SUMMARY
The purinergic transmitter ATP (adenosine 5’-triphosphate) plays an essential role in both the central and peripheral nervous systems, and the ability to directly measure extracellular ATP in real time will increase our understanding of its physiological functions. We developed an ultrasensitive GPCR Activation‒Based ATP sensor called GRABATP1.0, with a robust fluorescence response to extracellular ATP when expressed in several cell types. This sensor has sub-second kinetics, ATP affinity in the range of tens of nanomolar, and can be used to localize ATP release with subcellular resolution. Using this sensor, we monitored ATP release under a variety of in vitro and in vivo conditions, including primary hippocampal neurons, a zebrafish model of injury-induced ATP release, and LPS-induced ATP-release events in individual astrocytes in the mouse cortex measured using in vivo two-photon imaging. Thus, the GRABATP1.0 sensor is a sensitive, versatile tool for monitoring ATP release and dynamics under both physiological and pathophysiological conditions.
INTRODUCTION
Adenosine 5′-triphosphate (ATP) is a universal energy-storing molecule used by virtually all living organisms. In addition to its metabolic function intracellularly, growing evidence suggest that released ATP into the extracellular space can serve as a signaling molecule (termed purinergic transmitter) (Burnstock, 1972), by binding and activating ionotropic P2X receptors and metabotropic P2Y receptors (Abbracchio et al., 2006; Khakh and North, 2012). In the nervous system, a wide range of functions are regulated by ATP, including pain sensation (Burnstock, 1996; Collier et al., 1966), mechanosensory and chemosensory transduction (Burnstock, 2009; Gourine et al., 2005), and synaptic transmission (Burnstock, 2006). Notably, noxious stimuli in the central nervous system (e.g., injury, low osmolality, and inflammation) can trigger a sustained increase in extracellular ATP (Davalos et al., 2005; Wang et al., 2004), which is considered as a multi-target “danger” signal (Rodrigues et al., 2015). Not surprisingly, impaired ATP signaling has been associated with pathological processes (Burnstock, 2007, 2008; Cheffer et al., 2018). Despite the central role that ATP plays in both health and disease, the detailed mechanisms underlying the release and extracellular distribution of ATP are poorly understood, especially in vivo.
A significant number of advances in the last few decades culminated in a variety of techniques and tools for measuring extracellular ATP (Dale, 2021; Wu and Li, 2020). Unfortunately, despite their advantages, these techniques have several key limitations. For example, methods such as microdialysis, electrochemistry-based probes, reporter cells, and bioluminescent assays can measure ATP both in vitro and in vivo (Pellegatti et al., 2008), but are severely limited with respect to precisely detecting ATP due to their relatively low spatial and/or temporal resolution. On the other hand, fluorescent sensor‒based imaging can provide excellent spatiotemporal resolution (Giepmans et al., 2006), and several fluorescent protein‒based sensors have been developed for measuring extracellular ATP, including the recent ecAT3.10 (Conley et al., 2017) and pm-iATPSnFR (Lobas et al., 2019) sensors; however, these sensors are not compatible with measuring extracellular ATP in vivo, mainly due to their limited sensitivity and/or signal-to-noise ratio. A recently developed ATP sensor known as ATPOS (ATP Optical Sensor) has a high affinity for ATP and has been used to image extracellular ATP in the mouse cortex (Kitajima et al., 2020); however, this sensor must be injected as a recombinant protein (Kitajima et al., 2020), which can cause tissue damage and is difficult to measure ATP in a cell type specific manner. In addition to adapting soluble bacterial F0F1-ATP synthase as an ATP-binding protein (e.g., ecAT3.10, pm-iATPSnFR and ATPOS), the natural-evolved extracellular ATP “detectors”—ATP receptors—were also used to engineer ATP sensors. For example, taking advantage of the permeability to Ca2+ ions during ATP-gated P2X channel opening, versatile tools were developed by fusing the genetically-encoded Ca2+ indicators to the C-terminal of P2X subunits (Ollivier et al., 2021; Richler et al., 2008). These sensors display fast kinetics and/or sensitivity allowing to detect ATP release; however, it might be difficult to exclude that in some conditions, especially under in vivo systems, an ATP-P2X-independent activation of GCaMP6s may occur. Overall, the lack of genetically encoded tools that can sense a change in extracellular ATP concentration with high spatiotemporal resolution, high specificity, and high sensitivity has limited our ability to study purinergic signaling under both physiological and pathophysiological conditions.
