TABLE 1.

Comparison of the advantage and the limitations of models for studying bacteria-pathogen and bacteria-host interaction.

Type of model	Examples of models*	Advantages	Limitations	Example references
“Simpler” in vitro methods	Agar spot test, broth microdilution, in vitro biofilms	Fast, inexpensive, high throughput, easy-to-perform Flexibility to change parameters easily Wealth of literature available for comparison Standardized protocols across laboratories Good for initial screening purposes	Does not represent in vivo responses Host response is missing Oversimplified models	Balouiri et al., 2016; Fijan, 2016
2D models	Caco-2, HT-29, HT29-MTX, T84, IEC-18 and IPEC-J2 tissue culture cells	Reproducible in lab environment Easy-to-perform Good for initial screening Simple model; well characterized in literature Several visualization methods have been optimized for 2D models	Cell lines mostly derived from cancer cells, thus different from healthy cells Does not include most cell types Hard to culture obligate anaerobes in co-culture due to oxygen requirements Grown as monolayer so they lack 3D structure	McGuckin et al., 2011; Vergauwen, 2015; Devriese et al., 2017; Jose et al., 2017; Yang et al., 2017; Gharbi et al., 2019
3D models	Organoids (e.g., enteroids and colonoids)	Mimics in vivo conditions Multicellular model Possibility of long-term cultures Possibility to investigate cell-cell interaction	Expensive and requires specialized expertise May, need biopsy/tissue samples Variability between models Difficult to study obligate anaerobes because of oxygen requirement Absence of shear forces and intestinal peristaltic movements to help cell differentiation Require complex media formulation and supplements	Werner et al., 2016; Costa and Ahluwalia, 2019; Bédard et al., 2020; Zhang et al., 2020
Chip based models	Microfluidic and multi-channel models	Non-transformed cell lines used and includes all cell types Patient specific tissue biopsies can be used to simulate disease conditions Peristalsis like movement can be included to enhance cell differentiation	Requires access to tissue biopsies Expensive, laborious and requires specialized expertise Variability between donors Small sampling size for downstream analysis	Ingber, 2016; Bein et al., 2018; Sontheimer-Phelps et al., 2020; Baddal and Marrazzo, 2021
In vitro digestive models	Fecal batch-culture, SHIME, TIM, Enteromix, Reading, PolyFermS	Allow the study of interactions with intestinal microbiota Study microbiota from different populations Include other models to study host interaction	No or limited ethical concerns Usually fecal inoculum Expensive to run	Minekus, 2015; Van de Wiele et al., 2015; Lamichhane et al., 2016; Piatek et al., 2020
In silico models		Fast forecast of host-microbiota-probiotic interactions	Only as good as theoretical knowledge of interactions	Geng et al., 2021
Ex vivo models	InteTESTine, Ussing chamber, IVOC	In vivo like multilayered structure Can be used to investigate bacteria interaction in diseased organs” Results can be translated to in vivo conditions	Requires access to human tissue biopsies and can be difficult to obtain Costly Tissues obtained need to be viable and fresh	Haque et al., 2004; Randall et al., 2011; Tsilingiri et al., 2013; Van Krimpen et al., 2014; Newburg et al., 2016; Thomson et al., 2019
Animal models	Mice, rats and pigs	Physiological model Allows investigation in the presence of intact gut microbiota with host cell interactions Possibility to conduct long term investigations Microbiota can be manipulated through diet Innate and adaptive immune response similar to humans	Ethically concerning Requires housing and care In some cases, cannot be translated to human responses Expensive experiments	Heinritz et al., 2013; Jiminez et al., 2015; Sciascia et al., 2016
“Simpler” animal models	C. elegans, honey bee, Ciona robusta, fruit fly, greater wax moth	Relatively fast, inexpensive Includes host response May include defined microbiota	Low/no ethical concerns Simplified host physiology Simplified microbiota	Vilela et al., 2015; Zanni et al., 2015; Zheng et al., 2018; Kumar et al., 2020; Poupet et al., 2020; Liberti et al., 2021

*Please see text for more detailed description of the models.