided Student’s t test, P = 0.8164).

e. Frequency of sEPSC recording in control and NPY-GOF mice (n=4 mice per group, mean±SEM, two-sided Wilcoxon rank sum test, *P = 0.0286).

f. Traces of sIPSC recording in control and NPY-GOF mice

g. Amplitude of sIPSC recording in control and NPY-GOF mice (n=4 mice per group, mean±SEM, unpaired two-sided Student’s t test, P = 0.9893).

h. Frequency of sIPSC recording in control and NPY-GOF mice (n=4 mice per group, mean±SEM, unpaired two-sided Student’s t test, *P = 0.0377).

i. Seizure incidence curves across EEG recording sessions (control: n=8, NPY-GOF: n=6).

j. Quantification of spikes number per hour over 48 hours period at 10-day intervals from P30 to P62 (data were derived from at least 3 mice in each group and are presented in violin plot with all individual data points, unpaired two-sided Student’s t test, P32: *P = 0.0290, P42: P = 0.7582, P52: P = 0.7903).

k. Representative EEG traces from mice bearing control and NPY-GOF tumors.

l. Model figure

Panel a, b, l were created using Biorender.com

Discussion

Investigating neuron-tumor interactions in autochthonous models of ZR^FUS driven EPN, we found that activation of specific neuronal subpopulations differentially impacts tumor progression: local cortical neurons drive progression, while serotonergic neurons in the dRN suppress progression (Fig.5l). Prior studies identified cortical projection neurons, glutamatergic signaling, sensory input, and remote areas with high-activity as driving glioma progression, supporting the notion that heightened neuronal activity is pro-tumorigenic^17-25. In contrast, our findings indicate that activation of discrete subtypes of neurons suppresses EPN tumorigenesis, highlighting the need to decipher how neuronal subtype- and circuit- specific interactions impact brain tumor progression. While our results suggest that serotonin from dRN neurons is directly transported into EPN cells, it remains possible that these effects are mediated through subsets of cortical interneurons that express the serotonin receptor Htr3a ^37-40 (Fig.5l). Mechanistically, we found that serotonin serves as an essential histone modification in EPN, associating with oncogenic transcriptional networks (Fig.5l). Given that histone serotonylation occurs in astrocytes and neurons^15,16, it is possible that this phenomenon is operant in other types of brain tumors.

Examining how neuronal signaling influences epigenomic states in EPN, we found that the developmental TF ETV5 is a target of histone serotonylation and key regulator of EPN tumorigenesis (Fig.5l). ETV5 plays a central role in astrocyte development⁴¹, while interactions between progenitor populations and neurons play a key role in brain development^42-45. Consequently, these relationships may influence other epigenomic- and developmental- states manifest in EPN. Moreover, we found that ETV5 promotes repressive chromatin states, implicating gene repression as a driver of EPN progression. Among the repressed ETV5 targets is NPY, which functions to suppress EPN progression and brain hyperactivity through paracrine signaling with neurons (Fig.5l). However, we cannot rule out a possible autocrine mechanism, as NPY functions in this manner in other systems^46-48. Brain tumors release factors that remodel synapses towards hyperactivity²³, whereas we found that EPN tumors can release factors that suppress excitatory synaptic remodeling and that repressing this mechanism is essential for progression. Collectively, these studies highlight the intersection between neurodevelopment, chromatin regulation, and neuronal signaling, reinforcing the importance of defining specific interactions between neuroactive compounds, circuits, and brain tumors.

Methods

Human RNA-seq data analysis

The DEGs form human EPN samples were obtained from our previous study¹⁰. To find genes enriched in ZR^FUS versus non-ZR^FUS EPN, GO term enrichment was performed using the R package Clusterprofiler⁵¹. For further detailed analysis of monoamine transport related genes, the gene set corresponding to the GO term was obtained from GO Project datasets from Mouse Genome Informatics (MGI).

EPN Patient sample acquisition

EPN samples were obtained from patients undergoing surgical resection at LeBonheur Children’s Hospital and St. Jude Children Hospital. Informed consent was obtained from all patients with approval for human tissue use from local IRB review. Diagnoses were confirmed by a neuropathologist. For histology, samples were paraffin embedded.

In utero electroporation (IUE) mouse ZR^FUS EPN model

To generate mouse ZR^FUS EPN tumors, we performed IUE in the CD-1 IGS mouse background at embryonic day 16.5 as previously described^11,52. Briefly, the uterine horns of the time pregnancy mice were exposed for injecting plasmid mixture into the embryonic lateral ventricles with Fast Green dye as the indicator, followed by electroporation using BTW Tweezertrodes connected to the pulse generator (BTX 8300; parameters set at 33 V and 55 ms per pulse six times at 100 ms intervals). The plasmid mixture included helper plasmid pGLAST-PBase (2.0 μg/μL), PBCAG-GFP (1.0 μg/μL), PBCAG-ZR^FUS-HA (1.0 μg/μL), and CRISPR-Cas9 construct with gRNA against Trp53 (1.5 μg/μL). Co-electroporation of PBCAG-GFP allowed fluorescent visualization of tumors. GOF studies were performed by co-electroporating human variant gene of interest in PBCAG constructs (1.0 μg/μL). LOF studies utilized CRISPR-Cas9 construct with Trp53 gRNA and gRNA against gene of interest (1.5 μg/μL). The knockout efficiency was validated by the mismatch-cleavage SURVEYOR assay (IDT, 706020) on genomic DNA acquired from mouse tumors. Primer sequences for SURVEYOR assay and functional studies validation are listed in Supplementary Table 17. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at Baylor College of Medicine and conform to the US Public Health Service Policy on Human Care and Use of Laboratory Animals.

