Modeling alveolar soft part sarcomagenesis in the mouse: a role for lactate in the tumor microenvironment

Abstract

Alveolar soft part sarcoma (ASPS), a deadly soft tissue malignancy with a predilection for adolescents and young adults, associates consistently with t(X;17) translocations that generate the fusion gene ASPSCR1-TFE3. We proved the oncogenic capacity of this fusion gene by driving sarcomagenesis in mice from conditional ASPSCR1-TFE3 expression. The completely penetrant tumors were indistinguishable from human ASPS by histology and gene expression. They formed preferentially in the anatomic environment highest in lactate, the cranial vault, expressed high levels of lactate importers, harbored abundant mitochondria, metabolized lactate as a metabolic substrate, and responded to the administration of exogenous lactate with tumor cell proliferation and angiogenesis. These data demonstrate lactate's role as a driver of alveolar soft part sarcomagenesis.

Figure 1. A mouse allele with conditional expression of human ASPSCR1-TFE3

(A) Schematic demonstrating generation of the fusion transcript in human alveolar soft part sarcoma, which was reverse transcribed to a complementary DNA (cDNA) and targeted to the Rosa26 locus, separated from its promoter by a loxP-flanked stop sequence consisting of the neomycin resistance gene (neoR) and a poly-adenylation transcription termination sequence. The cDNA is followed by enhanced green fluorescent protein (eGFP) and internal ribosomal entry site (IRES), such that both the fusion gene and eGFP will be translated from the same transcript following Cre-mediated recombination to remove the stop. (B) Photomicrographs of embryonic day 12.5 Rosa26^LSL-AT3/_WT fibroblasts 48 hours after exposure to TATCre or control, demonstrating GFP fluorescence from the recombined allele and (C) reverse transcriptase polymerase chain reaction (RT-PCR) showing presence of the fusion gene transcript following TATCre, but not control. (Scale bars are 100µm in length.)

Figure 2. ASPSCR1-TFE3 induces rapid tumorigenesis, restricted to the cranium

(A) Chart depicting the relative rates of tumorigenesis and (B) diagrams of the anatomic distribution of tumors among Rosa26-^{LSL-EWSR1-ATF1}/_CreER, Rosa26-^{LSL-SS18-SSX2}/_CreER, and Rosa26-^LSL-AT3/_CreER mice. Lateral cranium radiographs (C–D), gross dissection mid-sagittal brain photos (E–F), and GFP fluorescence photos (G–H) of 3 month old Rosa26-^LSL-AT3/_CreER (C, E, G) and Rosa26-^LSL-AT3/_WT (D, F, H) mice showing occipital expansion (between black arrows, C) from a tumor that developed in the brain (white arrows, E and G) and demonstrates GFP fluorescence from the active AT3 allele. Representative photomicrographs of the gamut of intra-cranial tumor presentations in Rosa26-^LSLAT3/_CreER mice, in which most were intra-parenchymal (I), but one encroached on the choroid plexus (J) and another developed within the eye (K). (Scale bars are 500µm in length.)

Figure 3. ASPSCR1-TFE3-induced mouse tumors recapitulate human alveolar soft part sarcoma diagnostic histopathology

(A) Low- and (B) high-power hematoxylin and eosin photomicrographs from tumors arising in Rosa26-^LSL-AT3/_CreER mice match those from human ASPSs (C and D) in the overall tissue architecture of loosely nesting cells between arcading capillary networks and cellular features of round large nuclei with prominent nucleoli. (E) Photomicrograph of diastase-resistant periodic acid Schiff staining identifies the pathognomonic cytoplasmic granules (black arrows) in Rosa26-^LSL-AT3/_CreER mouse tumors and human ASPSs (F). Clinical immunohistochemistry against human TFE3 demonstrates strong nuclear staining in both Rosa26-^LSL-AT3/_CreER mouse tumors (G) and human ASPSs (H). H&E photomicrographs demonstrate a tumor growing between sulci near the investing pia mater (arrows in I), which was sometimes maintained between the tumor cells and the surrounding tissues (arrows in J); other invading tumor cells into the cerebellum (K), cerebral cortex (L), and choroid plexus (M) maintained no intervening leptomeningeal layer, but often associated closely with small vessels (arrows in N). (All scale bars and the width of panels B, D, E, and F are 40µm in length.)

Figure 4. ASPSCR1-TFE3-induced mouse tumors recapitulate human alveolar soft part sarcoma by transcriptome profile

(A) Series of Broad Institute gene set enrichment analysis (GSEA) plots testing the 500 most significantly up-regulated (left, in each pair) and down-regulated (right, in each pair) genes from the Rosa26-^LSL-AT3/_CreER mouse tumor (mASPS) to muscle RNAseq comparison, the human alveolar soft part sarcoma (hASPS) to muscle RNAseq comparison, and the human ASPS to muscle microarray comparison (GEO), in each of the other data sets. All of the up-regulated gene lists had high normalized enrichment scores (NES) and were statistically significant with false discover rate q-values less than 0.0005. None of the down-regulated gene lists showed enrichment. (B) Heat map of log2 transformed fold-changes from the control mean of the 539 genes with at least 2-fold and significant mean up-regulation in all three data sets. (C) Chart of the Z-scores of the most significant biological process gene ontologies from GO-Elite analysis of the 539 genes up-regulated in all three data sets. See also Table S1.

