Introduction

During neuronal development, glial cells play critical roles in many aspects of the growth and establishment of neural networks. Astrocytes in particular are integral for regulating neuronal extension, synaptic formation, and circuitry homeostasis via reciprocal astrocyte-neuronal signalling systems within the central nervous system (CNS) (reviewed in Gradisnik & Velnar, 2023). Indeed, prior to the establishment of neural circuits, neurons must first be prompted by astrocytes to undergo a developmental switch that allows them to form synapses (Barker & Ullian, 2008). In addition, astrocytes regulate the timing of structural synapse formation across brain regions, as well as the activity and function of synapses once formed (Liu et al. 1996; Pfrieger and Barres 1997; Li et al. 1999; Ullian et al. 2001). Given this, it is not surprising that astrocyte dysfunction has been associated with neurodevelopmental disorders characterized by either morphological or functional synaptic abnormalities, such as autism spectrum disorders (ASDs).

In many ASDs, genetic mutations lead to either the enhanced formation or reduced elimination of dendritic spines that cause greater densities of synaptic connections within the developing CNS (Tang et al., 2014; Guang et al., 2018). In fragile X syndrome (FXS), the leading monogenic cause of intellectual disability and ASD, higher densities of dendritic spines with immature morphologies and excitatory activity result in hyperexcitable cortical connections, sensory hyperresponsivity, attention deficit, and seizures (Farzin et al., 2006; Musumeci et al., 1999). In FXS, an expansion mutation in the Fmr1 gene results in the absence of the fragile X messenger ribonucleoprotein (FMRP) in both neurons and astrocytes (Pieretti et al., 1991; Sutcliffe et al., 1992; Turner et al., 1996; Verkerk et al., 1991; Youings et al., 2000). Deletion of FMRP in astrocytes specifically affects the expression and release of synaptogenic factors, glutamate transmission, and neuronal morphology, as well as the activity and function of neuronal circuits (Cheng et al., 2016; Higashimori et al., 2016; Hodges et al., 2017; Jacobs & Doering, 2010; Wallingford et al., 2017). While the astrocyte-mediated molecular mechanisms involved in FXS dysfunction continue to be elucidated, several studies have demonstrated a potential role for the purinergic signalling system.

Gliotransmission of purines and pyrimidines (ex. ATP, UTP) and their metabolites facilitate widespread astrocyte-astrocyte and astrocyte-neuron communication within the CNS (Abbracchio et al., 2009). This communication is driven by the activation of two large receptor families, P1 and P2, that are expressed by both astrocytes and neurons, and classified according to their structure and ligand affinity. The P1 purinergic receptor family contains four members, A1, A2A, A2B, and A3, which are all metabotropic and bind adenosine. The P2 family is subdivided into ionotropic P2X receptors (P2X1–7) and metabotropic P2Y receptors (P2Y1,2,4,6,11–14) that are activated by nucleoside di- and triphosphates to increase intracellular calcium levels and promote signal transduction (Abbracchio et al., 2009). The uridine-based signalling ligands utilized by P2Y receptors are generated primarily by astrocytes through a combination of de novo synthesis pathways and the action of intracellular kinases that transfer phosphate residues from ATP to form UMP, UDP, and UTP (Dobolyi et al., 2011). Using an Fmr1 knockout (KO) mouse model of FXS, our work has demonstrated that ATP/UTP-specific P2Y2 and UDP/UTP-sensitive P2Y6 receptor levels are elevated in Fmr1 KO cortical astrocytes and correspond with prolonged Fmr1 KO astrocyte activation, secretion of synaptogenic protein TSP-1, and hyperexcitability of neuronal circuits (Reynolds et al., 2021, 2024). Cortical astrocytes derived from FXS patients also exhibit elevated activation in response to ATP stimulus, further supporting a purinergic link to FXS astrocyte pathogenesis (Ren et al., 2023).

While our understanding of purinergic receptors in disorders such as FXS continues to expand, relatively little is known about the production, secretion, or metabolism of the purine or pyrimidine ligands. ATP/UTP nucleotides are known to stimulate the release of nucleoside triphosphates by astrocytes in an autocrine manner that can further drive activation of both glial and neuronal purinergic receptors (Abbracchio et al., 2009). Unbound ATP and UTP are rapidly hydrolysed to their di- and monophosphate forms by the astrocyte membrane-bound ectonucleoside triphosphate diphosphohydrolase enzyme CD39; while CD73, or ecto-5′-nucleotidase, removes single phosphates from AMP and UMP to produce adenosine and uridine (Fig. 1) (Dobolyi et al., 2011; Zimmermann, 1996). Purines and pyrimidines can also undergo re-uptake via astrocyte nucleoside transporters for recycling and re-release (Lie et al., 1999; Wink et al., 2003). The action of astrocyte-based ectonucleotidases and transporters thereby helps to modulate ligand availability for both P2X and P2Y receptors, as well as those of the P1 receptor family. In this study, we explored the extent of purinergic dysregulation in FXS astrocytes by utilizing hydrophilic interaction liquid chromatography (HILIC) coupled with mass spectrometry (MS) to compare the intracellular and extracellular levels of purinergic metabolites by wildtype (WT) versus Fmr1 KO mouse astrocytes. In addition, we also investigated the expression of astrocyte-bound CD39 and CD73 ectonucleotidases responsible for the extracellular metabolism of secreted ATP and UTP. Our findings demonstrate significant variations in purine production and metabolism in Fmr1 KO astrocytes and support previous reports of purinergic signalling dysregulation in FXS.

