Dynamic and rapid detection of guanosine during ischemia

Abstract

Guanosine acts in both neuroprotective and neuro-signaling pathways in the central nervous system; in this paper we present the first fast voltammetric measurements of endogenous guanosine release during pre- and post-ischemic conditions. We discuss the metric of our measurements via analysis of event concentration, duration, and interevent time of rapid guanosine release. We observe changes across all three metrics from our normoxic to ischemic conditions. Pharmacological studies were performed to confirm that guanosine release is a calcium-dependent process and that the signaling observed is purinergic. Finally, we show the validity of our ischemic model via staining and florescent imaging. Overall, this paper sets the tone for rapid monitoring of guanosine and provides a platform to investigate the extent to which guanosine accumulates at the site of brain injury, i.e. ischemia.

Graphical Abstract

Introduction

Ischemic stroke accounts for over eighty percent of strokes and is a leading and rapidly rising cause of death globally. Insufficient oxygen and glucose supply to sensitive brain tissues quickly upsets a delicate metabolic balance, evoking membrane depolarization, excitotoxicity, inflammation, and widespread cell injury and death. Even transient ischemic attacks can have permanent, debilitating impacts on memory, motor control, and self-sufficiency^1,2. Therefore, understanding the neurochemistry – both injurious and protective – occurring at the point and time of ischemic assault is essential to creating robust and effective therapeutic intervention strategies. Here, we measure subsecond fluctuations in endogenous guanosine release prior to and after ischemic assault for the first time, providing a significant advance in our understanding of the neurobiological role of purines during neuroprotection.

Purines have emerged as a leading class of neuromodulatory signaling molecules involved in a host of protective and regenerative functions both during and following ischemic attack^3-6. Both ATP and GTP are well-known for their roles in neurotransmission and intercellular communication^6-9, but their nucleoside counterparts have emerged as increasingly important regulatory molecules active at the point of assault. Adenosine’s role, both as a rapid and chronic protective agent, has been well characterized in recent years^10-13. However, its sister molecule guanosine remains far less understood but increasingly appears to play a central role in neuroprotection during ischemic attack.

Guanosine’s role as a neuroprotector extends from the molecular to cellular to system level. Following ischemic injury, extracellular guanosine levels rise significantly and remain elevated for up to a week after the insult occurred¹⁴. High endogenous guanosine is associated with improved survival from ischemic stroke, and is correlated with higher recovery rates for memory, cognition, and motor skills^15-19. Elevated extracellular guanosine levels are also associated with neuronal rescue and survival post-stroke^17,20-22. Similarly, the amount of infarction is substantially decreased following guanosine administration administered during reperfusion injury²². Guanosine administration also lowers microglial activation and decreases inflammatory cytokine levels after injury^19,23-28. On a molecular level, guanosine is intimately entwined with several essential neurochemical responses to injury. Guanosine administration lowers accumulation of reactive oxygen species and decreases oxidative stress in a variety of pathological models^19,29,30. It facilitates increased glutamate clearance from the extracellular space, an essential role for minimizing the effects of glutamatergic excitotoxicity during ischemia, and can decrease membrane permeability through antagonistic binding at NMDA receptors^16,31-35. Guanosine is also involved in the broader purinergic response to ischemia. Extracellular adenosine levels rise rapidly following guanosine administration, and guanosine will interact at adenosine receptors involved in cell recovery pathways^30,36,37. While guanosine’s importance is clear, no study has examined guanosine’s endogenous response at the point and time of ischemic assault. Although guanosine is known to be a potent neuromodulator, its ability to respond rapidly and dynamically has gone uninvestigated. With severe cell damage initiating within minutes of ischemic assault, a robust understanding of guanosine’s immediate neuroprotective action is essential to developing effective therapeutic strategies. Our lab previously developed a method for spatially resolved detection of low nanomolar levels of guanosine released on a sub-second timescale³⁸. Here, we apply fast scan cyclic voltammetry (FSCV) to measure dynamic guanosine signaling for the first time in both healthy and ischemic hippocampal tissue. We establish that guanosine is released on a subsecond timescale and is immediately upregulated in quantity, duration, and frequency during injury. This work provides essential insight into the timescale and mode of purinergic, and more specifically guanosine, neuroprotection during ischemic assault.

Results and Discussion

Rapid guanosine signaling in healthy tissue

Guanosine’s significance as a potent neuromodulator during and after ischemic damage is well established^19-21,27,39. Tonic guanosine levels rise sharply in the hours following injury and remain elevated for up to seven days post damage^19,26,39. Sustained high purine levels contribute to tissue recovery., improve mobility and memory, and decrease neurological deficit even in severe ischemic models^{6,10,11,20,39}. What remains unclear is the dynamic response of guanosine to initial aggravation, and how it changes compared to modulation patterns in healthy tissue.

