Targeting fibroblast-endothelial cell interactions in LAM pathogenesis using 3D spheroid models and spatial transcriptomics

Abstract

Lymphangioleiomyomatosis (LAM) is a progressive lung disease with limited treatments, largely because of an incomplete understanding of its pathogenesis. Lymphatic endothelial cells (LECs) invade LAM cell clusters, which include human melanoma black-45-positive epithelioid cells and smooth muscle α-actin-expressing LAM-associated fibroblasts (LAMFs). Recent evidence shows that LAMFs resemble cancer-associated fibroblasts, with LAMF-LEC interactions contributing to disease progression. To explore these mechanisms, we used spatial transcriptomics on LAM lung tissues and identified a gene cluster enriched in kinase signaling pathways linked to myofibroblasts and coexpressed with LEC markers. Kinase arrays revealed elevated PDGFR and FGFR in LAMFs. Using a 3D coculture spheroid model of primary LAMFs and LECs, we observed increased invasion in LAMF-LEC spheroids compared with non-LAM fibroblasts. Treatment with sorafenib, a multikinase inhibitor, significantly reduced invasion, outperforming rapamycin. We also verified tuberous sclerosis complex 2-deficient renal angiomyolipoma (TSC2-null AML) cells as key VEGF-A secretors; VEGF-A was suppressed by sorafenib in both TSC2-null AML cells and LAMFs. These findings highlight VEGF-A and basic FGF as potential therapeutic targets and suggest multikinase inhibition as a promising strategy for LAM.

Keywords: Cell biology; Cell migration/adhesion; Genetic diseases; Protein kinases; Pulmonology.

Figure 1. Identification of LAM nodules in patient LAM lung tissue.

(A) Tiled 20× image of H&E staining of LAM lung tissue; red arrows indicate simplified alveoli. (B) Representative images of LAM lung tissues stained with VEGFR3 (green), podoplanin (PDPN) (yellow), HMB-45 (cyan), and SMαA (red). Nuclei are counterstained with DAPI (blue) in all images. White arrows indicate alveolar cysts, orange arrows highlight LAM nodules forming near cysts, and yellow arrows in C indicate PDPN and VEGFR3 expressing LEC recruited to LAM nodules. Scale bars represent 2 mm in A, 500 μm in B, and 50 μm in C.

Figure 2. Representative staining of LAM lung tissue highlighting distinct cell populations.

(A and B) Representative images of LAM lung tissues stained with VEGFR3 (green), PDPN (yellow), HMB-45 (cyan), and SMαA (red). Nuclei are counterstained with DAPI (blue) in all images. White arrows indicate alveolar cysts, characteristic structural features of LAM. Orange arrows highlight LAM nodules forming near cysts, with HMB-45 staining serving as the gold standard for diagnosing LAM by identifying melanocytic lineage cells. Yellow arrows indicate LECs expressing both PDPN and VEGFR3, recruited to and integrated within LAM nodules, illustrating the involvement of LECs in LAM pathology.

Figure 3. Unbiased clustering of LAM tissue and marker expression.

(A and E) H&E staining of LAM lung tissues used for spatial transcriptomics. (B and F) Spatial mapping of the gene clusters for each of the lung tissues identified in the uniform manifold approximation and projections (UMAPs) provided in C and G, where colors represent distinct cellular clusters categorized by gene expression at individual spatial transcriptomic spots, for the combined dataset from both LAM tissues. (D and H) Violin plots for established LAM-associated genes represented across each cell cluster for LAM_D1 and LAM_D2 lung tissues. TGFB1I1, TGF-β1 interacting protein 1.

Figure 4. Spatial transcriptomic analysis of LAM tissue highlights LAM-core regions.