Fig. 1
figure 1

Metabolism of adenosine- and uridine-based nucleosides and nucleotides. Astrocytes produce, secrete, and internalize purine and pyrimidine metabolites through the action of intracellular or membrane-bound enzymes, as well as pannexin channels and nucleoside transporters, respectively. ATP, adenosine triphosphate; UTP, uridine triphosphate; ADP, adenosine diphosphate; UDP, uridine diphosphate; AMP, adenosine monophosphate; UMP, uridine monophosphate; NMP, nucleoside monophosphate; NDP, nucleoside diphosphate; CD39, ectonucleoside triphosphate diphosphohydrolase-1; CD73, ecto-5'-nucleotidase

Full size image

Methods

Animals

Two genotypes of mice, wildtype (WT) and Fmr1−/− (Fmr1 KO; FVB.129P2[B6]-Fmr1tm1Cgr), were bred and housed at the McMaster Health Sciences Central Animal Facility. All mouse housing, handling, and euthanization protocols followed the standards set by the Canadian Council on Animal Care and were authorized by the McMaster Animal Ethics Board (Animal Utilization Protocol 17-04-11). Pups of both sexes were euthanized by decapitation at postnatal day (P) 1–3 for astrocyte dissection and primary culture.

Primary Astrocyte Culture

WT and Fmr1 KO primary cortical astrocyte cultures were prepared and maintained following protocols described previously by Jacobs and Doering (2009). Six cortical hemispheres, obtained from three postnatal day (P) 1-3 mouse pups, were pooled per genotype to form a single primary culture (n). Cells of the developing cerebral cortex were isolated by removing the hippocampus and meninges from each cortical hemisphere under a Zeiss Stemi SR stereo microscope (Carl Zeiss, Oberkochen, Germany), then dissociating cortical tissue in ice-cold Hanks’ buffered saline solution (HBSS; Invitrogen, Waltham, MA, USA) with 1 mg/mL DNase (Roche Applied Science) and 0.25% trypsin (Invitrogen). Cells were plated in a glial-selective medium consisting of minimum essential medium (Gibco, Waltham, MA, USA) supplemented with 10% horse serum (Gibco) and 0.6% d-(+)-glucose (Sigma-Aldrich, St. Louis, MO, USA), and were maintained in a humidified incubator at 37 °C and 5% CO2 (NuAire, Plymouth, MN, USA). Primary cultures received half media changes every 2–3 days, starting at 24h post-dissection, until cells were 75–90% confluent (~ 6–8 days).

After reaching 75–90% confluence, primary astrocyte cultures intended for liquid chromatography/mass spectrometry (LC/MS) were lifted with 0.05% trypsin-ethylenediaminetetraacetic acid (trypsin–EDTA; Gibco) for reseeding. Astrocytes were reseeded in six-well plates (Corning, NY, USA) pre-coated with 1 mg/mL poly-l-lysine (Sigma-Aldrich) and 10 µg/mL laminin (Invitrogen), at a density of 85 000 cells/well. After 24h, glial media were replaced with a serum-free glial media comprised of minimum essential media (Gibco) plus 0.6% d-(+)-glucose (Sigma-Aldrich), then cultures were maintained at 37 °C and 5% CO2 for an additional 48 h.

Liquid Chromatography/Mass Spectrometry Sample Preparation for Intracellular Metabolites

For sample collection, plates were placed on ice and washed 2 × with ice-cold phosphate buffered saline (Gibco, Waltham, MA, USA) to halt ectonucleotidase activity. All six wells per plate were collected and pooled per sample to obtain sufficient quantities for nucleoside triphosphate detection, while a seventh well was collected to obtain a representative cell count for normalization. Astrocytes intended for LC/MS were mechanically lifted with a cell scraper (Corning) in 500 μL ice-cold LC/MS-grade methanol (Fisher Scientific, Waltham, MA, USA), while those intended for representative cell counts were lifted with 0.05% trypsin–EDTA. LC/MS cell suspensions were vortexed vigorously for 2 min, then stored at -80 °C until extraction.

