Loss of Fmr1 reorganizes the multi-elemental composition across tissues in Fragile X Syndrome mice

Sabiha Alam; Jamie T. Reeves; Shawn M. Wilder; Elizabeth A. McCullagh

doi:10.1371/journal.pone.0352693

Abstract

Fragile X Syndrome (FXS) results from a genetic mutation which silences the expression of Fragile X Messenger Ribonucleoprotein (FMRP). FMRP serves various roles regulating cellular protein synthesis including mRNAs that code for proteins regulating ion flux. However, there are few studies measuring the elemental balance between FXS genotypes and tissues. Here, we measured the multivariate balance of 10 elements in tissues of wild-type and Fmr1-knockout mice to compare elemental composition of brain and somatic tissues within and across genotypes. Using a Bayesian mixed model approach, we found that the main differences between groups were between tissues, with significant effects of genotype and interaction of tissue on genotype. Wild-type feces were significantly higher in magnesium and sodium than knockout. Fur was significantly higher in potassium in wild-type, which was supported by the interaction effect of genotype with tissue. These results align with previous work showing FXS pathologies alter electrolytic and metal ion regulation, neuronal excitability, and gastrointestinal function. Future work should additionally test how elemental differences relate to function at the cellular level, as well as patterns of individual intake, digestion, assimilation, and/or excretion.

Citation: Alam S, Reeves JT, Wilder SM, McCullagh EA (2026) Loss of Fmr1 reorganizes the multi-elemental composition across tissues in Fragile X Syndrome mice. PLoS One 21(7): e0352693. https://doi.org/10.1371/journal.pone.0352693

Editor: Nien-Pei Tsai, University of Illinois at Urbana-Champaign, UNITED STATES OF AMERICA

Received: March 27, 2026; Accepted: June 13, 2026; Published: July 10, 2026

Copyright: © 2026 Alam et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting information files as well as linked external data support (Figshare https://doi.org/10.6084/m9.figshare.32592354).

Funding: This work was partially funded by an NIH NIGMS COBRE (5P20GM152333-02) pilot award through the Oklahoma Center for Microbiome Research to EAM. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. There was no additional external funding received for this study.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Fragile X Syndrome (FXS) is a neurodevelopmental disorder which results from a mutation in a single gene, Fmr1 (Fragile X Messenger Ribonucleoprotein 1), on the X chromosome [1,2]. Mutation of this gene reduces the expression of its protein product, Fragile X Messenger Ribonucleoprotein (FMRP) [3,4]. Because FMRP is an RNA-binding protein, it affects a wide range of biological processes, including gastrointestinal function, synaptic activity, and neural development, among others. The diversity of these effects makes it difficult to catalog all downstream consequences of Fmr1 disruption at the molecular level. Because all biological systems (e.g., cell, tissue) are made up of approximately 20 chemical elements (depending on structure, and process; [5], the relative abundance of all elements encompassing a system (i.e., the ionome) are appropriately conceptualized as a working unit [6,7]. Responses of the ionome to genetic perturbations have been used to advance understanding of various phenotypes [8] including disease [9–13] which could be considered complementary to other techniques such as transcriptomics [14,15] and proteomic analyses [16,17].

Fmr1 encodes an RNA-binding protein (FMRP), and its loss affects diverse aspects of cellular physiology by disrupting post-transcriptional regulation of many target mRNAs. FMRP associates with ribosomes and polyribosomes to control the translation of transcripts involved in synaptic signaling, ion channel regulation, and metabolic processes [18,19]. Consequently, Fmr1 disruption alters both the synthesis and localization of proteins responsible for ion transport and storage (e.g., voltage-gated channels, metalloregulatory enzymes; [20–22]. Thus, FXS may secondarily influence elemental balance across tissues. These alterations in translational efficiency and post translational modifications can cascade into systemic changes in ionomic composition, potentially influencing neuronal function.

Individuals with FXS are often diagnosed with autism spectrum disorder (ASD), and exhibit overlapping clinical features such as antisocial behavior, communication deficits, cognitive impairments, and language delays [23]. Several studies have identified associations between ASD and specific elements: supplementation with zinc promotes restoration of synaptic proteins such as Shank3 and Shank2 and helps restore excitatory–inhibitory balance [24]. Magnesium, particularly in combination with vitamin B6, further mitigates neurobehavioral disorders in ASD [25,26]. Moreover, recent research demonstrates that ASD-related symptoms extend beyond neural dysfunction, including disruptions in balance across gut and peripheral tissues (iron - [27,28]; zinc – [29]; various trace metals – [30]. However, focusing on single elements or tissues, while informative, risks missing the broader systemic interactions that emerge from the coordination of multiple elements. Without a comprehensive ionomic perspective, the integrative patterns linking neural, gastrointestinal, and metabolic functions may remain unresolved.

In this context, we conducted an experiment to examine how the multielemental composition, the ionome, of neural and somatic tissues as well as digestive biproducts (feces) differs in the context of FXS between wild-type and knockout strains of mice. Because elemental concentrations are interdependent, these data were analyzed within a compositional framework that captures relative, rather than absolute, changes among elements. Using this approach, we explored the potential influence of Fmr1 loss on overall elemental balance among brain regions and somatic tissues. We hypothesized that male mice lacking Fmr1 would exhibit distinct multielement compositions relative to wild-type males. To test this hypothesis, we quantified and compared the ionomic composition of gut, brain, and noninvasive tissues such as fur and feces between genotypes. This compositional data analysis provides insight into systemic elemental reorganization in FXS and identifies tissues that may serve as noninvasive proxies for elemental diagnostics in clinical contexts [31].

Methodology

Experimental animals and design

In our study, we used an Fmr1 KO (stock #003025, Fmr1 KO) mouse model which recapitulates some core symptoms of FXS patients along with C57BL/6J (stock #000664, B6) wild-type background (control animals), which were obtained from the Jackson Laboratory and bred at Oklahoma State University [32]. All animals included in this study were male WT (n = 8) and KO (n = 6) mice ranging from 99–156 days of age at the time of tissue collection. Experimental groups were age-matched as closely as possible to minimize potential age-related effects on tissue elemental composition. Animals were group housed with littermates and generated through genotype exclusive mating (wildtype to wildtype and FXS to FXS). The results therefore reflect data from two cages of animals per genotype. All mice were on a 12-hour light cycle (6 AM- 6 PM) and received the same chow (LabDiet 500, Lab Supply, Fort Worth, TX) and bedding (alpha-dri, Innovive, San Diego, CA) and additional nestlets (cotton NST, Lab Supply, Fort Worth, TX). Since FXS is an X-linked trait and therefore more common in males, we limited the study to male animals. All experimental procedures were conducted under appropriate laws and NIH guidelines and our study principles received approval from the Oklahoma State University Institutional Animal Care and Use Committee (protocol 20−07).

Tissue collection and preparation

Feces were collected from animals as they were passively in the holding cage, or we gently held the mice by the base of their tails and waited around 5 minutes until they defecated. After defecation, we used forceps to collect fresh fecal pellets and immediately placed those pellets into pre-labeled sterile tubes. Next, we euthanized mice by exposing them to isoflurane overdose. After confirming the death of the mice by lack of respiration and toe pinch reflex, we decapitated them and harvested their brains. We placed brains in a Petri dish and dissected the brains into specific brain regions including olfactory bulb and striatum. After removing the above-mentioned dissected brain regions, we pooled tissue from the pons, medulla, hippocampus, thalamus, hypothalamus as brain PMHTH (Figure S1 in S1 File). Since the brains were dissected fresh, these segregations (or pooling) indicated regions we were confident were isolated from other regions. We then transferred all dissected brain regions into a 2 mL Eppendorf tube for subsequent processing. Additionally, we collected samples from cecum and included fur for each individual mouse. To collect the cecum, we opened the abdominal cavity with a midline incision from the lower abdomen to the sternum, which exteriorized the intestine to expose the cecum—a large, pouch-like structure located at the junction between the ileum and the colon. We immediately transferred the cecum to a sterile, 2 mL Eppendorf tube. Later, to separate the cecal contents from the cecum tissue, we opened the cecum longitudinally on a sterile Pedri dish, then collected cecal contents and discarded the cecal tissue. For fur samples, we used clean, sterile scissors to trim the fur from the abdomen. Then, we transferred the clipped fur into a pre-weighed and labeled Eppendorf tube using sterile tweezers.

