A chemical enterocolitis model in zebrafish larvae that is dependent on microbiota and responsive to pharmacological agents


  • Stefan H. Oehlers,

    1. Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
    Search for more papers by this author
  • Maria Vega Flores,

    1. Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
    Search for more papers by this author
  • Kazuhide S. Okuda,

    1. Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
    Search for more papers by this author
  • Chris J. Hall,

    1. Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
    Search for more papers by this author
  • Kathryn E. Crosier,

    1. Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
    Search for more papers by this author
  • Philip S. Crosier

    Corresponding author
    1. Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Auckland, New Zealand
    • Department of Molecular Medicine and Pathology, School of Medical Sciences, The University of Auckland, Private Bag 92019, Auckland, New Zealand
    Search for more papers by this author


Inflammatory bowel disease (IBD) results from dysfunctional interactions between the intestinal immune system and microbiota, influenced by host genetic susceptibility. Because a key feature of the pathology is intestinal epithelial damage, potential disease factors have been traditionally analyzed within the background of chemical colitis models in mice. The zebrafish has greatly complemented the mouse for modeling aspects of disease processes, with an advantage for high content drug screens. Larval zebrafish exposed to the haptenizing agent trinitrobenzene sulfonic acid (TNBS) displayed impaired intestinal homeostasis and inflammation reminiscent of human IBD. There was a marked induction of pro-inflammatory cytokines, the degradative enzyme mmp9 and leukocytosis. Enterocolitis was dependent on microbiota and Toll-like receptor signaling, that can be ameliorated by antibiotic and anti-inflammatory drug treatments. This system will be useful to rapidly interrogate in vivo the biological significance of the IBD candidate genes so far identified and to carry out pharmacological modifier screens. Developmental Dynamics, 2011. © 2010 Wiley-Liss, Inc.


Intestinal epithelial damage is thought to be important for the development and maintenance of inflammatory bowel disease (IBD) in humans and for colitis in animal models (Sansonetti,2004). The intestinal epithelium provides a physical barrier between the intestinal lumen and sterile tissues, as well as an immunological barrier capable of sensing and responding to microbial incursion. It has been proposed that loss of these two functions through combinations of genetic (host-derived) and environmental (mainly microbial-derived) events underlies the etiology of IBD. Human IBD is a complex genetic disease with multiple hits seemingly required during its pathogenesis. The use of animal models of spontaneous and induced colitis has been, and will continue to be, pivotal in ascribing function to novel IBD candidate genes (Braun and Wei,2007). Furthermore, an animal model where genetic and environmental determinants of disease are tractable in a medium- to large-scale context would be ideal for screening novel disease modifiers.

The zebrafish (Danio rerio) model system, an established developmental biology model, has recently come to the fore in the fields of microbial and mucosal immunity (Meeker and Trede,2008; Sullivan and Kim,2008). Larval zebrafish are experimentally accessible for the study of mucosal immunology because of their transparency and the production of large numbers of embryos. Research into zebrafish gastrointestinal function has identified many orthologous gene and signaling pathways necessary for digestive and mucosal immune function (Wallace and Pack,2003; Ng et al.,2005; Bates et al.,2007; Flores et al.,2008).

Given the impressive depth and scope of current IBD research, the zebrafish is unlikely to replace current mammalian models of colitis for studies of disease development and progress (Strober et al.,2002; Brugman et al.,2009; Fleming et al.,2010). The zebrafish platform is instead proposed to complement existing research pathways in subfields of IBD research such as candidate gene function characterization, real time analysis of host–microbe interactions and, importantly, in drug screening for disease modifiers.

While recent advances in immune modulation have resulted in the application of biological therapies to effect and maintain remission in active IBD, conventional anti-inflammatory drugs such as 5-aminosalicylic acid (5-ASA) and prednisone remain important treatments (Rutgeerts et al.,2009). However, as many patients are refractory to current classes of therapeutics there is a need to develop novel anti-inflammatory drugs with relevance to enterocolitis (Priest et al.,2006). Zebrafish models have been successfully used to identify small molecules with therapeutic application, including oncogene suppression and hematopoietic stem cell expansion (North et al.,2007; Yeh et al.,2009).

The IBD genetics field is identifying new candidate genes linked to disease phenotypes in subpopulations of patients around the world (Budarf et al.,2009). There is a need for approaches that use whole animal in vivo functional studies. As the larval zebrafish is genetically tractable, it represents a model that will permit the rapid in vivo analysis of IBD candidate gene function.

