Complex 33-beam simulated galactic cosmic radiation exposure impacts cognitive function and prefrontal cortex neurotransmitter networks in male mice

Desai, Rajeev I.; Kangas, Brian D.; Luc, Oanh T.; Solakidou, Eleana; Smith, Evan C.; Dawes, Monica H.; Ma, Xiaoyu; Makriyannis, Alexandros; Chatterjee, Subhamoy; Dayeh, Maher A.; Muñoz-Jaramillo, Andrés; Desai, Mihir I.; Limoli, Charles L.

doi:10.1038/s41467-023-42173-x

Download PDF

Article
Open access
Published: 27 November 2023

Complex 33-beam simulated galactic cosmic radiation exposure impacts cognitive function and prefrontal cortex neurotransmitter networks in male mice

Rajeev I. Desai ORCID: orcid.org/0000-0002-4053-5996^1,2,3,
Brian D. Kangas ORCID: orcid.org/0000-0002-4597-9801^1,2,
Oanh T. Luc^1,2,
Eleana Solakidou^3,4,
Evan C. Smith³,
Monica H. Dawes ORCID: orcid.org/0000-0002-4729-5526^1,2,
Xiaoyu Ma³,
Alexandros Makriyannis³,
Subhamoy Chatterjee⁵,
Maher A. Dayeh^6,7,
Andrés Muñoz-Jaramillo ORCID: orcid.org/0000-0002-4716-0840⁵,
Mihir I. Desai^6,7 &
…
Charles L. Limoli⁸

Nature Communications volume 14, Article number: 7779 (2023) Cite this article

1504 Accesses
12 Altmetric
Metrics details

Subjects

Abstract

Astronauts will encounter extended exposure to galactic cosmic radiation (GCR) during deep space exploration, which could impair brain function. Here, we report that in male mice, acute or chronic GCR exposure did not modify reward sensitivity but did adversely affect attentional processes and increased reaction times. Potassium (K+)-stimulation in the prefrontal cortex (PFC) elevated dopamine (DA) but abolished temporal DA responsiveness after acute and chronic GCR exposure. Unlike acute GCR, chronic GCR increased levels of all other neurotransmitters, with differences evident between groups after higher K+-stimulation. Correlational and machine learning analysis showed that acute and chronic GCR exposure differentially reorganized the connection strength and causation of DA and other PFC neurotransmitter networks compared to controls which may explain space radiation-induced neurocognitive deficits.

Life-long brain compensatory responses to galactic cosmic radiation exposure

Article Open access 22 February 2021

Multi-domain cognitive assessment of male mice shows space radiation is not harmful to high-level cognition and actually improves pattern separation

Article Open access 17 February 2020

Modeling space radiation induced cognitive dysfunction using targeted and non-targeted effects

Article Open access 23 April 2021

Introduction

Human deep space exploration to the Moon and Mars will require travel beyond the Earth’s protective magnetic field and involve significant health challenges for astronauts. This includes extended exposure to Galactic Cosmic Radiation (GCR) consisting of particles of high atomic number, energy, and charge (HZE). Due to shielding constraints, charged particles penetrate the spacecraft rendering astronauts unprotected against GCR exposure. During interplanetary travel, each cell within an astronaut’s body is likely to be hit by either a proton or secondary particle every few days and by a HZE ion approximately once a month^1,2. Mounting evidence suggests that exposure to space radiation can cause variable damage to the central nervous system (CNS)^3,4,5,6,7,8. Estimates indicate that during a Mars mission, nearly every neural cell will be transversed once every three days^3,4,5,6,7,8 by various combinations of protons and/or HZE nuclei. These estimates have raised concerns that continuous exposure to GCR may damage CNS functionality and jeopardize mission success^{5,6,7,8,9,10,11}.

Investigations into how extended GCR exposure adversely affects the brain have been impeded by the limited number of humans that have traveled in deep space. This fact has necessitated ground-based studies in which laboratory animals are exposed to different simulated patterns of space radiation to probe the multitude of CNS effects. Early CNS studies in rodent models primarily focused on acute effects following simplified single ion exposures that were not accurate representations of the low dose and low dose rate deep space radiation environment in which astronauts are continuously subjected^{5,6,7,8,9,10,11}. More recent results obtained with advancements in beam delivery technology have indicated that space-relevant doses and dose rates of various single-ion, mixed-field GCR, and neutron exposures lead to cellular and molecular damage to the CNS that likely results in short- and long-term decrements in neurocognitive performance^{5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26}.

Although the precise mechanisms through which GCR produces damaging effects on the brain remains unknown, past studies have linked behavioral decrements to compromised structural plasticity and network level connectivity, as well as elevated microgliosis, astrogliosis, and inflammatory signaling in the irradiated brain^{5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26}. In nearly every instance, these studies found poor correlation with the particle linear energy transfer (LET: keV/µm) or dose-responsiveness when realistic exposures (≤50 cGy) were evaluated^{5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26}. Investigators have also attempted to relate neurocognitive deficits to the neurochemical sequelae of high doses of single-ion HZE particle and proton radiation. However, assay limitations and a focus on the neurotransmitter dopamine (DA) and resultant behavioral effects have yielded mixed results, i.e., DA activity related to some but not all behavioral deficits^8,13,14,15. More recently, studies have reported that space radiation exposure also produces persistent deleterious changes in other neurotransmitters involved in cognition²⁷ such as glutamate (Glu), γ-aminobutyric acid (GABA), and serotonin (5-HT)^{17,18,19,20,28,29} as well as monoamine metabolism³⁰.

