Registered report: the microRNA miR-34a inhibits prostate cancer stem cells and metastasis by directly repressing CD44

The Reproducibility Project: Cancer Biology seeks to address growing concerns about reproducibility in scientific research by conducting replications of selected experiments from a number of high-profile papers in the field of cancer biology. The papers, which were published between 2010 and 2012, were selected on the basis of citations and Altimetric scores (Errington et al., 2014). This Registered report describes the proposed replication plan of key experiments from ‘The microRNA miR-34a inhibits prostate cancer stem cells and metastasis by directly repressing CD44’ by Liu and colleagues published in Nature Medicine in 2011 (Liu et al., 2011). Liu and colleagues first demonstrated that miR-34a levels were reduced in CD44+ prostate cancer cells (Figure 1B). They then showed that xenograft tumors from cells expressing exogenous miR-34a were smaller in size than control tumors (Supplemental Figure 5C). Tumors with exogenous miR-34a showed reduced levels of CD44 expression (Figure 4A), and mutation of two putative miR-34a binding sites in the CD33 3′ UTR partially abrogated signal repression in a luciferase assay (Figure 4D). 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.06434.001


Introduction
Many cancers display characteristic growth patterns associated with cancer stem cells (CSCs). In recent years, the importance of microRNAs in the initiation and maintenance of various cancers has become more widely explored (Maroof et al., 2014). In particular, miR-34 family members (miR-34a, miR-34b and miR-34c) have been identified as playing a role in the p53 pathway as well as influencing Notch and Met signaling (Misso et al., 2014), and are associated with inhibition of glioblastoma (Abel et al., 2009;Wiggins et al., 2010), pancreatic tumor cells (Corney et al., 2010) and gastric cancer (Ji et al., 2008). In their 2011 Nature Medicine paper, Liu and colleagues elucidated the importance of a particular microRNA, miR-34a, on the action of CD44 + prostate cancer putative CSCs, demonstrating that miR-34a was barely expressed in CD44 + CSCs and had tumor suppressing properties. They also showed that miR-34a affected the levels of CD44, a widespread marker of putative cancer stem cells (Liu et al., 2011).
In Figure 1B, Liu and colleagues purified CSCs from three xenograft cancer models; LAPC9, LAPC4 and Du145 cells. They assessed the levels of miR-34a expression as a percentage of expression from CD44 + cells vs CD44 − in each model by quantitative RT-PCR, as compared to levels of let-7b, a known tumor suppressive miRNA that is underexpressed in CD44 + cells. They found that miR-34a levels were markedly low in these xenograft models as compared to let-7b levels. This experiment will be replicated in Protocol 3.
In Supplemental Figure 5C, the authors demonstrated that the addition of exogenous miR-34a-through lentiviral infection of LAPC4 cells-decreased the size of resultant xenograft tumors, as compared to LAPC4 cells infected with a control lentivirus. This experiment demonstrates the functional relevance of miR-34a, and will be replicated in Protocol 4.
In Figure 4, Liu and colleagues present data supporting the hypothesis that CD44 is a target of miR-34a. In Figure 4D, they demonstrate that mutation of two putative miR-34a binding sites in the 3′ UTR of CD44 decreased signal in an in vitro luciferase assay (replicated in Protocol 7). They show evidence that tumors overexpressing miR-34a by lentiviral infection with miR-34a (Supplemental Figure 4A, replicated in Protocol 6) have reduced expression of CD44 ( Figure 4A, right panel (Western blots), conceptually replicated in Protocol 5).
Since the publication of Liu and colleagues' work, additional groups have identified roles for miR-34 family members in gallbladder cancer (Jin et al., 2013), mesothelioma (Toyooka, 2013) and breast cancer (Achari et al., 2014). In particular, Basak and colleagues explored the relationship of CD44 and miR-34a in breast cancer cells and showed that addition of exogenous miR-34a to CD44 hi cells reduced their colony forming efficiency (Basak et al., 2013). Fan and colleagues showed that miR-34a reduced the metastatic potential, invasion and migration of osteosarcoma cells (Fan, 2013). Siemens and colleagues also provide evidence that miR-34a regulates the expression of CD44 (Siemens et al., 2013(Siemens et al., , 2014.

