Introduction

In 2022, there were approximately 4.1 million women with a history of breast cancer living in the United States, and it is estimated that nearly 317,000 new cases of breast cancer will be reported in 20251,2. Female breast cancer makes up approximately 15.5% of new cancer cases, and is the second most diagnosed cancer in United States women behind non-melanoma skin cancer2. There is a 13% chance of being diagnosed with breast cancer over the duration of a woman’s lifetime, with the greatest risk occurring between the ages of 55–74 and a five-year survival rate of 90.8%2. Current and ongoing advances in detection and treatment of breast cancer have led to increased survival rates. However, research into improving quality of life (QoL) and long-term management of patients and their mental health is still relatively nascent3.

Breast cancer survivors are at an increased risk for depression, anxiety, suicidal behaviors, and relationship issues compared to women with no history of breast cancer. Survivors also report issues with fatigue, sleep disturbance, muscle or joint aches, and cognitive difficulties4. Side effects in patients vary based on the treatment received; lingering pain, lymphedema, neuropathy, infertility or early menopause, cardiovascular disease, and osteoporosis are among reported long term side effects5. These adverse side effects can impact the QoL of survivors3,6. QoL is multifaceted, but generally focuses on how a patient’s physical, mental, social, and autonomy are changed during and after cancer interventions7. QoL is often negatively affected by chemotherapy induced cognitive impairment (CICI). More commonly referred to as “chemobrain” or “chemo-fog,” CICI is characterized by an impairment in a patient’s memory, learning, concentration, executive function, and attention during and following chemotherapy treatment8,9. Chemobrain was first reported as a possible side effect of chemotherapy treatment in the 1970 s, and today it is estimated that 17–78% of patients undergoing chemotherapy report having symptoms of CICI10,11. Objective neuropsychological tests are typically used to measure cognitive impairment, but they may not detect the oftentimes subtle impairment experienced in CICI12. Clinically, the Functional Assessment of Cancer Therapy-Cognition (FACT-Cog) test was developed as a tool to assess CICI, but it lacks established cut-off scores to define diagnoses, making it hard to compare results across studies13,14. The inability to properly diagnose CICI limits treatment options to managing co-occurring symptoms rather than addressing the root cause10.

One of the standard treatments for breast cancer is four cycles of doxorubicin, brand name Adriamycin, (DOX) and cyclophosphamide (CYP) (AC), followed by four doses of paclitaxel (PTX)15,16. This combination therapy is often referred to as AC-T regimen. CYP is an alkylating agent that disrupts cellular function by creating cross linkage in RNA and DNA strands17,18. Its small molecular size allows it to cross the blood brain barrier (BBB). Disruption of the BBB can lead to neuroinflammation, synaptic dysfunction, and impaired cognitive health19,20. CYP can induce oxidative stress, inflammatory pathways, and activate proinflammatory cytokines associated with cognitive impairment21. DOX, an isolation from Streptomyces peucetius in the 1960’s, was found to have the ability to cause intercalation of DNA and RNA strands. It also interferes with the ability of topoisomerase II to repair DNA, resulting in apoptosis of effected cells22,23. PTX acts by promoting assembly of stable microtubules which inhibits depolymerization, leading to cell cycle arrest and apoptosis24. PTX has been shown to cause brain microvascular endothelial cell senescence which disrupts the BBB19. The addition of taxanes (PTX and docetaxel (DTX)) in conjunction with AC treatment has been shown to significantly improve the efficacy of treatment and the overall survival among women with breast cancer compared to just AC alone16,25, yet few studies exist that explore the effects of the AC-T combination together.

Chemotherapy-associated intestinal inflammation and subsequent breakdown of the mucosal barrier has been one of the major challenges in clinical cancer treatment26. Mucositis has been attributed to the high physiological proliferation and turnover of cells in the intestine that is affected by cancer chemotherapy26,27. Chemotherapeutic agents have also been linked to inducing gut microbiome disturbances that can contribute to mental health conditions28. PTX was reported to induce mice behavior issues such as fatigue and cognitive dysfunction together with low microglia immunoreactivity in addition to an increase in inflammatory markers gene expression in the brain29. Alterations in colonic tissue structure were also reported to be correlated with microglial activation in areas of the brain such as the dentate gyrus and prefrontal cortex30. DTX has also been reported to have neurotoxic effects such as increased neuroinflammation, and oxidative stress in rats31. DTX has been implicated in disrupting intestinal barrier integrity and altering gut microbiota composition, leading to increased intestinal permeability and shifts in microbial diversity and abundance. These changes may contribute to systemic inflammation and neuroimmune signaling alterations that underlie chemotherapy induced cognitive impairment and fatigue32. Although a clear connection exists between the gut microbiome and cognitive function, the details of the process are not fully understood.

Many studies have reported that dysbiosis of normal gut flora plays a key role in modulating different neurological and psychological disorders such as anxiety and depression33, autism spectrum disease34,35, Alzheimer’s disease (AD)36,37,38, Parkinson’s disease39, multiple sclerosis40, and stroke41. A cross talk between gut microbiome and central nervous system also known as Gut-Brain-Axis is getting increased attention primarily to investigate the role of gut microbiome in modulating brain functions, behavior, and cognition42,43,44. The gut-brain axis acts as a communication channel that establishes a reciprocal relationship between the gut and the central nervous system. The gut microbiome can affect brain function, and reciprocally, the brain can affect microbial composition45. The gut microbiota plays a pivotal role in regulating neurochemical signaling pathways that influence brain function and behavior. Certain bacterial species are capable of producing neurotransmitters and neuroactive compounds, including gamma-aminobutyric acid (GABA), serotonin (5-HT), dopamine, norepinephrine, and acetylcholine, which can act locally on the enteric nervous system (ENS) or communicate with the central nervous system (CNS) through neural, endocrine, and immune pathways46. The gut microbiota also contribute to the regulation of cerebrovascular integrity by producing short-chain fatty acids (SCFAs) such as butyrate and propionate, which serve as energy sources for endothelial cells and help maintain the BBB by strengthening tight-junction proteins and reducing neuroinflammation47. A comprehensive understanding of the mechanisms underlying gut–brain axis communication is crucial for developing novel microbiome-based therapeutic strategies aimed at preventing or treating a range of neurological and cognitive disorders.

It has been hypothesized that proposed causes of CICI include disruption of the BBB, oxidative stress and inflammatory response, and alterations in neuronal activity48. Presently, most studies focus on the cognitive and dysbiosis caused by one chemotherapy; however, over 50% of patients with node-positive breast cancer receive combination chemotherapy treatment49,50. In the current study, we use behavioral, proteogenomic, and microbial data to better understand the cognitive deficits and associated brain and gut microbial changes after combination chemotherapy treatment using clinically relevant doses of AC-T.

Methods

Ethical approval and statements

Animal studies reported in this manuscript complied with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines 2.0 (https://arriveguidelines.org/arrive-guidelines) to ensure methodological rigor and reproducibility of methods and findings. All animal procedures (protocol #4178) were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Arkansas for Medical Sciences in compliance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals, the National Institutes of Health Guide for the Care and Use of Laboratory Animals (2011) as well as the American Veterinary Medical Association Animal Euthanasia Guidelines (2020). All the experiments and methods were performed in accordance with the relevant guidelines and regulations.

