Lymphangiogenesis is induced by mycobacterial granulomas via vascular endothelial growth factor receptor-3 and supports systemic T-cell responses against mycobacterial antigen

Abstract

Granulomatous inflammation is characteristic of many autoimmune and infectious diseases. The lymphatic drainage of these inflammatory sites remains poorly understood, despite an expanding understanding of lymphatic role in inflammation and disease. Here, we show that the lymph vessel growth factor Vegf-c is up-regulated in Bacillus Calmette-Guerin- and Mycobacterium tuberculosis-induced granulomas, and that infection results in lymph vessel sprouting and increased lymphatic area in granulomatous tissue. The observed lymphangiogenesis during infection was reduced by inhibition of vascular endothelial growth factor receptor 3. By using a model of chronic granulomatous infection, we also show that lymphatic remodeling of tissue persists despite resolution of acute infection and a 10- to 100-fold reduction in the number of bacteria and tissue-infiltrating leukocytes. Inhibition of vascular endothelial growth factor receptor 3 decreased the growth of new vessels, but also reduced the proliferation of antigen-specific T cells. Together, our data show that granuloma-up-regulated factors increase granuloma access to secondary lymph organs by lymphangiogenesis, and that this process facilitates the generation of systemic T-cell responses to granuloma-contained antigens.

Figure 1

Lymphangiogenesis during acute Bacillus Calmette-Guerin (BCG)-induced infection by CD11b⁺ granulomatous inflammation. A: Micrographs of liver sections (35 μm thick) from uninfected and BCG-infected mice shows the extent of granulomatous inflammation and CD11b⁺ cell recruitment into infected tissue. B: LYVE-1 (a lymphatic marker) staining shows a representative area with increased lymphatic endothelium in an area with proximity to a granulomatous CD11b⁺ cell inflammation. White arrowheads identify the wall of the lymphoid vessel, and orange arrowheads show the lumen. White dashed lines in the far right panels outline CD11b⁺ granuloma. C: Flow cytometry plots of granuloma-isolated cells from the liver of BCG-infected mice stained with CD11b, CD11c, and CD4 antibodies. Plots highlight the dominant CD11b⁺ granuloma population. Original magnification: ×40 (A); ×200 (B, main image); ×400 (right panels of area outlined in white dashed box, B).

Figure 2

Characterization of inflammation around LYVE-1⁺ vessels during Bacillus Calmette-Guerin (BCG) infection. A: Confocal microscopy of liver sections (60 μm thick) from CD11c-eYFP–expressing mice during acute BCG infection. Areas with granulomatous inflammation and representative lymphatic vessel are shown. Dendritic cell reporter CD11c-eYFP mice received DsRed-expressing P25 T cells that are specific for mycobacterial antigen (orange arrowheads) 1 week before harvest at 21 days postinfection. Yellow arrowheads show BCG bacilli. Right panels show enlarged area outlined by white dashed lines in left panels. Areas selected for enlargement show dendritic cells, BCG bacilli, mycobacterial-specific T cells, and BCG in a representative granuloma. B: Colocalization of LYVE-1 and ICAM-1 (white arrowheads) was detected on lymphatic endothelium in the liver of BCG-infected mice, but not uninfected controls.

Figure 3

Continued lymphangiogenesis in the liver despite resolution of bacillus Calmette-Guerin (BCG) infection. A: Proportion of liver enclosed by LYVE-1⁺ staining, as well as total number of LYVE-1⁺ areas, in liver sections (35 μm thick) from uninfected and BCG-infected mice. The proportion and number of detected LYVE-1⁺ areas were calculated based on measurements from three to five ×40 magnification fields (among sections taken from the right, median, and left liver lobes) for each mouse. B: Micrograph of a representative area of lymphatic endothelium (LYVE-1, indicated by white arrows) in uninfected and BCG-infected mice. C and D: Number of liver granulomas (C) and CFU in uninfected and BCG-infected mice (D). n = 6 to 8 mice per time point (A, C, and D).

Figure 4

Vegfr3 regulates lymphangiogenesis during mycobacterial infection. A: Vegf-c concentration from cell culture supernatants of isolated granuloma and splenic leukocytes (1 × 10⁶ cells cultured for 24 hours). Leukocytes from Bacillus Calmette-Guerin (BCG)-infected mice were isolated 3 weeks after i.p. infection (acute). Vegf-c from splenic leukocytes of uninfected mice is below the detection limit of the ELISA. B and C: Mice were treated with 10 mg/kg/day of MAZ51 (VEGFR3 blocker) after BCG infection for 7 days before harvest. B: Total normalized area of LYVE-1⁺ staining in MAZ51- and vehicle-treated mice. C: Representative area of lymphatics shows reduced surface area of LYVE-1⁺ endothelium in MAZ51-treated mice compared to vehicle-treated controls. D: Frequency of CD11b⁺ cells among entire granuloma population in MAZ51-treated and vehicle-treated groups. Representative flow plots from each group are shown. E: Representative granulomas from treated and untreated groups. F: Total number of cells and CD11b⁺ cells among treated and untreated groups. G: Total granuloma number between treated and untreated groups. One-way analysis of variance with Tukey's post hoc test (B) and t-test (A, D, F, and G) were used to determine statistical significance. Data are expressed as means ± SEM (A, B, D, F, and G). ^∗P < 0.05, ^∗∗P < 0.01 (A and B).