Ionome measurement

Once tissues were collected and stored, we dried them at room temperature in an oven at 55 °C for 72 hours. Then, we homogenized dried tissues using a mortar and pestle and weighed 4–12 mg for each sample. We digested homogenized samples using 100% trace metal grade HNO3 and 100% trace metal grade H2O2 in a 2:1 ratio; allowing them at least 24 hours for complete digestion before performing elemental analysis using inductively coupled plasma optical emission spectroscopy (ICP-OES, iCAP7400; ThermoScientific, Waltham, MA). We measured the concentrations of 10 biologically relevant elements in brain tissue using ICP-OES: Calcium (Ca), Copper (Cu), Iron (Fe), Potassium (K), Magnesium (Mg), Manganese (Mn), sodium (Na), Phosphorus (P), Sulfur (S), and Zinc (Zn). We retained wavelengths with 90% of measured samples within the limits of detection established by standard curves and replaced values over and under limits of detection with upper and lower limits. Then, we averaged emittance of multiple wavelengths when more than one wavelength quantified an individual element.

Of the 84 total samples analyzed for 10 elements (840 readings), 4 readings of sulfur, and 8 readings of zinc were missing values. Missing sulfur values included WT olfactory bulb #170, KO olfactory bulb #157, WT fur #162, KO PMHTH #142. Missing Zn values included KO PMHTH # 142 and 157, WT fur # 150, KO fur #151, WT olfactory bulb #169 and 170, KO olfactory bulb # 157 and 158. Thus, we retained n_WT = 6 and n_KO = 4 for sulfur, and n = 3 for zinc in both genotypes.

To account for missing values, we imputed missing values “impCoda” function from the “robCompositions” R package, which is specifically designed for compositional data [33]. A few missing values are common in ICP-based elemental analyses due to matrix effects and instrument detection limits [34]. Tissue elemental concentrations represent compositional data and underwent additive log-ratio transformation to account for their inherent constraints.

Concentrations of each element (µg/mg) were converted into raw masses, and we used the total original mass of each sample as the denominator in ALR transformation. For each element, we then calculated its proportion of the total sample mass (Element/Total) to standardize comparisons across samples in log space. These proportional values represent the relative abundances of each element within the sampled elemental pools suitable for robust statistical analysis and inference [44]. Here, we chose to use the total sample mass as opposed to measured mass because the proportions of original sample mass quantified by measurement were quite low in all tissues (3–20%). Given that we did not quantify C and N due to the constraints of ICP-OES and sample availability, we use total sample mass in the denominator to avoid assuming independence between measured elements and the mass of unmeasured elements. Although one could measure all known elements and still not quantify the entire mass (i.e., 100%) of a sample given the dynamic condition of these systems, we chose this quantity to reflect our certain knowledge that there are many elements in unmeasured mass which we chose not to quantify, including elements which often compose substantial proportions of biological tissue (e.g., C and N).

Mixed modeling

We used a Bayesian multivariate mixed linear model to evaluate the effects of genotype (WT vs KO) and tissue (i.e., feces, striatum, cecum, olfactory bulb, fur, and PMHTH) on patterns of observed elemental variance. We used R package ‘brms’ to conduct the mixed modeling workflow (Bürkner, 2017) and used default values for all prior distributions. We conducted all analyses using statistical program R ver 4.5.2 (R Core Team, 2025).

We modeled each observation (y_i) as a multivariate normal response variable with identity link, linear predictor f(η_i) and deviance (σ) (Equation 1). An observation represents one measurement of 10 elements (y_i = [y_iCa, y_iCu, y_iFe, … y_iZn]) from a single tissue within an individual, with expectation (η_i) and residual deviance (σ). Although we sampled six tissue types, we chose not to include a flexible deviance parameter for each tissue due to limitations in sample size (n_WT = 8; n_KO = 6).

The linear predictor (η_i) includes the design vector (X_i) which maps population-level coefficients (tissue, genotype, and tissue x genotype interaction) onto individual observations (Equation 2). The linear predictor also includes the vector of element-wise group-level coefficients (μ_j[i]) for observation i of individual j (i.e., random effect of individual on baseline concentrations across all tissues; Equation 3). Group-level coefficients were derived from a multivariate normal distribution centered at zero with covariance matrix (Σ_μ), which is computed as the squared deviance of random element intercepts (σ_μ²) times the correlation matrix of all elements (R_μ; Equation 4).

The residual error term (ϵ_i) is a multivariate normal variable centered at zero with covariance matrix (Σ_ϵ - Equation 5). Residual error includes all variance not explained by the linear predictor. The residual covariance matrix, like that in the group-level coefficients, was computed as the product of the squared residual deviance of random element intercepts (D_ϵ²) times the residual correlation matrix of all elements (R_ϵ - Equation 6).

As is standard in the ‘brms’ package, we initialized a Markov Chain Monte Carlo (MCMC) sampler with 4 chains of 2,000 iterations each. In each chain, we used the standard burn-in of 1,000 iterations and retained 1,000 iterations from each chain. Thus, the maximum possible effective sample size was 4,000 for each parameter.

Model validation

To evaluate MCMC convergence, we assessed whether rank-normalized split Gelman-Rubin diagnostics (Ȓ) for all estimated parameters met the validation threshold of Ȓ < 1.05 (Vehtari et al., 2021). Additionally, we evaluated the effective sample sizes of model parameters, including the bulk and tail measures (ESS_bulk and ESS_tail). These measures quantify the exploration of high-probability and tail posterior parameter spaces by all chains. We used a standard threshold of ESS_tail > 400 and ESS_bulk > 400 to assess chain convergence and stability.

After obtaining model results, we performed posterior predictive checks and assessed residual dispersion. Posterior predictive checks allow visual comparison of the observed and predicted distributions. We performed posterior checks only for elements which displayed genotype-level contrasts in at least one tissue with 95% Highest Posterior Density (HPD) interval not overlapping zero. To assess residual dispersion, we extracted residuals of conditional predictions for all samples and stratified residual error post-hoc based on observation-level genotype and tissue factors. Then, we calculated the standard deviation (SD) and interquartile range (IQR) of residuals for each group and compared values across groups.

Model evaluation

We computed the posterior R² distributions for each element to quantify explanatory power and related uncertainty. Then, we inverted the residual covariance matrix (Σ_ϵ) to obtain the precision matrix (Ω_ϵ = Σ_ϵ⁻¹) and extracted isolated element-wise error as the inverse of values on the diagonal of the precision matrix (i.e., element error independent of covariance). We thus decomposed posterior variance into population-level components (tissue, genotype, and tissue x genotype interaction), group-level components (individual), shared residual error (elemental covariance), and isolated residual error.

We performed post-hoc contrasts of predicted values across experimental groupings to compute estimated marginal means. We performed pairwise contrasts across all population-level coefficients (tissue, genotype, tissue x genotype) by directly computing the difference between predicted group-level values across each iteration in retained samples of all four chains. We consider contrasts indicative of an observed effect when 95% HPD intervals do not cross zero and report the posterior mean magnitude and direction of the effect on the log-ratio scale.

(1)

(2)

(3)

(4)

(5)

(6)

Results

Global model

Overall, our model demonstrated strong evidence of convergence (maximum Ȓ = 1.006) and stability (minimum ESS_bulk = 514, ESS_tail = 880; Table S1 in S6 File). In addition, predictive distributions were generally well-aligned with observations (Figure S1 and S2 in S1 and S2 Files, Table S2 in S7 File). Residual dispersion from posterior predictions (SD and IQR) did not differ between KO and WT mice (Table S1 in S6 File), but striatum demonstrated significant overdispersion relative to other tissues (~ 4 times higher; Table S2 in S7 File). Estimates and credible intervals for model parameters are available in Table S3 and S4 in S8 and S9 Files).

The overall model predicted effects of genotype, tissue, and interaction on elemental ratios (Figs 1 and 2). Total variance explained by population-level (fixed) effects varied among elements, but median estimates exceeded 0.75 in 70% of elements (Table 1). The population-level effects explained nearly all variance in Mg and Mn (median 98 and 99%, respectively) with high confidence. However, Ca, S, and Zn generally were not reliably predictable, as median estimates were between 0.28 and 0.46 and uncertainty was substantially higher than other elements (Table 1).

