Lactobacillus Rhamnosus GG Affects the BDNF System in Brain Samples of Wistar Rats with Pepsin-Trypsin-Digested Gliadin (PTG)-Induced Enteropathy

Celiac disease (CD) presents as chronic low-grade inflammation of the small intestine often characterized by psychiatric comorbidities. The brain-derived neurotrophic factor (BDNF), which we have shown to be reduced in the serum of CD patients, acts as the bridge between immune activation and the nervous system adaptive response. Since Lactobacillus has been shown to upregulate BDNF, this study aimed to evaluate whether the administration of Lactobacillus rhamnosus GG (L.GG) could positively affect the brain BDNF system in rats mimicking the CD lesions. Data have shown that the administration of pepsin-trypsin digested gliadin (PTG) and L.GG alter the levels of mature BDNF (mBDNF), as evaluated by Western blotting. PTG provoked a reduction of mBDNF compared to controls, and a compensatory increase of its receptor TrkB. L.GG induced a slight positive effect on mBDNF levels under normal conditions, while it was able to rescue the PTG-induced reduced expression of mBDNF. The curative effect of L.GG was finely tuned, accompanied by the reduction of TrkB, probably to avoid the effect of excessive BDNF.


Introduction
Celiac disease (CD) presents as a chronic low-grade inflammatory condition of the small intestine that develops in genetically susceptible individuals after the ingestion of gliadin [1]. Apart from classical gastrointestinal (GI) (such as diarrhea, bloating, abdominal pain, constipation) and extra-GI symptoms (e.g., iron deficiency anemia, weight loss, osteoporosis, and fatigue), it may often be characterized by psychiatric comorbidities [2], such as anxiety and depression [3]. Indeed, a link exists between inflammation and neuroplasticity: pro-inflammatory cytokines act as key mediators, with the brain-derived neurotrophic factor (BDNF) representing the molecular bridge between immune activation and the capacity of the nervous system to adapt to environmental challenges [4]. BDNF has been determined [33], the possible response of the BDNF system to L.GG administration was investigated in brain samples of new-born male Wistar rats. Thus, the levels of pro and mBDNF isoforms, of different BDNF mature transcripts, and the TrkB and p75NTR receptors were evaluated.

Gliadin Digest
PTG was obtained following a previously published procedure [34]. Briefly, 50 g wheat gliadin (Sigma-Aldrich, Milan, Italy) was dissolved in 500 mL 0.2 N HCl and digested for 2 h at 37 • C with 1 g pepsin (P6887, Sigma-Aldrich, Milan, Italy). After adjusting to pH 7.4, the peptic digest was incubated with 1 g trypsin (T7418, Sigma-Aldrich, Milan, Italy) for 4 h at 37 • C. After boiling for 30 min, the solution was freeze-dried, lyophilized, and stored at −20 • C until used.

Animals and Experimental Design
The study was approved by the Italian Ministry of Health (approval date: 15 December 2016; n. 1178/2016-PR). All the applied procedures followed the International Guidelines for the use of laboratory animals. Brain samples used in this study were from new-born Wistar rats housed at the animal housing room of the National Institute of Gastroenterology "S. De Bellis" Research Hospital, Castellana Grotte, Bari, Italy.
The experimental design included a control and four different treatments, and every litter of at least ten puppies represented a different treatment group (Table 1)  Animals previously sensitized with 1000 U IFN-γ administered intraperitoneally after birth and with a following oral administration of PTG 50 µg/day for 10 days 10 3 L.GG Animals treated with an oral administration of L.GG 1 × 10 9 CFU for 10 days 10 4

Co-administered
Animals previously sensitized with 1000 U IFN-γ administered intraperitoneally after birth and with a following oral co-administration of PTG 50 µg/day and L.GG 10 9 CFU for 10 days 10 5 Pre-treated Animals previously sensitized with 1000 U IFN-γ administered intraperitoneally after birth with a following administration of PTG for 10 days and successively treated with L.GG 10 9 CFU for further 10 days 20 Ctrl = controls. PTG = pepsin-trypsin-digested gliadin. L.GG = Lactobacillus rhamnosus GG.
In detail, as concerns PTG treatment, new-born rats received orally 50 µg PTG/day in a single dose for ten days, and, finally, a provocative dose of PTG 100 µg two hours before sacrifice.
The probiotic Lactobacillus rhamnosus GG (ATCC 53103) (Dicoflor, Dicofarm, Rome, Italy) was administered orally at a concentration of 10 9 CFU/day in a single dose for ten days.
After treatments, the puppies were sacrificed by anesthetic overdose, and brain samples were immediately removed and stored at −80 • C until assayed.

