miR-23b-5p: A potential biomarker in fluoride-induced neurological injury.

Abstract

Full-text online study on Science Direect at,
https://www.sciencedirect.com/science/article/pii/S0147651325016161

Highlights

Fluoride exposure induced cognitive and memory impairment in rats by downregulating serum BDNF levels.
miR-23b-5p may play a mediating role in fluoride-induced injury to SH-SY5Y cells.
miR-23b-5p is expected to be a potential biomarker for fluoride-induced neurological injury.

Fluoride-induced neurological injury has garnered widespread attention, however, its underlying mechanisms remain unclear and there is still a lack of effective biomarkers for early identification. To investigate the molecular mechanism of fluoride neurotoxicity and identify potential intervention targets, we established a fluoride-exposed rat model and an SH-SY5Y cell model. We found that 25, 50, and 100 mg/L NaF resulted in impaired spatial memory capacity, disturbed neuronal alignment in the cerebral cortex and hippocampus, and reduced BDNF levels in rats’ cerebral cortex. In vitro experiments demonstrated that fluoride increased apoptosis in SH-SY5Y cells. RNA sequencing revealed a significant elevation in miR-23b-5p levels in the brain tissues of fluoride-exposed rats. Subsequent validation experiments showed that fluoride upregulated miR-23b-5p expression in both rat brain tissues and SH-SY5Y cells. Notably, inhibition of miR-23b-5p led to increased brain-derived neurotrophic factor(BDNF) protein levels and reduced apoptosis in SH-SY5Y cells, indicating that miR-23b-5p served as a key regulator of fluoride-induced neurological injury. To further explore the potential of miR-23b-5p as a biomarker of fluoride-induced neurological injury, we conducted an epidemiologic survey in Wenshui County, Lvliang City, Shanxi Province, China. The results revealed that the expression level of miR-23b-5p in blood increased with the elevation of urinary fluoride, while serum BDNF expression decreased with increasing urinary fluoride, and miR-23b-5p exhibited a partial mediating effect between urinary fluoride and BDNF. This study identified a novel biomarker and potential intervention targets for fluoride-induced neurological injury.

Graphical Abstract

Keywords
Fluorosis
Neurological injury
MiR-23b-5p
Biomarker

EXCERPTS:

1. Introduction

Endemic fluorosis is a chronic systemic disorder characterized by long-term excessive fluoride exposure, with its typical clinical symptoms including dental fluorosis and skeletal fluorosis. The World Health Organization (WHO) estimates that over 200 million people worldwide are at risk of drinking-water type fluorosis, with tens of millions already exhibiting clinical symptoms (Wang et al., 2023a, Taher et al., 2024). Among various clinical manifestations, neurological injury is particularly significant (Scientific Reports, 2021, Lopes et al., 2020, Ren et al., 2022, Zhao et al., 2024, Wang et al., 2023b), manifesting as cognitive impairment, memory decline and reduced attention (Bashash et al., 2017, Retinasamy et al., 2019). Accumulating evidence suggested that fluoride exposure induced irreversible neurological injury, contributing to reduced intelligence quotient (IQ) in children and increased risk of neurodegenerative diseases in adults (Bashash et al., 2017). These impairments seriously compromise patients’ quality of life, resulting in delayed language development, loss of working capacity, and difficulties with activities of daily living (Taylor et al., 2025, Adkins and Brunst, 2021, Valdez-Jiménez et al., 2011). Due to its insidious onset, specific biomarkers for fluoride-induced neurological injury are urgently needed to facilitate early detection. However, the current lack of such biomarkers poses a significant challenge.

Previous studies explored the mechanisms of fluoride-induced neurological injury (Preston and Eichenbaum, 2013, Raud et al., 2023). Epidemiologically, prenatal fluoride exposure was linked to offspring cognitive impairment (Bashash et al., 2017, Green et al., 2019). Animal studies showed fluoride induced pathological changes including neuronal abnormalities, reduced density, apical dendrite atrophy and synaptic disconnection (Ning et al., 2021). In vitro, fluoride induced neurological injury via oxidative stress and dysregulation of key neuroinjury genes, such as BDNF (Ferreira et al., 2021). With the advancement of research, biomarkers associated with fluoride-induced neurological injury were refined: synaptophysin and postsynaptic density protein 95 (PSD-95) could serve as potential biomarkers for synaptic structural damage in fluoride-induced neurological injury (Li et al., 2022a, Zhu et al., 2011, Yang et al., 2018), while tight junction protein (ZO-1) could reflect blood-brain barrier dysfunction in the context of such neuroinjury (Xin et al., 2021). However, current research on these biomarkers was mostly limited to animal models and in vitro cell experiments, and had not yet been systematically validated in populations with different levels of fluoride exposure. Recent epigenetic research, particularly findings on abnormal microRNAs(miRNAs) expression, offered new insights—miRNAs integrated multi-level injuries and were detectable early. For example, fluoride disrupted neuronal function by altering miR-124 methylation and maturation (Dar et al., 2019). While progress in fluoride neurotoxicity mechanisms existed, specific biomarkers for population screening or early diagnosis remained lacking. In-depth study of epigenetic markers, particularly miRNAs, and their relationship with fluoride neurotoxicity would guide molecular risk assessment and public health strategies.

