Ileitis-associated tertiary lymphoid organs arise at lymphatic valves and impede mesenteric lymph flow in response to tumor necrosis factor

Abstract

Lymphangitis and the formation of tertiary lymphoid organs (TLOs) in the mesentery are features of Crohn's disease. Here, we examined the genesis of these TLOs and their impact on disease progression. Whole-mount and intravital imaging of the ileum and ileum-draining collecting lymphatic vessels (CLVs) draining to mesenteric lymph nodes from TNF^ΔARE mice, a model of ileitis, revealed TLO formation at valves of CLVs. TLOs obstructed cellular and molecular outflow from the gut and were sites of lymph leakage and backflow. Tumor necrosis factor (TNF) neutralization begun at early stages of TLO formation restored lymph transport. However, robustly developed, chronic TLOs resisted regression and restoration of flow after TNF neutralization. TNF stimulation of cultured lymphatic endothelial cells reprogrammed responses to oscillatory shear stress, preventing the induction of valve-associated genes. Disrupted transport of immune cells, driven by loss of valve integrity and TLO formation, may contribute to the pathology of Crohn's disease.

Keywords: Crohn’s disease; cytokine signaling; endothelium; inflammation; inflammatory bowel disease; lympho-organogenesis; lymphocyte migration.

Figure 1.. Characterization of mesenteric TLOs in murine ileitis and their association with CLVs.

(A-B) Quantification of immune cells in ileum lamina propria or associated mesentery at 4, 8 or 24 weeks-of-age (n=4 mice per condition). Student’s t-tests compared TNF^ΔARE/+ mice (dark red bars) to WT littermates (white bars). * p<0.05; ** p <0.01; *** p<0.001; **** p<0.0001. (C) Frequency of B (top) and T (bottom) lymphocytes among CD45+ cells whole mesentery (gray) or isolated TLOs in TNF^ΔARE/+ mice (black) or WT littermate mesentery (white) at 20–25 weeks. Experiments were performed twice, n=3–4 per group (D) Confocal imaging of TLO in Prox1cre^ERT2 Rosa26 tdTomato^fl/fl x TNF^ΔARE/+ mouse at 24 weeks. (E) Whole-mount image of a 24-week-old TNF^ΔARE/+ mouse stained for Prox1 (white) and αSMA (red). (F) Macroscopic view of the ileum-associated mesentery after i.p. injection of anti-MHC II mAb. (G-I) Confocal images of ileum-associated mesentery at 20 weeks stained for (G) MHC II (green), αSMA (red), PROX1 (white); (H) F4/80 (blue), Lyve1 (white), MHC II (green), αSMA (red); (I) B220 (blue), LYVE1 (green) and CD115 (red). (J) Mesenteric TLO H&E section. Black arrows marks vessel-like structures. (K) Serial section from panel J stained for B cells (B220, white), adipocytes (perilipin, orange), lymphatics (Lyve1, green), and nuclei (DAPI, blue). Yellow arrows mirror black arrows in panel J.

Figure 2.. Effect of TLOs on leukocyte trafficking from the ileum to distal organs.

(A) Flow cytometry plots and gating scheme showing photoconversion of ileal cells from KikGREEN to KikRED after light exposure. (B) Percent conversion among different cell types in the ileum without (white) or with light (dark red) (n=3; **** p<0.001, two-way ANOVA, Bonferroni post hoc test). (C) Last 4 cm of the ileum before (no light) and after (+ UV light) photoconversion in the KikGR mice. (D) Bright-field and fluorescent image of the mesentery bearing TLOs draining the proximal bowel (top image) or ileum-draining mesentery (bottom image) 24 h after photoconversion of the gut as in panel C. Arrow marks lymphatic flow direction (E). Pie chart shows the relative distribution of where KikRED CD45⁺ immune cells localized 24 h after photoconversion in KikGR-TNF^ΔARE/+ mice or KikGR littermates. (F) Data as in panel E, total CD45⁺ cells, DCs, T cells or B cells, leaving the gut and entering LNs, spleen, or lodging in the mesentery for KikGR littermates (white bars) and KikGR TNF^ΔARE/+ mice (dark red). (G) Photoconversion of the duodenum in KikGR TNF^ΔARE/+ mice or KikGR littermates. (H) Photoconversion of individual TLOs. (I) Comparison of KikRED CD45+ cell trafficking fate after photoconversion of TLOs in KikGR TNF^ΔARE/+ mice (dark red) near the gut wall and analysis of all downstream TLOs (mesentery) versus downstream LNs or spleen. Peyer’s patches were photoconverted as comparison (white). (J) Adoptive transfer of TFP+ CD45.1+ naïve T cells intravenously into TNF^ΔARE/+ mice (dark red) or WT littermates (white) (n=5–6 per genotype). (F-J) Experiments performed twice, n= 4 per genotype at 24–28 weeks-old mice (unless stated). Statistics show p-value from Student’s t-test when comparing both genotypes in each tissue analyzed. * p<0.05; ** p <0.01; *** p<0.001; **** p<0.0001.