Recently, our group and others developed a series of genetically encoded GPCR activation‒based (GRAB) sensors to measure a variety of neuromodulators—including acetylcholine (Jing et al., 2020; Jing et al., 2018), dopamine (Patriarchi et al., 2018; Patriarchi et al., 2020; Sun et al., 2018; Sun et al., 2020), norepinephrine (Feng et al., 2019), serotonin (Wan et al., 2020), and adenosine (Peng et al., 2020)—with high sensitivity, selectivity, and spatiotemporal resolution, providing the ability to monitor these neuromodulators in targeted cells under in vivo setting. Here, we report the development and application of a new GFP-based GRABATP sensor using a P2Y receptor as the ATP-binding scaffold. This sensor, which we call GRABATP1.0 (short as ATP1.0), can be expressed in a wide range of cell types, producing a robust fluorescence response (with a ΔF/F0 of 500-1000%), and with high selectivity for both ATP and ADP; moreover, this sensor can be used to detect changes of extracellular ATP both in vitro and in vivo under a variety of conditions.
RESULTS
Development and characterization of a new GRAB sensor for detecting ATP
To develop a genetically encoded GRAB sensor for detecting ATP, we first systematically screened a series of candidate G protein‒couple receptors (GPCRs) known to be activated by ATP, including the human P2Y1, P2Y2, P2Y4, P2Y11, P2Y12, and P2Y13 receptors (Xing et al., 2016). Using these GPCRs as the scaffold, we inserted cpEGFP into the receptor flanked by short linker peptides at both the N- and C-terminus (Figure S1A); we selected the hP2Y1-based chimera ATP0.1 for further optimization based on its good membrane trafficking and high fluorescence response upon application of 100 μM ATP (Figure S1B). We then optimized the length and amino acid composition of the linkers between the hP2Y1 receptor and the cpEGFP moiety (Figure 1A) and identified the candidate with the largest fluorescence response (Figure 1B); we call this sensor GRABATP1.0 (hereafter referred to as ATP1.0). When expressed in HEK293T cells, ATP1.0 trafficked to the plasma membrane (Figure 1C) and produced a peak ΔF/F0 value of 500% in response to 100 μM extracellular ATP (Figure 1B and 1C). As a negative control, we also generated a mutant version of this sensor called ATP1.0mut, which contains the N283A mutation in the receptor’s ATP-binding pocket (Zhang et al., 2015), thus is non-sensitive to ATP (Figures S2 and S3).
We then characterized the specificity, kinetics, brightness and spectrum of the ATP1.0 sensor. With respect to specificity, the ATP-induced response was fully blocked by the P2Y1 receptor antagonist MRS-2500, and no measurable response was produced by any other neurotransmitters or neuromodulators tested, including glutamate, GABA, glycine, dopamine, norepinephrine, serotonin, histamine, and acetylcholine (Figure 1D). ADP and ATP produced a similar response, whereas structurally similar purinergic molecules or derivatives such as AMP, adenosine, UDP, and UDP-glucose virtually produced no response (Figure 1E). ATP1.0 has rapid response kinetics, with a rise time constant (τon) of ~250 milliseconds and a decay time constant (τoff) of ~9 seconds upon application of ATP and subsequent application of MRS-2500, respectively (Figure 1F). With respect to the sensor’s brightness, ATP increased the brightness of ATP1.0 to approximately 64% of the brightness measured in cells expressing an hP2Y1-EGFP fusion protein (Figure S4A). Finally, ATP1.0 shows similar spectrum as EGFP under one-photon excitation, with the excitation peak at ~500nm and emission peak at ~520nm (Figure. 1G).
To compare the performance of ATP1.0 with other extracellular ATP sensors, including single wavelength‒based iATPSnFR sensors (Lobas et al., 2019) and FRET-based ecAT3.10 sensors (Conley et al., 2017), we expressed these sensors in HEK293T cells and performed confocal imaging. Although ATP1.0 and iATPSnFR1.0 were expressed at similar levels at the plasma membrane expression (Figure 1H) and have similar brightness (data not shown), cells expressing ATP1.0 had a 50-fold larger dynamic range to ATP compared to cells expressing iATPSnFR1.0 (Figure 1H-1J). Moreover, compared to cells expressing ATP1.0, cells expressing ecAT3.10 had an extremely small response (Figure 1K) and a significantly smaller signal-to-noise ratio (Figure 1L).
