Synovial sarcoma is a soft-tissue malignancy with mesenchymal origins [1
]. These tumors are predominantly found in the extremities, can arise anywhere in the body, and have a predilection for developing adjacent to bone [1
]. One hypothesis is that the bone provides an anti-apoptotic niche for transformed cells to grow by secreting the decoy receptor osteoprotegerin (OPG) [4
]. Furthermore, we traced one cell of origin to a mesenchymal progenitor found in the periosteum [4
]. Not only do synovial sarcomas arise next to bone, but ossification can occur within human synovial sarcomas [5
]. This is a rare event (< 10%) in an already rare cancer type: 3 patients per 1 million [1
]. The impact of ossification in synovial sarcoma remains unclear. However, we hypothesized that ossification in synovial sarcoma is correlated with a worse prognosis after observing it in our metastatic mouse model.
To understand the prognostic impact of ossification in synovial sarcoma, we evaluated the prevalence and outcomes in mice with synovial sarcoma, using genetically engineered mice that express either the fusion oncogene SS18-SSX1
]. These mice have been studied extensively for their pathophysiology, histology, and molecular expression profiles [2
]. Iterations with accompanying genetic manipulations of tumor suppressor genes and oncogenes have been evaluated to approach the most relevant preclinical model. The model that best recapitulates the metastatic biology of human synovial sarcoma is the combined expression of SS18-SSX1
and the deletion of Pten
via the injection of TATCre, a critical tumor suppressor gene that acts to suppress the over-proliferation of tumor cells [12
]. This genetic manipulation is temporally and spatially controlled by the injection of the protein TATCre. In this study, we looked at the prevalence of ossification within genetically engineered mouse models of synovial sarcoma and correlated the survival and tumor progression with the presence of ossification.
Prior to our investigation, there was uncertainty about the ossification phenotype observed in synovial sarcoma and its correlation to outcome [5
]. Using a mouse model of synovial sarcoma that exhibits an overall penetrance of 7.2% ossification, we examined the correlation of ossification with pulmonary metastasis and outcome. While we did see more ossification within our metastatic model of synovial sarcoma than for the nonmetastatic model, we demonstrated that a nonsignificant trend in mice with ossifying primary tumors showed gross metastatic lesions in the lungs upon necropsy. However, the evaluation of survival was more favorable for mice with detectable ossification. This led us to conclude that both the development of pulmonary metastasis and ossifying centers in the primary tumor are independent from each other but that both are dependent on time.
Synovial sarcomas were initially thought to arise from muscle precursors [2
] but have been shown to originate more readily within the periosteum, the highly vascularized connective tissue that surrounds bone [4
]. Though it is a phenotype with a low penetrance and which is not readily apparent, most synovial sarcomas express genes involved in the development of bone [23
]. As examined above, ossification and metastasis were not immediately correlated with a worse prognosis, but this can likely be attributed to the morbidity of primary tumor growth necessitating humane euthanasia in faster growing tumors before significant ossification could occur.