A recent study has explored the feasibility of the larval zebrafish as a model for IBD for high-content drug screening (Fleming et al.,2010). In this model, enterocolitis was assessed by examination of gut morphology and function. In our current study, we analyzed the zebrafish larval trinitrobenzene sulfonic acid (TNBS)-induced enterocolitis in greater detail and demonstrate that the model recapitulates hallmark aspects of human IBD including induction of pro-inflammatory pathways and degradative enzymes, and leukocytosis around the intestine. Importantly, because of the contribution of genetic and microbial components to IBD etiology, we combined bacterial and genetic manipulation to investigate the genetic pathways involved in pathogenesis. Furthermore, as in human IBD, these TNBS-induced effects could be prevented or diminished by administration of antibiotics or anti-inflammatory drugs. The data generated in the development of this model is also applicable to the general study of intestinal development and function.


Histological Analysis of TNBS-Induced Enterocolitis

A range of TNBS doses were investigated with the aim of establishing a dose that would not cause overt morphological changes to larvae while reproducibly inducing inflammation (Fleming et al.,2010; Oehlers et al.,2010). Initial titration studies showed a dose of up to 50 μg/ml TNBS could be tolerated for three continuous days with greater than 90% cohort survival while doses of 75 μg/ml or greater resulted in less than 50% survival by three post-exposure (data not shown).

While it is possible to control the route of exposure in mammalian models, exposure of zebrafish larvae will lead to potential interactions with any epithelial surface (O'Toole et al.,2004; Pressley et al.,2005). Our first analysis used Neutral Red vital staining to detect damage to the skin (Herbomel et al.,2001; McLeish et al.,2010). Larvae exposed to 75 μg/ml TNBS, the highest dose investigated, for 3 days demonstrated widespread skin damage (Fig. 1A). The 75 μg/ml dose for 3 days appeared to be the threshold dose and exposure length for the manifestation of skin damage as larvae exposed to lower doses of TNBS or to the 75 μg/ml dose for duration less than three days did not manifest any Neutral Red staining of the skin (Fig. 1A and Supp. Figs. S1A, S2A, which are available online).

Figure 1.

Histological analysis of trinitrobenzene sulfonic acid (TNBS) -induced enterocolitis. A: Transmitted light microscopy of 6 days postfertilization (dpf) larvae stained with Neutral Red. Red lines indicate regions sampled by transverse sectioning. Arrowheads in the bottom half of the panel indicate ectopic Neutral Red staining. B: Alcian blue stained 6 dpf larvae. Arrowheads in the bottom half of the panel indicate ectopic Alcian blue staining. C: Transverse sections stained with hematoxylin and eosin, anatomical locations are described in Panel A. Scale bar = 1 mm in A,B, 50 μm in C.

Changes to mucus producing cells have been associated with both TNBS-induced enterocolitis and skin damage (Iger et al.,1994; Fleming et al.,2010). We analyzed the same series of TNBS doses with whole-mount Alcian blue staining to detect acidic mucus producing cells. Contrary to previous characterization, we did not observe any increase in Goblet cells at any of the time points or doses that we analyzed (Fig. 1B and Supp. Figs. S1B, S2B). However, we observed an increase in skin Alcian blue staining in the larvae exposed to 75 μg/ml TNBS for three days with a similar distribution to the Neutral Red staining.

Exposure to a high dose of TNBS for 5 days has been documented to cause gross changes to intestinal morphology (Fleming et al.,2010). However, with the doses and durations that we analyzed, we did not observe any TNBS-induced changes to intestinal cell morphology (Fig. 1C and Supp. Figs. S1C, S2C).

Based on larval survival and the appearance of skin damage, we carried out subsequent analyzes at 3 days of exposure to 50 μg/ml TNBS, unless otherwise indicated. Development of the eye, containing rapidly proliferative tissue, was used as an indicator of extra-intestinal effects of TNBS exposure on growth rate (de Jong-Curtain et al.,2009; Parichy et al.,2009). Measurement of larval eye diameter revealed no significant difference between TNBS-treated and control samples (97.4% ± 1.0% and 100% ± 1.2%, respectively, P = 0.10).

TNBS Haptenization Is Concentrated to the Intestinal Tract

Live imaging under trans-illumination showed that TNBS was concentrated within the intestinal tract of TNBS-exposed larvae. Increasing concentrations of TNBS were associated with greater abundance of luminal yellow substance in the larval intestinal tract (Fig. 2A). As increased bile secretion may be responsible for the observed yellow substance, we incubated larvae with 22-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-23,24-bisnor-5-cholen-3-ol (NBD)-cholesterol to visualize biliary function (Farber et al.,2001). We observed similar distribution of fluorescence in control and TNBS-exposed larvae (Supp. Fig. S3A).

Figure 2.

