HYPOTHESES AND AIMS OF THE PROJECT - Patient specific stem cell models: focus on glioblastoma a

! 33!

2. HYPOTHESES AND AIMS OF THE PROJECT

Most of the neurological studies in humans are based on post mortem tissue or brain imaging. This leads to numerous biases in the analysis: postmortem tissue represents most of the time the final stage of the disease, collected tissue is extremely heterogeneous in term of treatment, age and delay in collection after death among others^96,97. Post mortem tissue has limited use to study neurodevelopmental diseases and the molecular mechanisms underlying pathological progression of disease⁹⁸. To overcome these limitations, new relevant models are needed to study specific cell types affected by disease.

The main objective in this study was to use stem cells from different origins in order to study specific critical aspects of human CNS pathology and physiology in vitro. More precisely, we used pluripotent stem cells in two different projects: (i) generate a model of glioblastoma development and (ii) generate a model to study the oxidant-generating enzyme NOX2 during neuronal differentiation:

Project 1. In the first project, we used two types of stem cells to develop an in vitro model for glioblastoma development. For this, we used an in vitro co-culture system with glioblastoma brain tumor stem cells isolated from patients and embryonic stem cell derived-engineered neural tissue. Using this co-culture system, we studied the following specific aspects:

• Molecular identification of specific genes induced during tumor-host tissue interaction in vitro.

• Comparison of main tumorigenic and malignant pathological aspects of glioblastomas multiform in a three-dimensional co-culture system with human engineered neural tissue.

• Correlation of the gene expression profile during in vitro glioblastoma tumorigenesis and patient survival.

! ! !

! 34!

Project 2. As discussed in the introduction, ROS has an important role in various CNS pathologies but also in physiological conditions. In the second part of the project we focused on the physiological role of ROS by specifically studying the ROS generating NOX2 enzyme during neural differentiation.

For this we used specific-iPSC from a NOX2-deficient patient with chronic granulomatous disease, a genetic disease due to a mutation in the CYBB gene coding for the oxidant generating NOX2 enzyme. We used these iPSC to study the role of NOX2 during neuronal differentiation. During this project we focused on the following aspects:

• Neuronal differentiation of iPSC lines in vitro.

• Comparing of disease specific iPSC lines (NOX2-/-) with control at different stages of in vitro neuronal differentiation.

• Address the potential role of NOX2-derived oxidants in vivo by focusing on adult neural stem cells in neurogenic rich brain regions of normal and genetically modified mice (NOX2-/-).

! ! !

! 35!

FIRST PART

!

Eram Garden, Shiraz

“The knowledge of anything, since all things have causes, is not acquired or complete unless it is known by its causes”

Avicenna (980 – 1037)

! ! !

! 36!

DESCRIPTION OF THE STUDY; GENE EXPRESSION PROFILE ANALYSIS OF GLIOBLASTOMA SPHERES CO-CULTURED WITH ENGINNERED NEURAL TISSUE

In the field of glioblastoma research, in vitro models are extensively used to explore tumor biology. However the great challenge in developing such models is to determine how close they resemble the human glioblastoma. This is particularly important for tumor development and responses to chemotherapy and radiotherapy. For example the discovery of tumor stem cells⁹⁹ has allowed challenging specific aspects of tumor treatment. The in vitro tumor models based on gliobtastoma stem cells have shown that they are resistant to various tumor treatment strategies including the chemotherapy and radiotherapy used in glioblastoma patients¹⁰⁰. In addition these studies have revealed a direct correlation between glioblastoma stem cell growth in tumor lesion in vitro and poor clinical outcome and patient disease progression¹⁰¹.

In the course of this thesis, we have used human stem cells to develop a three-dimensional in vitro model based on air-liquid interface culture method to study some of the aspect of glioblastoma development. For this, we have co-cultured glioblastoma spheroids derived from patients with an engineered human neural tissue (ENT) developed previously in our laboratory¹⁰². Glioblastoma spheroid cells were transiently transfected with tomato fluorescent protein (TFP) in order to track the glioblastomas cell development during the co-culture with ENT.

In the first part of this study, we aimed at the validation of the in vitro model and especially the development and survival of the tumor cells during the period of co-culture.

! ! !

! 37!