Also, worth noting is the consideration that, as observed in Figure 2
b, mice heterozygous for hSS1
showed a longer survival period than homozygous mice. Having a heterozygous genotype may lead to a greater amount of differentiation within the tumor environment, translating into a better overall survival or prognosis. A higher degree of differentiation often relates to less migration and an easier identification of the original tumor source [24
An interesting element was the presence of ossification genes, even in the absence of an osteoid matrix and calcium crystal deposits in the primary tumors. While it is apparent that bone development genes from our list of interest are responsible for ossification within synovial sarcoma, it is less clear what factors are driving these bone development genes to be upregulated. One possible consideration is inflammation within the tumor environment. Such inflammation incites an immune response, galvanizing immune cells such as lymphocytes, macrophages, and neutrophils to infiltrate the primary site and secrete growth signaling molecules that are responsible for transforming mesenchymal stem cells into differentiated osteoblast cells [25
silencing acts as a major step in promoting and maintaining baseline inflammation through PI3K/AKT signal transduction, resulting in the recruitment of macrophages and neutrophils to the tumor microenvironment [27
]. The accumulation of these immune cells leads to the deactivation of T-cells and natural killer cells, which acts to limit cancer cell elimination while simultaneously secreting growth factors such as VEGF and TGF-α, thus promoting angiogenesis and facilitating the endothelial-mesenchymal transition (EndMT) [28
]. Both of these features act to promote metastasis, while also simultaneously triggering bone reactivity and development. This corresponds well to the patterns of ossification observed in our mouse tumor samples
The most interesting finding was perhaps the PTHLH
upregulation within the sequencing results among human patients between nonmetastatic and metastatic groups. PTHLH
aids in regulating cell differentiation and proliferation, particularly in bone development, and promotes cell migration and invasion [16
]. A significant prevention of apoptosis has also been observed with PTHLH
(runt-related transcription factor 2), another gene involved in osteoblast differentiation, may stimulate PTHLH
expression through IHH
(Indian Hedgehog) [15
], both of whose expressions are upregulated in our metastatic model when compared to the muscle model (Figure 4
a). The increased expression of RUNX2
, as seen above, correlates with an increase in PTHLH
expression. Coupled with the upregulation of PTH1R
(parathyroid hormone type 1 receptor) seen in our mouse models, PTHLH
likely plays a significant role in cancer cell proliferation, migration, and invasion [15
While thoroughly discussed here in the context of synovial sarcoma, soft tissue ossification is not unique to neoplastic disease. Indeed, there are a number of non-neoplastic, progressive, inherited, and acquired disorders of inappropriate bone formation in soft tissues, collectively called heterotopic ossification (HO). Several lab-generated mouse models for both ossifying sarcomas (such as those used in the current study) and HO disorders, have been developed to provide insights into the mechanistic beginnings of HO. Recently, the American alligator (Alligator mississippiensis
) has been established as a natural animal model for HO because of the development of ectopic bone in some scales [30
]. Osteoderm formation shares histologic and, presumably, mechanistic similarities to HO. The development of a boney matrix within the mature dermis initiates approximately 9–12 months post-hatch, when mesenchymal stem cells differentiate into osteoblasts to form qualitatively normal bone [30
]. Most forms of HO are triggered and exacerbated by trauma, when faulty mechanisms of inflammation initiate the aberrant cell differentiation of mesenchymal stem cells involved in tissue regeneration during wound healing [31
]. Consequently, a local osteogenic program is expressed and leads to the production of heterotopic bone in the affected tissue. The initiating and driving factors for ossification in both osteoderms and HO are the subject of ongoing investigations.
Ossifying synovial sarcomas might undergo an ossification process that is similar to HO lesions, as indicated by the presence of a histologically typical bone matrix, the expression by tumor cells of osteogenic genes, and the contribution of the inflammatory tumor environment to the ossification phenotype. As such, the American alligator may also become an appropriate model for soft tissue ossification in neoplastic disease as well as for HO disorders. Future work to describe the initiation, formation, and prognosis of synovial sarcoma along a protracted developmental time scale will be needed to identify the link between aberrant cell fate mechanisms and tissue environment cues that initiate bone formation in synovial sarcoma. A comparative model between HO and synovial sarcoma could provide a platform to deconstruct conserved components and identify common therapeutic targets for the treatment of both disease states.
4. Materials and Methods
Mouse experiments were conducted with the approval of the University of Utah’s Institutional Animal Care Committee in accordance with legal and ethical standards established by the National Research Council and published in the Guide for the Care and Use of Laboratory Animals (protocol # 14-01016). The previously described Rosa26-LSL-SS18-SSX1;Ptenfl/fl and Rosa26-LSL-SS18-SSX2;Ptenfl/fl mice were maintained on a mixed strain background, C57BL/6 and SvJ. Mice were genotyped with the following primers: Rosa26-LSL-SS18-SSX (F flox—AAACCGCGAAGAGTTTGTCCTC, F wt—GTTATCAGTAAGGGAGCTGCAGTGG, R—GGCGGATCACAAGCAATAATAACC) Pten (F flox—CAAGCACTCTGCGAACTGAG, R—AAGTTTTTGAAGGCAAGATGC). TATCre was dosed by 10 μL intramuscular injections at 50 μM at 1 month of age.
