Excessive release of inorganic polyphosphate by ALS/FTD astrocytes causes non-cell autonomous toxicity to motoneurons

SUMMARY

Non-cell autonomous mechanisms contribute to neurodegenerative diseases like amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), in which astrocytes release unidentified factors that are toxic to motoneurons (MNs). We report here that mouse and patient iPSC-derived astrocytes with diverse ALS/FTD-linked mutations (SOD1, TARDBP, C9ORF72) display elevated levels of intracellular inorganic polyphosphate (polyP), a ubiquitous, negatively charged biopolymer. PolyP levels are also increased in astrocyte conditioned media (ACM) from ALS/FTD astrocytes. ACM-mediated MN death is prevented by degrading or neutralizing polyP in ALS/FTD astrocytes or ACM. Studies further reveal that postmortem familial and sporadic ALS spinal cord sections display enriched polyP staining signals and that ALS cerebrospinal fluid (CSF) exhibits increased polyP concentrations. Our in vitro results establish excessive astrocyte-derived polyP as a critical factor in non-cell autonomous MN degeneration and a potential therapeutic target for ALS/FTD. The CSF data indicate that polyP might serve as a new biomarker for ALS/FTD.

Graphical Abstract

eTOC

Arredondo et al., demonstrate that inorganic polyphosphate (polyP) levels are increased in human and mouse ALS/FTD astrocytes in culture and in tissue, as well as in astrocyte conditioned media (ACM) and cerebrospinal fluid. Targeting polyP in ALS/FTD astrocytes or in derived ACM prevents motoneuron (MN) death. These findings reveal that polyP released by ALS/FTD astrocytes is a critical factor in non-cell autonomous MN degeneration and a potential therapeutic target for ALS/FTD.

INTRODUCTION

Astrocytes play a critical role in the maintenance of the health and function of neurons. Derangements of astrocyte function have been implicated in various neurodegenerative disorders, including amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), C9ORF72-related ALS-FTD (C9ALS-FTD), Alzheimer’s disease (AD), Parkinson’s disease (PD) and Huntington’s disease (HD). Some neurotoxic factors released by the diseased astrocytes have been identified, such as glutamate and ATP in AD (Orellana et al., 2011), TAU in FTD (Hallmann et al., 2017), α-synuclein in PD (di Domenico et al., 2019), and long-chain saturated lipids in inflammation (Guttenplan et al., 2021). For ALS and C9ALS-FTD (together termed ALS/FTD), there is compelling evidence that astrocytic non-cell autonomous toxicity is a major cause of motoneuron (MN) cell death, which involves the release of soluble factor(s) by mouse and human astrocytes harboring disease-causing mutant genes (Van Harten et al., 2021). To date, the identity of the neurotoxic factors released by ALS/FTD astrocytes has not been established.

ALS and FTD form a continuous spectrum of aggressive neurodegenerative diseases, affecting primarily MNs and frontotemporal neurons, respectively (Ling et al., 2013; Ferrari et al., 2019). The majority of ALS patients have the sporadic form of the disease (sALS), but about 10% of the cases have familial ALS (fALS), which is associated with pathogenic mutations in genes such as superoxide dismutase 1 (hereafter mutSOD1), transactive response DNA-binding protein 43 (TARDBP encoding TDP-43; mutTDP43), and C9ORF72 (mutC9ORF72) (Taylor et al., 2016; Renton et al., 2014), the latter of which is characterized by an intronic hexanucleotide expansion. Patients carrying mutC9ORF72 and mutTDP43 can suffer from FTD, ALS or simultaneously both diseases, and thus exhibit motor dysfunction as well as cognitive impairment (Ling et al., 2013; Ferrari et al., 2019). Independently of genotype, sporadic and familial ALS and ALS/FTD patients share many clinical and histopathological features, an observation that suggests a convergent, common mechanistic pathway in different subsets of ALS/FTD. Elucidating the common elements in ALS/FTD is thus an important objective en route to developing widely applicable and effective diagnostic and treatment options for these devastating diseases.

Increasing evidence from animal models implicates astrocytes in the pathology of ALS/FTD (Clement et al., 2003; Lepore et al., 2008; Papadeas et al., 2011; Qian et al., 2017; Tong et al., 2013; Wang et al., 2011; Yamanaka et al., 2008). Astrocytic non-cellautonomous processes have been demonstrated in many studies using cocultures of healthy wild-type MNs together with mouse or human astrocytes harboring disease-causing mutant genes (Di Giorgio et al., 2007; Haidet-Phillips et al., 2011; Nagai et al., 2007; Vargas et al., 2006; reviewed in Van Harten et al., 2021). Moreover, using astrocyte-conditioned media (ACM) from mutated astrocytes, it has been demonstrated that ALS/FTD astrocytes release soluble factors that induce non-cell autonomous toxicity in MNs. Thus, ACM from mouse and human astrocytes with known mutations (SOD1, TARDBP, C9ORF72 genes) (Birger et al., 2019; Fritz et al., 2013; Haidet-Phillips et al., 2011; Ikiz et al., 2015; Jury et al., 2020; Mishra et al., 2020; Nagai et al., 2007; Rojas et al., 2014; Varcianna et al., 2019) and unknown causes (sporadic cases) (Haidet-Phillips et al., 2011; Re et al., 2014) kill MNs in vitro. Chronic infusion of ACM from mutSOD1 astrocytes also led to spinal MN degeneration and neuromuscular dysfunction in healthy rats (Ramírez-Jarquín et al., 2017). Mechanistically, ALS-ACM triggers MN death by inducing hyperexcitability that leads to Ca²⁺ overload and oxidative stress (Fritz et al., 2013; Rojas et al., 2014). MN hyperexcitability is an early and critical feature in ALS models (Devlin et al., 2015; Wainger et al., 2014; van Zundert et al., 2008, 2012) and patients (Geevasinga et al., 2016), and is mediated by increased activation of voltage-sensitive sodium (Na_v) channels and/or inactivation of voltage-sensitive potassium (K_v) channels. While the generation of ALS/FTD-ACM has increased the potential for identifying the toxic factors that kill MNs, standard biochemical approaches searching for the presence of candidate small organic molecules, such as neuro- and glio-transmitters and cytokines/chemokines (i.e. glutamate, ATP, and TNFα) (Nagai et al., 2007; Re et al., 2014) or of peptides/proteins by unbiased mass spectrometry (Mishra et al., 2021; our unpublished data), thus far have been inconclusive. Hence, there is a need to further expand our knowledge of potentially neurotoxic molecules secreted by ALS/FTD astrocytes. Here, we address the role of astrocyte-secreted polyphosphate (polyP) as an adverse pathogenic molecule in ALS.

PolyP is a simple, linear (unbranched), inorganic polymer that comprises 3 to hundreds of ortho-phosphate (Pi) residues linked by high-energy phosphor-anhydride bonds. PolyP is found in every tested cell type in nature, conserved across more than a billion years of evolution (Kornberg et al., 1999; Xie and Jakob, 2019). Initial studies from Kornberg and colleagues in bacteria and yeast, which contain high concentrations of cytoplasmic polyP, revealed numerous vital physiological functions for polyP, including serving as a source of energy, a Pi reservoir, and a chelator for divalent cations (Kornberg et al., 1999; Rao et al., 2009). In contrast to the specific yeast exopolyphosphatase 1 (PPX1) and the bacterial PPX, that hydrolyze polyP into inorganic phosphate (Pi), an equivalent mammalian polyphosphatase enzyme has not so far been identified (but see Baev et al. 2020 and Samper-Martín et al., 2021 for mammalian proteins with polyP phosphatase activity). Recently developed methods for detecting and manipulating the level of endogenous polyP have yielded insights into the function of polyP in brain cells (Desfougères et al., 2020; Xie and Jakob, 2019). Results from primary neuronal cultures indicate that polyP is a neuroactive compound that modulates the activity of several ion channels, including the Na_v and K_v channels, thereby enhancing neuronal excitability (Stotz et al., 2014). Additional reports show that polyP can be released from primary wild-type mouse astrocytes, supporting the concept that polyP acts as a glial transmitter to mediate communications between astrocytes and neurons (Angelova et al., 2018; Holmström et al., 2013).

In this study, we systematically investigated and confirmed that the inorganic molecule polyP is enriched in diverse human and mouse ALS/FTD astrocytes and that its excessive release kills MNs by non-cell autonomous mechanisms.

RESULTS

PolyP is enriched in primary mouse ALS/FTD astrocytes that carry mutations in SOD1, TARDBP and C9ORF72

We have reported that conditioned media generated from primary mouse ALS astrocytes (ACM-ALS from SOD1^G93A and TDP43^A315T astrocytes, see below) causes MN death through neuronal hyperexcitability, which can be prevented by applying the polyamine spermidine to the ALS-ACM (Fritz et al., 2013; Rojas et al., 2014) or by incubating ALS-ACM with glass beads (data not shown). Interestingly, it has been shown that inorganic polyP is released from astrocytes (Angelova et al., 2018; Holmström et al., 2013), enhances neuronal excitability (Stotz et al., 2014), interacts with positively charged polyamines through molecular complementarity and has a high affinity for glass (Kornberg et al., 1999). These findings, along with the longstanding difficulty in identifying toxic organic molecules within ALS-ACM (Mishra et al., 2021; our unpublished data), led us to hypothesize that excessive inorganic polyP is released by ALS/FTD astrocytes to induce neuronal hyperexcitability and subsequent MN death.

To confirm the presence of polyP (Figure 1A) in ALS/FTD astrocytes, we evaluated three staining methods that use fluorescence microscopy to detect this polymer (Desfougères et al., 2020). (i) We used DAPI (4′,6-diamidino-2-phenylindole) (Figure 1B), a non-selective fluorescent dye that labels DNA (DAPI-DNA) at the blue emission spectrum (Ex: 405; Em: 450-485 nm), but that shifts its emission to the green spectrum (Ex: 488 nm; Em: 510-560 nm) when interacting with polyP (DAPI-polyP) (Aschar-Sobbi et al., 2008; Martin and Van Mooy, 2013). A caveat of this widely used polyP probe is that cell-free assays demonstrated that it can interact with a variety of molecules, particularly nucleic acids, and hence interfere with the DAPI-polyP detection. (ii) We therefore also used the recently developed fluorescent dye JC-D8 (Figure 1B; Ex: 488 nm; Em: 510-560 nm), which reveals polyP in living mammalian cells and has been shown in cell-free assays to exhibit a sharp increase in fluorescence intensity upon binding to polyP. This increased signal is not detected for other P-containing molecules or polyanions, except for heparin (Angelova et al., 2014). (iii) We additionally used a recombinant polyP binding domain (termed “recPPBD”) (Figure 1B), derived from the E. coli PPX and that is fused to an Xpress epitope tag; this tag permits detection of polyP in yeast (Saito et al., 2005) and mammalian cells (Samper-Martín et al., 2021) by an immunostaining assay (on fixed and permeabilized cells) with an antibody against Xpress. Competitive binding assays for recPPBD with ³²P-labeled long-chain synthetic polyP and other P-containing molecules (i.e. DNA, RNA, sodium phosphate [P1]), and sodium polytriphosphate [P3]) revealed that recPPBD is highly selectivity for long-chain polyP (Saito et al., 2005).

Figure 1. Intracellular polyP level is elevated in primary mutSOD1, mutTDP43 and mutC9ORF72 mouse ALS/FTD astrocytes.

(A) Schematic of polyP (n=10-1000). (B) Schematics of the recPPBD immunostaining assay and dyes DAPI-polyP and JC-D8 to detect polyP. Confocal images and Pearson’s correlation coefficients showing co-localization (yellow) of recPPBD (white/red) with DAPI-polyP (white/green) or JC-D8 (white/green) in mutSOD1 astrocytes. Nuclei (blue) are detected with DAPI-DNA or TOPRO3. Scale bar 10 μm. (C) Confocal images of mutSOD1, mutTDP43, mutC9ORF72 and non-transgenic littermate (NTg from mutSOD1 shown here) astrocytes stained with recPPBD and TOPRO3. Scale bar 10 μm. (D-F) Quantification of polyP levels in cytoplasm of mutSOD1 (D), mutTDP43 (E) and mutC9ORF72 (F) astrocytes, versus their NTg astrocytes, determined with recPPBD (upper graphs), JC-D8 (middle graphs) or DAPI-polyP (lower graphs). Graphs show mean±S.E.M. *P<0.05, **P<0,01; unpaired Student’s t-test versus NTg (n=3-4 cultures). See also Figures S1–4.

Prior to analyzing primary astrocytes, we expanded the cell-free assays with JC-D8, DAPI and recPPBD to further confirm their specificity to detect polyP (Figure S1) and to identify potential interactions with ‘interfering’ molecules, including other polyanionic molecules (RNA, DNA, heparin, chondroitin sulfate [CS]), small P-containing molecules (ATP, GTP, P1), proteins (BSA, FBS and in some experiments syndecan-4 and brevican). Conversely, we performed competition assays with cell-based systems (N2a cells) to test whether the fore-mentioned molecules could interfere with the detection of synthetic polyP by the three probes (Figure S2). Given that polyP size can vary and polyP chains as large as 800 Pi residues have been reported in the brain using ³²P-based labeling (Kumble and Kornberg, 1995), synthetic polyP of different polymer lengths were used in these assays, including: P3 (the shortest polyP), polyPS (short; 14), polyP_M (medium; 60), polyP_L (large; 130), and polyP_XL (extra-large; 700, ranging from 200-1300 Pi). In agreement with previous studies, our cell-free and cell-based assays showed that JC-D8, DAPI and recPPBD display increased interaction with longer polyP polymers (Figure S1 and S2). Regarding the potential interfering molecules, recPPBD only bound significantly to heparin (but not to the heparan sulphate proteoglycan syndecan-4) in the cell-free assay (Figure S1D); none of the molecules under evaluation affected the recPPBD-polyP signal in the competitive cell-based assays (Figure S2A). JC-D8 was less selective, as several molecules significantly interfered with the polyP signal in the cell-free (i.e. heparin) and cell-based (e.g. heparin, DNA and GTP) assays (Figure S1B and S2B). DAPI-polyP showed the lowest specificity for polyP, as all the tested non-proteinaceous molecules altered the fluorescent signal in both assays (Figure S1C and S2C). Together, these data indicate that recPPBD, and to a lesser extend JC-D8, can be used to detect polyP in cell-free and cell-based assays with N2a cells. Given that the in vitro data also indicate that heparin (and possible heparin-bound proteins other than syndecan-4) can potentially lead to a recPPBD signal, we evaluated the recPPBD staining in astrocytes expressing the specific polyP phosphatase PPX1, which binds to but does not hydrolyze heparin (Labberton et al., 2016; Andreeva et al., 2019; data not shown). AAV9-mediated expression of PPX1 (Abramov et al., 2007) in astrocytes in culture (Figure S3A) and in spinal cord tissue (Figure S15A) robustly decreased the recPPBD signal, indicating that this probe selectively detects polyP in our study. Similar results were found in neurons (Figure S15A) and HEK cells (Samper-Martín et al., 2021).

