Novel preclinical models, therapies and biomarkers for testicular cancer
Rosas Plaza, Fernanda
DOI:
10.33612/diss.119056452
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Rosas Plaza, F. (2020). Novel preclinical models, therapies and biomarkers for testicular cancer. Rijksuniversiteit Groningen. https://doi.org/10.33612/diss.119056452
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Chapter 4
Establishment and characterization of testicular
cancer patient-derived xenograft models for
preclinical evaluation of novel therapeutic strategies
Ximena Rosas-Plaza, Gerda de Vries, Gert Jan Meersma, Vincent C. Leeuwenburgh, Albert J.H. Suurmeijer, Marcel A. van Vugt, Jourik A. Gietema, and Steven de Jong.
Testicular cancer (TC) is the most common solid tumor in young men. While cisplatin-based chemotherapy is highly effective in TC patients, chemoresistance still accounts for 10% of disease related deaths. Pre-clinical models that faithfully reflect the patient tumor are needed to assist in target discovery and advanced drug development. Patient derived xenograft (PDX) models of TC were established by subcutaneous implantation of solid tumor pieces in NOD scid gamma (NSG) mice. From 8 TC tumors that were implanted, 3 were established as a PDX model. PDX models and matched patient tumors were characterized using immunohistochemistry, showing retention of histological subtypes over several passages. Copy number variation analysis and RNA-sequencing was performed on PDX tumors to assess the effect of passaging, showing high concordance between passages. Chemosensitivity of PDX models corresponded with patients’ response to chemotherapy. In conclusion, we describe the establishment and characterization of 3 TC PDX models. These models faithfully reflected chemosensitivity and can now be used for mechanistic studies, therapeutic development and pre-clinical validation of novel therapeutic strategies in testicular cancer.
Introduction
Testicular cancer is one of the most common solid tumors in young men between
20-40 years of age and the incidence is rising worldwide1. TC can be divided in two types,
seminomas and non-seminomas, each accounting for approximately 50% of cases2.
Seminomas resemble undifferentiated spermatogonia with low metastatic potential. Non-seminomas can be further divided in 4 histological subtypes displaying varying stages of differentiation. Embryonal carcinoma (EC) is the most undifferentiated type expressing some pluripotency markers. Yolk sac carcinoma (YS) and choriocarcinomas (CC) resemble extraembryonic differentiated tissues and express alpha-fetoprotein (AFP) and human chorionic gonadotropin, respectively. The last subtype, teratoma, shows several patterns of somatic differentiation, which can be either incomplete
(immature teratoma) or well differentiated (mature teratoma)3. For non-seminomas,
usually a mixture of histological subtypes is present while tumors with pure histology are less frequent. Mixtures of both seminoma and non-seminoma components are also frequently observed.
TC is one of the few solid tumors that can effectively be treated with cisplatin-based chemotherapy. For metastatic disease, first line chemotherapy will achieve cure rates as high as 80%, and salvage therapy cures another 10% of patients. The IGCCC stratification classifies patients with different survival rates and poor risk patients have the lowest 5-year survival, with a survival rate of only a 50%. However, oncologists cannot predict which patients will not respond to chemotherapy or will
develop a relapse4.
Preclinical testing of new therapeutic agents or combination strategies in testicular cancer is mainly performed in cell lines or cell line xenograft models. While cell lines can be great tools for target discovery and mechanistic studies, they also have important limitations when it comes to preclinical drug testing. For TC, the number of cell lines available is limited, with around 20 cell lines described in
literature5,6. In addition, not all TC subtypes are well represented in these cell line
models, because most of them are of the EC subtype. Another limitation of using cell line models, especially in the context of TC, is the lack of tumor heterogeneity as most non-seminoma tumors are mixed tumors, a characteristic that will not be recapitulated by cell lines. To overcome the limitations presented by cell line models and cell line xenografts, much preclinical research is now being performed using patient derived xenograft (PDX) models. Advantages of PDX models include the histological preservation of the tumor when serially transplanted in different generations of mice, and the molecular resemblance to the original tumor looking
at genomic features and expression levels7–12.
Methodologies for establishing PDX models are diverse in terms of implantation site,
have been established and described14–17. These TC PDX models, implanted either orthotopically or subcutaneously, were derived from both primary and metastatic tissue and represent all non-seminoma subtypes. We studied the feasibility of establishing subcutaneous TC PDX models. Next, we characterized these tumors and evaluated concordance of primary tumor tissue with the established PDX tumors over serial passaging.
Results
Establishment and biobanking of testicular cancer PDX models
Between March 2016 and March 2019, tumor samples from 8 patients were collected and implanted in male NOD-scid gamma mice, of which 7 were diagnosed with TC (Table 1). Most tumors (7/8) were obtained via orchiectomy and implanted within 4 hours after surgery. One tumor (TC5) was stored overnight at room temperature in RPMI/10% FCS before implantation. PDX model TC4 was established from a needle biopsy of a metastatic tumor taken from a patient with refractory disease. Of these tumors, 3 PDXs were successfully developed with an engraftment rate of 27% (Table 1). Median latency time, the time between tumor implantation and tumor growth, was 25 days (range 12-97) with large variation between tumors from different patients and between tumor pieces from individual patients (Table 1).
Table 1 Successfully established TC PDX models from March 2016 – March 2019.
Once first generation PDX models were established (F1) serial passaging to second and third generation (F2 and F3) was successful for all 3 models (Fig. 1A). Overall, tumor growth was observed between 10-15 weeks after implantation
Model Sample Site Stage Treatment* Status at last follow up Histology patient tumor Tumor take rate (F1) Tumor latency time (F1)
TC1 Surgery Primary IV (metastatic disease)
Naive Complete response Mixed GCT: EC, YS, TER, seminoma
3/6 12 – 97 days TC3† Surgery Primary Premalignant
lesion
N/A N/A GCNIS N/A N/A
TC4 Biopsy Peritoneal metastasis II (metastatic disease) CEB,TIP, TICE, carbo/pacli
Refractory disease YS 1/1 21 days
TC5 Surgery Primary IV (metastatic disease)
Naive Complete response Mixed GCT: YS, TER 5/6 15 – 29 days TC6 Surgery Primary II (metastatic
disease)
Naive Complete response Mixed GCT: Seminoma, EC,YS,TER
0/6 N/A
TC7 Surgery Primary I (localized disease)
Naive Complete response Mixed GCT: EC, seminoma
0/6 N/A
TC8 Surgery Primary IV (metastatic disease)
VIP Complete response TER 0/6 N/A
TC10 Surgery Primary II (metastatic disease)
Naive Complete response Seminoma 0/2 N/A
GCT: germ cell tumor; EC: embryonal carcinoma; YS: yolk sac tumor; TER: teratoma; GCNIS: Germ cell neoplasia in situ. *Treatment that patient had undergone before tumor tissue
was collected for implantation. CEB: carboplatin, bleomycin, and etoposide containing chemotherapy; TIP paclitaxel, ifosfamide and cisplatin containing chemotherapy, VIP: ifosfamide, cisplatin and etoposide containing chemotherapy; TICE: high-dose chemotherapy with stem cell transplant; carbo/pacli: carboplatin and paclitaxel containing chemothera-py. † This tissue material was not included in take rate calculation since it was not a cancer lesion.