AAV generation, AAV-DREADD delivery, and CNO treatment

All AAVs are generated at the Neuroconnectivity Core at BCM and the detailed information are included in Supplementary Table 17 AAV was diluted with loading dye (10x Fast Green, 2 mg/ml) and loaded into a microdispenser (Drummond Scientific, no.13-681-460) before injection. For ipsilateral cortical neural stimulation, AAV was injected at a rate of 7 nl/s to give a total volume of 1.5 μl, into the ventricle of P5 mice. For dRN neural stimulation experiment, the dRN was injected with 750 nL of AAV at a rate of 200 nL/minute (injection coordinates: −4.5 mm A/P, 0 mm M/L, and −2.2 mm ventral⁵³) and the micropipette was left in place for 5 min after injection before being slowly removed. Mice were weighed daily before injecting with CNO (Tocris, 4936) at 0.5 mg/kg or 5 mg/kg intraperitoneally, or an equivalent volume of saline for controls. Mice received CNO or saline injection twice per day before collecting tissues for further analyses.

Animal Survival Analyses, Bromodeoxyuridine labeling, and Mouse Tumor Collection

Animals were monitored for suggestive symptoms of tumors, including lethargy, hunched posture, decreased grooming, trembling, partial limb paralysis, abnormal gait, and hydrocephalus, representing the IACUC permitted end point. Male and female mice were both included in this study. Mice were humanely euthanized once the disease symptoms showed. Survival dates were recorded for Kaplan-Meier Survival curve analyses. For BrdU labeling, mouse brains at postnatal day P70 were harvested 4 hrs after BrdU pulsing (100 mg/kg body weight in phosphate-buffered saline [PBS]) by intraperitoneal injection. For biochemical and molecular studies, tumors tissues were dissected using GFP before further processing in other experiments described as below.

Immunofluorescence staining

For paraffin sectioning, mouse brains were fixed through intracardial perfusion and overnight incubation with 4% paraformaldehyde. After dehydration using 70% EtOH, samples were sent to Pathology Core of Breast Center in Baylor College of Medicine for paraffin embedding.10 μm of sections were deparaffinized as follows before further treatments: 3 × 3 min in xylene, 3 × 3 min in 100% EtOH, 3 × 3 min in 95% EtOH, 3 min in 80% EtOH, 3 min in 70% EtOH, 3 min in 50% EtOH, 5 min in ddH₂O, and 5 min in PBS.

For cryosectioning, mouse brains were fixed through intracardial perfusion and incubated overnight with 4% paraformaldehyde. Subsequent incubation with 20% sucrose was done for another overnight before embedding tissues in Tissue Plus O.C.T. compound (Fisher Scientific, 4585). 30 μm of sections were mounted onto slides before further processing.

Antigen exposure step was done by incubating the sections in 10 mM sodium citrate (pH 6.5) at 95 °C for 20 min. After 1 hr of serum blocking, slides were incubated with primary antibodies overnight at 4 °C. The next day, slides were rinsed with PBS, incubated with secondary antibodies conjugated with fluorophore for 1 hr, and rinsed with PBS again. After counterstaining with Hoechst 33342 nucleic acid stain (ThermoFisher, H3570), coverslips were mounted with VECTASHIELD^® PLUS Antifade Mounting Medium (Vector Laboratories, H-1900). Antibody information is listed in Supplementary Table 17.

Confocal imaging

Immunofluorescent images were acquired using a Zeiss Axio Zeiss LSM980 confocal microscope with 20x or 63x oil objective. A total of 3 images per sample were acquired in each experiment at 0.5-μm intervals over a 5-μm depth. Immunostaining experiment was performed on the same day for both control and experimental groups and images were acquired using identical laser power setting for comparison. BrdU⁺, Ki67⁺ nuclei, and total nuclei number were recorded using the Analyze Particles function for each field of view in ImageJ/Fiji. Synapse counts and colocalization were calculated using Synapse Counter plugin (SynPuCo) in ImageJ/Fiji (https://github.com/SynPuCo/SynapseCounter). NPY intensity were measured within each field of view, while H3-5HT, Etv5, Tph2, and Npy2r intensity were measured from at least 10 cells from each field of view in each sample followed by normalizing to DAPI.

Selective serotonin reuptake inhibitor treatment

Mice started to receive SSRI intraperitoneally from P28 for 3 weeks. Mice received Sertraline (2.5 mg/kg, in 7% dimethyl sulfoxide in PBS, MedChem Express, HY-B0176A) or vehicle injection 3 days per week before collecting tissues for further analyses.

