Registered report: Melanoma genome sequencing reveals frequent PREX2 mutations

The Reproducibility Project: Cancer Biology seeks to address growing concerns about reproducibility in scientific research by conducting replications of 50 papers in the field of cancer biology published between 2010 and 2012. This Registered Report describes the proposed replication plan of key experiments from ‘Melanoma genome sequencing reveals frequent PREX2 mutations’ by Berger and colleagues, published in Nature in 2012 (Berger et al., 2012). The key experiments that will be replicated are those reported in Figure 3B and Supplementary Figure S6. In these experiments, Berger and colleagues show that somatic PREX2 mutations identified through whole-genome sequencing of human melanoma can contribute to enhanced lethality of tumor xenografts in nude mice (Figure 3B, S6B, and S6C; Berger et al., 2012). The Reproducibility Project: Cancer Biology is a collaboration between the Center for Open Science and Science Exchange, and the results of the replications will be published by eLife. DOI: http://dx.doi.org/10.7554/eLife.04180.001


Introduction
Melanoma is a highly aggressive tumor with poor prognosis in the metastatic stage. Based on their association with UV-induced DNA damage, melanomas are often hypermutated and considerable efforts have been made to sequence such tumors in order to better understand their molecular basis. Many well-known oncogenes are frequently involved in melanoma pathogenesis, including BRAF and NRAS, and significant work has been done to develop targeted kinase inhibitors against the protein products of these genes (Kunz, 2014). However, even with treatment, melanoma has an extremely high rate of recurrence; thus, there is great interest in identifying novel candidate genes that promote oncogenesis in melanoma, thereby providing additional therapeutic targets.
One such candidate is Phosphatidylinositol-3,4,5-trisphosphate RAC Exchanger 2 (PREX2), a 183-kDa protein known to inhibit PTEN phosphatase activity, stimulate PI3K signaling, and suspected to regulate the small GTPase RAC1 (Fine et al., 2009;Cerami et al., 2012). Using wholegenome sequencing of 25 metastatic tumors, Berger and colleagues identified PREX2 as being a highly mutated gene in melanoma. Apart from observing a large subset of BRAF and NRAS mutations, the authors found PREX2 to have a mutation frequency of approximately 14%, with 13 detected non-synonymous point mutations, including four nonsense truncation mutations (Berger et al., 2012). In order to demonstrate the biological relevance of specific PREX2 mutations, the authors created transformed melanocyte cell lines that stably expressed various mutated and truncated forms of PREX2. By using these cell lines to create tumor xenografts in nude mice, the authors showed that ectopic expression of mutant PREX2 accelerated tumor formation.
Berger and colleagues chose to analyze six representative PREX2 mutations derived from their whole-genome sequencing screen. These variants included three truncation variants and three nonsynonymous point mutations predicted to carry functional impact. These mutant PREX2 constructs were packaged into lentiviruses and transduced into TERT-immortalized human melanocytes engineered to express NRAS G12D . Ectopic expression of various mutant PREX2 isoforms was confirmed by Western blot ( Figure 6A). These experiments will be replicated in Protocols 1 and 2. Berger and colleagues next transplanted the melanocytic lines into immunodeficient mice alongside control melanocytes expressing either wild-type PREX2 or GFP (green fluorescent protein). They found that overexpression of all three truncated variants, as well as the point mutation G844D, significantly accelerated tumor growth in vivo, thus affirming the biological relevance of their genomic data ( Figure 3B, S6B, and S6C). These key experiments, which support the hypothesis that mutant PREX2 promotes oncogenesis in melanoma, will be replicated in Protocol 3. There is some debate over which mutations observed in various melanoma samples are biologically relevant, including PREX2. Potentially, mutational heterogeneity across tumor samples may contribute to false-positive findings (Lawrence et al., 2013). Various genome-wide screens have yielded conflicting results about which genes are frequently mutated in melanoma. Recently, mutated PREX2 was identified in both the primary tumor and in metastatic tumor tissue from a genomic analysis of a single melanoma patient (Turajlic et al., 2012). However, five studies failed to identify PREX2 in their genome-wide melanoma screens, including a meta-analysis study that analyzed hundreds of published datasets Krauthammer et al., 2012;Ni et al., 2013;Marzese et al., 2014;Xia et al., 2014). To date, there have been no replication attempts assessing the biological significance of PREX2 mutant isoforms in melanoma.

Materials and methods
Unless otherwise noted, all protocol information was derived from the original paper, references from the original paper, or information obtained directly from the authors.

Protocol 1: generation of NRAS G12D melanocyte cells expressing various mutated forms of PREX2
This protocol describes the generation of pMEL/hTERT/CDK4(R24C)/p53DD/NRAS G12D (NRAS G12D ) melanocytes that stably express various mutated forms of PREX2. This protocol details the production of lentivirus for each mutated PREX2 isoform, as well as the viral transduction of melanocytes, and selection for stable-expressing lines using antibiotic resistance.