The recent availability of complex 33-beam terrestrial GCR simulation, advances in rodent touchscreen technology, and development of in vivo microdialysis combined with biochemical techniques using liquid chromatography mass spectrometry (LC-MS/MS), offers a unique opportunity to delineate how space-relevant GCR exposure alters in vivo neurotransmitter function to impact neurocognitive performance. Some recent studies utilized various touchscreen-based neurocognitive assays in rodents to find that exposure to single-ion or 6-mixed-ion beam either improved, impaired, or did not alter performance^31,32,33. Likewise, we found that acute or chronic delivery of the 33-beam GCR exposure produced less severe neurocognitive and electrophysiological impairments in mice, suggesting that the impact of space-relevant radiation exposure may be less detrimental than simplified irradiation paradigms^34,35,36. Collectively, such data highlight the critical need to study the effects of more space-relevant radiation exposures on multiple within-subject neurocognitive and neurotransmitter endpoints to better estimate the risk of space radiation exposure to CNS functionality.

Given that brain neurotransmitters provide informative readouts of neurocognitive function, it is critical to delineate how exposure to space-relevant radiation alters neurotransmitter activity in vivo to impact neurocognitive performance. Here, we conducted studies in adult mice to examine the impact of the complex 33-beam terrestrial GCR simulation on prefrontal cortex (PFC) neurotransmitters involved in translational touchscreen-based complex behavior. Based on well-established reports, the PFC was selected because of the extensive body of evidence showing its involvement in higher order behaviors and cognitive function, including motivation and psychomotor vigilance that are likely required for mission success during spaceflight^{37,38,39,40,41}. With regard to neurotransmitters, we primarily focused on documenting changes in monoamines, Glu, and GABA as a group of key neurotransmitters in the PFC because prior work has shown: a) that they all play an important role in various cognitive processes, including motivation and attention and b) that their dysregulation is directly linked to a variety of neuropsychiatric conditions^{27,37,38,39,40,41,42,43,44}. We report findings describing how the PFC neurotransmitter networks are reorganized by realistic complex GCR simulations and how that informs on neurocognitive outcomes to provide a foundational and comprehensive outlook regarding the impact of space radiation on CNS function.

Results

Figure 1 shows a detailed timeline of experimental protocols. Mice received a 33-beam GCR simulation for a total dose of 40 cGy, either all-at-once (acute) or 49.9 cGy over 4 weeks (chronic). Next, economic demand functions for a palatable food reinforcer were determined to assess changes in reward sensitivity. Thereafter, sustained attention was examined during a psychomotor vigilance task. The same mice were then probed by combined in vivo microdialysis with LC-MS/MS in order to quantify potassium (K+)-evoked (i.e., depolarization elicited) changes in multiple neurotransmitters in the PFC. Finally, a combination of statistical and machine learning techniques (correlation analysis, Granger causality test, and iterative random forests) was used to determine how acute and chronic GCR altered PFC neurotransmitter networks to impact neurocognitive performance.

**Fig. 1: Timeline of experimental protocols in mice.**

Effects of acute or chronic GCR exposure on motivation and sustained vigilance

Economic demand

An important goal for evaluating the impact of GCR on cognition is to employ tasks in laboratory animals that meaningfully translate to humans. Thus, we first investigated the impact of acute or chronic GCR exposure in mice engaged in a touchscreen-based economic demand task designed to measure motivation (Fig. 2a). In this assay, subjects respond to a geometric stimulus on the touchscreen under escalating response requirements (price) for small amounts of a highly palatable liquid reward (commodity) to examine the extent to which subjects will defend consumption in the face of increasing cost (Fig. 2a; Online Methods). As shown in Fig. 2b, the exponential demand equation provided a strong fit to the findings of decreased consumption as a function of increased price in all three treatment groups (R² = 0.98–0.99 [control], 0.97–1.0 [acute], 0.94–0.99 [chronic]). A milk concentration-related rank order of consumption was apparent in the control and acute treatment groups. Accordingly, for these treatment groups, consumption of 50% milk was associated with the most inelastic curve relative to other milk concentrations, followed by 20%, 5%, and 0%. The chronic treatment group generally followed this rank order, with the exception of responding for the 20% milk concentration, which was comparatively more inelastic at higher response requirements than 50%. Moreover, Fig. 2c shows that although there was an orderly positive relationship between milk concentration and both Q₀ and essential value, there were no systematic differences across treatment groups. These outcomes are confirmed statistically by a significant interaction for milk concentration (Q₀: F[3,84] = 54.74, p < 0.0001; essential value: F[3,84] = 155.90, p < 0.0001), but not for the treatment group (Q₀: F[2,84] = 0.99, p = 0.38; essential value: F[2,84] = 2.89, p = 0.06).

Psychomotor vigilance

Next, we then determined whether these same mice would display any deficits on sustained attention using the psychomotor vigilance task. In this assay, subjects were required to carefully attend and respond effectively to intermittent geometric stimuli presented briefly in variable locations and after variable intervals on the touchscreen to obtain the palatable liquid reward (Fig. 3a; Online Methods). Figure 3b presents the mean titrated reaction time for each treatment group across the 15-day testing condition. Significant differences in task performance across treatment groups were especially prominent during the first week of testing, with the control group quickly approaching the asymptotic level in mean titrated reaction times of approximately 2.5 s. Both the acute and chronic groups also displayed quicker titrated reaction times across the 15-day testing condition, but this developed at a slower rate, which was confirmed by a significant main effect of treatment (F[2,315] = 30.56, p < 0.0001). Moreover, although differences in reaction times among treatment groups did not reach statistical criteria after day 6, acute and chronic group mean titrated reaction times failed to fully approximate levels observed in the control group by the end of 15-day testing condition, with both GCR treatment groups having reaction times plateau at values approximately 1 s longer than the average control subject.

