The TNF^ΔARE Mouse as a Model of Intestinal Fibrosis

Abstract

Crohn disease (CD) is a highly morbid chronic inflammatory disease. Although many patients with CD also develop fibrostenosing complications, there are no medical therapies for intestinal fibrosis. This is due, in part, to a lack of high-fidelity biomimetic models to enhance understanding and drug development, which highlights the need for developing in vivo models of inflammatory bowel disease–related intestinal fibrosis. This study investigates whether the TNF^ΔARE mouse, a model of ileal inflammation, also develops intestinal fibrosis. Several clinically relevant outcomes were studied, including features of structural fibrosis, histologic fibrosis, and gene expression. These include the use of a new luminal casting technique, traditional histologic outcomes, use of second harmonic imaging, and quantitative PCR. These features were studied in aged TNF^ΔARE mice as well as in cohorts of numerous ages. At >24 weeks of age, TNF^ΔARE mice developed structural, histologic, and transcriptional changes of ileal fibrosis. Protein and RNA expression profiles showed changes as early as 6 weeks, coinciding with histologic changes as early as 14 to 15 weeks. Overt structural fibrosis was delayed until at least 16 weeks and was most developed after 24 weeks. This study found that the TNF^ΔARE mouse is a viable and highly tractable model of ileal fibrosis. This model and the techniques used herein can be leveraged for both mechanistic studies and therapeutic development for the treatment of intestinal fibrosis.

Graphical abstract

Inflammatory bowel disease (IBD) is a family of disorders characterized by chronic, relapsing inflammation of the gastrointestinal tract.¹ The prevalence of IBD varies worldwide and corresponds roughly with the degree of industrialization. As a result, developed countries report the highest rates of IBD with accelerating prevalence rates in the developing world due to burgeoning industrialization.² IBD is currently subdivided into two major classes, Crohn disease (CD) and ulcerative colitis. There are >780,000 patients with CD in the United States, and most will eventually develop fibrostenotic or structural complications of the intestine.³ Billions of dollars are spent on powerful biologics for CD, but these powerful anti-inflammatory therapies have not significantly altered the incidence of fibrostenosing CD.⁴ Currently, there are no accepted medical antifibrotic therapies for CD, making the development of such a treatment a research priority.

A significant barrier to the development of antifibrotic therapies is an incomplete understanding of the pathophysiology of IBD-related intestinal fibrosis. This aspect is further limited by a lack of in vivo models of intestinal fibrosis that are reliable and appropriately mimic the human condition. Murine models have been a mainstay of in vivo modeling of intestinal inflammation. It is notable that CD-associated fibrosis is most frequently located in the terminal ileum, but most murine models of IBD investigate colonic inflammation.⁵ This represents a significant limitation to the study of intestinal fibrosis in CD. Currently, there is only one described spontaneous model of ileal inflammation and fibrosis in strain SAMP1/YitFc. However, this strain has significant breeding limitations, which hamper its use.⁶ Thus, an opportunity exists to enhance understanding of fibrostenosing CD through development of new in vivo model systems that closely mimic the complex pathophysiology in human CD and are also highly reliable, reproducible, and amenable to both mechanistic and translational applications.

The TNF^ΔARE mouse is a well-described murine model of spontaneous inflammation that develops Crohn-like ileitis and a number of extraintestinal manifestations that mimic CD.^7,8 Due to a deletion of AU-rich elements (ARE) in the 3′ untranslated region of the tumor necrosis factor (Tnf) α mRNA transcript, Tnfα mRNA has greater stability resulting in systemic overproduction. This overproduction of Tnfα leads to highly penetrant ileal inflammation that is localized to the terminal ileum, as well as inflammatory arthritis.7, 8, 9, 10 Typically, in heterozygous mice, the changes of ileitis begin to occur at 6 to 8 weeks of age.^7,9,10 Homozygous mutations lead to early demise, either in utero or shortly after birth.

Given the highly biomimetic characteristics of the TNF^ΔARE+/– mouse to CD, the current study was designed to determine whether these animals represent an appropriate model of ileal fibrosis. Histologic and biochemical analyses of TNF^ΔARE+/– tissues identified age-related increases in biomarkers of overt fibrosis, including fibrosis-associated gene expression, increased collagen deposition and disorganization, and overt ileal strictures in older mice. These studies identified the TNF^ΔARE+/– mouse model as a relevant and tractable model of intestinal fibrosis with potential use in uncovering novel mechanisms of fibrostenotic disease in human CD.

Materials and Methods

Animal Handling

The B6.129S-Tnf^tm2GKI/Jarn (previously described^7,11) and wild-type littermates that were all bred in house were used (this mouse is referred to as TNF^ΔARE+/– for the remainder of the article). All animals were maintained and handled according to protocols approved by the Institutional Animal Care and Use Committee protocol #00182.

Ileum Collection and Processing

After euthanization per Institutional Animal Care and Use Committee protocols using carbon dioxide (with secondary cervical dislocation), the abdominal cavity was carefully opened, and the cecum, colon, and small bowel were identified. The small intestine, cecum, and proximal colon were then removed, and a section spanning mid-small bowel to mid colon was then positioned next to a ruler and, in most cases, a photograph was obtained. Removal of some peri-intestinal fat was performed for clarity of visualization and to be able to straighten the specimen. Typically, in TNF^{ΔARE +/−} animals, the distal 3.0 to 4.0 cm of the ileum was grossly affected/diseased, and thus roughly 1.0 to 5.0 cm of ileum was used for analysis in TNF^{ΔARE +/−} and TNF^{ΔARE −/−} control animals. Tissue was then used for RNA purification and ileal casting.

Luminal Casting

Once the ileum, cecum, and proximal colon were removed, they were processed, examined for gross stricturing, and photographed. The ileum was then cut at the ileocecal junction and approximately 5 to 6 cm proximally with either a scalpel or sharp dissecting scissors. Any stool was gently flushed out using phosphate-buffered saline (PBS) that was injected into the proximal end of the ileum tissue using a 3 or 5 mL syringe and needle that was between 22 and 26.5 gauge. Stool can also be gently expelled using blunt tweezers, with care taken not to damage the tissue.

An agarose-based compound was prepared by heating until it was liquid and flows freely. These experiments primarily used Epredia HistoGel (REF HG-4000 to 012; Thermo Fisher Scientific, Waltham, MA), with some experiments using a 1% to 2% agarose composed of tap water mixed with Fisher BP160-500 molecular biology grade, low electroendosmosis/multipurpose agarose. In a separate area, cold PBS was placed into a holding tray (in this case, weigh boats) and placed on ice. Each tray was labeled with the sample for the animal. The cold PBS serves to hold the casted ileum and preserve the tissue while the agarose hardens.

