Long-term molecular effects of ¹⁶O-ion exposure in rat brain and implications for space radiation risk

Abstract

Exposure to high-energy charged particles like ¹⁶O ions poses risks to brain function during deep space travel. This study examined long-term neuroinflammatory and synaptic changes in rats 6 months after low-dose total-body ¹⁶O irradiation (1 cGy and 10 cGy), focusing on the frontal cortex, striatum, hippocampus, and thalamus. Using Western blotting, qPCR, and digital PCR, we assessed expression levels of neuroinflammatory (GFAP and IBA1) and synaptic (GAP43, PSD95, SYP, SPINO, and GPHN) markers. Correlation analyses to examine the relationship between the assessed molecular changes and behavioral performance in social odor recognition and psychomotor vigilance tasks were performed. In the hippocampus, 1 cGy exposure resulted in increased IBA1, GAP43, and PSD95, whereas 10 cGy exposure led to decreased GFAP and increased PSD95, SPINO, and GPHN. In the striatum, synaptic markers were elevated after both 1 cGy and 10 cGy exposure, though region-specific differences in gene expression levels were observed (qPCR/dPCR). In the frontal cortex, no changes across any of the targeted markers were detected, except for increased SYP in 1 cGy rats. In the thalamus, GFAP was reduced in 10 cGy rats, while GAP43 was increased in 1 cGy and PSD95 was increased in both 1 cGy and 10 cGy exposed animals. Correlation analyses revealed significant associations between molecular changes and behavior, including a negative correlation between hippocampal SYP protein levels and social memory performance in the 1 cGy group. Taken together, these results suggest dose- and region-specific brain responses to low-dose ¹⁶O ion exposure—a vital component of space radiation—culminating in an enhanced synaptic remodeling and possible neurological alterations. The data also highlight potential molecular mechanisms underlying region-based (i.e., striatum) cognitive vulnerability following low-dose particle exposure.

Introduction

Long-duration space missions will expose astronauts to a unique and chronic form of ionizing radiation: galactic cosmic rays (GCRs). GCRs are composed primarily of protons (~87%) and helium ions (~12%), as well as high-energy, high-charge (HZE) nuclei (~1–2%), including oxygen (¹⁶O), carbon, and iron ions, which readily penetrate shielding and interact with biological tissue, including the central nervous system (CNS)¹. Unlike terrestrial radiation exposures, space radiation involves chronic low-dose exposure to these HZE particles characterized by their high-linear energy transfer (LET), which is known to induce persistent oxidative stress, neuroinflammation, and synaptic dysfunction^2,3. High-LET particles are more harmful to biological matter due to the deep penetration and deposition of energy known to destabilize and damage DNA and other cellular components, which is further compounded by the fact that there has been no development of effective shielding or biological countermeasures for this type of radiation. The effects of HZE exposure have been linked to long-term cognitive deficits, including impairments in memory, attention, and executive function, posing significant risks to astronaut health and mission success^4–8. More specifically, one of the critical behavioral domains negatively affected by space radiation is social recognition memory^9–14. This form of recognition memory is essential for environmental awareness and social behavior in mammals and serves as a sensitive readout for hippocampal and striatal integrity. Attentional performance is also a crucial area of cognitive function that is shown to be affected by space radiation^15–19. The International Space Station (ISS) (https://www.nasa.gov/international-space-station) has established the use of the psychomotor vigilance task (PVT) by astronauts to provide performance feedback and detect changes in alertness in the operational space environment. Recently, worse in-flight PVT performance was found to be associated with higher radiation dose, suggesting that radiation exposure negatively affects sustained attention in astronauts²⁰. Importantly, simulated space radiation exposure (like protons and the GCR simulation) have been shown to cause deficits in reaction times and decreased sustained attention in rats and mice^16–19.

Rodent models exposed to low doses of HZE particles show region-specific neurobiological and behavioral disruptions months after exposure (including impaired operant conditioning, learning and memory, and attention set shifting), making them ideal for investigating the long-term CNS effects of space radiation exposure²¹. Moreover, whereas many of the previous experiments utilized ions like iron (⁵⁶Fe), oxygen ions (¹⁶O) are also a significant component of the GCR spectrum. Indeed, while ¹⁶O ions have a lower LET than ⁵⁶Fe, they are more abundant and have a relatively high fluence per dose, arguably consistent with a hypothesized fluence-dependent signal amplification in brain behavioral networks²², and possibly inducing various neurobiological effects at low doses. For instance, studies have demonstrated that ¹⁶O exposure is linked to hippocampal-mediated memory deficits and altered physiology, including modified spine morphology and synaptic densities^23–26. Further, dose modeling behind realistic thick shielding structures suggests that larger ions like ⁵⁶Fe will fragment into smaller ions, and that ions with Z between 3-10 (which includes ¹⁶O ions) will comprise almost 15-20% of the total dose delivered to the CNS²⁷, underscoring the importance of investigating the effects of ¹⁶O ion exposure on this organ.

Leveraging previous behavioral findings of social odor recognition memory (SORM) deficits in the same animals¹⁰, as well as newly identified dose-specific deficits in sustained attention (measured by the rodent psychomotor vigilance task (rPVT)), this study aimed to determine if these behavioral changes were indeed associated with region-specific molecular alterations. We assessed if a set of brain-related proteins (neuroinflammatory: glial fibrillary acidic protein (GFAP), ionized calcium-binding adapter molecule 1 (IBA1); and synaptic markers: synaptophysin (SYP), spinophilin (SPINO), postsynaptic density protein 95 (PSD95), gephyrin (GPHN), growth-associated protein 43 (GAP43)) from specific brain regions (frontal (FCtx), striatum (Str), hippocampus (Hippo), thalamus (Thal)) of ¹⁶O ion exposed (1000 MeV/n; 1 cGy and 10 cGy) vs. non-exposed (control) rats showed protein abundance level changes (Western Blotting (WB) analyses). Moreover, we associated WB data with corresponding mRNA expression measured by quantitative polymerase chain reaction (qPCR) and digital polymerase chain reaction (dPCR) methods using tissues from the same brain regions to possibly reveal region- and dose- specific gene activation differences. Finally, based on the previously reported SORM deficits and more recently analyzed rPVT findings from the same animals, we assessed if those behavioral changes were associated with molecular changes of any of the selected markers in any of the specific regions of the brain examined. All molecular and behavioral analyses were performed at 6 months post 1 cGy or 10 cGy total-body ¹⁶O exposure.

