Influence of ApoE genotype on doxorubicin-induced cognitive impairment in juvenile rats Authors: Chadni Patel, Frank Diglio, Benjamin H Durham, Peter D Cole Published: 2025 ABSTRACT AbstractMany pediatric cancer survivors experience chemotherapy-induced cognitive impairment (CICI), which negatively impacts their quality of life. However, while some patients develop CICI, others do not, suggesting that genetic variants may contribute to the development of CICI. The Apolipoprotein E (ApoE) E4 allele has been identified as a risk variant for CICI among pediatric cancer survivors. However, the mechanisms by which ApoE4 contributes to the development of CICI remain unknown. Using a commonly used chemotherapeutic agent known to induce CICI, doxorubicin, we treated five-week-old rats homozygous for either the human ApoE3 or ApoE4 allele with doxorubicin (2 mg/kg/week for 4 weeks) or saline. Behavioral assessments revealed that ApoE4 rats were more susceptible to doxorubicin-induced impairments in visual and spatial memory compared to ApoE3 rats. Pathophysiological analyses showed a significant reduction in hippocampal neurogenesis of ApoE4 doxorubicin-treated rats relative to the other groups. Serum levels of GFAP were significantly increased in ApoE4 doxorubicin-treated rats. These findings suggest that the ApoE genotype influences vulnerability to CICI and highlight a potential mechanistic link through impaired neurogenesis, laying the groundwork for genotype-specific therapeutic strategies.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33104-5.Keywords: Chemotherapy-induced cognitive impairment, Chemotherapy-induced neurotoxicity, Doxorubicin, ApoE4, ChemobrainSubject terms: Cancer, Neuroscience Introduction The survival rate for pediatric cancers has significantly increased in the past several decades1,2. However, many pediatric cancer survivors experience toxicity from therapy, including chemotherapy-induced cognitive impairment (CICI)3,4, which can affect more than a third of childhood cancer survivors5,6. Since children with cancer are exposed to chemotherapy while they are still undergoing active brain development, CICI can continue well into adulthood. CICI includes a range of symptoms and measurable deficits, including impairment of executive functioning, concentration, processing speed, and short-term memory7,8. For pediatric survivors, these impairments negatively affect their school performance9, leading to an increased need for special education classes10. Symptoms associated with CICI lead to a reduction in quality of life for the survivors as well as their caregivers11. Not all pediatric cancer patients are at equal risk for CICI, and the interpatient variability in the susceptibility to CICI is not well understood12. Our laboratory and other researchers have previously identified common polymorphisms in genes related to neuroinflammation and oxidative stress that were associated with inferior intelligence quotient or cognitive function among survivors of childhood acute lymphoblastic leukemia (ALL), supporting the hypothesis that these genes increase the susceptibility of CICI13. These polymorphisms in genes include endothelial nitric oxide synthase (NOS3), catechol-O-methyltransferase (COMT), hemochromatosis (HFE), glutathione S-transferase pi 1 (GSTP1), prostaglandin transporter (SLCO2A1), and apolipoprotein E4 (APOE4)13,14. However, preclinical studies investigating the effects of gene variants on CICI primarily focus on adult animal models and include only one sex15–17, and therefore cannot establish direct relevance to pediatric cancer patients. To address this gap, we used a juvenile rat model, with animals of both sexes, homozygous for either the ApoE3 or ApoE4 allele. For this initial study, we focused on doxorubicin, a chemotherapy agent that is broadly used in the treatment of pediatric and adult cancer patients. Multiple in vivo studies showed that doxorubicin induces cognitive dysfunction, along with increased neurodegeneration, decreased neurogenesis, and increased pro-inflammatory markers18–22. However, many of these preclinical models for doxorubicin-induced CICI do not accurately reflect pediatric dosing used clinically, explore the acute effects of doxorubicin, utilize animals of only one sex, and do not account for genetic susceptibility. Therefore, it is essential to not only develop a pediatric CICI model but also to better understand the neuropathology and sub-acute effects of doxorubicin to improve the quality of life of patients. Several well-described factors that can account for CICI include a leaky blood–brain barrier (BBB), proinflammatory neuroimmune system, and heightened neurogenesis5. However, these mechanisms have not been explored in an ApoE4 juvenile CICI rat model. Here, we mimicked a curative pediatric treatment regimen utilizing five-week-old rats homozygous for either the human ApoE3 or ApoE4 allele and treated with doxorubicin to study whether harboring the ApoE4 genotype will increase sensitivity to CICI and exacerbate the associated neuropathology. Materials and methods Animal model Four-week-old Sprague Dawley rats with a homozygous replacement of the rat ApoE gene with either human ApoE3 (hApoE3) or human ApoE4 (hApoE4) gene were purchased from Envigo (an Inotiv company). Both the hApoE3 and hApoE4 isoforms were created utilizing Zinc Finger Nuclease (ZFN). The rats were allowed one week to acclimate to the vivarium. The rats were housed in same-sex groups of two or three rats per cage with a 12/12 h light/dark cycle and ad libitum access to food (LabDiet 5053) and water. At 5 weeks of age, the rats were treated via intraperitoneal (i.p.) route of administration with 2 mg/kg/week of doxorubicin for 4 weeks (Doxorubicin HCl 2 mg/mL, Fresenius Kabi, LLC.—63323010161; Henry Schein, Catalog No. 1164027) or equal volume of saline (1X PBS (Phosphate Buffered Saline pH 7.4, Gibco, Catalog No. 10010049). All injected solutions were sterilized using 0.22 μM syringe filters (VWR, Catalog No. 76479-016). The rats were randomly divided into 4 experimental groups, with an equal number of males and females: ApoE3 + saline, ApoE3 + doxorubicin, ApoE4 + saline, and ApoE4 + doxorubicin. Each group had 9 males and 9 females, creating a total of n = 18 per group. The weights of the rats were monitored weekly during the injections. This regimen is approximately equivalent to 25 mg/m2 per course23, which is typical dosing for some pediatric cancer regimens. All experiments were approved by the Animal Institute and Use Committee of Rutgers University and were conducted following the “Guide for the Care and Use of Laboratory Animals.” The ARRIVE guidelines were followed for all experiments described here. Behavioral testing Behavioral testing was conducted to assess the sub-acute effects (10–13 weeks of age; 2–5 weeks after the last doxorubicin injection) (Fig. 1A) of doxorubicin treatment at a juvenile age. To assess the cognitive functioning of the rat model, we used the following tests, open field (OF) (10 weeks of age), object placement (OP) (11–12 weeks of age), and object recognition (OR) (12–13 weeks of age), as previously published by our laboratory24–26 with slight modifications. First, the OF test was performed to confirm that doxorubicin does not induce locomotor deficits or anxiety-like behavior. In the OF test, we recorded the total track length, center track length, center time, and center entries. The novel object recognition test (OR) is a test for visual and recognition memory27. In the OR test, each rat was exposed to two identical objects for 5 min and returned to its home cage. After a defined interval (60–90 min), the rat was allowed to explore the same arena, but now with a novel object for 3 min. Our second detection method is the object placement test (OP), which tests the spatial memory. Similar to the OR test, in the OP test, the rat was exposed to two similar objects for 3 min, and then, after a defined interval (7.5–15 min), the rat explored the same arena, but with a relocated object, and the rat spent 3 min testing. The time the rat spent with each object was recorded. The preference score is the ratio of the time spent exploring the novel or relocated object divided by the total time exploring both objects, and that is calculated as a percentage24–26. A preference score threshold of 55% was used for the OR and OP tests, where percentages above this threshold indicate intact memory24. Statistical analysis was performed to compare the different groups through GraphPad Prism 10.4 (GraphPad Software, San Diego, CA). The Shapiro–Wilk test was used to test for normality within each group. A one sample t test (parametric) with a hypothetical mean of 50 was used to determine whether the PBS-treated control rats displayed a preference score higher than chance, which indicates intact memory. Rats that received a preference score above the threshold preferred the novel or relocated object. Rats that performed below the threshold had no preference for the novel or relocated object. The number of rats that either had a preference or had no preference for the novel or relocated object was determined, and a pairwise contingency was performed using the two-sided Fisher’s exact test, and the significance displayed on the graph is based on the number of rats. However, the graph itself displays the percentage. All the objects utilized for the OR and OP tests were validated prior to testing. For the behavioral analyses, the data from males and females from the same group were combined. ApoE4 doxorubicin-treated rats have increased sensitivity to cognitive impairment. (A) Schematic showing the doxorubicin (DOXO) treatment regimen via intraperitoneal (i.p.) injections in ApoE3 (E3) and ApoE4 (E4) rats, followed by behavioral testing and rat euthanasia and sample collection. (B) The growth curve shows the weights of rats in grams during and after doxorubicin treatment. The groups are separated by sex, genotype, and treatment (n = 9 per group). Two weeks after the last doxorubicin dose, the open field test was conducted on the rats: (C) bar graph shows the mean total track length in centimeters (cm) for each group, and (D) bar graph shows the mean center time for each group (n = 18 per group displayed). (E) The bar graph represents the object placement (OP) preference score percentage for each group, and it is calculated by the time spent exploring the relocated object divided by total time exploring both objects and displayed as a percentage. The red dotted line is the 55% threshold. (F) The stacked bar graph represents the percentage of the total number of rats for that group that either performed above the threshold (55%) and had a preference or that performed below and had no preference for the OP test. Groups are separated by genotype and treatment (n = 15–18 per group displayed). (G) The bar graph represents the object recognition (OR) preference score percentage for each group, and it is calculated by time spent exploring the novel object divided by total time exploring both objects and displayed as a percentage. The red dotted line is the 55% threshold. (H) The stacked bar graph represents the percentage of the total number of rats for that group that either performed above the threshold (55%) and had a preference or that performed below and had no preference for the OR test. Groups are separated by genotype and treatment (n = 15–18 per group displayed). Data represent mean ± SEM, and biological replicates per group are displayed as a circle on the graph. 2way ANOVA with Tukey’s multiple comparisons test (B), one sample t test with hypothetical mean at 50 (E, G). Contingency Fisher’s exact test was utilized on the number of animals per group, and the significance displayed here is based on the number of rats per group that had a preference or no preference for each genotype, which was conducted individually (F, H). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.001, n.s.: not significant. Biorender.com used for Panel A. Sample collection At approximately 13–14 weeks of age (5–6 weeks after the last doxorubicin injection), the rats were euthanized. The rats were deeply anesthetized with 2–5% isoflurane. After the animal was completely anesthetized (~ 2–3 min), the anesthetic depth was checked by showing lack of response to a toe pinch reflex. An incision was made at the level of the xiphoid process, the diaphragm punctured, and ribs were cut to open the thoracic cavity. Blood was collected from the left ventricle of the heart using a 25gx5/8″ 1 cc syringe needle, allowed to coagulate for 20 min at room temperature, then immediately put on ice, centrifuged at 2000 g for 10 min at 4 °C, and the serum was transferred to a new tube and stored in − 80 °C until further processing. Following the cardiac blood collection, an incision was made in the rat’s right atrium, followed by cardiac perfusion with PBS, subsequently the brain was removed, and a sagittal cut was done. One hemisphere of the brain was immediately placed in 4% paraformaldehyde (PFA) (Formaldehyde Aqueous Solution (Paraformaldehyde Aqueous Solution) EM Grade, Electron Microscopy Sciences, Catalog No. 15714, diluted in 1X PBS accordingly to make 4% PFA) for fixation for 4–6 h at room temperature, followed by 24 h in 4 °C on shaker. This was followed by 70% ethanol washes at 4 °C, and kept in 70% ethanol, and sent to the Rutgers Cancer Institute Histopathology Core facility for processing and for paraffin-embedded blocks for the sagittal sections. Brain sections were cut at a 5 μm thickness. The contralateral hemisphere was dissected into 3 different regions: hippocampus, cortex, and cerebellum, and all were snap-frozen separately in liquid nitrogen and stored at -80 °C until further processing. Additionally, after the PBS perfusion, the apex of the heart’s left ventricle was collected, and fixed in 4% PFA and processed similarly to the brain samples described above. A part of the left ventricle was also removed and snap-frozen in liquid nitrogen and stored at -80 °C until further processing. For a small cohort of rats, liver samples were also collected similarly, where 2 parts of the right lobe of the liver were collected, one was fixed in 4% PFA and processed similarly to the brain samples described above. Another piece of the right lobe of the liver was collected and snap-frozen in liquid nitrogen and stored at − 80 °C until further processing. All experiments were approved by the Animal Institute and Use Committee of Rutgers University and were conducted following the “Guide for the Care and Use of Laboratory Animals.” RNA sequencing Hippocampal and cortical tissue collected at the time of the animal euthanasia were pulverized into a fine powder using liquid nitrogen and the traditional mortar and pestle. Approximately 10–20 mg of powdered hippocampal and cortical tissue was weighed on dry ice and shipped to GENEWIZ from Azenta Life Sciences (South Plainfield, New Jersey, USA). RNA extraction, quality control, library preparation, sequencing, and data analysis were conducted at Azenta Life Sciences (South Plainfield, NJ, USA) (https://www.azenta.com/). From the frozen brain samples provided to Azenta, the total RNA was extracted using the Qiagen RNeasy Plus Universal mini kit following manufacturer’s instructions (Qiagen, Hilden, Germany). RNA integrity was and total RNA was normalized prior to library preparation following manufacturer’s protocol. RNA sequencing libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina using manufacturer’s instructions (NEB, Ipswich, MA, USA). Using a 2 × 150 bp Paired End (PE) configuration the samples were sequenced. The quality of the raw data was assessed and the sequence reads were trimmed to remove possible adapter sequences and nucleotides with poor quality. Using the STAR aligner v.2.5.2b, trimmed reads were mapped to the reference genome available on ENSEMBL and BAM files were generated. Using feature counts from the Subread package v.1.5.2, the unique gene hit counts were calculated and only unique reads that fell within exon regions were counted. The gene hit counts table was used for downstream differential expression analysis. Using DESeq2, a comparison of gene expression between the groups of samples was performed. And the Wald test was used to generate p-values and Log2 fold changes. Volcano plots were made in R using the differentially expressed genes (DEG) RNA-sequencing data file provided by Azenta. To conduct pathway analysis, the analyzed DEGs RNA-sequencing data file was uploaded to Qiagen’s Ingenuity Pathway Analysis (IPA) platform, and the threshold selection for the dataset was |log2(Fold Change)|= 0.25 and p-value < 0.05. The comparisons uploaded were: ApoE3 PBS vs ApoE3 doxorubicin and ApoE4 PBS vs ApoE4 doxorubicin, both for the hippocampus region. IPA was utilized to identify canonical pathways impacted 5–6 weeks following the last doxorubicin dose. IPA platform calculated the activation z-score at a -log10 (p-value) threshold of 1.3. The comparison analysis feature was used to allow for visualization across multiple analyses, and in our case, it was utilized to compare: ApoE3 doxorubicin vs ApoE3 PBS and ApoE4 doxorubicin vs ApoE4 PBS. Immunofluorescence staining The brain sagittal paraffin-embedded sections were cut at 5 μm thickness. The de-paraffinization and rehydration of the slides were performed for 10 min each (Hemo-D, Hemo-D, 100% Ethanol, 100% Ethanol, 95% Ethanol, 70% Ethanol, 50% Ethanol, ddH2O). This was followed by heated antigen retrieval with citrate buffer (abcam, Catalog No. ab93678 diluted in ddH2O) for 15 min and cooled down to room temperature. The sections were washed with PBS before blocking. The sections were blocked with 20% Horse Serum (HS) (Millipore Sigma, Catalog No. H0146-10ML) with 0.3% Triton X-100 (Sigma, Catalog No. T8787-100ML) for 90 min at room temperature. The sections were then washed with PBS. The sections were incubated overnight at 4 °C in a humidified chamber with primary antibody diluted in 5% HS in PBS: rabbit doublecortin (DCX, 1:250) (abcam, Catalog No. ab18723). The next day the sections were incubated at room temperature for 90 min with the secondary antibody: donkey anti-rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 594 (1:600) (Invitrogen, Catalog No. A32754). The sections were washed with PBS-T (0.1% Tween) and then stained with DAPI (1:1000) for 5–10 min. Followed by the PBST washes and ddH2O wash. The slides were mounted with Fluormount G (Southern Biotech, Catalog No. 0100-01) and coverslip #1.5. The sections were imaged using a Leica THUNDER Imaging fluorescence microscope and Zeiss Axioscan 7 with Zen Blue version 3.7.97.05. The Zeiss Axioscan slide scanner tissue was scanned with the 20X objective using DAPI as the focus reference plane. For Cohort 1 (71 rats), DCX-positive cells were counted from two sections of the same rat in the dentate gyrus of the hippocampus of the rat brain and averaged, except in the case of a section being damaged. For Cohort 2 (31 rats), DCX-positive cells were counted from one section of each rat in the dentate gyrus of the hippocampus. For each rat, the area of the dentate gyrus was measured to express the DCX-positive cells as DCX + /mm2. For the DCX staining, each cohort was normalized to the ApoE3 PBS group to account for differences