Where needed, a portion of terminal ileum was removed from the end for RNA or protein analysis using sharp dissecting scissors or scalpel. The distal end of the remaining ileum was securely sealed using either a suture to tie it off or by clamping with a hemostat. If the hemostat was used, it was carefully positioned and supported by using blocks or by wedging between other tools so as to be upright, keeping the ileum straight and without putting strain on the tissue. Two to three milliliters of agarose gel were then drawn into a 5 mL syringe with a 16- to 26.5-gauge needle. Select the largest bore needle that will fit into the lumen of the ileum; needles <26.5 gauge have difficulty passing the gel due to viscosity. The proximal ileum was then gently grasped with either hemostats or flat tweezers, and the needle with the agarose-filled syringe attached was carefully inserted into the open lumen, taking extreme care not to puncture the wall of the ileum. The grasping hemostat was then gently tightened around the needle manually to create a seal, and the agarose gel was gently injected into the lumen until the ileum was completely filled, revealing any defects/strictures that were present.

The needle was gently removed while the grasping hemostat was then tightened to seal the distal end, and the entire piece was carefully placed into the cold PBS. It was manually held in the cold PBS for about 10 seconds, until the agarose was hard enough that the hemostat(s) could be removed and agarose did not leak out the end. The ileum was now casted, and it was left in the cold PBS for approximately 30 to 60 minutes to allow the gel to fully cure. Once cured and hardened, the ileum was removed, analyzed, and photographed. Using very fine dissecting scissors, the ileum was gently cut lengthwise to flay open the ileum and reveal the cast within. Great care was taken to cut open the ileum and nudge the cast out of the flayed ileum without damaging the cast. The cast was then analyzed and photographed, and the ileal tissue was fixed in 10% neutral buffered formalin for further histologic analysis. If needed, a small amount of methylene blue was dripped onto the cast to enhance visualization.

Measurements of Casted Ileum for Determination of Presence of Stricture and Ileal Narrowing

All measurements were performed by using (FIJI Is Just) ImageJ version 2.9.0/1.53 t/Java 1.8.0_172 (FIJI; NIH, Bethesda, MD; https://imagej.nih.gov/ij). Measurement scales were set internally by using a ruler that was imaged with each piece.

Determination of Luminal Area

Luminal casting was performed as described earlier in the Materials and Methods. The casts were then removed by cutting the ileum longitudinally and removing the cast within, attempted to keep the cast as intact as possible. The area of each cast, and thus intraluminal area, was measured. Area was then normalized to length of the cast. All measurements were performed by using FIJI. Measurement scales were set internally for each image by using a ruler that was imaged with each piece.

Histology

After casting and cast removal, the ileum was flayed open and fixed in 10% neutral buffered formalin at room temperature or 4°C for a minimum of 24 hours. Tissue was then processed via formalin fixation paraffin embedding and stained by using hematoxylin and eosin (H&E), picrosirius red (PSR), and trichrome. Slides were imaged by using a BX41 microscope (Olympus, Tokyo, Japan) with a SPOT Insight FireWire 2 megasample camera and processed by using SPOT 5.3 software (SPOT Imaging, Sterling Heights, MI). For polarized images of PSR stains, an Olympus polarizing filter was used. Images were acquired and analyzed in a de-identified manner.

Wall Width Measurements

For evaluation of wall layer thickness, H&E and/or PSR were used. Samples were analyzed in a de-identified manner. The entire slide was examined, and the thickest representative portion was centered into one field of view. Three measurements roughly equidistant were then obtained from this field of view using FIJI and averaged to generate a wall thickness value for the ileum of each animal, measured from the serosa to the crypt base.

Fibrosis Scoring

Samples were analyzed in a de-identified manner. Each sample was graded by using H&E and PSR in two domains on a scale from 0 to 3. These scores included the following factors.

Disorganization/Disarray of Collagen Fibers

• •0 – none, physiologically organized
• •1 – minimal disorganization, few small breaks
• •2 – moderately disorganized but structure still easy to see
• •3 – severely disorganized, no evidence of normal physiological pattern, collagen infiltrates muscular layers

Fibromuscular Hyperplasia/Muscularization/Muscular Obliteration of the Submucosa

• •0 – none, normal muscle submucosa content
• •1 – mild increase in muscle content
• •2 – large amount of muscle in submucosa, affects the morphology but roughly normal structure
• •3 – muscle has obliterated the entire submucosal space

Second Harmonic Generation Collagen Calculation Methods

Slides were de-paraffinized and re-hydrated by using xylene and progressive concentrations of ethyl alcohol. Slides were imaged by using label-free second harmonic generation with a laser-scanning confocal microscope (LSM 780 spectral; Carl Zeiss, Thornwood, NY). Once images were obtained, further analysis was performed by using FIJI. Threshold was set to 50 to best eliminate artifacts. The red/collagen area of the submucosa and wall, less the epithelium, was then measured and normalized to ileum length (in micrometers). Samples were analyzed in a de-identified manner.

Enzyme-Linked Immunoassay Methods

Whole ileal tissue was collected from the distal 0.5 cm of the terminal ileum and placed in cold PBS. Tissue was homogenized by using a Bio-Gen Series PRO200 homogenizer (PRO Scientific, Oxford, CT). An enzyme-linked immunosorbent assay was then performed by using the Mouse Collagen Type 1 ELISA Kit (catalog #NBP2-75822; Novus Biologicals, Littleton, CO) as per the manufacturer’s instructions. Total protein concentration per sample was measured by using the Pierce BCA Protein Assay Kit (catalog #23227; Thermo Fisher Scientific) as per the manufacturer’s instructions, and samples were normalized to protein concentration.

Transcriptional Analysis

Whole ileal tissue was collected and snap-frozen in RNAlater solution (Invitrogen Ref AM7021; Thermo Fisher Scientific) and later thawed and homogenized by using a Bio-Gen Series PRO200 homogenizer. EZ-10 DNAaway Miniprep RNA isolation kits (catalog #BS88136; Bio Basic, Markham, ON, Canada) were used to isolate RNA from the homogenate according to the manufacturer’s instructions. cDNA was then generated via reverse transcription using the iScript cDNA Synthesis kit (Bio-Rad, Hercules, CA) according to the manufacturer’s instructions. SYBR Green (Applied Biosystems, Foster City, CA) was then used for PCR analysis using the primer sequences presented in Table 1.

Table 1.