Trinitrobenzene sulfonic acid (TNBS) haptenization is concentrated to the intestinal tract and also causes extra-intestinal effects. A: Transmitted light microscopy of 6 days postfertilization (dpf) larvae exposed to increasing concentrations of TNBS. B: Confocal microscope images of whole mount immunofluorescent detection of TNBS hapenization by anti-dinitrophenyl (DNP) antibody in TG(ifabp:RFP) larvae (left: control, middle: high haptenization, right: low haptenization). C: Immunofluorescent detection of TNBS haptenization in a physical section of the intestinal bulb of a larva exposed to TNBS. Slides were counterstained with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) to mark nuclei. D: Comparison of 5-bromo-2′-deoxyuridine (BrdU) incorporation in 6 dpf larvae following TNBS exposure. Counts were performed as described in the methods (n = 10, P values calculated by t-test, error bars represent a 95% confidence interval of the mean). E: Analysis of pro-inflammatory cytokine expression in 6 dpf larval dissected intestines and larval carcasses, measured by quantitative polymerase chain reaction (qPCR) after TNBS exposure and compared with controls. P values calculated by one-sided t-test against a hypothetic value of 1, error bars represent a 95% confidence interval of the mean, n = 3, pools of 15 larvae. *P < 0.05, **P < 0.01, ***P < 0.001.

To define the extent of TNBS haptenization on larvae, spatial localization of trinitrophenyl haptens was analyzed using an anti-dinitrophenyl fluorescently-tagged antibody (Comoglio et al.,1975). Whereas a marked fluorescent signal was detected within the intestinal region in TNBS-exposed larvae, this was absent in control samples (Fig. 2B). As also shown in Figure 2B, a weaker and variable amount of epidermal TNBS haptenization was detected in some TNBS-exposed samples. The location of fluorescent signal in the TNBS-exposed larvae was confirmed by examination of sections through the intestine (Fig. 2C).

TNBS Exposure Primarily Evokes an Inflammatory Response Within the Intestine

As a global increase in proliferating cell nuclear antigen (pcna) transcription was observed following TNBS exposure (Supp. Fig. S3B), we undertook analysis of proliferation using bromodeoxyuridine (BrdU) labeling to visualize the tissue distribution of proliferation in larvae. A significant increase in the number of proliferating cells in both the gut and trunk of TNBS-exposed larvae was observed (Fig. 2D). BrdU-positive cells were observed in the intestinal epithelium (Supp. Fig. S3C) and throughout the radial axis of the trunk (data not shown). TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) staining was carried out to examine the prevalence of apoptotic cell death. Apoptosis was unchanged following exposure to TNBS (Supp. Fig. S3D).

Previous study of the TNBS-exposure model had revealed an up-regulation of pro-inflammatory cytokine genes at the whole organism level (Oehlers et al.,2010), we carried out dissections to resolve the relative responsiveness of the body and gut to TNBS exposure. We found both compartments significantly up-regulated transcription of il-1b (Fig. 2E). We did not observe an induction of ccl20 transcription in the gut of zebrafish larvae exposed to TNBS, in contrast with that reported from studies with IBD patients (Kwon et al.,2002; Kaser et al.,2004). Transcription of the chemotactic cytokine il-8, known to be up-regulated in IBD (Daig et al.,1996), was significantly increased in the intestine but not the body of larval zebrafish exposed to TNBS.

To determine the effect of TNBS exposure on intestinal epithelial cell gene expression, we analyzed a panel of intestinal homeostatic and enterocolitis-related genes by quantitative polymerase chain reaction (qPCR). Despite the observed global increase in cell proliferation upon exposure to TNBS, expression of markers of intestinal epithelial differentiation (barx1, cdx1b, foxa3, gata5, gata6, and sox2), hedgehog signaling (ihha, shha, and shhb), peroxisome proliferator-activated receptor (pparda, ppardb, and pparg), serotonin reuptake (serta and sertb), and a second housekeeping gene (sdh) were found to be unaffected (Supp. Fig. S4A). The expression of intestinal fatty acid binding protein (ifabp or fabp2) and fatty acid binding protein 4 (fabp4) were also found to be unaffected by TNBS exposure. However, a statistically significant reduction in fabp6 expression was observed, suggesting altered lipid metabolism after TNBS exposure (Fig. 3A). This potential defect was further investigated with Nile Red staining, a vital dye for lipids (Jones et al.,2008). Exposure to TNBS resulted in increased retention of Nile Red (Fig. 3B and see Supp. Figs. S4B, S4C).

Figure 3.

