Assessing the Effects of Redox Modifier MnTnBuOE-2-PyP 5+ on Cognition and Hippocampal Physiology Following Doxorubicin, Cyclophosphamide, and Paclitaxel Treatment

Background: Chemotherapy treatment for breast cancer can induce cognitive impairments often involving oxidative stress. The brain, as a whole, is susceptible to oxidative stress due to its high-energy requirements, limited anaerobic respiration capacities, and limited antioxidant defenses. The goal of the current study was to determine if the manganese porphyrin superoxide dismutase mimetic MnTnBuOE-2-PyP (MnBuOE) could ameliorate the effects of doxorubicin, cyclophosphamide, and paclitaxel (AC-T) on mature dendrite morphology and cognitive function. Methods: Four-month-old female C57BL/6 mice received intraperitoneal injections of chemotherapy followed by subcutaneous injections of MnBuOE. Four weeks following chemotherapy treatment, mice were tested for hippocampus-dependent cognitive performance in the Morris water maze. After testing, brains were collected for Golgi staining and molecular analyses. Results: MnBuOE treatment preserved spatial memory during the Morris water-maze. MnBuOE/AC-T showed spatial memory retention during all probe trials. AC-T treatment significantly impaired spatial memory retention in the first and third probe trial (no platform). AC-T treatment decreased dendritic length in the Cornu Ammonis 1 (CA1) and dentate gyrus (DG) areas of the hippocampus while AC-T/MnBuOE maintained dendritic length. Comparative proteomic analysis revealed affected protein networks associated with cell morphology and behavior functions in both the AC-T and AC-T/MnBuOE treatment groups.


Introduction
During the last 20 years, breast cancer (BC) patients have self-reported cognitive impairments after chemotherapy bringing awareness to both the clinical and scientific communities [1,2].

Distance Moved
Hippocampal-dependent spatial learning and memory were assessed with the 5-day MWM. Improvement in spatial learning and memory is represented by a decrease in path latency (i.e., time) to reach the platform. A two-way repeated measures ANOVA of velocity revealed significant differences in treatment-by-day interactions (F (4,112) = 27.21; p < 0.0001; Figure 1a), but Holm's correction revealed that there were no significant differences between treatments on either training day. An ordinary two-way ANOVA of latency revealed no significant differences in treatment effect (F (3,140) = 2.24; p = 0.09; Figure 1b). There were significant differences in treatment-by-day interactions (F (4,140) = 23.15; p < 0.0001; Figure 1b) meaning mouse performance of the test improved as testing progressed which suggests that the AC-T does not affect the ability of mice to learn during the MWM task. redox modifier, can ameliorate the effects of AC-T chemotherapy on cognition and hippocampal morphology.

Distance Moved
Hippocampal-dependent spatial learning and memory were assessed with the 5-day MWM. Improvement in spatial learning and memory is represented by a decrease in path latency (i.e., time) to reach the platform. A two-way repeated measures ANOVA of velocity revealed significant differences in treatment-by-day interactions (F (4,112) = 27.21; p < 0.0001; Figure 1a), but Holm's correction revealed that there were no significant differences between treatments on either training day. An ordinary twoway ANOVA of latency revealed no significant differences in treatment effect (F (3,140) = 2.24; p = 0.09; Figure 1b). There were significant differences in treatment-by-day interactions (F (4,140) = 23.15; p < 0.0001; Figure 1b) meaning mouse performance of the test improved as testing progressed which suggests that the AC-T does not affect the ability of mice to learn during the MWM task. There were no significant differences in latency or distance moved between the treatments throughout testing. Mean ± SEM (n = 8).