Download:

Table 1. Predicted explanatory power (R-squared distribution) of total population-level (i.e., fixed) effects on individual elements. R-squared was iteratively calculated from posterior predictions using 4,000 draws for each element. Columns represent median estimate, estimated error, and quantiles.

https://doi.org/10.1371/journal.pone.0352693.t001

Download:

Fig 1. Estimated marginal concentration profiles across tissues and genotypes.

Facets track model-predicted elemental concentrations (µg/mg dry mass) back-transformed from the multivariate log-ratio scale via exponentiation. Points denote posterior medians, and error bars represent the 95% highest posterior density (HPD) intervals.

https://doi.org/10.1371/journal.pone.0352693.g001

Download:

Fig 2. Estimated marginal concentration profiles across tissues and genotypes.

Facets track model-predicted elemental ratios (ALR-transformed). Points denote posterior medians, and error bars represent the 95% HPD intervals.

https://doi.org/10.1371/journal.pone.0352693.g002

Tissue in general explained the most variance in elemental ratios (especially in Mg and Mn), but genotype and tissue-genotype interaction also influenced some elements (Fig 3). Group-level (i.e., random) individual mouse-level coefficients explained very little observed variance in any element, but some elements displayed substantial isolated and correlated residual error (Fig 3; Table 2a and 2b).

Download:

Table 2. A) Global variance decomposition showing percentage contributions of experimental factors to total ionomic variance observed in original data. Fixed effects include Genotype, Tissue, and Genotype x Tissue Interaction (Interaction). Random effects include mouse identity (Individual). Residual variance is partitioned into isolated unexplained error and shared pairwise variance allocations calculated using Squared Multiple Correlation (SMC). B) Percentage contribution of residual correlations between individual pairs of elements.

https://doi.org/10.1371/journal.pone.0352693.t002

Download:

Fig 3. Global variance decomposition across elements.

Stacked bars illustrate the relative percentage of total variance allocated to distinct hierarchical model components for each element. To account for the mathematical closure properties of compositional data, residual variance is partitioned into isolated error (grey) and residual correlations (yellow) using a Squared Multiple Correlation (SMC) projection of the residual matrix. All profiles were normalized to a strict 100% variance closure pool per element.

https://doi.org/10.1371/journal.pone.0352693.g003

Contrasts

Contrasts were performed to detect posterior differences between population-level (i.e., fixed) groupings found genotype-level effects within individual tissues for three elements: K, Mg, and Na (Fig 4). The posterior log-ratios of Mg and Na in feces were respectively 0.5 and 0.3 units lower in KO than WT mice (Fig 4). The posterior log-ratio of K in fur was 0.5 units lower in KO than WT mice (Fig 4).

Download:

Fig 4. Pairwise contrasts of genotype effects on tissue-level ionomes.

Points represent the median of the posterior difference distribution on the native additive log-ratio (ALR) scale. Horizontal bars represent the 95% HPD intervals for the pairwise differences between wild-type (WT) and knockout (KO) genotypes across tissue types.

https://doi.org/10.1371/journal.pone.0352693.g004

All elements demonstrated at least one significant tissue-level contrast, but a substantial number of elements demonstrated multiple differences in log-ratio at the tissue level (Figs 1, 2 and 5). However, only K displayed significant effects of the interaction between tissue and genotype (Figs 1, 2 and 6). Further, pairwise contrast within individual mice of elemental intercepts failed to detect significant within-individual correlations between elements (Figure S3 in S3 File).

Download:

Fig 5. Global tissue fixed-effect coefficients on the log-ratio scale.

Forest plot isolating the structural baseline shifts caused by anatomical tissue differentiation, independent of genotype. Coefficients represent the unique log-ratio shift relative to the designated reference tissue. Points denote posterior medians and horizontal bars indicate 95% HPD.

https://doi.org/10.1371/journal.pone.0352693.g005

Download:

Fig 6. Plot displaying the fixed-effect interaction terms (Tissue x Genotype) from the global joint model.

Points denote posterior medians and horizontal bars represent the 95% HPD.

https://doi.org/10.1371/journal.pone.0352693.g006

Residuals

Predicted isolated error (from inverted covariance matrix) was generally low (< 10%) but reached a maximum in K and a minimum in Mn (Fig 3). However, residual correlation explained the most variance of any source in S and Zn (Fig 3). These two elements displayed the strongest residual correlation with each other, but other pairs of elements also displayed strong residual correlation, including Ca – P, Ca – S, Ca – Zn, P – S, and P – Zn (Fig 7).

Download:

Fig 7. Faceted forest plots displaying the median coefficients and 95% HPD for residual correlation between pairs of elements.

Values for each pairwise comparison are available in Table 1.

https://doi.org/10.1371/journal.pone.0352693.g007

Post-hoc binning of posterior residual error by extracting correlation coefficients from MCMC iterations revealed differences in residual correlation among tissues (Figure S4 in S4 File). For example, residual correlation in Fe-P and Fe-K only occurred in fur. Alternatively, the residual Fe-Na correlation was higher and more certain in fur than that in olfactory bulb or PMHTH (Figure S4 in S4 File).

Post-hoc assessment of differences in elemental correlation also revealed differences among genotype (Figure S5 in S5 File). Correlations between Mg – Na, Fe – Mn, and Fe – K were moderately positive (median 0.3–0.4) for WT, but 95% HPD intervals of KO mice overlapped with 0 (Figure S5 in S5 File).

Discussion

The results of this work provided limited support for our hypothesis that FXS affects the compositional ionome of ten elements (Ca, Cu, Fe, K, Mg, Mn, Na, S, P, and Zn) in neural and somatic tissues. In this study, multivariate analyses demonstrated that elemental composition patterns were driven mainly by tissue type rather than genotype, with by far the most variation attributable to tissue origin. However, multiple elements differed between WT and KO animals in feces (Mg, Na) and fur (K), as well as differences in residual correlation of elements among tissues and genotype. Notably, we found that the striatum had substantially higher residual dispersion than other tissues, indicating ~4 times more elemental variance in this tissue. Further, high residual correlation and heavily weighted residual error in Ca, P, S, and Zn suggest that unaccounted factors drive a substantial proportion of variance in these elements, in some cases exceeding variance among tissues. Overall, these differences support previous observations that Fmr1-KO influences gut and broader epidermal function through electrolyte (e.g., Na, K, Mg) dysregulation [20,21,35,36].

We found several differences in elemental ratios between individual tissues of Fmr1-KO and WT mice. For example, we detected an interactive effect of tissue: genotype on potassium contribution, but mean potassium contribution differed between Fmr1-KO and WT only in fur. Interestingly, feces from WT mice contained relatively higher Mg and Na levels compared with Fmr1-KO mice, suggesting potential differences in elemental handling or excretion between genotypes. Because Mg and Na are important regulators of numerous cellular and neuronal processes, altered homeostasis of these elements may contribute to broader physiological dysfunction associated with Fmr1 deficiency.

Previous work further showed children with ASD self-select diets higher in Mg and lower in protein and Ca than children without ASD, and that selective diets were more likely to result in substantial deficiency of at least one nutrient [37]. Interestingly, evidence from children with ASD suggested Mg supplementation might alleviate neuro-behavioral symptoms [26]. Although this remains to be rigorously tested, previous authors advocate for the relevance of understanding Mg action in FMR1 pathways [38].

Limitations and future directions

In conclusion, we found differences in ionomic composition between tissues in all 10 individual elements; although, we found genotype-level differences in overall composition and individual elements only in some tissues. Yet, several limitations must be acknowledged. First, our study did not use mechanistic assays to link elemental changes directly to FMRP function or downstream molecular pathways. Additionally, other elements contribute to tissue-level ionomes and are required for cellular processes and structures (e.g., carbon, nitrogen, silicon, etc.) which we did not measure here. As a result, the total percentage of original sample mass measured was in most cases quite small. Furthermore, although experimental groups were age-matched as closely as possible, age-related physiological and metabolic changes may also influence tissue elemental composition. Future studies with larger cohorts specifically designed to evaluate age-dependent variability in tissue ionomic profiles would therefore be valuable. Additionally we only used male mice in our study, previous work has shown that there are sex differences physiologically [39,40] and behaviorally [41] in FXS mice suggesting that sex may also be an important factor for future work. Although isoflurane exposure was applied uniformly across experimental groups, previous studies have reported that inhalation of anesthetics can affect iron homeostasis, ferroptosis-related pathways, intracellular metabolite diffusion, and mineral-regulatory processes [42,43]. Together, these findings suggest that isoflurane may affect elemental and metabolic homeostasis and therefore represents a potential confounding factor when interpreting tissue mineral composition data, particularly iron-related measurements and is an important difference between our work and previous published literature [28]. Additionally, in this study, all tissues were collected without any vascular perfusion, and serum elemental composition was not measured; therefore, tissue ionomic profiles may partially reflect circulation-associated elemental content in addition to tissue-specific organization.

