Enteric innervation combined with proteomics for the evaluation of the effects of chronic fluoride exposure on the duodenum of rats

Ingested fluoride (F) is absorbed mainly in the small intestine, which is controlled by the Enteric Nervous System (ENS). Although important intestinal symptomatology has been described after excessive F exposure, there have been no studies reporting the effects of F on the ENS. In this study, the effects of chronic F exposure were evaluated on the duodenums of rats through proteomic and morphological analyses. Concentrations of 0, 10, or 50 ppm of F were applied to the drinking water for 30 days. Immunofluorescence techniques were performed in the myenteric plexus of the duodenum to detect HuC/D, neuronal nitric oxide (nNOS), vasoactive intestinal peptide (VIP), calcitonin gene related peptide (CGRP), and substance P (SP). The 50 ppm F group presented a significant decrease in the density of nNOS-IR neurons. Significant morphological alterations were also observed in HUC/D-IR and nNOS-IR neurons; VIP-IR, CGRP-IR, and SP-IR varicosities for both groups (10 and 50 ppm F). Proteomic analysis of the duodenum demonstrated alterations in the expression of several proteins, especially those related to important biological processes, such as protein polymerization, which helps to explain the downregulation of many proteins upon exposure to 50 ppm of F.

Histological analysis. The intestinal lumen was washed, fixed in 10% neutral-buffered formalin, embedded in paraffin, cut into 5 µm sections and stained with haematoxylin and eosin (HE) using a standard technique. Images were obtained (10X objective) using a high resolution camera (Moticam 2500, Motic China Group Co, Shanghai, China) coupled to a microscope (Olympus BX40, Olympus Co., Japan). The thickness of the tunica muscularis and of the total wall of the duodenum were measured using Image-Pro Plus software in 10 sections per animal in 5 different regions (images not provided).
Myenteric plexus immunofluorescence. After laparotomy the duodenum was washed with 0.1 M PBS (pH 7.4) and filled with Zamboni's fixative 28 for 18 h at 4 °C. This segment was opened along the mesenteric border and washed in 80% ethanol to remove the fixative. The samples of duodenum were then dehydrated in an increasing ethanol series (95% and 100%), cleared with xylene, gradually rehydrated in a decreasing ethanol series (100%, 90%, 80%, 50%), and stored in PBS containing 0.08% sodium azide at 4 °C. The samples were cut into pieces measuring approximately 1 cm 2 and were microdissected under a Stemi DV4 stereo microscope (Zeiss, Jena, Germany) to obtain whole-mount preparations of the tunica muscularis (since study of the myenteric plexus requires removal of the mucosal and submucosal layers). The whole mounts were processed for immunofluorescence techniques to detect Human neuronal protein (HuC/D), Vasoactive Intestinal Peptide (VIP), Calcitonin gene-related peptide (CGRP), and Substance P (SP). The HuC/D was used, since it is an important marker of the general enteric neuronal population 29 . To identify the neurons that present Nitric Oxide (NO) as a neurotransmitter, an antibody for the neuronal Nitric Oxide (nNOS) was applied 15 , and the immunoreactive cell bodies were named HuC/D-IR and nNOS-IR, respectively. In the myenteric plexus cell bodies of neurons that present VIP are rarely found, but present varicose axons that also exhibits immunoreactivity to VIP. So, in the present study, the VIP neurons were identified by their varicosities (VIP-IR), as also CGRP and SP neurons (CGRP-IR and SP-IR, respectively). For the immunofluorescence techniques, the samples were processed as described by Hermes-Uliana et al. 30 . Briefly, the dissected samples were washed in PBS with 0.5% Triton X-100 and were incubated for 1 h in blocking solution (PBS with 0.5% Triton X-100, 2% BSA, and 10% donkey serum). For the double-labelling technique, the following primary antibodies, in determined concentrations, were applied for 48 hours: anti-HuC/D (mouse, 1:500), and anti-nNOS (rabbit, 1:500). After