in each cohort, and then the fold change or relative values were combined. Serum assays Serum collected from a smaller cohort was utilized to conduct the following assay’s: Creatine Kinase (Ray Biotech, Catalog No. MA-CK), Alanine Transaminase (Ray Biotech, Catalog No. MA-ALT), Superoxide Dismutase (Ray Biotech, Catalog No. MA-SOD), and Triglyceride (Ray Biotech, Catalog No. MA-TG). Serum collected from the larger cohort was utilized to conduct the S-PLEX Neurology Panel 1 Kit (Meso Scale Discovery (MSD), Catalog No K15639S-1), which measured glial fibrillary acidic protein (GFAP), Tau, and Neurofilament Light Chain (NfL). Additionally, the serum protein levels were measured using Rat TIMP-1 ELISA Kit (Ray Biotech, Catalog No. ELR-TIMP1-1), Rat ICAM-1 ELISA Kit (Ray Biotech, Catalog No. ELR-ICAM1-1), Rat BDNF ELISA Kit (Ray Biotech, Catalog No. ELR-BDNF-1), and Rat S100B ELISA kit (AFG Bioscience, Catalog No. EK720458). The assays were conducted according to the manufacturer’s protocol and guidelines. Concentration of samples were interpolated from the standard curve. Western blots Brain lysates were prepared by homogenizing either cortex or hippocampus samples in the standard lysis buffer: 1X RIPA buffer (RIPA Lysis and Extraction Buffer, Thermo Scientific, Catalog No. 89901) with protease inhibitors (Halt Protease Inhibitor Cocktail, EDTA-Free (100X), Thermo Scientific, Catalog No. 78425) and phosphatase inhibitors (Pierce Phosphatase Inhibitor Mini Tablets, Thermo Scientific, Catalog No. A32957). Protein concentrations were determined using the Pierce BCA Protein Assay Kits (Thermo Scientific, Catalog No. 23225) and samples were prepared using 10–25 μg of protein lysates, RIPA buffer, 4X Bolt LDS Sample Buffer (Invitrogen Thermo Fisher, Catalog No. B0007), 10X Bolt Sample Reducing Agent (Invitrogen Thermo Fisher, Catalog No. B0009) and heated at 95 °C for 5–7 min. Proteins were separated by Bolt™ Bis–Tris Plus Mini Protein Gels, 4–12%, 1.0 mm, 15 well (Invitrogen Thermo Fisher, Catalog No NW04125BOX) with pre-stained protein ladder (Precision Plus Protein Dual Color Standards, Bio-Rad, Catalog No. 1610374). 1X of Running Buffer was prepared using the Bolt MES Running Buffer, 20X (Invitrogen Thermo Fisher, Catalog No. B0002). Proteins were transferred onto BrightStar™ PVDF Membrane, Pore Size 0.45 µm (Asi Alkali Scientific, Catalog No. XR732) in the BioRad Trans-Blot Turbo Transfer System using the standard predefined protocol. Ponceau S. staining confirmed the transfer of proteins onto the membrane. Blots were blocked either in 5% Milk (Blotting-Grade Blocker, Biorad, Catalog No. 1706404) or 5% bovine serum albumin (BSA) in PBS-T with 0.1% Tween (PBS-T) or Tris-buffered saline with 0.1% Tween (TBS-T) for 60–120 min at room temperature on a shaker/rocker. The blots were cut prior to hybridization with primary antibodies and the protein ladder was used to assess the correct location to cut the blots. The primary antibody incubation was in 5% Milk or 5% BSA at 1:500–1:1000 dilution at 4 °C overnight. The following primary antibodies were used: p38 MAPK (1:1000; Cell Signaling Technology, Catalog No. 9212S), Phospho-p38 MAPK (Thr180/Tyr182) (1:500–1:1000; abcam, Catalog No. ab4822), phospho-Histone H2A.X (Ser139) (1:500–1:1000, Sigma-Aldrich Millipore Sigma, Catalog No. 05-636), myelin basic protein (MBP, 1:500) (Millipore Sigma, Catalog No. NE1019), β-Actin (1:1000; Cell Signaling Technology, Catalog No. 3700S), Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (1:1000; Cell Signaling Technology, Catalog No. 9101S), and p44/42 MAPK (Erk1/2) (1:1000; Cell Signaling Technology, Catalog No. 4695S). Blots were washed 3 times in PBS-T or TBS-T for 15 min each, followed by secondary antibody incubation in 5% Milk at 1:5000 dilution [(Goat anti-Rabbit IgG (H + L) Secondary Antibody, HRP, Invitrogen, Catalog No. 31460) or (Goat anti-Mouse IgG (H + L) Secondary Antibody, HRP, Invitrogen, Catalog No. 31430)]. Blots were washed 3 times in PBS-T or TBS-T for 15 min each, and blots were developed in the ChemiDoc Biorad Imaging System or Azure Biosystems Imager using the SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific, Catalog No. 34096). All images were analyzed using ImageJ. Enzyme-linked immunosorbent assays (ELISA) Brain lysates were prepared by homogenizing either cortex or hippocampus samples in the standard lysis buffer (described above in the Western Blots section), followed by centrifugation and protein quantitation using the Pierce BCA Protein Assay Kit (Thermo Fisher, Catalog No. 23227). Equal amount of protein was assessed per sample, and the assay was conducted according to manufacturer’s protocol for the following ELISA kits: Transthyretin (Prealbumin) (TTR) (Ray Biotech, Catalog No. ELR-Ttr-1) (40–50 µg protein per sample), Brain-derived neurotrophic factor (BDNF) (Ray Biotech, Catalog No. ELR-BDNF-1) (25 µg protein per sample), Tau (Novus bio-techne, Catalog No. NBP2-81,164) (20–25 µg protein per sample), cAMP responsive element binding protein (CREB) (Novus bio-techne, Catalog No. NBP2-67251) (20–25 µg protein per sample), and Phospho-CREB (S133) DuoSet (bio-techne R&D, Catalog No. DYC2510-2) (30 µg protein per sample). Concentration of samples were interpolated from the standard curve and normalized to the protein load per well. For phospho-CREB, the samples from 2 cohorts were utilized, and each cohort was normalized to the ApoE3 PBS group to account for cohort-specific differences, and then the fold change was combined. Oxidative stress assays Brain lysates were prepared by homogenizing the cortex samples in standard lysis buffer with butylated hydroxytoluene (BHT) followed by centrifugation and protein quantitation using the Pierce 660 nm Protein Reagent (Thermo Fisher, Catalog No. 22660). An equal amount of protein was assessed per sample according to the manufacturer’s protocol for the following: Lipid Peroxidation Malondialdehyde (MDA) Assay Kit (Ray Biotech, Catalog #: MA-MDA-2) (100-150 µg protein per sample), Total Antioxidant Capacity Assay kit (abcam, Catalog No. ab65329) (50 µg protein per sample), GSH/GSSG Ratio Detection Assay Kit II (abcam, Catalog No. ab205811) (25 µg protein per sample), Prostaglandin E2 (PGE2) Assay (biotechne R&D, Catalog No. KGE004B). For MDA, total antioxidant capacity, GSH, and PGE2 assays, cortex samples from 2 cohorts were utilized, and each cohort was normalized to the ApoE3 PBS group to account for cohort-specific differences, and then the fold change was combined. Immunohistochemical staining Hematoxylin and eosin (H&E) was conducted on liver and heart tissue sections using the Sakura Prisma automated stainer according to the manufacturer’s guidelines. Picric Acid Sirius Red staining was performed on liver and heart tissue sections following the manufacturer’s protocol. Both heart and liver tissue were paraffin-embedded as explained above and the sections were cut at 5 μm thickness. The Zeiss Axioscan 7 was used to capture the brightfield scans. A pathologist performed formal interstitial fibrosis scoring for the heart and liver tissue. Phosphorylation assay Heart and liver lysates were prepared by homogenizing samples in the lysis buffer (as described above in the Western Blot section), followed by centrifugation and protein quantitation using the Pierce BCA Protein Assay Kit (Thermo Fisher, Catalog No. 23227). An equal amount of protein was assessed per sample, and the assay was conducted according to the protocol provided by MSD for the following semi-quantitative assays: Phospho/Total ERK1/2 Kit (MSD, Catalog No K15107D-1) and Phospho/Total p38 Kit (MSD, Catalog No K15112D-1). Samples were normalized to the ApoE3 PBS group to determine the relative changes in phosphorylation. Statistical analysis GraphPad Prism 10.4 (GraphPad Software, San Diego, CA) was utilized to perform all the analyses. Data were analyzed by the Shapiro–Wilk test to assess for normality within each group or normality of residuals. A two-way ANOVA full model with Tukey’s multiple comparison was utilized for most panels. If the residuals did not pass the normality test, a natural log transformation was performed, followed by a two-way ANOVA full model with Tukey’s multiple comparison. In certain cases, where the residuals did not pass the normality test after a natural log transformation, we further evaluated the distribution of the residuals using histograms. When the residuals appeared normal and reasonable, a two-way ANOVA full model followed by Tukey’s multiple comparison was conducted. If a log transformation was conducted, the data in the graphs was presented in original units, however, the pairwise comparison and p-value significance displayed was based on the two-way ANOVA results of the log-transformed data. In the behavioral testing analysis, a one sample t test with a hypothetical mean of 50 was used to determine whether the PBS-treated control rats displayed a preference score higher than chance. Additionally, the contingencies in the behavioral testing were analyzed by Fisher’s exact test. For all statistical tests, results are considered statistically significant when the p < 0.05. Results ApoE4 doxorubicin-treated rats have increased sensitivity to cognitive impairment To mimic a pediatric regimen in vivo, male and female juvenile ApoE3 and ApoE4 rats were treated with doxorubicin (2 mg/kg/week for 4 weeks) (Fig. 1A). We observed that ApoE4 doxorubicin-treated male rats have a significantly lower average weight at the completion of the treatment regimen at 9 weeks of age (Fig. 1B). A significant difference in weight for the ApoE4 doxorubicin-treated male rats relative to the ApoE4 PBS-treated male rats