Antibiotic treatment induces microbiome dysbiosis and reduction of neuroinflammation following traumatic brain injury in mice

Abstract Background The gut microbiome is linked to brain pathology in cases of traumatic brain injury (TBI), yet the specific bacteria that are implicated are not well characterized. To address this gap, in this study, we induced traumatic brain injury (TBI) in male C57BL/6J mice using the controlled cortical impact (CCI) injury model. After 35 days, we administered a broad-spectrum antibiotics (ABX) cocktail (ampicillin, gentamicin, metronidazole, vancomycin) through oral gavage for 2 days to diminish existing microbiota. Subsequently, we inflicted a second TBI on the mice and analyzed the neuropathological outcomes five days later. Results Longitudinal analysis of the microbiome showed significant shifts in the diversity and abundance of bacterial genera during both acute and chronic inflammation. These changes were particularly dramatic following treatment with ABX and after the second TBI. ABX treatment did not affect the production of short-chain fatty acids (SCFA) but did alter intestinal morphology, characterized by reduced villus width and a lower count of goblet cells, suggesting potential negative impacts on intestinal integrity. Nevertheless, diminishing the intestinal microbiome reduced cortical damage, apoptotic cell density, and microglial/macrophage activation in the cortical and thalamic regions of the brain. Conclusions Our findings suggest that eliminating colonized gut bacteria via broad-spectrum ABX reduces neuroinflammation and enhances neurological outcomes in TBI despite implications to gut health.


Background
Traumatic brain injury (TBI) is a prevalent cause of disability and death in adults, frequently resulting from contact sports, military activities, or vehicular accidents [1].TBI is generally categorized into acute and chronic stages [2], each with unique consequences and in ammatory reactions that in uence the secondary in ammatory cascade and the recovery of the patient [3].Neuroin ammation plays a critical and potentially alterable role in this secondary injury, as supported by animal and human research [4].Individuals with moderate to severe single TBI might endure a persistent neuroin ammatory state, which can accelerate neurodegenerative processes in the brain.This condition may be exacerbated by a subsequent TBI [5].
The neuroin ammatory response involving microglia and macrophages to brain injury can damage the intestinal mucosa.This damage may alter the gut microbiota's composition and increase the intestinal barrier's permeability [6].Such disruptions can lead to intestinal in ammation, often described as a "leaky gut" [7].Furthermore, intestinal bacteria can release components, and their byproducts can enter the bloodstream, triggering in ammation through both peripheral and central immune pathways that directly activate the microglia in the brain [8,9].Bacteria can in uence various neurological processes through the microbiota-gut-brain axis, including myelination, neurogenesis, cytokine release, and microglial activation, including mood and cognition [10,11].The diverse microbial populations colonizing the gut have a symbiotic relationship with their host, essential for fermenting undigested carbohydrates, producing short-chain free fatty acids (SCFAs) like acetate, butyrate, and propionate from intestinal bacteria, metabolizing crucial substances, and protecting against pathogens.SCFAs support microglial maturation, innate immune responses [12], and energy metabolism in normal conditions and offer anti-in ammatory bene ts [13].However, disruptions to this symbiosis caused by antibiotic use or illness can lead to dysbiosis.
Gut dysbiosis is often characterized by a reduced abundance or complete loss of SCFA-producing bacteria such as Faecalibacterium, Christensenellaceae, Collinsella, Roseburia, certain strains of Ruminococcus, Bi dobacterium, Bacteroides, Parabacteroides, Oscillospira, some Clostridium species, and the mucin-degrading bacterium Akkermansia.There is also often an increase in potential pathogens, including members of the Enterobacteriaceae family, as well as Campylobacter, Enterococcus, Streptococcus, Staphylococcus, Fusobacterium, Veillonella, Ruminococcus, Megasphaera, and Deltaproteobacteria [14].TBI, through the brain-gut-microbiome axis, has been shown to cause changes in microbiota during both the acute and chronic phases [15][16][17][18][19][20].Our previous research found that TBI leads to microbiota dysbiosis, with a notable decrease in Lactobacillus gasseri, Ruminococcus avefaciens, and Eubacterium ventriosum [21].Besides, the Firmicutes-to-Bacteroidetes ratio decreased as part of the stress response after TBI [22].Clinical ndings from a recent study suggest that within 72 hours after severe trauma, patients with multiple injuries showed a decline in bene cial bacteria but an increase in Clostridiales and Enterococcus [23].
Antibiotics (ABX) are frequently administered in intensive care units to prevent infections and sepsis in TBI patients, thereby reducing injury-related and hospital-acquired infections [24].Early administration of ABX is associated with improved survival in TBI patients [25], and improves the TBI outcome [26].In preclinical studies, a standard approach to deplete enteric bacteria with minimal systemic absorption is to administer a broad-spectrum ABX cocktail [27].ABX can have rapid and potentially lasting effects on the microbiota.However, ABX can lead to gut microbiota dysbiosis, associated with altered signaling molecules and cognitive impairments [28].The administration of ABX in TBI patients is not without some risk due to potential reductions in healthy gut microbiota and its link to increased ABX resistance among bacteria [29].Further research is needed to understand the full impact of ABX on recovery from brain injuries and its relationship with microbiome dysbiosis.
In this study, we administered a broad-spectrum antibiotic cocktail to young male mice and noted that the elimination of intestinal bacteria leads to dysbiosis, decreased neuroin ammation, and enhanced outcomes after TBI.These results underscore the importance of further exploring the connections between the gut and brain microbiomes to identify possible therapeutic strategies for TBI patients.

