Mouse Models of Ulcerative Colitis: What Can They Really Tell Us?

Summary: Animal models of colitis are invaluable to ulcerative colitis (and more broadly, IBD) research, both in helping to elucidate potential disease mechanisms and in the process of testing and approving new therapies. A huge variety of mouse models of colitis have been developed, but they can be grouped into just a few categories based on the mechanism of disease induction, and a small handful of models are the most widely used by far.

Despite the importance of these models, no animal model yet exists that is an accurate mimic of human ulcerative colitis, and therefore each particular model and its limitations must be taken into account both when deciding which model to use and when drawing conclusions based on the findings.

This article is linked in the IBD Index as a special topic. Last updated on April 20, 2022.

Much of what we know about the etiology of ulcerative colitis (UC) comes from animal research, and we owe all of our current UC therapies to the intensive drug-approval process that begins with preclinical animal studies.

Models of colitis have been developed for many different lab animals from fruit flies (really) to pigs, but since mouse models are by far the most common, we’ll focus on those.

A semantic note: If you read any UC research in mice, you’ll notice that the term “ulcerative colitis” isn’t used to describe what’s happening in the mice; the more general term “colitis” is used instead. That’s because while the mice are exhibiting inflammation of the colon (ie, colitis), it’s not accurate to liken it to human ulcerative colitis. So yes, the title of this article is a bit of a misnomer. More on this below.

Table of Contents
Strengths and Weaknesses of the Most Common Mouse Models of Colitis
        DSS-Induced Colitis
        Oxazolone-Induced Colitis
        IL10 Knockout Colitis
        T-cell Transfer (Adoptive Transfer) Colitis
Development of Colitis in Mice is Highly Dependent on Gut Bacteria
Mouse Models of Colitis Have Limited Translational Relevance to Human Ulcerative Colitis

The earliest mouse models of UC were knockout mice: IL-10, IL-2, and TCRα knockouts, all developed in 1993. (1) Now, at least 66 different mouse models of IBD have been developed, but mouse models of colitis can be broadly categorized into three groups:

Chemically-induced colitis: Chemicals such as dextran sulfate sodium (DSS), 2,4,6-trinitrobenzene sulfonic acid (TNBS), oxazolone, or acetic acid are administered either rectally or in drinking water to induce colitis in wild-type or genetically engineered mice.
GEM models of spontaneous colitis: Colitis occurs spontaneously due to gene knockout and response to the microbiome. Examples of GEMs that spontaneously develop colitis include Il-10, Mdr1a, and TCRα knockout mice. (As you’ll see in the image below, this category can be further broken down based on whether the knocked-out gene affects immune function or epithelial barrier function.)
T-cell transfer colitis: Naïve T-cells are harvested from a donor mouse and injected into an immunodeficient mouse of the same strain background. The naïve T-cells don’t adequately differentiate into regulatory T-cells in the host mouse, leading to the development of spontaneous colitis mediated by effector T-cells. (4)

(Source: Taconic Biosciences, which is a provider of mouse models for human disease research.)

A reasonable chart showing some of the UC and Crohn’s mouse models, although this is seven years old at the time of this writing.
Source: Animal models of ulcerative colitis and their application in drug research, 2013

**Examples of GEM models** **(category 2 in above list)**. Forty different immune cell-specific KO mice (left) spontaneously develop intestinal inflammation due to dysregulated innate and adaptive immune responses. Eighteen different intestinal epithelial cell-specific KO mice (right) develop spontaneous colitis caused by impaired mucosal barriers, unregulated necroptosis/apoptosis, reduction of antibacterial peptide, and/or abnormal response to endoplasmic reticulum stress. (Source)

A couple things to note:

These mechanisms aren’t mutually exclusive. For example, piroxicam (an NSAID that causes oxidative stress) is often used to initiate colitis in IL-10 knockout mice, and DSS is often administered to genetically susceptible mice rather than wild-type mice.
Another mechanism that is sometimes used that isn’t included in the above is infection with specific strains of pathogenic or opportunistic bacteria (such as Salmonella typhimurium or adherent-invasive e. coli). This could conceivably fall under the first category, especially because additional treatment (such as antibiotics to disrupt the native microbiome, or a chemical irritant) is often necessary. (2)

Strengths and Weaknesses of the Most Common Mouse Models of Colitis

The sheer volume of different models that have been developed thus far should tell you that none alone are an adequate replica of human ulcerative colitis. Each has strengths and weaknesses, and if you want the full complexity you can check out the references I’ve linked throughout this article and listed at the bottom, but for the purposes of this article I’ll briefly review some of the strengths, weaknesses, and considerations of a few of the most widely used models.

Note: Although TNBS is very commonly used, it’s generally thought to be a better model for Crohn’s than UC, so I’ve left it out here.