Next, we examined the performance of ATP1.0 in cultured rat primary astrocytes using an adeno-associated virus (AAV) expressing the sensor under the control of the astrocyte-specific GfaABC1D promoter (Lee et al., 2008). We found that ATP1.0 was widely distributed throughout the plasma membrane, including the soma and cell processes (Figures 2A and S5A). Similarly, when expressed in cultured cortical neurons under the control of the neuron-specific hSyn promotor, ATP1.0 was widely distributed throughout the plasma membrane, including the soma and neurites (Figures 2D and S5B). Both astrocytic and neuronally expressed ATP1.0 responded robustly to ATP application, with peak ΔF/F0 values of approximately 1000% and 780%, respectively (Figure 2A-2F). Moreover, the ATP-induced fluorescence response was blocked by the P2Y1 receptor antagonist MRS-2500 (Figure 2G and 2H), and virtually no response was observed in neurons expressing the control ATP1.0mut sensor (Figure S2B). In addition, similar to our results obtained with HEK293T cells, ATP1.0 expressed in neurons responded to both ATP and ADP, but did not respond to AMP, adenosine, UTP, or GTP (Figure 2G-2I). Importantly, the ATP1.0 sensor was stable at the cell surface, as we observed no decrease in fluorescence in ATP1.0-expressing neurons during a 2-hour application of 10 μM ATP (Figure 2J-2L).
Taken together, these results indicate that the ATP1.0 sensor is suitable for the use in several cell types, providing a sensitive, specific, and stable fluorescence increase in response to extracellular ATP.
ATP1.0 can be used to monitor the release of ATP from cultured hippocampal neurons
Next, we examined whether the ATP1.0 sensor could be used to detect the release of endogenous ATP in neuron-glia co-cultures (Figure 3A), a widely used system for studying ATP signaling (Fields, 2011; Koizumi et al., 2003; Zhang et al., 2003). First, we tested whether ATP1.0 could detect stimulus-evoked ATP release. In the brain, ATP is released in response to mechanical stimulation and cell swelling (Newman, 2001; Xia et al., 2012). To induce a mechanical stimulus, we pressed a glass pipette against the cultured cells; when the ATP1.0 signal increased, we then removed the pipette to end the stimulus. We found that mechanical stimulation induced a rapid, localized increase in ΔF/F0, reflecting the release of ATP (Figure 3B). To induce cell swelling, we bathed the cells in a hypotonic solution (130 mOsm/kg); within one minute, a robust increase in ΔF/F0 was observed (Figure 3D). Importantly, the responses induced by both stimuli were abolished by the application of MRS-2500 and were absent in cells expressing the control ATP1.0mut sensor (Figure 3B-3E), confirming the specificity of ATP1.0. We also found that the hypotonic stimulus‒induced release of ATP may not require classical SNARE-dependent vesicular releasing machinery, as expressing tetanus toxin light chain (TeNT), which cleaves synaptobrevin and prevents exocytosis (Patterson et al., 2010; Schiavo et al., 1992), had no effect on the response in cells expressing hSyn-ATP1.0 (Figure 3F1 and 3G1); as a control, expressing TeNT abolished the stimulation-evoked release of glutamate (Glu) release measured using the Glu sensor SF-iGluSnFR.A184V (Figure 3F2 and 3G2).
In addition to stimulus-evoked ATP release, we also observed spontaneous, localized, transient ATP1.0 signals in our neuron-glia co-cultures even in the absence of external stimulation (Figure 3H and 3I). In the 1.6-mm2 imaging field, these events occurred at a rate of 1.2/min and had an average peak ΔF/F0 of approximately 210% (Figure 3K). The average rise time (τon) and decay time (τoff) of spontaneous ATP-releasing events were 11 s and 43 s (Fig. 3L), respectively. The average diameter of spontaneous ATP-releasing events was 32 μm based on our analysis of full width at half maximum (FWHM) (Fig. 3M). In contrast, no spontaneous events were observed in the presence of MRS-2500 or in cells expressing ATP1.0mut (Figure 3H, 3I and 3K). To confirm that the ATP1.0 signal reflects extracellular ATP dynamics, we imaged cells in the presence of the ATP degrading enzyme apyrase. We observed that apyrase (30 U/ml) treatment significantly blocked spontaneous events (Figure 3H, 3I and 3K).
ATP1.0 can be used to measure the injury-induced in vivo propagation of ATP in zebrafish larvae
Having shown that the ATP1.0 sensor is suitable for use in in vitro systems, we then examined whether it could be applied to monitor ATP in in vivo systems such as zebrafish. We therefore transiently expressed either ATP1.0 or ATP1.0mut in neurons of larval zebrafish under the control of the neuron-specific elval3 promoter (Figure 4A and 4B). Local puffing of ATP, but not saline, elicited a robust transient increase in ΔF/F0 in the optic tectum. These signals were blocked by MRS-2500 and not observed in zebrafish larvae expressing the control ATP1.0mut sensor (Figure 4C).