Methods
Mice, administration of antibiotics and TBI model.
Young adult (14-week-old) C57BL/6J male mice (Jackson Laboratories, Bar Harbor, ME) were housed at the Houston Methodist Research Institute animal facilities under a standard 12-hour light and dark cycle with access to food and water ad libitum.All in vivo experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at Houston Methodist Research Institute, Houston (Texas, USA).To achieve microbiome depletion, an antibiotic (ABX) cocktail was made consisting of ampicillin (1 mg/mL), gentamicin (1 mg/mL), metronidazole (1 mg/mL), and vancomycin (0.5 mg/mL), dissolved in autoclaved drinking water.The cocktail was administered to ABX-treated male mice via oral gavage in a dosage of 200 mL for 2 consecutive days.The vehicle (VH)-treated mice underwent all procedures except microbiota depletion and were administered water as control VH via oral gavage.Experimental groups consisted of an ABX and VH treatment (n = 10 per group).Mice from both treatment groups were anesthetized with iso urane before receiving an initial controlled cortical impact (CCI) injury on the left hemisphere at the primary motor and somatosensory cortex using an electromagnetic Impact One stereotaxic impactor (Leica Microsystems, Buffalo Grove, IL, USA).The impact site was localized at 2 mm lateral and 2 mm posterior to Bregma with a 3 mm diameter at impact tip.The impact was made at a velocity of 3.2 m/s and impact depth of 1.5 mm.As speci ed in our earlier studies, these parameters were determined to induce an in ammatory response post-TBI [2,30].A second CCI was performed following the same parameters 38 days after the initial injury.Mice were anesthetized and sacri ced 5 days post-injury (dpi) (after the second CCI), and brains, blood, and small intestines were collected for further analysis (Fig. 1a).

Fecal Microbiome DNA extraction for 16S rRNA Sequencing
Gut microbiome DNA concentrations were obtained by collecting stool samples from the VH and ABX groups (Fig. 1a).After collection, stool samples were stored at -80°C.Genomic bacterial DNA was extracted from frozen stool samples using the QIAamp PowerFecal Pro DNA Kit (Qiagen, Germantown, MD).Bead beating was implemented in three cycles for DNA extraction, each lasting one minute at a speed of 6.5 m/s and a rest period of ve minutes between cycles.This mechanical disruption was conducted using a FastPrep-24 system (MP Biomedicals, Irvine, CA).Following the bead-beating process, DNA isolation was completed according to the manufacturer's instructions.The concentration of the extracted genomic DNA was then measured using a DS-11 Series Spectrophotometer/Fluorometer (DeNovix, Wilmington, DE).Extracted genomic bacterial DNA from collected fecal samples was analyzed for microbiota colonization and diversity by 16S rRNA gene compositional analysis.Illumina MiSeq was performed using the adapters and single-index barcodes so that the polymerase chain reaction (PCR) products could be pooled and sequenced directly, targeting at least 10,000 reads per sample [33].

