Liver proteome of mice with different genetic susceptibilities to the effects of fluoride

ABSTRACT A/J and 129P3/J mice strains have been widely studied over the last few years because they respond quite differently to fluoride (F) exposure. 129P3/J mice are remarkably resistant to the development of dental fluorosis, despite excreting less F in urine and having higher circulating F levels. These two strains also present different characteristics regardless of F exposure. Objective In this study, we investigated the differential pattern of protein expression in the liver of these mice to provide insights on why they have different responses to F. Material and Methods Weanling male A/J and 129P3/J mice (n=10 from each strain) were pared and housed in metabolic cages with ad libitum access to low-F food and deionized water for 42 days. Liver proteome profiles were examined using nLC-MS/MS. Protein function was classified by GO biological process (Cluego v2.0.7 + Clupedia v1.0.8) and protein-protein interaction network was constructed (PSICQUIC, Cytoscape). Results Most proteins with fold change were increased in A/J mice. The functional category with the highest percentage of altered genes was oxidation-reduction process (20%). Subnetwork analysis revealed that proteins with fold change interacted with Disks large homolog 4 and Calcium-activated potassium channel subunit alpha-1. A/J mice had an increase in proteins related to energy flux and oxidative stress. Conclusion This could be a possible explanation for the high susceptibility of these mice to the effects of F, since the exposure also induces oxidative stress.


INTRODUCTION
A/J and 129P3/J mice strains have been widely studied over the last few years because they respond given the same dose of F, the A/J strain responds with a rapid onset and severe development of dental 8 . This was believed to be a consequence of the faster excretion of F by the 129P3/J strain. Surprisingly, a metabolic study showed that the 129P3/J mice excrete less F in urine, have higher circulating F levels and, consequently, higher bone F levels, however, they still are remarkably resistant 5 . Some differences between these strains are intrinsic to themselves and do not depend on the F exposure. For example, the A/J mice drink 129P3/J counterparts 4 , which can be explained by the increased expression of Alpha-aminoadipic semialdehyde dehydrogenase in the kidney of 129P3/J mice, regardless of F exposure. This enzyme metabolyzes irreversibly betaine aldehyde to betaine that is the most effective osmoprotectant accumulated by eukaryotic organisms to cope with osmotic stress 4 . In addition, exclusive proteins expressed in the kidney of A/J or 129P3/J mice This suggests that the genetic background per se accounts for such differences between these two strains of mice.
Liver represents the main detoxifying tissue in the body by processing, neutralizing, and eliminating toxins from the digestive tract through hepatocyte-mediated enzymatic detoxification systems. Due to these important functions, liver is one of the body's organs most subject to injury. Thus, it is believed that the differential pattern of protein expression in the liver of A/J and 129P3/J mice can provide new insights that could explain why they respond differently when exposed to F. To achieve this, state-of-the-art shotgun proteomics combined to bioinformatics approaches were used.

Animals and samples collection
Weanling male mice from the A/J and 129P3/J inbred strains (3-week-old; n=10 from each strain) were pared and housed in metabolic cages with ad libitum access to low-F food (AIN76A, PMI Nutrition, Richmond, IN, USA, 0.95 mg/Kg F) and deionized water for 42 days. The temperature and humidity in the climate-controlled room, which had a 12 h light/dark cycle, were 23±1°C and 40%-80%, respectively. All experimental protocols were approved by the Ethics Committee for Animal Experiments of Bauru School of Dentistry, University of São Paulo (Protocol # 031/2013). At the end of the study, the mice were anesthetized with ketamine/xylazine and livers were collected. Samples designated for proteomic analysis were stored at -80°C, while those designated for F analysis were stored at -20°C.

Fluoride analysis in liver
Fluoride analysis was done with the ion-sensitive electrode, after hexamethyldisiloxane-facilitated diffusion 22 , exactly as previously described 20 .

Statistical analysis
For liver F concentration, the GraphPad InStat software version 4.0 for Windows (GraphPad software Inc., La Jolla, California USA) was used. Data were analyzed by unpaired t test (p<0.05).

