The hypothalamus is the major centre of interaction of the numerous anabolic and catabolic factors involved in the control of energy balance. The serotonergic system plays a key role in these mechanisms, inhibiting food intake and stimulating energy expenditure(Reference Saper, Chou and Elmquist1–Reference Meguid, Fetissov, Varma, Sato, Zhang and Rossi-Fanelli4).
Serotonin (5-hydroxytryptamine; 5-HT) release is stimulated by the ingestion of food in hypothalamic sites related to feeding control, a finding that points to the physiological relevance of the system(Reference Schwartz, Hernandez and Hoebel5–Reference De Fanti, Hamilton and Horwitz10). Moreover, several investigations indicate that alterations of central serotonergic activity are present in obese rodents and humans(Reference Mori, Guimarães, Nascimento and Ribeiro6, Reference De Fanti, Hamilton and Horwitz10–Reference Meguid, Fetissov, Blaha and Yang12) and drugs that act as serotonin receptor agonists or serotonin reuptake inhibitors are used in the treatment of obesity(Reference Hainer, Kabrnova, Aldhoon, Kunesova and Wagenknecht13, Reference Halford, Harrold, Boyland, Lawton and Blundell14).
Obesity, with its associated conditions, such as CVD, hypertension, insulin resistance, type 2 diabetes and the metabolic syndrome, represents nowadays a major health issue worldwide. Malnutrition during intra-uterine life has been linked to the development of obesity in later life. The ‘thrifty phenotype’ hypothesis suggests that the adaptive strategies developed during the period of poor nutrient or energy delivery may become programmed, influencing the homeostatic mechanisms towards maximisation of energy uptake and storage(Reference Cripps, Martin-Gronert and Ozanne3, Reference Taylor and Poston15).
We have recently demonstrated that female rats exposed to protein–energy malnutrition during early intra-uterine life presented with non-hyperphagic obesity as adults. Evidence of a deranged central control of feeding was provided by the findings of impairment of insulin-induced hypophagia and hypothalamic signal transduction. The absence of overt hyperphagia pointed to the suggestion that adaptations took place which, at least until that point in life, were able to counterbalance the hypothalamic resistance to insulin, maintaining food intake and limiting the severity of obesity(Reference Sardinha, Telles, Albuquerque, Oyama, Guimarães, Santos and Ribeiro16).
The serotonergic system has been shown to participate in such compensatory mechanisms. For example, down-regulation of serotonergic activity reportedly occurred in neuropeptide Y-deficient mice(Reference Gehlert, Thompson, Hemrick-Luecke and Shaw17). Moreover, the status of the serotonergic system is highly influenced by diet and a nutritional stress during gestation could affect its development, bringing consequences to the control of feeding in adulthood. Data on the effects of gestational undernutrition upon the serotonergic system are not conclusive. Prenatal protein malnutrition failed to affect serotonin and 5-hydroxyindoleacetic acid (5-HIAA) levels in various brain areas, including the hypothalamus, of adult rats(Reference Chen, Turiak, Galler and Volicer18). Serotonin turnover increased in the hypothalamus of infant rats protein malnourished during prenatal life(Reference Kehoe, Mallison, Bronzino and McCormick19). Contrastingly, a decreased density of serotonin binding sites has been found in the cerebral cortex and brainstem of infant rats born from dams fed only 50 % of the normal nutritional intake during the whole gestation(Reference Manjarrez, Manuel, Mercado and Hernandez20). Impairment of brain serotonergic transmission has also been seen in human newborns with intra-uterine growth restriction(Reference Manjarrez, Cisneros, Herrera, Vazquez, Robles and Hernandez21).
In view of the important participation of serotonin on the central control of energy homeostasis, a better knowledge of its functional status after a prenatal malnutrition is of interest. The present work aimed to contribute to this subject by assessing some aspects related to the serotonergic control of feeding in adult female rats, either control or undernourished during the first 2 weeks in utero. We assessed serotonin's ability to inhibit food intake after intracerebroventricular administration, the hypothalamic levels of the serotonin transporter (ST) and the 5-HT1B and 5-HT2C receptor proteins, and the levels of serotonin in microdialysates of the medial hypothalamus, in response to the ingestion of food.