Western Immunoblotting
Proteins were extracted from brain samples using lysis buffer (Pierce Ripa buffer, Thermo  Scientific, Rockford, IL, USA) supplemented with protease and phosphatase inhibitors (Thermo  Scientific, Rockford, IL, USA).

PCR
In this study, the expression of BDNF total transcripts and transcripts containing exon III, IV and VI was evaluated.
Total cell RNA from brain samples was extracted using Tri-Reagent (Mol. Res. Center Inc., Cincinnati, OH, USA), following the manufacture's instruction. For cDNA synthesis, about 2 µg total cell RNA was used. Reverse transcription (RT) was carried out using the iScript Advanced cDNA Synthesis Kit (BioRad Laboratories Inc., Hercules, CA, USA).
Expression of BDNF total transcripts was evaluated by the quantitative PCR (qPCR) method. Real-time PCRs were performed in 25 µL final volume containing 2 µL of cDNA, 1x master mix with SYBR Green (iQ SYBR Green Supermix, BioRad Laboratories Inc., Hercules, CA, USA) and sense and antisense primers for the total BDNF and the GAPDH gene [35]. Reactions were performed in a CFX96 Real-Time PCR Detection System (Bio-Rad, Milan, Italy) using the following protocol: 45 cycles at 95 • C for 30 s, 95 • C for 5 s, 60 • C for 30 s followed by a melting curve step at 65-95 • C with a heating rate of 0.5 • C per cycle for 80 cycles. Relative quantification was done using the ∆∆Ct method.
Expression of BDNF transcripts containing exon III, IV, and VI, respectively, were evaluated by end point-PCR using Dream Taq PCR 1x Master Mix (Thermo Fisher Scientific Inc, Waltham, MA, USA), 25 pmoles forward and reverse primers and cDNA template (2 µL) in 25 µL of final volume. PCR parameters were initial denaturation at 95 • C for 7 min, followed by 35 cycles of 30 s at 95 • C, 20 s at 60 • C, 30 s at 72 • C, and 7 min of final extension at 72 • C. Primers were as those in Smiljanic et al. [36].
An aliquot (10 µL) of each PCR amplification was loaded on 1.5% agarose gel. Ethidium bromide-stained bands were visualized using the Gel Doc XR documentation system (BioRad Laboratories Inc., Hercules, CA, USA) and Image Lab Software (BioRad Laboratories Inc., Hercules, CA, USA) was used to analyze band intensity. The expression of BDNF transcripts was expressed as relative values (OD of BDNF normalized to the OD of the corresponding GAPDH band).

Statistical Analysis
Due to the non-normal distribution of the data, non-parametric tests were performed. Data were analyzed by Kruskal-Wallis analysis of variance and Dunn's Multiple Comparison Test. Correlation between pro and mBDNF levels was analyzed by the Spearman regression test. All data are expressed as mean and SEM. Differences were considered significant at p < 0.05. A specific software package (SigmaStat for Windows version 3.00 SPSS Inc. San Jose, CA, USA) was used.

BDNF Analysis
The levels of the mature BDNF and the BDNF precursor, reported as the expression relative to ß-actin, were evaluated in samples of the brain from untreated and treated rats. Western blot analyses showed the presence of both mBDNF and proBDNF isoforms ( Figure 1). Experimental treatments affected the expression of the protein, with a statistically significant difference as concerns the mature isoform of BDNF (p: 0.0315, Kruskal-Wallis test). In rats sensitized with IFN-γ, PTG treatment induced a 33% reduction, and the administration of L.GG caused a small increase (29%) of mBDNF levels compared to Ctrl. When rats were co-administered with PTG and L.GG, a 91% increase in mBDNF levels compared to Ctrl occurred, whereas the administration of L.GG in pre-treated rats almost doubled the levels of mBDNF in comparison with Ctrl (194% increase); levels in co-treated rats were significantly different from those in PTG treated ones (p < 0.05, Dunn's multiple comparison post-test) ( Figure 1).
Nutrients 2020, 12, 629 5 of 12 concerns the mature isoform of BDNF (p: 0.0315, Kruskal-Wallis test). In rats sensitized with IFN-γ, PTG treatment induced a 33% reduction, and the administration of L.GG caused a small increase (29%) of mBDNF levels compared to Ctrl. When rats were co-administered with PTG and L.GG, a 91% increase in mBDNF levels compared to Ctrl occurred, whereas the administration of L.GG in pre-treated rats almost doubled the levels of mBDNF in comparison with Ctrl (194% increase); levels in co-treated rats were significantly different from those in PTG treated ones (p < 0.05, Dunn's multiple comparison post-test) ( Figure 1). The precursor proBDNF was affected by the different treatments; the administration of PTG and L.GG, both alone and in combination, induced an increase in the levels of this isoform, although without reaching statistical significance (p: 0.234 Kruskal-Wallis test) ( Figure 1).