Sampling
• Outline of experimental endpoints: 1. At the end of this protocol, we will have generated NRAS G12D melanocytes overexpressing the following protein products:

Confirmatory analysis plan
• Statistical analysis of the Replication Data: 1. Not applicable.

Known differences from the original study
This replication is only generating stable melanocyte lines for GFP, wild-type PREX2, PREX2 Q1430*, and PREX2 G844D. The original study also included several other PREX2 mutants, including PREX2 K278*, E824*, P948S, and G106E. This replication will include the additional step of sequencing the endogenous PREX2 gene in the NRAS G12D melanocyte cell line to determine its mutational status. All known differences in reagents and supplies are listed in the materials and reagents section above, with the originally used item listed in the comments section. All differences have the same capabilities as the original and are not expected to alter the experimental design.

Provisions for quality control
The cell line used in this experiment will undergo STR profiling to confirm its identity and will be sent for mycoplasma testing to ensure there is no contamination. PREX2 expression constructs obtained from the original authors will be verified for sequence identity and DNA integrity. The endogenous mutational status of PREX2 in NRAS G12D melanocytes will be assessed. All data obtained from the experiment will be made publicly available, either in the published manuscript or as an open access dataset available on the Open Science Framework (https://osf.io/jvpnw/).

Protocol 2: confirming ectopic expression of PREX2 mutant isoforms by Western blot
This protocol investigates the expression levels of mutant PREX2 isoforms in virally transduced NRAS G12D melanocytes that were generated in Protocol 1. This protocol uses an anti-V5 antibody to recognize tagged forms of wild-type and mutant PREX2 (as well as the GFP control), thus verifying the successful lentiviral transduction of expression constructs and providing information about ectopic protein expression levels (as was demonstrated in Figure 6A). Membranes will also be probed with anti-α-tubulin to provide normalized values of relative protein expression. Three original cell lines produced by the original authors will also be included so that protein expression levels can be compared between the two studies.

Sampling
1. The original data presented is qualitative and this prevents power calculations being performed a priori to determine sample size (number of biological replicates). Instead, we will be including three cell lines originally derived by the authors and analyzing these cell lines in parallel to the newly derived cell lines from Protocol 1. 2. Three separate lysates will be prepared from each cell line: • GFP vector stable NRAS G12D cells (control) • Previously generated PREX2 Q1430* stable NRAS G12D cells (control from original study authors) • Previously generated PREX2 G844D stable NRAS G12D cells (control from original study authors) Registered report • Previously generated WT PREX2 stable NRAS G12D cells (control from original study authors) • PREX2 WT stable NRAS G12D cells (from Protocol 1) • PREX2 Q1430* stable NRAS G12D cells (from Protocol 1) • PREX2 G844D stable NRAS G12D cells (from Protocol 1) 1. Blots will be probed with the following antibodies:

Deliverables
• Data to be collected: 1. Images of probed membranes (full images with ladder) 2. Scanned images of Ponceau-stained membranes, post-transfer 3. Densitometric analyses of normalized bands, presented in a bar graph showing standard deviation across replicates for each cell line

Confirmatory analysis plan
• Statistical analysis of the Replication Data: 1. Means and standard deviations will be computed across replicates for each cell line. 2. We will perform a 2-way ANOVA (2 × 3 factorial analysis), comparing expression levels of the three PREX2 variants and the two cell-line cohorts (the originally-derived cell lines and the newly-derived cell lines from Protocol 1). This analysis will test two parameters: a) whether the original and replication values are different and b) if the three PREX2 variants are different. Because our hypothesis is that they are all the same, no individual follow-up tests are needed.

Known differences from the original study
This replication is only analyzing protein expression from cell lines engineered to express GFP, wildtype PREX2, PREX2 Q1430*, and PREX2 G844D. The original study also included several other PREX2 mutants, including PREX2 K278*, E824*, P948S, and G106E. This replication includes an antibody probing for PREX2, so that we can better determine its endogenous expression level. Additionally, we are also testing protein expression in the original PREX2 cells lines derived by the original authors, so that we can compare expression levels between the original lines and the replication lines. All known differences in reagents and supplies are listed in the materials and reagents section above, with the originally used item listed in the comments section. All differences have the same capabilities as the original and are not expected to alter the experimental design.

Provisions for quality control
The endogenous expression of PREX2 will be assessed in cell lines not overexpressing PREX2 variants. An image of Ponceau-stained membranes (post-transfer) will be included to verify successful protein transfer. All of the raw data, including the image files and quantified bands from the Western blot, will be uploaded to the project page on the OSF (https://osf.io/jvpnw/) and made publically available. This experiment is also the quality control for the other replication protocols as it assesses the levels of ectopic PREX2 variant expression in the utilized cell lines.