Probe Trials
To assess spatial memory retention, probe trials were done where the platform was removed after the hidden platform training on days 3-5. Normal, non-impaired animals demonstrate spatial memory retention by spending more time in the target quadrant as compared to the other quadrants. During the first probe trial on day 3, significant differences were revealed for the saline treated group (F (3,28) Figure 2a). The AC-T treated mice were able to discriminate between the target, right, and opposite quadrants on day 3 (F (3,28) = 9.35, p < 0.001; Figure  2a). However, Holm's correction revealed there was no significant difference between the target and left quadrant (p = 0.51, Figure 2a). During the second probe trial on day 4, all treatment groups showed a significant difference in preference for the target quadrant: saline (F (3,28) = 39.80, p < 0.0001; Figure 2b), AC-T (F (3,28) = 17.96, p < 0.0001; Figure 2b), AC-T/MnBuOE (F (3,28) = 37.84, p< 0.0001; Figure 2b), and saline/MnBuOE (F (3,28) = 11.89, p < 0.0001; Figure 2b). Similar to the first probe trial results, the results for the third probe trial on day 5 revealed significant differences for the saline treated group (F (3,28)  There were no significant differences in latency or distance moved between the treatments throughout testing. Mean ± SEM (n = 8).

Probe Trials
To assess spatial memory retention, probe trials were done where the platform was removed after the hidden platform training on days 3-5. Normal, non-impaired animals demonstrate spatial memory retention by spending more time in the target quadrant as compared to the other quadrants. During the first probe trial on day 3, significant differences were revealed for the saline treated group (F (3,28) = 14.34, p < 0.0001; Figure 2a), the AC-T/MnBuOE treated group (F (3,28) = 31.09, p < 0.0001; Figure 2a), and the saline/MnBuOE group (F (3,28) = 18.10, p < 0.0001; Figure 2a). The AC-T treated mice were able to discriminate between the target, right, and opposite quadrants on day 3 (F (3,28) = 9.35, p < 0.001; Figure 2a). However, Holm's correction revealed there was no significant difference between the target and left quadrant (p = 0.51, Figure 2a). During the second probe trial on day 4, all treatment groups showed a significant difference in preference for the target quadrant: saline (F (3,28) = 39.80, p < 0.0001; Figure 2b), AC-T (F (3,28) = 17.96, p < 0.0001; Figure 2b), AC-T/MnBuOE (F (3,28) = 37.84, p< 0.0001; Figure 2b), and saline/MnBuOE (F (3,28) = 11.89, p < 0.0001; Figure 2b). Similar to the first probe trial results, the results for the third probe trial on day 5 revealed significant differences for the saline treated group (F (3,28) = 12.77, p < 0.0001; Figure 2c), the AC-T/MnBuOE-treated group (F (3,28) = 22.16, p < 0.0001; Figure 2c), and the saline/MnBuOE-group (F (3,28) = 12.55, p < 0.0001; Figure 2c). The AC-T treated mice were able to discriminate between the target, right, and opposite quadrants on day 5 (F (3,28) = 6.67, p < 0.01; Figure 2c). However, Holm's correction revealed there was no significant difference between the target and left quadrant (p = 0.40, Figure 2c) demonstrating a memory impairment. We have calculated platform crossing in the first probe trial of the water maze test. We compared all treated mice and found there was a significant difference day 3 and 5 between the treatment groups (Figure 2d, f). Following a second day of hidden platform training (Probe 2), all groups showed memory retention and spent more time searching in the target quadrant than in any other quadrant (Figure 2e). saline/MnBuOE-group (F (3,28) = 12.55, p < 0.0001; Figure 2c). The AC-T treated mice were able to discriminate between the target, right, and opposite quadrants on day 5 (F (3,28) = 6.67, p < 0.01; Figure 2c). However, Holm's correction revealed there was no significant difference between the target and left quadrant (p = 0.40, Figure 2c) demonstrating a memory impairment. We have calculated platform crossing in the first probe trial of the water maze test. We compared all treated mice and found there was a significant difference day 3 and 5 between the treatment groups ( Figure  (d-f) After removal of the platform on day 3-5, the number of platform crossings. Each bar represents the mean of 8 mice; error bars are the SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
After Holm-Sidak multiple comparisons, MnBuOE treatment did not show a significant interaction between treatment and dendritic length, dendritic complexity, nodes, end points, or surface area ( Figure  4). There was a main effect of distance (F ( After Holm-Sidak multiple comparisons, MnBuOE treatment did not show a significant interaction between treatment and dendritic length, dendritic complexity, nodes, end points, or surface area. There was a main effect of treatment (F (1,4) = 13.2 p = 0.0221; Figure 3) and distance (F (32,128) = 57.33, p < 0.0001; Figure 3c). Post-hoc analysis revealed significant changes in dendritic length at 130 µm (Sidak, p = 0.0308). AC-T/MnBuOE treatment as compared with saline alone did show a significant interaction between treatment and dendritic length (F =8.514, p = 0.0013; Figure 3), end points (F = 6.75, p < 0.01), and surface area (F = 9.69, p < 0.001) that remained significant after Holm-Sidak analysis. There was a main effect of Sholl length (F (1.55,6.19) = 0.0484 p = 0.0001: Figure 3b). Post-hoc analysis revealed no significant changes in dendritic length.

