Characterization of genetically defined sporadic and hereditary type 1 papillary renal cell carcinoma cell lines

Abstract Renal cell carcinoma (RCC) is not a single disease but is made up of several different histologically defined subtypes that are associated with distinct genetic alterations which require subtype specific management and treatment. Papillary renal cell carcinoma (pRCC) is the second most common subtype after conventional/clear cell RCC (ccRCC), representing ~20% of cases, and is subcategorized into type 1 and type 2 pRCC. It is important for preclinical studies to have cell lines that accurately represent each specific RCC subtype. This study characterizes seven cell lines derived from both primary and metastatic sites of type 1 pRCC, including the first cell line derived from a hereditary papillary renal carcinoma (HPRC)‐associated tumor. Complete or partial gain of chromosome 7 was observed in all cell lines and other common gains of chromosomes 16, 17, or 20 were seen in several cell lines. Activating mutations of MET were present in three cell lines that all demonstrated increased MET phosphorylation in response to HGF and abrogation of MET phosphorylation in response to MET inhibitors. CDKN2A loss due to mutation or gene deletion, associated with poor outcomes in type 1 pRCC patients, was observed in all cell line models. Six cell lines formed tumor xenografts in athymic nude mice and thus provide in vivo models of type 1 pRCC. These type 1 pRCC cell lines provide a comprehensive representation of the genetic alterations associated with pRCC that will give insight into the biology of this disease and be ideal preclinical models for therapeutic studies.


| INTRODUCTION
Renal cell carcinoma (RCC) does not exist as a single disease but is made up of multiple histologically defined disease subtypes that are each associated with distinctive genetic and genomic alterations.
These differences mean that each subtype may require specific management and treatment. Papillary renal cell carcinoma (pRCC) is the second most common subtype of RCC after conventional or clear cell renal cell carcinoma (ccRCC), accounting for~20% of cases. Papillary RCC is further subcategorized into type 1 and type 2 pRCC. It is important for preclinical studies that cell line models which accurately represent each specific RCC subtype are produced and characterized.
RCC, also referred to as kidney cancer, can occur as either a sporadic or inherited disease, and studies of familial kidney cancer syndromes have elucidated important genetic features of RCC. Investigations of the inherited von Hippel-Lindau (VHL) clear cell RCC (ccRCC) predisposition syndrome led to the discovery of the VHL tumor suppressor gene and loss of function of this gene in VHL renal tumors, as well as in a significant percentage of tumors from patients with sporadic ccRCC (ranging from 59% to 92%). [1][2][3][4][5] The knowledge gained about the function of the VHL protein product in controlling oxygen sensing and the cells response to hypoxia and the effect of VHL loss on cellular biology has led to the development of targeted therapies. [6][7][8][9] Similarly, studies of the hereditary pRCC syndromes identified germline activating mutations within the MET oncogene in hereditary papillary renal carcinoma (HPRC), associated with type 1 pRCC, as well as germline inactivating mutations in the FH tumor suppressor gene in hereditary leiomyomatosis and renal cell carcinoma (HLRCC), often associated with an aggressive variant of type 2 pRCC. [10][11][12] Alongside these familial studies, analysis of large cohorts of sporadic pRCC, such as those performed by the Cancer Genome Atlas (TCGA), have provided further insights into the biology of these tumors that need to be reflected in cell line models. 13 Altered chromosomal copy number patterns represent the most common somatic changes in sporadic type 1 pRCC with gains of chromosome 7 (encoding the MET gene) and 17 being most frequent, followed by gain of chromosomes 12, 16, and 20. 13-15 In addition to gain, somatic mutation of the MET gene occurs in approximately 13% of sporadic type 1 pRCC. 10,16,17 TCGA analysis of sporadic type 1 pRCC identified additional features including mutations in chromatin remodeling genes (eg, BAP1) and that loss of CDKN2A (p16) associated with poorer patient outcome. 13,18 Cell line models provide an important tool for investigating the importance of these discoveries and a preclinical model for testing novel therapeutic agents that may target specific alterations within these tumors. The NCI has a long history of generating genetically defined cell line models from primary or metastatic kidney tumor tissue excised from patients with either sporadic or germline mutation associated RCCs. [19][20][21][22] In previous publications we have described two cell lines, (UOK262 and UOK268), derived from HLRCC-associated type 2 pRCC-like tumors, that have been used to demonstrate the potential preclinical effectiveness of both proteasome inhibitors and ABL1 inhibitors in targeting these cells adaption to oxidative stress. [23][24][25][26] Another recent publication highlighted a type 2 pRCC cell line derived from a sporadic tumor that demonstrated a NF2 mutation that dysregulated the HIPPO pathway and created a targetable therapeutic susceptibility to dasatinib, thereby inhibiting the downstream effects of HIPPO pathway dysregulation. 27 This study characterizes the genetic and genomic alterations in seven type 1 pRCC cell lines derived from either primary kidney tumors or metastatic material excised from patients, including the first cell line derived from a germline MET-mutated HPRC patient. These cell line models provide examples of CDKN2A (p16) loss, BAP1, and KRAS mutation, and type 1 pRCC-associated chromosomal copy number alterations that should enable the effective evaluation of precision therapies.