A, Tumor growth of freshly implanted tumor tissue of TC1 in F1, F2 and F3 generations, TC4 in F1 and F2 and TC5 in F1 and F3 generations. B, Tumor growth of F0 and F1 material from TC1, F2 from TC4 and F1 material from TC5, stored in FCS/DMSO before (re-)implantation. Each line represents the individual tumor volume (mm3) of several mice.
Biobanked 0 5 10 15 20 0 500 1000 1500 2000 F0 -> F1 Weeks F1 -> F2 Biobanked 0 5 10 15 0 500 1000 1500 2000 2500 Weeks F2 -> F3 TC4 0 5 10 15 0 1000 2000 3000 4000 F1 F2 Weeks Biobanked 0 2 4 6 8 0 500 1000 1500 2000 2500 Weeks F1 -> F2 TC5 0 5 10 15 0 500 1000 1500 2000 2500 F1 Weeks F3 A B TC1 0 5 10 15 0 500 1000 1500 2000 2500 F1 F2 Weeks F3 Tumor volume (mm 3) Tumor volume (mm 3) Tumor volume (mm 3) Tumor volume (mm 3) Tumor volume (mm 3) Tumor volume (mm 3)
for the 3 PDX models and their different passages. Biobanking was evaluated at different generations (Fig. 1B). Thawed F0 material was able to engraft into an F1 (1/2 mice) and thawed F1 material in F2 (5/5 mice). Tumor take rates were 25% with 1 out of 4 implanted tumor pieces showing growth for F1 and 70% (7 out of 10 tumor pieces) for F2 generation. Latency time of stored material for this model, 7 days for F1 and 25 days for F2, was comparable to freshly implanted material, 12 days for F1 and 35 days for F2 generation. Biobanked F2 tumor tissue from TC4 was re-implanted into F3 and tumor growth was observed in 7/7 mice, with 8/14 pieces showing growth, which indicates a take rate of 57%.Frozen F1 tumor tissue from PDX TC5 was re-implanted in F2. Tumor growth was observed in 5/5 mice, with 9/10 pieces showing growth, resulting in a take rate of 90%. Figure 1. Growth curves of three TC PDX models and biobanking possibilities.
PDX models partly retain immunohistochemical characteristics of the primary tumor
Histology of the primary patient tumor and 3 subsequent PDX generations was determined by H&E staining (Fig. 2). PDX models TC1 and TC5 originated from mixed germ cell tumors (Table 1). Histology of the patient tumor TC1 reported by pathology mentioned YS, EC, teratoma and seminoma components. However, histological examination of the implanted tumor pieces (F0) showed YS and immature teratoma components only, suggesting a sampling bias from the patient tumor. After passaging of the tumor to F1, the histological subtypes YS and immature teratoma remained, as well as in the F2 and F3 generation. TC4 originated from a pure YS and histology was retained over 3 subsequent passages. Patient
tumor TC5 consisted of YS and immature teratoma components, subtypes that were retained in subsequent passages F1, F2 and F3. From these 3 PDX models we conclude that passaging of tumors in mouse does not affect histological subtypes. Figure 2. Histopathological characteristics of three established TC PDX models.
HE stainings at 10X and 20X magnification of patient tissue and tumors belonging to F1 to F3 generations of PDX TC1, TC4 and TC5. 20x magnification fields highlight the teratoma and yolk sac components of TC1 and TC5 and the yolk sac histology of TC4. Mixed histology was conserved across all passages of TC1 and TC5 PDX models. Scale bars in the upper panels represent 200 μm, and scale bars in the smaller inserts represent 100 μm.
TER YS TER YS Patient F1 F2 F3 TER YS YS YS YS YS YS YS TER YS YS TER YS TER TER YS TC1 TC4 TC5
As a marker of proliferation, tumors were stained for Ki-67. All 3 PDX models showed high positivity for Ki-67, remaining stable over the 3 passages (Fig. 3A-C). Mouse specific infiltration in the PDX tumors was identified by staining against Cyclophilin A using a mouse specific antibody targeting a region with low homology to the human protein. Cyclophilin A is a housekeeping protein involved in protein folding and recently identified as a sensitive target for detection of the
murine microenvironment in human tumor xenografts18,19. The patient tumors
from TC1 and TC5 were negative for cyclophilin A. Mouse specific infiltration was observed in all three PDX generations (F1-F3) of TC1, TC4 and TC5 (Fig. 3A-C). Five tumors samples failed to engraft of which one was classified as a germ cell neoplasia in situ, a pre-malignant lesion. Other tumor samples consisted of either mixed tumor histology, pure teratoma or seminoma components (Table 1, Suppl. Fig. 1). Proliferation index of these samples was determined indicating that all samples contained proliferating (Ki-67+) cells (Suppl. Fig. 1).
Figure 3. Proliferation index and mouse specific tumor infiltration across several passages.
Images at 20X magnification of Ki-67 and Cyclophillin A (mouse specific) IHC stainings performed with patient and PDX tumors of TC1 (A), TC4 (B) and TC5 (C). Insertions in Ki-67 stainings state proliferation index of each tumor. Scale bars represent 100 μm.
Patient F1 F2 F3 Ki-67 F1 F2 F3 Ki-67 F1 F2 F3 Patient Ki-67 A B C 55% 18% 49% 37% 37% 32% 18% 29% 16% 39% 38% Cyclophillin A Cyclophillin A Cyclophillin A TC1 TC4 TC5
Genomic stability of TC PDX models
Genome wide copy number alterations (CNAs) were measured in different passages of PDX models, and on primary tumor if available (Fig. 4A, B). The primary tumor of PDX TC1 did not show many genome wide gains and losses, whereas considerable more gains and losses were observed in subsequent passages F1, F2 and F3, suggesting tumor cell enrichment and clonal selection bias. CNAs observed in the primary tumor were retained in the different generations of PDX TC1. Additional CNAs observed in the F1 generation, but not in the primary tumor, were retained in the subsequent F2 and F3 generation. Unfortunately, no DNA was available from the patient tumors of TC4 and TC5 to be able to assess changes in copy number alterations in F0 and F1 generation tumors. For both TC4 and TC5, the F1 PDX tumors showed a genome wide distribution of CNAs which were retained in the F2 and F3 generation (Fig. 4A). The correlation between the different PDX passages was determined based on thresholded CNAs of all SNP probes (n=691687). Heterogeneity was observed between the three different models. A strong correlation was observed between passages within each PDX model with pearson r > 0.68. Correlation of the primary tumor of TC1 with its subsequent PDX passages was moderate to strong with pearson r > 0.57. Hierarchical clustering revealed that passages within a PDX model cluster more closely together than unrelated samples (Fig. 4B).
Figure 4. Copy number alterations across different PDX passages.