CA1 Pyramidal Neurons
Similar analyses were done with the apical and basal regions of the CA1 region. In the apical CA1 neurons, AC-T treatment resulted in significant interaction between treatment and dendritic length (F = 17.79, p < 0.0001; Figure 4), dendritic complexity (F (3,16)   After Holm-Sidak multiple comparisons, MnBuOE treatment did not show a significant interaction between treatment and dendritic length, dendritic complexity, nodes, end points, or surface area ( Figure 4). There was a main effect of distance (F (32,128) = 119.3, p < 0.0001; Figure 4c). Post-hoc analysis revealed no significant changes in dendritic length. When compared to AC-T alone, MnBuOE showed a significant interaction between treatment and dendritic complexity that was significant after multiple comparisons (p < 0.01; Figure 6b). AC-T/MnBuOE treatment did not show a significant interaction between treatment and dendritic length, dendritic complexity, nodes, end points, or surface area after Holm-Sidak multiple comparisons ( Figure 4). There were main effects of distance (F (2.837,11.35) = 92.65, p < 0.0001; Figure 4c). Post-hoc analysis revealed no significant changes in dendritic length.
Looking at the comparison between the AC-T-treated mice and the AC-T/MnBuOE-treated mice in the basal CA1, there were significant interactions between treatment and dendritic length, dendritic complexity (p < 0.001; Figure 6c), nodes, end points, and surface area even after Holm-Sidak post-hoc analysis. There were main effects of distance (F (32,128) = 116.3, p < 0.0001; Figure 5d) and treatment (F

Proteomics
A set of 211 proteins were identified as differentially expressed between saline-treated and chemotherapy (AC-T) treated mice (Table 1 and Table S1). The proteins were annotated with their respective cellular localization and protein type. For the comparison between AC-T treated mice and AC-T/MnBuOE treated mice, a set of 147 proteins was identified as differentially expressed (Table 2 and  Table S1). Table 1. Top dysregulated proteins associated with AC-T treatment. A comparison of AC-T treated mice versus saline-treated mice identified 211 proteins as meeting statistical significance for differential expression. See Table S1 for a full list of proteins and Table S2 for specific peptide count information.

