Background
Plexiform fibromyxoma (PF) is a rare submucosal gastric tumor with unknown incidence that can be confused with myxoid gastrointestinal stromal tumors (GIST) [
1]. While slow growth and lack of metastases suggest an indolent natural history, these so-called “benign” tumors often present with upper gastrointestinal bleeding and gastric outlet obstruction. Although the cell of origin remains unknown, PF generally occurs in the gastric muscularis propria and frequently invades into the mucosa and submucosa [
1‐
4]. Histologically, PF is characterized by a multinodular plexiform growth pattern, diffuse myxoid stroma and prominent thin arborizing capillaries. Immunohistochemically, PF is characterized by expression of α-SMA (alpha smooth muscle actin) while lacking expression for CD117 (KIT, c-KIT), DOG-1 (discovered on GIST-1), CD34, S-100, desmin (or focal), and cytokeratins. On this basis, PF can be differentiated from GIST, as well as leiomyoma, leiomyosarcoma, schwannoma, gastric carcinoids, and desmoids [
1,
3].
To date, there has been only one report investigating the molecular biology of PF. The authors demonstrated that 5 of 16 tumors had activation of the
GLI1 oncogene, a transcription factor in the hedgehog (Hh) pathway [
5]. Two tumors were found to have
GLI1 amplification, while three had
MALAT1-
GLI1 oncogenic fusions. Both types of
GLI1 genomic alterations resulted in overexpression of GLI1 protein and activation of Hh signaling. This highly conserved pathway has been implicated in the biology of several types of cancers, including gastroblastoma [
6], basal cell carcinoma [
7], medulloblastoma [
8], rhabdomyosarcoma [
9], hepatocellular carcinoma [
10], and GIST [
11], as well as tumors with plexiform and fibromyxoid histologies [
12]. The pathway is activated when the hedgehog ligands, namely sonic hedgehog (SHH), indian hedgehog (IHH) and desert hedgehog (DHH) ligands, bind to their receptor, Patched 1 (PTCH1), a multi-pass transmembrane protein. In an unbound state, PTCH1 negatively regulates smoothened (SMO), a seven-transmembrane domain proto-oncoprotein. Following ligand binding, the PTCH1 tumor suppressor protein releases SMO inhibition, which then leads to activation of the GLI family of transcription factors that includes the GLI1 proto-oncoprotein. In turn, GLI1 regulates expression of many genes involved in cell cycle progression, including Cyclin D1 (
CCND1), as well as members of the Hh pathway itself, including PTCH1, GLI1, and hedgehog-interacting protein (HHIP) [
13]. Thus, oncogenic mutation of
SMO or
GLI1, as well as inactivating mutations of
PTCH1 can activate the pathway and downstream transcription [
13]. However, only SMO and PTCH1-altered tumors can be targeted with the three FDA-approved SMO inhibitors, namely sonidegib (Novartis/Sun), vismodegib (Genetech-Roche), and glasdegib (Pfizer).
The
PTCH1 tumor suppressor gene is located on the long arm of chromosome 9 (9q22.32). Over 500 different
PTCH1 mutations have been implicated in Gorlin syndrome (i.e., basal cell carcinoma and rhabdomyosarcoma), sporadic basal cell carcinoma, holoprosencephaly, keratocystic odontogenic tumors, and ocular developmental anomalies [
14]. Herein, we present the first report of recurrent
PTCH1 loss in plexiform fibromyxoma based on next generation sequencing. This newly identified genetic alteration is the first tumor suppressor associated with PF tumorigenesis. In turn, we examined the role of
PTCH1 inactivation on Hh signaling and investigated a rational approach to pharmacologically treating
PTCH1-, but not
GLI1-mutated PF, with SMO inhibitors.
Methods
Human subjects
Written informed consent was obtained for all study participants, including publication of clinical data. Patient tissue collection, acquisition of clinical data, and conducting experimental procedures on biological samples was approved or exempt by each institutional IRB [UC San Diego Human Research Protections Program Institutional Review Board (IRB) Protocol #090401, NCI Office of Human Subjects Research (OHSR) IRB exemption was granted for work with anonymized annotated human samples, and OHSU IRB exemption was granted for contribution of one patient to the study cohort]. All experiments were conducted in accordance with appropriate regulatory guidelines for use of human tissue. Excess explant tissue was collected for study. An index case from UCSD served as the initial observation for this study. Additional patients were retrospectively identified from the NCI and OHSU. These were analyzed in a validation cohort.
Pathologic diagnosis
Pathologic diagnosis was performed at each institution. In general, pathologic diagnosis of PF requires histopathology showing spindle shaped cells, diffuse or focally plexiform architecture, prominent thin capillaries and a bland myxoid background. Additionally, immunohistochemical staining is characteristically positive for α-SMA while negative for CD117, CD34, DOG-1, and S-100.