Isotopically labelled adenosine-13C5 (Toronto Research Chemicals, North York, ON, Canada) was added to each cell suspension to act as an internal standard, achieving a final concentration of 2.5 ng/µL. Cell suspensions were vigorously vortexed for 2 min and centrifuged at 10,000 rpm for 5 min (4 °C), then supernatants were retained for LC/MS analysis. The extraction process was repeated twice with an additional 500 µL methanol, for a total of 3 pooled extractions per sample. The final pooled sample was dried with nitrogen gas and stored at − 80 °C until analysis. Samples were resuspended in 5 mM NH4OAc (50 µL; Sigma-Aldrich) immediately prior to LC/MS.

Solid Phase Extraction (SPE) for Extracellular Metabolites

For detection of secreted nucleotides, each confluent primary astrocyte culture was passaged (as above) and replated into two T75 flasks. The cultures were maintained for approximately 4 additional days in vitro (DIV) or until they reached > 75% confluence. Glial media was then replaced with 6 mL of serum-free glial media with 100 µM NTPDase inhibitor ARL 67156 trisodium salt (Tocris) per flask and incubated for 48 h. The media was then collected and pooled from both flasks per experimental sample. Adenosine-13C5 was added to each sample (2.755 µL of 100 µM stock) to act as an internal standard. To normalize samples to corresponding cell counts, astrocytes were removed from T75 flasks, centrifuged for 5 min (1500 rpm), and resulting pellets were resuspended in 1 mL of glial media. A small aliquot of the cell suspension was extracted, and cell numbers were analysed using a hemocytometer under 10X magnification.

A STRATA X-AW SPE column was connected to a vacuum manifold, which applied ~ 5 mmHg of vacuum pressure. The column was first conditioned with 1 mL each of MS-grade methanol (MeOH), 2:25:73 formic acid/MeOH/H2O, and H2O alone, and positioned over a collection tube. Astrocyte-conditioned media samples were added to the column at a rate of 1 drop/sec. The column was thoroughly washed with 1 mL MS-grade H2O, the vacuum dried for 5 min prior to sample elution in 1 mL 2:98 NH4OH/methanol. Collected samples were flash frozen in liquid nitrogen and stored at -80°C prior to lyophilization using a 6L Labconco FreeZone® Freeze Dryer System (Centre for Microbial Chemical Biology, McMaster University). Immediately prior to LC/MS, lyophilized samples were reconstituted in 30 μL of 5 mM NH4OAc (MS-grade) and transferred to spring loaded inserts for appropriate MS autosampler loading.

Liquid Chromatography/Mass Spectrometry

A Luna NH2 HILIC column with dimensions of 2.0 × 150 mm, 100 Å pore size, and 5 µm particle size (Cat# 00F-4378-B0, Phenomenex, Torrance, CA, USA) was utilized to separate purinergic targets for analysis. LC was performed on an Agilent 1260 HPLC system with a flow rate of 0.4 mL/min and an injection volume of 2 µL. Nucleoside mono-, di-, and triphosphates, with the exception of UMP, were able to be separated using the following aqueous and organic mobile phases: (A) 100 mM ammonium acetate (NH4OAc; Sigma Aldrich) in water, adjusted to pH 9 with ammonium hydroxide (NH4OH; Caledon); and (B) 100% acetonitrile (CH3CN; Fisher Scientific), respectively. Adenosine, uridine, cytidine, and UMP were poorly detected using these parameters but were robustly detected when the concentration of mobile phase A was reduced to 10 mM NH4OAc (pH 9). As this lower concentration of NH4OAc could not be optimized to produce sharp nucleoside triphosphate peaks, samples were run twice, first with A: 100 mM NH4OAc (pH 9) to detect ATP, ADP, AMP, UTP, UDP, and inosine, and again with A: 10 mM NH4OAc (pH 9), to permit quantification of UMP, adenosine, uridine, and cytidine. The same LC/MS gradient was utilized for both applications and is summarized in Table 1. The LC system was paired with an Agilent 6550 iFunnel Q-TOF Mass Spectrometer (Agilent, Santa Clara, CA, USA) for purinergic detection. MS parameters included negative electrospray ionization, and high sensitivity detection with 2GHz extended dynamic range.