While previous work has also measured elemental differences in FXS, direct quantitative comparison between the present study and others should be interpreted cautiously because the studies differed substantially in tissue selection, analytical methodology, and normalization strategy. Talvio et al. quantified absolute elemental concentrations (µg/g wet weight) in cortex, cerebellum, heart, liver, and spleen using ICP-MS, whereas our study analyzed dried tissue samples using ICP-OES, which is less sensitive, within a compositional framework focused on relative multielement organization across olfactory bulb, striatum, PMHTH, fur, feces, and cecal contents [28]. Additionally, the studies differed in tissue preparation and experimental workflow, including anesthetic exposure, tissue pooling strategy, drying procedures, digestion workflows, and elemental normalization approaches. These methodological differences are expected to influence absolute elemental concentrations, particularly for redox-sensitive elements such as iron. Importantly, Talvio et al. similarly emphasized cautious interpretation due to limited statistical power and multiple comparisons [28]. Our study was designed to evaluate whether loss of Fmr1 alters relative multielement organization across neural and somatic tissues. In this compositional framework, interpretation primarily depends on relative elemental relationships rather than direct equivalence of absolute concentrations.

Nevertheless, analyzing multi-tissue, multi-elemental data allowed us to identify systematic and genotype-dependent ionomic shifts in the Fmr1 knockout male mouse model. The combined use of additive log-ratio (ALR) transformation and compositional data analysis provided a robust framework for handling the constrained and interdependent nature of multi-elemental data [44]. Genome-wide RNAi ionomics screens in human cells have further demonstrated that multielemental profiles can reveal previously unrecognized genes and regulatory networks governing trace element metabolism [45], in the context of this study, ionomic profiling of non-invasive tissues such as feces or fur/hair could be linked to behavioral deficits enabling biomarker discovery.

Even though, in this study we measured overall relative concentration of an element in different neural as well as somatic tissues, future studies should assess not only total elemental abundance but also labile mineral content. The reason behind this logic is total tissue mineral levels do not unveil whether an element is biologically in active form, safely stored, protein-bound, redox-active, or localized within specific cellular compartments in tissues. This especially carries crucial insights for iron, because iron toxicity depends strongly on its biochemical form and intracellular localization. For example- ferritin-bound or transferrin-bound iron is relatively protected, whereas labile Fe² ⁺ can promote Fenton chemistry, oxidative stress, lipid peroxidation, and ferroptosis [46]. Thus, two tissues may show similar total iron concentrations, yet differ substantially in toxicity or physiological impact if one contains a larger labile/redox-active iron pool. Therefore, measuring only total tissue iron could not fully capture biologically relevant alterations in iron homeostasis in our study.

For the other elements measured in this study — their labile pools may provide more mechanistic insight than total elemental concentration alone. For example, labile Ca² ⁺ regulates ion transport, secretion, epithelial signaling, and cellular stress responses [47]; labile Zn² ⁺ functions as a signaling ion [48]; and labile Cu and Mn may influence redox biology, mitochondrial function, and oxidative stress [49].

Future studies should investigate labile metal pools within the gut lumen and microbial environment, because total elemental abundance in luminal contents may not accurately reflect the fraction of minerals that is biologically available to microbes or host epithelial cells. Since the concept of “labile metals” refers to exchangeable, bioavailable metal ions that actively participate in microbial metabolism, redox reactions, signaling, and host–microbe interactions rather than metals that are tightly bound or inertly stored. Previous studies also emphasize that bacteria dynamically regulate labile metal pools to maintain metabolic homeostasis, oxidative stress resistance, and nutrient acquisition [50].

This is particularly relevant to our study because we quantified elemental concentrations in intestinal luminal contents, where metals are simultaneously influenced by diet, microbial utilization, epithelial transport, mucus interactions, and host inflammatory responses [51,52]. Therefore, future studies should examine not only total elemental levels but also the biologically active and compartment-specific metal pools, as these may better explain how altered mineral homeostasis contributes to gut microbiome changes and other neural as well as peripheral phenotypes in FXS. Future work is needed to combine bulk elemental profiling with approaches that measure bioavailable and compartment-specific mineral pools. This would help determine whether altered mineral composition in FXS reflects changes in total abundance, altered intracellular distribution, or shifts in the biologically reactive mineral pools most likely to affect mitochondrial metabolism, oxidative stress and cellular toxicity.

While challenging, future work integrating ionomic patterns with other biochemical systems (e.g., transcriptome, proteome) would help identify mechanistic pathways underlying loss of FMRP and disrupted elemental homeostasis. Validation of tissue-specific elemental signatures (e.g., Mg and Na in feces; K in fur) may further enable biomarker discovery and the development of ion-targeted therapeutic strategies for Fragile X Syndrome particularly if supplementation or severity could be linked with ionomic variation.

Supporting information

S1 File. Figure S1. Multivariate posterior predictive checks (PPC) for core elemental indicators.

Density overlays validating global model fit for representative elements (Mg, Na, and K). The solid black line y defines the empirical distribution of the observed log-ratio data; thin turquoise lines y_rep represent 50 independent simulated datasets generated from the posterior predictive distribution. High alignment between empirical and simulated densities verifies structural model adequacy and distributional assumptions.

https://doi.org/10.1371/journal.pone.0352693.s001

(DOCX)

S2 File. Figure S2. Markov Chain Monte Carlo (MCMC) sampling diagnostics.

Standalone posterior histograms split by individual sampling chains for intercept parameters (Mg, Na, and K). Overlapping unimodal distributions across all four computational chains visually confirm robust parameter space exploration, chain mixing, and global convergence.

https://doi.org/10.1371/journal.pone.0352693.s002

(DOCX)

S3 File. Figure S3. Forest plot displaying the median residual correlation coefficients and corresponding 95% HPD intervals for baseline elemental co-accumulation patterns among individual mice.

Points represent median and lines represent 95% HPD.

https://doi.org/10.1371/journal.pone.0352693.s003

(DOCX)

S4 File. Figure S4. Forest plot displaying the tissue-specific median residual correlation coefficients and corresponding 95% HPD intervals.

Points represent median and lines represent 95% HPD.

https://doi.org/10.1371/journal.pone.0352693.s004

(DOCX)

S5 File. Figure S5. Genotype-specific forest plot displaying the posterior median residual correlations and corresponding 95% HPD intervals for all elemental pairs, isolated by genotype.

Points represent median and lines represent 95% HPD.

https://doi.org/10.1371/journal.pone.0352693.s005

(DOCX)

S6 File. Table S1. Global multivariate Bayesian model parameter coordinates and chain diagnostics.

Posterior summaries (median estimates, standard errors, and 95% Bayesian Credible Intervals) for all fixed-effect parameters, residual standard deviations (sigma), and residual correlations (rescor) across the 10-element ionomic suite. Chain convergence and sampling efficiency are verified via the Gelman-Rubin diagnostic (Rhat) alongside bulk and tail Effective Sample Sizes (Bulk_ESS, Tail_ESS).

https://doi.org/10.1371/journal.pone.0352693.s006

(DOCX)

S7 File. Table S2. Verification matrix of model predictions.

Columns represent predicted median value, lower, and upper quartile, as well as observed median for each tissue, genotype, and element combination.

https://doi.org/10.1371/journal.pone.0352693.s007

(DOCX)

S8 File. Table S3. Post-hoc assessment of multivariate residual dispersion and homoscedasticity.

Values represent the Interquartile Range (IQR) and Standard Deviation (SD) of the model-derived residual error on the Additive Log-Ratio (ALR) scale, isolated by genotype.

https://doi.org/10.1371/journal.pone.0352693.s008

(DOCX)

S9 File. Table S4. Post-hoc assessment of multivariate residual dispersion and homoscedasticity.