persisted until the conclusion of the experiment (Fig. 1B). A couple of male doxorubicin-treated rats died before the completion of the study. To test whether harboring the ApoE4 genotype increases sensitivity to cognitive impairment post-doxorubicin treatment, we investigated the impact of doxorubicin on cognition by conducting behavioral testing. We observed no differences in the total track length (Fig. 1C; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,68) = 0.001639, p = 0.9678), ApoE genotype (F (DFn, DFd), F(1,68) = 0.4556, p = 0.5020), treatment (F (DFn, DFd), F(1,68) = 0.02443, p = 0.8763)) and center time (Fig. 1D; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,51) = 0.3738, p = 0.5436), ApoE genotype (F (DFn, DFd), F(1,51) = 2.420, p = 0.1260), treatment (F (DFn, DFd), F(1,51) = 0.01196, p = 0.9134)), indicating no locomotor deficits or anxiety-like behavior after doxorubicin treatment in the rats. These results provided evidence to proceed with behavioral testing as the rats do not have any motor skill deficiency and allowed us to interpret the results of the subsequent cognitive tests accurately. Both the ApoE3 and ApoE4 PBS-treated rats demonstrated mean OP preference scores significantly higher than chance, indicating intact spatial memory (Fig. 1E; One sample t test: E3 PBS, t = 6.830, df = 15, p < 0.0001; E4 PBS, t = 2.689, df = 17, p = 0.0155). In contrast, ApoE4 doxorubicin-treated rats were more likely to exhibit spatial memory deficits compared to the ApoE4 PBS-treated rats in the OP test (Fig. 1F; E4 PBS compared to E4 DOXO: Fisher’s exact test, p = 0.0143). Similarly, in the OR test, we found that both the ApoE3 and ApoE4 PBS-treated rats attained a mean preference score significantly higher than chance, while the doxorubicin-treated rats did not (Fig. 1G; One sample t test: E3 PBS, t = 6.822, df = 17, p < 0.0001; E4 PBS, t = 3.948, df = 17, p = 0.0010). The results of the OR test showed that ApoE4 doxorubicin-treated rats were more likely to display visual memory deficits compared to the ApoE4 PBS-treated rats (Fig. 1H; E4 PBS compared to E4 DOXO: Fisher’s exact test, p = 0.0004). We noted that while not significant, for both the OP (Fig. 1F; E3 PBS compared to E3 DOXO: Fisher’s exact test, p = 0.0707) and OR (Fig. 1H; E3 PBS compared to E3 DOXO: Fisher’s exact test, p = 0.0750) tests, we observed a considerable number of ApoE3 doxorubicin-treated rats that had no preference for the relocated or new object, respectively, when compared to the ApoE3 PBS-treated rats. Our results clearly suggest that harboring the ApoE4 genotype increased vulnerability to cognitive impairment induced by doxorubicin treatment. ApoE4 doxorubicin-treated rats have decreased hippocampal neurogenesis To further deepen our understanding of the neuropathological mechanisms associated with doxorubicin-induced cognitive impairment, we explored a key marker of neurogenesis. Hippocampal neurogenesis is an important process for cognition; doublecortin (DCX) is the marker of immature neurons28. We found a significant suppression in the number of DCX-positive cells in the dentate gyrus of the hippocampus of the ApoE4 doxorubicin-treated rats compared to all the other groups (Fig. 2A and B). We observed significant effects of ApoE genotype (two-way ANOVA, F (DFn, DFd), F(1,90) = 16.50, p = 0.0001) and treatment (two-way ANOVA, F (DFn, DFd), F(1,90) = 8.656, p = 0.0041) on DCX expression in the hippocampus (Fig. 2B). However, a two-way ANOVA revealed no significant interaction (F (DFn, DFd), F(1,90) = 1.343, p = 0.2497). Although not statistically significant, we observed that ApoE4 PBS-treated rats had comparatively fewer DCX-positive cells compared to ApoE3 PBS-treated rats. Overall, these results suggest that doxorubicin treatment decreased neurogenesis in the hippocampus of ApoE4 rats. Doxorubicin treatment impacts hippocampal neurogenesis in ApoE4 rats. (A) Representative images of ApoE3 (E3) and ApoE4 (E4) rats treated with doxorubicin (DOXO) or PBS in the dentate gyrus region of the hippocampus. Representative images are merged, and the blue color represents the DAPI and the red color represents the doublecortin (DCX) positive cells. Scale bar = 100 μm. (B) The bar graph shows the quantitation of the DCX-positive cells per mm2 in the dentate gyrus of the hippocampus between all groups, normalized to the ApoE3 PBS group by cohort and displayed as the mean fold change per group. Groups are separated by genotype and treatment (n = 21–25 per group displayed). Data represent mean ± SEM, and biological replicates per group are displayed as an individual circle on the graph. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.001. Ordinary two-way ANOVA full model with Tukey’s multiple comparisons test (B). Transcriptomics alterations after doxorubicin treatment in ApoE3 and ApoE4 rats To investigate the neuropathology underlying the chemotherapy-induced cognitive deficits, RNA-sequencing was performed on two regions of the brain: the hippocampus and the cortex. The hippocampus is a region of the brain that plays an essential role in learning and memory formation29. Less stringent thresholds were utilized to analyze the RNA-sequencing datasets to ensure subtle yet meaningful changes, and each doxorubicin group was compared to the respective PBS-treated group. First, we showed the top upregulated or downregulated genes after doxorubicin treatment in the hippocampus for each comparison, which we illustrated by a volcano plot for ApoE3 (Supplemental Fig. 1A) and ApoE4 (Supplemental Fig. 1B) rats. We found 606 downregulated genes and 135 upregulated genes in the hippocampus of ApoE3 doxorubicin-treated rats (Supplemental Fig. 1A). There were 218 downregulated genes and 131 upregulated genes in the hippocampus of ApoE4 doxorubicin-treated rats (Supplemental Fig. 1B). We also performed the principal component analysis (PCA) on the hippocampus samples of ApoE3 (Supplemental Fig. 1C) and ApoE4 (Supplemental Fig. 1D) rats. To delve deeper into these hippocampal differences, we utilized the IPA platform to identify the canonical pathways dysregulated by doxorubicin treatment in ApoE3 and ApoE4 rats. The IPA comparison analysis feature allowed us to have a side-by-side comparison of ApoE3 and ApoE4 doxorubicin-treated rats (after each of them had been compared to their genotype-matched PBS-treated rats) (Supplemental Fig. 2A). Interestingly, for many of the canonical pathways, ApoE3 and ApoE4 rats had an opposite response after doxorubicin treatment. These pathways included: CREB signaling in neurons, S100 family signaling pathway, synaptic long-term depression, and synaptic long-term potentiation, possibly highlighting ApoE4-specific pathway dysregulation in the hippocampus following doxorubicin exposure. Similarly, we also investigated transcriptomics alterations in a different region of the brain, the cortex, which is a critical region of the brain for high-order functions like decision making. Using a volcano plot, we illustrated the top upregulated or downregulated genes after doxorubicin treatment in the cortex for ApoE3 (Supplemental Fig. 3A) and ApoE4 (Supplemental Fig. 3B) rats. We found 100 downregulated genes and 249 upregulated genes in the cortex of ApoE3 doxorubicin-treated rats (Supplemental Fig. 3A). There were 177 downregulated genes and 145 upregulated genes in the cortex of ApoE4 doxorubicin-treated rats (Supplemental Fig. 3B). We conducted PCA on the cortical samples of the ApoE3 (Supplemental Fig. 3C) and ApoE4 (Supplemental Fig. 3D) groups. We compared the canonical pathways between the cortex and hippocampus in either the ApoE3 or ApoE4 rats (after each of them had been compared to their genotype-matched PBS-treated rats). We observed that the “Glutaminergic Receptor Signaling Pathway (Enhanced)” displayed a positive z-score for the hippocampus and cortex regions of the brain in the ApoE4 doxorubicin-treated rats. However, in the ApoE3 doxorubicin-treated rats, we observed that the “Glutaminergic Receptor Signaling Pathway (Enhanced)” displayed a negative z-score for the hippocampus and cortex regions. Additionally, we observed that the “PPAR Signaling” canonical pathway displayed a positive z-score for the hippocampus and cortex regions of the brain in the ApoE3 doxorubicin-treated rats. Interestingly, we observed that transthyretin (Ttr) was a top downregulated gene in both the hippocampus and cortex regions of the brain in the ApoE4 doxorubicin-treated rats. The impact of doxorubicin treatment on circulating cytokines in serum Circulating markers can be utilized as a rapid measurement of identifying which patients may experience these CICI symptoms. Utilizing serum in preclinical studies is extremely beneficial for translational purposes, as this is a specimen that can be collected regularly during treatment; therefore, these measurements can be integrated into the protocol. Recent literature provides evidence of neurological biomarkers including neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP), and tau. Although these markers are highly expressed in the cerebrospinal fluid (CSF), this is not a routine collection in all pediatric cancers. NfL is a neuron-specific intermediate filament scaffolding protein that is utilized for identifying injury to the nervous system30. GFAP is an intermediate filament of astrocytes and recent evidence supports GFAP as a biomarker of CNS diseases, neurodegenerative diseases31 and measuring the severity of traumatic brain injury (TBI)32. Tau protein plays an important role in neuronal processes and stabilizing microtubules33 and has also been a key biomarker of cognitive decline34. Here, we found a