This paper introduces a new type of system to simulate conditions in the large intestine. This system combines removal of metabolites and water with peristaltic mixing to obtain and handle physiological concentrations of microorganisms, dry matter and microbial metabolites. The system has been designed to be complementary to the dynamic multi-compartmental system that simulates conditions in the stomach and small intestine described by Minekus et al. [Minekus M, Marteau P, Havenaar R, Huis in't Veld JHJ (1995) ATLA 23:197-209]. High densities of microorganisms, comparable to those found in the colon in vivo, were achieved by absorption of water and dialysis of metabolites through hollow-fibre membranes inside the reactor compartments. The dense chyme was mixed and transported by peristaltic movements. The potential of the system as a tool to study fermentation was demonstrated in experiments with pectin, fructo-oligosaccharide, lactulose and lactitol as substrates. Parameters such as total acid production and short-chain fatty acid (SCFA) patterns were determined with time to characterize the fermentation. The stability of the microflora in the system was tested after inoculation with fresh fecal samples and after inoculation with a microflora that was maintained in a fermenter. Both approaches resulted in total anaerobic bacterial counts higher than 10(10) colony-forming units/ml with physiological levels of Bifidobacterium, Lactobacillus, Enterobacteriaceae and Clostridium. The dry matter content was approximately 10%, while the total SCFA concentration was maintained at physiological concentrations with similar molar ratios for acetic acid, propionic acid and butyric acid as measured in vivo.

Food and nutrition studies in animals and human beings often meet with technical difficulties and sometimes with ethical questions. An alternative to research in living animals is the dynamic multicompartmental in vitro model for the gastrointestinal tract described by Minekus el al. (1995) and . The dynamic conditions that are simulated in this model are peristaltic movements, transit times, pH responses, secretion of enzymes and electrolytes and absorption of nutrients and water. To obtain data for an in vitro model of the dog gastrointestinal tract, the literature was surveyed for physiological responses to different types of dog food. These included: values of enzyme activities, electrolyte concentrations, gastric emptying and intestinal transit times, pH values, secretion and composition of bile and absorption rates in different parts of the dog gastrointestinal tract. The review focuses on research carried out on healthy, adult dogs of 10-20 kg and on parameters related to the oral cavity, stomach and small intestine. This literature research gives sufficient data on the physiology of the canine digestive tract for the development of an in vitro dynamic model that adequately simulates the functions of the stomach and small intestine of the dog.

In order to manufacture complete and balanced dog diets, it is important to know the nutrient requirements of dogs and the availability of these nutrients from food. As pet food manufacturers are restricted in their options for (invasive) animal studies, due to ethical constraints, it is important to have alternative methods for researching the effects of various dog diets. To simulate the gastrointestinal tract of the dog, the dynamic gastrointestinal tract model developed by Minekus et al. was further developed and modified in this study. The model consists of four compartments which simulate the stomach and small intestine (duodenum, jejunum and ileum). Each compartment is made of glass, with a flexible inner wall. This wall can be compressed by increasing the pressure of the surrounding water, mimicking the peristaltic movements and mixing seen in vivo. The model is computer-controlled to simulate physiological parameters such as pH, transit time and secretion of digestive juices, as derived from the literature. Gastric meal delivery and the effects of intestinal transit time on protein digestibility and availability for absorption of calcium from dog food were studied to evaluate the model. The gastric meal delivery of dry dog food was identical to a preset curve, which was based on in vivo data from healthy dogs. The emptying time for canned dog food was somewhat slower than the preset values, probably due to the viscosity of the meal. The differences between the preset values and the measured delivery were not significant. The digestibility of protein and the availability of calcium for absorption increased with a longer transit time. A significant difference was found between medium and slow transit times for the nitrogen content in the ileal delivery effluent and the jejunal dialysates (p < 0.05). The same trend was seen for calcium (not significant). The overall conclusion is that the model is a useful tool for mimicking the gastrointestinal tract of dogs. Parameters such as pH, transit time and enzyme activity can be mimicked and can be kept within a physiological range.

Little is known about the effects of gelling agents in canned dog food on nutrient digestibilities and fecal characteristics. Dogs were fed canned diets with either no gelling agent (control) or one of three gelling agents, wheat starch, a guar gum/carrageenan mixture (50:50), and a locust bean meal (LBM)/carrageenan mixture (50:50) incorporated at two levels (0.2 and 0.5% of the diet on a wet weight basis). Six dogs were fed each diet in a 6 x 7 Youden square design. Dogs fed diets containing gelling agents had higher ileal digestibilities of OM (P = 0.05), fat (P < 0.01), GE (P = 0.02), and total amino acids (P = 0.04) and lower (P < 0.01) total tract DM digestibilities when compared to dogs fed the control diet. Fecal output by dogs fed the control diet, expressed on both an as-is (73.4 g/d) and DM (45.6 g/d) basis, was lower (P < 0.01) than for dogs fed gelling agent-containing diets (mean 102.3 g/d as-is and 57.6 g/d DM). Dogs fed the control diet also had lower (P < 0.01) fecal DM percentages and higher (P = 0.02) fecal scores than dogs fed gelling agent-containing diets.