Sample preparation for proteomic analysis
Samples were prepared for analysis as previously described 17 . The frozen tissue was homogenized in a cryogenic mill (model 6770, Spex, Metuchen, NJ, EUA). For protein extraction, liver homogenate was incubated in lysis buffer containing 7 M urea, 2 M thiourea, 4% CHAPS, 1% IPG buffer pH 3-10, 40 mM DTT for 1 h at 4°C with occasional shaking. After this period, the homogenate was centrifuged at 15,000 rpm for 30 min at 4°C and the supernatant containing soluble proteins was recovered. The proteins were precipitated using the kit PlusOne 2D Cleanup (GE Healthcare, Uppsala, Sweden), as recommended by the manufacturer. Pellets were resuspended in rehydration buffer (7 M urea, 2 M thiourea, 0.5% CHAPS, 0.5% IPG buffer pH 3-10, 18 mM DTT, 0.002% bromophenol blue).
of the same group were combined to constitute a pool, 50 mM AMBIC, containing 3 M urea, were AMICON (Millipore, St Charles, MO, USA). Protein by Bradford protein assay 3 . To each sample (50 μg of total protein for each pool in a volume of 50 μL), 10 μL of 50 mM AMBIC were added. In sequence, 25 μL of 0.2% RapiGEST™ (Waters Co., Manchester, UK) were added and incubated at 80°C for 15 min. Following, 2.5 μL of 100 mM DTT were added and incubated at 60°C for 30 min. Also, 2.5 μL of 300 mM IAA were added and incubated for 30 min at room temperature (under dark). Then, 10 μL of trypsin (100 ng; Trypsin Gold Mass Spectrometry, Promega, Madison, USA) were added and digestion occurred for 14 h at 37°C. After digestion, 10 μl of 5 % TFA were added, incubated for 90 min at 37°C and the sample was centrifuged (14,000 rpm for 30 min). The supernatant was collected and 5 μL of ADH (1 pmol/μL) plus 85 μL 3% ACN were added.

LC-MS/MS and bioinformatics analyses
performed on a nanoAcquity UPLC-Xevo QTof MS system (Waters, Manchester, UK), exactly as previously described 15 . Difference in expression among the groups was obtained using PLGS software and expressed as p<0.05 for downregulated proteins 1-p>0.95 for up-regulated proteins ( Table 1). Bioinformatics analysis was performed, as reported earlier 1,15,[17][18][19] . Briefly, Uniprot protein ID accession numbers were mapped back to their associated encoding Uniprot gene entries for the comparison A/J X 129P3/J. Gene Ontology annotation of Broad Biological Process was performed using Cluego v2.0.7 + Clupedia v1.0.8, a Cytoscape plugin. Uniprot IDs were uploaded to Table 1 and analyzed with default parameters, which specify a Enrichment (right-sided hypergeometric test) correction method using Bonferroni step down, analysis mode "Function" and load gene cluster list for Mus musculus, Evidence Codes "All", set 8) and KappaScoreThreshold 0.03. The proteinprotein interaction network was downloaded from PSICQUIC, built in Cytoscape version 3.0.2 and constructed as proposed by Millan 18 (2013). A network was then created, providing global view of potentially relevant interacting partners of proteins whose abundances change.
differentially expressed proteins Table 1 shows proteins with expression changes in A/J and 129P3/J mice. In general, most proteins with fold change were increased in A/J mice. Figure 1 shows the functional classification according to the biological process with the most Among them, the category with the highest percentage of genes was oxidation-reduction process (20%), followed by cellular amino acid metabolic process (16%) and response to oxidative stress (12%).

Protein-protein interaction network
For the comparison displayed above, a network was created; employing all the interactions found in the search conducted using PSICQUIC. After the global network was created, nodes and edges were Mus musculus taxonomy (10090). The value of fold change and also the p-value were added in new columns. The ActiveModules 1.8 plug-in to Cytoscape was used to make active modules connected subnetworks within the molecular interaction network whose fold changes and p-value, as shown in the original proteomic analysis. Figure 2 shows the subnetwork generated by VizMapper. As can be seen, most proteins with fold change present interaction with Disks large homolog 4 (Q62108; 11 proteins) and Calcium-activated potassium channel subunit alpha-1 (Q08460; 18 proteins).