significant increase in serum GFAP concentrations in the ApoE4 doxorubicin-treated rats (Fig. 3A), indicating possible brain damage and cognitive dysfunction. A two-way ANOVA on log-transformed data revealed significant effects of ApoE genotype (F (DFn, DFd), F(1,67) = 9.726, p = 0.0027) on serum GFAP levels, but no significant effects of treatment (F (DFn, DFd), F(1,67) = 2.609, p = 0.1109) and no significant interaction (F (DFn, DFd), F(1,67) = 0.9669, p = 0.3290) (Fig. 3A). In both ApoE3 and ApoE4 rats, doxorubicin treatment induced a significant increase in serum NfL levels, however, there is no difference in the serum NfL levels between the ApoE3 and ApoE4 doxorubicin-treated rats (Fig. 3B). A two-way ANOVA on log-transformed data showed significant effects of treatment (F (DFn, DFd), F(1,67) = 104.8, p < 0.0001) on serum NfL levels, but no significant effects of ApoE genotype (F (DFn, DFd), F(1,67) = 0.6745, p = 0.4144) and no significant interaction (F (DFn, DFd), F(1,67) = 0.01807, p = 0.8935) (Fig. 3B). We observed a non-significant increase in the serum tau levels in the ApoE4 doxorubicin-treated rats compared to all the other groups (Fig. 3C; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,67) = 0.1940, p = 0.6610), ApoE genotype (F (DFn, DFd), F(1,67) = 2.835, p = 0.0969), treatment (F (DFn, DFd), F(1,67) = 0.01760, p = 0.8949)). Next, we found a significant increase in the intercellular adhesion molecule-1 (ICAM-1) serum levels after doxorubicin treatment in the ApoE3 rats, which remained elevated in the ApoE4 rats independent of treatment (Fig. 3D). A two-way ANOVA revealed significant effects of treatment (F (DFn, DFd), F(1,67) = 6.889, p = 0.0107) on serum ICAM-1 levels, but no significant effects of ApoE genotype (F (DFn, DFd), F(1,67) = 3.461, p = 0.0672) and no significant interaction (F (DFn, DFd), F(1,67) = 1.316, p = 0.2554) (Fig. 3D). Elevated ICAM-1 levels have been linked with cognitive dysfunction in glioma patients35. Additionally, we investigated changes in brain-derived neurotrophic factor (BDNF), which has previously been linked with CICI36 and Alzheimer’s disease37. Interestingly, we found lower serum BDNF levels in the ApoE4 rats compared to the ApoE3 rats independent of doxorubicin treatment (Fig. 3E). A two-way ANOVA showed significant effects of ApoE genotype (F (DFn, DFd), F(1,67) = 8.493, p = 0.0048), but no significant effect of treatment (F (DFn, DFd), F(1,67) = 0.7877, p = 0.3780) on serum BDNF levels and no significant interaction (F (DFn, DFd), F(1,67) = 0.5532, p = 0.4596) (Fig. 3E). Lastly, serum levels of calcium binding protein B (S100B), a marker of brain damage38, has also been linked with cognitive impairment in other diseases39–41. However, we found no difference in the serum S100B levels between all the groups (Fig. 3F; two-way ANOVA: interaction (F (DFn, DFd), F(1,67) = 0.08703, p = 0.7689), ApoE genotype (F (DFn, DFd), F(1,67) = 2.054, p = 0.1564), treatment (F (DFn, DFd), F(1,67) = 0.09769, p = 0.7556)). These serum level changes in key translatable markers are vital to this study provided that very limited preclinical studies investigate the long-lasting effects of doxorubicin treatment, and here, we reported changes 5–6 weeks after the last doxorubicin treatment in a pediatric CICI model. Serum markers impacted 5–6 weeks after doxorubicin treatment in ApoE3 and ApoE4 rats. The bar graphs displayed here represent the mean serum concentration after doxorubicin (DOXO) in ApoE3 (E3) and ApoE4 (E4) rats for the following markers: (A) glial fibrillary acidic protein (GFAP) in fg/mL, (B) neurofilament light chain (NfL) in fg/mL, (C) tau in fg/mL, (D) intercellular adhesion molecule-1 (ICAM-1) in ng/mL, (E) brain-derived neurotrophic factor (BDNF) in pg/mL, (F) S100B in pg/mL. Data represent mean ± SEM, and n = 17–18 biological replicates per group that are displayed as an individual circle on the graph. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.001. ordinary two-way ANOVA full model with Tukey’s multiple comparisons test (A, B, D, E). Panels A and B were performed on log-transformed data; and the data in the graphs for panels A and B are presented in original units, however, the pairwise comparison and p-value significance displayed here is based on the results of the log-transformed data. The impact of doxorubicin treatment on oxidative stress markers in the brain of ApoE4 rats Since doxorubicin induces the production of reactive oxygen species (ROS), which may play a role in cytotoxicity and neurotoxicity42–44, we investigated various measures of oxidative stress and lipid peroxidation in the cortex region of the brain after doxorubicin treatment. ROS can induce lipid peroxidation, which can produce byproducts like malondialdehyde (MDA) and be used as an indicator of oxidative stress in tissues45,46. Although we observed higher levels of MDA in ApoE4 rats compared to ApoE3 rats, there was no increased sensitivity to doxorubicin treatment (Supplemental Fig. 4A; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,36) = 0.1312, p = 0.7193), ApoE genotype (F (DFn, DFd), F(1,36) = 8.818, p = 0.0053), treatment (F (DFn, DFd), F(1,36) = 0.2018, p = 0.6560)). Antioxidants may help protect against oxidative stress47; however, we observed no significant differences in total antioxidant capacity (Supplemental Fig. 4B; two-way ANOVA: interaction (F (DFn, DFd), F(1,36) = 0.04803, p = 0.8278), ApoE genotype (F (DFn, DFd), F(1,36) = 0.1173, p = 0.7340), treatment (F (DFn, DFd), F(1,36) = 6.766e-005, p = 0.9935)). Moreover, glutathione is an antioxidant that can mitigate the toxic effects of ROS48,49. ApoE4 doxorubicin-treated rats had slightly higher levels of reduced glutathione in the cortex, although not significant (Supplemental Fig. 4C; two-way ANOVA: interaction (F (DFn, DFd), F(1,35) = 1.207, p = 0.2795), ApoE genotype (F (DFn, DFd), F(1,35) = 0.2104, p = 0.6493), treatment (F (DFn, DFd), F(1,35) = 0.1906, p = 0.6651)). Lastly, in the brain, we investigated whether prostaglandin E2 (PGE2), a lipid mediator that plays an important role in inflammation50, and found no differences between the groups (Supplemental Fig. 4D; two-way ANOVA: interaction (F (DFn, DFd), F(1,27) = 0.4602, p = 0.5033), ApoE genotype (F (DFn, DFd), F(1,27) = 0.02924, p = 0.8655), treatment (F (DFn, DFd), F(1,27) = 3.383, p = 0.0769)). Further, we also explored oxidative stress markers in the serum. Although there were no significant differences between the ApoE4 PBS-treated and ApoE4 doxorubicin-treated, we reported a significant increase in the superoxide dismutase (SOD) levels in the ApoE4 doxorubicin-treated rats compared to the ApoE3 PBS-treated rats (Supplemental Fig. 4E; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,27) = 1.698, p = 0.2036), ApoE genotype (F (DFn, DFd), F(1,27) = 7.565, p = 0.0105), treatment (F (DFn, DFd), F(1,27) = 3.677, p = 0.0658)). Our results suggest that oxidative stress is possibly increased in ApoE4 rats; however, 5–6 weeks after the last doxorubicin treatment, there were no significant differences in oxidative stress markers. The impact of doxorubicin treatment on key neurobiological markers and signaling pathways in the brain To delve into the mechanism of doxorubicin-induced cognitive impairment, we tested neurodegenerative markers in both the hippocampus and cortex regions of the brain. A decrease in plasma brain-derived neurotropic factor (BDNF) is linked to cognitive impairment in cancer patients treated with doxorubicin51,52. Doxorubicin decreases BDNF levels in the hippocampus of mice36. It is possible that a decrease in neurogenesis can be related to BDNF. We found that in the ApoE4 doxorubicin-treated rats, the levels of BDNF levels in the hippocampus were lower compared to the ApoE4 PBS-treated and ApoE3 doxorubicin-treated rats (Supplemental Fig. 5A; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 9.046, p = 0.0070), ApoE genotype (F (DFn, DFd), F(1,20) = 0.4441, p = 0.5127), treatment (F (DFn, DFd), F(1,20) = 0.2445, p = 0.6264)). However, in the cortex, we observed slightly elevated BDNF levels for the ApoE4 rats compared to the ApoE3 rats, independent of doxorubicin treatment (Supplemental Fig. 6A; two-way ANOVA: interaction (F (DFn, DFd), F(1,12) = 0.01006, p = 0.9217), ApoE genotype (F (DFn, DFd), F(1,12) = 5.117, p = 0.0430), treatment (F (DFn, DFd), F(1,12) = 0.4537, p = 0.5133)). Another important regulator that might play a role in cognitive impairment is cAMP Response Element-Binding Protein (CREB)53. In the hippocampus, we observed slightly higher CREB levels in the ApoE4 rats, however, not significant (Supplemental Fig. 5B; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 0.04527, p = 0.8337), ApoE genotype (F (DFn, DFd), F(1,20) = 2.087, p = 0.1641), treatment (F (DFn, DFd), F(1,20) = 0.02344, p = 0.8798)). In the cortex of the ApoE4 doxorubicin-treated rats, we observed lower protein levels of CREB (Supplemental Fig. 6B; two-way ANOVA: interaction (F (DFn, DFd), F(1,12) = 1.227, p = 0.2897), ApoE genotype (F (DFn, DFd), F(1,12) = 1.771, p = 0.2080), treatment (F (DFn, DFd), F(1,12) = 6.526, p = 0.0252)). However, we found no differences in phospho-CREB levels in the cortex (Supplemental Fig. 6C; two-way ANOVA: interaction (F (DFn, DFd), F(1,36) = 0.04686, p = 0.8298), ApoE genotype (F (DFn, DFd), F(1,36) = 2.034, p = 0.1624), treatment (F (DFn, DFd), F(1,36) = 1.079, p = 0.3060)). Surprisingly, in the hippocampus, we found marginally elevated levels of total tau in the ApoE4 rats, however, doxorubicin does not further increase the levels (Supplemental Fig. 5C; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 0.005285, p = 0.9428), ApoE genotype (F (DFn, DFd), F(1,20) = 4.968, p = 0.0375), treatment (F (DFn, DFd), F(1,20) = 0.03728, p = 0.8488)). However, in the cortex, we observed a trend towards a significant decrease of total tau in ApoE4 doxorubicin-treated rats (Supplemental Fig. 6D; two-way ANOVA: interaction (F (DFn, DFd), F(1,12) = 5.271, p = 0.0405), ApoE genotype (F (DFn, DFd), F(1,12) = 0.9401, p = 0.3514), treatment (F (DFn, DFd), F(1,12) = 1.262, p = 0.2832)). Lastly, we investigated the expression of transthyretin (Ttr), which may play a protective role in neurodegenerative disease54. Our transcriptomic analyses displayed a decrease in the Ttr gene in the ApoE4 doxorubicin-treated rats when compared to ApoE4 PBS-treated. However, we found no TTR protein level differences in both the hippocampus (Supplemental Fig. 5D; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 0.0004360, p = 0.9835), ApoE genotype (F (DFn, DFd), F(1,20) = 9.349, p = 0.0062), treatment (F (DFn, DFd), F(1,20) = 0.4190, p = 0.5248)) and in the cortex (Supplemental Fig. 6E; two-way ANOVA: interaction (F (DFn, DFd), F(1,12) = 10.20, p = 0.0077), ApoE genotype (F (DFn, DFd), F(1,12) = 1.408, p = 0.2583), treatment (F (DFn, DFd), F(1,12) = 0.1023, p = 0.7546)). Next, we investigated whether p38 mitogen activated protein kinase (p38 MAPK) is activated in the brain post-doxorubicin treatment. As highlighted previously, p38 MAPK plays a pivotal role in inflammatory stress response55. In human cerebral microvascular endothelial cells, we previously found a robust activation of p38 MAPK after doxorubicin treatment56. There is evidence that ApoE4 mice showed activation of p38 MAPK in the hippocampus57. We found that a slight increase in the phosphorylation of p38 MAPK in the cortex of ApoE4 doxorubicin-treated rats compared to all other groups (Fig. 4A and 4B; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,20) = 0.3424, p = 0.5650), ApoE genotype (F (DFn, DFd), F(1,20) = 2.150, p = 0.1581), treatment (F (DFn, DFd), F(1,20) = 4.240, p = 0.0527). Doxorubicin induces double-stranded breaks through its inhibition of topoisomerase II. Hence, we investigated whether doxorubicin treatment induces DNA damage in the brain by assessing whether histone H2A variant will be phosphorylated in response to double-stranded DNA breaks (γH2AX)58. We discovered an increase with doxorubicin treatment alone in the ApoE3 rats in γH2AX protein expression. However, in the ApoE4 rats, doxorubicin treatment further induced a significant increase in γH2AX protein expression in the cortex (Fig. 4A and 4C). A two-way ANOVA revealed significant effects of ApoE genotype (F (DFn, DFd), F(1,20) = 59.40, p < 0.0001) and significant effects of treatment (F (DFn, DFd), F(1,20) = 13.34, p = 0.0016) on γH2AX protein expression in the cortex region of the brain, but no significant interaction (F (DFn, DFd), F(1,20) = 0.2039, p = 0.6565) (Fig. 4C). Given that we saw the difference with p38 MAPK, we next delved into investigating extracellular signal-regulated kinase (ERK), which is a pathway activated upon DNA-damaging agents, and it has been implicated in many neurodegenerative pathologies59. Although not statistically significant, we observed that ERK is robustly activated in the hippocampus of ApoE3 doxorubicin-treated rats. While harboring the ApoE4 genotype also increases the activation of ERK, doxorubicin treatment does not further exacerbate this activation (Supplemental Fig. 7A and Supplemental Fig. 7B; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,43) = 0.06662, p = 0.7976), ApoE genotype (F (DFn, DFd), F(1,43) = 0.8266, p = 0.3683), treatment (F (DFn, DFd), F(1,43) = 1.428, p = 0.2387)). Further, the role of myelination in a juvenile model of CICI remains underexplored. There is also evidence that p38 MAPK acts as a negative regulator of myelination60. We examined the hippocampal region of the brain to determine whether myelination is impacted, however, we found no significant differences (Supplemental Fig. 7C and Supplemental Fig. 7D; two-way ANOVA: interaction (F (DFn, DFd), F(1,43) = 0.7221, p = 0.4002), ApoE genotype (F (DFn, DFd), F(1,43) = 0.07064, p = 0.7917), treatment (F (DFn, DFd), F(1,43) = 0.3733, p = 0.5444)). Taken together, we observed increased protein expression of γH2AX and activated p38 MAPK in the brain after doxorubicin treatment in ApoE4 rats. The impact of doxorubicin on p38 MAPK signaling and γH2AX expression in the cortex of ApoE3 and ApoE4 rats. (A) Representative Western Blots for the cortex region of the brain of ApoE3 (E3) and ApoE4 (E4) rats treated with doxorubicin (DOXO) or PBS for the following proteins: phosphorylated p38 mitogen activated protein kinase (MAPK) Thr180/Tyr182 (T180/Y182) (p-p38 MAPK), total p38 MAPK, phosphorylated Histone H2AX S139 (γH2AX) and β-Actin, with the molecular weight in kDa on the left side of the blot. Uncropped blots are presented in Supplemental Figs. 10–14. (B) The bar graph shows the protein expression of p-p38 MAPK, which is calculated by dividing p-p38 MAPK by total p38 MAPK and then normalized to the ApoE3 PBS-treated rats and displayed as fold change. The bar graph shows the protein expression for (C) γH2AX and is normalized to the loading control, β-Actin, and displayed as fold change to ApoE3 PBS-treated rats. Panels (B, C) show western blot quantifications that include male and female rats. The groups in panels (B, C) are separated by genotype and treatment (n = 6 per group displayed, where biological replicates are shown as an individual circle on the graph). Data represent mean ± SEM. *p < 0.05, ***p < 0.001, ****p < 0.001. ordinary two-way ANOVA full model with Tukey’s multiple comparisons test (B, C). For Panel B, the ordinary two-way ANOVA full model Tukey’s multiple comparisons test was performed on log-transformed data. The data in the Panel B is presented as back-transformed values, however, the pairwise comparison and p-value displayed here is based on the results of the log-transformed data. The impact of doxorubicin treatment on cardiotoxicity in ApoE3 and ApoE4 rats Neurotoxicity is not the only debilitating side effect of chemotherapy treatments, but cardiotoxicity is also highly relevant in patients with doxorubicin treatment, therefore, we also investigated cardiac tissue after doxorubicin treatment in ApoE3 and ApoE4 rats. We found no significant difference in the serum creatine kinase levels among the groups (Supplemental Fig. 8A; two-way ANOVA: interaction (F (DFn, DFd), F(1,27) = 0.6770, p = 0.4178), ApoE genotype (F (DFn, DFd), F(1,27) = 0.5972, p = 0.4464), treatment (F (DFn, DFd), F(1,27) = 0.04891, p = 0.8266)). Surprisingly, we found a stepwise increase in the serum tissue inhibitor of metalloproteinase-1 (TIMP-1) levels, with the most significant increase in the ApoE4 doxorubicin-treated rats (Supplemental Fig. 8B; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,67) = 2.987, p = 0.0885), ApoE genotype (F (DFn, DFd), F(1,67) = 19.39, p < 0.0001), treatment (F (DFn, DFd), F(1,67) = 11.82, p = 0.0010)). An increase in TIMP-1 may provide neuroprotection and attenuate cognitive dysfunction61. However, TIMP-1 is largely known to be used as a marker for fibrosis62,63 and is utilized as a marker of cardiotoxicity. To determine the critical mediators that are impacted by doxorubicin treatment in the heart tissue, we investigated phosphorylation changes in proteins that have been implicated in doxorubicin-induced cardiotoxicity. p38 MAPK plays an important role in the inflammatory stress response and production of ROS. However, we observed no significant difference in the p38 MAPK phosphorylation pattern (Supplemental Fig. 8C; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 0.2295, p = 0.6371), ApoE genotype (F (DFn, DFd), F(1,20) = 0.1386, p = 0.7136), treatment (F (DFn, DFd), F(1,20) = 0.1795, p = 0.6763)). ERK is known to be activated by DNA-damaging agents, and doxorubicin is a DNA-damaging agent. Interestingly, we found that doxorubicin treatment slightly increases the phosphorylation of ERK in ApoE3 rats, and we further saw a significant increase in ERK phosphorylation in the ApoE4 doxorubicin-treated rats compared to the ApoE3 PBS-treated rats (Supplemental Fig. 8D; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 0.001844, p = 0.9662), ApoE genotype (F (DFn, DFd), F(1,20) = 7.714, p = 0.0116), treatment (F (DFn, DFd), F(1,20) = 4.315, p = 0.0509)). Lastly, we performed histopathological assessment on the heart tissue to determine whether harboring the ApoE4 genotype increases sensitivity to cardiac damage after doxorubicin treatment. We found increased interstitial fibrosis after doxorubicin treatment in both the ApoE3 and ApoE4 rats, but no clear evidence of cardiac myocyte degeneration, necrosis, hypertrophy, or inflammatory cell infiltration by H&E (Supplemental Fig. 8E) and collagen staining (Supplemental Fig. 8F). All rats from the ApoE4 doxorubicin-treated rats evaluated for cardiac interstitial fibrosis scoring had a score of 2, possibly indicating increased sensitivity (Supplemental Table 1). The ApoE3 PBS-treated rats had no interstitial fibrosis and a lower score than ApoE4 PBS-treated rats (Supplemental Table 1). These studies evaluating doxorubicin-induced cardiotoxicity imply ApoE4 as a potential risk factor, however, further investigation is needed. The impact of doxorubicin treatment on hepatotoxicity in ApoE3 and ApoE4 rats Similar to neurotoxicity and cardiotoxicity, hepatoxicity is also a side effect impacting patients treated with doxorubicin. To deepen our understanding of genetic susceptibility in doxorubicin-induced hepatoxicity, we assessed the serum and liver tissue for key markers. In male rats specifically, serum