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. An asterisk (*) indicates data or information provided by the Reproducibility Project: Cancer Biology core team. A hashtag (#) indicates information provided by the replicating lab.

Protocol 1: Maintenance of LAPC4 xenograft prostate tumors
This protocol describes the maintenance of the LAPC4 xenograft tumor tissue by serial implantation in NOD/SCID mice.

Confirmatory analysis plan
c Statistical Analysis of the Replication Data: ○ None required. c Meta-analysis of original and replication attempt effect sizes: ○ None required.

Known differences from the original study
The replicating lab will use a thoracentesis puncture needle to implant tumor tissue into recipient mice instead of making incisions. The replicating lab routinely uses this technique to implant tumor tissue into mice.
Tumor development characteristics will be recorded for each mouse. 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/gb7sr/).

Protocol 2: Purification of LAPC4 cells from xenograft tumors
This protocol describes how to isolate and separate human LAPC4 cells from xenografted tumors for further use in downstream protocols.
Sampling c No power calculations required.

Materials and reagents Procedure
Note: This protocol contains information described in papers from Patrawala and colleagues (2005), a. Pre-wet strainer with medium. b. Collect into a 50 ml centrifuge tube and resuspend in 1-3 ml IMDM with GLUTAMAX + 10% FBS. 8. Count viability of cells on a hemocytometer using erythrosine B.
a. If a large number of cells are obtained (>50 million live cells), then only a portion of that number should be used for further processing (flow cytometry time and sorting speed will put a limit on analyzing a maximum of ∼20 million cells). b. Adjust cell suspension to 1-1.5 million cells (dead + live) per ml with IMDM with GLUTAMAX + 10% FBS. 9. Load 3 ml of cell suspension gently onto a 3 ml layer of Histopaque-1077 gradient in a 15 ml tube and centrifuge at 400×g for 30 min at room temperature. a. Load with a 1 ml pipette tip. b. For larger volumes, 15 ml of cell suspension can be added to 15 ml of Histopaque in a 50 ml tube at a concentration of no more than 1 × 10 6 cells/ml (dead + live). 10. Collect live nucleated epithelial cells at the interface of the two layers (opaque layer) using a 1 ml pipette tip to transfer to a new tube. Spin at 130×g for 5 min at RT. a. If multiple tubes were used in Step 9, combine the live cells from all tubes before spinning. 11. Resuspend cells in 0.1-5 ml of IMDM with GLUTAMAX + 10% FBS, count the cell number, and spin again at 130×g for 5 min at RT. a. Wash cells again as per Miltenyi's instructions. 12. Deplete cell mixture of lineage-positive host (mouse) cells using the MACS Lineage Cell Depletion Kit. Refer to additional details in protocol from Miltenyi. (Volumes given are for 5 million cells.) a. Suspend cell pellet in 40 μl of staining buffer (PBS, pH 7.2, 0.5% FBS, 5 μg/ml insulin, 0.5% BSA). b. Add 10 μl of supplied biotin-antibody cocktail (CD5, CD45R, CD11b, anti-Ly-6G, 7-4, and Ter-119). c. Mix well and incubate for 10 min at 4˚C in the dark. d. Add 30 μl of staining buffer. e. Add 20 μl supplied anti-biotin microbeads. f. Mix well and incubate for additional 15 min at 4˚C in the dark. g. Wash cells by adding 10-20× labeling volume in cold PBS and centrifuge at 130×g for 5 min at 4˚C. Pipette off supernatant completely. h. Resuspend up to 10 7 cells in 500 μl staining buffer. i. Assemble MACS apparatus by attaching the MACS Separator on the stand and placing the MACS MS column in the magnetic field with a 15 ml collection tube below the column. j. Prepare the column by rinsing with 500 μl of staining buffer. k. Apply cell suspension to the column. Allow cells to pass by gravity and collect the effluent as fraction with unlabeled cells, representing the enriched human tumor cell fraction depleted of mouse cells. l. Wash the column with 500 μl of staining buffer and collect the effluent in the same tube as effluent of step k. Perform 3×, each time once the column reservoir is empty. m. Centrifuge the effluent at 130×g for 5 min at 4˚C and discard supernatant. 13. # Confirm the purity of the cells using FACS.
a. Prepare the PE-labeled human EpCAM following the manufacture's instruction for the primary and isotype staining, calculating enough total volume to add 100 μl per sample Ab stain in cold Stain buffer. b. Use cold Stain buffer (PBS, pH 7.2, 0.5% FBS, 0.5% BSA) when preparing primary antibodies and isotypes. c. Add 100 μl per sample of diluted antibodies to the 5 ml FACS tube and resuspend cells by gently pipetting. d. Incubate for 45 min at 4˚C, protected from light. e. Add 200 μl cold Stain Buffer to each tube and centrifuge at 1300 rpm (240×g) × 5 min at 4˚C. f. Remove supernantant and resuspend in cold BD Stain Buffer at 200 μl/well and then pellet at 1300 rpm (240×g) × 5 min at 4˚C. Repeat wash two times. g. Add 100 μl cold Stain Buffer/tube to resuspend cells and analyze samples immediately using BD FACSCalibur. Cell debris and cell aggregates will be gated out based on the FSC/SSC parameters. 14. Resuspend cells in 0.1-5 ml of IMDM with GLUTAMAX + 10% FBS and count the cell number. 15. Use resuspended cells for further analysis (Protocol 3, Protocol 4, and Protocol 7).
Deliverables c Data to be collected: ○ Weight of tumor tissue. ○ Cell viability counts before and after Histopaque gradient and after MACS depletion. ○ All FACS plots from assessing the purity of the cells. ○ Percentage of LAPC4 cells in enriched cell population. c Sample delivered for further analysis: ○ Resuspended purified LAPC4 cells (for Protocols 3, 4 and 8).