DISCUSSION
129P3/J mice interestingly have been reported to excrete less F and as consequence to have higher circulating F levels, bone and enamel F levels and they still are remarkably resistant to 5,[7][8]12 . In this study, even without administration of F through the drinking water and with consumption of a low-F concentrations, which might have been due to the residual amounts of F present in their diets and is in-line with the metabolic characteristics of this strain regarding F [4][5] .
In this study, proteomic analysis of liver of 129P3/J and A/J mice was employed to provide insights into the possible mechanisms that could protein is up-regulated and by -sign, when the protein is down-regulated in the respective comparison. a based on protein ID from UniProt protein database (http://www.uniprot.org/) explain the differential metabolic handling and effects of F in these two strains. It has been shown that even without exposure to F, A/J mice present a higher retention of proteins in the maturing enamel 9 . For this reason, the mice were not treated with F, because we wanted to were intrinsic to the strains. Most proteins with fold change were increased in the A/J mice (Table  1), with fold changes ranging between 1 and 2.
Formimidoyltransferase-cyclodeaminase, however, was increased 3.82 times in A/J mice. This enzyme patients with autoimmune hepatitis 14 and is found down-regulated in hepatocellular carcinoma 16 .
Formimidoyltransferase-cyclodeaminase has two enzymatic functions. In one of them, formiminotetrahydrofolate and glutamate are produced. Through its cyclodeaminase function, the enzyme breaks down formiminotetrahydrofolate, involved in the synthesis of purines and pyrimidines, and amino acids (UNIPROT). Thus, the increase in this enzyme might explain the increased expression of other liver proteins in A/J mice due to higher supply of nucleotides and amino acids.
Remarkably, the functional category with the highest percentage of altered genes was oxidationreduction process. The increase of proteins such as ATP synthase subunit alpha, mitochondrial, Heat shock cognate 71 kDa protein, Electron Beta-enolase, Gamma-enolase and, Malate dehydrogenase in the A/J mice indicate an increased concomitant increase in GRP78, which suggests endoplasmic reticulum (ER) stress 20 . ER stress occurs when nascent proteins are misfolded or not folded properly, leading to the initiation of the unfolded protein response, as the unfolded proteins accumulate in the ER 13 . It has been demonstrated that F is able to induce an ER stress response in the LS8 ameloblast-derived cell line, which could be 13 .
In addition, administration of F through the drinking water is able to increase the expression of GRP78 in the liver of rats 20 . Thus, considering that A/J mice oxidative stress even without exposure to F, this exposure has been shown to worsen oxidative stress 20 , which can implicate in the pathogenesis 8 , this can be a hypothesis for the high susceptibility of the A/J to the effects of F.
The proteins in the center of the protein-protein interaction network are related to potassium channels. One of them (calcium-activated potassium channel subunit alpha-1) is a potassium channel activated either by membrane depolarization or increase in cytosolic Ca 2+ that mediates export of K + . It is also activated by the concentration of cytosolic Mg 2+ . Its activation dampens the excitatory events that elevate the cytosolic Ca 2+ concentration and/or depolarize the cell membrane. Therefore, it contributes to the repolarization of the membrane potential and plays a key role in controlling excitability in a number of systems, such as regulation of the contraction of smooth muscle 21 , the tuning of hair cells in the cochlea 6 , regulation of transmitter release 6 and innate immunity 2 . The other one is Disks large homolog 4 that is required for synaptic plasticity associated with NMDA (N-methyl-D-aspartate) receptor signaling 11 . It interacts with shaker-type potassium channels and the cytoplasmic tail of NMDA receptor presence of a protein associated with the nervous system in the center of the network in this study. However, we must consider that liver failure leaves to the accumulation of ammonia, which affects the cerebral function 10 . As mentioned above, A/J mice presented several proteins related to the have caused oxidative stress and contributed to liver damage, which in turn might have provoked cerebral alterations. Since this was a preliminary exploratory work, future studies comparing the should be conducted to add new light into this topic. Also, additional studies should be done to quantify, by other techniques, the proteins with changing expression in this study. Despite being an exploratory study, the lack of additional techniques limitation of this study.

CONCLUSIONS
In conclusion, A/J mice had an increase in This could be a possible explanation for the high susceptibility of these mice to the effects of F, since