Proteomics
A set of 211 proteins were identified as differentially expressed between saline-treated and chemotherapy (AC-T) treated mice (Table 1 and Table S1). The proteins were annotated with their respective cellular localization and protein type. For the comparison between AC-T treated mice and AC-T/MnBuOE treated mice, a set of 147 proteins was identified as differentially expressed (Table 2  and Table S1). Table 1. Top dysregulated proteins associated with AC-T treatment. A comparison of AC-T treated mice versus saline-treated mice identified 211 proteins as meeting statistical significance for differential expression. See Table S1 for a full list of proteins and Table S2 for specific peptide count information.  For an unbiased perspective of the molecular changes behind chemobrain, the datasets of differentially expressed proteins were uploaded to Ingenuity Pathway Analysis (IPA) application for pathway and network analysis. IPA bioinformatics identified the top protein networks associated with AC-T treatment with functions associated with neurological disease, cell morphology, behavior, and cell-to-cell signaling ( Table 3 and Figure 8). IPA identified molybdenum cofactor and selenocysteine biosynthesis as top canonical pathways associated with AC-T treatment (Table 4 and Figure 9). Next, IPA ran the same analysis using the AC-T versus AC-T/MnBuOE differential protein dataset. The top protein networks associated with addition of MnBuOE included networks with functions associated with cell morphology, cell-to-cell signaling, nervous system development, and behavior (Table 5 and Figure 10). Top canonical pathways associated with MnBuOE treatment included the mitochondrial L-carnitine shuttle pathway, the super pathway of inositol phosphate compounds, and sirtuin signaling ( Table 6 and Figure 11). Table 3. Top five IPA protein networks associated with AC-T treatment. Functions associated with the network include neurological disease, cell morphology, and cell-to-cell signaling and interaction. The color of the node depicts differential expression. Red represents upregulated proteins. Green represents downregulated proteins. The intensity of the color denotes the degree of regulation where brighter colors are more regulated. Gray node color reflects proteins that were found in the data set but were insignificant expression wise. Uncolored nodes were not identified as differentially expressed in our data, but were incorporated into the computational network based on evidence stored in the Ingenuity Knowledge Base. Known direct and indirect interactions between network proteins, as well as the direction of the interaction, are indicated by arrows or blocked lines. Central to the network are p38 MAPK, Akt, Jnk, and Creb.  Functions associated with the network include neurological disease, cell morphology, and cell-to-cell signaling and interaction. The color of the node depicts differential expression. Red represents upregulated proteins. Green represents downregulated proteins. The intensity of the color denotes the degree of regulation where brighter colors are more regulated. Gray node color reflects proteins that were found in the data set but were insignificant expression wise. Uncolored nodes were not identified as differentially expressed in our data, but were incorporated into the computational network based on evidence stored in the Ingenuity Knowledge Base. Known direct and indirect interactions between network proteins, as well as the direction of the interaction, are indicated by arrows or blocked lines. Central to the network are p38 MAPK, Akt, Jnk, and Creb. The IPA ratio divides the number of molecules that meet criteria by the total number of pathway proteins listed in the IPA database. The p-value represents the probability of the ratio occurring by chance. Figure 9. Graph of the number of proteins identified in the proteomics data set that map to the top 5 identified canonical pathways affected by AC-T treatment.    The color of the node depicts differential expression. Red represents upregulated proteins. Green represents downregulated proteins. The intensity of the color denotes the degree of regulation where brighter colors are more regulated. Gray node color reflects proteins that were found in the data set but were insignificant. The IPA ratio divides the number of molecules that meet criteria by the total number of pathway proteins listed in the IPA database. The p-value represents the probability of the ratio occurring by chance. tified Figure 10. Graphic representation of mouse hippocampus protein network 1, identified by IPA as being affected by AC-T/MnBuOE treatment as compared to AC-T alone. Functions associated with the network include cardiovascular disease, cardiovascular system development and function, and cell morphology. The color of the node depicts differential expression. Red represents upregulated proteins. Green represents downregulated proteins. The intensity of the color denotes the degree of regulation where brighter colors are more regulated. Gray node color reflects proteins that were found in the data set but were insignificant. The IPA ratio divides the number of molecules that meet criteria by the total number of pathway proteins listed in the IPA database. The p-value represents the probability of the ratio occurring by chance. The IPA ratio divides the number of molecules that meet criteria by the total number of pathway proteins listed in the IPA database. The p-value represents the probability of the ratio occurring by chance. Figure 11. Graph of the number of proteins identified in the proteomics data set that map to the top 5 identified canonical pathways affected by AC-T/MnBuOE treatment as compared to AC-T treatment.