Figure 1.
Western blot analysis of mBDNF and proBDNF in brain samples from control and treated rats, each group consisting of six rats. Immunoreactive bands were quantified using Image Lab Software (BioRad Laboratories Inc., Hercules, CA, USA). The graphs on the right show quantification of the intensity of bands calibrated to the intensity of the ß-actin band. All data represent the results of at least three independent experiments (mean plus SEM). * p < 0.05. Data were analyzed by Kruskal-Wallis analysis of variance and Dunn's Multiple Comparison Test.
The levels of precursor and mature BDNF positively correlated, reaching statistical significance in the Ctrl and pre-treated groups (r: 0.94, p: 0.01, Spearman test) (Figure 2A, E, respectively) and approached significance in PTG and L.GG rats (r: 0.83, p: 0.058, Spearman test) ( Figure 2B, C, respectively). Western blot analysis of mBDNF and proBDNF in brain samples from control and treated rats, each group consisting of six rats. Immunoreactive bands were quantified using Image Lab Software (BioRad Laboratories Inc., Hercules, CA, USA). The graphs on the right show quantification of the intensity of bands calibrated to the intensity of the ß-actin band. All data represent the results of at least three independent experiments (mean plus SEM). * p < 0.05. Data were analyzed by Kruskal-Wallis analysis of variance and Dunn's Multiple Comparison Test.
The precursor proBDNF was affected by the different treatments; the administration of PTG and L.GG, both alone and in combination, induced an increase in the levels of this isoform, although without reaching statistical significance (p: 0.234 Kruskal-Wallis test) (Figure 1).
The levels of precursor and mature BDNF positively correlated, reaching statistical significance in the Ctrl and pre-treated groups (r: 0.94, p: 0.01, Spearman test) (Figure 2A,E, respectively) and approached significance in PTG and L.GG rats (r: 0.83, p: 0.058, Spearman test) ( Figure 2B,C, respectively). The levels of BDNF total transcripts in brain samples from untreated and treated rats were determined. No statistically significant difference was found ( Figure 3).  The levels of BDNF total transcripts in brain samples from untreated and treated rats were determined. No statistically significant difference was found ( Figure 3). The levels of BDNF total transcripts in brain samples from untreated and treated rats were determined. No statistically significant difference was found ( Figure 3).   Figure 4 is a representative agarose gel showing end point-PCR results for exon III and IV respectively. Only faint bands were obtained using primers specific for the exon VI (data not shown). A statistically significant difference was found as concerns the transcript containing the exon III (p: 0.0293, Kruskal-Wallis test) ( Figure 4A), but not for exon IV ( Figure 4B).
Nutrients 2020, 12, 629 7 of 12 Figure 4 is a representative agarose gel showing end point-PCR results for exon III and IV respectively. Only faint bands were obtained using primers specific for the exon VI (data not shown). A statistically significant difference was found as concerns the transcript containing the exon III (p: 0.0293, Kruskal-Wallis test) ( Figure 4A), but not for exon IV ( Figure 4B).

Figure 4. End point-PCR analysis of BDNF transcripts containing exon III (A) and IV (B) in brain
samples from control and treated rats, each group consisting of six rats. An aliquot (10 µL) of each PCR amplification was loaded on 1.5% agarose gel. Bands were quantified using Image Lab Software (BioRad Laboratories Inc., Hercules, CA, USA). The graphs on the right show quantification of the intensity of bands calibrated to the intensity of the GAPDH band. All data represent the results of at least three independent experiments (mean plus SEM). Data were analyzed by Kruskal-Wallis analysis of variance and Dunn's Multiple Comparison Test.

TrkB and p75NTR Analysis
The effect of PTG and L.GG administration on the levels of TrkB was evaluated. In brain samples, a 140-145 kDa and a 95 kDa TrkB isoforms were detected. The experimental treatments affected the levels of the 140-145 isoform (p: 0.0433, Kruskal-Wallis test). The administration of PTG induced a 39% increase in protein levels in comparison with Ctrl; an increase was also observed when L.GG was administered, both alone and in combination with PTG (5% and 12% increase, respectively). In contrast, the administration of L.GG after the PTG treatment induced a significant 59% reduction in protein levels compared to PTG (p < 0.05, Dunn's multiple comparison post-test). Changes in the levels of the non-glycosylated form were not statistically significant ( Figure 5A).
Levels of the proBDNF p75NTR receptor were evaluated: experimental treatments did not significantly affect this protein (p: 0.0625, Kruskal-Wallis test) ( Figure 5B).

