The studies were carried out using newly born female Wistar rats, kept at 25 ± 1°C and under a 12/12 h light-dark cycle. The animals were fed on a special purified diet for rodents prepared in our laboratory (following AIN-93G ) and they were allowed water ad libitum. Food and water intake as well as the body weight of all animals were recorded weekly. All animal procedures complied with the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes of the Council of Europe, n 123, Strasbourg, 1985.
Administration of alloxan
On reaching the age of 6 days, 16 female rats were injected intraperitoneally with alloxan monohydrate (Sigma-Aldrich Inc., St Louis, MO, USA) dissolved in citrate buffer 0.01 M, pH 4.5 to a concentration of 250 mg/kg of body weight. The animals were fasted for 16 h prior to drug administration . Sixteen rats of the same age injected with the same volume of citrate buffer were used as controls. All the experimental procedures were carried out as originally proposed by Kodama et al , except that those authors injected a lower dose of alloxan (200 mg/kg of body weight). Previous studies in our laboratory  showed that using alloxan 200 mg/kg of body weight resulted in spontaneous recovery from glucose intolerance in all rats after 90 days of experimentation.
The animals (both controls and alloxan-treated) were randomly divided into groups of 8 after weaning (28 days), and were treated up to 120 days old as follows:
• Controls (C): no exercise programme;
• Trained Controls (TC): underwent a swimming training programme;
• Alloxan (A): no exercise programme;
• Trained Alloxan (TA): underwent a swimming training programme;
Glucose tolerance test (GTT)
All rats were fasted for 16 h when they reached the age of 28, 60, 90, and 120 days, then a glucose solution (200 g/L) was administered via a catheter through the mouth into the stomach at a dose of 2.0 g/kg of body weight. Blood samples (25 μL) were then taken from a cut on the tip of their tails to determine the levels of serum glucose at 0, 30, 60, and 120 min after glucose administration. In addition, blood samples (75 μL) were obtained at 0 min to obtain insulin levels in order to calculate the HOMA (Homeostasis Model Assessment) insulin sensitivity index.
The levels of serum glucose were determined using the glucose-oxidase colorimetric enzymatic method (Laborlab Kit, Guarulhos, São Paulo, Brazil), while levels of insulin were measured by radioimmunoassay . The blood response to glucose during the GTT was evaluated based on the total area under the curve of serum glucose (mM × 120 min) .
Sensitivity to insulin
Peripheral insulin sensitivity was evaluated on days 28, 60, 90, and 120 based on the HOMA insulin sensitivity index using the following equation: serum insulin (pmol/L) × serum glucose mmol/L/22.5. Using this method, higher HOMA indices denoted lower insulin sensitivity .
On the 28th day, all animals were subjected to an effort trial to determine their individual exercise intensity necessary to reach the aerobic/anaerobic metabolic transition, following the Maximal Lactate Steady State protocol (MLSS). This protocol was designed to detect the highest blood lactate concentration at which the entrance of lactate into the blood stream was counterbalanced by its removal, maintaining a stable concentration during exercises of constant intensity . This has proved useful in prescribing exercises to and determining the aerobic capacity of humans , rats  and mice . Our research group has recently designed a MLSS protocol for rats using swimming exercises , which was employed in this study. In short, to determine the MLSS, a series of 25 min swimming exercises were performed, which supported increasing overloads in relation to body weight, and were fixed in each series, with intervals of 48 h between them. Blood samples were taken every 5 min, from a cut on the tail tip, for lactate determination. The blood lactate concentration representative of the MLSS was considered at the highest overload in which there was no variation in the blood lactate higher than 1.0 mmol/L between 10 and 25 min of exercise [18, 19]
The animals were then subjected to a training programme of 60 min/day, 5 days a week, over 12 weeks. The swimming sessions were run in individual tanks with the established relative overloads according to the rat's weight in order to reach the individual aerobic/anaerobic metabolic transition . Every 30 days, new MLSS tests were conducted to assign new overloads.
Results are presented as means ± standard deviation and were analyzed using the Students t-test and two-way ANOVA followed by a Newman-Keuls test, where appropriate. Significance level was 5%.