Animals

Twelve-week-old female C57/BL6J mice were purchased from Jackson Laboratory. Female mice were used as this was a breast cancer drug paradigm, and diagnosis of breast cancer in males makes up less than 1% of all new cancer diagnoses51. Mice were housed 4 per cage in a climate-controlled facility on a 12 h light/dark cycle (lights on 6:00–18:00). Water and standard lab chow was provided ad libitum.

Drug paradigm

Chemotherapeutic agents were purchased from the UAMS Inpatient Pharmacy, diluted into stock solutions and stored according to the manufacturer’s instructions. Chemotherapeutic agent dosages used were previously shown by our lab to induce hippocampal morphological and behavioral changes while being translatable to human dosages52. DOX (A) was administered via intraperitoneal (I.P.) injection at 2 mg/kg dosage in combination with CYP (C) at 50 mg/kg constituting the AC chemotherapy regimen. AC was given once a week for four weeks. Upon completion of the AC injections, PTX (5 mg/kg) was administered once weekly via IP injections for four weeks, completing the AC-T regimen. Control mice received 0.9% saline for all injections52. Behavior testing began 30 days following the last injection, and euthanasia by cervical dislocation occurred 30 min after the last behavior test. Experimental timeline is represented in Fig. 1.

Fig. 1
figure 1

Experimental timeline. figure generated with BioRender.com.

Full size image

Behavior testing- sociability

The three-chamber sociability test is depicted below (Fig. 2)53,54. Briefly, a polycarbonate box was divided into three chambers with removable partitions that had openings to allow free movement throughout the chambers. During the habituation stage, the test mouse was placed in the center chamber and was allowed to freely explore for 10 min (Stage 1). After 10 min, a mouse of similar age, sex, and strain (Stranger 1), was placed inside a small wire cage in one of the side chambers (Stage 2). The test mouse was allowed to familiarize itself with Stranger 1 for 10 min before a second stranger (Stranger 2) was placed in the previously empty wire cage in the other side chamber (Stage 3). The test mouse was again allowed to explore Stranger 2 for 10 min (Social Novelty phase). Data analyzed compared the percentage of time spent with Stranger 1 versus Stranger 2. Location of Stranger 1 versus Stranger 2 was appropriately counterbalanced55. The apparatus was cleaned with 20% ethanol and allowed to air dry before the next trial began. Behavior was recorded and tracked using the Ethovision tracking system for all behavior sessions (Ethovision XT v17.5, Noldus Information Technology, Wageningen, Netherlands).

Fig. 2
figure 2

Three chamber sociability test. created with BioRender.com.

Full size image

Open-Field (OF) test

To assess the unconditioned anxiety of mice the OF behavioral test was employed (Fig. 3)56. The mice were placed in the middle of a square Plexiglass arena (40 cm x 40 cm x 35 cm) with a gray floor. The walls of the arena were covered with white posterboard to prevent the mouse from seeing the room’s surroundings. The mouse was allowed to roam freely for 10 min. At the end of the trial, the number of feces were counted. The arena was cleaned with 20% ethanol between animals to minimize olfactory signals between trials. Locomotor activity was measured using total distance traveled (cm) and velocity (cm/s). The center region of the arena (20 cm x 20 cm) is considered anxiogenic. Time spent in the center zone was recorded.

Fig. 3
figure 3

Open field apparatus: Mice with increased anxiety tend to spend more time along the outer edge of the apparatus compared to those with less anxiety who freely explore the apparatus. created with BioRender.com.

Full size image

RNA sequencing (hippocampal tissue)

After behavior testing, whole brains were harvested and flash frozen in 2-methylbutane. Hippocampal samples were collected from coronal slices using a 0.2 mm tissue punch and were sent to the RNA Sequencing Core (UAMS, Little Rock, AR) for processing (n = 5–6). RNA was extracted from samples using the Zymo Quick-DNA/RNA miniprep Plus Kit (Cat # D7003). mRNA Library was constructed using the Illumina TruSeq Stranded mRNA Library Prep kit (Cat # 20020594). Samples were run on a 200-cycle NextSeq 2000 P2 flow cell. Dragen (v 4.3.6) was used for binary base call conversion and FastQC (v 0.11.9) was used for quality control. Quantification was performed using Salmon (v 1.9.0). Fastq files were uploaded into Qiagen Ingenuity Pathway Analysis (IPA) via the RNA-Seq Analysis Portal57.). All analysis was performed using an FDR p-value < 0.05, and a fold change > 1.5 or < −1.5.

Western blot (hippocampal tissue)

A Potter-Elvehjem mechanical compact stirrer (BDC2002, Caframo LabSolutions), was used to homogenize frozen hippocampal samples in a 1% Triton-X100 radioimmunoprecipitation buffer (MilliporeSigma) containing protease inhibitors. Bicinchoninic acid (BCA) protein assay (Bio-Rad) was used to determine the protein concentration. 30 µg of protein was added to a 2x Laemmli buffer containing beta-mercaptoethanol (5%) and gel electrophoresis was performed before transferring proteins to a PVDF (0.22 μm) membrane. Membranes were incubated in either rabbit anti-c-Fos (Abcam, ab208942), mouse anti-zif268 (Egr1) (Santa Cruz Biotechnology, sc-515830), mouse anti-Arc (Santa Cruz Biotechnology, sc-17839), mouse anti- GFAP (Abcam, ab68428), mouse anti-Iba-1 (Abcam, ab178846), mouse anti-doublecortin (Abcam, ab18723), or mouse anti-glyceraldehyde 3-phosphate dehydrogenase (GADPH) (Santa Cruz Biotechnology, sc-32233) overnight in TBS with 0.1% Tween-20 and 5% non-fat dry milk at 4 C. Membranes were again incubated with horseradish peroxidase-conjugated goat anti-mouse or (Cell Signaling, 7074 S) anti-rabbit IgG (Cell Signaling, 7076 S) and then covered in enhanced chemiluminescence Plus Western Blotting Detection Reagent (MilliporeSigma). AlphaImager gel documentation system (ProteinSimple) was used to develop and image films. Densitometry was performed using ImageJ software, and antibodies were normalized to GAPDH and calculated relative to saline58. Full length western blots can be found in the supplementary information document.

Proteomics (intestinal tissue)

Protein isolation

Intestines were homogenized in RIPA buffer on ice. Homogenates were centrifuged and supernatants collected. Protein concentrations were determined with the Bradford assay kit (Thermo Fisher) following the manufacturer’s instructions.

CME bHPLC TMT Methods – Orbitrap eclipse

Total protein from each sample was reduced, alkylated, and purified by chloroform/methanol extraction prior to digestion with sequencing grade trypsin (Promega). The resulting peptides were labeled using a tandem mass tag 10-plex isobaric label reagent set (Thermo) and combined into one multiplex sample group. Labeled peptides were separated into 46 fractions on a 100 × 1.0 mm Acquity BEH C18 column (Waters) using an UltiMate 3000 UHPLC system (Thermo) with a 50 min gradient from 99:1 to 60:40 buffer A: B ratio under basic pH conditions, then consolidated into 18 super-fractions. Buffer A = 10 mM ammonium hydroxide, 0.5% acetonitrile, Buffer B = 10 mM ammonium hydroxide, 99.9% acetonitrile, both buffers adjusted to pH 10 for offline separation.