Primary tumor dissociation and single cell suspension
Excess fresh tumor tissue was dissociated into single cell suspensions using the gentleMACS Dissociator (Miltenyi Biotec, San Diego, CA) as previously described [
15]. Solid tissues were cut into 5-mm size pieces and were transferred to a gentleMACS C-Tube containing RPMI 1640 media and MACS human tumor dissociation enzyme solution (Miltenyi Biotec) according to manufacturer’s instructions for tough tumor tissue (h_Tumor_01). The sample was then passed through a 70 μm filter, and tumor cells were collected following centrifugation. The single cells were cultured in RPMI 1640 supplemented with 20% fetal bovine serum (FBS), 1% penicillin/streptomycin (Mediatech, Manassas, VA) and 2 mM glutamine (Mediatech).
Next generation sequencing
Sequencing was performed on formalin-fixed paraffin-embedded (FFPE) sections (6 tumors from National Cancer Institute (NCI), 1 tumor from Oregon Health & Science University (OHSU) and 1 tumor from University of California, San Diego (UCSD). FoundationOne-Heme Next Generation Sequencing (NGS) was performed on Tumor 1 as previously described [
16]. Briefly, DNA was extracted from FFPE sections with a minimum of 20% tumor tissue. DNA was submitted for comprehensive genomic profiling with hybridization-captured, adaptor ligation-based libraries. The FoundationOne-Heme test interrogates a cancer-related library of 400 gene-associated genes, 30 introns and 250 RNA transcripts tailored for hematologic malignancy and sarcoma [
17]. Tumors 2–5 were analyzed by the GeneTrails Comprehensive Solid Tumor Panel (Knight Diagnostic Laboratories, OHSU), which is a DNA sequencing panel that screens for alterations in 124 known oncogenes and tumor suppressor genes. DNA extracted from FFPE tumor tissue must provide a minimum of 50 ng DNA content, which is required to achieve a minimum depth of 250 reads per amplicon. Samples that had less than this minimum required DNA content did not undergo NGS analysis. In addition, Tumors 2–5 were subjected to RNA sequencing using the GeneTrails Solid Tumor Fusion Gene Panel (Knight Diagnostic Laboratories, OHSU), which is a partner-agnostic assay for gene fusions across 22 cancer-related target genes. RNA extracted from FFPE tumor tissue required a minimum of 30 ng RNA content for adequate analysis. Samples with less than the required amount were not analyzed by RNASeq.
RNA preparation and expression quantification
Total RNA for in vitro experiments was prepared from single cell suspension using the RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA quality was assessed with the Nanodrop 2000 (Thermo Scientific, Waltham, MA). Reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) was performed on a CFX96 real-time system (Bio-Rad Laboratories, Hercules, CA) using SsoFast EvaGreen Supermix (Bio-Rad Laboratories). Each sample was tested in triplicate. Forward and reverse primer sequences are as previously described [
10,
18]. The threshold cycle (Ct) of target genes were normalized to
ACTB according to delta-Ct method [
18]. PCR transcript expression and appropriate size were validated by gel electrophoresis using a 2.0% agarose gel in 1× Tris–acetate-EDTA buffer.
Cell viability assay
Single cell suspensions of tumor cells were seeded at 2500 cells per well on a 96-well plate (Corning, Lowell, MA). The cells were grown for 48 h and subsequently treated with Sonidegib (LDE225, Novartis, Basel, Switzerland) with a titration of 50-, 25-, 12.5- and 6.25-μM for 72 h based upon prior reports in mantle cell lymphoma, rhabdomyosarcoma, and chronic myeloid leukemia [
19‐
21]. The CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI) was performed and luminescence measured on the Tecan Infinite 200 microplate reader (Tecan, Männedorf, Switzerland).
Statistical analysis
Statistical analyses were performed using Prism GraphPad 7 (GraphPad Software, La Jolla, CA). Results are expressed as mean ± standard deviation. Comparison were made using student’s t-test and statistical significance was accepted at the 5% level. Inhibitor concentration-50 (IC50) was calculated using GraphPad 7.
Discussion
In a recent study,
GLI1 upregulation through a
MALAT1-
GLI1 fusion or copy number amplification was detected in a subset of PF [
5]. Spans et al. demonstrated the critical nature of Hh signaling in this disease. Here, we identified two cases of plexiform fibromyxoma (PF) with genomic loss of
PTCH1, a tumor suppressor gene in the hedgehog pathway. Quantitative PCR of one of these tumors supported ligand-independent activation of the Hh pathway. In addition, targeting smoothened in the context of
PTCH1 inactivation resulted in cell killing of primary PF cells. This approach represents the first in vitro use of a targeted therapeutic strategy for this rare tumor.
