Patient-derived organoids (PDOs) as a novel in vitro model for neuroblastoma tumours

PDOs were established from 4 NB patients stage M (HR-NB with metastatic disease) (N691, N700, N711, N772) cells kindly provided by J. Molenaar (Academic Medical Centre, University of Amsterdam). Tumour biopsies were handled according to procedures reported by Bate-Eya et al. [10] to obtain single-cells and for their molecular and genomic characterization. The tumour sample was classified as NB Schwannian stroma-poor according to the International NB Pathology Committee [3] and contained more than 60% of neuroblasts in tumour cells. To initiate PDOs culture, we modified the procedure reported by Hubert CG et al., 2016 [11]. Cells suspension was embedded in Matrigel (5,250,005, Sacco S.r.l., Como, Italy) and 20 μl droplets were placed on parafilm molds and incubated for 1 h at 37 °C. Then, droplets were transferred in 12-well plates and cultured in DMEM-F12 (Aurogene, Rome, Italy) supplemented with 1% Glutamine (Life Technologies, Carlsbad, CA, USA), 1% Penicillin/Streptomycin antibiotics (Life Technologies, Carlsbad, CA, USA), 40 ng/ml basic fibroblast growth factor (bFGF) (Sigma-Aldrich, Missouri, USA), 1% B27 (Gibco, USA), 20 ng/ml epidermal growth factor (EGF) (Cell Guidance System Ltd., Cambridge, UK), 1% N2 (ThermoFisher Scientific, Massachusetts, USA), 10% BIT 9500 Serum Substitute (STEMCELL Technologies, Canada Inc.), 5% MEM non essential amino acids (Biowest, Nauillé, France), 55 μM β-mercaptoethanol (Sigma Aldrich, Missouri, USA) and 1000 U/ml leukaemia inhibitory factor (LIF) (Voden medical instruments, Monza-Brianza, Italy). PDOs were cultured for 40 days before characterisation. Images of growing organoids were acquired using DeltaPix DP 200 program (Exacta-Optech Labcenter, Modena, Italy). To quantify viable cells numbers, PDOs were collected after 30 and 60 days and dissociated by means of a solution containing PBS 1X (Carlo Erba reagents, Cornaredo, Italy), DNase (Roche, Basilea, Switzerland) and Collagenase/Dispase (Roche, Basilea, Switzerland) enzymes. Cell viability and number was evaluated with Trypan Blue (Invitrogen, California, USA) and Countess Automated Cell Counter (Invitrogen, California, USA). Two different tests of cryoconservation and expansion were performed on PDOs grown for 3 weeks. i. PDOs were dissociated to obtain single-cell suspension; cells were resuspended in cryoprotective medium (12-132A, Lonza, Walkersville, MD, USA), frozen and maintained in the vapor phase of liquid nitrogen. After 1 month, cells were thawed, resuspended in fresh organoids’ medium, tested for their viability with Trypan Blue (Invitrogen, California, USA) and directly embedded in Matrigel to re-establish the organoids, according to protocols previously described. ii. whole PDOs were frozen in 50% conditioned medium and 50% cryoprotective medium (12-132A, Lonza, Walkersville, MD, USA). Organoids were maintained for 1 month in the vapor phase of liquid nitrogen, then thawed and cultured in organoids’ complete medium. Graphs and statistical analyses were performed using GraphPad Prism software (GraphPad, La Jolla, CA, USA). All data in graphs represent the mean of at least three independent experiments ± SEM.