Each super-fraction was then further separated by reverse phase XSelect CSH C18 2.5 μm resin (Waters) on an in-line 150 × 0.075 mm column using an UltiMate 3000 RSLCnano system (Thermo). Peptides were eluted using a 75 min gradient from 98:2 to 60:40 buffer A: B ratio. Eluted peptides were ionized by electrospray (2.2 kV) followed by mass spectrometric analysis on an Orbitrap Eclipse Tribrid mass spectrometer (Thermo) using multi-notch MS3 parameters. MS data were acquired using the FTMS analyzer in top-speed profile mode at a resolution of 120,000 over a range of 375 to 1500 m/z. Following CID activation with normalized collision energy of 31.0, MS/MS data were acquired using the ion trap analyzer in centroid mode and normal mass range. Using synchronous precursor selection, up to 10 MS/MS precursors were selected for HCD activation with normalized collision energy of 55.0, followed by acquisition of MS3 reporter ion data using the FTMS analyzer in profile mode at a resolution of 50,000 over a range of 100–500 m/z. Buffer A = 0.1% formic acid, 0.5% acetonitrile, Buffer B = 0.1% formic acid, 99.9% acetonitrile.

Data analysis

Proteins were identified and MS3 reporter ions quantified using MaxQuant (Max Planck Institute) against the UniprotKB Mus musculus (March 2021) database with a parent ion tolerance of 3 ppm, a fragment ion tolerance of 0.5 Da, and a reporter ion tolerance of 0.003 Da. Scaffold Q + S (Proteome Software) was used to verify MS/MS based peptide and protein identifications Protein identifications were accepted if they could be established with less than 1.0% false discovery and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet al.gorithm59 and to perform reporter ion-based statistical analysis. Protein TMT MS3 reporter ion intensity values are assessed for quality and normalized using ProteiNorm60. Normalized counts were uploaded into QIAGEN IPA for network analysis evaluated with Fold change > 1.5, < −1.5 and an FDR p-value < 0.05.

Villus height and crypt depth

We assessed the villus height and the crypt depth. Segments of proximal jejunum were obtained, fixed, and embedded so that four transverse sections were obtained per specimen, cut at 5 μm, and stained with hematoxylin and eosin (H&E). H&E-stained slides were used for villi length and crypt depth determination. Each stained section was examined for histopathological abnormalities on a microscope supported with a digital camera. Images were captured at 4x magnification for villus measurements and 10x magnification for crypt measurements. An average of 12 villi were analyzed for villous height and 12 crypts were analyzed for crypt depth measurements per animal at 20x magnification. The villus height was measured from the tip to the villus-crypt junction and the crypt depth from the base of the villus to the mucosa with ImageJ software v.1.53 (n = 5 per treatment group)61.

16 S rRNA gene amplicon sequencing

The fecal samples were sent to RTL Genomics sequencing facility for sequencing. The V3-V4 hypervariable regions of the bacterial 16S small subunit rRNA gene were amplified using primers 357 F (5’- CCTACGGGNGGCWGCAG − 3’) and 785R (5’- GACTACHVGGGTATCTAATCC-3’) and the amplicons were subsequently sequenced on an Illumina MiSeq sequencing platform (Illumina) by generating paired-end reads from amplicon libraries with 300-bp inserts.

Raw sequencing reads obtained from the sequencing facility were demultiplexed and primer trimmed. Processing of raw sequencing reads, and subsequent microbiome analysis was performed using QIIME2 version 2024.1062. Any references prefixed with q2- are QIIME 2 plugins. Demultiplexed FASTQ files were imported in QIIME2 as QIIME Zipped Artifacts (qza) using q2-import and visualized using q2-demux summarize via QIIME2 View (QIIME 2 View) to check the quality of the forward and the reverse reads. Amplicon Sequence Variants (ASVs)/Exact Sequence Variants (ESVs)63 were generated from forward reads using DADA264 via q2-DADA2 plugin truncating forward reads to 270 bases and the reverse reads to 185 bases. For denoising using DADA2, ‘pooled’ method was used for chimera detection and ‘pseudo’ method was used to pool chimeras, while also setting the ‘--p-min-fold-parent-over-abundance’ to 16.

Taxonomy was assigned to ASVs using a Naïve Bayes classifier trained65,66 on the V3-V4 amplicon region of the SSU SILVA NR99 reference database67 using RESCRIPt68. ASVs that were classified as ‘Unclassified’, ‘Mitochondria’, ‘Eukaryota’, and those not having at least phylum level classification were removed. Quality of the sequences were evaluated using q2-quality-control plugin by comparing the feature sequences to the RESCRIPt curated SILVA V3-V4 region of 16 S rRNA gene69. Any ASVs that contained less than 90% identity and had less than 90% query coverage, were removed70,71,72. ASVs were placed into and existing SILVA v128 reference tree with q2-fragment-insertion73,74,75.

Alpha diversity (within sample diversity)

Various alpha diversity metrics (within sample diversity) were estimated using various rarefaction approaches and samples were rarefied approximately 8800-10,000 sequences per samples to determine Observed features, Shannon index, Faith phylogenetic diversity. Statistical significance test of various alpha diversity measures was estimated using PERMANOVA test.

Beta diversity (between sample diversity- i.e., differences/distances between samples)

Between sample diversity was estimated using UniFrac (weighted and unweighted)76 and Bray-Curtis dissimilarity using q2-diversity after rarefying samples to 8898 sequences per samples77,78. We used permutation-based statistical testing (PERMANOVA) via the q2 diversity beta-group-significance within QIIME 2 to test for statistical significance in Bray-Curtis, weighted and unweighted UniFrac distance between the groups.

ANCOM-BC79 was used, via q2-composition, to estimate the differential abundance of non-rarefied features, which were subsequently collapsed by taxonomy at the genus level. Only significant (q ≤ 0.05) features with a log fold change ≥ 1 were retained.

Results

Behavior- sociability: AC-T decreases social memory

Fig. 4
figure 4

Social memory test (a) Neither Saline or AC-T treated mice displayed significant bias of chamber exploration. (b) both Saline and AC-T displayed normal sociability. (c) Saline treated mice were able to discriminate between Stranger 1 and Stranger 2 when both mice were present, whereas mice treated AC-T failed to distinguish between the novel and familiar stimulus. Denotes significant difference between saline and AC-T treated † p < 0.05; ‡ p < 0.001. Bars represent average ± SEM, N = 12.

Full size image

Social behavior and memory were assessed using the three-chamber sociability test (Fig. 4). Mice in both saline and AC-T displayed normal habituation by spending approximately equal time exploring both lateral chambers (F(3, 44) = 2.169, P = 0.1051). During familiarization, both treatment groups spent significantly more time with the stranger mouse versus the empty chamber (F(3, 44) = 9.163, P < 0.0001). We performed Bonferroni’s post-hoc analyses, which revealed mice treated with saline spent significantly more time with the novel mouse (Stranger 2) during the Stage 3 (Saline Novel vs. Familiar P < 0.05), whereas mice treated with AC-T showed no significant difference in the time spent with either the familiar or the stranger mouse (AC-T Novel vs. Familiar P = 0.1949).

Fig. 5
figure 5

There was a significant decrease in central tendency in AC-T treated mice. bars represent average ± SEM, N = 12 ⁎p < 0.05.

Full size image

Behavior- open field: AC-T increases anxiety-like behavior

Anxiogenic behavior in the OF test is measured by a parameter known as central tendency, or the amount of time each mouse spends in the center of the arena. Central tendency was used to measure anxiogenic behavior in the OF test. Decreased central tendency indicates increased anxiety. There was a significant decrease in central tendency between the saline and AC-T groups (Fig. 5) (p = 0.0349).