(A) UMAP where colors represent distinct cellular clusters categorized by gene expression at individual spatial transcriptomic spots, for the combined dataset from both LAM tissues. (B–D) Spatial localization of well-established LAM genes ACTA2 and VEGFD from LAM_D1 lungs showing LAM nodule localization correlating to high gene expression. (E) Violin plots of LAM-core genes in the combined dataset identifying cluster 4 as the LAM-core. (F) Heatmap for gene expression for the combined dataset where each column represents the color coded cell cluster for all differentially expressed LAM-core–enriched genes with key up- and downregulated genes highlighted. (G) Significant IPA canonical pathways’ enrichment in the LAM-core. Orange represents positive z-scores, and blue represents negative z-scores.

Figure 5. LAM-core–enriched gene signature maps to myofibroblasts in Azimuth: Human Lung v2 (HLCA) database.

(A and C) Spatial transcriptomic gene clusters from LAM_D1 and LAM_D2 tissues, respectively mapped to Lung v2 dataset for level 3 annotation, which refers to the classification of gene expression profiles into more refined cellular subtypes or functional states within the broader lung tissue hierarchy. (B and D) Spatial gene clusters represented by cell type signatures in human lung tissues for LAM_D1 and LAM_D2, respectively. (E and F) RCTD images representing LAM_D1 and LAM_D2 cell-associated gene expression. AF, alveolar fibroblasts; AdF, adventitial fibroblasts; AT1, alveolar type 1 cells; SM, smooth muscle cells; PF, peribronchial fibroblasts; AMs, alveolar macrophages. (G and H) Violin plots for LAM_D1 tissues showing relative gene expression of LAM-core genes and LEC gene PDPN with the highest expression in cluster 7 (G) mapping to myofibroblasts (H).

Figure 6. LAMFs represent an activated lung fibroblast phenotype compared with HLFs.

(A) Relative expression of SMαA comparing HLFs and LAMFs from 3 independent donors with a representative Western blot, with quantification normalized to β-actin. (B and C) Representative SMαA (green, B) and phase contrast (C) images of spheroids generated from HLFs and LAMFs; scale bars = 100 μm. (D and E) Quantification of changes in the compactness (D) and solidity (E) of spheroids over 7 days comparing LAMFs and HLFs. Each dot indicates a spheroid and a minimum of 11 (range 11–78) spheroids were evaluated. N = 3 biological replicates per cell type. (F) Kinase array for LAMFs representing expression in LAMFs relative to the average signal intensity across all proteins evaluated. (G) Heatmap of canonical pathways comparing the integrated spatial transcriptomics data with the kinase array data showing kinase-related pathway data shown have a cutoff z-score > 1 and log₁₀P value > 1.5; orange is higher pathway activation and blue is pathway inhibition. (H) quantitative reverse transcription PCR comparing gene expression of PDGFRB and TGFB1I1 in HLFs and LAMFs. Data represent mean ± SEM. Panels A and H are analyzed with an unpaired Student’s t test and panels D and E with a Mann-Whitney U 2-tailed test with significance represented by *P < 0.05, **P < 0.01, ****P < 0.0001.

Figure 7. LAMF-LEC organoids have increased invasion into the ECM.

(A) Spatial heatmap of localization of LEC genes in LAM_D1 tissue (SOX18, PDPN, LYVE1, and VEGFR3). Scale bars represent 2 mm. (B) Colocalization of core LEC gene signature and LAM-core signature genes in LAM lung tissue LAM_D1. Scale bars represent 2 mm. (C) Violin plots showing highest expression of both LEC signature genes and LAM-core signature genes in blue cluster 4, which spatially maps to histological regions of the lung tissue representing LAM nodules. The blue arrow is highlighting the cluster that is represented by the blue dots on the image above (original magnification, ×10). (D) Representative immunofluorescence (IF) images of LAMF-LEC spheroids with CellTracker Red–labeled LECs and CellTracker Green–labeled LAMFs 24 hours after seeding in 3D culture conditions. Scale bars represent 100 μm. (E) Quantification of changes in the compactness, perimeter, and solidity of the cocultured spheroids over 3 days comparing LAMFs and HLFs. Each dot indicates a spheroid and a minimum of 11 (range 11–78) spheroids were evaluated. (F) Representative images of LAMF-LEC spheroids embedded in ECM after 7 days. Scale bars represent 100 μm. (G) Quantification of changes in the compactness, perimeter, and solidity of the cocultured spheroids over 7 days comparing LAMFs and HLFs. Each dot indicates a spheroid and a minimum of 11 (range 11–78) spheroids were evaluated. (H) Representative phase contrast images of HLFs and LAMFs after 7 days of 3D culture. (Original magnification, ×10.) In all experiments N = 3, n = 9 experimental repeats. Data shown represent mean ± SEM. Panels E and G are analyzed using a Mann-Whitney U 2-tailed test significance represented by *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 8. Sorafenib treatment inhibits invasion of LAMF spheroids.

