Introduction

Fluoride (F) is considered one of the essential elements for the maintenance of the normal cellular processes in the organism1 and is largely employed in dentistry to control dental caries2. However, when excessive amounts are ingested, F can induce oxidative stress and lipid peroxidation, alter intracellular homeostasis and cell cycle, disrupt communication between cells and signal transduction and induce apoptosis3.

Nearly 25% of ingested F is absorbed from the stomach as an undissociated molecule (HF) in a process that is inversely related to pH4, while the remainder is absorbed in the ionic form (F−) from the small intestine, in a pH-independent manner5. Due to its major role in F absorption, the gastrointestinal tract (GIT) is regarded as the principal way of exposure to F6. Thus, gastrointestinal symptoms, including motion sickness, vomiting, diarrhea and abdominal pain are the first signs of F toxicity7,8,9,10.

The Enteric Nervous System (ENS) is an interlinked network of neurons disposed in the intestinal walls that controls the function of the GIT11. Due to its control function, changes in ENS affect the absorption, secretion, permeability and motility of the GIT12. Recently, immunofluorescence techniques revealed important alterations in the morphology of different types of enteric neurons and proteomic analysis demonstrated changes in the expression of several proteins of the duodenum of rats13 after chronic exposure to F, providing the first insights for the comprehension of the mechanisms underlying the actions of F on the bowel. However, the effect of F on the ENS and proteomic profile of the jejunum has never been reported. Considering that each segment of the small intestine has distinct anatomical, histological and physiological characteristics with functional implications14, this study evaluated the morphology of distinct subtypes of enteric neurons of the jejunum after chronic exposure to F. Quantitative label-free proteomics tools were employed to evaluate the changes on the pattern of protein profile of the jejunum, after exposure to F, in attempt to provide mechanistic explanations for the effects of this ion in the intestine.

Material and Methods

Animals and treatment

The Ethics Committee for Animal Experiments of Bauru Dental School, University of São Paulo approved all experimental protocols (#014/2011 and #012/2016). All experimental protocols were approved by. The assays conformed with the guidelines of the National Research Council. Eighteen adult male rats (60 days of life - Rattus norvegicus, Wistar type) were randomly assigned to 3 groups (n = 6/group). They remained one by one in metabolic cages, having access to water and food ad libitum under standard conditions of light and temperature. The animals received deionized water (0 mgF/L), 10 mgF/L and 50 mgF/L for 30 days as sodium fluoride (NaF) dissolved in deionized water, in order to simulate chronic intoxication with F. Since rodents metabolize F 5 times faster than humans, these F concentrations correspond to ~2 and 10 mg/L in the drinking water of humans15. After the experimental period, the animals received an intramuscular injection of anesthetic and muscle relaxant (ketamine chlorhydrate and xylazine chlorhydrate, respectively). While the rats were anesthetized, the peritoneal and thoracic cavities were exposed, and the heart was punctured for blood collection, using a heparinized syringe. Plasma was obtained by centrifugation at 800 g for 5 minutes for quantification of F, described in a previous publication13. After blood collection, the jejunum was collected for histological, immunofluorescence and proteomic analysis. For the collection of the jejunum, animal chow was removed from the animals 18 hours prior the euthanasia to decrease the volume of fecal material inside the small intestine, facilitating the cleaning process for posterior processing. After laparotomy, to remove the jejunum, initially the small intestine was localized, and the jejunum proximal limit was identified by the portion after the duodenojejunal flexure that is attached to the posterior abdominal wall by the ligament of Treitz. After incisions in the flexure and ligament, 20 centimeters distally to the incision were despised and then 15 centimeters of the jejunum were harvested for processing. After harvesting, the jejunum was washed with PBS solution applied several times with a syringe in the lumen to remove completely any residue of fecal material.

Histological analysis

This analysis was performed exactly as described by Melo, et al.13.

Myenteric plexus immunohistochemistry, morphometric and semi-quantitative analysis

These analyses were performed exactly as described by Melo, et al.13.

Proteomic analysis

The frozen jejunum was homogenized in a cryogenic mill (model 6770, SPEX, Metuchen, NJ, EUA). Samples from 2 animals were pooled and analyses were carried out in triplicates. Protein extraction was performed by incubation in lysis buffer (7 M urea, 2 M thiourea, 40 mM DTT, all diluted in AMBIC solution) under constant stirring at 4 °C. Samples were centrifuged at 14000 rpm for 30 min at 4 °C and the supernatant was collected. Protein quantification was performed16. To 50 μL of each sample (containing 50 μg protein) 25 μL of 0.2% Rapigest (WATERS cat#186001861) was added, followed by agitation and then 10 μL 50 mM AMBIC were added. Samples were incubated for 30 min at 37 °C. Samples were reduced (2.5 μL 100 mM DTT; BIORAD, cat# 161–0611) and alkylated (2.5 μL 300 mM IAA; GE, cat# RPN 6302 V) under dark at room temperature for 30 min. Digestion was performed at 37 °C overnight by adding 100 ng trypsin (PROMEGA, cat #V5280). After digestion, 10 µL of 5% TFA were added, incubated for 90 min at 37 °C and sample was centrifuged (14000 rpm at 6 °C for 30 min). Supernatant was purified using C 18 Spin columns (PIERCE, cat #89870). Samples were resuspended in 200 μL 3% acetonitrile.