RNA sequencing of hippocampal tissue: AC-T alters gene expression in the hippocampus

FastQ files were uploaded and analyzed using the QIAGEN RNASeq Analysis Portal. Using the parameters of Fold change > 1.5, < −1.5 and an FDR p-value < 0.05, 460 differently expressed genes were identified. Of these, 290 were downregulated and 170 were upregulated in AC-T treated mice compared to saline controls. In Figs. 6, 7, 8 and 9, we present multiple complementary approaches to visualize the RNA sequencing data. Figure 6a shows a principal component analysis (PCA) plot comparing AC-T–treated and saline-treated samples. PCA reduces the dimensionality of large datasets while preserving the majority of the variance80. The clear separation of AC-T and saline samples along the first two principal components indicates distinct transcriptional profiles between the treatment groups. Volcano plots (Fig. 6b) provide a rapid visualization of differential gene expression by simultaneously displaying fold change and statistical significance81. Figure 7 presents a heatmap of the most highly upregulated and downregulated genes, highlighting treatment-associated expression patterns across samples. In addition, we identified differentially expressed genes associated with neuronal development and morphogenesis, which are visualized in Figs. 8 and 9 using volcano plots and heatmaps of these functional gene subsets. Collectively, the PCA plot, volcano plots, and heatmaps (Figs. 6, 7, 8 and 9) demonstrate that AC-T treatment induces broad and biologically relevant changes in gene expression.

Fig. 6
figure 6

PCA and Volcano plot of results. PCA plot of PC1:PC2. Volcano plot has an x cut off (0,15). Gray dots indicate no significance; red dots indicate upregulated genes (Fold Change > 1.5); blue dots indicate down regulated genes (Fold Change < −1.5).

Full size image
Fig. 7
figure 7

Heatmap generated from Qiagen RNA Seq Analysis Portal. A darker blue color indicates gene expression that was decreased in AC-T treated animals, red indicates increased expression of genes.

Full size image
Fig. 8
figure 8

Volcano plot highlighting the genes associated with the development of neurons as indicated by Qiagen. Dark blue and red points on the volcano plot indicate where downregulated genes associated with development of neurons are in relation to all differentially expressed genes. In the heatmap, low expression genes are indicated in blue and high expression is indicated in red relative to saline controls.

Full size image
Fig. 9
figure 9

Volcano plot highlighting the genes indicated to be associated with the morphogenesis of neurons. Dark blue and red points on the volcano plot indicate where associated are in relation to all differentially expressed genes (Log Fold Change > or < 1.5). In the heatmap, decreased expression are indicated in blue and increased expression is indicated in red relative to saline controls.

Full size image

RNA Sequencing- IPA networks: AC-T disrupts nervous system development

For tertiary analysis, the dataset was uploaded to Qiagen IPA where 25 functional networks were identified. The top 10 can be found in Table 1. Networks 3 (Fig. 10), 4 (Fig. 11), and 8 (Figs. 12 and 13 were associated with cellular growth and development and nervous system development and function.

Table 1 Top 10 networks disrupted by AC-T treatment compared to saline controls. IPA p-score is calculated as the equal to the -log10(p-value) where the p-value is calculated from the right-tailed fisher’s exact Test. A higher IPA score indicates that a higher number of molecules from our dataset belong to that network.
Full size table
Fig. 10
figure 10

Representation of hippocampal RNA network 3. Functions associated with this network include Cellular Development, Cellular Growth and Proliferation and Nervous System Development and Function. Overlay of the disease and function tool highlighted specific functions such as complexity of apical processes, abnormal morphology of neurons and formation of axon branches. Green molecules indicate downregulated expression and red molecule nodes are upregulated in AC-T samples compared to saline controls.

Full size image
Fig. 11
figure 11

Network 4 representation of RNA from hippocampal tissues. Functions include Cellular Development, Cellular Growth and Proliferation and Nervous System Development and Function. The disease and function overlay tool highlighted alterations in genes associated with the development of neurons, abnormal brain morphology, and formation of the hippocampus, among others. Green molecules indicate downregulated expression and red molecule nodes are upregulated in AC-T samples compared to saline controls.

Full size image
Fig. 12
figure 12

Network 8: Network functions associated with Network 8 included Cellular Development, Cellular Growth and Proliferation, and Nervous System Development and Function. Disease and function overlay tool showed alterations in density of dendrites, morphology of neurons and development of neurons. Green molecules are downregulated and red is increased in AC-T tissues compared to saline controls.

Full size image
Fig. 13
figure 13

Legend for Qiagen IPA networks.

Full size image

GO: cellular components: AC-T alters neuronal development and function

3,134 genes were identified with a p-value < 0.05. This list was uploaded to the STRING database82 for functional enrichment analysis where the “ENSEMBLE ID” was used as the identifier and “LFC” was used as the value. 36 Gene Ontology (GO) Cellular Components were identified. Top 10 are depicted in the functional enrichment visualization (Fig. 14), and Table 2 contains those associated with neurogenesis and neuronal function. GO Cellular Components describes genes that affect both the cellular structures (such as the cytoskeleton) and the macromolecular complexes of these components83.

Fig. 14
figure 14

Enrichment Top 10 Cellular Components. Among the top affected components in the hippocampus after AC-T treatment are postsynaptic density, neuron to neuron synapse, and asymmetric synapses.

Full size image
Table 2 GO: cellular components associated with neuronal function and neurogenesis. “Genes Mapped” indicates how many genes from our dataset are associated with this term. The enrichment score compares to the mean value of genes associated with that term to the largest deviation from the overall input mean. Direction indicates where in the dataset the term was enriched. Top indicates upregulated genes, bottom indicates downregulated genes, and both ends indicates that the term was found to be both up and down regulated. False discovery rate (FDR) displays the adjusted p-value to account for multiple testing and is a metric for Understanding the significance of the enrichment.
Full size table

Western blots: AC-T alters neurogenesis, astrocytic activity, and neuroinflammation

In the current study, doublecortin (DCX) was used as a marker for immature neurons. DCX is a microtubule stabilizer and a marker for migratory immature neurons and neurogenesis84,85,86. There was a significant decrease in the expression of DCX (t = 4.975, p = 0.0004; Fig. 15a) after AC-T treatment. Glial fibrillary acidic protein (GFAP) was used to mark reactive astrocytes, which are dynamic glial cells that respond to changes in the neuronal environment to maintain normal neuronal function and support neurogenesis87,88. AC-T chemotherapy also significantly decreased the expression of GFAP (t = 3.836, p = 0.0020; Fig. 15b). Iba1 is a marker for microglia activation which are important for maintaining homeostasis in the brain89. There was a significant increase in Iba1 (t = 4.353, p = 0.0010; Fig. 15c) in the AC-T group when compared to the saline group in a t-test.

Fig. 15
figure 15

Doublecortin, GFAP and Iba1 in the hippocampus were measured using western blot analysis A). Protein expression was normalized to GAPDH B-D). Mice treated with AC-T showed a significant reduction in Doublecortin and GFAP. However, there was a significant increase in Iba1 expression in the hippocampus. Bars represent Average ± SEM, N = 5–6. Denotes significant difference between saline and AC-T treated. **p < 0.01, *** p = 0.001. Full length western blots can be found in supplementary file.