Primer Sequences

Primer	Sequence
Actb (mouse)	Forward: 5′-AACCCTAAGGCCAACCGTGAA-3′
	Reverse: 5′-TCACGCACGATTTCCCTCTCA-3′
Tnfα (mouse)	Forward: 5′-CCCTCACACTCAGATCATCTTCT-3′
	Reverse: 5′-GCTACGACGTGGGCTACAG-3′
Il1b (mouse)	Forward: 5′-GGATGATGATGATAACCTGC-3′
	Reverse: 5′-CATGGAGAATATCACTTGTTGG-3′
Tgfb1 (mouse)	Forward: 5′-GGATACCAACTATTGCTTCAG-3′
	Reverse: 5′-TGTCCAGGCTCCAAATATAG-3′
Axl (mouse)	Forward: 5′-AGAGCTGAAGGAGAAACTAC-3′
	Reverse: 5′-TGGCAATTTTCATGGTCTTC-3′
Col1a1 (mouse)	Forward: 5′-GATCTGTATCTGCCACAATG-3′
	Reverse: 5′-TGGTGATACGTATTCTTCCG-3′
Col1a2 (mouse)	Forward: 5′-CAGCGATTACTACTGGATTG-3′
	Reverse: 5′-GATAGTCTCTCCTAACCAGAC-3′

QuantStudio machine and software version 1.5.2 (Applied Biosystems, Waltham MA) were used to perform quantitative real-time PCR and analyzed as fold change to control using the ΔΔC_T method and β-actin as the housekeeping gene.¹²

Statistical Analysis

Statistical analysis was performed by using GraphPad Prism version 9.5.1 (733) (GraphPad Software, La Jolla, CA). Data were assessed for normal distribution by using the Shapiro-Wilk test. For data with normal distribution, the appropriate parametric test was used, and an appropriate nonparametric test was used when data were not normally distributed. Depending on the data, an unpaired t-test (with Welch's correction where variance was unequal), U-test, ordinary one-way analysis of variance (Welch’s analysis of variance in which variance was unequal using a Brown-Forsythe test), Kruskal-Wallis test, receiver-operating characteristic (ROC) curve analysis, or binomial testing was used. Outlier analysis using Robust regression and Outlier Removal (ROUT) method with Q = 1% was performed by using GraphPad Prism, and any identified outliers were removed. For data analyzed with an ordinary one-way analysis of variance, Tukey’s multiple comparisons test was used to compare groups. When the Kruskal-Wallis test was used, Dunn’s multiple comparisons test was used to perform multiple comparisons between groups. Results were deemed statistically significant when P ≤ 0.05.

Results

Structural Fibrotic Changes of Intestinal Fibrosis

Intestinal stricturing and luminal narrowing are some of the most significant clinical sequelae of CD.^13,14 These were analyzed to determine whether clinically analogous findings developed in the terminal ileum of aged TNF^{ΔARE +/−} mice.

Presence of Strictures in the Terminal Ileum of Aged (>24 Weeks) TNF^{ΔARE +/−} Mice

Determining the presence of a stricture in human IBD patients is a clinical challenge.¹⁵ Examination of surgical specimens remains the most concrete method of determining the presence of strictures, but the use of cross-sectional imaging has emerged as another useful tool.¹⁶ There is heterogeneity in the interpretation of what constitutes a stricture on cross-sectional imaging based on luminal diameter, full bowel diameter, up-stream dilation, and wall thickness. To determine the presence of strictures, the terminal ileum of wild-type and TNF^{ΔARE +/−} mice ex vivo was evaluated by using the above two methods. This analysis was inclusive of six wild-type and seven TNF^{ΔARE +/−} mice at >24 weeks of age.

The terminal ileum was manually inspected both before and after injection with an agarose-based gel as described in Luminal Casting. This method was needed to distend the intestines fully for accurate assessment, as distension is recognized as an important feature for accurate determination of stricture in the small intestine.¹⁷ The terminal ileum of TNF^{ΔARE +/−} mice aged >24 weeks contained gross stricture(s) with upstream dilation and ragged and uneven contours. Once the luminal cast was removed, these features were again revealed, indicating the lumen contained strictures with upstream dilation and overall decreased luminal volume. Wild-type ileum was smooth and uniform, with no stricture or upstream dilation (Figure 1 A and B). The number of strictures observed on manual inspection of specimens was counted, analogous to surgical determination of presence of stricture. No strictures were detected in wild-type specimens compared with an average of 1.14 ± 0.90 strictures detected in aged TNF^{ΔARE +/−} specimens (P = 0.015) (Figure 1D).

Figure 1.

Aged (>24 weeks) TNF^{ΔARE +/–} mice develop clinically analogous structural fibrosis, including stricture and luminal narrowing, while age-matched littermate control wild-type mice do not. A: Terminal ileum that has been filled with agarose casts. B: Casts from (A) that have been removed from the ileum. C: Quantification of gross ileal strictures in aged TNF^{ΔARE +/–} mice and wild-type controls. D: Manual inspection of gross strictures in aged TNF^{ΔARE +/–} mice and wild-type controls. E: Ratio of the width of the widest section to the narrowest section of the terminal ileum of wild-type controls and aged TNF^{ΔARE +/–} mice. F: Luminal cast area/length was measured. Dotted line outlines the ileum in (A) and the ileal casts in (B) to assist in visualization of the size and shape. Data were analyzed by using binomial test (C), Welch’s t-test (D and E), or unpaired t-test (F) based on normality and variance of the sample and are expressed as means ± SD. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗∗P < 0.0001.

To more objectively quantify the number of strictures present, measurements were performed in terminal ileum that had undergone luminal casting as described in Luminal Casting. This method was needed to distend the intestines fully for accurate measurement of diameter, as distension is recognized as an important feature for accurate determination of stricture in the small intestine.¹⁷ To our knowledge, no such consensus on stricture determination exists for murine models. For this study, expert consensus from human imaging was extrapolated to create a definition of ileal stricture. An expert consensus on definition of naive small bowel stricture in humans suggests that stricture be defined as luminal narrowing with bowel wall thickening and upstream dilation, with luminal diameter reduced by at least 50% relative to an adjacent bowel loop.¹⁷ FIJI was used to obtain measurements to determine whether the terminal ileum contained any stricture meeting the rigorous and clinically analogous parameter of a narrowing with diameter ≤50% of an adjacent loop. Aged TNF^{ΔARE +/−} mice developed stricture in 57% of animals (P < 0.0001) compared with 0% wild-type mice (Figure 1C). To further characterize the ileal narrowing, the ratio of widest to narrowest measurement was compared. Aged TNF^{ΔARE +/−} mice had significant ileal narrowing compared with wild-type controls. The ratio of the width of the widest section to the narrowest section of the terminal ileum of wild-type mice compared with aged TNF^{ΔARE +/−} mice was 1.3 ± 0.12 versus 1.9 ± 0.37, respectively (P = 0.005) (Figure 1E).

