Mistakes in mouse models of IBD and how to avoid them

April 27, 2016 By: Pim J. Koelink and Anje A. te Velde

Image courtesy of P.J. Koelink.

View and download pdf version

Mistakes in mouse models of IBD and how to avoid them 

Learn how to get the most from experimental colitis models!

In general, mouse models of colitis are used to study its pathophysiology and for the development of new treatment modalities for inflammatory bowel disease (IBD). For the latter it is essential to select a mouse model that has many overlapping features with human IBD.

More than 50 experimental colitis models have been developed and they have provided us with very useful insights into IBD physiology, as reviewed by Bouma and Strober1 and others,2–4 but they have limited use in predicting the clinical relevance of therapeutic targets in IBD.5

Experimental colitis models broadly fit into four different groups. First is spontaneous colitis, resulting from a naturally occurring genetic abnormality. Second is induced colitis occurring as a consequence of a targeted mutation or the introduction of a transgene. Third is induced colitis resulting from administration of different exogenous causative agents. Fourth is induction of colitis by manipulation of the immune system. We have learned a great deal from these models about the involvement of genetics, the microbiota and the role of different cells and the mucus layer in the development of IBD.

Here we discuss the major mistakes that are made using experimental colitis models, based on our own experience and the scientific literature. Recently increased awareness has developed for the necessity to improve the methodological quality of animal studies. 

Mistake 1 | Choosing an inadequate or obsolete model

Choosing the right mouse model is a major issue in studies of experimental colitis. In previous years some quite extensive overviews of the different models available have been published,3,6–8 but most authors refrain from giving advice on the best model to choose. Indeed, if any advice is given it is very limited. For example, Goyal et al.7 concluded that the “…currently available animal models are relevant to human IBD if they are chosen carefully (chronic, immune mediated)” and DeVoss and Diehl6 indicated that a successful approach requires “…careful utilization of pathway models to query specific scientific or efficacy questions.”

There are a number of chemically induced acute colitis models that are easy to use, rapid and of low cost and therefore widely used. However, these models may not be the best models to study IBD, because chemical damage to the gut epithelium results in self-limiting inflammation rather than chronic inflammation. Comparative analysis of colonic gene expression in the 2,4,6-trinitrobenzenesulfonic acid (TNBS), dextran sulfate sodium (DSS) and T-cell transfer models with human IBD revealed that the pattern of gene expression in the T-cell transfer model most closely reflects altered gene expression in IBD.9 Chemically induced models should only be used if the intention is to study the physiology of epithelial regeneration or intestinal wound healing.

In general, the different mouse models of colitis may reflect human IBD subtypes as described in table 1.  However, there is not one single experimental colitis model that resembles all aspects of human IBD. Making the choice of which model to use should combine the research question and the IBD subtype to achieve the best outcome. 

Mistake 2 | Not using standard protocols for induction of colitis consistently

In 2006, a critical appraisal of experimental colitis induction using TNBS revealed that the protocol followed differed in each of the studies included.10 Indeed, the mouse strains used, mouse age, dosing and times of TNBS administration, percentage of ethanol used and the duration of the experiment all varied. In 2007 Wirtz et al.11 published experimental protocols for the chemical induction of colitis in mouse models, thereby setting the gold standard for this methodology. Unfortunately, since then not many studies using these models seem to have followed the procedure described in this protocol.5 For the T-cell transfer model an excellent protocol, including critical parameters and troubleshooting, has been published in Current Protocols in Immunology12 and by Ostanin et al.13

The lack of consistency in the experimental protocols used for the chemical induction of colitis in mouse models hampers reproducibility, which is fundamental for any scientific experiment.14,15 In addition, standardization of environmental factors, such as circadian rhythms, nutrition, age, sex and strain are important confounders that have to be identified and acknowledged.16,17 To ensure consistency and reproducibility, the same protocol and environmental circumstances should be secured in every experiment and preferably shared by several laboratories.