Microbiome Data Analysis
Raw data les in binary base call (BCL) format were converted into FASTQs and demultiplexed based on the single-index barcodes using the Illumina 'bcl2fastq' software.Demultiplexed read pairs underwent quality ltering using bbduk.sh(BBMap version 38.82), removing Illumina adapters, PhiX reads, and sequences with a Phred quality score below 15 and length below 100 bp after trimming.16S V1-V3 quality-controlled reads were then merged using bbmerge.sh(BBMap version 38.82), with merge parameters optimized for the 16S V1-V3 amplicon type (vstrict = t qtrim = t trimq = 15).Further processing was performed using nf-core/ampliseq version 2.8.0 of the nf-core community work ows, implementing reproducible software environments from the Bioconda and Biocontainers projects [35][36][37][38].Data quality was evaluated with FastQC (version 0.12.1) and summarized with MultiQC (version 1.18) [39].Sequences were processed sample-wise (independent) with DADA2 (version 1.28) to eliminate any residual PhiX contamination, trim reads (forward reads at 275 bp and reverse reads at 265 bp; reads shorter than this were discarded), discard reads with > 2 expected errors, to correct errors, to merge read pairs, and to remove PCR chimeras.After clustering, 2097 amplicon sequencing variants (ASVs) were obtained across all samples.Between 6.77% and 34.22% of reads per sample (average 15.3%) were retained.The ASV count table contained a total of 1663253 counts, at least 4360 and at most 36800 per sample (average 16972).Taxonomic classi cation was performed by DADA2 and the database 'Silva 138.1 prokaryotic SSU' [41].ASV sequences, abundance, and DADA2 taxonomic assignments were loaded into QIIME2 (version 2023.7)[42].Within QIIME2, the nal microbial community data were collected into an .rdsle found within the phyloseq (version 1.44) folder of the nfcore ampliseq output [43].The resulting phyloseq data frame object was merged with the phylogenetic tree calculated within QIIME2.The phyloseq object was used in the creation of alpha and beta diversity plots, PERMANOVA calculations with vegan::adonis2 (version 2.6-5), relative abundance bar plots with microViz (version 0.12.1), next ow, and R (version 4.3.3)scripts, and differentially abundant taxa calculations using ANCOMBC2 [44].OTU-based analysis was also performed using Lotus2 [31].Samples were quality ltered, demultiplexed, and prepared for downstream processing using sdm.Taxonomic alignment to the 'Silva 138.1 prokaryotic SSU' was performed using the RDPclassifer [32].OTUs were clustered using VSEARCH and a nal phylogenetic tree was constructed from a MAFFT multiple sequence alignment using FastTree2 [33][34][35].Resulting OTU data analysis and visualization was performed using the.Rdata le produced from the Lotus2 work ow with the same R code written for the ASV analysis (Supplementary Fig. 1).All scripts used in both data processing and visualization can be found at https://github.com/microbemarsh/abx_depletion.Data are stored in the SRA database from NIH with the BioProject number: PRJNA1104663.Serum SCFA analysis.
Separation of metabolites was performed on Acquity UPLC HSS T3 1.8 um (2.1×100mM).The SCFA were measured in ESI negative mode using a 6495 triple quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA) coupled to an HPLC system (Agilent Technologies, Santa Clara, CA) with multiple reaction monitoring (MRM).The acquired data was analyzed using Agilent Mass Hunter quantitative software (Agilent Technologies, Santa Clara, CA).Raw peak intensity for each SCFA was normalized by sum, log transformed, and auto-scaled (mean centered and divided by standard deviation).

Rotarod Test
The sensorimotor function was evaluated by means of a Rotarod behavior test using the system from Ugo Basile (Gemonio, Italy).All mice underwent an initial training session of 3 trial runs, two days before behavior testing began.Following the completion of training, mice were assessed at the following time points: 24 h after ABX/VH treatment, 3 dpi, and 5 dpi to assess motor function following the second TBI.
Mice were placed on the stationary rod and allowed to explore for 30 sec.The stationary rod was then rotated, accelerating from 4 to 40 rpm for 5 min.Each trial ended when the animal fell off the rod.The latency to fall was recorded, and the mean value was established for each mouse.All behavior testing was conducted by an experimenter blinded to the animal groups.