| Material acquisition
All patients from which procured materials were acquired were evalu-

| Cell line production protocol
All novel cell lines were established from either tumor tissue excised during surgery or from a cell pellet derived from ascites. Spontaneously immortal cell lines were generated using the protocols and techniques previously described. 22 In brief, the tumor tissue was diced into small pieces (~1-2 mm 3 ) and smeared across a 10 cm tissue culture dish or 75 cm 2 flask and placed under 5 to 8 mL of DMEM media in sterile conditions. Ascites fluid was centrifuged at 4 C at 1000g for 5 minutes to produce a cell pellet which was mixed with a small volume of media (5-8 mL) and placed in a tissue culture flask under sterile conditions. All cells were initially cultivated in DMEM medium containing 25 mM D-glucose with 2 mM L-glutamine (ThermoFisher Scientific Inc, Massachusetts) and supplemented with 10% fetal calf serum (Sigma-Aldrich, Missouri) and x1 Antibiotic-Antimycotic solution (ThermoFisher Scientific Inc). Cell lines were propagated for over 20 passages to demonstrate immortalization with a passage being performed every 2 to 5 days. Once established, the Antibiotic-Antimycotic solution was removed from the media to demonstrate that no infection was present within the cell line.

| Short tandem repeat analysis and mycoplasma testing
All cell lines were evaluated by short tandem repeat (STR) analysis (Genetica Cell Line Testing, North Carolina) to provide a distinct genetic fingerprint for each line (Table S1). When possible, this was compared with the STR analysis of any normal or tumor tissue derived DNA available from the originating patient. In addition, all cell lines were evaluated for mycoplasma contamination and shown to be negative (Genetica Cell Line Testing).

| Spectral karyotyping and FISH analysis
Metaphase spreads were prepared for each cell line and spreads were subsequently hybridized with custom chromosome-specific fluorescence spectral karyotyping (SKY) probes, imaged, and analyzed using the SkyView software (Applied Spectral Imaging, California, CA), as previously described. 28 The chromosome complements of 15 to 20 metaphase spreads were analyzed for each cell line, and karyotype descriptions were produced in accordance with the human chromosome nomenclature standards described in ISCN (2013). 29 A structural aberration or chromosomal gain was considered clonal if two or more metaphase spreads contained the same change, while chromosomal losses were considered clonal if three or more metaphase spreads demonstrated the same loss. 28,29 A minimum of 15 metaphase spreads were analyzed by fluorescence in situ hybridization (FISH) using probes for two oncogenes MET (7q31), EGFR (7p11.2), and a probe for the centromere of chromosome 7 (CEP 7). FISH image analysis was performed using the Q-FISH TM software (Leica).