Full size image

Western blots: AC-T decreases expression of immediate early genes

There was a significant decrease in the expression of Arc (t = 2.243, p = 0.0470; Fig. 16a), c-Fos (t = 3.412, p = 0.007; Fig. 16b), and Egr1 (t = 7.717, p = 0.0010; Fig. 16c) in the AC-T group when compared to the saline group using a t-test.

Fig. 16
figure 16

Immediate-early gene expression in the hippocampus. IEG expression was measured using western blot analysis. Protein expression was normalized to GAPDH a -c). Mice in the AC-T group showed a significant reduction in IEG expression in the hippocampus. Bars represent average ± SEM, N = 5–6. Denotes significant difference between saline and AC-T treated. * p < 0.05, ** p < 0.01, *** p < 0.001. Full length western blot images can be found in supplementary file.

Full size image

Differential analysis of protein expression: AC-T alters IL-15 signaling in intestinal tissue

A total of 61 proteins were differentially expressed in AC-T intestinal tissue vs. saline. Table 3 shows the top dysregulated proteins expressed.

Table 3 Top dysregulated proteins in intestinal tissue associated with AC-T treatment in comparison to saline control.
Full size table

Ingenuity Pathway Analysis tool was used to analyze proteomics networks affected by the treatment regimen in intestinal tissue. Qiagen IPA algorithm overlays focus molecules from the experimental dataset to the Global Molecular Network and generates a connectivity map. The p-score (-log10(p-value)) is calculated by Fisher’s exact test and is indicative of the probability of focus molecules in a network being selected randomly from the Global Molecular Network. Table 4 Fig. 17 contains the top 4 networks.

Table 4 Top IPA protein networks associated with AC-T treatment in the intestines. The IPA p-score is calculated as the -log 10 (p-value).
Full size table
Fig. 17
figure 17

Depiction of proteomics network 1, identified by Qiagen IPA. Molecules involved in IL-15 signaling are present. Node color is indicative of differential expression where red represents upregulation, and green represents downregulation. Color intensity directly corresponds to the degree of regulation. Gray nodes represent proteins found in the data set but were not significantly expressed. Uncolored nodes represent proteins not differentially expressed but were incorporated into the computational network based on evidence stored in the Ingenuity Knowledge Base. The arrows and blocked lines indicate known direct and indirect interaction as well as the direction of the interaction.

Full size image

Microbiome analysis: AC-T chemotherapy alters gut Microbiome

Differences in overall microbial taxonomic profiles between the saline and AC-T–treated groups were pronounced, revealing a marked increase in the relative abundance of Akkermansia in AC-T–treated animals compared to saline controls (Figure 18). These shifts in microbial community structure were further supported by analyses of alpha diversity (Figure 19), beta diversity (Figure 20), and differential abundance (Figure 21), collectively indicating that AC-T chemotherapy significantly altered gut microbial composition and diversity.

Fig. 18
figure 18

Microbiome composition of different groups at phylum level and genus level. The most abundant phyla were Firmicutes in saline controls, whereas Akkermansia were the most dominant bacterial genus in AC-T treated animals.

Full size image

Alpha diversity

Fecal microbial diversity and richness were evaluated using alpha diversity (within sample diversity). Any features that contained less than 50 reads and did not occur in at least 2 samples were discarded. The final table was rarefied to 8,000 reads per sample for downstream alpha & beta diversity analyses. The significance of each diversity metric was tested with the Pairwise Kruskal-Wallis test between the AC-T group and saline group. Kruskal-Wallis identified significant difference. Chemotherapy significantly decreased Peilou Evenness (H = 7.5. p = 0.006; Figure. 19a), Faith’s Phylodiversity (H = 5.63, p = 0.01; Figure. 19b) Jaccard (H = 17.05, p = 0.003: Figure. 19c). However, there was a significant increase in Observed Features (H = 6.10, p = 0.04: Figure. 19d).

Fig. 19
figure 19

Box and whisker plot showing various Alpha diversity metrics between AC-T treated animals and saline controls. Statistical significance differences were observed in (A) Pileou’s evenness, (B) Faith’s phylodiversity, (C) Shannon entropy, and (D) observed features. Kruskal-Wallis pairwise test was used for statistical analysis (* p < 0.05).

Full size image

Beta Diversity

Between sample diversity (beta diversity) was estimated by Bray-Curtis dissimilarity matrix (abundance based and non-phylogenetic method), Weighted UniFrac distance (abundance based and phylogenetic method), Unweighted UniFrac distance (abundance based and non-phylogenetic), and Jaccard distance (abundance based and non-phylogenetic method). The significance of each diversity metric was tested with a PERMANOVA, pseudo-F, between the AC-T group and saline group. The statistical package identified significant differences in Bray-Curtis dissimilarity matrix (F = 29.64, p = 0.003: Figure. 20a) and Jaccard distance (F = 17.05, p = 0.001: Figure. 20b), Unweighted UniFrac distance (F = 23.65, p = 0.004: Figure. 20c), and Weighted-UniFrac distance (F = 42.90, p = 0.003; Figure. 20d).

Fig. 20
figure 20

Comparison of Beta diversity between chemo and the control groups. a) Bray-Curtis, b) Jaccard, c) Unweighted UniFrac, and d) Weighted UniFrac shown. Plots are relative to saline, that is, blue is saline vs. saline distances and green is saline vs. AC-T distances (* p < 0.05).

Full size image

Differential abundance

Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC), which estimates the unknown sampling fractions and corrects the bias induced by their differences among samples. ANCOM-BC is based on Aitchison’s methodology, uses relative abundances to infer about absolute abundances. Results showed that at the genus level, the abundance of Lachnospiraceae; Lachnospiraceae_UCG-004, Lachnospiraceae; Lachnospiraceae_NC2004_group, Lactobacillaceae; g_Limosilactobacillus, Bifidobacteriaceae; g_Incertae_Sedis, Lachnospiraceae_Anaerostipes, Lactobacillaceae; g_Ligilactobacillus, Lactobacillaceae were enriched in the saline treatment group. However, Lachnospiraceae; g_Lachnoclostridium, Ruminococcaceae; g_Ruminococcus, Rikenellaceae; g_Alistipes and Akkermansiaceae; g_Akkermansia were enriched in AC-T treated mice compared to saline (Fig. 21).

Fig. 21
figure 21

Heatmap of differentially abundant genera between Saline vs. AC-T, as determined via ANCOM-BC. A lighter color indicates low abundance, whereas a darker color indicates highly abundance or enriched taxa.

Full size image

Crypt depth and villus height: AC-T significantly decreases crypt depth

The intestinal derangement was observed 30 days after AC-T treatment. When the mean depth of crypts in mice treated with (145.00 ± 6.23) AC-T was compared to the (125.6 ± 2.44) saline group, a significant decrease in average depth was observed (t = 3.042, p < 0.01; Fig. 22). Villus length in mice treated with (326.7 ± 14.94 μm) AC-T was not significantly affected compared to (305.3 ± 7.66 μm) saline (t = 1.287, p = 0.20; Fig. 22).

Fig. 22
figure 22

AC-T treatment significantly decreased crypt depth, but villus height was not greatly impacted. (** p < 0.01).

Full size image

Discussion

Behavioral effects: AC-T decreases social memory and increases anxiety-like behavior

Cognitive impairments associated with chemotherapy administration can negatively affect the following cognitive domains: attention and concentration; executive function; visuospatial skills; and learning and memory90.A previous study showed that A previous study has shown that AC-T treatment impacts hippocampal dependent spatial memory in the Morris Water Maze52. Another study stated that rats treated with DOX or CYP, either alone or in combination with one another, showed deficits in spatial cognition as indicated by the novel recognition test91. In the current study, first, we investigated how a combination chemotherapy regimen (AC-T) affected hippocampal-dependent social memory using the three chamber sociability test92.