Immunohistochemical staining of PDOs was performed on 5 μm formalin fixed, paraffin-embedded tissue sections using a fully automated system (Bond-maX, Leica, Newcastle Upon Tyne, UK). Sections were de-waxed, rehydrated and incubated in retrieval buffer solution (Leica, Newcastle Upon Tyne, UK) for antigen recovery. Specimens were then washed with phosphate-buffered saline (pH 7.0) and incubated with the Bond Polymer Refine Detection Kit (Leica, Newcastle Upon Tyne, UK) according to the manufacturer’s protocols. Staining of proteins was performed with 3,3′-diaminobenzidine (DAB), and the slides were counterstained with Mayer’s haematoxylin (Diapath, Martinengo, Italy). Immunostains for NB84a (Leica; Newcastle, Clone NB84a, dilution 1:50), Synaptophysin (Dako, Clone DAK-SYNAP; dilution 1:100), Chromogranin A (Dako, Clone DAK-A3 dilution 1:200) and pHH3 (Thermo Scientific, dilution 1:100) were performed using an automated immunostainer. Images were acquired using Leica DM4000B microscope with Leica DFC295 camera. Measurement of the mitotic index was performed as follows: the percentage of positive tumour cell nuclei was counted in 5 × 100 cells for each case (magnification 400x, field size 0.18 mm²) in areas that are defined as “hot spots” (exactly where there are more positive nuclei).

aCGH was performed according to Agilent oligonucleotide array-based CGH for Genomic DNA Analysis Protocol (version 7.3). Samples were compared with commercial genomic DNA (Promega, Madison, Wisconsin, USA) representative for female and male. 1200 ng/μl of DNA were digested at high temperature and labelled by random priming with CY5-dCTP for the patients and CY3-dCTP for controls. Purification of the samples was performed by means of Micron filters YM-30 (Sigma-Aldrich, Missouri, USA) DNA was hybridised to a 244 K platform, at 65 °C for about 40 h. Chips were scanned on DNA microarray Agilent Scanner and digital analysis was carried out with Agilent Genomic Workbench 7.0 using the Aberration Algorithm ADM-1 with an Aberration Threshold of 5.0.

PDOs were collected and dissociated in cold lysis buffer (Biosource International, USA) supplemented with Phosphatase Inhibitor Cocktail 2 (P5726-Sigma Aldrich, Missouri, USA), Phosphatase Inhibitor Cocktail 3 (P044-Sigma Aldrich, Missouri, USA), Protease Inhibitor Cocktail (P8340-Sigma-Aldrich, Missouri, USA) and PMSF (Sigma-Aldrich, Missouri, USA). Protein concentration was determined using Pierce BCA Protein assay kit (ThermoFisher Scientific, USA) and Wallac Victor spectrophotometer (Perkin Elmer, Massachusetts, USA) and loaded in Criterion TGX Precast gel (BioRad, California, USA). Membranes were incubated overnight at 4 °C with the following primary antibodies: anti-DBH (CSB-PA958833, Flarebio biotech LLC., MD, USA); anti-LGR-5 (CSB-PA267899, Flarebio biotech LLC., MD, USA); anti-MycN (CSB-PA965036, Flarebio biotech LLC., MD, USA); anti-p75 NGF receptor (ab93934, Clone 2F1C2, Abcam, Cambridge, UK); anti-Vimentin (NB100–74564, Clone J144, Novusbiologicals, Colorado, USA); anti-GAPDH (NB300–221, Clone 1D4, Novusbiologicals, Colorado, USA); anti-YAP (#4912, Cell Signaling, Massachusetts, USA). Images were acquired using Alliance 9.7 Western Blot Imaging system (Uvitec limited).

PDOs were collected and dissociated to obtain single-cell suspension as described above. Cells were plated into 96-well plates with decreasing numbers of cells per well (1000, 750, 500, 250, 125, 100, 50, 25, 10, 5, 2 and 1). Every dilution was replicated for 24 wells for three independent experiments. Cultures were monitored at light microscope for colony formation and, after 15 days of culture, wells were scored positive or negative for the presence of at least one colony. Data were analysed using extreme limiting dilution assay. Extreme limiting dilution analysis was performed using available software (http://​bioinf.​wehi.​edu.​au/​software/​elda/​).