this first incubation period, the samples were washed 3    Morphometric and quantitative analysis. The quantitative analysis was performed for the HuC-D/ nNOS technique based on the neuronal density (neurons/cm 2 ) and the morphometric analysis based on the area (µm 2 ) of 100 neuronal cell bodies/animal. For the VIP, SP and CGRP techniques the morphometric analysis was performed in 400 varicosities from each animal (2,400/group). Images were obtained using a 20X microscope objective for the HuC-D/nNOS technique and a 40X objective for the VIP, SP and CGRP techniques. All images were analysed using the Image-Pro Plus ® software, version 4.5.0. 29  Sample preparation for proteomic analysis. The duodenum was harvested and frozen in liquid nitrogen. The frozen samples were homogenized in a cryogenic mill (model 6770, Spex, Metuchen, NJ, USA). Protein preparation for proteomic analysis was performed as previously described 31 . Briefly, proteins were extracted by incubation in lysis buffer (7 M urea, 2 M thiourea, 4% CHAPS, 1% IPG buffer pH 3-10, 40 mM DTT) for 1 h at 4 °C and then were precipitated using the PlusOne 2D Cleanup kit. Duodenum proteins (25 µL) from each animal of the same group were combined to constitute a pool. Then 50 mM AMBIC containing 3 M urea was added to each pool, the sample was filtered in 3 kDa AMICON and the protein was quantified using the Bradford method 31 . Samples were reduced (DTT), alkylated (IAA) and digested with 100 ng trypsin for 14 h at 37 °C. After digestion 10 µl of 5% TFA were added, the sample was incubated for 90 min at 37 °C and centrifuged (14,000 rpm for 30 min). The supernatant was harvested, and 5 µL of ADH (1pmol/µL) plus 85 µL of 3% ACN were added.

LC-MS/MS and bioinformatics analyses.
The peptide identification was performed on a nanoAcquity UPLC-Xevo QTof MS system (Waters Corporation, Manchester, UK), as previously described 32 . Differences in expression among groups were obtained using the ProteinLynx Global Server (PLGS) software provided by Waters Corporation and are expressed as p < 0.05 for downregulated proteins and 1 − p > 0.95 for upregulated proteins. Bioinformatics analysis was performed for comparison of the groups exposed to F relative to the control group (Tables S1-S6), as reported earlier [32][33][34][35] .

Results
Plasma F concentrations. The validation of the exposure was performed through analysis of the plasma F concentrations. Significant differences were found among the groups (F = 53.57, p < 0.0001; Table 1), confirming the exposure. Both groups exposed to F (10 and 50 ppm F) presented plasma F concentrations significantly higher than in the control group.
Morphological analysis of the duodenum wall thickness. The mean (±SD) thickness of the duodenum tunica muscularis was significantly increased in the 50 ppm F group (161.55 ± 3.73 µm 2 ) compared to control (129.65 ± 1.73 µm 2 ) and the 10 ppm F group (137.17 ± 2.24 µm 2 ) (p < 0.05). However, the total thickness of the duodenum wall did not significantly differ among the groups (data not shown).

Morphometric and quantitative analysis of myenteric neurons of the duodenum. Myenteric
neurons HuC/D-IR analysis. In the morphometric analysis of the general population of neurons, the cell body areas (µm 2 ) of the HuC/D-IR neurons of the duodenum presented a statistically significant difference among the groups (p < 0.05), with a decrease in the average value of the areas for the 50 ppm F group, compared to the control group (p < 0.05). In the quantitative analyses, no significant change was observed in the density of the HuC/D-IR neurons (p > 0.05, Table 2).
Myenteric nNOS-IR neuron analysis. In the morphometric analyses of the nNOS-IR neurons the values of the cell body areas showed the same pattern as the HuC/D-IR neurons, with a decrease in the mean value of the areas for the 50 ppm F group compared to the control group (p < 0.05). In the quantitative analyses, a significant reduction in the density of nNOS-IR neurons was found for the 50 ppm F group, compared to the control group (p < 0.05) ( Table 3).