A growing number of studies showed miRNAs regulated fluorosis pathogenesis. As critical non-coding RNAs, they participated in diverse physiological and pathological processes. Epidemiologically, fluorosis patients showed abnormal serum miRNAs such as miR-155, miR-21 and miR-146a. Animal studies demonstrated that perinatal fluoride exposure upregulated brain miR-124 and miR-132 in offspring, causing learning and memory impairments (Wang et al., 2018), and altered 35 miRNAs in the liver (Zhao et al., 2022).Our prior work identified miR-200c-3p and miR-1213 in fluoride-induced skeletal fluorosis and cardiovascular dysfunction (Pei et al., 2018, Jiang et al., 2018). However, few miRNAs linked to fluoride-induced neurological injury were known, and no specific biomarkers for early diagnosis and screening existed. Thus, comprehensive study of miRNAs dynamic expression in this injury was imperative—it would clarify toxic mechanisms, provide targets for miRNA-based non-invasive early diagnosis, and benefit fluorosis early detection and precision prevention.

As key regulators, miRNAs played pivotal roles in neurological injury, with their expression changes promising as biomarkers. In Alzheimer’s disease (AD), cerebrospinal fluid miR-451a was reduced and positively correlated with cognitive scores (Feng et al., 2023); in Parkinson’s disease, abnormal peripheral blood miR-7 and miR-153 were linked to disease progression (Benoit et al., 2020)—supporting miRNAs as potential neurological disease diagnostic markers. Previous studies found specific miRNAs’ neuroprotective effects: miR-15a and miR-126 promoted vascular regeneration and neuronal survival in stroke models (Wu et al., 2017), while miR-122–5 p mitigated radiation-induced neuronal injury by suppressing neuroinflammatory pathways (Zhou et al., 2021). In AD, plasma miR-23b-3p was downregulated, positively correlated with cognitive scores, suggesting diagnostic value, and might treat via targeting GSK-3B to inhibit tau phosphorylation and neuronal apoptosis (Jiang et al., 2022). Given miRNAs’ role in neurological injury and peripheral detectability, investigating the expression of specific miRNAs—such as miR-23b—in fluoride-induced neurological injury would clarify toxic mechanisms and lay a foundation for developing blood-based miRNA detection systems for early biomarker identification.

In summary, this study demonstrated that fluoride exposure induced neurological injury through in vivo and in vitro experiments. RNA sequencing showed that miR-23b-5p was significantly upregulated in the cerebral cortex and hippocampus of fluoride-exposed rats, which was validated in animal and cell models. Cellular experiments confirmed that miR-23b-5p down-regulated BDNF expression and promoted apoptosis in SH-SY5Y cells. The population study revealed that the expression level of miR-23b-5p in blood increased with the elevation of urinary fluoride, while serum BDNF expression decreased with increasing urinary fluoride. miR-23b-5p played a partial mediating role in the association between urinary fluoride and serum BDNF, suggesting its potential as a latent early biomarker for fluoride-induced neurological injury.

… 2.14. Sample collection and testing of survey respondents

This study was conducted in three endemic villages of drinking-water type fluorosis in Wenshui County, Lvliang City, Shanxi Province. The inclusion criteria included: permanent residents aged >18 years who had lived locally for >5 years, primarily consumed local water sources, and had no major chronic diseases. Exclusion criteria comprised: individuals with missing samples, those who had recently used fluoride-containing medications/antioxidants, pregnant/lactating women, those with a history of neurological surgery/radiotherapy, and other medically unstable conditions. The questionnaire mainly includes basic data (age, gender, ethnicity), social characteristics (educational level), and lifestyle habits (smoking, alcohol consumption). Physical examinations mainly include height, weight, waist circumference, etc. Two hundred subjects were randomly selected from 573 screening participants. After excluding outliers and missing data, 140 subjects were finally included in the analysis. Fasting morning urine samples were collected from all subjects for the detection of urinary fluoride concentration. Five milliliters of fasting venous blood was collected from each subject and immediately transferred into a vacuum blood collection tube containing EDTA anticoagulant, which was gently inverted and mixed to prevent blood coagulation. After standing and centrifugation, the formed blood clots were collected and stored in a -80 °C refrigerator for subsequent detection of miR-23b-5 p content. Before detection, the frozen blood clots were taken out, cut into pieces with sterile scissors, added with 1?mL of TRIzol (Invitrogen, America) reagent, and homogenized in an ice bath using a tissue homogenizer until there were no obvious lumps. Total RNA was extracted following the standard RNA extraction protocol, and then miR-23b-5 p was detected by stem-loop quantitative real-time polymerase chain reaction (ABI, USA). Neuroinjury-related proteins were detected using a multiplex assay technology (Luminex Corporation, USA). The study protocol was approved by the Ethics Committee of the Chinese Center for Disease Control and Prevention (Approval No.: HRBMUECDC20210410), and written informed consent was obtained from all participants.

