Special Features of the Model

The emphasis of the SHIME® system is primarily put on the simulation of the colon microbial community. Because of the inaccessibility of the human colon region to take a representative microbial inoculum, the fecal microbiota is chosen as inoculum to the colon compartments of the SHIME reactor. The fecal microbiome is significantly different from the in vivo colon microbiome, both in terms of composition as metabolic activity. Yet, the colon being considered as a plug-flow system, the fecal microbiome is nothing else than a colon microbiome that has undergone community and metabolic shifts during transit from the proximal colon to the rectum. The idea of the SHIME system—and other multi-stage colon compartment reactors such as the Reading model—is to allow a suitable adaptation period for the fecal microbiome to adapt to the conditions that prevail in the respective colon compartments. From an engineering perspective a suitable adaptation time for starting up a reactor is around 5–10 times its residence time. Taking the example of a male individual with a gut residence time of 48 h and imposing that in the SHIME system, this would entail an adaptation time of 10–20 days upon inoculation with the fecal microbiota in the colon compartments. We will come back to this stabilization aspect in the next paragraph.

A second aspect is the choice of inoculum. The SHIME is typically inoculated with the fecal microbiome, derived from one individual. There has been or there still is a lot of debate on what is the most suitable inoculum for mimicking the human gut microbiome in the most representative way. Some research groups specifically opt for pooling fecal microbiota from for example ten different individuals (Minekus et al. 1999), thereby (partly) accounting for the huge interindividual variability that exists in microbiome composition and to incorporate properties from different microbiomes in order to create an 'average' 11th microbiome. Given the enormous functional redundancy of the gut microbiome such pooled microbiome will indeed take on a normal fermentation profile which generally not that different from the microbial fermentation profile of a single individual. While such approach may work when investigating hydrolysis and fermentation of carbohydrates, it fails to accurately mimic microbial processes that lack this functional redundancy. To exemplify, it has become clear that the microbial metabolic potency towards polyphenols such as daidzein, isoxanthohumol, catechins and others is highly dependent on an individual's microbiome (van Duynhoven et al. 2011). The existence of different 'metabotypes' has therefore even been proposed to distinguish a bioactive metabolite producing phenotype from a non-producing phenotype (Bolca et al. 2013). The above element is the primary reason why a SHIME reactor is inoculated with the fecal microbiome from one individual and succeeds in maintaining the microbial metabolic phenotype towards specific polyphenols during the in vivo/in vitro transition and several weeks after (Decroos et al. 2006; Possemiers et al. 2006).

The choice for individual inocula also appears to be a crucial aspect in recent studies where the SHIME was fundamentally optimized to enable colonization of the mucosal microbiome (Van den Abbeele et al. 2013). This concerns a third special feature of the SHIME model. The mucosal microbiome is this part of the gut microbial ecosystem that is able to colonize the mucus overlying the gut epithelium. Due to its close proximity to host epithelial cells, the mucosal microbiome is thought to have an intrinsically higher potency to modulate gut health, and by extension, human health (Van den Abbeele et al. 2011). The mucosal microbiome was already known to fundamentally differ from the luminal microbiome in composition and interestingly, presence of important mucosal colonizers such as Faecalibacterium prausnitzii seems to negatively correlate with occurrence (Willing et al. 2009) and postoperative recurrence of ileal Crohn's disease (Sokol et al. 2008). Given the difficult access to the mucosal environment, the development of gut simulators that accurately mimic mucosal microbial colonization is considered a strong asset to obtain a better understanding of the host-microbe interactome. It is in this philosophy that Van den Abbeele et al. (2013) decided to optimize the SHIME for mimicking mucosal microbial colonization by incorporation of mucin-covered microcosms. The major finding of this so-called M-SHIME, or mucosal SHIME, was that colonization of the mucosal environment was characterized by a higher abundance of butyrate producing Clostridium clusters IV and XIVa. This phylogenetic group is considered crucial for delivering butyrate as primary energy source to colonocytes and improves gut barrier function by strengthening the tight junctions. Coming back to the importance of individual inocula and avoiding pooled samples, it was also demonstrated that the M-SHIME was able to maintain the unique features of an individual's microbiome in terms of its mucosal composition (Van den Abbeele et al. 2013).

The fourth feature is the flexibility of the SHIME model and the ease with which reactor compartments can be added or left away. This modular setup is useful when a placebo-controlled study needs to be conducted, when different prebiotics need to be compared (Grootaert et al. 2009) or when a microbiome phenotype producing a bioactive metabolite is compared with a non-producing phenotype (Possemiers et al. 2006). Moreover, it is even possible to explore the interindividual variability in microbiome behavior upon specific treatments by having a common upper digestive tract simulation in the gastric and intestine compartments and subsequently split up the system in several parallel colon compartments, each of which are inoculated with the fecal microbiota from separate individuals.

A fifth feature of the SHIME model refers to the possibility of simulating the microbiome from different human target groups such as adult vs. infant, healthy vs. diseased (e.g. ulcerative colitis patients: Vermeiren et al. 2012) as well as the simulation of animal (pig, dog) microbiomes. In each of these specific cases, the microbial inocula, residence times of the different gastric compartments, composition of the gastric juices, region specific pH's, feed, feeding regimes and body temperatures are adapted in the SHIME set-up leading to an accurate and relevant simulation of the targeted human or animal host. Finally, other features of the SHIME include the gradual emptying of the gastric digest into the intestine compartment, the option of running a dynamic pH profiles in the gastric compartment and the possibility of putting a dialysis unit behind the intestine compartment to enable running experiments with real food matrices or food constituents that need to undergo predigestion and removal of sugar monomers or amino acids and peptides before the digest is transferred to the colon compartment.