Feeding and food availability modulate brain-derived neurotrophic factor, an orexigen with metabolic roles in zebrafish

Emerging findings point to a role for brain-derived neurotrophic factor (BDNF) on feeding in mammals. However, its role on energy balance is unclear. Moreover, whether BDNF regulates energy homeostasis in non-mammals remain unknown. This research aimed to determine whether BDNF is a metabolic peptide in zebrafish. Our results demonstrate that BDNF mRNAs and protein, as well as mRNAs encoding its receptors trkb2, p75ntra and p75ntrb, are detectable in the zebrafish brain, foregut and liver. Intraperitoneal injection of BDNF increased food intake at 1, 2 and 6 h post-administration, and caused an upregulation of brain npy, agrp and orexin, foregut ghrelin, and hepatic leptin mRNAs, and a reduction in brain nucb2. Fasting for 7 days increased bdnf and p75ntrb mRNAs in the foregut, while decreased bdnf, trkb2, p75ntra and p75ntrb mRNAs in the brain and liver. Additionally, the expression of bdnf and its receptors increased preprandially, and decreased after a meal in the foregut and liver. Finally, we observed BDNF-induced changes in the expression and/or activity of enzymes involved in glucose and lipid metabolism in the liver. Overall, present results indicate that BDNF is a novel regulator of appetite and metabolism in fish, which is modulated by energy intake and food availability.

showing BDNF immunofluorescence (green). A magnified image of representative cells immunopositive for BDNF is shown in a square inset for both foregut and liver. In insets, nuclei are stained blue (DAPI). No or small immunoreactivity was detected in negative (d,g) or preabsorption (e,h) controls. Scale bars are indicated in each image. (i-k) Tissue distribution of mRNAs encoding BDNF receptors in zebrafish. Data obtained by RT-qPCR are expressed as mean + SEM (n = 6), relative to the tissue with the lowest mRNA expression. Ac absorptive cell, BDNF brain-derived neurotrophic factor, Ep epithelium, Lp lamina propria, p75ntr neurotrophin receptor p75, trkb tropomyosin receptor kinase B.
Scientific RepoRtS | (2020) 10:10727 | https://doi.org/10.1038/s41598-020-67535-z www.nature.com/scientificreports/ Figure 2. Periprandial changes in the levels of mRNAs encoding BDNF and its receptors in the zebrafish brain (a-d), liver (e-h) and foregut (i,j). Samples were collected before scheduled feeding time (− 3 h and − 1 h), at feeding time (0 h) and after scheduled feeding time (+ 1 h and + 3 h) in both fed and unfed fish. Data are expressed as mean ± SEM (n = 6) relative to the lowest average expression. Arrows denote feeding time. Different letters indicate significant differences (p < 0.05) among the different time points in fed (black dots) or unfed (white dots) groups, while asterisks indicate significant differences (*p < 0.05, **p < 0.01) between groups at the same time point. Bdnf brain-derived neurotrophic factor, p75ntr neurotrophin receptor p75, trkb tropomyosin receptor kinase B.
In the liver, a significant preprandial increase in bdnf, trkb2 and p75ntra mRNA levels was observed at 1 h before the scheduled feeding (− 1 h) compared to − 3 h values [ANOVA significance values: bdnf: F (4, 25) = 2.912, p = 0.042; trkb2: F (4, 25) = 6.218, p = 0.001; p75ntra: F (4, 25) = 3.597, p = 0.019]. For these three genes, feeding decreased mRNA up to levels similar or slightly lower than its expression at 3 h before the regular feeding time. Such postprandial mRNA levels in fed fish were, in all cases, significantly lower than in fish that missed their scheduled feeding (bdnf: p = 0.002 and 0.029; trkb2: p = 0.017; p75ntra: p = 0.023) ( Fig. 2e-g). Expression levels of p75ntrb in the liver were significantly lower at 3 h after the scheduled feeding compared to the rest of the values [ANOVA significance values: F (4, 25) = 4.685, p = 0.006], which remained unaltered in both fed and unfed fish (Fig. 2h).