DSS-Induced Colitis

This type of colitis is induced in either wild-type or genetically engineered mice by adding dextran sodium sulfate (DSS) to drinking water at concentrations typically around 2-5%. The accepted mechanism is that the DSS causes injury to the intestinal epithelium, causing bacterial antigens to come into contact with the immune cells in the lamina propria, eliciting an inflammatory response in the colon.

This is the most widely used model of colitis, yet one of the papers I used in researching this topic says “this colitis is a simple model of acute chemical injury rather than chronic inflammation, and researchers should not use it as a model for human UC.” (1) Indeed, mice begin to recover from their colitis once DSS is removed from their water, although a chronic model of colitis can be achieved by cycling DSS administration.

Not only is the initiating event (massive chemically-induced epithelial damage) not an accurate reflection of human UC, but the immune response seen in chronic DSS colitis is different – it’s characterized by a mixed T-cell response, where human colitis is characterized by polarized T-cell responses. (4) Of note, DSS colitis also appears even in the absence of adaptive (T-cell mediated) immunity, so it’s a helpful model for studying the role of innate immune cells (such as macrophages and neutrophils) in intestinal inflammation.

Despite perhaps not being an ideal proxy for human UC, the ease of administration can’t be beat, and the ability to control onset, duration, and severity of symptoms is a huge strength. Due to its etiology, it’s widely used to study epithelial healing post-injury, as well as innate immunity (as previously mentioned).

And although epithelial injury is still the widely accepted mechanism behind DSS-induced colitis, a paper from 2017 delving deeper into the etiology indicates that the mechanisms of action may be more complex than previously thought, and that the DSS model may provide valuable insight into mechanisms such as dysregulation of mucin and interaction with the intestinal microbiome and metabolome. (3)

Oxazolone-Induced Colitis

Oxazolone-induced colitis is an interesting one, because even a cursory look at the literature reveals disagreement as to its relevancy to human UC.

One 2015 paper states that it resembles human ulcerative colitis in both morphology and immunopathogenis. (4) You can see they even highlighted this in their graphic showing five of the common murine colitis models, which I included below.

Meanwhile, a more recent paper from 2018 questions its relevancy to human UC at all, saying that at best it perhaps could model extremely severe UC (as observed in patients with toxic megacolon), but is really more a model for sepsis than for UC. (6)

In an effort to stay out of the weeds, I’ll leave that discussion for another time (or maybe never), but at the very least, this is a good example of why any findings based on animal models of a human disease should be taken with a grain of salt.

IL10 Knockout Colitis

The IL10 knockout (IL10-/-) model was one of the earliest mouse models of colitis, developed by Taconic Biosciences and first reported in Kühn et. al in 1993, and is still the most widely used GEM model. (8) IL10 is produced by a number of immune cells, including macrophages, T-cells, and B-cells, and is an important anti-inflammatory and regulatory cytokine. Mice without it spontaneously develop chronic colitis, with onset timing and severity depending on genetic background, microbiome, and other environmental factors.

Importantly, as I’ll discuss below, spontaneous development of colitis does not happen in germ-free animals.

I couldn’t get a great read on it from the most recent literature I reviewed, but the general consensus seems to be that immune-mediated models of colitis such as this one and the cell-transfer model (discussed below) are more relevant to true UC than chemically-induced models. (7) And this makes sense, knowing that the generally accepted mechanism for human IBD is a dysregulated immune response to commensal intestinal flora. Of course, the trade-off is that these models are not nearly as controllable as models such as DSS, and they’re still a long way from accurately mimicking true human ulcerative colitis.

That said, the IL10-/- model in particular does have one potential causal similarity with human IBD: the presence of polymorphisms at the IL10 locus in humans has been identified as a risk factor for both UC and Crohns. (4)

T-cell Transfer (Adoptive Transfer) Colitis

As described above, in this model, naïve T-cells are injected into an immunodeficient mouse, and the lack of regulatory T-cells leads to the development of spontaneous colitis mediated by effector T-cells. (4) It was initially thought that inflammation in this model was limited to the colon, but it appears to result in inflammation of the small intestine as well, making this model also useful for Crohn’s disease research. (7)

T-cell transfer colitis is a much newer model of colitis compared with IL10-/- and DSS, but has quickly gained popularity. In particular, this model is helpful for studying the role of Treg cells in intestinal inflammation (which should come as no surprise given the disease mechanism). This model is also more reliable and replicable than GEM models like IL10-/-, and is not strain-specific (in other words, you can use it on any type of mouse). (7)

A graphic showing some of the common mouse models of colitis plus the area of research they’re most useful in. Note that the top three are chemically-induced and the bottom is a GEM model. Source: Experimental Models of Inflammatory Bowel Diseases, 2015

Development of Colitis in Mice is Highly Dependent on Gut Bacteria

If all these mouse models show us one thing about intestinal inflammation, it’s this: intestinal inflammation doesn’t occur in the absence of intestinal bacteria.

In all the reading I’ve done for this article, I haven’t come across a single model of colitis that develops in germ-free mice, and most have specified that enteric bacteria is necessary for the development of colitis in that model, including DSS, TNBS, IL-10 knockout, IL-2 knockout, Mdr1a knockout, and adoptive transfer.