Mistake 3 | Failing to randomly allocate animals to their experimental group

Randomly allocating animals to groups is a relevant issue when studying intestinal inflammation, because it ensures that subtle differences between the animals are unlikely to influence the experimental outcome. Usually, the body weight (or body weight change) of the animal, besides sex and age, is the most important parameter to account for in the randomization process. As the composition of the microbiome has a great influence on the development of intestinal inflammation,18 and this can differ between cages, the animals in each experimental group should be co-housed, so that representatives of the different experimental groups are within a single cage, and the groups are replicated across a series of cages.19 This way the differences between the groups and the cages can be measured independently.

In some cases it may be impossible to mix the experimental groups in one cage. For example, mice may not be mixed due to gender differences. Also, it is impossible to mix DSS animals and control animals (i.e. non-DSS-treated mice) in one cage. If this is the case it is recommended to randomly allocate the cages within the same room. In addition, when the animals are sacrificed the sequence should be randomized to avoid the introduction of bias. There are several ways to randomly allocate the animals used in an experiment. A random sequence generator for randomization of animals and the random integer set generator for randomization of an intervention can be found at www.random.org

Mistake 4 | Not blinding the study

Another important consideration when setting up an accurate animal experiment is that it should be blinded at several levels. The division between experimental and control groups should be blinded to avoid selection bias. The person who is responsible for the daily care of the animals should be unaware of the intervention(s) to avoid performance bias. Moreover, the people involved in determining the outcome parameters should not be aware of the intervention(s) to avoid detection bias. In general, blinding can be realised by having an independent outsider give each animal an individual mark/number coupled to the intervention(s) and only disclosing the mark/number at the end of the experiment. In general, the same care and quality control should be incorporated in animal experimentation as is customary in human clinical trials. Hooijmans et al. recently described a tool that can be used to assess the risk of bias for animal studies.20

Mistake 5 | Inadequate use of outcome parameters

As with research in human patients, one problem when using animal models is deciding what the most important disease parameter/primary endpoint is in fundamental and/or translational studies. Semi-quantitative evaluation of intestinal histo(patho)logy is considered to be the gold standard in animal models of intestinal inflammation. However, these histology-based scoring systems are not uniformly used in the literature. Most of these scoring systems include different sub-scores of histological aberrancies that are present in the animal model, such as crypt loss or immune cell infiltration.

For different models different sub-parameters are relevant, for example epithelial destruction in the DSS model, or epithelial hyperplasia in the T-cell transfer model.21 Therefore the most reliable scoring system for the model should be chosen. The slides should be blinded comprehensively without any reference to an individual animal or experimental group. As histopathological scores are given as an ordinal read-out (i.e. 0,1,2….) the median value is the most appropriate measure for central tendency within groups. This hampers the calculation of the number of animals needed per group, as the mean is used to calculate group sizes.22

Mistake 6 | Insufficient matching of control animals

When transgenic or knockout animals are used to study the effect of the transgenic/knockout genes wild-type animals are often used for comparison. As the microbiome has a great influence on the experimental outcome in these mouse models of IBD this has to be taken into account.18,23–25 There is good evidence that the composition of the microbiota differs in animals that are not co-housed. Jacobsson et al. observed that two C57BL/6 mice colonies maintained in different rooms at the same facility had a different gut microbiota.26 In addition, Ivanov et al. found that C57BL/6 mice obtained from different commercial vendors displayed differences in the numbers of Th17 T cells that could be related to the presence of specific bacterial taxa.27 This difference in microbiota was recently confirmed in another study for additional strains of mice.28 Another aspect that has to be taken in consideration is that certain drugs can have an effect on the microbiota composition and in this way affect disease development.

When using wild-type animals as controls in experiments with genetically modified mice, litter-mate wild-type animals should be used. As genetic modification can result in an altered microbiota,29 and this can be transferred to co-housed control mice together with increased susceptibility to colitis, the use of co-housing to ensure similar microbial composition should be done with precaution. To avoid the possible bias introduced by the microbial composition, models with specific microbiota can be used. Regardless, in the future it may be obligatory to characterize the microbiota in every study and incorporate this information into data evaluation.23

Mistake 7 | Not being aware of the susceptibility differences of the available mouse strains

One of the insights in IBD physiology reviewed by Bouma and Strober1 is that the host genetic background determines susceptibility to colitis. Various studies have described that the differences in susceptibility to chemically induced colitis is strain dependent.