A, CNA plots of PDX TC1, TC4 and TC5 representing the genome wide location of gains (blue) and losses (red) in patient tumor (F0), and/ or 3 subsequent PDX passages (F1, F2 and F3). The colored bar below each CNA plot indicates loss of heterozygosity (yellow) or allelic imbalance (purple). B, CNA concordance analysis between patient and/or different passages of PDX tumors by hierarchical clustering.
TC1.F0 TC1.F1 TC1.F3 TC1.F2 TC4.F3 TC4.F2 TC4.F1 TC5.F3 TC5.F2 TC5.F1 TC1.F0 TC1.F1 TC1.F2 TC1.F3 TC4.F3 TC4.F2 TC4.F1 TC5.F3 TC5.F1 TC5.F2 TC1.F0 TC1.F1 TC1.F3 TC1.F2 TC4.F3 TC4.F2 TC4.F1 TC5.F3 TC5.F2 TC5.F1 0 0.2 0.4 0.6 0.8 1 Pearson value A B
Transcriptional stability of TC PDX models
Whole-transcriptome analysis (RNA-seq) was performed on F1 and F3 generation PDX tumors of TC1 and TC5. Due to low quality RNA, only the F3 generation of TC4 was included. In total 5 tumors were analyzed by RNA-sequencing. Differential expression analysis was performed on paired PDX tumors of generation F1 and F3. A high concordance between paired PDX tumors was observed where 211 genes were significantly differentially expressed for TC1 and 201 genes were significantly differentially expressed for TC5. Gene ontology analysis on DEGs between paired tumors (F1-F3) from TC1 and TC5 showed enrichment of genes that are mainly involved in the extracellular region, including extracellular matrix organization, cell adhesion, secretion and immune response (Fig. 5A). This suggests that with increasing passaging in vivo, these tumors lose the human immune and stromal components.
To gain insight in differences between and also within tumors, correlation analysis on all DEGs was performed. Within TC1 and TC5, a moderate correlation between F1 and F3 was observed with spearman r>0.48. Weak to moderate correlation was observed between the F3 tumors of TC1, TC4 and TC5. Hierarchical clustering on DEGs of all samples revealed that F1 and F3 tumors from TC1 clustered together, but not F1 and F3 tumors from TC5. Largest distance between all samples was observed for PDX TC4 (Fig. 5B).
Figure 5. Differentially expressed genes (DEGs) in different PDX generations. A TC4.F3 TC5.F3 TC5.F1 TC1.F3 TC1.F1 TC4.F3 TC5.F3 TC5.F1 TC1.F3 TC1.F1 0 0.4 0.6 0.8 0.2 1 Spearman value 0 10 20 30
signaling receptor binding collagen bindingintegrin binding extracellular matrix structural constituent extracellular region collagen trimer extracellular region part collagen-containing extracellular matrix extracellular matrix biological adhesion cell adhesion anatomical structure morphogenesis extracellular structure organizationextracellular matrix organization
DEGs TC1.F1 vs. TC1.F3 Enrichment score (-log10(p-value)) 0 10 20 30 heparin binding transporter activity peptide antigen binding identical protein binding signaling receptor binding extracellular organelle extracellular vesicle extracellular region partextracellular space extracellular region transport secretion localization immune response defense response DEGs TC5.F1 vs. TC5.F3 Enrichment score (-log10(p-value)) B
Biological Process Cellular Compartment Molecular Function
ECM structural constituent conferring tensile strength
A, Gene Ontology (GO) analysis of significantly differentially expressed genes between the F1 and F3 generation PDX tumors of TC1 or TC5. B, Relative distances by hierarchical clustering based on significantly DEGs between all samples.
A TC4.F3 TC5.F3 TC5.F1 TC1.F3 TC1.F1 TC4.F3 TC5.F3 TC5.F1 TC1.F3 TC1.F1 0 0.4 0.6 0.8 0.2 1 Spearman value 0 10 20 30
signaling receptor binding collagen binding integrin binding extracellular matrix structural constituent extracellular region collagen trimer extracellular region part collagen-containing extracellular matrix extracellular matrix biological adhesion cell adhesion anatomical structure morphogenesis extracellular structure organizationextracellular matrix organization
DEGs TC1.F1 vs. TC1.F3 Enrichment score (-log10(p-value)) 0 10 20 30 heparin binding transporter activity peptide antigen binding identical protein binding signaling receptor binding extracellular organelle extracellular vesicle extracellular region part extracellular space extracellular region transport secretion localization immune response defense response DEGs TC5.F1 vs. TC5.F3 Enrichment score (-log10(p-value)) B
Biological Process Cellular Compartment Molecular Function
ECM structural constituent conferring tensile strength
PDX models mimic response to conventional chemotherapy seen in patients
Cisplatin is the main chemotherapeutic agent in the treatment of testicular cancer. We assessed whether cisplatin sensitivity of the established PDX models corresponded to the patients’ response in the clinic. TC1 and TC5 were obtained from patients that had a complete response with cisplatin-based chemotherapy, while TC4 was obtained from a patient refractory to cisplatin treatment. In both TC1 and TC5, the lower dose of cisplatin (1 mg/kg) gave a significant tumor growth delay (Figure 6A), and small differences (not significant) were observed in tumor weight. High dose cisplatin (4 mg/kg) completely abolished tumor growth in TC1 and TC5 (Figure 6A). Tumor weight at the end of treatment was significantly lower in the high dose cisplatin group compared to vehicle (Fig. 6A). In the high dose cisplatin group (4 mg/kg) of TC4, tumor growth was slightly delayed over the course of treatment compared to vehicle controls (Fig. 6A). A small difference (not significant) was observed in tumor weight (Fig. 6A). IHC staining of these tumors showed a decrease in Ki-67 positive cells with increasing cisplatin concentrations in the cisplatin sensitive PDX models (TC1 and TC5),
while no difference was observed for the cisplatin resistant model TC4. An increase in cleaved caspase-3 staining, a well-known marker of apoptosis, after cisplatin treatment was only observed in TC1 (Fig. 6B). Mouse body weight was monitored during the course of treatment to measure treatment related toxicity. Only mice in the high dose cisplatin group lost weight (Suppl. Fig. 2).
Figure 6. Cisplatin sensitivity of three TC PDX models.