LC-MS/MS and bioinformatics analyses

The peptides identification was done on a nanoAcquity UPLC-Xevo QTof MS system (WATERS, Manchester, UK), using the PLGS software, as previously described17. Difference in abundance among the groups was obtained using the Monte-Carlo algorithm in the ProteinLynx Global Server (PLGS) software and displayed as p < 0.05 for down-regulated proteins and 1 − p > 0.95 for up-regulated proteins. Bioinformatics analysis was done to compare the treated groups with the control group (Tables S1–S5), as previously reported17,18,19,20. The software CYTOSCAPE 3.0.4 (JAVA) was used to build networks of molecular interaction between the identified proteins, with the aid of ClueGo and ClusterMarker applications.

Results

Morphological analysis of the jejunum wall thickness

The mean (±SD) thickness of the jejunum tunica muscularis was significantly higher in the 50 mgF/L (93.0 ± 1.4 µm2) when compared to control (81.5 ± 1.1 µm2) and 10 mgF/L (84.2 ± 2.5 µm2) groups (Bonferroni’s test, p < 0.05). The total thickness of the jejunum wall was significantly lower in the 50 mgF/L (742.25 ± 7.8 µm2) and 10 mgF/L (734.4 ± 11.8 µm2) when compared to control (783.15 ± 5.8 µm2) (Bonferroni’s test, p < 0.05).

Myenteric HuC/D – IR neurons analysis

When the general population of neuron was morphometrically analyzed, the cell bodies areas (µm2) of the HuC/D–IR neurons were not significantly different among the groups (p > 0.05). In the semi-quantitative analyses (neurons/cm2), a significant decrease in the density was observed in the treated groups when compared with control (p < 0.05) (Table 1).

Table 1 Means and standard errors of the values of the cell bodies areas and density of HUC/D-IR and nNOS-IR neurons and VIP-IR, CGRP-IR, and SP-IR values of myenteric neurons varicosities areas of the jejunum of rats chronically exposed or not to fluoride in the drinking water.
Full size table

Myenteric nNOS –IR neurons analysis

The cell bodies areas (µm2) of the nNOS-IR neurons did not present a significant difference among the groups (p > 0.05) in the morphometric analysis. As for the semi-quantitative analyses, a decrease in the mean value of the density for the group treated with 50 mgF/L when compared with the other groups was observed (p < 0.05; Table 1).

Myenteric varicosities VIP-IR, CGRP-IR or SP-IR morphometric analysis

A significant increase in the VIP-IR varicosity areas (µm2) was detected in the group treated with 50 mgF/L when compared with the control group (p < 0.05). For the CGRP-IR varicosity areas, the groups did not differ significantly (p > 0.05). However, SP-IP varicosity areas were significantly increased in the treated groups when compared with control. In addition, the group treated with 10mgF/L presented an area significantly higher than the group treated with 50 mgF/L (Table 1).

Typical images of the immunofluorescences are shown in Figs 1 and 2.

Figure 1
figure 1

Photomicrography of myenteric neurons of the rats jejunum stained for HuC/D (green), nNOS (red), and double-labeling (HuC/D and nNOS) for the control group (0 mgF/L) and for the groups treated with 10 and 50 mgF/L. 20x Objective.

Full size image
Figure 2
figure 2

Photomicrography of myenteric varicosities of the rats jejunum after F chronic exposure (0, 10 or 50 mgF/L) for VIP-IR, SP-IR CGRP-IR. 40x Objective.

Full size image

Proteomic analysis of the jejunum

The total numbers of proteins identified in the control, 10 and 50 mgF/L groups were 294, 343 and 322, respectively. These proteins were present in the 3 pooled samples for each group. Among them, 81 (Table S1), 120 (Table S2) and 99 (Table S3) proteins were uniquely identified in the control, 10 mgF/L and 50 mgF/L groups, respectively. In the quantitative analysis of the 10 mgF/L vs. control group, 30 proteins with change in expression were detected (Table S4). As for the comparison 50 mgF/L vs. control group, 40 proteins with change in expression were found (Table S5). Most of the proteins with changed expression were upregulated in the groups treated with F when compared with the control group (21 and 23 proteins in the groups treated with 10 mgF/L and 50 mgF/L, Tables S4 and S5, respectively).