Here, we report the first subsecond measurements of spontaneous guanosine release in both healthy and ischemic brain tissue. Guanosine was monitored with a carbon-fiber microelectrode in the CA1 region of the rat hippocampus in an ex vivo slice model. A typical FSCV measurement of transient guanosine events is shown in Fig 1A. Event concentration, duration, and interevent time is determined for each discrete transient. The cyclic voltammogram (Fig. 1B), a vertical slice of the color plot representative of a single scan, is used to selectively identify guanosine in a complex environment (Fig. S2, S3). At the applied waveform, guanosine oxidizes at +1.3 V, just past the switching potential, and has a much smaller secondary oxidation at +0.8 V on the forward scan³⁸. Guanosine detection is sensitive over other relevant ischemia-activated neurochemicals at this waveform; interference from adenosine, histamine, H2O2, and other guanine-based purines is minimal (Fig. S2, S3). Additionally, due to background subtraction, FSCV is not capable of measuring basal level neurochemicals or neurochemicals that are not rapidly released, which eliminates many potential interferences including downstream metabolites. Potentials exceeding 1.4 V are necessary to oxidize adenine-based purines. Our prior work characterizing guanosine detection with FSCV fully investigated and established the selectivity of guanosine detection at this waveform.³⁸ Dynamic guanosine signaling was monitored in slices for 50 minutes after electrode implantation, and data recorded for 45 minutes, allowing a 5-minute acclimation. Each slice was divided into three 15-minute sections, with the first 15 minutes as the control and the last 15 minutes as treatment (Fig. S1). Ischemia or pharmaceutical treatments were administered for the last 30 minutes of measurement, allowing a 15-minute equilibration period before comparison of control and treatment groups. To determine baseline signaling in a healthy, normoxic environment, we perfused only oxygenated artificial cerebrospinal fluid (aCSF) for the entire 45 minute recording. The last 15 minutes of data collection is referred to as “sham,” similar to vehicle controls (i.e. saline injections) in in vivo models.

Figure 1.

Guanosine is released on a subsecond timescale in healthy hippocampal tissue. Transient release events are observable at 2 s, 4s, 7s, 16s, 17s, and 26s on the false color plot (A). Event concentration, duration, and interevent time are determined from the current vs. time trace seen above the color plot. Guanosine is identified through its cyclic voltammogram (B), which has a primary peak at 1.3 V and a secondary peak at 0.8 V on the applied waveform.

In a normoxic environment, base guanosine signaling remains low unless aggravated by an assault. No change was observed in transient concentration over 45 minutes (Fig. 2A, C, mean 168.1 ± 19.3 nM control vs. 146.8 ± 19.0 nM sham, Mann-Whitney test, p = 0.8212, n = 18 slices). Concentration distributions were comparable for both control and sham groups, with few events larger than 500 nM observed (Fig. 2B). The calculated concentration ranges were similar to those reported for baseline adenosine signaling in normoxic conditions^12,40.

Figure 2.

Rapid guanosine signaling is slow in healthy tissue. Average event concentrations are stable over 45 minutes (A-C). Guanosine transients are longer during later measurements (D-F). Signaling slows significantly following initial implantation and time between events increases (G-I). Similarly, fewer transients are released over time (J-L) (n = 18 slices).

Event duration, however, increases in the sham compared to average transient length immediately following implantation (Fig. 2D, F, mean 0.91 ± 0.04 s control vs. 1.18 ± 0.11 s, Mann-Whitney test, p < 0.0001, n = 18 slices). Eighty-three percent of events in the control period lasted one second or less, while only fifty-nine percent of events in the sham period were less than a second long (Fig. 2E). Notably, we do not see a correlation between event concentration and duration (Fig 5). This suggests a shift not in regulation of guanosine’s release, but rather its reuptake. The wider distribution of event times, in conjunction with the changes observed in signaling frequency, indicates a shift in guanosine’s homeostatic signaling baseline seemingly relative to electrode implantation.

Figure 5.

No significant correlations between event concentration, duration, and interevent time is observed in healthy groups (A, B). In ischemia, concentration and interevent time are mildly negatively correlated, while a moderate positive correlation between event length and time until the next is observed (C).

As mentioned previously, guanosine release is far more frequent in the first several minutes following electrode implantation than it is towards the end of data collection (Fig. 2G, mean 43.5 ± 5.2 s control vs. 176.6 ± 28.4 s sham, Mann-Whitney test, p < 0.0001, n = 18 slices). Quiet periods between transients increase steadily with time, averaging just 25.4 s between guanosine events for the first three minutes of the control. By the fifteen minute mark, interevent time (IET) quintupled to 120.3 s, and for the last three minutes of the sham period, average time between signals reached 387.1 s (Fig. 2I). Eighty-two percent of events in the control period had less than one minute between them, while fifty percent of events in the sham period were separated by under a minute (Fig. 2H). Of note is that the median IETs for both control and sham groups are much smaller than the means (16.7 s control, 46.9 s sham), which is indicative of multiple events clustered tightly together followed by long periods of silence.

Concomitant with interevent frequency is the counted number of rapid guanosine events, which decreases over time (Fig. 2L). The observed number of events is inversely proportional to IET. Significantly more guanosine is released in the control period compared to the sham (Fig. 2J, mean 20 ± 4.3 control vs. 4.6 ± 1.8 sham, Wilcoxon test, p < 0.0001, n = 18 slices). More than half (55%) of slices had five events or fewer in the sham period (Fig. 2K). Taken together with the steady increase in time between release and the duration of guanosine transients, this paints a picture of slow and steady release in the absence of assault on the tissue.