Sirtuin Signaling Pathway
No of proteins identified Canonical Pathways Figure 11. Graph of the number of proteins identified in the proteomics data set that map to the top 5 identified canonical pathways affected by AC-T/MnBuOE treatment as compared to AC-T treatment.

Discussion
We investigated the ability of MnBuOE to ameliorate the effects of AC-T chemotherapy on the hippocampus. Previously, it was demonstrated that female BALB/C mice receiving combination therapy (AC) showed deficits in spatial memory marked by a decrease in entries and movement in the target zone [26]. Our findings of behavioral deficits with the Morris water maze align with previous studies examining rodent models of cognitive decline after mono-or poly-chemotherapy treatment; however, this is the first time it was shown using the AC-T regimen. Another study noted the protective capacity of antioxidants, using the antioxidant astaxanthin, which reversed doxorubicin-induced cognitive impairment in mice [27]. In our case, the addition of MnBuOE to AC-T treated mice maintained intact spatial memory retention.
Dendritic branching alterations can disrupt synapse formation and/or stability, which ultimately can lead to neurological and cognitive disorders. Interestingly, chemotherapy drugs have been found to negatively impact dendritic branching. Chronic AC treatment significantly reduces total dendritic length and complexity in the hippocampus of female C57BL/6 mice [28]. We found that AC-T chemotherapy decreased dendritic length and complexity in the DG and CA1 regions of the hippocampus. MnBuOE treatment by itself did not affect dendritic length or spatial memory retention. The combination AC-T/MnBuOE maintained dendritic length in the DG and CA1 areas. Changes in the shape or size of the dendritic spines affect their stability and synaptic transmission [29]. Kang et al. found that AC treatment significantly decreased number of mushroom spines in the CA1 neurons of female black mice [28]. We did not see a change in the proportion of mushroom spines after AC-T treatment in the DG or CA1 neurons. Yet, MnBuOE treatment itself increased the number of mushroom spines which are considered the most stable type of spine [30,31]. Increasing the number of dendritic mitochondria or increasing their activity is known to enhance the number and plasticity of spines [32]. A plausible explanation is under normal redox circumstances, MnBuOE could contribute to the formation of mushroom spines by increasing mitochondrial health and/or activity since it accumulates in the mitochondria [20]. Further experiments would be needed to investigate this possibility.
We took a shot-gun proteomic approach to look at how AC-T treatment affected the hippocampus of adult female mice. Mitochondrial specific proteins were both up-and downregulated. Brain mitochondrial carrier protein 1 was found to be upregulated and serves to reduce mitochondrial membrane potential. Conversely, downregulation of the protein C19orf12 homolog is associated with response to oxidative stress. Another downregulated protein was an accessory subunit for NADH dehydrogenase or complex I of the respiratory chain.
The canonical pathways significantly affected by AC-T treatment were molybdenum biosynthesis and selenocysteine biosynthesis which are both involved in redox reactions [33,34]. Fatty acid activation was also affected which can occur in the outer mitochondrial membrane. This indicates that our chemotherapy regimen affects multiple mitochondrial processes. Curiously, the super pathway of inositol phosphate compounds was affected. Despite it being a broad pathway, inositol phosphate compounds, especially inositol 1, 4, 5-trisphosphate (IP3), play an important role in internal calcium ion signaling. Dysregulation of the IP3 receptor pathway has been implicated in neurodegenerative disorders and could provide insight into the mechanisms behind chemobrain [35]. Network analysis revealed affected networks whose broad functions include behavior and cell morphology both of which we found to be affected by AC-T.
The next proteomic comparison involved the effects of MnBuOE on AC-T treated mice. Protein C19orf12 homolog was also found to be downregulated. We found that among the most significantly affected canonical pathways was again the super pathway of inositol phosphate compounds. We also saw the mitochondrial L-carnitine shuttle pathway affected. Doxorubicin has been shown to disrupt the carnitine system contributing to impaired energy metabolism [36] Chemotherapy can lead to cachexia involving perturbed energy balance and mitochondrial dysfunction [37]. We also saw the polyamine regulation in the colon cancer pathway as significantly affected. In regards to brain tissue, polyamines serve as modulators of NMDA (N-methyl-d-aspartate) and AMPA (α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate) glutamate receptors [38]. These receptors are key regulators of long-term potentiation. Sirtuin signaling was also found to be affected. Sirtuins are NAD+ dependent deacetylases that regulate cell responses to stress and are considered to play critical roles in cancer, aging, and neurodegenerative diseases [39,40]. Network analysis revealed the top network to have functions associated with the cardiovascular system and cell morphology. Both doxorubicin and cyclophosphamide have cardiotoxic side effects, so it is not surprising to see proteins associated with cardiovascular signaling. The other top networks also had broad functions associated with behavior. The proteomics analysis suggests mitochondrial dysfunction, oxidative stress, and energy metabolism as possible mechanisms behind chemobrain.

