Whole-organ analysis of TGF-β-mediated remodelling of the tumour microenvironment by tissue clearing

Abstract

Tissue clearing is one of the most powerful strategies for a comprehensive analysis of disease progression. Here, we established an integrated pipeline that combines tissue clearing, 3D imaging, and machine learning and applied to a mouse tumour model of experimental lung metastasis using human lung adenocarcinoma A549 cells. This pipeline provided the spatial information of the tumour microenvironment. We further explored the role of transforming growth factor-β (TGF-β) in cancer metastasis. TGF-β-stimulated cancer cells enhanced metastatic colonization of unstimulated-cancer cells in vivo when both cells were mixed. RNA-sequencing analysis showed that expression of the genes related to coagulation and inflammation were up-regulated in TGF-β-stimulated cancer cells. Further, whole-organ analysis revealed accumulation of platelets or macrophages with TGF-β-stimulated cancer cells, suggesting that TGF-β might promote remodelling of the tumour microenvironment, enhancing the colonization of cancer cells. Hence, our integrated pipeline for 3D profiling will help the understanding of the tumour microenvironment.

Fig. 1. Visualization of the tumour microenvironment using a one-day whole-organ clearing protocol.

a Scheme of whole-organ profiling of the tumour microenvironment using tissue clearing, image acquisition, and automated image analysis. LSFM light-sheet fluorescence microscope. b Protocol of one-day whole-organ clearing (top). Bright-field images of organs (lung, brain, heart, liver, kidney, pancreas, and spleen) after fixation (RI = 1.33) and clearing (RI = 1.52). Fixed organs were stocked in PBS buffer after PFA fixation. c Visualization of the tumour microenvironment in the experimental lung metastasis model. A549-mCherry cells were intravenously injected in mice (day 0). Then, the lung was subjected to whole-organ clearing protocol and immunostained with FITC-conjugated anti-α-SMA antibody (day 7), anti-VEGFR3 antibody, Alexa 546-conjugated anti-goat IgG antibody (day 1), Red Fluorochrome (635)-conjugated anti-Iba1 antibody (day 14), or FITC-conjugated anti-Ki67 antibody (day 14). d Visualization of the platelets in the experimental lung metastasis model. A549-mCherry cells were intravenously injected in mice (hour 0). Mice were administered with DyLight 649-conjugated anti-CD42c antibody immediately. Then, the lung was subjected to whole-organ clearing protocol, followed by 3D imaging (hour 1). e Visualization of TGF-β in the experimental lung metastasis model. A549-mCherry cells were intravenously injected in mice (day 0). Mice were administered with Alexa 647-conjugated anti-TGF-β antibody (day 6). Then, the lung was subjected to whole-organ clearing protocol, followed by 3D imaging (day 7). Representative images are shown. 3D image (whole), scale bar = 2000 μm. 3D image (enlarged), scale bar = 200 μm. 2D image, scale bar = 200 μm. Figure schematic created with biorender.com.

Fig. 2. Analysis of intercellular distance using whole-organ clearing protocol.

Whole-lung imaging of cancer cells and the tumour microenvironment. A549-mCherry cells were intravenously injected into mice (day 0). Then, the lung was subjected to whole-organ clearing protocol (day 7) and immunostained with anti-VEGFR3 antibody, followed by co-immunostaining with Alexa 546-conjugated anti-goat IgG antibody and FITC-conjugated anti-α-SMA antibody. After pixel classification by ilastik, the original 16-bit images were converted into binary images. a The visualization of pixel classification processes. We set four annotations: yellow annotations are true signal, blue annotations are signal leakage along the Z-axis, red annotations are tissue autofluorescence, and green annotations are background signal. Scale bar = 50 μm (purple). b Representative images. Yellow signals show classified signals of α-SMA, VEGFR3, and mCherry. c Quantification of minimal distance from cancer cells to α-SMA, VEGFR3, or mCherry-positive cells. $, #, and * indicate α-SMA, VEGFR3, and mCherry signal at the minimal distance from A549 in 2D plane, respectively. d The empirical cumulative distribution function (ECDF) of the minimal distance from cancer cells to α-SMA or VEGFR3 signals. 3D image (whole), scale bar = 2000 μm. 3D image (enlarged), scale bar = 200 μm. 2D image, scale bar = 200 μm. Mouse number in each group is n = 3. Representative result of two independent experiments.

Fig. 3. Metastatic ability of a mixed cancer cell population of TGF-β-stimulated and unstimulated cells.

Analysis of the cooperative metastasis between the TGF-β-stimulated cancer cells and the unstimulated cancer cells. A549-mCherry or A549-GFP cells were pre-stimulated with or without TGF-β1 for 3 days, respectively, and intravenously injected in mice at the indicated number (day 0). Then, the lung was subjected to whole-organ clearing protocol (day 14), followed by 3D imaging. a Representative images. b Quantification of the metastatic colony number and the total metastatic tumour volume of A549-GFP cells in the lungs of mice. c Quantification of the metastatic colony number and the total metastatic tumour volume of A549-mCherry cells in the lungs of mice. 3D image (whole), scale bar = 2000 μm. 2D image, scale bar = 200 μm. Mouse number in each group is n = 5. Representative result of two independent experiments. Data represent the effect size as a bootstrap 95% confidence interval.

Fig. 4. TGF-β-stimulated cancer cells enhance colonization of unstimulated-cancer cells.