Effects of acute or chronic GCR exposure on neurotransmitters in the PFC

Baseline and high K+-evoked extracellular levels of neurotransmitters in the PFC

In the same mice, we investigated how GCR impacts neurotransmitters in the PFC–a key brain region implicated in reward sensitivity and attention^{37,38,39,40,41,43}. Using microdialysis probes (Fig. 4a, b) targeting the PFC (Fig. 4b, c) and LC-MS/MS techniques (Fig. 4d, e), we determined real-time baseline (4 mM K+) and high K+ (30, 60, and 120 mM)-evoked changes in multiple neurotransmitters (Fig. 4a–e; Online Methods). See Supplementary Table 1 for cumulative basal and high K+-evoked neurotransmitter levels in each group. We found that levels of DA and 5-HT were significantly higher in acute [2.71- and 1.75-fold, respectively] and chronic [2.14- and 1.96, respectively] groups compared the controls (Fig. 5a, b, respectively). Baseline levels of norepinephrine (NE) were also significantly greater (1.60-fold) in the chronic group compared to controls (Fig. 5c). In contrast, neither acute or chronic GCR significantly changed baseline levels of Glu and GABA compared to control values (Supplementary Table 1, Fig. 5d, e, respectively). A significant difference between levels of NE and Glu in the acute vs chronic GCR groups was observed (2.19-fold and 1.45-fold, Fig. 5c, d, respectively); for the other neurotransmitters, the basal levels in acute vs chronic GCR groups were unchanged (Fig. 5a–e).

**Fig. 4: Flowchart illustration of in vivo microdialysis experimental set-up, design, sample collection, sample analysis, and histology.**

**Fig. 5: Acute and chronic GCR exposure produced differences in K+-evoked neurotransmitter levels in the PFC.**

Increasing K+-stimulation uncovered distinct differences in neurotransmitters among the treatment groups. DA (1.87–2.41-fold) and 5-HT (1.78–1.86-fold) were significantly higher in the acute group compared to controls during two or more higher K+ concentrations (Fig. 5a, b, respectively), whereas other neurotransmitters did not significantly differ between acute and controls (Fig. 5c–e, respectively). In contrast, 5-HT (1.87–2.19-fold), NE (1.34–1.40-fold), Glu (1.45–1.65-fold), and GABA (1.50–1.92-fold) levels were significantly higher in the chronic group compared to controls during stimulation with two or more higher K+ concentrations (Fig. 5b–e, respectively). DA levels in the chronic group were generally higher compared to controls with significant differences observed at 120 mM K+ (1.69-fold; Fig. 5a). While NE (60 mM K+: 1.45-fold), Glu (120 mM K+: 1.56-fold), and GABA (120 mM K+: 1.62-fold) differed between the two GCR groups (Fig. 5d, e, respectively), the other neurotransmitters were unchanged (Fig. 5a–e).

Time course of neurotransmitter responsiveness in the PFC

In control mice, 30 mM induced a rapid increase in DA (~394 ± 122% basal nM levels) and NE (~157 ± 37% basal nM levels) within 20-min that rapidly returned to near baseline values (40–80 min; Fig. 6a, c, respectively). Increasing the K+ to 60- or 120-mM also elevated DA (~259–293 ± 98–102% basal nM levels) and NE (207–221 ± 31–57% basal nM levels) about 40-min after each K+ stimulation with levels of both neurotransmitters returning to baseline values about 20–40 min after peak effects (Fig. 6a, c, respectively). In contrast, acute or chronic GCR exposure completely abolished normal rapid elevations and decreases in DA responsiveness during each increasing K+ stimulation (Fig. 6a). DA levels in both GCR groups were generally lower than controls throughout the experiment (Fig. 6a). Further analysis confirmed that DA levels in the control group were significantly greater than basal values and levels in the acute and chronic GCR groups at 20 min during 30 mM (Fig. 6a; ~394 ± 122 vs 116 or 94 ± 45 or 39% basal nM levels, respectively; p values < 0.05). Unlike DA, acute GCR mice showed normal K+-stimulated rapid increases and decline in NE responsiveness (Fig. 6c). NE changes in chronic group were generally lower but were not significantly different compared to controls (Fig. 6c). Levels of 5-HT were not consistently altered in all groups during 30- or 60-mM, but a steady rise in 5-HT in all three groups was observed at 120 mM; except for some separation between acute and control groups at 240 min (p < 0.05) no differences were observed (Fig. 6b: p values > 0.05).

**Fig. 6: Acute and chronic GCR exposure abolished K+-evoked DA response in PFC, but not other neurotransmitters.**

In control mice, 30- and 60-mM decreased Glu levels (Fig. 6d: 120–160 min, ~53 ±9.1% basal nM levels; p values < 0.05), whereas GABA was unaffected (Fig. 6e). Increasing the K+ to 120 mM restored Glu levels towards basal values (Fig. 6d), but GABA was significantly elevated above basal values in control mice (Fig. 6e; 200–240 min: ~515 ± 156% basal nM levels; p values < 0.05). In acute and chronic groups, 30- and 60-mM did not change Glu or GABA levels compared to controls (Fig. 6d–e, respectively). However, 120 mM K+ increased Glu in the chronic group compared to controls and acute GCR mice (200–220 min: ~134 ± 33% basal nM levels; p values < 0.05); controls and acute GCR mice produced similar effects on Glu (Fig. 6d). Compared to the controls, GABA levels were significantly lower (200 min: ~223, ±60% basal nM levels; p values < 0.05) and higher (220 min: ~957, ±283% basal nM levels; p values < 0.05) in, respectively, acute and chronic GCR mice during 120 mM (Fig. 6e). The increased GABA levels from 200 to 240 min in chronic mice were also significantly different than in the acute group (p < 0.05; Fig. 6e).

Reorganization of neurotransmitter network in the PFC

We used Pearson correlation analysis to determine the relationships among neurotransmitters (Supplemental Tables S1–S4). Under baseline 4 mM, correlations between neurotransmitters become more positive and negative after acute and chronic GCR exposure, respectively, relative to controls (Fig. 7a). Increasing K+ from 30- to 120-mM produced small increases in positive correlations between the neurotransmitters in controls, whereas correlations between neurotransmitters in the GCR groups compared to 4 mM and controls at higher K+ concentrations were significantly higher and more positive (Fig. 7a).