Confirmatory analysis plan
c Statistical Analysis of the Replication Data: ○ None required. c Meta-analysis of original and replication attempt effect sizes: ○ None required.
Known differences from the original study c None noted.

Provisions for quality control
The purity of the cells after depletion of lineage-positive cells will be determined by FACS. 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/gb7sr/).

Protocol 3: Expression of miR-34a in LAPC4 sub-populations
This protocol describes how to assess expression levels of miR-34a by qRT-PCR in isolated LAPC4 xenograft tumor cells, as seen in Figure 1B.
Sampling c The original data presented is qualitative. Based on power calculations with a range of possible variance, we will perform the experiment three times for a final estimated power of at least 82.2%. ○ See Power calculations for details. c Each experiment consists of two cohorts: ○ Cohort 1: CD44 + LAPC4 cells. ○ Cohort 2: CD44 − LAPC4 cells. ○ Within each cohort, assess levels of three miRNAs: ■ Let-7b. ■ miR-34a. ■ miR-103. c These qRT-PCR reactions are run in technical triplicate.
1. CD44-based purification by FACS (volumes given below are for 5 million cells). Note: This step contains information described in colleagues (2006), (2007) and Li and colleagues (2009). a. Centrifuge the single cell suspension from Protocol 2 at 380×g for 5 min at 4˚C and resuspend cells in 90 μl of cold staining buffer (PBS, pH 7.2, 0.5% BSA, 5 μg/ml insulin). i. Remove 10 μl from cell suspension and add to 90 μl staining buffer in another tube for isotype control staining. b. Add 20 μl of FcR blocking reagent to the remaining 80 μl of cell suspension and incubate at 4˚C in the dark for 10 min. c. Add 10 μl of FITC-conjugated monoclonal anti-CD44 to the cell suspension, mix by tapping, and incubate at 4˚C in the dark for 15 min. i. Note: antibody comes prediluted from the manufacturer and no additional dilution is necessary. ii. Gently tap every 5 min to prevent cells from settling down. iii. For isotype control use 10 μl FITC-mouse IgG2b isotype control immunoglobulin. d. Add 5 ml of cold PBS and spin at 380×g for 5 min at 4˚C. Resuspend cells in cold serum-free IMDM with GLUTAMAX + 1% pen/strep at a concentration of 2 × 10 6 cells/ml in a 5 ml polystyrene tube at no more than 2.5 ml/tube. i. Keep on ice and shielded from light. e. Prepare collection tubes using 5 ml polypropylene tubes containing 1 ml of FBS + 1% pen/strep. f. Add 7-AAD to 1 μg/ml (stock = 100 μg/ml) to the sort tube 10 min before analysis.