Presence of Luminal Narrowing in the Terminal Ileum of Aged (24+ week) TNF^{ΔARE +/−} Mice

Luminal area, a surrogate for luminal volume, was evaluated to determine whether it differed in aged TNF^{ΔARE +/−} mice. Once the luminal casts were removed, the area per length of each cast was calculated by using FIJI, incorporating five wild-type mice and six TNF^{ΔARE +/−} mice. These data show significantly reduced luminal area and thus luminal volume in aged TNF^{ΔARE +/−} mice compared with wild-type controls. Luminal cast area quantified as area/length revealed a mean area of 22.57 ± 2.46 mm²/cm in wild-type mice compared with 16.72 ± 3.73 mm²/cm in aged TNF^{ΔARE +/−} mice (P = 0.015) (Figure 1F).

At >24 weeks of age, structural changes such as stricture and luminal narrowing were observed in the terminal ileum of TNF^{ΔARE +/−} mice that were clinically analogous to the intestinal fibrosis seen in human patients with CD.

Histologic Features of Intestinal Fibrosis

Histologic evidence of intestinal fibrosis is difficult to characterize. Based on histologic features identified as important indicators of fibrosis in human CD,^18,19 increased wall thickness (measured from serosa to mucosal layer/crypt base), increased collagen content/collagen expansion, disorganization/disarray of collagen fibers, fibromuscular hyperplasia, and muscular obliteration of the submucosa are key features of small intestinal fibrosis. A fibrosis scoring system was created to capture features of disorganization/disarray of collagen fibers, fibromuscular hyperplasia, and muscular obliteration of the submucosa, and objective measurements of wall thickness and amount of collagen content in the terminal ileum of TNF^{ΔARE +/−} mice were performed.

Wall Width

Wall width of the terminal ileum of wild-type and TNF^{ΔARE +/−} mice was measured after formalin fixation paraffin embedding and staining with H&E and/or trichrome. This analysis was inclusive of 13 wild-type and 17 TNF^{ΔARE +/−} mice at >24 weeks of age. The mean wall width of wild-type and aged TNF^{ΔARE +/−} mice was 106.6 ± 26.61 μm versus 467.8 ± 171.30 μm, respectively (P < 0.0001) (Figure 2, A and B). These data support the contention that the wall of the terminal ileum was significantly increased in aged TNF^{ΔARE +/−} mice.

Figure 2.

Aged (>24 weeks) TNF^{ΔARE +/–} mice develop histologic fibrosis, whereas age-matched littermate control wild-type mice do not. A: Representative histology of the terminal ileum of wild-type and aged TNF^{ΔARE +/–} mice using hematoxylin and eosin staining. B: Wall width measurements of wild-type and aged TNF^{ΔARE +/–} mice. C: Representative histology of the terminal ileum of wild-type and aged TNF^{ΔARE +/–} mice using picrosirius red staining. Top panel was under white light, lower panel was under polarized light. D: Fibrosis score of wild-type and aged TNF^{ΔARE +/–} mice based on histologic scoring. E: Enzyme-linked immunosorbent assay quantification of collagen type I. Data were analyzed by using Welch’s t-test (B), U-test (D), or unpaired t-test (E) based on normality and variance of the sample and are expressed as means ± SD. n = 13 wild type and n =16 TNF^{ΔARE +/–} (A). ∗∗∗∗P < 0.0001. Scale bars: 100 μm (A and C).

Fibrosis Score

To evaluate whether the TNF^{ΔARE +/−} mice develop histologic changes of fibrosis consistent with human disease, fibrosis scoring criteria using features published in a large systematic review and international consensus manuscripts were created.^18,19 For fibrosis scoring, each slide was graded by using H&E and/or trichrome stain as well as PSR in two domains on a scale as described in Materials and Methods. This analysis was inclusive of 13 wild-type and 16 TNF^{ΔARE +/−} mice at >24 weeks of age.

At >24 weeks of age, TNF^{ΔARE +/−} mice developed all of the aforementioned features of intestinal fibrosis in the terminal ileum. Representative images of H&E (Figure 2A) and PSR under both white light and polarized light (Figure 2C) revealed increased wall thickness, increased collagen content/collagen expansion, disorganization/disarray of collagen fibers, fibromuscular hyperplasia, and muscular obliteration of the submucosa. The average fibrosis score of wild-type to aged TNF^{ΔARE +/−} mice was 0.46 ± 0.52 to 5.13 ± 1.09, respectively (P < 0.0001) (Figure 2D). These data support substantial fibrotic changes occurring histologically in aged TNF^{ΔARE +/−} mice after 24 weeks of age.

Collagen Content in the Terminal Ileum of Aged TNF^{ΔARE +/−} Mice

The aforementioned data show disorganization of collagen. Changes in collagen content were objectively quantified by using enzyme-linked immunosorbent assay and second harmonic generation. The use of second harmonic generation is more sensitive than collagen immunohistochemical staining and hydroxyproline measures, and it sensitively measures actual fibrillar collagen deposition.20, 21, 22, 23, 24

Using enzyme-linked immunosorbent assay, the amount of collagen type I was quantified in a subset of wild-type mice versus aged TNF^ΔARE+/− mice. These data included seven wild-type and five aged TNF^{ΔARE +/−} mice. The average concentration of collagen type I in wild-type and aged TNF^{ΔARE +/−} mice was 0.19 ± 1.06 ng/mL and 8.47 ± 2.58 ng/mL, respectively (P < 0.0001) (Figure 2E). Using second harmonic generation imaging, collagen amount was quantified in a subset of the wild-type mice versus aged TNF^ΔARE+/− mice. These data included 20 mice, 10 per group, and indicated an average collagen area (square micrometers) per length (micrometers) of 24.28 ± 7.25 in the wild-type mice and 67.05 ± 40.38 in the aged TNF^{ΔARE +/−} mice (P = 0.0085) (Figure 3).

Figure 3.

The increased collagen content deposition of aged TNF^{ΔARE +/–} mice in the terminal ileum wall is presented using second harmonic generation on a subset of our >24-week–old age-matched littermate controls and aged TNF^{ΔARE +/–} mice. Representative images are shown. These data are inclusive of 10 age-matched littermate control wild-type mice and 10 aged TNF^{ΔARE +/–} mice. A: Representative images of the terminal ileum of wild-type and aged TNF^{ΔARE +/–} mice. B: Collagen content in the ileal wall quantified using second harmonic generation and reported as pixels of collagen (count) per length. Data were analyzed by using Welch’s t-test (B) based on normality and variance of the sample and are expressed as means ± SD. ∗∗P < 0.01. Scale bar: 100 μm (A).