A, Tumor growth, and final tumor weight of PDX models TC1, TC4 and TC5 treated with vehicle or cisplatin. Data shows average ± SD. B, Representative images from tumors shown in (A) at 10X magnification and quantification of Ki-67 and cleaved caspase-3. Scale bars represent 200 μm. 0 3 5 7 10 12 14 17 19 21 0 500 1000 1500 2000 TC1 Days of treatment 0 3 5 7 10 12 13 17 19 21 0 500 1000 1500 2000 TC5 Days of treatment Vehicle Cisplatin (1 mg/kg) Cisplatin (4 mg/kg) A B *** ** *** 0.0 0.5 1.0 1.5 2.0 0.0 0.2 0.4 0.6 *** 0 3 5 7 10 12 13 17 19 21 0 500 1000 1500 2000 2500 3000 TC4 Days of treatment * * Ki-67 1 mg/kg 4 mg/kg 0.0 0.2 0.4 0.6 0.8 1.0 Tumor weight (gr) * Vehicle Cisplatin (1 mg/kg) Cisplatin (4 mg/kg) Tumor weight (gr) Vehicle Cisplatin (1 mg/kg) Cisplatin (4 mg/kg) Vehicle Cisplatin (4 mg/kg) Tumor weight (gr) Vehicle Cl. casp-3 0 0.005 0.010 0.015 0.020 0.025 Cl. Casp-3
Positive pixel count / µm
2 Vehicle 4 mg/kg Ki-67 Cl. casp-3 1 mg/kg 4 mg/kg Vehicle Ki-67 Cl. casp-3 0 20 40 60 80 100 Ki-67 Ki-67 index (%) 0 0.05 0.10 0.15 0.20 Cl. Casp-3
Positive pixel count / µm
2 0 20 40 60 80 100 Ki-67 0 0.005 0.010 0.015 0.020 Cl. Casp-3 Vehicle Cisplatin 1 mg/kg Cisplatin 4 mg/kg
Positive pixel count / µm
2 Ki-67 index (%) Tumor volume (mm 3) Tumor volume (mm 3) Tumor volume (mm 3) * *** ns ns ns * ns ns Sensitive Resistant ns * 0 20 40 60 80 100 Ki-67 Ki-67 index (%) ns
TC PDX models are sensitive to mTORC1/2 and MDM2 inhibition
Preclinical efficacy of two novel targeted agents for TC was assessed in three p53 wild-type PDX models. Blocking the interaction between MDM2 and p53 is therapeutically interesting as it leads to re-activation of p53 and induces a p53 dependent apoptotic response. Nutlin-3a, a small molecule inhibitor of MDM2, has shown anti-tumor activity in vitro and sensitized TC cells to cisplatin
treatment20,21. Efficacy of an alternative and clinically interesting MDM2 inhibitor
RG7388 was tested in two different doses in TC1. A small effect on tumor growth was observed at the end of treatment in the group receiving the lower dose of RG7388 (50 mg/kg) (Fig. 7A). A higher dose of RG7388 (75 mg/kg) resulted in considerable smaller tumor volumes compared to the vehicle treated group. No significant differences in tumor weight were observed (Fig. 7A). IHC analysis of the tumors showed a trend towards a lower Ki-67 in both RG7388 treatment arms and an induction of cleaved caspase 3 with the high dose of RG7388 (75 mg/kg) (Fig. 7B). In TC4 only the highest dose of RG7388 (75 mg/kg) was tested. No effect was observed on tumor growth and tumor weight (Fig. 7C). Ki-67 and cleaved caspase-3 IHC stainings showed no difference between control and RG7388 treated tumors (Fig. 7D).
Another interesting pre-clinical target for TC is mTOR, as the PI3K/AKT/
mTOR pathway was shown to be highly active TC models22–25. The mTORC1/2
inhibitor AZD8055 was tested in 2 PDX models and the effect on tumor growth was assessed. Two dosages of AZD8055 were assessed in TC5, where the lower dose (2.7 mg/kg) had no effect on tumor growth and tumor weight (Fig. 7E). In this PDX model a delayed tumor growth, reflected by lower tumor weight, was observed in mice treated with the highest dose of AZD8055 (10 mg/kg) compared to the vehicle group, whereas no effect of the high dose was observed in the cisplatin resistant PDX model TC4 (Fig. 7G). IHC stainings of TC4 and TC5 showed that increasing concentrations of the mTORC1/2 inhibitor did not induce any changes in Ki-67 positive cells or cleaved caspase-3 positivity (Fig. 7F). Mouse weights, monitored during the 3 week treatment period, were not affected by AZD8055 or RG7388 treatment (Suppl. Fig. 4).
A and C, Tumor growth, and final tumor weight of PDX TC1 and TC4 treated with vehicle or RG7388. Data shows average ± SD.B and D, Representative images of tumors shown in (A) and (C) respectively at 10X magnification, scale bars represents 200 μm, and quantification of Ki-67 and cleaved caspase-3. Data shows average ± SEM. E and G, Tumor growth, and final tumor weight of PDX TC5 and TC4 treated with vehicle or AZD8055. Data shows average ± SD. F and H, Representative images of tumors shown in (E) and (G) respectively at 10X magnification, scale bars represents 200 μm, and quantification of Ki-67 and cleaved caspase-3.
0 2 4 7 9 11 14 16 18 21 0 500 1000 1500 2000 2500 Days of treatment Tu m or v ol um e (m m 3 ) VehicleRG7388 (50 mg/kg) RG7388 (75 mg/kg) 0 3 5 7 10 12 13 17 19 21 0 500 1000 1500 2000 2500 Days of treatment Tu m or v ol um e (m m 3 ) VehicleAZD8055 (2.7 mg/kg) AZD8055 (10 mg/kg) A B * *** * Vehicle 50 mg/kg 75 mg/kg Ki-67 Cl. casp-3 Vehicle 50 mg/kg 75 mg/kg 0 0.05 0.10 0.15 0.20 Cl. Casp-3 Vehicle RG7388 50 mg/kg RG7388 75 mg/kg
Positive pixel count /
µm 2 0 0.002 0.004 0.006 0.008 Cl. Casp-3 Vehicle AZD8055 2.7 mg/kg AZD8055 10 mg/kg ns ns ns * ns ns ns ns 0.0 0.5 1.0 1.5 Tu m or w ei gh t ( gr ) 0.0 0.2 0.4 0.6 0.8 1.0 Tu m or w ei gh t ( gr )
Positive pixel count / mm
2 Ki-67 Cl. casp-3 Tu m or w ei gh t ( gr ) Tu m or w ei gh t ( gr ) C E F G H TC4 TC1 TC4 TC5 0.00 0.02 0.04 0.06 Cl. Casp-3 0.00 0.05 0.10 0.15 0.20 Cl. Casp-3 0 20 40 60 80 100 Ki-67 Ki -6 7 in de x (% ) Vehicle AZD8055 10 mg/kg 0 20 40 60 80 100 Ki-67 Ki -6 7 in de x (% ) Vehicle RG7388 75 mg/kg
Positive pixel count / mm
2
Positive pixel count / mm
2 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 0 2 4 7 9 11 14 16 18 21 0 500 1000 1500 Days of treatment Tu m or v ol um e (m m 3 ) Vehicle AZD8055 (10 mg/kg) 0 2 4 7 9 11 14 16 18 21 0 500 1000 1500 2000 2500 Days of treatment Tu m or v ol um e (m m 3 ) Vehicle RG7388 (75 mg/kg) 0 20 40 60 80 100 Ki-67 Ki -6 7 in de x (% ) 0 20 40 60 80 100 Ki-67 K i-6 7 in de x (% ) ns ns ns ns ns ns ns ns Vehicle 75 mg/kg Ki-67 Cl. casp-3 Vehicle 75 mg/kg Ki-67 Cl. casp-3 D 0 2 4 7 9 11 14 16 18 21 0 500 1000 1500 2000 2500 Days of treatment Tu m or v ol um e (m m 3 ) VehicleRG7388 (50 mg/kg) RG7388 (75 mg/kg) 0 3 5 7 10 12 13 17 19 21 0 500 1000 1500 2000 2500 Days of treatment Tu m or v ol um e (m m 3 ) VehicleAZD8055 (2.7 mg/kg) AZD8055 (10 mg/kg) A B * *** * Vehicle 50 mg/kg 75 mg/kg Ki-67 Cl. casp-3 Vehicle 50 mg/kg 75 mg/kg 0 0.05 0.10 0.15 0.20 Cl. Casp-3 Vehicle RG7388 50 mg/kg RG7388 75 mg/kg
Positive pixel count /