Figure 3.. Differentiation and trafficking of Helicobacter-specific T cells in TNF^ΔARE/+ mice.

(A) Flow cytometry gating in the colon-draining LN and mesentery to track adoptively transferred CT2 and CT6 transgenic T cells. Gating details, Fig. S3. (B) Total CT2/CT6 T cells recovered in recipient littermate controls (WT, white) or TNF^ΔARE/+ mice (dark red) and (C) the number recovered that differentiated into FoxP3⁺ Tregs. Distribution of total recovered (D), Foxp3⁺ (E) or RORγt ⁺Foxp3⁻ (F) CT2/CT6 cells in different lymphoid tissues and the mesentery. (B-F) Data represent n=6–8 mice per group combined from 2 independent experiments. Statistics show p-value from Student’s t-test when comparing both genotypes in each tissue analyzed. * p<0.05; ** p <0.01.

Figure 4.. Effect of anti-TNF therapy on reducing intestinal inflammation and TLO progression or regression.

(A) Schematic representation of TNF^ΔARE/+ disease burden and treatments performed. (B) Impact of anti-TNF neutralizing mAb (blue) or isotype-matched control mAb (dark red) on bodyweight changes in 12-week-old TNF^ΔARE/+ mice or WT littermates (white) over an ensuing 12 weeks. Data represent n=16–18 mice per group from 3 independent experiments. Statistics show p-value from two-way ANOVA with Sidak’s post hoc test when comparing genotypes at each age. * p<0.05; ** p <0.01; *** p<0.001; **** p<0.0001 for anti-TNF mAb versus isotype mAb, and §§ p <0.01; §§§§ p<0.0001 for WT littermates versus isotype mAb treated mice. Genotypes were compared by Student’s t-test. (C) Terminal ileum histology of WT littermate controls, TNF^ΔARE/+ mice treated with control mAb or anti-TNF mAb at 24 weeks of age (after 12 weeks of treatment). (D) Quantification of mesenteric MHCII+ clusters, according to mouse genotype, anti-TNF treatment and age. Data represent n=5–10 mice per group, repeated twice. Statistics show p-value from one-way ANOVA with Tukey’s post hoc test when comparing all groups. * p<0.05; *** p<0.001; **** p<0.0001. (E) En face, whole-mount view of the mesentery of a representative TNF^ΔARE/+ mouse at the 24-week endpoint following anti-TNF therapy for 12 weeks. Prox1 (white) staining shows ancillary lymphatic vessels emerging from the CLVs where valves would be expected. Yellow arrowheads point to valves that do not have lymphangiogenic sprouts, but that nonetheless accumulate MHC II+ cells (green), αSMA marks CLVs (red). (F-G) Representative FOXC2 (green) and Prox1 (red) staining of CLV valves in WT littermates (F) versus TNF^ΔARE/+ mice (G) that show wildly distorted FOXC2 staining at TLO sites.

Figure 5.. In vivo mesenteric collecting lymphatic vessel cannulation to examine the impact of TLOs on lymph flow.