Table 1 Optimized gradient utilized for LC/MS nucleotide analysis
Full size table

Prior to sample analysis, ESI- source parameters were optimized with authentic purinergic standards (Sigma-Aldrich). The following purinergic targets were detected at the specified observed m/z and retention times: ATP (505.9985 m/z; 8.7 min), ADP (426.0221 m/z; 8.4 min), AMP (346.0558 m/z; 7.9 min), UTP (482.9613 m/z; 8.5 min), UDP (402.9949 m/z; 5.7 min), UMP (323.0286 m/z; 7.7 min), adenosine (266.0895 m/z; 4.2 min), uridine (243.0623; 4.2 min), cytidine (242.0782; 5.7 min), inosine (267.0735 m/z; 5.1 min), and the internal standard adenosine-13C5 (271.1068 m/z). Within each genotype, the experimental sample consisted of 6–7 individual cultures (n = 6 or 7) prepared from 4 separate litters and run in triplicate. Selected targets were identified using Agilent MassHunter Qualitative Analysis B.07.00 software (Agilent), then analysed using Agilent MassHunter Quantitative Analysis B.07.01 software (Agilent). Mean nucleoside abundance was normalized to representative cell counts, as well as to the relative abundance of adenosine-13C5 internal standard.

Western Blotting

After reaching 75–90% confluence, primary astrocyte cultures intended for western blotting were lifted with 0.05% trypsin–EDTA and pelleted, then flash frozen in liquid nitrogen and stored at -80°C until homogenization. A total of 15 WT (n = 15) and 14 Fmr1 KO (n = 14) samples were used for CD39 western blotting, and 8 samples per genotype (n = 8) were used for CD73 western blotting, obtained from individual cultures prepared from separate litters. Samples were homogenized on ice in brain extraction buffer (25 mM HEPES pH 7.3, 150 mM KCl, 8% glycerol, 0.1% NP-40, one Roche ULTRA protease inhibitor tablet, and one Roche PhoSTOP phosphatase inhibitor tablet) (Reynolds et al., 2021), and then total protein content was quantified within lysates (DC protein assay; BioRad, Mississauga, ON, Canada) to determine gel loading dilutions.

Western blotting samples were prepared with Laemmli sample buffer plus 2.5% ß-mercaptoethanol (BioRad) to achieve a dilution of 10 µg protein per lane. Samples were run on polyacrylamide TGX Stain-Free 4–12% gradient gels (BioRad) to separate proteins, activated with UV light for 45 s to permit visualization of “total protein” (tryptophan), then transferred to polyvinylidene difluoride (BioRad) membranes. Stain-free total protein loading control images were acquired using a ChemiDoc imaging system (BioRad), then membranes were blocked with 5% non-fat milk for 1 h. Primary antibodies against CD39 (rabbit monoclonal; 1:500; Abcam, Cambridge, UK Cat# ab223842, RRID:AB_2889212) or CD73 (mouse monoclonal; 1:500; R7D Systems, Minneapolis, MN, USA Cat# MAB57951) were applied overnight (4 °C), followed by the corresponding secondary antibodies donkey anti-rabbit horseradish peroxidase (1:2500; GE Healthcare, Chicago, IL, USA) or donkey anti-mouse horseradish peroxidase (1:5000; GE Healthcare). Bands for astrocyte membrane-bound CD39 were detected at ~ 75 and ~ 52 kDa, corresponding to glycosylated (active) and unglycosylated (immature) forms of the enzyme, respectively (Zhong et al., 2001). Membrane-bound glycosylated CD73 was detected at ~ 62 kDa (Adzic & Nedeljkovic, 2018; Zhou et al., 2019). Membranes were developed using Clarity MAX enhanced chemiluminescence (ECL) substrate (BioRad), and chemiluminescence images were acquired using the ChemiDoc imaging system (BioRad). Densitometry was performed using ImageLab 6.0.1 software (BioRad) to normalize bands of interest to the density of total protein bands.

Statistics

All statistical analyses were conducted using GraphPad Prism 8.0 software (GraphPad Software, San Diego, CA, USA). Comparisons between WT and Fmr1 KO are expressed as fold change relative to WT means and were obtained using unpaired two-tailed t tests with significance at p < 0.05. For the metabolite analysis in WT and KO astrocyte-conditioned media, One-Way ANOVA was used with Tukey’s posthoc multiple comparison test, comparing each mean. Relationships of interest were annotated, with p < 0.05 indicating significance.

Results

Intracellular UDP is Elevated in Fmr1 KO Astrocytes

Here, we utilized a hydrophilic interaction liquid chromatography protocol paired with mass spectrometry (HILIC-LC/MS) to detect intracellular purinergic signalling molecules within Fmr1 KO and WT primary astrocytes. The internal standard adenosine-13C5 was used to normalize all ligand measurements and showed good reproducibility during the LC/MS sequence (data not shown). We observed several differences in intracellular nucleoside tri-, di-, and monophosphate abundance. Within the uridine-based family of ligands, intracellular UTP levels were unchanged in WT vs Fmr1 KO astrocytes (n = 6, p = 0.2673; Fig. 2A, B), while levels of intracellular UDP were elevated in Fmr1 KO astrocytes relative to WT (n = 6, p = 0.0118; Fig. 2C, D). The abundance of astrocyte intracellular UMP was also unchanged between genotypes (WT n = 5/Fmr1 KO n = 6, p = 0.7292; Fig. 2E, F).