Scientific RepoRts | 7: 1070 | DOI:10.1038/s41598-017-01090-y Myenteric varicosities VIP-IR, CGRP-IR, or SP-IR morphometric analysis. In the morphometric analyses of VIP-IR varicosity area (µm 2 ), a significant increase was detected in both groups exposed to F (10 and 50 ppm F), compared to the control group (p < 0.05 when all groups were compared) ( Table 4). In the morphometric analyses of the CGRP-IR varicosity areaa (µm 2 ), a significant increase was observed in the 10 ppm F group and a decrease was observed in the 50 ppm F group, compared to the control group (p < 0.05) ( Table 4). In the morphometric analyses of SP-IR varicosity areas (µm 2 ), a significant increase was observed in both groups exposed to F, compared to the control group (p < 0.05) ( Table 4). Representative images of the immunofluorescence are displayed in the supplementary information (Supplementary Figs S1 and S2).
Proteomic analysis of the duodenum. After chronic F exposure, totals of 699, 643, and 591 proteins were identified by mass spectrometry in the control, 10, and 50 ppm F groups, respectively. In the quantitative analysis of the 10 ppm F vs. control groups, 229 proteins with altered expression were observed (Supplementary Table S1), and the majority of these proteins were upregulated (143 proteins). For the 50 ppm F vs. control group comparison a total of 284 proteins with altered expression were identified (Supplementary Table S2), and the majority (270 proteins) presented downregulation. Totals of 220, 179 and 155 proteins were identified exclusively in the control, 10, and 50 ppm F groups, respectively (Supplementary Tables S3-S5). The functional classification according to the most affected biological processes is presented in Figs 1 and 2 for the comparisons of 10 ppm F vs. control and 50 ppm F vs. control, respectively. Exposure to 10 ppm F led to the most pronounced alterations, with changes in 43 functional categories (Fig. 1). Among them, the categories that presented the highest percentage of affected genes were the pyridine nucleotide metabolic process (41%), the carboxylic acid metabolic process (38%), the nicotinamide nucleotide metabolic process (36%), cellular component assembly (29%), myofibril assembly (24%), actomyosin structure organization (20%), and cytoskeleton organization (20%). Exposure to the highest F concentration affected 27 functional categories (Fig. 2). Among them, the categories with the highest percentage of associated genes affected were protein polymerization (33%), positive regulation of organelle organization (33%), actin filament organization (29%), the response to metal ions (23%), the response to inorganic substances (23%), the nicotinamide nucleotide metabolic process (19%), the purine ribonucleoside triphosphate metabolic   Table 2. Means and standard errors of the values of the cell bodies areas and density of the HUC/D-IR myenteric neurons of the duodenum of rats exposed or not to F chronically, through the drinking water for 30 days. Animal groups: Control (deionized water-0 ppm F), 10 ppm F, and 50 ppm de F. In the quantitative analysis the Fisher's test was applied but no significant difference was observed among the groups. In the morphometric analysis, means followed by different letters in the same column are statistically different according to Tukey's test (p < 0.05).  Table 3. Means and standard errors of the values of the cell bodies areas and density of the nNOS-IR myenteric neurons of the duodenum of rats exposed or not to F chronically, through the drinking water for 30 days. Animal groups: Control (deionized water-0 ppm F), 10 ppm F, and 50 ppm de F. In the quantitative analysis the Fisher's test was applied but no significant difference was observed among the groups. In the morphometric analysis, means followed by different letters in the same column are statistically different according to Tukey's test (p < 0.05).  Table 4. Means and standard errors of the VIP-IR, CGRP-IR, and SP-IR values of myenteric neurons varicosities areas of the duodenum of rats exposed or not to F chronically, through the drinking water for 30 days. Animal groups: Control (deionized water-0 ppm F), 10 ppm F, and 50 ppm de F. Means followed by different letters in the same column are statistically different according to Tukey's test (p < 0.05). n = 6.