… 3. Result

3.1. General conditions of fluoride-exposed rats and fluoride contents in urine, blood and brain tissues

The results showed that compared with the control group, the body weight of fluoride-exposed rats decreased in a dose-dependent manner (p<0.01, Fig. 1A), while the brain-to-body weight ratio increased with increasing levels of fluoride (p<0.01, Fig. 1B). This may be because the inhibitory effect of fluoride on overall growth was significantly stronger than its effect on the brain. Therefore, even if brain weight decreased slightly, its proportion relative to body weight would still increase. To assess fluoride exposure, we examined urinary fluoride, serum fluoride, brain fluoride and dental fluorosis in rats. As shown in Fig. 1C-E, the levels of urinary fluoride, serum fluoride and brain fluoride were higher in rats than in the control group after 12 weeks of free water intake (p<0.05). The teeth of rats in the control group were translucent and shiny. In contrast, rats exposed to 25 mg/L NaF developed a small number of white plaques on the tooth surface. For rats exposed to 50 mg/L and 100 mg/L NaF, their teeth showed horizontal stripes with varying degrees of yellow and white coloration, accompanied by reduced translucency and obvious damage to the tooth structure (Fig. 1F). The above results indicated that the fluoride-exposed rat model was successfully established.

3.2. Fluoride exposure induced brain tissues injury in ratsTo investigate the effects of fluoride on brain injury, the MWM was used to assess the spatial learning and memory abilities of rats. The results showed that the latency to find the platform was significantly longer in the fluoride exposed group than in the control group (p<0.01, Fig. 2A). After removing the platform, the frequency of crossing the target quadrant in the fluoride exposed group was significantly lower than that in the control group (p<0.01, Fig. 2B). The results of daily training showed that the average latency of rats in the fluoride-exposed group was significantly longer than that in the control group every day(Fig. 2C). The swimming paths of the fluoride-exposed group were more dispersed and disorganized compared with those of the control group (Fig. 2D). These results indicated that fluoride exposure impaired spatial learning and memory in rats.

To investigate in depth the neurological injury caused by fluoride exposure, histopathological changes in rat brain tissue were systematically observed in this study. The results showed that cells in the control group were arranged in an orderly manner with uniform cytoplasmic staining. In the 25 mg/L group, the cerebral cortex and hippocampal structure of rats were clear without obvious structural damage; however, in the 50 mg/L and 100 mg/L groups, the cell bodies in the rat cerebral cortex showed significant shrinkage, with blurred tissue stratification and disordered cell arrangement. Additionally, a large number of pyramidal cells in the hippocampal region were reduced, and the cell bodies were fragmented with vacuole formation(Fig. 2E). Nissl staining showed that neurons in the hippocampal CA1 region of the control group were intact, closely arranged, and rich in Nissl bodies, while the high-dose group exhibited a significant reduction in the number of neurons and loose arrangement (Fig. 2F). Quantitative analysis indicated that the neuronal density in the fluoride-exposed groups was significantly lower (p<0.001) and showed a dose-dependent decreasing trend with increasing fluoride exposure (Fig. 2G). Under electron microscopy, mitochondrial swelling and mitochondrial cristae fragmentation were observed in the 25 mg/L group. In the 50 mg/L group, mitochondrial cristae fragmentation and partial loss of the double-layer structure of the nuclear envelope were noted in neurons. In the 100 mg/L group, chromatin margination occurred; additionally, microtubules showed poor continuity with severe breakage and loose tissue structure, along with cytoplasmic vacuoles and mitochondrial cristae fragmentation or disappearance(Fig. 2H).Subsequently, the protein level of BDNF, a key factor of brain injury in rat cerebral cortex, was detected (Fig. 2I, J). The results showed that the protein level of BDNF was significantly reduced in the 100 mg/L NaF group compared with the control group (p<0.001).The above results suggested that fluoride exposure could lead to dose-dependent neurological injury, the extent of which is closely related to the exposure dosage.

3.3. Fluoride exposure induced SH-SY5Y cell injuryBased on the RTCA assay, 30 mg/L NaF was determined to be used for the fluorine-induced SH-SY5Y cell injury model construction(Fig. 3A). At this dose, the activity of SH-SY5Y cells was significantly lower than that of the control group and was in the safe survival range (activity > IC₅₀), which could ensure the cell state for subsequent detection and induce SH-SY5Y cell injury. Further detection of BDNF protein levels in SH-SY5Y cells showed that BDNF protein expression in the 30 mg/L NaF group was significantly lower than that in the control group (p<0.05, Fig. 3B, C).

The effects of fluoride exposure on apoptosis of SH-SY5Y cells were shown in Fig. 3D-G. The results indicated that 30 mg/L NaF significantly increased the apoptosis of SH-SY5Y cells to 2.31 times of the control group (p<0.01). These data confirmed the impairing effect of fluoride exposure on SH-SY5Y cells.