Protein lysates preparation, histone extracts preparation, and Western blot

For preparing whole cell lysates, tissues were washed with cold PBS three times, followed by dissociation in Radio ImmunoPrecipitation Assay (RIPA) lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% deoxycholate, 0.1% SDS, 1% NP-40; with protease inhibitors) using a pellet homogenizer on ice.

For histone lysates extraction, tissues were washed with cold PBS three times, followed by dissociation using Histone Extraction Kit (Abcam, ab113476) according to the manufacturer’s instructions.

Before performing Western blot analysis, protein quantification was performed using Bradford Protein Assay (BioRad, 5000006) and 40 μg of the protein lysates were applied to SDS gel electrophoresis, and wet transfer to nitrocellulose membrane at 350 mA for 65 min. The membranes were blocked in 5% non-fat milk in Tris-buffered-saline with Tween20 (TBST) and incubated with primary antibodies overnight at 4 °C. The next day, membranes were washed with TBST, incubated with horseradish peroxidase–conjugated secondary antibodies for 1 hr at room temperature, and washed again with TBST before subsequent development using luminol reagent (Santa Cruz Biotechnology, sc2048). Antibody information is listed in Supplementary Table 17.

ChIP

Mouse tumors were dissociated in cold PBS using a pellet homogenizer on ice. Chromatin cross-linking was performed by incubating samples with freshly prepared 1.1% formaldehyde solution at room temperature for 10 min, followed by stopping reaction with 0.1 M glycine addition. After collecting cell pellets by centrifugation, samples were washed with PBS and store at −80 °C until further processing. Pellets were resuspended in PBS/PMSF with 0.5% Igepal to release nuclei before washing with cold ChIP-Buffer (0.25% TritonX, 10 mM EDTA, 0.5 mM EGTA, 10 mM Hepes [pH 6.5]). Nuclei were lysed with ChIP lysis buffer (0.5% SDS, 5 mM EDTA, 25 mM Tris-HCl [pH 8.0]) for 15-20 min at room temperature, sonicated into 250-350 bp using the Diagenode Bioruptor. Fragmented chromatin concentration was quantified using the Quant-iT double-stranded DNA (dsDNA) assay kit (ThermoFisher, Q33120), diluted 5-fold with ChIP-dilution buffer (2 mM EDTA, 150 mM NaCl, 1% Triton X-100, 20 mM Tris-HCl [pH 8.0]; with protease inhibitors). Immunoprecipitation was then performed by incubating fragmented chromatin with antibodies overnight at 4 °C with rotation. The next day, Protein A/G magnetic beads (ThermoFisher, 88802) were added for 6 hrs incubation at 4 °C with rotation. Beads were washed with Tris-SDS-EDTA-I buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris-HCl [pH 8.0]), Tris-SDS-EDTA-II buffer (TSEI buffer with 500 mM NaCl), LiCl buffer (250 mM LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl [pH 8.0]), and Tris-EDTA buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA). Beads were incubated in ChIP-elution buffer (1% SDS, 0.1 M NaHCO₃) for 20 min at 65 °C twice for eluting pull-down samples. Eluates from beads were treated with proteinase K (0.4 mg/mL; ThermoFisher, AM2546) and NaCl (0.125M) overnight at 65 °C for reverse cross-linking. ChIP-DNA was purified using PCR purification kit (Qiagen, 28104) and quantified using the Quant-iT dsDNA assay kit. 10-12 ng of ChIP-DNA was used for ChIP-seq library preparation as described below. Antibody information and antibody:chromatin ratios used for immunoprecipitation are listed in Supplementary Table 17.

ChIP-Seq Library Preparation, Sequencing, and Bioinformatic Analysis

TruSeq ChIP Library Preparation Kit (Illumina, IP-202-1012) was used to prepare ChIP-seq libraries. 250-350 bp of libraries were purified from agarose gel using the QIAquick Gel Extraction Kit (QIAGEN, 28706), amplified with PCR, and extracted using AMPure XP beads (Beckman Coulter Life Science, A63882). The quality of the libraries was analyzed using the Standard Sensitivity NGS Fragment Analysis Kit (Agilent formerly AATI, DNF-473-0500) on 12-Capillary Fragment Analyzer. After using Quant-iT dsDNA assay kit for quantification, equal concentration (2 nM) of each library was pooled for single end (1 × 150) sequencing using High Output v2 kit (Illumina, FC-404-2002) on an illumine NextSeq550 System (~60-80 million reads per sample). All ChIP-Seq experiments were performed on at least two independent biological replicates.