Values represent the Interquartile Range (IQR) and Standard Deviation (SD) of the model-derived residual error on the Additive Log-Ratio (ALR) scale, isolated by tissue type.

https://doi.org/10.1371/journal.pone.0352693.s009

(DOCX)

References

1. Hagerman RJ, Berry-Kravis E, Hazlett HC, Bailey DB Jr, Moine H, Kooy RF, et al. Fragile X syndrome. Nat Rev Dis Primers. 2017;3:17065. pmid:28960184
- View Article
- PubMed/NCBI
- Google Scholar
2. Salcedo-Arellano MJ, Hagerman RJ, Martínez-Cerdeño V. Fragile X syndrome: clinical presentation, pathology and treatment. Gac Med Mex. 2020;156(1):60–6. pmid:32026885
- View Article
- PubMed/NCBI
- Google Scholar
3. Bagni C, Tassone F, Neri G, Hagerman R. Fragile X syndrome: causes, diagnosis, mechanisms, and therapeutics. J Clin Invest. 2012;122(12):4314–22. pmid:23202739
- View Article
- PubMed/NCBI
- Google Scholar
4. Mila M, Alvarez-Mora MI, Madrigal I, Rodriguez-Revenga L. Fragile X syndrome: an overview and update of the FMR1 gene. Clin Genet. 2018;93(2):197–205. pmid:28617938
- View Article
- PubMed/NCBI
- Google Scholar
5. Williams RJP, Fraústo Da Silva JJR. Evolution was chemically constrained. J Theor Biol. 2003;220(3):323–43. pmid:12468283
- View Article
- PubMed/NCBI
- Google Scholar
6. Baxter I. Ionomics: the functional genomics of elements. Brief Funct Genom. 2010;9(2):149–56. pmid:20081216
- View Article
- PubMed/NCBI
- Google Scholar
7. Salt DE, Baxter I, Lahner B. Ionomics and the study of the plant ionome. Annu Rev Plant Biol. 2008;59:709–33. pmid:18251712
- View Article
- PubMed/NCBI
- Google Scholar
8. Huang X-Y, Salt DE. Plant ionomics: from elemental profiling to environmental adaptation. Mol Plant. 2016;9(6):787–97. pmid:27212388
- View Article
- PubMed/NCBI
- Google Scholar
9. Cabral M, Kuxhaus O, Eichelmann F, Kopp JF, Alker W, Hackler J, et al. Trace element profile and incidence of type 2 diabetes, cardiovascular disease and colorectal cancer: results from the EPIC-Potsdam cohort study. Eur J Nutr. 2021;60(6):3267–78. pmid:33590281
- View Article
- PubMed/NCBI
- Google Scholar
10. Dubey P, Thakur V, Chattopadhyay M. Role of minerals and trace elements in diabetes and insulin resistance. Nutrients. 2020;12(6):1864. pmid:32585827
- View Article
- PubMed/NCBI
- Google Scholar
11. Sarafanov AG, Todorov TI, Centeno JA, Macias V, Gao W, Liang W-M, et al. Prostate cancer outcome and tissue levels of metal ions. Prostate. 2011;71(11):1231–8. pmid:21271612
- View Article
- PubMed/NCBI
- Google Scholar
12. Zhang Y, Xu Y, Zheng L. Disease ionomics: understanding the role of ions in complex disease. Int J Mol Sci. 2020;21(22):8646. pmid:33212764
- View Article
- PubMed/NCBI
- Google Scholar
13. Zhang Y, Huang B, Jin J, Xiao Y, Ying H. Recent advances in the application of ionomics in metabolic diseases. Front Nutr. 2023;9:1111933. pmid:36726817
- View Article
- PubMed/NCBI
- Google Scholar
14. Ding Q, Sethna F, Wu X-T, Miao Z, Chen P, Zhang Y, et al. Transcriptome signature analysis repurposes trifluoperazine for the treatment of fragile X syndrome in mouse model. Commun Biol. 2020;3(1):127. pmid:32179850
- View Article
- PubMed/NCBI
- Google Scholar
15. Ebrahimiazar S, Kikkawa T, Minakuchi Y, Miyashita S, Manabe S, Hoshino M, et al. A transcriptomic dataset of embryonic murine telencephalon of fmr1-deficient mice. Sci Data. 2025;12(1):927. pmid:40456813
- View Article
- PubMed/NCBI
- Google Scholar
16. Abbasi DA, Berry-Kravis E, Zhao X, Cologna SM. Proteomics insights into fragile X syndrome: unraveling molecular mechanisms and therapeutic avenues. Neurobiol Dis. 2024;194:106486. pmid:38548140
- View Article
- PubMed/NCBI
- Google Scholar
17. Gao M-M, Shi H, Yan H-J, Long Y-S. Proteome profiling of the prefrontal cortex of Fmr1 knockout mouse reveals enhancement of complement and coagulation cascades. J Proteom. 2023;274:104822. pmid:36646274
- View Article
- PubMed/NCBI
- Google Scholar
18. Darnell JC, Klann E. The translation of translational control by FMRP: therapeutic targets for FXS. Nat Neurosci. 2013;16(11):1530–6. pmid:23584741
- View Article
- PubMed/NCBI
- Google Scholar
19. Ferron L. Fragile X mental retardation protein controls ion channel expression and activity. J Physiol. 2016;594(20):5861–7. pmid:26864773
- View Article
- PubMed/NCBI
- Google Scholar
20. Cheng P, Qiu Z, Du Y. Potassium channels and autism spectrum disorder: an overview. Int J Dev Neurosci. 2021;81(6):479–91. pmid:34008235
- View Article
- PubMed/NCBI
- Google Scholar
21. Deng P-Y, Klyachko VA. Channelopathies in fragile X syndrome. Nat Rev Neurosci. 2021;22(5):275–89. pmid:33828309
- View Article
- PubMed/NCBI
- Google Scholar
22. Napoli E, Ross-Inta C, Wong S, Omanska-Klusek A, Barrow C, Iwahashi C, et al. Altered zinc transport disrupts mitochondrial protein processing/import in fragile X-associated tremor/ataxia syndrome. Hum Mol Genet. 2011;20(15):3079–92. pmid:21558427
- View Article
- PubMed/NCBI
- Google Scholar
23. Kaufmann WE, Kidd SA, Andrews HF, Budimirovic DB, Esler A, Haas-Givler B, et al. Autism spectrum disorder in fragile X syndrome: cooccurring conditions and current treatment. Pediatrics. 2017;139(Suppl 3):S194–206. pmid:28814540
- View Article
- PubMed/NCBI
- Google Scholar
24. Hagmeyer S, Sauer AK, Grabrucker AM. Prospects of zinc supplementation in autism spectrum disorders and Shankopathies such as Phelan McDermid Syndrome. Front Synaptic Neurosci. 2018;10:11. pmid:29875651
- View Article
- PubMed/NCBI
- Google Scholar
25. Khan F, Rahman MS, Akhter S, Momen ABI, Raihan SG. Vitamin B6 and magnesium on neurobehavioral status of autism spectrum disorder: a randomized, double-blind, placebo controlled study. Bangladesh J Med. 2021;32(1):12–8.
- View Article
- Google Scholar
26. Mousain-Bosc M. Magnesium, hyperactivity and autism in children. 2006;19:53–62.
27. Lin P, Zhang Q, Sun J, Li Q, Li D, Zhu M, et al. A comparison between children and adolescents with autism spectrum disorders and healthy controls in biomedical factors, trace elements, and microbiota biomarkers: a meta-analysis. Front Psychiatry. 2024;14:1318637. pmid:38283894
- View Article
- PubMed/NCBI
- Google Scholar
28. Talvio K, Kanninen KM, White AR, Koistinaho J, Castrén ML. Increased iron content in the heart of the Fmr1 knockout mouse. Biometals. 2021;34(4):947–54. pmid:34089433
- View Article
- PubMed/NCBI
- Google Scholar
29. Vela G, Stark P, Socha M, Sauer AK, Hagmeyer S, Grabrucker AM. Zinc in gut-brain interaction in autism and neurological disorders. Neural Plast. 2015;2015:972791. pmid:25878905
- View Article
- PubMed/NCBI
- Google Scholar
30. Grabrucker A. Front matter. In: Biometals in autism spectrum disorders. Elsevier; 2020. iii p.
- View Article
- Google Scholar
31. Austin C, Curtin P, Arora M, Reichenberg A, Curtin A, Iwai-Shimada M, et al. Elemental dynamics in hair accurately predict future autism spectrum disorder diagnosis: an international multi-center study. J Clin Med. 2022;11(23):7154. pmid:36498727
- View Article
- PubMed/NCBI
- Google Scholar
32. Bakker CE, Verheij C, Willemsen R, Helm R van der, Oerlemans F, The Dutch-Belgian Fragile X Consortium. Fmr1 knockout mice: a model to study fragile X mental retardation. Cell. 1994;78:23–33.
- View Article
- Google Scholar
33. Templ M, Hron K, Filzmoser P. robCompositions: an R‐package for robust statistical analysis of compositional data. In: Pawlowsky‐Glahn V, Buccianti A, editors. Compositional data analysis. 1st ed. Wiley; 2011. p. 341–55.
- View Article
- Google Scholar
34. Reimann C, Filzmoser P, Garrett RG, Dutter R. Statistical data analysis explained: applied environmental statistics with R. 1st ed. Wiley; 2008.
- View Article
- Google Scholar
35. D’Antoni S, Spatuzza M, Bonaccorso CM, Catania MV. Role of fragile X messenger ribonucleoprotein 1 in the pathophysiology of brain disorders: a glia perspective. Neurosci Biobehav Rev. 2024;162:105731. pmid:38763180
- View Article
- PubMed/NCBI
- Google Scholar
36. Napoli E, Ross-Inta C, Song G, Wong S, Hagerman R, Gane LW. Premutation in the Fragile X Mental Retardation 1 (FMR1) gene affects maternal Zn-milk and perinatal brain bioenergetics and scaffolding. Front Neurosci. 2016;10.
- View Article
- Google Scholar
37. Zimmer MH, Hart LC, Manning-Courtney P, Murray DS, Bing NM, Summer S. Food variety as a predictor of nutritional status among children with autism. J Autism Dev Disord. 2012;42(4):549–56. pmid:21556968
- View Article
- PubMed/NCBI
- Google Scholar
38. Mousain-Bosc M. Magnesium in the central nervous system. Adelaide, South Australia, Australia: University of Adelaide Press; 2011. p. 283–302.
39. Chawla A, McCullagh EA. Auditory brain stem responses in the C57BL/6J fragile X syndrome-knockout mouse model. Front Integr Neurosci. 2022;15:803483. pmid:35111002
- View Article
- PubMed/NCBI
- Google Scholar
40. Smith EG, Pedapati EV, Liu R, Schmitt LM, Dominick KC, Shaffer RC, et al. Sex differences in resting EEG power in Fragile X syndrome. J Psychiatr Res. 2021;138:89–95. pmid:33836434
- View Article
- PubMed/NCBI
- Google Scholar
41. Petroni V, Subashi E, Premoli M, Wöhr M, Crusio WE, Lemaire V, et al. Autistic-like behavioral effects of prenatal stress in juvenile Fmr1 mice: the relevance of sex differences and gene-environment interactions. Sci Rep. 2022;12(1):7269. pmid:35508566
- View Article
- PubMed/NCBI
- Google Scholar
42. Miao M, Han Y, Wang Y, Wang J, Zhu R, Yang Y, et al. Dysregulation of iron homeostasis and ferroptosis in sevoflurane and isoflurane associated perioperative neurocognitive disorders. CNS Neurosci Ther. 2024;30(2):e14553. pmid:38334231
- View Article
- PubMed/NCBI
- Google Scholar
43. Wang J, Sun J, Qiao S, Li H, Che T, Wang C, et al. Effects of isoflurane on complex II‑associated mitochondrial respiration and reactive oxygen species production: roles of nitric oxide and mitochondrial KATP channels. Mol Med Rep. 2019;20(5):4383–90. pmid:31545457
- View Article
- PubMed/NCBI
- Google Scholar
44. Greenacre M. Compositional data analysis. Annu Rev Stat Its Appl. 2021;8:271–99.
- View Article
- Google Scholar
45. Malinouski M, Hasan NM, Zhang Y, Seravalli J, Lin J, Avanesov A, et al. Genome-wide RNAi ionomics screen reveals new genes and regulation of human trace element metabolism. Nat Commun. 2014;5:3301. pmid:24522796
- View Article
- PubMed/NCBI
- Google Scholar
46. Kakhlon O, Cabantchik ZI. The labile iron pool: characterization, measurement, and participation in cellular processes1 1This article is part of a series of reviews on “Iron and Cellular Redox Status.” The full list of papers may be found on the homepage of the journal. Free Radic Biol Med. 2002;33(8):1037–46.
- View Article
- Google Scholar
47. Berridge MJ, Bootman MD, Roderick HL. Calcium signalling: dynamics, homeostasis and remodelling. Nat Rev Mol Cell Biol. 2003;4(7):517–29. pmid:12838335
- View Article
- PubMed/NCBI
- Google Scholar
48. De Leon-Rodriguez L, Lubag AJM, Sherry AD. Imaging free zinc levels in vivo - what can be learned? Inorg Chim Acta. 2012;393:12–23.
- View Article
- Google Scholar
49. Lindahl PA, Moore MJ. Labile low-molecular-mass metal complexes in mitochondria: trials and tribulations of a burgeoning field. Biochemistry. 2016;55(30):4140–53. pmid:27433847
- View Article
- PubMed/NCBI
- Google Scholar
50. Helmann JD. Metals in motion: understanding labile metal pools in bacteria. Biochemistry. 2025;64(2):329–45. pmid:39755956
- View Article
- PubMed/NCBI
- Google Scholar
51. Hood MI, Skaar EP. Nutritional immunity: transition metals at the pathogen-host interface. Nat Rev Microbiol. 2012;10(8):525–37. pmid:22796883
- View Article
- PubMed/NCBI
- Google Scholar
52. Lopez CA, Skaar EP. The impact of dietary transition metals on host-bacterial interactions. Cell Host Microbe. 2018;23(6):737–48. pmid:29902439
- View Article
- PubMed/NCBI
- Google Scholar