3.4. RNA sequencing revealed fluoride increased miR-23b-5p expression in brain tissues of fluoride-exposed ratsThe miRNA expression profiles of brain tissues from fluoride-exposed and control rats were analyzed by high-throughput sequencing. Results showed that 40 differentially expressed miRNAs were identified in the cerebral cortex (19 upregulated and 21 downregulated, Fig. 4A), and 33 differentially expressed miRNAs were detected in the hippocampus (16 upregulated and 17 downregulated, Fig. 4B). Heatmap analysis revealed specific expression patterns of these miRNAs (Fig. 4C-D). Nine differentially expressed miRNAs were shared between the cerebral cortex and hippocampus, suggesting their potential involvement in the common regulatory mechanisms of fluoride neurotoxicity. Among them, miR-23b-5p showed the most significant expression change and was speculated to be a core regulatory molecule mediating fluoride-induced neurological injury. Fig. 4

Fig. 4. RNA sequencing revealed fluoride increased miR-23b-5p expression in brain tissues of fluoride-exposed rats. (A) Volcano map of differentially expressed miRNAs in the cerebral cortex. (B) Volcano map of differentially expressed miRNAs in the hippocampus. (C) Heatmap of differentially expressed miRNAs in the cerebral cortex. (D) Heatmap of differentially expressed miRNAs in the hippocampus. n=3.

3.5. Functional enrichment analysis of GO and KEGG for target genes of differentially expressed miRNAs

To further explore the potential regulatory functions of the differential miRNAs, GO and KEGG enrichment analyses were performed on the target genes of differential miRNAs in brain tissues between fluoride-exposed rats and control rats. The GO functions of target genes of differential miRNAs in the cerebral cortex were significantly enriched in “cellular response to UV-C” and “translesion synthesis” (Fig. 5A), and their KEGG pathways were significantly enriched in “Fanconi anemia pathway” and “Fc epsilon RI signaling pathway” (Fig. 5C). The GO functions of target genes of differential miRNAs in the hippocampus were significantly enriched in “single stranded viral RNA replication via double stranded DNA intermediate” and “regulation of glycogen (starch) synthase activity” (Fig. 5B), and their KEGG pathways were significantly enriched in “Adipocytokine signaling pathway” and “Vascular smooth muscle contraction” (Fig. 5D). Fig. 5

Fig. 5. GO and KEGG enrichment analyses of target genes of differentially expressed miRNAs in rats. (A) Go enrichment plot of target genes of differential miRNAs in the cerebral cortex. (B) Go enrichment plot of target genes of differential miRNAs in the hippocampus. (C) KEGG metabolic pathway map of target genes of differential miRNAs in the cerebral cortex. (D) KEGG metabolic pathway map of target genes of differential miRNAs in hippocampus.

3.6. Fluoride exposure upregulated miR-23b-5p expression in rat brain tissues and SH-SY5Y cells

To verify the RNA sequencing results, miR-23b-5p expression in rat brain tissues and SH-SY5Y cells was examined. The results showed that miR-23b-5p expression was significantly up-regulated in the cerebral cortex and hippocampal tissues of 50 and 100mg/L group compared with the control group (p<0.01, Fig. 6A-B).

For SH-SY5Y cells, miR-23b-5p expression level was significantly increased after 30 mg/L NaF treatment compared with control (p<0.001, Fig. 6C). The results of the animal and cellular models were consistent, confirming that fluoride exposure upregulated miR-23b-5p expression.

3.7. miR-23b-5p mediated fluoride-induced apoptosis in SH-SY5Y cells

To investigate whether miR-23b-5p mediated fluoride-induced apoptosis in SH-SY5Y cells, functional validation experiments were performed by inhibiting miR-23b-5p expression. The results showed that exposure to 30 mg/L NaF for 12 h significantly inhibited BDNF expression in SH-SY5Y cells, while inhibition of miR-23b-5p significantly increased the expression of BDNF (p<0.05). The BDNF expression level in the inhibitor group was significantly higher than that in the co-treatment group (p<0.01, Fig. 7A-B), indicating that inhibition of miR-23b-5p can more effectively promote BDNF expression in the absence of fluoride exposure. Neuronal apoptosis assays revealed that NaF treatment significantly increased apoptosis (p<0.05, Fig. 7C-F), and the apoptosis was reduced in the inhibitor group (p<0.05). The apoptosis in the co-treatment group was significantly lower than that in the NaF group(p<0.0001). The results suggested that inhibition of miR-23b-5p could partially reverse fluoride-induced apoptosis in SH-SY5Y cells. Taken together, miR-23b-5p may mediate fluoride-induced apoptosis in SH-SY5Y cells.

Fig. 7

Fig. 7. miR-23b-5p mediated fluoride-induced apoptosis in SH-SY5Y cells. (A) Protein expression of BDNF in SH-SY5Y cells after transfection with miR-23b-5p inhibitor. (B) Quantification of BDNF protein expression in SH-SY5Y cells after transfection with miR-23b-5p inhibitor. (C) Apoptosis of SH-SY5Y cells after transfection with miR-23b-5p inhibitor. (D-F) Quantification of early, late and total apoptosis in SH-SY5Y cells after transfection with miR-23b-5p inhibitor. *p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001 versus the control group. ### p<0.001, #### p<0.0001 versus the NaF group. ++ p<0.01 versus the Inhibitor group.