**Fig. 7: Pearson’s correlation analysis showed that acute and chronic GCR exposure differentially reorganized PFC neurotransmitter networks.**

We plotted heat maps of pairwise neurotransmitter correlations (Fig. 7b) and constructed corresponding neurotransmitter networks based on values that only include Pearson’s correlation ≥ 0.5 with a significance p-value of <0.05 (Fig. 7c; Supplemental Tables S1–S4). During 4 mM, there was a significant and strong positive correlation between Glu–GABA in controls (Fig. 7b, c). In control mice, this Glu-GABA network was no longer linked at 30 mM and 60 mM but was preserved at 120 mM with a significant and positive correlation between the two neurotransmitters (Fig. 7b, c). In the acute group, these neurotransmitter interactions were altered during 4 mM baseline conditions; the Glu–GABA interaction was not significant and no longer correlated (Fig. 7b, c). In acute GCR mice stimulated with 30 mM, there were significant and strong positive correlations between three neurotransmitter pairs (DA–NE, NE–Glu, and Glu–GABA) that were not present in control mice at 30 mM (Fig. 7b, c). Glu and GABA were also significantly and positively correlated, and no interactions were observed between GABA and monoamines (Fig. 7b, c). 60 mM also resulted in significant and positive correlations between two neurotransmitter pairs in the acute GCR mice (DA–NE and NE–GABA). However, this differed from 30 mM, in that no neurotransmitter correlated with Glu; instead DA was positively correlated with NE and NE directly interacted with GABA (Fig. 7b, c). Except for Glu, 120 mM induced significant and strong positive correlations between all neurotransmitters resulting in multiple connecting lines for DA, other monoamines, and GABA in the acute GCR group (i.e., DA–5-HT/NE/GABA, GABA–5-HT/NE Fig. 7b, c). After chronic GCR, the Glu–GABA interaction was preserved during 4 mM and a significant and positive interaction was observed between DA–NE (Fig. 7a–c). In general, increasing K+ resulted in distinct neurotransmitter interactions in the chronic GCR group compared to others. 30 mM produced significant positive correlations between DA and 5-HT/NE and retained the Glu–GABA networks that were evident in control mice during 4 mM, but not at 30 mM (Fig. 7b, c). Unlike control and acute groups, significant positive correlations were only observed between certain neurotransmitters in the chronic group during 60 (DA–5-HT and Glu–GABA) and 120 (DA–5-HT, 5-HT–GABA, and Glu–GABA) mM K+ (Fig. 7b, c).

Machine learning prediction models to determine PFC neurotransmitter networks

Simplified correlation-based neurotransmitter networks that rely on Pearson correlation do not evaluate either nonlinear relationships or causality between neurotransmitter levels for different target classes. Thus, we used an iterative random forest (iRF) based regression model called iRF-LOOP to predict how acute or chronic GCR exposure alters PFC neurotransmitter interactions compared to controls (Fig. 8). This analysis revealed a highly complex and interconnected neurotransmitter network that changes with increasing K+-stimulation across treatment groups (Fig. 8a, b). Overall, during each K+ stimulation, GABA strongly predicted Glu levels but not necessarily the reverse. This connection strength was generally not impacted after acute or chronic GCR compared to controls (Fig. 8a, b). While Glu predicted levels of 5-HT/NE and GABA predicted levels of DA/5-HT/NE during stimulation with one or more K+ concentrations in control mice, these interactions were reorganized in the acute GCR group. Evidence of this is apparent by the elimination of the GABA–DA connection, a reduction (except for 120 mM) of the GABA–5-HT interaction, the emergence of a stronger DA–NE or DA–Glu relationship, where DA predicted levels of NE and vice versa or DA predicted levels of Glu (Fig. 8a, b). Likewise, a reduction in interactions between GABA and DA/5-HT/NE were observed in the chronic group compared to controls. In general, the interactions between Glu and monoamines did not significantly alter compared to the control group except for some small variations in connection strength and direction during one or more K+ concentration (Fig. 8a, b). Unlike the acute group, no DA–Glu connections were found in chronic GCR mice (Fig. 8a, b). Strong interactions were also predicted between NE and DA/5-HT and vice versa in control mice during stimulation with one or more K+ concentrations, but DA–5-HT were not connected in these mice except at the 120 mM K+ (Fig. 8a, b). These monoamine relationships were completely reorganized after acute GCR exposure and DA now strongly predicted 5-HT/NE levels and vice versa (Fig. 8a, b). Similarly, chronic exposure to GCR caused a strong bidirectional relationship to develop between DA and 5-HT, strengthened the 5-HT–NE interaction, and conserved the NE–DA interaction (Fig. 8a, b). In general, at the 30–120 mM, neurotransmitter networks were more balanced and well connected in control mice compared to mice exposed to acute and chronic GCR that exhibited fewer connections with more monoamine reorganization. In addition, we conducted further linear regression analysis and added a ‘+’ sign of impact to reflect that an increase in one neurotransmitter promotes the increase in the levels of another neurotransmitter, whereas a ‘–’ sign reflects suppression (Fig. 8a, b).

**Fig. 8: Machine learning-derived networks unravel the impact of neurotransmitters on each other and their rearrangement after acute and chronic GCR exposure.**

It is noteworthy that the above approaches do not treat the input data as time series. Thus, we performed a causality analysis to better appreciate the causal impact of the neurotransmitter changes induced by different GCR exposure levels. We performed the Granger Causality test to understand the significance of those causal impacts and generated boxplots depicting the distribution of p-values over all the subjects for impact of each neurotransmitter on DA and vice versa (Fig. 9a, b). We primarily focused on DA, as our data indicate this neurotransmitter to be a key player in the reorganized neurotransmitter networks. We found that for the control group, the median p-values were greater than 0.05 for all the neurotransmitters impacting DA and vice versa (Fig. 9a, b), implying that the other neurotransmitters do not impact DA and vice versa in the control subjects. However, based on median p-values, in subjects exposed to acute GCR only the neurotransmitter GABA impacts DA (Fig. 9a) and in acute GCR group the causal impact of other neurotransmitters (except GABA) on DA cannot be meaningfully established. Moreover, Fig. 9 shows that DA in turn impacts GABA, GLU, and 5-HT, but not NE. Likewise, in subjects exposed to chronic GCR, the median p-values show that GABA and NE impact DA and vice versa (Fig. 9).