Fig. 2
figure 2

LC/MS quantification of intracellular UTP, UDP, and UMP in WT and Fmr1 KO primary cortical astrocytes. Representative extracted ion chromatograms of WT and Fmr1 KO astrocyte intracellular UTP (A), UDP (C), and UMP (E). Relative abundance of intracellular UTP (B), UDP (D), and UMP (F) in WT and Fmr1 KO cultured astrocytes, showing increased UDP levels in Fmr1 KO astrocytes. Abundance is presented as fold change relative to mean WT signal intensity and is normalized to both representative cell count and adenosine-13C5 internal standard signal intensity. Data presented as means ± SEM. n = 6, with the exception of WT UMP n = 5; *p < 0.05

Full size image

Intracellular ATP and Adenosine are Reduced in Fmr1 KO Astrocytes

In contrast, the adenosine-based family of ligands showed additional differential patterns of intracellular levels between WT and Fmr1 KO astrocytes. Intracellular ATP levels were notably reduced in Fmr1 KO astrocytes in comparison to WT (n = 6, p = 0.0425; Fig. 3A, B), while intracellular ADP quantities did not change between genotypes (n = 6, p = 0.4195; Fig. 3C, D). However, intracellular levels of AMP were elevated in Fmr1 KO astrocytes relative to WT (n = 6, p = 0.0122; Fig. 3E, F).

Fig. 3
figure 3

LC/MS quantification of intracellular ATP, ADP, and AMP in WT and Fmr1 KO primary cortical astrocytes. Representative extracted ion chromatograms of WT and Fmr1 KO astrocyte intracellular ATP (A), ADP (C), and AMP (E). Relative abundance of intracellular ATP (B), ADP (D), and AMP (F) in WT and Fmr1 KO cultured astrocytes, showing reduced ATP levels and elevated AMP levels in Fmr1 KO astrocytes. Abundance is presented as fold change relative to mean WT signal intensity and is normalized to both representative cell count and adenosine-13C5 internal standard intensity. Data presented as means ± SEM. n = 6; *p < 0.05

Full size image

Intracellular nucleosides function as mono-, di-, and triphosphate precursors, while the nucleoside adenosine also acts as a gliotransmitter by binding to the P1 purinergic receptor subfamily to exert both excitatory and inhibitory effects. Here, adenosine abundance was reduced in Fmr1 KO astrocytes (WT n = 5/Fmr1 KO n = 6, p = 0.0494; Fig. 4A, B). Intracellular levels of uridine (n = 5, p = 0.6328; Fig. 4C, D), inosine (n = 6, p = 0.6111; Fig. 4E, F), and cytidine (WT n = 5/Fmr1 KO n = 6, p = 0.7062; Fig. 4G, H) were unchanged.

Fig. 4
figure 4

LC/MS quantification of intracellular nucleosides in WT and Fmr1 KO primary cortical astrocytes. Representative extracted ion chromatograms of adenosine (A), uridine (C), inosine (E), and cytidine (G). Relative abundance of intracellular adenosine (B), uridine (D), inosine (F), and cytidine (H) in WT and Fmr1 KO cultured astrocytes, show decreased adenosine levels in Fmr1 KO astrocytes only. Abundance is presented as fold change relative to mean WT signal intensity and is normalized to both representative cell counts and adenosine-13C5 internal standard signal intensity. Data presented as means ± SEM. WT n = 5, with the exception of WT inosine n = 6; Fmr1 KO n = 6; *p < 0.05

Full size image

Glycosylated CD39, but not CD73, is Elevated in Fmr1 KO Astrocytes

The abundance of purinergic ligands is governed in part by the action of the CD39 ectonucleoside triphosphate diphosphohydrolase enzyme, which breaks down UTP and ADP into their di- and monophosphate forms (Huang et al., 2021). Active CD39 is localized to the plasma membrane, where it was detectable at the commonly observed molecular weight of ~ 75 kDa. The ~ 52 kDa CD39 band we observed here has previously been shown to be indicative of an intracellularly localized form that only becomes active once it is glycosylated and incorporated within the plasma membrane (Zhong et al., 2001). Here, we found that the glycosylated, active form of CD39 (~ 75 kDa) was more highly expressed on the surface of Fmr1 KO astrocytes (n = 14) than WT (p = 0.0489; n = 15; Fig. 5A). In contrast, levels of the immature, unglycosylated form (~ 52 kDa) were reduced on the Fmr1 KO astrocyte surface relative to WT (p = 0.0386; Fig. 5A). Despite these differences, the total abundance of glycosylated plus immature CD39 did not differ between genotypes (p = 0.5799), indicating that glycosylation rather than protein synthesis is altered in FXS. The membrane-bound ecto-5’-ectonucleotidase enzyme, also known as CD73, converts UMP into uridine and is therefore also important in regulating purinergic ligand availability. Here, no differences were observed in the level of astrocyte-associated CD73 (p = 0.5185; n = 8) between WT and Fmr1 KO cultures (Fig. 5B).