(A) Schematic representation of the lymphatic cannulation. (B-D) After cannulation, CLVs were perfused with high molecular weight FITC-Dextran (green). (B) Frame images from Movie S1, from intravital cannulation imaging of ileum-draining CLV at input pressures of 5 cm-H₂O (B^i-iii) or 34 cm-H₂O (B^iv). (C) Sequence of cannulation images (from Movie S2) of a 24-week-old TNF^ΔARE/+ (C^i-ii) CLV. Reverse lymph tracer progression towards the intestinal wall (red box, C^v-vi) starting in the third image, first row (Cⁱⁱⁱ). Last 2 images in the second row (C^vii-viii) are a different field of view in the same mouse, when high pressure was applied (C^viii). (D) Top row (D^i-iii) depicts images from Movie S2, where CLV cannulation leading to a small TLO at low input pressure (4 cm-H₂O) results in tracer extravasation (Dⁱⁱⁱ). Bottom row (D^iv-vi) shows images from Movie S2, where low input pressure fails to induce perfusion (D^iv) and increasing the pressure promoted intense leakage (D^vi). (E) CLV pressure readings from TNF^ΔARE/+ mouse (red line) and WT littermate (gray line). (F) Summary of pressure measurements made according to genotype and measurement location, n=3–5 mice per group. Statistics show p-value from one-way ANOVA with Tukey’s post hoc test when comparing both WT and TNF^ΔARE/+ (adjacent to TLO) to TNF^ΔARE/+ (afferent TLO). **** p<0.0001.

Figure 6.. Lymph flow assessment in response to anti-TNF therapy.

(A) Schematic representation of the procedure used in this figure. (B) Images of an injected Peyer’s patch (top panel) and the draining muscularis submucosa lymphatic network (bottom panel) after the injection of FITC-dextran tracer dye (green). (C) Summary of time taken for the tracer to arrive at mesLN after deposition in the Peyer’s patch according to mouse age, genotype, and anti-TNF treatment. Data represent n=4–8 mice per group, repeated twice. Statistics show p-value from one-way ANOVA with Tukey’s post hoc test when comparing all genotypes and ages. * p<0.05; *** p<0.001. (D) Fluorescence images of the tracer (FITC-Dextran, green) flow at endpoints for WT littermate (as in Movie S3) or TNF^ΔARE/+ mice without (as in Movie S4) or with anti-TNF therapy (as in Movie S6). Mesenteric lymph nodes (mesLN, magenta dashed line) and TLOs (white dashed line) are highlighted. (E) Sequence depicting backflow from a TLO toward the intestinal wall in a TNF^ΔARE/+ mouse, as shown in Movie S5. (F) Images showing lymph tracer leakage in TNF^ΔARE/+ mice, as in Movie S4. (G) Transit tracer time to arrive in mesLN after injection according to mouse genotype, two courses of anti-TNF treatment (2 weeks or 5 weeks), sex and age. Data represent n=4–7 mice per group. Statistics show p-value from one-way ANOVA with Tukey’s post hoc test when comparing all genotypes and ages. * p<0.05; **** p<0.0001. Further information is shown in Table 1.

Figure 7.. TNF impairs OSS-induced gene expression profile necessary for lymphatic valve integrity.

(A-D) Primary human LECs were culture under static or oscillatory shear stress (OSS) conditions, in the absence (vehicle control media, white bars) or presence of TNF (red bars). (A) Expression of different genes in LECs was verified by qPCR. Data represent n=3 replicates per condition, performed independently. (B) LECs under static (left columns) or OSS (right columns) condition, with or without TNF, were immunoblotted to verify the presence of the following protein at different conditions: PROX1, FOXC2, CX37. Beta-actin (ACTB, bottom) was used an input assay control. C-D) Images and the fluorescent intensity analysis (average intensity) of LECs cultured in similar conditions of panel A. Cells were stained with PROX1 (green) and β-catenin (red) (C) or FOXC2 (magenta) and β-catenin (white) (D). All values are mean ± SEM of n = 3 experiments. Statistics show p-value from two-way ANOVA with Tukey’s post hoc test comparing all stimulations. * p<0.05; ** p <0.01; *** p<0.001; **** p<0.0001. Scale bars: 50μm.