In the foregut, bdnf mRNAs increased significantly during -3 to − 1 h to the scheduled feeding time, and drastically decreased at 1 h post-feeding in fed fish [ANOVA significance values: F (4, 25) = 12.690, p ≤ 0.001]. Meanwhile, the levels of bdnf remained high at + 1 h in those fish that missed the scheduled feeding, although a significant drop was detected at + 3 h in this group [ANOVA significance values: F (4, 25) = 16.458, p ≤ 0.001] (Fig. 2i). Expression of p75ntrb also rose preprandially in the foregut. However, no significant postprandial variations were detected neither in fed nor unfed fish when compared to values at scheduled feeding time, and the values at + 3 h returned to levels seen at − 3 h [ANOVA significance values: fed fish: F (4, 25) = 6.038, p = 0.002; unfed fish: F (4, 25) = 5.680, p = 0.002] (Fig. 2j).
BDNF is an orexigen in zebrafish. BDNF Fig. 4d-g). There was also a significant upregulation of proopiomelanocortin (pomc) mRNA expression in the brain upon the injection of 10 ng/g bw BDNF [ANOVA: F (3, 20) = 3.235, p = 0.044; SNK: p = 0.045] (Fig. 4b), and of cocaine-and amphetamine-regulated transcript (cart) and orexin mRNA expression after the injection of 10 Table 1). The effects of in vitro treatment of ZFL cells with BDNF on the expression of genes involved in glucose and lipid metabolism, mitochondrial activity and transcription factors are shown in Table 2. Almost all of the genes studied (glut2, sglt1, gck, pklr, pck2, fbp1a, g6pcb, glycogen phosphorylase, acly, acaca, fasn, cpt1a, hmgcl, acat1, ppargc1α and pparα) were significantly upregulated in response to exposure of cells to 0.  Expression of mRNAs encoding key appetite-regulating peptides in the zebrafish brain (b-h), foregut (i-j) and liver (k-l) 2 h after intraperitoneal administration of saline alone (control) or containing 1, 10 or 100 ng/g bw of BDNF. Data obtained by RT-qPCR are expressed as mean + SEM (n = 6). Different letters indicate significant differences (p < 0.05) among groups assessed by one-way ANOVA and SNK test. Agrp agouti-related protein, BDNF brain-derived neurotrophic factor, cart cocaine-and amphetamine-regulated transcript, cck cholecystokinin, grl ghrelin, npy neuropeptide Y, nucb2 nucleobindin 2, pomc proopiomelanocortin.   ,   Table 1. Effects of intraperitoneal administration of BDNF on the expression of genes involved in glucose and lipid transport and metabolism and of key transcription factors in the zebrafish liver. Fish were ip injected with saline alone (control) or containing 1, 10 or 100 ng/g bw of BDNF and samples were collected 1 h postinjection. Data obtained by RT-qPCR are expressed as mean + SEM (n = 6  Genes that were most significantly induced include glut2, sglt1, pklr and hmgcl (magnitude of induction of ≈ 6 to 7 fold in response to the highest BDNF concentration), while the smallest magnitudes of induction were detected for pck2, cpt1a and pparα (≈ 2 to 3 fold).

BDNF modulates the hepatic expression of genes involved in glucose and lipid metabolism in vivo and in vitro.
In vitro treatment with BDNF also caused a significant reduction in the mRNA levels of acadm [1 h-ANOVA:   (Fig. 5i).