Our new findings lend further support for a primary role of aberrant Hh signaling in PF pathophysiology. In contrast to the previously reported
GLI1 alteration, we observed loss of
PTCH1, indicating that events occurring upstream in the Hh pathway can also drive PF growth [
5]. Although it is possible that non-canonical activation of the Hh pathway accounts for the expression of Hh pathway components, we believe this is unlikely since non-canonical Hh activation usually results in low SMO levels whereas
Smo mRNA was found to be high in Tumor 1 [
24,
25]. We also show that targeted SMO inhibition in vitro, which lies downstream of PTCH1, but upstream of GLI1, leads to dose-dependent cell killing of PF cells. This supports the role of PTCH1/SMO-axis oncogenesis since non-canonical Hh activation is SMO-independent and usually resistant to SMO inhibition. In contrast to
PTCH1-mutated PF,
GLI1 upregulated PFs are predicted to be resistant to SMO inhibition. Thus, our new findings provide evidence for a possible therapeutic role of Hh inhibition in the treatment of
PTCH1-inactivated PF. Additional testing using in vivo models will be necessary to determine whether SMO inhibition is a viable strategy for treating patients with
PTCH1-inactivated PF.
Hedgehog signaling has been associated with several tumor types. In fact, germline
PTCH1 mutations underlie nevoid basal-cell carcinoma syndrome, also known as Gorlin Syndrome. This is an autosomal dominant condition associated with distinctive facial abnormalities, as well as, basal cell carcinomas and mesenchymal tumors. Interestingly, histopathologic features of tumors associated with Gorlin Syndrome bear striking similarities to plexiform fibromyxoma. For example, Gorlin-associated odontogenic myxofibrous tumors have a bland myxoid appearance in a plexiform pattern [
26]. Similarly, a subset of pediatric gastric pericytomas have been associated with an
ACTB-
GLI1 gene fusion [
27]. Additionally, six cases of malignant epithelioid neoplasms with frequent S100-positivity were all found to have
GLI1 fusions involving
ACTB,
MALAT1 or
PTCH1 [
28]. These tumors may represent a histologically novel class of pericytoma or soft tissue sarcoma distinct from plexiform fibromyxoma. Collectively, these observations underscore a strong genotypic-to-histopathologic/morphologic relationship in
PTCH1 associated diseases.
In addition to syndromic diseases,
PTCH1 mutations have been implicated in several cancers including sporadic basal cell carcinoma (BCC) [
29], squamous cell carcinoma [
30], medulloblastoma [
8], and embryonal rhabdomyosarcoma [
31]. There are over 500
PTCH1 mutations described in human disease [
14]. The mono-allelic exon 15–24 deletion of
PTCH1 in Tumor 1 would be predicted to truncate the C-terminal region. Interestingly, elegant mutagenesis experiments in
Drosophila have shown that the C-terminus of
ptch1 is necessary for inhibition of the Hh pathway and that deletion of the last 156 residues produces a ligand-independent dominant negative phenotype [
32,
33]. Consistent with these data,
PTCH1 haploinsufficiency has been reported in variety of cancer types, including basal cell carcinoma, medulloblastoma, and rhabdomyosarcoma [
34]. Furthermore, “one-hit”
PTCH1 inactivation has been described in up to one-third of patients with nevoid basal cell carcinoma syndrome (NBCCS) who have mono-allelic
PTCH1 mutations [
35]. Taken together, this suggests that single allele inactivation of
PTCH1 can have a dominant negative effect and may account for Hh activation, as well as tumorigenesis, in a subset of PF.
Based on the prior observation of GLI1 overexpression in PF and the present identification of PTCH1 inactivation, it appears that there are two distinct subsets of PF identified to date. It is interesting to note that the previously reported MALAT1-GLI1 fusion protein implicates chromosomal loci 11q13 (MALAT1) and 12q13 (GLI1). Tumor 3 from our cohort showed loss of ERBB3 (12q13; close to GLI1) and CHEK1 (11q24). Unfortunately, the RNA from this case was too degraded to screen for a possible GLI1 fusion.
There are several limitations to our study. First, primary cells were expanded in culture for several days before performing cell killing assays. Therefore, it is possible that these cells do not fully recapitulate in vivo biology. Second, the PF cells were not immortalized and we were limited in the amount of in vitro studies that could be performed before viable cells were exhausted. Thus, additional mechanistic studies including genetic PTCH1 rescue experiments, SMO knockdown experiments, and assessment of Hh target gene expression following SMO inhibitor treatment were not feasible. Development of an immortalized PF cell line would prove an extremely useful tool in further characterization of PF pathophysiology.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.