[ref1] 1. Hagerman RJ, Berry-Kravis E, Hazlett HC, Bailey DB Jr, Moine H, Kooy RF, et al. Fragile X syndrome. Nat Rev Dis Primers. 2017;3:17065. pmid:28960184
View Article
PubMed/NCBI
Google Scholar

[2] View Article

[3] PubMed/NCBI

[4] Google Scholar

[ref2] 2. Salcedo-Arellano MJ, Hagerman RJ, Martínez-Cerdeño V. Fragile X syndrome: clinical presentation, pathology and treatment. Gac Med Mex. 2020;156(1):60–6. pmid:32026885
View Article
PubMed/NCBI
Google Scholar

[6] View Article

[7] PubMed/NCBI

[8] Google Scholar

[ref3] 3. Bagni C, Tassone F, Neri G, Hagerman R. Fragile X syndrome: causes, diagnosis, mechanisms, and therapeutics. J Clin Invest. 2012;122(12):4314–22. pmid:23202739
View Article
PubMed/NCBI
Google Scholar

[10] View Article

[11] PubMed/NCBI

[12] Google Scholar

[ref4] 4. Mila M, Alvarez-Mora MI, Madrigal I, Rodriguez-Revenga L. Fragile X syndrome: an overview and update of the FMR1 gene. Clin Genet. 2018;93(2):197–205. pmid:28617938
View Article
PubMed/NCBI
Google Scholar

[14] View Article

[15] PubMed/NCBI

[16] Google Scholar

[ref5] 5. Williams RJP, Fraústo Da Silva JJR. Evolution was chemically constrained. J Theor Biol. 2003;220(3):323–43. pmid:12468283
View Article
PubMed/NCBI
Google Scholar

[18] View Article

[19] PubMed/NCBI

[20] Google Scholar

[ref6] 6. Baxter I. Ionomics: the functional genomics of elements. Brief Funct Genom. 2010;9(2):149–56. pmid:20081216
View Article
PubMed/NCBI
Google Scholar

[22] View Article

[23] PubMed/NCBI

[24] Google Scholar

[ref7] 7. Salt DE, Baxter I, Lahner B. Ionomics and the study of the plant ionome. Annu Rev Plant Biol. 2008;59:709–33. pmid:18251712
View Article
PubMed/NCBI
Google Scholar

[26] View Article

[27] PubMed/NCBI

[28] Google Scholar

[ref8] 8. Huang X-Y, Salt DE. Plant ionomics: from elemental profiling to environmental adaptation. Mol Plant. 2016;9(6):787–97. pmid:27212388
View Article
PubMed/NCBI
Google Scholar