Animals
Sixteen-week-old female C57Bl6/J wild-type mice (n = 48) were purchased from Jackson Laboratory (Bar Harbor, ME). The mice were housed either five or two mice per cage on constant 12:12 light-dark cycle. Food (Tekland Rodent Diet 8604, Envigo) and water were provided ad libitum. All experiments carried out were approved by the Institutional Animal Care and Use Committee at UAMS (number 3831, approval date 28 March 2018).

Chemotherapy Paradigm
The chemotherapeutic drugs were purchased from the UAMS Inpatient Pharmacy and stored according to manufacturer's instructions. Doxorubicin (A) and cyclophosphamide (C) were diluted using sterile saline to make stock solutions. The mice received weekly intraperitoneal injections of saline (0.9% saline; n =12) or doxorubicin (2 mg/kg; n =12) and cyclophosphamide (50 mg/kg: n =12). The AC regimen was given over four weeks totaling four injections ( Day 1,8,15,and 22) (Figure 12). The next cycle of chemotherapy included weekly intraperitoneal injections of paclitaxel (T) (5 mg/kg; n = 12) for 4 weeks. This constituted the AC-T regimen. MnBuOE (a kind gift from Dr. James Crapo, National Jewish Health, Denver, CO, 3 mg/kg; n =12) was then administered twice weekly subcutaneously for eight injections over 4 weeks. All animals were injected between 1400 h-1700 h.

Chemotherapy Paradigm
The chemotherapeutic drugs were purchased from the UAMS Inpatient Pharmacy and stored according to manufacturer's instructions. Doxorubicin (A) and cyclophosphamide (C) were diluted using sterile saline to make stock solutions. The mice received weekly intraperitoneal injections of saline (0.9% saline; n =12) or doxorubicin (2 mg/kg; n =12) and cyclophosphamide (50 mg/kg: n =12). The AC regimen was given over four weeks totaling four injections ( Day 1,8,15,and 22) (Figure 12). The next cycle of chemotherapy included weekly intraperitoneal injections of paclitaxel (T) (5 mg/kg; n =12) for 4 weeks. This constituted the AC-T regimen. MnBuOE (a kind gift from Dr. James Crapo, National Jewish Health, Denver, CO, 3 mg/kg; n =12) was then administered twice weekly subcutaneously for eight injections over 4 weeks. All animals were injected between 1400 h-1700 h.  Dosages were selected based on standard human chemotherapeutic doses. Patients receive four cycles of AC (doxorubicin 60 mg/m 2 and cyclophosphamide 600 mg/m 2 ) followed by four cycles of paclitaxel (175 mg/m 2 ). To translate the clinical dose to animals, the body surface area normalization was used. Previous work by Flanigan et al. showed that doxorubicin (2 mg/kg) and cyclophosphamide (50 mg/kg) was sufficient to cause changes in hippocampal morphology without acute toxicity. The dosages of doxorubicin and paclitaxel translate to human equivalent doses that are below clinical dose. Paclitaxel was then chosen to be given at 5 mg/kg for a cumulative dose of 20 mg/kg, the maximum tolerated single dose for mice [41]. Cyclophosphamide was given at a clinically relevant dose.