3.8. miR-23b-5p as a potential biomarker for fluoride-induced neurological injury in populations

In this study, urinary fluoride concentration was used as the indicator to evaluate fluoride exposure levels. The distribution characteristics of urinary fluoride concentrations among the 140 subjects were as follows: the median was 1.59 mg/L, with an interquartile range of (0.93 mg/L, 2.08 mg/L). For the grouping criteria, the study population was divided into 3 groups according to urinary fluoride concentrations using the tertile method: the low-exposure group (UF <0.93 mg/L, n=47), the medium-exposure group (0.93 mg/L < UF < 2.08 mg/L, n=47), and the high-exposure group (UF > 2.08 mg/L, n=46). The median urinary fluoride concentration in each group was 0.77 mg/L (low fluoride group), 1.62 mg/L (medium fluoride group), and 2.86 mg/L (high fluoride group), and the differences between the groups were statistically significant (p<0.001). Analysis of the baseline data showed that there were no statistically significant differences among the three study groups in terms of demographic characteristics and health indicators, including age, gender, education level, smoking history, drinking history and body mass index (BMI) (p>0.05), indicating that the groups were comparable (Table 1).

Table 1. Basic characteristics of study population.

Empty Cell	ALL (n=140)	Low UF (n=47)	Medium UF(n=47)	High UF (n=46)	p
Sex					0.251
Male	50(35.7%)	19(40.4%)	19(40.4%)	12(26.1%)
Female	90(64.3%)	28(59.6%)	28(59.6%)	34(73.9%)
Age (years)	55.76±10.62	55.32±10.82	54.74±10.71	57.26±10.39	0.492
Education					0.379
Primary and below	48(34.3%)	20(42.6%)	12(25.5%)	16(34.8%)
Junior high school	81(57.9%)	23(48.9%)	30(63.8%)	28(60.9%)
High school and above	11(7.9%)	4(8.5%)	5(10.6%)	2(4.3%)
Smoking					0.574
No	116(82.9%)	39(83.0%)	37(78.7%)	40(87.0%)
Yes	24(17.1%)	8(17.0%)	10(21.3%)	6(13.0%)
Drinking					0.079
No	111(79.3%)	37(78.7%)	33(70.2%)	41(89.1%)
Yes	29(20.7%)	10(21.3%)	14(29.8%)	5(10.9%)
BMI (kg/m²)	25.04±3.46	24.85±3.20	24.75±4.08	25.52±3.04	0.507
UF (mg/L)	1.59(0.93,2.08)	0.77(0.58,0.94)	1.62(1.34,1.73)	2.86(2.08,3.88)	<0.001

The results of the mediation analysis (Fig. 8) showed that urinary fluoride levels had a significant positive effect on miR-23b-5p (a=1.28, p<0.001), while miR-23b-5p exerted a significant negative effect on BDNF (b=-1099, p<0.05). Further analysis revealed that the direct inhibitory effect of urinary fluoride on BDNF was c’= -1581.9 (p<0.05), and the total effect including the mediating role was c= -2992.1 (p<0.001). Among these, the indirect effect mediated by miR-23b-5p (ab=-1410, p<0.05) accounted for 47.1% of the total effect. These findings indicated that fluoride exposure not only directly inhibited BDNF expression but also exacerbated BDNF decline by upregulating miR-23b-5p, suggesting that miR-23b-5p served as a key mediator in the association between fluoride exposure and neurological injury markers. Its value as a potential biomarker therefore warranted attention.

4. Discussion

The early detection and mechanistic analysis of fluoride-induced neurological injury have become international research hotspots. Epidemiological studies have shown that long-term fluoride exposure is significantly associated with cognitive dysfunction, memory loss, and an increased risk of neurodegenerative diseases (Russ et al., 2020, Li et al., 2016, NTP monograph, 2024). Due to the latency and irreversibility of neurological injury, early diagnosis and intervention face significant challenges. Therefore, exploring highly sensitive and specific biomarkers is of great significance for the early warning and mechanistic study of fluoride-induced neurological injury. Focusing on the screening of biomarkers for fluoride-induced neurological injury, this study confirmed through multi-dimensional research that miR-23b-5p could serve as a potential biomarker for fluoride-induced neurological injury. This study provided critical theoretical evidence for the early diagnosis of fluorosis, the formulation of preventive strategies, and the screening of precise intervention targets, offering novel directions and robust scientific support for the prevention and treatment of fluoride-induced neurological injury.

The fluoride-induced neurological injury have been confirmed across multiple research levels. Epidemiologic investigations have shown that fluoride exposure is negatively associated with cognitive impairment (Grandjean and Landrigan, 2014), with long-term consumption of high-fluoride water linked to significantly reduced IQ scores in children and a dose-response relationship between urinary fluoride levels and cognitive dysfunction (Choi et al., 2015). In animal experiments, the results of our study were highly consistent with those of Rocha-Amador et al (Rocha-Amador et al., 2007).: MWM tests revealed prolonged escape latency and fewer platform crossings in fluoride-exposed rats, consistent with spatial learning and memory deficits reported by Valdez-Jiménez et al (Valdez-Jiménez et al., 2011). Histopathological examinations showed disorganized neuron arrangement and reduced Nissl bodies in the hippocampal CA1 region of fluoride-exposed rats, while electron microscopy revealed ultrastructural changes such as neuronal mitochondrial swelling and cristae fragmentation. In vitro experiments showed that 30 mg/L NaF exposure of SH-SY5Y cells for 12 h resulted in an increase in apoptosis level, which is consistent with the fluorine-induced neuronal apoptosis results reported by Ke et al. and Zhou et al (Ke et al., 2016). and Zhou et al (Zhou et al., 2019). Collectively, these findings indicated that fluoride exposure induced cognitive and neurological injury in rats.