Figures 3 and 4 show the functional classification according to the biological process with the most significant term, for the comparisons 10 mgF/L vs. control and 50 mgF/L vs. control, respectively. The group exposed to the highest F concentration had the largest alteration, with change in 15 functional categories (Fig. 4). Among them, the categories with the highest percentage of associated genes were: Cellular respiration (14.3%), NAD metabolic process (10.2%), Oxygen transport (10.2%), Chromatin silencing (8.2%) and ER-associated ubiquitin-dependent protein catabolic process (8.2%). Exposure to the lowest F concentration influenced 12 functional categories (Fig. 3). The biological processes with the highest percentage of affected genes were: Nicotinamide nucleotide metabolic process (25%), Regulation of neuronal synaptic plasticity (11.4%), NAD metabolic process (15.9%) and Positive regulation of response to wounding (9.1%). It should be highlighted that Regulation of oxidative stress-induced intrinsic apoptotic signaling pathway was also identified, with 4.5% of affected genes (4.5%).

Figure 3
figure 3

Functional distribution of proteins identified with differential expression in the jejunum of rats exposed to the chronic dose of 10 mgF/L vs. Control Group (0 mgF/L). Categories of proteins based on GO annotation Biological Process. Terms significant (Kappa = 0.04) and distribution according to percentage of number of genes association. Proteins access number was provided by UNIPROT. The gene ontology was evaluated according to ClueGo® pluggins of Cytoscape® software 3.4.089,90.

Full size image
Figure 4
figure 4

Functional distribution of proteins identified with differential expression in the jejunum of rats exposed to the chronic dose of 50 mgF/L vs. Control Group (0 mgF/L). Categories of proteins based on GO annotation Biological Process. Terms significant (Kappa = 0.04) and distribution according to percentage of number of genes association. Proteins access number was provided by UNIPROT. The gene ontology was evaluated according to ClueGo® pluggins of Cytoscape® software 3.4.089,90.

Full size image

Figures 5 and 6 show the subnetworks created by CLUSTERMARK for the comparisons 10 mgF/L vs. control and 50 mgF/L vs. control, respectively. For the 10 mgF/L group (Fig. 5), most of the proteins with change in expression interacted with Solute carrier family 2, facilitated glucose transporter member 4 (GLUT4; P19357) and Small ubiquitin-related modifier 3 (Q5XIF4) (Fig. 5A) or with Polyubiquitin-C (Q63429) and Elongation factor 2 (P05197) (Fig. 5B). As for the group treated with 50 mgF/L, most of the proteins with change in expression interacted with GLUT4 (P19357) and Mitogen-activated protein kinase 3 (MAPK3; P21708) (Fig. 6A) or Polyubiquitin-C (Q63429) (Fig. 6B).

Figure 5
figure 5

Subnetworks created by ClusterMark® to establish the interaction between proteins identified with differential expression in the 10 mgF/L group in relation to the control group. The color of the nodes indicates the differential expression of the respective named protein with its access code. The dark red and dark green nodes indicate proteins unique to the control and 10 mgF/L groups, respectively. The nodes in gray indicate the interaction proteins that are offered by CYTOSCAPE®, which were not identified in the present study and the light red and light green nodes indicate downregulation and upregulation, respectively. In (A), the access numbers in the gray nodes correspond to: Dynein light chain 1, cytoplasmic (P63170), Poly [ADP-ribose] polymerase 1 (P27008), E3 ubiquitin-protein ligase RNF4 (O88846), Small ubiquitin-related modifier 3 (Q5XIF4), Nischarin (Q4G017), Heterogeneous nuclear ribonucleoprotein K (P61980), Peroxisomal bifunctional enzyme (P07896), Lethal(2) giant larvae protein homolog 1 (Q8K4K5), Rab GDP dissociation inhibitor alpha (P50398) and Solute carrier family 2, facilitated glucose transporter member 4 (P19357). The access numbers of the unique proteins of the control (dark red nodes) correspond to: Aconitate hydratase, mitochondrial (Q9ER34), Cytochrome c oxidase subunit 4 isoform 1, mitochondrial (P10888) and NADH dehydrogenase [ubiquinone] flavoprotein 2, mitochondrial (P19234). The accession numbers of the unique 10 mgF/L (dark green nodes) proteins correspond to: Aspartyl aminopeptidase (Q4V8H5), Ras-related protein Rab-1B (P10536), Vigilin (Q9Z1A6), Ras-related protein Rab-12 (P35284), Ras-related protein Rab-10 (P35281), Triosephosphate isomerase (P48500), Annexin A2 (Q07936) and Ras-related protein Rab-14 (P61107). The access numbers of the downregulated proteins (light red nodes) correspond to: Malate dehydrogenase, mitochondrial (P04636), Glutathione S-transferase P (P04906) and Histone H4 (P62804). The accession numbers of the upregulated proteins (light green nodes) correspond to: L-lactate dehydrogenase A chain (P04642). In (B), the access numbers in the gray nodes correspond to: Protein Svil (D3ZEZ9), Polyubiquitin-C (Q63429), Apoptosis-inducing factor 1, mitochondrial (Q9JM53), Protein deglycase DJ-1 (O88767), RAC-beta serine/threonine-protein kinase (P47197), RAC-alpha serine/threonine-protein kinase (P47196) and Gap junction alpha-1 protein (P08050). The access numbers of the unique proteins of the control (dark red nodes) correspond to: Vinculin (P85972), Eukaryotic initiation factor 4A-II (Q5RKI1), Malate dehydrogenase, cytoplasmic (O88989) and 40 S ribosomal protein S10 (P63326). The accession numbers of the single 10 mgF/L (dark green nodes) proteins correspond to: Elongation factor 2 (P05197) and Sodium- and chloride-dependent GABA transporter 3 (P31647). The access numbers of the downregulated proteins (light red nodes) correspond to: Peptidyl-prolyl cis-trans isomerase A (P10111).