Results

To comprehensively characterize the long-term sequelae of low-dose ¹⁶O irradiation, we first summarize the behavioral outcomes from the previously reported deficits in SORM¹⁰ and the newly reported rPVT data. Subsequently, we detail the dose- and region-specific alterations in neuroinflammatory and synaptic protein levels and gene expression (summarized in Table 1). Then, we integrate these datasets through correlation analyses to determine the relationship between the observed molecular changes and behavioral performance.

Table 1.

Summary of hippocampal and striatal Western blot and PCR results (6 months post irradiation)

	Hippocampus				Striatum
Rad dose	Marker	WB	qPCR	dPCR	Marker	WB	qPCR	dPCR
1 cGy	GFAP	ns	ns	ns	GFAP	ns	ns	Increased (p ≤ 0.001)
	IBA1	Increased (p ≤ 0.05)	ns	ns	IBA1	ns	ns	Increased (p ≤ 0.01)
	GAP43	Increased (p ≤ 0.01)	ns	ns	GAP43	Increased (p ≤ 0.01)	ns	ns
	PSD95	Increased (p ≤ 0.01)	ns	ns	PSD95	Increased (p ≤ 0.0001)	ns	ns
	SPINO	ns	ns	ns	SPINO	Increased (p ≤ 0.05)	Decreased (p ≤ 0.01)	ns
	SYP	ns	ns	ns	SYP	Increased (p ≤ 0.01)	Decreased (p ≤ 0.05)	ns
	GPHN	ns	ns	ns	GPHN	Increased (p ≤ 0.05)	ns	Increased (p ≤ 0.01)
10 cGy	GFAP	Decreased (p ≤ 0.01)	ns	ns	GFAP	ns	ns	ns
	IBA1	ns	ns	ns	IBA1	ns	ns	Increased (p ≤ 0.05)
	GAP43	ns	ns	ns	GAP43	ns	ns	ns
	PSD95	Increased (p ≤ 0.05)	ns	ns	PSD95	Increased (p ≤ 0.05)	ns	ns
	SPINO	Increased (p ≤ 0.05)	ns	ns	SPINO	Increased (p ≤ 0.05)	ns	ns
	SYP	ns	ns	ns	SYP	Increased (p ≤ 0.001)	ns	ns
	GPHN	Increased (p ≤ 0.05)	ns	ns	GPHN	Increased (p ≤ 0.05)	ns	Increased (p ≤ 0.05)

Effects of 16O exposure on memory and attention performance

Social object recognition memory

Memory deficits in these rats were previously reported¹⁰ (results not shown here), with rats in both the 1 cGy and 10 cGy exposure groups displaying discrimination indices significantly less than the sham control rats at 6 months post exposure. ¹⁶O exposure caused similar levels of recall memory impairment for both exposure levels during the recognition test phase, where irradiated rats did not demonstrate a preference for the “novel 2” odor over the “novel 1” odor presented during the habituation phase 24 hours earlier. These previously reported discrimination indices are newly utilized here in the correlation analyses described below.

Rodent Psychomotor Vigilance Task (rPVT): ¹⁶O ion exposure altered lapses in attention and false alarm responses in the 10 cGy exposed subjects. While the average lapses emitted during the session did not differ between groups over time (p = 0.079), a response-stimulus interval (RSI) analysis showed that 10 cGy rats increased lapses emitted at the shortest response interval (RSI = 4, p = 0.008), with a trend for a similar increase at the next RSI (RSI = 5, p = 0.078). The 10 cGy exposed rats significantly increased false alarm responses over time following radiation exposure compared to controls (p < 0.001), whereas the control (non-exposed) and 1 cGy exposed rats showed downward trends indicative of fewer false alarm responses over time. In contrast to the increase in total false alarms over time, the pattern of false alarm responses emitted across the different RSIs did not change following exposure in the 10 cGy group. Other performance parameters on the rPVT did not differ between the groups (Fig. 1).

Fig. 1. Effects of ¹⁶O on rPVT at 6 months post irradiation.

Mean daily percent lapses of attention (top) and false alarms (bottom) emitted in the rPVT over time, in addition to the lapses emitted at each response-stimulus interval (RSI) of the rPVT at baseline (2 weeks prior to radiation) and at 6 months following radiation (last 2 weeks). Lapses over time do not differ between groups, but 10 cGy exposed rats emitted more lapses at the shortest RSI (**p = 0.008), with a trend for more lapses at the second RSI (^p = 0.078), as well. 10 cGy exposed rats displayed increased impulsive responding over time following radiation compared to sham control rats (**p < 0.001), but the pattern of false alarm responses emitted at each RSI was not affected by radiation exposure.

Region and dose-specific changes in inflammatory and synaptic protein levels

WB results for all regions are summarized in Fig. 2. Representative full blots are provided in Supplementary Fig. 1. With respect to the neuroinflammatory markers, IBA1 increased in the Hippo of 1 cGy rats, while GFAP decreased in the Hippo of 10 cGy rats. GFAP levels were also decreased in the Thal of 10 cGy rats (Fig. 2A). Regarding synaptic markers, an increase of SYP levels was detected in the FCtx of 1 cGy rats (Fig. 2B); in the Str, levels of PSD95, SPINO, SYP and GPHN were all increased in both 1 cGy and 10 cGy rats, while GAP43 was only increased in the 1 cGy rats (Fig. 2B). In the Hippo, GAP43 and PSD95 were increased in 1 cGy rats, while PSD95, SPINO and GPHN were increased in 10 cGy rats (Fig. 2B). In the Thal, GAP43 and PSD95 increased in 1 cGy rats, but only PSD95 was increased in 10 cGy animals (Fig. 2B).

Fig. 2. ¹⁶O affects protein abundance in a region and dose-specific manner 6 months post irradiation.

