Inhibition of the cGAS‐STING pathway ameliorates the premature senescence hallmarks of Ataxia‐Telangiectasia brain organoids

Abstract Ataxia‐telangiectasia (A‐T) is a genetic disorder caused by the lack of functional ATM kinase. A‐T is characterized by chronic inflammation, neurodegeneration and premature ageing features that are associated with increased genome instability, nuclear shape alterations, micronuclei accumulation, neuronal defects and premature entry into cellular senescence. The causal relationship between the detrimental inflammatory signature and the neurological deficiencies of A‐T remains elusive. Here, we utilize human pluripotent stem cell‐derived cortical brain organoids to study A‐T neuropathology. Mechanistically, we show that the cGAS‐STING pathway is required for the recognition of micronuclei and induction of a senescence‐associated secretory phenotype (SASP) in A‐T olfactory neurosphere‐derived cells and brain organoids. We further demonstrate that cGAS and STING inhibition effectively suppresses self‐DNA‐triggered SASP expression in A‐T brain organoids, inhibits astrocyte senescence and neurodegeneration, and ameliorates A‐T brain organoid neuropathology. Our study thus reveals that increased cGAS and STING activity is an important contributor to chronic inflammation and premature senescence in the central nervous system of A‐T and constitutes a novel therapeutic target for treating neuropathology in A‐T patients.


| INTRODUC TI ON
Ataxia-telangiectasia (A-T) is a rare human genetic disease caused by homozygous or compound heterozygous mutations in the ataxiatelangiectasia-mutated (ATM) gene that encodes the ATM kinase. In agreement with ATM's well-documented role in orchestrating the cellular response to DNA double-strand breaks (DSBs) (Shiloh & Ziv, 2013) A-T patient cells show markedly increased radiosensitivity (Taylor et al., 2019), increased genome instability, elevated inflammation profiles, accelerated telomere attrition, mitochondrial dysfunction and premature entry into cellular senescence as compared to their healthy control counterparts (Shiloh & Lederman, 2017).
Uncontrolled or sustained inflammation is strongly linked to a number of age-related chronic neurodegenerative diseases (Baker & Petersen, 2018), and a systemic elevation of inflammatory cytokine levels was previously linked to detrimental outcomes in both A-T patients and mouse models. Interestingly, senescent cells adopt a pro-inflammatory secretome known as the senescence-associated secretory phenotype (SASP) (Campisi & d'Adda di Fagagna, 2007), which is proposed to play an important role in promoting neurological deficiencies (Bussian et al., 2018).
As cells enter into senescence, a major hallmark of human ageing (Lopez-Otin et al., 2013), DNA fragments leak out from the nucleus into the cytoplasm of mammalian cells (Dou et al., 2017).
Such cytoplasmic chromatin fragments-often seen in the form of micronuclei-are proposed to arise as a result of replication stress (Ho et al., 2016), DNA damage (Ahn et al., 2014), mitochondrial dysfunction (Vizioli et al., 2020) or intermediates of reverse-transcribed retroelements (De Cecco et al., 2019). This results in activation of the cGAS-STING pathway (Dou et al., 2017), a cytosolic DNA sensing signalling response that plays essential roles in activating proinflammatory genes (Barber, 2015). Interestingly, A-T patient skin fibroblasts prematurely undergo cellular senescence and accumulate micronuclei in culture (Lan et al., 2019). In agreement with these data, a recent report postulated a role for cGAS and STING in sensing cytoplasmic DNA in Atm −/− mice, resulting in a systemic inflammatory phenotype (Hartlova et al., 2015) reminiscent of the clinical picture of A-T patients.
Neurodegeneration remains however an important yet poorly understood aspect of A-T (Rothblum-Oviatt et al., 2016). Given that Atm −/− mice generally fail to display clear symptoms of neurological deficiencies including neurodegeneration (Lavin, 2013), understanding how a lack of ATM kinase activity impacts the nervous system in humans is instrumental for understanding and treating A-T-associated neurodegeneration. In an effort to more accurately model the neurological detrimental phenotypes of A-T, in this study, we used A-T patient olfactory neurosphere-derived (ONS) cells (Stewart et al., 2013) and generated induced pluripotent stem cells (iPSC) from these cells in order to generate human cortical brain organoids. Here, we show that primary A-T patient ONS cells retain misshapen nuclei and exhibit significantly increased amounts of cytoplasmic DNA in the form of micronuclei, which in turn prime a constitutive SASP response that is dependent on the cGAS-STING pathway. Furthermore, we then demonstrate that A-T patient iPSC-derived brain organoids show a similar increase in micronuclei, elevated numbers of senescent cells located mostly within astrocyte populations, increased expression of pro-inflammatory genes, as well as premature neuronal degeneration and dysfunction. Importantly, inhibition of cGAS and STINGdependent inflammation effectively suppresses self-DNA-induced SASP activation and rescues neuropathological features in A-T brain organoids. Collectively, our results reveal a major contribution of the cGAS-STING innate immune signalling pathway to detrimental neurological phenotypes of A-T and validate its functional inhibition as a promising target for treating A-T neuropathogenesis.