Glioblastoma spheres were co-cultured with ENT for two weeks in vitro. During this period of co-culture, glioblastoma were developing within engineered neural tissue as evidenced by immunostaining for glioma markers and presence of TFP fluorescence. Histological sections at different time points during co-culture revealed an invasion phenotype of glioblastoma tumor cells into the ENT as well as secondary tumor side formation and presence of peritumoral necrotic cells. Our result showed that our glioblastoma co-culture in vitro model displays important histopathological hallmarks of glioblastoma attributed to therapeutic and disease progression!failure of therapies in the clinic^82,103-105.

In glioblastoma patients, malignant progression of tumor into healthy adjacent tissue involves molecular changes¹⁰⁶. Once the in vitro glioblastoma model was validated, we sought to investigate molecular changes occurring during the co-culture system using comparative cDNA microarray. For this we used the following conditions; glioblastoma spheres before and after differentiation on air-liquid interface, ENT alone and glioblastoma spheres co-culture with ENT for two weeks. This analysis revealed important transcriptional changes between the different conditions. Notably, the gene expression profile of the condition glioblastoma spheres co-culture with ENT was different from that of ENT alone or glioblastoma spheres alone suggesting transcriptional changes due to tumor-host interaction.

Fold change and functional analysis revealed a significant induction and up-regulation of regulatory genes belonging to type I interferon-signaling cascade specifically in the co-culture of glioblastoma spheres with ENT. These genes are involved in host defense and known to be downstream genes of the interferon-STAT-1 signaling cascade¹⁰⁷.

In order to confirm that the observed changes in gene expression is specific for glioblastoma development into ENT, the same culture conditions were used for breast carcinoma tumor cell co-culture on ENTs. A panel of induced and up-regulated genes was

! ! !

! 38!

selected and their expression was assessed by Q-PCR for glioblastoma co-culture with ENT and breast carcinoma co-culture with ENT. This analysis confirmed that the identified profile in gene expression is restricted to glioblastoma sphere interaction with host tissue in vitro and is not induced by another type of solid tumor. Interestingly, this panel of interferon genes signaling has been identified as predictors of glioblastoma resistance to radiotherapy^108,109 and poor patient survival¹¹⁰.

The gene expression changes observed in glioblastoma-ENT co-culture were compared to data from cancer genomic atlas dataset. The significantly up-regulated genes fell into three main functional groups; i) interferon response related genes, ii) genes related to extracellular matrix and iii) cholesterol-related genes. Notably, genes related to interferon and cholesterol-related genes showed significant correlation with glioblastoma patient survival.

Taken together, development of glioblastoma within engineered neural tissue in vitro, could be used as a relevant model to study gene expression mechanism related to tumor-host interaction. This study is described in detail in a research article published in Biomaterial¹¹¹ and reported hereafter.

! ! !

! 39!

RESEARCH ARTICLE 1

!

The relationship between brain tumor cell invasion of engineered neural tissues and in vivo features of glioblastoma

Zeynab Nayerniaâ, Laurent Turchi^f^,^g, Erika Cosset^b, Hedi Petersonâ, Valérie Dutoit^c, Pierre-Yves Dietrich^c, Diderik Tirefortâ^,^b, Hervé Chneiweiss^d,

Johannes-Alexander Lobrinus^e, Karl-Heinz Krause^a, Thierry Virolle^f^,^g^,^**^,¹, Olivier Preynat-Seauve^b^,^*^,¹

aDepartment of Pathology and Immunology, Faculty of Medicine, University of Geneva, CH-1211 Geneva, Switzerland

bLaboratory of Immunohematology, Hematology Unit, Geneva University Hospital, CH-1211 Geneva, Switzerland

cLaboratory of Tumor Immunology, Centre of Oncology, Geneva University Hospital, CH-1211 Geneva, Switzerland

dINSERM U752, Faculty of Medicine, University of Paris Descartes, 75014 Paris, France