Water Maze
The Morris water maze (MWM) was used to assess hippocampal-dependent spatial learning and memory [42]. A circular pool (133 cm diameter) was filled with opaque water (24 • C). The MWM is a 5-day task consisting of two sessions per day (2 h apart). Each session consisted of three trials, resulting in six trials daily. Trials were conducted with 10-min intertrial intervals. The mice were placed into the water facing the edge of the pool. Trials ended when the mouse located the platform. If a mouse failed to locate the platform within 60 s, they were led to the platform by the experimenter and required to sit on the platform for 10 s. Days 1 and 2 were training days using a visible platform. Days 3-5 consisted of hidden platform training which required the mice to rely on extra-maze cues. Following the second session on days 3-5, spatial memory retention was tested using a probe trial. The hidden platform was removed for this trial, and the mice were placed in the quadrant opposite to where the platform was at (target quadrant). The mice were allowed to swim for 60 s. Time spent in the target quadrant was compared to time spent in the non-target quadrants. Other measures of performance in the maze included average velocity and distance to the platform [43]. Ethovision video tracking system (Ethovision XT, Noldus Information Technology, Wageningen, Netherlands) was used to record all sessions.

Tissue Collection
The mice were sacrificed by cervical dislocation followed by decapitation. The brains were removed and dissected. The left hemisphere was dissected down to the olfactory bulb, cerebellum, hippocampus, prefrontal cortex, and cortex in cold phosphate buffered saline (1× PBS). The samples were snap frozen in liquid nitrogen followed by storage at −80 • C. The right hemisphere was stored in Golgi solution.

Golgi Staining
In order to look at dendrite morphology and spine dynamics, Golgi staining was used to randomly stain neurons in the hippocampus [44]. An adapted protocol was used with the reagents of the superGolgi kit (Bioenno Tech LLC, Santa Ana, CA, USA) [45]. Right hemispheres were impregnated with potassium dichromate solution for two weeks (n = 32) followed by 48 h in post-impregnation buffer. The brain was sectioned in the coronal plane at 200 µm thick slices using a vibratome. The slices were transferred into a 24-well plate for staining. The slices were washed in PBS-T (1× PBS, 0.3% Triton TM X-100) before staining with ammonium hydroxide and washed in post-stain buffer. After another PBS-T wash, the slices were mounted on 1% gelatin-coated slides and left to dry overnight in Coplin jars. The next morning, the slides were dehydrated with ethanol, cleaned in xylene, and cover-slipped with Permount TM , Thermo Fisher Scientific TM Inc., Waltham, MA, USA).

Dendritic Morphology
An experimenter blinded to treatment groups traced 5 randomly stained neurons per animal and averaged the values to include in the analysis. Specifically, neurons were traced in the DG and CA1 of the hippocampus. Sholl analysis was used to quantify the amount and distribution of the dendritic arbor [45,46]. A series of circles with increasing radii (10 µm intervals) were centered on the cell body [47]. Statistical analysis was done with GraphPad Prism 8 using a two-way mixed model with Sidak multiple comparisons. Branch-point analysis was used to quantify the complexity of the dendritic. First, the number of bifurcations (where one branch divides into two branches) is counted. Then, the branch-points were counted where first order points were defined where a primary branch splits into second-order branches and so on. Dendritic complexity analysis (DCI) = (sum (branch tip orders + number of branch tips)) X (total dendritic length/total number of primary dendrites). The Neuroexplorer component of the Neurolucida ® program (Microbright-field Science, Williston, VT, USA) was used for the analysis. For complexity, statistical analysis was done with an ordinary 1-way ANOVA with the Holm-Sidak correction for multiple comparisons.