Confirmatory analysis plan
c Statistical Analysis of the Replication Data: ○ At the time of analysis, we will perform the Shapiro-Wilk test and generate a quantile-quantile (q-q) plot to attempt to assess the normality of the data and also perform Levene's test to assess homoscedasiticity. If the data appears skewed, we will attempt a transformation in order to proceed with the proposed statistical analysis listed below and possibly perform the appropriate non-parametric test. ■ Compare percentage expression levels of miR-34a to levels of let-7b. c unpaired two-tailed Student's t-test. c Meta-analysis of original and replication attempt effect sizes: ○ Because the original data is qualitative, a meta-analysis cannot be performed. The replication data will be plotted as means and 95% confidence intervals and the original data will be presented as a single point on the same plot for comparison.
Known differences from the original study c The replication will be restricted to LAPC4 cells only. c The replicating lab will use a Stratagene MX3005P Real-time PCR system in place of the ABI Prism 7900 SDS used by the original authors. c Based on the recommendation of the original authors, total RNA will be isolated in Step 2a instead of isolating only the small RNA fraction.

Provisions for quality control
RNA sample purity (A 260/280 and A 260/230 ratios) will be reported for each sample. An isotype control will be used during the FACS experiment. 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/gb7sr/).

Protocol 4: Effects of miR-34a on tumor growth
This protocol describes how to inject LAPC4 tumor cells infected with a lentivirus containing miR-34a into NOD/SCID mice and measure the resultant tumor size, as seen in Supplemental Figure 5C.
Sampling c This experiment will be performed with 7 mice per group for a final power of 81.2%. ○ See Power Calculations section for details. c The experiment consists of three cohorts of NOD/SCID mice: ○ Cohort 1: injected with uninfected LAPC4 cells (additional control). ■ N = 7. c Based on power calculations performed for the original two groups. ○ Cohort 2: injected with control-infected LAPC4 cells.
■ N = 7. i. This initial collected media can be stored briefly at 4˚C. e. After an additional 12-24 hr of culture, collect viral supernatants again and pool with first collection. f. Concentrate viral stock.

Materials and reagents
i. Centrifuge the viral supernatant at 3000 rpm for 15 min to remove any cell debris. ii. Filter the supernatant through a 0.45 μm syringe filter. iii. Ultracentrifuge at 22,000 rpm for 2 hr at 4˚C to produce concentrated viral stocks. g. Determine lentivirus titer for GFP using HT1080 cells.
i. 1 day before harvesting viral supernatant, plate 1.2 × 10 5 HT1080 cells per well of a six well dish. ii. On the day of viral supernatant harvesting, count the number of cells in one well to determine cell number at time of infection. iii. Add a range of volumes between 2 and 5 μl of concentrated viral supernatant to the wells.
Incubate for 72 hr. iv. Using a fluorescent microscope, count the number of GFP-positive cells per well; roughly 1-10% of cells will be GFP positive.   Figure 5C). c Sample delivered for further analysis: ○ Tumor tissue processed for protein extraction (Protocol 5) and RNA extraction (Protocol 6). ○ Snap frozen cells for RNA extraction (Protocol 6).