Discussion
BDNF has emerged as an important regulator of feeding and energy balance in mammals [40][41][42] . However, no evidence is available in fish regarding the existence of such a link between BDNF and the regulation of food intake and energy balance. The results of this research address this paucity of information in the fish literature. First, we studied whether BDNF and its receptors are present in tissues involved in the regulation of appetite and energy homeostasis (namely the brain, gut and liver) in zebrafish. The presence and distribution of the BDNF system in the zebrafish brain have been reported 2,12,14 . Thus, the presence of BDNF (both gene and protein) observed in this study in the zebrafish brain confirms these earlier findings. Our results also demonstrate that BDNF mRNA and protein are expressed in the liver and, although at low levels, are also present in the gut of zebrafish. This indicates the presence of BDNF in peripheral tissues. Interestingly, Western blot analysis detected not only a single band corresponding to the mature BDNF but also additional bands, which may correspond to different isoforms and/or glycosylated forms of BDNF and/or pro-BDNF. Indeed, three isoforms of pro-BDNF have been identified in zebrafish, with around 95% of identity 49 . The fact that different band patterns were observed among tissues might indicate a tissue-specific expression of such putative isoforms and/or post-translationally modified forms of BDNF in zebrafish. Besides BDNF, the presence of mRNAs encoding for receptors for this peptide (TrkB2, p75NTRa and p75NTRb) was also observed in the zebrafish gut and liver, supporting a putative role for the peptide in these tissues. Additionally, results from this study described that BDNF, TrkB2, p75NTRa and  www.nature.com/scientificreports/ p75NTRb mRNAs and/or protein are present in other tissues of zebrafish, including the eye, gill, heart, spleen and muscle. This wide tissue distribution suggests a multifunctional role for the BDNF system in fish. It is worth mentioning that only the expression of trkb2 (and not trkb1 or other forms of this receptor) was analyzed in this research. We chose to focus on trkb2 based on previous studies suggesting that this isoform of TrkB is the main BDNF receptor in zebrafish, at least in the brain 27 and the lateral line 28 . This is a limitation of our research, and future studies analyzing the expression pattern of other Trk receptors in zebrafish should be performed. Second, we hypothesized that if BDNF regulates feeding, it might display periprandial variations in the expression and be regulated by food availability. Major findings from our study described an overall preprandial rise and a postprandial decrease in the expression of bdnf and its receptor mRNAs in the zebrafish foregut and liver. These periprandial profiles are typical of orexigenic peptides, including NPY 50 and ghrelin 51,52 . However, we observed a postprandial increase in the abundance of bdnf and trkb2 transcripts in the brain, which are in disagreement with observations in rest of the tissues. This may suggest a different biological response of the BDNF system to the peripheral nutritional status, or be linked to a different physiological action of this system in the zebrafish. The existence of periprandial variations in the expression of the BDNF system led us to the next study on whether BDNF administration influences appetite in zebrafish. Our results demonstrated that BDNF dose-dependently increases food intake in zebrafish at 1, 2 and 6 h post-IP injection. This result is concordant with the periprandial variations observed for the BDNF system in zebrafish peripheral tissues in this study. However, it differs from previous reports in mammals describing an anorexigenic role for BDNF [53][54][55][56] , which strongly suggests different physiological functions for BDNF between fish and mammals. Findings from the present study also indicate that the orexigenic role of BDNF in zebrafish may be mediated by the modulation of brain peptide circuitries, i.e. the upregulation of the orexigens NPY, AgRP and orexin, and the downregulation of the anorexigen NUCB2/ nesfatin-1. Studies in mammals have shown that BDNF appears also to interact with NPY in its action on food intake, although in an opposite direction that the one described here given the anorexigenic action of BDNF in mammals 55 . In order to modulate central appetite-regulating hormones, peripherally administered BDNF must cross the blood-brain barrier. Although no studies are available in fish, BDNF was reported to be able to cross the blood-brain barrier in both directions in mammals 57 . BDNF actions in the zebrafish brain are likely mediated by the TrkB2 or p75NTRa receptors, but not p75NTRb, as suggested by the high expression of the two former ones and the lower expression of the latter in the hypothalamus. Besides brain circuitries, the increased expression of ghrelin in the foregut in response to 100 ng/g bw BDNF suggests that peripheral ghrelin might also contribute to the BDNF-induced increase in appetite in the zebrafish. While not conclusive, among the BDNF receptors studied here, only p75ntrb was detected in the zebrafish foregut, and so it could be a potential candidate that mediates the BDNF-induced upregulation of ghrelin expression. Notably, BDNF administration also caused a significant upregulation of brain pomc and cart, and of hepatic leptin a and leptin b mRNAs. Such observations may appear controversial given the anorexigenic nature of the peptides encoded by these three genes, and might be related to other functions of BDNF in zebrafish. In accordance with the role of BDNF in stimulating food intake in zebrafish, we also observed that expression of bdnf in the foregut is enhanced by food deprivation. Nevertheless, the effects of fasting were opposite in the brain and liver, where it downregulates the expression of the BDNF system. A similar effect of fasting on BDNF was described in the mammalian brain 58,59 . Based on our observations, it is plausible that the increased expression of bdnf in the foregut (directly implicated in sensing the levels of food intake) is enough for eliciting an increase in food intake in fasting states, while the reduced expression in the brain and gut might correlate with other effects of BDNF. However, further studies are needed to understand the physiological meaning of the fasting-evoked downregulation of the BDNF system in the zebrafish brain and liver.