PDOs were dissociated to obtain single-cells as described above. 200.000 cells/tube were stained for 30′ at room temperature in the dark with: anti-CD15 FITC (IM1423U, Clone 80H5, Beckman Coulter, California, USA), anti-CD44 PE (550,989, Clone 515, Becton Dickinson, California, USA), anti-CD133 PE (130-1,/1, Miltenyi Biotech, Cologne, Germany), anti-CD29 PE (556,049, Clone HUTS-21, Becton Dickinson, California, USA), anti-CD24 ECD (IM2645, Clone ALB9, Beckman Coulter, California, USA) and anti-CD56 PeCy7 (A21692, Clone N901, Beckman Coulter, California, USA). Samples were analyzed on FC500 flow cytometer (Beckman Coulter, USA): percentages of different populations were calculated based on live-gated cells relative to physical parameters, side scatter and forward scatter. At least 30.000 live gate events were acquired. Analyses were performed using Kaluza Flow Cytometry analysis software (Beckman Coulter, USA). Five independent experiments were performed.

We generated 6 independent NB organoids (PDO1-PDO6) corresponding to 4 different HR-NB patients stage M (Additional file 1: Table S1). The number of cells for PDOs, culture conditions and the time of culture were established for PDOs. PDOs were generated starting from 5 ×10⁴, 7.5×10⁴ and 1×10⁵ cells embedded in 20 μl Matrigel and cultured for a maximum of 2 months without passaging. We observed an efficient cell growth at all cellular concentrations tested, and we chose 1 × 10⁵ cells for the following experiments of the study. The formation and growth of PDOs was again monitored for 2 months. The morphology on day 7, 14 and 40 is shown in Fig. 1. The isolated cells are self-organized in a three-dimensional structure in which both spheres and single-cells are present (Fig. 1, left panels). As cells populate the Matrigel matrix, the organoids grow with cultivation time. To quantify viable cell number, a trypan blue assay was performed after 30 days and 60 days of culturing. Our data showed an increase in viable cell numbers within day 30 of culturing in all PDOs analysed (PDO1: 68.0% ± 7.1; PDO2: 79.4% ± 0.9; PDO3: 72.1% ± 2.2; PDO4: 72.6% ± 10.0; PDO5s; 85.4% ± 1.5; PDO6: 69.0% ± 6.6) (Additional file 3: Figure S1a). Although growth rates slowed in 5/6 PDOs (PDO1: 69.9% ± 11.8; PDO2: 77.7% ± 6.7; PDO3: 59.1% ± 20.7; PDO4: 87.8% ± 1.1; PDO5: 87.3% ± 1.1; PDO6: 76.6% ± 7.3), all PDOs were stable and viable after 2 months of continuous culture without passaging. To further characterise the proliferative feature of the PDOs, we evaluated the expression of the cell cycle markers phospho-Histone H3 (pHH3) by immunohistochemistry (Additional file 3: Figure S1b). Furthermore, to strengthen the findings on PDOs ability to organise themselves as parental tumours, we compared PDOs structure versus cells growing as spheres. We used AMC691B cells for the analysis since they have a high sphere-forming capacity. Although both PDOs and spheroids well reproduced NB’s cytology, spheroids did not resemble tumours structure, while in PDOs the architecture was well recognised and mimicking NB morphology (Additional file 4: Figure S2a). This morphology had the typical appearance observed in surgical specimens from NB patients. The analysis of the cell cycle markers pHH3 by immunohistochemistry was also performed on sphere samples (Additional file 4: Figure S2b). pHH3 expression revealed a comparable mitotic index of PDOs with NB tumours compared to spheroid culture.

In order to investigate the possibility of expanding and maintaining the organoids as reservoirs of patient-derived materials, we performed tests to evaluate the ability of PDOs to be cryopreserved and expanded. PDOs were dissociated obtaining single-cell suspension and cryopreserved. After 1 month, cells were thawed, tested for their viability (PDO1: 72% ± 5; PDO2: 79% ± 2; PDO3: 61% ± 5) and directly embedded in Matrigel to re-establish the PDO (Additional file 5: Figure S3a). In addition, whole organoids were also cryopreserved and thawed after 1 month (PDO1: 50% ± 3; PDO2: 63% ± 4; PDO3: 49.5% ± 6.5) (Additional file 5: Figure S3b). In both cases, the frozen-thawed PDOs were able to grow, indicating that cryopreservation did not affect the ability to proliferate in vitro, allowing their long-term storage and retrieval.