Cresyl Violet staining
Fixed whole brain samples were sectioned using a cryostat (Epredia Cryostar NX50, Fisher Scienti c, Waltham, MA, US), in which 15 µm slices were processed and mounted directly onto glass slides or submerged in a free-oating cryo-protective solution (30% sucrose, 1% polyvinylpyrrolidone, 30% ethylene glycol, and 0.01M PBS).For immunohistochemical usage, collected sections were taken at coronal planes from the frontal cortex throughout the dorsal hippocampus.A cresyl-violet solution was prepared under a ventilated hood by combining 0.1% cresyl-violet (Sigma-Aldrich, St. Louis, MO, US), acetic acid, and distilled water.Pre-mounted brain sections were stained with cresyl violet solution for 10-20 min before dehydrating in ethanol dilutes.Lastly, slides were submerged in xylene before being covered with a xylene based Permount (ThermoFisher Scienti c) mounting media and coverslipped.The lesion volume was calculated as a percentage of the lesion area.Collected data was then averaged for each of the 9-12 brain sections per slide using ImageJ software.
Immunohistochemistry, cell death assay and quantitative analysis.
Free-oating brain sections were rst washed with a series of PBS and 0.5% PBS-Triton X-100 (PBS-T) before applying a 3% normal goat serum (NGS, #1000, Vector Laboratories, Burlingame, CA) blocking solution for 1 hour at room temperature (RT).A primary antibody solution made from blocking solution (PBS-T and 3% NGS) and dilutions of the following primary antibodies: anti-rabbit Iba-1 (1:500, Wako), anti-mouse CD68 (1:200, Biorad), anti-rat F4/80 (1:200, R&D Systems), anti-rabbit P2Y12 (1:500, Anaspec), anti-rabbit GFAP (1:500, Dako), and anti-rat Ly6B2 (1:500, Biorad), were incubated at 4°C overnight.The next day, brain sections were washed in PBS-T 3x5min and incubated with the corresponding anti-rabbit or anti-mouse Alexa Fluor 568-conjugated and anti-rat Alexa Fluor 488conjugated IgG secondary antibody (all 1:1000, Thermo Fisher Scienti c, Waltham, MA, USA) for 2 hours at RT.Samples were washed with distilled water 3x5min before being counterstained with DAPI solution diluted in PBS (1:50,000, Sigma-Aldrich) for 5 min.Sections were mounted and cover slipped using Tris Buffer mounting medium (Electron Microscopy Sections, Hat eld, PA).Cell death was evaluated separately using the Fluorescence In Situ Cell Death Detection kit (Roche Diagnostic, Indianapolis, IN, US).Brain sections were analyzed for DNA strand breakage using Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) following the manufacturer's instructions.All histological images were acquired on a Nikon uorescence microscope (Eclipse Ni-U, Melville, NY, USA) and confocal imaging system (Leica Microsystems, Deer eld, IL, USA).Areas of interest were the somatosensory cortex and thalamus regions of the brain.For the quantitative analysis of immunolabeled sections, we employed unbiased, standardized sampling methods to evaluate tissue areas in the cortex and thalamus that exhibited positive immunoreactivity.To measure the number of Iba-1 positive cells, we analyzed an average of ve single-plane sections from the lesion center (ranging from − 1.34 to -2.30 mm from bregma) for each animal, blind to the conditions, across each brain region.Within each region, all cells positive for Iba-1 CD68, F4/80, P2Y12, and Ly6B2 in ve speci c elds in the cortex and two speci c elds in the thalamus (x20, 151.894 mm²) near the impact site were counted.For proportional area measurements, we quanti ed the extent of the reaction for microglial and astroglial cells as the percentage of the area occupied by immunohistochemically stained cellular pro les within the injured cortex and thalamus regions.The data were presented as the percentage of area showing Iba-1 or GFAP positive immunoreactivity relative to the total studied area.We performed a quantitative evaluation of immunoreactive regions for Iba-1 and GFAP across 15 cortical and thalamus areas at the impact level.Quantitative image analysis of the staining's in the cortical and thalamic regions was performed using ImageJ64 software (NIH, Bethesda, MD, USA) as previously described [30] for inversion, thresholding, and densitometric analysis.The threshold function was utilized to set a black and white threshold corresponding to the imaged eld, subtracting the averaged background.The "Analyze Particles" function was then used to calculate both the total area of positive staining and the proportion of the total area.

Microglia Morphological analysis.
Immuno-stained anti-rabbit Iba-1 brain sections were imaged at a 40x oil objective using a confocal imaging system (Leica Microsystems, Deer eld, IL, USA) to further investigate microglia structure and morphology.Microglia in the somatosensory cortex and thalamus were reconstructed with Neurolucida morphometric software (MBF Biosciences, VT, USA) by using a 3D sectional plane.First, microglia somas were individually labeled to identify central points of each microglia.Dendrite branch mapping was then performed using the program's tracing tool.Once mapped, tracings were rendered into a 2D diagram using NeuroExplorer software (MBP Biosciences, VT, USA).A Sholl analysis was performed to assess cell complexity in relation to soma size and distance.Concentric circles spaced 5 µm apart, originating from the soma, were placed over each microglia.The number of dendrite branches that intersected the radius, the average dendrite length, the number of nodes (branching points), and the average surface area of individual microglia were measured as a function of the distance from the cell soma for each radius (Fig. 5c-f).The total data was graphed using GraphPad Prism 8 software (GraphPad; San Diego, CA, US) and mapped for signi cance in the area under the curve (AUD).

Gut Histology Staining
The small intestines of ABX and VH grouped mice were isolated and xed in 4% paraformaldehyde for 48 hours before being transferred to a 70% ethanol solution for dehydration.Samples were processed using a Shandon Exelsion ES Tissue Processor and embedded in para n on a Shandon HistoCenter Embedding System, as speci ed by the manufacturer's standard processing and protocols.Samples were sectioned at 5 µm thickness and mounted onto glass slides.Intestinal samples were depara nized in xylene before being rehydrated in water and then stained with a hematoxylin solution for 6 hours at 60-70 • C. The tissue was then rinsed with tap water to remove excess stain before being treated with 0.3% acid alcohol in water.It was then used to differentiate the tissue before being rinsed in tap water and stained with eosin for 2 min.Slides were then rinsed and mounted with a xylene-based Permount mounting medium and allowed to dry overnight.Alcian Blue staining was then performed to evaluate mucin production facilitated by goblet cells in the intestines.Intestinal samples were depara nized in xylene before being dehydrated in dilutions of ethanol and water.Slides were washed in distilled water before applying the Alcian blue solution for 30 minutes.Excess stain was removed using tap water before Nuclear Fast Red Solution was applied for 5 min.Samples were then rinsed and dehydrated before being dipped in xylene and mounted as previously described.