Figure 5

Mycobacterial-specific T-cell activation decreases with blockade of lymphangiogenesis. Bacillus Calmette-Guerin (BCG)–infected mice were treated with MAZ51 (Vegfr3 inhibitor) for 7 days before harvest at 21 days post infection. On the first day of MAZ51 treatment, mice received CFSE-labeled mycobacterial-specific P25 T cells. A: Representative flow plot from mice shows identification of transferred P25 T cells in spleen and liver draining lymph nodes (top panels), as well as CFSE dilution (bottom panels, from gate shown in A). B: Quantification of data from A. t-Test used to determine statistical significance. Data are expressed as means ± SEM (B). n = 3 to 5 (B, mice per group). ^∗P < 0.05.

Figure 6

Mycobacterium tuberculosis (MTB)-induced lymphangiogenesis in the lung. MTB infection results in murine lung lymphangiogenesis, and blockade of Vegfr3 reduces mycobacterial-specific T-cell proliferation. A: Sixty micrometer thick sections from the left lung of uninfected and MTB-infected mice shows CD11b⁺ granulomatous inflammation and increased LYVE1⁺ area that results 4 weeks after aerosol infection with MTB. B: Quantification of increased lymph endothelium, measured by LYVE-1, from image shown in A. C: Vegf-c concentration from cell culture supernatants of isolated lung and splenic leukocytes (1 × 10⁶ cells cultured for 24 hours). Leukocytes were isolated 6 weeks after aerosol infection with MTB. Vegf-c concentration from splenic leukocytes was below the detection limit of the ELISA. D: Representative area of uninfected and MTB-infected lung showing LYVE-1⁺ vessels. Dashed lines show CD11b cell–containing granulomas. E: Higher magnification shows proximity of LYVE-1 staining with CD11b⁺ cells, some of which contain tdTomato-expressing MTB bacilli (white arrows in top panel; bacilli can be distinguished by rod-like morphology). CD11b⁺ fluorescent cells can be found inside the lumen of lymphatic vessels (bottom two rows, orange arrows). F: Mycobacterial-specific, DsRed P25 T cells were transferred into MTB-infected mice that were treated with the VEGFR3 blocker MAZ51 for 10 days. Frequency of P25 T cells in proliferation (measured by CFSE dilution), as well as the frequency of P25 T cells of total leukocytes, was measured in the lung, spleen, and lung-draining lymph nodes. Representative flow cytometry plots show CFSE dilution of P25 T cells. t-Test was used to determine statistical significance. G: Cells from the draining lymph nodes of Bacillus Calmette-Guerin–infected mice were co-cultured with CFSE-labeled P25 T cells and 1 μg/mL purified mycobacterial 85b protein, along with 5 μg/mL MAZ51 (in 0.2% dimethyl sulfoxide). The proliferation of P25 cells was measured 3 days later. Data are expressed as means ± SEM (B, C, and F). n = 4 to 5 (B and C, animals per group); n = 3 (G, animals per group). ^∗P < 0.05, ^∗∗P < 0.01. Original magnification: ×40 (A, 60-μm tissue cuts); ×100 (D). LN, lymph node.

Figure 7

Obstructed lymphatics observed in a subset of granulomas. A: H&E staining of Mycobacterium tuberculosis (MTB)-infected lung (granulomas outlined in black dashed lines), which show phenotypically different granulomas in the same infected host. Non-obstructed lymphatics (area 1) and obstructed lymphatics (area 2) were further analyzed in panels B and C respectively. B and C: Fluorescent micrographs showing the differential expression of LYVE-1⁺ staining depending on the size and density of the granuloma. Granulomas are outlined in white dashed lines, and right panels in B and C show magnifications of areas indicated by white arrowheads (B, non-obstructed lymphatics), and white arrows (C, obstructed lymphatics). Most granulomatous inflammation is associated directly with or in proximity to LYVE-1⁺ vessels (B), but a few large granulomas (C) have little or no proximity to LYVE-1⁺ cells. No lymphatic obstruction is apparent in smaller granulomas, and lymphatic vessels are not plugged with cells (yellow arrowhead). Certain large lesions induce lymphatic obstruction, and LYVE-1⁺ cells are plugged with granuloma cells (orange arrowhead) and/or have low LYVE-1 staining intensity with perforated-like morphology (green arrowhead).

Figure 8

Model of lymphangiogenesis and lymphatic obstruction during granulomatous inflammation. Mycobacterial-induced granulomas around lymphatic vessels in infected tissues result in Vegf-c–mediated lymphangiogenesis via vessel-expressed Vegfr3. Increased lymphatic endothelium provides increased surface area for antigen presenting cells from the granuloma to emigrate to the lymph nodes with bacteria or bacterial antigens and prime mycobacterium-specific T cells. After adaptive immunity has been generated (indicated by white dotted line), granulomatous inflammation can resolve, though certain granulomas may continue to accumulate cells. These granulomas are of a scale and density that can result in obstructed lymphatics. Granulomas with this phenotype likely result from sustained and increasing inflammation that correlates with poor protection or deficient elimination of mycobacteria.