Next, we examined whether ATP1.0 could be used to measure the release of endogenous ATP in live zebrafish. It is known that ATP signaling plays key roles in promoting the migration of microglia to injury site (Li et al., 2012; Sieger et al., 2012). We found that injury induced by laser ablation in the optic tectum caused a robust increase in fluorescence in ATP1.0-expressing zebrafish (Figure 4D and 4E). Moreover, the response propagated in a radial pattern outward from the site of injury (Figure 4E, 4H, and 4I). Next, we simultaneously monitored ATP release and the migration of microglia by expressing ATP1.0 in the optic tectum of a transgenic zebrafish line in which the microglia are labeled with the red fluorescent protein DsRed (Figure 4F). We found that following laser ablation, microglia gradually migrated to the site of injury along the path of ATP propagation measured using ATP1.0 (Figure 4G and 4J). Thus, our ATP1.0 sensor is well-suited for in vivo application in zebrafish larvae, providing high spatiotemporal resolution.
ATP1.0 can be used to monitor localized ATP release during LPS-induced systemic inflammation in mice
Purinergic signaling molecules, including ATP, are considered critical extracellular messengers in response to acute and chronic inflammation, acting via paracrine or autocrine processes on immune cells in the peripheral nervous system and on neurons and glia cells in the central nervous system (Idzko et al., 2014). To date, however, the pattern by which ATP is released during systemic inflammation, as well as the relationship between this release and inflammatory status, are poorly understood. We therefore used a mouse model of systemic inflammation induced by an intraperitoneal injection of bacterial lipopolysaccharides (LPS; 10 mg/kg), and directly observed ATP dynamics in the visual cortex using two-photon imaging of ATP1.0 fluorescence (Figure 5A); this inflammation model caused a robust increase in expression of the inflammatory cytokines IL-1β and IL-10 in the brain (Figure S6). Twenty-four hours after LPS injection, we observed multiple localized ATP-release events in the cortex, with a frequency of approximately 5-10 events/min measured during 20 minutes of recording (Figure 5B2 and 5D). In contrast, fewer events occurred prior to LPS injection (data not shown), in saline-injected controls (Figure 5B1, 5C and 5E), and no events were observed in LPS-injected mice expressing the mutant ATP1.0mut sensor (Figure 5B3 and 5C).
Next, we used the Astrocyte Quantitative Analysis (AQuA) software (Wang et al., 2019) to characterize the individual events. The ATP-release events had broadly distributed signal kinetics, although the majority of events have a relatively fast rise time (<5 s) and a slower decay time (10-20 s) (Figure 5F and 5G). In addition, the events had a spatially selective pattern, with an average signal diameter (determined using the maximum diameter of each event) of approximately 9.9 μm (Figure 5H), smaller than the average diameter of a typical astrocyte (10-20 μm) (Chai et al., 2017). To detailly examine the correlation between the ATP-release events and the progression of inflammation, we recorded cortical ATP events at various time points after LPS injection. We found an increase in ATP-release events within 30 min of LPS injection, and the number of events increased progressively with time, reaching a plateau 6 hours after injection; in contrast, no events were detected in saline-injected mice at any time point up to 24 hours (Figure 5I). Interestingly, an analysis of the location of the ATP-release events within the cortex revealed that the early events occurred relatively close to the blood vessels, and the distance between the events and the nearest vessels increased with time (Figure 5J). These data suggest that the brain can sense inflammation and respond in the form of spatially selective ATP-release events, demonstrating that the ATP1.0 sensor is compatible with in vivo imaging in mice, with unprecedented sensitivity and spatiotemporal resolution.
DISCUSSION
Here, we report the development and characterization of a new, ultrasensitive, genetically encoded ATP sensor called GRABATP1.0. We also show that this sensor can be expressed reliably in a variety of cell types, including cell lines, astrocytes, and neurons, providing a robust tool for measuring extracellular ATP. Moreover, we show that this sensor can be used to visualize the real-time release of endogenous ATP in vitro, as well as ATP signaling in two in vivo models under several conditions.
Our GRABATP sensors have at least four distinct advantages over other sensors with respect to monitoring the dynamics of extracellular ATP. First, ATP1.0 has extremely high sensitivity for extracellular ATP compared to other ATP sensors such as the recently developed, genetically-encoded single-wavelength ATP sensor, iATPSnFR1.0. When expressed in HEK293T cells, GRABATP1.0 displayed an EC50 of ~6.7 μM with a maximum ΔF/F0 of ~500% (Figure. 1J). Under the same condition, iATPSnFR1.0 displayed excellent plasma membrane localization (Figure 1H), yielding an EC50 ~381 μM (Figure 1G), which was consistent with the published data (Lobas et al., 2019). However, the maximum ΔF/F0 of iATPSnFR1.0 is ~10%, ~10-fold lower than the reported data (Lobas et al., 2019), presumably because of different imaging conditions. Curiously, we found that the GRABATP1.0 exhibits apparently different affinities to ATP in HEK293T cells (apparent EC50 ~6.7 μM) vs. neurons (apparent EC50 ~45 nM). One reason we speculated is due to the existence of enzymes that degraded the ATP in cultured HEK293T cells, which reduced the apparent affinities. Given the high sensitivity of GRABATP1.0 sensors, particularly when expressed in neurons and astrocytes, ATP1.0 will be useful for studying both pathological and physiological processes. Second, the ATP1.0 sensor is genetically encoded and can be expressed selectively in a variety of cell types, providing cell type‒specific measurements of ATP transmission. Third, ATP1.0 has high spatial resolution, suitable for measuring highly localized, transient ATP-release events in hippocampal cultures and in the mouse cortex. Lastly, our results demonstrated ATP1.0 can be used to monitor ATP dynamics in vivo using a variety of animal models, including zebrafish and mice.