Statistical analysis
Statistical signi cance from SCFAs analysis was determined using multiple comparison t-tests within MetaboAnalyst6 [57].Orthogonal Partial least squares discrimination analysis (oPLS-DA) calculation and visualization was performed using the MetaboAnalystR [56, 58] package within R (version 4.3.3).Correlation heatmaps of key bacterial taxa to SCFAs were performed using normalized metabolomic data and microbial counts within a custom R script at https://github.com/microbemarsh/abx_depletion.Analysis of the rotarod test and histochemical/immuno uorescence utilized a one-way ANOVA followed by a Tukey's multiple comparison to compare the time after injury and sex as the independent variables.A post hoc test with Bonferroni multiple test correction was then applied.All data in the study were presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 were considered statistically signi cant.GraphPad Prism 8 Software (GraphPad; San Diego, CA, US) was used for statistical analysis.

Results
ABX treatment induce shifts in gut taxonomy composition.
The experimental design involved inducing a rst TBI, followed by microbiome sequencing at various timepoints: baseline, 6 days post-rst TBI (acute phase), 35 days post-rst TBI (chronic phase), after a course of ABX or VH, and 5 days post-second TBI (Fig. 1a).Bacterial DNA concentration in fecal samples showed signi cant decreases following ABX treatment, exhibiting lower DNA concentrations than VH after treatment (Fig. 1b, ***p < 0.001) ensuring the effectiveness of the ABX depletion.
The relative abundance of bacterial taxa at the phylum and genus levels are represented in two experimental groups: Group I (VH) had relatively stable microbiomes across the time points, while Group II (ABX) showed signi cant shifts in composition (Fig. 1c-f).TBI induced the change of gut microbiota components in both groups.Relative abundance at the phylum level indicated changes in the balance of major phyla like Bacteroidota, Firmicutes, Verrucomicrobiota, Proteobacteria, and Actinobacteria (Fig. 1c).Relative abundance at the genus level (Fig. 1e, f) demonstrated signi cant changes following ABX treatment, revealing a more diverse microbial composition, with notable alterations post-second TBI.
Gut bacterium diversity and concentrations shift following TBI and ABX treatment.
SCFA pro les and bacterial genus correlations were examined following ABX treatment and VH in a TBI mouse model.The heatmap shows the normalized peak intensity of various SCFAs in blood samples from two groups of mice, VH and ABX-treated (Fig. 3a).Each row corresponds to a speci c SCFA, and each column represents an individual mouse identi ed by unique Mouse IDs.Differences in SCFA pro les are evident, with each group displaying a distinct pattern of SCFA production.Orthogonal partial least squares differential analysis (oPLS-DA) of SCFAs (Fig. 3b) between VH and ABX mice showed clear separation, with each point representing an individual mouse sample.The axes represent the maximal variance in SCFA levels between groups, indicating signi cant differences in SCFA pro les.The percentage of variance for the T score is 17.6%, and for the orthogonal T score is 46.5%, providing insights into the underlying variance structure.Additionally, Pearson's correlation heatmap and hierarchical clustering show the relationship between key bacterial genera and SCFA production for VH and ABX (Fig. 3c, d).Speci cally, the production of butyrate, propionate, 3-methyl-valeate and are associated with the abundance of Erysipelatoclostridium (q = 6.2e-3, q = 0.045, and 3.3e-3 respectively), and the increment of 2-methyl-butyrate levels are associated to Alistipes (q = 0.039) (Fig. 3c).This analysis highlights the intricate relationships between SCFA pro les and bacterial genera, suggesting that ABX treatment following TBI does not signi cantly affect these dynamics (Fig. 3d).
Deletion of the gut microbiota alleviates the neuroprotective effect in mice with TBI.
To eliminate the in uence of gut microbiota differences on experimental results, animals were treated with ABX for 2 days before the second TBI.Subsequently, the effects of ABX treatment on lesion volume and apoptotic cell death after TBI were examined using TUNEL staining and histological analysis.The cortical lesion size was signi cantly 25% smaller in the ABX-treated group compared to the VH group (Fig. 4a) at 5 dpi.Representative histological images of the cresyl-violet brain sections showed the reduced lesion area in the ABX group (Fig. 4a, c).TUNEL staining, used to detect apoptotic cells in the injured brain.In the cortex, the ABX group had fewer TUNEL-positive cells (18.2 ± 1.9) than the VH group (26.8 ± 2.4), indicating reduced cellular loss (Fig. 4d).Similar trends were observed in the thalamus (Fig. 4g), where the ABX group exhibited signi cantly fewer TUNEL-positive cells (104.8 ± 9.8), than the VH group (186.2 ± 17.3).The representative TUNEL staining images (Fig. 4e-f, h-i) further demonstrate these ndings, showing reduced apoptotic cells in the ABX group.These results suggests that ABX therapy might offer neuroprotective bene ts in mitigating brain damage and cell death post-TBI.
Microglial density and morphology were analyzed in the cortex and thalamus following the second TBI in VH and ABX-treated groups.Quanti cation of Iba-1-positive cells showed a signi cant decrease in both the cortex (101.9 ± 4.9, p = 0.03) (Fig. 5a) and thalamus (263.6 ± 20.7, p < 0.001) (Fig. 5b) of ABX mice as compared to the VH (121.9 ± 6.4 and 364.7 ± 13.6, respectively).Representative Iba-1 staining images in the cortex and thalamus are shown in panels (Fig. 5a1-a4) and (Fig. 5b1-b4), respectively, demonstrating the morphological differences between VH and ABX-treated groups.To further analyze microglia processes and morphology following TBI, a Sholl analysis of microglial structural characteristics was assessed by plotting the radial distance of corresponding dendrite intersections, lengths, and nodes, from the soma.First, a representative Iba-1 image was acquired through confocal imaging (Fig. 5c).Individual microglia somas were established (Fig. 5d) before manually mapping dendrite branching (Fig. 5e).Lastly, the Sholl analysis was conducted for each microglia by adding concentric circles 2um apart, starting at each microglia soma (Fig. 5f) to evaluate process complexity.The speci c measurements of the number of dendritic intersections, dendrite branching length, nodule processes, and microglia surface area were measured and plotted using Neurolucida software.The Sholl analysis results (Fig. 5g-l) in the cortex and thalamus indicate signi cant differences in microglial process intersections, average process length, and average nodes.There was a signi cant decrease in the overall average number of intersections (24.6 ± 9.5, p = 0.03) and average dendrite length (70.7 ± 29.4, p = 0.04) in the cortical region of mice that received ABX compared to the VH mice (Fig. 5g, h).No other signi cant morphological changes were present in the average number of nodes and average surface area in the somatosensory cortex (Fig. 5i), and no signi cant changes overall in the thalamus (Fig. 5j, k,  l).These results suggest that ABX treatment after TBI reduces microglial/macrophage density and alters morphology, with fewer processes and reduced complexity, indicating a potentially reduced in ammatory response compared to the VH-treated group.
ABX treatment reduces microglia/macrophage activation and astrogliosis but does not affect neutrophil in ltration following TBI.