Changes to intestinal physiology caused by trinitrobenzene sulfonic acid (TNBS) exposure. A: Analysis of the fatty acid binding protein family expression measured by quantitative polymerase chain reaction (qPCR) after TNBS exposure and compared with control (n = 10, pools of 20 larvae, P value calculated by one-sided t-test against a hypothetical mean of 1). B: Nile Red fluorescence in larvae exposed to TNBS expressed as a percentage of signal detected in control larvae (n = 10, pools of 5 larvae, P value calculated by t-test). C,D: Comparison of mid-intestinal markers in control (left column) and TNBS-exposed (right column) larvae, in situ hybridization for fabp6 transcripts (C), bar indicates length of staining, and neutral red staining of highly endocytic epithelium (D). Scale bar represents 100 μm. E: Quantification of 6 days postfertilization (dpf) larval mid-intestinal length marked by neutral red staining. All error bars represent a 95% confidence interval of the mean. *P < 0.05, ***P < 0.001.

Reduced spatial expression of fabp6 transcripts were observed by in situ hybridization suggesting shortening of the larval mid-intestine (Fig. 3C). As further evidence that a specific segment of the gut manifested this phenotype, we used the vital dye neutral red to stain the highly endocytic epithelium of the larval mid-intestine (Fig. 3D). Quantification of mid-intestine length demonstrated significant shortening of the larval mid-intestine by exposure to TNBS (Fig. 2E).

TNBS Exposure Affects the Development of Intestinal Vasculature

To examine the effects of TNBS exposure on other features of intestinal development, we analyzed the formation of the intestinal vasculature.

Intestinal vascularization may be altered in abnormalities of the gut, including inflammatory disorders. We, therefore, examined the effects of TNBS on vasculature using a TG (fli1:EGFP)γ1 transgenic zebrafish line (Lawson and Weinstein,2002). As we did not observe any changes to the architecture of the intestinal vasculature after TNBS exposure (Fig. 4A,B), we enumerated the number of subintestinal branches (between the supraintestinal artery and the subintestinal vessel) along the length of the intestinal bulb (Fig. 4C). There was a consistent reduction in the number of branches in TNBS-treated larvae compared with controls at 6 days postfertilization (dpf). No change in the patterning or number of trunk vessels was observed after exposure to TNBS. To determine if reduced pro-angiogenic signaling played a role in the reduced intestinal vascularization, we analyzed the expression of the major angiogenic factor, vascular endothelial growth factor a (vegfa) by qPCR. This analysis revealed a decrease in expression after exposure to TNBS (Fig. 4D). Expression of the vascular-specific transcription factor fli1 was unchanged by exposure to TNBS.

Figure 4.

Intestinal vasculature is disrupted by trinitrobenzene sulfonic acid (TNBS) exposure. A: Comparison of vasculature in live 6 days postfertilization (dpf) TG(fli1a:EGFP)γ1 transgenic larvae imaged under epifluorescence with a GFP filter and (B) representative close up images. Scale bars represent 100 μm. C: Quantification of intestinal bulb vasculature at 6 dpf; mean numbers of subintestinal branches as indicated (n = 45). D: Quantitative polymerase chain reaction analysis of fli1 and vegfa transcription in 6 dpf larvae exposed to TNBS, shown relative to controls (RQ = 1; n = 10, pools of 20 larvae). All error bars represent a 95% confidence interval of the mean, P values calculated by t-test. **P < 0.01, ***P < 0.001.

Zebrafish Larvae Mount a Leukocytic Response During TNBS-Induced Inflammation

Leukocytosis is a recognized component of enterocolitis pathogenesis and has been previously noted in an adult zebrafish enterocolitis model (Meuret et al.,1978; Heits et al.,1999; Brugman et al.,2009). Live imaging studies were performed to analyze the distribution of leukocytes after exposure to TNBS. TG(mpx:EGFP)Fig. zebrafish demonstrated a significant redistribution of enhanced green fluorescent protein (EGFP) -positive cells from the caudal hematopoietic tissue (CHT) region to the periphery (Fig. 5A). Confocal live imaging of double transgenic TG(mpx:EGFP114, ifabp:RFP) zebrafish confirmed an accumulation of EGFP-positive cells around the intestine (Fig. 5B,C, Supp. Figs. S5, S6).

Figure 5.

Zebrafish larvae mount a leukocytic response to trinitrobenzene sulfonic acid (TNBS) exposure. Left column is control and right column is TNBS-exposed. A: Comparison of neutrophil distribution in 6 days postfertilization (dpf) TG(mpx:EGFP)114. Larvae exposed to TNBS demonstrated a redistribution of leukocytes away from the CHT region. B: Comparison of confocal microscopy images of double transgenic TG(mpx:EGFP114, ifabp:RFP) larvae expressing EGFP in neutrophils and RFP in intestinal epithelial cells. C: Three-dimensional reconstruction of confocal microscopy data illustrating two angles. Scale bars in A and B represent 100 μm. CHT, caudal hematopoietic tissue. EGFP, enhanced green fluorescent protein. RFP, red fluorescent protein.