| Nucleic acid extraction
Cell pellets were generated for each cell line from 80-90% confluent 10 cm dishes and DNAs were extracted using a Promega Maxwell

| DNA sequencing and mitochondrial (mt)DNA sequencing
Cell line DNA was sequenced using the OncoVar assay that is a targeted hybrid capture sequencing analysis which detects genomic variants in a panel of 240 cancer-related genes, including known pRCC associated genes, and was performed as previously published. 30 Validations were compared to a control DNA sample derived from the peripheral blood leukocytes of a control patient. The entire mitochondrial genome was sequenced and analyzed as previously described. 32 In brief, the entire mitochondrial genome was amplified from whole genomic DNA as overlapping PCR fragments using KAPA2G Fast Readymix

| Two site immunoassay analysis of MET protein and phosphoprotein
Analysis of the total and phospho-MET levels in Triton X-100 cell line extracts was performed by an electrochemiluminescent two site immunoassay using a SectorImager 2400 plate reader (Meso Scale Discovery, Gaithersburg, Maryland) as previously described. 33 The assay has attomole sensitivity for total MET; pMET was expressed as

| Clinical features of patients and cell line derivation
Seven independent spontaneously immortalized cell lines were derived from material procured from six individuals with sporadic type 1 pRCC and one member from a family with HPRC (Table 1). All seven patients had a histologic diagnosis of type 1 pRCC based on analysis of biopsies or surgical materials ( Figure S1). All cell lines were analyzed for their STR genetic fingerprint and matched to original patient normal tissue or tumor DNA where available (Table S1). UOK208, UOK274, UOK342, and UOK345 all matched their respective patient normal tissue STR profiles and UOK112, UOK332, and UOK345 matched their respective tumor DNA STR profiles (Table S1). UOK112 and UOK342 represent the only previously published cell lines from this cohort. 19,27 UOK112 was derived from a 67-year-old male patient who presented with a right-sided 10.5 cm renal mass and multiple metastatic lung nodules and underwent a radical nephrectomy to remove the kidney tumor that generated the cell line. UOK208 was made from a 64-year-old male patient who presented with a left-sided 8.0 cm renal mass and a metastatic neck mass that demonstrated type 1 pRCC and clear cell RCC features.
The renal mass was excised by radical nephrectomy, that also demonstrated positive local lymph nodes, and used to generate a cell line. UOK274 originated from primary tumor from a 53-year-old male patient who presented with a left-sided 13.0 cm renal mass and multiple metastatic lung nodules. UOK332, UOK337, and UOK342 were derived from ascites fluids obtained from three patients with metastatic type 1 pRCC with primary renal masses measuring 10.0, 8.0, and 3.2 cm, respectively. UOK345 was derived from a male patient who initially presented at age 41 and whose mother and maternal grandfather also had kidney cancer ( Figure 1A). Germline analysis of the HPRC-associated MET oncogene identified a known pathogenic missense mutation, p.H1112R, thus confirming his diagnosis of HPRC ( Figure 1B). 16 The patient presented with multifocal, bilateral kidney tumors (15 were surgically excised ranging from 0.7 to 3.0 cm) and metastasis to both local lymph nodes and the lung ( Figure 1C); he was managed for several years before the UOK345 cell line was generated from a sample of ascites fluid.

| Spectral karyotyping of type 1 pRCC cell lines
Spectral karyotyping demonstrated that five of the seven lines were near diploid exhibiting few copy number chromosomal gains or losses and few translocations ( Figures 1D, 2, and S2; Table 2). UOK342 is a hypo-triploid (>3n) cell line, and UOK337 is a hyper-triploid cell line with 10 different unbalanced chromosomal translocations, some duplicated, others only observed as one copy (Figures 1D, 2, and S2).

| Targeted mutation analysis of type 1 pRCC cell lines
Potential driver mutations within the cell lines were identified utilizing a targeted sequencing assay, OncoVar v3, which includes approximately 240 known cancer genes ( Table 2). In addition to the germline MET p.H1112R mutation present in UOK345, UOK208 was found to have a known pathogenic MET mutation, p.V1088A, and UOK337 demonstrated a novel missense mutation, p.H1106Q. All of these mutations occur in the MET tyrosine kinase domain. 17 In all three cell lines, confirmatory Sanger sequencing showed increased signal for the mutant allele over the wild-type indicating that the gained copies of chromosome 7 contained the mutation ( Figure 3A).  Figure S4). 18 UOK345 was the only cell line that demonstrated a pathogenic change within the mitochondrial genome with a homoplasmic missense mutation in ND5 (p.R436H) ( Figure S4).