Western blotting identified a significant increase in IBA1 in the Hippo after 1 cGy exposure, and a significant decrease in GFAP in both the Hippo and Thal after 10 cGy (A). Among synaptic targets that were assessed (B), FCtx only shows a significant increase in SYP after 1 cGy. Str shows a significant increase after both 1 cGy and 10 cGy exposure for SYP as well as GPHN, SPINO, PSD95, and a significant increase in GAP43 only after 1 cGy. Hippo and Thal also showed a significant increase in GAP43 after 1 cGy. They also showed significant increases in PSD95 after both 1 and 10 cGy exposure, while Hippo showed increases in SPINO and GPHN after only 10 cGy exposure. (n = 5–6 rats per group; mean ± SEM).

Hippocampal and striatal mRNA expression levels

Due to the observation that the Str and Hippo obviously exhibit higher sensitivity to ¹⁶O exposure, subsequent molecular analyses focused only on these two brain regions. Following normalization with ERK2 (a.k.a. MAPK1; see Methods), qPCR analysis revealed no significant changes in hippocampal gene expression for any selected targets (GFAP, IBA1, GAP43, PSD95, SPINO, SYP, and GPHN) (Fig. 3A, B). In the Str, a decrease in SPINO expression was observed after 1 cGy exposure (Fig. 3C), but no changes were detected at 10 cGy (Fig. 3D). For absolute quantification measures, additional dPCR analyses were performed. In the Hippo, dPCR data showed no significant changes in gene activation in the Hippo for either 1 cGy or 10 cGy exposed rats (Fig. 4A). In the Str, dPCR showed a significant increase in GFAP expression in 1 cGy rats (Fig. 4B). Moreover, IBA1 and GPHN expression levels were increased in both 1 cGy and 10 cGy rats (Fig. 4B).

Fig. 3. qPCR mRNA expression in the Str was altered 6 months post irradiation, only after 1 cGy ¹⁶O exposure.

There were no detected changes in hippocampal mRNA expression for any of the tested targets (GFAP, IBA1, GAP43, PSD95, SPINO, SYP, and GPHN, all normalized to ERK2) after 1 cGy (A) or 10 cGy exposure (B). In the Str, qPCR detected a significant decrease in SYP and SPINO mRNA expression after 1 cGy exposure (C), but no changes were detected following 10 cGy exposure (D) (n = 5–6 rats per group; mean ± SEM).

Fig. 4. dPCR mRNA expression in the Str altered 6 months post irradiation.

No significant changes in mRNA expression were detected in the Hippo after 1 cGy or 10 cGy for any of the tested targets (GFAP, IBA1, GAP43, PSD95, SPINO, SYP, GPHN, ERK2) (A). After 1 cGy exposure, Str levels of GFAP significantly increased, while levels of IBA1, GPHN and ERK2 all increased after both 1 cGy and 10 cGy exposure (B). Following correlation analyses, a matrix of p-values showing correlations between mRNA expression and behavior is presented for the Hippo (C) and Str (D), with significant correlations indicated by bold, italicized, right-justified text. Strength of correlation is indicated by a color scale, with blue indicating positive and red indicating negative correlations. 1 cGy values are presented in the lower left half of the matrix, and 10 cGy values are presented in the upper right half of the matrix. (See Supplementary Figs. 6, 7 for complete p value and R² coefficients) (n = 5–6 rats per group; mean ± SEM).

Correlations between molecular targets and behavioral outcomes

Correlation analyses were performed between the assessed molecular targets and behavioral performance (SORM discrimination index and rPVT overall performance score (OPS)). The findings are presented as matrices of p-values with R² represented in color-coded cells (Fig. 4C, D (dPCR-behavioral correlations) and Fig. 5C, D (WB-behavioral correlations)). Behavioral scores (SORM and rPVT) were positively correlated with each other among animals exposed to 1 cGy, while no correlations were identified between memory (SORM) and attention (rPVT) performance in 10 cGy. SORM performance was negatively correlated with SYP protein levels in the Hippo of 1 cGy rats (Fig. 5C, WB-behavioral correlations). SORM was also positively correlated with IBA1 mRNA expression in the Hippo in 10 cGy rats (Fig. 4C, dPCR-behavioral correlations). Moreover, in the Str, rPVT was positively correlated with GPHN protein levels in 10 cGy rats (Fig. 5D, WB-behavior correlations). Neither behavioral assessment correlated with the mRNA expression of any targets in the Str (Fig. 4D, dPCR-behavior correlations). A positive correlation was observed between IBA1 and GFAP protein levels in both Hippo and Str for 1 cGy- exposed animals (Fig. 5C, D). A positive correlation was also noted among synaptic mRNA expressions in the Hippo and Str after both 1 cGy and 10 cGy exposure (Fig. 4C, D, dPCR-behavioral correlations).

Fig. 5. Western blot protein levels in the Hippo and Str altered 6 months post irradiation.

After 1 cGy exposure, Hippo levels of IBA1 increased along with synaptic proteins GAP43 and PSD95; after 10 cGy exposure, we identified decreased GFAP as well as increased PSD95, SPINO and GPHN (A). After 1 cGy exposure, Str levels of GAP43 increased, while levels of PSD95, SPINO, SYP, and GPHN all increased after both 1 cGy and 10 cGy exposure (B). Following correlation analyses, a matrix of p values showing correlations between protein levels and behavior is presented for the Hippo (C) and Str (D), with significant correlations indicated by bold, italicized, right-justified text. Strength of correlation is indicated by a color scale, with blue indicating positive and red indicating negative correlations. 1 cGy values are presented in the lower left half of the matrix, and 10 cGy values are presented in the upper right half of the matrix. (See Supplementary Figs. 4 and 5 for complete p value and R² coefficients) (n = 5–6 rats per group; mean ± SEM).

Qualitative assessment of microglial and astroglial morphology

A qualitative neuropathological assessment was performed on two randomly chosen sections per each animal for each condition (Control/not-exposed, 1 cGy, 10 cGy exposure) at low and high magnification to assess possible overt cell morphology changes or reactivity responses. The digitally-scanned qualitative inspection of sections IHC-stained for GFAP revealed no obvious evidence of astrogliosis or astrocytic hypertrophy in the Hippo or its subregions across the three conditions (Fig. 6). Similarly, evaluation of IHC-IBA1 staining did not demonstrate overt differences in microglial reactivity or morphology among Control/not-exposed, 1 cGy, and 10 cGy ¹⁶O-exposed rats (Fig. 7). In the absence of clear astroglial or microglial cells’ morphological changes, no additional histological quantifications (i.e., stereology-based measurements) were performed.