At >24 weeks of age, TNF^ΔARE +/− mice had developed the key features of histologic fibrosis in the terminal ileum that mimic classical features of intestinal fibrosis in human patients with CD.

Gene Expression

Pro-inflammatory and Profibrotic Gene Expression in the Terminal Ileum of Aged and Young TNF^{ΔARE +/−} Mice

After determining that structural and histologic fibrosis develops in >24-week–old TNF^{ΔARE +/−} mice, select pro-inflammatory and profibrotic gene expression was analyzed. RNA from whole-tissue samples of terminal ileum was purified and analyzed with quantitative PCR (wild type, n = 8; TNF^{ΔARE +/−}, n = 10) (Figure 4A). Aged TNF^{ΔARE +/−} mice exhibited a 29.37 ± 16.68-fold increase compared with wild-type mice in Tnfα mRNA expression (P < 0.0001), and a 37.89 ± 26.74-fold increase compared with wild-type mice in Il1β expression (P = 0.002). This analysis confirmed that Tnfα was appropriately elevated, and the increase in Il1β further shows that the terminal ileum was in an inflamed state. Aged TNF^{ΔARE +/−} mice also exhibited a 5.12 ± 2.39-fold increase compared with wild-type mice in Tgfβ1 expression (P = 0.0004) and a 2.85 ± 0.79-fold increase compared with wild-type mice in Axl expression (P < 0.0001). Tgfβ1 is well recognized as a pivotal regulator of fibrosis,25, 26, 27, 28 and Axl has been shown to be a regulatory pathway in fibrosis in many organs, including the small intestine.²⁹ Thus, two pivotal pro-inflammatory genes and two profibrotic genes showed increased expression in TNF^{ΔARE +/−} mice at an age >24 weeks.

Figure 4.

Aged TNF^{ΔARE +/–} mice express increases in several pro-inflammatory and profibrotic genes in the terminal ileum, and these changes occur as early as 6 to 7 weeks. A: Gene expression in mice >24 weeks old. B: Gene expression in 6- to 7-week–old mice. Data were analyzed by using U-test, Welch’s t-test, or unpaired t-test based on normality and variance of the sample and are expressed as means ± SD. ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

To determine whether these genes represent appropriate biomarkers of ileal fibrosis, receiver-operating characteristic curves were used to look for associations between fold-change of gene expression and fibrosis score as detailed in Fibrosis Score and Fibrosis Scoring in >24-week–old TNF^{ΔARE +/−} mice (Supplemental Figure S1). These analyses show a strong association between fibrosis score and increased mRNA expression of Tnfα (P = 0.0002), Il1β (P = 0.0005), and Axl (P = 0.002). No association was found with Tgfβ1, which may be due to the complexity of Tgfβ1 signaling and heterogeneity of expression in TNF^{ΔARE +/−} mice. Ultimately, this receiver-operating characteristic analysis supports this gene profiling approach as appropriately diagnostic for frank tissue fibrosis in >24-week–old TNF^{ΔARE +/–} mice.

To corroborate the increase in collagen seen histologically, Col1a1 and Col1a2 mRNA was analyzed in a subset of animals (for Col1a1 wild-type, n = 6; TNF^{ΔARE +/−}, n = 7; for Col1a2 wild type, n = 6; TNF^{ΔARE +/−}, n = 8). There was no difference in Col1a1 expression, but Col1a2 exhibited a 2.74 ± 0.97-fold increase compared with wild type (P = 0.003) (Figure 4A).

To determine whether these expression profiles occur at younger ages, mice at 6 to 7 weeks of age were analyzed (wild type, n = 8; TNF^{ΔARE +/−}, n = 8). The increased expression of Tnfα, Il1β, Tgfβ1, and Axl was present as early as 6 to 7 weeks (Figure 4B). Six- to seven-week–old TNF^{ΔARE +/−} mice exhibited a 32.1 ± 8.27-fold increase in Tnfα mRNA expression (P < 0.0001) and a 25.3 ± 11.90-fold increase in Il1β expression (P = 0.0007). Six- to seven-week–old TNF^{ΔARE +/−} mice also exhibited a 3.1 ± 0.75-fold increase in Tgfβ1 expression (P < 0.0001) and a 2.3 ± 0.46-fold increase in Axl expression (P = 0.001). Expression was not different for Col1a1 or Col1a2 (Figure 4B).

Thus, it seems that while the genetic milieu that ultimately results in fibrosis may be present as early as 6 to 7 weeks, increased collagen type I mRNA is not yet elevated at this age.

Profiling the Progression to Ileal Fibrosis in TNF^{ΔARE +/−} Mice

Age of Emergence of Histologic Fibrotic Changes in the Terminal Ileum of TNF^{ΔARE +/−} Mice

Given the young age at which profibrotic gene expression occurs, characterization of the age at which structural and histologic fibrosis emerges in the ileum of TNF^{ΔARE +/−} mice was performed. To determine at what age histologic fibrosis emerges in the terminal ileum, H&E, PSR, wall width measurements, and fibrosis scoring performed on mice >24 weeks old were repeated on three younger age cohorts. These cohorts were aged 7 to 10 weeks, 14 to 15 weeks, and 18 to 24 weeks. Representative images of H&E and PSR staining under white light and polarized light showed that at 7 to 10 weeks, TNF^{ΔARE +/−} terminal ileum was nearly indistinguishable from wild type (representative of wild type, n = 10; TNF^{ΔARE +/−}, n = 8), but pro-inflammatory and profibrotic changes began to emerge at 14 to 15 weeks (representative of wild type, n = 5; TNF^{ΔARE +/−}, n = 5) and continued at 18 to 24 weeks (representative of wild type, n = 7; TNF^{ΔARE +/−}, n = 7) (Figure 5, A and B). These findings are visually less pronounced than those seen in animals aged >24 weeks.

Figure 5.

Age-related histologic changes of intestinal fibrosis. A: Representative histology of the terminal ileum of wild-type and aged TNF^{ΔARE +/–} mice using hematoxylin and eosin staining. B: Representative histology of the terminal ileum of wild-type and TNF^{ΔARE +/–} mice of varying ages using picrosirius red staining. Top panel was under white light, lower panel was under polarized light. C: Wall width measurements of wild-type and TNF^{ΔARE +/–} mice aged 7 to 10 weeks, aged 14 to 15 weeks, and aged 18 to 24 weeks. D: Fibrosis score based on histologic scoring of wild-type and TNF^{ΔARE +/–} mice aged 7 to 10 weeks, aged 14 to 15 weeks, and aged 18 to 24 weeks. Data were analyzed by using U-test (C) or U-test test, Welch’s t-test, unpaired t-test based on normality, and variance of the sample (D) and are expressed as means ± SD. ∗P < 0.05; ∗∗P < 0.01. Scale bars: 100 μm (A and B).