Of more interest is the rapid uptick in guanosine release for the several minutes following electrode implantation. FSCV traditionally operates on the assumption of minimal tissue damage after insertion of the electrode due to the thinness of the carbon fiber sensing surface^41-43. Compared to other invasive techniques for measuring signaling neurochemical levels, FSCV indeed minimizes damage to surrounding tissue and maintains baseline signaling patterns with little perturbation⁴⁴. These previous studies, however, have primarily examined cell survival and neurotransmitter release pathways. Guanosine, like adenosine, is a neuromodulator that increases in response to injury and downregulates expression of proinflammatory markers that lead to increased long-term damage^6,8,23,27. The immediate and sustained increase in the number and frequency of dynamic guanosine events following implantation, followed by a return to slower signaling, suggests that our microelectrodes are aggravating the tissue more than originally thought. Interestingly, the concentration of the events does not change, while the time between events drops sharply compared to the sham state. This implies that rather than an increase in the amount of guanosine prepared for export intracellularly, the purine release is instead driven stimulated release. As glutamate is a known upregulator of guanosine release³¹, causes an influx of cations into cells, increasing excitability⁴⁵, and is ubiquitous in the CA1 region of the hippocampus where these experiments were conducted⁴⁶, we suggest that guanosine’s increase in frequency of release is caused by increased glutamate release following aggravation of neurons from electrode implantation. Guanosine is implied as a noncompetitive inhibitor of NMDA receptors^32,47. This increase in extracellular guanosine will then, in turn, decrease the excitatory impacts of glutamate and return the tissue to homeostatic release levels. This glutamate-guanosine interaction likely also impacts the observed duration of dynamic guanosine release, with increased duration correlated with slower reuptake. Recent studies have demonstrated calcium-dependent regulation of equilibrative nucleoside transporters⁴⁸, the proteins responsible for removal of purines from the extracellular matrix. The increase in intracellular calcium caused by rising glutamate levels facilitates faster purinergic reuptake, which would correspond with shorter guanosine events. Together, our data suggests that guanosine responds rapidly to even minute tissue aggravation, possibly through glutamatergic mediation. Our lab is currently in the process of examining the relationship between glutamate release, its receptors, and dynamic guanosine signaling in the hippocampus and is currently out of the scope of the work presented here.

It is also possible that the increase in guanosine release following electrode implantation is caused by mechanical stimulation. Other rapidly signaling neurochemicals, including guanosine’s sister purine adenosine, have been demonstrated to release following local mechanical stimulation in a wide range of brain regions^40,44,49,50. However, we suspect that if guanosine is responding to mechanical stimulation, it is following a separate pathway than adenosine. Previous studies documenting adenosine release following electrode implantation show singular large, long events immediately after application of the stimulus^40,50,51. We do not observe similar events. Rather, elevated guanosine release begins several seconds to two minutes following implantation, and is not characterized by high concentrations or prolonged release. Instead, as previously described, we see rapid, short, consecutive events which slow over time, suggesting a less direct mechanism than observed for other purines.

Pharmacological confirmation of biological basis for guanosine release

As confirmation of the biological origin of guanosine transients, we pharmacologically demonstrate both calcium dependent signaling and the purinergic nature of these events. Calcium is a ubiquitous intracellular second messenger and essential to both exocytosis and purine transporter protein activation. To demonstrate calcium dependence, we perfused hippocampal slices with unmodified oxygenated aCSF for 15 minutes following electrode implantation, then switched to a modified calcium-free aCSF solution with 1mM EDTA to chelate any remaining extracellular free calcium in the slice. Guanosine release following an initial 15-minute treatment period is compared to the isochronous periods of control and sham data comparison. Particularly, the “sham” and “EDTA” datasets were collected an equal amount of time following electrode implantation, and any differences observed are pharmacological in nature. EDTA also is not known to decrease sensitivity to purines on carbon surfaces⁵⁰.

Calcium deprivation

Calcium chelation had no significant impact on the quantity of guanosine released in individual transient events, although the mean did decrease by approximately 50 nM (Fig. 3A, EDTA mean = 114.0 ± 11.6 nM, Kruskal-Wallis test, 3 groups, p = 0.8733, n = 16 slices). This reduction in average event size is not attributable to a loss of sensitivity, however; as calcium deprivation has varying effects on low, moderate, and high-concentration sets of events, rather than simply shifting the concentration distribution towards smaller events. While the relative frequency curves are similar for the control and sham groups as previously noted, EDTA treatment increases the frequency of mid-concentration events while lowering the number of single high-release transients (Fig. 3B). Approximately fifty percent of each group is composed of events smaller than 100 nM (51.6% control, 53.4% sham, 48.0% EDTA). Midsize events, 100-200 nM, account for approximately twenty-five to thirty percent of control and sham data (29.8% control, 24.3% sham), which is increased to forty percent for the EDTA group. Accordingly, large events (larger than 200 nM) account for twenty percent of both sham and control (17.3% control vs. 19.4%) but decrease to twelve percent during calcium deprivation. This suggests not merely a direct calcium-based blockage of guanosine signaling from EDTA treatment, but also indirect interference in guanosine packaging or loading caused by calcium removal. EDTA, a large, highly charged ion, cannot pass into the intracellular space⁵². We speculate that as calcium deprivation impacts a variety of neurochemical release pathways, including and most importantly glutamate, the changes we observe in guanosine release are likely influenced indirectly by lack of glutamatergic modulation.

Figure 3.