Full size image
Figure 6
figure 6

Subnetworks created by ClusterMark® to establish the interaction between proteins identified with differential expression in the 50 mgF/L group in relation to the control group. The color of the nodes indicates the differential expression of the respective named protein with its access code. The dark red and dark green nodes indicate proteins unique to the control and 50 mgF/L groups, respectively. The nodes in gray indicate the interaction proteins that are offered by CYTOSCAPE®, which were not identified in the present study and the light red and light green nodes indicate downregulation and upregulation, respectively. In (A), the access numbers in the gray nodes correspond to: PTEN induced putative kinase 1 (Predicted) (D3Z9M9), Mitogen-activated protein kinase 3 (P21708), T-complex protein 1 subunit beta (Q5XIM9), Gap junction alpha-1 protein (P08050), Cartilage oligomeric matrix protein (P35444), Acid ceramidase (Q6P7S1), Integrin alpha-7 (Q63258), Stress-70 protein, mitochondrial (P48721), Solute carrier family 2, facilitated glucose transporter member 4 (P19357), Histone H3.1 (Q6LED0), Heterogeneous nuclear ribonucleoprotein K (P61980), Ankyrin-3 (O70511), Plectin (P30427) and Ankyrin-3 (O70511-7). The access numbers of the unique proteins of the control (dark red nodes) correspond to: Malate dehydrogenase, cytoplasmic (O88989), 40 S ribosomal protein S10 (P63326), Eukaryotic initiation factor 4A-II (Q5RKI1) and Vinculin (P85972). The accession numbers of the unique 50 mgF/L (dark green nodes) proteins correspond to: Tektin-2 (Q6AYM2), Mitochondrial fission 1 protein (P84817) and Paralemmin-1 (Q920Q0). The access numbers of the downregulated proteins (light red nodes) correspond to: Histone H2A type 2-A (P0CC09) and Histone H2A.Z (P0C0S7). The accession numbers of the upregulated proteins (light green nodes) correspond to: Hemoglobin subunit beta-1 (P02091) and Hemoglobin subunit alpha-1/2 (P01946). In (B), the access numbers in the gray nodes correspond to: Polyubiquitin-C (Q63429), Protein Svil (D3ZEZ9) e Glucocorticoid receptor (P06536). The access numbers of the unique proteins of the control (dark red nodes) correspond to: Heat shock protein 75 kDa, mitochondrial (Q5XHZ0), Cytochrome c oxidase subunit 4 isoform 1, mitochondrial (P10888) and NADH dehydrogenase [ubiquinone] flavoprotein 2, mitochondrial (P19234). The accession numbers of the single 50 mgF/L (dark green nodes) proteins correspond to: Mitogen-activated protein kinase 4 (Q63454), Synaptic vesicle membrane protein VAT-1 homolog (Q3MIE4), ATP-dependent 6-phosphofructokinase, liver type (P30835), Tissue alpha-L-fucosidase (P17164) and Peroxiredoxin-6 (O35244). The access numbers of the downregulated proteins (light red nodes) correspond to: ATP synthase subunit beta, mitochondrial (P10719) and e Histone H2A type 1-F (Q64598). The accession numbers of the upregulated proteins (light green nodes) correspond to: Tubulin beta-2A chain (P85108).