GROUPS Area VIP-IR varicosities (µm 2 ) Area CGRP-IR varicosities (µm 2 ) Area SP-IR varicosities (µm 2 )
process (18%) and regulation of translation (18%). Figures 3 and 4 show the subnetworks generated by JActive Modules for the comparisons of 10 ppm F vs. control and 50 ppm F vs. control, respectively. When the animals were exposed to 10 ppm F (Fig. 3), most of the proteins with altered expression interacted with com Pleiotrophin (P63090) and Rab GDP dissociation inhibitor alpha (P50398), while some of them interacted with RNA-binding protein PNO1 (Q6VBQ8) and other ones with Mitogen-activated protein kinase 14 (MAPK14; P70618). Exposure to the highest F concentration caused alterations in many proteins, generating a complex interaction subnetwork (Fig. 4). In this case, many of the proteins with altered expression interacted with partners involved in general and specialized signalling networks such as

Discussion
Despite reported symptoms in cases of excessive F ingestion, such as nausea, vomiting, diarrhoea, and abdominal pain [38][39][40] , the mechanisms leading to these alterations in the GIT are largely unknown. To our knowledge, this study was the first to provide mechanistic insights on this topic, taking advantage of the combination of morphometric and proteomic analyses. The evaluation of enteric neurons is considered an important analysis of pathological effects on the enteric ganglia with the aim of more precisely identifying and describing disorders related to enteric innervation. In the ENS, morphological changes in the myenteric plexus can induce motor disorders 41 , can be correlated with inflammatory conditions 42 , or can reflect physiological processes, such as aging 43 .
An increase in the release of neurotransmitters that stimulate muscle contraction such as SP 44 , or a decrease in the release of neurotransmitters that promote muscle relaxation, such as NO 45 , can consequently modify the behaviour of the intestinal smooth muscle contraction. Both events might have occurred in the present study since we observed a significant increase and a significant decrease, respectively, in the average values of the SP varicosity area (Table 4) and the nNOS-IR neuron area, for example ( Table 3). As the enteric neurons have an essential role in the intestinal function, these alterations in their morphology and/or density observed in the present study could help to explain the intestinal discomfort and symptoms caused by F, which is also corroborated by the significant decrease in the area of HuC/D-IR cell body areas.
Some authors have described the correlation between an increase in F ingestion and elevated serum levels of NO with disturbances of NO metabolism in chicks 46 . Other important connections were presented between excessive F intake and NO production 47 and increased NOS activity in the brain 48 , leading to alterations of some mechanisms, including glutamate reuptake and lipid peroxidation 49 . The latter is extremely important in the development of oxidative stress 50 and degenerative diseases 51,52 . Specially the oxidative stress seems to be the mechanism involved in the neurodegeneration caused by F, and it has been found as a consequence of F toxicity in the brain and also in the intestine 47,48,53 . Our results offer evidence of the correlation between F intake and NO production in the enteric innervation of the duodenum since we observed that F could affect nNOS-IR myenteric neurons at the morphometric and quantitative levels. The results observed in this study could be explained by an alteration of NO production or nNOS activity, which occurs in the CNS 47 , since F is also an enzyme inhibitor and its interaction with enzymes has been described as its main cellular toxic effect 54 . Another important effect of NO on the myenteric plexus is the induction of VIP release 55 . This VIP release by the myenteric varicosities, in turn, induces the production of NO by nNOS-IR neurons 56,57 . This mechanism could explain the results in the present study, as a significant increase in the area of the VIP-IR varicosities was found for both F concentrations compared to the control, which could be an attempt to compensate for the reduction in NO production that might be characterized by a decrease in the area of the nNOS-IR neurons 30 . This result could also be explained by VIP action being complementary to NO in the control of gastrointestinal motility 58 by interacting with NO in the inhibition of smooth muscle contraction of the intestinal wall 59,60 .
Another possible explanation of our results lies in both plexi (myenteric and submucous) working synchronously during intestinal function 58 and the morphological alterations observed in our study in the myenteric varicosities VIP-IR could also reflect a mechanism involving VIP production in the submucous plexus, where its physiological role is correlated with immunological functions 61 and anti-inflammatory action 62 . Therefore, it is possible that F could affect VIP varicosities in the myenteric plexus through alterations in the submucous plexus as an part of inflammatory process caused by the ingestion of high F concentrations. Associated with these important findings of the morphological analysis, the proteomic approach revealed that the group exposed to 10 ppm of F, compared with control, presented most of the proteins with downregulated expression interacting with Pleiotrophin (P63090). Among these proteins were Fasciculation and elongation protein zeta-1 (P97577), Protein disulfide-isomerase (P04785), Guanine nucleotide-binding protein subunit beta-2-like 1 (P63245), Glycerol-3-phosphate dehydrogenase [NAD(+)]-cytoplasmic (O35077), and Glutathione S-transferase P (P04906). This connection is relevant to the present study since Pleiotrophin is a neurotrophic factor expressed during CNS development 63 ; its upregulation promotes peripheral nerve regeneration 64 and is also related to increased neuronal density with the aim of restoring brain function 65 . Its correlation with the CNS has also been described in other studies showing its involvement in cellular differentiation 66 and proliferation 67 , as well as postsynaptic specialization 63 . Because this protein plays an important role in neuronal homeostasis in the CNS and peripheral nervous system, downregulation of its interaction partners could indicate that the concentration of 10 ppm of F in drinking water could have a similar toxic effect on the ENS although this F concentration did not cause alterations in the density of the general population or the nitrergic neurons of the duodenum myenteric plexus.