Toll-like Receptor (TLR) Adaptor myd88 Is Necessary for Zebrafish Larval Resilience to TNBS Challenge

The activation of innate immune signaling pathways during chemically-induced enterocolitis is necessary to limit both direct cellular damage and bacterial overgrowth larvae is unknown, we used morpholino injection to reduce the expression of Myd88, the major adaptor molecule for TLR signaling. Experimental knockdown of Myd88 did not affect larval survival or development (van der Sar et al.,2006). However, exposure to TNBS at a concentration of 75 μg/ml resulted in a significant decrease in survival compared with control larvae exposed to TNBS (Fig. 6A). Myd88 knockdown larvae exposed to TNBS were more susceptible to mortality over the first 24 to 48 hr of TNBS exposure. Similar rates of mortality between knockdown and control larvae were observed thereafter (data not shown).

Figure 6.

The Toll-like signaling adaptor molecule Myd88 and antibiotics protect against trinitrobenzene sulfonic acid (TNBS) -induced mortality. A: Comparison of Myd88 morphant survival with that of control larvae exposed to TNBS at a concentration of 75 μg/ml for 48 hr. Survival is expressed as a percentage of the original cohort (n = 5, groups of 20 larvae). B: Time course comparison of survival between larvae co-treatment with broad-spectrum antibiotics and control larvae exposed TNBS at a concentration of 100 μg/ml (n = 4, groups of 20 larvae). C: Quantitative polymerase chain reaction analysis of pro-inflammatory cytokine transcription in larvae exposed to TNBS with or without antibiotic administration (n = 5, pools of 20 larvae). All error bars represent at 95% confidence interval of the mean, P values calculated by t-test. *P < 0.05, **P < 0.01, ***P < 0.001.

Microbiota Is Necessary for the Induction of Inflammation by TNBS Exposure

To determine if increased TNBS-induced mortality in Myd88 morphants might be due to a loss of host–microbiota homeostasis, zebrafish larvae were treated with broad-spectrum antibiotics (ampicillin and kanamycin). While high mortality was observed in the conventionally-reared cohort exposed to TNBS (100 μg/ml), antibiotic administration before the addition of TNBS into the media resulted in increased survival (Fig. 6B). Administration of antibiotics to Myd88 morphants increased their survival rates to control levels (data not shown). Plating of E3 media onto Luria agar for microbiota enumeration confirmed sensitivity of the microbiota to the antibiotics used with typical colony forming unit (CFU) counts reduced from 106 CFU/ml to 104 CFU/ml or lower.

To exclude any direct effect of antibiotic administration on the activity of TNBS, immunofluorescent detection was used to quantify the level of TNBS haptenization in unprotected and antibiotic-protected larvae. There was no discernable difference in signal strength between unprotected TNBS-exposed and antibiotic-treated TNBS-exposed larvae (Supp. Fig. S7).

Previous analysis of inflammatory cytokine production in zebrafish larvae exposed to TNBS used a TNF-α antibody method (Fleming et al.,2010). Because of the substantial difference in the biological role of TNF-α in mammals and zebrafish, we examined the transcription of a panel of pro-inflammatory cytokines as a more accurate measure of inflammation (Roca et al.,2008). Consistent with previous studies of zebrafish larval intestinal infection, an early induction of pro-inflammatory cytokine transcription has been observed in this model (Oehlers et al.,2010). Because antibiotics were able to protect larvae from TNBS-related mortality, we investigated the effect of antibiotic administration on the induction of a pro-inflammatory immune response. Compared with conventionally-reared larvae exposed to TNBS, those co-treated with ampicillin and kanamycin did not initiate the transcription of pro-inflammatory cytokines il-1b, tnf-a, ccl20, and il-8 (Fig. 6C).

Administration of Pharmacological Agents Can Prevent and Ameliorate the Induction of Inflammatory Markers Following TNBS Exposure

As TNBS exposure caused inflammation in a microbiota-dependent manner (relevant to human IBD), we next tested the responsiveness of the model to human IBD medications. Anti-inflammatory agents such as 5-ASA and prednisolone are important therapies for human IBD and administration of either drug immediately before the addition of TNBS reduced the induction of pro-inflammatory cytokines in response to TNBS (Fig. 7A,B). The effectiveness of anti-inflammatory drugs within the context of established inflammation was assessed by the administration of anti-inflammatory drugs to 5 dpf larvae (exposed to TNBS for 2 days) for a period of 24 hr (Oehlers et al.,2010). We found this 24-hr or “overnight” treatment with either drug to be effective in reducing the expression of pro-inflammatory cytokines (Fig. 7C,D).

Figure 7.