eNeuropathology Unit, Geneva University Hospital, CH-1211 Geneva, Switzerland

fUniversité de Nice-Sophia Antipolis, 06108 Nice, France

gInstitut de Biologie Valrose, CNRS UMR 7277eINSERM UMR1091, 06108 Nice, France

a r t i c l e i n f o

Glioblastoma is an aggressive brain tumor characterized by its high propensity for local invasion, for-mation of secondary foci within the brain, as well as areas of necrosis. This study aims to (i) provide a technical approach to reproduce features of the diseasein vitroand (ii) characterize the tumor/host brain tissue interaction at the molecular level. Human engineered neural tissue (ENT) obtained from plurip-otent stem cells was generated and co-cultured with human glioblastoma-initiating cells. Within two weeks, glioblastoma cells invaded the nervous tissue. This invasion displayed features of the disease in vivo: a primary tumor mass, diffuse migration of invading single cells into the nervous tissue, sec-ondary foci, as well as peritumoral cell death. Through comparative molecular analyses, this model allowed the identification of more than 100 genes that are specifically induced and up-regulated by the nervous tissue/tumor interaction. Notably the type I interferon response, extracellular matrix-related genes were most highly represented and showed a significant correlation with patient survival. In conclusion, glioblastoma development within a nervous tissue can be engineeredin vitro, providing a relevant model to study the disease and allows the identification of clinically-relevant genes induced by the tumor/host tissue interaction.

1. Introduction

Human gliomas account for more than 70% of all brain tumors and of these, glioblastoma is the most frequent and malignant histological type. The prognosis of glioblastoma is poor and fewer than 3% of glioblastoma patients are still alive 5 years after diag-nosis. Glioblastoma is characterized by inﬁltrative growth into the surrounding healthy brain, leading to secondary foci formation.

The dispersed glioblastoma cells are out of reach of surgery, chemotherapy and radiation. Thus, the tumor cannot be completely removed and these remaining invading tumor cells are thought to be responsible for the poor prognosis of glioblastoma[1].

In order to understand the biological aspects of glioblastoma development, several experimental models have been developed including derivation of glioblastoma cell lines, organotypic slices and animal models [2]. A major breakthrough in glioblastoma research was the isolation of a subpopulation of cells with stem cell features, the glioblastoma initiating cells (GIC), harboring long term self-renewal ability and the multipotency capacity to differentiate into other cell types and form heterogeneous spherical structures in suspension called gliomaspheres [3]. These spheres could be differentiated in culture into cells that phenotypically resemble the tumor from the patient[4,5]are used to study glioblastoma/brain

*Corresponding author. Geneva University Hospital, CH-1211 Geneva, Switzerland.

**Corresponding author. Université de Nice-Sophia Antipolis, 06108 Nice, France.

E-mail addresses:virolle@unice.fr(T. Virolle),olivier.preynat-seauve@hcuge.ch (O. Preynat-Seauve).

1 Equal contribution.

Contents lists available atSciVerse ScienceDirect

Biomaterials

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / b i o m a t e r i a l s

http://dx.doi.org/10.1016/j.biomaterials.2013.07.006

interactionin vitrowhen co-cultured with rodent or human orga-notypic brain slicesin vitro[6,7]or after injection into the brain of immunocompromised micein vivo[8]. Although signiﬁcant prog-ress has been achieved by foregoing experimental models, the vast majority do not mimic the pathophysiological microenvironment of human glioblastoma.

In this study, we have developed and characterized a three dimensional approach to imitatein vitro features of tumor/host nervous tissue interaction in an exclusively human environment. To achieve this goal, we used a nervous tissue engineering method developed previously[9]which allows the differentiation of human pluripotent stem cell into nervous tissue by using aireliquid interface culture. By co-culturing glioblastoma stem-like cells on the engineered nervous tissue, we establish a human tissue sharing similarities with glioblastoma development in the brain environ-ment. This includes diffuse invasion, formation of secondary foci and necrosis areas. Using expression proﬁle analysis, this model allowed us to identify general alterations in gene expression upon interaction of tumor cells with non-tumoral brain tissue.

2. Material and methods 2.1. Culture of undifferentiated ESC

The ESC cell line H1 was from WiCell Research Institute (Madison, WI,http://

www.wicell.org). H1 was maintained in 80% Dulbecco’s modiﬁed Eagle’s medium (DMEM)/Ham’s F-12 medium (F12), 20% knockout serum replacement, 2 mM L -glutamine, 1% nonessential amino acids, 0.1 m^Mb-mercaptoethanol, and 10 ng/ml bFGF. Mouse embryonicﬁbroblasts were used as feeders and isolated from embryos of pregnant CF-1 mice (Charles River Laboratories, Wilmington, MA,http://www.

criver.com). Fibroblasts were cultured in DMEM, 10% fetal bovine serum (FBS), and 1% penicillin/streptomycin. Feeders were mitotically inactivated by irradiation (45e 50 Gy) before being seeded on a 0.1% gelatin-coated plate.