Calcium chelation with EDTA prevents rapid guanosine release. No significant change in event concentration occurs, although the average event size decreases (A, B). Transient length also remains stable when accounting for time-dependent changes (C, D). Signaling slows further following calcium chelation, increasing time between events further (E, F). However, no significant change in the number of events is observed (G, H) (n = 16 slices).

The duration of individual guanosine releases is similarly unchanged with calcium deprivation; while events last longer than observed with the control, no significant change is observed between the sham and EDTA groups (Fig. 3C, EDTA mean 1.01 ± 0.06 s, Kruskal-Wallis test with Dunn’s post test, 3 groups, p_kw < 0.0001, p_{sham vs EDTA} > 0.9999, n = 16 slices). No discernible change in the distribution of event durations was observed between the sham and EDTA groups (Fig. 3D). This suggests guanosine’s reuptake pathways are unaffected or redirected by calcium deprivation or downstream effects thereof. Notably, as multiple members of the equilibrative nucleoside transporter family are calcium activated, particularly ENT1 and ENT2^48,53, this suggests that guanosine’s rapid clearance from the extracellular space may be modulated at least partially by non-calcium dependent transporter proteins. However, as intracellular calcium levels play a more direct role in modulating nucleoside transport than extracellular calcium levels do, it is also likely that internal calcium levels, and thus ENT1 and ENT2 activity, are not appreciably impacted by extracellular Ca²⁺ deprivation.

Calcium removal does, however, cause a demonstrable decrease in guanosine release, even past the notable drop in event frequency observed over time previously in the unmodified aCSF conditions (Fig. 3E, EDTA mean 264.9 ± 35.7 s, Kruskal-Wallis test with Dunn’s post test, 3 groups, p_kw < 0.0001, p_{sham vs EDTA} = 0.0002, n = 16 slices). This accounts for an average ninety-second increase in time between guanosine events. As previously discussed, transients in the control and sham groups were frequently closely bunched followed by extended periods of silence. This phenomenon was not observed during calcium deprivation (Fig. 3F). While a majority of events for both control and sham groups were spaced less than a minute apart (81.9% control, 50.5% sham), only sixteen percent of transients in the EDTA group occur within 60 seconds of each other. The relative frequency of interevent times does not undergo a substantial decrease with increasing time between events, as observed with the control and sham data. Instead, time between events is equivalently distributed up through and past 300 s. This is corroborated by a similar decrease in the number of events observed in each period. While no significant difference is observed between the sham and EDTA groups, the mean number of events is halved from 4.6 ± 1.8 to 2.6 ± 0.5 (Fig. 3G, Kruskal-Wallis test with Dunn’s post test, 3 groups, p_kw < 0.0001, p_{sham vs EDTA} > 0.9999, n = 16 slices). The distribution reflects this; no calcium-deprived slice had more than 7 events in the treatment period, while over half the slices in the sham group had 10 or more.

As anticipated of a calcium-dependent process, dynamic guanosine release is slowed substantially beyond the decrease in release observed with tissue equilibration following perturbation from the electrode. As previously discussed with changes in concentration distribution, EDTA does not decrease electrode sensitivity. Rather, guanosine release is prevented, either directly through inadequate calcium for channel activation or vesicular release, or indirectly through decreased cell excitability from blockage of glutamate release.

Nucleotide metabolism inhibition

We further confirm the purinergic nature of the events measured through treatment with POM1, an ecto-NTPase inhibitor. ATP’s ubiquitous role as a neurotransmitter and neuromodulator is well-established, and GTP has been similarly implicated, although less thoroughly explored^8,54,55. Both purinergic nucleotides metabolize within milliseconds of release into the extracellular space, quickly breaking down into their respective nucleosides, which take considerably longer to convert to downstream metabolites^17,56. The first step of this process is catalyzed by ecto-NTPDases, which cleave an initial phosphate, preparing the molecule for further rapid dephosphorylation. ATP serves as the primary source for a significant amount of dynamically signaling extracellular adenosine, and GTP is anticipated to serve a similar role in guanosine signaling^8,50. Addition of POM1, a competitive NTPDase1 inhibitor, prevents the initial step in the metabolism of GTP to guanosine. Like previously described with EDTA-based calcium chelation, oxygenated aCSF containing 100 μM POM1 was perfused following the initial 15-minute control in unmodified aCSF, and data in the POM1 group is comparable to the sham data set. Here, we confirm that we are measuring a mixture of GTP and guanosine release. We do not measure statistically significant changes between our sham data and the POM1 group for concentration, duration, interevent time, or average number of events (Fig. S4, Kruskal-Wallis with Dunn’s post test, p > 0.05 A, B, C, D; n = 4 slices); however, we do observe a 100 nM decrease in average event concentration for the POM1 group, and a 200 s increase in time between transient events. We are significantly more sensitive to guanosine than GTP^38,57 (Fig. S2), so the observed decrease in event concentration corresponds to an increased GTP to guanosine ratio when the metabolic pathway is inhibited. Additionally, while the limit of detection for GTP has not previously been determined with FSCV, its sister molecule ATP, which interacts similarly with our sensors⁵⁸, is not detectable below 500 nM. We suggest that the observed increase in time between events likewise corresponds to many GTP release events being undetectable. This indicates that a portion of rapid extracellular guanosine release is packaged and released as GTP; we find this unsurprising given GTP’s known activity as a cotransmitter^55,59. Direct guanosine release from glial cells has previously been reported,⁶⁰ as has a complex interactive relationship of adenosine and ATP signaling between neurons and microglia⁶¹. We suggest that further investigation may reveal a similarly intricate set of interactions for GTP and guanosine.