Administration of pharmacological agents to modulate inflammation. A,B: Quantitative polymerase chain reaction analysis of pro-inflammatory cytokine transcription in larvae exposed to trinitrobenzene sulfonic acid (TNBS) with or without continuous (cont.) administration of 5-aminosalicylic acid (5-ASA; A) or prednisolone (B). C,D: Treated overnight (O/N) with 5-ASA or prednisolone, respectively (n ≥ 3, pools of 10 to 20 larvae). E: Fluorescence-activated cell sorting (FACS) analysis of green fluorescent protein (GFP) -positive cell counts of 6 days postfertilization (dpf) TG(lyzC:EGFP) 117 larvae expressing enhanced GFP (EGFP) in leukocytes (n ≥ 3, pools of 10 to 20 larvae). F: Quantitative polymerase chain reaction analysis of matrix metalloproteinase (mmp) 9 transcription in 6 dpf larvae exposed to TNBS with or without broad-spectrum antibiotics or 5-ASA or prednisolone (n ≥ 3, pools of 10 to 20 larvae). G: Enumeration of intestinal leukocyte recruitment in Tg(mpx:EGFP)114 larvae (n = at least 5 per treatment). All error bars represent a 95% confidence interval of the mean, P values calculated by matched t-test. *P < 0.05, **P < 0.01, ***P < 0.001.

To determine if the redistribution of leukocytes around the inflamed intestinal region was also associated with an increase in leukocyte numbers, fluorescence-activated cell sorting (FACS) analysis of transgenic TG(mpx: EGFP)114 and TG(lyz:EGFP) 117 larvae following relevant exposures was undertaken (Supp. Figs. S8A, S8B). These transgenic lines have marked cells within the myeloid compartment. GFP-positive counts were increased in both TG(mpx:EGFP) 114 and TG(lyz:EGFP) 117 lines after TNBS exposure (Supp. Fig. S8C). Co-incubation with 5-ASA, but not prednisolone, prevented an increase in lyz:EGFP marked GFP-positive cells after TNBS exposure (Fig. 7E).

Macrophages and neutrophils are potent sources of cytokines and tissue-destructive enzymes such as matrix metalloproteinases (MMP). MMP9 has been identified as a mediator of intestinal damage in colitis models (Garg et al.,2009). In zebrafish larvae exposed to TNBS, an increase in the transcription of mmp9 was observed from 24 hr after exposure, whole-mount in situ hybridization analysis of expression demonstrated increased mmp9 expression in epidermal tissues (Supp. Fig. S8D). Induction of mmp9 expression was reduced by the administration of antibiotics or anti-inflammatory drugs (Fig. 7F).

To illustrate the usefulness of the transgenic zebrafish lines for performing chemical screens, we enumerated the recruitment of leukocytes to the larval intestine using Tg(mpx:EGFP) 114 larvae. Consistent with that illustrated in Figure 5, a clear increase in leukocytes recruited to the intestine was observed (Fig. 7G and see Supp. Fig. S9). Co-administration of antibiotics, 5-ASA or prednisolone significantly reduced the recruitment of leukocytes to the intestine and skin (for representative images, see Supp. Fig. S9). Administration of prednisolone, but not 5-ASA, after the induction of inflammation was similarly successful at reducing the recruitment of leukocytes to the intestine and skin.


We describe a new technology platform for the rapid analysis of intestinal epithelial damage, using the zebrafish model organism. Biologically important parameters for modeling human IBD include the necessity of the microbiota to trigger inflammation, production of pro-inflammatory cytokines, recruitment of leukocytes and the loss of intestinal function. This study presents a detailed characterization of the inflammatory events following TNBS exposure in larval zebrafish and provides evidence supporting the development of intestinal inflammation. The results establish a basis for using larval zebrafish to explore genetic and environmental factors involved in the response to epithelial damage with potential relevance to human IBD.

Data from murine models of colitis have suggested that the microbiota is beneficial in providing resilience to acute intestinal injury but detrimental to chronic intestinal injury as seen in IBD (Rakoff-Nahoum et al.,2004; Prantera,2008). The role of the microbiota in the larval zebrafish model has not been previously described. Our findings show this larval zebrafish model shares important parallels with enterocolitis described in adult zebrafish and mammalian systems (Sansonetti,2004; Bates et al.,2007; Prantera,2008; Brugman et al.,2009).

The development of robust visual phenotypes in an optically transparent model system makes this model an attractive platform for carrying out relatively large-scale screens for chemical or genetic modifiers of enterocolitis in vivo (Kaufman et al.,2009). Because the pathological changes in leukocyte distribution correlated with pro-inflammatory cytokine production, transgenic lines could be used as a “live indicator” of inflammation in larval zebrafish. Furthermore, the ability to alter the inflammatory status of the zebrafish gastrointestinal tract independent of an introduced pathogen will allow the comparison of host–microbe interactions between physiological and pathological host backgrounds in real time, a parameter presently absent from mammalian studies (Rawls et al.,2007).