| Focal CDKN2A deletion and expression analysis of CDKN2A, MET, and EGFR
The mutation of CDKN2A, the gene that encodes p16, in UOK112 is important as loss of CDKN2A has been shown to associate with more aggressive pRCC. However, CDKN2A is more frequently lost due to specific gene deletion or promoter methylation. 13,18 To evaluate the potential for focal gene deletion that would not be detectable by SKY analysis, Taqman-based gene copy number analysis was performed and demonstrated complete loss of CDKN2A in UOK208, UOK337, UOK342, and UOK345. UOK112 had heterozygous loss of CDKN2A and both UOK274 and UOK332 demonstrated a high degree of CDKN2A copy loss that was not complete, potentially suggesting some degree of heterogeneity within those cell lines ( Figure 4A, Table 2). Expression analysis revealed no CDKN2A mRNA expression in 6 out of 7 cell lines including UOK274 and UOK332. Expression was only present in the CDKN2A-mutant UOK112 cell line ( Figure 4A, Table 2).
Gain of a complete copy of chromosome 7 results in the gain of several potential oncogenes including MET at 7p31.2 and EGFR at 7p11.2 ( Figure 4B)  xenografts growing slowly and taking over 6 months to reach over 300 mm 2 , while UOK342 xenografts grew rapidly producing 2000 mm 2 tumors within~40 days ( Figure 6, Table 2). UOK208 and F I G U R E 6 Mouse xenografts of the UOK342 and UOK345 type 1 pRCC cell line models. Five NCI athymic NCr-nu/nu mice were subcutaneously injected in the flank with approximately 1 million cells of either UOK342 or UOK345 to evaluate the rate of xenograft tumor growth for these type 1 pRCC cell line models. Representative hematoxylin and eosin stained slides from xenografts for each cell line show typical type 1 pRCC histology UOK274 also produced slow growing xenografts, while UOK332 and UOK337 xenografts grew more rapidly ( Figure S6, Table 2). UOK208, UOK332, and UOK345 xenograft tumors demonstrated similar histopathologic patterns to those seen in the original tumors (Figures 6, S1, and S7). UOK274 and UOK337 showed evidence of papillary histology, but no images of the originating tumors were available for comparison ( Figure S7). The rapidly growing UOK342 xenografts demonstrated little evidence of papillary structure and looked like solid tumors ( Figure 6). Xenografts were analyzed for the presence of key mutations to confirm that they were derived from the appropriate cell line ( Figure S8).

| DISCUSSION
The MET mutations are associated with HPRC, a disease defined by bilateral, multifocal type 1 pRCCs. 10,13,18,35 18 Notably, two cell lines (UOK274 and UOK342) had known pathogenic activating mutation of the KRAS gene, p.G12D and p.
G12C respectively. Pathogenic mutation of KRAS was a rare event within the primary pRCC tumors investigated by the TCGA, present in only 2 out of 155 type 1 pRCC (1.3%) and 3 out of 67 type 2 pRCC (4.5%). 13 Investigation of RAS inhibitors is a very active area of therapeutic research and could provide a therapeutic option for a fraction of type 1 pRCC patients that could be evaluated in these model cell lines. 38 The presence of these less frequent events in our cell lines may reflect the aggressive nature of the tumors from which they were derived. The TCGA type 1 pRCC samples were largely lower stage tumors and these cell lines were all derived from patients with high stage, metastatic disease. In particular, loss of CDKN2A was shown by the TCGA to be associated with poor patient outcome and loss of CDKN2A was observed in all of our cell line models. 13 Understanding the role of CDKN2A loss in advanced disease by studying these cell line models could elucidate potential new therapies, such as targeting the cyclin dependent kinases 4 and 6 (CDK4 and CDK6) that are the inhibitor targets of the p16 protein encoded by CDKN2A.
Finally, both in vitro and in vivo analysis of cell line models is essential to provide convincing data for the translation of potential new therapies into patients. Six of these cell lines were capable of producing xenograft tumors when injected into the flank of athymic nude mice, providing models for the vast array of genetic alterations that occur within type 1 pRCC and representing both primary and metastatic disease.
It is important to note that there are several models for pRCC currently available that have been characterized to varying degrees, including UOK112 and UOK342 that are characterized in this study.
UOK112 was initially published over a decade ago with minimal genetic characterization and UOK342 was first published recently as an example of a pRCC cell line with a somatic NF2 mutation that responds to dasatinib treatment. 19

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article or available from the corresponding author upon request.