In our network, CAP-1 interacted with Debrin-like protein (Q9JHL4) which plays a role in reorganization of the actin cytoskeleton, formation of cell projections in neuron morphogenesis, and synapse formation. In non-muscle cells, Tropomyosin alpha-1 chain is implicated in stabilizing cytoskeleton actin filaments. Annexin A2 is a calcium-regulated membrane-binding protein that binds to actin filaments. Also downregulated by F, Coronin-1A (Q91ZN1) is involved in cytoskeleton organization and rearrangement in neuronal cells 72 , and GTP-binding nuclear protein Ran (RAN; P62828) contributes to nucleocytoplasmic transport, RNA export, chromatin condensation, and cell cycle control (UNIPROT). Downregulation of all of these key proteins might disturb cell signalling and important cellular processes such as protein synthesis, cell motility and vesicle trafficking leading to intracellular accumulation of proteins. This protein concentration could be related to the results observed here, such as increased thickness of the tunica muscularis and increased average values of the enteric neuron and varicosity areas, and it could possibly be related to enteric neurotransmission production and release.
Also involved in cellular transport, Fasciculation and elongation protein zeta-1 (FEZ-1; P97577) was absent in the group exposed to 10 ppm of F and was downregulated by 50 ppm of F, compared with control. This protein is an adapter of the transport mediated by kinesins considered "motor" proteins, transporting not only molecules but also organelles through a process that is dependent on ATP and microtubules. Intracellular protein transport is essential for neuronal differentiation, gene expression and cytoskeleton rearrangement 73 . Supression of FEZ-1 by si-RNA affects mitochondrial motility and neuronal morphology 74 in the CNS, and its downregulation observed after F exposure could affect enteric neurons by compromising the intracellular transport.
Phosphatidylethanolamine-binding protein 1 (PEBP1; P31044), a synaptic signalling protein, was also downregulated by 50 ppm of F in the present study. This molecule is downregulated in several neurological conditions such as Alzheimer's 75 and other memory-related disorders 76 . Its localization at the synapse suggests that it may be a critical regulator of neuronal survival 77 . Downregulation of PEBP1 could be associated with cognitive impairment which has been suggested to occur under exposure to high doses of F 78,79 , despite this theory still being debated.
In summary, chronic exposure to the highest F concentration promoted in the duodenums of rats a significant increase in the thickness of the tunica muscularis and alteration in the protein expression profile. This alteration was more pronounced than in previous proteomic studies that employed similar F concentrations and that evaluated other organs such as the kidney 80-82 brain 83,84 , and liver 26,31 . This difference might have occurred because nearly 75% of ingested F is absorbed in the small intestine, especially in the proximal portions 4 . In addition, when F is ingested, the cells from the intestinal wall are directly exposed to higher F concentrations, which is different from other cell types distributed in several organs that will have contact only with absorbed F 85 . Upon exposure to the lowest F concentration, most of the proteins had upregulated expression (Table S1), while the opposite occurred for the highest F concentration ( Table S2). All of the data presented suggest an increase in protein synthesis upon exposure to low F doses and impairment of protein synthesis when high F doses are ingested, suggesting important alterations in many cellular processes due to exposure to F. These alterations might help to explain the important gastrointestinal symptoms reported in cases of excessive F exposure, especially those involving the ENS, since the morphological alterations observed on enteric neurons present a pattern that is similar to enteric neuropathies caused by important pathologies that affect GIT function.