µm 2 0 0.002 0.004 0.006 0.008 Cl. Casp-3 Vehicle AZD8055 2.7 mg/kg AZD8055 10 mg/kg ns ns ns * ns ns ns ns 0.0 0.5 1.0 1.5 Tu m or w ei gh t ( gr ) 0.0 0.2 0.4 0.6 0.8 1.0 Tu m or w ei gh t ( gr )
Positive pixel count / mm
2 Ki-67 Cl. casp-3 Tu m or w ei gh t ( gr ) Tu m or w ei gh t ( gr ) C E F G H TC4 TC1 TC4 TC5 0.00 0.02 0.04 0.06 Cl. Casp-3 0.00 0.05 0.10 0.15 0.20 Cl. Casp-3 0 20 40 60 80 100 Ki-67 Ki -6 7 in de x (% ) Vehicle AZD8055 10 mg/kg 0 20 40 60 80 100 Ki-67 Ki -6 7 in de x (% ) Vehicle RG7388 75 mg/kg
Positive pixel count / mm
2
Positive pixel count / mm
2 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 0 2 4 7 9 11 14 16 18 21 0 500 1000 1500 Days of treatment Tu m or v ol um e (m m 3 ) Vehicle AZD8055 (10 mg/kg) 0 2 4 7 9 11 14 16 18 21 0 500 1000 1500 2000 2500 Days of treatment Tu m or v ol um e (m m 3 ) Vehicle RG7388 (75 mg/kg) 0 20 40 60 80 100 Ki-67 Ki -6 7 in de x (% ) 0 20 40 60 80 100 Ki-67 K i-6 7 in de x (% ) ns ns ns ns ns ns ns ns Vehicle 75 mg/kg Ki-67 Cl. casp-3 Vehicle 75 mg/kg Ki-67 Cl. casp-3 D
Discussion
In the present study, we describe the development of subcutaneous TC PDX models. Efficient tissue biobanking of both patient and PDX TC tumor material that can be used for future experiments was demonstrated. In addition, three TC PDX models were characterized in depth indicating that PDX tumors retain important tumor characteristics like (mixed) tumor histology and sensitivity to conventional chemotherapy. Sensitivity to RG7388 and AZD8055 was tested in TC PDXs highlighting the potential use of PDX models to test novel treatment strategies targeting MDM2 or mTORC1/2 in TC.
TC cell line and xenograft models are scarce. Public data shows there are only 30
established TC PDX models17,26, 18 developed orthotopically and 12 subcutaneously.
The small number of TC PDX models compared to other tumor types can be partially explained by the rarity of this cancer type and the small number of research groups working on TC. It has been proposed that orthotopic implantation of TC not only
increases take rate but that the testis niche is necessary for tumor establishment13.
Orthotopic tumor implantation may be favorable, however the implantation is more laborious and complex compared to subcutaneous implantation and monitoring of tumor growth requires imaging techniques and luciferase expressing tumors.
Reported establishment rate of TC PDX models implanted orthotopically was 35%26,
and 25% for tumors implanted subcutaneously17. Here, we show that engraftment
rate of subcutaneously implanted TC tumors from different patients was comparable, with 27%. Median latency time of establishment of TC PDXs has not been described before, but our study shows that TC PDX median latency time was shorter (25 days)
compared to several other tumor types, e.g. gastric cancer (94 days)27 and ovarian
cancer (45 days)9. Importantly, we evaluated storage possibilities by implanting
thawed material coming from patient tumor, F1 and F2 generations. Engraftment of the aforementioned tumor pieces was successful with no impairment on take rate or latency time. All three PDX models contained yolk sac component with a solid and papillary or solid and glandular architecture. Histological evaluation showed that two successfully established mixed tumors retained the non-seminoma components that were present in the implanted tumor pieces. Nonetheless, we cannot exclude the possibility that selection on specific histological subtypes takes place in the mice. IHC evaluation also indicated that mouse stroma takes over the human stroma, a
phenomenon already described in other tumor types28. Tumor adaptation was also
described using RNAseq analysis, that showed extracellular matrix organization as the most differencitally expressed genes when different passages were compared. A crucial feature that has brought much attention to PDX models compared to cell line xenografts is the larger heterogeneity of PDX models and therefore closer
resemblance to human tumors29. Even though clonal selection has been found to
occur after PDX establishment30,31, genomic landscape analysis showed higher
shows that PDX models retain this valuable tumor characteristic, heterogeneity. Therefore even with their limitations, PDXs are superior to cell line models.
In addition to histological evaluation, we also investigated the preservation of molecular features over different passages of PDX models. We show high genetic and transcriptional stability over time. On copy number variation level, passaging of PDX tumor led to an accumulation of CNAs that were undetected in the patient tumor, suggesting that positive selection of pre-existing clones has occurred. In addition, the normal tumor microenvironment of patient tumors is lost during passaging in mice. Our results clearly indicate that the tumor microenvironment is taken over by mouse specific infiltration as demonstrated with mouse-specific cyclophilin A staining. Furthermore, it has been described that the acquisition of CNAs during PDX passaging
differs from the acquisition of CNAs during tumor progression in patients33. To
maintain PDX tumors genetically resembling patient tumors, it is important to use low passage-PDX tumors as much as possible when studying for example drug-sensitivity. Clinical trials for TC are difficult to perform due to small number of patients and overall good response to cisplatin-based chemotherapy. Therefore, use of PDX models is highly valuable in rare diseases like TC. Here, we confirmed that all three TC PDX models had similar cisplatin sensitivity compared to the patient tumor they were derived from, adding to growing evidence that PDX models are superior at
predicting drug response in the clinic than cell line models32, 34, 35. Two targeted drugs,
MDM2 inhibitor RG7388 and mTORC1/2 inhibitor AZD8055, were tested using our TC PDX models. RG7388 was studied as a single agent in two PDX models, showing an effect on tumor growth in TC while AZD8055 induced an impaired tumor growth in TC5. Both RG7388 and AZD8055 treatment did not alter tumor growth in TC4. This tumor was derived from a patient refractory to cisplatin treatment that now appears to be a multi-drug resistant tumor as demonstrated by the lack of response to cisplatin but also to MDM2 or mTORC1/2 inhibitors.