Dynamic guanosine signaling increases in ischemia

Guanosine has repeatedly been implicated as a potent neuromodulator and protective agent during and after ischemia^{6,19-21,26,27}. Extracellular levels rise rapidly following initial injury and remain elevated for up to a week after ischemic injury¹⁴, and are involved in increased glutamate reuptake and modulated damage following excitotoxicity^16,31. Guanosine administration following application of severe ischemic models decreases volume of infarction, lowers pro-inflammatory marker levels, and improves neuron survival levels markedly^19,39. It also promotes improved behavioral, memory, and motor outcomes following damage^9,24,27. While significant work has been done to demonstrate the value guanosine provides in improved ischemic outcomes, all previous studies have focused on basal-level measurements in the hours, days, and weeks following initial assault. However, little is known about its response during tissue aggravation and how it modulates damage pathways at the point and time of ischemic attack. We demonstrate for the first time that guanosine responds rapidly and dynamically to ischemia, with significant changes to the amount, duration, and frequency of release compared to a normoxic environment.

As described in previous sections, we modeled ischemic assault with superfusion of oxygen-glucose deprivation (OGD) aCSF, following initial superfusion with oxygenated aCSF for the control period. The comparison times shown are following 15 minutes of ischemia for consistency with the sham group; however, guanosine upregulation was observed immediately upon administration of OGD. We observed a substantial increase in the size of guanosine transients in ischemic conditions (Fig. 4A, mean 254.4 ± 11.28 nM ischemia, Kruskal-Wallis test with Dunn’s post test, 3 groups, p_kw < 0.0001, p_{sham vs ischemia} < 0.0001, n = 6 slices). The mean event concentration rose by over 100 nM compared to the sham group, as did the median (203.6 nM), which more than doubled that for either the control or sham groups. The distribution of event concentrations also widened considerably; only 31.6% of the ischemic events were 100 nM or smaller, compared to 50% for the control and sham groups (Fig. 4B). Moderate sized events accounted for only 16.4% of events released in ischemia; the majority (52.0%) of ischemic guanosine transients were categorized as large, compared to 19.4% in the sham. A substantial percentage (13.4%) of ischemic transients were larger than 500 nM. Extracellular guanosine is noted to rise to micromolar levels within hours of ischemic assault, suggesting that the local accumulation of rapidly released guanosine may play a role in prolonging elevated purine levels.

Figure 4.

Ischemic conditions induce increased guanosine signaling. Transient concentrations increase following ischemic conditions, and large events make up a larger proportion of signaling (A, B). Signal duration also increases significantly, more than doubling on average (C, D). Frequency of release increases sharply. Average time between events drops to less than half the time observed immediately following implantation (E, F). The number of events also increases drastically, more than tripling the number observed in control (G, H) (n = 6).

This marked increase in the quantity of guanosine released in ischemia continues with prolonged event durations. Mean transient length increases from 0.91 s in the normoxic sham to 2.86 ± 0.09 s during ischemic injury, almost tripling their duration (Fig. 4C, D, Kruskal-Wallis with Dunn’s post test, 3 groups, p_kw < 0.0001, p_{sham vs ischemia} < 0.0001, n = 6 slices). While a portion of this increase is attributable to the increase in concentration of the individual events, we do not see any correlation between event concentration and duration for ischemic events (Fig. 5C and Figure S5). This suggests that the time individual guanosine release events spend in the extracellular space is prolonged during immediate ischemic attack. As previously discussed, purinergic reuptake is primarily governed by equilibrative nucleoside transporter proteins (ENTs). ENT activity is dependent on the nucleoside gradient; as bidirectional transporters, they are capable of both release and reuptake⁵³. We suggest that the increased event durations observed may correlate with saturation of ENTs, causing kinetic limitation of reuptake. This limitation may impact the elevated levels of guanosine sustained past the initial assault, as guanosine continues to flood the extracellular space and is returned far more slowly.

Guanosine is not only released more during ischemic assault; it is released more frequently. Previously, we noted transient events slowed following initial tissue perturbation in normoxia and remained low in a healthy, undisturbed environment. Initiation of ischemia immediately elevates frequency to near-constant release (Fig. 4E, mean IET 15.3 ± 2.2 s, Kruskal-Wallis test, 3 groups, p_kw < 0.0001, p_{sham vs ischemia} < 0.0001, n = 6 slices). The median IET for ischemic tissue drops even further to 6.9 s, as events clustered and frequently overlapped. A majority of events (96.5%) were separated by less than a minute in ischemia, compared to fifty percent in the sham (Fig. 4F). We see similar effects when quantifying the number of guanosine events tabulated during ischemia. An average of 62.3 ± 23.8 events were recorded per slice following ischemic treatment, compared to 4.6 events in the sham, increasing the number of events by an order of magnitude (Fig. 4G, H, Kruskal-Wallis test, 3 groups, p_kw < 0.0001, p_{sham vs ischemia} = 0.0008, n = 6 slices). Release frequency was increased even over the control groups, which saw elevated guanosine following tissue perturbation. This rapid increase in release events implies a correlation with increased cell excitability. As previously discussed, guanosine and glutamate are intimately entwined as cyclical modulators of the other’s activity^16,31. Elevated glutamate release is well-documented in the initial stages of ischemic assault^62,63, leading to excitotoxicity and increased cell sensitivity. Guanosine, it is believed, responds to excitotoxicity and facilitates glutamate reuptake during stress to relieve cells from constant self-propagated stimulation. Subsequently, we expect guanosine’s intermittent release in healthy conditions to be significantly upregulated during the beginning stages of excitotoxicity.