During the habituation stage, mice showed no preference to either side of the chamber. In the second stage, both saline and AC-T treated mice spent more time with the stranger mouse versus the empty chamber, indicating no deficit in their ability to socialize. This stage of the test is often used in mouse models of autism where autism model mice spend significantly less time interacting with stranger mouse compared to wildtype controls93. During the final stage of the test, AC-T treated mice spent equal time with both the familiar and novel mice, whereas saline mice spent significantly more time with the novel mouse. This indicates a deficit in social memory after AC-T treatment.

These findings are consistent with others who have explored the effects of various breast cancer chemotherapy regimens on cognition. Mice treated with doxorubicin, cyclophosphamide, and docetaxel (TAC) showed a deficit in social memory, as did mice treated with a combination of cyclophosphamide, methotrexate, and 5-fluoruracil (CMF)54,94. Social memory is the ability to recognize and interact with others in a social group and is important for maintaining relationships95. Social relationships have been shown to attribute to the Health-Related QoLin breast cancer survivors96. Further, studies showed that breast cancer patients with strong social relationships showed fewer depressive symptoms97,98.

Next, we used the open field (OF) test as an observational method to assess unconditioned responses such as general locomotor activity, anxiety-like behavior, and exploratory tendencies. The OF test provides a straightforward and rapid assessment that does not require prior training or conditioning of the test subject, contributing to its widespread use in behavioral neuroscience research for evaluating spontaneous exploration and emotional reactivity99. We assume evolutionary forces have selected for a common anxiety-mediated response in animals in response to certain stimuli100. Rodents are normally active at night and prefer darkness thus, they show aversion to large, brightly lit, open and unknown environments101. Two measures of emotionality are deduced from the OF test: locomotor activity in each predefined area as well as autonomic activity such defecation102. Mice displaying more anxiety will result in less exploration time in the center of the arena. Increased defecation could also be an indicator of anxiety. When comparing mouse models, studies have conclusively linked results from the OF test with other measures of anxiety supporting its continued use. In the present study, OF results showed that mice treated with AC-T had a decrease in central tendency, indicating an increase in anxiety behaviors. Rats treated with DOX, CYP, or the AC combination also displayed increased anxiety as measured by the light-dark box apparatus91. These findings, along with those of the three chamber sociability test, both demonstrate changes that can occur after chemotherapy that impact the QoL of survivors103.

Molecular mechanisms: AC-T treatment disrupts neuronal function and neurogenesis in hippocampal tissue

RNA sequencing (RNA-Seq) technology provides comprehensive insights into gene expression patterns, molecular signaling pathways, and genomic variations that can reveal potential biomarkers for clinical diagnosis and therapeutic intervention. By enabling high-resolution quantification of transcripts across thousands of genes simultaneously, RNA-Seq allows for the identification of disease-associated molecular signatures that can be translated into targeted therapeutic strategies or prognostic tools104. The resulting datasets were uploaded into QIAGEN Ingenuity Pathway Analysis (IPA) for in depth functional annotation and biological network analysis, facilitating the identification of differentially regulated pathways involved in neuroinflammation, synaptic signaling, oxidative stress, and immune regulation. Qiagen IPA predicts networks that may be altered based on known direct or indirect interactions of molecules within the QIAGEN database. In network 3 (Fig. 10), genes involved with differentiation, specification, and morphogenesis of neurons were highlighted. Decreased expression, indicated by the green color, included genes such as ISL1, POU4F1, and the IRX family of genes. Iroquois (Iro/Irx) homeobox genes are important for cell specification and differentiation of various tissues including the heart and brain105. IRX3 and IRX5 have been linked to energy homeostasis and neurogenesis in the hypothalamus106. ISL1, a homeobox transcription factor, is necessary for the development of sensory and sympathetic neurons107. POU4F1 is a member of the Pit-Oct-Unc transcription factor family and is important for neurogenesis and neuronal survival108. Changes in the expression of these genes after AC-T treatment present possible therapeutic targets for addressing the cognitive deficits presented here.

Network 4 (Fig. 11) focused on development of neurons, brain morphology, and development of the hippocampus. In the present study, FOXB1, SLC5A7, and ZFHX3 were all decreased in AC-T treated mice. FOXB1 is a transcription factor that plays an important part in the neurogenesis/gliogenesis switch; FOXB1 promotes the differentiation of mature oligodendrocytes, which are essential for maintaining normal neuronal and synaptic function109,110. FOXB1 knockout mice exhibited deficits in spatial navigation and working memory111. SLC5A7 is a marker for cholinergic neurons which innervate the dentate gyrus and are believed to in part regulate adult neurogenesis in the hippocampus112,113. ZFHX3, also known as ATGF1, is a zinc finger homeobox transcription factor that regulates genes for neuronal survival and cellular adhesion. Expression promotes neuronal differentiation114. These findings suggest that disruptions in neurogenesis and homeostatic balance contribute to the cognitive deficits observed following chemotherapy treatment115.

Network 8 (Fig. 12) comprised genes involved in neuronal morphology and development, including those regulating interneuron differentiation and migration as well as the development of cholinergic and GABAergic neuronal populations. Among them, the expression of LHX5 was significantly decreased after AC-T treatment. LHX5 is a member of the LIM homeobox genes which are key players in differentiation and proliferation of cells during development. LHX5 is specific for the developing nervous system; deletion of LHX5 resulted in a malformed hippocampus and impaired spatial navigation in the Barnes maze116. This suggest that alterations in neurogenesis and neuronal morphology may contribute to the behavioral changes observed, offering potential biomarkers for treating CICI-related impairments.

Gene ontology (GO) functional enrichment analysis (Fig. 13) highlighted alterations in neuronal structures important for typical neuronal function. Postsynaptic density is necessary for normal synaptic transmission117. Neuron to neuron synapse was also altered; synaptic dysfunction due to an increase in reactive oxygen species is believed to contribute to the development of AD with increases in inflammation and abnormal axon function among affected systems118. In our western blot analysis, we quantified the protein expression of GFAP, a marker for astrocytes, as astrocytes are known to play an important part in maintaining normal synaptic transmission119.

Adult neurogenesis is crucial for learning and memory formation by forming new synaptic connections with existing neurons120. DCX is a marker for immature neurons and neurogenesis121. Several chemotherapy drugs including CYP and DOX have been shown to alter neurogenesis in the hippocampus, and taxane agents like paclitaxel have been shown to alter axonal transport and neuronal function122. Ablation of doublecortin positive cells resulted in deficits in hippocampal dependent spatial memory and working memory tasks123. In the current study we found a significant decrease in the expression of DCX after AC-T treatment (Fig. 15a). Our data aligns with a previous study where mice that received either a single dose of CYP or DOX had significantly less DCX positive cells compared to saline controls and had impaired spatial recognition memory124. In another, mice treated with paclitaxel also had a significant decrease in the number of DCX positive cells and exhibited impairment in the Morris water maze125. The decrease in DCX in the present study indicates an alteration in neurogenesis after AC-T treatment.