Our next aim was to elucidate whether BDNF has a role in glucose and lipid metabolism in the zebrafish, as supported by the significant presence of the peptide and its receptor in the liver. Our in vivo and in vitro results demonstrated important changes in the expression and/or activity of metabolic enzymes in response to BDNF. The glucoregulatory role of BDNF in the zebrafish liver appears to be linked to increasing the levels of glucose in the hepatic cells, both by stimulating glucose entrance into the cells, or by gluconeogenesis and glycogen phosphorylation. The stimulation of glucose entrance into hepatocytes is suggested by the increase in mRNA and protein levels of the glucose transporters Glut2 and Sglt1 in vitro in response to BDNF exposure. The increased activity and/or mRNA expression of genes encoding the gluconeogenic enzymes Pepck, Fbpase and G6pase in vivo and in vitro points to an increase in gluconeogenesis. Finally, an increase in glycogen phosphorylation is suggested by the upregulation of glycogen phosphorylase. The role of BDNF in stimulating gluconeogenesis and glycogenolysis might be mediated by leptin, which is known to increase the expression of hepatic gluconeogenic enzymes 60 and to induce glycogenolysis 61 in fish. This would explain the increased mRNA expression of leptin a and leptin b in the zebrafish liver after IP administration of BDNF, which seems not to agree with the BDNF-evoked induction of food intake. The suggested capacity of BDNF to increase hepatic levels of glucose in the zebrafish seems to be contradictory with the reported effect of BDNF lowering blood glucose levels in mammals 45,46 . This points out once again that BDNF functions differently in fish and mammals, and warrant further studies on BDNF and its role on glucose homeostasis in vertebrates.
Interestingly, our results suggest that not only gluconeogenic but also glycolytic pathways are upregulated by BDNF. This is suggested by the increased expression of pfkla and pfklb in vivo, increased expression of gck and pklr in vitro, and increased activity of Gck and Pk in vitro. Together, such observations point to a situation in which gluconeogenesis and glycolysis appear to be simultaneously stimulated in the zebrafish hepatic cells in response to BDNF. This hypothesis needs, however, to be further confirmed, as our results in vivo showed no change in the levels of gck, and indeed decreased levels of pklr mRNA, in response to BDNF administration. It is possible that BDNF is simultaneously stimulating gluconeogenesis and glycolysis directly in the liver, but in an in vivo situation where multiple factors are interacting in the modulation of physiological processes, the result is an inhibition of the glycolytic pathways. Gluconeogenesis and glycolysis are typically regulated in order not to Scientific RepoRtS | (2020) 10:10727 | https://doi.org/10.1038/s41598-020-67535-z www.nature.com/scientificreports/ occur at the same time, however the simultaneous stimulation of both processes have been previously reported in a fish species 62 . If both gluconeogenesis and glycolysis are being upregulated by BDNF, it could be hypothesized that the high amount of glucose that is entering into the hepatic cells and that is being synthesized in response to BDNF is at the same time being hydrolyzed. Glucose hydrolysis could be related to either the obtention of energy through the Krebs cycle, or to the synthesis of fatty acids. An increase in fatty acid synthesis in response to BDNF is suggested by our results, given the increased expression and/or activity of the enzymes Acly, Acc and Fas in BDNF-treated groups compared to control groups. Therefore, it would be possible that the pyruvate that results from the glycolysis and is converted into citrate, instead of being used for the Krebs cycle, is being released into the cytoplasm and used as a substrate for fatty acid biosynthesis 63 . Thus, in the cytoplasm, citrate would be converted into acetyl-CoA by Acly, then into malonyl-CoA by Acc and finally into palmitate by Fas. Palmitate can be either β-oxidized in order to obtain more energy or be used for