To further investigate the nuanced relationships within guanosine’s release profile, we examined the correlation between event concentration, duration, and interevent time in both our healthy and ischemic groups. Scatter plots showing the concentration, duration, and time until next event for each transient were generated for the control, sham, and ischemia data sets (Fig. S5). Linear regression yielded no linear relationships (R² ≥ 0.7). Next, Spearman correlation matrices were performed. No significant correlations were observed between event concentration and duration for the control, sham, or ischemic data sets (Fig. 5, Spearman correlation, r_control(337) = −0.01, p_control = 0.840, r_sham(90) = 0.17, p_sham = 0.096; r_ishemia(364) = 0.05, p_ischemia = 0.302). From this, we observe no relationship between the amount of guanosine released per event and the length of time it spends in the extracellular space before reuptake. As concentration and duration are independent, this demonstrates that the increased duration observed for ischemic events is not an artifact of larger quantities of guanosine being released. In both our control and sham groups, we similarly saw no correlation between event concentration and time to the next event (Spearman correlation, r_control(337) = −0.09, p_control = 0.104, r_sham(90) = 0.09, p_sham = 0.392). However, in our ischemic group we observed a mild negative correlation (Spearman correlation, r_ishemia(364) = −0.34, p_ischemia < 0.001), with larger guanosine releases having shorter times until the next event. As previously established, both release quantity and frequency are increased significantly during severe ischemic aggravation, so a mild correlation is unsurprising. Of greater interest, however, is the interaction observed between time spent in the extracellular space and guanosine’s release frequency. Following induction of ischemia, duration and interevent time exhibit a moderate positive correlation (Spearman correlation, r_ishemia(364) = 0.45, p_ischemia < 0.001). Longer events which spend more time in the extracellular space, then, are frequently observed with longer times until the next guanosine event. Notably, our sham and control groups both exhibited complete independence for duration and interevent time, indicating this interaction is only present during aggravation and highly upregulated signaling (Spearman correlation, r_control(337) = −0.01, p_control = 0.863, r_sham(90) = 0.03, p_sham = 0.791). This behavior is indicative of the negative feedback effects of an inhibitory autoreceptor with a high activation threshold. Guanosine is currently considered an orphan neuromodulator, while significant evidence for a guanosine-specific receptor has been established, the precise receptor has yet to be identified^8,64. Adenosine’s A₁ receptor is well-known for its inhibitory effects and has been demonstrated to negatively regulate rapid adenosine release^65-67. Guanosine can interact non-specifically at the A₁ receptor to activate a variety of neuroprotective pathways³⁰, however, abolition of adenosine receptors does not fully abolish guanosine’s protective impact, suggesting a role for its individual receptor^8,68,69.

Staining and imaging confirmation of ischemic model

In order to confirm our ischemic model, we used a combination of 2,3,5-Triphenyltetrazolium chloride (TTC) staining and immunohistochemistry (IHC). The overarching goal of the combination of these two staining methods was to confirm delivery of ischemic injury to tissue so that accurate correlation between changes in guanosine event can be attributed to ischemic damage. Sagittal slices of rat brains (400 μm) were separated into control and experimental groups. Slices were allowed a 1 hour recovery period after slicing to return to homeostasis in aCSF. The control slices were treated for an additional 15 mins in the nutrient and oxygen rich aCSF (normoxic aCSF), whereas the experimental slices were bathed for 30 mins in nutrient and oxygen lacking aCSF (ischemic aCSF); mimicking the experimental paradigms as described previously in this text. A subset of tissue was fixed and stained with TTC, a white compound that is converted (enzymatically) to red in living tissue. We observed negligible damage in the control group (Fig. 6). The infarct volume was calculated to be only 11.2 ± 2.5 % compared to the ischemic condition where an average of 65.5 ± 7.1 % of infarcted tissue. This ~6-fold increase (unpaired t-test, p <0.0001; n = 6) in infarcted tissue validates measurable metabolic damage in the tissue from our ischemia model.

Figure 6:

TTC staining reveals a significant increase in infarct tissue volume under ischemic conditions. Image shows the control slices (normoxia) compared to experimental conditions (ischemia); where red is indicative of healthy intact tissue while white is indicative of metabolically challenged tissue (A). A significant ~6-fold increase in damaged/dead infarct volume from normoxic to ischemic conditions was measred (B) (unpaired t-test, p < 0.0001, n = 6).

IHC revealed minimal to no change in the presence of neurons and glial cells immediately after ischemia (Fig. 7A,B). For IHC, 400 μm tissue slices were fixed immediately after the normoxia control or ischemic condition. After fixing, tissue was re-sliced to 75 μm, permeabilized and stained with recombinant anti-NeuN, an antibody used to mark neuronal bodies (stomas), and anti-GFAP, an antibody used to mark acidic proteins located on glial cells. We observed no changes in neuronal or glial cell florescent intensity under ischemic conditions (Fig. 7C unpaired t-test, p = 0.8551; n = 4 and unpaired t-test, p = 0.8544; n = 4, respectively). The combined imaging approaches (TTC and IHC) demonstrate that the cells are actively undergoing metabolic stress due to the ischemic condition; however, minimal cell death occurs immediately after the 30 minute experimental paradigm. This data supports prior work which states that ischemic assault occurs early in the CA1 region of the hippocampus: however, it takes up to 2-3 days to fully succumb to this damage^70,71. Overall, we validate that the changes observed in transient guanosine signaling are a result of tissue stress due to ischemia.

