The Prochilodus lineatus juveniles used in this experiment (60.7 ± 1.3 g and 8.0 ± 1.5 cm) were purchased from Piscicultura Polettini, Mogi Mirim/SP, Braziland transported to the Histology Laboratory of UNESP, Campus Rio Claro, Sao Paulo, Brazil. The animals were previously climatized in polyethylene boxes (500 liters) with constant aeration and fed with appropriate commercial food (325 g of crude protein) once a day.

The B. dracunculifolia leaves used in this experiment were collected in Rio Claro-SP, Brazil (22˚22'30.0''S, 47˚28'31.5''W) and, after identification, exsiccates of the vegetal material were deposited and registered in the herbarium “Herbário Rioclarense”, Botany Department, UNESP, Campus Rio Claro (number 58140).

The leaf compound extraction followed the protocol established by ANVISA—(Brazilian Health Surveillance Agency) for the Preparation mother tinctures from dry plants through maceration. The leaves were macerated with grain alcohol, 30% and 70% for nine days. The product of the 30% maceration was mixed to the 70% and vice-versa. After nine days, all the leaf compounds were obtained, those soluble in alcohol 70% and 30%. For each treatment group, 1.5 mL (amount ingested in treatment in folk medicine) of the extract was added to 1.2 g of commercial food (Poytara®). The material was kept in microbiological incubator at 37˚C for alcohol evaporation and stored in amber jars.

Two experimental groups of 30 individuals were used, the experiment was made in duplicate. The control group (Ctrl) and B. dracunculifolia Treated group (BdT). The animals were randomly distributed into four 70-liter tanks (8 animals each) with air pumps, cooling, thermostat (to maintain constant temperature) and covered with UV blocking material to reduce stress. Both groups were fed for a maximum of 21 days: Control group with regular commercial food and B. dracunculifolia Treated group with the food added with B. dracunculifolia extract. The animals were collected 14 and 21 days after the experiment (21-day feeding period), 6 individuals per treatment were collected. The fish were kept in semi-static system (every day about 20% of the water was renewed) and the water physical and chemical parameters (pH, ammonia, hardness and temperature) were measured at each collection.

Fragments of the liver were fixed in formalin 10%, transferred to sodium phosphate pH = 7.4, dehydrated in crescent ethanol series, included in Leica historesin and sectioned (6 µm thickness) using microtome Leica RM2245. The sections were subjected to specific reactions and mounted on slides. For lipofuscin, the slides were mounted using Entellan, without the need of specific reactions, once lipofuscin is fluorescent [17]. The material was analyzed using fluorescence microscope Olympus-BX51 and photographed using software DP-Controller, light filter 450 - 490 nm.

The morphological alterations were evaluated through semi-quantitative methods: Mean Value of Alterations (MVA) and Histopathological Alteration Index (HAI). The MVA was calculated based on the incidence of lesions, according to18where a numeric value is attributed to each animal according to the scale: 1 (absence of histopathological alterations), 2 (localized lesions), and 3 (widely distributed lesions) and the HAI was based on the severity of each lesion. [19]. The HAI value was calculated for each animal, according to the formula:

HAI = ( 1 × Σ I ) + ( 10 × Σ II ) + ( 100 × Σ III )

where ΣI, II and III correspond to the number of the stages: I, II and III, respectively. The HAI values between 0 and 10 indicate normal tissue functioning; between 11 and 20 indicate mild damage to the organ; between 21 and 50 indicate moderate damage; from 51 to 99, severe damage and greater than 100 indicate irreversible tissue damage [18].

For collagen quantification, six liver sections of each individual were analyzed after Picrosirius red reaction, according to [20]. The collagen was isolated using software Image J® version 1.51p and the collagen total area was quantified.

For the glycogen and bile stagnation quantification, six sections of each individual were analyzed following PAS reaction, according to the protocol stablished by [21]. Ten fields from each section were photographed, five for bile stagnation and five for glycogen. Glycogen was semi-quantitatively evaluated and bile stagnation was analyzed using software Image J®. The total bile stagnation area was quantified according to [22].

To analyze the amount of lipofuscin in the tissues, ten photographs of six liver fragments of each individual were taken, according to [23]. The images were analyzed using program ImageJ®, the lipofuscin granules were isolated and the total area was quantified.

Total proteins detection was performed using six sections of each animal, which were subjected to Xylidine Ponceau reaction according to [24]. Macrophages detection was carried out subjecting the same number of sections to Gomori reaction [25]. Total proteins were analized using semi-quantitative method, and machophages were quantified using the software Image J® following [23].

The data obtained through the analyses were submitted to Shapiro-Wilk to verify normality and to ANOVA/Tukey test to obtain parametric results. The groups that did not satisfy normality assumptions were submitted to Kruskal-Wallis/Dunn, with significance level p < 0.05. Statistic test was performed using software Bioestat 5.0® and Graph Pad Prism 5.0®.

All experiments were performed in accordance with relevant guidelines and regulations and approved by the Ethics Committee on Animal Use (CEUA)—from Estadual University of São Pauolo, Rio Claro, license number 10/2017. Before the euthanize the animals were anesthetized with benzocaine solution (0.1 g of benzocaine in 1 mL of ethanol for each 100 mL of deionized water).

The water parameters remained within the acceptable levels [23]. The liver of the specimens presented the following alterations: cytoplasmic vacuolation, nuclear hypertrophy, sinusoid capillary dilatation and congestion, and the relative frequency in which they occurred are displayed in Table 1. The most frequent alterations are displayed in Figure 1.

The MVA and HAI obtained for the hepatic alterations were significantly higher in comparison with the control in the 21-day feeding period—ANOVA/Tukey test (Figure 2).

Some collagen staining was observed in vessel walls, in insufficient amounts for quantification.

The animals fed for 21 days presented significant increase in bile stagnation in comparison with the control group, with p < 0.05 for Kruskal-Wallis/Dunntest (Table 2).

The animals fed during both treatment periods presented significant difference of macrophage number, as well as of total proteins in comparison with the control group, with p < 0.05 for ANOVA/Tukey test.

Lipofuscin granules were identified as red punctuate cytoplasmic fluorescence. The number of granules in the liver increased in both treatments; however, not significantly in comparison with the control group (Figure 3), with p < 0.05 (Kruskal-Wallis/Dunn) (Table 2).

Ctrl—Control group; BdT—B. dracunculifolia Treated group.