For H3-5HT and ETV5 ChIP-seq analysis, the raw paired-end reads were adapter and quality trimmed using trimgalore (v0.6.7). The resulting reads were aligned to mouse genome mm10 using bowtie2 (v2.3.5.1, parameters: –local -D 20 -R 3 -N 0 -L 20 -i S,1,0.50 –no-unal –no-mixed –no-discordant –phred33 -I 10 -X 700) ⁵⁴. Duplicated reads were then marked and removed using picard MarkDuplicates (v2.26.6) and samtools (v1.14) ⁵⁵, respectively. DeepTools (v3.5.1) was used to convert all the resulting BAM files to Bigwig format for visualization⁵⁶. MACS2 (v2.2.7.1) was used to call peaks, on the resulting BAM files, with a q-value threshold of 0.05⁵⁷. The ZR^FUS binding sites obtained from previous study was used to determine overlap with H3-5HT sites using computematrix¹¹. De novo motif analysis was performed using HOMER (v4.4) software suite.

For identifying core TFs, H3K27ac data were obtained from our previous study¹¹. and super-enhancers were first annotated using ROSE⁵⁸ with a stitching distance of 12.5 kb and exclusion of peaks within 2.5 kb of a promoter. To infer core TFs, the super-enhancers overlapping open chromatin regions marked by ATAC peaks were utilized. The resulting TFs were compared against the TFs identified in ZR^FUS-specific context in our previous work¹⁰.

For H3K27ac and H3K27me3 ChIP-seq analysis in Fig.3, sequencing files were downloaded and merged before performing quality control using fastQC (v0.11.7) and MultiQC (v1.6). Reads were then mapped to mouse genome (mm10 assembly) using bowtie2 (v 2.2.6) ⁵⁴ before generating bedgraph files and tag directories using HOMER (v4.4) software suite⁵⁹. To filter ChIP peaks enriched over input control, findPeaks command in histone mode was conducted using parameters set as follows: H3K27ac ChIP-seq: default; H3K27me3 ChIP-seq: -L 0 -C 0 -size 2000 -minDist 4000. Enriched peaks were annotated using annotatePeaks with mm10 assembly. To visualize peak distribution along the mouse genome, Integrated Genome Browser-compatible files were made using samtools (v0.1.19), sort and index, deepTools (v3.2.0), and bamCompare (v3.2.0) ^55,56. ETS-specific motif analysis was done using seq2profile.pl at 100 bp from the peak center. Overlapping and differentially bound peaks between biological replicates were obtained using getDifferentialPeaks, and peaks were visualized using computeMatrix and plotHeatmap.

In vivo barcoded screen

A piggyBac transposable vector (piggyBac-CAG-GFP-T2A-attR1-attR4) was constructed from the piggyBac-eGFP plasmid⁶⁰. The eGFP stop codon was removed for inserting an in frame T2A sequence followed by attR1 and attR4 Gateway cloning sites, which flanked with chloramphenicol and ccdB selection cassettes. A V5 tag sequence was inserted downstream of the attR4 site. Human variant open reading frames (ORFs) of the developmental TFs were cloned with unique barcode sequence using the HiTMMoB method⁶¹. 38 barcoded TF ORFs and 12 barcoded mCherry controls were pooled at equal moles, ethanol precipitated and reconcentrated in ddH₂O to make the pooled plasmid library for IUE. This plasmid library was diluted to 1 μg/μL for final injection cocktail.

Mouse tumors were dissected and rinsed with PBS for subsequent genomic DNA (gDNA) preparation using the EZNA Tissue DNA Kit (Omega BioTek, D3396), according to the manufacturer’s instructions. The purified gDNA was used for constructing sequencing libraries as previously described²². Briefly, 50 ng of gDNA from mouse tumors and 2 ng of pooled IUE injection cocktail (input control) were used for amplifying barcoded sequences using PCR. Platinum Super Mix (ThermoFisher, 12532016) with primers targeting up- and downstream of the barcodes were used in this PCR reactions with the cycling parameters stated as follows: (94°C, 4 mins) × 1; (94°C, 1 min; 54°C, 1 min; 68°C, 1 min) × 35; (68°C, 10 mins) × 1; hold at 4°C. PureLink PCR Purification Kit (ThermoFisher, K310001) was used to purify the PCR products for further processing using the Ion Plus Library Kit (ThermoFisher, 4471252). Samples were purified and ligated to unique Ion Xpress Barcode Adaptors (ThermoFisher, 4474517) to generate the barcoded libraries. After amplification and purification, the resulting library of each sample were pooled for PGM sequencing (318 V2 Chip) according to the manufacturer’s recommendations. Raw data were concatenated into a single reference file and indexed using the Burrows-Wheeler alignment tool for aligning amplicon barcode sequences. Barcode enrichment level was calculated by quantifying the number of detected reads for each barcode relative to the total number of sequencing reads. Data were presented by normalizing values to input control and standard error of the mean (SEM) was calculated across replicates and plotted as error bars on the graphs.

Immunoprecipitation-mass spectrometry and co-immunoprecipitation

Tissues were washed with cold PBS three times and dissociated on ice using a pellet homogenizer. Nuclear lysates were extracted using NE-PER Nuclear and Cytoplasmic Extraction Reagents (ThermoFisher, 78833) according to the manufacturer’s instructions.