In this in vivo study, we employed three exposure doses of NaF: 25, 50, and 100 mg/L. After conversion, the corresponding human equivalent exposure doses were approximately 3.9, 7.9, and 15.8 mg/L, respectively. The WHO recommends an upper limit of fluoride concentration in drinking water of 1.5 mg/L. However, in some high-fluoride areas, fluoride levels in drinking water can reach up to 20 mg/L, significantly exceeding the WHO guideline. Therefore, the use of these doses was justified when studying health effects associated with high fluoride exposure. Additionally, the three-dose design was implemented to determine the minimum effective dose required to induce neural damage and alter miR-23b-5p expression, to observe whether the effects intensify with increasing dose, and to ensure that the 100 mg/L dose can reliably replicate typical neuropathological phenotypes related to fluorosis within a relatively short experimental period. Meanwhile, the reasons for setting the fluoride exposure duration to 12 weeks in this study are as follows: A 12 week (3 month) period is generally recognized as a subchronic exposure duration across toxicological research fields. Its core advantage lies in its ability to stably induce typical pathological phenotypes of fluoride toxicity in rats. Existing studies have shown that fluoride exposure for either 3 months or 6 months can cause damage to rats, such as increased urinary fluoride levels, obvious dental fluorosis, and cognitive dysfunction (Wang et al., 2023b, Dong et al., 2024). From the perspective of economic efficiency, a 12-week exposure duration can significantly reduce animal feeding costs and shorten the experimental cycle. In summary, the selection of a 12 week fluoride exposure duration in this study not only conforms to the classical design standards for subchronic toxicity models but also balances the feasibility and economy of the experiment, thereby providing a stable model foundation for subsequent investigations into the toxic mechanisms associated with fluoride exposure. Urinary fluoride level is a core indicator reflecting internal fluoride exposure. The results of this study showed that urinary fluoride levels in rats increased with higher exposure doses, directly confirming the successful establishment of the fluoride exposure dose model and laying a foundation for subsequent research. However, this differed from the actual scenario of long-term low-concentration fluoride exposure in humans. Therefore, in future research, models with lower concentrations and longer exposure durations can be adopted to conduct relevant studies.

BDNF is a critical molecular marker reflecting neurological injury (Autry and Monteggia, 2012). Clinical studies showed that serum BDNF levels in Alzheimer’s disease patients were negatively correlated with disease severity (Laske et al., 2007), chronic stress led to sustained downregulation of hippocampal BDNF expression (Duman and Monteggia, 2006), and multiple neurotoxins induce BDNF expression abnormalities that precede clinical symptoms. This study found that fluoride exposure decreased BDNF levels both in the cerebral cortex and SH-SY5Y cells, and population analysis revealed a negative correlation between urinary fluoride and BDNF, consistent with previous findings and validating the universality of BDNF as a neurological injury marker. Existing research demonstrated that subchronic fluoride exposure reduced hippocampal BDNF protein expression by 40% in rats, accompanied by spatial memory impairment (Niu et al., 2009), which aligned with the results of this study. Previous studies pointed out that fluoride-induced neurological injury may also cause changes in a number of other indicators, including oxidative stress indicators (e.g., ROS, MDA, SOD) (Barbier et al., 2010), inflammatory cytokines (TNF-a, IL-6, etc.) (Fu et al., 2022), synaptic plasticity proteins (PSD-95, etc.) (Zhu et al., 2011), and apoptosis-related proteins. In this study, BDNF was selected as a marker protein for neurological injury because it is directly related to cognitive function as a core regulator of neuronal survival and synaptic plasticity, and because peripheral BDNF levels can reflect neurological injury. Notably, while BDNF has marker advantages, neurological injury involves synergistic interactions within multi-molecular networks. Future studies may combine other neurological injury-related molecules (e.g., NGF, NT-3, oxidative stress and inflammation indicators) for joint testing to establish a more comprehensive fluoride- induced neurological injury evaluation system.