Despite these advantages of genetically encoded GRABATP sensors, a potential caveat is that ATP1.0 is based on the scaffold P2Y1 receptor (Waldo et al., 2002) and therefore responds to both ATP and ADP. Given that ATP and ADP may regulate distinct processes, particularly in the peripheral nervous system (Gaarder et al., 1961), next-generation GRABATP sensors should be developed with improved molecular specificity, for example by engineering the GPCR scaffold to increase the sensor’s selectivity for ATP over ADP, and vice versa. Alternatively, other P2Y receptors, such as P2Y11 and P2Y12, which are more specific for ATP (Communi et al., 1997) and ADP (Hollopeter et al., 2001), respectively, can be used as scaffolds in developing future ATP or ADP sensors (Fig. S1B).
In hippocampal cultures, the ATP1.0 sensor readily resolved both evoked and spontaneous ATP release. Moreover, our study revealed that the hypotonicity‒induced ATP release was not sensitive to TeNT, supporting a non‒ vesicular mechanism of ATP release (Lazarowski, 2012). Interestingly, several molecules are proposed to mediate stimulus-induced ATP release (Taruno, 2018), including calcium homeostasis modulator (CALHM) (Taruno et al., 2013), pannexin/connexin, P2X7 receptors (Pellegatti et al., 2005), Leucine Rich Repeat Containing 8 VRAC Subunit A (LRRC8A)/SWELL1 (Qiu et al., 2014; Voss et al., 2014) and SLCO2A1 (Sabirov et al., 2017). We anticipate the new developed ATP1.0 sensor will provide a good tool to further dissect the relative contributions of these channels on ATP release under different stimulation conditions.
By combining the ATP1.0 sensor with in vivo two-photon imaging, we detected highly localized ATP-release events in the mouse brain following a systemic injection of LPS, and we found that these events were smaller in size than the diameter of a single astrocyte (Chai et al., 2017), indicating that the brain can sense systemic inflammation and respond with ATP signaling at cellular level. Further combining ATP1.0 imaging with genetic and pharmacological tools may facilitate the identification of cell types and molecules required for ATP signaling during these processes. A growing body of experimental evidence suggests that neuroinflammation is a key pathological event triggering and perpetuating the neurodegenerative processes associated with many neurological diseases, including Alzheimer's disease, Parkinson's disease, and amyotrophic lateral sclerosis (Amor et al., 2014; Nguyen et al., 2002). Thus, our GRABATP sensor can be a powerful tool for studying dynamic changes in ATP release and the role of these changes in the neuroinflammatory processes that underlie neurodegeneration.
ATP plays an important role in neuron-glia interactions, which has complex interaction with other signaling such as calcium or glutamate. For example, the release of ATP can trigger calcium waves in astrocytes and affect neuronal glutamate release (Bazargani and Attwell, 2016; Fields and Burnstock, 2006; Guthrie et al., 1999; Illes et al., 2019; Zhang et al., 2003). Thus, the ATP1.0 sensor can be combined with a spectrally compatible calcium indicator, glutamate sensor, and/or other fluorescent indicators, providing an orthogonal readout of ATP with extremely high spatial and temporal resolution, yielding new insights into the role of ATP signaling under both physiological and pathophysiological processes.
AUTHOR CONTRIBUTIONS
Y.L. supervised the project. Z.W. and K.H. performed the experiments related to the development, optimization, and characterization of the sensors in cultured cells, with contributions from S.P., B.L., and H.W. Z.W. performed the imaging of ATP release in cultured cells. H.L. and T.L. performed the in vivo zebrafish experiments under the supervision of J.D. Y.C. and M.J. performed the in vivo two-photon imaging experiments in mice. All authors contributed to the interpretation and analysis of the data. Z.W. and Y.L. wrote the manuscript with input from all authors.