Fig. 6. GFAP reactivity in the hippocampus at 6 months post-O16 irradiation (0 cGy, 1 cGy, and 10 cGy).

This figure shows the immunohistochemical (IHC) staining for glial fibrillary acidic protein (GFAP) in the rat hippocampus six months after exposure to ¹⁶O ion irradiation. Among others, GFAP is a marker for astrocytes. A–C Show the hippocampus from a sham-irradiated (0 cGy) rat, while (D–F) show sections from rats exposed to 1 cGy and 10 cGy of radiation, respectively. The IHC staining reveals no obvious astrogliosis or astrocytic hypertrophy across all three conditions. In (A), the right bottom corner, the scalebar corresponds to 200 µm for all panels.

Fig. 7. IBA1 reactivity in the hippocampus at 6 months post-O16 Irradiation (0 cGy, 1 cGy, and 10 cGy).

This figure displays the immunohistochemical (IHC) staining for ionized calcium-binding adapter molecule 1 (IBA1) in the rat hippocampus 6 months after exposure to ¹⁶O ion irradiation. IBA1 is a marker for microglia. A–C Show the hippocampus from a sham-irradiated (0 cGy) rat, while (D–F) show sections from rats exposed to 1 cGy and 10 cGy of radiation, respectively. The staining indicates no significant microglial reactivity differences among the 0 cGy, 1 cGy1 cGy, and 10 cGy10 cGy groups. In (A), the right bottom corner, the scalebar corresponds to 200 µm for all panels.

Discussion

This investigation demonstrates that low-dose ¹⁶O ion irradiation induces persistent, region-specific alterations in neuroinflammatory and synaptic protein levels in the rat brain 6 months post exposure. These brain region-based neuroinflammatory, synaptic and molecular-behavioral correlation findings are novel and further support the current body of research indicating that the severity and persistence of cognitive deficits following cosmic radiation exposure are dependent on the linear energy transfer (LET) of specific high-energy, high-charge (HZE) particles²⁸. Unlike low-LET radiation (e.g., X-rays), high-LET particles such as ¹⁶O induce complex DNA damage and oxidative stress that are difficult to repair, leading to lasting neurobiological perturbations^29–31. Our data revealed that even at low doses (1 cGy and 10 cGy), ¹⁶O ion exposure triggers a nonlinear response characterized by distinct molecular signatures and degrees of changes in different brain regions (i.e., frontal cortex, striatum, hippocampus, and thalamus). Moreover, although the small sample size of our experiments requires a cautious interpretation of correlations between our biological targets, it is important to point out the positive correlation between IBA1 and GFAP protein levels in the hippocampus and striatum of 1 cGy exposed animals, molecular signals which are possibly indicative of a long-term inflammatory response.

Furthermore, among the primary findings of this study, there is the discordance observed between the expression levels of glial markers measured by WB and dPCR methods vs. the normal morphological appearance of glial cells assessed by IHC techniques. In fact, we observed increased IBA1 levels in the hippocampus at 1 cGy, and an increased IBA1 mRNA expression in the striatum at 1 cGy and 10 cGy doses, yet IHC analysis revealed no overt signs of microglial hypertrophy or activation, even at higher magnification. While this discrepancy was based only on qualitative neuropathological evaluations, it is likely that this reflects the different biological pools measured by these different assays. IHC primarily visualizes the structural distribution of proteins, whereas WB quantifies the total soluble protein fraction. However, the upregulation of IBA1 protein expression and mRNA in the absence of obvious morphological changes suggests that microglial cells may be in a “primed” or sensitized state. Indeed, in this state, microglial cells upregulate functional proteins in response to low-level stress without transitioning to a fully ameboid, phagocytic phenotype. This explanation aligns with prior studies showing that HZE radiation can induce chronic neuroinflammation and alter microglial physiology without classical morphological activation³².

Likewise, we observed decreased GFAP protein levels in the hippocampus and thalamus following 10 cGy exposure, despite stable GFAP mRNA levels and unchanged astrocytic morphology. As is known, GFAP exists in insoluble filamentous forms (cytoskeleton) and soluble pools involved in remodeling. Consequently, the reduction in soluble GFAP detected by WB could indicate a disruption in cytoskeletal turnover or a specific vulnerability of non-filamentous GFAP pools and not necessarily associated with a corresponding detectable enhanced gene activation. It would also be relevant to acknowledge that GFAP is normally expressed in neural stem and progenitor cells (NSPCs) in the subgranular zone, as well³³. The observed reduction in hippocampal GFAP could partially reflect a loss or alteration of the radiosensitive stem cell population rather than solely or mainly of the number of the mature astrocytic cell population or reactivity. Future studies utilizing co-labeling (e.g., Sox2 and Nestin) will be necessary to distinguish amongst all GFAP-positive cell populations that are particularly affected by low doses of ¹⁶O or other heavy ion exposure.

Importantly, our results highlight the striatum as a region of higher vulnerability and responsiveness to ¹⁶O ion irradiation. We observed, in fact, a robust and consistent upregulation of multiple synaptic proteins (GAP43, PSD95, SPINO, SYP, and GPHN) in this brain region at either 1 cGy or 10 cGy exposure. These findings were corroborated by dPCR data showing trends of increased mRNA expression levels for most of these markers, with significant levels for GPHN in 1 cGy and 10 cGy exposed rats. Functionally, the striatum is a critical region for motor planning, habit formation, and decision-making, and spaceflight stressors, including HZE ions, have been linked to striatal dysfunction and dopaminergic alterations^18,34–36. At a more speculative level, the upregulation of different synaptic markers 6 months after ¹⁶O ion exposure may suggest a long-term, possibly compensatory, remodeling of striatal circuits. More specifically, the persistent elevation of these proteins could represent a “synaptic scaling” response, where neural circuits increase synaptic strength or number, to compensate for radiation-induced neuronal damage or functional deficits. Interestingly, in the hippocampus, the molecular response was less uniform among those synaptic markers in comparison to the striatum. Indeed, while 1 cGy exposure increased presynaptic (GAP43) and postsynaptic (PSD95) markers, 10 cGy exposure resulted in a distinct profile characterized by elevated SPINO and GPHN. The upregulation of GPHN is of particular interest. GPHN is a scaffolding protein mainly located at inhibitory synapses and regulates GABAergic synaptic transmission. A GPHN increase suggests an alteration in the excitatory/inhibitory balance of hippocampal circuits. This last finding aligns well with previous reports describing altered oscillatory activity and enhanced inhibitory signaling following mixed-field GCR exposure³⁷.

