Paragangliomas arise through an autonomous vasculo-angio-neurogenic program inhibited by imatinib

Tumours can be viewed as aberrant tissues or organs sustained by tumorigenic stem-like cells that engage into dysregulated histo/organogenetic processes. Paragangliomas, prototypical organoid tumours constituted by dysmorphic variants of the vascular and neural tissues found in normal paraganglia, provide a model to test this hypothesis. To understand the origin of paragangliomas, we built a biobank comprising 77 cases, 18 primary cultures, 4 derived cell lines, 80 patient-derived xenografts and 11 cell-derived xenografts. We comparatively investigated these unique complementary materials using morphofunctional, ultrastructural and flow cytometric assays accompanied by microRNA studies. We found that paragangliomas contain stem-like cells with hybrid mesenchymal/vasculoneural phenotype, stabilized and expanded in the derived cultures. The viability and growth of such cultures depended on the downregulation of the miR-200 and miR-34 families, which allowed high PDGFRA and ZEB1 protein expression levels. Both tumour tissue- and cell culture-derived xenografts recapitulated the vasculoneural paraganglioma structure and arose from mesenchymal-like cells through a fixed developmental sequence. First, vasculoangiogenesis organized the microenvironment, building a perivascular niche which in turn supported neurogenesis. Neuroepithelial differentiation was associated with severe mitochondrial dysfunction, not present in cultured paraganglioma cells, but acquired in vivo during xenograft formation. Vasculogenesis was the Achilles’ heel of xenograft development. In fact, imatinib, that targets endothelial-mural signalling, blocked paraganglioma xenograft formation (11 xenografts from 12 cell transplants in the control group versus 2 out of 10 in the treated group, P = 0.0015). Overall our key results were unaffected by the SDHx gene carrier status of the patient, characterized for 70 out of 77 cases. In conclusion, we explain the biphasic vasculoneural structure of paragangliomas and identify an early and pharmacologically actionable phase of paraganglioma organization. Electronic supplementary material The online version of this article (10.1007/s00401-017-1799-2) contains supplementary material, which is available to authorized users.


Table S1
List of the 77 paraganglioma cases included in the study with selected characteristics of the patients, including sex, age at surgery (for the tumor analyzed here), tumor localization, germline SDHA, SDHB, SDHC, SDHD and SDHAF2 mutation status, type and effect of the mutation, if present, and SDHB protein immunostaining All the patients were admitted for elective surgery without any previous therapy. Clinical genetic testing for germline SDHx mutations, including large deletions and rearrangements, was performed as described [1] on blood for a subset of 70 patients recruited from 11-2009 to 3-2016. SDHx mutational data for the 7 cases recruited from 5-2016 to 6-2017 are at present not available. One case (PTJ34) had lymph node metastases, one case had two independent tumors (5PC/PV), 9 patients presented with paraganglioma recurrency. Preoperative urinary and plasma catecholamines were negative, in agreement with the mostly parasympathetic (non secretory) origin of head and neck paragangliomas [2,3]. Only two patients (PTJ1 and PTJ86) reported a family history of paraganglioma, and none reported synchronous or metachronous thoracoabdominal paragangliomas or pheochromocytomas. Sites of tumor origin, encoded in the case acronyms, include carotid body (paraganglioma, carotid body: PC, 6 cases), cervical portion of the vagus nerve (paraganglioma, vagus nerve: PV, 7 cases), tympano-jugular region (paraganglioma, tympano-jugular: PTJ, 45 cases), tympanic nerve (paraganglioma, tympanic: PT, 19 cases). The median age at surgery for the series of 77 cases was 47.6 years (13-76 years), 30 patients (39.5%) were males, 46 patients (60.5%) females; median age at surgery for males was 47.8 (15-76), for females 47.6 (13-74). Overall, germline SDHx mutations were detected in 24/70 (34.3%) examined patients (5PC and 5PV are metachronous tumors of a single patient). Notably 10/24 mutations (41.6%) appear to be novel.  Particular features of our case series include higher frequency of mutations in SDHB and not in SDHD [2] and relatively high frequency of mutations in SDHA, the latter also noted in a recent independent study [4]. These features might reflect the rarity of familial PGL/PC history (see Table S1) [2,3]. Statistical differences were calculated using 2-tailed unpaired t-test (age) or 2-tailed Fisher exact test (SDHB loss).

Table S4 Flow cytometric analysis of dissociated total paraganglioma cells reveals cell populations positive for the mesenchymal surface markers CD73, CD90 and CD105
The columns in bold report the total fractions of dissociated cells positive for CD73, CD90, and CD105, the last column in italics reports the cell fractions positive for both CD73 and CD105 (CD73+/CD105+). PTJ121 1 and PTJ121 2 are distinct samples from a single tumor. Mutation status for these cases is currently not available.

Table S5 Flow cytometric analysis of dissociated total paraganglioma cells shows that the CD133/CD44-positive subset contains variable fractions of cells with CD34++/CD45-phenotype
NA: not available.
The first two columns report the total fractions of cells positive for surface CD34 (CD34++/CD45-) and for CD133/CD44 (CD133+/CD44+). The last column reports the cell fractions with CD34++/CD45 phenotype found within the CD133+/CD44+ subset. Variability is expected, because of the heterogeneous cellular composition of the tumors and of preoperative embolization.  Table S6 Flow cytometric analysis of dissociated total paraganglioma cells shows that the NCAM-and the GFAP-positive subsets contain variable fractions of cells with CD34++/CD45-phenotype NA: not available; a : also characterized for GFAP/NCAM double positivity (0.2%, 40.7%, 6.6% respectively). Furthermore, in PV87 and PV88 a fraction of the GFAP/NCAM double positive cells (14.3% and 22% respectively) was inside the CD34++ population.

SDHB
The columns in bold report the total fractions of dissociated cells positive for surface CD34 (CD34++/CD45-), surface NCAM and, when available, intracellular GFAP (variability may reflect tissue embolization and heterogeneous vascular and neural tumor tissue composition). The columns in italics report the fractions of NCAM+ or GFAP+ cells found inside the CD34++ brilliant population.   strongly positive (+++). Bimodal antigen distribution was observed only in 4 instances, for which % of positive cells are provided in the notes. The Mut/SDHB IHC column details if a germline SDHx mutation (Mut) was detected (affected gene) or not and if SDHB immunohistochemistry was positive (+) or negative (-) in the tumor from which the culture was developed. As evident here, germline SDHx mutation status did not affect the phenotype of the cell cultures. Low cell numbers precluded systematic testing of the primary cultures for all markers.

Table S8
Xenograft formation from patient-derived paraganglioma samples transplanted into NSG mice NA: not available; a : gene specified for cases with identified SDHx mutation; b : PDX formation was always confimed by epon-embedded semithin section light microscopy, electron microscopy and/or frozen section ApoTome immunofluorescence.
Relevant SDHx mutation status of the donor patient, graft sites, growth times, number of transplanted paraganglioma (PGL) samples and xenograft formation are indicated. PDX formation was obtained from 80/90 transplanted paraganglioma samples (89%). No significant differences in engraftment rates were observed between samples derived from SDHx mutation carriers versus noncarriers. The rates of engraftment were not affected by the anatomic localizations of the original tumors. Control normal tissue transplants (3 from abdominal skin of paraganglioma patients, not listed in the Table) underwent regression and calcification.