synthesizing phospholipids for maintaining cell membrane structure. Previous reports in mammals indicated that only 20-30% of palmitate is β-oxidized, while about 60-70% of palmitate in the liver and most bodily tissues is incorporated into phospholipids 64 . Results from the present study seem to be in accordance with such an effect in mammals, as the reduced expression of acadm, echs1 and hadh by BDNF suggests that β-oxidation is being inhibited by this peptide in the zebrafish liver. Liver-specific BDNF mutant mice contain the same expression levels of Acadl (long-chain acyl-coenzyme A dehydrogenase) and Acadm compared to wild-type mice when fed a standard chow, but higher levels were found when fed a high-fat diet 65 . This suggests that BDNF does not regulate β-oxidation under a normal diet in mammals, but might diminish the effects of enhanced lipid oxidation, ultimately resulting in hepatic steatosis, under a high-fat diet challenge. While our gene expression results point to a situation in which BDNF would inhibit β-oxidation in zebrafish fed a normal diet, it would be interesting to study the effects of BDNF in fish fed a high-fat diet. In our study, the increase in the mRNA levels and activity of Cpt1a, in charge of facilitating the entrance of fatty acids into the mitochondria, appears to be in disagreement with BDNF not promoting β-oxidation. To study whether this putative increase in the entrance of fatty acids into the mitochondria might be instead associated with an increase in the formation of ketonic bodies, we studied the expression of key enzymes involved in such process. Levels of hmgcl and acat1 mRNAs were upregulated by BDNF in vivo and in vitro, supporting a BDNF-derived induction of ketogenesis in the zebrafish liver. No previous reports are available on the direct effects of BDNF on ketogenesis. However, it has been described that a ketogenic diet alters BDNF expression, although both an upregulation 66 and a downregulation 67 were reported. This observation in mammals together with our results might suggest a reciprocal regulatory loop between BDNF and ketogenesis, although further studies are required, especially given the differences we are here describing between the physiology of BDNF in fish and mammals. Ketogenesis is a typical response to a situation of very low glucose levels and depletion of glycogen stores, such as a prolonged fasting 68 . While we do not have evidence for any cellular metabolic changes after BDNF, the indications from the gene expression changes affecting metabolic machinery suggest an increase in hepatic glucose levels and ketogenesis in response to BDNF. This might suggest that BDNF mimics a metabolic state similar to the one induced by fasting in the zebrafish. This is in accordance with observations in mammals reporting that BDNF mediates the adaptive responses of brain and body to energetic challenges (e.g., food deprivation and exercise) 40 . Based on the tissue distribution profiles shown in this study, all proposed actions for BDNF in the liver are likely mediated by TrkB2 and/or p75ntra, but not p75ntrb. Nevertheless, further studies are needed to confirm the involvement of the different receptor subtypes in the actions of BDNF in the zebrafish.
Finally, we investigated whether Ppargc1α and Pparα mediate the actions of BDNF in the zebrafish liver. PPARGC1α is a key transcriptional regulator of energy homeostasis known to stimulate the transcription of genes involved in gluconeogenesis, Krebs cycle flux, fatty acid oxidation and mitochondrial oxidative phosphorylation in mammals [69][70][71] . PPARα acts as a ligand-activated receptor, controlling the transcription of genes involved in lipid homeostasis 72,73 . In mammals, it has been shown that BDNF is a downstream of PPARGC1α 74-76 , but whether BDNF also activates such coactivator remains to be studied. Present observations showed that ppargc1α and pparα mRNA levels are upregulated by in vivo and in vitro treatment with BDNF in the zebrafish liver, suggesting that BDNF actions in this tissue are mediated (at least partially) by the induction of these two transcriptional regulators.