Next, we utilized recPPBD together with the JC-D8 or DAPI-polyP dyes to determine the subcellular localization of endogenous polyP in primary spinal cord astrocyte cultures derived from mutant SOD1^G93A (mutSOD1) mice; these ALS glial cells were selected as they display significantly more intracellular polyP staining signal when compared to wild-type astrocytes (see below). For nuclear counterstains, DAPI-DNA or TOPRO3 was used. Note that TOPRO3 and NucBlue (Hoechst 33342), unlike DAPI-polyP, do not shift their emission wavelength in the presence of polyP. Confocal microscopy (Figure 1B) revealed that the distribution of the polyP signal detected with recPPBD in primary spinal cord astrocytes was compartmentalized, with high-intensity puncta in the cytoplasm (middle images). Co-staining assays along with confocal microscopy further showed that the recPPBD signal co-localized with cytoplasmic DAPI-polyP (upper images) and JC-D8 in astrocytes (lower images). In addition to the expression of PPX1 (Figure S3A; discussed above), another specificity experiment was performed where the probes recPPBD, JC-D8 and DAPI-polyP were pre-incubated with an excess of synthetic polyP_L before their application to mutSOD1 astrocytes. Following polyP_L pre-absorption, the detection of endogenous polyP in mutSOD1 astrocytes was strongly reduced (Figure S3B–D). These findings indicate that endogenous polyP in the cytoplasm of cultured astrocytes (i.e., vesicles, see below) can be reliably detected by three independent polyP staining methods.

Using the staining methods, we next assessed the intracellular polyP level in primary astrocytes cultures derived from ALS mutSOD1 mice and two other well-studied transgenic ALS/FTD mouse models: ALS mutTDP43 and C9-ALS/FTD mutC9ORF72 (see Methods). Given the diversity of the backgrounds of these ALS/FTD mice, astrocytes from non-transgenic (NTg) littermates were used as control astrocytes for each ALS/FTD model. Weak polyP signals were detected in all of these NTg astrocytes (Figure S4A–C), as well as in astrocytes derived from wild-type mice and from transgenic mice over-expressing wild-type human SOD1 gene (Figure S4D). Importantly, confocal microscopy revealed that polyP staining with recPPBD, JC-D8 and DAPI-polyP was substantially higher in mutSOD1, mutTDP43 and mutC9ORF72 astrocytes (Figure 1C, Figure S4A–C). Quantification indicated that the average fluorescence intensity obtained with recPPBD (upper graph), JC-D8 (middle graph) and DAPI-polyP (lower graph) was ~2-fold higher in the cytoplasm of astrocytes that harbor mutSOD1 (Figure 1D), mutTDP43 (Figure 1E) or mutC9ORF72 (Figure 1F), compared to their NTg controls. We conclude that polyP is enriched in primary astrocyte cultures derived from three distinct and unrelated ALS/FTD mouse models.

Primary mouse ALS/FTD astrocytes release increased levels of polyP

Recent studies using co-cultures of neurons and glia have shown that polyP localizes within lysosomal vesicles that can be released from astrocytes into the extracellular space upon treatment with Ca²⁺ ionophores (Angelova et al., 2018; Holmström et al., 2013), after which polyP can be subsequently taken up by neurons (Holmström et al., 2013). In agreement with these findings, we showed using total internal reflection fluorescence (TIRF) microscopy that JC-D8-labeled mutSOD1 and NTg astrocytes can release vesicular polyP after treatment with the Ca²⁺ ionophore ionomycin (Figure S5). Next, we used a polyP deposition assay to gain insight into the level of polyP in conditioned media from the three categories of mutant ALS/FTD astrocytes -mutSOD1-ACM, mutTDP43-ACM and mutC9ORF72-ACM- each of which are highly toxic to MNs (Fritz et al., 2013; Ikiz et al., 2015; Jury et al., 2020; Mishra et al., 2020; Nagai et al., 2007; Rojas et al., 2014). ACM from NTg astrocytes (NTg-ACM) and feeding medium (Ctrl-medium) were included (internal controls) and shown to be innocuous for cultured MNs. To visualize the morphology of neurons, cultures were infected with AAV2 virus carrying the gene coding the fluorescent protein tdTomato under the neuronal synapsin promoter (AAV-synapsin-tdTomato). At 7-8 DIV, cultures were treated with the diverse ACMs or positive control polyP_L (50 μM) for 5 min and incubated with the JC-D8 probe to detect polyP. We detected an accumulation of the JC-D8 signal on or near neuronal membranes of tdTomato-positive neurons, including MNs, when exposed to ACM from mutSOD1, mutTDP43 or mutC9ORF72 astrocytes, relative to their corresponding control media (Figure 2A). Signal quantification revealed that the average JC-D8-positive fluorescence intensity of tdTomato-positive spinal cord neurons was ~2-3-fold higher when exposed to ALS/FTD-ACMs compared to NTg controls (Figure 2B). Together, these data indicate that the concentration of polyP is increased in ALS/FTD-ACMs compared to control media.

Figure 2. Primary mouse ALS/FTD astrocytes release higher levels of polyP, which is deposited on, or near to, neuronal plasma membranes.

(A) Confocal images of synapsin-tdTomato-expressing spinal cord neurons (red) stained with JC-D8 (white) showing increased polyP deposition on, or in proximity to, neuronal membranes (arrows: MNs) when treated with mutSOD1-ACM, mutTDP43-ACM, mutC9ORF72-ACM or synthetic polyP (50 μM polyP_L) versus ACMs from controls (NTg-ACM, Ctrl-medium). Scale bar 10 μm. (B) Quantification of JC-D8 fluorescence after applying the ACMs and controls, as indicated (mean±S.E.M. *P<0.05, **P<0.01; unpaired Student’s t-test; n=3 cultures). (C) Confocal images showing immunofluorescence for MAP2 (blue), SMI32 (green) and ChAT (red) in ventral spinal cord cultures. All 3 neuronal markers co-localize (white arrows); while ChAT and SMI32 can be used to identify MNs, MAP2 can detect INs (yellow arrowheads). Scale bar 20 μm. (D-E) Confocal images for the polyP probe recPPBD (white), for the IN marker MAP2 (blue), and for the MN markers ChAT (red) (D) or SMI32 (green). Scale bar 10 μm. (E). Images and quantification show an equal deposition of polyP on, or near, membranes of MNs (ChAT⁺ or SMI32⁺; white arrows) and INs (MAP2⁺ only; yellow arrowheads) when treated with mutSOD1-ACM (mean±S.E.M. *P>0.05 (not significant, ns); unpaired Student’s t-test MN versus IN). (F-G) P/C-extracted polyP from NTg-ACM or mutSOD1-ACM, either digested by recPPX/PPase and polyP-Pi levels colorimetrical quantified by malachite (F), or not digested to fluorometrical measure polyP-polymer levels by JC-D8 (G) (mean±S.E.M. **P<0.01; unpaired Student’s t-test; n=3-4 independent ACM analyzed in duplicate). See also Figure S5.

Given that ALS/FTD-ACM specifically kills MNs in spinal cord cultures, we next asked if the polyP released by mutSOD1 astrocytes is preferentially targeting MNs versus interneurons (INs). A drawback of the JC-D8-based staining method is that this dye largely loses its fluorescent signal when cells are processed for immunostaining analyses. Hence, we used the recPPBD probe to detect the polyP signal on, or in proximity to, membranes of MNs and INs exposed to mutSOD1-ACM; note that only limited recPPBD signal was detected in the absence of ACM (data not shown). Typically, cells in primary spinal cord cultures are fixed at 7 DIV and double immunostained for unphosphorylated neurofilament-H (SMI32) and MAP2 to identify MNs (SMI32⁺/MAP2⁺-cells) and INs (SMI32⁻/MAP2⁺-cells) (Fritz et al., 2013; Mishra et al., 2020; Nagai et al., 2007). We also identified MNs by using an antibody against choline acetyltransferase (ChAT; Figure 2C). Confocal images and subsequent quantification revealed that both MNs (ChAT⁺/MAP2⁺) and INs (ChAT⁻/MAP2⁺) exhibited comparable recPPBD signals when treated with ACM-mutSOD1 for 5 min (Figure 2D). Similar results were obtained when MNs were identified with SMI32 staining (Figure 2E).

Next, we used two approaches to quantify polyP levels in mutSOD1-ACM. Determining the polyP concentration in mammalian samples, particularly in ACM, is appealing but represents a unique challenge due to the low polyP concentration (in the μM range versus mM range in yeast and bacteria) and the long polyP size chains reported in brain tissue. Therefore, using methods such as ³¹P NMR that require high sample concentrations (mM range) or electrospray ionization mass spectrometry (ESI-MS) that can determine only very short-chain polyP are not suitable for mammalian samples (Christ et al., 2020). Alternatively, enzymatic assays have been used to measure the concentration of long-chain polyP in diverse types of organisms. These assays include three steps: i) extraction of polyP from the sample, ii) digestion of polyP with recombinant PPX (recPPX1), either alone or in combination with recombinant pyrophosphatase 1 (recPPX/PPase), and iii) radioactive, colorimetric or fluorometric detection of the released Pi (Christ et al., 2020). In the rodent brain, radioactive (Kumble and Kornberg, 1995) and fluorometric (Lee et al., 2018) methods detected 95 μM and 48 μM polyP (measured as Pi), respectively. To isolate and purify polyP from mutSOD1-ACM and NTg-ACM, we developed a modified version of a phenol-chloroform (P/C) extraction protocol originally devised for yeast (Christ and Blank, 2018). Next, polyP was enzymatically hydrolyzed with recPPX/PPase and the released Pi measured by a malachite-based colorimetric assay (Bru et al., 2016; Christ and Blank, 2018). Standard curves were generated using synthetic polyP_L (dissolved in water) and Pi to then calculate the ACM polyP concentration and estimate the polyP recovery ratio (RR, 73±9%). Malachite quantification of P/C-extracted polyP (Figure 2F) confirmed that mutSOD1-ACM contained a ~4.0-fold higher level of polyP-Pi compared to NTg-ACM. In an alternative assay, P/C-extracted polyP was not digested with PPX/PPase; instead, the polyP polymer was directly quantified fluorometrically using the JC-D8 probe. Quantification demonstrated that mutSOD1-ACM contained ~1.4-fold higher levels of polyP-polymer compared to NTg-ACM (Figure 2G). Together, and in agreement with the polyP deposition staining assay, our two analytical strategies demonstrate that there is a significantly higher polyP concentration in mutSOD1-ACM compared to control ACM.

PolyP is enriched in spinal cord and motor cortex sections of ALS/FTD mouse models

To determine whether polyP is also elevated within spinal cord tissue samples (ventral horn in lumbar 4-6) derived from the three symptomatic ALS/FTD transgenic mice described above, we used recPPBD fluorescent staining along with confocal microcopy. Next, we examined mutSOD1 mice (4-month-old) and their NTg littermates and performed double immunostaining using the polyP probe recPPBD and antibodies that target ChAT (Figure 3A), NeuN (Figure 3B) and the glial fibrillary acidic protein (GFAP; Figure 3C) to identify MNs, neurons and astrocytes, respectively. The NeuN-positive cells are predominantly MNs based on their large size and localization in the ventral horn of the spinal cord. As expected for the mutSOD1 ALS mouse model, there was a significant loss of ChAT- and NeuN-positive MNs in ventral horn of the spinal cord. The remaining ChAT- (Figure 3A) and NeuN- (Figure 3B) positive MNs revealed an increased recPPBD signal consisting of high-intensity cytoplasmic puncta. We also found that mutSOD1 spinal cord, unlike those in control tissue, exhibited astrogliosis and a marked cytoplasmic recPPBD immunoreactivity (Figure 3C). The average recPPBD fluorescence intensity signal in the cytoplasm of GFAP- (Figure 3D₁) and NeuN- (Figure 3D₂) positive cells was ~5-fold higher in tissue from mutSOD1 mice compared to NTg controls. Evident recPPBD labeling in the mutSOD1 spinal cord samples was also detected outside GFAP-positive cells as well as in small GFAP-negative cells, potentially microglial cells (Figure 3C).

Figure 3. PolyP is enriched in spinal cord sections of ALS/FTD mouse models.

(A-C) Confocal images for recPPBD (white) with MN (ChAT, red) (A), neuronal (NeuN, red) (B), and astrocyte (GFAP, green) (C) markers in lumbar (4-6) ventral lumbar spinal cords of NTg and mutSOD1 mice. Scale bar 20 μm. Lower images/insets: higher magnification images showing recPPBD-immunoreactive puncta in ChAT, NeuN-, and GFAP-positive cells in mutSOD1 sections. Insets: scale bar 10 μm. For the low-magnification images in A, the dashed lines indicate the borders between the gray and white matters. Arrowheads in C indicate small GFAP-negative cells, potentially microglial. (D-F) Quantification of recPPBD fluorescence in cytoplasm of GFAP- (D₁,E₁,F₁) and NeuN-positive cells (D₂,E₂,F₂) in sections from mutSOD1 (D), mutC9ORF72 (E), and mutTDP43 (F) mice and each corresponding NTg. Graphs show mean±S.E.M. **P<0.01; unpaired Student’s t-test versus NTg (n=3-4 animals). See also Figures S6–8.

We also examined spinal tissue samples derived from symptomatic mutTDP43 (2-month-old) and mutC9ORF72 (9-month-old) mice, and their NTg littermates as controls. Quantification revealed significantly increased recPPBD immunoreactivity in GFAP-positive and NeuN-positive cells in spinal cord sections of mutC9ORF72 (Figure 3E) and mutTDP43 (Figure 3F) (see Figure S6 and S7 for additional confocal microscopy images and controls, respectively). Another set of experiments, performed with JC-D8 live staining, also revealed higher polyP labeling in NeuN-positive and -negative cells in the motor cortex of mutSOD1 and mutC9ORF72 animals relative to NTg littermates and transgenic wtSOD1 mice (Figure S8).