Fig. 5
figure 5

Western blotting quantification of WT and Fmr1 KO primary cortical astrocyte membrane-bound ectonucleotidases. A. Representative western blot of astrocyte membrane-bound CD39 ectonucleotidase and corresponding total protein control, showing bands for both glycosylated (~ 75 kDa) and unglycosylated (~ 52 kDa) CD39. B. Glycosylated (i.e. active) CD39 levels were elevated in Fmr1 KO astrocyte cultures (n = 14) relative to WT (n = 15), while unglycosylated (i.e. immature) CD39 levels (C) were reduced in Fmr1 KO; when both forms were added together, the total levels of CD39 were unchanged between genotypes. D. Representative western blot and corresponding total protein image of WT and Fmr1 KO astrocyte-associated CD73. E. Astrocyte-bound CD73 was present at similar levels in WT and Fmr1 KO (n = 8) cultures. Enzyme abundance was normalized to total protein and presented as fold change relative to mean WT levels. Data presented as means ± SEM; *p < 0.05

Full size image

Secreted Adenosine Levels are Elevated by Fmr1 KO Cortical Astrocytes

Levels of extracellular purines and pyrimidines were also measured in the conditioned media of Fmr1 KO and WT astrocyte cultures. Extracellular nucleosides are the result of either direct secretion by astrocytes or the metabolism of their phosphate bound precursors. Here, we found that the concentration of uridine present within WT astrocyte-conditioned media was significantly higher than the levels of other nucleosides, including adenosine (p = 0.0021), inosine (p = 0.0029), guanosine (p = 0.006), and cytosine (p = 0.006) (one-way ANOVA; Fig. 6A). Conversely, there were higher average concentrations of uridine present in media collected from Fmr1 KO astrocytes in comparison to inosine (p = 0.0118), guanosine (p = 0.0060), and cytosine (p = 0.0061), but not adenosine (p = 0.4721) (one-way ANOVA; Fig. 6B). Across genotypic comparisons, only adenosine was measured in significantly higher concentrations in Fmr1 KO astrocyte media (n = 5) than in WT (n = 7; p = 0.0462; Fig. 6C). All other genotypic comparisons showed similar concentrations of uridine (WT n = 7/KO n = 6, p = 0.7578; Fig. 6D), inosine (WT n = 6/KO n = 6, p = 0.8734; Fig. 6D), guanosine (WT n = 7/KO n = 6, p = 0.5185; Fig. 6D), and cytidine (WT n = 7/KO n = 6, p = 0.6918; Fig. 6D) within the collected media of WT and Fmr1 KO astrocyte cultures (via Student t tests).

Fig. 6
figure 6

LC/MS quantification of extracellular nucleosides in conditioned media of WT and Fmr1 KO primary cortical astrocyte cultures. Relative abundance of each secreted factor within each culture genotype is given for WT (A) and Fmr1 KO (B) astrocytes. Direct comparisons between secreted levels of adenosine (C), uridine (D), inosine (E), guanosine (F), and cytidine (G) from WT and Fmr1 KO cultured astrocytes. Concentrations are normalized to both representative cell counts and adenosine-13C5 internal standard signal intensity. Data presented as means ± SEM. WT n = 7, with the exception of WT inosine n = 6; Fmr1 KO n = 6, with the exception of KO adenosine n = 5; *p < 0.05; **p < 0.01; ***p < 0.001

Full size image

Discussion

Recent research has shown that purinergic signalling pathways are elevated in Fmr1 KO cortical astrocytes, which has functional consequences for astrocytic and neuronal excitation (Reynolds et al., 2021, 2024). To further elucidate the extent of purinergic astrocyte dysfunction, it is important to consider the secretion and metabolism of purinergic ligands by these cells. Here, we compared the levels of Fmr1 KO and WT intracellular nucleoside triphosphates and successive metabolites using HILIC-LC/MS, either secreted or metabolized, within the media of WT and Fmr1 KO astrocyte primary cultures. We confirmed that in addition to receptor expression differences reported previously, several nucleoside-based ligands for these receptors were also dysregulated in Fmr1 KO astrocytes. This included significant differences in levels of UDP, ATP, AMP, and adenosine intracellular stores within Fmr1 KO astrocytes, and increased concentrations of extracellular adenosine. In addition, Fmr1 KO astrocytes showed enhanced levels of glycosylation of the membrane-bound CD39 ectonucleotidase, which is responsible for the metabolism of both ATP and UTP. Taken together, the loss of FMRP in astrocytes leads to widespread purinergic dysregulation affecting not only receptor expression but also the production and availability of several purines and pyrimidines.