Figure 7:

No significant changes in the presence of neurons or glial cells during ischemia are observed. Neurons are tagged with anti-NeuN (red) and glial cells are tagged with anti-GFAP (green) (A, B). There is no change in relative fluorescent intensity across experimental conditions in the hippocampus for both neurons and glial cells (C). (unpaired t-test, p > 0.05, n = 4).

Conclusion

Guanosine’s dynamic signaling profile during normoxia leading into ischemic assault paints a picture of a neuromodulator primed and ready for action immediately following stress. Absent injury, guanosine maintains a pattern of low, sporadic release maintaining homeostatic levels in the extracellular space. Immediately following aggravation, however, guanosine release elevates in both amount and frequency and remains heightened for the duration of ischemic conditions. Of particular interest is how guanosine release and reuptake is modulated by its fluctuating environment, with special regard to its interactions with glutamate and excitotoxic conditions. While its immediate response to injury is clear, its role in reperfusion and recovery remains shrouded. Guanosine has garnered little attention compared to its more flamboyant sister molecule adenosine, and further investigation in its dynamic signaling is essential for establishing a complete profile of the purinergic response to ischemia.

Materials and Methods

Reagents

All reagents were purchased from Fisher Scientific (Fair Lawn, NJ) unless otherwise noted. Aqueous solutions were made with Milli-Q deionized water (Millipore, Billerica, MA). Stock solutions of 10 mM guanosine, GTP, adenosine, and histamine were prepared in 0.1 M HCl and diluted to 5 μM the day of the experiment for interference studies. 10 mM H₂O₂ was similarly prepared in water. Brain slice experiments were conducted with artificial cerebrospinal fluid (aCSF). Two types of aCSF were used: “normoxia” and “ischemia” aCSF. The normoxia aCSF consisted of 2.5 mM KCl, 1.2 mM NaH₂PO₄, 2.4 mM CaCl₂, 1.2 mM MgCl₂, 126 mM NaCl, 11 mM D-glucose, 25 mM sodium bicarbonate, and 15 mM Tris(hydroxymethyl)aminomethane. The ischemia aCSF was constituted the same with the exception of D-glucose. Normoxia aCSF was oxygenated with a 95% O₂ and 5% CO₂ mix during both slice preparation and the normoxia portion of the experiment. The ischemia aCSF was deoxygenated using ultrapure N₂ gas during the injury portion of the experiment. For calcium deprivation experiments, 1 mM Ethylenediaminetetraacetic acid (EDTA, Sigma-Aldrich, St. Louis, MO, United States) in CaCl2-deficient normoxia aCSF was applied to the slice for 30 minutes following 15 minutes in unmodified normoxic aCSF. For NTPase inhibition experiments, 100 μM sodium polyoxotungstate (POM1, Tocris Bioscience, Bristol, England) in normoxia aCSF was applied to the slice for 30 mins, again following 15 minutes in unmodified normoxia aCSF. Blocking buffer for immunohistochemical experiments included 10% fetal bovine serum (FBS) (Vanvantor Radnor, Pennsylvania, USA) and 1% Triton X-100 (Sigma-Aldrich Saint Louis, Missouri, USA) in 1x Dulbecco's Phosphate Buffered Saline (DPBS) (Cytiva Marlborough, Massachusetts, USA).

Carbon-Fiber Microelectrode Fabrication

Microelectrodes were made from 7-μm T-650 carbon fibers (Mitsubishi Chemical Carbon Fiber and Composites, Sacramento, CA) and were aspirated into 1.2 × 0.68 mm glass capillaries (A&M Systems, Sequim, WA) and pulled into two using a vertical PE-22 Electrode Puller (Narishige, Tokyo, Japan). Fibers were manually cut 150-200 μm from the glass seal using a scalpel under a microscope. Electrode fibers were cut on the day of the experiment and used directly after backfilling with 1 M KCl supporting electrolyte.

Fast-Scan Cyclic Voltammetry

Cyclic voltammograms were obtained using a Dagan Chem-Clamp 5-MEG potentiostat (Dagan Corporation, Minneapolis, MN) coupled to a UNC breakout box (UNC Electronics Shop, Chapel Hill, NC). High-Definition Cyclic Voltammetry (HDCV) software (UNC at Chapel Hill) with a multifunction I/O device (PC1e-6363, National Instruments, Austin, TX) was used for data acquisition and analysis. The waveform used to detect guanosine scanned from −0.4 V to 1.3 V and back at 400 V/s and a frequency of 10 Hz. A 3 kHz lowpass filter was used and all data were background-subtracted to eliminate non-faradaic current. Guanosine current was converted to concentration using the slope of our calibration curve (8.3 nA/μM) as previously reported.⁶¹