DECLARATION OF INTEREST
H.W., M.J., and Y.L. have filed patent applications, the value of which may be affected by this publication.
SUPPLEMENTAL INFORMATION
MATERIALS AND METHODS
Molecular biology
Plasmids were generated using Gibson assembly. DNA fragments were generated using PCR amplification with primers (Tsingke) with ~25-bp overlap, and all sequences were verified using Sanger sequencing. All cDNAs encoding the candidate GRABATP sensors were cloned into the pDisplay vector (Invitrogen) with an upstream IgK leader sequence and a downstream IRES-mCherry-CAAX cassette (to label the cell membrane). The cDNAs encoding the ATP receptor subtypes were amplified from the human GPCR cDNA library (hORFeome database 8.1), and the third intracellular loop (ICL3) of each ATP receptor was swapped with the corresponding ICL3 in the GRABNE sensor. The swapping sites in the P2Y1 receptor and the amino acid composition between the P2Y1 receptor and the ICL3 of GRABNE were then screened. The plasmids used to express the GRABATP sensors in mammalian neurons and astrocytes were cloned into the pAAV vector under the control of human synapsin promoter (hSyn) or the GfaABC1D promoter, respectively. The plasmids used to express the GRABATP sensors in zebrafish were cloned into Elval3: Tetoff vectors.
The pm-iATPSnFR1.0 sensor was a gift from Baljit Khakh (Addgene plasmid #102548). The ecAT3.10 sensor was a gift from Mathew Tantama (Addgene plasmid #107215). The SF-iGluSnFR.A184V sensor was a gift from Loren Looger (Addgene plasmid #106175). The Rncp-iGluSnFR sensor was a gift from Robert Campbell (Addgene plasmid #107336) and was subcloned into the pAAV-hSyn vector. Finally, the plasmid encoding TeNT was a gift from Dr. Peng Cao and was subcloned into the pAAV-CAG vector.
Cell cultures, zebrafish, and mice
HEK293T cells and primary neuron-glia co-cultures were prepared and cultured as described previously (Peng et al., 2020). In brief, HEK293T cells were cultured at 37°C in 5% CO2 in DMEM (Biological Industries) supplemented with 10% (v/v) fetal bovine serum (FBS, Gibco) and 1% penicillin-streptomycin (Biological Industries). Rat primary neuron-glia co-cultures were prepared from 0-day old (P0) rat pups (male and female, randomly selected) purchased from Charles River Laboratories (Beijing, China). Cortical or hippocampal cells were dissociated from the dissected brains in 0.25% Trypsin-EDTA (Gibco) and plated on 12-mm glass coverslips coated with poly-D-lysine (Sigma-Aldrich) in neurobasal medium (Gibco) containing 2% B-27 supplement (Gibco), 1% GlutaMAX (Gibco), and 1% penicillin-streptomycin (Gibco). Based on glial cell density, after approximately 4 days in culture (DIV4) cytosine β-D-arabinofuranoside (Sigma) was added to the hippocampal cultures in a 50% growth media exchange, with a final concentration of 2 μM.
Rat primary astrocytes were prepared as previously described (Schildge et al., 2013). In brief, the cortex or hippocampi were dissected from P0 rat pups, and the cells were dissociated using trypsin digestion for 10 mins at 37°C and plated on a poly-D-lysine‒coated T25 flask. The plating and culture media contained DMEM supplemented with 10% (v/v) FBS and 1% penicillin-streptomycin. The next day, and every 2 days thereafter, the medium was changed. At DIV 7-8, the flask was shaken on an orbital shaker at 180 rpm for 30 min, and the supernatant containing the microglia was discarded; 10 ml of fresh astrocyte culture medium was then added to the flask, which was shaken at 240 rpm for ≥6 h to remove oligodendrocyte precursor cells. The remaining astrocytes were dissociated with trypsin and plated on 12-mm glass coverslips in 24-well plates containing culture medium. Both the neurons and astrocytes were cultured at 37°C in 5% CO2.
For zebrafish experiments, zebrafish larvae at 4-6 days post-fertilization (4-6 dpf) were used for all experiments in this study. As the sex of zebrafish cannot be determined in the larval stage, sex discrimination was not a factor in our study. Wild-type (AB background) and Tg(coro1a: DsRed) zebrafish strains were used in this study. Adult zebrafish and larvae were maintained and raised under standard laboratory protocols (Yu et al., 2010), and all procedures were approved by the Institute of Neuroscience, Chinese Academy of Sciences.
Wild-type C57BL/6J mice were housed under a 12-h/12-h light/dark cycle. All protocols for animal surgery and maintenance were approved by the Animal Care and Use Committees at Peking University and the Chinese Institute for Brain Research, and were performed in accordance with the guidelines established by the US National Institutes of Health. Adult mice (>6 weeks of age) were used for the in vivo experiments.