For IP-MS, the immunoprecipitation and mass spectrometry were carried out as described earlier⁵². Briefly, 3mg of nuclear extracts were immuno-precipitated with anti-ETV5 antibody for 1 hour at 4 °C. The protein complex was run on SDS-PAGE and in-gel digested using trypsin protease. The LC-MS analysis was carried out on a nanoLC1000 system coupled to Orbitrap Fusion mass spectrometer (Thermo Scientific, San Jose, CA). The MS raw data was searched against the mus musculus protein database from NCBI refseq (updated 2020_03_24). The ZR^FUS protein and human ETV5 protein sequence was added to the database as well. The raw data processing, label-free based quantification and differential analysis was carried out as described before⁵².

For co-IP, nuclear lysates were prepared from tumor tissues as described above. IP was performed by incubating nuclear lysates with antibodies or anti-IgG overnight at 4 °C with rotation. The next day, Recombinant Protein G Agarose (ThermoFisher, 15920010) were added for subsequent pull-down at 4 °C for 4 hrs. After collecting and washing steps, beads were boiled in 2× SDS gel loading dye for 10 min at 95 °C to elute immunoprecipitated proteins. Eluates were subjected to further analyses by Western blot described above. Antibody information is listed in Supplementary Table 17.

RNA purification and reverse transcription quantitative PCR

Mouse tumors were dissected, rinsed with PBS and dissociated in TRIZOL (ThermoFisher, 15596018). Total RNA was purified using the RNeasy Mini Kit (Qiagen, 74106) following the manufacturer’s instructions. 500 ng of RNA was converted to complementary DNA (cDNA) using iScript Reverse Transcriptase Supermix (BioRad, 1708841). qPCR reactions mix included 2 ng cDNA, 250 nM primers, and 1× PerfeCTa SYBR Fast Mix (Quantabio, 95072-012), and the reaction was conducted on a BioRad CFX96 Touch Real-Time PCR Detection System (95 °C for 30 s, 40 cycles of 95 °C for 5 s and 60°C for 30 s, with subsequent melting curve analysis). The transcript expression of target genes was normalized to Actb. Primer sequences are listed in Supplementary Table 17.

RNA-seq library preparation, sequencing, and bioinformatic analysis

RNA extraction was performed as described above. 300 ng of total RNA was used for constructing Illumina sequencing libraries with 6-bp single indices by the TruSeq Stranded mRNA kit (Illumina, RS-122-2101). Equal concentration (2 nM) of each library was pooled together for paired-end sequencing using Mid Output v2 kit (Illumina, 20024904) on an Illumina NextSeq550 System (~20-30 million reads per sample).

Sequencing files were downloaded in fastq files format and quality control was performed using fastQC (v0.11.7) and MultiQC (v1.6) before mapping reads to mouse genome (mm10 assembly) using STAR (v2.5.0a) ⁶². Mapped reads were then used to build count matrices using Bioconductor packages GenomicAlignments (v1.26.0) and GenomicFeatures (v1.42.3) ⁶³ in R (v4.0.3). University of California Santa Cruz transcripts were downloaded from Illumina iGenomes in GTF file format. DESeq2 (v1.30.1) ⁶⁴ was used to normalize and analyze differential gene expression. Enrichr was used to determine GO terms with P value of < 0.05 and ggplot2 (v3.3.5) was used to visualize the results.

Slice recording for EPSC, IPSC

Animals were anesthetized with isoflurane at P30 to isolate and submerge the brains in ice-cold ACSF solution (130 mM NaCl, 24 mM NaHCO₃,1.25 mM NaH₂PO₄, 3.5 mM KCl, 1.5 mM CaCl₂,1.5 mM MgCl₂, and 10 mM D (+)-glucose, pH 7.4). 300-mm brain slices were cut using a vibratome (DSK Linear Slicer) oxygenated in ACSF at room temperature for 1 hour and then acclimated at room temperature with continuous perfusion with ASCF solution (2 ml/min). Slices were placed in the recording chamber, and target cells were identified with an upright Olympus microscope with a 60× water immersion objective with infrared differential interference contrast optics. Whole-cell recording was performed from cortical neurons at room temperature with pCLAMP10 and Multi-clamp 700B amplifier (Axon Instrument, Molecular Devices). The holding potential was −60 mV and pipette resistance was typically 5-8 megohm. The pipette was filled with an internal solution: 135 mM CsMeSO₄, 8 mM NaCl, 10 mM HEPES, 0.25 mM EGTA, 1 mM Mg-ATP, 0.25 mM Na₂-GTP, 30 mM QX-314, pH 7.2 with CsOH (278-285 mOsmol) for EPSC measurement; 135 mM CsCl, 4 mM NaCl, 0.5 mM CaCl₂, 10 mM HEPES, 5 mM EGTA, 2 mM Mg-ATP, 0.5 mM Na₂-GTP, 30 mM QX-314, pH 7.2 with CsOH (278-285 mOsmol) for IPSC measurement. IPSC measurement was performed with ionotropic glutamate receptor antagonists, (2R)-amino-5-phosphonovaleric acid (APV) (50 mM, Tocris), and cyanquixaline (CNQX) (20 mM, Tocris). Electrical signals were digitized and sampled at 50 ms intervals with Digidata 1550B and Multi-clamp 700B amplifier (Molecular Devices) on pCLAMP 10.7 software. Data were filtered at 2 kHz and the recorded current was analyzed with ClampFit 10.7 software.