Mouse tissues were fixed in 4% paraformaldehyde overnight and embedded in paraffin. Paraffin-embedded tissues were stained by immunohistochemistry by rehydrating slides through a citrosolv and ethanol dilution wash. Hematoxylin and eosin staining was performed as previously described [12
4.3. Transcriptome Analyses
Total RNA was isolated from murine synovial sarcomas and from normal muscle taken from the Sartorius and Rectus femoris with the RNeasy mini kit (QIAGEN, Germantown, MD, USA). For the transcriptome sequencing of TATCre tumors, RNA was prepared using the Illumina TruSeq RNA kit (Illumina, San Diego, CA, USA), checked with the Bioanalyzer RNA 6000 chip (Agilent Technologies), captured using the RiboZero method (Illumina, San Diego, CA, USA), and 50-cycle end-read sequenced on an Illumina HiSeq 2000. Reference fasta files were generated by combining the chromosome sequences from mm10 with splice junction sequences generated by USeq (v8.8.8, Source Forge, Salt Lake City, UT, USA) MakeTranscriptome using Ensembl transcript models (build 74). Reads were aligned with Novoalign (v2.08.01, Novocraft, Selangor, Malaysia), allowing up to 50 alignments per read. USeq’s SamTranscriptomeParser selected the best alignment for each and converted the coordinates of reads aligning to splices back to genomic space. The differential gene expression was measured using USeq DefinedRegionDifferentialSeq, which counts the number of reads aligned to each gene and then calls DESeq2 (v1.4.5, Bioconductor) using default settings.
For the NanoString gene expression analysis, RNA was isolated from three different tumor cell populations taken from Rosa26-LSL-SS18-SSX2;Ptenfl/fl
: GFP+, CD11b+/Ly6C+/Ly6G+, and CD11b+/Ly6Cmid/F4-80+/MHC IIhigh as previously described [12
]. In brief, specimens were minced, enzymatically digested (Tumor Dissociation kit; Miltenyi Biotec, San Diego, CA, USA), and mechanically dissociated (GentleMACS Tissue Homogenizer; Miltenyi Biotec, San Diego, CA, USA). Tissue debris was removed using 70-µm MACS Smart strainers (Miltenyi Biotec, San Diego, CA, USA). Cells from all tissues were washed with PBS containing 0.5 mg/mL bovine serum albumin, and then stained using a cocktail of rat anti–mouse antibodies from BD: Ly6C, CD11b, I-A/I-E, F4/80, Ly-6G, and DAPI (Invitrogen, Carlsbad, CA, USA). The cell expression was determined by multicolor flow cytometry on a FACSCanto flow cytometer (BD, San Jose, CA, USA), and cell sorting was performed using a FACSAria (BD, San Jose, CA, USA) and FlowJo 8.7.1 (FlowJo, Ashland, OR, USA) for analysis. 10 ng of RNA was combined with the mouse immunology panel of 545 gene probes and analyzed on the nCounter (NanoString, Seattle, WA, USA).
4.4. Statistics and Analysis
Genes of interest were selected from a list of bone development genes that were then verified to have bone expression using https://www.genecards.org/
. After confirming the gene, the expression was checked under the TISSUES section. Any genes that had expression in bone or bone marrow were selected as genes of interest.
The Pten loss-induced tumor group, our metastatic model (n = 5), and the muscle tissue group (n = 3) were sampled, and the tissues were processed to determine the gene expression. Fragments per Kilobase of transcript per Million mapped reads (FPKM), Log2, and Adjusted p-values were obtained from analysis, and the results were ordered from highest to lowest adjusted p-value. Using these FPKM values, a heatmap was generated for all genes of interest.
Shared genes of interest between metastatic vs. muscle and metastatic vs. nonmetastatic tissues were found by sorting columns alphabetically and finding the overlap and were then plotted using Adobe Illustrator in an area-proportional Venn diagram showing the overlap of genes shared between both groups.
Human tumor sample data sets were accessed using NCBI GEO, Series GSE54187, from the Stanford University Department of Pathology. The gene expression data was extracted and analyzed using an unpaired t-test to determine if a statistical significance between metastatic and nonmetastatic tumor samples existed for each gene of interest.
The NanoString data was analyzed for significance by averaging the fluorescence signal from the tumor cells (n = 4) and MDSC cells (neutrophils, macrophages, monocytes, n = 8) for ossifying genes and performing a one-way ANOVA to determine the statistical significance.