Full size image

Discussion

The small intestine is responsible for absorption of around 70–75% of F5,21. As consequence, gastrointestinal symptoms, such as abdominal pain, nausea, vomiting and diarrhea, are the most common occurrence in cases of excessive ingestion of F22,23,24,25. The mechanisms underlying these changes remain to be determined. Recently, our group took advantage of immunofluorescence and proteomics techniques to evaluate changes in the duodenum of rats after chronic exposure to F13. The group treated with 50 mgF/L had a significant decrease in the density of nNOS-IR neurons. Additionally, important morphological changes were seen in HUC/D-IR and nNOS-IR neurons, as well as in VIP-IR, CGRP-IR, and SP-IR varicosities for the groups treated with both 10 and 50 mgF/L. Moreover, profound proteomic alterations were observed in both treated groups. In the group treated with 10 mgF/L, most of the proteins with altered expression were upregulated. On the other hand, downregulation of several proteins was found in the group treated with the highest F concentration13.

Many proteins observed in the previous study were correlated with the neurotransmission process, which is essential for the function of the GIT through ENS control. For example, the pattern of intestinal smooth muscle contraction can be modified when the release of neurotransmitters stimulating muscle contraction, such as SP26 is increased or when the release of neurotransmitters promoting muscle relaxation, such as NO27, is decreased. In the present study, both conditions might have occurred, because we found a significance increase and decrease in the mean values of the SP varicosities area and the density of nNOS-IR neurons, respectively (Table 1), which is in accordance with our previous findings for the duodenum13. This finding can be also associated with the significant decrease in the density of HUC/D-IR neurons (Table 1), and it could contribute to the intestinal discomfort and symptoms, such as abdominal pain and diarrhea, observed upon excessive exposure to F.

Another important neurotransmitter that also participates in the control of intestinal motility is VIP. In our study, it was observed a statistically significant increase in the mean value of the areas of VIP-IR myenteric varicosities in the 50 mgF/L group when compared with control. This finding is similar to what was observed in our previous study where duodenum was analyzed13 and confirms that this dose of F can compromise the vipergic innervation of the small intestine. For the inhibitory control of motility, the main neurotransmitters involved are NO and VIP28, so basically any changes in the vipergic innervation can alter the intestinal motility, leading to a decrease in the tone of the intestinal smooth muscle, which could trigger diarrhea or even increased susceptibility to intestinal infections by decreased intestinal transit29. We can also suggest, in this case, that this increase may mean upregulation in the expression of the VIP, as a response to F toxicity since other processes such as axotomy and blocking of axonal transport or hypertrophic alterations promote upregulation of VIP in enteric neurons30. This increase can also be related to a neuroprotective role of VIP, because it acts as a potent anti-inflammatory molecule and presents an important antioxidant activity31,32,33. In addition to this, VIP is one of the most important elements involved in enteric neuroplasticity33, which is the ENS ability to adapt to any change in its microenvironment34. Due to the morphological changes that we observed in our study in the vipergic varicosities, we can suggest that F can induce important neuroplastic changes in the GIT.

Since alterations in the morphology of the intestinal wall infer important pathophysiological processes, we analyzed the total thickness of the intestinal wall, as well as the tunica muscularis separately. The group treated with 50 mgF/L presented a significant decrease in the total thickness of the intestinal wall and an increase in the thickness of the tunica muscularis, indicating that F can alter morphologically the jejunum wall. The finding for the tunica muscularis of the jejunum is in-line with our previous findings for the duodenum13, despite the total thickness of the duodenum wall was not altered. Changes in the number and morphology of myenteric cell bodies may be related to variations of the tunica muscularis thickness, which presents the structures responsible for the maintenance, development and plasticity of these neurons35. Similar increase in the thickness of the intestinal wall and tunica muscularis have been reported in the duodenum and jejunum of rats fed with a high fat diet for 8 weeks, where morphological alterations in the general population of enteric neurons and in the nitrergic population were also detected36, emphasizing that intestinal physiology comprises many interconnected mechanisms.

As in our study F caused morphological alterations in different enteric neuronal subtypes, which present several neurotransmitters involved in the GIT motility, it is possible that these alterations affect the GIT function, and promote the important symptomatology of F toxicity on the GIT, such as abdominal pain and diarrhea. We also believe that our results are quite relevant regarding the ENS, since mechanisms of neurodegeneration associated to enteric neuropathies are characterized basically by alterations, damage or loss of enteric neurons, as observed in several important pathologies37 and also in our study. Thus, in order to better investigate these findings involving the enteric innervation, we performed the proteomic analysis.