Exposure of wild-type spinal cord neurons to polyP reproduces the toxic effects of mutSOD1-ACM

Our earlier experiments demonstrated that application of mutSOD1-ACM to cultures of wild-type spinal cord neurons rapidly induces hyperexcitability, leading to excessive Ca²⁺ influx and MN death within a few days (Fritz et al., 2013; Rojas et al., 2014). We therefore explored whether application of synthetic polyP mimics the toxicity of ACM derived from mutSOD1 astrocytes. Because the chain length of polyP can be as large as 800 Pi residues in rodent tissue (Kumble and Kornberg, 1995), we used synthetic polyP with varying polymer size ranges: polyP_S, polyP_M, polyP_L. (Figure 4A) These synthetic polyP polymers were applied to 3 DIV primary spinal cord cultures, at concentrations up to 100 μM, to mimic the conditions found in CNS tissue (Kumble and Kornberg, 1995; Lee et al., 2018) and in mutSOD1-ACM (Figure 2). The survival of MNs and INs was assessed at 7 DIV using SMI-32 and MAP2 immunostaining. Quantification of MN survival revealed that a single application of 5 μM polyP_L significantly reduces MN survival and that 50 to 100 μM polyP_L kills ~40% of MNs (Figure 4A). The toxic activity of the shorter synthetic polyP polymers, polyP_s and polyP_M, was less evident, specially at lower concentrations.

Figure 4. Application of synthetic polyP to wild-type spinal cord neurons reproduces the toxicity effect of mutSOD1-ACM.

(A) Fewer MNs (SMI32⁺/MAP2⁺ cells; white arrow) found in 7 DIV spinal cord cultures treated (4 days) with mutSOD1-ACM or synthetic polyP versus controls (NTg-ACM and wtSOD1-ACM) as indicated (n=3 cultures). (B) Venn diagram and functional annotation analysis of the 97 commonly expressed transcripts in spinal cord cultures: polyP_L versus wtSOD1-ACM and mutSOD1-ACM versus wtSOD1-ACM. (C-D) Representative calcium activity fingerprint (C) and whole-cell patch-clamp recordings (D) showing in individual cells that polyP_L (100 μM) and mutSOD1-ACM increase intracellular Ca²⁺ transients and spontaneous AP firing, respectively. KCl (50 mM) induces Ca²⁺ rises in all cells. Graphs show mean±S.E.M. *P<0.05, **P<0.01, and ***P<0.001; one-way ANOVA (A,C) and paired Student’s t-test (D) versus controls (NTg-ACM, Ctr-medium or wtSOD1-ACM). See also Figures S9–10.

Transcriptomic analyses further revealed that treatment of spinal cord cultures for 90 min with polyP_L (100 μM) or mutSOD1-ACM leads to a common set of differentially expressed mRNAs (97 genes); functional annotation analysis (GO for biological process, molecular function, and cellular components) underscored that these genes are functionally enriched in inflammatory-related events (Figure 4B and Figure S9).

Whole-cell patch clamp electrophysiological recordings and Ca²⁺ imaging (measured with fura-2) of primary spinal cord neurons were used to determine whether polyP_L (100 μM) mimics the membrane action potential (AP) and cytoplasmic Ca²⁺ dynamics induced by mutSOD1-ACM. Application of synthetic polyP to dorsal root ganglion and hippocampal neurons has been shown to enhance neuronal excitability and Ca²⁺ signaling (Stotz et al., 2014). In line with this report, we found that polyP_L rapidly and significantly increased the intracellular Ca²⁺ transients (Figure 4C) as well as the firing of spontaneous APs (Figure 4D), and at a similar extent in both in MNs and INs (Figure S10 and see discussion).Together, these data show that addition of synthetic polyP to spinal cord neurons reproduces the effects of ALS-ACM by both increasing excitability and Ca²⁺ transients, along with the altered expression of inflammatory-related gene transcripts, effects that can trigger MNs death.

Targeting polyP in mutSOD1 mouse astrocytes or in ALS/FTD-ACM attenuates the intracellular Ca²⁺ transients and prevents MN death

To confirm that polyP is a toxic component in mutSOD1-ACM, we assayed the toxicity of ACM from mutSOD1 astrocytes after targeting polyP within the ALS astrocytes or in the derived ACM. To reduce polyP in mutSOD1 astrocytes, cells were transduced with AAV9 vectors carrying different versions of the yeast PPX1 gene, fused to GFP, and driven by a CMV promoter. Our experimental approach included the following two constructs: (i) PPX-GFP, where the wild-type enzyme is ubiquitously expressed in the cytoplasm (Abramov et al., 2007); and (ii) PPX-PLong-GFP, where PPX-GFP was genetically fused to the membrane-permeant peptide PTEN-Long–specific region (Hopkins et al., 2013). This latter chimeric protein was designed to impart PPX-GFP with a mechanism for secretion and re-uptake into cells, and thus establish a bystander effect. Three days after mutSOD1 astrocytes were infected with AAV9 vectors, ACM was collected and added to the primary spinal cord cultures. It was found that ACM derived from mutSOD1 astrocytes transduced with the PPX-coding constructs, particularly with PPX-PLong-GFP, was significantly less toxic to MNs than ACM from untreated mutSOD1 astrocytes (Figure 5A).

Figure 5. Reducing polyP in ALS astrocytes, or targeting polyP in ALS/FTD-ACM, prevents MN death and restores intracellular Ca²⁺ transients.

(A) Expression of PPX1 in mutSOD1 astrocytes (using AAV9-PPX-GFP or AAV9-PPX-PLong-GFP) lowers toxicity of mutSOD1-ACM to MNs. (B-E) Treatment with the polyP degrading enzymes recPPX/PPase or CIP (B), or with the polyP neutralizing molecules recPPBD (but not BSA) (C), G4-PAMAM-NH2 (D), or UHRA9/10 (E) rescues MNs from mutSOD1-ACM and polyPL (100 μM). Schematics of the different polyP sequestering molecules are depicted. (F-J) Application of UHRA10 to mutSOD1-ACM (F), mutTDP43-ACM (G-H) or mutC9ORF72-ACM (I-J) reduces MN death (G,I) and Ca²⁺ transients (F,H,J). Graphs show means±S.E.M. **P<0.01, and ***P<0.001; one-way ANOVA versus controls (NTg-ACM or Ctrl-medium). #P<0.05, ##P<0.01, and ###P<0.001; one-way ANOVA versus mutSOD1-ACM or polyP_L (n=3 cultures). See also Figure S11.

Next, we determined if MN death can be prevented by degrading or neutralizing polyP in ALS/FTD-ACM, using two exopolyphosphatases: recPPX/PPase from yeast and the alkaline phosphatase from calf intestine (CIP), known to hydrolyze polyP in vitro. We found that pre-treatment of mutSOD1-ACM with either recPPX/PPase or CIP significantly protected MNs from death (Figure 5B). In another set of experiments, we applied the recPPBD protein to mutSOD1-ACM, aiming to sequester and neutralize extracellular polyP released by mutSOD1 astrocytes and hence prevent its toxic effects on MNs. Treatment of mutSOD1-ACM with recPPBD (20-40 μg/ml), but not with a control protein (40 μg/ml BSA), prior to applying it to spinal cord cultures effectively prevented, in a concentration-dependent manner, the death of MNs by mutSOD1-ACM (Figure 5C). We further explored the beneficial effects of neutralizing extracellular polyP by using synthetic nano-sized tree-like branched polycationic compounds. In particular, we evaluated poly-amidoamine (PAMAM) and universal heparin reversal agents (UHRAs), which are nano-sized cationic branched polymers shown to efficiently bind and neutralize polyP (Travers et al., 2014; Smith et al., 2012). Application of the dendrimer G4-PAMAM-NH2 (14 kDa, 1 μg/mL), containing an ethylenediamine core and 64 positively charged primary -NH2 groups on the surface, significantly reduced MN death when applied together with mutSOD-ACM to spinal cord cultures (Figure 5D). Additionally, we evaluated UHRA9 (16 kDa) and UHRA10 (10 kDa), which were engineered to contain 16 and 11 polyP-binding groups (with each containing positively charged tertiary amines), respectively (Travers et al., 2014). MN death was abolished when UHRA9 (250 nM) or UHRA10 (250-500 nM) were added to the mutSOD-ACM (Figure 5E). We also found that application of UHRA10 to mutSOD-ACM reduced the number of events of the intracellular Ca²⁺ transients measured with GCaMP6 to control values (Figure 5F). Additional studies with a mutant SOD1 Drosophila model provide in vivo evidence for the importance of polyP as a neurotoxin in the presence of this mutant ALS gene. Thus, administration of the nano-sized branch polymer UHRA10, which binds and neutralizes extracellular polyP in the blood of treated mice (Travers et al., 2014), rescues in adult flies locomotion deficits and prolongs mean lifespan without causing toxicity (Figure S11).

We also assessed whether the toxic component present in mutTDP43-ACM and mutC9ORF72-ACM can be inhibited by the treatment with these nano-sized poly-cationic compounds. Treatment of mutTDP43-ACM (Figure 5G–H) or mutC9ORF72-ACM (Figure 5I–J) with UHRA10 (500 nM) significantly reduced the numbers of ACM-induced intracellular Ca²⁺ transient events in neurons and protected MNs from death. Together, these findings show that the polyP present in astrocyte-conditioned media derived from diverse ALS/FTD astrocytes is a critical neurotoxic factor in non-cell autonomous MN degeneration.

Human mutTDP43 astrocytes and derived ACM is toxic to MNs and exhibit a higher polyP concentration

Next, we evaluated the toxicity of ACM from human astrocytes derived from induced pluripotent stem cells (iPSCs) from an ALS/FTD subject carrying a A90V mutation in TARDBP (mutTDP43) and a healthy family member (control) (Zhang et al., 2013). An efficient mature astrocyte phenotype was achieved for both control and mutTDP43 patient-derived iPSCs, a stage that is characterized by a high percentage of cells expressing the following mature astrocyte markers (Roybon et al., 2013): Cx43 (>91% positive cells), EAAT2 (>93%) and s100β (>78%) (Figure S12). Comparable to mouse mutTDP43 astrocytes (Figure 1E), confocal microscopy and subsequent quantification demonstrated that the average fluorescence intensity obtained with the recPPBD-based labeling was ~2-fold higher in the cytoplasm of mutTDP43 patient astrocytes compared to control subject astrocytes (Figure 6A–B). Additionally, we also found that MN death was prevented when mutTDP43-ACM was pre-treated with recPPX/PPase or CIP (Figure 6C). JC-D8 quantification of P/C-extracted polyP (68±8% RR) further confirmed that human mutTDP43-ACM contained a ~2-fold higher level of polyP-polymer compared to control (Figure 6D).

Figure 6. PolyP levels are elevated in mutTDP43 patient iPSC-derived astrocytes and degrading polyP in mutTDP43-ACM prevents MN death.

(A) Mature astrocytes were generated from iPSCs derived from an ALS/FTD TDP43 ^A90V patient (mutTDP43) and a healthy control subject (control). Confocal images for recPPBD (white), Nucblue (blue), phalloidin (pink) and mature astrocyte marker s100β (red). Scale bar 10 μm. (B) Quantification of cytoplasmic polyP levels of individual human control and mutTDP43 astrocytes determined with recPPBD (mean±S.E.M. ***P<0.001; unpaired Student’s t-test mutTDP43 versus control; ≥20 cells/condition from 2 independent differentiations). (C) Treatment with of mutTDP43-ACM with the polyP degrading enzymes recPPX/PPase or CIP rescues MNs from death (means±S.E.M. ***P<0.001; one-way ANOVA versus controls control-ACM. ###P<0.001; one-way ANOVA versus mutTDP43-ACM (n=3 cultures). (D) P/C-extracted polyP-polymer from control-ACM and mutTDP43-ACM fluorometrical quantified by JC-D8 (mean±S.E,M. *P<0.05; unpaired Student’s t-test; n=4 from two independent differentiations). See also Figure S12.

ALS patients exhibit polyP enrichment in spinal cord sections and in CSF

To address whether polyP is enriched in fALS (mutSOD1 and mutC9ORF72) and sALS patients, we performed immunohistochemical staining assays with recPPBD on formalin-fixed paraffin-embedded section through the ventral spinal cord. Based on the morphology and size of cells (Cooper-Knock et al., 2012; Forsberg et al., 2011; Kiernan and Hudson, 1991), large MN-like cells and smaller non-neuronal glia-like cells were identified throughout the ventral cord gray matter of control human spinal cord sections. As expected, the spinal cord ventral horn of most, but not all (e.g. sALS#8), ALS patients showed fewer MNs than controls; residual MN-like cells were typically atrophied (Figure 7B,E; Figure S13 for all images, staining controls and patient information). In spinal cord sections from both control (Figure 7A) and ALS (Figure 7B) subjects, recPPBD immunostaining was detected intracellularly and extracellularly. Importantly, the intensity of the polyP staining was higher in ALS patients and quantification revealed that glia-like cells (Figure 7C), but not MN-like cells (Figure 7D), from ALS patients displayed a significantly higher recPPBD immunoreactivity relative to control subjects. These findings indicate that polyP enrichment in spinal and cortical astrocytes represents a novel hallmark of ALS/FTD.

Figure 7. PolyP level is higher in postmortem spinal cord tissues and CSF from fALS and sALS patients.

(A-B) Representative immunohistochemistry micrographs for recPPBD in spinal cords from control (A) and ALS (C9ORF72) (B) subjects. In ALS, magnifications show recPPBD-immunoreactivity in glial-like cells (box#1, red arrows) and smaller MN-like cells (box#2). Scale bar 50 μm (main) and 10 μm (boxes). (C-D) Quantification of recPPBD-immunoreactivity in cytoplasm of glial-like (C) and MN-like (D) cells from control (n=5) and ALS (n=9) tissues. Graphs show mean±S.E.M. *P<0.05; Mann-Whitney U-test versus control. (E) Soma area (y-axis) versus number (x-axis) of MN-like cells. (F) Silica extracted polyP from the CSF of healthy control subjects (Ctrl, n=15) and ALS patients (n=16) and colorimetrical quantified as Pi by malachite (mean±S.E.M. **P<0.01; unpaired two-tailed St6udent’s t-test). See also Figures S13–14.