Electroencephalogram

Mice at the age of P21-25 were implanted with intracranial EEG electrodes as previously published protocols²³. 48 hrs of continuous monitoring session was performed every 10 days from P30. In each recording session, mice were connected to a tethered wire and housed in a recording chamber where food and water were provided ad libitum. The recording room were maintained at 20-22°C with 40% humidity and featured 12 hrs light/dark cycles. EEG signals were acquired at 2 kHz using BioAmp (ADI), digitized, and analyzed using LabChart Pro software. Cortical spikes were quantified using a built-in function in LabChart Pro with optimized detection parameters for each event. Seizure event was calculated from the waveform data and corresponding video-recorded behaviors by visual inspection.

Quantification and statistical analysis

Sample sizes and statistical details are listed in all figure legends. All analyses were performed blind to experimental conditions. Animals from each cohort were randomly distributed to experimental groups. Analyses were conducted using ImageJ/Fiji, Graphpad Prism 9, and R-studio. For immunofluorescent imaging, each experiment was performed in at least 3 independent pairs of animals using the same sets of staining and confocal imaging conditions. Representative images with similar results were showed in the figures. For Kaplan–Meier survival analysis, we used the log-rank test to compare the differences between groups. For the other datasets, data were tested for normal distribution using the Shapiro-Wilk tests and for homogeneity of variance using the Levene test. Parametric tests were used for normally distributed datasets and for data with small sample size (n=3). For comparison of two groups, unpaired or paired Student’s t test was used, and for comparison of three groups, one-way analysis of variance (ANOVA) followed by Tukey’s tests was used. If data were not normally distributed, non-parametric Wilcoxon rank sum tests were performed. Data are presented as ± SEM. N refers to animal number. Levels of statistical significance are indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001

Extended Data

Extended Data Figure 1. DREADD-based activation of neurons via CNO treatment.

a. Schematic of DREADD-hM3Dq activation of ipsilateral cortical neurons in ZR^FUS EPN mice.

b. Low magnification image of EPN tumors and representative BrdU staining of EPN tumors after DREADD-hM3Dq activation of ipsilateral cortical neurons via CNO (scale bar=50 μm).

c. Quantification of BrdU staining in saline versus CNO treated EPN tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, P_{SalinevsAAV_CNO} = 0.4522, P_{CNOvsAAV_CNO} = 0.1221, *P_{AAV_SalinevsAAV_CNO} = 0.0337).

d. Schematic of DREADD-hM3Dq activation of ipsilateral inhibitory neurons in ZR^FUS EPN mice.

e. Low magnification image of EPN tumors and representative BrdU staining of EPN tumors after DREADD-based activation of ipsilateral inhibitory neurons via CNO (scale bar=50 μm).

f. Quantification of BrdU staining in saline versus CNO treated EPN tumors (Saline: n=3, CNO: n=3, AAV_Saline: n=3, AAV_CNO: n=4, mean±SEM, unpaired two-sided Student’s t test, P_{SalinevsAAV_CNO} = 0.0770, P_{CNOvsAAV_CNO} = 0.3896, P_{AAV_SalinevsAAV_CNO} = 0.8727).

g. Representative BrdU staining in saline and CNO (0.5 and 5 mg/kg) treated EPN tumors (scale bar=50 μm).

h. Immunofluorescence staining of c-Fos in the ipsilateral cortical neurons in saline versus CNO treated EPN tumors (scale bar=50 μm).

i. Quantification of c-Fos positive neurons in the ipsilateral cortex in saline versus CNO treated EPN tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, *P = 0.0155).

j. Immunofluorescence staining of mCherry and Tph2 in the AAV injected dRN (scale bar=30 μm). Tph2: tryptophan hydroxylase 2.

k. Immunofluorescence staining of c-Fos in the dRN serotonergic neurons in saline versus CNO treated EPN tumors (scale bar=50 μm).

l. Quantification of c-Fos positive neurons in the dRN serotonergic neurons in saline versus CNO treated EPN tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, ***P = 0.0005).

Panel a, d were created using Biorender.com

Extended Data Figure 2. Expression of monoamine transporters in EPN.

a. Expression levels of genes under monoamine transporter GO term in human ZR^FUS versus non-ZR^FUS EPN tumors.

b. Expression levels of genes under monoamine transporter GO term in mouse EPN tumors versus non-tumor tissues.

c. Log₂(FC) of serotonin transporters in human ZR^FUS tumors versus non-ZR^FUS tumor tissues.

d. Log₂(FC) of serotonin transporters in mouse EPN tumors versus non-tumor tissues.

e. Normalized read counts of serotonin transporters in mouse EPN tumors versus non-tumor tissues (at least n=10 per group, median±upper/lower limits, box boundary states upper/lower quartiles).

f. Immunofluorescence staining of Slc6a4 in mouse non-tumor cortex and EPN tumors (scale bar=50 μm).

g. Immunofluorescence staining of dopamine transporter (Slc6a3) in mouse substantia nigra, non-tumor cortex and EPN tumors (scale bar=50 μm).