[30] View Article

[31] PubMed/NCBI

[32] Google Scholar

[ref9] 9. Cabral M, Kuxhaus O, Eichelmann F, Kopp JF, Alker W, Hackler J, et al. Trace element profile and incidence of type 2 diabetes, cardiovascular disease and colorectal cancer: results from the EPIC-Potsdam cohort study. Eur J Nutr. 2021;60(6):3267–78. pmid:33590281
View Article
PubMed/NCBI
Google Scholar

[34] View Article

[35] PubMed/NCBI

[36] Google Scholar

[ref10] 10. Dubey P, Thakur V, Chattopadhyay M. Role of minerals and trace elements in diabetes and insulin resistance. Nutrients. 2020;12(6):1864. pmid:32585827
View Article
PubMed/NCBI
Google Scholar

[38] View Article

[39] PubMed/NCBI

[40] Google Scholar

[ref11] 11. Sarafanov AG, Todorov TI, Centeno JA, Macias V, Gao W, Liang W-M, et al. Prostate cancer outcome and tissue levels of metal ions. Prostate. 2011;71(11):1231–8. pmid:21271612
View Article
PubMed/NCBI
Google Scholar

[42] View Article

[43] PubMed/NCBI

[44] Google Scholar

[ref12] 12. Zhang Y, Xu Y, Zheng L. Disease ionomics: understanding the role of ions in complex disease. Int J Mol Sci. 2020;21(22):8646. pmid:33212764
View Article
PubMed/NCBI
Google Scholar

[46] View Article

[47] PubMed/NCBI

[48] Google Scholar

[ref13] 13. Zhang Y, Huang B, Jin J, Xiao Y, Ying H. Recent advances in the application of ionomics in metabolic diseases. Front Nutr. 2023;9:1111933. pmid:36726817
View Article
PubMed/NCBI
Google Scholar

[50] View Article

[51] PubMed/NCBI

[52] Google Scholar

[ref14] 14. Ding Q, Sethna F, Wu X-T, Miao Z, Chen P, Zhang Y, et al. Transcriptome signature analysis repurposes trifluoperazine for the treatment of fragile X syndrome in mouse model. Commun Biol. 2020;3(1):127. pmid:32179850
View Article
PubMed/NCBI
Google Scholar

[54] View Article

[55] PubMed/NCBI

[56] Google Scholar

[ref15] 15. Ebrahimiazar S, Kikkawa T, Minakuchi Y, Miyashita S, Manabe S, Hoshino M, et al. A transcriptomic dataset of embryonic murine telencephalon of fmr1-deficient mice. Sci Data. 2025;12(1):927. pmid:40456813
View Article
PubMed/NCBI
Google Scholar

[58] View Article

[59] PubMed/NCBI

[60] Google Scholar

[ref16] 16. Abbasi DA, Berry-Kravis E, Zhao X, Cologna SM. Proteomics insights into fragile X syndrome: unraveling molecular mechanisms and therapeutic avenues. Neurobiol Dis. 2024;194:106486. pmid:38548140
View Article
PubMed/NCBI
Google Scholar

[62] View Article

[63] PubMed/NCBI

[64] Google Scholar

[ref17] 17. Gao M-M, Shi H, Yan H-J, Long Y-S. Proteome profiling of the prefrontal cortex of Fmr1 knockout mouse reveals enhancement of complement and coagulation cascades. J Proteom. 2023;274:104822. pmid:36646274
View Article
PubMed/NCBI
Google Scholar

[66] View Article

[67] PubMed/NCBI

[68] Google Scholar

[ref18] 18. Darnell JC, Klann E. The translation of translational control by FMRP: therapeutic targets for FXS. Nat Neurosci. 2013;16(11):1530–6. pmid:23584741
View Article
PubMed/NCBI
Google Scholar

[70] View Article

[71] PubMed/NCBI

[72] Google Scholar

[ref19] 19. Ferron L. Fragile X mental retardation protein controls ion channel expression and activity. J Physiol. 2016;594(20):5861–7. pmid:26864773
View Article
PubMed/NCBI
Google Scholar

[74] View Article

[75] PubMed/NCBI

[76] Google Scholar

[ref20] 20. Cheng P, Qiu Z, Du Y. Potassium channels and autism spectrum disorder: an overview. Int J Dev Neurosci. 2021;81(6):479–91. pmid:34008235
View Article
PubMed/NCBI
Google Scholar

[78] View Article

[79] PubMed/NCBI

[80] Google Scholar

[ref21] 21. Deng P-Y, Klyachko VA. Channelopathies in fragile X syndrome. Nat Rev Neurosci. 2021;22(5):275–89. pmid:33828309
View Article
PubMed/NCBI
Google Scholar

[82] View Article

[83] PubMed/NCBI

[84] Google Scholar

[ref22] 22. Napoli E, Ross-Inta C, Wong S, Omanska-Klusek A, Barrow C, Iwahashi C, et al. Altered zinc transport disrupts mitochondrial protein processing/import in fragile X-associated tremor/ataxia syndrome. Hum Mol Genet. 2011;20(15):3079–92. pmid:21558427
View Article
PubMed/NCBI
Google Scholar

[86] View Article

[87] PubMed/NCBI

[88] Google Scholar

[ref23] 23. Kaufmann WE, Kidd SA, Andrews HF, Budimirovic DB, Esler A, Haas-Givler B, et al. Autism spectrum disorder in fragile X syndrome: cooccurring conditions and current treatment. Pediatrics. 2017;139(Suppl 3):S194–206. pmid:28814540
View Article
PubMed/NCBI
Google Scholar

[90] View Article

[91] PubMed/NCBI

[92] Google Scholar

[ref24] 24. Hagmeyer S, Sauer AK, Grabrucker AM. Prospects of zinc supplementation in autism spectrum disorders and Shankopathies such as Phelan McDermid Syndrome. Front Synaptic Neurosci. 2018;10:11. pmid:29875651
View Article
PubMed/NCBI
Google Scholar

[94] View Article

[95] PubMed/NCBI

[96] Google Scholar

[ref25] 25. Khan F, Rahman MS, Akhter S, Momen ABI, Raihan SG. Vitamin B6 and magnesium on neurobehavioral status of autism spectrum disorder: a randomized, double-blind, placebo controlled study. Bangladesh J Med. 2021;32(1):12–8.
View Article
Google Scholar

[98] View Article

[99] Google Scholar

[ref26] 26. Mousain-Bosc M. Magnesium, hyperactivity and autism in children. 2006;19:53–62.

[ref27] 27. Lin P, Zhang Q, Sun J, Li Q, Li D, Zhu M, et al. A comparison between children and adolescents with autism spectrum disorders and healthy controls in biomedical factors, trace elements, and microbiota biomarkers: a meta-analysis. Front Psychiatry. 2024;14:1318637. pmid:38283894
View Article
PubMed/NCBI
Google Scholar

[102] View Article

[103] PubMed/NCBI

[104] Google Scholar

[ref28] 28. Talvio K, Kanninen KM, White AR, Koistinaho J, Castrén ML. Increased iron content in the heart of the Fmr1 knockout mouse. Biometals. 2021;34(4):947–54. pmid:34089433
View Article
PubMed/NCBI
Google Scholar

[106] View Article

[107] PubMed/NCBI

[108] Google Scholar

[ref29] 29. Vela G, Stark P, Socha M, Sauer AK, Hagmeyer S, Grabrucker AM. Zinc in gut-brain interaction in autism and neurological disorders. Neural Plast. 2015;2015:972791. pmid:25878905
View Article
PubMed/NCBI
Google Scholar

[110] View Article

[111] PubMed/NCBI

[112] Google Scholar

[ref30] 30. Grabrucker A. Front matter. In: Biometals in autism spectrum disorders. Elsevier; 2020. iii p.
View Article
Google Scholar

[114] View Article

[115] Google Scholar

[ref31] 31. Austin C, Curtin P, Arora M, Reichenberg A, Curtin A, Iwai-Shimada M, et al. Elemental dynamics in hair accurately predict future autism spectrum disorder diagnosis: an international multi-center study. J Clin Med. 2022;11(23):7154. pmid:36498727
View Article
PubMed/NCBI
Google Scholar