Astrocytes are one of the most abundant glial cell in the CNS and play an important role in blood brain barrier integrity, homeostasis, response to injury, and neuronal function126,127. Astrocytic end feet support the endothelial cells that comprise the blood brain barrier and soluble factors released by astrocytes maintain the BBB128,129. Astrocytes can store glucose, ensuring adequate energy for normal brain function, memory, and learning; and they respond to both neurotransmitters and glial transmitters127. Astrocytes are important for the regulation of neuroinflammation and blood brain barrier maintenance; disruptions to the blood brain barrier and an increase in neuroinflammation are proposed causes of chemobrain88,130,131. Astrocyte dysfunction has been implicated in a range of neurological and psychiatric disorders, including major depressive disorder, anxiety, and other conditions characterized by disrupted neuroglial signaling and impaired synaptic homeostasis132,133,134. Mice treated with DOX showed a significant increase in the expression of GFAP, as did mice who were treated with paclitaxel had a significant increase in GFAP 6 h after the last PTX29. In the current study, western blot analysis revealed there was a significant decrease in GFAP, indicating a decrease in astrocyte activity after chemotherapy treatment (Fig. 15b). These results contrast with previous reports in which chemotherapy has been shown to either increase GFAP expression suggesting astrocyte activation and gliosis in response to neuroinflammation. While both types of findings differ from our observations, it is important to note that we assessed GFAP expression 30 days after the final chemotherapy administration, a time point at which astrocytic recovery or replenishment may have occurred. This delayed assessment window may therefore account for the apparent reduction in GFAP observed in our study relative to earlier reports that measured acute or subacute responses.

Microglia is another abundant glial cell that is responsible for x, y,z135,136. After injury, activated microglia play an important role in the phagocytosis of dead cells, but sustained microglial activation contributes to the chronic inflammatory state in the brain137. Microglia are important for memory and emotion, but overactivation of microglia results in an increase in the release of pro-inflammatory cytokines, affecting learning and memory29,138. Mice treated with either PTX or DOX showed a significant increase in the expression of Iba1 in the hippocampus29,139. In the current study, there was a significant increase in Iba1 in AC-T mice versus saline controls (Fig. 15c). This suggests that an increase in neuroinflammation, as indicated by Iba1, may contribute to the cognitive dysfunction observed after AC-T treatment. Inflammation, changes in neurogenesis, and reduced levels of neurotransmitters are all proposed causes of chemobrain140. The hippocampus is densely populated with microglia. In the context of our own study, increases in Iba1 and GFAP expression may suggest long-term dysregulation of neuroinflammatory pathways.

IEGs are rapidly transcribed during neuronal activity and behavior tasks141. To identify molecular alterations linked to learning and memory processes after AC-T chemotherapy, we quantified the expression of IEGs: c-Fos, Arc, and Egr1which serve as established indicators of neuronal activation, synaptic plasticity, and experience-dependent cellular activity. Activity-regulated cytoskeleton-associated protein (Arc) is an effector IEG that is activated during synaptic activity and learning. It plays an important role in long-term potentiation (LTP) and long-term depression (LTD) and in synaptic plasticity142,143. Arc knockout mice showed impairment in long term memory consolidation144. Egr1, also known as Zif268, encodes for a zinc finger transcription factor and plays an important role in maintenance of LTP and consolidation of memories in the hippocampus145. C-Fos alters gene transcription in response to cell signals and can be used as a marker for neuronal activation146. The expression of c-Fos increases during behavior training and learning tasks as well as in memory retrieval147. In the current study, there was a significant decrease in expression of IEGs (c-Fos, Arc, and Egr1) after chemotherapy treatment, indicating a decrease in synaptic plasticity and neuroactivity (Fig. 16). These findings are consistent with work by McElroy et al.. where there were significant decreases in dendritic length and complexity in the CA1 and DG regions of the hippocampus; changes in dendritic length and complexity alters synaptic formation, stability, and transmission52. Further, mice treated with CMF also exhibited alterations in hippocampal-dependent behavior and decreases in the expression of IEGs148. Taken together, our current research indicates that AC-T treatment alters normal neuronal function and cognition.

Gut-Brain axis

In the small intestine, villi are tiny, finger-like projections that greatly increase the surface area for nutrient absorption. Each villus is covered by epithelial cells that contain microvilli, forming a brush border rich in digestive enzymes. These enzymes help break down complex organic matter such as carbohydrates, proteins, and lipids into smaller, absorbable molecules like monosaccharides, amino acids, and fatty acids. Destruction of intestinal villi during intestinal mucositis reduces the small intestine’s absorptive surface. Crypts located at the base of the villi, contain stem cells moving from the crypts to the villi generating new cells for the intestinal lining, and secrete antimicrobial peptides such as defenses and lysosomes which provide immune protection to the host organism149,150,151. Injury to normal proliferating cells in the intestine often complicates the treatment of patients with neoplastic disease. Because of their high-proliferative potential, crypt cells within the epithelium are most susceptible to DNA damage and cell death following chemotherapy. Chemotherapy-induced mucositis (CIM) arises because of DNA injury in the rapidly dividing stem cells residing in the crypt-villus axis. It affects up to 40–100% of all cancer patients dosed with chemotherapeutics, depending drug and dosing regimen152. It involves inflammation and ulceration, causing symptoms like diarrhea, abdominal pain, and nausea. The consequences of uncontrolled CIM can be physically, psychologically, and economically devastating26,153.

Intestinal mucosal barrier dysfunction is one of the main adverse effects induced by DOX, CYP and PTX which impairs the functional integrity and structure of the intestinal mucosal154,155,156. A previous study with single dose of DOX show that apoptosis in the crypts peaks around 6–24 h, whereas the maximum effects of the villi height and crypt depth peaks at about 72–96 h in mouse and rat models157. When mice are treated with a single combined dose PTX and CYP, apoptosis induced in the crypts peaks at 24 h158. One of the major goals of the current study was to characterize that late effects AC-T on crypt and villi morphology using clinically relevant treatment regimen. Here, we saw a significant decrease in crypt depth in AC-T mice compared to saline but no difference in villi length.

To gain insight into potential molecular changes occurring in the intestines, TMT proteomics was performed. The top dysregulated proteins in intestinal tissue associated with AC-T treatment in comparison to saline control were cytokines associated with the Interleukin-15 (IL-15) pathway. IL-15 is a proinflammatory cytokine that plays a role in the development, proliferation and activation of different immune cell types such as natural killer, T lymphocytes and B lymphocytes. IL-15 induces is produced in multiple tissue types including the intestinal inner lining and triggers several signaling events via the common receptor α chain that characterizes the structure complex of the IL-2 family of cytokines159. IL-15 is a key player in gut microbiome regulation and has been implicated in several disease conditions such as inflammatory bowel disease, celiac disease, and other inflammatory conditions159. The regulation of IL-15 expression is affected at the transcription, translation, or posttranslational levels, and dysfunction of the negative feedback mediation has been associated with autoimmune and chronic inflammatory issues160. Data from the current study warrants further investigation into examining whether modulating IL-15 would ameliorate the negative effects of AC-T treatment on gut physiology.