triglyceride levels were increased in ApoE3 doxorubicin-treated rats, and harboring the ApoE4 genotype alone did not further increase triglyceride levels (Supplemental Fig. 9A; three-way ANOVA: sex (F (DFn, DFd), F(1,23) = 49.88, p < 0.0001), ApoE genotype (F (DFn, DFd), F(1,23) = 1.600, p = 0.2185), treatment (F (DFn, DFd), F(1,23) = 7.305, p = 0.0127, sex x ApoE genotype (F (DFn, DFd), F(1,23) = 0.3213, p = 0.5763), sex x treatment (F (DFn, DFd), F(1,23) = 8.380, p = 0.0082), ApoE genotype x treatment (F (DFn, DFd), F(1,23) = 12.80, p = 0.0016), sex x ApoE genotype x treatment (F (DFn, DFd), F(1,23) = 1.481, p = 0.2359)). To assess liver toxicity, we utilized a common marker of liver damage in the serum, alanine transaminase (ALT). Interestingly, we observed that only in ApoE3 male rats were the ALT levels significantly higher after doxorubicin treatment (Supplemental Fig. 9B; three-way ANOVA: sex (F (DFn, DFd), F(1,23) = 23.58, p < 0.0001), ApoE genotype (F (DFn, DFd), F(1,23) = 0.001910, p = 0.9655), treatment (F (DFn, DFd), F(1,23) = 6.663, p = 0.0167, sex x ApoE genotype (F (DFn, DFd), F(1,23) = 0.001910, p = 0.9655), sex x treatment (F (DFn, DFd), F(1,23) = 6.663, p = 0.0167), ApoE genotype x treatment (F (DFn, DFd), F(1,23) = 3.217, p = 0.0860), sex x ApoE genotype x treatment (F (DFn, DFd), F(1,23) = 0.5533, p = 0.4645)). Although not significant, in ApoE4 male rats, we found an increase in ALT levels after doxorubicin treatment, but not to the same magnitude as ApoE3 rats (Supplemental Fig. 9B). Similar to the heart tissue evaluation, in liver tissue, we explored the same phospho-proteins; however, we observed no significant differences, although we observed a trend towards a significant decrease in the phosphorylation of p38 MAPK (Supplemental Fig. 9C; two-way ANOVA on log-transformed data: interaction (F (DFn, DFd), F(1,20) = 1.823, p = 0.1921), ApoE genotype (F (DFn, DFd), F(1,20) = 3.125, p = 0.0924), treatment (F (DFn, DFd), F(1,20) = 0.02855, p = 0.8675)) and ERK (Supplemental Fig. 9D; two-way ANOVA: interaction (F (DFn, DFd), F(1,20) = 0.1720, p = 0.6828), ApoE genotype (F (DFn, DFd), F(1,20) = 2.538, p = 0.1268), treatment (F (DFn, DFd), F(1,20) = 1.580, p = 0.2232)) in the ApoE4 doxorubicin-treated rats. We also performed histopathological assessment on the liver tissue. We found no evidence of hepatocyte degeneration, necrosis, abnormal central veins or abnormal congestion of the central vein by H&E staining after doxorubicin treatment in ApoE3 and ApoE4 rats (Supplemental Fig. 9E). Although we found a clear increase in collagen staining around the liver capsule after doxorubicin treatment and expansion of the portal tracts by fibrosis, there was no difference between the ApoE3 and ApoE4 doxorubicin-treated groups (Supplemental Fig. 9F and Supplemental Table 1). These studies evaluating doxorubicin-induced hepatotoxicity suggest that ApoE4 may not be a potential risk factor. Discussion Over the past decade, many preclinical studies have aimed to understand the interpatient variability observed in CICI by exploring the role of the ApoE4 genotype using adult rodent models 16,64,65. However, the contribution of ApoE4 as a pediatric genetic risk factor for CICI has not previously been investigated in preclinical studies, despite clinical evidence that childhood cancer survivors with the ApoE4 allele are at greater risk of CICI14. To address this gap, we developed a juvenile ApoE4 CICI model including both male and female rats to investigate sub-acute effects of doxorubicin treatment. While most preclinical models of CICI investigate the neuropathological changes shortly after the final chemotherapy dose, our model examines these changes 5–6 weeks after the last doxorubicin dose, providing insights into more persistent and delayed consequences of doxorubicin treatment. Given the urgent need for susceptibility or risk biomarkers in CICI, our study provides valuable insights into the potential genetic and associated molecular mediators. Here, we reported that ApoE4 doxorubicin-treated rats were more likely to exhibit both visual and spatial memory impairments than ApoE3 doxorubicin-treated rats when compared to their genotype-matched PBS-treated control rats. We demonstrated a role for ApoE4 in doxorubicin-related CICI and showed that the mechanism may relate to the reduction in hippocampal neurogenesis. Transcriptomic analyses revealed differentially regulated pathways following doxorubicin treatment, with several pathways showing opposite trends between ApoE3 and ApoE4 rats. We further analyzed translatable serum markers of neurotoxicity and reported higher serum levels of GFAP in the ApoE4 doxorubicin-treated rats. Moreover, we found an increase in p38 MAPK activation and γH2AX protein expression in the cortex of ApoE4 doxorubicin-treated rats. Overall, ApoE genotype contributes to differential susceptibility to doxorubicin-induced toxicity. Within this study, doxorubicin treatment led to weight loss in male rats, with significant weight loss in ApoE4 doxorubicin-treated male rats compared to the age-matched ApoE4 PBS-treated male rats. Other studies reported weight loss after doxorubicin treatment in male rats, supporting the consistency with our study66,67. However, whether harboring the ApoE4 genotype increases sensitivity to weight loss is not known. Further, we observed increased sensitivity to visual and spatial memory impairment in the ApoE4 doxorubicin-treated rats. This is consistent with another study that reported spatial memory deficits in ApoE4 doxorubicin-treated female mice15. As expected and hypothesized, ApoE3 doxorubicin-treated rats exhibited some degree of visual and spatial memory deficits, which is consistent with other studies reporting a doxorubicin-induced decrease in spatial memory in wild-type rats without a genetic risk background18,19. In our study, ApoE4 doxorubicin-treated rats showed a clearly heightened sensitivity for visual and spatial memory deficits, highlighting the importance of further exploring the role of genetic susceptibility. In terms of neuropathological changes, we observed a significant reduction in hippocampal neurogenesis among ApoE4 rats 5–6 weeks after the last doxorubicin exposure. Doxorubicin does not cross the BBB68, yet through indirect mechanisms suppresses hippocampal neurogenesis. In 2-month-old male athymic nude rats, injected with a similar dose of doxorubicin, showed cognitive impairment and displayed a significant decrease in neurogenesis22. While we also observed a slight reduction in DCX-positive staining in the ApoE3 rats after doxorubicin treatment, there was a substantially greater reduction in the ApoE4 doxorubicin-treated rats. As previously described, spatial memory is a hippocampus-dependent test. While the relationship between the spatial memory impairment and hippocampal neurogenesis was not directly established in this study, these parallel trends imply an association between increased sensitivity to spatial memory impairments and impaired hippocampal neurogenesis in ApoE4 doxorubicin-treated rats. Attention issues in survivors of childhood ALL harboring the E4 allele of ApoE were associated with changes in GFAP levels in the CSF14,69. Blood GFAP levels were higher in patients with mild cognitive impairment or Alzheimer’s disease when compared to the healthy controls70. Blood GFAP levels have also been correlated with dementia and cognitive decline71. Plasma GFAP and NfL may serve as early biomarkers for dementia72. Similarly, plasma tau also serves as a biomarker for dementia73. Although different pathologies, these studies relate to cognitive decline. These clinical findings align with our preclinical results, in which we reported an increase in GFAP and tau serum levels 5–6 weeks after doxorubicin treatment in ApoE4 rats. However, we observed a robust increase in serum NfL levels after doxorubicin treatment, regardless of the genotype. Clinical studies demonstrated that NfL levels can be used as a measure to monitor chemotherapy-induced peripheral neuropathy (within a different class of chemotherapy)74,75. Further investigation is required to determine the onset of these serum level changes and how long these serum levels changes will persist. Further, we observed lower BDNF serum levels in ApoE4 rats independent of doxorubicin treatment. This is consistent with clinical studies, in which female Alzheimer’s Disease patients had lower serum BDNF levels in APOE4 carriers compared to non-APOE4 carriers76. Although we observed no significant differences among the groups in oxidative stress markers after doxorubicin treatment, we reported elevated MDA levels in the ApoE4 rats, while doxorubicin does not further heighten this response. This finding is consistent with prior reports showing increased MDA levels in ApoE4 mice cortices compared to ApoE3 mice77. However, our finding is inconsistent with many prior preclinical studies that reported an increase in oxidative stress markers within the brain after doxorubicin treatment. Male doxorubicin-treated rats demonstrated increased lipid peroxidation and decreased GSH in both the cortex and hippocampus18. While we found no significant changes in the total antioxidant capacity, others reported that doxorubicin treatment increased antioxidant enzyme levels20. We speculate that the lack of differences in oxidative stress markers may be due to the timing of our analysis, which we conducted 5–6 weeks post the last doxorubicin injection. Although a different chemotherapeutic agent, oxaliplatin treatment in the rat brain endothelial cells (RBE4) presented an increase in ROS initially within 6 h, followed by a reduction in ROS