**Fig. 9: Pair-wise Granger causality tests identify dopamine as an important player in driving or being driven by other neurotransmitters after acute and chronic GCR exposure.**

Discussion

Over the next decade, NASA’s planned Artemis missions intend to establish a long-term human footprint on the Moon in preparation for future human deep space exploration to Mars. To successfully realize this goal, it is imperative to fully understand the impending CNS health risks associated with incessant exposure to deep space radiation that will be encountered by astronauts during long-duration missions to the Moon and Mars. Variations across Earth-based simulated irradiation paradigms, age, and behavioral history at the time of exposure and cognitive testing, and specifics of behavioral testing have further complicated our ability to translate findings from preclinical studies in rodents to astronauts^{1,2,3,4,5,6,7,8,9,10,45}. These issues have impeded our ability to adequately document a clear link between in vivo neurobiological abnormalities and neurocognitive deficits.

Here, we evaluated whether whole-body acute or chronic exposure to a complex 33-beam terrestrial GCR simulation deteriorated performance on two distinct touchscreen tasks that appraised reward sensitivity/motivation or sustained attention in the same mature male mice irradiated at ~6 months old and, if so, whether any deficits can be linked to in vivo neurotransmitter abnormalities in the PFC. Contrary to earlier work using simplified irradiation paradigms^{5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26}, we found that acute or chronic GCR did not appreciably modify motivation for reward via economic demand assessments. From a risk-assessment perspective, this is a desirable outcome. Indeed, although performance on an economic demand procedure is not cognitive behavior per se, it is encouraging to document no blunted sensitivity or motivation for reinforcing consequences. Moreover, it provides evidence that other cognitive deficits, for example, the adverse effects on attentional processes and reaction time that were observed, were not simply a product of GCR-related deficits in motivation for reward. As such despite observing no deleterious effects on economic demand indices in the same mice, the development of psychomotor vigilance performance was slowed as was asymptotic reaction time levels. That is, GCR-treatment group differences subsided over time with extended daily training/testing conditions (i.e., test days 10–15). However, although differences in reaction time among treatment groups did not reach statistical criteria after day 6, in subjects exposed to acute or chronic GCR, performance on the vigilance task never fully approximated that of control animals (i.e., ~1 s inferior reaction time). Projected to astronauts, albeit statistically insignificant, these consequences could nevertheless adversely impact critical time sensitive decision making in-flight and jeopardize mission success.

The above observations are consistent with recent studies that showed exposure to simplified irradiation paradigms either selectively impaired specific aspects of cognition (attentional set shifting) or did not alter various touchscreen-based discrimination and learning assessments in male rodents^{31,32,33,34,35,36}. Interestingly, Eisch and colleagues reported that single-ion exposure selectively improved touchscreen pattern separation in male mice without affecting other outcomes and improved discrimination learning in female mice but impaired stimulus-response rule-based habit learning^35,36, suggesting certain sex differences. Work published from a similarly exposed cohort found female mice somewhat more resistant to cognitive deficits³⁴, which paralleled past results obtained with single helium ion exposures⁴⁶. This work has suggested that the higher levels of baseline inflammation may precondition the female rodent brain, thereby attenuating the extent of space radiation injury and possibly ameliorating resultant cognitive impairments.

In previous work, cohorts of male and female mice exposed to the same acute or chronic delivery of the 33-beam GCR were found to exhibit similar and less severe neurocognitive and electrophysiological impairments³⁴. Importantly, this study was not designed to discriminate differences in dose rate but rather to elucidate how a complex GCR simulation might impact CNS functionality. For the chronic irradiations, dose rates for individual energetic particles varied between ~0.04 and 0.9 cGy/min, with a collective mean dose rate of ~2.08 cGy/h. For the acute irradiations, dose rates for individual energetic particles varied between ~0.16 and 4 cGy/min, with a collective mean dose rate of 25 cGy/h. Unfortunately, behavioral tasks are simply not sensitive enough or designed to pick up these small differences in radiation dose rates. It is important to note that few studies have systematically and rigorously evaluated the impact of space relevant dose rates on CNS functionality. The first study that implemented a dose rate of 1 mGy/day over the course of six months, reported an array of adverse behavioral, electrophysiological, and molecular outcomes compared to unirradiated controls⁴⁷.

Collectively, these results suggest that exposure to space-relevant radiation might be less detrimental on neurocognitive functionality than expected from earlier reports using more simplified irradiation paradigms³⁴. The complexity and differences in CNS outcomes across studies highlights the critical need for NASA to further evaluate the impact of space-relevant radiation exposures on multiple within-subject neurocognitive endpoints in male and female subjects. Adapting approaches used here can only improve estimates of space radiation exposure risk to the neurological well-being of astronauts^{9,31,32,33,35,36,45,48}.

Several aspects of the present results are noteworthy. First, the finding that vigilance performance improved after acute or chronic GCR exposure with extended daily touchscreen training/testing is intriguing. Here, mice were first trained/tested on the economic demand task followed by the vigilance task. Projecting these data to astronauts suggests that tasks involving touchscreen-based vigilance training or even gaming might help forestall further cognitive decline caused by GCR. In support of the foregoing, data from human studies show that virtual environmental enrichment can enhance hippocampal memory across astronaut relevant age groups^49,50. Second, it is possible that the detrimental effects of both acute or chronic GCR on both touchscreen tasks may have dissipated and might be more pronounced early (i.e., 1–2 months) after GCR exposure. Although, additional work is needed to document these early effects of acute or chronic GCR exposure, the adverse effects on attention observed 5–6 months post-GCR exposure are consistent with prior work showing that harmful CNS effects persist long (i.e., up to 1 year) after irradiation^{7,8,9,10,11,12,21,22,23,24}. Thus, the adverse cognitive effects of GCR may persist during long-duration missions and upon return to Earth, which might compromise subsequent terrestrial activities. Third, the same mice were tested on two distinct tasks that varied in the type of cognitive challenges required to overcome for optimal performance, i.e., obtaining reward for increased price vs maintaining attention to obtain reward. Prior studies have shown that adjusting conditions that varied in cognitive difficulty could reveal more subtle yet significant effects of radiation exposure on cognition that might not otherwise be observed using simple tasks^{6,7,8,9,45,48}. Future studies that manipulate cognitive load in subjects pre-trained on neurocognitive tasks will be helpful for elucidating spaceflight stressor effects on astronaut performance during (i.e., evolving mission scenarios) and after the mission^{6,7,8,9,45,48}.