| Micronuclei in primary A-T ONS cells activate the cGAS-STING pathway
To explore the potential generation of cytoplasmic DNA and senescence hallmarks in A-T human cells and study its role in neuroinflammation, we employed primary human ONS cells obtained from olfactory mucosa biopsies, comparing ONS cells from five A-T patients and five healthy controls. Consistent with previous reports F I G U R E 1 Lack of functional ATM in ONS patient cells induces a cGAS-STING-dependent SASP response driven by accumulation of cytoplasmic DNA. (a-e) 5 wild type (WT) and 5 A-T human ONS cell lines were used and analysed at identical passage number (passage 20). (a) Representative images of senescence-associated β-galactosidase (SAβ-gal) positive assays shown in b. Scale bar, 100 μm. (b-d) Each point in the bar plot in (b) shows the average percentage of SAβ-gal positive cells of one cell line, (c) the average percentage of cells with misshapen nuclei and in (d) the average percentage of cells with micronuclei. Error bars represent SD; n = 5 independent patient samples; Student's t test. (e) Total RNA from human A-T ONS patient cells was used to quantify the mRNA expression levels of the indicated SASP genes and normalized to RPLP0 mRNA and compared to WT ONS controls. Error bars represent SD; n = 5 independent biological samples; Student's t test. (f, g) A-T ONS cells were transfected with the indicated siRNA for 48hr. (f) Representative Western blot analysis of cGAS and STING of whole cell extracts from siRNA-transfected cells. α-tubulin was used as loading control. (g) Total RNA from siRNA-transfected A-T ONS patient cells was used to quantify the mRNA expression levels of the indicated SASP genes and normalized to RPLP0 mRNA and compared to siControl. Error bars represent SD; n = 3 independent experiments; one-way ANOVA with Tukey's multiple-comparison post hoc corrections. (h) Representative Western blot analysis of JNK1/2 and ATM of whole cell extracts from siRNA-transfected WT and A-T ONS cells. α-tubulin was used as loading control. (i) WT and A-T ONS cells were either transfected with the indicated siRNAs, treated with JNK inhibitor (JNKi, SP600125, 20 μM) or MitoQ (100 nM) for ten days. Cells were thereafter assessed for micronuclei formation. Bar graphs show the percentage of micronuclei-positive cells. Error bars represent SD; n = 3 independent experiments; one-way ANOVA with Tukey's multiple-comparison post hoc corrections in human A-T skin fibroblasts (Lan et al., 2019), we detected a premature entry into cellular senescence (Figure 1a, b)-as measured by senescence-associated β-galactosidase (SAβ-gal) activity-and heightened numbers of misshapen nuclei (Figure 1c) in A-T ONS cells compared to their wild-type counterparts at identical passage number. This was associated with a concomitant increase in cytoplasmic chromatin content, as measured by elevated numbers of micronuclei ( Figure 1d). To further substantiate the direct involvement of ATM in this phenomenon, we exposed primary healthy control dermal fibroblasts to the specific ATM inhibitor KU-55933 over a three-day period. Our data showed that acute ATM inhibition also led to a progressive increase in nuclear shape abnormalities ( Figure S1a) and micronuclei counts ( Figure S1b, c).
Because increased micronuclei have been reported to prime a SASP gene response (Dou et al., 2017), we next quantified mRNA abundance of the SASP genes IL-6, IL-8, IL1α and CCL20. Remarkably, all A-T ONS cells consistently displayed elevated levels of the SASP genes tested as compared to their wild-type (WT) counterparts ( Figure 1e). Since the mechanisms that activate SASP involve a series of events that are connected to DNA damage and the cGAS-STING pathway (Hartlova et al., 2015), we hypothesized that the elevated levels of micronuclei observed in ONS cells of A-T patients are recognized by the cGAS-STING pathway to promote the SASP program.
To test this, we stably reduced cGAS or STING expression in WT directly test this, we exposed WT and A-T proliferating ONS cells to either concomitant downregulation of JNK1 and JNK2 expression via RNA interference (Figure 1h), the JNK inhibitor SP600125 (Bennett et al., 2001) or the mitochondrial antioxidant MitoQ (Kelso et al., 2001) over a ten-day period. Interestingly, these three inter-