The C3H/HeJ, C3H/HeJBir30 and C57BL/6 strains are highly susceptible to DSS-induced acute colitis, while BALB/c mice only develop colitis when higher percentages of DSS are administered.31 Also, the recovery phase of the disease after 5 days of administering DSS differs between C57BL/6 and BALB/c mice—C57BL/6 mice develop a severe chronic inflammation, whereas BALB/c mice resolve the colitis after the acute phase.31

In TNBS colitis the difference in susceptibility to colitis between SJL/J (susceptible) and C57BL/6 (resistant) mice is associated with the ability to mount an IL-12 response to lipopolysaccharide (LPS).32,33 IL-12 is the major cytokine for the differentiation of Th1-CD4+ T cells. For the mouse models in which T cells play a role it is important to realize that, in general, mice with a C57BL/6 background are more prone to develop a Th1 response, whereas BALB/c mice have a tendency to develop a Th2 response34 when exposed to pathogens.

In the T-cell transfer model mice both C57BL/613 and BALB/c35 backgrounds are used. In IL-10 knockout mice severe intestinal lesions develop in mice with a 129SvEv or BALB/c background, while C57BL/6 strains are relatively resistant to the development of colitis.36,37 In C57BL/6 mice colitis induction can be accelerated by peroral administration of piroxicam, a nonselective nonsteroidal anti-inflammatory drug (NSAID).38

To avoid the differences in susceptibility introduced by these extreme phenotypes, it might be an option to introduce the use of a collaborative cross-mouse genetic reference population as a new less biased resource in IBD research.39,40

Mistake 8 | Not being aware of the differences in disease susceptibility between the sexes

eereeFor most autoimmune diseases there is a clear difference in susceptibility between the sexes, with females more frequently affected than males.41 In experimental models of colitis sex-specific effects have also been described. For DSS colitis greater male susceptibility has been observed,30,42 and for TNBS the wasting disease has been shown to have a greater effect on female mice.33

Most experiments are performed with either male or female mice. However, in incidental experiments in which both sexes have been used,43–45 or a comparison was made between experiments,5 differences can be observed. In the T-cell transfer model both male13 and female35 mice are used. In general, DeVoss et al.6 recommend using female animals if possible. They indicate that male animals are more prone to display aggressive behaviour resulting in fighting, with the resulting stress and wounds potentially having a negative impact on a study. This finding hampers the random allocation of the mice because non-littermates cannot be housed together. However, single housing of male animals also has an effect on wellbeing46 and is expensive. In a study in which several aspects of the current usage of experimental colitis models was analysed, the predominant use of male animals was observed.5

Mistake 9 | Poor reporting quality

Experimental colitis models are frequently used to try to answer several biomedical research questions in IBD research. For successful translation of the knowledge from these studies to the clinic they should be well designed and reported, which does not seem to be the case.5,47

Quality assessment of animal experiments includes several different features and questions, and should at least include the items discussed previously. Is the research question specified and clear? Are animals randomly allocated across groups and is the outcome assessment randomly allocated across groups? Are the group characteristics clearly described and do they use a correct control group? Do they use a blinded outcome assessment? Is the timing clear? Which scoring system is used for histology? Are the treatment protocols clearly described? Are the number of animals per group clear and what is reported about the animals excluded from analysis? If mentioned, is it clear what the exclusion criteria are? Do the authors report complete outcome data?

Several tools are available to improve the reporting of outcomes in experimental colitis models. With the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines, which consists of a 20-item checklist, the reporting quality of all specific characteristics of the animals (including species, strain, sex, age, genetic background), housing details and methodology will be boosted. Encouragingly, more and more editors of scientific journals have adopted these guidelines and urge authors of submitted papers to use them.48 In addition, Bramhall et al. have published a checklist of essential and desirable criteria specified for reporting in animal models of colitis.47

Mistake 10 | Inadequate administration of therapeutic agents

 Experimental colitis models are frequently used for preclinical drug evaluation. The pharmacological approach is an important topic, and several aspects and considerations have been reviewed by Koboziev et al.49 Here, we focus on some of the main issues.