In summary, this study described the presence of the BDNF system in several tissues of zebrafish, some of which are key regulators of appetite and energy homeostasis. Indeed, we demonstrated that BDNF is an important meal-responsive orexigen in zebrafish. This is the first study reporting the role of BDNF in food intake in fish, and its periprandial profiles in a vertebrate. Additionally, we provided evidence in favor of an important involvement of BDNF in glucose and lipid metabolism in the zebrafish liver, by acting on the expression and/or activity of enzymes implicated in key processes regulating energy balance. Such roles of BDNF in energy homeostasis points to an increase in hepatic glucose levels, lipogenesis and ketogenesis. However, further studies analyzing changes in metabolite levels in response to BDNF are needed to corroborate this putative catabolic role for BDNF in the zebrafish liver. This is especially true considering the fact that BDNF has an orexigenic role in zebrafish. Results presented here describe novel aspects on the physiology of BDNF of both fish and mammals, and add significant new information to our growing knowledge on regulators of metabolic and endocrine functions in vertebrates. Future lines of investigation should evaluate the metabolic effects of BDNF under a high-fat diet challenge, and should focus on the mechanism of action and signaling pathways activated by BDNF.

Methods
Animals. Zebrafish (Danio rerio), with a body weight (bw) of ~ 1 g, were obtained from the Aquatic Toxicology Research Facility at the University of Saskatchewan and housed in 10 L aquaria with a constant flow of temperature-controlled water ( experimental design. Tissue distribution of BDNF and its receptors. The distribution of the BDNF system within the zebrafish was studied both at mRNA and protein levels. For mRNA, the following tissues were collected from six zebrafish: brain (without the hypothalamus), hypothalamus, eye, skin, gills, heart, foregut (intestinal bulb and anterior most portion of the intestine, ≈ 0.5 cm), hindgut (posteriormost portion of the intestine, ≈ 0.5 cm), liver, spleen, ovary, testis and muscle. Tissues were immediately frozen in liquid nitrogen and stored at − 80 °C until quantification of gene expression (see "Quantification of mRNA abundance by real-time quantitative PCR (RT-qPCR)"). For protein determination, samples of brain, hypothalamus, eye, gill, foregut, liver and spleen were dissected, frozen in liquid nitrogen and stored at − 80 °C until Western blot analysis (see "Determination of protein levels by Western blot"). Additionally, two samples of foregut and liver were collected and immediately transferred to 4% paraformaldehyde for immunohistochemistry aimed to determine BDNFlike immunoreactivity (see "Detection of BDNF-like immunoreactivity by immunohistochemistry").
Periprandial expression of the BDNF system. Fish were divided into seven groups (n = 6/group) and acclimated for 2 weeks. On the day of the experiment, samples of whole brain, liver and foregut were collected at 3 h prior to feeding (− 3 h), 1 h prior to feeding (− 1 h), at the regular feeding time (0 h), 1 h after feeding (+ 1 h) and 3 h after feeding (+ 3 h). Two groups remained unfed and were sampled at + 1 h and + 3 h. Tissue samples were collected and stored at − 80 °C for mRNA expression analysis (see "Quantification of mRNA abundance by real-time quantitative PCR (RT-qPCR)").
Fasting-induced changes in the mRNA expression of the BDNF system. Two groups of fish (n = 6/group) were acclimated to tank conditions during 2 weeks. One group of fish was then not fed or exposed to food for 7 days, while the other group continued to receive food at the regular feeding time. At the end of 7 days, brain, liver and foregut were sampled at 11:30 h from both fed and fasted fish. The expression of BDNF system mRNAs was quantified as described below (see "Quantification of mRNA abundance by real-time quantitative PCR (RT-qPCR)").