In terms of molecular-behavioral correlations, we identified dose-dependent changes between specific region-based molecular alterations and the assessed behavioral performance. For example, at 1 cGy, animals exhibited deficits in SORM but preserved sustained attention (rPVT). Interestingly, though, SORM performance in the 1 cGy group was negatively correlated with hippocampal synaptophysin (SYP) levels. This negative correlation seems to imply that animals with the highest upregulation of synaptic proteins performed most poorly, suggesting that this synaptic remodeling may be a maladaptive phenomenon or be indicative of a more severe underlying injury that the brain is attempting to repair.

In contrast, at 10 cGy, we detected deficits in rPVT performance, characterized by increased impulsive responding (false alarms) and lapses in attention at short response intervals. Intriguingly, the performance on the rPVT was positively correlated with striatal GPHN levels. We hypothesize that, given the role of the striatum in inhibitory control, the upregulation of inhibitory synaptic markers like GPHN may be a signal of compensatory mechanisms attempting to restore proper gating of motor responses. The failure of these mechanisms to fully preserve behavior (as deficits were still observed) highlights the limits of this plasticity.

Importantly, the dissociation between observed molecular findings of 1 cGy and 10 cGy animals—where 1 cGy animals showed correlated behavioral and molecular changes, while 10 cGy animals showed less coordinated responses—challenges the commonly accepted linear no-threshold model of postulated radiation risk^38,39. The distinct molecular profiles detected suggest that 1 cGy may trigger specific adaptive or maladaptive plasticity pathways that may be already overwhelmed, or almost totally dysregulated, at the higher 10 cGy dose.

Some limitations need to be considered when interpreting these new findings. First, the dose rate used in this study is orders of magnitude higher than the low dose rate characteristic of the space environment, however this is common for studies of simulated space radiation exposure due to logistical constraints of the BNL-NSRL (Brookhaven National Laboratory, NASA Space Radiation Laboratory, U.S. Department of Enegy). While the total doses (1 cGy and 10 cGy) are relevant to mission totals, the acute delivery may elicit biological responses distinct from chronic, low-fluence exposure. Consequently, caution is warranted when extrapolating these results directly to astronaut risk models. Second, while we observed “multidirectional” effects across regions, the lack of a linear dose-response relationship introduces complexity in interpretation. It is possible that the variations observed between 1 cGy and 10 cGy reflect the inherent variability of low-dose radiation effects rather than a strictly deterministic mechanism. As noted by others, low-dose outcomes can be stochastic, and the small sample size (n = 5–6 for molecular assays) may influence the statistical robustness of these correlations. Third, the discordance between protein (WB) and gene expression (qPCR/dPCR) data emphasizes the complexity of radiation responses. For example, in the striatum, we observed increased GFAP mRNA via dPCR despite no change in protein levels. However, these discrepancies should not be interpreted as “post-transcriptional modifications” per se, but rather as evidence of complex regulatory dynamics, such as differences in mRNA stability, translation rates, or protein turnover. Additionally, our evaluation of “housekeeping” or “reference” genes revealed that radiation affects the expression of canonical controls like GAPDH, necessitating the use of absolute quantification methods like dPCR to ensure accuracy^40,41. Finally, although we identified correlations between molecular markers and behavior, these associations do not establish causality. The concurrent changes in synaptic proteins and cognitive performance may be downstream effects of a common upstream driver, such as oxidative stress or DNA damage, which were not directly measured in this study at the 6-month time point.

In summary, this study provides evidence that acute, low-dose ¹⁶O ion irradiation results in lasting, region-specific changes in the adult rat brain that are associated with persistent cognitive deficits. We identified a persistent remodeling of synaptic architecture, particularly in the striatum, and alterations in glial protein expression that occur in the absence of overt morphological inflammation. These molecular shifts correlate with specific deficits in social recognition memory and sustained attention, suggesting that even low doses of HZE particles can induce long-term reorganization of neural circuits. These findings underscore the importance of investigating LET-dependent and region-specific risks to the CNS to develop targeted mitigation strategies for future deep space missions.

Methods

Animal and irradiation procedures

A detailed description of animal and radiation procedures can be found in ref. ¹⁰. Briefly, 10–12-week-old male rats (Long-Evans, Envigo, East Millstone, NJ, USA) were procured and singly housed on a reverse light cycle with access to enrichment toys. Water was provided ad libitum, and measured chow was fed each day to maintain 90% free feeding weight.

At 5–6 months old, rats were shipped to BNL-NSRL (https://www.bnl.gov) seven days prior to irradiation. All rats were irradiated on the same day, receiving 1 cGy or 10 cGy whole body exposure to ¹⁶O ions (1000 MeV/n) generated at the BNL-NSRL facility. Awake rats were placed in plexiglass holders (~20 × 10 × 10 cm) and taken into the beam line (Fig. 8). Ions of this energy have an average linear energy transfer (LET) of 14 keV/um, with a mean range in water of ~81 cm. Rats were positioned in the beam line such that exposures were performed in the entrance (plateau) region of ¹⁶O ions, avoiding deposition of the Bragg peak into rat tissue. Dose rates ranged from 1.78–4.09 cGy/min, with the actual delivered dose varying no more than 0.1 cGy relative to the target dose. A sham control group experienced the same procedures, except they were not brought into the beam line. Following exposure, rats were returned to Johns Hopkins University (JHU) for follow-up testing.

Fig. 8. Diagram of irradiation setup.

Rats were irradiated awake in plexiglass holders positioned in the beam line (represented by the arrow). Created in BioRender. Davis, C. (2025) https://BioRender.com/p7xi7ei.