Impact of antibiotic treatment on gut histology following traumatic brain injury.
We sought to determine the effects of ABX treatment on overall small intestinal microarchitecture (villi length/width and crypt depth/width) that impacts nutrient absorption.Small intestine samples were stained with hematoxylin and eosin to label intestinal villi and crypt structure and with Alcian blue to measure goblet cell and mucin production.ABX treatment induced a signi cantly shorter villi width (Fig. 7a, 43.3 ± 1.8, p = 0.04) versus VH (48.7 ± 1.3).No signi cant differences in crypt and villi length (Fig. 7c, d) were found between ABX and VH groups.Next, we examined the impact of ABX treatment on goblet cells and mucin production.These cells are key components of the mucosal barrier and play important roles in host defense against intestinal pathogens.Compared to controls, ABX-treated mice had a signi cantly fewer goblet cells per eld (Fig. 7g, 7h), in both the villi (7.3 ± 0.4, p < 0.001) and crypt (1.5 ± 0.06, p < 0.001) regions of the small intestines versus VH-treated (10.5 ± 0.2 and 2.3 ± 0.1, respectively).These results suggest that ABX treatment after TBI induces signi cant changes in gut histology, notably broader villi and a reduction in goblet cells.
(a) Villi width and (b) length were quanti ed, demonstrating a reduction in villi width in antibiotic treatment (ABX) compared to vehicle (VH) mice.Crypt width (c) depth (d) measurements show no signi cant differences between treatments.Representative hematoxylin and eosin-stained sections of the damage small intestines after TBI from VH (e) and ABX (f) treated mice, illustrating the overall architecture and cell morphology.Lines in the image in red indicated the length of the villi.The ABXtreated group shows a signi cant decrease in goblet cell count in the villi (g) and crypts (h) of the ileum.Alcian blue (blue) and nuclear fast red (pink) staining of the small intestine sections from VH (i) and ABX (j) treated mice, highlighting the mucin-producing goblet cells per crypt (yellow head arrow) and goblet per villi (orange arrow).Scale bars: (e, f, I, j) = 100 µm.Mean ± SEM. n = 9/group.*(p < 0.05), and *** (p < 0.001).