Next, we assessed whether polyP concentrations are elevated in the cerebrospinal fluid (CSF) of sALS and fALS patients. To quantify CSF polyP concentration, we developed a modified version of a silica extraction protocol originally devised for mammalian tissues (Lee et al., 2018). Specifically, the polyP from CSF (1 ml) of 16 ALS patients and of 15 control subjects (see Figure S14 for additional graphs and patient information) was extracted with silica columns and PPX/PPase-digested Pi was measured with malachite. Quantification revealed that the polyP concentration in ALS patient CSF contained ~1.5-fold higher level of polyP-Pi compared to control subject CSF (75±3% RR with artificial CSF) (Figure 7F). Importantly, these findings indicate that polyP enrichment in CSF may represent a novel biomarker of ALS.

DISCUSSION

We have taken advantage of several independent experimental approaches to demonstrate systematically in vitro and in vivo that levels of inorganic polyP are elevated in mouse astrocytes expressing mutant SOD1, TARDBP or C9ORF72, three genes whose mutations encompass the majority of familial ALS/FTD patients. Similar results were found with human astrocytes in vitro and in vivo. Thus, polyP is enriched in iPSC-derived astrocytes harboring mutant TARDBP and in astrocyte-like cells in post-mortem spinal cord tissue from fALS (SOD1 or C9ORF72) and sALS patients. Our biochemical quantification assays further demonstrate that polyP concentrations are significantly increased in ACM from human and mouse ALS/FTD astrocytes and intriguingly, in CSF of ALS patients. These novel findings collectively indicate that polyP enrichment in astrocytes is a widely present hallmark of ALS/FTD and that an elevated concentration of polyP in CSF might serve as a new biomarker for ALS/FTD.

Importantly, by using four distinct approaches to reduce or sequester intracellular and extracellular polyP, respectively, our in vitro studies document that MNs can be rescued from the toxicity induced by the media conditioned by human and mouse ALS/FTD astrocytes. Conversely, exposure of spinal cord neurons to polyP reproduces the toxic effects of ALS-ACM, causing increased neuronal excitability, increased Ca²⁺ transients, enhanced MN lethality and augmented transcription of genes related to inflammation. These results indicate that excessive inorganic polyP released from cultured human and mouse ALS/FTD astrocytes is both necessary and sufficient to cause non-cell autonomous toxicity to MNs. It is important to emphasize that excessive polyP is not necessarily toxic to other types of neurons. Thus, our data show that extracellular polyP does not kill interneurons (INs). Moreover, recent studies with cultured cortical neurons have shown that application of synthetic polyP (20 μM polyP_S-M-L) protected these neurons from glutamate-induced cell death by activating P2Y receptors (Maiolino et al., 2019). In our present study, we have not identified the specific ion channel involved in the MN toxicity. However, Na_v and K_v channels are strong candidates as they have been shown to be regulated by polyP (Stotz et al., 2014) and reducing neuronal excitability by targeting these channels rescues MNs from death (Fritz et al., 2013; Wainger et., 2014). Given that polyP_L similarly increases the neuronal excitability of MNs and INs (Figure S10), it is likely that this polymer activates lethal intracellular processes in MNs, but not in INs. In this respect, the limited calcium buffering capacity and proteostasis stress response of MNs have been implicated in their intrinsic vulnerability in ALS (van Zundert et al., 2012; Saxena and Caroni, 2011).

Our results demonstrate that enrichment and excessive release of polyP represent a convergent neurotoxic pathway that contributes to the clinical and histopathological overlap in different categories of ALS/FTD patients. We hypothesized that interventions that attenuate levels of polyP in patients may be an innovative therapeutic strategy for diverse types of ALS/FTD. To address this long-term goal, we have begun lowering intracellular polyP levels in brain cells of neonatal (P0) mutSOD1 ALS mice, using intracerebroventricular injections with AAV9-CMV-PPX1-PLong-GFP (PPX1-PLong) or AAV9-CMV-GFP (control) (used in vitro in Figure 5A). While transduction of PPX1-PLong markedly reduced polyP staining signals in astrocytes, neurons, and the neuropil, neither disease onset nor survival of the treated mutSOD1 mice improved significantly (Figure S15). Arguably, in these mutSOD1 mice, MN death might be mediated by converging mechanisms that involves polyP released from quiescent astrocytes as well as by inflammatory factors secreted from reactive astrocytes (Guttenplan et al., 2020). Additionally, our results with mutSOD1 mice likely reflects the complexity of cellular pathways associated with the synthesis, storage, degradation, and secretion of polyP in brain cells. PolyP mediates several critical physiological functions, serving as an energy source, a Pi reservoir, and as a chelator for divalent cations, among other functions (Rao et al., 2009; Xie and Jakob, 2019; Bondy-Chorney et al., 2020). Moreover, recent reports indicate that intracellular polyP provides a scaffold for the amyloidogenic proteins Aβ, tau-P, and α-synuclein, thereby enhancing the nucleation of fibril formation (Cremers et al., 2016). This process appears to be beneficial for brain cells in neurodegenerative diseases, as insoluble aggregates and fibrils found in AD (Aβ, tau-P), PD (α-synuclein) and ALS (SOD1) are less toxic than soluble oligomeric species of the same proteins (Xie and Jakob, 2019; Zhu et al., 2018). These observations suggest that reductions in intracellular polyP levels may have adverse (especially in neurons) as well as beneficial influences in our mutSOD1 mice.

The mechanisms underlying polyP metabolism and secretion of polyP-containing vesicles from astrocytes are not well defined. While yeast and bacterial polyP-degrading and synthesizing enzymes have been studied for decades (Rao et al., 2009; Xie and Jakob, 2019), only recent studies have unveiled mammalian proteins with previous unknown polyP metabolic activity (Baev et al. 2020; Samper-Martín et al., 2021). Identifying the regulatory machinery, including the polyP enzymes and transporters, that control the levels of polyP in astrocytes, and particularly in vesicles, will be critical to understanding the molecular mechanisms of polyP-induced toxicity in ALS/FTD.

In conclusion, our study demonstrates that an inorganic molecule, polyP, contributes to neurodegeneration in ALS/FTD and that the polyP pathway is potentially a novel target for therapy development in ALS/FTD. In the long term, improved understanding of the biological properties and function of inorganic polyP, and the determinants of its levels in the CNS, will enhance the efforts to develop therapy for ALS/FTD and related neurodegenerative disorders.

STAR ★ METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Brigitte van Zundert (bvanzundert@unab.cl).

Materials availability

All unique/stable reagents generated in this study are available from the Lead Contact without restriction.

Data and code availability

• RNA-seq data have been deposited at GEO (https://www.ncbi.nlm.nih.gov/geo/, Accession GEO: GSE196664) and is publicly available.

• No new software or code was generated in this study.

• Any additional information required to reanalyze the data reported in this work paper is available from the Lead Contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Human spinal cord tissues

Human spinal cord tissues were provided by the Target ALS Multicenter Human Postmortem Tissue Core and its staff members, including Lyle Ostrow, M.D., Ph.D., and Kathleen Wilsbach, Ph.D. (Johns Hopkins University), Neil Shneider, M.D., Ph.D. (Columbia University), and John Ravits, M.D. (University of California, San Diego). All available de-identified clinical and pathological records were collected and used together with ALS genotypes to summarize patient demographics and disease features (see Table in Figure S13).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-Xpress	Invitrogen	Cat# R910-25 RRID:AB_2556552
Mouse anti-MAP2A 2B	Sigma-Aldrich	Cat# MAB378 RRID:AB_94967
Rabbit anti-MAP2	Invitrogen	Cat# OSM00030W RRID:AB_10987424
Rabbit anti-SMI32	Abcam	Cat# ab8135 RRID:AB_306298
Mouse anti-SMI32	Abcam	Cat# ab187374 RRID:AB_2904559
Goat anti-ChAT	Sigma-Aldrich	Cat# AB144P RRID:AB_2079751
Rabbit anti-GFAP	Dako	Cat# Z0334 RRID:AB_10013382
Rabbit anti-GFAP	Sigma-Aldrich	Cat# G3893 RRID:AB_477010
Rabbit anti-NeuN	Sigma-Aldrich	Cat# ABN78 RRID:AB_10807945
Mouse anti-NeuN	Millipore	Cat# MAB377 RRID:AB_177621
Rabbit anti-S100	Dako	Cat# Z0311 RRID:AB_10013383
Mouse anti-SOD1	Abcam	Cat# ab238052 RRID:AB_2904560
Mouse anti-misfolded SOD1	Medimabs	Cat# MM-0070-p RRID:AB_10015296
Rabbit anti-GFP	Invitrogen	Cat# A-11122 RRID:AB_221569
Mouse anti-REPO	DSHB	Cat# 8D12 RRID:AB_528448
Mouse anti-Cx43	Invitrogen	Cat# 13-8300 RRID:AB_2533038
Rabbit anti-EAAT2	Invitrogen	Cat# PA5-17099 RRID:AB_10978571
Mouse anti-Aldh1L1	NeuroMab	Cat# 75-140 RRID:AB_10673448
Goat anti-mouse IgG (H+L), Alexa fluor 488 conjugated	Invitrogen	Cat# A-11029 RRID:AB_138404
Goat anti-mouse IgG (H+L), Alexa fluor 546 conjugated	Invitrogen	Cat# A-11003 RRID:AB_141370
Donkey anti-goat IgG (H+L), Alexa fluor 546 conjugated	Invitrogen	Cat# A-11056 RRID:AB_142628
Goat anti-rabbit IgG (H+L), Alexa fluor 488 conjugated	Invitrogen	Cat# A-11034 RRID:AB_2576217
Goat anti-rabbit IgG (H+L), Alexa fluor 546 conjugated	Invitrogen	Cat# A-11035 RRID:AB_143051
Goat anti-mouse IgG (H+L), Alexa fluor 633 conjugated	Invitrogen	Cat# A-21052 RRID:AB_2535719
Phalloidin, Alexa fluor 633 conjugated	Invitrogen	Cat# A-22284 RRID:AB_2904561
Bacterial and virus strains
AAV2-synapsin-tdTomato	This paper	N/A
AAV1-hSyn1-mRuby2-GSG-P2A-GCaMP6s-WPRE-pA	Rose T, Jaepel J, Hubener M, Bonhoeffer T. Science. 2016 Jun 10;352(6291):1319-22. doi: 10.1126/science.aad3358. 10.1126/science.aad3358 PubMed 27284193	Addgene Cat# 50942-AAV1
AAV9-PPX-GFP	This paper	N/A
AAV9-PPX-Plong	This paper	N/A
Biological samples
Human spinal cord tissue	Target ALS human postmortem tissue core Table in Figure S13	N/A
Human CSF	Massachusetts General Hospital Neurological Clinical Research Institute and Harvard Medical School Table in Figure S14	N/A
Chemicals, peptides, and recombinant proteins
PolyPS – PolyPM - PolyPL	Dr. T. Shiba, Kitasato University, Japan	N/A
PolyPXL	Kerafast	Cat# EUI002
recPPBD	This paper	N/A
recPPX/PPase mix	Aminoverse	https://www.aminoverse.com/phosfinity/
CIP	NEB	Cat# M0525
Goat Serum	Thermo Scientific	Cat# 50062Z
JC-D8	Angelova et al., 2014	N/A
DAPI	Sigma-Aldrich	Cat# D9542
TO-PRO™-3	Invitrogen	Cat# T3605
Nissl stain	Invitrogen	Cat# N21479
Nucblue (Hoechst 33342)	Thermo Scientific	Cat# R37605
Donkey serum	Sigma-Aldrich	Cat# D9663
Fura-2-AM	Invitrogen	Cat# F1221
Pluronic acid	Invitrogen	Cat# P6867
DMEM	Hyclone	Cat# SH30081.01
MEM	Gibco	Cat# 11090-073
Neurobasal media	Gibco	Cat# 21103-049
L-glutamine	Gibco	Cat# 25030-081
Penicillin-Streptomycin	Gibco	Cat# 15140-122
Sodium pyruvate	Gibco	Cat# 11360-070
Horse serum	Gibco	Cat# 15060-114
Fetal bovine serum	Gibco	Cat# 16000-044
N2 supplement	Gibco	Cat# 17502-048
Fluoromount-G™	EMS	Cat# 17984-25
Ni-NTA-agarose	Invitrogen	Cat# R90110
HBSS	Gibco	Cat# 14025-053
HEPES	Gibco	Cat# 15630-56
UHRA-10	Travers et al., 2014	N/A
UHRA-9	Travers et al., 2014	N/A
G4-PAMAM-NH2 dendrimer	Sigma-Aldrich	Cat# 412449
Trypsin	Gibco	Cat# 15090046
Proteinase K	Millipore	Cat# 70663-4
BSA	Winkler	BM-0150
DMSO	Sigma-Aldrich	Cat# D2650
Phenol	Sigma-Aldrich	Cat# 77607
Sodium Phosphate	Sigma-Aldrich	Cat# 342483
Sodium triphosphate	Sigma-Aldrich	Cat# 238503
ATP	Sigma-Aldrich	Cat# A9187
GTP	Sigma-Aldrich	Cat# G8877
Ionomycin	Sigma-Aldrich	Cat# I0634
Heparin	Sigma-Aldrich	Cat# H3149
Chondoitin sulfate	Sigma-Aldrich	Cat# 1133570
Syndecan-4	R&D systems	Cat# 6267-SD-050
Brevican	R&D systems	Cat# 7188-BC-050
Heparinase III	R&D systems	Cat# 6145-GH
Chondroitinase ABC	R&D systems	Cat# 6877-GH
tRNA	Sigma-Aldrich	Cat# R5636
BSA	Sigma-Aldrich	Cat# P5369
Critical comercial assays
Malaquite green Phosphate Assay kit	BioAssay Systems	Cat# POMG-25H
Phosphinity–Total polyphosphate quantification kit	Aminoverse	https://www.aminoverse.com/phosfinity/
Vectastain elite ABC-HRP kit, peroxidase	Vector laboratories	PK-6101
RNeasy kit	Qiagen	Cat# 74004
Deposited data
RNA-Seq raw	Tdis paper	GEO: GSE196664
Experimental models: Cell lines
HEK293T cell line	ATCC	Cat# CRL-3216
Experimental models: Organisms/strains
Mouse: human SOD1G93A	Jackson Laboratories	Cat# 002726 RRID:IMSR_JAX:002726
Mouse: human SOD1WT	Jackson Laboratories	Cat# 002297 RRID:IMSR_JAX:002297
Mouse: TARDBPA315T	Jackson Laboratories	Cat# 010700 RRID:IMSR_JAX:010700
Mouse: C9orf72 hexanucleotide repeats (~500)	Peters et al., 2015	N/A
Drosophila: repoGal4	Bloomington Stock Center	Cat# 7415 RRID:BDSC_7415
Drosophila: human SOD1A4V	Bloomington Stock Center	Cat# 33607 RRID:BDSC_33607
Drosophila: human SOD1WT	Bloomington Stock Center	Cat# 33606 RRID:BDSC_33606
Drosophila: dSOD1	Bloomington Stock Center	Cat# 33605 RRID:BDSC_33605
Recombinant DNA
Plasmid: pTrc-PPBD	Saito et al., 2005	N/A
Plasmid: PPX-GFP	Abramov et al., 2007	N/A
Plasmid: PPX-GFP-PLong	This paper	N/A
Plasmid: pET14b	Merck	Cat# 69660
Software and algorithms
Fiji	Image J	https://fiji.sc/
GraphPad Prism 6.01	GraphPad	https://www.graphpad.com/
Huygens	SCI	https://svi.nl/Huygens-Software
pClamp 10.7	Molecular Devices	https://support.moleculardevices.com/s/
HCImage v2	Hamamatsu	https://www.hamamatsu.com/us/en/product/type/U11158/index.html
Others
Epifluorescence microscope	Nikon	TE2000e
Confocal microscope	Leica	TCS SP8
Vibratome	Leica	VT 1000GS
Cryostat	Leica	152S
Fluorescence microscope	Olympus	BX51
Patch-clamp amplifier	Axon Instruments	Axopatch 200B
Patch-clamp amplifier	HEKA	N/A
Borosilicate glass pipettes	WPI	Cat# TW150F-4
Zeba spin desalting column	Thermo Scientific	Cat# 89883
Econospin silica spin column	Epoch Life Science	Cat# 1910-050
Dow Corning® high-vacuum silicone grease	Sigma	Cat# Z273554-1EA