Subtle changes to intestinal function were noted including altered lipid metabolism, decreased intestinal vasculature, increased leukocyte recruitment and increased cellular proliferation without a concomitant increase in cell death. Thus this model may provide an appropriate platform to probe the intersection of metabolism and immune function using small molecules and genetics. However, it is known that the larval zebrafish gut undergoes significant morphological maturation after 3 dpf as the linear gut tube undergoes convolution and folding (Wallace and Pack,2003; Ng et al.,2005). As such, it may not be appropriate to analyze inflammation-driven changes of intestinal differentiation within such a developmentally active system (Lees et al.,2008; Noble et al.,2008).

A dysfunctional intestinal vasculature has been implicated in the maintenance of colitis, with studies noting an increase in vascularization around sites of inflammation (Sandor et al.,2006; Danese,2007). Thus, reduced intestinal vascularization was unexpected in the context of findings in mammalian systems and our observation of increased cellular proliferation in the gut. Because an intestine-specific decrease in vasculature has also been noted in cdx1b morphants (unpublished data), reduced intestinal vascularization may reflect an inflammation-driven reduction in functional capacity within the developing zebrafish intestine (Flores et al.,2008).

Because zebrafish larvae have not yet developed an adaptive immune system to mediate a T-cell response to an introduced hapten, TNBS may act by directly disrupting intestinal epithelial cell function or indirectly disrupting mucin protection of the intestinal epithelium resulting in aberrant microbial contact with host cells (Van der Sluis et al.,2006; Johansson et al.,2008). The concentration of TNBS haptenizing activity to the larval intestine may reflect a preference for intestinal mucins. It may also be the result of the accumulation of TNBS in the larval intestine.

Neutral red staining has been used as a visual marker of epithelial damage in zebrafish larvae (McLeish et al.,2010). Our assays dispute earlier observations that immersion in TNBS does not cause a chemical burn on surface epithelial layers when used at a dose of 75 μg/ml (Fleming et al.,2010). As we did not observe skin damage for the primary conditions used in our study and extra-intestinal manifestations of IBD are relatively common (for a detail review, see Larsen et al.,2010), it is possible that our observations of leukocyte recruitment to epithelial layers other than the intestine and skin expression of mmp9 are systemic manifestations of induced inflammation.

As the pace of genome-wide association studies of IBD candidate genes has increased dramatically over the last few years, the capacity to investigate in vivo function using traditional mammalian model systems has significantly lagged behind (Budarf et al.,2009). The amenability of zebrafish larvae to the use of antisense morpholino technology allows the reproduction of a gene depletion background for analysis in a fraction of the time required to generate and breed a colony of knockout mice. We are currently identifying functional orthologs of IBD susceptibility genes in the zebrafish and reproducing biologically important facets of their known and unexpected functions in vivo. A zebrafish chemical enterocolitis platform will thus add a simple intestinal damage model to the growing toolbox of assays to rapidly assess the in vivo function of novel disease-associated genes in the zebrafish.


Animal Husbandry

Zebrafish (Danio rerio) embryos were obtained from natural spawning and raised at 28.5°C in embryo medium (E3; Westerfield,2000).

Morpholino Microinjection

Morpholinos (MOs; GeneTools, LLC, Philomath, OR) were designed to target the splice donor site at the 3′ end of exon 2 of the myd88 gene (5′-GGTTAAACACTGACCCTGTGGATC-3′). MOs were injected into 1 to 4 cell embryos as described (Nasevicius and Ekker,2000).

Induction of Enterocolitis by TNBS Exposure

Zebrafish (3 dpf) were placed into groups of 20 larvae in 20 ml of E3. TNBS (Sigma-Aldrich) was added to achieve the required final concentrations of 50–100 μg/ml. Larvae were anesthetized in M-222 (Tricaine Sigma-Aldrich, St. Louis, MO), and live imaged with stereomicroscopy. Antibiotics consisted of ampicillin (AppliChem, Darmstadt, Germany, 100 μg/ml final concentration), and kanamycin (Sigma-Aldrich, 5 μg/ml final concentration), or a cell culture penicillin streptomycin pre-mix (Invitrogen, Carlsbad, CA), diluted 1:100 for a final concentration of 100 units/ml penicillin and 100 μg/ml streptomycin. Co-incubations were performed with addition of antibiotics immediately before TNBS exposure. The 5-ASA rescue solution was prepared by dissolving 5-ASA (Sigma-Aldrich, 50 μg/ml final concentration) in E3 adjusted to pH 7. The prednisolone rescue solution was prepared by dissolving 6α-methylprednisolone (Sigma-Aldrich, 25 μg/ml final concentration) in dimethyl sulfoxide (DMSO).