Effects of BDNF on food intake and on the expression of appetite regulators and genes involved in glucose and lipid metabolism. Twelve groups of fish (n = 3/group) were acclimated to tank conditions for 3 weeks. For one week prior to the experiment, daily basal food intake in each tank was recorded. For this, a pre-weighed amount of food was offered to fish in each tank. After 30 min, the uneaten food was collected, dried for 24 h and weighed. Quantification of food intake was determined by subtracting the dry weight of the amount of food retrieved from the tank after 30 min of feeding from the dry weight of the total amount of food originally provided. For each group of fish, levels of food intake during the 3 days prior to the experiment were averaged and considered the baseline food intake. On the day of experiment, fish were anesthetized prior to their daily scheduled feeding time and intraperitoneally (IP) injected with either sterile saline (0.9% NaCl; control group) or saline containing 1, 10 or 100 ng/g bw of BDNF (Recombinant human BDNF, Catalog # ab206642; Abcam, Toronto, ON, Canada). Of the total of 12 fish groups initially set up, 3 groups were used as control and 3 groups were used for each of the BDNF dose. Even though peptide used here is human BDNF, it was selected based on a BLAST analysis that showed 91.6% identity with the zebrafish peptide. After injections, fish were allowed to recover (5 min) and were fed a pre-weighed quantity of food. Uneaten food pellets were removed 1 h post-feeding, and the amount of food intake was calculated as described above. At 2 h post-injection, a new amount of pre-weighed food was offered again to fish, and, after 30 min, uneaten pellets were collected to calculate food intake levels as described above. The same procedure was repeated at 6 and 24 h post-injection. Food intake levels were calculated for every group of fish (tank) as percentage of food ingested with respect to their corresponding baseline levels (considered as 100%). The complete experiment was repeated three times. At the fourth repetition, injected fish were only allowed to eat at 2 h post-injection. Then, fish were anaesthetized again, sacrificed by decapitation, and brain, foregut and liver were collected. Samples were kept at − 80 °C until gene expression analysis (see "Quantification of mRNA abundance by real-time quantitative PCR (RT-qPCR)").
Effects of BDNF on the mRNA levels of genes involved in glucose and lipid metabolism in vitro. ZFL cells were seeded at 5 × 10 5 cells/well in 24-well plates and grown to confluency as described earlier. Once 80-90% conflu-Scientific RepoRtS | (2020) 10:10727 | https://doi.org/10.1038/s41598-020-67535-z www.nature.com/scientificreports/ ency was achieved, media was replaced by 1 mL of fresh media alone (6 wells) or containing BDNF (0.1, 1 or 10 nM; 6 wells each). After an incubation period of 1 h and 6 h, media was removed and 500 µL of PureZOL™ RNA Isolation Reagent (Bio-Rad, Mississauga, ON, Canada) was added to each well. Cells were then scraped, transferred to tubes and stored at − 80 °C until total RNA was extracted (see "Quantification of mRNA abundance by real-time quantitative PCR (RT-qPCR)"). This experiment was repeated twice.

BDNF effects on the abundance of glucose transporters and the activity of enzymes implicated in glucose and lipid metabolism in vitro.
For this assay we chose the concentrations and time in which BDNF exerts the most significant inductions in mRNA expression. ZFL cells were seeded at 1 × 10 6 cells/well in 6-well plates and grown to confluency. Then, culture media was replaced by 1 mL of fresh media alone (4 wells for Western blot and 8 wells for assessment of enzymatic activity) or containing 1 or 10 nM BDNF (4 wells each for Western blot and 8 wells each for assessment of enzymatic activity), and plates were incubated for 1 h. At the end of the culture time, media was removed and 300 µL of lysis buffer was added to each well. For Western blot analysis, the lysis buffer used was T-PER Tissue Protein Extraction Reagent (Thermo Fisher Scientific), while samples for enzymatic activity assessment were lysed in a 80 mM Trizma buffer (pH 7.6) containing 5 mM EDTA, 2.6 mM DTT and protease inhibitor cocktail (Thermo Fisher Scientific). After the addition of the buffer, cells were scraped, collected and frozen at − 80 °C until further analysis (see "Determination of protein levels by Western blot", "Enzymatic activity determination").
