Streptozotocin, Dinitrophenylhydrazine (DNPH), 2’-7’-dichloro-fluorescein-diacetate (DCFH-DA), oxidized di- chlorofluorescein (DCF), 1,1’,3,3’tetraethoxy-propane, Horse-radish peroxidase (HRP), o-phthalaldehyde (OPT), reduced glutathione (GSH), oxidized glutathione (GSSG), glutathione redutase and 2,6-bis(1,1-dimethylethyl)- 4-methylphenol (BHT) were purchased from Sigma Chemical Co (St. Louis, MO, USA). Commercial kits for albumin, AST and ALT assays were purchased from Gold Analisa Diagnóstica Ltda^®, Belo Horizonte, Brazil. All other chemicals were of analytical.

Dried yacon leaves and fresh yacon roots were acquired from Takashi Kakihara & Cia Ltda., Capão Bonito-São Paulo State, Brazil. A voucher specimen was deposited in the Herbarium of the State University of Maringá under the number HUEM 13021 (Identifier: Silmara Baroni). The roots in natura were firstly dehydrated as previously described [28] . The hydroalcoholic extracts of both leaves and roots were obtained from a 10% suspension (w:v) with 70% ethanol (10 g of dried leaves or dehydrated roots to a final volume of 100 ml 70% ethanol), under mechanical stirring for 5 hours. The extracts were then filtered (filter paper Whatman no. 1), slowly evaporated in a rotary evaporator (Technal, Piracicaba, SP, Brazil) at 600 mmHg to remove the solvent, lyophilized and stored at −20˚C. For the experiments the dried powders (hydroalcoholic extracts) of roots and leaves were suspended in water immediately before use. The lyophilized powder was a green solid and the total yields (w:w) of extraction were 58% for dried roots and 11% for dried leaves.

Male Wistar rats were fed ad libitum with a standard laboratory diet (Nuvilab^®, Colombo, Brazil) and maintained on a regulated light-dark cycle. For the induction of diabetes, animals weighing 180 - 210 g were injected intraperitoneally with a single dose of streptozotocin (50 mg per Kg body weight), dissolved in 0.1 M citrate buffer (pH 4.6). Diabetes was confirmed by measuring blood glucose (tail incision) after 48 hours. Animals with glycemia above 250 mg∙dL⁻¹ (14.0 mM) in the fed condition were included in the study. Control animals received a single intraperitoneal dose of 0.1 M citrate buffer. All experiments of the present work were done in accordance with the world-wide accepted ethical guidelines for animal experimentation and previously approved by the Ethics Committee for Animal Experimentation of the University of Maringá (Protocol 098/10-CEEA).

Rats were randomly distributed into six groups: controls (C); controls treated with yacon root extract (C + RE); controls treated with yacon leaf extract (C + LE); diabetic (D); diabetic treated with yacon root extract (D + RE); and diabetic treated with yacon leaf extract (D + LE). The animals were treated daily by oral gavage with a single dose of the extracts (previously dissolved in water) during 14 days. Control (C) and diabetic (D) groups received saline in the same manner. The dose of the hydroalcoholic extracts of the leaves was 400 mg per Kg body weight as in a previous study [12] . The dose of the root extract was 1.06 mg per Kg, corresponding to 340 mg fructooligosaccharides per Kg, a value also based on previous investigations [9] [20] [29] .

At the end of the treatment period (14 days), 12 hours fasted rats were anesthetized by intraperitoneal injection of sodium pentobarbital (50 mg∙Kg⁻¹). The criterion of anesthesia was the lack of body or limb movement in response to a standardized tail clamping stimulus. The abdominal cavity was exposed, the liver removed, freeze- clamped and stored in liquid nitrogen. The tissue was then homogenized in a Dounce homogenizer with 10 volumes of ice-cold 0.1 M potassium phosphate buffer (pH 7.4) and an aliquot was separated as the total homogenate. The remaining homogenate was centrifuged at 11,000 g during 20 min and the supernatant separated as the soluble fraction of the homogenate.

The mitochondria were isolated by differential centrifugation as previously described [30] . Briefly, the removed fresh liver was minced, washed and homogenized in the ice-cold buffer containing 230 mM mannitol, 70 mM sucrose, 1 mM ethylenediamine tetraacetic acid (EDTA) and 3 mM HEPES (pH 7.4). The homogenate was then centrifuged sequentially at 600 g (10 min) and 7000 g (10 min). The pellet with the intact mitochondria was washed twice and resuspended with the buffer without EDTA. The protein content in the homogenate and mitochondrial fraction was measured using the Folin phenol reagent [31] .