The proteomic approach revealed for both F doses that the majority of the proteins presenting changed expression interacted with Solute carrier family 2, facilitated glucose transporter member 4 (GLUT4) (P19357) and Polyubiquitin C (Q63429). In the network comparing 10 mgF/L vs. control groups, 17 members of the Ras-related Rab proteins (isoforms 1A, 1B, 3A, 3C, 3D, 4A, 4B, 5A, 8A, 8B, 10, 12, 14, 26, 35, 37, and 43) were uniquely found in the group treated with 10 mgF/L (Table S2), despite some not being present in the network. The GTPases Rab proteins are known as key regulators of intracellular membrane trafficking, from the formation of transport vesicles to their fusion with membranes. Rabs modulate between an inactive form (GDP-bound) and an active form (GTP-bound). The latter can attract to membranes distinct downstream effectors that will lead to vesicle formation, movement, tethering and fusion (UNIPROT). Generally, many studies report Rab proteins as molecules present in the CNS and their specific roles. Although marked differences distinguish the neuronal function between the ENS and CNS, their similarities allow the use of some principles established for the brain environment to be reapplied in the enteric context38. Several cellular processes can be altered and promote the enteric neuronal alterations caused by F effects through mechanisms involving the Rab proteins, which are considered neuronal regulators involved in the traffic and signaling of different molecules that promote neurons homeostasis, such as the neurotrophins family of growth factors. The neurotrophins-receptors complexes trigger important signaling pathways that promote development, survival and other neuronal functions through intracellular transport mechanisms mediate by the Rab proteins39.

Rab 1A is a regulator of specific vesicular trafficking from the ER to Golgi complex, and in dopaminergic neurons its expression presents a protective effect enhancing the control of motor function in surviving neurons of hemiparkinsonian animals40. From the family of the Rab 3 proteins, 3 members were present in the 10 mgF/L group, Rab 3A, Rab 3C, and Rab 3D. The Rab 3 family is observed in different cell types with high exocytic function41, in which they function as exocytosis regulators42 correlated with neuronal traffic39, and are present in synaptic vesicles, modulating the neurotransmitter release42. Rab 3A is the most abundant isoform in the brain, where it presents a modulatory function in synaptic membrane fusion through a Ca2+-dependent manner43. In the peripheral nervous system Rab 3A has increased expression in sciatic nerve lesion area associated to an increase in the expression of two other important proteins that contribute to neurotransmission, synaptophysin and synapsin I44. Rab 3C is highly expressed in primary hippocampal neurons, mediating regulated exocytosis45, while Rab 3D is present in secretory granules and vesicles of other cell types, such as adipocytes, exocrine glands, hematopoietic cells46, and low levels of expression were already identified in the duodenum, confirming its presence in exocrine cells of the GIT47.

The Rabs 4A and 4B were also identified as exclusive for the 10 mgF/L, and Rab4 is described as a regulator of early endosomes in the synapses, contributing to neurotransmitter receptor recycling through endosomes acting associated to other molecules in the later steps of the endocytic recycling pathway in dendrites, directing the neuronal membrane receptor trafficking48. This process is extremely important for the delivery of neurotransmitter receptors into the synaptic membrane, determining the synaptic function and plasticity. Rab 5A presents a role in axonal and dendritic endocytosis, contributing to the biogenesis of synaptic vesicles49. Rab 8 presents the same role as Rab 4, being required to direct into synapses neurotransmitter receptors as the AMPA-type glutamatergic receptors, presenting an important role in the control of synaptic function and plasticity at the postsynaptic membrane50.

Rab 10 is required for the secretion of neuropeptides through the release of dense core vesicles, which is a mechanism that modulates neuronal activity51. It is also a regulator of membrane trafficking during dendrite morphogenesis, and loss of RAB 10 decreases proximal dendritic arborization in the multi-dendritic PVD neurons52. In the CNS Rab 12 is colocalized with M98K, and overexpression of the latter induces cell death in retinal glial cells, while knockdown of Rab 12 reduces M98K-induced cell death in the same cells through the autophagy mechanism53.

Rab 26 promotes in the brain the formation of clusters of vesicles in neuritis54, and the authors suggest a new mechanism for degradation of synaptic vesicles in which Rab 26 selectively conducts synaptic and secretory vesicles into preautophagosomal structures. In neuronal immortalized cells, Rab 35 promotes neurite differentiation and favors axon elongation in rat primary neurons in an activity-dependent manner55.