Laboratory animal care and handling was conducted in compliance with Public Health Service (PHS) Policy on the Humane Care and Use of Laboratory Animals, the NIH Guide for the Care and Use of Laboratory Animals, ARRIVE guidelines and all applicable Federal regulations governing the protection of animals in research, and the Institutional Animal Care, and Use Committee of the Johns Hopkins University (RA17M78) approved all procedures. All animal work was approved by the Institutional Animal Care and Use Committee at the Uniformed Services University (USU, Bethesda, MD, USA) and the Johns Hopkins University in compliance with the PHS Policy on Humane Care and Use of Laboratory Animals, the NIH Guide for the Care and Use of Laboratory Animals, and all applicable Federal regulations governing the protection of animals in research.

Tissue collection

Six months after irradiation, rats (n = 5–6/group) were euthanized by live decapitation, brains removed and the right hemispheres immersed in 30% sucrose for 3 h at 4 °C (the left hemispheres were used in ref. ¹⁰), after which they were embedded in Tissue-Tek OCT compound (product#4583, Sakura Finetek USA, Inc., Torrance, CA) and flash frozen in liquid nitrogen for storage at -80°C until processing. The brain was microdissected into discrete regions (including frontal cortex (FCtx), striatum (Str), thalamus (Thal), and hippocampus (Hippo)) from 100 µm thick coronal sections cut on a cryostat from fresh frozen brains, utilizing a rat brain atlas to identify precise anatomical delineations. Dissected regions from each animal were collected (~100 mg per region) and divided for WB and RNA procedures.

Protein extraction and Western blot (WB) procedures

For WB analyses, the collected tissue was homogenized using glass dounce homogenizers with ice-cold lysis buffer (1 mL/100 mg tissue) containing: 50 mM Tris-HCl (pH = 8), 1% Igepal, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1 mM NaF, 1:100 protease inhibitor cocktail (Sigma-Aldrich, P2714, St. Louis, MO, USA). Following centrifugation for 20 min at 12,000×g, supernatants were collected and aliquoted for freezing at −80 °C. Micro BCA assay (Thermo Fisher Scientific, cat. #23235, Waltham, MA, USA) was used to determine total protein content from each brain region.

Protein samples were mixed with dH₂O and 10% by volume 10x NuPAGE Sample Reducing Agent (Life Technologies, cat#2353153, Carlsbad, CA, USA), 25% by volume 4x Laemmli Sample Buffer (Bio-Rad Laboratories, Inc., cat#1610747, Hercules, CA, USA), and a calculated volume of protein to achieve 2 µg/µl concentration, denatured for 10 min at 70 °C. For all brain regions, 10 µg of protein per sample (Control n = 5, 1 cGy n = 6, 10 cGy n = 5) were loaded into Novex Nupage 4–12% Bis-Tris Gels (Life Technologies, cat#NP0329, Carlsbad, CA, USA) for 30 min electrophoresis at constant 200 V, with 2.5 µl Precision Plus Protein Dual Color Standards (Bio-Rad Laboratories, cat#1610374, Hercules, CA, USA) loaded into one well per gel. Gels were transferred using the iBlot2 dry transfer method (Life Technologies, cat#IB21001, Carlsbad, CA, USA) onto PVDF membranes and then rinsed with dH2O and incubated at RT for 10 min with No-Stain Protein Labeling Reagent (Thermo-Fisher, cat#A44449, Waltham, MA, USA) according to the manufacturer’s instructions for total protein fluorescent visualization. Membranes were then rinsed with dH₂O and blocked for 1 h at room temperature in 5% milk in 1x TBST.

Primary antibodies (Table 2) were diluted in 5% milk in 1x TBST to appropriate working concentrations and incubated on the membranes at 4 °C overnight. After rinsing the membranes 3 × 5 min in 1x TBST, they were incubated for 2 h at RT in appropriate HRP-tagged secondary antibodies (Table 2) diluted 1:2000 in 5% milk in 1x TBST. Membranes were rinsed in 1x TBST for 3 × 5 min and in 1x TBS for 1 × 5 min, then incubated for 1 min in chemiluminescent substrate (SuperSignal West Pico Chemiluminescent Substrate, Thermo Fisher Scientific, cat#34577, Waltham, MA, USA) and imaged on iBright Imaging System (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). Densitometry was performed using iBright Analysis Software (Thermo Fisher Scientific, Waltham, MA, USA), with all protein intensities normalized to total protein signal intensity.

Table 2.

Western blot antibodies

Antibody	Concentration	Catalog #	Company
Primary antibodies
IBA1 (AIF1)	1:1000	Ab178847	Abcam
GFAP	1:10,000	NCL-L-GFAP-GA5	Leica
GAP43	1:40,000	Ab75810	Abcam
PSD95 (DLG4)	1:1000	75-028	NeuroMab
GPHN	1:2000	MA5-25368	Thermo
SYN	1:2000	Ab8049	Abcam
SPINO (PPP1R9B)	1:1000	14136	Cell Signaling
MAPK (ERK1/ERK2)	1:2000	M5670	Millipore Sigma
Secondary antibodies
HRP-tagged Goat anti-Mouse	1:2000	ab97040	Abcam
HRP-tagged Goat anti-Rabbit	1:2000	ab97080	Abcam

It is important to note that during our experiments, the housekeeping/reference genes (or proteins) were first assessed as an alternative (or rather parallel analysis) to normalization with total protein in case irradiation in the experiment had a significant impact on total protein levels. Our assessment revealed that radiation does affect many of the canonical reference proteins (like GAPDH), and importantly, that this effect varies by dose and region. We were unable to replicate the assessment for qPCR due to logistical constraints, and so we decided to select a reference gene that demonstrated stable protein levels in the regions of interest (hippocampus and striatum) in the hope that this would have a greater likelihood of mRNA stability, which unfortunately, we determined was likely not the case. Thus, we included our secondary application of dPCR, in which it was not necessary to include a reference gene for normalization.