Confirmatory analysis plan
c Statistical Analysis of the Replication Data: ○ At the time of analysis, we will perform the Shapiro-Wilk test and generate a quantile-quantile (q-q) plot to attempt to assess the normality of the data and also perform Levene's test to assess homoscedasiticity. If the data appears skewed, we will attempt a transformation in order to proceed with the proposed statistical analysis listed below and possibly perform the appropriate non-parametric test. ■ Compare mean tumor weight in miR-34a expressed tumors relative to lenti-ctl control tumors. Also compare to uninfected control tumors (additional control). c One-way ANOVA followed by planned comparisons using Fisher's LSD.
○ Lenti-control vs miR-34a. ○ Uninfected vs miR-34a. ■ Compare tumor incidence in miR-34a expressed tumors relative to lenti-ctl tumors. Also compare to uninfected control tumors (additional control). c One-way ANOVA followed by planned comparisons using Fisher's LSD. ○ Lenti-control vs miR-34a. ○ Uninfected vs miR-34a. ■ Compare tumor latency in miR-34a expressed tumors relative to lenti-ctl tumors. Also compare to uninfected control tumors (additional control). c Fisher's Exact Test. ■ Compare tumor volume over time in miR-34a expressed tumors relative to lenti-ctl tumors. Also compare to uninfected control tumors (additional control). c Calculate area under the curve (AUC) for each mouse and perform a one-way ANOVA followed by planned comparisons using Fisher's LSD. c Lenti-control vs miR-34a. c Uninfected vs miR-34a. c Meta-analysis of original and replication attempt effect sizes: ○ This replication attempt will perform the statistical analysis listed above, compute the effects sizes, compare them against the reported effect size in the original paper and use a metaanalytic approach to combine the original and replication effects, which will be presented as a forest plot.
Known differences from the original study

Provisions for quality control
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/gb7sr/). c We are confirming that >90% of injected LAPC4 cells have been successfully transduced with the vector. c We are recording tumor latency and tumor volume as additional parameters. c Only control tumors' weight will be used to determine the experimental endpoint for tumor growth.
Protocol 5: Western blot analysis of miR34a infected LAPC4 tumor tissue (additional related experiment) This protocol describes Western blot analysis of CD44 levels from tumor tissue derived from Protocol 4. In the original study, this was performed in Figure 4A (right panels) on tumor tissues derived from DU145 and PC3 xenograft tumors. This experiment, however, will be performed on LAPC4 xenograft tumor tissue to maintain consistency of tumor materials throughout the replication plan and is thus exploratory in nature.
Sampling c This protocol will use the samples derived from Protocol 4. ○ Original data presented is qualitative; representative images. The replication attempt is exploratory in nature and not looking for a specific effect. c The experiment consists of three cohorts: ○ Cohort 1: tumors from uninfected LAPC4 cells. ○ Cohort 2: tumors from control-infected LAPC4 cells. ○ Cohort 3: tumors from miR-34-infected LAPC4 cells. ○ Each cohort is probed by Western blot for: ■ Anti-CD44. ■ Anti-beta-actin.
Note: at authors' recommendation, the replicating lab will use their standard western blot protocol.
1. Make a total protein lysate from tumor tissue from Protocol 4. a. Transfer a piece of tumor sample into a 5 ml polypropylene tube containing 800-1000 μl lysis buffer. i. Lysis buffer: Add 1 tablet of complete-EDTA free tablet, 100 μl Phosphatase Inhibitor Cocktail II and II to 10 ml RIPA buffer. b. Homogenize the tumor sample on ice with setting on 6 (equal to 30,000 rpm) for 5 s on and 5 s off until it is homogenized completely. c. Transfer the lysate into a 1.5 ml tube and sonicate on ice for 15 s. d. Spin at 13,000 rpm for 20 min at 4˚C to remove tissue debris. e. Spin the resultant supernatant at 13,000 rpm for 20 min at 4˚C. f. Aliquot the supernatant and store at −80˚C. 2. Quantify total protein concentration.
Deliverables c Data to be collected: ○ Full images, including ladder, of western blots for CD44 and beta-actin for each tumor sample. ○ Calculate band intensity for each band as ratio to loading control band.

Exploratory analysis plan
c Statistical Analysis of the Replication Data: ○ At the time of analysis, we will perform the Shapiro-Wilk test and generate a quantile-quantile (q-q) plot to attempt to assess the normality of the data and also perform Levene's test to assess homoscedasiticity. If the data appears skewed, we will attempt a transformation in order to proceed with the proposed statistical analysis listed below and possibly perform the appropriate non-parametric test. ■ Compare CD44 protein across all three conditions. c One way ANOVA followed by planned comparisons using Fisher's LSD: ○ CD44 protein levels in miR-34a expressed tumors vs to lenti-ctl tumors. ○ CD44 protein levels in miR-34a expressed tumors vs to uninfected control tumors (additional control). c Meta-analysis of original and replication attempt effect sizes: ○ Meta-analysis will not be performed as this experiment is exploratory in nature and there is no matching original data with which to statistically compare the replication data.
Known differences from the original study c While the original study performed Western blot analysis on DU145 and PC3 xenograft tumor tissue, this experiment will perform the same analysis on LAPC4 xenograft tumor tissue. c Uninjected LAPC4 cells and tumors derived from uninjected LAPC4 cells have been added as an additional control. c Because of these changes, the experiment is not directly confirmatory in nature, but rather exploratory. Thus, we will not directly power the experiment, but we have determined our sensitivity based on the number of samples we will use.