Human cerebrospinal fluid

Human CSF samples were obtained from the Northeast ALS Consortium (NEALS) ALS repository and the Dominant Inherited ALS Network. All available de-identified clinical records were collected and used together with ALS genotypes to summarize patient demographics and disease features (see Table in Figure S14).

Control and ALS/FTD transgenic mice

All protocols involving rodents (including rat spinal cord cultures; see below) were carried out according to the NIH and ARRIVE guidelines and were approved by the Ethical and Bio-security Committees of Universidad Andres Bello. Hemizygous transgenic mice carrying mutant human SOD1^G93A (high copy number; B6SJL; Cat. # 002726), wild-type human SOD1^WT (B6SJL; Cat. # 002297) and TARDBP^A315T (C57BL/6J; Cat. # 010700) transgenes were obtained from Jackson Laboratories (Bar Harbor, Maine, USA). Hemizygous transgenic mice (background B6SJLF/J) carrying a bacterial artificial chromosome containing part of the human C9ORF72 gene (exon 1-6) with ~500 G₄C₂ hexanucleotide repeats in intron 1 were generated by Robert Brown (Peters et al., 2015). All mice were maintained and bred at the animal facility of Universidad Andres Bello. The presence of the human SOD1 transgene (Gurney et al., 1994; van Zundert et al., 2008), the human TARDBP transgene (Wegorzewska et al., 2009) and the human C9ORF72 transgene (Peters et al., 2015) were validated by polymerase chain reaction as previously described. Transgenic males (SOD1, C9ORF72 and TARDBP) were crossed with WT females (B6SJL or C57BL/6J) from the corresponding background; these non-transgenic littermates (NTg) were used as internal controls for all experiments. The ALS SOD1^G93A (mutSOD1) mice, but not the hSOD1^WT (wtSOD1) mice, develop signs of neuromuscular deficits (tremor of the legs and loss of extension reflex of the hind paws) starting at 3 months of age and have an average lifespan of 133-147 days (Gurney et al., 1994). ALS mice carrying TDP43^A315T (mutTDP43) develop similar loss of motor function between 2-4 months and do not survive to the age of 4 months (Wegorzewska et al., 2009). Mice carrying C9ORF72 hexanucleotide repeats (~500) (mutC9ORF72) show pathological hallmarks (RNA foci and dinucleotide peptides) of ALS and FTD (Peters et al., 2015) as well as mild cognitive deficits at 6 and 9 months of age (Jury et al., 2020).

For astrocytes cultures, randomized mixed-gender cohorts were used at P1-P2. For the polyP staining assays, mutSOD1 (4-month-old), mutTDP43 (2-month-old) and mutC9ORF72 (9-month-old) along with their corresponding NTg littermates were examined. At these stages, the animals showed symptoms of ALS (mutSOD1 and mutTDP43) and FTD (mutC9ORF72). Only males were used for these experiments; since we have not analyzed female mice, we do not know whether the sex of the animals impacts the findings. For the in vivo experiments in which mutSOD1 mice overexpressed the yeast PPX1 construct, P0-1 mutSOD1 mice were randomly injected ICV with the AAV9-CMVPPX1-PLong-GFP (total 8: 5 males and 3 females); AAV9-CMV-PPX-GFP (total 3: 1 male and 2 females) or not treated (total 8: 5 males and 3 females). We did not detect that the sex of the animals impacted the findings (i.e., survival and rotarod). Sprague-Dawley rats, to obtain embryonic ventral spinal cord cultures (E14), were originally obtained from the Pontifical Catholic University of Chile (Santiago) and maintained and bred at the animal facility of Universidad Andres Bello.

Primary mouse astrocyte cultures and ACM

Primary astrocyte cultures and ACM were prepared as previously described from mutSOD1 (Fritz et al., 2013; Rojas et al., 2014), mutTDP43 (Rojas et al., 2014) and mutC9ORF72 (Jury et al., 2020) spinal cords derived from neonatal transgenic mice. Briefly, cultures of astrocytes were prepared from whole spinal cords of these transgenic ALS mice at P1-2. Non-transgenic astrocytes from littermates (Ctrl) and astrocytes expressing wtSOD1 (wtSOD1) were used as controls. Cells were maintained in DMEM (Hyclone, Cat. # SH30081.01) containing 10% FBS (Gibco, Cat. # 16000-044), 1% L-glutamine (Gibco, Cat. # 25030-081) and 1% penicillin-streptomycin (Gibco, Cat. # 15140-122) at 37°C, 5% CO₂. Cultures reached confluence after ~3 weeks. Residual microglia were removed by shaking overnight, leaving astrocyte cultures enriched for S100β and Aldh1L1. Astrocytes were used to determine intracellular polyP content or to generate ACM.

As previously described (Fritz et al., 2013; Rojas et al., 2014; Jury et al., 2020), after primary spinal cord astrocyte cultures reached confluence, media was replaced by neuronal growth media: 70% MEM (Gibco, Cat. # 11090-073), 25% Neurobasal media (Gibco, Cat. # 21103-049), 1% N2 supplement (Gibco, Cat. # 17502-048), 1% L-glutamine (Gibco, Cat. # 25030-081), 1% penicillin-streptomycin (Gibco, Cat. # 15140-122), 2% horse serum (Gibco, Cat. # 15060-114; lot 1517711) and 1% sodium pyruvate (Gibco, Cat. # 11360-070). After 7 days, ACM was collected, supplemented with 4.5 mg ml/L D-glucose (final concentration), filtered and stored at −80°C. The ACM used to evaluate ACM-polyP deposition assay and survival of MNs were diluted: The mutSOD1-ACM, wtSOD1-ACM, mutTDP43-ACM, and their respective NTg-ACM were diluted 9-fold. The mutC9ORF72-ACM and its NTg-ACM were diluted 8-fold. At the selected dilutions, ACM derived from the astrocytes expressing the ALS-causing genes robustly killed MNs and resulted in detectable deposits of polyP at neuronal membranes relative to the control media.

Human iPSC-induced astrocytes cultures and ACM

We analyzed fully reprogrammed iPSC lines previously generated by retroviral transduction with the four Yamanaka factors (OCT4, SOX2, KLF4, and cMYC) from skin biopsies derived from an ALS/FTD subject (75 years old male, termed patient) carrying a A90V mutation in TARDBP (TDP43^A90V) and a healthy subject (56 years old female, termed control), a control family member without mutations (termed control) (Zhang et al., 2013). Maintenance of iPSCs and differentiation to NPCs were performed as previously described (Almeida et al., 2012; Zhang et al., 2013). Note that because same reagents are used for different differentiation steps, the company and catalogue number is indicated only the first time. Briefly, iPSCs were maintained in feeder-free conditions using mTeSR1 medium and EBs were generated and maintained in suspension for one week in EB differentiation media: knock-out (KO) DMEM/F12 (Gibco Cat. # 12660-012) media supplemented with 10% KO serum replacement (Gibco Cat. # 10828-028), 1x GlutaMax (Gibco Cat. # 35050-061), 1x NEAA (Gibco Cat. # 11140-050), and 2-mercaptoethanol (Sigma-Aldrich Cat. # M3148). To obtain rosette-shaped neuroepithelial cells, the EBs were plated on poly-L-ornithine-laminin-coated plates (Sigma-Aldrich Cat. # P4957, Cat. # L2020) and were grown for one week in neural induction media: KO-DMEM/F12 supplemented with N2 (Gibco Cat. # 17502-048), NEAA, 2 μg/mL heparin (Sigma-Aldrich Cat.# H3149), and 10 ng/mL βFGF (Gibco Cat.# PHG0021). Fully formed rosettes were manually isolated under the microscope, re-plated on matrigel-coated plates (Coming Cat. # 354277) and grown for one more week in neural expansion media: neurobasal supplemented with Glutamax, NEAA, B-27 (Gibco Cat. # 17504-044), and βFGF. The disaggregation of rosettes using accutase (EMD Millipore Cat. # SCR005) generated a stable, proliferative, monolayer culture of NPCs. As previously described, NPCs were differentiated to astrocytes by culturing cells first for 2 weeks in astrocyte precursor medium (Shaltouki et al., 2013): KO DMEM/F12, 1x StemPro NSCs Supplement (Gibco Cat. # A10508-01), 10 ng/mL Activin A (Gibco Cat. # PHC9564), 10 ng/mL Heregulin 1β (R&D Systems Cat. # 377-HB-050), 200 ng/mL IGF1 (R&D System Cat. # P 291-G1-200), 20 ng/mL βFGF, 20 ng/mL EGF (Gibco Cat. # PHG0311), 1x GlutaMAX. And then for 2 additional weeks in astrocyte maturation and maintenance medium (Malik et al., 2014): DMEM/F12, B27, 10 ng/mL Heregulin, 5 ng/mL BMP2 (BioVision Cat. # 4577-50) and 2 ng/mL CNTF (R&D Systems Cat. # 257-NT-010). Next, astrocyte cultures were fixed with 4% paraformaldehyde and immunostained with a rabbit polyclonal antibody against EAAT2/GLT1 (1:50; Invitrogen, Cat# PA5-17099), a mouse monoclonal antibody against Cx43 (1:200; Invitrogen, Cat# 13-8300), a rabbit polyclonal antibody against S100β (1:1000; Dako, Cat# Z0311) and a mouse monoclonal antibody against ALDH1L1 (1:50; NeuroMab, Cat# 75-140).

After primary human iPSC-induced astrocyte cultures reached day 28 of differentiation, media was replaced by neuronal growth media as described for mouse astrocytes: 70% MEM (Gibco, Cat. # 11090-073), 25% Neurobasal media (Gibco, Cat. # 21103-049), 1% N2 supplement (Gibco, Cat. # 17502-048), 1% L-glutamine (Gibco, Cat. # 25030-081), 1% penicillin-streptomycin (Gibco, Cat. # 15140-122), 2% horse serum (Gibco, Cat. # 15060-114; lot 1517711) and 1% sodium pyruvate (Gibco, Cat. # 11360-070). After 7 days with this media (day 35 of differentiation), ACM was collected, supplemented with 4.5 mg ml/L D-glucose (final concentration), filtered and stored at −80°C. The control-ACM and human TDP43-ACM used to evaluate MN survival was diluted 6-fold. At the selected dilution, ACM derived from the patient astrocytes robustly killed MNs.

Primary ventral spinal cord cultures

Pregnant Sprague-Dawley rats were deeply anesthetized with CO₂ and primary spinal cultures were prepared from E14 pups as described (Fritz et al., 2013; Rojas et al., 2014). Briefly, cultures of ventral spinal cords were prepared and maintained in neuronal growth media: 70% MEM (Gibco, Cat. # 11090-073), 25% Neurobasal media (Gibco, Cat. # 21103-049), 1% N2 supplement (Gibco, Cat. # 17502-048), 1% L-glutamine (Gibco, Cat. # 25030-081), 1% penicillin-streptomycin (Gibco, Cat. # 15140-122), 2% horse serum (Gibco, Cat. # 15060-114; lot 1517711) and 1% sodium pyruvate (Gibco, Cat. # 11360-070). Cultures were maintained for 7-9 DIV at 37°C under 5% CO₂ and supplemented with 45 μg/ml E18 chick leg extract; media was renewed every 3 days. Presence of MNs was verified with immunostaining. Briefly, primary spinal cultures were fixed at 7 DIV with 4% paraformaldehyde, and immunostained with a goat polyclonal antibody against ChAT (1:200; Sigma-Aldrich; Cat# AB144P). In addition, neurons were immunostained with antibodies against SMI-32, MAP2 (see below in MN survival assay).

mutSOD1 Drosophila

RepoGal4 (BL: 7415), mutSOD1 (is UAS-hSOD1^A4V, human mutant SOD1, BL: 33607), wtSOD1 (is UAS-hSOD1, human wild-type SOD1, BL: 33606), dSOD1 (Drosophila wild-type SOD1 isoform, BL:33605) were obtained from the Bloomington Stock Centre. The human SOD original lines were deposited by Bonini’s lab in BL (Watson et al., 2008). w1118 and UAS-GFP were a gift of John Ewer. For climbing and survival assays, the data is shown only for female flies. Similar data was obtained with male flies. For survival assays, flies were grown and aged at 29°C. Flies were reared at a maximum of 20 individuals per tube, transferred every other day into fresh media and dead flies were counted. Flies were maintained on standard cornmeal/molasses/yeast/ agar media with 80% humidity on a 12:12-h light–dark cycle. For each experiment the repoGal4 line alone or driving GFP was used as control. All lines were outcrossed to w1118 to homogenize the genetic background before starting any experiment. Lifespan assays, we performed using IBM® SPSS® Statistics. The logrank test was used to compare survival distributions between groups. At least 3 independent cohorts were used for each genotype with controls and queries running in parallel each time.