Increased Intracellular UDP is Consistent with Elevated P2Y Activity in Fmr1 KO Astrocytes

One of the most prominent genotypic differences we observed in this study was a nearly twofold increase in Fmr1 KO astrocyte intracellular UDP relative to WT. This increase in UDP is particularly interesting given recent reports of pyrimidine dysregulation in FXS and ASD. Elevated levels of uridine, a UDP nucleoside precursor, were detected within the plasma of children with ASD (Adams et al., 2011). Circulating uridine has been previously correlated with cerebral UDP levels, as intravenously administered uridine has been shown to pass through the blood–brain barrier to increase the concentration of UDP within the mouse brain (Steculorum et al., 2015). Thus, the elevated plasma uridine in ASD is consistent with our findings of elevated UDP in Fmr1 KO astrocytes. Increases to uridine or UMP via supplementation, dietary or otherwise, also result in elevated levels of UTP and CTP, which in turn raises phosphatide synthesis, synaptic membrane formation, and synaptic protein density (Wurtman et al., 2009). While we did not detect elevated UTP abundance in Fmr1 KO compared to WT astrocytes, the fact that UTP is metabolized and secreted very quickly within these cells makes any changes difficult to detect. Given the enhanced growth of dendritic spines and the hyperactive circuitry reported in the Fmr1 KO cortex, it is possible that enhancement of pyrimidine derivatives, such as UDP or UTP, in Fmr1 KO astrocytes contributes to aberrant synaptogenesis.

The increased presence of astrocyte intracellular UDP also suggests the potential for substantial Fmr1 KO UDP release to stimulate the activity of astrocytic or neuronal P2Y6 receptors. UDP uniquely acts as a strong agonist of metabotropic P2Y6 receptors, with weaker or negligible binding to other purinergic receptor types (Abbracchio et al., 2009). P2Y6 receptors are in fact one of the two P2Y receptor types found to be elevated in Fmr1 KO cortical astrocytes (Reynolds et al., 2021), so the increase in intracellular UDP may be indicative of upregulated P2Y6-mediated signalling. Activation of P2Y6 and other P2Y receptors leads to PLC/IP3-driven endoplasmic reticulum intracellular calcium release, thereby promoting the propagation of intercellular calcium waves, calcium-dependent gliotransmitter release, and transcription factor activation via signal transduction cascades including the MAPK, Akt, and STAT3 pathways (reviewed in Burnstock, 2009; Napier et al., 2023). Thus, the combination of increased receptor expression paired with increased intracellular ligand stores suggests the potential for widespread calcium wave propagation, neuronal firing, and modulation of gene expression, all of which are generally consistent with the FXS phenotype of cortical hyperexcitation.

Significant Alterations of Fmr1 KO Adenosine-Based Purinergic Signalling Molecules

The adenosine-based family of purinergic molecules was altered in Fmr1 KO astrocytes, with significant dysregulations of the intracellular levels of ATP, AMP, and adenosine. Recent work has shown reduced ATP synthesis within the Fmr1 KO cortex (D'Antoni et al., 2020), which is consistent with the reduced ATP levels we observed here. Mitochondria are important sources of ATP production, and our detection methods did not discriminate between intracellular ATP used for cellular metabolism versus gliotransmission. Mitochondrial dysfunction, including decreased mitochondrial fusion and elevated mitochondrial fragmentation in Fmr1 KO neurons (Shen et al., 2019, 2023), as well as increased production of reactive oxygen species (ROS) in Fmr1 KO astrocytes (Vandenberg et al., 2021, 2022), has been recently reported in the Fmr1 KO model. Oxidative stress, arising from high calcium loading and the dysregulation of mitochondrial permeability, manifests elevations of cytoplasmic ROS and a depletion of ATP (reviewed in Brookes et al., 2004). Given the accumulating evidence for oxidative stress and mitochondrial dysfunction in FXS and related disorders, such as fragile X-associated tremor/ataxia syndrome, ATP reductions may not be surprising. Given the innumerable roles ATP plays within the cell, it is likely that both mitochondrial dependent and independent mechanisms contribute to the reduction of ATP in Fmr1 KO astrocytes.