Of great importance is being aware that in chemical models of colitis the administered compound can potentially interfere with the DSS or TNBS and result in a reduced colitis induction. Also, in DSS colitis it must be confirmed that the treatment regimen does not influence the water consumption. So, this should be carefully monitored. Drugs can also have an effect on the microbiota composition and in this way affect disease development.

In experimental colitis models in which the induction of the colitis is fixed on a specific time point, as is the case in the chemically induced models, it is calculated that 78% of the treatments are applied before or within 24 h after the induction of colitis.5 In this situation it can be questioned whether a positive effect is due to actual treatment or interference with induction of colitis. It is actually essential to treat established disease. Koboziev et al49 indicate that “…one of the best predictors of clinical efficacy of a drug is its ability to reverse established disease in at least two different animal models of chronic intestinal inflammation.” This idea is also advocated in a recent commentary on reproducibility.17 With the introduction of endoscopy,50 researchers are able to investigate the effect on established disease more efficiently, enabling the comparison of disease characteristics (semi-quantitative score of endoscopy) before and after treatment for each individual animal (and do paired statistical analysis). 

Article information

© UEG 2016 Koelink and te Velde.

Cite this article as: Koelink PJ and te Velde AA. Mistakes in mouse models of IBD and how to avoid them. UEG Education 2016: 16: 11–14.

Correspondence to: a.a.tevelde@amc.nl

Acknowledgements: The authors would like to thank Dr. Manon Wildenberg for her helpful comments.

Conflicts of interest: The authors declare there are no conflicts of interest. 

A pdf of this article can be found in the UEG Education Library.

About the authors

Pim J. Koelink and Anje A. te Velde are at the Tytgat Institute for Liver and Intestinal Research. AMC, Amsterdam, the Netherlands.

Dr Koelink works in the group of Gijs van den Brink and focusses on animal models of chronic intestinal inflammation to unravel both fundamental and translational research questions.

Dr te Velde is a Principle Investigator in the Tytgat Institute. She is a trained immunologist with 20 years experience in experimental colitis models. Using this knowledge, her present focus is to disseminate information and increase public awareness of experimental colitis models. She is also interested in subtypes of IBD incorporated in disease-overarching, multidisciplinary research (she is a board member of the Immunowell Foundation, Immunowell.com).