2.2. ESC-derived engineered neural tissue

It was performed as previously described[9], including some minor modiﬁca-tions. Brieﬂy, ESC colonies were detached with type IV collagenase (1 mg/ml) and cultured in suspension in ultra-low attachment plates (Corning Costar, Acton, MA, http://www.corning.com/lifesciences) for 1 week in neural induction medium (DMEM/F12, N2 supplement (Gibco)) supplemented with 10m^Mphenazopyridine.

Approximately 5e10 ESC-derived clusters were plated on a hydrophilic polytetra-ﬂuoroethylene (PTFE) membrane (6 mm diameter, 0.4 lm; BioCell-Interface, La Chaux-de-Fonds, Switzerland,http://www.biocell-interface.com). This membrane was then deposited on a Millicell-CM (0.4mm) culture plate insert designed for 6-well plates (Millipore, Billerica, MA,http://www.millipore.com). One milliliter of neural induction medium supplemented with 10 m^M phenazopyridine[10]was added to each well under the membrane insert for differentiation.

2.3. Isolation and cultivation of gliomaspheres

Viable fragments of high-grade human glioblastoma were transferred to a beaker containing 0.25% trypsin in 0.1 mMEDTA solution containing and slowly stirred at 37^!C for 30e60 min. In some isolation processes, 0.6 mg/ml papaine and 10 mg/ml DNase were used instead of trypsin/EDTA. Dissociated cells were plated in 75-cm²tissue cultureﬂasks plated at 2500e5000 cells per cm²in DMEM/F-12 medium (1:1) containing the N2, G5, and B27 supplements (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com). After several passage clusters of adherent cells detach and form spheres in different size in low attachmentﬂasks (Corning Costar, Acton, MA,http://www.corning.com/lifesciences).

2.4. Immunohistochemistry

Immunohistochemistry analyses were carried out according to standard pro-tocols. Briefly, samples werefixed in 4% paraformaldehyde in Dulbecco’s BPS for 30 min at room temperature. The tissue was thenfixed into selected agar gelatine for easier handling and embedded into paraffin. Sections (10mm) were prepared with microtome deparaffinised and rehydrated. Slides were then incubated in citrate buffer (620-W microwave) for 15 min (0.01^M; pH 6.0) followed by three washes with PBS and incubated with appropriate dilutions of primary antibodies in PBS containing 0.3% Triton X-100 overnight at 4^!C. After several washes in PBS, the sections were incubated with a secondary biotinylated antibodies and ABC reagent solution for 1 h, separately (Vector laboratories,http://www.vectorlabs.com). Color was developed with 3,3⁰-diaminobenzidine (DAB, SigmaeAldrich) incubation for 10e20 min.

2.5. Antibodies

The following primary antibodies against human antigens were used: rabbit anti-cleaved caspase-3 (Cell Signaling Technology, Beverly, MA, http://www.

cellsignal.com), rabbit anti-Musashi-1, rabbit anti-nestin, mouse anti-vimentin, rabbit anti-Sox-2 (all from Chemicon, Temecula, CA,http://www.chemicon.com), rabbit anti-glial ﬁbrillary acidic protein (anti-GFAP) (all from Dako, Glostrup, Denmark,http://www.dako.com), mouse anti-bIII-tubulin (SigmaeAldrich, St. Louis, http://www.sigmaaldrich.com), rabbit anti-acrystallin B (from Abcam,http:www.

abcam.com).

2.6. RNA isolation

Isolation of total RNA was performed by using RNeasy mini kit from Qiagene following the manufacturer’s protocol. RNA quality was veriﬁed by 2100 Bioanalyzer (Agilent) and the concentration was determined by the ND-1000 Spectrophotom-eter (NanoDrop).