The wall width was not increased in TNF^{ΔARE +/−} mice aged 7 to 10 weeks (wild type, n = 10; TNF^{ΔARE +/−}, n = 7) but was significantly thicker in animals aged 14 to 15 weeks (wild type, n = 5; TNF^{ΔARE +/−}, n = 5) and 18 to 24 weeks (wild type, n = 7; TNF^{ΔARE +/−}, n = 7) (Figure 5C). Wall width measurements of wild-type compared with TNF^{ΔARE +/−} mice at 7 to 10 weeks was 71.51 ± 16.3 μm and 85.04 ± 10.0 μm, respectively (P = 0.07); at 14 to 15 weeks, mean width was 114.7 ± 39.2 μm and 240.0 ± 94.3 μm (P = 0.03); and at 18 to 24 weeks, mean width was 91.74 ± 16.9 μm and 249.0 ± 87.9 μm (P = 0.003). Thus, wall width began to increase by 14 to 15 weeks. This was likely due to a combination of developing fibrosis but also strongly influenced by immune cell infiltration, which is a characteristic of the TNF^{ΔARE +/−} mice by this age.

The fibrosis score based on the histologic scoring described in Materials and Methods of wild-type mice compared with TNF^{ΔARE +/−} mice was unchanged in the youngest cohort (wild type, n = 9; TNF^{ΔARE +/−}, n = 8) but was significantly increased compared with wild-type mice in the 14 to 15 weeks of age cohort (wild type, n = 5; TNF^{ΔARE +/−}, n = 5) and the 18 to 24 weeks of age cohort (wild type, n = 7; TNF^{ΔARE +/−}, n = 7) (Figure 5D). Wild-type to TNF^{ΔARE +/−} mean scores in mice aged 7 to 10 weeks were 0 ± 0 and 0.63 ± 1.06, respectively (P = 0.07); mean scores in those aged 14 to 15 weeks were 0 ± 0.00 and 2 ± 1.41 (P = 0.05); and mean scores in those aged 18 to 24 weeks were 0 ± 0.00 and 2.57 ± 1.40 (P = 0.005).

At age 7 to 10 weeks, TNF^{ΔARE +/−} mice did not exhibit histologic evidence of fibrosis. Beginning at around 14 to 15 weeks, wall thickening and fibrotic changes started to occur, and continued through the 14- to 15-week and 18- to 24-week age range.

Age of Emergence of Structural Fibrotic Changes in the Terminal Ileum of TNF^{ΔARE +/−} Mice

To determine at what age clinically analogous strictures emerge, the ileal casting and measurements performed on >24-week–old mice were repeated on two younger age cohorts. These cohorts were aged 7 to 10 weeks or 16 to 24 weeks. Grossly, the terminal ileum did not exhibit any strictures by observation or measurement in the cohort aged 7 to 10 weeks (wild type, n = 8; TNF^{ΔARE +/−}, n = 6) (Figure 6, A, B, and D). Furthermore, 7- to 10-week–old TNF^{ΔARE +/−} mice did not develop increased widest/narrowest measurement, indicating ileal narrowing (P = 0.36), or smaller luminal area volumes (P = 0.86) (Figure 6, C and E). At the age of 16 to 24 weeks, TNF^{ΔARE +/−} mice exhibited evidence of gross ileal strictures (wild type, n = 7; TNF^{ΔARE +/−}, n = 8) (mean of 0.63 ± 0.52 stricture in TNF^{ΔARE +/−} mice; none in wild type; P = 0.03) (Figure 6, F, G, and I) but no increased ratio of widest to narrowest ileum (ileal narrowing) (wild type, n = 6; TNF^{ΔARE +/−}, n = 8) (P = 0.11) (Figure 6H) or smaller luminal area volumes (wild type, n = 7; TNF^{ΔARE +/−}, n = 8) (Figure 6J) (P = 0.39).

Figure 6.

TNF^ΔARE+/– mice do not develop robust clinically analogous structural fibrosis until after 24 weeks of age. These cohorts are aged 7 to 10 weeks (A–E) or aged 16 to 24 weeks (F–J). (A and F) Terminal ileum that has been filled with agarose casts at ages 7 to 10 weeks (A) and 16 to 24 weeks (F). Once removed, the casts of the same mice are shown at ages 7 to 10 weeks (B) and 16 to 24 weeks (G). C–E: Quantification of gross ileal strictures, ileal narrowing, and smaller luminal area volumes in wild-type and 7- to 10-week–old TNF^{ΔARE +/–} mice. H–J: Quantification of gross ileal strictures, ileal narrowing, and luminal area volumes in wild-type and 16- to 24-week–old TNF^{ΔARE +/–} mice. Dotted line outlines the ileum in (A and F) and the ileal casts in (B and G) assist in visualization of the size and shape. Data were analyzed by using an unpaired t-test (C, E, H, and J) or the U-test (I) and are expressed as means ± SD. ∗P < 0.05.

Thus, there was evidence that ileal stricturing was beginning to develop at 16 to 24 weeks of age, but there was no significant change in ileal narrowing, and no change in luminal cast area between wild-type and TNF^{ΔARE +/−} mice. These results suggest that TNF^{ΔARE +/−} mice do not begin to develop clinically analogous stricturing until around 16 to 24 weeks of age, but more robust structural fibrosis occurs after 24 weeks.

Comparison of Clinically Analogous Structural Changes of Ileal Fibrosis across Age Groups

Having discovered that TNF^{ΔARE +/−} mice at >24 weeks of age exhibit several aspects of structural ileal fibrosis not seen in younger TNF^{ΔARE +/−} mice, a time course for the development of ileal fibrosis in TNF^{ΔARE +/−} mice was established. For the time course, the data presented earlier were re-organized for direct comparison. For each of the following endpoints examined, the wild-type cohorts of each age group compared with each other showed no significant difference (Supplemental Figure S2). Wild-type data were pooled into one group regardless of age and compared with TNF^{ΔARE +/−} mice of each age cohort. Structural parameters of observed strictures, ileal narrowing, and luminal cast area were compared.