The fact that several members of the Rab proteins were expressed exclusively in the 10 mgF/L group might indicate that this F concentration could affect the neuronal functions, since different Rab proteins regulate distinct processes in the neuronal environment. Since the 10 mgF/L concentration caused a decrease in the enteric neuronal density, which can compromise the enteric neuronal activity, the expression of several Rab proteins can reflect an attempt to keep the neurotransmission unaltered in the presence of F. Besides the neuronal activity, other important biological mechanisms involve the Rab proteins action. In the network comparing 10 mgF/L vs. control groups, the isoforms 1B, 10, 12 and 14 interact with GLUT4, and especially Rab 10 and Rab 14, are required in GLUT4 translocation to the plasma membrane56,57. Their increased expression might help to explain the increased sensitivity to insulin recently reported to occur in rats with diabetes induced by streptozotocin exposed to 10 mgF/L in the drinking water58. The increased expression of Rab 10 and Rab 14 might facilitate glucose uptake. Rab 37 and Rab 3A, also present among the proteins exclusively expressed in the 10 mgF/L group, are involved in the insulin release. Rab 3A has an important role in the hormone release from pancreatic β-cells with a regulatory control on insulin-containing secretion59. Rab 37, with a high sequence homology with Rab 3A, has also been reported to participate in regulated secretion in mammalian cells in the control of insulin exocytosis through a different mechanism of Rab 3A60. According to the authors, impairment of Rab 37 expression may contribute to abnormal insulin release in pre-diabetic and diabetic conditions. We can infer that the expression of both proteins indicates that the insulin release mechanism could be altered with this F dose. We also observed an increase in L-lactate dehydrogenase A chain (LDH) (P04642) upon exposure to 10 mgF/L. This enzyme converts pyruvate to lactate with regeneration of NADH into NAD+. It is an alternative way to supply the lack of oxygen for aerobic oxidation of pyruvate and NADH produced in glycolysis61. In fact, the categories nicotinamide nucleotide metabolic process and NAD metabolic process were among the ones with the highest percentage of affected genes when the 10 mgF/L group was compared with control. Previous studies have reported increase in the LDH activity in the serum of infants who consumed water containing more than 2 mgF/L62. It was also overexpressed in the brain of rats treated with F63. When pyruvate is converted into lactate by LDH, less pyruvate is available to enter into the mitochondria and form acetyl-CoA, which is consistent with the reduction of Malate dehydrogenase, mitochondrial (P04636) and of enzymes related to the oxidative phosphorylation, such as Cytochrome c oxidase subunit 4 isoform 1, mitochondrial (P10888) and NADH dehydrogenase [ubiquinone] flavoprotein 2, mitochondrial (P19234). According to Barbier, et al.3, F has an inhibitory effect on the activity of citric acid cycle enzymes, in agreement with our finding of reduction in Malate dehydrogenase, mitochondrial. Another protein with altered expression (downregulation) in the group treated with 10 mgF/L that interacts with GLUT4 was Glutathione S-transferase P (P04906) that was also found downregulated in the duodenum of rats treated with the same dose of F13. This enzyme is involved in the metabolism and detoxification of xenobiotics64. Many proteins with altered expression in the network comparing 10 mgF/L vs. control groups interact with Polyubiquitin C (Q63429), a highly conserved polypeptide that is covalently bound to other cellular proteins to signal processes such as protein degradation, protein/protein interaction and protein intracellular trafficking65. Among them are proteins related to translation, that were absent in the group treated with 10 mgF/L, such as Eukaryotic initiation factor 4A-II (Q5RKI1) and 40 S ribosomal protein S10 (P63326). The latter was also reduced in the group treated with 50 mgF/L both in the present study and in a previous study where duodenum was analyzed13. In addition, Peptidyl-prolyl cis-trans isomerase A (P10111) was reduced in the group treated with 10 mgF/L compared to control, which might impair protein folding. Also involved in protein synthesis, Elongation factor 2 (P05197) presented altered expression upon exposure to 10 mgF/L. This protein was present only in the group treated with 10 mgF/L, and catalyzes the GTP-dependent ribosomal translocation step during translation elongation (UNIPROT). Differences in expression of all these proteins indicate alterations in distinct steps of protein synthesis upon exposure to 10 mgF/L. Changes in protein synthesis might help to explain the alterations in the thickness of the jejunum wall observed in this group. Interestingly, Elongation factor 2 interacted with two of the 3 isoforms of the protein kinase AKT, namely RAC-alpha serine/threonine-protein kinase (AKT1; P47196) and RAC-beta serine/threonine-protein kinase (AKT2; P47197) that mediate protein synthesis and glucose metabolism66.

In the network comparing the 50 mgF/L vs. control groups (Fig. 6), some proteins with relevance for the neuronal homeostasis were expressed uniquely in the 50 mgF/L, such as Tektin-2 (Q6AYM2), Perforin-1 (Q5FVS5), and Mitochondrial fission 1 protein (Fis1-P84817). The Tektins family has significant expression in adult brain and in embryonic stages of the choroid plexus, the forming retina, and olfactory receptor neurons, and can be considered a molecular target for the comprehension of neural development67. Although not present in the subnetwork, Perforin participates in the CD8+ T cells response, promoting granule cytotoxicity leading to a fast cellular necrosis of the target cell in minutes68 or apoptosis in a period of hours through a mechanism in which the target cell collaborates with perforin to deliver granzymes into the cytosol69. Using these mechanisms perforin-dependent, CD8+ T cells promote neuronal damage in inflammatory CNS disorders70.