[117] View Article

[118] PubMed/NCBI

[119] Google Scholar

[ref32] 32. Bakker CE, Verheij C, Willemsen R, Helm R van der, Oerlemans F, The Dutch-Belgian Fragile X Consortium. Fmr1 knockout mice: a model to study fragile X mental retardation. Cell. 1994;78:23–33.
View Article
Google Scholar

[121] View Article

[122] Google Scholar

[ref33] 33. Templ M, Hron K, Filzmoser P. robCompositions: an R‐package for robust statistical analysis of compositional data. In: Pawlowsky‐Glahn V, Buccianti A, editors. Compositional data analysis. 1st ed. Wiley; 2011. p. 341–55.
View Article
Google Scholar

[124] View Article

[125] Google Scholar

[ref34] 34. Reimann C, Filzmoser P, Garrett RG, Dutter R. Statistical data analysis explained: applied environmental statistics with R. 1st ed. Wiley; 2008.
View Article
Google Scholar

[127] View Article

[128] Google Scholar

[ref35] 35. D’Antoni S, Spatuzza M, Bonaccorso CM, Catania MV. Role of fragile X messenger ribonucleoprotein 1 in the pathophysiology of brain disorders: a glia perspective. Neurosci Biobehav Rev. 2024;162:105731. pmid:38763180
View Article
PubMed/NCBI
Google Scholar

[130] View Article

[131] PubMed/NCBI

[132] Google Scholar

[ref36] 36. Napoli E, Ross-Inta C, Song G, Wong S, Hagerman R, Gane LW. Premutation in the Fragile X Mental Retardation 1 (FMR1) gene affects maternal Zn-milk and perinatal brain bioenergetics and scaffolding. Front Neurosci. 2016;10.
View Article
Google Scholar

[134] View Article

[135] Google Scholar

[ref37] 37. Zimmer MH, Hart LC, Manning-Courtney P, Murray DS, Bing NM, Summer S. Food variety as a predictor of nutritional status among children with autism. J Autism Dev Disord. 2012;42(4):549–56. pmid:21556968
View Article
PubMed/NCBI
Google Scholar

[137] View Article

[138] PubMed/NCBI

[139] Google Scholar

[ref38] 38. Mousain-Bosc M. Magnesium in the central nervous system. Adelaide, South Australia, Australia: University of Adelaide Press; 2011. p. 283–302.

[ref39] 39. Chawla A, McCullagh EA. Auditory brain stem responses in the C57BL/6J fragile X syndrome-knockout mouse model. Front Integr Neurosci. 2022;15:803483. pmid:35111002
View Article
PubMed/NCBI
Google Scholar

[142] View Article

[143] PubMed/NCBI

[144] Google Scholar

[ref40] 40. Smith EG, Pedapati EV, Liu R, Schmitt LM, Dominick KC, Shaffer RC, et al. Sex differences in resting EEG power in Fragile X syndrome. J Psychiatr Res. 2021;138:89–95. pmid:33836434
View Article
PubMed/NCBI
Google Scholar

[146] View Article

[147] PubMed/NCBI

[148] Google Scholar

[ref41] 41. Petroni V, Subashi E, Premoli M, Wöhr M, Crusio WE, Lemaire V, et al. Autistic-like behavioral effects of prenatal stress in juvenile Fmr1 mice: the relevance of sex differences and gene-environment interactions. Sci Rep. 2022;12(1):7269. pmid:35508566
View Article
PubMed/NCBI
Google Scholar

[150] View Article

[151] PubMed/NCBI

[152] Google Scholar

[ref42] 42. Miao M, Han Y, Wang Y, Wang J, Zhu R, Yang Y, et al. Dysregulation of iron homeostasis and ferroptosis in sevoflurane and isoflurane associated perioperative neurocognitive disorders. CNS Neurosci Ther. 2024;30(2):e14553. pmid:38334231
View Article
PubMed/NCBI
Google Scholar

[154] View Article

[155] PubMed/NCBI

[156] Google Scholar

[ref43] 43. Wang J, Sun J, Qiao S, Li H, Che T, Wang C, et al. Effects of isoflurane on complex II‑associated mitochondrial respiration and reactive oxygen species production: roles of nitric oxide and mitochondrial KATP channels. Mol Med Rep. 2019;20(5):4383–90. pmid:31545457
View Article
PubMed/NCBI
Google Scholar

[158] View Article

[159] PubMed/NCBI

[160] Google Scholar

[ref44] 44. Greenacre M. Compositional data analysis. Annu Rev Stat Its Appl. 2021;8:271–99.
View Article
Google Scholar

[162] View Article

[163] Google Scholar

[ref45] 45. Malinouski M, Hasan NM, Zhang Y, Seravalli J, Lin J, Avanesov A, et al. Genome-wide RNAi ionomics screen reveals new genes and regulation of human trace element metabolism. Nat Commun. 2014;5:3301. pmid:24522796
View Article
PubMed/NCBI
Google Scholar

[165] View Article

[166] PubMed/NCBI

[167] Google Scholar

[ref46] 46. Kakhlon O, Cabantchik ZI. The labile iron pool: characterization, measurement, and participation in cellular processes1 1This article is part of a series of reviews on “Iron and Cellular Redox Status.” The full list of papers may be found on the homepage of the journal. Free Radic Biol Med. 2002;33(8):1037–46.
View Article
Google Scholar

[169] View Article

[170] Google Scholar

[ref47] 47. Berridge MJ, Bootman MD, Roderick HL. Calcium signalling: dynamics, homeostasis and remodelling. Nat Rev Mol Cell Biol. 2003;4(7):517–29. pmid:12838335
View Article
PubMed/NCBI
Google Scholar

[172] View Article

[173] PubMed/NCBI

[174] Google Scholar

[ref48] 48. De Leon-Rodriguez L, Lubag AJM, Sherry AD. Imaging free zinc levels in vivo - what can be learned? Inorg Chim Acta. 2012;393:12–23.
View Article
Google Scholar

[176] View Article

[177] Google Scholar

[ref49] 49. Lindahl PA, Moore MJ. Labile low-molecular-mass metal complexes in mitochondria: trials and tribulations of a burgeoning field. Biochemistry. 2016;55(30):4140–53. pmid:27433847
View Article
PubMed/NCBI
Google Scholar

[179] View Article

[180] PubMed/NCBI

[181] Google Scholar

[ref50] 50. Helmann JD. Metals in motion: understanding labile metal pools in bacteria. Biochemistry. 2025;64(2):329–45. pmid:39755956
View Article
PubMed/NCBI
Google Scholar

[183] View Article

[184] PubMed/NCBI

[185] Google Scholar

[ref51] 51. Hood MI, Skaar EP. Nutritional immunity: transition metals at the pathogen-host interface. Nat Rev Microbiol. 2012;10(8):525–37. pmid:22796883
View Article
PubMed/NCBI
Google Scholar

[187] View Article

[188] PubMed/NCBI

[189] Google Scholar

[ref52] 52. Lopez CA, Skaar EP. The impact of dietary transition metals on host-bacterial interactions. Cell Host Microbe. 2018;23(6):737–48. pmid:29902439
View Article
PubMed/NCBI
Google Scholar

[191] View Article

[192] PubMed/NCBI

[193] Google Scholar

Figures

Abstract

Introduction

Methodology

Experimental animals and design

Tissue collection and preparation

Ionome measurement

Mixed modeling

Model validation

Model evaluation

Results

Global model

Contrasts

Residuals

Discussion

Limitations and future directions

Supporting information

S1 File. Figure S1. Multivariate posterior predictive checks (PPC) for core elemental indicators.

S2 File. Figure S2. Markov Chain Monte Carlo (MCMC) sampling diagnostics.

S3 File. Figure S3. Forest plot displaying the median residual correlation coefficients and corresponding 95% HPD intervals for baseline elemental co-accumulation patterns among individual mice.

S4 File. Figure S4. Forest plot displaying the tissue-specific median residual correlation coefficients and corresponding 95% HPD intervals.

S5 File. Figure S5. Genotype-specific forest plot displaying the posterior median residual correlations and corresponding 95% HPD intervals for all elemental pairs, isolated by genotype.

S6 File. Table S1. Global multivariate Bayesian model parameter coordinates and chain diagnostics.

S7 File. Table S2. Verification matrix of model predictions.

S8 File. Table S3. Post-hoc assessment of multivariate residual dispersion and homoscedasticity.

S9 File. Table S4. Post-hoc assessment of multivariate residual dispersion and homoscedasticity.

References