Extended Data Figure 3. Synaptic staining and H3-5HT staining after dRN manipulation.

a. Low magnification image of tumor margin and representative higher magnification images (derived from dashed box) of peri-tumoral excitatory synaptic staining in saline versus CNO treated tumors from Fig. 1f (scale bar=10 μm).

b. Quantification of excitatory synaptic staining in saline versus CNO treated tumors Fig. 1f (AAV_saline: n=4, AAV_CNO: n=3, mean±SEM, two-sided Wilcoxon rank sum test, P = 0.6286).

c. Low magnification view of tumor margin and representative higher magnification images (derived from dashed box) of peri-tumoral inhibitory synaptic staining in saline versus CNO treated tumors from Fig. 1f (scale bar = 10 μm).

d. Quantification of inhibitory synaptic staining in saline versus CNO treated tumors from Fig. 1f (AAV_saline: n=4, AAV_CNO: n=3, mean±SEM, unpaired two-sided Student’s t test, P = 0.7481).

e. Low magnification image of tumor margin and representative higher magnification images (derived from dashed box) of peri-tumoral excitatory synaptic staining in saline versus CNO treated tumors from Fig. 1i (scale bar=10 μm).

f. Quantification of excitatory synaptic staining in saline versus CNO treated tumors from Fig. 1i (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, P = 0.6460).

g. Low magnification image of tumor margin and representative higher magnification images (derived from dashed box) of peri-tumoral inhibitory synaptic staining in saline versus CNO treated tumors from Fig. 1i (scale bar=10 μm).

h. Quantification of inhibitory synaptic staining in saline versus CNO treated tumors from Fig. 1i (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, P = 0.9864).

i. Immunofluorescence staining of H3-5HT in saline versus CNO treated tumors from Fig. 1f (scale bar=10 μm).

j. Quantification of H3-5HT intensity in saline versus CNO treated tumors from Fig. 1f (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, P = 0.2894).

k. Immunofluorescence staining of H3-5HT in saline versus CNO treated tumors from Fig. 1i (scale bar=10 μm).

l. Quantification of H3-5HT intensity in saline versus CNO treated tumors from Fig. 1i (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, *P = 0.0465).

Extended Data Figure 4. Tumor proliferation and Etv5 expression in H3.3-Q5A EPN tumors.

a. Immunofluorescence staining of H3-5HT in H3.3-WT and H3.3-Q5A EPN tumors (scale bar=25 μm).

b. Quantification of H3-5HT intensity in H3.3 control and H3.3-Q5A (n=4 per group, mean±SEM, two-sided Wilcoxon rank sum test, *P = 0.0286)

c. Representative Ki67 staining of EPN control versus H3.3 wild-type and H3.3-Q5A tumors (scale bar=50 μm).

d. Quantification of Ki67 staining in EPN control versus H3.3 wild-type and H3.3-Q5A tumors (EPN control: n=3, H3.3 wild-type: n=5, H3.3-Q5A: n=4, mean±SEM, unpaired two-sided Student’s t test, P_{EPNcontrolvsH3.3wild-type} = 0.9040, **P_{EPNcontrolvsH3.3-Q5A} = 0.0015, ****P_{H3.3wild-typevsH3.3-Q5A} = 1.59E-05).

e. Representative ETV5 staining of EPN control versus H3.3 wild-type and H3.3-Q5A tumors (scale bar=25 μm).

f. Quantification of Etv5 staining in EPN control versus H3.3 wild-type and H3.3-Q5A tumors (EPN control: n=3, H3.3 wild-type: n=3, H3.3-Q5A: n=4, mean±SEM, unpaired two-sided Student’s t test, P_{EPNcontrolvsH3.3wild-type} = 0.6180, **P_{EPNcontrolvsH3.3-Q5A} = 0.0053, *P_{H3.3wild-typevsH3.3-Q5A} = 0.0398).

Extended Data Figure 5. Slc6a4-LOF EPN analyses, expression of serotonin synthetase, and SSRI treatment in EPN.

a. Kaplan–Meier survival curve of EPN control (n=19, median=74 days) and Slc6a4-LOF (n=14, median=95 days, log-rank test, P = 0.1177).

b. Representative Ki67 staining of EPN control versus Slc6a4-LOF tumors (scale bar=50 μm).

c. Quantification of Ki67 staining in EPN control versus Slc6a4-LOF tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, **P = 0.0055).

d. Representative H3-5HT staining of EPN control versus Slc6a4-LOF tumors (scale bar=25 μm).

e. Quantification of H3-5HT intensity in EPN control versus Slc6a4-LOF tumors (n=4 per group, mean±SEM, two-sided Wilcoxon rank sum test, *P = 0.0286).