Experimental methods
Animals and experimental design
The Committee on Animal Research Ethics of the Federal University of São Paulo approved the procedures used in the present experiments. Virgin Wistar rats were mated and the first day of pregnancy was determined by examination of vaginal smears for the presence of sperm. Since day 1 of pregnancy, the dams were randomly assigned to be a control or a restricted dam. The control dams (n 11) were fed ad libitum throughout pregnancy and lactation. The restricted dams (n 16) were food restricted, during the first 2 weeks of pregnancy, to 50 % of the intake of the control dams at the same pregnancy day. During the week 3 of pregnancy and lactation, the restricted dams were pair-fed to the control dams. On the day of delivery, the pups were adjusted to eight per dam, including both male and female pups, with a male:female ratio of 2:6. At weaning, two female pups from each litter were randomly selected and assigned to the different studies.
Upon weaning (postnatal day 21), the female offspring from control dams (control group) and from restricted dams (restricted group) were caged five per cage. Both the control and restricted rats were fed ad libitum, from weaning until age 4 months, when all data presented herein were collected. Only female offspring were used in the present study.
All animals were maintained in controlled conditions of lighting (12 h–12 h light–dark cycle, lights off at 18.00 hours) and temperature (24 ± 1°C) and had free access to water throughout the experimental period. The food provided to dams and offspring consisted of standard rat chow (Nuvital Nutrients, Columbo, PR, Brazil) containing (w/w) 4·5 % fat, 23 % protein and 33 % carbohydrate, with 11·3 kJ/g (2·7 kcal/g), as determined at the Bromatology Division of the Federal University of São Paulo.
Carcass and blood determinations
The control and restricted rats were decapitated after an overnight fast. Trunk blood was collected and serum stored at − 70°C until analysed. Insulin was assayed by RIA using a commercially available kit (DPC, Los Angeles, CA, USA) and glucose levels were determined by the glucose oxidase method (Labtest Diagnóstica, Vista Alegre, MG, Brazil).
For determination of carcass lipid content, carcasses were shaved, softened and homogenised after removal of the gastrointestinal tract(Reference Leshner, Litwin and Squibb22). Lipid was extracted from 5 g samples with petroleum ether and determined gravimetrically(Reference Stansbie, Browsey, Crettaz and Demton23).
Microdialysis experiments
The animals were anaesthetised with ketamine–xylazine (67/13 mg/kg) and stereotaxicaly implanted with a 21-gauge guide cannula aimed at the right ventromedial hypothalamus (VMH) (from the bregma: anterior − 2·5 mm, lateral − 0·6 mm and ventral − 7·9 mm; Paxinos & Watson(Reference Paxinos and Watson24)). The cannula was secured to the skull with screws and dental cement and the animals were individually caged thereafter.
After at least 1 week of recovery, a concentric custom-constructed microdialysis probe (1·5 mm of effective membrane length) was inserted through the VMH guide and fixed to it with a small drop of dental cement. The details of probe construction have been described previously(Reference Mori, Guimarães, Nascimento and Ribeiro6). The animals were connected to a swivel system, which allowed continuous probe perfusion with artificial cerebrospinal fluid by a microperfusion pump (Carnegie Medicin, Solna, Sweden). Cerebrospinal fluid composition was: 145 mm-NaCl, 2·7 mm-KCl, 1·0 mm-MgCl2, 1·2 mm-CaCl2, 2·0 mm-Na2HPO4 (pH 7·4). Overnight perfusion was performed at 1·0 μl/min.
On the next morning (after an overnight fast) probe flow rate was adjusted to 1·5 μl/min and collection of 20 min dialysate samples was started after at least 2 h. Samples were collected into 10 μl 0·5 m-perchloric acid and immediately injected into an HPLC system. Baseline samples were collected until 5-HT levels were stable; the last three samples were averaged to yield the mean baseline level (100 % value). Food pellets (2·0 g) were then introduced into the cages and six additional 20 min microdialysate samples were collected. The amount of food consumed was recorded every 20 min, coinciding with the interval of microdialysate collection.
High-performance liquid chromatography analysis
Dialysate levels of 5-HT and 5-HIAA were measured by HPLC with electrochemical detection. The system (ESA Inc., Chelmsford, MA, USA) consisted of a model 580 pump with two PEEK pulse dampers in series, a 50 μl Rheodyne PEEK sample loop, a 3 μm MD150 C column, a model 5020 guard cell set at 300 mV, a model 5014B analytical cell set at − 175 and 150 mV, and a model 5200A detector. The mobile phase consisted of 75 mm-sodium phosphate, 1·5 mm-octanesulfonic acid, 50 μm-EDTA, triethylamine (100 μl/l) and 10 % (v/v) acetonitrile at pH 3·0. The flow rate was 0·6 ml/min. The detection limit for 5-HT was 1·5 pg/50 μl at a signal:noise ratio of 3:1.