Although past studies have related neurocognitive deficits to compromised structural plasticity, network level connectivity, elevated microgliosis, astrogliosis, and/or inflammatory signaling in the irradiated brain^{5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26}, the precise neurobiological mechanisms responsible for GCR-induced neurocognitive damage remains uncertain. Given that brain neurotransmitters provide relevant readouts of neurocognitive function²⁷, it is surprising that the impact of space-relevant radiation on brain neurotransmitter dynamics has not been delineated. Here, we focused on identifying a link between GCR-induced changes in reward sensitivity and attention and in vivo neurotransmitter alterations in the PFC, a key brain region involved in both reward and attention^{27,37,38,39,40,41}. We found that exposure to acute or chronic GCR selectively increased normal basal levels of DA, 5-HT, and NE, but not Glu or GABA compared to controls. These results are consistent with previous in vitro findings in rodents showing long-lasting alterations in DA, other monoamines, and their metabolites after exposure to simplified irradiation paradigms, but are somewhat contrary to alterations in Glu and GABA activity observed following simplified radiation exposures^{8,14,15,17,18,19,20,33}. From a broader perspective, previous studies have used simplified single-ion exposures which may not produce the physiologically relevant changes desired to simulate deep space radiation. By using a complex 33-beam GCR simulation we more accurately reproduce the physiologically relevant conditions, which may produce differing results from simpler, less representative models. However, we emphasize that past work using complex GCR³⁴, mixed field 5 ion studies^26,51 and even acute⁵² and chronic low dose neutron⁴⁷ studies suggest multiple disruptions across interconnected brain regions. Moreover, chronic GCR exposure increased levels of all K+-stimulated neurotransmitters compared to controls, indicating global PFC changes that compromised all neurotransmitter systems; these effects were less significant after acute GCR. Notably, both GCR paradigms similarly altered DA levels compared to other neurotransmitters. In addition, time course analysis showed that the ability of the PFC DA system to respond normally to changing physiological conditions was severely and equally compromised after exposure to both paradigms, where mice were unresponsive to K+ stimulation. In contrast, the temporal patterns of other neurotransmitters were less affected, requiring greater K+-stimulation to observe any differences. Thus, we posit that acute and chronic GCR did not alter reward sensitivity/motivation because significant amounts of PFC DA was available to engender rewarding properties. However, the adverse effects on attention might be attributed to radiation-induced changes in the temporal pattern of DA dynamics in the PFC, i.e., compromised normal DA responsiveness that is typically necessary for attentional processes. This is consistent with the view that optimal extracellular DA levels in the PFC is critical for sustaining normal neurocognitive functioning, and that any DA disruptions can lead to cortical dysfunction^42,43. That DA mechanisms are compromised to differentially impact aspects of neurocognitive function is in agreement with prior work showing associations between changes in DA activity and some but not all DA-mediated behaviors^8,13,14,15.

It is important to note that the other factors (e.g., behavioral training and age) in addition to GCR exposure could also contribute to the overall observed PFC neurotransmitter changes. While assessment of age and behavioral training history were not part of the overall study goals, it is noteworthy that the animals in the control group were of the same age with the comparable behavioral history as the GCR-exposed animals, meaning that while age and behavioral training may be important variables, it is one that we considered in the experimental design. Nevertheless, we appreciate that the neurotransmitter responses may be a consequence of how these controlled variables may interact with the neurotransmitter changes elicited by GCR simulation. From a more global perspective, the entire brain is susceptible to radiation injury, but clearly will not exhibit equivalent levels of sensitivity. Behavior is a multifaceted response to a variety of molecular and biochemical changes impacting neurotransmission, and GCR exposure produces persistent changes in functional and molecular outcomes in the CNS. Other work has found exposure to energetic iron ions to cause persistent reductions in the glutamatergic readily releasable pool in rat hippocampal synaptosomes¹⁸. Here complex GCR exposures may portend changes in neurotransmitter availability and/or alter the dynamic response of neurotransmitters available which can greatly impact neurocognitive function. Increased levels of neurotransmitters (Fig. 5) may well predispose certain excitatory and inhibitory circuits to fire asynchronously, which may not lead to overt changes in behavior under basal operating conditions, but under situations of exaggerated spaceflight stress, critical, accurate and timely decision making may be compromised. For example, if we translate increasing physiological stress (i.e., K+ levels) to increasing levels of spaceflight stress, then muted dynamic neurotransmitter responses (particularly with DA, Fig. 6a) could easily compromise reaction times and adaptive decision making, a central role for the PFC. Together these data suggest that interventions targeting DA pathways may help forestall and/or restore homeostatic signaling in the irradiated brain³⁰. These are the first within-subject in vivo studies in mice to document the impact of space-relevant GCR exposures on multiple touchscreen-based neurocognitive assays and associated neurotransmitter dynamics in the PFC.