production78. This suggests that the oxidative stress response should be measured earlier. Our transcriptomic analyses revealed a significant decrease in Ttr transcript level in the hippocampus and cortex of ApoE4 doxorubicin-treated rats compared to ApoE4 PBS-treated rats. However, at the protein level analysis, we did not find any differences in the TTR expression in the hippocampus, although the TTR levels were slightly elevated in the ApoE4 rats, independent of doxorubicin treatment. In contrast, the cortex, we found a decrease in the TTR protein levels in the ApoE4 doxorubicin-treated rats. In the brain, TTR is synthesized by the choroid plexus79, and TTR might play a protective role in Alzheimer’s disease by mitigating the aggregation of amyloid beta in the brain79. Our results are consistent with previous studies that reported a decrease in the Ttr gene for ApoE4 doxorubicin-treated female mice at both 7 days and 21 days following treatment65. Here, we collected samples 5–6 weeks following the final doxorubicin exposure, and it is possible that at earlier time points, we would be able to better capture TTR protein level changes. We speculate that the decrease in TTR at both transcript and protein levels observed in ApoE4 doxorubicin-treated rats may contribute to the increased sensitivity seen in our study, as the loss of TTR’s protective function may exacerbate the susceptibility to CICI and requires further investigation. Our results suggest further exploration of the role of TTR at earlier timepoints. In our study, we found robust activation of γH2AX and an increase in p38 MAPK activation in the ApoE4 doxorubicin-treated rats. Doxorubicin does not cross the BBB, yet 5–6 weeks following the last doxorubicin treatment, there was a significant increase in γH2AX protein expression, indicating DNA damage within the brain. While we were not able to decipher the exact mechanism resulting in DNA damage, we hypothesized that the BBB disruption may contribute to the heightened DNA damage response. Moreover, we found that doxorubicin does not significantly alter the myelination reactions. This finding suggests that the pathophysiology is different between methotrexate-induced neurotoxicity and doxorubicin-induced neurotoxicity. Preclinical studies showed that methotrexate is associated with white matter changes80–82, which are presumably due to inhibition of methylation reactions and result in impaired myelination. Here, we not only reported assessment of neurotoxicity, but also cardiotoxicity and hepatotoxicity. In the heart tissue, we observed ERK activation in the ApoE4 doxorubicin-treated rats. We also observed a significant increase in serum TIMP-1 levels, indicating fibrosis and cardiotoxicity62,63. Although there is evidence that an increase in TIMP-1 may provide neuroprotection and attenuate cognitive dysfunction61, it is primarily used as a cardiotoxicity marker. Immunohistochemical staining suggested an increase in interstitial fibrosis in cardiac tissue after doxorubicin treatment. Our results indicate that ApoE4 doxorubicin-treated rats have increased sensitivity to cardiotoxicity. Moreover, our hepatotoxicity investigations did not reveal genotype-specific differences. Doxorubicin treatment significantly increased serum triglyceride and ALT levels in male ApoE3 rats, and these levels remained high in the ApoE4 rats regardless of treatment, suggesting a potential sex-dependent and genotype-dependent role in these markers. However, our sample size was small and future studies will utilize a larger cohort to study these serum level changes. Additionally, in the liver, we observed a clear increase in collagen staining 5–6 weeks after doxorubicin treatment, indicating liver fibrosis and hepatotoxicity, but no genotype-dependent differences were found. This is consistent with other studies that reported an increase in liver fibrosis after doxorubicin treatment83,84. Future studies to evaluate cardiotoxicity and hepatotoxicity will include assessment of common serum markers like troponin and liver-type arginase 1 (ARG1). While our study provides valuable insights into how the ApoE genotype contributes to increased vulnerability of CICI after doxorubicin treatment in a juvenile rat model, we have many limitations to note. First, our behavioral tests included only OR and OP tests to detect cognitive impairment, and this does not allow for a comprehensive assessment of all cognitive domains. Future studies should also include the 5-choice serial reaction time task (5-CSRTT). Second, we did not find any differences in oxidative stress markers within the brain, which may be attributed to the timing of brain tissue collection, which was 5–6 weeks after the last doxorubicin treatment, and it is likely that at this time, the acute oxidative stress response is resolved. Future studies will include earlier timepoints for tissue collection and incorporate more assessments of oxidative stress, such as fluorescence-activated cell sorting (FACS) using a redox-sensitive fluoroprobe dye. Third, some of the neurology markers we selected in this study lacked sensitivity and did not detect subtle differences. Another potential future direction for this study that our laboratory aims for is employing non-invasive techniques, as it allows for more translatable research in the clinical setting. Future studies will employ dynamic contrast-enhanced magnetic resonance imaging, which provides an influx constant to assess the BBB leakage by allowing the calculation of the K-trans. Furthermore, a serial section transmission electron microscopy (ssTEM) will provide a visualization representation of the BBB breakdown85. Follow-up studies will also explore modified dosing regimens to better recapitulate the neurotoxicity and include intermediate tissue collection points. Conclusions Collectively, our experimental outcomes provide a better understanding of the molecular mechanisms and genetic susceptibility related to doxorubicin-induced cognitive impairment. We developed a doxorubicin-induced cognitive impairment ApoE4 juvenile rat model including both male and female rats. ApoE4 doxorubicin-treated rats were more likely to exhibit both visual and spatial memory deficits. Doxorubicin treatment reduced hippocampal neurogenesis, and this neuropathological change was exacerbated in ApoE4 rats, suggesting genotype-dependent vulnerability. Serum levels of GFAP were increased 5–6 weeks following doxorubicin treatment in ApoE4 rats, suggesting its potential to serve as biomarker of doxorubicin-induced cognitive impairment. These findings help identify a subset of patients more susceptible to CICI. To our knowledge, our behavioral research and associated neuropathological changes in a juvenile rat model with the combination ApoE4 and doxorubicin has not been previously conducted. Our study was the first to explore doxorubicin-induced sub-acute changes in an ApoE4 juvenile model with both male and female rats. Taken together, the ApoE genotype contributes to differential vulnerability to doxorubicin-induced toxicity. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Acknowledgements We would like to acknowledge past and current members of the Cole Laboratory who may have assisted in animal injections, dissections, and/or provided insights during discussions regarding this project, including Jeremy Willekens and Shengguo Li. Figure 1A was created in Biorender.com. Abbreviations 5-Choice serial reaction time task Acute lymphoblastic leukemia Alanine transaminase Apolipoprotein E Blood–brain barrier Brain-derived neurotropic factor Chemotherapy-induced cognitive impairment CAMP Response Element-Binding Protein Cerebrospinal fluid Doublecortin Doxorubicin Extracellular signal-regulated kinase Fluorescence-activated cell sorting Glial fibrillary acidic protein Intercellular Adhesion Molecule-1 International Cancer and Cognition Task Force Mitogen-activated protein kinase Myelin basic protein Malondialdehyde Macrophage inflammatory protein 1 alpha Magnetic resonance imaging Neurofilament light chain Open field Object placement Object recognition Reactive oxygen species Superoxide dismutase Serial section transmission electron microscopy Calcium Binding Protein B Traumatic brain injury Tissue inhibitor of metalloproteinase-1 Author contributions Conceptualization: C.P, P.C.; Methodology: C.P.; Formal analysis: C.P., B.D.; Investigation: C.P., F.D.; Data curation: C.P.; Figure Preparation: C.P.; Writing—original draft: C.P.; Writing—review & editing: C.P., P.C., F.D., B.D.; Supervision: P.C.; Funding acquisition: C.P., P.C. Funding We acknowledge funding from the New Jersey Commission of Cancer Research Predoctoral Fellowship COCR24PRF009, the Biotechnology Training Program NIH T32 GM135141, and the New Jersey Pediatric Hematology and Oncology Research Center of Excellence (NJ PHORCE) Pilot Grant. Data availability The RNA sequencing data discussed in this publication has been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE305230 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE305230). Declarations Competing interests The author(s) declare that they have no competing interests. Ethics approval and consent to participate All animal studies were performed in compliance with the Animal Institute and Use Committee of Rutgers University and were conducted following the “Guide for the Care and Use of Laboratory Animals.” Consent for publication Not applicable. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The RNA sequencing data discussed in this publication has been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE305230 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE305230).