TNF^{ΔARE +/−} mice did not develop significant gross strictures until 16 to >24 weeks (Figure 7A). The mean number of strictures in wild-type mice, 7- to 10-week–old TNF^{ΔARE +/−} mice, 16- to 24-week–old TNF^{ΔARE +/−} mice, and >24-week–old TNF^{ΔARE +/−} mice was 0 ± 0.00, 0 ± 0.00, 0.6250 ± 0.52, and 1.143 ± 0.90, respectively (P < 0.0001). There was a significant increase in the number of strictures in >24-week–old TNF^{ΔARE +/−} mice compared with wild-type mice (P = 0.0003) and compared with the youngest cohort of aged TNF^{ΔARE +/−} mice (P = 0.009). There was also a significant increase in the number of strictures in 16- to 24-week–old TNF^{ΔARE +/−} mice compared with wild-type mice (P = 0.0008). No significant differences were seen when comparing wild-type versus 7- to 10-week–old TNF^{ΔARE +/−} mice, when comparing 7- to 10-week–old versus 16 to 24-week old TNF^{ΔARE +/−} mice, or when comparing 16- to 24-week–old mice versus >24-week–old TNF^{ΔARE +/−} mice.

Figure 7.

TNF^ΔARE+/– mice develop structural and histologic changes of ileal fibrosis over time. These data have been presented in other figures and are re-organized here for direct comparison. All wild-type data across age groups have been pooled. These show the progression of number of ileal strictures (A), ileal narrowing (B), luminal cast area (C), wall width (D), and fibrosis score (E) in wild-type and TNF^{ARE +/–} mice at 7 to 10, 16 to 20, and >24 weeks. Data were analyzed by using a Kruskal-Wallis test (A and E) or ordinary one-way analysis of variance (with Welch’s correction when variance was significant) (B–D) and are expressed as means ± SD. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

TNF^{ΔARE +/−} mice did not develop significant ileal narrowing until 16 to >24 weeks (Figure 7B). The mean ratio of widest to narrowest section of terminal ileum in wild-type mice, 7- to 10-week–old TNF^{ΔARE +/−} mice, 16- to 24-week–old TNF^{ΔARE +/−} mice, and >24-week–old TNF^{ΔARE +/−} mice was 1.40 ± 0.18, 1.513 ± 0.24, 1.730 ± 0.17, and 1.892 ± 0.37, respectively (P = 0.0003). There was a significant increase in the ratio of widest to narrowest section of terminal ileum in >24-week–old TNF^{ΔARE +/−} mice compared with wild-type mice (P = 0.0005) and in the wild-type mice compared with the 16- to 24-week–old TNF^{ΔARE +/−} mice (P = 0.02). No other significant differences were seen.

TNF^{ΔARE +/−} mice did not develop significant reductions in luminal cast area until >24 weeks (Figure 7C). The mean luminal cast area of terminal ileum in wild-type, 7- to 10-week–old TNF^{ΔARE +/−} mice, 16- to 20-week–old TNF^{ΔARE +/−} mice, and >24-week–old TNF^{ΔARE +/−} mice was 25.20 ± 5.17 mm²/cm, 28.46 mm²/cm ± 4.17, 26.28 ± 5.19 mm²/cm, and 16.72 ± 3.73 mm²/cm, respectively (P = 0.0008). There was a significant decrease in cast area (a surrogate for luminal volume) in >24-week–old TNF^{ΔARE +/−} mice compared with wild-type mice (P = 0.003), compared with the youngest cohort of aged TNF^{ΔARE +/−} mice (P = 0.001), and compared with the 16- to 24-week–old group (P = 0.005). No significant differences were seen when comparing wild-type versus 7- to 10-week–old or 16- to 20-week–old TNF^{ΔARE +/−} mice, nor when comparing 7- to 10-week–old versus 16- to 24-week–old TNF^{ΔARE +/−} mice.

These data indicate that structural parameters of observed strictures, ileal narrowing, and luminal cast do not significantly change until TNF^{ΔARE +/−} mice reach at least 24 weeks of age.

Comparison of Histologic Changes of Ileal Fibrosis across Age Groups

Having found that TNF^{ΔARE +/−} mice >24 weeks of age exhibited several histologic aspects of ileal fibrosis, steps were taken to establish a time course for the development of histologic features of ileal fibrosis in TNF^{ΔARE +/−} mice. For each of these endpoints, the data discussed earlier that were re-organized for direct comparison were used. The wild-type cohorts of each age group were compared for each of the following endpoints examined. The wall width and fibrosis scores of wild-type mice had significant differences (P = 0.005 and 0.001), with the >24-week–old group and the 14- to 15-week–old group having greater wall width than the 7- to 10-week–old group, with average widths of 106.6 μm, 114.7 μm, and 71.51 μm, respectively, and the >24-week–old group having a greater score than the 7- to 10-week–old group (mean score 0.46 compared with 0) (Supplemental Figure S2). This is likely due to the increasing size of the animals as they age and is not believed to be clinically relevant. Wild-type data were pooled into one group regardless of age and compared versus TNF^{ΔARE +/−} mice of each age cohort. Histologic parameters of wall width and fibrosis score were compared.

When making comparisons across all age groups, the ileal wall width of TNF^{ΔARE +/−} mice was significantly thicker at 18 to 24 weeks. The mean wall width of the terminal ileum of the wild-type mice, the 7- to 10-week–old TNF^{ΔARE +/−} mice, the 14- to 15-week–old TNF^{ΔARE +/−} mice, the 18- to 24-week–old TNF^{ΔARE +/−} mice, and the >24-week–old TNF^{ΔARE +/−} mice was 94.8 ± 28.7 μm, 85.0 ± 10.0 μm, 240.0 ± 94.3 μm, 249.0 ± 87.9 μm, and 467.8 ± 171.3 μm, respectively (P < 0.0001) (Figure 7D). There was a significant increase in wall width of the terminal ileum in >24-week–old TNF^{ΔARE +/–} mice compared with the wild-type (P < 0.0001), 7- to 10-week–old (P < 0.0001), 14- to 15-week–old (P = 0.0185), and 18- to 24-week–old (P = 0.005) cohorts. The 18- to 24-week age group had a significantly increased score compared with the wild-type group (P = 0.03) and the 7- to 10-week–old group (P = 0.02). No other significant changes were seen. When combined with the data directly comparing wall width to wild type in the 14- to 15-week age group, these data show a progression of terminal ileum wall thickening that was not present in younger animals but started to show in TNF^{ΔARE +/−} mice aged 14 to 24 weeks.