Brain Slice Preparation

All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Cincinnati and were performed in accordance with The Guide for the Care and Use of Laboratory Animals (“The Guide”) by the National Research Council. Adolescent male Sprague-Dawley rats weighing 170-180 g (Charles River Laboratories, Wilmington, MA) were housed in a vivarium accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) and provided food and water ad libitum. Rats were anesthetized with isoflurane (Henry Shrein, Melville, NY) and euthanized via decapitation immediately prior to the experiment. The brain was then removed and placed in ice-cold oxygenated (95% O₂ and 5% CO₂) aCSF for 2 mins. The brain was then mounted onto the vibratome stage for slicing using super glue. Sagittal slices of the hippocampus (400 μm thickness) were obtained using a Leica VT1000S vibratome (Chicago, IL) set to a speed of 90 and frequency of 3. The tissue was allowed to recover undisturbed in room temperature oxygenated aCSF for approximately 1 hour prior to the experiment.

For testing, slices were placed in a standard superfusion chamber (Warner Instruments, Hamden, CT) maintained at 34-37° C and superfused with oxygenated aCSF at a rate of 2 ml min-1 using a Watson-Marlow 205S peristaltic pump (Wilmington, MA). A freshly cut carbon-fiber microelectrode was then implanted into the stratum radiatum immediately adjacent to the stratum pyramidale region of the CA1 in the dorsal hippocampus and lowered approximately 75 μm into the tissue using a Narishige MM-3 micromanipulator (Fig. S1B). The electrode was allowed to equilibrate for approximately 10 mins in tissue prior to data collection. FSCV data was then recorded for a total of 45 mins in each slice: 15 mins of normoxic control immediately followed by 30 mins of either ischemia or drug (Fig S1A).

Immunohistochemistry and Imaging

Slices were obtained following the above animal protocol and slicing protocol. The 400μm tissue slices were then separated into control (normal aCSF) and experimental group (ischemic aCSF). After the 1-hour acclamation period, the tissue slices were left in either control or experimental condition solution for 30 minutes. After the 30 minutes, slices were submerged in ice cold 99.8% methanol, anhydrous (Sigma-Aldrich Saint Louis, Missouri, USA) for 1 hour for fixation in a petri dish. They were then removed from the solution washed 3 times for 15 minutes each in DPBS without calcium or magnesium in a clean petri dish. Hippocampi were then sectioned out using a scapple and placed into 6% agarose of NuSieve GTG Agarose (Lonza, Alpharetta, Georgia USA) in 1x PBS (Corning, Corning, New York, U.S) and left to solidify. Once solidified, a tissue puncher was used to punch out hippocampal slices embedded in agarose and then were mounted on a vibratome tray where they were sliced again at 75 μm in ice-cold 1x PBS. Slices were then transferred to a glass slide. Slices were incubated in 300 μL of blocking buffer per slide for 90 minutes. Slides were washed 3 times, 15 min each in DPBS. Slices were then dried and incubated with Anti-NeuN (1:100) antibody Neuronal Marker (Alexa Fluor^® 647) (Abcam Waltham, Massachusetts, USA) and Anti-GFAP (1:1000) antibody (Abcam Waltham, Massachusetts, USA) in the blocking solution overnight. Slides were again washed 3 times 15 min each with DPBS. Slides were dried and then incubated at room temp in Goat anti-Chicken IgY (H+L) Cross-Adsorbed Secondary Antibody tagged with Alexa Fluor Plus 488 (1:200) (Thermo Fisher Scientific, Waltham, Massachusetts, USA) in the blocking solution for 1hr 30 min. After incubation, the slides were then rinsed with DPBS three times for 15 min each and then dried. Slices were imaged with an AxioZoom macroscope (Carl Zeiss Microscopy, Oberkochen, Germany) with an Axiocam 506 mono camera and Cy5 and EGFP filter cubes. 14-bit images were captured and analyzed in Zen software (Carl Zeiss Microscopy) and analysis was done using Image J.

TTC Staining and Infarct Area Calculations

2,3,5-triphenyltetrazolium chloride (TTC) powder (Sigma-Aldrich, St Louis, MO) was dissolved in 1x PBS (Corning, US) to a w/v percentage of 2%. Since TTC dye is water-soluble and sensitive to light, the solution was covered in aluminum foil and prewarmed in an incubator for 30min at 37°C. Both normal and ischemic brain slices (n = 6 for each group) were placed in separate petri dishes containing the TTC solution. The slices were spread flat by a set of paint brushes to the bottom of the dishes; they were then covered by aluminum foil and incubated at 37°C for 30 min in the dark. After 30 min, slices were washed with the 1x PBS solution and transferred to a petri dish containing 10% formalin neutral buffered solution (Sigma-Aldrich, St. Louis, MO, USA) and fixed for 24 hrs. After fixation, slices were washed three times with 1X PBS buffer. Slices were placed on a white plastic board and the excess PBS was removed. Rulers were placed on the board both horizontally and vertically for scale. The picture of the slices was taken by using an iPhone 14 Pro Max camera and transferred to the computer where the infarct area and volume calculations were performed using ImageJ software (NIH, Bethesda, MD, USA).

Statistics

All statistical analyses were performed in GraphPad Prism v. 9.0 (GraphPad Software, La Jolla, CA). Statistical p-values were significant at 95% confidence level (p < 0.05). Values on plots are reported as the mean ± standard error of the mean (SEM), with n representing the number of brain slices.