Total phenolic compounds of the yacon extracts were measured by means of the Folin-Ciocalteu reagent and the results expressed as mg of catechin equivalents (g extract)⁻¹ [32] . The antioxidant activity in vitro was measured spectrophotometrically at 515 nm as the ability of the extracts in scavenging the DPPH radicals (1,1-diphenyl-2- picrylhydrazyl) [33] . The antioxidant activity was calculated as the percentage of scavenging efficiency and the results expressed as the half-maximal effective concentration of the extract (EC₅₀) in relation to the 0.2 mg∙ml⁻¹ BHT, the reference compound. The EC₅₀ was calculated by the Scientist software from MicroMath^® (Salt Lake City, Utah) using Stineman’s interpolation formula [34] .

The levels of protein carbonyl groups in the homogenate were used as oxidative injury marker. Protein carbonyl groups contents were measured spectrophotometrically using 2,4-dinitrophenylhydrazine (DNPH) (ε₃₇₀ = 22 ×10³ M⁻¹∙cm⁻¹) and the values expressed as nmol∙(mg∙protein)⁻¹ [35] .

The total ROS content was quantified in the supernatant of the homogenate via the 2’-7’-dichlorofluorescein- diacetate (DCFH-DA) assay as described [36] . Briefly, the acetate groups are removed by cell esterases producing the reduced DCFH, which can be oxidized by peroxides to form the fluorescent oxidized dichlorofluorescein (DCF). The formation of DCF was measured spectrofluorimetrically and the excitation and emission wavelengths were set at 504 and 529 nm, respectively. A standard curve with oxidized dichlorofluorescein (DCF) was used to express the results as nmol・(mg∙protein)⁻¹.

The rate of mitochondrial ROS production (real time ROS production), basically H₂O₂, was estimated by measuring the linear fluorescence increase (504 nm for excitation and 529 nm for emission) due to DCF formation from DCFH via oxidation by H₂O₂ in the presence of horseradish peroxidase [37] . Briefly, intact mitochondria (0.5 mg) were suspended in 2 ml of a mixture containing 250 mM mannitol, 1.36 mM DCFA-DA, 10 mM HEPES buffer (pH 7.2), and 10 mM succinate as respiratory substrate. The experiments were performed in the presence and absence of 10 μM rotenone, a mitochondrial electron transport blocker. The ROS generation is stimulated when isolated mitochondria are energized with succinate and it is further enhanced when the mitochondrial electron transport chain is blocked by rotenone [38] . Fluorescence was recorded during 10 min under agitation. The results were expressed as nmol∙min⁻¹∙(mg∙protein)⁻¹ and, alternatively, as the effective concentration of the extract that inhibits 50% (EC₅₀) of the maximum ROS generation. The EC₅₀ was calculated by numerical interpolation using Stineman’s interpolation formula [34] .

Reduced glutathione (GSH) and oxidized glutathione (GSSG) were measured spectrofluorimetrically in the total homogenate (excitation 350 nm and emission 420 nm) by means of the o-phthalaldehyde (OPT) assay as described [39] . The fluorescence was estimated as GSH. For the GSSG assay, the sample was previously incubated with 10 mM N-ethylmaleimide (NEM) and subsequently with 1 M NaOH and 0.4 µM o-phthalaldehyde to detect the fluorescence. The results were calculated using a standard curve prepared with GSH or GSSG and the values expressed as nmol∙(mg∙protein)⁻¹.

Antioxidant enzymatic activities were assessed in the supernatant of the homogenate. The catalase activity was estimated by measuring changes in absorbance at 240 nm using H₂O₂ as substrate and expressed as µmol∙min⁻¹∙(mg∙protein)⁻¹ [37] . The glutathione reductase activity was estimated by measuring changes in absorbance at 340 nm using NADPH and GSSG as substrates and expressed as nmol∙min⁻¹∙(mg∙protein)⁻¹ [40] . The superoxide dismutase (SOD) activity was estimated by its capacity of inhibiting the pyrogallol autoxidation in alkaline medium. The latter was measured at 420 nm [41] . One SOD unit was considered the quantity of enzyme able to promote 50% inhibition and the results were expressed as U∙(mg∙protein)⁻¹. The glucose 6-phos- phate dehydrogenase (G6PDH) activity was estimated by measuring the increase in absorbance at 340 nm due to NADP⁺-dependent glucose 6-phosphate transformation and expressed as nmol∙min⁻¹∙(mg∙protein)⁻¹ [40] . The glutathione peroxidase activity was estimated by measuring changes in absorbance at 340 nm due to NADPH consumption in the presence of H₂O₂, GSH and glutathione reductase and expressed as nmol∙min⁻¹∙(mg∙protein)⁻¹ [42] .

The glucose concentration and the activities of aspartate amino-transferase (AST) and alanine aminotransferase (ALT) were measured in the plasma. The peritoneal cavity of anesthetized rats was exposed and the blood was collected from the cava vein. After centrifugation at 3000 g for 10 min, plasma glucose, AST and ALT were measured spectrophotometrically using commercial Kits.