TNF^{ΔARE +/−} mice begin to develop observable histologic evidence of fibrosis emerging at 14 to 15 weeks (Figures 5D and 7E). The mean fibrosis scores of wild-type, 7- to 10-week–old TNF^{ΔARE +/−} mice, 14- to 15-week–old TNF^{ΔARE +/−} mice, 18- to 24-week–old TNF^{ΔARE +/−} mice, and >24-week–old TNF^{ΔARE +/−} mice were 0.00 ± 0.00, 0.6250 ± 1.06, 2 ± 1.41, 2.57 ± 1.40, and 5.125 ± 1.09, respectively (P < 0.0001). There was a significant increase in fibrosis score of the terminal ileum in >24-week–old TNF^{ΔARE +/−} mice compared with wild-type mice (P < 0.0001) and mice aged 7 to 10 weeks (P = 0.0005), but no difference compared with the groups aged 14 to 15 weeks or 18 to 24 weeks. The 18- to 24-week age group had a significantly increased score compared with the wild-type group (P = 0.013), but no difference compared with the group aged 7 to 10 weeks nor the group aged 14 to 15 weeks. No significant differences were seen when comparing wild-type versus the 7- to 10-week–old cohort nor the cohort aged 14 to 15 weeks. In summary, these data show a progression of histologic fibrosis in TNF^{ΔARE +/−} mice aged 14 to 24 weeks which was most pronounced after 24 weeks of age.

Discussion

There is significant interest in the development of antifibrotic therapies for diseases such as CD.^17,30 The field is currently limited by a lack of understanding of the pathophysiology of IBD-related intestinal fibrosis. A significant void is the existence of appropriate in vivo models that mimic human disease. Current models of intestinal fibrosis include spontaneous mutations, targeted gene deletions, chemically induced fibrosis, and infectious inflammatory models.⁶ Although each model provides valuable insight into the fibrotic response, each is also fraught with inconsistency and sporadic appearance of fibrotic lesions. The goal of the current study was to determine whether TNF^{ΔARE +/−} mice may provide insight into the development of structural fibrosis and, as such, serve as a relevant and reproducible model to define principles of fibrostenosis.

The TNF^ΔARE model was originally developed by targeted deletion of 69 base pairs of AU-rich elements in the 3′ untranslated region of the TNF gene.⁷ This deletion results in increased TNF mRNA stability, increased TNF protein synthesis, and elevated systemic TNF levels. The disease is mediated by leukocytes that intensely infiltrate the intestinal lamina propria. T cells isolated from the mesenteric lymph nodes of TNF^ΔARE mice, but not those from C57BL6/J mice, produce high levels of TNF-α during early disease (3 to 4 weeks of age) and a mixed Th1:Th2 cytokine profile during the later stages of chronic ileitis. TNF^ΔARE T cells can adoptively transfer disease to naive, non-predisposed RAG^−/− recipients. The adoptively transferred disease is different from that transferred by CD45Rb^high cells, as TNF^ΔARE+/− T cells predominantly induce ileitis, showing an inherent capacity to recirculate preferentially to the small intestine.⁸ Moreover, TNF^ΔARE mice develop intestinal lymphangiectasia, lymphangiogenesis, and tertiary lymph node organs in the mesentery.³¹ These features bear significant similarity to the lymphatic changes seen in human CD.31, 32, 33, 34 Important in this regard, the TNF^ΔARE mouse recapitulates many features of human CD, including: (1) the spontaneous nature and chronic course of the disease; (2) its pathogenetic dependence on T cells and proinflammatory cytokines; (3) its specific small intestinal (ileal) localization; and (4) abnormal lymphatics with development of mesenteric tertiary lymphoid organs.

Analysis of TNF^{ΔARE +/−} mice revealed development of ileal fibrosis similar to that observed in patients with CD. As was noted in human patients, a time-dependent component accompanied by chronic inflammation preceded the development of fibrosis in TNF^{ΔARE +/−} mice. The progression of pro-inflammatory gene induction occurred at a time point before 6 weeks, with the first signs of histologic fibrosis appearing around 14 to 15 weeks, and profound histologic and structural fibrosis occurring at >24 weeks of age. These analyses have several strengths. The investigations and outcome measures were specifically designed to model the fibrotic nature of human patients with CD. Although histologic features of intestinal fibrosis can be difficult to determine, multiple aspects of intestinal fibrosis were incorporated, including wall thickness, disorganization of collagen, fibromuscular hypertrophy and obliteration, and increased collagen content. TNF^{ΔARE +/−} mice developed clear fibrotic changes in all of aspects of this analysis.

Clinically analogous outcomes of intestinal stricture, including decreased luminal area and luminal narrowing, were specifically evaluated. This approach was enabled by using an agarose-based casting technique that was developed to allow for characterization of these outcomes. This lends relevance to the current model as a biomimetic of human disease. This casting technique was also transferrable to the study disease in any part of the gastrointestinal tract and may also be adaptable to study of other organs. The observation of a progression of changes at distinct ages, starting with gene expression and progressing through histologic changes and lastly structural changes, lends itself to the tailored study of interventions ranging from preventative therapy to agents aimed at reversal of fibrosis.

Although the TNF^{ΔARE +/−} model has many strengths, it also has some limitations. While in the heterozygote state these mice are consistently viable and more robust than for some genetic models, there are logistical limitations because of the need to age the mice to see full intestinal fibrosis. Moreover, these animals did not develop overt intestinal fistulization and/or life-threatening obstruction often observed in patients with CD.^3,35 It may be possible that if these mice were allowed to age further (ie, beyond 24 weeks), obstruction and fistulization could occur. It is notable that these changes were not observed in our cohorts with multiple animals between 30 and 35 weeks of age. TNF^{ΔARE +/−} mice meet humane endpoints for termination before seeing these changes. It will be interesting to determine whether this phenotype could be enhanced or progressed. It is possible, for example, that the combinatorial use of chemical fibrosis inducers (eg, bleomycin)36, 37, 38 in the TNF^ΔARE+/− could provide earlier and more aggressive fibrotic endpoints. More work is necessary to uncover the complex mechanisms of fibrosis in this model. Although the genetic modification is mono-genetic, the subsequent increases in other pro-inflammatory as well as profibrotic cytokines suggest that multiple pathways may be involved, which is corroborated by the known complexity and incomplete understanding of TNF signaling.^39,40

Taken together, this model has several clinically relevant, measurable, and targetable outcomes that can be studied. This model and the outlined techniques and outcome measures are conducive to the study of potential medical therapies as well as mechanistic studies in intestinal fibrosis. Thus, the TNF^{ΔARE +/−} mouse is a tractable and viable model for the study of intestinal fibrosis such as that seen in CD.

Supplemental Data

Supplemental Figure S1

Receiver-operating characteristic curves showing gene expression correlation to fibrosis score for each of the four genes we analyzed.

Supplemental Figure S2

All wild-type data from each age group for observed strictures (A), ileal narrowing (B), luminal cast area (C), ileal wall width (D), and fibrosis score (E). Wild types across all ages. ∗P < 0.05.