Histological analysis
For verification of positioning of the microdialysis membranes, at the termination of the experiments all animals were deeply anaesthetised and perfused with 0·9 % saline followed by 10 % formalin. Brains were removed and 40 μm sections were examined under a microscope, following staining with Cresyl violet. Only data from rats with correct membrane placements were included in the analysis.
Intracerebroventricular cannula implantation and measurement of food intake
At 1 week before reaching age 4 months, the animals were stereotaxicaly implanted with a 23-gauge guide cannula aimed at the lateral cerebral ventricle (from the bregma: anterior − 0·9 mm, lateral+1·6 mm and ventral − 2·5 mm; Paxinos & Watson(Reference Paxinos and Watson24)). The cannula was secured to the skull with screws and dental cement. The animals were individually caged and maintained with food and water ad libitum for 1 week after surgery. After this recovery period, they were fasted for 6 h and then intracerebroventricularly injected with either 2·0 μl saline or 2·0 μl saline containing 200 μg serotonin (Sigma, St Louis, MO, USA). The injections were performed in the animal room, immediately before lights out. The rats were returned to their home cages and pre-weighed food cups were introduced into the cages. The amount of food consumed was determined 24 h later, by weighing the amount of food remaining in the cup. Measurements of food intake were corrected for spillage.
After these measurements, the placement of the intracerebroventricular cannula was examined. For this, the animals were deeply anaesthetised, received an intracerebroventricular injection of 5 μl Evans blue dye, were killed by decapitation and had their brains removed for inspection of dye distribution. Only the results from animals in which the placement of the intracerebroventricular cannula was correct were included in the analysis.
Immunoprecipitation and immunoblotting
The animals were decapitated under anaesthesia (ketamine–xylazine, 33/6·5 mg/kg) and their hypothalami quickly removed and homogenised in 1·0 ml solubilisation buffer (100 mm 2-amino-2-hydroxymethyl-propane-1,3-diol (pH 7·5), aprotinin (0·1 mg/ml), 2 mm-phenylmethanesulfonylfluoride, 10 mm-sodium orthovanadate, 100 mm-sodium fluoride, 10 mm-sodium pyrophosphate and 10 mm-EDTA). Two hypothalami were pooled in each vial. Triton X-100 was added to a final concentration of 1 %. These extracts were clarified by centrifugation and equal amounts of protein (about 1 mg) were immunoprecipitated overnight with antibody against either the 5-HT1B receptor or the ST (SC-1460 or SC-1458, respectively; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and resolved in 8 % SDS-PAGE. The 5-HT2C receptor protein was determined in total extracts of hypothalamus (160 μg protein), which were directly applied to 10 % SDS-PAGE after extraction.
In all cases, the resolved proteins were transferred to nitrocellulose membranes and probed with the primary antibody against 5-HT1B, ST or 5-HT2C (SC-15 081). All membranes were then incubated with anti-goat secondary antibody conjugated with horseradish peroxidase (A9452; Sigma, St Louis, MO, USA). The detection was performed by chemiluminescence (ECL reagent; Amersham Biosciences, Piscataway, NJ, USA) and quantitative analysis was performed with Scion Image software (Scion Corporation, Frederick, MD, USA).
Statistical analysis
The results are expressed as means with their standard errors. Basal levels of 5-HT and 5-HIAA were expressed as their absolute dialysate content. A mean baseline level (100 % value) was obtained by averaging the three samples collected just before the introduction of food and the subsequent data were expressed as percentage of the mean baseline level. The microdialysis data were submitted to ANOVA for repeated measures followed by Duncan's test(Reference Duncan25), for comparisons among the samples collected throughout the experimental period, in the same group. For comparisons between correspondent samples from the control and restricted groups, the independent Student's t test was used. The area under the curve relating serotonin to time was calculated by the trapezoidal rule.
The other comparisons between the control and restricted groups were performed by the independent Student's t test. The same test was used to compare food intake between vehicle-treated and serotonin-treated rats. Significance was set at P < 0·05.