2.7. Microarray

For each condition, 255 ng of total RNA was used to synthesize cRNA using the Illumina TotalPrep RNA amplification kit (Ambion). cRNA concentration was measured with a spectrophotometer and cRNA quality was determined by 2100 Bioanalyzer (Agilent). For microarray 750 ng of amplified cRNA product was hy-bridized to human HT-12 v4.0 Illumina microarrays (Illumina) according to the manufacturer’s instructions. Microarray preprocessing was done using R (R Devel-opment Core Team, 2011) and BioConductor packages. Microarrays were normalized and differentially expressed genes were identified using Limma package. Probes having FDR correctedp-values less than 0.05 and fold change larger than 2.5 be-tween compared conditions were considered significantly differentially expressed.

Each such gene set ordered by fold change was characterized by Gene Ontology terms, KEGG and Reactome pathways, TRANSFAC binding predictions, MicroCosm miRNA sites, BIOGRID interaction data, human disease genes from Human Pheno-type Ontology using g:Proﬁler webtool (default settings).

2.8. Q-PCR

Isolation of total RNA was performed by using RNeasy mini kit from Qiagene following the manufacturer’s protocol. RNA concentration was assessed by nano-drop and 500 ng of total RNA was used to synthesize cDNA by using TAKARA kit following the manufactures protocol. Primer sequences were designed by Invitrogen primer design tools and are summarized inSup. Table 2. RT-PCR was carried out in optical 384-well plates and labeled by using the SYBR green master mix (Applied Biosystems), and theﬂuorescence was quantiﬁed with a Prism 7900 HT sequence detection system (Applied Biosystems). The expression of MX1, OAS1-3, IFIT-3, ISG15, C1QB, DHX58 genes was examined in a total volume of 10 _l containing 1_

SYBR green reagent, 0.86Mof each gene-speciﬁc primer pair, and 1 ng of cDNA. For ampliﬁcation the program recommended by Applied Biosystems was used (50^!C for 2 min, 95^!C for 10 min, 40 cycles of 95^!C for 15 s and 60^!C for 1 min). The relative level of each RNA was normalized to the corresponding GAPDH RNA levels. All TaqMan RT-PCR reactions were carried out in technical and biological triplicates, and the average cycle threshold (CT) values were determined.

2.9. Survival analysis

We performed survival analysis as a function of gene expression data. We used 518 normalized glioblastoma samples from The Cancer Genome Atlas (TCGA, downloaded 11.02.2013) for 10,113 genes that overlap between the TCGA and our Illumina dataset. We created single gene models using Cox Proportional Hazards regression models with coxph routine from Survival package in R. Genes were considered affecting survival if the coxph Wald test FDR correctedp-value was smaller than 0.05. Out of the 10,113 genes 349 have an effect with ap-value smaller than the threshold (1807 before FDR correction). We concentrated our analysis on genes that are up-expressed in ENTþGIC compared to ENT. We studied further genes that had fold change above 2.5. There were 132 such Illumina probes that corresponded to 115 unique gene identiﬁers, out of which 97 are present in TCGA dataset.

3. Results

The initial goal of this study was thein vitro generation of a human glioblastoma-like tissue. Glioblastoma initiating cells (GIC) were isolated from surgically removed glioblastoma and main-tained in culture as described inMaterial and methods. These cells were able to form gliomaspheres in suspension when maintained in low attachment plates (Fig. 1A). Weﬁrst addressed if a differ-entiation program into a tumor-like tissue could be reproduced Z. Nayernia et al. / Biomaterials xxx (2013) 1e12

in vitro. For this, an aireliquid interface culture method we described previously was used to differentiate GIC [9]. 15e20 gliomaspheres were placed on a polytetraﬂuoroethylene (PTFE) membrane and maintained for two weeks in neural induction medium. Once gliomaspheres were plated (Fig. 1B), the spheres merged rapidly and formed a compact cell mass after 24 h (Fig. 1C).

As we described previously for neural stem cells, after two weeks of culture, a three dimensional tissue was obtained[9]. Hemalune eosin coloration of tissue sections revealed histological features similar to the tumorin vivosuch as pleiomorphic nuclei, indistinct

Dans le document Patient specific stem cell models: focus on glioblastoma and neuronal differentiation aspects of chronic granulomatous disease (Page 35-185)