Protocol 3: generation of tumor xenografts expressing mutated forms of PREX2
This protocol assesses the propensity of ectopically expressed PREX2 mutations to accelerate tumor formation of immortalized human melanocytes in vivo. This protocol utilizes stably transfected NRAS G12D human melanocyte lines that were previously generated and analyzed in Protocols 1 and 2. The melanocytic lines are transplanted into immunodeficient mice alongside control melanocytes expressing wild-type PREX2 or GFP (green fluorescent protein). Tumor growth is assessed for 16 weeks, and tumor-free survival is monitored, as depicted in Figure 3B, S6B. Further, confirmatory staining and analysis of tumor tissue will be completed, as depicted in Figure 6C.

Sampling
• These experiments will utilize 7, 8, or 14 mice per treatment group, for a total power of ≥80%.  A. Perform H&E staining by hand using the following procedure: i. Deparaffinize sections twice in xylene, then rehydrate through graded alcohols (95%, 70%, 50% ETOH) to water. ii. Stain sections with Carazzi's hematoxylin, then rinse slides in water. iii. Stain sections with eosin. iv. Dehydrate sections through graded alcohols (50%, 70%, 90%) and then place in xylene. v. Apply coverslips to slides with Permount and store slides at room temperature.

Blindly image stained sections and have images blindly analyzed by a Board Certified Veterinary
Pathologist to verify the tumor composition of the tissue sections.
Data to be collected  Figure 6C) 5. Pathologist's report of tissue section evaluation

Confirmatory analysis plan
This replication attempt will perform the statistical analyses listed below, compute the effects sizes, compare them against the reported effect size in the original paper, and use a meta-analytic approach to combine the original and replication effects, which will be presented as a Forest plot.
• Statistical analysis of the Replication Data: 1. Comparison of Kaplan-Meier survival curves tracking tumor incidence using Bonferroni's correction for multiple comparisons.
• The authors originally examined the Kaplan-Meier curves for PREX2 mutants and compared the endpoint values of the mutant curves to the endpoint values of the wild-type PREX2 curve using an unpaired two-tailed t-test. We will replicate their t-tests but also compare the entire survival curves (each mutant curve vs both wild-type and GFP control) using the log-rank Mantel-Cox test with Bonferroni's alpha correction, which we believe is a more appropriate statistical approach.

Comparison of tumor growth rates
• We will measure tumor growth rates across all mouse cohorts over the length of the study. These data were collected but not reported or analyzed in the original study. We will plot growth curves for each treatment group and use area under the curve analysis to calculate the mean and std. error. We will then use the means, std. error, and n to perform a 1-way ANOVA. Further, we will perform corrected t-tests (Bonferroni correction) to perform pairwise comparisons between PREX2 mutants and either GFP or wild-type controls.

Known differences from the original study
This replication is only generating and analyzing xenografts based on the stable melanocyte lines for GFP, wild-type PREX2, PREX2 Q1430*, and PREX2 G844D. The original study also generated and analyzed tumor xenografts using other PREX2 mutant-expressing melanocyte lines, including PREX2 K278*, E824*, P948S, and G106E. In order to sufficiently power all experiments and achieve the necessary number of events for Kaplan-Meier analysis, the duration of this replication will be extended from 9 weeks in the original paper to 16 weeks in the replication. All known differences in reagents and supplies are listed in the materials and reagents section above, with the originally used item listed in the comments section. All differences have the same capabilities as the original and are not expected to alter the experimental design.

Provisions for quality control
The genetic integrity, mycoplasma-free purity, and levels of exogenous expression of each NrasG12V melanocyte line used in this experiment have been previously validated in Protocols 1 and 2. All mice will be handled and housed in accordance with the Institutional Animal Care and Use Committee (IACUC). All data obtained from the experiment-raw data, data analysis, control data, and quality control data-will be made publicly available, either in the published manuscript or as an open access dataset available on the Open Science Framework (https://osf.io/82nfe/)

Power calculations Protocol 3 Summary of original data
Note: Mantel-Haenszel hazard ratios were generated in Graphpad Prism v. 6.0 following analysis of Kaplan-Meier curves with the log-rank (Mantel-Cox) test using the Mantel-Haenszel method.

Test family
• Log-rank (Mantel-Cox) test with Bonferroni alpha correction for multiple comparisons

Power calculations
• Performed with the Sample Size Calculator hosted by the Clinical & Translational Science Institute (CTSI) at the University of California-San Francisco (http://www.sample-size.net/sample-size-survivalanalysis/) (Rubinstein et al., 1981;Schoenfeld, 1983