Results
As shown in Table 1, the body weight of the control and restricted rats was similar but body fat was significantly higher in the restricted than in the control animals.
(Mean values with their standard errors for six rats from three litters)
* Estimated by Student's t test.
As shown in Table 2, basal levels of 5-HT and 5-HIAA in VMH microdialysates were similar between the control and restricted groups.
(Mean values with their standard errors for six control and seven restricted rats from three to four litters)
* Estimated by Student's t test.
Figure 1(a) shows the six 20 min food intakes of the control and restricted groups during collection of VMH microdialysates. Both groups ate the most part of the food in the first 20 min interval that followed food presentation, with the restricted group eating an amount significantly higher (90 %) than that eaten by the control rats (P = 0·020). Figure 1(b) shows that the intake of food significantly stimulated serotonin levels in VMH microdialysates in the control (F (5,36) = 3·55; P = 0·018) and restricted (F (5,36) = 3·09; P = 0·036) rats. In both groups, serotonin levels showed a progressive increase until the third 20 min microdialysate sample collected after food presentation. The increase was significantly more pronounced in the restricted than in the control rats in sample 1 (P = 0·04). From sample 3 to sample 6, the levels declined in the control group, returning to values similar to baseline. On the other hand, in the restricted group the levels remained significantly elevated, with no decline being apparent until the end of the experimental period. The area under the curve relating serotonin to time was 173 (se 15) in the control rats and 198 (se 17) (P = 0·14) in the restricted rats for the 120 min after food presentation. In either the control or restricted rats, the levels of 5-HIAA in the VMH microdialysates failed to change significantly (data not shown).
(a) Food intake during the six consecutive 20 min periods after food presentation to control (
) and restricted (
) rats. (b) 5-Hydroxytryptamine (5-HT) levels in 20 min ventromedial hypothalamus microdialysates collected before (basal) and up to 120 min after the presentation of food to control (–●–) and restricted (
) rats. Values are means for six control and seven restricted rats from three to four litters, with their standard errors represented by vertical bars. * Mean value was significantly different from basal (P < 0·05). † Mean value of the restricted rats was significantly different from that of the control rats (P < 0·05).
In the control rats, an intracerebroventricular injection of serotonin caused a significant inhibition of 24 h food intake, in comparison with the intracerebroventricular injection of vehicle (P = 0·04). On the contrary, the injection of serotonin failed to significantly inhibit feeding in the restricted rats, as the intake was similar to that seen after vehicle injection (P = 0·18) (Fig. 2).
Food intake (g/24 h) of control () and restricted () rats intracerebroventricularly treated with either vehicle or 300 μg serotonin. Values are means for eight control and sixteen restricted rats for each treatment from four to eight litters, with their standard errors represented by vertical bars. * Mean value was significantly different from that of the vehicle treatment (P < 0·05).
The Western blot analysis of hypothalamic immunoprecipitates identified the bands corresponding to the ST and 5-HT1B receptor proteins with 75 kDa and 48 kDa, respectively. The levels of the ST protein were slightly (36 %) but significantly (P = 0·03) increased in the hypothalamus of the restricted rats, in comparison with the control rats (Fig. 3 (a)). Hypothalamic levels of the 5-HT1B receptor subtype were 80 % higher in the restricted than in the control females (P = 0·01; Fig. 3 (b)). The 5-HT2C receptor protein, identified in the total extracts with 58 kDa, showed a 28 % non-significant (P = 0·07) decrease in the hypothalamus of the restricted rats, in relation to the control ones (Fig. 3 (c)).
Hypothalamic protein levels of (a) the serotonin transporter (ST; n 6), (b) 5-hydroxytryptamine (5-HT)1B receptor (n 7–9) and (c) 5-HT2C receptor (n 7) in control and restricted rats. Data are are expressed in densitometric arbitrary units. Values are means, with their standard errors represented by vertical bars. * Mean value was significantly different from that of the restricted rats (P < 0·05).