References

  1. Bouma G and Strober W. The immunological and genetic basis of inflammatory bowel disease. Nat Rev Immunol 2003; 3: 521–533. 
  2. Khanna PV, et al. Use of animal models in elucidating disease pathogenesis in IBD. Semin Immunopathol 2014; 36: 541–551. 
  3. Mizoguchi A and Mizoguchi E. Animal models of IBD: linkage to human disease. Curr Opin Pharmacol 2010; 10: 578–587. 
  4. Uhlig HH and Powrie F. Mouse models of intestinal inflammation as tools to understand the pathogenesis of inflammatory bowel disease. Eur J Immunol 2009; 39: 2021–2026. 
  5. Zeeff SB, Kunne C, Bouma G, et al. Actual usage and quality of experimental colitis models in preclinical efficacy testing: a scoping review. Inflamm Bowel Dis Prepublished April 21, 20016, DOI: 10.1097/MIB.0000000000000758. 
  6. DeVoss J and Diehl L. Murine models of inflammatory bowel disease (IBD): challenges of modeling human disease. Toxicol Pathol 2014; 42: 99–110. 
  7. Goyal N, et al. Animal models of inflammatory bowel disease: a review. Inflammopharmacology 2014; 22: 219–33.
  8. Jones-Hall YL and Grisham MB. Immunopathological characterization of selected mouse models of inflammatory bowel disease: Comparison to human disease. Pathophysiology 2014; 21: 267–88. 
  9. te Velde AA, et al. Comparative analysis of colonic gene expression of three experimental colitis models mimicking inflammatory bowel disease. Inflamm Bowel Dis 2007; 13: 325–330. 
  10. te Velde AA, Verstege MI and Hommes DW. Critical appraisal of the current practice in murine TNBS-induced colitis. Inflamm Bowel Dis 2006; 12: 995–999. 
  11. Wirtz S, et al. Chemically induced mouse models of intestinal inflammation. Nat Protoc 2007; 2: 541–546. 
  12. Read S and Powrie F. Induction of inflammatory bowel disease in immunodeficient mice by depletion of regulatory T cells. Curr Protoc Immunol 1999; 30 (suppl): 15.13.1–15.13.10. 
  13. Ostanin DV, et al. T cell transfer model of chronic colitis: concepts, considerations, and tricks of the trade. Am J Physiol Gastrointest Liver Physiol 2009; 296: G135–G146. 
  14. National Academies of Sciences, Engineering, and Medicine. Reproducibility Issues in Research with Animals and Animal Models: Workshop in Brief. Washington, DC: The National Academies Press, 2015.
  15. Prinz F, Schlange T and Asadullah K. Believe it or not: how much can we rely on published data on potential drug targets? Nat Rev Drug Discov 2011; 10: 712. 
  16. Nature. Chow down. Nature 2016; 530: 254. 
  17. Reardon S. A mouse's house may ruin experiments. Nature 2016; 530: 264. 
  18. Gkouskou KK, et al. The gut microbiota in mouse models of inflammatory bowel disease. Front Cell Infect Microbiol 2014; 4: 28. 
  19. Macpherson AJ and McCoy KD. Standardised animal models of host microbial mutualism. Mucosal Immunol 2015; 8: 476–486. 
  20. Hooijmans CR, et al. SYRCLE's risk of bias tool for animal studies. BMC Med Res Methodol 2014; 14: 43. 
  21. Erben U, et al. A guide to histomorphological evaluation of intestinal inflammation in mouse models. Int J Clin Exp Pathol 2014; 7: 4557–4576. 
  22. Gibson-Corley KN, Olivier AK and Meyerholz DK.  Principles for valid histopathologic scoring in research. Vet Pathol 2013; 50: 1007–1015. 
  23. Hansen AK, et al. Impact of the gut microbiota on rodent models of human disease. World J Gastroenterol 2014; 20: 17727–17736. 
  24. Schoeb TR and Bullard DC. Microbial and histopathologic considerations in the use of mouse models of inflammatory bowel diseases. Inflamm Bowel Dis 2012; 18: 1558–1565. 
  25. Hansen AK, et al. A review of applied aspects of dealing with gut microbiota impact on rodent models. ILAR J, 2015; 56: 250–264. 
  26. Jakobsson HE, et al. The composition of the gut microbiota shapes the colon mucus barrier. EMBO Rep 2015; 16: 164–177. 
  27. Ivanov II, et al. Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell 2009; 139: 485–498. 
  28. Ericsson AC, et al. Effects of vendor and genetic background on the composition of the fecal microbiota of inbred mice. PLoS One 2015: 10: e0116704. 
  29. Zenewicz LA, et al. IL-22 deficiency alters colonoc microbiota to be transmissable and colitogenic. J Immunol 2013; 190: 5306–5312. 
  30. Mahler M, et al. Differential susceptibility of inbred mouse strains to dextran sulfate sodium-induced colitis. Am J Physiol 1998; 274: G544–G551. 