Discussion
In our research, we observed that a broad-spectrum ABX treatment signi cantly altered the composition of the fecal microbiota following TBI in mice (Figs. 1, 2).Our results revealed that ABX treatment resulted in signi cant shifts in the relative abundance of various bacterial taxa at both the phylum and genus levels, underscoring the profound impact of ABX therapy on gut microbial communities.
We administered a combination of antibiotics (ampicillin, gentamicin, metronidazole, and vancomycin) to deplete the gut microbiota.Ampicilin-induced changes in the microbiota, decreasing the bacterial diversity, the expression of MHC class I and II genes, and elevated mast cell protease expression in the intestine [36].Mice treated with gentamicin shifted intestinal community and impacts in the gut metabolome decreasing SCFA [37].Metronidazole is an antimicrobial that selectively targets anaerobic bacteria [38], and it has been shown to affect the gut microbiota by slightly decreasing the Firmicutes to Bacteroidetes ratio, primarily due to an increase in Bacteroidetes [39].Finally, vancomycin, a narrowspectrum ABX effective against Gram-positive bacteria like Staphylococcus and Clostridium and has been shown to diminish microbiome diversity and change its composition, reducing the production of SCFAs [14].
We found a clear split in bacterial diversity one day after ABX treatment, with the VH group exhibiting more uniformity while the ABX-treated group displayed greater disparity by 5 dpi.These shifts after ABX treatment indicate a transition from a balanced gut microbiome toward a more dysbiotic state.We noted that alpha and beta diversity differed only following ABX treatment, maintaining uniformity throughout the study.However, pronounced beta diversity changes following ABX treatment has been previously observed by others in studies on microbiome disruption in adolescent and adult rats following TBI [40].In clinical studies, factors such as ABX exposure and the occurrence of infection were linked to larger disparities in the rectal and oral microbiomes between TBI patients and healthy controls [41].
Our ndings revealed several signi cant shifts in the fecal microbiome in the acute and chronic phases after the rst TBI.Notably, we observed an increase in the abundance of genus Lachnospiraceae UCG-006 during both phases compared to baseline levels.The genus Lachnospiraceae UCG-006 is known to produce SCFAs, such as butyrate, which have anti-in ammatory properties [42].This aligns with previous studies that suggest a positive correlation between Lachnospiraceae UCG-006 and antioxidant enzyme activity [43].A higher relative abundance of the genus Lachnospiraceae UCG-006 is associated with positive health outcomes after infection with S. Typhimurium [32].Thus, the increase in this genus could suggest its contribution to repair processes after TBI, potentially lessening in ammation and enhancing recovery.This increase was accompanied by a signi cant reduction in Bi dobacterium and Lactobacillus, corroborated by our previous research [44].Bi dobacteria were signi cantly negatively correlated with the expression of cortical genes related to in ammatory responses.[45].
In the acute stage of the rst TBI, we found a marked decrease in bene cial bacteria like Bi dobacterium and Lactobacillus (Fig. 2i), as we have previously published [44].Similarly, we observed a reduction in Enterorhabdus during the chronic phase (Fig. 2j).Interestingly, the Enterorhabdus genus includes pathogenic bacteria that were rst isolated from a mouse model of intestinal in ammation [46].In an animal model of dietary-induced nonalcoholic fatty liver disease (NAFLD), the probiotic L. plantarum NA136 treatment relieved insulin resistance and signi cantly increased relative proportions of Enterorhabdus [47].Studies using mouse models have demonstrated that speci c probiotic bacteria, like Lactobacillus, can enhance vagus nerve activity, modify brain neurotransmitter levels of GABA, glutamate, and serotonin, and reduce anxiety-related behaviors [48].Similarly, since both Bi dobacterium and Lactobacillus can produce GABA following TBI in mice [45], it is plausible to consider that they may also act as neuroprotective probiotics as we have demonstrated in our previous work [49].Another study found that Lactobacillus may play a neuroprotective role by reshaping the intestinal microbiota of TBI mice [50].These changes suggest that TBI can disrupt the balance of bene cial and harmful bacteria, potentially impacting brain recovery.In Group I of mice, which were not treated with ABX 38 days postrst TBI, there was an observed increase in Ruminococcus.This genus is also recognized for producing SCFAs, especially butyrate, suggesting a potential role in reducing in ammation and aiding brain recovery [51].However, our metabolite data did not show a correlation with any of the increased SCFAs analyzed.
After the second TBI, the ABX-treated group experienced acute dysbiosis, with a decrease in Lolidextribacter, Akkermansia, Bacteroides, Lachnoclostridium, and Parasutterella, alongside a slight increase in Duboisella.A decline in gut barrier-protective bacteria like Akkermansia could lead to a more permeable gut lining [52], allowing in ammatory mediators to enter the bloodstream and potentially reach the brain, further exacerbating neuroin ammation.Treatment with A. muciniphila was shown to improve intestinal injury, neurological dysfunction, and neuroin ammation in the cerebral cortex of mice with TBI [53].Previous clinical studies have shown that individuals with lower bacterial diversity before receiving ABX may be more vulnerable to opportunistic or pathogenic species, such as Lachnoclostridium [54].An increase in Duboisella, given the overall context of dysbiosis, could indicate a shift in microbial balance, indicating ongoing disruption and a potential risk of further in ammatory conditions [55].These ndings show that depleting the pathogenic microbiome could be a therapeutic target to decrease in ammation and help to the brain recovery.