The PFC consists of a complex network of multiple neurotransmitter systems that modulate each other’s actions and are interconnected across brain regions to maintain normal CNS function, including reward/motivation and attention^27,42,43. Indeed, aberrant neurotransmission in the PFC has been closely linked to various neurological diseases^27,42,43 raising concerns that neurotransmitter disruptions may adversely impact astronaut health/well-being during and after deep space missions. Since it is likely that our measurements reflect global changes in neurotransmitter availability and neurocircuit activity, we embarked on a detailed analyses of pairwise neurotransmitter networks. Pearson’s correlation analysis revealed a profound reorganization of relationships between PFC neurotransmitters following acute or chronic GCR exposure relative to controls. In particular, the pairwise relationships at each K+-evoked concentrations were reorganized and strengthened among the monoamines DA, 5-HT, and NE as well as monoamines and GABA, with some weakening in the pairwise relationships between monoamines and Glu following both GCR paradigms compared to controls (Fig. 7). Collectively, our findings of disruption in monoamine neurotransmitter networks from correlation analyses were corroborated by our predictive machine learning model that uncovered significant differences in PFC monoamine neurotransmitter networks among groups both in terms of connection strength and causation (Fig. 8). More specifically, a pronounced increase in the impact of 5-HT or NE on DA and vice versa coupled with a significant weakening of GABA and Glu’s impact on these monoamines in the irradiated PFC compared to controls. This is in line with the existence of functional interactions between 5-HT and DA where 5-HT either positively or negatively modulates DA levels to impact cognition⁴⁴. The capability of GCR exposures to elicit marked neurotransmitter network reorganization is noteworthy and certainly not restricted to the PFC, and likely represents some of the fundamental biochemical changes that are contributory if not causal to space radiation-induced behavioral and neurocognitive decrements.

The Granger Causality test provided deeper insight into the PFC neurotransmitter network that was altered over protracted acute and chronic post-exposure times. Evident from this analysis is that DA has a significant impact on irradiated neurotransmitter levels (most notably GABA, Fig. 9a), while the other neurotransmitters also impact DA levels (most notably after chronic exposures, Fig. 9b) when compared to controls. The reciprocity of the neurotransmitter network strengthens our arguments for the functional reorganization of these networks following space radiation exposure. While it remains difficult to pinpoint the incipient event that triggered such widespread change, it may also prove relatively unimportant in the context of chronic exposures encountered during extended deep space travel. We are also cognizant that changing levels of K+ might have off-target or more subtle effects on network connectivity that may confound current interpretations, especially at high levels of K+ stimulations that may have induced oscillations in the GLU-GABA networks. To the extent that our machine learning and causality analyses are predictive of network-level reorganization, then it seems plausible that such long-term alterations will be an inevitable consequence of long-duration space travel. What is less certain, is whether such reorganization of neurotransmitter interactions will continue to evolve and if/how they will functionally impact critical decision making and adaptive reasoning under more stressful situations in space.

It is possible that probing neurotransmitter network changes in the PFC at multiple time points (e.g., early, middle, and late) after GCR exposure may reveal further insights into how in vivo neurotransmitter network dynamics are altered over time to impact behavior. Unfortunately, such an endeavor would require a substantially larger cohort of subjects that were simply not available and therefore was beyond the scope of this work. Nevertheless, it is noteworthy, that our data showing long-lasting impact of acute and chronic GCR exposure on behavior and neurotransmitter consequences is highly consistent with the long-term adverse effects on other behavioral and molecular endpoints that have been documented following exposure to space radiation^{7,8,9,10,11,12,21,22,23,24}. Thus, the data presented here sets a solid benchmark for an expanded evaluation of in vivo neurotransmitter network dynamics over time following GCR exposure. From a broader perspective, it is likely that the neurotransmitter network reorganization may be different immediately after GCR exposure vs at later time points. The detrimental effects of both acute or chronic GCR on both touchscreen task and neurotransmitter function may have dissipated and might be even more pronounced early after GCR exposure. However, in the context of a long duration spaceflight, the changes we measure are likely more reflective of the accumulated GCR dose and acute changes (if they do occur) have not been documented and would not translate to late onset radiation-induced cognitive decrements known to take time (months) to manifest after such low dose complex radiation exposures. We therefore contend that our protracted measurements are much more relevant to NASA risk assessments where potential decrements are more likely to manifest on the return from Mars rather than en route to Mars. Thus, while late changes in neurotransmitter levels may reflect the combined impact of all experimental stressors, under our experimental conditions, we believe these are most likely the result of permanent radiation-induced changes to the basal neurochemical state of the irradiated brain.

The fact that low dose, space relevant GCR exposures can elicit functional reorganization of neurotransmitter networks in a single brain region is noteworthy and suggests that similar changes likely transpire throughout the brain after whole body exposures. Indeed, smaller scale changes in pair-wise cell recordings conducted in the hippocampal CA1^17,26,53 have not revealed the more global network level disruptions found after larger scale recordings across multiple brain regions²³ or by LTP mesurements^34,47,51. Importantly, at the space relevant doses used (≤0.5 Gy), cell death in not the cause of the functional changes reported here, or in past studies conducted under similar dosing paradigms^{1,2,3,4,5,6,7,8}. This raises the intriguing idea that the whole brain may be considered the critical target for understanding general CNS radiation effects. Under this conceptual framework, much of current and past studies can be placed in context, where global disruptions in neurotransmitter networks alter synaptic connections across broad brain regions to elicit changes in functional outcomes that cannot be rationalized by smaller scale measurements. Thus, these findings of how the PFC neurotransmitter networks are reorganized by realistic complex GCR simulations provide insights into how such changes impact diverse neurocognitive outcomes and could help re-shape how we envision the impact of space radiation on CNS functionality.

Methods

Subjects and irradiation

All animal procedures were approved by the Institutional Animal Care and Use Committee at McLean Hospital and NASA Space Radiation Laboratory (NSRL) at Brookhaven National Laboratory and were conducted in accordance with guidelines provided by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, Commission on Life Sciences, National Research Council⁵⁴.