Phase I clinical trials with the mTORC1/2 inhibitor TAK-228 showed good
tolerability of the inhibitor alone36 or in combination with other drugs37. Phase II
trials with TAK-228 are currently ongoing. Phase I/II trials with RG7388, known as idasanutlin, are ongoing in solid tumor patients, as are trials with several other MDM2 inhibitors, including AMG-232, ds3032b and ALRN-6924. Functional p53 protein is necessary for MDM2 inhibition efficacy and TP53 mutations are rarely
observed in TC patients38–40, making them eligible for this therapy.
In summary, we have established three PDX models of testicular cancer and characterized them on genomic, transcriptomic and histological level. These models showed accurate drug sensitivity prediction for cisplatin and can now be used for mechanistic studies, therapeutic development and pre-clinical validation of novel therapeutic strategies in testicular cancer.
Methods
Establishing of tumor xenografts
Written informed consent was obtained before surgery from all patients of which tumor samples were used for PDX modeling. Mice were kept under pathogen free conditions and received sterilized food and water ad libitum. All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Groningen (Groningen, the Netherlands) in accordance with the approved guideline “code of practice: animal experiments in cancer research” (Netherlands Inspectorate for Health Protection, Commodities and Veterinary Public Health, 1999). Subcutaneous TC PDX models were established as described previously9. Taking into account the heterogeneity of TC, sampling of the tumor was assisted by a pathologist who aimed at selecting different areas guided by macroscopic examination. Histology of each tumor piece was evaluated subsequently by an experienced oncological pathologist who determined TC components. In short, primary tumor or biopsy material
was cut into ~3x3x3 mm3 pieces and implanted on both flanks of 4-12 week old
male NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice (internal breed, Central Animal Facility, University Medical Centre Groningen). When sufficient material available, the tumor was also biobanked in liquid nitrogen using freeze media: fetal calf serum (FCS) with 5% dimethyl sulfoxyde (DMSO), paraffin embedded and snap frozen. Tumor growth was monitored by caliper measurements once a week and tumor volume was calculated using the following formula (width2 x
length)/2. Once tumor volume reached >1500 – 2500 mm3, mice were sacrificed
and tumors harvested. These tumors were used for immediate re-implantation into the next generation, as well as biobanked in liquid nitrogen using freeze media (FCS/5% DMSO), paraffin embedded and snap frozen.
Immunohistochemistry
Formalin-fixed and paraffin embedded material was cut into 4 µm sections and mounted on glass slides. Hematoxylin and eosin (H&E) stainings were used to look at tumor histology. Immunohistochemical (IHC) stainings were done for Ki-67, cyclophillin A, and cleaved caspase-3. Tissue slides were deparaffinized in xylene and rehydrated in ethanol. Antigen retrieval was done for 15 minutes as listed in Supplementary Table 1. Endogenous peroxidase was blocked
for 30 minutes with 0.3% H2O2. Tissue slides were then incubated with the
primary antibodies diluted in phosphate buffered saline (PBS) with 1% bovine serum albumin (BSA) for 1 hour at room temperature or at 4°C overnight (Suppl. Table 1). Slides were stained with HRP labeled secondary antibodies (DAKO, Germany), staining was visualized by DAB and counterstained with hematoxylin.
Stained sections were scanned using the NanoZoomer 2.0-HT multi slide scanner (Hamamatsu, Japan). Automated scoring of scanned images was done with QuPath, an open source, digital image analysis software41. Scoring of Ki-67 and cleaved caspase-3 were done in different ways. For Ki-67 scoring, QuPath software was trained using an automated cell detection classifier to separate tumor from stromal background populations. The simple tissue detection command was applied to detect the total tissue area of each image. Then the cell detection command, followed by the automated cell detection classifier was used, automatically calculating the percentage of Ki-67-positive tumor cells. For cleaved caspase-3, the first step was using the simple tissue detection command. Next, the positive pixel count command was applied, giving the positive pixel
count. Data are represented as positive pixel count / area µm2.
DNA and RNA isolation of PDX tumors
DNA and RNA from PDX tumors of different generations were isolated from snap frozen tumor tissue. Frozen section slides were used for quantifying the amount of vital tumor cells. Cryostat sections of 10 um were cut (~25/tumor) for DNA and RNA isolation, with additional 4 µm sections for H&E staining. Simultaneous isolation of DNA and RNA from the same sample was done using the AllPrep DNA/RNA mini kit (Qiagen, Germany).
Single nucleotide polymorphism genotyping
The Infinium Global Screening Array-24 v1.0 BeadChip from Illumina containing ~700.000 SNPs to determine copy number variation was used for genome-wide SNP genotyping of the different generations PDX tumors of TC1, TC4 and TC5. DNA processing, tagging and hybridization to the CHIP were performed according to the manufacturer's protocol (Illumina). Primary assessment and SNP call rate quality control of SNP intensity output files were performed using GenomeStudio software. Samples passed the inclusion quality control criteria, including call rates > 95%. Further analysis was performed with Nexus Copy number software (BioDiscovery) to generate copy number alteration (CNA) profiles. Quantitative CNA correlative analysis of the different PDX passages
were performed as described previously9.
RNA sequencing
The NEBNext Ultra Directional RNA Library Prep Kit (New England Biolabs, NEB), was used to process the samples. Ribosomal RNA was depleted from total RNA using the rRNA depletion kit (NEB). After fragmentation of the RNA, cDNA synthesis was performed. cDNA was ligated with the sequencing adapters followed by a PCR amplification. The quality and yield after sample preparation
were measured with the Fragment Analyzer. Clustering and DNA sequencing using the NovaSeq 6000 was performed according to manufacturer's protocols. A concentration of 1.5 nM of DNA was used. Sequencing was performed by GenomeScan (Leiden, the Netherlands).
Primary data analysis and quality score calculations were performed with RTA3 and Bcl2fastq v2.20 (Illumina). Prior to alignment, the reads were trimmed for
adapter sequences using Trimmomatic v0.3046. Reads were mapped to both the
mouse (GRCm38.p4) and human (GRCh37.75) reference to segregate mouse and human reads using BBMap (v36.62; http://sourceforge.net/projects/bbmap/) Alignment of reads was done by mapping to the human reference sequence
(GRCh37.75) using Tophat v2.0.1447 with default settings and were quantified
on gene level. For differential expression analysis, read counts were loaded
into the DESeq2 (v1.14.1)48 package within the R platform (v3.3.0). Average
hierarchical clustering of differentially expressed genes and visualization was
performed within R (v3.4.1) using gplots package49. Clusters were defined
visually setting the cut off at 0.7. In order to analyze the differentially expressed genes (DEGs) at the functional level, we used the PANTHER classification system to obtain the enriched biological processes (BPs), molecular function (MF) and
cellular compartment (CC)50. Correlative analysis on DEGs was performed
using complete hierarchical clustering of Spearman correlation metrics. All computations and heatmap generation were performed using R (v3.4.1).