After 40 days of culture, each PDO was paraffin-embedded, and paraffin blocks were cut to obtain 5 μm slices using a microtome. Haematoxylin-eosin (H&E) staining was performed in order to histologically characterise the PDOs in accordance with the procedure performed on the tumours of origin to classify its subtype. All primary tumours from which PDOs derived were classified as NB Schwannian stroma-poor, according to the International Neuroblastoma Pathology Committee (INPC). Moreover, at the diagnosis all tumours resulted positive for the NB markers NB84, SYP and CHGA. Histological images from primary tumour tissues can be found in the literature [10]. Similarly, all PDOs were correctly classified as Neuroblastoma, Schwannian stroma poor [3]. H&E staining highlighted the typical appearance of NB, characterised by uniformly sized cells, containing round to oval hyperchromatic nuclei and scant cytoplasm (Fig. 2, left panels). In all cases, histological examination showed a proliferation of primitive cells (arrows) with one or few prominent nucleoli, sometimes with recognisable neurite formation (asterisks) and variable mitotic and karyorrhectic activities (MKI areas, arrowhead). To further characterise the PDOs, we evaluated the expression of NB specific markers by immunohistochemistry. NB84 is a commonly employed diagnostic marker, and it was largely expressed in all the organoids examined, as evidenced by the brown cytoplasmatic staining (Fig. 2, middle panels). SYP is a transmembrane glycoprotein, also evaluated at diagnosis, and usually located at presynaptic vesicles of neurons and in vesicles in adrenal medulla. All PDOs resulted positive for this protein, depicted by the intense brown staining at the cytoplasm (Fig. 2, middle panels). CHGA is found in secretory vesicles of neurons and endocrine cells, but also widely expressed among NB tumours (Fig. 2, right panels). The positive staining for these specific markers confirmed that PDOs retained histological NB tumours features.

To identify chromosomal abnormalities and to explore the genomic features of PDOs, we performed an array comparative genomic hybridization analysis (aCGH) on NB organoids. The genomic profiles of the PDOs recapitulated the majority of the common genomic alterations in NB: MYCN amplification, 1p deletion, 11q loss and 17q gain. We confirmed MYCN amplification, a NB hallmark related to poor outcome of the HR patients in patients older than 1 year [12], 1p deletion, 11q loss and 17q gain, that occurs in about 50% of the tumours [13]. These loci are linked to the aggressiveness of this tumour [14‐17]. The PDOs aCGH profiles have been compared to the aCGH profiles of the tumours from which they derived, and the principal alterations are reported in Additional file 2: Table S2. Comparison between organoids and corresponding parental tumours showed that both their profiles are highly concordant (Fig. 3). aCGH profiles of the tumours of origin have been previously published [10]. Five out of six PDOs (PDO1, PDO2, PDO3, PDO4, PDO5) harbour MYCN amplification (green peaks in Fig. 3, Chr.2 p24.3), in accordance with the pattern observed in the original tumours. Furthermore, we detected 1p36 loss in 5/6 PDOs and the loss of the long arm of chromosome 11 only in 1/6 PDOs (violet peaks in Fig. 3). 17q gain was found in all the PDOs, whereas it was identified in 3/4 parental tumours (red peaks in Fig. 3). These apparently discordant data can be due to the presence of different sub-clones within primary NB tumours.