| Premature senescence and inflammation in A-T brain organoids
The brain is one of the major organs to exhibit typical signs of disease in A-T patients. These include microcephaly, progressive neurodegeneration, chronic inflammation and premature ageing (Rothblum-Oviatt et al., 2016;Shiloh & Lederman, 2017). Furthermore, A-T patients likewise exhibit defects in the cortex, such as cortical neuropathology and cerebral white-matter abnormalities (Ciemins & Horowitz, 2000;Pizzamiglio et al., 2016). To explore whether cGAS-STING signalling contributes to A-T neuropathology, we generated cortical brain organoids (BOs) from pluripotent stem cells (PSCs) of WT and A-T genetic backgrounds. Importantly, droplet digital PCR on genomic DNA of WT and A-T PSCs confirmed that both cell lines carried wild-type TP53 sequences ( Figure S1g) in regions where mutations are known to affect the ability of p53 to bind to the DNA binding domain and therefore diminish p53-mediated regulation of apoptosis, cell cycle progression and genomic stability, as previously described (Merkle et al., 2017). Consistent with our data in ONS cells, showing WT versus A-T brain organoid differential expression of upregulated (green) and downregulated (orange) genes. (e) Gene Set Enrichment Analysis using ageing hallmark gene sets from the Molecular Signature Database was carried out. The statistically significant signatures were selected (FDR < 0.25) and placed in order of normalized enrichment score, which represents the strength of the relationship between the phenotype and gene signature. Bars indicate the pathways enriched in genes that are downregulated (orange) and upregulated (green) in A-T BOs as compared to the WT BO group. (f) Total RNA from WT and A-T BOs were used for RT-qPCR analysis. B2M mRNA was used as normalizer. n = 5 independent biological samples. Exact P values can be found in Figure S2. (g) Immunofluorescence representative images of WT and A-T BO sections stained for NF-κB phosphorylated on serine 536 (pNF-κB). Scale bar, 0.7 mm. (h) Quantification of data presented in (g). Bar graphs show the percentage of pNF-κB positive cells. Each point in the scatter plot represents a single BO section analysed. Error bars represent SD; n = 3 independent experiments; Student's t test
Given that transcription of the SASP upon GAS and STING activation is mediated by phosphorylation of the transcription factors NF-κB on serine 536 (pNF-κB) and IRF3 on serine 386 (pIRF3), we next examined the in situ accumulation of pNF-κB and pIRF3 in WT Early Genes (IEGs) FOS, NPAS4 and ARC, as previously reported (Madabhushi et al., 2015). Interestingly, A-T BOs displayed significantly lower expression of IEGs than WT BOs (Figure 2f and Figure   S3g) that could not be upregulated by exposure to excitatory stimulation with KCl, NMDA or Bicuculline, whereas healthy control BOs showed a robust expression of IEGs that were further increased by excitatory stimulation (Figure 3h). Altogether, these results demonstrate that A-T BOs exhibit a significant accumulation of senescent astrocytes that promote a pro-inflammatory environment, which in turn results in neurons that display dysfunctional neuronal activity.

| cGAS or STING inhibition reduces A-T brain organoid senescence
Since the above data support the putative role for the cGAS-STING pathway in driving premature senescence in A-T BOs, we next examined the long-term consequence of premature senescence on neuronal survival. We then generated BOs and allowed them to To study the contribution of the cGAS and STING pathways to this detrimental neuronal phenotype in A-T BOs, we next investigated whether we could ameliorate SASP and senescence with previously reported specific cGAS and STING inhibitors (Dai et al., 2019;Haag et al., 2018). To this end, we treated BOs weekly with the cGAS inhibitor aspirin (Dai et al., 2019) or the STING inhibitor H-151 (Haag et al., 2018) for a period of one month prior to reaching the sixth month of BOs development ( Figure S4a). Excitingly, we observed that incubation with either of these inhibitors significantly rescued the loss of organoid size in A-T BOs, while WT's size remained unaltered (Figure 4a). Furthermore, both cGAS and STING inhibition significantly reduced the abundance of cells with SAβ-gal activity in A-T BOs to levels comparable to their control counterparts, (Figure 4b, c), and simultaneously led to a concomitant reduction in p21 and SASP expression in A-T BOs (Figure 4d and Figure   S4b-h). Taken together, these results strongly indicate that inhibition of cGAS-STING signalling in AT-BOs decreases senescence phenotypes, reduces inflammation and has beneficial effects on brain tissue homeostasis. suggesting that neither cGAS nor STING inhibition has detrimental effects on healthy WT neurons in this model system. In summary, these results demonstrate that inhibition of cGAS or STING with aspirin or H-151 respectively, mitigates astrocyte senescence in A-T BOs, resulting in improved neuronal function and survival and a significant reduction in brain organoid inflammatory signatures.