Provisions for quality control
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/gb7sr/).

Protocol 6: qRT-PCR analysis of miR34 infected LAPC4 cells and tumor tissue
This protocol describes how to assess miR-34a levels in tumor tissue derived from lenti-miR-34a infected LAPC4 cells. This is an additional experiment added by the RP:CB core team and is partially based on Supplemental Figure 4A. It is a quality control check to ensure that increased levels of miR-34a are still present even after the infected LAPC4 cells have been injected into mice to form tumors.
Sampling c This protocol will use samples derived from Protocol 4. ○ Original data presented is qualitative; representative images. The replication attempt is exploratory in nature and not looking for a specific effect. c The experiment consists of six cohorts: ○ RNA derived from uninfected LAPC4 cells. ○ RNA derived from lenti-ctl LAPC4 cells. ○ RNA derived from lenti-miR-34a infected LAPC4 cells. ■ Each cohort of cells has three technical replicates. ○ RNA derived from uninfected LAPC4 xenograft tumor tissue. ○ RNA derived from lenti-ctl LAPC4 xenograft tumor tissue. ○ RNA derived from lenti-miR-34a infected LAPC4 xenograft tumor tissue.

Materials and reagents Procedure
Note: This procedure contains information described in Wiggins and colleagues (2010). Exploratory analysis plan c Statistical Analysis of the Replication Data: ○ At the time of analysis, we will perform the Shapiro-Wilk test and generate a quantile-quantile (q-q) plot to attempt to assess the normality of the data and also perform Levene's test to assess homoscedasiticity. If the data appears skewed, we will attempt a transformation in order to proceed with the proposed statistical analysis listed below and possibly perform the appropriate non-parametric test. ■ One way ANOVA of miR-34a levels across all three conditions in cells before injection followed by planned comparisons using Fisher's LSD: c Lenti-clt control cells compared to miR-34a expressed cells. c Uninfected cells compared to miR-34a expressed cells. ■ One way ANOVA of miR-34a levels across all three conditions in tumors followed by planned comparisons using Fisher's LSD: c Lenti-clt control tumors compared to miR-34a expressed tumors. c Uninfected tumors compared to miR-34a expressed tumors. c Meta-analysis of original and replication attempt effect sizes: ○ Meta-analysis will not be performed as this experiment is exploratory in nature and there is no matching original data with which to statistically compare the replication data.
Known differences from the original study c The replication will be restricted to LAPC4 cells. c We will run the qRT-PCR analysis on tumors derived from the treated LAPC4 cells as well as on the infected LAPC4 cells directly. c Uninjected LAPC4 cells and tumors derived from uninjected LAPC4 cells have been added as an additional control. c Because of these changes, the experiment is not directly confirmatory in nature, but rather exploratory. Thus, we will not directly power the experiment, but we have determined our sensitivity based on the number of samples we will use.

Provisions for quality control
RNA sample purity (A 260/280 and A 260/230 ratios) will be reported for each sample. 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/gb7sr/).
Protocol 7: Luciferase assays confirming binding of miR-34a to putative binding sites in the 3′ UTR of CD44 This protocol describes how to perform luciferase assays in order to demonstrate that binding of miR-34a to its binding sites in the 3′ UTR of CD44 decreases luciferase activity. It also tests whether this decrease is abrogated when the seed regions of the putative miR-34a binding sites are mutated, as seen in Figure 4D. The replication will use only the pMIR-CD443′ UTR and pMIR-CD44M1M23′ UTR constructs.
5. 48 hr later, read the ratio of luciferase to renilla using the Promega dual luciferase assay according to the manufacturer's instructions. 6. Repeat experiment from Step 3 onwards an additional 15 times.
Deliverables c Data to be collected: ○ Sequence data confirming vector identity. ○ Whole gel images of agarose gels confirming vector integrity. ○ Raw luciferase and renilla readings for each well. ○ Normalized renilla and luciferase readings based on empty well and cells-only wells. ○ Ratio of luciferase to renilla averaged across the four wells of each condition.