Discussion
Data in the literature indicate that intra-uterine undernutrition may lead to disturbances of appetite and energy homeostasis regulation and adulthood obesity(Reference Cripps, Martin-Gronert and Ozanne3, Reference Taylor and Poston15). We have previously shown that 4-month-old female rats whose dams were undernourished during weeks 1 and 2 of pregnancy had low birth weight and developed non-hyperphagic obesity(Reference Sardinha, Telles, Albuquerque, Oyama, Guimarães, Santos and Ribeiro16). This indicated that a high energetic efficiency contributed to the excess fat deposition, agreeing with the report of a low brown adipose tissue activity in rats exposed to intra-uterine undernutrition(Reference Anguita, Sigulem and Sawaya26).
With the purpose of understanding the participation of disturbances in the hypothalamic systems that control energy homeostasis on the establishment of the normophagic obesity in animals malnourished in utero, we have previously examined the action of insulin. We demonstrated impairment of insulin-induced hypophagia and blunting of insulin signalling activation in the hypothalamus of restricted females. Additionally, we found their circulating leptin levels to be increased(Reference Sardinha, Telles, Albuquerque, Oyama, Guimarães, Santos and Ribeiro16). Other authors have described hypothalamic resistance to leptin in adult rats born from mothers protein–energy restricted during lactation(Reference Passos, Vicente, Lisboa and Moura27). Those data indicated that, despite the observed normophagia, multiple aspects of the central control of energy homeostasis might be compromised in adult female rats prenatally exposed to undernutrition. The present study was aimed at further examining this control, focusing on aspects of the hypothalamic serotonergic function. An important consideration to the rationale of the present study was that both hyperphagia and hypometabolism, common disturbances present in different obesity syndromes, could be consequences of serotonergic dysfunction, as serotonin inhibits food intake and stimulates thermogenesis(Reference Ribeiro, Telles, Oyama, Silveira, Nascimento and Starks2, Reference Meguid, Fetissov, Varma, Sato, Zhang and Rossi-Fanelli4, Reference Rothwell and Stock28). Moreover, although pharmacological activation of central serotonin has long been used as a tool in the treatment of obesity, not much is known about the actual status of the hypothalamic serotonergic system in obese animals and humans.
The present brain microdialysis experiments showed that, during food ingestion, serotonin levels in microdialysates of the medial hypothalamus increased earlier and remained elevated longer in the restricted rats when compared with the control rats. Moreover, the present finding of increased levels of the ST protein is compatible with an elevated hypothalamic turnover of the amine. These observations are in agreement with the reports of increased hippocampal serotonin release in response to raphe electrical stimulation and of elevated free plasma l-tryptophan and brain tryptophan hydroxylase activity in rats prenatally protein malnourished(Reference Mokler, Bronzino, Galler and Morgane29, Reference Manjarrez, Chagoya and Hernandez30).
Exacerbation of the serotonergic response to food has been previously found in hyperphagic obese rats and suggested to reflect a compensatory mechanism aimed at overcoming a putative functional resistance to the released amine(Reference Mori, Guimarães, Nascimento and Ribeiro6, Reference De Fanti, Hamilton and Horwitz10, Reference Orosco, Rouch, Meile and Nicolaidis11). In contrast, in a model of non-hyperphagic and hypometabolic obesity, induced by neonatal treatment with the excitotoxin monosodium glutamate(Reference Ribeiro, Nascimento, Andrade, Hirata and Dolnikoff31), we have previously demonstrated that the pattern of serotonin release in the lateral hypothalamus, as stimulated by the ingestion of food, was similar to that seen in normal rats(Reference Mori, Guimarães, Nascimento and Ribeiro6). Serotonin released at hypothalamic sites during food intake plays a relevant part in the physiological response directed towards satiety and also contributes to thermogenesis(Reference Schwartz, Hernandez and Hoebel5–Reference De Fanti, Hamilton and Horwitz10).
To ascertain whether a functional resistance to serotonin was present in the restricted females, we measured the effect, on the 24 h food intake, of an intracerebroventricular injection of 300 μg serotonin. The findings confirmed the suspected impairment of serotonin action in the restricted rats, by showing that serotonin reduced food intake in the control rats while it failed to significantly inhibit ingestion in the restricted ones. A lower dose of serotonin (200 μg) was ineffective in inhibiting intake in either group (data not shown). These observations reinforce the probability that the excess amount of serotonin released in response to feeding, as observed in the present microdialysis experiments, was developed as a compensation for the impaired ability of the amine to properly induce its biological effect.