We explored the molecular features of the PDOs model testing their functional properties by quantifying the stem cell population and tumour-initiating capability in vitro. After 2 months of culture, we performed limiting dilution assay (LDA) of 5 of 6 PDOs. Sphere-forming cell frequency calculated using ELDA software was as following: PDO1: 1/133; PDO2: 1/209; PDO3: 1/121; PDO4: 1/369; PDO5: 1/225 (Fig. 4a). We obtained different values of stem cell frequency that reflected the different ability to invade the Matrigel and fill its boundaries. Interestingly, PDOs that displayed a higher stem cell frequency (PDO1 and PDO3) were also able to populate earlier Matrigel droplet (Fig. 1).

We performed western blot analysis of 5 out of 6 PDOs. First, we examined LGR-5 marker since it is correlated to the stemness and it was recently presented as a cancer stem cell marker for NB cells [18]. We observed the cancer stem cell marker LGR-5 expressed in all PDOs. Next, we confirmed the protein expression of MYCN in all PDOs (Fig. 4b).

NB tumours includes two types of tumour cells: undifferentiated mesenchymal-like cells and committed adrenergic tumour cells, thus resembling cells from subsequent developmental stages of the adrenergic lineage [19]. Therefore, to explore mesenchymal and adrenergic features within the PDOs, we performed western blot analyses for the expression of the sympatho-adrenal lineage markers p75 (CD271) and dopamine beta-hydroxylase (DBH), and the mesenchymal markers VIMENTIN and YAP1. The analysis of these markers revealed two general categories: PDO2 and PDO4 displayed expression of adrenergic markers DBH and p75 and were negative for the mesenchymal markers VIMENTIN and YAP1; PDO1 and PDO3 were positive for VIMENTIN and YAP1, with low protein level of the adrenergic marker DBH and p75 (Fig. 4b). Interestingly, we evidenced that PDOs enriched in cells expressing VIMENTIN and YAP1 at higher levels have the highest stem cell frequencies, as highlighted by the LDA assay (graph in Fig. 4a, black and yellow lines). To further characterise the cell subtypes we performed flow cytometry analysis. In particular, we evaluated the expression of CD56 (NCAM), a neural cell adhesion molecule, CD133, a well-known neural stem cells marker [20], CD17, a type III receptor tyrosine kinase operating in cell signal transduction, CD15 and CD29 as neural stem cells markers, and CD24 as marker of differentiated cells and neuroblasts [21]. All PDOs except PDO1, characterised by a lower percentage, showed high positivity to CD56 marker, with more than 90% of positive cells (Table 1). Consistent with the observation that a subset of NB tumours with a favourable outcome is associated with CD117 (c-kit) expression [22], we investigated the expression pattern of this molecule. All PDOs resulted nearly negative for CD117, ranging from 0.3 to 9.5% of positive cells (Table 1). PDOs showed varying levels of expression of CD133. CD133 has been described to be associated with a mesenchymal phenotype [19]. Interestingly, CD133 resulted highly expressed in PDO1 (CD133+ cells: 58,4%) (Fig. 5a, left panel), the NB organoid we previously described as a mesenchymal subtype. Pruszak et al. [21] reported the identification, based on different expression of CD15, CD29 and CD24, of three distinct subpopulations of neural lineage cells derived from human embryonic stem cells: neural stem cells (CCD15+/CD29^HI/CD24^LO), neural crest-like/mesenchymal cells (CD15−/CCD29^HI/CD24^LO) and neuroblasts (CD15−/CD29^LO/CD24^HI). The PDOs we have analysed presented remarkable percentages of CD15−/CD29^LO/CD24^HI cells, with PDO5 showing the highest proportion of neuroblasts (91.47%), followed by PDO2, PDO3, and PDO4 ranging from 70.8 to 77.6% (Fig. 5b, left panel). The lowest proportion was instead observed in PDO1, which is indeed characterised by the highest expression of the neural stem cell marker CD133 (Fig. 5b, right panel). This result again correlates with the observation that PDO1 is classified as undifferentiated mesenchymal subtype with respect to PDO2 and PDO4, characterised by a committed adrenergic phenotype. Lastly, to exclude the contamination of haematopoietic lineage cells, specific markers were investigated: CD14, CD45 and CD20. All PDOs resulted negative for these markers (Table 1).