| DISCUSS ION
Ageing in the brain is characterized by increased inflammation, a decline in DNA repair, accumulation of senescent cells, and these each represent major risk factors for neurodegenerative diseases In this study, we first showed that primary ONS cells from nasal biopsies of five A-T patients consistently display nuclear shape abnormalities and increased micronuclei with concomitant elevated pro-inflammatory SASP levels. Importantly, delivery of siRNAs against cGAS and STING in A-T ONS cells significantly reduced the expression of a number of SASP genes, indicating that the cGAS-

ONS cells.
Furthermore, we showed that human BOs of A-T accumulate increased numbers of senescent cells that are largely comprised of astrocyte populations. While senescent OPCs and microglia are reported to contribute to age-related neurodegeneration in the mouse brain (Ogrodnik et al., 2021;Zhang et al., 2019), our data suggest that OPCs undergo minimal cellular senescence onset in A-T BOs.
As regards to microglia, we did not characterise their role in this  the phenotypes associated with cellular senescence in human brain models. Indeed, this family of senomorphic approaches designed to reduce or alter the SASP have shown promise in vivo (Aguado et al., 2019;Dai et al., 2019;Haag et al., 2018;Milanovic et al., 2018), suggesting that novel therapeutic compounds targeting the drivers of SASP, including cGAS and STING, could greatly advance this area of research.
It is worth highlighting that these conclusions may have impact

| Olfactory biopsies
As previously described (Stewart et al., 2013), nasal biopsies were collected from controls and A-T patients. Biopsies were obtained with informed consent from the parents in accordance with the   for subsequent analysis. First, the raw count was transformed to count-permillion (cpm) value. Then, genes with 0 cpm value across all ten samples were excluded from DGE analysis. The raw count contained 38557 genes and 25391 passed the filtering process and were used for statistical analysis. A total of 13166 genes failed to meet the filtering criteria. Benjamini-Hochberg multiple testing correction was applied to the DGE to obtain the adjusted P value for all genes. Genes were defined as significantly up-or downregulated when A-T brain organoids presented a p value <0.05 compared to wild-type brain organoids. Gene expression was analysed by comparison between 5 WT and 5 A-T brain organoids. Samples were mounted in mowiol solution (Calbiochem).

| RNA in situ hybridization
RNA in situ hybridization (ISH) was performed using an RNAscope HiPlex Kit (Advanced Cell Diagnostics) with detection probes against GFAP, IL1B, IL-8 and IL-6 mRNAs following directions of the manufacturer. Brain organoids for RNA ISH were sectioned as described above and sections were consecutively exposed to target retrieval conditions followed by digestion with protease, hybridization with target probes, amplification and labelling with fluorophores. Samples were counterstained with DAPI to visualize nuclei, and slides were imaged with a Zeiss AxioScan Z1 Fluorescent Imager.

| Imaging and analysis
Immunofluorescence images were acquired using a multiphoton confocal Leica TCS SP8MP fluorescence microscope or a Zeiss AxioScan Z1 Fluorescent Imager. For organoid stainings, the number of positive cells per organoid for senescence, astrocyte and neuronal markers tested was analysed by the imaging software CellProfiler, using the same pipeline for each sample in the same experiment.
To analyse image circularity, processing of each confocal microscope high-resolution image consisted of a consecutive series of algorithms implemented as plugins in the freely available software ImageJ (http://imagej.nih.gov/ij/). The outlines of segmented nuclei are determined using edge detection algorithms based on differential brightness cut-offs. Circularity indexes range from 1.0 (representing a perfect circle) to 0 (representing a straight line).

| Measurement of organoid-secreted proteins
Media concentrations of IL-1α, IL-1β, IL-6, IL-8, MMP3 and CCL20 were assessed using a custom-made magnetic luminex screening assay (R&D Systems) according to the manufacturer's instructions and read on a MAGPIX instrument (Luminex Corp).

| Antibodies
Anti

| Statistical analysis
Results are shown as mean ± standard error of the mean (SEM) or standard deviation (SD) or as percentages ±95% confidence interval as indicated. p Value was calculated by the indicated statistical tests, using Prism software. In figure legends, n indicates the number of independent experiments.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

AUTH O R CO NTR I B UTI O N S
H.C. generated human brain organoids and data on Figures  the study and wrote the paper. All authors edited the paper.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support this study are available from the corresponding authors upon reasonable request. Sequence data from brain organoids bulk RNA-seq that support the findings of this study have been deposited in the European Nucleotide Archive with the primary accession code PRJEB43363.