Quantification of mRNA abundance by real-time quantitative PCR (RT-qPCR). Total RNA was isolated using PureZOL™ RNA Isolation Reagent (Bio-Rad). RNA purity was validated by optical density (OD) absorption ratio (OD 260 nm/280 nm) using a NanoDrop 2000c (Thermo, Vantaa, Finland). Then, an aliquot of 1 µg of total RNA was reverse transcribed into cDNA in a 20 µL reaction volume using iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad, Mississauga, ON, Canada) according to the manufacturer's instructions. Real-time quantitative PCRs were performed using SensiFAST SYBR No-ROX Kit (FroggaBio, Toronto, ON, Canada). The specific primer sequences used for target genes, and reference gene (β-actin) are shown in SI Table S1 and were ordered from IDT (Toronto, ON, Canada). Genes were amplified in duplicated RT-qPCR runs using a 96-well plate loaded with 1 µL of cDNA and 500 nM of each forward and reverse primer in a final volume of 10 µL. Each PCR run included a standard curve for the corresponding gene made of two replicates of four serial dilution points, and water instead of cDNA as control in order to ensure that the reagents were not contaminated. RT-qPCR cycling conditions consisted of an initial step of 95 °C for 3 min, and 35 cycles of 95 °C for 10 s and 60 °C for 25 s. A melting curve was systematically monitored (temperature gradient at 0.5 °C/5 s from 65 to 95 °C) at the end of each run to confirm specificity of the amplification reaction. All runs were performed using a CFX Connect Real-Time System (Bio-Rad). The 2-ΔΔCt method 77 was used to determine the relative mRNA expression.

Determination of protein levels by Western blot. Proteins were extracted from tissues and cells using
T-PER Tissue Protein Extraction Reagent (Thermo Fisher Scientific) as directed by the manufacturer. Bradford assay (Bio-Rad) was used to determine protein concentration. The samples (containing 20 µg protein) were prepared in 4× Laemmli buffer containing 0.2% of 2-mercaptoethanol (Bio-Rad) and were subjected to boiling at 95 °C for 10 min prior to loading. Samples were then run on a gradient gel (Bio-Rad) and transferred to a 0.2 μm nitrocellulose membrane (Bio-Rad). After blocking using 1× RapidBlock solution (AMRESCO, Toronto, ON, Canada), target proteins within the membrane were detected by overnight incubation with specific primary antibody: mouse monoclonal to BDNF (1:500 dilution; Catalog # ab203573; Abcam), goat polyclonal to GLUT2 (1:500 dilution; Catalog # ab111117; Abcam) and rabbit polyclonal to SGLT1 (1:500 dilution; Catalog # ab14686; Abcam). Vinculin protein was used for normalization and was detected using rabbit antiserum directed against mouse vinculin (1:1,000 dilution; Catalog # ab129002, Abcam). Secondary antibodies used were: sheep antimouse, goat anti-rabbit or rabbit anti-goat IgG (H+L) HRP conjugate (1:2,000 dilution; Bio-Rad). For visualization of protein, the membrane was incubated for 5 min in Clarity Western ECL substrate (Bio-Rad) and imaged using ChemiDoc MP imaging system (Bio-Rad). Blot images were analysed using ImageLab software and band density of vinculin was used to normalize glucose transporter protein density.

Detection of BDNF-like immunoreactivity by immunohistochemistry. Samples from zebrafish
foregut and liver were collected as previously described and processed (dehydrated and embedded in paraffin) at the Prairie Diagnostic Services, University of Saskatchewan. Paraffin blocks were then sectioned at 7 µm thickness, and transverse sections were mounted onto Superfrost slides (Thermo Fisher Scientific). The protocol for IHC was performed as previously described 78 . Mouse monoclonal human BDNF antibody (1:200 dilution; Catalog # ab203573; Abcam) was used to detect BDNF-like immunoreactivity. As with the peptide, even though this antibody is specifically designed to react with mammalian species, a BLAST analysis of the immunogen showed more than 90% identity with BDNF zebrafish sequence. Additionally, Western blot analysis here performed detected a band of the expected size. Nevertheless, since it is likely that a certain degree of non-specificity exists in our findings, the suffix "-like" was used to refer to immunostaining obtained. Secondary antibody used was goat anti-mouse IgG Alexa Fluor 488 (Invitrogen, Burlington, ON, Canada). Separate sets of slides were used for negative and preabsorption controls. Negative control slides were only treated with the secondary antibody. Preabsorption controls were carried out by incubating slides with a preabsorption mixture of BDNF and primary antibody, using a protocol previously described 79 . All primary and secondary antibodies were diluted in an antibody diluent (Dako, Mississauga, ON, Canada). Slides were mounted using VECTASHIELD mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlington, ON, Canada) and assessed