Microbiome dysbiosis

To further characterize changes in the gut after AC-T chemotherapy, we evaluated the alpha- and beta-diversity of the microbiome. Significant differences in alpha-diversity between the AC-T and saline groups was observed. Chemotherapy treatment significantly decreased intestinal alpha-diversity of AC-T vs. Saline groups, as measured by Pielou’s evenness, Faith’s phylodiversity, and Shannon Entropy (Fig. 19). Decreases in alpha diversity have been shown in other neurological diseases like Alzheimer disease161; decreases in alpha diversity may indicate a less resilient gut microbiome, leaving the host more susceptible to disruptions, such as chemotherapy162. The number of observed features was significantly increased, but these features are from phylogenetically constrained and taxonomically similar groups, as shown in (Fig. 18). We analysed beta-diversity quantitatively (Bray-Curtis and Weighted UniFrac) and qualitatively (Jaccard and Unweighted UniFrac)163. Both qualitative and quantitative beta-diversity analysis revealed a distinct separation between the AC-T and saline treatment group (Fig. 20), indicating further dysbiosis of the gut microbiome after AC-T treatment. Palanivelu observed significant alterations in beta diversity metrics, indicating reduced microbial diversity and distinct clustering of microbial communities in Alzheimer’s disease (AD) model mice compared to wild-type controls164. Similarly, mice treated with CMF chemotherapy exhibited qualitative shifts in beta diversity, characterized by distinct separation of microbial community composition from untreated controls and a loss of overall microbial richness and evenness148. These findings suggest that both neurodegenerative disease and chemotherapy can disrupt the balance and structure of the gut microbial ecosystem, potentially influencing host neuroimmune and metabolic pathways involved in cognitive function.

The taxonomic composition of the community and relative abundance were remarkably different among groups. Saline treated mice presented a higher abundance of 3 Lachnospiraceae, 4 Lactobacillaceae and 1 Bifidobactericeae species. Lachnospiraceae account for 10%–45% of the total bacteria in feces of healthy adults; in the colon they ferment dietary fiber to promote healthy gut and immune function165. Lactobacillaceae species hydrolyse starch and other sugars to produce butyrate and other SCFAs166. Bifidobactericeae family are gut microbiota that exhibit probiotic or health promoting effects by fermenting carbohydrates, enhancing nutrient absorption, and synthesizing vitamins167. Lactobacillus spp. and Bifidobacterium spp. are commonly used as probiotics to improve gut health, known for their anti-depressive potential and safety profiles168.

Several studies have highlighted the effect of gut microbiome perturbations on social behavior in mice41. Along with the host genetic factors, gut microbiota can interact with the host and contribute to behavioral deficits42. The gut microbiome is a key player in shaping the development and function of the brain and influences social behavior in mice169,170. AC-T treated mice had a greater abundance Ruminococcaceae, Rikenellaceae and Akkermansiaceae. Over-abundance of some members of Ruminococcaceae have been linked to colorectal cancer, inflammatory bowel disease, and neurodegeneration171,172. Ruminococcaceae under normal health conditions play an important role in the maintenance of gut health through their ability to produce butyrate and other short-chain fatty acids (SCFAs)173,174. Microbially derived butyrate and SCFAs are known to lower the Treg (anti-inflammatory)/Th17 (pro-inflammatory) cell ratio175,176. Interestingly, butyrate is not thought to be the preferred energy source in neonate colons but better utilized within adults177. While some members of the Ruminococcaceae can drive inflammation, they can also promote cognitive dysfunction via the production of isoamylamine and mild cognitive impairment in general178,179.

Rikenellaceae are hydrogen-producing bacteria that selectively neutralize cytotoxic reactive oxygen species and protect cells from oxidative stress under healthy conditions180. Akkermansiaceae typically contributes to the integrity of the gut barrier, preventing harmful substances from leaking into the body and potentially reducing inflammation and improving overall health181. However, it should be noted that under certain conditions of dysbiosis, over-abundance of Akkermancia spp. can lead to the thinning of the mucosal layer182, likely exacerbating the significant decrease in observed crypt depth (Fig. 22). Excessive amounts of Akkermansia arise, often due either to low fermentable fiber diets183 or resulting from therapeutic treatments that negatively affect mucin production and gut homeostasis, leading to dysbiosis184 These conditions can result in gut permeability which can drive auto-inflammatory responses, such as neuro-inflammation185.

Rikenellaceae supplementation has been shown to ameliorates the symptoms of inflammatory bowel disease186. Similarly, an over-abundance of Rikenellaceae has been linked to other health issues such as obesity, diabetes, and inflammatory bowel disease187,188,189. Over-abundance of some members of Ruminococcaceae have been linked to colorectal cancer, inflammatory bowel disease, and neurodegeneration171,172. Ruminococcaceae and Akkermansia have been found to be enriched in patients with lupus and AD in mice and colorectal cancer in humans190. We suspect that the direct and indirect negative effects of AC-T treatment result in dysbiotic microbial communities may further exacerbate the negative effects of chemotherapy on the host.

Future directions

Future studies will focus on elucidating the immunoregulatory mechanisms that mediate the relationship between gut dysbiosis and CICI. Increasing evidence suggests that the gut microbiota can influence peripheral and central immune responses through modulation of cytokine networks, including IL-15, a pleiotropic cytokine that regulates T cell, NK cell, and microglial activity191,192,193. Altered IL-15 signaling has been implicated in neuroinflammation, blood brain barrier (BBB) disruption, and synaptic dysfunction processes that are also observed in preclinical models of CICI193,194. To investigate these interactions, we can employ RNASeq based transcriptomic profiling of brain and gut tissues to identify immune and inflammatory gene expression signatures associated with microbiome perturbations and chemotherapy exposure. In parallel, fecal microbiota transplantation (FMT) experiments can be conducted to test whether restoration of microbial balance can normalize IL-15 mediated signaling pathways and mitigate neuroinflammatory responses195,196. Collectively, these studies would define how immune regulation within the gut brain axis contributes to cognitive decline following chemotherapy and may identify novel microbiome- or cytokine-targeted interventions to prevent or reverse CICI.

Conclusion, limitations

Morphologically, the mouse and human digestive tract is conserved in both species. Both human and mouse gut microbiota are both dominated by phylum Bacteriodetes and Firmictues, but deeper taxonomic classifications shows that 85% of bacterial genera in the mouse microbiome are not present in humans197. To address this limitation, we could explore human fecal transplantation into humanized mouse models to better replicate the taxonomic make-up of the human microbiome198. Additionally, humans consume a diverse diet compared to standard rodent chow; specialized rodent diets could be used to better represent human diets198. We assessed physiological changes 30 days after completion of chemotherapy treatment, which is roughly 3.5 human years199. 50% of patients who report experiencing CICI report long-lasting cognitive effects, making our 30 day model translational for these long term side effects200.

In the current study, female mice underwent a clinically relevant dose regimen of AC-T to better understand the effects of these drugs on breast cancer survivors. While changes in other hippocampal dependent cognitive tasks have been previously reported, our study is the first to report alterations in social memory and anxiety-like behavior after AC-T treatment. RNA sequencing provided an overview of changes in the expression of genes in the hippocampus; heatmaps showed changes associated with neurogenesis and neuronal morphology, and IPA network analysis predicted disruptions normal neuronal proliferation and differentiation, morphogenesis, and signaling. Our findings are consistent with others who revealed changes in neurogenesis using single agent chemotherapies; however, in the present study we used a clinically relevant dose regimen124. The gut microbiome showed clear signs of dysbiosis within the AC-T group, potentially exacerbating the negative effects of chemotherapy. The observed effects of our chemotherapy regimen taken together with previous studies involving the interplay between the gut microbiome and behavior heavily point to the involvement of an inflammatory response through IL-15 signaling via the gut-brain axis.