AAV virus preparation
The following AAV viruses were used to infect cultured cells and for in vivo expression (all packaged at Vigene Biosciences): AAV2/9-hSyn-ATP1.0, AAV2/9-GfaABC1D-ATP1.0, AAV2/9-hSyn-ATP1.0mut, AAV2/9-GfaABC1D-ATP1.0mut, AAV2/9-CAG-EBFP2-iP2A-TeNT, AAV2/9-SF-iGluSnFR.A184V, and AAV2/9-hSyn-Rncp-iGluSnFR.
Expression of GRABATP in cultured cells and in vivo
For screening, HEK293T cells expressing the candidate GRABATP sensors were plated in 96-well plates (PerkinElmer). For confocal imaging, HEK293T cells were plated on 12-mm glass coverslips in 24-well plates and grown to 60-80% confluence for transfection. Cells were transfected using a mixture containing 1 μg DNA and 1 μg PEI for 4-6 h and imaged 24-48 hours after transfection. For diffuse in vitro expression, the viruses were added to neuron-glia co-cultures or cultured astrocytes at DIV 5-9, and the cells were characterized ≥48 hours after infection; DIV ≥13 cells were used for physiological analyses.
For in vivo expression in zebrafish, plasmids encoding either ATP1.0 or ATP1.0mut were co-injected (25 ng/μl) with Tol2 transposase mRNA (25 ng/μl) into one-cell stage wild-type (AB background) or Tg(coro1a: DsRed) embryos.
To express GRABATP in mice in vivo, the mice were anesthetized with an i.p. injection of Avertin (500 mg/kg, Sigma); the skin was retracted from the head, and a metal recording chamber was affixed. After the mice recovered for 1-2 days, the mice were re-anesthetized, the cranial window on the visual cortex was opened, and 400-500 nl of AAV was injected using a microsyringe pump (Nanoliter 2000 injector, WPI) at the following coordinates: AP: −2.2 mm relative to Bregma, ML: 2.0 mm relative to Bregma, and DV: 0.5 mm below the dura at an angle of 30°. A 4 mm × 4 mm square coverslip was used to replace the skull after AAV injection, and in vivo two-photon imaging was performed 3 weeks after injection.
Confocal imaging of cultured cells
Before imaging, the culture medium was replaced with Tyrode’s solution contained (in mM): 150 NaCl, 4 KCl, 2 MgCl2, 2 CaCl2, 10 HEPES, and 10 glucose (pH 7.3-7.4). For inducing cell swelling, the hypotonic Tyrode’s solution (osmolality: 130 mOsm/kg) contained (in mM): 50 NaCl, 75 KCl, 2 MgCl2, 2 CaCl2, 10 HEPES, and 10 glucose (pH 7.3-7.4). HEK293T cells grown in 96-well plates were imaged using an Opera Phenix high-content screening system (PerkinElmer) equipped with a 20x/0.4 NA objective, a 40x/0.6 NA objective, a 40x/1.15 NA water-immersion objective, a 488-nm laser, and a 561-nm laser; green fluorescence (GRABATP sensors and P2Y1R-EGFP) and red fluorescence (mCherry-CAAX) were recorded using a 525/50-nm and 600/30-nm emission filter, respectively. Cells grown on 12-mm coverslips were imaged using a Ti-E A1 confocal microscope (Nikon) equipped with a 10x/0.45 NA objective, a 20x/0.75 NA objective, a 40x/1.35 NA oil-immersion objective, a 405-nm laser, a 488-nm laser, and a 561-nm laser; blue fluorescence (BFP2-TeNT), green fluorescence (GRABATP sensors, iATPSnFR1.0, P2Y1R-EGFP, and SF-iGluSnFR.A184V), and red fluorescence (mCherry-CAAX and Rncp-iGlu) were recorded using a 450/25-nm, 525/50-nm, and 595/50-nm emission filter, respectively.
The following compounds were applied by replacing the Tyrode’s solution (for imaging cells in 96-well plates) or by either bath application or a custom-made perfusion system (for imaging cells cultured on 12-mm coverslips): ATP (Sigma), ADP (Sigma), AMP (Sigma), adenosine (Ado, Sigma), UDP (Sigma), UTP (Sigma), GTP (Sigma), UDP-glucose (Tocris), MRS-2500 (Tocris), Glu (Sigma), GABA (Tocris), Gly (Sigma), DA (Sigma), NE (Tocris), 5-HT (Tocris), HA (Tocris), ACh (Solarbio), and apyrase (Sigma, 15 U/ml apyrase grade VI plus 15 U/ml apyrase grade VII). Between experiments, the recording chamber was cleaned thoroughly using Tyrode’s solution and 75% ethanol. The micropressure application of drugs was controlled using a Pneumatic PicoPump PV800 (World Precision Instruments). Hypotonic solutions were delivered by perfusion. For mechanical stimulation, a glass pipette was placed above the cultured cells. For field stimulation of cultured neurons, parallel platinum electrodes positioned 1 cm apart were controlled using a Grass S88 stimulator (Grass Instruments), and 1-ms pulses were applied at 80 V. Except where indicated otherwise, all experiments were performed at room temperature.