TrkB and p75NTR Analysis
The effect of PTG and L.GG administration on the levels of TrkB was evaluated. In brain samples, a 140-145 kDa and a 95 kDa TrkB isoforms were detected. The experimental treatments affected the levels of the 140-145 isoform (p: 0.0433, Kruskal-Wallis test). The administration of PTG induced a 39% increase in protein levels in comparison with Ctrl; an increase was also observed when L.GG was administered, both alone and in combination with PTG (5% and 12% increase, respectively). In contrast, the administration of L.GG after the PTG treatment induced a significant 59% reduction in protein levels compared to PTG (p < 0.05, Dunn's multiple comparison post-test). Changes in the levels of the non-glycosylated form were not statistically significant ( Figure 5A).
Levels of the proBDNF p75NTR receptor were evaluated: experimental treatments did not significantly affect this protein (p: 0.0625, Kruskal-Wallis test) ( Figure 5B).

Discussion
This study was performed to evaluate the effect of administration of the probiotic L.GG on the BDNF system in the course of PTG-induced enteropathy. An animal model able to mimic the CD lesions in vivo was used, namely, newborn rats sensitized with IFN-γ immediately after birth and administered with PTG [32,37].
In a previous study, our group reported not only lower circulating BDNF levels in CD patients than healthy controls, but also a significant correlation between low BDNF levels and impaired quality of life [21]. It has already been demonstrated in animal models that chronic GI inflammation induces anxiety-like behavior through the alteration of CNS biochemistry [38]. BDNF is a key molecule in the CNS, able to affect mood, behavior, learning, and memory and a reduced circulating level of this neurotrophin is related to the development of psychiatric disorders [39].
Since circulating BDNF levels reflect those in the brain [23], in the present study, the regulation of the BDNF system was evaluated in rat brain samples. Specifically, our aim was to investigate whether the administration of L.GG could modulate the expression of the mature, proteolytically cleaved mBDNF protein and the precursor proBDNF isoforms in the course of PTG induced enteropathy. Besides, total and differently spliced BDNF transcripts, the TrkB, evaluated as full-length and glycosylated forms, and p75NTR receptors levels were investigated.
This study revealed that the experimental treatments (administration of PTG and L.GG, alone or in combination) significantly altered the levels of mBDNF, as evaluated by Western blotting.
The administration of PTG to rats sensitized with IFN-γ downregulated the expression of the mature isoform of the neurotrophin, which is involved in promoting neuronal survival and plasticity [15]. This result is consistent with the already demonstrated relationship between peripheral inflammation and decreased brain neurotrophic factors [17], which can affect cognition, anxiety and depression [38][39][40], as well as with our previous findings in CD patients [21].
Increasing evidence shows that the gut microbiota can influence the development and maintenance of the CNS and enteric nervous system (ENS), through microbial metabolites, such as short-chain fatty acids and neurotransmitters, able to cross the intestinal and blood-brain barriers [41]. The effectiveness of probiotic supplementation in the reduction of psychological symptoms [42][43] and as an alternative treatment for CD [25] are nowadays the objects of investigation. Indeed, Figure 5. Western blot analysis of TrkB (A) and p75NTR (B) in brain samples from control and treated rats, each group consisting of six rats. Immunoreactive bands were quantified using Image Lab Software (BioRad Laboratories Inc., Hercules, CA, USA). The graphs on the right show quantification of the intensity of bands calibrated to the intensity of the ß-actin band. All data represent the results of at least three independent experiments (mean plus SEM). * p < 0.05. The data were analyzed by Kruskal-Wallis analysis of variance and Dunn's Multiple Comparison Test.