Histological analysis was performed as described (Ng et al.,2005).

Live Imaging

Neutral red staining was performed by incubating larvae in 2.5 μg/ml neutral red (Sigma-Aldrich) in E3 for at least 5 hr as described (Herbomel et al.,2001). Nile Red (Sigma-Aldrich) staining was performed as described (Jones et al.,2008). NBD-cholesterol staining was performed as described (Farber et al.,2001). Larvae were incubated in 2 μg/ml acridine orange (Sigma-Aldrich) in E3 for 30 min and washed three times in E3 before imaging. Live larvae were anesthetized in tricaine and mounted in 3% methylcellulose for imaging on either a Nikon SMZ1500 stereomicroscope equipped with a DS-U2/L2 or a Nikon D-Eclipse C1 confocal microscope. Further image manipulation was carried out with either ImageJ software Version 1.43 (National Institutes of Health) or Volocity 5.0 image analysis software (Improvision/PerkinElmer Life and Analytical Sciences).

Immunofluorescence Detection

Anti-dinitrophenyl (DNP) antibodies were used to detect TNBS activity (Invitrogen; Comoglio et al.,1975). Whole mount immunofluorescence detection was performed as described (Trotter et al.,2009). Whole animal specimens were imaged and photographed with a Leica (Heerbrugg, Switzerland) MZ16 FA stereomicroscope and a Leica DFC490 camera. Histological sections were imaged and photographed with a Leica DMR compound microscope and a DFC420C camera.

Detection of Cell Proliferation and Apoptosis

Cell proliferation was assessed by detection of 5-bromo-2′-deoxyuridine (BrdU) using the In Situ Cell Proliferation Kit, FLUOS (Roche, Mannheim, Germany). TUNEL assays were performed to assess apoptosis using the In Situ Cell Death Kit, TMR red (Roche). Samples were imaged as above. Gut proliferation was scored by manual counting along the length of the gut from the intestinal bulb to the anus and trunk proliferation was scored by manual counting of a single field of view immediately posterior to the swim bladder.

RNA Extraction and qPCR

Total RNA was isolated from pools of 10 to 30 zebrafish larvae using Trizol (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized from total RNA by reverse transcriptase (Applied Biosystems, Foster City, CA). The qPCR was carried out with Platinum SYBR Green qPCR SuperMix-UDG with ROX (Invitrogen) in an ABI PRISM 7900HT Fast sequence detection system (Applied Biosystems). Primers were designed by Primer Express software (for sequences, see Supp. Table S1, which is available online). PCR cycling conditions were 50°C for 2 min, 95°C for 2 min and 40 cycles of 95°C for 15 s, and 60°C for 30 s. Each reaction was carried out in quadruplicate, and a melting curve analysis was performed to confirm the specificity of the reactions. Analysis used the comparative CT method with ef1α as the endogenous control. The average ΔCT value was calculated by subtracting the average ef1α CT from the average target gene CT. The -ΔΔCT was calculated by subtracting the control ΔCT from the treated ΔCT. The relative quantity (RQ) of mRNA was calculated as 2-(ΔΔCT).


Larval dissociation was performed as described (Covassin et al.,2006). Briefly, larvae were killed in a bath of ice-cold E3 and rinsed in calcium-free Ringer solution before trypsin (Invitrogen, final concentration of 0.25% trypsin-EDTA [ethylenediaminetetraacetic protein]) digestion. Cells were washed several times and resuspended in 0.9× PBS/5% fetal calf serum. FACS analyses for the TG(lyzC:EGFP)117 and TG(mpx:EGFP)114 lines were based on forward and side scatter characteristics and GFP expression. All FACS analysis experiments were carried out on a BD LSRII.


Results are presented as mean ± 95% confidence interval. All statistical analyses were performed with GraphPad Prism version 5.0a for Mac (GraphPad Software, San Diego, CA).


We thank Alhad Mahagaonkar for his expert management of our zebrafish facility. We thank Vicki Scott and Stephen Edgar for FACS analysis support. We also thank Dr. J. Wu for the TG(ifabp:RFP) transgenic zebrafish line, Dr. B. Weinstein for the TG(fli1a:EGFP)γ1 and Dr. S. Renshaw for the TG(mpx:EGFP)114 transgenic zebrafish line. Funding for this research was provided by the Foundation for Research, Science and Technology, New Zealand (to P.S.C.) and a scholarship from the Tertiary Education Commission, New Zealand (to S.H.O.).