In vivo confocal imaging of GRABATP in larval zebrafish
GRABATP responses induced by local puffing of drugs was performed in zebrafish larvae expressing either Elval3: Tetoff-ATP1.0 or Elval3: Tetoff-ATP1.0mut. GRABATP responses and microglia movement following laser ablation‒ induced injury were performed by using Tg(coro1a:DsRed) zebrafish larvae expressing Elval3: Tetoff-ATP1.0.
In vivo confocal imaging experiments were performed using an FN1 confocal microscope (Nikon) equipped with a 40x (NA 0.8) or 25 x (NA 1.1) water-immersion objective. Before imaging, the larvae were immobilized in 1.2% low-melting point agarose. Time-series imaging was carried out at 28°C using a heating system. A 488-nm or 561-nm excitation laser and a 525/50-nm or 595/50-nm emission filter were used to excite and collect the GFP and DsRed signals, respectively.
To monitor the GRABATP responses to locally puffed drugs, the larvae were paralyzed with 1 mg/ml α-bungarotoxin (Tocris), the agorae around the tectum region were removed, and a small incision in the skin around the top tectum was made for introducing the micropipette. The larvae were incubated in external solution (ES) either with or without 90 μM MRS-2500 (Tocris). Local puff application of ES with or without 5 mM ATP (Sigma) was performed using a micropipette with a tip diameter of 1-2 μm introduced via the contralateral optic tectum to the target tectum region. For each zebrafish larva, the solution contained in the micropipette was puffed using 2 pulses of gas pressure (3 psi, 50-ms duration, 1-s interval), with 5 local puffing sessions in total applied at a 2-min interval. Single optical section confocal imaging was performed with an interval of 2.2 s.
To monitor the GRABATP responses and microglial dynamics following laser ablation in larval zebrafish, the larvae were paralyzed and imaged as described above. Time-series images were captured before and immediately after laser ablation. For laser ablation, target regions (5 μm in diameter) were illuminated at 800 nm for 7 s using a two-photon laser.
Two-photon in vivo imaging in mice
Two-photon imaging was performed using a FluoView FVMPE-RS microscope (Olympus) equipped a laser (Spectra-Physics). For experiments involving lipopolysaccharides (LPS), 10 mg/kg LPS from Escherichia coli O111:B4 (Sigma, L4130) was dissolved in sterile saline and injected intraperitoneally (i.p.) into the mice. ATP1.0 was imaged using a 920-nm laser, and the imaging frequency was set at 32 Hz with 512×512-pixel resolution.
Data analysis
Imaging data obtained from cultured cells and zebrafish were first processed using ImageJ software (NIH) or MATLAB 2018 (MathWorks); traces were generated using OriginPro 2019 or MATLAB, and pseudocolor images were generated using ImageJ. For the mouse 2-photon microscopy images, 10 images were first averaged and processed using AQuA software in MATLAB, and the detail information regarding individual ATP-release events were plotted using either MATLAB with custom-written programs or OriginPro 2019.
Except where indicated otherwise, all summary data are presented as the mean ± SEM. Groups were analyzed using either the Student’s t-test or a one-way ANOVA.
Data and software availability
The plasmids for expressing ATP1.0 and ATP1.0mut used in this study have been deposited at Addgene (https://www.addgene.org/Yulong_Li/).
ACKNOWLEDGMENTS
We thank the members of the Li lab for helpful suggestions and comments. This research was supported by the Beijing Municipal Science & Technology Commission (Z181100001318002 and Z181100001518004 to Y.L., and the Beijing Nova Program Z20111000680000 to M.J.); grants from the NSFC (81821092 and 92032000 to Y.L.); grants from National Key R&D Program of China (2020YFE0204000 to Y.L. and 2019YFA0801603 to J.D.); and grants from the Peking-Tsinghua Center for Life Sciences and the State Key Laboratory of Membrane Biology at Peking University School of Life Sciences (to Y.L.). Z.W. is supported by the Boehringer Ingelheim-Peking University Postdoctoral Program. We thank Xiaoguang Lei at PKU-CLS and the National Center for Protein Sciences at Peking University for their support and assistance using the Opera Phenix high-content screening system.