(A–F) Changes in perimeter (A and D), compactness (B and E), and solidity (C and F) comparing day 3 and day 7 for HLF and LAMF spheroids treated with either 20 nM rapamycin (Rapa) or 7 μM sorafenib (Sora), compared with vehicle (Veh). Each dot indicates a spheroid and a minimum of 11 (range 11–78) spheroids were evaluated for each of 3 independent donors (N = 3). (G) Representative phase contrast images of spheroids at day 7 of treatment for LAMFs comparing Veh with Rapa and Sora treatments. (Original magnification, ×10.) (H) Presto Blue viability assays for HLFs and LAMFs in response to increasing doses of Sora; data are expressed as a percentage of the Veh mean from 3 experimental repeats. Data shown represent mean ± SEM; A–F are analyzed with the Kruskal-Wallis test and post hoc Dunn’s multiple comparisons test. Panel H is analyzed with a 1-way ANOVA with post hoc Dunnett’s multiple comparisons tests. Significance is represented by *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, for N = 3 independent donor cells.

Figure 9. Sora treatment inhibits migration and invasion of LAMF-LEC spheroids.

(A–D) Representative confocal images of spheroids at day 4 (A and B) and day 7 (C and D) of treatment with 7 μM Sora or Veh comparing HLFs and LAMFs cocultured with LECs. Fibroblasts are stained with CellTracker Green and LECs with CellTracker Red. Scale bars in all images are 200 μm. (E) Changes in solidity, perimeter, and compactness at day 7 for HLF-LEC and LAMF-LEC spheroids treated with either 20 nM Rapa or 20 nM or 7 μM Sora, compared with Veh. Each dot indicates a spheroid and a minimum of 11 (range 11–78) spheroids were evaluated per donor. Data shown represent mean ± SEM. Panel E is analyzed with the Kruskal-Wallis test and post hoc Dunn’s multiple comparisons test, and significance is represented by *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 for each of N = 3 donors.

Figure 10. Sora inhibits secretion of pro-angiogenic cytokines from LAMF and TSC2 AML cells.

(A and B) Secreted VEGF-A, VEGF-C, and bFGF from HLFs (red) or LAMFs (blue) (A) and AML S102 (red) and S103 (blue) cells (B). (C) Gene expression of activated fibroblast markers, FAP, TGFB1, ACTA2, and PDGFRA, in HLFs induced by supernatants from either HLFs (black, control) or from AML S103 (red, TSC2⁺) and S102 (blue, TSC2^–/–) cells. Data shown represent mean ± SEM. All panels are analyzed with a 1-way ANOVA with post hoc Tukey’s multiple comparisons test. Significance is represented by *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 for each of N = 3 donors.

Figure 11. Schematic depicting the proposed cellular interactions leading to LAM pathogenesis.

LAM cells (#1) (TSC2^–/–) secrete high levels of VEGF-A, VEGF-D, and FGF2 (#2), which contribute to the activation of resident lung fibroblasts, generating activated myofibroblasts (#3). Activated fibroblasts can lead to dynamic changes in cellular motility and influence the composition of the ECM (#4), creating a unique LAM nodule niche. Activated fibroblasts can also influence alveolar stem cell behavior (#5), and these same secreted factors can recruit LECs (#6) to LAM nodules, which may also provide a pathway for LAM metastasis to other organs (#7).