Figure 5. ASPSCR1-TFE3-induced tumorigenesis is most efficient within the cranium

(A) Diagram depicting the relative incidence and representative histopathology of the two grouped anatomic locations of tumors arising between ages 13 and 16 weeks among 6 Prx1-CreERT2;Rosa26-^LSL-AT3/_WT littermate mice after receiving tamoxifen at age 2 weeks. (B) Photomicrographs of X-gal-stained sections from the cranium’s inner and outer tables, as well as the proximal humerus in 3 week old Prx1-CreERT2;Rosa26-^LacZ/_WT mice after receiving tamoxifen at age 2 weeks, demonstrating the Prx1-CreERT2 lineage in each location (blue cells identified by black arrows), only the first of which led to identifiable tumorigenesis when initiating the AT3 allele. (Scale bars in A and panel widths in B are each 100µm in length.)

Figure 6. ASPSCR1-TFE3-induced mouse tumor lactate transport gene expression and tissue lactate levels in the mouse

(A) Chart of the gene expression by fragments per kilobase of exon per million fragments mapped (FPKM) in Rosa26-^LSL-AT3/_CreER mouse tumors (black) and muscle samples (red), showing high and significantly up-regulated expression of both Cd147 and Mct1 in tumors and low and significantly down-regulated expression of Mct4. (B) Plot of the individual (dots) and mean (bars) concentrations of lactate in the tissues of 5 mice, determined biochemically.

Figure 7. ASPSCR1-TFE3-induced mouse tumors and human alveolar soft part sarcomas can metabolize lactate

Heat maps from Rosa26-^LSL-AT3/_CreER mouse tumors (A) and human alveolar soft part sarcomas (B) showing the log2 ratios of expression normalized to the mean for the Broad Institute’s curated MitoCarta gene list. (C) Chart of the mean ± standard deviation of the oxygen consumption rate (normalized by DNA content) of tissue sections from tumors arising in mice induced by each of three fusion oncogenes, representing alveolar soft part sarcoma, synovial sarcoma, and clear cell sarcoma (n = 10 for each). (D) Chart of the mean ± standard deviation of the percent change in oxygen consumption among fresh tissue section from three mouse tumor types upon administration of 10mM lactate substrate (n = 5 for each).

Figure 8. ASPSCR1-TFE3-induced mouse tumors express hypoxia genes in the absence of hypoxia and respond to exogenous lactate

(A) Explanatory diagram of the pathway by which either hypoxia or imported lactate can decrease 2-oxoglutarate-mediated Hif1α hydroxylation and degradation (Lu et al., 2002) (Lu et al., 2005). (B) Photomicrographs of pimonidazole-treated Rosa26-^LSL-AT3/_CreER mouse tumor and kidney tissue sections stained with immunohistochemistry against pimonidazole adducts (above; brown depicts presence of hypoxia) and Hif1α (below), demonstrating nuclear Hif1α, but no hypoxia in tumors, and Hif1α correlating to hypoxia in normal renal tubules. (C) Broad Institute gene set enrichment analysis for the established Harris hypoxia gene list in the Rosa26-^LSL-AT3/_CreER mouse tumor (mASPS) to muscle RNAseq comparison, showing upregulation of many ostensibly hypoxia-driven genes in the tumors. (Scale bars are 10µm in length.) (D) Representative photomicrographs of immunohistochemistry against Ki-67, indicating proliferating nuclei (brown) or Cd31, indicating vascular endothelial cells (brown) after 2 weeks of lactate or saline administration to Rosa26-^LSL-AT3/_CreER mice bearing tumors. (E) Chart of the mean ± standard deviation of the ratio of Ki-67-positive tumor cell nuclei to total tumor cell nuclei between the two groups (n = 5 lactate and 7 saline controls). (F) Chart of the mean ± standard deviation of the ratio of the linear vessel perimeter to total tumor cell nuclei between the two groups (n = 5 lactate and 7 saline controls). (G) Charts presenting mean ± standard deviation of the relative viability of ASPSCR1-TFE3-expressing human cancer cell lines FU-UR-1 (at 48 hours, G) and ASPS-1 (at 72 hours, H) in increasing concentrations of sodium lactate (n = 3 for each point; experiments independently repeated thrice). (I) Western blots showing increased nuclear HIF1α when exposed to 20mM sodium lactate that is blocked by the MCT1-inhibitor, CHC at 0.1mM in FU-UR-1 cells and 1mM in ASPS-1 cells.