Confirmatory analysis plan
c At the time of analysis, we will perform the Shapiro-Wilk test and generate a quantile-quantile (q-q) plot to attempt to assess the normality of the data and also perform Levene's test to assess homoscedasiticity. If the data appears skewed, we will attempt a transformation in order to proceed with the proposed statistical analysis listed below and possibly perform the appropriate nonparametric test. ○ Statistical Analysis of the Replication Data: ■ Two way ANOVA (2 × 2 factorial) followed by Bonferroni corrected pairwise comparisons: c Wt+NC vs wt+34a. c Meta-analysis of original and replication attempt effect sizes: ○ This replication attempt will perform the statistical analysis listed above, compute the effects sizes, compare them against the reported effect size in the original paper and use a metaanalytic approach to combine the original and replication effects, which will be presented as a forest plot.
Known differences from the original study c The replication attempt will use only the wild-type CD44 3′ UTR anad the 3′ UTR with both miR-34a seed regions mutated; it will exclude the two constructs containing one mutated seed region.

Provisions for quality control
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/gb7sr/). c STR profiling and mycoplasma testing results. c Sequencing results and agarose gel images confirming vector integrity.

Protocol 3
Summary of original data c Original data is qualitative; only one experiment is shown without any error bars. c Values were estimated from graph. c Since no variance or sample size data were provided, power calculations will be performed with a range of variances and an assumed N per group of 3.

Test family
c Unpaired two-tailed Student's t-test, alpha error = 0.05.
In order to produce quantitative replication data, we will run the replication experiment three times. We will then determine the replication attempt's mean and variance from those three replicates. We will combine the replication variance with the original data, whose single datapoint we are assuming represents a mean. Combining the original mean and the replication variance will generate a simulated effect size. Using this simulated effect size, we will then determine the number of replicates required to reach 80% power and will perform additional replicates if necessary to ensure the replication is powered to at least 80%.

Protocol 4
Summary of original data c Note: Original values presented in Supplemental Figure 5C. Authors confirmed tumor incidence was 100%.
Test family c One-way ANOVA followed by planned comparisons using Fisher's LSD. Supp. Figure  Power calculations c Calculations were performed using R software, version 3.1.2 (R Core Team, 2014) and G*Power software (version 3.1.7) (Faul et al., 2007).
c Note: the additional control group of uninfected cells was assumed to have the same mean and variance as the lenti-control group for calculation purposes.

Protocol 5
c This experiment differs from the original it is based on, and is thus exploratory in nature. The number of sample is derived from Protocol 4. c Based on the sample size from Protocol 4, with α of 0.05, we will be powered to 80% to detect an effect size f of 0.7379139 (ANOVA: Fixed effects, omnibus, one-way) and with α of 0.05, we will be powered to 80% to detect an effect size d of 1.6317141 (unpaired two-tailed t-test).

Protocol 6
c This experiment differs from the original it is based on, and is thus exploratory in nature. The number of samples is derived from Protocol 4.

Sensitivity calculations
c Calculations were performed using R software, version 3.1.2 (R Core Team, 2014) and G*Power software (version 3.1.7) (Faul et al., 2007). c Based on the sample size from Protocol 4, with α of 0.05, we will be powered to 80% to detect an effect size f of 0.7379139 (ANOVA: Fixed effects, omnibus, one-way) and with α of 0.05, we will be powered to 80% to detect an effect size d of 1.6317141 (unpaired two-tailed t-test) for tumors. c Based on the sample size from Protocol 4, with α of 0.05, we will be powered to 80% to detect an effect size f of 1.3573433 (ANOVA: Fixed effects, omnibus, one-way) and with α of 0.05, we will be powered to 80% to detect an effect size d of 3.0708923 (unpaired two-tailed t-test) for tumors.

Protocol 7
Summary of original data c Note: data provided by original authors.