In terms of purine and pyrimidine biosynthesis and metabolism within Fmr1 KO astrocytes, reduced intracellular ATP levels may also be the direct result of enhanced UDP formation, as both uridine kinase and UMP kinase utilize ATP as a substrate to promote UMP and UDP production, respectively. The same is true for elevations in AMP, as adenosine kinase also utilizes ATP to phosphorylate adenosine to AMP (as shown in Fig. 1). Another consideration is nucleoside or nucleotide transport and secretion. While the secretion of ATP from astrocytes could not be assessed directly due to rapid rates of metabolism, elevated levels of adenosine in the conditioned media of Fmr1 KO astrocytes suggest that either adenosine-based nucleotides or nucleosides are released at greater capacities. ATP can be released from astrocytes via vesicular release, or connexin hemi-channel and pannexin channel activation. Under pathological conditions, connexin-43, connexin-30, and pannexin-1 are permeable to ATP and lead to the secretion of nucleotides to either bind purinergic receptors or be efficiently metabolized by membrane-bound enzymes (Giaume et al., 2020). Extracellular adenosine is then cleared and replenished within the cell by equilibrative nucleoside transporters, such as ENT1 and ENT2, that mediate uptake by facilitated diffusion. In addition, nucleosides are recycled into astrocytes via CN2 sodium-dependent concentrative transporters that have a particularly strong affinity for adenosine (Peng et al., 2005). The increased concentration of extracellular adenosine within the Fmr1 KO astrocyte media, in combination with reductions of both intracellular ATP and adenosine levels, suggests that the rate of ATP metabolism outpaces re-uptake of adenosine by Fmr1 KO astrocytes.

Increased Glycosylation of CD39 Ectonucleotidase in Fmr1 KO Astrocytes

Membrane-bound ectonucleotidases control the strength and duration of purinergic agonism by metabolizing secreted gliotransmitters to facilitate their re-uptake by nucleoside transporters. Increased glycosylation of astrocyte membrane-bound CD39 further supports that ectonucleotidase-mediated nucleotide metabolism prior to re-uptake is enhanced in the FXS cortex. In the context of extracellular ATP metabolism, CD39 has been shown to more rapidly metabolize ATP than ADP. Extracellular ATP is almost completely metabolized within 30 min in vitro to produce a 40% accumulation of ADP; however, ADP is not converted to AMP until 80% of the initially released ATP had been metabolized, taking an additional 2.5 h (Wink et al., 2003). Though the dynamics of UTP metabolism were not examined by Wink et al. (2003), ATP and UTP are both metabolized by the ectonucleotidase CD39, suggesting that UTP/UDP metabolism may follow a similar pattern. As receptors such as P2Y1 and P2Y6 have higher affinity for diphosphates than triphosphates (ADP > ATP and UDP > UTP, respectively) (von Kügelgen, 2019), the action of CD39 may therefore increase the likelihood that these receptors are agonised following nucleotide release. While our recent work suggests that aberrant activation of P2Y2 receptors contributes to cellular pathology, such as hyperactivity of astrocytes and neurons, in the Fmr1 KO mouse (Reynolds et al., 2024), further consideration of other receptor dysregulation (ex. P2Y6) will help elucidate the impact of this signalling system in FXS.

Hydrophilic Interaction LC/MS is a Robust Method for Nucleoside Triphosphate Detection

Previous studies have successfully reported relative ATP concentrations using techniques such as ELISA or luciferase assays, but commercially available methodology for the reliable detection of UTP is much more limited. LC/MS approaches for UTP detection have also remained challenging due to significant restrictions in the optimization of the separation of hydrophilic analytes. Here, we used a NH2 column in HILIC mode that permitted polar and hydrophilic molecules to be temporarily retained on a polar stationary phase with a higher percentage of organic solvent and eluted with an aqueous mobile phase gradient (Alpert, 1990; Hemström & Irgum, 2006). HILIC has previously been utilized to detect UTP, ATP, and their metabolites with some success (Narayan, 2017), though detection of UTP was still limited (Galeano Garcia et al., 2019; Kong et al., 2018; Zhang et al., 2014). In contrast, our use of a basic and highly concentrated aqueous mobile phase (100 mM NH4OAc pH 9) ensured the timely elution of ATP and UTP, while nucleosides were readily and reproducibly detected under lower aqueous phase concentrations (10 mM NH4OAc pH 9). In addition, we optimized solid phase extraction from culture media prior to liquid chromatography in order to concentrate extracellular purine and pyrimidine metabolites while removing matrix interferences. In all, this work offers a novel methodology for the examination of astrocyte-mediated purinergic dysregulation and further elucidates the pathological changes to this signalling system in FXS.