Discussion
This study was performed to evaluate the effect of administration of the probiotic L.GG on the BDNF system in the course of PTG-induced enteropathy. An animal model able to mimic the CD lesions in vivo was used, namely, newborn rats sensitized with IFN-γ immediately after birth and administered with PTG [32,37].
In a previous study, our group reported not only lower circulating BDNF levels in CD patients than healthy controls, but also a significant correlation between low BDNF levels and impaired quality of life [21]. It has already been demonstrated in animal models that chronic GI inflammation induces anxiety-like behavior through the alteration of CNS biochemistry [38]. BDNF is a key molecule in the CNS, able to affect mood, behavior, learning, and memory and a reduced circulating level of this neurotrophin is related to the development of psychiatric disorders [39].
Since circulating BDNF levels reflect those in the brain [23], in the present study, the regulation of the BDNF system was evaluated in rat brain samples. Specifically, our aim was to investigate whether the administration of L.GG could modulate the expression of the mature, proteolytically cleaved mBDNF protein and the precursor proBDNF isoforms in the course of PTG induced enteropathy. Besides, total and differently spliced BDNF transcripts, the TrkB, evaluated as full-length and glycosylated forms, and p75NTR receptors levels were investigated.
This study revealed that the experimental treatments (administration of PTG and L.GG, alone or in combination) significantly altered the levels of mBDNF, as evaluated by Western blotting.
The administration of PTG to rats sensitized with IFN-γ downregulated the expression of the mature isoform of the neurotrophin, which is involved in promoting neuronal survival and plasticity [15]. This result is consistent with the already demonstrated relationship between peripheral inflammation and decreased brain neurotrophic factors [17], which can affect cognition, anxiety and depression [38][39][40], as well as with our previous findings in CD patients [21].
Increasing evidence shows that the gut microbiota can influence the development and maintenance of the CNS and enteric nervous system (ENS), through microbial metabolites, such as short-chain fatty acids and neurotransmitters, able to cross the intestinal and blood-brain barriers [41]. The effectiveness of probiotic supplementation in the reduction of psychological symptoms [42,43] and as an alternative treatment for CD [25] are nowadays the objects of investigation. Indeed, recent results from clinical trials [44], animal models [45], and in vitro studies [46] indicate that the Lactobacillus administration up-regulates BDNF expression.
The results of the present study show that the administration of L.GG was able to increase mBDNF compared to Ctrl rats slightly. When L.GG was co-administered with PTG, the increase in mBDNF was higher than that observed in rats treated with L.GG alone. This increase was even higher in PTG-pretreated rats, leading to significantly different neurotrophin levels from those in PTG rats. Therefore, L.GG showed only a slight positive effect on BDNF levels under normal conditions, while it was able to rescue the reduced expression of BDNF, thus repairing the adverse impact induced by a ten-day PTG administration.
The variations of proBDNF in all the experimental conditions followed a trend quite similar to that of the mature isoform, although the differences of proBDNF levels were not statistically significant. The existence of the balance between pro and mBDNF, shown by the positive correlation between the two isoforms, can be considered as suggestive of a major role of proBDNF as the precursor of the proteolytically cleaved mBDNF. Notably, an imbalance between the two isoforms and the up-regulation of the expression of proBDNF/p75NTR has been reported in patients with major depressive disorders [47].
The impact of experimental treatments on the levels of BDNF transcripts was also evaluated. It is noteworthy to remember the complexity of the regulation of BDNF expression, which involves transcriptional and post-transcriptional mechanisms, such as miRNA regulation [48]. Results from this study show the significant difference in the levels of the specific transcript containing exon III, one of the major activity-dependent transcripts [35]. Further studies are needed to dissect the complexity of the regulation of BDNF better. mBDNF exerts its effect through the interaction with its specific TrkB receptor, which exists as multiple differently spliced isoforms, glycosylated and non-glycosylated [49]. The N-glycosylation, allowing the localization of the protein to the cell surface, enables ligand-specific activation [50]. Two forms of the TrkB protein were present in brain samples from differently treated rats, a 140-145 kDa one, representing the glycosylated receptor, and a 95 kDa form, corresponding to the full-length TrkB. The levels of the 95 kDa protein were not significantly affected by experimental treatments, whereas significant changes were observed in levels of the glycosylated protein. The most relevant increase in the 140-145 kDa protein was found in the PTG group, which could be indicative of a compensatory response to the reduction of its ligand mBDNF evoked by PTG treatment. Conversely, the reduction of the protein observed in the pre-treated group could be considered as protective against the exposure to possibly excessive mBDNF levels. Indeed, TrkB mediates neuronal necrosis in the case of elevated levels of mBDNF [51].
Consistently with the non-significant variation of pro-BDNF levels in all the experimental conditions, the levels of p75NTR also showed the same behavior.
In conclusion, data from this study demonstrate that the induced enteropathy affects the expression of the BDNF system in the rat brain. Interestingly, PTG treatment was able to provoke a balanced response consisting of a reduction of the neurotrophin compared to untreated controls along with a compensatory increase in the TrkB receptor. The curative effect of L.GG also appeared to be finely tuned in this experimental model, since the reduction of the cognate receptor accompanied the increase of BDNF in pre-treated rats to probably avoid the negative effect of an excess of BDNF [51].
On the basis of our results, the actual role of the BDNF system should be further investigated in gluten-related human disorders, as well as in the course of probiotic treatments. The obtained results could be helpful for the more efficient management of the celiac patient.