ABX decrease levels of acetate, n-butyrate, and propionate, which are products of microbial fermentation, and reduce adenine, cytosine, guanine, and uracil due to an overall reduction in bacterial load [28].In our study, we did not nd no changes in neuroprotective SCFA levels after ABX treatment, though neuroin ammation and injured brain regions were decreased, likely through mechanisms not directly tied to SCFAs.Erysipelatoclostridium are in ammation-associated microbes, and they contributed signi cantly to metabolites synthesis in the gut [56].Butyrate, considered the most signi cant SCFA, is produced by a broad phylogenetic array of bacteria.This includes many types of Firmicutes, such as Ruminococcus, Clostridial Clusters IV and XIVa, Eubacterium, Anaerostipes, Coprococcus, Faecalibacterium, and Roseburia, as well as certain species of Bacteroides and Bi dobacterium [57].Our analysis also shown butyrate and propionate as key metabolites associated with Erysipelatoclostridium as an essential gut bacterial genus.Furthermore, in TBI mice treated with VH, serum butyrate and propionate levels were positively correlated with the abundance of Erysipelatoclostridium and negatively correlated with levels of 2-methyl-butyrate.This indicates that ve days after TBI, the microbiota capable of producing SCFAs is replenished in animals not subjected to antibiotic treatment, promoting recolonization with bene cial bacteria.This restoration does not happen in the group treated with ABX.
We utilized short read 16S rRNA amplicon pro ling to analyze microbial populations, which allowed us to identify and assess the relative abundance of individual bacterial taxa.While we employed best practice microbiome analysis tools for short read 16S analysis [58] analyzing changes in microbiome data due to limited resolution of short regions of 16S limit our analysis down to genus levels [59].While this provides a high-level view of the bacterial communities inhabiting the mouse gut, it does not allow us to interrogate speci c species and/or strains directly responsible for metabolite and SCFA production interfacing with the gut-brain axis Also, given the compositional nature of microbiome analysis, relative abundance analysis can introduce bias and measurement error [60].
The gut microbiota has been shown to in uence BBB permeability [61].The host-microbiota in uences brain levels of numerous neurochemical mediators crucial for neuronal function [62].Gut microbiota regulates the innate immune functions of microglia, which prepares the brain to respond rapidly to pathogens or threats.After administering ABX, it was observed that the intestinal microbiota in uenced the neuroin ammatory response and brain injury following neonatal hypoxia-ischemia in mice [63].Another study showed that administering ABX prior to TBI in mice helped lessen early neuroin ammatory responses, yet it did not signi cantly affect brain histopathology [64].Furthermore, the lack of a complex host microbiota leads to de ciencies in microglia maturation, differentiation, and function.Consequently, mice treated with germ-free (GF) and ABX conditions showed impairments in microglial functionality [12].Our neuropathological analysis revealed that microglia/macrophage cells and apoptotic cells were signi cantly lower in ABX-treated mice, indicating potential neuroprotective bene ts, align with existing studies showing that the depletion of gut bacteria can attenuate the neuroin ammatory response [65].Other studies have shown that although ABX may contribute to a reduction in lesion volume following TBI, it adversely reduces monocyte in ltration, worsens neuronal loss, and increases microglial pro-in ammatory markers following a single TBI [66].Despite the neuroprotective effects of ABX, our results did not show signi cantly alter neutrophil in ltration in the cortex and thalamus, indicating that certain aspects of the in ammatory response remain unaffected.ABX treatment had a notable impact on gut pathology, leading to signi cant changes in the small intestine's structure.Clinical data on mucosal barrier function following TBI are sparse and limited to ICU patients [17].Goblet cells, specialized epithelial cells in the gastrointestinal tract responsible for producing and secreting mucin, play a critical role in maintaining the mucus barrier.A reduction in goblet cell count can weaken this barrier, which could result in immune dysregulation and chronic in ammation.
Our ndings showed a signi cant decrease in goblet cell count indicating a disruption in the gut's protective barrier.This shift suggests that ABX treatment can impact gut homeostasis, possibly disrupting mucin production and other protective mucus components.

Conclusions
Collectively, this study indicates that ABX treatment following TBI has complex effects, including shifts in gut microbiome diversity and composition, reduced neuroin ammation, and changes in gut pathology.Further research is needed to understand the long-term effects of ABX treatment on brain-gut interactions and post-TBI recovery to help optimize therapeutic strategies for TBI patients, aiming to reduce neuroin ammation while preserving gut health.(c, d) Relative abundance at the phylum level, color-coded for major phyla including Bacteroidota, Firmicutes, Verrucomicrobiota, Proteobacteria, Actinobacteria, and others.(e, f) Relative abundance at the genus level, showing detailed shifts in microbial composition with a diverse set of genera colorcoded for ease of differentiation.Both sets of bar charts reveal the impact of ABX on microbial diversity and the subsequent changes following a second TBI (5 dpi).Alcian blue (blue) and nuclear fast red (pink) staining of the small intestine sections from VH (i) and ABX (j)treated mice, highlighting the mucin-producing goblet cells per crypt (yellow head arrow) and goblet

Figures
Figures

Figure 1 Effects
Figure 1

Figure 2 Changes
Figure 2

Figure 3 Short
Figure 3