Considering the challenge of endogenous controls in radiation research, our efforts to accurately assess protein and gene expression levels have led us to explore and acknowledge certain important points: the idea of housekeeping/reference genes has increasingly come into question in the literature, particularly for experimental conditions like radiation, which is known to cause DNA damage^42,43. We confirmed that many of these endogenous proteins and genes change differentially depending on radiation dose. Perhaps even more important is the consideration that this is region-dependent, with these proteins/genes showing distinct alterations in expression levels even just in the four regions that we assessed. It has become progressively more apparent that accurate measurement of mRNA expression, particularly but not exclusively in experiments involving radiation exposure, requires either absolute quantification (to the extent that this is possible through dPCR) or extensive verification of endogenous controls for every condition and brain region being assessed. Indeed, proteins that previously would have been considered unsuitable could make ideal normalization candidates if their stability can be appropriately verified. It is crucial for the field moving forward to carefully consider previous findings that may benefit from reinterpretation in this light. WB findings must also be cautiously interpreted, where experimental conditions like radiation may influence proteins commonly used for normalization or even total protein levels.

RNA isolation, quantitative (qPCR), and digital polymerase chain reaction (dPCR) procedures

RNA was extracted from the collected tissue samples with miRNeasy Tissue/Cells Advanced Mini Kit (Qiagen, cat#217604, Hilden, Germany) according to the manufacturer’s directions and retrotranscribed (1000 ng RNA per sample) to cDNA with High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Thermo Fisher Scientific, cat#4368814, Waltham, MA, USA). Quantification of cDNA was performed using TaqMan gene expression assays (Table 3) following TaqMan Fast Advanced Master Mix for qPCR protocol (Applied Biosystems, Thermo Fisher Scientific, cat#4444963, Waltham, MA, USA) and analyzed through qPCR on a QuantStudio 7 Pro Real-time PCR System (Thermo Fisher Scientific, Waltham, MA, USA), normalized to MAPK1 as an endogenous control gene. Samples were run in triplicate, and Ct values were analyzed using the ΔΔCT method. Differences in RNA expression were evaluated by a two-tailed Student’s t-test of mean ΔΔCT values, with statistical significance p < 0.05. For our normalization strategy, we assessed potential reference proteins using WB to identify candidates that remained stable across irradiation conditions. We determined that Mapk1 (ERK2) protein levels were stable in the hippocampus and striatum after 1 cGy and 10 cGy exposure (Supplementary Fig. 2). Consequently, Mapk1 was selected as the reference gene for qPCR normalization. We acknowledge that radiation can induce complex genomic instability, potentially affecting the expression of canonical reference genes^42–44. Due to these constraints, we also employed digital PCR (dPCR), which allows for absolute quantification without reliance on endogenous normalization controls.

Table 3.

TaqMan gene expression assays for PCR

Gene	Dye	Assay ID	Catalog #
AIF1 (IBA1)	FAM	Rn03993468_m1	4331182
GFAP	FAM	Rn01253033_m1	4331182
GAP43	FAM	Rn01474579_m1	4331182
DLG4 (PSD95)	FAM	Rn00571479_m1	4331182
GPHN	FAM	Rn00575867_m1	4331182
SYP	FAM	Rn01528256_m1	4331182
PPP1R9B (SPINO)	FAM	Rn00585021_m1	4331182
MAPK1 (ERK2)	FAM (qPCR only)	Rn00671828_m1	4351370
	VIC (dPCR only)	Rn00671828_m1	4448485

The same cDNA samples were used with TaqMan gene expression assays (Table 3) for quantification and analysis with digital PCR (dPCR) through the Johns Hopkins Genetic Resources Core Facility (GRCF), SOM, JHU; RRID:SCR_018669. One water control was used with each multiplex on each plate. Quality and proper dilution of the samples for each assay were assessed. It was determined that 1:20 dilutions were appropriate for all assays except for GPHN and IBA1 (AIF1), which required diluting 1:10. Samples were run on the QIAcuity 8 instrument, firmware version 2.5, analysis software version 2.5.0. Each FAM-labeled assay was run in multiplex with MAPK1 (VIC labeled). Each sample was run in triplicate, loading 2 µl into 10.2 µl of mastermix. The 12.2 µl of mastermix was then loaded into the nanoplate, and samples were thermocycled using the standard probe cycling conditions in the QIAcuity user manual: hold 95 °C for 120 s, and then 40x cycle of 95 °C for 15 s and 60 °C for 30 s.

Immunohistochemistry procedures

Sagittal tissue blocks from each animal were uniformly processed using an automated tissue processor (ASP 6025, Leica Biosystems, Nussloch, Germany). After processing, each block was embedded in paraffin and serially sectioned at 5 µm thickness. Two randomly selected sections per block were used for immunohistochemistry (IHC). IHC was performed with a Leica Bond III automated immunostainer and a diaminobenzidine chromogen detection system (DS9800, Leica Biosystems, Buffalo Grove, IL). The following antibodies were applied: anti-glial fibrillary acidic protein (GFAP; mouse anti-rat monoclonal antibody GA5, 1:250; Bond heat-induced epitope retrieval, 10 min; PA0026, Leica Biosystems, Wetzlar, Germany); anti-ionized calcium-binding adapter molecule 1 (IBA1; rabbit polyclonal, 1:100; epitope retrieval, 10 min; 016-20001, FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan).

Stained sections were scanned with the Aperio AT2 whole-slide scanner (Leica Biosystems, Richmond, IL) and archived in the Biolucida system (version 2017, MBF Bioscience, Williston, VT, USA) for subsequent assessment. Neuropathological evaluation was performed with Aperio ImageScope (version 2016, Leica Biosystems). A blinded qualitative neuropathological evaluation was performed by an expert neuropathologist (DI). Analyses focused mainly on the hippocampus as a probable region to identify morphological evidence of astroglial and microglial reactivity. Specifically, sections were examined for signs of astrocytic hypertrophy, scarring (astrogliosis), or microglial activation (e.g., ameboid morphology, process retraction). An initial bright-light microscopy inspection was conducted at a lower (5x objective) and a higher magnification (40x objective). Then, the full cross-sectional neurohistological assessments were conducted using the digital section scanning system (APERIO AT2 system) at a digital magnification equivalent to a bright-light microscopy 5x objective. The neuropathological assessment was qualitative in nature, and it was intended as a preliminary screening for overt pathological changes rather than to provide a stereological quantification for the possible morphometric changes detected (i.e., cell number, cell volume, etc.), if any.

