Methods

Animals and experimental protocol
This study was approved by the Ethics Committee for the Analysis of Research Projects (Protocol 065/13). All animals received humane care in compliance with the experimental protocols of the Ethical Principles in Animal Experiments adopted by the Brazilian Association of Animal Testing.
Thirty adult male Rattus norvegicus albinus Wistar animals weighing 250 to 300 g were housed in the Universidade de São Paulo, School of Medicine’s facilities. Five animals were housed per cage and were given food and purified tap water ad libitum. The animals were operated on in the Center for Study and Research in Surgery (CEPEC), Department of Urology of the Faculty of Medicine. The animals were divided into three groups: the stomach restriction group (R10), the sham group (S10), which underwent the same procedure except for the loofah insertion, and the control group (C10). Animals were fed and kept in separate cages. They were weighed every other day until being sacrificed on day 10.

Surgery to promote gastric reduction
To reduce the gastric capacity and cause significant weight loss in rats without developing anemia, a surgical procedure was performed, similar to that described by Tolosa et al.30. After 12 hours of fasting, the rats were anesthetized by isoflurane 4% vaporization. Anesthesia was continued through a mask with 1.5-3% isoflurane inspired fraction, and the rats were submitted to median laparotomy and gastrostomy in the greater curvature of the stomach for the implantation of a cylindrical Luffa bush (loofah; 1.5 cm in diameter) made of cylindrical Luffa (an experimental phytobezoar). The gastrostomy was closed with continuous mononylon suture yarn 6.0, and the abdominal wall was closed in two planes: the aponeurosis, with 4.0 mononylon suture yarn, and the skin, with 3.0 plain cotton stitches. The animals in the sham group were exposed to the same surgical procedure described above but were not submitted to phytobezoar implantation. Tramadol hydrochloride (20-40 mg/kg) was administered intraperitoneally, with a frequency of at least 12 hours for 5 days31.

Experimental groups
On day 1, the animals were divided into three groups of 10 animals: one group underwent surgery for stomach restriction (R10), with both the sham controls (S10) and normal controls (C10). Each animal was weighed every two days until the day of euthanasia.

Euthanasia and material collection
After the experimental period, the animals were euthanized with carbon dioxide (CO2). This gas is lethal because it causes depression of the central nervous system. The animals were kept in the chamber for more than 10 minutes to confirm their death32. Two colon samples were collected 2 cm from the cecum and were individually packed into containers. Samples were submitted to routine histological processing for Hematoxylin-Eosin (HE), Alcian Blue (AB), Periodic Acid-Schiff (PAS), High Iron Diamine (HID-AB) and Picrosirius Red staining.

Histological morphometric analysis
The Zeiss Microscope Imager.A2 was used for the analysis of the slides with HE. The images were captured by a camera coupled with the AxioVision Release 4.8.1 program. The entire length of the crypts of the colon mucosa was measured and the goblet cells were counted, with three crypts per field. We used six fields per sheet from the control, sham and experimental groups, which were analyzed at a magnification of x400. We also analyzed the inflammatory infiltration in the colon mucosa. Polymorphonuclear and lymphomononuclear cell counts were performed in three areas of the inflammatory infiltration. Inflammatory cell counts were performed in a fraction of the inflammatory cell area. Polymorphonuclear and lymphomononuclear cells were counted in six fields of the area of inflammatory infiltration. The number of inflammatory cells was expressed by dividing the number of cells by the area of the infiltration.
Mucins were quantified in the colon of the control, sham and experimental groups using a computer-assisted image analysis. Briefly, an image analysis system consisting of an Olympus camera mounted on a microscope sent the images to a monitor using a computer-controlled (Pentium 1330 MHz) digitizing system (Oculus TCX, Coreco Inc., St. Laurent, Quebec, Canada). The images were processed by Image-Pro Plus 7.0 software (Media Cybernetics, Inc., Bethesda, MD, USA). For each lamina, three crypts per field were measured, and six fields per sheet from the control, sham and experimental groups were analyzed at a magnification of x400. The mucin content in this compartment was expressed as the number of mucins divided by the total area studied. The final results were expressed as the number of mucins per total area. The measure of the tissue content of Picrosirius Red was analyzed with the same methodology as described above utilizing a polarized light.

Statistical analysis
Weight, goblet and inflammatory cell and mucin tissue expression (PAS, AB, and HID-AB) were presented by groups as the mean, median, minimum and maximum.
Between-group comparisons were made using generalized estimation equations (GEE) for data with a normal distribution, as well as an identity link function, supposing an exchangeable correlation matrix between fields33. Inflammatory cells were compared between groups and types using GEE with Poisson distribution due to cell absence. Bonferroni’s multiple comparison was applied to identify differences between groups34. For weight, a multiple-group comparison was performed using an analysis of variance (ANOVA) followed by Bonferroni’s post hoc test. To verify the correlation between the weight and height of the crypt and the cell quantity, Spearman’s correlation was calculated35. To perform the analysis, IBM-SPSS software for Windows version 22 was used, and the data were tabulated using Microsoft Excel 2013 software. All tests were carried out with a significance level of 5%.