In recent years, research on fluoride exposure and neurological injury has made progress, with multiple biomarkers being used to assess fluoride-induced neurological injury. Traditional studies focused on oxidative stress markers (e.g., ROS, MDA) and inflammatory factors (e.g., TNF-a, IL-6), while indicative of cellular injury from fluoride exposure, lack tissue specificity and appear late in the process. With the deepening of epigenetics research, microRNAs have emerged as novel biomarkers with advantages, and their associations with neurological disorders have been widely reported (Chandran et al., 2024, Banack et al., 2024). For example, serum miR-124–3p is downregulated in Parkinson’s disease patients (Esteves et al., 2022), miR-155 modulates microglial function via IFN-y signaling to improve cognition in Alzheimer’s disease (Yin et al., 2023), and miR-29c-3p has also been linked to Alzheimer’s disease (Sha et al., 2021). Notably, in neurological injury models, changes in miR-23b expression were closely associated with the progression of nerve injury. Studies showed that miR-23b could target GnT-III, activate the Akt/GSK-3B signaling pathway to inhibit tau protein pathology, and altered the processing of amyloid precursor protein (ABPP) to suppress oxidative stress, thereby improving cognitive impairment and pathological symptoms related to AD (Pan et al., 2021); in a rat model of cerebral ischemia-reperfusion-induced neurological injury, miR-23b expression was downregulated, and it promoted hippocampal neuron apoptosis by regulating the TAB3/NF-_kB/p53 axis (Roshan-Milani et al., 2022); Schwann cells subjected to mechanical stimulation deliver miR-23b-3p to neurons via their extracellular vesicles, downregulating the expression of neuropilin 1 (Nrp1) in neurons and promoting axonal regeneration (Xia et al., 2020). This study identified a critical role for miR-23b-5p in fluoride-induced neurological injury: its expression was significantly elevated in both animal models and SH-SY5Y cell models of fluoride exposure. A critical follow-up question is: how does fluoride regulate the expression of miR-23b-5p? Although this study did not directly investigate its upstream mechanisms, based on the known toxicological effects of fluoride and the regulatory properties of miR-23b-5p, we propose several plausible possibilities for validation in future research. Firstly, oxidative stress is likely a key upstream event. Numerous studies have confirmed that fluoride exposure rapidly induces a sharp increase in intracellular reactive oxygen species (ROS) levels (Li et al., 2024). ROS act not only as effector molecules but also as important signaling messengers, which can activate transcription factors such as NF-_kB and p53 (Li et al., 2022b, Zhang et al., 2025). These transcription factors may directly bind to the promoter region of miR-23b-5p to drive its transcription. Secondly, neuroinflammation may play a bridging role. Fluoride can activate glial cells, leading to the release of pro-inflammatory cytokines such as TNF-a and IL-1B (Wang et al., 2024). These inflammatory mediators may act on neurons in a paracrine manner, indirectly regulating the expression of miR-23b-5p by activating their surface receptors (e.g., TNFR) (Li et al., 2021) and downstream signaling pathways (Yang et al., 2022). Additionally, epigenetic reprogramming cannot be ignored. Fluoride may alter the chromatin accessibility of the miR-23b-5p genomic locus through mechanisms such as regulating DNA methylation or histone modification, thereby relieving the inhibition of its transcription and promoting its expression (Yamashita et al., 2024, Dey Bhowmik et al., 2023). In the population analysis, the negative correlation between urinary fluoride levels and BDNF reflected not only a direct inhibitory effect but also an exacerbation of BDNF decline via upregulation of miR-23b-5p. Mediation analysis showed that approximately 47.1% of the fluoride-induced BDNF reduction was mediated by miR-23b-5p. This finding provided human evidence for the pathway of “fluoride exposure – non-coding RNA regulation-neurological injury”, suggesting that miR-23b-5p might serve as a potential biomarker linking fluoride exposure to neurological injury. Although this study attempted to reduce confounding bias through strict inclusion and exclusion criteria (e.g., controlling for years of residence and excluding major chronic diseases), as an observational study, caution must still be exercised regarding the potential interference of uncontrolled confounding factors on the results. In terms of dietary nutrition, minerals such as calcium and magnesium may alter fluoride burden in the body by affecting fluoride absorption and metabolism, while antioxidant nutrients may alleviate fluoride-induced oxidative stress (Tao et al., 2022, Tefera et al., 2022). Failure to fully adjust for dietary differences could interfere with the authenticity of the association between fluoride and miR-23b-5p. Differences in genetic background are also not to be overlooked: genetic polymorphisms in fluoride metabolism-related enzymes (e.g., UGT, GST) and in the regulatory regions of miR-23b-5p may lead to individual variations in responses to fluoride exposure, thereby influencing the interpretation of results (He et al., 2023). Future research should integrate biomarker profiling and genomic analyses to precisely control for nutritional and co-exposure confounds and elucidate genetic susceptibility, thereby clarifying the causal relationship and individual variability in fluoride-induced neurological injury. Notably, bioinformatics analysis (TargetScan and miRDB databases) revealed potential binding sites for miR-23b-5p to target cAMP response rlement binding protein 1(CREB1). Proteomics assays showed that CREB1 protein levels and BDNF expression were synchronously decreased in the fluoride-exposed group, while protein-protein interaction analysis indicated a functional association between CREB1 and BDNF. Based on these data, we speculate that miR-23b-5p may be involved in fluoride-induced neurological injury by affecting the CREB1/BDNF pathway, providing an important direction for follow-up mechanistic studies. However, their regulatory relationships were not verified by gene overexpression or knockdown in this study. Given the high stability and detection sensitivity of miRNAs in bodily fluids, and the specific association of miR-23b-5p with neurological injury, we propose miR-23b-5p as a novel molecular marker for assessing fluoride neurotoxicity. This could facilitate early identification of high-risk populations and provide new targets for developing intervention strategies.