Efficacy studies of standard of care and novel therapeutics
Mice were implanted with PDX tumors as described above. When tumors demonstrated sustained growth, mice were randomized into vehicle control or treatment groups (n=3-4 mice/group). Cisplatin (1 mg/kg or 4 mg/kg) or vehicle (saline) was administered weekly via intraperitoneal injection. AZD8055 (2.7 mg/kg or 10 mg/kg in 10% DMSO, 40% Polyethylene glycol 300) or vehicle was administered daily via intraperitoneal injection. RG7388 (50 mg/kg or 75 mg/kg in 2% hydroxypropylcellulose, 0.2% Tween-80) or vehicle was administered daily via oral gavage. Tumor growth was assessed 3 times a week using caliper measurements. All mice were sacrificed after 21 days of treatment. For future analysis the tumors were resected and paraffin embedded. Tumor volumes are plotted as mean ± SEM for all groups. Two-way ANOVA with Dunnett post hoc test was used to determine the significance of all pair-wise comparisons using GraphPad Prism (US).
Acknowledgements
The authors would like to thank Joost J. Caumanns, Shang Li and Ilse Koole for help with the PDX models, and Phuong Le for help with tissue processing. Steven de Jong is a member of the EuroPDX consortium.
Author contributions
M.A.T.M.V., J.A.G. and S.J. conceived and supervised the project. G.V. and X.R. performed the majority of experiments and data analysis with the assistance of G.J.M., V.C.L. and A.J.H.S.; G.V., X.R., M.A.T.M.V., J.A.G., and S.J. wrote the manuscript. All authors discussed the results and commented on the manuscript.
Conflict of interest
References
1. Trabert, B., Chen, J., Devesa, S. S., Bray, F. & Mcglynn, K. A. International patterns and trends in testicular cancer incidence, overall and by histologic subtype, 1973-2007. Andrology 3, 4–12 (2015).
2. Oosterhuis, J. W. & Looijenga, L. H. J. Testicular germ-cell tumours in a broader perspective. Nat Rev Cancer 5, 210–222 (2005).
3. Houldsworth, J., Korkola, J. E., Bosl, G. J. & Chaganti, R. S. K. Biology and Genetics of Adult Male Germ Cell Tumors. J Clin Oncol 24, 5512–5518 (2006).
4. Mead, G. M. & Stenning, S. P. The international germ cell consensus classification: A new prognostic factor-based staging classification for metastatic germ cell tumours. Clin Oncol 9, 207–209 (1997).
5. Rahman, N. A. & Huhtaniemi, I. T. Testicular cell lines. Mol Cell Endocrinol 228, 53–65 (2004).
6. Andrews, P. W., Bronson, D. L., Benham, F., Strickland, S. & Knowles, B. B. A comparative study of eight cell lines derived from human testicular teratocarcinoma. Int J Cancer 26, 269–280 (1980).
7. Bertotti, A. et al. A Molecularly Annotated Platform of Patient-Derived Xenografts ("Xenopatients") Identifies HER2 as an Effective Therapeutic Target in Cetuximab-Resistant Colorectal Cancer. Cancer Discov 1, 508–523 (2011). 8. Sivanand, S. et al. A Validated Tumorgraft Model Reveals Activity of Dovitinib
Against Renal Cell Carcinoma. Sci Transl Med 4, 137ra75-137ra75 (2012). 9. Alkema, N. G. et al. Biobanking of patient and patient-derived xenograft ovarian
tumour tissue: Efficient preservation with low and high fetal calf serum based methods. Sci Rep (2015). doi:10.1038/srep14495
10. Liu, J. F. et al. Establishment of Patient-Derived Tumor Xenograft Models of Epithelial Ovarian Cancer for Preclinical Evaluation of Novel Therapeutics. Clin Cancer Res 23, 1263–1273 (2017).
11. Kageyama, K. et al. Establishment of an orthotopic patient-derived xenograft mouse model using uveal melanoma hepatic metastasis. J Transl Med (2017). doi:10.1186/s12967-017-1247-z
12. Zhang, C., Awasthi, N., Schwarz, M. A. & Schwarz, R. E. Establishing a peritoneal dissemination xenograft mouse model for survival outcome assessment of experimental gastric cancer. J Surg Res 182, 227–234 (2013).
13. Hidalgo, M. et al. Patient-derived Xenograft models: An emerging platform for translational cancer research. Cancer Discov (2014). doi:10.1158/2159-8290. CD-14-0001
14. Castillo-Avila, W. et al. Sunitinib Inhibits Tumor Growth and Synergizes with Cisplatin in Orthotopic Models of Cisplatin-Sensitive and Cisplatin-Resistant Human Testicular Germ Cell Tumors. Clin Cancer Res 15, 3384–3395 (2009). 15. Juliachs, M. et al. ErbBs inhibition by lapatinib blocks tumor growth in an
orthotopic model of human testicular germ cell tumor. Int J Cancer 133, 235–246 (2013).
16. Juliachs, M. et al. Effectivity of pazopanib treatment in orthotopic models of human testicular germ cell tumors. BMC Cancer 13, 382 (2013).
17. Burger, A. M. & Fiebig, H.-H. Screening using animal systems. in Anticancer Drug Development 285–299 (2002).
18. Wang, P. & Heitman, J. The cyclophilins. Genome Biol 6, 226 (2005).
19. Hengstschläger, M., Rosner, M. & Hengstschläger, M. An antibody-based approach for the in situ evaluation of mouse contribution in human stem cell-derived xenografts. Protoc Exch (2018). doi:10.1038/protex.2018.009 20. Koster, R., Timmer-Bosscha, H., Bischoff, R., Gietema, J. A. & de Jong, S. Disruption
of the MDM2–p53 interaction strongly potentiates p53-dependent apoptosis in cisplatin-resistant human testicular carcinoma cells via the Fas/FasL pathway. Cell Death Dis 2, e148–e148 (2011).
21. Bauer, S. et al. Therapeutic Potential of Mdm2 Inhibition in Malignant Germ Cell Tumours. Eur Urol 57, 679–687 (2010).
22. Selfe, J. et al. IGF1R signalling in testicular germ cell tumour cells impacts on cell survival and acquired cisplatin resistance. J Pathol 244, 242–253 (2018). 23. Juliachs, M. et al. Effectivity of pazopanib treatment in orthotopic models of
human testicular germ cell tumors. BMC Cancer 13, 382 (2013).
24. Juliachs, M. et al. The PDGFRß–AKT Pathway Contributes to CDDP-Acquired Resistance in Testicular Germ Cell Tumors. Clin Cancer Res 20, 658–667 (2014). 25. Chen, K. S. et al. EGF Receptor and mTORC1 Are Novel Therapeutic Targets in
Nonseminomatous Germ Cell Tumors. Mol Cancer Ther 17, 1079–1089 (2018). 26. Byrne, A. T. et al. Interrogating open issues in cancer precision medicine with
patient-derived xenografts. Nat Rev Cancer 17, 254–268 (2017).