Experimentally-naïve male C57BL/6 mice (Jackson Laboratories, Bar Harbor, MA) ~6–9 months old (~30–40 years in human age) and weighing 26–35 g at the time of testing were used. Mice were exposed to a complex 33-beam composition of galactic cosmic radiation simulation (GCRSim) in the NASA Space Radiation Laboratory (NSRL) at Brookhaven National Laboratory (BNL, Upton, NY). The ionic composition of GCRSim contained a mixture of ion species consisting of protons, helium, oxygen, silicon, titanium, carbon, and iron that recapitulated many of the most abundant elements in GCR⁷. Separate groups of mice (n = 8–10/group) were exposed to either an acute or chronic GCRSim irradiation regimen as described previously³⁴. Subjects in the acute group were exposed to the 33-beam GCRSim delivered in a single day over a 2-hr period for a total dose of 40 cGy. Mice in the chronic group were exposed to the same 33-beam GCRSim delivered over ~1 h for 4 weeks (6 days/week) for a total dose of 49.9 cGy (2.08 cGy/day). The total dose delivered is similar to what an astronaut would experience during a voyage to Mars^{1,2,3,4,5,6,7,8}. Individual mice were loosely restrained within acrylic enclosures (3 × 1.5 × 1.5 in) mounted perpendicular to the beam line during the duration of whole-body exposures. Irradiation was overseen by NSRL staff, who also performed all radiation dosimetry and confirmed spatial beam uniformity. The low total GCR doses administered produced no observed changes in body weight. Further details on the operation of the NSRL facility have previously been described⁵⁵. Age-matched control mice underwent all aspects of the study in parallel to those receiving irradiation, were housed under similar conditions and handled equivalently, aside from not receiving GCR irradiation. Following completion of irradiation procedures, all mice were transported from NSRL BNL to McLean Hospital (Belmont, MA) and underwent an 8-week quarantine period upon arrival. Thereafter, mice were housed in a temperature- (~18–23 °C) and humidity (~40–60%)-controlled vivarium with a 12-h light/dark cycle (lights on from 7 AM to 7 PM). Subjects were each fed 2.5–3.0 g of rodent chow post-session and had unrestricted access to water in their home cage. Experimental sessions were conducted five days a week (Mon-Fri). All behavioral and neurochemical testing in mice was conducted over the course of several months (i.e., 6–9 months old) post-irradiation and post-quarantine period at McLean Hospital.

Behavioral procedures

Apparatus

Details and schematics of the touch-sensitive experimental chamber for rodents can be found in Kangas and Bergman⁵⁶. Briefly, a custom-built Plexiglas chamber (35 × 30 × 25 cm) was situated in a sound- and light-attenuating enclosure (60 × 40 × 45 cm). A 17” touch-sensitive screen (1739L, ELO TouchSystems, Menlo Park, CA) was mounted on the right-hand inside wall of the chamber. An infusion pump (PHM-100-5, Med Associates, St. Albans, VT) outside the enclosure was used to deliver a sweetened condensed milk solution into the shallow reservoir (diameter: 3 cm) of an aluminum receptacle (5x4x1 cm) that was mounted slightly above the Plexiglas floor and centered on the left-hand inside wall. A speaker bar (NQ576AT, Hewlett-Packard, Palo Alto, CA) mounted above the touchscreen was used to emit audible feedback. White noise was present outside the chamber to mask extraneous sounds. All experimental events and data collection were programmed in E-Prime Professional 2.0 (Psychology Software Tools, Inc., Sharpsburg, PA).

Touchscreen training

Modified response shaping techniques were first used to train mice to engage with the touchscreen⁵⁷. Subjects first learned to touch a 5 × 5 cm blue square on a black background centered on the touchscreen, with its lower edge 1.5 cm above the floor. This required the mouse to rear on its hind legs and make a response on the box with its paw. Each response was followed with 0.02 ml of a 20% sweetened condensed milk concentration, paired with a 600-ms yellow screen flash and a 440 Hz tone, and followed by a 5-s blackout period. Following reliable responding, presentation of the box alternated in a quasi-random fashion across one of three positions: center, 3.5 cm to the left of center, or 3.5 cm to the right of center. After subjects completed two consecutive 100-trial training sessions within 60 min, the economic demand task commenced.

Economic demand task

Consumption of varying milk concentrations was assessed using a within-session economic demand task to determine the effects of GCR exposure on sensitivity to reward. Demand for 4 different milk concentrations were each tested during 3-session conditions in all mice using the following fixed sequence: 20%, 5%, 50%, and 0% sweetened condensed milk in water. Reward delivery was contingent on responses to a centered 7 × 7 cm green box on a purple background and consisted of 0.02 ml milk paired with a 600-ms yellow screen flash and a 440 Hz tone, followed by a 10-s blackout period. The response requirement increased in an ascending order (2, 5, 10, 30, 50) within-session across five 10-min components that were separated by a 2-min inter-component blackout periods.

Psychomotor vigilance task

After completion of the economic demand task, subjects were exposed to a touchscreen-based variant of a psychomotor vigilance task to evaluate the effects of GCR exposure on attentional processes. This task variant is based on previous studies examining attentional processes by responding to blips on a radar screen that vary in location and intermittency⁵⁸. Trials began with the presentation of a 7 × 7 cm stimulus (pink square) on the screen (black background) in one of six locations (evenly spaced from left to right, 2.5 cm above the floor). The intermittency of stimulus presentation (after a 15-, 30-, or 45-s intertrial interval) as well as the stimulus location were randomized across trials. The duration of stimulus presentation on the first trial of the first session was 10 s for all subjects. Touching the stimulus within this set duration resulted in delivery of 0.02 ml of 20% milk (standard concentration for mice) paired with a 600-ms yellow screen flash and 440 Hz tone. The maximum stimulus duration on the following trial then decreased by 0.25 s. If the subject failed to respond to the stimulus before the duration had elapsed, the stimulus disappeared, and the maximum stimulus duration increased by 0.25 s on the following trial. Daily sessions were comprised of 48 trials and the titrated duration of stimulus presentation on the last trial of a session determined the initial stimulus duration for the next session. There was no upper limit for the stimulus duration, but the lower limit was set to 0.25 s to avoid a 0-s duration stimulus presentation. These titrating contingencies were designed to dial in the attentional capabilities of the subject via their reaction time and ability to maintain task performance across a 48-trial session (i.e., focused vigilance) by making the durations of stimulus presentations briefer as the subject mastered the task. This also allowed for the assessment of potential GCR-treatment effects on both task development as well as asymptotic attentional performance. Each subject completed a total of 15 sessions under these task conditions.