f. Immunofluorescent staining of Tph2 in mouse dRN (positive control), non-tumor cortex, and EPN tumor (scale bar=50 μm)

g. Quantification of Tph2 intensity in mouse dRN, non-tumor cortex, and EPN tumor (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, ****P_{dRNvsnon-tumorcortex} = 3.76E-05, **P_{dRNvsEPNtumor} = 3.73E-05, P_{non-tumorcortexvsEPNtumor} = 0.7262).

h. Schematic of DMSO/SSRI treatment in EPN

i. Representative H3-5HT staining of DMSO versus Sertraline-HCl treated tumors (scale bar=25 μm).

j. Quantification of H3-5HT intensity in DMSOl versus Sertraline-HCl treated tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, P = 0.0743)

k. Panel h was created using Biorender.com

Extended Data Figure 6. H3-5HT ChIP-seq and ETV5 ChIP-seq analyses in EPN.

a. Venn diagram depicting core TFs with H3-5HT peaks in mouse EPN tumors.

b. ETV5 ChIP-seq heatmap profiles in mouse EPN tumors and venn diagram depicting genes overlapped between downregulated DEGs acquiring H3K27me3 peaks in ETV5-GOF tumors and ETV5 annotated genes.

Extended Data Figure 7. Validation of candidates from functional screen.

a. Immunoblots of Lhx2, Lhx4, Etv5, and Klf12 in mouse non-tumor cortex versus EPN tumors (n=3 per group).

b. Kaplan–Meier survival curves of EPN control (n=51, median=70 days), LHX2-GOF (n=23, median=82 days, log-rank test, P = 0.9640), LHX4-GOF (n=9, median=114 days, log-rank test, P = 0.3532), and KLF12-GOF (n=23, median=77 days, log-rank test, P = 0.7588).

c. Immunoblots of LHX2, LHX4, and KLF12 in control versus corresponding GOF tumors.

d. Immunoblots of ETV5 in control versus ETV5-GOF and ETV5-LOF tumors.

e. RT-qPCR fold-enrichment of ETV5 and Etv5 transcript (ddCt) in control versus ETV5-GOF and ETV5-LOF tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, GOF: **P = 0.0068, LOF: ***P = 0.0007).

f. Surveyor Nuclease Digestion assay of Etv5 gRNA efficiency. Mouse non-tumor cortex: negative control. Two primer sets were used, and gel images are presented in left and right panel. Asterisks label the nuclease digested bands.

g. Representative BrdU staining of control versus ETV5-GOF and ETV5-LOF tumors (scale bar=50 μm).

h. Quantification of BrdU staining in control versus ETV5-GOF and ETV5-LOF tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, GOF: **P = 0.00258, LOF: ***P = 0.000976).

i. Immunoblots of H3K27ac and H3K27me3 in control versus ETV5GOF tumors. Total H3: loading control.

j. Ring chart showing percentage of H3K27me3 sites in ETV5-GOF tumors carrying ETS motif allowing 0 mismatch at 1,000 bp from peak center.

k. GO-terms analysis of ETV5 binding partners in mouse EPN tumors performed using Enrichr (two-sided Fisher’s exact test).

l. RT-qPCR fold-enrichment of gene transcript (ddCt) in control versus ETV5GOF tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, Npy: ***P = 0.00024, Gabra5: P = 0.4342, Chrm4: *P = 0.0239, Kcnmb4: *P = 0.0109, Nptx2: P = 0.2549).

m. H3K27me3 ChIP-seq peaks at Npy and Chrm4 locus in control and ETV5-GOF tumors.

Extended Data Figure 8. Kaplan–Meier survival curves comparison between sexes.

a-c. All Kaplan–Meier survival curves, table of median survival, and log-rank test results comparison between groups divided by sex.

Extended Data Figure 9. Analysis of NPY-GOF EPN tumors.

a. RT-qPCR fold-enrichment of NPY transcript (ddCt) in control versus NPY-GOF tumors (n=3 per group, mean±SEM, unpaired two-sided Student’s t test, *P = 0.0255).

b. RT-qPCR fold-enrichment of Npy2r transcript (ddCt) in mouse non-tumor cortex and EPN tumors (n=3 per group, as mean±SEM, unpaired two-sided Student’s t test, *P = 0.0215).

c. Immunofluorescence staining of Npy2r in mouse cortex. NeuN: neuronal marker, Aldh1l1: astrocytic marker (scale bar=50 μm).

d. Low magnification image of the tumor margin and representative images of Npy2r staining in mouse non-tumor and EPN tumors (scale bar=50 μm). Top right panel: Quantification of Npy2r intensity normalized to DAPI in mouse non-tumor cortex versus EPN tumors. (n=4 per group, mean±SEM, unpaired two-sided Student’s t test, **P = 0.0014).

e. GO-terms analysis of the downregulated DEGs in NPY-GOF tumors versus control performed using Enrichr (two-sided Fisher’s exact test).

f. Zoom-out EEG traces from mice bearing control and NPY-GOF tumors.

Data availability

ChIP-seq and RNA-seq data have been deposited in the NCBI Gene Expression Omnibus under accession no. GSE246033. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD048170. Source data are provided with this paper. All other data in this article are available from the corresponding author on reasonable request.