METHOD DETAILS

MN survival assay

To measure survival of MNs and INs in ventral spinal cord cultures treated with ACM, cultures were immunolabeled and counted as described (Fritz et al., 2013; Jury et al., 2020; Nagai et al., 2007; Rojas et al., 2014; Sepulveda et al., 2010; Urushitani et al., 2006; Zhu et al., 2018; Mishra et al., 2020). Briefly, primary spinal cultures were fixed at 7 DIV with 4% paraformaldehyde, and immunostained with a rabbit polyclonal antibody against MAP2 (1:200; Invitrogen; Cat. # OSM00030W) to label all neurons (INs plus MNs) and with a mouse monoclonal SMI-32 antibody (1:600, Abeam; Cat. # ab187374) to reveal the presence of unphosphorylated neurofilament-H, which is expressed specifically in MNs in spinal cord cultures (Nagai et al., 2007; Urushitani et al., 2006; Zhu et al., 2018), previously we found that our WT primary spinal cultures typically contain at least 8–10% MNs until 12 DIV (Sepulveda et al., 2010). Fluorescently labeled neurons were visualized with epifluorescent illumination on an Olympus IX81 microscope (equipped with a Q-Imaging Micropublisher 3.3 Real-Time Viewing camera) using a 20x objective; MAP2- and SMI32-positive neurons were counted off-line using Fuji ImageJ. Per condition, ≥ 12 randomly chosen fields (≥ 400 cells) were analyzed to calculate the percentage of SMI-32-positive MNs within the total number of MAP2-positive cells. Each condition was replicated in 3-4 independent cultures.

Intracellular polyP staining assays with DAPI-polyP, JC-D8 and recPPBD in astrocyte cultures

To evaluate intracellular polyP levels in primary astrocyte cultures, polyP labeling assays were performed with DAPI-polyP (Aschar-Sobbi et al., 2008; Holmström et al., 2013; Stotz et al., 2014; Tijssen et al., 1982), JC-D8 (Angelova et al., 2014) or recPPBD (Saito et al., 2005). All steps were performed at room temperature, unless indicated. All images were taken on a Leica TCS SP8 confocal microscope with a 63x oil objective (NA = 1.4; HC PL APO CS2) and a z-step of 0.5 μm optical sections (velocity scan 600 Hz; resolution 1024x1024 pixels, equivalent to 185 μm x 185 μm):

(i) DAPI-polyP: cultured astrocytes were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton-X-100 for 20 min (only if additional immunofluorescent labeling was used), loaded with the fluorescent dye DAPI (0,1 μg/ml, Sigma, Cat. # D9542) in PBS for 30 min, carefully washed 3 times and mounted on coverslips with non-fluorescing fluoromount-G™ (Electron Microscopy Sciences, Cat. # 17984-25). Note that it is not recommended to use mounting media that already includes DAPI (e.g. DAPI-Fluoromount-G or Prolong Gold Antifade Mountant with DAPI). The following laser excitation (Ex) and emission (Em) wavelengths were used to detect DAPI-polyP (Ex: 488 nm and Em: 510-560 nm) and DAPI-DNA (Ex: 405 nm and Em: 410-470 nm).

(ii) JC-D8: live cultured astrocytes were incubated with HEPES-buffered salt solution (HBSS) containing the fluorescent probe JC-D8 (5 μM) for 30 min in the CO2 incubator at 37°C. Next, cells were carefully washed 3 times with HBSS and fixed with 4% paraformaldehyde for 20 min. Cells were permeabilized with 0.1% Triton-X-100 for 20 min (only if additional immunofluorescent labeling was used), loaded with the nuclear dye To-Pro-3 (termed here TOPRO3) (1:2000, Cat. # T3605, Invitrogen), washed and mounted on coverslips with non-fluorescing Fluoromount-G™ The following laser excitation and emission wavelengths were used to detect JC-D8 (Ex: 488 nm and Em: 510-560 nm) and TOPRO3 (Ex: 638 nm and Em: 643-776 nm).

(iii) recPPBD: cultured astrocytes were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton-X-100 for 20 min, washed with PBS, blocked with goat serum and treated with the yeast recombinant PPBD protein (recPPBD containing an Xpress epitope tag) (20-30 μg/ml) in goat serum for 30 min. RecPPBD was produced in E. coli BL21 bacteria, exactly as described (Saito et al., 2005). Next, cultures were incubated with a mouse anti-Xpress antibody (1:500, Cat. # R910-25, Invitrogen) overnight at 4°C, washed, and incubated with an Alexa-conjugated goat anti-mouse antibody (1:500, Alexa⁴⁸⁸, Invitrogen, Cat. # A-11029) for 1.5 h. Cells were finally loaded with the nuclear dye TOPRO3, washed 3 times and mounted on coverslips with non-fluorescing Fluoromount-G™. The following laser wavelengths were used to detect recPPBD-Xpress-Alexa⁴⁸⁸ (Ex: 488 nm and Em: 550 nm) and TOPRO3 (Ex: 638 nm and Em: 643-776 nm).

For quantifying the intracellular polyP signal in individual astrocytes, the polyP fluorescent signal intensity detected by recPBD, JC-D8 or DAPI-polyP was measured from the maximum intensity projection of confocal z-stack images (4 μm total) by Fiji ImageJ software (8 bits, measuring intensity from 0 to 225). Specifically, the signal located within a circular region of interest (ROI) was analyzed; this ROI had a diameter twice the size of the analyzed nucleus. Four randomly chosen fields (185 μm x 185 μm) were analyzed per independent experiment for each experimental condition; per field, 2 isolated astrocytes (containing a nuclear diameter ranging from 10-14 μm) were randomly quantified per condition. Each condition was replicated in 3-4 independent cultures, leading to the analysis of 24-32 astrocytes per condition.

Colocalization of DAPI-polyP, JC-D8 and recPPBD in astrocyte cultures

To determine the colocalization between (i) recPPBD and DAPI-PolyP, and (ii) recPPBD and JC-D8 in mutSOD1 astrocytes, labeling was performed as described in the above section, with some modifications. For both co-labeling experiments, the Xpress epitope of recPPBD was detected with Alexa⁵⁴⁶-conjugated goat anti-mouse antibody (1:500, Alexa⁵⁴⁶, Invitrogen, Cat. # A-11003). The following laser wavelengths were used for the co-labeling of (i) recPPBD and DAPI-polyP: recPPBD-Xpress-Alexa⁵⁴⁶ (Ex: 552 nm and Em: 560-620 nm), DAPI-PolyP (Ex: 488 nm and Em: 510-560 nm), and DAPI-DNA (Ex: 405 nm and Em: 410-470 nm), and (ii) recPPBD and JC-D8: recPPBD-Xpress-Alexa⁵⁴⁶ (Ex: 552 nm and Em: 560-620 nm), JC-D8 (Ex: 488 nm and Em: 510-560 nm), and TOPRO (Ex: 638 nm and Em: 643-776 nm). All images were taken on a Leica TCS SP8 confocal microscope with a 63x oil objective and 3x digital zoom (NA = 1.4; HC PL APO CS2) and a z-step of 0.1 μm optical sections (velocity scan 600 Hz; resolution 1024x1024 pixels, equivalent to 62 μm x 62 μm). Pearson correlation coefficients were calculated by Huygens Professional software. Each colocalization was analyzed in 6-8 astrocytes per condition.

ACM-polyP deposition assay on neuronal plasma membranes

Ventral spinal cord cultures (2 DIV) were infected with AAV2-synapsin-tdTomato. At 7-8 DIV, live cultures were incubated with HEPES-buffered salt solution (HBSS) containing the fluorescent probe JC-D8 (5 μM) for 30 min in the CO2 incubator at 37°C. Next, cultures were treated with ACM, feeding media or polyP_L for 5 min, carefully washed 3 times with HBSS, fixed with 4% paraformaldehyde for 20 min and mounted on coverslips with non-fluorescing fluoromount-G™ The following laser excitation and emission wavelengths were used to detect JC-D8 (Ex: 488 nm and Em: 510-560 nm) and tdTomato (Ex: 638 nm and Em: 653-776 nm). All images were taken on a Leica TCS SP8 confocal microscope with a 63x oil objective (NA = 1.4; HC PL APO CS2) and a z-step of 0.5 μm optical sections (velocity scan 600 Hz; resolution 1024x1024 pixels, equivalent to 185 μm x 185 μm). For quantifying the polyP deposition on individual tdTomato-positive spinal cord neurons, the polyP fluorescent signal intensity detected by JC-D8 was measured from the maximum intensity projection of confocal z-stack images (4-5 μm total) by Fiji ImageJ software (8 bits, measuring intensity from 0 to 225). To measure the polyP signal on the soma of tdTomato-positive spinal cord neurons, ROIs were drawn on the cell bodies. Six randomly chosen fields (185 μm x 185 μm) were analyzed per independent experiment for each experimental condition; per field, 6 isolated tdTomato-positive neurons were randomly quantified per condition. Each condition was replicated in 3 independent cultures.

Quantification of polyP in ACM

To extract polyP from ACM, we developed a modified version of a phenol-chloroform (P/C) extraction protocol originally devised for yeast (Bru et al, 2014; Christ and Blank, 2018). To measure the polyP-Pi content in ACM, isolated polyP was degraded with recPPX/PPase (Christ and Blank, 2018) available from the commercial Phosfinity kit (Aminoverse); we then quantified the liberated free Pi by colorimetrically by Malachite green or quantified the polyP-polymer fluorometrically with the JC-D8 probe. For both methods, standard curves were generated using synthetic polyP_L (in water) to calculate the ACM polyP-Pi/polymer concentration and estimate polyP recovery ratio.

Briefly, to isolate and purify polyP from ACM, 400 μl of ACM or synthetic polyP_L (as control) were incubated with 1X trypsin (Gibco, Cat# 15090046) for 1 h at 37°C. The digested samples were transferred to a 1.5 ml microcentrifuge tube with 300 μl of phenol (Sigma, Cat# 77607) and 40 μl of 2% SDS. The samples were mixed by inverting the tubes four times and vortexing for 5 sec. Then, the samples were incubated for 5 min at 65°C and immediately transferred to ice for 1 min. Next, 300 μl of chloroform:isoamilic alcohol (24:1) was added to each sample and again the solutions were mixed and vortexed. A low-density phase lock gel (Dow Corning®) was added to each tube and the samples were centrifugated at 13.000 x g for 2 min at RT. The top aqueous phase was transferred to a new tube and 350 μl of chloroform:isoamilic alcohol (24:1) was added. After mixing, vortexing and add a new low-density phase lock gel, the samples were centrifugated at 13.000 x g for 2 min at RT. Finally, the top aqueous phase was desalted using a Zeba™ spin desalting column (Thermo Scientific, Cat# 89883). The final eluate was recovered and quantified for polyP-Pi or polyP-polymer detection.

For polyP-Pi quantification, final eluates from ACM (and synthetic polyP_L) were digested with recPPX/PPase enzyme mix (Amminoverse, Phosphinity kit) for 1 h at 37°C. The Pi determination was performed with the Malachite green phosphate Assay kit (BioAssay Systems, Cat# POMG-25H). Briefly, 50 μl of each sample was incubated with 12.5 μl of mix A:B reagents (100:1) and incubated for 30 min at RT. Then, absorbance was read at 620 nm in a Synergy LX multi-model reader (BioTek). Finally, in every sample, Pi concentration was estimated subtracting PPX/PPase un-digested from the PPX/PPase digested situation. For polyP-polymer quantification, 100 μl of ACM (and synthetic polyP_L) samples were incubated with 100 μl of JC-D8 to a final 10 μM concentration in a solution of 20 mM HEPES pH 7.4 - 1% DMSO buffer, for 15 min at RT. Each sample was transferred to a dark 96-wells microplate (Thermo Scientific, Cat# 165305) and exposed for 20 sec to an UV transilluminator. Then, a Synergy H1 hybrid multi-model reader (BioTek) was set at 390 nm for excitation and emission was recorded from 420 to 700 nm to detect the JC-D8 fluorescence. The maximum value between 500 and 560 nm was selected to estimate the polyP-polymer concentration. The recovery ratio for polyP_L was 73 ± 9 % (n=4).