27. Choi, Y. Y. et al. Establishment and characterisation of patient-derived xenografts as paraclinical models for gastric cancer. Sci Rep 6, 22172 (2016).
28. Alkema, N. G. et al. Biobanking of patient and patient-derived xenograft ovarian tumour tissue: efficient preservation with low and high fetal calf serum based methods. Sci Rep 5, 14495 (2015).
29. Cassidy, J. W., Caldas, C. & Bruna, A. Maintaining Tumor Heterogeneity in Patient-Derived Tumor Xenografts. Cancer Res 75, 2963 LP – 2968 (2015).
30. Eirew, P. et al. Dynamics of genomic clones in breast cancer patient xenografts at single-cell resolution. Nature 518, 422–426 (2015).
31. Ben-David, U. et al. Patient-derived xenografts undergo mouse-specific tumor evolution. Nat Genet 49, 1567–1575 (2017).
32. Gao, H. et al. High-throughput screening using patient-derived tumor xenografts to predict clinical trial drug response. Nat Med 21, 1318–1325 (2015).
33. Ben-David, U. et al. Patient-derived xenografts undergo mouse-specific tumor evolution. Nat Genet 49, 1567–1575 (2017).
34. Pompili, L., Porru, M., Caruso, C., Biroccio, A. & Leonetti, C. Patient-derived xenografts: a relevant preclinical model for drug development. J Exp Clin Cancer Res 35, 189 (2016). 35. Ricci, F. et al. Patient-Derived Ovarian Tumor Xenografts Recapitulate Human
36. Moore, K. N. et al. Phase I study of the investigational oral mTORC1/2 inhibitor sapanisertib (TAK-228): tolerability and food effects of a milled formulation in patients with advanced solid tumours. ESMO Open 3, e000291 (2018). 37. Burris, H. A. et al. TAK-228 (formerly MLN0128), an investigational dual
TORC1/2 inhibitor plus paclitaxel, with/without trastuzumab, in patients with advanced solid malignancies. Cancer Chemother Pharmacol 80, 261–273 (2017).
38. Feldman, D. R. et al. Presence of Somatic Mutations within PIK3CA, AKT, RAS, and FGFR3 but not BRAF in Cisplatin-Resistant Germ Cell Tumors. Clin Cancer Res 20, 3712–3720 (2014).
39. Litchfield, K. et al. Whole-exome sequencing reveals the mutational spectrum of testicular germ cell tumours. Nat Commun 6, 5973 (2015).
40. Bagrodia, A. et al. Genetic Determinants of Cisplatin Resistance in Patients With Advanced Germ Cell Tumors. J Clin Oncol 34, 4000–4007 (2016). 41. Bankhead, P. et al. QuPath: Open source software for digital pathology image
analysis. Sci Rep 7, 1–7 (2017).
42. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014).
43. McKenna, A. et al. The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res (2010). doi:10.1101/gr.107524.110
44. Garrison, E. & Marth, G. Haplotype-based variant detection from short-read sequencing. (2012).
45. Ferronika, P. et al. Comprehensive Profiling of Primary and Metastatic ccRCC Reveals a High Homology of the Metastases to a Subregion of the Primary Tumour. Cancers (Basel) 11, 812 (2019).
46. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014).
47. Kim Daehwanand Pertea, G. and T. C. and P. H. and K. R. and S. S. L. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol 14, R36 (2013).
48. Love Michael I.and Huber, W. and A. S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550 (2014). 49. Gregory R. Warnes, Ben Bolker, Lodewijk Bonebakker, Robert Gentleman,
Wolfgang Huber Andy Liaw, Thomas Lumley, Martin Maechler, Arni Magnusson, Steffen Moeller, Marc Schwartz, B. V. gplots: Various R Programming Tools for Plotting Data. R Packag version 301
50. Mi, H. et al. PANTHER version 11: Expanded annotation data from Gene Ontology and Reactome pathways, and data analysis tool enhancements. Nucleic Acids Res 45, D183–D189 (2017).
Supplementary Figure 1. Histopathological characteristics of tumor that failed to engraft.
HE and Ki-67 stainings at 20X and 10X magnification of patient tissue that failed to engraft as PDX. Insertions in Ki-67 stainings state proliferation index of each tumor. Scale bars for the HE pictures represent 100 μm, and for Ki-67 pictures scale bars represent 200 μm.
HE Ki-67 TC6 TC7 TC8 TC10 TC3 11.3% 28.3% 37.6% 16.3% 61.6%
Supplementary Figure 2. Effect of cisplatin treatment on mouse weight over time. TC1 0 5 10 15 20 0 10 20 30 40 Days TC4 0 5 10 15 20 0 10 20 30 40 Days TC5 0 5 10 15 20 0 10 20 30 40 Vehicle Cisplatin (1 mg/kg) Cisplatin (4 mg/kg) Days W eight (gr) W eight (gr) W eight (gr)
Mouse weight was measured 3 times a week while animals were in treatment. Animals were treated with cisplatin for 21 days according to the experimental set-up of Figure 6A.
Supplementary Figure 3. Effect of AZD8055 or RG7388 treatment on mouse weight over time.
Supplementary Table 1. Antibodies and antigen retrieval used for immunohistochemical stainings.
Mouse weight was measured 3 times a week while animals were in treatment. Animals were treated with AZD8055 (TC5 and TC4) or RG7388 (TC1 and TC4) for 21 days according to the experimental set-up of Figure 7A, C, E and G.
RAMhrp = Rabbit -anti-Mouse horseradish peroxidase, GARhrp = Goat-anti-Rabbit horseradish peroxidase, o/n = overnight. TC1 0 5 10 15 20 0 10 20 30 40 Vehicle RG7388 (50 mg/kg) Days RG7388 (75 mg/kg) TC5 0 5 10 15 20 0 10 20 30 40 Vehicle AZD8055 (2.7 mg/kg) AZD8055 (10 mg/kg) Days TC4 0 5 10 15 20 0 10 20 30 40 Days Vehicle AZD8055 (10 mg/kg) RG7388 (75 mg/kg) W eight (gr) W eight (gr) W eight (gr)
Ki-67 Tris/EDTA (pH 9.0) DAKO (M7240) 1:350 60 minutes 20 ºC RAMhrp - GARhrp P53 Tris/EDTA (pH 9.0) DAKO (M7001) 1:1000 60 minutes 20 ºC RAMhrp - GARhrp AFP Tris/EDTA (pH 9.0) Dako(A0008) 1:800 ? ?
Cyclophillin A EDTA (pH 8.0) Cell signaling
(51418) 1:500 60 minutes 20 ºC GARhrp - RAGhrp Cleaved caspase-3 EDTA (pH 8.0) Cell signaling
(9661) 1:100 60 minutes 20 ºC GARhrp - RAGhrp Abbreviations: RAMhrp = Rabbit -anti-Mouse horseradish peroxidase, GARhrp = Goat-anti-Rabbit horseradish peroxidase, o/n = overnight.
S upplementary table 1. Antibodies and antigen retrieval used for immunohistochemical stainings Antigen Antigen retrieval Company