The 5-HT1B and 5-HT2C receptor subtypes are important mediators of serotonin hypophagia but it is not completely known how they interact to support this serotonergic function. They have been recently shown to, respectively, inhibit neuropeptide Y/Agouti-related peptide neurons and stimulate pro-opiomelanocortin neurons in the arcuate nucleus(Reference Heisler, Jobst and Sutton32). 5-HT1B has been shown to be hyper-responsive to agonist stimulation in 5-HT2C knockout mice, demonstrating that these serotonergic receptors are able to accomplish compensatory adaptations(Reference De Vry and Schreiber33).
Since the blunting of serotonin-induced hypophagia could be due to diminished receptor density, we measured the hypothalamic amount of these receptors. While 5-HT2C levels were decreased, although non-significantly, those of the 5-HT1B receptor were increased in the restricted rats. This observation does point to a compensatory adaptation, potentially able to overcome serotonin inefficacy. We also found a significantly higher amount of the ST protein in the hypothalamus of the restricted group, a finding compatible with the high local activity of the system, indicated by the exacerbated release elicited by a meal. It is interesting to note that increased 5-HT1B and ST binding sites have been described in the hypothalamus of rats made obese by cafeteria feeding(Reference Park, Harrold, Widdowson and Williams34).
It is possible that either an impaired interaction of serotonin with its receptors or a defective stimulation of the transducing pathways after receptor binding was involved in the altered ability of the neurotransmitter to reduce food intake. Experiments specially designed to explore these aspects will help ascertain to what extent each of these putative abnormalities contributed to the alterations in serotonin action observed in the restricted rats.
The serotonergic system has been shown to be influenced by oestrogen hormones(Reference Raap, DonCarlos, Garcia, Zhang, Muma, Battaglia and Van de Kar35) and we have reported sex differences in the late effects of intra-uterine malnutrition, with obesity and hypothalamic insulin resistance having developed in the female but not in the male offspring(Reference Mori, Guimarães, Nascimento and Ribeiro6). Results from our laboratory (FLC Sardinha, MM Telles, KT Albuquerque, LM Oyama, CMO Nascimento and EB Ribeiro, unpublished results) demonstrated that, unlike the females, the male restricted offspring had a normal hypophagic response to serotonin. Thus, although the relevance of the ovarian hormones to the present findings was not specifically examined, as this aspect was beyond the scope of the present study, they are likely to have played a role.
In newborn and infant animals and humans, there are reports of alterations in the central serotonergic system, induced by intra-uterine or early life malnutrition, which could lead to serotonin-related disorders in later life(Reference Kehoe, Mallison, Bronzino and McCormick19–Reference Manjarrez, Cisneros, Herrera, Vazquez, Robles and Hernandez21). In the present experiments, we were able to demonstrate that female rats exposed to intra-uterine malnutrition retain, as adults, abnormalities in central serotonin physiology, indicating a programming effect of the early undernutrition upon the brain serotonergic system. The present data reinforce our previous demonstration that the brain neural network regulating appetite and energy homeostasis is affected by the exposure to a mild nutritional deficit, confined in a very precocious period of intra-uterine life(Reference Sardinha, Telles, Albuquerque, Oyama, Guimarães, Santos and Ribeiro16).
It is noteworthy that compensatory responses developed to counteract the functional inefficacy of 5-HT, which included increased capacity to release serotonin and increased receptor density. Acting in conjunction, these adaptive responses were probably relevant towards the establishment of a normophagic phenotype. However, the presence of increased body adiposity indicates that the regulation of energy balance was abnormal in the restricted animals and it cannot be ruled out that impairment of the serotonergic component of energy expenditure contributed to it.
Acknowledgements
The present study was supported by a grant from the National Council for Scientific and Technological Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico; CNPq). L. C. J. P., F. L. C. S. and M. M. T. were recipients of fellowships from the Coordination for the Enhancement of Higher Education Personnel (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; CAPES). R. B. G. was recipient of a fellowship from the State of São Paulo Research Foundation (Fundação de Amparo à Pesquisa do Estado de São Paulo; FAPESP). The laboratories of L. M. O., C. M. O. N., O. F. P. S. and E. B. R. were supported by research grants from CNPq and FAPESP.
The experiments of the present study were performed by L. C. J. P., F. L. C. S., M. M. T., R. B. G., K. T. A., I. S. A. and L. M. O. All authors contributed to the discussion and writing of the paper. L. C. J. P. and F. L. C. S. contributed equally to the study.
The authors declare no conflict of interest.