Quantification of polyP in human CSF

To extract polyP from human CSF, we adapted an analytical silica column protocol originally devised for tissues (Lee et al., 2018). To measure the polyP content, isolated polyP was degraded with recPPX/PPase (Christ and Blank, 2018), available from the commercial Phosfinity kit (Aminoverse); we then used Malachite green to quantify liberated free Pi. Briefly, 1 ml aliquots of individual human CSF samples were treated with 0.5 mg/ml proteinase K (Millipore, Cat # 70663-4) and 10 mM EDTA (pH 8) (Thermo Scientific, Cat #AAJ15694AE) and then were heated for 2 h at 56 °C in a hybridization oven. Next, CSF was then separated into 4 x 1.5 ml Eppendorf tubes, and 2 volumes of binding buffer was added: 5 M guanidine thiocyanate (Thermo Fisher Scientific cat # AM9422) in 50 mM Tris-Cl pH 6.8 (Fisher Scientific, Cat# BP152-5), and 25 mM EDTA, 0.9 M Na-acetate (pH 5.3) (Thermo Fisher Scientific cat # AM9740), 1% BME). Tubes were then placed on a heat block at 95 °C for 10 min, cooled down to RT, and 2 volumes of 100% ethanol were added. Samples were mixed by inversion, and the content of each tube was loaded twice on EconoSpin column (Epoch Lifescience) and centrifuged (10,000 x g, 30 sec, RT). The flow-through was discarded, and 0.7 ml of washing buffer I (1 M guanidine thiocyanate in 80% ethanol) was added. The solution was centrifuged (11,000 x g, 1 min, RT) and the flow-throw discarded. The column was washed twice with washing buffer II (150 mM NaCl, 10 mM Tris-Cl pH 7.5 in 80% ethanol), centrifuged (12,000 x g, 2 min, RT) and the column was transferred into a clean 1.5 ml microcentrifuge tube. PolyP was eluted by incubating the filter with elution buffer (60 μl of 10 mM Tris-HCl (pH 8)) for 15 min at RT and centrifuged (12,000 x g, 2 min, RT). The elution step was repeated twice more, with a 10 min incubation in the two. The final eluates from the 4 tubes containing the extracts from the same CSF samples were pooled, and concentrated by using an Amicon-ultracel 3K filter (cat #UFC5003-96, Millipore). The ultrafiltrate was brought to a final volume of about 190 μl, and analyzed for its polyP content. During this process, each analyzed CSF sample was concentrated about 5-fold; the volumes were measured to determine the concentration factor needed to calculate the final polyP concentration after completing the digestion and the Pi measurements (see below). To quantify Pi, we used the Phosfinity kit (Amminoverse) to digest polyP, according to the manufacturer’s instructions, with the following modifications. Briefly, 110 μl of concentrated eluate was incubated at 37 °C for 1 h in the presence and absence of 1.1 μl of a recombinant PPX1/PPase1 and 55 μL of enzyme buffer provided in the kit. The incubation was stopped by adding 10 μM EDTA and the free Pi liberated by the enzymatic reaction was measured in triplicate by loading 50 μl aliquots in 96 multi-wells plates (Greiner Bio-One cat # 650101), together with 12.5 μl of the Malachite green reagent (BioAssay System, Cat # POMG-25H). The plate was shaken for 30 sec and incubated for 30 min at RT; the absorbance (ABS) at 620 nm was read using a Spectramax M3 plate reader (Molecular Devices). CSF Pi content was calculated by subtracting the ABS of the recPPX/PPase-untreated samples from the ABS of recPPX/PPase-treated samples. The Pi standard calibration curve generated for each assay was used to calculate the polyP concentration for each averaged ABS/sample. Finally, the amount of polyP was corrected for the concentrating factor of each sample and was expressed in [μM]. As a control protocol (P-ctrl), 2.5 μM PolyP_L in artificial CSF (125 mM NaCl, 25 mM NaH3CO4, 2.5 mM KCl, 1.25 mM Na-phosphate buffer (pH 7.2), 2 mM CaCl2, 1 mM MgCl2, 0.1 mg/ml BSA) was incubated with 0.5 mg/ml proteinase K and 10 mM EDTA (pH 8) and was processed as for all the other samples to assess the recovery ratio as well as the ability of the recPPX/PPase to digest the polyP in the extracts. The recovery ratio for artificial CSF was 75 ± 3% (n=6).

PolyP degradation and sequestering molecules

PolyP was degraded with recPPX/PPase from the Phosfinity kit (Aminoverse) or with CIP (NEB, Cat# M0525). To sequester and neutralize polyP in astrocyte-conditioned media, the following molecules were used: (i) recPPBD (20-30 μg/ml), produced in E. coli BL21 bacteria as described (Saito et al., 2005); (ii) UHRA9 and UHRA10 (5-500 nM), donated (Smith et al., 2012; Travers et al., 2014); (iii) G4-PAMAM-NH2 (0.1 - 1 μg/ml, Cat. #412449, Sigma). The capacity of these molecules to bind polyP was tested in TBE urea-PAGE gels (not shown).

PolyP fluorescent staining assays in fixed mouse spinal cord sections with recPPBD

MutSOD1 (4 mo), mutTDP43 (2 mo) and mutC9ORF72 (9 mo) mice and their corresponding non-transgenic littermates were anesthetized with ketamine/xylazine and then transcardially perfused with Ringer’s solution followed by a perfusion with cold 4% paraformaldehyde. The spinal cord was carefully removed and post-fixed in 4% paraformaldehyde overnight at room temperature. Next, samples were cryoprotected in a sucrose gradient (10, 20 and 30%) for 3 days at 4°C. Spinal cord cross sections (40 μm thick) were cut using a cryostat (Leica, CM 152S, Germany). Lumbar spinal cord sections were washed with PBS and incubated next with blocking/permeabilization solution (3% donkey serum and 3% BSA in 0.5% Triton X-100 in PBS) for 4 h at room temperature and then incubated with recPPBD (30 μg/ml) containing an Xpress epitope tag for 40 min at room temperature. Next, slices were washed three times with 0.5% Triton X-100 in PBS and incubated with the following primary antibodies: anti-ChAT (1:200, goat polyclonal antibody, Sigma-Aldrich; Cat# AB144P), anti-NeuN (1:200, rabbit, Cat. #MAB377, Millipore), anti-GFAP (1:1000, rabbit, Cat. # G3893, Sigma) and anti-Xpress (1:500, mouse, Cat. # R910-25, Invitrogen) in blocking/permeabilization solution overnight at 4°C. After rinsing 3 times with 0.5% Triton X-100 in PBS, the spinal cord slices were incubated with Alexa-conjugated goat secondary antibodies and the nuclear dye TOPRO3 (1:1000, Cat. # T3605, Invitrogen) in secondary solution (3%BSA; 5% Triton X-100 in PBS) for 2 h at room temperature. The following Alexa-conjugated goat secondary antibodies were used: for recPPBD (1:500, anti-mouse Alexa⁴⁸⁸, Cat. # A-11029, Invitrogen), for GFAP and NeuN (1:500, anti-rabbit Alexa⁵⁴⁶ Cat. # A-11029 and anti-rabbit Alexa⁵⁴⁶, Cat. # A-11081, Invitrogen). Finally, slices were washed twice with 0.5% Triton X-100 in PBS and once with PBS and mounted on coverslips with non-fluorescing fluoromount-G™ (Electron Microscopy Sciences, Cat. # 17984-25). Images of the ventral horn of the lumbar spinal cord were taken on a Leica TCS SP8 confocal microscope with a 63x oil objective (NA = 1.4; HC PL APO CS2) and a z-step of 0.5 μm optical sections (velocity scan 600 Hz; resolution 1024x1024 pixels, equivalent to 185 μm x 185 μm). The following laser wavelengths were used to detect RecPPBD-Xpress-Alexa⁴⁸⁸ (Ex: 488 nm and Em: 510-550 nm), NeuN-Alexa⁵⁴⁶ (Ex: 552 nm and Em: 564-620 nm), GFAP-Alexa⁵⁴⁶ (Ex: 552 nm and Em: 564-620 nm). Maximum intensity projections of confocal z-stack images of whole cells (containing 12-17 images) were analyzed. Fiji ImageJ software (8 bits, measuring intensity from 0 to 225) was used to quantify the intracellular recPPBD-polyP fluorescence intensity. Specifically, the recPPBD-polyP signal located within the soma of individual NeuN-positive and GFAP-positive cells was quantified by drawing an ROI on the cell bodies of these cells. Three randomly chosen fields (200 μm x 200 μm) within ventral horn were analyzed per independent experiment for each experimental condition; per field, 8-10 NeuN-positive and 8-10 GFAP-positive cells were randomly quantified. Each condition was replicated in 3-4 mice.

PolyP fluorescent staining assays in unfixed mouse motor cortex sections with JC-D8

Acute brain slices of mutSOD1 (4 mo) and mutC9ORF72 (9 mo) mice and their corresponding non-transgenic littermates were generated as described (Maturana et al., 2017). Briefly, animals were anesthetized with isoflurane and euthanized by decapitation. Next, brains were dissected and placed in ice-cold artificial cerebral spinal fluid (ACSF). Cortical coronal slices (300 μm thick) were obtained by using a vibratome (VT 1000GS, Leica, Germany) filled with ice- cold ACSF. The slices were transferred to a holding chamber and immersed in oxygenated ACSF (20°C–22°C) in the presence of 2 mM pyruvate, pH 7.4. For the detection of polyP, slices were stained for 30 min with JCD8 (10 μM), carefully washed with ACSF and fixed in 4% paraformaldehyde for 1 h at room temperature. Samples were cryoprotected in a sucrose gradient (10, 20 and 30%) for 3 days at 4°C. Next, slices were cut serially in 40 μm sections on a cryostat (CM 152S, Leica, Germany). To identify neurons and layer V in the motor cortex, slices were washed three times with PBS, incubated with blocking/permeabilization solution (3% donkey serum; 3% BSA; 0.5% Triton X-100 in PBS) for 4 h at room temperature and then incubated with the primary antibody anti-NeuN (1:300, Cat. # MAB377, Millipore) and Nissl staining (NeuroTrace™ 435/455 Blue Fluorescent, Cat. # N21479, Invitrogen) in blocking/permeabilization solution overnight at 4°C. Slices were then rinsed 3 times with 0.5% Triton X-100 in PBS and incubated with Alexa Fluor 546 conjugated secondary antibody (Alexa⁵⁴⁶; 1:500, Cat. # R37115, Invitrogen) for 2 h at room temperature. Finally, slices were washed twice with 0.5% Triton X-100 in PBS and once with PBS and mounted on coverslips with non-fluorescing fluoromount-G™ (Electron Microscopy Sciences, Cat. # 17984-25). Images of the motor cortex were taken on an Olympus Fluoview FV1000 confocal microscope with a 60x oil objective (NA = 1.3, C-Apochromat) at a z-step of 0.5 μm (resolution 1024x1024). The following laser wavelengths were used to detect JC-D8 (Ex: 488 and Em: 510 nm), Nissl (Ex:405 and Em: 455 nm) and NeuN-Alexa⁵⁴⁶ (Ex:546 and Em: 580 nm). Maximum intensity projections of confocal z-stack images of whole cells (containing 25 images) were analyzed. Fiji ImageJ software (8 bits, measuring intensity from 0 to 225) was used to quantify the intracellular JCD8 fluorescence intensity. Specifically, the JCD8 signal located within the soma of individual Nissl⁺/NeuN⁺ cells and Nissl⁺/NeuN⁻ cells located within the motor cortex layer V was quantified by drawing an ROI on Nissl⁺ cell bodies. To determine the number of JC-D8⁺/NeuN⁺ and JC-D8⁺/NeuN⁻ cells per 100 μm², JC-D8 fluorescence intensity of >20 a.u. (8 bit) and containing more than one “JCD8-spot” (size: >0.1 μm²) was defined as a JC-D8-positive cell. The measurements of the ROIs on Nissl⁺/NeuN⁺ cells and Nissl⁺/NeuN⁻ cells were also used to quantify the size of the soma area of the cells. For all analyses, three randomly chosen fields of 212 μm x 212 μm were examined per independent experiment for each experimental condition; per field, 8-10 Nissl⁺/NeuN⁺ cells and 8-10 Nissle⁺/NeuN⁻ cells were randomly quantified, leading to the analysis of at least 60 NeuN⁺ and 30 NeuN⁻ cells per animal. Each condition was replicated in 3 mice.

PolyP immunohistochemical staining assays with recPPBD on human post-mortem spinal cord sections

The autopsy spinal cord samples from 9 ALS cases and 5 non-neurologic control subjects were provided by the Target ALS Human Postmortem Tissue Core (see details on patients in Figure S11). Formalin-fixed paraffin-embedded tissue sections (3 μm) were deparaffmized in Xylene and rehydrated in a graded alcohol series (100%, 96%, 70% and 50%) for 10 min in each bath and rinsed in running tap water before pre-treatment with sodium citrate buffer 0.01M, pH 6 for 10 min at 90°C for antigen retrieval. Samples were washed 3 times 5 min each in a solution of 0.01M PBS 0.15% Triton X-100 and incubated with 0.3% H₂O₂ at room temperature for 30 min, then incubated 1 h with blocking solution containing 5% BSA in 0.01M PBS 0.15% Triton X-100 and 1 h with recPPBD (30 μg/ml) followed by overnight incubation at 4°C with the primary antibody anti-Xpress (1:500, Cat. # R910-25, Invitrogen) diluted in blocking solution. After being rinsed 3 times with 0.01M PBS 0.15% Triton X-100, sections were incubated for 1 h with a mouse biotinylated secondary antibody (1:250, Cat. # BA-9200, Vectastain, ABC Elite kit PK-6101, Vector Laboratories) and diluted in blocking solution followed by 1 h incubation with avidin–biotin–peroxidase complex (1:125, Vectastain, ABC Elite kit PK-6101, Vector Laboratories). Finally, a 30% H₂O₂ solution was added in the presence of 3,3’-diaminobenzidine (DAB, 1 mg/ml) to visualize the antigen–antibody complex. Staining for cresyl violet was performed in tissue sections to identify the different regions in human spinal cord. Next, images were captured on a BX51 microscope (Olympus; Tokyo, Japan) with a halogen light source and light-balancing daylight. Photoshop software version 13.0 (Adobe Systems Inc., San Jose, CA) and ImageJ Fiji version 1.52i (National Institutes of Health, USA) were used to analyze and post-process images.

To count the number of MN-like cells with their respective areas, two pictures per ventral spinal cord section were taken for every sample at 10x objective (682 μm x 512 μm area) and all cells with ≥ 110 μm² area and a visible nucleolus typically found in MNs (Kiernan and Hudson, 1991) were counted. To count the number of glial-like cells with their respective area, two pictures per ventral spinal cord section were taken for every sample at 100x (350 μm x 260 μm area). For the glia-like cells, all cells with an area ranging from 5 μm² to 16 μm² were counted and the following phenotypes as previously described (Forsberg et al., 2011): astrocytes, characterized by their round nuclear shape; microglia, characterized by their round nuclear shape but smaller area than astrocytes; and oligodendrocytes, characterized by their elongated nuclear shape. To measure the size of the soma of cells, the cresyl violet staining served to draw an ROI using Fiji ImageJ software. The drawn ROIs were used to quantify the DAB staining of the soma of individual cells. To accurately quantify only the intensity of the DAB signal without the interference of the cresyl violet staining the images were processed as described (Linden et al., 2015). Briefly, brown color (corresponding to DAB signal) was selected from the background area (blue, white and non-tissue areas) with the use of the eyedropper tool and transformed in grayscale 8 bits images with intensity ranging from 0 to 225. Ten pictures at 100x were analyzed per sample. At least 10 MN-like cell and 25 glia-like cells were analyzed per subject in the lumbar spinal cord samples. In thoracic and cervical spinal cord samples, 2-6 MN-like cell and 25-30 glia-like cells were analyzed per subject.