Mitochondrial fission is implicated in the cell death through a pathway that involves caspase activation71, and Mitochondrial fission 1 protein (Fis1) is considered essential for mitochondrial fission72. Overexpression of Fis1 caused increase of mitochondrial fragmentation, which conducted to apoptosis or triggered autophagy73,74, and neuroprotective effects are correlated with inhibition of Fis175.

The fact that these proteins presented increased expression in relation to the control group can reflect F neurotoxicity on the ENS with the concentration of 50 mgF/L, and could result in the decrease in the density of the general population of neurons since these 3 proteins are involved in pathways that conduct to cell death by distinct mechanisms.

Other proteins with altered expression interacted mainly with GLUT4 (P19357) and Polyubiquitin C (Q63429), which was also observed for the network comparing the 10 mgF/L vs. control groups (Fig. 5). In addition, Mitogen-activated protein kinase 3 (MAPK3; P21708) was also an interacting partner as in the duodenum of rats treated with the same concentration of F in the drinking water13. Among the proteins that interacted with GLUT4, Peroxiredoxin-6 (O35244) was present only in the group treated with 50 mgF/L, when compared with control (Fig. 6). This enzyme, located in the cytoplasm, protects cells against oxidative stress, in addition to modulating intracellular signaling pathways. Peroxiredoxins catalyze the reduction of H2O2 and hydroxyperoxide in water and alcohol76. Thus, changes in these proteins expression could be linked to fluoride-induced oxidative stress that has been extensively described in the literature3,77,78,79,80,81. In the group treated with 50 mgF/L, there was a remarkable downregulation in several isoforms of Histones, in comparison with control (Fig. 6 and Table S5). The major role described for histones is DNA “packaging”, however, it is also well described that these proteins confer variations in chromatin structure to ensure dynamic processes of transcriptional regulation in eukaryotes82. Epigenetic modifications of DNA and histones are fundamental mechanisms by which neurons adapt their transcriptional response to developmental and environmental factors. Modifications in the chromatin of neurons contribute dramatically to changes in the neuronal circuits, and it is possible that histone activity is involved in disorders that compromise neuronal function83. Thus, changes in the expression of histones might have contributed to the alterations found in the morphology of enteric neurons in response to F exposure. In addition, structural muscle proteins such as different isoforms of actin and myosin were increased or exclusive in the group treated with 50 mgF/L (Tables S3 and S5), which helps to explain the increase in the thickness of the jejunum tunica muscularis.

Probably the most remarkable finding of the present study was that when the groups treated with 10 and 50 mgF/L are compared with control, some proteins related to energetic metabolism presented similar alterations in expression, regardless the dose of F, such as: Cytochrome c oxidase subunit 4 isoform 1, mitochondrial (P10888), NADH dehydrogenase [ubiquinone] flavoprotein 2, mitochondrial (P19234), Malate dehydrogenase, mitochondrial (P04636), Malate dehydrogenase, cytoplasmic (O88989) and L-lactate dehydrogenase A chain (P04642). The absence of Malate dehydrogenase, mitochondrial (P04636), Malate dehydrogenase, cytoplasmic (O88989), that form oxaloacetate, absence of NADH dehydrogenase [ubiquinone] flavoprotein 2, mitochondrial (P19234) that transfers electrons from NADH to respiratory chain in both groups treated with F, as well as the reduction of ATP synthase subunit beta, mitochondrial (P10719) (only in the group treated with the highest F dose), as well as the increase in L-lactate dehydrogenase A chain (P04642) in both groups treated with F indicate an increase in anaerobic metabolism in attempt to obtain energy, since aerobic metabolism is impaired in the presence of F. However, the rate of production of ATP through anaerobic pathways is much lower than that of aerobic pathways, which is in-line with the reports of reduction in the production of ATP induced by exposure to high F doses3,84. It is important to highlight that these changes in the expression of proteins associated to energy metabolism induced by exposure to 10 and 50 mgF/L in the drinking water are more pronounced than those observed previously in other organs exposed to roughly the same doses of F58,63,80,85,86,87,88. This might be due to the fact that the small intestine is responsible for the absorption of around 75% of ingested F5, which makes the cells of the intestinal wall exposed to higher doses of F than the cells from the other organs.

In conclusion, chronic exposure to F, especially to the highest concentration evaluated, increased the thickness of the tunica muscularis and altered the pattern of protein expression. Extensive downregulation of several isoforms of histones might have contributed to the alterations found in the morphology of enteric neurons in response to F exposure. Additionally, changes in proteins involved in energy metabolism indicate a shift from aerobic to anaerobic metabolism upon exposure to the highest F concentration. These findings provide new insights into the mechanisms involved in F toxicity in the intestine.