The error parameters presented in graphs and tables are standard errors of the means. Statistical analysis was done by means of the GraphPad Prism Software (version 5.0). The statistical significance was analyzed by means of ANOVA with Newman-Keuls post-hoc testing. The significance level was 5% (p < 0.05).

Total phenolic compounds and the antioxidant activity in vitro (DPPH assay) of the hydroalcoholic extracts were evaluated in order to compare the antioxidant potential of the root extracts with those of the leaf extracts and also to compare them with other types of yacon extracts. The results are shown in Table 1. The content in total phenolic compounds of the leaf extract was approximately twenty times higher than that of the root extract. The antioxidant activity via DPPH assay was nearly twenty times higher in the leaf extract. The antioxidant activity of the extracts was additionally explored by measuring their capacity of inhibiting ROS generation in rat liver mitochondria, which constitute a true biological system as opposed to the chemical DPPH assay. Figure 1 shows the concentration dependence of the inhibition of the mitochondrial ROS generation by the yacon extracts and Table 1 lists the EC₅₀ values. Measurements in the presence of both extracts were done in the presence of rotenone, which enhances the production of ROS in isolated mitochondria. Both root and leaf extracts inhibited ROS generation in a well defined concentration-dependent manner. Here again, the leaf extract was much more effective as revealed by the lower concentrations that were required to inhibit ROS generation. Quantitatively this can be inferred from the concentrations producing 50% inhibition (EC₅₀ values) given in Table 1 which reveal that the leaf extract is approximately 45 times more effective than the root extract.

The effects of treating fasted rats with the hydroalcoholic yacon extracts on their blood glucose levels are shown in Figure 2. The antihyperglycemic effect of yacon extracts is well known [7] [8] [17] and these experiments were done in order to ascertain that this effect was preserved in both extracts that were used and to establish doses of both extracts capable of normalizing blood glucose concentration. Glycemia of the diabetic animals was 170% higher when compared to the controls. Treatment of the rats with 1060 mg∙Kg⁻¹ root extract or 400 mg∙Kg⁻¹ leaf extract during 14 days maintained the blood glucose concentration of diabetic rats close to normality. The treatment of control rats did not modify glycemia.

The plasma AST and ALT activities were measured to evaluate if the treatment with the extracts is hepatotoxic.

The results are shown in Table 2. The plasma AST activities in all groups were the same. The plasma ALT activity, however, was higher in diabetic rats when compared to the controls. The treatment of both diabetic and control rats with both extracts did not modify the plasma AST and ALT activities.

Oxidative injury of the liver was evaluated by the levels of protein carbonyl groups in the tissue homogenate and the results are shown in Figure 3. The protein carbonyl levels of non-treated diabetic animals were 40% higher when compared to the controls. It should be mentioned that there is a general agreement that protein carbonyls are a reliable indicator for oxidative damage to macromolecules, especially in the liver [37] . Treatment of the diabetic animals with the leaf and root extracts decreased these levels by 34% and 39%, respectively. The treatment of control rats did not modify the levels of protein carbonyl groups.

The tissue oxidative injury is normally caused by increases in the levels of reactive oxygen species (ROS) whose contents in the total homogenate are shown in Figure 4. The ROS content was 60% higher in the diabetic rats when compared to the controls. The treatment of the control rats with the yacon extracts did not modify the ROS content of the liver. The yacon treatment of diabetic rats, on the other hand, was effective in decreasing the ROS content only when the root extract was administered.

Table 3 shows the levels of oxidized (GSSG) and reduced (GSH) glutathione in the liver. Diabetes did not modify the GSH or GSSG concentrations. Treatment of healthy rats with the leaf extract produced a significant increase of 48% in the GSH content. The root extract, on the contrary, tended to diminish the GSH content, the

change, however, lacking statistical significance. Diabetes did not produce any change in the GSH or GSSG contents. Treatment of diabetic animals with both extracts produced significant increases in the GSH concentrations. The root extract in this case was somewhat more effective (+47%) than the leaf extract (+39%).

Five antioxidant enzymatic activities were measured and the results are shown in Table 4. Diabetes diminished the activities of catalase (30%), superoxide dismutase (50%), glutathione peroxidase (21%) and glucose 6- phosphate dehydrogenase (G6PDH; 34%). The glutathione reductase activity also tended to be smaller, a tendency lacking statistical significance. Treatment of healthy rats with both yacon extracts did not affect the enzymatic activities, although a strong tendency toward stimulation was apparent for the glutathione reductase activity upon the root extract treatment. Treatment of the diabetic rats with both the leaf and root extracts, on the other hand, had a positive effect on all the enzymatic activities listed in Table 4, but with several quantitative differences. The most remarkable difference was found for G6PDH, which was 77% increased by the root extract and only 26% by the leaf extract. The glutathione peroxidase activity was also considerably more increased by the root extract than by the leaf extract treatment, namely by 51% and 16%, respectively. The other enzymes were similarly increased by both extracts. In the case of superoxide dismutase this increase reached 100% or more.