  31. Melgar S, Karlsson A and Michaelsson E. Acute colitis induced by dextran sulfate sodium progresses to chronicity in C57BL/6 but not in BALB/c mice: correlation between symptoms and inflammation. Am J Physiol Gastrointest Liver Physiol 2005; 288: G1328–G1338. 
  32. Scheiffele F and Fuss IJ. Induction of TNBS colitis in mice. Curr Protoc Immunol 2002; Chapter 15: Unit 15.19. 
  33. Bouma G, Kaushiva A and Strober W. Experimental murine colitis is regulated by two genetic loci, including one on chromosome 11 that regulates IL-12 responses. Gastroenterology 2002; 123: 554–565. 
  34. Hsieh CS, et al. T cell genetic background determines default T helper phenotype development in vitro. J Exp Med 1995; 181: 713–721. 
  35. Powrie F, et al. Inhibition of Th1 responses prevents inflammatory bowel disease in scid mice reconstituted with CD45RBhi CD4+ T cells. Immunity 1994; 1: 553–562. 
  36. Berg DJ, et al. Enterocolitis and colon cancer in interleukin-10-deficient mice are associated with aberrant cytokine production and CD4(+) TH1-like responses. J Clin Invest 1996; 98: 1010–1020. 
  37. Beckwith J, et al. Cdcs1, a major colitogenic locus in mice, regulates innate and adaptive immune response to enteric bacterial antigens. Gastroenterology 2005; 129: 1473–1484. 
  38. Holgersen K, et al. Characterisation of enterocolitis in the piroxicam-accelerated interleukin-10 knock out mouse—a model mimicking inflammatory bowel disease. J Crohns Colitis 2014; 8: 147–160. 
  39. Atamni HJ, et al. High-fat-diet induced development of increased fasting glucose levels and impaired response to intraperitoneal glucose challenge in the collaborative cross mouse genetic reference population. BMC Genet 2016; 17: 10. 
  40. Aylor DL, et al. Genetic analysis of complex traits in the emerging Collaborative Cross. Genome Res 2011; 21:  1213–1222. 
  41. Ngo ST, Steyn FJ and McCombe PA. Gender differences in autoimmune disease. Front Neuroendocrinol 2014; 35: 347–369. 
  42. Babickova J, et al. Sex differences in experimentally induced colitis in mice: a role for estrogens. Inflammation 2015; 38: 1996–2006. 
  43. te Velde AA, et al. Effects of dietary plant sterols and stanol esters with low- and high-fat diets in chronic and acute models for experimental colitis. Nutrients 2015; 7: 8518–8531. 
  44. Berglund M, et al. Gender dependent importance of IRAK-1 in dextran sulfate sodium induced colitis. Cell Immunol 2009; 259: 27–32. 
  45. Alex P, et al. Distinct cytokine patterns identified from multiplex profiles of murine DSS and TNBS-induced colitis. Inflamm Bowel Dis 2009; 15: 341–352. 
  46. Gonder JC and Laber K. A renewed look at laboratory rodent housing and management. ILAR J 2007; 48: 29–36. 
  47. Bramhall M, et al. Quality of methods reporting in animal models of colitis. Inflamm Bowel Dis 2015; 21: 1248–1259. 
  48. Kilkenny C, et al. Improving bioscience research reporting: the ARRIVE guidelines for reporting animal research. PLoS Biol 2010; 8: e1000412. 
  49. Koboziev I, et al. Pharmacological intervention studies using mouse models of the inflammatory bowel diseases: translating preclinical data into new drug therapies. Inflamm Bowel Dis 2011; 17: 1229–1245. 
  50. Becker C, et al. In vivo imaging of colitis and colon cancer development in mice using high resolution chromoendoscopy. Gut 2005; 54: 950–954. 
  51. Valatas V, Vakas M and Kolios G. The value of experimental models of colitis in predicting efficacy of biological therapies for inflammatory bowel diseases. Am J Physiol Gastrointest Liver Physiol 2013; 305: G763–G785. 
  52. Prattis S and Jurjus A. Spontaneous and transgenic rodent models of inflammatory bowel disease. Lab Anim Res 2015; 31: 47–68. 


 

Comments

Marta Afonso, October 18, 2016 11:00
A really good iniative. I would like to have access to a section in mistakes in hepatotology and in mouse models of liver disease.

Good job. :)
Wieland Hettrich, October 18, 2016 10:27
Useful update
Carolijn Smids, October 17, 2016 09:53
Very important and interesting article!! Notably, the time from drug administration until sacrification is often only a few hours, after which conclusions are sometimes drawn that there are no adverse events in the animals. However, the short observational period impede these statements.
Maurice DUBIAN, October 16, 2016 09:45
Good formations and interessting

More Education News