Temporal analysis of the effect of TGF-β-stimulated cancer cells on the metastasis of unstimulated cancer cells. Unstimulated A549-GFP cells were intravenously injected in mice (day 0). Same number of TGF-β-stimulated A549-mCherry cells were intravenously injected in the same mice at the indicated time point. Then, the lung was subjected to whole-organ clearing protocol (day 14), followed by 3D imaging. a Representative images. b Quantification of the metastatic colony number and the metastatic tumour volume of A549-GFP cells in lungs of mice. c Quantification of the colonies in which unstimulated cells were co-localized with TGF-β-stimulated cells. The ratios of GFP-positive and mCherry-positive colony number to the total GFP-positive colony number were indicated. d Density plot of the minimal distance from GFP-positive colony to mCherry-positive colony is shown. # and $ indicate the peaks around 50 µm and 250 µm, respectively. 3D image (whole), scale bar = 2000 μm. 2D image, scale bar = 200 μm. Mouse number in each group is n = 3. Representative result of two independent experiments. Data represent the effect size as a bootstrap 95% confidence interval.

Fig. 5. Expression of the tumour microenvironment-related genes in TGF-β-stimulated cancer cells.

Identification of targets of TGF-β in A549 cells using RNA-seq analysis. a Expression of epithelial-mesenchymal transition (EMT) markers in A549 cells. A549 cells were cultured in the absence (“Tβ−”) or presence of TGF-β1 (“Tβ+”) for 3 days. Then, gene expression level was determined by RNA-seq analysis. Fragments per kilobase of exon per million reads mapped (FPKM) values of each gene are shown. b–d Gene expression data obtained in a were used for gene set enrichment analysis (GSEA). A list of the most enriched hallmark gene in “Tβ+” sets is presented in b. The normalized enrichment score (NES) and false discovery rate (FDR) are shown. Representative enrichment plots of the hallmarks are shown in c. Heat map showing the expression of genes listed in the hallmarks in d. Genes which were extracted in either or both hallmarks are listed.

Fig. 6. TGF-β-stimulated cancer cells enhance metastasis of unstimulated-cancer cells through platelets.

a Visualization of cancer cells and platelets in vivo. Mice were pre-treated with DyLight 649-conjugated anti-CD42c antibody. TGF-β-stimulated A549-mCherry and unstimulated A549-GFP cells were injected into nude mice immediately (hour 0). Then, the lung was subjected to whole-organ clearing protocol (hour 1), followed by 2D imaging. The arrows indicate the co-localization of cancer cells and platelets. b, c Platelet depletion decreased cancer metastasis. Mice were pre-treated with anti-CD42b neutralizing antibody or control IgG. One hour later, TGF-β-stimulated A549-mCherry cells were injected into nude mice (day 0). Then, the lung was subjected to whole-organ clearing protocol (day 14), followed by 3D imaging. Representative images are shown in b. Quantification of the metastatic colony number and the metastatic tumour volume of cancer cells are shown in c. Mouse number in each group is n = 4. Representative result of two independent experiments. d–f Distribution of the distance between cancer cells and platelets. Mice were pretreated with DyLight 649-conjugated anti-CD42c antibody. TGF-β-stimulated A549-mCherry or unstimulated A549-mCherry cells were injected into nude mice immediately (hour 0). Then, the lung was subjected to whole-organ clearing protocol (hour 1), followed by 3D imaging. d Representative images. The arrows indicate co-localization of cancer cells and platelets. e A histogram of the minimal distance between cancer cells and platelets. f Quantification of the colonies in which stimulated or unstimulated cells were co-localized with platelets. The number of mCherry-positive and DyLight 649-positive colonies is indicated. 3D image (whole), scale bar = 2000 μm. 2D image, scale bar = 200 μm. Mouse number in each group is n = 3. Representative result of two independent experiments. Data represent the effect size as a bootstrap 95% confidence interval.

Fig. 7. TGF-β-stimulated cancer cells enhance metastasis of unstimulated-cancer cells through the activation of macrophages.

a Cytokine array of the culture supernatants of the cells obtained from the lung. Unstimulated or TGF-β-stimulated A549-mCherry cells were injected into nude mice. Cells were obtained from the lung of the nude mice 1 day after injection of cancer cells and cultured for one day. b Visualization of cancer cells and macrophages. TGF-β-stimulated A549-mCherry (Tβ+) and unstimulated A549-GFP (Tβ−) cells were injected (day 0). Then, the lung was subjected to whole-organ clearing protocol (day 1) and immunostained with anti-Iba1 antibody, followed by 2D imaging. The arrows indicate co-localization of cancer cells and macrophages. c, d Macrophage depletion decreased cancer metastasis. Mice were pretreated with clodronate or control liposome 3 and 1 day before cancer cell injections. TGF-β-stimulated A549-mCherry cells were injected into nude mice (day 0). Then, the lung was subjected to whole-organ clearing protocol (day 14), followed by 3D imaging. Representative images are shown in c. Quantification of the metastatic colony number and the metastatic tumour volume are shown in d. Mouse number in each group is n = 4. Representative result of two independent experiments. Data represent the effect size as a bootstrap 95% confidence interval. e–g Distribution of the minimal distance between cancer cells, macrophages, and VEGFR3-positive vessels. TGF-β-stimulated A549-mCherry or unstimulated A549-mCherry cells were injected (day 0). Then, the lung was subjected to whole-organ clearing protocol (day 1), followed by 3D imaging. Representative images are shown in e. The arrows indicate the co-localization of cancer cells and macrophages. A histogram of the minimal distance from macrophages to TGF-β-stimulated A549-mCherry cells or unstimulated A549-mCherry cells is shown in f. A histogram of the minimal distance from macrophages to VEGFR3-positive vessels is shown in g. 3D image (whole), scale bar = 2000 μm. 2D image, scale bar = 200 μm. Mouse number in each group is n = 4. Representative result of two independent experiments.