Dendritic Spine Morphology
Dendritic spines were also analyzed using Golgi-stained brain sections featuring the hippocampus. The criteria for neurons chosen for analysis were the following: 1. non-truncated dendrites; 2. consistent Golgi staining along dendrites; 3. relative isolation from other neurons to avoid interference [48]. The neurons were traced using Neurolucida software version 11. Six to seven neurons per brain with 5 dendritic segments (20 nm long) per neuron were analyzed [47].

Data Analysis
Proteins were identified and quantified by searching the UniprotKB database restricted to Mus musculus using MaxQuant (version 1.6.5.0, Max Planck Institute). The database search parameters included selecting the MS1 reporter type, trypsin digestion with up to two missed cleavages, fixed modifications for carbamidomethyl of cysteine, variable modifications for oxidation on methionine and acetyl on N-terminus, the precursor ion tolerance of 5 ppm for the first search and 3 ppm for the main search, and label-free quantitation with iBAQ normalized intensities. Peptide and protein identifications were accepted using the 1.0% false discovery rate identification threshold. Protein probabilities were assigned by the Protein Prophet algorithm [49].
MaxQuant iBAQ intensities for each sample were median-normalized so the medians were equal to the sample with the maximum median. Median-normalized iBAQ intensities were then imported into Perseus (version 1.6.1.3, Max Planck Institute) to perform log2 transformation and impute the missing values using a normal distribution with a width of 0.3 and a downshift of 2 standard deviations. The Linear Models for Microarray Data (limma) Bioconductor package was used to calculate differential expression among the experimental conditions using the lmFit and eBayes functions [49,50]. Proteins were considered to be significantly different with a fold change >2 and a false discovery rate adjusted p-value < 0.05. Differentially expressed proteins were analyzed using Ingenuity Pathway Analysis (IPA, QIAGEN Redwood City, www.qiagen.com/ingenuity) to identify pathways and networks.
Visible-and hidden-platform water-maze learning curves were analyzed by two-way repeated-measures ANOVA. We used Holm's correction to control for multiple comparisons, and separate analyses were conducted for the visible-and hidden-platform learning curves. For analysis of performance in the MWM probe trials, we used one-way ANOVAs with Holm's post-hoc test, when appropriate. For measures of dendritic intersections, a mixed-factors ANOVA tested for the effects of chemotherapy (as the between-subjects variable) and distance from the cell body (Sholl radius, repeated measures variable). ANOVAs were followed by Holm's post hoc tests when appropriate.

Conclusions
We used adult female mice to measure the effects of AC-T chemotherapy on hippocampal morphology and spatial memory. We observed decreased spatial memory retention in the Morris water maze for AC-T treated animals as compared to controls. AC-T/MnBuOE was observed to have intact spatial memory retention. We also found alterations in dendritic morphology in the hippocampus of AC-T treated mice. AC-T/MnBuOE preserved dendritic morphology. MnBuOE treatment alone increased the number of mushroom spines. Proteomic network analysis revealed networks associated with cell morphology and behavior, in which we found deficits with AC-T treatment. Further work elucidating the contribution of oxidative stress to cognitive impairment after chemotherapy will help inform us of potential therapeutics such as the SOD mimetic, MnBuOE. Next, it is important to look at the effects of chemotherapy on cognition, dendritic morphology, mitochondrial dysfunction, and oxidative stress in tumor-bearing animals in order to see the effects of MnBuOE on cancer. Funding: This work was supported by the Translational Research Institute for Space Health (TRISH) funded project #T0401 and NIH P20 GM109005. This work was also supported by core facilities of the Center for Translational Neuroscience IDeA program award P30 GM110702 and by the Center for Translational Pediatric Research Genomics Core supported by P20 GM121293.