Correlations between biology and behavior

The animals used in this study were assessed for social odor recognition memory (SORM) performance at 1- and 6 months post irradiation in a previously published study¹⁰. These same rats were also trained on the rodent psychomotor vigilance task (rPVT), which is a food-reinforced operant test of sustained attention, where rats must monitor and respond to the illumination of a response key that occurs randomly between 3 and 10 s. Detailed training methods for the rodent psychomotor vigilance test (rPVT) have been reported in other studies^45–47, but briefly, rats were trained to respond on an illuminated nose-poke key for food pellets. rPVT sessions began with the onset of the house light, and after a variable delay from 3 to 10 s, the nose-poke was illuminated. Correct responses were defined as a response on the nose-poke while it was illuminated (1.5 s limited hold) and were reinforced with a food pellet. Responses prior to the light onset were not reinforced, punished with an 8 s time out, and counted as premature responses. No response during illumination of the nose-poke key had no programmed consequences and was counted as an omission. Lapses in responding, an important measure for tests of sustained attention, were calculated as omissions plus any correct trial where the response time was greater than twice each rat’s mean response latency for that session. To acquire a single value from each behavioral test that represented each rat’s performance, the rPVT overall performance score (OPS) and the SORM discrimination index was used. The SORM discrimination indices were reported in ref. ¹⁰, and were calculated using total exploration time summed separately for the N2, N1 and familiar beads as follows: DI = (N2 – N1)/(N2 + N1 + F1 + F2) * 100. Discrimination index values greater than zero demonstrate a significant preference for the N2 odor bead, whereas values less than zero show preference for the N1 odor bead. In addition, DI values equal to zero demonstrate that the N1 and N2 odor beads were explored to a similar degree. To acquire a single value that described daily rPVT performance for each subject, we computed an overall performance score that is like the overall performance score used in the human PVT literature: Overall performance score (OPS) = 1 − (# of premature responses + # of lapses)/total trials (including premature responses). The OPS ranges from 1 (perfect performance) to 0 (worst possible performance). The discrimination indices for SORM and rPVT OPS for each rat were plotted against their WB protein levels and dPCR absolute gene expression levels in the Hippo and Str of 1 cGy and 10 cGy exposure groups to determine Pearson correlation coefficients, with resulting matrices presented in Figs. 5C, D and 4C, D showing p values and color-coded by strength and direction of correlation.

Statistics

For Western blot and PCR datasets, the primary experimental objective was to determine if specific radiation doses (1 cGy or 10 cGy) induced deviation from the baseline sham control. Consequently, statistical comparisons were performed using two-tailed unpaired Student’s t-tests to compare each radiation dose group (1 cGy or 10 cGy) separately against the control group. This approach was selected to isolate dose-specific signatures, particularly given the nonmonotonic (nonlinear) biological responses frequently observed in low-dose radiation studies. All molecular assays were performed in technical triplicate, and the average of these replicates was used for statistical analysis. Sample sizes were n = 5–6 rats per group for all molecular experiments.

For the rodent psychomotor vigilance task (rPVT), daily averages of lapses in attention and false alarm responses (measures of impulsivity) were analyzed using linear regression to assess performance trends over time. Additionally, performance metrics at specific response-stimulus intervals (RSI) were compared between the baseline period (final two weeks prior to shipping and radiation exposure) and the post irradiation period (final two weeks of testing) within each group using paired t-tests. Sample sizes for behavioral experiments included n = 11–13 rats per group.

To assess the relationship between molecular changes and cognitive outcomes, Pearson correlation matrices were generated. These analyses correlated the normalized protein or gene expression levels in the hippocampus and striatum with individual behavioral scores (SORM Discrimination Index and rPVT overall performance score) for animals within the 1 cGy and 10 cGy cohorts.

Of note, we carefully considered the use of ANOVA vs. unpaired t-tests. We selected unpaired t-tests to compare each dose independently against control (not-irradiated) animals based on the now well-established nonlinear and nonmonotonic nature of low-dose radiation responses (e.g., hormesis or adaptive responses at low doses vs. damage at higher doses). An ANOVA is indeed best suited for detecting variance across a group or testing for a linear dose-response relationship. Instead, as our data and the broader literature on high-LET radiation suggest, 1 cGy and 10 cGy do not exist on a simple linear continuum but rather induce distinct, and often opposing, biological mechanisms (e.g., synaptic upregulation/priming at 1 cGy vs. downregulation/injury at 10 cGy). Moreover, these neurological responses may vary across different regions of the brain. For example, when using an ANOVA followed by post hoc tests, there is a risk of masking significant effects at the specific low-dose threshold (1 cGy) if the variance is driven primarily by the higher dose, or conversely, if the opposing directions of the effects cancel out the overall group variance. Additionally, when making only two comparisons (rather than three comparisons or more), there is less concern of inflated Type I error, so an ANOVA is unnecessary. Therefore, we treated the comparison of 1 cGy vs. Control and 10 cGy vs. Control as discrete, independent hypotheses to precisely characterize the molecular “fingerprint” of each specific exposure level.

Data values reported are as mean ± SEM. Differences with p value ≤0.05 were considered significant in all cases (*p ≤ 0.05, **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001). Sample sizes of n = 5–6 rats per group were used for all molecular experiments; behavioral experiments included n = 11–13 rats per group. Statistical tests were performed using GraphPad Prism version 10.2.3 for Windows (GraphPad Software, La Jolla, CA). Pearson correlation matrices were generated with GraphPad Prism for MacOS (v10.4.1), with p values ≤0.05 considered significant.

Author contributions

D.I., C.M.D. and K.H. conceptualized the project and wrote the manuscript. K.H. and C.M.D. performed experimental investigations and statistical analyses. E.K.M. validated methods and administrated the project. R.G.I. provided resources and technical expertise. C.M.D. and D.P.P. acquired funding and provided resources. D.I. supervised the project. All authors reviewed the manuscript. D.I. and C.M.D. contributed equally to this work.

Data availability

The datasets used and/or analyzed during the current study and supporting the conclusions of this article are included in this article and in all supplementary materials provided. These datasets are also available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41526-026-00622-z.