This study had the following limitations: Firstly, only male rats were included in the experiment, which was not only due to economic considerations but also aimed to improve data homogeneity and reproducibility, as male rats generally exhibited more stable fluoride pharmacokinetics and neurotoxic responses. However, the importance of gender differences was undeniable. Existing studies confirmed that estrogen could mediate the expression regulation of miRNAs through nuclear receptors (e.g., ERa, ERB) (Klinge, 2020). For instance, it could upregulate the miR-200 family to inhibit neuroinflammatory responses (Márton et al., 2020), or regulate the expression of neuron differentiation-related genes via miR-124 (Varshney et al., 2017). Future studies might include female rats to more comprehensively reveal the gender-specific regulatory mechanisms underlying fluoride-induced neurological injury. Secondly, although flow cytometry was used to detect trends in cell apoptosis rates, quantitative techniques such as TUNEL staining were not employed for single-cell localization and counting. This limitation arises from the research’s initial focus on molecular marker screening, with prioritization of high-throughput flow cytometry for rapid data acquisition, while TUNEL staining requires longer processing time, thus lacking direct morphological evidence of the association between fluoride exposure and neuronal apoptosis. Additionally, the human population analysis only explored the correlation between urinary fluoride levels and biomarkers (e.g., miR-23b-5p, BDNF), without conducting standardized cognitive function assessments (e.g., MMSE, MoCA). This is mainly due to limited resources in the cross-sectional survey and the difficulty of completing large-scale testing within a short period. Meanwhile, cognitive function assessments are significantly influenced by educational level and participant cooperation, so the research team prioritized objective biological sample testing as the primary endpoint, resulting in insufficient evidence for the association between fluoride exposure and human neurological injury. Finally, in vitro experiments, only a single dose (30 mg/L NaF) was used, and the downstream regulatory pathways of miR-23b-5p (e.g., CREB1/BDNF signaling axis) were not investigated. Future studies intend to deepen the mechanistic research through multi-dose design and downstream target detection.

In conclusion, this study revealed the key regulatory role of miR-23b-5p in fluoride-induced neurological injury. Experimental results showed that fluoride exposure significantly upregulated the expression of miR-23b-5p, while concomitantly downregulating BDNF expression. The significant mediating effect of miR-23b-5p between fluoride exposure and neurological injury was confirmed at the population level, and this finding provides key epidemiological evidence for the establishment of miR-23b-5p as a novel biomarker of fluorine-induced neurological injury. Certainly, fluoride-induced neurological injury was a complex process involving multiple pathways. Beyond the miR-23b-5p/BDNF axis revealed in this study, previous research confirmed that signaling pathways such as RhoA/ROCK (Chen et al., 2023) and oxidative stress/JNK (Liu et al., 2011) also played important roles. Therefore, miR-23b-5p was more likely to be one of the key factors in this complex regulatory network rather than the sole factor. The potential interactive relationships between miR-23b-5p and other pathways would be an important direction for future research.

5. Conclusion

This study demonstrated that fluoride exposure contributed to fluoride-induced neurological injury by upregulating miR-23b-5p, which suppressed BDNF expression and promoted apoptosis. These findings suggested that miR-23b-5p was one of the key regulatory factors in fluoride-induced neurological injury and could serve as a potential biomarker. However, since fluoride-induced neurological injury involved a complex signaling network, the role of miR-23b-5p likely represented one critical component within this intricate network. Further studies were needed to explore additional regulatory factors to fully elucidate the molecular mechanisms underlying fluoride-induced neurological injury. This study provided new theoretical insights into the pathogenic mechanisms of fluoride-induced neurological injury and identified potential targets for preventive and therapeutic strategies.

CRediT authorship contribution statement

Wei Huang: Conceptualization, Data curation, Investigation, Formal analysis, Methodology. Yunzhu Liu: Writing – original draft, Writing – review and editing. Shuaifei Yang: Writing – review and editing. Xirui Feng: Writing – review and editing. Jianguo Feng: Writing – review and editing. Yuting Jiang: Writing – review and editing. Yanhui Gao: Supervision, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by National Key Research and Development Program of China (grant number 2022YFC2503000).

Appendix A. Supplementary material

Download: Download Word document (17KB)

Data availability

The data that has been used is confidential.

References

Adkins and Brunst, 2021

E.A. Adkins, K.J. Brunst

Impacts of fluoride neurotoxicity and mitochondrial dysfunction on cognition and mental health: a literature review

Int. J. Environ. Res. Public Health, 18 (24) (2021), p. 12884

Crossref

View in ScopusGoogle ScholarAutry and Monteggia, 2012

A.E. Autry, L.M. Monteggia

Brain-derived neurotrophic factor and neuropsychiatric disorders

Pharm. Rev., 64 (2) (2012), pp. 238-258

View PDF View article CrossrefView in ScopusGoogle ScholarBanack et al., 2024

S.A. Banack, R.A. Dunlop, P. Mehta, et al.

A microRNA diagnostic biomarker for amyotrophic lateral sclerosis

Brain Commun., 6 (5) (2024), Article fcae268

View in Scopus

Google ScholarBarbier et al., 2010

O. Barbier, L. Arreola-Mendoza, L.M. Del Razo

Molecular mechanisms of fluoride toxicity