Cell-free assays with JC-D8, DAPI and recPPBD

(i) The JC-D8 binding assays were performed in a 96 well black/clear bottom plate (Thermo Scientific, Cat# 165305), where synthetic polyP or the potential interfering molecule (see Table in Figure S1) was mixed with JC-D8 at 10 μM final concentration in a solution of 20 mM HEPES and 1% DMSO buffer. The reaction was incubated for 15 minutes at RT and exposed to an UV transilluminator for 20 sec. Then, the plate was read in a Synergy H1 hybrid multi-model reader (BioTek) at 390 nm excitation and 510-560 nm emission. (ii) The DAPI binding assays were performed as the JC-D8 assays, this time incubating synthetic polyP or the interfering molecule with 15 μM DAPI in a PBS buffer. The microplate was read at 415 nm excitation and 510-560 nm emission. (iii) For the recPPBD binding assay, gel shift assays were performed similarly as previously described (Müller et al, 2009; Labberton et al, 2016), with some modifications. RecPPBD (9 μg) was incubated with synthetic polyP or the potential interfering molecule for 30 minutes at RT. Then the samples were loaded and run in a 7,5% native PAGE. Proteins were visualized using Instant blue Coomassie protein stain (Abeam, Cat# 119211).

Cell-based assays with JC-D8, DAPI and recPPBD

With some modifications (see below), N2a cells were initially pre-incubated with synthetic polyP 100 μM for 5 minutes. Subsequently, cells were fixed with 4% PFA, permeabilized in 0.1% Triton X-100 and blocked with goat serum. Next, cells were incubated with the following primary antibodies: anti-NeuN (1:200, rabbit, Cat. # MAB377, Millipore) and anti-Xpress (1:500, mouse, Cat. # R910-25, Invitrogen) overnight at 4°C. Next, cells were incubated with Alexa-conjugated secondary antibodies (1:500, anti-mouse Alexa⁴⁸⁸, Cat. # A-11029, Invitrogen and 1:500, anti-rabbit Alexa⁵⁴⁶ Cat. # A-11029). Finally, cells were mounted on coverslips with fluoromount-G™ (Electron Microscopy Sciences, Cat. # 17984-25). Cells images were taken from an epifluorescence microscope (Nikon, Eclipse TE2000e) and images were analyzed with the Fiji ImageJ software (8 bits, measuring intensity from 0 to 225).

(i) In the recPPBD assays, after fixation and blocking, the N2a cells were incubated for 40 min with recPPBD (30 μg/ml) alone or pre-treated with the different polyP chain-sizes or the interfering molecules for 1 h at RT (see Table Figure S1 for concentrations). Next, the primary and secondary antibody incubations were performed together with the nuclear dye TOPRO3 (1:1000, Cat. # T3605, Invitrogen). (ii) In the JC-D8 assays, unfixed N2a cells were incubated for 1 h with JC-D8 (7 μM) alone or pre-treated with the different polyP chain-sizes or the interfering molecules (as above). Next, cells were fixed, permeabilized, blocked and incubated with the antibodies and nuclear dye TOPRO3 (as above). (iii) In the DAPI assays, after fixation and blocking, the N2a cells were incubated for 1 h with DAPI (0.5 μg/ml Sigma, Cat. # D9542) alone or pre-treated with the different polyP chain-sizes or the interfering molecules, and then treated with the antibodies.

Intracellular Ca²⁺ imaging assays

Ratiometric Ca²⁺ imaging experiments with fura-2AM were performed in primary spinal cord neurons as described (Fritz et al., 2013). Briefly, neurons were incubated with 5 μM fura-2-AM (F1221, Invitrogen-Thermo Fisher Scientific, Waltham, MA, USA) dissolved in standard extracellular solution supplemented with 0.02% pluronic acid (P6867, Invitrogen-Thermo Fisher Scientific, Waltham, MA, USA) for 50 min at 37°C in darkness. Fluorescence measurements were made using an inverted Nikon Ti microscope fitted with a 12-bit cooled ORCA C8484-03G02 CCD camera (Hamamatsu, Hamamatsu City, Japan). Fura-2 was excited at 340 nm and 380 nm at 0.5 Hz with a Polychrome V monochromator (Till Photonics), with exposure times no longer than 40 ms; the emitted fluorescence was filtered with a 510 nm longpass filter. Calibrated ratios were displayed online with HCImage v2 software (Hamamatsu, Japan). A solution containing elevated K⁺ (50 mM KC1) was perfused at the end of the protocol to determine the viability of the neurons in the entire field. Only cells showing a [Ca²⁺]_i increase in response to a 50 mM elevation in extracellular K⁺ were included in the analysis. A random pool of 40 cells were selected to generate a “calcium activity fingerprint graph” using the GraphPad software. Additionally, Ca²⁺ imaging was also performed in spinal cord neurons using the genetically encoded calcium indicator GCaMP6s. For this, 3-4 DIV neurons were infected with AAV1/2 viral particles containing the insert hSyn1-mRuby2-GSG-P2A-GCaMP6s-WPRE-pA (Addgene). At 6-7 DIV, Ca²⁺ frequency was recorded using an epifluorescence microscope (Nikon TE2000e, 20X objective and Andor Zyla 5.5 camera). Images were taken every 50 ms for 1-1.5 min using an excitation wavelength at 480 nm and emission at 510 nm. For the quantification of the calcium event frequency, the soma of isolated spinal cord neurons was selected and analyzed using the ‘”Z profiler” plugin (https://imagej.nih.gov/ij/plugins/z-profiler.html) of the ImageJ Fiji software. Each intensity peak was counted manually. For both calcium imaging assays, at least 40 neurons per culture were analyzed. Each condition was replicated in 3-4 independent cultures.

TIRF imaging

To visualize vesicular fusion events, we used total internal reflection fluorescence microscopy (TIRF) (Márquez-Miranda et al., 2016). The imaging setup was equipped with a 100X objective (Olympus, Japan), and an EM-CCD camera (Luca-S, Andor, Ireland). Ctrl and mutSOD1 primary astrocytes were loaded with JC-D8 (5 μM) for 30 min at room temperature in HEPES-buffered salt solution (HBSS). Videos were acquired using WinFluor software and later converted and analyzed using Z Profiler plugin in ImageJ software. For the TIRF assays, 4 neurons per condition were quantified.

Transcriptomic assays

Spinal cord cultures (4 DIV) were treated for 90 min with polyPL (100 μM), mutSOD1-ACM or wtSOD1-ACM. Total RNA was extracted from four independent cultures using RNeasy kit (Qiagen). RNA quality was verified on a bioanalyzer before library preparation. TruSeq stranded mRNA protocol (Illumina) was used to generate libraries that were sequenced on an Illumina HiSeq4000 sequencing instrument. Single-end 50bp reads were aligned to the mouse genome and differentially expressed genes were analyzed using NOISeq method (BGI). Functional analysis of differentially expressed genes dataset was performed using DAVID database of those genes with a cut-off fold change of 2. Data are expressed as polyP_L vs. wtSOD1-ACM and mutSOD1-ACM vs. wtSOD1-ACM.

Electrophysiological assays

Action potentials (APs) were recorded by conventional whole-cell patch clamp (Fritz et al., 2013; Hermosilla et al., 2017; van Zundert et al., 2008). For the recording of spontaneous Aps, neurons were hold under the current-clamp mode (I=0 pA). Borosilicate glass pipettes (World Precision Instruments, Cat. # TW150F-4) were pulled to 2–5 MΩ resistance and filled with internal solution containing (mM): 105 K-Gluconate, 35 KCl, 8.8 NaCl, 0.5 EGTA 10 HEPES, 30 sucrose (pH 7.2 adjusted with KOH).The bath solution (artificial cerebrospinal fluid solution, ACSF) contained (mM): 150 NaCl, 5.4 KCl, 2 CaCl₂, 2 MgCl₂, 10 HEPES, 10 glucose (pH 7.4 adjusted with NaOH). Data were acquired at room temperature (20-23°C) using an Axopatch 200B (Molecular Device) or EPC7 (HEKA) amplifier and pClamp 10.7 software (Axon Instruments), low pass-filtered at 1 or 5 kHz, and digitized at 10 kHz using a 1322a Digidata (Axon Instruments). Spontaneous APs were recorded in the absence of injected current for at least 2 min at each paired condition in cells with R_access<10 MΩ. Cells were discarded when they had no APs (measured the first 2 min), showed excessive hyperactivity (firing frequency higher than 1 Hz) or when displaying an unstable baseline. The frequency of AP was calculated off-line using the Threshold Search option of the Event Detection of Clampfit 10.7 software. The baseline was manually adjusted and the AP was counted if the membrane potential changed 30 mV or more. The number of APs 1 min before and 1 min after ACM application were calculated. All neurons with frequencies between 0.1 and 1.0 Hz and with no change in R_access were analyzed. For each electrophysiological analysis, at least 10 neurons were analyzed per condition.

AAV9 vector design and packaging

AAV9 vectors were produced by triple transient transfection of HEK293T cells and purified by iodixanol gradient centrifugation as previously described (Weismann et al., 2015). AAV vectors carry two AAV2 inverted terminal repeats flanking a transgene expression cassette composed of the cytomegalovirus enhancer fused to a chicken β-actin promoter/rabbit β globin intron, cDNA of the yeast scPPX1 (PPX) gene, fused to GFP, and an SV40 poly A. Two different GFP-PPX versions were generated: (i) PPX-GFP (Abramov et al., 2007) (pTR-CB-Cyto.GFP.PPX; 8.73 x 10¹²vg/ml), and (ii) PLong-GFP-PPX (pTR-CB-PLong.GFP.PPX; 5.12 x 10¹²vg/ml), where the native GFP-PPX chimeric protein was genetically fused to the PTEN membrane-permeant peptide (Hopkins et al., 2013).

Treatment of mutSOD1 mice with AAV9-CMV-PPX-GFP

AAV9 vectors were delivered through an intracerebral ventricular (ICV) injection as previously described (Stoica et al., 2016). Briefly, mutSOD1 mice received 1 μl of undiluted AAV9-pTR-CB-PLong.GFP.PPX (5.12 x 10¹² vg/ml) per cerebral lateral ventricle at P0-1 As controls, mutSOD1 were not treated or treated with AAV9-CMV-GFP. Disease onset was defined as the age of first appearance of limb weakness detected by failure of the mice to remain of a rotarod apparatus (fixed speed, 14 rpm for 180 sec) (Harvard Apparatus touchscreen PanLab, cat# LE8305). Three trials were performed for each animal (since P60) and the longest time taken to fall was recorded. Once animals developed partial paralysis, hydrated gelatin with food was placed on the bottom of the cage to combat dehydration. To determine ‘survival’ reliably and humanely, an artificial endpoint was used, in which animals were euthanized when they could not right themselves within 30 sec when placed on their sides, or groom their faces (detected by the development of infection in one or both eyes).

Treatment of mutSOD1 Drosophila with UHRA10

For drug treatment and fly rescue experiments, UHRA10 was dissolved in water at final 500 nM. We used 100 μl covering the whole surface of the food of the rearing tube to ensure equal access to the drug and adult flies were transferred after absorption of the liquid the same days, some hours later. Control flies were treated in the same manner using water (drug vehicle). Flies were flipped every 2 days. At least three independent cohorts were used for each genotype with controls and queries running in parallel each time.

For fly climbing assays, measurements were taken from three independent aged cohorts (50-60 aleatory individuals each time). Fifteen to 20 flies per assay were placed in a clean tube 24 h before the experiment. Tubes were tapped and recorded. The trajectory of flies in the tube was traced as a segmented line. Kymograph plugin for ImageJ was used to calculate the speed of each moving object. The slope of each trajectory in the tube was calculated manually using the end point coordinates, multiplied by the length of the tube (in mm) and divided by the speed of the movie (30 frames/s) to obtain an individual speed in mm/s. All procedures were performed on ImageJ. An average speed per genotype per time point was obtained from at least two different crosses. Average speeds were compared using unpaired t-test.

For JC-D8 staining, 3- and 30-day old fly brains were hand dissected, fixed in formalin 4% for 5 min, and JC-D8 (10 μM) was added for 1 hour at RT, and TOPRO3 (Invitrogen, T3605; dilution 1:5000) for 10 min. The brains were washed with PBS and then mounted for visualization. Experiments were performed 3 times and individuals for each sample were taken from independent aged cohorts. Images of Drosophila lobal globes were taken on a Leica TCS SP8 confocal microscope with a × 63 oil objective and × 2 digital zoom, and a z-step of 0.5 μm optical sections. The following laser wavelengths were used to detect TOPRO3 (Ex 638 nm and Em 643-776 nm) and JC-D8 (Ex 488 nm and Em 510-560 nm). Maximum intensity projections of confocal z-stack images (containing 8–10 stacks) were analyzed.

For western blot of Drosophila brain samples, for each sample twenty frozen head flies were homogenized in 50 μl of RIPA buffer. For total hSOD1 detection, samples were boiled 10 min at 95°C with laemmli buffer and run on 10% SDS-PAGE. For misfolded hSOD1 detection, samples were treated under non-reducing condition (no SDS, no β-mercaptoethanol and no heat), run on 10% naïve PAGE and transferred to nitrocellulose membranes following standard procedures for western blot. Membranes were blocked in blocking buffer (PBS, 0.1% tween, 5% milk) and were incubated overnight with primary antibodies for hSOD1 (Abeam, ab238052; 1:1000), misfolded hSOD1 (Medimabs, MM-0070-p; 1:1000), GFP (Invitrogen, A-11122; 1:2000) and Repo (DSHB, dilution 1:100). The latter was used as loading control. The next day, membranes were washed and incubated with HRP-conjugated second antibody at room temperature for 2 h. Experiments were performed three times with 20-30 heads per sample obtained from independent aged cohorts.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using GraphPad Prism 5 software. One-way ANOVA followed by the Bonferroni post-hoc was utilized when making multiple (three or more) comparisons. Student’s t-test (rodent samples) and Mann-Whitney U-test (human samples) were performed when two populations were examined. In all figures, data is reported as mean±S.E.M.; *P < 0.05, **P < 0.01, ***P < 0.001 compared to control. #P<0.05, ##P<0.01, and ###P<0.001, compared to mutSOD1-ACM or polyP_L (as indicated in legends). Each Figure shows the exact size of populations (n) used for each experiment and the legend explains its significance.

Highlights.

• PolyP is enriched in human and mouse ALS/FTD astrocytes in vitro and in vivo

• Excessive polyP released by ALS/FTD astrocytes is toxic to primary MNs

• In vitro studies indicate that polyP is a new therapeutic target of ALS/FTD

• Study of human samples indicate that polyP is new hallmark and biomarker of ALS/FTD