DSS, to me, is a particularly interesting case. Germ-free mice given DSS exhibit massive intestinal bleeding and high mortality compared with conventional mice given DSS, but the bleeding isn’t from colitis – in fact, they have very little intestinal inflammation. (5) They do, however, have markedly weakened intestinal barrier function, so the DSS essentially rips apart their epithelium causing bleeding and death largely in the absence of an inflammatory response.

This indicates that enteric bacteria are necessary for intestinal inflammation (bad), but are also necessary for proper intestinal barrier function (good). This is a good example to keep in mind if you start to think “hey, if inflammation doesn’t occur in the absence of bacteria, can we just eliminate all the bacteria?” A) you probably couldn’t anyway, but B) if you did, your gut wouldn’t work right!

(And of course, intestinal inflammation itself isn’t exclusively “bad”; it’s part of our defense against invaders. It’s only bad in the modern context of chronic dysregulated immune responses.)

Mouse Models of Colitis Have Limited Translational Relevance to Human Ulcerative Colitis

Now for the relevant part. Namely: what does all this have to do with us? I’ll let a couple researchers speak for themselves, then share my thoughts.

No single model captures the complexity of human IBD, but each model provides valuable insights into one or another major aspect of disease, and together they have led to the establishment of now a generally accepted set of principles of human IBD pathogenesis. Chief among these are that (1) different causes of induced or genetically based inflammation give rise to a finite number of common pathways of immunopathogenesis; (2) normal resident gut microbiota can drive intestinal inflammation; (3) loss of oral tolerance and disruption of the epithelial barrier contribute to the development of intestinal inflammation; and (4) polarized T helper (T_H) cell responses as well as defects in innate immunity mediate disease.”
–Kiesler et al. 2015. “Experimental Models of Inflammatory Bowel Diseases”

Certainly, there are limitations inferring observations from animal models. Particularly, human IBD is polygenic and multifactorial in nature, in contrast to experimental IBD animal model that usually involves disruption or overexpression of single genes and thus limiting systemic pathways. Another limitation of extrapolating animal models into human safety and efficacy is the qualitative and quantitative measurement of outcomes. Human clinical trials for IBD measures disease activities through various indexes that quantify symptoms related to active disease, including the CDAI and the Mayo Clinic score for UC. In contrast, efficacies of interventions in animal models are usually assessed through histopathology score of colonic inflammation and clinical score. These parameters might not be sufficient to make inference to clinical remission in heterogeneous groups of patients.”
Mizoguchi et al. 2020. “Recent updates on the basic mechanisms and pathogenesis of inflammatory bowel diseases in experimental animal models”

While there are many in vivo models of IBD, none adequately predicts response to therapeutics.”
Pizarro et al. 2019. “Challenges in IBD Research: Preclinical Human IBD Mechanisms”

As an example of this, biologics that target TNF-alpha (including popular drugs like Remicade, Humira, etc) initially didn’t show promising results in animal models; these treatments failed to show benefit in either acute-DSS and IL10-/- models, although a chronic DSS model and several others later showed promise. And of course, human RCTs were successful enough that this class of drugs is now a mainstay in the treatment of UC.

There’s also the example of IL-17 inhibition discussed here, which was extremely promising in mouse models but failed clinically (and may even worsen Crohn’s disease).

All that said, even based on the relatively cursory summary I’ve given in this article, it should be apparent that all of these models are contrived in a way that isn’t reflective of human UC. In other words, ulcerative colitis in humans isn’t caused by intentional chemical injury, full gene knockouts, or the expansion of immature T-cells in an immunodeficient host. So of course the broad translational relevance of these models is going to be extremely limited!

This table on the Taconic Biosciences website gives an overview of several of the popular mouse models of colitis, including most I’ve discussed here. It’s a helpful overview in many respects, but I’d like to draw your attention to the “Translational relevance” row in particular: you can see that for the DSS and adoptive transfer models, there is “no evidence on the relevance to human disease.”

Think about that for a minute. For two of the most commonly-used animal models of colitis, there’s no evidence you can point to to confidently say that a finding based on those animal models will be relevant in humans.

I emphasize this not to diminish the contributions of animal research to our current understanding of ulcerative colitis, which is substantial, but to give you a realistic picture of what IBD animal research looks like so you can take in (and pass on) information responsibly, and not be bamboozled by media headlines (or even research summaries) that try to whip up excitement through irresponsible translation and extrapolation of animal results to human disease.

I’ll close with another Pizarro et al. quote that sums things up pretty well: “Every in vivo model has its strengths and weaknesses, and the key to robust scientific investigation remains the selection of the correct model for the scientific question under inquiry with focus on mechanistic questions rather than direct translatability to efficacy in the clinic.” (9)

Mouse Models of Ulcerative Colitis: What Can They Really Tell Us?